Mercurial > vim
annotate src/spell.c @ 6999:dc1b678f0e4e v7.4.817
patch 7.4.817
Problem: Invalid memory access in file_pat_to_reg_pat().
Solution: Use vim_isspace() instead of checking for a space only. (Dominique
Pelle)
| author | Bram Moolenaar <bram@vim.org> |
|---|---|
| date | Tue, 11 Aug 2015 16:20:05 +0200 |
| parents | 1e621b31948b |
| children | ad432f8f68fb |
| rev | line source |
|---|---|
| 223 | 1 /* vi:set ts=8 sts=4 sw=4: |
| 2 * | |
| 3 * VIM - Vi IMproved by Bram Moolenaar | |
| 4 * | |
| 5 * Do ":help uganda" in Vim to read copying and usage conditions. | |
| 6 * Do ":help credits" in Vim to see a list of people who contributed. | |
| 7 * See README.txt for an overview of the Vim source code. | |
| 8 */ | |
| 9 | |
| 10 /* | |
| 11 * spell.c: code for spell checking | |
| 226 | 12 * |
| 300 | 13 * The spell checking mechanism uses a tree (aka trie). Each node in the tree |
| 14 * has a list of bytes that can appear (siblings). For each byte there is a | |
| 15 * pointer to the node with the byte that follows in the word (child). | |
| 324 | 16 * |
| 17 * A NUL byte is used where the word may end. The bytes are sorted, so that | |
| 18 * binary searching can be used and the NUL bytes are at the start. The | |
| 19 * number of possible bytes is stored before the list of bytes. | |
| 20 * | |
| 21 * The tree uses two arrays: "byts" stores the characters, "idxs" stores | |
| 22 * either the next index or flags. The tree starts at index 0. For example, | |
| 23 * to lookup "vi" this sequence is followed: | |
| 24 * i = 0 | |
| 25 * len = byts[i] | |
| 26 * n = where "v" appears in byts[i + 1] to byts[i + len] | |
| 27 * i = idxs[n] | |
| 28 * len = byts[i] | |
| 29 * n = where "i" appears in byts[i + 1] to byts[i + len] | |
| 30 * i = idxs[n] | |
| 31 * len = byts[i] | |
| 32 * find that byts[i + 1] is 0, idxs[i + 1] has flags for "vi". | |
| 300 | 33 * |
| 339 | 34 * There are two word trees: one with case-folded words and one with words in |
| 300 | 35 * original case. The second one is only used for keep-case words and is |
| 36 * usually small. | |
| 37 * | |
| 481 | 38 * There is one additional tree for when not all prefixes are applied when |
| 339 | 39 * generating the .spl file. This tree stores all the possible prefixes, as |
| 40 * if they were words. At each word (prefix) end the prefix nr is stored, the | |
| 41 * following word must support this prefix nr. And the condition nr is | |
| 42 * stored, used to lookup the condition that the word must match with. | |
| 43 * | |
| 300 | 44 * Thanks to Olaf Seibert for providing an example implementation of this tree |
| 45 * and the compression mechanism. | |
| 625 | 46 * LZ trie ideas: |
| 47 * http://www.irb.hr/hr/home/ristov/papers/RistovLZtrieRevision1.pdf | |
| 48 * More papers: http://www-igm.univ-mlv.fr/~laporte/publi_en.html | |
| 243 | 49 * |
| 50 * Matching involves checking the caps type: Onecap ALLCAP KeepCap. | |
| 51 * | |
| 236 | 52 * Why doesn't Vim use aspell/ispell/myspell/etc.? |
| 53 * See ":help develop-spell". | |
| 54 */ | |
| 55 | |
| 469 | 56 /* Use SPELL_PRINTTREE for debugging: dump the word tree after adding a word. |
| 482 | 57 * Only use it for small word lists! */ |
| 469 | 58 #if 0 |
| 59 # define SPELL_PRINTTREE | |
| 482 | 60 #endif |
| 61 | |
| 633 | 62 /* Use DEBUG_TRIEWALK to print the changes made in suggest_trie_walk() for a |
| 63 * specific word. */ | |
| 625 | 64 #if 0 |
| 65 # define DEBUG_TRIEWALK | |
| 66 #endif | |
| 67 | |
| 300 | 68 /* |
| 324 | 69 * Use this to adjust the score after finding suggestions, based on the |
| 70 * suggested word sounding like the bad word. This is much faster than doing | |
| 71 * it for every possible suggestion. | |
| 625 | 72 * Disadvantage: When "the" is typed as "hte" it sounds quite different ("@" |
| 73 * vs "ht") and goes down in the list. | |
| 344 | 74 * Used when 'spellsuggest' is set to "best". |
| 75 */ | |
| 76 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4) | |
| 77 | |
| 78 /* | |
| 625 | 79 * Do the opposite: based on a maximum end score and a known sound score, |
| 1705 | 80 * compute the maximum word score that can be used. |
| 625 | 81 */ |
| 82 #define MAXSCORE(word_score, sound_score) ((4 * word_score - sound_score) / 3) | |
| 83 | |
| 84 /* | |
| 339 | 85 * Vim spell file format: <HEADER> |
| 483 | 86 * <SECTIONS> |
| 339 | 87 * <LWORDTREE> |
| 88 * <KWORDTREE> | |
| 89 * <PREFIXTREE> | |
| 300 | 90 * |
| 483 | 91 * <HEADER>: <fileID> <versionnr> |
| 92 * | |
| 93 * <fileID> 8 bytes "VIMspell" | |
| 94 * <versionnr> 1 byte VIMSPELLVERSION | |
| 95 * | |
| 96 * | |
| 97 * Sections make it possible to add information to the .spl file without | |
| 98 * making it incompatible with previous versions. There are two kinds of | |
| 99 * sections: | |
| 100 * 1. Not essential for correct spell checking. E.g. for making suggestions. | |
| 101 * These are skipped when not supported. | |
| 102 * 2. Optional information, but essential for spell checking when present. | |
| 103 * E.g. conditions for affixes. When this section is present but not | |
| 104 * supported an error message is given. | |
| 300 | 105 * |
| 483 | 106 * <SECTIONS>: <section> ... <sectionend> |
| 107 * | |
| 108 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents) | |
| 109 * | |
| 110 * <sectionID> 1 byte number from 0 to 254 identifying the section | |
| 111 * | |
| 112 * <sectionflags> 1 byte SNF_REQUIRED: this section is required for correct | |
| 113 * spell checking | |
| 114 * | |
| 115 * <sectionlen> 4 bytes length of section contents, MSB first | |
| 116 * | |
| 117 * <sectionend> 1 byte SN_END | |
| 118 * | |
| 119 * | |
| 714 | 120 * sectionID == SN_INFO: <infotext> |
| 121 * <infotext> N bytes free format text with spell file info (version, | |
| 122 * website, etc) | |
| 123 * | |
| 483 | 124 * sectionID == SN_REGION: <regionname> ... |
| 125 * <regionname> 2 bytes Up to 8 region names: ca, au, etc. Lower case. | |
| 300 | 126 * First <regionname> is region 1. |
| 127 * | |
| 483 | 128 * sectionID == SN_CHARFLAGS: <charflagslen> <charflags> |
| 129 * <folcharslen> <folchars> | |
| 300 | 130 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128). |
| 131 * <charflags> N bytes List of flags (first one is for character 128): | |
| 324 | 132 * 0x01 word character CF_WORD |
| 133 * 0x02 upper-case character CF_UPPER | |
| 483 | 134 * <folcharslen> 2 bytes Number of bytes in <folchars>. |
| 135 * <folchars> N bytes Folded characters, first one is for character 128. | |
| 300 | 136 * |
| 483 | 137 * sectionID == SN_MIDWORD: <midword> |
| 138 * <midword> N bytes Characters that are word characters only when used | |
| 366 | 139 * in the middle of a word. |
| 140 * | |
| 483 | 141 * sectionID == SN_PREFCOND: <prefcondcnt> <prefcond> ... |
| 339 | 142 * <prefcondcnt> 2 bytes Number of <prefcond> items following. |
| 143 * <prefcond> : <condlen> <condstr> | |
| 144 * <condlen> 1 byte Length of <condstr>. | |
| 145 * <condstr> N bytes Condition for the prefix. | |
| 146 * | |
| 483 | 147 * sectionID == SN_REP: <repcount> <rep> ... |
| 148 * <repcount> 2 bytes number of <rep> items, MSB first. | |
| 323 | 149 * <rep> : <repfromlen> <repfrom> <reptolen> <repto> |
| 483 | 150 * <repfromlen> 1 byte length of <repfrom> |
| 151 * <repfrom> N bytes "from" part of replacement | |
| 152 * <reptolen> 1 byte length of <repto> | |
| 153 * <repto> N bytes "to" part of replacement | |
| 323 | 154 * |
| 625 | 155 * sectionID == SN_REPSAL: <repcount> <rep> ... |
| 156 * just like SN_REP but for soundfolded words | |
| 157 * | |
| 483 | 158 * sectionID == SN_SAL: <salflags> <salcount> <sal> ... |
| 159 * <salflags> 1 byte flags for soundsalike conversion: | |
| 323 | 160 * SAL_F0LLOWUP |
| 161 * SAL_COLLAPSE | |
| 162 * SAL_REM_ACCENTS | |
| 483 | 163 * <salcount> 2 bytes number of <sal> items following |
| 323 | 164 * <sal> : <salfromlen> <salfrom> <saltolen> <salto> |
| 483 | 165 * <salfromlen> 1 byte length of <salfrom> |
| 166 * <salfrom> N bytes "from" part of soundsalike | |
| 167 * <saltolen> 1 byte length of <salto> | |
| 168 * <salto> N bytes "to" part of soundsalike | |
| 323 | 169 * |
| 483 | 170 * sectionID == SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto> |
| 171 * <sofofromlen> 2 bytes length of <sofofrom> | |
| 172 * <sofofrom> N bytes "from" part of soundfold | |
| 173 * <sofotolen> 2 bytes length of <sofoto> | |
| 174 * <sofoto> N bytes "to" part of soundfold | |
| 323 | 175 * |
| 625 | 176 * sectionID == SN_SUGFILE: <timestamp> |
| 177 * <timestamp> 8 bytes time in seconds that must match with .sug file | |
| 178 * | |
| 698 | 179 * sectionID == SN_NOSPLITSUGS: nothing |
| 180 * | |
| 625 | 181 * sectionID == SN_WORDS: <word> ... |
| 182 * <word> N bytes NUL terminated common word | |
| 183 * | |
| 483 | 184 * sectionID == SN_MAP: <mapstr> |
| 185 * <mapstr> N bytes String with sequences of similar characters, | |
| 323 | 186 * separated by slashes. |
| 300 | 187 * |
| 809 | 188 * sectionID == SN_COMPOUND: <compmax> <compminlen> <compsylmax> <compoptions> |
| 189 * <comppatcount> <comppattern> ... <compflags> | |
| 483 | 190 * <compmax> 1 byte Maximum nr of words in compound word. |
| 191 * <compminlen> 1 byte Minimal word length for compounding. | |
| 192 * <compsylmax> 1 byte Maximum nr of syllables in compound word. | |
| 809 | 193 * <compoptions> 2 bytes COMP_ flags. |
| 194 * <comppatcount> 2 bytes number of <comppattern> following | |
| 714 | 195 * <compflags> N bytes Flags from COMPOUNDRULE items, separated by |
| 483 | 196 * slashes. |
| 197 * | |
| 809 | 198 * <comppattern>: <comppatlen> <comppattext> |
| 199 * <comppatlen> 1 byte length of <comppattext> | |
| 200 * <comppattext> N bytes end or begin chars from CHECKCOMPOUNDPATTERN | |
| 201 * | |
| 202 * sectionID == SN_NOBREAK: (empty, its presence is what matters) | |
| 492 | 203 * |
| 483 | 204 * sectionID == SN_SYLLABLE: <syllable> |
| 205 * <syllable> N bytes String from SYLLABLE item. | |
| 300 | 206 * |
| 207 * <LWORDTREE>: <wordtree> | |
| 208 * | |
| 339 | 209 * <KWORDTREE>: <wordtree> |
| 210 * | |
| 211 * <PREFIXTREE>: <wordtree> | |
| 212 * | |
| 213 * | |
| 300 | 214 * <wordtree>: <nodecount> <nodedata> ... |
| 215 * | |
| 216 * <nodecount> 4 bytes Number of nodes following. MSB first. | |
| 217 * | |
| 218 * <nodedata>: <siblingcount> <sibling> ... | |
| 219 * | |
| 220 * <siblingcount> 1 byte Number of siblings in this node. The siblings | |
| 221 * follow in sorted order. | |
| 222 * | |
| 339 | 223 * <sibling>: <byte> [ <nodeidx> <xbyte> |
| 481 | 224 * | <flags> [<flags2>] [<region>] [<affixID>] |
| 225 * | [<pflags>] <affixID> <prefcondnr> ] | |
| 300 | 226 * |
| 227 * <byte> 1 byte Byte value of the sibling. Special cases: | |
| 228 * BY_NOFLAGS: End of word without flags and for all | |
| 229 * regions. | |
| 481 | 230 * For PREFIXTREE <affixID> and |
| 366 | 231 * <prefcondnr> follow. |
| 390 | 232 * BY_FLAGS: End of word, <flags> follow. |
| 481 | 233 * For PREFIXTREE <pflags>, <affixID> |
| 455 | 234 * and <prefcondnr> follow. |
| 390 | 235 * BY_FLAGS2: End of word, <flags> and <flags2> |
| 236 * follow. Not used in PREFIXTREE. | |
| 237 * BY_INDEX: Child of sibling is shared, <nodeidx> | |
| 300 | 238 * and <xbyte> follow. |
| 239 * | |
| 240 * <nodeidx> 3 bytes Index of child for this sibling, MSB first. | |
| 241 * | |
| 242 * <xbyte> 1 byte byte value of the sibling. | |
| 243 * | |
| 244 * <flags> 1 byte bitmask of: | |
| 245 * WF_ALLCAP word must have only capitals | |
| 246 * WF_ONECAP first char of word must be capital | |
| 389 | 247 * WF_KEEPCAP keep-case word |
| 248 * WF_FIXCAP keep-case word, all caps not allowed | |
| 300 | 249 * WF_RARE rare word |
| 389 | 250 * WF_BANNED bad word |
| 300 | 251 * WF_REGION <region> follows |
| 481 | 252 * WF_AFX <affixID> follows |
| 300 | 253 * |
| 500 | 254 * <flags2> 1 byte Bitmask of: |
| 390 | 255 * WF_HAS_AFF >> 8 word includes affix |
| 500 | 256 * WF_NEEDCOMP >> 8 word only valid in compound |
| 698 | 257 * WF_NOSUGGEST >> 8 word not used for suggestions |
| 809 | 258 * WF_COMPROOT >> 8 word already a compound |
| 819 | 259 * WF_NOCOMPBEF >> 8 no compounding before this word |
| 260 * WF_NOCOMPAFT >> 8 no compounding after this word | |
| 390 | 261 * |
| 455 | 262 * <pflags> 1 byte bitmask of: |
| 263 * WFP_RARE rare prefix | |
| 264 * WFP_NC non-combining prefix | |
| 265 * WFP_UP letter after prefix made upper case | |
| 266 * | |
| 300 | 267 * <region> 1 byte Bitmask for regions in which word is valid. When |
| 268 * omitted it's valid in all regions. | |
| 269 * Lowest bit is for region 1. | |
| 270 * | |
| 481 | 271 * <affixID> 1 byte ID of affix that can be used with this word. In |
| 339 | 272 * PREFIXTREE used for the required prefix ID. |
| 273 * | |
| 274 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list | |
| 275 * from HEADER. | |
| 300 | 276 * |
| 277 * All text characters are in 'encoding', but stored as single bytes. | |
| 278 */ | |
| 279 | |
| 625 | 280 /* |
| 281 * Vim .sug file format: <SUGHEADER> | |
| 282 * <SUGWORDTREE> | |
| 283 * <SUGTABLE> | |
| 284 * | |
| 285 * <SUGHEADER>: <fileID> <versionnr> <timestamp> | |
| 286 * | |
| 287 * <fileID> 6 bytes "VIMsug" | |
| 288 * <versionnr> 1 byte VIMSUGVERSION | |
| 289 * <timestamp> 8 bytes timestamp that must match with .spl file | |
| 290 * | |
| 291 * | |
| 292 * <SUGWORDTREE>: <wordtree> (see above, no flags or region used) | |
| 293 * | |
| 294 * | |
| 295 * <SUGTABLE>: <sugwcount> <sugline> ... | |
| 296 * | |
| 297 * <sugwcount> 4 bytes number of <sugline> following | |
| 298 * | |
| 299 * <sugline>: <sugnr> ... NUL | |
| 300 * | |
| 301 * <sugnr>: X bytes word number that results in this soundfolded word, | |
| 302 * stored as an offset to the previous number in as | |
| 303 * few bytes as possible, see offset2bytes()) | |
| 304 */ | |
| 305 | |
| 223 | 306 #include "vim.h" |
| 307 | |
| 737 | 308 #if defined(FEAT_SPELL) || defined(PROTO) |
| 223 | 309 |
| 625 | 310 #ifndef UNIX /* it's in os_unix.h for Unix */ |
| 311 # include <time.h> /* for time_t */ | |
| 312 #endif | |
| 313 | |
| 6606 | 314 #define MAXWLEN 254 /* Assume max. word len is this many bytes. |
| 323 | 315 Some places assume a word length fits in a |
| 6606 | 316 byte, thus it can't be above 255. |
| 317 Must be >= PFD_NOTSPECIAL. */ | |
| 226 | 318 |
| 490 | 319 /* Type used for indexes in the word tree need to be at least 4 bytes. If int |
| 324 | 320 * is 8 bytes we could use something smaller, but what? */ |
| 5684 | 321 #if VIM_SIZEOF_INT > 3 |
| 324 | 322 typedef int idx_T; |
| 323 #else | |
| 324 typedef long idx_T; | |
| 325 #endif | |
| 326 | |
| 2660 | 327 #ifdef VMS |
| 328 # define SPL_FNAME_TMPL "%s_%s.spl" | |
| 329 # define SPL_FNAME_ADD "_add." | |
| 330 # define SPL_FNAME_ASCII "_ascii." | |
| 331 #else | |
| 332 # define SPL_FNAME_TMPL "%s.%s.spl" | |
| 333 # define SPL_FNAME_ADD ".add." | |
| 334 # define SPL_FNAME_ASCII ".ascii." | |
| 335 #endif | |
| 336 | |
| 324 | 337 /* Flags used for a word. Only the lowest byte can be used, the region byte |
| 338 * comes above it. */ | |
| 300 | 339 #define WF_REGION 0x01 /* region byte follows */ |
| 340 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */ | |
| 341 #define WF_ALLCAP 0x04 /* word must be all capitals */ | |
| 342 #define WF_RARE 0x08 /* rare word */ | |
| 307 | 343 #define WF_BANNED 0x10 /* bad word */ |
| 481 | 344 #define WF_AFX 0x20 /* affix ID follows */ |
| 389 | 345 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */ |
| 323 | 346 #define WF_KEEPCAP 0x80 /* keep-case word */ |
| 347 | |
| 390 | 348 /* for <flags2>, shifted up one byte to be used in wn_flags */ |
| 349 #define WF_HAS_AFF 0x0100 /* word includes affix */ | |
| 500 | 350 #define WF_NEEDCOMP 0x0200 /* word only valid in compound */ |
| 698 | 351 #define WF_NOSUGGEST 0x0400 /* word not to be suggested */ |
| 809 | 352 #define WF_COMPROOT 0x0800 /* already compounded word, COMPOUNDROOT */ |
| 819 | 353 #define WF_NOCOMPBEF 0x1000 /* no compounding before this word */ |
| 354 #define WF_NOCOMPAFT 0x2000 /* no compounding after this word */ | |
| 390 | 355 |
| 633 | 356 /* only used for su_badflags */ |
| 357 #define WF_MIXCAP 0x20 /* mix of upper and lower case: macaRONI */ | |
| 358 | |
| 389 | 359 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP) |
| 300 | 360 |
| 455 | 361 /* flags for <pflags> */ |
| 822 | 362 #define WFP_RARE 0x01 /* rare prefix */ |
| 363 #define WFP_NC 0x02 /* prefix is not combining */ | |
| 364 #define WFP_UP 0x04 /* to-upper prefix */ | |
| 365 #define WFP_COMPPERMIT 0x08 /* prefix with COMPOUNDPERMITFLAG */ | |
| 366 #define WFP_COMPFORBID 0x10 /* prefix with COMPOUNDFORBIDFLAG */ | |
| 367 | |
| 368 /* Flags for postponed prefixes in "sl_pidxs". Must be above affixID (one | |
| 369 * byte) and prefcondnr (two bytes). */ | |
| 370 #define WF_RAREPFX (WFP_RARE << 24) /* rare postponed prefix */ | |
| 371 #define WF_PFX_NC (WFP_NC << 24) /* non-combining postponed prefix */ | |
| 372 #define WF_PFX_UP (WFP_UP << 24) /* to-upper postponed prefix */ | |
| 373 #define WF_PFX_COMPPERMIT (WFP_COMPPERMIT << 24) /* postponed prefix with | |
| 374 * COMPOUNDPERMITFLAG */ | |
| 375 #define WF_PFX_COMPFORBID (WFP_COMPFORBID << 24) /* postponed prefix with | |
| 376 * COMPOUNDFORBIDFLAG */ | |
| 377 | |
| 390 | 378 |
| 809 | 379 /* flags for <compoptions> */ |
| 380 #define COMP_CHECKDUP 1 /* CHECKCOMPOUNDDUP */ | |
| 381 #define COMP_CHECKREP 2 /* CHECKCOMPOUNDREP */ | |
| 382 #define COMP_CHECKCASE 4 /* CHECKCOMPOUNDCASE */ | |
| 383 #define COMP_CHECKTRIPLE 8 /* CHECKCOMPOUNDTRIPLE */ | |
| 384 | |
| 390 | 385 /* Special byte values for <byte>. Some are only used in the tree for |
| 386 * postponed prefixes, some only in the other trees. This is a bit messy... */ | |
| 387 #define BY_NOFLAGS 0 /* end of word without flags or region; for | |
| 455 | 388 * postponed prefix: no <pflags> */ |
| 389 #define BY_INDEX 1 /* child is shared, index follows */ | |
| 390 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for | |
| 391 * postponed prefix: <pflags> follows */ | |
| 392 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes | |
| 393 * follow; never used in prefix tree */ | |
| 394 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */ | |
| 236 | 395 |
| 625 | 396 /* Info from "REP", "REPSAL" and "SAL" entries in ".aff" file used in si_rep, |
| 397 * si_repsal, sl_rep, and si_sal. Not for sl_sal! | |
| 323 | 398 * One replacement: from "ft_from" to "ft_to". */ |
| 399 typedef struct fromto_S | |
| 236 | 400 { |
| 323 | 401 char_u *ft_from; |
| 402 char_u *ft_to; | |
| 403 } fromto_T; | |
| 236 | 404 |
| 344 | 405 /* Info from "SAL" entries in ".aff" file used in sl_sal. |
| 406 * The info is split for quick processing by spell_soundfold(). | |
| 407 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */ | |
| 408 typedef struct salitem_S | |
| 409 { | |
| 410 char_u *sm_lead; /* leading letters */ | |
| 411 int sm_leadlen; /* length of "sm_lead" */ | |
| 375 | 412 char_u *sm_oneof; /* letters from () or NULL */ |
| 344 | 413 char_u *sm_rules; /* rules like ^, $, priority */ |
| 414 char_u *sm_to; /* replacement. */ | |
| 372 | 415 #ifdef FEAT_MBYTE |
| 416 int *sm_lead_w; /* wide character copy of "sm_lead" */ | |
| 375 | 417 int *sm_oneof_w; /* wide character copy of "sm_oneof" */ |
| 372 | 418 int *sm_to_w; /* wide character copy of "sm_to" */ |
| 419 #endif | |
| 344 | 420 } salitem_T; |
| 421 | |
| 375 | 422 #ifdef FEAT_MBYTE |
| 423 typedef int salfirst_T; | |
| 424 #else | |
| 425 typedef short salfirst_T; | |
| 426 #endif | |
| 427 | |
| 483 | 428 /* Values for SP_*ERROR are negative, positive values are used by |
| 429 * read_cnt_string(). */ | |
| 430 #define SP_TRUNCERROR -1 /* spell file truncated error */ | |
| 431 #define SP_FORMERROR -2 /* format error in spell file */ | |
| 495 | 432 #define SP_OTHERERROR -3 /* other error while reading spell file */ |
| 483 | 433 |
| 236 | 434 /* |
| 243 | 435 * Structure used to store words and other info for one language, loaded from |
| 436 * a .spl file. | |
| 300 | 437 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the |
| 438 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words. | |
| 439 * | |
| 440 * The "byts" array stores the possible bytes in each tree node, preceded by | |
| 441 * the number of possible bytes, sorted on byte value: | |
| 442 * <len> <byte1> <byte2> ... | |
| 443 * The "idxs" array stores the index of the child node corresponding to the | |
| 444 * byte in "byts". | |
| 445 * Exception: when the byte is zero, the word may end here and "idxs" holds | |
| 481 | 446 * the flags, region mask and affixID for the word. There may be several |
| 447 * zeros in sequence for alternative flag/region/affixID combinations. | |
| 236 | 448 */ |
| 449 typedef struct slang_S slang_T; | |
| 450 struct slang_S | |
| 451 { | |
| 452 slang_T *sl_next; /* next language */ | |
| 453 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */ | |
| 310 | 454 char_u *sl_fname; /* name of .spl file */ |
| 323 | 455 int sl_add; /* TRUE if it's a .add file. */ |
| 339 | 456 |
| 300 | 457 char_u *sl_fbyts; /* case-folded word bytes */ |
| 324 | 458 idx_T *sl_fidxs; /* case-folded word indexes */ |
| 300 | 459 char_u *sl_kbyts; /* keep-case word bytes */ |
| 324 | 460 idx_T *sl_kidxs; /* keep-case word indexes */ |
| 339 | 461 char_u *sl_pbyts; /* prefix tree word bytes */ |
| 462 idx_T *sl_pidxs; /* prefix tree word indexes */ | |
| 463 | |
| 714 | 464 char_u *sl_info; /* infotext string or NULL */ |
| 465 | |
| 236 | 466 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */ |
| 323 | 467 |
| 376 | 468 char_u *sl_midword; /* MIDWORD string or NULL */ |
| 469 | |
| 625 | 470 hashtab_T sl_wordcount; /* hashtable with word count, wordcount_T */ |
| 471 | |
| 809 | 472 int sl_compmax; /* COMPOUNDWORDMAX (default: MAXWLEN) */ |
| 501 | 473 int sl_compminlen; /* COMPOUNDMIN (default: 0) */ |
| 483 | 474 int sl_compsylmax; /* COMPOUNDSYLMAX (default: MAXWLEN) */ |
| 809 | 475 int sl_compoptions; /* COMP_* flags */ |
| 476 garray_T sl_comppat; /* CHECKCOMPOUNDPATTERN items */ | |
| 714 | 477 regprog_T *sl_compprog; /* COMPOUNDRULE turned into a regexp progrm |
| 483 | 478 * (NULL when no compounding) */ |
| 1762 | 479 char_u *sl_comprules; /* all COMPOUNDRULE concatenated (or NULL) */ |
| 483 | 480 char_u *sl_compstartflags; /* flags for first compound word */ |
| 485 | 481 char_u *sl_compallflags; /* all flags for compound words */ |
| 492 | 482 char_u sl_nobreak; /* When TRUE: no spaces between words */ |
| 483 | 483 char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */ |
| 484 garray_T sl_syl_items; /* syllable items */ | |
| 481 | 485 |
| 339 | 486 int sl_prefixcnt; /* number of items in "sl_prefprog" */ |
| 487 regprog_T **sl_prefprog; /* table with regprogs for prefixes */ | |
| 488 | |
| 323 | 489 garray_T sl_rep; /* list of fromto_T entries from REP lines */ |
| 490 short sl_rep_first[256]; /* indexes where byte first appears, -1 if | |
| 491 there is none */ | |
| 344 | 492 garray_T sl_sal; /* list of salitem_T entries from SAL lines */ |
| 375 | 493 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if |
| 323 | 494 there is none */ |
| 625 | 495 int sl_followup; /* SAL followup */ |
| 496 int sl_collapse; /* SAL collapse_result */ | |
| 497 int sl_rem_accents; /* SAL remove_accents */ | |
| 375 | 498 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items: |
| 499 * "sl_sal_first" maps chars, when has_mbyte | |
| 500 * "sl_sal" is a list of wide char lists. */ | |
| 625 | 501 garray_T sl_repsal; /* list of fromto_T entries from REPSAL lines */ |
| 502 short sl_repsal_first[256]; /* sl_rep_first for REPSAL lines */ | |
| 698 | 503 int sl_nosplitsugs; /* don't suggest splitting a word */ |
| 625 | 504 |
| 505 /* Info from the .sug file. Loaded on demand. */ | |
| 506 time_t sl_sugtime; /* timestamp for .sug file */ | |
| 507 char_u *sl_sbyts; /* soundfolded word bytes */ | |
| 508 idx_T *sl_sidxs; /* soundfolded word indexes */ | |
| 509 buf_T *sl_sugbuf; /* buffer with word number table */ | |
| 510 int sl_sugloaded; /* TRUE when .sug file was loaded or failed to | |
| 511 load */ | |
| 512 | |
| 330 | 513 int sl_has_map; /* TRUE if there is a MAP line */ |
| 514 #ifdef FEAT_MBYTE | |
| 515 hashtab_T sl_map_hash; /* MAP for multi-byte chars */ | |
| 516 int sl_map_array[256]; /* MAP for first 256 chars */ | |
| 517 #else | |
| 518 char_u sl_map_array[256]; /* MAP for first 256 chars */ | |
| 519 #endif | |
| 625 | 520 hashtab_T sl_sounddone; /* table with soundfolded words that have |
| 521 handled, see add_sound_suggest() */ | |
| 236 | 522 }; |
| 523 | |
| 243 | 524 /* First language that is loaded, start of the linked list of loaded |
| 525 * languages. */ | |
| 236 | 526 static slang_T *first_lang = NULL; |
| 527 | |
| 323 | 528 /* Flags used in .spl file for soundsalike flags. */ |
| 529 #define SAL_F0LLOWUP 1 | |
| 530 #define SAL_COLLAPSE 2 | |
| 531 #define SAL_REM_ACCENTS 4 | |
| 532 | |
| 236 | 533 /* |
| 534 * Structure used in "b_langp", filled from 'spelllang'. | |
| 535 */ | |
| 536 typedef struct langp_S | |
| 537 { | |
| 503 | 538 slang_T *lp_slang; /* info for this language */ |
| 539 slang_T *lp_sallang; /* language used for sound folding or NULL */ | |
| 540 slang_T *lp_replang; /* language used for REP items or NULL */ | |
| 236 | 541 int lp_region; /* bitmask for region or REGION_ALL */ |
| 542 } langp_T; | |
| 543 | |
| 544 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i)) | |
| 545 | |
| 307 | 546 #define REGION_ALL 0xff /* word valid in all regions */ |
| 547 | |
| 483 | 548 #define VIMSPELLMAGIC "VIMspell" /* string at start of Vim spell file */ |
| 549 #define VIMSPELLMAGICL 8 | |
| 550 #define VIMSPELLVERSION 50 | |
| 551 | |
| 625 | 552 #define VIMSUGMAGIC "VIMsug" /* string at start of Vim .sug file */ |
| 553 #define VIMSUGMAGICL 6 | |
| 554 #define VIMSUGVERSION 1 | |
| 555 | |
| 483 | 556 /* Section IDs. Only renumber them when VIMSPELLVERSION changes! */ |
| 557 #define SN_REGION 0 /* <regionname> section */ | |
| 558 #define SN_CHARFLAGS 1 /* charflags section */ | |
| 559 #define SN_MIDWORD 2 /* <midword> section */ | |
| 560 #define SN_PREFCOND 3 /* <prefcond> section */ | |
| 561 #define SN_REP 4 /* REP items section */ | |
| 562 #define SN_SAL 5 /* SAL items section */ | |
| 563 #define SN_SOFO 6 /* soundfolding section */ | |
| 564 #define SN_MAP 7 /* MAP items section */ | |
| 565 #define SN_COMPOUND 8 /* compound words section */ | |
| 566 #define SN_SYLLABLE 9 /* syllable section */ | |
| 492 | 567 #define SN_NOBREAK 10 /* NOBREAK section */ |
| 625 | 568 #define SN_SUGFILE 11 /* timestamp for .sug file */ |
| 569 #define SN_REPSAL 12 /* REPSAL items section */ | |
| 570 #define SN_WORDS 13 /* common words */ | |
| 698 | 571 #define SN_NOSPLITSUGS 14 /* don't split word for suggestions */ |
| 714 | 572 #define SN_INFO 15 /* info section */ |
| 483 | 573 #define SN_END 255 /* end of sections */ |
| 574 | |
| 575 #define SNF_REQUIRED 1 /* <sectionflags>: required section */ | |
| 576 | |
| 307 | 577 /* Result values. Lower number is accepted over higher one. */ |
| 578 #define SP_BANNED -1 | |
| 236 | 579 #define SP_OK 0 |
| 307 | 580 #define SP_RARE 1 |
| 581 #define SP_LOCAL 2 | |
| 582 #define SP_BAD 3 | |
| 236 | 583 |
| 381 | 584 /* file used for "zG" and "zW" */ |
| 385 | 585 static char_u *int_wordlist = NULL; |
| 381 | 586 |
| 625 | 587 typedef struct wordcount_S |
| 588 { | |
| 589 short_u wc_count; /* nr of times word was seen */ | |
| 590 char_u wc_word[1]; /* word, actually longer */ | |
| 591 } wordcount_T; | |
| 592 | |
| 593 static wordcount_T dumwc; | |
| 835 | 594 #define WC_KEY_OFF (unsigned)(dumwc.wc_word - (char_u *)&dumwc) |
| 625 | 595 #define HI2WC(hi) ((wordcount_T *)((hi)->hi_key - WC_KEY_OFF)) |
| 596 #define MAXWORDCOUNT 0xffff | |
| 597 | |
| 236 | 598 /* |
| 323 | 599 * Information used when looking for suggestions. |
| 600 */ | |
| 601 typedef struct suginfo_S | |
| 602 { | |
| 603 garray_T su_ga; /* suggestions, contains "suggest_T" */ | |
| 344 | 604 int su_maxcount; /* max. number of suggestions displayed */ |
| 323 | 605 int su_maxscore; /* maximum score for adding to su_ga */ |
| 625 | 606 int su_sfmaxscore; /* idem, for when doing soundfold words */ |
| 344 | 607 garray_T su_sga; /* like su_ga, sound-folded scoring */ |
| 323 | 608 char_u *su_badptr; /* start of bad word in line */ |
| 609 int su_badlen; /* length of detected bad word in line */ | |
| 346 | 610 int su_badflags; /* caps flags for bad word */ |
| 323 | 611 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */ |
| 612 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */ | |
| 534 | 613 char_u su_sal_badword[MAXWLEN]; /* su_badword soundfolded */ |
| 323 | 614 hashtab_T su_banned; /* table with banned words */ |
| 503 | 615 slang_T *su_sallang; /* default language for sound folding */ |
| 323 | 616 } suginfo_T; |
| 617 | |
| 618 /* One word suggestion. Used in "si_ga". */ | |
| 619 typedef struct suggest_S | |
| 620 { | |
| 621 char_u *st_word; /* suggested word, allocated string */ | |
| 625 | 622 int st_wordlen; /* STRLEN(st_word) */ |
| 323 | 623 int st_orglen; /* length of replaced text */ |
| 624 int st_score; /* lower is better */ | |
| 344 | 625 int st_altscore; /* used when st_score compares equal */ |
| 626 int st_salscore; /* st_score is for soundalike */ | |
| 324 | 627 int st_had_bonus; /* bonus already included in score */ |
| 503 | 628 slang_T *st_slang; /* language used for sound folding */ |
| 323 | 629 } suggest_T; |
| 630 | |
| 344 | 631 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i]) |
| 323 | 632 |
| 625 | 633 /* TRUE if a word appears in the list of banned words. */ |
| 634 #define WAS_BANNED(su, word) (!HASHITEM_EMPTY(hash_find(&su->su_banned, word))) | |
| 635 | |
| 1705 | 636 /* Number of suggestions kept when cleaning up. We need to keep more than |
| 625 | 637 * what is displayed, because when rescore_suggestions() is called the score |
| 638 * may change and wrong suggestions may be removed later. */ | |
| 639 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 130 ? 150 : (su)->su_maxcount + 20) | |
| 324 | 640 |
| 641 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots | |
| 642 * of suggestions that are not going to be displayed. */ | |
| 625 | 643 #define SUG_MAX_COUNT(su) (SUG_CLEAN_COUNT(su) + 50) |
| 323 | 644 |
| 645 /* score for various changes */ | |
| 344 | 646 #define SCORE_SPLIT 149 /* split bad word */ |
| 698 | 647 #define SCORE_SPLIT_NO 249 /* split bad word with NOSPLITSUGS */ |
| 323 | 648 #define SCORE_ICASE 52 /* slightly different case */ |
| 483 | 649 #define SCORE_REGION 200 /* word is for different region */ |
| 323 | 650 #define SCORE_RARE 180 /* rare word */ |
| 625 | 651 #define SCORE_SWAP 75 /* swap two characters */ |
| 323 | 652 #define SCORE_SWAP3 110 /* swap two characters in three */ |
| 532 | 653 #define SCORE_REP 65 /* REP replacement */ |
| 323 | 654 #define SCORE_SUBST 93 /* substitute a character */ |
| 655 #define SCORE_SIMILAR 33 /* substitute a similar character */ | |
| 479 | 656 #define SCORE_SUBCOMP 33 /* substitute a composing character */ |
| 324 | 657 #define SCORE_DEL 94 /* delete a character */ |
| 532 | 658 #define SCORE_DELDUP 66 /* delete a duplicated character */ |
| 479 | 659 #define SCORE_DELCOMP 28 /* delete a composing character */ |
| 324 | 660 #define SCORE_INS 96 /* insert a character */ |
| 532 | 661 #define SCORE_INSDUP 67 /* insert a duplicate character */ |
| 475 | 662 #define SCORE_INSCOMP 30 /* insert a composing character */ |
| 366 | 663 #define SCORE_NONWORD 103 /* change non-word to word char */ |
| 323 | 664 |
| 372 | 665 #define SCORE_FILE 30 /* suggestion from a file */ |
| 323 | 666 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower. |
| 667 * 350 allows for about three changes. */ | |
| 344 | 668 |
| 625 | 669 #define SCORE_COMMON1 30 /* subtracted for words seen before */ |
| 670 #define SCORE_COMMON2 40 /* subtracted for words often seen */ | |
| 671 #define SCORE_COMMON3 50 /* subtracted for words very often seen */ | |
| 672 #define SCORE_THRES2 10 /* word count threshold for COMMON2 */ | |
| 673 #define SCORE_THRES3 100 /* word count threshold for COMMON3 */ | |
| 674 | |
| 675 /* When trying changed soundfold words it becomes slow when trying more than | |
| 676 * two changes. With less then two changes it's slightly faster but we miss a | |
| 677 * few good suggestions. In rare cases we need to try three of four changes. | |
| 678 */ | |
| 679 #define SCORE_SFMAX1 200 /* maximum score for first try */ | |
| 680 #define SCORE_SFMAX2 300 /* maximum score for second try */ | |
| 681 #define SCORE_SFMAX3 400 /* maximum score for third try */ | |
| 682 | |
| 344 | 683 #define SCORE_BIG SCORE_INS * 3 /* big difference */ |
| 625 | 684 #define SCORE_MAXMAX 999999 /* accept any score */ |
| 685 #define SCORE_LIMITMAX 350 /* for spell_edit_score_limit() */ | |
| 686 | |
| 687 /* for spell_edit_score_limit() we need to know the minimum value of | |
| 688 * SCORE_ICASE, SCORE_SWAP, SCORE_DEL, SCORE_SIMILAR and SCORE_INS */ | |
| 689 #define SCORE_EDIT_MIN SCORE_SIMILAR | |
| 323 | 690 |
| 691 /* | |
| 236 | 692 * Structure to store info for word matching. |
| 693 */ | |
| 694 typedef struct matchinf_S | |
| 695 { | |
| 696 langp_T *mi_lp; /* info for language and region */ | |
| 243 | 697 |
| 698 /* pointers to original text to be checked */ | |
| 236 | 699 char_u *mi_word; /* start of word being checked */ |
| 339 | 700 char_u *mi_end; /* end of matching word so far */ |
| 243 | 701 char_u *mi_fend; /* next char to be added to mi_fword */ |
| 300 | 702 char_u *mi_cend; /* char after what was used for |
| 703 mi_capflags */ | |
| 243 | 704 |
| 705 /* case-folded text */ | |
| 706 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */ | |
| 300 | 707 int mi_fwordlen; /* nr of valid bytes in mi_fword */ |
| 243 | 708 |
| 339 | 709 /* for when checking word after a prefix */ |
| 710 int mi_prefarridx; /* index in sl_pidxs with list of | |
| 481 | 711 affixID/condition */ |
| 339 | 712 int mi_prefcnt; /* number of entries at mi_prefarridx */ |
| 713 int mi_prefixlen; /* byte length of prefix */ | |
| 455 | 714 #ifdef FEAT_MBYTE |
| 715 int mi_cprefixlen; /* byte length of prefix in original | |
| 716 case */ | |
| 717 #else | |
| 718 # define mi_cprefixlen mi_prefixlen /* it's the same value */ | |
| 719 #endif | |
| 339 | 720 |
| 481 | 721 /* for when checking a compound word */ |
| 722 int mi_compoff; /* start of following word offset */ | |
| 483 | 723 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */ |
| 724 int mi_complen; /* nr of compound words used */ | |
| 809 | 725 int mi_compextra; /* nr of COMPOUNDROOT words */ |
| 481 | 726 |
| 243 | 727 /* others */ |
| 236 | 728 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */ |
| 300 | 729 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
730 win_T *mi_win; /* buffer being checked */ |
| 492 | 731 |
| 732 /* for NOBREAK */ | |
| 733 int mi_result2; /* "mi_resul" without following word */ | |
| 734 char_u *mi_end2; /* "mi_end" without following word */ | |
| 236 | 735 } matchinf_T; |
| 736 | |
| 307 | 737 /* |
| 738 * The tables used for recognizing word characters according to spelling. | |
| 739 * These are only used for the first 256 characters of 'encoding'. | |
| 740 */ | |
| 741 typedef struct spelltab_S | |
| 742 { | |
| 743 char_u st_isw[256]; /* flags: is word char */ | |
| 744 char_u st_isu[256]; /* flags: is uppercase char */ | |
| 745 char_u st_fold[256]; /* chars: folded case */ | |
| 324 | 746 char_u st_upper[256]; /* chars: upper case */ |
| 307 | 747 } spelltab_T; |
| 748 | |
| 749 static spelltab_T spelltab; | |
| 750 static int did_set_spelltab; | |
| 751 | |
| 324 | 752 #define CF_WORD 0x01 |
| 753 #define CF_UPPER 0x02 | |
| 307 | 754 |
| 755 static void clear_spell_chartab __ARGS((spelltab_T *sp)); | |
| 756 static int set_spell_finish __ARGS((spelltab_T *new_st)); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
757 static int spell_iswordp __ARGS((char_u *p, win_T *wp)); |
| 5477 | 758 static int spell_iswordp_nmw __ARGS((char_u *p, win_T *wp)); |
| 759 #ifdef FEAT_MBYTE | |
| 760 static int spell_mb_isword_class __ARGS((int cl, win_T *wp)); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
761 static int spell_iswordp_w __ARGS((int *p, win_T *wp)); |
| 376 | 762 #endif |
| 483 | 763 static int write_spell_prefcond __ARGS((FILE *fd, garray_T *gap)); |
| 307 | 764 |
| 765 /* | |
| 344 | 766 * For finding suggestions: At each node in the tree these states are tried: |
| 330 | 767 */ |
| 768 typedef enum | |
| 769 { | |
| 344 | 770 STATE_START = 0, /* At start of node check for NUL bytes (goodword |
| 771 * ends); if badword ends there is a match, otherwise | |
| 772 * try splitting word. */ | |
| 375 | 773 STATE_NOPREFIX, /* try without prefix */ |
| 344 | 774 STATE_SPLITUNDO, /* Undo splitting. */ |
| 330 | 775 STATE_ENDNUL, /* Past NUL bytes at start of the node. */ |
| 776 STATE_PLAIN, /* Use each byte of the node. */ | |
| 777 STATE_DEL, /* Delete a byte from the bad word. */ | |
| 625 | 778 STATE_INS_PREP, /* Prepare for inserting bytes. */ |
| 330 | 779 STATE_INS, /* Insert a byte in the bad word. */ |
| 780 STATE_SWAP, /* Swap two bytes. */ | |
| 344 | 781 STATE_UNSWAP, /* Undo swap two characters. */ |
| 782 STATE_SWAP3, /* Swap two characters over three. */ | |
| 783 STATE_UNSWAP3, /* Undo Swap two characters over three. */ | |
| 784 STATE_UNROT3L, /* Undo rotate three characters left */ | |
| 785 STATE_UNROT3R, /* Undo rotate three characters right */ | |
| 330 | 786 STATE_REP_INI, /* Prepare for using REP items. */ |
| 787 STATE_REP, /* Use matching REP items from the .aff file. */ | |
| 788 STATE_REP_UNDO, /* Undo a REP item replacement. */ | |
| 789 STATE_FINAL /* End of this node. */ | |
| 790 } state_T; | |
| 791 | |
| 792 /* | |
| 346 | 793 * Struct to keep the state at each level in suggest_try_change(). |
| 323 | 794 */ |
| 795 typedef struct trystate_S | |
| 796 { | |
| 330 | 797 state_T ts_state; /* state at this level, STATE_ */ |
| 323 | 798 int ts_score; /* score */ |
| 344 | 799 idx_T ts_arridx; /* index in tree array, start of node */ |
| 330 | 800 short ts_curi; /* index in list of child nodes */ |
| 801 char_u ts_fidx; /* index in fword[], case-folded bad word */ | |
| 802 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */ | |
| 803 char_u ts_twordlen; /* valid length of tword[] */ | |
| 482 | 804 char_u ts_prefixdepth; /* stack depth for end of prefix or |
| 485 | 805 * PFD_PREFIXTREE or PFD_NOPREFIX */ |
| 806 char_u ts_flags; /* TSF_ flags */ | |
| 330 | 807 #ifdef FEAT_MBYTE |
| 808 char_u ts_tcharlen; /* number of bytes in tword character */ | |
| 809 char_u ts_tcharidx; /* current byte index in tword character */ | |
| 810 char_u ts_isdiff; /* DIFF_ values */ | |
| 811 char_u ts_fcharstart; /* index in fword where badword char started */ | |
| 812 #endif | |
| 483 | 813 char_u ts_prewordlen; /* length of word in "preword[]" */ |
| 814 char_u ts_splitoff; /* index in "tword" after last split */ | |
| 492 | 815 char_u ts_splitfidx; /* "ts_fidx" at word split */ |
| 483 | 816 char_u ts_complen; /* nr of compound words used */ |
| 485 | 817 char_u ts_compsplit; /* index for "compflags" where word was spit */ |
| 346 | 818 char_u ts_save_badflags; /* su_badflags saved here */ |
| 625 | 819 char_u ts_delidx; /* index in fword for char that was deleted, |
| 820 valid when "ts_flags" has TSF_DIDDEL */ | |
| 323 | 821 } trystate_T; |
| 822 | |
| 330 | 823 /* values for ts_isdiff */ |
| 824 #define DIFF_NONE 0 /* no different byte (yet) */ | |
| 825 #define DIFF_YES 1 /* different byte found */ | |
| 826 #define DIFF_INSERT 2 /* inserting character */ | |
| 827 | |
| 485 | 828 /* values for ts_flags */ |
| 829 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */ | |
| 830 #define TSF_DIDSPLIT 2 /* tried split at this point */ | |
| 625 | 831 #define TSF_DIDDEL 4 /* did a delete, "ts_delidx" has index */ |
| 485 | 832 |
| 375 | 833 /* special values ts_prefixdepth */ |
| 485 | 834 #define PFD_NOPREFIX 0xff /* not using prefixes */ |
| 482 | 835 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */ |
| 625 | 836 #define PFD_NOTSPECIAL 0xfd /* highest value that's not special */ |
| 375 | 837 |
| 339 | 838 /* mode values for find_word */ |
| 481 | 839 #define FIND_FOLDWORD 0 /* find word case-folded */ |
| 840 #define FIND_KEEPWORD 1 /* find keep-case word */ | |
| 841 #define FIND_PREFIX 2 /* find word after prefix */ | |
| 842 #define FIND_COMPOUND 3 /* find case-folded compound word */ | |
| 843 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */ | |
| 339 | 844 |
| 236 | 845 static slang_T *slang_alloc __ARGS((char_u *lang)); |
| 846 static void slang_free __ARGS((slang_T *lp)); | |
| 310 | 847 static void slang_clear __ARGS((slang_T *lp)); |
| 625 | 848 static void slang_clear_sug __ARGS((slang_T *lp)); |
| 339 | 849 static void find_word __ARGS((matchinf_T *mip, int mode)); |
| 1762 | 850 static int match_checkcompoundpattern __ARGS((char_u *ptr, int wlen, garray_T *gap)); |
| 483 | 851 static int can_compound __ARGS((slang_T *slang, char_u *word, char_u *flags)); |
| 1762 | 852 static int can_be_compound __ARGS((trystate_T *sp, slang_T *slang, char_u *compflags, int flag)); |
| 853 static int match_compoundrule __ARGS((slang_T *slang, char_u *compflags)); | |
| 455 | 854 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req)); |
| 485 | 855 static void find_prefix __ARGS((matchinf_T *mip, int mode)); |
| 339 | 856 static int fold_more __ARGS((matchinf_T *mip)); |
| 389 | 857 static int spell_valid_case __ARGS((int wordflags, int treeflags)); |
| 498 | 858 static int no_spell_checking __ARGS((win_T *wp)); |
| 307 | 859 static void spell_load_lang __ARGS((char_u *lang)); |
| 310 | 860 static char_u *spell_enc __ARGS((void)); |
| 385 | 861 static void int_wordlist_spl __ARGS((char_u *fname)); |
| 310 | 862 static void spell_load_cb __ARGS((char_u *fname, void *cookie)); |
| 323 | 863 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent)); |
| 389 | 864 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp)); |
| 483 | 865 static int read_region_section __ARGS((FILE *fd, slang_T *slang, int len)); |
| 866 static int read_charflags_section __ARGS((FILE *fd)); | |
| 867 static int read_prefcond_section __ARGS((FILE *fd, slang_T *lp)); | |
| 625 | 868 static int read_rep_section __ARGS((FILE *fd, garray_T *gap, short *first)); |
| 483 | 869 static int read_sal_section __ARGS((FILE *fd, slang_T *slang)); |
| 625 | 870 static int read_words_section __ARGS((FILE *fd, slang_T *lp, int len)); |
| 871 static void count_common_word __ARGS((slang_T *lp, char_u *word, int len, int count)); | |
| 872 static int score_wordcount_adj __ARGS((slang_T *slang, int score, char_u *word, int split)); | |
| 483 | 873 static int read_sofo_section __ARGS((FILE *fd, slang_T *slang)); |
| 874 static int read_compound __ARGS((FILE *fd, slang_T *slang, int len)); | |
| 495 | 875 static int byte_in_str __ARGS((char_u *str, int byte)); |
| 483 | 876 static int init_syl_tab __ARGS((slang_T *slang)); |
| 877 static int count_syllables __ARGS((slang_T *slang, char_u *word)); | |
| 381 | 878 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to)); |
| 879 static void set_sal_first __ARGS((slang_T *lp)); | |
| 372 | 880 #ifdef FEAT_MBYTE |
| 881 static int *mb_str2wide __ARGS((char_u *s)); | |
| 882 #endif | |
| 625 | 883 static int spell_read_tree __ARGS((FILE *fd, char_u **bytsp, idx_T **idxsp, int prefixtree, int prefixcnt)); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
884 static idx_T read_tree_node __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, idx_T startidx, int prefixtree, int maxprefcondnr)); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
885 static void clear_midword __ARGS((win_T *buf)); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
886 static void use_midword __ARGS((slang_T *lp, win_T *buf)); |
| 236 | 887 static int find_region __ARGS((char_u *rp, char_u *region)); |
| 888 static int captype __ARGS((char_u *word, char_u *end)); | |
| 474 | 889 static int badword_captype __ARGS((char_u *word, char_u *end)); |
| 323 | 890 static void spell_reload_one __ARGS((char_u *fname, int added_word)); |
| 483 | 891 static void set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp)); |
| 307 | 892 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp)); |
| 893 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen)); | |
| 475 | 894 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col)); |
| 638 | 895 static void spell_find_suggest __ARGS((char_u *badptr, int badlen, suginfo_T *su, int maxcount, int banbadword, int need_cap, int interactive)); |
| 372 | 896 #ifdef FEAT_EVAL |
| 897 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr)); | |
| 898 #endif | |
| 899 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname)); | |
| 625 | 900 static void spell_suggest_intern __ARGS((suginfo_T *su, int interactive)); |
| 901 static void suggest_load_files __ARGS((void)); | |
| 902 static void tree_count_words __ARGS((char_u *byts, idx_T *idxs)); | |
| 344 | 903 static void spell_find_cleanup __ARGS((suginfo_T *su)); |
| 324 | 904 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper)); |
| 344 | 905 static void allcap_copy __ARGS((char_u *word, char_u *wcopy)); |
| 346 | 906 static void suggest_try_special __ARGS((suginfo_T *su)); |
| 907 static void suggest_try_change __ARGS((suginfo_T *su)); | |
| 625 | 908 static void suggest_trie_walk __ARGS((suginfo_T *su, langp_T *lp, char_u *fword, int soundfold)); |
| 909 static void go_deeper __ARGS((trystate_T *stack, int depth, int score_add)); | |
| 455 | 910 #ifdef FEAT_MBYTE |
| 911 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word)); | |
| 912 #endif | |
| 323 | 913 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword)); |
| 344 | 914 static void score_comp_sal __ARGS((suginfo_T *su)); |
| 915 static void score_combine __ARGS((suginfo_T *su)); | |
| 351 | 916 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound)); |
| 625 | 917 static void suggest_try_soundalike_prep __ARGS((void)); |
| 346 | 918 static void suggest_try_soundalike __ARGS((suginfo_T *su)); |
| 625 | 919 static void suggest_try_soundalike_finish __ARGS((void)); |
| 920 static void add_sound_suggest __ARGS((suginfo_T *su, char_u *goodword, int score, langp_T *lp)); | |
| 921 static int soundfold_find __ARGS((slang_T *slang, char_u *word)); | |
| 323 | 922 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags)); |
| 330 | 923 static void set_map_str __ARGS((slang_T *lp, char_u *map)); |
| 323 | 924 static int similar_chars __ARGS((slang_T *slang, int c1, int c2)); |
| 625 | 925 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus, slang_T *slang, int maxsf)); |
| 926 static void check_suggestions __ARGS((suginfo_T *su, garray_T *gap)); | |
| 323 | 927 static void add_banned __ARGS((suginfo_T *su, char_u *word)); |
| 324 | 928 static void rescore_suggestions __ARGS((suginfo_T *su)); |
| 534 | 929 static void rescore_one __ARGS((suginfo_T *su, suggest_T *stp)); |
| 344 | 930 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep)); |
| 375 | 931 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res)); |
| 932 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res)); | |
| 933 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res)); | |
| 372 | 934 #ifdef FEAT_MBYTE |
| 375 | 935 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res)); |
| 372 | 936 #endif |
| 344 | 937 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound)); |
| 625 | 938 static int spell_edit_score __ARGS((slang_T *slang, char_u *badword, char_u *goodword)); |
| 939 static int spell_edit_score_limit __ARGS((slang_T *slang, char_u *badword, char_u *goodword, int limit)); | |
| 940 #ifdef FEAT_MBYTE | |
| 941 static int spell_edit_score_limit_w __ARGS((slang_T *slang, char_u *badword, char_u *goodword, int limit)); | |
| 942 #endif | |
| 701 | 943 static void dump_word __ARGS((slang_T *slang, char_u *word, char_u *pat, int *dir, int round, int flags, linenr_T lnum)); |
| 944 static linenr_T dump_prefixes __ARGS((slang_T *slang, char_u *word, char_u *pat, int *dir, int round, int flags, linenr_T startlnum)); | |
| 625 | 945 static buf_T *open_spellbuf __ARGS((void)); |
| 946 static void close_spellbuf __ARGS((buf_T *buf)); | |
| 323 | 947 |
| 324 | 948 /* |
| 949 * Use our own character-case definitions, because the current locale may | |
| 950 * differ from what the .spl file uses. | |
| 951 * These must not be called with negative number! | |
| 952 */ | |
| 953 #ifndef FEAT_MBYTE | |
| 954 /* Non-multi-byte implementation. */ | |
| 1883 | 955 # define SPELL_TOFOLD(c) ((c) < 256 ? (int)spelltab.st_fold[c] : (c)) |
| 956 # define SPELL_TOUPPER(c) ((c) < 256 ? (int)spelltab.st_upper[c] : (c)) | |
| 324 | 957 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE) |
| 958 #else | |
| 447 | 959 # if defined(HAVE_WCHAR_H) |
| 960 # include <wchar.h> /* for towupper() and towlower() */ | |
| 961 # endif | |
| 324 | 962 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do |
| 963 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use | |
| 964 * the "w" library function for characters above 255 if available. */ | |
| 965 # ifdef HAVE_TOWLOWER | |
| 966 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \ | |
| 1883 | 967 : (c) < 256 ? (int)spelltab.st_fold[c] : (int)towlower(c)) |
| 324 | 968 # else |
| 969 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \ | |
| 1883 | 970 : (c) < 256 ? (int)spelltab.st_fold[c] : (c)) |
| 324 | 971 # endif |
| 972 | |
| 973 # ifdef HAVE_TOWUPPER | |
| 974 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \ | |
| 1883 | 975 : (c) < 256 ? (int)spelltab.st_upper[c] : (int)towupper(c)) |
| 324 | 976 # else |
| 977 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \ | |
| 1883 | 978 : (c) < 256 ? (int)spelltab.st_upper[c] : (c)) |
| 324 | 979 # endif |
| 980 | |
| 981 # ifdef HAVE_ISWUPPER | |
| 982 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \ | |
| 983 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c)) | |
| 984 # else | |
| 985 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \ | |
| 474 | 986 : (c) < 256 ? spelltab.st_isu[c] : (FALSE)) |
| 324 | 987 # endif |
| 988 #endif | |
| 989 | |
| 307 | 990 |
| 991 static char *e_format = N_("E759: Format error in spell file"); | |
| 381 | 992 static char *e_spell_trunc = N_("E758: Truncated spell file"); |
| 482 | 993 static char *e_afftrailing = N_("Trailing text in %s line %d: %s"); |
| 495 | 994 static char *e_affname = N_("Affix name too long in %s line %d: %s"); |
| 995 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP"); | |
| 996 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range"); | |
| 469 | 997 static char *msg_compressing = N_("Compressing word tree..."); |
| 236 | 998 |
| 626 | 999 /* Remember what "z?" replaced. */ |
| 1000 static char_u *repl_from = NULL; | |
| 1001 static char_u *repl_to = NULL; | |
| 1002 | |
| 236 | 1003 /* |
| 1004 * Main spell-checking function. | |
| 300 | 1005 * "ptr" points to a character that could be the start of a word. |
| 534 | 1006 * "*attrp" is set to the highlight index for a badly spelled word. For a |
| 1007 * non-word or when it's OK it remains unchanged. | |
| 236 | 1008 * This must only be called when 'spelllang' is not empty. |
| 323 | 1009 * |
| 385 | 1010 * "capcol" is used to check for a Capitalised word after the end of a |
| 1011 * sentence. If it's zero then perform the check. Return the column where to | |
| 1012 * check next, or -1 when no sentence end was found. If it's NULL then don't | |
| 1013 * worry. | |
| 323 | 1014 * |
| 236 | 1015 * Returns the length of the word in bytes, also when it's OK, so that the |
| 1016 * caller can skip over the word. | |
| 1017 */ | |
| 1018 int | |
| 625 | 1019 spell_check(wp, ptr, attrp, capcol, docount) |
| 236 | 1020 win_T *wp; /* current window */ |
| 1021 char_u *ptr; | |
| 534 | 1022 hlf_T *attrp; |
| 385 | 1023 int *capcol; /* column to check for Capital */ |
| 625 | 1024 int docount; /* count good words */ |
| 236 | 1025 { |
| 1026 matchinf_T mi; /* Most things are put in "mi" so that it can | |
| 1027 be passed to functions quickly. */ | |
| 344 | 1028 int nrlen = 0; /* found a number first */ |
| 385 | 1029 int c; |
| 483 | 1030 int wrongcaplen = 0; |
| 500 | 1031 int lpi; |
| 625 | 1032 int count_word = docount; |
| 236 | 1033 |
| 307 | 1034 /* A word never starts at a space or a control character. Return quickly |
| 1035 * then, skipping over the character. */ | |
| 1036 if (*ptr <= ' ') | |
| 1037 return 1; | |
| 690 | 1038 |
| 1039 /* Return here when loading language files failed. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1040 if (wp->w_s->b_langp.ga_len == 0) |
| 690 | 1041 return 1; |
| 1042 | |
| 483 | 1043 vim_memset(&mi, 0, sizeof(matchinf_T)); |
| 236 | 1044 |
| 344 | 1045 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and |
| 584 | 1046 * 0X99FF. But always do check spelling to find "3GPP" and "11 |
| 1047 * julifeest". */ | |
| 300 | 1048 if (*ptr >= '0' && *ptr <= '9') |
| 1049 { | |
| 316 | 1050 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X')) |
| 1051 mi.mi_end = skiphex(ptr + 2); | |
| 300 | 1052 else |
| 344 | 1053 mi.mi_end = skipdigits(ptr); |
| 835 | 1054 nrlen = (int)(mi.mi_end - ptr); |
| 584 | 1055 } |
| 346 | 1056 |
| 1057 /* Find the normal end of the word (until the next non-word character). */ | |
| 344 | 1058 mi.mi_word = ptr; |
| 584 | 1059 mi.mi_fend = ptr; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1060 if (spell_iswordp(mi.mi_fend, wp)) |
| 344 | 1061 { |
| 1062 do | |
| 1063 { | |
| 307 | 1064 mb_ptr_adv(mi.mi_fend); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1065 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp)); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1066 |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1067 if (capcol != NULL && *capcol == 0 && wp->w_s->b_cap_prog != NULL) |
| 385 | 1068 { |
| 1069 /* Check word starting with capital letter. */ | |
| 455 | 1070 c = PTR2CHAR(ptr); |
| 385 | 1071 if (!SPELL_ISUPPER(c)) |
| 483 | 1072 wrongcaplen = (int)(mi.mi_fend - ptr); |
| 385 | 1073 } |
| 1074 } | |
| 1075 if (capcol != NULL) | |
| 1076 *capcol = -1; | |
| 344 | 1077 |
| 1078 /* We always use the characters up to the next non-word character, | |
| 1079 * also for bad words. */ | |
| 1080 mi.mi_end = mi.mi_fend; | |
| 1081 | |
| 1082 /* Check caps type later. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1083 mi.mi_capflags = 0; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1084 mi.mi_cend = NULL; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1085 mi.mi_win = wp; |
| 344 | 1086 |
| 483 | 1087 /* case-fold the word with one non-word character, so that we can check |
| 1088 * for the word end. */ | |
| 344 | 1089 if (*mi.mi_fend != NUL) |
| 1090 mb_ptr_adv(mi.mi_fend); | |
| 1091 | |
| 1092 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword, | |
| 1093 MAXWLEN + 1); | |
| 835 | 1094 mi.mi_fwordlen = (int)STRLEN(mi.mi_fword); |
| 344 | 1095 |
| 1096 /* The word is bad unless we recognize it. */ | |
| 1097 mi.mi_result = SP_BAD; | |
| 492 | 1098 mi.mi_result2 = SP_BAD; |
| 344 | 1099 |
| 1100 /* | |
| 1101 * Loop over the languages specified in 'spelllang'. | |
| 625 | 1102 * We check them all, because a word may be matched longer in another |
| 1103 * language. | |
| 344 | 1104 */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1105 for (lpi = 0; lpi < wp->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1106 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1107 mi.mi_lp = LANGP_ENTRY(wp->w_s->b_langp, lpi); |
| 500 | 1108 |
| 1109 /* If reloading fails the language is still in the list but everything | |
| 1110 * has been cleared. */ | |
| 1111 if (mi.mi_lp->lp_slang->sl_fidxs == NULL) | |
| 1112 continue; | |
| 1113 | |
| 344 | 1114 /* Check for a matching word in case-folded words. */ |
| 1115 find_word(&mi, FIND_FOLDWORD); | |
| 1116 | |
| 1117 /* Check for a matching word in keep-case words. */ | |
| 1118 find_word(&mi, FIND_KEEPWORD); | |
| 1119 | |
| 1120 /* Check for matching prefixes. */ | |
| 485 | 1121 find_prefix(&mi, FIND_FOLDWORD); |
| 492 | 1122 |
| 1123 /* For a NOBREAK language, may want to use a word without a following | |
| 1124 * word as a backup. */ | |
| 1125 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD | |
| 1126 && mi.mi_result2 != SP_BAD) | |
| 1127 { | |
| 1128 mi.mi_result = mi.mi_result2; | |
| 1129 mi.mi_end = mi.mi_end2; | |
| 1130 } | |
| 625 | 1131 |
| 1132 /* Count the word in the first language where it's found to be OK. */ | |
| 1133 if (count_word && mi.mi_result == SP_OK) | |
| 1134 { | |
| 1135 count_common_word(mi.mi_lp->lp_slang, ptr, | |
| 1136 (int)(mi.mi_end - ptr), 1); | |
| 1137 count_word = FALSE; | |
| 1138 } | |
| 344 | 1139 } |
| 1140 | |
| 1141 if (mi.mi_result != SP_OK) | |
| 1142 { | |
| 346 | 1143 /* If we found a number skip over it. Allows for "42nd". Do flag |
| 1144 * rare and local words, e.g., "3GPP". */ | |
| 344 | 1145 if (nrlen > 0) |
| 346 | 1146 { |
| 1147 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED) | |
| 1148 return nrlen; | |
| 1149 } | |
| 344 | 1150 |
| 1151 /* When we are at a non-word character there is no error, just | |
| 1152 * skip over the character (try looking for a word after it). */ | |
| 5477 | 1153 else if (!spell_iswordp_nmw(ptr, wp)) |
| 243 | 1154 { |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1155 if (capcol != NULL && wp->w_s->b_cap_prog != NULL) |
| 385 | 1156 { |
| 1157 regmatch_T regmatch; | |
| 6375 | 1158 int r; |
| 385 | 1159 |
| 1160 /* Check for end of sentence. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1161 regmatch.regprog = wp->w_s->b_cap_prog; |
| 385 | 1162 regmatch.rm_ic = FALSE; |
| 6375 | 1163 r = vim_regexec(®match, ptr, 0); |
| 1164 wp->w_s->b_cap_prog = regmatch.regprog; | |
| 1165 if (r) | |
| 385 | 1166 *capcol = (int)(regmatch.endp[0] - ptr); |
| 1167 } | |
| 1168 | |
| 344 | 1169 #ifdef FEAT_MBYTE |
| 1170 if (has_mbyte) | |
| 474 | 1171 return (*mb_ptr2len)(ptr); |
| 344 | 1172 #endif |
| 1173 return 1; | |
| 300 | 1174 } |
| 483 | 1175 else if (mi.mi_end == ptr) |
| 1176 /* Always include at least one character. Required for when there | |
| 1177 * is a mixup in "midword". */ | |
| 1178 mb_ptr_adv(mi.mi_end); | |
| 492 | 1179 else if (mi.mi_result == SP_BAD |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1180 && LANGP_ENTRY(wp->w_s->b_langp, 0)->lp_slang->sl_nobreak) |
| 492 | 1181 { |
| 1182 char_u *p, *fp; | |
| 1183 int save_result = mi.mi_result; | |
| 1184 | |
| 1185 /* First language in 'spelllang' is NOBREAK. Find first position | |
| 1186 * at which any word would be valid. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1187 mi.mi_lp = LANGP_ENTRY(wp->w_s->b_langp, 0); |
| 500 | 1188 if (mi.mi_lp->lp_slang->sl_fidxs != NULL) |
| 1189 { | |
| 1190 p = mi.mi_word; | |
| 1191 fp = mi.mi_fword; | |
| 1192 for (;;) | |
| 1193 { | |
| 1194 mb_ptr_adv(p); | |
| 1195 mb_ptr_adv(fp); | |
| 1196 if (p >= mi.mi_end) | |
| 1197 break; | |
| 835 | 1198 mi.mi_compoff = (int)(fp - mi.mi_fword); |
| 500 | 1199 find_word(&mi, FIND_COMPOUND); |
| 1200 if (mi.mi_result != SP_BAD) | |
| 1201 { | |
| 1202 mi.mi_end = p; | |
| 1203 break; | |
| 1204 } | |
| 1205 } | |
| 1206 mi.mi_result = save_result; | |
| 1207 } | |
| 492 | 1208 } |
| 243 | 1209 |
| 344 | 1210 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED) |
| 534 | 1211 *attrp = HLF_SPB; |
| 344 | 1212 else if (mi.mi_result == SP_RARE) |
| 534 | 1213 *attrp = HLF_SPR; |
| 344 | 1214 else |
| 534 | 1215 *attrp = HLF_SPL; |
| 243 | 1216 } |
| 1217 | |
| 483 | 1218 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE)) |
| 1219 { | |
| 1220 /* Report SpellCap only when the word isn't badly spelled. */ | |
| 534 | 1221 *attrp = HLF_SPC; |
| 483 | 1222 return wrongcaplen; |
| 1223 } | |
| 1224 | |
| 300 | 1225 return (int)(mi.mi_end - ptr); |
| 236 | 1226 } |
| 1227 | |
| 1228 /* | |
| 300 | 1229 * Check if the word at "mip->mi_word" is in the tree. |
| 339 | 1230 * When "mode" is FIND_FOLDWORD check in fold-case word tree. |
| 1231 * When "mode" is FIND_KEEPWORD check in keep-case word tree. | |
| 1232 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word | |
| 1233 * tree. | |
| 300 | 1234 * |
| 1235 * For a match mip->mi_result is updated. | |
| 243 | 1236 */ |
| 1237 static void | |
| 339 | 1238 find_word(mip, mode) |
| 243 | 1239 matchinf_T *mip; |
| 339 | 1240 int mode; |
| 243 | 1241 { |
| 324 | 1242 idx_T arridx = 0; |
| 300 | 1243 int endlen[MAXWLEN]; /* length at possible word endings */ |
| 324 | 1244 idx_T endidx[MAXWLEN]; /* possible word endings */ |
| 300 | 1245 int endidxcnt = 0; |
| 1246 int len; | |
| 1247 int wlen = 0; | |
| 1248 int flen; | |
| 1249 int c; | |
| 1250 char_u *ptr; | |
| 324 | 1251 idx_T lo, hi, m; |
| 243 | 1252 #ifdef FEAT_MBYTE |
| 300 | 1253 char_u *s; |
| 490 | 1254 #endif |
| 339 | 1255 char_u *p; |
| 307 | 1256 int res = SP_BAD; |
| 300 | 1257 slang_T *slang = mip->mi_lp->lp_slang; |
| 1258 unsigned flags; | |
| 1259 char_u *byts; | |
| 324 | 1260 idx_T *idxs; |
| 481 | 1261 int word_ends; |
| 485 | 1262 int prefix_found; |
| 492 | 1263 int nobreak_result; |
| 481 | 1264 |
| 1265 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND) | |
| 236 | 1266 { |
| 300 | 1267 /* Check for word with matching case in keep-case tree. */ |
| 1268 ptr = mip->mi_word; | |
| 1269 flen = 9999; /* no case folding, always enough bytes */ | |
| 1270 byts = slang->sl_kbyts; | |
| 1271 idxs = slang->sl_kidxs; | |
| 481 | 1272 |
| 1273 if (mode == FIND_KEEPCOMPOUND) | |
| 1274 /* Skip over the previously found word(s). */ | |
| 1275 wlen += mip->mi_compoff; | |
| 236 | 1276 } |
| 1277 else | |
| 1278 { | |
| 300 | 1279 /* Check for case-folded in case-folded tree. */ |
| 1280 ptr = mip->mi_fword; | |
| 1281 flen = mip->mi_fwordlen; /* available case-folded bytes */ | |
| 1282 byts = slang->sl_fbyts; | |
| 1283 idxs = slang->sl_fidxs; | |
| 339 | 1284 |
| 1285 if (mode == FIND_PREFIX) | |
| 1286 { | |
| 1287 /* Skip over the prefix. */ | |
| 1288 wlen = mip->mi_prefixlen; | |
| 1289 flen -= mip->mi_prefixlen; | |
| 1290 } | |
| 481 | 1291 else if (mode == FIND_COMPOUND) |
| 1292 { | |
| 1293 /* Skip over the previously found word(s). */ | |
| 1294 wlen = mip->mi_compoff; | |
| 1295 flen -= mip->mi_compoff; | |
| 1296 } | |
| 1297 | |
| 243 | 1298 } |
| 1299 | |
| 300 | 1300 if (byts == NULL) |
| 1301 return; /* array is empty */ | |
| 236 | 1302 |
| 1303 /* | |
| 307 | 1304 * Repeat advancing in the tree until: |
| 1305 * - there is a byte that doesn't match, | |
| 1306 * - we reach the end of the tree, | |
| 1307 * - or we reach the end of the line. | |
| 236 | 1308 */ |
| 300 | 1309 for (;;) |
| 236 | 1310 { |
| 346 | 1311 if (flen <= 0 && *mip->mi_fend != NUL) |
| 339 | 1312 flen = fold_more(mip); |
| 300 | 1313 |
| 1314 len = byts[arridx++]; | |
| 1315 | |
| 1316 /* If the first possible byte is a zero the word could end here. | |
| 1317 * Remember this index, we first check for the longest word. */ | |
| 1318 if (byts[arridx] == 0) | |
| 1319 { | |
| 307 | 1320 if (endidxcnt == MAXWLEN) |
| 1321 { | |
| 1322 /* Must be a corrupted spell file. */ | |
| 1323 EMSG(_(e_format)); | |
| 1324 return; | |
| 1325 } | |
| 300 | 1326 endlen[endidxcnt] = wlen; |
| 1327 endidx[endidxcnt++] = arridx++; | |
| 1328 --len; | |
| 1329 | |
| 1330 /* Skip over the zeros, there can be several flag/region | |
| 1331 * combinations. */ | |
| 1332 while (len > 0 && byts[arridx] == 0) | |
| 1333 { | |
| 1334 ++arridx; | |
| 1335 --len; | |
| 1336 } | |
| 1337 if (len == 0) | |
| 1338 break; /* no children, word must end here */ | |
| 1339 } | |
| 1340 | |
| 1341 /* Stop looking at end of the line. */ | |
| 1342 if (ptr[wlen] == NUL) | |
| 1343 break; | |
| 1344 | |
| 1345 /* Perform a binary search in the list of accepted bytes. */ | |
| 1346 c = ptr[wlen]; | |
| 346 | 1347 if (c == TAB) /* <Tab> is handled like <Space> */ |
| 1348 c = ' '; | |
| 300 | 1349 lo = arridx; |
| 1350 hi = arridx + len - 1; | |
| 1351 while (lo < hi) | |
| 1352 { | |
| 1353 m = (lo + hi) / 2; | |
| 1354 if (byts[m] > c) | |
| 1355 hi = m - 1; | |
| 1356 else if (byts[m] < c) | |
| 1357 lo = m + 1; | |
| 1358 else | |
| 1359 { | |
| 1360 lo = hi = m; | |
| 1361 break; | |
| 236 | 1362 } |
| 1363 } | |
| 300 | 1364 |
| 1365 /* Stop if there is no matching byte. */ | |
| 1366 if (hi < lo || byts[lo] != c) | |
| 1367 break; | |
| 1368 | |
| 1369 /* Continue at the child (if there is one). */ | |
| 1370 arridx = idxs[lo]; | |
| 1371 ++wlen; | |
| 1372 --flen; | |
| 346 | 1373 |
| 1374 /* One space in the good word may stand for several spaces in the | |
| 1375 * checked word. */ | |
| 1376 if (c == ' ') | |
| 1377 { | |
| 1378 for (;;) | |
| 1379 { | |
| 1380 if (flen <= 0 && *mip->mi_fend != NUL) | |
| 1381 flen = fold_more(mip); | |
| 1382 if (ptr[wlen] != ' ' && ptr[wlen] != TAB) | |
| 1383 break; | |
| 1384 ++wlen; | |
| 1385 --flen; | |
| 1386 } | |
| 1387 } | |
| 236 | 1388 } |
| 1389 | |
| 300 | 1390 /* |
| 1391 * Verify that one of the possible endings is valid. Try the longest | |
| 1392 * first. | |
| 1393 */ | |
| 1394 while (endidxcnt > 0) | |
| 1395 { | |
| 1396 --endidxcnt; | |
| 1397 arridx = endidx[endidxcnt]; | |
| 1398 wlen = endlen[endidxcnt]; | |
| 236 | 1399 |
| 300 | 1400 #ifdef FEAT_MBYTE |
| 1401 if ((*mb_head_off)(ptr, ptr + wlen) > 0) | |
| 1402 continue; /* not at first byte of character */ | |
| 1403 #endif | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1404 if (spell_iswordp(ptr + wlen, mip->mi_win)) |
| 481 | 1405 { |
| 492 | 1406 if (slang->sl_compprog == NULL && !slang->sl_nobreak) |
| 481 | 1407 continue; /* next char is a word character */ |
| 1408 word_ends = FALSE; | |
| 1409 } | |
| 1410 else | |
| 1411 word_ends = TRUE; | |
| 485 | 1412 /* The prefix flag is before compound flags. Once a valid prefix flag |
| 1413 * has been found we try compound flags. */ | |
| 1414 prefix_found = FALSE; | |
| 300 | 1415 |
| 1416 #ifdef FEAT_MBYTE | |
| 339 | 1417 if (mode != FIND_KEEPWORD && has_mbyte) |
| 300 | 1418 { |
| 1419 /* Compute byte length in original word, length may change | |
| 339 | 1420 * when folding case. This can be slow, take a shortcut when the |
| 1421 * case-folded word is equal to the keep-case word. */ | |
| 300 | 1422 p = mip->mi_word; |
| 339 | 1423 if (STRNCMP(ptr, p, wlen) != 0) |
| 1424 { | |
| 1425 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s)) | |
| 1426 mb_ptr_adv(p); | |
| 835 | 1427 wlen = (int)(p - mip->mi_word); |
| 339 | 1428 } |
| 300 | 1429 } |
| 1430 #endif | |
| 236 | 1431 |
| 339 | 1432 /* Check flags and region. For FIND_PREFIX check the condition and |
| 1433 * prefix ID. | |
| 1434 * Repeat this if there are more flags/region alternatives until there | |
| 1435 * is a match. */ | |
| 1436 res = SP_BAD; | |
| 1437 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0; | |
| 1438 --len, ++arridx) | |
| 300 | 1439 { |
| 1440 flags = idxs[arridx]; | |
| 324 | 1441 |
| 339 | 1442 /* For the fold-case tree check that the case of the checked word |
| 1443 * matches with what the word in the tree requires. | |
| 1444 * For keep-case tree the case is always right. For prefixes we | |
| 1445 * don't bother to check. */ | |
| 1446 if (mode == FIND_FOLDWORD) | |
| 300 | 1447 { |
| 1448 if (mip->mi_cend != mip->mi_word + wlen) | |
| 1449 { | |
| 323 | 1450 /* mi_capflags was set for a different word length, need |
| 1451 * to do it again. */ | |
| 300 | 1452 mip->mi_cend = mip->mi_word + wlen; |
| 323 | 1453 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend); |
| 300 | 1454 } |
| 1455 | |
| 346 | 1456 if (mip->mi_capflags == WF_KEEPCAP |
| 1457 || !spell_valid_case(mip->mi_capflags, flags)) | |
| 339 | 1458 continue; |
| 300 | 1459 } |
| 236 | 1460 |
| 339 | 1461 /* When mode is FIND_PREFIX the word must support the prefix: |
| 1462 * check the prefix ID and the condition. Do that for the list at | |
| 366 | 1463 * mip->mi_prefarridx that find_prefix() filled. */ |
| 485 | 1464 else if (mode == FIND_PREFIX && !prefix_found) |
| 481 | 1465 { |
| 366 | 1466 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx, |
| 390 | 1467 flags, |
| 455 | 1468 mip->mi_word + mip->mi_cprefixlen, slang, |
| 1469 FALSE); | |
| 366 | 1470 if (c == 0) |
| 339 | 1471 continue; |
| 366 | 1472 |
| 1473 /* Use the WF_RARE flag for a rare prefix. */ | |
| 1474 if (c & WF_RAREPFX) | |
| 1475 flags |= WF_RARE; | |
| 485 | 1476 prefix_found = TRUE; |
| 339 | 1477 } |
| 1478 | |
| 492 | 1479 if (slang->sl_nobreak) |
| 1480 { | |
| 1481 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND) | |
| 1482 && (flags & WF_BANNED) == 0) | |
| 1483 { | |
| 1484 /* NOBREAK: found a valid following word. That's all we | |
| 1485 * need to know, so return. */ | |
| 1486 mip->mi_result = SP_OK; | |
| 1487 break; | |
| 1488 } | |
| 1489 } | |
| 1490 | |
| 1491 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND | |
| 1492 || !word_ends)) | |
| 481 | 1493 { |
| 974 | 1494 /* If there is no compound flag or the word is shorter than |
| 483 | 1495 * COMPOUNDMIN reject it quickly. |
| 1496 * Makes you wonder why someone puts a compound flag on a word | |
| 481 | 1497 * that's too short... Myspell compatibility requires this |
| 1498 * anyway. */ | |
| 490 | 1499 if (((unsigned)flags >> 24) == 0 |
| 1500 || wlen - mip->mi_compoff < slang->sl_compminlen) | |
| 483 | 1501 continue; |
| 500 | 1502 #ifdef FEAT_MBYTE |
| 1503 /* For multi-byte chars check character length against | |
| 1504 * COMPOUNDMIN. */ | |
| 1505 if (has_mbyte | |
| 501 | 1506 && slang->sl_compminlen > 0 |
| 500 | 1507 && mb_charlen_len(mip->mi_word + mip->mi_compoff, |
| 1508 wlen - mip->mi_compoff) < slang->sl_compminlen) | |
| 1509 continue; | |
| 1510 #endif | |
| 483 | 1511 |
| 809 | 1512 /* Limit the number of compound words to COMPOUNDWORDMAX if no |
| 490 | 1513 * maximum for syllables is specified. */ |
| 809 | 1514 if (!word_ends && mip->mi_complen + mip->mi_compextra + 2 |
| 1515 > slang->sl_compmax | |
| 490 | 1516 && slang->sl_compsylmax == MAXWLEN) |
| 483 | 1517 continue; |
| 1518 | |
| 819 | 1519 /* Don't allow compounding on a side where an affix was added, |
| 1520 * unless COMPOUNDPERMITFLAG was used. */ | |
| 1521 if (mip->mi_complen > 0 && (flags & WF_NOCOMPBEF)) | |
| 1522 continue; | |
| 1523 if (!word_ends && (flags & WF_NOCOMPAFT)) | |
| 1524 continue; | |
| 1525 | |
| 485 | 1526 /* Quickly check if compounding is possible with this flag. */ |
| 495 | 1527 if (!byte_in_str(mip->mi_complen == 0 |
| 485 | 1528 ? slang->sl_compstartflags |
| 1529 : slang->sl_compallflags, | |
| 495 | 1530 ((unsigned)flags >> 24))) |
| 481 | 1531 continue; |
| 1532 | |
| 1762 | 1533 /* If there is a match with a CHECKCOMPOUNDPATTERN rule |
| 1534 * discard the compound word. */ | |
| 1535 if (match_checkcompoundpattern(ptr, wlen, &slang->sl_comppat)) | |
| 1536 continue; | |
| 1537 | |
| 490 | 1538 if (mode == FIND_COMPOUND) |
| 1539 { | |
| 1540 int capflags; | |
| 1541 | |
| 1542 /* Need to check the caps type of the appended compound | |
| 1543 * word. */ | |
| 1544 #ifdef FEAT_MBYTE | |
| 1545 if (has_mbyte && STRNCMP(ptr, mip->mi_word, | |
| 1546 mip->mi_compoff) != 0) | |
| 1547 { | |
| 1548 /* case folding may have changed the length */ | |
| 1549 p = mip->mi_word; | |
| 1550 for (s = ptr; s < ptr + mip->mi_compoff; mb_ptr_adv(s)) | |
| 1551 mb_ptr_adv(p); | |
| 1552 } | |
| 1553 else | |
| 1554 #endif | |
| 1555 p = mip->mi_word + mip->mi_compoff; | |
| 1556 capflags = captype(p, mip->mi_word + wlen); | |
| 1557 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP | |
| 1558 && (flags & WF_FIXCAP) != 0)) | |
| 1559 continue; | |
| 1560 | |
| 1561 if (capflags != WF_ALLCAP) | |
| 1562 { | |
| 1563 /* When the character before the word is a word | |
| 1564 * character we do not accept a Onecap word. We do | |
| 1565 * accept a no-caps word, even when the dictionary | |
| 1566 * word specifies ONECAP. */ | |
| 1567 mb_ptr_back(mip->mi_word, p); | |
| 5477 | 1568 if (spell_iswordp_nmw(p, mip->mi_win) |
| 490 | 1569 ? capflags == WF_ONECAP |
| 1570 : (flags & WF_ONECAP) != 0 | |
| 1571 && capflags != WF_ONECAP) | |
| 1572 continue; | |
| 1573 } | |
| 1574 } | |
| 1575 | |
| 483 | 1576 /* If the word ends the sequence of compound flags of the |
| 714 | 1577 * words must match with one of the COMPOUNDRULE items and |
| 483 | 1578 * the number of syllables must not be too large. */ |
| 1579 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24); | |
| 1580 mip->mi_compflags[mip->mi_complen + 1] = NUL; | |
| 1581 if (word_ends) | |
| 1582 { | |
| 1583 char_u fword[MAXWLEN]; | |
| 1584 | |
| 1585 if (slang->sl_compsylmax < MAXWLEN) | |
| 1586 { | |
| 1587 /* "fword" is only needed for checking syllables. */ | |
| 1588 if (ptr == mip->mi_word) | |
| 1589 (void)spell_casefold(ptr, wlen, fword, MAXWLEN); | |
| 1590 else | |
| 1591 vim_strncpy(fword, ptr, endlen[endidxcnt]); | |
| 1592 } | |
| 1593 if (!can_compound(slang, fword, mip->mi_compflags)) | |
| 1594 continue; | |
| 1595 } | |
| 1762 | 1596 else if (slang->sl_comprules != NULL |
| 1597 && !match_compoundrule(slang, mip->mi_compflags)) | |
| 1598 /* The compound flags collected so far do not match any | |
| 1599 * COMPOUNDRULE, discard the compounded word. */ | |
| 1600 continue; | |
| 481 | 1601 } |
| 1602 | |
| 500 | 1603 /* Check NEEDCOMPOUND: can't use word without compounding. */ |
| 1604 else if (flags & WF_NEEDCOMP) | |
| 1605 continue; | |
| 1606 | |
| 492 | 1607 nobreak_result = SP_OK; |
| 1608 | |
| 481 | 1609 if (!word_ends) |
| 1610 { | |
| 492 | 1611 int save_result = mip->mi_result; |
| 1612 char_u *save_end = mip->mi_end; | |
| 501 | 1613 langp_T *save_lp = mip->mi_lp; |
| 1614 int lpi; | |
| 492 | 1615 |
| 1616 /* Check that a valid word follows. If there is one and we | |
| 1617 * are compounding, it will set "mi_result", thus we are | |
| 1618 * always finished here. For NOBREAK we only check that a | |
| 1619 * valid word follows. | |
| 481 | 1620 * Recursive! */ |
| 492 | 1621 if (slang->sl_nobreak) |
| 1622 mip->mi_result = SP_BAD; | |
| 481 | 1623 |
| 1624 /* Find following word in case-folded tree. */ | |
| 1625 mip->mi_compoff = endlen[endidxcnt]; | |
| 1626 #ifdef FEAT_MBYTE | |
| 1627 if (has_mbyte && mode == FIND_KEEPWORD) | |
| 1628 { | |
| 1629 /* Compute byte length in case-folded word from "wlen": | |
| 1630 * byte length in keep-case word. Length may change when | |
| 1631 * folding case. This can be slow, take a shortcut when | |
| 1632 * the case-folded word is equal to the keep-case word. */ | |
| 1633 p = mip->mi_fword; | |
| 1634 if (STRNCMP(ptr, p, wlen) != 0) | |
| 1635 { | |
| 1636 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s)) | |
| 1637 mb_ptr_adv(p); | |
| 835 | 1638 mip->mi_compoff = (int)(p - mip->mi_fword); |
| 481 | 1639 } |
| 1640 } | |
| 1641 #endif | |
| 485 | 1642 c = mip->mi_compoff; |
| 483 | 1643 ++mip->mi_complen; |
| 809 | 1644 if (flags & WF_COMPROOT) |
| 1645 ++mip->mi_compextra; | |
| 501 | 1646 |
| 1647 /* For NOBREAK we need to try all NOBREAK languages, at least | |
| 1648 * to find the ".add" file(s). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1649 for (lpi = 0; lpi < mip->mi_win->w_s->b_langp.ga_len; ++lpi) |
| 501 | 1650 { |
| 1651 if (slang->sl_nobreak) | |
| 1652 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1653 mip->mi_lp = LANGP_ENTRY(mip->mi_win->w_s->b_langp, lpi); |
| 501 | 1654 if (mip->mi_lp->lp_slang->sl_fidxs == NULL |
| 1655 || !mip->mi_lp->lp_slang->sl_nobreak) | |
| 1656 continue; | |
| 1657 } | |
| 1658 | |
| 1659 find_word(mip, FIND_COMPOUND); | |
| 1660 | |
| 1661 /* When NOBREAK any word that matches is OK. Otherwise we | |
| 1662 * need to find the longest match, thus try with keep-case | |
| 1663 * and prefix too. */ | |
| 492 | 1664 if (!slang->sl_nobreak || mip->mi_result == SP_BAD) |
| 1665 { | |
| 501 | 1666 /* Find following word in keep-case tree. */ |
| 1667 mip->mi_compoff = wlen; | |
| 1668 find_word(mip, FIND_KEEPCOMPOUND); | |
| 1669 | |
| 819 | 1670 #if 0 /* Disabled, a prefix must not appear halfway a compound word, |
| 1671 unless the COMPOUNDPERMITFLAG is used and then it can't be a | |
| 1672 postponed prefix. */ | |
| 501 | 1673 if (!slang->sl_nobreak || mip->mi_result == SP_BAD) |
| 1674 { | |
| 1675 /* Check for following word with prefix. */ | |
| 1676 mip->mi_compoff = c; | |
| 1677 find_prefix(mip, FIND_COMPOUND); | |
| 1678 } | |
| 819 | 1679 #endif |
| 492 | 1680 } |
| 501 | 1681 |
| 1682 if (!slang->sl_nobreak) | |
| 1683 break; | |
| 492 | 1684 } |
| 483 | 1685 --mip->mi_complen; |
| 809 | 1686 if (flags & WF_COMPROOT) |
| 1687 --mip->mi_compextra; | |
| 501 | 1688 mip->mi_lp = save_lp; |
| 485 | 1689 |
| 492 | 1690 if (slang->sl_nobreak) |
| 1691 { | |
| 1692 nobreak_result = mip->mi_result; | |
| 1693 mip->mi_result = save_result; | |
| 1694 mip->mi_end = save_end; | |
| 1695 } | |
| 1696 else | |
| 1697 { | |
| 1698 if (mip->mi_result == SP_OK) | |
| 1699 break; | |
| 1700 continue; | |
| 1701 } | |
| 481 | 1702 } |
| 1703 | |
| 339 | 1704 if (flags & WF_BANNED) |
| 1705 res = SP_BANNED; | |
| 1706 else if (flags & WF_REGION) | |
| 1707 { | |
| 1708 /* Check region. */ | |
| 390 | 1709 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0) |
| 300 | 1710 res = SP_OK; |
| 339 | 1711 else |
| 1712 res = SP_LOCAL; | |
| 300 | 1713 } |
| 339 | 1714 else if (flags & WF_RARE) |
| 1715 res = SP_RARE; | |
| 307 | 1716 else |
| 339 | 1717 res = SP_OK; |
| 1718 | |
| 492 | 1719 /* Always use the longest match and the best result. For NOBREAK |
| 1720 * we separately keep the longest match without a following good | |
| 1721 * word as a fall-back. */ | |
| 1722 if (nobreak_result == SP_BAD) | |
| 1723 { | |
| 1724 if (mip->mi_result2 > res) | |
| 1725 { | |
| 1726 mip->mi_result2 = res; | |
| 1727 mip->mi_end2 = mip->mi_word + wlen; | |
| 1728 } | |
| 1729 else if (mip->mi_result2 == res | |
| 1730 && mip->mi_end2 < mip->mi_word + wlen) | |
| 1731 mip->mi_end2 = mip->mi_word + wlen; | |
| 1732 } | |
| 1733 else if (mip->mi_result > res) | |
| 339 | 1734 { |
| 1735 mip->mi_result = res; | |
| 1736 mip->mi_end = mip->mi_word + wlen; | |
| 1737 } | |
| 351 | 1738 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen) |
| 339 | 1739 mip->mi_end = mip->mi_word + wlen; |
| 1740 | |
| 492 | 1741 if (mip->mi_result == SP_OK) |
| 339 | 1742 break; |
| 300 | 1743 } |
| 1744 | |
| 492 | 1745 if (mip->mi_result == SP_OK) |
| 300 | 1746 break; |
| 1747 } | |
| 236 | 1748 } |
| 1749 | |
| 323 | 1750 /* |
| 1762 | 1751 * Return TRUE if there is a match between the word ptr[wlen] and |
| 1752 * CHECKCOMPOUNDPATTERN rules, assuming that we will concatenate with another | |
| 1753 * word. | |
| 1754 * A match means that the first part of CHECKCOMPOUNDPATTERN matches at the | |
| 1755 * end of ptr[wlen] and the second part matches after it. | |
| 1756 */ | |
| 1757 static int | |
| 1758 match_checkcompoundpattern(ptr, wlen, gap) | |
| 1759 char_u *ptr; | |
| 1760 int wlen; | |
| 1761 garray_T *gap; /* &sl_comppat */ | |
| 1762 { | |
| 1763 int i; | |
| 1764 char_u *p; | |
| 1765 int len; | |
| 1766 | |
| 1767 for (i = 0; i + 1 < gap->ga_len; i += 2) | |
| 1768 { | |
| 1769 p = ((char_u **)gap->ga_data)[i + 1]; | |
| 1770 if (STRNCMP(ptr + wlen, p, STRLEN(p)) == 0) | |
| 1771 { | |
| 1772 /* Second part matches at start of following compound word, now | |
| 1773 * check if first part matches at end of previous word. */ | |
| 1774 p = ((char_u **)gap->ga_data)[i]; | |
| 1771 | 1775 len = (int)STRLEN(p); |
| 1762 | 1776 if (len <= wlen && STRNCMP(ptr + wlen - len, p, len) == 0) |
| 1777 return TRUE; | |
| 1778 } | |
| 1779 } | |
| 1780 return FALSE; | |
| 1781 } | |
| 1782 | |
| 1783 /* | |
| 626 | 1784 * Return TRUE if "flags" is a valid sequence of compound flags and "word" |
| 1785 * does not have too many syllables. | |
| 482 | 1786 */ |
| 1787 static int | |
| 483 | 1788 can_compound(slang, word, flags) |
| 482 | 1789 slang_T *slang; |
| 483 | 1790 char_u *word; |
| 1791 char_u *flags; | |
| 1792 { | |
| 495 | 1793 #ifdef FEAT_MBYTE |
| 1794 char_u uflags[MAXWLEN * 2]; | |
| 1795 int i; | |
| 1796 #endif | |
| 1797 char_u *p; | |
| 483 | 1798 |
| 1799 if (slang->sl_compprog == NULL) | |
| 1800 return FALSE; | |
| 495 | 1801 #ifdef FEAT_MBYTE |
| 1802 if (enc_utf8) | |
| 1803 { | |
| 1804 /* Need to convert the single byte flags to utf8 characters. */ | |
| 1805 p = uflags; | |
| 1806 for (i = 0; flags[i] != NUL; ++i) | |
| 1807 p += mb_char2bytes(flags[i], p); | |
| 1808 *p = NUL; | |
| 1809 p = uflags; | |
| 1810 } | |
| 1811 else | |
| 1812 #endif | |
| 1813 p = flags; | |
| 6375 | 1814 if (!vim_regexec_prog(&slang->sl_compprog, FALSE, p, 0)) |
| 483 | 1815 return FALSE; |
| 1816 | |
| 490 | 1817 /* Count the number of syllables. This may be slow, do it last. If there |
| 1818 * are too many syllables AND the number of compound words is above | |
| 809 | 1819 * COMPOUNDWORDMAX then compounding is not allowed. */ |
| 483 | 1820 if (slang->sl_compsylmax < MAXWLEN |
| 1821 && count_syllables(slang, word) > slang->sl_compsylmax) | |
| 495 | 1822 return (int)STRLEN(flags) < slang->sl_compmax; |
| 483 | 1823 return TRUE; |
| 482 | 1824 } |
| 1825 | |
| 1826 /* | |
| 1762 | 1827 * Return TRUE when the sequence of flags in "compflags" plus "flag" can |
| 1828 * possibly form a valid compounded word. This also checks the COMPOUNDRULE | |
| 1829 * lines if they don't contain wildcards. | |
| 1830 */ | |
| 1831 static int | |
| 1832 can_be_compound(sp, slang, compflags, flag) | |
| 1833 trystate_T *sp; | |
| 1834 slang_T *slang; | |
| 1835 char_u *compflags; | |
| 1836 int flag; | |
| 1837 { | |
| 1838 /* If the flag doesn't appear in sl_compstartflags or sl_compallflags | |
| 1839 * then it can't possibly compound. */ | |
| 1840 if (!byte_in_str(sp->ts_complen == sp->ts_compsplit | |
| 1841 ? slang->sl_compstartflags : slang->sl_compallflags, flag)) | |
| 1842 return FALSE; | |
| 1843 | |
| 1844 /* If there are no wildcards, we can check if the flags collected so far | |
| 1845 * possibly can form a match with COMPOUNDRULE patterns. This only | |
| 1846 * makes sense when we have two or more words. */ | |
| 1847 if (slang->sl_comprules != NULL && sp->ts_complen > sp->ts_compsplit) | |
| 1848 { | |
| 1849 int v; | |
| 1850 | |
| 1851 compflags[sp->ts_complen] = flag; | |
| 1852 compflags[sp->ts_complen + 1] = NUL; | |
| 1853 v = match_compoundrule(slang, compflags + sp->ts_compsplit); | |
| 1854 compflags[sp->ts_complen] = NUL; | |
| 1855 return v; | |
| 1856 } | |
| 1857 | |
| 1858 return TRUE; | |
| 1859 } | |
| 1860 | |
| 1861 | |
| 1862 /* | |
| 1863 * Return TRUE if the compound flags in compflags[] match the start of any | |
| 1864 * compound rule. This is used to stop trying a compound if the flags | |
| 1865 * collected so far can't possibly match any compound rule. | |
| 1866 * Caller must check that slang->sl_comprules is not NULL. | |
| 1867 */ | |
| 1868 static int | |
| 1869 match_compoundrule(slang, compflags) | |
| 1870 slang_T *slang; | |
| 1871 char_u *compflags; | |
| 1872 { | |
| 1873 char_u *p; | |
| 1874 int i; | |
| 1875 int c; | |
| 1876 | |
| 1877 /* loop over all the COMPOUNDRULE entries */ | |
| 1878 for (p = slang->sl_comprules; *p != NUL; ++p) | |
| 1879 { | |
| 1880 /* loop over the flags in the compound word we have made, match | |
| 1881 * them against the current rule entry */ | |
| 1882 for (i = 0; ; ++i) | |
| 1883 { | |
| 1884 c = compflags[i]; | |
| 1885 if (c == NUL) | |
| 1886 /* found a rule that matches for the flags we have so far */ | |
| 1887 return TRUE; | |
| 1888 if (*p == '/' || *p == NUL) | |
| 1889 break; /* end of rule, it's too short */ | |
| 1890 if (*p == '[') | |
| 1891 { | |
| 1892 int match = FALSE; | |
| 1893 | |
| 1894 /* compare against all the flags in [] */ | |
| 1895 ++p; | |
| 1896 while (*p != ']' && *p != NUL) | |
| 1897 if (*p++ == c) | |
| 1898 match = TRUE; | |
| 1899 if (!match) | |
| 1900 break; /* none matches */ | |
| 1901 } | |
| 1902 else if (*p != c) | |
| 1903 break; /* flag of word doesn't match flag in pattern */ | |
| 1904 ++p; | |
| 1905 } | |
| 1906 | |
| 1907 /* Skip to the next "/", where the next pattern starts. */ | |
| 1908 p = vim_strchr(p, '/'); | |
| 1909 if (p == NULL) | |
| 1910 break; | |
| 1911 } | |
| 1912 | |
| 1913 /* Checked all the rules and none of them match the flags, so there | |
| 1914 * can't possibly be a compound starting with these flags. */ | |
| 1915 return FALSE; | |
| 1916 } | |
| 1917 | |
| 1918 /* | |
| 390 | 1919 * Return non-zero if the prefix indicated by "arridx" matches with the prefix |
| 1920 * ID in "flags" for the word "word". | |
| 366 | 1921 * The WF_RAREPFX flag is included in the return value for a rare prefix. |
| 351 | 1922 */ |
| 1923 static int | |
| 455 | 1924 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req) |
| 351 | 1925 int totprefcnt; /* nr of prefix IDs */ |
| 1926 int arridx; /* idx in sl_pidxs[] */ | |
| 390 | 1927 int flags; |
| 351 | 1928 char_u *word; |
| 1929 slang_T *slang; | |
| 455 | 1930 int cond_req; /* only use prefixes with a condition */ |
| 351 | 1931 { |
| 1932 int prefcnt; | |
| 1933 int pidx; | |
| 6375 | 1934 regprog_T **rp; |
| 390 | 1935 int prefid; |
| 1936 | |
| 1937 prefid = (unsigned)flags >> 24; | |
| 351 | 1938 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt) |
| 1939 { | |
| 1940 pidx = slang->sl_pidxs[arridx + prefcnt]; | |
| 1941 | |
| 1942 /* Check the prefix ID. */ | |
| 1943 if (prefid != (pidx & 0xff)) | |
| 1944 continue; | |
| 1945 | |
| 390 | 1946 /* Check if the prefix doesn't combine and the word already has a |
| 1947 * suffix. */ | |
| 1948 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC)) | |
| 1949 continue; | |
| 1950 | |
| 351 | 1951 /* Check the condition, if there is one. The condition index is |
| 366 | 1952 * stored in the two bytes above the prefix ID byte. */ |
| 6375 | 1953 rp = &slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff]; |
| 1954 if (*rp != NULL) | |
| 1955 { | |
| 1956 if (!vim_regexec_prog(rp, FALSE, word, 0)) | |
| 351 | 1957 continue; |
| 1958 } | |
| 455 | 1959 else if (cond_req) |
| 1960 continue; | |
| 1961 | |
| 1962 /* It's a match! Return the WF_ flags. */ | |
| 366 | 1963 return pidx; |
| 1964 } | |
| 1965 return 0; | |
| 351 | 1966 } |
| 1967 | |
| 1968 /* | |
| 339 | 1969 * Check if the word at "mip->mi_word" has a matching prefix. |
| 1970 * If it does, then check the following word. | |
| 1971 * | |
| 485 | 1972 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a |
| 1973 * prefix in a compound word. | |
| 1974 * | |
| 339 | 1975 * For a match mip->mi_result is updated. |
| 1976 */ | |
| 1977 static void | |
| 485 | 1978 find_prefix(mip, mode) |
| 339 | 1979 matchinf_T *mip; |
| 485 | 1980 int mode; |
| 339 | 1981 { |
| 1982 idx_T arridx = 0; | |
| 1983 int len; | |
| 1984 int wlen = 0; | |
| 1985 int flen; | |
| 1986 int c; | |
| 1987 char_u *ptr; | |
| 1988 idx_T lo, hi, m; | |
| 1989 slang_T *slang = mip->mi_lp->lp_slang; | |
| 1990 char_u *byts; | |
| 1991 idx_T *idxs; | |
| 1992 | |
| 375 | 1993 byts = slang->sl_pbyts; |
| 1994 if (byts == NULL) | |
| 1995 return; /* array is empty */ | |
| 1996 | |
| 339 | 1997 /* We use the case-folded word here, since prefixes are always |
| 1998 * case-folded. */ | |
| 1999 ptr = mip->mi_fword; | |
| 2000 flen = mip->mi_fwordlen; /* available case-folded bytes */ | |
| 485 | 2001 if (mode == FIND_COMPOUND) |
| 2002 { | |
| 2003 /* Skip over the previously found word(s). */ | |
| 2004 ptr += mip->mi_compoff; | |
| 2005 flen -= mip->mi_compoff; | |
| 2006 } | |
| 339 | 2007 idxs = slang->sl_pidxs; |
| 2008 | |
| 2009 /* | |
| 2010 * Repeat advancing in the tree until: | |
| 2011 * - there is a byte that doesn't match, | |
| 2012 * - we reach the end of the tree, | |
| 2013 * - or we reach the end of the line. | |
| 2014 */ | |
| 2015 for (;;) | |
| 2016 { | |
| 2017 if (flen == 0 && *mip->mi_fend != NUL) | |
| 2018 flen = fold_more(mip); | |
| 2019 | |
| 2020 len = byts[arridx++]; | |
| 2021 | |
| 2022 /* If the first possible byte is a zero the prefix could end here. | |
| 2023 * Check if the following word matches and supports the prefix. */ | |
| 2024 if (byts[arridx] == 0) | |
| 2025 { | |
| 2026 /* There can be several prefixes with different conditions. We | |
| 2027 * try them all, since we don't know which one will give the | |
| 2028 * longest match. The word is the same each time, pass the list | |
| 2029 * of possible prefixes to find_word(). */ | |
| 2030 mip->mi_prefarridx = arridx; | |
| 2031 mip->mi_prefcnt = len; | |
| 2032 while (len > 0 && byts[arridx] == 0) | |
| 2033 { | |
| 2034 ++arridx; | |
| 2035 --len; | |
| 2036 } | |
| 2037 mip->mi_prefcnt -= len; | |
| 2038 | |
| 2039 /* Find the word that comes after the prefix. */ | |
| 2040 mip->mi_prefixlen = wlen; | |
| 485 | 2041 if (mode == FIND_COMPOUND) |
| 2042 /* Skip over the previously found word(s). */ | |
| 2043 mip->mi_prefixlen += mip->mi_compoff; | |
| 2044 | |
| 455 | 2045 #ifdef FEAT_MBYTE |
| 2046 if (has_mbyte) | |
| 2047 { | |
| 2048 /* Case-folded length may differ from original length. */ | |
| 485 | 2049 mip->mi_cprefixlen = nofold_len(mip->mi_fword, |
| 2050 mip->mi_prefixlen, mip->mi_word); | |
| 455 | 2051 } |
| 2052 else | |
| 485 | 2053 mip->mi_cprefixlen = mip->mi_prefixlen; |
| 455 | 2054 #endif |
| 339 | 2055 find_word(mip, FIND_PREFIX); |
| 2056 | |
| 2057 | |
| 2058 if (len == 0) | |
| 2059 break; /* no children, word must end here */ | |
| 2060 } | |
| 2061 | |
| 2062 /* Stop looking at end of the line. */ | |
| 2063 if (ptr[wlen] == NUL) | |
| 2064 break; | |
| 2065 | |
| 2066 /* Perform a binary search in the list of accepted bytes. */ | |
| 2067 c = ptr[wlen]; | |
| 2068 lo = arridx; | |
| 2069 hi = arridx + len - 1; | |
| 2070 while (lo < hi) | |
| 2071 { | |
| 2072 m = (lo + hi) / 2; | |
| 2073 if (byts[m] > c) | |
| 2074 hi = m - 1; | |
| 2075 else if (byts[m] < c) | |
| 2076 lo = m + 1; | |
| 2077 else | |
| 2078 { | |
| 2079 lo = hi = m; | |
| 2080 break; | |
| 2081 } | |
| 2082 } | |
| 2083 | |
| 2084 /* Stop if there is no matching byte. */ | |
| 2085 if (hi < lo || byts[lo] != c) | |
| 2086 break; | |
| 2087 | |
| 2088 /* Continue at the child (if there is one). */ | |
| 2089 arridx = idxs[lo]; | |
| 2090 ++wlen; | |
| 2091 --flen; | |
| 2092 } | |
| 2093 } | |
| 2094 | |
| 2095 /* | |
| 2096 * Need to fold at least one more character. Do until next non-word character | |
| 626 | 2097 * for efficiency. Include the non-word character too. |
| 339 | 2098 * Return the length of the folded chars in bytes. |
| 2099 */ | |
| 2100 static int | |
| 2101 fold_more(mip) | |
| 2102 matchinf_T *mip; | |
| 2103 { | |
| 2104 int flen; | |
| 2105 char_u *p; | |
| 2106 | |
| 2107 p = mip->mi_fend; | |
| 2108 do | |
| 2109 { | |
| 2110 mb_ptr_adv(mip->mi_fend); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2111 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_win)); |
| 339 | 2112 |
| 626 | 2113 /* Include the non-word character so that we can check for the word end. */ |
| 339 | 2114 if (*mip->mi_fend != NUL) |
| 2115 mb_ptr_adv(mip->mi_fend); | |
| 2116 | |
| 2117 (void)spell_casefold(p, (int)(mip->mi_fend - p), | |
| 2118 mip->mi_fword + mip->mi_fwordlen, | |
| 2119 MAXWLEN - mip->mi_fwordlen); | |
| 835 | 2120 flen = (int)STRLEN(mip->mi_fword + mip->mi_fwordlen); |
| 339 | 2121 mip->mi_fwordlen += flen; |
| 2122 return flen; | |
| 2123 } | |
| 2124 | |
| 2125 /* | |
| 323 | 2126 * Check case flags for a word. Return TRUE if the word has the requested |
| 2127 * case. | |
| 2128 */ | |
| 2129 static int | |
| 389 | 2130 spell_valid_case(wordflags, treeflags) |
| 2131 int wordflags; /* flags for the checked word. */ | |
| 323 | 2132 int treeflags; /* flags for the word in the spell tree */ |
| 2133 { | |
| 389 | 2134 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0) |
| 323 | 2135 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0 |
| 474 | 2136 && ((treeflags & WF_ONECAP) == 0 |
| 2137 || (wordflags & WF_ONECAP) != 0))); | |
| 323 | 2138 } |
| 2139 | |
| 351 | 2140 /* |
| 2141 * Return TRUE if spell checking is not enabled. | |
| 2142 */ | |
| 2143 static int | |
| 498 | 2144 no_spell_checking(wp) |
| 2145 win_T *wp; | |
| 2146 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2147 if (!wp->w_p_spell || *wp->w_s->b_p_spl == NUL |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2148 || wp->w_s->b_langp.ga_len == 0) |
| 351 | 2149 { |
| 2150 EMSG(_("E756: Spell checking is not enabled")); | |
| 2151 return TRUE; | |
| 2152 } | |
| 2153 return FALSE; | |
| 2154 } | |
| 300 | 2155 |
| 236 | 2156 /* |
| 2157 * Move to next spell error. | |
| 500 | 2158 * "curline" is FALSE for "[s", "]s", "[S" and "]S". |
| 2159 * "curline" is TRUE to find word under/after cursor in the same line. | |
| 483 | 2160 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move |
| 2161 * to after badly spelled word before the cursor. | |
| 495 | 2162 * Return 0 if not found, length of the badly spelled word otherwise. |
| 236 | 2163 */ |
| 2164 int | |
| 498 | 2165 spell_move_to(wp, dir, allwords, curline, attrp) |
| 2166 win_T *wp; | |
| 236 | 2167 int dir; /* FORWARD or BACKWARD */ |
| 500 | 2168 int allwords; /* TRUE for "[s"/"]s", FALSE for "[S"/"]S" */ |
| 323 | 2169 int curline; |
| 534 | 2170 hlf_T *attrp; /* return: attributes of bad word or NULL |
| 2171 (only when "dir" is FORWARD) */ | |
| 236 | 2172 { |
| 249 | 2173 linenr_T lnum; |
| 2174 pos_T found_pos; | |
| 495 | 2175 int found_len = 0; |
| 236 | 2176 char_u *line; |
| 2177 char_u *p; | |
| 346 | 2178 char_u *endp; |
| 534 | 2179 hlf_T attr; |
| 236 | 2180 int len; |
| 5519 | 2181 #ifdef FEAT_SYN_HL |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2182 int has_syntax = syntax_present(wp); |
| 5519 | 2183 #endif |
| 249 | 2184 int col; |
| 2185 int can_spell; | |
| 346 | 2186 char_u *buf = NULL; |
| 2187 int buflen = 0; | |
| 2188 int skip = 0; | |
| 385 | 2189 int capcol = -1; |
| 500 | 2190 int found_one = FALSE; |
| 2191 int wrapped = FALSE; | |
| 236 | 2192 |
| 498 | 2193 if (no_spell_checking(wp)) |
| 495 | 2194 return 0; |
| 236 | 2195 |
| 249 | 2196 /* |
| 2197 * Start looking for bad word at the start of the line, because we can't | |
| 817 | 2198 * start halfway a word, we don't know where it starts or ends. |
| 249 | 2199 * |
| 2200 * When searching backwards, we continue in the line to find the last | |
| 2201 * bad word (in the cursor line: before the cursor). | |
| 346 | 2202 * |
| 2203 * We concatenate the start of the next line, so that wrapped words work | |
| 2204 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards | |
| 2205 * though... | |
| 249 | 2206 */ |
| 498 | 2207 lnum = wp->w_cursor.lnum; |
| 698 | 2208 clearpos(&found_pos); |
| 236 | 2209 |
| 2210 while (!got_int) | |
| 2211 { | |
| 498 | 2212 line = ml_get_buf(wp->w_buffer, lnum, FALSE); |
| 346 | 2213 |
| 835 | 2214 len = (int)STRLEN(line); |
| 346 | 2215 if (buflen < len + MAXWLEN + 2) |
| 2216 { | |
| 2217 vim_free(buf); | |
| 2218 buflen = len + MAXWLEN + 2; | |
| 2219 buf = alloc(buflen); | |
| 2220 if (buf == NULL) | |
| 2221 break; | |
| 2222 } | |
| 2223 | |
| 385 | 2224 /* In first line check first word for Capital. */ |
| 2225 if (lnum == 1) | |
| 2226 capcol = 0; | |
| 2227 | |
| 2228 /* For checking first word with a capital skip white space. */ | |
| 2229 if (capcol == 0) | |
| 835 | 2230 capcol = (int)(skipwhite(line) - line); |
| 2231 else if (curline && wp == curwin) | |
| 2232 { | |
| 2233 /* For spellbadword(): check if first word needs a capital. */ | |
| 944 | 2234 col = (int)(skipwhite(line) - line); |
| 835 | 2235 if (check_need_cap(lnum, col)) |
| 2236 capcol = col; | |
| 2237 | |
| 2238 /* Need to get the line again, may have looked at the previous | |
| 2239 * one. */ | |
| 2240 line = ml_get_buf(wp->w_buffer, lnum, FALSE); | |
| 2241 } | |
| 385 | 2242 |
| 346 | 2243 /* Copy the line into "buf" and append the start of the next line if |
| 2244 * possible. */ | |
| 2245 STRCPY(buf, line); | |
| 498 | 2246 if (lnum < wp->w_buffer->b_ml.ml_line_count) |
| 883 | 2247 spell_cat_line(buf + STRLEN(buf), |
| 2248 ml_get_buf(wp->w_buffer, lnum + 1, FALSE), MAXWLEN); | |
| 346 | 2249 |
| 2250 p = buf + skip; | |
| 2251 endp = buf + len; | |
| 2252 while (p < endp) | |
| 236 | 2253 { |
| 500 | 2254 /* When searching backward don't search after the cursor. Unless |
| 2255 * we wrapped around the end of the buffer. */ | |
| 300 | 2256 if (dir == BACKWARD |
| 498 | 2257 && lnum == wp->w_cursor.lnum |
| 500 | 2258 && !wrapped |
| 498 | 2259 && (colnr_T)(p - buf) >= wp->w_cursor.col) |
| 300 | 2260 break; |
| 249 | 2261 |
| 300 | 2262 /* start of word */ |
| 534 | 2263 attr = HLF_COUNT; |
| 625 | 2264 len = spell_check(wp, p, &attr, &capcol, FALSE); |
| 249 | 2265 |
| 534 | 2266 if (attr != HLF_COUNT) |
| 300 | 2267 { |
| 2268 /* We found a bad word. Check the attribute. */ | |
| 534 | 2269 if (allwords || attr == HLF_SPB) |
| 236 | 2270 { |
| 300 | 2271 /* When searching forward only accept a bad word after |
| 2272 * the cursor. */ | |
| 2273 if (dir == BACKWARD | |
| 500 | 2274 || lnum != wp->w_cursor.lnum |
| 498 | 2275 || (lnum == wp->w_cursor.lnum |
| 500 | 2276 && (wrapped |
| 2277 || (colnr_T)(curline ? p - buf + len | |
| 346 | 2278 : p - buf) |
| 500 | 2279 > wp->w_cursor.col))) |
| 236 | 2280 { |
| 5519 | 2281 #ifdef FEAT_SYN_HL |
| 300 | 2282 if (has_syntax) |
| 249 | 2283 { |
| 835 | 2284 col = (int)(p - buf); |
| 498 | 2285 (void)syn_get_id(wp, lnum, (colnr_T)col, |
| 1504 | 2286 FALSE, &can_spell, FALSE); |
| 857 | 2287 if (!can_spell) |
| 2288 attr = HLF_COUNT; | |
| 300 | 2289 } |
| 2290 else | |
| 737 | 2291 #endif |
| 300 | 2292 can_spell = TRUE; |
| 249 | 2293 |
| 300 | 2294 if (can_spell) |
| 2295 { | |
| 857 | 2296 found_one = TRUE; |
| 300 | 2297 found_pos.lnum = lnum; |
| 835 | 2298 found_pos.col = (int)(p - buf); |
| 249 | 2299 #ifdef FEAT_VIRTUALEDIT |
| 300 | 2300 found_pos.coladd = 0; |
| 249 | 2301 #endif |
| 300 | 2302 if (dir == FORWARD) |
| 2303 { | |
| 2304 /* No need to search further. */ | |
| 498 | 2305 wp->w_cursor = found_pos; |
| 346 | 2306 vim_free(buf); |
| 498 | 2307 if (attrp != NULL) |
| 2308 *attrp = attr; | |
| 495 | 2309 return len; |
| 249 | 2310 } |
| 483 | 2311 else if (curline) |
| 2312 /* Insert mode completion: put cursor after | |
| 2313 * the bad word. */ | |
| 2314 found_pos.col += len; | |
| 495 | 2315 found_len = len; |
| 249 | 2316 } |
| 236 | 2317 } |
| 857 | 2318 else |
| 2319 found_one = TRUE; | |
| 236 | 2320 } |
| 2321 } | |
| 2322 | |
| 300 | 2323 /* advance to character after the word */ |
| 2324 p += len; | |
| 385 | 2325 capcol -= len; |
| 236 | 2326 } |
| 2327 | |
| 483 | 2328 if (dir == BACKWARD && found_pos.lnum != 0) |
| 2329 { | |
| 500 | 2330 /* Use the last match in the line (before the cursor). */ |
| 498 | 2331 wp->w_cursor = found_pos; |
| 483 | 2332 vim_free(buf); |
| 495 | 2333 return found_len; |
| 483 | 2334 } |
| 2335 | |
| 323 | 2336 if (curline) |
| 346 | 2337 break; /* only check cursor line */ |
| 323 | 2338 |
| 236 | 2339 /* Advance to next line. */ |
| 249 | 2340 if (dir == BACKWARD) |
| 2341 { | |
| 500 | 2342 /* If we are back at the starting line and searched it again there |
| 2343 * is no match, give up. */ | |
| 2344 if (lnum == wp->w_cursor.lnum && wrapped) | |
| 346 | 2345 break; |
| 500 | 2346 |
| 2347 if (lnum > 1) | |
| 2348 --lnum; | |
| 2349 else if (!p_ws) | |
| 2350 break; /* at first line and 'nowrapscan' */ | |
| 2351 else | |
| 2352 { | |
| 2353 /* Wrap around to the end of the buffer. May search the | |
| 2354 * starting line again and accept the last match. */ | |
| 2355 lnum = wp->w_buffer->b_ml.ml_line_count; | |
| 2356 wrapped = TRUE; | |
| 503 | 2357 if (!shortmess(SHM_SEARCH)) |
| 2358 give_warning((char_u *)_(top_bot_msg), TRUE); | |
| 500 | 2359 } |
| 385 | 2360 capcol = -1; |
| 249 | 2361 } |
| 2362 else | |
| 2363 { | |
| 500 | 2364 if (lnum < wp->w_buffer->b_ml.ml_line_count) |
| 2365 ++lnum; | |
| 2366 else if (!p_ws) | |
| 2367 break; /* at first line and 'nowrapscan' */ | |
| 2368 else | |
| 2369 { | |
| 2370 /* Wrap around to the start of the buffer. May search the | |
| 2371 * starting line again and accept the first match. */ | |
| 2372 lnum = 1; | |
| 2373 wrapped = TRUE; | |
| 503 | 2374 if (!shortmess(SHM_SEARCH)) |
| 2375 give_warning((char_u *)_(bot_top_msg), TRUE); | |
| 500 | 2376 } |
| 2377 | |
| 2378 /* If we are back at the starting line and there is no match then | |
| 2379 * give up. */ | |
| 1808 | 2380 if (lnum == wp->w_cursor.lnum && (!found_one || wrapped)) |
| 346 | 2381 break; |
| 2382 | |
| 2383 /* Skip the characters at the start of the next line that were | |
| 2384 * included in a match crossing line boundaries. */ | |
| 534 | 2385 if (attr == HLF_COUNT) |
| 835 | 2386 skip = (int)(p - endp); |
| 346 | 2387 else |
| 2388 skip = 0; | |
| 385 | 2389 |
| 835 | 2390 /* Capcol skips over the inserted space. */ |
| 385 | 2391 --capcol; |
| 2392 | |
| 2393 /* But after empty line check first word in next line */ | |
| 2394 if (*skipwhite(line) == NUL) | |
| 2395 capcol = 0; | |
| 249 | 2396 } |
| 236 | 2397 |
| 2398 line_breakcheck(); | |
| 2399 } | |
| 2400 | |
| 346 | 2401 vim_free(buf); |
| 495 | 2402 return 0; |
| 346 | 2403 } |
| 2404 | |
| 2405 /* | |
| 2406 * For spell checking: concatenate the start of the following line "line" into | |
| 2407 * "buf", blanking-out special characters. Copy less then "maxlen" bytes. | |
| 1577 | 2408 * Keep the blanks at the start of the next line, this is used in win_line() |
| 2409 * to skip those bytes if the word was OK. | |
| 346 | 2410 */ |
| 2411 void | |
| 2412 spell_cat_line(buf, line, maxlen) | |
| 2413 char_u *buf; | |
| 2414 char_u *line; | |
| 2415 int maxlen; | |
| 2416 { | |
| 2417 char_u *p; | |
| 2418 int n; | |
| 2419 | |
| 2420 p = skipwhite(line); | |
| 2421 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL) | |
| 2422 p = skipwhite(p + 1); | |
| 2423 | |
| 2424 if (*p != NUL) | |
| 2425 { | |
| 1577 | 2426 /* Only worth concatenating if there is something else than spaces to |
| 2427 * concatenate. */ | |
| 2428 n = (int)(p - line) + 1; | |
| 2429 if (n < maxlen - 1) | |
| 2430 { | |
| 2431 vim_memset(buf, ' ', n); | |
| 2432 vim_strncpy(buf + n, p, maxlen - 1 - n); | |
| 2433 } | |
| 346 | 2434 } |
| 236 | 2435 } |
| 2436 | |
| 626 | 2437 /* |
| 2438 * Structure used for the cookie argument of do_in_runtimepath(). | |
| 2439 */ | |
| 501 | 2440 typedef struct spelload_S |
| 2441 { | |
| 2442 char_u sl_lang[MAXWLEN + 1]; /* language name */ | |
| 2443 slang_T *sl_slang; /* resulting slang_T struct */ | |
| 2444 int sl_nobreak; /* NOBREAK language found */ | |
| 2445 } spelload_T; | |
| 2446 | |
| 236 | 2447 /* |
| 307 | 2448 * Load word list(s) for "lang" from Vim spell file(s). |
| 310 | 2449 * "lang" must be the language without the region: e.g., "en". |
| 236 | 2450 */ |
| 307 | 2451 static void |
| 236 | 2452 spell_load_lang(lang) |
| 2453 char_u *lang; | |
| 2454 { | |
| 310 | 2455 char_u fname_enc[85]; |
| 236 | 2456 int r; |
| 501 | 2457 spelload_T sl; |
| 649 | 2458 #ifdef FEAT_AUTOCMD |
| 2459 int round; | |
| 2460 #endif | |
| 307 | 2461 |
| 310 | 2462 /* Copy the language name to pass it to spell_load_cb() as a cookie. |
| 307 | 2463 * It's truncated when an error is detected. */ |
| 501 | 2464 STRCPY(sl.sl_lang, lang); |
| 2465 sl.sl_slang = NULL; | |
| 2466 sl.sl_nobreak = FALSE; | |
| 307 | 2467 |
| 649 | 2468 #ifdef FEAT_AUTOCMD |
| 2469 /* We may retry when no spell file is found for the language, an | |
| 2470 * autocommand may load it then. */ | |
| 2471 for (round = 1; round <= 2; ++round) | |
| 2472 #endif | |
| 2473 { | |
| 2474 /* | |
| 2475 * Find the first spell file for "lang" in 'runtimepath' and load it. | |
| 2476 */ | |
| 2477 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 2478 #ifdef VMS |
| 2479 "spell/%s_%s.spl", | |
| 2480 #else | |
| 2481 "spell/%s.%s.spl", | |
| 2482 #endif | |
| 2483 lang, spell_enc()); | |
| 649 | 2484 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl); |
| 2485 | |
| 2486 if (r == FAIL && *sl.sl_lang != NUL) | |
| 2487 { | |
| 2488 /* Try loading the ASCII version. */ | |
| 2489 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 2490 #ifdef VMS |
| 2491 "spell/%s_ascii.spl", | |
| 2492 #else | |
| 2493 "spell/%s.ascii.spl", | |
| 2494 #endif | |
| 2495 lang); | |
| 649 | 2496 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl); |
| 2497 | |
| 2498 #ifdef FEAT_AUTOCMD | |
| 2499 if (r == FAIL && *sl.sl_lang != NUL && round == 1 | |
| 2500 && apply_autocmds(EVENT_SPELLFILEMISSING, lang, | |
| 2501 curbuf->b_fname, FALSE, curbuf)) | |
| 2502 continue; | |
| 2503 break; | |
| 2504 #endif | |
| 2505 } | |
| 714 | 2506 #ifdef FEAT_AUTOCMD |
| 2507 break; | |
| 2508 #endif | |
| 307 | 2509 } |
| 2510 | |
| 2511 if (r == FAIL) | |
| 649 | 2512 { |
| 2660 | 2513 smsg((char_u *) |
| 2514 #ifdef VMS | |
| 2515 _("Warning: Cannot find word list \"%s_%s.spl\" or \"%s_ascii.spl\""), | |
| 2516 #else | |
| 2517 _("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""), | |
| 2518 #endif | |
| 483 | 2519 lang, spell_enc(), lang); |
| 649 | 2520 } |
| 501 | 2521 else if (sl.sl_slang != NULL) |
| 2522 { | |
| 626 | 2523 /* At least one file was loaded, now load ALL the additions. */ |
| 310 | 2524 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl"); |
| 501 | 2525 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &sl); |
| 310 | 2526 } |
| 2527 } | |
| 2528 | |
| 2529 /* | |
| 2530 * Return the encoding used for spell checking: Use 'encoding', except that we | |
| 2531 * use "latin1" for "latin9". And limit to 60 characters (just in case). | |
| 2532 */ | |
| 2533 static char_u * | |
| 2534 spell_enc() | |
| 2535 { | |
| 2536 | |
| 2537 #ifdef FEAT_MBYTE | |
| 2538 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0) | |
| 2539 return p_enc; | |
| 2540 #endif | |
| 2541 return (char_u *)"latin1"; | |
| 236 | 2542 } |
| 2543 | |
| 2544 /* | |
| 385 | 2545 * Get the name of the .spl file for the internal wordlist into |
| 2546 * "fname[MAXPATHL]". | |
| 2547 */ | |
| 2548 static void | |
| 2549 int_wordlist_spl(fname) | |
| 2550 char_u *fname; | |
| 2551 { | |
| 2660 | 2552 vim_snprintf((char *)fname, MAXPATHL, SPL_FNAME_TMPL, |
| 385 | 2553 int_wordlist, spell_enc()); |
| 2554 } | |
| 2555 | |
| 2556 /* | |
| 625 | 2557 * Allocate a new slang_T for language "lang". "lang" can be NULL. |
| 236 | 2558 * Caller must fill "sl_next". |
| 2559 */ | |
| 2560 static slang_T * | |
| 2561 slang_alloc(lang) | |
| 2562 char_u *lang; | |
| 2563 { | |
| 2564 slang_T *lp; | |
| 2565 | |
| 300 | 2566 lp = (slang_T *)alloc_clear(sizeof(slang_T)); |
| 236 | 2567 if (lp != NULL) |
| 2568 { | |
| 625 | 2569 if (lang != NULL) |
| 2570 lp->sl_name = vim_strsave(lang); | |
| 323 | 2571 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10); |
| 625 | 2572 ga_init2(&lp->sl_repsal, sizeof(fromto_T), 10); |
| 483 | 2573 lp->sl_compmax = MAXWLEN; |
| 2574 lp->sl_compsylmax = MAXWLEN; | |
| 625 | 2575 hash_init(&lp->sl_wordcount); |
| 2576 } | |
| 2577 | |
| 236 | 2578 return lp; |
| 2579 } | |
| 2580 | |
| 2581 /* | |
| 2582 * Free the contents of an slang_T and the structure itself. | |
| 2583 */ | |
| 2584 static void | |
| 2585 slang_free(lp) | |
| 2586 slang_T *lp; | |
| 2587 { | |
| 2588 vim_free(lp->sl_name); | |
| 310 | 2589 vim_free(lp->sl_fname); |
| 2590 slang_clear(lp); | |
| 2591 vim_free(lp); | |
| 2592 } | |
| 2593 | |
| 2594 /* | |
| 2595 * Clear an slang_T so that the file can be reloaded. | |
| 2596 */ | |
| 2597 static void | |
| 2598 slang_clear(lp) | |
| 2599 slang_T *lp; | |
| 2600 { | |
| 339 | 2601 garray_T *gap; |
| 2602 fromto_T *ftp; | |
| 344 | 2603 salitem_T *smp; |
| 339 | 2604 int i; |
| 625 | 2605 int round; |
| 323 | 2606 |
| 300 | 2607 vim_free(lp->sl_fbyts); |
| 310 | 2608 lp->sl_fbyts = NULL; |
| 300 | 2609 vim_free(lp->sl_kbyts); |
| 310 | 2610 lp->sl_kbyts = NULL; |
| 339 | 2611 vim_free(lp->sl_pbyts); |
| 2612 lp->sl_pbyts = NULL; | |
| 2613 | |
| 300 | 2614 vim_free(lp->sl_fidxs); |
| 310 | 2615 lp->sl_fidxs = NULL; |
| 300 | 2616 vim_free(lp->sl_kidxs); |
| 310 | 2617 lp->sl_kidxs = NULL; |
| 339 | 2618 vim_free(lp->sl_pidxs); |
| 2619 lp->sl_pidxs = NULL; | |
| 323 | 2620 |
| 625 | 2621 for (round = 1; round <= 2; ++round) |
| 2622 { | |
| 2623 gap = round == 1 ? &lp->sl_rep : &lp->sl_repsal; | |
| 2624 while (gap->ga_len > 0) | |
| 2625 { | |
| 2626 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len]; | |
| 2627 vim_free(ftp->ft_from); | |
| 2628 vim_free(ftp->ft_to); | |
| 2629 } | |
| 2630 ga_clear(gap); | |
| 2631 } | |
| 344 | 2632 |
| 2633 gap = &lp->sl_sal; | |
| 375 | 2634 if (lp->sl_sofo) |
| 376 | 2635 { |
| 2636 /* "ga_len" is set to 1 without adding an item for latin1 */ | |
| 2637 if (gap->ga_data != NULL) | |
| 2638 /* SOFOFROM and SOFOTO items: free lists of wide characters. */ | |
| 2639 for (i = 0; i < gap->ga_len; ++i) | |
| 2640 vim_free(((int **)gap->ga_data)[i]); | |
| 2641 } | |
| 375 | 2642 else |
| 2643 /* SAL items: free salitem_T items */ | |
| 2644 while (gap->ga_len > 0) | |
| 2645 { | |
| 2646 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len]; | |
| 2647 vim_free(smp->sm_lead); | |
| 2648 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */ | |
| 2649 vim_free(smp->sm_to); | |
| 2650 #ifdef FEAT_MBYTE | |
| 2651 vim_free(smp->sm_lead_w); | |
| 2652 vim_free(smp->sm_oneof_w); | |
| 2653 vim_free(smp->sm_to_w); | |
| 2654 #endif | |
| 2655 } | |
| 344 | 2656 ga_clear(gap); |
| 323 | 2657 |
| 339 | 2658 for (i = 0; i < lp->sl_prefixcnt; ++i) |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4352
diff
changeset
|
2659 vim_regfree(lp->sl_prefprog[i]); |
| 376 | 2660 lp->sl_prefixcnt = 0; |
| 339 | 2661 vim_free(lp->sl_prefprog); |
| 376 | 2662 lp->sl_prefprog = NULL; |
| 2663 | |
| 714 | 2664 vim_free(lp->sl_info); |
| 2665 lp->sl_info = NULL; | |
| 2666 | |
| 376 | 2667 vim_free(lp->sl_midword); |
| 2668 lp->sl_midword = NULL; | |
| 339 | 2669 |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4352
diff
changeset
|
2670 vim_regfree(lp->sl_compprog); |
| 1762 | 2671 vim_free(lp->sl_comprules); |
| 483 | 2672 vim_free(lp->sl_compstartflags); |
| 485 | 2673 vim_free(lp->sl_compallflags); |
| 483 | 2674 lp->sl_compprog = NULL; |
| 1762 | 2675 lp->sl_comprules = NULL; |
| 483 | 2676 lp->sl_compstartflags = NULL; |
| 485 | 2677 lp->sl_compallflags = NULL; |
| 483 | 2678 |
| 2679 vim_free(lp->sl_syllable); | |
| 2680 lp->sl_syllable = NULL; | |
| 2681 ga_clear(&lp->sl_syl_items); | |
| 481 | 2682 |
| 809 | 2683 ga_clear_strings(&lp->sl_comppat); |
| 2684 | |
| 625 | 2685 hash_clear_all(&lp->sl_wordcount, WC_KEY_OFF); |
| 2686 hash_init(&lp->sl_wordcount); | |
| 2687 | |
| 2688 #ifdef FEAT_MBYTE | |
| 2689 hash_clear_all(&lp->sl_map_hash, 0); | |
| 2690 #endif | |
| 2691 | |
| 2692 /* Clear info from .sug file. */ | |
| 2693 slang_clear_sug(lp); | |
| 483 | 2694 |
| 2695 lp->sl_compmax = MAXWLEN; | |
| 501 | 2696 lp->sl_compminlen = 0; |
| 483 | 2697 lp->sl_compsylmax = MAXWLEN; |
| 2698 lp->sl_regions[0] = NUL; | |
| 236 | 2699 } |
| 2700 | |
| 2701 /* | |
| 625 | 2702 * Clear the info from the .sug file in "lp". |
| 2703 */ | |
| 2704 static void | |
| 2705 slang_clear_sug(lp) | |
| 2706 slang_T *lp; | |
| 2707 { | |
| 2708 vim_free(lp->sl_sbyts); | |
| 2709 lp->sl_sbyts = NULL; | |
| 2710 vim_free(lp->sl_sidxs); | |
| 2711 lp->sl_sidxs = NULL; | |
| 2712 close_spellbuf(lp->sl_sugbuf); | |
| 2713 lp->sl_sugbuf = NULL; | |
| 2714 lp->sl_sugloaded = FALSE; | |
| 2715 lp->sl_sugtime = 0; | |
| 2716 } | |
| 2717 | |
| 2718 /* | |
| 307 | 2719 * Load one spell file and store the info into a slang_T. |
| 236 | 2720 * Invoked through do_in_runtimepath(). |
| 2721 */ | |
| 2722 static void | |
| 310 | 2723 spell_load_cb(fname, cookie) |
| 236 | 2724 char_u *fname; |
| 501 | 2725 void *cookie; |
| 2726 { | |
| 2727 spelload_T *slp = (spelload_T *)cookie; | |
| 2728 slang_T *slang; | |
| 2729 | |
| 2730 slang = spell_load_file(fname, slp->sl_lang, NULL, FALSE); | |
| 2731 if (slang != NULL) | |
| 2732 { | |
| 2733 /* When a previously loaded file has NOBREAK also use it for the | |
| 2734 * ".add" files. */ | |
| 2735 if (slp->sl_nobreak && slang->sl_add) | |
| 2736 slang->sl_nobreak = TRUE; | |
| 2737 else if (slang->sl_nobreak) | |
| 2738 slp->sl_nobreak = TRUE; | |
| 2739 | |
| 2740 slp->sl_slang = slang; | |
| 2741 } | |
| 310 | 2742 } |
| 2743 | |
| 2744 /* | |
| 2745 * Load one spell file and store the info into a slang_T. | |
| 2746 * | |
| 625 | 2747 * This is invoked in three ways: |
| 310 | 2748 * - From spell_load_cb() to load a spell file for the first time. "lang" is |
| 2749 * the language name, "old_lp" is NULL. Will allocate an slang_T. | |
| 2750 * - To reload a spell file that was changed. "lang" is NULL and "old_lp" | |
| 2751 * points to the existing slang_T. | |
| 625 | 2752 * - Just after writing a .spl file; it's read back to produce the .sug file. |
| 626 | 2753 * "old_lp" is NULL and "lang" is NULL. Will allocate an slang_T. |
| 2754 * | |
| 323 | 2755 * Returns the slang_T the spell file was loaded into. NULL for error. |
| 310 | 2756 */ |
| 323 | 2757 static slang_T * |
| 2758 spell_load_file(fname, lang, old_lp, silent) | |
| 310 | 2759 char_u *fname; |
| 2760 char_u *lang; | |
| 2761 slang_T *old_lp; | |
| 323 | 2762 int silent; /* no error if file doesn't exist */ |
| 310 | 2763 { |
| 236 | 2764 FILE *fd; |
| 483 | 2765 char_u buf[VIMSPELLMAGICL]; |
| 236 | 2766 char_u *p; |
| 2767 int i; | |
| 339 | 2768 int n; |
| 300 | 2769 int len; |
| 236 | 2770 char_u *save_sourcing_name = sourcing_name; |
| 2771 linenr_T save_sourcing_lnum = sourcing_lnum; | |
| 307 | 2772 slang_T *lp = NULL; |
| 344 | 2773 int c = 0; |
| 483 | 2774 int res; |
| 236 | 2775 |
| 310 | 2776 fd = mch_fopen((char *)fname, "r"); |
| 236 | 2777 if (fd == NULL) |
| 2778 { | |
| 323 | 2779 if (!silent) |
| 2780 EMSG2(_(e_notopen), fname); | |
| 2781 else if (p_verbose > 2) | |
| 2782 { | |
| 2783 verbose_enter(); | |
| 2784 smsg((char_u *)e_notopen, fname); | |
| 2785 verbose_leave(); | |
| 2786 } | |
| 255 | 2787 goto endFAIL; |
| 236 | 2788 } |
| 310 | 2789 if (p_verbose > 2) |
| 2790 { | |
| 2791 verbose_enter(); | |
| 2792 smsg((char_u *)_("Reading spell file \"%s\""), fname); | |
| 2793 verbose_leave(); | |
| 2794 } | |
| 2795 | |
| 2796 if (old_lp == NULL) | |
| 2797 { | |
| 2798 lp = slang_alloc(lang); | |
| 2799 if (lp == NULL) | |
| 2800 goto endFAIL; | |
| 2801 | |
| 2802 /* Remember the file name, used to reload the file when it's updated. */ | |
| 2803 lp->sl_fname = vim_strsave(fname); | |
| 2804 if (lp->sl_fname == NULL) | |
| 2805 goto endFAIL; | |
| 2806 | |
| 2660 | 2807 /* Check for .add.spl (_add.spl for VMS). */ |
| 2808 lp->sl_add = strstr((char *)gettail(fname), SPL_FNAME_ADD) != NULL; | |
| 310 | 2809 } |
| 2810 else | |
| 2811 lp = old_lp; | |
| 307 | 2812 |
| 236 | 2813 /* Set sourcing_name, so that error messages mention the file name. */ |
| 2814 sourcing_name = fname; | |
| 2815 sourcing_lnum = 0; | |
| 2816 | |
| 625 | 2817 /* |
| 2818 * <HEADER>: <fileID> | |
| 339 | 2819 */ |
| 236 | 2820 for (i = 0; i < VIMSPELLMAGICL; ++i) |
| 2821 buf[i] = getc(fd); /* <fileID> */ | |
| 2822 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0) | |
| 2823 { | |
| 483 | 2824 EMSG(_("E757: This does not look like a spell file")); |
| 381 | 2825 goto endFAIL; |
| 483 | 2826 } |
| 2827 c = getc(fd); /* <versionnr> */ | |
| 2828 if (c < VIMSPELLVERSION) | |
| 2829 { | |
| 2830 EMSG(_("E771: Old spell file, needs to be updated")); | |
| 381 | 2831 goto endFAIL; |
| 2832 } | |
| 483 | 2833 else if (c > VIMSPELLVERSION) |
| 2834 { | |
| 2835 EMSG(_("E772: Spell file is for newer version of Vim")); | |
| 481 | 2836 goto endFAIL; |
| 483 | 2837 } |
| 2838 | |
| 2839 | |
| 2840 /* | |
| 2841 * <SECTIONS>: <section> ... <sectionend> | |
| 2842 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents) | |
| 2843 */ | |
| 2844 for (;;) | |
| 2845 { | |
| 2846 n = getc(fd); /* <sectionID> or <sectionend> */ | |
| 2847 if (n == SN_END) | |
| 2848 break; | |
| 2849 c = getc(fd); /* <sectionflags> */ | |
| 695 | 2850 len = get4c(fd); /* <sectionlen> */ |
| 483 | 2851 if (len < 0) |
| 2852 goto truncerr; | |
| 2853 | |
| 2854 res = 0; | |
| 2855 switch (n) | |
| 2856 { | |
| 714 | 2857 case SN_INFO: |
| 2858 lp->sl_info = read_string(fd, len); /* <infotext> */ | |
| 2859 if (lp->sl_info == NULL) | |
| 2860 goto endFAIL; | |
| 2861 break; | |
| 2862 | |
| 483 | 2863 case SN_REGION: |
| 2864 res = read_region_section(fd, lp, len); | |
| 2865 break; | |
| 2866 | |
| 2867 case SN_CHARFLAGS: | |
| 2868 res = read_charflags_section(fd); | |
| 2869 break; | |
| 2870 | |
| 2871 case SN_MIDWORD: | |
| 2872 lp->sl_midword = read_string(fd, len); /* <midword> */ | |
| 2873 if (lp->sl_midword == NULL) | |
| 2874 goto endFAIL; | |
| 2875 break; | |
| 2876 | |
| 2877 case SN_PREFCOND: | |
| 2878 res = read_prefcond_section(fd, lp); | |
| 2879 break; | |
| 2880 | |
| 2881 case SN_REP: | |
| 625 | 2882 res = read_rep_section(fd, &lp->sl_rep, lp->sl_rep_first); |
| 2883 break; | |
| 2884 | |
| 2885 case SN_REPSAL: | |
| 2886 res = read_rep_section(fd, &lp->sl_repsal, lp->sl_repsal_first); | |
| 483 | 2887 break; |
| 2888 | |
| 2889 case SN_SAL: | |
| 2890 res = read_sal_section(fd, lp); | |
| 2891 break; | |
| 2892 | |
| 2893 case SN_SOFO: | |
| 2894 res = read_sofo_section(fd, lp); | |
| 2895 break; | |
| 2896 | |
| 2897 case SN_MAP: | |
| 2898 p = read_string(fd, len); /* <mapstr> */ | |
| 2899 if (p == NULL) | |
| 2900 goto endFAIL; | |
| 2901 set_map_str(lp, p); | |
| 2902 vim_free(p); | |
| 2903 break; | |
| 2904 | |
| 625 | 2905 case SN_WORDS: |
| 2906 res = read_words_section(fd, lp, len); | |
| 2907 break; | |
| 2908 | |
| 2909 case SN_SUGFILE: | |
|
2229
d45902a5c61c
Fix a few more things for persistent undo.
Bram Moolenaar <bram@vim.org>
parents:
2214
diff
changeset
|
2910 lp->sl_sugtime = get8ctime(fd); /* <timestamp> */ |
| 625 | 2911 break; |
| 2912 | |
| 698 | 2913 case SN_NOSPLITSUGS: |
| 2914 lp->sl_nosplitsugs = TRUE; /* <timestamp> */ | |
| 2915 break; | |
| 2916 | |
| 483 | 2917 case SN_COMPOUND: |
| 2918 res = read_compound(fd, lp, len); | |
| 2919 break; | |
| 2920 | |
| 492 | 2921 case SN_NOBREAK: |
| 2922 lp->sl_nobreak = TRUE; | |
| 2923 break; | |
| 2924 | |
| 483 | 2925 case SN_SYLLABLE: |
| 2926 lp->sl_syllable = read_string(fd, len); /* <syllable> */ | |
| 2927 if (lp->sl_syllable == NULL) | |
| 2928 goto endFAIL; | |
| 2929 if (init_syl_tab(lp) == FAIL) | |
| 2930 goto endFAIL; | |
| 2931 break; | |
| 2932 | |
| 2933 default: | |
| 2934 /* Unsupported section. When it's required give an error | |
| 2935 * message. When it's not required skip the contents. */ | |
| 2936 if (c & SNF_REQUIRED) | |
| 2937 { | |
| 2938 EMSG(_("E770: Unsupported section in spell file")); | |
| 2939 goto endFAIL; | |
| 2940 } | |
| 2941 while (--len >= 0) | |
| 2942 if (getc(fd) < 0) | |
| 2943 goto truncerr; | |
| 2944 break; | |
| 2945 } | |
| 625 | 2946 someerror: |
| 483 | 2947 if (res == SP_FORMERROR) |
| 2948 { | |
| 2949 EMSG(_(e_format)); | |
| 381 | 2950 goto endFAIL; |
| 344 | 2951 } |
| 483 | 2952 if (res == SP_TRUNCERROR) |
| 2953 { | |
| 2954 truncerr: | |
| 2955 EMSG(_(e_spell_trunc)); | |
| 375 | 2956 goto endFAIL; |
| 483 | 2957 } |
| 495 | 2958 if (res == SP_OTHERERROR) |
| 483 | 2959 goto endFAIL; |
| 389 | 2960 } |
| 236 | 2961 |
| 625 | 2962 /* <LWORDTREE> */ |
| 2963 res = spell_read_tree(fd, &lp->sl_fbyts, &lp->sl_fidxs, FALSE, 0); | |
| 2964 if (res != 0) | |
| 2965 goto someerror; | |
| 2966 | |
| 2967 /* <KWORDTREE> */ | |
| 2968 res = spell_read_tree(fd, &lp->sl_kbyts, &lp->sl_kidxs, FALSE, 0); | |
| 2969 if (res != 0) | |
| 2970 goto someerror; | |
| 2971 | |
| 2972 /* <PREFIXTREE> */ | |
| 2973 res = spell_read_tree(fd, &lp->sl_pbyts, &lp->sl_pidxs, TRUE, | |
| 2974 lp->sl_prefixcnt); | |
| 2975 if (res != 0) | |
| 2976 goto someerror; | |
| 243 | 2977 |
| 310 | 2978 /* For a new file link it in the list of spell files. */ |
| 626 | 2979 if (old_lp == NULL && lang != NULL) |
| 310 | 2980 { |
| 2981 lp->sl_next = first_lang; | |
| 2982 first_lang = lp; | |
| 2983 } | |
| 307 | 2984 |
| 255 | 2985 goto endOK; |
| 2986 | |
| 2987 endFAIL: | |
| 310 | 2988 if (lang != NULL) |
| 2989 /* truncating the name signals the error to spell_load_lang() */ | |
| 2990 *lang = NUL; | |
| 2991 if (lp != NULL && old_lp == NULL) | |
| 307 | 2992 slang_free(lp); |
| 500 | 2993 lp = NULL; |
| 255 | 2994 |
| 2995 endOK: | |
| 236 | 2996 if (fd != NULL) |
| 2997 fclose(fd); | |
| 2998 sourcing_name = save_sourcing_name; | |
| 2999 sourcing_lnum = save_sourcing_lnum; | |
| 323 | 3000 |
| 3001 return lp; | |
| 236 | 3002 } |
| 3003 | |
| 376 | 3004 /* |
| 3005 * Read a length field from "fd" in "cnt_bytes" bytes. | |
| 381 | 3006 * Allocate memory, read the string into it and add a NUL at the end. |
| 376 | 3007 * Returns NULL when the count is zero. |
| 483 | 3008 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result |
| 3009 * otherwise. | |
| 376 | 3010 */ |
| 3011 static char_u * | |
| 389 | 3012 read_cnt_string(fd, cnt_bytes, cntp) |
| 376 | 3013 FILE *fd; |
| 3014 int cnt_bytes; | |
| 389 | 3015 int *cntp; |
| 376 | 3016 { |
| 3017 int cnt = 0; | |
| 3018 int i; | |
| 3019 char_u *str; | |
| 3020 | |
| 3021 /* read the length bytes, MSB first */ | |
| 3022 for (i = 0; i < cnt_bytes; ++i) | |
| 3023 cnt = (cnt << 8) + getc(fd); | |
| 3024 if (cnt < 0) | |
| 3025 { | |
| 483 | 3026 *cntp = SP_TRUNCERROR; |
| 376 | 3027 return NULL; |
| 3028 } | |
| 389 | 3029 *cntp = cnt; |
| 3030 if (cnt == 0) | |
| 3031 return NULL; /* nothing to read, return NULL */ | |
| 376 | 3032 |
| 483 | 3033 str = read_string(fd, cnt); |
| 3034 if (str == NULL) | |
| 495 | 3035 *cntp = SP_OTHERERROR; |
| 483 | 3036 return str; |
| 3037 } | |
| 3038 | |
| 3039 /* | |
| 3040 * Read SN_REGION: <regionname> ... | |
| 3041 * Return SP_*ERROR flags. | |
| 3042 */ | |
| 3043 static int | |
| 3044 read_region_section(fd, lp, len) | |
| 3045 FILE *fd; | |
| 3046 slang_T *lp; | |
| 3047 int len; | |
| 3048 { | |
| 3049 int i; | |
| 3050 | |
| 3051 if (len > 16) | |
| 3052 return SP_FORMERROR; | |
| 3053 for (i = 0; i < len; ++i) | |
| 3054 lp->sl_regions[i] = getc(fd); /* <regionname> */ | |
| 3055 lp->sl_regions[len] = NUL; | |
| 3056 return 0; | |
| 3057 } | |
| 3058 | |
| 3059 /* | |
| 3060 * Read SN_CHARFLAGS section: <charflagslen> <charflags> | |
| 3061 * <folcharslen> <folchars> | |
| 3062 * Return SP_*ERROR flags. | |
| 3063 */ | |
| 3064 static int | |
| 3065 read_charflags_section(fd) | |
| 3066 FILE *fd; | |
| 3067 { | |
| 3068 char_u *flags; | |
| 3069 char_u *fol; | |
| 3070 int flagslen, follen; | |
| 3071 | |
| 3072 /* <charflagslen> <charflags> */ | |
| 3073 flags = read_cnt_string(fd, 1, &flagslen); | |
| 3074 if (flagslen < 0) | |
| 3075 return flagslen; | |
| 3076 | |
| 3077 /* <folcharslen> <folchars> */ | |
| 3078 fol = read_cnt_string(fd, 2, &follen); | |
| 3079 if (follen < 0) | |
| 3080 { | |
| 3081 vim_free(flags); | |
| 3082 return follen; | |
| 3083 } | |
| 3084 | |
| 3085 /* Set the word-char flags and fill SPELL_ISUPPER() table. */ | |
| 3086 if (flags != NULL && fol != NULL) | |
| 3087 set_spell_charflags(flags, flagslen, fol); | |
| 3088 | |
| 3089 vim_free(flags); | |
| 3090 vim_free(fol); | |
| 3091 | |
| 3092 /* When <charflagslen> is zero then <fcharlen> must also be zero. */ | |
| 3093 if ((flags == NULL) != (fol == NULL)) | |
| 3094 return SP_FORMERROR; | |
| 3095 return 0; | |
| 3096 } | |
| 3097 | |
| 3098 /* | |
| 3099 * Read SN_PREFCOND section. | |
| 3100 * Return SP_*ERROR flags. | |
| 3101 */ | |
| 3102 static int | |
| 3103 read_prefcond_section(fd, lp) | |
| 3104 FILE *fd; | |
| 3105 slang_T *lp; | |
| 3106 { | |
| 3107 int cnt; | |
| 3108 int i; | |
| 3109 int n; | |
| 3110 char_u *p; | |
| 3111 char_u buf[MAXWLEN + 1]; | |
| 3112 | |
| 3113 /* <prefcondcnt> <prefcond> ... */ | |
| 695 | 3114 cnt = get2c(fd); /* <prefcondcnt> */ |
| 483 | 3115 if (cnt <= 0) |
| 3116 return SP_FORMERROR; | |
| 3117 | |
| 3118 lp->sl_prefprog = (regprog_T **)alloc_clear( | |
| 3119 (unsigned)sizeof(regprog_T *) * cnt); | |
| 3120 if (lp->sl_prefprog == NULL) | |
| 495 | 3121 return SP_OTHERERROR; |
| 483 | 3122 lp->sl_prefixcnt = cnt; |
| 3123 | |
| 376 | 3124 for (i = 0; i < cnt; ++i) |
| 483 | 3125 { |
| 3126 /* <prefcond> : <condlen> <condstr> */ | |
| 3127 n = getc(fd); /* <condlen> */ | |
| 3128 if (n < 0 || n >= MAXWLEN) | |
| 3129 return SP_FORMERROR; | |
| 3130 | |
| 3131 /* When <condlen> is zero we have an empty condition. Otherwise | |
| 3132 * compile the regexp program used to check for the condition. */ | |
| 3133 if (n > 0) | |
| 3134 { | |
| 3135 buf[0] = '^'; /* always match at one position only */ | |
| 3136 p = buf + 1; | |
| 3137 while (n-- > 0) | |
| 3138 *p++ = getc(fd); /* <condstr> */ | |
| 3139 *p = NUL; | |
| 3140 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING); | |
| 3141 } | |
| 3142 } | |
| 3143 return 0; | |
| 3144 } | |
| 3145 | |
| 3146 /* | |
| 625 | 3147 * Read REP or REPSAL items section from "fd": <repcount> <rep> ... |
| 483 | 3148 * Return SP_*ERROR flags. |
| 3149 */ | |
| 3150 static int | |
| 625 | 3151 read_rep_section(fd, gap, first) |
| 483 | 3152 FILE *fd; |
| 625 | 3153 garray_T *gap; |
| 3154 short *first; | |
| 483 | 3155 { |
| 3156 int cnt; | |
| 3157 fromto_T *ftp; | |
| 3158 int i; | |
| 3159 | |
| 695 | 3160 cnt = get2c(fd); /* <repcount> */ |
| 483 | 3161 if (cnt < 0) |
| 3162 return SP_TRUNCERROR; | |
| 3163 | |
| 3164 if (ga_grow(gap, cnt) == FAIL) | |
| 495 | 3165 return SP_OTHERERROR; |
| 483 | 3166 |
| 3167 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */ | |
| 3168 for (; gap->ga_len < cnt; ++gap->ga_len) | |
| 3169 { | |
| 3170 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len]; | |
| 3171 ftp->ft_from = read_cnt_string(fd, 1, &i); | |
| 3172 if (i < 0) | |
| 3173 return i; | |
| 3174 if (i == 0) | |
| 3175 return SP_FORMERROR; | |
| 3176 ftp->ft_to = read_cnt_string(fd, 1, &i); | |
| 3177 if (i <= 0) | |
| 3178 { | |
| 3179 vim_free(ftp->ft_from); | |
| 3180 if (i < 0) | |
| 3181 return i; | |
| 3182 return SP_FORMERROR; | |
| 3183 } | |
| 3184 } | |
| 3185 | |
| 3186 /* Fill the first-index table. */ | |
| 3187 for (i = 0; i < 256; ++i) | |
| 3188 first[i] = -1; | |
| 3189 for (i = 0; i < gap->ga_len; ++i) | |
| 3190 { | |
| 3191 ftp = &((fromto_T *)gap->ga_data)[i]; | |
| 3192 if (first[*ftp->ft_from] == -1) | |
| 3193 first[*ftp->ft_from] = i; | |
| 3194 } | |
| 3195 return 0; | |
| 3196 } | |
| 3197 | |
| 3198 /* | |
| 3199 * Read SN_SAL section: <salflags> <salcount> <sal> ... | |
| 3200 * Return SP_*ERROR flags. | |
| 3201 */ | |
| 3202 static int | |
| 3203 read_sal_section(fd, slang) | |
| 3204 FILE *fd; | |
| 3205 slang_T *slang; | |
| 3206 { | |
| 3207 int i; | |
| 3208 int cnt; | |
| 3209 garray_T *gap; | |
| 3210 salitem_T *smp; | |
| 3211 int ccnt; | |
| 3212 char_u *p; | |
| 485 | 3213 int c = NUL; |
| 483 | 3214 |
| 3215 slang->sl_sofo = FALSE; | |
| 3216 | |
| 3217 i = getc(fd); /* <salflags> */ | |
| 3218 if (i & SAL_F0LLOWUP) | |
| 3219 slang->sl_followup = TRUE; | |
| 3220 if (i & SAL_COLLAPSE) | |
| 3221 slang->sl_collapse = TRUE; | |
| 3222 if (i & SAL_REM_ACCENTS) | |
| 3223 slang->sl_rem_accents = TRUE; | |
| 3224 | |
| 695 | 3225 cnt = get2c(fd); /* <salcount> */ |
| 483 | 3226 if (cnt < 0) |
| 3227 return SP_TRUNCERROR; | |
| 3228 | |
| 3229 gap = &slang->sl_sal; | |
| 3230 ga_init2(gap, sizeof(salitem_T), 10); | |
| 548 | 3231 if (ga_grow(gap, cnt + 1) == FAIL) |
| 495 | 3232 return SP_OTHERERROR; |
| 483 | 3233 |
| 3234 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */ | |
| 3235 for (; gap->ga_len < cnt; ++gap->ga_len) | |
| 3236 { | |
| 3237 smp = &((salitem_T *)gap->ga_data)[gap->ga_len]; | |
| 3238 ccnt = getc(fd); /* <salfromlen> */ | |
| 3239 if (ccnt < 0) | |
| 3240 return SP_TRUNCERROR; | |
| 3241 if ((p = alloc(ccnt + 2)) == NULL) | |
| 495 | 3242 return SP_OTHERERROR; |
| 483 | 3243 smp->sm_lead = p; |
| 3244 | |
| 3245 /* Read up to the first special char into sm_lead. */ | |
| 3246 for (i = 0; i < ccnt; ++i) | |
| 3247 { | |
| 3248 c = getc(fd); /* <salfrom> */ | |
| 3249 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL) | |
| 3250 break; | |
| 3251 *p++ = c; | |
| 3252 } | |
| 835 | 3253 smp->sm_leadlen = (int)(p - smp->sm_lead); |
| 483 | 3254 *p++ = NUL; |
| 3255 | |
| 3256 /* Put (abc) chars in sm_oneof, if any. */ | |
| 3257 if (c == '(') | |
| 3258 { | |
| 3259 smp->sm_oneof = p; | |
| 3260 for (++i; i < ccnt; ++i) | |
| 3261 { | |
| 3262 c = getc(fd); /* <salfrom> */ | |
| 3263 if (c == ')') | |
| 3264 break; | |
| 3265 *p++ = c; | |
| 3266 } | |
| 3267 *p++ = NUL; | |
| 3268 if (++i < ccnt) | |
| 3269 c = getc(fd); | |
| 3270 } | |
| 3271 else | |
| 3272 smp->sm_oneof = NULL; | |
| 3273 | |
| 3274 /* Any following chars go in sm_rules. */ | |
| 3275 smp->sm_rules = p; | |
| 3276 if (i < ccnt) | |
| 3277 /* store the char we got while checking for end of sm_lead */ | |
| 3278 *p++ = c; | |
| 3279 for (++i; i < ccnt; ++i) | |
| 3280 *p++ = getc(fd); /* <salfrom> */ | |
| 3281 *p++ = NUL; | |
| 3282 | |
| 3283 /* <saltolen> <salto> */ | |
| 3284 smp->sm_to = read_cnt_string(fd, 1, &ccnt); | |
| 3285 if (ccnt < 0) | |
| 3286 { | |
| 3287 vim_free(smp->sm_lead); | |
| 3288 return ccnt; | |
| 3289 } | |
| 3290 | |
| 3291 #ifdef FEAT_MBYTE | |
| 3292 if (has_mbyte) | |
| 3293 { | |
| 3294 /* convert the multi-byte strings to wide char strings */ | |
| 3295 smp->sm_lead_w = mb_str2wide(smp->sm_lead); | |
| 3296 smp->sm_leadlen = mb_charlen(smp->sm_lead); | |
| 3297 if (smp->sm_oneof == NULL) | |
| 3298 smp->sm_oneof_w = NULL; | |
| 3299 else | |
| 3300 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof); | |
| 3301 if (smp->sm_to == NULL) | |
| 3302 smp->sm_to_w = NULL; | |
| 3303 else | |
| 3304 smp->sm_to_w = mb_str2wide(smp->sm_to); | |
| 3305 if (smp->sm_lead_w == NULL | |
| 3306 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL) | |
| 3307 || (smp->sm_to_w == NULL && smp->sm_to != NULL)) | |
| 3308 { | |
| 3309 vim_free(smp->sm_lead); | |
| 3310 vim_free(smp->sm_to); | |
| 3311 vim_free(smp->sm_lead_w); | |
| 3312 vim_free(smp->sm_oneof_w); | |
| 3313 vim_free(smp->sm_to_w); | |
| 495 | 3314 return SP_OTHERERROR; |
| 483 | 3315 } |
| 3316 } | |
| 3317 #endif | |
| 3318 } | |
| 3319 | |
| 548 | 3320 if (gap->ga_len > 0) |
| 3321 { | |
| 3322 /* Add one extra entry to mark the end with an empty sm_lead. Avoids | |
| 3323 * that we need to check the index every time. */ | |
| 3324 smp = &((salitem_T *)gap->ga_data)[gap->ga_len]; | |
| 3325 if ((p = alloc(1)) == NULL) | |
| 3326 return SP_OTHERERROR; | |
| 3327 p[0] = NUL; | |
| 3328 smp->sm_lead = p; | |
| 3329 smp->sm_leadlen = 0; | |
| 3330 smp->sm_oneof = NULL; | |
| 3331 smp->sm_rules = p; | |
| 3332 smp->sm_to = NULL; | |
| 3333 #ifdef FEAT_MBYTE | |
| 3334 if (has_mbyte) | |
| 3335 { | |
| 3336 smp->sm_lead_w = mb_str2wide(smp->sm_lead); | |
| 3337 smp->sm_leadlen = 0; | |
| 3338 smp->sm_oneof_w = NULL; | |
| 3339 smp->sm_to_w = NULL; | |
| 3340 } | |
| 3341 #endif | |
| 3342 ++gap->ga_len; | |
| 3343 } | |
| 3344 | |
| 483 | 3345 /* Fill the first-index table. */ |
| 3346 set_sal_first(slang); | |
| 3347 | |
| 3348 return 0; | |
| 3349 } | |
| 3350 | |
| 3351 /* | |
| 625 | 3352 * Read SN_WORDS: <word> ... |
| 3353 * Return SP_*ERROR flags. | |
| 3354 */ | |
| 3355 static int | |
| 3356 read_words_section(fd, lp, len) | |
| 3357 FILE *fd; | |
| 3358 slang_T *lp; | |
| 3359 int len; | |
| 3360 { | |
| 3361 int done = 0; | |
| 3362 int i; | |
| 1619 | 3363 int c; |
| 625 | 3364 char_u word[MAXWLEN]; |
| 3365 | |
| 3366 while (done < len) | |
| 3367 { | |
| 3368 /* Read one word at a time. */ | |
| 3369 for (i = 0; ; ++i) | |
| 3370 { | |
| 1619 | 3371 c = getc(fd); |
| 3372 if (c == EOF) | |
| 3373 return SP_TRUNCERROR; | |
| 3374 word[i] = c; | |
| 625 | 3375 if (word[i] == NUL) |
| 3376 break; | |
| 3377 if (i == MAXWLEN - 1) | |
| 3378 return SP_FORMERROR; | |
| 3379 } | |
| 3380 | |
| 3381 /* Init the count to 10. */ | |
| 3382 count_common_word(lp, word, -1, 10); | |
| 3383 done += i + 1; | |
| 3384 } | |
| 3385 return 0; | |
| 3386 } | |
| 3387 | |
| 3388 /* | |
| 3389 * Add a word to the hashtable of common words. | |
| 3390 * If it's already there then the counter is increased. | |
| 3391 */ | |
| 3392 static void | |
| 3393 count_common_word(lp, word, len, count) | |
| 3394 slang_T *lp; | |
| 3395 char_u *word; | |
| 3396 int len; /* word length, -1 for upto NUL */ | |
| 3397 int count; /* 1 to count once, 10 to init */ | |
| 3398 { | |
| 3399 hash_T hash; | |
| 3400 hashitem_T *hi; | |
| 3401 wordcount_T *wc; | |
| 3402 char_u buf[MAXWLEN]; | |
| 3403 char_u *p; | |
| 3404 | |
| 3405 if (len == -1) | |
| 3406 p = word; | |
| 3407 else | |
| 3408 { | |
| 3409 vim_strncpy(buf, word, len); | |
| 3410 p = buf; | |
| 3411 } | |
| 3412 | |
| 3413 hash = hash_hash(p); | |
| 3414 hi = hash_lookup(&lp->sl_wordcount, p, hash); | |
| 3415 if (HASHITEM_EMPTY(hi)) | |
| 3416 { | |
| 835 | 3417 wc = (wordcount_T *)alloc((unsigned)(sizeof(wordcount_T) + STRLEN(p))); |
| 625 | 3418 if (wc == NULL) |
| 3419 return; | |
| 3420 STRCPY(wc->wc_word, p); | |
| 3421 wc->wc_count = count; | |
| 3422 hash_add_item(&lp->sl_wordcount, hi, wc->wc_word, hash); | |
| 3423 } | |
| 3424 else | |
| 3425 { | |
| 3426 wc = HI2WC(hi); | |
| 3427 if ((wc->wc_count += count) < (unsigned)count) /* check for overflow */ | |
| 3428 wc->wc_count = MAXWORDCOUNT; | |
| 3429 } | |
| 3430 } | |
| 3431 | |
| 3432 /* | |
| 3433 * Adjust the score of common words. | |
| 3434 */ | |
| 3435 static int | |
| 3436 score_wordcount_adj(slang, score, word, split) | |
| 3437 slang_T *slang; | |
| 3438 int score; | |
| 3439 char_u *word; | |
| 3440 int split; /* word was split, less bonus */ | |
| 3441 { | |
| 3442 hashitem_T *hi; | |
| 3443 wordcount_T *wc; | |
| 3444 int bonus; | |
| 3445 int newscore; | |
| 3446 | |
| 3447 hi = hash_find(&slang->sl_wordcount, word); | |
| 3448 if (!HASHITEM_EMPTY(hi)) | |
| 3449 { | |
| 3450 wc = HI2WC(hi); | |
| 3451 if (wc->wc_count < SCORE_THRES2) | |
| 3452 bonus = SCORE_COMMON1; | |
| 3453 else if (wc->wc_count < SCORE_THRES3) | |
| 3454 bonus = SCORE_COMMON2; | |
| 3455 else | |
| 3456 bonus = SCORE_COMMON3; | |
| 3457 if (split) | |
| 3458 newscore = score - bonus / 2; | |
| 3459 else | |
| 3460 newscore = score - bonus; | |
| 3461 if (newscore < 0) | |
| 3462 return 0; | |
| 3463 return newscore; | |
| 3464 } | |
| 3465 return score; | |
| 3466 } | |
| 3467 | |
| 3468 /* | |
| 483 | 3469 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto> |
| 3470 * Return SP_*ERROR flags. | |
| 3471 */ | |
| 3472 static int | |
| 3473 read_sofo_section(fd, slang) | |
| 3474 FILE *fd; | |
| 3475 slang_T *slang; | |
| 3476 { | |
| 3477 int cnt; | |
| 3478 char_u *from, *to; | |
| 3479 int res; | |
| 3480 | |
| 3481 slang->sl_sofo = TRUE; | |
| 3482 | |
| 3483 /* <sofofromlen> <sofofrom> */ | |
| 3484 from = read_cnt_string(fd, 2, &cnt); | |
| 3485 if (cnt < 0) | |
| 3486 return cnt; | |
| 3487 | |
| 3488 /* <sofotolen> <sofoto> */ | |
| 3489 to = read_cnt_string(fd, 2, &cnt); | |
| 3490 if (cnt < 0) | |
| 3491 { | |
| 3492 vim_free(from); | |
| 3493 return cnt; | |
| 3494 } | |
| 3495 | |
| 3496 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */ | |
| 3497 if (from != NULL && to != NULL) | |
| 3498 res = set_sofo(slang, from, to); | |
| 3499 else if (from != NULL || to != NULL) | |
| 3500 res = SP_FORMERROR; /* only one of two strings is an error */ | |
| 3501 else | |
| 3502 res = 0; | |
| 3503 | |
| 3504 vim_free(from); | |
| 3505 vim_free(to); | |
| 3506 return res; | |
| 3507 } | |
| 3508 | |
| 3509 /* | |
| 3510 * Read the compound section from the .spl file: | |
| 809 | 3511 * <compmax> <compminlen> <compsylmax> <compoptions> <compflags> |
| 483 | 3512 * Returns SP_*ERROR flags. |
| 3513 */ | |
| 3514 static int | |
| 3515 read_compound(fd, slang, len) | |
| 3516 FILE *fd; | |
| 3517 slang_T *slang; | |
| 3518 int len; | |
| 3519 { | |
| 3520 int todo = len; | |
| 3521 int c; | |
| 3522 int atstart; | |
| 3523 char_u *pat; | |
| 3524 char_u *pp; | |
| 3525 char_u *cp; | |
| 485 | 3526 char_u *ap; |
| 1762 | 3527 char_u *crp; |
| 809 | 3528 int cnt; |
| 3529 garray_T *gap; | |
| 483 | 3530 |
| 3531 if (todo < 2) | |
| 3532 return SP_FORMERROR; /* need at least two bytes */ | |
| 3533 | |
| 3534 --todo; | |
| 3535 c = getc(fd); /* <compmax> */ | |
| 3536 if (c < 2) | |
| 3537 c = MAXWLEN; | |
| 3538 slang->sl_compmax = c; | |
| 3539 | |
| 3540 --todo; | |
| 3541 c = getc(fd); /* <compminlen> */ | |
| 3542 if (c < 1) | |
| 501 | 3543 c = 0; |
| 483 | 3544 slang->sl_compminlen = c; |
| 3545 | |
| 3546 --todo; | |
| 3547 c = getc(fd); /* <compsylmax> */ | |
| 3548 if (c < 1) | |
| 3549 c = MAXWLEN; | |
| 3550 slang->sl_compsylmax = c; | |
| 3551 | |
| 809 | 3552 c = getc(fd); /* <compoptions> */ |
| 3553 if (c != 0) | |
| 3554 ungetc(c, fd); /* be backwards compatible with Vim 7.0b */ | |
| 3555 else | |
| 3556 { | |
| 3557 --todo; | |
| 3558 c = getc(fd); /* only use the lower byte for now */ | |
| 3559 --todo; | |
| 3560 slang->sl_compoptions = c; | |
| 3561 | |
| 3562 gap = &slang->sl_comppat; | |
| 3563 c = get2c(fd); /* <comppatcount> */ | |
| 3564 todo -= 2; | |
| 3565 ga_init2(gap, sizeof(char_u *), c); | |
| 3566 if (ga_grow(gap, c) == OK) | |
| 3567 while (--c >= 0) | |
| 3568 { | |
| 3569 ((char_u **)(gap->ga_data))[gap->ga_len++] = | |
| 3570 read_cnt_string(fd, 1, &cnt); | |
| 3571 /* <comppatlen> <comppattext> */ | |
| 3572 if (cnt < 0) | |
| 3573 return cnt; | |
| 822 | 3574 todo -= cnt + 1; |
| 3575 } | |
| 3576 } | |
| 3577 if (todo < 0) | |
| 3578 return SP_FORMERROR; | |
| 809 | 3579 |
| 714 | 3580 /* Turn the COMPOUNDRULE items into a regexp pattern: |
| 483 | 3581 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$". |
| 495 | 3582 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes. |
| 3583 * Conversion to utf-8 may double the size. */ | |
| 3584 c = todo * 2 + 7; | |
| 3585 #ifdef FEAT_MBYTE | |
| 3586 if (enc_utf8) | |
| 3587 c += todo * 2; | |
| 3588 #endif | |
| 3589 pat = alloc((unsigned)c); | |
| 483 | 3590 if (pat == NULL) |
| 495 | 3591 return SP_OTHERERROR; |
| 483 | 3592 |
| 485 | 3593 /* We also need a list of all flags that can appear at the start and one |
| 3594 * for all flags. */ | |
| 483 | 3595 cp = alloc(todo + 1); |
| 3596 if (cp == NULL) | |
| 3597 { | |
| 3598 vim_free(pat); | |
| 495 | 3599 return SP_OTHERERROR; |
| 483 | 3600 } |
| 3601 slang->sl_compstartflags = cp; | |
| 3602 *cp = NUL; | |
| 3603 | |
| 485 | 3604 ap = alloc(todo + 1); |
| 3605 if (ap == NULL) | |
| 3606 { | |
| 3607 vim_free(pat); | |
| 495 | 3608 return SP_OTHERERROR; |
| 485 | 3609 } |
| 3610 slang->sl_compallflags = ap; | |
| 3611 *ap = NUL; | |
| 3612 | |
| 1762 | 3613 /* And a list of all patterns in their original form, for checking whether |
| 3614 * compounding may work in match_compoundrule(). This is freed when we | |
| 3615 * encounter a wildcard, the check doesn't work then. */ | |
| 3616 crp = alloc(todo + 1); | |
| 3617 slang->sl_comprules = crp; | |
| 3618 | |
| 483 | 3619 pp = pat; |
| 3620 *pp++ = '^'; | |
| 3621 *pp++ = '\\'; | |
| 3622 *pp++ = '('; | |
| 3623 | |
| 3624 atstart = 1; | |
| 3625 while (todo-- > 0) | |
| 3626 { | |
| 3627 c = getc(fd); /* <compflags> */ | |
| 1619 | 3628 if (c == EOF) |
| 3629 { | |
| 3630 vim_free(pat); | |
| 3631 return SP_TRUNCERROR; | |
| 3632 } | |
| 485 | 3633 |
| 3634 /* Add all flags to "sl_compallflags". */ | |
| 2691 | 3635 if (vim_strchr((char_u *)"?*+[]/", c) == NULL |
| 495 | 3636 && !byte_in_str(slang->sl_compallflags, c)) |
| 485 | 3637 { |
| 3638 *ap++ = c; | |
| 3639 *ap = NUL; | |
| 3640 } | |
| 3641 | |
| 483 | 3642 if (atstart != 0) |
| 3643 { | |
| 3644 /* At start of item: copy flags to "sl_compstartflags". For a | |
| 3645 * [abc] item set "atstart" to 2 and copy up to the ']'. */ | |
| 3646 if (c == '[') | |
| 3647 atstart = 2; | |
| 3648 else if (c == ']') | |
| 3649 atstart = 0; | |
| 3650 else | |
| 3651 { | |
| 495 | 3652 if (!byte_in_str(slang->sl_compstartflags, c)) |
| 483 | 3653 { |
| 3654 *cp++ = c; | |
| 3655 *cp = NUL; | |
| 3656 } | |
| 3657 if (atstart == 1) | |
| 3658 atstart = 0; | |
| 3659 } | |
| 3660 } | |
| 1762 | 3661 |
| 3662 /* Copy flag to "sl_comprules", unless we run into a wildcard. */ | |
| 3663 if (crp != NULL) | |
| 3664 { | |
| 2691 | 3665 if (c == '?' || c == '+' || c == '*') |
| 1762 | 3666 { |
| 3667 vim_free(slang->sl_comprules); | |
| 3668 slang->sl_comprules = NULL; | |
| 3669 crp = NULL; | |
| 3670 } | |
| 3671 else | |
| 3672 *crp++ = c; | |
| 3673 } | |
| 3674 | |
| 483 | 3675 if (c == '/') /* slash separates two items */ |
| 3676 { | |
| 3677 *pp++ = '\\'; | |
| 3678 *pp++ = '|'; | |
| 3679 atstart = 1; | |
| 3680 } | |
| 3681 else /* normal char, "[abc]" and '*' are copied as-is */ | |
| 3682 { | |
| 2691 | 3683 if (c == '?' || c == '+' || c == '~') |
| 3684 *pp++ = '\\'; /* "a?" becomes "a\?", "a+" becomes "a\+" */ | |
| 495 | 3685 #ifdef FEAT_MBYTE |
| 3686 if (enc_utf8) | |
| 3687 pp += mb_char2bytes(c, pp); | |
| 3688 else | |
| 3689 #endif | |
| 3690 *pp++ = c; | |
| 483 | 3691 } |
| 3692 } | |
| 3693 | |
| 3694 *pp++ = '\\'; | |
| 3695 *pp++ = ')'; | |
| 3696 *pp++ = '$'; | |
| 3697 *pp = NUL; | |
| 3698 | |
| 1762 | 3699 if (crp != NULL) |
| 3700 *crp = NUL; | |
| 3701 | |
| 483 | 3702 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT); |
| 3703 vim_free(pat); | |
| 3704 if (slang->sl_compprog == NULL) | |
| 3705 return SP_FORMERROR; | |
| 3706 | |
| 3707 return 0; | |
| 3708 } | |
| 3709 | |
| 495 | 3710 /* |
| 498 | 3711 * Return TRUE if byte "n" appears in "str". |
| 495 | 3712 * Like strchr() but independent of locale. |
| 3713 */ | |
| 3714 static int | |
| 498 | 3715 byte_in_str(str, n) |
| 495 | 3716 char_u *str; |
| 498 | 3717 int n; |
| 495 | 3718 { |
| 3719 char_u *p; | |
| 3720 | |
| 3721 for (p = str; *p != NUL; ++p) | |
| 498 | 3722 if (*p == n) |
| 495 | 3723 return TRUE; |
| 3724 return FALSE; | |
| 3725 } | |
| 3726 | |
| 483 | 3727 #define SY_MAXLEN 30 |
| 3728 typedef struct syl_item_S | |
| 3729 { | |
| 3730 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */ | |
| 3731 int sy_len; | |
| 3732 } syl_item_T; | |
| 3733 | |
| 3734 /* | |
| 3735 * Truncate "slang->sl_syllable" at the first slash and put the following items | |
| 3736 * in "slang->sl_syl_items". | |
| 3737 */ | |
| 3738 static int | |
| 3739 init_syl_tab(slang) | |
| 3740 slang_T *slang; | |
| 3741 { | |
| 3742 char_u *p; | |
| 3743 char_u *s; | |
| 3744 int l; | |
| 3745 syl_item_T *syl; | |
| 3746 | |
| 3747 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4); | |
| 3748 p = vim_strchr(slang->sl_syllable, '/'); | |
| 3749 while (p != NULL) | |
| 3750 { | |
| 3751 *p++ = NUL; | |
| 495 | 3752 if (*p == NUL) /* trailing slash */ |
| 483 | 3753 break; |
| 3754 s = p; | |
| 3755 p = vim_strchr(p, '/'); | |
| 3756 if (p == NULL) | |
| 835 | 3757 l = (int)STRLEN(s); |
| 483 | 3758 else |
| 835 | 3759 l = (int)(p - s); |
| 483 | 3760 if (l >= SY_MAXLEN) |
| 3761 return SP_FORMERROR; | |
| 3762 if (ga_grow(&slang->sl_syl_items, 1) == FAIL) | |
| 495 | 3763 return SP_OTHERERROR; |
| 483 | 3764 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) |
| 3765 + slang->sl_syl_items.ga_len++; | |
| 3766 vim_strncpy(syl->sy_chars, s, l); | |
| 3767 syl->sy_len = l; | |
| 3768 } | |
| 3769 return OK; | |
| 3770 } | |
| 3771 | |
| 3772 /* | |
| 3773 * Count the number of syllables in "word". | |
| 3774 * When "word" contains spaces the syllables after the last space are counted. | |
| 3775 * Returns zero if syllables are not defines. | |
| 3776 */ | |
| 3777 static int | |
| 3778 count_syllables(slang, word) | |
| 3779 slang_T *slang; | |
| 3780 char_u *word; | |
| 3781 { | |
| 3782 int cnt = 0; | |
| 3783 int skip = FALSE; | |
| 3784 char_u *p; | |
| 3785 int len; | |
| 3786 int i; | |
| 3787 syl_item_T *syl; | |
| 3788 int c; | |
| 3789 | |
| 3790 if (slang->sl_syllable == NULL) | |
| 3791 return 0; | |
| 3792 | |
| 3793 for (p = word; *p != NUL; p += len) | |
| 3794 { | |
| 3795 /* When running into a space reset counter. */ | |
| 3796 if (*p == ' ') | |
| 3797 { | |
| 3798 len = 1; | |
| 3799 cnt = 0; | |
| 3800 continue; | |
| 3801 } | |
| 3802 | |
| 3803 /* Find longest match of syllable items. */ | |
| 3804 len = 0; | |
| 3805 for (i = 0; i < slang->sl_syl_items.ga_len; ++i) | |
| 3806 { | |
| 3807 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i; | |
| 3808 if (syl->sy_len > len | |
| 3809 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0) | |
| 3810 len = syl->sy_len; | |
| 3811 } | |
| 3812 if (len != 0) /* found a match, count syllable */ | |
| 3813 { | |
| 3814 ++cnt; | |
| 3815 skip = FALSE; | |
| 3816 } | |
| 3817 else | |
| 3818 { | |
| 3819 /* No recognized syllable item, at least a syllable char then? */ | |
| 3820 #ifdef FEAT_MBYTE | |
| 3821 c = mb_ptr2char(p); | |
| 3822 len = (*mb_ptr2len)(p); | |
| 3823 #else | |
| 3824 c = *p; | |
| 3825 len = 1; | |
| 3826 #endif | |
| 3827 if (vim_strchr(slang->sl_syllable, c) == NULL) | |
| 3828 skip = FALSE; /* No, search for next syllable */ | |
| 3829 else if (!skip) | |
| 3830 { | |
| 3831 ++cnt; /* Yes, count it */ | |
| 3832 skip = TRUE; /* don't count following syllable chars */ | |
| 3833 } | |
| 3834 } | |
| 3835 } | |
| 3836 return cnt; | |
| 376 | 3837 } |
| 3838 | |
| 381 | 3839 /* |
| 3840 * Set the SOFOFROM and SOFOTO items in language "lp". | |
| 483 | 3841 * Returns SP_*ERROR flags when there is something wrong. |
| 381 | 3842 */ |
| 3843 static int | |
| 3844 set_sofo(lp, from, to) | |
| 3845 slang_T *lp; | |
| 3846 char_u *from; | |
| 3847 char_u *to; | |
| 3848 { | |
| 3849 int i; | |
| 3850 | |
| 3851 #ifdef FEAT_MBYTE | |
| 3852 garray_T *gap; | |
| 3853 char_u *s; | |
| 3854 char_u *p; | |
| 3855 int c; | |
| 3856 int *inp; | |
| 3857 | |
| 3858 if (has_mbyte) | |
| 3859 { | |
| 3860 /* Use "sl_sal" as an array with 256 pointers to a list of wide | |
| 3861 * characters. The index is the low byte of the character. | |
| 3862 * The list contains from-to pairs with a terminating NUL. | |
| 3863 * sl_sal_first[] is used for latin1 "from" characters. */ | |
| 3864 gap = &lp->sl_sal; | |
| 3865 ga_init2(gap, sizeof(int *), 1); | |
| 3866 if (ga_grow(gap, 256) == FAIL) | |
| 495 | 3867 return SP_OTHERERROR; |
| 381 | 3868 vim_memset(gap->ga_data, 0, sizeof(int *) * 256); |
| 3869 gap->ga_len = 256; | |
| 3870 | |
| 3871 /* First count the number of items for each list. Temporarily use | |
| 3872 * sl_sal_first[] for this. */ | |
| 3873 for (p = from, s = to; *p != NUL && *s != NUL; ) | |
| 3874 { | |
| 474 | 3875 c = mb_cptr2char_adv(&p); |
| 3876 mb_cptr_adv(s); | |
| 381 | 3877 if (c >= 256) |
| 3878 ++lp->sl_sal_first[c & 0xff]; | |
| 3879 } | |
| 3880 if (*p != NUL || *s != NUL) /* lengths differ */ | |
| 483 | 3881 return SP_FORMERROR; |
| 381 | 3882 |
| 3883 /* Allocate the lists. */ | |
| 3884 for (i = 0; i < 256; ++i) | |
| 3885 if (lp->sl_sal_first[i] > 0) | |
| 3886 { | |
| 3887 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1)); | |
| 3888 if (p == NULL) | |
| 495 | 3889 return SP_OTHERERROR; |
| 381 | 3890 ((int **)gap->ga_data)[i] = (int *)p; |
| 3891 *(int *)p = 0; | |
| 3892 } | |
| 3893 | |
| 3894 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal | |
| 3895 * list. */ | |
| 3896 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256); | |
| 3897 for (p = from, s = to; *p != NUL && *s != NUL; ) | |
| 3898 { | |
| 474 | 3899 c = mb_cptr2char_adv(&p); |
| 3900 i = mb_cptr2char_adv(&s); | |
| 381 | 3901 if (c >= 256) |
| 3902 { | |
| 3903 /* Append the from-to chars at the end of the list with | |
| 3904 * the low byte. */ | |
| 3905 inp = ((int **)gap->ga_data)[c & 0xff]; | |
| 3906 while (*inp != 0) | |
| 3907 ++inp; | |
| 3908 *inp++ = c; /* from char */ | |
| 3909 *inp++ = i; /* to char */ | |
| 3910 *inp++ = NUL; /* NUL at the end */ | |
| 3911 } | |
| 3912 else | |
| 3913 /* mapping byte to char is done in sl_sal_first[] */ | |
| 3914 lp->sl_sal_first[c] = i; | |
| 3915 } | |
| 3916 } | |
| 3917 else | |
| 3918 #endif | |
| 3919 { | |
| 3920 /* mapping bytes to bytes is done in sl_sal_first[] */ | |
| 3921 if (STRLEN(from) != STRLEN(to)) | |
| 483 | 3922 return SP_FORMERROR; |
| 381 | 3923 |
| 3924 for (i = 0; to[i] != NUL; ++i) | |
| 3925 lp->sl_sal_first[from[i]] = to[i]; | |
| 3926 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */ | |
| 3927 } | |
| 3928 | |
| 483 | 3929 return 0; |
| 381 | 3930 } |
| 3931 | |
| 3932 /* | |
| 3933 * Fill the first-index table for "lp". | |
| 3934 */ | |
| 3935 static void | |
| 3936 set_sal_first(lp) | |
| 3937 slang_T *lp; | |
| 3938 { | |
| 3939 salfirst_T *sfirst; | |
| 3940 int i; | |
| 3941 salitem_T *smp; | |
| 3942 int c; | |
| 3943 garray_T *gap = &lp->sl_sal; | |
| 3944 | |
| 3945 sfirst = lp->sl_sal_first; | |
| 3946 for (i = 0; i < 256; ++i) | |
| 3947 sfirst[i] = -1; | |
| 3948 smp = (salitem_T *)gap->ga_data; | |
| 3949 for (i = 0; i < gap->ga_len; ++i) | |
| 3950 { | |
| 3951 #ifdef FEAT_MBYTE | |
| 3952 if (has_mbyte) | |
| 3953 /* Use the lowest byte of the first character. For latin1 it's | |
| 3954 * the character, for other encodings it should differ for most | |
| 3955 * characters. */ | |
| 3956 c = *smp[i].sm_lead_w & 0xff; | |
| 3957 else | |
| 3958 #endif | |
| 3959 c = *smp[i].sm_lead; | |
| 3960 if (sfirst[c] == -1) | |
| 3961 { | |
| 3962 sfirst[c] = i; | |
| 3963 #ifdef FEAT_MBYTE | |
| 3964 if (has_mbyte) | |
| 3965 { | |
| 3966 int n; | |
| 3967 | |
| 3968 /* Make sure all entries with this byte are following each | |
| 3969 * other. Move the ones that are in the wrong position. Do | |
| 3970 * keep the same ordering! */ | |
| 3971 while (i + 1 < gap->ga_len | |
| 3972 && (*smp[i + 1].sm_lead_w & 0xff) == c) | |
| 3973 /* Skip over entry with same index byte. */ | |
| 3974 ++i; | |
| 3975 | |
| 3976 for (n = 1; i + n < gap->ga_len; ++n) | |
| 3977 if ((*smp[i + n].sm_lead_w & 0xff) == c) | |
| 3978 { | |
| 3979 salitem_T tsal; | |
| 3980 | |
| 3981 /* Move entry with same index byte after the entries | |
| 3982 * we already found. */ | |
| 3983 ++i; | |
| 3984 --n; | |
| 3985 tsal = smp[i + n]; | |
| 3986 mch_memmove(smp + i + 1, smp + i, | |
| 3987 sizeof(salitem_T) * n); | |
| 3988 smp[i] = tsal; | |
| 3989 } | |
| 3990 } | |
| 3991 #endif | |
| 3992 } | |
| 3993 } | |
| 3994 } | |
| 376 | 3995 |
| 372 | 3996 #ifdef FEAT_MBYTE |
| 3997 /* | |
| 3998 * Turn a multi-byte string into a wide character string. | |
| 3999 * Return it in allocated memory (NULL for out-of-memory) | |
| 4000 */ | |
| 4001 static int * | |
| 4002 mb_str2wide(s) | |
| 4003 char_u *s; | |
| 4004 { | |
| 4005 int *res; | |
| 4006 char_u *p; | |
| 4007 int i = 0; | |
| 4008 | |
| 4009 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1)); | |
| 4010 if (res != NULL) | |
| 4011 { | |
| 4012 for (p = s; *p != NUL; ) | |
| 4013 res[i++] = mb_ptr2char_adv(&p); | |
| 4014 res[i] = NUL; | |
| 4015 } | |
| 4016 return res; | |
| 4017 } | |
| 4018 #endif | |
| 4019 | |
| 236 | 4020 /* |
| 625 | 4021 * Read a tree from the .spl or .sug file. |
| 4022 * Allocates the memory and stores pointers in "bytsp" and "idxsp". | |
| 4023 * This is skipped when the tree has zero length. | |
| 4024 * Returns zero when OK, SP_ value for an error. | |
| 4025 */ | |
| 4026 static int | |
| 4027 spell_read_tree(fd, bytsp, idxsp, prefixtree, prefixcnt) | |
| 4028 FILE *fd; | |
| 4029 char_u **bytsp; | |
| 4030 idx_T **idxsp; | |
| 4031 int prefixtree; /* TRUE for the prefix tree */ | |
| 4032 int prefixcnt; /* when "prefixtree" is TRUE: prefix count */ | |
| 4033 { | |
| 4034 int len; | |
| 4035 int idx; | |
| 4036 char_u *bp; | |
| 4037 idx_T *ip; | |
| 4038 | |
| 4039 /* The tree size was computed when writing the file, so that we can | |
| 4040 * allocate it as one long block. <nodecount> */ | |
| 695 | 4041 len = get4c(fd); |
| 625 | 4042 if (len < 0) |
| 4043 return SP_TRUNCERROR; | |
| 4044 if (len > 0) | |
| 4045 { | |
| 4046 /* Allocate the byte array. */ | |
| 4047 bp = lalloc((long_u)len, TRUE); | |
| 4048 if (bp == NULL) | |
| 4049 return SP_OTHERERROR; | |
| 4050 *bytsp = bp; | |
| 4051 | |
| 4052 /* Allocate the index array. */ | |
| 4053 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE); | |
| 4054 if (ip == NULL) | |
| 4055 return SP_OTHERERROR; | |
| 4056 *idxsp = ip; | |
| 4057 | |
| 4058 /* Recursively read the tree and store it in the array. */ | |
| 4059 idx = read_tree_node(fd, bp, ip, len, 0, prefixtree, prefixcnt); | |
| 4060 if (idx < 0) | |
| 4061 return idx; | |
| 4062 } | |
| 4063 return 0; | |
| 4064 } | |
| 4065 | |
| 4066 /* | |
| 300 | 4067 * Read one row of siblings from the spell file and store it in the byte array |
| 4068 * "byts" and index array "idxs". Recursively read the children. | |
| 4069 * | |
| 625 | 4070 * NOTE: The code here must match put_node()! |
| 4071 * | |
| 4072 * Returns the index (>= 0) following the siblings. | |
| 4073 * Returns SP_TRUNCERROR if the file is shorter than expected. | |
| 4074 * Returns SP_FORMERROR if there is a format error. | |
| 236 | 4075 */ |
| 324 | 4076 static idx_T |
| 625 | 4077 read_tree_node(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr) |
| 300 | 4078 FILE *fd; |
| 4079 char_u *byts; | |
| 324 | 4080 idx_T *idxs; |
| 300 | 4081 int maxidx; /* size of arrays */ |
| 324 | 4082 idx_T startidx; /* current index in "byts" and "idxs" */ |
| 339 | 4083 int prefixtree; /* TRUE for reading PREFIXTREE */ |
| 4084 int maxprefcondnr; /* maximum for <prefcondnr> */ | |
| 236 | 4085 { |
| 300 | 4086 int len; |
| 4087 int i; | |
| 4088 int n; | |
| 324 | 4089 idx_T idx = startidx; |
| 300 | 4090 int c; |
| 366 | 4091 int c2; |
| 300 | 4092 #define SHARED_MASK 0x8000000 |
| 236 | 4093 |
| 300 | 4094 len = getc(fd); /* <siblingcount> */ |
| 4095 if (len <= 0) | |
| 625 | 4096 return SP_TRUNCERROR; |
| 300 | 4097 |
| 4098 if (startidx + len >= maxidx) | |
| 625 | 4099 return SP_FORMERROR; |
| 300 | 4100 byts[idx++] = len; |
| 4101 | |
| 4102 /* Read the byte values, flag/region bytes and shared indexes. */ | |
| 4103 for (i = 1; i <= len; ++i) | |
| 236 | 4104 { |
| 300 | 4105 c = getc(fd); /* <byte> */ |
| 4106 if (c < 0) | |
| 625 | 4107 return SP_TRUNCERROR; |
| 300 | 4108 if (c <= BY_SPECIAL) |
| 4109 { | |
| 366 | 4110 if (c == BY_NOFLAGS && !prefixtree) |
| 300 | 4111 { |
| 4112 /* No flags, all regions. */ | |
| 4113 idxs[idx] = 0; | |
| 4114 c = 0; | |
| 4115 } | |
| 390 | 4116 else if (c != BY_INDEX) |
| 300 | 4117 { |
| 339 | 4118 if (prefixtree) |
| 4119 { | |
| 455 | 4120 /* Read the optional pflags byte, the prefix ID and the |
| 4121 * condition nr. In idxs[] store the prefix ID in the low | |
| 4122 * byte, the condition index shifted up 8 bits, the flags | |
| 4123 * shifted up 24 bits. */ | |
| 4124 if (c == BY_FLAGS) | |
| 4125 c = getc(fd) << 24; /* <pflags> */ | |
| 4126 else | |
| 4127 c = 0; | |
| 4128 | |
| 481 | 4129 c |= getc(fd); /* <affixID> */ |
| 455 | 4130 |
| 695 | 4131 n = get2c(fd); /* <prefcondnr> */ |
| 339 | 4132 if (n >= maxprefcondnr) |
| 625 | 4133 return SP_FORMERROR; |
| 455 | 4134 c |= (n << 8); |
| 339 | 4135 } |
| 390 | 4136 else /* c must be BY_FLAGS or BY_FLAGS2 */ |
| 339 | 4137 { |
| 4138 /* Read flags and optional region and prefix ID. In | |
| 390 | 4139 * idxs[] the flags go in the low two bytes, region above |
| 4140 * that and prefix ID above the region. */ | |
| 4141 c2 = c; | |
| 339 | 4142 c = getc(fd); /* <flags> */ |
| 390 | 4143 if (c2 == BY_FLAGS2) |
| 4144 c = (getc(fd) << 8) + c; /* <flags2> */ | |
| 339 | 4145 if (c & WF_REGION) |
| 390 | 4146 c = (getc(fd) << 16) + c; /* <region> */ |
| 481 | 4147 if (c & WF_AFX) |
| 4148 c = (getc(fd) << 24) + c; /* <affixID> */ | |
| 339 | 4149 } |
| 4150 | |
| 300 | 4151 idxs[idx] = c; |
| 4152 c = 0; | |
| 4153 } | |
| 4154 else /* c == BY_INDEX */ | |
| 4155 { | |
| 4156 /* <nodeidx> */ | |
| 695 | 4157 n = get3c(fd); |
| 300 | 4158 if (n < 0 || n >= maxidx) |
| 625 | 4159 return SP_FORMERROR; |
| 300 | 4160 idxs[idx] = n + SHARED_MASK; |
| 4161 c = getc(fd); /* <xbyte> */ | |
| 4162 } | |
| 4163 } | |
| 4164 byts[idx++] = c; | |
| 236 | 4165 } |
| 4166 | |
| 300 | 4167 /* Recursively read the children for non-shared siblings. |
| 4168 * Skip the end-of-word ones (zero byte value) and the shared ones (and | |
| 4169 * remove SHARED_MASK) */ | |
| 4170 for (i = 1; i <= len; ++i) | |
| 4171 if (byts[startidx + i] != 0) | |
| 4172 { | |
| 4173 if (idxs[startidx + i] & SHARED_MASK) | |
| 4174 idxs[startidx + i] &= ~SHARED_MASK; | |
| 4175 else | |
| 4176 { | |
| 4177 idxs[startidx + i] = idx; | |
| 625 | 4178 idx = read_tree_node(fd, byts, idxs, maxidx, idx, |
| 339 | 4179 prefixtree, maxprefcondnr); |
| 300 | 4180 if (idx < 0) |
| 4181 break; | |
| 4182 } | |
| 4183 } | |
| 236 | 4184 |
| 300 | 4185 return idx; |
| 236 | 4186 } |
| 4187 | |
| 4188 /* | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4189 * Parse 'spelllang' and set w_s->b_langp accordingly. |
| 351 | 4190 * Returns NULL if it's OK, an error message otherwise. |
| 236 | 4191 */ |
| 4192 char_u * | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4193 did_set_spelllang(wp) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4194 win_T *wp; |
| 236 | 4195 { |
| 4196 garray_T ga; | |
| 351 | 4197 char_u *splp; |
| 236 | 4198 char_u *region; |
| 416 | 4199 char_u region_cp[3]; |
| 355 | 4200 int filename; |
| 236 | 4201 int region_mask; |
| 503 | 4202 slang_T *slang; |
| 236 | 4203 int c; |
| 351 | 4204 char_u lang[MAXWLEN + 1]; |
| 323 | 4205 char_u spf_name[MAXPATHL]; |
| 351 | 4206 int len; |
| 4207 char_u *p; | |
| 381 | 4208 int round; |
| 385 | 4209 char_u *spf; |
| 389 | 4210 char_u *use_region = NULL; |
| 4211 int dont_use_region = FALSE; | |
| 501 | 4212 int nobreak = FALSE; |
| 503 | 4213 int i, j; |
| 4214 langp_T *lp, *lp2; | |
| 1185 | 4215 static int recursive = FALSE; |
| 4216 char_u *ret_msg = NULL; | |
| 4217 char_u *spl_copy; | |
| 4218 | |
| 4219 /* We don't want to do this recursively. May happen when a language is | |
| 4220 * not available and the SpellFileMissing autocommand opens a new buffer | |
| 4221 * in which 'spell' is set. */ | |
| 4222 if (recursive) | |
| 4223 return NULL; | |
| 4224 recursive = TRUE; | |
| 236 | 4225 |
| 4226 ga_init2(&ga, sizeof(langp_T), 2); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4227 clear_midword(wp); |
| 236 | 4228 |
| 4352 | 4229 /* Make a copy of 'spelllang', the SpellFileMissing autocommands may change |
| 1185 | 4230 * it under our fingers. */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4231 spl_copy = vim_strsave(wp->w_s->b_p_spl); |
| 1185 | 4232 if (spl_copy == NULL) |
| 4233 goto theend; | |
| 4234 | |
| 5485 | 4235 #ifdef FEAT_MBYTE |
| 5477 | 4236 wp->w_s->b_cjk = 0; |
| 5485 | 4237 #endif |
| 5477 | 4238 |
| 4239 /* Loop over comma separated language names. */ | |
| 1185 | 4240 for (splp = spl_copy; *splp != NUL; ) |
| 351 | 4241 { |
| 4242 /* Get one language name. */ | |
| 4243 copy_option_part(&splp, lang, MAXWLEN, ","); | |
| 240 | 4244 region = NULL; |
| 835 | 4245 len = (int)STRLEN(lang); |
| 355 | 4246 |
| 5477 | 4247 if (STRCMP(lang, "cjk") == 0) |
| 4248 { | |
| 5485 | 4249 #ifdef FEAT_MBYTE |
| 5477 | 4250 wp->w_s->b_cjk = 1; |
| 5485 | 4251 #endif |
| 5477 | 4252 continue; |
| 4253 } | |
| 4254 | |
| 355 | 4255 /* If the name ends in ".spl" use it as the name of the spell file. |
| 4256 * If there is a region name let "region" point to it and remove it | |
| 4257 * from the name. */ | |
| 4258 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0) | |
| 236 | 4259 { |
| 355 | 4260 filename = TRUE; |
| 4261 | |
| 416 | 4262 /* Locate a region and remove it from the file name. */ |
| 4263 p = vim_strchr(gettail(lang), '_'); | |
| 4264 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2]) | |
| 4265 && !ASCII_ISALPHA(p[3])) | |
| 4266 { | |
| 4267 vim_strncpy(region_cp, p + 1, 2); | |
| 4268 mch_memmove(p, p + 3, len - (p - lang) - 2); | |
| 4269 len -= 3; | |
| 4270 region = region_cp; | |
| 4271 } | |
| 4272 else | |
| 4273 dont_use_region = TRUE; | |
| 4274 | |
| 355 | 4275 /* Check if we loaded this language before. */ |
| 503 | 4276 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 4277 if (fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME) | |
| 355 | 4278 break; |
| 236 | 4279 } |
| 355 | 4280 else |
| 4281 { | |
| 4282 filename = FALSE; | |
| 4283 if (len > 3 && lang[len - 3] == '_') | |
| 4284 { | |
| 4285 region = lang + len - 2; | |
| 4286 len -= 3; | |
| 4287 lang[len] = NUL; | |
| 389 | 4288 } |
| 4289 else | |
| 4290 dont_use_region = TRUE; | |
| 355 | 4291 |
| 4292 /* Check if we loaded this language before. */ | |
| 503 | 4293 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 4294 if (STRICMP(lang, slang->sl_name) == 0) | |
| 355 | 4295 break; |
| 4296 } | |
| 236 | 4297 |
| 416 | 4298 if (region != NULL) |
| 4299 { | |
| 4300 /* If the region differs from what was used before then don't | |
| 4301 * use it for 'spellfile'. */ | |
| 4302 if (use_region != NULL && STRCMP(region, use_region) != 0) | |
| 4303 dont_use_region = TRUE; | |
| 4304 use_region = region; | |
| 4305 } | |
| 4306 | |
| 351 | 4307 /* If not found try loading the language now. */ |
| 503 | 4308 if (slang == NULL) |
| 355 | 4309 { |
| 4310 if (filename) | |
| 4311 (void)spell_load_file(lang, lang, NULL, FALSE); | |
| 4312 else | |
| 1185 | 4313 { |
| 355 | 4314 spell_load_lang(lang); |
| 1185 | 4315 #ifdef FEAT_AUTOCMD |
| 4316 /* SpellFileMissing autocommands may do anything, including | |
| 4317 * destroying the buffer we are using... */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4318 if (!buf_valid(wp->w_buffer)) |
| 1185 | 4319 { |
| 4320 ret_msg = (char_u *)"E797: SpellFileMissing autocommand deleted buffer"; | |
| 4321 goto theend; | |
| 4322 } | |
| 4323 #endif | |
| 4324 } | |
| 355 | 4325 } |
| 236 | 4326 |
| 307 | 4327 /* |
| 351 | 4328 * Loop over the languages, there can be several files for "lang". |
| 307 | 4329 */ |
| 503 | 4330 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 4331 if (filename ? fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME | |
| 4332 : STRICMP(lang, slang->sl_name) == 0) | |
| 236 | 4333 { |
| 316 | 4334 region_mask = REGION_ALL; |
| 355 | 4335 if (!filename && region != NULL) |
| 236 | 4336 { |
| 307 | 4337 /* find region in sl_regions */ |
| 503 | 4338 c = find_region(slang->sl_regions, region); |
| 307 | 4339 if (c == REGION_ALL) |
| 4340 { | |
| 503 | 4341 if (slang->sl_add) |
| 389 | 4342 { |
| 503 | 4343 if (*slang->sl_regions != NUL) |
| 389 | 4344 /* This addition file is for other regions. */ |
| 4345 region_mask = 0; | |
| 4346 } | |
| 4347 else | |
| 4348 /* This is probably an error. Give a warning and | |
| 4349 * accept the words anyway. */ | |
| 351 | 4350 smsg((char_u *) |
| 4351 _("Warning: region %s not supported"), | |
| 4352 region); | |
| 307 | 4353 } |
| 4354 else | |
| 4355 region_mask = 1 << c; | |
| 236 | 4356 } |
| 307 | 4357 |
| 389 | 4358 if (region_mask != 0) |
| 307 | 4359 { |
| 389 | 4360 if (ga_grow(&ga, 1) == FAIL) |
| 4361 { | |
| 4362 ga_clear(&ga); | |
| 1185 | 4363 ret_msg = e_outofmem; |
| 4364 goto theend; | |
| 389 | 4365 } |
| 503 | 4366 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 389 | 4367 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 4368 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4369 use_midword(slang, wp); |
| 503 | 4370 if (slang->sl_nobreak) |
| 501 | 4371 nobreak = TRUE; |
| 307 | 4372 } |
| 385 | 4373 } |
| 4374 } | |
| 4375 | |
| 4376 /* round 0: load int_wordlist, if possible. | |
| 4377 * round 1: load first name in 'spellfile'. | |
| 4378 * round 2: load second name in 'spellfile. | |
| 4379 * etc. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4380 spf = curwin->w_s->b_p_spf; |
| 385 | 4381 for (round = 0; round == 0 || *spf != NUL; ++round) |
| 4382 { | |
| 4383 if (round == 0) | |
| 4384 { | |
| 4385 /* Internal wordlist, if there is one. */ | |
| 4386 if (int_wordlist == NULL) | |
| 381 | 4387 continue; |
| 385 | 4388 int_wordlist_spl(spf_name); |
| 381 | 4389 } |
| 4390 else | |
| 4391 { | |
| 385 | 4392 /* One entry in 'spellfile'. */ |
| 4393 copy_option_part(&spf, spf_name, MAXPATHL - 5, ","); | |
| 4394 STRCAT(spf_name, ".spl"); | |
| 4395 | |
| 4396 /* If it was already found above then skip it. */ | |
| 4397 for (c = 0; c < ga.ga_len; ++c) | |
| 500 | 4398 { |
| 4399 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname; | |
| 4400 if (p != NULL && fullpathcmp(spf_name, p, FALSE) == FPC_SAME) | |
| 385 | 4401 break; |
| 500 | 4402 } |
| 385 | 4403 if (c < ga.ga_len) |
| 381 | 4404 continue; |
| 4405 } | |
| 4406 | |
| 351 | 4407 /* Check if it was loaded already. */ |
| 503 | 4408 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 4409 if (fullpathcmp(spf_name, slang->sl_fname, FALSE) == FPC_SAME) | |
| 323 | 4410 break; |
| 503 | 4411 if (slang == NULL) |
| 323 | 4412 { |
| 351 | 4413 /* Not loaded, try loading it now. The language name includes the |
| 385 | 4414 * region name, the region is ignored otherwise. for int_wordlist |
| 4415 * use an arbitrary name. */ | |
| 4416 if (round == 0) | |
| 4417 STRCPY(lang, "internal wordlist"); | |
| 4418 else | |
| 4419 { | |
| 4420 vim_strncpy(lang, gettail(spf_name), MAXWLEN); | |
| 381 | 4421 p = vim_strchr(lang, '.'); |
| 4422 if (p != NULL) | |
| 4423 *p = NUL; /* truncate at ".encoding.add" */ | |
| 4424 } | |
| 503 | 4425 slang = spell_load_file(spf_name, lang, NULL, TRUE); |
| 501 | 4426 |
| 4427 /* If one of the languages has NOBREAK we assume the addition | |
| 4428 * files also have this. */ | |
| 503 | 4429 if (slang != NULL && nobreak) |
| 4430 slang->sl_nobreak = TRUE; | |
| 4431 } | |
| 4432 if (slang != NULL && ga_grow(&ga, 1) == OK) | |
| 323 | 4433 { |
| 389 | 4434 region_mask = REGION_ALL; |
| 4435 if (use_region != NULL && !dont_use_region) | |
| 4436 { | |
| 4437 /* find region in sl_regions */ | |
| 503 | 4438 c = find_region(slang->sl_regions, use_region); |
| 389 | 4439 if (c != REGION_ALL) |
| 4440 region_mask = 1 << c; | |
| 503 | 4441 else if (*slang->sl_regions != NUL) |
| 389 | 4442 /* This spell file is for other regions. */ |
| 4443 region_mask = 0; | |
| 4444 } | |
| 4445 | |
| 4446 if (region_mask != 0) | |
| 4447 { | |
| 503 | 4448 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 4449 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL; | |
| 4450 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL; | |
| 389 | 4451 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 4452 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4453 use_midword(slang, wp); |
| 389 | 4454 } |
| 323 | 4455 } |
| 4456 } | |
| 4457 | |
| 236 | 4458 /* Everything is fine, store the new b_langp value. */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4459 ga_clear(&wp->w_s->b_langp); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4460 wp->w_s->b_langp = ga; |
| 236 | 4461 |
| 503 | 4462 /* For each language figure out what language to use for sound folding and |
| 4463 * REP items. If the language doesn't support it itself use another one | |
| 4464 * with the same name. E.g. for "en-math" use "en". */ | |
| 4465 for (i = 0; i < ga.ga_len; ++i) | |
| 4466 { | |
| 4467 lp = LANGP_ENTRY(ga, i); | |
| 4468 | |
| 4469 /* sound folding */ | |
| 4470 if (lp->lp_slang->sl_sal.ga_len > 0) | |
| 4471 /* language does sound folding itself */ | |
| 4472 lp->lp_sallang = lp->lp_slang; | |
| 4473 else | |
| 4474 /* find first similar language that does sound folding */ | |
| 4475 for (j = 0; j < ga.ga_len; ++j) | |
| 4476 { | |
| 4477 lp2 = LANGP_ENTRY(ga, j); | |
| 4478 if (lp2->lp_slang->sl_sal.ga_len > 0 | |
| 4479 && STRNCMP(lp->lp_slang->sl_name, | |
| 4480 lp2->lp_slang->sl_name, 2) == 0) | |
| 4481 { | |
| 4482 lp->lp_sallang = lp2->lp_slang; | |
| 4483 break; | |
| 4484 } | |
| 4485 } | |
| 4486 | |
| 4487 /* REP items */ | |
| 4488 if (lp->lp_slang->sl_rep.ga_len > 0) | |
| 4489 /* language has REP items itself */ | |
| 4490 lp->lp_replang = lp->lp_slang; | |
| 4491 else | |
| 625 | 4492 /* find first similar language that has REP items */ |
| 503 | 4493 for (j = 0; j < ga.ga_len; ++j) |
| 4494 { | |
| 4495 lp2 = LANGP_ENTRY(ga, j); | |
| 4496 if (lp2->lp_slang->sl_rep.ga_len > 0 | |
| 4497 && STRNCMP(lp->lp_slang->sl_name, | |
| 4498 lp2->lp_slang->sl_name, 2) == 0) | |
| 4499 { | |
| 4500 lp->lp_replang = lp2->lp_slang; | |
| 4501 break; | |
| 4502 } | |
| 4503 } | |
| 4504 } | |
| 4505 | |
| 1185 | 4506 theend: |
| 4507 vim_free(spl_copy); | |
| 4508 recursive = FALSE; | |
| 5891 | 4509 redraw_win_later(wp, NOT_VALID); |
| 1185 | 4510 return ret_msg; |
| 236 | 4511 } |
| 4512 | |
| 4513 /* | |
| 376 | 4514 * Clear the midword characters for buffer "buf". |
| 4515 */ | |
| 4516 static void | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4517 clear_midword(wp) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4518 win_T *wp; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4519 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4520 vim_memset(wp->w_s->b_spell_ismw, 0, 256); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4521 #ifdef FEAT_MBYTE |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4522 vim_free(wp->w_s->b_spell_ismw_mb); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4523 wp->w_s->b_spell_ismw_mb = NULL; |
| 376 | 4524 #endif |
| 4525 } | |
| 4526 | |
| 4527 /* | |
| 4528 * Use the "sl_midword" field of language "lp" for buffer "buf". | |
| 4529 * They add up to any currently used midword characters. | |
| 4530 */ | |
| 4531 static void | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4532 use_midword(lp, wp) |
| 376 | 4533 slang_T *lp; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4534 win_T *wp; |
| 376 | 4535 { |
| 4536 char_u *p; | |
| 4537 | |
| 389 | 4538 if (lp->sl_midword == NULL) /* there aren't any */ |
| 4539 return; | |
| 4540 | |
| 376 | 4541 for (p = lp->sl_midword; *p != NUL; ) |
| 4542 #ifdef FEAT_MBYTE | |
| 4543 if (has_mbyte) | |
| 4544 { | |
| 4545 int c, l, n; | |
| 4546 char_u *bp; | |
| 4547 | |
| 4548 c = mb_ptr2char(p); | |
| 474 | 4549 l = (*mb_ptr2len)(p); |
| 4550 if (c < 256 && l <= 2) | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4551 wp->w_s->b_spell_ismw[c] = TRUE; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4552 else if (wp->w_s->b_spell_ismw_mb == NULL) |
| 376 | 4553 /* First multi-byte char in "b_spell_ismw_mb". */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4554 wp->w_s->b_spell_ismw_mb = vim_strnsave(p, l); |
| 376 | 4555 else |
| 4556 { | |
| 4557 /* Append multi-byte chars to "b_spell_ismw_mb". */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4558 n = (int)STRLEN(wp->w_s->b_spell_ismw_mb); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4559 bp = vim_strnsave(wp->w_s->b_spell_ismw_mb, n + l); |
| 376 | 4560 if (bp != NULL) |
| 4561 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4562 vim_free(wp->w_s->b_spell_ismw_mb); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4563 wp->w_s->b_spell_ismw_mb = bp; |
| 376 | 4564 vim_strncpy(bp + n, p, l); |
| 4565 } | |
| 4566 } | |
| 4567 p += l; | |
| 4568 } | |
| 4569 else | |
| 4570 #endif | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4571 wp->w_s->b_spell_ismw[*p++] = TRUE; |
| 376 | 4572 } |
| 4573 | |
| 4574 /* | |
| 236 | 4575 * Find the region "region[2]" in "rp" (points to "sl_regions"). |
| 4576 * Each region is simply stored as the two characters of it's name. | |
| 381 | 4577 * Returns the index if found (first is 0), REGION_ALL if not found. |
| 236 | 4578 */ |
| 4579 static int | |
| 4580 find_region(rp, region) | |
| 4581 char_u *rp; | |
| 4582 char_u *region; | |
| 4583 { | |
| 4584 int i; | |
| 4585 | |
| 4586 for (i = 0; ; i += 2) | |
| 4587 { | |
| 4588 if (rp[i] == NUL) | |
| 4589 return REGION_ALL; | |
| 4590 if (rp[i] == region[0] && rp[i + 1] == region[1]) | |
| 4591 break; | |
| 4592 } | |
| 4593 return i / 2; | |
| 4594 } | |
| 4595 | |
| 4596 /* | |
| 323 | 4597 * Return case type of word: |
| 236 | 4598 * w word 0 |
| 300 | 4599 * Word WF_ONECAP |
| 4600 * W WORD WF_ALLCAP | |
| 4601 * WoRd wOrd WF_KEEPCAP | |
| 236 | 4602 */ |
| 4603 static int | |
| 4604 captype(word, end) | |
| 4605 char_u *word; | |
| 323 | 4606 char_u *end; /* When NULL use up to NUL byte. */ |
| 236 | 4607 { |
| 4608 char_u *p; | |
| 4609 int c; | |
| 4610 int firstcap; | |
| 4611 int allcap; | |
| 4612 int past_second = FALSE; /* past second word char */ | |
| 4613 | |
| 4614 /* find first letter */ | |
| 5477 | 4615 for (p = word; !spell_iswordp_nmw(p, curwin); mb_ptr_adv(p)) |
| 323 | 4616 if (end == NULL ? *p == NUL : p >= end) |
| 236 | 4617 return 0; /* only non-word characters, illegal word */ |
| 4618 #ifdef FEAT_MBYTE | |
| 310 | 4619 if (has_mbyte) |
| 4620 c = mb_ptr2char_adv(&p); | |
| 4621 else | |
| 236 | 4622 #endif |
| 310 | 4623 c = *p++; |
| 324 | 4624 firstcap = allcap = SPELL_ISUPPER(c); |
| 236 | 4625 |
| 4626 /* | |
| 4627 * Need to check all letters to find a word with mixed upper/lower. | |
| 4628 * But a word with an upper char only at start is a ONECAP. | |
| 4629 */ | |
| 323 | 4630 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p)) |
| 5477 | 4631 if (spell_iswordp_nmw(p, curwin)) |
| 236 | 4632 { |
| 455 | 4633 c = PTR2CHAR(p); |
| 324 | 4634 if (!SPELL_ISUPPER(c)) |
| 236 | 4635 { |
| 4636 /* UUl -> KEEPCAP */ | |
| 4637 if (past_second && allcap) | |
| 300 | 4638 return WF_KEEPCAP; |
| 236 | 4639 allcap = FALSE; |
| 4640 } | |
| 4641 else if (!allcap) | |
| 4642 /* UlU -> KEEPCAP */ | |
| 300 | 4643 return WF_KEEPCAP; |
| 236 | 4644 past_second = TRUE; |
| 4645 } | |
| 4646 | |
| 4647 if (allcap) | |
| 300 | 4648 return WF_ALLCAP; |
| 236 | 4649 if (firstcap) |
| 300 | 4650 return WF_ONECAP; |
| 236 | 4651 return 0; |
| 4652 } | |
| 4653 | |
| 474 | 4654 /* |
| 4655 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a | |
| 4656 * capital. So that make_case_word() can turn WOrd into Word. | |
| 4657 * Add ALLCAP for "WOrD". | |
| 4658 */ | |
| 4659 static int | |
| 4660 badword_captype(word, end) | |
| 4661 char_u *word; | |
| 4662 char_u *end; | |
| 4663 { | |
| 4664 int flags = captype(word, end); | |
| 475 | 4665 int c; |
| 474 | 4666 int l, u; |
| 4667 int first; | |
| 4668 char_u *p; | |
| 4669 | |
| 4670 if (flags & WF_KEEPCAP) | |
| 4671 { | |
| 4672 /* Count the number of UPPER and lower case letters. */ | |
| 4673 l = u = 0; | |
| 4674 first = FALSE; | |
| 4675 for (p = word; p < end; mb_ptr_adv(p)) | |
| 4676 { | |
| 475 | 4677 c = PTR2CHAR(p); |
| 4678 if (SPELL_ISUPPER(c)) | |
| 474 | 4679 { |
| 4680 ++u; | |
| 4681 if (p == word) | |
| 4682 first = TRUE; | |
| 4683 } | |
| 4684 else | |
| 4685 ++l; | |
| 4686 } | |
| 4687 | |
| 4688 /* If there are more UPPER than lower case letters suggest an | |
| 4689 * ALLCAP word. Otherwise, if the first letter is UPPER then | |
| 4690 * suggest ONECAP. Exception: "ALl" most likely should be "All", | |
| 4691 * require three upper case letters. */ | |
| 4692 if (u > l && u > 2) | |
| 4693 flags |= WF_ALLCAP; | |
| 4694 else if (first) | |
| 4695 flags |= WF_ONECAP; | |
| 633 | 4696 |
| 4697 if (u >= 2 && l >= 2) /* maCARONI maCAroni */ | |
| 4698 flags |= WF_MIXCAP; | |
| 474 | 4699 } |
| 4700 return flags; | |
| 4701 } | |
| 4702 | |
| 5519 | 4703 /* |
| 4704 * Delete the internal wordlist and its .spl file. | |
| 4705 */ | |
| 4706 void | |
| 4707 spell_delete_wordlist() | |
| 4708 { | |
| 4709 char_u fname[MAXPATHL]; | |
| 4710 | |
| 4711 if (int_wordlist != NULL) | |
| 4712 { | |
| 4713 mch_remove(int_wordlist); | |
| 4714 int_wordlist_spl(fname); | |
| 4715 mch_remove(fname); | |
| 4716 vim_free(int_wordlist); | |
| 4717 int_wordlist = NULL; | |
| 4718 } | |
| 4719 } | |
| 4720 | |
| 4721 #if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO) | |
| 355 | 4722 /* |
| 4723 * Free all languages. | |
| 4724 */ | |
| 4725 void | |
| 4726 spell_free_all() | |
| 4727 { | |
| 503 | 4728 slang_T *slang; |
| 355 | 4729 buf_T *buf; |
| 4730 | |
| 2580 | 4731 /* Go through all buffers and handle 'spelllang'. <VN> */ |
| 355 | 4732 for (buf = firstbuf; buf != NULL; buf = buf->b_next) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4733 ga_clear(&buf->b_s.b_langp); |
| 355 | 4734 |
| 4735 while (first_lang != NULL) | |
| 4736 { | |
| 503 | 4737 slang = first_lang; |
| 4738 first_lang = slang->sl_next; | |
| 4739 slang_free(slang); | |
| 355 | 4740 } |
| 366 | 4741 |
| 5519 | 4742 spell_delete_wordlist(); |
| 381 | 4743 |
| 626 | 4744 vim_free(repl_to); |
| 4745 repl_to = NULL; | |
| 4746 vim_free(repl_from); | |
| 4747 repl_from = NULL; | |
| 355 | 4748 } |
| 5519 | 4749 #endif |
| 4750 | |
| 4751 #if defined(FEAT_MBYTE) || defined(PROTO) | |
| 236 | 4752 /* |
| 4753 * Clear all spelling tables and reload them. | |
| 307 | 4754 * Used after 'encoding' is set and when ":mkspell" was used. |
| 236 | 4755 */ |
| 4756 void | |
| 4757 spell_reload() | |
| 4758 { | |
| 316 | 4759 win_T *wp; |
| 236 | 4760 |
| 358 | 4761 /* Initialize the table for spell_iswordp(). */ |
| 236 | 4762 init_spell_chartab(); |
| 4763 | |
| 4764 /* Unload all allocated memory. */ | |
| 355 | 4765 spell_free_all(); |
| 236 | 4766 |
| 4767 /* Go through all buffers and handle 'spelllang'. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4768 for (wp = firstwin; wp != NULL; wp = wp->w_next) |
| 236 | 4769 { |
| 316 | 4770 /* Only load the wordlists when 'spelllang' is set and there is a |
| 4771 * window for this buffer in which 'spell' is set. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4772 if (*wp->w_s->b_p_spl != NUL) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4773 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4774 if (wp->w_p_spell) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4775 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4776 (void)did_set_spelllang(wp); |
| 316 | 4777 # ifdef FEAT_WINDOWS |
| 4778 break; | |
| 4779 # endif | |
| 4780 } | |
| 4781 } | |
| 236 | 4782 } |
| 4783 } | |
| 5519 | 4784 #endif |
| 236 | 4785 |
| 310 | 4786 /* |
| 4787 * Reload the spell file "fname" if it's loaded. | |
| 4788 */ | |
| 4789 static void | |
| 323 | 4790 spell_reload_one(fname, added_word) |
| 310 | 4791 char_u *fname; |
| 323 | 4792 int added_word; /* invoked through "zg" */ |
| 310 | 4793 { |
| 503 | 4794 slang_T *slang; |
| 323 | 4795 int didit = FALSE; |
| 310 | 4796 |
| 503 | 4797 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 4798 { | |
| 4799 if (fullpathcmp(fname, slang->sl_fname, FALSE) == FPC_SAME) | |
| 4800 { | |
| 4801 slang_clear(slang); | |
| 4802 if (spell_load_file(fname, NULL, slang, FALSE) == NULL) | |
| 500 | 4803 /* reloading failed, clear the language */ |
| 503 | 4804 slang_clear(slang); |
| 737 | 4805 redraw_all_later(SOME_VALID); |
| 323 | 4806 didit = TRUE; |
| 310 | 4807 } |
| 500 | 4808 } |
| 323 | 4809 |
| 4810 /* When "zg" was used and the file wasn't loaded yet, should redo | |
| 626 | 4811 * 'spelllang' to load it now. */ |
| 323 | 4812 if (added_word && !didit) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4813 did_set_spelllang(curwin); |
| 310 | 4814 } |
| 4815 | |
| 4816 | |
| 236 | 4817 /* |
| 4818 * Functions for ":mkspell". | |
| 4819 */ | |
| 4820 | |
| 300 | 4821 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff |
| 236 | 4822 and .dic file. */ |
| 4823 /* | |
| 4824 * Main structure to store the contents of a ".aff" file. | |
| 4825 */ | |
| 4826 typedef struct afffile_S | |
| 4827 { | |
| 4828 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */ | |
| 498 | 4829 int af_flagtype; /* AFT_CHAR, AFT_LONG, AFT_NUM or AFT_CAPLONG */ |
| 621 | 4830 unsigned af_rare; /* RARE ID for rare word */ |
| 4831 unsigned af_keepcase; /* KEEPCASE ID for keep-case word */ | |
| 495 | 4832 unsigned af_bad; /* BAD ID for banned word */ |
| 4833 unsigned af_needaffix; /* NEEDAFFIX ID */ | |
| 820 | 4834 unsigned af_circumfix; /* CIRCUMFIX ID */ |
| 500 | 4835 unsigned af_needcomp; /* NEEDCOMPOUND ID */ |
| 809 | 4836 unsigned af_comproot; /* COMPOUNDROOT ID */ |
| 4837 unsigned af_compforbid; /* COMPOUNDFORBIDFLAG ID */ | |
| 4838 unsigned af_comppermit; /* COMPOUNDPERMITFLAG ID */ | |
| 698 | 4839 unsigned af_nosuggest; /* NOSUGGEST ID */ |
| 809 | 4840 int af_pfxpostpone; /* postpone prefixes without chop string and |
| 4841 without flags */ | |
| 6224 | 4842 int af_ignoreextra; /* IGNOREEXTRA present */ |
| 236 | 4843 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */ |
| 4844 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */ | |
| 495 | 4845 hashtab_T af_comp; /* hashtable for compound flags, compitem_T */ |
| 236 | 4846 } afffile_T; |
| 4847 | |
| 495 | 4848 #define AFT_CHAR 0 /* flags are one character */ |
| 498 | 4849 #define AFT_LONG 1 /* flags are two characters */ |
| 4850 #define AFT_CAPLONG 2 /* flags are one or two characters */ | |
| 4851 #define AFT_NUM 3 /* flags are numbers, comma separated */ | |
| 495 | 4852 |
| 236 | 4853 typedef struct affentry_S affentry_T; |
| 4854 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */ | |
| 4855 struct affentry_S | |
| 4856 { | |
| 4857 affentry_T *ae_next; /* next affix with same name/number */ | |
| 4858 char_u *ae_chop; /* text to chop off basic word (can be NULL) */ | |
| 4859 char_u *ae_add; /* text to add to basic word (can be NULL) */ | |
| 809 | 4860 char_u *ae_flags; /* flags on the affix (can be NULL) */ |
| 236 | 4861 char_u *ae_cond; /* condition (NULL for ".") */ |
| 4862 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */ | |
| 822 | 4863 char ae_compforbid; /* COMPOUNDFORBIDFLAG found */ |
| 4864 char ae_comppermit; /* COMPOUNDPERMITFLAG found */ | |
| 300 | 4865 }; |
| 4866 | |
| 495 | 4867 #ifdef FEAT_MBYTE |
| 4868 # define AH_KEY_LEN 17 /* 2 x 8 bytes + NUL */ | |
| 4869 #else | |
| 498 | 4870 # define AH_KEY_LEN 7 /* 6 digits + NUL */ |
| 495 | 4871 #endif |
| 455 | 4872 |
| 300 | 4873 /* Affix header from ".aff" file. Used for af_pref and af_suff. */ |
| 4874 typedef struct affheader_S | |
| 4875 { | |
| 495 | 4876 char_u ah_key[AH_KEY_LEN]; /* key for hashtab == name of affix */ |
| 4877 unsigned ah_flag; /* affix name as number, uses "af_flagtype" */ | |
| 4878 int ah_newID; /* prefix ID after renumbering; 0 if not used */ | |
| 300 | 4879 int ah_combine; /* suffix may combine with prefix */ |
| 498 | 4880 int ah_follows; /* another affix block should be following */ |
| 300 | 4881 affentry_T *ah_first; /* first affix entry */ |
| 4882 } affheader_T; | |
| 4883 | |
| 4884 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key) | |
| 4885 | |
| 495 | 4886 /* Flag used in compound items. */ |
| 4887 typedef struct compitem_S | |
| 4888 { | |
| 4889 char_u ci_key[AH_KEY_LEN]; /* key for hashtab == name of compound */ | |
| 4890 unsigned ci_flag; /* affix name as number, uses "af_flagtype" */ | |
| 4891 int ci_newID; /* affix ID after renumbering. */ | |
| 4892 } compitem_T; | |
| 4893 | |
| 4894 #define HI2CI(hi) ((compitem_T *)(hi)->hi_key) | |
| 4895 | |
| 300 | 4896 /* |
| 4897 * Structure that is used to store the items in the word tree. This avoids | |
| 474 | 4898 * the need to keep track of each allocated thing, everything is freed all at |
| 4899 * once after ":mkspell" is done. | |
| 1808 | 4900 * Note: "sb_next" must be just before "sb_data" to make sure the alignment of |
| 4901 * "sb_data" is correct for systems where pointers must be aligned on | |
| 4902 * pointer-size boundaries and sizeof(pointer) > sizeof(int) (e.g., Sparc). | |
| 300 | 4903 */ |
| 4904 #define SBLOCKSIZE 16000 /* size of sb_data */ | |
| 4905 typedef struct sblock_S sblock_T; | |
| 4906 struct sblock_S | |
| 4907 { | |
| 1808 | 4908 int sb_used; /* nr of bytes already in use */ |
| 300 | 4909 sblock_T *sb_next; /* next block in list */ |
| 4910 char_u sb_data[1]; /* data, actually longer */ | |
| 236 | 4911 }; |
| 4912 | |
| 4913 /* | |
| 300 | 4914 * A node in the tree. |
| 236 | 4915 */ |
| 300 | 4916 typedef struct wordnode_S wordnode_T; |
| 4917 struct wordnode_S | |
| 236 | 4918 { |
| 346 | 4919 union /* shared to save space */ |
| 4920 { | |
| 469 | 4921 char_u hashkey[6]; /* the hash key, only used while compressing */ |
| 346 | 4922 int index; /* index in written nodes (valid after first |
| 4923 round) */ | |
| 4924 } wn_u1; | |
| 4925 union /* shared to save space */ | |
| 4926 { | |
| 4927 wordnode_T *next; /* next node with same hash key */ | |
| 4928 wordnode_T *wnode; /* parent node that will write this node */ | |
| 4929 } wn_u2; | |
| 300 | 4930 wordnode_T *wn_child; /* child (next byte in word) */ |
| 4931 wordnode_T *wn_sibling; /* next sibling (alternate byte in word, | |
| 4932 always sorted) */ | |
| 474 | 4933 int wn_refs; /* Nr. of references to this node. Only |
| 4934 relevant for first node in a list of | |
| 4935 siblings, in following siblings it is | |
| 4936 always one. */ | |
| 300 | 4937 char_u wn_byte; /* Byte for this node. NUL for word end */ |
| 625 | 4938 |
| 4939 /* Info for when "wn_byte" is NUL. | |
| 4940 * In PREFIXTREE "wn_region" is used for the prefcondnr. | |
| 4941 * In the soundfolded word tree "wn_flags" has the MSW of the wordnr and | |
| 4942 * "wn_region" the LSW of the wordnr. */ | |
| 4943 char_u wn_affixID; /* supported/required prefix ID or 0 */ | |
| 4944 short_u wn_flags; /* WF_ flags */ | |
| 4945 short wn_region; /* region mask */ | |
| 4946 | |
| 469 | 4947 #ifdef SPELL_PRINTTREE |
| 4948 int wn_nr; /* sequence nr for printing */ | |
| 4949 #endif | |
| 236 | 4950 }; |
| 4951 | |
| 390 | 4952 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */ |
| 4953 | |
| 300 | 4954 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key) |
| 236 | 4955 |
| 300 | 4956 /* |
| 4957 * Info used while reading the spell files. | |
| 4958 */ | |
| 4959 typedef struct spellinfo_S | |
| 249 | 4960 { |
| 300 | 4961 wordnode_T *si_foldroot; /* tree with case-folded words */ |
| 334 | 4962 long si_foldwcount; /* nr of words in si_foldroot */ |
| 474 | 4963 |
| 300 | 4964 wordnode_T *si_keeproot; /* tree with keep-case words */ |
| 334 | 4965 long si_keepwcount; /* nr of words in si_keeproot */ |
| 474 | 4966 |
| 339 | 4967 wordnode_T *si_prefroot; /* tree with postponed prefixes */ |
| 474 | 4968 |
| 625 | 4969 long si_sugtree; /* creating the soundfolding trie */ |
| 4970 | |
| 300 | 4971 sblock_T *si_blocks; /* memory blocks used */ |
| 482 | 4972 long si_blocks_cnt; /* memory blocks allocated */ |
| 2691 | 4973 int si_did_emsg; /* TRUE when ran out of memory */ |
| 4974 | |
| 482 | 4975 long si_compress_cnt; /* words to add before lowering |
| 4976 compression limit */ | |
| 474 | 4977 wordnode_T *si_first_free; /* List of nodes that have been freed during |
| 4978 compression, linked by "wn_child" field. */ | |
| 482 | 4979 long si_free_count; /* number of nodes in si_first_free */ |
| 474 | 4980 #ifdef SPELL_PRINTTREE |
| 4981 int si_wordnode_nr; /* sequence nr for nodes */ | |
| 4982 #endif | |
| 625 | 4983 buf_T *si_spellbuf; /* buffer used to store soundfold word table */ |
| 474 | 4984 |
| 300 | 4985 int si_ascii; /* handling only ASCII words */ |
| 310 | 4986 int si_add; /* addition file */ |
| 351 | 4987 int si_clear_chartab; /* when TRUE clear char tables */ |
| 300 | 4988 int si_region; /* region mask */ |
| 4989 vimconv_T si_conv; /* for conversion to 'encoding' */ | |
| 302 | 4990 int si_memtot; /* runtime memory used */ |
| 310 | 4991 int si_verbose; /* verbose messages */ |
| 474 | 4992 int si_msg_count; /* number of words added since last message */ |
| 714 | 4993 char_u *si_info; /* info text chars or NULL */ |
| 316 | 4994 int si_region_count; /* number of regions supported (1 when there |
| 4995 are no regions) */ | |
| 2593 | 4996 char_u si_region_name[17]; /* region names; used only if |
| 483 | 4997 * si_region_count > 1) */ |
| 323 | 4998 |
| 4999 garray_T si_rep; /* list of fromto_T entries from REP lines */ | |
| 625 | 5000 garray_T si_repsal; /* list of fromto_T entries from REPSAL lines */ |
| 323 | 5001 garray_T si_sal; /* list of fromto_T entries from SAL lines */ |
| 375 | 5002 char_u *si_sofofr; /* SOFOFROM text */ |
| 5003 char_u *si_sofoto; /* SOFOTO text */ | |
| 625 | 5004 int si_nosugfile; /* NOSUGFILE item found */ |
| 698 | 5005 int si_nosplitsugs; /* NOSPLITSUGS item found */ |
| 323 | 5006 int si_followup; /* soundsalike: ? */ |
| 5007 int si_collapse; /* soundsalike: ? */ | |
| 625 | 5008 hashtab_T si_commonwords; /* hashtable for common words */ |
| 5009 time_t si_sugtime; /* timestamp for .sug file */ | |
| 323 | 5010 int si_rem_accents; /* soundsalike: remove accents */ |
| 5011 garray_T si_map; /* MAP info concatenated */ | |
| 495 | 5012 char_u *si_midword; /* MIDWORD chars or NULL */ |
| 483 | 5013 int si_compmax; /* max nr of words for compounding */ |
| 481 | 5014 int si_compminlen; /* minimal length for compounding */ |
| 483 | 5015 int si_compsylmax; /* max nr of syllables for compounding */ |
| 809 | 5016 int si_compoptions; /* COMP_ flags */ |
| 5017 garray_T si_comppat; /* CHECKCOMPOUNDPATTERN items, each stored as | |
| 5018 a string */ | |
| 481 | 5019 char_u *si_compflags; /* flags used for compounding */ |
| 492 | 5020 char_u si_nobreak; /* NOBREAK */ |
| 483 | 5021 char_u *si_syllable; /* syllable string */ |
| 339 | 5022 garray_T si_prefcond; /* table with conditions for postponed |
| 5023 * prefixes, each stored as a string */ | |
| 495 | 5024 int si_newprefID; /* current value for ah_newID */ |
| 500 | 5025 int si_newcompID; /* current value for compound ID */ |
| 300 | 5026 } spellinfo_T; |
| 249 | 5027 |
| 474 | 5028 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname)); |
| 1762 | 5029 static int is_aff_rule __ARGS((char_u **items, int itemcnt, char *rulename, int mincount)); |
| 822 | 5030 static void aff_process_flags __ARGS((afffile_T *affile, affentry_T *entry)); |
| 714 | 5031 static int spell_info_item __ARGS((char_u *s)); |
| 495 | 5032 static unsigned affitem2flag __ARGS((int flagtype, char_u *item, char_u *fname, int lnum)); |
| 5033 static unsigned get_affitem __ARGS((int flagtype, char_u **pp)); | |
| 5034 static void process_compflags __ARGS((spellinfo_T *spin, afffile_T *aff, char_u *compflags)); | |
| 500 | 5035 static void check_renumber __ARGS((spellinfo_T *spin)); |
| 495 | 5036 static int flag_in_afflist __ARGS((int flagtype, char_u *afflist, unsigned flag)); |
| 5037 static void aff_check_number __ARGS((int spinval, int affval, char *name)); | |
| 5038 static void aff_check_string __ARGS((char_u *spinval, char_u *affval, char *name)); | |
| 339 | 5039 static int str_equal __ARGS((char_u *s1, char_u *s2)); |
| 323 | 5040 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to)); |
| 5041 static int sal_to_bool __ARGS((char_u *s)); | |
| 300 | 5042 static void spell_free_aff __ARGS((afffile_T *aff)); |
| 474 | 5043 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile)); |
| 820 | 5044 static int get_affix_flags __ARGS((afffile_T *affile, char_u *afflist)); |
| 483 | 5045 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist)); |
| 495 | 5046 static void get_compflags __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist)); |
| 820 | 5047 static int store_aff_word __ARGS((spellinfo_T *spin, char_u *word, char_u *afflist, afffile_T *affile, hashtab_T *ht, hashtab_T *xht, int condit, int flags, char_u *pfxlist, int pfxlen)); |
| 474 | 5048 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname)); |
| 5049 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align)); | |
| 5050 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s)); | |
| 300 | 5051 static void free_blocks __ARGS((sblock_T *bl)); |
| 474 | 5052 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin)); |
| 483 | 5053 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix)); |
| 481 | 5054 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID)); |
| 474 | 5055 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin)); |
| 625 | 5056 static int deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node)); |
| 474 | 5057 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n)); |
| 5058 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root)); | |
| 5059 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot)); | |
| 300 | 5060 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2)); |
| 500 | 5061 static int write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname)); |
| 346 | 5062 static void clear_node __ARGS((wordnode_T *node)); |
| 1757 | 5063 static int put_node __ARGS((FILE *fd, wordnode_T *node, int idx, int regionmask, int prefixtree)); |
| 625 | 5064 static void spell_make_sugfile __ARGS((spellinfo_T *spin, char_u *wfname)); |
| 5065 static int sug_filltree __ARGS((spellinfo_T *spin, slang_T *slang)); | |
| 5066 static int sug_maketable __ARGS((spellinfo_T *spin)); | |
| 5067 static int sug_filltable __ARGS((spellinfo_T *spin, wordnode_T *node, int startwordnr, garray_T *gap)); | |
| 5068 static int offset2bytes __ARGS((int nr, char_u *buf)); | |
| 5069 static int bytes2offset __ARGS((char_u **pp)); | |
| 5070 static void sug_write __ARGS((spellinfo_T *spin, char_u *fname)); | |
| 3263 | 5071 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int over_write, int added_word)); |
| 625 | 5072 static void spell_message __ARGS((spellinfo_T *spin, char_u *str)); |
| 310 | 5073 static void init_spellfile __ARGS((void)); |
| 236 | 5074 |
| 455 | 5075 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags, |
| 5076 * but it must be negative to indicate the prefix tree to tree_add_word(). | |
| 5077 * Use a negative number with the lower 8 bits zero. */ | |
| 5078 #define PFX_FLAGS -256 | |
| 390 | 5079 |
| 820 | 5080 /* flags for "condit" argument of store_aff_word() */ |
| 5081 #define CONDIT_COMB 1 /* affix must combine */ | |
| 5082 #define CONDIT_CFIX 2 /* affix must have CIRCUMFIX flag */ | |
| 5083 #define CONDIT_SUF 4 /* add a suffix for matching flags */ | |
| 5084 #define CONDIT_AFF 8 /* word already has an affix */ | |
| 5085 | |
| 483 | 5086 /* |
| 5087 * Tunable parameters for when the tree is compressed. See 'mkspellmem'. | |
| 5088 */ | |
| 5089 static long compress_start = 30000; /* memory / SBLOCKSIZE */ | |
| 5090 static long compress_inc = 100; /* memory / SBLOCKSIZE */ | |
| 5091 static long compress_added = 500000; /* word count */ | |
| 5092 | |
| 469 | 5093 #ifdef SPELL_PRINTTREE |
| 5094 /* | |
| 5095 * For debugging the tree code: print the current tree in a (more or less) | |
| 5096 * readable format, so that we can see what happens when adding a word and/or | |
| 5097 * compressing the tree. | |
| 5098 * Based on code from Olaf Seibert. | |
| 5099 */ | |
| 5100 #define PRINTLINESIZE 1000 | |
| 5101 #define PRINTWIDTH 6 | |
| 5102 | |
| 5103 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \ | |
| 5104 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2) | |
| 5105 | |
| 5106 static char line1[PRINTLINESIZE]; | |
| 5107 static char line2[PRINTLINESIZE]; | |
| 5108 static char line3[PRINTLINESIZE]; | |
| 5109 | |
| 5110 static void | |
| 5111 spell_clear_flags(wordnode_T *node) | |
| 5112 { | |
| 5113 wordnode_T *np; | |
| 5114 | |
| 5115 for (np = node; np != NULL; np = np->wn_sibling) | |
| 5116 { | |
| 5117 np->wn_u1.index = FALSE; | |
| 5118 spell_clear_flags(np->wn_child); | |
| 5119 } | |
| 5120 } | |
| 5121 | |
| 5122 static void | |
| 5123 spell_print_node(wordnode_T *node, int depth) | |
| 5124 { | |
| 5125 if (node->wn_u1.index) | |
| 5126 { | |
| 5127 /* Done this node before, print the reference. */ | |
| 5128 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0); | |
| 5129 PRINTSOME(line2, depth, " ", 0, 0); | |
| 5130 PRINTSOME(line3, depth, " ", 0, 0); | |
| 5131 msg(line1); | |
| 5132 msg(line2); | |
| 5133 msg(line3); | |
| 5134 } | |
| 5135 else | |
| 5136 { | |
| 5137 node->wn_u1.index = TRUE; | |
| 5138 | |
| 5139 if (node->wn_byte != NUL) | |
| 5140 { | |
| 5141 if (node->wn_child != NULL) | |
| 5142 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0); | |
| 5143 else | |
| 5144 /* Cannot happen? */ | |
| 5145 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0); | |
| 5146 } | |
| 5147 else | |
| 5148 PRINTSOME(line1, depth, " $ ", 0, 0); | |
| 5149 | |
| 5150 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs); | |
| 5151 | |
| 5152 if (node->wn_sibling != NULL) | |
| 5153 PRINTSOME(line3, depth, " | ", 0, 0); | |
| 5154 else | |
| 5155 PRINTSOME(line3, depth, " ", 0, 0); | |
| 5156 | |
| 5157 if (node->wn_byte == NUL) | |
| 5158 { | |
| 5159 msg(line1); | |
| 5160 msg(line2); | |
| 5161 msg(line3); | |
| 5162 } | |
| 5163 | |
| 5164 /* do the children */ | |
| 5165 if (node->wn_byte != NUL && node->wn_child != NULL) | |
| 5166 spell_print_node(node->wn_child, depth + 1); | |
| 5167 | |
| 5168 /* do the siblings */ | |
| 5169 if (node->wn_sibling != NULL) | |
| 5170 { | |
| 5171 /* get rid of all parent details except | */ | |
| 5172 STRCPY(line1, line3); | |
| 5173 STRCPY(line2, line3); | |
| 5174 spell_print_node(node->wn_sibling, depth); | |
| 5175 } | |
| 5176 } | |
| 5177 } | |
| 5178 | |
| 5179 static void | |
| 5180 spell_print_tree(wordnode_T *root) | |
| 5181 { | |
| 5182 if (root != NULL) | |
| 5183 { | |
| 5184 /* Clear the "wn_u1.index" fields, used to remember what has been | |
| 5185 * done. */ | |
| 5186 spell_clear_flags(root); | |
| 5187 | |
| 5188 /* Recursively print the tree. */ | |
| 5189 spell_print_node(root, 0); | |
| 5190 } | |
| 5191 } | |
| 5192 #endif /* SPELL_PRINTTREE */ | |
| 5193 | |
| 236 | 5194 /* |
| 323 | 5195 * Read the affix file "fname". |
| 316 | 5196 * Returns an afffile_T, NULL for complete failure. |
| 236 | 5197 */ |
| 5198 static afffile_T * | |
| 474 | 5199 spell_read_aff(spin, fname) |
| 5200 spellinfo_T *spin; | |
| 236 | 5201 char_u *fname; |
| 5202 { | |
| 5203 FILE *fd; | |
| 5204 afffile_T *aff; | |
| 5205 char_u rline[MAXLINELEN]; | |
| 5206 char_u *line; | |
| 5207 char_u *pc = NULL; | |
| 625 | 5208 #define MAXITEMCNT 30 |
| 334 | 5209 char_u *(items[MAXITEMCNT]); |
| 236 | 5210 int itemcnt; |
| 5211 char_u *p; | |
| 5212 int lnum = 0; | |
| 5213 affheader_T *cur_aff = NULL; | |
| 495 | 5214 int did_postpone_prefix = FALSE; |
| 236 | 5215 int aff_todo = 0; |
| 5216 hashtab_T *tp; | |
| 255 | 5217 char_u *low = NULL; |
| 5218 char_u *fol = NULL; | |
| 5219 char_u *upp = NULL; | |
| 323 | 5220 int do_rep; |
| 625 | 5221 int do_repsal; |
| 323 | 5222 int do_sal; |
| 944 | 5223 int do_mapline; |
| 323 | 5224 int found_map = FALSE; |
| 324 | 5225 hashitem_T *hi; |
| 455 | 5226 int l; |
| 495 | 5227 int compminlen = 0; /* COMPOUNDMIN value */ |
| 5228 int compsylmax = 0; /* COMPOUNDSYLMAX value */ | |
| 809 | 5229 int compoptions = 0; /* COMP_ flags */ |
| 5230 int compmax = 0; /* COMPOUNDWORDMAX value */ | |
| 714 | 5231 char_u *compflags = NULL; /* COMPOUNDFLAG and COMPOUNDRULE |
| 495 | 5232 concatenated */ |
| 5233 char_u *midword = NULL; /* MIDWORD value */ | |
| 5234 char_u *syllable = NULL; /* SYLLABLE value */ | |
| 5235 char_u *sofofrom = NULL; /* SOFOFROM value */ | |
| 5236 char_u *sofoto = NULL; /* SOFOTO value */ | |
| 236 | 5237 |
| 300 | 5238 /* |
| 5239 * Open the file. | |
| 5240 */ | |
| 310 | 5241 fd = mch_fopen((char *)fname, "r"); |
| 236 | 5242 if (fd == NULL) |
| 5243 { | |
| 5244 EMSG2(_(e_notopen), fname); | |
| 5245 return NULL; | |
| 5246 } | |
| 5247 | |
| 625 | 5248 vim_snprintf((char *)IObuff, IOSIZE, _("Reading affix file %s ..."), fname); |
| 5249 spell_message(spin, IObuff); | |
| 236 | 5250 |
| 323 | 5251 /* Only do REP lines when not done in another .aff file already. */ |
| 5252 do_rep = spin->si_rep.ga_len == 0; | |
| 5253 | |
| 625 | 5254 /* Only do REPSAL lines when not done in another .aff file already. */ |
| 5255 do_repsal = spin->si_repsal.ga_len == 0; | |
| 5256 | |
| 323 | 5257 /* Only do SAL lines when not done in another .aff file already. */ |
| 5258 do_sal = spin->si_sal.ga_len == 0; | |
| 5259 | |
| 5260 /* Only do MAP lines when not done in another .aff file already. */ | |
| 944 | 5261 do_mapline = spin->si_map.ga_len == 0; |
| 323 | 5262 |
| 300 | 5263 /* |
| 5264 * Allocate and init the afffile_T structure. | |
| 5265 */ | |
| 474 | 5266 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE); |
| 236 | 5267 if (aff == NULL) |
| 840 | 5268 { |
| 5269 fclose(fd); | |
| 236 | 5270 return NULL; |
| 840 | 5271 } |
| 236 | 5272 hash_init(&aff->af_pref); |
| 5273 hash_init(&aff->af_suff); | |
| 495 | 5274 hash_init(&aff->af_comp); |
| 236 | 5275 |
| 5276 /* | |
| 5277 * Read all the lines in the file one by one. | |
| 5278 */ | |
| 255 | 5279 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int) |
| 236 | 5280 { |
| 255 | 5281 line_breakcheck(); |
| 236 | 5282 ++lnum; |
| 5283 | |
| 5284 /* Skip comment lines. */ | |
| 5285 if (*rline == '#') | |
| 5286 continue; | |
| 5287 | |
| 5288 /* Convert from "SET" to 'encoding' when needed. */ | |
| 5289 vim_free(pc); | |
| 310 | 5290 #ifdef FEAT_MBYTE |
| 300 | 5291 if (spin->si_conv.vc_type != CONV_NONE) |
| 236 | 5292 { |
| 300 | 5293 pc = string_convert(&spin->si_conv, rline, NULL); |
| 255 | 5294 if (pc == NULL) |
| 5295 { | |
| 5296 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"), | |
| 5297 fname, lnum, rline); | |
| 5298 continue; | |
| 5299 } | |
| 236 | 5300 line = pc; |
| 5301 } | |
| 5302 else | |
| 310 | 5303 #endif |
| 236 | 5304 { |
| 5305 pc = NULL; | |
| 5306 line = rline; | |
| 5307 } | |
| 5308 | |
| 5309 /* Split the line up in white separated items. Put a NUL after each | |
| 5310 * item. */ | |
| 5311 itemcnt = 0; | |
| 5312 for (p = line; ; ) | |
| 5313 { | |
| 5314 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */ | |
| 5315 ++p; | |
| 5316 if (*p == NUL) | |
| 5317 break; | |
| 334 | 5318 if (itemcnt == MAXITEMCNT) /* too many items */ |
| 300 | 5319 break; |
| 236 | 5320 items[itemcnt++] = p; |
| 714 | 5321 /* A few items have arbitrary text argument, don't split them. */ |
| 5322 if (itemcnt == 2 && spell_info_item(items[0])) | |
| 5323 while (*p >= ' ' || *p == TAB) /* skip until CR/NL */ | |
| 5324 ++p; | |
| 5325 else | |
| 5326 while (*p > ' ') /* skip until white space or CR/NL */ | |
| 5327 ++p; | |
| 236 | 5328 if (*p == NUL) |
| 5329 break; | |
| 5330 *p++ = NUL; | |
| 5331 } | |
| 5332 | |
| 5333 /* Handle non-empty lines. */ | |
| 5334 if (itemcnt > 0) | |
| 5335 { | |
| 1762 | 5336 if (is_aff_rule(items, itemcnt, "SET", 2) && aff->af_enc == NULL) |
| 236 | 5337 { |
| 310 | 5338 #ifdef FEAT_MBYTE |
| 300 | 5339 /* Setup for conversion from "ENC" to 'encoding'. */ |
| 5340 aff->af_enc = enc_canonize(items[1]); | |
| 5341 if (aff->af_enc != NULL && !spin->si_ascii | |
| 5342 && convert_setup(&spin->si_conv, aff->af_enc, | |
| 5343 p_enc) == FAIL) | |
| 5344 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"), | |
| 5345 fname, aff->af_enc, p_enc); | |
| 375 | 5346 spin->si_conv.vc_fail = TRUE; |
| 310 | 5347 #else |
| 5348 smsg((char_u *)_("Conversion in %s not supported"), fname); | |
| 5349 #endif | |
| 236 | 5350 } |
| 1762 | 5351 else if (is_aff_rule(items, itemcnt, "FLAG", 2) |
| 495 | 5352 && aff->af_flagtype == AFT_CHAR) |
| 5353 { | |
| 5354 if (STRCMP(items[1], "long") == 0) | |
| 498 | 5355 aff->af_flagtype = AFT_LONG; |
| 495 | 5356 else if (STRCMP(items[1], "num") == 0) |
| 498 | 5357 aff->af_flagtype = AFT_NUM; |
| 5358 else if (STRCMP(items[1], "caplong") == 0) | |
| 5359 aff->af_flagtype = AFT_CAPLONG; | |
| 495 | 5360 else |
| 5361 smsg((char_u *)_("Invalid value for FLAG in %s line %d: %s"), | |
| 5362 fname, lnum, items[1]); | |
| 621 | 5363 if (aff->af_rare != 0 |
| 5364 || aff->af_keepcase != 0 | |
| 5365 || aff->af_bad != 0 | |
| 500 | 5366 || aff->af_needaffix != 0 |
| 820 | 5367 || aff->af_circumfix != 0 |
| 500 | 5368 || aff->af_needcomp != 0 |
| 809 | 5369 || aff->af_comproot != 0 |
| 698 | 5370 || aff->af_nosuggest != 0 |
| 500 | 5371 || compflags != NULL |
| 495 | 5372 || aff->af_suff.ht_used > 0 |
| 5373 || aff->af_pref.ht_used > 0) | |
| 5374 smsg((char_u *)_("FLAG after using flags in %s line %d: %s"), | |
| 5375 fname, lnum, items[1]); | |
| 5376 } | |
| 714 | 5377 else if (spell_info_item(items[0])) |
| 5378 { | |
| 5379 p = (char_u *)getroom(spin, | |
| 5380 (spin->si_info == NULL ? 0 : STRLEN(spin->si_info)) | |
| 5381 + STRLEN(items[0]) | |
| 5382 + STRLEN(items[1]) + 3, FALSE); | |
| 5383 if (p != NULL) | |
| 5384 { | |
| 5385 if (spin->si_info != NULL) | |
| 5386 { | |
| 5387 STRCPY(p, spin->si_info); | |
| 5388 STRCAT(p, "\n"); | |
| 5389 } | |
| 5390 STRCAT(p, items[0]); | |
| 5391 STRCAT(p, " "); | |
| 5392 STRCAT(p, items[1]); | |
| 5393 spin->si_info = p; | |
| 5394 } | |
| 5395 } | |
| 1762 | 5396 else if (is_aff_rule(items, itemcnt, "MIDWORD", 2) |
| 495 | 5397 && midword == NULL) |
| 5398 { | |
| 5399 midword = getroom_save(spin, items[1]); | |
| 366 | 5400 } |
| 1762 | 5401 else if (is_aff_rule(items, itemcnt, "TRY", 2)) |
| 300 | 5402 { |
| 323 | 5403 /* ignored, we look in the tree for what chars may appear */ |
| 300 | 5404 } |
| 621 | 5405 /* TODO: remove "RAR" later */ |
| 1762 | 5406 else if ((is_aff_rule(items, itemcnt, "RAR", 2) |
| 5407 || is_aff_rule(items, itemcnt, "RARE", 2)) | |
| 5408 && aff->af_rare == 0) | |
| 621 | 5409 { |
| 5410 aff->af_rare = affitem2flag(aff->af_flagtype, items[1], | |
| 495 | 5411 fname, lnum); |
| 307 | 5412 } |
| 621 | 5413 /* TODO: remove "KEP" later */ |
| 1762 | 5414 else if ((is_aff_rule(items, itemcnt, "KEP", 2) |
| 5415 || is_aff_rule(items, itemcnt, "KEEPCASE", 2)) | |
| 621 | 5416 && aff->af_keepcase == 0) |
| 5417 { | |
| 5418 aff->af_keepcase = affitem2flag(aff->af_flagtype, items[1], | |
| 495 | 5419 fname, lnum); |
| 307 | 5420 } |
| 1762 | 5421 else if ((is_aff_rule(items, itemcnt, "BAD", 2) |
| 5422 || is_aff_rule(items, itemcnt, "FORBIDDENWORD", 2)) | |
| 5423 && aff->af_bad == 0) | |
| 346 | 5424 { |
| 495 | 5425 aff->af_bad = affitem2flag(aff->af_flagtype, items[1], |
| 5426 fname, lnum); | |
| 346 | 5427 } |
| 1762 | 5428 else if (is_aff_rule(items, itemcnt, "NEEDAFFIX", 2) |
| 483 | 5429 && aff->af_needaffix == 0) |
| 5430 { | |
| 495 | 5431 aff->af_needaffix = affitem2flag(aff->af_flagtype, items[1], |
| 5432 fname, lnum); | |
| 483 | 5433 } |
| 1762 | 5434 else if (is_aff_rule(items, itemcnt, "CIRCUMFIX", 2) |
| 820 | 5435 && aff->af_circumfix == 0) |
| 5436 { | |
| 5437 aff->af_circumfix = affitem2flag(aff->af_flagtype, items[1], | |
| 5438 fname, lnum); | |
| 5439 } | |
| 1762 | 5440 else if (is_aff_rule(items, itemcnt, "NOSUGGEST", 2) |
| 698 | 5441 && aff->af_nosuggest == 0) |
| 5442 { | |
| 5443 aff->af_nosuggest = affitem2flag(aff->af_flagtype, items[1], | |
| 5444 fname, lnum); | |
| 5445 } | |
| 1762 | 5446 else if ((is_aff_rule(items, itemcnt, "NEEDCOMPOUND", 2) |
| 5447 || is_aff_rule(items, itemcnt, "ONLYINCOMPOUND", 2)) | |
| 500 | 5448 && aff->af_needcomp == 0) |
| 5449 { | |
| 5450 aff->af_needcomp = affitem2flag(aff->af_flagtype, items[1], | |
| 5451 fname, lnum); | |
| 5452 } | |
| 1762 | 5453 else if (is_aff_rule(items, itemcnt, "COMPOUNDROOT", 2) |
| 809 | 5454 && aff->af_comproot == 0) |
| 5455 { | |
| 5456 aff->af_comproot = affitem2flag(aff->af_flagtype, items[1], | |
| 5457 fname, lnum); | |
| 5458 } | |
| 1762 | 5459 else if (is_aff_rule(items, itemcnt, "COMPOUNDFORBIDFLAG", 2) |
| 5460 && aff->af_compforbid == 0) | |
| 809 | 5461 { |
| 5462 aff->af_compforbid = affitem2flag(aff->af_flagtype, items[1], | |
| 5463 fname, lnum); | |
| 822 | 5464 if (aff->af_pref.ht_used > 0) |
| 5465 smsg((char_u *)_("Defining COMPOUNDFORBIDFLAG after PFX item may give wrong results in %s line %d"), | |
| 5466 fname, lnum); | |
| 809 | 5467 } |
| 1762 | 5468 else if (is_aff_rule(items, itemcnt, "COMPOUNDPERMITFLAG", 2) |
| 5469 && aff->af_comppermit == 0) | |
| 809 | 5470 { |
| 5471 aff->af_comppermit = affitem2flag(aff->af_flagtype, items[1], | |
| 5472 fname, lnum); | |
| 822 | 5473 if (aff->af_pref.ht_used > 0) |
| 5474 smsg((char_u *)_("Defining COMPOUNDPERMITFLAG after PFX item may give wrong results in %s line %d"), | |
| 5475 fname, lnum); | |
| 809 | 5476 } |
| 1762 | 5477 else if (is_aff_rule(items, itemcnt, "COMPOUNDFLAG", 2) |
| 495 | 5478 && compflags == NULL) |
| 5479 { | |
| 714 | 5480 /* Turn flag "c" into COMPOUNDRULE compatible string "c+", |
| 495 | 5481 * "Na" into "Na+", "1234" into "1234+". */ |
| 5482 p = getroom(spin, STRLEN(items[1]) + 2, FALSE); | |
| 483 | 5483 if (p != NULL) |
| 5484 { | |
| 495 | 5485 STRCPY(p, items[1]); |
| 5486 STRCAT(p, "+"); | |
| 5487 compflags = p; | |
| 5488 } | |
| 481 | 5489 } |
| 1762 | 5490 else if (is_aff_rule(items, itemcnt, "COMPOUNDRULES", 2)) |
| 5491 { | |
| 5492 /* We don't use the count, but do check that it's a number and | |
| 5493 * not COMPOUNDRULE mistyped. */ | |
| 5494 if (atoi((char *)items[1]) == 0) | |
| 5495 smsg((char_u *)_("Wrong COMPOUNDRULES value in %s line %d: %s"), | |
| 5496 fname, lnum, items[1]); | |
| 5497 } | |
| 5498 else if (is_aff_rule(items, itemcnt, "COMPOUNDRULE", 2)) | |
| 483 | 5499 { |
| 2691 | 5500 /* Don't use the first rule if it is a number. */ |
| 5501 if (compflags != NULL || *skipdigits(items[1]) != NUL) | |
| 5502 { | |
| 5503 /* Concatenate this string to previously defined ones, | |
| 5504 * using a slash to separate them. */ | |
| 5505 l = (int)STRLEN(items[1]) + 1; | |
| 495 | 5506 if (compflags != NULL) |
| 2691 | 5507 l += (int)STRLEN(compflags) + 1; |
| 5508 p = getroom(spin, l, FALSE); | |
| 5509 if (p != NULL) | |
| 483 | 5510 { |
| 2691 | 5511 if (compflags != NULL) |
| 5512 { | |
| 5513 STRCPY(p, compflags); | |
| 5514 STRCAT(p, "/"); | |
| 5515 } | |
| 5516 STRCAT(p, items[1]); | |
| 5517 compflags = p; | |
| 483 | 5518 } |
| 5519 } | |
| 5520 } | |
| 1762 | 5521 else if (is_aff_rule(items, itemcnt, "COMPOUNDWORDMAX", 2) |
| 495 | 5522 && compmax == 0) |
| 5523 { | |
| 5524 compmax = atoi((char *)items[1]); | |
| 5525 if (compmax == 0) | |
| 809 | 5526 smsg((char_u *)_("Wrong COMPOUNDWORDMAX value in %s line %d: %s"), |
| 483 | 5527 fname, lnum, items[1]); |
| 481 | 5528 } |
| 1762 | 5529 else if (is_aff_rule(items, itemcnt, "COMPOUNDMIN", 2) |
| 495 | 5530 && compminlen == 0) |
| 5531 { | |
| 5532 compminlen = atoi((char *)items[1]); | |
| 5533 if (compminlen == 0) | |
| 481 | 5534 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"), |
| 5535 fname, lnum, items[1]); | |
| 5536 } | |
| 1762 | 5537 else if (is_aff_rule(items, itemcnt, "COMPOUNDSYLMAX", 2) |
| 495 | 5538 && compsylmax == 0) |
| 5539 { | |
| 5540 compsylmax = atoi((char *)items[1]); | |
| 5541 if (compsylmax == 0) | |
| 483 | 5542 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"), |
| 5543 fname, lnum, items[1]); | |
| 5544 } | |
| 1762 | 5545 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDDUP", 1)) |
| 809 | 5546 { |
| 5547 compoptions |= COMP_CHECKDUP; | |
| 5548 } | |
| 1762 | 5549 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDREP", 1)) |
| 809 | 5550 { |
| 5551 compoptions |= COMP_CHECKREP; | |
| 5552 } | |
| 1762 | 5553 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDCASE", 1)) |
| 809 | 5554 { |
| 5555 compoptions |= COMP_CHECKCASE; | |
| 5556 } | |
| 1762 | 5557 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDTRIPLE", 1)) |
| 809 | 5558 { |
| 5559 compoptions |= COMP_CHECKTRIPLE; | |
| 5560 } | |
| 1762 | 5561 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDPATTERN", 2)) |
| 809 | 5562 { |
| 5563 if (atoi((char *)items[1]) == 0) | |
| 5564 smsg((char_u *)_("Wrong CHECKCOMPOUNDPATTERN value in %s line %d: %s"), | |
| 5565 fname, lnum, items[1]); | |
| 5566 } | |
| 1762 | 5567 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDPATTERN", 3)) |
| 809 | 5568 { |
| 5569 garray_T *gap = &spin->si_comppat; | |
| 5570 int i; | |
| 5571 | |
| 5572 /* Only add the couple if it isn't already there. */ | |
| 5573 for (i = 0; i < gap->ga_len - 1; i += 2) | |
| 5574 if (STRCMP(((char_u **)(gap->ga_data))[i], items[1]) == 0 | |
| 5575 && STRCMP(((char_u **)(gap->ga_data))[i + 1], | |
| 5576 items[2]) == 0) | |
| 5577 break; | |
| 5578 if (i >= gap->ga_len && ga_grow(gap, 2) == OK) | |
| 5579 { | |
| 5580 ((char_u **)(gap->ga_data))[gap->ga_len++] | |
| 5581 = getroom_save(spin, items[1]); | |
| 5582 ((char_u **)(gap->ga_data))[gap->ga_len++] | |
| 5583 = getroom_save(spin, items[2]); | |
| 5584 } | |
| 5585 } | |
| 1762 | 5586 else if (is_aff_rule(items, itemcnt, "SYLLABLE", 2) |
| 495 | 5587 && syllable == NULL) |
| 5588 { | |
| 5589 syllable = getroom_save(spin, items[1]); | |
| 483 | 5590 } |
| 1762 | 5591 else if (is_aff_rule(items, itemcnt, "NOBREAK", 1)) |
| 492 | 5592 { |
| 5593 spin->si_nobreak = TRUE; | |
| 5594 } | |
| 1762 | 5595 else if (is_aff_rule(items, itemcnt, "NOSPLITSUGS", 1)) |
| 698 | 5596 { |
| 5597 spin->si_nosplitsugs = TRUE; | |
| 5598 } | |
| 1762 | 5599 else if (is_aff_rule(items, itemcnt, "NOSUGFILE", 1)) |
| 625 | 5600 { |
| 5601 spin->si_nosugfile = TRUE; | |
| 5602 } | |
| 1762 | 5603 else if (is_aff_rule(items, itemcnt, "PFXPOSTPONE", 1)) |
| 339 | 5604 { |
| 5605 aff->af_pfxpostpone = TRUE; | |
| 5606 } | |
| 6224 | 5607 else if (is_aff_rule(items, itemcnt, "IGNOREEXTRA", 1)) |
| 5608 { | |
| 5609 aff->af_ignoreextra = TRUE; | |
| 5610 } | |
| 236 | 5611 else if ((STRCMP(items[0], "PFX") == 0 |
| 5612 || STRCMP(items[0], "SFX") == 0) | |
| 5613 && aff_todo == 0 | |
| 334 | 5614 && itemcnt >= 4) |
| 236 | 5615 { |
| 498 | 5616 int lasti = 4; |
| 5617 char_u key[AH_KEY_LEN]; | |
| 5618 | |
| 5619 if (*items[0] == 'P') | |
| 5620 tp = &aff->af_pref; | |
| 5621 else | |
| 5622 tp = &aff->af_suff; | |
| 5623 | |
| 5624 /* Myspell allows the same affix name to be used multiple | |
| 5625 * times. The affix files that do this have an undocumented | |
| 5626 * "S" flag on all but the last block, thus we check for that | |
| 5627 * and store it in ah_follows. */ | |
| 5628 vim_strncpy(key, items[1], AH_KEY_LEN - 1); | |
| 5629 hi = hash_find(tp, key); | |
| 5630 if (!HASHITEM_EMPTY(hi)) | |
| 5631 { | |
| 5632 cur_aff = HI2AH(hi); | |
| 5633 if (cur_aff->ah_combine != (*items[2] == 'Y')) | |
| 5634 smsg((char_u *)_("Different combining flag in continued affix block in %s line %d: %s"), | |
| 5635 fname, lnum, items[1]); | |
| 5636 if (!cur_aff->ah_follows) | |
| 5637 smsg((char_u *)_("Duplicate affix in %s line %d: %s"), | |
| 5638 fname, lnum, items[1]); | |
| 5639 } | |
| 5640 else | |
| 5641 { | |
| 5642 /* New affix letter. */ | |
| 5643 cur_aff = (affheader_T *)getroom(spin, | |
| 5644 sizeof(affheader_T), TRUE); | |
| 5645 if (cur_aff == NULL) | |
| 5646 break; | |
| 5647 cur_aff->ah_flag = affitem2flag(aff->af_flagtype, items[1], | |
| 5648 fname, lnum); | |
| 5649 if (cur_aff->ah_flag == 0 || STRLEN(items[1]) >= AH_KEY_LEN) | |
| 5650 break; | |
| 5651 if (cur_aff->ah_flag == aff->af_bad | |
| 621 | 5652 || cur_aff->ah_flag == aff->af_rare |
| 5653 || cur_aff->ah_flag == aff->af_keepcase | |
| 500 | 5654 || cur_aff->ah_flag == aff->af_needaffix |
| 820 | 5655 || cur_aff->ah_flag == aff->af_circumfix |
| 698 | 5656 || cur_aff->ah_flag == aff->af_nosuggest |
| 809 | 5657 || cur_aff->ah_flag == aff->af_needcomp |
| 5658 || cur_aff->ah_flag == aff->af_comproot) | |
| 698 | 5659 smsg((char_u *)_("Affix also used for BAD/RARE/KEEPCASE/NEEDAFFIX/NEEDCOMPOUND/NOSUGGEST in %s line %d: %s"), |
| 498 | 5660 fname, lnum, items[1]); |
| 5661 STRCPY(cur_aff->ah_key, items[1]); | |
| 5662 hash_add(tp, cur_aff->ah_key); | |
| 5663 | |
| 5664 cur_aff->ah_combine = (*items[2] == 'Y'); | |
| 5665 } | |
| 5666 | |
| 5667 /* Check for the "S" flag, which apparently means that another | |
| 5668 * block with the same affix name is following. */ | |
| 5669 if (itemcnt > lasti && STRCMP(items[lasti], "S") == 0) | |
| 5670 { | |
| 5671 ++lasti; | |
| 5672 cur_aff->ah_follows = TRUE; | |
| 5673 } | |
| 5674 else | |
| 5675 cur_aff->ah_follows = FALSE; | |
| 5676 | |
| 334 | 5677 /* Myspell allows extra text after the item, but that might |
| 5678 * mean mistakes go unnoticed. Require a comment-starter. */ | |
| 498 | 5679 if (itemcnt > lasti && *items[lasti] != '#') |
| 809 | 5680 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]); |
| 334 | 5681 |
| 498 | 5682 if (STRCMP(items[2], "Y") != 0 && STRCMP(items[2], "N") != 0) |
| 236 | 5683 smsg((char_u *)_("Expected Y or N in %s line %d: %s"), |
| 5684 fname, lnum, items[2]); | |
| 339 | 5685 |
| 498 | 5686 if (*items[0] == 'P' && aff->af_pfxpostpone) |
| 5687 { | |
| 5688 if (cur_aff->ah_newID == 0) | |
| 495 | 5689 { |
| 5690 /* Use a new number in the .spl file later, to be able | |
| 5691 * to handle multiple .aff files. */ | |
| 500 | 5692 check_renumber(spin); |
| 495 | 5693 cur_aff->ah_newID = ++spin->si_newprefID; |
| 5694 | |
| 5695 /* We only really use ah_newID if the prefix is | |
| 5696 * postponed. We know that only after handling all | |
| 5697 * the items. */ | |
| 5698 did_postpone_prefix = FALSE; | |
| 5699 } | |
| 498 | 5700 else |
| 5701 /* Did use the ID in a previous block. */ | |
| 5702 did_postpone_prefix = TRUE; | |
| 5703 } | |
| 5704 | |
| 300 | 5705 aff_todo = atoi((char *)items[3]); |
| 236 | 5706 } |
| 5707 else if ((STRCMP(items[0], "PFX") == 0 | |
| 5708 || STRCMP(items[0], "SFX") == 0) | |
| 5709 && aff_todo > 0 | |
| 5710 && STRCMP(cur_aff->ah_key, items[1]) == 0 | |
| 334 | 5711 && itemcnt >= 5) |
| 236 | 5712 { |
| 5713 affentry_T *aff_entry; | |
| 455 | 5714 int upper = FALSE; |
| 366 | 5715 int lasti = 5; |
| 5716 | |
| 334 | 5717 /* Myspell allows extra text after the item, but that might |
| 6224 | 5718 * mean mistakes go unnoticed. Require a comment-starter, |
| 5719 * unless IGNOREEXTRA is used. Hunspell uses a "-" item. */ | |
| 5720 if (itemcnt > lasti | |
| 5721 && !aff->af_ignoreextra | |
| 5722 && *items[lasti] != '#' | |
| 809 | 5723 && (STRCMP(items[lasti], "-") != 0 |
| 5724 || itemcnt != lasti + 1)) | |
| 482 | 5725 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]); |
| 334 | 5726 |
| 236 | 5727 /* New item for an affix letter. */ |
| 5728 --aff_todo; | |
| 474 | 5729 aff_entry = (affentry_T *)getroom(spin, |
| 447 | 5730 sizeof(affentry_T), TRUE); |
| 236 | 5731 if (aff_entry == NULL) |
| 5732 break; | |
| 240 | 5733 |
| 236 | 5734 if (STRCMP(items[2], "0") != 0) |
| 474 | 5735 aff_entry->ae_chop = getroom_save(spin, items[2]); |
| 236 | 5736 if (STRCMP(items[3], "0") != 0) |
| 809 | 5737 { |
| 474 | 5738 aff_entry->ae_add = getroom_save(spin, items[3]); |
| 236 | 5739 |
| 822 | 5740 /* Recognize flags on the affix: abcd/XYZ */ |
| 809 | 5741 aff_entry->ae_flags = vim_strchr(aff_entry->ae_add, '/'); |
| 5742 if (aff_entry->ae_flags != NULL) | |
| 822 | 5743 { |
| 809 | 5744 *aff_entry->ae_flags++ = NUL; |
| 822 | 5745 aff_process_flags(aff, aff_entry); |
| 5746 } | |
| 809 | 5747 } |
| 5748 | |
| 300 | 5749 /* Don't use an affix entry with non-ASCII characters when |
| 5750 * "spin->si_ascii" is TRUE. */ | |
| 5751 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop) | |
| 240 | 5752 || has_non_ascii(aff_entry->ae_add))) |
| 5753 { | |
| 5754 aff_entry->ae_next = cur_aff->ah_first; | |
| 5755 cur_aff->ah_first = aff_entry; | |
| 300 | 5756 |
| 5757 if (STRCMP(items[4], ".") != 0) | |
| 5758 { | |
| 5759 char_u buf[MAXLINELEN]; | |
| 5760 | |
| 474 | 5761 aff_entry->ae_cond = getroom_save(spin, items[4]); |
| 300 | 5762 if (*items[0] == 'P') |
| 5763 sprintf((char *)buf, "^%s", items[4]); | |
| 5764 else | |
| 5765 sprintf((char *)buf, "%s$", items[4]); | |
| 5766 aff_entry->ae_prog = vim_regcomp(buf, | |
| 481 | 5767 RE_MAGIC + RE_STRING + RE_STRICT); |
| 5768 if (aff_entry->ae_prog == NULL) | |
| 5769 smsg((char_u *)_("Broken condition in %s line %d: %s"), | |
| 5770 fname, lnum, items[4]); | |
| 300 | 5771 } |
| 339 | 5772 |
| 5773 /* For postponed prefixes we need an entry in si_prefcond | |
| 809 | 5774 * for the condition. Use an existing one if possible. |
| 822 | 5775 * Can't be done for an affix with flags, ignoring |
| 5776 * COMPOUNDFORBIDFLAG and COMPOUNDPERMITFLAG. */ | |
| 809 | 5777 if (*items[0] == 'P' && aff->af_pfxpostpone |
| 5778 && aff_entry->ae_flags == NULL) | |
| 339 | 5779 { |
| 455 | 5780 /* When the chop string is one lower-case letter and |
| 5781 * the add string ends in the upper-case letter we set | |
| 5782 * the "upper" flag, clear "ae_chop" and remove the | |
| 5783 * letters from "ae_add". The condition must either | |
| 5784 * be empty or start with the same letter. */ | |
| 5785 if (aff_entry->ae_chop != NULL | |
| 5786 && aff_entry->ae_add != NULL | |
| 5787 #ifdef FEAT_MBYTE | |
| 474 | 5788 && aff_entry->ae_chop[(*mb_ptr2len)( |
| 455 | 5789 aff_entry->ae_chop)] == NUL |
| 5790 #else | |
| 5791 && aff_entry->ae_chop[1] == NUL | |
| 5792 #endif | |
| 5793 ) | |
| 339 | 5794 { |
| 455 | 5795 int c, c_up; |
| 5796 | |
| 5797 c = PTR2CHAR(aff_entry->ae_chop); | |
| 5798 c_up = SPELL_TOUPPER(c); | |
| 5799 if (c_up != c | |
| 5800 && (aff_entry->ae_cond == NULL | |
| 5801 || PTR2CHAR(aff_entry->ae_cond) == c)) | |
| 5802 { | |
| 5803 p = aff_entry->ae_add | |
| 5804 + STRLEN(aff_entry->ae_add); | |
| 5805 mb_ptr_back(aff_entry->ae_add, p); | |
| 5806 if (PTR2CHAR(p) == c_up) | |
| 5807 { | |
| 5808 upper = TRUE; | |
| 5809 aff_entry->ae_chop = NULL; | |
| 5810 *p = NUL; | |
| 5811 | |
| 5812 /* The condition is matched with the | |
| 5813 * actual word, thus must check for the | |
| 5814 * upper-case letter. */ | |
| 5815 if (aff_entry->ae_cond != NULL) | |
| 5816 { | |
| 5817 char_u buf[MAXLINELEN]; | |
| 5818 #ifdef FEAT_MBYTE | |
| 5819 if (has_mbyte) | |
| 5820 { | |
| 5821 onecap_copy(items[4], buf, TRUE); | |
| 5822 aff_entry->ae_cond = getroom_save( | |
| 474 | 5823 spin, buf); |
| 455 | 5824 } |
| 5825 else | |
| 5826 #endif | |
| 5827 *aff_entry->ae_cond = c_up; | |
| 5828 if (aff_entry->ae_cond != NULL) | |
| 5829 { | |
| 5830 sprintf((char *)buf, "^%s", | |
| 339 | 5831 aff_entry->ae_cond); |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4352
diff
changeset
|
5832 vim_regfree(aff_entry->ae_prog); |
| 455 | 5833 aff_entry->ae_prog = vim_regcomp( |
| 5834 buf, RE_MAGIC + RE_STRING); | |
| 5835 } | |
| 5836 } | |
| 5837 } | |
| 5838 } | |
| 339 | 5839 } |
| 5840 | |
| 809 | 5841 if (aff_entry->ae_chop == NULL |
| 5842 && aff_entry->ae_flags == NULL) | |
| 390 | 5843 { |
| 455 | 5844 int idx; |
| 5845 char_u **pp; | |
| 5846 int n; | |
| 5847 | |
| 495 | 5848 /* Find a previously used condition. */ |
| 455 | 5849 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0; |
| 5850 --idx) | |
| 5851 { | |
| 5852 p = ((char_u **)spin->si_prefcond.ga_data)[idx]; | |
| 5853 if (str_equal(p, aff_entry->ae_cond)) | |
| 5854 break; | |
| 5855 } | |
| 5856 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK) | |
| 5857 { | |
| 5858 /* Not found, add a new condition. */ | |
| 5859 idx = spin->si_prefcond.ga_len++; | |
| 5860 pp = ((char_u **)spin->si_prefcond.ga_data) | |
| 5861 + idx; | |
| 5862 if (aff_entry->ae_cond == NULL) | |
| 5863 *pp = NULL; | |
| 5864 else | |
| 474 | 5865 *pp = getroom_save(spin, |
| 455 | 5866 aff_entry->ae_cond); |
| 5867 } | |
| 5868 | |
| 5869 /* Add the prefix to the prefix tree. */ | |
| 5870 if (aff_entry->ae_add == NULL) | |
| 5871 p = (char_u *)""; | |
| 390 | 5872 else |
| 455 | 5873 p = aff_entry->ae_add; |
| 809 | 5874 |
| 455 | 5875 /* PFX_FLAGS is a negative number, so that |
| 5876 * tree_add_word() knows this is the prefix tree. */ | |
| 5877 n = PFX_FLAGS; | |
| 5878 if (!cur_aff->ah_combine) | |
| 5879 n |= WFP_NC; | |
| 5880 if (upper) | |
| 5881 n |= WFP_UP; | |
| 822 | 5882 if (aff_entry->ae_comppermit) |
| 5883 n |= WFP_COMPPERMIT; | |
| 5884 if (aff_entry->ae_compforbid) | |
| 5885 n |= WFP_COMPFORBID; | |
| 474 | 5886 tree_add_word(spin, p, spin->si_prefroot, n, |
| 5887 idx, cur_aff->ah_newID); | |
| 495 | 5888 did_postpone_prefix = TRUE; |
| 5889 } | |
| 5890 | |
| 5891 /* Didn't actually use ah_newID, backup si_newprefID. */ | |
| 5892 if (aff_todo == 0 && !did_postpone_prefix) | |
| 5893 { | |
| 5894 --spin->si_newprefID; | |
| 5895 cur_aff->ah_newID = 0; | |
| 390 | 5896 } |
| 339 | 5897 } |
| 240 | 5898 } |
| 236 | 5899 } |
| 1762 | 5900 else if (is_aff_rule(items, itemcnt, "FOL", 2) && fol == NULL) |
| 495 | 5901 { |
| 5902 fol = vim_strsave(items[1]); | |
| 5903 } | |
| 1762 | 5904 else if (is_aff_rule(items, itemcnt, "LOW", 2) && low == NULL) |
| 495 | 5905 { |
| 5906 low = vim_strsave(items[1]); | |
| 5907 } | |
| 1762 | 5908 else if (is_aff_rule(items, itemcnt, "UPP", 2) && upp == NULL) |
| 495 | 5909 { |
| 5910 upp = vim_strsave(items[1]); | |
| 255 | 5911 } |
| 1762 | 5912 else if (is_aff_rule(items, itemcnt, "REP", 2) |
| 5913 || is_aff_rule(items, itemcnt, "REPSAL", 2)) | |
| 625 | 5914 { |
| 5915 /* Ignore REP/REPSAL count */; | |
| 323 | 5916 if (!isdigit(*items[1])) |
| 625 | 5917 smsg((char_u *)_("Expected REP(SAL) count in %s line %d"), |
| 323 | 5918 fname, lnum); |
| 5919 } | |
| 625 | 5920 else if ((STRCMP(items[0], "REP") == 0 |
| 5921 || STRCMP(items[0], "REPSAL") == 0) | |
| 5922 && itemcnt >= 3) | |
| 5923 { | |
| 5924 /* REP/REPSAL item */ | |
| 482 | 5925 /* Myspell ignores extra arguments, we require it starts with |
| 5926 * # to detect mistakes. */ | |
| 5927 if (itemcnt > 3 && items[3][0] != '#') | |
| 5928 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]); | |
| 625 | 5929 if (items[0][3] == 'S' ? do_repsal : do_rep) |
| 532 | 5930 { |
| 5931 /* Replace underscore with space (can't include a space | |
| 5932 * directly). */ | |
| 5933 for (p = items[1]; *p != NUL; mb_ptr_adv(p)) | |
| 5934 if (*p == '_') | |
| 5935 *p = ' '; | |
| 5936 for (p = items[2]; *p != NUL; mb_ptr_adv(p)) | |
| 5937 if (*p == '_') | |
| 5938 *p = ' '; | |
| 625 | 5939 add_fromto(spin, items[0][3] == 'S' |
| 5940 ? &spin->si_repsal | |
| 5941 : &spin->si_rep, items[1], items[2]); | |
| 532 | 5942 } |
| 323 | 5943 } |
| 1762 | 5944 else if (is_aff_rule(items, itemcnt, "MAP", 2)) |
| 323 | 5945 { |
| 5946 /* MAP item or count */ | |
| 5947 if (!found_map) | |
| 5948 { | |
| 5949 /* First line contains the count. */ | |
| 5950 found_map = TRUE; | |
| 5951 if (!isdigit(*items[1])) | |
| 5952 smsg((char_u *)_("Expected MAP count in %s line %d"), | |
| 5953 fname, lnum); | |
| 5954 } | |
| 944 | 5955 else if (do_mapline) |
| 323 | 5956 { |
| 346 | 5957 int c; |
| 5958 | |
| 5959 /* Check that every character appears only once. */ | |
| 5960 for (p = items[1]; *p != NUL; ) | |
| 5961 { | |
| 5962 #ifdef FEAT_MBYTE | |
| 5963 c = mb_ptr2char_adv(&p); | |
| 5964 #else | |
| 5965 c = *p++; | |
| 5966 #endif | |
| 5967 if ((spin->si_map.ga_len > 0 | |
| 5968 && vim_strchr(spin->si_map.ga_data, c) | |
| 5969 != NULL) | |
| 5970 || vim_strchr(p, c) != NULL) | |
| 5971 smsg((char_u *)_("Duplicate character in MAP in %s line %d"), | |
| 5972 fname, lnum); | |
| 5973 } | |
| 5974 | |
| 323 | 5975 /* We simply concatenate all the MAP strings, separated by |
| 5976 * slashes. */ | |
| 5977 ga_concat(&spin->si_map, items[1]); | |
| 5978 ga_append(&spin->si_map, '/'); | |
| 5979 } | |
| 5980 } | |
| 1762 | 5981 /* Accept "SAL from to" and "SAL from to #comment". */ |
| 5982 else if (is_aff_rule(items, itemcnt, "SAL", 3)) | |
| 323 | 5983 { |
| 5984 if (do_sal) | |
| 5985 { | |
| 5986 /* SAL item (sounds-a-like) | |
| 5987 * Either one of the known keys or a from-to pair. */ | |
| 5988 if (STRCMP(items[1], "followup") == 0) | |
| 5989 spin->si_followup = sal_to_bool(items[2]); | |
| 5990 else if (STRCMP(items[1], "collapse_result") == 0) | |
| 5991 spin->si_collapse = sal_to_bool(items[2]); | |
| 5992 else if (STRCMP(items[1], "remove_accents") == 0) | |
| 5993 spin->si_rem_accents = sal_to_bool(items[2]); | |
| 5994 else | |
| 5995 /* when "to" is "_" it means empty */ | |
| 5996 add_fromto(spin, &spin->si_sal, items[1], | |
| 5997 STRCMP(items[2], "_") == 0 ? (char_u *)"" | |
| 5998 : items[2]); | |
| 5999 } | |
| 236 | 6000 } |
| 1762 | 6001 else if (is_aff_rule(items, itemcnt, "SOFOFROM", 2) |
| 495 | 6002 && sofofrom == NULL) |
| 6003 { | |
| 6004 sofofrom = getroom_save(spin, items[1]); | |
| 375 | 6005 } |
| 1762 | 6006 else if (is_aff_rule(items, itemcnt, "SOFOTO", 2) |
| 495 | 6007 && sofoto == NULL) |
| 6008 { | |
| 6009 sofoto = getroom_save(spin, items[1]); | |
| 375 | 6010 } |
| 625 | 6011 else if (STRCMP(items[0], "COMMON") == 0) |
| 6012 { | |
| 6013 int i; | |
| 6014 | |
| 6015 for (i = 1; i < itemcnt; ++i) | |
| 6016 { | |
| 6017 if (HASHITEM_EMPTY(hash_find(&spin->si_commonwords, | |
| 6018 items[i]))) | |
| 6019 { | |
| 6020 p = vim_strsave(items[i]); | |
| 6021 if (p == NULL) | |
| 6022 break; | |
| 6023 hash_add(&spin->si_commonwords, p); | |
| 6024 } | |
| 6025 } | |
| 6026 } | |
| 300 | 6027 else |
| 481 | 6028 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"), |
| 236 | 6029 fname, lnum, items[0]); |
| 6030 } | |
| 6031 } | |
| 6032 | |
| 255 | 6033 if (fol != NULL || low != NULL || upp != NULL) |
| 6034 { | |
| 351 | 6035 if (spin->si_clear_chartab) |
| 6036 { | |
| 6037 /* Clear the char type tables, don't want to use any of the | |
| 6038 * currently used spell properties. */ | |
| 6039 init_spell_chartab(); | |
| 6040 spin->si_clear_chartab = FALSE; | |
| 6041 } | |
| 6042 | |
| 316 | 6043 /* |
| 6044 * Don't write a word table for an ASCII file, so that we don't check | |
| 6045 * for conflicts with a word table that matches 'encoding'. | |
| 324 | 6046 * Don't write one for utf-8 either, we use utf_*() and |
| 316 | 6047 * mb_get_class(), the list of chars in the file will be incomplete. |
| 6048 */ | |
| 6049 if (!spin->si_ascii | |
| 6050 #ifdef FEAT_MBYTE | |
| 6051 && !enc_utf8 | |
| 6052 #endif | |
| 6053 ) | |
| 260 | 6054 { |
| 6055 if (fol == NULL || low == NULL || upp == NULL) | |
| 6056 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname); | |
| 6057 else | |
| 316 | 6058 (void)set_spell_chartab(fol, low, upp); |
| 260 | 6059 } |
| 255 | 6060 |
| 6061 vim_free(fol); | |
| 6062 vim_free(low); | |
| 6063 vim_free(upp); | |
| 6064 } | |
| 6065 | |
| 481 | 6066 /* Use compound specifications of the .aff file for the spell info. */ |
| 495 | 6067 if (compmax != 0) |
| 6068 { | |
| 809 | 6069 aff_check_number(spin->si_compmax, compmax, "COMPOUNDWORDMAX"); |
| 495 | 6070 spin->si_compmax = compmax; |
| 6071 } | |
| 6072 | |
| 6073 if (compminlen != 0) | |
| 6074 { | |
| 6075 aff_check_number(spin->si_compminlen, compminlen, "COMPOUNDMIN"); | |
| 6076 spin->si_compminlen = compminlen; | |
| 6077 } | |
| 6078 | |
| 6079 if (compsylmax != 0) | |
| 6080 { | |
| 6081 if (syllable == NULL) | |
| 6082 smsg((char_u *)_("COMPOUNDSYLMAX used without SYLLABLE")); | |
| 6083 aff_check_number(spin->si_compsylmax, compsylmax, "COMPOUNDSYLMAX"); | |
| 6084 spin->si_compsylmax = compsylmax; | |
| 6085 } | |
| 6086 | |
| 809 | 6087 if (compoptions != 0) |
| 6088 { | |
| 6089 aff_check_number(spin->si_compoptions, compoptions, "COMPOUND options"); | |
| 6090 spin->si_compoptions |= compoptions; | |
| 6091 } | |
| 6092 | |
| 495 | 6093 if (compflags != NULL) |
| 6094 process_compflags(spin, aff, compflags); | |
| 6095 | |
| 6096 /* Check that we didn't use too many renumbered flags. */ | |
| 500 | 6097 if (spin->si_newcompID < spin->si_newprefID) |
| 6098 { | |
| 6099 if (spin->si_newcompID == 127 || spin->si_newcompID == 255) | |
| 495 | 6100 MSG(_("Too many postponed prefixes")); |
| 500 | 6101 else if (spin->si_newprefID == 0 || spin->si_newprefID == 127) |
| 495 | 6102 MSG(_("Too many compound flags")); |
| 481 | 6103 else |
| 1705 | 6104 MSG(_("Too many postponed prefixes and/or compound flags")); |
| 495 | 6105 } |
| 6106 | |
| 6107 if (syllable != NULL) | |
| 6108 { | |
| 6109 aff_check_string(spin->si_syllable, syllable, "SYLLABLE"); | |
| 6110 spin->si_syllable = syllable; | |
| 6111 } | |
| 6112 | |
| 6113 if (sofofrom != NULL || sofoto != NULL) | |
| 6114 { | |
| 6115 if (sofofrom == NULL || sofoto == NULL) | |
| 6116 smsg((char_u *)_("Missing SOFO%s line in %s"), | |
| 6117 sofofrom == NULL ? "FROM" : "TO", fname); | |
| 6118 else if (spin->si_sal.ga_len > 0) | |
| 6119 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname); | |
| 483 | 6120 else |
| 495 | 6121 { |
| 6122 aff_check_string(spin->si_sofofr, sofofrom, "SOFOFROM"); | |
| 6123 aff_check_string(spin->si_sofoto, sofoto, "SOFOTO"); | |
| 6124 spin->si_sofofr = sofofrom; | |
| 6125 spin->si_sofoto = sofoto; | |
| 6126 } | |
| 6127 } | |
| 6128 | |
| 6129 if (midword != NULL) | |
| 6130 { | |
| 6131 aff_check_string(spin->si_midword, midword, "MIDWORD"); | |
| 6132 spin->si_midword = midword; | |
| 483 | 6133 } |
| 6134 | |
| 236 | 6135 vim_free(pc); |
| 6136 fclose(fd); | |
| 6137 return aff; | |
| 6138 } | |
| 6139 | |
| 6140 /* | |
| 1762 | 6141 * Return TRUE when items[0] equals "rulename", there are "mincount" items or |
| 6142 * a comment is following after item "mincount". | |
| 6143 */ | |
| 6144 static int | |
| 6145 is_aff_rule(items, itemcnt, rulename, mincount) | |
| 6146 char_u **items; | |
| 6147 int itemcnt; | |
| 6148 char *rulename; | |
| 6149 int mincount; | |
| 6150 { | |
| 6151 return (STRCMP(items[0], rulename) == 0 | |
| 6152 && (itemcnt == mincount | |
| 6153 || (itemcnt > mincount && items[mincount][0] == '#'))); | |
| 6154 } | |
| 6155 | |
| 6156 /* | |
| 822 | 6157 * For affix "entry" move COMPOUNDFORBIDFLAG and COMPOUNDPERMITFLAG from |
| 6158 * ae_flags to ae_comppermit and ae_compforbid. | |
| 6159 */ | |
| 6160 static void | |
| 6161 aff_process_flags(affile, entry) | |
| 6162 afffile_T *affile; | |
| 6163 affentry_T *entry; | |
| 6164 { | |
| 6165 char_u *p; | |
| 6166 char_u *prevp; | |
| 825 | 6167 unsigned flag; |
| 822 | 6168 |
| 6169 if (entry->ae_flags != NULL | |
| 6170 && (affile->af_compforbid != 0 || affile->af_comppermit != 0)) | |
| 6171 { | |
| 6172 for (p = entry->ae_flags; *p != NUL; ) | |
| 6173 { | |
| 6174 prevp = p; | |
| 6175 flag = get_affitem(affile->af_flagtype, &p); | |
| 6176 if (flag == affile->af_comppermit || flag == affile->af_compforbid) | |
| 6177 { | |
| 1619 | 6178 STRMOVE(prevp, p); |
| 822 | 6179 p = prevp; |
| 6180 if (flag == affile->af_comppermit) | |
| 6181 entry->ae_comppermit = TRUE; | |
| 6182 else | |
| 6183 entry->ae_compforbid = TRUE; | |
| 6184 } | |
| 6185 if (affile->af_flagtype == AFT_NUM && *p == ',') | |
| 6186 ++p; | |
| 6187 } | |
| 6188 if (*entry->ae_flags == NUL) | |
| 6189 entry->ae_flags = NULL; /* nothing left */ | |
| 6190 } | |
| 6191 } | |
| 6192 | |
| 6193 /* | |
| 714 | 6194 * Return TRUE if "s" is the name of an info item in the affix file. |
| 6195 */ | |
| 6196 static int | |
| 6197 spell_info_item(s) | |
| 6198 char_u *s; | |
| 6199 { | |
| 6200 return STRCMP(s, "NAME") == 0 | |
| 6201 || STRCMP(s, "HOME") == 0 | |
| 6202 || STRCMP(s, "VERSION") == 0 | |
| 6203 || STRCMP(s, "AUTHOR") == 0 | |
| 6204 || STRCMP(s, "EMAIL") == 0 | |
| 6205 || STRCMP(s, "COPYRIGHT") == 0; | |
| 6206 } | |
| 6207 | |
| 6208 /* | |
| 495 | 6209 * Turn an affix flag name into a number, according to the FLAG type. |
| 6210 * returns zero for failure. | |
| 6211 */ | |
| 6212 static unsigned | |
| 6213 affitem2flag(flagtype, item, fname, lnum) | |
| 6214 int flagtype; | |
| 6215 char_u *item; | |
| 6216 char_u *fname; | |
| 6217 int lnum; | |
| 6218 { | |
| 6219 unsigned res; | |
| 6220 char_u *p = item; | |
| 6221 | |
| 6222 res = get_affitem(flagtype, &p); | |
| 6223 if (res == 0) | |
| 6224 { | |
| 498 | 6225 if (flagtype == AFT_NUM) |
| 495 | 6226 smsg((char_u *)_("Flag is not a number in %s line %d: %s"), |
| 6227 fname, lnum, item); | |
| 6228 else | |
| 6229 smsg((char_u *)_("Illegal flag in %s line %d: %s"), | |
| 6230 fname, lnum, item); | |
| 6231 } | |
| 6232 if (*p != NUL) | |
| 6233 { | |
| 6234 smsg((char_u *)_(e_affname), fname, lnum, item); | |
| 6235 return 0; | |
| 6236 } | |
| 6237 | |
| 6238 return res; | |
| 6239 } | |
| 6240 | |
| 6241 /* | |
| 6242 * Get one affix name from "*pp" and advance the pointer. | |
| 6243 * Returns zero for an error, still advances the pointer then. | |
| 6244 */ | |
| 6245 static unsigned | |
| 6246 get_affitem(flagtype, pp) | |
| 6247 int flagtype; | |
| 6248 char_u **pp; | |
| 6249 { | |
| 6250 int res; | |
| 6251 | |
| 498 | 6252 if (flagtype == AFT_NUM) |
| 495 | 6253 { |
| 6254 if (!VIM_ISDIGIT(**pp)) | |
| 6255 { | |
| 498 | 6256 ++*pp; /* always advance, avoid getting stuck */ |
| 495 | 6257 return 0; |
| 6258 } | |
| 6259 res = getdigits(pp); | |
| 6260 } | |
| 6261 else | |
| 6262 { | |
| 6263 #ifdef FEAT_MBYTE | |
| 6264 res = mb_ptr2char_adv(pp); | |
| 6265 #else | |
| 6266 res = *(*pp)++; | |
| 6267 #endif | |
| 498 | 6268 if (flagtype == AFT_LONG || (flagtype == AFT_CAPLONG |
| 495 | 6269 && res >= 'A' && res <= 'Z')) |
| 6270 { | |
| 6271 if (**pp == NUL) | |
| 6272 return 0; | |
| 6273 #ifdef FEAT_MBYTE | |
| 6274 res = mb_ptr2char_adv(pp) + (res << 16); | |
| 6275 #else | |
| 6276 res = *(*pp)++ + (res << 16); | |
| 6277 #endif | |
| 6278 } | |
| 6279 } | |
| 6280 return res; | |
| 6281 } | |
| 6282 | |
| 6283 /* | |
| 6284 * Process the "compflags" string used in an affix file and append it to | |
| 6285 * spin->si_compflags. | |
| 6286 * The processing involves changing the affix names to ID numbers, so that | |
| 6287 * they fit in one byte. | |
| 6288 */ | |
| 6289 static void | |
| 6290 process_compflags(spin, aff, compflags) | |
| 6291 spellinfo_T *spin; | |
| 6292 afffile_T *aff; | |
| 6293 char_u *compflags; | |
| 6294 { | |
| 6295 char_u *p; | |
| 6296 char_u *prevp; | |
| 6297 unsigned flag; | |
| 6298 compitem_T *ci; | |
| 6299 int id; | |
| 6300 int len; | |
| 6301 char_u *tp; | |
| 6302 char_u key[AH_KEY_LEN]; | |
| 6303 hashitem_T *hi; | |
| 6304 | |
| 6305 /* Make room for the old and the new compflags, concatenated with a / in | |
| 6306 * between. Processing it makes it shorter, but we don't know by how | |
| 6307 * much, thus allocate the maximum. */ | |
| 835 | 6308 len = (int)STRLEN(compflags) + 1; |
| 495 | 6309 if (spin->si_compflags != NULL) |
| 835 | 6310 len += (int)STRLEN(spin->si_compflags) + 1; |
| 495 | 6311 p = getroom(spin, len, FALSE); |
| 6312 if (p == NULL) | |
| 6313 return; | |
| 6314 if (spin->si_compflags != NULL) | |
| 6315 { | |
| 6316 STRCPY(p, spin->si_compflags); | |
| 6317 STRCAT(p, "/"); | |
| 6318 } | |
| 6319 spin->si_compflags = p; | |
| 6320 tp = p + STRLEN(p); | |
| 6321 | |
| 6322 for (p = compflags; *p != NUL; ) | |
| 6323 { | |
| 2691 | 6324 if (vim_strchr((char_u *)"/?*+[]", *p) != NULL) |
| 495 | 6325 /* Copy non-flag characters directly. */ |
| 6326 *tp++ = *p++; | |
| 6327 else | |
| 6328 { | |
| 6329 /* First get the flag number, also checks validity. */ | |
| 6330 prevp = p; | |
| 6331 flag = get_affitem(aff->af_flagtype, &p); | |
| 6332 if (flag != 0) | |
| 6333 { | |
| 6334 /* Find the flag in the hashtable. If it was used before, use | |
| 6335 * the existing ID. Otherwise add a new entry. */ | |
| 6336 vim_strncpy(key, prevp, p - prevp); | |
| 6337 hi = hash_find(&aff->af_comp, key); | |
| 6338 if (!HASHITEM_EMPTY(hi)) | |
| 6339 id = HI2CI(hi)->ci_newID; | |
| 6340 else | |
| 6341 { | |
| 6342 ci = (compitem_T *)getroom(spin, sizeof(compitem_T), TRUE); | |
| 6343 if (ci == NULL) | |
| 6344 break; | |
| 6345 STRCPY(ci->ci_key, key); | |
| 6346 ci->ci_flag = flag; | |
| 6347 /* Avoid using a flag ID that has a special meaning in a | |
| 6348 * regexp (also inside []). */ | |
| 6349 do | |
| 6350 { | |
| 500 | 6351 check_renumber(spin); |
| 6352 id = spin->si_newcompID--; | |
| 2691 | 6353 } while (vim_strchr((char_u *)"/?*+[]\\-^", id) != NULL); |
| 495 | 6354 ci->ci_newID = id; |
| 6355 hash_add(&aff->af_comp, ci->ci_key); | |
| 6356 } | |
| 6357 *tp++ = id; | |
| 6358 } | |
| 498 | 6359 if (aff->af_flagtype == AFT_NUM && *p == ',') |
| 495 | 6360 ++p; |
| 6361 } | |
| 6362 } | |
| 6363 | |
| 6364 *tp = NUL; | |
| 6365 } | |
| 6366 | |
| 6367 /* | |
| 500 | 6368 * Check that the new IDs for postponed affixes and compounding don't overrun |
| 6369 * each other. We have almost 255 available, but start at 0-127 to avoid | |
| 6370 * using two bytes for utf-8. When the 0-127 range is used up go to 128-255. | |
| 6371 * When that is used up an error message is given. | |
| 6372 */ | |
| 6373 static void | |
| 6374 check_renumber(spin) | |
| 6375 spellinfo_T *spin; | |
| 6376 { | |
| 6377 if (spin->si_newprefID == spin->si_newcompID && spin->si_newcompID < 128) | |
| 6378 { | |
| 6379 spin->si_newprefID = 127; | |
| 6380 spin->si_newcompID = 255; | |
| 6381 } | |
| 6382 } | |
| 6383 | |
| 6384 /* | |
| 495 | 6385 * Return TRUE if flag "flag" appears in affix list "afflist". |
| 6386 */ | |
| 6387 static int | |
| 6388 flag_in_afflist(flagtype, afflist, flag) | |
| 6389 int flagtype; | |
| 6390 char_u *afflist; | |
| 6391 unsigned flag; | |
| 6392 { | |
| 6393 char_u *p; | |
| 6394 unsigned n; | |
| 6395 | |
| 6396 switch (flagtype) | |
| 6397 { | |
| 6398 case AFT_CHAR: | |
| 6399 return vim_strchr(afflist, flag) != NULL; | |
| 6400 | |
| 498 | 6401 case AFT_CAPLONG: |
| 6402 case AFT_LONG: | |
| 495 | 6403 for (p = afflist; *p != NUL; ) |
| 6404 { | |
| 6405 #ifdef FEAT_MBYTE | |
| 6406 n = mb_ptr2char_adv(&p); | |
| 6407 #else | |
| 6408 n = *p++; | |
| 6409 #endif | |
| 498 | 6410 if ((flagtype == AFT_LONG || (n >= 'A' && n <= 'Z')) |
| 495 | 6411 && *p != NUL) |
| 6412 #ifdef FEAT_MBYTE | |
| 6413 n = mb_ptr2char_adv(&p) + (n << 16); | |
| 6414 #else | |
| 6415 n = *p++ + (n << 16); | |
| 6416 #endif | |
| 6417 if (n == flag) | |
| 6418 return TRUE; | |
| 6419 } | |
| 6420 break; | |
| 6421 | |
| 498 | 6422 case AFT_NUM: |
| 495 | 6423 for (p = afflist; *p != NUL; ) |
| 6424 { | |
| 6425 n = getdigits(&p); | |
| 6426 if (n == flag) | |
| 6427 return TRUE; | |
| 6428 if (*p != NUL) /* skip over comma */ | |
| 6429 ++p; | |
| 6430 } | |
| 6431 break; | |
| 6432 } | |
| 6433 return FALSE; | |
| 6434 } | |
| 6435 | |
| 6436 /* | |
| 6437 * Give a warning when "spinval" and "affval" numbers are set and not the same. | |
| 6438 */ | |
| 6439 static void | |
| 6440 aff_check_number(spinval, affval, name) | |
| 6441 int spinval; | |
| 6442 int affval; | |
| 6443 char *name; | |
| 6444 { | |
| 6445 if (spinval != 0 && spinval != affval) | |
| 6446 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name); | |
| 6447 } | |
| 6448 | |
| 6449 /* | |
| 6450 * Give a warning when "spinval" and "affval" strings are set and not the same. | |
| 6451 */ | |
| 6452 static void | |
| 6453 aff_check_string(spinval, affval, name) | |
| 6454 char_u *spinval; | |
| 6455 char_u *affval; | |
| 6456 char *name; | |
| 6457 { | |
| 6458 if (spinval != NULL && STRCMP(spinval, affval) != 0) | |
| 6459 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name); | |
| 6460 } | |
| 6461 | |
| 6462 /* | |
| 339 | 6463 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being |
| 6464 * NULL as equal. | |
| 6465 */ | |
| 6466 static int | |
| 6467 str_equal(s1, s2) | |
| 6468 char_u *s1; | |
| 6469 char_u *s2; | |
| 6470 { | |
| 6471 if (s1 == NULL || s2 == NULL) | |
| 6472 return s1 == s2; | |
| 6473 return STRCMP(s1, s2) == 0; | |
| 6474 } | |
| 6475 | |
| 6476 /* | |
| 323 | 6477 * Add a from-to item to "gap". Used for REP and SAL items. |
| 6478 * They are stored case-folded. | |
| 6479 */ | |
| 6480 static void | |
| 6481 add_fromto(spin, gap, from, to) | |
| 6482 spellinfo_T *spin; | |
| 6483 garray_T *gap; | |
| 6484 char_u *from; | |
| 6485 char_u *to; | |
| 6486 { | |
| 6487 fromto_T *ftp; | |
| 6488 char_u word[MAXWLEN]; | |
| 6489 | |
| 6490 if (ga_grow(gap, 1) == OK) | |
| 6491 { | |
| 6492 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len; | |
| 835 | 6493 (void)spell_casefold(from, (int)STRLEN(from), word, MAXWLEN); |
| 474 | 6494 ftp->ft_from = getroom_save(spin, word); |
| 835 | 6495 (void)spell_casefold(to, (int)STRLEN(to), word, MAXWLEN); |
| 474 | 6496 ftp->ft_to = getroom_save(spin, word); |
| 323 | 6497 ++gap->ga_len; |
| 6498 } | |
| 6499 } | |
| 6500 | |
| 6501 /* | |
| 6502 * Convert a boolean argument in a SAL line to TRUE or FALSE; | |
| 6503 */ | |
| 6504 static int | |
| 6505 sal_to_bool(s) | |
| 6506 char_u *s; | |
| 6507 { | |
| 6508 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0; | |
| 6509 } | |
| 6510 | |
| 6511 /* | |
| 236 | 6512 * Free the structure filled by spell_read_aff(). |
| 6513 */ | |
| 6514 static void | |
| 6515 spell_free_aff(aff) | |
| 6516 afffile_T *aff; | |
| 6517 { | |
| 6518 hashtab_T *ht; | |
| 6519 hashitem_T *hi; | |
| 6520 int todo; | |
| 6521 affheader_T *ah; | |
| 300 | 6522 affentry_T *ae; |
| 236 | 6523 |
| 6524 vim_free(aff->af_enc); | |
| 6525 | |
| 339 | 6526 /* All this trouble to free the "ae_prog" items... */ |
| 236 | 6527 for (ht = &aff->af_pref; ; ht = &aff->af_suff) |
| 6528 { | |
| 835 | 6529 todo = (int)ht->ht_used; |
| 236 | 6530 for (hi = ht->ht_array; todo > 0; ++hi) |
| 6531 { | |
| 6532 if (!HASHITEM_EMPTY(hi)) | |
| 6533 { | |
| 6534 --todo; | |
| 6535 ah = HI2AH(hi); | |
| 300 | 6536 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next) |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4352
diff
changeset
|
6537 vim_regfree(ae->ae_prog); |
| 236 | 6538 } |
| 6539 } | |
| 6540 if (ht == &aff->af_suff) | |
| 6541 break; | |
| 6542 } | |
| 300 | 6543 |
| 236 | 6544 hash_clear(&aff->af_pref); |
| 6545 hash_clear(&aff->af_suff); | |
| 495 | 6546 hash_clear(&aff->af_comp); |
| 236 | 6547 } |
| 6548 | |
| 6549 /* | |
| 300 | 6550 * Read dictionary file "fname". |
| 236 | 6551 * Returns OK or FAIL; |
| 6552 */ | |
| 6553 static int | |
| 474 | 6554 spell_read_dic(spin, fname, affile) |
| 6555 spellinfo_T *spin; | |
| 236 | 6556 char_u *fname; |
| 300 | 6557 afffile_T *affile; |
| 236 | 6558 { |
| 300 | 6559 hashtab_T ht; |
| 236 | 6560 char_u line[MAXLINELEN]; |
| 481 | 6561 char_u *p; |
| 300 | 6562 char_u *afflist; |
| 483 | 6563 char_u store_afflist[MAXWLEN]; |
| 6564 int pfxlen; | |
| 6565 int need_affix; | |
| 300 | 6566 char_u *dw; |
| 236 | 6567 char_u *pc; |
| 6568 char_u *w; | |
| 6569 int l; | |
| 6570 hash_T hash; | |
| 6571 hashitem_T *hi; | |
| 6572 FILE *fd; | |
| 6573 int lnum = 1; | |
| 300 | 6574 int non_ascii = 0; |
| 6575 int retval = OK; | |
| 6576 char_u message[MAXLINELEN + MAXWLEN]; | |
| 307 | 6577 int flags; |
| 481 | 6578 int duplicate = 0; |
| 236 | 6579 |
| 300 | 6580 /* |
| 6581 * Open the file. | |
| 6582 */ | |
| 310 | 6583 fd = mch_fopen((char *)fname, "r"); |
| 236 | 6584 if (fd == NULL) |
| 6585 { | |
| 6586 EMSG2(_(e_notopen), fname); | |
| 6587 return FAIL; | |
| 6588 } | |
| 6589 | |
| 300 | 6590 /* The hashtable is only used to detect duplicated words. */ |
| 6591 hash_init(&ht); | |
| 6592 | |
| 625 | 6593 vim_snprintf((char *)IObuff, IOSIZE, |
| 6594 _("Reading dictionary file %s ..."), fname); | |
| 6595 spell_message(spin, IObuff); | |
| 236 | 6596 |
| 474 | 6597 /* start with a message for the first line */ |
| 6598 spin->si_msg_count = 999999; | |
| 6599 | |
| 236 | 6600 /* Read and ignore the first line: word count. */ |
| 6601 (void)vim_fgets(line, MAXLINELEN, fd); | |
| 324 | 6602 if (!vim_isdigit(*skipwhite(line))) |
| 236 | 6603 EMSG2(_("E760: No word count in %s"), fname); |
| 6604 | |
| 6605 /* | |
| 6606 * Read all the lines in the file one by one. | |
| 6607 * The words are converted to 'encoding' here, before being added to | |
| 6608 * the hashtable. | |
| 6609 */ | |
| 255 | 6610 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int) |
| 236 | 6611 { |
| 255 | 6612 line_breakcheck(); |
| 236 | 6613 ++lnum; |
| 455 | 6614 if (line[0] == '#' || line[0] == '/') |
| 351 | 6615 continue; /* comment line */ |
| 236 | 6616 |
| 300 | 6617 /* Remove CR, LF and white space from the end. White space halfway |
| 6618 * the word is kept to allow e.g., "et al.". */ | |
| 835 | 6619 l = (int)STRLEN(line); |
| 236 | 6620 while (l > 0 && line[l - 1] <= ' ') |
| 6621 --l; | |
| 6622 if (l == 0) | |
| 6623 continue; /* empty line */ | |
| 6624 line[l] = NUL; | |
| 6625 | |
| 310 | 6626 #ifdef FEAT_MBYTE |
| 236 | 6627 /* Convert from "SET" to 'encoding' when needed. */ |
| 300 | 6628 if (spin->si_conv.vc_type != CONV_NONE) |
| 236 | 6629 { |
| 300 | 6630 pc = string_convert(&spin->si_conv, line, NULL); |
| 255 | 6631 if (pc == NULL) |
| 6632 { | |
| 6633 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"), | |
| 6634 fname, lnum, line); | |
| 6635 continue; | |
| 6636 } | |
| 236 | 6637 w = pc; |
| 6638 } | |
| 6639 else | |
| 310 | 6640 #endif |
| 236 | 6641 { |
| 6642 pc = NULL; | |
| 6643 w = line; | |
| 6644 } | |
| 6645 | |
| 822 | 6646 /* Truncate the word at the "/", set "afflist" to what follows. |
| 6647 * Replace "\/" by "/" and "\\" by "\". */ | |
| 6648 afflist = NULL; | |
| 6649 for (p = w; *p != NUL; mb_ptr_adv(p)) | |
| 6650 { | |
| 6651 if (*p == '\\' && (p[1] == '\\' || p[1] == '/')) | |
| 1619 | 6652 STRMOVE(p, p + 1); |
| 822 | 6653 else if (*p == '/') |
| 6654 { | |
| 6655 *p = NUL; | |
| 6656 afflist = p + 1; | |
| 6657 break; | |
| 6658 } | |
| 6659 } | |
| 6660 | |
| 6661 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */ | |
| 6662 if (spin->si_ascii && has_non_ascii(w)) | |
| 6663 { | |
| 6664 ++non_ascii; | |
| 825 | 6665 vim_free(pc); |
| 822 | 6666 continue; |
| 6667 } | |
| 6668 | |
| 474 | 6669 /* This takes time, print a message every 10000 words. */ |
| 6670 if (spin->si_verbose && spin->si_msg_count > 10000) | |
| 6671 { | |
| 6672 spin->si_msg_count = 0; | |
| 339 | 6673 vim_snprintf((char *)message, sizeof(message), |
| 6674 _("line %6d, word %6d - %s"), | |
| 6675 lnum, spin->si_foldwcount + spin->si_keepwcount, w); | |
| 6676 msg_start(); | |
| 6677 msg_puts_long_attr(message, 0); | |
| 6678 msg_clr_eos(); | |
| 6679 msg_didout = FALSE; | |
| 6680 msg_col = 0; | |
| 6681 out_flush(); | |
| 6682 } | |
| 6683 | |
| 300 | 6684 /* Store the word in the hashtable to be able to find duplicates. */ |
| 474 | 6685 dw = (char_u *)getroom_save(spin, w); |
| 236 | 6686 if (dw == NULL) |
| 825 | 6687 { |
| 300 | 6688 retval = FAIL; |
| 825 | 6689 vim_free(pc); |
| 236 | 6690 break; |
| 825 | 6691 } |
| 236 | 6692 |
| 300 | 6693 hash = hash_hash(dw); |
| 6694 hi = hash_lookup(&ht, dw, hash); | |
| 236 | 6695 if (!HASHITEM_EMPTY(hi)) |
| 481 | 6696 { |
| 6697 if (p_verbose > 0) | |
| 6698 smsg((char_u *)_("Duplicate word in %s line %d: %s"), | |
| 375 | 6699 fname, lnum, dw); |
| 481 | 6700 else if (duplicate == 0) |
| 6701 smsg((char_u *)_("First duplicate word in %s line %d: %s"), | |
| 6702 fname, lnum, dw); | |
| 6703 ++duplicate; | |
| 6704 } | |
| 236 | 6705 else |
| 300 | 6706 hash_add_item(&ht, hi, dw, hash); |
| 6707 | |
| 307 | 6708 flags = 0; |
| 483 | 6709 store_afflist[0] = NUL; |
| 6710 pfxlen = 0; | |
| 6711 need_affix = FALSE; | |
| 307 | 6712 if (afflist != NULL) |
| 6713 { | |
| 820 | 6714 /* Extract flags from the affix list. */ |
| 6715 flags |= get_affix_flags(affile, afflist); | |
| 6716 | |
| 495 | 6717 if (affile->af_needaffix != 0 && flag_in_afflist( |
| 6718 affile->af_flagtype, afflist, affile->af_needaffix)) | |
| 483 | 6719 need_affix = TRUE; |
| 339 | 6720 |
| 6721 if (affile->af_pfxpostpone) | |
| 6722 /* Need to store the list of prefix IDs with the word. */ | |
| 483 | 6723 pfxlen = get_pfxlist(affile, afflist, store_afflist); |
| 6724 | |
| 6725 if (spin->si_compflags != NULL) | |
| 6726 /* Need to store the list of compound flags with the word. | |
| 6727 * Concatenate them to the list of prefix IDs. */ | |
| 495 | 6728 get_compflags(affile, afflist, store_afflist + pfxlen); |
| 307 | 6729 } |
| 6730 | |
| 300 | 6731 /* Add the word to the word tree(s). */ |
| 483 | 6732 if (store_word(spin, dw, flags, spin->si_region, |
| 6733 store_afflist, need_affix) == FAIL) | |
| 300 | 6734 retval = FAIL; |
| 236 | 6735 |
| 300 | 6736 if (afflist != NULL) |
| 6737 { | |
| 6738 /* Find all matching suffixes and add the resulting words. | |
| 6739 * Additionally do matching prefixes that combine. */ | |
| 474 | 6740 if (store_aff_word(spin, dw, afflist, affile, |
| 307 | 6741 &affile->af_suff, &affile->af_pref, |
| 820 | 6742 CONDIT_SUF, flags, store_afflist, pfxlen) == FAIL) |
| 300 | 6743 retval = FAIL; |
| 6744 | |
| 6745 /* Find all matching prefixes and add the resulting words. */ | |
| 474 | 6746 if (store_aff_word(spin, dw, afflist, affile, |
| 339 | 6747 &affile->af_pref, NULL, |
| 820 | 6748 CONDIT_SUF, flags, store_afflist, pfxlen) == FAIL) |
| 300 | 6749 retval = FAIL; |
| 6750 } | |
| 825 | 6751 |
| 6752 vim_free(pc); | |
| 236 | 6753 } |
| 6754 | |
| 481 | 6755 if (duplicate > 0) |
| 6756 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname); | |
| 300 | 6757 if (spin->si_ascii && non_ascii > 0) |
| 481 | 6758 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"), |
| 6759 non_ascii, fname); | |
| 300 | 6760 hash_clear(&ht); |
| 6761 | |
| 236 | 6762 fclose(fd); |
| 300 | 6763 return retval; |
| 236 | 6764 } |
| 6765 | |
| 6766 /* | |
| 820 | 6767 * Check for affix flags in "afflist" that are turned into word flags. |
| 6768 * Return WF_ flags. | |
| 6769 */ | |
| 6770 static int | |
| 6771 get_affix_flags(affile, afflist) | |
| 6772 afffile_T *affile; | |
| 6773 char_u *afflist; | |
| 6774 { | |
| 6775 int flags = 0; | |
| 6776 | |
| 6777 if (affile->af_keepcase != 0 && flag_in_afflist( | |
| 6778 affile->af_flagtype, afflist, affile->af_keepcase)) | |
| 6779 flags |= WF_KEEPCAP | WF_FIXCAP; | |
| 6780 if (affile->af_rare != 0 && flag_in_afflist( | |
| 6781 affile->af_flagtype, afflist, affile->af_rare)) | |
| 6782 flags |= WF_RARE; | |
| 6783 if (affile->af_bad != 0 && flag_in_afflist( | |
| 6784 affile->af_flagtype, afflist, affile->af_bad)) | |
| 6785 flags |= WF_BANNED; | |
| 6786 if (affile->af_needcomp != 0 && flag_in_afflist( | |
| 6787 affile->af_flagtype, afflist, affile->af_needcomp)) | |
| 6788 flags |= WF_NEEDCOMP; | |
| 6789 if (affile->af_comproot != 0 && flag_in_afflist( | |
| 6790 affile->af_flagtype, afflist, affile->af_comproot)) | |
| 6791 flags |= WF_COMPROOT; | |
| 6792 if (affile->af_nosuggest != 0 && flag_in_afflist( | |
| 6793 affile->af_flagtype, afflist, affile->af_nosuggest)) | |
| 6794 flags |= WF_NOSUGGEST; | |
| 6795 return flags; | |
| 6796 } | |
| 6797 | |
| 6798 /* | |
| 339 | 6799 * Get the list of prefix IDs from the affix list "afflist". |
| 6800 * Used for PFXPOSTPONE. | |
| 483 | 6801 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL |
| 6802 * and return the number of affixes. | |
| 6803 */ | |
| 6804 static int | |
| 6805 get_pfxlist(affile, afflist, store_afflist) | |
| 339 | 6806 afffile_T *affile; |
| 6807 char_u *afflist; | |
| 483 | 6808 char_u *store_afflist; |
| 339 | 6809 { |
| 6810 char_u *p; | |
| 495 | 6811 char_u *prevp; |
| 483 | 6812 int cnt = 0; |
| 495 | 6813 int id; |
| 6814 char_u key[AH_KEY_LEN]; | |
| 339 | 6815 hashitem_T *hi; |
| 6816 | |
| 495 | 6817 for (p = afflist; *p != NUL; ) |
| 6818 { | |
| 6819 prevp = p; | |
| 6820 if (get_affitem(affile->af_flagtype, &p) != 0) | |
| 6821 { | |
| 6822 /* A flag is a postponed prefix flag if it appears in "af_pref" | |
| 6823 * and it's ID is not zero. */ | |
| 6824 vim_strncpy(key, prevp, p - prevp); | |
| 6825 hi = hash_find(&affile->af_pref, key); | |
| 6826 if (!HASHITEM_EMPTY(hi)) | |
| 6827 { | |
| 6828 id = HI2AH(hi)->ah_newID; | |
| 6829 if (id != 0) | |
| 6830 store_afflist[cnt++] = id; | |
| 6831 } | |
| 6832 } | |
| 498 | 6833 if (affile->af_flagtype == AFT_NUM && *p == ',') |
| 495 | 6834 ++p; |
| 483 | 6835 } |
| 6836 | |
| 6837 store_afflist[cnt] = NUL; | |
| 6838 return cnt; | |
| 339 | 6839 } |
| 6840 | |
| 6841 /* | |
| 495 | 6842 * Get the list of compound IDs from the affix list "afflist" that are used |
| 6843 * for compound words. | |
| 483 | 6844 * Puts the flags in "store_afflist[]". |
| 6845 */ | |
| 6846 static void | |
| 495 | 6847 get_compflags(affile, afflist, store_afflist) |
| 6848 afffile_T *affile; | |
| 481 | 6849 char_u *afflist; |
| 483 | 6850 char_u *store_afflist; |
| 481 | 6851 { |
| 6852 char_u *p; | |
| 495 | 6853 char_u *prevp; |
| 483 | 6854 int cnt = 0; |
| 495 | 6855 char_u key[AH_KEY_LEN]; |
| 6856 hashitem_T *hi; | |
| 6857 | |
| 6858 for (p = afflist; *p != NUL; ) | |
| 6859 { | |
| 6860 prevp = p; | |
| 6861 if (get_affitem(affile->af_flagtype, &p) != 0) | |
| 6862 { | |
| 6863 /* A flag is a compound flag if it appears in "af_comp". */ | |
| 6864 vim_strncpy(key, prevp, p - prevp); | |
| 6865 hi = hash_find(&affile->af_comp, key); | |
| 6866 if (!HASHITEM_EMPTY(hi)) | |
| 6867 store_afflist[cnt++] = HI2CI(hi)->ci_newID; | |
| 6868 } | |
| 498 | 6869 if (affile->af_flagtype == AFT_NUM && *p == ',') |
| 495 | 6870 ++p; |
| 6871 } | |
| 483 | 6872 |
| 6873 store_afflist[cnt] = NUL; | |
| 481 | 6874 } |
| 6875 | |
| 6876 /* | |
| 300 | 6877 * Apply affixes to a word and store the resulting words. |
| 6878 * "ht" is the hashtable with affentry_T that need to be applied, either | |
| 6879 * prefixes or suffixes. | |
| 6880 * "xht", when not NULL, is the prefix hashtable, to be used additionally on | |
| 6881 * the resulting words for combining affixes. | |
| 6882 * | |
| 6883 * Returns FAIL when out of memory. | |
| 236 | 6884 */ |
| 300 | 6885 static int |
| 820 | 6886 store_aff_word(spin, word, afflist, affile, ht, xht, condit, flags, |
| 483 | 6887 pfxlist, pfxlen) |
| 474 | 6888 spellinfo_T *spin; /* spell info */ |
| 300 | 6889 char_u *word; /* basic word start */ |
| 6890 char_u *afflist; /* list of names of supported affixes */ | |
| 339 | 6891 afffile_T *affile; |
| 300 | 6892 hashtab_T *ht; |
| 6893 hashtab_T *xht; | |
| 820 | 6894 int condit; /* CONDIT_SUF et al. */ |
| 307 | 6895 int flags; /* flags for the word */ |
| 339 | 6896 char_u *pfxlist; /* list of prefix IDs */ |
| 483 | 6897 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest |
| 6898 * is compound flags */ | |
| 236 | 6899 { |
| 6900 int todo; | |
| 6901 hashitem_T *hi; | |
| 300 | 6902 affheader_T *ah; |
| 6903 affentry_T *ae; | |
| 6904 char_u newword[MAXWLEN]; | |
| 6905 int retval = OK; | |
| 820 | 6906 int i, j; |
| 300 | 6907 char_u *p; |
| 366 | 6908 int use_flags; |
| 390 | 6909 char_u *use_pfxlist; |
| 820 | 6910 int use_pfxlen; |
| 6911 int need_affix; | |
| 6912 char_u store_afflist[MAXWLEN]; | |
| 483 | 6913 char_u pfx_pfxlist[MAXWLEN]; |
| 6914 size_t wordlen = STRLEN(word); | |
| 820 | 6915 int use_condit; |
| 236 | 6916 |
| 835 | 6917 todo = (int)ht->ht_used; |
| 300 | 6918 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi) |
| 236 | 6919 { |
| 6920 if (!HASHITEM_EMPTY(hi)) | |
| 6921 { | |
| 6922 --todo; | |
| 300 | 6923 ah = HI2AH(hi); |
| 236 | 6924 |
| 300 | 6925 /* Check that the affix combines, if required, and that the word |
| 6926 * supports this affix. */ | |
| 820 | 6927 if (((condit & CONDIT_COMB) == 0 || ah->ah_combine) |
| 6928 && flag_in_afflist(affile->af_flagtype, afflist, | |
| 6929 ah->ah_flag)) | |
| 236 | 6930 { |
| 300 | 6931 /* Loop over all affix entries with this name. */ |
| 6932 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next) | |
| 236 | 6933 { |
| 300 | 6934 /* Check the condition. It's not logical to match case |
| 6935 * here, but it is required for compatibility with | |
| 339 | 6936 * Myspell. |
| 481 | 6937 * Another requirement from Myspell is that the chop |
| 6938 * string is shorter than the word itself. | |
| 339 | 6939 * For prefixes, when "PFXPOSTPONE" was used, only do |
| 820 | 6940 * prefixes with a chop string and/or flags. |
| 6941 * When a previously added affix had CIRCUMFIX this one | |
| 6942 * must have it too, if it had not then this one must not | |
| 6943 * have one either. */ | |
| 339 | 6944 if ((xht != NULL || !affile->af_pfxpostpone |
| 809 | 6945 || ae->ae_chop != NULL |
| 6946 || ae->ae_flags != NULL) | |
| 481 | 6947 && (ae->ae_chop == NULL |
| 6948 || STRLEN(ae->ae_chop) < wordlen) | |
| 339 | 6949 && (ae->ae_prog == NULL |
| 6375 | 6950 || vim_regexec_prog(&ae->ae_prog, FALSE, |
| 6951 word, (colnr_T)0)) | |
| 820 | 6952 && (((condit & CONDIT_CFIX) == 0) |
| 6953 == ((condit & CONDIT_AFF) == 0 | |
| 6954 || ae->ae_flags == NULL | |
| 6955 || !flag_in_afflist(affile->af_flagtype, | |
| 6956 ae->ae_flags, affile->af_circumfix)))) | |
| 300 | 6957 { |
| 6958 /* Match. Remove the chop and add the affix. */ | |
| 6959 if (xht == NULL) | |
| 240 | 6960 { |
| 300 | 6961 /* prefix: chop/add at the start of the word */ |
| 6962 if (ae->ae_add == NULL) | |
| 6963 *newword = NUL; | |
| 6964 else | |
| 2768 | 6965 vim_strncpy(newword, ae->ae_add, MAXWLEN - 1); |
| 300 | 6966 p = word; |
| 6967 if (ae->ae_chop != NULL) | |
| 310 | 6968 { |
| 300 | 6969 /* Skip chop string. */ |
| 310 | 6970 #ifdef FEAT_MBYTE |
| 6971 if (has_mbyte) | |
| 324 | 6972 { |
| 310 | 6973 i = mb_charlen(ae->ae_chop); |
| 324 | 6974 for ( ; i > 0; --i) |
| 6975 mb_ptr_adv(p); | |
| 6976 } | |
| 310 | 6977 else |
| 6978 #endif | |
| 324 | 6979 p += STRLEN(ae->ae_chop); |
| 310 | 6980 } |
| 300 | 6981 STRCAT(newword, p); |
| 6982 } | |
| 6983 else | |
| 6984 { | |
| 6985 /* suffix: chop/add at the end of the word */ | |
| 2768 | 6986 vim_strncpy(newword, word, MAXWLEN - 1); |
| 300 | 6987 if (ae->ae_chop != NULL) |
| 6988 { | |
| 6989 /* Remove chop string. */ | |
| 6990 p = newword + STRLEN(newword); | |
| 835 | 6991 i = (int)MB_CHARLEN(ae->ae_chop); |
| 310 | 6992 for ( ; i > 0; --i) |
| 300 | 6993 mb_ptr_back(newword, p); |
| 6994 *p = NUL; | |
| 6995 } | |
| 6996 if (ae->ae_add != NULL) | |
| 6997 STRCAT(newword, ae->ae_add); | |
| 240 | 6998 } |
| 6999 | |
| 820 | 7000 use_flags = flags; |
| 7001 use_pfxlist = pfxlist; | |
| 7002 use_pfxlen = pfxlen; | |
| 7003 need_affix = FALSE; | |
| 7004 use_condit = condit | CONDIT_COMB | CONDIT_AFF; | |
| 7005 if (ae->ae_flags != NULL) | |
| 7006 { | |
| 7007 /* Extract flags from the affix list. */ | |
| 7008 use_flags |= get_affix_flags(affile, ae->ae_flags); | |
| 7009 | |
| 7010 if (affile->af_needaffix != 0 && flag_in_afflist( | |
| 7011 affile->af_flagtype, ae->ae_flags, | |
| 7012 affile->af_needaffix)) | |
| 7013 need_affix = TRUE; | |
| 7014 | |
| 7015 /* When there is a CIRCUMFIX flag the other affix | |
| 7016 * must also have it and we don't add the word | |
| 7017 * with one affix. */ | |
| 7018 if (affile->af_circumfix != 0 && flag_in_afflist( | |
| 7019 affile->af_flagtype, ae->ae_flags, | |
| 7020 affile->af_circumfix)) | |
| 7021 { | |
| 7022 use_condit |= CONDIT_CFIX; | |
| 7023 if ((condit & CONDIT_CFIX) == 0) | |
| 7024 need_affix = TRUE; | |
| 7025 } | |
| 7026 | |
| 7027 if (affile->af_pfxpostpone | |
| 7028 || spin->si_compflags != NULL) | |
| 7029 { | |
| 7030 if (affile->af_pfxpostpone) | |
| 7031 /* Get prefix IDS from the affix list. */ | |
| 7032 use_pfxlen = get_pfxlist(affile, | |
| 7033 ae->ae_flags, store_afflist); | |
| 7034 else | |
| 7035 use_pfxlen = 0; | |
| 7036 use_pfxlist = store_afflist; | |
| 7037 | |
| 7038 /* Combine the prefix IDs. Avoid adding the | |
| 7039 * same ID twice. */ | |
| 7040 for (i = 0; i < pfxlen; ++i) | |
| 7041 { | |
| 7042 for (j = 0; j < use_pfxlen; ++j) | |
| 7043 if (pfxlist[i] == use_pfxlist[j]) | |
| 7044 break; | |
| 7045 if (j == use_pfxlen) | |
| 7046 use_pfxlist[use_pfxlen++] = pfxlist[i]; | |
| 7047 } | |
| 7048 | |
| 7049 if (spin->si_compflags != NULL) | |
| 7050 /* Get compound IDS from the affix list. */ | |
| 7051 get_compflags(affile, ae->ae_flags, | |
| 7052 use_pfxlist + use_pfxlen); | |
| 7053 | |
| 7054 /* Combine the list of compound flags. | |
| 7055 * Concatenate them to the prefix IDs list. | |
| 7056 * Avoid adding the same ID twice. */ | |
| 7057 for (i = pfxlen; pfxlist[i] != NUL; ++i) | |
| 7058 { | |
| 7059 for (j = use_pfxlen; | |
| 7060 use_pfxlist[j] != NUL; ++j) | |
| 7061 if (pfxlist[i] == use_pfxlist[j]) | |
| 7062 break; | |
| 7063 if (use_pfxlist[j] == NUL) | |
| 7064 { | |
| 7065 use_pfxlist[j++] = pfxlist[i]; | |
| 7066 use_pfxlist[j] = NUL; | |
| 7067 } | |
| 7068 } | |
| 7069 } | |
| 7070 } | |
| 483 | 7071 |
| 819 | 7072 /* Obey a "COMPOUNDFORBIDFLAG" of the affix: don't |
| 809 | 7073 * use the compound flags. */ |
| 822 | 7074 if (use_pfxlist != NULL && ae->ae_compforbid) |
| 483 | 7075 { |
| 820 | 7076 vim_strncpy(pfx_pfxlist, use_pfxlist, use_pfxlen); |
| 483 | 7077 use_pfxlist = pfx_pfxlist; |
| 7078 } | |
| 390 | 7079 |
| 7080 /* When there are postponed prefixes... */ | |
| 393 | 7081 if (spin->si_prefroot != NULL |
| 7082 && spin->si_prefroot->wn_sibling != NULL) | |
| 390 | 7083 { |
| 7084 /* ... add a flag to indicate an affix was used. */ | |
| 7085 use_flags |= WF_HAS_AFF; | |
| 7086 | |
| 7087 /* ... don't use a prefix list if combining | |
| 483 | 7088 * affixes is not allowed. But do use the |
| 7089 * compound flags after them. */ | |
| 827 | 7090 if (!ah->ah_combine && use_pfxlist != NULL) |
| 820 | 7091 use_pfxlist += use_pfxlen; |
| 390 | 7092 } |
| 366 | 7093 |
| 819 | 7094 /* When compounding is supported and there is no |
| 7095 * "COMPOUNDPERMITFLAG" then forbid compounding on the | |
| 7096 * side where the affix is applied. */ | |
| 822 | 7097 if (spin->si_compflags != NULL && !ae->ae_comppermit) |
| 819 | 7098 { |
| 7099 if (xht != NULL) | |
| 7100 use_flags |= WF_NOCOMPAFT; | |
| 7101 else | |
| 7102 use_flags |= WF_NOCOMPBEF; | |
| 7103 } | |
| 7104 | |
| 300 | 7105 /* Store the modified word. */ |
| 474 | 7106 if (store_word(spin, newword, use_flags, |
| 820 | 7107 spin->si_region, use_pfxlist, |
| 7108 need_affix) == FAIL) | |
| 300 | 7109 retval = FAIL; |
| 236 | 7110 |
| 820 | 7111 /* When added a prefix or a first suffix and the affix |
| 7112 * has flags may add a(nother) suffix. RECURSIVE! */ | |
| 7113 if ((condit & CONDIT_SUF) && ae->ae_flags != NULL) | |
| 7114 if (store_aff_word(spin, newword, ae->ae_flags, | |
| 7115 affile, &affile->af_suff, xht, | |
| 7116 use_condit & (xht == NULL | |
| 7117 ? ~0 : ~CONDIT_SUF), | |
| 7118 use_flags, use_pfxlist, pfxlen) == FAIL) | |
| 7119 retval = FAIL; | |
| 7120 | |
| 300 | 7121 /* When added a suffix and combining is allowed also |
| 820 | 7122 * try adding a prefix additionally. Both for the |
| 7123 * word flags and for the affix flags. RECURSIVE! */ | |
| 300 | 7124 if (xht != NULL && ah->ah_combine) |
| 820 | 7125 { |
| 7126 if (store_aff_word(spin, newword, | |
| 7127 afflist, affile, | |
| 7128 xht, NULL, use_condit, | |
| 7129 use_flags, use_pfxlist, | |
| 7130 pfxlen) == FAIL | |
| 7131 || (ae->ae_flags != NULL | |
| 7132 && store_aff_word(spin, newword, | |
| 7133 ae->ae_flags, affile, | |
| 7134 xht, NULL, use_condit, | |
| 7135 use_flags, use_pfxlist, | |
| 7136 pfxlen) == FAIL)) | |
| 300 | 7137 retval = FAIL; |
| 820 | 7138 } |
| 236 | 7139 } |
| 7140 } | |
| 7141 } | |
| 7142 } | |
| 7143 } | |
| 7144 | |
| 7145 return retval; | |
| 7146 } | |
| 7147 | |
| 7148 /* | |
| 300 | 7149 * Read a file with a list of words. |
| 236 | 7150 */ |
| 7151 static int | |
| 474 | 7152 spell_read_wordfile(spin, fname) |
| 7153 spellinfo_T *spin; | |
| 300 | 7154 char_u *fname; |
| 236 | 7155 { |
| 300 | 7156 FILE *fd; |
| 7157 long lnum = 0; | |
| 7158 char_u rline[MAXLINELEN]; | |
| 7159 char_u *line; | |
| 7160 char_u *pc = NULL; | |
| 381 | 7161 char_u *p; |
| 300 | 7162 int l; |
| 7163 int retval = OK; | |
| 7164 int did_word = FALSE; | |
| 7165 int non_ascii = 0; | |
| 307 | 7166 int flags; |
| 316 | 7167 int regionmask; |
| 236 | 7168 |
| 300 | 7169 /* |
| 7170 * Open the file. | |
| 7171 */ | |
| 310 | 7172 fd = mch_fopen((char *)fname, "r"); |
| 300 | 7173 if (fd == NULL) |
| 236 | 7174 { |
| 300 | 7175 EMSG2(_(e_notopen), fname); |
| 7176 return FAIL; | |
| 236 | 7177 } |
| 7178 | |
| 625 | 7179 vim_snprintf((char *)IObuff, IOSIZE, _("Reading word file %s ..."), fname); |
| 7180 spell_message(spin, IObuff); | |
| 300 | 7181 |
| 7182 /* | |
| 7183 * Read all the lines in the file one by one. | |
| 7184 */ | |
| 7185 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int) | |
| 7186 { | |
| 7187 line_breakcheck(); | |
| 7188 ++lnum; | |
| 7189 | |
| 7190 /* Skip comment lines. */ | |
| 7191 if (*rline == '#') | |
| 7192 continue; | |
| 7193 | |
| 7194 /* Remove CR, LF and white space from the end. */ | |
| 835 | 7195 l = (int)STRLEN(rline); |
| 300 | 7196 while (l > 0 && rline[l - 1] <= ' ') |
| 7197 --l; | |
| 7198 if (l == 0) | |
| 7199 continue; /* empty or blank line */ | |
| 7200 rline[l] = NUL; | |
| 7201 | |
| 859 | 7202 /* Convert from "/encoding={encoding}" to 'encoding' when needed. */ |
| 300 | 7203 vim_free(pc); |
| 310 | 7204 #ifdef FEAT_MBYTE |
| 300 | 7205 if (spin->si_conv.vc_type != CONV_NONE) |
| 7206 { | |
| 7207 pc = string_convert(&spin->si_conv, rline, NULL); | |
| 7208 if (pc == NULL) | |
| 7209 { | |
| 7210 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"), | |
| 7211 fname, lnum, rline); | |
| 7212 continue; | |
| 7213 } | |
| 7214 line = pc; | |
| 7215 } | |
| 7216 else | |
| 310 | 7217 #endif |
| 300 | 7218 { |
| 7219 pc = NULL; | |
| 7220 line = rline; | |
| 7221 } | |
| 7222 | |
| 307 | 7223 if (*line == '/') |
| 300 | 7224 { |
| 307 | 7225 ++line; |
| 7226 if (STRNCMP(line, "encoding=", 9) == 0) | |
| 300 | 7227 { |
| 7228 if (spin->si_conv.vc_type != CONV_NONE) | |
| 316 | 7229 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"), |
| 7230 fname, lnum, line - 1); | |
| 300 | 7231 else if (did_word) |
| 316 | 7232 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"), |
| 7233 fname, lnum, line - 1); | |
| 300 | 7234 else |
| 7235 { | |
| 310 | 7236 #ifdef FEAT_MBYTE |
| 7237 char_u *enc; | |
| 7238 | |
| 300 | 7239 /* Setup for conversion to 'encoding'. */ |
| 859 | 7240 line += 9; |
| 316 | 7241 enc = enc_canonize(line); |
| 300 | 7242 if (enc != NULL && !spin->si_ascii |
| 7243 && convert_setup(&spin->si_conv, enc, | |
| 7244 p_enc) == FAIL) | |
| 7245 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"), | |
| 316 | 7246 fname, line, p_enc); |
| 300 | 7247 vim_free(enc); |
| 375 | 7248 spin->si_conv.vc_fail = TRUE; |
| 310 | 7249 #else |
| 7250 smsg((char_u *)_("Conversion in %s not supported"), fname); | |
| 7251 #endif | |
| 300 | 7252 } |
| 307 | 7253 continue; |
| 300 | 7254 } |
| 307 | 7255 |
| 316 | 7256 if (STRNCMP(line, "regions=", 8) == 0) |
| 7257 { | |
| 7258 if (spin->si_region_count > 1) | |
| 7259 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"), | |
| 7260 fname, lnum, line); | |
| 7261 else | |
| 7262 { | |
| 7263 line += 8; | |
| 7264 if (STRLEN(line) > 16) | |
| 7265 smsg((char_u *)_("Too many regions in %s line %d: %s"), | |
| 7266 fname, lnum, line); | |
| 7267 else | |
| 7268 { | |
| 835 | 7269 spin->si_region_count = (int)STRLEN(line) / 2; |
| 316 | 7270 STRCPY(spin->si_region_name, line); |
| 389 | 7271 |
| 7272 /* Adjust the mask for a word valid in all regions. */ | |
| 7273 spin->si_region = (1 << spin->si_region_count) - 1; | |
| 316 | 7274 } |
| 7275 } | |
| 7276 continue; | |
| 7277 } | |
| 7278 | |
| 381 | 7279 smsg((char_u *)_("/ line ignored in %s line %d: %s"), |
| 7280 fname, lnum, line - 1); | |
| 7281 continue; | |
| 7282 } | |
| 7283 | |
| 7284 flags = 0; | |
| 7285 regionmask = spin->si_region; | |
| 7286 | |
| 7287 /* Check for flags and region after a slash. */ | |
| 7288 p = vim_strchr(line, '/'); | |
| 7289 if (p != NULL) | |
| 7290 { | |
| 7291 *p++ = NUL; | |
| 7292 while (*p != NUL) | |
| 7293 { | |
| 7294 if (*p == '=') /* keep-case word */ | |
| 389 | 7295 flags |= WF_KEEPCAP | WF_FIXCAP; |
| 381 | 7296 else if (*p == '!') /* Bad, bad, wicked word. */ |
| 7297 flags |= WF_BANNED; | |
| 7298 else if (*p == '?') /* Rare word. */ | |
| 7299 flags |= WF_RARE; | |
| 7300 else if (VIM_ISDIGIT(*p)) /* region number(s) */ | |
| 316 | 7301 { |
| 381 | 7302 if ((flags & WF_REGION) == 0) /* first one */ |
| 7303 regionmask = 0; | |
| 7304 flags |= WF_REGION; | |
| 7305 | |
| 7306 l = *p - '0'; | |
| 316 | 7307 if (l > spin->si_region_count) |
| 7308 { | |
| 7309 smsg((char_u *)_("Invalid region nr in %s line %d: %s"), | |
| 381 | 7310 fname, lnum, p); |
| 316 | 7311 break; |
| 7312 } | |
| 7313 regionmask |= 1 << (l - 1); | |
| 7314 } | |
| 381 | 7315 else |
| 7316 { | |
| 7317 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"), | |
| 7318 fname, lnum, p); | |
| 7319 break; | |
| 7320 } | |
| 7321 ++p; | |
| 307 | 7322 } |
| 300 | 7323 } |
| 7324 | |
| 7325 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */ | |
| 7326 if (spin->si_ascii && has_non_ascii(line)) | |
| 7327 { | |
| 7328 ++non_ascii; | |
| 7329 continue; | |
| 7330 } | |
| 7331 | |
| 7332 /* Normal word: store it. */ | |
| 483 | 7333 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL) |
| 300 | 7334 { |
| 7335 retval = FAIL; | |
| 7336 break; | |
| 7337 } | |
| 7338 did_word = TRUE; | |
| 7339 } | |
| 7340 | |
| 7341 vim_free(pc); | |
| 7342 fclose(fd); | |
| 7343 | |
| 625 | 7344 if (spin->si_ascii && non_ascii > 0) |
| 7345 { | |
| 7346 vim_snprintf((char *)IObuff, IOSIZE, | |
| 7347 _("Ignored %d words with non-ASCII characters"), non_ascii); | |
| 7348 spell_message(spin, IObuff); | |
| 7349 } | |
| 7350 | |
| 300 | 7351 return retval; |
| 236 | 7352 } |
| 7353 | |
| 7354 /* | |
| 300 | 7355 * Get part of an sblock_T, "len" bytes long. |
| 474 | 7356 * This avoids calling free() for every little struct we use (and keeping |
| 7357 * track of them). | |
| 300 | 7358 * The memory is cleared to all zeros. |
| 7359 * Returns NULL when out of memory. | |
| 7360 */ | |
| 7361 static void * | |
| 474 | 7362 getroom(spin, len, align) |
| 7363 spellinfo_T *spin; | |
| 447 | 7364 size_t len; /* length needed */ |
| 7365 int align; /* align for pointer */ | |
| 300 | 7366 { |
| 7367 char_u *p; | |
| 474 | 7368 sblock_T *bl = spin->si_blocks; |
| 300 | 7369 |
| 447 | 7370 if (align && bl != NULL) |
| 7371 /* Round size up for alignment. On some systems structures need to be | |
| 7372 * aligned to the size of a pointer (e.g., SPARC). */ | |
| 7373 bl->sb_used = (bl->sb_used + sizeof(char *) - 1) | |
| 7374 & ~(sizeof(char *) - 1); | |
| 7375 | |
| 300 | 7376 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE) |
| 7377 { | |
| 2691 | 7378 if (len >= SBLOCKSIZE) |
| 7379 bl = NULL; | |
| 7380 else | |
| 7381 /* Allocate a block of memory. It is not freed until much later. */ | |
| 7382 bl = (sblock_T *)alloc_clear( | |
| 7383 (unsigned)(sizeof(sblock_T) + SBLOCKSIZE)); | |
| 300 | 7384 if (bl == NULL) |
| 2691 | 7385 { |
| 7386 if (!spin->si_did_emsg) | |
| 7387 { | |
| 7388 EMSG(_("E845: Insufficient memory, word list will be incomplete")); | |
| 7389 spin->si_did_emsg = TRUE; | |
| 7390 } | |
| 300 | 7391 return NULL; |
| 2691 | 7392 } |
| 474 | 7393 bl->sb_next = spin->si_blocks; |
| 7394 spin->si_blocks = bl; | |
| 300 | 7395 bl->sb_used = 0; |
| 482 | 7396 ++spin->si_blocks_cnt; |
| 300 | 7397 } |
| 7398 | |
| 7399 p = bl->sb_data + bl->sb_used; | |
| 835 | 7400 bl->sb_used += (int)len; |
| 300 | 7401 |
| 7402 return p; | |
| 7403 } | |
| 7404 | |
| 7405 /* | |
| 7406 * Make a copy of a string into memory allocated with getroom(). | |
| 2691 | 7407 * Returns NULL when out of memory. |
| 300 | 7408 */ |
| 7409 static char_u * | |
| 474 | 7410 getroom_save(spin, s) |
| 7411 spellinfo_T *spin; | |
| 300 | 7412 char_u *s; |
| 7413 { | |
| 7414 char_u *sc; | |
| 7415 | |
| 474 | 7416 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE); |
| 300 | 7417 if (sc != NULL) |
| 7418 STRCPY(sc, s); | |
| 7419 return sc; | |
| 7420 } | |
| 7421 | |
| 7422 | |
| 7423 /* | |
| 7424 * Free the list of allocated sblock_T. | |
| 236 | 7425 */ |
| 7426 static void | |
| 300 | 7427 free_blocks(bl) |
| 7428 sblock_T *bl; | |
| 236 | 7429 { |
| 300 | 7430 sblock_T *next; |
| 236 | 7431 |
| 300 | 7432 while (bl != NULL) |
| 236 | 7433 { |
| 300 | 7434 next = bl->sb_next; |
| 7435 vim_free(bl); | |
| 7436 bl = next; | |
| 236 | 7437 } |
| 7438 } | |
| 7439 | |
| 7440 /* | |
| 300 | 7441 * Allocate the root of a word tree. |
| 2691 | 7442 * Returns NULL when out of memory. |
| 236 | 7443 */ |
| 300 | 7444 static wordnode_T * |
| 474 | 7445 wordtree_alloc(spin) |
| 7446 spellinfo_T *spin; | |
| 7447 { | |
| 7448 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE); | |
| 236 | 7449 } |
| 7450 | |
| 7451 /* | |
| 300 | 7452 * Store a word in the tree(s). |
| 390 | 7453 * Always store it in the case-folded tree. For a keep-case word this is |
| 7454 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and | |
| 7455 * used to find suggestions. | |
| 300 | 7456 * For a keep-case word also store it in the keep-case tree. |
| 482 | 7457 * When "pfxlist" is not NULL store the word for each postponed prefix ID and |
| 7458 * compound flag. | |
| 236 | 7459 */ |
| 7460 static int | |
| 483 | 7461 store_word(spin, word, flags, region, pfxlist, need_affix) |
| 474 | 7462 spellinfo_T *spin; |
| 300 | 7463 char_u *word; |
| 307 | 7464 int flags; /* extra flags, WF_BANNED */ |
| 316 | 7465 int region; /* supported region(s) */ |
| 339 | 7466 char_u *pfxlist; /* list of prefix IDs or NULL */ |
| 483 | 7467 int need_affix; /* only store word with affix ID */ |
| 236 | 7468 { |
| 835 | 7469 int len = (int)STRLEN(word); |
| 300 | 7470 int ct = captype(word, word + len); |
| 7471 char_u foldword[MAXWLEN]; | |
| 339 | 7472 int res = OK; |
| 7473 char_u *p; | |
| 236 | 7474 |
| 323 | 7475 (void)spell_casefold(word, len, foldword, MAXWLEN); |
| 339 | 7476 for (p = pfxlist; res == OK; ++p) |
| 7477 { | |
| 483 | 7478 if (!need_affix || (p != NULL && *p != NUL)) |
| 7479 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags, | |
| 474 | 7480 region, p == NULL ? 0 : *p); |
| 339 | 7481 if (p == NULL || *p == NUL) |
| 7482 break; | |
| 7483 } | |
| 334 | 7484 ++spin->si_foldwcount; |
| 307 | 7485 |
| 474 | 7486 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP))) |
| 334 | 7487 { |
| 339 | 7488 for (p = pfxlist; res == OK; ++p) |
| 7489 { | |
| 483 | 7490 if (!need_affix || (p != NULL && *p != NUL)) |
| 7491 res = tree_add_word(spin, word, spin->si_keeproot, flags, | |
| 474 | 7492 region, p == NULL ? 0 : *p); |
| 339 | 7493 if (p == NULL || *p == NUL) |
| 7494 break; | |
| 7495 } | |
| 334 | 7496 ++spin->si_keepwcount; |
| 7497 } | |
| 300 | 7498 return res; |
| 236 | 7499 } |
| 7500 | |
| 7501 /* | |
| 300 | 7502 * Add word "word" to a word tree at "root". |
| 625 | 7503 * When "flags" < 0 we are adding to the prefix tree where "flags" is used for |
| 366 | 7504 * "rare" and "region" is the condition nr. |
| 255 | 7505 * Returns FAIL when out of memory. |
| 236 | 7506 */ |
| 255 | 7507 static int |
| 481 | 7508 tree_add_word(spin, word, root, flags, region, affixID) |
| 474 | 7509 spellinfo_T *spin; |
| 300 | 7510 char_u *word; |
| 7511 wordnode_T *root; | |
| 7512 int flags; | |
| 7513 int region; | |
| 481 | 7514 int affixID; |
| 236 | 7515 { |
| 300 | 7516 wordnode_T *node = root; |
| 7517 wordnode_T *np; | |
| 469 | 7518 wordnode_T *copyp, **copyprev; |
| 300 | 7519 wordnode_T **prev = NULL; |
| 7520 int i; | |
| 255 | 7521 |
| 300 | 7522 /* Add each byte of the word to the tree, including the NUL at the end. */ |
| 7523 for (i = 0; ; ++i) | |
| 255 | 7524 { |
| 469 | 7525 /* When there is more than one reference to this node we need to make |
| 7526 * a copy, so that we can modify it. Copy the whole list of siblings | |
| 7527 * (we don't optimize for a partly shared list of siblings). */ | |
| 7528 if (node != NULL && node->wn_refs > 1) | |
| 7529 { | |
| 7530 --node->wn_refs; | |
| 7531 copyprev = prev; | |
| 7532 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling) | |
| 7533 { | |
| 7534 /* Allocate a new node and copy the info. */ | |
| 474 | 7535 np = get_wordnode(spin); |
| 469 | 7536 if (np == NULL) |
| 7537 return FAIL; | |
| 7538 np->wn_child = copyp->wn_child; | |
| 7539 if (np->wn_child != NULL) | |
| 7540 ++np->wn_child->wn_refs; /* child gets extra ref */ | |
| 7541 np->wn_byte = copyp->wn_byte; | |
| 7542 if (np->wn_byte == NUL) | |
| 7543 { | |
| 7544 np->wn_flags = copyp->wn_flags; | |
| 7545 np->wn_region = copyp->wn_region; | |
| 481 | 7546 np->wn_affixID = copyp->wn_affixID; |
| 469 | 7547 } |
| 7548 | |
| 7549 /* Link the new node in the list, there will be one ref. */ | |
| 7550 np->wn_refs = 1; | |
| 840 | 7551 if (copyprev != NULL) |
| 7552 *copyprev = np; | |
| 469 | 7553 copyprev = &np->wn_sibling; |
| 7554 | |
| 7555 /* Let "node" point to the head of the copied list. */ | |
| 7556 if (copyp == node) | |
| 7557 node = np; | |
| 7558 } | |
| 7559 } | |
| 7560 | |
| 300 | 7561 /* Look for the sibling that has the same character. They are sorted |
| 7562 * on byte value, thus stop searching when a sibling is found with a | |
| 339 | 7563 * higher byte value. For zero bytes (end of word) the sorting is |
| 481 | 7564 * done on flags and then on affixID. */ |
| 339 | 7565 while (node != NULL |
| 7566 && (node->wn_byte < word[i] | |
| 7567 || (node->wn_byte == NUL | |
| 7568 && (flags < 0 | |
| 625 | 7569 ? node->wn_affixID < (unsigned)affixID |
| 7570 : (node->wn_flags < (unsigned)(flags & WN_MASK) | |
| 390 | 7571 || (node->wn_flags == (flags & WN_MASK) |
| 625 | 7572 && (spin->si_sugtree |
| 7573 ? (node->wn_region & 0xffff) < region | |
| 7574 : node->wn_affixID | |
| 7575 < (unsigned)affixID))))))) | |
| 236 | 7576 { |
| 300 | 7577 prev = &node->wn_sibling; |
| 7578 node = *prev; | |
| 236 | 7579 } |
| 339 | 7580 if (node == NULL |
| 7581 || node->wn_byte != word[i] | |
| 7582 || (word[i] == NUL | |
| 7583 && (flags < 0 | |
| 625 | 7584 || spin->si_sugtree |
| 390 | 7585 || node->wn_flags != (flags & WN_MASK) |
| 481 | 7586 || node->wn_affixID != affixID))) |
| 255 | 7587 { |
| 300 | 7588 /* Allocate a new node. */ |
| 474 | 7589 np = get_wordnode(spin); |
| 300 | 7590 if (np == NULL) |
| 7591 return FAIL; | |
| 7592 np->wn_byte = word[i]; | |
| 469 | 7593 |
| 7594 /* If "node" is NULL this is a new child or the end of the sibling | |
| 7595 * list: ref count is one. Otherwise use ref count of sibling and | |
| 7596 * make ref count of sibling one (matters when inserting in front | |
| 7597 * of the list of siblings). */ | |
| 7598 if (node == NULL) | |
| 7599 np->wn_refs = 1; | |
| 7600 else | |
| 7601 { | |
| 7602 np->wn_refs = node->wn_refs; | |
| 7603 node->wn_refs = 1; | |
| 7604 } | |
|
2299
92e31a1b0c2f
Fix crash in generating spell .sug file.
Bram Moolenaar <bram@vim.org>
parents:
2250
diff
changeset
|
7605 if (prev != NULL) |
|
92e31a1b0c2f
Fix crash in generating spell .sug file.
Bram Moolenaar <bram@vim.org>
parents:
2250
diff
changeset
|
7606 *prev = np; |
| 300 | 7607 np->wn_sibling = node; |
| 7608 node = np; | |
| 255 | 7609 } |
| 300 | 7610 |
| 7611 if (word[i] == NUL) | |
| 7612 { | |
| 7613 node->wn_flags = flags; | |
| 7614 node->wn_region |= region; | |
| 481 | 7615 node->wn_affixID = affixID; |
| 300 | 7616 break; |
| 7617 } | |
| 7618 prev = &node->wn_child; | |
| 7619 node = *prev; | |
| 255 | 7620 } |
| 469 | 7621 #ifdef SPELL_PRINTTREE |
| 7622 smsg("Added \"%s\"", word); | |
| 7623 spell_print_tree(root->wn_sibling); | |
| 7624 #endif | |
| 7625 | |
| 474 | 7626 /* count nr of words added since last message */ |
| 7627 ++spin->si_msg_count; | |
| 7628 | |
| 482 | 7629 if (spin->si_compress_cnt > 1) |
| 7630 { | |
| 7631 if (--spin->si_compress_cnt == 1) | |
| 7632 /* Did enough words to lower the block count limit. */ | |
| 483 | 7633 spin->si_blocks_cnt += compress_inc; |
| 482 | 7634 } |
| 7635 | |
| 469 | 7636 /* |
| 482 | 7637 * When we have allocated lots of memory we need to compress the word tree |
| 7638 * to free up some room. But compression is slow, and we might actually | |
| 7639 * need that room, thus only compress in the following situations: | |
| 7640 * 1. When not compressed before (si_compress_cnt == 0): when using | |
| 483 | 7641 * "compress_start" blocks. |
| 7642 * 2. When compressed before and used "compress_inc" blocks before | |
| 7643 * adding "compress_added" words (si_compress_cnt > 1). | |
| 7644 * 3. When compressed before, added "compress_added" words | |
| 482 | 7645 * (si_compress_cnt == 1) and the number of free nodes drops below the |
| 7646 * maximum word length. | |
| 469 | 7647 */ |
| 482 | 7648 #ifndef SPELL_PRINTTREE |
| 7649 if (spin->si_compress_cnt == 1 | |
| 7650 ? spin->si_free_count < MAXWLEN | |
| 483 | 7651 : spin->si_blocks_cnt >= compress_start) |
| 482 | 7652 #endif |
| 7653 { | |
| 7654 /* Decrement the block counter. The effect is that we compress again | |
| 483 | 7655 * when the freed up room has been used and another "compress_inc" |
| 7656 * blocks have been allocated. Unless "compress_added" words have | |
| 482 | 7657 * been added, then the limit is put back again. */ |
| 483 | 7658 spin->si_blocks_cnt -= compress_inc; |
| 7659 spin->si_compress_cnt = compress_added; | |
| 482 | 7660 |
| 474 | 7661 if (spin->si_verbose) |
| 7662 { | |
| 7663 msg_start(); | |
| 7664 msg_puts((char_u *)_(msg_compressing)); | |
| 7665 msg_clr_eos(); | |
| 7666 msg_didout = FALSE; | |
| 7667 msg_col = 0; | |
| 7668 out_flush(); | |
| 7669 } | |
| 482 | 7670 |
| 7671 /* Compress both trees. Either they both have many nodes, which makes | |
| 7672 * compression useful, or one of them is small, which means | |
| 4352 | 7673 * compression goes fast. But when filling the soundfold word tree |
| 625 | 7674 * there is no keep-case tree. */ |
| 482 | 7675 wordtree_compress(spin, spin->si_foldroot); |
| 625 | 7676 if (affixID >= 0) |
| 7677 wordtree_compress(spin, spin->si_keeproot); | |
| 469 | 7678 } |
| 255 | 7679 |
| 7680 return OK; | |
| 236 | 7681 } |
| 7682 | |
| 469 | 7683 /* |
| 483 | 7684 * Check the 'mkspellmem' option. Return FAIL if it's wrong. |
| 7685 * Sets "sps_flags". | |
| 7686 */ | |
| 7687 int | |
| 7688 spell_check_msm() | |
| 7689 { | |
| 7690 char_u *p = p_msm; | |
| 7691 long start = 0; | |
| 944 | 7692 long incr = 0; |
| 483 | 7693 long added = 0; |
| 7694 | |
| 7695 if (!VIM_ISDIGIT(*p)) | |
| 7696 return FAIL; | |
| 7697 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/ | |
| 7698 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102); | |
| 7699 if (*p != ',') | |
| 7700 return FAIL; | |
| 7701 ++p; | |
| 7702 if (!VIM_ISDIGIT(*p)) | |
| 7703 return FAIL; | |
| 944 | 7704 incr = (getdigits(&p) * 102) / (SBLOCKSIZE / 10); |
| 483 | 7705 if (*p != ',') |
| 7706 return FAIL; | |
| 7707 ++p; | |
| 7708 if (!VIM_ISDIGIT(*p)) | |
| 7709 return FAIL; | |
| 7710 added = getdigits(&p) * 1024; | |
| 7711 if (*p != NUL) | |
| 7712 return FAIL; | |
| 7713 | |
| 944 | 7714 if (start == 0 || incr == 0 || added == 0 || incr > start) |
| 483 | 7715 return FAIL; |
| 7716 | |
| 7717 compress_start = start; | |
| 944 | 7718 compress_inc = incr; |
| 483 | 7719 compress_added = added; |
| 7720 return OK; | |
| 7721 } | |
| 7722 | |
| 7723 | |
| 7724 /* | |
| 469 | 7725 * Get a wordnode_T, either from the list of previously freed nodes or |
| 7726 * allocate a new one. | |
| 2691 | 7727 * Returns NULL when out of memory. |
| 469 | 7728 */ |
| 7729 static wordnode_T * | |
| 474 | 7730 get_wordnode(spin) |
| 7731 spellinfo_T *spin; | |
| 469 | 7732 { |
| 7733 wordnode_T *n; | |
| 7734 | |
| 474 | 7735 if (spin->si_first_free == NULL) |
| 7736 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE); | |
| 469 | 7737 else |
| 7738 { | |
| 474 | 7739 n = spin->si_first_free; |
| 7740 spin->si_first_free = n->wn_child; | |
| 469 | 7741 vim_memset(n, 0, sizeof(wordnode_T)); |
| 482 | 7742 --spin->si_free_count; |
| 469 | 7743 } |
| 7744 #ifdef SPELL_PRINTTREE | |
| 2691 | 7745 if (n != NULL) |
| 7746 n->wn_nr = ++spin->si_wordnode_nr; | |
| 469 | 7747 #endif |
| 7748 return n; | |
| 7749 } | |
| 7750 | |
| 7751 /* | |
| 474 | 7752 * Decrement the reference count on a node (which is the head of a list of |
| 7753 * siblings). If the reference count becomes zero free the node and its | |
| 7754 * siblings. | |
| 625 | 7755 * Returns the number of nodes actually freed. |
| 7756 */ | |
| 7757 static int | |
| 474 | 7758 deref_wordnode(spin, node) |
| 7759 spellinfo_T *spin; | |
| 7760 wordnode_T *node; | |
| 7761 { | |
| 625 | 7762 wordnode_T *np; |
| 7763 int cnt = 0; | |
| 474 | 7764 |
| 7765 if (--node->wn_refs == 0) | |
| 625 | 7766 { |
| 474 | 7767 for (np = node; np != NULL; np = np->wn_sibling) |
| 7768 { | |
| 7769 if (np->wn_child != NULL) | |
| 625 | 7770 cnt += deref_wordnode(spin, np->wn_child); |
| 474 | 7771 free_wordnode(spin, np); |
| 625 | 7772 ++cnt; |
| 7773 } | |
| 7774 ++cnt; /* length field */ | |
| 7775 } | |
| 7776 return cnt; | |
| 474 | 7777 } |
| 7778 | |
| 7779 /* | |
| 469 | 7780 * Free a wordnode_T for re-use later. |
| 474 | 7781 * Only the "wn_child" field becomes invalid. |
| 469 | 7782 */ |
| 7783 static void | |
| 474 | 7784 free_wordnode(spin, n) |
| 7785 spellinfo_T *spin; | |
| 7786 wordnode_T *n; | |
| 7787 { | |
| 7788 n->wn_child = spin->si_first_free; | |
| 7789 spin->si_first_free = n; | |
| 482 | 7790 ++spin->si_free_count; |
| 469 | 7791 } |
| 7792 | |
| 236 | 7793 /* |
| 300 | 7794 * Compress a tree: find tails that are identical and can be shared. |
| 7795 */ | |
| 7796 static void | |
| 474 | 7797 wordtree_compress(spin, root) |
| 7798 spellinfo_T *spin; | |
| 300 | 7799 wordnode_T *root; |
| 7800 { | |
| 7801 hashtab_T ht; | |
| 7802 int n; | |
| 7803 int tot = 0; | |
| 469 | 7804 int perc; |
| 300 | 7805 |
| 474 | 7806 /* Skip the root itself, it's not actually used. The first sibling is the |
| 7807 * start of the tree. */ | |
| 7808 if (root->wn_sibling != NULL) | |
| 300 | 7809 { |
| 7810 hash_init(&ht); | |
| 474 | 7811 n = node_compress(spin, root->wn_sibling, &ht, &tot); |
| 469 | 7812 |
| 7813 #ifndef SPELL_PRINTTREE | |
| 310 | 7814 if (spin->si_verbose || p_verbose > 2) |
| 469 | 7815 #endif |
| 310 | 7816 { |
| 469 | 7817 if (tot > 1000000) |
| 7818 perc = (tot - n) / (tot / 100); | |
| 482 | 7819 else if (tot == 0) |
| 7820 perc = 0; | |
| 469 | 7821 else |
| 7822 perc = (tot - n) * 100 / tot; | |
| 625 | 7823 vim_snprintf((char *)IObuff, IOSIZE, |
| 7824 _("Compressed %d of %d nodes; %d (%d%%) remaining"), | |
| 7825 n, tot, tot - n, perc); | |
| 7826 spell_message(spin, IObuff); | |
| 310 | 7827 } |
| 469 | 7828 #ifdef SPELL_PRINTTREE |
| 474 | 7829 spell_print_tree(root->wn_sibling); |
| 469 | 7830 #endif |
| 300 | 7831 hash_clear(&ht); |
| 7832 } | |
| 7833 } | |
| 7834 | |
| 7835 /* | |
| 7836 * Compress a node, its siblings and its children, depth first. | |
| 7837 * Returns the number of compressed nodes. | |
| 236 | 7838 */ |
| 255 | 7839 static int |
| 474 | 7840 node_compress(spin, node, ht, tot) |
| 7841 spellinfo_T *spin; | |
| 300 | 7842 wordnode_T *node; |
| 7843 hashtab_T *ht; | |
| 7844 int *tot; /* total count of nodes before compressing, | |
| 7845 incremented while going through the tree */ | |
| 236 | 7846 { |
| 300 | 7847 wordnode_T *np; |
| 7848 wordnode_T *tp; | |
| 7849 wordnode_T *child; | |
| 7850 hash_T hash; | |
| 236 | 7851 hashitem_T *hi; |
| 300 | 7852 int len = 0; |
| 7853 unsigned nr, n; | |
| 7854 int compressed = 0; | |
| 236 | 7855 |
| 300 | 7856 /* |
| 7857 * Go through the list of siblings. Compress each child and then try | |
| 7858 * finding an identical child to replace it. | |
| 7859 * Note that with "child" we mean not just the node that is pointed to, | |
| 625 | 7860 * but the whole list of siblings of which the child node is the first. |
| 300 | 7861 */ |
| 482 | 7862 for (np = node; np != NULL && !got_int; np = np->wn_sibling) |
| 236 | 7863 { |
| 300 | 7864 ++len; |
| 7865 if ((child = np->wn_child) != NULL) | |
| 7866 { | |
| 625 | 7867 /* Compress the child first. This fills hashkey. */ |
| 474 | 7868 compressed += node_compress(spin, child, ht, tot); |
| 300 | 7869 |
| 7870 /* Try to find an identical child. */ | |
| 346 | 7871 hash = hash_hash(child->wn_u1.hashkey); |
| 7872 hi = hash_lookup(ht, child->wn_u1.hashkey, hash); | |
| 300 | 7873 if (!HASHITEM_EMPTY(hi)) |
| 7874 { | |
| 625 | 7875 /* There are children we encountered before with a hash value |
| 7876 * identical to the current child. Now check if there is one | |
| 7877 * that is really identical. */ | |
| 346 | 7878 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next) |
| 300 | 7879 if (node_equal(child, tp)) |
| 7880 { | |
| 7881 /* Found one! Now use that child in place of the | |
| 469 | 7882 * current one. This means the current child and all |
| 7883 * its siblings is unlinked from the tree. */ | |
| 7884 ++tp->wn_refs; | |
| 625 | 7885 compressed += deref_wordnode(spin, child); |
| 300 | 7886 np->wn_child = tp; |
| 7887 break; | |
| 7888 } | |
| 7889 if (tp == NULL) | |
| 7890 { | |
| 7891 /* No other child with this hash value equals the child of | |
| 7892 * the node, add it to the linked list after the first | |
| 7893 * item. */ | |
| 7894 tp = HI2WN(hi); | |
| 346 | 7895 child->wn_u2.next = tp->wn_u2.next; |
| 7896 tp->wn_u2.next = child; | |
| 300 | 7897 } |
| 7898 } | |
| 7899 else | |
| 7900 /* No other child has this hash value, add it to the | |
| 7901 * hashtable. */ | |
| 346 | 7902 hash_add_item(ht, hi, child->wn_u1.hashkey, hash); |
| 300 | 7903 } |
| 236 | 7904 } |
| 625 | 7905 *tot += len + 1; /* add one for the node that stores the length */ |
| 300 | 7906 |
| 7907 /* | |
| 7908 * Make a hash key for the node and its siblings, so that we can quickly | |
| 7909 * find a lookalike node. This must be done after compressing the sibling | |
| 7910 * list, otherwise the hash key would become invalid by the compression. | |
| 7911 */ | |
| 346 | 7912 node->wn_u1.hashkey[0] = len; |
| 300 | 7913 nr = 0; |
| 7914 for (np = node; np != NULL; np = np->wn_sibling) | |
| 236 | 7915 { |
| 300 | 7916 if (np->wn_byte == NUL) |
| 481 | 7917 /* end node: use wn_flags, wn_region and wn_affixID */ |
| 7918 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16); | |
| 300 | 7919 else |
| 7920 /* byte node: use the byte value and the child pointer */ | |
| 835 | 7921 n = (unsigned)(np->wn_byte + ((long_u)np->wn_child << 8)); |
| 300 | 7922 nr = nr * 101 + n; |
| 236 | 7923 } |
| 300 | 7924 |
| 7925 /* Avoid NUL bytes, it terminates the hash key. */ | |
| 7926 n = nr & 0xff; | |
| 346 | 7927 node->wn_u1.hashkey[1] = n == 0 ? 1 : n; |
| 300 | 7928 n = (nr >> 8) & 0xff; |
| 346 | 7929 node->wn_u1.hashkey[2] = n == 0 ? 1 : n; |
| 300 | 7930 n = (nr >> 16) & 0xff; |
| 346 | 7931 node->wn_u1.hashkey[3] = n == 0 ? 1 : n; |
| 300 | 7932 n = (nr >> 24) & 0xff; |
| 346 | 7933 node->wn_u1.hashkey[4] = n == 0 ? 1 : n; |
| 7934 node->wn_u1.hashkey[5] = NUL; | |
| 300 | 7935 |
| 482 | 7936 /* Check for CTRL-C pressed now and then. */ |
| 7937 fast_breakcheck(); | |
| 7938 | |
| 300 | 7939 return compressed; |
| 7940 } | |
| 7941 | |
| 7942 /* | |
| 7943 * Return TRUE when two nodes have identical siblings and children. | |
| 7944 */ | |
| 7945 static int | |
| 7946 node_equal(n1, n2) | |
| 7947 wordnode_T *n1; | |
| 7948 wordnode_T *n2; | |
| 7949 { | |
| 7950 wordnode_T *p1; | |
| 7951 wordnode_T *p2; | |
| 7952 | |
| 7953 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL; | |
| 7954 p1 = p1->wn_sibling, p2 = p2->wn_sibling) | |
| 7955 if (p1->wn_byte != p2->wn_byte | |
| 7956 || (p1->wn_byte == NUL | |
| 7957 ? (p1->wn_flags != p2->wn_flags | |
| 339 | 7958 || p1->wn_region != p2->wn_region |
| 481 | 7959 || p1->wn_affixID != p2->wn_affixID) |
| 300 | 7960 : (p1->wn_child != p2->wn_child))) |
| 7961 break; | |
| 7962 | |
| 7963 return p1 == NULL && p2 == NULL; | |
| 236 | 7964 } |
| 7965 | |
| 323 | 7966 static int |
| 7967 #ifdef __BORLANDC__ | |
| 7968 _RTLENTRYF | |
| 7969 #endif | |
| 7970 rep_compare __ARGS((const void *s1, const void *s2)); | |
| 7971 | |
| 7972 /* | |
| 7973 * Function given to qsort() to sort the REP items on "from" string. | |
| 7974 */ | |
| 7975 static int | |
| 7976 #ifdef __BORLANDC__ | |
| 7977 _RTLENTRYF | |
| 7978 #endif | |
| 7979 rep_compare(s1, s2) | |
| 7980 const void *s1; | |
| 7981 const void *s2; | |
| 7982 { | |
| 7983 fromto_T *p1 = (fromto_T *)s1; | |
| 7984 fromto_T *p2 = (fromto_T *)s2; | |
| 7985 | |
| 7986 return STRCMP(p1->ft_from, p2->ft_from); | |
| 7987 } | |
| 7988 | |
| 236 | 7989 /* |
| 483 | 7990 * Write the Vim .spl file "fname". |
| 500 | 7991 * Return FAIL or OK; |
| 7992 */ | |
| 7993 static int | |
| 474 | 7994 write_vim_spell(spin, fname) |
| 7995 spellinfo_T *spin; | |
| 236 | 7996 char_u *fname; |
| 7997 { | |
| 300 | 7998 FILE *fd; |
| 7999 int regionmask; | |
| 236 | 8000 int round; |
| 300 | 8001 wordnode_T *tree; |
| 8002 int nodecount; | |
| 323 | 8003 int i; |
| 8004 int l; | |
| 8005 garray_T *gap; | |
| 8006 fromto_T *ftp; | |
| 8007 char_u *p; | |
| 8008 int rr; | |
| 500 | 8009 int retval = OK; |
| 1759 | 8010 size_t fwv = 1; /* collect return value of fwrite() to avoid |
| 1753 | 8011 warnings from picky compiler */ |
| 236 | 8012 |
| 310 | 8013 fd = mch_fopen((char *)fname, "w"); |
| 300 | 8014 if (fd == NULL) |
| 236 | 8015 { |
| 8016 EMSG2(_(e_notopen), fname); | |
| 500 | 8017 return FAIL; |
| 236 | 8018 } |
| 8019 | |
| 483 | 8020 /* <HEADER>: <fileID> <versionnr> */ |
| 300 | 8021 /* <fileID> */ |
| 1753 | 8022 fwv &= fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd); |
| 1759 | 8023 if (fwv != (size_t)1) |
| 8024 /* Catch first write error, don't try writing more. */ | |
| 8025 goto theend; | |
| 8026 | |
| 483 | 8027 putc(VIMSPELLVERSION, fd); /* <versionnr> */ |
| 8028 | |
| 8029 /* | |
| 8030 * <SECTIONS>: <section> ... <sectionend> | |
| 8031 */ | |
| 8032 | |
| 714 | 8033 /* SN_INFO: <infotext> */ |
| 8034 if (spin->si_info != NULL) | |
| 8035 { | |
| 8036 putc(SN_INFO, fd); /* <sectionID> */ | |
| 8037 putc(0, fd); /* <sectionflags> */ | |
| 8038 | |
| 835 | 8039 i = (int)STRLEN(spin->si_info); |
| 714 | 8040 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */ |
| 1753 | 8041 fwv &= fwrite(spin->si_info, (size_t)i, (size_t)1, fd); /* <infotext> */ |
| 714 | 8042 } |
| 8043 | |
| 483 | 8044 /* SN_REGION: <regionname> ... |
| 8045 * Write the region names only if there is more than one. */ | |
| 316 | 8046 if (spin->si_region_count > 1) |
| 236 | 8047 { |
| 483 | 8048 putc(SN_REGION, fd); /* <sectionID> */ |
| 8049 putc(SNF_REQUIRED, fd); /* <sectionflags> */ | |
| 8050 l = spin->si_region_count * 2; | |
| 8051 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */ | |
| 1753 | 8052 fwv &= fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd); |
| 483 | 8053 /* <regionname> ... */ |
| 316 | 8054 regionmask = (1 << spin->si_region_count) - 1; |
| 236 | 8055 } |
| 8056 else | |
| 300 | 8057 regionmask = 0; |
| 483 | 8058 |
| 8059 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars> | |
| 8060 * | |
| 8061 * The table with character flags and the table for case folding. | |
| 8062 * This makes sure the same characters are recognized as word characters | |
| 8063 * when generating an when using a spell file. | |
| 260 | 8064 * Skip this for ASCII, the table may conflict with the one used for |
| 323 | 8065 * 'encoding'. |
| 8066 * Also skip this for an .add.spl file, the main spell file must contain | |
| 8067 * the table (avoids that it conflicts). File is shorter too. | |
| 8068 */ | |
| 483 | 8069 if (!spin->si_ascii && !spin->si_add) |
| 8070 { | |
| 8071 char_u folchars[128 * 8]; | |
| 8072 int flags; | |
| 8073 | |
| 485 | 8074 putc(SN_CHARFLAGS, fd); /* <sectionID> */ |
| 483 | 8075 putc(SNF_REQUIRED, fd); /* <sectionflags> */ |
| 8076 | |
| 8077 /* Form the <folchars> string first, we need to know its length. */ | |
| 8078 l = 0; | |
| 8079 for (i = 128; i < 256; ++i) | |
| 8080 { | |
| 8081 #ifdef FEAT_MBYTE | |
| 8082 if (has_mbyte) | |
| 8083 l += mb_char2bytes(spelltab.st_fold[i], folchars + l); | |
| 8084 else | |
| 8085 #endif | |
| 8086 folchars[l++] = spelltab.st_fold[i]; | |
| 8087 } | |
| 8088 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */ | |
| 8089 | |
| 8090 fputc(128, fd); /* <charflagslen> */ | |
| 8091 for (i = 128; i < 256; ++i) | |
| 8092 { | |
| 8093 flags = 0; | |
| 8094 if (spelltab.st_isw[i]) | |
| 8095 flags |= CF_WORD; | |
| 8096 if (spelltab.st_isu[i]) | |
| 8097 flags |= CF_UPPER; | |
| 8098 fputc(flags, fd); /* <charflags> */ | |
| 8099 } | |
| 8100 | |
| 8101 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */ | |
| 1753 | 8102 fwv &= fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */ |
| 483 | 8103 } |
| 8104 | |
| 8105 /* SN_MIDWORD: <midword> */ | |
| 8106 if (spin->si_midword != NULL) | |
| 8107 { | |
| 8108 putc(SN_MIDWORD, fd); /* <sectionID> */ | |
| 8109 putc(SNF_REQUIRED, fd); /* <sectionflags> */ | |
| 8110 | |
| 835 | 8111 i = (int)STRLEN(spin->si_midword); |
| 483 | 8112 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */ |
| 1753 | 8113 fwv &= fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); |
| 8114 /* <midword> */ | |
| 366 | 8115 } |
| 8116 | |
| 483 | 8117 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */ |
| 8118 if (spin->si_prefcond.ga_len > 0) | |
| 8119 { | |
| 8120 putc(SN_PREFCOND, fd); /* <sectionID> */ | |
| 8121 putc(SNF_REQUIRED, fd); /* <sectionflags> */ | |
| 8122 | |
| 8123 l = write_spell_prefcond(NULL, &spin->si_prefcond); | |
| 8124 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */ | |
| 8125 | |
| 8126 write_spell_prefcond(fd, &spin->si_prefcond); | |
| 8127 } | |
| 8128 | |
| 8129 /* SN_REP: <repcount> <rep> ... | |
| 625 | 8130 * SN_SAL: <salflags> <salcount> <sal> ... |
| 8131 * SN_REPSAL: <repcount> <rep> ... */ | |
| 323 | 8132 |
| 483 | 8133 /* round 1: SN_REP section |
| 625 | 8134 * round 2: SN_SAL section (unless SN_SOFO is used) |
| 8135 * round 3: SN_REPSAL section */ | |
| 8136 for (round = 1; round <= 3; ++round) | |
| 323 | 8137 { |
| 8138 if (round == 1) | |
| 8139 gap = &spin->si_rep; | |
| 625 | 8140 else if (round == 2) |
| 8141 { | |
| 8142 /* Don't write SN_SAL when using a SN_SOFO section */ | |
| 8143 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL) | |
| 8144 continue; | |
| 8145 gap = &spin->si_sal; | |
| 483 | 8146 } |
| 323 | 8147 else |
| 625 | 8148 gap = &spin->si_repsal; |
| 8149 | |
| 8150 /* Don't write the section if there are no items. */ | |
| 8151 if (gap->ga_len == 0) | |
| 8152 continue; | |
| 8153 | |
| 8154 /* Sort the REP/REPSAL items. */ | |
| 8155 if (round != 2) | |
| 8156 qsort(gap->ga_data, (size_t)gap->ga_len, | |
| 8157 sizeof(fromto_T), rep_compare); | |
| 8158 | |
| 8159 i = round == 1 ? SN_REP : (round == 2 ? SN_SAL : SN_REPSAL); | |
| 8160 putc(i, fd); /* <sectionID> */ | |
| 483 | 8161 |
| 8162 /* This is for making suggestions, section is not required. */ | |
| 8163 putc(0, fd); /* <sectionflags> */ | |
| 8164 | |
| 8165 /* Compute the length of what follows. */ | |
| 8166 l = 2; /* count <repcount> or <salcount> */ | |
| 8167 for (i = 0; i < gap->ga_len; ++i) | |
| 8168 { | |
| 8169 ftp = &((fromto_T *)gap->ga_data)[i]; | |
| 835 | 8170 l += 1 + (int)STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */ |
| 8171 l += 1 + (int)STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */ | |
| 483 | 8172 } |
| 8173 if (round == 2) | |
| 8174 ++l; /* count <salflags> */ | |
| 8175 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */ | |
| 8176 | |
| 8177 if (round == 2) | |
| 8178 { | |
| 323 | 8179 i = 0; |
| 8180 if (spin->si_followup) | |
| 8181 i |= SAL_F0LLOWUP; | |
| 8182 if (spin->si_collapse) | |
| 8183 i |= SAL_COLLAPSE; | |
| 8184 if (spin->si_rem_accents) | |
| 8185 i |= SAL_REM_ACCENTS; | |
| 8186 putc(i, fd); /* <salflags> */ | |
| 8187 } | |
| 8188 | |
| 8189 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */ | |
| 8190 for (i = 0; i < gap->ga_len; ++i) | |
| 8191 { | |
| 8192 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */ | |
| 8193 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */ | |
| 8194 ftp = &((fromto_T *)gap->ga_data)[i]; | |
| 8195 for (rr = 1; rr <= 2; ++rr) | |
| 8196 { | |
| 8197 p = rr == 1 ? ftp->ft_from : ftp->ft_to; | |
| 835 | 8198 l = (int)STRLEN(p); |
| 323 | 8199 putc(l, fd); |
| 1758 | 8200 if (l > 0) |
| 8201 fwv &= fwrite(p, l, (size_t)1, fd); | |
| 323 | 8202 } |
| 8203 } | |
| 483 | 8204 |
| 8205 } | |
| 8206 | |
| 8207 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto> | |
| 8208 * This is for making suggestions, section is not required. */ | |
| 375 | 8209 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL) |
| 8210 { | |
| 483 | 8211 putc(SN_SOFO, fd); /* <sectionID> */ |
| 8212 putc(0, fd); /* <sectionflags> */ | |
| 375 | 8213 |
| 835 | 8214 l = (int)STRLEN(spin->si_sofofr); |
| 483 | 8215 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4); |
| 8216 /* <sectionlen> */ | |
| 8217 | |
| 8218 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */ | |
| 1753 | 8219 fwv &= fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */ |
| 375 | 8220 |
| 835 | 8221 l = (int)STRLEN(spin->si_sofoto); |
| 483 | 8222 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */ |
| 1753 | 8223 fwv &= fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */ |
| 483 | 8224 } |
| 8225 | |
| 625 | 8226 /* SN_WORDS: <word> ... |
| 8227 * This is for making suggestions, section is not required. */ | |
| 8228 if (spin->si_commonwords.ht_used > 0) | |
| 8229 { | |
| 8230 putc(SN_WORDS, fd); /* <sectionID> */ | |
| 8231 putc(0, fd); /* <sectionflags> */ | |
| 8232 | |
| 8233 /* round 1: count the bytes | |
| 8234 * round 2: write the bytes */ | |
| 8235 for (round = 1; round <= 2; ++round) | |
| 8236 { | |
| 8237 int todo; | |
| 8238 int len = 0; | |
| 8239 hashitem_T *hi; | |
| 8240 | |
| 835 | 8241 todo = (int)spin->si_commonwords.ht_used; |
| 625 | 8242 for (hi = spin->si_commonwords.ht_array; todo > 0; ++hi) |
| 8243 if (!HASHITEM_EMPTY(hi)) | |
| 8244 { | |
| 835 | 8245 l = (int)STRLEN(hi->hi_key) + 1; |
| 625 | 8246 len += l; |
| 8247 if (round == 2) /* <word> */ | |
| 1753 | 8248 fwv &= fwrite(hi->hi_key, (size_t)l, (size_t)1, fd); |
| 625 | 8249 --todo; |
| 8250 } | |
| 8251 if (round == 1) | |
| 8252 put_bytes(fd, (long_u)len, 4); /* <sectionlen> */ | |
| 8253 } | |
| 8254 } | |
| 8255 | |
| 483 | 8256 /* SN_MAP: <mapstr> |
| 8257 * This is for making suggestions, section is not required. */ | |
| 8258 if (spin->si_map.ga_len > 0) | |
| 8259 { | |
| 8260 putc(SN_MAP, fd); /* <sectionID> */ | |
| 8261 putc(0, fd); /* <sectionflags> */ | |
| 8262 l = spin->si_map.ga_len; | |
| 8263 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */ | |
| 1753 | 8264 fwv &= fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd); |
| 483 | 8265 /* <mapstr> */ |
| 8266 } | |
| 8267 | |
| 625 | 8268 /* SN_SUGFILE: <timestamp> |
| 8269 * This is used to notify that a .sug file may be available and at the | |
| 8270 * same time allows for checking that a .sug file that is found matches | |
| 8271 * with this .spl file. That's because the word numbers must be exactly | |
| 8272 * right. */ | |
| 8273 if (!spin->si_nosugfile | |
| 8274 && (spin->si_sal.ga_len > 0 | |
| 8275 || (spin->si_sofofr != NULL && spin->si_sofoto != NULL))) | |
| 8276 { | |
| 8277 putc(SN_SUGFILE, fd); /* <sectionID> */ | |
| 8278 putc(0, fd); /* <sectionflags> */ | |
| 8279 put_bytes(fd, (long_u)8, 4); /* <sectionlen> */ | |
| 8280 | |
| 8281 /* Set si_sugtime and write it to the file. */ | |
| 8282 spin->si_sugtime = time(NULL); | |
|
2229
d45902a5c61c
Fix a few more things for persistent undo.
Bram Moolenaar <bram@vim.org>
parents:
2214
diff
changeset
|
8283 put_time(fd, spin->si_sugtime); /* <timestamp> */ |
| 625 | 8284 } |
| 8285 | |
| 698 | 8286 /* SN_NOSPLITSUGS: nothing |
| 8287 * This is used to notify that no suggestions with word splits are to be | |
| 8288 * made. */ | |
| 8289 if (spin->si_nosplitsugs) | |
| 8290 { | |
| 8291 putc(SN_NOSPLITSUGS, fd); /* <sectionID> */ | |
| 8292 putc(0, fd); /* <sectionflags> */ | |
| 8293 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */ | |
| 8294 } | |
| 8295 | |
| 483 | 8296 /* SN_COMPOUND: compound info. |
| 8297 * We don't mark it required, when not supported all compound words will | |
| 8298 * be bad words. */ | |
| 8299 if (spin->si_compflags != NULL) | |
| 8300 { | |
| 8301 putc(SN_COMPOUND, fd); /* <sectionID> */ | |
| 8302 putc(0, fd); /* <sectionflags> */ | |
| 8303 | |
| 835 | 8304 l = (int)STRLEN(spin->si_compflags); |
| 809 | 8305 for (i = 0; i < spin->si_comppat.ga_len; ++i) |
| 835 | 8306 l += (int)STRLEN(((char_u **)(spin->si_comppat.ga_data))[i]) + 1; |
| 809 | 8307 put_bytes(fd, (long_u)(l + 7), 4); /* <sectionlen> */ |
| 8308 | |
| 483 | 8309 putc(spin->si_compmax, fd); /* <compmax> */ |
| 8310 putc(spin->si_compminlen, fd); /* <compminlen> */ | |
| 8311 putc(spin->si_compsylmax, fd); /* <compsylmax> */ | |
| 809 | 8312 putc(0, fd); /* for Vim 7.0b compatibility */ |
| 8313 putc(spin->si_compoptions, fd); /* <compoptions> */ | |
| 8314 put_bytes(fd, (long_u)spin->si_comppat.ga_len, 2); | |
| 8315 /* <comppatcount> */ | |
| 8316 for (i = 0; i < spin->si_comppat.ga_len; ++i) | |
| 8317 { | |
| 8318 p = ((char_u **)(spin->si_comppat.ga_data))[i]; | |
| 835 | 8319 putc((int)STRLEN(p), fd); /* <comppatlen> */ |
| 1753 | 8320 fwv &= fwrite(p, (size_t)STRLEN(p), (size_t)1, fd); |
| 8321 /* <comppattext> */ | |
| 809 | 8322 } |
| 483 | 8323 /* <compflags> */ |
| 1753 | 8324 fwv &= fwrite(spin->si_compflags, (size_t)STRLEN(spin->si_compflags), |
| 809 | 8325 (size_t)1, fd); |
| 483 | 8326 } |
| 8327 | |
| 492 | 8328 /* SN_NOBREAK: NOBREAK flag */ |
| 8329 if (spin->si_nobreak) | |
| 8330 { | |
| 8331 putc(SN_NOBREAK, fd); /* <sectionID> */ | |
| 8332 putc(0, fd); /* <sectionflags> */ | |
| 8333 | |
| 1188 | 8334 /* It's empty, the presence of the section flags the feature. */ |
| 492 | 8335 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */ |
| 8336 } | |
| 8337 | |
| 483 | 8338 /* SN_SYLLABLE: syllable info. |
| 8339 * We don't mark it required, when not supported syllables will not be | |
| 8340 * counted. */ | |
| 8341 if (spin->si_syllable != NULL) | |
| 8342 { | |
| 8343 putc(SN_SYLLABLE, fd); /* <sectionID> */ | |
| 8344 putc(0, fd); /* <sectionflags> */ | |
| 8345 | |
| 835 | 8346 l = (int)STRLEN(spin->si_syllable); |
| 483 | 8347 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */ |
| 1753 | 8348 fwv &= fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); |
| 8349 /* <syllable> */ | |
| 483 | 8350 } |
| 8351 | |
| 8352 /* end of <SECTIONS> */ | |
| 8353 putc(SN_END, fd); /* <sectionend> */ | |
| 8354 | |
| 302 | 8355 |
| 236 | 8356 /* |
| 339 | 8357 * <LWORDTREE> <KWORDTREE> <PREFIXTREE> |
| 236 | 8358 */ |
| 323 | 8359 spin->si_memtot = 0; |
| 339 | 8360 for (round = 1; round <= 3; ++round) |
| 236 | 8361 { |
| 339 | 8362 if (round == 1) |
| 469 | 8363 tree = spin->si_foldroot->wn_sibling; |
| 339 | 8364 else if (round == 2) |
| 469 | 8365 tree = spin->si_keeproot->wn_sibling; |
| 339 | 8366 else |
| 469 | 8367 tree = spin->si_prefroot->wn_sibling; |
| 236 | 8368 |
| 346 | 8369 /* Clear the index and wnode fields in the tree. */ |
| 8370 clear_node(tree); | |
| 8371 | |
| 300 | 8372 /* Count the number of nodes. Needed to be able to allocate the |
| 346 | 8373 * memory when reading the nodes. Also fills in index for shared |
| 300 | 8374 * nodes. */ |
| 346 | 8375 nodecount = put_node(NULL, tree, 0, regionmask, round == 3); |
| 236 | 8376 |
| 300 | 8377 /* number of nodes in 4 bytes */ |
| 8378 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */ | |
| 302 | 8379 spin->si_memtot += nodecount + nodecount * sizeof(int); |
| 236 | 8380 |
| 300 | 8381 /* Write the nodes. */ |
| 346 | 8382 (void)put_node(fd, tree, 0, regionmask, round == 3); |
| 236 | 8383 } |
| 8384 | |
| 1753 | 8385 /* Write another byte to check for errors (file system full). */ |
| 500 | 8386 if (putc(0, fd) == EOF) |
| 8387 retval = FAIL; | |
| 1759 | 8388 theend: |
| 500 | 8389 if (fclose(fd) == EOF) |
| 8390 retval = FAIL; | |
| 8391 | |
| 1759 | 8392 if (fwv != (size_t)1) |
| 1753 | 8393 retval = FAIL; |
| 8394 if (retval == FAIL) | |
| 8395 EMSG(_(e_write)); | |
| 8396 | |
| 500 | 8397 return retval; |
| 236 | 8398 } |
| 8399 | |
| 8400 /* | |
| 346 | 8401 * Clear the index and wnode fields of "node", it siblings and its |
| 8402 * children. This is needed because they are a union with other items to save | |
| 8403 * space. | |
| 8404 */ | |
| 8405 static void | |
| 8406 clear_node(node) | |
| 8407 wordnode_T *node; | |
| 8408 { | |
| 8409 wordnode_T *np; | |
| 8410 | |
| 8411 if (node != NULL) | |
| 8412 for (np = node; np != NULL; np = np->wn_sibling) | |
| 8413 { | |
| 8414 np->wn_u1.index = 0; | |
| 8415 np->wn_u2.wnode = NULL; | |
| 8416 | |
| 8417 if (np->wn_byte != NUL) | |
| 8418 clear_node(np->wn_child); | |
| 8419 } | |
| 8420 } | |
| 8421 | |
| 8422 | |
| 8423 /* | |
| 300 | 8424 * Dump a word tree at node "node". |
| 8425 * | |
| 8426 * This first writes the list of possible bytes (siblings). Then for each | |
| 8427 * byte recursively write the children. | |
| 8428 * | |
| 625 | 8429 * NOTE: The code here must match the code in read_tree_node(), since |
| 8430 * assumptions are made about the indexes (so that we don't have to write them | |
| 8431 * in the file). | |
| 236 | 8432 * |
| 300 | 8433 * Returns the number of nodes used. |
| 236 | 8434 */ |
| 300 | 8435 static int |
| 944 | 8436 put_node(fd, node, idx, regionmask, prefixtree) |
| 339 | 8437 FILE *fd; /* NULL when only counting */ |
| 300 | 8438 wordnode_T *node; |
| 944 | 8439 int idx; |
| 300 | 8440 int regionmask; |
| 339 | 8441 int prefixtree; /* TRUE for PREFIXTREE */ |
| 236 | 8442 { |
| 944 | 8443 int newindex = idx; |
| 300 | 8444 int siblingcount = 0; |
| 8445 wordnode_T *np; | |
| 236 | 8446 int flags; |
| 300 | 8447 |
| 8448 /* If "node" is zero the tree is empty. */ | |
| 8449 if (node == NULL) | |
| 8450 return 0; | |
| 8451 | |
| 8452 /* Store the index where this node is written. */ | |
| 944 | 8453 node->wn_u1.index = idx; |
| 236 | 8454 |
| 300 | 8455 /* Count the number of siblings. */ |
| 8456 for (np = node; np != NULL; np = np->wn_sibling) | |
| 8457 ++siblingcount; | |
| 236 | 8458 |
| 300 | 8459 /* Write the sibling count. */ |
| 8460 if (fd != NULL) | |
| 8461 putc(siblingcount, fd); /* <siblingcount> */ | |
| 236 | 8462 |
| 300 | 8463 /* Write each sibling byte and optionally extra info. */ |
| 8464 for (np = node; np != NULL; np = np->wn_sibling) | |
| 236 | 8465 { |
| 300 | 8466 if (np->wn_byte == 0) |
| 8467 { | |
| 8468 if (fd != NULL) | |
| 8469 { | |
| 339 | 8470 /* For a NUL byte (end of word) write the flags etc. */ |
| 8471 if (prefixtree) | |
| 300 | 8472 { |
| 481 | 8473 /* In PREFIXTREE write the required affixID and the |
| 390 | 8474 * associated condition nr (stored in wn_region). The |
| 8475 * byte value is misused to store the "rare" and "not | |
| 8476 * combining" flags */ | |
| 455 | 8477 if (np->wn_flags == (short_u)PFX_FLAGS) |
| 8478 putc(BY_NOFLAGS, fd); /* <byte> */ | |
| 8479 else | |
| 8480 { | |
| 390 | 8481 putc(BY_FLAGS, fd); /* <byte> */ |
| 455 | 8482 putc(np->wn_flags, fd); /* <pflags> */ |
| 8483 } | |
| 481 | 8484 putc(np->wn_affixID, fd); /* <affixID> */ |
| 339 | 8485 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */ |
| 300 | 8486 } |
| 8487 else | |
| 8488 { | |
| 339 | 8489 /* For word trees we write the flag/region items. */ |
| 8490 flags = np->wn_flags; | |
| 8491 if (regionmask != 0 && np->wn_region != regionmask) | |
| 8492 flags |= WF_REGION; | |
| 481 | 8493 if (np->wn_affixID != 0) |
| 8494 flags |= WF_AFX; | |
| 339 | 8495 if (flags == 0) |
| 8496 { | |
| 8497 /* word without flags or region */ | |
| 8498 putc(BY_NOFLAGS, fd); /* <byte> */ | |
| 8499 } | |
| 8500 else | |
| 8501 { | |
| 390 | 8502 if (np->wn_flags >= 0x100) |
| 8503 { | |
| 8504 putc(BY_FLAGS2, fd); /* <byte> */ | |
| 8505 putc(flags, fd); /* <flags> */ | |
| 8506 putc((unsigned)flags >> 8, fd); /* <flags2> */ | |
| 8507 } | |
| 8508 else | |
| 8509 { | |
| 8510 putc(BY_FLAGS, fd); /* <byte> */ | |
| 8511 putc(flags, fd); /* <flags> */ | |
| 8512 } | |
| 339 | 8513 if (flags & WF_REGION) |
| 8514 putc(np->wn_region, fd); /* <region> */ | |
| 481 | 8515 if (flags & WF_AFX) |
| 8516 putc(np->wn_affixID, fd); /* <affixID> */ | |
| 339 | 8517 } |
| 300 | 8518 } |
| 8519 } | |
| 8520 } | |
| 8521 else | |
| 8522 { | |
| 346 | 8523 if (np->wn_child->wn_u1.index != 0 |
| 8524 && np->wn_child->wn_u2.wnode != node) | |
| 300 | 8525 { |
| 8526 /* The child is written elsewhere, write the reference. */ | |
| 8527 if (fd != NULL) | |
| 8528 { | |
| 8529 putc(BY_INDEX, fd); /* <byte> */ | |
| 8530 /* <nodeidx> */ | |
| 346 | 8531 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3); |
| 300 | 8532 } |
| 8533 } | |
| 346 | 8534 else if (np->wn_child->wn_u2.wnode == NULL) |
| 300 | 8535 /* We will write the child below and give it an index. */ |
| 346 | 8536 np->wn_child->wn_u2.wnode = node; |
| 236 | 8537 |
| 300 | 8538 if (fd != NULL) |
| 8539 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */ | |
| 8540 { | |
| 8541 EMSG(_(e_write)); | |
| 8542 return 0; | |
| 8543 } | |
| 8544 } | |
| 236 | 8545 } |
| 8546 | |
| 300 | 8547 /* Space used in the array when reading: one for each sibling and one for |
| 8548 * the count. */ | |
| 8549 newindex += siblingcount + 1; | |
| 249 | 8550 |
| 300 | 8551 /* Recursively dump the children of each sibling. */ |
| 8552 for (np = node; np != NULL; np = np->wn_sibling) | |
| 346 | 8553 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node) |
| 8554 newindex = put_node(fd, np->wn_child, newindex, regionmask, | |
| 339 | 8555 prefixtree); |
| 249 | 8556 |
| 300 | 8557 return newindex; |
| 236 | 8558 } |
| 8559 | |
| 8560 | |
| 8561 /* | |
| 310 | 8562 * ":mkspell [-ascii] outfile infile ..." |
| 8563 * ":mkspell [-ascii] addfile" | |
| 236 | 8564 */ |
| 8565 void | |
| 8566 ex_mkspell(eap) | |
| 8567 exarg_T *eap; | |
| 8568 { | |
| 8569 int fcount; | |
| 8570 char_u **fnames; | |
| 310 | 8571 char_u *arg = eap->arg; |
| 8572 int ascii = FALSE; | |
| 8573 | |
| 8574 if (STRNCMP(arg, "-ascii", 6) == 0) | |
| 8575 { | |
| 8576 ascii = TRUE; | |
| 8577 arg = skipwhite(arg + 6); | |
| 8578 } | |
| 8579 | |
| 8580 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */ | |
| 3620 | 8581 if (get_arglist_exp(arg, &fcount, &fnames, FALSE) == OK) |
| 310 | 8582 { |
| 323 | 8583 mkspell(fcount, fnames, ascii, eap->forceit, FALSE); |
| 310 | 8584 FreeWild(fcount, fnames); |
| 8585 } | |
| 8586 } | |
| 8587 | |
| 8588 /* | |
| 625 | 8589 * Create the .sug file. |
| 8590 * Uses the soundfold info in "spin". | |
| 8591 * Writes the file with the name "wfname", with ".spl" changed to ".sug". | |
| 8592 */ | |
| 8593 static void | |
| 8594 spell_make_sugfile(spin, wfname) | |
| 8595 spellinfo_T *spin; | |
| 8596 char_u *wfname; | |
| 8597 { | |
| 2770 | 8598 char_u *fname = NULL; |
| 625 | 8599 int len; |
| 8600 slang_T *slang; | |
| 8601 int free_slang = FALSE; | |
| 8602 | |
| 8603 /* | |
| 8604 * Read back the .spl file that was written. This fills the required | |
| 8605 * info for soundfolding. This also uses less memory than the | |
| 8606 * pointer-linked version of the trie. And it avoids having two versions | |
| 8607 * of the code for the soundfolding stuff. | |
| 8608 * It might have been done already by spell_reload_one(). | |
| 8609 */ | |
| 8610 for (slang = first_lang; slang != NULL; slang = slang->sl_next) | |
| 8611 if (fullpathcmp(wfname, slang->sl_fname, FALSE) == FPC_SAME) | |
| 8612 break; | |
| 8613 if (slang == NULL) | |
| 8614 { | |
| 8615 spell_message(spin, (char_u *)_("Reading back spell file...")); | |
| 8616 slang = spell_load_file(wfname, NULL, NULL, FALSE); | |
| 8617 if (slang == NULL) | |
| 8618 return; | |
| 8619 free_slang = TRUE; | |
| 8620 } | |
| 8621 | |
| 8622 /* | |
| 8623 * Clear the info in "spin" that is used. | |
| 8624 */ | |
| 8625 spin->si_blocks = NULL; | |
| 8626 spin->si_blocks_cnt = 0; | |
| 8627 spin->si_compress_cnt = 0; /* will stay at 0 all the time*/ | |
| 8628 spin->si_free_count = 0; | |
| 8629 spin->si_first_free = NULL; | |
| 8630 spin->si_foldwcount = 0; | |
| 8631 | |
| 8632 /* | |
| 8633 * Go through the trie of good words, soundfold each word and add it to | |
| 8634 * the soundfold trie. | |
| 8635 */ | |
| 8636 spell_message(spin, (char_u *)_("Performing soundfolding...")); | |
| 8637 if (sug_filltree(spin, slang) == FAIL) | |
| 8638 goto theend; | |
| 8639 | |
| 8640 /* | |
| 8641 * Create the table which links each soundfold word with a list of the | |
| 8642 * good words it may come from. Creates buffer "spin->si_spellbuf". | |
| 8643 * This also removes the wordnr from the NUL byte entries to make | |
| 8644 * compression possible. | |
| 8645 */ | |
| 8646 if (sug_maketable(spin) == FAIL) | |
| 8647 goto theend; | |
| 8648 | |
| 8649 smsg((char_u *)_("Number of words after soundfolding: %ld"), | |
| 8650 (long)spin->si_spellbuf->b_ml.ml_line_count); | |
| 8651 | |
| 8652 /* | |
| 8653 * Compress the soundfold trie. | |
| 8654 */ | |
| 8655 spell_message(spin, (char_u *)_(msg_compressing)); | |
| 8656 wordtree_compress(spin, spin->si_foldroot); | |
| 8657 | |
| 8658 /* | |
| 8659 * Write the .sug file. | |
| 8660 * Make the file name by changing ".spl" to ".sug". | |
| 8661 */ | |
| 2770 | 8662 fname = alloc(MAXPATHL); |
| 8663 if (fname == NULL) | |
| 8664 goto theend; | |
| 2768 | 8665 vim_strncpy(fname, wfname, MAXPATHL - 1); |
| 835 | 8666 len = (int)STRLEN(fname); |
| 625 | 8667 fname[len - 2] = 'u'; |
| 8668 fname[len - 1] = 'g'; | |
| 8669 sug_write(spin, fname); | |
| 8670 | |
| 8671 theend: | |
| 2770 | 8672 vim_free(fname); |
| 625 | 8673 if (free_slang) |
| 8674 slang_free(slang); | |
| 8675 free_blocks(spin->si_blocks); | |
| 8676 close_spellbuf(spin->si_spellbuf); | |
| 8677 } | |
| 8678 | |
| 8679 /* | |
| 8680 * Build the soundfold trie for language "slang". | |
| 8681 */ | |
| 8682 static int | |
| 8683 sug_filltree(spin, slang) | |
| 8684 spellinfo_T *spin; | |
| 8685 slang_T *slang; | |
| 8686 { | |
| 8687 char_u *byts; | |
| 8688 idx_T *idxs; | |
| 8689 int depth; | |
| 8690 idx_T arridx[MAXWLEN]; | |
| 8691 int curi[MAXWLEN]; | |
| 8692 char_u tword[MAXWLEN]; | |
| 8693 char_u tsalword[MAXWLEN]; | |
| 8694 int c; | |
| 8695 idx_T n; | |
| 8696 unsigned words_done = 0; | |
| 8697 int wordcount[MAXWLEN]; | |
| 8698 | |
| 4352 | 8699 /* We use si_foldroot for the soundfolded trie. */ |
| 625 | 8700 spin->si_foldroot = wordtree_alloc(spin); |
| 8701 if (spin->si_foldroot == NULL) | |
| 8702 return FAIL; | |
| 8703 | |
| 8704 /* let tree_add_word() know we're adding to the soundfolded tree */ | |
| 8705 spin->si_sugtree = TRUE; | |
| 8706 | |
| 8707 /* | |
| 8708 * Go through the whole case-folded tree, soundfold each word and put it | |
| 8709 * in the trie. | |
| 8710 */ | |
| 8711 byts = slang->sl_fbyts; | |
| 8712 idxs = slang->sl_fidxs; | |
| 8713 | |
| 8714 arridx[0] = 0; | |
| 8715 curi[0] = 1; | |
| 8716 wordcount[0] = 0; | |
| 8717 | |
| 8718 depth = 0; | |
| 8719 while (depth >= 0 && !got_int) | |
| 8720 { | |
| 8721 if (curi[depth] > byts[arridx[depth]]) | |
| 8722 { | |
| 8723 /* Done all bytes at this node, go up one level. */ | |
| 8724 idxs[arridx[depth]] = wordcount[depth]; | |
| 8725 if (depth > 0) | |
| 8726 wordcount[depth - 1] += wordcount[depth]; | |
| 8727 | |
| 8728 --depth; | |
| 8729 line_breakcheck(); | |
| 8730 } | |
| 8731 else | |
| 8732 { | |
| 8733 | |
| 8734 /* Do one more byte at this node. */ | |
| 8735 n = arridx[depth] + curi[depth]; | |
| 8736 ++curi[depth]; | |
| 8737 | |
| 8738 c = byts[n]; | |
| 8739 if (c == 0) | |
| 8740 { | |
| 8741 /* Sound-fold the word. */ | |
| 8742 tword[depth] = NUL; | |
| 8743 spell_soundfold(slang, tword, TRUE, tsalword); | |
| 8744 | |
| 8745 /* We use the "flags" field for the MSB of the wordnr, | |
| 8746 * "region" for the LSB of the wordnr. */ | |
| 8747 if (tree_add_word(spin, tsalword, spin->si_foldroot, | |
| 8748 words_done >> 16, words_done & 0xffff, | |
| 8749 0) == FAIL) | |
| 8750 return FAIL; | |
| 8751 | |
| 8752 ++words_done; | |
| 8753 ++wordcount[depth]; | |
| 8754 | |
| 8755 /* Reset the block count each time to avoid compression | |
| 8756 * kicking in. */ | |
| 8757 spin->si_blocks_cnt = 0; | |
| 8758 | |
| 8759 /* Skip over any other NUL bytes (same word with different | |
| 8760 * flags). */ | |
| 8761 while (byts[n + 1] == 0) | |
| 8762 { | |
| 8763 ++n; | |
| 8764 ++curi[depth]; | |
| 8765 } | |
| 8766 } | |
| 8767 else | |
| 8768 { | |
| 8769 /* Normal char, go one level deeper. */ | |
| 8770 tword[depth++] = c; | |
| 8771 arridx[depth] = idxs[n]; | |
| 8772 curi[depth] = 1; | |
| 8773 wordcount[depth] = 0; | |
| 8774 } | |
| 8775 } | |
| 8776 } | |
| 8777 | |
| 8778 smsg((char_u *)_("Total number of words: %d"), words_done); | |
| 8779 | |
| 8780 return OK; | |
| 8781 } | |
| 8782 | |
| 8783 /* | |
| 8784 * Make the table that links each word in the soundfold trie to the words it | |
| 8785 * can be produced from. | |
| 8786 * This is not unlike lines in a file, thus use a memfile to be able to access | |
| 8787 * the table efficiently. | |
| 8788 * Returns FAIL when out of memory. | |
| 8789 */ | |
| 8790 static int | |
| 8791 sug_maketable(spin) | |
| 8792 spellinfo_T *spin; | |
| 8793 { | |
| 8794 garray_T ga; | |
| 8795 int res = OK; | |
| 8796 | |
| 8797 /* Allocate a buffer, open a memline for it and create the swap file | |
| 8798 * (uses a temp file, not a .swp file). */ | |
| 8799 spin->si_spellbuf = open_spellbuf(); | |
| 8800 if (spin->si_spellbuf == NULL) | |
| 8801 return FAIL; | |
| 8802 | |
| 8803 /* Use a buffer to store the line info, avoids allocating many small | |
| 8804 * pieces of memory. */ | |
| 8805 ga_init2(&ga, 1, 100); | |
| 8806 | |
| 8807 /* recursively go through the tree */ | |
| 8808 if (sug_filltable(spin, spin->si_foldroot->wn_sibling, 0, &ga) == -1) | |
| 8809 res = FAIL; | |
| 8810 | |
| 8811 ga_clear(&ga); | |
| 8812 return res; | |
| 8813 } | |
| 8814 | |
| 8815 /* | |
| 8816 * Fill the table for one node and its children. | |
| 8817 * Returns the wordnr at the start of the node. | |
| 8818 * Returns -1 when out of memory. | |
| 8819 */ | |
| 8820 static int | |
| 8821 sug_filltable(spin, node, startwordnr, gap) | |
| 8822 spellinfo_T *spin; | |
| 8823 wordnode_T *node; | |
| 8824 int startwordnr; | |
| 8825 garray_T *gap; /* place to store line of numbers */ | |
| 8826 { | |
| 8827 wordnode_T *p, *np; | |
| 8828 int wordnr = startwordnr; | |
| 8829 int nr; | |
| 8830 int prev_nr; | |
| 8831 | |
| 8832 for (p = node; p != NULL; p = p->wn_sibling) | |
| 8833 { | |
| 8834 if (p->wn_byte == NUL) | |
| 8835 { | |
| 8836 gap->ga_len = 0; | |
| 8837 prev_nr = 0; | |
| 8838 for (np = p; np != NULL && np->wn_byte == NUL; np = np->wn_sibling) | |
| 8839 { | |
| 8840 if (ga_grow(gap, 10) == FAIL) | |
| 8841 return -1; | |
| 8842 | |
| 8843 nr = (np->wn_flags << 16) + (np->wn_region & 0xffff); | |
| 8844 /* Compute the offset from the previous nr and store the | |
| 8845 * offset in a way that it takes a minimum number of bytes. | |
| 8846 * It's a bit like utf-8, but without the need to mark | |
| 8847 * following bytes. */ | |
| 8848 nr -= prev_nr; | |
| 8849 prev_nr += nr; | |
| 8850 gap->ga_len += offset2bytes(nr, | |
| 8851 (char_u *)gap->ga_data + gap->ga_len); | |
| 8852 } | |
| 8853 | |
| 8854 /* add the NUL byte */ | |
| 8855 ((char_u *)gap->ga_data)[gap->ga_len++] = NUL; | |
| 8856 | |
| 8857 if (ml_append_buf(spin->si_spellbuf, (linenr_T)wordnr, | |
| 8858 gap->ga_data, gap->ga_len, TRUE) == FAIL) | |
| 8859 return -1; | |
| 8860 ++wordnr; | |
| 8861 | |
| 8862 /* Remove extra NUL entries, we no longer need them. We don't | |
| 8863 * bother freeing the nodes, the won't be reused anyway. */ | |
| 8864 while (p->wn_sibling != NULL && p->wn_sibling->wn_byte == NUL) | |
| 8865 p->wn_sibling = p->wn_sibling->wn_sibling; | |
| 8866 | |
| 8867 /* Clear the flags on the remaining NUL node, so that compression | |
| 8868 * works a lot better. */ | |
| 8869 p->wn_flags = 0; | |
| 8870 p->wn_region = 0; | |
| 8871 } | |
| 8872 else | |
| 8873 { | |
| 8874 wordnr = sug_filltable(spin, p->wn_child, wordnr, gap); | |
| 8875 if (wordnr == -1) | |
| 8876 return -1; | |
| 8877 } | |
| 8878 } | |
| 8879 return wordnr; | |
| 8880 } | |
| 8881 | |
| 8882 /* | |
| 8883 * Convert an offset into a minimal number of bytes. | |
| 8884 * Similar to utf_char2byters, but use 8 bits in followup bytes and avoid NUL | |
| 8885 * bytes. | |
| 8886 */ | |
| 8887 static int | |
| 8888 offset2bytes(nr, buf) | |
| 8889 int nr; | |
| 8890 char_u *buf; | |
| 8891 { | |
| 8892 int rem; | |
| 8893 int b1, b2, b3, b4; | |
| 8894 | |
| 8895 /* Split the number in parts of base 255. We need to avoid NUL bytes. */ | |
| 8896 b1 = nr % 255 + 1; | |
| 8897 rem = nr / 255; | |
| 8898 b2 = rem % 255 + 1; | |
| 8899 rem = rem / 255; | |
| 8900 b3 = rem % 255 + 1; | |
| 8901 b4 = rem / 255 + 1; | |
| 8902 | |
| 8903 if (b4 > 1 || b3 > 0x1f) /* 4 bytes */ | |
| 8904 { | |
| 8905 buf[0] = 0xe0 + b4; | |
| 8906 buf[1] = b3; | |
| 8907 buf[2] = b2; | |
| 8908 buf[3] = b1; | |
| 8909 return 4; | |
| 8910 } | |
| 8911 if (b3 > 1 || b2 > 0x3f ) /* 3 bytes */ | |
| 8912 { | |
| 8913 buf[0] = 0xc0 + b3; | |
| 8914 buf[1] = b2; | |
| 8915 buf[2] = b1; | |
| 8916 return 3; | |
| 8917 } | |
| 8918 if (b2 > 1 || b1 > 0x7f ) /* 2 bytes */ | |
| 8919 { | |
| 8920 buf[0] = 0x80 + b2; | |
| 8921 buf[1] = b1; | |
| 8922 return 2; | |
| 8923 } | |
| 8924 /* 1 byte */ | |
| 8925 buf[0] = b1; | |
| 8926 return 1; | |
| 8927 } | |
| 8928 | |
| 8929 /* | |
| 8930 * Opposite of offset2bytes(). | |
| 8931 * "pp" points to the bytes and is advanced over it. | |
| 8932 * Returns the offset. | |
| 8933 */ | |
| 8934 static int | |
| 8935 bytes2offset(pp) | |
| 8936 char_u **pp; | |
| 8937 { | |
| 8938 char_u *p = *pp; | |
| 8939 int nr; | |
| 8940 int c; | |
| 8941 | |
| 8942 c = *p++; | |
| 8943 if ((c & 0x80) == 0x00) /* 1 byte */ | |
| 8944 { | |
| 8945 nr = c - 1; | |
| 8946 } | |
| 8947 else if ((c & 0xc0) == 0x80) /* 2 bytes */ | |
| 8948 { | |
| 8949 nr = (c & 0x3f) - 1; | |
| 8950 nr = nr * 255 + (*p++ - 1); | |
| 8951 } | |
| 8952 else if ((c & 0xe0) == 0xc0) /* 3 bytes */ | |
| 8953 { | |
| 8954 nr = (c & 0x1f) - 1; | |
| 8955 nr = nr * 255 + (*p++ - 1); | |
| 8956 nr = nr * 255 + (*p++ - 1); | |
| 8957 } | |
| 8958 else /* 4 bytes */ | |
| 8959 { | |
| 8960 nr = (c & 0x0f) - 1; | |
| 8961 nr = nr * 255 + (*p++ - 1); | |
| 8962 nr = nr * 255 + (*p++ - 1); | |
| 8963 nr = nr * 255 + (*p++ - 1); | |
| 8964 } | |
| 8965 | |
| 8966 *pp = p; | |
| 8967 return nr; | |
| 8968 } | |
| 8969 | |
| 8970 /* | |
| 8971 * Write the .sug file in "fname". | |
| 8972 */ | |
| 8973 static void | |
| 8974 sug_write(spin, fname) | |
| 8975 spellinfo_T *spin; | |
| 8976 char_u *fname; | |
| 8977 { | |
| 8978 FILE *fd; | |
| 8979 wordnode_T *tree; | |
| 8980 int nodecount; | |
| 8981 int wcount; | |
| 8982 char_u *line; | |
| 8983 linenr_T lnum; | |
| 8984 int len; | |
| 8985 | |
| 8986 /* Create the file. Note that an existing file is silently overwritten! */ | |
| 8987 fd = mch_fopen((char *)fname, "w"); | |
| 8988 if (fd == NULL) | |
| 8989 { | |
| 8990 EMSG2(_(e_notopen), fname); | |
| 8991 return; | |
| 8992 } | |
| 8993 | |
| 8994 vim_snprintf((char *)IObuff, IOSIZE, | |
| 8995 _("Writing suggestion file %s ..."), fname); | |
| 8996 spell_message(spin, IObuff); | |
| 8997 | |
| 8998 /* | |
| 8999 * <SUGHEADER>: <fileID> <versionnr> <timestamp> | |
| 9000 */ | |
| 9001 if (fwrite(VIMSUGMAGIC, VIMSUGMAGICL, (size_t)1, fd) != 1) /* <fileID> */ | |
| 9002 { | |
| 9003 EMSG(_(e_write)); | |
| 9004 goto theend; | |
| 9005 } | |
| 9006 putc(VIMSUGVERSION, fd); /* <versionnr> */ | |
| 9007 | |
| 9008 /* Write si_sugtime to the file. */ | |
|
2229
d45902a5c61c
Fix a few more things for persistent undo.
Bram Moolenaar <bram@vim.org>
parents:
2214
diff
changeset
|
9009 put_time(fd, spin->si_sugtime); /* <timestamp> */ |
| 625 | 9010 |
| 9011 /* | |
| 9012 * <SUGWORDTREE> | |
| 9013 */ | |
| 9014 spin->si_memtot = 0; | |
| 9015 tree = spin->si_foldroot->wn_sibling; | |
| 9016 | |
| 9017 /* Clear the index and wnode fields in the tree. */ | |
| 9018 clear_node(tree); | |
| 9019 | |
| 9020 /* Count the number of nodes. Needed to be able to allocate the | |
| 9021 * memory when reading the nodes. Also fills in index for shared | |
| 9022 * nodes. */ | |
| 9023 nodecount = put_node(NULL, tree, 0, 0, FALSE); | |
| 9024 | |
| 9025 /* number of nodes in 4 bytes */ | |
| 9026 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */ | |
| 9027 spin->si_memtot += nodecount + nodecount * sizeof(int); | |
| 9028 | |
| 9029 /* Write the nodes. */ | |
| 9030 (void)put_node(fd, tree, 0, 0, FALSE); | |
| 9031 | |
| 9032 /* | |
| 9033 * <SUGTABLE>: <sugwcount> <sugline> ... | |
| 9034 */ | |
| 9035 wcount = spin->si_spellbuf->b_ml.ml_line_count; | |
| 9036 put_bytes(fd, (long_u)wcount, 4); /* <sugwcount> */ | |
| 9037 | |
| 9038 for (lnum = 1; lnum <= (linenr_T)wcount; ++lnum) | |
| 9039 { | |
| 9040 /* <sugline>: <sugnr> ... NUL */ | |
| 9041 line = ml_get_buf(spin->si_spellbuf, lnum, FALSE); | |
| 835 | 9042 len = (int)STRLEN(line) + 1; |
| 625 | 9043 if (fwrite(line, (size_t)len, (size_t)1, fd) == 0) |
| 9044 { | |
| 9045 EMSG(_(e_write)); | |
| 9046 goto theend; | |
| 9047 } | |
| 9048 spin->si_memtot += len; | |
| 9049 } | |
| 9050 | |
| 9051 /* Write another byte to check for errors. */ | |
| 9052 if (putc(0, fd) == EOF) | |
| 9053 EMSG(_(e_write)); | |
| 9054 | |
| 9055 vim_snprintf((char *)IObuff, IOSIZE, | |
| 9056 _("Estimated runtime memory use: %d bytes"), spin->si_memtot); | |
| 9057 spell_message(spin, IObuff); | |
| 9058 | |
| 9059 theend: | |
| 9060 /* close the file */ | |
| 9061 fclose(fd); | |
| 9062 } | |
| 9063 | |
| 9064 /* | |
| 9065 * Open a spell buffer. This is a nameless buffer that is not in the buffer | |
| 9066 * list and only contains text lines. Can use a swapfile to reduce memory | |
| 9067 * use. | |
| 9068 * Most other fields are invalid! Esp. watch out for string options being | |
| 9069 * NULL and there is no undo info. | |
| 9070 * Returns NULL when out of memory. | |
| 9071 */ | |
| 9072 static buf_T * | |
| 9073 open_spellbuf() | |
| 9074 { | |
| 9075 buf_T *buf; | |
| 9076 | |
| 9077 buf = (buf_T *)alloc_clear(sizeof(buf_T)); | |
| 9078 if (buf != NULL) | |
| 9079 { | |
| 9080 buf->b_spell = TRUE; | |
| 9081 buf->b_p_swf = TRUE; /* may create a swap file */ | |
|
5204
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
9082 #ifdef FEAT_CRYPT |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
9083 buf->b_p_key = empty_option; |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
9084 #endif |
| 625 | 9085 ml_open(buf); |
| 9086 ml_open_file(buf); /* create swap file now */ | |
| 9087 } | |
| 9088 return buf; | |
| 9089 } | |
| 9090 | |
| 9091 /* | |
| 9092 * Close the buffer used for spell info. | |
| 9093 */ | |
| 9094 static void | |
| 9095 close_spellbuf(buf) | |
| 9096 buf_T *buf; | |
| 9097 { | |
| 9098 if (buf != NULL) | |
| 9099 { | |
| 9100 ml_close(buf, TRUE); | |
| 9101 vim_free(buf); | |
| 9102 } | |
| 9103 } | |
| 9104 | |
| 9105 | |
| 9106 /* | |
| 310 | 9107 * Create a Vim spell file from one or more word lists. |
| 9108 * "fnames[0]" is the output file name. | |
| 9109 * "fnames[fcount - 1]" is the last input file name. | |
| 9110 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name | |
| 9111 * and ".spl" is appended to make the output file name. | |
| 9112 */ | |
| 9113 static void | |
| 3263 | 9114 mkspell(fcount, fnames, ascii, over_write, added_word) |
| 310 | 9115 int fcount; |
| 9116 char_u **fnames; | |
| 9117 int ascii; /* -ascii argument given */ | |
| 3263 | 9118 int over_write; /* overwrite existing output file */ |
| 323 | 9119 int added_word; /* invoked through "zg" */ |
| 310 | 9120 { |
| 2770 | 9121 char_u *fname = NULL; |
| 9122 char_u *wfname; | |
| 310 | 9123 char_u **innames; |
| 9124 int incount; | |
| 236 | 9125 afffile_T *(afile[8]); |
| 9126 int i; | |
| 9127 int len; | |
| 9128 struct stat st; | |
| 255 | 9129 int error = FALSE; |
| 300 | 9130 spellinfo_T spin; |
| 9131 | |
| 9132 vim_memset(&spin, 0, sizeof(spin)); | |
| 323 | 9133 spin.si_verbose = !added_word; |
| 310 | 9134 spin.si_ascii = ascii; |
| 323 | 9135 spin.si_followup = TRUE; |
| 9136 spin.si_rem_accents = TRUE; | |
| 9137 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20); | |
| 625 | 9138 ga_init2(&spin.si_repsal, (int)sizeof(fromto_T), 20); |
| 323 | 9139 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20); |
| 9140 ga_init2(&spin.si_map, (int)sizeof(char_u), 100); | |
| 809 | 9141 ga_init2(&spin.si_comppat, (int)sizeof(char_u *), 20); |
| 339 | 9142 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50); |
| 625 | 9143 hash_init(&spin.si_commonwords); |
| 500 | 9144 spin.si_newcompID = 127; /* start compound ID at first maximum */ |
| 310 | 9145 |
| 9146 /* default: fnames[0] is output file, following are input files */ | |
| 9147 innames = &fnames[1]; | |
| 9148 incount = fcount - 1; | |
| 9149 | |
| 2770 | 9150 wfname = alloc(MAXPATHL); |
| 9151 if (wfname == NULL) | |
| 9152 return; | |
| 9153 | |
| 310 | 9154 if (fcount >= 1) |
| 240 | 9155 { |
| 835 | 9156 len = (int)STRLEN(fnames[0]); |
| 310 | 9157 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0) |
| 9158 { | |
| 9159 /* For ":mkspell path/en.latin1.add" output file is | |
| 9160 * "path/en.latin1.add.spl". */ | |
| 9161 innames = &fnames[0]; | |
| 9162 incount = 1; | |
| 2770 | 9163 vim_snprintf((char *)wfname, MAXPATHL, "%s.spl", fnames[0]); |
| 310 | 9164 } |
| 366 | 9165 else if (fcount == 1) |
| 9166 { | |
| 9167 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */ | |
| 9168 innames = &fnames[0]; | |
| 9169 incount = 1; | |
| 2770 | 9170 vim_snprintf((char *)wfname, MAXPATHL, SPL_FNAME_TMPL, |
| 2660 | 9171 fnames[0], spin.si_ascii ? (char_u *)"ascii" : spell_enc()); |
| 366 | 9172 } |
| 310 | 9173 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0) |
| 9174 { | |
| 9175 /* Name ends in ".spl", use as the file name. */ | |
| 2770 | 9176 vim_strncpy(wfname, fnames[0], MAXPATHL - 1); |
| 310 | 9177 } |
| 9178 else | |
| 9179 /* Name should be language, make the file name from it. */ | |
| 2770 | 9180 vim_snprintf((char *)wfname, MAXPATHL, SPL_FNAME_TMPL, |
| 2660 | 9181 fnames[0], spin.si_ascii ? (char_u *)"ascii" : spell_enc()); |
| 310 | 9182 |
| 9183 /* Check for .ascii.spl. */ | |
| 2660 | 9184 if (strstr((char *)gettail(wfname), SPL_FNAME_ASCII) != NULL) |
| 310 | 9185 spin.si_ascii = TRUE; |
| 9186 | |
| 9187 /* Check for .add.spl. */ | |
| 2660 | 9188 if (strstr((char *)gettail(wfname), SPL_FNAME_ADD) != NULL) |
| 310 | 9189 spin.si_add = TRUE; |
| 240 | 9190 } |
| 9191 | |
| 310 | 9192 if (incount <= 0) |
| 236 | 9193 EMSG(_(e_invarg)); /* need at least output and input names */ |
| 351 | 9194 else if (vim_strchr(gettail(wfname), '_') != NULL) |
| 9195 EMSG(_("E751: Output file name must not have region name")); | |
| 310 | 9196 else if (incount > 8) |
| 236 | 9197 EMSG(_("E754: Only up to 8 regions supported")); |
| 9198 else | |
| 9199 { | |
| 9200 /* Check for overwriting before doing things that may take a lot of | |
| 9201 * time. */ | |
| 3263 | 9202 if (!over_write && mch_stat((char *)wfname, &st) >= 0) |
| 236 | 9203 { |
| 9204 EMSG(_(e_exists)); | |
| 2770 | 9205 goto theend; |
| 236 | 9206 } |
| 310 | 9207 if (mch_isdir(wfname)) |
| 236 | 9208 { |
| 310 | 9209 EMSG2(_(e_isadir2), wfname); |
| 2770 | 9210 goto theend; |
| 9211 } | |
| 9212 | |
| 9213 fname = alloc(MAXPATHL); | |
| 9214 if (fname == NULL) | |
| 9215 goto theend; | |
| 236 | 9216 |
| 9217 /* | |
| 9218 * Init the aff and dic pointers. | |
| 9219 * Get the region names if there are more than 2 arguments. | |
| 9220 */ | |
| 310 | 9221 for (i = 0; i < incount; ++i) |
| 236 | 9222 { |
| 310 | 9223 afile[i] = NULL; |
| 300 | 9224 |
| 316 | 9225 if (incount > 1) |
| 236 | 9226 { |
| 835 | 9227 len = (int)STRLEN(innames[i]); |
| 310 | 9228 if (STRLEN(gettail(innames[i])) < 5 |
| 9229 || innames[i][len - 3] != '_') | |
| 236 | 9230 { |
| 310 | 9231 EMSG2(_("E755: Invalid region in %s"), innames[i]); |
| 2770 | 9232 goto theend; |
| 236 | 9233 } |
| 316 | 9234 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]); |
| 9235 spin.si_region_name[i * 2 + 1] = | |
| 9236 TOLOWER_ASC(innames[i][len - 1]); | |
| 236 | 9237 } |
| 9238 } | |
| 316 | 9239 spin.si_region_count = incount; |
| 236 | 9240 |
| 474 | 9241 spin.si_foldroot = wordtree_alloc(&spin); |
| 9242 spin.si_keeproot = wordtree_alloc(&spin); | |
| 9243 spin.si_prefroot = wordtree_alloc(&spin); | |
| 339 | 9244 if (spin.si_foldroot == NULL |
| 9245 || spin.si_keeproot == NULL | |
| 9246 || spin.si_prefroot == NULL) | |
| 300 | 9247 { |
| 469 | 9248 free_blocks(spin.si_blocks); |
| 2770 | 9249 goto theend; |
| 300 | 9250 } |
| 9251 | |
| 351 | 9252 /* When not producing a .add.spl file clear the character table when |
| 9253 * we encounter one in the .aff file. This means we dump the current | |
| 9254 * one in the .spl file if the .aff file doesn't define one. That's | |
| 9255 * better than guessing the contents, the table will match a | |
| 9256 * previously loaded spell file. */ | |
| 9257 if (!spin.si_add) | |
| 9258 spin.si_clear_chartab = TRUE; | |
| 9259 | |
| 236 | 9260 /* |
| 9261 * Read all the .aff and .dic files. | |
| 9262 * Text is converted to 'encoding'. | |
| 300 | 9263 * Words are stored in the case-folded and keep-case trees. |
| 236 | 9264 */ |
| 310 | 9265 for (i = 0; i < incount && !error; ++i) |
| 236 | 9266 { |
| 300 | 9267 spin.si_conv.vc_type = CONV_NONE; |
| 310 | 9268 spin.si_region = 1 << i; |
| 9269 | |
| 2770 | 9270 vim_snprintf((char *)fname, MAXPATHL, "%s.aff", innames[i]); |
| 300 | 9271 if (mch_stat((char *)fname, &st) >= 0) |
| 9272 { | |
| 9273 /* Read the .aff file. Will init "spin->si_conv" based on the | |
| 9274 * "SET" line. */ | |
| 474 | 9275 afile[i] = spell_read_aff(&spin, fname); |
| 310 | 9276 if (afile[i] == NULL) |
| 300 | 9277 error = TRUE; |
| 9278 else | |
| 9279 { | |
| 9280 /* Read the .dic file and store the words in the trees. */ | |
| 2770 | 9281 vim_snprintf((char *)fname, MAXPATHL, "%s.dic", |
| 310 | 9282 innames[i]); |
| 474 | 9283 if (spell_read_dic(&spin, fname, afile[i]) == FAIL) |
| 300 | 9284 error = TRUE; |
| 9285 } | |
| 9286 } | |
| 9287 else | |
| 9288 { | |
| 9289 /* No .aff file, try reading the file as a word list. Store | |
| 9290 * the words in the trees. */ | |
| 474 | 9291 if (spell_read_wordfile(&spin, innames[i]) == FAIL) |
| 300 | 9292 error = TRUE; |
| 9293 } | |
| 236 | 9294 |
| 310 | 9295 #ifdef FEAT_MBYTE |
| 236 | 9296 /* Free any conversion stuff. */ |
| 300 | 9297 convert_setup(&spin.si_conv, NULL, NULL); |
| 310 | 9298 #endif |
| 236 | 9299 } |
| 9300 | |
| 492 | 9301 if (spin.si_compflags != NULL && spin.si_nobreak) |
| 9302 MSG(_("Warning: both compounding and NOBREAK specified")); | |
| 9303 | |
| 625 | 9304 if (!error && !got_int) |
| 236 | 9305 { |
| 9306 /* | |
| 300 | 9307 * Combine tails in the tree. |
| 236 | 9308 */ |
| 625 | 9309 spell_message(&spin, (char_u *)_(msg_compressing)); |
| 474 | 9310 wordtree_compress(&spin, spin.si_foldroot); |
| 9311 wordtree_compress(&spin, spin.si_keeproot); | |
| 9312 wordtree_compress(&spin, spin.si_prefroot); | |
| 236 | 9313 } |
| 9314 | |
| 625 | 9315 if (!error && !got_int) |
| 300 | 9316 { |
| 9317 /* | |
| 9318 * Write the info in the spell file. | |
| 9319 */ | |
| 625 | 9320 vim_snprintf((char *)IObuff, IOSIZE, |
| 9321 _("Writing spell file %s ..."), wfname); | |
| 9322 spell_message(&spin, IObuff); | |
| 310 | 9323 |
| 500 | 9324 error = write_vim_spell(&spin, wfname) == FAIL; |
| 474 | 9325 |
| 625 | 9326 spell_message(&spin, (char_u *)_("Done!")); |
| 9327 vim_snprintf((char *)IObuff, IOSIZE, | |
| 9328 _("Estimated runtime memory use: %d bytes"), spin.si_memtot); | |
| 9329 spell_message(&spin, IObuff); | |
| 9330 | |
| 9331 /* | |
| 9332 * If the file is loaded need to reload it. | |
| 9333 */ | |
| 500 | 9334 if (!error) |
| 9335 spell_reload_one(wfname, added_word); | |
| 300 | 9336 } |
| 9337 | |
| 9338 /* Free the allocated memory. */ | |
| 323 | 9339 ga_clear(&spin.si_rep); |
| 625 | 9340 ga_clear(&spin.si_repsal); |
| 323 | 9341 ga_clear(&spin.si_sal); |
| 9342 ga_clear(&spin.si_map); | |
| 809 | 9343 ga_clear(&spin.si_comppat); |
| 339 | 9344 ga_clear(&spin.si_prefcond); |
| 625 | 9345 hash_clear_all(&spin.si_commonwords, 0); |
| 300 | 9346 |
| 9347 /* Free the .aff file structures. */ | |
| 310 | 9348 for (i = 0; i < incount; ++i) |
| 9349 if (afile[i] != NULL) | |
| 9350 spell_free_aff(afile[i]); | |
| 339 | 9351 |
| 9352 /* Free all the bits and pieces at once. */ | |
| 9353 free_blocks(spin.si_blocks); | |
| 625 | 9354 |
| 9355 /* | |
| 9356 * If there is soundfolding info and no NOSUGFILE item create the | |
| 9357 * .sug file with the soundfolded word trie. | |
| 9358 */ | |
| 9359 if (spin.si_sugtime != 0 && !error && !got_int) | |
| 9360 spell_make_sugfile(&spin, wfname); | |
| 9361 | |
| 9362 } | |
| 2770 | 9363 |
| 9364 theend: | |
| 9365 vim_free(fname); | |
| 9366 vim_free(wfname); | |
| 625 | 9367 } |
| 9368 | |
| 9369 /* | |
| 9370 * Display a message for spell file processing when 'verbose' is set or using | |
| 9371 * ":mkspell". "str" can be IObuff. | |
| 9372 */ | |
| 9373 static void | |
| 9374 spell_message(spin, str) | |
| 9375 spellinfo_T *spin; | |
| 9376 char_u *str; | |
| 9377 { | |
| 9378 if (spin->si_verbose || p_verbose > 2) | |
| 9379 { | |
| 9380 if (!spin->si_verbose) | |
| 9381 verbose_enter(); | |
| 9382 MSG(str); | |
| 9383 out_flush(); | |
| 9384 if (!spin->si_verbose) | |
| 9385 verbose_leave(); | |
| 9386 } | |
| 9387 } | |
| 310 | 9388 |
| 9389 /* | |
| 385 | 9390 * ":[count]spellgood {word}" |
| 9391 * ":[count]spellwrong {word}" | |
| 710 | 9392 * ":[count]spellundo {word}" |
| 310 | 9393 */ |
| 9394 void | |
| 9395 ex_spell(eap) | |
| 9396 exarg_T *eap; | |
| 9397 { | |
| 835 | 9398 spell_add_word(eap->arg, (int)STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong, |
| 710 | 9399 eap->forceit ? 0 : (int)eap->line2, |
| 9400 eap->cmdidx == CMD_spellundo); | |
| 236 | 9401 } |
| 9402 | |
| 310 | 9403 /* |
| 9404 * Add "word[len]" to 'spellfile' as a good or bad word. | |
| 9405 */ | |
| 9406 void | |
| 944 | 9407 spell_add_word(word, len, bad, idx, undo) |
| 310 | 9408 char_u *word; |
| 9409 int len; | |
| 9410 int bad; | |
| 944 | 9411 int idx; /* "zG" and "zW": zero, otherwise index in |
| 385 | 9412 'spellfile' */ |
| 710 | 9413 int undo; /* TRUE for "zug", "zuG", "zuw" and "zuW" */ |
| 310 | 9414 { |
| 977 | 9415 FILE *fd = NULL; |
| 385 | 9416 buf_T *buf = NULL; |
| 351 | 9417 int new_spf = FALSE; |
| 381 | 9418 char_u *fname; |
| 2770 | 9419 char_u *fnamebuf = NULL; |
| 385 | 9420 char_u line[MAXWLEN * 2]; |
| 9421 long fpos, fpos_next = 0; | |
| 9422 int i; | |
| 9423 char_u *spf; | |
| 9424 | |
| 944 | 9425 if (idx == 0) /* use internal wordlist */ |
| 385 | 9426 { |
| 9427 if (int_wordlist == NULL) | |
| 9428 { | |
| 6721 | 9429 int_wordlist = vim_tempname('s', FALSE); |
| 385 | 9430 if (int_wordlist == NULL) |
| 381 | 9431 return; |
| 9432 } | |
| 385 | 9433 fname = int_wordlist; |
| 381 | 9434 } |
| 310 | 9435 else |
| 9436 { | |
| 381 | 9437 /* If 'spellfile' isn't set figure out a good default value. */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9438 if (*curwin->w_s->b_p_spf == NUL) |
| 381 | 9439 { |
| 9440 init_spellfile(); | |
| 9441 new_spf = TRUE; | |
| 9442 } | |
| 9443 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9444 if (*curwin->w_s->b_p_spf == NUL) |
| 381 | 9445 { |
| 523 | 9446 EMSG2(_(e_notset), "spellfile"); |
| 381 | 9447 return; |
| 9448 } | |
| 2770 | 9449 fnamebuf = alloc(MAXPATHL); |
| 9450 if (fnamebuf == NULL) | |
| 9451 return; | |
| 381 | 9452 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9453 for (spf = curwin->w_s->b_p_spf, i = 1; *spf != NUL; ++i) |
| 385 | 9454 { |
| 9455 copy_option_part(&spf, fnamebuf, MAXPATHL, ","); | |
| 944 | 9456 if (i == idx) |
| 385 | 9457 break; |
| 9458 if (*spf == NUL) | |
| 9459 { | |
| 944 | 9460 EMSGN(_("E765: 'spellfile' does not have %ld entries"), idx); |
| 2770 | 9461 vim_free(fnamebuf); |
| 385 | 9462 return; |
| 9463 } | |
| 9464 } | |
| 9465 | |
| 310 | 9466 /* Check that the user isn't editing the .add file somewhere. */ |
| 385 | 9467 buf = buflist_findname_exp(fnamebuf); |
| 310 | 9468 if (buf != NULL && buf->b_ml.ml_mfp == NULL) |
| 9469 buf = NULL; | |
| 9470 if (buf != NULL && bufIsChanged(buf)) | |
| 381 | 9471 { |
| 310 | 9472 EMSG(_(e_bufloaded)); |
| 2770 | 9473 vim_free(fnamebuf); |
| 381 | 9474 return; |
| 9475 } | |
| 9476 | |
| 385 | 9477 fname = fnamebuf; |
| 9478 } | |
| 9479 | |
| 710 | 9480 if (bad || undo) |
| 9481 { | |
| 9482 /* When the word appears as good word we need to remove that one, | |
| 385 | 9483 * since its flags sort before the one with WF_BANNED. */ |
| 9484 fd = mch_fopen((char *)fname, "r"); | |
| 9485 if (fd != NULL) | |
| 9486 { | |
| 9487 while (!vim_fgets(line, MAXWLEN * 2, fd)) | |
| 9488 { | |
| 9489 fpos = fpos_next; | |
| 9490 fpos_next = ftell(fd); | |
| 9491 if (STRNCMP(word, line, len) == 0 | |
| 9492 && (line[len] == '/' || line[len] < ' ')) | |
| 9493 { | |
| 9494 /* Found duplicate word. Remove it by writing a '#' at | |
| 9495 * the start of the line. Mixing reading and writing | |
| 9496 * doesn't work for all systems, close the file first. */ | |
| 9497 fclose(fd); | |
| 9498 fd = mch_fopen((char *)fname, "r+"); | |
| 9499 if (fd == NULL) | |
| 9500 break; | |
| 9501 if (fseek(fd, fpos, SEEK_SET) == 0) | |
| 710 | 9502 { |
| 385 | 9503 fputc('#', fd); |
| 710 | 9504 if (undo) |
| 974 | 9505 { |
| 9506 home_replace(NULL, fname, NameBuff, MAXPATHL, TRUE); | |
| 5374 | 9507 smsg((char_u *)_("Word '%.*s' removed from %s"), |
| 9508 len, word, NameBuff); | |
| 974 | 9509 } |
| 710 | 9510 } |
| 385 | 9511 fseek(fd, fpos_next, SEEK_SET); |
| 9512 } | |
| 9513 } | |
|
2445
04dae202d316
Fixes for coverity warnings.
Bram Moolenaar <bram@vim.org>
parents:
2299
diff
changeset
|
9514 if (fd != NULL) |
|
04dae202d316
Fixes for coverity warnings.
Bram Moolenaar <bram@vim.org>
parents:
2299
diff
changeset
|
9515 fclose(fd); |
| 385 | 9516 } |
| 381 | 9517 } |
| 975 | 9518 |
| 9519 if (!undo) | |
| 710 | 9520 { |
| 9521 fd = mch_fopen((char *)fname, "a"); | |
| 9522 if (fd == NULL && new_spf) | |
| 9523 { | |
| 975 | 9524 char_u *p; |
| 9525 | |
| 710 | 9526 /* We just initialized the 'spellfile' option and can't open the |
| 9527 * file. We may need to create the "spell" directory first. We | |
| 9528 * already checked the runtime directory is writable in | |
| 9529 * init_spellfile(). */ | |
| 975 | 9530 if (!dir_of_file_exists(fname) && (p = gettail_sep(fname)) != fname) |
| 9531 { | |
| 9532 int c = *p; | |
| 9533 | |
| 710 | 9534 /* The directory doesn't exist. Try creating it and opening |
| 9535 * the file again. */ | |
| 975 | 9536 *p = NUL; |
| 9537 vim_mkdir(fname, 0755); | |
| 9538 *p = c; | |
| 710 | 9539 fd = mch_fopen((char *)fname, "a"); |
| 9540 } | |
| 9541 } | |
| 9542 | |
| 9543 if (fd == NULL) | |
| 9544 EMSG2(_(e_notopen), fname); | |
| 310 | 9545 else |
| 710 | 9546 { |
| 9547 if (bad) | |
| 9548 fprintf(fd, "%.*s/!\n", len, word); | |
| 9549 else | |
| 9550 fprintf(fd, "%.*s\n", len, word); | |
| 9551 fclose(fd); | |
| 9552 | |
| 9553 home_replace(NULL, fname, NameBuff, MAXPATHL, TRUE); | |
| 5374 | 9554 smsg((char_u *)_("Word '%.*s' added to %s"), len, word, NameBuff); |
| 710 | 9555 } |
| 9556 } | |
| 9557 | |
| 9558 if (fd != NULL) | |
| 9559 { | |
| 381 | 9560 /* Update the .add.spl file. */ |
| 9561 mkspell(1, &fname, FALSE, TRUE, TRUE); | |
| 9562 | |
| 9563 /* If the .add file is edited somewhere, reload it. */ | |
| 9564 if (buf != NULL) | |
| 627 | 9565 buf_reload(buf, buf->b_orig_mode); |
| 381 | 9566 |
| 737 | 9567 redraw_all_later(SOME_VALID); |
| 310 | 9568 } |
| 2770 | 9569 vim_free(fnamebuf); |
| 310 | 9570 } |
| 9571 | |
| 9572 /* | |
| 9573 * Initialize 'spellfile' for the current buffer. | |
| 9574 */ | |
| 9575 static void | |
| 9576 init_spellfile() | |
| 9577 { | |
| 2770 | 9578 char_u *buf; |
| 310 | 9579 int l; |
| 500 | 9580 char_u *fname; |
| 310 | 9581 char_u *rtp; |
| 351 | 9582 char_u *lend; |
| 501 | 9583 int aspath = FALSE; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9584 char_u *lstart = curbuf->b_s.b_p_spl; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9585 |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9586 if (*curwin->w_s->b_p_spl != NUL && curwin->w_s->b_langp.ga_len > 0) |
| 310 | 9587 { |
| 2770 | 9588 buf = alloc(MAXPATHL); |
| 9589 if (buf == NULL) | |
| 9590 return; | |
| 9591 | |
| 501 | 9592 /* Find the end of the language name. Exclude the region. If there |
| 9593 * is a path separator remember the start of the tail. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9594 for (lend = curwin->w_s->b_p_spl; *lend != NUL |
| 351 | 9595 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend) |
| 501 | 9596 if (vim_ispathsep(*lend)) |
| 9597 { | |
| 9598 aspath = TRUE; | |
| 9599 lstart = lend + 1; | |
| 9600 } | |
| 351 | 9601 |
| 9602 /* Loop over all entries in 'runtimepath'. Use the first one where we | |
| 9603 * are allowed to write. */ | |
| 310 | 9604 rtp = p_rtp; |
| 9605 while (*rtp != NUL) | |
| 9606 { | |
| 501 | 9607 if (aspath) |
| 9608 /* Use directory of an entry with path, e.g., for | |
| 9609 * "/dir/lg.utf-8.spl" use "/dir". */ | |
| 2593 | 9610 vim_strncpy(buf, curbuf->b_s.b_p_spl, |
| 9611 lstart - curbuf->b_s.b_p_spl - 1); | |
| 501 | 9612 else |
| 9613 /* Copy the path from 'runtimepath' to buf[]. */ | |
| 9614 copy_option_part(&rtp, buf, MAXPATHL, ","); | |
| 310 | 9615 if (filewritable(buf) == 2) |
| 9616 { | |
| 316 | 9617 /* Use the first language name from 'spelllang' and the |
| 9618 * encoding used in the first loaded .spl file. */ | |
| 501 | 9619 if (aspath) |
| 2593 | 9620 vim_strncpy(buf, curbuf->b_s.b_p_spl, |
| 9621 lend - curbuf->b_s.b_p_spl); | |
| 501 | 9622 else |
| 9623 { | |
| 819 | 9624 /* Create the "spell" directory if it doesn't exist yet. */ |
| 835 | 9625 l = (int)STRLEN(buf); |
| 819 | 9626 vim_snprintf((char *)buf + l, MAXPATHL - l, "/spell"); |
| 2593 | 9627 if (filewritable(buf) != 2) |
| 819 | 9628 vim_mkdir(buf, 0755); |
| 9629 | |
| 835 | 9630 l = (int)STRLEN(buf); |
| 501 | 9631 vim_snprintf((char *)buf + l, MAXPATHL - l, |
| 819 | 9632 "/%.*s", (int)(lend - lstart), lstart); |
| 501 | 9633 } |
| 835 | 9634 l = (int)STRLEN(buf); |
| 2770 | 9635 fname = LANGP_ENTRY(curwin->w_s->b_langp, 0) |
| 9636 ->lp_slang->sl_fname; | |
| 501 | 9637 vim_snprintf((char *)buf + l, MAXPATHL - l, ".%s.add", |
| 9638 fname != NULL | |
| 9639 && strstr((char *)gettail(fname), ".ascii.") != NULL | |
| 9640 ? (char_u *)"ascii" : spell_enc()); | |
| 310 | 9641 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL); |
| 9642 break; | |
| 9643 } | |
| 501 | 9644 aspath = FALSE; |
| 310 | 9645 } |
| 2770 | 9646 |
| 9647 vim_free(buf); | |
| 310 | 9648 } |
| 9649 } | |
| 236 | 9650 |
| 300 | 9651 |
| 307 | 9652 /* |
| 9653 * Init the chartab used for spelling for ASCII. | |
| 9654 * EBCDIC is not supported! | |
| 9655 */ | |
| 9656 static void | |
| 9657 clear_spell_chartab(sp) | |
| 9658 spelltab_T *sp; | |
| 9659 { | |
| 324 | 9660 int i; |
| 307 | 9661 |
| 9662 /* Init everything to FALSE. */ | |
| 9663 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw)); | |
| 9664 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu)); | |
| 9665 for (i = 0; i < 256; ++i) | |
| 324 | 9666 { |
| 307 | 9667 sp->st_fold[i] = i; |
| 324 | 9668 sp->st_upper[i] = i; |
| 9669 } | |
| 307 | 9670 |
| 9671 /* We include digits. A word shouldn't start with a digit, but handling | |
| 9672 * that is done separately. */ | |
| 9673 for (i = '0'; i <= '9'; ++i) | |
| 9674 sp->st_isw[i] = TRUE; | |
| 9675 for (i = 'A'; i <= 'Z'; ++i) | |
| 9676 { | |
| 9677 sp->st_isw[i] = TRUE; | |
| 9678 sp->st_isu[i] = TRUE; | |
| 9679 sp->st_fold[i] = i + 0x20; | |
| 9680 } | |
| 9681 for (i = 'a'; i <= 'z'; ++i) | |
| 324 | 9682 { |
| 307 | 9683 sp->st_isw[i] = TRUE; |
| 324 | 9684 sp->st_upper[i] = i - 0x20; |
| 9685 } | |
| 307 | 9686 } |
| 9687 | |
| 9688 /* | |
| 9689 * Init the chartab used for spelling. Only depends on 'encoding'. | |
| 9690 * Called once while starting up and when 'encoding' changes. | |
| 9691 * The default is to use isalpha(), but the spell file should define the word | |
| 9692 * characters to make it possible that 'encoding' differs from the current | |
| 390 | 9693 * locale. For utf-8 we don't use isalpha() but our own functions. |
| 307 | 9694 */ |
| 9695 void | |
| 9696 init_spell_chartab() | |
| 9697 { | |
| 9698 int i; | |
| 9699 | |
| 9700 did_set_spelltab = FALSE; | |
| 9701 clear_spell_chartab(&spelltab); | |
| 9702 #ifdef FEAT_MBYTE | |
| 9703 if (enc_dbcs) | |
| 9704 { | |
| 9705 /* DBCS: assume double-wide characters are word characters. */ | |
| 9706 for (i = 128; i <= 255; ++i) | |
| 9707 if (MB_BYTE2LEN(i) == 2) | |
| 9708 spelltab.st_isw[i] = TRUE; | |
| 9709 } | |
| 324 | 9710 else if (enc_utf8) |
| 9711 { | |
| 9712 for (i = 128; i < 256; ++i) | |
| 9713 { | |
|
2140
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9714 int f = utf_fold(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9715 int u = utf_toupper(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9716 |
| 324 | 9717 spelltab.st_isu[i] = utf_isupper(i); |
| 9718 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i); | |
|
2140
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9719 /* The folded/upper-cased value is different between latin1 and |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9720 * utf8 for 0xb5, causing E763 for no good reason. Use the latin1 |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9721 * value for utf-8 to avoid this. */ |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9722 spelltab.st_fold[i] = (f < 256) ? f : i; |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9723 spelltab.st_upper[i] = (u < 256) ? u : i; |
| 324 | 9724 } |
| 9725 } | |
| 307 | 9726 else |
| 9727 #endif | |
| 9728 { | |
| 324 | 9729 /* Rough guess: use locale-dependent library functions. */ |
| 307 | 9730 for (i = 128; i < 256; ++i) |
| 9731 { | |
| 9732 if (MB_ISUPPER(i)) | |
| 9733 { | |
| 324 | 9734 spelltab.st_isw[i] = TRUE; |
| 307 | 9735 spelltab.st_isu[i] = TRUE; |
| 9736 spelltab.st_fold[i] = MB_TOLOWER(i); | |
| 9737 } | |
| 324 | 9738 else if (MB_ISLOWER(i)) |
| 9739 { | |
| 9740 spelltab.st_isw[i] = TRUE; | |
| 9741 spelltab.st_upper[i] = MB_TOUPPER(i); | |
| 9742 } | |
| 307 | 9743 } |
| 9744 } | |
| 9745 } | |
| 9746 | |
| 9747 /* | |
| 9748 * Set the spell character tables from strings in the affix file. | |
| 9749 */ | |
| 9750 static int | |
| 9751 set_spell_chartab(fol, low, upp) | |
| 9752 char_u *fol; | |
| 9753 char_u *low; | |
| 9754 char_u *upp; | |
| 9755 { | |
| 9756 /* We build the new tables here first, so that we can compare with the | |
| 9757 * previous one. */ | |
| 9758 spelltab_T new_st; | |
| 9759 char_u *pf = fol, *pl = low, *pu = upp; | |
| 9760 int f, l, u; | |
| 9761 | |
| 9762 clear_spell_chartab(&new_st); | |
| 9763 | |
| 9764 while (*pf != NUL) | |
| 9765 { | |
| 9766 if (*pl == NUL || *pu == NUL) | |
| 9767 { | |
| 9768 EMSG(_(e_affform)); | |
| 9769 return FAIL; | |
| 9770 } | |
| 9771 #ifdef FEAT_MBYTE | |
| 9772 f = mb_ptr2char_adv(&pf); | |
| 9773 l = mb_ptr2char_adv(&pl); | |
| 9774 u = mb_ptr2char_adv(&pu); | |
| 9775 #else | |
| 9776 f = *pf++; | |
| 9777 l = *pl++; | |
| 9778 u = *pu++; | |
| 9779 #endif | |
| 9780 /* Every character that appears is a word character. */ | |
| 9781 if (f < 256) | |
| 9782 new_st.st_isw[f] = TRUE; | |
| 9783 if (l < 256) | |
| 9784 new_st.st_isw[l] = TRUE; | |
| 9785 if (u < 256) | |
| 9786 new_st.st_isw[u] = TRUE; | |
| 9787 | |
| 9788 /* if "LOW" and "FOL" are not the same the "LOW" char needs | |
| 9789 * case-folding */ | |
| 9790 if (l < 256 && l != f) | |
| 9791 { | |
| 9792 if (f >= 256) | |
| 9793 { | |
| 9794 EMSG(_(e_affrange)); | |
| 9795 return FAIL; | |
| 9796 } | |
| 9797 new_st.st_fold[l] = f; | |
| 9798 } | |
| 9799 | |
| 9800 /* if "UPP" and "FOL" are not the same the "UPP" char needs | |
| 324 | 9801 * case-folding, it's upper case and the "UPP" is the upper case of |
| 9802 * "FOL" . */ | |
| 307 | 9803 if (u < 256 && u != f) |
| 9804 { | |
| 9805 if (f >= 256) | |
| 9806 { | |
| 9807 EMSG(_(e_affrange)); | |
| 9808 return FAIL; | |
| 9809 } | |
| 9810 new_st.st_fold[u] = f; | |
| 9811 new_st.st_isu[u] = TRUE; | |
| 324 | 9812 new_st.st_upper[f] = u; |
| 307 | 9813 } |
| 9814 } | |
| 9815 | |
| 9816 if (*pl != NUL || *pu != NUL) | |
| 9817 { | |
| 9818 EMSG(_(e_affform)); | |
| 9819 return FAIL; | |
| 9820 } | |
| 9821 | |
| 9822 return set_spell_finish(&new_st); | |
| 9823 } | |
| 9824 | |
| 9825 /* | |
| 9826 * Set the spell character tables from strings in the .spl file. | |
| 9827 */ | |
| 483 | 9828 static void |
| 389 | 9829 set_spell_charflags(flags, cnt, fol) |
| 307 | 9830 char_u *flags; |
| 389 | 9831 int cnt; /* length of "flags" */ |
| 9832 char_u *fol; | |
| 307 | 9833 { |
| 9834 /* We build the new tables here first, so that we can compare with the | |
| 9835 * previous one. */ | |
| 9836 spelltab_T new_st; | |
| 9837 int i; | |
| 389 | 9838 char_u *p = fol; |
| 324 | 9839 int c; |
| 307 | 9840 |
| 9841 clear_spell_chartab(&new_st); | |
| 9842 | |
| 389 | 9843 for (i = 0; i < 128; ++i) |
| 9844 { | |
| 9845 if (i < cnt) | |
| 9846 { | |
| 9847 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0; | |
| 9848 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0; | |
| 9849 } | |
| 9850 | |
| 9851 if (*p != NUL) | |
| 9852 { | |
| 307 | 9853 #ifdef FEAT_MBYTE |
| 389 | 9854 c = mb_ptr2char_adv(&p); |
| 307 | 9855 #else |
| 389 | 9856 c = *p++; |
| 307 | 9857 #endif |
| 389 | 9858 new_st.st_fold[i + 128] = c; |
| 9859 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256) | |
| 9860 new_st.st_upper[c] = i + 128; | |
| 9861 } | |
| 307 | 9862 } |
| 9863 | |
| 483 | 9864 (void)set_spell_finish(&new_st); |
| 307 | 9865 } |
| 9866 | |
| 9867 static int | |
| 9868 set_spell_finish(new_st) | |
| 9869 spelltab_T *new_st; | |
| 9870 { | |
| 9871 int i; | |
| 9872 | |
| 9873 if (did_set_spelltab) | |
| 9874 { | |
| 9875 /* check that it's the same table */ | |
| 9876 for (i = 0; i < 256; ++i) | |
| 9877 { | |
| 9878 if (spelltab.st_isw[i] != new_st->st_isw[i] | |
| 9879 || spelltab.st_isu[i] != new_st->st_isu[i] | |
| 324 | 9880 || spelltab.st_fold[i] != new_st->st_fold[i] |
| 9881 || spelltab.st_upper[i] != new_st->st_upper[i]) | |
| 307 | 9882 { |
| 9883 EMSG(_("E763: Word characters differ between spell files")); | |
| 9884 return FAIL; | |
| 9885 } | |
| 9886 } | |
| 9887 } | |
| 9888 else | |
| 9889 { | |
| 9890 /* copy the new spelltab into the one being used */ | |
| 9891 spelltab = *new_st; | |
| 9892 did_set_spelltab = TRUE; | |
| 9893 } | |
| 9894 | |
| 9895 return OK; | |
| 9896 } | |
| 9897 | |
| 9898 /* | |
| 358 | 9899 * Return TRUE if "p" points to a word character. |
| 366 | 9900 * As a special case we see "midword" characters as word character when it is |
| 358 | 9901 * followed by a word character. This finds they'there but not 'they there'. |
| 366 | 9902 * Thus this only works properly when past the first character of the word. |
| 358 | 9903 */ |
| 9904 static int | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9905 spell_iswordp(p, wp) |
| 358 | 9906 char_u *p; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9907 win_T *wp; /* buffer used */ |
| 358 | 9908 { |
| 366 | 9909 #ifdef FEAT_MBYTE |
| 358 | 9910 char_u *s; |
| 366 | 9911 int l; |
| 9912 int c; | |
| 9913 | |
| 9914 if (has_mbyte) | |
| 9915 { | |
| 9916 l = MB_BYTE2LEN(*p); | |
| 358 | 9917 s = p; |
| 366 | 9918 if (l == 1) |
| 9919 { | |
| 9920 /* be quick for ASCII */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9921 if (wp->w_s->b_spell_ismw[*p]) |
| 366 | 9922 s = p + 1; /* skip a mid-word character */ |
| 9923 } | |
| 9924 else | |
| 9925 { | |
| 9926 c = mb_ptr2char(p); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9927 if (c < 256 ? wp->w_s->b_spell_ismw[c] |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9928 : (wp->w_s->b_spell_ismw_mb != NULL |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9929 && vim_strchr(wp->w_s->b_spell_ismw_mb, c) != NULL)) |
| 366 | 9930 s = p + l; |
| 9931 } | |
| 9932 | |
| 390 | 9933 c = mb_ptr2char(s); |
| 9934 if (c > 255) | |
| 5477 | 9935 return spell_mb_isword_class(mb_get_class(s), wp); |
| 390 | 9936 return spelltab.st_isw[c]; |
| 366 | 9937 } |
| 358 | 9938 #endif |
| 366 | 9939 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9940 return spelltab.st_isw[wp->w_s->b_spell_ismw[*p] ? p[1] : p[0]]; |
| 376 | 9941 } |
| 9942 | |
| 9943 /* | |
| 9944 * Return TRUE if "p" points to a word character. | |
| 9945 * Unlike spell_iswordp() this doesn't check for "midword" characters. | |
| 9946 */ | |
| 9947 static int | |
| 5477 | 9948 spell_iswordp_nmw(p, wp) |
| 376 | 9949 char_u *p; |
| 5477 | 9950 win_T *wp; |
| 376 | 9951 { |
| 9952 #ifdef FEAT_MBYTE | |
| 390 | 9953 int c; |
| 9954 | |
| 9955 if (has_mbyte) | |
| 9956 { | |
| 9957 c = mb_ptr2char(p); | |
| 9958 if (c > 255) | |
| 5477 | 9959 return spell_mb_isword_class(mb_get_class(p), wp); |
| 390 | 9960 return spelltab.st_isw[c]; |
| 9961 } | |
| 376 | 9962 #endif |
| 9963 return spelltab.st_isw[*p]; | |
| 358 | 9964 } |
| 9965 | |
| 372 | 9966 #ifdef FEAT_MBYTE |
| 9967 /* | |
| 1580 | 9968 * Return TRUE if word class indicates a word character. |
| 9969 * Only for characters above 255. | |
| 9970 * Unicode subscript and superscript are not considered word characters. | |
| 5477 | 9971 * See also dbcs_class() and utf_class() in mbyte.c. |
| 9972 */ | |
| 9973 static int | |
| 9974 spell_mb_isword_class(cl, wp) | |
| 9975 int cl; | |
| 9976 win_T *wp; | |
| 9977 { | |
| 9978 if (wp->w_s->b_cjk) | |
| 9979 /* East Asian characters are not considered word characters. */ | |
| 9980 return cl == 2 || cl == 0x2800; | |
| 1580 | 9981 return cl >= 2 && cl != 0x2070 && cl != 0x2080; |
| 9982 } | |
| 9983 | |
| 9984 /* | |
| 372 | 9985 * Return TRUE if "p" points to a word character. |
| 9986 * Wide version of spell_iswordp(). | |
| 9987 */ | |
| 9988 static int | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9989 spell_iswordp_w(p, wp) |
| 372 | 9990 int *p; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9991 win_T *wp; |
| 372 | 9992 { |
| 9993 int *s; | |
| 9994 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9995 if (*p < 256 ? wp->w_s->b_spell_ismw[*p] |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9996 : (wp->w_s->b_spell_ismw_mb != NULL |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9997 && vim_strchr(wp->w_s->b_spell_ismw_mb, *p) != NULL)) |
| 372 | 9998 s = p + 1; |
| 9999 else | |
| 10000 s = p; | |
| 10001 | |
| 390 | 10002 if (*s > 255) |
| 372 | 10003 { |
| 10004 if (enc_utf8) | |
| 5477 | 10005 return spell_mb_isword_class(utf_class(*s), wp); |
| 372 | 10006 if (enc_dbcs) |
| 5477 | 10007 return spell_mb_isword_class( |
| 10008 dbcs_class((unsigned)*s >> 8, *s & 0xff), wp); | |
| 372 | 10009 return 0; |
| 10010 } | |
| 10011 return spelltab.st_isw[*s]; | |
| 10012 } | |
| 10013 #endif | |
| 10014 | |
| 358 | 10015 /* |
| 339 | 10016 * Write the table with prefix conditions to the .spl file. |
| 483 | 10017 * When "fd" is NULL only count the length of what is written. |
| 10018 */ | |
| 10019 static int | |
| 339 | 10020 write_spell_prefcond(fd, gap) |
| 10021 FILE *fd; | |
| 10022 garray_T *gap; | |
| 10023 { | |
| 10024 int i; | |
| 10025 char_u *p; | |
| 10026 int len; | |
| 483 | 10027 int totlen; |
| 1759 | 10028 size_t x = 1; /* collect return value of fwrite() */ |
| 483 | 10029 |
| 10030 if (fd != NULL) | |
| 10031 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */ | |
| 10032 | |
| 10033 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */ | |
| 339 | 10034 |
| 10035 for (i = 0; i < gap->ga_len; ++i) | |
| 10036 { | |
| 10037 /* <prefcond> : <condlen> <condstr> */ | |
| 10038 p = ((char_u **)gap->ga_data)[i]; | |
| 483 | 10039 if (p != NULL) |
| 339 | 10040 { |
| 835 | 10041 len = (int)STRLEN(p); |
| 483 | 10042 if (fd != NULL) |
| 10043 { | |
| 10044 fputc(len, fd); | |
| 1753 | 10045 x &= fwrite(p, (size_t)len, (size_t)1, fd); |
| 483 | 10046 } |
| 10047 totlen += len; | |
| 10048 } | |
| 10049 else if (fd != NULL) | |
| 10050 fputc(0, fd); | |
| 10051 } | |
| 10052 | |
| 10053 return totlen; | |
| 307 | 10054 } |
| 10055 | |
| 10056 /* | |
| 324 | 10057 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated. |
| 10058 * Uses the character definitions from the .spl file. | |
| 307 | 10059 * When using a multi-byte 'encoding' the length may change! |
| 10060 * Returns FAIL when something wrong. | |
| 10061 */ | |
| 10062 static int | |
| 324 | 10063 spell_casefold(str, len, buf, buflen) |
| 10064 char_u *str; | |
| 307 | 10065 int len; |
| 10066 char_u *buf; | |
| 10067 int buflen; | |
| 10068 { | |
| 10069 int i; | |
| 10070 | |
| 10071 if (len >= buflen) | |
| 10072 { | |
| 10073 buf[0] = NUL; | |
| 10074 return FAIL; /* result will not fit */ | |
| 10075 } | |
| 10076 | |
| 10077 #ifdef FEAT_MBYTE | |
| 10078 if (has_mbyte) | |
| 10079 { | |
| 324 | 10080 int outi = 0; |
| 10081 char_u *p; | |
| 307 | 10082 int c; |
| 10083 | |
| 10084 /* Fold one character at a time. */ | |
| 324 | 10085 for (p = str; p < str + len; ) |
| 307 | 10086 { |
| 10087 if (outi + MB_MAXBYTES > buflen) | |
| 10088 { | |
| 10089 buf[outi] = NUL; | |
| 10090 return FAIL; | |
| 10091 } | |
| 474 | 10092 c = mb_cptr2char_adv(&p); |
| 324 | 10093 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi); |
| 307 | 10094 } |
| 10095 buf[outi] = NUL; | |
| 10096 } | |
| 10097 else | |
| 10098 #endif | |
| 10099 { | |
| 10100 /* Be quick for non-multibyte encodings. */ | |
| 10101 for (i = 0; i < len; ++i) | |
| 324 | 10102 buf[i] = spelltab.st_fold[str[i]]; |
| 307 | 10103 buf[i] = NUL; |
| 10104 } | |
| 10105 | |
| 10106 return OK; | |
| 10107 } | |
| 10108 | |
| 625 | 10109 /* values for sps_flags */ |
| 372 | 10110 #define SPS_BEST 1 |
| 10111 #define SPS_FAST 2 | |
| 10112 #define SPS_DOUBLE 4 | |
| 10113 | |
| 625 | 10114 static int sps_flags = SPS_BEST; /* flags from 'spellsuggest' */ |
| 10115 static int sps_limit = 9999; /* max nr of suggestions given */ | |
| 372 | 10116 |
| 10117 /* | |
| 10118 * Check the 'spellsuggest' option. Return FAIL if it's wrong. | |
| 483 | 10119 * Sets "sps_flags" and "sps_limit". |
| 372 | 10120 */ |
| 10121 int | |
| 10122 spell_check_sps() | |
| 10123 { | |
| 10124 char_u *p; | |
| 483 | 10125 char_u *s; |
| 372 | 10126 char_u buf[MAXPATHL]; |
| 10127 int f; | |
| 10128 | |
| 10129 sps_flags = 0; | |
| 483 | 10130 sps_limit = 9999; |
| 372 | 10131 |
| 10132 for (p = p_sps; *p != NUL; ) | |
| 10133 { | |
| 10134 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 10135 | |
| 10136 f = 0; | |
| 483 | 10137 if (VIM_ISDIGIT(*buf)) |
| 10138 { | |
| 10139 s = buf; | |
| 10140 sps_limit = getdigits(&s); | |
| 10141 if (*s != NUL && !VIM_ISDIGIT(*s)) | |
| 10142 f = -1; | |
| 10143 } | |
| 10144 else if (STRCMP(buf, "best") == 0) | |
| 372 | 10145 f = SPS_BEST; |
| 10146 else if (STRCMP(buf, "fast") == 0) | |
| 10147 f = SPS_FAST; | |
| 10148 else if (STRCMP(buf, "double") == 0) | |
| 10149 f = SPS_DOUBLE; | |
| 10150 else if (STRNCMP(buf, "expr:", 5) != 0 | |
| 10151 && STRNCMP(buf, "file:", 5) != 0) | |
| 10152 f = -1; | |
| 10153 | |
| 10154 if (f == -1 || (sps_flags != 0 && f != 0)) | |
| 10155 { | |
| 10156 sps_flags = SPS_BEST; | |
| 483 | 10157 sps_limit = 9999; |
| 372 | 10158 return FAIL; |
| 10159 } | |
| 10160 if (f != 0) | |
| 10161 sps_flags = f; | |
| 10162 } | |
| 10163 | |
| 10164 if (sps_flags == 0) | |
| 10165 sps_flags = SPS_BEST; | |
| 10166 | |
| 10167 return OK; | |
| 10168 } | |
| 10169 | |
| 323 | 10170 /* |
| 5374 | 10171 * "z=": Find badly spelled word under or after the cursor. |
| 323 | 10172 * Give suggestions for the properly spelled word. |
| 638 | 10173 * In Visual mode use the highlighted word as the bad word. |
| 485 | 10174 * When "count" is non-zero use that suggestion. |
| 323 | 10175 */ |
| 10176 void | |
| 485 | 10177 spell_suggest(count) |
| 10178 int count; | |
| 323 | 10179 { |
| 10180 char_u *line; | |
| 10181 pos_T prev_cursor = curwin->w_cursor; | |
| 10182 char_u wcopy[MAXWLEN + 2]; | |
| 10183 char_u *p; | |
| 10184 int i; | |
| 10185 int c; | |
| 10186 suginfo_T sug; | |
| 10187 suggest_T *stp; | |
| 372 | 10188 int mouse_used; |
| 388 | 10189 int need_cap; |
| 483 | 10190 int limit; |
| 485 | 10191 int selected = count; |
| 638 | 10192 int badlen = 0; |
| 1938 | 10193 int msg_scroll_save = msg_scroll; |
| 638 | 10194 |
| 10195 if (no_spell_checking(curwin)) | |
| 10196 return; | |
| 10197 | |
| 10198 if (VIsual_active) | |
| 10199 { | |
| 10200 /* Use the Visually selected text as the bad word. But reject | |
| 10201 * a multi-line selection. */ | |
| 10202 if (curwin->w_cursor.lnum != VIsual.lnum) | |
| 10203 { | |
| 6949 | 10204 vim_beep(BO_SPELL); |
| 638 | 10205 return; |
| 10206 } | |
| 10207 badlen = (int)curwin->w_cursor.col - (int)VIsual.col; | |
| 10208 if (badlen < 0) | |
| 10209 badlen = -badlen; | |
| 10210 else | |
| 10211 curwin->w_cursor.col = VIsual.col; | |
| 10212 ++badlen; | |
| 10213 end_visual_mode(); | |
| 10214 } | |
| 5735 | 10215 /* Find the start of the badly spelled word. */ |
| 10216 else if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0 | |
| 346 | 10217 || curwin->w_cursor.col > prev_cursor.col) |
| 10218 { | |
| 10219 /* No bad word or it starts after the cursor: use the word under the | |
| 10220 * cursor. */ | |
| 10221 curwin->w_cursor = prev_cursor; | |
| 10222 line = ml_get_curline(); | |
| 10223 p = line + curwin->w_cursor.col; | |
| 10224 /* Backup to before start of word. */ | |
| 5477 | 10225 while (p > line && spell_iswordp_nmw(p, curwin)) |
| 346 | 10226 mb_ptr_back(line, p); |
| 10227 /* Forward to start of word. */ | |
| 5477 | 10228 while (*p != NUL && !spell_iswordp_nmw(p, curwin)) |
| 346 | 10229 mb_ptr_adv(p); |
| 10230 | |
| 5477 | 10231 if (!spell_iswordp_nmw(p, curwin)) /* No word found. */ |
| 346 | 10232 { |
| 10233 beep_flush(); | |
| 10234 return; | |
| 10235 } | |
| 835 | 10236 curwin->w_cursor.col = (colnr_T)(p - line); |
| 323 | 10237 } |
| 10238 | |
| 344 | 10239 /* Get the word and its length. */ |
| 10240 | |
| 388 | 10241 /* Figure out if the word should be capitalised. */ |
| 475 | 10242 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col); |
| 10243 | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10244 /* Make a copy of current line since autocommands may free the line. */ |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10245 line = vim_strsave(ml_get_curline()); |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10246 if (line == NULL) |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10247 goto skip; |
| 388 | 10248 |
| 483 | 10249 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in |
| 10250 * 'spellsuggest', whatever is smaller. */ | |
| 10251 if (sps_limit > (int)Rows - 2) | |
| 10252 limit = (int)Rows - 2; | |
| 10253 else | |
| 10254 limit = sps_limit; | |
| 638 | 10255 spell_find_suggest(line + curwin->w_cursor.col, badlen, &sug, limit, |
| 625 | 10256 TRUE, need_cap, TRUE); |
| 323 | 10257 |
| 10258 if (sug.su_ga.ga_len == 0) | |
| 10259 MSG(_("Sorry, no suggestions")); | |
| 485 | 10260 else if (count > 0) |
| 10261 { | |
| 10262 if (count > sug.su_ga.ga_len) | |
| 10263 smsg((char_u *)_("Sorry, only %ld suggestions"), | |
| 10264 (long)sug.su_ga.ga_len); | |
| 10265 } | |
| 323 | 10266 else |
| 10267 { | |
| 372 | 10268 vim_free(repl_from); |
| 10269 repl_from = NULL; | |
| 10270 vim_free(repl_to); | |
| 10271 repl_to = NULL; | |
| 10272 | |
| 474 | 10273 #ifdef FEAT_RIGHTLEFT |
| 10274 /* When 'rightleft' is set the list is drawn right-left. */ | |
| 10275 cmdmsg_rl = curwin->w_p_rl; | |
| 10276 if (cmdmsg_rl) | |
| 10277 msg_col = Columns - 1; | |
| 10278 #endif | |
| 10279 | |
| 323 | 10280 /* List the suggestions. */ |
| 10281 msg_start(); | |
| 914 | 10282 msg_row = Rows - 1; /* for when 'cmdheight' > 1 */ |
| 372 | 10283 lines_left = Rows; /* avoid more prompt */ |
| 323 | 10284 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"), |
| 10285 sug.su_badlen, sug.su_badptr); | |
| 474 | 10286 #ifdef FEAT_RIGHTLEFT |
| 10287 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0) | |
| 10288 { | |
| 10289 /* And now the rabbit from the high hat: Avoid showing the | |
| 10290 * untranslated message rightleft. */ | |
| 10291 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC", | |
| 10292 sug.su_badlen, sug.su_badptr); | |
| 10293 } | |
| 10294 #endif | |
| 323 | 10295 msg_puts(IObuff); |
| 10296 msg_clr_eos(); | |
| 10297 msg_putchar('\n'); | |
| 346 | 10298 |
| 323 | 10299 msg_scroll = TRUE; |
| 10300 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 10301 { | |
| 344 | 10302 stp = &SUG(sug.su_ga, i); |
| 323 | 10303 |
| 10304 /* The suggested word may replace only part of the bad word, add | |
| 10305 * the not replaced part. */ | |
| 2768 | 10306 vim_strncpy(wcopy, stp->st_word, MAXWLEN); |
| 323 | 10307 if (sug.su_badlen > stp->st_orglen) |
| 625 | 10308 vim_strncpy(wcopy + stp->st_wordlen, |
| 323 | 10309 sug.su_badptr + stp->st_orglen, |
| 10310 sug.su_badlen - stp->st_orglen); | |
| 474 | 10311 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1); |
| 10312 #ifdef FEAT_RIGHTLEFT | |
| 10313 if (cmdmsg_rl) | |
| 10314 rl_mirror(IObuff); | |
| 10315 #endif | |
| 10316 msg_puts(IObuff); | |
| 10317 | |
| 10318 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy); | |
| 346 | 10319 msg_puts(IObuff); |
| 10320 | |
| 10321 /* The word may replace more than "su_badlen". */ | |
| 10322 if (sug.su_badlen < stp->st_orglen) | |
| 10323 { | |
| 10324 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""), | |
| 10325 stp->st_orglen, sug.su_badptr); | |
| 10326 msg_puts(IObuff); | |
| 10327 } | |
| 10328 | |
| 324 | 10329 if (p_verbose > 0) |
| 344 | 10330 { |
| 346 | 10331 /* Add the score. */ |
| 351 | 10332 if (sps_flags & (SPS_DOUBLE | SPS_BEST)) |
| 474 | 10333 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)", |
| 344 | 10334 stp->st_salscore ? "s " : "", |
| 10335 stp->st_score, stp->st_altscore); | |
| 10336 else | |
| 474 | 10337 vim_snprintf((char *)IObuff, IOSIZE, " (%d)", |
| 346 | 10338 stp->st_score); |
| 474 | 10339 #ifdef FEAT_RIGHTLEFT |
| 10340 if (cmdmsg_rl) | |
| 10341 /* Mirror the numbers, but keep the leading space. */ | |
| 10342 rl_mirror(IObuff + 1); | |
| 10343 #endif | |
| 346 | 10344 msg_advance(30); |
| 10345 msg_puts(IObuff); | |
| 344 | 10346 } |
| 323 | 10347 msg_putchar('\n'); |
| 10348 } | |
| 10349 | |
| 474 | 10350 #ifdef FEAT_RIGHTLEFT |
| 10351 cmdmsg_rl = FALSE; | |
| 10352 msg_col = 0; | |
| 10353 #endif | |
| 323 | 10354 /* Ask for choice. */ |
| 485 | 10355 selected = prompt_for_number(&mouse_used); |
| 372 | 10356 if (mouse_used) |
| 485 | 10357 selected -= lines_left; |
| 1938 | 10358 lines_left = Rows; /* avoid more prompt */ |
| 10359 /* don't delay for 'smd' in normal_cmd() */ | |
| 10360 msg_scroll = msg_scroll_save; | |
| 485 | 10361 } |
| 10362 | |
| 10363 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK) | |
| 10364 { | |
| 10365 /* Save the from and to text for :spellrepall. */ | |
| 10366 stp = &SUG(sug.su_ga, selected - 1); | |
| 548 | 10367 if (sug.su_badlen > stp->st_orglen) |
| 10368 { | |
| 10369 /* Replacing less than "su_badlen", append the remainder to | |
| 10370 * repl_to. */ | |
| 10371 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen); | |
| 10372 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word, | |
| 10373 sug.su_badlen - stp->st_orglen, | |
| 10374 sug.su_badptr + stp->st_orglen); | |
| 10375 repl_to = vim_strsave(IObuff); | |
| 10376 } | |
| 10377 else | |
| 10378 { | |
| 10379 /* Replacing su_badlen or more, use the whole word. */ | |
| 10380 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen); | |
| 10381 repl_to = vim_strsave(stp->st_word); | |
| 10382 } | |
| 485 | 10383 |
| 10384 /* Replace the word. */ | |
| 1938 | 10385 p = alloc((unsigned)STRLEN(line) - stp->st_orglen |
| 10386 + stp->st_wordlen + 1); | |
| 485 | 10387 if (p != NULL) |
| 10388 { | |
| 835 | 10389 c = (int)(sug.su_badptr - line); |
| 485 | 10390 mch_memmove(p, line, c); |
| 10391 STRCPY(p + c, stp->st_word); | |
| 10392 STRCAT(p, sug.su_badptr + stp->st_orglen); | |
| 10393 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 10394 curwin->w_cursor.col = c; | |
| 10395 | |
| 10396 /* For redo we use a change-word command. */ | |
| 10397 ResetRedobuff(); | |
| 10398 AppendToRedobuff((char_u *)"ciw"); | |
| 620 | 10399 AppendToRedobuffLit(p + c, |
| 625 | 10400 stp->st_wordlen + sug.su_badlen - stp->st_orglen); |
| 485 | 10401 AppendCharToRedobuff(ESC); |
| 819 | 10402 |
| 10403 /* After this "p" may be invalid. */ | |
| 10404 changed_bytes(curwin->w_cursor.lnum, c); | |
| 485 | 10405 } |
| 10406 } | |
| 10407 else | |
| 10408 curwin->w_cursor = prev_cursor; | |
| 323 | 10409 |
| 344 | 10410 spell_find_cleanup(&sug); |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10411 skip: |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10412 vim_free(line); |
| 344 | 10413 } |
| 10414 | |
| 10415 /* | |
| 475 | 10416 * Check if the word at line "lnum" column "col" is required to start with a |
| 10417 * capital. This uses 'spellcapcheck' of the current buffer. | |
| 10418 */ | |
| 10419 static int | |
| 10420 check_need_cap(lnum, col) | |
| 10421 linenr_T lnum; | |
| 10422 colnr_T col; | |
| 10423 { | |
| 10424 int need_cap = FALSE; | |
| 10425 char_u *line; | |
| 10426 char_u *line_copy = NULL; | |
| 10427 char_u *p; | |
| 10428 colnr_T endcol; | |
| 10429 regmatch_T regmatch; | |
| 10430 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10431 if (curwin->w_s->b_cap_prog == NULL) |
| 475 | 10432 return FALSE; |
| 10433 | |
| 10434 line = ml_get_curline(); | |
| 10435 endcol = 0; | |
| 10436 if ((int)(skipwhite(line) - line) >= (int)col) | |
| 10437 { | |
| 10438 /* At start of line, check if previous line is empty or sentence | |
| 10439 * ends there. */ | |
| 10440 if (lnum == 1) | |
| 10441 need_cap = TRUE; | |
| 10442 else | |
| 10443 { | |
| 10444 line = ml_get(lnum - 1); | |
| 10445 if (*skipwhite(line) == NUL) | |
| 10446 need_cap = TRUE; | |
| 10447 else | |
| 10448 { | |
| 10449 /* Append a space in place of the line break. */ | |
| 10450 line_copy = concat_str(line, (char_u *)" "); | |
| 10451 line = line_copy; | |
| 835 | 10452 endcol = (colnr_T)STRLEN(line); |
| 475 | 10453 } |
| 10454 } | |
| 10455 } | |
| 10456 else | |
| 10457 endcol = col; | |
| 10458 | |
| 10459 if (endcol > 0) | |
| 10460 { | |
| 10461 /* Check if sentence ends before the bad word. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10462 regmatch.regprog = curwin->w_s->b_cap_prog; |
| 475 | 10463 regmatch.rm_ic = FALSE; |
| 10464 p = line + endcol; | |
| 10465 for (;;) | |
| 10466 { | |
| 10467 mb_ptr_back(line, p); | |
| 5477 | 10468 if (p == line || spell_iswordp_nmw(p, curwin)) |
| 475 | 10469 break; |
| 10470 if (vim_regexec(®match, p, 0) | |
| 10471 && regmatch.endp[0] == line + endcol) | |
| 10472 { | |
| 10473 need_cap = TRUE; | |
| 10474 break; | |
| 10475 } | |
| 10476 } | |
| 6375 | 10477 curwin->w_s->b_cap_prog = regmatch.regprog; |
| 475 | 10478 } |
| 10479 | |
| 10480 vim_free(line_copy); | |
| 10481 | |
| 10482 return need_cap; | |
| 10483 } | |
| 10484 | |
| 10485 | |
| 10486 /* | |
| 372 | 10487 * ":spellrepall" |
| 10488 */ | |
| 10489 void | |
| 10490 ex_spellrepall(eap) | |
| 1883 | 10491 exarg_T *eap UNUSED; |
| 372 | 10492 { |
| 10493 pos_T pos = curwin->w_cursor; | |
| 10494 char_u *frompat; | |
| 10495 int addlen; | |
| 10496 char_u *line; | |
| 10497 char_u *p; | |
| 10498 int save_ws = p_ws; | |
| 483 | 10499 linenr_T prev_lnum = 0; |
| 372 | 10500 |
| 10501 if (repl_from == NULL || repl_to == NULL) | |
| 10502 { | |
| 10503 EMSG(_("E752: No previous spell replacement")); | |
| 10504 return; | |
| 10505 } | |
| 835 | 10506 addlen = (int)(STRLEN(repl_to) - STRLEN(repl_from)); |
| 10507 | |
| 10508 frompat = alloc((unsigned)STRLEN(repl_from) + 7); | |
| 372 | 10509 if (frompat == NULL) |
| 10510 return; | |
| 10511 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from); | |
| 10512 p_ws = FALSE; | |
| 10513 | |
| 483 | 10514 sub_nsubs = 0; |
| 10515 sub_nlines = 0; | |
| 372 | 10516 curwin->w_cursor.lnum = 0; |
| 10517 while (!got_int) | |
| 10518 { | |
| 1521 | 10519 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP, NULL) == 0 |
| 372 | 10520 || u_save_cursor() == FAIL) |
| 10521 break; | |
| 10522 | |
| 10523 /* Only replace when the right word isn't there yet. This happens | |
| 10524 * when changing "etc" to "etc.". */ | |
| 10525 line = ml_get_curline(); | |
| 10526 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col, | |
| 10527 repl_to, STRLEN(repl_to)) != 0) | |
| 10528 { | |
| 835 | 10529 p = alloc((unsigned)STRLEN(line) + addlen + 1); |
| 372 | 10530 if (p == NULL) |
| 10531 break; | |
| 10532 mch_memmove(p, line, curwin->w_cursor.col); | |
| 10533 STRCPY(p + curwin->w_cursor.col, repl_to); | |
| 10534 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from)); | |
| 10535 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 10536 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col); | |
| 483 | 10537 |
| 10538 if (curwin->w_cursor.lnum != prev_lnum) | |
| 10539 { | |
| 10540 ++sub_nlines; | |
| 10541 prev_lnum = curwin->w_cursor.lnum; | |
| 10542 } | |
| 10543 ++sub_nsubs; | |
| 372 | 10544 } |
| 835 | 10545 curwin->w_cursor.col += (colnr_T)STRLEN(repl_to); |
| 372 | 10546 } |
| 10547 | |
| 10548 p_ws = save_ws; | |
| 10549 curwin->w_cursor = pos; | |
| 10550 vim_free(frompat); | |
| 10551 | |
| 483 | 10552 if (sub_nsubs == 0) |
| 372 | 10553 EMSG2(_("E753: Not found: %s"), repl_from); |
| 483 | 10554 else |
| 10555 do_sub_msg(FALSE); | |
| 372 | 10556 } |
| 10557 | |
| 10558 /* | |
| 344 | 10559 * Find spell suggestions for "word". Return them in the growarray "*gap" as |
| 10560 * a list of allocated strings. | |
| 10561 */ | |
| 10562 void | |
| 625 | 10563 spell_suggest_list(gap, word, maxcount, need_cap, interactive) |
| 344 | 10564 garray_T *gap; |
| 10565 char_u *word; | |
| 10566 int maxcount; /* maximum nr of suggestions */ | |
| 475 | 10567 int need_cap; /* 'spellcapcheck' matched */ |
| 625 | 10568 int interactive; |
| 344 | 10569 { |
| 10570 suginfo_T sug; | |
| 10571 int i; | |
| 10572 suggest_T *stp; | |
| 10573 char_u *wcopy; | |
| 10574 | |
| 638 | 10575 spell_find_suggest(word, 0, &sug, maxcount, FALSE, need_cap, interactive); |
| 344 | 10576 |
| 10577 /* Make room in "gap". */ | |
| 10578 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1); | |
| 626 | 10579 if (ga_grow(gap, sug.su_ga.ga_len) == OK) |
| 10580 { | |
| 10581 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 10582 { | |
| 10583 stp = &SUG(sug.su_ga, i); | |
| 10584 | |
| 10585 /* The suggested word may replace only part of "word", add the not | |
| 10586 * replaced part. */ | |
| 10587 wcopy = alloc(stp->st_wordlen | |
| 835 | 10588 + (unsigned)STRLEN(sug.su_badptr + stp->st_orglen) + 1); |
| 626 | 10589 if (wcopy == NULL) |
| 10590 break; | |
| 10591 STRCPY(wcopy, stp->st_word); | |
| 10592 STRCPY(wcopy + stp->st_wordlen, sug.su_badptr + stp->st_orglen); | |
| 10593 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy; | |
| 10594 } | |
| 344 | 10595 } |
| 10596 | |
| 10597 spell_find_cleanup(&sug); | |
| 10598 } | |
| 10599 | |
| 10600 /* | |
| 10601 * Find spell suggestions for the word at the start of "badptr". | |
| 10602 * Return the suggestions in "su->su_ga". | |
| 10603 * The maximum number of suggestions is "maxcount". | |
| 10604 * Note: does use info for the current window. | |
| 10605 * This is based on the mechanisms of Aspell, but completely reimplemented. | |
| 10606 */ | |
| 10607 static void | |
| 638 | 10608 spell_find_suggest(badptr, badlen, su, maxcount, banbadword, need_cap, interactive) |
| 344 | 10609 char_u *badptr; |
| 638 | 10610 int badlen; /* length of bad word or 0 if unknown */ |
| 344 | 10611 suginfo_T *su; |
| 10612 int maxcount; | |
| 358 | 10613 int banbadword; /* don't include badword in suggestions */ |
| 388 | 10614 int need_cap; /* word should start with capital */ |
| 625 | 10615 int interactive; |
| 344 | 10616 { |
| 534 | 10617 hlf_T attr = HLF_COUNT; |
| 372 | 10618 char_u buf[MAXPATHL]; |
| 10619 char_u *p; | |
| 10620 int do_combine = FALSE; | |
| 10621 char_u *sps_copy; | |
| 10622 #ifdef FEAT_EVAL | |
| 10623 static int expr_busy = FALSE; | |
| 10624 #endif | |
| 385 | 10625 int c; |
| 503 | 10626 int i; |
| 10627 langp_T *lp; | |
| 344 | 10628 |
| 10629 /* | |
| 10630 * Set the info in "*su". | |
| 10631 */ | |
| 10632 vim_memset(su, 0, sizeof(suginfo_T)); | |
| 10633 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10); | |
| 10634 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10); | |
| 355 | 10635 if (*badptr == NUL) |
| 10636 return; | |
| 344 | 10637 hash_init(&su->su_banned); |
| 10638 | |
| 10639 su->su_badptr = badptr; | |
| 638 | 10640 if (badlen != 0) |
| 10641 su->su_badlen = badlen; | |
| 10642 else | |
| 10643 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL, FALSE); | |
| 344 | 10644 su->su_maxcount = maxcount; |
| 372 | 10645 su->su_maxscore = SCORE_MAXINIT; |
| 344 | 10646 |
| 10647 if (su->su_badlen >= MAXWLEN) | |
| 10648 su->su_badlen = MAXWLEN - 1; /* just in case */ | |
| 10649 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen); | |
| 10650 (void)spell_casefold(su->su_badptr, su->su_badlen, | |
| 10651 su->su_fbadword, MAXWLEN); | |
| 346 | 10652 /* get caps flags for bad word */ |
| 474 | 10653 su->su_badflags = badword_captype(su->su_badptr, |
| 10654 su->su_badptr + su->su_badlen); | |
| 388 | 10655 if (need_cap) |
| 10656 su->su_badflags |= WF_ONECAP; | |
| 344 | 10657 |
| 503 | 10658 /* Find the default language for sound folding. We simply use the first |
| 10659 * one in 'spelllang' that supports sound folding. That's good for when | |
| 10660 * using multiple files for one language, it's not that bad when mixing | |
| 10661 * languages (e.g., "pl,en"). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10662 for (i = 0; i < curbuf->b_s.b_langp.ga_len; ++i) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10663 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10664 lp = LANGP_ENTRY(curbuf->b_s.b_langp, i); |
| 503 | 10665 if (lp->lp_sallang != NULL) |
| 10666 { | |
| 10667 su->su_sallang = lp->lp_sallang; | |
| 10668 break; | |
| 10669 } | |
| 10670 } | |
| 10671 | |
| 534 | 10672 /* Soundfold the bad word with the default sound folding, so that we don't |
| 10673 * have to do this many times. */ | |
| 10674 if (su->su_sallang != NULL) | |
| 10675 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE, | |
| 10676 su->su_sal_badword); | |
| 10677 | |
| 385 | 10678 /* If the word is not capitalised and spell_check() doesn't consider the |
| 10679 * word to be bad then it might need to be capitalised. Add a suggestion | |
| 10680 * for that. */ | |
| 455 | 10681 c = PTR2CHAR(su->su_badptr); |
| 534 | 10682 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT) |
| 385 | 10683 { |
| 10684 make_case_word(su->su_badword, buf, WF_ONECAP); | |
| 10685 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE, | |
| 625 | 10686 0, TRUE, su->su_sallang, FALSE); |
| 385 | 10687 } |
| 10688 | |
| 344 | 10689 /* Ban the bad word itself. It may appear in another region. */ |
| 358 | 10690 if (banbadword) |
| 10691 add_banned(su, su->su_badword); | |
| 344 | 10692 |
| 372 | 10693 /* Make a copy of 'spellsuggest', because the expression may change it. */ |
| 10694 sps_copy = vim_strsave(p_sps); | |
| 10695 if (sps_copy == NULL) | |
| 10696 return; | |
| 10697 | |
| 10698 /* Loop over the items in 'spellsuggest'. */ | |
| 10699 for (p = sps_copy; *p != NUL; ) | |
| 10700 { | |
| 10701 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 10702 | |
| 10703 if (STRNCMP(buf, "expr:", 5) == 0) | |
| 10704 { | |
| 10705 #ifdef FEAT_EVAL | |
| 375 | 10706 /* Evaluate an expression. Skip this when called recursively, |
| 10707 * when using spellsuggest() in the expression. */ | |
| 372 | 10708 if (!expr_busy) |
| 10709 { | |
| 10710 expr_busy = TRUE; | |
| 10711 spell_suggest_expr(su, buf + 5); | |
| 10712 expr_busy = FALSE; | |
| 10713 } | |
| 10714 #endif | |
| 10715 } | |
| 10716 else if (STRNCMP(buf, "file:", 5) == 0) | |
| 10717 /* Use list of suggestions in a file. */ | |
| 10718 spell_suggest_file(su, buf + 5); | |
| 10719 else | |
| 10720 { | |
| 10721 /* Use internal method. */ | |
| 625 | 10722 spell_suggest_intern(su, interactive); |
| 372 | 10723 if (sps_flags & SPS_DOUBLE) |
| 10724 do_combine = TRUE; | |
| 10725 } | |
| 10726 } | |
| 10727 | |
| 10728 vim_free(sps_copy); | |
| 10729 | |
| 10730 if (do_combine) | |
| 10731 /* Combine the two list of suggestions. This must be done last, | |
| 10732 * because sorting changes the order again. */ | |
| 10733 score_combine(su); | |
| 10734 } | |
| 10735 | |
| 10736 #ifdef FEAT_EVAL | |
| 10737 /* | |
| 10738 * Find suggestions by evaluating expression "expr". | |
| 10739 */ | |
| 10740 static void | |
| 10741 spell_suggest_expr(su, expr) | |
| 10742 suginfo_T *su; | |
| 10743 char_u *expr; | |
| 10744 { | |
| 10745 list_T *list; | |
| 10746 listitem_T *li; | |
| 10747 int score; | |
| 10748 char_u *p; | |
| 10749 | |
| 10750 /* The work is split up in a few parts to avoid having to export | |
| 10751 * suginfo_T. | |
| 10752 * First evaluate the expression and get the resulting list. */ | |
| 10753 list = eval_spell_expr(su->su_badword, expr); | |
| 10754 if (list != NULL) | |
| 10755 { | |
| 10756 /* Loop over the items in the list. */ | |
| 10757 for (li = list->lv_first; li != NULL; li = li->li_next) | |
| 10758 if (li->li_tv.v_type == VAR_LIST) | |
| 10759 { | |
| 10760 /* Get the word and the score from the items. */ | |
| 10761 score = get_spellword(li->li_tv.vval.v_list, &p); | |
| 625 | 10762 if (score >= 0 && score <= su->su_maxscore) |
| 10763 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 10764 score, 0, TRUE, su->su_sallang, FALSE); | |
| 372 | 10765 } |
| 10766 list_unref(list); | |
| 10767 } | |
| 10768 | |
| 625 | 10769 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 10770 check_suggestions(su, &su->su_ga); | |
| 372 | 10771 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 10772 } | |
| 10773 #endif | |
| 10774 | |
| 10775 /* | |
| 474 | 10776 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'. |
| 372 | 10777 */ |
| 10778 static void | |
| 10779 spell_suggest_file(su, fname) | |
| 10780 suginfo_T *su; | |
| 10781 char_u *fname; | |
| 10782 { | |
| 10783 FILE *fd; | |
| 10784 char_u line[MAXWLEN * 2]; | |
| 10785 char_u *p; | |
| 10786 int len; | |
| 10787 char_u cword[MAXWLEN]; | |
| 10788 | |
| 10789 /* Open the file. */ | |
| 10790 fd = mch_fopen((char *)fname, "r"); | |
| 10791 if (fd == NULL) | |
| 10792 { | |
| 10793 EMSG2(_(e_notopen), fname); | |
| 10794 return; | |
| 10795 } | |
| 10796 | |
| 10797 /* Read it line by line. */ | |
| 10798 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int) | |
| 10799 { | |
| 10800 line_breakcheck(); | |
| 10801 | |
| 10802 p = vim_strchr(line, '/'); | |
| 10803 if (p == NULL) | |
| 10804 continue; /* No Tab found, just skip the line. */ | |
| 10805 *p++ = NUL; | |
| 10806 if (STRICMP(su->su_badword, line) == 0) | |
| 10807 { | |
| 10808 /* Match! Isolate the good word, until CR or NL. */ | |
| 10809 for (len = 0; p[len] >= ' '; ++len) | |
| 10810 ; | |
| 10811 p[len] = NUL; | |
| 10812 | |
| 10813 /* If the suggestion doesn't have specific case duplicate the case | |
| 10814 * of the bad word. */ | |
| 10815 if (captype(p, NULL) == 0) | |
| 10816 { | |
| 10817 make_case_word(p, cword, su->su_badflags); | |
| 10818 p = cword; | |
| 10819 } | |
| 10820 | |
| 10821 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 625 | 10822 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE); |
| 372 | 10823 } |
| 10824 } | |
| 10825 | |
| 10826 fclose(fd); | |
| 10827 | |
| 625 | 10828 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 10829 check_suggestions(su, &su->su_ga); | |
| 372 | 10830 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 10831 } | |
| 10832 | |
| 10833 /* | |
| 10834 * Find suggestions for the internal method indicated by "sps_flags". | |
| 10835 */ | |
| 10836 static void | |
| 625 | 10837 spell_suggest_intern(su, interactive) |
| 372 | 10838 suginfo_T *su; |
| 625 | 10839 int interactive; |
| 10840 { | |
| 10841 /* | |
| 10842 * Load the .sug file(s) that are available and not done yet. | |
| 10843 */ | |
| 10844 suggest_load_files(); | |
| 10845 | |
| 344 | 10846 /* |
| 346 | 10847 * 1. Try special cases, such as repeating a word: "the the" -> "the". |
| 344 | 10848 * |
| 10849 * Set a maximum score to limit the combination of operations that is | |
| 10850 * tried. | |
| 10851 */ | |
| 346 | 10852 suggest_try_special(su); |
| 10853 | |
| 10854 /* | |
| 10855 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries | |
| 10856 * from the .aff file and inserting a space (split the word). | |
| 10857 */ | |
| 10858 suggest_try_change(su); | |
| 344 | 10859 |
| 10860 /* For the resulting top-scorers compute the sound-a-like score. */ | |
| 10861 if (sps_flags & SPS_DOUBLE) | |
| 10862 score_comp_sal(su); | |
| 10863 | |
| 10864 /* | |
| 346 | 10865 * 3. Try finding sound-a-like words. |
| 344 | 10866 */ |
| 625 | 10867 if ((sps_flags & SPS_FAST) == 0) |
| 10868 { | |
| 10869 if (sps_flags & SPS_BEST) | |
| 10870 /* Adjust the word score for the suggestions found so far for how | |
| 10871 * they sounds like. */ | |
| 10872 rescore_suggestions(su); | |
| 10873 | |
| 10874 /* | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10875 * While going through the soundfold tree "su_maxscore" is the score |
| 625 | 10876 * for the soundfold word, limits the changes that are being tried, |
| 10877 * and "su_sfmaxscore" the rescored score, which is set by | |
| 10878 * cleanup_suggestions(). | |
| 10879 * First find words with a small edit distance, because this is much | |
| 10880 * faster and often already finds the top-N suggestions. If we didn't | |
| 10881 * find many suggestions try again with a higher edit distance. | |
| 10882 * "sl_sounddone" is used to avoid doing the same word twice. | |
| 10883 */ | |
| 10884 suggest_try_soundalike_prep(); | |
| 10885 su->su_maxscore = SCORE_SFMAX1; | |
| 10886 su->su_sfmaxscore = SCORE_MAXINIT * 3; | |
| 346 | 10887 suggest_try_soundalike(su); |
| 625 | 10888 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) |
| 10889 { | |
| 10890 /* We didn't find enough matches, try again, allowing more | |
| 10891 * changes to the soundfold word. */ | |
| 10892 su->su_maxscore = SCORE_SFMAX2; | |
| 10893 suggest_try_soundalike(su); | |
| 10894 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) | |
| 10895 { | |
| 10896 /* Still didn't find enough matches, try again, allowing even | |
| 10897 * more changes to the soundfold word. */ | |
| 10898 su->su_maxscore = SCORE_SFMAX3; | |
| 10899 suggest_try_soundalike(su); | |
| 10900 } | |
| 10901 } | |
| 10902 su->su_maxscore = su->su_sfmaxscore; | |
| 10903 suggest_try_soundalike_finish(); | |
| 10904 } | |
| 10905 | |
| 10906 /* When CTRL-C was hit while searching do show the results. Only clear | |
| 10907 * got_int when using a command, not for spellsuggest(). */ | |
| 344 | 10908 ui_breakcheck(); |
| 625 | 10909 if (interactive && got_int) |
| 344 | 10910 { |
| 10911 (void)vgetc(); | |
| 10912 got_int = FALSE; | |
| 10913 } | |
| 10914 | |
| 372 | 10915 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0) |
| 344 | 10916 { |
| 10917 if (sps_flags & SPS_BEST) | |
| 10918 /* Adjust the word score for how it sounds like. */ | |
| 10919 rescore_suggestions(su); | |
| 10920 | |
| 625 | 10921 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 10922 check_suggestions(su, &su->su_ga); | |
| 372 | 10923 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 344 | 10924 } |
| 10925 } | |
| 10926 | |
| 10927 /* | |
| 625 | 10928 * Load the .sug files for languages that have one and weren't loaded yet. |
| 10929 */ | |
| 10930 static void | |
| 10931 suggest_load_files() | |
| 10932 { | |
| 10933 langp_T *lp; | |
| 10934 int lpi; | |
| 10935 slang_T *slang; | |
| 10936 char_u *dotp; | |
| 10937 FILE *fd; | |
| 10938 char_u buf[MAXWLEN]; | |
| 10939 int i; | |
| 10940 time_t timestamp; | |
| 10941 int wcount; | |
| 10942 int wordnr; | |
| 10943 garray_T ga; | |
| 10944 int c; | |
| 10945 | |
| 10946 /* Do this for all languages that support sound folding. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10947 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10948 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10949 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 10950 slang = lp->lp_slang; |
| 10951 if (slang->sl_sugtime != 0 && !slang->sl_sugloaded) | |
| 10952 { | |
| 10953 /* Change ".spl" to ".sug" and open the file. When the file isn't | |
| 10954 * found silently skip it. Do set "sl_sugloaded" so that we | |
| 10955 * don't try again and again. */ | |
| 10956 slang->sl_sugloaded = TRUE; | |
| 10957 | |
| 10958 dotp = vim_strrchr(slang->sl_fname, '.'); | |
| 10959 if (dotp == NULL || fnamecmp(dotp, ".spl") != 0) | |
| 10960 continue; | |
| 10961 STRCPY(dotp, ".sug"); | |
| 822 | 10962 fd = mch_fopen((char *)slang->sl_fname, "r"); |
| 625 | 10963 if (fd == NULL) |
| 10964 goto nextone; | |
| 10965 | |
| 10966 /* | |
| 10967 * <SUGHEADER>: <fileID> <versionnr> <timestamp> | |
| 10968 */ | |
| 10969 for (i = 0; i < VIMSUGMAGICL; ++i) | |
| 10970 buf[i] = getc(fd); /* <fileID> */ | |
| 10971 if (STRNCMP(buf, VIMSUGMAGIC, VIMSUGMAGICL) != 0) | |
| 10972 { | |
| 626 | 10973 EMSG2(_("E778: This does not look like a .sug file: %s"), |
| 625 | 10974 slang->sl_fname); |
| 10975 goto nextone; | |
| 10976 } | |
| 10977 c = getc(fd); /* <versionnr> */ | |
| 10978 if (c < VIMSUGVERSION) | |
| 10979 { | |
| 626 | 10980 EMSG2(_("E779: Old .sug file, needs to be updated: %s"), |
| 625 | 10981 slang->sl_fname); |
| 10982 goto nextone; | |
| 10983 } | |
| 10984 else if (c > VIMSUGVERSION) | |
| 10985 { | |
| 626 | 10986 EMSG2(_("E780: .sug file is for newer version of Vim: %s"), |
| 625 | 10987 slang->sl_fname); |
| 10988 goto nextone; | |
| 10989 } | |
| 10990 | |
| 10991 /* Check the timestamp, it must be exactly the same as the one in | |
| 10992 * the .spl file. Otherwise the word numbers won't match. */ | |
|
2229
d45902a5c61c
Fix a few more things for persistent undo.
Bram Moolenaar <bram@vim.org>
parents:
2214
diff
changeset
|
10993 timestamp = get8ctime(fd); /* <timestamp> */ |
| 625 | 10994 if (timestamp != slang->sl_sugtime) |
| 10995 { | |
| 626 | 10996 EMSG2(_("E781: .sug file doesn't match .spl file: %s"), |
| 625 | 10997 slang->sl_fname); |
| 10998 goto nextone; | |
| 10999 } | |
| 11000 | |
| 11001 /* | |
| 11002 * <SUGWORDTREE>: <wordtree> | |
| 11003 * Read the trie with the soundfolded words. | |
| 11004 */ | |
| 11005 if (spell_read_tree(fd, &slang->sl_sbyts, &slang->sl_sidxs, | |
| 11006 FALSE, 0) != 0) | |
| 11007 { | |
| 11008 someerror: | |
| 626 | 11009 EMSG2(_("E782: error while reading .sug file: %s"), |
| 625 | 11010 slang->sl_fname); |
| 11011 slang_clear_sug(slang); | |
| 11012 goto nextone; | |
| 11013 } | |
| 11014 | |
| 11015 /* | |
| 11016 * <SUGTABLE>: <sugwcount> <sugline> ... | |
| 11017 * | |
| 11018 * Read the table with word numbers. We use a file buffer for | |
| 11019 * this, because it's so much like a file with lines. Makes it | |
| 11020 * possible to swap the info and save on memory use. | |
| 11021 */ | |
| 11022 slang->sl_sugbuf = open_spellbuf(); | |
| 11023 if (slang->sl_sugbuf == NULL) | |
| 11024 goto someerror; | |
| 11025 /* <sugwcount> */ | |
| 695 | 11026 wcount = get4c(fd); |
| 625 | 11027 if (wcount < 0) |
| 11028 goto someerror; | |
| 11029 | |
| 11030 /* Read all the wordnr lists into the buffer, one NUL terminated | |
| 11031 * list per line. */ | |
| 11032 ga_init2(&ga, 1, 100); | |
| 11033 for (wordnr = 0; wordnr < wcount; ++wordnr) | |
| 11034 { | |
| 11035 ga.ga_len = 0; | |
| 11036 for (;;) | |
| 11037 { | |
| 11038 c = getc(fd); /* <sugline> */ | |
| 11039 if (c < 0 || ga_grow(&ga, 1) == FAIL) | |
| 11040 goto someerror; | |
| 11041 ((char_u *)ga.ga_data)[ga.ga_len++] = c; | |
| 11042 if (c == NUL) | |
| 11043 break; | |
| 11044 } | |
| 11045 if (ml_append_buf(slang->sl_sugbuf, (linenr_T)wordnr, | |
| 11046 ga.ga_data, ga.ga_len, TRUE) == FAIL) | |
| 11047 goto someerror; | |
| 11048 } | |
| 11049 ga_clear(&ga); | |
| 11050 | |
| 11051 /* | |
| 11052 * Need to put word counts in the word tries, so that we can find | |
| 11053 * a word by its number. | |
| 11054 */ | |
| 11055 tree_count_words(slang->sl_fbyts, slang->sl_fidxs); | |
| 11056 tree_count_words(slang->sl_sbyts, slang->sl_sidxs); | |
| 11057 | |
| 11058 nextone: | |
| 11059 if (fd != NULL) | |
| 11060 fclose(fd); | |
| 11061 STRCPY(dotp, ".spl"); | |
| 11062 } | |
| 11063 } | |
| 11064 } | |
| 11065 | |
| 11066 | |
| 11067 /* | |
| 11068 * Fill in the wordcount fields for a trie. | |
| 11069 * Returns the total number of words. | |
| 11070 */ | |
| 11071 static void | |
| 11072 tree_count_words(byts, idxs) | |
| 11073 char_u *byts; | |
| 11074 idx_T *idxs; | |
| 11075 { | |
| 11076 int depth; | |
| 11077 idx_T arridx[MAXWLEN]; | |
| 11078 int curi[MAXWLEN]; | |
| 11079 int c; | |
| 11080 idx_T n; | |
| 11081 int wordcount[MAXWLEN]; | |
| 11082 | |
| 11083 arridx[0] = 0; | |
| 11084 curi[0] = 1; | |
| 11085 wordcount[0] = 0; | |
| 11086 depth = 0; | |
| 11087 while (depth >= 0 && !got_int) | |
| 11088 { | |
| 11089 if (curi[depth] > byts[arridx[depth]]) | |
| 11090 { | |
| 11091 /* Done all bytes at this node, go up one level. */ | |
| 11092 idxs[arridx[depth]] = wordcount[depth]; | |
| 11093 if (depth > 0) | |
| 11094 wordcount[depth - 1] += wordcount[depth]; | |
| 11095 | |
| 11096 --depth; | |
| 11097 fast_breakcheck(); | |
| 11098 } | |
| 11099 else | |
| 11100 { | |
| 11101 /* Do one more byte at this node. */ | |
| 11102 n = arridx[depth] + curi[depth]; | |
| 11103 ++curi[depth]; | |
| 11104 | |
| 11105 c = byts[n]; | |
| 11106 if (c == 0) | |
| 11107 { | |
| 11108 /* End of word, count it. */ | |
| 11109 ++wordcount[depth]; | |
| 11110 | |
| 11111 /* Skip over any other NUL bytes (same word with different | |
| 11112 * flags). */ | |
| 11113 while (byts[n + 1] == 0) | |
| 11114 { | |
| 11115 ++n; | |
| 11116 ++curi[depth]; | |
| 11117 } | |
| 11118 } | |
| 11119 else | |
| 11120 { | |
| 11121 /* Normal char, go one level deeper to count the words. */ | |
| 11122 ++depth; | |
| 11123 arridx[depth] = idxs[n]; | |
| 11124 curi[depth] = 1; | |
| 11125 wordcount[depth] = 0; | |
| 11126 } | |
| 11127 } | |
| 11128 } | |
| 11129 } | |
| 11130 | |
| 11131 /* | |
| 344 | 11132 * Free the info put in "*su" by spell_find_suggest(). |
| 11133 */ | |
| 11134 static void | |
| 11135 spell_find_cleanup(su) | |
| 11136 suginfo_T *su; | |
| 11137 { | |
| 11138 int i; | |
| 11139 | |
| 11140 /* Free the suggestions. */ | |
| 11141 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 11142 vim_free(SUG(su->su_ga, i).st_word); | |
| 11143 ga_clear(&su->su_ga); | |
| 11144 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 11145 vim_free(SUG(su->su_sga, i).st_word); | |
| 11146 ga_clear(&su->su_sga); | |
| 323 | 11147 |
| 11148 /* Free the banned words. */ | |
| 625 | 11149 hash_clear_all(&su->su_banned, 0); |
| 323 | 11150 } |
| 11151 | |
| 11152 /* | |
| 324 | 11153 * Make a copy of "word", with the first letter upper or lower cased, to |
| 11154 * "wcopy[MAXWLEN]". "word" must not be empty. | |
| 11155 * The result is NUL terminated. | |
| 323 | 11156 */ |
| 11157 static void | |
| 324 | 11158 onecap_copy(word, wcopy, upper) |
| 323 | 11159 char_u *word; |
| 11160 char_u *wcopy; | |
| 11161 int upper; /* TRUE: first letter made upper case */ | |
| 11162 { | |
| 11163 char_u *p; | |
| 11164 int c; | |
| 11165 int l; | |
| 11166 | |
| 11167 p = word; | |
| 11168 #ifdef FEAT_MBYTE | |
| 11169 if (has_mbyte) | |
| 474 | 11170 c = mb_cptr2char_adv(&p); |
| 323 | 11171 else |
| 11172 #endif | |
| 11173 c = *p++; | |
| 11174 if (upper) | |
| 324 | 11175 c = SPELL_TOUPPER(c); |
| 323 | 11176 else |
| 324 | 11177 c = SPELL_TOFOLD(c); |
| 323 | 11178 #ifdef FEAT_MBYTE |
| 11179 if (has_mbyte) | |
| 11180 l = mb_char2bytes(c, wcopy); | |
| 11181 else | |
| 11182 #endif | |
| 11183 { | |
| 11184 l = 1; | |
| 11185 wcopy[0] = c; | |
| 11186 } | |
| 376 | 11187 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1); |
| 323 | 11188 } |
| 11189 | |
| 11190 /* | |
| 324 | 11191 * Make a copy of "word" with all the letters upper cased into |
| 11192 * "wcopy[MAXWLEN]". The result is NUL terminated. | |
| 323 | 11193 */ |
| 11194 static void | |
| 11195 allcap_copy(word, wcopy) | |
| 11196 char_u *word; | |
| 11197 char_u *wcopy; | |
| 11198 { | |
| 11199 char_u *s; | |
| 11200 char_u *d; | |
| 11201 int c; | |
| 11202 | |
| 11203 d = wcopy; | |
| 11204 for (s = word; *s != NUL; ) | |
| 11205 { | |
| 11206 #ifdef FEAT_MBYTE | |
| 11207 if (has_mbyte) | |
| 474 | 11208 c = mb_cptr2char_adv(&s); |
| 323 | 11209 else |
| 11210 #endif | |
| 11211 c = *s++; | |
| 492 | 11212 |
| 11213 #ifdef FEAT_MBYTE | |
| 3042 | 11214 /* We only change 0xdf to SS when we are certain latin1 is used. It |
| 492 | 11215 * would cause weird errors in other 8-bit encodings. */ |
| 11216 if (enc_latin1like && c == 0xdf) | |
| 11217 { | |
| 11218 c = 'S'; | |
| 11219 if (d - wcopy >= MAXWLEN - 1) | |
| 11220 break; | |
| 11221 *d++ = c; | |
| 11222 } | |
| 11223 else | |
| 11224 #endif | |
| 11225 c = SPELL_TOUPPER(c); | |
| 323 | 11226 |
| 11227 #ifdef FEAT_MBYTE | |
| 11228 if (has_mbyte) | |
| 11229 { | |
| 11230 if (d - wcopy >= MAXWLEN - MB_MAXBYTES) | |
| 11231 break; | |
| 11232 d += mb_char2bytes(c, d); | |
| 11233 } | |
| 11234 else | |
| 11235 #endif | |
| 11236 { | |
| 11237 if (d - wcopy >= MAXWLEN - 1) | |
| 11238 break; | |
| 11239 *d++ = c; | |
| 11240 } | |
| 11241 } | |
| 11242 *d = NUL; | |
| 11243 } | |
| 11244 | |
| 11245 /* | |
| 346 | 11246 * Try finding suggestions by recognizing specific situations. |
| 11247 */ | |
| 11248 static void | |
| 11249 suggest_try_special(su) | |
| 11250 suginfo_T *su; | |
| 11251 { | |
| 11252 char_u *p; | |
| 376 | 11253 size_t len; |
| 346 | 11254 int c; |
| 11255 char_u word[MAXWLEN]; | |
| 11256 | |
| 11257 /* | |
| 11258 * Recognize a word that is repeated: "the the". | |
| 11259 */ | |
| 11260 p = skiptowhite(su->su_fbadword); | |
| 11261 len = p - su->su_fbadword; | |
| 11262 p = skipwhite(p); | |
| 11263 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0) | |
| 11264 { | |
| 11265 /* Include badflags: if the badword is onecap or allcap | |
| 11266 * use that for the goodword too: "The the" -> "The". */ | |
| 11267 c = su->su_fbadword[len]; | |
| 11268 su->su_fbadword[len] = NUL; | |
| 11269 make_case_word(su->su_fbadword, word, su->su_badflags); | |
| 11270 su->su_fbadword[len] = c; | |
| 534 | 11271 |
| 11272 /* Give a soundalike score of 0, compute the score as if deleting one | |
| 11273 * character. */ | |
| 11274 add_suggestion(su, &su->su_ga, word, su->su_badlen, | |
| 625 | 11275 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE); |
| 346 | 11276 } |
| 11277 } | |
| 11278 | |
| 11279 /* | |
| 323 | 11280 * Try finding suggestions by adding/removing/swapping letters. |
| 11281 */ | |
| 11282 static void | |
| 346 | 11283 suggest_try_change(su) |
| 323 | 11284 suginfo_T *su; |
| 11285 { | |
| 11286 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */ | |
| 625 | 11287 int n; |
| 11288 char_u *p; | |
| 11289 int lpi; | |
| 323 | 11290 langp_T *lp; |
| 11291 | |
| 11292 /* We make a copy of the case-folded bad word, so that we can modify it | |
| 346 | 11293 * to find matches (esp. REP items). Append some more text, changing |
| 11294 * chars after the bad word may help. */ | |
| 323 | 11295 STRCPY(fword, su->su_fbadword); |
| 835 | 11296 n = (int)STRLEN(fword); |
| 346 | 11297 p = su->su_badptr + su->su_badlen; |
| 835 | 11298 (void)spell_casefold(p, (int)STRLEN(p), fword + n, MAXWLEN - n); |
| 323 | 11299 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11300 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11301 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11302 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 482 | 11303 |
| 500 | 11304 /* If reloading a spell file fails it's still in the list but |
| 11305 * everything has been cleared. */ | |
| 625 | 11306 if (lp->lp_slang->sl_fbyts == NULL) |
| 500 | 11307 continue; |
| 11308 | |
| 625 | 11309 /* Try it for this language. Will add possible suggestions. */ |
| 11310 suggest_trie_walk(su, lp, fword, FALSE); | |
| 11311 } | |
| 11312 } | |
| 11313 | |
| 11314 /* Check the maximum score, if we go over it we won't try this change. */ | |
| 11315 #define TRY_DEEPER(su, stack, depth, add) \ | |
| 11316 (stack[depth].ts_score + (add) < su->su_maxscore) | |
| 11317 | |
| 11318 /* | |
| 11319 * Try finding suggestions by adding/removing/swapping letters. | |
| 11320 * | |
| 11321 * This uses a state machine. At each node in the tree we try various | |
| 11322 * operations. When trying if an operation works "depth" is increased and the | |
| 11323 * stack[] is used to store info. This allows combinations, thus insert one | |
| 11324 * character, replace one and delete another. The number of changes is | |
| 11325 * limited by su->su_maxscore. | |
| 11326 * | |
| 11327 * After implementing this I noticed an article by Kemal Oflazer that | |
| 11328 * describes something similar: "Error-tolerant Finite State Recognition with | |
| 11329 * Applications to Morphological Analysis and Spelling Correction" (1996). | |
| 11330 * The implementation in the article is simplified and requires a stack of | |
| 11331 * unknown depth. The implementation here only needs a stack depth equal to | |
| 11332 * the length of the word. | |
| 11333 * | |
| 11334 * This is also used for the sound-folded word, "soundfold" is TRUE then. | |
| 11335 * The mechanism is the same, but we find a match with a sound-folded word | |
| 11336 * that comes from one or more original words. Each of these words may be | |
| 11337 * added, this is done by add_sound_suggest(). | |
| 11338 * Don't use: | |
| 11339 * the prefix tree or the keep-case tree | |
| 11340 * "su->su_badlen" | |
| 11341 * anything to do with upper and lower case | |
| 11342 * anything to do with word or non-word characters ("spell_iswordp()") | |
| 11343 * banned words | |
| 11344 * word flags (rare, region, compounding) | |
| 11345 * word splitting for now | |
| 11346 * "similar_chars()" | |
| 11347 * use "slang->sl_repsal" instead of "lp->lp_replang->sl_rep" | |
| 11348 */ | |
| 11349 static void | |
| 11350 suggest_trie_walk(su, lp, fword, soundfold) | |
| 11351 suginfo_T *su; | |
| 11352 langp_T *lp; | |
| 11353 char_u *fword; | |
| 11354 int soundfold; | |
| 11355 { | |
| 11356 char_u tword[MAXWLEN]; /* good word collected so far */ | |
| 11357 trystate_T stack[MAXWLEN]; | |
| 11358 char_u preword[MAXWLEN * 3]; /* word found with proper case; | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
11359 * concatenation of prefix compound |
| 625 | 11360 * words and split word. NUL terminated |
| 11361 * when going deeper but not when coming | |
| 11362 * back. */ | |
| 11363 char_u compflags[MAXWLEN]; /* compound flags, one for each word */ | |
| 11364 trystate_T *sp; | |
| 11365 int newscore; | |
| 11366 int score; | |
| 11367 char_u *byts, *fbyts, *pbyts; | |
| 11368 idx_T *idxs, *fidxs, *pidxs; | |
| 11369 int depth; | |
| 11370 int c, c2, c3; | |
| 11371 int n = 0; | |
| 11372 int flags; | |
| 11373 garray_T *gap; | |
| 11374 idx_T arridx; | |
| 11375 int len; | |
| 11376 char_u *p; | |
| 11377 fromto_T *ftp; | |
| 11378 int fl = 0, tl; | |
| 11379 int repextra = 0; /* extra bytes in fword[] from REP item */ | |
| 11380 slang_T *slang = lp->lp_slang; | |
| 11381 int fword_ends; | |
| 11382 int goodword_ends; | |
| 11383 #ifdef DEBUG_TRIEWALK | |
| 11384 /* Stores the name of the change made at each level. */ | |
| 11385 char_u changename[MAXWLEN][80]; | |
| 11386 #endif | |
| 11387 int breakcheckcount = 1000; | |
| 11388 int compound_ok; | |
| 11389 | |
| 11390 /* | |
| 11391 * Go through the whole case-fold tree, try changes at each node. | |
| 11392 * "tword[]" contains the word collected from nodes in the tree. | |
| 11393 * "fword[]" the word we are trying to match with (initially the bad | |
| 11394 * word). | |
| 11395 */ | |
| 11396 depth = 0; | |
| 11397 sp = &stack[0]; | |
| 11398 vim_memset(sp, 0, sizeof(trystate_T)); | |
| 11399 sp->ts_curi = 1; | |
| 11400 | |
| 11401 if (soundfold) | |
| 11402 { | |
| 11403 /* Going through the soundfold tree. */ | |
| 11404 byts = fbyts = slang->sl_sbyts; | |
| 11405 idxs = fidxs = slang->sl_sidxs; | |
| 11406 pbyts = NULL; | |
| 11407 pidxs = NULL; | |
| 11408 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 11409 sp->ts_state = STATE_START; | |
| 11410 } | |
| 11411 else | |
| 11412 { | |
| 330 | 11413 /* |
| 375 | 11414 * When there are postponed prefixes we need to use these first. At |
| 11415 * the end of the prefix we continue in the case-fold tree. | |
| 11416 */ | |
| 482 | 11417 fbyts = slang->sl_fbyts; |
| 11418 fidxs = slang->sl_fidxs; | |
| 11419 pbyts = slang->sl_pbyts; | |
| 11420 pidxs = slang->sl_pidxs; | |
| 375 | 11421 if (pbyts != NULL) |
| 11422 { | |
| 11423 byts = pbyts; | |
| 11424 idxs = pidxs; | |
| 482 | 11425 sp->ts_prefixdepth = PFD_PREFIXTREE; |
| 375 | 11426 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */ |
| 11427 } | |
| 11428 else | |
| 11429 { | |
| 11430 byts = fbyts; | |
| 11431 idxs = fidxs; | |
| 482 | 11432 sp->ts_prefixdepth = PFD_NOPREFIX; |
| 485 | 11433 sp->ts_state = STATE_START; |
| 375 | 11434 } |
| 625 | 11435 } |
| 11436 | |
| 11437 /* | |
| 11438 * Loop to find all suggestions. At each round we either: | |
| 11439 * - For the current state try one operation, advance "ts_curi", | |
| 11440 * increase "depth". | |
| 11441 * - When a state is done go to the next, set "ts_state". | |
| 11442 * - When all states are tried decrease "depth". | |
| 11443 */ | |
| 11444 while (depth >= 0 && !got_int) | |
| 11445 { | |
| 11446 sp = &stack[depth]; | |
| 11447 switch (sp->ts_state) | |
| 11448 { | |
| 11449 case STATE_START: | |
| 11450 case STATE_NOPREFIX: | |
| 11451 /* | |
| 11452 * Start of node: Deal with NUL bytes, which means | |
| 11453 * tword[] may end here. | |
| 11454 */ | |
| 11455 arridx = sp->ts_arridx; /* current node in the tree */ | |
| 11456 len = byts[arridx]; /* bytes in this node */ | |
| 11457 arridx += sp->ts_curi; /* index of current byte */ | |
| 11458 | |
| 11459 if (sp->ts_prefixdepth == PFD_PREFIXTREE) | |
| 11460 { | |
| 11461 /* Skip over the NUL bytes, we use them later. */ | |
| 11462 for (n = 0; n < len && byts[arridx + n] == 0; ++n) | |
| 11463 ; | |
| 11464 sp->ts_curi += n; | |
| 11465 | |
| 11466 /* Always past NUL bytes now. */ | |
| 11467 n = (int)sp->ts_state; | |
| 11468 sp->ts_state = STATE_ENDNUL; | |
| 11469 sp->ts_save_badflags = su->su_badflags; | |
| 11470 | |
| 11471 /* At end of a prefix or at start of prefixtree: check for | |
| 11472 * following word. */ | |
| 11473 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX) | |
| 11474 { | |
| 11475 /* Set su->su_badflags to the caps type at this position. | |
| 11476 * Use the caps type until here for the prefix itself. */ | |
| 11477 #ifdef FEAT_MBYTE | |
| 11478 if (has_mbyte) | |
| 11479 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); | |
| 11480 else | |
| 11481 #endif | |
| 11482 n = sp->ts_fidx; | |
| 11483 flags = badword_captype(su->su_badptr, su->su_badptr + n); | |
| 11484 su->su_badflags = badword_captype(su->su_badptr + n, | |
| 455 | 11485 su->su_badptr + su->su_badlen); |
| 625 | 11486 #ifdef DEBUG_TRIEWALK |
| 11487 sprintf(changename[depth], "prefix"); | |
| 11488 #endif | |
| 11489 go_deeper(stack, depth, 0); | |
| 11490 ++depth; | |
| 11491 sp = &stack[depth]; | |
| 11492 sp->ts_prefixdepth = depth - 1; | |
| 11493 byts = fbyts; | |
| 11494 idxs = fidxs; | |
| 11495 sp->ts_arridx = 0; | |
| 11496 | |
| 11497 /* Move the prefix to preword[] with the right case | |
| 11498 * and make find_keepcap_word() works. */ | |
| 11499 tword[sp->ts_twordlen] = NUL; | |
| 11500 make_case_word(tword + sp->ts_splitoff, | |
| 11501 preword + sp->ts_prewordlen, flags); | |
| 835 | 11502 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 625 | 11503 sp->ts_splitoff = sp->ts_twordlen; |
| 11504 } | |
| 11505 break; | |
| 11506 } | |
| 11507 | |
| 11508 if (sp->ts_curi > len || byts[arridx] != 0) | |
| 11509 { | |
| 11510 /* Past bytes in node and/or past NUL bytes. */ | |
| 11511 sp->ts_state = STATE_ENDNUL; | |
| 11512 sp->ts_save_badflags = su->su_badflags; | |
| 11513 break; | |
| 11514 } | |
| 11515 | |
| 11516 /* | |
| 11517 * End of word in tree. | |
| 11518 */ | |
| 11519 ++sp->ts_curi; /* eat one NUL byte */ | |
| 11520 | |
| 11521 flags = (int)idxs[arridx]; | |
| 698 | 11522 |
| 11523 /* Skip words with the NOSUGGEST flag. */ | |
| 11524 if (flags & WF_NOSUGGEST) | |
| 11525 break; | |
| 11526 | |
| 625 | 11527 fword_ends = (fword[sp->ts_fidx] == NUL |
| 11528 || (soundfold | |
| 11529 ? vim_iswhite(fword[sp->ts_fidx]) | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11530 : !spell_iswordp(fword + sp->ts_fidx, curwin))); |
| 625 | 11531 tword[sp->ts_twordlen] = NUL; |
| 11532 | |
| 11533 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL | |
| 485 | 11534 && (sp->ts_flags & TSF_PREFIXOK) == 0) |
| 625 | 11535 { |
| 11536 /* There was a prefix before the word. Check that the prefix | |
| 11537 * can be used with this word. */ | |
| 11538 /* Count the length of the NULs in the prefix. If there are | |
| 11539 * none this must be the first try without a prefix. */ | |
| 11540 n = stack[sp->ts_prefixdepth].ts_arridx; | |
| 11541 len = pbyts[n++]; | |
| 11542 for (c = 0; c < len && pbyts[n + c] == 0; ++c) | |
| 11543 ; | |
| 11544 if (c > 0) | |
| 11545 { | |
| 11546 c = valid_word_prefix(c, n, flags, | |
| 483 | 11547 tword + sp->ts_splitoff, slang, FALSE); |
| 625 | 11548 if (c == 0) |
| 11549 break; | |
| 11550 | |
| 11551 /* Use the WF_RARE flag for a rare prefix. */ | |
| 11552 if (c & WF_RAREPFX) | |
| 11553 flags |= WF_RARE; | |
| 11554 | |
| 11555 /* Tricky: when checking for both prefix and compounding | |
| 11556 * we run into the prefix flag first. | |
| 11557 * Remember that it's OK, so that we accept the prefix | |
| 11558 * when arriving at a compound flag. */ | |
| 11559 sp->ts_flags |= TSF_PREFIXOK; | |
| 11560 } | |
| 11561 } | |
| 11562 | |
| 11563 /* Check NEEDCOMPOUND: can't use word without compounding. Do try | |
| 11564 * appending another compound word below. */ | |
| 11565 if (sp->ts_complen == sp->ts_compsplit && fword_ends | |
| 500 | 11566 && (flags & WF_NEEDCOMP)) |
| 625 | 11567 goodword_ends = FALSE; |
| 11568 else | |
| 11569 goodword_ends = TRUE; | |
| 11570 | |
| 11571 p = NULL; | |
| 11572 compound_ok = TRUE; | |
| 11573 if (sp->ts_complen > sp->ts_compsplit) | |
| 11574 { | |
| 11575 if (slang->sl_nobreak) | |
| 11576 { | |
| 11577 /* There was a word before this word. When there was no | |
| 11578 * change in this word (it was correct) add the first word | |
| 11579 * as a suggestion. If this word was corrected too, we | |
| 11580 * need to check if a correct word follows. */ | |
| 11581 if (sp->ts_fidx - sp->ts_splitfidx | |
| 11582 == sp->ts_twordlen - sp->ts_splitoff | |
| 11583 && STRNCMP(fword + sp->ts_splitfidx, | |
| 11584 tword + sp->ts_splitoff, | |
| 11585 sp->ts_fidx - sp->ts_splitfidx) == 0) | |
| 492 | 11586 { |
| 625 | 11587 preword[sp->ts_prewordlen] = NUL; |
| 11588 newscore = score_wordcount_adj(slang, sp->ts_score, | |
| 11589 preword + sp->ts_prewordlen, | |
| 11590 sp->ts_prewordlen > 0); | |
| 11591 /* Add the suggestion if the score isn't too bad. */ | |
| 11592 if (newscore <= su->su_maxscore) | |
| 492 | 11593 add_suggestion(su, &su->su_ga, preword, |
| 503 | 11594 sp->ts_splitfidx - repextra, |
| 625 | 11595 newscore, 0, FALSE, |
| 11596 lp->lp_sallang, FALSE); | |
| 11597 break; | |
| 492 | 11598 } |
| 490 | 11599 } |
| 11600 else | |
| 625 | 11601 { |
| 11602 /* There was a compound word before this word. If this | |
| 11603 * word does not support compounding then give up | |
| 11604 * (splitting is tried for the word without compound | |
| 11605 * flag). */ | |
| 11606 if (((unsigned)flags >> 24) == 0 | |
| 11607 || sp->ts_twordlen - sp->ts_splitoff | |
| 11608 < slang->sl_compminlen) | |
| 11609 break; | |
| 11610 #ifdef FEAT_MBYTE | |
| 11611 /* For multi-byte chars check character length against | |
| 11612 * COMPOUNDMIN. */ | |
| 11613 if (has_mbyte | |
| 11614 && slang->sl_compminlen > 0 | |
| 11615 && mb_charlen(tword + sp->ts_splitoff) | |
| 11616 < slang->sl_compminlen) | |
| 11617 break; | |
| 11618 #endif | |
| 11619 | |
| 11620 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 11621 compflags[sp->ts_complen + 1] = NUL; | |
| 11622 vim_strncpy(preword + sp->ts_prewordlen, | |
| 11623 tword + sp->ts_splitoff, | |
| 11624 sp->ts_twordlen - sp->ts_splitoff); | |
| 1762 | 11625 |
| 11626 /* Verify CHECKCOMPOUNDPATTERN rules. */ | |
| 11627 if (match_checkcompoundpattern(preword, sp->ts_prewordlen, | |
| 11628 &slang->sl_comppat)) | |
| 11629 compound_ok = FALSE; | |
| 11630 | |
| 11631 if (compound_ok) | |
| 11632 { | |
| 11633 p = preword; | |
| 11634 while (*skiptowhite(p) != NUL) | |
| 11635 p = skipwhite(skiptowhite(p)); | |
| 11636 if (fword_ends && !can_compound(slang, p, | |
| 625 | 11637 compflags + sp->ts_compsplit)) |
| 1762 | 11638 /* Compound is not allowed. But it may still be |
| 11639 * possible if we add another (short) word. */ | |
| 11640 compound_ok = FALSE; | |
| 11641 } | |
| 625 | 11642 |
| 11643 /* Get pointer to last char of previous word. */ | |
| 11644 p = preword + sp->ts_prewordlen; | |
| 11645 mb_ptr_back(preword, p); | |
| 11646 } | |
| 11647 } | |
| 11648 | |
| 11649 /* | |
| 11650 * Form the word with proper case in preword. | |
| 11651 * If there is a word from a previous split, append. | |
| 11652 * For the soundfold tree don't change the case, simply append. | |
| 11653 */ | |
| 11654 if (soundfold) | |
| 11655 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff); | |
| 11656 else if (flags & WF_KEEPCAP) | |
| 11657 /* Must find the word in the keep-case tree. */ | |
| 11658 find_keepcap_word(slang, tword + sp->ts_splitoff, | |
| 483 | 11659 preword + sp->ts_prewordlen); |
| 625 | 11660 else |
| 11661 { | |
| 11662 /* Include badflags: If the badword is onecap or allcap | |
| 11663 * use that for the goodword too. But if the badword is | |
| 11664 * allcap and it's only one char long use onecap. */ | |
| 11665 c = su->su_badflags; | |
| 11666 if ((c & WF_ALLCAP) | |
| 11667 #ifdef FEAT_MBYTE | |
| 11668 && su->su_badlen == (*mb_ptr2len)(su->su_badptr) | |
| 346 | 11669 #else |
| 625 | 11670 && su->su_badlen == 1 |
| 11671 #endif | |
| 11672 ) | |
| 11673 c = WF_ONECAP; | |
| 11674 c |= flags; | |
| 11675 | |
| 11676 /* When appending a compound word after a word character don't | |
| 11677 * use Onecap. */ | |
| 5477 | 11678 if (p != NULL && spell_iswordp_nmw(p, curwin)) |
| 625 | 11679 c &= ~WF_ONECAP; |
| 11680 make_case_word(tword + sp->ts_splitoff, | |
| 490 | 11681 preword + sp->ts_prewordlen, c); |
| 625 | 11682 } |
| 11683 | |
| 11684 if (!soundfold) | |
| 11685 { | |
| 323 | 11686 /* Don't use a banned word. It may appear again as a good |
| 11687 * word, thus remember it. */ | |
| 11688 if (flags & WF_BANNED) | |
| 11689 { | |
| 483 | 11690 add_banned(su, preword + sp->ts_prewordlen); |
| 323 | 11691 break; |
| 11692 } | |
| 534 | 11693 if ((sp->ts_complen == sp->ts_compsplit |
| 625 | 11694 && WAS_BANNED(su, preword + sp->ts_prewordlen)) |
| 11695 || WAS_BANNED(su, preword)) | |
| 534 | 11696 { |
| 11697 if (slang->sl_compprog == NULL) | |
| 11698 break; | |
| 11699 /* the word so far was banned but we may try compounding */ | |
| 11700 goodword_ends = FALSE; | |
| 11701 } | |
| 625 | 11702 } |
| 11703 | |
| 11704 newscore = 0; | |
| 11705 if (!soundfold) /* soundfold words don't have flags */ | |
| 11706 { | |
| 323 | 11707 if ((flags & WF_REGION) |
| 390 | 11708 && (((unsigned)flags >> 16) & lp->lp_region) == 0) |
| 323 | 11709 newscore += SCORE_REGION; |
| 11710 if (flags & WF_RARE) | |
| 11711 newscore += SCORE_RARE; | |
| 11712 | |
| 346 | 11713 if (!spell_valid_case(su->su_badflags, |
| 483 | 11714 captype(preword + sp->ts_prewordlen, NULL))) |
| 323 | 11715 newscore += SCORE_ICASE; |
| 625 | 11716 } |
| 11717 | |
| 11718 /* TODO: how about splitting in the soundfold tree? */ | |
| 11719 if (fword_ends | |
| 11720 && goodword_ends | |
| 11721 && sp->ts_fidx >= sp->ts_fidxtry | |
| 11722 && compound_ok) | |
| 11723 { | |
| 11724 /* The badword also ends: add suggestions. */ | |
| 11725 #ifdef DEBUG_TRIEWALK | |
| 11726 if (soundfold && STRCMP(preword, "smwrd") == 0) | |
| 11727 { | |
| 11728 int j; | |
| 11729 | |
| 11730 /* print the stack of changes that brought us here */ | |
| 11731 smsg("------ %s -------", fword); | |
| 11732 for (j = 0; j < depth; ++j) | |
| 11733 smsg("%s", changename[j]); | |
| 11734 } | |
| 11735 #endif | |
| 11736 if (soundfold) | |
| 11737 { | |
| 11738 /* For soundfolded words we need to find the original | |
| 1188 | 11739 * words, the edit distance and then add them. */ |
| 625 | 11740 add_sound_suggest(su, preword, sp->ts_score, lp); |
| 11741 } | |
|
2457
6c1d10a85398
Avoid illegal memory access in spell suggestion. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2455
diff
changeset
|
11742 else if (sp->ts_fidx > 0) |
| 625 | 11743 { |
| 11744 /* Give a penalty when changing non-word char to word | |
| 11745 * char, e.g., "thes," -> "these". */ | |
| 366 | 11746 p = fword + sp->ts_fidx; |
| 625 | 11747 mb_ptr_back(fword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11748 if (!spell_iswordp(p, curwin)) |
| 366 | 11749 { |
| 11750 p = preword + STRLEN(preword); | |
| 625 | 11751 mb_ptr_back(preword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11752 if (spell_iswordp(p, curwin)) |
| 366 | 11753 newscore += SCORE_NONWORD; |
| 11754 } | |
| 11755 | |
| 625 | 11756 /* Give a bonus to words seen before. */ |
| 11757 score = score_wordcount_adj(slang, | |
| 11758 sp->ts_score + newscore, | |
| 11759 preword + sp->ts_prewordlen, | |
| 11760 sp->ts_prewordlen > 0); | |
| 11761 | |
| 11762 /* Add the suggestion if the score isn't too bad. */ | |
| 11763 if (score <= su->su_maxscore) | |
| 633 | 11764 { |
| 625 | 11765 add_suggestion(su, &su->su_ga, preword, |
| 11766 sp->ts_fidx - repextra, | |
| 11767 score, 0, FALSE, lp->lp_sallang, FALSE); | |
| 633 | 11768 |
| 11769 if (su->su_badflags & WF_MIXCAP) | |
| 11770 { | |
| 11771 /* We really don't know if the word should be | |
| 11772 * upper or lower case, add both. */ | |
| 11773 c = captype(preword, NULL); | |
| 11774 if (c == 0 || c == WF_ALLCAP) | |
| 11775 { | |
| 11776 make_case_word(tword + sp->ts_splitoff, | |
| 11777 preword + sp->ts_prewordlen, | |
| 11778 c == 0 ? WF_ALLCAP : 0); | |
| 11779 | |
| 11780 add_suggestion(su, &su->su_ga, preword, | |
| 11781 sp->ts_fidx - repextra, | |
| 11782 score + SCORE_ICASE, 0, FALSE, | |
| 11783 lp->lp_sallang, FALSE); | |
| 11784 } | |
| 11785 } | |
| 11786 } | |
| 625 | 11787 } |
| 11788 } | |
| 11789 | |
| 11790 /* | |
| 11791 * Try word split and/or compounding. | |
| 11792 */ | |
| 11793 if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends) | |
| 11794 #ifdef FEAT_MBYTE | |
| 11795 /* Don't split halfway a character. */ | |
| 11796 && (!has_mbyte || sp->ts_tcharlen == 0) | |
| 11797 #endif | |
| 11798 ) | |
| 11799 { | |
| 11800 int try_compound; | |
| 11801 int try_split; | |
| 11802 | |
| 11803 /* If past the end of the bad word don't try a split. | |
| 11804 * Otherwise try changing the next word. E.g., find | |
| 11805 * suggestions for "the the" where the second "the" is | |
| 11806 * different. It's done like a split. | |
| 11807 * TODO: word split for soundfold words */ | |
| 11808 try_split = (sp->ts_fidx - repextra < su->su_badlen) | |
| 11809 && !soundfold; | |
| 11810 | |
| 11811 /* Get here in several situations: | |
| 11812 * 1. The word in the tree ends: | |
| 11813 * If the word allows compounding try that. Otherwise try | |
| 11814 * a split by inserting a space. For both check that a | |
| 11815 * valid words starts at fword[sp->ts_fidx]. | |
| 11816 * For NOBREAK do like compounding to be able to check if | |
| 11817 * the next word is valid. | |
| 11818 * 2. The badword does end, but it was due to a change (e.g., | |
| 11819 * a swap). No need to split, but do check that the | |
| 11820 * following word is valid. | |
| 11821 * 3. The badword and the word in the tree end. It may still | |
| 11822 * be possible to compound another (short) word. | |
| 11823 */ | |
| 11824 try_compound = FALSE; | |
| 11825 if (!soundfold | |
| 11826 && slang->sl_compprog != NULL | |
| 11827 && ((unsigned)flags >> 24) != 0 | |
| 11828 && sp->ts_twordlen - sp->ts_splitoff | |
| 11829 >= slang->sl_compminlen | |
| 11830 #ifdef FEAT_MBYTE | |
| 11831 && (!has_mbyte | |
| 11832 || slang->sl_compminlen == 0 | |
| 11833 || mb_charlen(tword + sp->ts_splitoff) | |
| 11834 >= slang->sl_compminlen) | |
| 11835 #endif | |
| 11836 && (slang->sl_compsylmax < MAXWLEN | |
| 11837 || sp->ts_complen + 1 - sp->ts_compsplit | |
| 11838 < slang->sl_compmax) | |
| 1762 | 11839 && (can_be_compound(sp, slang, |
| 11840 compflags, ((unsigned)flags >> 24)))) | |
| 11841 | |
| 625 | 11842 { |
| 11843 try_compound = TRUE; | |
| 11844 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 11845 compflags[sp->ts_complen + 1] = NUL; | |
| 11846 } | |
| 11847 | |
| 11848 /* For NOBREAK we never try splitting, it won't make any word | |
| 11849 * valid. */ | |
| 11850 if (slang->sl_nobreak) | |
| 11851 try_compound = TRUE; | |
| 11852 | |
| 11853 /* If we could add a compound word, and it's also possible to | |
| 11854 * split at this point, do the split first and set | |
| 11855 * TSF_DIDSPLIT to avoid doing it again. */ | |
| 11856 else if (!fword_ends | |
| 11857 && try_compound | |
| 11858 && (sp->ts_flags & TSF_DIDSPLIT) == 0) | |
| 11859 { | |
| 485 | 11860 try_compound = FALSE; |
| 625 | 11861 sp->ts_flags |= TSF_DIDSPLIT; |
| 11862 --sp->ts_curi; /* do the same NUL again */ | |
| 11863 compflags[sp->ts_complen] = NUL; | |
| 11864 } | |
| 11865 else | |
| 11866 sp->ts_flags &= ~TSF_DIDSPLIT; | |
| 11867 | |
| 11868 if (try_split || try_compound) | |
| 11869 { | |
| 534 | 11870 if (!try_compound && (!fword_ends || !goodword_ends)) |
| 485 | 11871 { |
| 11872 /* If we're going to split need to check that the | |
| 501 | 11873 * words so far are valid for compounding. If there |
| 11874 * is only one word it must not have the NEEDCOMPOUND | |
| 11875 * flag. */ | |
| 11876 if (sp->ts_complen == sp->ts_compsplit | |
| 11877 && (flags & WF_NEEDCOMP)) | |
| 11878 break; | |
| 490 | 11879 p = preword; |
| 11880 while (*skiptowhite(p) != NUL) | |
| 11881 p = skipwhite(skiptowhite(p)); | |
| 485 | 11882 if (sp->ts_complen > sp->ts_compsplit |
| 490 | 11883 && !can_compound(slang, p, |
| 485 | 11884 compflags + sp->ts_compsplit)) |
| 11885 break; | |
| 698 | 11886 |
| 11887 if (slang->sl_nosplitsugs) | |
| 11888 newscore += SCORE_SPLIT_NO; | |
| 11889 else | |
| 11890 newscore += SCORE_SPLIT; | |
| 625 | 11891 |
| 11892 /* Give a bonus to words seen before. */ | |
| 11893 newscore = score_wordcount_adj(slang, newscore, | |
| 11894 preword + sp->ts_prewordlen, TRUE); | |
| 482 | 11895 } |
| 11896 | |
| 625 | 11897 if (TRY_DEEPER(su, stack, depth, newscore)) |
| 323 | 11898 { |
| 625 | 11899 go_deeper(stack, depth, newscore); |
| 11900 #ifdef DEBUG_TRIEWALK | |
| 11901 if (!try_compound && !fword_ends) | |
| 11902 sprintf(changename[depth], "%.*s-%s: split", | |
| 11903 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 11904 else | |
| 11905 sprintf(changename[depth], "%.*s-%s: compound", | |
| 11906 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 11907 #endif | |
| 323 | 11908 /* Save things to be restored at STATE_SPLITUNDO. */ |
| 346 | 11909 sp->ts_save_badflags = su->su_badflags; |
| 376 | 11910 sp->ts_state = STATE_SPLITUNDO; |
| 11911 | |
| 11912 ++depth; | |
| 11913 sp = &stack[depth]; | |
| 323 | 11914 |
| 482 | 11915 /* Append a space to preword when splitting. */ |
| 11916 if (!try_compound && !fword_ends) | |
| 11917 STRCAT(preword, " "); | |
| 835 | 11918 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 483 | 11919 sp->ts_splitoff = sp->ts_twordlen; |
| 492 | 11920 sp->ts_splitfidx = sp->ts_fidx; |
| 376 | 11921 |
| 11922 /* If the badword has a non-word character at this | |
| 11923 * position skip it. That means replacing the | |
| 482 | 11924 * non-word character with a space. Always skip a |
| 534 | 11925 * character when the word ends. But only when the |
| 11926 * good word can end. */ | |
| 625 | 11927 if (((!try_compound && !spell_iswordp_nmw(fword |
| 5477 | 11928 + sp->ts_fidx, |
| 11929 curwin)) | |
| 625 | 11930 || fword_ends) |
| 11931 && fword[sp->ts_fidx] != NUL | |
| 11932 && goodword_ends) | |
| 376 | 11933 { |
| 482 | 11934 int l; |
| 11935 | |
| 376 | 11936 #ifdef FEAT_MBYTE |
| 11937 if (has_mbyte) | |
| 482 | 11938 l = MB_BYTE2LEN(fword[sp->ts_fidx]); |
| 376 | 11939 else |
| 11940 #endif | |
| 482 | 11941 l = 1; |
| 11942 if (fword_ends) | |
| 11943 { | |
| 11944 /* Copy the skipped character to preword. */ | |
| 483 | 11945 mch_memmove(preword + sp->ts_prewordlen, |
| 482 | 11946 fword + sp->ts_fidx, l); |
| 483 | 11947 sp->ts_prewordlen += l; |
| 11948 preword[sp->ts_prewordlen] = NUL; | |
| 482 | 11949 } |
| 11950 else | |
| 11951 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST; | |
| 11952 sp->ts_fidx += l; | |
| 376 | 11953 } |
| 455 | 11954 |
| 485 | 11955 /* When compounding include compound flag in |
| 11956 * compflags[] (already set above). When splitting we | |
| 11957 * may start compounding over again. */ | |
| 482 | 11958 if (try_compound) |
| 483 | 11959 ++sp->ts_complen; |
| 482 | 11960 else |
| 485 | 11961 sp->ts_compsplit = sp->ts_complen; |
| 11962 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 482 | 11963 |
| 455 | 11964 /* set su->su_badflags to the caps type at this |
| 11965 * position */ | |
| 324 | 11966 #ifdef FEAT_MBYTE |
| 11967 if (has_mbyte) | |
| 455 | 11968 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); |
| 324 | 11969 else |
| 11970 #endif | |
| 455 | 11971 n = sp->ts_fidx; |
| 474 | 11972 su->su_badflags = badword_captype(su->su_badptr + n, |
| 455 | 11973 su->su_badptr + su->su_badlen); |
| 323 | 11974 |
| 11975 /* Restart at top of the tree. */ | |
| 376 | 11976 sp->ts_arridx = 0; |
| 485 | 11977 |
| 11978 /* If there are postponed prefixes, try these too. */ | |
| 11979 if (pbyts != NULL) | |
| 11980 { | |
| 11981 byts = pbyts; | |
| 11982 idxs = pidxs; | |
| 11983 sp->ts_prefixdepth = PFD_PREFIXTREE; | |
| 11984 sp->ts_state = STATE_NOPREFIX; | |
| 11985 } | |
| 323 | 11986 } |
| 11987 } | |
| 625 | 11988 } |
| 11989 break; | |
| 11990 | |
| 11991 case STATE_SPLITUNDO: | |
| 11992 /* Undo the changes done for word split or compound word. */ | |
| 11993 su->su_badflags = sp->ts_save_badflags; | |
| 11994 | |
| 11995 /* Continue looking for NUL bytes. */ | |
| 11996 sp->ts_state = STATE_START; | |
| 11997 | |
| 11998 /* In case we went into the prefix tree. */ | |
| 11999 byts = fbyts; | |
| 12000 idxs = fidxs; | |
| 12001 break; | |
| 12002 | |
| 12003 case STATE_ENDNUL: | |
| 12004 /* Past the NUL bytes in the node. */ | |
| 12005 su->su_badflags = sp->ts_save_badflags; | |
| 12006 if (fword[sp->ts_fidx] == NUL | |
| 12007 #ifdef FEAT_MBYTE | |
| 12008 && sp->ts_tcharlen == 0 | |
| 12009 #endif | |
| 12010 ) | |
| 12011 { | |
| 12012 /* The badword ends, can't use STATE_PLAIN. */ | |
| 12013 sp->ts_state = STATE_DEL; | |
| 323 | 12014 break; |
| 625 | 12015 } |
| 12016 sp->ts_state = STATE_PLAIN; | |
| 12017 /*FALLTHROUGH*/ | |
| 12018 | |
| 12019 case STATE_PLAIN: | |
| 12020 /* | |
| 12021 * Go over all possible bytes at this node, add each to tword[] | |
| 12022 * and use child node. "ts_curi" is the index. | |
| 12023 */ | |
| 12024 arridx = sp->ts_arridx; | |
| 12025 if (sp->ts_curi > byts[arridx]) | |
| 12026 { | |
| 12027 /* Done all bytes at this node, do next state. When still at | |
| 12028 * already changed bytes skip the other tricks. */ | |
| 12029 if (sp->ts_fidx >= sp->ts_fidxtry) | |
| 323 | 12030 sp->ts_state = STATE_DEL; |
| 12031 else | |
| 625 | 12032 sp->ts_state = STATE_FINAL; |
| 12033 } | |
| 12034 else | |
| 12035 { | |
| 12036 arridx += sp->ts_curi++; | |
| 12037 c = byts[arridx]; | |
| 12038 | |
| 12039 /* Normal byte, go one level deeper. If it's not equal to the | |
| 12040 * byte in the bad word adjust the score. But don't even try | |
| 12041 * when the byte was already changed. And don't try when we | |
| 5695 | 12042 * just deleted this byte, accepting it is always cheaper than |
| 625 | 12043 * delete + substitute. */ |
| 12044 if (c == fword[sp->ts_fidx] | |
| 12045 #ifdef FEAT_MBYTE | |
| 12046 || (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE) | |
| 12047 #endif | |
| 12048 ) | |
| 12049 newscore = 0; | |
| 12050 else | |
| 12051 newscore = SCORE_SUBST; | |
| 12052 if ((newscore == 0 | |
| 12053 || (sp->ts_fidx >= sp->ts_fidxtry | |
| 12054 && ((sp->ts_flags & TSF_DIDDEL) == 0 | |
| 12055 || c != fword[sp->ts_delidx]))) | |
| 12056 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 12057 { | |
| 12058 go_deeper(stack, depth, newscore); | |
| 12059 #ifdef DEBUG_TRIEWALK | |
| 12060 if (newscore > 0) | |
| 12061 sprintf(changename[depth], "%.*s-%s: subst %c to %c", | |
| 12062 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12063 fword[sp->ts_fidx], c); | |
| 323 | 12064 else |
| 625 | 12065 sprintf(changename[depth], "%.*s-%s: accept %c", |
| 12066 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12067 fword[sp->ts_fidx]); | |
| 12068 #endif | |
| 323 | 12069 ++depth; |
| 625 | 12070 sp = &stack[depth]; |
| 12071 ++sp->ts_fidx; | |
| 12072 tword[sp->ts_twordlen++] = c; | |
| 12073 sp->ts_arridx = idxs[arridx]; | |
| 12074 #ifdef FEAT_MBYTE | |
| 12075 if (newscore == SCORE_SUBST) | |
| 12076 sp->ts_isdiff = DIFF_YES; | |
| 330 | 12077 if (has_mbyte) |
| 358 | 12078 { |
| 625 | 12079 /* Multi-byte characters are a bit complicated to |
| 12080 * handle: They differ when any of the bytes differ | |
| 12081 * and then their length may also differ. */ | |
| 12082 if (sp->ts_tcharlen == 0) | |
| 12083 { | |
| 12084 /* First byte. */ | |
| 12085 sp->ts_tcharidx = 0; | |
| 12086 sp->ts_tcharlen = MB_BYTE2LEN(c); | |
| 12087 sp->ts_fcharstart = sp->ts_fidx - 1; | |
| 12088 sp->ts_isdiff = (newscore != 0) | |
| 12089 ? DIFF_YES : DIFF_NONE; | |
| 12090 } | |
| 12091 else if (sp->ts_isdiff == DIFF_INSERT) | |
| 12092 /* When inserting trail bytes don't advance in the | |
| 12093 * bad word. */ | |
| 12094 --sp->ts_fidx; | |
| 12095 if (++sp->ts_tcharidx == sp->ts_tcharlen) | |
| 12096 { | |
| 12097 /* Last byte of character. */ | |
| 12098 if (sp->ts_isdiff == DIFF_YES) | |
| 12099 { | |
| 12100 /* Correct ts_fidx for the byte length of the | |
| 12101 * character (we didn't check that before). */ | |
| 12102 sp->ts_fidx = sp->ts_fcharstart | |
| 12103 + MB_BYTE2LEN( | |
| 12104 fword[sp->ts_fcharstart]); | |
| 12105 | |
| 12106 /* For changing a composing character adjust | |
| 12107 * the score from SCORE_SUBST to | |
| 12108 * SCORE_SUBCOMP. */ | |
| 12109 if (enc_utf8 | |
| 12110 && utf_iscomposing( | |
| 12111 mb_ptr2char(tword | |
| 12112 + sp->ts_twordlen | |
| 12113 - sp->ts_tcharlen)) | |
| 12114 && utf_iscomposing( | |
| 12115 mb_ptr2char(fword | |
| 12116 + sp->ts_fcharstart))) | |
| 12117 sp->ts_score -= | |
| 12118 SCORE_SUBST - SCORE_SUBCOMP; | |
| 12119 | |
| 12120 /* For a similar character adjust score from | |
| 12121 * SCORE_SUBST to SCORE_SIMILAR. */ | |
| 12122 else if (!soundfold | |
| 12123 && slang->sl_has_map | |
| 12124 && similar_chars(slang, | |
| 12125 mb_ptr2char(tword | |
| 12126 + sp->ts_twordlen | |
| 12127 - sp->ts_tcharlen), | |
| 12128 mb_ptr2char(fword | |
| 12129 + sp->ts_fcharstart))) | |
| 12130 sp->ts_score -= | |
| 12131 SCORE_SUBST - SCORE_SIMILAR; | |
| 12132 } | |
| 12133 else if (sp->ts_isdiff == DIFF_INSERT | |
| 12134 && sp->ts_twordlen > sp->ts_tcharlen) | |
| 12135 { | |
| 12136 p = tword + sp->ts_twordlen - sp->ts_tcharlen; | |
| 12137 c = mb_ptr2char(p); | |
| 12138 if (enc_utf8 && utf_iscomposing(c)) | |
| 12139 { | |
| 12140 /* Inserting a composing char doesn't | |
| 12141 * count that much. */ | |
| 12142 sp->ts_score -= SCORE_INS - SCORE_INSCOMP; | |
| 12143 } | |
| 12144 else | |
| 12145 { | |
| 12146 /* If the previous character was the same, | |
| 12147 * thus doubling a character, give a bonus | |
| 12148 * to the score. Also for the soundfold | |
| 12149 * tree (might seem illogical but does | |
| 12150 * give better scores). */ | |
| 12151 mb_ptr_back(tword, p); | |
| 12152 if (c == mb_ptr2char(p)) | |
| 12153 sp->ts_score -= SCORE_INS | |
| 12154 - SCORE_INSDUP; | |
| 12155 } | |
| 12156 } | |
| 12157 | |
| 12158 /* Starting a new char, reset the length. */ | |
| 12159 sp->ts_tcharlen = 0; | |
| 12160 } | |
| 358 | 12161 } |
| 330 | 12162 else |
| 12163 #endif | |
| 358 | 12164 { |
| 625 | 12165 /* If we found a similar char adjust the score. |
| 12166 * We do this after calling go_deeper() because | |
| 12167 * it's slow. */ | |
| 12168 if (newscore != 0 | |
| 12169 && !soundfold | |
| 12170 && slang->sl_has_map | |
| 12171 && similar_chars(slang, | |
| 12172 c, fword[sp->ts_fidx - 1])) | |
| 12173 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR; | |
| 358 | 12174 } |
| 625 | 12175 } |
| 12176 } | |
| 12177 break; | |
| 12178 | |
| 12179 case STATE_DEL: | |
| 12180 #ifdef FEAT_MBYTE | |
| 12181 /* When past the first byte of a multi-byte char don't try | |
| 12182 * delete/insert/swap a character. */ | |
| 12183 if (has_mbyte && sp->ts_tcharlen > 0) | |
| 12184 { | |
| 12185 sp->ts_state = STATE_FINAL; | |
| 12186 break; | |
| 12187 } | |
| 12188 #endif | |
| 12189 /* | |
| 12190 * Try skipping one character in the bad word (delete it). | |
| 12191 */ | |
| 12192 sp->ts_state = STATE_INS_PREP; | |
| 12193 sp->ts_curi = 1; | |
| 12194 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*') | |
| 12195 /* Deleting a vowel at the start of a word counts less, see | |
| 12196 * soundalike_score(). */ | |
| 12197 newscore = 2 * SCORE_DEL / 3; | |
| 12198 else | |
| 12199 newscore = SCORE_DEL; | |
| 12200 if (fword[sp->ts_fidx] != NUL | |
| 12201 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 12202 { | |
| 12203 go_deeper(stack, depth, newscore); | |
| 12204 #ifdef DEBUG_TRIEWALK | |
| 12205 sprintf(changename[depth], "%.*s-%s: delete %c", | |
| 12206 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12207 fword[sp->ts_fidx]); | |
| 12208 #endif | |
| 12209 ++depth; | |
| 12210 | |
| 12211 /* Remember what character we deleted, so that we can avoid | |
| 12212 * inserting it again. */ | |
| 12213 stack[depth].ts_flags |= TSF_DIDDEL; | |
| 12214 stack[depth].ts_delidx = sp->ts_fidx; | |
| 12215 | |
| 12216 /* Advance over the character in fword[]. Give a bonus to the | |
| 12217 * score if the same character is following "nn" -> "n". It's | |
| 12218 * a bit illogical for soundfold tree but it does give better | |
| 12219 * results. */ | |
| 12220 #ifdef FEAT_MBYTE | |
| 12221 if (has_mbyte) | |
| 12222 { | |
| 12223 c = mb_ptr2char(fword + sp->ts_fidx); | |
| 12224 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]); | |
| 12225 if (enc_utf8 && utf_iscomposing(c)) | |
| 12226 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP; | |
| 12227 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx)) | |
| 12228 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 12229 } |
| 12230 else | |
| 625 | 12231 #endif |
| 12232 { | |
| 12233 ++stack[depth].ts_fidx; | |
| 12234 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1]) | |
| 12235 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 12236 } |
| 12237 break; | |
| 625 | 12238 } |
| 12239 /*FALLTHROUGH*/ | |
| 12240 | |
| 12241 case STATE_INS_PREP: | |
| 12242 if (sp->ts_flags & TSF_DIDDEL) | |
| 12243 { | |
| 12244 /* If we just deleted a byte then inserting won't make sense, | |
| 12245 * a substitute is always cheaper. */ | |
| 12246 sp->ts_state = STATE_SWAP; | |
| 12247 break; | |
| 12248 } | |
| 12249 | |
| 12250 /* skip over NUL bytes */ | |
| 12251 n = sp->ts_arridx; | |
| 12252 for (;;) | |
| 12253 { | |
| 12254 if (sp->ts_curi > byts[n]) | |
| 12255 { | |
| 12256 /* Only NUL bytes at this node, go to next state. */ | |
| 12257 sp->ts_state = STATE_SWAP; | |
| 12258 break; | |
| 12259 } | |
| 12260 if (byts[n + sp->ts_curi] != NUL) | |
| 12261 { | |
| 12262 /* Found a byte to insert. */ | |
| 12263 sp->ts_state = STATE_INS; | |
| 330 | 12264 break; |
| 12265 } | |
| 625 | 12266 ++sp->ts_curi; |
| 12267 } | |
| 12268 break; | |
| 12269 | |
| 12270 /*FALLTHROUGH*/ | |
| 12271 | |
| 12272 case STATE_INS: | |
| 12273 /* Insert one byte. Repeat this for each possible byte at this | |
| 12274 * node. */ | |
| 12275 n = sp->ts_arridx; | |
| 12276 if (sp->ts_curi > byts[n]) | |
| 12277 { | |
| 12278 /* Done all bytes at this node, go to next state. */ | |
| 12279 sp->ts_state = STATE_SWAP; | |
| 12280 break; | |
| 12281 } | |
| 12282 | |
| 12283 /* Do one more byte at this node, but: | |
| 12284 * - Skip NUL bytes. | |
| 12285 * - Skip the byte if it's equal to the byte in the word, | |
| 12286 * accepting that byte is always better. | |
| 12287 */ | |
| 12288 n += sp->ts_curi++; | |
| 12289 c = byts[n]; | |
| 12290 if (soundfold && sp->ts_twordlen == 0 && c == '*') | |
| 12291 /* Inserting a vowel at the start of a word counts less, | |
| 12292 * see soundalike_score(). */ | |
| 12293 newscore = 2 * SCORE_INS / 3; | |
| 12294 else | |
| 12295 newscore = SCORE_INS; | |
| 12296 if (c != fword[sp->ts_fidx] | |
| 12297 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 12298 { | |
| 12299 go_deeper(stack, depth, newscore); | |
| 12300 #ifdef DEBUG_TRIEWALK | |
| 12301 sprintf(changename[depth], "%.*s-%s: insert %c", | |
| 12302 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12303 c); | |
| 12304 #endif | |
| 12305 ++depth; | |
| 12306 sp = &stack[depth]; | |
| 12307 tword[sp->ts_twordlen++] = c; | |
| 12308 sp->ts_arridx = idxs[n]; | |
| 330 | 12309 #ifdef FEAT_MBYTE |
| 12310 if (has_mbyte) | |
| 12311 { | |
| 625 | 12312 fl = MB_BYTE2LEN(c); |
| 12313 if (fl > 1) | |
| 330 | 12314 { |
| 625 | 12315 /* There are following bytes for the same character. |
| 12316 * We must find all bytes before trying | |
| 12317 * delete/insert/swap/etc. */ | |
| 12318 sp->ts_tcharlen = fl; | |
| 12319 sp->ts_tcharidx = 1; | |
| 12320 sp->ts_isdiff = DIFF_INSERT; | |
| 330 | 12321 } |
| 323 | 12322 } |
| 12323 else | |
| 625 | 12324 fl = 1; |
| 12325 if (fl == 1) | |
| 12326 #endif | |
| 12327 { | |
| 12328 /* If the previous character was the same, thus doubling a | |
| 12329 * character, give a bonus to the score. Also for | |
| 12330 * soundfold words (illogical but does give a better | |
| 12331 * score). */ | |
| 12332 if (sp->ts_twordlen >= 2 | |
| 12333 && tword[sp->ts_twordlen - 2] == c) | |
| 12334 sp->ts_score -= SCORE_INS - SCORE_INSDUP; | |
| 12335 } | |
| 12336 } | |
| 12337 break; | |
| 12338 | |
| 12339 case STATE_SWAP: | |
| 12340 /* | |
| 12341 * Swap two bytes in the bad word: "12" -> "21". | |
| 12342 * We change "fword" here, it's changed back afterwards at | |
| 12343 * STATE_UNSWAP. | |
| 12344 */ | |
| 12345 p = fword + sp->ts_fidx; | |
| 12346 c = *p; | |
| 12347 if (c == NUL) | |
| 12348 { | |
| 12349 /* End of word, can't swap or replace. */ | |
| 12350 sp->ts_state = STATE_FINAL; | |
| 12351 break; | |
| 12352 } | |
| 12353 | |
| 12354 /* Don't swap if the first character is not a word character. | |
| 12355 * SWAP3 etc. also don't make sense then. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12356 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 12357 { |
| 12358 sp->ts_state = STATE_REP_INI; | |
| 323 | 12359 break; |
| 625 | 12360 } |
| 12361 | |
| 12362 #ifdef FEAT_MBYTE | |
| 12363 if (has_mbyte) | |
| 12364 { | |
| 12365 n = mb_cptr2len(p); | |
| 12366 c = mb_ptr2char(p); | |
| 1337 | 12367 if (p[n] == NUL) |
| 12368 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12369 else if (!soundfold && !spell_iswordp(p + n, curwin)) |
| 625 | 12370 c2 = c; /* don't swap non-word char */ |
| 12371 else | |
| 12372 c2 = mb_ptr2char(p + n); | |
| 12373 } | |
| 12374 else | |
| 12375 #endif | |
| 12376 { | |
| 1337 | 12377 if (p[1] == NUL) |
| 12378 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12379 else if (!soundfold && !spell_iswordp(p + 1, curwin)) |
| 625 | 12380 c2 = c; /* don't swap non-word char */ |
| 12381 else | |
| 12382 c2 = p[1]; | |
| 12383 } | |
| 12384 | |
| 1337 | 12385 /* When the second character is NUL we can't swap. */ |
| 12386 if (c2 == NUL) | |
| 12387 { | |
| 12388 sp->ts_state = STATE_REP_INI; | |
| 12389 break; | |
| 12390 } | |
| 12391 | |
| 625 | 12392 /* When characters are identical, swap won't do anything. |
| 12393 * Also get here if the second char is not a word character. */ | |
| 12394 if (c == c2) | |
| 12395 { | |
| 12396 sp->ts_state = STATE_SWAP3; | |
| 12397 break; | |
| 12398 } | |
| 12399 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP)) | |
| 12400 { | |
| 12401 go_deeper(stack, depth, SCORE_SWAP); | |
| 12402 #ifdef DEBUG_TRIEWALK | |
| 12403 sprintf(changename[depth], "%.*s-%s: swap %c and %c", | |
| 12404 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12405 c, c2); | |
| 12406 #endif | |
| 12407 sp->ts_state = STATE_UNSWAP; | |
| 12408 ++depth; | |
| 330 | 12409 #ifdef FEAT_MBYTE |
| 12410 if (has_mbyte) | |
| 12411 { | |
| 625 | 12412 fl = mb_char2len(c2); |
| 12413 mch_memmove(p, p + n, fl); | |
| 12414 mb_char2bytes(c, p + fl); | |
| 12415 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 12416 } |
| 12417 else | |
| 12418 #endif | |
| 12419 { | |
| 625 | 12420 p[0] = c2; |
| 330 | 12421 p[1] = c; |
| 625 | 12422 stack[depth].ts_fidxtry = sp->ts_fidx + 2; |
| 12423 } | |
| 12424 } | |
| 12425 else | |
| 12426 /* If this swap doesn't work then SWAP3 won't either. */ | |
| 12427 sp->ts_state = STATE_REP_INI; | |
| 12428 break; | |
| 12429 | |
| 12430 case STATE_UNSWAP: | |
| 12431 /* Undo the STATE_SWAP swap: "21" -> "12". */ | |
| 12432 p = fword + sp->ts_fidx; | |
| 12433 #ifdef FEAT_MBYTE | |
| 12434 if (has_mbyte) | |
| 12435 { | |
| 12436 n = MB_BYTE2LEN(*p); | |
| 12437 c = mb_ptr2char(p + n); | |
| 12438 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n); | |
| 12439 mb_char2bytes(c, p); | |
| 12440 } | |
| 12441 else | |
| 12442 #endif | |
| 12443 { | |
| 12444 c = *p; | |
| 12445 *p = p[1]; | |
| 12446 p[1] = c; | |
| 12447 } | |
| 12448 /*FALLTHROUGH*/ | |
| 12449 | |
| 12450 case STATE_SWAP3: | |
| 12451 /* Swap two bytes, skipping one: "123" -> "321". We change | |
| 12452 * "fword" here, it's changed back afterwards at STATE_UNSWAP3. */ | |
| 12453 p = fword + sp->ts_fidx; | |
| 12454 #ifdef FEAT_MBYTE | |
| 12455 if (has_mbyte) | |
| 12456 { | |
| 12457 n = mb_cptr2len(p); | |
| 12458 c = mb_ptr2char(p); | |
| 12459 fl = mb_cptr2len(p + n); | |
| 12460 c2 = mb_ptr2char(p + n); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12461 if (!soundfold && !spell_iswordp(p + n + fl, curwin)) |
| 625 | 12462 c3 = c; /* don't swap non-word char */ |
| 12463 else | |
| 12464 c3 = mb_ptr2char(p + n + fl); | |
| 12465 } | |
| 12466 else | |
| 12467 #endif | |
| 12468 { | |
| 12469 c = *p; | |
| 12470 c2 = p[1]; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12471 if (!soundfold && !spell_iswordp(p + 2, curwin)) |
| 625 | 12472 c3 = c; /* don't swap non-word char */ |
| 12473 else | |
| 12474 c3 = p[2]; | |
| 12475 } | |
| 12476 | |
| 12477 /* When characters are identical: "121" then SWAP3 result is | |
| 12478 * identical, ROT3L result is same as SWAP: "211", ROT3L result is | |
| 12479 * same as SWAP on next char: "112". Thus skip all swapping. | |
| 12480 * Also skip when c3 is NUL. | |
| 12481 * Also get here when the third character is not a word character. | |
| 12482 * Second character may any char: "a.b" -> "b.a" */ | |
| 12483 if (c == c3 || c3 == NUL) | |
| 12484 { | |
| 12485 sp->ts_state = STATE_REP_INI; | |
| 12486 break; | |
| 12487 } | |
| 12488 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 12489 { | |
| 12490 go_deeper(stack, depth, SCORE_SWAP3); | |
| 12491 #ifdef DEBUG_TRIEWALK | |
| 12492 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c", | |
| 12493 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12494 c, c3); | |
| 12495 #endif | |
| 12496 sp->ts_state = STATE_UNSWAP3; | |
| 12497 ++depth; | |
| 12498 #ifdef FEAT_MBYTE | |
| 12499 if (has_mbyte) | |
| 12500 { | |
| 12501 tl = mb_char2len(c3); | |
| 12502 mch_memmove(p, p + n + fl, tl); | |
| 12503 mb_char2bytes(c2, p + tl); | |
| 12504 mb_char2bytes(c, p + fl + tl); | |
| 12505 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl; | |
| 12506 } | |
| 12507 else | |
| 12508 #endif | |
| 12509 { | |
| 12510 p[0] = p[2]; | |
| 12511 p[2] = c; | |
| 12512 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 12513 } | |
| 12514 } | |
| 12515 else | |
| 12516 sp->ts_state = STATE_REP_INI; | |
| 12517 break; | |
| 12518 | |
| 12519 case STATE_UNSWAP3: | |
| 12520 /* Undo STATE_SWAP3: "321" -> "123" */ | |
| 12521 p = fword + sp->ts_fidx; | |
| 12522 #ifdef FEAT_MBYTE | |
| 12523 if (has_mbyte) | |
| 12524 { | |
| 12525 n = MB_BYTE2LEN(*p); | |
| 12526 c2 = mb_ptr2char(p + n); | |
| 12527 fl = MB_BYTE2LEN(p[n]); | |
| 12528 c = mb_ptr2char(p + n + fl); | |
| 12529 tl = MB_BYTE2LEN(p[n + fl]); | |
| 12530 mch_memmove(p + fl + tl, p, n); | |
| 12531 mb_char2bytes(c, p); | |
| 12532 mb_char2bytes(c2, p + tl); | |
| 12533 p = p + tl; | |
| 12534 } | |
| 12535 else | |
| 12536 #endif | |
| 12537 { | |
| 12538 c = *p; | |
| 12539 *p = p[2]; | |
| 12540 p[2] = c; | |
| 12541 ++p; | |
| 12542 } | |
| 12543 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12544 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 12545 { |
| 12546 /* Middle char is not a word char, skip the rotate. First and | |
| 12547 * third char were already checked at swap and swap3. */ | |
| 12548 sp->ts_state = STATE_REP_INI; | |
| 12549 break; | |
| 12550 } | |
| 12551 | |
| 12552 /* Rotate three characters left: "123" -> "231". We change | |
| 12553 * "fword" here, it's changed back afterwards at STATE_UNROT3L. */ | |
| 12554 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 12555 { | |
| 12556 go_deeper(stack, depth, SCORE_SWAP3); | |
| 12557 #ifdef DEBUG_TRIEWALK | |
| 12558 p = fword + sp->ts_fidx; | |
| 12559 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c", | |
| 12560 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12561 p[0], p[1], p[2]); | |
| 12562 #endif | |
| 12563 sp->ts_state = STATE_UNROT3L; | |
| 12564 ++depth; | |
| 330 | 12565 p = fword + sp->ts_fidx; |
| 12566 #ifdef FEAT_MBYTE | |
| 12567 if (has_mbyte) | |
| 12568 { | |
| 474 | 12569 n = mb_cptr2len(p); |
| 330 | 12570 c = mb_ptr2char(p); |
| 474 | 12571 fl = mb_cptr2len(p + n); |
| 625 | 12572 fl += mb_cptr2len(p + n + fl); |
| 12573 mch_memmove(p, p + n, fl); | |
| 12574 mb_char2bytes(c, p + fl); | |
| 12575 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 12576 } |
| 12577 else | |
| 12578 #endif | |
| 323 | 12579 { |
| 330 | 12580 c = *p; |
| 625 | 12581 *p = p[1]; |
| 12582 p[1] = p[2]; | |
| 12583 p[2] = c; | |
| 12584 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 12585 } | |
| 12586 } | |
| 12587 else | |
| 12588 sp->ts_state = STATE_REP_INI; | |
| 12589 break; | |
| 12590 | |
| 12591 case STATE_UNROT3L: | |
| 12592 /* Undo ROT3L: "231" -> "123" */ | |
| 12593 p = fword + sp->ts_fidx; | |
| 12594 #ifdef FEAT_MBYTE | |
| 12595 if (has_mbyte) | |
| 12596 { | |
| 12597 n = MB_BYTE2LEN(*p); | |
| 12598 n += MB_BYTE2LEN(p[n]); | |
| 12599 c = mb_ptr2char(p + n); | |
| 12600 tl = MB_BYTE2LEN(p[n]); | |
| 12601 mch_memmove(p + tl, p, n); | |
| 12602 mb_char2bytes(c, p); | |
| 12603 } | |
| 12604 else | |
| 12605 #endif | |
| 12606 { | |
| 12607 c = p[2]; | |
| 12608 p[2] = p[1]; | |
| 12609 p[1] = *p; | |
| 12610 *p = c; | |
| 12611 } | |
| 12612 | |
| 12613 /* Rotate three bytes right: "123" -> "312". We change "fword" | |
| 12614 * here, it's changed back afterwards at STATE_UNROT3R. */ | |
| 12615 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 12616 { | |
| 12617 go_deeper(stack, depth, SCORE_SWAP3); | |
| 12618 #ifdef DEBUG_TRIEWALK | |
| 12619 p = fword + sp->ts_fidx; | |
| 12620 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c", | |
| 12621 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12622 p[0], p[1], p[2]); | |
| 12623 #endif | |
| 12624 sp->ts_state = STATE_UNROT3R; | |
| 12625 ++depth; | |
| 330 | 12626 p = fword + sp->ts_fidx; |
| 12627 #ifdef FEAT_MBYTE | |
| 12628 if (has_mbyte) | |
| 12629 { | |
| 625 | 12630 n = mb_cptr2len(p); |
| 12631 n += mb_cptr2len(p + n); | |
| 330 | 12632 c = mb_ptr2char(p + n); |
| 625 | 12633 tl = mb_cptr2len(p + n); |
| 330 | 12634 mch_memmove(p + tl, p, n); |
| 12635 mb_char2bytes(c, p); | |
| 625 | 12636 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl; |
| 330 | 12637 } |
| 12638 else | |
| 12639 #endif | |
| 12640 { | |
| 12641 c = p[2]; | |
| 12642 p[2] = p[1]; | |
| 12643 p[1] = *p; | |
| 12644 *p = c; | |
| 625 | 12645 stack[depth].ts_fidxtry = sp->ts_fidx + 3; |
| 12646 } | |
| 12647 } | |
| 12648 else | |
| 12649 sp->ts_state = STATE_REP_INI; | |
| 12650 break; | |
| 12651 | |
| 12652 case STATE_UNROT3R: | |
| 12653 /* Undo ROT3R: "312" -> "123" */ | |
| 12654 p = fword + sp->ts_fidx; | |
| 12655 #ifdef FEAT_MBYTE | |
| 12656 if (has_mbyte) | |
| 12657 { | |
| 12658 c = mb_ptr2char(p); | |
| 12659 tl = MB_BYTE2LEN(*p); | |
| 12660 n = MB_BYTE2LEN(p[tl]); | |
| 12661 n += MB_BYTE2LEN(p[tl + n]); | |
| 12662 mch_memmove(p, p + tl, n); | |
| 12663 mb_char2bytes(c, p + n); | |
| 12664 } | |
| 12665 else | |
| 12666 #endif | |
| 12667 { | |
| 12668 c = *p; | |
| 12669 *p = p[1]; | |
| 12670 p[1] = p[2]; | |
| 12671 p[2] = c; | |
| 12672 } | |
| 12673 /*FALLTHROUGH*/ | |
| 12674 | |
| 12675 case STATE_REP_INI: | |
| 12676 /* Check if matching with REP items from the .aff file would work. | |
| 12677 * Quickly skip if: | |
| 12678 * - there are no REP items and we are not in the soundfold trie | |
| 12679 * - the score is going to be too high anyway | |
| 12680 * - already applied a REP item or swapped here */ | |
| 12681 if ((lp->lp_replang == NULL && !soundfold) | |
| 12682 || sp->ts_score + SCORE_REP >= su->su_maxscore | |
| 12683 || sp->ts_fidx < sp->ts_fidxtry) | |
| 12684 { | |
| 12685 sp->ts_state = STATE_FINAL; | |
| 323 | 12686 break; |
| 625 | 12687 } |
| 12688 | |
| 12689 /* Use the first byte to quickly find the first entry that may | |
| 12690 * match. If the index is -1 there is none. */ | |
| 12691 if (soundfold) | |
| 12692 sp->ts_curi = slang->sl_repsal_first[fword[sp->ts_fidx]]; | |
| 12693 else | |
| 12694 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]]; | |
| 12695 | |
| 12696 if (sp->ts_curi < 0) | |
| 12697 { | |
| 12698 sp->ts_state = STATE_FINAL; | |
| 12699 break; | |
| 12700 } | |
| 12701 | |
| 12702 sp->ts_state = STATE_REP; | |
| 12703 /*FALLTHROUGH*/ | |
| 12704 | |
| 12705 case STATE_REP: | |
| 12706 /* Try matching with REP items from the .aff file. For each match | |
| 12707 * replace the characters and check if the resulting word is | |
| 12708 * valid. */ | |
| 12709 p = fword + sp->ts_fidx; | |
| 12710 | |
| 12711 if (soundfold) | |
| 12712 gap = &slang->sl_repsal; | |
| 12713 else | |
| 503 | 12714 gap = &lp->lp_replang->sl_rep; |
| 625 | 12715 while (sp->ts_curi < gap->ga_len) |
| 12716 { | |
| 12717 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++; | |
| 12718 if (*ftp->ft_from != *p) | |
| 12719 { | |
| 12720 /* past possible matching entries */ | |
| 12721 sp->ts_curi = gap->ga_len; | |
| 323 | 12722 break; |
| 12723 } | |
| 625 | 12724 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0 |
| 12725 && TRY_DEEPER(su, stack, depth, SCORE_REP)) | |
| 12726 { | |
| 12727 go_deeper(stack, depth, SCORE_REP); | |
| 12728 #ifdef DEBUG_TRIEWALK | |
| 12729 sprintf(changename[depth], "%.*s-%s: replace %s with %s", | |
| 12730 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12731 ftp->ft_from, ftp->ft_to); | |
| 12732 #endif | |
| 12733 /* Need to undo this afterwards. */ | |
| 12734 sp->ts_state = STATE_REP_UNDO; | |
| 12735 | |
| 12736 /* Change the "from" to the "to" string. */ | |
| 12737 ++depth; | |
| 835 | 12738 fl = (int)STRLEN(ftp->ft_from); |
| 12739 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 12740 if (fl != tl) |
| 323 | 12741 { |
| 1619 | 12742 STRMOVE(p + tl, p + fl); |
| 625 | 12743 repextra += tl - fl; |
| 323 | 12744 } |
| 625 | 12745 mch_memmove(p, ftp->ft_to, tl); |
| 12746 stack[depth].ts_fidxtry = sp->ts_fidx + tl; | |
| 12747 #ifdef FEAT_MBYTE | |
| 12748 stack[depth].ts_tcharlen = 0; | |
| 12749 #endif | |
| 12750 break; | |
| 12751 } | |
| 12752 } | |
| 12753 | |
| 12754 if (sp->ts_curi >= gap->ga_len && sp->ts_state == STATE_REP) | |
| 12755 /* No (more) matches. */ | |
| 12756 sp->ts_state = STATE_FINAL; | |
| 12757 | |
| 12758 break; | |
| 12759 | |
| 12760 case STATE_REP_UNDO: | |
| 12761 /* Undo a REP replacement and continue with the next one. */ | |
| 12762 if (soundfold) | |
| 12763 gap = &slang->sl_repsal; | |
| 12764 else | |
| 12765 gap = &lp->lp_replang->sl_rep; | |
| 12766 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1; | |
| 835 | 12767 fl = (int)STRLEN(ftp->ft_from); |
| 12768 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 12769 p = fword + sp->ts_fidx; |
| 12770 if (fl != tl) | |
| 12771 { | |
| 1619 | 12772 STRMOVE(p + fl, p + tl); |
| 625 | 12773 repextra -= tl - fl; |
| 12774 } | |
| 12775 mch_memmove(p, ftp->ft_from, fl); | |
| 12776 sp->ts_state = STATE_REP; | |
| 12777 break; | |
| 12778 | |
| 12779 default: | |
| 12780 /* Did all possible states at this level, go up one level. */ | |
| 12781 --depth; | |
| 12782 | |
| 12783 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE) | |
| 12784 { | |
| 12785 /* Continue in or go back to the prefix tree. */ | |
| 12786 byts = pbyts; | |
| 12787 idxs = pidxs; | |
| 12788 } | |
| 12789 | |
| 12790 /* Don't check for CTRL-C too often, it takes time. */ | |
| 12791 if (--breakcheckcount == 0) | |
| 12792 { | |
| 12793 ui_breakcheck(); | |
| 12794 breakcheckcount = 1000; | |
| 12795 } | |
| 12796 } | |
| 12797 } | |
| 12798 } | |
| 12799 | |
| 12800 | |
| 12801 /* | |
| 12802 * Go one level deeper in the tree. | |
| 12803 */ | |
| 12804 static void | |
| 12805 go_deeper(stack, depth, score_add) | |
| 323 | 12806 trystate_T *stack; |
| 12807 int depth; | |
| 12808 int score_add; | |
| 12809 { | |
| 330 | 12810 stack[depth + 1] = stack[depth]; |
| 323 | 12811 stack[depth + 1].ts_state = STATE_START; |
| 625 | 12812 stack[depth + 1].ts_score = stack[depth].ts_score + score_add; |
| 323 | 12813 stack[depth + 1].ts_curi = 1; /* start just after length byte */ |
| 485 | 12814 stack[depth + 1].ts_flags = 0; |
| 323 | 12815 } |
| 12816 | |
| 455 | 12817 #ifdef FEAT_MBYTE |
| 12818 /* | |
| 12819 * Case-folding may change the number of bytes: Count nr of chars in | |
| 12820 * fword[flen] and return the byte length of that many chars in "word". | |
| 12821 */ | |
| 12822 static int | |
| 12823 nofold_len(fword, flen, word) | |
| 12824 char_u *fword; | |
| 12825 int flen; | |
| 12826 char_u *word; | |
| 12827 { | |
| 12828 char_u *p; | |
| 12829 int i = 0; | |
| 12830 | |
| 12831 for (p = fword; p < fword + flen; mb_ptr_adv(p)) | |
| 12832 ++i; | |
| 12833 for (p = word; i > 0; mb_ptr_adv(p)) | |
| 12834 --i; | |
| 12835 return (int)(p - word); | |
| 12836 } | |
| 12837 #endif | |
| 12838 | |
| 323 | 12839 /* |
| 12840 * "fword" is a good word with case folded. Find the matching keep-case | |
| 12841 * words and put it in "kword". | |
| 12842 * Theoretically there could be several keep-case words that result in the | |
| 12843 * same case-folded word, but we only find one... | |
| 12844 */ | |
| 12845 static void | |
| 12846 find_keepcap_word(slang, fword, kword) | |
| 12847 slang_T *slang; | |
| 12848 char_u *fword; | |
| 12849 char_u *kword; | |
| 12850 { | |
| 12851 char_u uword[MAXWLEN]; /* "fword" in upper-case */ | |
| 12852 int depth; | |
| 324 | 12853 idx_T tryidx; |
| 323 | 12854 |
| 12855 /* The following arrays are used at each depth in the tree. */ | |
| 324 | 12856 idx_T arridx[MAXWLEN]; |
| 323 | 12857 int round[MAXWLEN]; |
| 12858 int fwordidx[MAXWLEN]; | |
| 12859 int uwordidx[MAXWLEN]; | |
| 12860 int kwordlen[MAXWLEN]; | |
| 12861 | |
| 12862 int flen, ulen; | |
| 12863 int l; | |
| 12864 int len; | |
| 12865 int c; | |
| 324 | 12866 idx_T lo, hi, m; |
| 323 | 12867 char_u *p; |
| 12868 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */ | |
| 324 | 12869 idx_T *idxs = slang->sl_kidxs; /* array with indexes */ |
| 323 | 12870 |
| 12871 if (byts == NULL) | |
| 12872 { | |
| 12873 /* array is empty: "cannot happen" */ | |
| 12874 *kword = NUL; | |
| 12875 return; | |
| 12876 } | |
| 12877 | |
| 12878 /* Make an all-cap version of "fword". */ | |
| 12879 allcap_copy(fword, uword); | |
| 12880 | |
| 12881 /* | |
| 12882 * Each character needs to be tried both case-folded and upper-case. | |
| 12883 * All this gets very complicated if we keep in mind that changing case | |
| 12884 * may change the byte length of a multi-byte character... | |
| 12885 */ | |
| 12886 depth = 0; | |
| 12887 arridx[0] = 0; | |
| 12888 round[0] = 0; | |
| 12889 fwordidx[0] = 0; | |
| 12890 uwordidx[0] = 0; | |
| 12891 kwordlen[0] = 0; | |
| 12892 while (depth >= 0) | |
| 12893 { | |
| 12894 if (fword[fwordidx[depth]] == NUL) | |
| 12895 { | |
| 12896 /* We are at the end of "fword". If the tree allows a word to end | |
| 12897 * here we have found a match. */ | |
| 12898 if (byts[arridx[depth] + 1] == 0) | |
| 12899 { | |
| 12900 kword[kwordlen[depth]] = NUL; | |
| 12901 return; | |
| 12902 } | |
| 12903 | |
| 12904 /* kword is getting too long, continue one level up */ | |
| 12905 --depth; | |
| 12906 } | |
| 12907 else if (++round[depth] > 2) | |
| 12908 { | |
| 12909 /* tried both fold-case and upper-case character, continue one | |
| 12910 * level up */ | |
| 12911 --depth; | |
| 12912 } | |
| 12913 else | |
| 12914 { | |
| 12915 /* | |
| 12916 * round[depth] == 1: Try using the folded-case character. | |
| 12917 * round[depth] == 2: Try using the upper-case character. | |
| 12918 */ | |
| 12919 #ifdef FEAT_MBYTE | |
| 12920 if (has_mbyte) | |
| 12921 { | |
| 474 | 12922 flen = mb_cptr2len(fword + fwordidx[depth]); |
| 12923 ulen = mb_cptr2len(uword + uwordidx[depth]); | |
| 323 | 12924 } |
| 12925 else | |
| 12926 #endif | |
| 12927 ulen = flen = 1; | |
| 12928 if (round[depth] == 1) | |
| 12929 { | |
| 12930 p = fword + fwordidx[depth]; | |
| 12931 l = flen; | |
| 12932 } | |
| 12933 else | |
| 12934 { | |
| 12935 p = uword + uwordidx[depth]; | |
| 12936 l = ulen; | |
| 12937 } | |
| 12938 | |
| 12939 for (tryidx = arridx[depth]; l > 0; --l) | |
| 12940 { | |
| 12941 /* Perform a binary search in the list of accepted bytes. */ | |
| 12942 len = byts[tryidx++]; | |
| 12943 c = *p++; | |
| 12944 lo = tryidx; | |
| 12945 hi = tryidx + len - 1; | |
| 12946 while (lo < hi) | |
| 12947 { | |
| 12948 m = (lo + hi) / 2; | |
| 12949 if (byts[m] > c) | |
| 12950 hi = m - 1; | |
| 12951 else if (byts[m] < c) | |
| 12952 lo = m + 1; | |
| 12953 else | |
| 12954 { | |
| 12955 lo = hi = m; | |
| 12956 break; | |
| 12957 } | |
| 12958 } | |
| 12959 | |
| 12960 /* Stop if there is no matching byte. */ | |
| 12961 if (hi < lo || byts[lo] != c) | |
| 12962 break; | |
| 12963 | |
| 12964 /* Continue at the child (if there is one). */ | |
| 12965 tryidx = idxs[lo]; | |
| 12966 } | |
| 12967 | |
| 12968 if (l == 0) | |
| 12969 { | |
| 12970 /* | |
| 12971 * Found the matching char. Copy it to "kword" and go a | |
| 12972 * level deeper. | |
| 12973 */ | |
| 12974 if (round[depth] == 1) | |
| 12975 { | |
| 12976 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth], | |
| 12977 flen); | |
| 12978 kwordlen[depth + 1] = kwordlen[depth] + flen; | |
| 12979 } | |
| 12980 else | |
| 12981 { | |
| 12982 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth], | |
| 12983 ulen); | |
| 12984 kwordlen[depth + 1] = kwordlen[depth] + ulen; | |
| 12985 } | |
| 12986 fwordidx[depth + 1] = fwordidx[depth] + flen; | |
| 12987 uwordidx[depth + 1] = uwordidx[depth] + ulen; | |
| 12988 | |
| 12989 ++depth; | |
| 12990 arridx[depth] = tryidx; | |
| 12991 round[depth] = 0; | |
| 12992 } | |
| 12993 } | |
| 12994 } | |
| 12995 | |
| 12996 /* Didn't find it: "cannot happen". */ | |
| 12997 *kword = NUL; | |
| 12998 } | |
| 12999 | |
| 13000 /* | |
| 344 | 13001 * Compute the sound-a-like score for suggestions in su->su_ga and add them to |
| 13002 * su->su_sga. | |
| 13003 */ | |
| 13004 static void | |
| 13005 score_comp_sal(su) | |
| 13006 suginfo_T *su; | |
| 13007 { | |
| 13008 langp_T *lp; | |
| 13009 char_u badsound[MAXWLEN]; | |
| 13010 int i; | |
| 13011 suggest_T *stp; | |
| 13012 suggest_T *sstp; | |
| 13013 int score; | |
| 500 | 13014 int lpi; |
| 344 | 13015 |
| 13016 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL) | |
| 13017 return; | |
| 13018 | |
| 13019 /* Use the sound-folding of the first language that supports it. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13020 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13021 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13022 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 13023 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 13024 { | |
| 13025 /* soundfold the bad word */ | |
| 375 | 13026 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound); |
| 344 | 13027 |
| 13028 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 13029 { | |
| 13030 stp = &SUG(su->su_ga, i); | |
| 13031 | |
| 351 | 13032 /* Case-fold the suggested word, sound-fold it and compute the |
| 13033 * sound-a-like score. */ | |
| 13034 score = stp_sal_score(stp, su, lp->lp_slang, badsound); | |
| 344 | 13035 if (score < SCORE_MAXMAX) |
| 13036 { | |
| 13037 /* Add the suggestion. */ | |
| 13038 sstp = &SUG(su->su_sga, su->su_sga.ga_len); | |
| 13039 sstp->st_word = vim_strsave(stp->st_word); | |
| 13040 if (sstp->st_word != NULL) | |
| 13041 { | |
| 625 | 13042 sstp->st_wordlen = stp->st_wordlen; |
| 344 | 13043 sstp->st_score = score; |
| 13044 sstp->st_altscore = 0; | |
| 13045 sstp->st_orglen = stp->st_orglen; | |
| 13046 ++su->su_sga.ga_len; | |
| 13047 } | |
| 13048 } | |
| 13049 } | |
| 13050 break; | |
| 13051 } | |
| 500 | 13052 } |
| 344 | 13053 } |
| 13054 | |
| 13055 /* | |
| 13056 * Combine the list of suggestions in su->su_ga and su->su_sga. | |
| 4352 | 13057 * They are entwined. |
| 344 | 13058 */ |
| 13059 static void | |
| 13060 score_combine(su) | |
| 13061 suginfo_T *su; | |
| 13062 { | |
| 13063 int i; | |
| 13064 int j; | |
| 13065 garray_T ga; | |
| 13066 garray_T *gap; | |
| 13067 langp_T *lp; | |
| 13068 suggest_T *stp; | |
| 13069 char_u *p; | |
| 13070 char_u badsound[MAXWLEN]; | |
| 13071 int round; | |
| 500 | 13072 int lpi; |
| 625 | 13073 slang_T *slang = NULL; |
| 344 | 13074 |
| 13075 /* Add the alternate score to su_ga. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13076 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13077 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13078 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 13079 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 13080 { | |
| 13081 /* soundfold the bad word */ | |
| 625 | 13082 slang = lp->lp_slang; |
| 13083 spell_soundfold(slang, su->su_fbadword, TRUE, badsound); | |
| 344 | 13084 |
| 13085 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 13086 { | |
| 13087 stp = &SUG(su->su_ga, i); | |
| 625 | 13088 stp->st_altscore = stp_sal_score(stp, su, slang, badsound); |
| 344 | 13089 if (stp->st_altscore == SCORE_MAXMAX) |
| 13090 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4; | |
| 13091 else | |
| 13092 stp->st_score = (stp->st_score * 3 | |
| 13093 + stp->st_altscore) / 4; | |
| 13094 stp->st_salscore = FALSE; | |
| 13095 } | |
| 13096 break; | |
| 13097 } | |
| 13098 } | |
| 13099 | |
| 844 | 13100 if (slang == NULL) /* Using "double" without sound folding. */ |
| 13101 { | |
| 13102 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, | |
| 13103 su->su_maxcount); | |
| 625 | 13104 return; |
| 844 | 13105 } |
| 625 | 13106 |
| 344 | 13107 /* Add the alternate score to su_sga. */ |
| 13108 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 13109 { | |
| 13110 stp = &SUG(su->su_sga, i); | |
| 625 | 13111 stp->st_altscore = spell_edit_score(slang, |
| 13112 su->su_badword, stp->st_word); | |
| 344 | 13113 if (stp->st_score == SCORE_MAXMAX) |
| 13114 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8; | |
| 13115 else | |
| 13116 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8; | |
| 13117 stp->st_salscore = TRUE; | |
| 13118 } | |
| 13119 | |
| 625 | 13120 /* Remove bad suggestions, sort the suggestions and truncate at "maxcount" |
| 13121 * for both lists. */ | |
| 13122 check_suggestions(su, &su->su_ga); | |
| 344 | 13123 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 625 | 13124 check_suggestions(su, &su->su_sga); |
| 344 | 13125 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount); |
| 13126 | |
| 13127 ga_init2(&ga, (int)sizeof(suginfo_T), 1); | |
| 13128 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL) | |
| 13129 return; | |
| 13130 | |
| 13131 stp = &SUG(ga, 0); | |
| 13132 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i) | |
| 13133 { | |
| 13134 /* round 1: get a suggestion from su_ga | |
| 13135 * round 2: get a suggestion from su_sga */ | |
| 13136 for (round = 1; round <= 2; ++round) | |
| 13137 { | |
| 13138 gap = round == 1 ? &su->su_ga : &su->su_sga; | |
| 13139 if (i < gap->ga_len) | |
| 13140 { | |
| 13141 /* Don't add a word if it's already there. */ | |
| 13142 p = SUG(*gap, i).st_word; | |
| 13143 for (j = 0; j < ga.ga_len; ++j) | |
| 13144 if (STRCMP(stp[j].st_word, p) == 0) | |
| 13145 break; | |
| 13146 if (j == ga.ga_len) | |
| 13147 stp[ga.ga_len++] = SUG(*gap, i); | |
| 13148 else | |
| 13149 vim_free(p); | |
| 13150 } | |
| 13151 } | |
| 13152 } | |
| 13153 | |
| 13154 ga_clear(&su->su_ga); | |
| 13155 ga_clear(&su->su_sga); | |
| 13156 | |
| 13157 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 13158 if (ga.ga_len > su->su_maxcount) | |
| 13159 { | |
| 13160 for (i = su->su_maxcount; i < ga.ga_len; ++i) | |
| 13161 vim_free(stp[i].st_word); | |
| 13162 ga.ga_len = su->su_maxcount; | |
| 13163 } | |
| 13164 | |
| 13165 su->su_ga = ga; | |
| 13166 } | |
| 13167 | |
| 13168 /* | |
| 351 | 13169 * For the goodword in "stp" compute the soundalike score compared to the |
| 13170 * badword. | |
| 13171 */ | |
| 13172 static int | |
| 13173 stp_sal_score(stp, su, slang, badsound) | |
| 13174 suggest_T *stp; | |
| 13175 suginfo_T *su; | |
| 13176 slang_T *slang; | |
| 13177 char_u *badsound; /* sound-folded badword */ | |
| 13178 { | |
| 13179 char_u *p; | |
| 534 | 13180 char_u *pbad; |
| 13181 char_u *pgood; | |
| 351 | 13182 char_u badsound2[MAXWLEN]; |
| 13183 char_u fword[MAXWLEN]; | |
| 13184 char_u goodsound[MAXWLEN]; | |
| 534 | 13185 char_u goodword[MAXWLEN]; |
| 13186 int lendiff; | |
| 13187 | |
| 13188 lendiff = (int)(su->su_badlen - stp->st_orglen); | |
| 13189 if (lendiff >= 0) | |
| 13190 pbad = badsound; | |
| 351 | 13191 else |
| 13192 { | |
| 13193 /* soundfold the bad word with more characters following */ | |
| 13194 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN); | |
| 13195 | |
| 13196 /* When joining two words the sound often changes a lot. E.g., "t he" | |
| 13197 * sounds like "t h" while "the" sounds like "@". Avoid that by | |
| 13198 * removing the space. Don't do it when the good word also contains a | |
| 13199 * space. */ | |
| 13200 if (vim_iswhite(su->su_badptr[su->su_badlen]) | |
| 13201 && *skiptowhite(stp->st_word) == NUL) | |
| 13202 for (p = fword; *(p = skiptowhite(p)) != NUL; ) | |
| 1619 | 13203 STRMOVE(p, p + 1); |
| 351 | 13204 |
| 375 | 13205 spell_soundfold(slang, fword, TRUE, badsound2); |
| 534 | 13206 pbad = badsound2; |
| 13207 } | |
| 13208 | |
| 2768 | 13209 if (lendiff > 0 && stp->st_wordlen + lendiff < MAXWLEN) |
| 534 | 13210 { |
| 13211 /* Add part of the bad word to the good word, so that we soundfold | |
| 13212 * what replaces the bad word. */ | |
| 13213 STRCPY(goodword, stp->st_word); | |
| 625 | 13214 vim_strncpy(goodword + stp->st_wordlen, |
| 13215 su->su_badptr + su->su_badlen - lendiff, lendiff); | |
| 534 | 13216 pgood = goodword; |
| 13217 } | |
| 13218 else | |
| 13219 pgood = stp->st_word; | |
| 351 | 13220 |
| 375 | 13221 /* Sound-fold the word and compute the score for the difference. */ |
| 534 | 13222 spell_soundfold(slang, pgood, FALSE, goodsound); |
| 13223 | |
| 13224 return soundalike_score(goodsound, pbad); | |
| 351 | 13225 } |
| 13226 | |
| 625 | 13227 /* structure used to store soundfolded words that add_sound_suggest() has |
| 13228 * handled already. */ | |
| 13229 typedef struct | |
| 13230 { | |
| 13231 short sft_score; /* lowest score used */ | |
| 13232 char_u sft_word[1]; /* soundfolded word, actually longer */ | |
| 13233 } sftword_T; | |
| 13234 | |
| 13235 static sftword_T dumsft; | |
| 13236 #define HIKEY2SFT(p) ((sftword_T *)(p - (dumsft.sft_word - (char_u *)&dumsft))) | |
| 13237 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key) | |
| 13238 | |
| 13239 /* | |
| 13240 * Prepare for calling suggest_try_soundalike(). | |
| 13241 */ | |
| 13242 static void | |
| 13243 suggest_try_soundalike_prep() | |
| 13244 { | |
| 13245 langp_T *lp; | |
| 13246 int lpi; | |
| 13247 slang_T *slang; | |
| 13248 | |
| 13249 /* Do this for all languages that support sound folding and for which a | |
| 13250 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13251 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13252 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13253 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 13254 slang = lp->lp_slang; |
| 13255 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 13256 /* prepare the hashtable used by add_sound_suggest() */ | |
| 13257 hash_init(&slang->sl_sounddone); | |
| 13258 } | |
| 13259 } | |
| 13260 | |
| 351 | 13261 /* |
| 323 | 13262 * Find suggestions by comparing the word in a sound-a-like form. |
| 503 | 13263 * Note: This doesn't support postponed prefixes. |
| 323 | 13264 */ |
| 13265 static void | |
| 346 | 13266 suggest_try_soundalike(su) |
| 323 | 13267 suginfo_T *su; |
| 13268 { | |
| 13269 char_u salword[MAXWLEN]; | |
| 13270 langp_T *lp; | |
| 500 | 13271 int lpi; |
| 503 | 13272 slang_T *slang; |
| 344 | 13273 |
| 625 | 13274 /* Do this for all languages that support sound folding and for which a |
| 13275 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13276 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13277 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13278 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 13279 slang = lp->lp_slang; |
| 13280 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 13281 { | |
| 13282 /* soundfold the bad word */ | |
| 13283 spell_soundfold(slang, su->su_fbadword, TRUE, salword); | |
| 13284 | |
| 13285 /* try all kinds of inserts/deletes/swaps/etc. */ | |
| 13286 /* TODO: also soundfold the next words, so that we can try joining | |
| 13287 * and splitting */ | |
| 13288 suggest_trie_walk(su, lp, salword, TRUE); | |
| 13289 } | |
| 13290 } | |
| 13291 } | |
| 13292 | |
| 13293 /* | |
| 13294 * Finish up after calling suggest_try_soundalike(). | |
| 13295 */ | |
| 13296 static void | |
| 13297 suggest_try_soundalike_finish() | |
| 13298 { | |
| 13299 langp_T *lp; | |
| 13300 int lpi; | |
| 13301 slang_T *slang; | |
| 13302 int todo; | |
| 13303 hashitem_T *hi; | |
| 13304 | |
| 13305 /* Do this for all languages that support sound folding and for which a | |
| 13306 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13307 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13308 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13309 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 503 | 13310 slang = lp->lp_slang; |
| 625 | 13311 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) |
| 13312 { | |
| 13313 /* Free the info about handled words. */ | |
| 835 | 13314 todo = (int)slang->sl_sounddone.ht_used; |
| 625 | 13315 for (hi = slang->sl_sounddone.ht_array; todo > 0; ++hi) |
| 13316 if (!HASHITEM_EMPTY(hi)) | |
| 13317 { | |
| 13318 vim_free(HI2SFT(hi)); | |
| 13319 --todo; | |
| 13320 } | |
| 1087 | 13321 |
| 13322 /* Clear the hashtable, it may also be used by another region. */ | |
| 625 | 13323 hash_clear(&slang->sl_sounddone); |
| 1087 | 13324 hash_init(&slang->sl_sounddone); |
| 625 | 13325 } |
| 13326 } | |
| 13327 } | |
| 13328 | |
| 13329 /* | |
| 13330 * A match with a soundfolded word is found. Add the good word(s) that | |
| 13331 * produce this soundfolded word. | |
| 13332 */ | |
| 13333 static void | |
| 13334 add_sound_suggest(su, goodword, score, lp) | |
| 13335 suginfo_T *su; | |
| 13336 char_u *goodword; | |
| 13337 int score; /* soundfold score */ | |
| 13338 langp_T *lp; | |
| 13339 { | |
| 13340 slang_T *slang = lp->lp_slang; /* language for sound folding */ | |
| 13341 int sfwordnr; | |
| 13342 char_u *nrline; | |
| 13343 int orgnr; | |
| 13344 char_u theword[MAXWLEN]; | |
| 13345 int i; | |
| 13346 int wlen; | |
| 13347 char_u *byts; | |
| 13348 idx_T *idxs; | |
| 13349 int n; | |
| 13350 int wordcount; | |
| 13351 int wc; | |
| 13352 int goodscore; | |
| 13353 hash_T hash; | |
| 13354 hashitem_T *hi; | |
| 13355 sftword_T *sft; | |
| 13356 int bc, gc; | |
| 13357 int limit; | |
| 13358 | |
| 13359 /* | |
| 13360 * It's very well possible that the same soundfold word is found several | |
| 13361 * times with different scores. Since the following is quite slow only do | |
| 13362 * the words that have a better score than before. Use a hashtable to | |
| 13363 * remember the words that have been done. | |
| 13364 */ | |
| 13365 hash = hash_hash(goodword); | |
| 13366 hi = hash_lookup(&slang->sl_sounddone, goodword, hash); | |
| 13367 if (HASHITEM_EMPTY(hi)) | |
| 13368 { | |
| 844 | 13369 sft = (sftword_T *)alloc((unsigned)(sizeof(sftword_T) |
| 13370 + STRLEN(goodword))); | |
| 625 | 13371 if (sft != NULL) |
| 13372 { | |
| 13373 sft->sft_score = score; | |
| 13374 STRCPY(sft->sft_word, goodword); | |
| 13375 hash_add_item(&slang->sl_sounddone, hi, sft->sft_word, hash); | |
| 13376 } | |
| 13377 } | |
| 13378 else | |
| 13379 { | |
| 13380 sft = HI2SFT(hi); | |
| 13381 if (score >= sft->sft_score) | |
| 13382 return; | |
| 13383 sft->sft_score = score; | |
| 13384 } | |
| 13385 | |
| 13386 /* | |
| 13387 * Find the word nr in the soundfold tree. | |
| 13388 */ | |
| 13389 sfwordnr = soundfold_find(slang, goodword); | |
| 13390 if (sfwordnr < 0) | |
| 13391 { | |
| 13392 EMSG2(_(e_intern2), "add_sound_suggest()"); | |
| 13393 return; | |
| 13394 } | |
| 13395 | |
| 13396 /* | |
| 13397 * go over the list of good words that produce this soundfold word | |
| 13398 */ | |
| 13399 nrline = ml_get_buf(slang->sl_sugbuf, (linenr_T)(sfwordnr + 1), FALSE); | |
| 13400 orgnr = 0; | |
| 13401 while (*nrline != NUL) | |
| 13402 { | |
| 13403 /* The wordnr was stored in a minimal nr of bytes as an offset to the | |
| 13404 * previous wordnr. */ | |
| 13405 orgnr += bytes2offset(&nrline); | |
| 13406 | |
| 13407 byts = slang->sl_fbyts; | |
| 13408 idxs = slang->sl_fidxs; | |
| 13409 | |
| 13410 /* Lookup the word "orgnr" one of the two tries. */ | |
| 13411 n = 0; | |
| 13412 wordcount = 0; | |
| 5506 | 13413 for (wlen = 0; wlen < MAXWLEN - 3; ++wlen) |
| 625 | 13414 { |
| 13415 i = 1; | |
| 13416 if (wordcount == orgnr && byts[n + 1] == NUL) | |
| 13417 break; /* found end of word */ | |
| 13418 | |
| 13419 if (byts[n + 1] == NUL) | |
| 13420 ++wordcount; | |
| 13421 | |
| 13422 /* skip over the NUL bytes */ | |
| 13423 for ( ; byts[n + i] == NUL; ++i) | |
| 13424 if (i > byts[n]) /* safety check */ | |
| 13425 { | |
| 13426 STRCPY(theword + wlen, "BAD"); | |
| 5506 | 13427 wlen += 3; |
| 625 | 13428 goto badword; |
| 13429 } | |
| 13430 | |
| 13431 /* One of the siblings must have the word. */ | |
| 13432 for ( ; i < byts[n]; ++i) | |
| 13433 { | |
| 13434 wc = idxs[idxs[n + i]]; /* nr of words under this byte */ | |
| 13435 if (wordcount + wc > orgnr) | |
| 13436 break; | |
| 13437 wordcount += wc; | |
| 13438 } | |
| 13439 | |
| 5506 | 13440 theword[wlen] = byts[n + i]; |
| 625 | 13441 n = idxs[n + i]; |
| 13442 } | |
| 13443 badword: | |
| 13444 theword[wlen] = NUL; | |
| 13445 | |
| 13446 /* Go over the possible flags and regions. */ | |
| 13447 for (; i <= byts[n] && byts[n + i] == NUL; ++i) | |
| 13448 { | |
| 13449 char_u cword[MAXWLEN]; | |
| 13450 char_u *p; | |
| 13451 int flags = (int)idxs[n + i]; | |
| 13452 | |
| 698 | 13453 /* Skip words with the NOSUGGEST flag */ |
| 13454 if (flags & WF_NOSUGGEST) | |
| 13455 continue; | |
| 13456 | |
| 625 | 13457 if (flags & WF_KEEPCAP) |
| 13458 { | |
| 13459 /* Must find the word in the keep-case tree. */ | |
| 13460 find_keepcap_word(slang, theword, cword); | |
| 13461 p = cword; | |
| 13462 } | |
| 13463 else | |
| 13464 { | |
| 13465 flags |= su->su_badflags; | |
| 13466 if ((flags & WF_CAPMASK) != 0) | |
| 13467 { | |
| 13468 /* Need to fix case according to "flags". */ | |
| 13469 make_case_word(theword, cword, flags); | |
| 13470 p = cword; | |
| 323 | 13471 } |
| 13472 else | |
| 625 | 13473 p = theword; |
| 13474 } | |
| 13475 | |
| 13476 /* Add the suggestion. */ | |
| 13477 if (sps_flags & SPS_DOUBLE) | |
| 13478 { | |
| 13479 /* Add the suggestion if the score isn't too bad. */ | |
| 13480 if (score <= su->su_maxscore) | |
| 13481 add_suggestion(su, &su->su_sga, p, su->su_badlen, | |
| 13482 score, 0, FALSE, slang, FALSE); | |
| 13483 } | |
| 13484 else | |
| 13485 { | |
| 13486 /* Add a penalty for words in another region. */ | |
| 13487 if ((flags & WF_REGION) | |
| 13488 && (((unsigned)flags >> 16) & lp->lp_region) == 0) | |
| 13489 goodscore = SCORE_REGION; | |
| 13490 else | |
| 13491 goodscore = 0; | |
| 13492 | |
| 13493 /* Add a small penalty for changing the first letter from | |
| 13494 * lower to upper case. Helps for "tath" -> "Kath", which is | |
| 4352 | 13495 * less common than "tath" -> "path". Don't do it when the |
| 625 | 13496 * letter is the same, that has already been counted. */ |
| 13497 gc = PTR2CHAR(p); | |
| 13498 if (SPELL_ISUPPER(gc)) | |
| 13499 { | |
| 13500 bc = PTR2CHAR(su->su_badword); | |
| 13501 if (!SPELL_ISUPPER(bc) | |
| 13502 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc)) | |
| 13503 goodscore += SCORE_ICASE / 2; | |
| 13504 } | |
| 13505 | |
| 13506 /* Compute the score for the good word. This only does letter | |
| 13507 * insert/delete/swap/replace. REP items are not considered, | |
| 13508 * which may make the score a bit higher. | |
| 13509 * Use a limit for the score to make it work faster. Use | |
| 13510 * MAXSCORE(), because RESCORE() will change the score. | |
| 13511 * If the limit is very high then the iterative method is | |
| 13512 * inefficient, using an array is quicker. */ | |
| 13513 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score); | |
| 13514 if (limit > SCORE_LIMITMAX) | |
| 13515 goodscore += spell_edit_score(slang, su->su_badword, p); | |
| 13516 else | |
| 13517 goodscore += spell_edit_score_limit(slang, su->su_badword, | |
| 13518 p, limit); | |
| 13519 | |
| 13520 /* When going over the limit don't bother to do the rest. */ | |
| 13521 if (goodscore < SCORE_MAXMAX) | |
| 13522 { | |
| 13523 /* Give a bonus to words seen before. */ | |
| 13524 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE); | |
| 13525 | |
| 13526 /* Add the suggestion if the score isn't too bad. */ | |
| 13527 goodscore = RESCORE(goodscore, score); | |
| 13528 if (goodscore <= su->su_sfmaxscore) | |
| 13529 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 13530 goodscore, score, TRUE, slang, TRUE); | |
| 13531 } | |
| 13532 } | |
| 13533 } | |
| 13534 /* smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr); */ | |
| 13535 } | |
| 13536 } | |
| 13537 | |
| 13538 /* | |
| 13539 * Find word "word" in fold-case tree for "slang" and return the word number. | |
| 13540 */ | |
| 13541 static int | |
| 13542 soundfold_find(slang, word) | |
| 13543 slang_T *slang; | |
| 13544 char_u *word; | |
| 13545 { | |
| 13546 idx_T arridx = 0; | |
| 13547 int len; | |
| 13548 int wlen = 0; | |
| 13549 int c; | |
| 13550 char_u *ptr = word; | |
| 13551 char_u *byts; | |
| 13552 idx_T *idxs; | |
| 13553 int wordnr = 0; | |
| 13554 | |
| 13555 byts = slang->sl_sbyts; | |
| 13556 idxs = slang->sl_sidxs; | |
| 13557 | |
| 13558 for (;;) | |
| 13559 { | |
| 13560 /* First byte is the number of possible bytes. */ | |
| 13561 len = byts[arridx++]; | |
| 13562 | |
| 13563 /* If the first possible byte is a zero the word could end here. | |
| 13564 * If the word ends we found the word. If not skip the NUL bytes. */ | |
| 13565 c = ptr[wlen]; | |
| 13566 if (byts[arridx] == NUL) | |
| 13567 { | |
| 13568 if (c == NUL) | |
| 13569 break; | |
| 13570 | |
| 13571 /* Skip over the zeros, there can be several. */ | |
| 13572 while (len > 0 && byts[arridx] == NUL) | |
| 13573 { | |
| 13574 ++arridx; | |
| 13575 --len; | |
| 13576 } | |
| 13577 if (len == 0) | |
| 13578 return -1; /* no children, word should have ended here */ | |
| 13579 ++wordnr; | |
| 13580 } | |
| 13581 | |
| 13582 /* If the word ends we didn't find it. */ | |
| 13583 if (c == NUL) | |
| 13584 return -1; | |
| 13585 | |
| 13586 /* Perform a binary search in the list of accepted bytes. */ | |
| 13587 if (c == TAB) /* <Tab> is handled like <Space> */ | |
| 13588 c = ' '; | |
| 13589 while (byts[arridx] < c) | |
| 13590 { | |
| 13591 /* The word count is in the first idxs[] entry of the child. */ | |
| 13592 wordnr += idxs[idxs[arridx]]; | |
| 13593 ++arridx; | |
| 13594 if (--len == 0) /* end of the bytes, didn't find it */ | |
| 13595 return -1; | |
| 13596 } | |
| 13597 if (byts[arridx] != c) /* didn't find the byte */ | |
| 13598 return -1; | |
| 13599 | |
| 13600 /* Continue at the child (if there is one). */ | |
| 13601 arridx = idxs[arridx]; | |
| 13602 ++wlen; | |
| 13603 | |
| 13604 /* One space in the good word may stand for several spaces in the | |
| 13605 * checked word. */ | |
| 13606 if (c == ' ') | |
| 13607 while (ptr[wlen] == ' ' || ptr[wlen] == TAB) | |
| 13608 ++wlen; | |
| 13609 } | |
| 13610 | |
| 13611 return wordnr; | |
| 323 | 13612 } |
| 13613 | |
| 13614 /* | |
| 324 | 13615 * Copy "fword" to "cword", fixing case according to "flags". |
| 323 | 13616 */ |
| 13617 static void | |
| 13618 make_case_word(fword, cword, flags) | |
| 13619 char_u *fword; | |
| 13620 char_u *cword; | |
| 13621 int flags; | |
| 13622 { | |
| 13623 if (flags & WF_ALLCAP) | |
| 13624 /* Make it all upper-case */ | |
| 13625 allcap_copy(fword, cword); | |
| 13626 else if (flags & WF_ONECAP) | |
| 13627 /* Make the first letter upper-case */ | |
| 324 | 13628 onecap_copy(fword, cword, TRUE); |
| 323 | 13629 else |
| 13630 /* Use goodword as-is. */ | |
| 13631 STRCPY(cword, fword); | |
| 13632 } | |
| 13633 | |
| 330 | 13634 /* |
| 13635 * Use map string "map" for languages "lp". | |
| 13636 */ | |
| 13637 static void | |
| 13638 set_map_str(lp, map) | |
| 13639 slang_T *lp; | |
| 13640 char_u *map; | |
| 13641 { | |
| 13642 char_u *p; | |
| 13643 int headc = 0; | |
| 13644 int c; | |
| 13645 int i; | |
| 13646 | |
| 13647 if (*map == NUL) | |
| 13648 { | |
| 13649 lp->sl_has_map = FALSE; | |
| 13650 return; | |
| 13651 } | |
| 13652 lp->sl_has_map = TRUE; | |
| 13653 | |
| 625 | 13654 /* Init the array and hash tables empty. */ |
| 330 | 13655 for (i = 0; i < 256; ++i) |
| 13656 lp->sl_map_array[i] = 0; | |
| 13657 #ifdef FEAT_MBYTE | |
| 13658 hash_init(&lp->sl_map_hash); | |
| 13659 #endif | |
| 13660 | |
| 13661 /* | |
| 13662 * The similar characters are stored separated with slashes: | |
| 13663 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and | |
| 13664 * before the same slash. For characters above 255 sl_map_hash is used. | |
| 13665 */ | |
| 13666 for (p = map; *p != NUL; ) | |
| 13667 { | |
| 13668 #ifdef FEAT_MBYTE | |
| 474 | 13669 c = mb_cptr2char_adv(&p); |
| 330 | 13670 #else |
| 13671 c = *p++; | |
| 13672 #endif | |
| 13673 if (c == '/') | |
| 13674 headc = 0; | |
| 13675 else | |
| 13676 { | |
| 13677 if (headc == 0) | |
| 13678 headc = c; | |
| 13679 | |
| 13680 #ifdef FEAT_MBYTE | |
| 13681 /* Characters above 255 don't fit in sl_map_array[], put them in | |
| 13682 * the hash table. Each entry is the char, a NUL the headchar and | |
| 13683 * a NUL. */ | |
| 13684 if (c >= 256) | |
| 13685 { | |
| 13686 int cl = mb_char2len(c); | |
| 13687 int headcl = mb_char2len(headc); | |
| 13688 char_u *b; | |
| 13689 hash_T hash; | |
| 13690 hashitem_T *hi; | |
| 13691 | |
| 13692 b = alloc((unsigned)(cl + headcl + 2)); | |
| 13693 if (b == NULL) | |
| 13694 return; | |
| 13695 mb_char2bytes(c, b); | |
| 13696 b[cl] = NUL; | |
| 13697 mb_char2bytes(headc, b + cl + 1); | |
| 13698 b[cl + 1 + headcl] = NUL; | |
| 13699 hash = hash_hash(b); | |
| 13700 hi = hash_lookup(&lp->sl_map_hash, b, hash); | |
| 13701 if (HASHITEM_EMPTY(hi)) | |
| 13702 hash_add_item(&lp->sl_map_hash, hi, b, hash); | |
| 13703 else | |
| 13704 { | |
| 13705 /* This should have been checked when generating the .spl | |
| 13706 * file. */ | |
| 626 | 13707 EMSG(_("E783: duplicate char in MAP entry")); |
| 330 | 13708 vim_free(b); |
| 13709 } | |
| 13710 } | |
| 13711 else | |
| 13712 #endif | |
| 13713 lp->sl_map_array[c] = headc; | |
| 13714 } | |
| 13715 } | |
| 13716 } | |
| 13717 | |
| 323 | 13718 /* |
| 13719 * Return TRUE if "c1" and "c2" are similar characters according to the MAP | |
| 13720 * lines in the .aff file. | |
| 13721 */ | |
| 13722 static int | |
| 13723 similar_chars(slang, c1, c2) | |
| 13724 slang_T *slang; | |
| 13725 int c1; | |
| 13726 int c2; | |
| 13727 { | |
| 330 | 13728 int m1, m2; |
| 13729 #ifdef FEAT_MBYTE | |
| 3549 | 13730 char_u buf[MB_MAXBYTES + 1]; |
| 330 | 13731 hashitem_T *hi; |
| 13732 | |
| 13733 if (c1 >= 256) | |
| 13734 { | |
| 13735 buf[mb_char2bytes(c1, buf)] = 0; | |
| 13736 hi = hash_find(&slang->sl_map_hash, buf); | |
| 13737 if (HASHITEM_EMPTY(hi)) | |
| 13738 m1 = 0; | |
| 13739 else | |
| 13740 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 13741 } | |
| 13742 else | |
| 13743 #endif | |
| 13744 m1 = slang->sl_map_array[c1]; | |
| 13745 if (m1 == 0) | |
| 323 | 13746 return FALSE; |
| 330 | 13747 |
| 13748 | |
| 13749 #ifdef FEAT_MBYTE | |
| 13750 if (c2 >= 256) | |
| 13751 { | |
| 13752 buf[mb_char2bytes(c2, buf)] = 0; | |
| 13753 hi = hash_find(&slang->sl_map_hash, buf); | |
| 13754 if (HASHITEM_EMPTY(hi)) | |
| 13755 m2 = 0; | |
| 13756 else | |
| 13757 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 13758 } | |
| 13759 else | |
| 13760 #endif | |
| 13761 m2 = slang->sl_map_array[c2]; | |
| 13762 | |
| 13763 return m1 == m2; | |
| 323 | 13764 } |
| 13765 | |
| 13766 /* | |
| 13767 * Add a suggestion to the list of suggestions. | |
| 625 | 13768 * For a suggestion that is already in the list the lowest score is remembered. |
| 13769 */ | |
| 13770 static void | |
| 13771 add_suggestion(su, gap, goodword, badlenarg, score, altscore, had_bonus, | |
| 13772 slang, maxsf) | |
| 323 | 13773 suginfo_T *su; |
| 625 | 13774 garray_T *gap; /* either su_ga or su_sga */ |
| 323 | 13775 char_u *goodword; |
| 534 | 13776 int badlenarg; /* len of bad word replaced with "goodword" */ |
| 324 | 13777 int score; |
| 351 | 13778 int altscore; |
| 344 | 13779 int had_bonus; /* value for st_had_bonus */ |
| 503 | 13780 slang_T *slang; /* language for sound folding */ |
| 625 | 13781 int maxsf; /* su_maxscore applies to soundfold score, |
| 13782 su_sfmaxscore to the total score. */ | |
| 13783 { | |
| 13784 int goodlen; /* len of goodword changed */ | |
| 13785 int badlen; /* len of bad word changed */ | |
| 323 | 13786 suggest_T *stp; |
| 534 | 13787 suggest_T new_sug; |
| 323 | 13788 int i; |
| 534 | 13789 char_u *pgood, *pbad; |
| 532 | 13790 |
| 534 | 13791 /* Minimize "badlen" for consistency. Avoids that changing "the the" to |
| 13792 * "thee the" is added next to changing the first "the" the "thee". */ | |
| 13793 pgood = goodword + STRLEN(goodword); | |
| 625 | 13794 pbad = su->su_badptr + badlenarg; |
| 13795 for (;;) | |
| 13796 { | |
| 835 | 13797 goodlen = (int)(pgood - goodword); |
| 13798 badlen = (int)(pbad - su->su_badptr); | |
| 625 | 13799 if (goodlen <= 0 || badlen <= 0) |
| 13800 break; | |
| 534 | 13801 mb_ptr_back(goodword, pgood); |
| 13802 mb_ptr_back(su->su_badptr, pbad); | |
| 13803 #ifdef FEAT_MBYTE | |
| 13804 if (has_mbyte) | |
| 13805 { | |
| 13806 if (mb_ptr2char(pgood) != mb_ptr2char(pbad)) | |
| 13807 break; | |
| 346 | 13808 } |
| 13809 else | |
| 534 | 13810 #endif |
| 13811 if (*pgood != *pbad) | |
| 13812 break; | |
| 625 | 13813 } |
| 13814 | |
| 534 | 13815 if (badlen == 0 && goodlen == 0) |
| 13816 /* goodword doesn't change anything; may happen for "the the" changing | |
| 13817 * the first "the" to itself. */ | |
| 13818 return; | |
| 346 | 13819 |
| 944 | 13820 if (gap->ga_len == 0) |
| 13821 i = -1; | |
| 13822 else | |
| 13823 { | |
| 13824 /* Check if the word is already there. Also check the length that is | |
| 13825 * being replaced "thes," -> "these" is a different suggestion from | |
| 13826 * "thes" -> "these". */ | |
| 13827 stp = &SUG(*gap, 0); | |
| 13828 for (i = gap->ga_len; --i >= 0; ++stp) | |
| 13829 if (stp->st_wordlen == goodlen | |
| 13830 && stp->st_orglen == badlen | |
| 13831 && STRNCMP(stp->st_word, goodword, goodlen) == 0) | |
| 13832 { | |
| 13833 /* | |
| 13834 * Found it. Remember the word with the lowest score. | |
| 13835 */ | |
| 13836 if (stp->st_slang == NULL) | |
| 13837 stp->st_slang = slang; | |
| 13838 | |
| 13839 new_sug.st_score = score; | |
| 13840 new_sug.st_altscore = altscore; | |
| 13841 new_sug.st_had_bonus = had_bonus; | |
| 13842 | |
| 13843 if (stp->st_had_bonus != had_bonus) | |
| 13844 { | |
| 13845 /* Only one of the two had the soundalike score computed. | |
| 13846 * Need to do that for the other one now, otherwise the | |
| 13847 * scores can't be compared. This happens because | |
| 13848 * suggest_try_change() doesn't compute the soundalike | |
| 13849 * word to keep it fast, while some special methods set | |
| 13850 * the soundalike score to zero. */ | |
| 13851 if (had_bonus) | |
| 13852 rescore_one(su, stp); | |
| 13853 else | |
| 13854 { | |
| 13855 new_sug.st_word = stp->st_word; | |
| 13856 new_sug.st_wordlen = stp->st_wordlen; | |
| 13857 new_sug.st_slang = stp->st_slang; | |
| 13858 new_sug.st_orglen = badlen; | |
| 13859 rescore_one(su, &new_sug); | |
| 13860 } | |
| 13861 } | |
| 13862 | |
| 13863 if (stp->st_score > new_sug.st_score) | |
| 13864 { | |
| 13865 stp->st_score = new_sug.st_score; | |
| 13866 stp->st_altscore = new_sug.st_altscore; | |
| 13867 stp->st_had_bonus = new_sug.st_had_bonus; | |
| 13868 } | |
| 13869 break; | |
| 13870 } | |
| 13871 } | |
| 625 | 13872 |
| 13873 if (i < 0 && ga_grow(gap, 1) == OK) | |
| 13874 { | |
| 13875 /* Add a suggestion. */ | |
| 13876 stp = &SUG(*gap, gap->ga_len); | |
| 13877 stp->st_word = vim_strnsave(goodword, goodlen); | |
| 13878 if (stp->st_word != NULL) | |
| 13879 { | |
| 13880 stp->st_wordlen = goodlen; | |
| 13881 stp->st_score = score; | |
| 13882 stp->st_altscore = altscore; | |
| 13883 stp->st_had_bonus = had_bonus; | |
| 13884 stp->st_orglen = badlen; | |
| 13885 stp->st_slang = slang; | |
| 13886 ++gap->ga_len; | |
| 13887 | |
| 13888 /* If we have too many suggestions now, sort the list and keep | |
| 13889 * the best suggestions. */ | |
| 13890 if (gap->ga_len > SUG_MAX_COUNT(su)) | |
| 13891 { | |
| 13892 if (maxsf) | |
| 13893 su->su_sfmaxscore = cleanup_suggestions(gap, | |
| 13894 su->su_sfmaxscore, SUG_CLEAN_COUNT(su)); | |
| 13895 else | |
| 13896 su->su_maxscore = cleanup_suggestions(gap, | |
| 13897 su->su_maxscore, SUG_CLEAN_COUNT(su)); | |
| 13898 } | |
| 13899 } | |
| 13900 } | |
| 13901 } | |
| 13902 | |
| 13903 /* | |
| 13904 * Suggestions may in fact be flagged as errors. Esp. for banned words and | |
| 13905 * for split words, such as "the the". Remove these from the list here. | |
| 13906 */ | |
| 13907 static void | |
| 13908 check_suggestions(su, gap) | |
| 13909 suginfo_T *su; | |
| 13910 garray_T *gap; /* either su_ga or su_sga */ | |
| 13911 { | |
| 13912 suggest_T *stp; | |
| 13913 int i; | |
| 13914 char_u longword[MAXWLEN + 1]; | |
| 13915 int len; | |
| 13916 hlf_T attr; | |
| 13917 | |
| 13918 stp = &SUG(*gap, 0); | |
| 13919 for (i = gap->ga_len - 1; i >= 0; --i) | |
| 13920 { | |
| 13921 /* Need to append what follows to check for "the the". */ | |
| 2768 | 13922 vim_strncpy(longword, stp[i].st_word, MAXWLEN); |
| 625 | 13923 len = stp[i].st_wordlen; |
| 13924 vim_strncpy(longword + len, su->su_badptr + stp[i].st_orglen, | |
| 13925 MAXWLEN - len); | |
| 13926 attr = HLF_COUNT; | |
| 13927 (void)spell_check(curwin, longword, &attr, NULL, FALSE); | |
| 13928 if (attr != HLF_COUNT) | |
| 13929 { | |
| 13930 /* Remove this entry. */ | |
| 13931 vim_free(stp[i].st_word); | |
| 13932 --gap->ga_len; | |
| 13933 if (i < gap->ga_len) | |
| 13934 mch_memmove(stp + i, stp + i + 1, | |
| 13935 sizeof(suggest_T) * (gap->ga_len - i)); | |
| 13936 } | |
| 13937 } | |
| 13938 } | |
| 13939 | |
| 323 | 13940 |
| 13941 /* | |
| 13942 * Add a word to be banned. | |
| 13943 */ | |
| 13944 static void | |
| 13945 add_banned(su, word) | |
| 13946 suginfo_T *su; | |
| 13947 char_u *word; | |
| 13948 { | |
| 626 | 13949 char_u *s; |
| 323 | 13950 hash_T hash; |
| 13951 hashitem_T *hi; | |
| 13952 | |
| 625 | 13953 hash = hash_hash(word); |
| 13954 hi = hash_lookup(&su->su_banned, word, hash); | |
| 13955 if (HASHITEM_EMPTY(hi)) | |
| 13956 { | |
| 13957 s = vim_strsave(word); | |
| 13958 if (s != NULL) | |
| 323 | 13959 hash_add_item(&su->su_banned, hi, s, hash); |
| 625 | 13960 } |
| 323 | 13961 } |
| 13962 | |
| 324 | 13963 /* |
| 534 | 13964 * Recompute the score for all suggestions if sound-folding is possible. This |
| 13965 * is slow, thus only done for the final results. | |
| 324 | 13966 */ |
| 13967 static void | |
| 13968 rescore_suggestions(su) | |
| 13969 suginfo_T *su; | |
| 13970 { | |
| 13971 int i; | |
| 534 | 13972 |
| 13973 if (su->su_sallang != NULL) | |
| 503 | 13974 for (i = 0; i < su->su_ga.ga_len; ++i) |
| 534 | 13975 rescore_one(su, &SUG(su->su_ga, i)); |
| 13976 } | |
| 13977 | |
| 13978 /* | |
| 13979 * Recompute the score for one suggestion if sound-folding is possible. | |
| 13980 */ | |
| 13981 static void | |
| 13982 rescore_one(su, stp) | |
| 535 | 13983 suginfo_T *su; |
| 13984 suggest_T *stp; | |
| 534 | 13985 { |
| 13986 slang_T *slang = stp->st_slang; | |
| 13987 char_u sal_badword[MAXWLEN]; | |
| 535 | 13988 char_u *p; |
| 534 | 13989 |
| 13990 /* Only rescore suggestions that have no sal score yet and do have a | |
| 13991 * language. */ | |
| 13992 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus) | |
| 13993 { | |
| 13994 if (slang == su->su_sallang) | |
| 535 | 13995 p = su->su_sal_badword; |
| 534 | 13996 else |
| 13997 { | |
| 13998 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword); | |
| 535 | 13999 p = sal_badword; |
| 14000 } | |
| 14001 | |
| 14002 stp->st_altscore = stp_sal_score(stp, su, slang, p); | |
| 534 | 14003 if (stp->st_altscore == SCORE_MAXMAX) |
| 14004 stp->st_altscore = SCORE_BIG; | |
| 14005 stp->st_score = RESCORE(stp->st_score, stp->st_altscore); | |
| 14006 stp->st_had_bonus = TRUE; | |
| 324 | 14007 } |
| 14008 } | |
| 14009 | |
| 323 | 14010 static int |
| 14011 #ifdef __BORLANDC__ | |
| 14012 _RTLENTRYF | |
| 14013 #endif | |
| 14014 sug_compare __ARGS((const void *s1, const void *s2)); | |
| 14015 | |
| 14016 /* | |
| 14017 * Function given to qsort() to sort the suggestions on st_score. | |
| 529 | 14018 * First on "st_score", then "st_altscore" then alphabetically. |
| 323 | 14019 */ |
| 14020 static int | |
| 14021 #ifdef __BORLANDC__ | |
| 14022 _RTLENTRYF | |
| 14023 #endif | |
| 14024 sug_compare(s1, s2) | |
| 14025 const void *s1; | |
| 14026 const void *s2; | |
| 14027 { | |
| 14028 suggest_T *p1 = (suggest_T *)s1; | |
| 14029 suggest_T *p2 = (suggest_T *)s2; | |
| 344 | 14030 int n = p1->st_score - p2->st_score; |
| 14031 | |
| 14032 if (n == 0) | |
| 529 | 14033 { |
| 14034 n = p1->st_altscore - p2->st_altscore; | |
| 14035 if (n == 0) | |
| 14036 n = STRICMP(p1->st_word, p2->st_word); | |
| 14037 } | |
| 344 | 14038 return n; |
| 323 | 14039 } |
| 14040 | |
| 14041 /* | |
| 14042 * Cleanup the suggestions: | |
| 14043 * - Sort on score. | |
| 14044 * - Remove words that won't be displayed. | |
| 344 | 14045 * Returns the maximum score in the list or "maxscore" unmodified. |
| 323 | 14046 */ |
| 344 | 14047 static int |
| 14048 cleanup_suggestions(gap, maxscore, keep) | |
| 14049 garray_T *gap; | |
| 14050 int maxscore; | |
| 324 | 14051 int keep; /* nr of suggestions to keep */ |
| 323 | 14052 { |
| 344 | 14053 suggest_T *stp = &SUG(*gap, 0); |
| 323 | 14054 int i; |
| 14055 | |
| 14056 /* Sort the list. */ | |
| 344 | 14057 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare); |
| 323 | 14058 |
| 14059 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 344 | 14060 if (gap->ga_len > keep) |
| 323 | 14061 { |
| 344 | 14062 for (i = keep; i < gap->ga_len; ++i) |
| 323 | 14063 vim_free(stp[i].st_word); |
| 344 | 14064 gap->ga_len = keep; |
| 14065 return stp[keep - 1].st_score; | |
| 323 | 14066 } |
| 344 | 14067 return maxscore; |
| 323 | 14068 } |
| 14069 | |
| 372 | 14070 #if defined(FEAT_EVAL) || defined(PROTO) |
| 14071 /* | |
| 14072 * Soundfold a string, for soundfold(). | |
| 14073 * Result is in allocated memory, NULL for an error. | |
| 14074 */ | |
| 14075 char_u * | |
| 14076 eval_soundfold(word) | |
| 14077 char_u *word; | |
| 14078 { | |
| 14079 langp_T *lp; | |
| 14080 char_u sound[MAXWLEN]; | |
| 500 | 14081 int lpi; |
| 372 | 14082 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14083 if (curwin->w_p_spell && *curwin->w_s->b_p_spl != NUL) |
| 372 | 14084 /* Use the sound-folding of the first language that supports it. */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14085 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14086 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14087 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 372 | 14088 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 14089 { | |
| 14090 /* soundfold the word */ | |
| 375 | 14091 spell_soundfold(lp->lp_slang, word, FALSE, sound); |
| 372 | 14092 return vim_strsave(sound); |
| 14093 } | |
| 500 | 14094 } |
| 372 | 14095 |
| 14096 /* No language with sound folding, return word as-is. */ | |
| 14097 return vim_strsave(word); | |
| 14098 } | |
| 14099 #endif | |
| 14100 | |
| 323 | 14101 /* |
| 14102 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 485 | 14103 * |
| 14104 * There are many ways to turn a word into a sound-a-like representation. The | |
| 14105 * oldest is Soundex (1918!). A nice overview can be found in "Approximate | |
| 14106 * swedish name matching - survey and test of different algorithms" by Klas | |
| 14107 * Erikson. | |
| 14108 * | |
| 14109 * We support two methods: | |
| 14110 * 1. SOFOFROM/SOFOTO do a simple character mapping. | |
| 14111 * 2. SAL items define a more advanced sound-folding (and much slower). | |
| 323 | 14112 */ |
| 14113 static void | |
| 375 | 14114 spell_soundfold(slang, inword, folded, res) |
| 14115 slang_T *slang; | |
| 14116 char_u *inword; | |
| 14117 int folded; /* "inword" is already case-folded */ | |
| 14118 char_u *res; | |
| 14119 { | |
| 14120 char_u fword[MAXWLEN]; | |
| 14121 char_u *word; | |
| 14122 | |
| 14123 if (slang->sl_sofo) | |
| 14124 /* SOFOFROM and SOFOTO used */ | |
| 14125 spell_soundfold_sofo(slang, inword, res); | |
| 14126 else | |
| 14127 { | |
| 14128 /* SAL items used. Requires the word to be case-folded. */ | |
| 14129 if (folded) | |
| 14130 word = inword; | |
| 14131 else | |
| 14132 { | |
| 835 | 14133 (void)spell_casefold(inword, (int)STRLEN(inword), fword, MAXWLEN); |
| 375 | 14134 word = fword; |
| 14135 } | |
| 14136 | |
| 14137 #ifdef FEAT_MBYTE | |
| 14138 if (has_mbyte) | |
| 14139 spell_soundfold_wsal(slang, word, res); | |
| 14140 else | |
| 14141 #endif | |
| 14142 spell_soundfold_sal(slang, word, res); | |
| 14143 } | |
| 14144 } | |
| 14145 | |
| 14146 /* | |
| 14147 * Perform sound folding of "inword" into "res" according to SOFOFROM and | |
| 14148 * SOFOTO lines. | |
| 14149 */ | |
| 14150 static void | |
| 14151 spell_soundfold_sofo(slang, inword, res) | |
| 14152 slang_T *slang; | |
| 14153 char_u *inword; | |
| 14154 char_u *res; | |
| 14155 { | |
| 14156 char_u *s; | |
| 14157 int ri = 0; | |
| 14158 int c; | |
| 14159 | |
| 14160 #ifdef FEAT_MBYTE | |
| 14161 if (has_mbyte) | |
| 14162 { | |
| 14163 int prevc = 0; | |
| 14164 int *ip; | |
| 14165 | |
| 14166 /* The sl_sal_first[] table contains the translation for chars up to | |
| 14167 * 255, sl_sal the rest. */ | |
| 14168 for (s = inword; *s != NUL; ) | |
| 14169 { | |
| 474 | 14170 c = mb_cptr2char_adv(&s); |
| 375 | 14171 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c)) |
| 14172 c = ' '; | |
| 14173 else if (c < 256) | |
| 14174 c = slang->sl_sal_first[c]; | |
| 14175 else | |
| 14176 { | |
| 14177 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff]; | |
| 14178 if (ip == NULL) /* empty list, can't match */ | |
| 14179 c = NUL; | |
| 14180 else | |
| 14181 for (;;) /* find "c" in the list */ | |
| 14182 { | |
| 14183 if (*ip == 0) /* not found */ | |
| 14184 { | |
| 14185 c = NUL; | |
| 14186 break; | |
| 14187 } | |
| 14188 if (*ip == c) /* match! */ | |
| 14189 { | |
| 14190 c = ip[1]; | |
| 14191 break; | |
| 14192 } | |
| 14193 ip += 2; | |
| 14194 } | |
| 14195 } | |
| 14196 | |
| 14197 if (c != NUL && c != prevc) | |
| 14198 { | |
| 14199 ri += mb_char2bytes(c, res + ri); | |
| 14200 if (ri + MB_MAXBYTES > MAXWLEN) | |
| 14201 break; | |
| 14202 prevc = c; | |
| 14203 } | |
| 14204 } | |
| 14205 } | |
| 14206 else | |
| 14207 #endif | |
| 14208 { | |
| 14209 /* The sl_sal_first[] table contains the translation. */ | |
| 14210 for (s = inword; (c = *s) != NUL; ++s) | |
| 14211 { | |
| 14212 if (vim_iswhite(c)) | |
| 14213 c = ' '; | |
| 14214 else | |
| 14215 c = slang->sl_sal_first[c]; | |
| 14216 if (c != NUL && (ri == 0 || res[ri - 1] != c)) | |
| 14217 res[ri++] = c; | |
| 14218 } | |
| 14219 } | |
| 14220 | |
| 14221 res[ri] = NUL; | |
| 14222 } | |
| 14223 | |
| 14224 static void | |
| 14225 spell_soundfold_sal(slang, inword, res) | |
| 323 | 14226 slang_T *slang; |
| 14227 char_u *inword; | |
| 14228 char_u *res; | |
| 14229 { | |
| 344 | 14230 salitem_T *smp; |
| 323 | 14231 char_u word[MAXWLEN]; |
| 375 | 14232 char_u *s = inword; |
| 323 | 14233 char_u *t; |
| 344 | 14234 char_u *pf; |
| 323 | 14235 int i, j, z; |
| 344 | 14236 int reslen; |
| 323 | 14237 int n, k = 0; |
| 14238 int z0; | |
| 14239 int k0; | |
| 14240 int n0; | |
| 14241 int c; | |
| 14242 int pri; | |
| 14243 int p0 = -333; | |
| 14244 int c0; | |
| 14245 | |
| 324 | 14246 /* Remove accents, if wanted. We actually remove all non-word characters. |
| 375 | 14247 * But keep white space. We need a copy, the word may be changed here. */ |
| 323 | 14248 if (slang->sl_rem_accents) |
| 14249 { | |
| 14250 t = word; | |
| 375 | 14251 while (*s != NUL) |
| 323 | 14252 { |
| 324 | 14253 if (vim_iswhite(*s)) |
| 344 | 14254 { |
| 14255 *t++ = ' '; | |
| 14256 s = skipwhite(s); | |
| 14257 } | |
| 323 | 14258 else |
| 14259 { | |
| 5477 | 14260 if (spell_iswordp_nmw(s, curwin)) |
| 323 | 14261 *t++ = *s; |
| 14262 ++s; | |
| 14263 } | |
| 14264 } | |
| 14265 *t = NUL; | |
| 14266 } | |
| 14267 else | |
| 2768 | 14268 vim_strncpy(word, s, MAXWLEN - 1); |
| 323 | 14269 |
| 344 | 14270 smp = (salitem_T *)slang->sl_sal.ga_data; |
| 323 | 14271 |
| 14272 /* | |
| 14273 * This comes from Aspell phonet.cpp. Converted from C++ to C. | |
| 324 | 14274 * Changed to keep spaces. |
| 323 | 14275 */ |
| 344 | 14276 i = reslen = z = 0; |
| 323 | 14277 while ((c = word[i]) != NUL) |
| 14278 { | |
| 344 | 14279 /* Start with the first rule that has the character in the word. */ |
| 323 | 14280 n = slang->sl_sal_first[c]; |
| 14281 z0 = 0; | |
| 14282 | |
| 14283 if (n >= 0) | |
| 14284 { | |
| 14285 /* check all rules for the same letter */ | |
| 344 | 14286 for (; (s = smp[n].sm_lead)[0] == c; ++n) |
| 323 | 14287 { |
| 344 | 14288 /* Quickly skip entries that don't match the word. Most |
| 14289 * entries are less then three chars, optimize for that. */ | |
| 14290 k = smp[n].sm_leadlen; | |
| 14291 if (k > 1) | |
| 323 | 14292 { |
| 344 | 14293 if (word[i + 1] != s[1]) |
| 14294 continue; | |
| 14295 if (k > 2) | |
| 14296 { | |
| 14297 for (j = 2; j < k; ++j) | |
| 14298 if (word[i + j] != s[j]) | |
| 14299 break; | |
| 14300 if (j < k) | |
| 14301 continue; | |
| 14302 } | |
| 323 | 14303 } |
| 14304 | |
| 375 | 14305 if ((pf = smp[n].sm_oneof) != NULL) |
| 323 | 14306 { |
| 375 | 14307 /* Check for match with one of the chars in "sm_oneof". */ |
| 344 | 14308 while (*pf != NUL && *pf != word[i + k]) |
| 14309 ++pf; | |
| 14310 if (*pf == NUL) | |
| 14311 continue; | |
| 14312 ++k; | |
| 323 | 14313 } |
| 344 | 14314 s = smp[n].sm_rules; |
| 14315 pri = 5; /* default priority */ | |
| 323 | 14316 |
| 14317 p0 = *s; | |
| 14318 k0 = k; | |
| 14319 while (*s == '-' && k > 1) | |
| 14320 { | |
| 14321 k--; | |
| 14322 s++; | |
| 14323 } | |
| 14324 if (*s == '<') | |
| 14325 s++; | |
| 344 | 14326 if (VIM_ISDIGIT(*s)) |
| 323 | 14327 { |
| 14328 /* determine priority */ | |
| 14329 pri = *s - '0'; | |
| 14330 s++; | |
| 14331 } | |
| 14332 if (*s == '^' && *(s + 1) == '^') | |
| 14333 s++; | |
| 14334 | |
| 14335 if (*s == NUL | |
| 14336 || (*s == '^' | |
| 324 | 14337 && (i == 0 || !(word[i - 1] == ' ' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14338 || spell_iswordp(word + i - 1, curwin))) |
| 323 | 14339 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14340 || (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 14341 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14342 && spell_iswordp(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14343 && (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 14344 { |
| 14345 /* search for followup rules, if: */ | |
| 14346 /* followup and k > 1 and NO '-' in searchstring */ | |
| 14347 c0 = word[i + k - 1]; | |
| 14348 n0 = slang->sl_sal_first[c0]; | |
| 14349 | |
| 14350 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 344 | 14351 && p0 != '-' && word[i + k] != NUL) |
| 323 | 14352 { |
| 14353 /* test follow-up rule for "word[i + k]" */ | |
| 344 | 14354 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0) |
| 323 | 14355 { |
| 344 | 14356 /* Quickly skip entries that don't match the word. |
| 14357 * */ | |
| 14358 k0 = smp[n0].sm_leadlen; | |
| 14359 if (k0 > 1) | |
| 323 | 14360 { |
| 344 | 14361 if (word[i + k] != s[1]) |
| 14362 continue; | |
| 14363 if (k0 > 2) | |
| 14364 { | |
| 14365 pf = word + i + k + 1; | |
| 14366 for (j = 2; j < k0; ++j) | |
| 14367 if (*pf++ != s[j]) | |
| 14368 break; | |
| 14369 if (j < k0) | |
| 14370 continue; | |
| 14371 } | |
| 323 | 14372 } |
| 344 | 14373 k0 += k - 1; |
| 14374 | |
| 375 | 14375 if ((pf = smp[n0].sm_oneof) != NULL) |
| 323 | 14376 { |
| 344 | 14377 /* Check for match with one of the chars in |
| 375 | 14378 * "sm_oneof". */ |
| 344 | 14379 while (*pf != NUL && *pf != word[i + k0]) |
| 14380 ++pf; | |
| 14381 if (*pf == NUL) | |
| 14382 continue; | |
| 14383 ++k0; | |
| 323 | 14384 } |
| 344 | 14385 |
| 14386 p0 = 5; | |
| 14387 s = smp[n0].sm_rules; | |
| 323 | 14388 while (*s == '-') |
| 14389 { | |
| 344 | 14390 /* "k0" gets NOT reduced because |
| 14391 * "if (k0 == k)" */ | |
| 323 | 14392 s++; |
| 14393 } | |
| 14394 if (*s == '<') | |
| 14395 s++; | |
| 344 | 14396 if (VIM_ISDIGIT(*s)) |
| 323 | 14397 { |
| 14398 p0 = *s - '0'; | |
| 14399 s++; | |
| 14400 } | |
| 14401 | |
| 14402 if (*s == NUL | |
| 14403 /* *s == '^' cuts */ | |
| 14404 || (*s == '$' | |
| 376 | 14405 && !spell_iswordp(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14406 curwin))) |
| 323 | 14407 { |
| 14408 if (k0 == k) | |
| 14409 /* this is just a piece of the string */ | |
| 14410 continue; | |
| 14411 | |
| 14412 if (p0 < pri) | |
| 14413 /* priority too low */ | |
| 14414 continue; | |
| 14415 /* rule fits; stop search */ | |
| 14416 break; | |
| 14417 } | |
| 14418 } | |
| 14419 | |
| 344 | 14420 if (p0 >= pri && smp[n0].sm_lead[0] == c0) |
| 323 | 14421 continue; |
| 14422 } | |
| 14423 | |
| 14424 /* replace string */ | |
| 344 | 14425 s = smp[n].sm_to; |
| 389 | 14426 if (s == NULL) |
| 14427 s = (char_u *)""; | |
| 344 | 14428 pf = smp[n].sm_rules; |
| 14429 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0; | |
| 323 | 14430 if (p0 == 1 && z == 0) |
| 14431 { | |
| 14432 /* rule with '<' is used */ | |
| 344 | 14433 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c |
| 14434 || res[reslen - 1] == *s)) | |
| 14435 reslen--; | |
| 323 | 14436 z0 = 1; |
| 14437 z = 1; | |
| 14438 k0 = 0; | |
| 372 | 14439 while (*s != NUL && word[i + k0] != NUL) |
| 323 | 14440 { |
| 14441 word[i + k0] = *s; | |
| 14442 k0++; | |
| 14443 s++; | |
| 14444 } | |
| 14445 if (k > k0) | |
| 1619 | 14446 STRMOVE(word + i + k0, word + i + k); |
| 323 | 14447 |
| 14448 /* new "actual letter" */ | |
| 14449 c = word[i]; | |
| 14450 } | |
| 14451 else | |
| 14452 { | |
| 14453 /* no '<' rule used */ | |
| 14454 i += k - 1; | |
| 14455 z = 0; | |
| 344 | 14456 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN) |
| 323 | 14457 { |
| 344 | 14458 if (reslen == 0 || res[reslen - 1] != *s) |
| 372 | 14459 res[reslen++] = *s; |
| 323 | 14460 s++; |
| 14461 } | |
| 14462 /* new "actual letter" */ | |
| 14463 c = *s; | |
| 344 | 14464 if (strstr((char *)pf, "^^") != NULL) |
| 323 | 14465 { |
| 14466 if (c != NUL) | |
| 372 | 14467 res[reslen++] = c; |
| 1619 | 14468 STRMOVE(word, word + i + 1); |
| 323 | 14469 i = 0; |
| 14470 z0 = 1; | |
| 14471 } | |
| 14472 } | |
| 14473 break; | |
| 14474 } | |
| 14475 } | |
| 14476 } | |
| 324 | 14477 else if (vim_iswhite(c)) |
| 14478 { | |
| 14479 c = ' '; | |
| 14480 k = 1; | |
| 14481 } | |
| 323 | 14482 |
| 14483 if (z0 == 0) | |
| 14484 { | |
| 344 | 14485 if (k && !p0 && reslen < MAXWLEN && c != NUL |
| 14486 && (!slang->sl_collapse || reslen == 0 | |
| 14487 || res[reslen - 1] != c)) | |
| 323 | 14488 /* condense only double letters */ |
| 372 | 14489 res[reslen++] = c; |
| 323 | 14490 |
| 14491 i++; | |
| 14492 z = 0; | |
| 14493 k = 0; | |
| 14494 } | |
| 14495 } | |
| 14496 | |
| 344 | 14497 res[reslen] = NUL; |
| 323 | 14498 } |
| 14499 | |
| 372 | 14500 #ifdef FEAT_MBYTE |
| 14501 /* | |
| 14502 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 14503 * Multi-byte version of spell_soundfold(). | |
| 14504 */ | |
| 14505 static void | |
| 375 | 14506 spell_soundfold_wsal(slang, inword, res) |
| 372 | 14507 slang_T *slang; |
| 14508 char_u *inword; | |
| 14509 char_u *res; | |
| 14510 { | |
| 375 | 14511 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data; |
| 372 | 14512 int word[MAXWLEN]; |
| 14513 int wres[MAXWLEN]; | |
| 14514 int l; | |
| 14515 char_u *s; | |
| 14516 int *ws; | |
| 14517 char_u *t; | |
| 14518 int *pf; | |
| 14519 int i, j, z; | |
| 14520 int reslen; | |
| 14521 int n, k = 0; | |
| 14522 int z0; | |
| 14523 int k0; | |
| 14524 int n0; | |
| 14525 int c; | |
| 14526 int pri; | |
| 14527 int p0 = -333; | |
| 14528 int c0; | |
| 14529 int did_white = FALSE; | |
| 3520 | 14530 int wordlen; |
| 14531 | |
| 372 | 14532 |
| 14533 /* | |
| 14534 * Convert the multi-byte string to a wide-character string. | |
| 14535 * Remove accents, if wanted. We actually remove all non-word characters. | |
| 14536 * But keep white space. | |
| 14537 */ | |
| 3520 | 14538 wordlen = 0; |
| 372 | 14539 for (s = inword; *s != NUL; ) |
| 14540 { | |
| 14541 t = s; | |
| 474 | 14542 c = mb_cptr2char_adv(&s); |
| 372 | 14543 if (slang->sl_rem_accents) |
| 14544 { | |
| 14545 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c)) | |
| 14546 { | |
| 14547 if (did_white) | |
| 14548 continue; | |
| 14549 c = ' '; | |
| 14550 did_white = TRUE; | |
| 14551 } | |
| 14552 else | |
| 14553 { | |
| 14554 did_white = FALSE; | |
| 5477 | 14555 if (!spell_iswordp_nmw(t, curwin)) |
| 372 | 14556 continue; |
| 14557 } | |
| 14558 } | |
| 3520 | 14559 word[wordlen++] = c; |
| 14560 } | |
| 14561 word[wordlen] = NUL; | |
| 372 | 14562 |
| 14563 /* | |
| 3520 | 14564 * This algorithm comes from Aspell phonet.cpp. |
| 372 | 14565 * Converted from C++ to C. Added support for multi-byte chars. |
| 14566 * Changed to keep spaces. | |
| 14567 */ | |
| 14568 i = reslen = z = 0; | |
| 14569 while ((c = word[i]) != NUL) | |
| 14570 { | |
| 14571 /* Start with the first rule that has the character in the word. */ | |
| 14572 n = slang->sl_sal_first[c & 0xff]; | |
| 14573 z0 = 0; | |
| 14574 | |
| 14575 if (n >= 0) | |
| 14576 { | |
|
2455
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
14577 /* Check all rules for the same index byte. |
|
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
14578 * If c is 0x300 need extra check for the end of the array, as |
|
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
14579 * (c & 0xff) is NUL. */ |
|
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
14580 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff) |
|
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
14581 && ws[0] != NUL; ++n) |
| 372 | 14582 { |
| 14583 /* Quickly skip entries that don't match the word. Most | |
| 14584 * entries are less then three chars, optimize for that. */ | |
| 375 | 14585 if (c != ws[0]) |
| 14586 continue; | |
| 372 | 14587 k = smp[n].sm_leadlen; |
| 14588 if (k > 1) | |
| 14589 { | |
| 14590 if (word[i + 1] != ws[1]) | |
| 14591 continue; | |
| 14592 if (k > 2) | |
| 14593 { | |
| 14594 for (j = 2; j < k; ++j) | |
| 14595 if (word[i + j] != ws[j]) | |
| 14596 break; | |
| 14597 if (j < k) | |
| 14598 continue; | |
| 14599 } | |
| 14600 } | |
| 14601 | |
| 375 | 14602 if ((pf = smp[n].sm_oneof_w) != NULL) |
| 372 | 14603 { |
| 375 | 14604 /* Check for match with one of the chars in "sm_oneof". */ |
| 372 | 14605 while (*pf != NUL && *pf != word[i + k]) |
| 14606 ++pf; | |
| 14607 if (*pf == NUL) | |
| 14608 continue; | |
| 14609 ++k; | |
| 14610 } | |
| 14611 s = smp[n].sm_rules; | |
| 14612 pri = 5; /* default priority */ | |
| 14613 | |
| 14614 p0 = *s; | |
| 14615 k0 = k; | |
| 14616 while (*s == '-' && k > 1) | |
| 14617 { | |
| 14618 k--; | |
| 14619 s++; | |
| 14620 } | |
| 14621 if (*s == '<') | |
| 14622 s++; | |
| 14623 if (VIM_ISDIGIT(*s)) | |
| 14624 { | |
| 14625 /* determine priority */ | |
| 14626 pri = *s - '0'; | |
| 14627 s++; | |
| 14628 } | |
| 14629 if (*s == '^' && *(s + 1) == '^') | |
| 14630 s++; | |
| 14631 | |
| 14632 if (*s == NUL | |
| 14633 || (*s == '^' | |
| 14634 && (i == 0 || !(word[i - 1] == ' ' | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14635 || spell_iswordp_w(word + i - 1, curwin))) |
| 372 | 14636 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14637 || (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 14638 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14639 && spell_iswordp_w(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14640 && (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 14641 { |
| 14642 /* search for followup rules, if: */ | |
| 14643 /* followup and k > 1 and NO '-' in searchstring */ | |
| 14644 c0 = word[i + k - 1]; | |
| 14645 n0 = slang->sl_sal_first[c0 & 0xff]; | |
| 14646 | |
| 14647 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 14648 && p0 != '-' && word[i + k] != NUL) | |
| 14649 { | |
| 375 | 14650 /* Test follow-up rule for "word[i + k]"; loop over |
| 14651 * all entries with the same index byte. */ | |
| 372 | 14652 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff) |
| 14653 == (c0 & 0xff); ++n0) | |
| 14654 { | |
| 14655 /* Quickly skip entries that don't match the word. | |
| 375 | 14656 */ |
| 14657 if (c0 != ws[0]) | |
| 14658 continue; | |
| 372 | 14659 k0 = smp[n0].sm_leadlen; |
| 14660 if (k0 > 1) | |
| 14661 { | |
| 14662 if (word[i + k] != ws[1]) | |
| 14663 continue; | |
| 14664 if (k0 > 2) | |
| 14665 { | |
| 14666 pf = word + i + k + 1; | |
| 14667 for (j = 2; j < k0; ++j) | |
| 14668 if (*pf++ != ws[j]) | |
| 14669 break; | |
| 14670 if (j < k0) | |
| 14671 continue; | |
| 14672 } | |
| 14673 } | |
| 14674 k0 += k - 1; | |
| 14675 | |
| 375 | 14676 if ((pf = smp[n0].sm_oneof_w) != NULL) |
| 372 | 14677 { |
| 14678 /* Check for match with one of the chars in | |
| 375 | 14679 * "sm_oneof". */ |
| 372 | 14680 while (*pf != NUL && *pf != word[i + k0]) |
| 14681 ++pf; | |
| 14682 if (*pf == NUL) | |
| 14683 continue; | |
| 14684 ++k0; | |
| 14685 } | |
| 14686 | |
| 14687 p0 = 5; | |
| 14688 s = smp[n0].sm_rules; | |
| 14689 while (*s == '-') | |
| 14690 { | |
| 14691 /* "k0" gets NOT reduced because | |
| 14692 * "if (k0 == k)" */ | |
| 14693 s++; | |
| 14694 } | |
| 14695 if (*s == '<') | |
| 14696 s++; | |
| 14697 if (VIM_ISDIGIT(*s)) | |
| 14698 { | |
| 14699 p0 = *s - '0'; | |
| 14700 s++; | |
| 14701 } | |
| 14702 | |
| 14703 if (*s == NUL | |
| 14704 /* *s == '^' cuts */ | |
| 14705 || (*s == '$' | |
| 376 | 14706 && !spell_iswordp_w(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14707 curwin))) |
| 372 | 14708 { |
| 14709 if (k0 == k) | |
| 14710 /* this is just a piece of the string */ | |
| 14711 continue; | |
| 14712 | |
| 14713 if (p0 < pri) | |
| 14714 /* priority too low */ | |
| 14715 continue; | |
| 14716 /* rule fits; stop search */ | |
| 14717 break; | |
| 14718 } | |
| 14719 } | |
| 14720 | |
| 14721 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff) | |
| 14722 == (c0 & 0xff)) | |
| 14723 continue; | |
| 14724 } | |
| 14725 | |
| 14726 /* replace string */ | |
| 14727 ws = smp[n].sm_to_w; | |
| 14728 s = smp[n].sm_rules; | |
| 14729 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0; | |
| 14730 if (p0 == 1 && z == 0) | |
| 14731 { | |
| 14732 /* rule with '<' is used */ | |
| 389 | 14733 if (reslen > 0 && ws != NULL && *ws != NUL |
| 14734 && (wres[reslen - 1] == c | |
| 372 | 14735 || wres[reslen - 1] == *ws)) |
| 14736 reslen--; | |
| 14737 z0 = 1; | |
| 14738 z = 1; | |
| 14739 k0 = 0; | |
| 389 | 14740 if (ws != NULL) |
| 14741 while (*ws != NUL && word[i + k0] != NUL) | |
| 14742 { | |
| 14743 word[i + k0] = *ws; | |
| 14744 k0++; | |
| 14745 ws++; | |
| 14746 } | |
| 372 | 14747 if (k > k0) |
| 14748 mch_memmove(word + i + k0, word + i + k, | |
| 3520 | 14749 sizeof(int) * (wordlen - (i + k) + 1)); |
| 372 | 14750 |
| 14751 /* new "actual letter" */ | |
| 14752 c = word[i]; | |
| 14753 } | |
| 14754 else | |
| 14755 { | |
| 14756 /* no '<' rule used */ | |
| 14757 i += k - 1; | |
| 14758 z = 0; | |
| 389 | 14759 if (ws != NULL) |
| 14760 while (*ws != NUL && ws[1] != NUL | |
| 14761 && reslen < MAXWLEN) | |
| 14762 { | |
| 14763 if (reslen == 0 || wres[reslen - 1] != *ws) | |
| 14764 wres[reslen++] = *ws; | |
| 14765 ws++; | |
| 14766 } | |
| 372 | 14767 /* new "actual letter" */ |
| 389 | 14768 if (ws == NULL) |
| 14769 c = NUL; | |
| 14770 else | |
| 14771 c = *ws; | |
| 372 | 14772 if (strstr((char *)s, "^^") != NULL) |
| 14773 { | |
| 14774 if (c != NUL) | |
| 14775 wres[reslen++] = c; | |
| 14776 mch_memmove(word, word + i + 1, | |
| 3520 | 14777 sizeof(int) * (wordlen - (i + 1) + 1)); |
| 372 | 14778 i = 0; |
| 14779 z0 = 1; | |
| 14780 } | |
| 14781 } | |
| 14782 break; | |
| 14783 } | |
| 14784 } | |
| 14785 } | |
| 14786 else if (vim_iswhite(c)) | |
| 14787 { | |
| 14788 c = ' '; | |
| 14789 k = 1; | |
| 14790 } | |
| 14791 | |
| 14792 if (z0 == 0) | |
| 14793 { | |
| 14794 if (k && !p0 && reslen < MAXWLEN && c != NUL | |
| 14795 && (!slang->sl_collapse || reslen == 0 | |
| 14796 || wres[reslen - 1] != c)) | |
| 14797 /* condense only double letters */ | |
| 14798 wres[reslen++] = c; | |
| 14799 | |
| 14800 i++; | |
| 14801 z = 0; | |
| 14802 k = 0; | |
| 14803 } | |
| 14804 } | |
| 14805 | |
| 14806 /* Convert wide characters in "wres" to a multi-byte string in "res". */ | |
| 14807 l = 0; | |
| 14808 for (n = 0; n < reslen; ++n) | |
| 14809 { | |
| 14810 l += mb_char2bytes(wres[n], res + l); | |
| 14811 if (l + MB_MAXBYTES > MAXWLEN) | |
| 14812 break; | |
| 14813 } | |
| 14814 res[l] = NUL; | |
| 14815 } | |
| 14816 #endif | |
| 14817 | |
| 324 | 14818 /* |
| 344 | 14819 * Compute a score for two sound-a-like words. |
| 14820 * This permits up to two inserts/deletes/swaps/etc. to keep things fast. | |
| 14821 * Instead of a generic loop we write out the code. That keeps it fast by | |
| 14822 * avoiding checks that will not be possible. | |
| 14823 */ | |
| 14824 static int | |
| 351 | 14825 soundalike_score(goodstart, badstart) |
| 14826 char_u *goodstart; /* sound-folded good word */ | |
| 14827 char_u *badstart; /* sound-folded bad word */ | |
| 344 | 14828 { |
| 351 | 14829 char_u *goodsound = goodstart; |
| 14830 char_u *badsound = badstart; | |
| 14831 int goodlen; | |
| 14832 int badlen; | |
| 344 | 14833 int n; |
| 14834 char_u *pl, *ps; | |
| 14835 char_u *pl2, *ps2; | |
| 351 | 14836 int score = 0; |
| 14837 | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14838 /* Adding/inserting "*" at the start (word starts with vowel) shouldn't be |
| 351 | 14839 * counted so much, vowels halfway the word aren't counted at all. */ |
| 14840 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound) | |
| 14841 { | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14842 if ((badsound[0] == NUL && goodsound[1] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14843 || (goodsound[0] == NUL && badsound[1] == NUL)) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14844 /* changing word with vowel to word without a sound */ |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14845 return SCORE_DEL; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14846 if (badsound[0] == NUL || goodsound[0] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14847 /* more than two changes */ |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14848 return SCORE_MAXMAX; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14849 |
| 625 | 14850 if (badsound[1] == goodsound[1] |
| 14851 || (badsound[1] != NUL | |
| 14852 && goodsound[1] != NUL | |
| 14853 && badsound[2] == goodsound[2])) | |
| 14854 { | |
| 14855 /* handle like a substitute */ | |
| 14856 } | |
| 351 | 14857 else |
| 625 | 14858 { |
| 14859 score = 2 * SCORE_DEL / 3; | |
| 14860 if (*badsound == '*') | |
| 14861 ++badsound; | |
| 14862 else | |
| 14863 ++goodsound; | |
| 14864 } | |
| 351 | 14865 } |
| 14866 | |
| 835 | 14867 goodlen = (int)STRLEN(goodsound); |
| 14868 badlen = (int)STRLEN(badsound); | |
| 344 | 14869 |
| 1188 | 14870 /* Return quickly if the lengths are too different to be fixed by two |
| 344 | 14871 * changes. */ |
| 14872 n = goodlen - badlen; | |
| 14873 if (n < -2 || n > 2) | |
| 14874 return SCORE_MAXMAX; | |
| 14875 | |
| 14876 if (n > 0) | |
| 14877 { | |
| 351 | 14878 pl = goodsound; /* goodsound is longest */ |
| 344 | 14879 ps = badsound; |
| 14880 } | |
| 14881 else | |
| 14882 { | |
| 351 | 14883 pl = badsound; /* badsound is longest */ |
| 344 | 14884 ps = goodsound; |
| 14885 } | |
| 14886 | |
| 14887 /* Skip over the identical part. */ | |
| 14888 while (*pl == *ps && *pl != NUL) | |
| 14889 { | |
| 14890 ++pl; | |
| 14891 ++ps; | |
| 14892 } | |
| 14893 | |
| 14894 switch (n) | |
| 14895 { | |
| 14896 case -2: | |
| 14897 case 2: | |
| 14898 /* | |
| 14899 * Must delete two characters from "pl". | |
| 14900 */ | |
| 14901 ++pl; /* first delete */ | |
| 14902 while (*pl == *ps) | |
| 14903 { | |
| 14904 ++pl; | |
| 14905 ++ps; | |
| 14906 } | |
| 14907 /* strings must be equal after second delete */ | |
| 14908 if (STRCMP(pl + 1, ps) == 0) | |
| 351 | 14909 return score + SCORE_DEL * 2; |
| 344 | 14910 |
| 14911 /* Failed to compare. */ | |
| 14912 break; | |
| 14913 | |
| 14914 case -1: | |
| 14915 case 1: | |
| 14916 /* | |
| 14917 * Minimal one delete from "pl" required. | |
| 14918 */ | |
| 14919 | |
| 14920 /* 1: delete */ | |
| 14921 pl2 = pl + 1; | |
| 14922 ps2 = ps; | |
| 14923 while (*pl2 == *ps2) | |
| 14924 { | |
| 14925 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 14926 return score + SCORE_DEL; |
| 344 | 14927 ++pl2; |
| 14928 ++ps2; | |
| 14929 } | |
| 14930 | |
| 14931 /* 2: delete then swap, then rest must be equal */ | |
| 14932 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 14933 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 14934 return score + SCORE_DEL + SCORE_SWAP; |
| 344 | 14935 |
| 14936 /* 3: delete then substitute, then the rest must be equal */ | |
| 14937 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 14938 return score + SCORE_DEL + SCORE_SUBST; |
| 344 | 14939 |
| 14940 /* 4: first swap then delete */ | |
| 14941 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 14942 { | |
| 14943 pl2 = pl + 2; /* swap, skip two chars */ | |
| 14944 ps2 = ps + 2; | |
| 14945 while (*pl2 == *ps2) | |
| 14946 { | |
| 14947 ++pl2; | |
| 14948 ++ps2; | |
| 14949 } | |
| 14950 /* delete a char and then strings must be equal */ | |
| 14951 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 14952 return score + SCORE_SWAP + SCORE_DEL; |
| 344 | 14953 } |
| 14954 | |
| 14955 /* 5: first substitute then delete */ | |
| 14956 pl2 = pl + 1; /* substitute, skip one char */ | |
| 14957 ps2 = ps + 1; | |
| 14958 while (*pl2 == *ps2) | |
| 14959 { | |
| 14960 ++pl2; | |
| 14961 ++ps2; | |
| 14962 } | |
| 14963 /* delete a char and then strings must be equal */ | |
| 14964 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 14965 return score + SCORE_SUBST + SCORE_DEL; |
| 344 | 14966 |
| 14967 /* Failed to compare. */ | |
| 14968 break; | |
| 14969 | |
| 14970 case 0: | |
| 14971 /* | |
| 1808 | 14972 * Lengths are equal, thus changes must result in same length: An |
| 344 | 14973 * insert is only possible in combination with a delete. |
| 14974 * 1: check if for identical strings | |
| 14975 */ | |
| 14976 if (*pl == NUL) | |
| 351 | 14977 return score; |
| 344 | 14978 |
| 14979 /* 2: swap */ | |
| 14980 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 14981 { | |
| 14982 pl2 = pl + 2; /* swap, skip two chars */ | |
| 14983 ps2 = ps + 2; | |
| 14984 while (*pl2 == *ps2) | |
| 14985 { | |
| 14986 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 14987 return score + SCORE_SWAP; |
| 344 | 14988 ++pl2; |
| 14989 ++ps2; | |
| 14990 } | |
| 14991 /* 3: swap and swap again */ | |
| 14992 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 14993 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 14994 return score + SCORE_SWAP + SCORE_SWAP; |
| 344 | 14995 |
| 14996 /* 4: swap and substitute */ | |
| 14997 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 14998 return score + SCORE_SWAP + SCORE_SUBST; |
| 344 | 14999 } |
| 15000 | |
| 15001 /* 5: substitute */ | |
| 15002 pl2 = pl + 1; | |
| 15003 ps2 = ps + 1; | |
| 15004 while (*pl2 == *ps2) | |
| 15005 { | |
| 15006 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 15007 return score + SCORE_SUBST; |
| 344 | 15008 ++pl2; |
| 15009 ++ps2; | |
| 15010 } | |
| 15011 | |
| 15012 /* 6: substitute and swap */ | |
| 15013 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 15014 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 15015 return score + SCORE_SUBST + SCORE_SWAP; |
| 344 | 15016 |
| 15017 /* 7: substitute and substitute */ | |
| 15018 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 15019 return score + SCORE_SUBST + SCORE_SUBST; |
| 344 | 15020 |
| 15021 /* 8: insert then delete */ | |
| 15022 pl2 = pl; | |
| 15023 ps2 = ps + 1; | |
| 15024 while (*pl2 == *ps2) | |
| 15025 { | |
| 15026 ++pl2; | |
| 15027 ++ps2; | |
| 15028 } | |
| 15029 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 15030 return score + SCORE_INS + SCORE_DEL; |
| 344 | 15031 |
| 15032 /* 9: delete then insert */ | |
| 15033 pl2 = pl + 1; | |
| 15034 ps2 = ps; | |
| 15035 while (*pl2 == *ps2) | |
| 15036 { | |
| 15037 ++pl2; | |
| 15038 ++ps2; | |
| 15039 } | |
| 15040 if (STRCMP(pl2, ps2 + 1) == 0) | |
| 351 | 15041 return score + SCORE_INS + SCORE_DEL; |
| 344 | 15042 |
| 15043 /* Failed to compare. */ | |
| 15044 break; | |
| 15045 } | |
| 15046 | |
| 15047 return SCORE_MAXMAX; | |
| 15048 } | |
| 324 | 15049 |
| 323 | 15050 /* |
| 15051 * Compute the "edit distance" to turn "badword" into "goodword". The less | |
| 344 | 15052 * deletes/inserts/substitutes/swaps are required the lower the score. |
| 324 | 15053 * |
| 485 | 15054 * The algorithm is described by Du and Chang, 1992. |
| 15055 * The implementation of the algorithm comes from Aspell editdist.cpp, | |
| 15056 * edit_distance(). It has been converted from C++ to C and modified to | |
| 15057 * support multi-byte characters. | |
| 323 | 15058 */ |
| 15059 static int | |
| 625 | 15060 spell_edit_score(slang, badword, goodword) |
| 15061 slang_T *slang; | |
| 323 | 15062 char_u *badword; |
| 15063 char_u *goodword; | |
| 15064 { | |
| 15065 int *cnt; | |
| 1188 | 15066 int badlen, goodlen; /* lengths including NUL */ |
| 323 | 15067 int j, i; |
| 15068 int t; | |
| 15069 int bc, gc; | |
| 324 | 15070 int pbc, pgc; |
| 15071 #ifdef FEAT_MBYTE | |
| 15072 char_u *p; | |
| 15073 int wbadword[MAXWLEN]; | |
| 15074 int wgoodword[MAXWLEN]; | |
| 15075 | |
| 15076 if (has_mbyte) | |
| 15077 { | |
| 15078 /* Get the characters from the multi-byte strings and put them in an | |
| 15079 * int array for easy access. */ | |
| 15080 for (p = badword, badlen = 0; *p != NUL; ) | |
| 474 | 15081 wbadword[badlen++] = mb_cptr2char_adv(&p); |
| 398 | 15082 wbadword[badlen++] = 0; |
| 324 | 15083 for (p = goodword, goodlen = 0; *p != NUL; ) |
| 474 | 15084 wgoodword[goodlen++] = mb_cptr2char_adv(&p); |
| 398 | 15085 wgoodword[goodlen++] = 0; |
| 324 | 15086 } |
| 15087 else | |
| 15088 #endif | |
| 15089 { | |
| 835 | 15090 badlen = (int)STRLEN(badword) + 1; |
| 15091 goodlen = (int)STRLEN(goodword) + 1; | |
| 324 | 15092 } |
| 323 | 15093 |
| 15094 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */ | |
| 15095 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)] | |
| 15096 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)), | |
| 15097 TRUE); | |
| 324 | 15098 if (cnt == NULL) |
| 15099 return 0; /* out of memory */ | |
| 323 | 15100 |
| 15101 CNT(0, 0) = 0; | |
| 15102 for (j = 1; j <= goodlen; ++j) | |
| 625 | 15103 CNT(0, j) = CNT(0, j - 1) + SCORE_INS; |
| 323 | 15104 |
| 15105 for (i = 1; i <= badlen; ++i) | |
| 15106 { | |
| 625 | 15107 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL; |
| 323 | 15108 for (j = 1; j <= goodlen; ++j) |
| 15109 { | |
| 324 | 15110 #ifdef FEAT_MBYTE |
| 15111 if (has_mbyte) | |
| 15112 { | |
| 15113 bc = wbadword[i - 1]; | |
| 15114 gc = wgoodword[j - 1]; | |
| 15115 } | |
| 15116 else | |
| 15117 #endif | |
| 15118 { | |
| 15119 bc = badword[i - 1]; | |
| 15120 gc = goodword[j - 1]; | |
| 15121 } | |
| 323 | 15122 if (bc == gc) |
| 15123 CNT(i, j) = CNT(i - 1, j - 1); | |
| 15124 else | |
| 15125 { | |
| 15126 /* Use a better score when there is only a case difference. */ | |
| 324 | 15127 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) |
| 323 | 15128 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1); |
| 15129 else | |
| 625 | 15130 { |
| 15131 /* For a similar character use SCORE_SIMILAR. */ | |
| 15132 if (slang != NULL | |
| 15133 && slang->sl_has_map | |
| 15134 && similar_chars(slang, gc, bc)) | |
| 15135 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1); | |
| 15136 else | |
| 15137 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1); | |
| 15138 } | |
| 323 | 15139 |
| 324 | 15140 if (i > 1 && j > 1) |
| 323 | 15141 { |
| 324 | 15142 #ifdef FEAT_MBYTE |
| 15143 if (has_mbyte) | |
| 15144 { | |
| 15145 pbc = wbadword[i - 2]; | |
| 15146 pgc = wgoodword[j - 2]; | |
| 15147 } | |
| 15148 else | |
| 15149 #endif | |
| 15150 { | |
| 15151 pbc = badword[i - 2]; | |
| 15152 pgc = goodword[j - 2]; | |
| 15153 } | |
| 15154 if (bc == pgc && pbc == gc) | |
| 15155 { | |
| 15156 t = SCORE_SWAP + CNT(i - 2, j - 2); | |
| 15157 if (t < CNT(i, j)) | |
| 15158 CNT(i, j) = t; | |
| 15159 } | |
| 323 | 15160 } |
| 15161 t = SCORE_DEL + CNT(i - 1, j); | |
| 15162 if (t < CNT(i, j)) | |
| 15163 CNT(i, j) = t; | |
| 15164 t = SCORE_INS + CNT(i, j - 1); | |
| 15165 if (t < CNT(i, j)) | |
| 15166 CNT(i, j) = t; | |
| 15167 } | |
| 15168 } | |
| 15169 } | |
| 344 | 15170 |
| 15171 i = CNT(badlen - 1, goodlen - 1); | |
| 15172 vim_free(cnt); | |
| 15173 return i; | |
| 323 | 15174 } |
| 307 | 15175 |
| 625 | 15176 typedef struct |
| 15177 { | |
| 15178 int badi; | |
| 15179 int goodi; | |
| 15180 int score; | |
| 15181 } limitscore_T; | |
| 15182 | |
| 15183 /* | |
| 15184 * Like spell_edit_score(), but with a limit on the score to make it faster. | |
| 15185 * May return SCORE_MAXMAX when the score is higher than "limit". | |
| 15186 * | |
| 15187 * This uses a stack for the edits still to be tried. | |
| 15188 * The idea comes from Aspell leditdist.cpp. Rewritten in C and added support | |
| 15189 * for multi-byte characters. | |
| 15190 */ | |
| 15191 static int | |
| 15192 spell_edit_score_limit(slang, badword, goodword, limit) | |
| 15193 slang_T *slang; | |
| 15194 char_u *badword; | |
| 15195 char_u *goodword; | |
| 15196 int limit; | |
| 15197 { | |
| 15198 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 15199 int stackidx; | |
| 15200 int bi, gi; | |
| 15201 int bi2, gi2; | |
| 15202 int bc, gc; | |
| 15203 int score; | |
| 15204 int score_off; | |
| 15205 int minscore; | |
| 15206 int round; | |
| 15207 | |
| 15208 #ifdef FEAT_MBYTE | |
| 15209 /* Multi-byte characters require a bit more work, use a different function | |
| 15210 * to avoid testing "has_mbyte" quite often. */ | |
| 15211 if (has_mbyte) | |
| 15212 return spell_edit_score_limit_w(slang, badword, goodword, limit); | |
| 15213 #endif | |
| 15214 | |
| 15215 /* | |
| 15216 * The idea is to go from start to end over the words. So long as | |
| 15217 * characters are equal just continue, this always gives the lowest score. | |
| 15218 * When there is a difference try several alternatives. Each alternative | |
| 15219 * increases "score" for the edit distance. Some of the alternatives are | |
| 15220 * pushed unto a stack and tried later, some are tried right away. At the | |
| 15221 * end of the word the score for one alternative is known. The lowest | |
| 15222 * possible score is stored in "minscore". | |
| 15223 */ | |
| 15224 stackidx = 0; | |
| 15225 bi = 0; | |
| 15226 gi = 0; | |
| 15227 score = 0; | |
| 15228 minscore = limit + 1; | |
| 15229 | |
| 15230 for (;;) | |
| 15231 { | |
| 15232 /* Skip over an equal part, score remains the same. */ | |
| 15233 for (;;) | |
| 15234 { | |
| 15235 bc = badword[bi]; | |
| 15236 gc = goodword[gi]; | |
| 15237 if (bc != gc) /* stop at a char that's different */ | |
| 15238 break; | |
| 15239 if (bc == NUL) /* both words end */ | |
| 15240 { | |
| 15241 if (score < minscore) | |
| 15242 minscore = score; | |
| 15243 goto pop; /* do next alternative */ | |
| 15244 } | |
| 15245 ++bi; | |
| 15246 ++gi; | |
| 15247 } | |
| 15248 | |
| 15249 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 15250 { | |
| 15251 do | |
| 15252 { | |
| 15253 if ((score += SCORE_DEL) >= minscore) | |
| 15254 goto pop; /* do next alternative */ | |
| 15255 } while (badword[++bi] != NUL); | |
| 15256 minscore = score; | |
| 15257 } | |
| 15258 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 15259 { | |
| 15260 do | |
| 15261 { | |
| 15262 if ((score += SCORE_INS) >= minscore) | |
| 15263 goto pop; /* do next alternative */ | |
| 15264 } while (goodword[++gi] != NUL); | |
| 15265 minscore = score; | |
| 15266 } | |
| 15267 else /* both words continue */ | |
| 15268 { | |
| 15269 /* If not close to the limit, perform a change. Only try changes | |
| 15270 * that may lead to a lower score than "minscore". | |
| 15271 * round 0: try deleting a char from badword | |
| 15272 * round 1: try inserting a char in badword */ | |
| 15273 for (round = 0; round <= 1; ++round) | |
| 15274 { | |
| 15275 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 15276 if (score_off < minscore) | |
| 15277 { | |
| 15278 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 15279 { | |
| 15280 /* Near the limit, rest of the words must match. We | |
| 15281 * can check that right now, no need to push an item | |
| 15282 * onto the stack. */ | |
| 15283 bi2 = bi + 1 - round; | |
| 15284 gi2 = gi + round; | |
| 15285 while (goodword[gi2] == badword[bi2]) | |
| 15286 { | |
| 15287 if (goodword[gi2] == NUL) | |
| 15288 { | |
| 15289 minscore = score_off; | |
| 15290 break; | |
| 15291 } | |
| 15292 ++bi2; | |
| 15293 ++gi2; | |
| 15294 } | |
| 15295 } | |
| 15296 else | |
| 15297 { | |
| 15298 /* try deleting/inserting a character later */ | |
| 15299 stack[stackidx].badi = bi + 1 - round; | |
| 15300 stack[stackidx].goodi = gi + round; | |
| 15301 stack[stackidx].score = score_off; | |
| 15302 ++stackidx; | |
| 15303 } | |
| 15304 } | |
| 15305 } | |
| 15306 | |
| 15307 if (score + SCORE_SWAP < minscore) | |
| 15308 { | |
| 15309 /* If swapping two characters makes a match then the | |
| 15310 * substitution is more expensive, thus there is no need to | |
| 15311 * try both. */ | |
| 15312 if (gc == badword[bi + 1] && bc == goodword[gi + 1]) | |
| 15313 { | |
| 15314 /* Swap two characters, that is: skip them. */ | |
| 15315 gi += 2; | |
| 15316 bi += 2; | |
| 15317 score += SCORE_SWAP; | |
| 15318 continue; | |
| 15319 } | |
| 15320 } | |
| 15321 | |
| 15322 /* Substitute one character for another which is the same | |
| 15323 * thing as deleting a character from both goodword and badword. | |
| 15324 * Use a better score when there is only a case difference. */ | |
| 15325 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 15326 score += SCORE_ICASE; | |
| 15327 else | |
| 15328 { | |
| 15329 /* For a similar character use SCORE_SIMILAR. */ | |
| 15330 if (slang != NULL | |
| 15331 && slang->sl_has_map | |
| 15332 && similar_chars(slang, gc, bc)) | |
| 15333 score += SCORE_SIMILAR; | |
| 15334 else | |
| 15335 score += SCORE_SUBST; | |
| 15336 } | |
| 15337 | |
| 15338 if (score < minscore) | |
| 15339 { | |
| 15340 /* Do the substitution. */ | |
| 15341 ++gi; | |
| 15342 ++bi; | |
| 15343 continue; | |
| 15344 } | |
| 15345 } | |
| 15346 pop: | |
| 15347 /* | |
| 15348 * Get here to try the next alternative, pop it from the stack. | |
| 15349 */ | |
| 15350 if (stackidx == 0) /* stack is empty, finished */ | |
| 15351 break; | |
| 15352 | |
| 15353 /* pop an item from the stack */ | |
| 15354 --stackidx; | |
| 15355 gi = stack[stackidx].goodi; | |
| 15356 bi = stack[stackidx].badi; | |
| 15357 score = stack[stackidx].score; | |
| 15358 } | |
| 15359 | |
| 15360 /* When the score goes over "limit" it may actually be much higher. | |
| 15361 * Return a very large number to avoid going below the limit when giving a | |
| 15362 * bonus. */ | |
| 15363 if (minscore > limit) | |
| 15364 return SCORE_MAXMAX; | |
| 15365 return minscore; | |
| 15366 } | |
| 15367 | |
| 15368 #ifdef FEAT_MBYTE | |
| 15369 /* | |
| 15370 * Multi-byte version of spell_edit_score_limit(). | |
| 15371 * Keep it in sync with the above! | |
| 15372 */ | |
| 15373 static int | |
| 15374 spell_edit_score_limit_w(slang, badword, goodword, limit) | |
| 15375 slang_T *slang; | |
| 15376 char_u *badword; | |
| 15377 char_u *goodword; | |
| 15378 int limit; | |
| 15379 { | |
| 15380 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 15381 int stackidx; | |
| 15382 int bi, gi; | |
| 15383 int bi2, gi2; | |
| 15384 int bc, gc; | |
| 15385 int score; | |
| 15386 int score_off; | |
| 15387 int minscore; | |
| 15388 int round; | |
| 15389 char_u *p; | |
| 15390 int wbadword[MAXWLEN]; | |
| 15391 int wgoodword[MAXWLEN]; | |
| 15392 | |
| 15393 /* Get the characters from the multi-byte strings and put them in an | |
| 15394 * int array for easy access. */ | |
| 15395 bi = 0; | |
| 15396 for (p = badword; *p != NUL; ) | |
| 15397 wbadword[bi++] = mb_cptr2char_adv(&p); | |
| 15398 wbadword[bi++] = 0; | |
| 15399 gi = 0; | |
| 15400 for (p = goodword; *p != NUL; ) | |
| 15401 wgoodword[gi++] = mb_cptr2char_adv(&p); | |
| 15402 wgoodword[gi++] = 0; | |
| 15403 | |
| 15404 /* | |
| 15405 * The idea is to go from start to end over the words. So long as | |
| 15406 * characters are equal just continue, this always gives the lowest score. | |
| 15407 * When there is a difference try several alternatives. Each alternative | |
| 15408 * increases "score" for the edit distance. Some of the alternatives are | |
| 15409 * pushed unto a stack and tried later, some are tried right away. At the | |
| 15410 * end of the word the score for one alternative is known. The lowest | |
| 15411 * possible score is stored in "minscore". | |
| 15412 */ | |
| 15413 stackidx = 0; | |
| 15414 bi = 0; | |
| 15415 gi = 0; | |
| 15416 score = 0; | |
| 15417 minscore = limit + 1; | |
| 15418 | |
| 15419 for (;;) | |
| 15420 { | |
| 15421 /* Skip over an equal part, score remains the same. */ | |
| 15422 for (;;) | |
| 15423 { | |
| 15424 bc = wbadword[bi]; | |
| 15425 gc = wgoodword[gi]; | |
| 15426 | |
| 15427 if (bc != gc) /* stop at a char that's different */ | |
| 15428 break; | |
| 15429 if (bc == NUL) /* both words end */ | |
| 15430 { | |
| 15431 if (score < minscore) | |
| 15432 minscore = score; | |
| 15433 goto pop; /* do next alternative */ | |
| 15434 } | |
| 15435 ++bi; | |
| 15436 ++gi; | |
| 15437 } | |
| 15438 | |
| 15439 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 15440 { | |
| 15441 do | |
| 15442 { | |
| 15443 if ((score += SCORE_DEL) >= minscore) | |
| 15444 goto pop; /* do next alternative */ | |
| 15445 } while (wbadword[++bi] != NUL); | |
| 15446 minscore = score; | |
| 15447 } | |
| 15448 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 15449 { | |
| 15450 do | |
| 15451 { | |
| 15452 if ((score += SCORE_INS) >= minscore) | |
| 15453 goto pop; /* do next alternative */ | |
| 15454 } while (wgoodword[++gi] != NUL); | |
| 15455 minscore = score; | |
| 15456 } | |
| 15457 else /* both words continue */ | |
| 15458 { | |
| 15459 /* If not close to the limit, perform a change. Only try changes | |
| 15460 * that may lead to a lower score than "minscore". | |
| 15461 * round 0: try deleting a char from badword | |
| 15462 * round 1: try inserting a char in badword */ | |
| 15463 for (round = 0; round <= 1; ++round) | |
| 15464 { | |
| 15465 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 15466 if (score_off < minscore) | |
| 15467 { | |
| 15468 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 15469 { | |
| 15470 /* Near the limit, rest of the words must match. We | |
| 15471 * can check that right now, no need to push an item | |
| 15472 * onto the stack. */ | |
| 15473 bi2 = bi + 1 - round; | |
| 15474 gi2 = gi + round; | |
| 15475 while (wgoodword[gi2] == wbadword[bi2]) | |
| 15476 { | |
| 15477 if (wgoodword[gi2] == NUL) | |
| 15478 { | |
| 15479 minscore = score_off; | |
| 15480 break; | |
| 15481 } | |
| 15482 ++bi2; | |
| 15483 ++gi2; | |
| 15484 } | |
| 15485 } | |
| 15486 else | |
| 15487 { | |
| 15488 /* try deleting a character from badword later */ | |
| 15489 stack[stackidx].badi = bi + 1 - round; | |
| 15490 stack[stackidx].goodi = gi + round; | |
| 15491 stack[stackidx].score = score_off; | |
| 15492 ++stackidx; | |
| 15493 } | |
| 15494 } | |
| 15495 } | |
| 15496 | |
| 15497 if (score + SCORE_SWAP < minscore) | |
| 15498 { | |
| 15499 /* If swapping two characters makes a match then the | |
| 15500 * substitution is more expensive, thus there is no need to | |
| 15501 * try both. */ | |
| 15502 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1]) | |
| 15503 { | |
| 15504 /* Swap two characters, that is: skip them. */ | |
| 15505 gi += 2; | |
| 15506 bi += 2; | |
| 15507 score += SCORE_SWAP; | |
| 15508 continue; | |
| 15509 } | |
| 15510 } | |
| 15511 | |
| 15512 /* Substitute one character for another which is the same | |
| 15513 * thing as deleting a character from both goodword and badword. | |
| 15514 * Use a better score when there is only a case difference. */ | |
| 15515 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 15516 score += SCORE_ICASE; | |
| 15517 else | |
| 15518 { | |
| 15519 /* For a similar character use SCORE_SIMILAR. */ | |
| 15520 if (slang != NULL | |
| 15521 && slang->sl_has_map | |
| 15522 && similar_chars(slang, gc, bc)) | |
| 15523 score += SCORE_SIMILAR; | |
| 15524 else | |
| 15525 score += SCORE_SUBST; | |
| 15526 } | |
| 15527 | |
| 15528 if (score < minscore) | |
| 15529 { | |
| 15530 /* Do the substitution. */ | |
| 15531 ++gi; | |
| 15532 ++bi; | |
| 15533 continue; | |
| 15534 } | |
| 15535 } | |
| 15536 pop: | |
| 15537 /* | |
| 15538 * Get here to try the next alternative, pop it from the stack. | |
| 15539 */ | |
| 15540 if (stackidx == 0) /* stack is empty, finished */ | |
| 15541 break; | |
| 15542 | |
| 15543 /* pop an item from the stack */ | |
| 15544 --stackidx; | |
| 15545 gi = stack[stackidx].goodi; | |
| 15546 bi = stack[stackidx].badi; | |
| 15547 score = stack[stackidx].score; | |
| 15548 } | |
| 15549 | |
| 15550 /* When the score goes over "limit" it may actually be much higher. | |
| 15551 * Return a very large number to avoid going below the limit when giving a | |
| 15552 * bonus. */ | |
| 15553 if (minscore > limit) | |
| 15554 return SCORE_MAXMAX; | |
| 15555 return minscore; | |
| 15556 } | |
| 15557 #endif | |
| 15558 | |
| 714 | 15559 /* |
| 15560 * ":spellinfo" | |
| 15561 */ | |
| 15562 void | |
| 15563 ex_spellinfo(eap) | |
| 1883 | 15564 exarg_T *eap UNUSED; |
| 714 | 15565 { |
| 15566 int lpi; | |
| 15567 langp_T *lp; | |
| 15568 char_u *p; | |
| 15569 | |
| 15570 if (no_spell_checking(curwin)) | |
| 15571 return; | |
| 15572 | |
| 15573 msg_start(); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15574 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len && !got_int; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15575 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15576 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 714 | 15577 msg_puts((char_u *)"file: "); |
| 15578 msg_puts(lp->lp_slang->sl_fname); | |
| 15579 msg_putchar('\n'); | |
| 15580 p = lp->lp_slang->sl_info; | |
| 15581 if (p != NULL) | |
| 15582 { | |
| 15583 msg_puts(p); | |
| 15584 msg_putchar('\n'); | |
| 15585 } | |
| 15586 } | |
| 15587 msg_end(); | |
| 15588 } | |
| 15589 | |
| 625 | 15590 #define DUMPFLAG_KEEPCASE 1 /* round 2: keep-case tree */ |
| 15591 #define DUMPFLAG_COUNT 2 /* include word count */ | |
| 701 | 15592 #define DUMPFLAG_ICASE 4 /* ignore case when finding matches */ |
| 710 | 15593 #define DUMPFLAG_ONECAP 8 /* pattern starts with capital */ |
| 15594 #define DUMPFLAG_ALLCAP 16 /* pattern is all capitals */ | |
| 625 | 15595 |
| 351 | 15596 /* |
| 15597 * ":spelldump" | |
| 15598 */ | |
| 15599 void | |
| 15600 ex_spelldump(eap) | |
| 15601 exarg_T *eap; | |
| 15602 { | |
| 5382 | 15603 char_u *spl; |
| 15604 long dummy; | |
| 15605 | |
| 701 | 15606 if (no_spell_checking(curwin)) |
| 15607 return; | |
| 5382 | 15608 get_option_value((char_u*)"spl", &dummy, &spl, OPT_LOCAL); |
| 15609 | |
| 15610 /* Create a new empty buffer in a new window. */ | |
| 701 | 15611 do_cmdline_cmd((char_u *)"new"); |
| 5382 | 15612 |
| 15613 /* enable spelling locally in the new window */ | |
| 15614 set_option_value((char_u*)"spell", TRUE, (char_u*)"", OPT_LOCAL); | |
| 15615 set_option_value((char_u*)"spl", dummy, spl, OPT_LOCAL); | |
| 15616 vim_free(spl); | |
| 15617 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15618 if (!bufempty() || !buf_valid(curbuf)) |
| 701 | 15619 return; |
| 15620 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15621 spell_dump_compl(NULL, 0, NULL, eap->forceit ? DUMPFLAG_COUNT : 0); |
| 701 | 15622 |
| 15623 /* Delete the empty line that we started with. */ | |
| 15624 if (curbuf->b_ml.ml_line_count > 1) | |
| 15625 ml_delete(curbuf->b_ml.ml_line_count, FALSE); | |
| 15626 | |
| 15627 redraw_later(NOT_VALID); | |
| 15628 } | |
| 15629 | |
| 15630 /* | |
| 15631 * Go through all possible words and: | |
| 15632 * 1. When "pat" is NULL: dump a list of all words in the current buffer. | |
| 15633 * "ic" and "dir" are not used. | |
| 15634 * 2. When "pat" is not NULL: add matching words to insert mode completion. | |
| 15635 */ | |
| 15636 void | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15637 spell_dump_compl(pat, ic, dir, dumpflags_arg) |
| 701 | 15638 char_u *pat; /* leading part of the word */ |
| 15639 int ic; /* ignore case */ | |
| 15640 int *dir; /* direction for adding matches */ | |
| 15641 int dumpflags_arg; /* DUMPFLAG_* */ | |
| 15642 { | |
| 351 | 15643 langp_T *lp; |
| 15644 slang_T *slang; | |
| 15645 idx_T arridx[MAXWLEN]; | |
| 15646 int curi[MAXWLEN]; | |
| 15647 char_u word[MAXWLEN]; | |
| 15648 int c; | |
| 15649 char_u *byts; | |
| 15650 idx_T *idxs; | |
| 15651 linenr_T lnum = 0; | |
| 15652 int round; | |
| 15653 int depth; | |
| 15654 int n; | |
| 15655 int flags; | |
| 381 | 15656 char_u *region_names = NULL; /* region names being used */ |
| 15657 int do_region = TRUE; /* dump region names and numbers */ | |
| 15658 char_u *p; | |
| 500 | 15659 int lpi; |
| 701 | 15660 int dumpflags = dumpflags_arg; |
| 15661 int patlen; | |
| 15662 | |
| 710 | 15663 /* When ignoring case or when the pattern starts with capital pass this on |
| 15664 * to dump_word(). */ | |
| 15665 if (pat != NULL) | |
| 15666 { | |
| 15667 if (ic) | |
| 15668 dumpflags |= DUMPFLAG_ICASE; | |
| 15669 else | |
| 15670 { | |
| 15671 n = captype(pat, NULL); | |
| 15672 if (n == WF_ONECAP) | |
| 15673 dumpflags |= DUMPFLAG_ONECAP; | |
| 15674 else if (n == WF_ALLCAP | |
| 15675 #ifdef FEAT_MBYTE | |
| 714 | 15676 && (int)STRLEN(pat) > mb_ptr2len(pat) |
| 710 | 15677 #else |
| 714 | 15678 && (int)STRLEN(pat) > 1 |
| 710 | 15679 #endif |
| 15680 ) | |
| 15681 dumpflags |= DUMPFLAG_ALLCAP; | |
| 15682 } | |
| 15683 } | |
| 351 | 15684 |
| 381 | 15685 /* Find out if we can support regions: All languages must support the same |
| 15686 * regions or none at all. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15687 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15688 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15689 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 381 | 15690 p = lp->lp_slang->sl_regions; |
| 15691 if (p[0] != 0) | |
| 15692 { | |
| 15693 if (region_names == NULL) /* first language with regions */ | |
| 15694 region_names = p; | |
| 15695 else if (STRCMP(region_names, p) != 0) | |
| 15696 { | |
| 15697 do_region = FALSE; /* region names are different */ | |
| 15698 break; | |
| 15699 } | |
| 15700 } | |
| 15701 } | |
| 15702 | |
| 15703 if (do_region && region_names != NULL) | |
| 15704 { | |
| 701 | 15705 if (pat == NULL) |
| 15706 { | |
| 15707 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names); | |
| 15708 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 15709 } | |
| 381 | 15710 } |
| 15711 else | |
| 15712 do_region = FALSE; | |
| 15713 | |
| 15714 /* | |
| 15715 * Loop over all files loaded for the entries in 'spelllang'. | |
| 15716 */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15717 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15718 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15719 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 351 | 15720 slang = lp->lp_slang; |
| 500 | 15721 if (slang->sl_fbyts == NULL) /* reloading failed */ |
| 15722 continue; | |
| 351 | 15723 |
| 701 | 15724 if (pat == NULL) |
| 15725 { | |
| 15726 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname); | |
| 15727 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 15728 } | |
| 15729 | |
| 15730 /* When matching with a pattern and there are no prefixes only use | |
| 15731 * parts of the tree that match "pat". */ | |
| 15732 if (pat != NULL && slang->sl_pbyts == NULL) | |
| 835 | 15733 patlen = (int)STRLEN(pat); |
| 701 | 15734 else |
| 840 | 15735 patlen = -1; |
| 351 | 15736 |
| 15737 /* round 1: case-folded tree | |
| 15738 * round 2: keep-case tree */ | |
| 15739 for (round = 1; round <= 2; ++round) | |
| 15740 { | |
| 15741 if (round == 1) | |
| 15742 { | |
| 701 | 15743 dumpflags &= ~DUMPFLAG_KEEPCASE; |
| 351 | 15744 byts = slang->sl_fbyts; |
| 15745 idxs = slang->sl_fidxs; | |
| 15746 } | |
| 15747 else | |
| 15748 { | |
| 701 | 15749 dumpflags |= DUMPFLAG_KEEPCASE; |
| 351 | 15750 byts = slang->sl_kbyts; |
| 15751 idxs = slang->sl_kidxs; | |
| 15752 } | |
| 15753 if (byts == NULL) | |
| 15754 continue; /* array is empty */ | |
| 15755 | |
| 15756 depth = 0; | |
| 15757 arridx[0] = 0; | |
| 15758 curi[0] = 1; | |
| 701 | 15759 while (depth >= 0 && !got_int |
| 15760 && (pat == NULL || !compl_interrupted)) | |
| 351 | 15761 { |
| 15762 if (curi[depth] > byts[arridx[depth]]) | |
| 15763 { | |
| 15764 /* Done all bytes at this node, go up one level. */ | |
| 15765 --depth; | |
| 15766 line_breakcheck(); | |
| 715 | 15767 ins_compl_check_keys(50); |
| 351 | 15768 } |
| 15769 else | |
| 15770 { | |
| 15771 /* Do one more byte at this node. */ | |
| 15772 n = arridx[depth] + curi[depth]; | |
| 15773 ++curi[depth]; | |
| 15774 c = byts[n]; | |
| 15775 if (c == 0) | |
| 15776 { | |
| 15777 /* End of word, deal with the word. | |
| 15778 * Don't use keep-case words in the fold-case tree, | |
| 15779 * they will appear in the keep-case tree. | |
| 15780 * Only use the word when the region matches. */ | |
| 15781 flags = (int)idxs[n]; | |
| 15782 if ((round == 2 || (flags & WF_KEEPCAP) == 0) | |
| 500 | 15783 && (flags & WF_NEEDCOMP) == 0 |
| 381 | 15784 && (do_region |
| 15785 || (flags & WF_REGION) == 0 | |
| 390 | 15786 || (((unsigned)flags >> 16) |
| 351 | 15787 & lp->lp_region) != 0)) |
| 15788 { | |
| 15789 word[depth] = NUL; | |
| 381 | 15790 if (!do_region) |
| 15791 flags &= ~WF_REGION; | |
| 355 | 15792 |
| 15793 /* Dump the basic word if there is no prefix or | |
| 15794 * when it's the first one. */ | |
| 390 | 15795 c = (unsigned)flags >> 24; |
| 355 | 15796 if (c == 0 || curi[depth] == 2) |
| 701 | 15797 { |
| 15798 dump_word(slang, word, pat, dir, | |
| 15799 dumpflags, flags, lnum); | |
| 15800 if (pat == NULL) | |
| 15801 ++lnum; | |
| 15802 } | |
| 351 | 15803 |
| 15804 /* Apply the prefix, if there is one. */ | |
| 355 | 15805 if (c != 0) |
| 701 | 15806 lnum = dump_prefixes(slang, word, pat, dir, |
| 15807 dumpflags, flags, lnum); | |
| 351 | 15808 } |
| 15809 } | |
| 15810 else | |
| 15811 { | |
| 15812 /* Normal char, go one level deeper. */ | |
| 15813 word[depth++] = c; | |
| 15814 arridx[depth] = idxs[n]; | |
| 15815 curi[depth] = 1; | |
| 701 | 15816 |
| 15817 /* Check if this characters matches with the pattern. | |
| 15818 * If not skip the whole tree below it. | |
| 710 | 15819 * Always ignore case here, dump_word() will check |
| 15820 * proper case later. This isn't exactly right when | |
| 15821 * length changes for multi-byte characters with | |
| 15822 * ignore case... */ | |
| 15823 if (depth <= patlen | |
| 15824 && MB_STRNICMP(word, pat, depth) != 0) | |
| 701 | 15825 --depth; |
| 351 | 15826 } |
| 15827 } | |
| 15828 } | |
| 15829 } | |
| 15830 } | |
| 15831 } | |
| 15832 | |
| 15833 /* | |
| 15834 * Dump one word: apply case modifications and append a line to the buffer. | |
| 701 | 15835 * When "lnum" is zero add insert mode completion. |
| 15836 */ | |
| 15837 static void | |
| 710 | 15838 dump_word(slang, word, pat, dir, dumpflags, wordflags, lnum) |
| 625 | 15839 slang_T *slang; |
| 351 | 15840 char_u *word; |
| 701 | 15841 char_u *pat; |
| 15842 int *dir; | |
| 625 | 15843 int dumpflags; |
| 710 | 15844 int wordflags; |
| 351 | 15845 linenr_T lnum; |
| 15846 { | |
| 15847 int keepcap = FALSE; | |
| 15848 char_u *p; | |
| 625 | 15849 char_u *tw; |
| 351 | 15850 char_u cword[MAXWLEN]; |
| 381 | 15851 char_u badword[MAXWLEN + 10]; |
| 15852 int i; | |
| 710 | 15853 int flags = wordflags; |
| 15854 | |
| 15855 if (dumpflags & DUMPFLAG_ONECAP) | |
| 15856 flags |= WF_ONECAP; | |
| 15857 if (dumpflags & DUMPFLAG_ALLCAP) | |
| 15858 flags |= WF_ALLCAP; | |
| 351 | 15859 |
| 625 | 15860 if ((dumpflags & DUMPFLAG_KEEPCASE) == 0 && (flags & WF_CAPMASK) != 0) |
| 351 | 15861 { |
| 15862 /* Need to fix case according to "flags". */ | |
| 15863 make_case_word(word, cword, flags); | |
| 15864 p = cword; | |
| 15865 } | |
| 15866 else | |
| 15867 { | |
| 15868 p = word; | |
| 625 | 15869 if ((dumpflags & DUMPFLAG_KEEPCASE) |
| 15870 && ((captype(word, NULL) & WF_KEEPCAP) == 0 | |
| 389 | 15871 || (flags & WF_FIXCAP) != 0)) |
| 351 | 15872 keepcap = TRUE; |
| 15873 } | |
| 625 | 15874 tw = p; |
| 351 | 15875 |
| 701 | 15876 if (pat == NULL) |
| 15877 { | |
| 15878 /* Add flags and regions after a slash. */ | |
| 15879 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap) | |
| 15880 { | |
| 15881 STRCPY(badword, p); | |
| 15882 STRCAT(badword, "/"); | |
| 15883 if (keepcap) | |
| 15884 STRCAT(badword, "="); | |
| 15885 if (flags & WF_BANNED) | |
| 15886 STRCAT(badword, "!"); | |
| 15887 else if (flags & WF_RARE) | |
| 15888 STRCAT(badword, "?"); | |
| 15889 if (flags & WF_REGION) | |
| 15890 for (i = 0; i < 7; ++i) | |
| 15891 if (flags & (0x10000 << i)) | |
| 15892 sprintf((char *)badword + STRLEN(badword), "%d", i + 1); | |
| 15893 p = badword; | |
| 15894 } | |
| 15895 | |
| 15896 if (dumpflags & DUMPFLAG_COUNT) | |
| 15897 { | |
| 15898 hashitem_T *hi; | |
| 15899 | |
| 15900 /* Include the word count for ":spelldump!". */ | |
| 15901 hi = hash_find(&slang->sl_wordcount, tw); | |
| 15902 if (!HASHITEM_EMPTY(hi)) | |
| 15903 { | |
| 15904 vim_snprintf((char *)IObuff, IOSIZE, "%s\t%d", | |
| 625 | 15905 tw, HI2WC(hi)->wc_count); |
| 701 | 15906 p = IObuff; |
| 15907 } | |
| 15908 } | |
| 15909 | |
| 15910 ml_append(lnum, p, (colnr_T)0, FALSE); | |
| 15911 } | |
| 710 | 15912 else if (((dumpflags & DUMPFLAG_ICASE) |
| 15913 ? MB_STRNICMP(p, pat, STRLEN(pat)) == 0 | |
| 15914 : STRNCMP(p, pat, STRLEN(pat)) == 0) | |
| 701 | 15915 && ins_compl_add_infercase(p, (int)STRLEN(p), |
| 942 | 15916 p_ic, NULL, *dir, 0) == OK) |
| 710 | 15917 /* if dir was BACKWARD then honor it just once */ |
| 15918 *dir = FORWARD; | |
| 351 | 15919 } |
| 15920 | |
| 15921 /* | |
| 372 | 15922 * For ":spelldump": Find matching prefixes for "word". Prepend each to |
| 15923 * "word" and append a line to the buffer. | |
| 701 | 15924 * When "lnum" is zero add insert mode completion. |
| 351 | 15925 * Return the updated line number. |
| 15926 */ | |
| 15927 static linenr_T | |
| 701 | 15928 dump_prefixes(slang, word, pat, dir, dumpflags, flags, startlnum) |
| 351 | 15929 slang_T *slang; |
| 15930 char_u *word; /* case-folded word */ | |
| 701 | 15931 char_u *pat; |
| 15932 int *dir; | |
| 625 | 15933 int dumpflags; |
| 351 | 15934 int flags; /* flags with prefix ID */ |
| 15935 linenr_T startlnum; | |
| 15936 { | |
| 15937 idx_T arridx[MAXWLEN]; | |
| 15938 int curi[MAXWLEN]; | |
| 15939 char_u prefix[MAXWLEN]; | |
| 455 | 15940 char_u word_up[MAXWLEN]; |
| 15941 int has_word_up = FALSE; | |
| 351 | 15942 int c; |
| 15943 char_u *byts; | |
| 15944 idx_T *idxs; | |
| 15945 linenr_T lnum = startlnum; | |
| 15946 int depth; | |
| 15947 int n; | |
| 15948 int len; | |
| 15949 int i; | |
| 15950 | |
| 626 | 15951 /* If the word starts with a lower-case letter make the word with an |
| 455 | 15952 * upper-case letter in word_up[]. */ |
| 15953 c = PTR2CHAR(word); | |
| 15954 if (SPELL_TOUPPER(c) != c) | |
| 15955 { | |
| 15956 onecap_copy(word, word_up, TRUE); | |
| 15957 has_word_up = TRUE; | |
| 15958 } | |
| 15959 | |
| 351 | 15960 byts = slang->sl_pbyts; |
| 15961 idxs = slang->sl_pidxs; | |
| 15962 if (byts != NULL) /* array not is empty */ | |
| 15963 { | |
| 15964 /* | |
| 15965 * Loop over all prefixes, building them byte-by-byte in prefix[]. | |
| 390 | 15966 * When at the end of a prefix check that it supports "flags". |
| 351 | 15967 */ |
| 15968 depth = 0; | |
| 15969 arridx[0] = 0; | |
| 15970 curi[0] = 1; | |
| 15971 while (depth >= 0 && !got_int) | |
| 15972 { | |
| 390 | 15973 n = arridx[depth]; |
| 15974 len = byts[n]; | |
| 15975 if (curi[depth] > len) | |
| 351 | 15976 { |
| 15977 /* Done all bytes at this node, go up one level. */ | |
| 15978 --depth; | |
| 15979 line_breakcheck(); | |
| 15980 } | |
| 15981 else | |
| 15982 { | |
| 15983 /* Do one more byte at this node. */ | |
| 390 | 15984 n += curi[depth]; |
| 351 | 15985 ++curi[depth]; |
| 15986 c = byts[n]; | |
| 15987 if (c == 0) | |
| 15988 { | |
| 15989 /* End of prefix, find out how many IDs there are. */ | |
| 15990 for (i = 1; i < len; ++i) | |
| 15991 if (byts[n + i] != 0) | |
| 15992 break; | |
| 15993 curi[depth] += i - 1; | |
| 15994 | |
| 455 | 15995 c = valid_word_prefix(i, n, flags, word, slang, FALSE); |
| 15996 if (c != 0) | |
| 351 | 15997 { |
| 376 | 15998 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1); |
| 701 | 15999 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 16000 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 16001 : flags, lnum); |
| 16002 if (lnum != 0) | |
| 16003 ++lnum; | |
| 351 | 16004 } |
| 455 | 16005 |
| 16006 /* Check for prefix that matches the word when the | |
| 16007 * first letter is upper-case, but only if the prefix has | |
| 16008 * a condition. */ | |
| 16009 if (has_word_up) | |
| 16010 { | |
| 16011 c = valid_word_prefix(i, n, flags, word_up, slang, | |
| 16012 TRUE); | |
| 16013 if (c != 0) | |
| 16014 { | |
| 16015 vim_strncpy(prefix + depth, word_up, | |
| 16016 MAXWLEN - depth - 1); | |
| 701 | 16017 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 16018 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 16019 : flags, lnum); |
| 16020 if (lnum != 0) | |
| 16021 ++lnum; | |
| 455 | 16022 } |
| 16023 } | |
| 351 | 16024 } |
| 16025 else | |
| 16026 { | |
| 16027 /* Normal char, go one level deeper. */ | |
| 16028 prefix[depth++] = c; | |
| 16029 arridx[depth] = idxs[n]; | |
| 16030 curi[depth] = 1; | |
| 16031 } | |
| 16032 } | |
| 16033 } | |
| 16034 } | |
| 16035 | |
| 16036 return lnum; | |
| 16037 } | |
| 16038 | |
| 498 | 16039 /* |
| 626 | 16040 * Move "p" to the end of word "start". |
| 16041 * Uses the spell-checking word characters. | |
| 498 | 16042 */ |
| 16043 char_u * | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
16044 spell_to_word_end(start, win) |
| 498 | 16045 char_u *start; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
16046 win_T *win; |
| 498 | 16047 { |
| 16048 char_u *p = start; | |
| 16049 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
16050 while (*p != NUL && spell_iswordp(p, win)) |
| 498 | 16051 mb_ptr_adv(p); |
| 16052 return p; | |
| 16053 } | |
| 16054 | |
| 475 | 16055 #if defined(FEAT_INS_EXPAND) || defined(PROTO) |
| 16056 /* | |
| 626 | 16057 * For Insert mode completion CTRL-X s: |
| 16058 * Find start of the word in front of column "startcol". | |
| 16059 * We don't check if it is badly spelled, with completion we can only change | |
| 16060 * the word in front of the cursor. | |
| 475 | 16061 * Returns the column number of the word. |
| 16062 */ | |
| 16063 int | |
| 16064 spell_word_start(startcol) | |
| 16065 int startcol; | |
| 16066 { | |
| 16067 char_u *line; | |
| 16068 char_u *p; | |
| 16069 int col = 0; | |
| 16070 | |
| 498 | 16071 if (no_spell_checking(curwin)) |
| 475 | 16072 return startcol; |
| 16073 | |
| 16074 /* Find a word character before "startcol". */ | |
| 16075 line = ml_get_curline(); | |
| 16076 for (p = line + startcol; p > line; ) | |
| 16077 { | |
| 16078 mb_ptr_back(line, p); | |
| 5477 | 16079 if (spell_iswordp_nmw(p, curwin)) |
| 475 | 16080 break; |
| 16081 } | |
| 16082 | |
| 16083 /* Go back to start of the word. */ | |
| 16084 while (p > line) | |
| 16085 { | |
| 835 | 16086 col = (int)(p - line); |
| 475 | 16087 mb_ptr_back(line, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
16088 if (!spell_iswordp(p, curwin)) |
| 475 | 16089 break; |
| 16090 col = 0; | |
| 16091 } | |
| 16092 | |
| 535 | 16093 return col; |
| 16094 } | |
| 16095 | |
| 16096 /* | |
| 16097 * Need to check for 'spellcapcheck' now, the word is removed before | |
| 16098 * expand_spelling() is called. Therefore the ugly global variable. | |
| 16099 */ | |
| 16100 static int spell_expand_need_cap; | |
| 16101 | |
| 16102 void | |
| 16103 spell_expand_check_cap(col) | |
| 16104 colnr_T col; | |
| 16105 { | |
| 475 | 16106 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col); |
| 16107 } | |
| 16108 | |
| 16109 /* | |
| 16110 * Get list of spelling suggestions. | |
| 16111 * Used for Insert mode completion CTRL-X ?. | |
| 16112 * Returns the number of matches. The matches are in "matchp[]", array of | |
| 16113 * allocated strings. | |
| 16114 */ | |
| 16115 int | |
| 1872 | 16116 expand_spelling(lnum, pat, matchp) |
| 1883 | 16117 linenr_T lnum UNUSED; |
| 475 | 16118 char_u *pat; |
| 16119 char_u ***matchp; | |
| 16120 { | |
| 16121 garray_T ga; | |
| 16122 | |
| 625 | 16123 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap, TRUE); |
| 475 | 16124 *matchp = ga.ga_data; |
| 16125 return ga.ga_len; | |
| 16126 } | |
| 16127 #endif | |
| 16128 | |
| 737 | 16129 #endif /* FEAT_SPELL */ |