Mercurial > vim
annotate src/spell.c @ 6375:27a36d1013a6 v7.4.519
updated for version 7.4.519
Problem: Crash when using syntax highlighting.
Solution: When regprog is freed and replaced, store the result.
| author | Bram Moolenaar <bram@vim.org> |
|---|---|
| date | Wed, 19 Nov 2014 16:38:07 +0100 |
| parents | 0d2c821cdc25 |
| children | f28c171348fb |
| 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 | |
| 323 | 314 #define MAXWLEN 250 /* Assume max. word len is this many bytes. |
| 315 Some places assume a word length fits in a | |
| 316 byte, thus it can't be above 255. */ | |
| 226 | 317 |
| 490 | 318 /* Type used for indexes in the word tree need to be at least 4 bytes. If int |
| 324 | 319 * is 8 bytes we could use something smaller, but what? */ |
| 5684 | 320 #if VIM_SIZEOF_INT > 3 |
| 324 | 321 typedef int idx_T; |
| 322 #else | |
| 323 typedef long idx_T; | |
| 324 #endif | |
| 325 | |
| 2660 | 326 #ifdef VMS |
| 327 # define SPL_FNAME_TMPL "%s_%s.spl" | |
| 328 # define SPL_FNAME_ADD "_add." | |
| 329 # define SPL_FNAME_ASCII "_ascii." | |
| 330 #else | |
| 331 # define SPL_FNAME_TMPL "%s.%s.spl" | |
| 332 # define SPL_FNAME_ADD ".add." | |
| 333 # define SPL_FNAME_ASCII ".ascii." | |
| 334 #endif | |
| 335 | |
| 324 | 336 /* Flags used for a word. Only the lowest byte can be used, the region byte |
| 337 * comes above it. */ | |
| 300 | 338 #define WF_REGION 0x01 /* region byte follows */ |
| 339 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */ | |
| 340 #define WF_ALLCAP 0x04 /* word must be all capitals */ | |
| 341 #define WF_RARE 0x08 /* rare word */ | |
| 307 | 342 #define WF_BANNED 0x10 /* bad word */ |
| 481 | 343 #define WF_AFX 0x20 /* affix ID follows */ |
| 389 | 344 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */ |
| 323 | 345 #define WF_KEEPCAP 0x80 /* keep-case word */ |
| 346 | |
| 390 | 347 /* for <flags2>, shifted up one byte to be used in wn_flags */ |
| 348 #define WF_HAS_AFF 0x0100 /* word includes affix */ | |
| 500 | 349 #define WF_NEEDCOMP 0x0200 /* word only valid in compound */ |
| 698 | 350 #define WF_NOSUGGEST 0x0400 /* word not to be suggested */ |
| 809 | 351 #define WF_COMPROOT 0x0800 /* already compounded word, COMPOUNDROOT */ |
| 819 | 352 #define WF_NOCOMPBEF 0x1000 /* no compounding before this word */ |
| 353 #define WF_NOCOMPAFT 0x2000 /* no compounding after this word */ | |
| 390 | 354 |
| 633 | 355 /* only used for su_badflags */ |
| 356 #define WF_MIXCAP 0x20 /* mix of upper and lower case: macaRONI */ | |
| 357 | |
| 389 | 358 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP) |
| 300 | 359 |
| 455 | 360 /* flags for <pflags> */ |
| 822 | 361 #define WFP_RARE 0x01 /* rare prefix */ |
| 362 #define WFP_NC 0x02 /* prefix is not combining */ | |
| 363 #define WFP_UP 0x04 /* to-upper prefix */ | |
| 364 #define WFP_COMPPERMIT 0x08 /* prefix with COMPOUNDPERMITFLAG */ | |
| 365 #define WFP_COMPFORBID 0x10 /* prefix with COMPOUNDFORBIDFLAG */ | |
| 366 | |
| 367 /* Flags for postponed prefixes in "sl_pidxs". Must be above affixID (one | |
| 368 * byte) and prefcondnr (two bytes). */ | |
| 369 #define WF_RAREPFX (WFP_RARE << 24) /* rare postponed prefix */ | |
| 370 #define WF_PFX_NC (WFP_NC << 24) /* non-combining postponed prefix */ | |
| 371 #define WF_PFX_UP (WFP_UP << 24) /* to-upper postponed prefix */ | |
| 372 #define WF_PFX_COMPPERMIT (WFP_COMPPERMIT << 24) /* postponed prefix with | |
| 373 * COMPOUNDPERMITFLAG */ | |
| 374 #define WF_PFX_COMPFORBID (WFP_COMPFORBID << 24) /* postponed prefix with | |
| 375 * COMPOUNDFORBIDFLAG */ | |
| 376 | |
| 390 | 377 |
| 809 | 378 /* flags for <compoptions> */ |
| 379 #define COMP_CHECKDUP 1 /* CHECKCOMPOUNDDUP */ | |
| 380 #define COMP_CHECKREP 2 /* CHECKCOMPOUNDREP */ | |
| 381 #define COMP_CHECKCASE 4 /* CHECKCOMPOUNDCASE */ | |
| 382 #define COMP_CHECKTRIPLE 8 /* CHECKCOMPOUNDTRIPLE */ | |
| 383 | |
| 390 | 384 /* Special byte values for <byte>. Some are only used in the tree for |
| 385 * postponed prefixes, some only in the other trees. This is a bit messy... */ | |
| 386 #define BY_NOFLAGS 0 /* end of word without flags or region; for | |
| 455 | 387 * postponed prefix: no <pflags> */ |
| 388 #define BY_INDEX 1 /* child is shared, index follows */ | |
| 389 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for | |
| 390 * postponed prefix: <pflags> follows */ | |
| 391 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes | |
| 392 * follow; never used in prefix tree */ | |
| 393 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */ | |
| 236 | 394 |
| 625 | 395 /* Info from "REP", "REPSAL" and "SAL" entries in ".aff" file used in si_rep, |
| 396 * si_repsal, sl_rep, and si_sal. Not for sl_sal! | |
| 323 | 397 * One replacement: from "ft_from" to "ft_to". */ |
| 398 typedef struct fromto_S | |
| 236 | 399 { |
| 323 | 400 char_u *ft_from; |
| 401 char_u *ft_to; | |
| 402 } fromto_T; | |
| 236 | 403 |
| 344 | 404 /* Info from "SAL" entries in ".aff" file used in sl_sal. |
| 405 * The info is split for quick processing by spell_soundfold(). | |
| 406 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */ | |
| 407 typedef struct salitem_S | |
| 408 { | |
| 409 char_u *sm_lead; /* leading letters */ | |
| 410 int sm_leadlen; /* length of "sm_lead" */ | |
| 375 | 411 char_u *sm_oneof; /* letters from () or NULL */ |
| 344 | 412 char_u *sm_rules; /* rules like ^, $, priority */ |
| 413 char_u *sm_to; /* replacement. */ | |
| 372 | 414 #ifdef FEAT_MBYTE |
| 415 int *sm_lead_w; /* wide character copy of "sm_lead" */ | |
| 375 | 416 int *sm_oneof_w; /* wide character copy of "sm_oneof" */ |
| 372 | 417 int *sm_to_w; /* wide character copy of "sm_to" */ |
| 418 #endif | |
| 344 | 419 } salitem_T; |
| 420 | |
| 375 | 421 #ifdef FEAT_MBYTE |
| 422 typedef int salfirst_T; | |
| 423 #else | |
| 424 typedef short salfirst_T; | |
| 425 #endif | |
| 426 | |
| 483 | 427 /* Values for SP_*ERROR are negative, positive values are used by |
| 428 * read_cnt_string(). */ | |
| 429 #define SP_TRUNCERROR -1 /* spell file truncated error */ | |
| 430 #define SP_FORMERROR -2 /* format error in spell file */ | |
| 495 | 431 #define SP_OTHERERROR -3 /* other error while reading spell file */ |
| 483 | 432 |
| 236 | 433 /* |
| 243 | 434 * Structure used to store words and other info for one language, loaded from |
| 435 * a .spl file. | |
| 300 | 436 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the |
| 437 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words. | |
| 438 * | |
| 439 * The "byts" array stores the possible bytes in each tree node, preceded by | |
| 440 * the number of possible bytes, sorted on byte value: | |
| 441 * <len> <byte1> <byte2> ... | |
| 442 * The "idxs" array stores the index of the child node corresponding to the | |
| 443 * byte in "byts". | |
| 444 * Exception: when the byte is zero, the word may end here and "idxs" holds | |
| 481 | 445 * the flags, region mask and affixID for the word. There may be several |
| 446 * zeros in sequence for alternative flag/region/affixID combinations. | |
| 236 | 447 */ |
| 448 typedef struct slang_S slang_T; | |
| 449 struct slang_S | |
| 450 { | |
| 451 slang_T *sl_next; /* next language */ | |
| 452 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */ | |
| 310 | 453 char_u *sl_fname; /* name of .spl file */ |
| 323 | 454 int sl_add; /* TRUE if it's a .add file. */ |
| 339 | 455 |
| 300 | 456 char_u *sl_fbyts; /* case-folded word bytes */ |
| 324 | 457 idx_T *sl_fidxs; /* case-folded word indexes */ |
| 300 | 458 char_u *sl_kbyts; /* keep-case word bytes */ |
| 324 | 459 idx_T *sl_kidxs; /* keep-case word indexes */ |
| 339 | 460 char_u *sl_pbyts; /* prefix tree word bytes */ |
| 461 idx_T *sl_pidxs; /* prefix tree word indexes */ | |
| 462 | |
| 714 | 463 char_u *sl_info; /* infotext string or NULL */ |
| 464 | |
| 236 | 465 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */ |
| 323 | 466 |
| 376 | 467 char_u *sl_midword; /* MIDWORD string or NULL */ |
| 468 | |
| 625 | 469 hashtab_T sl_wordcount; /* hashtable with word count, wordcount_T */ |
| 470 | |
| 809 | 471 int sl_compmax; /* COMPOUNDWORDMAX (default: MAXWLEN) */ |
| 501 | 472 int sl_compminlen; /* COMPOUNDMIN (default: 0) */ |
| 483 | 473 int sl_compsylmax; /* COMPOUNDSYLMAX (default: MAXWLEN) */ |
| 809 | 474 int sl_compoptions; /* COMP_* flags */ |
| 475 garray_T sl_comppat; /* CHECKCOMPOUNDPATTERN items */ | |
| 714 | 476 regprog_T *sl_compprog; /* COMPOUNDRULE turned into a regexp progrm |
| 483 | 477 * (NULL when no compounding) */ |
| 1762 | 478 char_u *sl_comprules; /* all COMPOUNDRULE concatenated (or NULL) */ |
| 483 | 479 char_u *sl_compstartflags; /* flags for first compound word */ |
| 485 | 480 char_u *sl_compallflags; /* all flags for compound words */ |
| 492 | 481 char_u sl_nobreak; /* When TRUE: no spaces between words */ |
| 483 | 482 char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */ |
| 483 garray_T sl_syl_items; /* syllable items */ | |
| 481 | 484 |
| 339 | 485 int sl_prefixcnt; /* number of items in "sl_prefprog" */ |
| 486 regprog_T **sl_prefprog; /* table with regprogs for prefixes */ | |
| 487 | |
| 323 | 488 garray_T sl_rep; /* list of fromto_T entries from REP lines */ |
| 489 short sl_rep_first[256]; /* indexes where byte first appears, -1 if | |
| 490 there is none */ | |
| 344 | 491 garray_T sl_sal; /* list of salitem_T entries from SAL lines */ |
| 375 | 492 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if |
| 323 | 493 there is none */ |
| 625 | 494 int sl_followup; /* SAL followup */ |
| 495 int sl_collapse; /* SAL collapse_result */ | |
| 496 int sl_rem_accents; /* SAL remove_accents */ | |
| 375 | 497 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items: |
| 498 * "sl_sal_first" maps chars, when has_mbyte | |
| 499 * "sl_sal" is a list of wide char lists. */ | |
| 625 | 500 garray_T sl_repsal; /* list of fromto_T entries from REPSAL lines */ |
| 501 short sl_repsal_first[256]; /* sl_rep_first for REPSAL lines */ | |
| 698 | 502 int sl_nosplitsugs; /* don't suggest splitting a word */ |
| 625 | 503 |
| 504 /* Info from the .sug file. Loaded on demand. */ | |
| 505 time_t sl_sugtime; /* timestamp for .sug file */ | |
| 506 char_u *sl_sbyts; /* soundfolded word bytes */ | |
| 507 idx_T *sl_sidxs; /* soundfolded word indexes */ | |
| 508 buf_T *sl_sugbuf; /* buffer with word number table */ | |
| 509 int sl_sugloaded; /* TRUE when .sug file was loaded or failed to | |
| 510 load */ | |
| 511 | |
| 330 | 512 int sl_has_map; /* TRUE if there is a MAP line */ |
| 513 #ifdef FEAT_MBYTE | |
| 514 hashtab_T sl_map_hash; /* MAP for multi-byte chars */ | |
| 515 int sl_map_array[256]; /* MAP for first 256 chars */ | |
| 516 #else | |
| 517 char_u sl_map_array[256]; /* MAP for first 256 chars */ | |
| 518 #endif | |
| 625 | 519 hashtab_T sl_sounddone; /* table with soundfolded words that have |
| 520 handled, see add_sound_suggest() */ | |
| 236 | 521 }; |
| 522 | |
| 243 | 523 /* First language that is loaded, start of the linked list of loaded |
| 524 * languages. */ | |
| 236 | 525 static slang_T *first_lang = NULL; |
| 526 | |
| 323 | 527 /* Flags used in .spl file for soundsalike flags. */ |
| 528 #define SAL_F0LLOWUP 1 | |
| 529 #define SAL_COLLAPSE 2 | |
| 530 #define SAL_REM_ACCENTS 4 | |
| 531 | |
| 236 | 532 /* |
| 533 * Structure used in "b_langp", filled from 'spelllang'. | |
| 534 */ | |
| 535 typedef struct langp_S | |
| 536 { | |
| 503 | 537 slang_T *lp_slang; /* info for this language */ |
| 538 slang_T *lp_sallang; /* language used for sound folding or NULL */ | |
| 539 slang_T *lp_replang; /* language used for REP items or NULL */ | |
| 236 | 540 int lp_region; /* bitmask for region or REGION_ALL */ |
| 541 } langp_T; | |
| 542 | |
| 543 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i)) | |
| 544 | |
| 307 | 545 #define REGION_ALL 0xff /* word valid in all regions */ |
| 546 | |
| 483 | 547 #define VIMSPELLMAGIC "VIMspell" /* string at start of Vim spell file */ |
| 548 #define VIMSPELLMAGICL 8 | |
| 549 #define VIMSPELLVERSION 50 | |
| 550 | |
| 625 | 551 #define VIMSUGMAGIC "VIMsug" /* string at start of Vim .sug file */ |
| 552 #define VIMSUGMAGICL 6 | |
| 553 #define VIMSUGVERSION 1 | |
| 554 | |
| 483 | 555 /* Section IDs. Only renumber them when VIMSPELLVERSION changes! */ |
| 556 #define SN_REGION 0 /* <regionname> section */ | |
| 557 #define SN_CHARFLAGS 1 /* charflags section */ | |
| 558 #define SN_MIDWORD 2 /* <midword> section */ | |
| 559 #define SN_PREFCOND 3 /* <prefcond> section */ | |
| 560 #define SN_REP 4 /* REP items section */ | |
| 561 #define SN_SAL 5 /* SAL items section */ | |
| 562 #define SN_SOFO 6 /* soundfolding section */ | |
| 563 #define SN_MAP 7 /* MAP items section */ | |
| 564 #define SN_COMPOUND 8 /* compound words section */ | |
| 565 #define SN_SYLLABLE 9 /* syllable section */ | |
| 492 | 566 #define SN_NOBREAK 10 /* NOBREAK section */ |
| 625 | 567 #define SN_SUGFILE 11 /* timestamp for .sug file */ |
| 568 #define SN_REPSAL 12 /* REPSAL items section */ | |
| 569 #define SN_WORDS 13 /* common words */ | |
| 698 | 570 #define SN_NOSPLITSUGS 14 /* don't split word for suggestions */ |
| 714 | 571 #define SN_INFO 15 /* info section */ |
| 483 | 572 #define SN_END 255 /* end of sections */ |
| 573 | |
| 574 #define SNF_REQUIRED 1 /* <sectionflags>: required section */ | |
| 575 | |
| 307 | 576 /* Result values. Lower number is accepted over higher one. */ |
| 577 #define SP_BANNED -1 | |
| 236 | 578 #define SP_OK 0 |
| 307 | 579 #define SP_RARE 1 |
| 580 #define SP_LOCAL 2 | |
| 581 #define SP_BAD 3 | |
| 236 | 582 |
| 381 | 583 /* file used for "zG" and "zW" */ |
| 385 | 584 static char_u *int_wordlist = NULL; |
| 381 | 585 |
| 625 | 586 typedef struct wordcount_S |
| 587 { | |
| 588 short_u wc_count; /* nr of times word was seen */ | |
| 589 char_u wc_word[1]; /* word, actually longer */ | |
| 590 } wordcount_T; | |
| 591 | |
| 592 static wordcount_T dumwc; | |
| 835 | 593 #define WC_KEY_OFF (unsigned)(dumwc.wc_word - (char_u *)&dumwc) |
| 625 | 594 #define HI2WC(hi) ((wordcount_T *)((hi)->hi_key - WC_KEY_OFF)) |
| 595 #define MAXWORDCOUNT 0xffff | |
| 596 | |
| 236 | 597 /* |
| 323 | 598 * Information used when looking for suggestions. |
| 599 */ | |
| 600 typedef struct suginfo_S | |
| 601 { | |
| 602 garray_T su_ga; /* suggestions, contains "suggest_T" */ | |
| 344 | 603 int su_maxcount; /* max. number of suggestions displayed */ |
| 323 | 604 int su_maxscore; /* maximum score for adding to su_ga */ |
| 625 | 605 int su_sfmaxscore; /* idem, for when doing soundfold words */ |
| 344 | 606 garray_T su_sga; /* like su_ga, sound-folded scoring */ |
| 323 | 607 char_u *su_badptr; /* start of bad word in line */ |
| 608 int su_badlen; /* length of detected bad word in line */ | |
| 346 | 609 int su_badflags; /* caps flags for bad word */ |
| 323 | 610 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */ |
| 611 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */ | |
| 534 | 612 char_u su_sal_badword[MAXWLEN]; /* su_badword soundfolded */ |
| 323 | 613 hashtab_T su_banned; /* table with banned words */ |
| 503 | 614 slang_T *su_sallang; /* default language for sound folding */ |
| 323 | 615 } suginfo_T; |
| 616 | |
| 617 /* One word suggestion. Used in "si_ga". */ | |
| 618 typedef struct suggest_S | |
| 619 { | |
| 620 char_u *st_word; /* suggested word, allocated string */ | |
| 625 | 621 int st_wordlen; /* STRLEN(st_word) */ |
| 323 | 622 int st_orglen; /* length of replaced text */ |
| 623 int st_score; /* lower is better */ | |
| 344 | 624 int st_altscore; /* used when st_score compares equal */ |
| 625 int st_salscore; /* st_score is for soundalike */ | |
| 324 | 626 int st_had_bonus; /* bonus already included in score */ |
| 503 | 627 slang_T *st_slang; /* language used for sound folding */ |
| 323 | 628 } suggest_T; |
| 629 | |
| 344 | 630 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i]) |
| 323 | 631 |
| 625 | 632 /* TRUE if a word appears in the list of banned words. */ |
| 633 #define WAS_BANNED(su, word) (!HASHITEM_EMPTY(hash_find(&su->su_banned, word))) | |
| 634 | |
| 1705 | 635 /* Number of suggestions kept when cleaning up. We need to keep more than |
| 625 | 636 * what is displayed, because when rescore_suggestions() is called the score |
| 637 * may change and wrong suggestions may be removed later. */ | |
| 638 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 130 ? 150 : (su)->su_maxcount + 20) | |
| 324 | 639 |
| 640 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots | |
| 641 * of suggestions that are not going to be displayed. */ | |
| 625 | 642 #define SUG_MAX_COUNT(su) (SUG_CLEAN_COUNT(su) + 50) |
| 323 | 643 |
| 644 /* score for various changes */ | |
| 344 | 645 #define SCORE_SPLIT 149 /* split bad word */ |
| 698 | 646 #define SCORE_SPLIT_NO 249 /* split bad word with NOSPLITSUGS */ |
| 323 | 647 #define SCORE_ICASE 52 /* slightly different case */ |
| 483 | 648 #define SCORE_REGION 200 /* word is for different region */ |
| 323 | 649 #define SCORE_RARE 180 /* rare word */ |
| 625 | 650 #define SCORE_SWAP 75 /* swap two characters */ |
| 323 | 651 #define SCORE_SWAP3 110 /* swap two characters in three */ |
| 532 | 652 #define SCORE_REP 65 /* REP replacement */ |
| 323 | 653 #define SCORE_SUBST 93 /* substitute a character */ |
| 654 #define SCORE_SIMILAR 33 /* substitute a similar character */ | |
| 479 | 655 #define SCORE_SUBCOMP 33 /* substitute a composing character */ |
| 324 | 656 #define SCORE_DEL 94 /* delete a character */ |
| 532 | 657 #define SCORE_DELDUP 66 /* delete a duplicated character */ |
| 479 | 658 #define SCORE_DELCOMP 28 /* delete a composing character */ |
| 324 | 659 #define SCORE_INS 96 /* insert a character */ |
| 532 | 660 #define SCORE_INSDUP 67 /* insert a duplicate character */ |
| 475 | 661 #define SCORE_INSCOMP 30 /* insert a composing character */ |
| 366 | 662 #define SCORE_NONWORD 103 /* change non-word to word char */ |
| 323 | 663 |
| 372 | 664 #define SCORE_FILE 30 /* suggestion from a file */ |
| 323 | 665 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower. |
| 666 * 350 allows for about three changes. */ | |
| 344 | 667 |
| 625 | 668 #define SCORE_COMMON1 30 /* subtracted for words seen before */ |
| 669 #define SCORE_COMMON2 40 /* subtracted for words often seen */ | |
| 670 #define SCORE_COMMON3 50 /* subtracted for words very often seen */ | |
| 671 #define SCORE_THRES2 10 /* word count threshold for COMMON2 */ | |
| 672 #define SCORE_THRES3 100 /* word count threshold for COMMON3 */ | |
| 673 | |
| 674 /* When trying changed soundfold words it becomes slow when trying more than | |
| 675 * two changes. With less then two changes it's slightly faster but we miss a | |
| 676 * few good suggestions. In rare cases we need to try three of four changes. | |
| 677 */ | |
| 678 #define SCORE_SFMAX1 200 /* maximum score for first try */ | |
| 679 #define SCORE_SFMAX2 300 /* maximum score for second try */ | |
| 680 #define SCORE_SFMAX3 400 /* maximum score for third try */ | |
| 681 | |
| 344 | 682 #define SCORE_BIG SCORE_INS * 3 /* big difference */ |
| 625 | 683 #define SCORE_MAXMAX 999999 /* accept any score */ |
| 684 #define SCORE_LIMITMAX 350 /* for spell_edit_score_limit() */ | |
| 685 | |
| 686 /* for spell_edit_score_limit() we need to know the minimum value of | |
| 687 * SCORE_ICASE, SCORE_SWAP, SCORE_DEL, SCORE_SIMILAR and SCORE_INS */ | |
| 688 #define SCORE_EDIT_MIN SCORE_SIMILAR | |
| 323 | 689 |
| 690 /* | |
| 236 | 691 * Structure to store info for word matching. |
| 692 */ | |
| 693 typedef struct matchinf_S | |
| 694 { | |
| 695 langp_T *mi_lp; /* info for language and region */ | |
| 243 | 696 |
| 697 /* pointers to original text to be checked */ | |
| 236 | 698 char_u *mi_word; /* start of word being checked */ |
| 339 | 699 char_u *mi_end; /* end of matching word so far */ |
| 243 | 700 char_u *mi_fend; /* next char to be added to mi_fword */ |
| 300 | 701 char_u *mi_cend; /* char after what was used for |
| 702 mi_capflags */ | |
| 243 | 703 |
| 704 /* case-folded text */ | |
| 705 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */ | |
| 300 | 706 int mi_fwordlen; /* nr of valid bytes in mi_fword */ |
| 243 | 707 |
| 339 | 708 /* for when checking word after a prefix */ |
| 709 int mi_prefarridx; /* index in sl_pidxs with list of | |
| 481 | 710 affixID/condition */ |
| 339 | 711 int mi_prefcnt; /* number of entries at mi_prefarridx */ |
| 712 int mi_prefixlen; /* byte length of prefix */ | |
| 455 | 713 #ifdef FEAT_MBYTE |
| 714 int mi_cprefixlen; /* byte length of prefix in original | |
| 715 case */ | |
| 716 #else | |
| 717 # define mi_cprefixlen mi_prefixlen /* it's the same value */ | |
| 718 #endif | |
| 339 | 719 |
| 481 | 720 /* for when checking a compound word */ |
| 721 int mi_compoff; /* start of following word offset */ | |
| 483 | 722 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */ |
| 723 int mi_complen; /* nr of compound words used */ | |
| 809 | 724 int mi_compextra; /* nr of COMPOUNDROOT words */ |
| 481 | 725 |
| 243 | 726 /* others */ |
| 236 | 727 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */ |
| 300 | 728 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
|
729 win_T *mi_win; /* buffer being checked */ |
| 492 | 730 |
| 731 /* for NOBREAK */ | |
| 732 int mi_result2; /* "mi_resul" without following word */ | |
| 733 char_u *mi_end2; /* "mi_end" without following word */ | |
| 236 | 734 } matchinf_T; |
| 735 | |
| 307 | 736 /* |
| 737 * The tables used for recognizing word characters according to spelling. | |
| 738 * These are only used for the first 256 characters of 'encoding'. | |
| 739 */ | |
| 740 typedef struct spelltab_S | |
| 741 { | |
| 742 char_u st_isw[256]; /* flags: is word char */ | |
| 743 char_u st_isu[256]; /* flags: is uppercase char */ | |
| 744 char_u st_fold[256]; /* chars: folded case */ | |
| 324 | 745 char_u st_upper[256]; /* chars: upper case */ |
| 307 | 746 } spelltab_T; |
| 747 | |
| 748 static spelltab_T spelltab; | |
| 749 static int did_set_spelltab; | |
| 750 | |
| 324 | 751 #define CF_WORD 0x01 |
| 752 #define CF_UPPER 0x02 | |
| 307 | 753 |
| 754 static void clear_spell_chartab __ARGS((spelltab_T *sp)); | |
| 755 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
|
756 static int spell_iswordp __ARGS((char_u *p, win_T *wp)); |
| 5477 | 757 static int spell_iswordp_nmw __ARGS((char_u *p, win_T *wp)); |
| 758 #ifdef FEAT_MBYTE | |
| 759 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
|
760 static int spell_iswordp_w __ARGS((int *p, win_T *wp)); |
| 376 | 761 #endif |
| 483 | 762 static int write_spell_prefcond __ARGS((FILE *fd, garray_T *gap)); |
| 307 | 763 |
| 764 /* | |
| 344 | 765 * For finding suggestions: At each node in the tree these states are tried: |
| 330 | 766 */ |
| 767 typedef enum | |
| 768 { | |
| 344 | 769 STATE_START = 0, /* At start of node check for NUL bytes (goodword |
| 770 * ends); if badword ends there is a match, otherwise | |
| 771 * try splitting word. */ | |
| 375 | 772 STATE_NOPREFIX, /* try without prefix */ |
| 344 | 773 STATE_SPLITUNDO, /* Undo splitting. */ |
| 330 | 774 STATE_ENDNUL, /* Past NUL bytes at start of the node. */ |
| 775 STATE_PLAIN, /* Use each byte of the node. */ | |
| 776 STATE_DEL, /* Delete a byte from the bad word. */ | |
| 625 | 777 STATE_INS_PREP, /* Prepare for inserting bytes. */ |
| 330 | 778 STATE_INS, /* Insert a byte in the bad word. */ |
| 779 STATE_SWAP, /* Swap two bytes. */ | |
| 344 | 780 STATE_UNSWAP, /* Undo swap two characters. */ |
| 781 STATE_SWAP3, /* Swap two characters over three. */ | |
| 782 STATE_UNSWAP3, /* Undo Swap two characters over three. */ | |
| 783 STATE_UNROT3L, /* Undo rotate three characters left */ | |
| 784 STATE_UNROT3R, /* Undo rotate three characters right */ | |
| 330 | 785 STATE_REP_INI, /* Prepare for using REP items. */ |
| 786 STATE_REP, /* Use matching REP items from the .aff file. */ | |
| 787 STATE_REP_UNDO, /* Undo a REP item replacement. */ | |
| 788 STATE_FINAL /* End of this node. */ | |
| 789 } state_T; | |
| 790 | |
| 791 /* | |
| 346 | 792 * Struct to keep the state at each level in suggest_try_change(). |
| 323 | 793 */ |
| 794 typedef struct trystate_S | |
| 795 { | |
| 330 | 796 state_T ts_state; /* state at this level, STATE_ */ |
| 323 | 797 int ts_score; /* score */ |
| 344 | 798 idx_T ts_arridx; /* index in tree array, start of node */ |
| 330 | 799 short ts_curi; /* index in list of child nodes */ |
| 800 char_u ts_fidx; /* index in fword[], case-folded bad word */ | |
| 801 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */ | |
| 802 char_u ts_twordlen; /* valid length of tword[] */ | |
| 482 | 803 char_u ts_prefixdepth; /* stack depth for end of prefix or |
| 485 | 804 * PFD_PREFIXTREE or PFD_NOPREFIX */ |
| 805 char_u ts_flags; /* TSF_ flags */ | |
| 330 | 806 #ifdef FEAT_MBYTE |
| 807 char_u ts_tcharlen; /* number of bytes in tword character */ | |
| 808 char_u ts_tcharidx; /* current byte index in tword character */ | |
| 809 char_u ts_isdiff; /* DIFF_ values */ | |
| 810 char_u ts_fcharstart; /* index in fword where badword char started */ | |
| 811 #endif | |
| 483 | 812 char_u ts_prewordlen; /* length of word in "preword[]" */ |
| 813 char_u ts_splitoff; /* index in "tword" after last split */ | |
| 492 | 814 char_u ts_splitfidx; /* "ts_fidx" at word split */ |
| 483 | 815 char_u ts_complen; /* nr of compound words used */ |
| 485 | 816 char_u ts_compsplit; /* index for "compflags" where word was spit */ |
| 346 | 817 char_u ts_save_badflags; /* su_badflags saved here */ |
| 625 | 818 char_u ts_delidx; /* index in fword for char that was deleted, |
| 819 valid when "ts_flags" has TSF_DIDDEL */ | |
| 323 | 820 } trystate_T; |
| 821 | |
| 330 | 822 /* values for ts_isdiff */ |
| 823 #define DIFF_NONE 0 /* no different byte (yet) */ | |
| 824 #define DIFF_YES 1 /* different byte found */ | |
| 825 #define DIFF_INSERT 2 /* inserting character */ | |
| 826 | |
| 485 | 827 /* values for ts_flags */ |
| 828 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */ | |
| 829 #define TSF_DIDSPLIT 2 /* tried split at this point */ | |
| 625 | 830 #define TSF_DIDDEL 4 /* did a delete, "ts_delidx" has index */ |
| 485 | 831 |
| 375 | 832 /* special values ts_prefixdepth */ |
| 485 | 833 #define PFD_NOPREFIX 0xff /* not using prefixes */ |
| 482 | 834 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */ |
| 625 | 835 #define PFD_NOTSPECIAL 0xfd /* highest value that's not special */ |
| 375 | 836 |
| 339 | 837 /* mode values for find_word */ |
| 481 | 838 #define FIND_FOLDWORD 0 /* find word case-folded */ |
| 839 #define FIND_KEEPWORD 1 /* find keep-case word */ | |
| 840 #define FIND_PREFIX 2 /* find word after prefix */ | |
| 841 #define FIND_COMPOUND 3 /* find case-folded compound word */ | |
| 842 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */ | |
| 339 | 843 |
| 236 | 844 static slang_T *slang_alloc __ARGS((char_u *lang)); |
| 845 static void slang_free __ARGS((slang_T *lp)); | |
| 310 | 846 static void slang_clear __ARGS((slang_T *lp)); |
| 625 | 847 static void slang_clear_sug __ARGS((slang_T *lp)); |
| 339 | 848 static void find_word __ARGS((matchinf_T *mip, int mode)); |
| 1762 | 849 static int match_checkcompoundpattern __ARGS((char_u *ptr, int wlen, garray_T *gap)); |
| 483 | 850 static int can_compound __ARGS((slang_T *slang, char_u *word, char_u *flags)); |
| 1762 | 851 static int can_be_compound __ARGS((trystate_T *sp, slang_T *slang, char_u *compflags, int flag)); |
| 852 static int match_compoundrule __ARGS((slang_T *slang, char_u *compflags)); | |
| 455 | 853 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req)); |
| 485 | 854 static void find_prefix __ARGS((matchinf_T *mip, int mode)); |
| 339 | 855 static int fold_more __ARGS((matchinf_T *mip)); |
| 389 | 856 static int spell_valid_case __ARGS((int wordflags, int treeflags)); |
| 498 | 857 static int no_spell_checking __ARGS((win_T *wp)); |
| 307 | 858 static void spell_load_lang __ARGS((char_u *lang)); |
| 310 | 859 static char_u *spell_enc __ARGS((void)); |
| 385 | 860 static void int_wordlist_spl __ARGS((char_u *fname)); |
| 310 | 861 static void spell_load_cb __ARGS((char_u *fname, void *cookie)); |
| 323 | 862 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent)); |
| 389 | 863 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp)); |
| 483 | 864 static int read_region_section __ARGS((FILE *fd, slang_T *slang, int len)); |
| 865 static int read_charflags_section __ARGS((FILE *fd)); | |
| 866 static int read_prefcond_section __ARGS((FILE *fd, slang_T *lp)); | |
| 625 | 867 static int read_rep_section __ARGS((FILE *fd, garray_T *gap, short *first)); |
| 483 | 868 static int read_sal_section __ARGS((FILE *fd, slang_T *slang)); |
| 625 | 869 static int read_words_section __ARGS((FILE *fd, slang_T *lp, int len)); |
| 870 static void count_common_word __ARGS((slang_T *lp, char_u *word, int len, int count)); | |
| 871 static int score_wordcount_adj __ARGS((slang_T *slang, int score, char_u *word, int split)); | |
| 483 | 872 static int read_sofo_section __ARGS((FILE *fd, slang_T *slang)); |
| 873 static int read_compound __ARGS((FILE *fd, slang_T *slang, int len)); | |
| 495 | 874 static int byte_in_str __ARGS((char_u *str, int byte)); |
| 483 | 875 static int init_syl_tab __ARGS((slang_T *slang)); |
| 876 static int count_syllables __ARGS((slang_T *slang, char_u *word)); | |
| 381 | 877 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to)); |
| 878 static void set_sal_first __ARGS((slang_T *lp)); | |
| 372 | 879 #ifdef FEAT_MBYTE |
| 880 static int *mb_str2wide __ARGS((char_u *s)); | |
| 881 #endif | |
| 625 | 882 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
|
883 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
|
884 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
|
885 static void use_midword __ARGS((slang_T *lp, win_T *buf)); |
| 236 | 886 static int find_region __ARGS((char_u *rp, char_u *region)); |
| 887 static int captype __ARGS((char_u *word, char_u *end)); | |
| 474 | 888 static int badword_captype __ARGS((char_u *word, char_u *end)); |
| 323 | 889 static void spell_reload_one __ARGS((char_u *fname, int added_word)); |
| 483 | 890 static void set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp)); |
| 307 | 891 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp)); |
| 892 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen)); | |
| 475 | 893 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col)); |
| 638 | 894 static void spell_find_suggest __ARGS((char_u *badptr, int badlen, suginfo_T *su, int maxcount, int banbadword, int need_cap, int interactive)); |
| 372 | 895 #ifdef FEAT_EVAL |
| 896 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr)); | |
| 897 #endif | |
| 898 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname)); | |
| 625 | 899 static void spell_suggest_intern __ARGS((suginfo_T *su, int interactive)); |
| 900 static void suggest_load_files __ARGS((void)); | |
| 901 static void tree_count_words __ARGS((char_u *byts, idx_T *idxs)); | |
| 344 | 902 static void spell_find_cleanup __ARGS((suginfo_T *su)); |
| 324 | 903 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper)); |
| 344 | 904 static void allcap_copy __ARGS((char_u *word, char_u *wcopy)); |
| 346 | 905 static void suggest_try_special __ARGS((suginfo_T *su)); |
| 906 static void suggest_try_change __ARGS((suginfo_T *su)); | |
| 625 | 907 static void suggest_trie_walk __ARGS((suginfo_T *su, langp_T *lp, char_u *fword, int soundfold)); |
| 908 static void go_deeper __ARGS((trystate_T *stack, int depth, int score_add)); | |
| 455 | 909 #ifdef FEAT_MBYTE |
| 910 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word)); | |
| 911 #endif | |
| 323 | 912 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword)); |
| 344 | 913 static void score_comp_sal __ARGS((suginfo_T *su)); |
| 914 static void score_combine __ARGS((suginfo_T *su)); | |
| 351 | 915 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound)); |
| 625 | 916 static void suggest_try_soundalike_prep __ARGS((void)); |
| 346 | 917 static void suggest_try_soundalike __ARGS((suginfo_T *su)); |
| 625 | 918 static void suggest_try_soundalike_finish __ARGS((void)); |
| 919 static void add_sound_suggest __ARGS((suginfo_T *su, char_u *goodword, int score, langp_T *lp)); | |
| 920 static int soundfold_find __ARGS((slang_T *slang, char_u *word)); | |
| 323 | 921 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags)); |
| 330 | 922 static void set_map_str __ARGS((slang_T *lp, char_u *map)); |
| 323 | 923 static int similar_chars __ARGS((slang_T *slang, int c1, int c2)); |
| 625 | 924 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)); |
| 925 static void check_suggestions __ARGS((suginfo_T *su, garray_T *gap)); | |
| 323 | 926 static void add_banned __ARGS((suginfo_T *su, char_u *word)); |
| 324 | 927 static void rescore_suggestions __ARGS((suginfo_T *su)); |
| 534 | 928 static void rescore_one __ARGS((suginfo_T *su, suggest_T *stp)); |
| 344 | 929 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep)); |
| 375 | 930 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res)); |
| 931 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res)); | |
| 932 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res)); | |
| 372 | 933 #ifdef FEAT_MBYTE |
| 375 | 934 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res)); |
| 372 | 935 #endif |
| 344 | 936 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound)); |
| 625 | 937 static int spell_edit_score __ARGS((slang_T *slang, char_u *badword, char_u *goodword)); |
| 938 static int spell_edit_score_limit __ARGS((slang_T *slang, char_u *badword, char_u *goodword, int limit)); | |
| 939 #ifdef FEAT_MBYTE | |
| 940 static int spell_edit_score_limit_w __ARGS((slang_T *slang, char_u *badword, char_u *goodword, int limit)); | |
| 941 #endif | |
| 701 | 942 static void dump_word __ARGS((slang_T *slang, char_u *word, char_u *pat, int *dir, int round, int flags, linenr_T lnum)); |
| 943 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 | 944 static buf_T *open_spellbuf __ARGS((void)); |
| 945 static void close_spellbuf __ARGS((buf_T *buf)); | |
| 323 | 946 |
| 324 | 947 /* |
| 948 * Use our own character-case definitions, because the current locale may | |
| 949 * differ from what the .spl file uses. | |
| 950 * These must not be called with negative number! | |
| 951 */ | |
| 952 #ifndef FEAT_MBYTE | |
| 953 /* Non-multi-byte implementation. */ | |
| 1883 | 954 # define SPELL_TOFOLD(c) ((c) < 256 ? (int)spelltab.st_fold[c] : (c)) |
| 955 # define SPELL_TOUPPER(c) ((c) < 256 ? (int)spelltab.st_upper[c] : (c)) | |
| 324 | 956 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE) |
| 957 #else | |
| 447 | 958 # if defined(HAVE_WCHAR_H) |
| 959 # include <wchar.h> /* for towupper() and towlower() */ | |
| 960 # endif | |
| 324 | 961 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do |
| 962 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use | |
| 963 * the "w" library function for characters above 255 if available. */ | |
| 964 # ifdef HAVE_TOWLOWER | |
| 965 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \ | |
| 1883 | 966 : (c) < 256 ? (int)spelltab.st_fold[c] : (int)towlower(c)) |
| 324 | 967 # else |
| 968 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \ | |
| 1883 | 969 : (c) < 256 ? (int)spelltab.st_fold[c] : (c)) |
| 324 | 970 # endif |
| 971 | |
| 972 # ifdef HAVE_TOWUPPER | |
| 973 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \ | |
| 1883 | 974 : (c) < 256 ? (int)spelltab.st_upper[c] : (int)towupper(c)) |
| 324 | 975 # else |
| 976 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \ | |
| 1883 | 977 : (c) < 256 ? (int)spelltab.st_upper[c] : (c)) |
| 324 | 978 # endif |
| 979 | |
| 980 # ifdef HAVE_ISWUPPER | |
| 981 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \ | |
| 982 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c)) | |
| 983 # else | |
| 984 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \ | |
| 474 | 985 : (c) < 256 ? spelltab.st_isu[c] : (FALSE)) |
| 324 | 986 # endif |
| 987 #endif | |
| 988 | |
| 307 | 989 |
| 990 static char *e_format = N_("E759: Format error in spell file"); | |
| 381 | 991 static char *e_spell_trunc = N_("E758: Truncated spell file"); |
| 482 | 992 static char *e_afftrailing = N_("Trailing text in %s line %d: %s"); |
| 495 | 993 static char *e_affname = N_("Affix name too long in %s line %d: %s"); |
| 994 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP"); | |
| 995 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range"); | |
| 469 | 996 static char *msg_compressing = N_("Compressing word tree..."); |
| 236 | 997 |
| 626 | 998 /* Remember what "z?" replaced. */ |
| 999 static char_u *repl_from = NULL; | |
| 1000 static char_u *repl_to = NULL; | |
| 1001 | |
| 236 | 1002 /* |
| 1003 * Main spell-checking function. | |
| 300 | 1004 * "ptr" points to a character that could be the start of a word. |
| 534 | 1005 * "*attrp" is set to the highlight index for a badly spelled word. For a |
| 1006 * non-word or when it's OK it remains unchanged. | |
| 236 | 1007 * This must only be called when 'spelllang' is not empty. |
| 323 | 1008 * |
| 385 | 1009 * "capcol" is used to check for a Capitalised word after the end of a |
| 1010 * sentence. If it's zero then perform the check. Return the column where to | |
| 1011 * check next, or -1 when no sentence end was found. If it's NULL then don't | |
| 1012 * worry. | |
| 323 | 1013 * |
| 236 | 1014 * Returns the length of the word in bytes, also when it's OK, so that the |
| 1015 * caller can skip over the word. | |
| 1016 */ | |
| 1017 int | |
| 625 | 1018 spell_check(wp, ptr, attrp, capcol, docount) |
| 236 | 1019 win_T *wp; /* current window */ |
| 1020 char_u *ptr; | |
| 534 | 1021 hlf_T *attrp; |
| 385 | 1022 int *capcol; /* column to check for Capital */ |
| 625 | 1023 int docount; /* count good words */ |
| 236 | 1024 { |
| 1025 matchinf_T mi; /* Most things are put in "mi" so that it can | |
| 1026 be passed to functions quickly. */ | |
| 344 | 1027 int nrlen = 0; /* found a number first */ |
| 385 | 1028 int c; |
| 483 | 1029 int wrongcaplen = 0; |
| 500 | 1030 int lpi; |
| 625 | 1031 int count_word = docount; |
| 236 | 1032 |
| 307 | 1033 /* A word never starts at a space or a control character. Return quickly |
| 1034 * then, skipping over the character. */ | |
| 1035 if (*ptr <= ' ') | |
| 1036 return 1; | |
| 690 | 1037 |
| 1038 /* 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
|
1039 if (wp->w_s->b_langp.ga_len == 0) |
| 690 | 1040 return 1; |
| 1041 | |
| 483 | 1042 vim_memset(&mi, 0, sizeof(matchinf_T)); |
| 236 | 1043 |
| 344 | 1044 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and |
| 584 | 1045 * 0X99FF. But always do check spelling to find "3GPP" and "11 |
| 1046 * julifeest". */ | |
| 300 | 1047 if (*ptr >= '0' && *ptr <= '9') |
| 1048 { | |
| 316 | 1049 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X')) |
| 1050 mi.mi_end = skiphex(ptr + 2); | |
| 300 | 1051 else |
| 344 | 1052 mi.mi_end = skipdigits(ptr); |
| 835 | 1053 nrlen = (int)(mi.mi_end - ptr); |
| 584 | 1054 } |
| 346 | 1055 |
| 1056 /* Find the normal end of the word (until the next non-word character). */ | |
| 344 | 1057 mi.mi_word = ptr; |
| 584 | 1058 mi.mi_fend = ptr; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1059 if (spell_iswordp(mi.mi_fend, wp)) |
| 344 | 1060 { |
| 1061 do | |
| 1062 { | |
| 307 | 1063 mb_ptr_adv(mi.mi_fend); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1064 } 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
|
1065 |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1066 if (capcol != NULL && *capcol == 0 && wp->w_s->b_cap_prog != NULL) |
| 385 | 1067 { |
| 1068 /* Check word starting with capital letter. */ | |
| 455 | 1069 c = PTR2CHAR(ptr); |
| 385 | 1070 if (!SPELL_ISUPPER(c)) |
| 483 | 1071 wrongcaplen = (int)(mi.mi_fend - ptr); |
| 385 | 1072 } |
| 1073 } | |
| 1074 if (capcol != NULL) | |
| 1075 *capcol = -1; | |
| 344 | 1076 |
| 1077 /* We always use the characters up to the next non-word character, | |
| 1078 * also for bad words. */ | |
| 1079 mi.mi_end = mi.mi_fend; | |
| 1080 | |
| 1081 /* Check caps type later. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1082 mi.mi_capflags = 0; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1083 mi.mi_cend = NULL; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1084 mi.mi_win = wp; |
| 344 | 1085 |
| 483 | 1086 /* case-fold the word with one non-word character, so that we can check |
| 1087 * for the word end. */ | |
| 344 | 1088 if (*mi.mi_fend != NUL) |
| 1089 mb_ptr_adv(mi.mi_fend); | |
| 1090 | |
| 1091 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword, | |
| 1092 MAXWLEN + 1); | |
| 835 | 1093 mi.mi_fwordlen = (int)STRLEN(mi.mi_fword); |
| 344 | 1094 |
| 1095 /* The word is bad unless we recognize it. */ | |
| 1096 mi.mi_result = SP_BAD; | |
| 492 | 1097 mi.mi_result2 = SP_BAD; |
| 344 | 1098 |
| 1099 /* | |
| 1100 * Loop over the languages specified in 'spelllang'. | |
| 625 | 1101 * We check them all, because a word may be matched longer in another |
| 1102 * language. | |
| 344 | 1103 */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1104 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
|
1105 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1106 mi.mi_lp = LANGP_ENTRY(wp->w_s->b_langp, lpi); |
| 500 | 1107 |
| 1108 /* If reloading fails the language is still in the list but everything | |
| 1109 * has been cleared. */ | |
| 1110 if (mi.mi_lp->lp_slang->sl_fidxs == NULL) | |
| 1111 continue; | |
| 1112 | |
| 344 | 1113 /* Check for a matching word in case-folded words. */ |
| 1114 find_word(&mi, FIND_FOLDWORD); | |
| 1115 | |
| 1116 /* Check for a matching word in keep-case words. */ | |
| 1117 find_word(&mi, FIND_KEEPWORD); | |
| 1118 | |
| 1119 /* Check for matching prefixes. */ | |
| 485 | 1120 find_prefix(&mi, FIND_FOLDWORD); |
| 492 | 1121 |
| 1122 /* For a NOBREAK language, may want to use a word without a following | |
| 1123 * word as a backup. */ | |
| 1124 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD | |
| 1125 && mi.mi_result2 != SP_BAD) | |
| 1126 { | |
| 1127 mi.mi_result = mi.mi_result2; | |
| 1128 mi.mi_end = mi.mi_end2; | |
| 1129 } | |
| 625 | 1130 |
| 1131 /* Count the word in the first language where it's found to be OK. */ | |
| 1132 if (count_word && mi.mi_result == SP_OK) | |
| 1133 { | |
| 1134 count_common_word(mi.mi_lp->lp_slang, ptr, | |
| 1135 (int)(mi.mi_end - ptr), 1); | |
| 1136 count_word = FALSE; | |
| 1137 } | |
| 344 | 1138 } |
| 1139 | |
| 1140 if (mi.mi_result != SP_OK) | |
| 1141 { | |
| 346 | 1142 /* If we found a number skip over it. Allows for "42nd". Do flag |
| 1143 * rare and local words, e.g., "3GPP". */ | |
| 344 | 1144 if (nrlen > 0) |
| 346 | 1145 { |
| 1146 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED) | |
| 1147 return nrlen; | |
| 1148 } | |
| 344 | 1149 |
| 1150 /* When we are at a non-word character there is no error, just | |
| 1151 * skip over the character (try looking for a word after it). */ | |
| 5477 | 1152 else if (!spell_iswordp_nmw(ptr, wp)) |
| 243 | 1153 { |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1154 if (capcol != NULL && wp->w_s->b_cap_prog != NULL) |
| 385 | 1155 { |
| 1156 regmatch_T regmatch; | |
| 6375 | 1157 int r; |
| 385 | 1158 |
| 1159 /* Check for end of sentence. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1160 regmatch.regprog = wp->w_s->b_cap_prog; |
| 385 | 1161 regmatch.rm_ic = FALSE; |
| 6375 | 1162 r = vim_regexec(®match, ptr, 0); |
| 1163 wp->w_s->b_cap_prog = regmatch.regprog; | |
| 1164 if (r) | |
| 385 | 1165 *capcol = (int)(regmatch.endp[0] - ptr); |
| 1166 } | |
| 1167 | |
| 344 | 1168 #ifdef FEAT_MBYTE |
| 1169 if (has_mbyte) | |
| 474 | 1170 return (*mb_ptr2len)(ptr); |
| 344 | 1171 #endif |
| 1172 return 1; | |
| 300 | 1173 } |
| 483 | 1174 else if (mi.mi_end == ptr) |
| 1175 /* Always include at least one character. Required for when there | |
| 1176 * is a mixup in "midword". */ | |
| 1177 mb_ptr_adv(mi.mi_end); | |
| 492 | 1178 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
|
1179 && LANGP_ENTRY(wp->w_s->b_langp, 0)->lp_slang->sl_nobreak) |
| 492 | 1180 { |
| 1181 char_u *p, *fp; | |
| 1182 int save_result = mi.mi_result; | |
| 1183 | |
| 1184 /* First language in 'spelllang' is NOBREAK. Find first position | |
| 1185 * 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
|
1186 mi.mi_lp = LANGP_ENTRY(wp->w_s->b_langp, 0); |
| 500 | 1187 if (mi.mi_lp->lp_slang->sl_fidxs != NULL) |
| 1188 { | |
| 1189 p = mi.mi_word; | |
| 1190 fp = mi.mi_fword; | |
| 1191 for (;;) | |
| 1192 { | |
| 1193 mb_ptr_adv(p); | |
| 1194 mb_ptr_adv(fp); | |
| 1195 if (p >= mi.mi_end) | |
| 1196 break; | |
| 835 | 1197 mi.mi_compoff = (int)(fp - mi.mi_fword); |
| 500 | 1198 find_word(&mi, FIND_COMPOUND); |
| 1199 if (mi.mi_result != SP_BAD) | |
| 1200 { | |
| 1201 mi.mi_end = p; | |
| 1202 break; | |
| 1203 } | |
| 1204 } | |
| 1205 mi.mi_result = save_result; | |
| 1206 } | |
| 492 | 1207 } |
| 243 | 1208 |
| 344 | 1209 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED) |
| 534 | 1210 *attrp = HLF_SPB; |
| 344 | 1211 else if (mi.mi_result == SP_RARE) |
| 534 | 1212 *attrp = HLF_SPR; |
| 344 | 1213 else |
| 534 | 1214 *attrp = HLF_SPL; |
| 243 | 1215 } |
| 1216 | |
| 483 | 1217 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE)) |
| 1218 { | |
| 1219 /* Report SpellCap only when the word isn't badly spelled. */ | |
| 534 | 1220 *attrp = HLF_SPC; |
| 483 | 1221 return wrongcaplen; |
| 1222 } | |
| 1223 | |
| 300 | 1224 return (int)(mi.mi_end - ptr); |
| 236 | 1225 } |
| 1226 | |
| 1227 /* | |
| 300 | 1228 * Check if the word at "mip->mi_word" is in the tree. |
| 339 | 1229 * When "mode" is FIND_FOLDWORD check in fold-case word tree. |
| 1230 * When "mode" is FIND_KEEPWORD check in keep-case word tree. | |
| 1231 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word | |
| 1232 * tree. | |
| 300 | 1233 * |
| 1234 * For a match mip->mi_result is updated. | |
| 243 | 1235 */ |
| 1236 static void | |
| 339 | 1237 find_word(mip, mode) |
| 243 | 1238 matchinf_T *mip; |
| 339 | 1239 int mode; |
| 243 | 1240 { |
| 324 | 1241 idx_T arridx = 0; |
| 300 | 1242 int endlen[MAXWLEN]; /* length at possible word endings */ |
| 324 | 1243 idx_T endidx[MAXWLEN]; /* possible word endings */ |
| 300 | 1244 int endidxcnt = 0; |
| 1245 int len; | |
| 1246 int wlen = 0; | |
| 1247 int flen; | |
| 1248 int c; | |
| 1249 char_u *ptr; | |
| 324 | 1250 idx_T lo, hi, m; |
| 243 | 1251 #ifdef FEAT_MBYTE |
| 300 | 1252 char_u *s; |
| 490 | 1253 #endif |
| 339 | 1254 char_u *p; |
| 307 | 1255 int res = SP_BAD; |
| 300 | 1256 slang_T *slang = mip->mi_lp->lp_slang; |
| 1257 unsigned flags; | |
| 1258 char_u *byts; | |
| 324 | 1259 idx_T *idxs; |
| 481 | 1260 int word_ends; |
| 485 | 1261 int prefix_found; |
| 492 | 1262 int nobreak_result; |
| 481 | 1263 |
| 1264 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND) | |
| 236 | 1265 { |
| 300 | 1266 /* Check for word with matching case in keep-case tree. */ |
| 1267 ptr = mip->mi_word; | |
| 1268 flen = 9999; /* no case folding, always enough bytes */ | |
| 1269 byts = slang->sl_kbyts; | |
| 1270 idxs = slang->sl_kidxs; | |
| 481 | 1271 |
| 1272 if (mode == FIND_KEEPCOMPOUND) | |
| 1273 /* Skip over the previously found word(s). */ | |
| 1274 wlen += mip->mi_compoff; | |
| 236 | 1275 } |
| 1276 else | |
| 1277 { | |
| 300 | 1278 /* Check for case-folded in case-folded tree. */ |
| 1279 ptr = mip->mi_fword; | |
| 1280 flen = mip->mi_fwordlen; /* available case-folded bytes */ | |
| 1281 byts = slang->sl_fbyts; | |
| 1282 idxs = slang->sl_fidxs; | |
| 339 | 1283 |
| 1284 if (mode == FIND_PREFIX) | |
| 1285 { | |
| 1286 /* Skip over the prefix. */ | |
| 1287 wlen = mip->mi_prefixlen; | |
| 1288 flen -= mip->mi_prefixlen; | |
| 1289 } | |
| 481 | 1290 else if (mode == FIND_COMPOUND) |
| 1291 { | |
| 1292 /* Skip over the previously found word(s). */ | |
| 1293 wlen = mip->mi_compoff; | |
| 1294 flen -= mip->mi_compoff; | |
| 1295 } | |
| 1296 | |
| 243 | 1297 } |
| 1298 | |
| 300 | 1299 if (byts == NULL) |
| 1300 return; /* array is empty */ | |
| 236 | 1301 |
| 1302 /* | |
| 307 | 1303 * Repeat advancing in the tree until: |
| 1304 * - there is a byte that doesn't match, | |
| 1305 * - we reach the end of the tree, | |
| 1306 * - or we reach the end of the line. | |
| 236 | 1307 */ |
| 300 | 1308 for (;;) |
| 236 | 1309 { |
| 346 | 1310 if (flen <= 0 && *mip->mi_fend != NUL) |
| 339 | 1311 flen = fold_more(mip); |
| 300 | 1312 |
| 1313 len = byts[arridx++]; | |
| 1314 | |
| 1315 /* If the first possible byte is a zero the word could end here. | |
| 1316 * Remember this index, we first check for the longest word. */ | |
| 1317 if (byts[arridx] == 0) | |
| 1318 { | |
| 307 | 1319 if (endidxcnt == MAXWLEN) |
| 1320 { | |
| 1321 /* Must be a corrupted spell file. */ | |
| 1322 EMSG(_(e_format)); | |
| 1323 return; | |
| 1324 } | |
| 300 | 1325 endlen[endidxcnt] = wlen; |
| 1326 endidx[endidxcnt++] = arridx++; | |
| 1327 --len; | |
| 1328 | |
| 1329 /* Skip over the zeros, there can be several flag/region | |
| 1330 * combinations. */ | |
| 1331 while (len > 0 && byts[arridx] == 0) | |
| 1332 { | |
| 1333 ++arridx; | |
| 1334 --len; | |
| 1335 } | |
| 1336 if (len == 0) | |
| 1337 break; /* no children, word must end here */ | |
| 1338 } | |
| 1339 | |
| 1340 /* Stop looking at end of the line. */ | |
| 1341 if (ptr[wlen] == NUL) | |
| 1342 break; | |
| 1343 | |
| 1344 /* Perform a binary search in the list of accepted bytes. */ | |
| 1345 c = ptr[wlen]; | |
| 346 | 1346 if (c == TAB) /* <Tab> is handled like <Space> */ |
| 1347 c = ' '; | |
| 300 | 1348 lo = arridx; |
| 1349 hi = arridx + len - 1; | |
| 1350 while (lo < hi) | |
| 1351 { | |
| 1352 m = (lo + hi) / 2; | |
| 1353 if (byts[m] > c) | |
| 1354 hi = m - 1; | |
| 1355 else if (byts[m] < c) | |
| 1356 lo = m + 1; | |
| 1357 else | |
| 1358 { | |
| 1359 lo = hi = m; | |
| 1360 break; | |
| 236 | 1361 } |
| 1362 } | |
| 300 | 1363 |
| 1364 /* Stop if there is no matching byte. */ | |
| 1365 if (hi < lo || byts[lo] != c) | |
| 1366 break; | |
| 1367 | |
| 1368 /* Continue at the child (if there is one). */ | |
| 1369 arridx = idxs[lo]; | |
| 1370 ++wlen; | |
| 1371 --flen; | |
| 346 | 1372 |
| 1373 /* One space in the good word may stand for several spaces in the | |
| 1374 * checked word. */ | |
| 1375 if (c == ' ') | |
| 1376 { | |
| 1377 for (;;) | |
| 1378 { | |
| 1379 if (flen <= 0 && *mip->mi_fend != NUL) | |
| 1380 flen = fold_more(mip); | |
| 1381 if (ptr[wlen] != ' ' && ptr[wlen] != TAB) | |
| 1382 break; | |
| 1383 ++wlen; | |
| 1384 --flen; | |
| 1385 } | |
| 1386 } | |
| 236 | 1387 } |
| 1388 | |
| 300 | 1389 /* |
| 1390 * Verify that one of the possible endings is valid. Try the longest | |
| 1391 * first. | |
| 1392 */ | |
| 1393 while (endidxcnt > 0) | |
| 1394 { | |
| 1395 --endidxcnt; | |
| 1396 arridx = endidx[endidxcnt]; | |
| 1397 wlen = endlen[endidxcnt]; | |
| 236 | 1398 |
| 300 | 1399 #ifdef FEAT_MBYTE |
| 1400 if ((*mb_head_off)(ptr, ptr + wlen) > 0) | |
| 1401 continue; /* not at first byte of character */ | |
| 1402 #endif | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1403 if (spell_iswordp(ptr + wlen, mip->mi_win)) |
| 481 | 1404 { |
| 492 | 1405 if (slang->sl_compprog == NULL && !slang->sl_nobreak) |
| 481 | 1406 continue; /* next char is a word character */ |
| 1407 word_ends = FALSE; | |
| 1408 } | |
| 1409 else | |
| 1410 word_ends = TRUE; | |
| 485 | 1411 /* The prefix flag is before compound flags. Once a valid prefix flag |
| 1412 * has been found we try compound flags. */ | |
| 1413 prefix_found = FALSE; | |
| 300 | 1414 |
| 1415 #ifdef FEAT_MBYTE | |
| 339 | 1416 if (mode != FIND_KEEPWORD && has_mbyte) |
| 300 | 1417 { |
| 1418 /* Compute byte length in original word, length may change | |
| 339 | 1419 * when folding case. This can be slow, take a shortcut when the |
| 1420 * case-folded word is equal to the keep-case word. */ | |
| 300 | 1421 p = mip->mi_word; |
| 339 | 1422 if (STRNCMP(ptr, p, wlen) != 0) |
| 1423 { | |
| 1424 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s)) | |
| 1425 mb_ptr_adv(p); | |
| 835 | 1426 wlen = (int)(p - mip->mi_word); |
| 339 | 1427 } |
| 300 | 1428 } |
| 1429 #endif | |
| 236 | 1430 |
| 339 | 1431 /* Check flags and region. For FIND_PREFIX check the condition and |
| 1432 * prefix ID. | |
| 1433 * Repeat this if there are more flags/region alternatives until there | |
| 1434 * is a match. */ | |
| 1435 res = SP_BAD; | |
| 1436 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0; | |
| 1437 --len, ++arridx) | |
| 300 | 1438 { |
| 1439 flags = idxs[arridx]; | |
| 324 | 1440 |
| 339 | 1441 /* For the fold-case tree check that the case of the checked word |
| 1442 * matches with what the word in the tree requires. | |
| 1443 * For keep-case tree the case is always right. For prefixes we | |
| 1444 * don't bother to check. */ | |
| 1445 if (mode == FIND_FOLDWORD) | |
| 300 | 1446 { |
| 1447 if (mip->mi_cend != mip->mi_word + wlen) | |
| 1448 { | |
| 323 | 1449 /* mi_capflags was set for a different word length, need |
| 1450 * to do it again. */ | |
| 300 | 1451 mip->mi_cend = mip->mi_word + wlen; |
| 323 | 1452 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend); |
| 300 | 1453 } |
| 1454 | |
| 346 | 1455 if (mip->mi_capflags == WF_KEEPCAP |
| 1456 || !spell_valid_case(mip->mi_capflags, flags)) | |
| 339 | 1457 continue; |
| 300 | 1458 } |
| 236 | 1459 |
| 339 | 1460 /* When mode is FIND_PREFIX the word must support the prefix: |
| 1461 * check the prefix ID and the condition. Do that for the list at | |
| 366 | 1462 * mip->mi_prefarridx that find_prefix() filled. */ |
| 485 | 1463 else if (mode == FIND_PREFIX && !prefix_found) |
| 481 | 1464 { |
| 366 | 1465 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx, |
| 390 | 1466 flags, |
| 455 | 1467 mip->mi_word + mip->mi_cprefixlen, slang, |
| 1468 FALSE); | |
| 366 | 1469 if (c == 0) |
| 339 | 1470 continue; |
| 366 | 1471 |
| 1472 /* Use the WF_RARE flag for a rare prefix. */ | |
| 1473 if (c & WF_RAREPFX) | |
| 1474 flags |= WF_RARE; | |
| 485 | 1475 prefix_found = TRUE; |
| 339 | 1476 } |
| 1477 | |
| 492 | 1478 if (slang->sl_nobreak) |
| 1479 { | |
| 1480 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND) | |
| 1481 && (flags & WF_BANNED) == 0) | |
| 1482 { | |
| 1483 /* NOBREAK: found a valid following word. That's all we | |
| 1484 * need to know, so return. */ | |
| 1485 mip->mi_result = SP_OK; | |
| 1486 break; | |
| 1487 } | |
| 1488 } | |
| 1489 | |
| 1490 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND | |
| 1491 || !word_ends)) | |
| 481 | 1492 { |
| 974 | 1493 /* If there is no compound flag or the word is shorter than |
| 483 | 1494 * COMPOUNDMIN reject it quickly. |
| 1495 * Makes you wonder why someone puts a compound flag on a word | |
| 481 | 1496 * that's too short... Myspell compatibility requires this |
| 1497 * anyway. */ | |
| 490 | 1498 if (((unsigned)flags >> 24) == 0 |
| 1499 || wlen - mip->mi_compoff < slang->sl_compminlen) | |
| 483 | 1500 continue; |
| 500 | 1501 #ifdef FEAT_MBYTE |
| 1502 /* For multi-byte chars check character length against | |
| 1503 * COMPOUNDMIN. */ | |
| 1504 if (has_mbyte | |
| 501 | 1505 && slang->sl_compminlen > 0 |
| 500 | 1506 && mb_charlen_len(mip->mi_word + mip->mi_compoff, |
| 1507 wlen - mip->mi_compoff) < slang->sl_compminlen) | |
| 1508 continue; | |
| 1509 #endif | |
| 483 | 1510 |
| 809 | 1511 /* Limit the number of compound words to COMPOUNDWORDMAX if no |
| 490 | 1512 * maximum for syllables is specified. */ |
| 809 | 1513 if (!word_ends && mip->mi_complen + mip->mi_compextra + 2 |
| 1514 > slang->sl_compmax | |
| 490 | 1515 && slang->sl_compsylmax == MAXWLEN) |
| 483 | 1516 continue; |
| 1517 | |
| 819 | 1518 /* Don't allow compounding on a side where an affix was added, |
| 1519 * unless COMPOUNDPERMITFLAG was used. */ | |
| 1520 if (mip->mi_complen > 0 && (flags & WF_NOCOMPBEF)) | |
| 1521 continue; | |
| 1522 if (!word_ends && (flags & WF_NOCOMPAFT)) | |
| 1523 continue; | |
| 1524 | |
| 485 | 1525 /* Quickly check if compounding is possible with this flag. */ |
| 495 | 1526 if (!byte_in_str(mip->mi_complen == 0 |
| 485 | 1527 ? slang->sl_compstartflags |
| 1528 : slang->sl_compallflags, | |
| 495 | 1529 ((unsigned)flags >> 24))) |
| 481 | 1530 continue; |
| 1531 | |
| 1762 | 1532 /* If there is a match with a CHECKCOMPOUNDPATTERN rule |
| 1533 * discard the compound word. */ | |
| 1534 if (match_checkcompoundpattern(ptr, wlen, &slang->sl_comppat)) | |
| 1535 continue; | |
| 1536 | |
| 490 | 1537 if (mode == FIND_COMPOUND) |
| 1538 { | |
| 1539 int capflags; | |
| 1540 | |
| 1541 /* Need to check the caps type of the appended compound | |
| 1542 * word. */ | |
| 1543 #ifdef FEAT_MBYTE | |
| 1544 if (has_mbyte && STRNCMP(ptr, mip->mi_word, | |
| 1545 mip->mi_compoff) != 0) | |
| 1546 { | |
| 1547 /* case folding may have changed the length */ | |
| 1548 p = mip->mi_word; | |
| 1549 for (s = ptr; s < ptr + mip->mi_compoff; mb_ptr_adv(s)) | |
| 1550 mb_ptr_adv(p); | |
| 1551 } | |
| 1552 else | |
| 1553 #endif | |
| 1554 p = mip->mi_word + mip->mi_compoff; | |
| 1555 capflags = captype(p, mip->mi_word + wlen); | |
| 1556 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP | |
| 1557 && (flags & WF_FIXCAP) != 0)) | |
| 1558 continue; | |
| 1559 | |
| 1560 if (capflags != WF_ALLCAP) | |
| 1561 { | |
| 1562 /* When the character before the word is a word | |
| 1563 * character we do not accept a Onecap word. We do | |
| 1564 * accept a no-caps word, even when the dictionary | |
| 1565 * word specifies ONECAP. */ | |
| 1566 mb_ptr_back(mip->mi_word, p); | |
| 5477 | 1567 if (spell_iswordp_nmw(p, mip->mi_win) |
| 490 | 1568 ? capflags == WF_ONECAP |
| 1569 : (flags & WF_ONECAP) != 0 | |
| 1570 && capflags != WF_ONECAP) | |
| 1571 continue; | |
| 1572 } | |
| 1573 } | |
| 1574 | |
| 483 | 1575 /* If the word ends the sequence of compound flags of the |
| 714 | 1576 * words must match with one of the COMPOUNDRULE items and |
| 483 | 1577 * the number of syllables must not be too large. */ |
| 1578 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24); | |
| 1579 mip->mi_compflags[mip->mi_complen + 1] = NUL; | |
| 1580 if (word_ends) | |
| 1581 { | |
| 1582 char_u fword[MAXWLEN]; | |
| 1583 | |
| 1584 if (slang->sl_compsylmax < MAXWLEN) | |
| 1585 { | |
| 1586 /* "fword" is only needed for checking syllables. */ | |
| 1587 if (ptr == mip->mi_word) | |
| 1588 (void)spell_casefold(ptr, wlen, fword, MAXWLEN); | |
| 1589 else | |
| 1590 vim_strncpy(fword, ptr, endlen[endidxcnt]); | |
| 1591 } | |
| 1592 if (!can_compound(slang, fword, mip->mi_compflags)) | |
| 1593 continue; | |
| 1594 } | |
| 1762 | 1595 else if (slang->sl_comprules != NULL |
| 1596 && !match_compoundrule(slang, mip->mi_compflags)) | |
| 1597 /* The compound flags collected so far do not match any | |
| 1598 * COMPOUNDRULE, discard the compounded word. */ | |
| 1599 continue; | |
| 481 | 1600 } |
| 1601 | |
| 500 | 1602 /* Check NEEDCOMPOUND: can't use word without compounding. */ |
| 1603 else if (flags & WF_NEEDCOMP) | |
| 1604 continue; | |
| 1605 | |
| 492 | 1606 nobreak_result = SP_OK; |
| 1607 | |
| 481 | 1608 if (!word_ends) |
| 1609 { | |
| 492 | 1610 int save_result = mip->mi_result; |
| 1611 char_u *save_end = mip->mi_end; | |
| 501 | 1612 langp_T *save_lp = mip->mi_lp; |
| 1613 int lpi; | |
| 492 | 1614 |
| 1615 /* Check that a valid word follows. If there is one and we | |
| 1616 * are compounding, it will set "mi_result", thus we are | |
| 1617 * always finished here. For NOBREAK we only check that a | |
| 1618 * valid word follows. | |
| 481 | 1619 * Recursive! */ |
| 492 | 1620 if (slang->sl_nobreak) |
| 1621 mip->mi_result = SP_BAD; | |
| 481 | 1622 |
| 1623 /* Find following word in case-folded tree. */ | |
| 1624 mip->mi_compoff = endlen[endidxcnt]; | |
| 1625 #ifdef FEAT_MBYTE | |
| 1626 if (has_mbyte && mode == FIND_KEEPWORD) | |
| 1627 { | |
| 1628 /* Compute byte length in case-folded word from "wlen": | |
| 1629 * byte length in keep-case word. Length may change when | |
| 1630 * folding case. This can be slow, take a shortcut when | |
| 1631 * the case-folded word is equal to the keep-case word. */ | |
| 1632 p = mip->mi_fword; | |
| 1633 if (STRNCMP(ptr, p, wlen) != 0) | |
| 1634 { | |
| 1635 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s)) | |
| 1636 mb_ptr_adv(p); | |
| 835 | 1637 mip->mi_compoff = (int)(p - mip->mi_fword); |
| 481 | 1638 } |
| 1639 } | |
| 1640 #endif | |
| 485 | 1641 c = mip->mi_compoff; |
| 483 | 1642 ++mip->mi_complen; |
| 809 | 1643 if (flags & WF_COMPROOT) |
| 1644 ++mip->mi_compextra; | |
| 501 | 1645 |
| 1646 /* For NOBREAK we need to try all NOBREAK languages, at least | |
| 1647 * to find the ".add" file(s). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1648 for (lpi = 0; lpi < mip->mi_win->w_s->b_langp.ga_len; ++lpi) |
| 501 | 1649 { |
| 1650 if (slang->sl_nobreak) | |
| 1651 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1652 mip->mi_lp = LANGP_ENTRY(mip->mi_win->w_s->b_langp, lpi); |
| 501 | 1653 if (mip->mi_lp->lp_slang->sl_fidxs == NULL |
| 1654 || !mip->mi_lp->lp_slang->sl_nobreak) | |
| 1655 continue; | |
| 1656 } | |
| 1657 | |
| 1658 find_word(mip, FIND_COMPOUND); | |
| 1659 | |
| 1660 /* When NOBREAK any word that matches is OK. Otherwise we | |
| 1661 * need to find the longest match, thus try with keep-case | |
| 1662 * and prefix too. */ | |
| 492 | 1663 if (!slang->sl_nobreak || mip->mi_result == SP_BAD) |
| 1664 { | |
| 501 | 1665 /* Find following word in keep-case tree. */ |
| 1666 mip->mi_compoff = wlen; | |
| 1667 find_word(mip, FIND_KEEPCOMPOUND); | |
| 1668 | |
| 819 | 1669 #if 0 /* Disabled, a prefix must not appear halfway a compound word, |
| 1670 unless the COMPOUNDPERMITFLAG is used and then it can't be a | |
| 1671 postponed prefix. */ | |
| 501 | 1672 if (!slang->sl_nobreak || mip->mi_result == SP_BAD) |
| 1673 { | |
| 1674 /* Check for following word with prefix. */ | |
| 1675 mip->mi_compoff = c; | |
| 1676 find_prefix(mip, FIND_COMPOUND); | |
| 1677 } | |
| 819 | 1678 #endif |
| 492 | 1679 } |
| 501 | 1680 |
| 1681 if (!slang->sl_nobreak) | |
| 1682 break; | |
| 492 | 1683 } |
| 483 | 1684 --mip->mi_complen; |
| 809 | 1685 if (flags & WF_COMPROOT) |
| 1686 --mip->mi_compextra; | |
| 501 | 1687 mip->mi_lp = save_lp; |
| 485 | 1688 |
| 492 | 1689 if (slang->sl_nobreak) |
| 1690 { | |
| 1691 nobreak_result = mip->mi_result; | |
| 1692 mip->mi_result = save_result; | |
| 1693 mip->mi_end = save_end; | |
| 1694 } | |
| 1695 else | |
| 1696 { | |
| 1697 if (mip->mi_result == SP_OK) | |
| 1698 break; | |
| 1699 continue; | |
| 1700 } | |
| 481 | 1701 } |
| 1702 | |
| 339 | 1703 if (flags & WF_BANNED) |
| 1704 res = SP_BANNED; | |
| 1705 else if (flags & WF_REGION) | |
| 1706 { | |
| 1707 /* Check region. */ | |
| 390 | 1708 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0) |
| 300 | 1709 res = SP_OK; |
| 339 | 1710 else |
| 1711 res = SP_LOCAL; | |
| 300 | 1712 } |
| 339 | 1713 else if (flags & WF_RARE) |
| 1714 res = SP_RARE; | |
| 307 | 1715 else |
| 339 | 1716 res = SP_OK; |
| 1717 | |
| 492 | 1718 /* Always use the longest match and the best result. For NOBREAK |
| 1719 * we separately keep the longest match without a following good | |
| 1720 * word as a fall-back. */ | |
| 1721 if (nobreak_result == SP_BAD) | |
| 1722 { | |
| 1723 if (mip->mi_result2 > res) | |
| 1724 { | |
| 1725 mip->mi_result2 = res; | |
| 1726 mip->mi_end2 = mip->mi_word + wlen; | |
| 1727 } | |
| 1728 else if (mip->mi_result2 == res | |
| 1729 && mip->mi_end2 < mip->mi_word + wlen) | |
| 1730 mip->mi_end2 = mip->mi_word + wlen; | |
| 1731 } | |
| 1732 else if (mip->mi_result > res) | |
| 339 | 1733 { |
| 1734 mip->mi_result = res; | |
| 1735 mip->mi_end = mip->mi_word + wlen; | |
| 1736 } | |
| 351 | 1737 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen) |
| 339 | 1738 mip->mi_end = mip->mi_word + wlen; |
| 1739 | |
| 492 | 1740 if (mip->mi_result == SP_OK) |
| 339 | 1741 break; |
| 300 | 1742 } |
| 1743 | |
| 492 | 1744 if (mip->mi_result == SP_OK) |
| 300 | 1745 break; |
| 1746 } | |
| 236 | 1747 } |
| 1748 | |
| 323 | 1749 /* |
| 1762 | 1750 * Return TRUE if there is a match between the word ptr[wlen] and |
| 1751 * CHECKCOMPOUNDPATTERN rules, assuming that we will concatenate with another | |
| 1752 * word. | |
| 1753 * A match means that the first part of CHECKCOMPOUNDPATTERN matches at the | |
| 1754 * end of ptr[wlen] and the second part matches after it. | |
| 1755 */ | |
| 1756 static int | |
| 1757 match_checkcompoundpattern(ptr, wlen, gap) | |
| 1758 char_u *ptr; | |
| 1759 int wlen; | |
| 1760 garray_T *gap; /* &sl_comppat */ | |
| 1761 { | |
| 1762 int i; | |
| 1763 char_u *p; | |
| 1764 int len; | |
| 1765 | |
| 1766 for (i = 0; i + 1 < gap->ga_len; i += 2) | |
| 1767 { | |
| 1768 p = ((char_u **)gap->ga_data)[i + 1]; | |
| 1769 if (STRNCMP(ptr + wlen, p, STRLEN(p)) == 0) | |
| 1770 { | |
| 1771 /* Second part matches at start of following compound word, now | |
| 1772 * check if first part matches at end of previous word. */ | |
| 1773 p = ((char_u **)gap->ga_data)[i]; | |
| 1771 | 1774 len = (int)STRLEN(p); |
| 1762 | 1775 if (len <= wlen && STRNCMP(ptr + wlen - len, p, len) == 0) |
| 1776 return TRUE; | |
| 1777 } | |
| 1778 } | |
| 1779 return FALSE; | |
| 1780 } | |
| 1781 | |
| 1782 /* | |
| 626 | 1783 * Return TRUE if "flags" is a valid sequence of compound flags and "word" |
| 1784 * does not have too many syllables. | |
| 482 | 1785 */ |
| 1786 static int | |
| 483 | 1787 can_compound(slang, word, flags) |
| 482 | 1788 slang_T *slang; |
| 483 | 1789 char_u *word; |
| 1790 char_u *flags; | |
| 1791 { | |
| 495 | 1792 #ifdef FEAT_MBYTE |
| 1793 char_u uflags[MAXWLEN * 2]; | |
| 1794 int i; | |
| 1795 #endif | |
| 1796 char_u *p; | |
| 483 | 1797 |
| 1798 if (slang->sl_compprog == NULL) | |
| 1799 return FALSE; | |
| 495 | 1800 #ifdef FEAT_MBYTE |
| 1801 if (enc_utf8) | |
| 1802 { | |
| 1803 /* Need to convert the single byte flags to utf8 characters. */ | |
| 1804 p = uflags; | |
| 1805 for (i = 0; flags[i] != NUL; ++i) | |
| 1806 p += mb_char2bytes(flags[i], p); | |
| 1807 *p = NUL; | |
| 1808 p = uflags; | |
| 1809 } | |
| 1810 else | |
| 1811 #endif | |
| 1812 p = flags; | |
| 6375 | 1813 if (!vim_regexec_prog(&slang->sl_compprog, FALSE, p, 0)) |
| 483 | 1814 return FALSE; |
| 1815 | |
| 490 | 1816 /* Count the number of syllables. This may be slow, do it last. If there |
| 1817 * are too many syllables AND the number of compound words is above | |
| 809 | 1818 * COMPOUNDWORDMAX then compounding is not allowed. */ |
| 483 | 1819 if (slang->sl_compsylmax < MAXWLEN |
| 1820 && count_syllables(slang, word) > slang->sl_compsylmax) | |
| 495 | 1821 return (int)STRLEN(flags) < slang->sl_compmax; |
| 483 | 1822 return TRUE; |
| 482 | 1823 } |
| 1824 | |
| 1825 /* | |
| 1762 | 1826 * Return TRUE when the sequence of flags in "compflags" plus "flag" can |
| 1827 * possibly form a valid compounded word. This also checks the COMPOUNDRULE | |
| 1828 * lines if they don't contain wildcards. | |
| 1829 */ | |
| 1830 static int | |
| 1831 can_be_compound(sp, slang, compflags, flag) | |
| 1832 trystate_T *sp; | |
| 1833 slang_T *slang; | |
| 1834 char_u *compflags; | |
| 1835 int flag; | |
| 1836 { | |
| 1837 /* If the flag doesn't appear in sl_compstartflags or sl_compallflags | |
| 1838 * then it can't possibly compound. */ | |
| 1839 if (!byte_in_str(sp->ts_complen == sp->ts_compsplit | |
| 1840 ? slang->sl_compstartflags : slang->sl_compallflags, flag)) | |
| 1841 return FALSE; | |
| 1842 | |
| 1843 /* If there are no wildcards, we can check if the flags collected so far | |
| 1844 * possibly can form a match with COMPOUNDRULE patterns. This only | |
| 1845 * makes sense when we have two or more words. */ | |
| 1846 if (slang->sl_comprules != NULL && sp->ts_complen > sp->ts_compsplit) | |
| 1847 { | |
| 1848 int v; | |
| 1849 | |
| 1850 compflags[sp->ts_complen] = flag; | |
| 1851 compflags[sp->ts_complen + 1] = NUL; | |
| 1852 v = match_compoundrule(slang, compflags + sp->ts_compsplit); | |
| 1853 compflags[sp->ts_complen] = NUL; | |
| 1854 return v; | |
| 1855 } | |
| 1856 | |
| 1857 return TRUE; | |
| 1858 } | |
| 1859 | |
| 1860 | |
| 1861 /* | |
| 1862 * Return TRUE if the compound flags in compflags[] match the start of any | |
| 1863 * compound rule. This is used to stop trying a compound if the flags | |
| 1864 * collected so far can't possibly match any compound rule. | |
| 1865 * Caller must check that slang->sl_comprules is not NULL. | |
| 1866 */ | |
| 1867 static int | |
| 1868 match_compoundrule(slang, compflags) | |
| 1869 slang_T *slang; | |
| 1870 char_u *compflags; | |
| 1871 { | |
| 1872 char_u *p; | |
| 1873 int i; | |
| 1874 int c; | |
| 1875 | |
| 1876 /* loop over all the COMPOUNDRULE entries */ | |
| 1877 for (p = slang->sl_comprules; *p != NUL; ++p) | |
| 1878 { | |
| 1879 /* loop over the flags in the compound word we have made, match | |
| 1880 * them against the current rule entry */ | |
| 1881 for (i = 0; ; ++i) | |
| 1882 { | |
| 1883 c = compflags[i]; | |
| 1884 if (c == NUL) | |
| 1885 /* found a rule that matches for the flags we have so far */ | |
| 1886 return TRUE; | |
| 1887 if (*p == '/' || *p == NUL) | |
| 1888 break; /* end of rule, it's too short */ | |
| 1889 if (*p == '[') | |
| 1890 { | |
| 1891 int match = FALSE; | |
| 1892 | |
| 1893 /* compare against all the flags in [] */ | |
| 1894 ++p; | |
| 1895 while (*p != ']' && *p != NUL) | |
| 1896 if (*p++ == c) | |
| 1897 match = TRUE; | |
| 1898 if (!match) | |
| 1899 break; /* none matches */ | |
| 1900 } | |
| 1901 else if (*p != c) | |
| 1902 break; /* flag of word doesn't match flag in pattern */ | |
| 1903 ++p; | |
| 1904 } | |
| 1905 | |
| 1906 /* Skip to the next "/", where the next pattern starts. */ | |
| 1907 p = vim_strchr(p, '/'); | |
| 1908 if (p == NULL) | |
| 1909 break; | |
| 1910 } | |
| 1911 | |
| 1912 /* Checked all the rules and none of them match the flags, so there | |
| 1913 * can't possibly be a compound starting with these flags. */ | |
| 1914 return FALSE; | |
| 1915 } | |
| 1916 | |
| 1917 /* | |
| 390 | 1918 * Return non-zero if the prefix indicated by "arridx" matches with the prefix |
| 1919 * ID in "flags" for the word "word". | |
| 366 | 1920 * The WF_RAREPFX flag is included in the return value for a rare prefix. |
| 351 | 1921 */ |
| 1922 static int | |
| 455 | 1923 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req) |
| 351 | 1924 int totprefcnt; /* nr of prefix IDs */ |
| 1925 int arridx; /* idx in sl_pidxs[] */ | |
| 390 | 1926 int flags; |
| 351 | 1927 char_u *word; |
| 1928 slang_T *slang; | |
| 455 | 1929 int cond_req; /* only use prefixes with a condition */ |
| 351 | 1930 { |
| 1931 int prefcnt; | |
| 1932 int pidx; | |
| 6375 | 1933 regprog_T **rp; |
| 390 | 1934 int prefid; |
| 1935 | |
| 1936 prefid = (unsigned)flags >> 24; | |
| 351 | 1937 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt) |
| 1938 { | |
| 1939 pidx = slang->sl_pidxs[arridx + prefcnt]; | |
| 1940 | |
| 1941 /* Check the prefix ID. */ | |
| 1942 if (prefid != (pidx & 0xff)) | |
| 1943 continue; | |
| 1944 | |
| 390 | 1945 /* Check if the prefix doesn't combine and the word already has a |
| 1946 * suffix. */ | |
| 1947 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC)) | |
| 1948 continue; | |
| 1949 | |
| 351 | 1950 /* Check the condition, if there is one. The condition index is |
| 366 | 1951 * stored in the two bytes above the prefix ID byte. */ |
| 6375 | 1952 rp = &slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff]; |
| 1953 if (*rp != NULL) | |
| 1954 { | |
| 1955 if (!vim_regexec_prog(rp, FALSE, word, 0)) | |
| 351 | 1956 continue; |
| 1957 } | |
| 455 | 1958 else if (cond_req) |
| 1959 continue; | |
| 1960 | |
| 1961 /* It's a match! Return the WF_ flags. */ | |
| 366 | 1962 return pidx; |
| 1963 } | |
| 1964 return 0; | |
| 351 | 1965 } |
| 1966 | |
| 1967 /* | |
| 339 | 1968 * Check if the word at "mip->mi_word" has a matching prefix. |
| 1969 * If it does, then check the following word. | |
| 1970 * | |
| 485 | 1971 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a |
| 1972 * prefix in a compound word. | |
| 1973 * | |
| 339 | 1974 * For a match mip->mi_result is updated. |
| 1975 */ | |
| 1976 static void | |
| 485 | 1977 find_prefix(mip, mode) |
| 339 | 1978 matchinf_T *mip; |
| 485 | 1979 int mode; |
| 339 | 1980 { |
| 1981 idx_T arridx = 0; | |
| 1982 int len; | |
| 1983 int wlen = 0; | |
| 1984 int flen; | |
| 1985 int c; | |
| 1986 char_u *ptr; | |
| 1987 idx_T lo, hi, m; | |
| 1988 slang_T *slang = mip->mi_lp->lp_slang; | |
| 1989 char_u *byts; | |
| 1990 idx_T *idxs; | |
| 1991 | |
| 375 | 1992 byts = slang->sl_pbyts; |
| 1993 if (byts == NULL) | |
| 1994 return; /* array is empty */ | |
| 1995 | |
| 339 | 1996 /* We use the case-folded word here, since prefixes are always |
| 1997 * case-folded. */ | |
| 1998 ptr = mip->mi_fword; | |
| 1999 flen = mip->mi_fwordlen; /* available case-folded bytes */ | |
| 485 | 2000 if (mode == FIND_COMPOUND) |
| 2001 { | |
| 2002 /* Skip over the previously found word(s). */ | |
| 2003 ptr += mip->mi_compoff; | |
| 2004 flen -= mip->mi_compoff; | |
| 2005 } | |
| 339 | 2006 idxs = slang->sl_pidxs; |
| 2007 | |
| 2008 /* | |
| 2009 * Repeat advancing in the tree until: | |
| 2010 * - there is a byte that doesn't match, | |
| 2011 * - we reach the end of the tree, | |
| 2012 * - or we reach the end of the line. | |
| 2013 */ | |
| 2014 for (;;) | |
| 2015 { | |
| 2016 if (flen == 0 && *mip->mi_fend != NUL) | |
| 2017 flen = fold_more(mip); | |
| 2018 | |
| 2019 len = byts[arridx++]; | |
| 2020 | |
| 2021 /* If the first possible byte is a zero the prefix could end here. | |
| 2022 * Check if the following word matches and supports the prefix. */ | |
| 2023 if (byts[arridx] == 0) | |
| 2024 { | |
| 2025 /* There can be several prefixes with different conditions. We | |
| 2026 * try them all, since we don't know which one will give the | |
| 2027 * longest match. The word is the same each time, pass the list | |
| 2028 * of possible prefixes to find_word(). */ | |
| 2029 mip->mi_prefarridx = arridx; | |
| 2030 mip->mi_prefcnt = len; | |
| 2031 while (len > 0 && byts[arridx] == 0) | |
| 2032 { | |
| 2033 ++arridx; | |
| 2034 --len; | |
| 2035 } | |
| 2036 mip->mi_prefcnt -= len; | |
| 2037 | |
| 2038 /* Find the word that comes after the prefix. */ | |
| 2039 mip->mi_prefixlen = wlen; | |
| 485 | 2040 if (mode == FIND_COMPOUND) |
| 2041 /* Skip over the previously found word(s). */ | |
| 2042 mip->mi_prefixlen += mip->mi_compoff; | |
| 2043 | |
| 455 | 2044 #ifdef FEAT_MBYTE |
| 2045 if (has_mbyte) | |
| 2046 { | |
| 2047 /* Case-folded length may differ from original length. */ | |
| 485 | 2048 mip->mi_cprefixlen = nofold_len(mip->mi_fword, |
| 2049 mip->mi_prefixlen, mip->mi_word); | |
| 455 | 2050 } |
| 2051 else | |
| 485 | 2052 mip->mi_cprefixlen = mip->mi_prefixlen; |
| 455 | 2053 #endif |
| 339 | 2054 find_word(mip, FIND_PREFIX); |
| 2055 | |
| 2056 | |
| 2057 if (len == 0) | |
| 2058 break; /* no children, word must end here */ | |
| 2059 } | |
| 2060 | |
| 2061 /* Stop looking at end of the line. */ | |
| 2062 if (ptr[wlen] == NUL) | |
| 2063 break; | |
| 2064 | |
| 2065 /* Perform a binary search in the list of accepted bytes. */ | |
| 2066 c = ptr[wlen]; | |
| 2067 lo = arridx; | |
| 2068 hi = arridx + len - 1; | |
| 2069 while (lo < hi) | |
| 2070 { | |
| 2071 m = (lo + hi) / 2; | |
| 2072 if (byts[m] > c) | |
| 2073 hi = m - 1; | |
| 2074 else if (byts[m] < c) | |
| 2075 lo = m + 1; | |
| 2076 else | |
| 2077 { | |
| 2078 lo = hi = m; | |
| 2079 break; | |
| 2080 } | |
| 2081 } | |
| 2082 | |
| 2083 /* Stop if there is no matching byte. */ | |
| 2084 if (hi < lo || byts[lo] != c) | |
| 2085 break; | |
| 2086 | |
| 2087 /* Continue at the child (if there is one). */ | |
| 2088 arridx = idxs[lo]; | |
| 2089 ++wlen; | |
| 2090 --flen; | |
| 2091 } | |
| 2092 } | |
| 2093 | |
| 2094 /* | |
| 2095 * Need to fold at least one more character. Do until next non-word character | |
| 626 | 2096 * for efficiency. Include the non-word character too. |
| 339 | 2097 * Return the length of the folded chars in bytes. |
| 2098 */ | |
| 2099 static int | |
| 2100 fold_more(mip) | |
| 2101 matchinf_T *mip; | |
| 2102 { | |
| 2103 int flen; | |
| 2104 char_u *p; | |
| 2105 | |
| 2106 p = mip->mi_fend; | |
| 2107 do | |
| 2108 { | |
| 2109 mb_ptr_adv(mip->mi_fend); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2110 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_win)); |
| 339 | 2111 |
| 626 | 2112 /* Include the non-word character so that we can check for the word end. */ |
| 339 | 2113 if (*mip->mi_fend != NUL) |
| 2114 mb_ptr_adv(mip->mi_fend); | |
| 2115 | |
| 2116 (void)spell_casefold(p, (int)(mip->mi_fend - p), | |
| 2117 mip->mi_fword + mip->mi_fwordlen, | |
| 2118 MAXWLEN - mip->mi_fwordlen); | |
| 835 | 2119 flen = (int)STRLEN(mip->mi_fword + mip->mi_fwordlen); |
| 339 | 2120 mip->mi_fwordlen += flen; |
| 2121 return flen; | |
| 2122 } | |
| 2123 | |
| 2124 /* | |
| 323 | 2125 * Check case flags for a word. Return TRUE if the word has the requested |
| 2126 * case. | |
| 2127 */ | |
| 2128 static int | |
| 389 | 2129 spell_valid_case(wordflags, treeflags) |
| 2130 int wordflags; /* flags for the checked word. */ | |
| 323 | 2131 int treeflags; /* flags for the word in the spell tree */ |
| 2132 { | |
| 389 | 2133 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0) |
| 323 | 2134 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0 |
| 474 | 2135 && ((treeflags & WF_ONECAP) == 0 |
| 2136 || (wordflags & WF_ONECAP) != 0))); | |
| 323 | 2137 } |
| 2138 | |
| 351 | 2139 /* |
| 2140 * Return TRUE if spell checking is not enabled. | |
| 2141 */ | |
| 2142 static int | |
| 498 | 2143 no_spell_checking(wp) |
| 2144 win_T *wp; | |
| 2145 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2146 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
|
2147 || wp->w_s->b_langp.ga_len == 0) |
| 351 | 2148 { |
| 2149 EMSG(_("E756: Spell checking is not enabled")); | |
| 2150 return TRUE; | |
| 2151 } | |
| 2152 return FALSE; | |
| 2153 } | |
| 300 | 2154 |
| 236 | 2155 /* |
| 2156 * Move to next spell error. | |
| 500 | 2157 * "curline" is FALSE for "[s", "]s", "[S" and "]S". |
| 2158 * "curline" is TRUE to find word under/after cursor in the same line. | |
| 483 | 2159 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move |
| 2160 * to after badly spelled word before the cursor. | |
| 495 | 2161 * Return 0 if not found, length of the badly spelled word otherwise. |
| 236 | 2162 */ |
| 2163 int | |
| 498 | 2164 spell_move_to(wp, dir, allwords, curline, attrp) |
| 2165 win_T *wp; | |
| 236 | 2166 int dir; /* FORWARD or BACKWARD */ |
| 500 | 2167 int allwords; /* TRUE for "[s"/"]s", FALSE for "[S"/"]S" */ |
| 323 | 2168 int curline; |
| 534 | 2169 hlf_T *attrp; /* return: attributes of bad word or NULL |
| 2170 (only when "dir" is FORWARD) */ | |
| 236 | 2171 { |
| 249 | 2172 linenr_T lnum; |
| 2173 pos_T found_pos; | |
| 495 | 2174 int found_len = 0; |
| 236 | 2175 char_u *line; |
| 2176 char_u *p; | |
| 346 | 2177 char_u *endp; |
| 534 | 2178 hlf_T attr; |
| 236 | 2179 int len; |
| 5519 | 2180 #ifdef FEAT_SYN_HL |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2181 int has_syntax = syntax_present(wp); |
| 5519 | 2182 #endif |
| 249 | 2183 int col; |
| 2184 int can_spell; | |
| 346 | 2185 char_u *buf = NULL; |
| 2186 int buflen = 0; | |
| 2187 int skip = 0; | |
| 385 | 2188 int capcol = -1; |
| 500 | 2189 int found_one = FALSE; |
| 2190 int wrapped = FALSE; | |
| 236 | 2191 |
| 498 | 2192 if (no_spell_checking(wp)) |
| 495 | 2193 return 0; |
| 236 | 2194 |
| 249 | 2195 /* |
| 2196 * Start looking for bad word at the start of the line, because we can't | |
| 817 | 2197 * start halfway a word, we don't know where it starts or ends. |
| 249 | 2198 * |
| 2199 * When searching backwards, we continue in the line to find the last | |
| 2200 * bad word (in the cursor line: before the cursor). | |
| 346 | 2201 * |
| 2202 * We concatenate the start of the next line, so that wrapped words work | |
| 2203 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards | |
| 2204 * though... | |
| 249 | 2205 */ |
| 498 | 2206 lnum = wp->w_cursor.lnum; |
| 698 | 2207 clearpos(&found_pos); |
| 236 | 2208 |
| 2209 while (!got_int) | |
| 2210 { | |
| 498 | 2211 line = ml_get_buf(wp->w_buffer, lnum, FALSE); |
| 346 | 2212 |
| 835 | 2213 len = (int)STRLEN(line); |
| 346 | 2214 if (buflen < len + MAXWLEN + 2) |
| 2215 { | |
| 2216 vim_free(buf); | |
| 2217 buflen = len + MAXWLEN + 2; | |
| 2218 buf = alloc(buflen); | |
| 2219 if (buf == NULL) | |
| 2220 break; | |
| 2221 } | |
| 2222 | |
| 385 | 2223 /* In first line check first word for Capital. */ |
| 2224 if (lnum == 1) | |
| 2225 capcol = 0; | |
| 2226 | |
| 2227 /* For checking first word with a capital skip white space. */ | |
| 2228 if (capcol == 0) | |
| 835 | 2229 capcol = (int)(skipwhite(line) - line); |
| 2230 else if (curline && wp == curwin) | |
| 2231 { | |
| 2232 /* For spellbadword(): check if first word needs a capital. */ | |
| 944 | 2233 col = (int)(skipwhite(line) - line); |
| 835 | 2234 if (check_need_cap(lnum, col)) |
| 2235 capcol = col; | |
| 2236 | |
| 2237 /* Need to get the line again, may have looked at the previous | |
| 2238 * one. */ | |
| 2239 line = ml_get_buf(wp->w_buffer, lnum, FALSE); | |
| 2240 } | |
| 385 | 2241 |
| 346 | 2242 /* Copy the line into "buf" and append the start of the next line if |
| 2243 * possible. */ | |
| 2244 STRCPY(buf, line); | |
| 498 | 2245 if (lnum < wp->w_buffer->b_ml.ml_line_count) |
| 883 | 2246 spell_cat_line(buf + STRLEN(buf), |
| 2247 ml_get_buf(wp->w_buffer, lnum + 1, FALSE), MAXWLEN); | |
| 346 | 2248 |
| 2249 p = buf + skip; | |
| 2250 endp = buf + len; | |
| 2251 while (p < endp) | |
| 236 | 2252 { |
| 500 | 2253 /* When searching backward don't search after the cursor. Unless |
| 2254 * we wrapped around the end of the buffer. */ | |
| 300 | 2255 if (dir == BACKWARD |
| 498 | 2256 && lnum == wp->w_cursor.lnum |
| 500 | 2257 && !wrapped |
| 498 | 2258 && (colnr_T)(p - buf) >= wp->w_cursor.col) |
| 300 | 2259 break; |
| 249 | 2260 |
| 300 | 2261 /* start of word */ |
| 534 | 2262 attr = HLF_COUNT; |
| 625 | 2263 len = spell_check(wp, p, &attr, &capcol, FALSE); |
| 249 | 2264 |
| 534 | 2265 if (attr != HLF_COUNT) |
| 300 | 2266 { |
| 2267 /* We found a bad word. Check the attribute. */ | |
| 534 | 2268 if (allwords || attr == HLF_SPB) |
| 236 | 2269 { |
| 300 | 2270 /* When searching forward only accept a bad word after |
| 2271 * the cursor. */ | |
| 2272 if (dir == BACKWARD | |
| 500 | 2273 || lnum != wp->w_cursor.lnum |
| 498 | 2274 || (lnum == wp->w_cursor.lnum |
| 500 | 2275 && (wrapped |
| 2276 || (colnr_T)(curline ? p - buf + len | |
| 346 | 2277 : p - buf) |
| 500 | 2278 > wp->w_cursor.col))) |
| 236 | 2279 { |
| 5519 | 2280 #ifdef FEAT_SYN_HL |
| 300 | 2281 if (has_syntax) |
| 249 | 2282 { |
| 835 | 2283 col = (int)(p - buf); |
| 498 | 2284 (void)syn_get_id(wp, lnum, (colnr_T)col, |
| 1504 | 2285 FALSE, &can_spell, FALSE); |
| 857 | 2286 if (!can_spell) |
| 2287 attr = HLF_COUNT; | |
| 300 | 2288 } |
| 2289 else | |
| 737 | 2290 #endif |
| 300 | 2291 can_spell = TRUE; |
| 249 | 2292 |
| 300 | 2293 if (can_spell) |
| 2294 { | |
| 857 | 2295 found_one = TRUE; |
| 300 | 2296 found_pos.lnum = lnum; |
| 835 | 2297 found_pos.col = (int)(p - buf); |
| 249 | 2298 #ifdef FEAT_VIRTUALEDIT |
| 300 | 2299 found_pos.coladd = 0; |
| 249 | 2300 #endif |
| 300 | 2301 if (dir == FORWARD) |
| 2302 { | |
| 2303 /* No need to search further. */ | |
| 498 | 2304 wp->w_cursor = found_pos; |
| 346 | 2305 vim_free(buf); |
| 498 | 2306 if (attrp != NULL) |
| 2307 *attrp = attr; | |
| 495 | 2308 return len; |
| 249 | 2309 } |
| 483 | 2310 else if (curline) |
| 2311 /* Insert mode completion: put cursor after | |
| 2312 * the bad word. */ | |
| 2313 found_pos.col += len; | |
| 495 | 2314 found_len = len; |
| 249 | 2315 } |
| 236 | 2316 } |
| 857 | 2317 else |
| 2318 found_one = TRUE; | |
| 236 | 2319 } |
| 2320 } | |
| 2321 | |
| 300 | 2322 /* advance to character after the word */ |
| 2323 p += len; | |
| 385 | 2324 capcol -= len; |
| 236 | 2325 } |
| 2326 | |
| 483 | 2327 if (dir == BACKWARD && found_pos.lnum != 0) |
| 2328 { | |
| 500 | 2329 /* Use the last match in the line (before the cursor). */ |
| 498 | 2330 wp->w_cursor = found_pos; |
| 483 | 2331 vim_free(buf); |
| 495 | 2332 return found_len; |
| 483 | 2333 } |
| 2334 | |
| 323 | 2335 if (curline) |
| 346 | 2336 break; /* only check cursor line */ |
| 323 | 2337 |
| 236 | 2338 /* Advance to next line. */ |
| 249 | 2339 if (dir == BACKWARD) |
| 2340 { | |
| 500 | 2341 /* If we are back at the starting line and searched it again there |
| 2342 * is no match, give up. */ | |
| 2343 if (lnum == wp->w_cursor.lnum && wrapped) | |
| 346 | 2344 break; |
| 500 | 2345 |
| 2346 if (lnum > 1) | |
| 2347 --lnum; | |
| 2348 else if (!p_ws) | |
| 2349 break; /* at first line and 'nowrapscan' */ | |
| 2350 else | |
| 2351 { | |
| 2352 /* Wrap around to the end of the buffer. May search the | |
| 2353 * starting line again and accept the last match. */ | |
| 2354 lnum = wp->w_buffer->b_ml.ml_line_count; | |
| 2355 wrapped = TRUE; | |
| 503 | 2356 if (!shortmess(SHM_SEARCH)) |
| 2357 give_warning((char_u *)_(top_bot_msg), TRUE); | |
| 500 | 2358 } |
| 385 | 2359 capcol = -1; |
| 249 | 2360 } |
| 2361 else | |
| 2362 { | |
| 500 | 2363 if (lnum < wp->w_buffer->b_ml.ml_line_count) |
| 2364 ++lnum; | |
| 2365 else if (!p_ws) | |
| 2366 break; /* at first line and 'nowrapscan' */ | |
| 2367 else | |
| 2368 { | |
| 2369 /* Wrap around to the start of the buffer. May search the | |
| 2370 * starting line again and accept the first match. */ | |
| 2371 lnum = 1; | |
| 2372 wrapped = TRUE; | |
| 503 | 2373 if (!shortmess(SHM_SEARCH)) |
| 2374 give_warning((char_u *)_(bot_top_msg), TRUE); | |
| 500 | 2375 } |
| 2376 | |
| 2377 /* If we are back at the starting line and there is no match then | |
| 2378 * give up. */ | |
| 1808 | 2379 if (lnum == wp->w_cursor.lnum && (!found_one || wrapped)) |
| 346 | 2380 break; |
| 2381 | |
| 2382 /* Skip the characters at the start of the next line that were | |
| 2383 * included in a match crossing line boundaries. */ | |
| 534 | 2384 if (attr == HLF_COUNT) |
| 835 | 2385 skip = (int)(p - endp); |
| 346 | 2386 else |
| 2387 skip = 0; | |
| 385 | 2388 |
| 835 | 2389 /* Capcol skips over the inserted space. */ |
| 385 | 2390 --capcol; |
| 2391 | |
| 2392 /* But after empty line check first word in next line */ | |
| 2393 if (*skipwhite(line) == NUL) | |
| 2394 capcol = 0; | |
| 249 | 2395 } |
| 236 | 2396 |
| 2397 line_breakcheck(); | |
| 2398 } | |
| 2399 | |
| 346 | 2400 vim_free(buf); |
| 495 | 2401 return 0; |
| 346 | 2402 } |
| 2403 | |
| 2404 /* | |
| 2405 * For spell checking: concatenate the start of the following line "line" into | |
| 2406 * "buf", blanking-out special characters. Copy less then "maxlen" bytes. | |
| 1577 | 2407 * Keep the blanks at the start of the next line, this is used in win_line() |
| 2408 * to skip those bytes if the word was OK. | |
| 346 | 2409 */ |
| 2410 void | |
| 2411 spell_cat_line(buf, line, maxlen) | |
| 2412 char_u *buf; | |
| 2413 char_u *line; | |
| 2414 int maxlen; | |
| 2415 { | |
| 2416 char_u *p; | |
| 2417 int n; | |
| 2418 | |
| 2419 p = skipwhite(line); | |
| 2420 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL) | |
| 2421 p = skipwhite(p + 1); | |
| 2422 | |
| 2423 if (*p != NUL) | |
| 2424 { | |
| 1577 | 2425 /* Only worth concatenating if there is something else than spaces to |
| 2426 * concatenate. */ | |
| 2427 n = (int)(p - line) + 1; | |
| 2428 if (n < maxlen - 1) | |
| 2429 { | |
| 2430 vim_memset(buf, ' ', n); | |
| 2431 vim_strncpy(buf + n, p, maxlen - 1 - n); | |
| 2432 } | |
| 346 | 2433 } |
| 236 | 2434 } |
| 2435 | |
| 626 | 2436 /* |
| 2437 * Structure used for the cookie argument of do_in_runtimepath(). | |
| 2438 */ | |
| 501 | 2439 typedef struct spelload_S |
| 2440 { | |
| 2441 char_u sl_lang[MAXWLEN + 1]; /* language name */ | |
| 2442 slang_T *sl_slang; /* resulting slang_T struct */ | |
| 2443 int sl_nobreak; /* NOBREAK language found */ | |
| 2444 } spelload_T; | |
| 2445 | |
| 236 | 2446 /* |
| 307 | 2447 * Load word list(s) for "lang" from Vim spell file(s). |
| 310 | 2448 * "lang" must be the language without the region: e.g., "en". |
| 236 | 2449 */ |
| 307 | 2450 static void |
| 236 | 2451 spell_load_lang(lang) |
| 2452 char_u *lang; | |
| 2453 { | |
| 310 | 2454 char_u fname_enc[85]; |
| 236 | 2455 int r; |
| 501 | 2456 spelload_T sl; |
| 649 | 2457 #ifdef FEAT_AUTOCMD |
| 2458 int round; | |
| 2459 #endif | |
| 307 | 2460 |
| 310 | 2461 /* Copy the language name to pass it to spell_load_cb() as a cookie. |
| 307 | 2462 * It's truncated when an error is detected. */ |
| 501 | 2463 STRCPY(sl.sl_lang, lang); |
| 2464 sl.sl_slang = NULL; | |
| 2465 sl.sl_nobreak = FALSE; | |
| 307 | 2466 |
| 649 | 2467 #ifdef FEAT_AUTOCMD |
| 2468 /* We may retry when no spell file is found for the language, an | |
| 2469 * autocommand may load it then. */ | |
| 2470 for (round = 1; round <= 2; ++round) | |
| 2471 #endif | |
| 2472 { | |
| 2473 /* | |
| 2474 * Find the first spell file for "lang" in 'runtimepath' and load it. | |
| 2475 */ | |
| 2476 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 2477 #ifdef VMS |
| 2478 "spell/%s_%s.spl", | |
| 2479 #else | |
| 2480 "spell/%s.%s.spl", | |
| 2481 #endif | |
| 2482 lang, spell_enc()); | |
| 649 | 2483 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl); |
| 2484 | |
| 2485 if (r == FAIL && *sl.sl_lang != NUL) | |
| 2486 { | |
| 2487 /* Try loading the ASCII version. */ | |
| 2488 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 2489 #ifdef VMS |
| 2490 "spell/%s_ascii.spl", | |
| 2491 #else | |
| 2492 "spell/%s.ascii.spl", | |
| 2493 #endif | |
| 2494 lang); | |
| 649 | 2495 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl); |
| 2496 | |
| 2497 #ifdef FEAT_AUTOCMD | |
| 2498 if (r == FAIL && *sl.sl_lang != NUL && round == 1 | |
| 2499 && apply_autocmds(EVENT_SPELLFILEMISSING, lang, | |
| 2500 curbuf->b_fname, FALSE, curbuf)) | |
| 2501 continue; | |
| 2502 break; | |
| 2503 #endif | |
| 2504 } | |
| 714 | 2505 #ifdef FEAT_AUTOCMD |
| 2506 break; | |
| 2507 #endif | |
| 307 | 2508 } |
| 2509 | |
| 2510 if (r == FAIL) | |
| 649 | 2511 { |
| 2660 | 2512 smsg((char_u *) |
| 2513 #ifdef VMS | |
| 2514 _("Warning: Cannot find word list \"%s_%s.spl\" or \"%s_ascii.spl\""), | |
| 2515 #else | |
| 2516 _("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""), | |
| 2517 #endif | |
| 483 | 2518 lang, spell_enc(), lang); |
| 649 | 2519 } |
| 501 | 2520 else if (sl.sl_slang != NULL) |
| 2521 { | |
| 626 | 2522 /* At least one file was loaded, now load ALL the additions. */ |
| 310 | 2523 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl"); |
| 501 | 2524 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &sl); |
| 310 | 2525 } |
| 2526 } | |
| 2527 | |
| 2528 /* | |
| 2529 * Return the encoding used for spell checking: Use 'encoding', except that we | |
| 2530 * use "latin1" for "latin9". And limit to 60 characters (just in case). | |
| 2531 */ | |
| 2532 static char_u * | |
| 2533 spell_enc() | |
| 2534 { | |
| 2535 | |
| 2536 #ifdef FEAT_MBYTE | |
| 2537 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0) | |
| 2538 return p_enc; | |
| 2539 #endif | |
| 2540 return (char_u *)"latin1"; | |
| 236 | 2541 } |
| 2542 | |
| 2543 /* | |
| 385 | 2544 * Get the name of the .spl file for the internal wordlist into |
| 2545 * "fname[MAXPATHL]". | |
| 2546 */ | |
| 2547 static void | |
| 2548 int_wordlist_spl(fname) | |
| 2549 char_u *fname; | |
| 2550 { | |
| 2660 | 2551 vim_snprintf((char *)fname, MAXPATHL, SPL_FNAME_TMPL, |
| 385 | 2552 int_wordlist, spell_enc()); |
| 2553 } | |
| 2554 | |
| 2555 /* | |
| 625 | 2556 * Allocate a new slang_T for language "lang". "lang" can be NULL. |
| 236 | 2557 * Caller must fill "sl_next". |
| 2558 */ | |
| 2559 static slang_T * | |
| 2560 slang_alloc(lang) | |
| 2561 char_u *lang; | |
| 2562 { | |
| 2563 slang_T *lp; | |
| 2564 | |
| 300 | 2565 lp = (slang_T *)alloc_clear(sizeof(slang_T)); |
| 236 | 2566 if (lp != NULL) |
| 2567 { | |
| 625 | 2568 if (lang != NULL) |
| 2569 lp->sl_name = vim_strsave(lang); | |
| 323 | 2570 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10); |
| 625 | 2571 ga_init2(&lp->sl_repsal, sizeof(fromto_T), 10); |
| 483 | 2572 lp->sl_compmax = MAXWLEN; |
| 2573 lp->sl_compsylmax = MAXWLEN; | |
| 625 | 2574 hash_init(&lp->sl_wordcount); |
| 2575 } | |
| 2576 | |
| 236 | 2577 return lp; |
| 2578 } | |
| 2579 | |
| 2580 /* | |
| 2581 * Free the contents of an slang_T and the structure itself. | |
| 2582 */ | |
| 2583 static void | |
| 2584 slang_free(lp) | |
| 2585 slang_T *lp; | |
| 2586 { | |
| 2587 vim_free(lp->sl_name); | |
| 310 | 2588 vim_free(lp->sl_fname); |
| 2589 slang_clear(lp); | |
| 2590 vim_free(lp); | |
| 2591 } | |
| 2592 | |
| 2593 /* | |
| 2594 * Clear an slang_T so that the file can be reloaded. | |
| 2595 */ | |
| 2596 static void | |
| 2597 slang_clear(lp) | |
| 2598 slang_T *lp; | |
| 2599 { | |
| 339 | 2600 garray_T *gap; |
| 2601 fromto_T *ftp; | |
| 344 | 2602 salitem_T *smp; |
| 339 | 2603 int i; |
| 625 | 2604 int round; |
| 323 | 2605 |
| 300 | 2606 vim_free(lp->sl_fbyts); |
| 310 | 2607 lp->sl_fbyts = NULL; |
| 300 | 2608 vim_free(lp->sl_kbyts); |
| 310 | 2609 lp->sl_kbyts = NULL; |
| 339 | 2610 vim_free(lp->sl_pbyts); |
| 2611 lp->sl_pbyts = NULL; | |
| 2612 | |
| 300 | 2613 vim_free(lp->sl_fidxs); |
| 310 | 2614 lp->sl_fidxs = NULL; |
| 300 | 2615 vim_free(lp->sl_kidxs); |
| 310 | 2616 lp->sl_kidxs = NULL; |
| 339 | 2617 vim_free(lp->sl_pidxs); |
| 2618 lp->sl_pidxs = NULL; | |
| 323 | 2619 |
| 625 | 2620 for (round = 1; round <= 2; ++round) |
| 2621 { | |
| 2622 gap = round == 1 ? &lp->sl_rep : &lp->sl_repsal; | |
| 2623 while (gap->ga_len > 0) | |
| 2624 { | |
| 2625 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len]; | |
| 2626 vim_free(ftp->ft_from); | |
| 2627 vim_free(ftp->ft_to); | |
| 2628 } | |
| 2629 ga_clear(gap); | |
| 2630 } | |
| 344 | 2631 |
| 2632 gap = &lp->sl_sal; | |
| 375 | 2633 if (lp->sl_sofo) |
| 376 | 2634 { |
| 2635 /* "ga_len" is set to 1 without adding an item for latin1 */ | |
| 2636 if (gap->ga_data != NULL) | |
| 2637 /* SOFOFROM and SOFOTO items: free lists of wide characters. */ | |
| 2638 for (i = 0; i < gap->ga_len; ++i) | |
| 2639 vim_free(((int **)gap->ga_data)[i]); | |
| 2640 } | |
| 375 | 2641 else |
| 2642 /* SAL items: free salitem_T items */ | |
| 2643 while (gap->ga_len > 0) | |
| 2644 { | |
| 2645 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len]; | |
| 2646 vim_free(smp->sm_lead); | |
| 2647 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */ | |
| 2648 vim_free(smp->sm_to); | |
| 2649 #ifdef FEAT_MBYTE | |
| 2650 vim_free(smp->sm_lead_w); | |
| 2651 vim_free(smp->sm_oneof_w); | |
| 2652 vim_free(smp->sm_to_w); | |
| 2653 #endif | |
| 2654 } | |
| 344 | 2655 ga_clear(gap); |
| 323 | 2656 |
| 339 | 2657 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
|
2658 vim_regfree(lp->sl_prefprog[i]); |
| 376 | 2659 lp->sl_prefixcnt = 0; |
| 339 | 2660 vim_free(lp->sl_prefprog); |
| 376 | 2661 lp->sl_prefprog = NULL; |
| 2662 | |
| 714 | 2663 vim_free(lp->sl_info); |
| 2664 lp->sl_info = NULL; | |
| 2665 | |
| 376 | 2666 vim_free(lp->sl_midword); |
| 2667 lp->sl_midword = NULL; | |
| 339 | 2668 |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4352
diff
changeset
|
2669 vim_regfree(lp->sl_compprog); |
| 1762 | 2670 vim_free(lp->sl_comprules); |
| 483 | 2671 vim_free(lp->sl_compstartflags); |
| 485 | 2672 vim_free(lp->sl_compallflags); |
| 483 | 2673 lp->sl_compprog = NULL; |
| 1762 | 2674 lp->sl_comprules = NULL; |
| 483 | 2675 lp->sl_compstartflags = NULL; |
| 485 | 2676 lp->sl_compallflags = NULL; |
| 483 | 2677 |
| 2678 vim_free(lp->sl_syllable); | |
| 2679 lp->sl_syllable = NULL; | |
| 2680 ga_clear(&lp->sl_syl_items); | |
| 481 | 2681 |
| 809 | 2682 ga_clear_strings(&lp->sl_comppat); |
| 2683 | |
| 625 | 2684 hash_clear_all(&lp->sl_wordcount, WC_KEY_OFF); |
| 2685 hash_init(&lp->sl_wordcount); | |
| 2686 | |
| 2687 #ifdef FEAT_MBYTE | |
| 2688 hash_clear_all(&lp->sl_map_hash, 0); | |
| 2689 #endif | |
| 2690 | |
| 2691 /* Clear info from .sug file. */ | |
| 2692 slang_clear_sug(lp); | |
| 483 | 2693 |
| 2694 lp->sl_compmax = MAXWLEN; | |
| 501 | 2695 lp->sl_compminlen = 0; |
| 483 | 2696 lp->sl_compsylmax = MAXWLEN; |
| 2697 lp->sl_regions[0] = NUL; | |
| 236 | 2698 } |
| 2699 | |
| 2700 /* | |
| 625 | 2701 * Clear the info from the .sug file in "lp". |
| 2702 */ | |
| 2703 static void | |
| 2704 slang_clear_sug(lp) | |
| 2705 slang_T *lp; | |
| 2706 { | |
| 2707 vim_free(lp->sl_sbyts); | |
| 2708 lp->sl_sbyts = NULL; | |
| 2709 vim_free(lp->sl_sidxs); | |
| 2710 lp->sl_sidxs = NULL; | |
| 2711 close_spellbuf(lp->sl_sugbuf); | |
| 2712 lp->sl_sugbuf = NULL; | |
| 2713 lp->sl_sugloaded = FALSE; | |
| 2714 lp->sl_sugtime = 0; | |
| 2715 } | |
| 2716 | |
| 2717 /* | |
| 307 | 2718 * Load one spell file and store the info into a slang_T. |
| 236 | 2719 * Invoked through do_in_runtimepath(). |
| 2720 */ | |
| 2721 static void | |
| 310 | 2722 spell_load_cb(fname, cookie) |
| 236 | 2723 char_u *fname; |
| 501 | 2724 void *cookie; |
| 2725 { | |
| 2726 spelload_T *slp = (spelload_T *)cookie; | |
| 2727 slang_T *slang; | |
| 2728 | |
| 2729 slang = spell_load_file(fname, slp->sl_lang, NULL, FALSE); | |
| 2730 if (slang != NULL) | |
| 2731 { | |
| 2732 /* When a previously loaded file has NOBREAK also use it for the | |
| 2733 * ".add" files. */ | |
| 2734 if (slp->sl_nobreak && slang->sl_add) | |
| 2735 slang->sl_nobreak = TRUE; | |
| 2736 else if (slang->sl_nobreak) | |
| 2737 slp->sl_nobreak = TRUE; | |
| 2738 | |
| 2739 slp->sl_slang = slang; | |
| 2740 } | |
| 310 | 2741 } |
| 2742 | |
| 2743 /* | |
| 2744 * Load one spell file and store the info into a slang_T. | |
| 2745 * | |
| 625 | 2746 * This is invoked in three ways: |
| 310 | 2747 * - From spell_load_cb() to load a spell file for the first time. "lang" is |
| 2748 * the language name, "old_lp" is NULL. Will allocate an slang_T. | |
| 2749 * - To reload a spell file that was changed. "lang" is NULL and "old_lp" | |
| 2750 * points to the existing slang_T. | |
| 625 | 2751 * - Just after writing a .spl file; it's read back to produce the .sug file. |
| 626 | 2752 * "old_lp" is NULL and "lang" is NULL. Will allocate an slang_T. |
| 2753 * | |
| 323 | 2754 * Returns the slang_T the spell file was loaded into. NULL for error. |
| 310 | 2755 */ |
| 323 | 2756 static slang_T * |
| 2757 spell_load_file(fname, lang, old_lp, silent) | |
| 310 | 2758 char_u *fname; |
| 2759 char_u *lang; | |
| 2760 slang_T *old_lp; | |
| 323 | 2761 int silent; /* no error if file doesn't exist */ |
| 310 | 2762 { |
| 236 | 2763 FILE *fd; |
| 483 | 2764 char_u buf[VIMSPELLMAGICL]; |
| 236 | 2765 char_u *p; |
| 2766 int i; | |
| 339 | 2767 int n; |
| 300 | 2768 int len; |
| 236 | 2769 char_u *save_sourcing_name = sourcing_name; |
| 2770 linenr_T save_sourcing_lnum = sourcing_lnum; | |
| 307 | 2771 slang_T *lp = NULL; |
| 344 | 2772 int c = 0; |
| 483 | 2773 int res; |
| 236 | 2774 |
| 310 | 2775 fd = mch_fopen((char *)fname, "r"); |
| 236 | 2776 if (fd == NULL) |
| 2777 { | |
| 323 | 2778 if (!silent) |
| 2779 EMSG2(_(e_notopen), fname); | |
| 2780 else if (p_verbose > 2) | |
| 2781 { | |
| 2782 verbose_enter(); | |
| 2783 smsg((char_u *)e_notopen, fname); | |
| 2784 verbose_leave(); | |
| 2785 } | |
| 255 | 2786 goto endFAIL; |
| 236 | 2787 } |
| 310 | 2788 if (p_verbose > 2) |
| 2789 { | |
| 2790 verbose_enter(); | |
| 2791 smsg((char_u *)_("Reading spell file \"%s\""), fname); | |
| 2792 verbose_leave(); | |
| 2793 } | |
| 2794 | |
| 2795 if (old_lp == NULL) | |
| 2796 { | |
| 2797 lp = slang_alloc(lang); | |
| 2798 if (lp == NULL) | |
| 2799 goto endFAIL; | |
| 2800 | |
| 2801 /* Remember the file name, used to reload the file when it's updated. */ | |
| 2802 lp->sl_fname = vim_strsave(fname); | |
| 2803 if (lp->sl_fname == NULL) | |
| 2804 goto endFAIL; | |
| 2805 | |
| 2660 | 2806 /* Check for .add.spl (_add.spl for VMS). */ |
| 2807 lp->sl_add = strstr((char *)gettail(fname), SPL_FNAME_ADD) != NULL; | |
| 310 | 2808 } |
| 2809 else | |
| 2810 lp = old_lp; | |
| 307 | 2811 |
| 236 | 2812 /* Set sourcing_name, so that error messages mention the file name. */ |
| 2813 sourcing_name = fname; | |
| 2814 sourcing_lnum = 0; | |
| 2815 | |
| 625 | 2816 /* |
| 2817 * <HEADER>: <fileID> | |
| 339 | 2818 */ |
| 236 | 2819 for (i = 0; i < VIMSPELLMAGICL; ++i) |
| 2820 buf[i] = getc(fd); /* <fileID> */ | |
| 2821 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0) | |
| 2822 { | |
| 483 | 2823 EMSG(_("E757: This does not look like a spell file")); |
| 381 | 2824 goto endFAIL; |
| 483 | 2825 } |
| 2826 c = getc(fd); /* <versionnr> */ | |
| 2827 if (c < VIMSPELLVERSION) | |
| 2828 { | |
| 2829 EMSG(_("E771: Old spell file, needs to be updated")); | |
| 381 | 2830 goto endFAIL; |
| 2831 } | |
| 483 | 2832 else if (c > VIMSPELLVERSION) |
| 2833 { | |
| 2834 EMSG(_("E772: Spell file is for newer version of Vim")); | |
| 481 | 2835 goto endFAIL; |
| 483 | 2836 } |
| 2837 | |
| 2838 | |
| 2839 /* | |
| 2840 * <SECTIONS>: <section> ... <sectionend> | |
| 2841 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents) | |
| 2842 */ | |
| 2843 for (;;) | |
| 2844 { | |
| 2845 n = getc(fd); /* <sectionID> or <sectionend> */ | |
| 2846 if (n == SN_END) | |
| 2847 break; | |
| 2848 c = getc(fd); /* <sectionflags> */ | |
| 695 | 2849 len = get4c(fd); /* <sectionlen> */ |
| 483 | 2850 if (len < 0) |
| 2851 goto truncerr; | |
| 2852 | |
| 2853 res = 0; | |
| 2854 switch (n) | |
| 2855 { | |
| 714 | 2856 case SN_INFO: |
| 2857 lp->sl_info = read_string(fd, len); /* <infotext> */ | |
| 2858 if (lp->sl_info == NULL) | |
| 2859 goto endFAIL; | |
| 2860 break; | |
| 2861 | |
| 483 | 2862 case SN_REGION: |
| 2863 res = read_region_section(fd, lp, len); | |
| 2864 break; | |
| 2865 | |
| 2866 case SN_CHARFLAGS: | |
| 2867 res = read_charflags_section(fd); | |
| 2868 break; | |
| 2869 | |
| 2870 case SN_MIDWORD: | |
| 2871 lp->sl_midword = read_string(fd, len); /* <midword> */ | |
| 2872 if (lp->sl_midword == NULL) | |
| 2873 goto endFAIL; | |
| 2874 break; | |
| 2875 | |
| 2876 case SN_PREFCOND: | |
| 2877 res = read_prefcond_section(fd, lp); | |
| 2878 break; | |
| 2879 | |
| 2880 case SN_REP: | |
| 625 | 2881 res = read_rep_section(fd, &lp->sl_rep, lp->sl_rep_first); |
| 2882 break; | |
| 2883 | |
| 2884 case SN_REPSAL: | |
| 2885 res = read_rep_section(fd, &lp->sl_repsal, lp->sl_repsal_first); | |
| 483 | 2886 break; |
| 2887 | |
| 2888 case SN_SAL: | |
| 2889 res = read_sal_section(fd, lp); | |
| 2890 break; | |
| 2891 | |
| 2892 case SN_SOFO: | |
| 2893 res = read_sofo_section(fd, lp); | |
| 2894 break; | |
| 2895 | |
| 2896 case SN_MAP: | |
| 2897 p = read_string(fd, len); /* <mapstr> */ | |
| 2898 if (p == NULL) | |
| 2899 goto endFAIL; | |
| 2900 set_map_str(lp, p); | |
| 2901 vim_free(p); | |
| 2902 break; | |
| 2903 | |
| 625 | 2904 case SN_WORDS: |
| 2905 res = read_words_section(fd, lp, len); | |
| 2906 break; | |
| 2907 | |
| 2908 case SN_SUGFILE: | |
|
2229
d45902a5c61c
Fix a few more things for persistent undo.
Bram Moolenaar <bram@vim.org>
parents:
2214
diff
changeset
|
2909 lp->sl_sugtime = get8ctime(fd); /* <timestamp> */ |
| 625 | 2910 break; |
| 2911 | |
| 698 | 2912 case SN_NOSPLITSUGS: |
| 2913 lp->sl_nosplitsugs = TRUE; /* <timestamp> */ | |
| 2914 break; | |
| 2915 | |
| 483 | 2916 case SN_COMPOUND: |
| 2917 res = read_compound(fd, lp, len); | |
| 2918 break; | |
| 2919 | |
| 492 | 2920 case SN_NOBREAK: |
| 2921 lp->sl_nobreak = TRUE; | |
| 2922 break; | |
| 2923 | |
| 483 | 2924 case SN_SYLLABLE: |
| 2925 lp->sl_syllable = read_string(fd, len); /* <syllable> */ | |
| 2926 if (lp->sl_syllable == NULL) | |
| 2927 goto endFAIL; | |
| 2928 if (init_syl_tab(lp) == FAIL) | |
| 2929 goto endFAIL; | |
| 2930 break; | |
| 2931 | |
| 2932 default: | |
| 2933 /* Unsupported section. When it's required give an error | |
| 2934 * message. When it's not required skip the contents. */ | |
| 2935 if (c & SNF_REQUIRED) | |
| 2936 { | |
| 2937 EMSG(_("E770: Unsupported section in spell file")); | |
| 2938 goto endFAIL; | |
| 2939 } | |
| 2940 while (--len >= 0) | |
| 2941 if (getc(fd) < 0) | |
| 2942 goto truncerr; | |
| 2943 break; | |
| 2944 } | |
| 625 | 2945 someerror: |
| 483 | 2946 if (res == SP_FORMERROR) |
| 2947 { | |
| 2948 EMSG(_(e_format)); | |
| 381 | 2949 goto endFAIL; |
| 344 | 2950 } |
| 483 | 2951 if (res == SP_TRUNCERROR) |
| 2952 { | |
| 2953 truncerr: | |
| 2954 EMSG(_(e_spell_trunc)); | |
| 375 | 2955 goto endFAIL; |
| 483 | 2956 } |
| 495 | 2957 if (res == SP_OTHERERROR) |
| 483 | 2958 goto endFAIL; |
| 389 | 2959 } |
| 236 | 2960 |
| 625 | 2961 /* <LWORDTREE> */ |
| 2962 res = spell_read_tree(fd, &lp->sl_fbyts, &lp->sl_fidxs, FALSE, 0); | |
| 2963 if (res != 0) | |
| 2964 goto someerror; | |
| 2965 | |
| 2966 /* <KWORDTREE> */ | |
| 2967 res = spell_read_tree(fd, &lp->sl_kbyts, &lp->sl_kidxs, FALSE, 0); | |
| 2968 if (res != 0) | |
| 2969 goto someerror; | |
| 2970 | |
| 2971 /* <PREFIXTREE> */ | |
| 2972 res = spell_read_tree(fd, &lp->sl_pbyts, &lp->sl_pidxs, TRUE, | |
| 2973 lp->sl_prefixcnt); | |
| 2974 if (res != 0) | |
| 2975 goto someerror; | |
| 243 | 2976 |
| 310 | 2977 /* For a new file link it in the list of spell files. */ |
| 626 | 2978 if (old_lp == NULL && lang != NULL) |
| 310 | 2979 { |
| 2980 lp->sl_next = first_lang; | |
| 2981 first_lang = lp; | |
| 2982 } | |
| 307 | 2983 |
| 255 | 2984 goto endOK; |
| 2985 | |
| 2986 endFAIL: | |
| 310 | 2987 if (lang != NULL) |
| 2988 /* truncating the name signals the error to spell_load_lang() */ | |
| 2989 *lang = NUL; | |
| 2990 if (lp != NULL && old_lp == NULL) | |
| 307 | 2991 slang_free(lp); |
| 500 | 2992 lp = NULL; |
| 255 | 2993 |
| 2994 endOK: | |
| 236 | 2995 if (fd != NULL) |
| 2996 fclose(fd); | |
| 2997 sourcing_name = save_sourcing_name; | |
| 2998 sourcing_lnum = save_sourcing_lnum; | |
| 323 | 2999 |
| 3000 return lp; | |
| 236 | 3001 } |
| 3002 | |
| 376 | 3003 /* |
| 3004 * Read a length field from "fd" in "cnt_bytes" bytes. | |
| 381 | 3005 * Allocate memory, read the string into it and add a NUL at the end. |
| 376 | 3006 * Returns NULL when the count is zero. |
| 483 | 3007 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result |
| 3008 * otherwise. | |
| 376 | 3009 */ |
| 3010 static char_u * | |
| 389 | 3011 read_cnt_string(fd, cnt_bytes, cntp) |
| 376 | 3012 FILE *fd; |
| 3013 int cnt_bytes; | |
| 389 | 3014 int *cntp; |
| 376 | 3015 { |
| 3016 int cnt = 0; | |
| 3017 int i; | |
| 3018 char_u *str; | |
| 3019 | |
| 3020 /* read the length bytes, MSB first */ | |
| 3021 for (i = 0; i < cnt_bytes; ++i) | |
| 3022 cnt = (cnt << 8) + getc(fd); | |
| 3023 if (cnt < 0) | |
| 3024 { | |
| 483 | 3025 *cntp = SP_TRUNCERROR; |
| 376 | 3026 return NULL; |
| 3027 } | |
| 389 | 3028 *cntp = cnt; |
| 3029 if (cnt == 0) | |
| 3030 return NULL; /* nothing to read, return NULL */ | |
| 376 | 3031 |
| 483 | 3032 str = read_string(fd, cnt); |
| 3033 if (str == NULL) | |
| 495 | 3034 *cntp = SP_OTHERERROR; |
| 483 | 3035 return str; |
| 3036 } | |
| 3037 | |
| 3038 /* | |
| 3039 * Read SN_REGION: <regionname> ... | |
| 3040 * Return SP_*ERROR flags. | |
| 3041 */ | |
| 3042 static int | |
| 3043 read_region_section(fd, lp, len) | |
| 3044 FILE *fd; | |
| 3045 slang_T *lp; | |
| 3046 int len; | |
| 3047 { | |
| 3048 int i; | |
| 3049 | |
| 3050 if (len > 16) | |
| 3051 return SP_FORMERROR; | |
| 3052 for (i = 0; i < len; ++i) | |
| 3053 lp->sl_regions[i] = getc(fd); /* <regionname> */ | |
| 3054 lp->sl_regions[len] = NUL; | |
| 3055 return 0; | |
| 3056 } | |
| 3057 | |
| 3058 /* | |
| 3059 * Read SN_CHARFLAGS section: <charflagslen> <charflags> | |
| 3060 * <folcharslen> <folchars> | |
| 3061 * Return SP_*ERROR flags. | |
| 3062 */ | |
| 3063 static int | |
| 3064 read_charflags_section(fd) | |
| 3065 FILE *fd; | |
| 3066 { | |
| 3067 char_u *flags; | |
| 3068 char_u *fol; | |
| 3069 int flagslen, follen; | |
| 3070 | |
| 3071 /* <charflagslen> <charflags> */ | |
| 3072 flags = read_cnt_string(fd, 1, &flagslen); | |
| 3073 if (flagslen < 0) | |
| 3074 return flagslen; | |
| 3075 | |
| 3076 /* <folcharslen> <folchars> */ | |
| 3077 fol = read_cnt_string(fd, 2, &follen); | |
| 3078 if (follen < 0) | |
| 3079 { | |
| 3080 vim_free(flags); | |
| 3081 return follen; | |
| 3082 } | |
| 3083 | |
| 3084 /* Set the word-char flags and fill SPELL_ISUPPER() table. */ | |
| 3085 if (flags != NULL && fol != NULL) | |
| 3086 set_spell_charflags(flags, flagslen, fol); | |
| 3087 | |
| 3088 vim_free(flags); | |
| 3089 vim_free(fol); | |
| 3090 | |
| 3091 /* When <charflagslen> is zero then <fcharlen> must also be zero. */ | |
| 3092 if ((flags == NULL) != (fol == NULL)) | |
| 3093 return SP_FORMERROR; | |
| 3094 return 0; | |
| 3095 } | |
| 3096 | |
| 3097 /* | |
| 3098 * Read SN_PREFCOND section. | |
| 3099 * Return SP_*ERROR flags. | |
| 3100 */ | |
| 3101 static int | |
| 3102 read_prefcond_section(fd, lp) | |
| 3103 FILE *fd; | |
| 3104 slang_T *lp; | |
| 3105 { | |
| 3106 int cnt; | |
| 3107 int i; | |
| 3108 int n; | |
| 3109 char_u *p; | |
| 3110 char_u buf[MAXWLEN + 1]; | |
| 3111 | |
| 3112 /* <prefcondcnt> <prefcond> ... */ | |
| 695 | 3113 cnt = get2c(fd); /* <prefcondcnt> */ |
| 483 | 3114 if (cnt <= 0) |
| 3115 return SP_FORMERROR; | |
| 3116 | |
| 3117 lp->sl_prefprog = (regprog_T **)alloc_clear( | |
| 3118 (unsigned)sizeof(regprog_T *) * cnt); | |
| 3119 if (lp->sl_prefprog == NULL) | |
| 495 | 3120 return SP_OTHERERROR; |
| 483 | 3121 lp->sl_prefixcnt = cnt; |
| 3122 | |
| 376 | 3123 for (i = 0; i < cnt; ++i) |
| 483 | 3124 { |
| 3125 /* <prefcond> : <condlen> <condstr> */ | |
| 3126 n = getc(fd); /* <condlen> */ | |
| 3127 if (n < 0 || n >= MAXWLEN) | |
| 3128 return SP_FORMERROR; | |
| 3129 | |
| 3130 /* When <condlen> is zero we have an empty condition. Otherwise | |
| 3131 * compile the regexp program used to check for the condition. */ | |
| 3132 if (n > 0) | |
| 3133 { | |
| 3134 buf[0] = '^'; /* always match at one position only */ | |
| 3135 p = buf + 1; | |
| 3136 while (n-- > 0) | |
| 3137 *p++ = getc(fd); /* <condstr> */ | |
| 3138 *p = NUL; | |
| 3139 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING); | |
| 3140 } | |
| 3141 } | |
| 3142 return 0; | |
| 3143 } | |
| 3144 | |
| 3145 /* | |
| 625 | 3146 * Read REP or REPSAL items section from "fd": <repcount> <rep> ... |
| 483 | 3147 * Return SP_*ERROR flags. |
| 3148 */ | |
| 3149 static int | |
| 625 | 3150 read_rep_section(fd, gap, first) |
| 483 | 3151 FILE *fd; |
| 625 | 3152 garray_T *gap; |
| 3153 short *first; | |
| 483 | 3154 { |
| 3155 int cnt; | |
| 3156 fromto_T *ftp; | |
| 3157 int i; | |
| 3158 | |
| 695 | 3159 cnt = get2c(fd); /* <repcount> */ |
| 483 | 3160 if (cnt < 0) |
| 3161 return SP_TRUNCERROR; | |
| 3162 | |
| 3163 if (ga_grow(gap, cnt) == FAIL) | |
| 495 | 3164 return SP_OTHERERROR; |
| 483 | 3165 |
| 3166 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */ | |
| 3167 for (; gap->ga_len < cnt; ++gap->ga_len) | |
| 3168 { | |
| 3169 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len]; | |
| 3170 ftp->ft_from = read_cnt_string(fd, 1, &i); | |
| 3171 if (i < 0) | |
| 3172 return i; | |
| 3173 if (i == 0) | |
| 3174 return SP_FORMERROR; | |
| 3175 ftp->ft_to = read_cnt_string(fd, 1, &i); | |
| 3176 if (i <= 0) | |
| 3177 { | |
| 3178 vim_free(ftp->ft_from); | |
| 3179 if (i < 0) | |
| 3180 return i; | |
| 3181 return SP_FORMERROR; | |
| 3182 } | |
| 3183 } | |
| 3184 | |
| 3185 /* Fill the first-index table. */ | |
| 3186 for (i = 0; i < 256; ++i) | |
| 3187 first[i] = -1; | |
| 3188 for (i = 0; i < gap->ga_len; ++i) | |
| 3189 { | |
| 3190 ftp = &((fromto_T *)gap->ga_data)[i]; | |
| 3191 if (first[*ftp->ft_from] == -1) | |
| 3192 first[*ftp->ft_from] = i; | |
| 3193 } | |
| 3194 return 0; | |
| 3195 } | |
| 3196 | |
| 3197 /* | |
| 3198 * Read SN_SAL section: <salflags> <salcount> <sal> ... | |
| 3199 * Return SP_*ERROR flags. | |
| 3200 */ | |
| 3201 static int | |
| 3202 read_sal_section(fd, slang) | |
| 3203 FILE *fd; | |
| 3204 slang_T *slang; | |
| 3205 { | |
| 3206 int i; | |
| 3207 int cnt; | |
| 3208 garray_T *gap; | |
| 3209 salitem_T *smp; | |
| 3210 int ccnt; | |
| 3211 char_u *p; | |
| 485 | 3212 int c = NUL; |
| 483 | 3213 |
| 3214 slang->sl_sofo = FALSE; | |
| 3215 | |
| 3216 i = getc(fd); /* <salflags> */ | |
| 3217 if (i & SAL_F0LLOWUP) | |
| 3218 slang->sl_followup = TRUE; | |
| 3219 if (i & SAL_COLLAPSE) | |
| 3220 slang->sl_collapse = TRUE; | |
| 3221 if (i & SAL_REM_ACCENTS) | |
| 3222 slang->sl_rem_accents = TRUE; | |
| 3223 | |
| 695 | 3224 cnt = get2c(fd); /* <salcount> */ |
| 483 | 3225 if (cnt < 0) |
| 3226 return SP_TRUNCERROR; | |
| 3227 | |
| 3228 gap = &slang->sl_sal; | |
| 3229 ga_init2(gap, sizeof(salitem_T), 10); | |
| 548 | 3230 if (ga_grow(gap, cnt + 1) == FAIL) |
| 495 | 3231 return SP_OTHERERROR; |
| 483 | 3232 |
| 3233 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */ | |
| 3234 for (; gap->ga_len < cnt; ++gap->ga_len) | |
| 3235 { | |
| 3236 smp = &((salitem_T *)gap->ga_data)[gap->ga_len]; | |
| 3237 ccnt = getc(fd); /* <salfromlen> */ | |
| 3238 if (ccnt < 0) | |
| 3239 return SP_TRUNCERROR; | |
| 3240 if ((p = alloc(ccnt + 2)) == NULL) | |
| 495 | 3241 return SP_OTHERERROR; |
| 483 | 3242 smp->sm_lead = p; |
| 3243 | |
| 3244 /* Read up to the first special char into sm_lead. */ | |
| 3245 for (i = 0; i < ccnt; ++i) | |
| 3246 { | |
| 3247 c = getc(fd); /* <salfrom> */ | |
| 3248 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL) | |
| 3249 break; | |
| 3250 *p++ = c; | |
| 3251 } | |
| 835 | 3252 smp->sm_leadlen = (int)(p - smp->sm_lead); |
| 483 | 3253 *p++ = NUL; |
| 3254 | |
| 3255 /* Put (abc) chars in sm_oneof, if any. */ | |
| 3256 if (c == '(') | |
| 3257 { | |
| 3258 smp->sm_oneof = p; | |
| 3259 for (++i; i < ccnt; ++i) | |
| 3260 { | |
| 3261 c = getc(fd); /* <salfrom> */ | |
| 3262 if (c == ')') | |
| 3263 break; | |
| 3264 *p++ = c; | |
| 3265 } | |
| 3266 *p++ = NUL; | |
| 3267 if (++i < ccnt) | |
| 3268 c = getc(fd); | |
| 3269 } | |
| 3270 else | |
| 3271 smp->sm_oneof = NULL; | |
| 3272 | |
| 3273 /* Any following chars go in sm_rules. */ | |
| 3274 smp->sm_rules = p; | |
| 3275 if (i < ccnt) | |
| 3276 /* store the char we got while checking for end of sm_lead */ | |
| 3277 *p++ = c; | |
| 3278 for (++i; i < ccnt; ++i) | |
| 3279 *p++ = getc(fd); /* <salfrom> */ | |
| 3280 *p++ = NUL; | |
| 3281 | |
| 3282 /* <saltolen> <salto> */ | |
| 3283 smp->sm_to = read_cnt_string(fd, 1, &ccnt); | |
| 3284 if (ccnt < 0) | |
| 3285 { | |
| 3286 vim_free(smp->sm_lead); | |
| 3287 return ccnt; | |
| 3288 } | |
| 3289 | |
| 3290 #ifdef FEAT_MBYTE | |
| 3291 if (has_mbyte) | |
| 3292 { | |
| 3293 /* convert the multi-byte strings to wide char strings */ | |
| 3294 smp->sm_lead_w = mb_str2wide(smp->sm_lead); | |
| 3295 smp->sm_leadlen = mb_charlen(smp->sm_lead); | |
| 3296 if (smp->sm_oneof == NULL) | |
| 3297 smp->sm_oneof_w = NULL; | |
| 3298 else | |
| 3299 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof); | |
| 3300 if (smp->sm_to == NULL) | |
| 3301 smp->sm_to_w = NULL; | |
| 3302 else | |
| 3303 smp->sm_to_w = mb_str2wide(smp->sm_to); | |
| 3304 if (smp->sm_lead_w == NULL | |
| 3305 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL) | |
| 3306 || (smp->sm_to_w == NULL && smp->sm_to != NULL)) | |
| 3307 { | |
| 3308 vim_free(smp->sm_lead); | |
| 3309 vim_free(smp->sm_to); | |
| 3310 vim_free(smp->sm_lead_w); | |
| 3311 vim_free(smp->sm_oneof_w); | |
| 3312 vim_free(smp->sm_to_w); | |
| 495 | 3313 return SP_OTHERERROR; |
| 483 | 3314 } |
| 3315 } | |
| 3316 #endif | |
| 3317 } | |
| 3318 | |
| 548 | 3319 if (gap->ga_len > 0) |
| 3320 { | |
| 3321 /* Add one extra entry to mark the end with an empty sm_lead. Avoids | |
| 3322 * that we need to check the index every time. */ | |
| 3323 smp = &((salitem_T *)gap->ga_data)[gap->ga_len]; | |
| 3324 if ((p = alloc(1)) == NULL) | |
| 3325 return SP_OTHERERROR; | |
| 3326 p[0] = NUL; | |
| 3327 smp->sm_lead = p; | |
| 3328 smp->sm_leadlen = 0; | |
| 3329 smp->sm_oneof = NULL; | |
| 3330 smp->sm_rules = p; | |
| 3331 smp->sm_to = NULL; | |
| 3332 #ifdef FEAT_MBYTE | |
| 3333 if (has_mbyte) | |
| 3334 { | |
| 3335 smp->sm_lead_w = mb_str2wide(smp->sm_lead); | |
| 3336 smp->sm_leadlen = 0; | |
| 3337 smp->sm_oneof_w = NULL; | |
| 3338 smp->sm_to_w = NULL; | |
| 3339 } | |
| 3340 #endif | |
| 3341 ++gap->ga_len; | |
| 3342 } | |
| 3343 | |
| 483 | 3344 /* Fill the first-index table. */ |
| 3345 set_sal_first(slang); | |
| 3346 | |
| 3347 return 0; | |
| 3348 } | |
| 3349 | |
| 3350 /* | |
| 625 | 3351 * Read SN_WORDS: <word> ... |
| 3352 * Return SP_*ERROR flags. | |
| 3353 */ | |
| 3354 static int | |
| 3355 read_words_section(fd, lp, len) | |
| 3356 FILE *fd; | |
| 3357 slang_T *lp; | |
| 3358 int len; | |
| 3359 { | |
| 3360 int done = 0; | |
| 3361 int i; | |
| 1619 | 3362 int c; |
| 625 | 3363 char_u word[MAXWLEN]; |
| 3364 | |
| 3365 while (done < len) | |
| 3366 { | |
| 3367 /* Read one word at a time. */ | |
| 3368 for (i = 0; ; ++i) | |
| 3369 { | |
| 1619 | 3370 c = getc(fd); |
| 3371 if (c == EOF) | |
| 3372 return SP_TRUNCERROR; | |
| 3373 word[i] = c; | |
| 625 | 3374 if (word[i] == NUL) |
| 3375 break; | |
| 3376 if (i == MAXWLEN - 1) | |
| 3377 return SP_FORMERROR; | |
| 3378 } | |
| 3379 | |
| 3380 /* Init the count to 10. */ | |
| 3381 count_common_word(lp, word, -1, 10); | |
| 3382 done += i + 1; | |
| 3383 } | |
| 3384 return 0; | |
| 3385 } | |
| 3386 | |
| 3387 /* | |
| 3388 * Add a word to the hashtable of common words. | |
| 3389 * If it's already there then the counter is increased. | |
| 3390 */ | |
| 3391 static void | |
| 3392 count_common_word(lp, word, len, count) | |
| 3393 slang_T *lp; | |
| 3394 char_u *word; | |
| 3395 int len; /* word length, -1 for upto NUL */ | |
| 3396 int count; /* 1 to count once, 10 to init */ | |
| 3397 { | |
| 3398 hash_T hash; | |
| 3399 hashitem_T *hi; | |
| 3400 wordcount_T *wc; | |
| 3401 char_u buf[MAXWLEN]; | |
| 3402 char_u *p; | |
| 3403 | |
| 3404 if (len == -1) | |
| 3405 p = word; | |
| 3406 else | |
| 3407 { | |
| 3408 vim_strncpy(buf, word, len); | |
| 3409 p = buf; | |
| 3410 } | |
| 3411 | |
| 3412 hash = hash_hash(p); | |
| 3413 hi = hash_lookup(&lp->sl_wordcount, p, hash); | |
| 3414 if (HASHITEM_EMPTY(hi)) | |
| 3415 { | |
| 835 | 3416 wc = (wordcount_T *)alloc((unsigned)(sizeof(wordcount_T) + STRLEN(p))); |
| 625 | 3417 if (wc == NULL) |
| 3418 return; | |
| 3419 STRCPY(wc->wc_word, p); | |
| 3420 wc->wc_count = count; | |
| 3421 hash_add_item(&lp->sl_wordcount, hi, wc->wc_word, hash); | |
| 3422 } | |
| 3423 else | |
| 3424 { | |
| 3425 wc = HI2WC(hi); | |
| 3426 if ((wc->wc_count += count) < (unsigned)count) /* check for overflow */ | |
| 3427 wc->wc_count = MAXWORDCOUNT; | |
| 3428 } | |
| 3429 } | |
| 3430 | |
| 3431 /* | |
| 3432 * Adjust the score of common words. | |
| 3433 */ | |
| 3434 static int | |
| 3435 score_wordcount_adj(slang, score, word, split) | |
| 3436 slang_T *slang; | |
| 3437 int score; | |
| 3438 char_u *word; | |
| 3439 int split; /* word was split, less bonus */ | |
| 3440 { | |
| 3441 hashitem_T *hi; | |
| 3442 wordcount_T *wc; | |
| 3443 int bonus; | |
| 3444 int newscore; | |
| 3445 | |
| 3446 hi = hash_find(&slang->sl_wordcount, word); | |
| 3447 if (!HASHITEM_EMPTY(hi)) | |
| 3448 { | |
| 3449 wc = HI2WC(hi); | |
| 3450 if (wc->wc_count < SCORE_THRES2) | |
| 3451 bonus = SCORE_COMMON1; | |
| 3452 else if (wc->wc_count < SCORE_THRES3) | |
| 3453 bonus = SCORE_COMMON2; | |
| 3454 else | |
| 3455 bonus = SCORE_COMMON3; | |
| 3456 if (split) | |
| 3457 newscore = score - bonus / 2; | |
| 3458 else | |
| 3459 newscore = score - bonus; | |
| 3460 if (newscore < 0) | |
| 3461 return 0; | |
| 3462 return newscore; | |
| 3463 } | |
| 3464 return score; | |
| 3465 } | |
| 3466 | |
| 3467 /* | |
| 483 | 3468 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto> |
| 3469 * Return SP_*ERROR flags. | |
| 3470 */ | |
| 3471 static int | |
| 3472 read_sofo_section(fd, slang) | |
| 3473 FILE *fd; | |
| 3474 slang_T *slang; | |
| 3475 { | |
| 3476 int cnt; | |
| 3477 char_u *from, *to; | |
| 3478 int res; | |
| 3479 | |
| 3480 slang->sl_sofo = TRUE; | |
| 3481 | |
| 3482 /* <sofofromlen> <sofofrom> */ | |
| 3483 from = read_cnt_string(fd, 2, &cnt); | |
| 3484 if (cnt < 0) | |
| 3485 return cnt; | |
| 3486 | |
| 3487 /* <sofotolen> <sofoto> */ | |
| 3488 to = read_cnt_string(fd, 2, &cnt); | |
| 3489 if (cnt < 0) | |
| 3490 { | |
| 3491 vim_free(from); | |
| 3492 return cnt; | |
| 3493 } | |
| 3494 | |
| 3495 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */ | |
| 3496 if (from != NULL && to != NULL) | |
| 3497 res = set_sofo(slang, from, to); | |
| 3498 else if (from != NULL || to != NULL) | |
| 3499 res = SP_FORMERROR; /* only one of two strings is an error */ | |
| 3500 else | |
| 3501 res = 0; | |
| 3502 | |
| 3503 vim_free(from); | |
| 3504 vim_free(to); | |
| 3505 return res; | |
| 3506 } | |
| 3507 | |
| 3508 /* | |
| 3509 * Read the compound section from the .spl file: | |
| 809 | 3510 * <compmax> <compminlen> <compsylmax> <compoptions> <compflags> |
| 483 | 3511 * Returns SP_*ERROR flags. |
| 3512 */ | |
| 3513 static int | |
| 3514 read_compound(fd, slang, len) | |
| 3515 FILE *fd; | |
| 3516 slang_T *slang; | |
| 3517 int len; | |
| 3518 { | |
| 3519 int todo = len; | |
| 3520 int c; | |
| 3521 int atstart; | |
| 3522 char_u *pat; | |
| 3523 char_u *pp; | |
| 3524 char_u *cp; | |
| 485 | 3525 char_u *ap; |
| 1762 | 3526 char_u *crp; |
| 809 | 3527 int cnt; |
| 3528 garray_T *gap; | |
| 483 | 3529 |
| 3530 if (todo < 2) | |
| 3531 return SP_FORMERROR; /* need at least two bytes */ | |
| 3532 | |
| 3533 --todo; | |
| 3534 c = getc(fd); /* <compmax> */ | |
| 3535 if (c < 2) | |
| 3536 c = MAXWLEN; | |
| 3537 slang->sl_compmax = c; | |
| 3538 | |
| 3539 --todo; | |
| 3540 c = getc(fd); /* <compminlen> */ | |
| 3541 if (c < 1) | |
| 501 | 3542 c = 0; |
| 483 | 3543 slang->sl_compminlen = c; |
| 3544 | |
| 3545 --todo; | |
| 3546 c = getc(fd); /* <compsylmax> */ | |
| 3547 if (c < 1) | |
| 3548 c = MAXWLEN; | |
| 3549 slang->sl_compsylmax = c; | |
| 3550 | |
| 809 | 3551 c = getc(fd); /* <compoptions> */ |
| 3552 if (c != 0) | |
| 3553 ungetc(c, fd); /* be backwards compatible with Vim 7.0b */ | |
| 3554 else | |
| 3555 { | |
| 3556 --todo; | |
| 3557 c = getc(fd); /* only use the lower byte for now */ | |
| 3558 --todo; | |
| 3559 slang->sl_compoptions = c; | |
| 3560 | |
| 3561 gap = &slang->sl_comppat; | |
| 3562 c = get2c(fd); /* <comppatcount> */ | |
| 3563 todo -= 2; | |
| 3564 ga_init2(gap, sizeof(char_u *), c); | |
| 3565 if (ga_grow(gap, c) == OK) | |
| 3566 while (--c >= 0) | |
| 3567 { | |
| 3568 ((char_u **)(gap->ga_data))[gap->ga_len++] = | |
| 3569 read_cnt_string(fd, 1, &cnt); | |
| 3570 /* <comppatlen> <comppattext> */ | |
| 3571 if (cnt < 0) | |
| 3572 return cnt; | |
| 822 | 3573 todo -= cnt + 1; |
| 3574 } | |
| 3575 } | |
| 3576 if (todo < 0) | |
| 3577 return SP_FORMERROR; | |
| 809 | 3578 |
| 714 | 3579 /* Turn the COMPOUNDRULE items into a regexp pattern: |
| 483 | 3580 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$". |
| 495 | 3581 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes. |
| 3582 * Conversion to utf-8 may double the size. */ | |
| 3583 c = todo * 2 + 7; | |
| 3584 #ifdef FEAT_MBYTE | |
| 3585 if (enc_utf8) | |
| 3586 c += todo * 2; | |
| 3587 #endif | |
| 3588 pat = alloc((unsigned)c); | |
| 483 | 3589 if (pat == NULL) |
| 495 | 3590 return SP_OTHERERROR; |
| 483 | 3591 |
| 485 | 3592 /* We also need a list of all flags that can appear at the start and one |
| 3593 * for all flags. */ | |
| 483 | 3594 cp = alloc(todo + 1); |
| 3595 if (cp == NULL) | |
| 3596 { | |
| 3597 vim_free(pat); | |
| 495 | 3598 return SP_OTHERERROR; |
| 483 | 3599 } |
| 3600 slang->sl_compstartflags = cp; | |
| 3601 *cp = NUL; | |
| 3602 | |
| 485 | 3603 ap = alloc(todo + 1); |
| 3604 if (ap == NULL) | |
| 3605 { | |
| 3606 vim_free(pat); | |
| 495 | 3607 return SP_OTHERERROR; |
| 485 | 3608 } |
| 3609 slang->sl_compallflags = ap; | |
| 3610 *ap = NUL; | |
| 3611 | |
| 1762 | 3612 /* And a list of all patterns in their original form, for checking whether |
| 3613 * compounding may work in match_compoundrule(). This is freed when we | |
| 3614 * encounter a wildcard, the check doesn't work then. */ | |
| 3615 crp = alloc(todo + 1); | |
| 3616 slang->sl_comprules = crp; | |
| 3617 | |
| 483 | 3618 pp = pat; |
| 3619 *pp++ = '^'; | |
| 3620 *pp++ = '\\'; | |
| 3621 *pp++ = '('; | |
| 3622 | |
| 3623 atstart = 1; | |
| 3624 while (todo-- > 0) | |
| 3625 { | |
| 3626 c = getc(fd); /* <compflags> */ | |
| 1619 | 3627 if (c == EOF) |
| 3628 { | |
| 3629 vim_free(pat); | |
| 3630 return SP_TRUNCERROR; | |
| 3631 } | |
| 485 | 3632 |
| 3633 /* Add all flags to "sl_compallflags". */ | |
| 2691 | 3634 if (vim_strchr((char_u *)"?*+[]/", c) == NULL |
| 495 | 3635 && !byte_in_str(slang->sl_compallflags, c)) |
| 485 | 3636 { |
| 3637 *ap++ = c; | |
| 3638 *ap = NUL; | |
| 3639 } | |
| 3640 | |
| 483 | 3641 if (atstart != 0) |
| 3642 { | |
| 3643 /* At start of item: copy flags to "sl_compstartflags". For a | |
| 3644 * [abc] item set "atstart" to 2 and copy up to the ']'. */ | |
| 3645 if (c == '[') | |
| 3646 atstart = 2; | |
| 3647 else if (c == ']') | |
| 3648 atstart = 0; | |
| 3649 else | |
| 3650 { | |
| 495 | 3651 if (!byte_in_str(slang->sl_compstartflags, c)) |
| 483 | 3652 { |
| 3653 *cp++ = c; | |
| 3654 *cp = NUL; | |
| 3655 } | |
| 3656 if (atstart == 1) | |
| 3657 atstart = 0; | |
| 3658 } | |
| 3659 } | |
| 1762 | 3660 |
| 3661 /* Copy flag to "sl_comprules", unless we run into a wildcard. */ | |
| 3662 if (crp != NULL) | |
| 3663 { | |
| 2691 | 3664 if (c == '?' || c == '+' || c == '*') |
| 1762 | 3665 { |
| 3666 vim_free(slang->sl_comprules); | |
| 3667 slang->sl_comprules = NULL; | |
| 3668 crp = NULL; | |
| 3669 } | |
| 3670 else | |
| 3671 *crp++ = c; | |
| 3672 } | |
| 3673 | |
| 483 | 3674 if (c == '/') /* slash separates two items */ |
| 3675 { | |
| 3676 *pp++ = '\\'; | |
| 3677 *pp++ = '|'; | |
| 3678 atstart = 1; | |
| 3679 } | |
| 3680 else /* normal char, "[abc]" and '*' are copied as-is */ | |
| 3681 { | |
| 2691 | 3682 if (c == '?' || c == '+' || c == '~') |
| 3683 *pp++ = '\\'; /* "a?" becomes "a\?", "a+" becomes "a\+" */ | |
| 495 | 3684 #ifdef FEAT_MBYTE |
| 3685 if (enc_utf8) | |
| 3686 pp += mb_char2bytes(c, pp); | |
| 3687 else | |
| 3688 #endif | |
| 3689 *pp++ = c; | |
| 483 | 3690 } |
| 3691 } | |
| 3692 | |
| 3693 *pp++ = '\\'; | |
| 3694 *pp++ = ')'; | |
| 3695 *pp++ = '$'; | |
| 3696 *pp = NUL; | |
| 3697 | |
| 1762 | 3698 if (crp != NULL) |
| 3699 *crp = NUL; | |
| 3700 | |
| 483 | 3701 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT); |
| 3702 vim_free(pat); | |
| 3703 if (slang->sl_compprog == NULL) | |
| 3704 return SP_FORMERROR; | |
| 3705 | |
| 3706 return 0; | |
| 3707 } | |
| 3708 | |
| 495 | 3709 /* |
| 498 | 3710 * Return TRUE if byte "n" appears in "str". |
| 495 | 3711 * Like strchr() but independent of locale. |
| 3712 */ | |
| 3713 static int | |
| 498 | 3714 byte_in_str(str, n) |
| 495 | 3715 char_u *str; |
| 498 | 3716 int n; |
| 495 | 3717 { |
| 3718 char_u *p; | |
| 3719 | |
| 3720 for (p = str; *p != NUL; ++p) | |
| 498 | 3721 if (*p == n) |
| 495 | 3722 return TRUE; |
| 3723 return FALSE; | |
| 3724 } | |
| 3725 | |
| 483 | 3726 #define SY_MAXLEN 30 |
| 3727 typedef struct syl_item_S | |
| 3728 { | |
| 3729 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */ | |
| 3730 int sy_len; | |
| 3731 } syl_item_T; | |
| 3732 | |
| 3733 /* | |
| 3734 * Truncate "slang->sl_syllable" at the first slash and put the following items | |
| 3735 * in "slang->sl_syl_items". | |
| 3736 */ | |
| 3737 static int | |
| 3738 init_syl_tab(slang) | |
| 3739 slang_T *slang; | |
| 3740 { | |
| 3741 char_u *p; | |
| 3742 char_u *s; | |
| 3743 int l; | |
| 3744 syl_item_T *syl; | |
| 3745 | |
| 3746 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4); | |
| 3747 p = vim_strchr(slang->sl_syllable, '/'); | |
| 3748 while (p != NULL) | |
| 3749 { | |
| 3750 *p++ = NUL; | |
| 495 | 3751 if (*p == NUL) /* trailing slash */ |
| 483 | 3752 break; |
| 3753 s = p; | |
| 3754 p = vim_strchr(p, '/'); | |
| 3755 if (p == NULL) | |
| 835 | 3756 l = (int)STRLEN(s); |
| 483 | 3757 else |
| 835 | 3758 l = (int)(p - s); |
| 483 | 3759 if (l >= SY_MAXLEN) |
| 3760 return SP_FORMERROR; | |
| 3761 if (ga_grow(&slang->sl_syl_items, 1) == FAIL) | |
| 495 | 3762 return SP_OTHERERROR; |
| 483 | 3763 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) |
| 3764 + slang->sl_syl_items.ga_len++; | |
| 3765 vim_strncpy(syl->sy_chars, s, l); | |
| 3766 syl->sy_len = l; | |
| 3767 } | |
| 3768 return OK; | |
| 3769 } | |
| 3770 | |
| 3771 /* | |
| 3772 * Count the number of syllables in "word". | |
| 3773 * When "word" contains spaces the syllables after the last space are counted. | |
| 3774 * Returns zero if syllables are not defines. | |
| 3775 */ | |
| 3776 static int | |
| 3777 count_syllables(slang, word) | |
| 3778 slang_T *slang; | |
| 3779 char_u *word; | |
| 3780 { | |
| 3781 int cnt = 0; | |
| 3782 int skip = FALSE; | |
| 3783 char_u *p; | |
| 3784 int len; | |
| 3785 int i; | |
| 3786 syl_item_T *syl; | |
| 3787 int c; | |
| 3788 | |
| 3789 if (slang->sl_syllable == NULL) | |
| 3790 return 0; | |
| 3791 | |
| 3792 for (p = word; *p != NUL; p += len) | |
| 3793 { | |
| 3794 /* When running into a space reset counter. */ | |
| 3795 if (*p == ' ') | |
| 3796 { | |
| 3797 len = 1; | |
| 3798 cnt = 0; | |
| 3799 continue; | |
| 3800 } | |
| 3801 | |
| 3802 /* Find longest match of syllable items. */ | |
| 3803 len = 0; | |
| 3804 for (i = 0; i < slang->sl_syl_items.ga_len; ++i) | |
| 3805 { | |
| 3806 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i; | |
| 3807 if (syl->sy_len > len | |
| 3808 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0) | |
| 3809 len = syl->sy_len; | |
| 3810 } | |
| 3811 if (len != 0) /* found a match, count syllable */ | |
| 3812 { | |
| 3813 ++cnt; | |
| 3814 skip = FALSE; | |
| 3815 } | |
| 3816 else | |
| 3817 { | |
| 3818 /* No recognized syllable item, at least a syllable char then? */ | |
| 3819 #ifdef FEAT_MBYTE | |
| 3820 c = mb_ptr2char(p); | |
| 3821 len = (*mb_ptr2len)(p); | |
| 3822 #else | |
| 3823 c = *p; | |
| 3824 len = 1; | |
| 3825 #endif | |
| 3826 if (vim_strchr(slang->sl_syllable, c) == NULL) | |
| 3827 skip = FALSE; /* No, search for next syllable */ | |
| 3828 else if (!skip) | |
| 3829 { | |
| 3830 ++cnt; /* Yes, count it */ | |
| 3831 skip = TRUE; /* don't count following syllable chars */ | |
| 3832 } | |
| 3833 } | |
| 3834 } | |
| 3835 return cnt; | |
| 376 | 3836 } |
| 3837 | |
| 381 | 3838 /* |
| 3839 * Set the SOFOFROM and SOFOTO items in language "lp". | |
| 483 | 3840 * Returns SP_*ERROR flags when there is something wrong. |
| 381 | 3841 */ |
| 3842 static int | |
| 3843 set_sofo(lp, from, to) | |
| 3844 slang_T *lp; | |
| 3845 char_u *from; | |
| 3846 char_u *to; | |
| 3847 { | |
| 3848 int i; | |
| 3849 | |
| 3850 #ifdef FEAT_MBYTE | |
| 3851 garray_T *gap; | |
| 3852 char_u *s; | |
| 3853 char_u *p; | |
| 3854 int c; | |
| 3855 int *inp; | |
| 3856 | |
| 3857 if (has_mbyte) | |
| 3858 { | |
| 3859 /* Use "sl_sal" as an array with 256 pointers to a list of wide | |
| 3860 * characters. The index is the low byte of the character. | |
| 3861 * The list contains from-to pairs with a terminating NUL. | |
| 3862 * sl_sal_first[] is used for latin1 "from" characters. */ | |
| 3863 gap = &lp->sl_sal; | |
| 3864 ga_init2(gap, sizeof(int *), 1); | |
| 3865 if (ga_grow(gap, 256) == FAIL) | |
| 495 | 3866 return SP_OTHERERROR; |
| 381 | 3867 vim_memset(gap->ga_data, 0, sizeof(int *) * 256); |
| 3868 gap->ga_len = 256; | |
| 3869 | |
| 3870 /* First count the number of items for each list. Temporarily use | |
| 3871 * sl_sal_first[] for this. */ | |
| 3872 for (p = from, s = to; *p != NUL && *s != NUL; ) | |
| 3873 { | |
| 474 | 3874 c = mb_cptr2char_adv(&p); |
| 3875 mb_cptr_adv(s); | |
| 381 | 3876 if (c >= 256) |
| 3877 ++lp->sl_sal_first[c & 0xff]; | |
| 3878 } | |
| 3879 if (*p != NUL || *s != NUL) /* lengths differ */ | |
| 483 | 3880 return SP_FORMERROR; |
| 381 | 3881 |
| 3882 /* Allocate the lists. */ | |
| 3883 for (i = 0; i < 256; ++i) | |
| 3884 if (lp->sl_sal_first[i] > 0) | |
| 3885 { | |
| 3886 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1)); | |
| 3887 if (p == NULL) | |
| 495 | 3888 return SP_OTHERERROR; |
| 381 | 3889 ((int **)gap->ga_data)[i] = (int *)p; |
| 3890 *(int *)p = 0; | |
| 3891 } | |
| 3892 | |
| 3893 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal | |
| 3894 * list. */ | |
| 3895 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256); | |
| 3896 for (p = from, s = to; *p != NUL && *s != NUL; ) | |
| 3897 { | |
| 474 | 3898 c = mb_cptr2char_adv(&p); |
| 3899 i = mb_cptr2char_adv(&s); | |
| 381 | 3900 if (c >= 256) |
| 3901 { | |
| 3902 /* Append the from-to chars at the end of the list with | |
| 3903 * the low byte. */ | |
| 3904 inp = ((int **)gap->ga_data)[c & 0xff]; | |
| 3905 while (*inp != 0) | |
| 3906 ++inp; | |
| 3907 *inp++ = c; /* from char */ | |
| 3908 *inp++ = i; /* to char */ | |
| 3909 *inp++ = NUL; /* NUL at the end */ | |
| 3910 } | |
| 3911 else | |
| 3912 /* mapping byte to char is done in sl_sal_first[] */ | |
| 3913 lp->sl_sal_first[c] = i; | |
| 3914 } | |
| 3915 } | |
| 3916 else | |
| 3917 #endif | |
| 3918 { | |
| 3919 /* mapping bytes to bytes is done in sl_sal_first[] */ | |
| 3920 if (STRLEN(from) != STRLEN(to)) | |
| 483 | 3921 return SP_FORMERROR; |
| 381 | 3922 |
| 3923 for (i = 0; to[i] != NUL; ++i) | |
| 3924 lp->sl_sal_first[from[i]] = to[i]; | |
| 3925 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */ | |
| 3926 } | |
| 3927 | |
| 483 | 3928 return 0; |
| 381 | 3929 } |
| 3930 | |
| 3931 /* | |
| 3932 * Fill the first-index table for "lp". | |
| 3933 */ | |
| 3934 static void | |
| 3935 set_sal_first(lp) | |
| 3936 slang_T *lp; | |
| 3937 { | |
| 3938 salfirst_T *sfirst; | |
| 3939 int i; | |
| 3940 salitem_T *smp; | |
| 3941 int c; | |
| 3942 garray_T *gap = &lp->sl_sal; | |
| 3943 | |
| 3944 sfirst = lp->sl_sal_first; | |
| 3945 for (i = 0; i < 256; ++i) | |
| 3946 sfirst[i] = -1; | |
| 3947 smp = (salitem_T *)gap->ga_data; | |
| 3948 for (i = 0; i < gap->ga_len; ++i) | |
| 3949 { | |
| 3950 #ifdef FEAT_MBYTE | |
| 3951 if (has_mbyte) | |
| 3952 /* Use the lowest byte of the first character. For latin1 it's | |
| 3953 * the character, for other encodings it should differ for most | |
| 3954 * characters. */ | |
| 3955 c = *smp[i].sm_lead_w & 0xff; | |
| 3956 else | |
| 3957 #endif | |
| 3958 c = *smp[i].sm_lead; | |
| 3959 if (sfirst[c] == -1) | |
| 3960 { | |
| 3961 sfirst[c] = i; | |
| 3962 #ifdef FEAT_MBYTE | |
| 3963 if (has_mbyte) | |
| 3964 { | |
| 3965 int n; | |
| 3966 | |
| 3967 /* Make sure all entries with this byte are following each | |
| 3968 * other. Move the ones that are in the wrong position. Do | |
| 3969 * keep the same ordering! */ | |
| 3970 while (i + 1 < gap->ga_len | |
| 3971 && (*smp[i + 1].sm_lead_w & 0xff) == c) | |
| 3972 /* Skip over entry with same index byte. */ | |
| 3973 ++i; | |
| 3974 | |
| 3975 for (n = 1; i + n < gap->ga_len; ++n) | |
| 3976 if ((*smp[i + n].sm_lead_w & 0xff) == c) | |
| 3977 { | |
| 3978 salitem_T tsal; | |
| 3979 | |
| 3980 /* Move entry with same index byte after the entries | |
| 3981 * we already found. */ | |
| 3982 ++i; | |
| 3983 --n; | |
| 3984 tsal = smp[i + n]; | |
| 3985 mch_memmove(smp + i + 1, smp + i, | |
| 3986 sizeof(salitem_T) * n); | |
| 3987 smp[i] = tsal; | |
| 3988 } | |
| 3989 } | |
| 3990 #endif | |
| 3991 } | |
| 3992 } | |
| 3993 } | |
| 376 | 3994 |
| 372 | 3995 #ifdef FEAT_MBYTE |
| 3996 /* | |
| 3997 * Turn a multi-byte string into a wide character string. | |
| 3998 * Return it in allocated memory (NULL for out-of-memory) | |
| 3999 */ | |
| 4000 static int * | |
| 4001 mb_str2wide(s) | |
| 4002 char_u *s; | |
| 4003 { | |
| 4004 int *res; | |
| 4005 char_u *p; | |
| 4006 int i = 0; | |
| 4007 | |
| 4008 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1)); | |
| 4009 if (res != NULL) | |
| 4010 { | |
| 4011 for (p = s; *p != NUL; ) | |
| 4012 res[i++] = mb_ptr2char_adv(&p); | |
| 4013 res[i] = NUL; | |
| 4014 } | |
| 4015 return res; | |
| 4016 } | |
| 4017 #endif | |
| 4018 | |
| 236 | 4019 /* |
| 625 | 4020 * Read a tree from the .spl or .sug file. |
| 4021 * Allocates the memory and stores pointers in "bytsp" and "idxsp". | |
| 4022 * This is skipped when the tree has zero length. | |
| 4023 * Returns zero when OK, SP_ value for an error. | |
| 4024 */ | |
| 4025 static int | |
| 4026 spell_read_tree(fd, bytsp, idxsp, prefixtree, prefixcnt) | |
| 4027 FILE *fd; | |
| 4028 char_u **bytsp; | |
| 4029 idx_T **idxsp; | |
| 4030 int prefixtree; /* TRUE for the prefix tree */ | |
| 4031 int prefixcnt; /* when "prefixtree" is TRUE: prefix count */ | |
| 4032 { | |
| 4033 int len; | |
| 4034 int idx; | |
| 4035 char_u *bp; | |
| 4036 idx_T *ip; | |
| 4037 | |
| 4038 /* The tree size was computed when writing the file, so that we can | |
| 4039 * allocate it as one long block. <nodecount> */ | |
| 695 | 4040 len = get4c(fd); |
| 625 | 4041 if (len < 0) |
| 4042 return SP_TRUNCERROR; | |
| 4043 if (len > 0) | |
| 4044 { | |
| 4045 /* Allocate the byte array. */ | |
| 4046 bp = lalloc((long_u)len, TRUE); | |
| 4047 if (bp == NULL) | |
| 4048 return SP_OTHERERROR; | |
| 4049 *bytsp = bp; | |
| 4050 | |
| 4051 /* Allocate the index array. */ | |
| 4052 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE); | |
| 4053 if (ip == NULL) | |
| 4054 return SP_OTHERERROR; | |
| 4055 *idxsp = ip; | |
| 4056 | |
| 4057 /* Recursively read the tree and store it in the array. */ | |
| 4058 idx = read_tree_node(fd, bp, ip, len, 0, prefixtree, prefixcnt); | |
| 4059 if (idx < 0) | |
| 4060 return idx; | |
| 4061 } | |
| 4062 return 0; | |
| 4063 } | |
| 4064 | |
| 4065 /* | |
| 300 | 4066 * Read one row of siblings from the spell file and store it in the byte array |
| 4067 * "byts" and index array "idxs". Recursively read the children. | |
| 4068 * | |
| 625 | 4069 * NOTE: The code here must match put_node()! |
| 4070 * | |
| 4071 * Returns the index (>= 0) following the siblings. | |
| 4072 * Returns SP_TRUNCERROR if the file is shorter than expected. | |
| 4073 * Returns SP_FORMERROR if there is a format error. | |
| 236 | 4074 */ |
| 324 | 4075 static idx_T |
| 625 | 4076 read_tree_node(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr) |
| 300 | 4077 FILE *fd; |
| 4078 char_u *byts; | |
| 324 | 4079 idx_T *idxs; |
| 300 | 4080 int maxidx; /* size of arrays */ |
| 324 | 4081 idx_T startidx; /* current index in "byts" and "idxs" */ |
| 339 | 4082 int prefixtree; /* TRUE for reading PREFIXTREE */ |
| 4083 int maxprefcondnr; /* maximum for <prefcondnr> */ | |
| 236 | 4084 { |
| 300 | 4085 int len; |
| 4086 int i; | |
| 4087 int n; | |
| 324 | 4088 idx_T idx = startidx; |
| 300 | 4089 int c; |
| 366 | 4090 int c2; |
| 300 | 4091 #define SHARED_MASK 0x8000000 |
| 236 | 4092 |
| 300 | 4093 len = getc(fd); /* <siblingcount> */ |
| 4094 if (len <= 0) | |
| 625 | 4095 return SP_TRUNCERROR; |
| 300 | 4096 |
| 4097 if (startidx + len >= maxidx) | |
| 625 | 4098 return SP_FORMERROR; |
| 300 | 4099 byts[idx++] = len; |
| 4100 | |
| 4101 /* Read the byte values, flag/region bytes and shared indexes. */ | |
| 4102 for (i = 1; i <= len; ++i) | |
| 236 | 4103 { |
| 300 | 4104 c = getc(fd); /* <byte> */ |
| 4105 if (c < 0) | |
| 625 | 4106 return SP_TRUNCERROR; |
| 300 | 4107 if (c <= BY_SPECIAL) |
| 4108 { | |
| 366 | 4109 if (c == BY_NOFLAGS && !prefixtree) |
| 300 | 4110 { |
| 4111 /* No flags, all regions. */ | |
| 4112 idxs[idx] = 0; | |
| 4113 c = 0; | |
| 4114 } | |
| 390 | 4115 else if (c != BY_INDEX) |
| 300 | 4116 { |
| 339 | 4117 if (prefixtree) |
| 4118 { | |
| 455 | 4119 /* Read the optional pflags byte, the prefix ID and the |
| 4120 * condition nr. In idxs[] store the prefix ID in the low | |
| 4121 * byte, the condition index shifted up 8 bits, the flags | |
| 4122 * shifted up 24 bits. */ | |
| 4123 if (c == BY_FLAGS) | |
| 4124 c = getc(fd) << 24; /* <pflags> */ | |
| 4125 else | |
| 4126 c = 0; | |
| 4127 | |
| 481 | 4128 c |= getc(fd); /* <affixID> */ |
| 455 | 4129 |
| 695 | 4130 n = get2c(fd); /* <prefcondnr> */ |
| 339 | 4131 if (n >= maxprefcondnr) |
| 625 | 4132 return SP_FORMERROR; |
| 455 | 4133 c |= (n << 8); |
| 339 | 4134 } |
| 390 | 4135 else /* c must be BY_FLAGS or BY_FLAGS2 */ |
| 339 | 4136 { |
| 4137 /* Read flags and optional region and prefix ID. In | |
| 390 | 4138 * idxs[] the flags go in the low two bytes, region above |
| 4139 * that and prefix ID above the region. */ | |
| 4140 c2 = c; | |
| 339 | 4141 c = getc(fd); /* <flags> */ |
| 390 | 4142 if (c2 == BY_FLAGS2) |
| 4143 c = (getc(fd) << 8) + c; /* <flags2> */ | |
| 339 | 4144 if (c & WF_REGION) |
| 390 | 4145 c = (getc(fd) << 16) + c; /* <region> */ |
| 481 | 4146 if (c & WF_AFX) |
| 4147 c = (getc(fd) << 24) + c; /* <affixID> */ | |
| 339 | 4148 } |
| 4149 | |
| 300 | 4150 idxs[idx] = c; |
| 4151 c = 0; | |
| 4152 } | |
| 4153 else /* c == BY_INDEX */ | |
| 4154 { | |
| 4155 /* <nodeidx> */ | |
| 695 | 4156 n = get3c(fd); |
| 300 | 4157 if (n < 0 || n >= maxidx) |
| 625 | 4158 return SP_FORMERROR; |
| 300 | 4159 idxs[idx] = n + SHARED_MASK; |
| 4160 c = getc(fd); /* <xbyte> */ | |
| 4161 } | |
| 4162 } | |
| 4163 byts[idx++] = c; | |
| 236 | 4164 } |
| 4165 | |
| 300 | 4166 /* Recursively read the children for non-shared siblings. |
| 4167 * Skip the end-of-word ones (zero byte value) and the shared ones (and | |
| 4168 * remove SHARED_MASK) */ | |
| 4169 for (i = 1; i <= len; ++i) | |
| 4170 if (byts[startidx + i] != 0) | |
| 4171 { | |
| 4172 if (idxs[startidx + i] & SHARED_MASK) | |
| 4173 idxs[startidx + i] &= ~SHARED_MASK; | |
| 4174 else | |
| 4175 { | |
| 4176 idxs[startidx + i] = idx; | |
| 625 | 4177 idx = read_tree_node(fd, byts, idxs, maxidx, idx, |
| 339 | 4178 prefixtree, maxprefcondnr); |
| 300 | 4179 if (idx < 0) |
| 4180 break; | |
| 4181 } | |
| 4182 } | |
| 236 | 4183 |
| 300 | 4184 return idx; |
| 236 | 4185 } |
| 4186 | |
| 4187 /* | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4188 * Parse 'spelllang' and set w_s->b_langp accordingly. |
| 351 | 4189 * Returns NULL if it's OK, an error message otherwise. |
| 236 | 4190 */ |
| 4191 char_u * | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4192 did_set_spelllang(wp) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4193 win_T *wp; |
| 236 | 4194 { |
| 4195 garray_T ga; | |
| 351 | 4196 char_u *splp; |
| 236 | 4197 char_u *region; |
| 416 | 4198 char_u region_cp[3]; |
| 355 | 4199 int filename; |
| 236 | 4200 int region_mask; |
| 503 | 4201 slang_T *slang; |
| 236 | 4202 int c; |
| 351 | 4203 char_u lang[MAXWLEN + 1]; |
| 323 | 4204 char_u spf_name[MAXPATHL]; |
| 351 | 4205 int len; |
| 4206 char_u *p; | |
| 381 | 4207 int round; |
| 385 | 4208 char_u *spf; |
| 389 | 4209 char_u *use_region = NULL; |
| 4210 int dont_use_region = FALSE; | |
| 501 | 4211 int nobreak = FALSE; |
| 503 | 4212 int i, j; |
| 4213 langp_T *lp, *lp2; | |
| 1185 | 4214 static int recursive = FALSE; |
| 4215 char_u *ret_msg = NULL; | |
| 4216 char_u *spl_copy; | |
| 4217 | |
| 4218 /* We don't want to do this recursively. May happen when a language is | |
| 4219 * not available and the SpellFileMissing autocommand opens a new buffer | |
| 4220 * in which 'spell' is set. */ | |
| 4221 if (recursive) | |
| 4222 return NULL; | |
| 4223 recursive = TRUE; | |
| 236 | 4224 |
| 4225 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
|
4226 clear_midword(wp); |
| 236 | 4227 |
| 4352 | 4228 /* Make a copy of 'spelllang', the SpellFileMissing autocommands may change |
| 1185 | 4229 * it under our fingers. */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4230 spl_copy = vim_strsave(wp->w_s->b_p_spl); |
| 1185 | 4231 if (spl_copy == NULL) |
| 4232 goto theend; | |
| 4233 | |
| 5485 | 4234 #ifdef FEAT_MBYTE |
| 5477 | 4235 wp->w_s->b_cjk = 0; |
| 5485 | 4236 #endif |
| 5477 | 4237 |
| 4238 /* Loop over comma separated language names. */ | |
| 1185 | 4239 for (splp = spl_copy; *splp != NUL; ) |
| 351 | 4240 { |
| 4241 /* Get one language name. */ | |
| 4242 copy_option_part(&splp, lang, MAXWLEN, ","); | |
| 240 | 4243 region = NULL; |
| 835 | 4244 len = (int)STRLEN(lang); |
| 355 | 4245 |
| 5477 | 4246 if (STRCMP(lang, "cjk") == 0) |
| 4247 { | |
| 5485 | 4248 #ifdef FEAT_MBYTE |
| 5477 | 4249 wp->w_s->b_cjk = 1; |
| 5485 | 4250 #endif |
| 5477 | 4251 continue; |
| 4252 } | |
| 4253 | |
| 355 | 4254 /* If the name ends in ".spl" use it as the name of the spell file. |
| 4255 * If there is a region name let "region" point to it and remove it | |
| 4256 * from the name. */ | |
| 4257 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0) | |
| 236 | 4258 { |
| 355 | 4259 filename = TRUE; |
| 4260 | |
| 416 | 4261 /* Locate a region and remove it from the file name. */ |
| 4262 p = vim_strchr(gettail(lang), '_'); | |
| 4263 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2]) | |
| 4264 && !ASCII_ISALPHA(p[3])) | |
| 4265 { | |
| 4266 vim_strncpy(region_cp, p + 1, 2); | |
| 4267 mch_memmove(p, p + 3, len - (p - lang) - 2); | |
| 4268 len -= 3; | |
| 4269 region = region_cp; | |
| 4270 } | |
| 4271 else | |
| 4272 dont_use_region = TRUE; | |
| 4273 | |
| 355 | 4274 /* Check if we loaded this language before. */ |
| 503 | 4275 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 4276 if (fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME) | |
| 355 | 4277 break; |
| 236 | 4278 } |
| 355 | 4279 else |
| 4280 { | |
| 4281 filename = FALSE; | |
| 4282 if (len > 3 && lang[len - 3] == '_') | |
| 4283 { | |
| 4284 region = lang + len - 2; | |
| 4285 len -= 3; | |
| 4286 lang[len] = NUL; | |
| 389 | 4287 } |
| 4288 else | |
| 4289 dont_use_region = TRUE; | |
| 355 | 4290 |
| 4291 /* Check if we loaded this language before. */ | |
| 503 | 4292 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 4293 if (STRICMP(lang, slang->sl_name) == 0) | |
| 355 | 4294 break; |
| 4295 } | |
| 236 | 4296 |
| 416 | 4297 if (region != NULL) |
| 4298 { | |
| 4299 /* If the region differs from what was used before then don't | |
| 4300 * use it for 'spellfile'. */ | |
| 4301 if (use_region != NULL && STRCMP(region, use_region) != 0) | |
| 4302 dont_use_region = TRUE; | |
| 4303 use_region = region; | |
| 4304 } | |
| 4305 | |
| 351 | 4306 /* If not found try loading the language now. */ |
| 503 | 4307 if (slang == NULL) |
| 355 | 4308 { |
| 4309 if (filename) | |
| 4310 (void)spell_load_file(lang, lang, NULL, FALSE); | |
| 4311 else | |
| 1185 | 4312 { |
| 355 | 4313 spell_load_lang(lang); |
| 1185 | 4314 #ifdef FEAT_AUTOCMD |
| 4315 /* SpellFileMissing autocommands may do anything, including | |
| 4316 * destroying the buffer we are using... */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4317 if (!buf_valid(wp->w_buffer)) |
| 1185 | 4318 { |
| 4319 ret_msg = (char_u *)"E797: SpellFileMissing autocommand deleted buffer"; | |
| 4320 goto theend; | |
| 4321 } | |
| 4322 #endif | |
| 4323 } | |
| 355 | 4324 } |
| 236 | 4325 |
| 307 | 4326 /* |
| 351 | 4327 * Loop over the languages, there can be several files for "lang". |
| 307 | 4328 */ |
| 503 | 4329 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 4330 if (filename ? fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME | |
| 4331 : STRICMP(lang, slang->sl_name) == 0) | |
| 236 | 4332 { |
| 316 | 4333 region_mask = REGION_ALL; |
| 355 | 4334 if (!filename && region != NULL) |
| 236 | 4335 { |
| 307 | 4336 /* find region in sl_regions */ |
| 503 | 4337 c = find_region(slang->sl_regions, region); |
| 307 | 4338 if (c == REGION_ALL) |
| 4339 { | |
| 503 | 4340 if (slang->sl_add) |
| 389 | 4341 { |
| 503 | 4342 if (*slang->sl_regions != NUL) |
| 389 | 4343 /* This addition file is for other regions. */ |
| 4344 region_mask = 0; | |
| 4345 } | |
| 4346 else | |
| 4347 /* This is probably an error. Give a warning and | |
| 4348 * accept the words anyway. */ | |
| 351 | 4349 smsg((char_u *) |
| 4350 _("Warning: region %s not supported"), | |
| 4351 region); | |
| 307 | 4352 } |
| 4353 else | |
| 4354 region_mask = 1 << c; | |
| 236 | 4355 } |
| 307 | 4356 |
| 389 | 4357 if (region_mask != 0) |
| 307 | 4358 { |
| 389 | 4359 if (ga_grow(&ga, 1) == FAIL) |
| 4360 { | |
| 4361 ga_clear(&ga); | |
| 1185 | 4362 ret_msg = e_outofmem; |
| 4363 goto theend; | |
| 389 | 4364 } |
| 503 | 4365 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 389 | 4366 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 4367 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4368 use_midword(slang, wp); |
| 503 | 4369 if (slang->sl_nobreak) |
| 501 | 4370 nobreak = TRUE; |
| 307 | 4371 } |
| 385 | 4372 } |
| 4373 } | |
| 4374 | |
| 4375 /* round 0: load int_wordlist, if possible. | |
| 4376 * round 1: load first name in 'spellfile'. | |
| 4377 * round 2: load second name in 'spellfile. | |
| 4378 * etc. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4379 spf = curwin->w_s->b_p_spf; |
| 385 | 4380 for (round = 0; round == 0 || *spf != NUL; ++round) |
| 4381 { | |
| 4382 if (round == 0) | |
| 4383 { | |
| 4384 /* Internal wordlist, if there is one. */ | |
| 4385 if (int_wordlist == NULL) | |
| 381 | 4386 continue; |
| 385 | 4387 int_wordlist_spl(spf_name); |
| 381 | 4388 } |
| 4389 else | |
| 4390 { | |
| 385 | 4391 /* One entry in 'spellfile'. */ |
| 4392 copy_option_part(&spf, spf_name, MAXPATHL - 5, ","); | |
| 4393 STRCAT(spf_name, ".spl"); | |
| 4394 | |
| 4395 /* If it was already found above then skip it. */ | |
| 4396 for (c = 0; c < ga.ga_len; ++c) | |
| 500 | 4397 { |
| 4398 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname; | |
| 4399 if (p != NULL && fullpathcmp(spf_name, p, FALSE) == FPC_SAME) | |
| 385 | 4400 break; |
| 500 | 4401 } |
| 385 | 4402 if (c < ga.ga_len) |
| 381 | 4403 continue; |
| 4404 } | |
| 4405 | |
| 351 | 4406 /* Check if it was loaded already. */ |
| 503 | 4407 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 4408 if (fullpathcmp(spf_name, slang->sl_fname, FALSE) == FPC_SAME) | |
| 323 | 4409 break; |
| 503 | 4410 if (slang == NULL) |
| 323 | 4411 { |
| 351 | 4412 /* Not loaded, try loading it now. The language name includes the |
| 385 | 4413 * region name, the region is ignored otherwise. for int_wordlist |
| 4414 * use an arbitrary name. */ | |
| 4415 if (round == 0) | |
| 4416 STRCPY(lang, "internal wordlist"); | |
| 4417 else | |
| 4418 { | |
| 4419 vim_strncpy(lang, gettail(spf_name), MAXWLEN); | |
| 381 | 4420 p = vim_strchr(lang, '.'); |
| 4421 if (p != NULL) | |
| 4422 *p = NUL; /* truncate at ".encoding.add" */ | |
| 4423 } | |
| 503 | 4424 slang = spell_load_file(spf_name, lang, NULL, TRUE); |
| 501 | 4425 |
| 4426 /* If one of the languages has NOBREAK we assume the addition | |
| 4427 * files also have this. */ | |
| 503 | 4428 if (slang != NULL && nobreak) |
| 4429 slang->sl_nobreak = TRUE; | |
| 4430 } | |
| 4431 if (slang != NULL && ga_grow(&ga, 1) == OK) | |
| 323 | 4432 { |
| 389 | 4433 region_mask = REGION_ALL; |
| 4434 if (use_region != NULL && !dont_use_region) | |
| 4435 { | |
| 4436 /* find region in sl_regions */ | |
| 503 | 4437 c = find_region(slang->sl_regions, use_region); |
| 389 | 4438 if (c != REGION_ALL) |
| 4439 region_mask = 1 << c; | |
| 503 | 4440 else if (*slang->sl_regions != NUL) |
| 389 | 4441 /* This spell file is for other regions. */ |
| 4442 region_mask = 0; | |
| 4443 } | |
| 4444 | |
| 4445 if (region_mask != 0) | |
| 4446 { | |
| 503 | 4447 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 4448 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL; | |
| 4449 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL; | |
| 389 | 4450 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 4451 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4452 use_midword(slang, wp); |
| 389 | 4453 } |
| 323 | 4454 } |
| 4455 } | |
| 4456 | |
| 236 | 4457 /* 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
|
4458 ga_clear(&wp->w_s->b_langp); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4459 wp->w_s->b_langp = ga; |
| 236 | 4460 |
| 503 | 4461 /* For each language figure out what language to use for sound folding and |
| 4462 * REP items. If the language doesn't support it itself use another one | |
| 4463 * with the same name. E.g. for "en-math" use "en". */ | |
| 4464 for (i = 0; i < ga.ga_len; ++i) | |
| 4465 { | |
| 4466 lp = LANGP_ENTRY(ga, i); | |
| 4467 | |
| 4468 /* sound folding */ | |
| 4469 if (lp->lp_slang->sl_sal.ga_len > 0) | |
| 4470 /* language does sound folding itself */ | |
| 4471 lp->lp_sallang = lp->lp_slang; | |
| 4472 else | |
| 4473 /* find first similar language that does sound folding */ | |
| 4474 for (j = 0; j < ga.ga_len; ++j) | |
| 4475 { | |
| 4476 lp2 = LANGP_ENTRY(ga, j); | |
| 4477 if (lp2->lp_slang->sl_sal.ga_len > 0 | |
| 4478 && STRNCMP(lp->lp_slang->sl_name, | |
| 4479 lp2->lp_slang->sl_name, 2) == 0) | |
| 4480 { | |
| 4481 lp->lp_sallang = lp2->lp_slang; | |
| 4482 break; | |
| 4483 } | |
| 4484 } | |
| 4485 | |
| 4486 /* REP items */ | |
| 4487 if (lp->lp_slang->sl_rep.ga_len > 0) | |
| 4488 /* language has REP items itself */ | |
| 4489 lp->lp_replang = lp->lp_slang; | |
| 4490 else | |
| 625 | 4491 /* find first similar language that has REP items */ |
| 503 | 4492 for (j = 0; j < ga.ga_len; ++j) |
| 4493 { | |
| 4494 lp2 = LANGP_ENTRY(ga, j); | |
| 4495 if (lp2->lp_slang->sl_rep.ga_len > 0 | |
| 4496 && STRNCMP(lp->lp_slang->sl_name, | |
| 4497 lp2->lp_slang->sl_name, 2) == 0) | |
| 4498 { | |
| 4499 lp->lp_replang = lp2->lp_slang; | |
| 4500 break; | |
| 4501 } | |
| 4502 } | |
| 4503 } | |
| 4504 | |
| 1185 | 4505 theend: |
| 4506 vim_free(spl_copy); | |
| 4507 recursive = FALSE; | |
| 5891 | 4508 redraw_win_later(wp, NOT_VALID); |
| 1185 | 4509 return ret_msg; |
| 236 | 4510 } |
| 4511 | |
| 4512 /* | |
| 376 | 4513 * Clear the midword characters for buffer "buf". |
| 4514 */ | |
| 4515 static void | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4516 clear_midword(wp) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4517 win_T *wp; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4518 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4519 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
|
4520 #ifdef FEAT_MBYTE |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4521 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
|
4522 wp->w_s->b_spell_ismw_mb = NULL; |
| 376 | 4523 #endif |
| 4524 } | |
| 4525 | |
| 4526 /* | |
| 4527 * Use the "sl_midword" field of language "lp" for buffer "buf". | |
| 4528 * They add up to any currently used midword characters. | |
| 4529 */ | |
| 4530 static void | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4531 use_midword(lp, wp) |
| 376 | 4532 slang_T *lp; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4533 win_T *wp; |
| 376 | 4534 { |
| 4535 char_u *p; | |
| 4536 | |
| 389 | 4537 if (lp->sl_midword == NULL) /* there aren't any */ |
| 4538 return; | |
| 4539 | |
| 376 | 4540 for (p = lp->sl_midword; *p != NUL; ) |
| 4541 #ifdef FEAT_MBYTE | |
| 4542 if (has_mbyte) | |
| 4543 { | |
| 4544 int c, l, n; | |
| 4545 char_u *bp; | |
| 4546 | |
| 4547 c = mb_ptr2char(p); | |
| 474 | 4548 l = (*mb_ptr2len)(p); |
| 4549 if (c < 256 && l <= 2) | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4550 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
|
4551 else if (wp->w_s->b_spell_ismw_mb == NULL) |
| 376 | 4552 /* 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
|
4553 wp->w_s->b_spell_ismw_mb = vim_strnsave(p, l); |
| 376 | 4554 else |
| 4555 { | |
| 4556 /* 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
|
4557 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
|
4558 bp = vim_strnsave(wp->w_s->b_spell_ismw_mb, n + l); |
| 376 | 4559 if (bp != NULL) |
| 4560 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4561 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
|
4562 wp->w_s->b_spell_ismw_mb = bp; |
| 376 | 4563 vim_strncpy(bp + n, p, l); |
| 4564 } | |
| 4565 } | |
| 4566 p += l; | |
| 4567 } | |
| 4568 else | |
| 4569 #endif | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4570 wp->w_s->b_spell_ismw[*p++] = TRUE; |
| 376 | 4571 } |
| 4572 | |
| 4573 /* | |
| 236 | 4574 * Find the region "region[2]" in "rp" (points to "sl_regions"). |
| 4575 * Each region is simply stored as the two characters of it's name. | |
| 381 | 4576 * Returns the index if found (first is 0), REGION_ALL if not found. |
| 236 | 4577 */ |
| 4578 static int | |
| 4579 find_region(rp, region) | |
| 4580 char_u *rp; | |
| 4581 char_u *region; | |
| 4582 { | |
| 4583 int i; | |
| 4584 | |
| 4585 for (i = 0; ; i += 2) | |
| 4586 { | |
| 4587 if (rp[i] == NUL) | |
| 4588 return REGION_ALL; | |
| 4589 if (rp[i] == region[0] && rp[i + 1] == region[1]) | |
| 4590 break; | |
| 4591 } | |
| 4592 return i / 2; | |
| 4593 } | |
| 4594 | |
| 4595 /* | |
| 323 | 4596 * Return case type of word: |
| 236 | 4597 * w word 0 |
| 300 | 4598 * Word WF_ONECAP |
| 4599 * W WORD WF_ALLCAP | |
| 4600 * WoRd wOrd WF_KEEPCAP | |
| 236 | 4601 */ |
| 4602 static int | |
| 4603 captype(word, end) | |
| 4604 char_u *word; | |
| 323 | 4605 char_u *end; /* When NULL use up to NUL byte. */ |
| 236 | 4606 { |
| 4607 char_u *p; | |
| 4608 int c; | |
| 4609 int firstcap; | |
| 4610 int allcap; | |
| 4611 int past_second = FALSE; /* past second word char */ | |
| 4612 | |
| 4613 /* find first letter */ | |
| 5477 | 4614 for (p = word; !spell_iswordp_nmw(p, curwin); mb_ptr_adv(p)) |
| 323 | 4615 if (end == NULL ? *p == NUL : p >= end) |
| 236 | 4616 return 0; /* only non-word characters, illegal word */ |
| 4617 #ifdef FEAT_MBYTE | |
| 310 | 4618 if (has_mbyte) |
| 4619 c = mb_ptr2char_adv(&p); | |
| 4620 else | |
| 236 | 4621 #endif |
| 310 | 4622 c = *p++; |
| 324 | 4623 firstcap = allcap = SPELL_ISUPPER(c); |
| 236 | 4624 |
| 4625 /* | |
| 4626 * Need to check all letters to find a word with mixed upper/lower. | |
| 4627 * But a word with an upper char only at start is a ONECAP. | |
| 4628 */ | |
| 323 | 4629 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p)) |
| 5477 | 4630 if (spell_iswordp_nmw(p, curwin)) |
| 236 | 4631 { |
| 455 | 4632 c = PTR2CHAR(p); |
| 324 | 4633 if (!SPELL_ISUPPER(c)) |
| 236 | 4634 { |
| 4635 /* UUl -> KEEPCAP */ | |
| 4636 if (past_second && allcap) | |
| 300 | 4637 return WF_KEEPCAP; |
| 236 | 4638 allcap = FALSE; |
| 4639 } | |
| 4640 else if (!allcap) | |
| 4641 /* UlU -> KEEPCAP */ | |
| 300 | 4642 return WF_KEEPCAP; |
| 236 | 4643 past_second = TRUE; |
| 4644 } | |
| 4645 | |
| 4646 if (allcap) | |
| 300 | 4647 return WF_ALLCAP; |
| 236 | 4648 if (firstcap) |
| 300 | 4649 return WF_ONECAP; |
| 236 | 4650 return 0; |
| 4651 } | |
| 4652 | |
| 474 | 4653 /* |
| 4654 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a | |
| 4655 * capital. So that make_case_word() can turn WOrd into Word. | |
| 4656 * Add ALLCAP for "WOrD". | |
| 4657 */ | |
| 4658 static int | |
| 4659 badword_captype(word, end) | |
| 4660 char_u *word; | |
| 4661 char_u *end; | |
| 4662 { | |
| 4663 int flags = captype(word, end); | |
| 475 | 4664 int c; |
| 474 | 4665 int l, u; |
| 4666 int first; | |
| 4667 char_u *p; | |
| 4668 | |
| 4669 if (flags & WF_KEEPCAP) | |
| 4670 { | |
| 4671 /* Count the number of UPPER and lower case letters. */ | |
| 4672 l = u = 0; | |
| 4673 first = FALSE; | |
| 4674 for (p = word; p < end; mb_ptr_adv(p)) | |
| 4675 { | |
| 475 | 4676 c = PTR2CHAR(p); |
| 4677 if (SPELL_ISUPPER(c)) | |
| 474 | 4678 { |
| 4679 ++u; | |
| 4680 if (p == word) | |
| 4681 first = TRUE; | |
| 4682 } | |
| 4683 else | |
| 4684 ++l; | |
| 4685 } | |
| 4686 | |
| 4687 /* If there are more UPPER than lower case letters suggest an | |
| 4688 * ALLCAP word. Otherwise, if the first letter is UPPER then | |
| 4689 * suggest ONECAP. Exception: "ALl" most likely should be "All", | |
| 4690 * require three upper case letters. */ | |
| 4691 if (u > l && u > 2) | |
| 4692 flags |= WF_ALLCAP; | |
| 4693 else if (first) | |
| 4694 flags |= WF_ONECAP; | |
| 633 | 4695 |
| 4696 if (u >= 2 && l >= 2) /* maCARONI maCAroni */ | |
| 4697 flags |= WF_MIXCAP; | |
| 474 | 4698 } |
| 4699 return flags; | |
| 4700 } | |
| 4701 | |
| 5519 | 4702 /* |
| 4703 * Delete the internal wordlist and its .spl file. | |
| 4704 */ | |
| 4705 void | |
| 4706 spell_delete_wordlist() | |
| 4707 { | |
| 4708 char_u fname[MAXPATHL]; | |
| 4709 | |
| 4710 if (int_wordlist != NULL) | |
| 4711 { | |
| 4712 mch_remove(int_wordlist); | |
| 4713 int_wordlist_spl(fname); | |
| 4714 mch_remove(fname); | |
| 4715 vim_free(int_wordlist); | |
| 4716 int_wordlist = NULL; | |
| 4717 } | |
| 4718 } | |
| 4719 | |
| 4720 #if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO) | |
| 355 | 4721 /* |
| 4722 * Free all languages. | |
| 4723 */ | |
| 4724 void | |
| 4725 spell_free_all() | |
| 4726 { | |
| 503 | 4727 slang_T *slang; |
| 355 | 4728 buf_T *buf; |
| 4729 | |
| 2580 | 4730 /* Go through all buffers and handle 'spelllang'. <VN> */ |
| 355 | 4731 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
|
4732 ga_clear(&buf->b_s.b_langp); |
| 355 | 4733 |
| 4734 while (first_lang != NULL) | |
| 4735 { | |
| 503 | 4736 slang = first_lang; |
| 4737 first_lang = slang->sl_next; | |
| 4738 slang_free(slang); | |
| 355 | 4739 } |
| 366 | 4740 |
| 5519 | 4741 spell_delete_wordlist(); |
| 381 | 4742 |
| 626 | 4743 vim_free(repl_to); |
| 4744 repl_to = NULL; | |
| 4745 vim_free(repl_from); | |
| 4746 repl_from = NULL; | |
| 355 | 4747 } |
| 5519 | 4748 #endif |
| 4749 | |
| 4750 #if defined(FEAT_MBYTE) || defined(PROTO) | |
| 236 | 4751 /* |
| 4752 * Clear all spelling tables and reload them. | |
| 307 | 4753 * Used after 'encoding' is set and when ":mkspell" was used. |
| 236 | 4754 */ |
| 4755 void | |
| 4756 spell_reload() | |
| 4757 { | |
| 316 | 4758 win_T *wp; |
| 236 | 4759 |
| 358 | 4760 /* Initialize the table for spell_iswordp(). */ |
| 236 | 4761 init_spell_chartab(); |
| 4762 | |
| 4763 /* Unload all allocated memory. */ | |
| 355 | 4764 spell_free_all(); |
| 236 | 4765 |
| 4766 /* 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
|
4767 for (wp = firstwin; wp != NULL; wp = wp->w_next) |
| 236 | 4768 { |
| 316 | 4769 /* Only load the wordlists when 'spelllang' is set and there is a |
| 4770 * 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
|
4771 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
|
4772 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4773 if (wp->w_p_spell) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4774 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4775 (void)did_set_spelllang(wp); |
| 316 | 4776 # ifdef FEAT_WINDOWS |
| 4777 break; | |
| 4778 # endif | |
| 4779 } | |
| 4780 } | |
| 236 | 4781 } |
| 4782 } | |
| 5519 | 4783 #endif |
| 236 | 4784 |
| 310 | 4785 /* |
| 4786 * Reload the spell file "fname" if it's loaded. | |
| 4787 */ | |
| 4788 static void | |
| 323 | 4789 spell_reload_one(fname, added_word) |
| 310 | 4790 char_u *fname; |
| 323 | 4791 int added_word; /* invoked through "zg" */ |
| 310 | 4792 { |
| 503 | 4793 slang_T *slang; |
| 323 | 4794 int didit = FALSE; |
| 310 | 4795 |
| 503 | 4796 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 4797 { | |
| 4798 if (fullpathcmp(fname, slang->sl_fname, FALSE) == FPC_SAME) | |
| 4799 { | |
| 4800 slang_clear(slang); | |
| 4801 if (spell_load_file(fname, NULL, slang, FALSE) == NULL) | |
| 500 | 4802 /* reloading failed, clear the language */ |
| 503 | 4803 slang_clear(slang); |
| 737 | 4804 redraw_all_later(SOME_VALID); |
| 323 | 4805 didit = TRUE; |
| 310 | 4806 } |
| 500 | 4807 } |
| 323 | 4808 |
| 4809 /* When "zg" was used and the file wasn't loaded yet, should redo | |
| 626 | 4810 * 'spelllang' to load it now. */ |
| 323 | 4811 if (added_word && !didit) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4812 did_set_spelllang(curwin); |
| 310 | 4813 } |
| 4814 | |
| 4815 | |
| 236 | 4816 /* |
| 4817 * Functions for ":mkspell". | |
| 4818 */ | |
| 4819 | |
| 300 | 4820 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff |
| 236 | 4821 and .dic file. */ |
| 4822 /* | |
| 4823 * Main structure to store the contents of a ".aff" file. | |
| 4824 */ | |
| 4825 typedef struct afffile_S | |
| 4826 { | |
| 4827 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */ | |
| 498 | 4828 int af_flagtype; /* AFT_CHAR, AFT_LONG, AFT_NUM or AFT_CAPLONG */ |
| 621 | 4829 unsigned af_rare; /* RARE ID for rare word */ |
| 4830 unsigned af_keepcase; /* KEEPCASE ID for keep-case word */ | |
| 495 | 4831 unsigned af_bad; /* BAD ID for banned word */ |
| 4832 unsigned af_needaffix; /* NEEDAFFIX ID */ | |
| 820 | 4833 unsigned af_circumfix; /* CIRCUMFIX ID */ |
| 500 | 4834 unsigned af_needcomp; /* NEEDCOMPOUND ID */ |
| 809 | 4835 unsigned af_comproot; /* COMPOUNDROOT ID */ |
| 4836 unsigned af_compforbid; /* COMPOUNDFORBIDFLAG ID */ | |
| 4837 unsigned af_comppermit; /* COMPOUNDPERMITFLAG ID */ | |
| 698 | 4838 unsigned af_nosuggest; /* NOSUGGEST ID */ |
| 809 | 4839 int af_pfxpostpone; /* postpone prefixes without chop string and |
| 4840 without flags */ | |
| 6224 | 4841 int af_ignoreextra; /* IGNOREEXTRA present */ |
| 236 | 4842 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */ |
| 4843 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */ | |
| 495 | 4844 hashtab_T af_comp; /* hashtable for compound flags, compitem_T */ |
| 236 | 4845 } afffile_T; |
| 4846 | |
| 495 | 4847 #define AFT_CHAR 0 /* flags are one character */ |
| 498 | 4848 #define AFT_LONG 1 /* flags are two characters */ |
| 4849 #define AFT_CAPLONG 2 /* flags are one or two characters */ | |
| 4850 #define AFT_NUM 3 /* flags are numbers, comma separated */ | |
| 495 | 4851 |
| 236 | 4852 typedef struct affentry_S affentry_T; |
| 4853 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */ | |
| 4854 struct affentry_S | |
| 4855 { | |
| 4856 affentry_T *ae_next; /* next affix with same name/number */ | |
| 4857 char_u *ae_chop; /* text to chop off basic word (can be NULL) */ | |
| 4858 char_u *ae_add; /* text to add to basic word (can be NULL) */ | |
| 809 | 4859 char_u *ae_flags; /* flags on the affix (can be NULL) */ |
| 236 | 4860 char_u *ae_cond; /* condition (NULL for ".") */ |
| 4861 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */ | |
| 822 | 4862 char ae_compforbid; /* COMPOUNDFORBIDFLAG found */ |
| 4863 char ae_comppermit; /* COMPOUNDPERMITFLAG found */ | |
| 300 | 4864 }; |
| 4865 | |
| 495 | 4866 #ifdef FEAT_MBYTE |
| 4867 # define AH_KEY_LEN 17 /* 2 x 8 bytes + NUL */ | |
| 4868 #else | |
| 498 | 4869 # define AH_KEY_LEN 7 /* 6 digits + NUL */ |
| 495 | 4870 #endif |
| 455 | 4871 |
| 300 | 4872 /* Affix header from ".aff" file. Used for af_pref and af_suff. */ |
| 4873 typedef struct affheader_S | |
| 4874 { | |
| 495 | 4875 char_u ah_key[AH_KEY_LEN]; /* key for hashtab == name of affix */ |
| 4876 unsigned ah_flag; /* affix name as number, uses "af_flagtype" */ | |
| 4877 int ah_newID; /* prefix ID after renumbering; 0 if not used */ | |
| 300 | 4878 int ah_combine; /* suffix may combine with prefix */ |
| 498 | 4879 int ah_follows; /* another affix block should be following */ |
| 300 | 4880 affentry_T *ah_first; /* first affix entry */ |
| 4881 } affheader_T; | |
| 4882 | |
| 4883 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key) | |
| 4884 | |
| 495 | 4885 /* Flag used in compound items. */ |
| 4886 typedef struct compitem_S | |
| 4887 { | |
| 4888 char_u ci_key[AH_KEY_LEN]; /* key for hashtab == name of compound */ | |
| 4889 unsigned ci_flag; /* affix name as number, uses "af_flagtype" */ | |
| 4890 int ci_newID; /* affix ID after renumbering. */ | |
| 4891 } compitem_T; | |
| 4892 | |
| 4893 #define HI2CI(hi) ((compitem_T *)(hi)->hi_key) | |
| 4894 | |
| 300 | 4895 /* |
| 4896 * Structure that is used to store the items in the word tree. This avoids | |
| 474 | 4897 * the need to keep track of each allocated thing, everything is freed all at |
| 4898 * once after ":mkspell" is done. | |
| 1808 | 4899 * Note: "sb_next" must be just before "sb_data" to make sure the alignment of |
| 4900 * "sb_data" is correct for systems where pointers must be aligned on | |
| 4901 * pointer-size boundaries and sizeof(pointer) > sizeof(int) (e.g., Sparc). | |
| 300 | 4902 */ |
| 4903 #define SBLOCKSIZE 16000 /* size of sb_data */ | |
| 4904 typedef struct sblock_S sblock_T; | |
| 4905 struct sblock_S | |
| 4906 { | |
| 1808 | 4907 int sb_used; /* nr of bytes already in use */ |
| 300 | 4908 sblock_T *sb_next; /* next block in list */ |
| 4909 char_u sb_data[1]; /* data, actually longer */ | |
| 236 | 4910 }; |
| 4911 | |
| 4912 /* | |
| 300 | 4913 * A node in the tree. |
| 236 | 4914 */ |
| 300 | 4915 typedef struct wordnode_S wordnode_T; |
| 4916 struct wordnode_S | |
| 236 | 4917 { |
| 346 | 4918 union /* shared to save space */ |
| 4919 { | |
| 469 | 4920 char_u hashkey[6]; /* the hash key, only used while compressing */ |
| 346 | 4921 int index; /* index in written nodes (valid after first |
| 4922 round) */ | |
| 4923 } wn_u1; | |
| 4924 union /* shared to save space */ | |
| 4925 { | |
| 4926 wordnode_T *next; /* next node with same hash key */ | |
| 4927 wordnode_T *wnode; /* parent node that will write this node */ | |
| 4928 } wn_u2; | |
| 300 | 4929 wordnode_T *wn_child; /* child (next byte in word) */ |
| 4930 wordnode_T *wn_sibling; /* next sibling (alternate byte in word, | |
| 4931 always sorted) */ | |
| 474 | 4932 int wn_refs; /* Nr. of references to this node. Only |
| 4933 relevant for first node in a list of | |
| 4934 siblings, in following siblings it is | |
| 4935 always one. */ | |
| 300 | 4936 char_u wn_byte; /* Byte for this node. NUL for word end */ |
| 625 | 4937 |
| 4938 /* Info for when "wn_byte" is NUL. | |
| 4939 * In PREFIXTREE "wn_region" is used for the prefcondnr. | |
| 4940 * In the soundfolded word tree "wn_flags" has the MSW of the wordnr and | |
| 4941 * "wn_region" the LSW of the wordnr. */ | |
| 4942 char_u wn_affixID; /* supported/required prefix ID or 0 */ | |
| 4943 short_u wn_flags; /* WF_ flags */ | |
| 4944 short wn_region; /* region mask */ | |
| 4945 | |
| 469 | 4946 #ifdef SPELL_PRINTTREE |
| 4947 int wn_nr; /* sequence nr for printing */ | |
| 4948 #endif | |
| 236 | 4949 }; |
| 4950 | |
| 390 | 4951 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */ |
| 4952 | |
| 300 | 4953 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key) |
| 236 | 4954 |
| 300 | 4955 /* |
| 4956 * Info used while reading the spell files. | |
| 4957 */ | |
| 4958 typedef struct spellinfo_S | |
| 249 | 4959 { |
| 300 | 4960 wordnode_T *si_foldroot; /* tree with case-folded words */ |
| 334 | 4961 long si_foldwcount; /* nr of words in si_foldroot */ |
| 474 | 4962 |
| 300 | 4963 wordnode_T *si_keeproot; /* tree with keep-case words */ |
| 334 | 4964 long si_keepwcount; /* nr of words in si_keeproot */ |
| 474 | 4965 |
| 339 | 4966 wordnode_T *si_prefroot; /* tree with postponed prefixes */ |
| 474 | 4967 |
| 625 | 4968 long si_sugtree; /* creating the soundfolding trie */ |
| 4969 | |
| 300 | 4970 sblock_T *si_blocks; /* memory blocks used */ |
| 482 | 4971 long si_blocks_cnt; /* memory blocks allocated */ |
| 2691 | 4972 int si_did_emsg; /* TRUE when ran out of memory */ |
| 4973 | |
| 482 | 4974 long si_compress_cnt; /* words to add before lowering |
| 4975 compression limit */ | |
| 474 | 4976 wordnode_T *si_first_free; /* List of nodes that have been freed during |
| 4977 compression, linked by "wn_child" field. */ | |
| 482 | 4978 long si_free_count; /* number of nodes in si_first_free */ |
| 474 | 4979 #ifdef SPELL_PRINTTREE |
| 4980 int si_wordnode_nr; /* sequence nr for nodes */ | |
| 4981 #endif | |
| 625 | 4982 buf_T *si_spellbuf; /* buffer used to store soundfold word table */ |
| 474 | 4983 |
| 300 | 4984 int si_ascii; /* handling only ASCII words */ |
| 310 | 4985 int si_add; /* addition file */ |
| 351 | 4986 int si_clear_chartab; /* when TRUE clear char tables */ |
| 300 | 4987 int si_region; /* region mask */ |
| 4988 vimconv_T si_conv; /* for conversion to 'encoding' */ | |
| 302 | 4989 int si_memtot; /* runtime memory used */ |
| 310 | 4990 int si_verbose; /* verbose messages */ |
| 474 | 4991 int si_msg_count; /* number of words added since last message */ |
| 714 | 4992 char_u *si_info; /* info text chars or NULL */ |
| 316 | 4993 int si_region_count; /* number of regions supported (1 when there |
| 4994 are no regions) */ | |
| 2593 | 4995 char_u si_region_name[17]; /* region names; used only if |
| 483 | 4996 * si_region_count > 1) */ |
| 323 | 4997 |
| 4998 garray_T si_rep; /* list of fromto_T entries from REP lines */ | |
| 625 | 4999 garray_T si_repsal; /* list of fromto_T entries from REPSAL lines */ |
| 323 | 5000 garray_T si_sal; /* list of fromto_T entries from SAL lines */ |
| 375 | 5001 char_u *si_sofofr; /* SOFOFROM text */ |
| 5002 char_u *si_sofoto; /* SOFOTO text */ | |
| 625 | 5003 int si_nosugfile; /* NOSUGFILE item found */ |
| 698 | 5004 int si_nosplitsugs; /* NOSPLITSUGS item found */ |
| 323 | 5005 int si_followup; /* soundsalike: ? */ |
| 5006 int si_collapse; /* soundsalike: ? */ | |
| 625 | 5007 hashtab_T si_commonwords; /* hashtable for common words */ |
| 5008 time_t si_sugtime; /* timestamp for .sug file */ | |
| 323 | 5009 int si_rem_accents; /* soundsalike: remove accents */ |
| 5010 garray_T si_map; /* MAP info concatenated */ | |
| 495 | 5011 char_u *si_midword; /* MIDWORD chars or NULL */ |
| 483 | 5012 int si_compmax; /* max nr of words for compounding */ |
| 481 | 5013 int si_compminlen; /* minimal length for compounding */ |
| 483 | 5014 int si_compsylmax; /* max nr of syllables for compounding */ |
| 809 | 5015 int si_compoptions; /* COMP_ flags */ |
| 5016 garray_T si_comppat; /* CHECKCOMPOUNDPATTERN items, each stored as | |
| 5017 a string */ | |
| 481 | 5018 char_u *si_compflags; /* flags used for compounding */ |
| 492 | 5019 char_u si_nobreak; /* NOBREAK */ |
| 483 | 5020 char_u *si_syllable; /* syllable string */ |
| 339 | 5021 garray_T si_prefcond; /* table with conditions for postponed |
| 5022 * prefixes, each stored as a string */ | |
| 495 | 5023 int si_newprefID; /* current value for ah_newID */ |
| 500 | 5024 int si_newcompID; /* current value for compound ID */ |
| 300 | 5025 } spellinfo_T; |
| 249 | 5026 |
| 474 | 5027 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname)); |
| 1762 | 5028 static int is_aff_rule __ARGS((char_u **items, int itemcnt, char *rulename, int mincount)); |
| 822 | 5029 static void aff_process_flags __ARGS((afffile_T *affile, affentry_T *entry)); |
| 714 | 5030 static int spell_info_item __ARGS((char_u *s)); |
| 495 | 5031 static unsigned affitem2flag __ARGS((int flagtype, char_u *item, char_u *fname, int lnum)); |
| 5032 static unsigned get_affitem __ARGS((int flagtype, char_u **pp)); | |
| 5033 static void process_compflags __ARGS((spellinfo_T *spin, afffile_T *aff, char_u *compflags)); | |
| 500 | 5034 static void check_renumber __ARGS((spellinfo_T *spin)); |
| 495 | 5035 static int flag_in_afflist __ARGS((int flagtype, char_u *afflist, unsigned flag)); |
| 5036 static void aff_check_number __ARGS((int spinval, int affval, char *name)); | |
| 5037 static void aff_check_string __ARGS((char_u *spinval, char_u *affval, char *name)); | |
| 339 | 5038 static int str_equal __ARGS((char_u *s1, char_u *s2)); |
| 323 | 5039 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to)); |
| 5040 static int sal_to_bool __ARGS((char_u *s)); | |
| 300 | 5041 static void spell_free_aff __ARGS((afffile_T *aff)); |
| 474 | 5042 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile)); |
| 820 | 5043 static int get_affix_flags __ARGS((afffile_T *affile, char_u *afflist)); |
| 483 | 5044 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist)); |
| 495 | 5045 static void get_compflags __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist)); |
| 820 | 5046 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 | 5047 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname)); |
| 5048 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align)); | |
| 5049 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s)); | |
| 300 | 5050 static void free_blocks __ARGS((sblock_T *bl)); |
| 474 | 5051 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin)); |
| 483 | 5052 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix)); |
| 481 | 5053 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID)); |
| 474 | 5054 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin)); |
| 625 | 5055 static int deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node)); |
| 474 | 5056 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n)); |
| 5057 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root)); | |
| 5058 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot)); | |
| 300 | 5059 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2)); |
| 500 | 5060 static int write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname)); |
| 346 | 5061 static void clear_node __ARGS((wordnode_T *node)); |
| 1757 | 5062 static int put_node __ARGS((FILE *fd, wordnode_T *node, int idx, int regionmask, int prefixtree)); |
| 625 | 5063 static void spell_make_sugfile __ARGS((spellinfo_T *spin, char_u *wfname)); |
| 5064 static int sug_filltree __ARGS((spellinfo_T *spin, slang_T *slang)); | |
| 5065 static int sug_maketable __ARGS((spellinfo_T *spin)); | |
| 5066 static int sug_filltable __ARGS((spellinfo_T *spin, wordnode_T *node, int startwordnr, garray_T *gap)); | |
| 5067 static int offset2bytes __ARGS((int nr, char_u *buf)); | |
| 5068 static int bytes2offset __ARGS((char_u **pp)); | |
| 5069 static void sug_write __ARGS((spellinfo_T *spin, char_u *fname)); | |
| 3263 | 5070 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int over_write, int added_word)); |
| 625 | 5071 static void spell_message __ARGS((spellinfo_T *spin, char_u *str)); |
| 310 | 5072 static void init_spellfile __ARGS((void)); |
| 236 | 5073 |
| 455 | 5074 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags, |
| 5075 * but it must be negative to indicate the prefix tree to tree_add_word(). | |
| 5076 * Use a negative number with the lower 8 bits zero. */ | |
| 5077 #define PFX_FLAGS -256 | |
| 390 | 5078 |
| 820 | 5079 /* flags for "condit" argument of store_aff_word() */ |
| 5080 #define CONDIT_COMB 1 /* affix must combine */ | |
| 5081 #define CONDIT_CFIX 2 /* affix must have CIRCUMFIX flag */ | |
| 5082 #define CONDIT_SUF 4 /* add a suffix for matching flags */ | |
| 5083 #define CONDIT_AFF 8 /* word already has an affix */ | |
| 5084 | |
| 483 | 5085 /* |
| 5086 * Tunable parameters for when the tree is compressed. See 'mkspellmem'. | |
| 5087 */ | |
| 5088 static long compress_start = 30000; /* memory / SBLOCKSIZE */ | |
| 5089 static long compress_inc = 100; /* memory / SBLOCKSIZE */ | |
| 5090 static long compress_added = 500000; /* word count */ | |
| 5091 | |
| 469 | 5092 #ifdef SPELL_PRINTTREE |
| 5093 /* | |
| 5094 * For debugging the tree code: print the current tree in a (more or less) | |
| 5095 * readable format, so that we can see what happens when adding a word and/or | |
| 5096 * compressing the tree. | |
| 5097 * Based on code from Olaf Seibert. | |
| 5098 */ | |
| 5099 #define PRINTLINESIZE 1000 | |
| 5100 #define PRINTWIDTH 6 | |
| 5101 | |
| 5102 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \ | |
| 5103 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2) | |
| 5104 | |
| 5105 static char line1[PRINTLINESIZE]; | |
| 5106 static char line2[PRINTLINESIZE]; | |
| 5107 static char line3[PRINTLINESIZE]; | |
| 5108 | |
| 5109 static void | |
| 5110 spell_clear_flags(wordnode_T *node) | |
| 5111 { | |
| 5112 wordnode_T *np; | |
| 5113 | |
| 5114 for (np = node; np != NULL; np = np->wn_sibling) | |
| 5115 { | |
| 5116 np->wn_u1.index = FALSE; | |
| 5117 spell_clear_flags(np->wn_child); | |
| 5118 } | |
| 5119 } | |
| 5120 | |
| 5121 static void | |
| 5122 spell_print_node(wordnode_T *node, int depth) | |
| 5123 { | |
| 5124 if (node->wn_u1.index) | |
| 5125 { | |
| 5126 /* Done this node before, print the reference. */ | |
| 5127 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0); | |
| 5128 PRINTSOME(line2, depth, " ", 0, 0); | |
| 5129 PRINTSOME(line3, depth, " ", 0, 0); | |
| 5130 msg(line1); | |
| 5131 msg(line2); | |
| 5132 msg(line3); | |
| 5133 } | |
| 5134 else | |
| 5135 { | |
| 5136 node->wn_u1.index = TRUE; | |
| 5137 | |
| 5138 if (node->wn_byte != NUL) | |
| 5139 { | |
| 5140 if (node->wn_child != NULL) | |
| 5141 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0); | |
| 5142 else | |
| 5143 /* Cannot happen? */ | |
| 5144 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0); | |
| 5145 } | |
| 5146 else | |
| 5147 PRINTSOME(line1, depth, " $ ", 0, 0); | |
| 5148 | |
| 5149 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs); | |
| 5150 | |
| 5151 if (node->wn_sibling != NULL) | |
| 5152 PRINTSOME(line3, depth, " | ", 0, 0); | |
| 5153 else | |
| 5154 PRINTSOME(line3, depth, " ", 0, 0); | |
| 5155 | |
| 5156 if (node->wn_byte == NUL) | |
| 5157 { | |
| 5158 msg(line1); | |
| 5159 msg(line2); | |
| 5160 msg(line3); | |
| 5161 } | |
| 5162 | |
| 5163 /* do the children */ | |
| 5164 if (node->wn_byte != NUL && node->wn_child != NULL) | |
| 5165 spell_print_node(node->wn_child, depth + 1); | |
| 5166 | |
| 5167 /* do the siblings */ | |
| 5168 if (node->wn_sibling != NULL) | |
| 5169 { | |
| 5170 /* get rid of all parent details except | */ | |
| 5171 STRCPY(line1, line3); | |
| 5172 STRCPY(line2, line3); | |
| 5173 spell_print_node(node->wn_sibling, depth); | |
| 5174 } | |
| 5175 } | |
| 5176 } | |
| 5177 | |
| 5178 static void | |
| 5179 spell_print_tree(wordnode_T *root) | |
| 5180 { | |
| 5181 if (root != NULL) | |
| 5182 { | |
| 5183 /* Clear the "wn_u1.index" fields, used to remember what has been | |
| 5184 * done. */ | |
| 5185 spell_clear_flags(root); | |
| 5186 | |
| 5187 /* Recursively print the tree. */ | |
| 5188 spell_print_node(root, 0); | |
| 5189 } | |
| 5190 } | |
| 5191 #endif /* SPELL_PRINTTREE */ | |
| 5192 | |
| 236 | 5193 /* |
| 323 | 5194 * Read the affix file "fname". |
| 316 | 5195 * Returns an afffile_T, NULL for complete failure. |
| 236 | 5196 */ |
| 5197 static afffile_T * | |
| 474 | 5198 spell_read_aff(spin, fname) |
| 5199 spellinfo_T *spin; | |
| 236 | 5200 char_u *fname; |
| 5201 { | |
| 5202 FILE *fd; | |
| 5203 afffile_T *aff; | |
| 5204 char_u rline[MAXLINELEN]; | |
| 5205 char_u *line; | |
| 5206 char_u *pc = NULL; | |
| 625 | 5207 #define MAXITEMCNT 30 |
| 334 | 5208 char_u *(items[MAXITEMCNT]); |
| 236 | 5209 int itemcnt; |
| 5210 char_u *p; | |
| 5211 int lnum = 0; | |
| 5212 affheader_T *cur_aff = NULL; | |
| 495 | 5213 int did_postpone_prefix = FALSE; |
| 236 | 5214 int aff_todo = 0; |
| 5215 hashtab_T *tp; | |
| 255 | 5216 char_u *low = NULL; |
| 5217 char_u *fol = NULL; | |
| 5218 char_u *upp = NULL; | |
| 323 | 5219 int do_rep; |
| 625 | 5220 int do_repsal; |
| 323 | 5221 int do_sal; |
| 944 | 5222 int do_mapline; |
| 323 | 5223 int found_map = FALSE; |
| 324 | 5224 hashitem_T *hi; |
| 455 | 5225 int l; |
| 495 | 5226 int compminlen = 0; /* COMPOUNDMIN value */ |
| 5227 int compsylmax = 0; /* COMPOUNDSYLMAX value */ | |
| 809 | 5228 int compoptions = 0; /* COMP_ flags */ |
| 5229 int compmax = 0; /* COMPOUNDWORDMAX value */ | |
| 714 | 5230 char_u *compflags = NULL; /* COMPOUNDFLAG and COMPOUNDRULE |
| 495 | 5231 concatenated */ |
| 5232 char_u *midword = NULL; /* MIDWORD value */ | |
| 5233 char_u *syllable = NULL; /* SYLLABLE value */ | |
| 5234 char_u *sofofrom = NULL; /* SOFOFROM value */ | |
| 5235 char_u *sofoto = NULL; /* SOFOTO value */ | |
| 236 | 5236 |
| 300 | 5237 /* |
| 5238 * Open the file. | |
| 5239 */ | |
| 310 | 5240 fd = mch_fopen((char *)fname, "r"); |
| 236 | 5241 if (fd == NULL) |
| 5242 { | |
| 5243 EMSG2(_(e_notopen), fname); | |
| 5244 return NULL; | |
| 5245 } | |
| 5246 | |
| 625 | 5247 vim_snprintf((char *)IObuff, IOSIZE, _("Reading affix file %s ..."), fname); |
| 5248 spell_message(spin, IObuff); | |
| 236 | 5249 |
| 323 | 5250 /* Only do REP lines when not done in another .aff file already. */ |
| 5251 do_rep = spin->si_rep.ga_len == 0; | |
| 5252 | |
| 625 | 5253 /* Only do REPSAL lines when not done in another .aff file already. */ |
| 5254 do_repsal = spin->si_repsal.ga_len == 0; | |
| 5255 | |
| 323 | 5256 /* Only do SAL lines when not done in another .aff file already. */ |
| 5257 do_sal = spin->si_sal.ga_len == 0; | |
| 5258 | |
| 5259 /* Only do MAP lines when not done in another .aff file already. */ | |
| 944 | 5260 do_mapline = spin->si_map.ga_len == 0; |
| 323 | 5261 |
| 300 | 5262 /* |
| 5263 * Allocate and init the afffile_T structure. | |
| 5264 */ | |
| 474 | 5265 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE); |
| 236 | 5266 if (aff == NULL) |
| 840 | 5267 { |
| 5268 fclose(fd); | |
| 236 | 5269 return NULL; |
| 840 | 5270 } |
| 236 | 5271 hash_init(&aff->af_pref); |
| 5272 hash_init(&aff->af_suff); | |
| 495 | 5273 hash_init(&aff->af_comp); |
| 236 | 5274 |
| 5275 /* | |
| 5276 * Read all the lines in the file one by one. | |
| 5277 */ | |
| 255 | 5278 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int) |
| 236 | 5279 { |
| 255 | 5280 line_breakcheck(); |
| 236 | 5281 ++lnum; |
| 5282 | |
| 5283 /* Skip comment lines. */ | |
| 5284 if (*rline == '#') | |
| 5285 continue; | |
| 5286 | |
| 5287 /* Convert from "SET" to 'encoding' when needed. */ | |
| 5288 vim_free(pc); | |
| 310 | 5289 #ifdef FEAT_MBYTE |
| 300 | 5290 if (spin->si_conv.vc_type != CONV_NONE) |
| 236 | 5291 { |
| 300 | 5292 pc = string_convert(&spin->si_conv, rline, NULL); |
| 255 | 5293 if (pc == NULL) |
| 5294 { | |
| 5295 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"), | |
| 5296 fname, lnum, rline); | |
| 5297 continue; | |
| 5298 } | |
| 236 | 5299 line = pc; |
| 5300 } | |
| 5301 else | |
| 310 | 5302 #endif |
| 236 | 5303 { |
| 5304 pc = NULL; | |
| 5305 line = rline; | |
| 5306 } | |
| 5307 | |
| 5308 /* Split the line up in white separated items. Put a NUL after each | |
| 5309 * item. */ | |
| 5310 itemcnt = 0; | |
| 5311 for (p = line; ; ) | |
| 5312 { | |
| 5313 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */ | |
| 5314 ++p; | |
| 5315 if (*p == NUL) | |
| 5316 break; | |
| 334 | 5317 if (itemcnt == MAXITEMCNT) /* too many items */ |
| 300 | 5318 break; |
| 236 | 5319 items[itemcnt++] = p; |
| 714 | 5320 /* A few items have arbitrary text argument, don't split them. */ |
| 5321 if (itemcnt == 2 && spell_info_item(items[0])) | |
| 5322 while (*p >= ' ' || *p == TAB) /* skip until CR/NL */ | |
| 5323 ++p; | |
| 5324 else | |
| 5325 while (*p > ' ') /* skip until white space or CR/NL */ | |
| 5326 ++p; | |
| 236 | 5327 if (*p == NUL) |
| 5328 break; | |
| 5329 *p++ = NUL; | |
| 5330 } | |
| 5331 | |
| 5332 /* Handle non-empty lines. */ | |
| 5333 if (itemcnt > 0) | |
| 5334 { | |
| 1762 | 5335 if (is_aff_rule(items, itemcnt, "SET", 2) && aff->af_enc == NULL) |
| 236 | 5336 { |
| 310 | 5337 #ifdef FEAT_MBYTE |
| 300 | 5338 /* Setup for conversion from "ENC" to 'encoding'. */ |
| 5339 aff->af_enc = enc_canonize(items[1]); | |
| 5340 if (aff->af_enc != NULL && !spin->si_ascii | |
| 5341 && convert_setup(&spin->si_conv, aff->af_enc, | |
| 5342 p_enc) == FAIL) | |
| 5343 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"), | |
| 5344 fname, aff->af_enc, p_enc); | |
| 375 | 5345 spin->si_conv.vc_fail = TRUE; |
| 310 | 5346 #else |
| 5347 smsg((char_u *)_("Conversion in %s not supported"), fname); | |
| 5348 #endif | |
| 236 | 5349 } |
| 1762 | 5350 else if (is_aff_rule(items, itemcnt, "FLAG", 2) |
| 495 | 5351 && aff->af_flagtype == AFT_CHAR) |
| 5352 { | |
| 5353 if (STRCMP(items[1], "long") == 0) | |
| 498 | 5354 aff->af_flagtype = AFT_LONG; |
| 495 | 5355 else if (STRCMP(items[1], "num") == 0) |
| 498 | 5356 aff->af_flagtype = AFT_NUM; |
| 5357 else if (STRCMP(items[1], "caplong") == 0) | |
| 5358 aff->af_flagtype = AFT_CAPLONG; | |
| 495 | 5359 else |
| 5360 smsg((char_u *)_("Invalid value for FLAG in %s line %d: %s"), | |
| 5361 fname, lnum, items[1]); | |
| 621 | 5362 if (aff->af_rare != 0 |
| 5363 || aff->af_keepcase != 0 | |
| 5364 || aff->af_bad != 0 | |
| 500 | 5365 || aff->af_needaffix != 0 |
| 820 | 5366 || aff->af_circumfix != 0 |
| 500 | 5367 || aff->af_needcomp != 0 |
| 809 | 5368 || aff->af_comproot != 0 |
| 698 | 5369 || aff->af_nosuggest != 0 |
| 500 | 5370 || compflags != NULL |
| 495 | 5371 || aff->af_suff.ht_used > 0 |
| 5372 || aff->af_pref.ht_used > 0) | |
| 5373 smsg((char_u *)_("FLAG after using flags in %s line %d: %s"), | |
| 5374 fname, lnum, items[1]); | |
| 5375 } | |
| 714 | 5376 else if (spell_info_item(items[0])) |
| 5377 { | |
| 5378 p = (char_u *)getroom(spin, | |
| 5379 (spin->si_info == NULL ? 0 : STRLEN(spin->si_info)) | |
| 5380 + STRLEN(items[0]) | |
| 5381 + STRLEN(items[1]) + 3, FALSE); | |
| 5382 if (p != NULL) | |
| 5383 { | |
| 5384 if (spin->si_info != NULL) | |
| 5385 { | |
| 5386 STRCPY(p, spin->si_info); | |
| 5387 STRCAT(p, "\n"); | |
| 5388 } | |
| 5389 STRCAT(p, items[0]); | |
| 5390 STRCAT(p, " "); | |
| 5391 STRCAT(p, items[1]); | |
| 5392 spin->si_info = p; | |
| 5393 } | |
| 5394 } | |
| 1762 | 5395 else if (is_aff_rule(items, itemcnt, "MIDWORD", 2) |
| 495 | 5396 && midword == NULL) |
| 5397 { | |
| 5398 midword = getroom_save(spin, items[1]); | |
| 366 | 5399 } |
| 1762 | 5400 else if (is_aff_rule(items, itemcnt, "TRY", 2)) |
| 300 | 5401 { |
| 323 | 5402 /* ignored, we look in the tree for what chars may appear */ |
| 300 | 5403 } |
| 621 | 5404 /* TODO: remove "RAR" later */ |
| 1762 | 5405 else if ((is_aff_rule(items, itemcnt, "RAR", 2) |
| 5406 || is_aff_rule(items, itemcnt, "RARE", 2)) | |
| 5407 && aff->af_rare == 0) | |
| 621 | 5408 { |
| 5409 aff->af_rare = affitem2flag(aff->af_flagtype, items[1], | |
| 495 | 5410 fname, lnum); |
| 307 | 5411 } |
| 621 | 5412 /* TODO: remove "KEP" later */ |
| 1762 | 5413 else if ((is_aff_rule(items, itemcnt, "KEP", 2) |
| 5414 || is_aff_rule(items, itemcnt, "KEEPCASE", 2)) | |
| 621 | 5415 && aff->af_keepcase == 0) |
| 5416 { | |
| 5417 aff->af_keepcase = affitem2flag(aff->af_flagtype, items[1], | |
| 495 | 5418 fname, lnum); |
| 307 | 5419 } |
| 1762 | 5420 else if ((is_aff_rule(items, itemcnt, "BAD", 2) |
| 5421 || is_aff_rule(items, itemcnt, "FORBIDDENWORD", 2)) | |
| 5422 && aff->af_bad == 0) | |
| 346 | 5423 { |
| 495 | 5424 aff->af_bad = affitem2flag(aff->af_flagtype, items[1], |
| 5425 fname, lnum); | |
| 346 | 5426 } |
| 1762 | 5427 else if (is_aff_rule(items, itemcnt, "NEEDAFFIX", 2) |
| 483 | 5428 && aff->af_needaffix == 0) |
| 5429 { | |
| 495 | 5430 aff->af_needaffix = affitem2flag(aff->af_flagtype, items[1], |
| 5431 fname, lnum); | |
| 483 | 5432 } |
| 1762 | 5433 else if (is_aff_rule(items, itemcnt, "CIRCUMFIX", 2) |
| 820 | 5434 && aff->af_circumfix == 0) |
| 5435 { | |
| 5436 aff->af_circumfix = affitem2flag(aff->af_flagtype, items[1], | |
| 5437 fname, lnum); | |
| 5438 } | |
| 1762 | 5439 else if (is_aff_rule(items, itemcnt, "NOSUGGEST", 2) |
| 698 | 5440 && aff->af_nosuggest == 0) |
| 5441 { | |
| 5442 aff->af_nosuggest = affitem2flag(aff->af_flagtype, items[1], | |
| 5443 fname, lnum); | |
| 5444 } | |
| 1762 | 5445 else if ((is_aff_rule(items, itemcnt, "NEEDCOMPOUND", 2) |
| 5446 || is_aff_rule(items, itemcnt, "ONLYINCOMPOUND", 2)) | |
| 500 | 5447 && aff->af_needcomp == 0) |
| 5448 { | |
| 5449 aff->af_needcomp = affitem2flag(aff->af_flagtype, items[1], | |
| 5450 fname, lnum); | |
| 5451 } | |
| 1762 | 5452 else if (is_aff_rule(items, itemcnt, "COMPOUNDROOT", 2) |
| 809 | 5453 && aff->af_comproot == 0) |
| 5454 { | |
| 5455 aff->af_comproot = affitem2flag(aff->af_flagtype, items[1], | |
| 5456 fname, lnum); | |
| 5457 } | |
| 1762 | 5458 else if (is_aff_rule(items, itemcnt, "COMPOUNDFORBIDFLAG", 2) |
| 5459 && aff->af_compforbid == 0) | |
| 809 | 5460 { |
| 5461 aff->af_compforbid = affitem2flag(aff->af_flagtype, items[1], | |
| 5462 fname, lnum); | |
| 822 | 5463 if (aff->af_pref.ht_used > 0) |
| 5464 smsg((char_u *)_("Defining COMPOUNDFORBIDFLAG after PFX item may give wrong results in %s line %d"), | |
| 5465 fname, lnum); | |
| 809 | 5466 } |
| 1762 | 5467 else if (is_aff_rule(items, itemcnt, "COMPOUNDPERMITFLAG", 2) |
| 5468 && aff->af_comppermit == 0) | |
| 809 | 5469 { |
| 5470 aff->af_comppermit = affitem2flag(aff->af_flagtype, items[1], | |
| 5471 fname, lnum); | |
| 822 | 5472 if (aff->af_pref.ht_used > 0) |
| 5473 smsg((char_u *)_("Defining COMPOUNDPERMITFLAG after PFX item may give wrong results in %s line %d"), | |
| 5474 fname, lnum); | |
| 809 | 5475 } |
| 1762 | 5476 else if (is_aff_rule(items, itemcnt, "COMPOUNDFLAG", 2) |
| 495 | 5477 && compflags == NULL) |
| 5478 { | |
| 714 | 5479 /* Turn flag "c" into COMPOUNDRULE compatible string "c+", |
| 495 | 5480 * "Na" into "Na+", "1234" into "1234+". */ |
| 5481 p = getroom(spin, STRLEN(items[1]) + 2, FALSE); | |
| 483 | 5482 if (p != NULL) |
| 5483 { | |
| 495 | 5484 STRCPY(p, items[1]); |
| 5485 STRCAT(p, "+"); | |
| 5486 compflags = p; | |
| 5487 } | |
| 481 | 5488 } |
| 1762 | 5489 else if (is_aff_rule(items, itemcnt, "COMPOUNDRULES", 2)) |
| 5490 { | |
| 5491 /* We don't use the count, but do check that it's a number and | |
| 5492 * not COMPOUNDRULE mistyped. */ | |
| 5493 if (atoi((char *)items[1]) == 0) | |
| 5494 smsg((char_u *)_("Wrong COMPOUNDRULES value in %s line %d: %s"), | |
| 5495 fname, lnum, items[1]); | |
| 5496 } | |
| 5497 else if (is_aff_rule(items, itemcnt, "COMPOUNDRULE", 2)) | |
| 483 | 5498 { |
| 2691 | 5499 /* Don't use the first rule if it is a number. */ |
| 5500 if (compflags != NULL || *skipdigits(items[1]) != NUL) | |
| 5501 { | |
| 5502 /* Concatenate this string to previously defined ones, | |
| 5503 * using a slash to separate them. */ | |
| 5504 l = (int)STRLEN(items[1]) + 1; | |
| 495 | 5505 if (compflags != NULL) |
| 2691 | 5506 l += (int)STRLEN(compflags) + 1; |
| 5507 p = getroom(spin, l, FALSE); | |
| 5508 if (p != NULL) | |
| 483 | 5509 { |
| 2691 | 5510 if (compflags != NULL) |
| 5511 { | |
| 5512 STRCPY(p, compflags); | |
| 5513 STRCAT(p, "/"); | |
| 5514 } | |
| 5515 STRCAT(p, items[1]); | |
| 5516 compflags = p; | |
| 483 | 5517 } |
| 5518 } | |
| 5519 } | |
| 1762 | 5520 else if (is_aff_rule(items, itemcnt, "COMPOUNDWORDMAX", 2) |
| 495 | 5521 && compmax == 0) |
| 5522 { | |
| 5523 compmax = atoi((char *)items[1]); | |
| 5524 if (compmax == 0) | |
| 809 | 5525 smsg((char_u *)_("Wrong COMPOUNDWORDMAX value in %s line %d: %s"), |
| 483 | 5526 fname, lnum, items[1]); |
| 481 | 5527 } |
| 1762 | 5528 else if (is_aff_rule(items, itemcnt, "COMPOUNDMIN", 2) |
| 495 | 5529 && compminlen == 0) |
| 5530 { | |
| 5531 compminlen = atoi((char *)items[1]); | |
| 5532 if (compminlen == 0) | |
| 481 | 5533 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"), |
| 5534 fname, lnum, items[1]); | |
| 5535 } | |
| 1762 | 5536 else if (is_aff_rule(items, itemcnt, "COMPOUNDSYLMAX", 2) |
| 495 | 5537 && compsylmax == 0) |
| 5538 { | |
| 5539 compsylmax = atoi((char *)items[1]); | |
| 5540 if (compsylmax == 0) | |
| 483 | 5541 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"), |
| 5542 fname, lnum, items[1]); | |
| 5543 } | |
| 1762 | 5544 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDDUP", 1)) |
| 809 | 5545 { |
| 5546 compoptions |= COMP_CHECKDUP; | |
| 5547 } | |
| 1762 | 5548 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDREP", 1)) |
| 809 | 5549 { |
| 5550 compoptions |= COMP_CHECKREP; | |
| 5551 } | |
| 1762 | 5552 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDCASE", 1)) |
| 809 | 5553 { |
| 5554 compoptions |= COMP_CHECKCASE; | |
| 5555 } | |
| 1762 | 5556 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDTRIPLE", 1)) |
| 809 | 5557 { |
| 5558 compoptions |= COMP_CHECKTRIPLE; | |
| 5559 } | |
| 1762 | 5560 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDPATTERN", 2)) |
| 809 | 5561 { |
| 5562 if (atoi((char *)items[1]) == 0) | |
| 5563 smsg((char_u *)_("Wrong CHECKCOMPOUNDPATTERN value in %s line %d: %s"), | |
| 5564 fname, lnum, items[1]); | |
| 5565 } | |
| 1762 | 5566 else if (is_aff_rule(items, itemcnt, "CHECKCOMPOUNDPATTERN", 3)) |
| 809 | 5567 { |
| 5568 garray_T *gap = &spin->si_comppat; | |
| 5569 int i; | |
| 5570 | |
| 5571 /* Only add the couple if it isn't already there. */ | |
| 5572 for (i = 0; i < gap->ga_len - 1; i += 2) | |
| 5573 if (STRCMP(((char_u **)(gap->ga_data))[i], items[1]) == 0 | |
| 5574 && STRCMP(((char_u **)(gap->ga_data))[i + 1], | |
| 5575 items[2]) == 0) | |
| 5576 break; | |
| 5577 if (i >= gap->ga_len && ga_grow(gap, 2) == OK) | |
| 5578 { | |
| 5579 ((char_u **)(gap->ga_data))[gap->ga_len++] | |
| 5580 = getroom_save(spin, items[1]); | |
| 5581 ((char_u **)(gap->ga_data))[gap->ga_len++] | |
| 5582 = getroom_save(spin, items[2]); | |
| 5583 } | |
| 5584 } | |
| 1762 | 5585 else if (is_aff_rule(items, itemcnt, "SYLLABLE", 2) |
| 495 | 5586 && syllable == NULL) |
| 5587 { | |
| 5588 syllable = getroom_save(spin, items[1]); | |
| 483 | 5589 } |
| 1762 | 5590 else if (is_aff_rule(items, itemcnt, "NOBREAK", 1)) |
| 492 | 5591 { |
| 5592 spin->si_nobreak = TRUE; | |
| 5593 } | |
| 1762 | 5594 else if (is_aff_rule(items, itemcnt, "NOSPLITSUGS", 1)) |
| 698 | 5595 { |
| 5596 spin->si_nosplitsugs = TRUE; | |
| 5597 } | |
| 1762 | 5598 else if (is_aff_rule(items, itemcnt, "NOSUGFILE", 1)) |
| 625 | 5599 { |
| 5600 spin->si_nosugfile = TRUE; | |
| 5601 } | |
| 1762 | 5602 else if (is_aff_rule(items, itemcnt, "PFXPOSTPONE", 1)) |
| 339 | 5603 { |
| 5604 aff->af_pfxpostpone = TRUE; | |
| 5605 } | |
| 6224 | 5606 else if (is_aff_rule(items, itemcnt, "IGNOREEXTRA", 1)) |
| 5607 { | |
| 5608 aff->af_ignoreextra = TRUE; | |
| 5609 } | |
| 236 | 5610 else if ((STRCMP(items[0], "PFX") == 0 |
| 5611 || STRCMP(items[0], "SFX") == 0) | |
| 5612 && aff_todo == 0 | |
| 334 | 5613 && itemcnt >= 4) |
| 236 | 5614 { |
| 498 | 5615 int lasti = 4; |
| 5616 char_u key[AH_KEY_LEN]; | |
| 5617 | |
| 5618 if (*items[0] == 'P') | |
| 5619 tp = &aff->af_pref; | |
| 5620 else | |
| 5621 tp = &aff->af_suff; | |
| 5622 | |
| 5623 /* Myspell allows the same affix name to be used multiple | |
| 5624 * times. The affix files that do this have an undocumented | |
| 5625 * "S" flag on all but the last block, thus we check for that | |
| 5626 * and store it in ah_follows. */ | |
| 5627 vim_strncpy(key, items[1], AH_KEY_LEN - 1); | |
| 5628 hi = hash_find(tp, key); | |
| 5629 if (!HASHITEM_EMPTY(hi)) | |
| 5630 { | |
| 5631 cur_aff = HI2AH(hi); | |
| 5632 if (cur_aff->ah_combine != (*items[2] == 'Y')) | |
| 5633 smsg((char_u *)_("Different combining flag in continued affix block in %s line %d: %s"), | |
| 5634 fname, lnum, items[1]); | |
| 5635 if (!cur_aff->ah_follows) | |
| 5636 smsg((char_u *)_("Duplicate affix in %s line %d: %s"), | |
| 5637 fname, lnum, items[1]); | |
| 5638 } | |
| 5639 else | |
| 5640 { | |
| 5641 /* New affix letter. */ | |
| 5642 cur_aff = (affheader_T *)getroom(spin, | |
| 5643 sizeof(affheader_T), TRUE); | |
| 5644 if (cur_aff == NULL) | |
| 5645 break; | |
| 5646 cur_aff->ah_flag = affitem2flag(aff->af_flagtype, items[1], | |
| 5647 fname, lnum); | |
| 5648 if (cur_aff->ah_flag == 0 || STRLEN(items[1]) >= AH_KEY_LEN) | |
| 5649 break; | |
| 5650 if (cur_aff->ah_flag == aff->af_bad | |
| 621 | 5651 || cur_aff->ah_flag == aff->af_rare |
| 5652 || cur_aff->ah_flag == aff->af_keepcase | |
| 500 | 5653 || cur_aff->ah_flag == aff->af_needaffix |
| 820 | 5654 || cur_aff->ah_flag == aff->af_circumfix |
| 698 | 5655 || cur_aff->ah_flag == aff->af_nosuggest |
| 809 | 5656 || cur_aff->ah_flag == aff->af_needcomp |
| 5657 || cur_aff->ah_flag == aff->af_comproot) | |
| 698 | 5658 smsg((char_u *)_("Affix also used for BAD/RARE/KEEPCASE/NEEDAFFIX/NEEDCOMPOUND/NOSUGGEST in %s line %d: %s"), |
| 498 | 5659 fname, lnum, items[1]); |
| 5660 STRCPY(cur_aff->ah_key, items[1]); | |
| 5661 hash_add(tp, cur_aff->ah_key); | |
| 5662 | |
| 5663 cur_aff->ah_combine = (*items[2] == 'Y'); | |
| 5664 } | |
| 5665 | |
| 5666 /* Check for the "S" flag, which apparently means that another | |
| 5667 * block with the same affix name is following. */ | |
| 5668 if (itemcnt > lasti && STRCMP(items[lasti], "S") == 0) | |
| 5669 { | |
| 5670 ++lasti; | |
| 5671 cur_aff->ah_follows = TRUE; | |
| 5672 } | |
| 5673 else | |
| 5674 cur_aff->ah_follows = FALSE; | |
| 5675 | |
| 334 | 5676 /* Myspell allows extra text after the item, but that might |
| 5677 * mean mistakes go unnoticed. Require a comment-starter. */ | |
| 498 | 5678 if (itemcnt > lasti && *items[lasti] != '#') |
| 809 | 5679 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]); |
| 334 | 5680 |
| 498 | 5681 if (STRCMP(items[2], "Y") != 0 && STRCMP(items[2], "N") != 0) |
| 236 | 5682 smsg((char_u *)_("Expected Y or N in %s line %d: %s"), |
| 5683 fname, lnum, items[2]); | |
| 339 | 5684 |
| 498 | 5685 if (*items[0] == 'P' && aff->af_pfxpostpone) |
| 5686 { | |
| 5687 if (cur_aff->ah_newID == 0) | |
| 495 | 5688 { |
| 5689 /* Use a new number in the .spl file later, to be able | |
| 5690 * to handle multiple .aff files. */ | |
| 500 | 5691 check_renumber(spin); |
| 495 | 5692 cur_aff->ah_newID = ++spin->si_newprefID; |
| 5693 | |
| 5694 /* We only really use ah_newID if the prefix is | |
| 5695 * postponed. We know that only after handling all | |
| 5696 * the items. */ | |
| 5697 did_postpone_prefix = FALSE; | |
| 5698 } | |
| 498 | 5699 else |
| 5700 /* Did use the ID in a previous block. */ | |
| 5701 did_postpone_prefix = TRUE; | |
| 5702 } | |
| 5703 | |
| 300 | 5704 aff_todo = atoi((char *)items[3]); |
| 236 | 5705 } |
| 5706 else if ((STRCMP(items[0], "PFX") == 0 | |
| 5707 || STRCMP(items[0], "SFX") == 0) | |
| 5708 && aff_todo > 0 | |
| 5709 && STRCMP(cur_aff->ah_key, items[1]) == 0 | |
| 334 | 5710 && itemcnt >= 5) |
| 236 | 5711 { |
| 5712 affentry_T *aff_entry; | |
| 455 | 5713 int upper = FALSE; |
| 366 | 5714 int lasti = 5; |
| 5715 | |
| 334 | 5716 /* Myspell allows extra text after the item, but that might |
| 6224 | 5717 * mean mistakes go unnoticed. Require a comment-starter, |
| 5718 * unless IGNOREEXTRA is used. Hunspell uses a "-" item. */ | |
| 5719 if (itemcnt > lasti | |
| 5720 && !aff->af_ignoreextra | |
| 5721 && *items[lasti] != '#' | |
| 809 | 5722 && (STRCMP(items[lasti], "-") != 0 |
| 5723 || itemcnt != lasti + 1)) | |
| 482 | 5724 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]); |
| 334 | 5725 |
| 236 | 5726 /* New item for an affix letter. */ |
| 5727 --aff_todo; | |
| 474 | 5728 aff_entry = (affentry_T *)getroom(spin, |
| 447 | 5729 sizeof(affentry_T), TRUE); |
| 236 | 5730 if (aff_entry == NULL) |
| 5731 break; | |
| 240 | 5732 |
| 236 | 5733 if (STRCMP(items[2], "0") != 0) |
| 474 | 5734 aff_entry->ae_chop = getroom_save(spin, items[2]); |
| 236 | 5735 if (STRCMP(items[3], "0") != 0) |
| 809 | 5736 { |
| 474 | 5737 aff_entry->ae_add = getroom_save(spin, items[3]); |
| 236 | 5738 |
| 822 | 5739 /* Recognize flags on the affix: abcd/XYZ */ |
| 809 | 5740 aff_entry->ae_flags = vim_strchr(aff_entry->ae_add, '/'); |
| 5741 if (aff_entry->ae_flags != NULL) | |
| 822 | 5742 { |
| 809 | 5743 *aff_entry->ae_flags++ = NUL; |
| 822 | 5744 aff_process_flags(aff, aff_entry); |
| 5745 } | |
| 809 | 5746 } |
| 5747 | |
| 300 | 5748 /* Don't use an affix entry with non-ASCII characters when |
| 5749 * "spin->si_ascii" is TRUE. */ | |
| 5750 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop) | |
| 240 | 5751 || has_non_ascii(aff_entry->ae_add))) |
| 5752 { | |
| 5753 aff_entry->ae_next = cur_aff->ah_first; | |
| 5754 cur_aff->ah_first = aff_entry; | |
| 300 | 5755 |
| 5756 if (STRCMP(items[4], ".") != 0) | |
| 5757 { | |
| 5758 char_u buf[MAXLINELEN]; | |
| 5759 | |
| 474 | 5760 aff_entry->ae_cond = getroom_save(spin, items[4]); |
| 300 | 5761 if (*items[0] == 'P') |
| 5762 sprintf((char *)buf, "^%s", items[4]); | |
| 5763 else | |
| 5764 sprintf((char *)buf, "%s$", items[4]); | |
| 5765 aff_entry->ae_prog = vim_regcomp(buf, | |
| 481 | 5766 RE_MAGIC + RE_STRING + RE_STRICT); |
| 5767 if (aff_entry->ae_prog == NULL) | |
| 5768 smsg((char_u *)_("Broken condition in %s line %d: %s"), | |
| 5769 fname, lnum, items[4]); | |
| 300 | 5770 } |
| 339 | 5771 |
| 5772 /* For postponed prefixes we need an entry in si_prefcond | |
| 809 | 5773 * for the condition. Use an existing one if possible. |
| 822 | 5774 * Can't be done for an affix with flags, ignoring |
| 5775 * COMPOUNDFORBIDFLAG and COMPOUNDPERMITFLAG. */ | |
| 809 | 5776 if (*items[0] == 'P' && aff->af_pfxpostpone |
| 5777 && aff_entry->ae_flags == NULL) | |
| 339 | 5778 { |
| 455 | 5779 /* When the chop string is one lower-case letter and |
| 5780 * the add string ends in the upper-case letter we set | |
| 5781 * the "upper" flag, clear "ae_chop" and remove the | |
| 5782 * letters from "ae_add". The condition must either | |
| 5783 * be empty or start with the same letter. */ | |
| 5784 if (aff_entry->ae_chop != NULL | |
| 5785 && aff_entry->ae_add != NULL | |
| 5786 #ifdef FEAT_MBYTE | |
| 474 | 5787 && aff_entry->ae_chop[(*mb_ptr2len)( |
| 455 | 5788 aff_entry->ae_chop)] == NUL |
| 5789 #else | |
| 5790 && aff_entry->ae_chop[1] == NUL | |
| 5791 #endif | |
| 5792 ) | |
| 339 | 5793 { |
| 455 | 5794 int c, c_up; |
| 5795 | |
| 5796 c = PTR2CHAR(aff_entry->ae_chop); | |
| 5797 c_up = SPELL_TOUPPER(c); | |
| 5798 if (c_up != c | |
| 5799 && (aff_entry->ae_cond == NULL | |
| 5800 || PTR2CHAR(aff_entry->ae_cond) == c)) | |
| 5801 { | |
| 5802 p = aff_entry->ae_add | |
| 5803 + STRLEN(aff_entry->ae_add); | |
| 5804 mb_ptr_back(aff_entry->ae_add, p); | |
| 5805 if (PTR2CHAR(p) == c_up) | |
| 5806 { | |
| 5807 upper = TRUE; | |
| 5808 aff_entry->ae_chop = NULL; | |
| 5809 *p = NUL; | |
| 5810 | |
| 5811 /* The condition is matched with the | |
| 5812 * actual word, thus must check for the | |
| 5813 * upper-case letter. */ | |
| 5814 if (aff_entry->ae_cond != NULL) | |
| 5815 { | |
| 5816 char_u buf[MAXLINELEN]; | |
| 5817 #ifdef FEAT_MBYTE | |
| 5818 if (has_mbyte) | |
| 5819 { | |
| 5820 onecap_copy(items[4], buf, TRUE); | |
| 5821 aff_entry->ae_cond = getroom_save( | |
| 474 | 5822 spin, buf); |
| 455 | 5823 } |
| 5824 else | |
| 5825 #endif | |
| 5826 *aff_entry->ae_cond = c_up; | |
| 5827 if (aff_entry->ae_cond != NULL) | |
| 5828 { | |
| 5829 sprintf((char *)buf, "^%s", | |
| 339 | 5830 aff_entry->ae_cond); |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4352
diff
changeset
|
5831 vim_regfree(aff_entry->ae_prog); |
| 455 | 5832 aff_entry->ae_prog = vim_regcomp( |
| 5833 buf, RE_MAGIC + RE_STRING); | |
| 5834 } | |
| 5835 } | |
| 5836 } | |
| 5837 } | |
| 339 | 5838 } |
| 5839 | |
| 809 | 5840 if (aff_entry->ae_chop == NULL |
| 5841 && aff_entry->ae_flags == NULL) | |
| 390 | 5842 { |
| 455 | 5843 int idx; |
| 5844 char_u **pp; | |
| 5845 int n; | |
| 5846 | |
| 495 | 5847 /* Find a previously used condition. */ |
| 455 | 5848 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0; |
| 5849 --idx) | |
| 5850 { | |
| 5851 p = ((char_u **)spin->si_prefcond.ga_data)[idx]; | |
| 5852 if (str_equal(p, aff_entry->ae_cond)) | |
| 5853 break; | |
| 5854 } | |
| 5855 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK) | |
| 5856 { | |
| 5857 /* Not found, add a new condition. */ | |
| 5858 idx = spin->si_prefcond.ga_len++; | |
| 5859 pp = ((char_u **)spin->si_prefcond.ga_data) | |
| 5860 + idx; | |
| 5861 if (aff_entry->ae_cond == NULL) | |
| 5862 *pp = NULL; | |
| 5863 else | |
| 474 | 5864 *pp = getroom_save(spin, |
| 455 | 5865 aff_entry->ae_cond); |
| 5866 } | |
| 5867 | |
| 5868 /* Add the prefix to the prefix tree. */ | |
| 5869 if (aff_entry->ae_add == NULL) | |
| 5870 p = (char_u *)""; | |
| 390 | 5871 else |
| 455 | 5872 p = aff_entry->ae_add; |
| 809 | 5873 |
| 455 | 5874 /* PFX_FLAGS is a negative number, so that |
| 5875 * tree_add_word() knows this is the prefix tree. */ | |
| 5876 n = PFX_FLAGS; | |
| 5877 if (!cur_aff->ah_combine) | |
| 5878 n |= WFP_NC; | |
| 5879 if (upper) | |
| 5880 n |= WFP_UP; | |
| 822 | 5881 if (aff_entry->ae_comppermit) |
| 5882 n |= WFP_COMPPERMIT; | |
| 5883 if (aff_entry->ae_compforbid) | |
| 5884 n |= WFP_COMPFORBID; | |
| 474 | 5885 tree_add_word(spin, p, spin->si_prefroot, n, |
| 5886 idx, cur_aff->ah_newID); | |
| 495 | 5887 did_postpone_prefix = TRUE; |
| 5888 } | |
| 5889 | |
| 5890 /* Didn't actually use ah_newID, backup si_newprefID. */ | |
| 5891 if (aff_todo == 0 && !did_postpone_prefix) | |
| 5892 { | |
| 5893 --spin->si_newprefID; | |
| 5894 cur_aff->ah_newID = 0; | |
| 390 | 5895 } |
| 339 | 5896 } |
| 240 | 5897 } |
| 236 | 5898 } |
| 1762 | 5899 else if (is_aff_rule(items, itemcnt, "FOL", 2) && fol == NULL) |
| 495 | 5900 { |
| 5901 fol = vim_strsave(items[1]); | |
| 5902 } | |
| 1762 | 5903 else if (is_aff_rule(items, itemcnt, "LOW", 2) && low == NULL) |
| 495 | 5904 { |
| 5905 low = vim_strsave(items[1]); | |
| 5906 } | |
| 1762 | 5907 else if (is_aff_rule(items, itemcnt, "UPP", 2) && upp == NULL) |
| 495 | 5908 { |
| 5909 upp = vim_strsave(items[1]); | |
| 255 | 5910 } |
| 1762 | 5911 else if (is_aff_rule(items, itemcnt, "REP", 2) |
| 5912 || is_aff_rule(items, itemcnt, "REPSAL", 2)) | |
| 625 | 5913 { |
| 5914 /* Ignore REP/REPSAL count */; | |
| 323 | 5915 if (!isdigit(*items[1])) |
| 625 | 5916 smsg((char_u *)_("Expected REP(SAL) count in %s line %d"), |
| 323 | 5917 fname, lnum); |
| 5918 } | |
| 625 | 5919 else if ((STRCMP(items[0], "REP") == 0 |
| 5920 || STRCMP(items[0], "REPSAL") == 0) | |
| 5921 && itemcnt >= 3) | |
| 5922 { | |
| 5923 /* REP/REPSAL item */ | |
| 482 | 5924 /* Myspell ignores extra arguments, we require it starts with |
| 5925 * # to detect mistakes. */ | |
| 5926 if (itemcnt > 3 && items[3][0] != '#') | |
| 5927 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]); | |
| 625 | 5928 if (items[0][3] == 'S' ? do_repsal : do_rep) |
| 532 | 5929 { |
| 5930 /* Replace underscore with space (can't include a space | |
| 5931 * directly). */ | |
| 5932 for (p = items[1]; *p != NUL; mb_ptr_adv(p)) | |
| 5933 if (*p == '_') | |
| 5934 *p = ' '; | |
| 5935 for (p = items[2]; *p != NUL; mb_ptr_adv(p)) | |
| 5936 if (*p == '_') | |
| 5937 *p = ' '; | |
| 625 | 5938 add_fromto(spin, items[0][3] == 'S' |
| 5939 ? &spin->si_repsal | |
| 5940 : &spin->si_rep, items[1], items[2]); | |
| 532 | 5941 } |
| 323 | 5942 } |
| 1762 | 5943 else if (is_aff_rule(items, itemcnt, "MAP", 2)) |
| 323 | 5944 { |
| 5945 /* MAP item or count */ | |
| 5946 if (!found_map) | |
| 5947 { | |
| 5948 /* First line contains the count. */ | |
| 5949 found_map = TRUE; | |
| 5950 if (!isdigit(*items[1])) | |
| 5951 smsg((char_u *)_("Expected MAP count in %s line %d"), | |
| 5952 fname, lnum); | |
| 5953 } | |
| 944 | 5954 else if (do_mapline) |
| 323 | 5955 { |
| 346 | 5956 int c; |
| 5957 | |
| 5958 /* Check that every character appears only once. */ | |
| 5959 for (p = items[1]; *p != NUL; ) | |
| 5960 { | |
| 5961 #ifdef FEAT_MBYTE | |
| 5962 c = mb_ptr2char_adv(&p); | |
| 5963 #else | |
| 5964 c = *p++; | |
| 5965 #endif | |
| 5966 if ((spin->si_map.ga_len > 0 | |
| 5967 && vim_strchr(spin->si_map.ga_data, c) | |
| 5968 != NULL) | |
| 5969 || vim_strchr(p, c) != NULL) | |
| 5970 smsg((char_u *)_("Duplicate character in MAP in %s line %d"), | |
| 5971 fname, lnum); | |
| 5972 } | |
| 5973 | |
| 323 | 5974 /* We simply concatenate all the MAP strings, separated by |
| 5975 * slashes. */ | |
| 5976 ga_concat(&spin->si_map, items[1]); | |
| 5977 ga_append(&spin->si_map, '/'); | |
| 5978 } | |
| 5979 } | |
| 1762 | 5980 /* Accept "SAL from to" and "SAL from to #comment". */ |
| 5981 else if (is_aff_rule(items, itemcnt, "SAL", 3)) | |
| 323 | 5982 { |
| 5983 if (do_sal) | |
| 5984 { | |
| 5985 /* SAL item (sounds-a-like) | |
| 5986 * Either one of the known keys or a from-to pair. */ | |
| 5987 if (STRCMP(items[1], "followup") == 0) | |
| 5988 spin->si_followup = sal_to_bool(items[2]); | |
| 5989 else if (STRCMP(items[1], "collapse_result") == 0) | |
| 5990 spin->si_collapse = sal_to_bool(items[2]); | |
| 5991 else if (STRCMP(items[1], "remove_accents") == 0) | |
| 5992 spin->si_rem_accents = sal_to_bool(items[2]); | |
| 5993 else | |
| 5994 /* when "to" is "_" it means empty */ | |
| 5995 add_fromto(spin, &spin->si_sal, items[1], | |
| 5996 STRCMP(items[2], "_") == 0 ? (char_u *)"" | |
| 5997 : items[2]); | |
| 5998 } | |
| 236 | 5999 } |
| 1762 | 6000 else if (is_aff_rule(items, itemcnt, "SOFOFROM", 2) |
| 495 | 6001 && sofofrom == NULL) |
| 6002 { | |
| 6003 sofofrom = getroom_save(spin, items[1]); | |
| 375 | 6004 } |
| 1762 | 6005 else if (is_aff_rule(items, itemcnt, "SOFOTO", 2) |
| 495 | 6006 && sofoto == NULL) |
| 6007 { | |
| 6008 sofoto = getroom_save(spin, items[1]); | |
| 375 | 6009 } |
| 625 | 6010 else if (STRCMP(items[0], "COMMON") == 0) |
| 6011 { | |
| 6012 int i; | |
| 6013 | |
| 6014 for (i = 1; i < itemcnt; ++i) | |
| 6015 { | |
| 6016 if (HASHITEM_EMPTY(hash_find(&spin->si_commonwords, | |
| 6017 items[i]))) | |
| 6018 { | |
| 6019 p = vim_strsave(items[i]); | |
| 6020 if (p == NULL) | |
| 6021 break; | |
| 6022 hash_add(&spin->si_commonwords, p); | |
| 6023 } | |
| 6024 } | |
| 6025 } | |
| 300 | 6026 else |
| 481 | 6027 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"), |
| 236 | 6028 fname, lnum, items[0]); |
| 6029 } | |
| 6030 } | |
| 6031 | |
| 255 | 6032 if (fol != NULL || low != NULL || upp != NULL) |
| 6033 { | |
| 351 | 6034 if (spin->si_clear_chartab) |
| 6035 { | |
| 6036 /* Clear the char type tables, don't want to use any of the | |
| 6037 * currently used spell properties. */ | |
| 6038 init_spell_chartab(); | |
| 6039 spin->si_clear_chartab = FALSE; | |
| 6040 } | |
| 6041 | |
| 316 | 6042 /* |
| 6043 * Don't write a word table for an ASCII file, so that we don't check | |
| 6044 * for conflicts with a word table that matches 'encoding'. | |
| 324 | 6045 * Don't write one for utf-8 either, we use utf_*() and |
| 316 | 6046 * mb_get_class(), the list of chars in the file will be incomplete. |
| 6047 */ | |
| 6048 if (!spin->si_ascii | |
| 6049 #ifdef FEAT_MBYTE | |
| 6050 && !enc_utf8 | |
| 6051 #endif | |
| 6052 ) | |
| 260 | 6053 { |
| 6054 if (fol == NULL || low == NULL || upp == NULL) | |
| 6055 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname); | |
| 6056 else | |
| 316 | 6057 (void)set_spell_chartab(fol, low, upp); |
| 260 | 6058 } |
| 255 | 6059 |
| 6060 vim_free(fol); | |
| 6061 vim_free(low); | |
| 6062 vim_free(upp); | |
| 6063 } | |
| 6064 | |
| 481 | 6065 /* Use compound specifications of the .aff file for the spell info. */ |
| 495 | 6066 if (compmax != 0) |
| 6067 { | |
| 809 | 6068 aff_check_number(spin->si_compmax, compmax, "COMPOUNDWORDMAX"); |
| 495 | 6069 spin->si_compmax = compmax; |
| 6070 } | |
| 6071 | |
| 6072 if (compminlen != 0) | |
| 6073 { | |
| 6074 aff_check_number(spin->si_compminlen, compminlen, "COMPOUNDMIN"); | |
| 6075 spin->si_compminlen = compminlen; | |
| 6076 } | |
| 6077 | |
| 6078 if (compsylmax != 0) | |
| 6079 { | |
| 6080 if (syllable == NULL) | |
| 6081 smsg((char_u *)_("COMPOUNDSYLMAX used without SYLLABLE")); | |
| 6082 aff_check_number(spin->si_compsylmax, compsylmax, "COMPOUNDSYLMAX"); | |
| 6083 spin->si_compsylmax = compsylmax; | |
| 6084 } | |
| 6085 | |
| 809 | 6086 if (compoptions != 0) |
| 6087 { | |
| 6088 aff_check_number(spin->si_compoptions, compoptions, "COMPOUND options"); | |
| 6089 spin->si_compoptions |= compoptions; | |
| 6090 } | |
| 6091 | |
| 495 | 6092 if (compflags != NULL) |
| 6093 process_compflags(spin, aff, compflags); | |
| 6094 | |
| 6095 /* Check that we didn't use too many renumbered flags. */ | |
| 500 | 6096 if (spin->si_newcompID < spin->si_newprefID) |
| 6097 { | |
| 6098 if (spin->si_newcompID == 127 || spin->si_newcompID == 255) | |
| 495 | 6099 MSG(_("Too many postponed prefixes")); |
| 500 | 6100 else if (spin->si_newprefID == 0 || spin->si_newprefID == 127) |
| 495 | 6101 MSG(_("Too many compound flags")); |
| 481 | 6102 else |
| 1705 | 6103 MSG(_("Too many postponed prefixes and/or compound flags")); |
| 495 | 6104 } |
| 6105 | |
| 6106 if (syllable != NULL) | |
| 6107 { | |
| 6108 aff_check_string(spin->si_syllable, syllable, "SYLLABLE"); | |
| 6109 spin->si_syllable = syllable; | |
| 6110 } | |
| 6111 | |
| 6112 if (sofofrom != NULL || sofoto != NULL) | |
| 6113 { | |
| 6114 if (sofofrom == NULL || sofoto == NULL) | |
| 6115 smsg((char_u *)_("Missing SOFO%s line in %s"), | |
| 6116 sofofrom == NULL ? "FROM" : "TO", fname); | |
| 6117 else if (spin->si_sal.ga_len > 0) | |
| 6118 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname); | |
| 483 | 6119 else |
| 495 | 6120 { |
| 6121 aff_check_string(spin->si_sofofr, sofofrom, "SOFOFROM"); | |
| 6122 aff_check_string(spin->si_sofoto, sofoto, "SOFOTO"); | |
| 6123 spin->si_sofofr = sofofrom; | |
| 6124 spin->si_sofoto = sofoto; | |
| 6125 } | |
| 6126 } | |
| 6127 | |
| 6128 if (midword != NULL) | |
| 6129 { | |
| 6130 aff_check_string(spin->si_midword, midword, "MIDWORD"); | |
| 6131 spin->si_midword = midword; | |
| 483 | 6132 } |
| 6133 | |
| 236 | 6134 vim_free(pc); |
| 6135 fclose(fd); | |
| 6136 return aff; | |
| 6137 } | |
| 6138 | |
| 6139 /* | |
| 1762 | 6140 * Return TRUE when items[0] equals "rulename", there are "mincount" items or |
| 6141 * a comment is following after item "mincount". | |
| 6142 */ | |
| 6143 static int | |
| 6144 is_aff_rule(items, itemcnt, rulename, mincount) | |
| 6145 char_u **items; | |
| 6146 int itemcnt; | |
| 6147 char *rulename; | |
| 6148 int mincount; | |
| 6149 { | |
| 6150 return (STRCMP(items[0], rulename) == 0 | |
| 6151 && (itemcnt == mincount | |
| 6152 || (itemcnt > mincount && items[mincount][0] == '#'))); | |
| 6153 } | |
| 6154 | |
| 6155 /* | |
| 822 | 6156 * For affix "entry" move COMPOUNDFORBIDFLAG and COMPOUNDPERMITFLAG from |
| 6157 * ae_flags to ae_comppermit and ae_compforbid. | |
| 6158 */ | |
| 6159 static void | |
| 6160 aff_process_flags(affile, entry) | |
| 6161 afffile_T *affile; | |
| 6162 affentry_T *entry; | |
| 6163 { | |
| 6164 char_u *p; | |
| 6165 char_u *prevp; | |
| 825 | 6166 unsigned flag; |
| 822 | 6167 |
| 6168 if (entry->ae_flags != NULL | |
| 6169 && (affile->af_compforbid != 0 || affile->af_comppermit != 0)) | |
| 6170 { | |
| 6171 for (p = entry->ae_flags; *p != NUL; ) | |
| 6172 { | |
| 6173 prevp = p; | |
| 6174 flag = get_affitem(affile->af_flagtype, &p); | |
| 6175 if (flag == affile->af_comppermit || flag == affile->af_compforbid) | |
| 6176 { | |
| 1619 | 6177 STRMOVE(prevp, p); |
| 822 | 6178 p = prevp; |
| 6179 if (flag == affile->af_comppermit) | |
| 6180 entry->ae_comppermit = TRUE; | |
| 6181 else | |
| 6182 entry->ae_compforbid = TRUE; | |
| 6183 } | |
| 6184 if (affile->af_flagtype == AFT_NUM && *p == ',') | |
| 6185 ++p; | |
| 6186 } | |
| 6187 if (*entry->ae_flags == NUL) | |
| 6188 entry->ae_flags = NULL; /* nothing left */ | |
| 6189 } | |
| 6190 } | |
| 6191 | |
| 6192 /* | |
| 714 | 6193 * Return TRUE if "s" is the name of an info item in the affix file. |
| 6194 */ | |
| 6195 static int | |
| 6196 spell_info_item(s) | |
| 6197 char_u *s; | |
| 6198 { | |
| 6199 return STRCMP(s, "NAME") == 0 | |
| 6200 || STRCMP(s, "HOME") == 0 | |
| 6201 || STRCMP(s, "VERSION") == 0 | |
| 6202 || STRCMP(s, "AUTHOR") == 0 | |
| 6203 || STRCMP(s, "EMAIL") == 0 | |
| 6204 || STRCMP(s, "COPYRIGHT") == 0; | |
| 6205 } | |
| 6206 | |
| 6207 /* | |
| 495 | 6208 * Turn an affix flag name into a number, according to the FLAG type. |
| 6209 * returns zero for failure. | |
| 6210 */ | |
| 6211 static unsigned | |
| 6212 affitem2flag(flagtype, item, fname, lnum) | |
| 6213 int flagtype; | |
| 6214 char_u *item; | |
| 6215 char_u *fname; | |
| 6216 int lnum; | |
| 6217 { | |
| 6218 unsigned res; | |
| 6219 char_u *p = item; | |
| 6220 | |
| 6221 res = get_affitem(flagtype, &p); | |
| 6222 if (res == 0) | |
| 6223 { | |
| 498 | 6224 if (flagtype == AFT_NUM) |
| 495 | 6225 smsg((char_u *)_("Flag is not a number in %s line %d: %s"), |
| 6226 fname, lnum, item); | |
| 6227 else | |
| 6228 smsg((char_u *)_("Illegal flag in %s line %d: %s"), | |
| 6229 fname, lnum, item); | |
| 6230 } | |
| 6231 if (*p != NUL) | |
| 6232 { | |
| 6233 smsg((char_u *)_(e_affname), fname, lnum, item); | |
| 6234 return 0; | |
| 6235 } | |
| 6236 | |
| 6237 return res; | |
| 6238 } | |
| 6239 | |
| 6240 /* | |
| 6241 * Get one affix name from "*pp" and advance the pointer. | |
| 6242 * Returns zero for an error, still advances the pointer then. | |
| 6243 */ | |
| 6244 static unsigned | |
| 6245 get_affitem(flagtype, pp) | |
| 6246 int flagtype; | |
| 6247 char_u **pp; | |
| 6248 { | |
| 6249 int res; | |
| 6250 | |
| 498 | 6251 if (flagtype == AFT_NUM) |
| 495 | 6252 { |
| 6253 if (!VIM_ISDIGIT(**pp)) | |
| 6254 { | |
| 498 | 6255 ++*pp; /* always advance, avoid getting stuck */ |
| 495 | 6256 return 0; |
| 6257 } | |
| 6258 res = getdigits(pp); | |
| 6259 } | |
| 6260 else | |
| 6261 { | |
| 6262 #ifdef FEAT_MBYTE | |
| 6263 res = mb_ptr2char_adv(pp); | |
| 6264 #else | |
| 6265 res = *(*pp)++; | |
| 6266 #endif | |
| 498 | 6267 if (flagtype == AFT_LONG || (flagtype == AFT_CAPLONG |
| 495 | 6268 && res >= 'A' && res <= 'Z')) |
| 6269 { | |
| 6270 if (**pp == NUL) | |
| 6271 return 0; | |
| 6272 #ifdef FEAT_MBYTE | |
| 6273 res = mb_ptr2char_adv(pp) + (res << 16); | |
| 6274 #else | |
| 6275 res = *(*pp)++ + (res << 16); | |
| 6276 #endif | |
| 6277 } | |
| 6278 } | |
| 6279 return res; | |
| 6280 } | |
| 6281 | |
| 6282 /* | |
| 6283 * Process the "compflags" string used in an affix file and append it to | |
| 6284 * spin->si_compflags. | |
| 6285 * The processing involves changing the affix names to ID numbers, so that | |
| 6286 * they fit in one byte. | |
| 6287 */ | |
| 6288 static void | |
| 6289 process_compflags(spin, aff, compflags) | |
| 6290 spellinfo_T *spin; | |
| 6291 afffile_T *aff; | |
| 6292 char_u *compflags; | |
| 6293 { | |
| 6294 char_u *p; | |
| 6295 char_u *prevp; | |
| 6296 unsigned flag; | |
| 6297 compitem_T *ci; | |
| 6298 int id; | |
| 6299 int len; | |
| 6300 char_u *tp; | |
| 6301 char_u key[AH_KEY_LEN]; | |
| 6302 hashitem_T *hi; | |
| 6303 | |
| 6304 /* Make room for the old and the new compflags, concatenated with a / in | |
| 6305 * between. Processing it makes it shorter, but we don't know by how | |
| 6306 * much, thus allocate the maximum. */ | |
| 835 | 6307 len = (int)STRLEN(compflags) + 1; |
| 495 | 6308 if (spin->si_compflags != NULL) |
| 835 | 6309 len += (int)STRLEN(spin->si_compflags) + 1; |
| 495 | 6310 p = getroom(spin, len, FALSE); |
| 6311 if (p == NULL) | |
| 6312 return; | |
| 6313 if (spin->si_compflags != NULL) | |
| 6314 { | |
| 6315 STRCPY(p, spin->si_compflags); | |
| 6316 STRCAT(p, "/"); | |
| 6317 } | |
| 6318 spin->si_compflags = p; | |
| 6319 tp = p + STRLEN(p); | |
| 6320 | |
| 6321 for (p = compflags; *p != NUL; ) | |
| 6322 { | |
| 2691 | 6323 if (vim_strchr((char_u *)"/?*+[]", *p) != NULL) |
| 495 | 6324 /* Copy non-flag characters directly. */ |
| 6325 *tp++ = *p++; | |
| 6326 else | |
| 6327 { | |
| 6328 /* First get the flag number, also checks validity. */ | |
| 6329 prevp = p; | |
| 6330 flag = get_affitem(aff->af_flagtype, &p); | |
| 6331 if (flag != 0) | |
| 6332 { | |
| 6333 /* Find the flag in the hashtable. If it was used before, use | |
| 6334 * the existing ID. Otherwise add a new entry. */ | |
| 6335 vim_strncpy(key, prevp, p - prevp); | |
| 6336 hi = hash_find(&aff->af_comp, key); | |
| 6337 if (!HASHITEM_EMPTY(hi)) | |
| 6338 id = HI2CI(hi)->ci_newID; | |
| 6339 else | |
| 6340 { | |
| 6341 ci = (compitem_T *)getroom(spin, sizeof(compitem_T), TRUE); | |
| 6342 if (ci == NULL) | |
| 6343 break; | |
| 6344 STRCPY(ci->ci_key, key); | |
| 6345 ci->ci_flag = flag; | |
| 6346 /* Avoid using a flag ID that has a special meaning in a | |
| 6347 * regexp (also inside []). */ | |
| 6348 do | |
| 6349 { | |
| 500 | 6350 check_renumber(spin); |
| 6351 id = spin->si_newcompID--; | |
| 2691 | 6352 } while (vim_strchr((char_u *)"/?*+[]\\-^", id) != NULL); |
| 495 | 6353 ci->ci_newID = id; |
| 6354 hash_add(&aff->af_comp, ci->ci_key); | |
| 6355 } | |
| 6356 *tp++ = id; | |
| 6357 } | |
| 498 | 6358 if (aff->af_flagtype == AFT_NUM && *p == ',') |
| 495 | 6359 ++p; |
| 6360 } | |
| 6361 } | |
| 6362 | |
| 6363 *tp = NUL; | |
| 6364 } | |
| 6365 | |
| 6366 /* | |
| 500 | 6367 * Check that the new IDs for postponed affixes and compounding don't overrun |
| 6368 * each other. We have almost 255 available, but start at 0-127 to avoid | |
| 6369 * using two bytes for utf-8. When the 0-127 range is used up go to 128-255. | |
| 6370 * When that is used up an error message is given. | |
| 6371 */ | |
| 6372 static void | |
| 6373 check_renumber(spin) | |
| 6374 spellinfo_T *spin; | |
| 6375 { | |
| 6376 if (spin->si_newprefID == spin->si_newcompID && spin->si_newcompID < 128) | |
| 6377 { | |
| 6378 spin->si_newprefID = 127; | |
| 6379 spin->si_newcompID = 255; | |
| 6380 } | |
| 6381 } | |
| 6382 | |
| 6383 /* | |
| 495 | 6384 * Return TRUE if flag "flag" appears in affix list "afflist". |
| 6385 */ | |
| 6386 static int | |
| 6387 flag_in_afflist(flagtype, afflist, flag) | |
| 6388 int flagtype; | |
| 6389 char_u *afflist; | |
| 6390 unsigned flag; | |
| 6391 { | |
| 6392 char_u *p; | |
| 6393 unsigned n; | |
| 6394 | |
| 6395 switch (flagtype) | |
| 6396 { | |
| 6397 case AFT_CHAR: | |
| 6398 return vim_strchr(afflist, flag) != NULL; | |
| 6399 | |
| 498 | 6400 case AFT_CAPLONG: |
| 6401 case AFT_LONG: | |
| 495 | 6402 for (p = afflist; *p != NUL; ) |
| 6403 { | |
| 6404 #ifdef FEAT_MBYTE | |
| 6405 n = mb_ptr2char_adv(&p); | |
| 6406 #else | |
| 6407 n = *p++; | |
| 6408 #endif | |
| 498 | 6409 if ((flagtype == AFT_LONG || (n >= 'A' && n <= 'Z')) |
| 495 | 6410 && *p != NUL) |
| 6411 #ifdef FEAT_MBYTE | |
| 6412 n = mb_ptr2char_adv(&p) + (n << 16); | |
| 6413 #else | |
| 6414 n = *p++ + (n << 16); | |
| 6415 #endif | |
| 6416 if (n == flag) | |
| 6417 return TRUE; | |
| 6418 } | |
| 6419 break; | |
| 6420 | |
| 498 | 6421 case AFT_NUM: |
| 495 | 6422 for (p = afflist; *p != NUL; ) |
| 6423 { | |
| 6424 n = getdigits(&p); | |
| 6425 if (n == flag) | |
| 6426 return TRUE; | |
| 6427 if (*p != NUL) /* skip over comma */ | |
| 6428 ++p; | |
| 6429 } | |
| 6430 break; | |
| 6431 } | |
| 6432 return FALSE; | |
| 6433 } | |
| 6434 | |
| 6435 /* | |
| 6436 * Give a warning when "spinval" and "affval" numbers are set and not the same. | |
| 6437 */ | |
| 6438 static void | |
| 6439 aff_check_number(spinval, affval, name) | |
| 6440 int spinval; | |
| 6441 int affval; | |
| 6442 char *name; | |
| 6443 { | |
| 6444 if (spinval != 0 && spinval != affval) | |
| 6445 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name); | |
| 6446 } | |
| 6447 | |
| 6448 /* | |
| 6449 * Give a warning when "spinval" and "affval" strings are set and not the same. | |
| 6450 */ | |
| 6451 static void | |
| 6452 aff_check_string(spinval, affval, name) | |
| 6453 char_u *spinval; | |
| 6454 char_u *affval; | |
| 6455 char *name; | |
| 6456 { | |
| 6457 if (spinval != NULL && STRCMP(spinval, affval) != 0) | |
| 6458 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name); | |
| 6459 } | |
| 6460 | |
| 6461 /* | |
| 339 | 6462 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being |
| 6463 * NULL as equal. | |
| 6464 */ | |
| 6465 static int | |
| 6466 str_equal(s1, s2) | |
| 6467 char_u *s1; | |
| 6468 char_u *s2; | |
| 6469 { | |
| 6470 if (s1 == NULL || s2 == NULL) | |
| 6471 return s1 == s2; | |
| 6472 return STRCMP(s1, s2) == 0; | |
| 6473 } | |
| 6474 | |
| 6475 /* | |
| 323 | 6476 * Add a from-to item to "gap". Used for REP and SAL items. |
| 6477 * They are stored case-folded. | |
| 6478 */ | |
| 6479 static void | |
| 6480 add_fromto(spin, gap, from, to) | |
| 6481 spellinfo_T *spin; | |
| 6482 garray_T *gap; | |
| 6483 char_u *from; | |
| 6484 char_u *to; | |
| 6485 { | |
| 6486 fromto_T *ftp; | |
| 6487 char_u word[MAXWLEN]; | |
| 6488 | |
| 6489 if (ga_grow(gap, 1) == OK) | |
| 6490 { | |
| 6491 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len; | |
| 835 | 6492 (void)spell_casefold(from, (int)STRLEN(from), word, MAXWLEN); |
| 474 | 6493 ftp->ft_from = getroom_save(spin, word); |
| 835 | 6494 (void)spell_casefold(to, (int)STRLEN(to), word, MAXWLEN); |
| 474 | 6495 ftp->ft_to = getroom_save(spin, word); |
| 323 | 6496 ++gap->ga_len; |
| 6497 } | |
| 6498 } | |
| 6499 | |
| 6500 /* | |
| 6501 * Convert a boolean argument in a SAL line to TRUE or FALSE; | |
| 6502 */ | |
| 6503 static int | |
| 6504 sal_to_bool(s) | |
| 6505 char_u *s; | |
| 6506 { | |
| 6507 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0; | |
| 6508 } | |
| 6509 | |
| 6510 /* | |
| 236 | 6511 * Free the structure filled by spell_read_aff(). |
| 6512 */ | |
| 6513 static void | |
| 6514 spell_free_aff(aff) | |
| 6515 afffile_T *aff; | |
| 6516 { | |
| 6517 hashtab_T *ht; | |
| 6518 hashitem_T *hi; | |
| 6519 int todo; | |
| 6520 affheader_T *ah; | |
| 300 | 6521 affentry_T *ae; |
| 236 | 6522 |
| 6523 vim_free(aff->af_enc); | |
| 6524 | |
| 339 | 6525 /* All this trouble to free the "ae_prog" items... */ |
| 236 | 6526 for (ht = &aff->af_pref; ; ht = &aff->af_suff) |
| 6527 { | |
| 835 | 6528 todo = (int)ht->ht_used; |
| 236 | 6529 for (hi = ht->ht_array; todo > 0; ++hi) |
| 6530 { | |
| 6531 if (!HASHITEM_EMPTY(hi)) | |
| 6532 { | |
| 6533 --todo; | |
| 6534 ah = HI2AH(hi); | |
| 300 | 6535 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
|
6536 vim_regfree(ae->ae_prog); |
| 236 | 6537 } |
| 6538 } | |
| 6539 if (ht == &aff->af_suff) | |
| 6540 break; | |
| 6541 } | |
| 300 | 6542 |
| 236 | 6543 hash_clear(&aff->af_pref); |
| 6544 hash_clear(&aff->af_suff); | |
| 495 | 6545 hash_clear(&aff->af_comp); |
| 236 | 6546 } |
| 6547 | |
| 6548 /* | |
| 300 | 6549 * Read dictionary file "fname". |
| 236 | 6550 * Returns OK or FAIL; |
| 6551 */ | |
| 6552 static int | |
| 474 | 6553 spell_read_dic(spin, fname, affile) |
| 6554 spellinfo_T *spin; | |
| 236 | 6555 char_u *fname; |
| 300 | 6556 afffile_T *affile; |
| 236 | 6557 { |
| 300 | 6558 hashtab_T ht; |
| 236 | 6559 char_u line[MAXLINELEN]; |
| 481 | 6560 char_u *p; |
| 300 | 6561 char_u *afflist; |
| 483 | 6562 char_u store_afflist[MAXWLEN]; |
| 6563 int pfxlen; | |
| 6564 int need_affix; | |
| 300 | 6565 char_u *dw; |
| 236 | 6566 char_u *pc; |
| 6567 char_u *w; | |
| 6568 int l; | |
| 6569 hash_T hash; | |
| 6570 hashitem_T *hi; | |
| 6571 FILE *fd; | |
| 6572 int lnum = 1; | |
| 300 | 6573 int non_ascii = 0; |
| 6574 int retval = OK; | |
| 6575 char_u message[MAXLINELEN + MAXWLEN]; | |
| 307 | 6576 int flags; |
| 481 | 6577 int duplicate = 0; |
| 236 | 6578 |
| 300 | 6579 /* |
| 6580 * Open the file. | |
| 6581 */ | |
| 310 | 6582 fd = mch_fopen((char *)fname, "r"); |
| 236 | 6583 if (fd == NULL) |
| 6584 { | |
| 6585 EMSG2(_(e_notopen), fname); | |
| 6586 return FAIL; | |
| 6587 } | |
| 6588 | |
| 300 | 6589 /* The hashtable is only used to detect duplicated words. */ |
| 6590 hash_init(&ht); | |
| 6591 | |
| 625 | 6592 vim_snprintf((char *)IObuff, IOSIZE, |
| 6593 _("Reading dictionary file %s ..."), fname); | |
| 6594 spell_message(spin, IObuff); | |
| 236 | 6595 |
| 474 | 6596 /* start with a message for the first line */ |
| 6597 spin->si_msg_count = 999999; | |
| 6598 | |
| 236 | 6599 /* Read and ignore the first line: word count. */ |
| 6600 (void)vim_fgets(line, MAXLINELEN, fd); | |
| 324 | 6601 if (!vim_isdigit(*skipwhite(line))) |
| 236 | 6602 EMSG2(_("E760: No word count in %s"), fname); |
| 6603 | |
| 6604 /* | |
| 6605 * Read all the lines in the file one by one. | |
| 6606 * The words are converted to 'encoding' here, before being added to | |
| 6607 * the hashtable. | |
| 6608 */ | |
| 255 | 6609 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int) |
| 236 | 6610 { |
| 255 | 6611 line_breakcheck(); |
| 236 | 6612 ++lnum; |
| 455 | 6613 if (line[0] == '#' || line[0] == '/') |
| 351 | 6614 continue; /* comment line */ |
| 236 | 6615 |
| 300 | 6616 /* Remove CR, LF and white space from the end. White space halfway |
| 6617 * the word is kept to allow e.g., "et al.". */ | |
| 835 | 6618 l = (int)STRLEN(line); |
| 236 | 6619 while (l > 0 && line[l - 1] <= ' ') |
| 6620 --l; | |
| 6621 if (l == 0) | |
| 6622 continue; /* empty line */ | |
| 6623 line[l] = NUL; | |
| 6624 | |
| 310 | 6625 #ifdef FEAT_MBYTE |
| 236 | 6626 /* Convert from "SET" to 'encoding' when needed. */ |
| 300 | 6627 if (spin->si_conv.vc_type != CONV_NONE) |
| 236 | 6628 { |
| 300 | 6629 pc = string_convert(&spin->si_conv, line, NULL); |
| 255 | 6630 if (pc == NULL) |
| 6631 { | |
| 6632 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"), | |
| 6633 fname, lnum, line); | |
| 6634 continue; | |
| 6635 } | |
| 236 | 6636 w = pc; |
| 6637 } | |
| 6638 else | |
| 310 | 6639 #endif |
| 236 | 6640 { |
| 6641 pc = NULL; | |
| 6642 w = line; | |
| 6643 } | |
| 6644 | |
| 822 | 6645 /* Truncate the word at the "/", set "afflist" to what follows. |
| 6646 * Replace "\/" by "/" and "\\" by "\". */ | |
| 6647 afflist = NULL; | |
| 6648 for (p = w; *p != NUL; mb_ptr_adv(p)) | |
| 6649 { | |
| 6650 if (*p == '\\' && (p[1] == '\\' || p[1] == '/')) | |
| 1619 | 6651 STRMOVE(p, p + 1); |
| 822 | 6652 else if (*p == '/') |
| 6653 { | |
| 6654 *p = NUL; | |
| 6655 afflist = p + 1; | |
| 6656 break; | |
| 6657 } | |
| 6658 } | |
| 6659 | |
| 6660 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */ | |
| 6661 if (spin->si_ascii && has_non_ascii(w)) | |
| 6662 { | |
| 6663 ++non_ascii; | |
| 825 | 6664 vim_free(pc); |
| 822 | 6665 continue; |
| 6666 } | |
| 6667 | |
| 474 | 6668 /* This takes time, print a message every 10000 words. */ |
| 6669 if (spin->si_verbose && spin->si_msg_count > 10000) | |
| 6670 { | |
| 6671 spin->si_msg_count = 0; | |
| 339 | 6672 vim_snprintf((char *)message, sizeof(message), |
| 6673 _("line %6d, word %6d - %s"), | |
| 6674 lnum, spin->si_foldwcount + spin->si_keepwcount, w); | |
| 6675 msg_start(); | |
| 6676 msg_puts_long_attr(message, 0); | |
| 6677 msg_clr_eos(); | |
| 6678 msg_didout = FALSE; | |
| 6679 msg_col = 0; | |
| 6680 out_flush(); | |
| 6681 } | |
| 6682 | |
| 300 | 6683 /* Store the word in the hashtable to be able to find duplicates. */ |
| 474 | 6684 dw = (char_u *)getroom_save(spin, w); |
| 236 | 6685 if (dw == NULL) |
| 825 | 6686 { |
| 300 | 6687 retval = FAIL; |
| 825 | 6688 vim_free(pc); |
| 236 | 6689 break; |
| 825 | 6690 } |
| 236 | 6691 |
| 300 | 6692 hash = hash_hash(dw); |
| 6693 hi = hash_lookup(&ht, dw, hash); | |
| 236 | 6694 if (!HASHITEM_EMPTY(hi)) |
| 481 | 6695 { |
| 6696 if (p_verbose > 0) | |
| 6697 smsg((char_u *)_("Duplicate word in %s line %d: %s"), | |
| 375 | 6698 fname, lnum, dw); |
| 481 | 6699 else if (duplicate == 0) |
| 6700 smsg((char_u *)_("First duplicate word in %s line %d: %s"), | |
| 6701 fname, lnum, dw); | |
| 6702 ++duplicate; | |
| 6703 } | |
| 236 | 6704 else |
| 300 | 6705 hash_add_item(&ht, hi, dw, hash); |
| 6706 | |
| 307 | 6707 flags = 0; |
| 483 | 6708 store_afflist[0] = NUL; |
| 6709 pfxlen = 0; | |
| 6710 need_affix = FALSE; | |
| 307 | 6711 if (afflist != NULL) |
| 6712 { | |
| 820 | 6713 /* Extract flags from the affix list. */ |
| 6714 flags |= get_affix_flags(affile, afflist); | |
| 6715 | |
| 495 | 6716 if (affile->af_needaffix != 0 && flag_in_afflist( |
| 6717 affile->af_flagtype, afflist, affile->af_needaffix)) | |
| 483 | 6718 need_affix = TRUE; |
| 339 | 6719 |
| 6720 if (affile->af_pfxpostpone) | |
| 6721 /* Need to store the list of prefix IDs with the word. */ | |
| 483 | 6722 pfxlen = get_pfxlist(affile, afflist, store_afflist); |
| 6723 | |
| 6724 if (spin->si_compflags != NULL) | |
| 6725 /* Need to store the list of compound flags with the word. | |
| 6726 * Concatenate them to the list of prefix IDs. */ | |
| 495 | 6727 get_compflags(affile, afflist, store_afflist + pfxlen); |
| 307 | 6728 } |
| 6729 | |
| 300 | 6730 /* Add the word to the word tree(s). */ |
| 483 | 6731 if (store_word(spin, dw, flags, spin->si_region, |
| 6732 store_afflist, need_affix) == FAIL) | |
| 300 | 6733 retval = FAIL; |
| 236 | 6734 |
| 300 | 6735 if (afflist != NULL) |
| 6736 { | |
| 6737 /* Find all matching suffixes and add the resulting words. | |
| 6738 * Additionally do matching prefixes that combine. */ | |
| 474 | 6739 if (store_aff_word(spin, dw, afflist, affile, |
| 307 | 6740 &affile->af_suff, &affile->af_pref, |
| 820 | 6741 CONDIT_SUF, flags, store_afflist, pfxlen) == FAIL) |
| 300 | 6742 retval = FAIL; |
| 6743 | |
| 6744 /* Find all matching prefixes and add the resulting words. */ | |
| 474 | 6745 if (store_aff_word(spin, dw, afflist, affile, |
| 339 | 6746 &affile->af_pref, NULL, |
| 820 | 6747 CONDIT_SUF, flags, store_afflist, pfxlen) == FAIL) |
| 300 | 6748 retval = FAIL; |
| 6749 } | |
| 825 | 6750 |
| 6751 vim_free(pc); | |
| 236 | 6752 } |
| 6753 | |
| 481 | 6754 if (duplicate > 0) |
| 6755 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname); | |
| 300 | 6756 if (spin->si_ascii && non_ascii > 0) |
| 481 | 6757 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"), |
| 6758 non_ascii, fname); | |
| 300 | 6759 hash_clear(&ht); |
| 6760 | |
| 236 | 6761 fclose(fd); |
| 300 | 6762 return retval; |
| 236 | 6763 } |
| 6764 | |
| 6765 /* | |
| 820 | 6766 * Check for affix flags in "afflist" that are turned into word flags. |
| 6767 * Return WF_ flags. | |
| 6768 */ | |
| 6769 static int | |
| 6770 get_affix_flags(affile, afflist) | |
| 6771 afffile_T *affile; | |
| 6772 char_u *afflist; | |
| 6773 { | |
| 6774 int flags = 0; | |
| 6775 | |
| 6776 if (affile->af_keepcase != 0 && flag_in_afflist( | |
| 6777 affile->af_flagtype, afflist, affile->af_keepcase)) | |
| 6778 flags |= WF_KEEPCAP | WF_FIXCAP; | |
| 6779 if (affile->af_rare != 0 && flag_in_afflist( | |
| 6780 affile->af_flagtype, afflist, affile->af_rare)) | |
| 6781 flags |= WF_RARE; | |
| 6782 if (affile->af_bad != 0 && flag_in_afflist( | |
| 6783 affile->af_flagtype, afflist, affile->af_bad)) | |
| 6784 flags |= WF_BANNED; | |
| 6785 if (affile->af_needcomp != 0 && flag_in_afflist( | |
| 6786 affile->af_flagtype, afflist, affile->af_needcomp)) | |
| 6787 flags |= WF_NEEDCOMP; | |
| 6788 if (affile->af_comproot != 0 && flag_in_afflist( | |
| 6789 affile->af_flagtype, afflist, affile->af_comproot)) | |
| 6790 flags |= WF_COMPROOT; | |
| 6791 if (affile->af_nosuggest != 0 && flag_in_afflist( | |
| 6792 affile->af_flagtype, afflist, affile->af_nosuggest)) | |
| 6793 flags |= WF_NOSUGGEST; | |
| 6794 return flags; | |
| 6795 } | |
| 6796 | |
| 6797 /* | |
| 339 | 6798 * Get the list of prefix IDs from the affix list "afflist". |
| 6799 * Used for PFXPOSTPONE. | |
| 483 | 6800 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL |
| 6801 * and return the number of affixes. | |
| 6802 */ | |
| 6803 static int | |
| 6804 get_pfxlist(affile, afflist, store_afflist) | |
| 339 | 6805 afffile_T *affile; |
| 6806 char_u *afflist; | |
| 483 | 6807 char_u *store_afflist; |
| 339 | 6808 { |
| 6809 char_u *p; | |
| 495 | 6810 char_u *prevp; |
| 483 | 6811 int cnt = 0; |
| 495 | 6812 int id; |
| 6813 char_u key[AH_KEY_LEN]; | |
| 339 | 6814 hashitem_T *hi; |
| 6815 | |
| 495 | 6816 for (p = afflist; *p != NUL; ) |
| 6817 { | |
| 6818 prevp = p; | |
| 6819 if (get_affitem(affile->af_flagtype, &p) != 0) | |
| 6820 { | |
| 6821 /* A flag is a postponed prefix flag if it appears in "af_pref" | |
| 6822 * and it's ID is not zero. */ | |
| 6823 vim_strncpy(key, prevp, p - prevp); | |
| 6824 hi = hash_find(&affile->af_pref, key); | |
| 6825 if (!HASHITEM_EMPTY(hi)) | |
| 6826 { | |
| 6827 id = HI2AH(hi)->ah_newID; | |
| 6828 if (id != 0) | |
| 6829 store_afflist[cnt++] = id; | |
| 6830 } | |
| 6831 } | |
| 498 | 6832 if (affile->af_flagtype == AFT_NUM && *p == ',') |
| 495 | 6833 ++p; |
| 483 | 6834 } |
| 6835 | |
| 6836 store_afflist[cnt] = NUL; | |
| 6837 return cnt; | |
| 339 | 6838 } |
| 6839 | |
| 6840 /* | |
| 495 | 6841 * Get the list of compound IDs from the affix list "afflist" that are used |
| 6842 * for compound words. | |
| 483 | 6843 * Puts the flags in "store_afflist[]". |
| 6844 */ | |
| 6845 static void | |
| 495 | 6846 get_compflags(affile, afflist, store_afflist) |
| 6847 afffile_T *affile; | |
| 481 | 6848 char_u *afflist; |
| 483 | 6849 char_u *store_afflist; |
| 481 | 6850 { |
| 6851 char_u *p; | |
| 495 | 6852 char_u *prevp; |
| 483 | 6853 int cnt = 0; |
| 495 | 6854 char_u key[AH_KEY_LEN]; |
| 6855 hashitem_T *hi; | |
| 6856 | |
| 6857 for (p = afflist; *p != NUL; ) | |
| 6858 { | |
| 6859 prevp = p; | |
| 6860 if (get_affitem(affile->af_flagtype, &p) != 0) | |
| 6861 { | |
| 6862 /* A flag is a compound flag if it appears in "af_comp". */ | |
| 6863 vim_strncpy(key, prevp, p - prevp); | |
| 6864 hi = hash_find(&affile->af_comp, key); | |
| 6865 if (!HASHITEM_EMPTY(hi)) | |
| 6866 store_afflist[cnt++] = HI2CI(hi)->ci_newID; | |
| 6867 } | |
| 498 | 6868 if (affile->af_flagtype == AFT_NUM && *p == ',') |
| 495 | 6869 ++p; |
| 6870 } | |
| 483 | 6871 |
| 6872 store_afflist[cnt] = NUL; | |
| 481 | 6873 } |
| 6874 | |
| 6875 /* | |
| 300 | 6876 * Apply affixes to a word and store the resulting words. |
| 6877 * "ht" is the hashtable with affentry_T that need to be applied, either | |
| 6878 * prefixes or suffixes. | |
| 6879 * "xht", when not NULL, is the prefix hashtable, to be used additionally on | |
| 6880 * the resulting words for combining affixes. | |
| 6881 * | |
| 6882 * Returns FAIL when out of memory. | |
| 236 | 6883 */ |
| 300 | 6884 static int |
| 820 | 6885 store_aff_word(spin, word, afflist, affile, ht, xht, condit, flags, |
| 483 | 6886 pfxlist, pfxlen) |
| 474 | 6887 spellinfo_T *spin; /* spell info */ |
| 300 | 6888 char_u *word; /* basic word start */ |
| 6889 char_u *afflist; /* list of names of supported affixes */ | |
| 339 | 6890 afffile_T *affile; |
| 300 | 6891 hashtab_T *ht; |
| 6892 hashtab_T *xht; | |
| 820 | 6893 int condit; /* CONDIT_SUF et al. */ |
| 307 | 6894 int flags; /* flags for the word */ |
| 339 | 6895 char_u *pfxlist; /* list of prefix IDs */ |
| 483 | 6896 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest |
| 6897 * is compound flags */ | |
| 236 | 6898 { |
| 6899 int todo; | |
| 6900 hashitem_T *hi; | |
| 300 | 6901 affheader_T *ah; |
| 6902 affentry_T *ae; | |
| 6903 char_u newword[MAXWLEN]; | |
| 6904 int retval = OK; | |
| 820 | 6905 int i, j; |
| 300 | 6906 char_u *p; |
| 366 | 6907 int use_flags; |
| 390 | 6908 char_u *use_pfxlist; |
| 820 | 6909 int use_pfxlen; |
| 6910 int need_affix; | |
| 6911 char_u store_afflist[MAXWLEN]; | |
| 483 | 6912 char_u pfx_pfxlist[MAXWLEN]; |
| 6913 size_t wordlen = STRLEN(word); | |
| 820 | 6914 int use_condit; |
| 236 | 6915 |
| 835 | 6916 todo = (int)ht->ht_used; |
| 300 | 6917 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi) |
| 236 | 6918 { |
| 6919 if (!HASHITEM_EMPTY(hi)) | |
| 6920 { | |
| 6921 --todo; | |
| 300 | 6922 ah = HI2AH(hi); |
| 236 | 6923 |
| 300 | 6924 /* Check that the affix combines, if required, and that the word |
| 6925 * supports this affix. */ | |
| 820 | 6926 if (((condit & CONDIT_COMB) == 0 || ah->ah_combine) |
| 6927 && flag_in_afflist(affile->af_flagtype, afflist, | |
| 6928 ah->ah_flag)) | |
| 236 | 6929 { |
| 300 | 6930 /* Loop over all affix entries with this name. */ |
| 6931 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next) | |
| 236 | 6932 { |
| 300 | 6933 /* Check the condition. It's not logical to match case |
| 6934 * here, but it is required for compatibility with | |
| 339 | 6935 * Myspell. |
| 481 | 6936 * Another requirement from Myspell is that the chop |
| 6937 * string is shorter than the word itself. | |
| 339 | 6938 * For prefixes, when "PFXPOSTPONE" was used, only do |
| 820 | 6939 * prefixes with a chop string and/or flags. |
| 6940 * When a previously added affix had CIRCUMFIX this one | |
| 6941 * must have it too, if it had not then this one must not | |
| 6942 * have one either. */ | |
| 339 | 6943 if ((xht != NULL || !affile->af_pfxpostpone |
| 809 | 6944 || ae->ae_chop != NULL |
| 6945 || ae->ae_flags != NULL) | |
| 481 | 6946 && (ae->ae_chop == NULL |
| 6947 || STRLEN(ae->ae_chop) < wordlen) | |
| 339 | 6948 && (ae->ae_prog == NULL |
| 6375 | 6949 || vim_regexec_prog(&ae->ae_prog, FALSE, |
| 6950 word, (colnr_T)0)) | |
| 820 | 6951 && (((condit & CONDIT_CFIX) == 0) |
| 6952 == ((condit & CONDIT_AFF) == 0 | |
| 6953 || ae->ae_flags == NULL | |
| 6954 || !flag_in_afflist(affile->af_flagtype, | |
| 6955 ae->ae_flags, affile->af_circumfix)))) | |
| 300 | 6956 { |
| 6957 /* Match. Remove the chop and add the affix. */ | |
| 6958 if (xht == NULL) | |
| 240 | 6959 { |
| 300 | 6960 /* prefix: chop/add at the start of the word */ |
| 6961 if (ae->ae_add == NULL) | |
| 6962 *newword = NUL; | |
| 6963 else | |
| 2768 | 6964 vim_strncpy(newword, ae->ae_add, MAXWLEN - 1); |
| 300 | 6965 p = word; |
| 6966 if (ae->ae_chop != NULL) | |
| 310 | 6967 { |
| 300 | 6968 /* Skip chop string. */ |
| 310 | 6969 #ifdef FEAT_MBYTE |
| 6970 if (has_mbyte) | |
| 324 | 6971 { |
| 310 | 6972 i = mb_charlen(ae->ae_chop); |
| 324 | 6973 for ( ; i > 0; --i) |
| 6974 mb_ptr_adv(p); | |
| 6975 } | |
| 310 | 6976 else |
| 6977 #endif | |
| 324 | 6978 p += STRLEN(ae->ae_chop); |
| 310 | 6979 } |
| 300 | 6980 STRCAT(newword, p); |
| 6981 } | |
| 6982 else | |
| 6983 { | |
| 6984 /* suffix: chop/add at the end of the word */ | |
| 2768 | 6985 vim_strncpy(newword, word, MAXWLEN - 1); |
| 300 | 6986 if (ae->ae_chop != NULL) |
| 6987 { | |
| 6988 /* Remove chop string. */ | |
| 6989 p = newword + STRLEN(newword); | |
| 835 | 6990 i = (int)MB_CHARLEN(ae->ae_chop); |
| 310 | 6991 for ( ; i > 0; --i) |
| 300 | 6992 mb_ptr_back(newword, p); |
| 6993 *p = NUL; | |
| 6994 } | |
| 6995 if (ae->ae_add != NULL) | |
| 6996 STRCAT(newword, ae->ae_add); | |
| 240 | 6997 } |
| 6998 | |
| 820 | 6999 use_flags = flags; |
| 7000 use_pfxlist = pfxlist; | |
| 7001 use_pfxlen = pfxlen; | |
| 7002 need_affix = FALSE; | |
| 7003 use_condit = condit | CONDIT_COMB | CONDIT_AFF; | |
| 7004 if (ae->ae_flags != NULL) | |
| 7005 { | |
| 7006 /* Extract flags from the affix list. */ | |
| 7007 use_flags |= get_affix_flags(affile, ae->ae_flags); | |
| 7008 | |
| 7009 if (affile->af_needaffix != 0 && flag_in_afflist( | |
| 7010 affile->af_flagtype, ae->ae_flags, | |
| 7011 affile->af_needaffix)) | |
| 7012 need_affix = TRUE; | |
| 7013 | |
| 7014 /* When there is a CIRCUMFIX flag the other affix | |
| 7015 * must also have it and we don't add the word | |
| 7016 * with one affix. */ | |
| 7017 if (affile->af_circumfix != 0 && flag_in_afflist( | |
| 7018 affile->af_flagtype, ae->ae_flags, | |
| 7019 affile->af_circumfix)) | |
| 7020 { | |
| 7021 use_condit |= CONDIT_CFIX; | |
| 7022 if ((condit & CONDIT_CFIX) == 0) | |
| 7023 need_affix = TRUE; | |
| 7024 } | |
| 7025 | |
| 7026 if (affile->af_pfxpostpone | |
| 7027 || spin->si_compflags != NULL) | |
| 7028 { | |
| 7029 if (affile->af_pfxpostpone) | |
| 7030 /* Get prefix IDS from the affix list. */ | |
| 7031 use_pfxlen = get_pfxlist(affile, | |
| 7032 ae->ae_flags, store_afflist); | |
| 7033 else | |
| 7034 use_pfxlen = 0; | |
| 7035 use_pfxlist = store_afflist; | |
| 7036 | |
| 7037 /* Combine the prefix IDs. Avoid adding the | |
| 7038 * same ID twice. */ | |
| 7039 for (i = 0; i < pfxlen; ++i) | |
| 7040 { | |
| 7041 for (j = 0; j < use_pfxlen; ++j) | |
| 7042 if (pfxlist[i] == use_pfxlist[j]) | |
| 7043 break; | |
| 7044 if (j == use_pfxlen) | |
| 7045 use_pfxlist[use_pfxlen++] = pfxlist[i]; | |
| 7046 } | |
| 7047 | |
| 7048 if (spin->si_compflags != NULL) | |
| 7049 /* Get compound IDS from the affix list. */ | |
| 7050 get_compflags(affile, ae->ae_flags, | |
| 7051 use_pfxlist + use_pfxlen); | |
| 7052 | |
| 7053 /* Combine the list of compound flags. | |
| 7054 * Concatenate them to the prefix IDs list. | |
| 7055 * Avoid adding the same ID twice. */ | |
| 7056 for (i = pfxlen; pfxlist[i] != NUL; ++i) | |
| 7057 { | |
| 7058 for (j = use_pfxlen; | |
| 7059 use_pfxlist[j] != NUL; ++j) | |
| 7060 if (pfxlist[i] == use_pfxlist[j]) | |
| 7061 break; | |
| 7062 if (use_pfxlist[j] == NUL) | |
| 7063 { | |
| 7064 use_pfxlist[j++] = pfxlist[i]; | |
| 7065 use_pfxlist[j] = NUL; | |
| 7066 } | |
| 7067 } | |
| 7068 } | |
| 7069 } | |
| 483 | 7070 |
| 819 | 7071 /* Obey a "COMPOUNDFORBIDFLAG" of the affix: don't |
| 809 | 7072 * use the compound flags. */ |
| 822 | 7073 if (use_pfxlist != NULL && ae->ae_compforbid) |
| 483 | 7074 { |
| 820 | 7075 vim_strncpy(pfx_pfxlist, use_pfxlist, use_pfxlen); |
| 483 | 7076 use_pfxlist = pfx_pfxlist; |
| 7077 } | |
| 390 | 7078 |
| 7079 /* When there are postponed prefixes... */ | |
| 393 | 7080 if (spin->si_prefroot != NULL |
| 7081 && spin->si_prefroot->wn_sibling != NULL) | |
| 390 | 7082 { |
| 7083 /* ... add a flag to indicate an affix was used. */ | |
| 7084 use_flags |= WF_HAS_AFF; | |
| 7085 | |
| 7086 /* ... don't use a prefix list if combining | |
| 483 | 7087 * affixes is not allowed. But do use the |
| 7088 * compound flags after them. */ | |
| 827 | 7089 if (!ah->ah_combine && use_pfxlist != NULL) |
| 820 | 7090 use_pfxlist += use_pfxlen; |
| 390 | 7091 } |
| 366 | 7092 |
| 819 | 7093 /* When compounding is supported and there is no |
| 7094 * "COMPOUNDPERMITFLAG" then forbid compounding on the | |
| 7095 * side where the affix is applied. */ | |
| 822 | 7096 if (spin->si_compflags != NULL && !ae->ae_comppermit) |
| 819 | 7097 { |
| 7098 if (xht != NULL) | |
| 7099 use_flags |= WF_NOCOMPAFT; | |
| 7100 else | |
| 7101 use_flags |= WF_NOCOMPBEF; | |
| 7102 } | |
| 7103 | |
| 300 | 7104 /* Store the modified word. */ |
| 474 | 7105 if (store_word(spin, newword, use_flags, |
| 820 | 7106 spin->si_region, use_pfxlist, |
| 7107 need_affix) == FAIL) | |
| 300 | 7108 retval = FAIL; |
| 236 | 7109 |
| 820 | 7110 /* When added a prefix or a first suffix and the affix |
| 7111 * has flags may add a(nother) suffix. RECURSIVE! */ | |
| 7112 if ((condit & CONDIT_SUF) && ae->ae_flags != NULL) | |
| 7113 if (store_aff_word(spin, newword, ae->ae_flags, | |
| 7114 affile, &affile->af_suff, xht, | |
| 7115 use_condit & (xht == NULL | |
| 7116 ? ~0 : ~CONDIT_SUF), | |
| 7117 use_flags, use_pfxlist, pfxlen) == FAIL) | |
| 7118 retval = FAIL; | |
| 7119 | |
| 300 | 7120 /* When added a suffix and combining is allowed also |
| 820 | 7121 * try adding a prefix additionally. Both for the |
| 7122 * word flags and for the affix flags. RECURSIVE! */ | |
| 300 | 7123 if (xht != NULL && ah->ah_combine) |
| 820 | 7124 { |
| 7125 if (store_aff_word(spin, newword, | |
| 7126 afflist, affile, | |
| 7127 xht, NULL, use_condit, | |
| 7128 use_flags, use_pfxlist, | |
| 7129 pfxlen) == FAIL | |
| 7130 || (ae->ae_flags != NULL | |
| 7131 && store_aff_word(spin, newword, | |
| 7132 ae->ae_flags, affile, | |
| 7133 xht, NULL, use_condit, | |
| 7134 use_flags, use_pfxlist, | |
| 7135 pfxlen) == FAIL)) | |
| 300 | 7136 retval = FAIL; |
| 820 | 7137 } |
| 236 | 7138 } |
| 7139 } | |
| 7140 } | |
| 7141 } | |
| 7142 } | |
| 7143 | |
| 7144 return retval; | |
| 7145 } | |
| 7146 | |
| 7147 /* | |
| 300 | 7148 * Read a file with a list of words. |
| 236 | 7149 */ |
| 7150 static int | |
| 474 | 7151 spell_read_wordfile(spin, fname) |
| 7152 spellinfo_T *spin; | |
| 300 | 7153 char_u *fname; |
| 236 | 7154 { |
| 300 | 7155 FILE *fd; |
| 7156 long lnum = 0; | |
| 7157 char_u rline[MAXLINELEN]; | |
| 7158 char_u *line; | |
| 7159 char_u *pc = NULL; | |
| 381 | 7160 char_u *p; |
| 300 | 7161 int l; |
| 7162 int retval = OK; | |
| 7163 int did_word = FALSE; | |
| 7164 int non_ascii = 0; | |
| 307 | 7165 int flags; |
| 316 | 7166 int regionmask; |
| 236 | 7167 |
| 300 | 7168 /* |
| 7169 * Open the file. | |
| 7170 */ | |
| 310 | 7171 fd = mch_fopen((char *)fname, "r"); |
| 300 | 7172 if (fd == NULL) |
| 236 | 7173 { |
| 300 | 7174 EMSG2(_(e_notopen), fname); |
| 7175 return FAIL; | |
| 236 | 7176 } |
| 7177 | |
| 625 | 7178 vim_snprintf((char *)IObuff, IOSIZE, _("Reading word file %s ..."), fname); |
| 7179 spell_message(spin, IObuff); | |
| 300 | 7180 |
| 7181 /* | |
| 7182 * Read all the lines in the file one by one. | |
| 7183 */ | |
| 7184 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int) | |
| 7185 { | |
| 7186 line_breakcheck(); | |
| 7187 ++lnum; | |
| 7188 | |
| 7189 /* Skip comment lines. */ | |
| 7190 if (*rline == '#') | |
| 7191 continue; | |
| 7192 | |
| 7193 /* Remove CR, LF and white space from the end. */ | |
| 835 | 7194 l = (int)STRLEN(rline); |
| 300 | 7195 while (l > 0 && rline[l - 1] <= ' ') |
| 7196 --l; | |
| 7197 if (l == 0) | |
| 7198 continue; /* empty or blank line */ | |
| 7199 rline[l] = NUL; | |
| 7200 | |
| 859 | 7201 /* Convert from "/encoding={encoding}" to 'encoding' when needed. */ |
| 300 | 7202 vim_free(pc); |
| 310 | 7203 #ifdef FEAT_MBYTE |
| 300 | 7204 if (spin->si_conv.vc_type != CONV_NONE) |
| 7205 { | |
| 7206 pc = string_convert(&spin->si_conv, rline, NULL); | |
| 7207 if (pc == NULL) | |
| 7208 { | |
| 7209 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"), | |
| 7210 fname, lnum, rline); | |
| 7211 continue; | |
| 7212 } | |
| 7213 line = pc; | |
| 7214 } | |
| 7215 else | |
| 310 | 7216 #endif |
| 300 | 7217 { |
| 7218 pc = NULL; | |
| 7219 line = rline; | |
| 7220 } | |
| 7221 | |
| 307 | 7222 if (*line == '/') |
| 300 | 7223 { |
| 307 | 7224 ++line; |
| 7225 if (STRNCMP(line, "encoding=", 9) == 0) | |
| 300 | 7226 { |
| 7227 if (spin->si_conv.vc_type != CONV_NONE) | |
| 316 | 7228 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"), |
| 7229 fname, lnum, line - 1); | |
| 300 | 7230 else if (did_word) |
| 316 | 7231 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"), |
| 7232 fname, lnum, line - 1); | |
| 300 | 7233 else |
| 7234 { | |
| 310 | 7235 #ifdef FEAT_MBYTE |
| 7236 char_u *enc; | |
| 7237 | |
| 300 | 7238 /* Setup for conversion to 'encoding'. */ |
| 859 | 7239 line += 9; |
| 316 | 7240 enc = enc_canonize(line); |
| 300 | 7241 if (enc != NULL && !spin->si_ascii |
| 7242 && convert_setup(&spin->si_conv, enc, | |
| 7243 p_enc) == FAIL) | |
| 7244 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"), | |
| 316 | 7245 fname, line, p_enc); |
| 300 | 7246 vim_free(enc); |
| 375 | 7247 spin->si_conv.vc_fail = TRUE; |
| 310 | 7248 #else |
| 7249 smsg((char_u *)_("Conversion in %s not supported"), fname); | |
| 7250 #endif | |
| 300 | 7251 } |
| 307 | 7252 continue; |
| 300 | 7253 } |
| 307 | 7254 |
| 316 | 7255 if (STRNCMP(line, "regions=", 8) == 0) |
| 7256 { | |
| 7257 if (spin->si_region_count > 1) | |
| 7258 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"), | |
| 7259 fname, lnum, line); | |
| 7260 else | |
| 7261 { | |
| 7262 line += 8; | |
| 7263 if (STRLEN(line) > 16) | |
| 7264 smsg((char_u *)_("Too many regions in %s line %d: %s"), | |
| 7265 fname, lnum, line); | |
| 7266 else | |
| 7267 { | |
| 835 | 7268 spin->si_region_count = (int)STRLEN(line) / 2; |
| 316 | 7269 STRCPY(spin->si_region_name, line); |
| 389 | 7270 |
| 7271 /* Adjust the mask for a word valid in all regions. */ | |
| 7272 spin->si_region = (1 << spin->si_region_count) - 1; | |
| 316 | 7273 } |
| 7274 } | |
| 7275 continue; | |
| 7276 } | |
| 7277 | |
| 381 | 7278 smsg((char_u *)_("/ line ignored in %s line %d: %s"), |
| 7279 fname, lnum, line - 1); | |
| 7280 continue; | |
| 7281 } | |
| 7282 | |
| 7283 flags = 0; | |
| 7284 regionmask = spin->si_region; | |
| 7285 | |
| 7286 /* Check for flags and region after a slash. */ | |
| 7287 p = vim_strchr(line, '/'); | |
| 7288 if (p != NULL) | |
| 7289 { | |
| 7290 *p++ = NUL; | |
| 7291 while (*p != NUL) | |
| 7292 { | |
| 7293 if (*p == '=') /* keep-case word */ | |
| 389 | 7294 flags |= WF_KEEPCAP | WF_FIXCAP; |
| 381 | 7295 else if (*p == '!') /* Bad, bad, wicked word. */ |
| 7296 flags |= WF_BANNED; | |
| 7297 else if (*p == '?') /* Rare word. */ | |
| 7298 flags |= WF_RARE; | |
| 7299 else if (VIM_ISDIGIT(*p)) /* region number(s) */ | |
| 316 | 7300 { |
| 381 | 7301 if ((flags & WF_REGION) == 0) /* first one */ |
| 7302 regionmask = 0; | |
| 7303 flags |= WF_REGION; | |
| 7304 | |
| 7305 l = *p - '0'; | |
| 316 | 7306 if (l > spin->si_region_count) |
| 7307 { | |
| 7308 smsg((char_u *)_("Invalid region nr in %s line %d: %s"), | |
| 381 | 7309 fname, lnum, p); |
| 316 | 7310 break; |
| 7311 } | |
| 7312 regionmask |= 1 << (l - 1); | |
| 7313 } | |
| 381 | 7314 else |
| 7315 { | |
| 7316 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"), | |
| 7317 fname, lnum, p); | |
| 7318 break; | |
| 7319 } | |
| 7320 ++p; | |
| 307 | 7321 } |
| 300 | 7322 } |
| 7323 | |
| 7324 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */ | |
| 7325 if (spin->si_ascii && has_non_ascii(line)) | |
| 7326 { | |
| 7327 ++non_ascii; | |
| 7328 continue; | |
| 7329 } | |
| 7330 | |
| 7331 /* Normal word: store it. */ | |
| 483 | 7332 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL) |
| 300 | 7333 { |
| 7334 retval = FAIL; | |
| 7335 break; | |
| 7336 } | |
| 7337 did_word = TRUE; | |
| 7338 } | |
| 7339 | |
| 7340 vim_free(pc); | |
| 7341 fclose(fd); | |
| 7342 | |
| 625 | 7343 if (spin->si_ascii && non_ascii > 0) |
| 7344 { | |
| 7345 vim_snprintf((char *)IObuff, IOSIZE, | |
| 7346 _("Ignored %d words with non-ASCII characters"), non_ascii); | |
| 7347 spell_message(spin, IObuff); | |
| 7348 } | |
| 7349 | |
| 300 | 7350 return retval; |
| 236 | 7351 } |
| 7352 | |
| 7353 /* | |
| 300 | 7354 * Get part of an sblock_T, "len" bytes long. |
| 474 | 7355 * This avoids calling free() for every little struct we use (and keeping |
| 7356 * track of them). | |
| 300 | 7357 * The memory is cleared to all zeros. |
| 7358 * Returns NULL when out of memory. | |
| 7359 */ | |
| 7360 static void * | |
| 474 | 7361 getroom(spin, len, align) |
| 7362 spellinfo_T *spin; | |
| 447 | 7363 size_t len; /* length needed */ |
| 7364 int align; /* align for pointer */ | |
| 300 | 7365 { |
| 7366 char_u *p; | |
| 474 | 7367 sblock_T *bl = spin->si_blocks; |
| 300 | 7368 |
| 447 | 7369 if (align && bl != NULL) |
| 7370 /* Round size up for alignment. On some systems structures need to be | |
| 7371 * aligned to the size of a pointer (e.g., SPARC). */ | |
| 7372 bl->sb_used = (bl->sb_used + sizeof(char *) - 1) | |
| 7373 & ~(sizeof(char *) - 1); | |
| 7374 | |
| 300 | 7375 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE) |
| 7376 { | |
| 2691 | 7377 if (len >= SBLOCKSIZE) |
| 7378 bl = NULL; | |
| 7379 else | |
| 7380 /* Allocate a block of memory. It is not freed until much later. */ | |
| 7381 bl = (sblock_T *)alloc_clear( | |
| 7382 (unsigned)(sizeof(sblock_T) + SBLOCKSIZE)); | |
| 300 | 7383 if (bl == NULL) |
| 2691 | 7384 { |
| 7385 if (!spin->si_did_emsg) | |
| 7386 { | |
| 7387 EMSG(_("E845: Insufficient memory, word list will be incomplete")); | |
| 7388 spin->si_did_emsg = TRUE; | |
| 7389 } | |
| 300 | 7390 return NULL; |
| 2691 | 7391 } |
| 474 | 7392 bl->sb_next = spin->si_blocks; |
| 7393 spin->si_blocks = bl; | |
| 300 | 7394 bl->sb_used = 0; |
| 482 | 7395 ++spin->si_blocks_cnt; |
| 300 | 7396 } |
| 7397 | |
| 7398 p = bl->sb_data + bl->sb_used; | |
| 835 | 7399 bl->sb_used += (int)len; |
| 300 | 7400 |
| 7401 return p; | |
| 7402 } | |
| 7403 | |
| 7404 /* | |
| 7405 * Make a copy of a string into memory allocated with getroom(). | |
| 2691 | 7406 * Returns NULL when out of memory. |
| 300 | 7407 */ |
| 7408 static char_u * | |
| 474 | 7409 getroom_save(spin, s) |
| 7410 spellinfo_T *spin; | |
| 300 | 7411 char_u *s; |
| 7412 { | |
| 7413 char_u *sc; | |
| 7414 | |
| 474 | 7415 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE); |
| 300 | 7416 if (sc != NULL) |
| 7417 STRCPY(sc, s); | |
| 7418 return sc; | |
| 7419 } | |
| 7420 | |
| 7421 | |
| 7422 /* | |
| 7423 * Free the list of allocated sblock_T. | |
| 236 | 7424 */ |
| 7425 static void | |
| 300 | 7426 free_blocks(bl) |
| 7427 sblock_T *bl; | |
| 236 | 7428 { |
| 300 | 7429 sblock_T *next; |
| 236 | 7430 |
| 300 | 7431 while (bl != NULL) |
| 236 | 7432 { |
| 300 | 7433 next = bl->sb_next; |
| 7434 vim_free(bl); | |
| 7435 bl = next; | |
| 236 | 7436 } |
| 7437 } | |
| 7438 | |
| 7439 /* | |
| 300 | 7440 * Allocate the root of a word tree. |
| 2691 | 7441 * Returns NULL when out of memory. |
| 236 | 7442 */ |
| 300 | 7443 static wordnode_T * |
| 474 | 7444 wordtree_alloc(spin) |
| 7445 spellinfo_T *spin; | |
| 7446 { | |
| 7447 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE); | |
| 236 | 7448 } |
| 7449 | |
| 7450 /* | |
| 300 | 7451 * Store a word in the tree(s). |
| 390 | 7452 * Always store it in the case-folded tree. For a keep-case word this is |
| 7453 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and | |
| 7454 * used to find suggestions. | |
| 300 | 7455 * For a keep-case word also store it in the keep-case tree. |
| 482 | 7456 * When "pfxlist" is not NULL store the word for each postponed prefix ID and |
| 7457 * compound flag. | |
| 236 | 7458 */ |
| 7459 static int | |
| 483 | 7460 store_word(spin, word, flags, region, pfxlist, need_affix) |
| 474 | 7461 spellinfo_T *spin; |
| 300 | 7462 char_u *word; |
| 307 | 7463 int flags; /* extra flags, WF_BANNED */ |
| 316 | 7464 int region; /* supported region(s) */ |
| 339 | 7465 char_u *pfxlist; /* list of prefix IDs or NULL */ |
| 483 | 7466 int need_affix; /* only store word with affix ID */ |
| 236 | 7467 { |
| 835 | 7468 int len = (int)STRLEN(word); |
| 300 | 7469 int ct = captype(word, word + len); |
| 7470 char_u foldword[MAXWLEN]; | |
| 339 | 7471 int res = OK; |
| 7472 char_u *p; | |
| 236 | 7473 |
| 323 | 7474 (void)spell_casefold(word, len, foldword, MAXWLEN); |
| 339 | 7475 for (p = pfxlist; res == OK; ++p) |
| 7476 { | |
| 483 | 7477 if (!need_affix || (p != NULL && *p != NUL)) |
| 7478 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags, | |
| 474 | 7479 region, p == NULL ? 0 : *p); |
| 339 | 7480 if (p == NULL || *p == NUL) |
| 7481 break; | |
| 7482 } | |
| 334 | 7483 ++spin->si_foldwcount; |
| 307 | 7484 |
| 474 | 7485 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP))) |
| 334 | 7486 { |
| 339 | 7487 for (p = pfxlist; res == OK; ++p) |
| 7488 { | |
| 483 | 7489 if (!need_affix || (p != NULL && *p != NUL)) |
| 7490 res = tree_add_word(spin, word, spin->si_keeproot, flags, | |
| 474 | 7491 region, p == NULL ? 0 : *p); |
| 339 | 7492 if (p == NULL || *p == NUL) |
| 7493 break; | |
| 7494 } | |
| 334 | 7495 ++spin->si_keepwcount; |
| 7496 } | |
| 300 | 7497 return res; |
| 236 | 7498 } |
| 7499 | |
| 7500 /* | |
| 300 | 7501 * Add word "word" to a word tree at "root". |
| 625 | 7502 * When "flags" < 0 we are adding to the prefix tree where "flags" is used for |
| 366 | 7503 * "rare" and "region" is the condition nr. |
| 255 | 7504 * Returns FAIL when out of memory. |
| 236 | 7505 */ |
| 255 | 7506 static int |
| 481 | 7507 tree_add_word(spin, word, root, flags, region, affixID) |
| 474 | 7508 spellinfo_T *spin; |
| 300 | 7509 char_u *word; |
| 7510 wordnode_T *root; | |
| 7511 int flags; | |
| 7512 int region; | |
| 481 | 7513 int affixID; |
| 236 | 7514 { |
| 300 | 7515 wordnode_T *node = root; |
| 7516 wordnode_T *np; | |
| 469 | 7517 wordnode_T *copyp, **copyprev; |
| 300 | 7518 wordnode_T **prev = NULL; |
| 7519 int i; | |
| 255 | 7520 |
| 300 | 7521 /* Add each byte of the word to the tree, including the NUL at the end. */ |
| 7522 for (i = 0; ; ++i) | |
| 255 | 7523 { |
| 469 | 7524 /* When there is more than one reference to this node we need to make |
| 7525 * a copy, so that we can modify it. Copy the whole list of siblings | |
| 7526 * (we don't optimize for a partly shared list of siblings). */ | |
| 7527 if (node != NULL && node->wn_refs > 1) | |
| 7528 { | |
| 7529 --node->wn_refs; | |
| 7530 copyprev = prev; | |
| 7531 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling) | |
| 7532 { | |
| 7533 /* Allocate a new node and copy the info. */ | |
| 474 | 7534 np = get_wordnode(spin); |
| 469 | 7535 if (np == NULL) |
| 7536 return FAIL; | |
| 7537 np->wn_child = copyp->wn_child; | |
| 7538 if (np->wn_child != NULL) | |
| 7539 ++np->wn_child->wn_refs; /* child gets extra ref */ | |
| 7540 np->wn_byte = copyp->wn_byte; | |
| 7541 if (np->wn_byte == NUL) | |
| 7542 { | |
| 7543 np->wn_flags = copyp->wn_flags; | |
| 7544 np->wn_region = copyp->wn_region; | |
| 481 | 7545 np->wn_affixID = copyp->wn_affixID; |
| 469 | 7546 } |
| 7547 | |
| 7548 /* Link the new node in the list, there will be one ref. */ | |
| 7549 np->wn_refs = 1; | |
| 840 | 7550 if (copyprev != NULL) |
| 7551 *copyprev = np; | |
| 469 | 7552 copyprev = &np->wn_sibling; |
| 7553 | |
| 7554 /* Let "node" point to the head of the copied list. */ | |
| 7555 if (copyp == node) | |
| 7556 node = np; | |
| 7557 } | |
| 7558 } | |
| 7559 | |
| 300 | 7560 /* Look for the sibling that has the same character. They are sorted |
| 7561 * on byte value, thus stop searching when a sibling is found with a | |
| 339 | 7562 * higher byte value. For zero bytes (end of word) the sorting is |
| 481 | 7563 * done on flags and then on affixID. */ |
| 339 | 7564 while (node != NULL |
| 7565 && (node->wn_byte < word[i] | |
| 7566 || (node->wn_byte == NUL | |
| 7567 && (flags < 0 | |
| 625 | 7568 ? node->wn_affixID < (unsigned)affixID |
| 7569 : (node->wn_flags < (unsigned)(flags & WN_MASK) | |
| 390 | 7570 || (node->wn_flags == (flags & WN_MASK) |
| 625 | 7571 && (spin->si_sugtree |
| 7572 ? (node->wn_region & 0xffff) < region | |
| 7573 : node->wn_affixID | |
| 7574 < (unsigned)affixID))))))) | |
| 236 | 7575 { |
| 300 | 7576 prev = &node->wn_sibling; |
| 7577 node = *prev; | |
| 236 | 7578 } |
| 339 | 7579 if (node == NULL |
| 7580 || node->wn_byte != word[i] | |
| 7581 || (word[i] == NUL | |
| 7582 && (flags < 0 | |
| 625 | 7583 || spin->si_sugtree |
| 390 | 7584 || node->wn_flags != (flags & WN_MASK) |
| 481 | 7585 || node->wn_affixID != affixID))) |
| 255 | 7586 { |
| 300 | 7587 /* Allocate a new node. */ |
| 474 | 7588 np = get_wordnode(spin); |
| 300 | 7589 if (np == NULL) |
| 7590 return FAIL; | |
| 7591 np->wn_byte = word[i]; | |
| 469 | 7592 |
| 7593 /* If "node" is NULL this is a new child or the end of the sibling | |
| 7594 * list: ref count is one. Otherwise use ref count of sibling and | |
| 7595 * make ref count of sibling one (matters when inserting in front | |
| 7596 * of the list of siblings). */ | |
| 7597 if (node == NULL) | |
| 7598 np->wn_refs = 1; | |
| 7599 else | |
| 7600 { | |
| 7601 np->wn_refs = node->wn_refs; | |
| 7602 node->wn_refs = 1; | |
| 7603 } | |
|
2299
92e31a1b0c2f
Fix crash in generating spell .sug file.
Bram Moolenaar <bram@vim.org>
parents:
2250
diff
changeset
|
7604 if (prev != NULL) |
|
92e31a1b0c2f
Fix crash in generating spell .sug file.
Bram Moolenaar <bram@vim.org>
parents:
2250
diff
changeset
|
7605 *prev = np; |
| 300 | 7606 np->wn_sibling = node; |
| 7607 node = np; | |
| 255 | 7608 } |
| 300 | 7609 |
| 7610 if (word[i] == NUL) | |
| 7611 { | |
| 7612 node->wn_flags = flags; | |
| 7613 node->wn_region |= region; | |
| 481 | 7614 node->wn_affixID = affixID; |
| 300 | 7615 break; |
| 7616 } | |
| 7617 prev = &node->wn_child; | |
| 7618 node = *prev; | |
| 255 | 7619 } |
| 469 | 7620 #ifdef SPELL_PRINTTREE |
| 7621 smsg("Added \"%s\"", word); | |
| 7622 spell_print_tree(root->wn_sibling); | |
| 7623 #endif | |
| 7624 | |
| 474 | 7625 /* count nr of words added since last message */ |
| 7626 ++spin->si_msg_count; | |
| 7627 | |
| 482 | 7628 if (spin->si_compress_cnt > 1) |
| 7629 { | |
| 7630 if (--spin->si_compress_cnt == 1) | |
| 7631 /* Did enough words to lower the block count limit. */ | |
| 483 | 7632 spin->si_blocks_cnt += compress_inc; |
| 482 | 7633 } |
| 7634 | |
| 469 | 7635 /* |
| 482 | 7636 * When we have allocated lots of memory we need to compress the word tree |
| 7637 * to free up some room. But compression is slow, and we might actually | |
| 7638 * need that room, thus only compress in the following situations: | |
| 7639 * 1. When not compressed before (si_compress_cnt == 0): when using | |
| 483 | 7640 * "compress_start" blocks. |
| 7641 * 2. When compressed before and used "compress_inc" blocks before | |
| 7642 * adding "compress_added" words (si_compress_cnt > 1). | |
| 7643 * 3. When compressed before, added "compress_added" words | |
| 482 | 7644 * (si_compress_cnt == 1) and the number of free nodes drops below the |
| 7645 * maximum word length. | |
| 469 | 7646 */ |
| 482 | 7647 #ifndef SPELL_PRINTTREE |
| 7648 if (spin->si_compress_cnt == 1 | |
| 7649 ? spin->si_free_count < MAXWLEN | |
| 483 | 7650 : spin->si_blocks_cnt >= compress_start) |
| 482 | 7651 #endif |
| 7652 { | |
| 7653 /* Decrement the block counter. The effect is that we compress again | |
| 483 | 7654 * when the freed up room has been used and another "compress_inc" |
| 7655 * blocks have been allocated. Unless "compress_added" words have | |
| 482 | 7656 * been added, then the limit is put back again. */ |
| 483 | 7657 spin->si_blocks_cnt -= compress_inc; |
| 7658 spin->si_compress_cnt = compress_added; | |
| 482 | 7659 |
| 474 | 7660 if (spin->si_verbose) |
| 7661 { | |
| 7662 msg_start(); | |
| 7663 msg_puts((char_u *)_(msg_compressing)); | |
| 7664 msg_clr_eos(); | |
| 7665 msg_didout = FALSE; | |
| 7666 msg_col = 0; | |
| 7667 out_flush(); | |
| 7668 } | |
| 482 | 7669 |
| 7670 /* Compress both trees. Either they both have many nodes, which makes | |
| 7671 * compression useful, or one of them is small, which means | |
| 4352 | 7672 * compression goes fast. But when filling the soundfold word tree |
| 625 | 7673 * there is no keep-case tree. */ |
| 482 | 7674 wordtree_compress(spin, spin->si_foldroot); |
| 625 | 7675 if (affixID >= 0) |
| 7676 wordtree_compress(spin, spin->si_keeproot); | |
| 469 | 7677 } |
| 255 | 7678 |
| 7679 return OK; | |
| 236 | 7680 } |
| 7681 | |
| 469 | 7682 /* |
| 483 | 7683 * Check the 'mkspellmem' option. Return FAIL if it's wrong. |
| 7684 * Sets "sps_flags". | |
| 7685 */ | |
| 7686 int | |
| 7687 spell_check_msm() | |
| 7688 { | |
| 7689 char_u *p = p_msm; | |
| 7690 long start = 0; | |
| 944 | 7691 long incr = 0; |
| 483 | 7692 long added = 0; |
| 7693 | |
| 7694 if (!VIM_ISDIGIT(*p)) | |
| 7695 return FAIL; | |
| 7696 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/ | |
| 7697 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102); | |
| 7698 if (*p != ',') | |
| 7699 return FAIL; | |
| 7700 ++p; | |
| 7701 if (!VIM_ISDIGIT(*p)) | |
| 7702 return FAIL; | |
| 944 | 7703 incr = (getdigits(&p) * 102) / (SBLOCKSIZE / 10); |
| 483 | 7704 if (*p != ',') |
| 7705 return FAIL; | |
| 7706 ++p; | |
| 7707 if (!VIM_ISDIGIT(*p)) | |
| 7708 return FAIL; | |
| 7709 added = getdigits(&p) * 1024; | |
| 7710 if (*p != NUL) | |
| 7711 return FAIL; | |
| 7712 | |
| 944 | 7713 if (start == 0 || incr == 0 || added == 0 || incr > start) |
| 483 | 7714 return FAIL; |
| 7715 | |
| 7716 compress_start = start; | |
| 944 | 7717 compress_inc = incr; |
| 483 | 7718 compress_added = added; |
| 7719 return OK; | |
| 7720 } | |
| 7721 | |
| 7722 | |
| 7723 /* | |
| 469 | 7724 * Get a wordnode_T, either from the list of previously freed nodes or |
| 7725 * allocate a new one. | |
| 2691 | 7726 * Returns NULL when out of memory. |
| 469 | 7727 */ |
| 7728 static wordnode_T * | |
| 474 | 7729 get_wordnode(spin) |
| 7730 spellinfo_T *spin; | |
| 469 | 7731 { |
| 7732 wordnode_T *n; | |
| 7733 | |
| 474 | 7734 if (spin->si_first_free == NULL) |
| 7735 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE); | |
| 469 | 7736 else |
| 7737 { | |
| 474 | 7738 n = spin->si_first_free; |
| 7739 spin->si_first_free = n->wn_child; | |
| 469 | 7740 vim_memset(n, 0, sizeof(wordnode_T)); |
| 482 | 7741 --spin->si_free_count; |
| 469 | 7742 } |
| 7743 #ifdef SPELL_PRINTTREE | |
| 2691 | 7744 if (n != NULL) |
| 7745 n->wn_nr = ++spin->si_wordnode_nr; | |
| 469 | 7746 #endif |
| 7747 return n; | |
| 7748 } | |
| 7749 | |
| 7750 /* | |
| 474 | 7751 * Decrement the reference count on a node (which is the head of a list of |
| 7752 * siblings). If the reference count becomes zero free the node and its | |
| 7753 * siblings. | |
| 625 | 7754 * Returns the number of nodes actually freed. |
| 7755 */ | |
| 7756 static int | |
| 474 | 7757 deref_wordnode(spin, node) |
| 7758 spellinfo_T *spin; | |
| 7759 wordnode_T *node; | |
| 7760 { | |
| 625 | 7761 wordnode_T *np; |
| 7762 int cnt = 0; | |
| 474 | 7763 |
| 7764 if (--node->wn_refs == 0) | |
| 625 | 7765 { |
| 474 | 7766 for (np = node; np != NULL; np = np->wn_sibling) |
| 7767 { | |
| 7768 if (np->wn_child != NULL) | |
| 625 | 7769 cnt += deref_wordnode(spin, np->wn_child); |
| 474 | 7770 free_wordnode(spin, np); |
| 625 | 7771 ++cnt; |
| 7772 } | |
| 7773 ++cnt; /* length field */ | |
| 7774 } | |
| 7775 return cnt; | |
| 474 | 7776 } |
| 7777 | |
| 7778 /* | |
| 469 | 7779 * Free a wordnode_T for re-use later. |
| 474 | 7780 * Only the "wn_child" field becomes invalid. |
| 469 | 7781 */ |
| 7782 static void | |
| 474 | 7783 free_wordnode(spin, n) |
| 7784 spellinfo_T *spin; | |
| 7785 wordnode_T *n; | |
| 7786 { | |
| 7787 n->wn_child = spin->si_first_free; | |
| 7788 spin->si_first_free = n; | |
| 482 | 7789 ++spin->si_free_count; |
| 469 | 7790 } |
| 7791 | |
| 236 | 7792 /* |
| 300 | 7793 * Compress a tree: find tails that are identical and can be shared. |
| 7794 */ | |
| 7795 static void | |
| 474 | 7796 wordtree_compress(spin, root) |
| 7797 spellinfo_T *spin; | |
| 300 | 7798 wordnode_T *root; |
| 7799 { | |
| 7800 hashtab_T ht; | |
| 7801 int n; | |
| 7802 int tot = 0; | |
| 469 | 7803 int perc; |
| 300 | 7804 |
| 474 | 7805 /* Skip the root itself, it's not actually used. The first sibling is the |
| 7806 * start of the tree. */ | |
| 7807 if (root->wn_sibling != NULL) | |
| 300 | 7808 { |
| 7809 hash_init(&ht); | |
| 474 | 7810 n = node_compress(spin, root->wn_sibling, &ht, &tot); |
| 469 | 7811 |
| 7812 #ifndef SPELL_PRINTTREE | |
| 310 | 7813 if (spin->si_verbose || p_verbose > 2) |
| 469 | 7814 #endif |
| 310 | 7815 { |
| 469 | 7816 if (tot > 1000000) |
| 7817 perc = (tot - n) / (tot / 100); | |
| 482 | 7818 else if (tot == 0) |
| 7819 perc = 0; | |
| 469 | 7820 else |
| 7821 perc = (tot - n) * 100 / tot; | |
| 625 | 7822 vim_snprintf((char *)IObuff, IOSIZE, |
| 7823 _("Compressed %d of %d nodes; %d (%d%%) remaining"), | |
| 7824 n, tot, tot - n, perc); | |
| 7825 spell_message(spin, IObuff); | |
| 310 | 7826 } |
| 469 | 7827 #ifdef SPELL_PRINTTREE |
| 474 | 7828 spell_print_tree(root->wn_sibling); |
| 469 | 7829 #endif |
| 300 | 7830 hash_clear(&ht); |
| 7831 } | |
| 7832 } | |
| 7833 | |
| 7834 /* | |
| 7835 * Compress a node, its siblings and its children, depth first. | |
| 7836 * Returns the number of compressed nodes. | |
| 236 | 7837 */ |
| 255 | 7838 static int |
| 474 | 7839 node_compress(spin, node, ht, tot) |
| 7840 spellinfo_T *spin; | |
| 300 | 7841 wordnode_T *node; |
| 7842 hashtab_T *ht; | |
| 7843 int *tot; /* total count of nodes before compressing, | |
| 7844 incremented while going through the tree */ | |
| 236 | 7845 { |
| 300 | 7846 wordnode_T *np; |
| 7847 wordnode_T *tp; | |
| 7848 wordnode_T *child; | |
| 7849 hash_T hash; | |
| 236 | 7850 hashitem_T *hi; |
| 300 | 7851 int len = 0; |
| 7852 unsigned nr, n; | |
| 7853 int compressed = 0; | |
| 236 | 7854 |
| 300 | 7855 /* |
| 7856 * Go through the list of siblings. Compress each child and then try | |
| 7857 * finding an identical child to replace it. | |
| 7858 * Note that with "child" we mean not just the node that is pointed to, | |
| 625 | 7859 * but the whole list of siblings of which the child node is the first. |
| 300 | 7860 */ |
| 482 | 7861 for (np = node; np != NULL && !got_int; np = np->wn_sibling) |
| 236 | 7862 { |
| 300 | 7863 ++len; |
| 7864 if ((child = np->wn_child) != NULL) | |
| 7865 { | |
| 625 | 7866 /* Compress the child first. This fills hashkey. */ |
| 474 | 7867 compressed += node_compress(spin, child, ht, tot); |
| 300 | 7868 |
| 7869 /* Try to find an identical child. */ | |
| 346 | 7870 hash = hash_hash(child->wn_u1.hashkey); |
| 7871 hi = hash_lookup(ht, child->wn_u1.hashkey, hash); | |
| 300 | 7872 if (!HASHITEM_EMPTY(hi)) |
| 7873 { | |
| 625 | 7874 /* There are children we encountered before with a hash value |
| 7875 * identical to the current child. Now check if there is one | |
| 7876 * that is really identical. */ | |
| 346 | 7877 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next) |
| 300 | 7878 if (node_equal(child, tp)) |
| 7879 { | |
| 7880 /* Found one! Now use that child in place of the | |
| 469 | 7881 * current one. This means the current child and all |
| 7882 * its siblings is unlinked from the tree. */ | |
| 7883 ++tp->wn_refs; | |
| 625 | 7884 compressed += deref_wordnode(spin, child); |
| 300 | 7885 np->wn_child = tp; |
| 7886 break; | |
| 7887 } | |
| 7888 if (tp == NULL) | |
| 7889 { | |
| 7890 /* No other child with this hash value equals the child of | |
| 7891 * the node, add it to the linked list after the first | |
| 7892 * item. */ | |
| 7893 tp = HI2WN(hi); | |
| 346 | 7894 child->wn_u2.next = tp->wn_u2.next; |
| 7895 tp->wn_u2.next = child; | |
| 300 | 7896 } |
| 7897 } | |
| 7898 else | |
| 7899 /* No other child has this hash value, add it to the | |
| 7900 * hashtable. */ | |
| 346 | 7901 hash_add_item(ht, hi, child->wn_u1.hashkey, hash); |
| 300 | 7902 } |
| 236 | 7903 } |
| 625 | 7904 *tot += len + 1; /* add one for the node that stores the length */ |
| 300 | 7905 |
| 7906 /* | |
| 7907 * Make a hash key for the node and its siblings, so that we can quickly | |
| 7908 * find a lookalike node. This must be done after compressing the sibling | |
| 7909 * list, otherwise the hash key would become invalid by the compression. | |
| 7910 */ | |
| 346 | 7911 node->wn_u1.hashkey[0] = len; |
| 300 | 7912 nr = 0; |
| 7913 for (np = node; np != NULL; np = np->wn_sibling) | |
| 236 | 7914 { |
| 300 | 7915 if (np->wn_byte == NUL) |
| 481 | 7916 /* end node: use wn_flags, wn_region and wn_affixID */ |
| 7917 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16); | |
| 300 | 7918 else |
| 7919 /* byte node: use the byte value and the child pointer */ | |
| 835 | 7920 n = (unsigned)(np->wn_byte + ((long_u)np->wn_child << 8)); |
| 300 | 7921 nr = nr * 101 + n; |
| 236 | 7922 } |
| 300 | 7923 |
| 7924 /* Avoid NUL bytes, it terminates the hash key. */ | |
| 7925 n = nr & 0xff; | |
| 346 | 7926 node->wn_u1.hashkey[1] = n == 0 ? 1 : n; |
| 300 | 7927 n = (nr >> 8) & 0xff; |
| 346 | 7928 node->wn_u1.hashkey[2] = n == 0 ? 1 : n; |
| 300 | 7929 n = (nr >> 16) & 0xff; |
| 346 | 7930 node->wn_u1.hashkey[3] = n == 0 ? 1 : n; |
| 300 | 7931 n = (nr >> 24) & 0xff; |
| 346 | 7932 node->wn_u1.hashkey[4] = n == 0 ? 1 : n; |
| 7933 node->wn_u1.hashkey[5] = NUL; | |
| 300 | 7934 |
| 482 | 7935 /* Check for CTRL-C pressed now and then. */ |
| 7936 fast_breakcheck(); | |
| 7937 | |
| 300 | 7938 return compressed; |
| 7939 } | |
| 7940 | |
| 7941 /* | |
| 7942 * Return TRUE when two nodes have identical siblings and children. | |
| 7943 */ | |
| 7944 static int | |
| 7945 node_equal(n1, n2) | |
| 7946 wordnode_T *n1; | |
| 7947 wordnode_T *n2; | |
| 7948 { | |
| 7949 wordnode_T *p1; | |
| 7950 wordnode_T *p2; | |
| 7951 | |
| 7952 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL; | |
| 7953 p1 = p1->wn_sibling, p2 = p2->wn_sibling) | |
| 7954 if (p1->wn_byte != p2->wn_byte | |
| 7955 || (p1->wn_byte == NUL | |
| 7956 ? (p1->wn_flags != p2->wn_flags | |
| 339 | 7957 || p1->wn_region != p2->wn_region |
| 481 | 7958 || p1->wn_affixID != p2->wn_affixID) |
| 300 | 7959 : (p1->wn_child != p2->wn_child))) |
| 7960 break; | |
| 7961 | |
| 7962 return p1 == NULL && p2 == NULL; | |
| 236 | 7963 } |
| 7964 | |
| 323 | 7965 static int |
| 7966 #ifdef __BORLANDC__ | |
| 7967 _RTLENTRYF | |
| 7968 #endif | |
| 7969 rep_compare __ARGS((const void *s1, const void *s2)); | |
| 7970 | |
| 7971 /* | |
| 7972 * Function given to qsort() to sort the REP items on "from" string. | |
| 7973 */ | |
| 7974 static int | |
| 7975 #ifdef __BORLANDC__ | |
| 7976 _RTLENTRYF | |
| 7977 #endif | |
| 7978 rep_compare(s1, s2) | |
| 7979 const void *s1; | |
| 7980 const void *s2; | |
| 7981 { | |
| 7982 fromto_T *p1 = (fromto_T *)s1; | |
| 7983 fromto_T *p2 = (fromto_T *)s2; | |
| 7984 | |
| 7985 return STRCMP(p1->ft_from, p2->ft_from); | |
| 7986 } | |
| 7987 | |
| 236 | 7988 /* |
| 483 | 7989 * Write the Vim .spl file "fname". |
| 500 | 7990 * Return FAIL or OK; |
| 7991 */ | |
| 7992 static int | |
| 474 | 7993 write_vim_spell(spin, fname) |
| 7994 spellinfo_T *spin; | |
| 236 | 7995 char_u *fname; |
| 7996 { | |
| 300 | 7997 FILE *fd; |
| 7998 int regionmask; | |
| 236 | 7999 int round; |
| 300 | 8000 wordnode_T *tree; |
| 8001 int nodecount; | |
| 323 | 8002 int i; |
| 8003 int l; | |
| 8004 garray_T *gap; | |
| 8005 fromto_T *ftp; | |
| 8006 char_u *p; | |
| 8007 int rr; | |
| 500 | 8008 int retval = OK; |
| 1759 | 8009 size_t fwv = 1; /* collect return value of fwrite() to avoid |
| 1753 | 8010 warnings from picky compiler */ |
| 236 | 8011 |
| 310 | 8012 fd = mch_fopen((char *)fname, "w"); |
| 300 | 8013 if (fd == NULL) |
| 236 | 8014 { |
| 8015 EMSG2(_(e_notopen), fname); | |
| 500 | 8016 return FAIL; |
| 236 | 8017 } |
| 8018 | |
| 483 | 8019 /* <HEADER>: <fileID> <versionnr> */ |
| 300 | 8020 /* <fileID> */ |
| 1753 | 8021 fwv &= fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd); |
| 1759 | 8022 if (fwv != (size_t)1) |
| 8023 /* Catch first write error, don't try writing more. */ | |
| 8024 goto theend; | |
| 8025 | |
| 483 | 8026 putc(VIMSPELLVERSION, fd); /* <versionnr> */ |
| 8027 | |
| 8028 /* | |
| 8029 * <SECTIONS>: <section> ... <sectionend> | |
| 8030 */ | |
| 8031 | |
| 714 | 8032 /* SN_INFO: <infotext> */ |
| 8033 if (spin->si_info != NULL) | |
| 8034 { | |
| 8035 putc(SN_INFO, fd); /* <sectionID> */ | |
| 8036 putc(0, fd); /* <sectionflags> */ | |
| 8037 | |
| 835 | 8038 i = (int)STRLEN(spin->si_info); |
| 714 | 8039 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */ |
| 1753 | 8040 fwv &= fwrite(spin->si_info, (size_t)i, (size_t)1, fd); /* <infotext> */ |
| 714 | 8041 } |
| 8042 | |
| 483 | 8043 /* SN_REGION: <regionname> ... |
| 8044 * Write the region names only if there is more than one. */ | |
| 316 | 8045 if (spin->si_region_count > 1) |
| 236 | 8046 { |
| 483 | 8047 putc(SN_REGION, fd); /* <sectionID> */ |
| 8048 putc(SNF_REQUIRED, fd); /* <sectionflags> */ | |
| 8049 l = spin->si_region_count * 2; | |
| 8050 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */ | |
| 1753 | 8051 fwv &= fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd); |
| 483 | 8052 /* <regionname> ... */ |
| 316 | 8053 regionmask = (1 << spin->si_region_count) - 1; |
| 236 | 8054 } |
| 8055 else | |
| 300 | 8056 regionmask = 0; |
| 483 | 8057 |
| 8058 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars> | |
| 8059 * | |
| 8060 * The table with character flags and the table for case folding. | |
| 8061 * This makes sure the same characters are recognized as word characters | |
| 8062 * when generating an when using a spell file. | |
| 260 | 8063 * Skip this for ASCII, the table may conflict with the one used for |
| 323 | 8064 * 'encoding'. |
| 8065 * Also skip this for an .add.spl file, the main spell file must contain | |
| 8066 * the table (avoids that it conflicts). File is shorter too. | |
| 8067 */ | |
| 483 | 8068 if (!spin->si_ascii && !spin->si_add) |
| 8069 { | |
| 8070 char_u folchars[128 * 8]; | |
| 8071 int flags; | |
| 8072 | |
| 485 | 8073 putc(SN_CHARFLAGS, fd); /* <sectionID> */ |
| 483 | 8074 putc(SNF_REQUIRED, fd); /* <sectionflags> */ |
| 8075 | |
| 8076 /* Form the <folchars> string first, we need to know its length. */ | |
| 8077 l = 0; | |
| 8078 for (i = 128; i < 256; ++i) | |
| 8079 { | |
| 8080 #ifdef FEAT_MBYTE | |
| 8081 if (has_mbyte) | |
| 8082 l += mb_char2bytes(spelltab.st_fold[i], folchars + l); | |
| 8083 else | |
| 8084 #endif | |
| 8085 folchars[l++] = spelltab.st_fold[i]; | |
| 8086 } | |
| 8087 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */ | |
| 8088 | |
| 8089 fputc(128, fd); /* <charflagslen> */ | |
| 8090 for (i = 128; i < 256; ++i) | |
| 8091 { | |
| 8092 flags = 0; | |
| 8093 if (spelltab.st_isw[i]) | |
| 8094 flags |= CF_WORD; | |
| 8095 if (spelltab.st_isu[i]) | |
| 8096 flags |= CF_UPPER; | |
| 8097 fputc(flags, fd); /* <charflags> */ | |
| 8098 } | |
| 8099 | |
| 8100 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */ | |
| 1753 | 8101 fwv &= fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */ |
| 483 | 8102 } |
| 8103 | |
| 8104 /* SN_MIDWORD: <midword> */ | |
| 8105 if (spin->si_midword != NULL) | |
| 8106 { | |
| 8107 putc(SN_MIDWORD, fd); /* <sectionID> */ | |
| 8108 putc(SNF_REQUIRED, fd); /* <sectionflags> */ | |
| 8109 | |
| 835 | 8110 i = (int)STRLEN(spin->si_midword); |
| 483 | 8111 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */ |
| 1753 | 8112 fwv &= fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); |
| 8113 /* <midword> */ | |
| 366 | 8114 } |
| 8115 | |
| 483 | 8116 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */ |
| 8117 if (spin->si_prefcond.ga_len > 0) | |
| 8118 { | |
| 8119 putc(SN_PREFCOND, fd); /* <sectionID> */ | |
| 8120 putc(SNF_REQUIRED, fd); /* <sectionflags> */ | |
| 8121 | |
| 8122 l = write_spell_prefcond(NULL, &spin->si_prefcond); | |
| 8123 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */ | |
| 8124 | |
| 8125 write_spell_prefcond(fd, &spin->si_prefcond); | |
| 8126 } | |
| 8127 | |
| 8128 /* SN_REP: <repcount> <rep> ... | |
| 625 | 8129 * SN_SAL: <salflags> <salcount> <sal> ... |
| 8130 * SN_REPSAL: <repcount> <rep> ... */ | |
| 323 | 8131 |
| 483 | 8132 /* round 1: SN_REP section |
| 625 | 8133 * round 2: SN_SAL section (unless SN_SOFO is used) |
| 8134 * round 3: SN_REPSAL section */ | |
| 8135 for (round = 1; round <= 3; ++round) | |
| 323 | 8136 { |
| 8137 if (round == 1) | |
| 8138 gap = &spin->si_rep; | |
| 625 | 8139 else if (round == 2) |
| 8140 { | |
| 8141 /* Don't write SN_SAL when using a SN_SOFO section */ | |
| 8142 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL) | |
| 8143 continue; | |
| 8144 gap = &spin->si_sal; | |
| 483 | 8145 } |
| 323 | 8146 else |
| 625 | 8147 gap = &spin->si_repsal; |
| 8148 | |
| 8149 /* Don't write the section if there are no items. */ | |
| 8150 if (gap->ga_len == 0) | |
| 8151 continue; | |
| 8152 | |
| 8153 /* Sort the REP/REPSAL items. */ | |
| 8154 if (round != 2) | |
| 8155 qsort(gap->ga_data, (size_t)gap->ga_len, | |
| 8156 sizeof(fromto_T), rep_compare); | |
| 8157 | |
| 8158 i = round == 1 ? SN_REP : (round == 2 ? SN_SAL : SN_REPSAL); | |
| 8159 putc(i, fd); /* <sectionID> */ | |
| 483 | 8160 |
| 8161 /* This is for making suggestions, section is not required. */ | |
| 8162 putc(0, fd); /* <sectionflags> */ | |
| 8163 | |
| 8164 /* Compute the length of what follows. */ | |
| 8165 l = 2; /* count <repcount> or <salcount> */ | |
| 8166 for (i = 0; i < gap->ga_len; ++i) | |
| 8167 { | |
| 8168 ftp = &((fromto_T *)gap->ga_data)[i]; | |
| 835 | 8169 l += 1 + (int)STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */ |
| 8170 l += 1 + (int)STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */ | |
| 483 | 8171 } |
| 8172 if (round == 2) | |
| 8173 ++l; /* count <salflags> */ | |
| 8174 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */ | |
| 8175 | |
| 8176 if (round == 2) | |
| 8177 { | |
| 323 | 8178 i = 0; |
| 8179 if (spin->si_followup) | |
| 8180 i |= SAL_F0LLOWUP; | |
| 8181 if (spin->si_collapse) | |
| 8182 i |= SAL_COLLAPSE; | |
| 8183 if (spin->si_rem_accents) | |
| 8184 i |= SAL_REM_ACCENTS; | |
| 8185 putc(i, fd); /* <salflags> */ | |
| 8186 } | |
| 8187 | |
| 8188 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */ | |
| 8189 for (i = 0; i < gap->ga_len; ++i) | |
| 8190 { | |
| 8191 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */ | |
| 8192 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */ | |
| 8193 ftp = &((fromto_T *)gap->ga_data)[i]; | |
| 8194 for (rr = 1; rr <= 2; ++rr) | |
| 8195 { | |
| 8196 p = rr == 1 ? ftp->ft_from : ftp->ft_to; | |
| 835 | 8197 l = (int)STRLEN(p); |
| 323 | 8198 putc(l, fd); |
| 1758 | 8199 if (l > 0) |
| 8200 fwv &= fwrite(p, l, (size_t)1, fd); | |
| 323 | 8201 } |
| 8202 } | |
| 483 | 8203 |
| 8204 } | |
| 8205 | |
| 8206 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto> | |
| 8207 * This is for making suggestions, section is not required. */ | |
| 375 | 8208 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL) |
| 8209 { | |
| 483 | 8210 putc(SN_SOFO, fd); /* <sectionID> */ |
| 8211 putc(0, fd); /* <sectionflags> */ | |
| 375 | 8212 |
| 835 | 8213 l = (int)STRLEN(spin->si_sofofr); |
| 483 | 8214 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4); |
| 8215 /* <sectionlen> */ | |
| 8216 | |
| 8217 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */ | |
| 1753 | 8218 fwv &= fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */ |
| 375 | 8219 |
| 835 | 8220 l = (int)STRLEN(spin->si_sofoto); |
| 483 | 8221 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */ |
| 1753 | 8222 fwv &= fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */ |
| 483 | 8223 } |
| 8224 | |
| 625 | 8225 /* SN_WORDS: <word> ... |
| 8226 * This is for making suggestions, section is not required. */ | |
| 8227 if (spin->si_commonwords.ht_used > 0) | |
| 8228 { | |
| 8229 putc(SN_WORDS, fd); /* <sectionID> */ | |
| 8230 putc(0, fd); /* <sectionflags> */ | |
| 8231 | |
| 8232 /* round 1: count the bytes | |
| 8233 * round 2: write the bytes */ | |
| 8234 for (round = 1; round <= 2; ++round) | |
| 8235 { | |
| 8236 int todo; | |
| 8237 int len = 0; | |
| 8238 hashitem_T *hi; | |
| 8239 | |
| 835 | 8240 todo = (int)spin->si_commonwords.ht_used; |
| 625 | 8241 for (hi = spin->si_commonwords.ht_array; todo > 0; ++hi) |
| 8242 if (!HASHITEM_EMPTY(hi)) | |
| 8243 { | |
| 835 | 8244 l = (int)STRLEN(hi->hi_key) + 1; |
| 625 | 8245 len += l; |
| 8246 if (round == 2) /* <word> */ | |
| 1753 | 8247 fwv &= fwrite(hi->hi_key, (size_t)l, (size_t)1, fd); |
| 625 | 8248 --todo; |
| 8249 } | |
| 8250 if (round == 1) | |
| 8251 put_bytes(fd, (long_u)len, 4); /* <sectionlen> */ | |
| 8252 } | |
| 8253 } | |
| 8254 | |
| 483 | 8255 /* SN_MAP: <mapstr> |
| 8256 * This is for making suggestions, section is not required. */ | |
| 8257 if (spin->si_map.ga_len > 0) | |
| 8258 { | |
| 8259 putc(SN_MAP, fd); /* <sectionID> */ | |
| 8260 putc(0, fd); /* <sectionflags> */ | |
| 8261 l = spin->si_map.ga_len; | |
| 8262 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */ | |
| 1753 | 8263 fwv &= fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd); |
| 483 | 8264 /* <mapstr> */ |
| 8265 } | |
| 8266 | |
| 625 | 8267 /* SN_SUGFILE: <timestamp> |
| 8268 * This is used to notify that a .sug file may be available and at the | |
| 8269 * same time allows for checking that a .sug file that is found matches | |
| 8270 * with this .spl file. That's because the word numbers must be exactly | |
| 8271 * right. */ | |
| 8272 if (!spin->si_nosugfile | |
| 8273 && (spin->si_sal.ga_len > 0 | |
| 8274 || (spin->si_sofofr != NULL && spin->si_sofoto != NULL))) | |
| 8275 { | |
| 8276 putc(SN_SUGFILE, fd); /* <sectionID> */ | |
| 8277 putc(0, fd); /* <sectionflags> */ | |
| 8278 put_bytes(fd, (long_u)8, 4); /* <sectionlen> */ | |
| 8279 | |
| 8280 /* Set si_sugtime and write it to the file. */ | |
| 8281 spin->si_sugtime = time(NULL); | |
|
2229
d45902a5c61c
Fix a few more things for persistent undo.
Bram Moolenaar <bram@vim.org>
parents:
2214
diff
changeset
|
8282 put_time(fd, spin->si_sugtime); /* <timestamp> */ |
| 625 | 8283 } |
| 8284 | |
| 698 | 8285 /* SN_NOSPLITSUGS: nothing |
| 8286 * This is used to notify that no suggestions with word splits are to be | |
| 8287 * made. */ | |
| 8288 if (spin->si_nosplitsugs) | |
| 8289 { | |
| 8290 putc(SN_NOSPLITSUGS, fd); /* <sectionID> */ | |
| 8291 putc(0, fd); /* <sectionflags> */ | |
| 8292 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */ | |
| 8293 } | |
| 8294 | |
| 483 | 8295 /* SN_COMPOUND: compound info. |
| 8296 * We don't mark it required, when not supported all compound words will | |
| 8297 * be bad words. */ | |
| 8298 if (spin->si_compflags != NULL) | |
| 8299 { | |
| 8300 putc(SN_COMPOUND, fd); /* <sectionID> */ | |
| 8301 putc(0, fd); /* <sectionflags> */ | |
| 8302 | |
| 835 | 8303 l = (int)STRLEN(spin->si_compflags); |
| 809 | 8304 for (i = 0; i < spin->si_comppat.ga_len; ++i) |
| 835 | 8305 l += (int)STRLEN(((char_u **)(spin->si_comppat.ga_data))[i]) + 1; |
| 809 | 8306 put_bytes(fd, (long_u)(l + 7), 4); /* <sectionlen> */ |
| 8307 | |
| 483 | 8308 putc(spin->si_compmax, fd); /* <compmax> */ |
| 8309 putc(spin->si_compminlen, fd); /* <compminlen> */ | |
| 8310 putc(spin->si_compsylmax, fd); /* <compsylmax> */ | |
| 809 | 8311 putc(0, fd); /* for Vim 7.0b compatibility */ |
| 8312 putc(spin->si_compoptions, fd); /* <compoptions> */ | |
| 8313 put_bytes(fd, (long_u)spin->si_comppat.ga_len, 2); | |
| 8314 /* <comppatcount> */ | |
| 8315 for (i = 0; i < spin->si_comppat.ga_len; ++i) | |
| 8316 { | |
| 8317 p = ((char_u **)(spin->si_comppat.ga_data))[i]; | |
| 835 | 8318 putc((int)STRLEN(p), fd); /* <comppatlen> */ |
| 1753 | 8319 fwv &= fwrite(p, (size_t)STRLEN(p), (size_t)1, fd); |
| 8320 /* <comppattext> */ | |
| 809 | 8321 } |
| 483 | 8322 /* <compflags> */ |
| 1753 | 8323 fwv &= fwrite(spin->si_compflags, (size_t)STRLEN(spin->si_compflags), |
| 809 | 8324 (size_t)1, fd); |
| 483 | 8325 } |
| 8326 | |
| 492 | 8327 /* SN_NOBREAK: NOBREAK flag */ |
| 8328 if (spin->si_nobreak) | |
| 8329 { | |
| 8330 putc(SN_NOBREAK, fd); /* <sectionID> */ | |
| 8331 putc(0, fd); /* <sectionflags> */ | |
| 8332 | |
| 1188 | 8333 /* It's empty, the presence of the section flags the feature. */ |
| 492 | 8334 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */ |
| 8335 } | |
| 8336 | |
| 483 | 8337 /* SN_SYLLABLE: syllable info. |
| 8338 * We don't mark it required, when not supported syllables will not be | |
| 8339 * counted. */ | |
| 8340 if (spin->si_syllable != NULL) | |
| 8341 { | |
| 8342 putc(SN_SYLLABLE, fd); /* <sectionID> */ | |
| 8343 putc(0, fd); /* <sectionflags> */ | |
| 8344 | |
| 835 | 8345 l = (int)STRLEN(spin->si_syllable); |
| 483 | 8346 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */ |
| 1753 | 8347 fwv &= fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); |
| 8348 /* <syllable> */ | |
| 483 | 8349 } |
| 8350 | |
| 8351 /* end of <SECTIONS> */ | |
| 8352 putc(SN_END, fd); /* <sectionend> */ | |
| 8353 | |
| 302 | 8354 |
| 236 | 8355 /* |
| 339 | 8356 * <LWORDTREE> <KWORDTREE> <PREFIXTREE> |
| 236 | 8357 */ |
| 323 | 8358 spin->si_memtot = 0; |
| 339 | 8359 for (round = 1; round <= 3; ++round) |
| 236 | 8360 { |
| 339 | 8361 if (round == 1) |
| 469 | 8362 tree = spin->si_foldroot->wn_sibling; |
| 339 | 8363 else if (round == 2) |
| 469 | 8364 tree = spin->si_keeproot->wn_sibling; |
| 339 | 8365 else |
| 469 | 8366 tree = spin->si_prefroot->wn_sibling; |
| 236 | 8367 |
| 346 | 8368 /* Clear the index and wnode fields in the tree. */ |
| 8369 clear_node(tree); | |
| 8370 | |
| 300 | 8371 /* Count the number of nodes. Needed to be able to allocate the |
| 346 | 8372 * memory when reading the nodes. Also fills in index for shared |
| 300 | 8373 * nodes. */ |
| 346 | 8374 nodecount = put_node(NULL, tree, 0, regionmask, round == 3); |
| 236 | 8375 |
| 300 | 8376 /* number of nodes in 4 bytes */ |
| 8377 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */ | |
| 302 | 8378 spin->si_memtot += nodecount + nodecount * sizeof(int); |
| 236 | 8379 |
| 300 | 8380 /* Write the nodes. */ |
| 346 | 8381 (void)put_node(fd, tree, 0, regionmask, round == 3); |
| 236 | 8382 } |
| 8383 | |
| 1753 | 8384 /* Write another byte to check for errors (file system full). */ |
| 500 | 8385 if (putc(0, fd) == EOF) |
| 8386 retval = FAIL; | |
| 1759 | 8387 theend: |
| 500 | 8388 if (fclose(fd) == EOF) |
| 8389 retval = FAIL; | |
| 8390 | |
| 1759 | 8391 if (fwv != (size_t)1) |
| 1753 | 8392 retval = FAIL; |
| 8393 if (retval == FAIL) | |
| 8394 EMSG(_(e_write)); | |
| 8395 | |
| 500 | 8396 return retval; |
| 236 | 8397 } |
| 8398 | |
| 8399 /* | |
| 346 | 8400 * Clear the index and wnode fields of "node", it siblings and its |
| 8401 * children. This is needed because they are a union with other items to save | |
| 8402 * space. | |
| 8403 */ | |
| 8404 static void | |
| 8405 clear_node(node) | |
| 8406 wordnode_T *node; | |
| 8407 { | |
| 8408 wordnode_T *np; | |
| 8409 | |
| 8410 if (node != NULL) | |
| 8411 for (np = node; np != NULL; np = np->wn_sibling) | |
| 8412 { | |
| 8413 np->wn_u1.index = 0; | |
| 8414 np->wn_u2.wnode = NULL; | |
| 8415 | |
| 8416 if (np->wn_byte != NUL) | |
| 8417 clear_node(np->wn_child); | |
| 8418 } | |
| 8419 } | |
| 8420 | |
| 8421 | |
| 8422 /* | |
| 300 | 8423 * Dump a word tree at node "node". |
| 8424 * | |
| 8425 * This first writes the list of possible bytes (siblings). Then for each | |
| 8426 * byte recursively write the children. | |
| 8427 * | |
| 625 | 8428 * NOTE: The code here must match the code in read_tree_node(), since |
| 8429 * assumptions are made about the indexes (so that we don't have to write them | |
| 8430 * in the file). | |
| 236 | 8431 * |
| 300 | 8432 * Returns the number of nodes used. |
| 236 | 8433 */ |
| 300 | 8434 static int |
| 944 | 8435 put_node(fd, node, idx, regionmask, prefixtree) |
| 339 | 8436 FILE *fd; /* NULL when only counting */ |
| 300 | 8437 wordnode_T *node; |
| 944 | 8438 int idx; |
| 300 | 8439 int regionmask; |
| 339 | 8440 int prefixtree; /* TRUE for PREFIXTREE */ |
| 236 | 8441 { |
| 944 | 8442 int newindex = idx; |
| 300 | 8443 int siblingcount = 0; |
| 8444 wordnode_T *np; | |
| 236 | 8445 int flags; |
| 300 | 8446 |
| 8447 /* If "node" is zero the tree is empty. */ | |
| 8448 if (node == NULL) | |
| 8449 return 0; | |
| 8450 | |
| 8451 /* Store the index where this node is written. */ | |
| 944 | 8452 node->wn_u1.index = idx; |
| 236 | 8453 |
| 300 | 8454 /* Count the number of siblings. */ |
| 8455 for (np = node; np != NULL; np = np->wn_sibling) | |
| 8456 ++siblingcount; | |
| 236 | 8457 |
| 300 | 8458 /* Write the sibling count. */ |
| 8459 if (fd != NULL) | |
| 8460 putc(siblingcount, fd); /* <siblingcount> */ | |
| 236 | 8461 |
| 300 | 8462 /* Write each sibling byte and optionally extra info. */ |
| 8463 for (np = node; np != NULL; np = np->wn_sibling) | |
| 236 | 8464 { |
| 300 | 8465 if (np->wn_byte == 0) |
| 8466 { | |
| 8467 if (fd != NULL) | |
| 8468 { | |
| 339 | 8469 /* For a NUL byte (end of word) write the flags etc. */ |
| 8470 if (prefixtree) | |
| 300 | 8471 { |
| 481 | 8472 /* In PREFIXTREE write the required affixID and the |
| 390 | 8473 * associated condition nr (stored in wn_region). The |
| 8474 * byte value is misused to store the "rare" and "not | |
| 8475 * combining" flags */ | |
| 455 | 8476 if (np->wn_flags == (short_u)PFX_FLAGS) |
| 8477 putc(BY_NOFLAGS, fd); /* <byte> */ | |
| 8478 else | |
| 8479 { | |
| 390 | 8480 putc(BY_FLAGS, fd); /* <byte> */ |
| 455 | 8481 putc(np->wn_flags, fd); /* <pflags> */ |
| 8482 } | |
| 481 | 8483 putc(np->wn_affixID, fd); /* <affixID> */ |
| 339 | 8484 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */ |
| 300 | 8485 } |
| 8486 else | |
| 8487 { | |
| 339 | 8488 /* For word trees we write the flag/region items. */ |
| 8489 flags = np->wn_flags; | |
| 8490 if (regionmask != 0 && np->wn_region != regionmask) | |
| 8491 flags |= WF_REGION; | |
| 481 | 8492 if (np->wn_affixID != 0) |
| 8493 flags |= WF_AFX; | |
| 339 | 8494 if (flags == 0) |
| 8495 { | |
| 8496 /* word without flags or region */ | |
| 8497 putc(BY_NOFLAGS, fd); /* <byte> */ | |
| 8498 } | |
| 8499 else | |
| 8500 { | |
| 390 | 8501 if (np->wn_flags >= 0x100) |
| 8502 { | |
| 8503 putc(BY_FLAGS2, fd); /* <byte> */ | |
| 8504 putc(flags, fd); /* <flags> */ | |
| 8505 putc((unsigned)flags >> 8, fd); /* <flags2> */ | |
| 8506 } | |
| 8507 else | |
| 8508 { | |
| 8509 putc(BY_FLAGS, fd); /* <byte> */ | |
| 8510 putc(flags, fd); /* <flags> */ | |
| 8511 } | |
| 339 | 8512 if (flags & WF_REGION) |
| 8513 putc(np->wn_region, fd); /* <region> */ | |
| 481 | 8514 if (flags & WF_AFX) |
| 8515 putc(np->wn_affixID, fd); /* <affixID> */ | |
| 339 | 8516 } |
| 300 | 8517 } |
| 8518 } | |
| 8519 } | |
| 8520 else | |
| 8521 { | |
| 346 | 8522 if (np->wn_child->wn_u1.index != 0 |
| 8523 && np->wn_child->wn_u2.wnode != node) | |
| 300 | 8524 { |
| 8525 /* The child is written elsewhere, write the reference. */ | |
| 8526 if (fd != NULL) | |
| 8527 { | |
| 8528 putc(BY_INDEX, fd); /* <byte> */ | |
| 8529 /* <nodeidx> */ | |
| 346 | 8530 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3); |
| 300 | 8531 } |
| 8532 } | |
| 346 | 8533 else if (np->wn_child->wn_u2.wnode == NULL) |
| 300 | 8534 /* We will write the child below and give it an index. */ |
| 346 | 8535 np->wn_child->wn_u2.wnode = node; |
| 236 | 8536 |
| 300 | 8537 if (fd != NULL) |
| 8538 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */ | |
| 8539 { | |
| 8540 EMSG(_(e_write)); | |
| 8541 return 0; | |
| 8542 } | |
| 8543 } | |
| 236 | 8544 } |
| 8545 | |
| 300 | 8546 /* Space used in the array when reading: one for each sibling and one for |
| 8547 * the count. */ | |
| 8548 newindex += siblingcount + 1; | |
| 249 | 8549 |
| 300 | 8550 /* Recursively dump the children of each sibling. */ |
| 8551 for (np = node; np != NULL; np = np->wn_sibling) | |
| 346 | 8552 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node) |
| 8553 newindex = put_node(fd, np->wn_child, newindex, regionmask, | |
| 339 | 8554 prefixtree); |
| 249 | 8555 |
| 300 | 8556 return newindex; |
| 236 | 8557 } |
| 8558 | |
| 8559 | |
| 8560 /* | |
| 310 | 8561 * ":mkspell [-ascii] outfile infile ..." |
| 8562 * ":mkspell [-ascii] addfile" | |
| 236 | 8563 */ |
| 8564 void | |
| 8565 ex_mkspell(eap) | |
| 8566 exarg_T *eap; | |
| 8567 { | |
| 8568 int fcount; | |
| 8569 char_u **fnames; | |
| 310 | 8570 char_u *arg = eap->arg; |
| 8571 int ascii = FALSE; | |
| 8572 | |
| 8573 if (STRNCMP(arg, "-ascii", 6) == 0) | |
| 8574 { | |
| 8575 ascii = TRUE; | |
| 8576 arg = skipwhite(arg + 6); | |
| 8577 } | |
| 8578 | |
| 8579 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */ | |
| 3620 | 8580 if (get_arglist_exp(arg, &fcount, &fnames, FALSE) == OK) |
| 310 | 8581 { |
| 323 | 8582 mkspell(fcount, fnames, ascii, eap->forceit, FALSE); |
| 310 | 8583 FreeWild(fcount, fnames); |
| 8584 } | |
| 8585 } | |
| 8586 | |
| 8587 /* | |
| 625 | 8588 * Create the .sug file. |
| 8589 * Uses the soundfold info in "spin". | |
| 8590 * Writes the file with the name "wfname", with ".spl" changed to ".sug". | |
| 8591 */ | |
| 8592 static void | |
| 8593 spell_make_sugfile(spin, wfname) | |
| 8594 spellinfo_T *spin; | |
| 8595 char_u *wfname; | |
| 8596 { | |
| 2770 | 8597 char_u *fname = NULL; |
| 625 | 8598 int len; |
| 8599 slang_T *slang; | |
| 8600 int free_slang = FALSE; | |
| 8601 | |
| 8602 /* | |
| 8603 * Read back the .spl file that was written. This fills the required | |
| 8604 * info for soundfolding. This also uses less memory than the | |
| 8605 * pointer-linked version of the trie. And it avoids having two versions | |
| 8606 * of the code for the soundfolding stuff. | |
| 8607 * It might have been done already by spell_reload_one(). | |
| 8608 */ | |
| 8609 for (slang = first_lang; slang != NULL; slang = slang->sl_next) | |
| 8610 if (fullpathcmp(wfname, slang->sl_fname, FALSE) == FPC_SAME) | |
| 8611 break; | |
| 8612 if (slang == NULL) | |
| 8613 { | |
| 8614 spell_message(spin, (char_u *)_("Reading back spell file...")); | |
| 8615 slang = spell_load_file(wfname, NULL, NULL, FALSE); | |
| 8616 if (slang == NULL) | |
| 8617 return; | |
| 8618 free_slang = TRUE; | |
| 8619 } | |
| 8620 | |
| 8621 /* | |
| 8622 * Clear the info in "spin" that is used. | |
| 8623 */ | |
| 8624 spin->si_blocks = NULL; | |
| 8625 spin->si_blocks_cnt = 0; | |
| 8626 spin->si_compress_cnt = 0; /* will stay at 0 all the time*/ | |
| 8627 spin->si_free_count = 0; | |
| 8628 spin->si_first_free = NULL; | |
| 8629 spin->si_foldwcount = 0; | |
| 8630 | |
| 8631 /* | |
| 8632 * Go through the trie of good words, soundfold each word and add it to | |
| 8633 * the soundfold trie. | |
| 8634 */ | |
| 8635 spell_message(spin, (char_u *)_("Performing soundfolding...")); | |
| 8636 if (sug_filltree(spin, slang) == FAIL) | |
| 8637 goto theend; | |
| 8638 | |
| 8639 /* | |
| 8640 * Create the table which links each soundfold word with a list of the | |
| 8641 * good words it may come from. Creates buffer "spin->si_spellbuf". | |
| 8642 * This also removes the wordnr from the NUL byte entries to make | |
| 8643 * compression possible. | |
| 8644 */ | |
| 8645 if (sug_maketable(spin) == FAIL) | |
| 8646 goto theend; | |
| 8647 | |
| 8648 smsg((char_u *)_("Number of words after soundfolding: %ld"), | |
| 8649 (long)spin->si_spellbuf->b_ml.ml_line_count); | |
| 8650 | |
| 8651 /* | |
| 8652 * Compress the soundfold trie. | |
| 8653 */ | |
| 8654 spell_message(spin, (char_u *)_(msg_compressing)); | |
| 8655 wordtree_compress(spin, spin->si_foldroot); | |
| 8656 | |
| 8657 /* | |
| 8658 * Write the .sug file. | |
| 8659 * Make the file name by changing ".spl" to ".sug". | |
| 8660 */ | |
| 2770 | 8661 fname = alloc(MAXPATHL); |
| 8662 if (fname == NULL) | |
| 8663 goto theend; | |
| 2768 | 8664 vim_strncpy(fname, wfname, MAXPATHL - 1); |
| 835 | 8665 len = (int)STRLEN(fname); |
| 625 | 8666 fname[len - 2] = 'u'; |
| 8667 fname[len - 1] = 'g'; | |
| 8668 sug_write(spin, fname); | |
| 8669 | |
| 8670 theend: | |
| 2770 | 8671 vim_free(fname); |
| 625 | 8672 if (free_slang) |
| 8673 slang_free(slang); | |
| 8674 free_blocks(spin->si_blocks); | |
| 8675 close_spellbuf(spin->si_spellbuf); | |
| 8676 } | |
| 8677 | |
| 8678 /* | |
| 8679 * Build the soundfold trie for language "slang". | |
| 8680 */ | |
| 8681 static int | |
| 8682 sug_filltree(spin, slang) | |
| 8683 spellinfo_T *spin; | |
| 8684 slang_T *slang; | |
| 8685 { | |
| 8686 char_u *byts; | |
| 8687 idx_T *idxs; | |
| 8688 int depth; | |
| 8689 idx_T arridx[MAXWLEN]; | |
| 8690 int curi[MAXWLEN]; | |
| 8691 char_u tword[MAXWLEN]; | |
| 8692 char_u tsalword[MAXWLEN]; | |
| 8693 int c; | |
| 8694 idx_T n; | |
| 8695 unsigned words_done = 0; | |
| 8696 int wordcount[MAXWLEN]; | |
| 8697 | |
| 4352 | 8698 /* We use si_foldroot for the soundfolded trie. */ |
| 625 | 8699 spin->si_foldroot = wordtree_alloc(spin); |
| 8700 if (spin->si_foldroot == NULL) | |
| 8701 return FAIL; | |
| 8702 | |
| 8703 /* let tree_add_word() know we're adding to the soundfolded tree */ | |
| 8704 spin->si_sugtree = TRUE; | |
| 8705 | |
| 8706 /* | |
| 8707 * Go through the whole case-folded tree, soundfold each word and put it | |
| 8708 * in the trie. | |
| 8709 */ | |
| 8710 byts = slang->sl_fbyts; | |
| 8711 idxs = slang->sl_fidxs; | |
| 8712 | |
| 8713 arridx[0] = 0; | |
| 8714 curi[0] = 1; | |
| 8715 wordcount[0] = 0; | |
| 8716 | |
| 8717 depth = 0; | |
| 8718 while (depth >= 0 && !got_int) | |
| 8719 { | |
| 8720 if (curi[depth] > byts[arridx[depth]]) | |
| 8721 { | |
| 8722 /* Done all bytes at this node, go up one level. */ | |
| 8723 idxs[arridx[depth]] = wordcount[depth]; | |
| 8724 if (depth > 0) | |
| 8725 wordcount[depth - 1] += wordcount[depth]; | |
| 8726 | |
| 8727 --depth; | |
| 8728 line_breakcheck(); | |
| 8729 } | |
| 8730 else | |
| 8731 { | |
| 8732 | |
| 8733 /* Do one more byte at this node. */ | |
| 8734 n = arridx[depth] + curi[depth]; | |
| 8735 ++curi[depth]; | |
| 8736 | |
| 8737 c = byts[n]; | |
| 8738 if (c == 0) | |
| 8739 { | |
| 8740 /* Sound-fold the word. */ | |
| 8741 tword[depth] = NUL; | |
| 8742 spell_soundfold(slang, tword, TRUE, tsalword); | |
| 8743 | |
| 8744 /* We use the "flags" field for the MSB of the wordnr, | |
| 8745 * "region" for the LSB of the wordnr. */ | |
| 8746 if (tree_add_word(spin, tsalword, spin->si_foldroot, | |
| 8747 words_done >> 16, words_done & 0xffff, | |
| 8748 0) == FAIL) | |
| 8749 return FAIL; | |
| 8750 | |
| 8751 ++words_done; | |
| 8752 ++wordcount[depth]; | |
| 8753 | |
| 8754 /* Reset the block count each time to avoid compression | |
| 8755 * kicking in. */ | |
| 8756 spin->si_blocks_cnt = 0; | |
| 8757 | |
| 8758 /* Skip over any other NUL bytes (same word with different | |
| 8759 * flags). */ | |
| 8760 while (byts[n + 1] == 0) | |
| 8761 { | |
| 8762 ++n; | |
| 8763 ++curi[depth]; | |
| 8764 } | |
| 8765 } | |
| 8766 else | |
| 8767 { | |
| 8768 /* Normal char, go one level deeper. */ | |
| 8769 tword[depth++] = c; | |
| 8770 arridx[depth] = idxs[n]; | |
| 8771 curi[depth] = 1; | |
| 8772 wordcount[depth] = 0; | |
| 8773 } | |
| 8774 } | |
| 8775 } | |
| 8776 | |
| 8777 smsg((char_u *)_("Total number of words: %d"), words_done); | |
| 8778 | |
| 8779 return OK; | |
| 8780 } | |
| 8781 | |
| 8782 /* | |
| 8783 * Make the table that links each word in the soundfold trie to the words it | |
| 8784 * can be produced from. | |
| 8785 * This is not unlike lines in a file, thus use a memfile to be able to access | |
| 8786 * the table efficiently. | |
| 8787 * Returns FAIL when out of memory. | |
| 8788 */ | |
| 8789 static int | |
| 8790 sug_maketable(spin) | |
| 8791 spellinfo_T *spin; | |
| 8792 { | |
| 8793 garray_T ga; | |
| 8794 int res = OK; | |
| 8795 | |
| 8796 /* Allocate a buffer, open a memline for it and create the swap file | |
| 8797 * (uses a temp file, not a .swp file). */ | |
| 8798 spin->si_spellbuf = open_spellbuf(); | |
| 8799 if (spin->si_spellbuf == NULL) | |
| 8800 return FAIL; | |
| 8801 | |
| 8802 /* Use a buffer to store the line info, avoids allocating many small | |
| 8803 * pieces of memory. */ | |
| 8804 ga_init2(&ga, 1, 100); | |
| 8805 | |
| 8806 /* recursively go through the tree */ | |
| 8807 if (sug_filltable(spin, spin->si_foldroot->wn_sibling, 0, &ga) == -1) | |
| 8808 res = FAIL; | |
| 8809 | |
| 8810 ga_clear(&ga); | |
| 8811 return res; | |
| 8812 } | |
| 8813 | |
| 8814 /* | |
| 8815 * Fill the table for one node and its children. | |
| 8816 * Returns the wordnr at the start of the node. | |
| 8817 * Returns -1 when out of memory. | |
| 8818 */ | |
| 8819 static int | |
| 8820 sug_filltable(spin, node, startwordnr, gap) | |
| 8821 spellinfo_T *spin; | |
| 8822 wordnode_T *node; | |
| 8823 int startwordnr; | |
| 8824 garray_T *gap; /* place to store line of numbers */ | |
| 8825 { | |
| 8826 wordnode_T *p, *np; | |
| 8827 int wordnr = startwordnr; | |
| 8828 int nr; | |
| 8829 int prev_nr; | |
| 8830 | |
| 8831 for (p = node; p != NULL; p = p->wn_sibling) | |
| 8832 { | |
| 8833 if (p->wn_byte == NUL) | |
| 8834 { | |
| 8835 gap->ga_len = 0; | |
| 8836 prev_nr = 0; | |
| 8837 for (np = p; np != NULL && np->wn_byte == NUL; np = np->wn_sibling) | |
| 8838 { | |
| 8839 if (ga_grow(gap, 10) == FAIL) | |
| 8840 return -1; | |
| 8841 | |
| 8842 nr = (np->wn_flags << 16) + (np->wn_region & 0xffff); | |
| 8843 /* Compute the offset from the previous nr and store the | |
| 8844 * offset in a way that it takes a minimum number of bytes. | |
| 8845 * It's a bit like utf-8, but without the need to mark | |
| 8846 * following bytes. */ | |
| 8847 nr -= prev_nr; | |
| 8848 prev_nr += nr; | |
| 8849 gap->ga_len += offset2bytes(nr, | |
| 8850 (char_u *)gap->ga_data + gap->ga_len); | |
| 8851 } | |
| 8852 | |
| 8853 /* add the NUL byte */ | |
| 8854 ((char_u *)gap->ga_data)[gap->ga_len++] = NUL; | |
| 8855 | |
| 8856 if (ml_append_buf(spin->si_spellbuf, (linenr_T)wordnr, | |
| 8857 gap->ga_data, gap->ga_len, TRUE) == FAIL) | |
| 8858 return -1; | |
| 8859 ++wordnr; | |
| 8860 | |
| 8861 /* Remove extra NUL entries, we no longer need them. We don't | |
| 8862 * bother freeing the nodes, the won't be reused anyway. */ | |
| 8863 while (p->wn_sibling != NULL && p->wn_sibling->wn_byte == NUL) | |
| 8864 p->wn_sibling = p->wn_sibling->wn_sibling; | |
| 8865 | |
| 8866 /* Clear the flags on the remaining NUL node, so that compression | |
| 8867 * works a lot better. */ | |
| 8868 p->wn_flags = 0; | |
| 8869 p->wn_region = 0; | |
| 8870 } | |
| 8871 else | |
| 8872 { | |
| 8873 wordnr = sug_filltable(spin, p->wn_child, wordnr, gap); | |
| 8874 if (wordnr == -1) | |
| 8875 return -1; | |
| 8876 } | |
| 8877 } | |
| 8878 return wordnr; | |
| 8879 } | |
| 8880 | |
| 8881 /* | |
| 8882 * Convert an offset into a minimal number of bytes. | |
| 8883 * Similar to utf_char2byters, but use 8 bits in followup bytes and avoid NUL | |
| 8884 * bytes. | |
| 8885 */ | |
| 8886 static int | |
| 8887 offset2bytes(nr, buf) | |
| 8888 int nr; | |
| 8889 char_u *buf; | |
| 8890 { | |
| 8891 int rem; | |
| 8892 int b1, b2, b3, b4; | |
| 8893 | |
| 8894 /* Split the number in parts of base 255. We need to avoid NUL bytes. */ | |
| 8895 b1 = nr % 255 + 1; | |
| 8896 rem = nr / 255; | |
| 8897 b2 = rem % 255 + 1; | |
| 8898 rem = rem / 255; | |
| 8899 b3 = rem % 255 + 1; | |
| 8900 b4 = rem / 255 + 1; | |
| 8901 | |
| 8902 if (b4 > 1 || b3 > 0x1f) /* 4 bytes */ | |
| 8903 { | |
| 8904 buf[0] = 0xe0 + b4; | |
| 8905 buf[1] = b3; | |
| 8906 buf[2] = b2; | |
| 8907 buf[3] = b1; | |
| 8908 return 4; | |
| 8909 } | |
| 8910 if (b3 > 1 || b2 > 0x3f ) /* 3 bytes */ | |
| 8911 { | |
| 8912 buf[0] = 0xc0 + b3; | |
| 8913 buf[1] = b2; | |
| 8914 buf[2] = b1; | |
| 8915 return 3; | |
| 8916 } | |
| 8917 if (b2 > 1 || b1 > 0x7f ) /* 2 bytes */ | |
| 8918 { | |
| 8919 buf[0] = 0x80 + b2; | |
| 8920 buf[1] = b1; | |
| 8921 return 2; | |
| 8922 } | |
| 8923 /* 1 byte */ | |
| 8924 buf[0] = b1; | |
| 8925 return 1; | |
| 8926 } | |
| 8927 | |
| 8928 /* | |
| 8929 * Opposite of offset2bytes(). | |
| 8930 * "pp" points to the bytes and is advanced over it. | |
| 8931 * Returns the offset. | |
| 8932 */ | |
| 8933 static int | |
| 8934 bytes2offset(pp) | |
| 8935 char_u **pp; | |
| 8936 { | |
| 8937 char_u *p = *pp; | |
| 8938 int nr; | |
| 8939 int c; | |
| 8940 | |
| 8941 c = *p++; | |
| 8942 if ((c & 0x80) == 0x00) /* 1 byte */ | |
| 8943 { | |
| 8944 nr = c - 1; | |
| 8945 } | |
| 8946 else if ((c & 0xc0) == 0x80) /* 2 bytes */ | |
| 8947 { | |
| 8948 nr = (c & 0x3f) - 1; | |
| 8949 nr = nr * 255 + (*p++ - 1); | |
| 8950 } | |
| 8951 else if ((c & 0xe0) == 0xc0) /* 3 bytes */ | |
| 8952 { | |
| 8953 nr = (c & 0x1f) - 1; | |
| 8954 nr = nr * 255 + (*p++ - 1); | |
| 8955 nr = nr * 255 + (*p++ - 1); | |
| 8956 } | |
| 8957 else /* 4 bytes */ | |
| 8958 { | |
| 8959 nr = (c & 0x0f) - 1; | |
| 8960 nr = nr * 255 + (*p++ - 1); | |
| 8961 nr = nr * 255 + (*p++ - 1); | |
| 8962 nr = nr * 255 + (*p++ - 1); | |
| 8963 } | |
| 8964 | |
| 8965 *pp = p; | |
| 8966 return nr; | |
| 8967 } | |
| 8968 | |
| 8969 /* | |
| 8970 * Write the .sug file in "fname". | |
| 8971 */ | |
| 8972 static void | |
| 8973 sug_write(spin, fname) | |
| 8974 spellinfo_T *spin; | |
| 8975 char_u *fname; | |
| 8976 { | |
| 8977 FILE *fd; | |
| 8978 wordnode_T *tree; | |
| 8979 int nodecount; | |
| 8980 int wcount; | |
| 8981 char_u *line; | |
| 8982 linenr_T lnum; | |
| 8983 int len; | |
| 8984 | |
| 8985 /* Create the file. Note that an existing file is silently overwritten! */ | |
| 8986 fd = mch_fopen((char *)fname, "w"); | |
| 8987 if (fd == NULL) | |
| 8988 { | |
| 8989 EMSG2(_(e_notopen), fname); | |
| 8990 return; | |
| 8991 } | |
| 8992 | |
| 8993 vim_snprintf((char *)IObuff, IOSIZE, | |
| 8994 _("Writing suggestion file %s ..."), fname); | |
| 8995 spell_message(spin, IObuff); | |
| 8996 | |
| 8997 /* | |
| 8998 * <SUGHEADER>: <fileID> <versionnr> <timestamp> | |
| 8999 */ | |
| 9000 if (fwrite(VIMSUGMAGIC, VIMSUGMAGICL, (size_t)1, fd) != 1) /* <fileID> */ | |
| 9001 { | |
| 9002 EMSG(_(e_write)); | |
| 9003 goto theend; | |
| 9004 } | |
| 9005 putc(VIMSUGVERSION, fd); /* <versionnr> */ | |
| 9006 | |
| 9007 /* 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
|
9008 put_time(fd, spin->si_sugtime); /* <timestamp> */ |
| 625 | 9009 |
| 9010 /* | |
| 9011 * <SUGWORDTREE> | |
| 9012 */ | |
| 9013 spin->si_memtot = 0; | |
| 9014 tree = spin->si_foldroot->wn_sibling; | |
| 9015 | |
| 9016 /* Clear the index and wnode fields in the tree. */ | |
| 9017 clear_node(tree); | |
| 9018 | |
| 9019 /* Count the number of nodes. Needed to be able to allocate the | |
| 9020 * memory when reading the nodes. Also fills in index for shared | |
| 9021 * nodes. */ | |
| 9022 nodecount = put_node(NULL, tree, 0, 0, FALSE); | |
| 9023 | |
| 9024 /* number of nodes in 4 bytes */ | |
| 9025 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */ | |
| 9026 spin->si_memtot += nodecount + nodecount * sizeof(int); | |
| 9027 | |
| 9028 /* Write the nodes. */ | |
| 9029 (void)put_node(fd, tree, 0, 0, FALSE); | |
| 9030 | |
| 9031 /* | |
| 9032 * <SUGTABLE>: <sugwcount> <sugline> ... | |
| 9033 */ | |
| 9034 wcount = spin->si_spellbuf->b_ml.ml_line_count; | |
| 9035 put_bytes(fd, (long_u)wcount, 4); /* <sugwcount> */ | |
| 9036 | |
| 9037 for (lnum = 1; lnum <= (linenr_T)wcount; ++lnum) | |
| 9038 { | |
| 9039 /* <sugline>: <sugnr> ... NUL */ | |
| 9040 line = ml_get_buf(spin->si_spellbuf, lnum, FALSE); | |
| 835 | 9041 len = (int)STRLEN(line) + 1; |
| 625 | 9042 if (fwrite(line, (size_t)len, (size_t)1, fd) == 0) |
| 9043 { | |
| 9044 EMSG(_(e_write)); | |
| 9045 goto theend; | |
| 9046 } | |
| 9047 spin->si_memtot += len; | |
| 9048 } | |
| 9049 | |
| 9050 /* Write another byte to check for errors. */ | |
| 9051 if (putc(0, fd) == EOF) | |
| 9052 EMSG(_(e_write)); | |
| 9053 | |
| 9054 vim_snprintf((char *)IObuff, IOSIZE, | |
| 9055 _("Estimated runtime memory use: %d bytes"), spin->si_memtot); | |
| 9056 spell_message(spin, IObuff); | |
| 9057 | |
| 9058 theend: | |
| 9059 /* close the file */ | |
| 9060 fclose(fd); | |
| 9061 } | |
| 9062 | |
| 9063 /* | |
| 9064 * Open a spell buffer. This is a nameless buffer that is not in the buffer | |
| 9065 * list and only contains text lines. Can use a swapfile to reduce memory | |
| 9066 * use. | |
| 9067 * Most other fields are invalid! Esp. watch out for string options being | |
| 9068 * NULL and there is no undo info. | |
| 9069 * Returns NULL when out of memory. | |
| 9070 */ | |
| 9071 static buf_T * | |
| 9072 open_spellbuf() | |
| 9073 { | |
| 9074 buf_T *buf; | |
| 9075 | |
| 9076 buf = (buf_T *)alloc_clear(sizeof(buf_T)); | |
| 9077 if (buf != NULL) | |
| 9078 { | |
| 9079 buf->b_spell = TRUE; | |
| 9080 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
|
9081 #ifdef FEAT_CRYPT |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
9082 buf->b_p_key = empty_option; |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
9083 #endif |
| 625 | 9084 ml_open(buf); |
| 9085 ml_open_file(buf); /* create swap file now */ | |
| 9086 } | |
| 9087 return buf; | |
| 9088 } | |
| 9089 | |
| 9090 /* | |
| 9091 * Close the buffer used for spell info. | |
| 9092 */ | |
| 9093 static void | |
| 9094 close_spellbuf(buf) | |
| 9095 buf_T *buf; | |
| 9096 { | |
| 9097 if (buf != NULL) | |
| 9098 { | |
| 9099 ml_close(buf, TRUE); | |
| 9100 vim_free(buf); | |
| 9101 } | |
| 9102 } | |
| 9103 | |
| 9104 | |
| 9105 /* | |
| 310 | 9106 * Create a Vim spell file from one or more word lists. |
| 9107 * "fnames[0]" is the output file name. | |
| 9108 * "fnames[fcount - 1]" is the last input file name. | |
| 9109 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name | |
| 9110 * and ".spl" is appended to make the output file name. | |
| 9111 */ | |
| 9112 static void | |
| 3263 | 9113 mkspell(fcount, fnames, ascii, over_write, added_word) |
| 310 | 9114 int fcount; |
| 9115 char_u **fnames; | |
| 9116 int ascii; /* -ascii argument given */ | |
| 3263 | 9117 int over_write; /* overwrite existing output file */ |
| 323 | 9118 int added_word; /* invoked through "zg" */ |
| 310 | 9119 { |
| 2770 | 9120 char_u *fname = NULL; |
| 9121 char_u *wfname; | |
| 310 | 9122 char_u **innames; |
| 9123 int incount; | |
| 236 | 9124 afffile_T *(afile[8]); |
| 9125 int i; | |
| 9126 int len; | |
| 9127 struct stat st; | |
| 255 | 9128 int error = FALSE; |
| 300 | 9129 spellinfo_T spin; |
| 9130 | |
| 9131 vim_memset(&spin, 0, sizeof(spin)); | |
| 323 | 9132 spin.si_verbose = !added_word; |
| 310 | 9133 spin.si_ascii = ascii; |
| 323 | 9134 spin.si_followup = TRUE; |
| 9135 spin.si_rem_accents = TRUE; | |
| 9136 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20); | |
| 625 | 9137 ga_init2(&spin.si_repsal, (int)sizeof(fromto_T), 20); |
| 323 | 9138 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20); |
| 9139 ga_init2(&spin.si_map, (int)sizeof(char_u), 100); | |
| 809 | 9140 ga_init2(&spin.si_comppat, (int)sizeof(char_u *), 20); |
| 339 | 9141 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50); |
| 625 | 9142 hash_init(&spin.si_commonwords); |
| 500 | 9143 spin.si_newcompID = 127; /* start compound ID at first maximum */ |
| 310 | 9144 |
| 9145 /* default: fnames[0] is output file, following are input files */ | |
| 9146 innames = &fnames[1]; | |
| 9147 incount = fcount - 1; | |
| 9148 | |
| 2770 | 9149 wfname = alloc(MAXPATHL); |
| 9150 if (wfname == NULL) | |
| 9151 return; | |
| 9152 | |
| 310 | 9153 if (fcount >= 1) |
| 240 | 9154 { |
| 835 | 9155 len = (int)STRLEN(fnames[0]); |
| 310 | 9156 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0) |
| 9157 { | |
| 9158 /* For ":mkspell path/en.latin1.add" output file is | |
| 9159 * "path/en.latin1.add.spl". */ | |
| 9160 innames = &fnames[0]; | |
| 9161 incount = 1; | |
| 2770 | 9162 vim_snprintf((char *)wfname, MAXPATHL, "%s.spl", fnames[0]); |
| 310 | 9163 } |
| 366 | 9164 else if (fcount == 1) |
| 9165 { | |
| 9166 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */ | |
| 9167 innames = &fnames[0]; | |
| 9168 incount = 1; | |
| 2770 | 9169 vim_snprintf((char *)wfname, MAXPATHL, SPL_FNAME_TMPL, |
| 2660 | 9170 fnames[0], spin.si_ascii ? (char_u *)"ascii" : spell_enc()); |
| 366 | 9171 } |
| 310 | 9172 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0) |
| 9173 { | |
| 9174 /* Name ends in ".spl", use as the file name. */ | |
| 2770 | 9175 vim_strncpy(wfname, fnames[0], MAXPATHL - 1); |
| 310 | 9176 } |
| 9177 else | |
| 9178 /* Name should be language, make the file name from it. */ | |
| 2770 | 9179 vim_snprintf((char *)wfname, MAXPATHL, SPL_FNAME_TMPL, |
| 2660 | 9180 fnames[0], spin.si_ascii ? (char_u *)"ascii" : spell_enc()); |
| 310 | 9181 |
| 9182 /* Check for .ascii.spl. */ | |
| 2660 | 9183 if (strstr((char *)gettail(wfname), SPL_FNAME_ASCII) != NULL) |
| 310 | 9184 spin.si_ascii = TRUE; |
| 9185 | |
| 9186 /* Check for .add.spl. */ | |
| 2660 | 9187 if (strstr((char *)gettail(wfname), SPL_FNAME_ADD) != NULL) |
| 310 | 9188 spin.si_add = TRUE; |
| 240 | 9189 } |
| 9190 | |
| 310 | 9191 if (incount <= 0) |
| 236 | 9192 EMSG(_(e_invarg)); /* need at least output and input names */ |
| 351 | 9193 else if (vim_strchr(gettail(wfname), '_') != NULL) |
| 9194 EMSG(_("E751: Output file name must not have region name")); | |
| 310 | 9195 else if (incount > 8) |
| 236 | 9196 EMSG(_("E754: Only up to 8 regions supported")); |
| 9197 else | |
| 9198 { | |
| 9199 /* Check for overwriting before doing things that may take a lot of | |
| 9200 * time. */ | |
| 3263 | 9201 if (!over_write && mch_stat((char *)wfname, &st) >= 0) |
| 236 | 9202 { |
| 9203 EMSG(_(e_exists)); | |
| 2770 | 9204 goto theend; |
| 236 | 9205 } |
| 310 | 9206 if (mch_isdir(wfname)) |
| 236 | 9207 { |
| 310 | 9208 EMSG2(_(e_isadir2), wfname); |
| 2770 | 9209 goto theend; |
| 9210 } | |
| 9211 | |
| 9212 fname = alloc(MAXPATHL); | |
| 9213 if (fname == NULL) | |
| 9214 goto theend; | |
| 236 | 9215 |
| 9216 /* | |
| 9217 * Init the aff and dic pointers. | |
| 9218 * Get the region names if there are more than 2 arguments. | |
| 9219 */ | |
| 310 | 9220 for (i = 0; i < incount; ++i) |
| 236 | 9221 { |
| 310 | 9222 afile[i] = NULL; |
| 300 | 9223 |
| 316 | 9224 if (incount > 1) |
| 236 | 9225 { |
| 835 | 9226 len = (int)STRLEN(innames[i]); |
| 310 | 9227 if (STRLEN(gettail(innames[i])) < 5 |
| 9228 || innames[i][len - 3] != '_') | |
| 236 | 9229 { |
| 310 | 9230 EMSG2(_("E755: Invalid region in %s"), innames[i]); |
| 2770 | 9231 goto theend; |
| 236 | 9232 } |
| 316 | 9233 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]); |
| 9234 spin.si_region_name[i * 2 + 1] = | |
| 9235 TOLOWER_ASC(innames[i][len - 1]); | |
| 236 | 9236 } |
| 9237 } | |
| 316 | 9238 spin.si_region_count = incount; |
| 236 | 9239 |
| 474 | 9240 spin.si_foldroot = wordtree_alloc(&spin); |
| 9241 spin.si_keeproot = wordtree_alloc(&spin); | |
| 9242 spin.si_prefroot = wordtree_alloc(&spin); | |
| 339 | 9243 if (spin.si_foldroot == NULL |
| 9244 || spin.si_keeproot == NULL | |
| 9245 || spin.si_prefroot == NULL) | |
| 300 | 9246 { |
| 469 | 9247 free_blocks(spin.si_blocks); |
| 2770 | 9248 goto theend; |
| 300 | 9249 } |
| 9250 | |
| 351 | 9251 /* When not producing a .add.spl file clear the character table when |
| 9252 * we encounter one in the .aff file. This means we dump the current | |
| 9253 * one in the .spl file if the .aff file doesn't define one. That's | |
| 9254 * better than guessing the contents, the table will match a | |
| 9255 * previously loaded spell file. */ | |
| 9256 if (!spin.si_add) | |
| 9257 spin.si_clear_chartab = TRUE; | |
| 9258 | |
| 236 | 9259 /* |
| 9260 * Read all the .aff and .dic files. | |
| 9261 * Text is converted to 'encoding'. | |
| 300 | 9262 * Words are stored in the case-folded and keep-case trees. |
| 236 | 9263 */ |
| 310 | 9264 for (i = 0; i < incount && !error; ++i) |
| 236 | 9265 { |
| 300 | 9266 spin.si_conv.vc_type = CONV_NONE; |
| 310 | 9267 spin.si_region = 1 << i; |
| 9268 | |
| 2770 | 9269 vim_snprintf((char *)fname, MAXPATHL, "%s.aff", innames[i]); |
| 300 | 9270 if (mch_stat((char *)fname, &st) >= 0) |
| 9271 { | |
| 9272 /* Read the .aff file. Will init "spin->si_conv" based on the | |
| 9273 * "SET" line. */ | |
| 474 | 9274 afile[i] = spell_read_aff(&spin, fname); |
| 310 | 9275 if (afile[i] == NULL) |
| 300 | 9276 error = TRUE; |
| 9277 else | |
| 9278 { | |
| 9279 /* Read the .dic file and store the words in the trees. */ | |
| 2770 | 9280 vim_snprintf((char *)fname, MAXPATHL, "%s.dic", |
| 310 | 9281 innames[i]); |
| 474 | 9282 if (spell_read_dic(&spin, fname, afile[i]) == FAIL) |
| 300 | 9283 error = TRUE; |
| 9284 } | |
| 9285 } | |
| 9286 else | |
| 9287 { | |
| 9288 /* No .aff file, try reading the file as a word list. Store | |
| 9289 * the words in the trees. */ | |
| 474 | 9290 if (spell_read_wordfile(&spin, innames[i]) == FAIL) |
| 300 | 9291 error = TRUE; |
| 9292 } | |
| 236 | 9293 |
| 310 | 9294 #ifdef FEAT_MBYTE |
| 236 | 9295 /* Free any conversion stuff. */ |
| 300 | 9296 convert_setup(&spin.si_conv, NULL, NULL); |
| 310 | 9297 #endif |
| 236 | 9298 } |
| 9299 | |
| 492 | 9300 if (spin.si_compflags != NULL && spin.si_nobreak) |
| 9301 MSG(_("Warning: both compounding and NOBREAK specified")); | |
| 9302 | |
| 625 | 9303 if (!error && !got_int) |
| 236 | 9304 { |
| 9305 /* | |
| 300 | 9306 * Combine tails in the tree. |
| 236 | 9307 */ |
| 625 | 9308 spell_message(&spin, (char_u *)_(msg_compressing)); |
| 474 | 9309 wordtree_compress(&spin, spin.si_foldroot); |
| 9310 wordtree_compress(&spin, spin.si_keeproot); | |
| 9311 wordtree_compress(&spin, spin.si_prefroot); | |
| 236 | 9312 } |
| 9313 | |
| 625 | 9314 if (!error && !got_int) |
| 300 | 9315 { |
| 9316 /* | |
| 9317 * Write the info in the spell file. | |
| 9318 */ | |
| 625 | 9319 vim_snprintf((char *)IObuff, IOSIZE, |
| 9320 _("Writing spell file %s ..."), wfname); | |
| 9321 spell_message(&spin, IObuff); | |
| 310 | 9322 |
| 500 | 9323 error = write_vim_spell(&spin, wfname) == FAIL; |
| 474 | 9324 |
| 625 | 9325 spell_message(&spin, (char_u *)_("Done!")); |
| 9326 vim_snprintf((char *)IObuff, IOSIZE, | |
| 9327 _("Estimated runtime memory use: %d bytes"), spin.si_memtot); | |
| 9328 spell_message(&spin, IObuff); | |
| 9329 | |
| 9330 /* | |
| 9331 * If the file is loaded need to reload it. | |
| 9332 */ | |
| 500 | 9333 if (!error) |
| 9334 spell_reload_one(wfname, added_word); | |
| 300 | 9335 } |
| 9336 | |
| 9337 /* Free the allocated memory. */ | |
| 323 | 9338 ga_clear(&spin.si_rep); |
| 625 | 9339 ga_clear(&spin.si_repsal); |
| 323 | 9340 ga_clear(&spin.si_sal); |
| 9341 ga_clear(&spin.si_map); | |
| 809 | 9342 ga_clear(&spin.si_comppat); |
| 339 | 9343 ga_clear(&spin.si_prefcond); |
| 625 | 9344 hash_clear_all(&spin.si_commonwords, 0); |
| 300 | 9345 |
| 9346 /* Free the .aff file structures. */ | |
| 310 | 9347 for (i = 0; i < incount; ++i) |
| 9348 if (afile[i] != NULL) | |
| 9349 spell_free_aff(afile[i]); | |
| 339 | 9350 |
| 9351 /* Free all the bits and pieces at once. */ | |
| 9352 free_blocks(spin.si_blocks); | |
| 625 | 9353 |
| 9354 /* | |
| 9355 * If there is soundfolding info and no NOSUGFILE item create the | |
| 9356 * .sug file with the soundfolded word trie. | |
| 9357 */ | |
| 9358 if (spin.si_sugtime != 0 && !error && !got_int) | |
| 9359 spell_make_sugfile(&spin, wfname); | |
| 9360 | |
| 9361 } | |
| 2770 | 9362 |
| 9363 theend: | |
| 9364 vim_free(fname); | |
| 9365 vim_free(wfname); | |
| 625 | 9366 } |
| 9367 | |
| 9368 /* | |
| 9369 * Display a message for spell file processing when 'verbose' is set or using | |
| 9370 * ":mkspell". "str" can be IObuff. | |
| 9371 */ | |
| 9372 static void | |
| 9373 spell_message(spin, str) | |
| 9374 spellinfo_T *spin; | |
| 9375 char_u *str; | |
| 9376 { | |
| 9377 if (spin->si_verbose || p_verbose > 2) | |
| 9378 { | |
| 9379 if (!spin->si_verbose) | |
| 9380 verbose_enter(); | |
| 9381 MSG(str); | |
| 9382 out_flush(); | |
| 9383 if (!spin->si_verbose) | |
| 9384 verbose_leave(); | |
| 9385 } | |
| 9386 } | |
| 310 | 9387 |
| 9388 /* | |
| 385 | 9389 * ":[count]spellgood {word}" |
| 9390 * ":[count]spellwrong {word}" | |
| 710 | 9391 * ":[count]spellundo {word}" |
| 310 | 9392 */ |
| 9393 void | |
| 9394 ex_spell(eap) | |
| 9395 exarg_T *eap; | |
| 9396 { | |
| 835 | 9397 spell_add_word(eap->arg, (int)STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong, |
| 710 | 9398 eap->forceit ? 0 : (int)eap->line2, |
| 9399 eap->cmdidx == CMD_spellundo); | |
| 236 | 9400 } |
| 9401 | |
| 310 | 9402 /* |
| 9403 * Add "word[len]" to 'spellfile' as a good or bad word. | |
| 9404 */ | |
| 9405 void | |
| 944 | 9406 spell_add_word(word, len, bad, idx, undo) |
| 310 | 9407 char_u *word; |
| 9408 int len; | |
| 9409 int bad; | |
| 944 | 9410 int idx; /* "zG" and "zW": zero, otherwise index in |
| 385 | 9411 'spellfile' */ |
| 710 | 9412 int undo; /* TRUE for "zug", "zuG", "zuw" and "zuW" */ |
| 310 | 9413 { |
| 977 | 9414 FILE *fd = NULL; |
| 385 | 9415 buf_T *buf = NULL; |
| 351 | 9416 int new_spf = FALSE; |
| 381 | 9417 char_u *fname; |
| 2770 | 9418 char_u *fnamebuf = NULL; |
| 385 | 9419 char_u line[MAXWLEN * 2]; |
| 9420 long fpos, fpos_next = 0; | |
| 9421 int i; | |
| 9422 char_u *spf; | |
| 9423 | |
| 944 | 9424 if (idx == 0) /* use internal wordlist */ |
| 385 | 9425 { |
| 9426 if (int_wordlist == NULL) | |
| 9427 { | |
| 9428 int_wordlist = vim_tempname('s'); | |
| 9429 if (int_wordlist == NULL) | |
| 381 | 9430 return; |
| 9431 } | |
| 385 | 9432 fname = int_wordlist; |
| 381 | 9433 } |
| 310 | 9434 else |
| 9435 { | |
| 381 | 9436 /* 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
|
9437 if (*curwin->w_s->b_p_spf == NUL) |
| 381 | 9438 { |
| 9439 init_spellfile(); | |
| 9440 new_spf = TRUE; | |
| 9441 } | |
| 9442 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9443 if (*curwin->w_s->b_p_spf == NUL) |
| 381 | 9444 { |
| 523 | 9445 EMSG2(_(e_notset), "spellfile"); |
| 381 | 9446 return; |
| 9447 } | |
| 2770 | 9448 fnamebuf = alloc(MAXPATHL); |
| 9449 if (fnamebuf == NULL) | |
| 9450 return; | |
| 381 | 9451 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9452 for (spf = curwin->w_s->b_p_spf, i = 1; *spf != NUL; ++i) |
| 385 | 9453 { |
| 9454 copy_option_part(&spf, fnamebuf, MAXPATHL, ","); | |
| 944 | 9455 if (i == idx) |
| 385 | 9456 break; |
| 9457 if (*spf == NUL) | |
| 9458 { | |
| 944 | 9459 EMSGN(_("E765: 'spellfile' does not have %ld entries"), idx); |
| 2770 | 9460 vim_free(fnamebuf); |
| 385 | 9461 return; |
| 9462 } | |
| 9463 } | |
| 9464 | |
| 310 | 9465 /* Check that the user isn't editing the .add file somewhere. */ |
| 385 | 9466 buf = buflist_findname_exp(fnamebuf); |
| 310 | 9467 if (buf != NULL && buf->b_ml.ml_mfp == NULL) |
| 9468 buf = NULL; | |
| 9469 if (buf != NULL && bufIsChanged(buf)) | |
| 381 | 9470 { |
| 310 | 9471 EMSG(_(e_bufloaded)); |
| 2770 | 9472 vim_free(fnamebuf); |
| 381 | 9473 return; |
| 9474 } | |
| 9475 | |
| 385 | 9476 fname = fnamebuf; |
| 9477 } | |
| 9478 | |
| 710 | 9479 if (bad || undo) |
| 9480 { | |
| 9481 /* When the word appears as good word we need to remove that one, | |
| 385 | 9482 * since its flags sort before the one with WF_BANNED. */ |
| 9483 fd = mch_fopen((char *)fname, "r"); | |
| 9484 if (fd != NULL) | |
| 9485 { | |
| 9486 while (!vim_fgets(line, MAXWLEN * 2, fd)) | |
| 9487 { | |
| 9488 fpos = fpos_next; | |
| 9489 fpos_next = ftell(fd); | |
| 9490 if (STRNCMP(word, line, len) == 0 | |
| 9491 && (line[len] == '/' || line[len] < ' ')) | |
| 9492 { | |
| 9493 /* Found duplicate word. Remove it by writing a '#' at | |
| 9494 * the start of the line. Mixing reading and writing | |
| 9495 * doesn't work for all systems, close the file first. */ | |
| 9496 fclose(fd); | |
| 9497 fd = mch_fopen((char *)fname, "r+"); | |
| 9498 if (fd == NULL) | |
| 9499 break; | |
| 9500 if (fseek(fd, fpos, SEEK_SET) == 0) | |
| 710 | 9501 { |
| 385 | 9502 fputc('#', fd); |
| 710 | 9503 if (undo) |
| 974 | 9504 { |
| 9505 home_replace(NULL, fname, NameBuff, MAXPATHL, TRUE); | |
| 5374 | 9506 smsg((char_u *)_("Word '%.*s' removed from %s"), |
| 9507 len, word, NameBuff); | |
| 974 | 9508 } |
| 710 | 9509 } |
| 385 | 9510 fseek(fd, fpos_next, SEEK_SET); |
| 9511 } | |
| 9512 } | |
|
2445
04dae202d316
Fixes for coverity warnings.
Bram Moolenaar <bram@vim.org>
parents:
2299
diff
changeset
|
9513 if (fd != NULL) |
|
04dae202d316
Fixes for coverity warnings.
Bram Moolenaar <bram@vim.org>
parents:
2299
diff
changeset
|
9514 fclose(fd); |
| 385 | 9515 } |
| 381 | 9516 } |
| 975 | 9517 |
| 9518 if (!undo) | |
| 710 | 9519 { |
| 9520 fd = mch_fopen((char *)fname, "a"); | |
| 9521 if (fd == NULL && new_spf) | |
| 9522 { | |
| 975 | 9523 char_u *p; |
| 9524 | |
| 710 | 9525 /* We just initialized the 'spellfile' option and can't open the |
| 9526 * file. We may need to create the "spell" directory first. We | |
| 9527 * already checked the runtime directory is writable in | |
| 9528 * init_spellfile(). */ | |
| 975 | 9529 if (!dir_of_file_exists(fname) && (p = gettail_sep(fname)) != fname) |
| 9530 { | |
| 9531 int c = *p; | |
| 9532 | |
| 710 | 9533 /* The directory doesn't exist. Try creating it and opening |
| 9534 * the file again. */ | |
| 975 | 9535 *p = NUL; |
| 9536 vim_mkdir(fname, 0755); | |
| 9537 *p = c; | |
| 710 | 9538 fd = mch_fopen((char *)fname, "a"); |
| 9539 } | |
| 9540 } | |
| 9541 | |
| 9542 if (fd == NULL) | |
| 9543 EMSG2(_(e_notopen), fname); | |
| 310 | 9544 else |
| 710 | 9545 { |
| 9546 if (bad) | |
| 9547 fprintf(fd, "%.*s/!\n", len, word); | |
| 9548 else | |
| 9549 fprintf(fd, "%.*s\n", len, word); | |
| 9550 fclose(fd); | |
| 9551 | |
| 9552 home_replace(NULL, fname, NameBuff, MAXPATHL, TRUE); | |
| 5374 | 9553 smsg((char_u *)_("Word '%.*s' added to %s"), len, word, NameBuff); |
| 710 | 9554 } |
| 9555 } | |
| 9556 | |
| 9557 if (fd != NULL) | |
| 9558 { | |
| 381 | 9559 /* Update the .add.spl file. */ |
| 9560 mkspell(1, &fname, FALSE, TRUE, TRUE); | |
| 9561 | |
| 9562 /* If the .add file is edited somewhere, reload it. */ | |
| 9563 if (buf != NULL) | |
| 627 | 9564 buf_reload(buf, buf->b_orig_mode); |
| 381 | 9565 |
| 737 | 9566 redraw_all_later(SOME_VALID); |
| 310 | 9567 } |
| 2770 | 9568 vim_free(fnamebuf); |
| 310 | 9569 } |
| 9570 | |
| 9571 /* | |
| 9572 * Initialize 'spellfile' for the current buffer. | |
| 9573 */ | |
| 9574 static void | |
| 9575 init_spellfile() | |
| 9576 { | |
| 2770 | 9577 char_u *buf; |
| 310 | 9578 int l; |
| 500 | 9579 char_u *fname; |
| 310 | 9580 char_u *rtp; |
| 351 | 9581 char_u *lend; |
| 501 | 9582 int aspath = FALSE; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9583 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
|
9584 |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9585 if (*curwin->w_s->b_p_spl != NUL && curwin->w_s->b_langp.ga_len > 0) |
| 310 | 9586 { |
| 2770 | 9587 buf = alloc(MAXPATHL); |
| 9588 if (buf == NULL) | |
| 9589 return; | |
| 9590 | |
| 501 | 9591 /* Find the end of the language name. Exclude the region. If there |
| 9592 * 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
|
9593 for (lend = curwin->w_s->b_p_spl; *lend != NUL |
| 351 | 9594 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend) |
| 501 | 9595 if (vim_ispathsep(*lend)) |
| 9596 { | |
| 9597 aspath = TRUE; | |
| 9598 lstart = lend + 1; | |
| 9599 } | |
| 351 | 9600 |
| 9601 /* Loop over all entries in 'runtimepath'. Use the first one where we | |
| 9602 * are allowed to write. */ | |
| 310 | 9603 rtp = p_rtp; |
| 9604 while (*rtp != NUL) | |
| 9605 { | |
| 501 | 9606 if (aspath) |
| 9607 /* Use directory of an entry with path, e.g., for | |
| 9608 * "/dir/lg.utf-8.spl" use "/dir". */ | |
| 2593 | 9609 vim_strncpy(buf, curbuf->b_s.b_p_spl, |
| 9610 lstart - curbuf->b_s.b_p_spl - 1); | |
| 501 | 9611 else |
| 9612 /* Copy the path from 'runtimepath' to buf[]. */ | |
| 9613 copy_option_part(&rtp, buf, MAXPATHL, ","); | |
| 310 | 9614 if (filewritable(buf) == 2) |
| 9615 { | |
| 316 | 9616 /* Use the first language name from 'spelllang' and the |
| 9617 * encoding used in the first loaded .spl file. */ | |
| 501 | 9618 if (aspath) |
| 2593 | 9619 vim_strncpy(buf, curbuf->b_s.b_p_spl, |
| 9620 lend - curbuf->b_s.b_p_spl); | |
| 501 | 9621 else |
| 9622 { | |
| 819 | 9623 /* Create the "spell" directory if it doesn't exist yet. */ |
| 835 | 9624 l = (int)STRLEN(buf); |
| 819 | 9625 vim_snprintf((char *)buf + l, MAXPATHL - l, "/spell"); |
| 2593 | 9626 if (filewritable(buf) != 2) |
| 819 | 9627 vim_mkdir(buf, 0755); |
| 9628 | |
| 835 | 9629 l = (int)STRLEN(buf); |
| 501 | 9630 vim_snprintf((char *)buf + l, MAXPATHL - l, |
| 819 | 9631 "/%.*s", (int)(lend - lstart), lstart); |
| 501 | 9632 } |
| 835 | 9633 l = (int)STRLEN(buf); |
| 2770 | 9634 fname = LANGP_ENTRY(curwin->w_s->b_langp, 0) |
| 9635 ->lp_slang->sl_fname; | |
| 501 | 9636 vim_snprintf((char *)buf + l, MAXPATHL - l, ".%s.add", |
| 9637 fname != NULL | |
| 9638 && strstr((char *)gettail(fname), ".ascii.") != NULL | |
| 9639 ? (char_u *)"ascii" : spell_enc()); | |
| 310 | 9640 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL); |
| 9641 break; | |
| 9642 } | |
| 501 | 9643 aspath = FALSE; |
| 310 | 9644 } |
| 2770 | 9645 |
| 9646 vim_free(buf); | |
| 310 | 9647 } |
| 9648 } | |
| 236 | 9649 |
| 300 | 9650 |
| 307 | 9651 /* |
| 9652 * Init the chartab used for spelling for ASCII. | |
| 9653 * EBCDIC is not supported! | |
| 9654 */ | |
| 9655 static void | |
| 9656 clear_spell_chartab(sp) | |
| 9657 spelltab_T *sp; | |
| 9658 { | |
| 324 | 9659 int i; |
| 307 | 9660 |
| 9661 /* Init everything to FALSE. */ | |
| 9662 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw)); | |
| 9663 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu)); | |
| 9664 for (i = 0; i < 256; ++i) | |
| 324 | 9665 { |
| 307 | 9666 sp->st_fold[i] = i; |
| 324 | 9667 sp->st_upper[i] = i; |
| 9668 } | |
| 307 | 9669 |
| 9670 /* We include digits. A word shouldn't start with a digit, but handling | |
| 9671 * that is done separately. */ | |
| 9672 for (i = '0'; i <= '9'; ++i) | |
| 9673 sp->st_isw[i] = TRUE; | |
| 9674 for (i = 'A'; i <= 'Z'; ++i) | |
| 9675 { | |
| 9676 sp->st_isw[i] = TRUE; | |
| 9677 sp->st_isu[i] = TRUE; | |
| 9678 sp->st_fold[i] = i + 0x20; | |
| 9679 } | |
| 9680 for (i = 'a'; i <= 'z'; ++i) | |
| 324 | 9681 { |
| 307 | 9682 sp->st_isw[i] = TRUE; |
| 324 | 9683 sp->st_upper[i] = i - 0x20; |
| 9684 } | |
| 307 | 9685 } |
| 9686 | |
| 9687 /* | |
| 9688 * Init the chartab used for spelling. Only depends on 'encoding'. | |
| 9689 * Called once while starting up and when 'encoding' changes. | |
| 9690 * The default is to use isalpha(), but the spell file should define the word | |
| 9691 * characters to make it possible that 'encoding' differs from the current | |
| 390 | 9692 * locale. For utf-8 we don't use isalpha() but our own functions. |
| 307 | 9693 */ |
| 9694 void | |
| 9695 init_spell_chartab() | |
| 9696 { | |
| 9697 int i; | |
| 9698 | |
| 9699 did_set_spelltab = FALSE; | |
| 9700 clear_spell_chartab(&spelltab); | |
| 9701 #ifdef FEAT_MBYTE | |
| 9702 if (enc_dbcs) | |
| 9703 { | |
| 9704 /* DBCS: assume double-wide characters are word characters. */ | |
| 9705 for (i = 128; i <= 255; ++i) | |
| 9706 if (MB_BYTE2LEN(i) == 2) | |
| 9707 spelltab.st_isw[i] = TRUE; | |
| 9708 } | |
| 324 | 9709 else if (enc_utf8) |
| 9710 { | |
| 9711 for (i = 128; i < 256; ++i) | |
| 9712 { | |
|
2140
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9713 int f = utf_fold(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9714 int u = utf_toupper(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9715 |
| 324 | 9716 spelltab.st_isu[i] = utf_isupper(i); |
| 9717 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
|
9718 /* 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
|
9719 * 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
|
9720 * value for utf-8 to avoid this. */ |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9721 spelltab.st_fold[i] = (f < 256) ? f : i; |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
9722 spelltab.st_upper[i] = (u < 256) ? u : i; |
| 324 | 9723 } |
| 9724 } | |
| 307 | 9725 else |
| 9726 #endif | |
| 9727 { | |
| 324 | 9728 /* Rough guess: use locale-dependent library functions. */ |
| 307 | 9729 for (i = 128; i < 256; ++i) |
| 9730 { | |
| 9731 if (MB_ISUPPER(i)) | |
| 9732 { | |
| 324 | 9733 spelltab.st_isw[i] = TRUE; |
| 307 | 9734 spelltab.st_isu[i] = TRUE; |
| 9735 spelltab.st_fold[i] = MB_TOLOWER(i); | |
| 9736 } | |
| 324 | 9737 else if (MB_ISLOWER(i)) |
| 9738 { | |
| 9739 spelltab.st_isw[i] = TRUE; | |
| 9740 spelltab.st_upper[i] = MB_TOUPPER(i); | |
| 9741 } | |
| 307 | 9742 } |
| 9743 } | |
| 9744 } | |
| 9745 | |
| 9746 /* | |
| 9747 * Set the spell character tables from strings in the affix file. | |
| 9748 */ | |
| 9749 static int | |
| 9750 set_spell_chartab(fol, low, upp) | |
| 9751 char_u *fol; | |
| 9752 char_u *low; | |
| 9753 char_u *upp; | |
| 9754 { | |
| 9755 /* We build the new tables here first, so that we can compare with the | |
| 9756 * previous one. */ | |
| 9757 spelltab_T new_st; | |
| 9758 char_u *pf = fol, *pl = low, *pu = upp; | |
| 9759 int f, l, u; | |
| 9760 | |
| 9761 clear_spell_chartab(&new_st); | |
| 9762 | |
| 9763 while (*pf != NUL) | |
| 9764 { | |
| 9765 if (*pl == NUL || *pu == NUL) | |
| 9766 { | |
| 9767 EMSG(_(e_affform)); | |
| 9768 return FAIL; | |
| 9769 } | |
| 9770 #ifdef FEAT_MBYTE | |
| 9771 f = mb_ptr2char_adv(&pf); | |
| 9772 l = mb_ptr2char_adv(&pl); | |
| 9773 u = mb_ptr2char_adv(&pu); | |
| 9774 #else | |
| 9775 f = *pf++; | |
| 9776 l = *pl++; | |
| 9777 u = *pu++; | |
| 9778 #endif | |
| 9779 /* Every character that appears is a word character. */ | |
| 9780 if (f < 256) | |
| 9781 new_st.st_isw[f] = TRUE; | |
| 9782 if (l < 256) | |
| 9783 new_st.st_isw[l] = TRUE; | |
| 9784 if (u < 256) | |
| 9785 new_st.st_isw[u] = TRUE; | |
| 9786 | |
| 9787 /* if "LOW" and "FOL" are not the same the "LOW" char needs | |
| 9788 * case-folding */ | |
| 9789 if (l < 256 && l != f) | |
| 9790 { | |
| 9791 if (f >= 256) | |
| 9792 { | |
| 9793 EMSG(_(e_affrange)); | |
| 9794 return FAIL; | |
| 9795 } | |
| 9796 new_st.st_fold[l] = f; | |
| 9797 } | |
| 9798 | |
| 9799 /* if "UPP" and "FOL" are not the same the "UPP" char needs | |
| 324 | 9800 * case-folding, it's upper case and the "UPP" is the upper case of |
| 9801 * "FOL" . */ | |
| 307 | 9802 if (u < 256 && u != f) |
| 9803 { | |
| 9804 if (f >= 256) | |
| 9805 { | |
| 9806 EMSG(_(e_affrange)); | |
| 9807 return FAIL; | |
| 9808 } | |
| 9809 new_st.st_fold[u] = f; | |
| 9810 new_st.st_isu[u] = TRUE; | |
| 324 | 9811 new_st.st_upper[f] = u; |
| 307 | 9812 } |
| 9813 } | |
| 9814 | |
| 9815 if (*pl != NUL || *pu != NUL) | |
| 9816 { | |
| 9817 EMSG(_(e_affform)); | |
| 9818 return FAIL; | |
| 9819 } | |
| 9820 | |
| 9821 return set_spell_finish(&new_st); | |
| 9822 } | |
| 9823 | |
| 9824 /* | |
| 9825 * Set the spell character tables from strings in the .spl file. | |
| 9826 */ | |
| 483 | 9827 static void |
| 389 | 9828 set_spell_charflags(flags, cnt, fol) |
| 307 | 9829 char_u *flags; |
| 389 | 9830 int cnt; /* length of "flags" */ |
| 9831 char_u *fol; | |
| 307 | 9832 { |
| 9833 /* We build the new tables here first, so that we can compare with the | |
| 9834 * previous one. */ | |
| 9835 spelltab_T new_st; | |
| 9836 int i; | |
| 389 | 9837 char_u *p = fol; |
| 324 | 9838 int c; |
| 307 | 9839 |
| 9840 clear_spell_chartab(&new_st); | |
| 9841 | |
| 389 | 9842 for (i = 0; i < 128; ++i) |
| 9843 { | |
| 9844 if (i < cnt) | |
| 9845 { | |
| 9846 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0; | |
| 9847 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0; | |
| 9848 } | |
| 9849 | |
| 9850 if (*p != NUL) | |
| 9851 { | |
| 307 | 9852 #ifdef FEAT_MBYTE |
| 389 | 9853 c = mb_ptr2char_adv(&p); |
| 307 | 9854 #else |
| 389 | 9855 c = *p++; |
| 307 | 9856 #endif |
| 389 | 9857 new_st.st_fold[i + 128] = c; |
| 9858 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256) | |
| 9859 new_st.st_upper[c] = i + 128; | |
| 9860 } | |
| 307 | 9861 } |
| 9862 | |
| 483 | 9863 (void)set_spell_finish(&new_st); |
| 307 | 9864 } |
| 9865 | |
| 9866 static int | |
| 9867 set_spell_finish(new_st) | |
| 9868 spelltab_T *new_st; | |
| 9869 { | |
| 9870 int i; | |
| 9871 | |
| 9872 if (did_set_spelltab) | |
| 9873 { | |
| 9874 /* check that it's the same table */ | |
| 9875 for (i = 0; i < 256; ++i) | |
| 9876 { | |
| 9877 if (spelltab.st_isw[i] != new_st->st_isw[i] | |
| 9878 || spelltab.st_isu[i] != new_st->st_isu[i] | |
| 324 | 9879 || spelltab.st_fold[i] != new_st->st_fold[i] |
| 9880 || spelltab.st_upper[i] != new_st->st_upper[i]) | |
| 307 | 9881 { |
| 9882 EMSG(_("E763: Word characters differ between spell files")); | |
| 9883 return FAIL; | |
| 9884 } | |
| 9885 } | |
| 9886 } | |
| 9887 else | |
| 9888 { | |
| 9889 /* copy the new spelltab into the one being used */ | |
| 9890 spelltab = *new_st; | |
| 9891 did_set_spelltab = TRUE; | |
| 9892 } | |
| 9893 | |
| 9894 return OK; | |
| 9895 } | |
| 9896 | |
| 9897 /* | |
| 358 | 9898 * Return TRUE if "p" points to a word character. |
| 366 | 9899 * As a special case we see "midword" characters as word character when it is |
| 358 | 9900 * followed by a word character. This finds they'there but not 'they there'. |
| 366 | 9901 * Thus this only works properly when past the first character of the word. |
| 358 | 9902 */ |
| 9903 static int | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9904 spell_iswordp(p, wp) |
| 358 | 9905 char_u *p; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9906 win_T *wp; /* buffer used */ |
| 358 | 9907 { |
| 366 | 9908 #ifdef FEAT_MBYTE |
| 358 | 9909 char_u *s; |
| 366 | 9910 int l; |
| 9911 int c; | |
| 9912 | |
| 9913 if (has_mbyte) | |
| 9914 { | |
| 9915 l = MB_BYTE2LEN(*p); | |
| 358 | 9916 s = p; |
| 366 | 9917 if (l == 1) |
| 9918 { | |
| 9919 /* be quick for ASCII */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9920 if (wp->w_s->b_spell_ismw[*p]) |
| 366 | 9921 s = p + 1; /* skip a mid-word character */ |
| 9922 } | |
| 9923 else | |
| 9924 { | |
| 9925 c = mb_ptr2char(p); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9926 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
|
9927 : (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
|
9928 && vim_strchr(wp->w_s->b_spell_ismw_mb, c) != NULL)) |
| 366 | 9929 s = p + l; |
| 9930 } | |
| 9931 | |
| 390 | 9932 c = mb_ptr2char(s); |
| 9933 if (c > 255) | |
| 5477 | 9934 return spell_mb_isword_class(mb_get_class(s), wp); |
| 390 | 9935 return spelltab.st_isw[c]; |
| 366 | 9936 } |
| 358 | 9937 #endif |
| 366 | 9938 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9939 return spelltab.st_isw[wp->w_s->b_spell_ismw[*p] ? p[1] : p[0]]; |
| 376 | 9940 } |
| 9941 | |
| 9942 /* | |
| 9943 * Return TRUE if "p" points to a word character. | |
| 9944 * Unlike spell_iswordp() this doesn't check for "midword" characters. | |
| 9945 */ | |
| 9946 static int | |
| 5477 | 9947 spell_iswordp_nmw(p, wp) |
| 376 | 9948 char_u *p; |
| 5477 | 9949 win_T *wp; |
| 376 | 9950 { |
| 9951 #ifdef FEAT_MBYTE | |
| 390 | 9952 int c; |
| 9953 | |
| 9954 if (has_mbyte) | |
| 9955 { | |
| 9956 c = mb_ptr2char(p); | |
| 9957 if (c > 255) | |
| 5477 | 9958 return spell_mb_isword_class(mb_get_class(p), wp); |
| 390 | 9959 return spelltab.st_isw[c]; |
| 9960 } | |
| 376 | 9961 #endif |
| 9962 return spelltab.st_isw[*p]; | |
| 358 | 9963 } |
| 9964 | |
| 372 | 9965 #ifdef FEAT_MBYTE |
| 9966 /* | |
| 1580 | 9967 * Return TRUE if word class indicates a word character. |
| 9968 * Only for characters above 255. | |
| 9969 * Unicode subscript and superscript are not considered word characters. | |
| 5477 | 9970 * See also dbcs_class() and utf_class() in mbyte.c. |
| 9971 */ | |
| 9972 static int | |
| 9973 spell_mb_isword_class(cl, wp) | |
| 9974 int cl; | |
| 9975 win_T *wp; | |
| 9976 { | |
| 9977 if (wp->w_s->b_cjk) | |
| 9978 /* East Asian characters are not considered word characters. */ | |
| 9979 return cl == 2 || cl == 0x2800; | |
| 1580 | 9980 return cl >= 2 && cl != 0x2070 && cl != 0x2080; |
| 9981 } | |
| 9982 | |
| 9983 /* | |
| 372 | 9984 * Return TRUE if "p" points to a word character. |
| 9985 * Wide version of spell_iswordp(). | |
| 9986 */ | |
| 9987 static int | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9988 spell_iswordp_w(p, wp) |
| 372 | 9989 int *p; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9990 win_T *wp; |
| 372 | 9991 { |
| 9992 int *s; | |
| 9993 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
9994 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
|
9995 : (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
|
9996 && vim_strchr(wp->w_s->b_spell_ismw_mb, *p) != NULL)) |
| 372 | 9997 s = p + 1; |
| 9998 else | |
| 9999 s = p; | |
| 10000 | |
| 390 | 10001 if (*s > 255) |
| 372 | 10002 { |
| 10003 if (enc_utf8) | |
| 5477 | 10004 return spell_mb_isword_class(utf_class(*s), wp); |
| 372 | 10005 if (enc_dbcs) |
| 5477 | 10006 return spell_mb_isword_class( |
| 10007 dbcs_class((unsigned)*s >> 8, *s & 0xff), wp); | |
| 372 | 10008 return 0; |
| 10009 } | |
| 10010 return spelltab.st_isw[*s]; | |
| 10011 } | |
| 10012 #endif | |
| 10013 | |
| 358 | 10014 /* |
| 339 | 10015 * Write the table with prefix conditions to the .spl file. |
| 483 | 10016 * When "fd" is NULL only count the length of what is written. |
| 10017 */ | |
| 10018 static int | |
| 339 | 10019 write_spell_prefcond(fd, gap) |
| 10020 FILE *fd; | |
| 10021 garray_T *gap; | |
| 10022 { | |
| 10023 int i; | |
| 10024 char_u *p; | |
| 10025 int len; | |
| 483 | 10026 int totlen; |
| 1759 | 10027 size_t x = 1; /* collect return value of fwrite() */ |
| 483 | 10028 |
| 10029 if (fd != NULL) | |
| 10030 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */ | |
| 10031 | |
| 10032 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */ | |
| 339 | 10033 |
| 10034 for (i = 0; i < gap->ga_len; ++i) | |
| 10035 { | |
| 10036 /* <prefcond> : <condlen> <condstr> */ | |
| 10037 p = ((char_u **)gap->ga_data)[i]; | |
| 483 | 10038 if (p != NULL) |
| 339 | 10039 { |
| 835 | 10040 len = (int)STRLEN(p); |
| 483 | 10041 if (fd != NULL) |
| 10042 { | |
| 10043 fputc(len, fd); | |
| 1753 | 10044 x &= fwrite(p, (size_t)len, (size_t)1, fd); |
| 483 | 10045 } |
| 10046 totlen += len; | |
| 10047 } | |
| 10048 else if (fd != NULL) | |
| 10049 fputc(0, fd); | |
| 10050 } | |
| 10051 | |
| 10052 return totlen; | |
| 307 | 10053 } |
| 10054 | |
| 10055 /* | |
| 324 | 10056 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated. |
| 10057 * Uses the character definitions from the .spl file. | |
| 307 | 10058 * When using a multi-byte 'encoding' the length may change! |
| 10059 * Returns FAIL when something wrong. | |
| 10060 */ | |
| 10061 static int | |
| 324 | 10062 spell_casefold(str, len, buf, buflen) |
| 10063 char_u *str; | |
| 307 | 10064 int len; |
| 10065 char_u *buf; | |
| 10066 int buflen; | |
| 10067 { | |
| 10068 int i; | |
| 10069 | |
| 10070 if (len >= buflen) | |
| 10071 { | |
| 10072 buf[0] = NUL; | |
| 10073 return FAIL; /* result will not fit */ | |
| 10074 } | |
| 10075 | |
| 10076 #ifdef FEAT_MBYTE | |
| 10077 if (has_mbyte) | |
| 10078 { | |
| 324 | 10079 int outi = 0; |
| 10080 char_u *p; | |
| 307 | 10081 int c; |
| 10082 | |
| 10083 /* Fold one character at a time. */ | |
| 324 | 10084 for (p = str; p < str + len; ) |
| 307 | 10085 { |
| 10086 if (outi + MB_MAXBYTES > buflen) | |
| 10087 { | |
| 10088 buf[outi] = NUL; | |
| 10089 return FAIL; | |
| 10090 } | |
| 474 | 10091 c = mb_cptr2char_adv(&p); |
| 324 | 10092 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi); |
| 307 | 10093 } |
| 10094 buf[outi] = NUL; | |
| 10095 } | |
| 10096 else | |
| 10097 #endif | |
| 10098 { | |
| 10099 /* Be quick for non-multibyte encodings. */ | |
| 10100 for (i = 0; i < len; ++i) | |
| 324 | 10101 buf[i] = spelltab.st_fold[str[i]]; |
| 307 | 10102 buf[i] = NUL; |
| 10103 } | |
| 10104 | |
| 10105 return OK; | |
| 10106 } | |
| 10107 | |
| 625 | 10108 /* values for sps_flags */ |
| 372 | 10109 #define SPS_BEST 1 |
| 10110 #define SPS_FAST 2 | |
| 10111 #define SPS_DOUBLE 4 | |
| 10112 | |
| 625 | 10113 static int sps_flags = SPS_BEST; /* flags from 'spellsuggest' */ |
| 10114 static int sps_limit = 9999; /* max nr of suggestions given */ | |
| 372 | 10115 |
| 10116 /* | |
| 10117 * Check the 'spellsuggest' option. Return FAIL if it's wrong. | |
| 483 | 10118 * Sets "sps_flags" and "sps_limit". |
| 372 | 10119 */ |
| 10120 int | |
| 10121 spell_check_sps() | |
| 10122 { | |
| 10123 char_u *p; | |
| 483 | 10124 char_u *s; |
| 372 | 10125 char_u buf[MAXPATHL]; |
| 10126 int f; | |
| 10127 | |
| 10128 sps_flags = 0; | |
| 483 | 10129 sps_limit = 9999; |
| 372 | 10130 |
| 10131 for (p = p_sps; *p != NUL; ) | |
| 10132 { | |
| 10133 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 10134 | |
| 10135 f = 0; | |
| 483 | 10136 if (VIM_ISDIGIT(*buf)) |
| 10137 { | |
| 10138 s = buf; | |
| 10139 sps_limit = getdigits(&s); | |
| 10140 if (*s != NUL && !VIM_ISDIGIT(*s)) | |
| 10141 f = -1; | |
| 10142 } | |
| 10143 else if (STRCMP(buf, "best") == 0) | |
| 372 | 10144 f = SPS_BEST; |
| 10145 else if (STRCMP(buf, "fast") == 0) | |
| 10146 f = SPS_FAST; | |
| 10147 else if (STRCMP(buf, "double") == 0) | |
| 10148 f = SPS_DOUBLE; | |
| 10149 else if (STRNCMP(buf, "expr:", 5) != 0 | |
| 10150 && STRNCMP(buf, "file:", 5) != 0) | |
| 10151 f = -1; | |
| 10152 | |
| 10153 if (f == -1 || (sps_flags != 0 && f != 0)) | |
| 10154 { | |
| 10155 sps_flags = SPS_BEST; | |
| 483 | 10156 sps_limit = 9999; |
| 372 | 10157 return FAIL; |
| 10158 } | |
| 10159 if (f != 0) | |
| 10160 sps_flags = f; | |
| 10161 } | |
| 10162 | |
| 10163 if (sps_flags == 0) | |
| 10164 sps_flags = SPS_BEST; | |
| 10165 | |
| 10166 return OK; | |
| 10167 } | |
| 10168 | |
| 323 | 10169 /* |
| 5374 | 10170 * "z=": Find badly spelled word under or after the cursor. |
| 323 | 10171 * Give suggestions for the properly spelled word. |
| 638 | 10172 * In Visual mode use the highlighted word as the bad word. |
| 485 | 10173 * When "count" is non-zero use that suggestion. |
| 323 | 10174 */ |
| 10175 void | |
| 485 | 10176 spell_suggest(count) |
| 10177 int count; | |
| 323 | 10178 { |
| 10179 char_u *line; | |
| 10180 pos_T prev_cursor = curwin->w_cursor; | |
| 10181 char_u wcopy[MAXWLEN + 2]; | |
| 10182 char_u *p; | |
| 10183 int i; | |
| 10184 int c; | |
| 10185 suginfo_T sug; | |
| 10186 suggest_T *stp; | |
| 372 | 10187 int mouse_used; |
| 388 | 10188 int need_cap; |
| 483 | 10189 int limit; |
| 485 | 10190 int selected = count; |
| 638 | 10191 int badlen = 0; |
| 1938 | 10192 int msg_scroll_save = msg_scroll; |
| 638 | 10193 |
| 10194 if (no_spell_checking(curwin)) | |
| 10195 return; | |
| 10196 | |
| 10197 if (VIsual_active) | |
| 10198 { | |
| 10199 /* Use the Visually selected text as the bad word. But reject | |
| 10200 * a multi-line selection. */ | |
| 10201 if (curwin->w_cursor.lnum != VIsual.lnum) | |
| 10202 { | |
| 10203 vim_beep(); | |
| 10204 return; | |
| 10205 } | |
| 10206 badlen = (int)curwin->w_cursor.col - (int)VIsual.col; | |
| 10207 if (badlen < 0) | |
| 10208 badlen = -badlen; | |
| 10209 else | |
| 10210 curwin->w_cursor.col = VIsual.col; | |
| 10211 ++badlen; | |
| 10212 end_visual_mode(); | |
| 10213 } | |
| 5735 | 10214 /* Find the start of the badly spelled word. */ |
| 10215 else if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0 | |
| 346 | 10216 || curwin->w_cursor.col > prev_cursor.col) |
| 10217 { | |
| 10218 /* No bad word or it starts after the cursor: use the word under the | |
| 10219 * cursor. */ | |
| 10220 curwin->w_cursor = prev_cursor; | |
| 10221 line = ml_get_curline(); | |
| 10222 p = line + curwin->w_cursor.col; | |
| 10223 /* Backup to before start of word. */ | |
| 5477 | 10224 while (p > line && spell_iswordp_nmw(p, curwin)) |
| 346 | 10225 mb_ptr_back(line, p); |
| 10226 /* Forward to start of word. */ | |
| 5477 | 10227 while (*p != NUL && !spell_iswordp_nmw(p, curwin)) |
| 346 | 10228 mb_ptr_adv(p); |
| 10229 | |
| 5477 | 10230 if (!spell_iswordp_nmw(p, curwin)) /* No word found. */ |
| 346 | 10231 { |
| 10232 beep_flush(); | |
| 10233 return; | |
| 10234 } | |
| 835 | 10235 curwin->w_cursor.col = (colnr_T)(p - line); |
| 323 | 10236 } |
| 10237 | |
| 344 | 10238 /* Get the word and its length. */ |
| 10239 | |
| 388 | 10240 /* Figure out if the word should be capitalised. */ |
| 475 | 10241 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col); |
| 10242 | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10243 /* 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
|
10244 line = vim_strsave(ml_get_curline()); |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10245 if (line == NULL) |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10246 goto skip; |
| 388 | 10247 |
| 483 | 10248 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in |
| 10249 * 'spellsuggest', whatever is smaller. */ | |
| 10250 if (sps_limit > (int)Rows - 2) | |
| 10251 limit = (int)Rows - 2; | |
| 10252 else | |
| 10253 limit = sps_limit; | |
| 638 | 10254 spell_find_suggest(line + curwin->w_cursor.col, badlen, &sug, limit, |
| 625 | 10255 TRUE, need_cap, TRUE); |
| 323 | 10256 |
| 10257 if (sug.su_ga.ga_len == 0) | |
| 10258 MSG(_("Sorry, no suggestions")); | |
| 485 | 10259 else if (count > 0) |
| 10260 { | |
| 10261 if (count > sug.su_ga.ga_len) | |
| 10262 smsg((char_u *)_("Sorry, only %ld suggestions"), | |
| 10263 (long)sug.su_ga.ga_len); | |
| 10264 } | |
| 323 | 10265 else |
| 10266 { | |
| 372 | 10267 vim_free(repl_from); |
| 10268 repl_from = NULL; | |
| 10269 vim_free(repl_to); | |
| 10270 repl_to = NULL; | |
| 10271 | |
| 474 | 10272 #ifdef FEAT_RIGHTLEFT |
| 10273 /* When 'rightleft' is set the list is drawn right-left. */ | |
| 10274 cmdmsg_rl = curwin->w_p_rl; | |
| 10275 if (cmdmsg_rl) | |
| 10276 msg_col = Columns - 1; | |
| 10277 #endif | |
| 10278 | |
| 323 | 10279 /* List the suggestions. */ |
| 10280 msg_start(); | |
| 914 | 10281 msg_row = Rows - 1; /* for when 'cmdheight' > 1 */ |
| 372 | 10282 lines_left = Rows; /* avoid more prompt */ |
| 323 | 10283 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"), |
| 10284 sug.su_badlen, sug.su_badptr); | |
| 474 | 10285 #ifdef FEAT_RIGHTLEFT |
| 10286 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0) | |
| 10287 { | |
| 10288 /* And now the rabbit from the high hat: Avoid showing the | |
| 10289 * untranslated message rightleft. */ | |
| 10290 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC", | |
| 10291 sug.su_badlen, sug.su_badptr); | |
| 10292 } | |
| 10293 #endif | |
| 323 | 10294 msg_puts(IObuff); |
| 10295 msg_clr_eos(); | |
| 10296 msg_putchar('\n'); | |
| 346 | 10297 |
| 323 | 10298 msg_scroll = TRUE; |
| 10299 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 10300 { | |
| 344 | 10301 stp = &SUG(sug.su_ga, i); |
| 323 | 10302 |
| 10303 /* The suggested word may replace only part of the bad word, add | |
| 10304 * the not replaced part. */ | |
| 2768 | 10305 vim_strncpy(wcopy, stp->st_word, MAXWLEN); |
| 323 | 10306 if (sug.su_badlen > stp->st_orglen) |
| 625 | 10307 vim_strncpy(wcopy + stp->st_wordlen, |
| 323 | 10308 sug.su_badptr + stp->st_orglen, |
| 10309 sug.su_badlen - stp->st_orglen); | |
| 474 | 10310 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1); |
| 10311 #ifdef FEAT_RIGHTLEFT | |
| 10312 if (cmdmsg_rl) | |
| 10313 rl_mirror(IObuff); | |
| 10314 #endif | |
| 10315 msg_puts(IObuff); | |
| 10316 | |
| 10317 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy); | |
| 346 | 10318 msg_puts(IObuff); |
| 10319 | |
| 10320 /* The word may replace more than "su_badlen". */ | |
| 10321 if (sug.su_badlen < stp->st_orglen) | |
| 10322 { | |
| 10323 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""), | |
| 10324 stp->st_orglen, sug.su_badptr); | |
| 10325 msg_puts(IObuff); | |
| 10326 } | |
| 10327 | |
| 324 | 10328 if (p_verbose > 0) |
| 344 | 10329 { |
| 346 | 10330 /* Add the score. */ |
| 351 | 10331 if (sps_flags & (SPS_DOUBLE | SPS_BEST)) |
| 474 | 10332 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)", |
| 344 | 10333 stp->st_salscore ? "s " : "", |
| 10334 stp->st_score, stp->st_altscore); | |
| 10335 else | |
| 474 | 10336 vim_snprintf((char *)IObuff, IOSIZE, " (%d)", |
| 346 | 10337 stp->st_score); |
| 474 | 10338 #ifdef FEAT_RIGHTLEFT |
| 10339 if (cmdmsg_rl) | |
| 10340 /* Mirror the numbers, but keep the leading space. */ | |
| 10341 rl_mirror(IObuff + 1); | |
| 10342 #endif | |
| 346 | 10343 msg_advance(30); |
| 10344 msg_puts(IObuff); | |
| 344 | 10345 } |
| 323 | 10346 msg_putchar('\n'); |
| 10347 } | |
| 10348 | |
| 474 | 10349 #ifdef FEAT_RIGHTLEFT |
| 10350 cmdmsg_rl = FALSE; | |
| 10351 msg_col = 0; | |
| 10352 #endif | |
| 323 | 10353 /* Ask for choice. */ |
| 485 | 10354 selected = prompt_for_number(&mouse_used); |
| 372 | 10355 if (mouse_used) |
| 485 | 10356 selected -= lines_left; |
| 1938 | 10357 lines_left = Rows; /* avoid more prompt */ |
| 10358 /* don't delay for 'smd' in normal_cmd() */ | |
| 10359 msg_scroll = msg_scroll_save; | |
| 485 | 10360 } |
| 10361 | |
| 10362 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK) | |
| 10363 { | |
| 10364 /* Save the from and to text for :spellrepall. */ | |
| 10365 stp = &SUG(sug.su_ga, selected - 1); | |
| 548 | 10366 if (sug.su_badlen > stp->st_orglen) |
| 10367 { | |
| 10368 /* Replacing less than "su_badlen", append the remainder to | |
| 10369 * repl_to. */ | |
| 10370 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen); | |
| 10371 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word, | |
| 10372 sug.su_badlen - stp->st_orglen, | |
| 10373 sug.su_badptr + stp->st_orglen); | |
| 10374 repl_to = vim_strsave(IObuff); | |
| 10375 } | |
| 10376 else | |
| 10377 { | |
| 10378 /* Replacing su_badlen or more, use the whole word. */ | |
| 10379 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen); | |
| 10380 repl_to = vim_strsave(stp->st_word); | |
| 10381 } | |
| 485 | 10382 |
| 10383 /* Replace the word. */ | |
| 1938 | 10384 p = alloc((unsigned)STRLEN(line) - stp->st_orglen |
| 10385 + stp->st_wordlen + 1); | |
| 485 | 10386 if (p != NULL) |
| 10387 { | |
| 835 | 10388 c = (int)(sug.su_badptr - line); |
| 485 | 10389 mch_memmove(p, line, c); |
| 10390 STRCPY(p + c, stp->st_word); | |
| 10391 STRCAT(p, sug.su_badptr + stp->st_orglen); | |
| 10392 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 10393 curwin->w_cursor.col = c; | |
| 10394 | |
| 10395 /* For redo we use a change-word command. */ | |
| 10396 ResetRedobuff(); | |
| 10397 AppendToRedobuff((char_u *)"ciw"); | |
| 620 | 10398 AppendToRedobuffLit(p + c, |
| 625 | 10399 stp->st_wordlen + sug.su_badlen - stp->st_orglen); |
| 485 | 10400 AppendCharToRedobuff(ESC); |
| 819 | 10401 |
| 10402 /* After this "p" may be invalid. */ | |
| 10403 changed_bytes(curwin->w_cursor.lnum, c); | |
| 485 | 10404 } |
| 10405 } | |
| 10406 else | |
| 10407 curwin->w_cursor = prev_cursor; | |
| 323 | 10408 |
| 344 | 10409 spell_find_cleanup(&sug); |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10410 skip: |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10411 vim_free(line); |
| 344 | 10412 } |
| 10413 | |
| 10414 /* | |
| 475 | 10415 * Check if the word at line "lnum" column "col" is required to start with a |
| 10416 * capital. This uses 'spellcapcheck' of the current buffer. | |
| 10417 */ | |
| 10418 static int | |
| 10419 check_need_cap(lnum, col) | |
| 10420 linenr_T lnum; | |
| 10421 colnr_T col; | |
| 10422 { | |
| 10423 int need_cap = FALSE; | |
| 10424 char_u *line; | |
| 10425 char_u *line_copy = NULL; | |
| 10426 char_u *p; | |
| 10427 colnr_T endcol; | |
| 10428 regmatch_T regmatch; | |
| 10429 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10430 if (curwin->w_s->b_cap_prog == NULL) |
| 475 | 10431 return FALSE; |
| 10432 | |
| 10433 line = ml_get_curline(); | |
| 10434 endcol = 0; | |
| 10435 if ((int)(skipwhite(line) - line) >= (int)col) | |
| 10436 { | |
| 10437 /* At start of line, check if previous line is empty or sentence | |
| 10438 * ends there. */ | |
| 10439 if (lnum == 1) | |
| 10440 need_cap = TRUE; | |
| 10441 else | |
| 10442 { | |
| 10443 line = ml_get(lnum - 1); | |
| 10444 if (*skipwhite(line) == NUL) | |
| 10445 need_cap = TRUE; | |
| 10446 else | |
| 10447 { | |
| 10448 /* Append a space in place of the line break. */ | |
| 10449 line_copy = concat_str(line, (char_u *)" "); | |
| 10450 line = line_copy; | |
| 835 | 10451 endcol = (colnr_T)STRLEN(line); |
| 475 | 10452 } |
| 10453 } | |
| 10454 } | |
| 10455 else | |
| 10456 endcol = col; | |
| 10457 | |
| 10458 if (endcol > 0) | |
| 10459 { | |
| 10460 /* 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
|
10461 regmatch.regprog = curwin->w_s->b_cap_prog; |
| 475 | 10462 regmatch.rm_ic = FALSE; |
| 10463 p = line + endcol; | |
| 10464 for (;;) | |
| 10465 { | |
| 10466 mb_ptr_back(line, p); | |
| 5477 | 10467 if (p == line || spell_iswordp_nmw(p, curwin)) |
| 475 | 10468 break; |
| 10469 if (vim_regexec(®match, p, 0) | |
| 10470 && regmatch.endp[0] == line + endcol) | |
| 10471 { | |
| 10472 need_cap = TRUE; | |
| 10473 break; | |
| 10474 } | |
| 10475 } | |
| 6375 | 10476 curwin->w_s->b_cap_prog = regmatch.regprog; |
| 475 | 10477 } |
| 10478 | |
| 10479 vim_free(line_copy); | |
| 10480 | |
| 10481 return need_cap; | |
| 10482 } | |
| 10483 | |
| 10484 | |
| 10485 /* | |
| 372 | 10486 * ":spellrepall" |
| 10487 */ | |
| 10488 void | |
| 10489 ex_spellrepall(eap) | |
| 1883 | 10490 exarg_T *eap UNUSED; |
| 372 | 10491 { |
| 10492 pos_T pos = curwin->w_cursor; | |
| 10493 char_u *frompat; | |
| 10494 int addlen; | |
| 10495 char_u *line; | |
| 10496 char_u *p; | |
| 10497 int save_ws = p_ws; | |
| 483 | 10498 linenr_T prev_lnum = 0; |
| 372 | 10499 |
| 10500 if (repl_from == NULL || repl_to == NULL) | |
| 10501 { | |
| 10502 EMSG(_("E752: No previous spell replacement")); | |
| 10503 return; | |
| 10504 } | |
| 835 | 10505 addlen = (int)(STRLEN(repl_to) - STRLEN(repl_from)); |
| 10506 | |
| 10507 frompat = alloc((unsigned)STRLEN(repl_from) + 7); | |
| 372 | 10508 if (frompat == NULL) |
| 10509 return; | |
| 10510 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from); | |
| 10511 p_ws = FALSE; | |
| 10512 | |
| 483 | 10513 sub_nsubs = 0; |
| 10514 sub_nlines = 0; | |
| 372 | 10515 curwin->w_cursor.lnum = 0; |
| 10516 while (!got_int) | |
| 10517 { | |
| 1521 | 10518 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP, NULL) == 0 |
| 372 | 10519 || u_save_cursor() == FAIL) |
| 10520 break; | |
| 10521 | |
| 10522 /* Only replace when the right word isn't there yet. This happens | |
| 10523 * when changing "etc" to "etc.". */ | |
| 10524 line = ml_get_curline(); | |
| 10525 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col, | |
| 10526 repl_to, STRLEN(repl_to)) != 0) | |
| 10527 { | |
| 835 | 10528 p = alloc((unsigned)STRLEN(line) + addlen + 1); |
| 372 | 10529 if (p == NULL) |
| 10530 break; | |
| 10531 mch_memmove(p, line, curwin->w_cursor.col); | |
| 10532 STRCPY(p + curwin->w_cursor.col, repl_to); | |
| 10533 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from)); | |
| 10534 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 10535 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col); | |
| 483 | 10536 |
| 10537 if (curwin->w_cursor.lnum != prev_lnum) | |
| 10538 { | |
| 10539 ++sub_nlines; | |
| 10540 prev_lnum = curwin->w_cursor.lnum; | |
| 10541 } | |
| 10542 ++sub_nsubs; | |
| 372 | 10543 } |
| 835 | 10544 curwin->w_cursor.col += (colnr_T)STRLEN(repl_to); |
| 372 | 10545 } |
| 10546 | |
| 10547 p_ws = save_ws; | |
| 10548 curwin->w_cursor = pos; | |
| 10549 vim_free(frompat); | |
| 10550 | |
| 483 | 10551 if (sub_nsubs == 0) |
| 372 | 10552 EMSG2(_("E753: Not found: %s"), repl_from); |
| 483 | 10553 else |
| 10554 do_sub_msg(FALSE); | |
| 372 | 10555 } |
| 10556 | |
| 10557 /* | |
| 344 | 10558 * Find spell suggestions for "word". Return them in the growarray "*gap" as |
| 10559 * a list of allocated strings. | |
| 10560 */ | |
| 10561 void | |
| 625 | 10562 spell_suggest_list(gap, word, maxcount, need_cap, interactive) |
| 344 | 10563 garray_T *gap; |
| 10564 char_u *word; | |
| 10565 int maxcount; /* maximum nr of suggestions */ | |
| 475 | 10566 int need_cap; /* 'spellcapcheck' matched */ |
| 625 | 10567 int interactive; |
| 344 | 10568 { |
| 10569 suginfo_T sug; | |
| 10570 int i; | |
| 10571 suggest_T *stp; | |
| 10572 char_u *wcopy; | |
| 10573 | |
| 638 | 10574 spell_find_suggest(word, 0, &sug, maxcount, FALSE, need_cap, interactive); |
| 344 | 10575 |
| 10576 /* Make room in "gap". */ | |
| 10577 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1); | |
| 626 | 10578 if (ga_grow(gap, sug.su_ga.ga_len) == OK) |
| 10579 { | |
| 10580 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 10581 { | |
| 10582 stp = &SUG(sug.su_ga, i); | |
| 10583 | |
| 10584 /* The suggested word may replace only part of "word", add the not | |
| 10585 * replaced part. */ | |
| 10586 wcopy = alloc(stp->st_wordlen | |
| 835 | 10587 + (unsigned)STRLEN(sug.su_badptr + stp->st_orglen) + 1); |
| 626 | 10588 if (wcopy == NULL) |
| 10589 break; | |
| 10590 STRCPY(wcopy, stp->st_word); | |
| 10591 STRCPY(wcopy + stp->st_wordlen, sug.su_badptr + stp->st_orglen); | |
| 10592 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy; | |
| 10593 } | |
| 344 | 10594 } |
| 10595 | |
| 10596 spell_find_cleanup(&sug); | |
| 10597 } | |
| 10598 | |
| 10599 /* | |
| 10600 * Find spell suggestions for the word at the start of "badptr". | |
| 10601 * Return the suggestions in "su->su_ga". | |
| 10602 * The maximum number of suggestions is "maxcount". | |
| 10603 * Note: does use info for the current window. | |
| 10604 * This is based on the mechanisms of Aspell, but completely reimplemented. | |
| 10605 */ | |
| 10606 static void | |
| 638 | 10607 spell_find_suggest(badptr, badlen, su, maxcount, banbadword, need_cap, interactive) |
| 344 | 10608 char_u *badptr; |
| 638 | 10609 int badlen; /* length of bad word or 0 if unknown */ |
| 344 | 10610 suginfo_T *su; |
| 10611 int maxcount; | |
| 358 | 10612 int banbadword; /* don't include badword in suggestions */ |
| 388 | 10613 int need_cap; /* word should start with capital */ |
| 625 | 10614 int interactive; |
| 344 | 10615 { |
| 534 | 10616 hlf_T attr = HLF_COUNT; |
| 372 | 10617 char_u buf[MAXPATHL]; |
| 10618 char_u *p; | |
| 10619 int do_combine = FALSE; | |
| 10620 char_u *sps_copy; | |
| 10621 #ifdef FEAT_EVAL | |
| 10622 static int expr_busy = FALSE; | |
| 10623 #endif | |
| 385 | 10624 int c; |
| 503 | 10625 int i; |
| 10626 langp_T *lp; | |
| 344 | 10627 |
| 10628 /* | |
| 10629 * Set the info in "*su". | |
| 10630 */ | |
| 10631 vim_memset(su, 0, sizeof(suginfo_T)); | |
| 10632 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10); | |
| 10633 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10); | |
| 355 | 10634 if (*badptr == NUL) |
| 10635 return; | |
| 344 | 10636 hash_init(&su->su_banned); |
| 10637 | |
| 10638 su->su_badptr = badptr; | |
| 638 | 10639 if (badlen != 0) |
| 10640 su->su_badlen = badlen; | |
| 10641 else | |
| 10642 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL, FALSE); | |
| 344 | 10643 su->su_maxcount = maxcount; |
| 372 | 10644 su->su_maxscore = SCORE_MAXINIT; |
| 344 | 10645 |
| 10646 if (su->su_badlen >= MAXWLEN) | |
| 10647 su->su_badlen = MAXWLEN - 1; /* just in case */ | |
| 10648 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen); | |
| 10649 (void)spell_casefold(su->su_badptr, su->su_badlen, | |
| 10650 su->su_fbadword, MAXWLEN); | |
| 346 | 10651 /* get caps flags for bad word */ |
| 474 | 10652 su->su_badflags = badword_captype(su->su_badptr, |
| 10653 su->su_badptr + su->su_badlen); | |
| 388 | 10654 if (need_cap) |
| 10655 su->su_badflags |= WF_ONECAP; | |
| 344 | 10656 |
| 503 | 10657 /* Find the default language for sound folding. We simply use the first |
| 10658 * one in 'spelllang' that supports sound folding. That's good for when | |
| 10659 * using multiple files for one language, it's not that bad when mixing | |
| 10660 * languages (e.g., "pl,en"). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10661 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
|
10662 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10663 lp = LANGP_ENTRY(curbuf->b_s.b_langp, i); |
| 503 | 10664 if (lp->lp_sallang != NULL) |
| 10665 { | |
| 10666 su->su_sallang = lp->lp_sallang; | |
| 10667 break; | |
| 10668 } | |
| 10669 } | |
| 10670 | |
| 534 | 10671 /* Soundfold the bad word with the default sound folding, so that we don't |
| 10672 * have to do this many times. */ | |
| 10673 if (su->su_sallang != NULL) | |
| 10674 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE, | |
| 10675 su->su_sal_badword); | |
| 10676 | |
| 385 | 10677 /* If the word is not capitalised and spell_check() doesn't consider the |
| 10678 * word to be bad then it might need to be capitalised. Add a suggestion | |
| 10679 * for that. */ | |
| 455 | 10680 c = PTR2CHAR(su->su_badptr); |
| 534 | 10681 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT) |
| 385 | 10682 { |
| 10683 make_case_word(su->su_badword, buf, WF_ONECAP); | |
| 10684 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE, | |
| 625 | 10685 0, TRUE, su->su_sallang, FALSE); |
| 385 | 10686 } |
| 10687 | |
| 344 | 10688 /* Ban the bad word itself. It may appear in another region. */ |
| 358 | 10689 if (banbadword) |
| 10690 add_banned(su, su->su_badword); | |
| 344 | 10691 |
| 372 | 10692 /* Make a copy of 'spellsuggest', because the expression may change it. */ |
| 10693 sps_copy = vim_strsave(p_sps); | |
| 10694 if (sps_copy == NULL) | |
| 10695 return; | |
| 10696 | |
| 10697 /* Loop over the items in 'spellsuggest'. */ | |
| 10698 for (p = sps_copy; *p != NUL; ) | |
| 10699 { | |
| 10700 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 10701 | |
| 10702 if (STRNCMP(buf, "expr:", 5) == 0) | |
| 10703 { | |
| 10704 #ifdef FEAT_EVAL | |
| 375 | 10705 /* Evaluate an expression. Skip this when called recursively, |
| 10706 * when using spellsuggest() in the expression. */ | |
| 372 | 10707 if (!expr_busy) |
| 10708 { | |
| 10709 expr_busy = TRUE; | |
| 10710 spell_suggest_expr(su, buf + 5); | |
| 10711 expr_busy = FALSE; | |
| 10712 } | |
| 10713 #endif | |
| 10714 } | |
| 10715 else if (STRNCMP(buf, "file:", 5) == 0) | |
| 10716 /* Use list of suggestions in a file. */ | |
| 10717 spell_suggest_file(su, buf + 5); | |
| 10718 else | |
| 10719 { | |
| 10720 /* Use internal method. */ | |
| 625 | 10721 spell_suggest_intern(su, interactive); |
| 372 | 10722 if (sps_flags & SPS_DOUBLE) |
| 10723 do_combine = TRUE; | |
| 10724 } | |
| 10725 } | |
| 10726 | |
| 10727 vim_free(sps_copy); | |
| 10728 | |
| 10729 if (do_combine) | |
| 10730 /* Combine the two list of suggestions. This must be done last, | |
| 10731 * because sorting changes the order again. */ | |
| 10732 score_combine(su); | |
| 10733 } | |
| 10734 | |
| 10735 #ifdef FEAT_EVAL | |
| 10736 /* | |
| 10737 * Find suggestions by evaluating expression "expr". | |
| 10738 */ | |
| 10739 static void | |
| 10740 spell_suggest_expr(su, expr) | |
| 10741 suginfo_T *su; | |
| 10742 char_u *expr; | |
| 10743 { | |
| 10744 list_T *list; | |
| 10745 listitem_T *li; | |
| 10746 int score; | |
| 10747 char_u *p; | |
| 10748 | |
| 10749 /* The work is split up in a few parts to avoid having to export | |
| 10750 * suginfo_T. | |
| 10751 * First evaluate the expression and get the resulting list. */ | |
| 10752 list = eval_spell_expr(su->su_badword, expr); | |
| 10753 if (list != NULL) | |
| 10754 { | |
| 10755 /* Loop over the items in the list. */ | |
| 10756 for (li = list->lv_first; li != NULL; li = li->li_next) | |
| 10757 if (li->li_tv.v_type == VAR_LIST) | |
| 10758 { | |
| 10759 /* Get the word and the score from the items. */ | |
| 10760 score = get_spellword(li->li_tv.vval.v_list, &p); | |
| 625 | 10761 if (score >= 0 && score <= su->su_maxscore) |
| 10762 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 10763 score, 0, TRUE, su->su_sallang, FALSE); | |
| 372 | 10764 } |
| 10765 list_unref(list); | |
| 10766 } | |
| 10767 | |
| 625 | 10768 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 10769 check_suggestions(su, &su->su_ga); | |
| 372 | 10770 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 10771 } | |
| 10772 #endif | |
| 10773 | |
| 10774 /* | |
| 474 | 10775 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'. |
| 372 | 10776 */ |
| 10777 static void | |
| 10778 spell_suggest_file(su, fname) | |
| 10779 suginfo_T *su; | |
| 10780 char_u *fname; | |
| 10781 { | |
| 10782 FILE *fd; | |
| 10783 char_u line[MAXWLEN * 2]; | |
| 10784 char_u *p; | |
| 10785 int len; | |
| 10786 char_u cword[MAXWLEN]; | |
| 10787 | |
| 10788 /* Open the file. */ | |
| 10789 fd = mch_fopen((char *)fname, "r"); | |
| 10790 if (fd == NULL) | |
| 10791 { | |
| 10792 EMSG2(_(e_notopen), fname); | |
| 10793 return; | |
| 10794 } | |
| 10795 | |
| 10796 /* Read it line by line. */ | |
| 10797 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int) | |
| 10798 { | |
| 10799 line_breakcheck(); | |
| 10800 | |
| 10801 p = vim_strchr(line, '/'); | |
| 10802 if (p == NULL) | |
| 10803 continue; /* No Tab found, just skip the line. */ | |
| 10804 *p++ = NUL; | |
| 10805 if (STRICMP(su->su_badword, line) == 0) | |
| 10806 { | |
| 10807 /* Match! Isolate the good word, until CR or NL. */ | |
| 10808 for (len = 0; p[len] >= ' '; ++len) | |
| 10809 ; | |
| 10810 p[len] = NUL; | |
| 10811 | |
| 10812 /* If the suggestion doesn't have specific case duplicate the case | |
| 10813 * of the bad word. */ | |
| 10814 if (captype(p, NULL) == 0) | |
| 10815 { | |
| 10816 make_case_word(p, cword, su->su_badflags); | |
| 10817 p = cword; | |
| 10818 } | |
| 10819 | |
| 10820 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 625 | 10821 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE); |
| 372 | 10822 } |
| 10823 } | |
| 10824 | |
| 10825 fclose(fd); | |
| 10826 | |
| 625 | 10827 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 10828 check_suggestions(su, &su->su_ga); | |
| 372 | 10829 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 10830 } | |
| 10831 | |
| 10832 /* | |
| 10833 * Find suggestions for the internal method indicated by "sps_flags". | |
| 10834 */ | |
| 10835 static void | |
| 625 | 10836 spell_suggest_intern(su, interactive) |
| 372 | 10837 suginfo_T *su; |
| 625 | 10838 int interactive; |
| 10839 { | |
| 10840 /* | |
| 10841 * Load the .sug file(s) that are available and not done yet. | |
| 10842 */ | |
| 10843 suggest_load_files(); | |
| 10844 | |
| 344 | 10845 /* |
| 346 | 10846 * 1. Try special cases, such as repeating a word: "the the" -> "the". |
| 344 | 10847 * |
| 10848 * Set a maximum score to limit the combination of operations that is | |
| 10849 * tried. | |
| 10850 */ | |
| 346 | 10851 suggest_try_special(su); |
| 10852 | |
| 10853 /* | |
| 10854 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries | |
| 10855 * from the .aff file and inserting a space (split the word). | |
| 10856 */ | |
| 10857 suggest_try_change(su); | |
| 344 | 10858 |
| 10859 /* For the resulting top-scorers compute the sound-a-like score. */ | |
| 10860 if (sps_flags & SPS_DOUBLE) | |
| 10861 score_comp_sal(su); | |
| 10862 | |
| 10863 /* | |
| 346 | 10864 * 3. Try finding sound-a-like words. |
| 344 | 10865 */ |
| 625 | 10866 if ((sps_flags & SPS_FAST) == 0) |
| 10867 { | |
| 10868 if (sps_flags & SPS_BEST) | |
| 10869 /* Adjust the word score for the suggestions found so far for how | |
| 10870 * they sounds like. */ | |
| 10871 rescore_suggestions(su); | |
| 10872 | |
| 10873 /* | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
10874 * While going through the soundfold tree "su_maxscore" is the score |
| 625 | 10875 * for the soundfold word, limits the changes that are being tried, |
| 10876 * and "su_sfmaxscore" the rescored score, which is set by | |
| 10877 * cleanup_suggestions(). | |
| 10878 * First find words with a small edit distance, because this is much | |
| 10879 * faster and often already finds the top-N suggestions. If we didn't | |
| 10880 * find many suggestions try again with a higher edit distance. | |
| 10881 * "sl_sounddone" is used to avoid doing the same word twice. | |
| 10882 */ | |
| 10883 suggest_try_soundalike_prep(); | |
| 10884 su->su_maxscore = SCORE_SFMAX1; | |
| 10885 su->su_sfmaxscore = SCORE_MAXINIT * 3; | |
| 346 | 10886 suggest_try_soundalike(su); |
| 625 | 10887 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) |
| 10888 { | |
| 10889 /* We didn't find enough matches, try again, allowing more | |
| 10890 * changes to the soundfold word. */ | |
| 10891 su->su_maxscore = SCORE_SFMAX2; | |
| 10892 suggest_try_soundalike(su); | |
| 10893 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) | |
| 10894 { | |
| 10895 /* Still didn't find enough matches, try again, allowing even | |
| 10896 * more changes to the soundfold word. */ | |
| 10897 su->su_maxscore = SCORE_SFMAX3; | |
| 10898 suggest_try_soundalike(su); | |
| 10899 } | |
| 10900 } | |
| 10901 su->su_maxscore = su->su_sfmaxscore; | |
| 10902 suggest_try_soundalike_finish(); | |
| 10903 } | |
| 10904 | |
| 10905 /* When CTRL-C was hit while searching do show the results. Only clear | |
| 10906 * got_int when using a command, not for spellsuggest(). */ | |
| 344 | 10907 ui_breakcheck(); |
| 625 | 10908 if (interactive && got_int) |
| 344 | 10909 { |
| 10910 (void)vgetc(); | |
| 10911 got_int = FALSE; | |
| 10912 } | |
| 10913 | |
| 372 | 10914 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0) |
| 344 | 10915 { |
| 10916 if (sps_flags & SPS_BEST) | |
| 10917 /* Adjust the word score for how it sounds like. */ | |
| 10918 rescore_suggestions(su); | |
| 10919 | |
| 625 | 10920 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 10921 check_suggestions(su, &su->su_ga); | |
| 372 | 10922 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 344 | 10923 } |
| 10924 } | |
| 10925 | |
| 10926 /* | |
| 625 | 10927 * Load the .sug files for languages that have one and weren't loaded yet. |
| 10928 */ | |
| 10929 static void | |
| 10930 suggest_load_files() | |
| 10931 { | |
| 10932 langp_T *lp; | |
| 10933 int lpi; | |
| 10934 slang_T *slang; | |
| 10935 char_u *dotp; | |
| 10936 FILE *fd; | |
| 10937 char_u buf[MAXWLEN]; | |
| 10938 int i; | |
| 10939 time_t timestamp; | |
| 10940 int wcount; | |
| 10941 int wordnr; | |
| 10942 garray_T ga; | |
| 10943 int c; | |
| 10944 | |
| 10945 /* 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
|
10946 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
|
10947 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
10948 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 10949 slang = lp->lp_slang; |
| 10950 if (slang->sl_sugtime != 0 && !slang->sl_sugloaded) | |
| 10951 { | |
| 10952 /* Change ".spl" to ".sug" and open the file. When the file isn't | |
| 10953 * found silently skip it. Do set "sl_sugloaded" so that we | |
| 10954 * don't try again and again. */ | |
| 10955 slang->sl_sugloaded = TRUE; | |
| 10956 | |
| 10957 dotp = vim_strrchr(slang->sl_fname, '.'); | |
| 10958 if (dotp == NULL || fnamecmp(dotp, ".spl") != 0) | |
| 10959 continue; | |
| 10960 STRCPY(dotp, ".sug"); | |
| 822 | 10961 fd = mch_fopen((char *)slang->sl_fname, "r"); |
| 625 | 10962 if (fd == NULL) |
| 10963 goto nextone; | |
| 10964 | |
| 10965 /* | |
| 10966 * <SUGHEADER>: <fileID> <versionnr> <timestamp> | |
| 10967 */ | |
| 10968 for (i = 0; i < VIMSUGMAGICL; ++i) | |
| 10969 buf[i] = getc(fd); /* <fileID> */ | |
| 10970 if (STRNCMP(buf, VIMSUGMAGIC, VIMSUGMAGICL) != 0) | |
| 10971 { | |
| 626 | 10972 EMSG2(_("E778: This does not look like a .sug file: %s"), |
| 625 | 10973 slang->sl_fname); |
| 10974 goto nextone; | |
| 10975 } | |
| 10976 c = getc(fd); /* <versionnr> */ | |
| 10977 if (c < VIMSUGVERSION) | |
| 10978 { | |
| 626 | 10979 EMSG2(_("E779: Old .sug file, needs to be updated: %s"), |
| 625 | 10980 slang->sl_fname); |
| 10981 goto nextone; | |
| 10982 } | |
| 10983 else if (c > VIMSUGVERSION) | |
| 10984 { | |
| 626 | 10985 EMSG2(_("E780: .sug file is for newer version of Vim: %s"), |
| 625 | 10986 slang->sl_fname); |
| 10987 goto nextone; | |
| 10988 } | |
| 10989 | |
| 10990 /* Check the timestamp, it must be exactly the same as the one in | |
| 10991 * 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
|
10992 timestamp = get8ctime(fd); /* <timestamp> */ |
| 625 | 10993 if (timestamp != slang->sl_sugtime) |
| 10994 { | |
| 626 | 10995 EMSG2(_("E781: .sug file doesn't match .spl file: %s"), |
| 625 | 10996 slang->sl_fname); |
| 10997 goto nextone; | |
| 10998 } | |
| 10999 | |
| 11000 /* | |
| 11001 * <SUGWORDTREE>: <wordtree> | |
| 11002 * Read the trie with the soundfolded words. | |
| 11003 */ | |
| 11004 if (spell_read_tree(fd, &slang->sl_sbyts, &slang->sl_sidxs, | |
| 11005 FALSE, 0) != 0) | |
| 11006 { | |
| 11007 someerror: | |
| 626 | 11008 EMSG2(_("E782: error while reading .sug file: %s"), |
| 625 | 11009 slang->sl_fname); |
| 11010 slang_clear_sug(slang); | |
| 11011 goto nextone; | |
| 11012 } | |
| 11013 | |
| 11014 /* | |
| 11015 * <SUGTABLE>: <sugwcount> <sugline> ... | |
| 11016 * | |
| 11017 * Read the table with word numbers. We use a file buffer for | |
| 11018 * this, because it's so much like a file with lines. Makes it | |
| 11019 * possible to swap the info and save on memory use. | |
| 11020 */ | |
| 11021 slang->sl_sugbuf = open_spellbuf(); | |
| 11022 if (slang->sl_sugbuf == NULL) | |
| 11023 goto someerror; | |
| 11024 /* <sugwcount> */ | |
| 695 | 11025 wcount = get4c(fd); |
| 625 | 11026 if (wcount < 0) |
| 11027 goto someerror; | |
| 11028 | |
| 11029 /* Read all the wordnr lists into the buffer, one NUL terminated | |
| 11030 * list per line. */ | |
| 11031 ga_init2(&ga, 1, 100); | |
| 11032 for (wordnr = 0; wordnr < wcount; ++wordnr) | |
| 11033 { | |
| 11034 ga.ga_len = 0; | |
| 11035 for (;;) | |
| 11036 { | |
| 11037 c = getc(fd); /* <sugline> */ | |
| 11038 if (c < 0 || ga_grow(&ga, 1) == FAIL) | |
| 11039 goto someerror; | |
| 11040 ((char_u *)ga.ga_data)[ga.ga_len++] = c; | |
| 11041 if (c == NUL) | |
| 11042 break; | |
| 11043 } | |
| 11044 if (ml_append_buf(slang->sl_sugbuf, (linenr_T)wordnr, | |
| 11045 ga.ga_data, ga.ga_len, TRUE) == FAIL) | |
| 11046 goto someerror; | |
| 11047 } | |
| 11048 ga_clear(&ga); | |
| 11049 | |
| 11050 /* | |
| 11051 * Need to put word counts in the word tries, so that we can find | |
| 11052 * a word by its number. | |
| 11053 */ | |
| 11054 tree_count_words(slang->sl_fbyts, slang->sl_fidxs); | |
| 11055 tree_count_words(slang->sl_sbyts, slang->sl_sidxs); | |
| 11056 | |
| 11057 nextone: | |
| 11058 if (fd != NULL) | |
| 11059 fclose(fd); | |
| 11060 STRCPY(dotp, ".spl"); | |
| 11061 } | |
| 11062 } | |
| 11063 } | |
| 11064 | |
| 11065 | |
| 11066 /* | |
| 11067 * Fill in the wordcount fields for a trie. | |
| 11068 * Returns the total number of words. | |
| 11069 */ | |
| 11070 static void | |
| 11071 tree_count_words(byts, idxs) | |
| 11072 char_u *byts; | |
| 11073 idx_T *idxs; | |
| 11074 { | |
| 11075 int depth; | |
| 11076 idx_T arridx[MAXWLEN]; | |
| 11077 int curi[MAXWLEN]; | |
| 11078 int c; | |
| 11079 idx_T n; | |
| 11080 int wordcount[MAXWLEN]; | |
| 11081 | |
| 11082 arridx[0] = 0; | |
| 11083 curi[0] = 1; | |
| 11084 wordcount[0] = 0; | |
| 11085 depth = 0; | |
| 11086 while (depth >= 0 && !got_int) | |
| 11087 { | |
| 11088 if (curi[depth] > byts[arridx[depth]]) | |
| 11089 { | |
| 11090 /* Done all bytes at this node, go up one level. */ | |
| 11091 idxs[arridx[depth]] = wordcount[depth]; | |
| 11092 if (depth > 0) | |
| 11093 wordcount[depth - 1] += wordcount[depth]; | |
| 11094 | |
| 11095 --depth; | |
| 11096 fast_breakcheck(); | |
| 11097 } | |
| 11098 else | |
| 11099 { | |
| 11100 /* Do one more byte at this node. */ | |
| 11101 n = arridx[depth] + curi[depth]; | |
| 11102 ++curi[depth]; | |
| 11103 | |
| 11104 c = byts[n]; | |
| 11105 if (c == 0) | |
| 11106 { | |
| 11107 /* End of word, count it. */ | |
| 11108 ++wordcount[depth]; | |
| 11109 | |
| 11110 /* Skip over any other NUL bytes (same word with different | |
| 11111 * flags). */ | |
| 11112 while (byts[n + 1] == 0) | |
| 11113 { | |
| 11114 ++n; | |
| 11115 ++curi[depth]; | |
| 11116 } | |
| 11117 } | |
| 11118 else | |
| 11119 { | |
| 11120 /* Normal char, go one level deeper to count the words. */ | |
| 11121 ++depth; | |
| 11122 arridx[depth] = idxs[n]; | |
| 11123 curi[depth] = 1; | |
| 11124 wordcount[depth] = 0; | |
| 11125 } | |
| 11126 } | |
| 11127 } | |
| 11128 } | |
| 11129 | |
| 11130 /* | |
| 344 | 11131 * Free the info put in "*su" by spell_find_suggest(). |
| 11132 */ | |
| 11133 static void | |
| 11134 spell_find_cleanup(su) | |
| 11135 suginfo_T *su; | |
| 11136 { | |
| 11137 int i; | |
| 11138 | |
| 11139 /* Free the suggestions. */ | |
| 11140 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 11141 vim_free(SUG(su->su_ga, i).st_word); | |
| 11142 ga_clear(&su->su_ga); | |
| 11143 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 11144 vim_free(SUG(su->su_sga, i).st_word); | |
| 11145 ga_clear(&su->su_sga); | |
| 323 | 11146 |
| 11147 /* Free the banned words. */ | |
| 625 | 11148 hash_clear_all(&su->su_banned, 0); |
| 323 | 11149 } |
| 11150 | |
| 11151 /* | |
| 324 | 11152 * Make a copy of "word", with the first letter upper or lower cased, to |
| 11153 * "wcopy[MAXWLEN]". "word" must not be empty. | |
| 11154 * The result is NUL terminated. | |
| 323 | 11155 */ |
| 11156 static void | |
| 324 | 11157 onecap_copy(word, wcopy, upper) |
| 323 | 11158 char_u *word; |
| 11159 char_u *wcopy; | |
| 11160 int upper; /* TRUE: first letter made upper case */ | |
| 11161 { | |
| 11162 char_u *p; | |
| 11163 int c; | |
| 11164 int l; | |
| 11165 | |
| 11166 p = word; | |
| 11167 #ifdef FEAT_MBYTE | |
| 11168 if (has_mbyte) | |
| 474 | 11169 c = mb_cptr2char_adv(&p); |
| 323 | 11170 else |
| 11171 #endif | |
| 11172 c = *p++; | |
| 11173 if (upper) | |
| 324 | 11174 c = SPELL_TOUPPER(c); |
| 323 | 11175 else |
| 324 | 11176 c = SPELL_TOFOLD(c); |
| 323 | 11177 #ifdef FEAT_MBYTE |
| 11178 if (has_mbyte) | |
| 11179 l = mb_char2bytes(c, wcopy); | |
| 11180 else | |
| 11181 #endif | |
| 11182 { | |
| 11183 l = 1; | |
| 11184 wcopy[0] = c; | |
| 11185 } | |
| 376 | 11186 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1); |
| 323 | 11187 } |
| 11188 | |
| 11189 /* | |
| 324 | 11190 * Make a copy of "word" with all the letters upper cased into |
| 11191 * "wcopy[MAXWLEN]". The result is NUL terminated. | |
| 323 | 11192 */ |
| 11193 static void | |
| 11194 allcap_copy(word, wcopy) | |
| 11195 char_u *word; | |
| 11196 char_u *wcopy; | |
| 11197 { | |
| 11198 char_u *s; | |
| 11199 char_u *d; | |
| 11200 int c; | |
| 11201 | |
| 11202 d = wcopy; | |
| 11203 for (s = word; *s != NUL; ) | |
| 11204 { | |
| 11205 #ifdef FEAT_MBYTE | |
| 11206 if (has_mbyte) | |
| 474 | 11207 c = mb_cptr2char_adv(&s); |
| 323 | 11208 else |
| 11209 #endif | |
| 11210 c = *s++; | |
| 492 | 11211 |
| 11212 #ifdef FEAT_MBYTE | |
| 3042 | 11213 /* We only change 0xdf to SS when we are certain latin1 is used. It |
| 492 | 11214 * would cause weird errors in other 8-bit encodings. */ |
| 11215 if (enc_latin1like && c == 0xdf) | |
| 11216 { | |
| 11217 c = 'S'; | |
| 11218 if (d - wcopy >= MAXWLEN - 1) | |
| 11219 break; | |
| 11220 *d++ = c; | |
| 11221 } | |
| 11222 else | |
| 11223 #endif | |
| 11224 c = SPELL_TOUPPER(c); | |
| 323 | 11225 |
| 11226 #ifdef FEAT_MBYTE | |
| 11227 if (has_mbyte) | |
| 11228 { | |
| 11229 if (d - wcopy >= MAXWLEN - MB_MAXBYTES) | |
| 11230 break; | |
| 11231 d += mb_char2bytes(c, d); | |
| 11232 } | |
| 11233 else | |
| 11234 #endif | |
| 11235 { | |
| 11236 if (d - wcopy >= MAXWLEN - 1) | |
| 11237 break; | |
| 11238 *d++ = c; | |
| 11239 } | |
| 11240 } | |
| 11241 *d = NUL; | |
| 11242 } | |
| 11243 | |
| 11244 /* | |
| 346 | 11245 * Try finding suggestions by recognizing specific situations. |
| 11246 */ | |
| 11247 static void | |
| 11248 suggest_try_special(su) | |
| 11249 suginfo_T *su; | |
| 11250 { | |
| 11251 char_u *p; | |
| 376 | 11252 size_t len; |
| 346 | 11253 int c; |
| 11254 char_u word[MAXWLEN]; | |
| 11255 | |
| 11256 /* | |
| 11257 * Recognize a word that is repeated: "the the". | |
| 11258 */ | |
| 11259 p = skiptowhite(su->su_fbadword); | |
| 11260 len = p - su->su_fbadword; | |
| 11261 p = skipwhite(p); | |
| 11262 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0) | |
| 11263 { | |
| 11264 /* Include badflags: if the badword is onecap or allcap | |
| 11265 * use that for the goodword too: "The the" -> "The". */ | |
| 11266 c = su->su_fbadword[len]; | |
| 11267 su->su_fbadword[len] = NUL; | |
| 11268 make_case_word(su->su_fbadword, word, su->su_badflags); | |
| 11269 su->su_fbadword[len] = c; | |
| 534 | 11270 |
| 11271 /* Give a soundalike score of 0, compute the score as if deleting one | |
| 11272 * character. */ | |
| 11273 add_suggestion(su, &su->su_ga, word, su->su_badlen, | |
| 625 | 11274 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE); |
| 346 | 11275 } |
| 11276 } | |
| 11277 | |
| 11278 /* | |
| 323 | 11279 * Try finding suggestions by adding/removing/swapping letters. |
| 11280 */ | |
| 11281 static void | |
| 346 | 11282 suggest_try_change(su) |
| 323 | 11283 suginfo_T *su; |
| 11284 { | |
| 11285 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */ | |
| 625 | 11286 int n; |
| 11287 char_u *p; | |
| 11288 int lpi; | |
| 323 | 11289 langp_T *lp; |
| 11290 | |
| 11291 /* We make a copy of the case-folded bad word, so that we can modify it | |
| 346 | 11292 * to find matches (esp. REP items). Append some more text, changing |
| 11293 * chars after the bad word may help. */ | |
| 323 | 11294 STRCPY(fword, su->su_fbadword); |
| 835 | 11295 n = (int)STRLEN(fword); |
| 346 | 11296 p = su->su_badptr + su->su_badlen; |
| 835 | 11297 (void)spell_casefold(p, (int)STRLEN(p), fword + n, MAXWLEN - n); |
| 323 | 11298 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11299 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
|
11300 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11301 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 482 | 11302 |
| 500 | 11303 /* If reloading a spell file fails it's still in the list but |
| 11304 * everything has been cleared. */ | |
| 625 | 11305 if (lp->lp_slang->sl_fbyts == NULL) |
| 500 | 11306 continue; |
| 11307 | |
| 625 | 11308 /* Try it for this language. Will add possible suggestions. */ |
| 11309 suggest_trie_walk(su, lp, fword, FALSE); | |
| 11310 } | |
| 11311 } | |
| 11312 | |
| 11313 /* Check the maximum score, if we go over it we won't try this change. */ | |
| 11314 #define TRY_DEEPER(su, stack, depth, add) \ | |
| 11315 (stack[depth].ts_score + (add) < su->su_maxscore) | |
| 11316 | |
| 11317 /* | |
| 11318 * Try finding suggestions by adding/removing/swapping letters. | |
| 11319 * | |
| 11320 * This uses a state machine. At each node in the tree we try various | |
| 11321 * operations. When trying if an operation works "depth" is increased and the | |
| 11322 * stack[] is used to store info. This allows combinations, thus insert one | |
| 11323 * character, replace one and delete another. The number of changes is | |
| 11324 * limited by su->su_maxscore. | |
| 11325 * | |
| 11326 * After implementing this I noticed an article by Kemal Oflazer that | |
| 11327 * describes something similar: "Error-tolerant Finite State Recognition with | |
| 11328 * Applications to Morphological Analysis and Spelling Correction" (1996). | |
| 11329 * The implementation in the article is simplified and requires a stack of | |
| 11330 * unknown depth. The implementation here only needs a stack depth equal to | |
| 11331 * the length of the word. | |
| 11332 * | |
| 11333 * This is also used for the sound-folded word, "soundfold" is TRUE then. | |
| 11334 * The mechanism is the same, but we find a match with a sound-folded word | |
| 11335 * that comes from one or more original words. Each of these words may be | |
| 11336 * added, this is done by add_sound_suggest(). | |
| 11337 * Don't use: | |
| 11338 * the prefix tree or the keep-case tree | |
| 11339 * "su->su_badlen" | |
| 11340 * anything to do with upper and lower case | |
| 11341 * anything to do with word or non-word characters ("spell_iswordp()") | |
| 11342 * banned words | |
| 11343 * word flags (rare, region, compounding) | |
| 11344 * word splitting for now | |
| 11345 * "similar_chars()" | |
| 11346 * use "slang->sl_repsal" instead of "lp->lp_replang->sl_rep" | |
| 11347 */ | |
| 11348 static void | |
| 11349 suggest_trie_walk(su, lp, fword, soundfold) | |
| 11350 suginfo_T *su; | |
| 11351 langp_T *lp; | |
| 11352 char_u *fword; | |
| 11353 int soundfold; | |
| 11354 { | |
| 11355 char_u tword[MAXWLEN]; /* good word collected so far */ | |
| 11356 trystate_T stack[MAXWLEN]; | |
| 11357 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
|
11358 * concatenation of prefix compound |
| 625 | 11359 * words and split word. NUL terminated |
| 11360 * when going deeper but not when coming | |
| 11361 * back. */ | |
| 11362 char_u compflags[MAXWLEN]; /* compound flags, one for each word */ | |
| 11363 trystate_T *sp; | |
| 11364 int newscore; | |
| 11365 int score; | |
| 11366 char_u *byts, *fbyts, *pbyts; | |
| 11367 idx_T *idxs, *fidxs, *pidxs; | |
| 11368 int depth; | |
| 11369 int c, c2, c3; | |
| 11370 int n = 0; | |
| 11371 int flags; | |
| 11372 garray_T *gap; | |
| 11373 idx_T arridx; | |
| 11374 int len; | |
| 11375 char_u *p; | |
| 11376 fromto_T *ftp; | |
| 11377 int fl = 0, tl; | |
| 11378 int repextra = 0; /* extra bytes in fword[] from REP item */ | |
| 11379 slang_T *slang = lp->lp_slang; | |
| 11380 int fword_ends; | |
| 11381 int goodword_ends; | |
| 11382 #ifdef DEBUG_TRIEWALK | |
| 11383 /* Stores the name of the change made at each level. */ | |
| 11384 char_u changename[MAXWLEN][80]; | |
| 11385 #endif | |
| 11386 int breakcheckcount = 1000; | |
| 11387 int compound_ok; | |
| 11388 | |
| 11389 /* | |
| 11390 * Go through the whole case-fold tree, try changes at each node. | |
| 11391 * "tword[]" contains the word collected from nodes in the tree. | |
| 11392 * "fword[]" the word we are trying to match with (initially the bad | |
| 11393 * word). | |
| 11394 */ | |
| 11395 depth = 0; | |
| 11396 sp = &stack[0]; | |
| 11397 vim_memset(sp, 0, sizeof(trystate_T)); | |
| 11398 sp->ts_curi = 1; | |
| 11399 | |
| 11400 if (soundfold) | |
| 11401 { | |
| 11402 /* Going through the soundfold tree. */ | |
| 11403 byts = fbyts = slang->sl_sbyts; | |
| 11404 idxs = fidxs = slang->sl_sidxs; | |
| 11405 pbyts = NULL; | |
| 11406 pidxs = NULL; | |
| 11407 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 11408 sp->ts_state = STATE_START; | |
| 11409 } | |
| 11410 else | |
| 11411 { | |
| 330 | 11412 /* |
| 375 | 11413 * When there are postponed prefixes we need to use these first. At |
| 11414 * the end of the prefix we continue in the case-fold tree. | |
| 11415 */ | |
| 482 | 11416 fbyts = slang->sl_fbyts; |
| 11417 fidxs = slang->sl_fidxs; | |
| 11418 pbyts = slang->sl_pbyts; | |
| 11419 pidxs = slang->sl_pidxs; | |
| 375 | 11420 if (pbyts != NULL) |
| 11421 { | |
| 11422 byts = pbyts; | |
| 11423 idxs = pidxs; | |
| 482 | 11424 sp->ts_prefixdepth = PFD_PREFIXTREE; |
| 375 | 11425 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */ |
| 11426 } | |
| 11427 else | |
| 11428 { | |
| 11429 byts = fbyts; | |
| 11430 idxs = fidxs; | |
| 482 | 11431 sp->ts_prefixdepth = PFD_NOPREFIX; |
| 485 | 11432 sp->ts_state = STATE_START; |
| 375 | 11433 } |
| 625 | 11434 } |
| 11435 | |
| 11436 /* | |
| 11437 * Loop to find all suggestions. At each round we either: | |
| 11438 * - For the current state try one operation, advance "ts_curi", | |
| 11439 * increase "depth". | |
| 11440 * - When a state is done go to the next, set "ts_state". | |
| 11441 * - When all states are tried decrease "depth". | |
| 11442 */ | |
| 11443 while (depth >= 0 && !got_int) | |
| 11444 { | |
| 11445 sp = &stack[depth]; | |
| 11446 switch (sp->ts_state) | |
| 11447 { | |
| 11448 case STATE_START: | |
| 11449 case STATE_NOPREFIX: | |
| 11450 /* | |
| 11451 * Start of node: Deal with NUL bytes, which means | |
| 11452 * tword[] may end here. | |
| 11453 */ | |
| 11454 arridx = sp->ts_arridx; /* current node in the tree */ | |
| 11455 len = byts[arridx]; /* bytes in this node */ | |
| 11456 arridx += sp->ts_curi; /* index of current byte */ | |
| 11457 | |
| 11458 if (sp->ts_prefixdepth == PFD_PREFIXTREE) | |
| 11459 { | |
| 11460 /* Skip over the NUL bytes, we use them later. */ | |
| 11461 for (n = 0; n < len && byts[arridx + n] == 0; ++n) | |
| 11462 ; | |
| 11463 sp->ts_curi += n; | |
| 11464 | |
| 11465 /* Always past NUL bytes now. */ | |
| 11466 n = (int)sp->ts_state; | |
| 11467 sp->ts_state = STATE_ENDNUL; | |
| 11468 sp->ts_save_badflags = su->su_badflags; | |
| 11469 | |
| 11470 /* At end of a prefix or at start of prefixtree: check for | |
| 11471 * following word. */ | |
| 11472 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX) | |
| 11473 { | |
| 11474 /* Set su->su_badflags to the caps type at this position. | |
| 11475 * Use the caps type until here for the prefix itself. */ | |
| 11476 #ifdef FEAT_MBYTE | |
| 11477 if (has_mbyte) | |
| 11478 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); | |
| 11479 else | |
| 11480 #endif | |
| 11481 n = sp->ts_fidx; | |
| 11482 flags = badword_captype(su->su_badptr, su->su_badptr + n); | |
| 11483 su->su_badflags = badword_captype(su->su_badptr + n, | |
| 455 | 11484 su->su_badptr + su->su_badlen); |
| 625 | 11485 #ifdef DEBUG_TRIEWALK |
| 11486 sprintf(changename[depth], "prefix"); | |
| 11487 #endif | |
| 11488 go_deeper(stack, depth, 0); | |
| 11489 ++depth; | |
| 11490 sp = &stack[depth]; | |
| 11491 sp->ts_prefixdepth = depth - 1; | |
| 11492 byts = fbyts; | |
| 11493 idxs = fidxs; | |
| 11494 sp->ts_arridx = 0; | |
| 11495 | |
| 11496 /* Move the prefix to preword[] with the right case | |
| 11497 * and make find_keepcap_word() works. */ | |
| 11498 tword[sp->ts_twordlen] = NUL; | |
| 11499 make_case_word(tword + sp->ts_splitoff, | |
| 11500 preword + sp->ts_prewordlen, flags); | |
| 835 | 11501 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 625 | 11502 sp->ts_splitoff = sp->ts_twordlen; |
| 11503 } | |
| 11504 break; | |
| 11505 } | |
| 11506 | |
| 11507 if (sp->ts_curi > len || byts[arridx] != 0) | |
| 11508 { | |
| 11509 /* Past bytes in node and/or past NUL bytes. */ | |
| 11510 sp->ts_state = STATE_ENDNUL; | |
| 11511 sp->ts_save_badflags = su->su_badflags; | |
| 11512 break; | |
| 11513 } | |
| 11514 | |
| 11515 /* | |
| 11516 * End of word in tree. | |
| 11517 */ | |
| 11518 ++sp->ts_curi; /* eat one NUL byte */ | |
| 11519 | |
| 11520 flags = (int)idxs[arridx]; | |
| 698 | 11521 |
| 11522 /* Skip words with the NOSUGGEST flag. */ | |
| 11523 if (flags & WF_NOSUGGEST) | |
| 11524 break; | |
| 11525 | |
| 625 | 11526 fword_ends = (fword[sp->ts_fidx] == NUL |
| 11527 || (soundfold | |
| 11528 ? vim_iswhite(fword[sp->ts_fidx]) | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11529 : !spell_iswordp(fword + sp->ts_fidx, curwin))); |
| 625 | 11530 tword[sp->ts_twordlen] = NUL; |
| 11531 | |
| 11532 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL | |
| 485 | 11533 && (sp->ts_flags & TSF_PREFIXOK) == 0) |
| 625 | 11534 { |
| 11535 /* There was a prefix before the word. Check that the prefix | |
| 11536 * can be used with this word. */ | |
| 11537 /* Count the length of the NULs in the prefix. If there are | |
| 11538 * none this must be the first try without a prefix. */ | |
| 11539 n = stack[sp->ts_prefixdepth].ts_arridx; | |
| 11540 len = pbyts[n++]; | |
| 11541 for (c = 0; c < len && pbyts[n + c] == 0; ++c) | |
| 11542 ; | |
| 11543 if (c > 0) | |
| 11544 { | |
| 11545 c = valid_word_prefix(c, n, flags, | |
| 483 | 11546 tword + sp->ts_splitoff, slang, FALSE); |
| 625 | 11547 if (c == 0) |
| 11548 break; | |
| 11549 | |
| 11550 /* Use the WF_RARE flag for a rare prefix. */ | |
| 11551 if (c & WF_RAREPFX) | |
| 11552 flags |= WF_RARE; | |
| 11553 | |
| 11554 /* Tricky: when checking for both prefix and compounding | |
| 11555 * we run into the prefix flag first. | |
| 11556 * Remember that it's OK, so that we accept the prefix | |
| 11557 * when arriving at a compound flag. */ | |
| 11558 sp->ts_flags |= TSF_PREFIXOK; | |
| 11559 } | |
| 11560 } | |
| 11561 | |
| 11562 /* Check NEEDCOMPOUND: can't use word without compounding. Do try | |
| 11563 * appending another compound word below. */ | |
| 11564 if (sp->ts_complen == sp->ts_compsplit && fword_ends | |
| 500 | 11565 && (flags & WF_NEEDCOMP)) |
| 625 | 11566 goodword_ends = FALSE; |
| 11567 else | |
| 11568 goodword_ends = TRUE; | |
| 11569 | |
| 11570 p = NULL; | |
| 11571 compound_ok = TRUE; | |
| 11572 if (sp->ts_complen > sp->ts_compsplit) | |
| 11573 { | |
| 11574 if (slang->sl_nobreak) | |
| 11575 { | |
| 11576 /* There was a word before this word. When there was no | |
| 11577 * change in this word (it was correct) add the first word | |
| 11578 * as a suggestion. If this word was corrected too, we | |
| 11579 * need to check if a correct word follows. */ | |
| 11580 if (sp->ts_fidx - sp->ts_splitfidx | |
| 11581 == sp->ts_twordlen - sp->ts_splitoff | |
| 11582 && STRNCMP(fword + sp->ts_splitfidx, | |
| 11583 tword + sp->ts_splitoff, | |
| 11584 sp->ts_fidx - sp->ts_splitfidx) == 0) | |
| 492 | 11585 { |
| 625 | 11586 preword[sp->ts_prewordlen] = NUL; |
| 11587 newscore = score_wordcount_adj(slang, sp->ts_score, | |
| 11588 preword + sp->ts_prewordlen, | |
| 11589 sp->ts_prewordlen > 0); | |
| 11590 /* Add the suggestion if the score isn't too bad. */ | |
| 11591 if (newscore <= su->su_maxscore) | |
| 492 | 11592 add_suggestion(su, &su->su_ga, preword, |
| 503 | 11593 sp->ts_splitfidx - repextra, |
| 625 | 11594 newscore, 0, FALSE, |
| 11595 lp->lp_sallang, FALSE); | |
| 11596 break; | |
| 492 | 11597 } |
| 490 | 11598 } |
| 11599 else | |
| 625 | 11600 { |
| 11601 /* There was a compound word before this word. If this | |
| 11602 * word does not support compounding then give up | |
| 11603 * (splitting is tried for the word without compound | |
| 11604 * flag). */ | |
| 11605 if (((unsigned)flags >> 24) == 0 | |
| 11606 || sp->ts_twordlen - sp->ts_splitoff | |
| 11607 < slang->sl_compminlen) | |
| 11608 break; | |
| 11609 #ifdef FEAT_MBYTE | |
| 11610 /* For multi-byte chars check character length against | |
| 11611 * COMPOUNDMIN. */ | |
| 11612 if (has_mbyte | |
| 11613 && slang->sl_compminlen > 0 | |
| 11614 && mb_charlen(tword + sp->ts_splitoff) | |
| 11615 < slang->sl_compminlen) | |
| 11616 break; | |
| 11617 #endif | |
| 11618 | |
| 11619 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 11620 compflags[sp->ts_complen + 1] = NUL; | |
| 11621 vim_strncpy(preword + sp->ts_prewordlen, | |
| 11622 tword + sp->ts_splitoff, | |
| 11623 sp->ts_twordlen - sp->ts_splitoff); | |
| 1762 | 11624 |
| 11625 /* Verify CHECKCOMPOUNDPATTERN rules. */ | |
| 11626 if (match_checkcompoundpattern(preword, sp->ts_prewordlen, | |
| 11627 &slang->sl_comppat)) | |
| 11628 compound_ok = FALSE; | |
| 11629 | |
| 11630 if (compound_ok) | |
| 11631 { | |
| 11632 p = preword; | |
| 11633 while (*skiptowhite(p) != NUL) | |
| 11634 p = skipwhite(skiptowhite(p)); | |
| 11635 if (fword_ends && !can_compound(slang, p, | |
| 625 | 11636 compflags + sp->ts_compsplit)) |
| 1762 | 11637 /* Compound is not allowed. But it may still be |
| 11638 * possible if we add another (short) word. */ | |
| 11639 compound_ok = FALSE; | |
| 11640 } | |
| 625 | 11641 |
| 11642 /* Get pointer to last char of previous word. */ | |
| 11643 p = preword + sp->ts_prewordlen; | |
| 11644 mb_ptr_back(preword, p); | |
| 11645 } | |
| 11646 } | |
| 11647 | |
| 11648 /* | |
| 11649 * Form the word with proper case in preword. | |
| 11650 * If there is a word from a previous split, append. | |
| 11651 * For the soundfold tree don't change the case, simply append. | |
| 11652 */ | |
| 11653 if (soundfold) | |
| 11654 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff); | |
| 11655 else if (flags & WF_KEEPCAP) | |
| 11656 /* Must find the word in the keep-case tree. */ | |
| 11657 find_keepcap_word(slang, tword + sp->ts_splitoff, | |
| 483 | 11658 preword + sp->ts_prewordlen); |
| 625 | 11659 else |
| 11660 { | |
| 11661 /* Include badflags: If the badword is onecap or allcap | |
| 11662 * use that for the goodword too. But if the badword is | |
| 11663 * allcap and it's only one char long use onecap. */ | |
| 11664 c = su->su_badflags; | |
| 11665 if ((c & WF_ALLCAP) | |
| 11666 #ifdef FEAT_MBYTE | |
| 11667 && su->su_badlen == (*mb_ptr2len)(su->su_badptr) | |
| 346 | 11668 #else |
| 625 | 11669 && su->su_badlen == 1 |
| 11670 #endif | |
| 11671 ) | |
| 11672 c = WF_ONECAP; | |
| 11673 c |= flags; | |
| 11674 | |
| 11675 /* When appending a compound word after a word character don't | |
| 11676 * use Onecap. */ | |
| 5477 | 11677 if (p != NULL && spell_iswordp_nmw(p, curwin)) |
| 625 | 11678 c &= ~WF_ONECAP; |
| 11679 make_case_word(tword + sp->ts_splitoff, | |
| 490 | 11680 preword + sp->ts_prewordlen, c); |
| 625 | 11681 } |
| 11682 | |
| 11683 if (!soundfold) | |
| 11684 { | |
| 323 | 11685 /* Don't use a banned word. It may appear again as a good |
| 11686 * word, thus remember it. */ | |
| 11687 if (flags & WF_BANNED) | |
| 11688 { | |
| 483 | 11689 add_banned(su, preword + sp->ts_prewordlen); |
| 323 | 11690 break; |
| 11691 } | |
| 534 | 11692 if ((sp->ts_complen == sp->ts_compsplit |
| 625 | 11693 && WAS_BANNED(su, preword + sp->ts_prewordlen)) |
| 11694 || WAS_BANNED(su, preword)) | |
| 534 | 11695 { |
| 11696 if (slang->sl_compprog == NULL) | |
| 11697 break; | |
| 11698 /* the word so far was banned but we may try compounding */ | |
| 11699 goodword_ends = FALSE; | |
| 11700 } | |
| 625 | 11701 } |
| 11702 | |
| 11703 newscore = 0; | |
| 11704 if (!soundfold) /* soundfold words don't have flags */ | |
| 11705 { | |
| 323 | 11706 if ((flags & WF_REGION) |
| 390 | 11707 && (((unsigned)flags >> 16) & lp->lp_region) == 0) |
| 323 | 11708 newscore += SCORE_REGION; |
| 11709 if (flags & WF_RARE) | |
| 11710 newscore += SCORE_RARE; | |
| 11711 | |
| 346 | 11712 if (!spell_valid_case(su->su_badflags, |
| 483 | 11713 captype(preword + sp->ts_prewordlen, NULL))) |
| 323 | 11714 newscore += SCORE_ICASE; |
| 625 | 11715 } |
| 11716 | |
| 11717 /* TODO: how about splitting in the soundfold tree? */ | |
| 11718 if (fword_ends | |
| 11719 && goodword_ends | |
| 11720 && sp->ts_fidx >= sp->ts_fidxtry | |
| 11721 && compound_ok) | |
| 11722 { | |
| 11723 /* The badword also ends: add suggestions. */ | |
| 11724 #ifdef DEBUG_TRIEWALK | |
| 11725 if (soundfold && STRCMP(preword, "smwrd") == 0) | |
| 11726 { | |
| 11727 int j; | |
| 11728 | |
| 11729 /* print the stack of changes that brought us here */ | |
| 11730 smsg("------ %s -------", fword); | |
| 11731 for (j = 0; j < depth; ++j) | |
| 11732 smsg("%s", changename[j]); | |
| 11733 } | |
| 11734 #endif | |
| 11735 if (soundfold) | |
| 11736 { | |
| 11737 /* For soundfolded words we need to find the original | |
| 1188 | 11738 * words, the edit distance and then add them. */ |
| 625 | 11739 add_sound_suggest(su, preword, sp->ts_score, lp); |
| 11740 } | |
|
2457
6c1d10a85398
Avoid illegal memory access in spell suggestion. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2455
diff
changeset
|
11741 else if (sp->ts_fidx > 0) |
| 625 | 11742 { |
| 11743 /* Give a penalty when changing non-word char to word | |
| 11744 * char, e.g., "thes," -> "these". */ | |
| 366 | 11745 p = fword + sp->ts_fidx; |
| 625 | 11746 mb_ptr_back(fword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11747 if (!spell_iswordp(p, curwin)) |
| 366 | 11748 { |
| 11749 p = preword + STRLEN(preword); | |
| 625 | 11750 mb_ptr_back(preword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
11751 if (spell_iswordp(p, curwin)) |
| 366 | 11752 newscore += SCORE_NONWORD; |
| 11753 } | |
| 11754 | |
| 625 | 11755 /* Give a bonus to words seen before. */ |
| 11756 score = score_wordcount_adj(slang, | |
| 11757 sp->ts_score + newscore, | |
| 11758 preword + sp->ts_prewordlen, | |
| 11759 sp->ts_prewordlen > 0); | |
| 11760 | |
| 11761 /* Add the suggestion if the score isn't too bad. */ | |
| 11762 if (score <= su->su_maxscore) | |
| 633 | 11763 { |
| 625 | 11764 add_suggestion(su, &su->su_ga, preword, |
| 11765 sp->ts_fidx - repextra, | |
| 11766 score, 0, FALSE, lp->lp_sallang, FALSE); | |
| 633 | 11767 |
| 11768 if (su->su_badflags & WF_MIXCAP) | |
| 11769 { | |
| 11770 /* We really don't know if the word should be | |
| 11771 * upper or lower case, add both. */ | |
| 11772 c = captype(preword, NULL); | |
| 11773 if (c == 0 || c == WF_ALLCAP) | |
| 11774 { | |
| 11775 make_case_word(tword + sp->ts_splitoff, | |
| 11776 preword + sp->ts_prewordlen, | |
| 11777 c == 0 ? WF_ALLCAP : 0); | |
| 11778 | |
| 11779 add_suggestion(su, &su->su_ga, preword, | |
| 11780 sp->ts_fidx - repextra, | |
| 11781 score + SCORE_ICASE, 0, FALSE, | |
| 11782 lp->lp_sallang, FALSE); | |
| 11783 } | |
| 11784 } | |
| 11785 } | |
| 625 | 11786 } |
| 11787 } | |
| 11788 | |
| 11789 /* | |
| 11790 * Try word split and/or compounding. | |
| 11791 */ | |
| 11792 if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends) | |
| 11793 #ifdef FEAT_MBYTE | |
| 11794 /* Don't split halfway a character. */ | |
| 11795 && (!has_mbyte || sp->ts_tcharlen == 0) | |
| 11796 #endif | |
| 11797 ) | |
| 11798 { | |
| 11799 int try_compound; | |
| 11800 int try_split; | |
| 11801 | |
| 11802 /* If past the end of the bad word don't try a split. | |
| 11803 * Otherwise try changing the next word. E.g., find | |
| 11804 * suggestions for "the the" where the second "the" is | |
| 11805 * different. It's done like a split. | |
| 11806 * TODO: word split for soundfold words */ | |
| 11807 try_split = (sp->ts_fidx - repextra < su->su_badlen) | |
| 11808 && !soundfold; | |
| 11809 | |
| 11810 /* Get here in several situations: | |
| 11811 * 1. The word in the tree ends: | |
| 11812 * If the word allows compounding try that. Otherwise try | |
| 11813 * a split by inserting a space. For both check that a | |
| 11814 * valid words starts at fword[sp->ts_fidx]. | |
| 11815 * For NOBREAK do like compounding to be able to check if | |
| 11816 * the next word is valid. | |
| 11817 * 2. The badword does end, but it was due to a change (e.g., | |
| 11818 * a swap). No need to split, but do check that the | |
| 11819 * following word is valid. | |
| 11820 * 3. The badword and the word in the tree end. It may still | |
| 11821 * be possible to compound another (short) word. | |
| 11822 */ | |
| 11823 try_compound = FALSE; | |
| 11824 if (!soundfold | |
| 11825 && slang->sl_compprog != NULL | |
| 11826 && ((unsigned)flags >> 24) != 0 | |
| 11827 && sp->ts_twordlen - sp->ts_splitoff | |
| 11828 >= slang->sl_compminlen | |
| 11829 #ifdef FEAT_MBYTE | |
| 11830 && (!has_mbyte | |
| 11831 || slang->sl_compminlen == 0 | |
| 11832 || mb_charlen(tword + sp->ts_splitoff) | |
| 11833 >= slang->sl_compminlen) | |
| 11834 #endif | |
| 11835 && (slang->sl_compsylmax < MAXWLEN | |
| 11836 || sp->ts_complen + 1 - sp->ts_compsplit | |
| 11837 < slang->sl_compmax) | |
| 1762 | 11838 && (can_be_compound(sp, slang, |
| 11839 compflags, ((unsigned)flags >> 24)))) | |
| 11840 | |
| 625 | 11841 { |
| 11842 try_compound = TRUE; | |
| 11843 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 11844 compflags[sp->ts_complen + 1] = NUL; | |
| 11845 } | |
| 11846 | |
| 11847 /* For NOBREAK we never try splitting, it won't make any word | |
| 11848 * valid. */ | |
| 11849 if (slang->sl_nobreak) | |
| 11850 try_compound = TRUE; | |
| 11851 | |
| 11852 /* If we could add a compound word, and it's also possible to | |
| 11853 * split at this point, do the split first and set | |
| 11854 * TSF_DIDSPLIT to avoid doing it again. */ | |
| 11855 else if (!fword_ends | |
| 11856 && try_compound | |
| 11857 && (sp->ts_flags & TSF_DIDSPLIT) == 0) | |
| 11858 { | |
| 485 | 11859 try_compound = FALSE; |
| 625 | 11860 sp->ts_flags |= TSF_DIDSPLIT; |
| 11861 --sp->ts_curi; /* do the same NUL again */ | |
| 11862 compflags[sp->ts_complen] = NUL; | |
| 11863 } | |
| 11864 else | |
| 11865 sp->ts_flags &= ~TSF_DIDSPLIT; | |
| 11866 | |
| 11867 if (try_split || try_compound) | |
| 11868 { | |
| 534 | 11869 if (!try_compound && (!fword_ends || !goodword_ends)) |
| 485 | 11870 { |
| 11871 /* If we're going to split need to check that the | |
| 501 | 11872 * words so far are valid for compounding. If there |
| 11873 * is only one word it must not have the NEEDCOMPOUND | |
| 11874 * flag. */ | |
| 11875 if (sp->ts_complen == sp->ts_compsplit | |
| 11876 && (flags & WF_NEEDCOMP)) | |
| 11877 break; | |
| 490 | 11878 p = preword; |
| 11879 while (*skiptowhite(p) != NUL) | |
| 11880 p = skipwhite(skiptowhite(p)); | |
| 485 | 11881 if (sp->ts_complen > sp->ts_compsplit |
| 490 | 11882 && !can_compound(slang, p, |
| 485 | 11883 compflags + sp->ts_compsplit)) |
| 11884 break; | |
| 698 | 11885 |
| 11886 if (slang->sl_nosplitsugs) | |
| 11887 newscore += SCORE_SPLIT_NO; | |
| 11888 else | |
| 11889 newscore += SCORE_SPLIT; | |
| 625 | 11890 |
| 11891 /* Give a bonus to words seen before. */ | |
| 11892 newscore = score_wordcount_adj(slang, newscore, | |
| 11893 preword + sp->ts_prewordlen, TRUE); | |
| 482 | 11894 } |
| 11895 | |
| 625 | 11896 if (TRY_DEEPER(su, stack, depth, newscore)) |
| 323 | 11897 { |
| 625 | 11898 go_deeper(stack, depth, newscore); |
| 11899 #ifdef DEBUG_TRIEWALK | |
| 11900 if (!try_compound && !fword_ends) | |
| 11901 sprintf(changename[depth], "%.*s-%s: split", | |
| 11902 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 11903 else | |
| 11904 sprintf(changename[depth], "%.*s-%s: compound", | |
| 11905 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 11906 #endif | |
| 323 | 11907 /* Save things to be restored at STATE_SPLITUNDO. */ |
| 346 | 11908 sp->ts_save_badflags = su->su_badflags; |
| 376 | 11909 sp->ts_state = STATE_SPLITUNDO; |
| 11910 | |
| 11911 ++depth; | |
| 11912 sp = &stack[depth]; | |
| 323 | 11913 |
| 482 | 11914 /* Append a space to preword when splitting. */ |
| 11915 if (!try_compound && !fword_ends) | |
| 11916 STRCAT(preword, " "); | |
| 835 | 11917 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 483 | 11918 sp->ts_splitoff = sp->ts_twordlen; |
| 492 | 11919 sp->ts_splitfidx = sp->ts_fidx; |
| 376 | 11920 |
| 11921 /* If the badword has a non-word character at this | |
| 11922 * position skip it. That means replacing the | |
| 482 | 11923 * non-word character with a space. Always skip a |
| 534 | 11924 * character when the word ends. But only when the |
| 11925 * good word can end. */ | |
| 625 | 11926 if (((!try_compound && !spell_iswordp_nmw(fword |
| 5477 | 11927 + sp->ts_fidx, |
| 11928 curwin)) | |
| 625 | 11929 || fword_ends) |
| 11930 && fword[sp->ts_fidx] != NUL | |
| 11931 && goodword_ends) | |
| 376 | 11932 { |
| 482 | 11933 int l; |
| 11934 | |
| 376 | 11935 #ifdef FEAT_MBYTE |
| 11936 if (has_mbyte) | |
| 482 | 11937 l = MB_BYTE2LEN(fword[sp->ts_fidx]); |
| 376 | 11938 else |
| 11939 #endif | |
| 482 | 11940 l = 1; |
| 11941 if (fword_ends) | |
| 11942 { | |
| 11943 /* Copy the skipped character to preword. */ | |
| 483 | 11944 mch_memmove(preword + sp->ts_prewordlen, |
| 482 | 11945 fword + sp->ts_fidx, l); |
| 483 | 11946 sp->ts_prewordlen += l; |
| 11947 preword[sp->ts_prewordlen] = NUL; | |
| 482 | 11948 } |
| 11949 else | |
| 11950 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST; | |
| 11951 sp->ts_fidx += l; | |
| 376 | 11952 } |
| 455 | 11953 |
| 485 | 11954 /* When compounding include compound flag in |
| 11955 * compflags[] (already set above). When splitting we | |
| 11956 * may start compounding over again. */ | |
| 482 | 11957 if (try_compound) |
| 483 | 11958 ++sp->ts_complen; |
| 482 | 11959 else |
| 485 | 11960 sp->ts_compsplit = sp->ts_complen; |
| 11961 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 482 | 11962 |
| 455 | 11963 /* set su->su_badflags to the caps type at this |
| 11964 * position */ | |
| 324 | 11965 #ifdef FEAT_MBYTE |
| 11966 if (has_mbyte) | |
| 455 | 11967 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); |
| 324 | 11968 else |
| 11969 #endif | |
| 455 | 11970 n = sp->ts_fidx; |
| 474 | 11971 su->su_badflags = badword_captype(su->su_badptr + n, |
| 455 | 11972 su->su_badptr + su->su_badlen); |
| 323 | 11973 |
| 11974 /* Restart at top of the tree. */ | |
| 376 | 11975 sp->ts_arridx = 0; |
| 485 | 11976 |
| 11977 /* If there are postponed prefixes, try these too. */ | |
| 11978 if (pbyts != NULL) | |
| 11979 { | |
| 11980 byts = pbyts; | |
| 11981 idxs = pidxs; | |
| 11982 sp->ts_prefixdepth = PFD_PREFIXTREE; | |
| 11983 sp->ts_state = STATE_NOPREFIX; | |
| 11984 } | |
| 323 | 11985 } |
| 11986 } | |
| 625 | 11987 } |
| 11988 break; | |
| 11989 | |
| 11990 case STATE_SPLITUNDO: | |
| 11991 /* Undo the changes done for word split or compound word. */ | |
| 11992 su->su_badflags = sp->ts_save_badflags; | |
| 11993 | |
| 11994 /* Continue looking for NUL bytes. */ | |
| 11995 sp->ts_state = STATE_START; | |
| 11996 | |
| 11997 /* In case we went into the prefix tree. */ | |
| 11998 byts = fbyts; | |
| 11999 idxs = fidxs; | |
| 12000 break; | |
| 12001 | |
| 12002 case STATE_ENDNUL: | |
| 12003 /* Past the NUL bytes in the node. */ | |
| 12004 su->su_badflags = sp->ts_save_badflags; | |
| 12005 if (fword[sp->ts_fidx] == NUL | |
| 12006 #ifdef FEAT_MBYTE | |
| 12007 && sp->ts_tcharlen == 0 | |
| 12008 #endif | |
| 12009 ) | |
| 12010 { | |
| 12011 /* The badword ends, can't use STATE_PLAIN. */ | |
| 12012 sp->ts_state = STATE_DEL; | |
| 323 | 12013 break; |
| 625 | 12014 } |
| 12015 sp->ts_state = STATE_PLAIN; | |
| 12016 /*FALLTHROUGH*/ | |
| 12017 | |
| 12018 case STATE_PLAIN: | |
| 12019 /* | |
| 12020 * Go over all possible bytes at this node, add each to tword[] | |
| 12021 * and use child node. "ts_curi" is the index. | |
| 12022 */ | |
| 12023 arridx = sp->ts_arridx; | |
| 12024 if (sp->ts_curi > byts[arridx]) | |
| 12025 { | |
| 12026 /* Done all bytes at this node, do next state. When still at | |
| 12027 * already changed bytes skip the other tricks. */ | |
| 12028 if (sp->ts_fidx >= sp->ts_fidxtry) | |
| 323 | 12029 sp->ts_state = STATE_DEL; |
| 12030 else | |
| 625 | 12031 sp->ts_state = STATE_FINAL; |
| 12032 } | |
| 12033 else | |
| 12034 { | |
| 12035 arridx += sp->ts_curi++; | |
| 12036 c = byts[arridx]; | |
| 12037 | |
| 12038 /* Normal byte, go one level deeper. If it's not equal to the | |
| 12039 * byte in the bad word adjust the score. But don't even try | |
| 12040 * when the byte was already changed. And don't try when we | |
| 5695 | 12041 * just deleted this byte, accepting it is always cheaper than |
| 625 | 12042 * delete + substitute. */ |
| 12043 if (c == fword[sp->ts_fidx] | |
| 12044 #ifdef FEAT_MBYTE | |
| 12045 || (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE) | |
| 12046 #endif | |
| 12047 ) | |
| 12048 newscore = 0; | |
| 12049 else | |
| 12050 newscore = SCORE_SUBST; | |
| 12051 if ((newscore == 0 | |
| 12052 || (sp->ts_fidx >= sp->ts_fidxtry | |
| 12053 && ((sp->ts_flags & TSF_DIDDEL) == 0 | |
| 12054 || c != fword[sp->ts_delidx]))) | |
| 12055 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 12056 { | |
| 12057 go_deeper(stack, depth, newscore); | |
| 12058 #ifdef DEBUG_TRIEWALK | |
| 12059 if (newscore > 0) | |
| 12060 sprintf(changename[depth], "%.*s-%s: subst %c to %c", | |
| 12061 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12062 fword[sp->ts_fidx], c); | |
| 323 | 12063 else |
| 625 | 12064 sprintf(changename[depth], "%.*s-%s: accept %c", |
| 12065 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12066 fword[sp->ts_fidx]); | |
| 12067 #endif | |
| 323 | 12068 ++depth; |
| 625 | 12069 sp = &stack[depth]; |
| 12070 ++sp->ts_fidx; | |
| 12071 tword[sp->ts_twordlen++] = c; | |
| 12072 sp->ts_arridx = idxs[arridx]; | |
| 12073 #ifdef FEAT_MBYTE | |
| 12074 if (newscore == SCORE_SUBST) | |
| 12075 sp->ts_isdiff = DIFF_YES; | |
| 330 | 12076 if (has_mbyte) |
| 358 | 12077 { |
| 625 | 12078 /* Multi-byte characters are a bit complicated to |
| 12079 * handle: They differ when any of the bytes differ | |
| 12080 * and then their length may also differ. */ | |
| 12081 if (sp->ts_tcharlen == 0) | |
| 12082 { | |
| 12083 /* First byte. */ | |
| 12084 sp->ts_tcharidx = 0; | |
| 12085 sp->ts_tcharlen = MB_BYTE2LEN(c); | |
| 12086 sp->ts_fcharstart = sp->ts_fidx - 1; | |
| 12087 sp->ts_isdiff = (newscore != 0) | |
| 12088 ? DIFF_YES : DIFF_NONE; | |
| 12089 } | |
| 12090 else if (sp->ts_isdiff == DIFF_INSERT) | |
| 12091 /* When inserting trail bytes don't advance in the | |
| 12092 * bad word. */ | |
| 12093 --sp->ts_fidx; | |
| 12094 if (++sp->ts_tcharidx == sp->ts_tcharlen) | |
| 12095 { | |
| 12096 /* Last byte of character. */ | |
| 12097 if (sp->ts_isdiff == DIFF_YES) | |
| 12098 { | |
| 12099 /* Correct ts_fidx for the byte length of the | |
| 12100 * character (we didn't check that before). */ | |
| 12101 sp->ts_fidx = sp->ts_fcharstart | |
| 12102 + MB_BYTE2LEN( | |
| 12103 fword[sp->ts_fcharstart]); | |
| 12104 | |
| 12105 /* For changing a composing character adjust | |
| 12106 * the score from SCORE_SUBST to | |
| 12107 * SCORE_SUBCOMP. */ | |
| 12108 if (enc_utf8 | |
| 12109 && utf_iscomposing( | |
| 12110 mb_ptr2char(tword | |
| 12111 + sp->ts_twordlen | |
| 12112 - sp->ts_tcharlen)) | |
| 12113 && utf_iscomposing( | |
| 12114 mb_ptr2char(fword | |
| 12115 + sp->ts_fcharstart))) | |
| 12116 sp->ts_score -= | |
| 12117 SCORE_SUBST - SCORE_SUBCOMP; | |
| 12118 | |
| 12119 /* For a similar character adjust score from | |
| 12120 * SCORE_SUBST to SCORE_SIMILAR. */ | |
| 12121 else if (!soundfold | |
| 12122 && slang->sl_has_map | |
| 12123 && similar_chars(slang, | |
| 12124 mb_ptr2char(tword | |
| 12125 + sp->ts_twordlen | |
| 12126 - sp->ts_tcharlen), | |
| 12127 mb_ptr2char(fword | |
| 12128 + sp->ts_fcharstart))) | |
| 12129 sp->ts_score -= | |
| 12130 SCORE_SUBST - SCORE_SIMILAR; | |
| 12131 } | |
| 12132 else if (sp->ts_isdiff == DIFF_INSERT | |
| 12133 && sp->ts_twordlen > sp->ts_tcharlen) | |
| 12134 { | |
| 12135 p = tword + sp->ts_twordlen - sp->ts_tcharlen; | |
| 12136 c = mb_ptr2char(p); | |
| 12137 if (enc_utf8 && utf_iscomposing(c)) | |
| 12138 { | |
| 12139 /* Inserting a composing char doesn't | |
| 12140 * count that much. */ | |
| 12141 sp->ts_score -= SCORE_INS - SCORE_INSCOMP; | |
| 12142 } | |
| 12143 else | |
| 12144 { | |
| 12145 /* If the previous character was the same, | |
| 12146 * thus doubling a character, give a bonus | |
| 12147 * to the score. Also for the soundfold | |
| 12148 * tree (might seem illogical but does | |
| 12149 * give better scores). */ | |
| 12150 mb_ptr_back(tword, p); | |
| 12151 if (c == mb_ptr2char(p)) | |
| 12152 sp->ts_score -= SCORE_INS | |
| 12153 - SCORE_INSDUP; | |
| 12154 } | |
| 12155 } | |
| 12156 | |
| 12157 /* Starting a new char, reset the length. */ | |
| 12158 sp->ts_tcharlen = 0; | |
| 12159 } | |
| 358 | 12160 } |
| 330 | 12161 else |
| 12162 #endif | |
| 358 | 12163 { |
| 625 | 12164 /* If we found a similar char adjust the score. |
| 12165 * We do this after calling go_deeper() because | |
| 12166 * it's slow. */ | |
| 12167 if (newscore != 0 | |
| 12168 && !soundfold | |
| 12169 && slang->sl_has_map | |
| 12170 && similar_chars(slang, | |
| 12171 c, fword[sp->ts_fidx - 1])) | |
| 12172 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR; | |
| 358 | 12173 } |
| 625 | 12174 } |
| 12175 } | |
| 12176 break; | |
| 12177 | |
| 12178 case STATE_DEL: | |
| 12179 #ifdef FEAT_MBYTE | |
| 12180 /* When past the first byte of a multi-byte char don't try | |
| 12181 * delete/insert/swap a character. */ | |
| 12182 if (has_mbyte && sp->ts_tcharlen > 0) | |
| 12183 { | |
| 12184 sp->ts_state = STATE_FINAL; | |
| 12185 break; | |
| 12186 } | |
| 12187 #endif | |
| 12188 /* | |
| 12189 * Try skipping one character in the bad word (delete it). | |
| 12190 */ | |
| 12191 sp->ts_state = STATE_INS_PREP; | |
| 12192 sp->ts_curi = 1; | |
| 12193 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*') | |
| 12194 /* Deleting a vowel at the start of a word counts less, see | |
| 12195 * soundalike_score(). */ | |
| 12196 newscore = 2 * SCORE_DEL / 3; | |
| 12197 else | |
| 12198 newscore = SCORE_DEL; | |
| 12199 if (fword[sp->ts_fidx] != NUL | |
| 12200 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 12201 { | |
| 12202 go_deeper(stack, depth, newscore); | |
| 12203 #ifdef DEBUG_TRIEWALK | |
| 12204 sprintf(changename[depth], "%.*s-%s: delete %c", | |
| 12205 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12206 fword[sp->ts_fidx]); | |
| 12207 #endif | |
| 12208 ++depth; | |
| 12209 | |
| 12210 /* Remember what character we deleted, so that we can avoid | |
| 12211 * inserting it again. */ | |
| 12212 stack[depth].ts_flags |= TSF_DIDDEL; | |
| 12213 stack[depth].ts_delidx = sp->ts_fidx; | |
| 12214 | |
| 12215 /* Advance over the character in fword[]. Give a bonus to the | |
| 12216 * score if the same character is following "nn" -> "n". It's | |
| 12217 * a bit illogical for soundfold tree but it does give better | |
| 12218 * results. */ | |
| 12219 #ifdef FEAT_MBYTE | |
| 12220 if (has_mbyte) | |
| 12221 { | |
| 12222 c = mb_ptr2char(fword + sp->ts_fidx); | |
| 12223 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]); | |
| 12224 if (enc_utf8 && utf_iscomposing(c)) | |
| 12225 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP; | |
| 12226 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx)) | |
| 12227 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 12228 } |
| 12229 else | |
| 625 | 12230 #endif |
| 12231 { | |
| 12232 ++stack[depth].ts_fidx; | |
| 12233 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1]) | |
| 12234 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 12235 } |
| 12236 break; | |
| 625 | 12237 } |
| 12238 /*FALLTHROUGH*/ | |
| 12239 | |
| 12240 case STATE_INS_PREP: | |
| 12241 if (sp->ts_flags & TSF_DIDDEL) | |
| 12242 { | |
| 12243 /* If we just deleted a byte then inserting won't make sense, | |
| 12244 * a substitute is always cheaper. */ | |
| 12245 sp->ts_state = STATE_SWAP; | |
| 12246 break; | |
| 12247 } | |
| 12248 | |
| 12249 /* skip over NUL bytes */ | |
| 12250 n = sp->ts_arridx; | |
| 12251 for (;;) | |
| 12252 { | |
| 12253 if (sp->ts_curi > byts[n]) | |
| 12254 { | |
| 12255 /* Only NUL bytes at this node, go to next state. */ | |
| 12256 sp->ts_state = STATE_SWAP; | |
| 12257 break; | |
| 12258 } | |
| 12259 if (byts[n + sp->ts_curi] != NUL) | |
| 12260 { | |
| 12261 /* Found a byte to insert. */ | |
| 12262 sp->ts_state = STATE_INS; | |
| 330 | 12263 break; |
| 12264 } | |
| 625 | 12265 ++sp->ts_curi; |
| 12266 } | |
| 12267 break; | |
| 12268 | |
| 12269 /*FALLTHROUGH*/ | |
| 12270 | |
| 12271 case STATE_INS: | |
| 12272 /* Insert one byte. Repeat this for each possible byte at this | |
| 12273 * node. */ | |
| 12274 n = sp->ts_arridx; | |
| 12275 if (sp->ts_curi > byts[n]) | |
| 12276 { | |
| 12277 /* Done all bytes at this node, go to next state. */ | |
| 12278 sp->ts_state = STATE_SWAP; | |
| 12279 break; | |
| 12280 } | |
| 12281 | |
| 12282 /* Do one more byte at this node, but: | |
| 12283 * - Skip NUL bytes. | |
| 12284 * - Skip the byte if it's equal to the byte in the word, | |
| 12285 * accepting that byte is always better. | |
| 12286 */ | |
| 12287 n += sp->ts_curi++; | |
| 12288 c = byts[n]; | |
| 12289 if (soundfold && sp->ts_twordlen == 0 && c == '*') | |
| 12290 /* Inserting a vowel at the start of a word counts less, | |
| 12291 * see soundalike_score(). */ | |
| 12292 newscore = 2 * SCORE_INS / 3; | |
| 12293 else | |
| 12294 newscore = SCORE_INS; | |
| 12295 if (c != fword[sp->ts_fidx] | |
| 12296 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 12297 { | |
| 12298 go_deeper(stack, depth, newscore); | |
| 12299 #ifdef DEBUG_TRIEWALK | |
| 12300 sprintf(changename[depth], "%.*s-%s: insert %c", | |
| 12301 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12302 c); | |
| 12303 #endif | |
| 12304 ++depth; | |
| 12305 sp = &stack[depth]; | |
| 12306 tword[sp->ts_twordlen++] = c; | |
| 12307 sp->ts_arridx = idxs[n]; | |
| 330 | 12308 #ifdef FEAT_MBYTE |
| 12309 if (has_mbyte) | |
| 12310 { | |
| 625 | 12311 fl = MB_BYTE2LEN(c); |
| 12312 if (fl > 1) | |
| 330 | 12313 { |
| 625 | 12314 /* There are following bytes for the same character. |
| 12315 * We must find all bytes before trying | |
| 12316 * delete/insert/swap/etc. */ | |
| 12317 sp->ts_tcharlen = fl; | |
| 12318 sp->ts_tcharidx = 1; | |
| 12319 sp->ts_isdiff = DIFF_INSERT; | |
| 330 | 12320 } |
| 323 | 12321 } |
| 12322 else | |
| 625 | 12323 fl = 1; |
| 12324 if (fl == 1) | |
| 12325 #endif | |
| 12326 { | |
| 12327 /* If the previous character was the same, thus doubling a | |
| 12328 * character, give a bonus to the score. Also for | |
| 12329 * soundfold words (illogical but does give a better | |
| 12330 * score). */ | |
| 12331 if (sp->ts_twordlen >= 2 | |
| 12332 && tword[sp->ts_twordlen - 2] == c) | |
| 12333 sp->ts_score -= SCORE_INS - SCORE_INSDUP; | |
| 12334 } | |
| 12335 } | |
| 12336 break; | |
| 12337 | |
| 12338 case STATE_SWAP: | |
| 12339 /* | |
| 12340 * Swap two bytes in the bad word: "12" -> "21". | |
| 12341 * We change "fword" here, it's changed back afterwards at | |
| 12342 * STATE_UNSWAP. | |
| 12343 */ | |
| 12344 p = fword + sp->ts_fidx; | |
| 12345 c = *p; | |
| 12346 if (c == NUL) | |
| 12347 { | |
| 12348 /* End of word, can't swap or replace. */ | |
| 12349 sp->ts_state = STATE_FINAL; | |
| 12350 break; | |
| 12351 } | |
| 12352 | |
| 12353 /* Don't swap if the first character is not a word character. | |
| 12354 * 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
|
12355 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 12356 { |
| 12357 sp->ts_state = STATE_REP_INI; | |
| 323 | 12358 break; |
| 625 | 12359 } |
| 12360 | |
| 12361 #ifdef FEAT_MBYTE | |
| 12362 if (has_mbyte) | |
| 12363 { | |
| 12364 n = mb_cptr2len(p); | |
| 12365 c = mb_ptr2char(p); | |
| 1337 | 12366 if (p[n] == NUL) |
| 12367 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12368 else if (!soundfold && !spell_iswordp(p + n, curwin)) |
| 625 | 12369 c2 = c; /* don't swap non-word char */ |
| 12370 else | |
| 12371 c2 = mb_ptr2char(p + n); | |
| 12372 } | |
| 12373 else | |
| 12374 #endif | |
| 12375 { | |
| 1337 | 12376 if (p[1] == NUL) |
| 12377 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12378 else if (!soundfold && !spell_iswordp(p + 1, curwin)) |
| 625 | 12379 c2 = c; /* don't swap non-word char */ |
| 12380 else | |
| 12381 c2 = p[1]; | |
| 12382 } | |
| 12383 | |
| 1337 | 12384 /* When the second character is NUL we can't swap. */ |
| 12385 if (c2 == NUL) | |
| 12386 { | |
| 12387 sp->ts_state = STATE_REP_INI; | |
| 12388 break; | |
| 12389 } | |
| 12390 | |
| 625 | 12391 /* When characters are identical, swap won't do anything. |
| 12392 * Also get here if the second char is not a word character. */ | |
| 12393 if (c == c2) | |
| 12394 { | |
| 12395 sp->ts_state = STATE_SWAP3; | |
| 12396 break; | |
| 12397 } | |
| 12398 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP)) | |
| 12399 { | |
| 12400 go_deeper(stack, depth, SCORE_SWAP); | |
| 12401 #ifdef DEBUG_TRIEWALK | |
| 12402 sprintf(changename[depth], "%.*s-%s: swap %c and %c", | |
| 12403 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12404 c, c2); | |
| 12405 #endif | |
| 12406 sp->ts_state = STATE_UNSWAP; | |
| 12407 ++depth; | |
| 330 | 12408 #ifdef FEAT_MBYTE |
| 12409 if (has_mbyte) | |
| 12410 { | |
| 625 | 12411 fl = mb_char2len(c2); |
| 12412 mch_memmove(p, p + n, fl); | |
| 12413 mb_char2bytes(c, p + fl); | |
| 12414 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 12415 } |
| 12416 else | |
| 12417 #endif | |
| 12418 { | |
| 625 | 12419 p[0] = c2; |
| 330 | 12420 p[1] = c; |
| 625 | 12421 stack[depth].ts_fidxtry = sp->ts_fidx + 2; |
| 12422 } | |
| 12423 } | |
| 12424 else | |
| 12425 /* If this swap doesn't work then SWAP3 won't either. */ | |
| 12426 sp->ts_state = STATE_REP_INI; | |
| 12427 break; | |
| 12428 | |
| 12429 case STATE_UNSWAP: | |
| 12430 /* Undo the STATE_SWAP swap: "21" -> "12". */ | |
| 12431 p = fword + sp->ts_fidx; | |
| 12432 #ifdef FEAT_MBYTE | |
| 12433 if (has_mbyte) | |
| 12434 { | |
| 12435 n = MB_BYTE2LEN(*p); | |
| 12436 c = mb_ptr2char(p + n); | |
| 12437 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n); | |
| 12438 mb_char2bytes(c, p); | |
| 12439 } | |
| 12440 else | |
| 12441 #endif | |
| 12442 { | |
| 12443 c = *p; | |
| 12444 *p = p[1]; | |
| 12445 p[1] = c; | |
| 12446 } | |
| 12447 /*FALLTHROUGH*/ | |
| 12448 | |
| 12449 case STATE_SWAP3: | |
| 12450 /* Swap two bytes, skipping one: "123" -> "321". We change | |
| 12451 * "fword" here, it's changed back afterwards at STATE_UNSWAP3. */ | |
| 12452 p = fword + sp->ts_fidx; | |
| 12453 #ifdef FEAT_MBYTE | |
| 12454 if (has_mbyte) | |
| 12455 { | |
| 12456 n = mb_cptr2len(p); | |
| 12457 c = mb_ptr2char(p); | |
| 12458 fl = mb_cptr2len(p + n); | |
| 12459 c2 = mb_ptr2char(p + n); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12460 if (!soundfold && !spell_iswordp(p + n + fl, curwin)) |
| 625 | 12461 c3 = c; /* don't swap non-word char */ |
| 12462 else | |
| 12463 c3 = mb_ptr2char(p + n + fl); | |
| 12464 } | |
| 12465 else | |
| 12466 #endif | |
| 12467 { | |
| 12468 c = *p; | |
| 12469 c2 = p[1]; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12470 if (!soundfold && !spell_iswordp(p + 2, curwin)) |
| 625 | 12471 c3 = c; /* don't swap non-word char */ |
| 12472 else | |
| 12473 c3 = p[2]; | |
| 12474 } | |
| 12475 | |
| 12476 /* When characters are identical: "121" then SWAP3 result is | |
| 12477 * identical, ROT3L result is same as SWAP: "211", ROT3L result is | |
| 12478 * same as SWAP on next char: "112". Thus skip all swapping. | |
| 12479 * Also skip when c3 is NUL. | |
| 12480 * Also get here when the third character is not a word character. | |
| 12481 * Second character may any char: "a.b" -> "b.a" */ | |
| 12482 if (c == c3 || c3 == NUL) | |
| 12483 { | |
| 12484 sp->ts_state = STATE_REP_INI; | |
| 12485 break; | |
| 12486 } | |
| 12487 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 12488 { | |
| 12489 go_deeper(stack, depth, SCORE_SWAP3); | |
| 12490 #ifdef DEBUG_TRIEWALK | |
| 12491 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c", | |
| 12492 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12493 c, c3); | |
| 12494 #endif | |
| 12495 sp->ts_state = STATE_UNSWAP3; | |
| 12496 ++depth; | |
| 12497 #ifdef FEAT_MBYTE | |
| 12498 if (has_mbyte) | |
| 12499 { | |
| 12500 tl = mb_char2len(c3); | |
| 12501 mch_memmove(p, p + n + fl, tl); | |
| 12502 mb_char2bytes(c2, p + tl); | |
| 12503 mb_char2bytes(c, p + fl + tl); | |
| 12504 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl; | |
| 12505 } | |
| 12506 else | |
| 12507 #endif | |
| 12508 { | |
| 12509 p[0] = p[2]; | |
| 12510 p[2] = c; | |
| 12511 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 12512 } | |
| 12513 } | |
| 12514 else | |
| 12515 sp->ts_state = STATE_REP_INI; | |
| 12516 break; | |
| 12517 | |
| 12518 case STATE_UNSWAP3: | |
| 12519 /* Undo STATE_SWAP3: "321" -> "123" */ | |
| 12520 p = fword + sp->ts_fidx; | |
| 12521 #ifdef FEAT_MBYTE | |
| 12522 if (has_mbyte) | |
| 12523 { | |
| 12524 n = MB_BYTE2LEN(*p); | |
| 12525 c2 = mb_ptr2char(p + n); | |
| 12526 fl = MB_BYTE2LEN(p[n]); | |
| 12527 c = mb_ptr2char(p + n + fl); | |
| 12528 tl = MB_BYTE2LEN(p[n + fl]); | |
| 12529 mch_memmove(p + fl + tl, p, n); | |
| 12530 mb_char2bytes(c, p); | |
| 12531 mb_char2bytes(c2, p + tl); | |
| 12532 p = p + tl; | |
| 12533 } | |
| 12534 else | |
| 12535 #endif | |
| 12536 { | |
| 12537 c = *p; | |
| 12538 *p = p[2]; | |
| 12539 p[2] = c; | |
| 12540 ++p; | |
| 12541 } | |
| 12542 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
12543 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 12544 { |
| 12545 /* Middle char is not a word char, skip the rotate. First and | |
| 12546 * third char were already checked at swap and swap3. */ | |
| 12547 sp->ts_state = STATE_REP_INI; | |
| 12548 break; | |
| 12549 } | |
| 12550 | |
| 12551 /* Rotate three characters left: "123" -> "231". We change | |
| 12552 * "fword" here, it's changed back afterwards at STATE_UNROT3L. */ | |
| 12553 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 12554 { | |
| 12555 go_deeper(stack, depth, SCORE_SWAP3); | |
| 12556 #ifdef DEBUG_TRIEWALK | |
| 12557 p = fword + sp->ts_fidx; | |
| 12558 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c", | |
| 12559 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12560 p[0], p[1], p[2]); | |
| 12561 #endif | |
| 12562 sp->ts_state = STATE_UNROT3L; | |
| 12563 ++depth; | |
| 330 | 12564 p = fword + sp->ts_fidx; |
| 12565 #ifdef FEAT_MBYTE | |
| 12566 if (has_mbyte) | |
| 12567 { | |
| 474 | 12568 n = mb_cptr2len(p); |
| 330 | 12569 c = mb_ptr2char(p); |
| 474 | 12570 fl = mb_cptr2len(p + n); |
| 625 | 12571 fl += mb_cptr2len(p + n + fl); |
| 12572 mch_memmove(p, p + n, fl); | |
| 12573 mb_char2bytes(c, p + fl); | |
| 12574 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 12575 } |
| 12576 else | |
| 12577 #endif | |
| 323 | 12578 { |
| 330 | 12579 c = *p; |
| 625 | 12580 *p = p[1]; |
| 12581 p[1] = p[2]; | |
| 12582 p[2] = c; | |
| 12583 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 12584 } | |
| 12585 } | |
| 12586 else | |
| 12587 sp->ts_state = STATE_REP_INI; | |
| 12588 break; | |
| 12589 | |
| 12590 case STATE_UNROT3L: | |
| 12591 /* Undo ROT3L: "231" -> "123" */ | |
| 12592 p = fword + sp->ts_fidx; | |
| 12593 #ifdef FEAT_MBYTE | |
| 12594 if (has_mbyte) | |
| 12595 { | |
| 12596 n = MB_BYTE2LEN(*p); | |
| 12597 n += MB_BYTE2LEN(p[n]); | |
| 12598 c = mb_ptr2char(p + n); | |
| 12599 tl = MB_BYTE2LEN(p[n]); | |
| 12600 mch_memmove(p + tl, p, n); | |
| 12601 mb_char2bytes(c, p); | |
| 12602 } | |
| 12603 else | |
| 12604 #endif | |
| 12605 { | |
| 12606 c = p[2]; | |
| 12607 p[2] = p[1]; | |
| 12608 p[1] = *p; | |
| 12609 *p = c; | |
| 12610 } | |
| 12611 | |
| 12612 /* Rotate three bytes right: "123" -> "312". We change "fword" | |
| 12613 * here, it's changed back afterwards at STATE_UNROT3R. */ | |
| 12614 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 12615 { | |
| 12616 go_deeper(stack, depth, SCORE_SWAP3); | |
| 12617 #ifdef DEBUG_TRIEWALK | |
| 12618 p = fword + sp->ts_fidx; | |
| 12619 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c", | |
| 12620 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12621 p[0], p[1], p[2]); | |
| 12622 #endif | |
| 12623 sp->ts_state = STATE_UNROT3R; | |
| 12624 ++depth; | |
| 330 | 12625 p = fword + sp->ts_fidx; |
| 12626 #ifdef FEAT_MBYTE | |
| 12627 if (has_mbyte) | |
| 12628 { | |
| 625 | 12629 n = mb_cptr2len(p); |
| 12630 n += mb_cptr2len(p + n); | |
| 330 | 12631 c = mb_ptr2char(p + n); |
| 625 | 12632 tl = mb_cptr2len(p + n); |
| 330 | 12633 mch_memmove(p + tl, p, n); |
| 12634 mb_char2bytes(c, p); | |
| 625 | 12635 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl; |
| 330 | 12636 } |
| 12637 else | |
| 12638 #endif | |
| 12639 { | |
| 12640 c = p[2]; | |
| 12641 p[2] = p[1]; | |
| 12642 p[1] = *p; | |
| 12643 *p = c; | |
| 625 | 12644 stack[depth].ts_fidxtry = sp->ts_fidx + 3; |
| 12645 } | |
| 12646 } | |
| 12647 else | |
| 12648 sp->ts_state = STATE_REP_INI; | |
| 12649 break; | |
| 12650 | |
| 12651 case STATE_UNROT3R: | |
| 12652 /* Undo ROT3R: "312" -> "123" */ | |
| 12653 p = fword + sp->ts_fidx; | |
| 12654 #ifdef FEAT_MBYTE | |
| 12655 if (has_mbyte) | |
| 12656 { | |
| 12657 c = mb_ptr2char(p); | |
| 12658 tl = MB_BYTE2LEN(*p); | |
| 12659 n = MB_BYTE2LEN(p[tl]); | |
| 12660 n += MB_BYTE2LEN(p[tl + n]); | |
| 12661 mch_memmove(p, p + tl, n); | |
| 12662 mb_char2bytes(c, p + n); | |
| 12663 } | |
| 12664 else | |
| 12665 #endif | |
| 12666 { | |
| 12667 c = *p; | |
| 12668 *p = p[1]; | |
| 12669 p[1] = p[2]; | |
| 12670 p[2] = c; | |
| 12671 } | |
| 12672 /*FALLTHROUGH*/ | |
| 12673 | |
| 12674 case STATE_REP_INI: | |
| 12675 /* Check if matching with REP items from the .aff file would work. | |
| 12676 * Quickly skip if: | |
| 12677 * - there are no REP items and we are not in the soundfold trie | |
| 12678 * - the score is going to be too high anyway | |
| 12679 * - already applied a REP item or swapped here */ | |
| 12680 if ((lp->lp_replang == NULL && !soundfold) | |
| 12681 || sp->ts_score + SCORE_REP >= su->su_maxscore | |
| 12682 || sp->ts_fidx < sp->ts_fidxtry) | |
| 12683 { | |
| 12684 sp->ts_state = STATE_FINAL; | |
| 323 | 12685 break; |
| 625 | 12686 } |
| 12687 | |
| 12688 /* Use the first byte to quickly find the first entry that may | |
| 12689 * match. If the index is -1 there is none. */ | |
| 12690 if (soundfold) | |
| 12691 sp->ts_curi = slang->sl_repsal_first[fword[sp->ts_fidx]]; | |
| 12692 else | |
| 12693 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]]; | |
| 12694 | |
| 12695 if (sp->ts_curi < 0) | |
| 12696 { | |
| 12697 sp->ts_state = STATE_FINAL; | |
| 12698 break; | |
| 12699 } | |
| 12700 | |
| 12701 sp->ts_state = STATE_REP; | |
| 12702 /*FALLTHROUGH*/ | |
| 12703 | |
| 12704 case STATE_REP: | |
| 12705 /* Try matching with REP items from the .aff file. For each match | |
| 12706 * replace the characters and check if the resulting word is | |
| 12707 * valid. */ | |
| 12708 p = fword + sp->ts_fidx; | |
| 12709 | |
| 12710 if (soundfold) | |
| 12711 gap = &slang->sl_repsal; | |
| 12712 else | |
| 503 | 12713 gap = &lp->lp_replang->sl_rep; |
| 625 | 12714 while (sp->ts_curi < gap->ga_len) |
| 12715 { | |
| 12716 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++; | |
| 12717 if (*ftp->ft_from != *p) | |
| 12718 { | |
| 12719 /* past possible matching entries */ | |
| 12720 sp->ts_curi = gap->ga_len; | |
| 323 | 12721 break; |
| 12722 } | |
| 625 | 12723 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0 |
| 12724 && TRY_DEEPER(su, stack, depth, SCORE_REP)) | |
| 12725 { | |
| 12726 go_deeper(stack, depth, SCORE_REP); | |
| 12727 #ifdef DEBUG_TRIEWALK | |
| 12728 sprintf(changename[depth], "%.*s-%s: replace %s with %s", | |
| 12729 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 12730 ftp->ft_from, ftp->ft_to); | |
| 12731 #endif | |
| 12732 /* Need to undo this afterwards. */ | |
| 12733 sp->ts_state = STATE_REP_UNDO; | |
| 12734 | |
| 12735 /* Change the "from" to the "to" string. */ | |
| 12736 ++depth; | |
| 835 | 12737 fl = (int)STRLEN(ftp->ft_from); |
| 12738 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 12739 if (fl != tl) |
| 323 | 12740 { |
| 1619 | 12741 STRMOVE(p + tl, p + fl); |
| 625 | 12742 repextra += tl - fl; |
| 323 | 12743 } |
| 625 | 12744 mch_memmove(p, ftp->ft_to, tl); |
| 12745 stack[depth].ts_fidxtry = sp->ts_fidx + tl; | |
| 12746 #ifdef FEAT_MBYTE | |
| 12747 stack[depth].ts_tcharlen = 0; | |
| 12748 #endif | |
| 12749 break; | |
| 12750 } | |
| 12751 } | |
| 12752 | |
| 12753 if (sp->ts_curi >= gap->ga_len && sp->ts_state == STATE_REP) | |
| 12754 /* No (more) matches. */ | |
| 12755 sp->ts_state = STATE_FINAL; | |
| 12756 | |
| 12757 break; | |
| 12758 | |
| 12759 case STATE_REP_UNDO: | |
| 12760 /* Undo a REP replacement and continue with the next one. */ | |
| 12761 if (soundfold) | |
| 12762 gap = &slang->sl_repsal; | |
| 12763 else | |
| 12764 gap = &lp->lp_replang->sl_rep; | |
| 12765 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1; | |
| 835 | 12766 fl = (int)STRLEN(ftp->ft_from); |
| 12767 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 12768 p = fword + sp->ts_fidx; |
| 12769 if (fl != tl) | |
| 12770 { | |
| 1619 | 12771 STRMOVE(p + fl, p + tl); |
| 625 | 12772 repextra -= tl - fl; |
| 12773 } | |
| 12774 mch_memmove(p, ftp->ft_from, fl); | |
| 12775 sp->ts_state = STATE_REP; | |
| 12776 break; | |
| 12777 | |
| 12778 default: | |
| 12779 /* Did all possible states at this level, go up one level. */ | |
| 12780 --depth; | |
| 12781 | |
| 12782 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE) | |
| 12783 { | |
| 12784 /* Continue in or go back to the prefix tree. */ | |
| 12785 byts = pbyts; | |
| 12786 idxs = pidxs; | |
| 12787 } | |
| 12788 | |
| 12789 /* Don't check for CTRL-C too often, it takes time. */ | |
| 12790 if (--breakcheckcount == 0) | |
| 12791 { | |
| 12792 ui_breakcheck(); | |
| 12793 breakcheckcount = 1000; | |
| 12794 } | |
| 12795 } | |
| 12796 } | |
| 12797 } | |
| 12798 | |
| 12799 | |
| 12800 /* | |
| 12801 * Go one level deeper in the tree. | |
| 12802 */ | |
| 12803 static void | |
| 12804 go_deeper(stack, depth, score_add) | |
| 323 | 12805 trystate_T *stack; |
| 12806 int depth; | |
| 12807 int score_add; | |
| 12808 { | |
| 330 | 12809 stack[depth + 1] = stack[depth]; |
| 323 | 12810 stack[depth + 1].ts_state = STATE_START; |
| 625 | 12811 stack[depth + 1].ts_score = stack[depth].ts_score + score_add; |
| 323 | 12812 stack[depth + 1].ts_curi = 1; /* start just after length byte */ |
| 485 | 12813 stack[depth + 1].ts_flags = 0; |
| 323 | 12814 } |
| 12815 | |
| 455 | 12816 #ifdef FEAT_MBYTE |
| 12817 /* | |
| 12818 * Case-folding may change the number of bytes: Count nr of chars in | |
| 12819 * fword[flen] and return the byte length of that many chars in "word". | |
| 12820 */ | |
| 12821 static int | |
| 12822 nofold_len(fword, flen, word) | |
| 12823 char_u *fword; | |
| 12824 int flen; | |
| 12825 char_u *word; | |
| 12826 { | |
| 12827 char_u *p; | |
| 12828 int i = 0; | |
| 12829 | |
| 12830 for (p = fword; p < fword + flen; mb_ptr_adv(p)) | |
| 12831 ++i; | |
| 12832 for (p = word; i > 0; mb_ptr_adv(p)) | |
| 12833 --i; | |
| 12834 return (int)(p - word); | |
| 12835 } | |
| 12836 #endif | |
| 12837 | |
| 323 | 12838 /* |
| 12839 * "fword" is a good word with case folded. Find the matching keep-case | |
| 12840 * words and put it in "kword". | |
| 12841 * Theoretically there could be several keep-case words that result in the | |
| 12842 * same case-folded word, but we only find one... | |
| 12843 */ | |
| 12844 static void | |
| 12845 find_keepcap_word(slang, fword, kword) | |
| 12846 slang_T *slang; | |
| 12847 char_u *fword; | |
| 12848 char_u *kword; | |
| 12849 { | |
| 12850 char_u uword[MAXWLEN]; /* "fword" in upper-case */ | |
| 12851 int depth; | |
| 324 | 12852 idx_T tryidx; |
| 323 | 12853 |
| 12854 /* The following arrays are used at each depth in the tree. */ | |
| 324 | 12855 idx_T arridx[MAXWLEN]; |
| 323 | 12856 int round[MAXWLEN]; |
| 12857 int fwordidx[MAXWLEN]; | |
| 12858 int uwordidx[MAXWLEN]; | |
| 12859 int kwordlen[MAXWLEN]; | |
| 12860 | |
| 12861 int flen, ulen; | |
| 12862 int l; | |
| 12863 int len; | |
| 12864 int c; | |
| 324 | 12865 idx_T lo, hi, m; |
| 323 | 12866 char_u *p; |
| 12867 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */ | |
| 324 | 12868 idx_T *idxs = slang->sl_kidxs; /* array with indexes */ |
| 323 | 12869 |
| 12870 if (byts == NULL) | |
| 12871 { | |
| 12872 /* array is empty: "cannot happen" */ | |
| 12873 *kword = NUL; | |
| 12874 return; | |
| 12875 } | |
| 12876 | |
| 12877 /* Make an all-cap version of "fword". */ | |
| 12878 allcap_copy(fword, uword); | |
| 12879 | |
| 12880 /* | |
| 12881 * Each character needs to be tried both case-folded and upper-case. | |
| 12882 * All this gets very complicated if we keep in mind that changing case | |
| 12883 * may change the byte length of a multi-byte character... | |
| 12884 */ | |
| 12885 depth = 0; | |
| 12886 arridx[0] = 0; | |
| 12887 round[0] = 0; | |
| 12888 fwordidx[0] = 0; | |
| 12889 uwordidx[0] = 0; | |
| 12890 kwordlen[0] = 0; | |
| 12891 while (depth >= 0) | |
| 12892 { | |
| 12893 if (fword[fwordidx[depth]] == NUL) | |
| 12894 { | |
| 12895 /* We are at the end of "fword". If the tree allows a word to end | |
| 12896 * here we have found a match. */ | |
| 12897 if (byts[arridx[depth] + 1] == 0) | |
| 12898 { | |
| 12899 kword[kwordlen[depth]] = NUL; | |
| 12900 return; | |
| 12901 } | |
| 12902 | |
| 12903 /* kword is getting too long, continue one level up */ | |
| 12904 --depth; | |
| 12905 } | |
| 12906 else if (++round[depth] > 2) | |
| 12907 { | |
| 12908 /* tried both fold-case and upper-case character, continue one | |
| 12909 * level up */ | |
| 12910 --depth; | |
| 12911 } | |
| 12912 else | |
| 12913 { | |
| 12914 /* | |
| 12915 * round[depth] == 1: Try using the folded-case character. | |
| 12916 * round[depth] == 2: Try using the upper-case character. | |
| 12917 */ | |
| 12918 #ifdef FEAT_MBYTE | |
| 12919 if (has_mbyte) | |
| 12920 { | |
| 474 | 12921 flen = mb_cptr2len(fword + fwordidx[depth]); |
| 12922 ulen = mb_cptr2len(uword + uwordidx[depth]); | |
| 323 | 12923 } |
| 12924 else | |
| 12925 #endif | |
| 12926 ulen = flen = 1; | |
| 12927 if (round[depth] == 1) | |
| 12928 { | |
| 12929 p = fword + fwordidx[depth]; | |
| 12930 l = flen; | |
| 12931 } | |
| 12932 else | |
| 12933 { | |
| 12934 p = uword + uwordidx[depth]; | |
| 12935 l = ulen; | |
| 12936 } | |
| 12937 | |
| 12938 for (tryidx = arridx[depth]; l > 0; --l) | |
| 12939 { | |
| 12940 /* Perform a binary search in the list of accepted bytes. */ | |
| 12941 len = byts[tryidx++]; | |
| 12942 c = *p++; | |
| 12943 lo = tryidx; | |
| 12944 hi = tryidx + len - 1; | |
| 12945 while (lo < hi) | |
| 12946 { | |
| 12947 m = (lo + hi) / 2; | |
| 12948 if (byts[m] > c) | |
| 12949 hi = m - 1; | |
| 12950 else if (byts[m] < c) | |
| 12951 lo = m + 1; | |
| 12952 else | |
| 12953 { | |
| 12954 lo = hi = m; | |
| 12955 break; | |
| 12956 } | |
| 12957 } | |
| 12958 | |
| 12959 /* Stop if there is no matching byte. */ | |
| 12960 if (hi < lo || byts[lo] != c) | |
| 12961 break; | |
| 12962 | |
| 12963 /* Continue at the child (if there is one). */ | |
| 12964 tryidx = idxs[lo]; | |
| 12965 } | |
| 12966 | |
| 12967 if (l == 0) | |
| 12968 { | |
| 12969 /* | |
| 12970 * Found the matching char. Copy it to "kword" and go a | |
| 12971 * level deeper. | |
| 12972 */ | |
| 12973 if (round[depth] == 1) | |
| 12974 { | |
| 12975 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth], | |
| 12976 flen); | |
| 12977 kwordlen[depth + 1] = kwordlen[depth] + flen; | |
| 12978 } | |
| 12979 else | |
| 12980 { | |
| 12981 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth], | |
| 12982 ulen); | |
| 12983 kwordlen[depth + 1] = kwordlen[depth] + ulen; | |
| 12984 } | |
| 12985 fwordidx[depth + 1] = fwordidx[depth] + flen; | |
| 12986 uwordidx[depth + 1] = uwordidx[depth] + ulen; | |
| 12987 | |
| 12988 ++depth; | |
| 12989 arridx[depth] = tryidx; | |
| 12990 round[depth] = 0; | |
| 12991 } | |
| 12992 } | |
| 12993 } | |
| 12994 | |
| 12995 /* Didn't find it: "cannot happen". */ | |
| 12996 *kword = NUL; | |
| 12997 } | |
| 12998 | |
| 12999 /* | |
| 344 | 13000 * Compute the sound-a-like score for suggestions in su->su_ga and add them to |
| 13001 * su->su_sga. | |
| 13002 */ | |
| 13003 static void | |
| 13004 score_comp_sal(su) | |
| 13005 suginfo_T *su; | |
| 13006 { | |
| 13007 langp_T *lp; | |
| 13008 char_u badsound[MAXWLEN]; | |
| 13009 int i; | |
| 13010 suggest_T *stp; | |
| 13011 suggest_T *sstp; | |
| 13012 int score; | |
| 500 | 13013 int lpi; |
| 344 | 13014 |
| 13015 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL) | |
| 13016 return; | |
| 13017 | |
| 13018 /* 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
|
13019 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
|
13020 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13021 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 13022 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 13023 { | |
| 13024 /* soundfold the bad word */ | |
| 375 | 13025 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound); |
| 344 | 13026 |
| 13027 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 13028 { | |
| 13029 stp = &SUG(su->su_ga, i); | |
| 13030 | |
| 351 | 13031 /* Case-fold the suggested word, sound-fold it and compute the |
| 13032 * sound-a-like score. */ | |
| 13033 score = stp_sal_score(stp, su, lp->lp_slang, badsound); | |
| 344 | 13034 if (score < SCORE_MAXMAX) |
| 13035 { | |
| 13036 /* Add the suggestion. */ | |
| 13037 sstp = &SUG(su->su_sga, su->su_sga.ga_len); | |
| 13038 sstp->st_word = vim_strsave(stp->st_word); | |
| 13039 if (sstp->st_word != NULL) | |
| 13040 { | |
| 625 | 13041 sstp->st_wordlen = stp->st_wordlen; |
| 344 | 13042 sstp->st_score = score; |
| 13043 sstp->st_altscore = 0; | |
| 13044 sstp->st_orglen = stp->st_orglen; | |
| 13045 ++su->su_sga.ga_len; | |
| 13046 } | |
| 13047 } | |
| 13048 } | |
| 13049 break; | |
| 13050 } | |
| 500 | 13051 } |
| 344 | 13052 } |
| 13053 | |
| 13054 /* | |
| 13055 * Combine the list of suggestions in su->su_ga and su->su_sga. | |
| 4352 | 13056 * They are entwined. |
| 344 | 13057 */ |
| 13058 static void | |
| 13059 score_combine(su) | |
| 13060 suginfo_T *su; | |
| 13061 { | |
| 13062 int i; | |
| 13063 int j; | |
| 13064 garray_T ga; | |
| 13065 garray_T *gap; | |
| 13066 langp_T *lp; | |
| 13067 suggest_T *stp; | |
| 13068 char_u *p; | |
| 13069 char_u badsound[MAXWLEN]; | |
| 13070 int round; | |
| 500 | 13071 int lpi; |
| 625 | 13072 slang_T *slang = NULL; |
| 344 | 13073 |
| 13074 /* 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
|
13075 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
|
13076 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13077 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 13078 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 13079 { | |
| 13080 /* soundfold the bad word */ | |
| 625 | 13081 slang = lp->lp_slang; |
| 13082 spell_soundfold(slang, su->su_fbadword, TRUE, badsound); | |
| 344 | 13083 |
| 13084 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 13085 { | |
| 13086 stp = &SUG(su->su_ga, i); | |
| 625 | 13087 stp->st_altscore = stp_sal_score(stp, su, slang, badsound); |
| 344 | 13088 if (stp->st_altscore == SCORE_MAXMAX) |
| 13089 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4; | |
| 13090 else | |
| 13091 stp->st_score = (stp->st_score * 3 | |
| 13092 + stp->st_altscore) / 4; | |
| 13093 stp->st_salscore = FALSE; | |
| 13094 } | |
| 13095 break; | |
| 13096 } | |
| 13097 } | |
| 13098 | |
| 844 | 13099 if (slang == NULL) /* Using "double" without sound folding. */ |
| 13100 { | |
| 13101 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, | |
| 13102 su->su_maxcount); | |
| 625 | 13103 return; |
| 844 | 13104 } |
| 625 | 13105 |
| 344 | 13106 /* Add the alternate score to su_sga. */ |
| 13107 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 13108 { | |
| 13109 stp = &SUG(su->su_sga, i); | |
| 625 | 13110 stp->st_altscore = spell_edit_score(slang, |
| 13111 su->su_badword, stp->st_word); | |
| 344 | 13112 if (stp->st_score == SCORE_MAXMAX) |
| 13113 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8; | |
| 13114 else | |
| 13115 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8; | |
| 13116 stp->st_salscore = TRUE; | |
| 13117 } | |
| 13118 | |
| 625 | 13119 /* Remove bad suggestions, sort the suggestions and truncate at "maxcount" |
| 13120 * for both lists. */ | |
| 13121 check_suggestions(su, &su->su_ga); | |
| 344 | 13122 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 625 | 13123 check_suggestions(su, &su->su_sga); |
| 344 | 13124 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount); |
| 13125 | |
| 13126 ga_init2(&ga, (int)sizeof(suginfo_T), 1); | |
| 13127 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL) | |
| 13128 return; | |
| 13129 | |
| 13130 stp = &SUG(ga, 0); | |
| 13131 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i) | |
| 13132 { | |
| 13133 /* round 1: get a suggestion from su_ga | |
| 13134 * round 2: get a suggestion from su_sga */ | |
| 13135 for (round = 1; round <= 2; ++round) | |
| 13136 { | |
| 13137 gap = round == 1 ? &su->su_ga : &su->su_sga; | |
| 13138 if (i < gap->ga_len) | |
| 13139 { | |
| 13140 /* Don't add a word if it's already there. */ | |
| 13141 p = SUG(*gap, i).st_word; | |
| 13142 for (j = 0; j < ga.ga_len; ++j) | |
| 13143 if (STRCMP(stp[j].st_word, p) == 0) | |
| 13144 break; | |
| 13145 if (j == ga.ga_len) | |
| 13146 stp[ga.ga_len++] = SUG(*gap, i); | |
| 13147 else | |
| 13148 vim_free(p); | |
| 13149 } | |
| 13150 } | |
| 13151 } | |
| 13152 | |
| 13153 ga_clear(&su->su_ga); | |
| 13154 ga_clear(&su->su_sga); | |
| 13155 | |
| 13156 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 13157 if (ga.ga_len > su->su_maxcount) | |
| 13158 { | |
| 13159 for (i = su->su_maxcount; i < ga.ga_len; ++i) | |
| 13160 vim_free(stp[i].st_word); | |
| 13161 ga.ga_len = su->su_maxcount; | |
| 13162 } | |
| 13163 | |
| 13164 su->su_ga = ga; | |
| 13165 } | |
| 13166 | |
| 13167 /* | |
| 351 | 13168 * For the goodword in "stp" compute the soundalike score compared to the |
| 13169 * badword. | |
| 13170 */ | |
| 13171 static int | |
| 13172 stp_sal_score(stp, su, slang, badsound) | |
| 13173 suggest_T *stp; | |
| 13174 suginfo_T *su; | |
| 13175 slang_T *slang; | |
| 13176 char_u *badsound; /* sound-folded badword */ | |
| 13177 { | |
| 13178 char_u *p; | |
| 534 | 13179 char_u *pbad; |
| 13180 char_u *pgood; | |
| 351 | 13181 char_u badsound2[MAXWLEN]; |
| 13182 char_u fword[MAXWLEN]; | |
| 13183 char_u goodsound[MAXWLEN]; | |
| 534 | 13184 char_u goodword[MAXWLEN]; |
| 13185 int lendiff; | |
| 13186 | |
| 13187 lendiff = (int)(su->su_badlen - stp->st_orglen); | |
| 13188 if (lendiff >= 0) | |
| 13189 pbad = badsound; | |
| 351 | 13190 else |
| 13191 { | |
| 13192 /* soundfold the bad word with more characters following */ | |
| 13193 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN); | |
| 13194 | |
| 13195 /* When joining two words the sound often changes a lot. E.g., "t he" | |
| 13196 * sounds like "t h" while "the" sounds like "@". Avoid that by | |
| 13197 * removing the space. Don't do it when the good word also contains a | |
| 13198 * space. */ | |
| 13199 if (vim_iswhite(su->su_badptr[su->su_badlen]) | |
| 13200 && *skiptowhite(stp->st_word) == NUL) | |
| 13201 for (p = fword; *(p = skiptowhite(p)) != NUL; ) | |
| 1619 | 13202 STRMOVE(p, p + 1); |
| 351 | 13203 |
| 375 | 13204 spell_soundfold(slang, fword, TRUE, badsound2); |
| 534 | 13205 pbad = badsound2; |
| 13206 } | |
| 13207 | |
| 2768 | 13208 if (lendiff > 0 && stp->st_wordlen + lendiff < MAXWLEN) |
| 534 | 13209 { |
| 13210 /* Add part of the bad word to the good word, so that we soundfold | |
| 13211 * what replaces the bad word. */ | |
| 13212 STRCPY(goodword, stp->st_word); | |
| 625 | 13213 vim_strncpy(goodword + stp->st_wordlen, |
| 13214 su->su_badptr + su->su_badlen - lendiff, lendiff); | |
| 534 | 13215 pgood = goodword; |
| 13216 } | |
| 13217 else | |
| 13218 pgood = stp->st_word; | |
| 351 | 13219 |
| 375 | 13220 /* Sound-fold the word and compute the score for the difference. */ |
| 534 | 13221 spell_soundfold(slang, pgood, FALSE, goodsound); |
| 13222 | |
| 13223 return soundalike_score(goodsound, pbad); | |
| 351 | 13224 } |
| 13225 | |
| 625 | 13226 /* structure used to store soundfolded words that add_sound_suggest() has |
| 13227 * handled already. */ | |
| 13228 typedef struct | |
| 13229 { | |
| 13230 short sft_score; /* lowest score used */ | |
| 13231 char_u sft_word[1]; /* soundfolded word, actually longer */ | |
| 13232 } sftword_T; | |
| 13233 | |
| 13234 static sftword_T dumsft; | |
| 13235 #define HIKEY2SFT(p) ((sftword_T *)(p - (dumsft.sft_word - (char_u *)&dumsft))) | |
| 13236 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key) | |
| 13237 | |
| 13238 /* | |
| 13239 * Prepare for calling suggest_try_soundalike(). | |
| 13240 */ | |
| 13241 static void | |
| 13242 suggest_try_soundalike_prep() | |
| 13243 { | |
| 13244 langp_T *lp; | |
| 13245 int lpi; | |
| 13246 slang_T *slang; | |
| 13247 | |
| 13248 /* Do this for all languages that support sound folding and for which a | |
| 13249 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13250 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
|
13251 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13252 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 13253 slang = lp->lp_slang; |
| 13254 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 13255 /* prepare the hashtable used by add_sound_suggest() */ | |
| 13256 hash_init(&slang->sl_sounddone); | |
| 13257 } | |
| 13258 } | |
| 13259 | |
| 351 | 13260 /* |
| 323 | 13261 * Find suggestions by comparing the word in a sound-a-like form. |
| 503 | 13262 * Note: This doesn't support postponed prefixes. |
| 323 | 13263 */ |
| 13264 static void | |
| 346 | 13265 suggest_try_soundalike(su) |
| 323 | 13266 suginfo_T *su; |
| 13267 { | |
| 13268 char_u salword[MAXWLEN]; | |
| 13269 langp_T *lp; | |
| 500 | 13270 int lpi; |
| 503 | 13271 slang_T *slang; |
| 344 | 13272 |
| 625 | 13273 /* Do this for all languages that support sound folding and for which a |
| 13274 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13275 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
|
13276 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13277 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 13278 slang = lp->lp_slang; |
| 13279 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 13280 { | |
| 13281 /* soundfold the bad word */ | |
| 13282 spell_soundfold(slang, su->su_fbadword, TRUE, salword); | |
| 13283 | |
| 13284 /* try all kinds of inserts/deletes/swaps/etc. */ | |
| 13285 /* TODO: also soundfold the next words, so that we can try joining | |
| 13286 * and splitting */ | |
| 13287 suggest_trie_walk(su, lp, salword, TRUE); | |
| 13288 } | |
| 13289 } | |
| 13290 } | |
| 13291 | |
| 13292 /* | |
| 13293 * Finish up after calling suggest_try_soundalike(). | |
| 13294 */ | |
| 13295 static void | |
| 13296 suggest_try_soundalike_finish() | |
| 13297 { | |
| 13298 langp_T *lp; | |
| 13299 int lpi; | |
| 13300 slang_T *slang; | |
| 13301 int todo; | |
| 13302 hashitem_T *hi; | |
| 13303 | |
| 13304 /* Do this for all languages that support sound folding and for which a | |
| 13305 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13306 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
|
13307 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
13308 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 503 | 13309 slang = lp->lp_slang; |
| 625 | 13310 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) |
| 13311 { | |
| 13312 /* Free the info about handled words. */ | |
| 835 | 13313 todo = (int)slang->sl_sounddone.ht_used; |
| 625 | 13314 for (hi = slang->sl_sounddone.ht_array; todo > 0; ++hi) |
| 13315 if (!HASHITEM_EMPTY(hi)) | |
| 13316 { | |
| 13317 vim_free(HI2SFT(hi)); | |
| 13318 --todo; | |
| 13319 } | |
| 1087 | 13320 |
| 13321 /* Clear the hashtable, it may also be used by another region. */ | |
| 625 | 13322 hash_clear(&slang->sl_sounddone); |
| 1087 | 13323 hash_init(&slang->sl_sounddone); |
| 625 | 13324 } |
| 13325 } | |
| 13326 } | |
| 13327 | |
| 13328 /* | |
| 13329 * A match with a soundfolded word is found. Add the good word(s) that | |
| 13330 * produce this soundfolded word. | |
| 13331 */ | |
| 13332 static void | |
| 13333 add_sound_suggest(su, goodword, score, lp) | |
| 13334 suginfo_T *su; | |
| 13335 char_u *goodword; | |
| 13336 int score; /* soundfold score */ | |
| 13337 langp_T *lp; | |
| 13338 { | |
| 13339 slang_T *slang = lp->lp_slang; /* language for sound folding */ | |
| 13340 int sfwordnr; | |
| 13341 char_u *nrline; | |
| 13342 int orgnr; | |
| 13343 char_u theword[MAXWLEN]; | |
| 13344 int i; | |
| 13345 int wlen; | |
| 13346 char_u *byts; | |
| 13347 idx_T *idxs; | |
| 13348 int n; | |
| 13349 int wordcount; | |
| 13350 int wc; | |
| 13351 int goodscore; | |
| 13352 hash_T hash; | |
| 13353 hashitem_T *hi; | |
| 13354 sftword_T *sft; | |
| 13355 int bc, gc; | |
| 13356 int limit; | |
| 13357 | |
| 13358 /* | |
| 13359 * It's very well possible that the same soundfold word is found several | |
| 13360 * times with different scores. Since the following is quite slow only do | |
| 13361 * the words that have a better score than before. Use a hashtable to | |
| 13362 * remember the words that have been done. | |
| 13363 */ | |
| 13364 hash = hash_hash(goodword); | |
| 13365 hi = hash_lookup(&slang->sl_sounddone, goodword, hash); | |
| 13366 if (HASHITEM_EMPTY(hi)) | |
| 13367 { | |
| 844 | 13368 sft = (sftword_T *)alloc((unsigned)(sizeof(sftword_T) |
| 13369 + STRLEN(goodword))); | |
| 625 | 13370 if (sft != NULL) |
| 13371 { | |
| 13372 sft->sft_score = score; | |
| 13373 STRCPY(sft->sft_word, goodword); | |
| 13374 hash_add_item(&slang->sl_sounddone, hi, sft->sft_word, hash); | |
| 13375 } | |
| 13376 } | |
| 13377 else | |
| 13378 { | |
| 13379 sft = HI2SFT(hi); | |
| 13380 if (score >= sft->sft_score) | |
| 13381 return; | |
| 13382 sft->sft_score = score; | |
| 13383 } | |
| 13384 | |
| 13385 /* | |
| 13386 * Find the word nr in the soundfold tree. | |
| 13387 */ | |
| 13388 sfwordnr = soundfold_find(slang, goodword); | |
| 13389 if (sfwordnr < 0) | |
| 13390 { | |
| 13391 EMSG2(_(e_intern2), "add_sound_suggest()"); | |
| 13392 return; | |
| 13393 } | |
| 13394 | |
| 13395 /* | |
| 13396 * go over the list of good words that produce this soundfold word | |
| 13397 */ | |
| 13398 nrline = ml_get_buf(slang->sl_sugbuf, (linenr_T)(sfwordnr + 1), FALSE); | |
| 13399 orgnr = 0; | |
| 13400 while (*nrline != NUL) | |
| 13401 { | |
| 13402 /* The wordnr was stored in a minimal nr of bytes as an offset to the | |
| 13403 * previous wordnr. */ | |
| 13404 orgnr += bytes2offset(&nrline); | |
| 13405 | |
| 13406 byts = slang->sl_fbyts; | |
| 13407 idxs = slang->sl_fidxs; | |
| 13408 | |
| 13409 /* Lookup the word "orgnr" one of the two tries. */ | |
| 13410 n = 0; | |
| 13411 wordcount = 0; | |
| 5506 | 13412 for (wlen = 0; wlen < MAXWLEN - 3; ++wlen) |
| 625 | 13413 { |
| 13414 i = 1; | |
| 13415 if (wordcount == orgnr && byts[n + 1] == NUL) | |
| 13416 break; /* found end of word */ | |
| 13417 | |
| 13418 if (byts[n + 1] == NUL) | |
| 13419 ++wordcount; | |
| 13420 | |
| 13421 /* skip over the NUL bytes */ | |
| 13422 for ( ; byts[n + i] == NUL; ++i) | |
| 13423 if (i > byts[n]) /* safety check */ | |
| 13424 { | |
| 13425 STRCPY(theword + wlen, "BAD"); | |
| 5506 | 13426 wlen += 3; |
| 625 | 13427 goto badword; |
| 13428 } | |
| 13429 | |
| 13430 /* One of the siblings must have the word. */ | |
| 13431 for ( ; i < byts[n]; ++i) | |
| 13432 { | |
| 13433 wc = idxs[idxs[n + i]]; /* nr of words under this byte */ | |
| 13434 if (wordcount + wc > orgnr) | |
| 13435 break; | |
| 13436 wordcount += wc; | |
| 13437 } | |
| 13438 | |
| 5506 | 13439 theword[wlen] = byts[n + i]; |
| 625 | 13440 n = idxs[n + i]; |
| 13441 } | |
| 13442 badword: | |
| 13443 theword[wlen] = NUL; | |
| 13444 | |
| 13445 /* Go over the possible flags and regions. */ | |
| 13446 for (; i <= byts[n] && byts[n + i] == NUL; ++i) | |
| 13447 { | |
| 13448 char_u cword[MAXWLEN]; | |
| 13449 char_u *p; | |
| 13450 int flags = (int)idxs[n + i]; | |
| 13451 | |
| 698 | 13452 /* Skip words with the NOSUGGEST flag */ |
| 13453 if (flags & WF_NOSUGGEST) | |
| 13454 continue; | |
| 13455 | |
| 625 | 13456 if (flags & WF_KEEPCAP) |
| 13457 { | |
| 13458 /* Must find the word in the keep-case tree. */ | |
| 13459 find_keepcap_word(slang, theword, cword); | |
| 13460 p = cword; | |
| 13461 } | |
| 13462 else | |
| 13463 { | |
| 13464 flags |= su->su_badflags; | |
| 13465 if ((flags & WF_CAPMASK) != 0) | |
| 13466 { | |
| 13467 /* Need to fix case according to "flags". */ | |
| 13468 make_case_word(theword, cword, flags); | |
| 13469 p = cword; | |
| 323 | 13470 } |
| 13471 else | |
| 625 | 13472 p = theword; |
| 13473 } | |
| 13474 | |
| 13475 /* Add the suggestion. */ | |
| 13476 if (sps_flags & SPS_DOUBLE) | |
| 13477 { | |
| 13478 /* Add the suggestion if the score isn't too bad. */ | |
| 13479 if (score <= su->su_maxscore) | |
| 13480 add_suggestion(su, &su->su_sga, p, su->su_badlen, | |
| 13481 score, 0, FALSE, slang, FALSE); | |
| 13482 } | |
| 13483 else | |
| 13484 { | |
| 13485 /* Add a penalty for words in another region. */ | |
| 13486 if ((flags & WF_REGION) | |
| 13487 && (((unsigned)flags >> 16) & lp->lp_region) == 0) | |
| 13488 goodscore = SCORE_REGION; | |
| 13489 else | |
| 13490 goodscore = 0; | |
| 13491 | |
| 13492 /* Add a small penalty for changing the first letter from | |
| 13493 * lower to upper case. Helps for "tath" -> "Kath", which is | |
| 4352 | 13494 * less common than "tath" -> "path". Don't do it when the |
| 625 | 13495 * letter is the same, that has already been counted. */ |
| 13496 gc = PTR2CHAR(p); | |
| 13497 if (SPELL_ISUPPER(gc)) | |
| 13498 { | |
| 13499 bc = PTR2CHAR(su->su_badword); | |
| 13500 if (!SPELL_ISUPPER(bc) | |
| 13501 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc)) | |
| 13502 goodscore += SCORE_ICASE / 2; | |
| 13503 } | |
| 13504 | |
| 13505 /* Compute the score for the good word. This only does letter | |
| 13506 * insert/delete/swap/replace. REP items are not considered, | |
| 13507 * which may make the score a bit higher. | |
| 13508 * Use a limit for the score to make it work faster. Use | |
| 13509 * MAXSCORE(), because RESCORE() will change the score. | |
| 13510 * If the limit is very high then the iterative method is | |
| 13511 * inefficient, using an array is quicker. */ | |
| 13512 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score); | |
| 13513 if (limit > SCORE_LIMITMAX) | |
| 13514 goodscore += spell_edit_score(slang, su->su_badword, p); | |
| 13515 else | |
| 13516 goodscore += spell_edit_score_limit(slang, su->su_badword, | |
| 13517 p, limit); | |
| 13518 | |
| 13519 /* When going over the limit don't bother to do the rest. */ | |
| 13520 if (goodscore < SCORE_MAXMAX) | |
| 13521 { | |
| 13522 /* Give a bonus to words seen before. */ | |
| 13523 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE); | |
| 13524 | |
| 13525 /* Add the suggestion if the score isn't too bad. */ | |
| 13526 goodscore = RESCORE(goodscore, score); | |
| 13527 if (goodscore <= su->su_sfmaxscore) | |
| 13528 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 13529 goodscore, score, TRUE, slang, TRUE); | |
| 13530 } | |
| 13531 } | |
| 13532 } | |
| 13533 /* smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr); */ | |
| 13534 } | |
| 13535 } | |
| 13536 | |
| 13537 /* | |
| 13538 * Find word "word" in fold-case tree for "slang" and return the word number. | |
| 13539 */ | |
| 13540 static int | |
| 13541 soundfold_find(slang, word) | |
| 13542 slang_T *slang; | |
| 13543 char_u *word; | |
| 13544 { | |
| 13545 idx_T arridx = 0; | |
| 13546 int len; | |
| 13547 int wlen = 0; | |
| 13548 int c; | |
| 13549 char_u *ptr = word; | |
| 13550 char_u *byts; | |
| 13551 idx_T *idxs; | |
| 13552 int wordnr = 0; | |
| 13553 | |
| 13554 byts = slang->sl_sbyts; | |
| 13555 idxs = slang->sl_sidxs; | |
| 13556 | |
| 13557 for (;;) | |
| 13558 { | |
| 13559 /* First byte is the number of possible bytes. */ | |
| 13560 len = byts[arridx++]; | |
| 13561 | |
| 13562 /* If the first possible byte is a zero the word could end here. | |
| 13563 * If the word ends we found the word. If not skip the NUL bytes. */ | |
| 13564 c = ptr[wlen]; | |
| 13565 if (byts[arridx] == NUL) | |
| 13566 { | |
| 13567 if (c == NUL) | |
| 13568 break; | |
| 13569 | |
| 13570 /* Skip over the zeros, there can be several. */ | |
| 13571 while (len > 0 && byts[arridx] == NUL) | |
| 13572 { | |
| 13573 ++arridx; | |
| 13574 --len; | |
| 13575 } | |
| 13576 if (len == 0) | |
| 13577 return -1; /* no children, word should have ended here */ | |
| 13578 ++wordnr; | |
| 13579 } | |
| 13580 | |
| 13581 /* If the word ends we didn't find it. */ | |
| 13582 if (c == NUL) | |
| 13583 return -1; | |
| 13584 | |
| 13585 /* Perform a binary search in the list of accepted bytes. */ | |
| 13586 if (c == TAB) /* <Tab> is handled like <Space> */ | |
| 13587 c = ' '; | |
| 13588 while (byts[arridx] < c) | |
| 13589 { | |
| 13590 /* The word count is in the first idxs[] entry of the child. */ | |
| 13591 wordnr += idxs[idxs[arridx]]; | |
| 13592 ++arridx; | |
| 13593 if (--len == 0) /* end of the bytes, didn't find it */ | |
| 13594 return -1; | |
| 13595 } | |
| 13596 if (byts[arridx] != c) /* didn't find the byte */ | |
| 13597 return -1; | |
| 13598 | |
| 13599 /* Continue at the child (if there is one). */ | |
| 13600 arridx = idxs[arridx]; | |
| 13601 ++wlen; | |
| 13602 | |
| 13603 /* One space in the good word may stand for several spaces in the | |
| 13604 * checked word. */ | |
| 13605 if (c == ' ') | |
| 13606 while (ptr[wlen] == ' ' || ptr[wlen] == TAB) | |
| 13607 ++wlen; | |
| 13608 } | |
| 13609 | |
| 13610 return wordnr; | |
| 323 | 13611 } |
| 13612 | |
| 13613 /* | |
| 324 | 13614 * Copy "fword" to "cword", fixing case according to "flags". |
| 323 | 13615 */ |
| 13616 static void | |
| 13617 make_case_word(fword, cword, flags) | |
| 13618 char_u *fword; | |
| 13619 char_u *cword; | |
| 13620 int flags; | |
| 13621 { | |
| 13622 if (flags & WF_ALLCAP) | |
| 13623 /* Make it all upper-case */ | |
| 13624 allcap_copy(fword, cword); | |
| 13625 else if (flags & WF_ONECAP) | |
| 13626 /* Make the first letter upper-case */ | |
| 324 | 13627 onecap_copy(fword, cword, TRUE); |
| 323 | 13628 else |
| 13629 /* Use goodword as-is. */ | |
| 13630 STRCPY(cword, fword); | |
| 13631 } | |
| 13632 | |
| 330 | 13633 /* |
| 13634 * Use map string "map" for languages "lp". | |
| 13635 */ | |
| 13636 static void | |
| 13637 set_map_str(lp, map) | |
| 13638 slang_T *lp; | |
| 13639 char_u *map; | |
| 13640 { | |
| 13641 char_u *p; | |
| 13642 int headc = 0; | |
| 13643 int c; | |
| 13644 int i; | |
| 13645 | |
| 13646 if (*map == NUL) | |
| 13647 { | |
| 13648 lp->sl_has_map = FALSE; | |
| 13649 return; | |
| 13650 } | |
| 13651 lp->sl_has_map = TRUE; | |
| 13652 | |
| 625 | 13653 /* Init the array and hash tables empty. */ |
| 330 | 13654 for (i = 0; i < 256; ++i) |
| 13655 lp->sl_map_array[i] = 0; | |
| 13656 #ifdef FEAT_MBYTE | |
| 13657 hash_init(&lp->sl_map_hash); | |
| 13658 #endif | |
| 13659 | |
| 13660 /* | |
| 13661 * The similar characters are stored separated with slashes: | |
| 13662 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and | |
| 13663 * before the same slash. For characters above 255 sl_map_hash is used. | |
| 13664 */ | |
| 13665 for (p = map; *p != NUL; ) | |
| 13666 { | |
| 13667 #ifdef FEAT_MBYTE | |
| 474 | 13668 c = mb_cptr2char_adv(&p); |
| 330 | 13669 #else |
| 13670 c = *p++; | |
| 13671 #endif | |
| 13672 if (c == '/') | |
| 13673 headc = 0; | |
| 13674 else | |
| 13675 { | |
| 13676 if (headc == 0) | |
| 13677 headc = c; | |
| 13678 | |
| 13679 #ifdef FEAT_MBYTE | |
| 13680 /* Characters above 255 don't fit in sl_map_array[], put them in | |
| 13681 * the hash table. Each entry is the char, a NUL the headchar and | |
| 13682 * a NUL. */ | |
| 13683 if (c >= 256) | |
| 13684 { | |
| 13685 int cl = mb_char2len(c); | |
| 13686 int headcl = mb_char2len(headc); | |
| 13687 char_u *b; | |
| 13688 hash_T hash; | |
| 13689 hashitem_T *hi; | |
| 13690 | |
| 13691 b = alloc((unsigned)(cl + headcl + 2)); | |
| 13692 if (b == NULL) | |
| 13693 return; | |
| 13694 mb_char2bytes(c, b); | |
| 13695 b[cl] = NUL; | |
| 13696 mb_char2bytes(headc, b + cl + 1); | |
| 13697 b[cl + 1 + headcl] = NUL; | |
| 13698 hash = hash_hash(b); | |
| 13699 hi = hash_lookup(&lp->sl_map_hash, b, hash); | |
| 13700 if (HASHITEM_EMPTY(hi)) | |
| 13701 hash_add_item(&lp->sl_map_hash, hi, b, hash); | |
| 13702 else | |
| 13703 { | |
| 13704 /* This should have been checked when generating the .spl | |
| 13705 * file. */ | |
| 626 | 13706 EMSG(_("E783: duplicate char in MAP entry")); |
| 330 | 13707 vim_free(b); |
| 13708 } | |
| 13709 } | |
| 13710 else | |
| 13711 #endif | |
| 13712 lp->sl_map_array[c] = headc; | |
| 13713 } | |
| 13714 } | |
| 13715 } | |
| 13716 | |
| 323 | 13717 /* |
| 13718 * Return TRUE if "c1" and "c2" are similar characters according to the MAP | |
| 13719 * lines in the .aff file. | |
| 13720 */ | |
| 13721 static int | |
| 13722 similar_chars(slang, c1, c2) | |
| 13723 slang_T *slang; | |
| 13724 int c1; | |
| 13725 int c2; | |
| 13726 { | |
| 330 | 13727 int m1, m2; |
| 13728 #ifdef FEAT_MBYTE | |
| 3549 | 13729 char_u buf[MB_MAXBYTES + 1]; |
| 330 | 13730 hashitem_T *hi; |
| 13731 | |
| 13732 if (c1 >= 256) | |
| 13733 { | |
| 13734 buf[mb_char2bytes(c1, buf)] = 0; | |
| 13735 hi = hash_find(&slang->sl_map_hash, buf); | |
| 13736 if (HASHITEM_EMPTY(hi)) | |
| 13737 m1 = 0; | |
| 13738 else | |
| 13739 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 13740 } | |
| 13741 else | |
| 13742 #endif | |
| 13743 m1 = slang->sl_map_array[c1]; | |
| 13744 if (m1 == 0) | |
| 323 | 13745 return FALSE; |
| 330 | 13746 |
| 13747 | |
| 13748 #ifdef FEAT_MBYTE | |
| 13749 if (c2 >= 256) | |
| 13750 { | |
| 13751 buf[mb_char2bytes(c2, buf)] = 0; | |
| 13752 hi = hash_find(&slang->sl_map_hash, buf); | |
| 13753 if (HASHITEM_EMPTY(hi)) | |
| 13754 m2 = 0; | |
| 13755 else | |
| 13756 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 13757 } | |
| 13758 else | |
| 13759 #endif | |
| 13760 m2 = slang->sl_map_array[c2]; | |
| 13761 | |
| 13762 return m1 == m2; | |
| 323 | 13763 } |
| 13764 | |
| 13765 /* | |
| 13766 * Add a suggestion to the list of suggestions. | |
| 625 | 13767 * For a suggestion that is already in the list the lowest score is remembered. |
| 13768 */ | |
| 13769 static void | |
| 13770 add_suggestion(su, gap, goodword, badlenarg, score, altscore, had_bonus, | |
| 13771 slang, maxsf) | |
| 323 | 13772 suginfo_T *su; |
| 625 | 13773 garray_T *gap; /* either su_ga or su_sga */ |
| 323 | 13774 char_u *goodword; |
| 534 | 13775 int badlenarg; /* len of bad word replaced with "goodword" */ |
| 324 | 13776 int score; |
| 351 | 13777 int altscore; |
| 344 | 13778 int had_bonus; /* value for st_had_bonus */ |
| 503 | 13779 slang_T *slang; /* language for sound folding */ |
| 625 | 13780 int maxsf; /* su_maxscore applies to soundfold score, |
| 13781 su_sfmaxscore to the total score. */ | |
| 13782 { | |
| 13783 int goodlen; /* len of goodword changed */ | |
| 13784 int badlen; /* len of bad word changed */ | |
| 323 | 13785 suggest_T *stp; |
| 534 | 13786 suggest_T new_sug; |
| 323 | 13787 int i; |
| 534 | 13788 char_u *pgood, *pbad; |
| 532 | 13789 |
| 534 | 13790 /* Minimize "badlen" for consistency. Avoids that changing "the the" to |
| 13791 * "thee the" is added next to changing the first "the" the "thee". */ | |
| 13792 pgood = goodword + STRLEN(goodword); | |
| 625 | 13793 pbad = su->su_badptr + badlenarg; |
| 13794 for (;;) | |
| 13795 { | |
| 835 | 13796 goodlen = (int)(pgood - goodword); |
| 13797 badlen = (int)(pbad - su->su_badptr); | |
| 625 | 13798 if (goodlen <= 0 || badlen <= 0) |
| 13799 break; | |
| 534 | 13800 mb_ptr_back(goodword, pgood); |
| 13801 mb_ptr_back(su->su_badptr, pbad); | |
| 13802 #ifdef FEAT_MBYTE | |
| 13803 if (has_mbyte) | |
| 13804 { | |
| 13805 if (mb_ptr2char(pgood) != mb_ptr2char(pbad)) | |
| 13806 break; | |
| 346 | 13807 } |
| 13808 else | |
| 534 | 13809 #endif |
| 13810 if (*pgood != *pbad) | |
| 13811 break; | |
| 625 | 13812 } |
| 13813 | |
| 534 | 13814 if (badlen == 0 && goodlen == 0) |
| 13815 /* goodword doesn't change anything; may happen for "the the" changing | |
| 13816 * the first "the" to itself. */ | |
| 13817 return; | |
| 346 | 13818 |
| 944 | 13819 if (gap->ga_len == 0) |
| 13820 i = -1; | |
| 13821 else | |
| 13822 { | |
| 13823 /* Check if the word is already there. Also check the length that is | |
| 13824 * being replaced "thes," -> "these" is a different suggestion from | |
| 13825 * "thes" -> "these". */ | |
| 13826 stp = &SUG(*gap, 0); | |
| 13827 for (i = gap->ga_len; --i >= 0; ++stp) | |
| 13828 if (stp->st_wordlen == goodlen | |
| 13829 && stp->st_orglen == badlen | |
| 13830 && STRNCMP(stp->st_word, goodword, goodlen) == 0) | |
| 13831 { | |
| 13832 /* | |
| 13833 * Found it. Remember the word with the lowest score. | |
| 13834 */ | |
| 13835 if (stp->st_slang == NULL) | |
| 13836 stp->st_slang = slang; | |
| 13837 | |
| 13838 new_sug.st_score = score; | |
| 13839 new_sug.st_altscore = altscore; | |
| 13840 new_sug.st_had_bonus = had_bonus; | |
| 13841 | |
| 13842 if (stp->st_had_bonus != had_bonus) | |
| 13843 { | |
| 13844 /* Only one of the two had the soundalike score computed. | |
| 13845 * Need to do that for the other one now, otherwise the | |
| 13846 * scores can't be compared. This happens because | |
| 13847 * suggest_try_change() doesn't compute the soundalike | |
| 13848 * word to keep it fast, while some special methods set | |
| 13849 * the soundalike score to zero. */ | |
| 13850 if (had_bonus) | |
| 13851 rescore_one(su, stp); | |
| 13852 else | |
| 13853 { | |
| 13854 new_sug.st_word = stp->st_word; | |
| 13855 new_sug.st_wordlen = stp->st_wordlen; | |
| 13856 new_sug.st_slang = stp->st_slang; | |
| 13857 new_sug.st_orglen = badlen; | |
| 13858 rescore_one(su, &new_sug); | |
| 13859 } | |
| 13860 } | |
| 13861 | |
| 13862 if (stp->st_score > new_sug.st_score) | |
| 13863 { | |
| 13864 stp->st_score = new_sug.st_score; | |
| 13865 stp->st_altscore = new_sug.st_altscore; | |
| 13866 stp->st_had_bonus = new_sug.st_had_bonus; | |
| 13867 } | |
| 13868 break; | |
| 13869 } | |
| 13870 } | |
| 625 | 13871 |
| 13872 if (i < 0 && ga_grow(gap, 1) == OK) | |
| 13873 { | |
| 13874 /* Add a suggestion. */ | |
| 13875 stp = &SUG(*gap, gap->ga_len); | |
| 13876 stp->st_word = vim_strnsave(goodword, goodlen); | |
| 13877 if (stp->st_word != NULL) | |
| 13878 { | |
| 13879 stp->st_wordlen = goodlen; | |
| 13880 stp->st_score = score; | |
| 13881 stp->st_altscore = altscore; | |
| 13882 stp->st_had_bonus = had_bonus; | |
| 13883 stp->st_orglen = badlen; | |
| 13884 stp->st_slang = slang; | |
| 13885 ++gap->ga_len; | |
| 13886 | |
| 13887 /* If we have too many suggestions now, sort the list and keep | |
| 13888 * the best suggestions. */ | |
| 13889 if (gap->ga_len > SUG_MAX_COUNT(su)) | |
| 13890 { | |
| 13891 if (maxsf) | |
| 13892 su->su_sfmaxscore = cleanup_suggestions(gap, | |
| 13893 su->su_sfmaxscore, SUG_CLEAN_COUNT(su)); | |
| 13894 else | |
| 13895 su->su_maxscore = cleanup_suggestions(gap, | |
| 13896 su->su_maxscore, SUG_CLEAN_COUNT(su)); | |
| 13897 } | |
| 13898 } | |
| 13899 } | |
| 13900 } | |
| 13901 | |
| 13902 /* | |
| 13903 * Suggestions may in fact be flagged as errors. Esp. for banned words and | |
| 13904 * for split words, such as "the the". Remove these from the list here. | |
| 13905 */ | |
| 13906 static void | |
| 13907 check_suggestions(su, gap) | |
| 13908 suginfo_T *su; | |
| 13909 garray_T *gap; /* either su_ga or su_sga */ | |
| 13910 { | |
| 13911 suggest_T *stp; | |
| 13912 int i; | |
| 13913 char_u longword[MAXWLEN + 1]; | |
| 13914 int len; | |
| 13915 hlf_T attr; | |
| 13916 | |
| 13917 stp = &SUG(*gap, 0); | |
| 13918 for (i = gap->ga_len - 1; i >= 0; --i) | |
| 13919 { | |
| 13920 /* Need to append what follows to check for "the the". */ | |
| 2768 | 13921 vim_strncpy(longword, stp[i].st_word, MAXWLEN); |
| 625 | 13922 len = stp[i].st_wordlen; |
| 13923 vim_strncpy(longword + len, su->su_badptr + stp[i].st_orglen, | |
| 13924 MAXWLEN - len); | |
| 13925 attr = HLF_COUNT; | |
| 13926 (void)spell_check(curwin, longword, &attr, NULL, FALSE); | |
| 13927 if (attr != HLF_COUNT) | |
| 13928 { | |
| 13929 /* Remove this entry. */ | |
| 13930 vim_free(stp[i].st_word); | |
| 13931 --gap->ga_len; | |
| 13932 if (i < gap->ga_len) | |
| 13933 mch_memmove(stp + i, stp + i + 1, | |
| 13934 sizeof(suggest_T) * (gap->ga_len - i)); | |
| 13935 } | |
| 13936 } | |
| 13937 } | |
| 13938 | |
| 323 | 13939 |
| 13940 /* | |
| 13941 * Add a word to be banned. | |
| 13942 */ | |
| 13943 static void | |
| 13944 add_banned(su, word) | |
| 13945 suginfo_T *su; | |
| 13946 char_u *word; | |
| 13947 { | |
| 626 | 13948 char_u *s; |
| 323 | 13949 hash_T hash; |
| 13950 hashitem_T *hi; | |
| 13951 | |
| 625 | 13952 hash = hash_hash(word); |
| 13953 hi = hash_lookup(&su->su_banned, word, hash); | |
| 13954 if (HASHITEM_EMPTY(hi)) | |
| 13955 { | |
| 13956 s = vim_strsave(word); | |
| 13957 if (s != NULL) | |
| 323 | 13958 hash_add_item(&su->su_banned, hi, s, hash); |
| 625 | 13959 } |
| 323 | 13960 } |
| 13961 | |
| 324 | 13962 /* |
| 534 | 13963 * Recompute the score for all suggestions if sound-folding is possible. This |
| 13964 * is slow, thus only done for the final results. | |
| 324 | 13965 */ |
| 13966 static void | |
| 13967 rescore_suggestions(su) | |
| 13968 suginfo_T *su; | |
| 13969 { | |
| 13970 int i; | |
| 534 | 13971 |
| 13972 if (su->su_sallang != NULL) | |
| 503 | 13973 for (i = 0; i < su->su_ga.ga_len; ++i) |
| 534 | 13974 rescore_one(su, &SUG(su->su_ga, i)); |
| 13975 } | |
| 13976 | |
| 13977 /* | |
| 13978 * Recompute the score for one suggestion if sound-folding is possible. | |
| 13979 */ | |
| 13980 static void | |
| 13981 rescore_one(su, stp) | |
| 535 | 13982 suginfo_T *su; |
| 13983 suggest_T *stp; | |
| 534 | 13984 { |
| 13985 slang_T *slang = stp->st_slang; | |
| 13986 char_u sal_badword[MAXWLEN]; | |
| 535 | 13987 char_u *p; |
| 534 | 13988 |
| 13989 /* Only rescore suggestions that have no sal score yet and do have a | |
| 13990 * language. */ | |
| 13991 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus) | |
| 13992 { | |
| 13993 if (slang == su->su_sallang) | |
| 535 | 13994 p = su->su_sal_badword; |
| 534 | 13995 else |
| 13996 { | |
| 13997 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword); | |
| 535 | 13998 p = sal_badword; |
| 13999 } | |
| 14000 | |
| 14001 stp->st_altscore = stp_sal_score(stp, su, slang, p); | |
| 534 | 14002 if (stp->st_altscore == SCORE_MAXMAX) |
| 14003 stp->st_altscore = SCORE_BIG; | |
| 14004 stp->st_score = RESCORE(stp->st_score, stp->st_altscore); | |
| 14005 stp->st_had_bonus = TRUE; | |
| 324 | 14006 } |
| 14007 } | |
| 14008 | |
| 323 | 14009 static int |
| 14010 #ifdef __BORLANDC__ | |
| 14011 _RTLENTRYF | |
| 14012 #endif | |
| 14013 sug_compare __ARGS((const void *s1, const void *s2)); | |
| 14014 | |
| 14015 /* | |
| 14016 * Function given to qsort() to sort the suggestions on st_score. | |
| 529 | 14017 * First on "st_score", then "st_altscore" then alphabetically. |
| 323 | 14018 */ |
| 14019 static int | |
| 14020 #ifdef __BORLANDC__ | |
| 14021 _RTLENTRYF | |
| 14022 #endif | |
| 14023 sug_compare(s1, s2) | |
| 14024 const void *s1; | |
| 14025 const void *s2; | |
| 14026 { | |
| 14027 suggest_T *p1 = (suggest_T *)s1; | |
| 14028 suggest_T *p2 = (suggest_T *)s2; | |
| 344 | 14029 int n = p1->st_score - p2->st_score; |
| 14030 | |
| 14031 if (n == 0) | |
| 529 | 14032 { |
| 14033 n = p1->st_altscore - p2->st_altscore; | |
| 14034 if (n == 0) | |
| 14035 n = STRICMP(p1->st_word, p2->st_word); | |
| 14036 } | |
| 344 | 14037 return n; |
| 323 | 14038 } |
| 14039 | |
| 14040 /* | |
| 14041 * Cleanup the suggestions: | |
| 14042 * - Sort on score. | |
| 14043 * - Remove words that won't be displayed. | |
| 344 | 14044 * Returns the maximum score in the list or "maxscore" unmodified. |
| 323 | 14045 */ |
| 344 | 14046 static int |
| 14047 cleanup_suggestions(gap, maxscore, keep) | |
| 14048 garray_T *gap; | |
| 14049 int maxscore; | |
| 324 | 14050 int keep; /* nr of suggestions to keep */ |
| 323 | 14051 { |
| 344 | 14052 suggest_T *stp = &SUG(*gap, 0); |
| 323 | 14053 int i; |
| 14054 | |
| 14055 /* Sort the list. */ | |
| 344 | 14056 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare); |
| 323 | 14057 |
| 14058 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 344 | 14059 if (gap->ga_len > keep) |
| 323 | 14060 { |
| 344 | 14061 for (i = keep; i < gap->ga_len; ++i) |
| 323 | 14062 vim_free(stp[i].st_word); |
| 344 | 14063 gap->ga_len = keep; |
| 14064 return stp[keep - 1].st_score; | |
| 323 | 14065 } |
| 344 | 14066 return maxscore; |
| 323 | 14067 } |
| 14068 | |
| 372 | 14069 #if defined(FEAT_EVAL) || defined(PROTO) |
| 14070 /* | |
| 14071 * Soundfold a string, for soundfold(). | |
| 14072 * Result is in allocated memory, NULL for an error. | |
| 14073 */ | |
| 14074 char_u * | |
| 14075 eval_soundfold(word) | |
| 14076 char_u *word; | |
| 14077 { | |
| 14078 langp_T *lp; | |
| 14079 char_u sound[MAXWLEN]; | |
| 500 | 14080 int lpi; |
| 372 | 14081 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14082 if (curwin->w_p_spell && *curwin->w_s->b_p_spl != NUL) |
| 372 | 14083 /* 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
|
14084 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
|
14085 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14086 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 372 | 14087 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 14088 { | |
| 14089 /* soundfold the word */ | |
| 375 | 14090 spell_soundfold(lp->lp_slang, word, FALSE, sound); |
| 372 | 14091 return vim_strsave(sound); |
| 14092 } | |
| 500 | 14093 } |
| 372 | 14094 |
| 14095 /* No language with sound folding, return word as-is. */ | |
| 14096 return vim_strsave(word); | |
| 14097 } | |
| 14098 #endif | |
| 14099 | |
| 323 | 14100 /* |
| 14101 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 485 | 14102 * |
| 14103 * There are many ways to turn a word into a sound-a-like representation. The | |
| 14104 * oldest is Soundex (1918!). A nice overview can be found in "Approximate | |
| 14105 * swedish name matching - survey and test of different algorithms" by Klas | |
| 14106 * Erikson. | |
| 14107 * | |
| 14108 * We support two methods: | |
| 14109 * 1. SOFOFROM/SOFOTO do a simple character mapping. | |
| 14110 * 2. SAL items define a more advanced sound-folding (and much slower). | |
| 323 | 14111 */ |
| 14112 static void | |
| 375 | 14113 spell_soundfold(slang, inword, folded, res) |
| 14114 slang_T *slang; | |
| 14115 char_u *inword; | |
| 14116 int folded; /* "inword" is already case-folded */ | |
| 14117 char_u *res; | |
| 14118 { | |
| 14119 char_u fword[MAXWLEN]; | |
| 14120 char_u *word; | |
| 14121 | |
| 14122 if (slang->sl_sofo) | |
| 14123 /* SOFOFROM and SOFOTO used */ | |
| 14124 spell_soundfold_sofo(slang, inword, res); | |
| 14125 else | |
| 14126 { | |
| 14127 /* SAL items used. Requires the word to be case-folded. */ | |
| 14128 if (folded) | |
| 14129 word = inword; | |
| 14130 else | |
| 14131 { | |
| 835 | 14132 (void)spell_casefold(inword, (int)STRLEN(inword), fword, MAXWLEN); |
| 375 | 14133 word = fword; |
| 14134 } | |
| 14135 | |
| 14136 #ifdef FEAT_MBYTE | |
| 14137 if (has_mbyte) | |
| 14138 spell_soundfold_wsal(slang, word, res); | |
| 14139 else | |
| 14140 #endif | |
| 14141 spell_soundfold_sal(slang, word, res); | |
| 14142 } | |
| 14143 } | |
| 14144 | |
| 14145 /* | |
| 14146 * Perform sound folding of "inword" into "res" according to SOFOFROM and | |
| 14147 * SOFOTO lines. | |
| 14148 */ | |
| 14149 static void | |
| 14150 spell_soundfold_sofo(slang, inword, res) | |
| 14151 slang_T *slang; | |
| 14152 char_u *inword; | |
| 14153 char_u *res; | |
| 14154 { | |
| 14155 char_u *s; | |
| 14156 int ri = 0; | |
| 14157 int c; | |
| 14158 | |
| 14159 #ifdef FEAT_MBYTE | |
| 14160 if (has_mbyte) | |
| 14161 { | |
| 14162 int prevc = 0; | |
| 14163 int *ip; | |
| 14164 | |
| 14165 /* The sl_sal_first[] table contains the translation for chars up to | |
| 14166 * 255, sl_sal the rest. */ | |
| 14167 for (s = inword; *s != NUL; ) | |
| 14168 { | |
| 474 | 14169 c = mb_cptr2char_adv(&s); |
| 375 | 14170 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c)) |
| 14171 c = ' '; | |
| 14172 else if (c < 256) | |
| 14173 c = slang->sl_sal_first[c]; | |
| 14174 else | |
| 14175 { | |
| 14176 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff]; | |
| 14177 if (ip == NULL) /* empty list, can't match */ | |
| 14178 c = NUL; | |
| 14179 else | |
| 14180 for (;;) /* find "c" in the list */ | |
| 14181 { | |
| 14182 if (*ip == 0) /* not found */ | |
| 14183 { | |
| 14184 c = NUL; | |
| 14185 break; | |
| 14186 } | |
| 14187 if (*ip == c) /* match! */ | |
| 14188 { | |
| 14189 c = ip[1]; | |
| 14190 break; | |
| 14191 } | |
| 14192 ip += 2; | |
| 14193 } | |
| 14194 } | |
| 14195 | |
| 14196 if (c != NUL && c != prevc) | |
| 14197 { | |
| 14198 ri += mb_char2bytes(c, res + ri); | |
| 14199 if (ri + MB_MAXBYTES > MAXWLEN) | |
| 14200 break; | |
| 14201 prevc = c; | |
| 14202 } | |
| 14203 } | |
| 14204 } | |
| 14205 else | |
| 14206 #endif | |
| 14207 { | |
| 14208 /* The sl_sal_first[] table contains the translation. */ | |
| 14209 for (s = inword; (c = *s) != NUL; ++s) | |
| 14210 { | |
| 14211 if (vim_iswhite(c)) | |
| 14212 c = ' '; | |
| 14213 else | |
| 14214 c = slang->sl_sal_first[c]; | |
| 14215 if (c != NUL && (ri == 0 || res[ri - 1] != c)) | |
| 14216 res[ri++] = c; | |
| 14217 } | |
| 14218 } | |
| 14219 | |
| 14220 res[ri] = NUL; | |
| 14221 } | |
| 14222 | |
| 14223 static void | |
| 14224 spell_soundfold_sal(slang, inword, res) | |
| 323 | 14225 slang_T *slang; |
| 14226 char_u *inword; | |
| 14227 char_u *res; | |
| 14228 { | |
| 344 | 14229 salitem_T *smp; |
| 323 | 14230 char_u word[MAXWLEN]; |
| 375 | 14231 char_u *s = inword; |
| 323 | 14232 char_u *t; |
| 344 | 14233 char_u *pf; |
| 323 | 14234 int i, j, z; |
| 344 | 14235 int reslen; |
| 323 | 14236 int n, k = 0; |
| 14237 int z0; | |
| 14238 int k0; | |
| 14239 int n0; | |
| 14240 int c; | |
| 14241 int pri; | |
| 14242 int p0 = -333; | |
| 14243 int c0; | |
| 14244 | |
| 324 | 14245 /* Remove accents, if wanted. We actually remove all non-word characters. |
| 375 | 14246 * But keep white space. We need a copy, the word may be changed here. */ |
| 323 | 14247 if (slang->sl_rem_accents) |
| 14248 { | |
| 14249 t = word; | |
| 375 | 14250 while (*s != NUL) |
| 323 | 14251 { |
| 324 | 14252 if (vim_iswhite(*s)) |
| 344 | 14253 { |
| 14254 *t++ = ' '; | |
| 14255 s = skipwhite(s); | |
| 14256 } | |
| 323 | 14257 else |
| 14258 { | |
| 5477 | 14259 if (spell_iswordp_nmw(s, curwin)) |
| 323 | 14260 *t++ = *s; |
| 14261 ++s; | |
| 14262 } | |
| 14263 } | |
| 14264 *t = NUL; | |
| 14265 } | |
| 14266 else | |
| 2768 | 14267 vim_strncpy(word, s, MAXWLEN - 1); |
| 323 | 14268 |
| 344 | 14269 smp = (salitem_T *)slang->sl_sal.ga_data; |
| 323 | 14270 |
| 14271 /* | |
| 14272 * This comes from Aspell phonet.cpp. Converted from C++ to C. | |
| 324 | 14273 * Changed to keep spaces. |
| 323 | 14274 */ |
| 344 | 14275 i = reslen = z = 0; |
| 323 | 14276 while ((c = word[i]) != NUL) |
| 14277 { | |
| 344 | 14278 /* Start with the first rule that has the character in the word. */ |
| 323 | 14279 n = slang->sl_sal_first[c]; |
| 14280 z0 = 0; | |
| 14281 | |
| 14282 if (n >= 0) | |
| 14283 { | |
| 14284 /* check all rules for the same letter */ | |
| 344 | 14285 for (; (s = smp[n].sm_lead)[0] == c; ++n) |
| 323 | 14286 { |
| 344 | 14287 /* Quickly skip entries that don't match the word. Most |
| 14288 * entries are less then three chars, optimize for that. */ | |
| 14289 k = smp[n].sm_leadlen; | |
| 14290 if (k > 1) | |
| 323 | 14291 { |
| 344 | 14292 if (word[i + 1] != s[1]) |
| 14293 continue; | |
| 14294 if (k > 2) | |
| 14295 { | |
| 14296 for (j = 2; j < k; ++j) | |
| 14297 if (word[i + j] != s[j]) | |
| 14298 break; | |
| 14299 if (j < k) | |
| 14300 continue; | |
| 14301 } | |
| 323 | 14302 } |
| 14303 | |
| 375 | 14304 if ((pf = smp[n].sm_oneof) != NULL) |
| 323 | 14305 { |
| 375 | 14306 /* Check for match with one of the chars in "sm_oneof". */ |
| 344 | 14307 while (*pf != NUL && *pf != word[i + k]) |
| 14308 ++pf; | |
| 14309 if (*pf == NUL) | |
| 14310 continue; | |
| 14311 ++k; | |
| 323 | 14312 } |
| 344 | 14313 s = smp[n].sm_rules; |
| 14314 pri = 5; /* default priority */ | |
| 323 | 14315 |
| 14316 p0 = *s; | |
| 14317 k0 = k; | |
| 14318 while (*s == '-' && k > 1) | |
| 14319 { | |
| 14320 k--; | |
| 14321 s++; | |
| 14322 } | |
| 14323 if (*s == '<') | |
| 14324 s++; | |
| 344 | 14325 if (VIM_ISDIGIT(*s)) |
| 323 | 14326 { |
| 14327 /* determine priority */ | |
| 14328 pri = *s - '0'; | |
| 14329 s++; | |
| 14330 } | |
| 14331 if (*s == '^' && *(s + 1) == '^') | |
| 14332 s++; | |
| 14333 | |
| 14334 if (*s == NUL | |
| 14335 || (*s == '^' | |
| 324 | 14336 && (i == 0 || !(word[i - 1] == ' ' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14337 || spell_iswordp(word + i - 1, curwin))) |
| 323 | 14338 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14339 || (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 14340 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14341 && spell_iswordp(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14342 && (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 14343 { |
| 14344 /* search for followup rules, if: */ | |
| 14345 /* followup and k > 1 and NO '-' in searchstring */ | |
| 14346 c0 = word[i + k - 1]; | |
| 14347 n0 = slang->sl_sal_first[c0]; | |
| 14348 | |
| 14349 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 344 | 14350 && p0 != '-' && word[i + k] != NUL) |
| 323 | 14351 { |
| 14352 /* test follow-up rule for "word[i + k]" */ | |
| 344 | 14353 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0) |
| 323 | 14354 { |
| 344 | 14355 /* Quickly skip entries that don't match the word. |
| 14356 * */ | |
| 14357 k0 = smp[n0].sm_leadlen; | |
| 14358 if (k0 > 1) | |
| 323 | 14359 { |
| 344 | 14360 if (word[i + k] != s[1]) |
| 14361 continue; | |
| 14362 if (k0 > 2) | |
| 14363 { | |
| 14364 pf = word + i + k + 1; | |
| 14365 for (j = 2; j < k0; ++j) | |
| 14366 if (*pf++ != s[j]) | |
| 14367 break; | |
| 14368 if (j < k0) | |
| 14369 continue; | |
| 14370 } | |
| 323 | 14371 } |
| 344 | 14372 k0 += k - 1; |
| 14373 | |
| 375 | 14374 if ((pf = smp[n0].sm_oneof) != NULL) |
| 323 | 14375 { |
| 344 | 14376 /* Check for match with one of the chars in |
| 375 | 14377 * "sm_oneof". */ |
| 344 | 14378 while (*pf != NUL && *pf != word[i + k0]) |
| 14379 ++pf; | |
| 14380 if (*pf == NUL) | |
| 14381 continue; | |
| 14382 ++k0; | |
| 323 | 14383 } |
| 344 | 14384 |
| 14385 p0 = 5; | |
| 14386 s = smp[n0].sm_rules; | |
| 323 | 14387 while (*s == '-') |
| 14388 { | |
| 344 | 14389 /* "k0" gets NOT reduced because |
| 14390 * "if (k0 == k)" */ | |
| 323 | 14391 s++; |
| 14392 } | |
| 14393 if (*s == '<') | |
| 14394 s++; | |
| 344 | 14395 if (VIM_ISDIGIT(*s)) |
| 323 | 14396 { |
| 14397 p0 = *s - '0'; | |
| 14398 s++; | |
| 14399 } | |
| 14400 | |
| 14401 if (*s == NUL | |
| 14402 /* *s == '^' cuts */ | |
| 14403 || (*s == '$' | |
| 376 | 14404 && !spell_iswordp(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14405 curwin))) |
| 323 | 14406 { |
| 14407 if (k0 == k) | |
| 14408 /* this is just a piece of the string */ | |
| 14409 continue; | |
| 14410 | |
| 14411 if (p0 < pri) | |
| 14412 /* priority too low */ | |
| 14413 continue; | |
| 14414 /* rule fits; stop search */ | |
| 14415 break; | |
| 14416 } | |
| 14417 } | |
| 14418 | |
| 344 | 14419 if (p0 >= pri && smp[n0].sm_lead[0] == c0) |
| 323 | 14420 continue; |
| 14421 } | |
| 14422 | |
| 14423 /* replace string */ | |
| 344 | 14424 s = smp[n].sm_to; |
| 389 | 14425 if (s == NULL) |
| 14426 s = (char_u *)""; | |
| 344 | 14427 pf = smp[n].sm_rules; |
| 14428 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0; | |
| 323 | 14429 if (p0 == 1 && z == 0) |
| 14430 { | |
| 14431 /* rule with '<' is used */ | |
| 344 | 14432 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c |
| 14433 || res[reslen - 1] == *s)) | |
| 14434 reslen--; | |
| 323 | 14435 z0 = 1; |
| 14436 z = 1; | |
| 14437 k0 = 0; | |
| 372 | 14438 while (*s != NUL && word[i + k0] != NUL) |
| 323 | 14439 { |
| 14440 word[i + k0] = *s; | |
| 14441 k0++; | |
| 14442 s++; | |
| 14443 } | |
| 14444 if (k > k0) | |
| 1619 | 14445 STRMOVE(word + i + k0, word + i + k); |
| 323 | 14446 |
| 14447 /* new "actual letter" */ | |
| 14448 c = word[i]; | |
| 14449 } | |
| 14450 else | |
| 14451 { | |
| 14452 /* no '<' rule used */ | |
| 14453 i += k - 1; | |
| 14454 z = 0; | |
| 344 | 14455 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN) |
| 323 | 14456 { |
| 344 | 14457 if (reslen == 0 || res[reslen - 1] != *s) |
| 372 | 14458 res[reslen++] = *s; |
| 323 | 14459 s++; |
| 14460 } | |
| 14461 /* new "actual letter" */ | |
| 14462 c = *s; | |
| 344 | 14463 if (strstr((char *)pf, "^^") != NULL) |
| 323 | 14464 { |
| 14465 if (c != NUL) | |
| 372 | 14466 res[reslen++] = c; |
| 1619 | 14467 STRMOVE(word, word + i + 1); |
| 323 | 14468 i = 0; |
| 14469 z0 = 1; | |
| 14470 } | |
| 14471 } | |
| 14472 break; | |
| 14473 } | |
| 14474 } | |
| 14475 } | |
| 324 | 14476 else if (vim_iswhite(c)) |
| 14477 { | |
| 14478 c = ' '; | |
| 14479 k = 1; | |
| 14480 } | |
| 323 | 14481 |
| 14482 if (z0 == 0) | |
| 14483 { | |
| 344 | 14484 if (k && !p0 && reslen < MAXWLEN && c != NUL |
| 14485 && (!slang->sl_collapse || reslen == 0 | |
| 14486 || res[reslen - 1] != c)) | |
| 323 | 14487 /* condense only double letters */ |
| 372 | 14488 res[reslen++] = c; |
| 323 | 14489 |
| 14490 i++; | |
| 14491 z = 0; | |
| 14492 k = 0; | |
| 14493 } | |
| 14494 } | |
| 14495 | |
| 344 | 14496 res[reslen] = NUL; |
| 323 | 14497 } |
| 14498 | |
| 372 | 14499 #ifdef FEAT_MBYTE |
| 14500 /* | |
| 14501 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 14502 * Multi-byte version of spell_soundfold(). | |
| 14503 */ | |
| 14504 static void | |
| 375 | 14505 spell_soundfold_wsal(slang, inword, res) |
| 372 | 14506 slang_T *slang; |
| 14507 char_u *inword; | |
| 14508 char_u *res; | |
| 14509 { | |
| 375 | 14510 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data; |
| 372 | 14511 int word[MAXWLEN]; |
| 14512 int wres[MAXWLEN]; | |
| 14513 int l; | |
| 14514 char_u *s; | |
| 14515 int *ws; | |
| 14516 char_u *t; | |
| 14517 int *pf; | |
| 14518 int i, j, z; | |
| 14519 int reslen; | |
| 14520 int n, k = 0; | |
| 14521 int z0; | |
| 14522 int k0; | |
| 14523 int n0; | |
| 14524 int c; | |
| 14525 int pri; | |
| 14526 int p0 = -333; | |
| 14527 int c0; | |
| 14528 int did_white = FALSE; | |
| 3520 | 14529 int wordlen; |
| 14530 | |
| 372 | 14531 |
| 14532 /* | |
| 14533 * Convert the multi-byte string to a wide-character string. | |
| 14534 * Remove accents, if wanted. We actually remove all non-word characters. | |
| 14535 * But keep white space. | |
| 14536 */ | |
| 3520 | 14537 wordlen = 0; |
| 372 | 14538 for (s = inword; *s != NUL; ) |
| 14539 { | |
| 14540 t = s; | |
| 474 | 14541 c = mb_cptr2char_adv(&s); |
| 372 | 14542 if (slang->sl_rem_accents) |
| 14543 { | |
| 14544 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c)) | |
| 14545 { | |
| 14546 if (did_white) | |
| 14547 continue; | |
| 14548 c = ' '; | |
| 14549 did_white = TRUE; | |
| 14550 } | |
| 14551 else | |
| 14552 { | |
| 14553 did_white = FALSE; | |
| 5477 | 14554 if (!spell_iswordp_nmw(t, curwin)) |
| 372 | 14555 continue; |
| 14556 } | |
| 14557 } | |
| 3520 | 14558 word[wordlen++] = c; |
| 14559 } | |
| 14560 word[wordlen] = NUL; | |
| 372 | 14561 |
| 14562 /* | |
| 3520 | 14563 * This algorithm comes from Aspell phonet.cpp. |
| 372 | 14564 * Converted from C++ to C. Added support for multi-byte chars. |
| 14565 * Changed to keep spaces. | |
| 14566 */ | |
| 14567 i = reslen = z = 0; | |
| 14568 while ((c = word[i]) != NUL) | |
| 14569 { | |
| 14570 /* Start with the first rule that has the character in the word. */ | |
| 14571 n = slang->sl_sal_first[c & 0xff]; | |
| 14572 z0 = 0; | |
| 14573 | |
| 14574 if (n >= 0) | |
| 14575 { | |
|
2455
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
14576 /* 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
|
14577 * 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
|
14578 * (c & 0xff) is NUL. */ |
|
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
14579 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
|
14580 && ws[0] != NUL; ++n) |
| 372 | 14581 { |
| 14582 /* Quickly skip entries that don't match the word. Most | |
| 14583 * entries are less then three chars, optimize for that. */ | |
| 375 | 14584 if (c != ws[0]) |
| 14585 continue; | |
| 372 | 14586 k = smp[n].sm_leadlen; |
| 14587 if (k > 1) | |
| 14588 { | |
| 14589 if (word[i + 1] != ws[1]) | |
| 14590 continue; | |
| 14591 if (k > 2) | |
| 14592 { | |
| 14593 for (j = 2; j < k; ++j) | |
| 14594 if (word[i + j] != ws[j]) | |
| 14595 break; | |
| 14596 if (j < k) | |
| 14597 continue; | |
| 14598 } | |
| 14599 } | |
| 14600 | |
| 375 | 14601 if ((pf = smp[n].sm_oneof_w) != NULL) |
| 372 | 14602 { |
| 375 | 14603 /* Check for match with one of the chars in "sm_oneof". */ |
| 372 | 14604 while (*pf != NUL && *pf != word[i + k]) |
| 14605 ++pf; | |
| 14606 if (*pf == NUL) | |
| 14607 continue; | |
| 14608 ++k; | |
| 14609 } | |
| 14610 s = smp[n].sm_rules; | |
| 14611 pri = 5; /* default priority */ | |
| 14612 | |
| 14613 p0 = *s; | |
| 14614 k0 = k; | |
| 14615 while (*s == '-' && k > 1) | |
| 14616 { | |
| 14617 k--; | |
| 14618 s++; | |
| 14619 } | |
| 14620 if (*s == '<') | |
| 14621 s++; | |
| 14622 if (VIM_ISDIGIT(*s)) | |
| 14623 { | |
| 14624 /* determine priority */ | |
| 14625 pri = *s - '0'; | |
| 14626 s++; | |
| 14627 } | |
| 14628 if (*s == '^' && *(s + 1) == '^') | |
| 14629 s++; | |
| 14630 | |
| 14631 if (*s == NUL | |
| 14632 || (*s == '^' | |
| 14633 && (i == 0 || !(word[i - 1] == ' ' | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14634 || spell_iswordp_w(word + i - 1, curwin))) |
| 372 | 14635 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14636 || (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 14637 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14638 && spell_iswordp_w(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14639 && (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 14640 { |
| 14641 /* search for followup rules, if: */ | |
| 14642 /* followup and k > 1 and NO '-' in searchstring */ | |
| 14643 c0 = word[i + k - 1]; | |
| 14644 n0 = slang->sl_sal_first[c0 & 0xff]; | |
| 14645 | |
| 14646 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 14647 && p0 != '-' && word[i + k] != NUL) | |
| 14648 { | |
| 375 | 14649 /* Test follow-up rule for "word[i + k]"; loop over |
| 14650 * all entries with the same index byte. */ | |
| 372 | 14651 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff) |
| 14652 == (c0 & 0xff); ++n0) | |
| 14653 { | |
| 14654 /* Quickly skip entries that don't match the word. | |
| 375 | 14655 */ |
| 14656 if (c0 != ws[0]) | |
| 14657 continue; | |
| 372 | 14658 k0 = smp[n0].sm_leadlen; |
| 14659 if (k0 > 1) | |
| 14660 { | |
| 14661 if (word[i + k] != ws[1]) | |
| 14662 continue; | |
| 14663 if (k0 > 2) | |
| 14664 { | |
| 14665 pf = word + i + k + 1; | |
| 14666 for (j = 2; j < k0; ++j) | |
| 14667 if (*pf++ != ws[j]) | |
| 14668 break; | |
| 14669 if (j < k0) | |
| 14670 continue; | |
| 14671 } | |
| 14672 } | |
| 14673 k0 += k - 1; | |
| 14674 | |
| 375 | 14675 if ((pf = smp[n0].sm_oneof_w) != NULL) |
| 372 | 14676 { |
| 14677 /* Check for match with one of the chars in | |
| 375 | 14678 * "sm_oneof". */ |
| 372 | 14679 while (*pf != NUL && *pf != word[i + k0]) |
| 14680 ++pf; | |
| 14681 if (*pf == NUL) | |
| 14682 continue; | |
| 14683 ++k0; | |
| 14684 } | |
| 14685 | |
| 14686 p0 = 5; | |
| 14687 s = smp[n0].sm_rules; | |
| 14688 while (*s == '-') | |
| 14689 { | |
| 14690 /* "k0" gets NOT reduced because | |
| 14691 * "if (k0 == k)" */ | |
| 14692 s++; | |
| 14693 } | |
| 14694 if (*s == '<') | |
| 14695 s++; | |
| 14696 if (VIM_ISDIGIT(*s)) | |
| 14697 { | |
| 14698 p0 = *s - '0'; | |
| 14699 s++; | |
| 14700 } | |
| 14701 | |
| 14702 if (*s == NUL | |
| 14703 /* *s == '^' cuts */ | |
| 14704 || (*s == '$' | |
| 376 | 14705 && !spell_iswordp_w(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
14706 curwin))) |
| 372 | 14707 { |
| 14708 if (k0 == k) | |
| 14709 /* this is just a piece of the string */ | |
| 14710 continue; | |
| 14711 | |
| 14712 if (p0 < pri) | |
| 14713 /* priority too low */ | |
| 14714 continue; | |
| 14715 /* rule fits; stop search */ | |
| 14716 break; | |
| 14717 } | |
| 14718 } | |
| 14719 | |
| 14720 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff) | |
| 14721 == (c0 & 0xff)) | |
| 14722 continue; | |
| 14723 } | |
| 14724 | |
| 14725 /* replace string */ | |
| 14726 ws = smp[n].sm_to_w; | |
| 14727 s = smp[n].sm_rules; | |
| 14728 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0; | |
| 14729 if (p0 == 1 && z == 0) | |
| 14730 { | |
| 14731 /* rule with '<' is used */ | |
| 389 | 14732 if (reslen > 0 && ws != NULL && *ws != NUL |
| 14733 && (wres[reslen - 1] == c | |
| 372 | 14734 || wres[reslen - 1] == *ws)) |
| 14735 reslen--; | |
| 14736 z0 = 1; | |
| 14737 z = 1; | |
| 14738 k0 = 0; | |
| 389 | 14739 if (ws != NULL) |
| 14740 while (*ws != NUL && word[i + k0] != NUL) | |
| 14741 { | |
| 14742 word[i + k0] = *ws; | |
| 14743 k0++; | |
| 14744 ws++; | |
| 14745 } | |
| 372 | 14746 if (k > k0) |
| 14747 mch_memmove(word + i + k0, word + i + k, | |
| 3520 | 14748 sizeof(int) * (wordlen - (i + k) + 1)); |
| 372 | 14749 |
| 14750 /* new "actual letter" */ | |
| 14751 c = word[i]; | |
| 14752 } | |
| 14753 else | |
| 14754 { | |
| 14755 /* no '<' rule used */ | |
| 14756 i += k - 1; | |
| 14757 z = 0; | |
| 389 | 14758 if (ws != NULL) |
| 14759 while (*ws != NUL && ws[1] != NUL | |
| 14760 && reslen < MAXWLEN) | |
| 14761 { | |
| 14762 if (reslen == 0 || wres[reslen - 1] != *ws) | |
| 14763 wres[reslen++] = *ws; | |
| 14764 ws++; | |
| 14765 } | |
| 372 | 14766 /* new "actual letter" */ |
| 389 | 14767 if (ws == NULL) |
| 14768 c = NUL; | |
| 14769 else | |
| 14770 c = *ws; | |
| 372 | 14771 if (strstr((char *)s, "^^") != NULL) |
| 14772 { | |
| 14773 if (c != NUL) | |
| 14774 wres[reslen++] = c; | |
| 14775 mch_memmove(word, word + i + 1, | |
| 3520 | 14776 sizeof(int) * (wordlen - (i + 1) + 1)); |
| 372 | 14777 i = 0; |
| 14778 z0 = 1; | |
| 14779 } | |
| 14780 } | |
| 14781 break; | |
| 14782 } | |
| 14783 } | |
| 14784 } | |
| 14785 else if (vim_iswhite(c)) | |
| 14786 { | |
| 14787 c = ' '; | |
| 14788 k = 1; | |
| 14789 } | |
| 14790 | |
| 14791 if (z0 == 0) | |
| 14792 { | |
| 14793 if (k && !p0 && reslen < MAXWLEN && c != NUL | |
| 14794 && (!slang->sl_collapse || reslen == 0 | |
| 14795 || wres[reslen - 1] != c)) | |
| 14796 /* condense only double letters */ | |
| 14797 wres[reslen++] = c; | |
| 14798 | |
| 14799 i++; | |
| 14800 z = 0; | |
| 14801 k = 0; | |
| 14802 } | |
| 14803 } | |
| 14804 | |
| 14805 /* Convert wide characters in "wres" to a multi-byte string in "res". */ | |
| 14806 l = 0; | |
| 14807 for (n = 0; n < reslen; ++n) | |
| 14808 { | |
| 14809 l += mb_char2bytes(wres[n], res + l); | |
| 14810 if (l + MB_MAXBYTES > MAXWLEN) | |
| 14811 break; | |
| 14812 } | |
| 14813 res[l] = NUL; | |
| 14814 } | |
| 14815 #endif | |
| 14816 | |
| 324 | 14817 /* |
| 344 | 14818 * Compute a score for two sound-a-like words. |
| 14819 * This permits up to two inserts/deletes/swaps/etc. to keep things fast. | |
| 14820 * Instead of a generic loop we write out the code. That keeps it fast by | |
| 14821 * avoiding checks that will not be possible. | |
| 14822 */ | |
| 14823 static int | |
| 351 | 14824 soundalike_score(goodstart, badstart) |
| 14825 char_u *goodstart; /* sound-folded good word */ | |
| 14826 char_u *badstart; /* sound-folded bad word */ | |
| 344 | 14827 { |
| 351 | 14828 char_u *goodsound = goodstart; |
| 14829 char_u *badsound = badstart; | |
| 14830 int goodlen; | |
| 14831 int badlen; | |
| 344 | 14832 int n; |
| 14833 char_u *pl, *ps; | |
| 14834 char_u *pl2, *ps2; | |
| 351 | 14835 int score = 0; |
| 14836 | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14837 /* Adding/inserting "*" at the start (word starts with vowel) shouldn't be |
| 351 | 14838 * counted so much, vowels halfway the word aren't counted at all. */ |
| 14839 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound) | |
| 14840 { | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14841 if ((badsound[0] == NUL && goodsound[1] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14842 || (goodsound[0] == NUL && badsound[1] == NUL)) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14843 /* 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
|
14844 return SCORE_DEL; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14845 if (badsound[0] == NUL || goodsound[0] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14846 /* more than two changes */ |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14847 return SCORE_MAXMAX; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
14848 |
| 625 | 14849 if (badsound[1] == goodsound[1] |
| 14850 || (badsound[1] != NUL | |
| 14851 && goodsound[1] != NUL | |
| 14852 && badsound[2] == goodsound[2])) | |
| 14853 { | |
| 14854 /* handle like a substitute */ | |
| 14855 } | |
| 351 | 14856 else |
| 625 | 14857 { |
| 14858 score = 2 * SCORE_DEL / 3; | |
| 14859 if (*badsound == '*') | |
| 14860 ++badsound; | |
| 14861 else | |
| 14862 ++goodsound; | |
| 14863 } | |
| 351 | 14864 } |
| 14865 | |
| 835 | 14866 goodlen = (int)STRLEN(goodsound); |
| 14867 badlen = (int)STRLEN(badsound); | |
| 344 | 14868 |
| 1188 | 14869 /* Return quickly if the lengths are too different to be fixed by two |
| 344 | 14870 * changes. */ |
| 14871 n = goodlen - badlen; | |
| 14872 if (n < -2 || n > 2) | |
| 14873 return SCORE_MAXMAX; | |
| 14874 | |
| 14875 if (n > 0) | |
| 14876 { | |
| 351 | 14877 pl = goodsound; /* goodsound is longest */ |
| 344 | 14878 ps = badsound; |
| 14879 } | |
| 14880 else | |
| 14881 { | |
| 351 | 14882 pl = badsound; /* badsound is longest */ |
| 344 | 14883 ps = goodsound; |
| 14884 } | |
| 14885 | |
| 14886 /* Skip over the identical part. */ | |
| 14887 while (*pl == *ps && *pl != NUL) | |
| 14888 { | |
| 14889 ++pl; | |
| 14890 ++ps; | |
| 14891 } | |
| 14892 | |
| 14893 switch (n) | |
| 14894 { | |
| 14895 case -2: | |
| 14896 case 2: | |
| 14897 /* | |
| 14898 * Must delete two characters from "pl". | |
| 14899 */ | |
| 14900 ++pl; /* first delete */ | |
| 14901 while (*pl == *ps) | |
| 14902 { | |
| 14903 ++pl; | |
| 14904 ++ps; | |
| 14905 } | |
| 14906 /* strings must be equal after second delete */ | |
| 14907 if (STRCMP(pl + 1, ps) == 0) | |
| 351 | 14908 return score + SCORE_DEL * 2; |
| 344 | 14909 |
| 14910 /* Failed to compare. */ | |
| 14911 break; | |
| 14912 | |
| 14913 case -1: | |
| 14914 case 1: | |
| 14915 /* | |
| 14916 * Minimal one delete from "pl" required. | |
| 14917 */ | |
| 14918 | |
| 14919 /* 1: delete */ | |
| 14920 pl2 = pl + 1; | |
| 14921 ps2 = ps; | |
| 14922 while (*pl2 == *ps2) | |
| 14923 { | |
| 14924 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 14925 return score + SCORE_DEL; |
| 344 | 14926 ++pl2; |
| 14927 ++ps2; | |
| 14928 } | |
| 14929 | |
| 14930 /* 2: delete then swap, then rest must be equal */ | |
| 14931 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 14932 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 14933 return score + SCORE_DEL + SCORE_SWAP; |
| 344 | 14934 |
| 14935 /* 3: delete then substitute, then the rest must be equal */ | |
| 14936 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 14937 return score + SCORE_DEL + SCORE_SUBST; |
| 344 | 14938 |
| 14939 /* 4: first swap then delete */ | |
| 14940 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 14941 { | |
| 14942 pl2 = pl + 2; /* swap, skip two chars */ | |
| 14943 ps2 = ps + 2; | |
| 14944 while (*pl2 == *ps2) | |
| 14945 { | |
| 14946 ++pl2; | |
| 14947 ++ps2; | |
| 14948 } | |
| 14949 /* delete a char and then strings must be equal */ | |
| 14950 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 14951 return score + SCORE_SWAP + SCORE_DEL; |
| 344 | 14952 } |
| 14953 | |
| 14954 /* 5: first substitute then delete */ | |
| 14955 pl2 = pl + 1; /* substitute, skip one char */ | |
| 14956 ps2 = ps + 1; | |
| 14957 while (*pl2 == *ps2) | |
| 14958 { | |
| 14959 ++pl2; | |
| 14960 ++ps2; | |
| 14961 } | |
| 14962 /* delete a char and then strings must be equal */ | |
| 14963 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 14964 return score + SCORE_SUBST + SCORE_DEL; |
| 344 | 14965 |
| 14966 /* Failed to compare. */ | |
| 14967 break; | |
| 14968 | |
| 14969 case 0: | |
| 14970 /* | |
| 1808 | 14971 * Lengths are equal, thus changes must result in same length: An |
| 344 | 14972 * insert is only possible in combination with a delete. |
| 14973 * 1: check if for identical strings | |
| 14974 */ | |
| 14975 if (*pl == NUL) | |
| 351 | 14976 return score; |
| 344 | 14977 |
| 14978 /* 2: swap */ | |
| 14979 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 14980 { | |
| 14981 pl2 = pl + 2; /* swap, skip two chars */ | |
| 14982 ps2 = ps + 2; | |
| 14983 while (*pl2 == *ps2) | |
| 14984 { | |
| 14985 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 14986 return score + SCORE_SWAP; |
| 344 | 14987 ++pl2; |
| 14988 ++ps2; | |
| 14989 } | |
| 14990 /* 3: swap and swap again */ | |
| 14991 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 14992 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 14993 return score + SCORE_SWAP + SCORE_SWAP; |
| 344 | 14994 |
| 14995 /* 4: swap and substitute */ | |
| 14996 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 14997 return score + SCORE_SWAP + SCORE_SUBST; |
| 344 | 14998 } |
| 14999 | |
| 15000 /* 5: substitute */ | |
| 15001 pl2 = pl + 1; | |
| 15002 ps2 = ps + 1; | |
| 15003 while (*pl2 == *ps2) | |
| 15004 { | |
| 15005 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 15006 return score + SCORE_SUBST; |
| 344 | 15007 ++pl2; |
| 15008 ++ps2; | |
| 15009 } | |
| 15010 | |
| 15011 /* 6: substitute and swap */ | |
| 15012 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 15013 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 15014 return score + SCORE_SUBST + SCORE_SWAP; |
| 344 | 15015 |
| 15016 /* 7: substitute and substitute */ | |
| 15017 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 15018 return score + SCORE_SUBST + SCORE_SUBST; |
| 344 | 15019 |
| 15020 /* 8: insert then delete */ | |
| 15021 pl2 = pl; | |
| 15022 ps2 = ps + 1; | |
| 15023 while (*pl2 == *ps2) | |
| 15024 { | |
| 15025 ++pl2; | |
| 15026 ++ps2; | |
| 15027 } | |
| 15028 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 15029 return score + SCORE_INS + SCORE_DEL; |
| 344 | 15030 |
| 15031 /* 9: delete then insert */ | |
| 15032 pl2 = pl + 1; | |
| 15033 ps2 = ps; | |
| 15034 while (*pl2 == *ps2) | |
| 15035 { | |
| 15036 ++pl2; | |
| 15037 ++ps2; | |
| 15038 } | |
| 15039 if (STRCMP(pl2, ps2 + 1) == 0) | |
| 351 | 15040 return score + SCORE_INS + SCORE_DEL; |
| 344 | 15041 |
| 15042 /* Failed to compare. */ | |
| 15043 break; | |
| 15044 } | |
| 15045 | |
| 15046 return SCORE_MAXMAX; | |
| 15047 } | |
| 324 | 15048 |
| 323 | 15049 /* |
| 15050 * Compute the "edit distance" to turn "badword" into "goodword". The less | |
| 344 | 15051 * deletes/inserts/substitutes/swaps are required the lower the score. |
| 324 | 15052 * |
| 485 | 15053 * The algorithm is described by Du and Chang, 1992. |
| 15054 * The implementation of the algorithm comes from Aspell editdist.cpp, | |
| 15055 * edit_distance(). It has been converted from C++ to C and modified to | |
| 15056 * support multi-byte characters. | |
| 323 | 15057 */ |
| 15058 static int | |
| 625 | 15059 spell_edit_score(slang, badword, goodword) |
| 15060 slang_T *slang; | |
| 323 | 15061 char_u *badword; |
| 15062 char_u *goodword; | |
| 15063 { | |
| 15064 int *cnt; | |
| 1188 | 15065 int badlen, goodlen; /* lengths including NUL */ |
| 323 | 15066 int j, i; |
| 15067 int t; | |
| 15068 int bc, gc; | |
| 324 | 15069 int pbc, pgc; |
| 15070 #ifdef FEAT_MBYTE | |
| 15071 char_u *p; | |
| 15072 int wbadword[MAXWLEN]; | |
| 15073 int wgoodword[MAXWLEN]; | |
| 15074 | |
| 15075 if (has_mbyte) | |
| 15076 { | |
| 15077 /* Get the characters from the multi-byte strings and put them in an | |
| 15078 * int array for easy access. */ | |
| 15079 for (p = badword, badlen = 0; *p != NUL; ) | |
| 474 | 15080 wbadword[badlen++] = mb_cptr2char_adv(&p); |
| 398 | 15081 wbadword[badlen++] = 0; |
| 324 | 15082 for (p = goodword, goodlen = 0; *p != NUL; ) |
| 474 | 15083 wgoodword[goodlen++] = mb_cptr2char_adv(&p); |
| 398 | 15084 wgoodword[goodlen++] = 0; |
| 324 | 15085 } |
| 15086 else | |
| 15087 #endif | |
| 15088 { | |
| 835 | 15089 badlen = (int)STRLEN(badword) + 1; |
| 15090 goodlen = (int)STRLEN(goodword) + 1; | |
| 324 | 15091 } |
| 323 | 15092 |
| 15093 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */ | |
| 15094 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)] | |
| 15095 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)), | |
| 15096 TRUE); | |
| 324 | 15097 if (cnt == NULL) |
| 15098 return 0; /* out of memory */ | |
| 323 | 15099 |
| 15100 CNT(0, 0) = 0; | |
| 15101 for (j = 1; j <= goodlen; ++j) | |
| 625 | 15102 CNT(0, j) = CNT(0, j - 1) + SCORE_INS; |
| 323 | 15103 |
| 15104 for (i = 1; i <= badlen; ++i) | |
| 15105 { | |
| 625 | 15106 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL; |
| 323 | 15107 for (j = 1; j <= goodlen; ++j) |
| 15108 { | |
| 324 | 15109 #ifdef FEAT_MBYTE |
| 15110 if (has_mbyte) | |
| 15111 { | |
| 15112 bc = wbadword[i - 1]; | |
| 15113 gc = wgoodword[j - 1]; | |
| 15114 } | |
| 15115 else | |
| 15116 #endif | |
| 15117 { | |
| 15118 bc = badword[i - 1]; | |
| 15119 gc = goodword[j - 1]; | |
| 15120 } | |
| 323 | 15121 if (bc == gc) |
| 15122 CNT(i, j) = CNT(i - 1, j - 1); | |
| 15123 else | |
| 15124 { | |
| 15125 /* Use a better score when there is only a case difference. */ | |
| 324 | 15126 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) |
| 323 | 15127 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1); |
| 15128 else | |
| 625 | 15129 { |
| 15130 /* For a similar character use SCORE_SIMILAR. */ | |
| 15131 if (slang != NULL | |
| 15132 && slang->sl_has_map | |
| 15133 && similar_chars(slang, gc, bc)) | |
| 15134 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1); | |
| 15135 else | |
| 15136 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1); | |
| 15137 } | |
| 323 | 15138 |
| 324 | 15139 if (i > 1 && j > 1) |
| 323 | 15140 { |
| 324 | 15141 #ifdef FEAT_MBYTE |
| 15142 if (has_mbyte) | |
| 15143 { | |
| 15144 pbc = wbadword[i - 2]; | |
| 15145 pgc = wgoodword[j - 2]; | |
| 15146 } | |
| 15147 else | |
| 15148 #endif | |
| 15149 { | |
| 15150 pbc = badword[i - 2]; | |
| 15151 pgc = goodword[j - 2]; | |
| 15152 } | |
| 15153 if (bc == pgc && pbc == gc) | |
| 15154 { | |
| 15155 t = SCORE_SWAP + CNT(i - 2, j - 2); | |
| 15156 if (t < CNT(i, j)) | |
| 15157 CNT(i, j) = t; | |
| 15158 } | |
| 323 | 15159 } |
| 15160 t = SCORE_DEL + CNT(i - 1, j); | |
| 15161 if (t < CNT(i, j)) | |
| 15162 CNT(i, j) = t; | |
| 15163 t = SCORE_INS + CNT(i, j - 1); | |
| 15164 if (t < CNT(i, j)) | |
| 15165 CNT(i, j) = t; | |
| 15166 } | |
| 15167 } | |
| 15168 } | |
| 344 | 15169 |
| 15170 i = CNT(badlen - 1, goodlen - 1); | |
| 15171 vim_free(cnt); | |
| 15172 return i; | |
| 323 | 15173 } |
| 307 | 15174 |
| 625 | 15175 typedef struct |
| 15176 { | |
| 15177 int badi; | |
| 15178 int goodi; | |
| 15179 int score; | |
| 15180 } limitscore_T; | |
| 15181 | |
| 15182 /* | |
| 15183 * Like spell_edit_score(), but with a limit on the score to make it faster. | |
| 15184 * May return SCORE_MAXMAX when the score is higher than "limit". | |
| 15185 * | |
| 15186 * This uses a stack for the edits still to be tried. | |
| 15187 * The idea comes from Aspell leditdist.cpp. Rewritten in C and added support | |
| 15188 * for multi-byte characters. | |
| 15189 */ | |
| 15190 static int | |
| 15191 spell_edit_score_limit(slang, badword, goodword, limit) | |
| 15192 slang_T *slang; | |
| 15193 char_u *badword; | |
| 15194 char_u *goodword; | |
| 15195 int limit; | |
| 15196 { | |
| 15197 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 15198 int stackidx; | |
| 15199 int bi, gi; | |
| 15200 int bi2, gi2; | |
| 15201 int bc, gc; | |
| 15202 int score; | |
| 15203 int score_off; | |
| 15204 int minscore; | |
| 15205 int round; | |
| 15206 | |
| 15207 #ifdef FEAT_MBYTE | |
| 15208 /* Multi-byte characters require a bit more work, use a different function | |
| 15209 * to avoid testing "has_mbyte" quite often. */ | |
| 15210 if (has_mbyte) | |
| 15211 return spell_edit_score_limit_w(slang, badword, goodword, limit); | |
| 15212 #endif | |
| 15213 | |
| 15214 /* | |
| 15215 * The idea is to go from start to end over the words. So long as | |
| 15216 * characters are equal just continue, this always gives the lowest score. | |
| 15217 * When there is a difference try several alternatives. Each alternative | |
| 15218 * increases "score" for the edit distance. Some of the alternatives are | |
| 15219 * pushed unto a stack and tried later, some are tried right away. At the | |
| 15220 * end of the word the score for one alternative is known. The lowest | |
| 15221 * possible score is stored in "minscore". | |
| 15222 */ | |
| 15223 stackidx = 0; | |
| 15224 bi = 0; | |
| 15225 gi = 0; | |
| 15226 score = 0; | |
| 15227 minscore = limit + 1; | |
| 15228 | |
| 15229 for (;;) | |
| 15230 { | |
| 15231 /* Skip over an equal part, score remains the same. */ | |
| 15232 for (;;) | |
| 15233 { | |
| 15234 bc = badword[bi]; | |
| 15235 gc = goodword[gi]; | |
| 15236 if (bc != gc) /* stop at a char that's different */ | |
| 15237 break; | |
| 15238 if (bc == NUL) /* both words end */ | |
| 15239 { | |
| 15240 if (score < minscore) | |
| 15241 minscore = score; | |
| 15242 goto pop; /* do next alternative */ | |
| 15243 } | |
| 15244 ++bi; | |
| 15245 ++gi; | |
| 15246 } | |
| 15247 | |
| 15248 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 15249 { | |
| 15250 do | |
| 15251 { | |
| 15252 if ((score += SCORE_DEL) >= minscore) | |
| 15253 goto pop; /* do next alternative */ | |
| 15254 } while (badword[++bi] != NUL); | |
| 15255 minscore = score; | |
| 15256 } | |
| 15257 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 15258 { | |
| 15259 do | |
| 15260 { | |
| 15261 if ((score += SCORE_INS) >= minscore) | |
| 15262 goto pop; /* do next alternative */ | |
| 15263 } while (goodword[++gi] != NUL); | |
| 15264 minscore = score; | |
| 15265 } | |
| 15266 else /* both words continue */ | |
| 15267 { | |
| 15268 /* If not close to the limit, perform a change. Only try changes | |
| 15269 * that may lead to a lower score than "minscore". | |
| 15270 * round 0: try deleting a char from badword | |
| 15271 * round 1: try inserting a char in badword */ | |
| 15272 for (round = 0; round <= 1; ++round) | |
| 15273 { | |
| 15274 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 15275 if (score_off < minscore) | |
| 15276 { | |
| 15277 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 15278 { | |
| 15279 /* Near the limit, rest of the words must match. We | |
| 15280 * can check that right now, no need to push an item | |
| 15281 * onto the stack. */ | |
| 15282 bi2 = bi + 1 - round; | |
| 15283 gi2 = gi + round; | |
| 15284 while (goodword[gi2] == badword[bi2]) | |
| 15285 { | |
| 15286 if (goodword[gi2] == NUL) | |
| 15287 { | |
| 15288 minscore = score_off; | |
| 15289 break; | |
| 15290 } | |
| 15291 ++bi2; | |
| 15292 ++gi2; | |
| 15293 } | |
| 15294 } | |
| 15295 else | |
| 15296 { | |
| 15297 /* try deleting/inserting a character later */ | |
| 15298 stack[stackidx].badi = bi + 1 - round; | |
| 15299 stack[stackidx].goodi = gi + round; | |
| 15300 stack[stackidx].score = score_off; | |
| 15301 ++stackidx; | |
| 15302 } | |
| 15303 } | |
| 15304 } | |
| 15305 | |
| 15306 if (score + SCORE_SWAP < minscore) | |
| 15307 { | |
| 15308 /* If swapping two characters makes a match then the | |
| 15309 * substitution is more expensive, thus there is no need to | |
| 15310 * try both. */ | |
| 15311 if (gc == badword[bi + 1] && bc == goodword[gi + 1]) | |
| 15312 { | |
| 15313 /* Swap two characters, that is: skip them. */ | |
| 15314 gi += 2; | |
| 15315 bi += 2; | |
| 15316 score += SCORE_SWAP; | |
| 15317 continue; | |
| 15318 } | |
| 15319 } | |
| 15320 | |
| 15321 /* Substitute one character for another which is the same | |
| 15322 * thing as deleting a character from both goodword and badword. | |
| 15323 * Use a better score when there is only a case difference. */ | |
| 15324 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 15325 score += SCORE_ICASE; | |
| 15326 else | |
| 15327 { | |
| 15328 /* For a similar character use SCORE_SIMILAR. */ | |
| 15329 if (slang != NULL | |
| 15330 && slang->sl_has_map | |
| 15331 && similar_chars(slang, gc, bc)) | |
| 15332 score += SCORE_SIMILAR; | |
| 15333 else | |
| 15334 score += SCORE_SUBST; | |
| 15335 } | |
| 15336 | |
| 15337 if (score < minscore) | |
| 15338 { | |
| 15339 /* Do the substitution. */ | |
| 15340 ++gi; | |
| 15341 ++bi; | |
| 15342 continue; | |
| 15343 } | |
| 15344 } | |
| 15345 pop: | |
| 15346 /* | |
| 15347 * Get here to try the next alternative, pop it from the stack. | |
| 15348 */ | |
| 15349 if (stackidx == 0) /* stack is empty, finished */ | |
| 15350 break; | |
| 15351 | |
| 15352 /* pop an item from the stack */ | |
| 15353 --stackidx; | |
| 15354 gi = stack[stackidx].goodi; | |
| 15355 bi = stack[stackidx].badi; | |
| 15356 score = stack[stackidx].score; | |
| 15357 } | |
| 15358 | |
| 15359 /* When the score goes over "limit" it may actually be much higher. | |
| 15360 * Return a very large number to avoid going below the limit when giving a | |
| 15361 * bonus. */ | |
| 15362 if (minscore > limit) | |
| 15363 return SCORE_MAXMAX; | |
| 15364 return minscore; | |
| 15365 } | |
| 15366 | |
| 15367 #ifdef FEAT_MBYTE | |
| 15368 /* | |
| 15369 * Multi-byte version of spell_edit_score_limit(). | |
| 15370 * Keep it in sync with the above! | |
| 15371 */ | |
| 15372 static int | |
| 15373 spell_edit_score_limit_w(slang, badword, goodword, limit) | |
| 15374 slang_T *slang; | |
| 15375 char_u *badword; | |
| 15376 char_u *goodword; | |
| 15377 int limit; | |
| 15378 { | |
| 15379 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 15380 int stackidx; | |
| 15381 int bi, gi; | |
| 15382 int bi2, gi2; | |
| 15383 int bc, gc; | |
| 15384 int score; | |
| 15385 int score_off; | |
| 15386 int minscore; | |
| 15387 int round; | |
| 15388 char_u *p; | |
| 15389 int wbadword[MAXWLEN]; | |
| 15390 int wgoodword[MAXWLEN]; | |
| 15391 | |
| 15392 /* Get the characters from the multi-byte strings and put them in an | |
| 15393 * int array for easy access. */ | |
| 15394 bi = 0; | |
| 15395 for (p = badword; *p != NUL; ) | |
| 15396 wbadword[bi++] = mb_cptr2char_adv(&p); | |
| 15397 wbadword[bi++] = 0; | |
| 15398 gi = 0; | |
| 15399 for (p = goodword; *p != NUL; ) | |
| 15400 wgoodword[gi++] = mb_cptr2char_adv(&p); | |
| 15401 wgoodword[gi++] = 0; | |
| 15402 | |
| 15403 /* | |
| 15404 * The idea is to go from start to end over the words. So long as | |
| 15405 * characters are equal just continue, this always gives the lowest score. | |
| 15406 * When there is a difference try several alternatives. Each alternative | |
| 15407 * increases "score" for the edit distance. Some of the alternatives are | |
| 15408 * pushed unto a stack and tried later, some are tried right away. At the | |
| 15409 * end of the word the score for one alternative is known. The lowest | |
| 15410 * possible score is stored in "minscore". | |
| 15411 */ | |
| 15412 stackidx = 0; | |
| 15413 bi = 0; | |
| 15414 gi = 0; | |
| 15415 score = 0; | |
| 15416 minscore = limit + 1; | |
| 15417 | |
| 15418 for (;;) | |
| 15419 { | |
| 15420 /* Skip over an equal part, score remains the same. */ | |
| 15421 for (;;) | |
| 15422 { | |
| 15423 bc = wbadword[bi]; | |
| 15424 gc = wgoodword[gi]; | |
| 15425 | |
| 15426 if (bc != gc) /* stop at a char that's different */ | |
| 15427 break; | |
| 15428 if (bc == NUL) /* both words end */ | |
| 15429 { | |
| 15430 if (score < minscore) | |
| 15431 minscore = score; | |
| 15432 goto pop; /* do next alternative */ | |
| 15433 } | |
| 15434 ++bi; | |
| 15435 ++gi; | |
| 15436 } | |
| 15437 | |
| 15438 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 15439 { | |
| 15440 do | |
| 15441 { | |
| 15442 if ((score += SCORE_DEL) >= minscore) | |
| 15443 goto pop; /* do next alternative */ | |
| 15444 } while (wbadword[++bi] != NUL); | |
| 15445 minscore = score; | |
| 15446 } | |
| 15447 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 15448 { | |
| 15449 do | |
| 15450 { | |
| 15451 if ((score += SCORE_INS) >= minscore) | |
| 15452 goto pop; /* do next alternative */ | |
| 15453 } while (wgoodword[++gi] != NUL); | |
| 15454 minscore = score; | |
| 15455 } | |
| 15456 else /* both words continue */ | |
| 15457 { | |
| 15458 /* If not close to the limit, perform a change. Only try changes | |
| 15459 * that may lead to a lower score than "minscore". | |
| 15460 * round 0: try deleting a char from badword | |
| 15461 * round 1: try inserting a char in badword */ | |
| 15462 for (round = 0; round <= 1; ++round) | |
| 15463 { | |
| 15464 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 15465 if (score_off < minscore) | |
| 15466 { | |
| 15467 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 15468 { | |
| 15469 /* Near the limit, rest of the words must match. We | |
| 15470 * can check that right now, no need to push an item | |
| 15471 * onto the stack. */ | |
| 15472 bi2 = bi + 1 - round; | |
| 15473 gi2 = gi + round; | |
| 15474 while (wgoodword[gi2] == wbadword[bi2]) | |
| 15475 { | |
| 15476 if (wgoodword[gi2] == NUL) | |
| 15477 { | |
| 15478 minscore = score_off; | |
| 15479 break; | |
| 15480 } | |
| 15481 ++bi2; | |
| 15482 ++gi2; | |
| 15483 } | |
| 15484 } | |
| 15485 else | |
| 15486 { | |
| 15487 /* try deleting a character from badword later */ | |
| 15488 stack[stackidx].badi = bi + 1 - round; | |
| 15489 stack[stackidx].goodi = gi + round; | |
| 15490 stack[stackidx].score = score_off; | |
| 15491 ++stackidx; | |
| 15492 } | |
| 15493 } | |
| 15494 } | |
| 15495 | |
| 15496 if (score + SCORE_SWAP < minscore) | |
| 15497 { | |
| 15498 /* If swapping two characters makes a match then the | |
| 15499 * substitution is more expensive, thus there is no need to | |
| 15500 * try both. */ | |
| 15501 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1]) | |
| 15502 { | |
| 15503 /* Swap two characters, that is: skip them. */ | |
| 15504 gi += 2; | |
| 15505 bi += 2; | |
| 15506 score += SCORE_SWAP; | |
| 15507 continue; | |
| 15508 } | |
| 15509 } | |
| 15510 | |
| 15511 /* Substitute one character for another which is the same | |
| 15512 * thing as deleting a character from both goodword and badword. | |
| 15513 * Use a better score when there is only a case difference. */ | |
| 15514 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 15515 score += SCORE_ICASE; | |
| 15516 else | |
| 15517 { | |
| 15518 /* For a similar character use SCORE_SIMILAR. */ | |
| 15519 if (slang != NULL | |
| 15520 && slang->sl_has_map | |
| 15521 && similar_chars(slang, gc, bc)) | |
| 15522 score += SCORE_SIMILAR; | |
| 15523 else | |
| 15524 score += SCORE_SUBST; | |
| 15525 } | |
| 15526 | |
| 15527 if (score < minscore) | |
| 15528 { | |
| 15529 /* Do the substitution. */ | |
| 15530 ++gi; | |
| 15531 ++bi; | |
| 15532 continue; | |
| 15533 } | |
| 15534 } | |
| 15535 pop: | |
| 15536 /* | |
| 15537 * Get here to try the next alternative, pop it from the stack. | |
| 15538 */ | |
| 15539 if (stackidx == 0) /* stack is empty, finished */ | |
| 15540 break; | |
| 15541 | |
| 15542 /* pop an item from the stack */ | |
| 15543 --stackidx; | |
| 15544 gi = stack[stackidx].goodi; | |
| 15545 bi = stack[stackidx].badi; | |
| 15546 score = stack[stackidx].score; | |
| 15547 } | |
| 15548 | |
| 15549 /* When the score goes over "limit" it may actually be much higher. | |
| 15550 * Return a very large number to avoid going below the limit when giving a | |
| 15551 * bonus. */ | |
| 15552 if (minscore > limit) | |
| 15553 return SCORE_MAXMAX; | |
| 15554 return minscore; | |
| 15555 } | |
| 15556 #endif | |
| 15557 | |
| 714 | 15558 /* |
| 15559 * ":spellinfo" | |
| 15560 */ | |
| 15561 void | |
| 15562 ex_spellinfo(eap) | |
| 1883 | 15563 exarg_T *eap UNUSED; |
| 714 | 15564 { |
| 15565 int lpi; | |
| 15566 langp_T *lp; | |
| 15567 char_u *p; | |
| 15568 | |
| 15569 if (no_spell_checking(curwin)) | |
| 15570 return; | |
| 15571 | |
| 15572 msg_start(); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15573 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
|
15574 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15575 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 714 | 15576 msg_puts((char_u *)"file: "); |
| 15577 msg_puts(lp->lp_slang->sl_fname); | |
| 15578 msg_putchar('\n'); | |
| 15579 p = lp->lp_slang->sl_info; | |
| 15580 if (p != NULL) | |
| 15581 { | |
| 15582 msg_puts(p); | |
| 15583 msg_putchar('\n'); | |
| 15584 } | |
| 15585 } | |
| 15586 msg_end(); | |
| 15587 } | |
| 15588 | |
| 625 | 15589 #define DUMPFLAG_KEEPCASE 1 /* round 2: keep-case tree */ |
| 15590 #define DUMPFLAG_COUNT 2 /* include word count */ | |
| 701 | 15591 #define DUMPFLAG_ICASE 4 /* ignore case when finding matches */ |
| 710 | 15592 #define DUMPFLAG_ONECAP 8 /* pattern starts with capital */ |
| 15593 #define DUMPFLAG_ALLCAP 16 /* pattern is all capitals */ | |
| 625 | 15594 |
| 351 | 15595 /* |
| 15596 * ":spelldump" | |
| 15597 */ | |
| 15598 void | |
| 15599 ex_spelldump(eap) | |
| 15600 exarg_T *eap; | |
| 15601 { | |
| 5382 | 15602 char_u *spl; |
| 15603 long dummy; | |
| 15604 | |
| 701 | 15605 if (no_spell_checking(curwin)) |
| 15606 return; | |
| 5382 | 15607 get_option_value((char_u*)"spl", &dummy, &spl, OPT_LOCAL); |
| 15608 | |
| 15609 /* Create a new empty buffer in a new window. */ | |
| 701 | 15610 do_cmdline_cmd((char_u *)"new"); |
| 5382 | 15611 |
| 15612 /* enable spelling locally in the new window */ | |
| 15613 set_option_value((char_u*)"spell", TRUE, (char_u*)"", OPT_LOCAL); | |
| 15614 set_option_value((char_u*)"spl", dummy, spl, OPT_LOCAL); | |
| 15615 vim_free(spl); | |
| 15616 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15617 if (!bufempty() || !buf_valid(curbuf)) |
| 701 | 15618 return; |
| 15619 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15620 spell_dump_compl(NULL, 0, NULL, eap->forceit ? DUMPFLAG_COUNT : 0); |
| 701 | 15621 |
| 15622 /* Delete the empty line that we started with. */ | |
| 15623 if (curbuf->b_ml.ml_line_count > 1) | |
| 15624 ml_delete(curbuf->b_ml.ml_line_count, FALSE); | |
| 15625 | |
| 15626 redraw_later(NOT_VALID); | |
| 15627 } | |
| 15628 | |
| 15629 /* | |
| 15630 * Go through all possible words and: | |
| 15631 * 1. When "pat" is NULL: dump a list of all words in the current buffer. | |
| 15632 * "ic" and "dir" are not used. | |
| 15633 * 2. When "pat" is not NULL: add matching words to insert mode completion. | |
| 15634 */ | |
| 15635 void | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15636 spell_dump_compl(pat, ic, dir, dumpflags_arg) |
| 701 | 15637 char_u *pat; /* leading part of the word */ |
| 15638 int ic; /* ignore case */ | |
| 15639 int *dir; /* direction for adding matches */ | |
| 15640 int dumpflags_arg; /* DUMPFLAG_* */ | |
| 15641 { | |
| 351 | 15642 langp_T *lp; |
| 15643 slang_T *slang; | |
| 15644 idx_T arridx[MAXWLEN]; | |
| 15645 int curi[MAXWLEN]; | |
| 15646 char_u word[MAXWLEN]; | |
| 15647 int c; | |
| 15648 char_u *byts; | |
| 15649 idx_T *idxs; | |
| 15650 linenr_T lnum = 0; | |
| 15651 int round; | |
| 15652 int depth; | |
| 15653 int n; | |
| 15654 int flags; | |
| 381 | 15655 char_u *region_names = NULL; /* region names being used */ |
| 15656 int do_region = TRUE; /* dump region names and numbers */ | |
| 15657 char_u *p; | |
| 500 | 15658 int lpi; |
| 701 | 15659 int dumpflags = dumpflags_arg; |
| 15660 int patlen; | |
| 15661 | |
| 710 | 15662 /* When ignoring case or when the pattern starts with capital pass this on |
| 15663 * to dump_word(). */ | |
| 15664 if (pat != NULL) | |
| 15665 { | |
| 15666 if (ic) | |
| 15667 dumpflags |= DUMPFLAG_ICASE; | |
| 15668 else | |
| 15669 { | |
| 15670 n = captype(pat, NULL); | |
| 15671 if (n == WF_ONECAP) | |
| 15672 dumpflags |= DUMPFLAG_ONECAP; | |
| 15673 else if (n == WF_ALLCAP | |
| 15674 #ifdef FEAT_MBYTE | |
| 714 | 15675 && (int)STRLEN(pat) > mb_ptr2len(pat) |
| 710 | 15676 #else |
| 714 | 15677 && (int)STRLEN(pat) > 1 |
| 710 | 15678 #endif |
| 15679 ) | |
| 15680 dumpflags |= DUMPFLAG_ALLCAP; | |
| 15681 } | |
| 15682 } | |
| 351 | 15683 |
| 381 | 15684 /* Find out if we can support regions: All languages must support the same |
| 15685 * regions or none at all. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15686 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
|
15687 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15688 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 381 | 15689 p = lp->lp_slang->sl_regions; |
| 15690 if (p[0] != 0) | |
| 15691 { | |
| 15692 if (region_names == NULL) /* first language with regions */ | |
| 15693 region_names = p; | |
| 15694 else if (STRCMP(region_names, p) != 0) | |
| 15695 { | |
| 15696 do_region = FALSE; /* region names are different */ | |
| 15697 break; | |
| 15698 } | |
| 15699 } | |
| 15700 } | |
| 15701 | |
| 15702 if (do_region && region_names != NULL) | |
| 15703 { | |
| 701 | 15704 if (pat == NULL) |
| 15705 { | |
| 15706 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names); | |
| 15707 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 15708 } | |
| 381 | 15709 } |
| 15710 else | |
| 15711 do_region = FALSE; | |
| 15712 | |
| 15713 /* | |
| 15714 * Loop over all files loaded for the entries in 'spelllang'. | |
| 15715 */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15716 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
|
15717 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
15718 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 351 | 15719 slang = lp->lp_slang; |
| 500 | 15720 if (slang->sl_fbyts == NULL) /* reloading failed */ |
| 15721 continue; | |
| 351 | 15722 |
| 701 | 15723 if (pat == NULL) |
| 15724 { | |
| 15725 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname); | |
| 15726 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 15727 } | |
| 15728 | |
| 15729 /* When matching with a pattern and there are no prefixes only use | |
| 15730 * parts of the tree that match "pat". */ | |
| 15731 if (pat != NULL && slang->sl_pbyts == NULL) | |
| 835 | 15732 patlen = (int)STRLEN(pat); |
| 701 | 15733 else |
| 840 | 15734 patlen = -1; |
| 351 | 15735 |
| 15736 /* round 1: case-folded tree | |
| 15737 * round 2: keep-case tree */ | |
| 15738 for (round = 1; round <= 2; ++round) | |
| 15739 { | |
| 15740 if (round == 1) | |
| 15741 { | |
| 701 | 15742 dumpflags &= ~DUMPFLAG_KEEPCASE; |
| 351 | 15743 byts = slang->sl_fbyts; |
| 15744 idxs = slang->sl_fidxs; | |
| 15745 } | |
| 15746 else | |
| 15747 { | |
| 701 | 15748 dumpflags |= DUMPFLAG_KEEPCASE; |
| 351 | 15749 byts = slang->sl_kbyts; |
| 15750 idxs = slang->sl_kidxs; | |
| 15751 } | |
| 15752 if (byts == NULL) | |
| 15753 continue; /* array is empty */ | |
| 15754 | |
| 15755 depth = 0; | |
| 15756 arridx[0] = 0; | |
| 15757 curi[0] = 1; | |
| 701 | 15758 while (depth >= 0 && !got_int |
| 15759 && (pat == NULL || !compl_interrupted)) | |
| 351 | 15760 { |
| 15761 if (curi[depth] > byts[arridx[depth]]) | |
| 15762 { | |
| 15763 /* Done all bytes at this node, go up one level. */ | |
| 15764 --depth; | |
| 15765 line_breakcheck(); | |
| 715 | 15766 ins_compl_check_keys(50); |
| 351 | 15767 } |
| 15768 else | |
| 15769 { | |
| 15770 /* Do one more byte at this node. */ | |
| 15771 n = arridx[depth] + curi[depth]; | |
| 15772 ++curi[depth]; | |
| 15773 c = byts[n]; | |
| 15774 if (c == 0) | |
| 15775 { | |
| 15776 /* End of word, deal with the word. | |
| 15777 * Don't use keep-case words in the fold-case tree, | |
| 15778 * they will appear in the keep-case tree. | |
| 15779 * Only use the word when the region matches. */ | |
| 15780 flags = (int)idxs[n]; | |
| 15781 if ((round == 2 || (flags & WF_KEEPCAP) == 0) | |
| 500 | 15782 && (flags & WF_NEEDCOMP) == 0 |
| 381 | 15783 && (do_region |
| 15784 || (flags & WF_REGION) == 0 | |
| 390 | 15785 || (((unsigned)flags >> 16) |
| 351 | 15786 & lp->lp_region) != 0)) |
| 15787 { | |
| 15788 word[depth] = NUL; | |
| 381 | 15789 if (!do_region) |
| 15790 flags &= ~WF_REGION; | |
| 355 | 15791 |
| 15792 /* Dump the basic word if there is no prefix or | |
| 15793 * when it's the first one. */ | |
| 390 | 15794 c = (unsigned)flags >> 24; |
| 355 | 15795 if (c == 0 || curi[depth] == 2) |
| 701 | 15796 { |
| 15797 dump_word(slang, word, pat, dir, | |
| 15798 dumpflags, flags, lnum); | |
| 15799 if (pat == NULL) | |
| 15800 ++lnum; | |
| 15801 } | |
| 351 | 15802 |
| 15803 /* Apply the prefix, if there is one. */ | |
| 355 | 15804 if (c != 0) |
| 701 | 15805 lnum = dump_prefixes(slang, word, pat, dir, |
| 15806 dumpflags, flags, lnum); | |
| 351 | 15807 } |
| 15808 } | |
| 15809 else | |
| 15810 { | |
| 15811 /* Normal char, go one level deeper. */ | |
| 15812 word[depth++] = c; | |
| 15813 arridx[depth] = idxs[n]; | |
| 15814 curi[depth] = 1; | |
| 701 | 15815 |
| 15816 /* Check if this characters matches with the pattern. | |
| 15817 * If not skip the whole tree below it. | |
| 710 | 15818 * Always ignore case here, dump_word() will check |
| 15819 * proper case later. This isn't exactly right when | |
| 15820 * length changes for multi-byte characters with | |
| 15821 * ignore case... */ | |
| 15822 if (depth <= patlen | |
| 15823 && MB_STRNICMP(word, pat, depth) != 0) | |
| 701 | 15824 --depth; |
| 351 | 15825 } |
| 15826 } | |
| 15827 } | |
| 15828 } | |
| 15829 } | |
| 15830 } | |
| 15831 | |
| 15832 /* | |
| 15833 * Dump one word: apply case modifications and append a line to the buffer. | |
| 701 | 15834 * When "lnum" is zero add insert mode completion. |
| 15835 */ | |
| 15836 static void | |
| 710 | 15837 dump_word(slang, word, pat, dir, dumpflags, wordflags, lnum) |
| 625 | 15838 slang_T *slang; |
| 351 | 15839 char_u *word; |
| 701 | 15840 char_u *pat; |
| 15841 int *dir; | |
| 625 | 15842 int dumpflags; |
| 710 | 15843 int wordflags; |
| 351 | 15844 linenr_T lnum; |
| 15845 { | |
| 15846 int keepcap = FALSE; | |
| 15847 char_u *p; | |
| 625 | 15848 char_u *tw; |
| 351 | 15849 char_u cword[MAXWLEN]; |
| 381 | 15850 char_u badword[MAXWLEN + 10]; |
| 15851 int i; | |
| 710 | 15852 int flags = wordflags; |
| 15853 | |
| 15854 if (dumpflags & DUMPFLAG_ONECAP) | |
| 15855 flags |= WF_ONECAP; | |
| 15856 if (dumpflags & DUMPFLAG_ALLCAP) | |
| 15857 flags |= WF_ALLCAP; | |
| 351 | 15858 |
| 625 | 15859 if ((dumpflags & DUMPFLAG_KEEPCASE) == 0 && (flags & WF_CAPMASK) != 0) |
| 351 | 15860 { |
| 15861 /* Need to fix case according to "flags". */ | |
| 15862 make_case_word(word, cword, flags); | |
| 15863 p = cword; | |
| 15864 } | |
| 15865 else | |
| 15866 { | |
| 15867 p = word; | |
| 625 | 15868 if ((dumpflags & DUMPFLAG_KEEPCASE) |
| 15869 && ((captype(word, NULL) & WF_KEEPCAP) == 0 | |
| 389 | 15870 || (flags & WF_FIXCAP) != 0)) |
| 351 | 15871 keepcap = TRUE; |
| 15872 } | |
| 625 | 15873 tw = p; |
| 351 | 15874 |
| 701 | 15875 if (pat == NULL) |
| 15876 { | |
| 15877 /* Add flags and regions after a slash. */ | |
| 15878 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap) | |
| 15879 { | |
| 15880 STRCPY(badword, p); | |
| 15881 STRCAT(badword, "/"); | |
| 15882 if (keepcap) | |
| 15883 STRCAT(badword, "="); | |
| 15884 if (flags & WF_BANNED) | |
| 15885 STRCAT(badword, "!"); | |
| 15886 else if (flags & WF_RARE) | |
| 15887 STRCAT(badword, "?"); | |
| 15888 if (flags & WF_REGION) | |
| 15889 for (i = 0; i < 7; ++i) | |
| 15890 if (flags & (0x10000 << i)) | |
| 15891 sprintf((char *)badword + STRLEN(badword), "%d", i + 1); | |
| 15892 p = badword; | |
| 15893 } | |
| 15894 | |
| 15895 if (dumpflags & DUMPFLAG_COUNT) | |
| 15896 { | |
| 15897 hashitem_T *hi; | |
| 15898 | |
| 15899 /* Include the word count for ":spelldump!". */ | |
| 15900 hi = hash_find(&slang->sl_wordcount, tw); | |
| 15901 if (!HASHITEM_EMPTY(hi)) | |
| 15902 { | |
| 15903 vim_snprintf((char *)IObuff, IOSIZE, "%s\t%d", | |
| 625 | 15904 tw, HI2WC(hi)->wc_count); |
| 701 | 15905 p = IObuff; |
| 15906 } | |
| 15907 } | |
| 15908 | |
| 15909 ml_append(lnum, p, (colnr_T)0, FALSE); | |
| 15910 } | |
| 710 | 15911 else if (((dumpflags & DUMPFLAG_ICASE) |
| 15912 ? MB_STRNICMP(p, pat, STRLEN(pat)) == 0 | |
| 15913 : STRNCMP(p, pat, STRLEN(pat)) == 0) | |
| 701 | 15914 && ins_compl_add_infercase(p, (int)STRLEN(p), |
| 942 | 15915 p_ic, NULL, *dir, 0) == OK) |
| 710 | 15916 /* if dir was BACKWARD then honor it just once */ |
| 15917 *dir = FORWARD; | |
| 351 | 15918 } |
| 15919 | |
| 15920 /* | |
| 372 | 15921 * For ":spelldump": Find matching prefixes for "word". Prepend each to |
| 15922 * "word" and append a line to the buffer. | |
| 701 | 15923 * When "lnum" is zero add insert mode completion. |
| 351 | 15924 * Return the updated line number. |
| 15925 */ | |
| 15926 static linenr_T | |
| 701 | 15927 dump_prefixes(slang, word, pat, dir, dumpflags, flags, startlnum) |
| 351 | 15928 slang_T *slang; |
| 15929 char_u *word; /* case-folded word */ | |
| 701 | 15930 char_u *pat; |
| 15931 int *dir; | |
| 625 | 15932 int dumpflags; |
| 351 | 15933 int flags; /* flags with prefix ID */ |
| 15934 linenr_T startlnum; | |
| 15935 { | |
| 15936 idx_T arridx[MAXWLEN]; | |
| 15937 int curi[MAXWLEN]; | |
| 15938 char_u prefix[MAXWLEN]; | |
| 455 | 15939 char_u word_up[MAXWLEN]; |
| 15940 int has_word_up = FALSE; | |
| 351 | 15941 int c; |
| 15942 char_u *byts; | |
| 15943 idx_T *idxs; | |
| 15944 linenr_T lnum = startlnum; | |
| 15945 int depth; | |
| 15946 int n; | |
| 15947 int len; | |
| 15948 int i; | |
| 15949 | |
| 626 | 15950 /* If the word starts with a lower-case letter make the word with an |
| 455 | 15951 * upper-case letter in word_up[]. */ |
| 15952 c = PTR2CHAR(word); | |
| 15953 if (SPELL_TOUPPER(c) != c) | |
| 15954 { | |
| 15955 onecap_copy(word, word_up, TRUE); | |
| 15956 has_word_up = TRUE; | |
| 15957 } | |
| 15958 | |
| 351 | 15959 byts = slang->sl_pbyts; |
| 15960 idxs = slang->sl_pidxs; | |
| 15961 if (byts != NULL) /* array not is empty */ | |
| 15962 { | |
| 15963 /* | |
| 15964 * Loop over all prefixes, building them byte-by-byte in prefix[]. | |
| 390 | 15965 * When at the end of a prefix check that it supports "flags". |
| 351 | 15966 */ |
| 15967 depth = 0; | |
| 15968 arridx[0] = 0; | |
| 15969 curi[0] = 1; | |
| 15970 while (depth >= 0 && !got_int) | |
| 15971 { | |
| 390 | 15972 n = arridx[depth]; |
| 15973 len = byts[n]; | |
| 15974 if (curi[depth] > len) | |
| 351 | 15975 { |
| 15976 /* Done all bytes at this node, go up one level. */ | |
| 15977 --depth; | |
| 15978 line_breakcheck(); | |
| 15979 } | |
| 15980 else | |
| 15981 { | |
| 15982 /* Do one more byte at this node. */ | |
| 390 | 15983 n += curi[depth]; |
| 351 | 15984 ++curi[depth]; |
| 15985 c = byts[n]; | |
| 15986 if (c == 0) | |
| 15987 { | |
| 15988 /* End of prefix, find out how many IDs there are. */ | |
| 15989 for (i = 1; i < len; ++i) | |
| 15990 if (byts[n + i] != 0) | |
| 15991 break; | |
| 15992 curi[depth] += i - 1; | |
| 15993 | |
| 455 | 15994 c = valid_word_prefix(i, n, flags, word, slang, FALSE); |
| 15995 if (c != 0) | |
| 351 | 15996 { |
| 376 | 15997 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1); |
| 701 | 15998 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 15999 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 16000 : flags, lnum); |
| 16001 if (lnum != 0) | |
| 16002 ++lnum; | |
| 351 | 16003 } |
| 455 | 16004 |
| 16005 /* Check for prefix that matches the word when the | |
| 16006 * first letter is upper-case, but only if the prefix has | |
| 16007 * a condition. */ | |
| 16008 if (has_word_up) | |
| 16009 { | |
| 16010 c = valid_word_prefix(i, n, flags, word_up, slang, | |
| 16011 TRUE); | |
| 16012 if (c != 0) | |
| 16013 { | |
| 16014 vim_strncpy(prefix + depth, word_up, | |
| 16015 MAXWLEN - depth - 1); | |
| 701 | 16016 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 16017 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 16018 : flags, lnum); |
| 16019 if (lnum != 0) | |
| 16020 ++lnum; | |
| 455 | 16021 } |
| 16022 } | |
| 351 | 16023 } |
| 16024 else | |
| 16025 { | |
| 16026 /* Normal char, go one level deeper. */ | |
| 16027 prefix[depth++] = c; | |
| 16028 arridx[depth] = idxs[n]; | |
| 16029 curi[depth] = 1; | |
| 16030 } | |
| 16031 } | |
| 16032 } | |
| 16033 } | |
| 16034 | |
| 16035 return lnum; | |
| 16036 } | |
| 16037 | |
| 498 | 16038 /* |
| 626 | 16039 * Move "p" to the end of word "start". |
| 16040 * Uses the spell-checking word characters. | |
| 498 | 16041 */ |
| 16042 char_u * | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
16043 spell_to_word_end(start, win) |
| 498 | 16044 char_u *start; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
16045 win_T *win; |
| 498 | 16046 { |
| 16047 char_u *p = start; | |
| 16048 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
16049 while (*p != NUL && spell_iswordp(p, win)) |
| 498 | 16050 mb_ptr_adv(p); |
| 16051 return p; | |
| 16052 } | |
| 16053 | |
| 475 | 16054 #if defined(FEAT_INS_EXPAND) || defined(PROTO) |
| 16055 /* | |
| 626 | 16056 * For Insert mode completion CTRL-X s: |
| 16057 * Find start of the word in front of column "startcol". | |
| 16058 * We don't check if it is badly spelled, with completion we can only change | |
| 16059 * the word in front of the cursor. | |
| 475 | 16060 * Returns the column number of the word. |
| 16061 */ | |
| 16062 int | |
| 16063 spell_word_start(startcol) | |
| 16064 int startcol; | |
| 16065 { | |
| 16066 char_u *line; | |
| 16067 char_u *p; | |
| 16068 int col = 0; | |
| 16069 | |
| 498 | 16070 if (no_spell_checking(curwin)) |
| 475 | 16071 return startcol; |
| 16072 | |
| 16073 /* Find a word character before "startcol". */ | |
| 16074 line = ml_get_curline(); | |
| 16075 for (p = line + startcol; p > line; ) | |
| 16076 { | |
| 16077 mb_ptr_back(line, p); | |
| 5477 | 16078 if (spell_iswordp_nmw(p, curwin)) |
| 475 | 16079 break; |
| 16080 } | |
| 16081 | |
| 16082 /* Go back to start of the word. */ | |
| 16083 while (p > line) | |
| 16084 { | |
| 835 | 16085 col = (int)(p - line); |
| 475 | 16086 mb_ptr_back(line, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
16087 if (!spell_iswordp(p, curwin)) |
| 475 | 16088 break; |
| 16089 col = 0; | |
| 16090 } | |
| 16091 | |
| 535 | 16092 return col; |
| 16093 } | |
| 16094 | |
| 16095 /* | |
| 16096 * Need to check for 'spellcapcheck' now, the word is removed before | |
| 16097 * expand_spelling() is called. Therefore the ugly global variable. | |
| 16098 */ | |
| 16099 static int spell_expand_need_cap; | |
| 16100 | |
| 16101 void | |
| 16102 spell_expand_check_cap(col) | |
| 16103 colnr_T col; | |
| 16104 { | |
| 475 | 16105 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col); |
| 16106 } | |
| 16107 | |
| 16108 /* | |
| 16109 * Get list of spelling suggestions. | |
| 16110 * Used for Insert mode completion CTRL-X ?. | |
| 16111 * Returns the number of matches. The matches are in "matchp[]", array of | |
| 16112 * allocated strings. | |
| 16113 */ | |
| 16114 int | |
| 1872 | 16115 expand_spelling(lnum, pat, matchp) |
| 1883 | 16116 linenr_T lnum UNUSED; |
| 475 | 16117 char_u *pat; |
| 16118 char_u ***matchp; | |
| 16119 { | |
| 16120 garray_T ga; | |
| 16121 | |
| 625 | 16122 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap, TRUE); |
| 475 | 16123 *matchp = ga.ga_data; |
| 16124 return ga.ga_len; | |
| 16125 } | |
| 16126 #endif | |
| 16127 | |
| 737 | 16128 #endif /* FEAT_SPELL */ |