Mercurial > vim
annotate src/spell.c @ 17653:cc68aca87c17 v8.1.1824
patch 8.1.1824: crash when correctly spelled word is very long
commit https://github.com/vim/vim/commit/5bcc5a1ff94bbab1b175e35a72e3df974106b393
Author: Bram Moolenaar <Bram@vim.org>
Date: Tue Aug 6 22:48:02 2019 +0200
patch 8.1.1824: crash when correctly spelled word is very long
Problem: Crash when correctly spelled word is very long. (Ben Kraft)
Solution: Check word length before copying. (closes https://github.com/vim/vim/issues/4778)
| author | Bram Moolenaar <Bram@vim.org> |
|---|---|
| date | Tue, 06 Aug 2019 23:00:05 +0200 |
| parents | 26e8d42987ca |
| children | 0f7ae8010787 |
| rev | line source |
|---|---|
|
10042
4aead6a9b7a9
commit https://github.com/vim/vim/commit/edf3f97ae2af024708ebb4ac614227327033ca47
Christian Brabandt <cb@256bit.org>
parents:
9953
diff
changeset
|
1 /* vi:set ts=8 sts=4 sw=4 noet: |
| 223 | 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 * |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
13 * See spellfile.c for the Vim spell file format. |
|
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
14 * |
| 300 | 15 * The spell checking mechanism uses a tree (aka trie). Each node in the tree |
| 16 * has a list of bytes that can appear (siblings). For each byte there is a | |
| 17 * pointer to the node with the byte that follows in the word (child). | |
| 324 | 18 * |
| 19 * A NUL byte is used where the word may end. The bytes are sorted, so that | |
| 20 * binary searching can be used and the NUL bytes are at the start. The | |
| 21 * number of possible bytes is stored before the list of bytes. | |
| 22 * | |
| 23 * The tree uses two arrays: "byts" stores the characters, "idxs" stores | |
| 24 * either the next index or flags. The tree starts at index 0. For example, | |
| 25 * to lookup "vi" this sequence is followed: | |
| 26 * i = 0 | |
| 27 * len = byts[i] | |
| 28 * n = where "v" appears in byts[i + 1] to byts[i + len] | |
| 29 * i = idxs[n] | |
| 30 * len = byts[i] | |
| 31 * n = where "i" appears in byts[i + 1] to byts[i + len] | |
| 32 * i = idxs[n] | |
| 33 * len = byts[i] | |
| 34 * find that byts[i + 1] is 0, idxs[i + 1] has flags for "vi". | |
| 300 | 35 * |
| 339 | 36 * There are two word trees: one with case-folded words and one with words in |
| 300 | 37 * original case. The second one is only used for keep-case words and is |
| 38 * usually small. | |
| 39 * | |
| 481 | 40 * There is one additional tree for when not all prefixes are applied when |
| 339 | 41 * generating the .spl file. This tree stores all the possible prefixes, as |
| 42 * if they were words. At each word (prefix) end the prefix nr is stored, the | |
| 43 * following word must support this prefix nr. And the condition nr is | |
| 44 * stored, used to lookup the condition that the word must match with. | |
| 45 * | |
| 300 | 46 * Thanks to Olaf Seibert for providing an example implementation of this tree |
| 47 * and the compression mechanism. | |
| 625 | 48 * LZ trie ideas: |
| 49 * http://www.irb.hr/hr/home/ristov/papers/RistovLZtrieRevision1.pdf | |
| 50 * More papers: http://www-igm.univ-mlv.fr/~laporte/publi_en.html | |
| 243 | 51 * |
| 52 * Matching involves checking the caps type: Onecap ALLCAP KeepCap. | |
| 53 * | |
| 236 | 54 * Why doesn't Vim use aspell/ispell/myspell/etc.? |
| 55 * See ":help develop-spell". | |
| 56 */ | |
| 57 | |
| 300 | 58 /* |
| 324 | 59 * Use this to adjust the score after finding suggestions, based on the |
| 60 * suggested word sounding like the bad word. This is much faster than doing | |
| 61 * it for every possible suggestion. | |
| 625 | 62 * Disadvantage: When "the" is typed as "hte" it sounds quite different ("@" |
| 63 * vs "ht") and goes down in the list. | |
| 344 | 64 * Used when 'spellsuggest' is set to "best". |
| 65 */ | |
| 66 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4) | |
| 67 | |
| 68 /* | |
| 625 | 69 * Do the opposite: based on a maximum end score and a known sound score, |
| 1705 | 70 * compute the maximum word score that can be used. |
| 625 | 71 */ |
| 72 #define MAXSCORE(word_score, sound_score) ((4 * word_score - sound_score) / 3) | |
| 73 | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
74 #define IN_SPELL_C |
| 223 | 75 #include "vim.h" |
| 76 | |
| 737 | 77 #if defined(FEAT_SPELL) || defined(PROTO) |
| 223 | 78 |
| 625 | 79 #ifndef UNIX /* it's in os_unix.h for Unix */ |
| 80 # include <time.h> /* for time_t */ | |
| 81 #endif | |
| 82 | |
| 633 | 83 /* only used for su_badflags */ |
| 84 #define WF_MIXCAP 0x20 /* mix of upper and lower case: macaRONI */ | |
| 85 | |
| 389 | 86 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP) |
| 300 | 87 |
| 307 | 88 #define REGION_ALL 0xff /* word valid in all regions */ |
| 89 | |
| 625 | 90 #define VIMSUGMAGIC "VIMsug" /* string at start of Vim .sug file */ |
| 91 #define VIMSUGMAGICL 6 | |
| 92 #define VIMSUGVERSION 1 | |
| 93 | |
| 307 | 94 /* Result values. Lower number is accepted over higher one. */ |
| 95 #define SP_BANNED -1 | |
| 236 | 96 #define SP_OK 0 |
| 307 | 97 #define SP_RARE 1 |
| 98 #define SP_LOCAL 2 | |
| 99 #define SP_BAD 3 | |
| 236 | 100 |
| 625 | 101 typedef struct wordcount_S |
| 102 { | |
| 103 short_u wc_count; /* nr of times word was seen */ | |
| 104 char_u wc_word[1]; /* word, actually longer */ | |
| 105 } wordcount_T; | |
| 106 | |
|
9570
695186e11daa
commit https://github.com/vim/vim/commit/840268400dc8fda62a14f8a084e8b1ea46619454
Christian Brabandt <cb@256bit.org>
parents:
9487
diff
changeset
|
107 #define WC_KEY_OFF offsetof(wordcount_T, wc_word) |
| 625 | 108 #define HI2WC(hi) ((wordcount_T *)((hi)->hi_key - WC_KEY_OFF)) |
| 109 #define MAXWORDCOUNT 0xffff | |
| 110 | |
| 236 | 111 /* |
| 323 | 112 * Information used when looking for suggestions. |
| 113 */ | |
| 114 typedef struct suginfo_S | |
| 115 { | |
| 116 garray_T su_ga; /* suggestions, contains "suggest_T" */ | |
| 344 | 117 int su_maxcount; /* max. number of suggestions displayed */ |
| 323 | 118 int su_maxscore; /* maximum score for adding to su_ga */ |
| 625 | 119 int su_sfmaxscore; /* idem, for when doing soundfold words */ |
| 344 | 120 garray_T su_sga; /* like su_ga, sound-folded scoring */ |
| 323 | 121 char_u *su_badptr; /* start of bad word in line */ |
| 122 int su_badlen; /* length of detected bad word in line */ | |
| 346 | 123 int su_badflags; /* caps flags for bad word */ |
| 323 | 124 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */ |
| 125 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */ | |
| 534 | 126 char_u su_sal_badword[MAXWLEN]; /* su_badword soundfolded */ |
| 323 | 127 hashtab_T su_banned; /* table with banned words */ |
| 503 | 128 slang_T *su_sallang; /* default language for sound folding */ |
| 323 | 129 } suginfo_T; |
| 130 | |
| 131 /* One word suggestion. Used in "si_ga". */ | |
| 132 typedef struct suggest_S | |
| 133 { | |
| 134 char_u *st_word; /* suggested word, allocated string */ | |
| 625 | 135 int st_wordlen; /* STRLEN(st_word) */ |
| 323 | 136 int st_orglen; /* length of replaced text */ |
| 137 int st_score; /* lower is better */ | |
| 344 | 138 int st_altscore; /* used when st_score compares equal */ |
| 139 int st_salscore; /* st_score is for soundalike */ | |
| 324 | 140 int st_had_bonus; /* bonus already included in score */ |
| 503 | 141 slang_T *st_slang; /* language used for sound folding */ |
| 323 | 142 } suggest_T; |
| 143 | |
| 344 | 144 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i]) |
| 323 | 145 |
| 625 | 146 /* TRUE if a word appears in the list of banned words. */ |
| 147 #define WAS_BANNED(su, word) (!HASHITEM_EMPTY(hash_find(&su->su_banned, word))) | |
| 148 | |
| 1705 | 149 /* Number of suggestions kept when cleaning up. We need to keep more than |
| 625 | 150 * what is displayed, because when rescore_suggestions() is called the score |
| 151 * may change and wrong suggestions may be removed later. */ | |
| 152 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 130 ? 150 : (su)->su_maxcount + 20) | |
| 324 | 153 |
| 154 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots | |
| 155 * of suggestions that are not going to be displayed. */ | |
| 625 | 156 #define SUG_MAX_COUNT(su) (SUG_CLEAN_COUNT(su) + 50) |
| 323 | 157 |
| 158 /* score for various changes */ | |
| 344 | 159 #define SCORE_SPLIT 149 /* split bad word */ |
| 698 | 160 #define SCORE_SPLIT_NO 249 /* split bad word with NOSPLITSUGS */ |
| 323 | 161 #define SCORE_ICASE 52 /* slightly different case */ |
| 483 | 162 #define SCORE_REGION 200 /* word is for different region */ |
| 323 | 163 #define SCORE_RARE 180 /* rare word */ |
| 625 | 164 #define SCORE_SWAP 75 /* swap two characters */ |
| 323 | 165 #define SCORE_SWAP3 110 /* swap two characters in three */ |
| 532 | 166 #define SCORE_REP 65 /* REP replacement */ |
| 323 | 167 #define SCORE_SUBST 93 /* substitute a character */ |
| 168 #define SCORE_SIMILAR 33 /* substitute a similar character */ | |
| 479 | 169 #define SCORE_SUBCOMP 33 /* substitute a composing character */ |
| 324 | 170 #define SCORE_DEL 94 /* delete a character */ |
| 532 | 171 #define SCORE_DELDUP 66 /* delete a duplicated character */ |
| 479 | 172 #define SCORE_DELCOMP 28 /* delete a composing character */ |
| 324 | 173 #define SCORE_INS 96 /* insert a character */ |
| 532 | 174 #define SCORE_INSDUP 67 /* insert a duplicate character */ |
| 475 | 175 #define SCORE_INSCOMP 30 /* insert a composing character */ |
| 366 | 176 #define SCORE_NONWORD 103 /* change non-word to word char */ |
| 323 | 177 |
| 372 | 178 #define SCORE_FILE 30 /* suggestion from a file */ |
| 323 | 179 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower. |
| 180 * 350 allows for about three changes. */ | |
| 344 | 181 |
| 625 | 182 #define SCORE_COMMON1 30 /* subtracted for words seen before */ |
| 183 #define SCORE_COMMON2 40 /* subtracted for words often seen */ | |
| 184 #define SCORE_COMMON3 50 /* subtracted for words very often seen */ | |
| 185 #define SCORE_THRES2 10 /* word count threshold for COMMON2 */ | |
| 186 #define SCORE_THRES3 100 /* word count threshold for COMMON3 */ | |
| 187 | |
| 188 /* When trying changed soundfold words it becomes slow when trying more than | |
| 189 * two changes. With less then two changes it's slightly faster but we miss a | |
| 190 * few good suggestions. In rare cases we need to try three of four changes. | |
| 191 */ | |
| 192 #define SCORE_SFMAX1 200 /* maximum score for first try */ | |
| 193 #define SCORE_SFMAX2 300 /* maximum score for second try */ | |
| 194 #define SCORE_SFMAX3 400 /* maximum score for third try */ | |
| 195 | |
| 344 | 196 #define SCORE_BIG SCORE_INS * 3 /* big difference */ |
| 625 | 197 #define SCORE_MAXMAX 999999 /* accept any score */ |
| 198 #define SCORE_LIMITMAX 350 /* for spell_edit_score_limit() */ | |
| 199 | |
| 200 /* for spell_edit_score_limit() we need to know the minimum value of | |
| 201 * SCORE_ICASE, SCORE_SWAP, SCORE_DEL, SCORE_SIMILAR and SCORE_INS */ | |
| 202 #define SCORE_EDIT_MIN SCORE_SIMILAR | |
| 323 | 203 |
| 204 /* | |
| 236 | 205 * Structure to store info for word matching. |
| 206 */ | |
| 207 typedef struct matchinf_S | |
| 208 { | |
| 209 langp_T *mi_lp; /* info for language and region */ | |
| 243 | 210 |
| 211 /* pointers to original text to be checked */ | |
| 236 | 212 char_u *mi_word; /* start of word being checked */ |
| 339 | 213 char_u *mi_end; /* end of matching word so far */ |
| 243 | 214 char_u *mi_fend; /* next char to be added to mi_fword */ |
| 300 | 215 char_u *mi_cend; /* char after what was used for |
| 216 mi_capflags */ | |
| 243 | 217 |
| 218 /* case-folded text */ | |
| 219 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */ | |
| 300 | 220 int mi_fwordlen; /* nr of valid bytes in mi_fword */ |
| 243 | 221 |
| 339 | 222 /* for when checking word after a prefix */ |
| 223 int mi_prefarridx; /* index in sl_pidxs with list of | |
| 481 | 224 affixID/condition */ |
| 339 | 225 int mi_prefcnt; /* number of entries at mi_prefarridx */ |
| 226 int mi_prefixlen; /* byte length of prefix */ | |
| 455 | 227 int mi_cprefixlen; /* byte length of prefix in original |
| 228 case */ | |
| 339 | 229 |
| 481 | 230 /* for when checking a compound word */ |
| 231 int mi_compoff; /* start of following word offset */ | |
| 483 | 232 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */ |
| 233 int mi_complen; /* nr of compound words used */ | |
| 809 | 234 int mi_compextra; /* nr of COMPOUNDROOT words */ |
| 481 | 235 |
| 243 | 236 /* others */ |
| 236 | 237 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */ |
| 300 | 238 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
|
239 win_T *mi_win; /* buffer being checked */ |
| 492 | 240 |
| 241 /* for NOBREAK */ | |
| 242 int mi_result2; /* "mi_resul" without following word */ | |
| 243 char_u *mi_end2; /* "mi_end" without following word */ | |
| 236 | 244 } matchinf_T; |
| 245 | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
246 |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
247 static int spell_iswordp(char_u *p, win_T *wp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
248 static int spell_mb_isword_class(int cl, win_T *wp); |
| 307 | 249 |
| 250 /* | |
| 344 | 251 * For finding suggestions: At each node in the tree these states are tried: |
| 330 | 252 */ |
| 253 typedef enum | |
| 254 { | |
| 344 | 255 STATE_START = 0, /* At start of node check for NUL bytes (goodword |
| 256 * ends); if badword ends there is a match, otherwise | |
| 257 * try splitting word. */ | |
| 375 | 258 STATE_NOPREFIX, /* try without prefix */ |
| 344 | 259 STATE_SPLITUNDO, /* Undo splitting. */ |
| 330 | 260 STATE_ENDNUL, /* Past NUL bytes at start of the node. */ |
| 261 STATE_PLAIN, /* Use each byte of the node. */ | |
| 262 STATE_DEL, /* Delete a byte from the bad word. */ | |
| 625 | 263 STATE_INS_PREP, /* Prepare for inserting bytes. */ |
| 330 | 264 STATE_INS, /* Insert a byte in the bad word. */ |
| 265 STATE_SWAP, /* Swap two bytes. */ | |
| 344 | 266 STATE_UNSWAP, /* Undo swap two characters. */ |
| 267 STATE_SWAP3, /* Swap two characters over three. */ | |
| 268 STATE_UNSWAP3, /* Undo Swap two characters over three. */ | |
| 269 STATE_UNROT3L, /* Undo rotate three characters left */ | |
| 270 STATE_UNROT3R, /* Undo rotate three characters right */ | |
| 330 | 271 STATE_REP_INI, /* Prepare for using REP items. */ |
| 272 STATE_REP, /* Use matching REP items from the .aff file. */ | |
| 273 STATE_REP_UNDO, /* Undo a REP item replacement. */ | |
| 274 STATE_FINAL /* End of this node. */ | |
| 275 } state_T; | |
| 276 | |
| 277 /* | |
| 346 | 278 * Struct to keep the state at each level in suggest_try_change(). |
| 323 | 279 */ |
| 280 typedef struct trystate_S | |
| 281 { | |
| 330 | 282 state_T ts_state; /* state at this level, STATE_ */ |
| 323 | 283 int ts_score; /* score */ |
| 344 | 284 idx_T ts_arridx; /* index in tree array, start of node */ |
| 330 | 285 short ts_curi; /* index in list of child nodes */ |
| 286 char_u ts_fidx; /* index in fword[], case-folded bad word */ | |
| 287 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */ | |
| 288 char_u ts_twordlen; /* valid length of tword[] */ | |
| 482 | 289 char_u ts_prefixdepth; /* stack depth for end of prefix or |
| 485 | 290 * PFD_PREFIXTREE or PFD_NOPREFIX */ |
| 291 char_u ts_flags; /* TSF_ flags */ | |
| 330 | 292 char_u ts_tcharlen; /* number of bytes in tword character */ |
| 293 char_u ts_tcharidx; /* current byte index in tword character */ | |
| 294 char_u ts_isdiff; /* DIFF_ values */ | |
| 295 char_u ts_fcharstart; /* index in fword where badword char started */ | |
| 483 | 296 char_u ts_prewordlen; /* length of word in "preword[]" */ |
| 297 char_u ts_splitoff; /* index in "tword" after last split */ | |
| 492 | 298 char_u ts_splitfidx; /* "ts_fidx" at word split */ |
| 483 | 299 char_u ts_complen; /* nr of compound words used */ |
| 485 | 300 char_u ts_compsplit; /* index for "compflags" where word was spit */ |
| 346 | 301 char_u ts_save_badflags; /* su_badflags saved here */ |
| 625 | 302 char_u ts_delidx; /* index in fword for char that was deleted, |
| 303 valid when "ts_flags" has TSF_DIDDEL */ | |
| 323 | 304 } trystate_T; |
| 305 | |
| 330 | 306 /* values for ts_isdiff */ |
| 307 #define DIFF_NONE 0 /* no different byte (yet) */ | |
| 308 #define DIFF_YES 1 /* different byte found */ | |
| 309 #define DIFF_INSERT 2 /* inserting character */ | |
| 310 | |
| 485 | 311 /* values for ts_flags */ |
| 312 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */ | |
| 313 #define TSF_DIDSPLIT 2 /* tried split at this point */ | |
| 625 | 314 #define TSF_DIDDEL 4 /* did a delete, "ts_delidx" has index */ |
| 485 | 315 |
| 375 | 316 /* special values ts_prefixdepth */ |
| 485 | 317 #define PFD_NOPREFIX 0xff /* not using prefixes */ |
| 482 | 318 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */ |
| 625 | 319 #define PFD_NOTSPECIAL 0xfd /* highest value that's not special */ |
| 375 | 320 |
| 339 | 321 /* mode values for find_word */ |
| 481 | 322 #define FIND_FOLDWORD 0 /* find word case-folded */ |
| 323 #define FIND_KEEPWORD 1 /* find keep-case word */ | |
| 324 #define FIND_PREFIX 2 /* find word after prefix */ | |
| 325 #define FIND_COMPOUND 3 /* find case-folded compound word */ | |
| 326 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */ | |
| 339 | 327 |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
328 static void find_word(matchinf_T *mip, int mode); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
329 static int match_checkcompoundpattern(char_u *ptr, int wlen, garray_T *gap); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
330 static int can_compound(slang_T *slang, char_u *word, char_u *flags); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
331 static int match_compoundrule(slang_T *slang, char_u *compflags); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
332 static int valid_word_prefix(int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
333 static void find_prefix(matchinf_T *mip, int mode); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
334 static int fold_more(matchinf_T *mip); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
335 static int spell_valid_case(int wordflags, int treeflags); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
336 static void spell_load_cb(char_u *fname, void *cookie); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
337 static int count_syllables(slang_T *slang, char_u *word); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
338 static void clear_midword(win_T *buf); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
339 static void use_midword(slang_T *lp, win_T *buf); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
340 static int find_region(char_u *rp, char_u *region); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
341 static int check_need_cap(linenr_T lnum, colnr_T col); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
342 static void spell_find_suggest(char_u *badptr, int badlen, suginfo_T *su, int maxcount, int banbadword, int need_cap, int interactive); |
| 372 | 343 #ifdef FEAT_EVAL |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
344 static void spell_suggest_expr(suginfo_T *su, char_u *expr); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
345 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
346 static void spell_suggest_file(suginfo_T *su, char_u *fname); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
347 static void spell_suggest_intern(suginfo_T *su, int interactive); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
348 static void spell_find_cleanup(suginfo_T *su); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
349 static void suggest_try_special(suginfo_T *su); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
350 static void suggest_try_change(suginfo_T *su); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
351 static void suggest_trie_walk(suginfo_T *su, langp_T *lp, char_u *fword, int soundfold); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
352 static void go_deeper(trystate_T *stack, int depth, int score_add); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
353 static int nofold_len(char_u *fword, int flen, char_u *word); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
354 static void find_keepcap_word(slang_T *slang, char_u *fword, char_u *kword); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
355 static void score_comp_sal(suginfo_T *su); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
356 static void score_combine(suginfo_T *su); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
357 static int stp_sal_score(suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
358 static void suggest_try_soundalike_prep(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
359 static void suggest_try_soundalike(suginfo_T *su); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
360 static void suggest_try_soundalike_finish(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
361 static void add_sound_suggest(suginfo_T *su, char_u *goodword, int score, langp_T *lp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
362 static int soundfold_find(slang_T *slang, char_u *word); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
363 static void make_case_word(char_u *fword, char_u *cword, int flags); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
364 static int similar_chars(slang_T *slang, int c1, int c2); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
365 static void add_suggestion(suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus, slang_T *slang, int maxsf); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
366 static void check_suggestions(suginfo_T *su, garray_T *gap); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
367 static void add_banned(suginfo_T *su, char_u *word); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
368 static void rescore_suggestions(suginfo_T *su); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
369 static void rescore_one(suginfo_T *su, suggest_T *stp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
370 static int cleanup_suggestions(garray_T *gap, int maxscore, int keep); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
371 static void spell_soundfold_sofo(slang_T *slang, char_u *inword, char_u *res); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
372 static void spell_soundfold_sal(slang_T *slang, char_u *inword, char_u *res); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
373 static void spell_soundfold_wsal(slang_T *slang, char_u *inword, char_u *res); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
374 static int soundalike_score(char_u *goodsound, char_u *badsound); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
375 static int spell_edit_score(slang_T *slang, char_u *badword, char_u *goodword); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
376 static int spell_edit_score_limit(slang_T *slang, char_u *badword, char_u *goodword, int limit); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
377 static int spell_edit_score_limit_w(slang_T *slang, char_u *badword, char_u *goodword, int limit); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
378 static void dump_word(slang_T *slang, char_u *word, char_u *pat, int *dir, int round, int flags, linenr_T lnum); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
379 static linenr_T dump_prefixes(slang_T *slang, char_u *word, char_u *pat, int *dir, int round, int flags, linenr_T startlnum); |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
380 |
| 236 | 381 |
| 626 | 382 /* Remember what "z?" replaced. */ |
| 383 static char_u *repl_from = NULL; | |
| 384 static char_u *repl_to = NULL; | |
| 385 | |
| 236 | 386 /* |
| 387 * Main spell-checking function. | |
| 300 | 388 * "ptr" points to a character that could be the start of a word. |
| 534 | 389 * "*attrp" is set to the highlight index for a badly spelled word. For a |
| 390 * non-word or when it's OK it remains unchanged. | |
| 236 | 391 * This must only be called when 'spelllang' is not empty. |
| 323 | 392 * |
| 385 | 393 * "capcol" is used to check for a Capitalised word after the end of a |
| 394 * sentence. If it's zero then perform the check. Return the column where to | |
| 395 * check next, or -1 when no sentence end was found. If it's NULL then don't | |
| 396 * worry. | |
| 323 | 397 * |
| 236 | 398 * Returns the length of the word in bytes, also when it's OK, so that the |
| 399 * caller can skip over the word. | |
| 400 */ | |
| 401 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
402 spell_check( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
403 win_T *wp, /* current window */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
404 char_u *ptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
405 hlf_T *attrp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
406 int *capcol, /* column to check for Capital */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
407 int docount) /* count good words */ |
| 236 | 408 { |
| 409 matchinf_T mi; /* Most things are put in "mi" so that it can | |
| 410 be passed to functions quickly. */ | |
| 344 | 411 int nrlen = 0; /* found a number first */ |
| 385 | 412 int c; |
| 483 | 413 int wrongcaplen = 0; |
| 500 | 414 int lpi; |
| 625 | 415 int count_word = docount; |
| 236 | 416 |
| 307 | 417 /* A word never starts at a space or a control character. Return quickly |
| 418 * then, skipping over the character. */ | |
| 419 if (*ptr <= ' ') | |
| 420 return 1; | |
| 690 | 421 |
| 422 /* 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
|
423 if (wp->w_s->b_langp.ga_len == 0) |
| 690 | 424 return 1; |
| 425 | |
| 483 | 426 vim_memset(&mi, 0, sizeof(matchinf_T)); |
| 236 | 427 |
| 344 | 428 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and |
| 584 | 429 * 0X99FF. But always do check spelling to find "3GPP" and "11 |
| 430 * julifeest". */ | |
| 300 | 431 if (*ptr >= '0' && *ptr <= '9') |
| 432 { | |
|
7447
ad432f8f68fb
commit https://github.com/vim/vim/commit/887c1fea4a114e7170091942d0446c8882701b5b
Christian Brabandt <cb@256bit.org>
parents:
6949
diff
changeset
|
433 if (*ptr == '0' && (ptr[1] == 'b' || ptr[1] == 'B')) |
|
ad432f8f68fb
commit https://github.com/vim/vim/commit/887c1fea4a114e7170091942d0446c8882701b5b
Christian Brabandt <cb@256bit.org>
parents:
6949
diff
changeset
|
434 mi.mi_end = skipbin(ptr + 2); |
|
ad432f8f68fb
commit https://github.com/vim/vim/commit/887c1fea4a114e7170091942d0446c8882701b5b
Christian Brabandt <cb@256bit.org>
parents:
6949
diff
changeset
|
435 else if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X')) |
| 316 | 436 mi.mi_end = skiphex(ptr + 2); |
| 300 | 437 else |
| 344 | 438 mi.mi_end = skipdigits(ptr); |
| 835 | 439 nrlen = (int)(mi.mi_end - ptr); |
| 584 | 440 } |
| 346 | 441 |
| 442 /* Find the normal end of the word (until the next non-word character). */ | |
| 344 | 443 mi.mi_word = ptr; |
| 584 | 444 mi.mi_fend = ptr; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
445 if (spell_iswordp(mi.mi_fend, wp)) |
| 344 | 446 { |
| 447 do | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
448 MB_PTR_ADV(mi.mi_fend); |
|
16162
cd5c83115ec6
patch 8.1.1086: too many curly braces
Bram Moolenaar <Bram@vim.org>
parents:
16142
diff
changeset
|
449 while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp)); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
450 |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
451 if (capcol != NULL && *capcol == 0 && wp->w_s->b_cap_prog != NULL) |
| 385 | 452 { |
| 453 /* Check word starting with capital letter. */ | |
| 455 | 454 c = PTR2CHAR(ptr); |
| 385 | 455 if (!SPELL_ISUPPER(c)) |
| 483 | 456 wrongcaplen = (int)(mi.mi_fend - ptr); |
| 385 | 457 } |
| 458 } | |
| 459 if (capcol != NULL) | |
| 460 *capcol = -1; | |
| 344 | 461 |
| 462 /* We always use the characters up to the next non-word character, | |
| 463 * also for bad words. */ | |
| 464 mi.mi_end = mi.mi_fend; | |
| 465 | |
| 466 /* Check caps type later. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
467 mi.mi_capflags = 0; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
468 mi.mi_cend = NULL; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
469 mi.mi_win = wp; |
| 344 | 470 |
| 483 | 471 /* case-fold the word with one non-word character, so that we can check |
| 472 * for the word end. */ | |
| 344 | 473 if (*mi.mi_fend != NUL) |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
474 MB_PTR_ADV(mi.mi_fend); |
| 344 | 475 |
| 476 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword, | |
| 477 MAXWLEN + 1); | |
| 835 | 478 mi.mi_fwordlen = (int)STRLEN(mi.mi_fword); |
| 344 | 479 |
| 480 /* The word is bad unless we recognize it. */ | |
| 481 mi.mi_result = SP_BAD; | |
| 492 | 482 mi.mi_result2 = SP_BAD; |
| 344 | 483 |
| 484 /* | |
| 485 * Loop over the languages specified in 'spelllang'. | |
| 625 | 486 * We check them all, because a word may be matched longer in another |
| 487 * language. | |
| 344 | 488 */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
489 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
|
490 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
491 mi.mi_lp = LANGP_ENTRY(wp->w_s->b_langp, lpi); |
| 500 | 492 |
| 493 /* If reloading fails the language is still in the list but everything | |
| 494 * has been cleared. */ | |
| 495 if (mi.mi_lp->lp_slang->sl_fidxs == NULL) | |
| 496 continue; | |
| 497 | |
| 344 | 498 /* Check for a matching word in case-folded words. */ |
| 499 find_word(&mi, FIND_FOLDWORD); | |
| 500 | |
| 501 /* Check for a matching word in keep-case words. */ | |
| 502 find_word(&mi, FIND_KEEPWORD); | |
| 503 | |
| 504 /* Check for matching prefixes. */ | |
| 485 | 505 find_prefix(&mi, FIND_FOLDWORD); |
| 492 | 506 |
| 507 /* For a NOBREAK language, may want to use a word without a following | |
| 508 * word as a backup. */ | |
| 509 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD | |
| 510 && mi.mi_result2 != SP_BAD) | |
| 511 { | |
| 512 mi.mi_result = mi.mi_result2; | |
| 513 mi.mi_end = mi.mi_end2; | |
| 514 } | |
| 625 | 515 |
| 516 /* Count the word in the first language where it's found to be OK. */ | |
| 517 if (count_word && mi.mi_result == SP_OK) | |
| 518 { | |
| 519 count_common_word(mi.mi_lp->lp_slang, ptr, | |
| 520 (int)(mi.mi_end - ptr), 1); | |
| 521 count_word = FALSE; | |
| 522 } | |
| 344 | 523 } |
| 524 | |
| 525 if (mi.mi_result != SP_OK) | |
| 526 { | |
| 346 | 527 /* If we found a number skip over it. Allows for "42nd". Do flag |
| 528 * rare and local words, e.g., "3GPP". */ | |
| 344 | 529 if (nrlen > 0) |
| 346 | 530 { |
| 531 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED) | |
| 532 return nrlen; | |
| 533 } | |
| 344 | 534 |
| 535 /* When we are at a non-word character there is no error, just | |
| 536 * skip over the character (try looking for a word after it). */ | |
| 5477 | 537 else if (!spell_iswordp_nmw(ptr, wp)) |
| 243 | 538 { |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
539 if (capcol != NULL && wp->w_s->b_cap_prog != NULL) |
| 385 | 540 { |
| 541 regmatch_T regmatch; | |
| 6375 | 542 int r; |
| 385 | 543 |
| 544 /* Check for end of sentence. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
545 regmatch.regprog = wp->w_s->b_cap_prog; |
| 385 | 546 regmatch.rm_ic = FALSE; |
| 6375 | 547 r = vim_regexec(®match, ptr, 0); |
| 548 wp->w_s->b_cap_prog = regmatch.regprog; | |
| 549 if (r) | |
| 385 | 550 *capcol = (int)(regmatch.endp[0] - ptr); |
| 551 } | |
| 552 | |
| 344 | 553 if (has_mbyte) |
| 474 | 554 return (*mb_ptr2len)(ptr); |
| 344 | 555 return 1; |
| 300 | 556 } |
| 483 | 557 else if (mi.mi_end == ptr) |
| 558 /* Always include at least one character. Required for when there | |
| 559 * is a mixup in "midword". */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
560 MB_PTR_ADV(mi.mi_end); |
| 492 | 561 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
|
562 && LANGP_ENTRY(wp->w_s->b_langp, 0)->lp_slang->sl_nobreak) |
| 492 | 563 { |
| 564 char_u *p, *fp; | |
| 565 int save_result = mi.mi_result; | |
| 566 | |
| 567 /* First language in 'spelllang' is NOBREAK. Find first position | |
| 568 * 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
|
569 mi.mi_lp = LANGP_ENTRY(wp->w_s->b_langp, 0); |
| 500 | 570 if (mi.mi_lp->lp_slang->sl_fidxs != NULL) |
| 571 { | |
| 572 p = mi.mi_word; | |
| 573 fp = mi.mi_fword; | |
| 574 for (;;) | |
| 575 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
576 MB_PTR_ADV(p); |
|
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
577 MB_PTR_ADV(fp); |
| 500 | 578 if (p >= mi.mi_end) |
| 579 break; | |
| 835 | 580 mi.mi_compoff = (int)(fp - mi.mi_fword); |
| 500 | 581 find_word(&mi, FIND_COMPOUND); |
| 582 if (mi.mi_result != SP_BAD) | |
| 583 { | |
| 584 mi.mi_end = p; | |
| 585 break; | |
| 586 } | |
| 587 } | |
| 588 mi.mi_result = save_result; | |
| 589 } | |
| 492 | 590 } |
| 243 | 591 |
| 344 | 592 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED) |
| 534 | 593 *attrp = HLF_SPB; |
| 344 | 594 else if (mi.mi_result == SP_RARE) |
| 534 | 595 *attrp = HLF_SPR; |
| 344 | 596 else |
| 534 | 597 *attrp = HLF_SPL; |
| 243 | 598 } |
| 599 | |
| 483 | 600 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE)) |
| 601 { | |
| 602 /* Report SpellCap only when the word isn't badly spelled. */ | |
| 534 | 603 *attrp = HLF_SPC; |
| 483 | 604 return wrongcaplen; |
| 605 } | |
| 606 | |
| 300 | 607 return (int)(mi.mi_end - ptr); |
| 236 | 608 } |
| 609 | |
| 610 /* | |
| 300 | 611 * Check if the word at "mip->mi_word" is in the tree. |
| 339 | 612 * When "mode" is FIND_FOLDWORD check in fold-case word tree. |
| 613 * When "mode" is FIND_KEEPWORD check in keep-case word tree. | |
| 614 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word | |
| 615 * tree. | |
| 300 | 616 * |
| 617 * For a match mip->mi_result is updated. | |
| 243 | 618 */ |
| 619 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
620 find_word(matchinf_T *mip, int mode) |
| 243 | 621 { |
| 324 | 622 idx_T arridx = 0; |
| 300 | 623 int endlen[MAXWLEN]; /* length at possible word endings */ |
| 324 | 624 idx_T endidx[MAXWLEN]; /* possible word endings */ |
| 300 | 625 int endidxcnt = 0; |
| 626 int len; | |
| 627 int wlen = 0; | |
| 628 int flen; | |
| 629 int c; | |
| 630 char_u *ptr; | |
| 324 | 631 idx_T lo, hi, m; |
| 300 | 632 char_u *s; |
| 339 | 633 char_u *p; |
| 307 | 634 int res = SP_BAD; |
| 300 | 635 slang_T *slang = mip->mi_lp->lp_slang; |
| 636 unsigned flags; | |
| 637 char_u *byts; | |
| 324 | 638 idx_T *idxs; |
| 481 | 639 int word_ends; |
| 485 | 640 int prefix_found; |
| 492 | 641 int nobreak_result; |
| 481 | 642 |
| 643 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND) | |
| 236 | 644 { |
| 300 | 645 /* Check for word with matching case in keep-case tree. */ |
| 646 ptr = mip->mi_word; | |
| 647 flen = 9999; /* no case folding, always enough bytes */ | |
| 648 byts = slang->sl_kbyts; | |
| 649 idxs = slang->sl_kidxs; | |
| 481 | 650 |
| 651 if (mode == FIND_KEEPCOMPOUND) | |
| 652 /* Skip over the previously found word(s). */ | |
| 653 wlen += mip->mi_compoff; | |
| 236 | 654 } |
| 655 else | |
| 656 { | |
| 300 | 657 /* Check for case-folded in case-folded tree. */ |
| 658 ptr = mip->mi_fword; | |
| 659 flen = mip->mi_fwordlen; /* available case-folded bytes */ | |
| 660 byts = slang->sl_fbyts; | |
| 661 idxs = slang->sl_fidxs; | |
| 339 | 662 |
| 663 if (mode == FIND_PREFIX) | |
| 664 { | |
| 665 /* Skip over the prefix. */ | |
| 666 wlen = mip->mi_prefixlen; | |
| 667 flen -= mip->mi_prefixlen; | |
| 668 } | |
| 481 | 669 else if (mode == FIND_COMPOUND) |
| 670 { | |
| 671 /* Skip over the previously found word(s). */ | |
| 672 wlen = mip->mi_compoff; | |
| 673 flen -= mip->mi_compoff; | |
| 674 } | |
| 675 | |
| 243 | 676 } |
| 677 | |
| 300 | 678 if (byts == NULL) |
| 679 return; /* array is empty */ | |
| 236 | 680 |
| 681 /* | |
| 307 | 682 * Repeat advancing in the tree until: |
| 683 * - there is a byte that doesn't match, | |
| 684 * - we reach the end of the tree, | |
| 685 * - or we reach the end of the line. | |
| 236 | 686 */ |
| 300 | 687 for (;;) |
| 236 | 688 { |
| 346 | 689 if (flen <= 0 && *mip->mi_fend != NUL) |
| 339 | 690 flen = fold_more(mip); |
| 300 | 691 |
| 692 len = byts[arridx++]; | |
| 693 | |
| 694 /* If the first possible byte is a zero the word could end here. | |
| 695 * Remember this index, we first check for the longest word. */ | |
| 696 if (byts[arridx] == 0) | |
| 697 { | |
| 307 | 698 if (endidxcnt == MAXWLEN) |
| 699 { | |
| 700 /* Must be a corrupted spell file. */ | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
701 emsg(_(e_format)); |
| 307 | 702 return; |
| 703 } | |
| 300 | 704 endlen[endidxcnt] = wlen; |
| 705 endidx[endidxcnt++] = arridx++; | |
| 706 --len; | |
| 707 | |
| 708 /* Skip over the zeros, there can be several flag/region | |
| 709 * combinations. */ | |
| 710 while (len > 0 && byts[arridx] == 0) | |
| 711 { | |
| 712 ++arridx; | |
| 713 --len; | |
| 714 } | |
| 715 if (len == 0) | |
| 716 break; /* no children, word must end here */ | |
| 717 } | |
| 718 | |
| 719 /* Stop looking at end of the line. */ | |
| 720 if (ptr[wlen] == NUL) | |
| 721 break; | |
| 722 | |
| 723 /* Perform a binary search in the list of accepted bytes. */ | |
| 724 c = ptr[wlen]; | |
| 346 | 725 if (c == TAB) /* <Tab> is handled like <Space> */ |
| 726 c = ' '; | |
| 300 | 727 lo = arridx; |
| 728 hi = arridx + len - 1; | |
| 729 while (lo < hi) | |
| 730 { | |
| 731 m = (lo + hi) / 2; | |
| 732 if (byts[m] > c) | |
| 733 hi = m - 1; | |
| 734 else if (byts[m] < c) | |
| 735 lo = m + 1; | |
| 736 else | |
| 737 { | |
| 738 lo = hi = m; | |
| 739 break; | |
| 236 | 740 } |
| 741 } | |
| 300 | 742 |
| 743 /* Stop if there is no matching byte. */ | |
| 744 if (hi < lo || byts[lo] != c) | |
| 745 break; | |
| 746 | |
| 747 /* Continue at the child (if there is one). */ | |
| 748 arridx = idxs[lo]; | |
| 749 ++wlen; | |
| 750 --flen; | |
| 346 | 751 |
| 752 /* One space in the good word may stand for several spaces in the | |
| 753 * checked word. */ | |
| 754 if (c == ' ') | |
| 755 { | |
| 756 for (;;) | |
| 757 { | |
| 758 if (flen <= 0 && *mip->mi_fend != NUL) | |
| 759 flen = fold_more(mip); | |
| 760 if (ptr[wlen] != ' ' && ptr[wlen] != TAB) | |
| 761 break; | |
| 762 ++wlen; | |
| 763 --flen; | |
| 764 } | |
| 765 } | |
| 236 | 766 } |
| 767 | |
| 300 | 768 /* |
| 769 * Verify that one of the possible endings is valid. Try the longest | |
| 770 * first. | |
| 771 */ | |
| 772 while (endidxcnt > 0) | |
| 773 { | |
| 774 --endidxcnt; | |
| 775 arridx = endidx[endidxcnt]; | |
| 776 wlen = endlen[endidxcnt]; | |
| 236 | 777 |
| 300 | 778 if ((*mb_head_off)(ptr, ptr + wlen) > 0) |
| 779 continue; /* not at first byte of character */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
780 if (spell_iswordp(ptr + wlen, mip->mi_win)) |
| 481 | 781 { |
| 492 | 782 if (slang->sl_compprog == NULL && !slang->sl_nobreak) |
| 481 | 783 continue; /* next char is a word character */ |
| 784 word_ends = FALSE; | |
| 785 } | |
| 786 else | |
| 787 word_ends = TRUE; | |
| 485 | 788 /* The prefix flag is before compound flags. Once a valid prefix flag |
| 789 * has been found we try compound flags. */ | |
| 790 prefix_found = FALSE; | |
| 300 | 791 |
| 339 | 792 if (mode != FIND_KEEPWORD && has_mbyte) |
| 300 | 793 { |
| 794 /* Compute byte length in original word, length may change | |
| 339 | 795 * when folding case. This can be slow, take a shortcut when the |
| 796 * case-folded word is equal to the keep-case word. */ | |
| 300 | 797 p = mip->mi_word; |
| 339 | 798 if (STRNCMP(ptr, p, wlen) != 0) |
| 799 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
800 for (s = ptr; s < ptr + wlen; MB_PTR_ADV(s)) |
|
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
801 MB_PTR_ADV(p); |
| 835 | 802 wlen = (int)(p - mip->mi_word); |
| 339 | 803 } |
| 300 | 804 } |
| 236 | 805 |
| 339 | 806 /* Check flags and region. For FIND_PREFIX check the condition and |
| 807 * prefix ID. | |
| 808 * Repeat this if there are more flags/region alternatives until there | |
| 809 * is a match. */ | |
| 810 res = SP_BAD; | |
| 811 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0; | |
| 812 --len, ++arridx) | |
| 300 | 813 { |
| 814 flags = idxs[arridx]; | |
| 324 | 815 |
| 339 | 816 /* For the fold-case tree check that the case of the checked word |
| 817 * matches with what the word in the tree requires. | |
| 818 * For keep-case tree the case is always right. For prefixes we | |
| 819 * don't bother to check. */ | |
| 820 if (mode == FIND_FOLDWORD) | |
| 300 | 821 { |
| 822 if (mip->mi_cend != mip->mi_word + wlen) | |
| 823 { | |
| 323 | 824 /* mi_capflags was set for a different word length, need |
| 825 * to do it again. */ | |
| 300 | 826 mip->mi_cend = mip->mi_word + wlen; |
| 323 | 827 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend); |
| 300 | 828 } |
| 829 | |
| 346 | 830 if (mip->mi_capflags == WF_KEEPCAP |
| 831 || !spell_valid_case(mip->mi_capflags, flags)) | |
| 339 | 832 continue; |
| 300 | 833 } |
| 236 | 834 |
| 339 | 835 /* When mode is FIND_PREFIX the word must support the prefix: |
| 836 * check the prefix ID and the condition. Do that for the list at | |
| 366 | 837 * mip->mi_prefarridx that find_prefix() filled. */ |
| 485 | 838 else if (mode == FIND_PREFIX && !prefix_found) |
| 481 | 839 { |
| 366 | 840 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx, |
| 390 | 841 flags, |
| 455 | 842 mip->mi_word + mip->mi_cprefixlen, slang, |
| 843 FALSE); | |
| 366 | 844 if (c == 0) |
| 339 | 845 continue; |
| 366 | 846 |
| 847 /* Use the WF_RARE flag for a rare prefix. */ | |
| 848 if (c & WF_RAREPFX) | |
| 849 flags |= WF_RARE; | |
| 485 | 850 prefix_found = TRUE; |
| 339 | 851 } |
| 852 | |
| 492 | 853 if (slang->sl_nobreak) |
| 854 { | |
| 855 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND) | |
| 856 && (flags & WF_BANNED) == 0) | |
| 857 { | |
| 858 /* NOBREAK: found a valid following word. That's all we | |
| 859 * need to know, so return. */ | |
| 860 mip->mi_result = SP_OK; | |
| 861 break; | |
| 862 } | |
| 863 } | |
| 864 | |
| 865 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND | |
| 866 || !word_ends)) | |
| 481 | 867 { |
| 974 | 868 /* If there is no compound flag or the word is shorter than |
| 483 | 869 * COMPOUNDMIN reject it quickly. |
| 870 * Makes you wonder why someone puts a compound flag on a word | |
| 481 | 871 * that's too short... Myspell compatibility requires this |
| 872 * anyway. */ | |
| 490 | 873 if (((unsigned)flags >> 24) == 0 |
| 874 || wlen - mip->mi_compoff < slang->sl_compminlen) | |
| 483 | 875 continue; |
| 500 | 876 /* For multi-byte chars check character length against |
| 877 * COMPOUNDMIN. */ | |
| 878 if (has_mbyte | |
| 501 | 879 && slang->sl_compminlen > 0 |
| 500 | 880 && mb_charlen_len(mip->mi_word + mip->mi_compoff, |
| 881 wlen - mip->mi_compoff) < slang->sl_compminlen) | |
| 882 continue; | |
| 483 | 883 |
| 809 | 884 /* Limit the number of compound words to COMPOUNDWORDMAX if no |
| 490 | 885 * maximum for syllables is specified. */ |
| 809 | 886 if (!word_ends && mip->mi_complen + mip->mi_compextra + 2 |
| 887 > slang->sl_compmax | |
| 490 | 888 && slang->sl_compsylmax == MAXWLEN) |
| 483 | 889 continue; |
| 890 | |
| 819 | 891 /* Don't allow compounding on a side where an affix was added, |
| 892 * unless COMPOUNDPERMITFLAG was used. */ | |
| 893 if (mip->mi_complen > 0 && (flags & WF_NOCOMPBEF)) | |
| 894 continue; | |
| 895 if (!word_ends && (flags & WF_NOCOMPAFT)) | |
| 896 continue; | |
| 897 | |
| 485 | 898 /* Quickly check if compounding is possible with this flag. */ |
| 495 | 899 if (!byte_in_str(mip->mi_complen == 0 |
| 485 | 900 ? slang->sl_compstartflags |
| 901 : slang->sl_compallflags, | |
| 495 | 902 ((unsigned)flags >> 24))) |
| 481 | 903 continue; |
| 904 | |
| 1762 | 905 /* If there is a match with a CHECKCOMPOUNDPATTERN rule |
| 906 * discard the compound word. */ | |
| 907 if (match_checkcompoundpattern(ptr, wlen, &slang->sl_comppat)) | |
| 908 continue; | |
| 909 | |
| 490 | 910 if (mode == FIND_COMPOUND) |
| 911 { | |
| 912 int capflags; | |
| 913 | |
| 914 /* Need to check the caps type of the appended compound | |
| 915 * word. */ | |
| 916 if (has_mbyte && STRNCMP(ptr, mip->mi_word, | |
| 917 mip->mi_compoff) != 0) | |
| 918 { | |
| 919 /* case folding may have changed the length */ | |
| 920 p = mip->mi_word; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
921 for (s = ptr; s < ptr + mip->mi_compoff; MB_PTR_ADV(s)) |
|
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
922 MB_PTR_ADV(p); |
| 490 | 923 } |
| 924 else | |
| 925 p = mip->mi_word + mip->mi_compoff; | |
| 926 capflags = captype(p, mip->mi_word + wlen); | |
| 927 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP | |
| 928 && (flags & WF_FIXCAP) != 0)) | |
| 929 continue; | |
| 930 | |
| 931 if (capflags != WF_ALLCAP) | |
| 932 { | |
| 933 /* When the character before the word is a word | |
| 934 * character we do not accept a Onecap word. We do | |
| 935 * accept a no-caps word, even when the dictionary | |
| 936 * word specifies ONECAP. */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
937 MB_PTR_BACK(mip->mi_word, p); |
| 5477 | 938 if (spell_iswordp_nmw(p, mip->mi_win) |
| 490 | 939 ? capflags == WF_ONECAP |
| 940 : (flags & WF_ONECAP) != 0 | |
| 941 && capflags != WF_ONECAP) | |
| 942 continue; | |
| 943 } | |
| 944 } | |
| 945 | |
| 483 | 946 /* If the word ends the sequence of compound flags of the |
| 714 | 947 * words must match with one of the COMPOUNDRULE items and |
| 483 | 948 * the number of syllables must not be too large. */ |
| 949 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24); | |
| 950 mip->mi_compflags[mip->mi_complen + 1] = NUL; | |
| 951 if (word_ends) | |
| 952 { | |
| 953 char_u fword[MAXWLEN]; | |
| 954 | |
| 955 if (slang->sl_compsylmax < MAXWLEN) | |
| 956 { | |
| 957 /* "fword" is only needed for checking syllables. */ | |
| 958 if (ptr == mip->mi_word) | |
| 959 (void)spell_casefold(ptr, wlen, fword, MAXWLEN); | |
| 960 else | |
| 961 vim_strncpy(fword, ptr, endlen[endidxcnt]); | |
| 962 } | |
| 963 if (!can_compound(slang, fword, mip->mi_compflags)) | |
| 964 continue; | |
| 965 } | |
| 1762 | 966 else if (slang->sl_comprules != NULL |
| 967 && !match_compoundrule(slang, mip->mi_compflags)) | |
| 968 /* The compound flags collected so far do not match any | |
| 969 * COMPOUNDRULE, discard the compounded word. */ | |
| 970 continue; | |
| 481 | 971 } |
| 972 | |
| 500 | 973 /* Check NEEDCOMPOUND: can't use word without compounding. */ |
| 974 else if (flags & WF_NEEDCOMP) | |
| 975 continue; | |
| 976 | |
| 492 | 977 nobreak_result = SP_OK; |
| 978 | |
| 481 | 979 if (!word_ends) |
| 980 { | |
| 492 | 981 int save_result = mip->mi_result; |
| 982 char_u *save_end = mip->mi_end; | |
| 501 | 983 langp_T *save_lp = mip->mi_lp; |
| 984 int lpi; | |
| 492 | 985 |
| 986 /* Check that a valid word follows. If there is one and we | |
| 987 * are compounding, it will set "mi_result", thus we are | |
| 988 * always finished here. For NOBREAK we only check that a | |
| 989 * valid word follows. | |
| 481 | 990 * Recursive! */ |
| 492 | 991 if (slang->sl_nobreak) |
| 992 mip->mi_result = SP_BAD; | |
| 481 | 993 |
| 994 /* Find following word in case-folded tree. */ | |
| 995 mip->mi_compoff = endlen[endidxcnt]; | |
| 996 if (has_mbyte && mode == FIND_KEEPWORD) | |
| 997 { | |
| 998 /* Compute byte length in case-folded word from "wlen": | |
| 999 * byte length in keep-case word. Length may change when | |
| 1000 * folding case. This can be slow, take a shortcut when | |
| 1001 * the case-folded word is equal to the keep-case word. */ | |
| 1002 p = mip->mi_fword; | |
| 1003 if (STRNCMP(ptr, p, wlen) != 0) | |
| 1004 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
1005 for (s = ptr; s < ptr + wlen; MB_PTR_ADV(s)) |
|
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
1006 MB_PTR_ADV(p); |
| 835 | 1007 mip->mi_compoff = (int)(p - mip->mi_fword); |
| 481 | 1008 } |
| 1009 } | |
|
8956
d9e671c5afe6
commit https://github.com/vim/vim/commit/ba53435144f46eaaa53c63a62e748b3feee9742c
Christian Brabandt <cb@256bit.org>
parents:
8524
diff
changeset
|
1010 #if 0 /* Disabled, see below */ |
| 485 | 1011 c = mip->mi_compoff; |
|
8956
d9e671c5afe6
commit https://github.com/vim/vim/commit/ba53435144f46eaaa53c63a62e748b3feee9742c
Christian Brabandt <cb@256bit.org>
parents:
8524
diff
changeset
|
1012 #endif |
| 483 | 1013 ++mip->mi_complen; |
| 809 | 1014 if (flags & WF_COMPROOT) |
| 1015 ++mip->mi_compextra; | |
| 501 | 1016 |
| 1017 /* For NOBREAK we need to try all NOBREAK languages, at least | |
| 1018 * to find the ".add" file(s). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1019 for (lpi = 0; lpi < mip->mi_win->w_s->b_langp.ga_len; ++lpi) |
| 501 | 1020 { |
| 1021 if (slang->sl_nobreak) | |
| 1022 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1023 mip->mi_lp = LANGP_ENTRY(mip->mi_win->w_s->b_langp, lpi); |
| 501 | 1024 if (mip->mi_lp->lp_slang->sl_fidxs == NULL |
| 1025 || !mip->mi_lp->lp_slang->sl_nobreak) | |
| 1026 continue; | |
| 1027 } | |
| 1028 | |
| 1029 find_word(mip, FIND_COMPOUND); | |
| 1030 | |
| 1031 /* When NOBREAK any word that matches is OK. Otherwise we | |
| 1032 * need to find the longest match, thus try with keep-case | |
| 1033 * and prefix too. */ | |
| 492 | 1034 if (!slang->sl_nobreak || mip->mi_result == SP_BAD) |
| 1035 { | |
| 501 | 1036 /* Find following word in keep-case tree. */ |
| 1037 mip->mi_compoff = wlen; | |
| 1038 find_word(mip, FIND_KEEPCOMPOUND); | |
| 1039 | |
| 819 | 1040 #if 0 /* Disabled, a prefix must not appear halfway a compound word, |
| 1041 unless the COMPOUNDPERMITFLAG is used and then it can't be a | |
| 1042 postponed prefix. */ | |
| 501 | 1043 if (!slang->sl_nobreak || mip->mi_result == SP_BAD) |
| 1044 { | |
| 1045 /* Check for following word with prefix. */ | |
| 1046 mip->mi_compoff = c; | |
| 1047 find_prefix(mip, FIND_COMPOUND); | |
| 1048 } | |
| 819 | 1049 #endif |
| 492 | 1050 } |
| 501 | 1051 |
| 1052 if (!slang->sl_nobreak) | |
| 1053 break; | |
| 492 | 1054 } |
| 483 | 1055 --mip->mi_complen; |
| 809 | 1056 if (flags & WF_COMPROOT) |
| 1057 --mip->mi_compextra; | |
| 501 | 1058 mip->mi_lp = save_lp; |
| 485 | 1059 |
| 492 | 1060 if (slang->sl_nobreak) |
| 1061 { | |
| 1062 nobreak_result = mip->mi_result; | |
| 1063 mip->mi_result = save_result; | |
| 1064 mip->mi_end = save_end; | |
| 1065 } | |
| 1066 else | |
| 1067 { | |
| 1068 if (mip->mi_result == SP_OK) | |
| 1069 break; | |
| 1070 continue; | |
| 1071 } | |
| 481 | 1072 } |
| 1073 | |
| 339 | 1074 if (flags & WF_BANNED) |
| 1075 res = SP_BANNED; | |
| 1076 else if (flags & WF_REGION) | |
| 1077 { | |
| 1078 /* Check region. */ | |
| 390 | 1079 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0) |
| 300 | 1080 res = SP_OK; |
| 339 | 1081 else |
| 1082 res = SP_LOCAL; | |
| 300 | 1083 } |
| 339 | 1084 else if (flags & WF_RARE) |
| 1085 res = SP_RARE; | |
| 307 | 1086 else |
| 339 | 1087 res = SP_OK; |
| 1088 | |
| 492 | 1089 /* Always use the longest match and the best result. For NOBREAK |
| 1090 * we separately keep the longest match without a following good | |
| 1091 * word as a fall-back. */ | |
| 1092 if (nobreak_result == SP_BAD) | |
| 1093 { | |
| 1094 if (mip->mi_result2 > res) | |
| 1095 { | |
| 1096 mip->mi_result2 = res; | |
| 1097 mip->mi_end2 = mip->mi_word + wlen; | |
| 1098 } | |
| 1099 else if (mip->mi_result2 == res | |
| 1100 && mip->mi_end2 < mip->mi_word + wlen) | |
| 1101 mip->mi_end2 = mip->mi_word + wlen; | |
| 1102 } | |
| 1103 else if (mip->mi_result > res) | |
| 339 | 1104 { |
| 1105 mip->mi_result = res; | |
| 1106 mip->mi_end = mip->mi_word + wlen; | |
| 1107 } | |
| 351 | 1108 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen) |
| 339 | 1109 mip->mi_end = mip->mi_word + wlen; |
| 1110 | |
| 492 | 1111 if (mip->mi_result == SP_OK) |
| 339 | 1112 break; |
| 300 | 1113 } |
| 1114 | |
| 492 | 1115 if (mip->mi_result == SP_OK) |
| 300 | 1116 break; |
| 1117 } | |
| 236 | 1118 } |
| 1119 | |
| 323 | 1120 /* |
| 1762 | 1121 * Return TRUE if there is a match between the word ptr[wlen] and |
| 1122 * CHECKCOMPOUNDPATTERN rules, assuming that we will concatenate with another | |
| 1123 * word. | |
| 1124 * A match means that the first part of CHECKCOMPOUNDPATTERN matches at the | |
| 1125 * end of ptr[wlen] and the second part matches after it. | |
| 1126 */ | |
| 1127 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1128 match_checkcompoundpattern( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1129 char_u *ptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1130 int wlen, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1131 garray_T *gap) /* &sl_comppat */ |
| 1762 | 1132 { |
| 1133 int i; | |
| 1134 char_u *p; | |
| 1135 int len; | |
| 1136 | |
| 1137 for (i = 0; i + 1 < gap->ga_len; i += 2) | |
| 1138 { | |
| 1139 p = ((char_u **)gap->ga_data)[i + 1]; | |
| 1140 if (STRNCMP(ptr + wlen, p, STRLEN(p)) == 0) | |
| 1141 { | |
| 1142 /* Second part matches at start of following compound word, now | |
| 1143 * check if first part matches at end of previous word. */ | |
| 1144 p = ((char_u **)gap->ga_data)[i]; | |
| 1771 | 1145 len = (int)STRLEN(p); |
| 1762 | 1146 if (len <= wlen && STRNCMP(ptr + wlen - len, p, len) == 0) |
| 1147 return TRUE; | |
| 1148 } | |
| 1149 } | |
| 1150 return FALSE; | |
| 1151 } | |
| 1152 | |
| 1153 /* | |
| 626 | 1154 * Return TRUE if "flags" is a valid sequence of compound flags and "word" |
| 1155 * does not have too many syllables. | |
| 482 | 1156 */ |
| 1157 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1158 can_compound(slang_T *slang, char_u *word, char_u *flags) |
| 483 | 1159 { |
| 495 | 1160 char_u uflags[MAXWLEN * 2]; |
| 1161 int i; | |
| 1162 char_u *p; | |
| 483 | 1163 |
| 1164 if (slang->sl_compprog == NULL) | |
| 1165 return FALSE; | |
| 495 | 1166 if (enc_utf8) |
| 1167 { | |
| 1168 /* Need to convert the single byte flags to utf8 characters. */ | |
| 1169 p = uflags; | |
| 1170 for (i = 0; flags[i] != NUL; ++i) | |
|
11269
121d29004998
patch 8.0.0520: using a function pointer while the function is known
Christian Brabandt <cb@256bit.org>
parents:
11129
diff
changeset
|
1171 p += utf_char2bytes(flags[i], p); |
| 495 | 1172 *p = NUL; |
| 1173 p = uflags; | |
| 1174 } | |
| 1175 else | |
| 1176 p = flags; | |
| 6375 | 1177 if (!vim_regexec_prog(&slang->sl_compprog, FALSE, p, 0)) |
| 483 | 1178 return FALSE; |
| 1179 | |
| 490 | 1180 /* Count the number of syllables. This may be slow, do it last. If there |
| 1181 * are too many syllables AND the number of compound words is above | |
| 809 | 1182 * COMPOUNDWORDMAX then compounding is not allowed. */ |
| 483 | 1183 if (slang->sl_compsylmax < MAXWLEN |
| 1184 && count_syllables(slang, word) > slang->sl_compsylmax) | |
| 495 | 1185 return (int)STRLEN(flags) < slang->sl_compmax; |
| 483 | 1186 return TRUE; |
| 482 | 1187 } |
| 1188 | |
| 1189 /* | |
| 1762 | 1190 * Return TRUE when the sequence of flags in "compflags" plus "flag" can |
| 1191 * possibly form a valid compounded word. This also checks the COMPOUNDRULE | |
| 1192 * lines if they don't contain wildcards. | |
| 1193 */ | |
| 1194 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1195 can_be_compound( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1196 trystate_T *sp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1197 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1198 char_u *compflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1199 int flag) |
| 1762 | 1200 { |
| 1201 /* If the flag doesn't appear in sl_compstartflags or sl_compallflags | |
| 1202 * then it can't possibly compound. */ | |
| 1203 if (!byte_in_str(sp->ts_complen == sp->ts_compsplit | |
| 1204 ? slang->sl_compstartflags : slang->sl_compallflags, flag)) | |
| 1205 return FALSE; | |
| 1206 | |
| 1207 /* If there are no wildcards, we can check if the flags collected so far | |
| 1208 * possibly can form a match with COMPOUNDRULE patterns. This only | |
| 1209 * makes sense when we have two or more words. */ | |
| 1210 if (slang->sl_comprules != NULL && sp->ts_complen > sp->ts_compsplit) | |
| 1211 { | |
| 1212 int v; | |
| 1213 | |
| 1214 compflags[sp->ts_complen] = flag; | |
| 1215 compflags[sp->ts_complen + 1] = NUL; | |
| 1216 v = match_compoundrule(slang, compflags + sp->ts_compsplit); | |
| 1217 compflags[sp->ts_complen] = NUL; | |
| 1218 return v; | |
| 1219 } | |
| 1220 | |
| 1221 return TRUE; | |
| 1222 } | |
| 1223 | |
| 1224 | |
| 1225 /* | |
| 1226 * Return TRUE if the compound flags in compflags[] match the start of any | |
| 1227 * compound rule. This is used to stop trying a compound if the flags | |
| 1228 * collected so far can't possibly match any compound rule. | |
| 1229 * Caller must check that slang->sl_comprules is not NULL. | |
| 1230 */ | |
| 1231 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1232 match_compoundrule(slang_T *slang, char_u *compflags) |
| 1762 | 1233 { |
| 1234 char_u *p; | |
| 1235 int i; | |
| 1236 int c; | |
| 1237 | |
| 1238 /* loop over all the COMPOUNDRULE entries */ | |
| 1239 for (p = slang->sl_comprules; *p != NUL; ++p) | |
| 1240 { | |
| 1241 /* loop over the flags in the compound word we have made, match | |
| 1242 * them against the current rule entry */ | |
| 1243 for (i = 0; ; ++i) | |
| 1244 { | |
| 1245 c = compflags[i]; | |
| 1246 if (c == NUL) | |
| 1247 /* found a rule that matches for the flags we have so far */ | |
| 1248 return TRUE; | |
| 1249 if (*p == '/' || *p == NUL) | |
| 1250 break; /* end of rule, it's too short */ | |
| 1251 if (*p == '[') | |
| 1252 { | |
| 1253 int match = FALSE; | |
| 1254 | |
| 1255 /* compare against all the flags in [] */ | |
| 1256 ++p; | |
| 1257 while (*p != ']' && *p != NUL) | |
| 1258 if (*p++ == c) | |
| 1259 match = TRUE; | |
| 1260 if (!match) | |
| 1261 break; /* none matches */ | |
| 1262 } | |
| 1263 else if (*p != c) | |
| 1264 break; /* flag of word doesn't match flag in pattern */ | |
| 1265 ++p; | |
| 1266 } | |
| 1267 | |
| 1268 /* Skip to the next "/", where the next pattern starts. */ | |
| 1269 p = vim_strchr(p, '/'); | |
| 1270 if (p == NULL) | |
| 1271 break; | |
| 1272 } | |
| 1273 | |
| 1274 /* Checked all the rules and none of them match the flags, so there | |
| 1275 * can't possibly be a compound starting with these flags. */ | |
| 1276 return FALSE; | |
| 1277 } | |
| 1278 | |
| 1279 /* | |
| 390 | 1280 * Return non-zero if the prefix indicated by "arridx" matches with the prefix |
| 1281 * ID in "flags" for the word "word". | |
| 366 | 1282 * The WF_RAREPFX flag is included in the return value for a rare prefix. |
| 351 | 1283 */ |
| 1284 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1285 valid_word_prefix( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1286 int totprefcnt, /* nr of prefix IDs */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1287 int arridx, /* idx in sl_pidxs[] */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1288 int flags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1289 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1290 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1291 int cond_req) /* only use prefixes with a condition */ |
| 351 | 1292 { |
| 1293 int prefcnt; | |
| 1294 int pidx; | |
| 6375 | 1295 regprog_T **rp; |
| 390 | 1296 int prefid; |
| 1297 | |
| 1298 prefid = (unsigned)flags >> 24; | |
| 351 | 1299 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt) |
| 1300 { | |
| 1301 pidx = slang->sl_pidxs[arridx + prefcnt]; | |
| 1302 | |
| 1303 /* Check the prefix ID. */ | |
| 1304 if (prefid != (pidx & 0xff)) | |
| 1305 continue; | |
| 1306 | |
| 390 | 1307 /* Check if the prefix doesn't combine and the word already has a |
| 1308 * suffix. */ | |
| 1309 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC)) | |
| 1310 continue; | |
| 1311 | |
| 351 | 1312 /* Check the condition, if there is one. The condition index is |
| 366 | 1313 * stored in the two bytes above the prefix ID byte. */ |
| 6375 | 1314 rp = &slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff]; |
| 1315 if (*rp != NULL) | |
| 1316 { | |
| 1317 if (!vim_regexec_prog(rp, FALSE, word, 0)) | |
| 351 | 1318 continue; |
| 1319 } | |
| 455 | 1320 else if (cond_req) |
| 1321 continue; | |
| 1322 | |
| 1323 /* It's a match! Return the WF_ flags. */ | |
| 366 | 1324 return pidx; |
| 1325 } | |
| 1326 return 0; | |
| 351 | 1327 } |
| 1328 | |
| 1329 /* | |
| 339 | 1330 * Check if the word at "mip->mi_word" has a matching prefix. |
| 1331 * If it does, then check the following word. | |
| 1332 * | |
| 485 | 1333 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a |
| 1334 * prefix in a compound word. | |
| 1335 * | |
| 339 | 1336 * For a match mip->mi_result is updated. |
| 1337 */ | |
| 1338 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1339 find_prefix(matchinf_T *mip, int mode) |
| 339 | 1340 { |
| 1341 idx_T arridx = 0; | |
| 1342 int len; | |
| 1343 int wlen = 0; | |
| 1344 int flen; | |
| 1345 int c; | |
| 1346 char_u *ptr; | |
| 1347 idx_T lo, hi, m; | |
| 1348 slang_T *slang = mip->mi_lp->lp_slang; | |
| 1349 char_u *byts; | |
| 1350 idx_T *idxs; | |
| 1351 | |
| 375 | 1352 byts = slang->sl_pbyts; |
| 1353 if (byts == NULL) | |
| 1354 return; /* array is empty */ | |
| 1355 | |
| 339 | 1356 /* We use the case-folded word here, since prefixes are always |
| 1357 * case-folded. */ | |
| 1358 ptr = mip->mi_fword; | |
| 1359 flen = mip->mi_fwordlen; /* available case-folded bytes */ | |
| 485 | 1360 if (mode == FIND_COMPOUND) |
| 1361 { | |
| 1362 /* Skip over the previously found word(s). */ | |
| 1363 ptr += mip->mi_compoff; | |
| 1364 flen -= mip->mi_compoff; | |
| 1365 } | |
| 339 | 1366 idxs = slang->sl_pidxs; |
| 1367 | |
| 1368 /* | |
| 1369 * Repeat advancing in the tree until: | |
| 1370 * - there is a byte that doesn't match, | |
| 1371 * - we reach the end of the tree, | |
| 1372 * - or we reach the end of the line. | |
| 1373 */ | |
| 1374 for (;;) | |
| 1375 { | |
| 1376 if (flen == 0 && *mip->mi_fend != NUL) | |
| 1377 flen = fold_more(mip); | |
| 1378 | |
| 1379 len = byts[arridx++]; | |
| 1380 | |
| 1381 /* If the first possible byte is a zero the prefix could end here. | |
| 1382 * Check if the following word matches and supports the prefix. */ | |
| 1383 if (byts[arridx] == 0) | |
| 1384 { | |
| 1385 /* There can be several prefixes with different conditions. We | |
| 1386 * try them all, since we don't know which one will give the | |
| 1387 * longest match. The word is the same each time, pass the list | |
| 1388 * of possible prefixes to find_word(). */ | |
| 1389 mip->mi_prefarridx = arridx; | |
| 1390 mip->mi_prefcnt = len; | |
| 1391 while (len > 0 && byts[arridx] == 0) | |
| 1392 { | |
| 1393 ++arridx; | |
| 1394 --len; | |
| 1395 } | |
| 1396 mip->mi_prefcnt -= len; | |
| 1397 | |
| 1398 /* Find the word that comes after the prefix. */ | |
| 1399 mip->mi_prefixlen = wlen; | |
| 485 | 1400 if (mode == FIND_COMPOUND) |
| 1401 /* Skip over the previously found word(s). */ | |
| 1402 mip->mi_prefixlen += mip->mi_compoff; | |
| 1403 | |
| 455 | 1404 if (has_mbyte) |
| 1405 { | |
| 1406 /* Case-folded length may differ from original length. */ | |
| 485 | 1407 mip->mi_cprefixlen = nofold_len(mip->mi_fword, |
| 1408 mip->mi_prefixlen, mip->mi_word); | |
| 455 | 1409 } |
| 1410 else | |
| 485 | 1411 mip->mi_cprefixlen = mip->mi_prefixlen; |
| 339 | 1412 find_word(mip, FIND_PREFIX); |
| 1413 | |
| 1414 | |
| 1415 if (len == 0) | |
| 1416 break; /* no children, word must end here */ | |
| 1417 } | |
| 1418 | |
| 1419 /* Stop looking at end of the line. */ | |
| 1420 if (ptr[wlen] == NUL) | |
| 1421 break; | |
| 1422 | |
| 1423 /* Perform a binary search in the list of accepted bytes. */ | |
| 1424 c = ptr[wlen]; | |
| 1425 lo = arridx; | |
| 1426 hi = arridx + len - 1; | |
| 1427 while (lo < hi) | |
| 1428 { | |
| 1429 m = (lo + hi) / 2; | |
| 1430 if (byts[m] > c) | |
| 1431 hi = m - 1; | |
| 1432 else if (byts[m] < c) | |
| 1433 lo = m + 1; | |
| 1434 else | |
| 1435 { | |
| 1436 lo = hi = m; | |
| 1437 break; | |
| 1438 } | |
| 1439 } | |
| 1440 | |
| 1441 /* Stop if there is no matching byte. */ | |
| 1442 if (hi < lo || byts[lo] != c) | |
| 1443 break; | |
| 1444 | |
| 1445 /* Continue at the child (if there is one). */ | |
| 1446 arridx = idxs[lo]; | |
| 1447 ++wlen; | |
| 1448 --flen; | |
| 1449 } | |
| 1450 } | |
| 1451 | |
| 1452 /* | |
| 1453 * Need to fold at least one more character. Do until next non-word character | |
| 626 | 1454 * for efficiency. Include the non-word character too. |
| 339 | 1455 * Return the length of the folded chars in bytes. |
| 1456 */ | |
| 1457 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1458 fold_more(matchinf_T *mip) |
| 339 | 1459 { |
| 1460 int flen; | |
| 1461 char_u *p; | |
| 1462 | |
| 1463 p = mip->mi_fend; | |
| 1464 do | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
1465 MB_PTR_ADV(mip->mi_fend); |
|
16162
cd5c83115ec6
patch 8.1.1086: too many curly braces
Bram Moolenaar <Bram@vim.org>
parents:
16142
diff
changeset
|
1466 while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_win)); |
| 339 | 1467 |
| 626 | 1468 /* Include the non-word character so that we can check for the word end. */ |
| 339 | 1469 if (*mip->mi_fend != NUL) |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
1470 MB_PTR_ADV(mip->mi_fend); |
| 339 | 1471 |
| 1472 (void)spell_casefold(p, (int)(mip->mi_fend - p), | |
| 1473 mip->mi_fword + mip->mi_fwordlen, | |
| 1474 MAXWLEN - mip->mi_fwordlen); | |
| 835 | 1475 flen = (int)STRLEN(mip->mi_fword + mip->mi_fwordlen); |
| 339 | 1476 mip->mi_fwordlen += flen; |
| 1477 return flen; | |
| 1478 } | |
| 1479 | |
| 1480 /* | |
| 323 | 1481 * Check case flags for a word. Return TRUE if the word has the requested |
| 1482 * case. | |
| 1483 */ | |
| 1484 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1485 spell_valid_case( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1486 int wordflags, /* flags for the checked word. */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1487 int treeflags) /* flags for the word in the spell tree */ |
| 323 | 1488 { |
| 389 | 1489 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0) |
| 323 | 1490 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0 |
| 474 | 1491 && ((treeflags & WF_ONECAP) == 0 |
| 1492 || (wordflags & WF_ONECAP) != 0))); | |
| 323 | 1493 } |
| 1494 | |
| 351 | 1495 /* |
| 1496 * Return TRUE if spell checking is not enabled. | |
| 1497 */ | |
| 1498 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1499 no_spell_checking(win_T *wp) |
| 498 | 1500 { |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1501 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
|
1502 || wp->w_s->b_langp.ga_len == 0) |
| 351 | 1503 { |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
1504 emsg(_("E756: Spell checking is not enabled")); |
| 351 | 1505 return TRUE; |
| 1506 } | |
| 1507 return FALSE; | |
| 1508 } | |
| 300 | 1509 |
| 236 | 1510 /* |
| 1511 * Move to next spell error. | |
| 500 | 1512 * "curline" is FALSE for "[s", "]s", "[S" and "]S". |
| 1513 * "curline" is TRUE to find word under/after cursor in the same line. | |
| 483 | 1514 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move |
| 1515 * to after badly spelled word before the cursor. | |
| 495 | 1516 * Return 0 if not found, length of the badly spelled word otherwise. |
| 236 | 1517 */ |
| 1518 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1519 spell_move_to( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1520 win_T *wp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1521 int dir, /* FORWARD or BACKWARD */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1522 int allwords, /* TRUE for "[s"/"]s", FALSE for "[S"/"]S" */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1523 int curline, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1524 hlf_T *attrp) /* return: attributes of bad word or NULL |
| 534 | 1525 (only when "dir" is FORWARD) */ |
| 236 | 1526 { |
| 249 | 1527 linenr_T lnum; |
| 1528 pos_T found_pos; | |
| 495 | 1529 int found_len = 0; |
| 236 | 1530 char_u *line; |
| 1531 char_u *p; | |
| 346 | 1532 char_u *endp; |
| 534 | 1533 hlf_T attr; |
| 236 | 1534 int len; |
| 5519 | 1535 #ifdef FEAT_SYN_HL |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1536 int has_syntax = syntax_present(wp); |
| 5519 | 1537 #endif |
| 249 | 1538 int col; |
| 1539 int can_spell; | |
| 346 | 1540 char_u *buf = NULL; |
| 1541 int buflen = 0; | |
| 1542 int skip = 0; | |
| 385 | 1543 int capcol = -1; |
| 500 | 1544 int found_one = FALSE; |
| 1545 int wrapped = FALSE; | |
| 236 | 1546 |
| 498 | 1547 if (no_spell_checking(wp)) |
| 495 | 1548 return 0; |
| 236 | 1549 |
| 249 | 1550 /* |
| 1551 * Start looking for bad word at the start of the line, because we can't | |
| 817 | 1552 * start halfway a word, we don't know where it starts or ends. |
| 249 | 1553 * |
| 1554 * When searching backwards, we continue in the line to find the last | |
| 1555 * bad word (in the cursor line: before the cursor). | |
| 346 | 1556 * |
| 1557 * We concatenate the start of the next line, so that wrapped words work | |
| 1558 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards | |
| 1559 * though... | |
| 249 | 1560 */ |
| 498 | 1561 lnum = wp->w_cursor.lnum; |
|
11121
778c10516955
patch 8.0.0448: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
10950
diff
changeset
|
1562 CLEAR_POS(&found_pos); |
| 236 | 1563 |
| 1564 while (!got_int) | |
| 1565 { | |
| 498 | 1566 line = ml_get_buf(wp->w_buffer, lnum, FALSE); |
| 346 | 1567 |
| 835 | 1568 len = (int)STRLEN(line); |
| 346 | 1569 if (buflen < len + MAXWLEN + 2) |
| 1570 { | |
| 1571 vim_free(buf); | |
| 1572 buflen = len + MAXWLEN + 2; | |
| 1573 buf = alloc(buflen); | |
| 1574 if (buf == NULL) | |
| 1575 break; | |
| 1576 } | |
| 1577 | |
| 385 | 1578 /* In first line check first word for Capital. */ |
| 1579 if (lnum == 1) | |
| 1580 capcol = 0; | |
| 1581 | |
| 1582 /* For checking first word with a capital skip white space. */ | |
| 1583 if (capcol == 0) | |
|
12323
4dba3e4f3b01
patch 8.0.1041: bogus characters when indenting during visual-block append
Christian Brabandt <cb@256bit.org>
parents:
11521
diff
changeset
|
1584 capcol = getwhitecols(line); |
| 835 | 1585 else if (curline && wp == curwin) |
| 1586 { | |
| 1587 /* For spellbadword(): check if first word needs a capital. */ | |
|
12323
4dba3e4f3b01
patch 8.0.1041: bogus characters when indenting during visual-block append
Christian Brabandt <cb@256bit.org>
parents:
11521
diff
changeset
|
1588 col = getwhitecols(line); |
| 835 | 1589 if (check_need_cap(lnum, col)) |
| 1590 capcol = col; | |
| 1591 | |
| 1592 /* Need to get the line again, may have looked at the previous | |
| 1593 * one. */ | |
| 1594 line = ml_get_buf(wp->w_buffer, lnum, FALSE); | |
| 1595 } | |
| 385 | 1596 |
| 346 | 1597 /* Copy the line into "buf" and append the start of the next line if |
| 1598 * possible. */ | |
| 1599 STRCPY(buf, line); | |
| 498 | 1600 if (lnum < wp->w_buffer->b_ml.ml_line_count) |
| 883 | 1601 spell_cat_line(buf + STRLEN(buf), |
| 1602 ml_get_buf(wp->w_buffer, lnum + 1, FALSE), MAXWLEN); | |
| 346 | 1603 |
| 1604 p = buf + skip; | |
| 1605 endp = buf + len; | |
| 1606 while (p < endp) | |
| 236 | 1607 { |
| 500 | 1608 /* When searching backward don't search after the cursor. Unless |
| 1609 * we wrapped around the end of the buffer. */ | |
| 300 | 1610 if (dir == BACKWARD |
| 498 | 1611 && lnum == wp->w_cursor.lnum |
| 500 | 1612 && !wrapped |
| 498 | 1613 && (colnr_T)(p - buf) >= wp->w_cursor.col) |
| 300 | 1614 break; |
| 249 | 1615 |
| 300 | 1616 /* start of word */ |
| 534 | 1617 attr = HLF_COUNT; |
| 625 | 1618 len = spell_check(wp, p, &attr, &capcol, FALSE); |
| 249 | 1619 |
| 534 | 1620 if (attr != HLF_COUNT) |
| 300 | 1621 { |
| 1622 /* We found a bad word. Check the attribute. */ | |
| 534 | 1623 if (allwords || attr == HLF_SPB) |
| 236 | 1624 { |
| 300 | 1625 /* When searching forward only accept a bad word after |
| 1626 * the cursor. */ | |
| 1627 if (dir == BACKWARD | |
| 500 | 1628 || lnum != wp->w_cursor.lnum |
| 498 | 1629 || (lnum == wp->w_cursor.lnum |
| 500 | 1630 && (wrapped |
| 1631 || (colnr_T)(curline ? p - buf + len | |
| 346 | 1632 : p - buf) |
| 500 | 1633 > wp->w_cursor.col))) |
| 236 | 1634 { |
| 5519 | 1635 #ifdef FEAT_SYN_HL |
| 300 | 1636 if (has_syntax) |
| 249 | 1637 { |
| 835 | 1638 col = (int)(p - buf); |
| 498 | 1639 (void)syn_get_id(wp, lnum, (colnr_T)col, |
| 1504 | 1640 FALSE, &can_spell, FALSE); |
| 857 | 1641 if (!can_spell) |
| 1642 attr = HLF_COUNT; | |
| 300 | 1643 } |
| 1644 else | |
| 737 | 1645 #endif |
| 300 | 1646 can_spell = TRUE; |
| 249 | 1647 |
| 300 | 1648 if (can_spell) |
| 1649 { | |
| 857 | 1650 found_one = TRUE; |
| 300 | 1651 found_pos.lnum = lnum; |
| 835 | 1652 found_pos.col = (int)(p - buf); |
| 300 | 1653 found_pos.coladd = 0; |
| 1654 if (dir == FORWARD) | |
| 1655 { | |
| 1656 /* No need to search further. */ | |
| 498 | 1657 wp->w_cursor = found_pos; |
| 346 | 1658 vim_free(buf); |
| 498 | 1659 if (attrp != NULL) |
| 1660 *attrp = attr; | |
| 495 | 1661 return len; |
| 249 | 1662 } |
| 483 | 1663 else if (curline) |
| 1664 /* Insert mode completion: put cursor after | |
| 1665 * the bad word. */ | |
| 1666 found_pos.col += len; | |
| 495 | 1667 found_len = len; |
| 249 | 1668 } |
| 236 | 1669 } |
| 857 | 1670 else |
| 1671 found_one = TRUE; | |
| 236 | 1672 } |
| 1673 } | |
| 1674 | |
| 300 | 1675 /* advance to character after the word */ |
| 1676 p += len; | |
| 385 | 1677 capcol -= len; |
| 236 | 1678 } |
| 1679 | |
| 483 | 1680 if (dir == BACKWARD && found_pos.lnum != 0) |
| 1681 { | |
| 500 | 1682 /* Use the last match in the line (before the cursor). */ |
| 498 | 1683 wp->w_cursor = found_pos; |
| 483 | 1684 vim_free(buf); |
| 495 | 1685 return found_len; |
| 483 | 1686 } |
| 1687 | |
| 323 | 1688 if (curline) |
| 346 | 1689 break; /* only check cursor line */ |
| 323 | 1690 |
|
10950
2297aae8e127
patch 8.0.0364: ]s does not move cursor with two spell errors in one line
Christian Brabandt <cb@256bit.org>
parents:
10359
diff
changeset
|
1691 /* If we are back at the starting line and searched it again there |
|
2297aae8e127
patch 8.0.0364: ]s does not move cursor with two spell errors in one line
Christian Brabandt <cb@256bit.org>
parents:
10359
diff
changeset
|
1692 * is no match, give up. */ |
|
2297aae8e127
patch 8.0.0364: ]s does not move cursor with two spell errors in one line
Christian Brabandt <cb@256bit.org>
parents:
10359
diff
changeset
|
1693 if (lnum == wp->w_cursor.lnum && wrapped) |
|
2297aae8e127
patch 8.0.0364: ]s does not move cursor with two spell errors in one line
Christian Brabandt <cb@256bit.org>
parents:
10359
diff
changeset
|
1694 break; |
|
2297aae8e127
patch 8.0.0364: ]s does not move cursor with two spell errors in one line
Christian Brabandt <cb@256bit.org>
parents:
10359
diff
changeset
|
1695 |
| 236 | 1696 /* Advance to next line. */ |
| 249 | 1697 if (dir == BACKWARD) |
| 1698 { | |
| 500 | 1699 if (lnum > 1) |
| 1700 --lnum; | |
| 1701 else if (!p_ws) | |
| 1702 break; /* at first line and 'nowrapscan' */ | |
| 1703 else | |
| 1704 { | |
| 1705 /* Wrap around to the end of the buffer. May search the | |
| 1706 * starting line again and accept the last match. */ | |
| 1707 lnum = wp->w_buffer->b_ml.ml_line_count; | |
| 1708 wrapped = TRUE; | |
| 503 | 1709 if (!shortmess(SHM_SEARCH)) |
| 1710 give_warning((char_u *)_(top_bot_msg), TRUE); | |
| 500 | 1711 } |
| 385 | 1712 capcol = -1; |
| 249 | 1713 } |
| 1714 else | |
| 1715 { | |
| 500 | 1716 if (lnum < wp->w_buffer->b_ml.ml_line_count) |
| 1717 ++lnum; | |
| 1718 else if (!p_ws) | |
| 1719 break; /* at first line and 'nowrapscan' */ | |
| 1720 else | |
| 1721 { | |
| 1722 /* Wrap around to the start of the buffer. May search the | |
| 1723 * starting line again and accept the first match. */ | |
| 1724 lnum = 1; | |
| 1725 wrapped = TRUE; | |
| 503 | 1726 if (!shortmess(SHM_SEARCH)) |
| 1727 give_warning((char_u *)_(bot_top_msg), TRUE); | |
| 500 | 1728 } |
| 1729 | |
| 1730 /* If we are back at the starting line and there is no match then | |
| 1731 * give up. */ | |
|
10950
2297aae8e127
patch 8.0.0364: ]s does not move cursor with two spell errors in one line
Christian Brabandt <cb@256bit.org>
parents:
10359
diff
changeset
|
1732 if (lnum == wp->w_cursor.lnum && !found_one) |
| 346 | 1733 break; |
| 1734 | |
| 1735 /* Skip the characters at the start of the next line that were | |
| 1736 * included in a match crossing line boundaries. */ | |
| 534 | 1737 if (attr == HLF_COUNT) |
| 835 | 1738 skip = (int)(p - endp); |
| 346 | 1739 else |
| 1740 skip = 0; | |
| 385 | 1741 |
| 835 | 1742 /* Capcol skips over the inserted space. */ |
| 385 | 1743 --capcol; |
| 1744 | |
| 1745 /* But after empty line check first word in next line */ | |
| 1746 if (*skipwhite(line) == NUL) | |
| 1747 capcol = 0; | |
| 249 | 1748 } |
| 236 | 1749 |
| 1750 line_breakcheck(); | |
| 1751 } | |
| 1752 | |
| 346 | 1753 vim_free(buf); |
| 495 | 1754 return 0; |
| 346 | 1755 } |
| 1756 | |
| 1757 /* | |
| 1758 * For spell checking: concatenate the start of the following line "line" into | |
| 1759 * "buf", blanking-out special characters. Copy less then "maxlen" bytes. | |
| 1577 | 1760 * Keep the blanks at the start of the next line, this is used in win_line() |
| 1761 * to skip those bytes if the word was OK. | |
| 346 | 1762 */ |
| 1763 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1764 spell_cat_line(char_u *buf, char_u *line, int maxlen) |
| 346 | 1765 { |
| 1766 char_u *p; | |
| 1767 int n; | |
| 1768 | |
| 1769 p = skipwhite(line); | |
| 1770 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL) | |
| 1771 p = skipwhite(p + 1); | |
| 1772 | |
| 1773 if (*p != NUL) | |
| 1774 { | |
| 1577 | 1775 /* Only worth concatenating if there is something else than spaces to |
| 1776 * concatenate. */ | |
| 1777 n = (int)(p - line) + 1; | |
| 1778 if (n < maxlen - 1) | |
| 1779 { | |
| 1780 vim_memset(buf, ' ', n); | |
| 1781 vim_strncpy(buf + n, p, maxlen - 1 - n); | |
| 1782 } | |
| 346 | 1783 } |
| 236 | 1784 } |
| 1785 | |
| 626 | 1786 /* |
| 1787 * Structure used for the cookie argument of do_in_runtimepath(). | |
| 1788 */ | |
| 501 | 1789 typedef struct spelload_S |
| 1790 { | |
| 1791 char_u sl_lang[MAXWLEN + 1]; /* language name */ | |
| 1792 slang_T *sl_slang; /* resulting slang_T struct */ | |
| 1793 int sl_nobreak; /* NOBREAK language found */ | |
| 1794 } spelload_T; | |
| 1795 | |
| 236 | 1796 /* |
| 307 | 1797 * Load word list(s) for "lang" from Vim spell file(s). |
| 310 | 1798 * "lang" must be the language without the region: e.g., "en". |
| 236 | 1799 */ |
| 307 | 1800 static void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1801 spell_load_lang(char_u *lang) |
| 236 | 1802 { |
| 310 | 1803 char_u fname_enc[85]; |
| 236 | 1804 int r; |
| 501 | 1805 spelload_T sl; |
| 649 | 1806 int round; |
| 307 | 1807 |
| 310 | 1808 /* Copy the language name to pass it to spell_load_cb() as a cookie. |
| 307 | 1809 * It's truncated when an error is detected. */ |
| 501 | 1810 STRCPY(sl.sl_lang, lang); |
| 1811 sl.sl_slang = NULL; | |
| 1812 sl.sl_nobreak = FALSE; | |
| 307 | 1813 |
| 649 | 1814 /* We may retry when no spell file is found for the language, an |
| 1815 * autocommand may load it then. */ | |
| 1816 for (round = 1; round <= 2; ++round) | |
| 1817 { | |
| 1818 /* | |
| 1819 * Find the first spell file for "lang" in 'runtimepath' and load it. | |
| 1820 */ | |
| 1821 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 1822 #ifdef VMS |
| 1823 "spell/%s_%s.spl", | |
| 1824 #else | |
| 1825 "spell/%s.%s.spl", | |
| 1826 #endif | |
| 1827 lang, spell_enc()); | |
|
8524
2f57bbe870ea
commit https://github.com/vim/vim/commit/7f8989dd8a627af2185df381195351a913f3777f
Christian Brabandt <cb@256bit.org>
parents:
7835
diff
changeset
|
1828 r = do_in_runtimepath(fname_enc, 0, spell_load_cb, &sl); |
| 649 | 1829 |
| 1830 if (r == FAIL && *sl.sl_lang != NUL) | |
| 1831 { | |
| 1832 /* Try loading the ASCII version. */ | |
| 1833 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 1834 #ifdef VMS |
| 1835 "spell/%s_ascii.spl", | |
| 1836 #else | |
| 1837 "spell/%s.ascii.spl", | |
| 1838 #endif | |
| 1839 lang); | |
|
8524
2f57bbe870ea
commit https://github.com/vim/vim/commit/7f8989dd8a627af2185df381195351a913f3777f
Christian Brabandt <cb@256bit.org>
parents:
7835
diff
changeset
|
1840 r = do_in_runtimepath(fname_enc, 0, spell_load_cb, &sl); |
| 649 | 1841 |
| 1842 if (r == FAIL && *sl.sl_lang != NUL && round == 1 | |
| 1843 && apply_autocmds(EVENT_SPELLFILEMISSING, lang, | |
| 1844 curbuf->b_fname, FALSE, curbuf)) | |
| 1845 continue; | |
| 1846 break; | |
|
13380
69517d67421f
patch 8.0.1564: too many #ifdefs
Christian Brabandt <cb@256bit.org>
parents:
13308
diff
changeset
|
1847 } |
| 714 | 1848 break; |
| 307 | 1849 } |
| 1850 | |
| 1851 if (r == FAIL) | |
| 649 | 1852 { |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
1853 smsg( |
| 2660 | 1854 #ifdef VMS |
| 1855 _("Warning: Cannot find word list \"%s_%s.spl\" or \"%s_ascii.spl\""), | |
| 1856 #else | |
| 1857 _("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""), | |
| 1858 #endif | |
| 483 | 1859 lang, spell_enc(), lang); |
| 649 | 1860 } |
| 501 | 1861 else if (sl.sl_slang != NULL) |
| 1862 { | |
| 626 | 1863 /* At least one file was loaded, now load ALL the additions. */ |
| 310 | 1864 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl"); |
|
8524
2f57bbe870ea
commit https://github.com/vim/vim/commit/7f8989dd8a627af2185df381195351a913f3777f
Christian Brabandt <cb@256bit.org>
parents:
7835
diff
changeset
|
1865 do_in_runtimepath(fname_enc, DIP_ALL, spell_load_cb, &sl); |
| 310 | 1866 } |
| 1867 } | |
| 1868 | |
| 1869 /* | |
| 1870 * Return the encoding used for spell checking: Use 'encoding', except that we | |
| 1871 * use "latin1" for "latin9". And limit to 60 characters (just in case). | |
| 1872 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1873 char_u * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1874 spell_enc(void) |
| 310 | 1875 { |
| 1876 | |
| 1877 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0) | |
| 1878 return p_enc; | |
| 1879 return (char_u *)"latin1"; | |
| 236 | 1880 } |
| 1881 | |
| 1882 /* | |
| 385 | 1883 * Get the name of the .spl file for the internal wordlist into |
| 1884 * "fname[MAXPATHL]". | |
| 1885 */ | |
| 1886 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1887 int_wordlist_spl(char_u *fname) |
| 385 | 1888 { |
| 2660 | 1889 vim_snprintf((char *)fname, MAXPATHL, SPL_FNAME_TMPL, |
| 385 | 1890 int_wordlist, spell_enc()); |
| 1891 } | |
| 1892 | |
| 1893 /* | |
| 625 | 1894 * Allocate a new slang_T for language "lang". "lang" can be NULL. |
| 236 | 1895 * Caller must fill "sl_next". |
| 1896 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1897 slang_T * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1898 slang_alloc(char_u *lang) |
| 236 | 1899 { |
| 1900 slang_T *lp; | |
| 1901 | |
|
16825
ce04ebdf26b8
patch 8.1.1414: alloc() returning "char_u *" causes a lot of type casts
Bram Moolenaar <Bram@vim.org>
parents:
16768
diff
changeset
|
1902 lp = ALLOC_CLEAR_ONE(slang_T); |
| 236 | 1903 if (lp != NULL) |
| 1904 { | |
| 625 | 1905 if (lang != NULL) |
| 1906 lp->sl_name = vim_strsave(lang); | |
| 323 | 1907 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10); |
| 625 | 1908 ga_init2(&lp->sl_repsal, sizeof(fromto_T), 10); |
| 483 | 1909 lp->sl_compmax = MAXWLEN; |
| 1910 lp->sl_compsylmax = MAXWLEN; | |
| 625 | 1911 hash_init(&lp->sl_wordcount); |
| 1912 } | |
| 1913 | |
| 236 | 1914 return lp; |
| 1915 } | |
| 1916 | |
| 1917 /* | |
| 1918 * Free the contents of an slang_T and the structure itself. | |
| 1919 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1920 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1921 slang_free(slang_T *lp) |
| 236 | 1922 { |
| 1923 vim_free(lp->sl_name); | |
| 310 | 1924 vim_free(lp->sl_fname); |
| 1925 slang_clear(lp); | |
| 1926 vim_free(lp); | |
| 1927 } | |
| 1928 | |
| 1929 /* | |
| 1930 * Clear an slang_T so that the file can be reloaded. | |
| 1931 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1932 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1933 slang_clear(slang_T *lp) |
| 310 | 1934 { |
| 339 | 1935 garray_T *gap; |
| 1936 fromto_T *ftp; | |
| 344 | 1937 salitem_T *smp; |
| 339 | 1938 int i; |
| 625 | 1939 int round; |
| 323 | 1940 |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1941 VIM_CLEAR(lp->sl_fbyts); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1942 VIM_CLEAR(lp->sl_kbyts); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1943 VIM_CLEAR(lp->sl_pbyts); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1944 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1945 VIM_CLEAR(lp->sl_fidxs); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1946 VIM_CLEAR(lp->sl_kidxs); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1947 VIM_CLEAR(lp->sl_pidxs); |
| 323 | 1948 |
| 625 | 1949 for (round = 1; round <= 2; ++round) |
| 1950 { | |
| 1951 gap = round == 1 ? &lp->sl_rep : &lp->sl_repsal; | |
| 1952 while (gap->ga_len > 0) | |
| 1953 { | |
| 1954 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len]; | |
| 1955 vim_free(ftp->ft_from); | |
| 1956 vim_free(ftp->ft_to); | |
| 1957 } | |
| 1958 ga_clear(gap); | |
| 1959 } | |
| 344 | 1960 |
| 1961 gap = &lp->sl_sal; | |
| 375 | 1962 if (lp->sl_sofo) |
| 376 | 1963 { |
| 1964 /* "ga_len" is set to 1 without adding an item for latin1 */ | |
| 1965 if (gap->ga_data != NULL) | |
| 1966 /* SOFOFROM and SOFOTO items: free lists of wide characters. */ | |
| 1967 for (i = 0; i < gap->ga_len; ++i) | |
| 1968 vim_free(((int **)gap->ga_data)[i]); | |
| 1969 } | |
| 375 | 1970 else |
| 1971 /* SAL items: free salitem_T items */ | |
| 1972 while (gap->ga_len > 0) | |
| 1973 { | |
| 1974 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len]; | |
| 1975 vim_free(smp->sm_lead); | |
| 1976 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */ | |
| 1977 vim_free(smp->sm_to); | |
| 1978 vim_free(smp->sm_lead_w); | |
| 1979 vim_free(smp->sm_oneof_w); | |
| 1980 vim_free(smp->sm_to_w); | |
| 1981 } | |
| 344 | 1982 ga_clear(gap); |
| 323 | 1983 |
| 339 | 1984 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
|
1985 vim_regfree(lp->sl_prefprog[i]); |
| 376 | 1986 lp->sl_prefixcnt = 0; |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1987 VIM_CLEAR(lp->sl_prefprog); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1988 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1989 VIM_CLEAR(lp->sl_info); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1990 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1991 VIM_CLEAR(lp->sl_midword); |
| 339 | 1992 |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4352
diff
changeset
|
1993 vim_regfree(lp->sl_compprog); |
| 483 | 1994 lp->sl_compprog = NULL; |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1995 VIM_CLEAR(lp->sl_comprules); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1996 VIM_CLEAR(lp->sl_compstartflags); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1997 VIM_CLEAR(lp->sl_compallflags); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1998 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1999 VIM_CLEAR(lp->sl_syllable); |
| 483 | 2000 ga_clear(&lp->sl_syl_items); |
| 481 | 2001 |
| 809 | 2002 ga_clear_strings(&lp->sl_comppat); |
| 2003 | |
| 625 | 2004 hash_clear_all(&lp->sl_wordcount, WC_KEY_OFF); |
| 2005 hash_init(&lp->sl_wordcount); | |
| 2006 | |
| 2007 hash_clear_all(&lp->sl_map_hash, 0); | |
| 2008 | |
| 2009 /* Clear info from .sug file. */ | |
| 2010 slang_clear_sug(lp); | |
| 483 | 2011 |
| 2012 lp->sl_compmax = MAXWLEN; | |
| 501 | 2013 lp->sl_compminlen = 0; |
| 483 | 2014 lp->sl_compsylmax = MAXWLEN; |
| 2015 lp->sl_regions[0] = NUL; | |
| 236 | 2016 } |
| 2017 | |
| 2018 /* | |
| 625 | 2019 * Clear the info from the .sug file in "lp". |
| 2020 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2021 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2022 slang_clear_sug(slang_T *lp) |
| 625 | 2023 { |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2024 VIM_CLEAR(lp->sl_sbyts); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2025 VIM_CLEAR(lp->sl_sidxs); |
| 625 | 2026 close_spellbuf(lp->sl_sugbuf); |
| 2027 lp->sl_sugbuf = NULL; | |
| 2028 lp->sl_sugloaded = FALSE; | |
| 2029 lp->sl_sugtime = 0; | |
| 2030 } | |
| 2031 | |
| 2032 /* | |
| 307 | 2033 * Load one spell file and store the info into a slang_T. |
| 236 | 2034 * Invoked through do_in_runtimepath(). |
| 2035 */ | |
| 2036 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2037 spell_load_cb(char_u *fname, void *cookie) |
| 501 | 2038 { |
| 2039 spelload_T *slp = (spelload_T *)cookie; | |
| 2040 slang_T *slang; | |
| 2041 | |
| 2042 slang = spell_load_file(fname, slp->sl_lang, NULL, FALSE); | |
| 2043 if (slang != NULL) | |
| 2044 { | |
| 2045 /* When a previously loaded file has NOBREAK also use it for the | |
| 2046 * ".add" files. */ | |
| 2047 if (slp->sl_nobreak && slang->sl_add) | |
| 2048 slang->sl_nobreak = TRUE; | |
| 2049 else if (slang->sl_nobreak) | |
| 2050 slp->sl_nobreak = TRUE; | |
| 2051 | |
| 2052 slp->sl_slang = slang; | |
| 2053 } | |
| 310 | 2054 } |
| 2055 | |
| 625 | 2056 |
| 2057 /* | |
| 2058 * Add a word to the hashtable of common words. | |
| 2059 * If it's already there then the counter is increased. | |
| 2060 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2061 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2062 count_common_word( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2063 slang_T *lp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2064 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2065 int len, /* word length, -1 for upto NUL */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2066 int count) /* 1 to count once, 10 to init */ |
| 625 | 2067 { |
| 2068 hash_T hash; | |
| 2069 hashitem_T *hi; | |
| 2070 wordcount_T *wc; | |
| 2071 char_u buf[MAXWLEN]; | |
| 2072 char_u *p; | |
| 2073 | |
| 2074 if (len == -1) | |
| 2075 p = word; | |
|
17653
cc68aca87c17
patch 8.1.1824: crash when correctly spelled word is very long
Bram Moolenaar <Bram@vim.org>
parents:
17434
diff
changeset
|
2076 else if (len >= MAXWLEN) |
|
cc68aca87c17
patch 8.1.1824: crash when correctly spelled word is very long
Bram Moolenaar <Bram@vim.org>
parents:
17434
diff
changeset
|
2077 return; |
| 625 | 2078 else |
| 2079 { | |
| 2080 vim_strncpy(buf, word, len); | |
| 2081 p = buf; | |
| 2082 } | |
| 2083 | |
| 2084 hash = hash_hash(p); | |
| 2085 hi = hash_lookup(&lp->sl_wordcount, p, hash); | |
| 2086 if (HASHITEM_EMPTY(hi)) | |
| 2087 { | |
|
16825
ce04ebdf26b8
patch 8.1.1414: alloc() returning "char_u *" causes a lot of type casts
Bram Moolenaar <Bram@vim.org>
parents:
16768
diff
changeset
|
2088 wc = alloc(sizeof(wordcount_T) + STRLEN(p)); |
| 625 | 2089 if (wc == NULL) |
| 2090 return; | |
| 2091 STRCPY(wc->wc_word, p); | |
| 2092 wc->wc_count = count; | |
| 2093 hash_add_item(&lp->sl_wordcount, hi, wc->wc_word, hash); | |
| 2094 } | |
| 2095 else | |
| 2096 { | |
| 2097 wc = HI2WC(hi); | |
| 2098 if ((wc->wc_count += count) < (unsigned)count) /* check for overflow */ | |
| 2099 wc->wc_count = MAXWORDCOUNT; | |
| 2100 } | |
| 2101 } | |
| 2102 | |
| 2103 /* | |
| 2104 * Adjust the score of common words. | |
| 2105 */ | |
| 2106 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2107 score_wordcount_adj( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2108 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2109 int score, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2110 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2111 int split) /* word was split, less bonus */ |
| 625 | 2112 { |
| 2113 hashitem_T *hi; | |
| 2114 wordcount_T *wc; | |
| 2115 int bonus; | |
| 2116 int newscore; | |
| 2117 | |
| 2118 hi = hash_find(&slang->sl_wordcount, word); | |
| 2119 if (!HASHITEM_EMPTY(hi)) | |
| 2120 { | |
| 2121 wc = HI2WC(hi); | |
| 2122 if (wc->wc_count < SCORE_THRES2) | |
| 2123 bonus = SCORE_COMMON1; | |
| 2124 else if (wc->wc_count < SCORE_THRES3) | |
| 2125 bonus = SCORE_COMMON2; | |
| 2126 else | |
| 2127 bonus = SCORE_COMMON3; | |
| 2128 if (split) | |
| 2129 newscore = score - bonus / 2; | |
| 2130 else | |
| 2131 newscore = score - bonus; | |
| 2132 if (newscore < 0) | |
| 2133 return 0; | |
| 2134 return newscore; | |
| 2135 } | |
| 2136 return score; | |
| 2137 } | |
| 2138 | |
| 483 | 2139 |
| 495 | 2140 /* |
| 498 | 2141 * Return TRUE if byte "n" appears in "str". |
| 495 | 2142 * Like strchr() but independent of locale. |
| 2143 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2144 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2145 byte_in_str(char_u *str, int n) |
| 495 | 2146 { |
| 2147 char_u *p; | |
| 2148 | |
| 2149 for (p = str; *p != NUL; ++p) | |
| 498 | 2150 if (*p == n) |
| 495 | 2151 return TRUE; |
| 2152 return FALSE; | |
| 2153 } | |
| 2154 | |
| 483 | 2155 #define SY_MAXLEN 30 |
| 2156 typedef struct syl_item_S | |
| 2157 { | |
| 2158 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */ | |
| 2159 int sy_len; | |
| 2160 } syl_item_T; | |
| 2161 | |
| 2162 /* | |
| 2163 * Truncate "slang->sl_syllable" at the first slash and put the following items | |
| 2164 * in "slang->sl_syl_items". | |
| 2165 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2166 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2167 init_syl_tab(slang_T *slang) |
| 483 | 2168 { |
| 2169 char_u *p; | |
| 2170 char_u *s; | |
| 2171 int l; | |
| 2172 syl_item_T *syl; | |
| 2173 | |
| 2174 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4); | |
| 2175 p = vim_strchr(slang->sl_syllable, '/'); | |
| 2176 while (p != NULL) | |
| 2177 { | |
| 2178 *p++ = NUL; | |
| 495 | 2179 if (*p == NUL) /* trailing slash */ |
| 483 | 2180 break; |
| 2181 s = p; | |
| 2182 p = vim_strchr(p, '/'); | |
| 2183 if (p == NULL) | |
| 835 | 2184 l = (int)STRLEN(s); |
| 483 | 2185 else |
| 835 | 2186 l = (int)(p - s); |
| 483 | 2187 if (l >= SY_MAXLEN) |
| 2188 return SP_FORMERROR; | |
| 2189 if (ga_grow(&slang->sl_syl_items, 1) == FAIL) | |
| 495 | 2190 return SP_OTHERERROR; |
| 483 | 2191 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) |
| 2192 + slang->sl_syl_items.ga_len++; | |
| 2193 vim_strncpy(syl->sy_chars, s, l); | |
| 2194 syl->sy_len = l; | |
| 2195 } | |
| 2196 return OK; | |
| 2197 } | |
| 2198 | |
| 2199 /* | |
| 2200 * Count the number of syllables in "word". | |
| 2201 * When "word" contains spaces the syllables after the last space are counted. | |
| 2202 * Returns zero if syllables are not defines. | |
| 2203 */ | |
| 2204 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2205 count_syllables(slang_T *slang, char_u *word) |
| 483 | 2206 { |
| 2207 int cnt = 0; | |
| 2208 int skip = FALSE; | |
| 2209 char_u *p; | |
| 2210 int len; | |
| 2211 int i; | |
| 2212 syl_item_T *syl; | |
| 2213 int c; | |
| 2214 | |
| 2215 if (slang->sl_syllable == NULL) | |
| 2216 return 0; | |
| 2217 | |
| 2218 for (p = word; *p != NUL; p += len) | |
| 2219 { | |
| 2220 /* When running into a space reset counter. */ | |
| 2221 if (*p == ' ') | |
| 2222 { | |
| 2223 len = 1; | |
| 2224 cnt = 0; | |
| 2225 continue; | |
| 2226 } | |
| 2227 | |
| 2228 /* Find longest match of syllable items. */ | |
| 2229 len = 0; | |
| 2230 for (i = 0; i < slang->sl_syl_items.ga_len; ++i) | |
| 2231 { | |
| 2232 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i; | |
| 2233 if (syl->sy_len > len | |
| 2234 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0) | |
| 2235 len = syl->sy_len; | |
| 2236 } | |
| 2237 if (len != 0) /* found a match, count syllable */ | |
| 2238 { | |
| 2239 ++cnt; | |
| 2240 skip = FALSE; | |
| 2241 } | |
| 2242 else | |
| 2243 { | |
| 2244 /* No recognized syllable item, at least a syllable char then? */ | |
| 2245 c = mb_ptr2char(p); | |
| 2246 len = (*mb_ptr2len)(p); | |
| 2247 if (vim_strchr(slang->sl_syllable, c) == NULL) | |
| 2248 skip = FALSE; /* No, search for next syllable */ | |
| 2249 else if (!skip) | |
| 2250 { | |
| 2251 ++cnt; /* Yes, count it */ | |
| 2252 skip = TRUE; /* don't count following syllable chars */ | |
| 2253 } | |
| 2254 } | |
| 2255 } | |
| 2256 return cnt; | |
| 376 | 2257 } |
| 2258 | |
| 381 | 2259 /* |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2260 * Parse 'spelllang' and set w_s->b_langp accordingly. |
| 351 | 2261 * Returns NULL if it's OK, an error message otherwise. |
| 236 | 2262 */ |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
2263 char * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2264 did_set_spelllang(win_T *wp) |
| 236 | 2265 { |
| 2266 garray_T ga; | |
| 351 | 2267 char_u *splp; |
| 236 | 2268 char_u *region; |
| 416 | 2269 char_u region_cp[3]; |
| 355 | 2270 int filename; |
| 236 | 2271 int region_mask; |
| 503 | 2272 slang_T *slang; |
| 236 | 2273 int c; |
| 351 | 2274 char_u lang[MAXWLEN + 1]; |
| 323 | 2275 char_u spf_name[MAXPATHL]; |
| 351 | 2276 int len; |
| 2277 char_u *p; | |
| 381 | 2278 int round; |
| 385 | 2279 char_u *spf; |
| 389 | 2280 char_u *use_region = NULL; |
| 2281 int dont_use_region = FALSE; | |
| 501 | 2282 int nobreak = FALSE; |
| 503 | 2283 int i, j; |
| 2284 langp_T *lp, *lp2; | |
| 1185 | 2285 static int recursive = FALSE; |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
2286 char *ret_msg = NULL; |
| 1185 | 2287 char_u *spl_copy; |
|
9487
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2288 bufref_T bufref; |
|
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2289 |
|
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2290 set_bufref(&bufref, wp->w_buffer); |
| 1185 | 2291 |
| 2292 /* We don't want to do this recursively. May happen when a language is | |
| 2293 * not available and the SpellFileMissing autocommand opens a new buffer | |
| 2294 * in which 'spell' is set. */ | |
| 2295 if (recursive) | |
| 2296 return NULL; | |
| 2297 recursive = TRUE; | |
| 236 | 2298 |
| 2299 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
|
2300 clear_midword(wp); |
| 236 | 2301 |
| 4352 | 2302 /* Make a copy of 'spelllang', the SpellFileMissing autocommands may change |
| 1185 | 2303 * it under our fingers. */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2304 spl_copy = vim_strsave(wp->w_s->b_p_spl); |
| 1185 | 2305 if (spl_copy == NULL) |
| 2306 goto theend; | |
| 2307 | |
| 5477 | 2308 wp->w_s->b_cjk = 0; |
| 2309 | |
| 2310 /* Loop over comma separated language names. */ | |
| 1185 | 2311 for (splp = spl_copy; *splp != NUL; ) |
| 351 | 2312 { |
|
16277
5ef25fa57f71
patch 8.1.1143: may pass weird strings to file name expansion
Bram Moolenaar <Bram@vim.org>
parents:
16239
diff
changeset
|
2313 // Get one language name. |
| 351 | 2314 copy_option_part(&splp, lang, MAXWLEN, ","); |
| 240 | 2315 region = NULL; |
| 835 | 2316 len = (int)STRLEN(lang); |
| 355 | 2317 |
|
16277
5ef25fa57f71
patch 8.1.1143: may pass weird strings to file name expansion
Bram Moolenaar <Bram@vim.org>
parents:
16239
diff
changeset
|
2318 if (!valid_spellang(lang)) |
|
5ef25fa57f71
patch 8.1.1143: may pass weird strings to file name expansion
Bram Moolenaar <Bram@vim.org>
parents:
16239
diff
changeset
|
2319 continue; |
|
5ef25fa57f71
patch 8.1.1143: may pass weird strings to file name expansion
Bram Moolenaar <Bram@vim.org>
parents:
16239
diff
changeset
|
2320 |
| 5477 | 2321 if (STRCMP(lang, "cjk") == 0) |
| 2322 { | |
| 2323 wp->w_s->b_cjk = 1; | |
| 2324 continue; | |
| 2325 } | |
| 2326 | |
| 355 | 2327 /* If the name ends in ".spl" use it as the name of the spell file. |
| 2328 * If there is a region name let "region" point to it and remove it | |
| 2329 * from the name. */ | |
| 2330 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0) | |
| 236 | 2331 { |
| 355 | 2332 filename = TRUE; |
| 2333 | |
| 416 | 2334 /* Locate a region and remove it from the file name. */ |
| 2335 p = vim_strchr(gettail(lang), '_'); | |
| 2336 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2]) | |
| 2337 && !ASCII_ISALPHA(p[3])) | |
| 2338 { | |
| 2339 vim_strncpy(region_cp, p + 1, 2); | |
| 2340 mch_memmove(p, p + 3, len - (p - lang) - 2); | |
| 2341 region = region_cp; | |
| 2342 } | |
| 2343 else | |
| 2344 dont_use_region = TRUE; | |
| 2345 | |
| 355 | 2346 /* Check if we loaded this language before. */ |
| 503 | 2347 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
|
16738
b52ea9c5f1db
patch 8.1.1371: cannot recover from a swap file
Bram Moolenaar <Bram@vim.org>
parents:
16606
diff
changeset
|
2348 if (fullpathcmp(lang, slang->sl_fname, FALSE, TRUE) == FPC_SAME) |
| 355 | 2349 break; |
| 236 | 2350 } |
| 355 | 2351 else |
| 2352 { | |
| 2353 filename = FALSE; | |
| 2354 if (len > 3 && lang[len - 3] == '_') | |
| 2355 { | |
| 2356 region = lang + len - 2; | |
| 2357 len -= 3; | |
| 2358 lang[len] = NUL; | |
| 389 | 2359 } |
| 2360 else | |
| 2361 dont_use_region = TRUE; | |
| 355 | 2362 |
| 2363 /* Check if we loaded this language before. */ | |
| 503 | 2364 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2365 if (STRICMP(lang, slang->sl_name) == 0) | |
| 355 | 2366 break; |
| 2367 } | |
| 236 | 2368 |
| 416 | 2369 if (region != NULL) |
| 2370 { | |
| 2371 /* If the region differs from what was used before then don't | |
| 2372 * use it for 'spellfile'. */ | |
| 2373 if (use_region != NULL && STRCMP(region, use_region) != 0) | |
| 2374 dont_use_region = TRUE; | |
| 2375 use_region = region; | |
| 2376 } | |
| 2377 | |
| 351 | 2378 /* If not found try loading the language now. */ |
| 503 | 2379 if (slang == NULL) |
| 355 | 2380 { |
| 2381 if (filename) | |
| 2382 (void)spell_load_file(lang, lang, NULL, FALSE); | |
| 2383 else | |
| 1185 | 2384 { |
| 355 | 2385 spell_load_lang(lang); |
| 1185 | 2386 /* SpellFileMissing autocommands may do anything, including |
| 2387 * destroying the buffer we are using... */ | |
|
9487
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2388 if (!bufref_valid(&bufref)) |
| 1185 | 2389 { |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
2390 ret_msg = N_("E797: SpellFileMissing autocommand deleted buffer"); |
| 1185 | 2391 goto theend; |
| 2392 } | |
| 2393 } | |
| 355 | 2394 } |
| 236 | 2395 |
| 307 | 2396 /* |
| 351 | 2397 * Loop over the languages, there can be several files for "lang". |
| 307 | 2398 */ |
| 503 | 2399 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
|
16738
b52ea9c5f1db
patch 8.1.1371: cannot recover from a swap file
Bram Moolenaar <Bram@vim.org>
parents:
16606
diff
changeset
|
2400 if (filename ? fullpathcmp(lang, slang->sl_fname, FALSE, TRUE) |
|
b52ea9c5f1db
patch 8.1.1371: cannot recover from a swap file
Bram Moolenaar <Bram@vim.org>
parents:
16606
diff
changeset
|
2401 == FPC_SAME |
| 503 | 2402 : STRICMP(lang, slang->sl_name) == 0) |
| 236 | 2403 { |
| 316 | 2404 region_mask = REGION_ALL; |
| 355 | 2405 if (!filename && region != NULL) |
| 236 | 2406 { |
| 307 | 2407 /* find region in sl_regions */ |
| 503 | 2408 c = find_region(slang->sl_regions, region); |
| 307 | 2409 if (c == REGION_ALL) |
| 2410 { | |
| 503 | 2411 if (slang->sl_add) |
| 389 | 2412 { |
| 503 | 2413 if (*slang->sl_regions != NUL) |
| 389 | 2414 /* This addition file is for other regions. */ |
| 2415 region_mask = 0; | |
| 2416 } | |
| 2417 else | |
| 2418 /* This is probably an error. Give a warning and | |
| 2419 * accept the words anyway. */ | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
2420 smsg(_("Warning: region %s not supported"), |
| 351 | 2421 region); |
| 307 | 2422 } |
| 2423 else | |
| 2424 region_mask = 1 << c; | |
| 236 | 2425 } |
| 307 | 2426 |
| 389 | 2427 if (region_mask != 0) |
| 307 | 2428 { |
| 389 | 2429 if (ga_grow(&ga, 1) == FAIL) |
| 2430 { | |
| 2431 ga_clear(&ga); | |
| 1185 | 2432 ret_msg = e_outofmem; |
| 2433 goto theend; | |
| 389 | 2434 } |
| 503 | 2435 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 389 | 2436 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 2437 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2438 use_midword(slang, wp); |
| 503 | 2439 if (slang->sl_nobreak) |
| 501 | 2440 nobreak = TRUE; |
| 307 | 2441 } |
| 385 | 2442 } |
| 2443 } | |
| 2444 | |
| 2445 /* round 0: load int_wordlist, if possible. | |
| 2446 * round 1: load first name in 'spellfile'. | |
| 2447 * round 2: load second name in 'spellfile. | |
| 2448 * etc. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2449 spf = curwin->w_s->b_p_spf; |
| 385 | 2450 for (round = 0; round == 0 || *spf != NUL; ++round) |
| 2451 { | |
| 2452 if (round == 0) | |
| 2453 { | |
| 2454 /* Internal wordlist, if there is one. */ | |
| 2455 if (int_wordlist == NULL) | |
| 381 | 2456 continue; |
| 385 | 2457 int_wordlist_spl(spf_name); |
| 381 | 2458 } |
| 2459 else | |
| 2460 { | |
| 385 | 2461 /* One entry in 'spellfile'. */ |
| 2462 copy_option_part(&spf, spf_name, MAXPATHL - 5, ","); | |
| 2463 STRCAT(spf_name, ".spl"); | |
| 2464 | |
| 2465 /* If it was already found above then skip it. */ | |
| 2466 for (c = 0; c < ga.ga_len; ++c) | |
| 500 | 2467 { |
| 2468 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname; | |
|
16738
b52ea9c5f1db
patch 8.1.1371: cannot recover from a swap file
Bram Moolenaar <Bram@vim.org>
parents:
16606
diff
changeset
|
2469 if (p != NULL && fullpathcmp(spf_name, p, FALSE, TRUE) |
|
b52ea9c5f1db
patch 8.1.1371: cannot recover from a swap file
Bram Moolenaar <Bram@vim.org>
parents:
16606
diff
changeset
|
2470 == FPC_SAME) |
| 385 | 2471 break; |
| 500 | 2472 } |
| 385 | 2473 if (c < ga.ga_len) |
| 381 | 2474 continue; |
| 2475 } | |
| 2476 | |
| 351 | 2477 /* Check if it was loaded already. */ |
| 503 | 2478 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
|
16738
b52ea9c5f1db
patch 8.1.1371: cannot recover from a swap file
Bram Moolenaar <Bram@vim.org>
parents:
16606
diff
changeset
|
2479 if (fullpathcmp(spf_name, slang->sl_fname, FALSE, TRUE) |
|
b52ea9c5f1db
patch 8.1.1371: cannot recover from a swap file
Bram Moolenaar <Bram@vim.org>
parents:
16606
diff
changeset
|
2480 == FPC_SAME) |
| 323 | 2481 break; |
| 503 | 2482 if (slang == NULL) |
| 323 | 2483 { |
| 351 | 2484 /* Not loaded, try loading it now. The language name includes the |
| 385 | 2485 * region name, the region is ignored otherwise. for int_wordlist |
| 2486 * use an arbitrary name. */ | |
| 2487 if (round == 0) | |
| 2488 STRCPY(lang, "internal wordlist"); | |
| 2489 else | |
| 2490 { | |
| 2491 vim_strncpy(lang, gettail(spf_name), MAXWLEN); | |
| 381 | 2492 p = vim_strchr(lang, '.'); |
| 2493 if (p != NULL) | |
| 2494 *p = NUL; /* truncate at ".encoding.add" */ | |
| 2495 } | |
| 503 | 2496 slang = spell_load_file(spf_name, lang, NULL, TRUE); |
| 501 | 2497 |
| 2498 /* If one of the languages has NOBREAK we assume the addition | |
| 2499 * files also have this. */ | |
| 503 | 2500 if (slang != NULL && nobreak) |
| 2501 slang->sl_nobreak = TRUE; | |
| 2502 } | |
| 2503 if (slang != NULL && ga_grow(&ga, 1) == OK) | |
| 323 | 2504 { |
| 389 | 2505 region_mask = REGION_ALL; |
| 2506 if (use_region != NULL && !dont_use_region) | |
| 2507 { | |
| 2508 /* find region in sl_regions */ | |
| 503 | 2509 c = find_region(slang->sl_regions, use_region); |
| 389 | 2510 if (c != REGION_ALL) |
| 2511 region_mask = 1 << c; | |
| 503 | 2512 else if (*slang->sl_regions != NUL) |
| 389 | 2513 /* This spell file is for other regions. */ |
| 2514 region_mask = 0; | |
| 2515 } | |
| 2516 | |
| 2517 if (region_mask != 0) | |
| 2518 { | |
| 503 | 2519 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 2520 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL; | |
| 2521 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL; | |
| 389 | 2522 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 2523 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2524 use_midword(slang, wp); |
| 389 | 2525 } |
| 323 | 2526 } |
| 2527 } | |
| 2528 | |
| 236 | 2529 /* 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
|
2530 ga_clear(&wp->w_s->b_langp); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2531 wp->w_s->b_langp = ga; |
| 236 | 2532 |
| 503 | 2533 /* For each language figure out what language to use for sound folding and |
| 2534 * REP items. If the language doesn't support it itself use another one | |
| 2535 * with the same name. E.g. for "en-math" use "en". */ | |
| 2536 for (i = 0; i < ga.ga_len; ++i) | |
| 2537 { | |
| 2538 lp = LANGP_ENTRY(ga, i); | |
| 2539 | |
| 2540 /* sound folding */ | |
| 2541 if (lp->lp_slang->sl_sal.ga_len > 0) | |
| 2542 /* language does sound folding itself */ | |
| 2543 lp->lp_sallang = lp->lp_slang; | |
| 2544 else | |
| 2545 /* find first similar language that does sound folding */ | |
| 2546 for (j = 0; j < ga.ga_len; ++j) | |
| 2547 { | |
| 2548 lp2 = LANGP_ENTRY(ga, j); | |
| 2549 if (lp2->lp_slang->sl_sal.ga_len > 0 | |
| 2550 && STRNCMP(lp->lp_slang->sl_name, | |
| 2551 lp2->lp_slang->sl_name, 2) == 0) | |
| 2552 { | |
| 2553 lp->lp_sallang = lp2->lp_slang; | |
| 2554 break; | |
| 2555 } | |
| 2556 } | |
| 2557 | |
| 2558 /* REP items */ | |
| 2559 if (lp->lp_slang->sl_rep.ga_len > 0) | |
| 2560 /* language has REP items itself */ | |
| 2561 lp->lp_replang = lp->lp_slang; | |
| 2562 else | |
| 625 | 2563 /* find first similar language that has REP items */ |
| 503 | 2564 for (j = 0; j < ga.ga_len; ++j) |
| 2565 { | |
| 2566 lp2 = LANGP_ENTRY(ga, j); | |
| 2567 if (lp2->lp_slang->sl_rep.ga_len > 0 | |
| 2568 && STRNCMP(lp->lp_slang->sl_name, | |
| 2569 lp2->lp_slang->sl_name, 2) == 0) | |
| 2570 { | |
| 2571 lp->lp_replang = lp2->lp_slang; | |
| 2572 break; | |
| 2573 } | |
| 2574 } | |
| 2575 } | |
| 2576 | |
| 1185 | 2577 theend: |
| 2578 vim_free(spl_copy); | |
| 2579 recursive = FALSE; | |
| 5891 | 2580 redraw_win_later(wp, NOT_VALID); |
| 1185 | 2581 return ret_msg; |
| 236 | 2582 } |
| 2583 | |
| 2584 /* | |
| 376 | 2585 * Clear the midword characters for buffer "buf". |
| 2586 */ | |
| 2587 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2588 clear_midword(win_T *wp) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2589 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2590 vim_memset(wp->w_s->b_spell_ismw, 0, 256); |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2591 VIM_CLEAR(wp->w_s->b_spell_ismw_mb); |
| 376 | 2592 } |
| 2593 | |
| 2594 /* | |
| 2595 * Use the "sl_midword" field of language "lp" for buffer "buf". | |
| 2596 * They add up to any currently used midword characters. | |
| 2597 */ | |
| 2598 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2599 use_midword(slang_T *lp, win_T *wp) |
| 376 | 2600 { |
| 2601 char_u *p; | |
| 2602 | |
| 389 | 2603 if (lp->sl_midword == NULL) /* there aren't any */ |
| 2604 return; | |
| 2605 | |
| 376 | 2606 for (p = lp->sl_midword; *p != NUL; ) |
| 2607 if (has_mbyte) | |
| 2608 { | |
| 2609 int c, l, n; | |
| 2610 char_u *bp; | |
| 2611 | |
| 2612 c = mb_ptr2char(p); | |
| 474 | 2613 l = (*mb_ptr2len)(p); |
| 2614 if (c < 256 && l <= 2) | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2615 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
|
2616 else if (wp->w_s->b_spell_ismw_mb == NULL) |
| 376 | 2617 /* 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
|
2618 wp->w_s->b_spell_ismw_mb = vim_strnsave(p, l); |
| 376 | 2619 else |
| 2620 { | |
| 2621 /* 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
|
2622 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
|
2623 bp = vim_strnsave(wp->w_s->b_spell_ismw_mb, n + l); |
| 376 | 2624 if (bp != NULL) |
| 2625 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2626 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
|
2627 wp->w_s->b_spell_ismw_mb = bp; |
| 376 | 2628 vim_strncpy(bp + n, p, l); |
| 2629 } | |
| 2630 } | |
| 2631 p += l; | |
| 2632 } | |
| 2633 else | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2634 wp->w_s->b_spell_ismw[*p++] = TRUE; |
| 376 | 2635 } |
| 2636 | |
| 2637 /* | |
| 236 | 2638 * Find the region "region[2]" in "rp" (points to "sl_regions"). |
|
15034
6e4e0d43b20b
patch 8.1.0528: various typos in comments
Bram Moolenaar <Bram@vim.org>
parents:
14862
diff
changeset
|
2639 * Each region is simply stored as the two characters of its name. |
| 381 | 2640 * Returns the index if found (first is 0), REGION_ALL if not found. |
| 236 | 2641 */ |
| 2642 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2643 find_region(char_u *rp, char_u *region) |
| 236 | 2644 { |
| 2645 int i; | |
| 2646 | |
| 2647 for (i = 0; ; i += 2) | |
| 2648 { | |
| 2649 if (rp[i] == NUL) | |
| 2650 return REGION_ALL; | |
| 2651 if (rp[i] == region[0] && rp[i + 1] == region[1]) | |
| 2652 break; | |
| 2653 } | |
| 2654 return i / 2; | |
| 2655 } | |
| 2656 | |
| 2657 /* | |
| 323 | 2658 * Return case type of word: |
| 236 | 2659 * w word 0 |
| 300 | 2660 * Word WF_ONECAP |
| 2661 * W WORD WF_ALLCAP | |
| 2662 * WoRd wOrd WF_KEEPCAP | |
| 236 | 2663 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2664 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2665 captype( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2666 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2667 char_u *end) /* When NULL use up to NUL byte. */ |
| 236 | 2668 { |
| 2669 char_u *p; | |
| 2670 int c; | |
| 2671 int firstcap; | |
| 2672 int allcap; | |
| 2673 int past_second = FALSE; /* past second word char */ | |
| 2674 | |
| 2675 /* find first letter */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2676 for (p = word; !spell_iswordp_nmw(p, curwin); MB_PTR_ADV(p)) |
| 323 | 2677 if (end == NULL ? *p == NUL : p >= end) |
| 236 | 2678 return 0; /* only non-word characters, illegal word */ |
| 310 | 2679 if (has_mbyte) |
| 2680 c = mb_ptr2char_adv(&p); | |
| 2681 else | |
| 2682 c = *p++; | |
| 324 | 2683 firstcap = allcap = SPELL_ISUPPER(c); |
| 236 | 2684 |
| 2685 /* | |
| 2686 * Need to check all letters to find a word with mixed upper/lower. | |
| 2687 * But a word with an upper char only at start is a ONECAP. | |
| 2688 */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2689 for ( ; end == NULL ? *p != NUL : p < end; MB_PTR_ADV(p)) |
| 5477 | 2690 if (spell_iswordp_nmw(p, curwin)) |
| 236 | 2691 { |
| 455 | 2692 c = PTR2CHAR(p); |
| 324 | 2693 if (!SPELL_ISUPPER(c)) |
| 236 | 2694 { |
| 2695 /* UUl -> KEEPCAP */ | |
| 2696 if (past_second && allcap) | |
| 300 | 2697 return WF_KEEPCAP; |
| 236 | 2698 allcap = FALSE; |
| 2699 } | |
| 2700 else if (!allcap) | |
| 2701 /* UlU -> KEEPCAP */ | |
| 300 | 2702 return WF_KEEPCAP; |
| 236 | 2703 past_second = TRUE; |
| 2704 } | |
| 2705 | |
| 2706 if (allcap) | |
| 300 | 2707 return WF_ALLCAP; |
| 236 | 2708 if (firstcap) |
| 300 | 2709 return WF_ONECAP; |
| 236 | 2710 return 0; |
| 2711 } | |
| 2712 | |
| 474 | 2713 /* |
| 2714 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a | |
| 2715 * capital. So that make_case_word() can turn WOrd into Word. | |
| 2716 * Add ALLCAP for "WOrD". | |
| 2717 */ | |
| 2718 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2719 badword_captype(char_u *word, char_u *end) |
| 474 | 2720 { |
| 2721 int flags = captype(word, end); | |
| 475 | 2722 int c; |
| 474 | 2723 int l, u; |
| 2724 int first; | |
| 2725 char_u *p; | |
| 2726 | |
| 2727 if (flags & WF_KEEPCAP) | |
| 2728 { | |
| 2729 /* Count the number of UPPER and lower case letters. */ | |
| 2730 l = u = 0; | |
| 2731 first = FALSE; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2732 for (p = word; p < end; MB_PTR_ADV(p)) |
| 474 | 2733 { |
| 475 | 2734 c = PTR2CHAR(p); |
| 2735 if (SPELL_ISUPPER(c)) | |
| 474 | 2736 { |
| 2737 ++u; | |
| 2738 if (p == word) | |
| 2739 first = TRUE; | |
| 2740 } | |
| 2741 else | |
| 2742 ++l; | |
| 2743 } | |
| 2744 | |
| 2745 /* If there are more UPPER than lower case letters suggest an | |
| 2746 * ALLCAP word. Otherwise, if the first letter is UPPER then | |
| 2747 * suggest ONECAP. Exception: "ALl" most likely should be "All", | |
| 2748 * require three upper case letters. */ | |
| 2749 if (u > l && u > 2) | |
| 2750 flags |= WF_ALLCAP; | |
| 2751 else if (first) | |
| 2752 flags |= WF_ONECAP; | |
| 633 | 2753 |
| 2754 if (u >= 2 && l >= 2) /* maCARONI maCAroni */ | |
| 2755 flags |= WF_MIXCAP; | |
| 474 | 2756 } |
| 2757 return flags; | |
| 2758 } | |
| 2759 | |
| 5519 | 2760 /* |
| 2761 * Delete the internal wordlist and its .spl file. | |
| 2762 */ | |
| 2763 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2764 spell_delete_wordlist(void) |
| 5519 | 2765 { |
| 2766 char_u fname[MAXPATHL]; | |
| 2767 | |
| 2768 if (int_wordlist != NULL) | |
| 2769 { | |
| 2770 mch_remove(int_wordlist); | |
| 2771 int_wordlist_spl(fname); | |
| 2772 mch_remove(fname); | |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2773 VIM_CLEAR(int_wordlist); |
| 5519 | 2774 } |
| 2775 } | |
| 2776 | |
| 355 | 2777 /* |
| 2778 * Free all languages. | |
| 2779 */ | |
| 2780 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2781 spell_free_all(void) |
| 355 | 2782 { |
| 503 | 2783 slang_T *slang; |
| 355 | 2784 buf_T *buf; |
| 2785 | |
| 2580 | 2786 /* Go through all buffers and handle 'spelllang'. <VN> */ |
|
9649
fd9727ae3c49
commit https://github.com/vim/vim/commit/2932359000b2f918d5fade79ea4d124d5943cd07
Christian Brabandt <cb@256bit.org>
parents:
9583
diff
changeset
|
2787 FOR_ALL_BUFFERS(buf) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2788 ga_clear(&buf->b_s.b_langp); |
| 355 | 2789 |
| 2790 while (first_lang != NULL) | |
| 2791 { | |
| 503 | 2792 slang = first_lang; |
| 2793 first_lang = slang->sl_next; | |
| 2794 slang_free(slang); | |
| 355 | 2795 } |
| 366 | 2796 |
| 5519 | 2797 spell_delete_wordlist(); |
| 381 | 2798 |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2799 VIM_CLEAR(repl_to); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2800 VIM_CLEAR(repl_from); |
| 355 | 2801 } |
|
15605
62b3805506b3
patch 8.1.0810: too many #ifdefs
Bram Moolenaar <Bram@vim.org>
parents:
15543
diff
changeset
|
2802 |
| 236 | 2803 /* |
| 2804 * Clear all spelling tables and reload them. | |
| 307 | 2805 * Used after 'encoding' is set and when ":mkspell" was used. |
| 236 | 2806 */ |
| 2807 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2808 spell_reload(void) |
| 236 | 2809 { |
| 316 | 2810 win_T *wp; |
| 236 | 2811 |
| 358 | 2812 /* Initialize the table for spell_iswordp(). */ |
| 236 | 2813 init_spell_chartab(); |
| 2814 | |
| 2815 /* Unload all allocated memory. */ | |
| 355 | 2816 spell_free_all(); |
| 236 | 2817 |
| 2818 /* Go through all buffers and handle 'spelllang'. */ | |
|
9649
fd9727ae3c49
commit https://github.com/vim/vim/commit/2932359000b2f918d5fade79ea4d124d5943cd07
Christian Brabandt <cb@256bit.org>
parents:
9583
diff
changeset
|
2819 FOR_ALL_WINDOWS(wp) |
| 236 | 2820 { |
| 316 | 2821 /* Only load the wordlists when 'spelllang' is set and there is a |
| 2822 * 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
|
2823 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
|
2824 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2825 if (wp->w_p_spell) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2826 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2827 (void)did_set_spelllang(wp); |
| 316 | 2828 break; |
| 2829 } | |
| 2830 } | |
| 236 | 2831 } |
| 2832 } | |
| 2833 | |
| 310 | 2834 /* |
| 625 | 2835 * Opposite of offset2bytes(). |
| 2836 * "pp" points to the bytes and is advanced over it. | |
| 2837 * Returns the offset. | |
| 2838 */ | |
| 2839 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2840 bytes2offset(char_u **pp) |
| 625 | 2841 { |
| 2842 char_u *p = *pp; | |
| 2843 int nr; | |
| 2844 int c; | |
| 2845 | |
| 2846 c = *p++; | |
| 2847 if ((c & 0x80) == 0x00) /* 1 byte */ | |
| 2848 { | |
| 2849 nr = c - 1; | |
| 2850 } | |
| 2851 else if ((c & 0xc0) == 0x80) /* 2 bytes */ | |
| 2852 { | |
| 2853 nr = (c & 0x3f) - 1; | |
| 2854 nr = nr * 255 + (*p++ - 1); | |
| 2855 } | |
| 2856 else if ((c & 0xe0) == 0xc0) /* 3 bytes */ | |
| 2857 { | |
| 2858 nr = (c & 0x1f) - 1; | |
| 2859 nr = nr * 255 + (*p++ - 1); | |
| 2860 nr = nr * 255 + (*p++ - 1); | |
| 2861 } | |
| 2862 else /* 4 bytes */ | |
| 2863 { | |
| 2864 nr = (c & 0x0f) - 1; | |
| 2865 nr = nr * 255 + (*p++ - 1); | |
| 2866 nr = nr * 255 + (*p++ - 1); | |
| 2867 nr = nr * 255 + (*p++ - 1); | |
| 2868 } | |
| 2869 | |
| 2870 *pp = p; | |
| 2871 return nr; | |
| 2872 } | |
| 2873 | |
| 2874 | |
| 2875 /* | |
| 2876 * Open a spell buffer. This is a nameless buffer that is not in the buffer | |
| 2877 * list and only contains text lines. Can use a swapfile to reduce memory | |
| 2878 * use. | |
| 2879 * Most other fields are invalid! Esp. watch out for string options being | |
| 2880 * NULL and there is no undo info. | |
| 2881 * Returns NULL when out of memory. | |
| 2882 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2883 buf_T * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2884 open_spellbuf(void) |
| 625 | 2885 { |
| 2886 buf_T *buf; | |
| 2887 | |
|
16825
ce04ebdf26b8
patch 8.1.1414: alloc() returning "char_u *" causes a lot of type casts
Bram Moolenaar <Bram@vim.org>
parents:
16768
diff
changeset
|
2888 buf = ALLOC_CLEAR_ONE(buf_T); |
| 625 | 2889 if (buf != NULL) |
| 2890 { | |
| 2891 buf->b_spell = TRUE; | |
| 2892 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
|
2893 #ifdef FEAT_CRYPT |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
2894 buf->b_p_key = empty_option; |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
2895 #endif |
| 625 | 2896 ml_open(buf); |
| 2897 ml_open_file(buf); /* create swap file now */ | |
| 2898 } | |
| 2899 return buf; | |
| 2900 } | |
| 2901 | |
| 2902 /* | |
| 2903 * Close the buffer used for spell info. | |
| 2904 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2905 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2906 close_spellbuf(buf_T *buf) |
| 625 | 2907 { |
| 2908 if (buf != NULL) | |
| 2909 { | |
| 2910 ml_close(buf, TRUE); | |
| 2911 vim_free(buf); | |
| 2912 } | |
| 2913 } | |
| 2914 | |
| 307 | 2915 /* |
| 2916 * Init the chartab used for spelling for ASCII. | |
| 2917 * EBCDIC is not supported! | |
| 2918 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2919 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2920 clear_spell_chartab(spelltab_T *sp) |
| 307 | 2921 { |
| 324 | 2922 int i; |
| 307 | 2923 |
| 2924 /* Init everything to FALSE. */ | |
| 2925 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw)); | |
| 2926 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu)); | |
| 2927 for (i = 0; i < 256; ++i) | |
| 324 | 2928 { |
| 307 | 2929 sp->st_fold[i] = i; |
| 324 | 2930 sp->st_upper[i] = i; |
| 2931 } | |
| 307 | 2932 |
| 2933 /* We include digits. A word shouldn't start with a digit, but handling | |
| 2934 * that is done separately. */ | |
| 2935 for (i = '0'; i <= '9'; ++i) | |
| 2936 sp->st_isw[i] = TRUE; | |
| 2937 for (i = 'A'; i <= 'Z'; ++i) | |
| 2938 { | |
| 2939 sp->st_isw[i] = TRUE; | |
| 2940 sp->st_isu[i] = TRUE; | |
| 2941 sp->st_fold[i] = i + 0x20; | |
| 2942 } | |
| 2943 for (i = 'a'; i <= 'z'; ++i) | |
| 324 | 2944 { |
| 307 | 2945 sp->st_isw[i] = TRUE; |
| 324 | 2946 sp->st_upper[i] = i - 0x20; |
| 2947 } | |
| 307 | 2948 } |
| 2949 | |
| 2950 /* | |
| 2951 * Init the chartab used for spelling. Only depends on 'encoding'. | |
| 2952 * Called once while starting up and when 'encoding' changes. | |
| 2953 * The default is to use isalpha(), but the spell file should define the word | |
| 2954 * characters to make it possible that 'encoding' differs from the current | |
| 390 | 2955 * locale. For utf-8 we don't use isalpha() but our own functions. |
| 307 | 2956 */ |
| 2957 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2958 init_spell_chartab(void) |
| 307 | 2959 { |
| 2960 int i; | |
| 2961 | |
| 2962 did_set_spelltab = FALSE; | |
| 2963 clear_spell_chartab(&spelltab); | |
| 2964 if (enc_dbcs) | |
| 2965 { | |
| 2966 /* DBCS: assume double-wide characters are word characters. */ | |
| 2967 for (i = 128; i <= 255; ++i) | |
| 2968 if (MB_BYTE2LEN(i) == 2) | |
| 2969 spelltab.st_isw[i] = TRUE; | |
| 2970 } | |
| 324 | 2971 else if (enc_utf8) |
| 2972 { | |
| 2973 for (i = 128; i < 256; ++i) | |
| 2974 { | |
|
2140
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
2975 int f = utf_fold(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
2976 int u = utf_toupper(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
2977 |
| 324 | 2978 spelltab.st_isu[i] = utf_isupper(i); |
| 2979 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
|
2980 /* 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
|
2981 * 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
|
2982 * value for utf-8 to avoid this. */ |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
2983 spelltab.st_fold[i] = (f < 256) ? f : i; |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
2984 spelltab.st_upper[i] = (u < 256) ? u : i; |
| 324 | 2985 } |
| 2986 } | |
| 307 | 2987 else |
| 2988 { | |
| 324 | 2989 /* Rough guess: use locale-dependent library functions. */ |
| 307 | 2990 for (i = 128; i < 256; ++i) |
| 2991 { | |
| 2992 if (MB_ISUPPER(i)) | |
| 2993 { | |
| 324 | 2994 spelltab.st_isw[i] = TRUE; |
| 307 | 2995 spelltab.st_isu[i] = TRUE; |
| 2996 spelltab.st_fold[i] = MB_TOLOWER(i); | |
| 2997 } | |
| 324 | 2998 else if (MB_ISLOWER(i)) |
| 2999 { | |
| 3000 spelltab.st_isw[i] = TRUE; | |
| 3001 spelltab.st_upper[i] = MB_TOUPPER(i); | |
| 3002 } | |
| 307 | 3003 } |
| 3004 } | |
| 3005 } | |
| 3006 | |
| 3007 | |
| 3008 /* | |
| 358 | 3009 * Return TRUE if "p" points to a word character. |
| 366 | 3010 * As a special case we see "midword" characters as word character when it is |
| 358 | 3011 * followed by a word character. This finds they'there but not 'they there'. |
| 366 | 3012 * Thus this only works properly when past the first character of the word. |
| 358 | 3013 */ |
| 3014 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3015 spell_iswordp( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3016 char_u *p, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3017 win_T *wp) /* buffer used */ |
| 358 | 3018 { |
| 3019 char_u *s; | |
| 366 | 3020 int l; |
| 3021 int c; | |
| 3022 | |
| 3023 if (has_mbyte) | |
| 3024 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
3025 l = MB_PTR2LEN(p); |
| 358 | 3026 s = p; |
| 366 | 3027 if (l == 1) |
| 3028 { | |
| 3029 /* be quick for ASCII */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3030 if (wp->w_s->b_spell_ismw[*p]) |
| 366 | 3031 s = p + 1; /* skip a mid-word character */ |
| 3032 } | |
| 3033 else | |
| 3034 { | |
| 3035 c = mb_ptr2char(p); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3036 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
|
3037 : (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
|
3038 && vim_strchr(wp->w_s->b_spell_ismw_mb, c) != NULL)) |
| 366 | 3039 s = p + l; |
| 3040 } | |
| 3041 | |
| 390 | 3042 c = mb_ptr2char(s); |
| 3043 if (c > 255) | |
| 5477 | 3044 return spell_mb_isword_class(mb_get_class(s), wp); |
| 390 | 3045 return spelltab.st_isw[c]; |
| 366 | 3046 } |
| 3047 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3048 return spelltab.st_isw[wp->w_s->b_spell_ismw[*p] ? p[1] : p[0]]; |
| 376 | 3049 } |
| 3050 | |
| 3051 /* | |
| 3052 * Return TRUE if "p" points to a word character. | |
| 3053 * Unlike spell_iswordp() this doesn't check for "midword" characters. | |
| 3054 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3055 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3056 spell_iswordp_nmw(char_u *p, win_T *wp) |
| 376 | 3057 { |
| 390 | 3058 int c; |
| 3059 | |
| 3060 if (has_mbyte) | |
| 3061 { | |
| 3062 c = mb_ptr2char(p); | |
| 3063 if (c > 255) | |
| 5477 | 3064 return spell_mb_isword_class(mb_get_class(p), wp); |
| 390 | 3065 return spelltab.st_isw[c]; |
| 3066 } | |
| 376 | 3067 return spelltab.st_isw[*p]; |
| 358 | 3068 } |
| 3069 | |
| 372 | 3070 /* |
| 1580 | 3071 * Return TRUE if word class indicates a word character. |
| 3072 * Only for characters above 255. | |
| 3073 * Unicode subscript and superscript are not considered word characters. | |
| 5477 | 3074 * See also dbcs_class() and utf_class() in mbyte.c. |
| 3075 */ | |
| 3076 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3077 spell_mb_isword_class(int cl, win_T *wp) |
| 5477 | 3078 { |
| 3079 if (wp->w_s->b_cjk) | |
| 3080 /* East Asian characters are not considered word characters. */ | |
| 3081 return cl == 2 || cl == 0x2800; | |
|
17434
26e8d42987ca
patch 8.1.1715: emoji characters are seen as word characters for spelling
Bram Moolenaar <Bram@vim.org>
parents:
16825
diff
changeset
|
3082 return cl >= 2 && cl != 0x2070 && cl != 0x2080 && cl != 3; |
| 1580 | 3083 } |
| 3084 | |
| 3085 /* | |
| 372 | 3086 * Return TRUE if "p" points to a word character. |
| 3087 * Wide version of spell_iswordp(). | |
| 3088 */ | |
| 3089 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3090 spell_iswordp_w(int *p, win_T *wp) |
| 372 | 3091 { |
| 3092 int *s; | |
| 3093 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3094 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
|
3095 : (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
|
3096 && vim_strchr(wp->w_s->b_spell_ismw_mb, *p) != NULL)) |
| 372 | 3097 s = p + 1; |
| 3098 else | |
| 3099 s = p; | |
| 3100 | |
| 390 | 3101 if (*s > 255) |
| 372 | 3102 { |
| 3103 if (enc_utf8) | |
| 5477 | 3104 return spell_mb_isword_class(utf_class(*s), wp); |
| 372 | 3105 if (enc_dbcs) |
| 5477 | 3106 return spell_mb_isword_class( |
| 3107 dbcs_class((unsigned)*s >> 8, *s & 0xff), wp); | |
| 372 | 3108 return 0; |
| 3109 } | |
| 3110 return spelltab.st_isw[*s]; | |
| 3111 } | |
| 3112 | |
| 358 | 3113 /* |
| 324 | 3114 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated. |
| 3115 * Uses the character definitions from the .spl file. | |
| 307 | 3116 * When using a multi-byte 'encoding' the length may change! |
| 3117 * Returns FAIL when something wrong. | |
| 3118 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3119 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3120 spell_casefold( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3121 char_u *str, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3122 int len, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3123 char_u *buf, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3124 int buflen) |
| 307 | 3125 { |
| 3126 int i; | |
| 3127 | |
| 3128 if (len >= buflen) | |
| 3129 { | |
| 3130 buf[0] = NUL; | |
| 3131 return FAIL; /* result will not fit */ | |
| 3132 } | |
| 3133 | |
| 3134 if (has_mbyte) | |
| 3135 { | |
| 324 | 3136 int outi = 0; |
| 3137 char_u *p; | |
| 307 | 3138 int c; |
| 3139 | |
| 3140 /* Fold one character at a time. */ | |
| 324 | 3141 for (p = str; p < str + len; ) |
| 307 | 3142 { |
| 3143 if (outi + MB_MAXBYTES > buflen) | |
| 3144 { | |
| 3145 buf[outi] = NUL; | |
| 3146 return FAIL; | |
| 3147 } | |
| 474 | 3148 c = mb_cptr2char_adv(&p); |
| 324 | 3149 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi); |
| 307 | 3150 } |
| 3151 buf[outi] = NUL; | |
| 3152 } | |
| 3153 else | |
| 3154 { | |
| 3155 /* Be quick for non-multibyte encodings. */ | |
| 3156 for (i = 0; i < len; ++i) | |
| 324 | 3157 buf[i] = spelltab.st_fold[str[i]]; |
| 307 | 3158 buf[i] = NUL; |
| 3159 } | |
| 3160 | |
| 3161 return OK; | |
| 3162 } | |
| 3163 | |
| 625 | 3164 /* values for sps_flags */ |
| 372 | 3165 #define SPS_BEST 1 |
| 3166 #define SPS_FAST 2 | |
| 3167 #define SPS_DOUBLE 4 | |
| 3168 | |
| 625 | 3169 static int sps_flags = SPS_BEST; /* flags from 'spellsuggest' */ |
| 3170 static int sps_limit = 9999; /* max nr of suggestions given */ | |
| 372 | 3171 |
| 3172 /* | |
| 3173 * Check the 'spellsuggest' option. Return FAIL if it's wrong. | |
| 483 | 3174 * Sets "sps_flags" and "sps_limit". |
| 372 | 3175 */ |
| 3176 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3177 spell_check_sps(void) |
| 372 | 3178 { |
| 3179 char_u *p; | |
| 483 | 3180 char_u *s; |
| 372 | 3181 char_u buf[MAXPATHL]; |
| 3182 int f; | |
| 3183 | |
| 3184 sps_flags = 0; | |
| 483 | 3185 sps_limit = 9999; |
| 372 | 3186 |
| 3187 for (p = p_sps; *p != NUL; ) | |
| 3188 { | |
| 3189 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 3190 | |
| 3191 f = 0; | |
| 483 | 3192 if (VIM_ISDIGIT(*buf)) |
| 3193 { | |
| 3194 s = buf; | |
| 3195 sps_limit = getdigits(&s); | |
| 3196 if (*s != NUL && !VIM_ISDIGIT(*s)) | |
| 3197 f = -1; | |
| 3198 } | |
| 3199 else if (STRCMP(buf, "best") == 0) | |
| 372 | 3200 f = SPS_BEST; |
| 3201 else if (STRCMP(buf, "fast") == 0) | |
| 3202 f = SPS_FAST; | |
| 3203 else if (STRCMP(buf, "double") == 0) | |
| 3204 f = SPS_DOUBLE; | |
| 3205 else if (STRNCMP(buf, "expr:", 5) != 0 | |
| 3206 && STRNCMP(buf, "file:", 5) != 0) | |
| 3207 f = -1; | |
| 3208 | |
| 3209 if (f == -1 || (sps_flags != 0 && f != 0)) | |
| 3210 { | |
| 3211 sps_flags = SPS_BEST; | |
| 483 | 3212 sps_limit = 9999; |
| 372 | 3213 return FAIL; |
| 3214 } | |
| 3215 if (f != 0) | |
| 3216 sps_flags = f; | |
| 3217 } | |
| 3218 | |
| 3219 if (sps_flags == 0) | |
| 3220 sps_flags = SPS_BEST; | |
| 3221 | |
| 3222 return OK; | |
| 3223 } | |
| 3224 | |
| 323 | 3225 /* |
| 5374 | 3226 * "z=": Find badly spelled word under or after the cursor. |
| 323 | 3227 * Give suggestions for the properly spelled word. |
| 638 | 3228 * In Visual mode use the highlighted word as the bad word. |
| 485 | 3229 * When "count" is non-zero use that suggestion. |
| 323 | 3230 */ |
| 3231 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3232 spell_suggest(int count) |
| 323 | 3233 { |
| 3234 char_u *line; | |
| 3235 pos_T prev_cursor = curwin->w_cursor; | |
| 3236 char_u wcopy[MAXWLEN + 2]; | |
| 3237 char_u *p; | |
| 3238 int i; | |
| 3239 int c; | |
| 3240 suginfo_T sug; | |
| 3241 suggest_T *stp; | |
| 372 | 3242 int mouse_used; |
| 388 | 3243 int need_cap; |
| 483 | 3244 int limit; |
| 485 | 3245 int selected = count; |
| 638 | 3246 int badlen = 0; |
| 1938 | 3247 int msg_scroll_save = msg_scroll; |
| 638 | 3248 |
| 3249 if (no_spell_checking(curwin)) | |
| 3250 return; | |
| 3251 | |
| 3252 if (VIsual_active) | |
| 3253 { | |
| 3254 /* Use the Visually selected text as the bad word. But reject | |
| 3255 * a multi-line selection. */ | |
| 3256 if (curwin->w_cursor.lnum != VIsual.lnum) | |
| 3257 { | |
| 6949 | 3258 vim_beep(BO_SPELL); |
| 638 | 3259 return; |
| 3260 } | |
| 3261 badlen = (int)curwin->w_cursor.col - (int)VIsual.col; | |
| 3262 if (badlen < 0) | |
| 3263 badlen = -badlen; | |
| 3264 else | |
| 3265 curwin->w_cursor.col = VIsual.col; | |
| 3266 ++badlen; | |
| 3267 end_visual_mode(); | |
| 3268 } | |
| 5735 | 3269 /* Find the start of the badly spelled word. */ |
| 3270 else if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0 | |
| 346 | 3271 || curwin->w_cursor.col > prev_cursor.col) |
| 3272 { | |
| 3273 /* No bad word or it starts after the cursor: use the word under the | |
| 3274 * cursor. */ | |
| 3275 curwin->w_cursor = prev_cursor; | |
| 3276 line = ml_get_curline(); | |
| 3277 p = line + curwin->w_cursor.col; | |
| 3278 /* Backup to before start of word. */ | |
| 5477 | 3279 while (p > line && spell_iswordp_nmw(p, curwin)) |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
3280 MB_PTR_BACK(line, p); |
| 346 | 3281 /* Forward to start of word. */ |
| 5477 | 3282 while (*p != NUL && !spell_iswordp_nmw(p, curwin)) |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
3283 MB_PTR_ADV(p); |
| 346 | 3284 |
| 5477 | 3285 if (!spell_iswordp_nmw(p, curwin)) /* No word found. */ |
| 346 | 3286 { |
| 3287 beep_flush(); | |
| 3288 return; | |
| 3289 } | |
| 835 | 3290 curwin->w_cursor.col = (colnr_T)(p - line); |
| 323 | 3291 } |
| 3292 | |
| 344 | 3293 /* Get the word and its length. */ |
| 3294 | |
| 388 | 3295 /* Figure out if the word should be capitalised. */ |
| 475 | 3296 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col); |
| 3297 | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3298 /* 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
|
3299 line = vim_strsave(ml_get_curline()); |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3300 if (line == NULL) |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3301 goto skip; |
| 388 | 3302 |
| 483 | 3303 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in |
| 3304 * 'spellsuggest', whatever is smaller. */ | |
| 3305 if (sps_limit > (int)Rows - 2) | |
| 3306 limit = (int)Rows - 2; | |
| 3307 else | |
| 3308 limit = sps_limit; | |
| 638 | 3309 spell_find_suggest(line + curwin->w_cursor.col, badlen, &sug, limit, |
| 625 | 3310 TRUE, need_cap, TRUE); |
| 323 | 3311 |
| 3312 if (sug.su_ga.ga_len == 0) | |
|
15543
dd725a8ab112
patch 8.1.0779: argument for message functions is inconsistent
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
3313 msg(_("Sorry, no suggestions")); |
| 485 | 3314 else if (count > 0) |
| 3315 { | |
| 3316 if (count > sug.su_ga.ga_len) | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
3317 smsg(_("Sorry, only %ld suggestions"), |
| 485 | 3318 (long)sug.su_ga.ga_len); |
| 3319 } | |
| 323 | 3320 else |
| 3321 { | |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
3322 VIM_CLEAR(repl_from); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
3323 VIM_CLEAR(repl_to); |
| 372 | 3324 |
| 474 | 3325 #ifdef FEAT_RIGHTLEFT |
| 3326 /* When 'rightleft' is set the list is drawn right-left. */ | |
| 3327 cmdmsg_rl = curwin->w_p_rl; | |
| 3328 if (cmdmsg_rl) | |
| 3329 msg_col = Columns - 1; | |
| 3330 #endif | |
| 3331 | |
| 323 | 3332 /* List the suggestions. */ |
| 3333 msg_start(); | |
| 914 | 3334 msg_row = Rows - 1; /* for when 'cmdheight' > 1 */ |
| 372 | 3335 lines_left = Rows; /* avoid more prompt */ |
| 323 | 3336 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"), |
| 3337 sug.su_badlen, sug.su_badptr); | |
| 474 | 3338 #ifdef FEAT_RIGHTLEFT |
| 3339 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0) | |
| 3340 { | |
| 3341 /* And now the rabbit from the high hat: Avoid showing the | |
| 3342 * untranslated message rightleft. */ | |
| 3343 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC", | |
| 3344 sug.su_badlen, sug.su_badptr); | |
| 3345 } | |
| 3346 #endif | |
|
15543
dd725a8ab112
patch 8.1.0779: argument for message functions is inconsistent
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
3347 msg_puts((char *)IObuff); |
| 323 | 3348 msg_clr_eos(); |
| 3349 msg_putchar('\n'); | |
| 346 | 3350 |
| 323 | 3351 msg_scroll = TRUE; |
| 3352 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 3353 { | |
| 344 | 3354 stp = &SUG(sug.su_ga, i); |
| 323 | 3355 |
| 3356 /* The suggested word may replace only part of the bad word, add | |
| 3357 * the not replaced part. */ | |
| 2768 | 3358 vim_strncpy(wcopy, stp->st_word, MAXWLEN); |
| 323 | 3359 if (sug.su_badlen > stp->st_orglen) |
| 625 | 3360 vim_strncpy(wcopy + stp->st_wordlen, |
| 323 | 3361 sug.su_badptr + stp->st_orglen, |
| 3362 sug.su_badlen - stp->st_orglen); | |
| 474 | 3363 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1); |
| 3364 #ifdef FEAT_RIGHTLEFT | |
| 3365 if (cmdmsg_rl) | |
| 3366 rl_mirror(IObuff); | |
| 3367 #endif | |
|
15543
dd725a8ab112
patch 8.1.0779: argument for message functions is inconsistent
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
3368 msg_puts((char *)IObuff); |
| 474 | 3369 |
| 3370 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy); | |
|
15543
dd725a8ab112
patch 8.1.0779: argument for message functions is inconsistent
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
3371 msg_puts((char *)IObuff); |
| 346 | 3372 |
| 3373 /* The word may replace more than "su_badlen". */ | |
| 3374 if (sug.su_badlen < stp->st_orglen) | |
| 3375 { | |
| 3376 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""), | |
| 3377 stp->st_orglen, sug.su_badptr); | |
|
15543
dd725a8ab112
patch 8.1.0779: argument for message functions is inconsistent
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
3378 msg_puts((char *)IObuff); |
| 346 | 3379 } |
| 3380 | |
| 324 | 3381 if (p_verbose > 0) |
| 344 | 3382 { |
| 346 | 3383 /* Add the score. */ |
| 351 | 3384 if (sps_flags & (SPS_DOUBLE | SPS_BEST)) |
| 474 | 3385 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)", |
| 344 | 3386 stp->st_salscore ? "s " : "", |
| 3387 stp->st_score, stp->st_altscore); | |
| 3388 else | |
| 474 | 3389 vim_snprintf((char *)IObuff, IOSIZE, " (%d)", |
| 346 | 3390 stp->st_score); |
| 474 | 3391 #ifdef FEAT_RIGHTLEFT |
| 3392 if (cmdmsg_rl) | |
| 3393 /* Mirror the numbers, but keep the leading space. */ | |
| 3394 rl_mirror(IObuff + 1); | |
| 3395 #endif | |
| 346 | 3396 msg_advance(30); |
|
15543
dd725a8ab112
patch 8.1.0779: argument for message functions is inconsistent
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
3397 msg_puts((char *)IObuff); |
| 344 | 3398 } |
| 323 | 3399 msg_putchar('\n'); |
| 3400 } | |
| 3401 | |
| 474 | 3402 #ifdef FEAT_RIGHTLEFT |
| 3403 cmdmsg_rl = FALSE; | |
| 3404 msg_col = 0; | |
| 3405 #endif | |
| 323 | 3406 /* Ask for choice. */ |
| 485 | 3407 selected = prompt_for_number(&mouse_used); |
| 372 | 3408 if (mouse_used) |
| 485 | 3409 selected -= lines_left; |
| 1938 | 3410 lines_left = Rows; /* avoid more prompt */ |
| 3411 /* don't delay for 'smd' in normal_cmd() */ | |
| 3412 msg_scroll = msg_scroll_save; | |
| 485 | 3413 } |
| 3414 | |
| 3415 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK) | |
| 3416 { | |
| 3417 /* Save the from and to text for :spellrepall. */ | |
| 3418 stp = &SUG(sug.su_ga, selected - 1); | |
| 548 | 3419 if (sug.su_badlen > stp->st_orglen) |
| 3420 { | |
| 3421 /* Replacing less than "su_badlen", append the remainder to | |
| 3422 * repl_to. */ | |
| 3423 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen); | |
| 3424 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word, | |
| 3425 sug.su_badlen - stp->st_orglen, | |
| 3426 sug.su_badptr + stp->st_orglen); | |
| 3427 repl_to = vim_strsave(IObuff); | |
| 3428 } | |
| 3429 else | |
| 3430 { | |
| 3431 /* Replacing su_badlen or more, use the whole word. */ | |
| 3432 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen); | |
| 3433 repl_to = vim_strsave(stp->st_word); | |
| 3434 } | |
| 485 | 3435 |
| 3436 /* Replace the word. */ | |
|
16764
ef00b6bc186b
patch 8.1.1384: using "int" for alloc() often results in compiler warnings
Bram Moolenaar <Bram@vim.org>
parents:
16738
diff
changeset
|
3437 p = alloc(STRLEN(line) - stp->st_orglen + stp->st_wordlen + 1); |
| 485 | 3438 if (p != NULL) |
| 3439 { | |
| 835 | 3440 c = (int)(sug.su_badptr - line); |
| 485 | 3441 mch_memmove(p, line, c); |
| 3442 STRCPY(p + c, stp->st_word); | |
| 3443 STRCAT(p, sug.su_badptr + stp->st_orglen); | |
| 3444 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 3445 curwin->w_cursor.col = c; | |
| 3446 | |
| 3447 /* For redo we use a change-word command. */ | |
| 3448 ResetRedobuff(); | |
| 3449 AppendToRedobuff((char_u *)"ciw"); | |
| 620 | 3450 AppendToRedobuffLit(p + c, |
| 625 | 3451 stp->st_wordlen + sug.su_badlen - stp->st_orglen); |
| 485 | 3452 AppendCharToRedobuff(ESC); |
| 819 | 3453 |
| 3454 /* After this "p" may be invalid. */ | |
| 3455 changed_bytes(curwin->w_cursor.lnum, c); | |
| 485 | 3456 } |
| 3457 } | |
| 3458 else | |
| 3459 curwin->w_cursor = prev_cursor; | |
| 323 | 3460 |
| 344 | 3461 spell_find_cleanup(&sug); |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3462 skip: |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3463 vim_free(line); |
| 344 | 3464 } |
| 3465 | |
| 3466 /* | |
| 475 | 3467 * Check if the word at line "lnum" column "col" is required to start with a |
| 3468 * capital. This uses 'spellcapcheck' of the current buffer. | |
| 3469 */ | |
| 3470 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3471 check_need_cap(linenr_T lnum, colnr_T col) |
| 475 | 3472 { |
| 3473 int need_cap = FALSE; | |
| 3474 char_u *line; | |
| 3475 char_u *line_copy = NULL; | |
| 3476 char_u *p; | |
| 3477 colnr_T endcol; | |
| 3478 regmatch_T regmatch; | |
| 3479 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3480 if (curwin->w_s->b_cap_prog == NULL) |
| 475 | 3481 return FALSE; |
| 3482 | |
| 3483 line = ml_get_curline(); | |
| 3484 endcol = 0; | |
|
12323
4dba3e4f3b01
patch 8.0.1041: bogus characters when indenting during visual-block append
Christian Brabandt <cb@256bit.org>
parents:
11521
diff
changeset
|
3485 if (getwhitecols(line) >= (int)col) |
| 475 | 3486 { |
| 3487 /* At start of line, check if previous line is empty or sentence | |
| 3488 * ends there. */ | |
| 3489 if (lnum == 1) | |
| 3490 need_cap = TRUE; | |
| 3491 else | |
| 3492 { | |
| 3493 line = ml_get(lnum - 1); | |
| 3494 if (*skipwhite(line) == NUL) | |
| 3495 need_cap = TRUE; | |
| 3496 else | |
| 3497 { | |
| 3498 /* Append a space in place of the line break. */ | |
| 3499 line_copy = concat_str(line, (char_u *)" "); | |
| 3500 line = line_copy; | |
| 835 | 3501 endcol = (colnr_T)STRLEN(line); |
| 475 | 3502 } |
| 3503 } | |
| 3504 } | |
| 3505 else | |
| 3506 endcol = col; | |
| 3507 | |
| 3508 if (endcol > 0) | |
| 3509 { | |
| 3510 /* 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
|
3511 regmatch.regprog = curwin->w_s->b_cap_prog; |
| 475 | 3512 regmatch.rm_ic = FALSE; |
| 3513 p = line + endcol; | |
| 3514 for (;;) | |
| 3515 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
3516 MB_PTR_BACK(line, p); |
| 5477 | 3517 if (p == line || spell_iswordp_nmw(p, curwin)) |
| 475 | 3518 break; |
| 3519 if (vim_regexec(®match, p, 0) | |
| 3520 && regmatch.endp[0] == line + endcol) | |
| 3521 { | |
| 3522 need_cap = TRUE; | |
| 3523 break; | |
| 3524 } | |
| 3525 } | |
| 6375 | 3526 curwin->w_s->b_cap_prog = regmatch.regprog; |
| 475 | 3527 } |
| 3528 | |
| 3529 vim_free(line_copy); | |
| 3530 | |
| 3531 return need_cap; | |
| 3532 } | |
| 3533 | |
| 3534 | |
| 3535 /* | |
| 372 | 3536 * ":spellrepall" |
| 3537 */ | |
| 3538 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3539 ex_spellrepall(exarg_T *eap UNUSED) |
| 372 | 3540 { |
| 3541 pos_T pos = curwin->w_cursor; | |
| 3542 char_u *frompat; | |
| 3543 int addlen; | |
| 3544 char_u *line; | |
| 3545 char_u *p; | |
| 3546 int save_ws = p_ws; | |
| 483 | 3547 linenr_T prev_lnum = 0; |
| 372 | 3548 |
| 3549 if (repl_from == NULL || repl_to == NULL) | |
| 3550 { | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
3551 emsg(_("E752: No previous spell replacement")); |
| 372 | 3552 return; |
| 3553 } | |
| 835 | 3554 addlen = (int)(STRLEN(repl_to) - STRLEN(repl_from)); |
| 3555 | |
|
16764
ef00b6bc186b
patch 8.1.1384: using "int" for alloc() often results in compiler warnings
Bram Moolenaar <Bram@vim.org>
parents:
16738
diff
changeset
|
3556 frompat = alloc(STRLEN(repl_from) + 7); |
| 372 | 3557 if (frompat == NULL) |
| 3558 return; | |
| 3559 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from); | |
| 3560 p_ws = FALSE; | |
| 3561 | |
| 483 | 3562 sub_nsubs = 0; |
| 3563 sub_nlines = 0; | |
| 372 | 3564 curwin->w_cursor.lnum = 0; |
| 3565 while (!got_int) | |
| 3566 { | |
|
11521
578df034735d
patch 8.0.0643: when a pattern search is slow Vim becomes unusable
Christian Brabandt <cb@256bit.org>
parents:
11394
diff
changeset
|
3567 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP, NULL, NULL) == 0 |
| 372 | 3568 || u_save_cursor() == FAIL) |
| 3569 break; | |
| 3570 | |
| 3571 /* Only replace when the right word isn't there yet. This happens | |
| 3572 * when changing "etc" to "etc.". */ | |
| 3573 line = ml_get_curline(); | |
| 3574 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col, | |
| 3575 repl_to, STRLEN(repl_to)) != 0) | |
| 3576 { | |
|
16764
ef00b6bc186b
patch 8.1.1384: using "int" for alloc() often results in compiler warnings
Bram Moolenaar <Bram@vim.org>
parents:
16738
diff
changeset
|
3577 p = alloc(STRLEN(line) + addlen + 1); |
| 372 | 3578 if (p == NULL) |
| 3579 break; | |
| 3580 mch_memmove(p, line, curwin->w_cursor.col); | |
| 3581 STRCPY(p + curwin->w_cursor.col, repl_to); | |
| 3582 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from)); | |
| 3583 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 3584 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col); | |
| 483 | 3585 |
| 3586 if (curwin->w_cursor.lnum != prev_lnum) | |
| 3587 { | |
| 3588 ++sub_nlines; | |
| 3589 prev_lnum = curwin->w_cursor.lnum; | |
| 3590 } | |
| 3591 ++sub_nsubs; | |
| 372 | 3592 } |
| 835 | 3593 curwin->w_cursor.col += (colnr_T)STRLEN(repl_to); |
| 372 | 3594 } |
| 3595 | |
| 3596 p_ws = save_ws; | |
| 3597 curwin->w_cursor = pos; | |
| 3598 vim_free(frompat); | |
| 3599 | |
| 483 | 3600 if (sub_nsubs == 0) |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
3601 semsg(_("E753: Not found: %s"), repl_from); |
| 483 | 3602 else |
| 3603 do_sub_msg(FALSE); | |
| 372 | 3604 } |
| 3605 | |
| 3606 /* | |
| 344 | 3607 * Find spell suggestions for "word". Return them in the growarray "*gap" as |
| 3608 * a list of allocated strings. | |
| 3609 */ | |
| 3610 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3611 spell_suggest_list( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3612 garray_T *gap, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3613 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3614 int maxcount, /* maximum nr of suggestions */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3615 int need_cap, /* 'spellcapcheck' matched */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3616 int interactive) |
| 344 | 3617 { |
| 3618 suginfo_T sug; | |
| 3619 int i; | |
| 3620 suggest_T *stp; | |
| 3621 char_u *wcopy; | |
| 3622 | |
| 638 | 3623 spell_find_suggest(word, 0, &sug, maxcount, FALSE, need_cap, interactive); |
| 344 | 3624 |
| 3625 /* Make room in "gap". */ | |
| 3626 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1); | |
| 626 | 3627 if (ga_grow(gap, sug.su_ga.ga_len) == OK) |
| 3628 { | |
| 3629 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 3630 { | |
| 3631 stp = &SUG(sug.su_ga, i); | |
| 3632 | |
| 3633 /* The suggested word may replace only part of "word", add the not | |
| 3634 * replaced part. */ | |
| 3635 wcopy = alloc(stp->st_wordlen | |
| 835 | 3636 + (unsigned)STRLEN(sug.su_badptr + stp->st_orglen) + 1); |
| 626 | 3637 if (wcopy == NULL) |
| 3638 break; | |
| 3639 STRCPY(wcopy, stp->st_word); | |
| 3640 STRCPY(wcopy + stp->st_wordlen, sug.su_badptr + stp->st_orglen); | |
| 3641 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy; | |
| 3642 } | |
| 344 | 3643 } |
| 3644 | |
| 3645 spell_find_cleanup(&sug); | |
| 3646 } | |
| 3647 | |
| 3648 /* | |
| 3649 * Find spell suggestions for the word at the start of "badptr". | |
| 3650 * Return the suggestions in "su->su_ga". | |
| 3651 * The maximum number of suggestions is "maxcount". | |
| 3652 * Note: does use info for the current window. | |
| 3653 * This is based on the mechanisms of Aspell, but completely reimplemented. | |
| 3654 */ | |
| 3655 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3656 spell_find_suggest( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3657 char_u *badptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3658 int badlen, /* length of bad word or 0 if unknown */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3659 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3660 int maxcount, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3661 int banbadword, /* don't include badword in suggestions */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3662 int need_cap, /* word should start with capital */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3663 int interactive) |
| 344 | 3664 { |
| 534 | 3665 hlf_T attr = HLF_COUNT; |
| 372 | 3666 char_u buf[MAXPATHL]; |
| 3667 char_u *p; | |
| 3668 int do_combine = FALSE; | |
| 3669 char_u *sps_copy; | |
| 3670 #ifdef FEAT_EVAL | |
| 3671 static int expr_busy = FALSE; | |
| 3672 #endif | |
| 385 | 3673 int c; |
| 503 | 3674 int i; |
| 3675 langp_T *lp; | |
| 344 | 3676 |
| 3677 /* | |
| 3678 * Set the info in "*su". | |
| 3679 */ | |
| 3680 vim_memset(su, 0, sizeof(suginfo_T)); | |
| 3681 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10); | |
| 3682 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10); | |
| 355 | 3683 if (*badptr == NUL) |
| 3684 return; | |
| 344 | 3685 hash_init(&su->su_banned); |
| 3686 | |
| 3687 su->su_badptr = badptr; | |
| 638 | 3688 if (badlen != 0) |
| 3689 su->su_badlen = badlen; | |
| 3690 else | |
| 3691 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL, FALSE); | |
| 344 | 3692 su->su_maxcount = maxcount; |
| 372 | 3693 su->su_maxscore = SCORE_MAXINIT; |
| 344 | 3694 |
| 3695 if (su->su_badlen >= MAXWLEN) | |
| 3696 su->su_badlen = MAXWLEN - 1; /* just in case */ | |
| 3697 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen); | |
| 3698 (void)spell_casefold(su->su_badptr, su->su_badlen, | |
| 3699 su->su_fbadword, MAXWLEN); | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
3700 /* TODO: make this work if the case-folded text is longer than the original |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
3701 * text. Currently an illegal byte causes wrong pointer computations. */ |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
3702 su->su_fbadword[su->su_badlen] = NUL; |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
3703 |
| 346 | 3704 /* get caps flags for bad word */ |
| 474 | 3705 su->su_badflags = badword_captype(su->su_badptr, |
| 3706 su->su_badptr + su->su_badlen); | |
| 388 | 3707 if (need_cap) |
| 3708 su->su_badflags |= WF_ONECAP; | |
| 344 | 3709 |
| 503 | 3710 /* Find the default language for sound folding. We simply use the first |
| 3711 * one in 'spelllang' that supports sound folding. That's good for when | |
| 3712 * using multiple files for one language, it's not that bad when mixing | |
| 3713 * languages (e.g., "pl,en"). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3714 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
|
3715 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3716 lp = LANGP_ENTRY(curbuf->b_s.b_langp, i); |
| 503 | 3717 if (lp->lp_sallang != NULL) |
| 3718 { | |
| 3719 su->su_sallang = lp->lp_sallang; | |
| 3720 break; | |
| 3721 } | |
| 3722 } | |
| 3723 | |
| 534 | 3724 /* Soundfold the bad word with the default sound folding, so that we don't |
| 3725 * have to do this many times. */ | |
| 3726 if (su->su_sallang != NULL) | |
| 3727 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE, | |
| 3728 su->su_sal_badword); | |
| 3729 | |
| 385 | 3730 /* If the word is not capitalised and spell_check() doesn't consider the |
| 3731 * word to be bad then it might need to be capitalised. Add a suggestion | |
| 3732 * for that. */ | |
| 455 | 3733 c = PTR2CHAR(su->su_badptr); |
| 534 | 3734 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT) |
| 385 | 3735 { |
| 3736 make_case_word(su->su_badword, buf, WF_ONECAP); | |
| 3737 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE, | |
| 625 | 3738 0, TRUE, su->su_sallang, FALSE); |
| 385 | 3739 } |
| 3740 | |
| 344 | 3741 /* Ban the bad word itself. It may appear in another region. */ |
| 358 | 3742 if (banbadword) |
| 3743 add_banned(su, su->su_badword); | |
| 344 | 3744 |
| 372 | 3745 /* Make a copy of 'spellsuggest', because the expression may change it. */ |
| 3746 sps_copy = vim_strsave(p_sps); | |
| 3747 if (sps_copy == NULL) | |
| 3748 return; | |
| 3749 | |
| 3750 /* Loop over the items in 'spellsuggest'. */ | |
| 3751 for (p = sps_copy; *p != NUL; ) | |
| 3752 { | |
| 3753 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 3754 | |
| 3755 if (STRNCMP(buf, "expr:", 5) == 0) | |
| 3756 { | |
| 3757 #ifdef FEAT_EVAL | |
| 375 | 3758 /* Evaluate an expression. Skip this when called recursively, |
| 3759 * when using spellsuggest() in the expression. */ | |
| 372 | 3760 if (!expr_busy) |
| 3761 { | |
| 3762 expr_busy = TRUE; | |
| 3763 spell_suggest_expr(su, buf + 5); | |
| 3764 expr_busy = FALSE; | |
| 3765 } | |
| 3766 #endif | |
| 3767 } | |
| 3768 else if (STRNCMP(buf, "file:", 5) == 0) | |
| 3769 /* Use list of suggestions in a file. */ | |
| 3770 spell_suggest_file(su, buf + 5); | |
| 3771 else | |
| 3772 { | |
| 3773 /* Use internal method. */ | |
| 625 | 3774 spell_suggest_intern(su, interactive); |
| 372 | 3775 if (sps_flags & SPS_DOUBLE) |
| 3776 do_combine = TRUE; | |
| 3777 } | |
| 3778 } | |
| 3779 | |
| 3780 vim_free(sps_copy); | |
| 3781 | |
| 3782 if (do_combine) | |
| 3783 /* Combine the two list of suggestions. This must be done last, | |
| 3784 * because sorting changes the order again. */ | |
| 3785 score_combine(su); | |
| 3786 } | |
| 3787 | |
| 3788 #ifdef FEAT_EVAL | |
| 3789 /* | |
| 3790 * Find suggestions by evaluating expression "expr". | |
| 3791 */ | |
| 3792 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3793 spell_suggest_expr(suginfo_T *su, char_u *expr) |
| 372 | 3794 { |
| 3795 list_T *list; | |
| 3796 listitem_T *li; | |
| 3797 int score; | |
| 3798 char_u *p; | |
| 3799 | |
| 3800 /* The work is split up in a few parts to avoid having to export | |
| 3801 * suginfo_T. | |
| 3802 * First evaluate the expression and get the resulting list. */ | |
| 3803 list = eval_spell_expr(su->su_badword, expr); | |
| 3804 if (list != NULL) | |
| 3805 { | |
| 3806 /* Loop over the items in the list. */ | |
| 3807 for (li = list->lv_first; li != NULL; li = li->li_next) | |
| 3808 if (li->li_tv.v_type == VAR_LIST) | |
| 3809 { | |
| 3810 /* Get the word and the score from the items. */ | |
| 3811 score = get_spellword(li->li_tv.vval.v_list, &p); | |
| 625 | 3812 if (score >= 0 && score <= su->su_maxscore) |
| 3813 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 3814 score, 0, TRUE, su->su_sallang, FALSE); | |
| 372 | 3815 } |
| 3816 list_unref(list); | |
| 3817 } | |
| 3818 | |
| 625 | 3819 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3820 check_suggestions(su, &su->su_ga); | |
| 372 | 3821 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 3822 } | |
| 3823 #endif | |
| 3824 | |
| 3825 /* | |
| 474 | 3826 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'. |
| 372 | 3827 */ |
| 3828 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3829 spell_suggest_file(suginfo_T *su, char_u *fname) |
| 372 | 3830 { |
| 3831 FILE *fd; | |
| 3832 char_u line[MAXWLEN * 2]; | |
| 3833 char_u *p; | |
| 3834 int len; | |
| 3835 char_u cword[MAXWLEN]; | |
| 3836 | |
| 3837 /* Open the file. */ | |
| 3838 fd = mch_fopen((char *)fname, "r"); | |
| 3839 if (fd == NULL) | |
| 3840 { | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
3841 semsg(_(e_notopen), fname); |
| 372 | 3842 return; |
| 3843 } | |
| 3844 | |
| 3845 /* Read it line by line. */ | |
| 3846 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int) | |
| 3847 { | |
| 3848 line_breakcheck(); | |
| 3849 | |
| 3850 p = vim_strchr(line, '/'); | |
| 3851 if (p == NULL) | |
| 3852 continue; /* No Tab found, just skip the line. */ | |
| 3853 *p++ = NUL; | |
| 3854 if (STRICMP(su->su_badword, line) == 0) | |
| 3855 { | |
| 3856 /* Match! Isolate the good word, until CR or NL. */ | |
| 3857 for (len = 0; p[len] >= ' '; ++len) | |
| 3858 ; | |
| 3859 p[len] = NUL; | |
| 3860 | |
| 3861 /* If the suggestion doesn't have specific case duplicate the case | |
| 3862 * of the bad word. */ | |
| 3863 if (captype(p, NULL) == 0) | |
| 3864 { | |
| 3865 make_case_word(p, cword, su->su_badflags); | |
| 3866 p = cword; | |
| 3867 } | |
| 3868 | |
| 3869 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 625 | 3870 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE); |
| 372 | 3871 } |
| 3872 } | |
| 3873 | |
| 3874 fclose(fd); | |
| 3875 | |
| 625 | 3876 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3877 check_suggestions(su, &su->su_ga); | |
| 372 | 3878 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 3879 } | |
| 3880 | |
| 3881 /* | |
| 3882 * Find suggestions for the internal method indicated by "sps_flags". | |
| 3883 */ | |
| 3884 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3885 spell_suggest_intern(suginfo_T *su, int interactive) |
| 625 | 3886 { |
| 3887 /* | |
| 3888 * Load the .sug file(s) that are available and not done yet. | |
| 3889 */ | |
| 3890 suggest_load_files(); | |
| 3891 | |
| 344 | 3892 /* |
| 346 | 3893 * 1. Try special cases, such as repeating a word: "the the" -> "the". |
| 344 | 3894 * |
| 3895 * Set a maximum score to limit the combination of operations that is | |
| 3896 * tried. | |
| 3897 */ | |
| 346 | 3898 suggest_try_special(su); |
| 3899 | |
| 3900 /* | |
| 3901 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries | |
| 3902 * from the .aff file and inserting a space (split the word). | |
| 3903 */ | |
| 3904 suggest_try_change(su); | |
| 344 | 3905 |
| 3906 /* For the resulting top-scorers compute the sound-a-like score. */ | |
| 3907 if (sps_flags & SPS_DOUBLE) | |
| 3908 score_comp_sal(su); | |
| 3909 | |
| 3910 /* | |
| 346 | 3911 * 3. Try finding sound-a-like words. |
| 344 | 3912 */ |
| 625 | 3913 if ((sps_flags & SPS_FAST) == 0) |
| 3914 { | |
| 3915 if (sps_flags & SPS_BEST) | |
| 3916 /* Adjust the word score for the suggestions found so far for how | |
| 3917 * they sounds like. */ | |
| 3918 rescore_suggestions(su); | |
| 3919 | |
| 3920 /* | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3921 * While going through the soundfold tree "su_maxscore" is the score |
| 625 | 3922 * for the soundfold word, limits the changes that are being tried, |
| 3923 * and "su_sfmaxscore" the rescored score, which is set by | |
| 3924 * cleanup_suggestions(). | |
| 3925 * First find words with a small edit distance, because this is much | |
| 3926 * faster and often already finds the top-N suggestions. If we didn't | |
| 3927 * find many suggestions try again with a higher edit distance. | |
| 3928 * "sl_sounddone" is used to avoid doing the same word twice. | |
| 3929 */ | |
| 3930 suggest_try_soundalike_prep(); | |
| 3931 su->su_maxscore = SCORE_SFMAX1; | |
| 3932 su->su_sfmaxscore = SCORE_MAXINIT * 3; | |
| 346 | 3933 suggest_try_soundalike(su); |
| 625 | 3934 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) |
| 3935 { | |
| 3936 /* We didn't find enough matches, try again, allowing more | |
| 3937 * changes to the soundfold word. */ | |
| 3938 su->su_maxscore = SCORE_SFMAX2; | |
| 3939 suggest_try_soundalike(su); | |
| 3940 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) | |
| 3941 { | |
| 3942 /* Still didn't find enough matches, try again, allowing even | |
| 3943 * more changes to the soundfold word. */ | |
| 3944 su->su_maxscore = SCORE_SFMAX3; | |
| 3945 suggest_try_soundalike(su); | |
| 3946 } | |
| 3947 } | |
| 3948 su->su_maxscore = su->su_sfmaxscore; | |
| 3949 suggest_try_soundalike_finish(); | |
| 3950 } | |
| 3951 | |
| 3952 /* When CTRL-C was hit while searching do show the results. Only clear | |
| 3953 * got_int when using a command, not for spellsuggest(). */ | |
| 344 | 3954 ui_breakcheck(); |
| 625 | 3955 if (interactive && got_int) |
| 344 | 3956 { |
| 3957 (void)vgetc(); | |
| 3958 got_int = FALSE; | |
| 3959 } | |
| 3960 | |
| 372 | 3961 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0) |
| 344 | 3962 { |
| 3963 if (sps_flags & SPS_BEST) | |
| 3964 /* Adjust the word score for how it sounds like. */ | |
| 3965 rescore_suggestions(su); | |
| 3966 | |
| 625 | 3967 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3968 check_suggestions(su, &su->su_ga); | |
| 372 | 3969 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 344 | 3970 } |
| 3971 } | |
| 3972 | |
| 3973 /* | |
| 3974 * Free the info put in "*su" by spell_find_suggest(). | |
| 3975 */ | |
| 3976 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3977 spell_find_cleanup(suginfo_T *su) |
| 344 | 3978 { |
| 3979 int i; | |
| 3980 | |
| 3981 /* Free the suggestions. */ | |
| 3982 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 3983 vim_free(SUG(su->su_ga, i).st_word); | |
| 3984 ga_clear(&su->su_ga); | |
| 3985 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 3986 vim_free(SUG(su->su_sga, i).st_word); | |
| 3987 ga_clear(&su->su_sga); | |
| 323 | 3988 |
| 3989 /* Free the banned words. */ | |
| 625 | 3990 hash_clear_all(&su->su_banned, 0); |
| 323 | 3991 } |
| 3992 | |
| 3993 /* | |
| 324 | 3994 * Make a copy of "word", with the first letter upper or lower cased, to |
| 3995 * "wcopy[MAXWLEN]". "word" must not be empty. | |
| 3996 * The result is NUL terminated. | |
| 323 | 3997 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3998 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3999 onecap_copy( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4000 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4001 char_u *wcopy, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4002 int upper) /* TRUE: first letter made upper case */ |
| 323 | 4003 { |
| 4004 char_u *p; | |
| 4005 int c; | |
| 4006 int l; | |
| 4007 | |
| 4008 p = word; | |
| 4009 if (has_mbyte) | |
| 474 | 4010 c = mb_cptr2char_adv(&p); |
| 323 | 4011 else |
| 4012 c = *p++; | |
| 4013 if (upper) | |
| 324 | 4014 c = SPELL_TOUPPER(c); |
| 323 | 4015 else |
| 324 | 4016 c = SPELL_TOFOLD(c); |
| 323 | 4017 if (has_mbyte) |
| 4018 l = mb_char2bytes(c, wcopy); | |
| 4019 else | |
| 4020 { | |
| 4021 l = 1; | |
| 4022 wcopy[0] = c; | |
| 4023 } | |
| 376 | 4024 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1); |
| 323 | 4025 } |
| 4026 | |
| 4027 /* | |
| 324 | 4028 * Make a copy of "word" with all the letters upper cased into |
| 4029 * "wcopy[MAXWLEN]". The result is NUL terminated. | |
| 323 | 4030 */ |
| 4031 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4032 allcap_copy(char_u *word, char_u *wcopy) |
| 323 | 4033 { |
| 4034 char_u *s; | |
| 4035 char_u *d; | |
| 4036 int c; | |
| 4037 | |
| 4038 d = wcopy; | |
| 4039 for (s = word; *s != NUL; ) | |
| 4040 { | |
| 4041 if (has_mbyte) | |
| 474 | 4042 c = mb_cptr2char_adv(&s); |
| 323 | 4043 else |
| 4044 c = *s++; | |
| 492 | 4045 |
| 3042 | 4046 /* We only change 0xdf to SS when we are certain latin1 is used. It |
| 492 | 4047 * would cause weird errors in other 8-bit encodings. */ |
| 4048 if (enc_latin1like && c == 0xdf) | |
| 4049 { | |
| 4050 c = 'S'; | |
| 4051 if (d - wcopy >= MAXWLEN - 1) | |
| 4052 break; | |
| 4053 *d++ = c; | |
| 4054 } | |
| 4055 else | |
| 4056 c = SPELL_TOUPPER(c); | |
| 323 | 4057 |
| 4058 if (has_mbyte) | |
| 4059 { | |
| 4060 if (d - wcopy >= MAXWLEN - MB_MAXBYTES) | |
| 4061 break; | |
| 4062 d += mb_char2bytes(c, d); | |
| 4063 } | |
| 4064 else | |
| 4065 { | |
| 4066 if (d - wcopy >= MAXWLEN - 1) | |
| 4067 break; | |
| 4068 *d++ = c; | |
| 4069 } | |
| 4070 } | |
| 4071 *d = NUL; | |
| 4072 } | |
| 4073 | |
| 4074 /* | |
| 346 | 4075 * Try finding suggestions by recognizing specific situations. |
| 4076 */ | |
| 4077 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4078 suggest_try_special(suginfo_T *su) |
| 346 | 4079 { |
| 4080 char_u *p; | |
| 376 | 4081 size_t len; |
| 346 | 4082 int c; |
| 4083 char_u word[MAXWLEN]; | |
| 4084 | |
| 4085 /* | |
| 4086 * Recognize a word that is repeated: "the the". | |
| 4087 */ | |
| 4088 p = skiptowhite(su->su_fbadword); | |
| 4089 len = p - su->su_fbadword; | |
| 4090 p = skipwhite(p); | |
| 4091 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0) | |
| 4092 { | |
| 4093 /* Include badflags: if the badword is onecap or allcap | |
| 4094 * use that for the goodword too: "The the" -> "The". */ | |
| 4095 c = su->su_fbadword[len]; | |
| 4096 su->su_fbadword[len] = NUL; | |
| 4097 make_case_word(su->su_fbadword, word, su->su_badflags); | |
| 4098 su->su_fbadword[len] = c; | |
| 534 | 4099 |
| 4100 /* Give a soundalike score of 0, compute the score as if deleting one | |
| 4101 * character. */ | |
| 4102 add_suggestion(su, &su->su_ga, word, su->su_badlen, | |
| 625 | 4103 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE); |
| 346 | 4104 } |
| 4105 } | |
| 4106 | |
| 4107 /* | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4108 * Change the 0 to 1 to measure how much time is spent in each state. |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4109 * Output is dumped in "suggestprof". |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4110 */ |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4111 #if 0 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4112 # define SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4113 proftime_T current; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4114 proftime_T total; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4115 proftime_T times[STATE_FINAL + 1]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4116 long counts[STATE_FINAL + 1]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4117 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4118 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4119 prof_init(void) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4120 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4121 for (int i = 0; i <= STATE_FINAL; ++i) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4122 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4123 profile_zero(×[i]); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4124 counts[i] = 0; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4125 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4126 profile_start(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4127 profile_start(&total); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4128 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4129 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4130 /* call before changing state */ |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4131 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4132 prof_store(state_T state) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4133 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4134 profile_end(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4135 profile_add(×[state], ¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4136 ++counts[state]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4137 profile_start(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4138 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4139 # define PROF_STORE(state) prof_store(state); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4140 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4141 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4142 prof_report(char *name) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4143 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4144 FILE *fd = fopen("suggestprof", "a"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4145 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4146 profile_end(&total); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4147 fprintf(fd, "-----------------------\n"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4148 fprintf(fd, "%s: %s\n", name, profile_msg(&total)); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4149 for (int i = 0; i <= STATE_FINAL; ++i) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4150 fprintf(fd, "%d: %s (%ld)\n", i, profile_msg(×[i]), counts[i]); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4151 fclose(fd); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4152 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4153 #else |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4154 # define PROF_STORE(state) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4155 #endif |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4156 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4157 /* |
| 323 | 4158 * Try finding suggestions by adding/removing/swapping letters. |
| 4159 */ | |
| 4160 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4161 suggest_try_change(suginfo_T *su) |
| 323 | 4162 { |
| 4163 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */ | |
| 625 | 4164 int n; |
| 4165 char_u *p; | |
| 4166 int lpi; | |
| 323 | 4167 langp_T *lp; |
| 4168 | |
| 4169 /* We make a copy of the case-folded bad word, so that we can modify it | |
| 346 | 4170 * to find matches (esp. REP items). Append some more text, changing |
| 4171 * chars after the bad word may help. */ | |
| 323 | 4172 STRCPY(fword, su->su_fbadword); |
| 835 | 4173 n = (int)STRLEN(fword); |
| 346 | 4174 p = su->su_badptr + su->su_badlen; |
| 835 | 4175 (void)spell_casefold(p, (int)STRLEN(p), fword + n, MAXWLEN - n); |
| 323 | 4176 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4177 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
|
4178 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4179 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 482 | 4180 |
| 500 | 4181 /* If reloading a spell file fails it's still in the list but |
| 4182 * everything has been cleared. */ | |
| 625 | 4183 if (lp->lp_slang->sl_fbyts == NULL) |
| 500 | 4184 continue; |
| 4185 | |
| 625 | 4186 /* Try it for this language. Will add possible suggestions. */ |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4187 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4188 prof_init(); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4189 #endif |
| 625 | 4190 suggest_trie_walk(su, lp, fword, FALSE); |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4191 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4192 prof_report("try_change"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4193 #endif |
| 625 | 4194 } |
| 4195 } | |
| 4196 | |
| 4197 /* Check the maximum score, if we go over it we won't try this change. */ | |
| 4198 #define TRY_DEEPER(su, stack, depth, add) \ | |
| 4199 (stack[depth].ts_score + (add) < su->su_maxscore) | |
| 4200 | |
| 4201 /* | |
| 4202 * Try finding suggestions by adding/removing/swapping letters. | |
| 4203 * | |
| 4204 * This uses a state machine. At each node in the tree we try various | |
| 4205 * operations. When trying if an operation works "depth" is increased and the | |
| 4206 * stack[] is used to store info. This allows combinations, thus insert one | |
| 4207 * character, replace one and delete another. The number of changes is | |
| 4208 * limited by su->su_maxscore. | |
| 4209 * | |
| 4210 * After implementing this I noticed an article by Kemal Oflazer that | |
| 4211 * describes something similar: "Error-tolerant Finite State Recognition with | |
| 4212 * Applications to Morphological Analysis and Spelling Correction" (1996). | |
| 4213 * The implementation in the article is simplified and requires a stack of | |
| 4214 * unknown depth. The implementation here only needs a stack depth equal to | |
| 4215 * the length of the word. | |
| 4216 * | |
| 4217 * This is also used for the sound-folded word, "soundfold" is TRUE then. | |
| 4218 * The mechanism is the same, but we find a match with a sound-folded word | |
| 4219 * that comes from one or more original words. Each of these words may be | |
| 4220 * added, this is done by add_sound_suggest(). | |
| 4221 * Don't use: | |
| 4222 * the prefix tree or the keep-case tree | |
| 4223 * "su->su_badlen" | |
| 4224 * anything to do with upper and lower case | |
| 4225 * anything to do with word or non-word characters ("spell_iswordp()") | |
| 4226 * banned words | |
| 4227 * word flags (rare, region, compounding) | |
| 4228 * word splitting for now | |
| 4229 * "similar_chars()" | |
| 4230 * use "slang->sl_repsal" instead of "lp->lp_replang->sl_rep" | |
| 4231 */ | |
| 4232 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4233 suggest_trie_walk( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4234 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4235 langp_T *lp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4236 char_u *fword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4237 int soundfold) |
| 625 | 4238 { |
| 4239 char_u tword[MAXWLEN]; /* good word collected so far */ | |
| 4240 trystate_T stack[MAXWLEN]; | |
| 4241 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
|
4242 * concatenation of prefix compound |
| 625 | 4243 * words and split word. NUL terminated |
| 4244 * when going deeper but not when coming | |
| 4245 * back. */ | |
| 4246 char_u compflags[MAXWLEN]; /* compound flags, one for each word */ | |
| 4247 trystate_T *sp; | |
| 4248 int newscore; | |
| 4249 int score; | |
| 4250 char_u *byts, *fbyts, *pbyts; | |
| 4251 idx_T *idxs, *fidxs, *pidxs; | |
| 4252 int depth; | |
| 4253 int c, c2, c3; | |
| 4254 int n = 0; | |
| 4255 int flags; | |
| 4256 garray_T *gap; | |
| 4257 idx_T arridx; | |
| 4258 int len; | |
| 4259 char_u *p; | |
| 4260 fromto_T *ftp; | |
| 4261 int fl = 0, tl; | |
| 4262 int repextra = 0; /* extra bytes in fword[] from REP item */ | |
| 4263 slang_T *slang = lp->lp_slang; | |
| 4264 int fword_ends; | |
| 4265 int goodword_ends; | |
| 4266 #ifdef DEBUG_TRIEWALK | |
| 4267 /* Stores the name of the change made at each level. */ | |
| 4268 char_u changename[MAXWLEN][80]; | |
| 4269 #endif | |
| 4270 int breakcheckcount = 1000; | |
| 4271 int compound_ok; | |
| 4272 | |
| 4273 /* | |
| 4274 * Go through the whole case-fold tree, try changes at each node. | |
| 4275 * "tword[]" contains the word collected from nodes in the tree. | |
| 4276 * "fword[]" the word we are trying to match with (initially the bad | |
| 4277 * word). | |
| 4278 */ | |
| 4279 depth = 0; | |
| 4280 sp = &stack[0]; | |
| 4281 vim_memset(sp, 0, sizeof(trystate_T)); | |
| 4282 sp->ts_curi = 1; | |
| 4283 | |
| 4284 if (soundfold) | |
| 4285 { | |
| 4286 /* Going through the soundfold tree. */ | |
| 4287 byts = fbyts = slang->sl_sbyts; | |
| 4288 idxs = fidxs = slang->sl_sidxs; | |
| 4289 pbyts = NULL; | |
| 4290 pidxs = NULL; | |
| 4291 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 4292 sp->ts_state = STATE_START; | |
| 4293 } | |
| 4294 else | |
| 4295 { | |
| 330 | 4296 /* |
| 375 | 4297 * When there are postponed prefixes we need to use these first. At |
| 4298 * the end of the prefix we continue in the case-fold tree. | |
| 4299 */ | |
| 482 | 4300 fbyts = slang->sl_fbyts; |
| 4301 fidxs = slang->sl_fidxs; | |
| 4302 pbyts = slang->sl_pbyts; | |
| 4303 pidxs = slang->sl_pidxs; | |
| 375 | 4304 if (pbyts != NULL) |
| 4305 { | |
| 4306 byts = pbyts; | |
| 4307 idxs = pidxs; | |
| 482 | 4308 sp->ts_prefixdepth = PFD_PREFIXTREE; |
| 375 | 4309 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */ |
| 4310 } | |
| 4311 else | |
| 4312 { | |
| 4313 byts = fbyts; | |
| 4314 idxs = fidxs; | |
| 482 | 4315 sp->ts_prefixdepth = PFD_NOPREFIX; |
| 485 | 4316 sp->ts_state = STATE_START; |
| 375 | 4317 } |
| 625 | 4318 } |
| 4319 | |
| 4320 /* | |
| 4321 * Loop to find all suggestions. At each round we either: | |
| 4322 * - For the current state try one operation, advance "ts_curi", | |
| 4323 * increase "depth". | |
| 4324 * - When a state is done go to the next, set "ts_state". | |
| 4325 * - When all states are tried decrease "depth". | |
| 4326 */ | |
| 4327 while (depth >= 0 && !got_int) | |
| 4328 { | |
| 4329 sp = &stack[depth]; | |
| 4330 switch (sp->ts_state) | |
| 4331 { | |
| 4332 case STATE_START: | |
| 4333 case STATE_NOPREFIX: | |
| 4334 /* | |
| 4335 * Start of node: Deal with NUL bytes, which means | |
| 4336 * tword[] may end here. | |
| 4337 */ | |
| 4338 arridx = sp->ts_arridx; /* current node in the tree */ | |
| 4339 len = byts[arridx]; /* bytes in this node */ | |
| 4340 arridx += sp->ts_curi; /* index of current byte */ | |
| 4341 | |
| 4342 if (sp->ts_prefixdepth == PFD_PREFIXTREE) | |
| 4343 { | |
| 4344 /* Skip over the NUL bytes, we use them later. */ | |
| 4345 for (n = 0; n < len && byts[arridx + n] == 0; ++n) | |
| 4346 ; | |
| 4347 sp->ts_curi += n; | |
| 4348 | |
| 4349 /* Always past NUL bytes now. */ | |
| 4350 n = (int)sp->ts_state; | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4351 PROF_STORE(sp->ts_state) |
| 625 | 4352 sp->ts_state = STATE_ENDNUL; |
| 4353 sp->ts_save_badflags = su->su_badflags; | |
| 4354 | |
| 4355 /* At end of a prefix or at start of prefixtree: check for | |
| 4356 * following word. */ | |
| 4357 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX) | |
| 4358 { | |
| 4359 /* Set su->su_badflags to the caps type at this position. | |
| 4360 * Use the caps type until here for the prefix itself. */ | |
| 4361 if (has_mbyte) | |
| 4362 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); | |
| 4363 else | |
| 4364 n = sp->ts_fidx; | |
| 4365 flags = badword_captype(su->su_badptr, su->su_badptr + n); | |
| 4366 su->su_badflags = badword_captype(su->su_badptr + n, | |
| 455 | 4367 su->su_badptr + su->su_badlen); |
| 625 | 4368 #ifdef DEBUG_TRIEWALK |
| 4369 sprintf(changename[depth], "prefix"); | |
| 4370 #endif | |
| 4371 go_deeper(stack, depth, 0); | |
| 4372 ++depth; | |
| 4373 sp = &stack[depth]; | |
| 4374 sp->ts_prefixdepth = depth - 1; | |
| 4375 byts = fbyts; | |
| 4376 idxs = fidxs; | |
| 4377 sp->ts_arridx = 0; | |
| 4378 | |
| 4379 /* Move the prefix to preword[] with the right case | |
| 4380 * and make find_keepcap_word() works. */ | |
| 4381 tword[sp->ts_twordlen] = NUL; | |
| 4382 make_case_word(tword + sp->ts_splitoff, | |
| 4383 preword + sp->ts_prewordlen, flags); | |
| 835 | 4384 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 625 | 4385 sp->ts_splitoff = sp->ts_twordlen; |
| 4386 } | |
| 4387 break; | |
| 4388 } | |
| 4389 | |
| 4390 if (sp->ts_curi > len || byts[arridx] != 0) | |
| 4391 { | |
| 4392 /* Past bytes in node and/or past NUL bytes. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4393 PROF_STORE(sp->ts_state) |
| 625 | 4394 sp->ts_state = STATE_ENDNUL; |
| 4395 sp->ts_save_badflags = su->su_badflags; | |
| 4396 break; | |
| 4397 } | |
| 4398 | |
| 4399 /* | |
| 4400 * End of word in tree. | |
| 4401 */ | |
| 4402 ++sp->ts_curi; /* eat one NUL byte */ | |
| 4403 | |
| 4404 flags = (int)idxs[arridx]; | |
| 698 | 4405 |
| 4406 /* Skip words with the NOSUGGEST flag. */ | |
| 4407 if (flags & WF_NOSUGGEST) | |
| 4408 break; | |
| 4409 | |
| 625 | 4410 fword_ends = (fword[sp->ts_fidx] == NUL |
| 4411 || (soundfold | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
4412 ? VIM_ISWHITE(fword[sp->ts_fidx]) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4413 : !spell_iswordp(fword + sp->ts_fidx, curwin))); |
| 625 | 4414 tword[sp->ts_twordlen] = NUL; |
| 4415 | |
| 4416 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL | |
| 485 | 4417 && (sp->ts_flags & TSF_PREFIXOK) == 0) |
| 625 | 4418 { |
| 4419 /* There was a prefix before the word. Check that the prefix | |
| 4420 * can be used with this word. */ | |
| 4421 /* Count the length of the NULs in the prefix. If there are | |
| 4422 * none this must be the first try without a prefix. */ | |
| 4423 n = stack[sp->ts_prefixdepth].ts_arridx; | |
| 4424 len = pbyts[n++]; | |
| 4425 for (c = 0; c < len && pbyts[n + c] == 0; ++c) | |
| 4426 ; | |
| 4427 if (c > 0) | |
| 4428 { | |
| 4429 c = valid_word_prefix(c, n, flags, | |
| 483 | 4430 tword + sp->ts_splitoff, slang, FALSE); |
| 625 | 4431 if (c == 0) |
| 4432 break; | |
| 4433 | |
| 4434 /* Use the WF_RARE flag for a rare prefix. */ | |
| 4435 if (c & WF_RAREPFX) | |
| 4436 flags |= WF_RARE; | |
| 4437 | |
| 4438 /* Tricky: when checking for both prefix and compounding | |
| 4439 * we run into the prefix flag first. | |
| 4440 * Remember that it's OK, so that we accept the prefix | |
| 4441 * when arriving at a compound flag. */ | |
| 4442 sp->ts_flags |= TSF_PREFIXOK; | |
| 4443 } | |
| 4444 } | |
| 4445 | |
| 4446 /* Check NEEDCOMPOUND: can't use word without compounding. Do try | |
| 4447 * appending another compound word below. */ | |
| 4448 if (sp->ts_complen == sp->ts_compsplit && fword_ends | |
| 500 | 4449 && (flags & WF_NEEDCOMP)) |
| 625 | 4450 goodword_ends = FALSE; |
| 4451 else | |
| 4452 goodword_ends = TRUE; | |
| 4453 | |
| 4454 p = NULL; | |
| 4455 compound_ok = TRUE; | |
| 4456 if (sp->ts_complen > sp->ts_compsplit) | |
| 4457 { | |
| 4458 if (slang->sl_nobreak) | |
| 4459 { | |
| 4460 /* There was a word before this word. When there was no | |
| 4461 * change in this word (it was correct) add the first word | |
| 4462 * as a suggestion. If this word was corrected too, we | |
| 4463 * need to check if a correct word follows. */ | |
| 4464 if (sp->ts_fidx - sp->ts_splitfidx | |
| 4465 == sp->ts_twordlen - sp->ts_splitoff | |
| 4466 && STRNCMP(fword + sp->ts_splitfidx, | |
| 4467 tword + sp->ts_splitoff, | |
| 4468 sp->ts_fidx - sp->ts_splitfidx) == 0) | |
| 492 | 4469 { |
| 625 | 4470 preword[sp->ts_prewordlen] = NUL; |
| 4471 newscore = score_wordcount_adj(slang, sp->ts_score, | |
| 4472 preword + sp->ts_prewordlen, | |
| 4473 sp->ts_prewordlen > 0); | |
| 4474 /* Add the suggestion if the score isn't too bad. */ | |
| 4475 if (newscore <= su->su_maxscore) | |
| 492 | 4476 add_suggestion(su, &su->su_ga, preword, |
| 503 | 4477 sp->ts_splitfidx - repextra, |
| 625 | 4478 newscore, 0, FALSE, |
| 4479 lp->lp_sallang, FALSE); | |
| 4480 break; | |
| 492 | 4481 } |
| 490 | 4482 } |
| 4483 else | |
| 625 | 4484 { |
| 4485 /* There was a compound word before this word. If this | |
| 4486 * word does not support compounding then give up | |
| 4487 * (splitting is tried for the word without compound | |
| 4488 * flag). */ | |
| 4489 if (((unsigned)flags >> 24) == 0 | |
| 4490 || sp->ts_twordlen - sp->ts_splitoff | |
| 4491 < slang->sl_compminlen) | |
| 4492 break; | |
| 4493 /* For multi-byte chars check character length against | |
| 4494 * COMPOUNDMIN. */ | |
| 4495 if (has_mbyte | |
| 4496 && slang->sl_compminlen > 0 | |
| 4497 && mb_charlen(tword + sp->ts_splitoff) | |
| 4498 < slang->sl_compminlen) | |
| 4499 break; | |
| 4500 | |
| 4501 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 4502 compflags[sp->ts_complen + 1] = NUL; | |
| 4503 vim_strncpy(preword + sp->ts_prewordlen, | |
| 4504 tword + sp->ts_splitoff, | |
| 4505 sp->ts_twordlen - sp->ts_splitoff); | |
| 1762 | 4506 |
| 4507 /* Verify CHECKCOMPOUNDPATTERN rules. */ | |
| 4508 if (match_checkcompoundpattern(preword, sp->ts_prewordlen, | |
| 4509 &slang->sl_comppat)) | |
| 4510 compound_ok = FALSE; | |
| 4511 | |
| 4512 if (compound_ok) | |
| 4513 { | |
| 4514 p = preword; | |
| 4515 while (*skiptowhite(p) != NUL) | |
| 4516 p = skipwhite(skiptowhite(p)); | |
| 4517 if (fword_ends && !can_compound(slang, p, | |
| 625 | 4518 compflags + sp->ts_compsplit)) |
| 1762 | 4519 /* Compound is not allowed. But it may still be |
| 4520 * possible if we add another (short) word. */ | |
| 4521 compound_ok = FALSE; | |
| 4522 } | |
| 625 | 4523 |
| 4524 /* Get pointer to last char of previous word. */ | |
| 4525 p = preword + sp->ts_prewordlen; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
4526 MB_PTR_BACK(preword, p); |
| 625 | 4527 } |
| 4528 } | |
| 4529 | |
| 4530 /* | |
| 4531 * Form the word with proper case in preword. | |
| 4532 * If there is a word from a previous split, append. | |
| 4533 * For the soundfold tree don't change the case, simply append. | |
| 4534 */ | |
| 4535 if (soundfold) | |
| 4536 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff); | |
| 4537 else if (flags & WF_KEEPCAP) | |
| 4538 /* Must find the word in the keep-case tree. */ | |
| 4539 find_keepcap_word(slang, tword + sp->ts_splitoff, | |
| 483 | 4540 preword + sp->ts_prewordlen); |
| 625 | 4541 else |
| 4542 { | |
| 4543 /* Include badflags: If the badword is onecap or allcap | |
| 4544 * use that for the goodword too. But if the badword is | |
| 4545 * allcap and it's only one char long use onecap. */ | |
| 4546 c = su->su_badflags; | |
| 4547 if ((c & WF_ALLCAP) | |
|
15605
62b3805506b3
patch 8.1.0810: too many #ifdefs
Bram Moolenaar <Bram@vim.org>
parents:
15543
diff
changeset
|
4548 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)) |
| 625 | 4549 c = WF_ONECAP; |
| 4550 c |= flags; | |
| 4551 | |
| 4552 /* When appending a compound word after a word character don't | |
| 4553 * use Onecap. */ | |
| 5477 | 4554 if (p != NULL && spell_iswordp_nmw(p, curwin)) |
| 625 | 4555 c &= ~WF_ONECAP; |
| 4556 make_case_word(tword + sp->ts_splitoff, | |
| 490 | 4557 preword + sp->ts_prewordlen, c); |
| 625 | 4558 } |
| 4559 | |
| 4560 if (!soundfold) | |
| 4561 { | |
| 323 | 4562 /* Don't use a banned word. It may appear again as a good |
| 4563 * word, thus remember it. */ | |
| 4564 if (flags & WF_BANNED) | |
| 4565 { | |
| 483 | 4566 add_banned(su, preword + sp->ts_prewordlen); |
| 323 | 4567 break; |
| 4568 } | |
| 534 | 4569 if ((sp->ts_complen == sp->ts_compsplit |
| 625 | 4570 && WAS_BANNED(su, preword + sp->ts_prewordlen)) |
| 4571 || WAS_BANNED(su, preword)) | |
| 534 | 4572 { |
| 4573 if (slang->sl_compprog == NULL) | |
| 4574 break; | |
| 4575 /* the word so far was banned but we may try compounding */ | |
| 4576 goodword_ends = FALSE; | |
| 4577 } | |
| 625 | 4578 } |
| 4579 | |
| 4580 newscore = 0; | |
| 4581 if (!soundfold) /* soundfold words don't have flags */ | |
| 4582 { | |
| 323 | 4583 if ((flags & WF_REGION) |
| 390 | 4584 && (((unsigned)flags >> 16) & lp->lp_region) == 0) |
| 323 | 4585 newscore += SCORE_REGION; |
| 4586 if (flags & WF_RARE) | |
| 4587 newscore += SCORE_RARE; | |
| 4588 | |
| 346 | 4589 if (!spell_valid_case(su->su_badflags, |
| 483 | 4590 captype(preword + sp->ts_prewordlen, NULL))) |
| 323 | 4591 newscore += SCORE_ICASE; |
| 625 | 4592 } |
| 4593 | |
| 4594 /* TODO: how about splitting in the soundfold tree? */ | |
| 4595 if (fword_ends | |
| 4596 && goodword_ends | |
| 4597 && sp->ts_fidx >= sp->ts_fidxtry | |
| 4598 && compound_ok) | |
| 4599 { | |
| 4600 /* The badword also ends: add suggestions. */ | |
| 4601 #ifdef DEBUG_TRIEWALK | |
| 4602 if (soundfold && STRCMP(preword, "smwrd") == 0) | |
| 4603 { | |
| 4604 int j; | |
| 4605 | |
| 4606 /* print the stack of changes that brought us here */ | |
| 4607 smsg("------ %s -------", fword); | |
| 4608 for (j = 0; j < depth; ++j) | |
| 4609 smsg("%s", changename[j]); | |
| 4610 } | |
| 4611 #endif | |
| 4612 if (soundfold) | |
| 4613 { | |
| 4614 /* For soundfolded words we need to find the original | |
| 1188 | 4615 * words, the edit distance and then add them. */ |
| 625 | 4616 add_sound_suggest(su, preword, sp->ts_score, lp); |
| 4617 } | |
|
2457
6c1d10a85398
Avoid illegal memory access in spell suggestion. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2455
diff
changeset
|
4618 else if (sp->ts_fidx > 0) |
| 625 | 4619 { |
| 4620 /* Give a penalty when changing non-word char to word | |
| 4621 * char, e.g., "thes," -> "these". */ | |
| 366 | 4622 p = fword + sp->ts_fidx; |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
4623 MB_PTR_BACK(fword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4624 if (!spell_iswordp(p, curwin)) |
| 366 | 4625 { |
| 4626 p = preword + STRLEN(preword); | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
4627 MB_PTR_BACK(preword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4628 if (spell_iswordp(p, curwin)) |
| 366 | 4629 newscore += SCORE_NONWORD; |
| 4630 } | |
| 4631 | |
| 625 | 4632 /* Give a bonus to words seen before. */ |
| 4633 score = score_wordcount_adj(slang, | |
| 4634 sp->ts_score + newscore, | |
| 4635 preword + sp->ts_prewordlen, | |
| 4636 sp->ts_prewordlen > 0); | |
| 4637 | |
| 4638 /* Add the suggestion if the score isn't too bad. */ | |
| 4639 if (score <= su->su_maxscore) | |
| 633 | 4640 { |
| 625 | 4641 add_suggestion(su, &su->su_ga, preword, |
| 4642 sp->ts_fidx - repextra, | |
| 4643 score, 0, FALSE, lp->lp_sallang, FALSE); | |
| 633 | 4644 |
| 4645 if (su->su_badflags & WF_MIXCAP) | |
| 4646 { | |
| 4647 /* We really don't know if the word should be | |
| 4648 * upper or lower case, add both. */ | |
| 4649 c = captype(preword, NULL); | |
| 4650 if (c == 0 || c == WF_ALLCAP) | |
| 4651 { | |
| 4652 make_case_word(tword + sp->ts_splitoff, | |
| 4653 preword + sp->ts_prewordlen, | |
| 4654 c == 0 ? WF_ALLCAP : 0); | |
| 4655 | |
| 4656 add_suggestion(su, &su->su_ga, preword, | |
| 4657 sp->ts_fidx - repextra, | |
| 4658 score + SCORE_ICASE, 0, FALSE, | |
| 4659 lp->lp_sallang, FALSE); | |
| 4660 } | |
| 4661 } | |
| 4662 } | |
| 625 | 4663 } |
| 4664 } | |
| 4665 | |
| 4666 /* | |
| 4667 * Try word split and/or compounding. | |
| 4668 */ | |
| 4669 if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends) | |
| 4670 /* Don't split halfway a character. */ | |
|
15605
62b3805506b3
patch 8.1.0810: too many #ifdefs
Bram Moolenaar <Bram@vim.org>
parents:
15543
diff
changeset
|
4671 && (!has_mbyte || sp->ts_tcharlen == 0)) |
| 625 | 4672 { |
| 4673 int try_compound; | |
| 4674 int try_split; | |
| 4675 | |
| 4676 /* If past the end of the bad word don't try a split. | |
| 4677 * Otherwise try changing the next word. E.g., find | |
| 4678 * suggestions for "the the" where the second "the" is | |
| 4679 * different. It's done like a split. | |
| 4680 * TODO: word split for soundfold words */ | |
| 4681 try_split = (sp->ts_fidx - repextra < su->su_badlen) | |
| 4682 && !soundfold; | |
| 4683 | |
| 4684 /* Get here in several situations: | |
| 4685 * 1. The word in the tree ends: | |
| 4686 * If the word allows compounding try that. Otherwise try | |
| 4687 * a split by inserting a space. For both check that a | |
| 4688 * valid words starts at fword[sp->ts_fidx]. | |
| 4689 * For NOBREAK do like compounding to be able to check if | |
| 4690 * the next word is valid. | |
| 4691 * 2. The badword does end, but it was due to a change (e.g., | |
| 4692 * a swap). No need to split, but do check that the | |
| 4693 * following word is valid. | |
| 4694 * 3. The badword and the word in the tree end. It may still | |
| 4695 * be possible to compound another (short) word. | |
| 4696 */ | |
| 4697 try_compound = FALSE; | |
| 4698 if (!soundfold | |
|
7526
08c1f73efcde
commit https://github.com/vim/vim/commit/7b877b360532713dc21a0ff3d55a76ac02eaf573
Christian Brabandt <cb@256bit.org>
parents:
7509
diff
changeset
|
4699 && !slang->sl_nocompoundsugs |
| 625 | 4700 && slang->sl_compprog != NULL |
| 4701 && ((unsigned)flags >> 24) != 0 | |
| 4702 && sp->ts_twordlen - sp->ts_splitoff | |
| 4703 >= slang->sl_compminlen | |
| 4704 && (!has_mbyte | |
| 4705 || slang->sl_compminlen == 0 | |
| 4706 || mb_charlen(tword + sp->ts_splitoff) | |
| 4707 >= slang->sl_compminlen) | |
| 4708 && (slang->sl_compsylmax < MAXWLEN | |
| 4709 || sp->ts_complen + 1 - sp->ts_compsplit | |
| 4710 < slang->sl_compmax) | |
| 1762 | 4711 && (can_be_compound(sp, slang, |
| 4712 compflags, ((unsigned)flags >> 24)))) | |
| 4713 | |
| 625 | 4714 { |
| 4715 try_compound = TRUE; | |
| 4716 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 4717 compflags[sp->ts_complen + 1] = NUL; | |
| 4718 } | |
| 4719 | |
| 4720 /* For NOBREAK we never try splitting, it won't make any word | |
| 4721 * valid. */ | |
|
7526
08c1f73efcde
commit https://github.com/vim/vim/commit/7b877b360532713dc21a0ff3d55a76ac02eaf573
Christian Brabandt <cb@256bit.org>
parents:
7509
diff
changeset
|
4722 if (slang->sl_nobreak && !slang->sl_nocompoundsugs) |
| 625 | 4723 try_compound = TRUE; |
| 4724 | |
| 4725 /* If we could add a compound word, and it's also possible to | |
| 4726 * split at this point, do the split first and set | |
| 4727 * TSF_DIDSPLIT to avoid doing it again. */ | |
| 4728 else if (!fword_ends | |
| 4729 && try_compound | |
| 4730 && (sp->ts_flags & TSF_DIDSPLIT) == 0) | |
| 4731 { | |
| 485 | 4732 try_compound = FALSE; |
| 625 | 4733 sp->ts_flags |= TSF_DIDSPLIT; |
| 4734 --sp->ts_curi; /* do the same NUL again */ | |
| 4735 compflags[sp->ts_complen] = NUL; | |
| 4736 } | |
| 4737 else | |
| 4738 sp->ts_flags &= ~TSF_DIDSPLIT; | |
| 4739 | |
| 4740 if (try_split || try_compound) | |
| 4741 { | |
| 534 | 4742 if (!try_compound && (!fword_ends || !goodword_ends)) |
| 485 | 4743 { |
| 4744 /* If we're going to split need to check that the | |
| 501 | 4745 * words so far are valid for compounding. If there |
| 4746 * is only one word it must not have the NEEDCOMPOUND | |
| 4747 * flag. */ | |
| 4748 if (sp->ts_complen == sp->ts_compsplit | |
| 4749 && (flags & WF_NEEDCOMP)) | |
| 4750 break; | |
| 490 | 4751 p = preword; |
| 4752 while (*skiptowhite(p) != NUL) | |
| 4753 p = skipwhite(skiptowhite(p)); | |
| 485 | 4754 if (sp->ts_complen > sp->ts_compsplit |
| 490 | 4755 && !can_compound(slang, p, |
| 485 | 4756 compflags + sp->ts_compsplit)) |
| 4757 break; | |
| 698 | 4758 |
| 4759 if (slang->sl_nosplitsugs) | |
| 4760 newscore += SCORE_SPLIT_NO; | |
| 4761 else | |
| 4762 newscore += SCORE_SPLIT; | |
| 625 | 4763 |
| 4764 /* Give a bonus to words seen before. */ | |
| 4765 newscore = score_wordcount_adj(slang, newscore, | |
| 4766 preword + sp->ts_prewordlen, TRUE); | |
| 482 | 4767 } |
| 4768 | |
| 625 | 4769 if (TRY_DEEPER(su, stack, depth, newscore)) |
| 323 | 4770 { |
| 625 | 4771 go_deeper(stack, depth, newscore); |
| 4772 #ifdef DEBUG_TRIEWALK | |
| 4773 if (!try_compound && !fword_ends) | |
| 4774 sprintf(changename[depth], "%.*s-%s: split", | |
| 4775 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 4776 else | |
| 4777 sprintf(changename[depth], "%.*s-%s: compound", | |
| 4778 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 4779 #endif | |
| 323 | 4780 /* Save things to be restored at STATE_SPLITUNDO. */ |
| 346 | 4781 sp->ts_save_badflags = su->su_badflags; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4782 PROF_STORE(sp->ts_state) |
| 376 | 4783 sp->ts_state = STATE_SPLITUNDO; |
| 4784 | |
| 4785 ++depth; | |
| 4786 sp = &stack[depth]; | |
| 323 | 4787 |
| 482 | 4788 /* Append a space to preword when splitting. */ |
| 4789 if (!try_compound && !fword_ends) | |
| 4790 STRCAT(preword, " "); | |
| 835 | 4791 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 483 | 4792 sp->ts_splitoff = sp->ts_twordlen; |
| 492 | 4793 sp->ts_splitfidx = sp->ts_fidx; |
| 376 | 4794 |
| 4795 /* If the badword has a non-word character at this | |
| 4796 * position skip it. That means replacing the | |
| 482 | 4797 * non-word character with a space. Always skip a |
| 534 | 4798 * character when the word ends. But only when the |
| 4799 * good word can end. */ | |
| 625 | 4800 if (((!try_compound && !spell_iswordp_nmw(fword |
| 5477 | 4801 + sp->ts_fidx, |
| 4802 curwin)) | |
| 625 | 4803 || fword_ends) |
| 4804 && fword[sp->ts_fidx] != NUL | |
| 4805 && goodword_ends) | |
| 376 | 4806 { |
| 482 | 4807 int l; |
| 4808 | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
4809 l = MB_PTR2LEN(fword + sp->ts_fidx); |
| 482 | 4810 if (fword_ends) |
| 4811 { | |
| 4812 /* Copy the skipped character to preword. */ | |
| 483 | 4813 mch_memmove(preword + sp->ts_prewordlen, |
| 482 | 4814 fword + sp->ts_fidx, l); |
| 483 | 4815 sp->ts_prewordlen += l; |
| 4816 preword[sp->ts_prewordlen] = NUL; | |
| 482 | 4817 } |
| 4818 else | |
| 4819 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST; | |
| 4820 sp->ts_fidx += l; | |
| 376 | 4821 } |
| 455 | 4822 |
| 485 | 4823 /* When compounding include compound flag in |
| 4824 * compflags[] (already set above). When splitting we | |
| 4825 * may start compounding over again. */ | |
| 482 | 4826 if (try_compound) |
| 483 | 4827 ++sp->ts_complen; |
| 482 | 4828 else |
| 485 | 4829 sp->ts_compsplit = sp->ts_complen; |
| 4830 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 482 | 4831 |
| 455 | 4832 /* set su->su_badflags to the caps type at this |
| 4833 * position */ | |
| 324 | 4834 if (has_mbyte) |
| 455 | 4835 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); |
| 324 | 4836 else |
| 455 | 4837 n = sp->ts_fidx; |
| 474 | 4838 su->su_badflags = badword_captype(su->su_badptr + n, |
| 455 | 4839 su->su_badptr + su->su_badlen); |
| 323 | 4840 |
| 4841 /* Restart at top of the tree. */ | |
| 376 | 4842 sp->ts_arridx = 0; |
| 485 | 4843 |
| 4844 /* If there are postponed prefixes, try these too. */ | |
| 4845 if (pbyts != NULL) | |
| 4846 { | |
| 4847 byts = pbyts; | |
| 4848 idxs = pidxs; | |
| 4849 sp->ts_prefixdepth = PFD_PREFIXTREE; | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4850 PROF_STORE(sp->ts_state) |
| 485 | 4851 sp->ts_state = STATE_NOPREFIX; |
| 4852 } | |
| 323 | 4853 } |
| 4854 } | |
| 625 | 4855 } |
| 4856 break; | |
| 4857 | |
| 4858 case STATE_SPLITUNDO: | |
| 4859 /* Undo the changes done for word split or compound word. */ | |
| 4860 su->su_badflags = sp->ts_save_badflags; | |
| 4861 | |
| 4862 /* Continue looking for NUL bytes. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4863 PROF_STORE(sp->ts_state) |
| 625 | 4864 sp->ts_state = STATE_START; |
| 4865 | |
| 4866 /* In case we went into the prefix tree. */ | |
| 4867 byts = fbyts; | |
| 4868 idxs = fidxs; | |
| 4869 break; | |
| 4870 | |
| 4871 case STATE_ENDNUL: | |
| 4872 /* Past the NUL bytes in the node. */ | |
| 4873 su->su_badflags = sp->ts_save_badflags; | |
|
15605
62b3805506b3
patch 8.1.0810: too many #ifdefs
Bram Moolenaar <Bram@vim.org>
parents:
15543
diff
changeset
|
4874 if (fword[sp->ts_fidx] == NUL && sp->ts_tcharlen == 0) |
| 625 | 4875 { |
| 4876 /* The badword ends, can't use STATE_PLAIN. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4877 PROF_STORE(sp->ts_state) |
| 625 | 4878 sp->ts_state = STATE_DEL; |
| 323 | 4879 break; |
| 625 | 4880 } |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4881 PROF_STORE(sp->ts_state) |
| 625 | 4882 sp->ts_state = STATE_PLAIN; |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
4883 /* FALLTHROUGH */ |
| 625 | 4884 |
| 4885 case STATE_PLAIN: | |
| 4886 /* | |
| 4887 * Go over all possible bytes at this node, add each to tword[] | |
| 4888 * and use child node. "ts_curi" is the index. | |
| 4889 */ | |
| 4890 arridx = sp->ts_arridx; | |
| 4891 if (sp->ts_curi > byts[arridx]) | |
| 4892 { | |
| 4893 /* Done all bytes at this node, do next state. When still at | |
| 4894 * already changed bytes skip the other tricks. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4895 PROF_STORE(sp->ts_state) |
| 625 | 4896 if (sp->ts_fidx >= sp->ts_fidxtry) |
| 323 | 4897 sp->ts_state = STATE_DEL; |
| 4898 else | |
| 625 | 4899 sp->ts_state = STATE_FINAL; |
| 4900 } | |
| 4901 else | |
| 4902 { | |
| 4903 arridx += sp->ts_curi++; | |
| 4904 c = byts[arridx]; | |
| 4905 | |
| 4906 /* Normal byte, go one level deeper. If it's not equal to the | |
| 4907 * byte in the bad word adjust the score. But don't even try | |
| 4908 * when the byte was already changed. And don't try when we | |
| 5695 | 4909 * just deleted this byte, accepting it is always cheaper than |
| 625 | 4910 * delete + substitute. */ |
| 4911 if (c == fword[sp->ts_fidx] | |
|
15605
62b3805506b3
patch 8.1.0810: too many #ifdefs
Bram Moolenaar <Bram@vim.org>
parents:
15543
diff
changeset
|
4912 || (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE)) |
| 625 | 4913 newscore = 0; |
| 4914 else | |
| 4915 newscore = SCORE_SUBST; | |
| 4916 if ((newscore == 0 | |
| 4917 || (sp->ts_fidx >= sp->ts_fidxtry | |
| 4918 && ((sp->ts_flags & TSF_DIDDEL) == 0 | |
| 4919 || c != fword[sp->ts_delidx]))) | |
| 4920 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 4921 { | |
| 4922 go_deeper(stack, depth, newscore); | |
| 4923 #ifdef DEBUG_TRIEWALK | |
| 4924 if (newscore > 0) | |
| 4925 sprintf(changename[depth], "%.*s-%s: subst %c to %c", | |
| 4926 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 4927 fword[sp->ts_fidx], c); | |
| 323 | 4928 else |
| 625 | 4929 sprintf(changename[depth], "%.*s-%s: accept %c", |
| 4930 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 4931 fword[sp->ts_fidx]); | |
| 4932 #endif | |
| 323 | 4933 ++depth; |
| 625 | 4934 sp = &stack[depth]; |
| 4935 ++sp->ts_fidx; | |
| 4936 tword[sp->ts_twordlen++] = c; | |
| 4937 sp->ts_arridx = idxs[arridx]; | |
| 4938 if (newscore == SCORE_SUBST) | |
| 4939 sp->ts_isdiff = DIFF_YES; | |
| 330 | 4940 if (has_mbyte) |
| 358 | 4941 { |
| 625 | 4942 /* Multi-byte characters are a bit complicated to |
| 4943 * handle: They differ when any of the bytes differ | |
| 4944 * and then their length may also differ. */ | |
| 4945 if (sp->ts_tcharlen == 0) | |
| 4946 { | |
| 4947 /* First byte. */ | |
| 4948 sp->ts_tcharidx = 0; | |
| 4949 sp->ts_tcharlen = MB_BYTE2LEN(c); | |
| 4950 sp->ts_fcharstart = sp->ts_fidx - 1; | |
| 4951 sp->ts_isdiff = (newscore != 0) | |
| 4952 ? DIFF_YES : DIFF_NONE; | |
| 4953 } | |
| 4954 else if (sp->ts_isdiff == DIFF_INSERT) | |
| 4955 /* When inserting trail bytes don't advance in the | |
| 4956 * bad word. */ | |
| 4957 --sp->ts_fidx; | |
| 4958 if (++sp->ts_tcharidx == sp->ts_tcharlen) | |
| 4959 { | |
| 4960 /* Last byte of character. */ | |
| 4961 if (sp->ts_isdiff == DIFF_YES) | |
| 4962 { | |
| 4963 /* Correct ts_fidx for the byte length of the | |
| 4964 * character (we didn't check that before). */ | |
| 4965 sp->ts_fidx = sp->ts_fcharstart | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
4966 + MB_PTR2LEN( |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
4967 fword + sp->ts_fcharstart); |
| 625 | 4968 /* For changing a composing character adjust |
| 4969 * the score from SCORE_SUBST to | |
| 4970 * SCORE_SUBCOMP. */ | |
| 4971 if (enc_utf8 | |
| 4972 && utf_iscomposing( | |
|
11269
121d29004998
patch 8.0.0520: using a function pointer while the function is known
Christian Brabandt <cb@256bit.org>
parents:
11129
diff
changeset
|
4973 utf_ptr2char(tword |
| 625 | 4974 + sp->ts_twordlen |
| 4975 - sp->ts_tcharlen)) | |
| 4976 && utf_iscomposing( | |
|
11269
121d29004998
patch 8.0.0520: using a function pointer while the function is known
Christian Brabandt <cb@256bit.org>
parents:
11129
diff
changeset
|
4977 utf_ptr2char(fword |
| 625 | 4978 + sp->ts_fcharstart))) |
| 4979 sp->ts_score -= | |
| 4980 SCORE_SUBST - SCORE_SUBCOMP; | |
| 4981 | |
| 4982 /* For a similar character adjust score from | |
| 4983 * SCORE_SUBST to SCORE_SIMILAR. */ | |
| 4984 else if (!soundfold | |
| 4985 && slang->sl_has_map | |
| 4986 && similar_chars(slang, | |
| 4987 mb_ptr2char(tword | |
| 4988 + sp->ts_twordlen | |
| 4989 - sp->ts_tcharlen), | |
| 4990 mb_ptr2char(fword | |
| 4991 + sp->ts_fcharstart))) | |
| 4992 sp->ts_score -= | |
| 4993 SCORE_SUBST - SCORE_SIMILAR; | |
| 4994 } | |
| 4995 else if (sp->ts_isdiff == DIFF_INSERT | |
| 4996 && sp->ts_twordlen > sp->ts_tcharlen) | |
| 4997 { | |
| 4998 p = tword + sp->ts_twordlen - sp->ts_tcharlen; | |
| 4999 c = mb_ptr2char(p); | |
| 5000 if (enc_utf8 && utf_iscomposing(c)) | |
| 5001 { | |
| 5002 /* Inserting a composing char doesn't | |
| 5003 * count that much. */ | |
| 5004 sp->ts_score -= SCORE_INS - SCORE_INSCOMP; | |
| 5005 } | |
| 5006 else | |
| 5007 { | |
| 5008 /* If the previous character was the same, | |
| 5009 * thus doubling a character, give a bonus | |
| 5010 * to the score. Also for the soundfold | |
| 5011 * tree (might seem illogical but does | |
| 5012 * give better scores). */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5013 MB_PTR_BACK(tword, p); |
| 625 | 5014 if (c == mb_ptr2char(p)) |
| 5015 sp->ts_score -= SCORE_INS | |
| 5016 - SCORE_INSDUP; | |
| 5017 } | |
| 5018 } | |
| 5019 | |
| 5020 /* Starting a new char, reset the length. */ | |
| 5021 sp->ts_tcharlen = 0; | |
| 5022 } | |
| 358 | 5023 } |
| 330 | 5024 else |
| 358 | 5025 { |
| 625 | 5026 /* If we found a similar char adjust the score. |
| 5027 * We do this after calling go_deeper() because | |
| 5028 * it's slow. */ | |
| 5029 if (newscore != 0 | |
| 5030 && !soundfold | |
| 5031 && slang->sl_has_map | |
| 5032 && similar_chars(slang, | |
| 5033 c, fword[sp->ts_fidx - 1])) | |
| 5034 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR; | |
| 358 | 5035 } |
| 625 | 5036 } |
| 5037 } | |
| 5038 break; | |
| 5039 | |
| 5040 case STATE_DEL: | |
| 5041 /* When past the first byte of a multi-byte char don't try | |
| 5042 * delete/insert/swap a character. */ | |
| 5043 if (has_mbyte && sp->ts_tcharlen > 0) | |
| 5044 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5045 PROF_STORE(sp->ts_state) |
| 625 | 5046 sp->ts_state = STATE_FINAL; |
| 5047 break; | |
| 5048 } | |
| 5049 /* | |
| 5050 * Try skipping one character in the bad word (delete it). | |
| 5051 */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5052 PROF_STORE(sp->ts_state) |
| 625 | 5053 sp->ts_state = STATE_INS_PREP; |
| 5054 sp->ts_curi = 1; | |
| 5055 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*') | |
| 5056 /* Deleting a vowel at the start of a word counts less, see | |
| 5057 * soundalike_score(). */ | |
| 5058 newscore = 2 * SCORE_DEL / 3; | |
| 5059 else | |
| 5060 newscore = SCORE_DEL; | |
| 5061 if (fword[sp->ts_fidx] != NUL | |
| 5062 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5063 { | |
| 5064 go_deeper(stack, depth, newscore); | |
| 5065 #ifdef DEBUG_TRIEWALK | |
| 5066 sprintf(changename[depth], "%.*s-%s: delete %c", | |
| 5067 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5068 fword[sp->ts_fidx]); | |
| 5069 #endif | |
| 5070 ++depth; | |
| 5071 | |
| 5072 /* Remember what character we deleted, so that we can avoid | |
| 5073 * inserting it again. */ | |
| 5074 stack[depth].ts_flags |= TSF_DIDDEL; | |
| 5075 stack[depth].ts_delidx = sp->ts_fidx; | |
| 5076 | |
| 5077 /* Advance over the character in fword[]. Give a bonus to the | |
| 5078 * score if the same character is following "nn" -> "n". It's | |
| 5079 * a bit illogical for soundfold tree but it does give better | |
| 5080 * results. */ | |
| 5081 if (has_mbyte) | |
| 5082 { | |
| 5083 c = mb_ptr2char(fword + sp->ts_fidx); | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5084 stack[depth].ts_fidx += MB_PTR2LEN(fword + sp->ts_fidx); |
| 625 | 5085 if (enc_utf8 && utf_iscomposing(c)) |
| 5086 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP; | |
| 5087 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx)) | |
| 5088 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 5089 } |
| 5090 else | |
| 625 | 5091 { |
| 5092 ++stack[depth].ts_fidx; | |
| 5093 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1]) | |
| 5094 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 5095 } |
| 5096 break; | |
| 625 | 5097 } |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5098 /* FALLTHROUGH */ |
| 625 | 5099 |
| 5100 case STATE_INS_PREP: | |
| 5101 if (sp->ts_flags & TSF_DIDDEL) | |
| 5102 { | |
| 5103 /* If we just deleted a byte then inserting won't make sense, | |
| 5104 * a substitute is always cheaper. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5105 PROF_STORE(sp->ts_state) |
| 625 | 5106 sp->ts_state = STATE_SWAP; |
| 5107 break; | |
| 5108 } | |
| 5109 | |
| 5110 /* skip over NUL bytes */ | |
| 5111 n = sp->ts_arridx; | |
| 5112 for (;;) | |
| 5113 { | |
| 5114 if (sp->ts_curi > byts[n]) | |
| 5115 { | |
| 5116 /* Only NUL bytes at this node, go to next state. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5117 PROF_STORE(sp->ts_state) |
| 625 | 5118 sp->ts_state = STATE_SWAP; |
| 5119 break; | |
| 5120 } | |
| 5121 if (byts[n + sp->ts_curi] != NUL) | |
| 5122 { | |
| 5123 /* Found a byte to insert. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5124 PROF_STORE(sp->ts_state) |
| 625 | 5125 sp->ts_state = STATE_INS; |
| 330 | 5126 break; |
| 5127 } | |
| 625 | 5128 ++sp->ts_curi; |
| 5129 } | |
| 5130 break; | |
| 5131 | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5132 /* FALLTHROUGH */ |
| 625 | 5133 |
| 5134 case STATE_INS: | |
| 5135 /* Insert one byte. Repeat this for each possible byte at this | |
| 5136 * node. */ | |
| 5137 n = sp->ts_arridx; | |
| 5138 if (sp->ts_curi > byts[n]) | |
| 5139 { | |
| 5140 /* Done all bytes at this node, go to next state. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5141 PROF_STORE(sp->ts_state) |
| 625 | 5142 sp->ts_state = STATE_SWAP; |
| 5143 break; | |
| 5144 } | |
| 5145 | |
| 5146 /* Do one more byte at this node, but: | |
| 5147 * - Skip NUL bytes. | |
| 5148 * - Skip the byte if it's equal to the byte in the word, | |
| 5149 * accepting that byte is always better. | |
| 5150 */ | |
| 5151 n += sp->ts_curi++; | |
| 5152 c = byts[n]; | |
| 5153 if (soundfold && sp->ts_twordlen == 0 && c == '*') | |
| 5154 /* Inserting a vowel at the start of a word counts less, | |
| 5155 * see soundalike_score(). */ | |
| 5156 newscore = 2 * SCORE_INS / 3; | |
| 5157 else | |
| 5158 newscore = SCORE_INS; | |
| 5159 if (c != fword[sp->ts_fidx] | |
| 5160 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5161 { | |
| 5162 go_deeper(stack, depth, newscore); | |
| 5163 #ifdef DEBUG_TRIEWALK | |
| 5164 sprintf(changename[depth], "%.*s-%s: insert %c", | |
| 5165 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5166 c); | |
| 5167 #endif | |
| 5168 ++depth; | |
| 5169 sp = &stack[depth]; | |
| 5170 tword[sp->ts_twordlen++] = c; | |
| 5171 sp->ts_arridx = idxs[n]; | |
| 330 | 5172 if (has_mbyte) |
| 5173 { | |
| 625 | 5174 fl = MB_BYTE2LEN(c); |
| 5175 if (fl > 1) | |
| 330 | 5176 { |
| 625 | 5177 /* There are following bytes for the same character. |
| 5178 * We must find all bytes before trying | |
| 5179 * delete/insert/swap/etc. */ | |
| 5180 sp->ts_tcharlen = fl; | |
| 5181 sp->ts_tcharidx = 1; | |
| 5182 sp->ts_isdiff = DIFF_INSERT; | |
| 330 | 5183 } |
| 323 | 5184 } |
| 5185 else | |
| 625 | 5186 fl = 1; |
| 5187 if (fl == 1) | |
| 5188 { | |
| 5189 /* If the previous character was the same, thus doubling a | |
| 5190 * character, give a bonus to the score. Also for | |
| 5191 * soundfold words (illogical but does give a better | |
| 5192 * score). */ | |
| 5193 if (sp->ts_twordlen >= 2 | |
| 5194 && tword[sp->ts_twordlen - 2] == c) | |
| 5195 sp->ts_score -= SCORE_INS - SCORE_INSDUP; | |
| 5196 } | |
| 5197 } | |
| 5198 break; | |
| 5199 | |
| 5200 case STATE_SWAP: | |
| 5201 /* | |
| 5202 * Swap two bytes in the bad word: "12" -> "21". | |
| 5203 * We change "fword" here, it's changed back afterwards at | |
| 5204 * STATE_UNSWAP. | |
| 5205 */ | |
| 5206 p = fword + sp->ts_fidx; | |
| 5207 c = *p; | |
| 5208 if (c == NUL) | |
| 5209 { | |
| 5210 /* End of word, can't swap or replace. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5211 PROF_STORE(sp->ts_state) |
| 625 | 5212 sp->ts_state = STATE_FINAL; |
| 5213 break; | |
| 5214 } | |
| 5215 | |
| 5216 /* Don't swap if the first character is not a word character. | |
| 5217 * 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
|
5218 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 5219 { |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5220 PROF_STORE(sp->ts_state) |
| 625 | 5221 sp->ts_state = STATE_REP_INI; |
| 323 | 5222 break; |
| 625 | 5223 } |
| 5224 | |
| 5225 if (has_mbyte) | |
| 5226 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5227 n = MB_CPTR2LEN(p); |
| 625 | 5228 c = mb_ptr2char(p); |
| 1337 | 5229 if (p[n] == NUL) |
| 5230 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5231 else if (!soundfold && !spell_iswordp(p + n, curwin)) |
| 625 | 5232 c2 = c; /* don't swap non-word char */ |
| 5233 else | |
| 5234 c2 = mb_ptr2char(p + n); | |
| 5235 } | |
| 5236 else | |
| 5237 { | |
| 1337 | 5238 if (p[1] == NUL) |
| 5239 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5240 else if (!soundfold && !spell_iswordp(p + 1, curwin)) |
| 625 | 5241 c2 = c; /* don't swap non-word char */ |
| 5242 else | |
| 5243 c2 = p[1]; | |
| 5244 } | |
| 5245 | |
| 1337 | 5246 /* When the second character is NUL we can't swap. */ |
| 5247 if (c2 == NUL) | |
| 5248 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5249 PROF_STORE(sp->ts_state) |
| 1337 | 5250 sp->ts_state = STATE_REP_INI; |
| 5251 break; | |
| 5252 } | |
| 5253 | |
| 625 | 5254 /* When characters are identical, swap won't do anything. |
| 5255 * Also get here if the second char is not a word character. */ | |
| 5256 if (c == c2) | |
| 5257 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5258 PROF_STORE(sp->ts_state) |
| 625 | 5259 sp->ts_state = STATE_SWAP3; |
| 5260 break; | |
| 5261 } | |
| 5262 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP)) | |
| 5263 { | |
| 5264 go_deeper(stack, depth, SCORE_SWAP); | |
| 5265 #ifdef DEBUG_TRIEWALK | |
| 5266 sprintf(changename[depth], "%.*s-%s: swap %c and %c", | |
| 5267 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5268 c, c2); | |
| 5269 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5270 PROF_STORE(sp->ts_state) |
| 625 | 5271 sp->ts_state = STATE_UNSWAP; |
| 5272 ++depth; | |
| 330 | 5273 if (has_mbyte) |
| 5274 { | |
| 625 | 5275 fl = mb_char2len(c2); |
| 5276 mch_memmove(p, p + n, fl); | |
| 5277 mb_char2bytes(c, p + fl); | |
| 5278 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 5279 } |
| 5280 else | |
| 5281 { | |
| 625 | 5282 p[0] = c2; |
| 330 | 5283 p[1] = c; |
| 625 | 5284 stack[depth].ts_fidxtry = sp->ts_fidx + 2; |
| 5285 } | |
| 5286 } | |
| 5287 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5288 { |
| 625 | 5289 /* If this swap doesn't work then SWAP3 won't either. */ |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5290 PROF_STORE(sp->ts_state) |
| 625 | 5291 sp->ts_state = STATE_REP_INI; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5292 } |
| 625 | 5293 break; |
| 5294 | |
| 5295 case STATE_UNSWAP: | |
| 5296 /* Undo the STATE_SWAP swap: "21" -> "12". */ | |
| 5297 p = fword + sp->ts_fidx; | |
| 5298 if (has_mbyte) | |
| 5299 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5300 n = MB_PTR2LEN(p); |
| 625 | 5301 c = mb_ptr2char(p + n); |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5302 mch_memmove(p + MB_PTR2LEN(p + n), p, n); |
| 625 | 5303 mb_char2bytes(c, p); |
| 5304 } | |
| 5305 else | |
| 5306 { | |
| 5307 c = *p; | |
| 5308 *p = p[1]; | |
| 5309 p[1] = c; | |
| 5310 } | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5311 /* FALLTHROUGH */ |
| 625 | 5312 |
| 5313 case STATE_SWAP3: | |
| 5314 /* Swap two bytes, skipping one: "123" -> "321". We change | |
| 5315 * "fword" here, it's changed back afterwards at STATE_UNSWAP3. */ | |
| 5316 p = fword + sp->ts_fidx; | |
| 5317 if (has_mbyte) | |
| 5318 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5319 n = MB_CPTR2LEN(p); |
| 625 | 5320 c = mb_ptr2char(p); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5321 fl = MB_CPTR2LEN(p + n); |
| 625 | 5322 c2 = mb_ptr2char(p + n); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5323 if (!soundfold && !spell_iswordp(p + n + fl, curwin)) |
| 625 | 5324 c3 = c; /* don't swap non-word char */ |
| 5325 else | |
| 5326 c3 = mb_ptr2char(p + n + fl); | |
| 5327 } | |
| 5328 else | |
| 5329 { | |
| 5330 c = *p; | |
| 5331 c2 = p[1]; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5332 if (!soundfold && !spell_iswordp(p + 2, curwin)) |
| 625 | 5333 c3 = c; /* don't swap non-word char */ |
| 5334 else | |
| 5335 c3 = p[2]; | |
| 5336 } | |
| 5337 | |
| 5338 /* When characters are identical: "121" then SWAP3 result is | |
| 5339 * identical, ROT3L result is same as SWAP: "211", ROT3L result is | |
| 5340 * same as SWAP on next char: "112". Thus skip all swapping. | |
| 5341 * Also skip when c3 is NUL. | |
| 5342 * Also get here when the third character is not a word character. | |
| 5343 * Second character may any char: "a.b" -> "b.a" */ | |
| 5344 if (c == c3 || c3 == NUL) | |
| 5345 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5346 PROF_STORE(sp->ts_state) |
| 625 | 5347 sp->ts_state = STATE_REP_INI; |
| 5348 break; | |
| 5349 } | |
| 5350 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5351 { | |
| 5352 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5353 #ifdef DEBUG_TRIEWALK | |
| 5354 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c", | |
| 5355 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5356 c, c3); | |
| 5357 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5358 PROF_STORE(sp->ts_state) |
| 625 | 5359 sp->ts_state = STATE_UNSWAP3; |
| 5360 ++depth; | |
| 5361 if (has_mbyte) | |
| 5362 { | |
| 5363 tl = mb_char2len(c3); | |
| 5364 mch_memmove(p, p + n + fl, tl); | |
| 5365 mb_char2bytes(c2, p + tl); | |
| 5366 mb_char2bytes(c, p + fl + tl); | |
| 5367 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl; | |
| 5368 } | |
| 5369 else | |
| 5370 { | |
| 5371 p[0] = p[2]; | |
| 5372 p[2] = c; | |
| 5373 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 5374 } | |
| 5375 } | |
| 5376 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5377 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5378 PROF_STORE(sp->ts_state) |
| 625 | 5379 sp->ts_state = STATE_REP_INI; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5380 } |
| 625 | 5381 break; |
| 5382 | |
| 5383 case STATE_UNSWAP3: | |
| 5384 /* Undo STATE_SWAP3: "321" -> "123" */ | |
| 5385 p = fword + sp->ts_fidx; | |
| 5386 if (has_mbyte) | |
| 5387 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5388 n = MB_PTR2LEN(p); |
| 625 | 5389 c2 = mb_ptr2char(p + n); |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5390 fl = MB_PTR2LEN(p + n); |
| 625 | 5391 c = mb_ptr2char(p + n + fl); |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5392 tl = MB_PTR2LEN(p + n + fl); |
| 625 | 5393 mch_memmove(p + fl + tl, p, n); |
| 5394 mb_char2bytes(c, p); | |
| 5395 mb_char2bytes(c2, p + tl); | |
| 5396 p = p + tl; | |
| 5397 } | |
| 5398 else | |
| 5399 { | |
| 5400 c = *p; | |
| 5401 *p = p[2]; | |
| 5402 p[2] = c; | |
| 5403 ++p; | |
| 5404 } | |
| 5405 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5406 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 5407 { |
| 5408 /* Middle char is not a word char, skip the rotate. First and | |
| 5409 * third char were already checked at swap and swap3. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5410 PROF_STORE(sp->ts_state) |
| 625 | 5411 sp->ts_state = STATE_REP_INI; |
| 5412 break; | |
| 5413 } | |
| 5414 | |
| 5415 /* Rotate three characters left: "123" -> "231". We change | |
| 5416 * "fword" here, it's changed back afterwards at STATE_UNROT3L. */ | |
| 5417 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5418 { | |
| 5419 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5420 #ifdef DEBUG_TRIEWALK | |
| 5421 p = fword + sp->ts_fidx; | |
| 5422 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c", | |
| 5423 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5424 p[0], p[1], p[2]); | |
| 5425 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5426 PROF_STORE(sp->ts_state) |
| 625 | 5427 sp->ts_state = STATE_UNROT3L; |
| 5428 ++depth; | |
| 330 | 5429 p = fword + sp->ts_fidx; |
| 5430 if (has_mbyte) | |
| 5431 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5432 n = MB_CPTR2LEN(p); |
| 330 | 5433 c = mb_ptr2char(p); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5434 fl = MB_CPTR2LEN(p + n); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5435 fl += MB_CPTR2LEN(p + n + fl); |
| 625 | 5436 mch_memmove(p, p + n, fl); |
| 5437 mb_char2bytes(c, p + fl); | |
| 5438 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 5439 } |
| 5440 else | |
| 323 | 5441 { |
| 330 | 5442 c = *p; |
| 625 | 5443 *p = p[1]; |
| 5444 p[1] = p[2]; | |
| 5445 p[2] = c; | |
| 5446 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 5447 } | |
| 5448 } | |
| 5449 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5450 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5451 PROF_STORE(sp->ts_state) |
| 625 | 5452 sp->ts_state = STATE_REP_INI; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5453 } |
| 625 | 5454 break; |
| 5455 | |
| 5456 case STATE_UNROT3L: | |
| 5457 /* Undo ROT3L: "231" -> "123" */ | |
| 5458 p = fword + sp->ts_fidx; | |
| 5459 if (has_mbyte) | |
| 5460 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5461 n = MB_PTR2LEN(p); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5462 n += MB_PTR2LEN(p + n); |
| 625 | 5463 c = mb_ptr2char(p + n); |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5464 tl = MB_PTR2LEN(p + n); |
| 625 | 5465 mch_memmove(p + tl, p, n); |
| 5466 mb_char2bytes(c, p); | |
| 5467 } | |
| 5468 else | |
| 5469 { | |
| 5470 c = p[2]; | |
| 5471 p[2] = p[1]; | |
| 5472 p[1] = *p; | |
| 5473 *p = c; | |
| 5474 } | |
| 5475 | |
| 5476 /* Rotate three bytes right: "123" -> "312". We change "fword" | |
| 5477 * here, it's changed back afterwards at STATE_UNROT3R. */ | |
| 5478 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5479 { | |
| 5480 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5481 #ifdef DEBUG_TRIEWALK | |
| 5482 p = fword + sp->ts_fidx; | |
| 5483 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c", | |
| 5484 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5485 p[0], p[1], p[2]); | |
| 5486 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5487 PROF_STORE(sp->ts_state) |
| 625 | 5488 sp->ts_state = STATE_UNROT3R; |
| 5489 ++depth; | |
| 330 | 5490 p = fword + sp->ts_fidx; |
| 5491 if (has_mbyte) | |
| 5492 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5493 n = MB_CPTR2LEN(p); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5494 n += MB_CPTR2LEN(p + n); |
| 330 | 5495 c = mb_ptr2char(p + n); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5496 tl = MB_CPTR2LEN(p + n); |
| 330 | 5497 mch_memmove(p + tl, p, n); |
| 5498 mb_char2bytes(c, p); | |
| 625 | 5499 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl; |
| 330 | 5500 } |
| 5501 else | |
| 5502 { | |
| 5503 c = p[2]; | |
| 5504 p[2] = p[1]; | |
| 5505 p[1] = *p; | |
| 5506 *p = c; | |
| 625 | 5507 stack[depth].ts_fidxtry = sp->ts_fidx + 3; |
| 5508 } | |
| 5509 } | |
| 5510 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5511 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5512 PROF_STORE(sp->ts_state) |
| 625 | 5513 sp->ts_state = STATE_REP_INI; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5514 } |
| 625 | 5515 break; |
| 5516 | |
| 5517 case STATE_UNROT3R: | |
| 5518 /* Undo ROT3R: "312" -> "123" */ | |
| 5519 p = fword + sp->ts_fidx; | |
| 5520 if (has_mbyte) | |
| 5521 { | |
| 5522 c = mb_ptr2char(p); | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5523 tl = MB_PTR2LEN(p); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5524 n = MB_PTR2LEN(p + tl); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5525 n += MB_PTR2LEN(p + tl + n); |
| 625 | 5526 mch_memmove(p, p + tl, n); |
| 5527 mb_char2bytes(c, p + n); | |
| 5528 } | |
| 5529 else | |
| 5530 { | |
| 5531 c = *p; | |
| 5532 *p = p[1]; | |
| 5533 p[1] = p[2]; | |
| 5534 p[2] = c; | |
| 5535 } | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5536 /* FALLTHROUGH */ |
| 625 | 5537 |
| 5538 case STATE_REP_INI: | |
| 5539 /* Check if matching with REP items from the .aff file would work. | |
| 5540 * Quickly skip if: | |
| 5541 * - there are no REP items and we are not in the soundfold trie | |
| 5542 * - the score is going to be too high anyway | |
| 5543 * - already applied a REP item or swapped here */ | |
| 5544 if ((lp->lp_replang == NULL && !soundfold) | |
| 5545 || sp->ts_score + SCORE_REP >= su->su_maxscore | |
| 5546 || sp->ts_fidx < sp->ts_fidxtry) | |
| 5547 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5548 PROF_STORE(sp->ts_state) |
| 625 | 5549 sp->ts_state = STATE_FINAL; |
| 323 | 5550 break; |
| 625 | 5551 } |
| 5552 | |
| 5553 /* Use the first byte to quickly find the first entry that may | |
| 5554 * match. If the index is -1 there is none. */ | |
| 5555 if (soundfold) | |
| 5556 sp->ts_curi = slang->sl_repsal_first[fword[sp->ts_fidx]]; | |
| 5557 else | |
| 5558 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]]; | |
| 5559 | |
| 5560 if (sp->ts_curi < 0) | |
| 5561 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5562 PROF_STORE(sp->ts_state) |
| 625 | 5563 sp->ts_state = STATE_FINAL; |
| 5564 break; | |
| 5565 } | |
| 5566 | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5567 PROF_STORE(sp->ts_state) |
| 625 | 5568 sp->ts_state = STATE_REP; |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5569 /* FALLTHROUGH */ |
| 625 | 5570 |
| 5571 case STATE_REP: | |
| 5572 /* Try matching with REP items from the .aff file. For each match | |
| 5573 * replace the characters and check if the resulting word is | |
| 5574 * valid. */ | |
| 5575 p = fword + sp->ts_fidx; | |
| 5576 | |
| 5577 if (soundfold) | |
| 5578 gap = &slang->sl_repsal; | |
| 5579 else | |
| 503 | 5580 gap = &lp->lp_replang->sl_rep; |
| 625 | 5581 while (sp->ts_curi < gap->ga_len) |
| 5582 { | |
| 5583 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++; | |
| 5584 if (*ftp->ft_from != *p) | |
| 5585 { | |
| 5586 /* past possible matching entries */ | |
| 5587 sp->ts_curi = gap->ga_len; | |
| 323 | 5588 break; |
| 5589 } | |
| 625 | 5590 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0 |
| 5591 && TRY_DEEPER(su, stack, depth, SCORE_REP)) | |
| 5592 { | |
| 5593 go_deeper(stack, depth, SCORE_REP); | |
| 5594 #ifdef DEBUG_TRIEWALK | |
| 5595 sprintf(changename[depth], "%.*s-%s: replace %s with %s", | |
| 5596 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5597 ftp->ft_from, ftp->ft_to); | |
| 5598 #endif | |
| 5599 /* Need to undo this afterwards. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5600 PROF_STORE(sp->ts_state) |
| 625 | 5601 sp->ts_state = STATE_REP_UNDO; |
| 5602 | |
| 5603 /* Change the "from" to the "to" string. */ | |
| 5604 ++depth; | |
| 835 | 5605 fl = (int)STRLEN(ftp->ft_from); |
| 5606 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 5607 if (fl != tl) |
| 323 | 5608 { |
| 1619 | 5609 STRMOVE(p + tl, p + fl); |
| 625 | 5610 repextra += tl - fl; |
| 323 | 5611 } |
| 625 | 5612 mch_memmove(p, ftp->ft_to, tl); |
| 5613 stack[depth].ts_fidxtry = sp->ts_fidx + tl; | |
| 5614 stack[depth].ts_tcharlen = 0; | |
| 5615 break; | |
| 5616 } | |
| 5617 } | |
| 5618 | |
| 5619 if (sp->ts_curi >= gap->ga_len && sp->ts_state == STATE_REP) | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5620 { |
| 625 | 5621 /* No (more) matches. */ |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5622 PROF_STORE(sp->ts_state) |
| 625 | 5623 sp->ts_state = STATE_FINAL; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5624 } |
| 625 | 5625 |
| 5626 break; | |
| 5627 | |
| 5628 case STATE_REP_UNDO: | |
| 5629 /* Undo a REP replacement and continue with the next one. */ | |
| 5630 if (soundfold) | |
| 5631 gap = &slang->sl_repsal; | |
| 5632 else | |
| 5633 gap = &lp->lp_replang->sl_rep; | |
| 5634 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1; | |
| 835 | 5635 fl = (int)STRLEN(ftp->ft_from); |
| 5636 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 5637 p = fword + sp->ts_fidx; |
| 5638 if (fl != tl) | |
| 5639 { | |
| 1619 | 5640 STRMOVE(p + fl, p + tl); |
| 625 | 5641 repextra -= tl - fl; |
| 5642 } | |
| 5643 mch_memmove(p, ftp->ft_from, fl); | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5644 PROF_STORE(sp->ts_state) |
| 625 | 5645 sp->ts_state = STATE_REP; |
| 5646 break; | |
| 5647 | |
| 5648 default: | |
| 5649 /* Did all possible states at this level, go up one level. */ | |
| 5650 --depth; | |
| 5651 | |
| 5652 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE) | |
| 5653 { | |
| 5654 /* Continue in or go back to the prefix tree. */ | |
| 5655 byts = pbyts; | |
| 5656 idxs = pidxs; | |
| 5657 } | |
| 5658 | |
| 5659 /* Don't check for CTRL-C too often, it takes time. */ | |
| 5660 if (--breakcheckcount == 0) | |
| 5661 { | |
| 5662 ui_breakcheck(); | |
| 5663 breakcheckcount = 1000; | |
| 5664 } | |
| 5665 } | |
| 5666 } | |
| 5667 } | |
| 5668 | |
| 5669 | |
| 5670 /* | |
| 5671 * Go one level deeper in the tree. | |
| 5672 */ | |
| 5673 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5674 go_deeper(trystate_T *stack, int depth, int score_add) |
| 323 | 5675 { |
| 330 | 5676 stack[depth + 1] = stack[depth]; |
| 323 | 5677 stack[depth + 1].ts_state = STATE_START; |
| 625 | 5678 stack[depth + 1].ts_score = stack[depth].ts_score + score_add; |
| 323 | 5679 stack[depth + 1].ts_curi = 1; /* start just after length byte */ |
| 485 | 5680 stack[depth + 1].ts_flags = 0; |
| 323 | 5681 } |
| 5682 | |
| 455 | 5683 /* |
| 5684 * Case-folding may change the number of bytes: Count nr of chars in | |
| 5685 * fword[flen] and return the byte length of that many chars in "word". | |
| 5686 */ | |
| 5687 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5688 nofold_len(char_u *fword, int flen, char_u *word) |
| 455 | 5689 { |
| 5690 char_u *p; | |
| 5691 int i = 0; | |
| 5692 | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5693 for (p = fword; p < fword + flen; MB_PTR_ADV(p)) |
| 455 | 5694 ++i; |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5695 for (p = word; i > 0; MB_PTR_ADV(p)) |
| 455 | 5696 --i; |
| 5697 return (int)(p - word); | |
| 5698 } | |
| 5699 | |
| 323 | 5700 /* |
| 5701 * "fword" is a good word with case folded. Find the matching keep-case | |
| 5702 * words and put it in "kword". | |
| 5703 * Theoretically there could be several keep-case words that result in the | |
| 5704 * same case-folded word, but we only find one... | |
| 5705 */ | |
| 5706 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5707 find_keepcap_word(slang_T *slang, char_u *fword, char_u *kword) |
| 323 | 5708 { |
| 5709 char_u uword[MAXWLEN]; /* "fword" in upper-case */ | |
| 5710 int depth; | |
| 324 | 5711 idx_T tryidx; |
| 323 | 5712 |
| 5713 /* The following arrays are used at each depth in the tree. */ | |
| 324 | 5714 idx_T arridx[MAXWLEN]; |
| 323 | 5715 int round[MAXWLEN]; |
| 5716 int fwordidx[MAXWLEN]; | |
| 5717 int uwordidx[MAXWLEN]; | |
| 5718 int kwordlen[MAXWLEN]; | |
| 5719 | |
| 5720 int flen, ulen; | |
| 5721 int l; | |
| 5722 int len; | |
| 5723 int c; | |
| 324 | 5724 idx_T lo, hi, m; |
| 323 | 5725 char_u *p; |
| 5726 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */ | |
| 324 | 5727 idx_T *idxs = slang->sl_kidxs; /* array with indexes */ |
| 323 | 5728 |
| 5729 if (byts == NULL) | |
| 5730 { | |
| 5731 /* array is empty: "cannot happen" */ | |
| 5732 *kword = NUL; | |
| 5733 return; | |
| 5734 } | |
| 5735 | |
| 5736 /* Make an all-cap version of "fword". */ | |
| 5737 allcap_copy(fword, uword); | |
| 5738 | |
| 5739 /* | |
| 5740 * Each character needs to be tried both case-folded and upper-case. | |
| 5741 * All this gets very complicated if we keep in mind that changing case | |
| 5742 * may change the byte length of a multi-byte character... | |
| 5743 */ | |
| 5744 depth = 0; | |
| 5745 arridx[0] = 0; | |
| 5746 round[0] = 0; | |
| 5747 fwordidx[0] = 0; | |
| 5748 uwordidx[0] = 0; | |
| 5749 kwordlen[0] = 0; | |
| 5750 while (depth >= 0) | |
| 5751 { | |
| 5752 if (fword[fwordidx[depth]] == NUL) | |
| 5753 { | |
| 5754 /* We are at the end of "fword". If the tree allows a word to end | |
| 5755 * here we have found a match. */ | |
| 5756 if (byts[arridx[depth] + 1] == 0) | |
| 5757 { | |
| 5758 kword[kwordlen[depth]] = NUL; | |
| 5759 return; | |
| 5760 } | |
| 5761 | |
| 5762 /* kword is getting too long, continue one level up */ | |
| 5763 --depth; | |
| 5764 } | |
| 5765 else if (++round[depth] > 2) | |
| 5766 { | |
| 5767 /* tried both fold-case and upper-case character, continue one | |
| 5768 * level up */ | |
| 5769 --depth; | |
| 5770 } | |
| 5771 else | |
| 5772 { | |
| 5773 /* | |
| 5774 * round[depth] == 1: Try using the folded-case character. | |
| 5775 * round[depth] == 2: Try using the upper-case character. | |
| 5776 */ | |
| 5777 if (has_mbyte) | |
| 5778 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5779 flen = MB_CPTR2LEN(fword + fwordidx[depth]); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5780 ulen = MB_CPTR2LEN(uword + uwordidx[depth]); |
| 323 | 5781 } |
| 5782 else | |
| 5783 ulen = flen = 1; | |
| 5784 if (round[depth] == 1) | |
| 5785 { | |
| 5786 p = fword + fwordidx[depth]; | |
| 5787 l = flen; | |
| 5788 } | |
| 5789 else | |
| 5790 { | |
| 5791 p = uword + uwordidx[depth]; | |
| 5792 l = ulen; | |
| 5793 } | |
| 5794 | |
| 5795 for (tryidx = arridx[depth]; l > 0; --l) | |
| 5796 { | |
| 5797 /* Perform a binary search in the list of accepted bytes. */ | |
| 5798 len = byts[tryidx++]; | |
| 5799 c = *p++; | |
| 5800 lo = tryidx; | |
| 5801 hi = tryidx + len - 1; | |
| 5802 while (lo < hi) | |
| 5803 { | |
| 5804 m = (lo + hi) / 2; | |
| 5805 if (byts[m] > c) | |
| 5806 hi = m - 1; | |
| 5807 else if (byts[m] < c) | |
| 5808 lo = m + 1; | |
| 5809 else | |
| 5810 { | |
| 5811 lo = hi = m; | |
| 5812 break; | |
| 5813 } | |
| 5814 } | |
| 5815 | |
| 5816 /* Stop if there is no matching byte. */ | |
| 5817 if (hi < lo || byts[lo] != c) | |
| 5818 break; | |
| 5819 | |
| 5820 /* Continue at the child (if there is one). */ | |
| 5821 tryidx = idxs[lo]; | |
| 5822 } | |
| 5823 | |
| 5824 if (l == 0) | |
| 5825 { | |
| 5826 /* | |
| 5827 * Found the matching char. Copy it to "kword" and go a | |
| 5828 * level deeper. | |
| 5829 */ | |
| 5830 if (round[depth] == 1) | |
| 5831 { | |
| 5832 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth], | |
| 5833 flen); | |
| 5834 kwordlen[depth + 1] = kwordlen[depth] + flen; | |
| 5835 } | |
| 5836 else | |
| 5837 { | |
| 5838 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth], | |
| 5839 ulen); | |
| 5840 kwordlen[depth + 1] = kwordlen[depth] + ulen; | |
| 5841 } | |
| 5842 fwordidx[depth + 1] = fwordidx[depth] + flen; | |
| 5843 uwordidx[depth + 1] = uwordidx[depth] + ulen; | |
| 5844 | |
| 5845 ++depth; | |
| 5846 arridx[depth] = tryidx; | |
| 5847 round[depth] = 0; | |
| 5848 } | |
| 5849 } | |
| 5850 } | |
| 5851 | |
| 5852 /* Didn't find it: "cannot happen". */ | |
| 5853 *kword = NUL; | |
| 5854 } | |
| 5855 | |
| 5856 /* | |
| 344 | 5857 * Compute the sound-a-like score for suggestions in su->su_ga and add them to |
| 5858 * su->su_sga. | |
| 5859 */ | |
| 5860 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5861 score_comp_sal(suginfo_T *su) |
| 344 | 5862 { |
| 5863 langp_T *lp; | |
| 5864 char_u badsound[MAXWLEN]; | |
| 5865 int i; | |
| 5866 suggest_T *stp; | |
| 5867 suggest_T *sstp; | |
| 5868 int score; | |
| 500 | 5869 int lpi; |
| 344 | 5870 |
| 5871 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL) | |
| 5872 return; | |
| 5873 | |
| 5874 /* 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
|
5875 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
|
5876 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5877 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 5878 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 5879 { | |
| 5880 /* soundfold the bad word */ | |
| 375 | 5881 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound); |
| 344 | 5882 |
| 5883 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 5884 { | |
| 5885 stp = &SUG(su->su_ga, i); | |
| 5886 | |
| 351 | 5887 /* Case-fold the suggested word, sound-fold it and compute the |
| 5888 * sound-a-like score. */ | |
| 5889 score = stp_sal_score(stp, su, lp->lp_slang, badsound); | |
| 344 | 5890 if (score < SCORE_MAXMAX) |
| 5891 { | |
| 5892 /* Add the suggestion. */ | |
| 5893 sstp = &SUG(su->su_sga, su->su_sga.ga_len); | |
| 5894 sstp->st_word = vim_strsave(stp->st_word); | |
| 5895 if (sstp->st_word != NULL) | |
| 5896 { | |
| 625 | 5897 sstp->st_wordlen = stp->st_wordlen; |
| 344 | 5898 sstp->st_score = score; |
| 5899 sstp->st_altscore = 0; | |
| 5900 sstp->st_orglen = stp->st_orglen; | |
| 5901 ++su->su_sga.ga_len; | |
| 5902 } | |
| 5903 } | |
| 5904 } | |
| 5905 break; | |
| 5906 } | |
| 500 | 5907 } |
| 344 | 5908 } |
| 5909 | |
| 5910 /* | |
| 5911 * Combine the list of suggestions in su->su_ga and su->su_sga. | |
| 4352 | 5912 * They are entwined. |
| 344 | 5913 */ |
| 5914 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5915 score_combine(suginfo_T *su) |
| 344 | 5916 { |
| 5917 int i; | |
| 5918 int j; | |
| 5919 garray_T ga; | |
| 5920 garray_T *gap; | |
| 5921 langp_T *lp; | |
| 5922 suggest_T *stp; | |
| 5923 char_u *p; | |
| 5924 char_u badsound[MAXWLEN]; | |
| 5925 int round; | |
| 500 | 5926 int lpi; |
| 625 | 5927 slang_T *slang = NULL; |
| 344 | 5928 |
| 5929 /* 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
|
5930 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
|
5931 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5932 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 5933 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 5934 { | |
| 5935 /* soundfold the bad word */ | |
| 625 | 5936 slang = lp->lp_slang; |
| 5937 spell_soundfold(slang, su->su_fbadword, TRUE, badsound); | |
| 344 | 5938 |
| 5939 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 5940 { | |
| 5941 stp = &SUG(su->su_ga, i); | |
| 625 | 5942 stp->st_altscore = stp_sal_score(stp, su, slang, badsound); |
| 344 | 5943 if (stp->st_altscore == SCORE_MAXMAX) |
| 5944 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4; | |
| 5945 else | |
| 5946 stp->st_score = (stp->st_score * 3 | |
| 5947 + stp->st_altscore) / 4; | |
| 5948 stp->st_salscore = FALSE; | |
| 5949 } | |
| 5950 break; | |
| 5951 } | |
| 5952 } | |
| 5953 | |
| 844 | 5954 if (slang == NULL) /* Using "double" without sound folding. */ |
| 5955 { | |
| 5956 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, | |
| 5957 su->su_maxcount); | |
| 625 | 5958 return; |
| 844 | 5959 } |
| 625 | 5960 |
| 344 | 5961 /* Add the alternate score to su_sga. */ |
| 5962 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 5963 { | |
| 5964 stp = &SUG(su->su_sga, i); | |
| 625 | 5965 stp->st_altscore = spell_edit_score(slang, |
| 5966 su->su_badword, stp->st_word); | |
| 344 | 5967 if (stp->st_score == SCORE_MAXMAX) |
| 5968 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8; | |
| 5969 else | |
| 5970 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8; | |
| 5971 stp->st_salscore = TRUE; | |
| 5972 } | |
| 5973 | |
| 625 | 5974 /* Remove bad suggestions, sort the suggestions and truncate at "maxcount" |
| 5975 * for both lists. */ | |
| 5976 check_suggestions(su, &su->su_ga); | |
| 344 | 5977 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 625 | 5978 check_suggestions(su, &su->su_sga); |
| 344 | 5979 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount); |
| 5980 | |
| 5981 ga_init2(&ga, (int)sizeof(suginfo_T), 1); | |
| 5982 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL) | |
| 5983 return; | |
| 5984 | |
| 5985 stp = &SUG(ga, 0); | |
| 5986 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i) | |
| 5987 { | |
| 5988 /* round 1: get a suggestion from su_ga | |
| 5989 * round 2: get a suggestion from su_sga */ | |
| 5990 for (round = 1; round <= 2; ++round) | |
| 5991 { | |
| 5992 gap = round == 1 ? &su->su_ga : &su->su_sga; | |
| 5993 if (i < gap->ga_len) | |
| 5994 { | |
| 5995 /* Don't add a word if it's already there. */ | |
| 5996 p = SUG(*gap, i).st_word; | |
| 5997 for (j = 0; j < ga.ga_len; ++j) | |
| 5998 if (STRCMP(stp[j].st_word, p) == 0) | |
| 5999 break; | |
| 6000 if (j == ga.ga_len) | |
| 6001 stp[ga.ga_len++] = SUG(*gap, i); | |
| 6002 else | |
| 6003 vim_free(p); | |
| 6004 } | |
| 6005 } | |
| 6006 } | |
| 6007 | |
| 6008 ga_clear(&su->su_ga); | |
| 6009 ga_clear(&su->su_sga); | |
| 6010 | |
| 6011 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 6012 if (ga.ga_len > su->su_maxcount) | |
| 6013 { | |
| 6014 for (i = su->su_maxcount; i < ga.ga_len; ++i) | |
| 6015 vim_free(stp[i].st_word); | |
| 6016 ga.ga_len = su->su_maxcount; | |
| 6017 } | |
| 6018 | |
| 6019 su->su_ga = ga; | |
| 6020 } | |
| 6021 | |
| 6022 /* | |
| 351 | 6023 * For the goodword in "stp" compute the soundalike score compared to the |
| 6024 * badword. | |
| 6025 */ | |
| 6026 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6027 stp_sal_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6028 suggest_T *stp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6029 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6030 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6031 char_u *badsound) /* sound-folded badword */ |
| 351 | 6032 { |
| 6033 char_u *p; | |
| 534 | 6034 char_u *pbad; |
| 6035 char_u *pgood; | |
| 351 | 6036 char_u badsound2[MAXWLEN]; |
| 6037 char_u fword[MAXWLEN]; | |
| 6038 char_u goodsound[MAXWLEN]; | |
| 534 | 6039 char_u goodword[MAXWLEN]; |
| 6040 int lendiff; | |
| 6041 | |
| 6042 lendiff = (int)(su->su_badlen - stp->st_orglen); | |
| 6043 if (lendiff >= 0) | |
| 6044 pbad = badsound; | |
| 351 | 6045 else |
| 6046 { | |
| 6047 /* soundfold the bad word with more characters following */ | |
| 6048 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN); | |
| 6049 | |
| 6050 /* When joining two words the sound often changes a lot. E.g., "t he" | |
| 6051 * sounds like "t h" while "the" sounds like "@". Avoid that by | |
| 6052 * removing the space. Don't do it when the good word also contains a | |
| 6053 * space. */ | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
6054 if (VIM_ISWHITE(su->su_badptr[su->su_badlen]) |
| 351 | 6055 && *skiptowhite(stp->st_word) == NUL) |
| 6056 for (p = fword; *(p = skiptowhite(p)) != NUL; ) | |
| 1619 | 6057 STRMOVE(p, p + 1); |
| 351 | 6058 |
| 375 | 6059 spell_soundfold(slang, fword, TRUE, badsound2); |
| 534 | 6060 pbad = badsound2; |
| 6061 } | |
| 6062 | |
| 2768 | 6063 if (lendiff > 0 && stp->st_wordlen + lendiff < MAXWLEN) |
| 534 | 6064 { |
| 6065 /* Add part of the bad word to the good word, so that we soundfold | |
| 6066 * what replaces the bad word. */ | |
| 6067 STRCPY(goodword, stp->st_word); | |
| 625 | 6068 vim_strncpy(goodword + stp->st_wordlen, |
| 6069 su->su_badptr + su->su_badlen - lendiff, lendiff); | |
| 534 | 6070 pgood = goodword; |
| 6071 } | |
| 6072 else | |
| 6073 pgood = stp->st_word; | |
| 351 | 6074 |
| 375 | 6075 /* Sound-fold the word and compute the score for the difference. */ |
| 534 | 6076 spell_soundfold(slang, pgood, FALSE, goodsound); |
| 6077 | |
| 6078 return soundalike_score(goodsound, pbad); | |
| 351 | 6079 } |
| 6080 | |
| 625 | 6081 /* structure used to store soundfolded words that add_sound_suggest() has |
| 6082 * handled already. */ | |
| 6083 typedef struct | |
| 6084 { | |
| 6085 short sft_score; /* lowest score used */ | |
| 6086 char_u sft_word[1]; /* soundfolded word, actually longer */ | |
| 6087 } sftword_T; | |
| 6088 | |
| 6089 static sftword_T dumsft; | |
| 6090 #define HIKEY2SFT(p) ((sftword_T *)(p - (dumsft.sft_word - (char_u *)&dumsft))) | |
| 6091 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key) | |
| 6092 | |
| 6093 /* | |
| 6094 * Prepare for calling suggest_try_soundalike(). | |
| 6095 */ | |
| 6096 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6097 suggest_try_soundalike_prep(void) |
| 625 | 6098 { |
| 6099 langp_T *lp; | |
| 6100 int lpi; | |
| 6101 slang_T *slang; | |
| 6102 | |
| 6103 /* Do this for all languages that support sound folding and for which a | |
| 6104 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6105 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
|
6106 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6107 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 6108 slang = lp->lp_slang; |
| 6109 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 6110 /* prepare the hashtable used by add_sound_suggest() */ | |
| 6111 hash_init(&slang->sl_sounddone); | |
| 6112 } | |
| 6113 } | |
| 6114 | |
| 351 | 6115 /* |
| 323 | 6116 * Find suggestions by comparing the word in a sound-a-like form. |
| 503 | 6117 * Note: This doesn't support postponed prefixes. |
| 323 | 6118 */ |
| 6119 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6120 suggest_try_soundalike(suginfo_T *su) |
| 323 | 6121 { |
| 6122 char_u salword[MAXWLEN]; | |
| 6123 langp_T *lp; | |
| 500 | 6124 int lpi; |
| 503 | 6125 slang_T *slang; |
| 344 | 6126 |
| 625 | 6127 /* Do this for all languages that support sound folding and for which a |
| 6128 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6129 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
|
6130 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6131 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 6132 slang = lp->lp_slang; |
| 6133 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 6134 { | |
| 6135 /* soundfold the bad word */ | |
| 6136 spell_soundfold(slang, su->su_fbadword, TRUE, salword); | |
| 6137 | |
| 6138 /* try all kinds of inserts/deletes/swaps/etc. */ | |
| 6139 /* TODO: also soundfold the next words, so that we can try joining | |
| 6140 * and splitting */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6141 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6142 prof_init(); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6143 #endif |
| 625 | 6144 suggest_trie_walk(su, lp, salword, TRUE); |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6145 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6146 prof_report("soundalike"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6147 #endif |
| 625 | 6148 } |
| 6149 } | |
| 6150 } | |
| 6151 | |
| 6152 /* | |
| 6153 * Finish up after calling suggest_try_soundalike(). | |
| 6154 */ | |
| 6155 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6156 suggest_try_soundalike_finish(void) |
| 625 | 6157 { |
| 6158 langp_T *lp; | |
| 6159 int lpi; | |
| 6160 slang_T *slang; | |
| 6161 int todo; | |
| 6162 hashitem_T *hi; | |
| 6163 | |
| 6164 /* Do this for all languages that support sound folding and for which a | |
| 6165 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6166 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
|
6167 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6168 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 503 | 6169 slang = lp->lp_slang; |
| 625 | 6170 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) |
| 6171 { | |
| 6172 /* Free the info about handled words. */ | |
| 835 | 6173 todo = (int)slang->sl_sounddone.ht_used; |
| 625 | 6174 for (hi = slang->sl_sounddone.ht_array; todo > 0; ++hi) |
| 6175 if (!HASHITEM_EMPTY(hi)) | |
| 6176 { | |
| 6177 vim_free(HI2SFT(hi)); | |
| 6178 --todo; | |
| 6179 } | |
| 1087 | 6180 |
| 6181 /* Clear the hashtable, it may also be used by another region. */ | |
| 625 | 6182 hash_clear(&slang->sl_sounddone); |
| 1087 | 6183 hash_init(&slang->sl_sounddone); |
| 625 | 6184 } |
| 6185 } | |
| 6186 } | |
| 6187 | |
| 6188 /* | |
| 6189 * A match with a soundfolded word is found. Add the good word(s) that | |
| 6190 * produce this soundfolded word. | |
| 6191 */ | |
| 6192 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6193 add_sound_suggest( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6194 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6195 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6196 int score, /* soundfold score */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6197 langp_T *lp) |
| 625 | 6198 { |
| 6199 slang_T *slang = lp->lp_slang; /* language for sound folding */ | |
| 6200 int sfwordnr; | |
| 6201 char_u *nrline; | |
| 6202 int orgnr; | |
| 6203 char_u theword[MAXWLEN]; | |
| 6204 int i; | |
| 6205 int wlen; | |
| 6206 char_u *byts; | |
| 6207 idx_T *idxs; | |
| 6208 int n; | |
| 6209 int wordcount; | |
| 6210 int wc; | |
| 6211 int goodscore; | |
| 6212 hash_T hash; | |
| 6213 hashitem_T *hi; | |
| 6214 sftword_T *sft; | |
| 6215 int bc, gc; | |
| 6216 int limit; | |
| 6217 | |
| 6218 /* | |
| 6219 * It's very well possible that the same soundfold word is found several | |
| 6220 * times with different scores. Since the following is quite slow only do | |
| 6221 * the words that have a better score than before. Use a hashtable to | |
| 6222 * remember the words that have been done. | |
| 6223 */ | |
| 6224 hash = hash_hash(goodword); | |
| 6225 hi = hash_lookup(&slang->sl_sounddone, goodword, hash); | |
| 6226 if (HASHITEM_EMPTY(hi)) | |
| 6227 { | |
|
16825
ce04ebdf26b8
patch 8.1.1414: alloc() returning "char_u *" causes a lot of type casts
Bram Moolenaar <Bram@vim.org>
parents:
16768
diff
changeset
|
6228 sft = alloc(sizeof(sftword_T) + STRLEN(goodword)); |
| 625 | 6229 if (sft != NULL) |
| 6230 { | |
| 6231 sft->sft_score = score; | |
| 6232 STRCPY(sft->sft_word, goodword); | |
| 6233 hash_add_item(&slang->sl_sounddone, hi, sft->sft_word, hash); | |
| 6234 } | |
| 6235 } | |
| 6236 else | |
| 6237 { | |
| 6238 sft = HI2SFT(hi); | |
| 6239 if (score >= sft->sft_score) | |
| 6240 return; | |
| 6241 sft->sft_score = score; | |
| 6242 } | |
| 6243 | |
| 6244 /* | |
| 6245 * Find the word nr in the soundfold tree. | |
| 6246 */ | |
| 6247 sfwordnr = soundfold_find(slang, goodword); | |
| 6248 if (sfwordnr < 0) | |
| 6249 { | |
|
10359
66f1b5bf3fa6
commit https://github.com/vim/vim/commit/95f096030ed1a8afea028f2ea295d6f6a70f466f
Christian Brabandt <cb@256bit.org>
parents:
10277
diff
changeset
|
6250 internal_error("add_sound_suggest()"); |
| 625 | 6251 return; |
| 6252 } | |
| 6253 | |
| 6254 /* | |
| 6255 * go over the list of good words that produce this soundfold word | |
| 6256 */ | |
| 6257 nrline = ml_get_buf(slang->sl_sugbuf, (linenr_T)(sfwordnr + 1), FALSE); | |
| 6258 orgnr = 0; | |
| 6259 while (*nrline != NUL) | |
| 6260 { | |
| 6261 /* The wordnr was stored in a minimal nr of bytes as an offset to the | |
| 6262 * previous wordnr. */ | |
| 6263 orgnr += bytes2offset(&nrline); | |
| 6264 | |
| 6265 byts = slang->sl_fbyts; | |
| 6266 idxs = slang->sl_fidxs; | |
| 6267 | |
| 6268 /* Lookup the word "orgnr" one of the two tries. */ | |
| 6269 n = 0; | |
| 6270 wordcount = 0; | |
| 5506 | 6271 for (wlen = 0; wlen < MAXWLEN - 3; ++wlen) |
| 625 | 6272 { |
| 6273 i = 1; | |
| 6274 if (wordcount == orgnr && byts[n + 1] == NUL) | |
| 6275 break; /* found end of word */ | |
| 6276 | |
| 6277 if (byts[n + 1] == NUL) | |
| 6278 ++wordcount; | |
| 6279 | |
| 6280 /* skip over the NUL bytes */ | |
| 6281 for ( ; byts[n + i] == NUL; ++i) | |
| 6282 if (i > byts[n]) /* safety check */ | |
| 6283 { | |
| 6284 STRCPY(theword + wlen, "BAD"); | |
| 5506 | 6285 wlen += 3; |
| 625 | 6286 goto badword; |
| 6287 } | |
| 6288 | |
| 6289 /* One of the siblings must have the word. */ | |
| 6290 for ( ; i < byts[n]; ++i) | |
| 6291 { | |
| 6292 wc = idxs[idxs[n + i]]; /* nr of words under this byte */ | |
| 6293 if (wordcount + wc > orgnr) | |
| 6294 break; | |
| 6295 wordcount += wc; | |
| 6296 } | |
| 6297 | |
| 5506 | 6298 theword[wlen] = byts[n + i]; |
| 625 | 6299 n = idxs[n + i]; |
| 6300 } | |
| 6301 badword: | |
| 6302 theword[wlen] = NUL; | |
| 6303 | |
| 6304 /* Go over the possible flags and regions. */ | |
| 6305 for (; i <= byts[n] && byts[n + i] == NUL; ++i) | |
| 6306 { | |
| 6307 char_u cword[MAXWLEN]; | |
| 6308 char_u *p; | |
| 6309 int flags = (int)idxs[n + i]; | |
| 6310 | |
| 698 | 6311 /* Skip words with the NOSUGGEST flag */ |
| 6312 if (flags & WF_NOSUGGEST) | |
| 6313 continue; | |
| 6314 | |
| 625 | 6315 if (flags & WF_KEEPCAP) |
| 6316 { | |
| 6317 /* Must find the word in the keep-case tree. */ | |
| 6318 find_keepcap_word(slang, theword, cword); | |
| 6319 p = cword; | |
| 6320 } | |
| 6321 else | |
| 6322 { | |
| 6323 flags |= su->su_badflags; | |
| 6324 if ((flags & WF_CAPMASK) != 0) | |
| 6325 { | |
| 6326 /* Need to fix case according to "flags". */ | |
| 6327 make_case_word(theword, cword, flags); | |
| 6328 p = cword; | |
| 323 | 6329 } |
| 6330 else | |
| 625 | 6331 p = theword; |
| 6332 } | |
| 6333 | |
| 6334 /* Add the suggestion. */ | |
| 6335 if (sps_flags & SPS_DOUBLE) | |
| 6336 { | |
| 6337 /* Add the suggestion if the score isn't too bad. */ | |
| 6338 if (score <= su->su_maxscore) | |
| 6339 add_suggestion(su, &su->su_sga, p, su->su_badlen, | |
| 6340 score, 0, FALSE, slang, FALSE); | |
| 6341 } | |
| 6342 else | |
| 6343 { | |
| 6344 /* Add a penalty for words in another region. */ | |
| 6345 if ((flags & WF_REGION) | |
| 6346 && (((unsigned)flags >> 16) & lp->lp_region) == 0) | |
| 6347 goodscore = SCORE_REGION; | |
| 6348 else | |
| 6349 goodscore = 0; | |
| 6350 | |
| 6351 /* Add a small penalty for changing the first letter from | |
| 6352 * lower to upper case. Helps for "tath" -> "Kath", which is | |
| 4352 | 6353 * less common than "tath" -> "path". Don't do it when the |
| 625 | 6354 * letter is the same, that has already been counted. */ |
| 6355 gc = PTR2CHAR(p); | |
| 6356 if (SPELL_ISUPPER(gc)) | |
| 6357 { | |
| 6358 bc = PTR2CHAR(su->su_badword); | |
| 6359 if (!SPELL_ISUPPER(bc) | |
| 6360 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc)) | |
| 6361 goodscore += SCORE_ICASE / 2; | |
| 6362 } | |
| 6363 | |
| 6364 /* Compute the score for the good word. This only does letter | |
| 6365 * insert/delete/swap/replace. REP items are not considered, | |
| 6366 * which may make the score a bit higher. | |
| 6367 * Use a limit for the score to make it work faster. Use | |
| 6368 * MAXSCORE(), because RESCORE() will change the score. | |
| 6369 * If the limit is very high then the iterative method is | |
| 6370 * inefficient, using an array is quicker. */ | |
| 6371 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score); | |
| 6372 if (limit > SCORE_LIMITMAX) | |
| 6373 goodscore += spell_edit_score(slang, su->su_badword, p); | |
| 6374 else | |
| 6375 goodscore += spell_edit_score_limit(slang, su->su_badword, | |
| 6376 p, limit); | |
| 6377 | |
| 6378 /* When going over the limit don't bother to do the rest. */ | |
| 6379 if (goodscore < SCORE_MAXMAX) | |
| 6380 { | |
| 6381 /* Give a bonus to words seen before. */ | |
| 6382 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE); | |
| 6383 | |
| 6384 /* Add the suggestion if the score isn't too bad. */ | |
| 6385 goodscore = RESCORE(goodscore, score); | |
| 6386 if (goodscore <= su->su_sfmaxscore) | |
| 6387 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 6388 goodscore, score, TRUE, slang, TRUE); | |
| 6389 } | |
| 6390 } | |
| 6391 } | |
| 6392 /* smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr); */ | |
| 6393 } | |
| 6394 } | |
| 6395 | |
| 6396 /* | |
| 6397 * Find word "word" in fold-case tree for "slang" and return the word number. | |
| 6398 */ | |
| 6399 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6400 soundfold_find(slang_T *slang, char_u *word) |
| 625 | 6401 { |
| 6402 idx_T arridx = 0; | |
| 6403 int len; | |
| 6404 int wlen = 0; | |
| 6405 int c; | |
| 6406 char_u *ptr = word; | |
| 6407 char_u *byts; | |
| 6408 idx_T *idxs; | |
| 6409 int wordnr = 0; | |
| 6410 | |
| 6411 byts = slang->sl_sbyts; | |
| 6412 idxs = slang->sl_sidxs; | |
| 6413 | |
| 6414 for (;;) | |
| 6415 { | |
| 6416 /* First byte is the number of possible bytes. */ | |
| 6417 len = byts[arridx++]; | |
| 6418 | |
| 6419 /* If the first possible byte is a zero the word could end here. | |
| 6420 * If the word ends we found the word. If not skip the NUL bytes. */ | |
| 6421 c = ptr[wlen]; | |
| 6422 if (byts[arridx] == NUL) | |
| 6423 { | |
| 6424 if (c == NUL) | |
| 6425 break; | |
| 6426 | |
| 6427 /* Skip over the zeros, there can be several. */ | |
| 6428 while (len > 0 && byts[arridx] == NUL) | |
| 6429 { | |
| 6430 ++arridx; | |
| 6431 --len; | |
| 6432 } | |
| 6433 if (len == 0) | |
| 6434 return -1; /* no children, word should have ended here */ | |
| 6435 ++wordnr; | |
| 6436 } | |
| 6437 | |
| 6438 /* If the word ends we didn't find it. */ | |
| 6439 if (c == NUL) | |
| 6440 return -1; | |
| 6441 | |
| 6442 /* Perform a binary search in the list of accepted bytes. */ | |
| 6443 if (c == TAB) /* <Tab> is handled like <Space> */ | |
| 6444 c = ' '; | |
| 6445 while (byts[arridx] < c) | |
| 6446 { | |
| 6447 /* The word count is in the first idxs[] entry of the child. */ | |
| 6448 wordnr += idxs[idxs[arridx]]; | |
| 6449 ++arridx; | |
| 6450 if (--len == 0) /* end of the bytes, didn't find it */ | |
| 6451 return -1; | |
| 6452 } | |
| 6453 if (byts[arridx] != c) /* didn't find the byte */ | |
| 6454 return -1; | |
| 6455 | |
| 6456 /* Continue at the child (if there is one). */ | |
| 6457 arridx = idxs[arridx]; | |
| 6458 ++wlen; | |
| 6459 | |
| 6460 /* One space in the good word may stand for several spaces in the | |
| 6461 * checked word. */ | |
| 6462 if (c == ' ') | |
| 6463 while (ptr[wlen] == ' ' || ptr[wlen] == TAB) | |
| 6464 ++wlen; | |
| 6465 } | |
| 6466 | |
| 6467 return wordnr; | |
| 323 | 6468 } |
| 6469 | |
| 6470 /* | |
| 324 | 6471 * Copy "fword" to "cword", fixing case according to "flags". |
| 323 | 6472 */ |
| 6473 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6474 make_case_word(char_u *fword, char_u *cword, int flags) |
| 323 | 6475 { |
| 6476 if (flags & WF_ALLCAP) | |
| 6477 /* Make it all upper-case */ | |
| 6478 allcap_copy(fword, cword); | |
| 6479 else if (flags & WF_ONECAP) | |
| 6480 /* Make the first letter upper-case */ | |
| 324 | 6481 onecap_copy(fword, cword, TRUE); |
| 323 | 6482 else |
| 6483 /* Use goodword as-is. */ | |
| 6484 STRCPY(cword, fword); | |
| 6485 } | |
| 6486 | |
| 330 | 6487 |
| 323 | 6488 /* |
| 6489 * Return TRUE if "c1" and "c2" are similar characters according to the MAP | |
| 6490 * lines in the .aff file. | |
| 6491 */ | |
| 6492 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6493 similar_chars(slang_T *slang, int c1, int c2) |
| 323 | 6494 { |
| 330 | 6495 int m1, m2; |
| 3549 | 6496 char_u buf[MB_MAXBYTES + 1]; |
| 330 | 6497 hashitem_T *hi; |
| 6498 | |
| 6499 if (c1 >= 256) | |
| 6500 { | |
| 6501 buf[mb_char2bytes(c1, buf)] = 0; | |
| 6502 hi = hash_find(&slang->sl_map_hash, buf); | |
| 6503 if (HASHITEM_EMPTY(hi)) | |
| 6504 m1 = 0; | |
| 6505 else | |
| 6506 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 6507 } | |
| 6508 else | |
| 6509 m1 = slang->sl_map_array[c1]; | |
| 6510 if (m1 == 0) | |
| 323 | 6511 return FALSE; |
| 330 | 6512 |
| 6513 | |
| 6514 if (c2 >= 256) | |
| 6515 { | |
| 6516 buf[mb_char2bytes(c2, buf)] = 0; | |
| 6517 hi = hash_find(&slang->sl_map_hash, buf); | |
| 6518 if (HASHITEM_EMPTY(hi)) | |
| 6519 m2 = 0; | |
| 6520 else | |
| 6521 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 6522 } | |
| 6523 else | |
| 6524 m2 = slang->sl_map_array[c2]; | |
| 6525 | |
| 6526 return m1 == m2; | |
| 323 | 6527 } |
| 6528 | |
| 6529 /* | |
| 6530 * Add a suggestion to the list of suggestions. | |
| 625 | 6531 * For a suggestion that is already in the list the lowest score is remembered. |
| 6532 */ | |
| 6533 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6534 add_suggestion( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6535 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6536 garray_T *gap, /* either su_ga or su_sga */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6537 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6538 int badlenarg, /* len of bad word replaced with "goodword" */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6539 int score, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6540 int altscore, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6541 int had_bonus, /* value for st_had_bonus */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6542 slang_T *slang, /* language for sound folding */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6543 int maxsf) /* su_maxscore applies to soundfold score, |
| 625 | 6544 su_sfmaxscore to the total score. */ |
| 6545 { | |
| 6546 int goodlen; /* len of goodword changed */ | |
| 6547 int badlen; /* len of bad word changed */ | |
| 323 | 6548 suggest_T *stp; |
| 534 | 6549 suggest_T new_sug; |
| 323 | 6550 int i; |
| 534 | 6551 char_u *pgood, *pbad; |
| 532 | 6552 |
| 534 | 6553 /* Minimize "badlen" for consistency. Avoids that changing "the the" to |
| 6554 * "thee the" is added next to changing the first "the" the "thee". */ | |
| 6555 pgood = goodword + STRLEN(goodword); | |
| 625 | 6556 pbad = su->su_badptr + badlenarg; |
| 6557 for (;;) | |
| 6558 { | |
| 835 | 6559 goodlen = (int)(pgood - goodword); |
| 6560 badlen = (int)(pbad - su->su_badptr); | |
| 625 | 6561 if (goodlen <= 0 || badlen <= 0) |
| 6562 break; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
6563 MB_PTR_BACK(goodword, pgood); |
|
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
6564 MB_PTR_BACK(su->su_badptr, pbad); |
| 534 | 6565 if (has_mbyte) |
| 6566 { | |
| 6567 if (mb_ptr2char(pgood) != mb_ptr2char(pbad)) | |
| 6568 break; | |
| 346 | 6569 } |
|
15605
62b3805506b3
patch 8.1.0810: too many #ifdefs
Bram Moolenaar <Bram@vim.org>
parents:
15543
diff
changeset
|
6570 else if (*pgood != *pbad) |
| 534 | 6571 break; |
| 625 | 6572 } |
| 6573 | |
| 534 | 6574 if (badlen == 0 && goodlen == 0) |
| 6575 /* goodword doesn't change anything; may happen for "the the" changing | |
| 6576 * the first "the" to itself. */ | |
| 6577 return; | |
| 346 | 6578 |
| 944 | 6579 if (gap->ga_len == 0) |
| 6580 i = -1; | |
| 6581 else | |
| 6582 { | |
| 6583 /* Check if the word is already there. Also check the length that is | |
| 6584 * being replaced "thes," -> "these" is a different suggestion from | |
| 6585 * "thes" -> "these". */ | |
| 6586 stp = &SUG(*gap, 0); | |
| 6587 for (i = gap->ga_len; --i >= 0; ++stp) | |
| 6588 if (stp->st_wordlen == goodlen | |
| 6589 && stp->st_orglen == badlen | |
| 6590 && STRNCMP(stp->st_word, goodword, goodlen) == 0) | |
| 6591 { | |
| 6592 /* | |
| 6593 * Found it. Remember the word with the lowest score. | |
| 6594 */ | |
| 6595 if (stp->st_slang == NULL) | |
| 6596 stp->st_slang = slang; | |
| 6597 | |
| 6598 new_sug.st_score = score; | |
| 6599 new_sug.st_altscore = altscore; | |
| 6600 new_sug.st_had_bonus = had_bonus; | |
| 6601 | |
| 6602 if (stp->st_had_bonus != had_bonus) | |
| 6603 { | |
| 6604 /* Only one of the two had the soundalike score computed. | |
| 6605 * Need to do that for the other one now, otherwise the | |
| 6606 * scores can't be compared. This happens because | |
| 6607 * suggest_try_change() doesn't compute the soundalike | |
| 6608 * word to keep it fast, while some special methods set | |
| 6609 * the soundalike score to zero. */ | |
| 6610 if (had_bonus) | |
| 6611 rescore_one(su, stp); | |
| 6612 else | |
| 6613 { | |
| 6614 new_sug.st_word = stp->st_word; | |
| 6615 new_sug.st_wordlen = stp->st_wordlen; | |
| 6616 new_sug.st_slang = stp->st_slang; | |
| 6617 new_sug.st_orglen = badlen; | |
| 6618 rescore_one(su, &new_sug); | |
| 6619 } | |
| 6620 } | |
| 6621 | |
| 6622 if (stp->st_score > new_sug.st_score) | |
| 6623 { | |
| 6624 stp->st_score = new_sug.st_score; | |
| 6625 stp->st_altscore = new_sug.st_altscore; | |
| 6626 stp->st_had_bonus = new_sug.st_had_bonus; | |
| 6627 } | |
| 6628 break; | |
| 6629 } | |
| 6630 } | |
| 625 | 6631 |
| 6632 if (i < 0 && ga_grow(gap, 1) == OK) | |
| 6633 { | |
| 6634 /* Add a suggestion. */ | |
| 6635 stp = &SUG(*gap, gap->ga_len); | |
| 6636 stp->st_word = vim_strnsave(goodword, goodlen); | |
| 6637 if (stp->st_word != NULL) | |
| 6638 { | |
| 6639 stp->st_wordlen = goodlen; | |
| 6640 stp->st_score = score; | |
| 6641 stp->st_altscore = altscore; | |
| 6642 stp->st_had_bonus = had_bonus; | |
| 6643 stp->st_orglen = badlen; | |
| 6644 stp->st_slang = slang; | |
| 6645 ++gap->ga_len; | |
| 6646 | |
| 6647 /* If we have too many suggestions now, sort the list and keep | |
| 6648 * the best suggestions. */ | |
| 6649 if (gap->ga_len > SUG_MAX_COUNT(su)) | |
| 6650 { | |
| 6651 if (maxsf) | |
| 6652 su->su_sfmaxscore = cleanup_suggestions(gap, | |
| 6653 su->su_sfmaxscore, SUG_CLEAN_COUNT(su)); | |
| 6654 else | |
| 6655 su->su_maxscore = cleanup_suggestions(gap, | |
| 6656 su->su_maxscore, SUG_CLEAN_COUNT(su)); | |
| 6657 } | |
| 6658 } | |
| 6659 } | |
| 6660 } | |
| 6661 | |
| 6662 /* | |
| 6663 * Suggestions may in fact be flagged as errors. Esp. for banned words and | |
| 6664 * for split words, such as "the the". Remove these from the list here. | |
| 6665 */ | |
| 6666 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6667 check_suggestions( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6668 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6669 garray_T *gap) /* either su_ga or su_sga */ |
| 625 | 6670 { |
| 6671 suggest_T *stp; | |
| 6672 int i; | |
| 6673 char_u longword[MAXWLEN + 1]; | |
| 6674 int len; | |
| 6675 hlf_T attr; | |
| 6676 | |
| 6677 stp = &SUG(*gap, 0); | |
| 6678 for (i = gap->ga_len - 1; i >= 0; --i) | |
| 6679 { | |
| 6680 /* Need to append what follows to check for "the the". */ | |
| 2768 | 6681 vim_strncpy(longword, stp[i].st_word, MAXWLEN); |
| 625 | 6682 len = stp[i].st_wordlen; |
| 6683 vim_strncpy(longword + len, su->su_badptr + stp[i].st_orglen, | |
| 6684 MAXWLEN - len); | |
| 6685 attr = HLF_COUNT; | |
| 6686 (void)spell_check(curwin, longword, &attr, NULL, FALSE); | |
| 6687 if (attr != HLF_COUNT) | |
| 6688 { | |
| 6689 /* Remove this entry. */ | |
| 6690 vim_free(stp[i].st_word); | |
| 6691 --gap->ga_len; | |
| 6692 if (i < gap->ga_len) | |
| 6693 mch_memmove(stp + i, stp + i + 1, | |
| 6694 sizeof(suggest_T) * (gap->ga_len - i)); | |
| 6695 } | |
| 6696 } | |
| 6697 } | |
| 6698 | |
| 323 | 6699 |
| 6700 /* | |
| 6701 * Add a word to be banned. | |
| 6702 */ | |
| 6703 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6704 add_banned( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6705 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6706 char_u *word) |
| 323 | 6707 { |
| 626 | 6708 char_u *s; |
| 323 | 6709 hash_T hash; |
| 6710 hashitem_T *hi; | |
| 6711 | |
| 625 | 6712 hash = hash_hash(word); |
| 6713 hi = hash_lookup(&su->su_banned, word, hash); | |
| 6714 if (HASHITEM_EMPTY(hi)) | |
| 6715 { | |
| 6716 s = vim_strsave(word); | |
| 6717 if (s != NULL) | |
| 323 | 6718 hash_add_item(&su->su_banned, hi, s, hash); |
| 625 | 6719 } |
| 323 | 6720 } |
| 6721 | |
| 324 | 6722 /* |
| 534 | 6723 * Recompute the score for all suggestions if sound-folding is possible. This |
| 6724 * is slow, thus only done for the final results. | |
| 324 | 6725 */ |
| 6726 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6727 rescore_suggestions(suginfo_T *su) |
| 324 | 6728 { |
| 6729 int i; | |
| 534 | 6730 |
| 6731 if (su->su_sallang != NULL) | |
| 503 | 6732 for (i = 0; i < su->su_ga.ga_len; ++i) |
| 534 | 6733 rescore_one(su, &SUG(su->su_ga, i)); |
| 6734 } | |
| 6735 | |
| 6736 /* | |
| 6737 * Recompute the score for one suggestion if sound-folding is possible. | |
| 6738 */ | |
| 6739 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6740 rescore_one(suginfo_T *su, suggest_T *stp) |
| 534 | 6741 { |
| 6742 slang_T *slang = stp->st_slang; | |
| 6743 char_u sal_badword[MAXWLEN]; | |
| 535 | 6744 char_u *p; |
| 534 | 6745 |
| 6746 /* Only rescore suggestions that have no sal score yet and do have a | |
| 6747 * language. */ | |
| 6748 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus) | |
| 6749 { | |
| 6750 if (slang == su->su_sallang) | |
| 535 | 6751 p = su->su_sal_badword; |
| 534 | 6752 else |
| 6753 { | |
| 6754 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword); | |
| 535 | 6755 p = sal_badword; |
| 6756 } | |
| 6757 | |
| 6758 stp->st_altscore = stp_sal_score(stp, su, slang, p); | |
| 534 | 6759 if (stp->st_altscore == SCORE_MAXMAX) |
| 6760 stp->st_altscore = SCORE_BIG; | |
| 6761 stp->st_score = RESCORE(stp->st_score, stp->st_altscore); | |
| 6762 stp->st_had_bonus = TRUE; | |
| 324 | 6763 } |
| 6764 } | |
| 6765 | |
|
16606
7e733046db1d
patch 8.1.1306: Borland support is outdated and doesn't work
Bram Moolenaar <Bram@vim.org>
parents:
16277
diff
changeset
|
6766 static int sug_compare(const void *s1, const void *s2); |
| 323 | 6767 |
| 6768 /* | |
| 6769 * Function given to qsort() to sort the suggestions on st_score. | |
| 529 | 6770 * First on "st_score", then "st_altscore" then alphabetically. |
| 323 | 6771 */ |
| 6772 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6773 sug_compare(const void *s1, const void *s2) |
| 323 | 6774 { |
| 6775 suggest_T *p1 = (suggest_T *)s1; | |
| 6776 suggest_T *p2 = (suggest_T *)s2; | |
| 344 | 6777 int n = p1->st_score - p2->st_score; |
| 6778 | |
| 6779 if (n == 0) | |
| 529 | 6780 { |
| 6781 n = p1->st_altscore - p2->st_altscore; | |
| 6782 if (n == 0) | |
| 6783 n = STRICMP(p1->st_word, p2->st_word); | |
| 6784 } | |
| 344 | 6785 return n; |
| 323 | 6786 } |
| 6787 | |
| 6788 /* | |
| 6789 * Cleanup the suggestions: | |
| 6790 * - Sort on score. | |
| 6791 * - Remove words that won't be displayed. | |
| 344 | 6792 * Returns the maximum score in the list or "maxscore" unmodified. |
| 323 | 6793 */ |
| 344 | 6794 static int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6795 cleanup_suggestions( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6796 garray_T *gap, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6797 int maxscore, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6798 int keep) /* nr of suggestions to keep */ |
| 323 | 6799 { |
| 344 | 6800 suggest_T *stp = &SUG(*gap, 0); |
| 323 | 6801 int i; |
| 6802 | |
| 6803 /* Sort the list. */ | |
| 344 | 6804 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare); |
| 323 | 6805 |
| 6806 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 344 | 6807 if (gap->ga_len > keep) |
| 323 | 6808 { |
| 344 | 6809 for (i = keep; i < gap->ga_len; ++i) |
| 323 | 6810 vim_free(stp[i].st_word); |
| 344 | 6811 gap->ga_len = keep; |
| 6812 return stp[keep - 1].st_score; | |
| 323 | 6813 } |
| 344 | 6814 return maxscore; |
| 323 | 6815 } |
| 6816 | |
| 372 | 6817 #if defined(FEAT_EVAL) || defined(PROTO) |
| 6818 /* | |
| 6819 * Soundfold a string, for soundfold(). | |
| 6820 * Result is in allocated memory, NULL for an error. | |
| 6821 */ | |
| 6822 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6823 eval_soundfold(char_u *word) |
| 372 | 6824 { |
| 6825 langp_T *lp; | |
| 6826 char_u sound[MAXWLEN]; | |
| 500 | 6827 int lpi; |
| 372 | 6828 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6829 if (curwin->w_p_spell && *curwin->w_s->b_p_spl != NUL) |
| 372 | 6830 /* 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
|
6831 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
|
6832 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6833 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 372 | 6834 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6835 { | |
| 6836 /* soundfold the word */ | |
| 375 | 6837 spell_soundfold(lp->lp_slang, word, FALSE, sound); |
| 372 | 6838 return vim_strsave(sound); |
| 6839 } | |
| 500 | 6840 } |
| 372 | 6841 |
| 6842 /* No language with sound folding, return word as-is. */ | |
| 6843 return vim_strsave(word); | |
| 6844 } | |
| 6845 #endif | |
| 6846 | |
| 323 | 6847 /* |
| 6848 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 485 | 6849 * |
| 6850 * There are many ways to turn a word into a sound-a-like representation. The | |
| 6851 * oldest is Soundex (1918!). A nice overview can be found in "Approximate | |
| 6852 * swedish name matching - survey and test of different algorithms" by Klas | |
| 6853 * Erikson. | |
| 6854 * | |
| 6855 * We support two methods: | |
| 6856 * 1. SOFOFROM/SOFOTO do a simple character mapping. | |
| 6857 * 2. SAL items define a more advanced sound-folding (and much slower). | |
| 323 | 6858 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
6859 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6860 spell_soundfold( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6861 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6862 char_u *inword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6863 int folded, /* "inword" is already case-folded */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6864 char_u *res) |
| 375 | 6865 { |
| 6866 char_u fword[MAXWLEN]; | |
| 6867 char_u *word; | |
| 6868 | |
| 6869 if (slang->sl_sofo) | |
| 6870 /* SOFOFROM and SOFOTO used */ | |
| 6871 spell_soundfold_sofo(slang, inword, res); | |
| 6872 else | |
| 6873 { | |
| 6874 /* SAL items used. Requires the word to be case-folded. */ | |
| 6875 if (folded) | |
| 6876 word = inword; | |
| 6877 else | |
| 6878 { | |
| 835 | 6879 (void)spell_casefold(inword, (int)STRLEN(inword), fword, MAXWLEN); |
| 375 | 6880 word = fword; |
| 6881 } | |
| 6882 | |
| 6883 if (has_mbyte) | |
| 6884 spell_soundfold_wsal(slang, word, res); | |
| 6885 else | |
| 6886 spell_soundfold_sal(slang, word, res); | |
| 6887 } | |
| 6888 } | |
| 6889 | |
| 6890 /* | |
| 6891 * Perform sound folding of "inword" into "res" according to SOFOFROM and | |
| 6892 * SOFOTO lines. | |
| 6893 */ | |
| 6894 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6895 spell_soundfold_sofo(slang_T *slang, char_u *inword, char_u *res) |
| 375 | 6896 { |
| 6897 char_u *s; | |
| 6898 int ri = 0; | |
| 6899 int c; | |
| 6900 | |
| 6901 if (has_mbyte) | |
| 6902 { | |
| 6903 int prevc = 0; | |
| 6904 int *ip; | |
| 6905 | |
| 6906 /* The sl_sal_first[] table contains the translation for chars up to | |
| 6907 * 255, sl_sal the rest. */ | |
| 6908 for (s = inword; *s != NUL; ) | |
| 6909 { | |
| 474 | 6910 c = mb_cptr2char_adv(&s); |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
6911 if (enc_utf8 ? utf_class(c) == 0 : VIM_ISWHITE(c)) |
| 375 | 6912 c = ' '; |
| 6913 else if (c < 256) | |
| 6914 c = slang->sl_sal_first[c]; | |
| 6915 else | |
| 6916 { | |
| 6917 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff]; | |
| 6918 if (ip == NULL) /* empty list, can't match */ | |
| 6919 c = NUL; | |
| 6920 else | |
| 6921 for (;;) /* find "c" in the list */ | |
| 6922 { | |
| 6923 if (*ip == 0) /* not found */ | |
| 6924 { | |
| 6925 c = NUL; | |
| 6926 break; | |
| 6927 } | |
| 6928 if (*ip == c) /* match! */ | |
| 6929 { | |
| 6930 c = ip[1]; | |
| 6931 break; | |
| 6932 } | |
| 6933 ip += 2; | |
| 6934 } | |
| 6935 } | |
| 6936 | |
| 6937 if (c != NUL && c != prevc) | |
| 6938 { | |
| 6939 ri += mb_char2bytes(c, res + ri); | |
| 6940 if (ri + MB_MAXBYTES > MAXWLEN) | |
| 6941 break; | |
| 6942 prevc = c; | |
| 6943 } | |
| 6944 } | |
| 6945 } | |
| 6946 else | |
| 6947 { | |
| 6948 /* The sl_sal_first[] table contains the translation. */ | |
| 6949 for (s = inword; (c = *s) != NUL; ++s) | |
| 6950 { | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
6951 if (VIM_ISWHITE(c)) |
| 375 | 6952 c = ' '; |
| 6953 else | |
| 6954 c = slang->sl_sal_first[c]; | |
| 6955 if (c != NUL && (ri == 0 || res[ri - 1] != c)) | |
| 6956 res[ri++] = c; | |
| 6957 } | |
| 6958 } | |
| 6959 | |
| 6960 res[ri] = NUL; | |
| 6961 } | |
| 6962 | |
| 6963 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6964 spell_soundfold_sal(slang_T *slang, char_u *inword, char_u *res) |
| 323 | 6965 { |
| 344 | 6966 salitem_T *smp; |
| 323 | 6967 char_u word[MAXWLEN]; |
| 375 | 6968 char_u *s = inword; |
| 323 | 6969 char_u *t; |
| 344 | 6970 char_u *pf; |
| 323 | 6971 int i, j, z; |
| 344 | 6972 int reslen; |
| 323 | 6973 int n, k = 0; |
| 6974 int z0; | |
| 6975 int k0; | |
| 6976 int n0; | |
| 6977 int c; | |
| 6978 int pri; | |
| 6979 int p0 = -333; | |
| 6980 int c0; | |
| 6981 | |
| 324 | 6982 /* Remove accents, if wanted. We actually remove all non-word characters. |
| 375 | 6983 * But keep white space. We need a copy, the word may be changed here. */ |
| 323 | 6984 if (slang->sl_rem_accents) |
| 6985 { | |
| 6986 t = word; | |
| 375 | 6987 while (*s != NUL) |
| 323 | 6988 { |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
6989 if (VIM_ISWHITE(*s)) |
| 344 | 6990 { |
| 6991 *t++ = ' '; | |
| 6992 s = skipwhite(s); | |
| 6993 } | |
| 323 | 6994 else |
| 6995 { | |
| 5477 | 6996 if (spell_iswordp_nmw(s, curwin)) |
| 323 | 6997 *t++ = *s; |
| 6998 ++s; | |
| 6999 } | |
| 7000 } | |
| 7001 *t = NUL; | |
| 7002 } | |
| 7003 else | |
| 2768 | 7004 vim_strncpy(word, s, MAXWLEN - 1); |
| 323 | 7005 |
| 344 | 7006 smp = (salitem_T *)slang->sl_sal.ga_data; |
| 323 | 7007 |
| 7008 /* | |
| 7009 * This comes from Aspell phonet.cpp. Converted from C++ to C. | |
| 324 | 7010 * Changed to keep spaces. |
| 323 | 7011 */ |
| 344 | 7012 i = reslen = z = 0; |
| 323 | 7013 while ((c = word[i]) != NUL) |
| 7014 { | |
| 344 | 7015 /* Start with the first rule that has the character in the word. */ |
| 323 | 7016 n = slang->sl_sal_first[c]; |
| 7017 z0 = 0; | |
| 7018 | |
| 7019 if (n >= 0) | |
| 7020 { | |
| 7021 /* check all rules for the same letter */ | |
| 344 | 7022 for (; (s = smp[n].sm_lead)[0] == c; ++n) |
| 323 | 7023 { |
| 344 | 7024 /* Quickly skip entries that don't match the word. Most |
| 7025 * entries are less then three chars, optimize for that. */ | |
| 7026 k = smp[n].sm_leadlen; | |
| 7027 if (k > 1) | |
| 323 | 7028 { |
| 344 | 7029 if (word[i + 1] != s[1]) |
| 7030 continue; | |
| 7031 if (k > 2) | |
| 7032 { | |
| 7033 for (j = 2; j < k; ++j) | |
| 7034 if (word[i + j] != s[j]) | |
| 7035 break; | |
| 7036 if (j < k) | |
| 7037 continue; | |
| 7038 } | |
| 323 | 7039 } |
| 7040 | |
| 375 | 7041 if ((pf = smp[n].sm_oneof) != NULL) |
| 323 | 7042 { |
| 375 | 7043 /* Check for match with one of the chars in "sm_oneof". */ |
| 344 | 7044 while (*pf != NUL && *pf != word[i + k]) |
| 7045 ++pf; | |
| 7046 if (*pf == NUL) | |
| 7047 continue; | |
| 7048 ++k; | |
| 323 | 7049 } |
| 344 | 7050 s = smp[n].sm_rules; |
| 7051 pri = 5; /* default priority */ | |
| 323 | 7052 |
| 7053 p0 = *s; | |
| 7054 k0 = k; | |
| 7055 while (*s == '-' && k > 1) | |
| 7056 { | |
| 7057 k--; | |
| 7058 s++; | |
| 7059 } | |
| 7060 if (*s == '<') | |
| 7061 s++; | |
| 344 | 7062 if (VIM_ISDIGIT(*s)) |
| 323 | 7063 { |
| 7064 /* determine priority */ | |
| 7065 pri = *s - '0'; | |
| 7066 s++; | |
| 7067 } | |
| 7068 if (*s == '^' && *(s + 1) == '^') | |
| 7069 s++; | |
| 7070 | |
| 7071 if (*s == NUL | |
| 7072 || (*s == '^' | |
| 324 | 7073 && (i == 0 || !(word[i - 1] == ' ' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7074 || spell_iswordp(word + i - 1, curwin))) |
| 323 | 7075 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7076 || (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 7077 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7078 && spell_iswordp(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7079 && (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 7080 { |
| 7081 /* search for followup rules, if: */ | |
| 7082 /* followup and k > 1 and NO '-' in searchstring */ | |
| 7083 c0 = word[i + k - 1]; | |
| 7084 n0 = slang->sl_sal_first[c0]; | |
| 7085 | |
| 7086 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 344 | 7087 && p0 != '-' && word[i + k] != NUL) |
| 323 | 7088 { |
| 7089 /* test follow-up rule for "word[i + k]" */ | |
| 344 | 7090 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0) |
| 323 | 7091 { |
| 344 | 7092 /* Quickly skip entries that don't match the word. |
| 7093 * */ | |
| 7094 k0 = smp[n0].sm_leadlen; | |
| 7095 if (k0 > 1) | |
| 323 | 7096 { |
| 344 | 7097 if (word[i + k] != s[1]) |
| 7098 continue; | |
| 7099 if (k0 > 2) | |
| 7100 { | |
| 7101 pf = word + i + k + 1; | |
| 7102 for (j = 2; j < k0; ++j) | |
| 7103 if (*pf++ != s[j]) | |
| 7104 break; | |
| 7105 if (j < k0) | |
| 7106 continue; | |
| 7107 } | |
| 323 | 7108 } |
| 344 | 7109 k0 += k - 1; |
| 7110 | |
| 375 | 7111 if ((pf = smp[n0].sm_oneof) != NULL) |
| 323 | 7112 { |
| 344 | 7113 /* Check for match with one of the chars in |
| 375 | 7114 * "sm_oneof". */ |
| 344 | 7115 while (*pf != NUL && *pf != word[i + k0]) |
| 7116 ++pf; | |
| 7117 if (*pf == NUL) | |
| 7118 continue; | |
| 7119 ++k0; | |
| 323 | 7120 } |
| 344 | 7121 |
| 7122 p0 = 5; | |
| 7123 s = smp[n0].sm_rules; | |
| 323 | 7124 while (*s == '-') |
| 7125 { | |
| 344 | 7126 /* "k0" gets NOT reduced because |
| 7127 * "if (k0 == k)" */ | |
| 323 | 7128 s++; |
| 7129 } | |
| 7130 if (*s == '<') | |
| 7131 s++; | |
| 344 | 7132 if (VIM_ISDIGIT(*s)) |
| 323 | 7133 { |
| 7134 p0 = *s - '0'; | |
| 7135 s++; | |
| 7136 } | |
| 7137 | |
| 7138 if (*s == NUL | |
| 7139 /* *s == '^' cuts */ | |
| 7140 || (*s == '$' | |
| 376 | 7141 && !spell_iswordp(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7142 curwin))) |
| 323 | 7143 { |
| 7144 if (k0 == k) | |
| 7145 /* this is just a piece of the string */ | |
| 7146 continue; | |
| 7147 | |
| 7148 if (p0 < pri) | |
| 7149 /* priority too low */ | |
| 7150 continue; | |
| 7151 /* rule fits; stop search */ | |
| 7152 break; | |
| 7153 } | |
| 7154 } | |
| 7155 | |
| 344 | 7156 if (p0 >= pri && smp[n0].sm_lead[0] == c0) |
| 323 | 7157 continue; |
| 7158 } | |
| 7159 | |
| 7160 /* replace string */ | |
| 344 | 7161 s = smp[n].sm_to; |
| 389 | 7162 if (s == NULL) |
| 7163 s = (char_u *)""; | |
| 344 | 7164 pf = smp[n].sm_rules; |
| 7165 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0; | |
| 323 | 7166 if (p0 == 1 && z == 0) |
| 7167 { | |
| 7168 /* rule with '<' is used */ | |
| 344 | 7169 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c |
| 7170 || res[reslen - 1] == *s)) | |
| 7171 reslen--; | |
| 323 | 7172 z0 = 1; |
| 7173 z = 1; | |
| 7174 k0 = 0; | |
| 372 | 7175 while (*s != NUL && word[i + k0] != NUL) |
| 323 | 7176 { |
| 7177 word[i + k0] = *s; | |
| 7178 k0++; | |
| 7179 s++; | |
| 7180 } | |
| 7181 if (k > k0) | |
| 1619 | 7182 STRMOVE(word + i + k0, word + i + k); |
| 323 | 7183 |
| 7184 /* new "actual letter" */ | |
| 7185 c = word[i]; | |
| 7186 } | |
| 7187 else | |
| 7188 { | |
| 7189 /* no '<' rule used */ | |
| 7190 i += k - 1; | |
| 7191 z = 0; | |
| 344 | 7192 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN) |
| 323 | 7193 { |
| 344 | 7194 if (reslen == 0 || res[reslen - 1] != *s) |
| 372 | 7195 res[reslen++] = *s; |
| 323 | 7196 s++; |
| 7197 } | |
| 7198 /* new "actual letter" */ | |
| 7199 c = *s; | |
| 344 | 7200 if (strstr((char *)pf, "^^") != NULL) |
| 323 | 7201 { |
| 7202 if (c != NUL) | |
| 372 | 7203 res[reslen++] = c; |
| 1619 | 7204 STRMOVE(word, word + i + 1); |
| 323 | 7205 i = 0; |
| 7206 z0 = 1; | |
| 7207 } | |
| 7208 } | |
| 7209 break; | |
| 7210 } | |
| 7211 } | |
| 7212 } | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7213 else if (VIM_ISWHITE(c)) |
| 324 | 7214 { |
| 7215 c = ' '; | |
| 7216 k = 1; | |
| 7217 } | |
| 323 | 7218 |
| 7219 if (z0 == 0) | |
| 7220 { | |
| 344 | 7221 if (k && !p0 && reslen < MAXWLEN && c != NUL |
| 7222 && (!slang->sl_collapse || reslen == 0 | |
| 7223 || res[reslen - 1] != c)) | |
| 323 | 7224 /* condense only double letters */ |
| 372 | 7225 res[reslen++] = c; |
| 323 | 7226 |
| 7227 i++; | |
| 7228 z = 0; | |
| 7229 k = 0; | |
| 7230 } | |
| 7231 } | |
| 7232 | |
| 344 | 7233 res[reslen] = NUL; |
| 323 | 7234 } |
| 7235 | |
| 372 | 7236 /* |
| 7237 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 7238 * Multi-byte version of spell_soundfold(). | |
| 7239 */ | |
| 7240 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7241 spell_soundfold_wsal(slang_T *slang, char_u *inword, char_u *res) |
| 372 | 7242 { |
| 375 | 7243 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data; |
| 372 | 7244 int word[MAXWLEN]; |
| 7245 int wres[MAXWLEN]; | |
| 7246 int l; | |
| 7247 char_u *s; | |
| 7248 int *ws; | |
| 7249 char_u *t; | |
| 7250 int *pf; | |
| 7251 int i, j, z; | |
| 7252 int reslen; | |
| 7253 int n, k = 0; | |
| 7254 int z0; | |
| 7255 int k0; | |
| 7256 int n0; | |
| 7257 int c; | |
| 7258 int pri; | |
| 7259 int p0 = -333; | |
| 7260 int c0; | |
| 7261 int did_white = FALSE; | |
| 3520 | 7262 int wordlen; |
| 7263 | |
| 372 | 7264 |
| 7265 /* | |
| 7266 * Convert the multi-byte string to a wide-character string. | |
| 7267 * Remove accents, if wanted. We actually remove all non-word characters. | |
| 7268 * But keep white space. | |
| 7269 */ | |
| 3520 | 7270 wordlen = 0; |
| 372 | 7271 for (s = inword; *s != NUL; ) |
| 7272 { | |
| 7273 t = s; | |
| 474 | 7274 c = mb_cptr2char_adv(&s); |
| 372 | 7275 if (slang->sl_rem_accents) |
| 7276 { | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7277 if (enc_utf8 ? utf_class(c) == 0 : VIM_ISWHITE(c)) |
| 372 | 7278 { |
| 7279 if (did_white) | |
| 7280 continue; | |
| 7281 c = ' '; | |
| 7282 did_white = TRUE; | |
| 7283 } | |
| 7284 else | |
| 7285 { | |
| 7286 did_white = FALSE; | |
| 5477 | 7287 if (!spell_iswordp_nmw(t, curwin)) |
| 372 | 7288 continue; |
| 7289 } | |
| 7290 } | |
| 3520 | 7291 word[wordlen++] = c; |
| 7292 } | |
| 7293 word[wordlen] = NUL; | |
| 372 | 7294 |
| 7295 /* | |
| 3520 | 7296 * This algorithm comes from Aspell phonet.cpp. |
| 372 | 7297 * Converted from C++ to C. Added support for multi-byte chars. |
| 7298 * Changed to keep spaces. | |
| 7299 */ | |
| 7300 i = reslen = z = 0; | |
| 7301 while ((c = word[i]) != NUL) | |
| 7302 { | |
| 7303 /* Start with the first rule that has the character in the word. */ | |
| 7304 n = slang->sl_sal_first[c & 0xff]; | |
| 7305 z0 = 0; | |
| 7306 | |
| 7307 if (n >= 0) | |
| 7308 { | |
|
2455
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
7309 /* 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
|
7310 * 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
|
7311 * (c & 0xff) is NUL. */ |
|
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
7312 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
|
7313 && ws[0] != NUL; ++n) |
| 372 | 7314 { |
| 7315 /* Quickly skip entries that don't match the word. Most | |
| 7316 * entries are less then three chars, optimize for that. */ | |
| 375 | 7317 if (c != ws[0]) |
| 7318 continue; | |
| 372 | 7319 k = smp[n].sm_leadlen; |
| 7320 if (k > 1) | |
| 7321 { | |
| 7322 if (word[i + 1] != ws[1]) | |
| 7323 continue; | |
| 7324 if (k > 2) | |
| 7325 { | |
| 7326 for (j = 2; j < k; ++j) | |
| 7327 if (word[i + j] != ws[j]) | |
| 7328 break; | |
| 7329 if (j < k) | |
| 7330 continue; | |
| 7331 } | |
| 7332 } | |
| 7333 | |
| 375 | 7334 if ((pf = smp[n].sm_oneof_w) != NULL) |
| 372 | 7335 { |
| 375 | 7336 /* Check for match with one of the chars in "sm_oneof". */ |
| 372 | 7337 while (*pf != NUL && *pf != word[i + k]) |
| 7338 ++pf; | |
| 7339 if (*pf == NUL) | |
| 7340 continue; | |
| 7341 ++k; | |
| 7342 } | |
| 7343 s = smp[n].sm_rules; | |
| 7344 pri = 5; /* default priority */ | |
| 7345 | |
| 7346 p0 = *s; | |
| 7347 k0 = k; | |
| 7348 while (*s == '-' && k > 1) | |
| 7349 { | |
| 7350 k--; | |
| 7351 s++; | |
| 7352 } | |
| 7353 if (*s == '<') | |
| 7354 s++; | |
| 7355 if (VIM_ISDIGIT(*s)) | |
| 7356 { | |
| 7357 /* determine priority */ | |
| 7358 pri = *s - '0'; | |
| 7359 s++; | |
| 7360 } | |
| 7361 if (*s == '^' && *(s + 1) == '^') | |
| 7362 s++; | |
| 7363 | |
| 7364 if (*s == NUL | |
| 7365 || (*s == '^' | |
| 7366 && (i == 0 || !(word[i - 1] == ' ' | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7367 || spell_iswordp_w(word + i - 1, curwin))) |
| 372 | 7368 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7369 || (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 7370 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7371 && spell_iswordp_w(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7372 && (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 7373 { |
| 7374 /* search for followup rules, if: */ | |
| 7375 /* followup and k > 1 and NO '-' in searchstring */ | |
| 7376 c0 = word[i + k - 1]; | |
| 7377 n0 = slang->sl_sal_first[c0 & 0xff]; | |
| 7378 | |
| 7379 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 7380 && p0 != '-' && word[i + k] != NUL) | |
| 7381 { | |
| 375 | 7382 /* Test follow-up rule for "word[i + k]"; loop over |
| 7383 * all entries with the same index byte. */ | |
| 372 | 7384 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff) |
| 7385 == (c0 & 0xff); ++n0) | |
| 7386 { | |
| 7387 /* Quickly skip entries that don't match the word. | |
| 375 | 7388 */ |
| 7389 if (c0 != ws[0]) | |
| 7390 continue; | |
| 372 | 7391 k0 = smp[n0].sm_leadlen; |
| 7392 if (k0 > 1) | |
| 7393 { | |
| 7394 if (word[i + k] != ws[1]) | |
| 7395 continue; | |
| 7396 if (k0 > 2) | |
| 7397 { | |
| 7398 pf = word + i + k + 1; | |
| 7399 for (j = 2; j < k0; ++j) | |
| 7400 if (*pf++ != ws[j]) | |
| 7401 break; | |
| 7402 if (j < k0) | |
| 7403 continue; | |
| 7404 } | |
| 7405 } | |
| 7406 k0 += k - 1; | |
| 7407 | |
| 375 | 7408 if ((pf = smp[n0].sm_oneof_w) != NULL) |
| 372 | 7409 { |
| 7410 /* Check for match with one of the chars in | |
| 375 | 7411 * "sm_oneof". */ |
| 372 | 7412 while (*pf != NUL && *pf != word[i + k0]) |
| 7413 ++pf; | |
| 7414 if (*pf == NUL) | |
| 7415 continue; | |
| 7416 ++k0; | |
| 7417 } | |
| 7418 | |
| 7419 p0 = 5; | |
| 7420 s = smp[n0].sm_rules; | |
| 7421 while (*s == '-') | |
| 7422 { | |
| 7423 /* "k0" gets NOT reduced because | |
| 7424 * "if (k0 == k)" */ | |
| 7425 s++; | |
| 7426 } | |
| 7427 if (*s == '<') | |
| 7428 s++; | |
| 7429 if (VIM_ISDIGIT(*s)) | |
| 7430 { | |
| 7431 p0 = *s - '0'; | |
| 7432 s++; | |
| 7433 } | |
| 7434 | |
| 7435 if (*s == NUL | |
| 7436 /* *s == '^' cuts */ | |
| 7437 || (*s == '$' | |
| 376 | 7438 && !spell_iswordp_w(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7439 curwin))) |
| 372 | 7440 { |
| 7441 if (k0 == k) | |
| 7442 /* this is just a piece of the string */ | |
| 7443 continue; | |
| 7444 | |
| 7445 if (p0 < pri) | |
| 7446 /* priority too low */ | |
| 7447 continue; | |
| 7448 /* rule fits; stop search */ | |
| 7449 break; | |
| 7450 } | |
| 7451 } | |
| 7452 | |
| 7453 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff) | |
| 7454 == (c0 & 0xff)) | |
| 7455 continue; | |
| 7456 } | |
| 7457 | |
| 7458 /* replace string */ | |
| 7459 ws = smp[n].sm_to_w; | |
| 7460 s = smp[n].sm_rules; | |
| 7461 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0; | |
| 7462 if (p0 == 1 && z == 0) | |
| 7463 { | |
| 7464 /* rule with '<' is used */ | |
| 389 | 7465 if (reslen > 0 && ws != NULL && *ws != NUL |
| 7466 && (wres[reslen - 1] == c | |
| 372 | 7467 || wres[reslen - 1] == *ws)) |
| 7468 reslen--; | |
| 7469 z0 = 1; | |
| 7470 z = 1; | |
| 7471 k0 = 0; | |
| 389 | 7472 if (ws != NULL) |
| 7473 while (*ws != NUL && word[i + k0] != NUL) | |
| 7474 { | |
| 7475 word[i + k0] = *ws; | |
| 7476 k0++; | |
| 7477 ws++; | |
| 7478 } | |
| 372 | 7479 if (k > k0) |
| 7480 mch_memmove(word + i + k0, word + i + k, | |
| 3520 | 7481 sizeof(int) * (wordlen - (i + k) + 1)); |
| 372 | 7482 |
| 7483 /* new "actual letter" */ | |
| 7484 c = word[i]; | |
| 7485 } | |
| 7486 else | |
| 7487 { | |
| 7488 /* no '<' rule used */ | |
| 7489 i += k - 1; | |
| 7490 z = 0; | |
| 389 | 7491 if (ws != NULL) |
| 7492 while (*ws != NUL && ws[1] != NUL | |
| 7493 && reslen < MAXWLEN) | |
| 7494 { | |
| 7495 if (reslen == 0 || wres[reslen - 1] != *ws) | |
| 7496 wres[reslen++] = *ws; | |
| 7497 ws++; | |
| 7498 } | |
| 372 | 7499 /* new "actual letter" */ |
| 389 | 7500 if (ws == NULL) |
| 7501 c = NUL; | |
| 7502 else | |
| 7503 c = *ws; | |
| 372 | 7504 if (strstr((char *)s, "^^") != NULL) |
| 7505 { | |
| 7506 if (c != NUL) | |
| 7507 wres[reslen++] = c; | |
| 7508 mch_memmove(word, word + i + 1, | |
| 3520 | 7509 sizeof(int) * (wordlen - (i + 1) + 1)); |
| 372 | 7510 i = 0; |
| 7511 z0 = 1; | |
| 7512 } | |
| 7513 } | |
| 7514 break; | |
| 7515 } | |
| 7516 } | |
| 7517 } | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7518 else if (VIM_ISWHITE(c)) |
| 372 | 7519 { |
| 7520 c = ' '; | |
| 7521 k = 1; | |
| 7522 } | |
| 7523 | |
| 7524 if (z0 == 0) | |
| 7525 { | |
| 7526 if (k && !p0 && reslen < MAXWLEN && c != NUL | |
| 7527 && (!slang->sl_collapse || reslen == 0 | |
| 7528 || wres[reslen - 1] != c)) | |
| 7529 /* condense only double letters */ | |
| 7530 wres[reslen++] = c; | |
| 7531 | |
| 7532 i++; | |
| 7533 z = 0; | |
| 7534 k = 0; | |
| 7535 } | |
| 7536 } | |
| 7537 | |
| 7538 /* Convert wide characters in "wres" to a multi-byte string in "res". */ | |
| 7539 l = 0; | |
| 7540 for (n = 0; n < reslen; ++n) | |
| 7541 { | |
| 7542 l += mb_char2bytes(wres[n], res + l); | |
| 7543 if (l + MB_MAXBYTES > MAXWLEN) | |
| 7544 break; | |
| 7545 } | |
| 7546 res[l] = NUL; | |
| 7547 } | |
| 7548 | |
| 324 | 7549 /* |
| 344 | 7550 * Compute a score for two sound-a-like words. |
| 7551 * This permits up to two inserts/deletes/swaps/etc. to keep things fast. | |
| 7552 * Instead of a generic loop we write out the code. That keeps it fast by | |
| 7553 * avoiding checks that will not be possible. | |
| 7554 */ | |
| 7555 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7556 soundalike_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7557 char_u *goodstart, /* sound-folded good word */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7558 char_u *badstart) /* sound-folded bad word */ |
| 344 | 7559 { |
| 351 | 7560 char_u *goodsound = goodstart; |
| 7561 char_u *badsound = badstart; | |
| 7562 int goodlen; | |
| 7563 int badlen; | |
| 344 | 7564 int n; |
| 7565 char_u *pl, *ps; | |
| 7566 char_u *pl2, *ps2; | |
| 351 | 7567 int score = 0; |
| 7568 | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7569 /* Adding/inserting "*" at the start (word starts with vowel) shouldn't be |
| 351 | 7570 * counted so much, vowels halfway the word aren't counted at all. */ |
| 7571 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound) | |
| 7572 { | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7573 if ((badsound[0] == NUL && goodsound[1] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7574 || (goodsound[0] == NUL && badsound[1] == NUL)) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7575 /* 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
|
7576 return SCORE_DEL; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7577 if (badsound[0] == NUL || goodsound[0] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7578 /* more than two changes */ |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7579 return SCORE_MAXMAX; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7580 |
| 625 | 7581 if (badsound[1] == goodsound[1] |
| 7582 || (badsound[1] != NUL | |
| 7583 && goodsound[1] != NUL | |
| 7584 && badsound[2] == goodsound[2])) | |
| 7585 { | |
| 7586 /* handle like a substitute */ | |
| 7587 } | |
| 351 | 7588 else |
| 625 | 7589 { |
| 7590 score = 2 * SCORE_DEL / 3; | |
| 7591 if (*badsound == '*') | |
| 7592 ++badsound; | |
| 7593 else | |
| 7594 ++goodsound; | |
| 7595 } | |
| 351 | 7596 } |
| 7597 | |
| 835 | 7598 goodlen = (int)STRLEN(goodsound); |
| 7599 badlen = (int)STRLEN(badsound); | |
| 344 | 7600 |
| 1188 | 7601 /* Return quickly if the lengths are too different to be fixed by two |
| 344 | 7602 * changes. */ |
| 7603 n = goodlen - badlen; | |
| 7604 if (n < -2 || n > 2) | |
| 7605 return SCORE_MAXMAX; | |
| 7606 | |
| 7607 if (n > 0) | |
| 7608 { | |
| 351 | 7609 pl = goodsound; /* goodsound is longest */ |
| 344 | 7610 ps = badsound; |
| 7611 } | |
| 7612 else | |
| 7613 { | |
| 351 | 7614 pl = badsound; /* badsound is longest */ |
| 344 | 7615 ps = goodsound; |
| 7616 } | |
| 7617 | |
| 7618 /* Skip over the identical part. */ | |
| 7619 while (*pl == *ps && *pl != NUL) | |
| 7620 { | |
| 7621 ++pl; | |
| 7622 ++ps; | |
| 7623 } | |
| 7624 | |
| 7625 switch (n) | |
| 7626 { | |
| 7627 case -2: | |
| 7628 case 2: | |
| 7629 /* | |
| 7630 * Must delete two characters from "pl". | |
| 7631 */ | |
| 7632 ++pl; /* first delete */ | |
| 7633 while (*pl == *ps) | |
| 7634 { | |
| 7635 ++pl; | |
| 7636 ++ps; | |
| 7637 } | |
| 7638 /* strings must be equal after second delete */ | |
| 7639 if (STRCMP(pl + 1, ps) == 0) | |
| 351 | 7640 return score + SCORE_DEL * 2; |
| 344 | 7641 |
| 7642 /* Failed to compare. */ | |
| 7643 break; | |
| 7644 | |
| 7645 case -1: | |
| 7646 case 1: | |
| 7647 /* | |
| 7648 * Minimal one delete from "pl" required. | |
| 7649 */ | |
| 7650 | |
| 7651 /* 1: delete */ | |
| 7652 pl2 = pl + 1; | |
| 7653 ps2 = ps; | |
| 7654 while (*pl2 == *ps2) | |
| 7655 { | |
| 7656 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7657 return score + SCORE_DEL; |
| 344 | 7658 ++pl2; |
| 7659 ++ps2; | |
| 7660 } | |
| 7661 | |
| 7662 /* 2: delete then swap, then rest must be equal */ | |
| 7663 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7664 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7665 return score + SCORE_DEL + SCORE_SWAP; |
| 344 | 7666 |
| 7667 /* 3: delete then substitute, then the rest must be equal */ | |
| 7668 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7669 return score + SCORE_DEL + SCORE_SUBST; |
| 344 | 7670 |
| 7671 /* 4: first swap then delete */ | |
| 7672 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 7673 { | |
| 7674 pl2 = pl + 2; /* swap, skip two chars */ | |
| 7675 ps2 = ps + 2; | |
| 7676 while (*pl2 == *ps2) | |
| 7677 { | |
| 7678 ++pl2; | |
| 7679 ++ps2; | |
| 7680 } | |
| 7681 /* delete a char and then strings must be equal */ | |
| 7682 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7683 return score + SCORE_SWAP + SCORE_DEL; |
| 344 | 7684 } |
| 7685 | |
| 7686 /* 5: first substitute then delete */ | |
| 7687 pl2 = pl + 1; /* substitute, skip one char */ | |
| 7688 ps2 = ps + 1; | |
| 7689 while (*pl2 == *ps2) | |
| 7690 { | |
| 7691 ++pl2; | |
| 7692 ++ps2; | |
| 7693 } | |
| 7694 /* delete a char and then strings must be equal */ | |
| 7695 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7696 return score + SCORE_SUBST + SCORE_DEL; |
| 344 | 7697 |
| 7698 /* Failed to compare. */ | |
| 7699 break; | |
| 7700 | |
| 7701 case 0: | |
| 7702 /* | |
| 1808 | 7703 * Lengths are equal, thus changes must result in same length: An |
| 344 | 7704 * insert is only possible in combination with a delete. |
| 7705 * 1: check if for identical strings | |
| 7706 */ | |
| 7707 if (*pl == NUL) | |
| 351 | 7708 return score; |
| 344 | 7709 |
| 7710 /* 2: swap */ | |
| 7711 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 7712 { | |
| 7713 pl2 = pl + 2; /* swap, skip two chars */ | |
| 7714 ps2 = ps + 2; | |
| 7715 while (*pl2 == *ps2) | |
| 7716 { | |
| 7717 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7718 return score + SCORE_SWAP; |
| 344 | 7719 ++pl2; |
| 7720 ++ps2; | |
| 7721 } | |
| 7722 /* 3: swap and swap again */ | |
| 7723 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7724 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7725 return score + SCORE_SWAP + SCORE_SWAP; |
| 344 | 7726 |
| 7727 /* 4: swap and substitute */ | |
| 7728 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7729 return score + SCORE_SWAP + SCORE_SUBST; |
| 344 | 7730 } |
| 7731 | |
| 7732 /* 5: substitute */ | |
| 7733 pl2 = pl + 1; | |
| 7734 ps2 = ps + 1; | |
| 7735 while (*pl2 == *ps2) | |
| 7736 { | |
| 7737 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7738 return score + SCORE_SUBST; |
| 344 | 7739 ++pl2; |
| 7740 ++ps2; | |
| 7741 } | |
| 7742 | |
| 7743 /* 6: substitute and swap */ | |
| 7744 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7745 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7746 return score + SCORE_SUBST + SCORE_SWAP; |
| 344 | 7747 |
| 7748 /* 7: substitute and substitute */ | |
| 7749 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7750 return score + SCORE_SUBST + SCORE_SUBST; |
| 344 | 7751 |
| 7752 /* 8: insert then delete */ | |
| 7753 pl2 = pl; | |
| 7754 ps2 = ps + 1; | |
| 7755 while (*pl2 == *ps2) | |
| 7756 { | |
| 7757 ++pl2; | |
| 7758 ++ps2; | |
| 7759 } | |
| 7760 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7761 return score + SCORE_INS + SCORE_DEL; |
| 344 | 7762 |
| 7763 /* 9: delete then insert */ | |
| 7764 pl2 = pl + 1; | |
| 7765 ps2 = ps; | |
| 7766 while (*pl2 == *ps2) | |
| 7767 { | |
| 7768 ++pl2; | |
| 7769 ++ps2; | |
| 7770 } | |
| 7771 if (STRCMP(pl2, ps2 + 1) == 0) | |
| 351 | 7772 return score + SCORE_INS + SCORE_DEL; |
| 344 | 7773 |
| 7774 /* Failed to compare. */ | |
| 7775 break; | |
| 7776 } | |
| 7777 | |
| 7778 return SCORE_MAXMAX; | |
| 7779 } | |
| 324 | 7780 |
| 323 | 7781 /* |
| 7782 * Compute the "edit distance" to turn "badword" into "goodword". The less | |
| 344 | 7783 * deletes/inserts/substitutes/swaps are required the lower the score. |
| 324 | 7784 * |
| 485 | 7785 * The algorithm is described by Du and Chang, 1992. |
| 7786 * The implementation of the algorithm comes from Aspell editdist.cpp, | |
| 7787 * edit_distance(). It has been converted from C++ to C and modified to | |
| 7788 * support multi-byte characters. | |
| 323 | 7789 */ |
| 7790 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7791 spell_edit_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7792 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7793 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7794 char_u *goodword) |
| 323 | 7795 { |
| 7796 int *cnt; | |
| 1188 | 7797 int badlen, goodlen; /* lengths including NUL */ |
| 323 | 7798 int j, i; |
| 7799 int t; | |
| 7800 int bc, gc; | |
| 324 | 7801 int pbc, pgc; |
| 7802 char_u *p; | |
| 7803 int wbadword[MAXWLEN]; | |
| 7804 int wgoodword[MAXWLEN]; | |
| 7805 | |
| 7806 if (has_mbyte) | |
| 7807 { | |
| 7808 /* Get the characters from the multi-byte strings and put them in an | |
| 7809 * int array for easy access. */ | |
| 7810 for (p = badword, badlen = 0; *p != NUL; ) | |
| 474 | 7811 wbadword[badlen++] = mb_cptr2char_adv(&p); |
| 398 | 7812 wbadword[badlen++] = 0; |
| 324 | 7813 for (p = goodword, goodlen = 0; *p != NUL; ) |
| 474 | 7814 wgoodword[goodlen++] = mb_cptr2char_adv(&p); |
| 398 | 7815 wgoodword[goodlen++] = 0; |
| 324 | 7816 } |
| 7817 else | |
| 7818 { | |
| 835 | 7819 badlen = (int)STRLEN(badword) + 1; |
| 7820 goodlen = (int)STRLEN(goodword) + 1; | |
| 324 | 7821 } |
| 323 | 7822 |
| 7823 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */ | |
| 7824 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)] | |
|
16825
ce04ebdf26b8
patch 8.1.1414: alloc() returning "char_u *" causes a lot of type casts
Bram Moolenaar <Bram@vim.org>
parents:
16768
diff
changeset
|
7825 cnt = ALLOC_MULT(int, (badlen + 1) * (goodlen + 1)); |
| 324 | 7826 if (cnt == NULL) |
| 7827 return 0; /* out of memory */ | |
| 323 | 7828 |
| 7829 CNT(0, 0) = 0; | |
| 7830 for (j = 1; j <= goodlen; ++j) | |
| 625 | 7831 CNT(0, j) = CNT(0, j - 1) + SCORE_INS; |
| 323 | 7832 |
| 7833 for (i = 1; i <= badlen; ++i) | |
| 7834 { | |
| 625 | 7835 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL; |
| 323 | 7836 for (j = 1; j <= goodlen; ++j) |
| 7837 { | |
| 324 | 7838 if (has_mbyte) |
| 7839 { | |
| 7840 bc = wbadword[i - 1]; | |
| 7841 gc = wgoodword[j - 1]; | |
| 7842 } | |
| 7843 else | |
| 7844 { | |
| 7845 bc = badword[i - 1]; | |
| 7846 gc = goodword[j - 1]; | |
| 7847 } | |
| 323 | 7848 if (bc == gc) |
| 7849 CNT(i, j) = CNT(i - 1, j - 1); | |
| 7850 else | |
| 7851 { | |
| 7852 /* Use a better score when there is only a case difference. */ | |
| 324 | 7853 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) |
| 323 | 7854 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1); |
| 7855 else | |
| 625 | 7856 { |
| 7857 /* For a similar character use SCORE_SIMILAR. */ | |
| 7858 if (slang != NULL | |
| 7859 && slang->sl_has_map | |
| 7860 && similar_chars(slang, gc, bc)) | |
| 7861 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1); | |
| 7862 else | |
| 7863 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1); | |
| 7864 } | |
| 323 | 7865 |
| 324 | 7866 if (i > 1 && j > 1) |
| 323 | 7867 { |
| 324 | 7868 if (has_mbyte) |
| 7869 { | |
| 7870 pbc = wbadword[i - 2]; | |
| 7871 pgc = wgoodword[j - 2]; | |
| 7872 } | |
| 7873 else | |
| 7874 { | |
| 7875 pbc = badword[i - 2]; | |
| 7876 pgc = goodword[j - 2]; | |
| 7877 } | |
| 7878 if (bc == pgc && pbc == gc) | |
| 7879 { | |
| 7880 t = SCORE_SWAP + CNT(i - 2, j - 2); | |
| 7881 if (t < CNT(i, j)) | |
| 7882 CNT(i, j) = t; | |
| 7883 } | |
| 323 | 7884 } |
| 7885 t = SCORE_DEL + CNT(i - 1, j); | |
| 7886 if (t < CNT(i, j)) | |
| 7887 CNT(i, j) = t; | |
| 7888 t = SCORE_INS + CNT(i, j - 1); | |
| 7889 if (t < CNT(i, j)) | |
| 7890 CNT(i, j) = t; | |
| 7891 } | |
| 7892 } | |
| 7893 } | |
| 344 | 7894 |
| 7895 i = CNT(badlen - 1, goodlen - 1); | |
| 7896 vim_free(cnt); | |
| 7897 return i; | |
| 323 | 7898 } |
| 307 | 7899 |
| 625 | 7900 typedef struct |
| 7901 { | |
| 7902 int badi; | |
| 7903 int goodi; | |
| 7904 int score; | |
| 7905 } limitscore_T; | |
| 7906 | |
| 7907 /* | |
| 7908 * Like spell_edit_score(), but with a limit on the score to make it faster. | |
| 7909 * May return SCORE_MAXMAX when the score is higher than "limit". | |
| 7910 * | |
| 7911 * This uses a stack for the edits still to be tried. | |
| 7912 * The idea comes from Aspell leditdist.cpp. Rewritten in C and added support | |
| 7913 * for multi-byte characters. | |
| 7914 */ | |
| 7915 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7916 spell_edit_score_limit( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7917 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7918 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7919 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7920 int limit) |
| 625 | 7921 { |
| 7922 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 7923 int stackidx; | |
| 7924 int bi, gi; | |
| 7925 int bi2, gi2; | |
| 7926 int bc, gc; | |
| 7927 int score; | |
| 7928 int score_off; | |
| 7929 int minscore; | |
| 7930 int round; | |
| 7931 | |
| 7932 /* Multi-byte characters require a bit more work, use a different function | |
| 7933 * to avoid testing "has_mbyte" quite often. */ | |
| 7934 if (has_mbyte) | |
| 7935 return spell_edit_score_limit_w(slang, badword, goodword, limit); | |
| 7936 | |
| 7937 /* | |
| 7938 * The idea is to go from start to end over the words. So long as | |
| 7939 * characters are equal just continue, this always gives the lowest score. | |
| 7940 * When there is a difference try several alternatives. Each alternative | |
| 7941 * increases "score" for the edit distance. Some of the alternatives are | |
| 7942 * pushed unto a stack and tried later, some are tried right away. At the | |
| 7943 * end of the word the score for one alternative is known. The lowest | |
| 7944 * possible score is stored in "minscore". | |
| 7945 */ | |
| 7946 stackidx = 0; | |
| 7947 bi = 0; | |
| 7948 gi = 0; | |
| 7949 score = 0; | |
| 7950 minscore = limit + 1; | |
| 7951 | |
| 7952 for (;;) | |
| 7953 { | |
| 7954 /* Skip over an equal part, score remains the same. */ | |
| 7955 for (;;) | |
| 7956 { | |
| 7957 bc = badword[bi]; | |
| 7958 gc = goodword[gi]; | |
| 7959 if (bc != gc) /* stop at a char that's different */ | |
| 7960 break; | |
| 7961 if (bc == NUL) /* both words end */ | |
| 7962 { | |
| 7963 if (score < minscore) | |
| 7964 minscore = score; | |
| 7965 goto pop; /* do next alternative */ | |
| 7966 } | |
| 7967 ++bi; | |
| 7968 ++gi; | |
| 7969 } | |
| 7970 | |
| 7971 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 7972 { | |
| 7973 do | |
| 7974 { | |
| 7975 if ((score += SCORE_DEL) >= minscore) | |
| 7976 goto pop; /* do next alternative */ | |
| 7977 } while (badword[++bi] != NUL); | |
| 7978 minscore = score; | |
| 7979 } | |
| 7980 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 7981 { | |
| 7982 do | |
| 7983 { | |
| 7984 if ((score += SCORE_INS) >= minscore) | |
| 7985 goto pop; /* do next alternative */ | |
| 7986 } while (goodword[++gi] != NUL); | |
| 7987 minscore = score; | |
| 7988 } | |
| 7989 else /* both words continue */ | |
| 7990 { | |
| 7991 /* If not close to the limit, perform a change. Only try changes | |
| 7992 * that may lead to a lower score than "minscore". | |
| 7993 * round 0: try deleting a char from badword | |
| 7994 * round 1: try inserting a char in badword */ | |
| 7995 for (round = 0; round <= 1; ++round) | |
| 7996 { | |
| 7997 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 7998 if (score_off < minscore) | |
| 7999 { | |
| 8000 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 8001 { | |
| 8002 /* Near the limit, rest of the words must match. We | |
| 8003 * can check that right now, no need to push an item | |
| 8004 * onto the stack. */ | |
| 8005 bi2 = bi + 1 - round; | |
| 8006 gi2 = gi + round; | |
| 8007 while (goodword[gi2] == badword[bi2]) | |
| 8008 { | |
| 8009 if (goodword[gi2] == NUL) | |
| 8010 { | |
| 8011 minscore = score_off; | |
| 8012 break; | |
| 8013 } | |
| 8014 ++bi2; | |
| 8015 ++gi2; | |
| 8016 } | |
| 8017 } | |
| 8018 else | |
| 8019 { | |
| 8020 /* try deleting/inserting a character later */ | |
| 8021 stack[stackidx].badi = bi + 1 - round; | |
| 8022 stack[stackidx].goodi = gi + round; | |
| 8023 stack[stackidx].score = score_off; | |
| 8024 ++stackidx; | |
| 8025 } | |
| 8026 } | |
| 8027 } | |
| 8028 | |
| 8029 if (score + SCORE_SWAP < minscore) | |
| 8030 { | |
| 8031 /* If swapping two characters makes a match then the | |
| 8032 * substitution is more expensive, thus there is no need to | |
| 8033 * try both. */ | |
| 8034 if (gc == badword[bi + 1] && bc == goodword[gi + 1]) | |
| 8035 { | |
| 8036 /* Swap two characters, that is: skip them. */ | |
| 8037 gi += 2; | |
| 8038 bi += 2; | |
| 8039 score += SCORE_SWAP; | |
| 8040 continue; | |
| 8041 } | |
| 8042 } | |
| 8043 | |
| 8044 /* Substitute one character for another which is the same | |
| 8045 * thing as deleting a character from both goodword and badword. | |
| 8046 * Use a better score when there is only a case difference. */ | |
| 8047 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 8048 score += SCORE_ICASE; | |
| 8049 else | |
| 8050 { | |
| 8051 /* For a similar character use SCORE_SIMILAR. */ | |
| 8052 if (slang != NULL | |
| 8053 && slang->sl_has_map | |
| 8054 && similar_chars(slang, gc, bc)) | |
| 8055 score += SCORE_SIMILAR; | |
| 8056 else | |
| 8057 score += SCORE_SUBST; | |
| 8058 } | |
| 8059 | |
| 8060 if (score < minscore) | |
| 8061 { | |
| 8062 /* Do the substitution. */ | |
| 8063 ++gi; | |
| 8064 ++bi; | |
| 8065 continue; | |
| 8066 } | |
| 8067 } | |
| 8068 pop: | |
| 8069 /* | |
| 8070 * Get here to try the next alternative, pop it from the stack. | |
| 8071 */ | |
| 8072 if (stackidx == 0) /* stack is empty, finished */ | |
| 8073 break; | |
| 8074 | |
| 8075 /* pop an item from the stack */ | |
| 8076 --stackidx; | |
| 8077 gi = stack[stackidx].goodi; | |
| 8078 bi = stack[stackidx].badi; | |
| 8079 score = stack[stackidx].score; | |
| 8080 } | |
| 8081 | |
| 8082 /* When the score goes over "limit" it may actually be much higher. | |
| 8083 * Return a very large number to avoid going below the limit when giving a | |
| 8084 * bonus. */ | |
| 8085 if (minscore > limit) | |
| 8086 return SCORE_MAXMAX; | |
| 8087 return minscore; | |
| 8088 } | |
| 8089 | |
| 8090 /* | |
| 8091 * Multi-byte version of spell_edit_score_limit(). | |
| 8092 * Keep it in sync with the above! | |
| 8093 */ | |
| 8094 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8095 spell_edit_score_limit_w( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8096 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8097 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8098 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8099 int limit) |
| 625 | 8100 { |
| 8101 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 8102 int stackidx; | |
| 8103 int bi, gi; | |
| 8104 int bi2, gi2; | |
| 8105 int bc, gc; | |
| 8106 int score; | |
| 8107 int score_off; | |
| 8108 int minscore; | |
| 8109 int round; | |
| 8110 char_u *p; | |
| 8111 int wbadword[MAXWLEN]; | |
| 8112 int wgoodword[MAXWLEN]; | |
| 8113 | |
| 8114 /* Get the characters from the multi-byte strings and put them in an | |
| 8115 * int array for easy access. */ | |
| 8116 bi = 0; | |
| 8117 for (p = badword; *p != NUL; ) | |
| 8118 wbadword[bi++] = mb_cptr2char_adv(&p); | |
| 8119 wbadword[bi++] = 0; | |
| 8120 gi = 0; | |
| 8121 for (p = goodword; *p != NUL; ) | |
| 8122 wgoodword[gi++] = mb_cptr2char_adv(&p); | |
| 8123 wgoodword[gi++] = 0; | |
| 8124 | |
| 8125 /* | |
| 8126 * The idea is to go from start to end over the words. So long as | |
| 8127 * characters are equal just continue, this always gives the lowest score. | |
| 8128 * When there is a difference try several alternatives. Each alternative | |
| 8129 * increases "score" for the edit distance. Some of the alternatives are | |
| 8130 * pushed unto a stack and tried later, some are tried right away. At the | |
| 8131 * end of the word the score for one alternative is known. The lowest | |
| 8132 * possible score is stored in "minscore". | |
| 8133 */ | |
| 8134 stackidx = 0; | |
| 8135 bi = 0; | |
| 8136 gi = 0; | |
| 8137 score = 0; | |
| 8138 minscore = limit + 1; | |
| 8139 | |
| 8140 for (;;) | |
| 8141 { | |
| 8142 /* Skip over an equal part, score remains the same. */ | |
| 8143 for (;;) | |
| 8144 { | |
| 8145 bc = wbadword[bi]; | |
| 8146 gc = wgoodword[gi]; | |
| 8147 | |
| 8148 if (bc != gc) /* stop at a char that's different */ | |
| 8149 break; | |
| 8150 if (bc == NUL) /* both words end */ | |
| 8151 { | |
| 8152 if (score < minscore) | |
| 8153 minscore = score; | |
| 8154 goto pop; /* do next alternative */ | |
| 8155 } | |
| 8156 ++bi; | |
| 8157 ++gi; | |
| 8158 } | |
| 8159 | |
| 8160 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 8161 { | |
| 8162 do | |
| 8163 { | |
| 8164 if ((score += SCORE_DEL) >= minscore) | |
| 8165 goto pop; /* do next alternative */ | |
| 8166 } while (wbadword[++bi] != NUL); | |
| 8167 minscore = score; | |
| 8168 } | |
| 8169 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 8170 { | |
| 8171 do | |
| 8172 { | |
| 8173 if ((score += SCORE_INS) >= minscore) | |
| 8174 goto pop; /* do next alternative */ | |
| 8175 } while (wgoodword[++gi] != NUL); | |
| 8176 minscore = score; | |
| 8177 } | |
| 8178 else /* both words continue */ | |
| 8179 { | |
| 8180 /* If not close to the limit, perform a change. Only try changes | |
| 8181 * that may lead to a lower score than "minscore". | |
| 8182 * round 0: try deleting a char from badword | |
| 8183 * round 1: try inserting a char in badword */ | |
| 8184 for (round = 0; round <= 1; ++round) | |
| 8185 { | |
| 8186 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 8187 if (score_off < minscore) | |
| 8188 { | |
| 8189 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 8190 { | |
| 8191 /* Near the limit, rest of the words must match. We | |
| 8192 * can check that right now, no need to push an item | |
| 8193 * onto the stack. */ | |
| 8194 bi2 = bi + 1 - round; | |
| 8195 gi2 = gi + round; | |
| 8196 while (wgoodword[gi2] == wbadword[bi2]) | |
| 8197 { | |
| 8198 if (wgoodword[gi2] == NUL) | |
| 8199 { | |
| 8200 minscore = score_off; | |
| 8201 break; | |
| 8202 } | |
| 8203 ++bi2; | |
| 8204 ++gi2; | |
| 8205 } | |
| 8206 } | |
| 8207 else | |
| 8208 { | |
| 8209 /* try deleting a character from badword later */ | |
| 8210 stack[stackidx].badi = bi + 1 - round; | |
| 8211 stack[stackidx].goodi = gi + round; | |
| 8212 stack[stackidx].score = score_off; | |
| 8213 ++stackidx; | |
| 8214 } | |
| 8215 } | |
| 8216 } | |
| 8217 | |
| 8218 if (score + SCORE_SWAP < minscore) | |
| 8219 { | |
| 8220 /* If swapping two characters makes a match then the | |
| 8221 * substitution is more expensive, thus there is no need to | |
| 8222 * try both. */ | |
| 8223 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1]) | |
| 8224 { | |
| 8225 /* Swap two characters, that is: skip them. */ | |
| 8226 gi += 2; | |
| 8227 bi += 2; | |
| 8228 score += SCORE_SWAP; | |
| 8229 continue; | |
| 8230 } | |
| 8231 } | |
| 8232 | |
| 8233 /* Substitute one character for another which is the same | |
| 8234 * thing as deleting a character from both goodword and badword. | |
| 8235 * Use a better score when there is only a case difference. */ | |
| 8236 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 8237 score += SCORE_ICASE; | |
| 8238 else | |
| 8239 { | |
| 8240 /* For a similar character use SCORE_SIMILAR. */ | |
| 8241 if (slang != NULL | |
| 8242 && slang->sl_has_map | |
| 8243 && similar_chars(slang, gc, bc)) | |
| 8244 score += SCORE_SIMILAR; | |
| 8245 else | |
| 8246 score += SCORE_SUBST; | |
| 8247 } | |
| 8248 | |
| 8249 if (score < minscore) | |
| 8250 { | |
| 8251 /* Do the substitution. */ | |
| 8252 ++gi; | |
| 8253 ++bi; | |
| 8254 continue; | |
| 8255 } | |
| 8256 } | |
| 8257 pop: | |
| 8258 /* | |
| 8259 * Get here to try the next alternative, pop it from the stack. | |
| 8260 */ | |
| 8261 if (stackidx == 0) /* stack is empty, finished */ | |
| 8262 break; | |
| 8263 | |
| 8264 /* pop an item from the stack */ | |
| 8265 --stackidx; | |
| 8266 gi = stack[stackidx].goodi; | |
| 8267 bi = stack[stackidx].badi; | |
| 8268 score = stack[stackidx].score; | |
| 8269 } | |
| 8270 | |
| 8271 /* When the score goes over "limit" it may actually be much higher. | |
| 8272 * Return a very large number to avoid going below the limit when giving a | |
| 8273 * bonus. */ | |
| 8274 if (minscore > limit) | |
| 8275 return SCORE_MAXMAX; | |
| 8276 return minscore; | |
| 8277 } | |
| 8278 | |
| 714 | 8279 /* |
| 8280 * ":spellinfo" | |
| 8281 */ | |
| 8282 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8283 ex_spellinfo(exarg_T *eap UNUSED) |
| 714 | 8284 { |
| 8285 int lpi; | |
| 8286 langp_T *lp; | |
| 8287 char_u *p; | |
| 8288 | |
| 8289 if (no_spell_checking(curwin)) | |
| 8290 return; | |
| 8291 | |
| 8292 msg_start(); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8293 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
|
8294 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8295 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
|
15543
dd725a8ab112
patch 8.1.0779: argument for message functions is inconsistent
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
8296 msg_puts("file: "); |
|
dd725a8ab112
patch 8.1.0779: argument for message functions is inconsistent
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
8297 msg_puts((char *)lp->lp_slang->sl_fname); |
| 714 | 8298 msg_putchar('\n'); |
| 8299 p = lp->lp_slang->sl_info; | |
| 8300 if (p != NULL) | |
| 8301 { | |
|
15543
dd725a8ab112
patch 8.1.0779: argument for message functions is inconsistent
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
8302 msg_puts((char *)p); |
| 714 | 8303 msg_putchar('\n'); |
| 8304 } | |
| 8305 } | |
| 8306 msg_end(); | |
| 8307 } | |
| 8308 | |
| 625 | 8309 #define DUMPFLAG_KEEPCASE 1 /* round 2: keep-case tree */ |
| 8310 #define DUMPFLAG_COUNT 2 /* include word count */ | |
| 701 | 8311 #define DUMPFLAG_ICASE 4 /* ignore case when finding matches */ |
| 710 | 8312 #define DUMPFLAG_ONECAP 8 /* pattern starts with capital */ |
| 8313 #define DUMPFLAG_ALLCAP 16 /* pattern is all capitals */ | |
| 625 | 8314 |
| 351 | 8315 /* |
| 8316 * ":spelldump" | |
| 8317 */ | |
| 8318 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8319 ex_spelldump(exarg_T *eap) |
| 351 | 8320 { |
| 5382 | 8321 char_u *spl; |
| 8322 long dummy; | |
| 8323 | |
| 701 | 8324 if (no_spell_checking(curwin)) |
| 8325 return; | |
| 5382 | 8326 get_option_value((char_u*)"spl", &dummy, &spl, OPT_LOCAL); |
| 8327 | |
| 8328 /* Create a new empty buffer in a new window. */ | |
| 701 | 8329 do_cmdline_cmd((char_u *)"new"); |
| 5382 | 8330 |
| 8331 /* enable spelling locally in the new window */ | |
| 8332 set_option_value((char_u*)"spell", TRUE, (char_u*)"", OPT_LOCAL); | |
|
7447
ad432f8f68fb
commit https://github.com/vim/vim/commit/887c1fea4a114e7170091942d0446c8882701b5b
Christian Brabandt <cb@256bit.org>
parents:
6949
diff
changeset
|
8333 set_option_value((char_u*)"spl", dummy, spl, OPT_LOCAL); |
| 5382 | 8334 vim_free(spl); |
| 8335 | |
|
11121
778c10516955
patch 8.0.0448: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
10950
diff
changeset
|
8336 if (!BUFEMPTY()) |
| 701 | 8337 return; |
| 8338 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8339 spell_dump_compl(NULL, 0, NULL, eap->forceit ? DUMPFLAG_COUNT : 0); |
| 701 | 8340 |
| 8341 /* Delete the empty line that we started with. */ | |
| 8342 if (curbuf->b_ml.ml_line_count > 1) | |
| 8343 ml_delete(curbuf->b_ml.ml_line_count, FALSE); | |
| 8344 | |
| 8345 redraw_later(NOT_VALID); | |
| 8346 } | |
| 8347 | |
| 8348 /* | |
| 8349 * Go through all possible words and: | |
| 8350 * 1. When "pat" is NULL: dump a list of all words in the current buffer. | |
| 8351 * "ic" and "dir" are not used. | |
| 8352 * 2. When "pat" is not NULL: add matching words to insert mode completion. | |
| 8353 */ | |
| 8354 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8355 spell_dump_compl( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8356 char_u *pat, /* leading part of the word */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8357 int ic, /* ignore case */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8358 int *dir, /* direction for adding matches */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8359 int dumpflags_arg) /* DUMPFLAG_* */ |
| 701 | 8360 { |
| 351 | 8361 langp_T *lp; |
| 8362 slang_T *slang; | |
| 8363 idx_T arridx[MAXWLEN]; | |
| 8364 int curi[MAXWLEN]; | |
| 8365 char_u word[MAXWLEN]; | |
| 8366 int c; | |
| 8367 char_u *byts; | |
| 8368 idx_T *idxs; | |
| 8369 linenr_T lnum = 0; | |
| 8370 int round; | |
| 8371 int depth; | |
| 8372 int n; | |
| 8373 int flags; | |
| 381 | 8374 char_u *region_names = NULL; /* region names being used */ |
| 8375 int do_region = TRUE; /* dump region names and numbers */ | |
| 8376 char_u *p; | |
| 500 | 8377 int lpi; |
| 701 | 8378 int dumpflags = dumpflags_arg; |
| 8379 int patlen; | |
| 8380 | |
| 710 | 8381 /* When ignoring case or when the pattern starts with capital pass this on |
| 8382 * to dump_word(). */ | |
| 8383 if (pat != NULL) | |
| 8384 { | |
| 8385 if (ic) | |
| 8386 dumpflags |= DUMPFLAG_ICASE; | |
| 8387 else | |
| 8388 { | |
| 8389 n = captype(pat, NULL); | |
| 8390 if (n == WF_ONECAP) | |
| 8391 dumpflags |= DUMPFLAG_ONECAP; | |
|
15605
62b3805506b3
patch 8.1.0810: too many #ifdefs
Bram Moolenaar <Bram@vim.org>
parents:
15543
diff
changeset
|
8392 else if (n == WF_ALLCAP && (int)STRLEN(pat) > mb_ptr2len(pat)) |
| 710 | 8393 dumpflags |= DUMPFLAG_ALLCAP; |
| 8394 } | |
| 8395 } | |
| 351 | 8396 |
| 381 | 8397 /* Find out if we can support regions: All languages must support the same |
| 8398 * regions or none at all. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8399 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
|
8400 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8401 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 381 | 8402 p = lp->lp_slang->sl_regions; |
| 8403 if (p[0] != 0) | |
| 8404 { | |
| 8405 if (region_names == NULL) /* first language with regions */ | |
| 8406 region_names = p; | |
| 8407 else if (STRCMP(region_names, p) != 0) | |
| 8408 { | |
| 8409 do_region = FALSE; /* region names are different */ | |
| 8410 break; | |
| 8411 } | |
| 8412 } | |
| 8413 } | |
| 8414 | |
| 8415 if (do_region && region_names != NULL) | |
| 8416 { | |
| 701 | 8417 if (pat == NULL) |
| 8418 { | |
| 8419 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names); | |
| 8420 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 8421 } | |
| 381 | 8422 } |
| 8423 else | |
| 8424 do_region = FALSE; | |
| 8425 | |
| 8426 /* | |
| 8427 * Loop over all files loaded for the entries in 'spelllang'. | |
| 8428 */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8429 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
|
8430 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8431 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 351 | 8432 slang = lp->lp_slang; |
| 500 | 8433 if (slang->sl_fbyts == NULL) /* reloading failed */ |
| 8434 continue; | |
| 351 | 8435 |
| 701 | 8436 if (pat == NULL) |
| 8437 { | |
| 8438 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname); | |
| 8439 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 8440 } | |
| 8441 | |
| 8442 /* When matching with a pattern and there are no prefixes only use | |
| 8443 * parts of the tree that match "pat". */ | |
| 8444 if (pat != NULL && slang->sl_pbyts == NULL) | |
| 835 | 8445 patlen = (int)STRLEN(pat); |
| 701 | 8446 else |
| 840 | 8447 patlen = -1; |
| 351 | 8448 |
| 8449 /* round 1: case-folded tree | |
| 8450 * round 2: keep-case tree */ | |
| 8451 for (round = 1; round <= 2; ++round) | |
| 8452 { | |
| 8453 if (round == 1) | |
| 8454 { | |
| 701 | 8455 dumpflags &= ~DUMPFLAG_KEEPCASE; |
| 351 | 8456 byts = slang->sl_fbyts; |
| 8457 idxs = slang->sl_fidxs; | |
| 8458 } | |
| 8459 else | |
| 8460 { | |
| 701 | 8461 dumpflags |= DUMPFLAG_KEEPCASE; |
| 351 | 8462 byts = slang->sl_kbyts; |
| 8463 idxs = slang->sl_kidxs; | |
| 8464 } | |
| 8465 if (byts == NULL) | |
| 8466 continue; /* array is empty */ | |
| 8467 | |
| 8468 depth = 0; | |
| 8469 arridx[0] = 0; | |
| 8470 curi[0] = 1; | |
| 701 | 8471 while (depth >= 0 && !got_int |
|
16142
570a296aa0b4
patch 8.1.1076: file for Insert mode is much too big
Bram Moolenaar <Bram@vim.org>
parents:
15636
diff
changeset
|
8472 && (pat == NULL || !ins_compl_interrupted())) |
| 351 | 8473 { |
| 8474 if (curi[depth] > byts[arridx[depth]]) | |
| 8475 { | |
| 8476 /* Done all bytes at this node, go up one level. */ | |
| 8477 --depth; | |
| 8478 line_breakcheck(); | |
|
10277
154d5a2e7395
commit https://github.com/vim/vim/commit/472e85970ee3a80abd824bef510df12e9cfe9e96
Christian Brabandt <cb@256bit.org>
parents:
10042
diff
changeset
|
8479 ins_compl_check_keys(50, FALSE); |
| 351 | 8480 } |
| 8481 else | |
| 8482 { | |
| 8483 /* Do one more byte at this node. */ | |
| 8484 n = arridx[depth] + curi[depth]; | |
| 8485 ++curi[depth]; | |
| 8486 c = byts[n]; | |
| 8487 if (c == 0) | |
| 8488 { | |
| 8489 /* End of word, deal with the word. | |
| 8490 * Don't use keep-case words in the fold-case tree, | |
| 8491 * they will appear in the keep-case tree. | |
| 8492 * Only use the word when the region matches. */ | |
| 8493 flags = (int)idxs[n]; | |
| 8494 if ((round == 2 || (flags & WF_KEEPCAP) == 0) | |
| 500 | 8495 && (flags & WF_NEEDCOMP) == 0 |
| 381 | 8496 && (do_region |
| 8497 || (flags & WF_REGION) == 0 | |
| 390 | 8498 || (((unsigned)flags >> 16) |
| 351 | 8499 & lp->lp_region) != 0)) |
| 8500 { | |
| 8501 word[depth] = NUL; | |
| 381 | 8502 if (!do_region) |
| 8503 flags &= ~WF_REGION; | |
| 355 | 8504 |
| 8505 /* Dump the basic word if there is no prefix or | |
| 8506 * when it's the first one. */ | |
| 390 | 8507 c = (unsigned)flags >> 24; |
| 355 | 8508 if (c == 0 || curi[depth] == 2) |
| 701 | 8509 { |
| 8510 dump_word(slang, word, pat, dir, | |
| 8511 dumpflags, flags, lnum); | |
| 8512 if (pat == NULL) | |
| 8513 ++lnum; | |
| 8514 } | |
| 351 | 8515 |
| 8516 /* Apply the prefix, if there is one. */ | |
| 355 | 8517 if (c != 0) |
| 701 | 8518 lnum = dump_prefixes(slang, word, pat, dir, |
| 8519 dumpflags, flags, lnum); | |
| 351 | 8520 } |
| 8521 } | |
| 8522 else | |
| 8523 { | |
| 8524 /* Normal char, go one level deeper. */ | |
| 8525 word[depth++] = c; | |
| 8526 arridx[depth] = idxs[n]; | |
| 8527 curi[depth] = 1; | |
| 701 | 8528 |
| 8529 /* Check if this characters matches with the pattern. | |
| 8530 * If not skip the whole tree below it. | |
| 710 | 8531 * Always ignore case here, dump_word() will check |
| 8532 * proper case later. This isn't exactly right when | |
| 8533 * length changes for multi-byte characters with | |
| 8534 * ignore case... */ | |
| 8535 if (depth <= patlen | |
| 8536 && MB_STRNICMP(word, pat, depth) != 0) | |
| 701 | 8537 --depth; |
| 351 | 8538 } |
| 8539 } | |
| 8540 } | |
| 8541 } | |
| 8542 } | |
| 8543 } | |
| 8544 | |
| 8545 /* | |
| 8546 * Dump one word: apply case modifications and append a line to the buffer. | |
| 701 | 8547 * When "lnum" is zero add insert mode completion. |
| 8548 */ | |
| 8549 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8550 dump_word( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8551 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8552 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8553 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8554 int *dir, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8555 int dumpflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8556 int wordflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8557 linenr_T lnum) |
| 351 | 8558 { |
| 8559 int keepcap = FALSE; | |
| 8560 char_u *p; | |
| 625 | 8561 char_u *tw; |
| 351 | 8562 char_u cword[MAXWLEN]; |
| 381 | 8563 char_u badword[MAXWLEN + 10]; |
| 8564 int i; | |
| 710 | 8565 int flags = wordflags; |
| 8566 | |
| 8567 if (dumpflags & DUMPFLAG_ONECAP) | |
| 8568 flags |= WF_ONECAP; | |
| 8569 if (dumpflags & DUMPFLAG_ALLCAP) | |
| 8570 flags |= WF_ALLCAP; | |
| 351 | 8571 |
| 625 | 8572 if ((dumpflags & DUMPFLAG_KEEPCASE) == 0 && (flags & WF_CAPMASK) != 0) |
| 351 | 8573 { |
| 8574 /* Need to fix case according to "flags". */ | |
| 8575 make_case_word(word, cword, flags); | |
| 8576 p = cword; | |
| 8577 } | |
| 8578 else | |
| 8579 { | |
| 8580 p = word; | |
| 625 | 8581 if ((dumpflags & DUMPFLAG_KEEPCASE) |
| 8582 && ((captype(word, NULL) & WF_KEEPCAP) == 0 | |
| 389 | 8583 || (flags & WF_FIXCAP) != 0)) |
| 351 | 8584 keepcap = TRUE; |
| 8585 } | |
| 625 | 8586 tw = p; |
| 351 | 8587 |
| 701 | 8588 if (pat == NULL) |
| 8589 { | |
| 8590 /* Add flags and regions after a slash. */ | |
| 8591 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap) | |
| 8592 { | |
| 8593 STRCPY(badword, p); | |
| 8594 STRCAT(badword, "/"); | |
| 8595 if (keepcap) | |
| 8596 STRCAT(badword, "="); | |
| 8597 if (flags & WF_BANNED) | |
| 8598 STRCAT(badword, "!"); | |
| 8599 else if (flags & WF_RARE) | |
| 8600 STRCAT(badword, "?"); | |
| 8601 if (flags & WF_REGION) | |
| 8602 for (i = 0; i < 7; ++i) | |
| 8603 if (flags & (0x10000 << i)) | |
| 8604 sprintf((char *)badword + STRLEN(badword), "%d", i + 1); | |
| 8605 p = badword; | |
| 8606 } | |
| 8607 | |
| 8608 if (dumpflags & DUMPFLAG_COUNT) | |
| 8609 { | |
| 8610 hashitem_T *hi; | |
| 8611 | |
| 8612 /* Include the word count for ":spelldump!". */ | |
| 8613 hi = hash_find(&slang->sl_wordcount, tw); | |
| 8614 if (!HASHITEM_EMPTY(hi)) | |
| 8615 { | |
| 8616 vim_snprintf((char *)IObuff, IOSIZE, "%s\t%d", | |
| 625 | 8617 tw, HI2WC(hi)->wc_count); |
| 701 | 8618 p = IObuff; |
| 8619 } | |
| 8620 } | |
| 8621 | |
| 8622 ml_append(lnum, p, (colnr_T)0, FALSE); | |
| 8623 } | |
| 710 | 8624 else if (((dumpflags & DUMPFLAG_ICASE) |
| 8625 ? MB_STRNICMP(p, pat, STRLEN(pat)) == 0 | |
| 8626 : STRNCMP(p, pat, STRLEN(pat)) == 0) | |
| 701 | 8627 && ins_compl_add_infercase(p, (int)STRLEN(p), |
|
16239
5df26b29e809
patch 8.1.1124: insert completion flags are mixed up
Bram Moolenaar <Bram@vim.org>
parents:
16162
diff
changeset
|
8628 p_ic, NULL, *dir, FALSE) == OK) |
| 710 | 8629 /* if dir was BACKWARD then honor it just once */ |
| 8630 *dir = FORWARD; | |
| 351 | 8631 } |
| 8632 | |
| 8633 /* | |
| 372 | 8634 * For ":spelldump": Find matching prefixes for "word". Prepend each to |
| 8635 * "word" and append a line to the buffer. | |
| 701 | 8636 * When "lnum" is zero add insert mode completion. |
| 351 | 8637 * Return the updated line number. |
| 8638 */ | |
| 8639 static linenr_T | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8640 dump_prefixes( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8641 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8642 char_u *word, /* case-folded word */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8643 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8644 int *dir, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8645 int dumpflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8646 int flags, /* flags with prefix ID */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8647 linenr_T startlnum) |
| 351 | 8648 { |
| 8649 idx_T arridx[MAXWLEN]; | |
| 8650 int curi[MAXWLEN]; | |
| 8651 char_u prefix[MAXWLEN]; | |
| 455 | 8652 char_u word_up[MAXWLEN]; |
| 8653 int has_word_up = FALSE; | |
| 351 | 8654 int c; |
| 8655 char_u *byts; | |
| 8656 idx_T *idxs; | |
| 8657 linenr_T lnum = startlnum; | |
| 8658 int depth; | |
| 8659 int n; | |
| 8660 int len; | |
| 8661 int i; | |
| 8662 | |
| 626 | 8663 /* If the word starts with a lower-case letter make the word with an |
| 455 | 8664 * upper-case letter in word_up[]. */ |
| 8665 c = PTR2CHAR(word); | |
| 8666 if (SPELL_TOUPPER(c) != c) | |
| 8667 { | |
| 8668 onecap_copy(word, word_up, TRUE); | |
| 8669 has_word_up = TRUE; | |
| 8670 } | |
| 8671 | |
| 351 | 8672 byts = slang->sl_pbyts; |
| 8673 idxs = slang->sl_pidxs; | |
| 8674 if (byts != NULL) /* array not is empty */ | |
| 8675 { | |
| 8676 /* | |
| 8677 * Loop over all prefixes, building them byte-by-byte in prefix[]. | |
| 390 | 8678 * When at the end of a prefix check that it supports "flags". |
| 351 | 8679 */ |
| 8680 depth = 0; | |
| 8681 arridx[0] = 0; | |
| 8682 curi[0] = 1; | |
| 8683 while (depth >= 0 && !got_int) | |
| 8684 { | |
| 390 | 8685 n = arridx[depth]; |
| 8686 len = byts[n]; | |
| 8687 if (curi[depth] > len) | |
| 351 | 8688 { |
| 8689 /* Done all bytes at this node, go up one level. */ | |
| 8690 --depth; | |
| 8691 line_breakcheck(); | |
| 8692 } | |
| 8693 else | |
| 8694 { | |
| 8695 /* Do one more byte at this node. */ | |
| 390 | 8696 n += curi[depth]; |
| 351 | 8697 ++curi[depth]; |
| 8698 c = byts[n]; | |
| 8699 if (c == 0) | |
| 8700 { | |
| 8701 /* End of prefix, find out how many IDs there are. */ | |
| 8702 for (i = 1; i < len; ++i) | |
| 8703 if (byts[n + i] != 0) | |
| 8704 break; | |
| 8705 curi[depth] += i - 1; | |
| 8706 | |
| 455 | 8707 c = valid_word_prefix(i, n, flags, word, slang, FALSE); |
| 8708 if (c != 0) | |
| 351 | 8709 { |
| 376 | 8710 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1); |
| 701 | 8711 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 8712 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 8713 : flags, lnum); |
| 8714 if (lnum != 0) | |
| 8715 ++lnum; | |
| 351 | 8716 } |
| 455 | 8717 |
| 8718 /* Check for prefix that matches the word when the | |
| 8719 * first letter is upper-case, but only if the prefix has | |
| 8720 * a condition. */ | |
| 8721 if (has_word_up) | |
| 8722 { | |
| 8723 c = valid_word_prefix(i, n, flags, word_up, slang, | |
| 8724 TRUE); | |
| 8725 if (c != 0) | |
| 8726 { | |
| 8727 vim_strncpy(prefix + depth, word_up, | |
| 8728 MAXWLEN - depth - 1); | |
| 701 | 8729 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 8730 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 8731 : flags, lnum); |
| 8732 if (lnum != 0) | |
| 8733 ++lnum; | |
| 455 | 8734 } |
| 8735 } | |
| 351 | 8736 } |
| 8737 else | |
| 8738 { | |
| 8739 /* Normal char, go one level deeper. */ | |
| 8740 prefix[depth++] = c; | |
| 8741 arridx[depth] = idxs[n]; | |
| 8742 curi[depth] = 1; | |
| 8743 } | |
| 8744 } | |
| 8745 } | |
| 8746 } | |
| 8747 | |
| 8748 return lnum; | |
| 8749 } | |
| 8750 | |
| 498 | 8751 /* |
| 626 | 8752 * Move "p" to the end of word "start". |
| 8753 * Uses the spell-checking word characters. | |
| 498 | 8754 */ |
| 8755 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8756 spell_to_word_end(char_u *start, win_T *win) |
| 498 | 8757 { |
| 8758 char_u *p = start; | |
| 8759 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8760 while (*p != NUL && spell_iswordp(p, win)) |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
8761 MB_PTR_ADV(p); |
| 498 | 8762 return p; |
| 8763 } | |
| 8764 | |
| 475 | 8765 #if defined(FEAT_INS_EXPAND) || defined(PROTO) |
| 8766 /* | |
| 626 | 8767 * For Insert mode completion CTRL-X s: |
| 8768 * Find start of the word in front of column "startcol". | |
| 8769 * We don't check if it is badly spelled, with completion we can only change | |
| 8770 * the word in front of the cursor. | |
| 475 | 8771 * Returns the column number of the word. |
| 8772 */ | |
| 8773 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8774 spell_word_start(int startcol) |
| 475 | 8775 { |
| 8776 char_u *line; | |
| 8777 char_u *p; | |
| 8778 int col = 0; | |
| 8779 | |
| 498 | 8780 if (no_spell_checking(curwin)) |
| 475 | 8781 return startcol; |
| 8782 | |
| 8783 /* Find a word character before "startcol". */ | |
| 8784 line = ml_get_curline(); | |
| 8785 for (p = line + startcol; p > line; ) | |
| 8786 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
8787 MB_PTR_BACK(line, p); |
| 5477 | 8788 if (spell_iswordp_nmw(p, curwin)) |
| 475 | 8789 break; |
| 8790 } | |
| 8791 | |
| 8792 /* Go back to start of the word. */ | |
| 8793 while (p > line) | |
| 8794 { | |
| 835 | 8795 col = (int)(p - line); |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
8796 MB_PTR_BACK(line, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8797 if (!spell_iswordp(p, curwin)) |
| 475 | 8798 break; |
| 8799 col = 0; | |
| 8800 } | |
| 8801 | |
| 535 | 8802 return col; |
| 8803 } | |
| 8804 | |
| 8805 /* | |
| 8806 * Need to check for 'spellcapcheck' now, the word is removed before | |
| 8807 * expand_spelling() is called. Therefore the ugly global variable. | |
| 8808 */ | |
| 8809 static int spell_expand_need_cap; | |
| 8810 | |
| 8811 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8812 spell_expand_check_cap(colnr_T col) |
| 535 | 8813 { |
| 475 | 8814 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col); |
| 8815 } | |
| 8816 | |
| 8817 /* | |
| 8818 * Get list of spelling suggestions. | |
| 8819 * Used for Insert mode completion CTRL-X ?. | |
| 8820 * Returns the number of matches. The matches are in "matchp[]", array of | |
| 8821 * allocated strings. | |
| 8822 */ | |
| 8823 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8824 expand_spelling( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8825 linenr_T lnum UNUSED, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8826 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8827 char_u ***matchp) |
| 475 | 8828 { |
| 8829 garray_T ga; | |
| 8830 | |
| 625 | 8831 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap, TRUE); |
| 475 | 8832 *matchp = ga.ga_data; |
| 8833 return ga.ga_len; | |
| 8834 } | |
| 8835 #endif | |
| 8836 | |
| 737 | 8837 #endif /* FEAT_SPELL */ |