Mercurial > vim
annotate src/spell.c @ 13244:ac42c4b11dbc v8.0.1496
patch 8.0.1496: clearing a pointer takes two lines
commit https://github.com/vim/vim/commit/d23a823669d93fb2a570a039173eefe4856ac806
Author: Bram Moolenaar <Bram@vim.org>
Date: Sat Feb 10 18:45:26 2018 +0100
patch 8.0.1496: clearing a pointer takes two lines
Problem: Clearing a pointer takes two lines.
Solution: Add VIM_CLEAR() and replace vim_clear(). (Hirohito Higashi,
closes #2629)
| author | Christian Brabandt <cb@256bit.org> |
|---|---|
| date | Sat, 10 Feb 2018 19:00:07 +0100 |
| parents | e769c912fcd9 |
| children | ae312df8d0ab |
| 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 #ifdef FEAT_MBYTE |
| 228 int mi_cprefixlen; /* byte length of prefix in original | |
| 229 case */ | |
| 230 #else | |
| 231 # define mi_cprefixlen mi_prefixlen /* it's the same value */ | |
| 232 #endif | |
| 339 | 233 |
| 481 | 234 /* for when checking a compound word */ |
| 235 int mi_compoff; /* start of following word offset */ | |
| 483 | 236 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */ |
| 237 int mi_complen; /* nr of compound words used */ | |
| 809 | 238 int mi_compextra; /* nr of COMPOUNDROOT words */ |
| 481 | 239 |
| 243 | 240 /* others */ |
| 236 | 241 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */ |
| 300 | 242 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
|
243 win_T *mi_win; /* buffer being checked */ |
| 492 | 244 |
| 245 /* for NOBREAK */ | |
| 246 int mi_result2; /* "mi_resul" without following word */ | |
| 247 char_u *mi_end2; /* "mi_end" without following word */ | |
| 236 | 248 } matchinf_T; |
| 249 | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
250 |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
251 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
|
252 #ifdef FEAT_MBYTE |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
253 static int spell_mb_isword_class(int cl, win_T *wp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
254 static int spell_iswordp_w(int *p, win_T *wp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
255 #endif |
| 307 | 256 |
| 257 /* | |
| 344 | 258 * For finding suggestions: At each node in the tree these states are tried: |
| 330 | 259 */ |
| 260 typedef enum | |
| 261 { | |
| 344 | 262 STATE_START = 0, /* At start of node check for NUL bytes (goodword |
| 263 * ends); if badword ends there is a match, otherwise | |
| 264 * try splitting word. */ | |
| 375 | 265 STATE_NOPREFIX, /* try without prefix */ |
| 344 | 266 STATE_SPLITUNDO, /* Undo splitting. */ |
| 330 | 267 STATE_ENDNUL, /* Past NUL bytes at start of the node. */ |
| 268 STATE_PLAIN, /* Use each byte of the node. */ | |
| 269 STATE_DEL, /* Delete a byte from the bad word. */ | |
| 625 | 270 STATE_INS_PREP, /* Prepare for inserting bytes. */ |
| 330 | 271 STATE_INS, /* Insert a byte in the bad word. */ |
| 272 STATE_SWAP, /* Swap two bytes. */ | |
| 344 | 273 STATE_UNSWAP, /* Undo swap two characters. */ |
| 274 STATE_SWAP3, /* Swap two characters over three. */ | |
| 275 STATE_UNSWAP3, /* Undo Swap two characters over three. */ | |
| 276 STATE_UNROT3L, /* Undo rotate three characters left */ | |
| 277 STATE_UNROT3R, /* Undo rotate three characters right */ | |
| 330 | 278 STATE_REP_INI, /* Prepare for using REP items. */ |
| 279 STATE_REP, /* Use matching REP items from the .aff file. */ | |
| 280 STATE_REP_UNDO, /* Undo a REP item replacement. */ | |
| 281 STATE_FINAL /* End of this node. */ | |
| 282 } state_T; | |
| 283 | |
| 284 /* | |
| 346 | 285 * Struct to keep the state at each level in suggest_try_change(). |
| 323 | 286 */ |
| 287 typedef struct trystate_S | |
| 288 { | |
| 330 | 289 state_T ts_state; /* state at this level, STATE_ */ |
| 323 | 290 int ts_score; /* score */ |
| 344 | 291 idx_T ts_arridx; /* index in tree array, start of node */ |
| 330 | 292 short ts_curi; /* index in list of child nodes */ |
| 293 char_u ts_fidx; /* index in fword[], case-folded bad word */ | |
| 294 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */ | |
| 295 char_u ts_twordlen; /* valid length of tword[] */ | |
| 482 | 296 char_u ts_prefixdepth; /* stack depth for end of prefix or |
| 485 | 297 * PFD_PREFIXTREE or PFD_NOPREFIX */ |
| 298 char_u ts_flags; /* TSF_ flags */ | |
| 330 | 299 #ifdef FEAT_MBYTE |
| 300 char_u ts_tcharlen; /* number of bytes in tword character */ | |
| 301 char_u ts_tcharidx; /* current byte index in tword character */ | |
| 302 char_u ts_isdiff; /* DIFF_ values */ | |
| 303 char_u ts_fcharstart; /* index in fword where badword char started */ | |
| 304 #endif | |
| 483 | 305 char_u ts_prewordlen; /* length of word in "preword[]" */ |
| 306 char_u ts_splitoff; /* index in "tword" after last split */ | |
| 492 | 307 char_u ts_splitfidx; /* "ts_fidx" at word split */ |
| 483 | 308 char_u ts_complen; /* nr of compound words used */ |
| 485 | 309 char_u ts_compsplit; /* index for "compflags" where word was spit */ |
| 346 | 310 char_u ts_save_badflags; /* su_badflags saved here */ |
| 625 | 311 char_u ts_delidx; /* index in fword for char that was deleted, |
| 312 valid when "ts_flags" has TSF_DIDDEL */ | |
| 323 | 313 } trystate_T; |
| 314 | |
| 330 | 315 /* values for ts_isdiff */ |
| 316 #define DIFF_NONE 0 /* no different byte (yet) */ | |
| 317 #define DIFF_YES 1 /* different byte found */ | |
| 318 #define DIFF_INSERT 2 /* inserting character */ | |
| 319 | |
| 485 | 320 /* values for ts_flags */ |
| 321 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */ | |
| 322 #define TSF_DIDSPLIT 2 /* tried split at this point */ | |
| 625 | 323 #define TSF_DIDDEL 4 /* did a delete, "ts_delidx" has index */ |
| 485 | 324 |
| 375 | 325 /* special values ts_prefixdepth */ |
| 485 | 326 #define PFD_NOPREFIX 0xff /* not using prefixes */ |
| 482 | 327 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */ |
| 625 | 328 #define PFD_NOTSPECIAL 0xfd /* highest value that's not special */ |
| 375 | 329 |
| 339 | 330 /* mode values for find_word */ |
| 481 | 331 #define FIND_FOLDWORD 0 /* find word case-folded */ |
| 332 #define FIND_KEEPWORD 1 /* find keep-case word */ | |
| 333 #define FIND_PREFIX 2 /* find word after prefix */ | |
| 334 #define FIND_COMPOUND 3 /* find case-folded compound word */ | |
| 335 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */ | |
| 339 | 336 |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
337 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
|
338 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
|
339 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
|
340 static int can_be_compound(trystate_T *sp, slang_T *slang, char_u *compflags, int flag); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
341 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
|
342 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
|
343 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
|
344 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
|
345 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
|
346 static int no_spell_checking(win_T *wp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
347 static void spell_load_lang(char_u *lang); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
348 static void int_wordlist_spl(char_u *fname); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
349 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
|
350 static int score_wordcount_adj(slang_T *slang, int score, char_u *word, int split); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
351 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
|
352 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
|
353 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
|
354 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
|
355 static int badword_captype(char_u *word, char_u *end); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
356 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
|
357 static void spell_find_suggest(char_u *badptr, int badlen, suginfo_T *su, int maxcount, int banbadword, int need_cap, int interactive); |
| 372 | 358 #ifdef FEAT_EVAL |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
359 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
|
360 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
361 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
|
362 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
|
363 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
|
364 static void allcap_copy(char_u *word, char_u *wcopy); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
365 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
|
366 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
|
367 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
|
368 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
|
369 #ifdef FEAT_MBYTE |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
370 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
|
371 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
372 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
|
373 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
|
374 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
|
375 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
|
376 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
|
377 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
|
378 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
|
379 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
|
380 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
|
381 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
|
382 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
|
383 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
|
384 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
|
385 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
|
386 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
|
387 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
|
388 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
|
389 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
|
390 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
|
391 #ifdef FEAT_MBYTE |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
392 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
|
393 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
394 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
|
395 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
|
396 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
|
397 #ifdef FEAT_MBYTE |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
398 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
|
399 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
400 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
|
401 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
|
402 |
| 236 | 403 |
| 626 | 404 /* Remember what "z?" replaced. */ |
| 405 static char_u *repl_from = NULL; | |
| 406 static char_u *repl_to = NULL; | |
| 407 | |
| 236 | 408 /* |
| 409 * Main spell-checking function. | |
| 300 | 410 * "ptr" points to a character that could be the start of a word. |
| 534 | 411 * "*attrp" is set to the highlight index for a badly spelled word. For a |
| 412 * non-word or when it's OK it remains unchanged. | |
| 236 | 413 * This must only be called when 'spelllang' is not empty. |
| 323 | 414 * |
| 385 | 415 * "capcol" is used to check for a Capitalised word after the end of a |
| 416 * sentence. If it's zero then perform the check. Return the column where to | |
| 417 * check next, or -1 when no sentence end was found. If it's NULL then don't | |
| 418 * worry. | |
| 323 | 419 * |
| 236 | 420 * Returns the length of the word in bytes, also when it's OK, so that the |
| 421 * caller can skip over the word. | |
| 422 */ | |
| 423 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
424 spell_check( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
425 win_T *wp, /* current window */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
426 char_u *ptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
427 hlf_T *attrp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
428 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
|
429 int docount) /* count good words */ |
| 236 | 430 { |
| 431 matchinf_T mi; /* Most things are put in "mi" so that it can | |
| 432 be passed to functions quickly. */ | |
| 344 | 433 int nrlen = 0; /* found a number first */ |
| 385 | 434 int c; |
| 483 | 435 int wrongcaplen = 0; |
| 500 | 436 int lpi; |
| 625 | 437 int count_word = docount; |
| 236 | 438 |
| 307 | 439 /* A word never starts at a space or a control character. Return quickly |
| 440 * then, skipping over the character. */ | |
| 441 if (*ptr <= ' ') | |
| 442 return 1; | |
| 690 | 443 |
| 444 /* 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
|
445 if (wp->w_s->b_langp.ga_len == 0) |
| 690 | 446 return 1; |
| 447 | |
| 483 | 448 vim_memset(&mi, 0, sizeof(matchinf_T)); |
| 236 | 449 |
| 344 | 450 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and |
| 584 | 451 * 0X99FF. But always do check spelling to find "3GPP" and "11 |
| 452 * julifeest". */ | |
| 300 | 453 if (*ptr >= '0' && *ptr <= '9') |
| 454 { | |
|
7447
ad432f8f68fb
commit https://github.com/vim/vim/commit/887c1fea4a114e7170091942d0446c8882701b5b
Christian Brabandt <cb@256bit.org>
parents:
6949
diff
changeset
|
455 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
|
456 mi.mi_end = skipbin(ptr + 2); |
|
ad432f8f68fb
commit https://github.com/vim/vim/commit/887c1fea4a114e7170091942d0446c8882701b5b
Christian Brabandt <cb@256bit.org>
parents:
6949
diff
changeset
|
457 else if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X')) |
| 316 | 458 mi.mi_end = skiphex(ptr + 2); |
| 300 | 459 else |
| 344 | 460 mi.mi_end = skipdigits(ptr); |
| 835 | 461 nrlen = (int)(mi.mi_end - ptr); |
| 584 | 462 } |
| 346 | 463 |
| 464 /* Find the normal end of the word (until the next non-word character). */ | |
| 344 | 465 mi.mi_word = ptr; |
| 584 | 466 mi.mi_fend = ptr; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
467 if (spell_iswordp(mi.mi_fend, wp)) |
| 344 | 468 { |
| 469 do | |
| 470 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
471 MB_PTR_ADV(mi.mi_fend); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
472 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp)); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
473 |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
474 if (capcol != NULL && *capcol == 0 && wp->w_s->b_cap_prog != NULL) |
| 385 | 475 { |
| 476 /* Check word starting with capital letter. */ | |
| 455 | 477 c = PTR2CHAR(ptr); |
| 385 | 478 if (!SPELL_ISUPPER(c)) |
| 483 | 479 wrongcaplen = (int)(mi.mi_fend - ptr); |
| 385 | 480 } |
| 481 } | |
| 482 if (capcol != NULL) | |
| 483 *capcol = -1; | |
| 344 | 484 |
| 485 /* We always use the characters up to the next non-word character, | |
| 486 * also for bad words. */ | |
| 487 mi.mi_end = mi.mi_fend; | |
| 488 | |
| 489 /* Check caps type later. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
490 mi.mi_capflags = 0; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
491 mi.mi_cend = NULL; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
492 mi.mi_win = wp; |
| 344 | 493 |
| 483 | 494 /* case-fold the word with one non-word character, so that we can check |
| 495 * for the word end. */ | |
| 344 | 496 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
|
497 MB_PTR_ADV(mi.mi_fend); |
| 344 | 498 |
| 499 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword, | |
| 500 MAXWLEN + 1); | |
| 835 | 501 mi.mi_fwordlen = (int)STRLEN(mi.mi_fword); |
| 344 | 502 |
| 503 /* The word is bad unless we recognize it. */ | |
| 504 mi.mi_result = SP_BAD; | |
| 492 | 505 mi.mi_result2 = SP_BAD; |
| 344 | 506 |
| 507 /* | |
| 508 * Loop over the languages specified in 'spelllang'. | |
| 625 | 509 * We check them all, because a word may be matched longer in another |
| 510 * language. | |
| 344 | 511 */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
512 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
|
513 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
514 mi.mi_lp = LANGP_ENTRY(wp->w_s->b_langp, lpi); |
| 500 | 515 |
| 516 /* If reloading fails the language is still in the list but everything | |
| 517 * has been cleared. */ | |
| 518 if (mi.mi_lp->lp_slang->sl_fidxs == NULL) | |
| 519 continue; | |
| 520 | |
| 344 | 521 /* Check for a matching word in case-folded words. */ |
| 522 find_word(&mi, FIND_FOLDWORD); | |
| 523 | |
| 524 /* Check for a matching word in keep-case words. */ | |
| 525 find_word(&mi, FIND_KEEPWORD); | |
| 526 | |
| 527 /* Check for matching prefixes. */ | |
| 485 | 528 find_prefix(&mi, FIND_FOLDWORD); |
| 492 | 529 |
| 530 /* For a NOBREAK language, may want to use a word without a following | |
| 531 * word as a backup. */ | |
| 532 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD | |
| 533 && mi.mi_result2 != SP_BAD) | |
| 534 { | |
| 535 mi.mi_result = mi.mi_result2; | |
| 536 mi.mi_end = mi.mi_end2; | |
| 537 } | |
| 625 | 538 |
| 539 /* Count the word in the first language where it's found to be OK. */ | |
| 540 if (count_word && mi.mi_result == SP_OK) | |
| 541 { | |
| 542 count_common_word(mi.mi_lp->lp_slang, ptr, | |
| 543 (int)(mi.mi_end - ptr), 1); | |
| 544 count_word = FALSE; | |
| 545 } | |
| 344 | 546 } |
| 547 | |
| 548 if (mi.mi_result != SP_OK) | |
| 549 { | |
| 346 | 550 /* If we found a number skip over it. Allows for "42nd". Do flag |
| 551 * rare and local words, e.g., "3GPP". */ | |
| 344 | 552 if (nrlen > 0) |
| 346 | 553 { |
| 554 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED) | |
| 555 return nrlen; | |
| 556 } | |
| 344 | 557 |
| 558 /* When we are at a non-word character there is no error, just | |
| 559 * skip over the character (try looking for a word after it). */ | |
| 5477 | 560 else if (!spell_iswordp_nmw(ptr, wp)) |
| 243 | 561 { |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
562 if (capcol != NULL && wp->w_s->b_cap_prog != NULL) |
| 385 | 563 { |
| 564 regmatch_T regmatch; | |
| 6375 | 565 int r; |
| 385 | 566 |
| 567 /* Check for end of sentence. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
568 regmatch.regprog = wp->w_s->b_cap_prog; |
| 385 | 569 regmatch.rm_ic = FALSE; |
| 6375 | 570 r = vim_regexec(®match, ptr, 0); |
| 571 wp->w_s->b_cap_prog = regmatch.regprog; | |
| 572 if (r) | |
| 385 | 573 *capcol = (int)(regmatch.endp[0] - ptr); |
| 574 } | |
| 575 | |
| 344 | 576 #ifdef FEAT_MBYTE |
| 577 if (has_mbyte) | |
| 474 | 578 return (*mb_ptr2len)(ptr); |
| 344 | 579 #endif |
| 580 return 1; | |
| 300 | 581 } |
| 483 | 582 else if (mi.mi_end == ptr) |
| 583 /* Always include at least one character. Required for when there | |
| 584 * 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
|
585 MB_PTR_ADV(mi.mi_end); |
| 492 | 586 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
|
587 && LANGP_ENTRY(wp->w_s->b_langp, 0)->lp_slang->sl_nobreak) |
| 492 | 588 { |
| 589 char_u *p, *fp; | |
| 590 int save_result = mi.mi_result; | |
| 591 | |
| 592 /* First language in 'spelllang' is NOBREAK. Find first position | |
| 593 * 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
|
594 mi.mi_lp = LANGP_ENTRY(wp->w_s->b_langp, 0); |
| 500 | 595 if (mi.mi_lp->lp_slang->sl_fidxs != NULL) |
| 596 { | |
| 597 p = mi.mi_word; | |
| 598 fp = mi.mi_fword; | |
| 599 for (;;) | |
| 600 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
601 MB_PTR_ADV(p); |
|
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
602 MB_PTR_ADV(fp); |
| 500 | 603 if (p >= mi.mi_end) |
| 604 break; | |
| 835 | 605 mi.mi_compoff = (int)(fp - mi.mi_fword); |
| 500 | 606 find_word(&mi, FIND_COMPOUND); |
| 607 if (mi.mi_result != SP_BAD) | |
| 608 { | |
| 609 mi.mi_end = p; | |
| 610 break; | |
| 611 } | |
| 612 } | |
| 613 mi.mi_result = save_result; | |
| 614 } | |
| 492 | 615 } |
| 243 | 616 |
| 344 | 617 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED) |
| 534 | 618 *attrp = HLF_SPB; |
| 344 | 619 else if (mi.mi_result == SP_RARE) |
| 534 | 620 *attrp = HLF_SPR; |
| 344 | 621 else |
| 534 | 622 *attrp = HLF_SPL; |
| 243 | 623 } |
| 624 | |
| 483 | 625 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE)) |
| 626 { | |
| 627 /* Report SpellCap only when the word isn't badly spelled. */ | |
| 534 | 628 *attrp = HLF_SPC; |
| 483 | 629 return wrongcaplen; |
| 630 } | |
| 631 | |
| 300 | 632 return (int)(mi.mi_end - ptr); |
| 236 | 633 } |
| 634 | |
| 635 /* | |
| 300 | 636 * Check if the word at "mip->mi_word" is in the tree. |
| 339 | 637 * When "mode" is FIND_FOLDWORD check in fold-case word tree. |
| 638 * When "mode" is FIND_KEEPWORD check in keep-case word tree. | |
| 639 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word | |
| 640 * tree. | |
| 300 | 641 * |
| 642 * For a match mip->mi_result is updated. | |
| 243 | 643 */ |
| 644 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
645 find_word(matchinf_T *mip, int mode) |
| 243 | 646 { |
| 324 | 647 idx_T arridx = 0; |
| 300 | 648 int endlen[MAXWLEN]; /* length at possible word endings */ |
| 324 | 649 idx_T endidx[MAXWLEN]; /* possible word endings */ |
| 300 | 650 int endidxcnt = 0; |
| 651 int len; | |
| 652 int wlen = 0; | |
| 653 int flen; | |
| 654 int c; | |
| 655 char_u *ptr; | |
| 324 | 656 idx_T lo, hi, m; |
| 243 | 657 #ifdef FEAT_MBYTE |
| 300 | 658 char_u *s; |
| 490 | 659 #endif |
| 339 | 660 char_u *p; |
| 307 | 661 int res = SP_BAD; |
| 300 | 662 slang_T *slang = mip->mi_lp->lp_slang; |
| 663 unsigned flags; | |
| 664 char_u *byts; | |
| 324 | 665 idx_T *idxs; |
| 481 | 666 int word_ends; |
| 485 | 667 int prefix_found; |
| 492 | 668 int nobreak_result; |
| 481 | 669 |
| 670 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND) | |
| 236 | 671 { |
| 300 | 672 /* Check for word with matching case in keep-case tree. */ |
| 673 ptr = mip->mi_word; | |
| 674 flen = 9999; /* no case folding, always enough bytes */ | |
| 675 byts = slang->sl_kbyts; | |
| 676 idxs = slang->sl_kidxs; | |
| 481 | 677 |
| 678 if (mode == FIND_KEEPCOMPOUND) | |
| 679 /* Skip over the previously found word(s). */ | |
| 680 wlen += mip->mi_compoff; | |
| 236 | 681 } |
| 682 else | |
| 683 { | |
| 300 | 684 /* Check for case-folded in case-folded tree. */ |
| 685 ptr = mip->mi_fword; | |
| 686 flen = mip->mi_fwordlen; /* available case-folded bytes */ | |
| 687 byts = slang->sl_fbyts; | |
| 688 idxs = slang->sl_fidxs; | |
| 339 | 689 |
| 690 if (mode == FIND_PREFIX) | |
| 691 { | |
| 692 /* Skip over the prefix. */ | |
| 693 wlen = mip->mi_prefixlen; | |
| 694 flen -= mip->mi_prefixlen; | |
| 695 } | |
| 481 | 696 else if (mode == FIND_COMPOUND) |
| 697 { | |
| 698 /* Skip over the previously found word(s). */ | |
| 699 wlen = mip->mi_compoff; | |
| 700 flen -= mip->mi_compoff; | |
| 701 } | |
| 702 | |
| 243 | 703 } |
| 704 | |
| 300 | 705 if (byts == NULL) |
| 706 return; /* array is empty */ | |
| 236 | 707 |
| 708 /* | |
| 307 | 709 * Repeat advancing in the tree until: |
| 710 * - there is a byte that doesn't match, | |
| 711 * - we reach the end of the tree, | |
| 712 * - or we reach the end of the line. | |
| 236 | 713 */ |
| 300 | 714 for (;;) |
| 236 | 715 { |
| 346 | 716 if (flen <= 0 && *mip->mi_fend != NUL) |
| 339 | 717 flen = fold_more(mip); |
| 300 | 718 |
| 719 len = byts[arridx++]; | |
| 720 | |
| 721 /* If the first possible byte is a zero the word could end here. | |
| 722 * Remember this index, we first check for the longest word. */ | |
| 723 if (byts[arridx] == 0) | |
| 724 { | |
| 307 | 725 if (endidxcnt == MAXWLEN) |
| 726 { | |
| 727 /* Must be a corrupted spell file. */ | |
| 728 EMSG(_(e_format)); | |
| 729 return; | |
| 730 } | |
| 300 | 731 endlen[endidxcnt] = wlen; |
| 732 endidx[endidxcnt++] = arridx++; | |
| 733 --len; | |
| 734 | |
| 735 /* Skip over the zeros, there can be several flag/region | |
| 736 * combinations. */ | |
| 737 while (len > 0 && byts[arridx] == 0) | |
| 738 { | |
| 739 ++arridx; | |
| 740 --len; | |
| 741 } | |
| 742 if (len == 0) | |
| 743 break; /* no children, word must end here */ | |
| 744 } | |
| 745 | |
| 746 /* Stop looking at end of the line. */ | |
| 747 if (ptr[wlen] == NUL) | |
| 748 break; | |
| 749 | |
| 750 /* Perform a binary search in the list of accepted bytes. */ | |
| 751 c = ptr[wlen]; | |
| 346 | 752 if (c == TAB) /* <Tab> is handled like <Space> */ |
| 753 c = ' '; | |
| 300 | 754 lo = arridx; |
| 755 hi = arridx + len - 1; | |
| 756 while (lo < hi) | |
| 757 { | |
| 758 m = (lo + hi) / 2; | |
| 759 if (byts[m] > c) | |
| 760 hi = m - 1; | |
| 761 else if (byts[m] < c) | |
| 762 lo = m + 1; | |
| 763 else | |
| 764 { | |
| 765 lo = hi = m; | |
| 766 break; | |
| 236 | 767 } |
| 768 } | |
| 300 | 769 |
| 770 /* Stop if there is no matching byte. */ | |
| 771 if (hi < lo || byts[lo] != c) | |
| 772 break; | |
| 773 | |
| 774 /* Continue at the child (if there is one). */ | |
| 775 arridx = idxs[lo]; | |
| 776 ++wlen; | |
| 777 --flen; | |
| 346 | 778 |
| 779 /* One space in the good word may stand for several spaces in the | |
| 780 * checked word. */ | |
| 781 if (c == ' ') | |
| 782 { | |
| 783 for (;;) | |
| 784 { | |
| 785 if (flen <= 0 && *mip->mi_fend != NUL) | |
| 786 flen = fold_more(mip); | |
| 787 if (ptr[wlen] != ' ' && ptr[wlen] != TAB) | |
| 788 break; | |
| 789 ++wlen; | |
| 790 --flen; | |
| 791 } | |
| 792 } | |
| 236 | 793 } |
| 794 | |
| 300 | 795 /* |
| 796 * Verify that one of the possible endings is valid. Try the longest | |
| 797 * first. | |
| 798 */ | |
| 799 while (endidxcnt > 0) | |
| 800 { | |
| 801 --endidxcnt; | |
| 802 arridx = endidx[endidxcnt]; | |
| 803 wlen = endlen[endidxcnt]; | |
| 236 | 804 |
| 300 | 805 #ifdef FEAT_MBYTE |
| 806 if ((*mb_head_off)(ptr, ptr + wlen) > 0) | |
| 807 continue; /* not at first byte of character */ | |
| 808 #endif | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
809 if (spell_iswordp(ptr + wlen, mip->mi_win)) |
| 481 | 810 { |
| 492 | 811 if (slang->sl_compprog == NULL && !slang->sl_nobreak) |
| 481 | 812 continue; /* next char is a word character */ |
| 813 word_ends = FALSE; | |
| 814 } | |
| 815 else | |
| 816 word_ends = TRUE; | |
| 485 | 817 /* The prefix flag is before compound flags. Once a valid prefix flag |
| 818 * has been found we try compound flags. */ | |
| 819 prefix_found = FALSE; | |
| 300 | 820 |
| 821 #ifdef FEAT_MBYTE | |
| 339 | 822 if (mode != FIND_KEEPWORD && has_mbyte) |
| 300 | 823 { |
| 824 /* Compute byte length in original word, length may change | |
| 339 | 825 * when folding case. This can be slow, take a shortcut when the |
| 826 * case-folded word is equal to the keep-case word. */ | |
| 300 | 827 p = mip->mi_word; |
| 339 | 828 if (STRNCMP(ptr, p, wlen) != 0) |
| 829 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
830 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
|
831 MB_PTR_ADV(p); |
| 835 | 832 wlen = (int)(p - mip->mi_word); |
| 339 | 833 } |
| 300 | 834 } |
| 835 #endif | |
| 236 | 836 |
| 339 | 837 /* Check flags and region. For FIND_PREFIX check the condition and |
| 838 * prefix ID. | |
| 839 * Repeat this if there are more flags/region alternatives until there | |
| 840 * is a match. */ | |
| 841 res = SP_BAD; | |
| 842 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0; | |
| 843 --len, ++arridx) | |
| 300 | 844 { |
| 845 flags = idxs[arridx]; | |
| 324 | 846 |
| 339 | 847 /* For the fold-case tree check that the case of the checked word |
| 848 * matches with what the word in the tree requires. | |
| 849 * For keep-case tree the case is always right. For prefixes we | |
| 850 * don't bother to check. */ | |
| 851 if (mode == FIND_FOLDWORD) | |
| 300 | 852 { |
| 853 if (mip->mi_cend != mip->mi_word + wlen) | |
| 854 { | |
| 323 | 855 /* mi_capflags was set for a different word length, need |
| 856 * to do it again. */ | |
| 300 | 857 mip->mi_cend = mip->mi_word + wlen; |
| 323 | 858 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend); |
| 300 | 859 } |
| 860 | |
| 346 | 861 if (mip->mi_capflags == WF_KEEPCAP |
| 862 || !spell_valid_case(mip->mi_capflags, flags)) | |
| 339 | 863 continue; |
| 300 | 864 } |
| 236 | 865 |
| 339 | 866 /* When mode is FIND_PREFIX the word must support the prefix: |
| 867 * check the prefix ID and the condition. Do that for the list at | |
| 366 | 868 * mip->mi_prefarridx that find_prefix() filled. */ |
| 485 | 869 else if (mode == FIND_PREFIX && !prefix_found) |
| 481 | 870 { |
| 366 | 871 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx, |
| 390 | 872 flags, |
| 455 | 873 mip->mi_word + mip->mi_cprefixlen, slang, |
| 874 FALSE); | |
| 366 | 875 if (c == 0) |
| 339 | 876 continue; |
| 366 | 877 |
| 878 /* Use the WF_RARE flag for a rare prefix. */ | |
| 879 if (c & WF_RAREPFX) | |
| 880 flags |= WF_RARE; | |
| 485 | 881 prefix_found = TRUE; |
| 339 | 882 } |
| 883 | |
| 492 | 884 if (slang->sl_nobreak) |
| 885 { | |
| 886 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND) | |
| 887 && (flags & WF_BANNED) == 0) | |
| 888 { | |
| 889 /* NOBREAK: found a valid following word. That's all we | |
| 890 * need to know, so return. */ | |
| 891 mip->mi_result = SP_OK; | |
| 892 break; | |
| 893 } | |
| 894 } | |
| 895 | |
| 896 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND | |
| 897 || !word_ends)) | |
| 481 | 898 { |
| 974 | 899 /* If there is no compound flag or the word is shorter than |
| 483 | 900 * COMPOUNDMIN reject it quickly. |
| 901 * Makes you wonder why someone puts a compound flag on a word | |
| 481 | 902 * that's too short... Myspell compatibility requires this |
| 903 * anyway. */ | |
| 490 | 904 if (((unsigned)flags >> 24) == 0 |
| 905 || wlen - mip->mi_compoff < slang->sl_compminlen) | |
| 483 | 906 continue; |
| 500 | 907 #ifdef FEAT_MBYTE |
| 908 /* For multi-byte chars check character length against | |
| 909 * COMPOUNDMIN. */ | |
| 910 if (has_mbyte | |
| 501 | 911 && slang->sl_compminlen > 0 |
| 500 | 912 && mb_charlen_len(mip->mi_word + mip->mi_compoff, |
| 913 wlen - mip->mi_compoff) < slang->sl_compminlen) | |
| 914 continue; | |
| 915 #endif | |
| 483 | 916 |
| 809 | 917 /* Limit the number of compound words to COMPOUNDWORDMAX if no |
| 490 | 918 * maximum for syllables is specified. */ |
| 809 | 919 if (!word_ends && mip->mi_complen + mip->mi_compextra + 2 |
| 920 > slang->sl_compmax | |
| 490 | 921 && slang->sl_compsylmax == MAXWLEN) |
| 483 | 922 continue; |
| 923 | |
| 819 | 924 /* Don't allow compounding on a side where an affix was added, |
| 925 * unless COMPOUNDPERMITFLAG was used. */ | |
| 926 if (mip->mi_complen > 0 && (flags & WF_NOCOMPBEF)) | |
| 927 continue; | |
| 928 if (!word_ends && (flags & WF_NOCOMPAFT)) | |
| 929 continue; | |
| 930 | |
| 485 | 931 /* Quickly check if compounding is possible with this flag. */ |
| 495 | 932 if (!byte_in_str(mip->mi_complen == 0 |
| 485 | 933 ? slang->sl_compstartflags |
| 934 : slang->sl_compallflags, | |
| 495 | 935 ((unsigned)flags >> 24))) |
| 481 | 936 continue; |
| 937 | |
| 1762 | 938 /* If there is a match with a CHECKCOMPOUNDPATTERN rule |
| 939 * discard the compound word. */ | |
| 940 if (match_checkcompoundpattern(ptr, wlen, &slang->sl_comppat)) | |
| 941 continue; | |
| 942 | |
| 490 | 943 if (mode == FIND_COMPOUND) |
| 944 { | |
| 945 int capflags; | |
| 946 | |
| 947 /* Need to check the caps type of the appended compound | |
| 948 * word. */ | |
| 949 #ifdef FEAT_MBYTE | |
| 950 if (has_mbyte && STRNCMP(ptr, mip->mi_word, | |
| 951 mip->mi_compoff) != 0) | |
| 952 { | |
| 953 /* case folding may have changed the length */ | |
| 954 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
|
955 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
|
956 MB_PTR_ADV(p); |
| 490 | 957 } |
| 958 else | |
| 959 #endif | |
| 960 p = mip->mi_word + mip->mi_compoff; | |
| 961 capflags = captype(p, mip->mi_word + wlen); | |
| 962 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP | |
| 963 && (flags & WF_FIXCAP) != 0)) | |
| 964 continue; | |
| 965 | |
| 966 if (capflags != WF_ALLCAP) | |
| 967 { | |
| 968 /* When the character before the word is a word | |
| 969 * character we do not accept a Onecap word. We do | |
| 970 * accept a no-caps word, even when the dictionary | |
| 971 * word specifies ONECAP. */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
972 MB_PTR_BACK(mip->mi_word, p); |
| 5477 | 973 if (spell_iswordp_nmw(p, mip->mi_win) |
| 490 | 974 ? capflags == WF_ONECAP |
| 975 : (flags & WF_ONECAP) != 0 | |
| 976 && capflags != WF_ONECAP) | |
| 977 continue; | |
| 978 } | |
| 979 } | |
| 980 | |
| 483 | 981 /* If the word ends the sequence of compound flags of the |
| 714 | 982 * words must match with one of the COMPOUNDRULE items and |
| 483 | 983 * the number of syllables must not be too large. */ |
| 984 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24); | |
| 985 mip->mi_compflags[mip->mi_complen + 1] = NUL; | |
| 986 if (word_ends) | |
| 987 { | |
| 988 char_u fword[MAXWLEN]; | |
| 989 | |
| 990 if (slang->sl_compsylmax < MAXWLEN) | |
| 991 { | |
| 992 /* "fword" is only needed for checking syllables. */ | |
| 993 if (ptr == mip->mi_word) | |
| 994 (void)spell_casefold(ptr, wlen, fword, MAXWLEN); | |
| 995 else | |
| 996 vim_strncpy(fword, ptr, endlen[endidxcnt]); | |
| 997 } | |
| 998 if (!can_compound(slang, fword, mip->mi_compflags)) | |
| 999 continue; | |
| 1000 } | |
| 1762 | 1001 else if (slang->sl_comprules != NULL |
| 1002 && !match_compoundrule(slang, mip->mi_compflags)) | |
| 1003 /* The compound flags collected so far do not match any | |
| 1004 * COMPOUNDRULE, discard the compounded word. */ | |
| 1005 continue; | |
| 481 | 1006 } |
| 1007 | |
| 500 | 1008 /* Check NEEDCOMPOUND: can't use word without compounding. */ |
| 1009 else if (flags & WF_NEEDCOMP) | |
| 1010 continue; | |
| 1011 | |
| 492 | 1012 nobreak_result = SP_OK; |
| 1013 | |
| 481 | 1014 if (!word_ends) |
| 1015 { | |
| 492 | 1016 int save_result = mip->mi_result; |
| 1017 char_u *save_end = mip->mi_end; | |
| 501 | 1018 langp_T *save_lp = mip->mi_lp; |
| 1019 int lpi; | |
| 492 | 1020 |
| 1021 /* Check that a valid word follows. If there is one and we | |
| 1022 * are compounding, it will set "mi_result", thus we are | |
| 1023 * always finished here. For NOBREAK we only check that a | |
| 1024 * valid word follows. | |
| 481 | 1025 * Recursive! */ |
| 492 | 1026 if (slang->sl_nobreak) |
| 1027 mip->mi_result = SP_BAD; | |
| 481 | 1028 |
| 1029 /* Find following word in case-folded tree. */ | |
| 1030 mip->mi_compoff = endlen[endidxcnt]; | |
| 1031 #ifdef FEAT_MBYTE | |
| 1032 if (has_mbyte && mode == FIND_KEEPWORD) | |
| 1033 { | |
| 1034 /* Compute byte length in case-folded word from "wlen": | |
| 1035 * byte length in keep-case word. Length may change when | |
| 1036 * folding case. This can be slow, take a shortcut when | |
| 1037 * the case-folded word is equal to the keep-case word. */ | |
| 1038 p = mip->mi_fword; | |
| 1039 if (STRNCMP(ptr, p, wlen) != 0) | |
| 1040 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
1041 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
|
1042 MB_PTR_ADV(p); |
| 835 | 1043 mip->mi_compoff = (int)(p - mip->mi_fword); |
| 481 | 1044 } |
| 1045 } | |
| 1046 #endif | |
|
8956
d9e671c5afe6
commit https://github.com/vim/vim/commit/ba53435144f46eaaa53c63a62e748b3feee9742c
Christian Brabandt <cb@256bit.org>
parents:
8524
diff
changeset
|
1047 #if 0 /* Disabled, see below */ |
| 485 | 1048 c = mip->mi_compoff; |
|
8956
d9e671c5afe6
commit https://github.com/vim/vim/commit/ba53435144f46eaaa53c63a62e748b3feee9742c
Christian Brabandt <cb@256bit.org>
parents:
8524
diff
changeset
|
1049 #endif |
| 483 | 1050 ++mip->mi_complen; |
| 809 | 1051 if (flags & WF_COMPROOT) |
| 1052 ++mip->mi_compextra; | |
| 501 | 1053 |
| 1054 /* For NOBREAK we need to try all NOBREAK languages, at least | |
| 1055 * to find the ".add" file(s). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1056 for (lpi = 0; lpi < mip->mi_win->w_s->b_langp.ga_len; ++lpi) |
| 501 | 1057 { |
| 1058 if (slang->sl_nobreak) | |
| 1059 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1060 mip->mi_lp = LANGP_ENTRY(mip->mi_win->w_s->b_langp, lpi); |
| 501 | 1061 if (mip->mi_lp->lp_slang->sl_fidxs == NULL |
| 1062 || !mip->mi_lp->lp_slang->sl_nobreak) | |
| 1063 continue; | |
| 1064 } | |
| 1065 | |
| 1066 find_word(mip, FIND_COMPOUND); | |
| 1067 | |
| 1068 /* When NOBREAK any word that matches is OK. Otherwise we | |
| 1069 * need to find the longest match, thus try with keep-case | |
| 1070 * and prefix too. */ | |
| 492 | 1071 if (!slang->sl_nobreak || mip->mi_result == SP_BAD) |
| 1072 { | |
| 501 | 1073 /* Find following word in keep-case tree. */ |
| 1074 mip->mi_compoff = wlen; | |
| 1075 find_word(mip, FIND_KEEPCOMPOUND); | |
| 1076 | |
| 819 | 1077 #if 0 /* Disabled, a prefix must not appear halfway a compound word, |
| 1078 unless the COMPOUNDPERMITFLAG is used and then it can't be a | |
| 1079 postponed prefix. */ | |
| 501 | 1080 if (!slang->sl_nobreak || mip->mi_result == SP_BAD) |
| 1081 { | |
| 1082 /* Check for following word with prefix. */ | |
| 1083 mip->mi_compoff = c; | |
| 1084 find_prefix(mip, FIND_COMPOUND); | |
| 1085 } | |
| 819 | 1086 #endif |
| 492 | 1087 } |
| 501 | 1088 |
| 1089 if (!slang->sl_nobreak) | |
| 1090 break; | |
| 492 | 1091 } |
| 483 | 1092 --mip->mi_complen; |
| 809 | 1093 if (flags & WF_COMPROOT) |
| 1094 --mip->mi_compextra; | |
| 501 | 1095 mip->mi_lp = save_lp; |
| 485 | 1096 |
| 492 | 1097 if (slang->sl_nobreak) |
| 1098 { | |
| 1099 nobreak_result = mip->mi_result; | |
| 1100 mip->mi_result = save_result; | |
| 1101 mip->mi_end = save_end; | |
| 1102 } | |
| 1103 else | |
| 1104 { | |
| 1105 if (mip->mi_result == SP_OK) | |
| 1106 break; | |
| 1107 continue; | |
| 1108 } | |
| 481 | 1109 } |
| 1110 | |
| 339 | 1111 if (flags & WF_BANNED) |
| 1112 res = SP_BANNED; | |
| 1113 else if (flags & WF_REGION) | |
| 1114 { | |
| 1115 /* Check region. */ | |
| 390 | 1116 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0) |
| 300 | 1117 res = SP_OK; |
| 339 | 1118 else |
| 1119 res = SP_LOCAL; | |
| 300 | 1120 } |
| 339 | 1121 else if (flags & WF_RARE) |
| 1122 res = SP_RARE; | |
| 307 | 1123 else |
| 339 | 1124 res = SP_OK; |
| 1125 | |
| 492 | 1126 /* Always use the longest match and the best result. For NOBREAK |
| 1127 * we separately keep the longest match without a following good | |
| 1128 * word as a fall-back. */ | |
| 1129 if (nobreak_result == SP_BAD) | |
| 1130 { | |
| 1131 if (mip->mi_result2 > res) | |
| 1132 { | |
| 1133 mip->mi_result2 = res; | |
| 1134 mip->mi_end2 = mip->mi_word + wlen; | |
| 1135 } | |
| 1136 else if (mip->mi_result2 == res | |
| 1137 && mip->mi_end2 < mip->mi_word + wlen) | |
| 1138 mip->mi_end2 = mip->mi_word + wlen; | |
| 1139 } | |
| 1140 else if (mip->mi_result > res) | |
| 339 | 1141 { |
| 1142 mip->mi_result = res; | |
| 1143 mip->mi_end = mip->mi_word + wlen; | |
| 1144 } | |
| 351 | 1145 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen) |
| 339 | 1146 mip->mi_end = mip->mi_word + wlen; |
| 1147 | |
| 492 | 1148 if (mip->mi_result == SP_OK) |
| 339 | 1149 break; |
| 300 | 1150 } |
| 1151 | |
| 492 | 1152 if (mip->mi_result == SP_OK) |
| 300 | 1153 break; |
| 1154 } | |
| 236 | 1155 } |
| 1156 | |
| 323 | 1157 /* |
| 1762 | 1158 * Return TRUE if there is a match between the word ptr[wlen] and |
| 1159 * CHECKCOMPOUNDPATTERN rules, assuming that we will concatenate with another | |
| 1160 * word. | |
| 1161 * A match means that the first part of CHECKCOMPOUNDPATTERN matches at the | |
| 1162 * end of ptr[wlen] and the second part matches after it. | |
| 1163 */ | |
| 1164 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1165 match_checkcompoundpattern( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1166 char_u *ptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1167 int wlen, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1168 garray_T *gap) /* &sl_comppat */ |
| 1762 | 1169 { |
| 1170 int i; | |
| 1171 char_u *p; | |
| 1172 int len; | |
| 1173 | |
| 1174 for (i = 0; i + 1 < gap->ga_len; i += 2) | |
| 1175 { | |
| 1176 p = ((char_u **)gap->ga_data)[i + 1]; | |
| 1177 if (STRNCMP(ptr + wlen, p, STRLEN(p)) == 0) | |
| 1178 { | |
| 1179 /* Second part matches at start of following compound word, now | |
| 1180 * check if first part matches at end of previous word. */ | |
| 1181 p = ((char_u **)gap->ga_data)[i]; | |
| 1771 | 1182 len = (int)STRLEN(p); |
| 1762 | 1183 if (len <= wlen && STRNCMP(ptr + wlen - len, p, len) == 0) |
| 1184 return TRUE; | |
| 1185 } | |
| 1186 } | |
| 1187 return FALSE; | |
| 1188 } | |
| 1189 | |
| 1190 /* | |
| 626 | 1191 * Return TRUE if "flags" is a valid sequence of compound flags and "word" |
| 1192 * does not have too many syllables. | |
| 482 | 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_compound(slang_T *slang, char_u *word, char_u *flags) |
| 483 | 1196 { |
| 495 | 1197 #ifdef FEAT_MBYTE |
| 1198 char_u uflags[MAXWLEN * 2]; | |
| 1199 int i; | |
| 1200 #endif | |
| 1201 char_u *p; | |
| 483 | 1202 |
| 1203 if (slang->sl_compprog == NULL) | |
| 1204 return FALSE; | |
| 495 | 1205 #ifdef FEAT_MBYTE |
| 1206 if (enc_utf8) | |
| 1207 { | |
| 1208 /* Need to convert the single byte flags to utf8 characters. */ | |
| 1209 p = uflags; | |
| 1210 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
|
1211 p += utf_char2bytes(flags[i], p); |
| 495 | 1212 *p = NUL; |
| 1213 p = uflags; | |
| 1214 } | |
| 1215 else | |
| 1216 #endif | |
| 1217 p = flags; | |
| 6375 | 1218 if (!vim_regexec_prog(&slang->sl_compprog, FALSE, p, 0)) |
| 483 | 1219 return FALSE; |
| 1220 | |
| 490 | 1221 /* Count the number of syllables. This may be slow, do it last. If there |
| 1222 * are too many syllables AND the number of compound words is above | |
| 809 | 1223 * COMPOUNDWORDMAX then compounding is not allowed. */ |
| 483 | 1224 if (slang->sl_compsylmax < MAXWLEN |
| 1225 && count_syllables(slang, word) > slang->sl_compsylmax) | |
| 495 | 1226 return (int)STRLEN(flags) < slang->sl_compmax; |
| 483 | 1227 return TRUE; |
| 482 | 1228 } |
| 1229 | |
| 1230 /* | |
| 1762 | 1231 * Return TRUE when the sequence of flags in "compflags" plus "flag" can |
| 1232 * possibly form a valid compounded word. This also checks the COMPOUNDRULE | |
| 1233 * lines if they don't contain wildcards. | |
| 1234 */ | |
| 1235 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1236 can_be_compound( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1237 trystate_T *sp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1238 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1239 char_u *compflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1240 int flag) |
| 1762 | 1241 { |
| 1242 /* If the flag doesn't appear in sl_compstartflags or sl_compallflags | |
| 1243 * then it can't possibly compound. */ | |
| 1244 if (!byte_in_str(sp->ts_complen == sp->ts_compsplit | |
| 1245 ? slang->sl_compstartflags : slang->sl_compallflags, flag)) | |
| 1246 return FALSE; | |
| 1247 | |
| 1248 /* If there are no wildcards, we can check if the flags collected so far | |
| 1249 * possibly can form a match with COMPOUNDRULE patterns. This only | |
| 1250 * makes sense when we have two or more words. */ | |
| 1251 if (slang->sl_comprules != NULL && sp->ts_complen > sp->ts_compsplit) | |
| 1252 { | |
| 1253 int v; | |
| 1254 | |
| 1255 compflags[sp->ts_complen] = flag; | |
| 1256 compflags[sp->ts_complen + 1] = NUL; | |
| 1257 v = match_compoundrule(slang, compflags + sp->ts_compsplit); | |
| 1258 compflags[sp->ts_complen] = NUL; | |
| 1259 return v; | |
| 1260 } | |
| 1261 | |
| 1262 return TRUE; | |
| 1263 } | |
| 1264 | |
| 1265 | |
| 1266 /* | |
| 1267 * Return TRUE if the compound flags in compflags[] match the start of any | |
| 1268 * compound rule. This is used to stop trying a compound if the flags | |
| 1269 * collected so far can't possibly match any compound rule. | |
| 1270 * Caller must check that slang->sl_comprules is not NULL. | |
| 1271 */ | |
| 1272 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1273 match_compoundrule(slang_T *slang, char_u *compflags) |
| 1762 | 1274 { |
| 1275 char_u *p; | |
| 1276 int i; | |
| 1277 int c; | |
| 1278 | |
| 1279 /* loop over all the COMPOUNDRULE entries */ | |
| 1280 for (p = slang->sl_comprules; *p != NUL; ++p) | |
| 1281 { | |
| 1282 /* loop over the flags in the compound word we have made, match | |
| 1283 * them against the current rule entry */ | |
| 1284 for (i = 0; ; ++i) | |
| 1285 { | |
| 1286 c = compflags[i]; | |
| 1287 if (c == NUL) | |
| 1288 /* found a rule that matches for the flags we have so far */ | |
| 1289 return TRUE; | |
| 1290 if (*p == '/' || *p == NUL) | |
| 1291 break; /* end of rule, it's too short */ | |
| 1292 if (*p == '[') | |
| 1293 { | |
| 1294 int match = FALSE; | |
| 1295 | |
| 1296 /* compare against all the flags in [] */ | |
| 1297 ++p; | |
| 1298 while (*p != ']' && *p != NUL) | |
| 1299 if (*p++ == c) | |
| 1300 match = TRUE; | |
| 1301 if (!match) | |
| 1302 break; /* none matches */ | |
| 1303 } | |
| 1304 else if (*p != c) | |
| 1305 break; /* flag of word doesn't match flag in pattern */ | |
| 1306 ++p; | |
| 1307 } | |
| 1308 | |
| 1309 /* Skip to the next "/", where the next pattern starts. */ | |
| 1310 p = vim_strchr(p, '/'); | |
| 1311 if (p == NULL) | |
| 1312 break; | |
| 1313 } | |
| 1314 | |
| 1315 /* Checked all the rules and none of them match the flags, so there | |
| 1316 * can't possibly be a compound starting with these flags. */ | |
| 1317 return FALSE; | |
| 1318 } | |
| 1319 | |
| 1320 /* | |
| 390 | 1321 * Return non-zero if the prefix indicated by "arridx" matches with the prefix |
| 1322 * ID in "flags" for the word "word". | |
| 366 | 1323 * The WF_RAREPFX flag is included in the return value for a rare prefix. |
| 351 | 1324 */ |
| 1325 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1326 valid_word_prefix( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1327 int totprefcnt, /* nr of prefix IDs */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1328 int arridx, /* idx in sl_pidxs[] */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1329 int flags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1330 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1331 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1332 int cond_req) /* only use prefixes with a condition */ |
| 351 | 1333 { |
| 1334 int prefcnt; | |
| 1335 int pidx; | |
| 6375 | 1336 regprog_T **rp; |
| 390 | 1337 int prefid; |
| 1338 | |
| 1339 prefid = (unsigned)flags >> 24; | |
| 351 | 1340 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt) |
| 1341 { | |
| 1342 pidx = slang->sl_pidxs[arridx + prefcnt]; | |
| 1343 | |
| 1344 /* Check the prefix ID. */ | |
| 1345 if (prefid != (pidx & 0xff)) | |
| 1346 continue; | |
| 1347 | |
| 390 | 1348 /* Check if the prefix doesn't combine and the word already has a |
| 1349 * suffix. */ | |
| 1350 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC)) | |
| 1351 continue; | |
| 1352 | |
| 351 | 1353 /* Check the condition, if there is one. The condition index is |
| 366 | 1354 * stored in the two bytes above the prefix ID byte. */ |
| 6375 | 1355 rp = &slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff]; |
| 1356 if (*rp != NULL) | |
| 1357 { | |
| 1358 if (!vim_regexec_prog(rp, FALSE, word, 0)) | |
| 351 | 1359 continue; |
| 1360 } | |
| 455 | 1361 else if (cond_req) |
| 1362 continue; | |
| 1363 | |
| 1364 /* It's a match! Return the WF_ flags. */ | |
| 366 | 1365 return pidx; |
| 1366 } | |
| 1367 return 0; | |
| 351 | 1368 } |
| 1369 | |
| 1370 /* | |
| 339 | 1371 * Check if the word at "mip->mi_word" has a matching prefix. |
| 1372 * If it does, then check the following word. | |
| 1373 * | |
| 485 | 1374 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a |
| 1375 * prefix in a compound word. | |
| 1376 * | |
| 339 | 1377 * For a match mip->mi_result is updated. |
| 1378 */ | |
| 1379 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1380 find_prefix(matchinf_T *mip, int mode) |
| 339 | 1381 { |
| 1382 idx_T arridx = 0; | |
| 1383 int len; | |
| 1384 int wlen = 0; | |
| 1385 int flen; | |
| 1386 int c; | |
| 1387 char_u *ptr; | |
| 1388 idx_T lo, hi, m; | |
| 1389 slang_T *slang = mip->mi_lp->lp_slang; | |
| 1390 char_u *byts; | |
| 1391 idx_T *idxs; | |
| 1392 | |
| 375 | 1393 byts = slang->sl_pbyts; |
| 1394 if (byts == NULL) | |
| 1395 return; /* array is empty */ | |
| 1396 | |
| 339 | 1397 /* We use the case-folded word here, since prefixes are always |
| 1398 * case-folded. */ | |
| 1399 ptr = mip->mi_fword; | |
| 1400 flen = mip->mi_fwordlen; /* available case-folded bytes */ | |
| 485 | 1401 if (mode == FIND_COMPOUND) |
| 1402 { | |
| 1403 /* Skip over the previously found word(s). */ | |
| 1404 ptr += mip->mi_compoff; | |
| 1405 flen -= mip->mi_compoff; | |
| 1406 } | |
| 339 | 1407 idxs = slang->sl_pidxs; |
| 1408 | |
| 1409 /* | |
| 1410 * Repeat advancing in the tree until: | |
| 1411 * - there is a byte that doesn't match, | |
| 1412 * - we reach the end of the tree, | |
| 1413 * - or we reach the end of the line. | |
| 1414 */ | |
| 1415 for (;;) | |
| 1416 { | |
| 1417 if (flen == 0 && *mip->mi_fend != NUL) | |
| 1418 flen = fold_more(mip); | |
| 1419 | |
| 1420 len = byts[arridx++]; | |
| 1421 | |
| 1422 /* If the first possible byte is a zero the prefix could end here. | |
| 1423 * Check if the following word matches and supports the prefix. */ | |
| 1424 if (byts[arridx] == 0) | |
| 1425 { | |
| 1426 /* There can be several prefixes with different conditions. We | |
| 1427 * try them all, since we don't know which one will give the | |
| 1428 * longest match. The word is the same each time, pass the list | |
| 1429 * of possible prefixes to find_word(). */ | |
| 1430 mip->mi_prefarridx = arridx; | |
| 1431 mip->mi_prefcnt = len; | |
| 1432 while (len > 0 && byts[arridx] == 0) | |
| 1433 { | |
| 1434 ++arridx; | |
| 1435 --len; | |
| 1436 } | |
| 1437 mip->mi_prefcnt -= len; | |
| 1438 | |
| 1439 /* Find the word that comes after the prefix. */ | |
| 1440 mip->mi_prefixlen = wlen; | |
| 485 | 1441 if (mode == FIND_COMPOUND) |
| 1442 /* Skip over the previously found word(s). */ | |
| 1443 mip->mi_prefixlen += mip->mi_compoff; | |
| 1444 | |
| 455 | 1445 #ifdef FEAT_MBYTE |
| 1446 if (has_mbyte) | |
| 1447 { | |
| 1448 /* Case-folded length may differ from original length. */ | |
| 485 | 1449 mip->mi_cprefixlen = nofold_len(mip->mi_fword, |
| 1450 mip->mi_prefixlen, mip->mi_word); | |
| 455 | 1451 } |
| 1452 else | |
| 485 | 1453 mip->mi_cprefixlen = mip->mi_prefixlen; |
| 455 | 1454 #endif |
| 339 | 1455 find_word(mip, FIND_PREFIX); |
| 1456 | |
| 1457 | |
| 1458 if (len == 0) | |
| 1459 break; /* no children, word must end here */ | |
| 1460 } | |
| 1461 | |
| 1462 /* Stop looking at end of the line. */ | |
| 1463 if (ptr[wlen] == NUL) | |
| 1464 break; | |
| 1465 | |
| 1466 /* Perform a binary search in the list of accepted bytes. */ | |
| 1467 c = ptr[wlen]; | |
| 1468 lo = arridx; | |
| 1469 hi = arridx + len - 1; | |
| 1470 while (lo < hi) | |
| 1471 { | |
| 1472 m = (lo + hi) / 2; | |
| 1473 if (byts[m] > c) | |
| 1474 hi = m - 1; | |
| 1475 else if (byts[m] < c) | |
| 1476 lo = m + 1; | |
| 1477 else | |
| 1478 { | |
| 1479 lo = hi = m; | |
| 1480 break; | |
| 1481 } | |
| 1482 } | |
| 1483 | |
| 1484 /* Stop if there is no matching byte. */ | |
| 1485 if (hi < lo || byts[lo] != c) | |
| 1486 break; | |
| 1487 | |
| 1488 /* Continue at the child (if there is one). */ | |
| 1489 arridx = idxs[lo]; | |
| 1490 ++wlen; | |
| 1491 --flen; | |
| 1492 } | |
| 1493 } | |
| 1494 | |
| 1495 /* | |
| 1496 * Need to fold at least one more character. Do until next non-word character | |
| 626 | 1497 * for efficiency. Include the non-word character too. |
| 339 | 1498 * Return the length of the folded chars in bytes. |
| 1499 */ | |
| 1500 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1501 fold_more(matchinf_T *mip) |
| 339 | 1502 { |
| 1503 int flen; | |
| 1504 char_u *p; | |
| 1505 | |
| 1506 p = mip->mi_fend; | |
| 1507 do | |
| 1508 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
1509 MB_PTR_ADV(mip->mi_fend); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1510 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_win)); |
| 339 | 1511 |
| 626 | 1512 /* Include the non-word character so that we can check for the word end. */ |
| 339 | 1513 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
|
1514 MB_PTR_ADV(mip->mi_fend); |
| 339 | 1515 |
| 1516 (void)spell_casefold(p, (int)(mip->mi_fend - p), | |
| 1517 mip->mi_fword + mip->mi_fwordlen, | |
| 1518 MAXWLEN - mip->mi_fwordlen); | |
| 835 | 1519 flen = (int)STRLEN(mip->mi_fword + mip->mi_fwordlen); |
| 339 | 1520 mip->mi_fwordlen += flen; |
| 1521 return flen; | |
| 1522 } | |
| 1523 | |
| 1524 /* | |
| 323 | 1525 * Check case flags for a word. Return TRUE if the word has the requested |
| 1526 * case. | |
| 1527 */ | |
| 1528 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1529 spell_valid_case( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1530 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
|
1531 int treeflags) /* flags for the word in the spell tree */ |
| 323 | 1532 { |
| 389 | 1533 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0) |
| 323 | 1534 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0 |
| 474 | 1535 && ((treeflags & WF_ONECAP) == 0 |
| 1536 || (wordflags & WF_ONECAP) != 0))); | |
| 323 | 1537 } |
| 1538 | |
| 351 | 1539 /* |
| 1540 * Return TRUE if spell checking is not enabled. | |
| 1541 */ | |
| 1542 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1543 no_spell_checking(win_T *wp) |
| 498 | 1544 { |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1545 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
|
1546 || wp->w_s->b_langp.ga_len == 0) |
| 351 | 1547 { |
| 1548 EMSG(_("E756: Spell checking is not enabled")); | |
| 1549 return TRUE; | |
| 1550 } | |
| 1551 return FALSE; | |
| 1552 } | |
| 300 | 1553 |
| 236 | 1554 /* |
| 1555 * Move to next spell error. | |
| 500 | 1556 * "curline" is FALSE for "[s", "]s", "[S" and "]S". |
| 1557 * "curline" is TRUE to find word under/after cursor in the same line. | |
| 483 | 1558 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move |
| 1559 * to after badly spelled word before the cursor. | |
| 495 | 1560 * Return 0 if not found, length of the badly spelled word otherwise. |
| 236 | 1561 */ |
| 1562 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1563 spell_move_to( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1564 win_T *wp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1565 int dir, /* FORWARD or BACKWARD */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1566 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
|
1567 int curline, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1568 hlf_T *attrp) /* return: attributes of bad word or NULL |
| 534 | 1569 (only when "dir" is FORWARD) */ |
| 236 | 1570 { |
| 249 | 1571 linenr_T lnum; |
| 1572 pos_T found_pos; | |
| 495 | 1573 int found_len = 0; |
| 236 | 1574 char_u *line; |
| 1575 char_u *p; | |
| 346 | 1576 char_u *endp; |
| 534 | 1577 hlf_T attr; |
| 236 | 1578 int len; |
| 5519 | 1579 #ifdef FEAT_SYN_HL |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1580 int has_syntax = syntax_present(wp); |
| 5519 | 1581 #endif |
| 249 | 1582 int col; |
| 1583 int can_spell; | |
| 346 | 1584 char_u *buf = NULL; |
| 1585 int buflen = 0; | |
| 1586 int skip = 0; | |
| 385 | 1587 int capcol = -1; |
| 500 | 1588 int found_one = FALSE; |
| 1589 int wrapped = FALSE; | |
| 236 | 1590 |
| 498 | 1591 if (no_spell_checking(wp)) |
| 495 | 1592 return 0; |
| 236 | 1593 |
| 249 | 1594 /* |
| 1595 * Start looking for bad word at the start of the line, because we can't | |
| 817 | 1596 * start halfway a word, we don't know where it starts or ends. |
| 249 | 1597 * |
| 1598 * When searching backwards, we continue in the line to find the last | |
| 1599 * bad word (in the cursor line: before the cursor). | |
| 346 | 1600 * |
| 1601 * We concatenate the start of the next line, so that wrapped words work | |
| 1602 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards | |
| 1603 * though... | |
| 249 | 1604 */ |
| 498 | 1605 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
|
1606 CLEAR_POS(&found_pos); |
| 236 | 1607 |
| 1608 while (!got_int) | |
| 1609 { | |
| 498 | 1610 line = ml_get_buf(wp->w_buffer, lnum, FALSE); |
| 346 | 1611 |
| 835 | 1612 len = (int)STRLEN(line); |
| 346 | 1613 if (buflen < len + MAXWLEN + 2) |
| 1614 { | |
| 1615 vim_free(buf); | |
| 1616 buflen = len + MAXWLEN + 2; | |
| 1617 buf = alloc(buflen); | |
| 1618 if (buf == NULL) | |
| 1619 break; | |
| 1620 } | |
| 1621 | |
| 385 | 1622 /* In first line check first word for Capital. */ |
| 1623 if (lnum == 1) | |
| 1624 capcol = 0; | |
| 1625 | |
| 1626 /* For checking first word with a capital skip white space. */ | |
| 1627 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
|
1628 capcol = getwhitecols(line); |
| 835 | 1629 else if (curline && wp == curwin) |
| 1630 { | |
| 1631 /* 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
|
1632 col = getwhitecols(line); |
| 835 | 1633 if (check_need_cap(lnum, col)) |
| 1634 capcol = col; | |
| 1635 | |
| 1636 /* Need to get the line again, may have looked at the previous | |
| 1637 * one. */ | |
| 1638 line = ml_get_buf(wp->w_buffer, lnum, FALSE); | |
| 1639 } | |
| 385 | 1640 |
| 346 | 1641 /* Copy the line into "buf" and append the start of the next line if |
| 1642 * possible. */ | |
| 1643 STRCPY(buf, line); | |
| 498 | 1644 if (lnum < wp->w_buffer->b_ml.ml_line_count) |
| 883 | 1645 spell_cat_line(buf + STRLEN(buf), |
| 1646 ml_get_buf(wp->w_buffer, lnum + 1, FALSE), MAXWLEN); | |
| 346 | 1647 |
| 1648 p = buf + skip; | |
| 1649 endp = buf + len; | |
| 1650 while (p < endp) | |
| 236 | 1651 { |
| 500 | 1652 /* When searching backward don't search after the cursor. Unless |
| 1653 * we wrapped around the end of the buffer. */ | |
| 300 | 1654 if (dir == BACKWARD |
| 498 | 1655 && lnum == wp->w_cursor.lnum |
| 500 | 1656 && !wrapped |
| 498 | 1657 && (colnr_T)(p - buf) >= wp->w_cursor.col) |
| 300 | 1658 break; |
| 249 | 1659 |
| 300 | 1660 /* start of word */ |
| 534 | 1661 attr = HLF_COUNT; |
| 625 | 1662 len = spell_check(wp, p, &attr, &capcol, FALSE); |
| 249 | 1663 |
| 534 | 1664 if (attr != HLF_COUNT) |
| 300 | 1665 { |
| 1666 /* We found a bad word. Check the attribute. */ | |
| 534 | 1667 if (allwords || attr == HLF_SPB) |
| 236 | 1668 { |
| 300 | 1669 /* When searching forward only accept a bad word after |
| 1670 * the cursor. */ | |
| 1671 if (dir == BACKWARD | |
| 500 | 1672 || lnum != wp->w_cursor.lnum |
| 498 | 1673 || (lnum == wp->w_cursor.lnum |
| 500 | 1674 && (wrapped |
| 1675 || (colnr_T)(curline ? p - buf + len | |
| 346 | 1676 : p - buf) |
| 500 | 1677 > wp->w_cursor.col))) |
| 236 | 1678 { |
| 5519 | 1679 #ifdef FEAT_SYN_HL |
| 300 | 1680 if (has_syntax) |
| 249 | 1681 { |
| 835 | 1682 col = (int)(p - buf); |
| 498 | 1683 (void)syn_get_id(wp, lnum, (colnr_T)col, |
| 1504 | 1684 FALSE, &can_spell, FALSE); |
| 857 | 1685 if (!can_spell) |
| 1686 attr = HLF_COUNT; | |
| 300 | 1687 } |
| 1688 else | |
| 737 | 1689 #endif |
| 300 | 1690 can_spell = TRUE; |
| 249 | 1691 |
| 300 | 1692 if (can_spell) |
| 1693 { | |
| 857 | 1694 found_one = TRUE; |
| 300 | 1695 found_pos.lnum = lnum; |
| 835 | 1696 found_pos.col = (int)(p - buf); |
| 249 | 1697 #ifdef FEAT_VIRTUALEDIT |
| 300 | 1698 found_pos.coladd = 0; |
| 249 | 1699 #endif |
| 300 | 1700 if (dir == FORWARD) |
| 1701 { | |
| 1702 /* No need to search further. */ | |
| 498 | 1703 wp->w_cursor = found_pos; |
| 346 | 1704 vim_free(buf); |
| 498 | 1705 if (attrp != NULL) |
| 1706 *attrp = attr; | |
| 495 | 1707 return len; |
| 249 | 1708 } |
| 483 | 1709 else if (curline) |
| 1710 /* Insert mode completion: put cursor after | |
| 1711 * the bad word. */ | |
| 1712 found_pos.col += len; | |
| 495 | 1713 found_len = len; |
| 249 | 1714 } |
| 236 | 1715 } |
| 857 | 1716 else |
| 1717 found_one = TRUE; | |
| 236 | 1718 } |
| 1719 } | |
| 1720 | |
| 300 | 1721 /* advance to character after the word */ |
| 1722 p += len; | |
| 385 | 1723 capcol -= len; |
| 236 | 1724 } |
| 1725 | |
| 483 | 1726 if (dir == BACKWARD && found_pos.lnum != 0) |
| 1727 { | |
| 500 | 1728 /* Use the last match in the line (before the cursor). */ |
| 498 | 1729 wp->w_cursor = found_pos; |
| 483 | 1730 vim_free(buf); |
| 495 | 1731 return found_len; |
| 483 | 1732 } |
| 1733 | |
| 323 | 1734 if (curline) |
| 346 | 1735 break; /* only check cursor line */ |
| 323 | 1736 |
|
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
|
1737 /* 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
|
1738 * 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
|
1739 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
|
1740 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
|
1741 |
| 236 | 1742 /* Advance to next line. */ |
| 249 | 1743 if (dir == BACKWARD) |
| 1744 { | |
| 500 | 1745 if (lnum > 1) |
| 1746 --lnum; | |
| 1747 else if (!p_ws) | |
| 1748 break; /* at first line and 'nowrapscan' */ | |
| 1749 else | |
| 1750 { | |
| 1751 /* Wrap around to the end of the buffer. May search the | |
| 1752 * starting line again and accept the last match. */ | |
| 1753 lnum = wp->w_buffer->b_ml.ml_line_count; | |
| 1754 wrapped = TRUE; | |
| 503 | 1755 if (!shortmess(SHM_SEARCH)) |
| 1756 give_warning((char_u *)_(top_bot_msg), TRUE); | |
| 500 | 1757 } |
| 385 | 1758 capcol = -1; |
| 249 | 1759 } |
| 1760 else | |
| 1761 { | |
| 500 | 1762 if (lnum < wp->w_buffer->b_ml.ml_line_count) |
| 1763 ++lnum; | |
| 1764 else if (!p_ws) | |
| 1765 break; /* at first line and 'nowrapscan' */ | |
| 1766 else | |
| 1767 { | |
| 1768 /* Wrap around to the start of the buffer. May search the | |
| 1769 * starting line again and accept the first match. */ | |
| 1770 lnum = 1; | |
| 1771 wrapped = TRUE; | |
| 503 | 1772 if (!shortmess(SHM_SEARCH)) |
| 1773 give_warning((char_u *)_(bot_top_msg), TRUE); | |
| 500 | 1774 } |
| 1775 | |
| 1776 /* If we are back at the starting line and there is no match then | |
| 1777 * 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
|
1778 if (lnum == wp->w_cursor.lnum && !found_one) |
| 346 | 1779 break; |
| 1780 | |
| 1781 /* Skip the characters at the start of the next line that were | |
| 1782 * included in a match crossing line boundaries. */ | |
| 534 | 1783 if (attr == HLF_COUNT) |
| 835 | 1784 skip = (int)(p - endp); |
| 346 | 1785 else |
| 1786 skip = 0; | |
| 385 | 1787 |
| 835 | 1788 /* Capcol skips over the inserted space. */ |
| 385 | 1789 --capcol; |
| 1790 | |
| 1791 /* But after empty line check first word in next line */ | |
| 1792 if (*skipwhite(line) == NUL) | |
| 1793 capcol = 0; | |
| 249 | 1794 } |
| 236 | 1795 |
| 1796 line_breakcheck(); | |
| 1797 } | |
| 1798 | |
| 346 | 1799 vim_free(buf); |
| 495 | 1800 return 0; |
| 346 | 1801 } |
| 1802 | |
| 1803 /* | |
| 1804 * For spell checking: concatenate the start of the following line "line" into | |
| 1805 * "buf", blanking-out special characters. Copy less then "maxlen" bytes. | |
| 1577 | 1806 * Keep the blanks at the start of the next line, this is used in win_line() |
| 1807 * to skip those bytes if the word was OK. | |
| 346 | 1808 */ |
| 1809 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1810 spell_cat_line(char_u *buf, char_u *line, int maxlen) |
| 346 | 1811 { |
| 1812 char_u *p; | |
| 1813 int n; | |
| 1814 | |
| 1815 p = skipwhite(line); | |
| 1816 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL) | |
| 1817 p = skipwhite(p + 1); | |
| 1818 | |
| 1819 if (*p != NUL) | |
| 1820 { | |
| 1577 | 1821 /* Only worth concatenating if there is something else than spaces to |
| 1822 * concatenate. */ | |
| 1823 n = (int)(p - line) + 1; | |
| 1824 if (n < maxlen - 1) | |
| 1825 { | |
| 1826 vim_memset(buf, ' ', n); | |
| 1827 vim_strncpy(buf + n, p, maxlen - 1 - n); | |
| 1828 } | |
| 346 | 1829 } |
| 236 | 1830 } |
| 1831 | |
| 626 | 1832 /* |
| 1833 * Structure used for the cookie argument of do_in_runtimepath(). | |
| 1834 */ | |
| 501 | 1835 typedef struct spelload_S |
| 1836 { | |
| 1837 char_u sl_lang[MAXWLEN + 1]; /* language name */ | |
| 1838 slang_T *sl_slang; /* resulting slang_T struct */ | |
| 1839 int sl_nobreak; /* NOBREAK language found */ | |
| 1840 } spelload_T; | |
| 1841 | |
| 236 | 1842 /* |
| 307 | 1843 * Load word list(s) for "lang" from Vim spell file(s). |
| 310 | 1844 * "lang" must be the language without the region: e.g., "en". |
| 236 | 1845 */ |
| 307 | 1846 static void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1847 spell_load_lang(char_u *lang) |
| 236 | 1848 { |
| 310 | 1849 char_u fname_enc[85]; |
| 236 | 1850 int r; |
| 501 | 1851 spelload_T sl; |
| 649 | 1852 #ifdef FEAT_AUTOCMD |
| 1853 int round; | |
| 1854 #endif | |
| 307 | 1855 |
| 310 | 1856 /* Copy the language name to pass it to spell_load_cb() as a cookie. |
| 307 | 1857 * It's truncated when an error is detected. */ |
| 501 | 1858 STRCPY(sl.sl_lang, lang); |
| 1859 sl.sl_slang = NULL; | |
| 1860 sl.sl_nobreak = FALSE; | |
| 307 | 1861 |
| 649 | 1862 #ifdef FEAT_AUTOCMD |
| 1863 /* We may retry when no spell file is found for the language, an | |
| 1864 * autocommand may load it then. */ | |
| 1865 for (round = 1; round <= 2; ++round) | |
| 1866 #endif | |
| 1867 { | |
| 1868 /* | |
| 1869 * Find the first spell file for "lang" in 'runtimepath' and load it. | |
| 1870 */ | |
| 1871 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 1872 #ifdef VMS |
| 1873 "spell/%s_%s.spl", | |
| 1874 #else | |
| 1875 "spell/%s.%s.spl", | |
| 1876 #endif | |
| 1877 lang, spell_enc()); | |
|
8524
2f57bbe870ea
commit https://github.com/vim/vim/commit/7f8989dd8a627af2185df381195351a913f3777f
Christian Brabandt <cb@256bit.org>
parents:
7835
diff
changeset
|
1878 r = do_in_runtimepath(fname_enc, 0, spell_load_cb, &sl); |
| 649 | 1879 |
| 1880 if (r == FAIL && *sl.sl_lang != NUL) | |
| 1881 { | |
| 1882 /* Try loading the ASCII version. */ | |
| 1883 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 1884 #ifdef VMS |
| 1885 "spell/%s_ascii.spl", | |
| 1886 #else | |
| 1887 "spell/%s.ascii.spl", | |
| 1888 #endif | |
| 1889 lang); | |
|
8524
2f57bbe870ea
commit https://github.com/vim/vim/commit/7f8989dd8a627af2185df381195351a913f3777f
Christian Brabandt <cb@256bit.org>
parents:
7835
diff
changeset
|
1890 r = do_in_runtimepath(fname_enc, 0, spell_load_cb, &sl); |
| 649 | 1891 |
| 1892 #ifdef FEAT_AUTOCMD | |
| 1893 if (r == FAIL && *sl.sl_lang != NUL && round == 1 | |
| 1894 && apply_autocmds(EVENT_SPELLFILEMISSING, lang, | |
| 1895 curbuf->b_fname, FALSE, curbuf)) | |
| 1896 continue; | |
| 1897 break; | |
| 1898 #endif | |
| 1899 } | |
| 714 | 1900 #ifdef FEAT_AUTOCMD |
| 1901 break; | |
| 1902 #endif | |
| 307 | 1903 } |
| 1904 | |
| 1905 if (r == FAIL) | |
| 649 | 1906 { |
| 2660 | 1907 smsg((char_u *) |
| 1908 #ifdef VMS | |
| 1909 _("Warning: Cannot find word list \"%s_%s.spl\" or \"%s_ascii.spl\""), | |
| 1910 #else | |
| 1911 _("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""), | |
| 1912 #endif | |
| 483 | 1913 lang, spell_enc(), lang); |
| 649 | 1914 } |
| 501 | 1915 else if (sl.sl_slang != NULL) |
| 1916 { | |
| 626 | 1917 /* At least one file was loaded, now load ALL the additions. */ |
| 310 | 1918 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
|
1919 do_in_runtimepath(fname_enc, DIP_ALL, spell_load_cb, &sl); |
| 310 | 1920 } |
| 1921 } | |
| 1922 | |
| 1923 /* | |
| 1924 * Return the encoding used for spell checking: Use 'encoding', except that we | |
| 1925 * use "latin1" for "latin9". And limit to 60 characters (just in case). | |
| 1926 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1927 char_u * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1928 spell_enc(void) |
| 310 | 1929 { |
| 1930 | |
| 1931 #ifdef FEAT_MBYTE | |
| 1932 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0) | |
| 1933 return p_enc; | |
| 1934 #endif | |
| 1935 return (char_u *)"latin1"; | |
| 236 | 1936 } |
| 1937 | |
| 1938 /* | |
| 385 | 1939 * Get the name of the .spl file for the internal wordlist into |
| 1940 * "fname[MAXPATHL]". | |
| 1941 */ | |
| 1942 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1943 int_wordlist_spl(char_u *fname) |
| 385 | 1944 { |
| 2660 | 1945 vim_snprintf((char *)fname, MAXPATHL, SPL_FNAME_TMPL, |
| 385 | 1946 int_wordlist, spell_enc()); |
| 1947 } | |
| 1948 | |
| 1949 /* | |
| 625 | 1950 * Allocate a new slang_T for language "lang". "lang" can be NULL. |
| 236 | 1951 * Caller must fill "sl_next". |
| 1952 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1953 slang_T * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1954 slang_alloc(char_u *lang) |
| 236 | 1955 { |
| 1956 slang_T *lp; | |
| 1957 | |
| 300 | 1958 lp = (slang_T *)alloc_clear(sizeof(slang_T)); |
| 236 | 1959 if (lp != NULL) |
| 1960 { | |
| 625 | 1961 if (lang != NULL) |
| 1962 lp->sl_name = vim_strsave(lang); | |
| 323 | 1963 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10); |
| 625 | 1964 ga_init2(&lp->sl_repsal, sizeof(fromto_T), 10); |
| 483 | 1965 lp->sl_compmax = MAXWLEN; |
| 1966 lp->sl_compsylmax = MAXWLEN; | |
| 625 | 1967 hash_init(&lp->sl_wordcount); |
| 1968 } | |
| 1969 | |
| 236 | 1970 return lp; |
| 1971 } | |
| 1972 | |
| 1973 /* | |
| 1974 * Free the contents of an slang_T and the structure itself. | |
| 1975 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1976 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1977 slang_free(slang_T *lp) |
| 236 | 1978 { |
| 1979 vim_free(lp->sl_name); | |
| 310 | 1980 vim_free(lp->sl_fname); |
| 1981 slang_clear(lp); | |
| 1982 vim_free(lp); | |
| 1983 } | |
| 1984 | |
| 1985 /* | |
| 1986 * Clear an slang_T so that the file can be reloaded. | |
| 1987 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1988 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1989 slang_clear(slang_T *lp) |
| 310 | 1990 { |
| 339 | 1991 garray_T *gap; |
| 1992 fromto_T *ftp; | |
| 344 | 1993 salitem_T *smp; |
| 339 | 1994 int i; |
| 625 | 1995 int round; |
| 323 | 1996 |
|
13244
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_fbyts); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1998 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
|
1999 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
|
2000 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2001 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
|
2002 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
|
2003 VIM_CLEAR(lp->sl_pidxs); |
| 323 | 2004 |
| 625 | 2005 for (round = 1; round <= 2; ++round) |
| 2006 { | |
| 2007 gap = round == 1 ? &lp->sl_rep : &lp->sl_repsal; | |
| 2008 while (gap->ga_len > 0) | |
| 2009 { | |
| 2010 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len]; | |
| 2011 vim_free(ftp->ft_from); | |
| 2012 vim_free(ftp->ft_to); | |
| 2013 } | |
| 2014 ga_clear(gap); | |
| 2015 } | |
| 344 | 2016 |
| 2017 gap = &lp->sl_sal; | |
| 375 | 2018 if (lp->sl_sofo) |
| 376 | 2019 { |
| 2020 /* "ga_len" is set to 1 without adding an item for latin1 */ | |
| 2021 if (gap->ga_data != NULL) | |
| 2022 /* SOFOFROM and SOFOTO items: free lists of wide characters. */ | |
| 2023 for (i = 0; i < gap->ga_len; ++i) | |
| 2024 vim_free(((int **)gap->ga_data)[i]); | |
| 2025 } | |
| 375 | 2026 else |
| 2027 /* SAL items: free salitem_T items */ | |
| 2028 while (gap->ga_len > 0) | |
| 2029 { | |
| 2030 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len]; | |
| 2031 vim_free(smp->sm_lead); | |
| 2032 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */ | |
| 2033 vim_free(smp->sm_to); | |
| 2034 #ifdef FEAT_MBYTE | |
| 2035 vim_free(smp->sm_lead_w); | |
| 2036 vim_free(smp->sm_oneof_w); | |
| 2037 vim_free(smp->sm_to_w); | |
| 2038 #endif | |
| 2039 } | |
| 344 | 2040 ga_clear(gap); |
| 323 | 2041 |
| 339 | 2042 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
|
2043 vim_regfree(lp->sl_prefprog[i]); |
| 376 | 2044 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
|
2045 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
|
2046 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2047 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
|
2048 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2049 VIM_CLEAR(lp->sl_midword); |
| 339 | 2050 |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4352
diff
changeset
|
2051 vim_regfree(lp->sl_compprog); |
| 483 | 2052 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
|
2053 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
|
2054 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
|
2055 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
|
2056 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2057 VIM_CLEAR(lp->sl_syllable); |
| 483 | 2058 ga_clear(&lp->sl_syl_items); |
| 481 | 2059 |
| 809 | 2060 ga_clear_strings(&lp->sl_comppat); |
| 2061 | |
| 625 | 2062 hash_clear_all(&lp->sl_wordcount, WC_KEY_OFF); |
| 2063 hash_init(&lp->sl_wordcount); | |
| 2064 | |
| 2065 #ifdef FEAT_MBYTE | |
| 2066 hash_clear_all(&lp->sl_map_hash, 0); | |
| 2067 #endif | |
| 2068 | |
| 2069 /* Clear info from .sug file. */ | |
| 2070 slang_clear_sug(lp); | |
| 483 | 2071 |
| 2072 lp->sl_compmax = MAXWLEN; | |
| 501 | 2073 lp->sl_compminlen = 0; |
| 483 | 2074 lp->sl_compsylmax = MAXWLEN; |
| 2075 lp->sl_regions[0] = NUL; | |
| 236 | 2076 } |
| 2077 | |
| 2078 /* | |
| 625 | 2079 * Clear the info from the .sug file in "lp". |
| 2080 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2081 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2082 slang_clear_sug(slang_T *lp) |
| 625 | 2083 { |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2084 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
|
2085 VIM_CLEAR(lp->sl_sidxs); |
| 625 | 2086 close_spellbuf(lp->sl_sugbuf); |
| 2087 lp->sl_sugbuf = NULL; | |
| 2088 lp->sl_sugloaded = FALSE; | |
| 2089 lp->sl_sugtime = 0; | |
| 2090 } | |
| 2091 | |
| 2092 /* | |
| 307 | 2093 * Load one spell file and store the info into a slang_T. |
| 236 | 2094 * Invoked through do_in_runtimepath(). |
| 2095 */ | |
| 2096 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2097 spell_load_cb(char_u *fname, void *cookie) |
| 501 | 2098 { |
| 2099 spelload_T *slp = (spelload_T *)cookie; | |
| 2100 slang_T *slang; | |
| 2101 | |
| 2102 slang = spell_load_file(fname, slp->sl_lang, NULL, FALSE); | |
| 2103 if (slang != NULL) | |
| 2104 { | |
| 2105 /* When a previously loaded file has NOBREAK also use it for the | |
| 2106 * ".add" files. */ | |
| 2107 if (slp->sl_nobreak && slang->sl_add) | |
| 2108 slang->sl_nobreak = TRUE; | |
| 2109 else if (slang->sl_nobreak) | |
| 2110 slp->sl_nobreak = TRUE; | |
| 2111 | |
| 2112 slp->sl_slang = slang; | |
| 2113 } | |
| 310 | 2114 } |
| 2115 | |
| 625 | 2116 |
| 2117 /* | |
| 2118 * Add a word to the hashtable of common words. | |
| 2119 * If it's already there then the counter is increased. | |
| 2120 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2121 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2122 count_common_word( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2123 slang_T *lp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2124 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2125 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
|
2126 int count) /* 1 to count once, 10 to init */ |
| 625 | 2127 { |
| 2128 hash_T hash; | |
| 2129 hashitem_T *hi; | |
| 2130 wordcount_T *wc; | |
| 2131 char_u buf[MAXWLEN]; | |
| 2132 char_u *p; | |
| 2133 | |
| 2134 if (len == -1) | |
| 2135 p = word; | |
| 2136 else | |
| 2137 { | |
| 2138 vim_strncpy(buf, word, len); | |
| 2139 p = buf; | |
| 2140 } | |
| 2141 | |
| 2142 hash = hash_hash(p); | |
| 2143 hi = hash_lookup(&lp->sl_wordcount, p, hash); | |
| 2144 if (HASHITEM_EMPTY(hi)) | |
| 2145 { | |
| 835 | 2146 wc = (wordcount_T *)alloc((unsigned)(sizeof(wordcount_T) + STRLEN(p))); |
| 625 | 2147 if (wc == NULL) |
| 2148 return; | |
| 2149 STRCPY(wc->wc_word, p); | |
| 2150 wc->wc_count = count; | |
| 2151 hash_add_item(&lp->sl_wordcount, hi, wc->wc_word, hash); | |
| 2152 } | |
| 2153 else | |
| 2154 { | |
| 2155 wc = HI2WC(hi); | |
| 2156 if ((wc->wc_count += count) < (unsigned)count) /* check for overflow */ | |
| 2157 wc->wc_count = MAXWORDCOUNT; | |
| 2158 } | |
| 2159 } | |
| 2160 | |
| 2161 /* | |
| 2162 * Adjust the score of common words. | |
| 2163 */ | |
| 2164 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2165 score_wordcount_adj( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2166 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2167 int score, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2168 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2169 int split) /* word was split, less bonus */ |
| 625 | 2170 { |
| 2171 hashitem_T *hi; | |
| 2172 wordcount_T *wc; | |
| 2173 int bonus; | |
| 2174 int newscore; | |
| 2175 | |
| 2176 hi = hash_find(&slang->sl_wordcount, word); | |
| 2177 if (!HASHITEM_EMPTY(hi)) | |
| 2178 { | |
| 2179 wc = HI2WC(hi); | |
| 2180 if (wc->wc_count < SCORE_THRES2) | |
| 2181 bonus = SCORE_COMMON1; | |
| 2182 else if (wc->wc_count < SCORE_THRES3) | |
| 2183 bonus = SCORE_COMMON2; | |
| 2184 else | |
| 2185 bonus = SCORE_COMMON3; | |
| 2186 if (split) | |
| 2187 newscore = score - bonus / 2; | |
| 2188 else | |
| 2189 newscore = score - bonus; | |
| 2190 if (newscore < 0) | |
| 2191 return 0; | |
| 2192 return newscore; | |
| 2193 } | |
| 2194 return score; | |
| 2195 } | |
| 2196 | |
| 483 | 2197 |
| 495 | 2198 /* |
| 498 | 2199 * Return TRUE if byte "n" appears in "str". |
| 495 | 2200 * Like strchr() but independent of locale. |
| 2201 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2202 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2203 byte_in_str(char_u *str, int n) |
| 495 | 2204 { |
| 2205 char_u *p; | |
| 2206 | |
| 2207 for (p = str; *p != NUL; ++p) | |
| 498 | 2208 if (*p == n) |
| 495 | 2209 return TRUE; |
| 2210 return FALSE; | |
| 2211 } | |
| 2212 | |
| 483 | 2213 #define SY_MAXLEN 30 |
| 2214 typedef struct syl_item_S | |
| 2215 { | |
| 2216 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */ | |
| 2217 int sy_len; | |
| 2218 } syl_item_T; | |
| 2219 | |
| 2220 /* | |
| 2221 * Truncate "slang->sl_syllable" at the first slash and put the following items | |
| 2222 * in "slang->sl_syl_items". | |
| 2223 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2224 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2225 init_syl_tab(slang_T *slang) |
| 483 | 2226 { |
| 2227 char_u *p; | |
| 2228 char_u *s; | |
| 2229 int l; | |
| 2230 syl_item_T *syl; | |
| 2231 | |
| 2232 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4); | |
| 2233 p = vim_strchr(slang->sl_syllable, '/'); | |
| 2234 while (p != NULL) | |
| 2235 { | |
| 2236 *p++ = NUL; | |
| 495 | 2237 if (*p == NUL) /* trailing slash */ |
| 483 | 2238 break; |
| 2239 s = p; | |
| 2240 p = vim_strchr(p, '/'); | |
| 2241 if (p == NULL) | |
| 835 | 2242 l = (int)STRLEN(s); |
| 483 | 2243 else |
| 835 | 2244 l = (int)(p - s); |
| 483 | 2245 if (l >= SY_MAXLEN) |
| 2246 return SP_FORMERROR; | |
| 2247 if (ga_grow(&slang->sl_syl_items, 1) == FAIL) | |
| 495 | 2248 return SP_OTHERERROR; |
| 483 | 2249 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) |
| 2250 + slang->sl_syl_items.ga_len++; | |
| 2251 vim_strncpy(syl->sy_chars, s, l); | |
| 2252 syl->sy_len = l; | |
| 2253 } | |
| 2254 return OK; | |
| 2255 } | |
| 2256 | |
| 2257 /* | |
| 2258 * Count the number of syllables in "word". | |
| 2259 * When "word" contains spaces the syllables after the last space are counted. | |
| 2260 * Returns zero if syllables are not defines. | |
| 2261 */ | |
| 2262 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2263 count_syllables(slang_T *slang, char_u *word) |
| 483 | 2264 { |
| 2265 int cnt = 0; | |
| 2266 int skip = FALSE; | |
| 2267 char_u *p; | |
| 2268 int len; | |
| 2269 int i; | |
| 2270 syl_item_T *syl; | |
| 2271 int c; | |
| 2272 | |
| 2273 if (slang->sl_syllable == NULL) | |
| 2274 return 0; | |
| 2275 | |
| 2276 for (p = word; *p != NUL; p += len) | |
| 2277 { | |
| 2278 /* When running into a space reset counter. */ | |
| 2279 if (*p == ' ') | |
| 2280 { | |
| 2281 len = 1; | |
| 2282 cnt = 0; | |
| 2283 continue; | |
| 2284 } | |
| 2285 | |
| 2286 /* Find longest match of syllable items. */ | |
| 2287 len = 0; | |
| 2288 for (i = 0; i < slang->sl_syl_items.ga_len; ++i) | |
| 2289 { | |
| 2290 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i; | |
| 2291 if (syl->sy_len > len | |
| 2292 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0) | |
| 2293 len = syl->sy_len; | |
| 2294 } | |
| 2295 if (len != 0) /* found a match, count syllable */ | |
| 2296 { | |
| 2297 ++cnt; | |
| 2298 skip = FALSE; | |
| 2299 } | |
| 2300 else | |
| 2301 { | |
| 2302 /* No recognized syllable item, at least a syllable char then? */ | |
| 2303 #ifdef FEAT_MBYTE | |
| 2304 c = mb_ptr2char(p); | |
| 2305 len = (*mb_ptr2len)(p); | |
| 2306 #else | |
| 2307 c = *p; | |
| 2308 len = 1; | |
| 2309 #endif | |
| 2310 if (vim_strchr(slang->sl_syllable, c) == NULL) | |
| 2311 skip = FALSE; /* No, search for next syllable */ | |
| 2312 else if (!skip) | |
| 2313 { | |
| 2314 ++cnt; /* Yes, count it */ | |
| 2315 skip = TRUE; /* don't count following syllable chars */ | |
| 2316 } | |
| 2317 } | |
| 2318 } | |
| 2319 return cnt; | |
| 376 | 2320 } |
| 2321 | |
| 381 | 2322 /* |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2323 * Parse 'spelllang' and set w_s->b_langp accordingly. |
| 351 | 2324 * Returns NULL if it's OK, an error message otherwise. |
| 236 | 2325 */ |
| 2326 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2327 did_set_spelllang(win_T *wp) |
| 236 | 2328 { |
| 2329 garray_T ga; | |
| 351 | 2330 char_u *splp; |
| 236 | 2331 char_u *region; |
| 416 | 2332 char_u region_cp[3]; |
| 355 | 2333 int filename; |
| 236 | 2334 int region_mask; |
| 503 | 2335 slang_T *slang; |
| 236 | 2336 int c; |
| 351 | 2337 char_u lang[MAXWLEN + 1]; |
| 323 | 2338 char_u spf_name[MAXPATHL]; |
| 351 | 2339 int len; |
| 2340 char_u *p; | |
| 381 | 2341 int round; |
| 385 | 2342 char_u *spf; |
| 389 | 2343 char_u *use_region = NULL; |
| 2344 int dont_use_region = FALSE; | |
| 501 | 2345 int nobreak = FALSE; |
| 503 | 2346 int i, j; |
| 2347 langp_T *lp, *lp2; | |
| 1185 | 2348 static int recursive = FALSE; |
| 2349 char_u *ret_msg = NULL; | |
| 2350 char_u *spl_copy; | |
|
9487
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2351 #ifdef FEAT_AUTOCMD |
|
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2352 bufref_T bufref; |
|
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2353 |
|
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2354 set_bufref(&bufref, wp->w_buffer); |
|
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2355 #endif |
| 1185 | 2356 |
| 2357 /* We don't want to do this recursively. May happen when a language is | |
| 2358 * not available and the SpellFileMissing autocommand opens a new buffer | |
| 2359 * in which 'spell' is set. */ | |
| 2360 if (recursive) | |
| 2361 return NULL; | |
| 2362 recursive = TRUE; | |
| 236 | 2363 |
| 2364 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
|
2365 clear_midword(wp); |
| 236 | 2366 |
| 4352 | 2367 /* Make a copy of 'spelllang', the SpellFileMissing autocommands may change |
| 1185 | 2368 * it under our fingers. */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2369 spl_copy = vim_strsave(wp->w_s->b_p_spl); |
| 1185 | 2370 if (spl_copy == NULL) |
| 2371 goto theend; | |
| 2372 | |
| 5485 | 2373 #ifdef FEAT_MBYTE |
| 5477 | 2374 wp->w_s->b_cjk = 0; |
| 5485 | 2375 #endif |
| 5477 | 2376 |
| 2377 /* Loop over comma separated language names. */ | |
| 1185 | 2378 for (splp = spl_copy; *splp != NUL; ) |
| 351 | 2379 { |
| 2380 /* Get one language name. */ | |
| 2381 copy_option_part(&splp, lang, MAXWLEN, ","); | |
| 240 | 2382 region = NULL; |
| 835 | 2383 len = (int)STRLEN(lang); |
| 355 | 2384 |
| 5477 | 2385 if (STRCMP(lang, "cjk") == 0) |
| 2386 { | |
| 5485 | 2387 #ifdef FEAT_MBYTE |
| 5477 | 2388 wp->w_s->b_cjk = 1; |
| 5485 | 2389 #endif |
| 5477 | 2390 continue; |
| 2391 } | |
| 2392 | |
| 355 | 2393 /* If the name ends in ".spl" use it as the name of the spell file. |
| 2394 * If there is a region name let "region" point to it and remove it | |
| 2395 * from the name. */ | |
| 2396 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0) | |
| 236 | 2397 { |
| 355 | 2398 filename = TRUE; |
| 2399 | |
| 416 | 2400 /* Locate a region and remove it from the file name. */ |
| 2401 p = vim_strchr(gettail(lang), '_'); | |
| 2402 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2]) | |
| 2403 && !ASCII_ISALPHA(p[3])) | |
| 2404 { | |
| 2405 vim_strncpy(region_cp, p + 1, 2); | |
| 2406 mch_memmove(p, p + 3, len - (p - lang) - 2); | |
| 2407 len -= 3; | |
| 2408 region = region_cp; | |
| 2409 } | |
| 2410 else | |
| 2411 dont_use_region = TRUE; | |
| 2412 | |
| 355 | 2413 /* Check if we loaded this language before. */ |
| 503 | 2414 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2415 if (fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME) | |
| 355 | 2416 break; |
| 236 | 2417 } |
| 355 | 2418 else |
| 2419 { | |
| 2420 filename = FALSE; | |
| 2421 if (len > 3 && lang[len - 3] == '_') | |
| 2422 { | |
| 2423 region = lang + len - 2; | |
| 2424 len -= 3; | |
| 2425 lang[len] = NUL; | |
| 389 | 2426 } |
| 2427 else | |
| 2428 dont_use_region = TRUE; | |
| 355 | 2429 |
| 2430 /* Check if we loaded this language before. */ | |
| 503 | 2431 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2432 if (STRICMP(lang, slang->sl_name) == 0) | |
| 355 | 2433 break; |
| 2434 } | |
| 236 | 2435 |
| 416 | 2436 if (region != NULL) |
| 2437 { | |
| 2438 /* If the region differs from what was used before then don't | |
| 2439 * use it for 'spellfile'. */ | |
| 2440 if (use_region != NULL && STRCMP(region, use_region) != 0) | |
| 2441 dont_use_region = TRUE; | |
| 2442 use_region = region; | |
| 2443 } | |
| 2444 | |
| 351 | 2445 /* If not found try loading the language now. */ |
| 503 | 2446 if (slang == NULL) |
| 355 | 2447 { |
| 2448 if (filename) | |
| 2449 (void)spell_load_file(lang, lang, NULL, FALSE); | |
| 2450 else | |
| 1185 | 2451 { |
| 355 | 2452 spell_load_lang(lang); |
| 1185 | 2453 #ifdef FEAT_AUTOCMD |
| 2454 /* SpellFileMissing autocommands may do anything, including | |
| 2455 * 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
|
2456 if (!bufref_valid(&bufref)) |
| 1185 | 2457 { |
|
9953
9560a2eb7968
commit https://github.com/vim/vim/commit/5b30291785e6b9be1a607504c14bd03c601b59a6
Christian Brabandt <cb@256bit.org>
parents:
9898
diff
changeset
|
2458 ret_msg = (char_u *)N_("E797: SpellFileMissing autocommand deleted buffer"); |
| 1185 | 2459 goto theend; |
| 2460 } | |
| 2461 #endif | |
| 2462 } | |
| 355 | 2463 } |
| 236 | 2464 |
| 307 | 2465 /* |
| 351 | 2466 * Loop over the languages, there can be several files for "lang". |
| 307 | 2467 */ |
| 503 | 2468 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2469 if (filename ? fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME | |
| 2470 : STRICMP(lang, slang->sl_name) == 0) | |
| 236 | 2471 { |
| 316 | 2472 region_mask = REGION_ALL; |
| 355 | 2473 if (!filename && region != NULL) |
| 236 | 2474 { |
| 307 | 2475 /* find region in sl_regions */ |
| 503 | 2476 c = find_region(slang->sl_regions, region); |
| 307 | 2477 if (c == REGION_ALL) |
| 2478 { | |
| 503 | 2479 if (slang->sl_add) |
| 389 | 2480 { |
| 503 | 2481 if (*slang->sl_regions != NUL) |
| 389 | 2482 /* This addition file is for other regions. */ |
| 2483 region_mask = 0; | |
| 2484 } | |
| 2485 else | |
| 2486 /* This is probably an error. Give a warning and | |
| 2487 * accept the words anyway. */ | |
| 351 | 2488 smsg((char_u *) |
| 2489 _("Warning: region %s not supported"), | |
| 2490 region); | |
| 307 | 2491 } |
| 2492 else | |
| 2493 region_mask = 1 << c; | |
| 236 | 2494 } |
| 307 | 2495 |
| 389 | 2496 if (region_mask != 0) |
| 307 | 2497 { |
| 389 | 2498 if (ga_grow(&ga, 1) == FAIL) |
| 2499 { | |
| 2500 ga_clear(&ga); | |
| 1185 | 2501 ret_msg = e_outofmem; |
| 2502 goto theend; | |
| 389 | 2503 } |
| 503 | 2504 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 389 | 2505 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 2506 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2507 use_midword(slang, wp); |
| 503 | 2508 if (slang->sl_nobreak) |
| 501 | 2509 nobreak = TRUE; |
| 307 | 2510 } |
| 385 | 2511 } |
| 2512 } | |
| 2513 | |
| 2514 /* round 0: load int_wordlist, if possible. | |
| 2515 * round 1: load first name in 'spellfile'. | |
| 2516 * round 2: load second name in 'spellfile. | |
| 2517 * etc. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2518 spf = curwin->w_s->b_p_spf; |
| 385 | 2519 for (round = 0; round == 0 || *spf != NUL; ++round) |
| 2520 { | |
| 2521 if (round == 0) | |
| 2522 { | |
| 2523 /* Internal wordlist, if there is one. */ | |
| 2524 if (int_wordlist == NULL) | |
| 381 | 2525 continue; |
| 385 | 2526 int_wordlist_spl(spf_name); |
| 381 | 2527 } |
| 2528 else | |
| 2529 { | |
| 385 | 2530 /* One entry in 'spellfile'. */ |
| 2531 copy_option_part(&spf, spf_name, MAXPATHL - 5, ","); | |
| 2532 STRCAT(spf_name, ".spl"); | |
| 2533 | |
| 2534 /* If it was already found above then skip it. */ | |
| 2535 for (c = 0; c < ga.ga_len; ++c) | |
| 500 | 2536 { |
| 2537 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname; | |
| 2538 if (p != NULL && fullpathcmp(spf_name, p, FALSE) == FPC_SAME) | |
| 385 | 2539 break; |
| 500 | 2540 } |
| 385 | 2541 if (c < ga.ga_len) |
| 381 | 2542 continue; |
| 2543 } | |
| 2544 | |
| 351 | 2545 /* Check if it was loaded already. */ |
| 503 | 2546 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2547 if (fullpathcmp(spf_name, slang->sl_fname, FALSE) == FPC_SAME) | |
| 323 | 2548 break; |
| 503 | 2549 if (slang == NULL) |
| 323 | 2550 { |
| 351 | 2551 /* Not loaded, try loading it now. The language name includes the |
| 385 | 2552 * region name, the region is ignored otherwise. for int_wordlist |
| 2553 * use an arbitrary name. */ | |
| 2554 if (round == 0) | |
| 2555 STRCPY(lang, "internal wordlist"); | |
| 2556 else | |
| 2557 { | |
| 2558 vim_strncpy(lang, gettail(spf_name), MAXWLEN); | |
| 381 | 2559 p = vim_strchr(lang, '.'); |
| 2560 if (p != NULL) | |
| 2561 *p = NUL; /* truncate at ".encoding.add" */ | |
| 2562 } | |
| 503 | 2563 slang = spell_load_file(spf_name, lang, NULL, TRUE); |
| 501 | 2564 |
| 2565 /* If one of the languages has NOBREAK we assume the addition | |
| 2566 * files also have this. */ | |
| 503 | 2567 if (slang != NULL && nobreak) |
| 2568 slang->sl_nobreak = TRUE; | |
| 2569 } | |
| 2570 if (slang != NULL && ga_grow(&ga, 1) == OK) | |
| 323 | 2571 { |
| 389 | 2572 region_mask = REGION_ALL; |
| 2573 if (use_region != NULL && !dont_use_region) | |
| 2574 { | |
| 2575 /* find region in sl_regions */ | |
| 503 | 2576 c = find_region(slang->sl_regions, use_region); |
| 389 | 2577 if (c != REGION_ALL) |
| 2578 region_mask = 1 << c; | |
| 503 | 2579 else if (*slang->sl_regions != NUL) |
| 389 | 2580 /* This spell file is for other regions. */ |
| 2581 region_mask = 0; | |
| 2582 } | |
| 2583 | |
| 2584 if (region_mask != 0) | |
| 2585 { | |
| 503 | 2586 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 2587 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL; | |
| 2588 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL; | |
| 389 | 2589 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 2590 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2591 use_midword(slang, wp); |
| 389 | 2592 } |
| 323 | 2593 } |
| 2594 } | |
| 2595 | |
| 236 | 2596 /* 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
|
2597 ga_clear(&wp->w_s->b_langp); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2598 wp->w_s->b_langp = ga; |
| 236 | 2599 |
| 503 | 2600 /* For each language figure out what language to use for sound folding and |
| 2601 * REP items. If the language doesn't support it itself use another one | |
| 2602 * with the same name. E.g. for "en-math" use "en". */ | |
| 2603 for (i = 0; i < ga.ga_len; ++i) | |
| 2604 { | |
| 2605 lp = LANGP_ENTRY(ga, i); | |
| 2606 | |
| 2607 /* sound folding */ | |
| 2608 if (lp->lp_slang->sl_sal.ga_len > 0) | |
| 2609 /* language does sound folding itself */ | |
| 2610 lp->lp_sallang = lp->lp_slang; | |
| 2611 else | |
| 2612 /* find first similar language that does sound folding */ | |
| 2613 for (j = 0; j < ga.ga_len; ++j) | |
| 2614 { | |
| 2615 lp2 = LANGP_ENTRY(ga, j); | |
| 2616 if (lp2->lp_slang->sl_sal.ga_len > 0 | |
| 2617 && STRNCMP(lp->lp_slang->sl_name, | |
| 2618 lp2->lp_slang->sl_name, 2) == 0) | |
| 2619 { | |
| 2620 lp->lp_sallang = lp2->lp_slang; | |
| 2621 break; | |
| 2622 } | |
| 2623 } | |
| 2624 | |
| 2625 /* REP items */ | |
| 2626 if (lp->lp_slang->sl_rep.ga_len > 0) | |
| 2627 /* language has REP items itself */ | |
| 2628 lp->lp_replang = lp->lp_slang; | |
| 2629 else | |
| 625 | 2630 /* find first similar language that has REP items */ |
| 503 | 2631 for (j = 0; j < ga.ga_len; ++j) |
| 2632 { | |
| 2633 lp2 = LANGP_ENTRY(ga, j); | |
| 2634 if (lp2->lp_slang->sl_rep.ga_len > 0 | |
| 2635 && STRNCMP(lp->lp_slang->sl_name, | |
| 2636 lp2->lp_slang->sl_name, 2) == 0) | |
| 2637 { | |
| 2638 lp->lp_replang = lp2->lp_slang; | |
| 2639 break; | |
| 2640 } | |
| 2641 } | |
| 2642 } | |
| 2643 | |
| 1185 | 2644 theend: |
| 2645 vim_free(spl_copy); | |
| 2646 recursive = FALSE; | |
| 5891 | 2647 redraw_win_later(wp, NOT_VALID); |
| 1185 | 2648 return ret_msg; |
| 236 | 2649 } |
| 2650 | |
| 2651 /* | |
| 376 | 2652 * Clear the midword characters for buffer "buf". |
| 2653 */ | |
| 2654 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2655 clear_midword(win_T *wp) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2656 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2657 vim_memset(wp->w_s->b_spell_ismw, 0, 256); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2658 #ifdef FEAT_MBYTE |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2659 VIM_CLEAR(wp->w_s->b_spell_ismw_mb); |
| 376 | 2660 #endif |
| 2661 } | |
| 2662 | |
| 2663 /* | |
| 2664 * Use the "sl_midword" field of language "lp" for buffer "buf". | |
| 2665 * They add up to any currently used midword characters. | |
| 2666 */ | |
| 2667 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2668 use_midword(slang_T *lp, win_T *wp) |
| 376 | 2669 { |
| 2670 char_u *p; | |
| 2671 | |
| 389 | 2672 if (lp->sl_midword == NULL) /* there aren't any */ |
| 2673 return; | |
| 2674 | |
| 376 | 2675 for (p = lp->sl_midword; *p != NUL; ) |
| 2676 #ifdef FEAT_MBYTE | |
| 2677 if (has_mbyte) | |
| 2678 { | |
| 2679 int c, l, n; | |
| 2680 char_u *bp; | |
| 2681 | |
| 2682 c = mb_ptr2char(p); | |
| 474 | 2683 l = (*mb_ptr2len)(p); |
| 2684 if (c < 256 && l <= 2) | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2685 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
|
2686 else if (wp->w_s->b_spell_ismw_mb == NULL) |
| 376 | 2687 /* 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
|
2688 wp->w_s->b_spell_ismw_mb = vim_strnsave(p, l); |
| 376 | 2689 else |
| 2690 { | |
| 2691 /* 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
|
2692 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
|
2693 bp = vim_strnsave(wp->w_s->b_spell_ismw_mb, n + l); |
| 376 | 2694 if (bp != NULL) |
| 2695 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2696 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
|
2697 wp->w_s->b_spell_ismw_mb = bp; |
| 376 | 2698 vim_strncpy(bp + n, p, l); |
| 2699 } | |
| 2700 } | |
| 2701 p += l; | |
| 2702 } | |
| 2703 else | |
| 2704 #endif | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2705 wp->w_s->b_spell_ismw[*p++] = TRUE; |
| 376 | 2706 } |
| 2707 | |
| 2708 /* | |
| 236 | 2709 * Find the region "region[2]" in "rp" (points to "sl_regions"). |
| 2710 * Each region is simply stored as the two characters of it's name. | |
| 381 | 2711 * Returns the index if found (first is 0), REGION_ALL if not found. |
| 236 | 2712 */ |
| 2713 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2714 find_region(char_u *rp, char_u *region) |
| 236 | 2715 { |
| 2716 int i; | |
| 2717 | |
| 2718 for (i = 0; ; i += 2) | |
| 2719 { | |
| 2720 if (rp[i] == NUL) | |
| 2721 return REGION_ALL; | |
| 2722 if (rp[i] == region[0] && rp[i + 1] == region[1]) | |
| 2723 break; | |
| 2724 } | |
| 2725 return i / 2; | |
| 2726 } | |
| 2727 | |
| 2728 /* | |
| 323 | 2729 * Return case type of word: |
| 236 | 2730 * w word 0 |
| 300 | 2731 * Word WF_ONECAP |
| 2732 * W WORD WF_ALLCAP | |
| 2733 * WoRd wOrd WF_KEEPCAP | |
| 236 | 2734 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2735 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2736 captype( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2737 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2738 char_u *end) /* When NULL use up to NUL byte. */ |
| 236 | 2739 { |
| 2740 char_u *p; | |
| 2741 int c; | |
| 2742 int firstcap; | |
| 2743 int allcap; | |
| 2744 int past_second = FALSE; /* past second word char */ | |
| 2745 | |
| 2746 /* find first letter */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2747 for (p = word; !spell_iswordp_nmw(p, curwin); MB_PTR_ADV(p)) |
| 323 | 2748 if (end == NULL ? *p == NUL : p >= end) |
| 236 | 2749 return 0; /* only non-word characters, illegal word */ |
| 2750 #ifdef FEAT_MBYTE | |
| 310 | 2751 if (has_mbyte) |
| 2752 c = mb_ptr2char_adv(&p); | |
| 2753 else | |
| 236 | 2754 #endif |
| 310 | 2755 c = *p++; |
| 324 | 2756 firstcap = allcap = SPELL_ISUPPER(c); |
| 236 | 2757 |
| 2758 /* | |
| 2759 * Need to check all letters to find a word with mixed upper/lower. | |
| 2760 * But a word with an upper char only at start is a ONECAP. | |
| 2761 */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2762 for ( ; end == NULL ? *p != NUL : p < end; MB_PTR_ADV(p)) |
| 5477 | 2763 if (spell_iswordp_nmw(p, curwin)) |
| 236 | 2764 { |
| 455 | 2765 c = PTR2CHAR(p); |
| 324 | 2766 if (!SPELL_ISUPPER(c)) |
| 236 | 2767 { |
| 2768 /* UUl -> KEEPCAP */ | |
| 2769 if (past_second && allcap) | |
| 300 | 2770 return WF_KEEPCAP; |
| 236 | 2771 allcap = FALSE; |
| 2772 } | |
| 2773 else if (!allcap) | |
| 2774 /* UlU -> KEEPCAP */ | |
| 300 | 2775 return WF_KEEPCAP; |
| 236 | 2776 past_second = TRUE; |
| 2777 } | |
| 2778 | |
| 2779 if (allcap) | |
| 300 | 2780 return WF_ALLCAP; |
| 236 | 2781 if (firstcap) |
| 300 | 2782 return WF_ONECAP; |
| 236 | 2783 return 0; |
| 2784 } | |
| 2785 | |
| 474 | 2786 /* |
| 2787 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a | |
| 2788 * capital. So that make_case_word() can turn WOrd into Word. | |
| 2789 * Add ALLCAP for "WOrD". | |
| 2790 */ | |
| 2791 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2792 badword_captype(char_u *word, char_u *end) |
| 474 | 2793 { |
| 2794 int flags = captype(word, end); | |
| 475 | 2795 int c; |
| 474 | 2796 int l, u; |
| 2797 int first; | |
| 2798 char_u *p; | |
| 2799 | |
| 2800 if (flags & WF_KEEPCAP) | |
| 2801 { | |
| 2802 /* Count the number of UPPER and lower case letters. */ | |
| 2803 l = u = 0; | |
| 2804 first = FALSE; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2805 for (p = word; p < end; MB_PTR_ADV(p)) |
| 474 | 2806 { |
| 475 | 2807 c = PTR2CHAR(p); |
| 2808 if (SPELL_ISUPPER(c)) | |
| 474 | 2809 { |
| 2810 ++u; | |
| 2811 if (p == word) | |
| 2812 first = TRUE; | |
| 2813 } | |
| 2814 else | |
| 2815 ++l; | |
| 2816 } | |
| 2817 | |
| 2818 /* If there are more UPPER than lower case letters suggest an | |
| 2819 * ALLCAP word. Otherwise, if the first letter is UPPER then | |
| 2820 * suggest ONECAP. Exception: "ALl" most likely should be "All", | |
| 2821 * require three upper case letters. */ | |
| 2822 if (u > l && u > 2) | |
| 2823 flags |= WF_ALLCAP; | |
| 2824 else if (first) | |
| 2825 flags |= WF_ONECAP; | |
| 633 | 2826 |
| 2827 if (u >= 2 && l >= 2) /* maCARONI maCAroni */ | |
| 2828 flags |= WF_MIXCAP; | |
| 474 | 2829 } |
| 2830 return flags; | |
| 2831 } | |
| 2832 | |
| 5519 | 2833 /* |
| 2834 * Delete the internal wordlist and its .spl file. | |
| 2835 */ | |
| 2836 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2837 spell_delete_wordlist(void) |
| 5519 | 2838 { |
| 2839 char_u fname[MAXPATHL]; | |
| 2840 | |
| 2841 if (int_wordlist != NULL) | |
| 2842 { | |
| 2843 mch_remove(int_wordlist); | |
| 2844 int_wordlist_spl(fname); | |
| 2845 mch_remove(fname); | |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2846 VIM_CLEAR(int_wordlist); |
| 5519 | 2847 } |
| 2848 } | |
| 2849 | |
| 2850 #if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO) | |
| 355 | 2851 /* |
| 2852 * Free all languages. | |
| 2853 */ | |
| 2854 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2855 spell_free_all(void) |
| 355 | 2856 { |
| 503 | 2857 slang_T *slang; |
| 355 | 2858 buf_T *buf; |
| 2859 | |
| 2580 | 2860 /* 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
|
2861 FOR_ALL_BUFFERS(buf) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2862 ga_clear(&buf->b_s.b_langp); |
| 355 | 2863 |
| 2864 while (first_lang != NULL) | |
| 2865 { | |
| 503 | 2866 slang = first_lang; |
| 2867 first_lang = slang->sl_next; | |
| 2868 slang_free(slang); | |
| 355 | 2869 } |
| 366 | 2870 |
| 5519 | 2871 spell_delete_wordlist(); |
| 381 | 2872 |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2873 VIM_CLEAR(repl_to); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2874 VIM_CLEAR(repl_from); |
| 355 | 2875 } |
| 5519 | 2876 #endif |
| 2877 | |
| 2878 #if defined(FEAT_MBYTE) || defined(PROTO) | |
| 236 | 2879 /* |
| 2880 * Clear all spelling tables and reload them. | |
| 307 | 2881 * Used after 'encoding' is set and when ":mkspell" was used. |
| 236 | 2882 */ |
| 2883 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2884 spell_reload(void) |
| 236 | 2885 { |
| 316 | 2886 win_T *wp; |
| 236 | 2887 |
| 358 | 2888 /* Initialize the table for spell_iswordp(). */ |
| 236 | 2889 init_spell_chartab(); |
| 2890 | |
| 2891 /* Unload all allocated memory. */ | |
| 355 | 2892 spell_free_all(); |
| 236 | 2893 |
| 2894 /* 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
|
2895 FOR_ALL_WINDOWS(wp) |
| 236 | 2896 { |
| 316 | 2897 /* Only load the wordlists when 'spelllang' is set and there is a |
| 2898 * 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
|
2899 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
|
2900 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2901 if (wp->w_p_spell) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2902 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2903 (void)did_set_spelllang(wp); |
| 316 | 2904 break; |
| 2905 } | |
| 2906 } | |
| 236 | 2907 } |
| 2908 } | |
| 5519 | 2909 #endif |
| 236 | 2910 |
| 310 | 2911 /* |
| 625 | 2912 * Opposite of offset2bytes(). |
| 2913 * "pp" points to the bytes and is advanced over it. | |
| 2914 * Returns the offset. | |
| 2915 */ | |
| 2916 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2917 bytes2offset(char_u **pp) |
| 625 | 2918 { |
| 2919 char_u *p = *pp; | |
| 2920 int nr; | |
| 2921 int c; | |
| 2922 | |
| 2923 c = *p++; | |
| 2924 if ((c & 0x80) == 0x00) /* 1 byte */ | |
| 2925 { | |
| 2926 nr = c - 1; | |
| 2927 } | |
| 2928 else if ((c & 0xc0) == 0x80) /* 2 bytes */ | |
| 2929 { | |
| 2930 nr = (c & 0x3f) - 1; | |
| 2931 nr = nr * 255 + (*p++ - 1); | |
| 2932 } | |
| 2933 else if ((c & 0xe0) == 0xc0) /* 3 bytes */ | |
| 2934 { | |
| 2935 nr = (c & 0x1f) - 1; | |
| 2936 nr = nr * 255 + (*p++ - 1); | |
| 2937 nr = nr * 255 + (*p++ - 1); | |
| 2938 } | |
| 2939 else /* 4 bytes */ | |
| 2940 { | |
| 2941 nr = (c & 0x0f) - 1; | |
| 2942 nr = nr * 255 + (*p++ - 1); | |
| 2943 nr = nr * 255 + (*p++ - 1); | |
| 2944 nr = nr * 255 + (*p++ - 1); | |
| 2945 } | |
| 2946 | |
| 2947 *pp = p; | |
| 2948 return nr; | |
| 2949 } | |
| 2950 | |
| 2951 | |
| 2952 /* | |
| 2953 * Open a spell buffer. This is a nameless buffer that is not in the buffer | |
| 2954 * list and only contains text lines. Can use a swapfile to reduce memory | |
| 2955 * use. | |
| 2956 * Most other fields are invalid! Esp. watch out for string options being | |
| 2957 * NULL and there is no undo info. | |
| 2958 * Returns NULL when out of memory. | |
| 2959 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2960 buf_T * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2961 open_spellbuf(void) |
| 625 | 2962 { |
| 2963 buf_T *buf; | |
| 2964 | |
| 2965 buf = (buf_T *)alloc_clear(sizeof(buf_T)); | |
| 2966 if (buf != NULL) | |
| 2967 { | |
| 2968 buf->b_spell = TRUE; | |
| 2969 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
|
2970 #ifdef FEAT_CRYPT |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
2971 buf->b_p_key = empty_option; |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
2972 #endif |
| 625 | 2973 ml_open(buf); |
| 2974 ml_open_file(buf); /* create swap file now */ | |
| 2975 } | |
| 2976 return buf; | |
| 2977 } | |
| 2978 | |
| 2979 /* | |
| 2980 * Close the buffer used for spell info. | |
| 2981 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2982 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2983 close_spellbuf(buf_T *buf) |
| 625 | 2984 { |
| 2985 if (buf != NULL) | |
| 2986 { | |
| 2987 ml_close(buf, TRUE); | |
| 2988 vim_free(buf); | |
| 2989 } | |
| 2990 } | |
| 2991 | |
| 307 | 2992 /* |
| 2993 * Init the chartab used for spelling for ASCII. | |
| 2994 * EBCDIC is not supported! | |
| 2995 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2996 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2997 clear_spell_chartab(spelltab_T *sp) |
| 307 | 2998 { |
| 324 | 2999 int i; |
| 307 | 3000 |
| 3001 /* Init everything to FALSE. */ | |
| 3002 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw)); | |
| 3003 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu)); | |
| 3004 for (i = 0; i < 256; ++i) | |
| 324 | 3005 { |
| 307 | 3006 sp->st_fold[i] = i; |
| 324 | 3007 sp->st_upper[i] = i; |
| 3008 } | |
| 307 | 3009 |
| 3010 /* We include digits. A word shouldn't start with a digit, but handling | |
| 3011 * that is done separately. */ | |
| 3012 for (i = '0'; i <= '9'; ++i) | |
| 3013 sp->st_isw[i] = TRUE; | |
| 3014 for (i = 'A'; i <= 'Z'; ++i) | |
| 3015 { | |
| 3016 sp->st_isw[i] = TRUE; | |
| 3017 sp->st_isu[i] = TRUE; | |
| 3018 sp->st_fold[i] = i + 0x20; | |
| 3019 } | |
| 3020 for (i = 'a'; i <= 'z'; ++i) | |
| 324 | 3021 { |
| 307 | 3022 sp->st_isw[i] = TRUE; |
| 324 | 3023 sp->st_upper[i] = i - 0x20; |
| 3024 } | |
| 307 | 3025 } |
| 3026 | |
| 3027 /* | |
| 3028 * Init the chartab used for spelling. Only depends on 'encoding'. | |
| 3029 * Called once while starting up and when 'encoding' changes. | |
| 3030 * The default is to use isalpha(), but the spell file should define the word | |
| 3031 * characters to make it possible that 'encoding' differs from the current | |
| 390 | 3032 * locale. For utf-8 we don't use isalpha() but our own functions. |
| 307 | 3033 */ |
| 3034 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3035 init_spell_chartab(void) |
| 307 | 3036 { |
| 3037 int i; | |
| 3038 | |
| 3039 did_set_spelltab = FALSE; | |
| 3040 clear_spell_chartab(&spelltab); | |
| 3041 #ifdef FEAT_MBYTE | |
| 3042 if (enc_dbcs) | |
| 3043 { | |
| 3044 /* DBCS: assume double-wide characters are word characters. */ | |
| 3045 for (i = 128; i <= 255; ++i) | |
| 3046 if (MB_BYTE2LEN(i) == 2) | |
| 3047 spelltab.st_isw[i] = TRUE; | |
| 3048 } | |
| 324 | 3049 else if (enc_utf8) |
| 3050 { | |
| 3051 for (i = 128; i < 256; ++i) | |
| 3052 { | |
|
2140
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3053 int f = utf_fold(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3054 int u = utf_toupper(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3055 |
| 324 | 3056 spelltab.st_isu[i] = utf_isupper(i); |
| 3057 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
|
3058 /* 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
|
3059 * 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
|
3060 * value for utf-8 to avoid this. */ |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3061 spelltab.st_fold[i] = (f < 256) ? f : i; |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3062 spelltab.st_upper[i] = (u < 256) ? u : i; |
| 324 | 3063 } |
| 3064 } | |
| 307 | 3065 else |
| 3066 #endif | |
| 3067 { | |
| 324 | 3068 /* Rough guess: use locale-dependent library functions. */ |
| 307 | 3069 for (i = 128; i < 256; ++i) |
| 3070 { | |
| 3071 if (MB_ISUPPER(i)) | |
| 3072 { | |
| 324 | 3073 spelltab.st_isw[i] = TRUE; |
| 307 | 3074 spelltab.st_isu[i] = TRUE; |
| 3075 spelltab.st_fold[i] = MB_TOLOWER(i); | |
| 3076 } | |
| 324 | 3077 else if (MB_ISLOWER(i)) |
| 3078 { | |
| 3079 spelltab.st_isw[i] = TRUE; | |
| 3080 spelltab.st_upper[i] = MB_TOUPPER(i); | |
| 3081 } | |
| 307 | 3082 } |
| 3083 } | |
| 3084 } | |
| 3085 | |
| 3086 | |
| 3087 /* | |
| 358 | 3088 * Return TRUE if "p" points to a word character. |
| 366 | 3089 * As a special case we see "midword" characters as word character when it is |
| 358 | 3090 * followed by a word character. This finds they'there but not 'they there'. |
| 366 | 3091 * Thus this only works properly when past the first character of the word. |
| 358 | 3092 */ |
| 3093 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3094 spell_iswordp( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3095 char_u *p, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3096 win_T *wp) /* buffer used */ |
| 358 | 3097 { |
| 366 | 3098 #ifdef FEAT_MBYTE |
| 358 | 3099 char_u *s; |
| 366 | 3100 int l; |
| 3101 int c; | |
| 3102 | |
| 3103 if (has_mbyte) | |
| 3104 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
3105 l = MB_PTR2LEN(p); |
| 358 | 3106 s = p; |
| 366 | 3107 if (l == 1) |
| 3108 { | |
| 3109 /* be quick for ASCII */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3110 if (wp->w_s->b_spell_ismw[*p]) |
| 366 | 3111 s = p + 1; /* skip a mid-word character */ |
| 3112 } | |
| 3113 else | |
| 3114 { | |
| 3115 c = mb_ptr2char(p); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3116 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
|
3117 : (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
|
3118 && vim_strchr(wp->w_s->b_spell_ismw_mb, c) != NULL)) |
| 366 | 3119 s = p + l; |
| 3120 } | |
| 3121 | |
| 390 | 3122 c = mb_ptr2char(s); |
| 3123 if (c > 255) | |
| 5477 | 3124 return spell_mb_isword_class(mb_get_class(s), wp); |
| 390 | 3125 return spelltab.st_isw[c]; |
| 366 | 3126 } |
| 358 | 3127 #endif |
| 366 | 3128 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3129 return spelltab.st_isw[wp->w_s->b_spell_ismw[*p] ? p[1] : p[0]]; |
| 376 | 3130 } |
| 3131 | |
| 3132 /* | |
| 3133 * Return TRUE if "p" points to a word character. | |
| 3134 * Unlike spell_iswordp() this doesn't check for "midword" characters. | |
| 3135 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3136 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3137 spell_iswordp_nmw(char_u *p, win_T *wp) |
| 376 | 3138 { |
| 3139 #ifdef FEAT_MBYTE | |
| 390 | 3140 int c; |
| 3141 | |
| 3142 if (has_mbyte) | |
| 3143 { | |
| 3144 c = mb_ptr2char(p); | |
| 3145 if (c > 255) | |
| 5477 | 3146 return spell_mb_isword_class(mb_get_class(p), wp); |
| 390 | 3147 return spelltab.st_isw[c]; |
| 3148 } | |
| 376 | 3149 #endif |
| 3150 return spelltab.st_isw[*p]; | |
| 358 | 3151 } |
| 3152 | |
| 372 | 3153 #ifdef FEAT_MBYTE |
| 3154 /* | |
| 1580 | 3155 * Return TRUE if word class indicates a word character. |
| 3156 * Only for characters above 255. | |
| 3157 * Unicode subscript and superscript are not considered word characters. | |
| 5477 | 3158 * See also dbcs_class() and utf_class() in mbyte.c. |
| 3159 */ | |
| 3160 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3161 spell_mb_isword_class(int cl, win_T *wp) |
| 5477 | 3162 { |
| 3163 if (wp->w_s->b_cjk) | |
| 3164 /* East Asian characters are not considered word characters. */ | |
| 3165 return cl == 2 || cl == 0x2800; | |
| 1580 | 3166 return cl >= 2 && cl != 0x2070 && cl != 0x2080; |
| 3167 } | |
| 3168 | |
| 3169 /* | |
| 372 | 3170 * Return TRUE if "p" points to a word character. |
| 3171 * Wide version of spell_iswordp(). | |
| 3172 */ | |
| 3173 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3174 spell_iswordp_w(int *p, win_T *wp) |
| 372 | 3175 { |
| 3176 int *s; | |
| 3177 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3178 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
|
3179 : (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
|
3180 && vim_strchr(wp->w_s->b_spell_ismw_mb, *p) != NULL)) |
| 372 | 3181 s = p + 1; |
| 3182 else | |
| 3183 s = p; | |
| 3184 | |
| 390 | 3185 if (*s > 255) |
| 372 | 3186 { |
| 3187 if (enc_utf8) | |
| 5477 | 3188 return spell_mb_isword_class(utf_class(*s), wp); |
| 372 | 3189 if (enc_dbcs) |
| 5477 | 3190 return spell_mb_isword_class( |
| 3191 dbcs_class((unsigned)*s >> 8, *s & 0xff), wp); | |
| 372 | 3192 return 0; |
| 3193 } | |
| 3194 return spelltab.st_isw[*s]; | |
| 3195 } | |
| 3196 #endif | |
| 3197 | |
| 358 | 3198 /* |
| 324 | 3199 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated. |
| 3200 * Uses the character definitions from the .spl file. | |
| 307 | 3201 * When using a multi-byte 'encoding' the length may change! |
| 3202 * Returns FAIL when something wrong. | |
| 3203 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3204 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3205 spell_casefold( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3206 char_u *str, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3207 int len, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3208 char_u *buf, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3209 int buflen) |
| 307 | 3210 { |
| 3211 int i; | |
| 3212 | |
| 3213 if (len >= buflen) | |
| 3214 { | |
| 3215 buf[0] = NUL; | |
| 3216 return FAIL; /* result will not fit */ | |
| 3217 } | |
| 3218 | |
| 3219 #ifdef FEAT_MBYTE | |
| 3220 if (has_mbyte) | |
| 3221 { | |
| 324 | 3222 int outi = 0; |
| 3223 char_u *p; | |
| 307 | 3224 int c; |
| 3225 | |
| 3226 /* Fold one character at a time. */ | |
| 324 | 3227 for (p = str; p < str + len; ) |
| 307 | 3228 { |
| 3229 if (outi + MB_MAXBYTES > buflen) | |
| 3230 { | |
| 3231 buf[outi] = NUL; | |
| 3232 return FAIL; | |
| 3233 } | |
| 474 | 3234 c = mb_cptr2char_adv(&p); |
| 324 | 3235 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi); |
| 307 | 3236 } |
| 3237 buf[outi] = NUL; | |
| 3238 } | |
| 3239 else | |
| 3240 #endif | |
| 3241 { | |
| 3242 /* Be quick for non-multibyte encodings. */ | |
| 3243 for (i = 0; i < len; ++i) | |
| 324 | 3244 buf[i] = spelltab.st_fold[str[i]]; |
| 307 | 3245 buf[i] = NUL; |
| 3246 } | |
| 3247 | |
| 3248 return OK; | |
| 3249 } | |
| 3250 | |
| 625 | 3251 /* values for sps_flags */ |
| 372 | 3252 #define SPS_BEST 1 |
| 3253 #define SPS_FAST 2 | |
| 3254 #define SPS_DOUBLE 4 | |
| 3255 | |
| 625 | 3256 static int sps_flags = SPS_BEST; /* flags from 'spellsuggest' */ |
| 3257 static int sps_limit = 9999; /* max nr of suggestions given */ | |
| 372 | 3258 |
| 3259 /* | |
| 3260 * Check the 'spellsuggest' option. Return FAIL if it's wrong. | |
| 483 | 3261 * Sets "sps_flags" and "sps_limit". |
| 372 | 3262 */ |
| 3263 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3264 spell_check_sps(void) |
| 372 | 3265 { |
| 3266 char_u *p; | |
| 483 | 3267 char_u *s; |
| 372 | 3268 char_u buf[MAXPATHL]; |
| 3269 int f; | |
| 3270 | |
| 3271 sps_flags = 0; | |
| 483 | 3272 sps_limit = 9999; |
| 372 | 3273 |
| 3274 for (p = p_sps; *p != NUL; ) | |
| 3275 { | |
| 3276 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 3277 | |
| 3278 f = 0; | |
| 483 | 3279 if (VIM_ISDIGIT(*buf)) |
| 3280 { | |
| 3281 s = buf; | |
| 3282 sps_limit = getdigits(&s); | |
| 3283 if (*s != NUL && !VIM_ISDIGIT(*s)) | |
| 3284 f = -1; | |
| 3285 } | |
| 3286 else if (STRCMP(buf, "best") == 0) | |
| 372 | 3287 f = SPS_BEST; |
| 3288 else if (STRCMP(buf, "fast") == 0) | |
| 3289 f = SPS_FAST; | |
| 3290 else if (STRCMP(buf, "double") == 0) | |
| 3291 f = SPS_DOUBLE; | |
| 3292 else if (STRNCMP(buf, "expr:", 5) != 0 | |
| 3293 && STRNCMP(buf, "file:", 5) != 0) | |
| 3294 f = -1; | |
| 3295 | |
| 3296 if (f == -1 || (sps_flags != 0 && f != 0)) | |
| 3297 { | |
| 3298 sps_flags = SPS_BEST; | |
| 483 | 3299 sps_limit = 9999; |
| 372 | 3300 return FAIL; |
| 3301 } | |
| 3302 if (f != 0) | |
| 3303 sps_flags = f; | |
| 3304 } | |
| 3305 | |
| 3306 if (sps_flags == 0) | |
| 3307 sps_flags = SPS_BEST; | |
| 3308 | |
| 3309 return OK; | |
| 3310 } | |
| 3311 | |
| 323 | 3312 /* |
| 5374 | 3313 * "z=": Find badly spelled word under or after the cursor. |
| 323 | 3314 * Give suggestions for the properly spelled word. |
| 638 | 3315 * In Visual mode use the highlighted word as the bad word. |
| 485 | 3316 * When "count" is non-zero use that suggestion. |
| 323 | 3317 */ |
| 3318 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3319 spell_suggest(int count) |
| 323 | 3320 { |
| 3321 char_u *line; | |
| 3322 pos_T prev_cursor = curwin->w_cursor; | |
| 3323 char_u wcopy[MAXWLEN + 2]; | |
| 3324 char_u *p; | |
| 3325 int i; | |
| 3326 int c; | |
| 3327 suginfo_T sug; | |
| 3328 suggest_T *stp; | |
| 372 | 3329 int mouse_used; |
| 388 | 3330 int need_cap; |
| 483 | 3331 int limit; |
| 485 | 3332 int selected = count; |
| 638 | 3333 int badlen = 0; |
| 1938 | 3334 int msg_scroll_save = msg_scroll; |
| 638 | 3335 |
| 3336 if (no_spell_checking(curwin)) | |
| 3337 return; | |
| 3338 | |
| 3339 if (VIsual_active) | |
| 3340 { | |
| 3341 /* Use the Visually selected text as the bad word. But reject | |
| 3342 * a multi-line selection. */ | |
| 3343 if (curwin->w_cursor.lnum != VIsual.lnum) | |
| 3344 { | |
| 6949 | 3345 vim_beep(BO_SPELL); |
| 638 | 3346 return; |
| 3347 } | |
| 3348 badlen = (int)curwin->w_cursor.col - (int)VIsual.col; | |
| 3349 if (badlen < 0) | |
| 3350 badlen = -badlen; | |
| 3351 else | |
| 3352 curwin->w_cursor.col = VIsual.col; | |
| 3353 ++badlen; | |
| 3354 end_visual_mode(); | |
| 3355 } | |
| 5735 | 3356 /* Find the start of the badly spelled word. */ |
| 3357 else if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0 | |
| 346 | 3358 || curwin->w_cursor.col > prev_cursor.col) |
| 3359 { | |
| 3360 /* No bad word or it starts after the cursor: use the word under the | |
| 3361 * cursor. */ | |
| 3362 curwin->w_cursor = prev_cursor; | |
| 3363 line = ml_get_curline(); | |
| 3364 p = line + curwin->w_cursor.col; | |
| 3365 /* Backup to before start of word. */ | |
| 5477 | 3366 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
|
3367 MB_PTR_BACK(line, p); |
| 346 | 3368 /* Forward to start of word. */ |
| 5477 | 3369 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
|
3370 MB_PTR_ADV(p); |
| 346 | 3371 |
| 5477 | 3372 if (!spell_iswordp_nmw(p, curwin)) /* No word found. */ |
| 346 | 3373 { |
| 3374 beep_flush(); | |
| 3375 return; | |
| 3376 } | |
| 835 | 3377 curwin->w_cursor.col = (colnr_T)(p - line); |
| 323 | 3378 } |
| 3379 | |
| 344 | 3380 /* Get the word and its length. */ |
| 3381 | |
| 388 | 3382 /* Figure out if the word should be capitalised. */ |
| 475 | 3383 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col); |
| 3384 | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3385 /* 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
|
3386 line = vim_strsave(ml_get_curline()); |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3387 if (line == NULL) |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3388 goto skip; |
| 388 | 3389 |
| 483 | 3390 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in |
| 3391 * 'spellsuggest', whatever is smaller. */ | |
| 3392 if (sps_limit > (int)Rows - 2) | |
| 3393 limit = (int)Rows - 2; | |
| 3394 else | |
| 3395 limit = sps_limit; | |
| 638 | 3396 spell_find_suggest(line + curwin->w_cursor.col, badlen, &sug, limit, |
| 625 | 3397 TRUE, need_cap, TRUE); |
| 323 | 3398 |
| 3399 if (sug.su_ga.ga_len == 0) | |
| 3400 MSG(_("Sorry, no suggestions")); | |
| 485 | 3401 else if (count > 0) |
| 3402 { | |
| 3403 if (count > sug.su_ga.ga_len) | |
| 3404 smsg((char_u *)_("Sorry, only %ld suggestions"), | |
| 3405 (long)sug.su_ga.ga_len); | |
| 3406 } | |
| 323 | 3407 else |
| 3408 { | |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
3409 VIM_CLEAR(repl_from); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
3410 VIM_CLEAR(repl_to); |
| 372 | 3411 |
| 474 | 3412 #ifdef FEAT_RIGHTLEFT |
| 3413 /* When 'rightleft' is set the list is drawn right-left. */ | |
| 3414 cmdmsg_rl = curwin->w_p_rl; | |
| 3415 if (cmdmsg_rl) | |
| 3416 msg_col = Columns - 1; | |
| 3417 #endif | |
| 3418 | |
| 323 | 3419 /* List the suggestions. */ |
| 3420 msg_start(); | |
| 914 | 3421 msg_row = Rows - 1; /* for when 'cmdheight' > 1 */ |
| 372 | 3422 lines_left = Rows; /* avoid more prompt */ |
| 323 | 3423 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"), |
| 3424 sug.su_badlen, sug.su_badptr); | |
| 474 | 3425 #ifdef FEAT_RIGHTLEFT |
| 3426 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0) | |
| 3427 { | |
| 3428 /* And now the rabbit from the high hat: Avoid showing the | |
| 3429 * untranslated message rightleft. */ | |
| 3430 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC", | |
| 3431 sug.su_badlen, sug.su_badptr); | |
| 3432 } | |
| 3433 #endif | |
| 323 | 3434 msg_puts(IObuff); |
| 3435 msg_clr_eos(); | |
| 3436 msg_putchar('\n'); | |
| 346 | 3437 |
| 323 | 3438 msg_scroll = TRUE; |
| 3439 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 3440 { | |
| 344 | 3441 stp = &SUG(sug.su_ga, i); |
| 323 | 3442 |
| 3443 /* The suggested word may replace only part of the bad word, add | |
| 3444 * the not replaced part. */ | |
| 2768 | 3445 vim_strncpy(wcopy, stp->st_word, MAXWLEN); |
| 323 | 3446 if (sug.su_badlen > stp->st_orglen) |
| 625 | 3447 vim_strncpy(wcopy + stp->st_wordlen, |
| 323 | 3448 sug.su_badptr + stp->st_orglen, |
| 3449 sug.su_badlen - stp->st_orglen); | |
| 474 | 3450 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1); |
| 3451 #ifdef FEAT_RIGHTLEFT | |
| 3452 if (cmdmsg_rl) | |
| 3453 rl_mirror(IObuff); | |
| 3454 #endif | |
| 3455 msg_puts(IObuff); | |
| 3456 | |
| 3457 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy); | |
| 346 | 3458 msg_puts(IObuff); |
| 3459 | |
| 3460 /* The word may replace more than "su_badlen". */ | |
| 3461 if (sug.su_badlen < stp->st_orglen) | |
| 3462 { | |
| 3463 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""), | |
| 3464 stp->st_orglen, sug.su_badptr); | |
| 3465 msg_puts(IObuff); | |
| 3466 } | |
| 3467 | |
| 324 | 3468 if (p_verbose > 0) |
| 344 | 3469 { |
| 346 | 3470 /* Add the score. */ |
| 351 | 3471 if (sps_flags & (SPS_DOUBLE | SPS_BEST)) |
| 474 | 3472 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)", |
| 344 | 3473 stp->st_salscore ? "s " : "", |
| 3474 stp->st_score, stp->st_altscore); | |
| 3475 else | |
| 474 | 3476 vim_snprintf((char *)IObuff, IOSIZE, " (%d)", |
| 346 | 3477 stp->st_score); |
| 474 | 3478 #ifdef FEAT_RIGHTLEFT |
| 3479 if (cmdmsg_rl) | |
| 3480 /* Mirror the numbers, but keep the leading space. */ | |
| 3481 rl_mirror(IObuff + 1); | |
| 3482 #endif | |
| 346 | 3483 msg_advance(30); |
| 3484 msg_puts(IObuff); | |
| 344 | 3485 } |
| 323 | 3486 msg_putchar('\n'); |
| 3487 } | |
| 3488 | |
| 474 | 3489 #ifdef FEAT_RIGHTLEFT |
| 3490 cmdmsg_rl = FALSE; | |
| 3491 msg_col = 0; | |
| 3492 #endif | |
| 323 | 3493 /* Ask for choice. */ |
| 485 | 3494 selected = prompt_for_number(&mouse_used); |
| 372 | 3495 if (mouse_used) |
| 485 | 3496 selected -= lines_left; |
| 1938 | 3497 lines_left = Rows; /* avoid more prompt */ |
| 3498 /* don't delay for 'smd' in normal_cmd() */ | |
| 3499 msg_scroll = msg_scroll_save; | |
| 485 | 3500 } |
| 3501 | |
| 3502 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK) | |
| 3503 { | |
| 3504 /* Save the from and to text for :spellrepall. */ | |
| 3505 stp = &SUG(sug.su_ga, selected - 1); | |
| 548 | 3506 if (sug.su_badlen > stp->st_orglen) |
| 3507 { | |
| 3508 /* Replacing less than "su_badlen", append the remainder to | |
| 3509 * repl_to. */ | |
| 3510 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen); | |
| 3511 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word, | |
| 3512 sug.su_badlen - stp->st_orglen, | |
| 3513 sug.su_badptr + stp->st_orglen); | |
| 3514 repl_to = vim_strsave(IObuff); | |
| 3515 } | |
| 3516 else | |
| 3517 { | |
| 3518 /* Replacing su_badlen or more, use the whole word. */ | |
| 3519 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen); | |
| 3520 repl_to = vim_strsave(stp->st_word); | |
| 3521 } | |
| 485 | 3522 |
| 3523 /* Replace the word. */ | |
| 1938 | 3524 p = alloc((unsigned)STRLEN(line) - stp->st_orglen |
| 3525 + stp->st_wordlen + 1); | |
| 485 | 3526 if (p != NULL) |
| 3527 { | |
| 835 | 3528 c = (int)(sug.su_badptr - line); |
| 485 | 3529 mch_memmove(p, line, c); |
| 3530 STRCPY(p + c, stp->st_word); | |
| 3531 STRCAT(p, sug.su_badptr + stp->st_orglen); | |
| 3532 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 3533 curwin->w_cursor.col = c; | |
| 3534 | |
| 3535 /* For redo we use a change-word command. */ | |
| 3536 ResetRedobuff(); | |
| 3537 AppendToRedobuff((char_u *)"ciw"); | |
| 620 | 3538 AppendToRedobuffLit(p + c, |
| 625 | 3539 stp->st_wordlen + sug.su_badlen - stp->st_orglen); |
| 485 | 3540 AppendCharToRedobuff(ESC); |
| 819 | 3541 |
| 3542 /* After this "p" may be invalid. */ | |
| 3543 changed_bytes(curwin->w_cursor.lnum, c); | |
| 485 | 3544 } |
| 3545 } | |
| 3546 else | |
| 3547 curwin->w_cursor = prev_cursor; | |
| 323 | 3548 |
| 344 | 3549 spell_find_cleanup(&sug); |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3550 skip: |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3551 vim_free(line); |
| 344 | 3552 } |
| 3553 | |
| 3554 /* | |
| 475 | 3555 * Check if the word at line "lnum" column "col" is required to start with a |
| 3556 * capital. This uses 'spellcapcheck' of the current buffer. | |
| 3557 */ | |
| 3558 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3559 check_need_cap(linenr_T lnum, colnr_T col) |
| 475 | 3560 { |
| 3561 int need_cap = FALSE; | |
| 3562 char_u *line; | |
| 3563 char_u *line_copy = NULL; | |
| 3564 char_u *p; | |
| 3565 colnr_T endcol; | |
| 3566 regmatch_T regmatch; | |
| 3567 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3568 if (curwin->w_s->b_cap_prog == NULL) |
| 475 | 3569 return FALSE; |
| 3570 | |
| 3571 line = ml_get_curline(); | |
| 3572 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
|
3573 if (getwhitecols(line) >= (int)col) |
| 475 | 3574 { |
| 3575 /* At start of line, check if previous line is empty or sentence | |
| 3576 * ends there. */ | |
| 3577 if (lnum == 1) | |
| 3578 need_cap = TRUE; | |
| 3579 else | |
| 3580 { | |
| 3581 line = ml_get(lnum - 1); | |
| 3582 if (*skipwhite(line) == NUL) | |
| 3583 need_cap = TRUE; | |
| 3584 else | |
| 3585 { | |
| 3586 /* Append a space in place of the line break. */ | |
| 3587 line_copy = concat_str(line, (char_u *)" "); | |
| 3588 line = line_copy; | |
| 835 | 3589 endcol = (colnr_T)STRLEN(line); |
| 475 | 3590 } |
| 3591 } | |
| 3592 } | |
| 3593 else | |
| 3594 endcol = col; | |
| 3595 | |
| 3596 if (endcol > 0) | |
| 3597 { | |
| 3598 /* 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
|
3599 regmatch.regprog = curwin->w_s->b_cap_prog; |
| 475 | 3600 regmatch.rm_ic = FALSE; |
| 3601 p = line + endcol; | |
| 3602 for (;;) | |
| 3603 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
3604 MB_PTR_BACK(line, p); |
| 5477 | 3605 if (p == line || spell_iswordp_nmw(p, curwin)) |
| 475 | 3606 break; |
| 3607 if (vim_regexec(®match, p, 0) | |
| 3608 && regmatch.endp[0] == line + endcol) | |
| 3609 { | |
| 3610 need_cap = TRUE; | |
| 3611 break; | |
| 3612 } | |
| 3613 } | |
| 6375 | 3614 curwin->w_s->b_cap_prog = regmatch.regprog; |
| 475 | 3615 } |
| 3616 | |
| 3617 vim_free(line_copy); | |
| 3618 | |
| 3619 return need_cap; | |
| 3620 } | |
| 3621 | |
| 3622 | |
| 3623 /* | |
| 372 | 3624 * ":spellrepall" |
| 3625 */ | |
| 3626 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3627 ex_spellrepall(exarg_T *eap UNUSED) |
| 372 | 3628 { |
| 3629 pos_T pos = curwin->w_cursor; | |
| 3630 char_u *frompat; | |
| 3631 int addlen; | |
| 3632 char_u *line; | |
| 3633 char_u *p; | |
| 3634 int save_ws = p_ws; | |
| 483 | 3635 linenr_T prev_lnum = 0; |
| 372 | 3636 |
| 3637 if (repl_from == NULL || repl_to == NULL) | |
| 3638 { | |
| 3639 EMSG(_("E752: No previous spell replacement")); | |
| 3640 return; | |
| 3641 } | |
| 835 | 3642 addlen = (int)(STRLEN(repl_to) - STRLEN(repl_from)); |
| 3643 | |
| 3644 frompat = alloc((unsigned)STRLEN(repl_from) + 7); | |
| 372 | 3645 if (frompat == NULL) |
| 3646 return; | |
| 3647 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from); | |
| 3648 p_ws = FALSE; | |
| 3649 | |
| 483 | 3650 sub_nsubs = 0; |
| 3651 sub_nlines = 0; | |
| 372 | 3652 curwin->w_cursor.lnum = 0; |
| 3653 while (!got_int) | |
| 3654 { | |
|
11521
578df034735d
patch 8.0.0643: when a pattern search is slow Vim becomes unusable
Christian Brabandt <cb@256bit.org>
parents:
11394
diff
changeset
|
3655 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP, NULL, NULL) == 0 |
| 372 | 3656 || u_save_cursor() == FAIL) |
| 3657 break; | |
| 3658 | |
| 3659 /* Only replace when the right word isn't there yet. This happens | |
| 3660 * when changing "etc" to "etc.". */ | |
| 3661 line = ml_get_curline(); | |
| 3662 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col, | |
| 3663 repl_to, STRLEN(repl_to)) != 0) | |
| 3664 { | |
| 835 | 3665 p = alloc((unsigned)STRLEN(line) + addlen + 1); |
| 372 | 3666 if (p == NULL) |
| 3667 break; | |
| 3668 mch_memmove(p, line, curwin->w_cursor.col); | |
| 3669 STRCPY(p + curwin->w_cursor.col, repl_to); | |
| 3670 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from)); | |
| 3671 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 3672 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col); | |
| 483 | 3673 |
| 3674 if (curwin->w_cursor.lnum != prev_lnum) | |
| 3675 { | |
| 3676 ++sub_nlines; | |
| 3677 prev_lnum = curwin->w_cursor.lnum; | |
| 3678 } | |
| 3679 ++sub_nsubs; | |
| 372 | 3680 } |
| 835 | 3681 curwin->w_cursor.col += (colnr_T)STRLEN(repl_to); |
| 372 | 3682 } |
| 3683 | |
| 3684 p_ws = save_ws; | |
| 3685 curwin->w_cursor = pos; | |
| 3686 vim_free(frompat); | |
| 3687 | |
| 483 | 3688 if (sub_nsubs == 0) |
| 372 | 3689 EMSG2(_("E753: Not found: %s"), repl_from); |
| 483 | 3690 else |
| 3691 do_sub_msg(FALSE); | |
| 372 | 3692 } |
| 3693 | |
| 3694 /* | |
| 344 | 3695 * Find spell suggestions for "word". Return them in the growarray "*gap" as |
| 3696 * a list of allocated strings. | |
| 3697 */ | |
| 3698 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3699 spell_suggest_list( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3700 garray_T *gap, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3701 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3702 int maxcount, /* maximum nr of suggestions */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3703 int need_cap, /* 'spellcapcheck' matched */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3704 int interactive) |
| 344 | 3705 { |
| 3706 suginfo_T sug; | |
| 3707 int i; | |
| 3708 suggest_T *stp; | |
| 3709 char_u *wcopy; | |
| 3710 | |
| 638 | 3711 spell_find_suggest(word, 0, &sug, maxcount, FALSE, need_cap, interactive); |
| 344 | 3712 |
| 3713 /* Make room in "gap". */ | |
| 3714 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1); | |
| 626 | 3715 if (ga_grow(gap, sug.su_ga.ga_len) == OK) |
| 3716 { | |
| 3717 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 3718 { | |
| 3719 stp = &SUG(sug.su_ga, i); | |
| 3720 | |
| 3721 /* The suggested word may replace only part of "word", add the not | |
| 3722 * replaced part. */ | |
| 3723 wcopy = alloc(stp->st_wordlen | |
| 835 | 3724 + (unsigned)STRLEN(sug.su_badptr + stp->st_orglen) + 1); |
| 626 | 3725 if (wcopy == NULL) |
| 3726 break; | |
| 3727 STRCPY(wcopy, stp->st_word); | |
| 3728 STRCPY(wcopy + stp->st_wordlen, sug.su_badptr + stp->st_orglen); | |
| 3729 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy; | |
| 3730 } | |
| 344 | 3731 } |
| 3732 | |
| 3733 spell_find_cleanup(&sug); | |
| 3734 } | |
| 3735 | |
| 3736 /* | |
| 3737 * Find spell suggestions for the word at the start of "badptr". | |
| 3738 * Return the suggestions in "su->su_ga". | |
| 3739 * The maximum number of suggestions is "maxcount". | |
| 3740 * Note: does use info for the current window. | |
| 3741 * This is based on the mechanisms of Aspell, but completely reimplemented. | |
| 3742 */ | |
| 3743 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3744 spell_find_suggest( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3745 char_u *badptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3746 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
|
3747 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3748 int maxcount, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3749 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
|
3750 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
|
3751 int interactive) |
| 344 | 3752 { |
| 534 | 3753 hlf_T attr = HLF_COUNT; |
| 372 | 3754 char_u buf[MAXPATHL]; |
| 3755 char_u *p; | |
| 3756 int do_combine = FALSE; | |
| 3757 char_u *sps_copy; | |
| 3758 #ifdef FEAT_EVAL | |
| 3759 static int expr_busy = FALSE; | |
| 3760 #endif | |
| 385 | 3761 int c; |
| 503 | 3762 int i; |
| 3763 langp_T *lp; | |
| 344 | 3764 |
| 3765 /* | |
| 3766 * Set the info in "*su". | |
| 3767 */ | |
| 3768 vim_memset(su, 0, sizeof(suginfo_T)); | |
| 3769 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10); | |
| 3770 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10); | |
| 355 | 3771 if (*badptr == NUL) |
| 3772 return; | |
| 344 | 3773 hash_init(&su->su_banned); |
| 3774 | |
| 3775 su->su_badptr = badptr; | |
| 638 | 3776 if (badlen != 0) |
| 3777 su->su_badlen = badlen; | |
| 3778 else | |
| 3779 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL, FALSE); | |
| 344 | 3780 su->su_maxcount = maxcount; |
| 372 | 3781 su->su_maxscore = SCORE_MAXINIT; |
| 344 | 3782 |
| 3783 if (su->su_badlen >= MAXWLEN) | |
| 3784 su->su_badlen = MAXWLEN - 1; /* just in case */ | |
| 3785 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen); | |
| 3786 (void)spell_casefold(su->su_badptr, su->su_badlen, | |
| 3787 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
|
3788 /* 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
|
3789 * 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
|
3790 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
|
3791 |
| 346 | 3792 /* get caps flags for bad word */ |
| 474 | 3793 su->su_badflags = badword_captype(su->su_badptr, |
| 3794 su->su_badptr + su->su_badlen); | |
| 388 | 3795 if (need_cap) |
| 3796 su->su_badflags |= WF_ONECAP; | |
| 344 | 3797 |
| 503 | 3798 /* Find the default language for sound folding. We simply use the first |
| 3799 * one in 'spelllang' that supports sound folding. That's good for when | |
| 3800 * using multiple files for one language, it's not that bad when mixing | |
| 3801 * languages (e.g., "pl,en"). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3802 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
|
3803 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3804 lp = LANGP_ENTRY(curbuf->b_s.b_langp, i); |
| 503 | 3805 if (lp->lp_sallang != NULL) |
| 3806 { | |
| 3807 su->su_sallang = lp->lp_sallang; | |
| 3808 break; | |
| 3809 } | |
| 3810 } | |
| 3811 | |
| 534 | 3812 /* Soundfold the bad word with the default sound folding, so that we don't |
| 3813 * have to do this many times. */ | |
| 3814 if (su->su_sallang != NULL) | |
| 3815 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE, | |
| 3816 su->su_sal_badword); | |
| 3817 | |
| 385 | 3818 /* If the word is not capitalised and spell_check() doesn't consider the |
| 3819 * word to be bad then it might need to be capitalised. Add a suggestion | |
| 3820 * for that. */ | |
| 455 | 3821 c = PTR2CHAR(su->su_badptr); |
| 534 | 3822 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT) |
| 385 | 3823 { |
| 3824 make_case_word(su->su_badword, buf, WF_ONECAP); | |
| 3825 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE, | |
| 625 | 3826 0, TRUE, su->su_sallang, FALSE); |
| 385 | 3827 } |
| 3828 | |
| 344 | 3829 /* Ban the bad word itself. It may appear in another region. */ |
| 358 | 3830 if (banbadword) |
| 3831 add_banned(su, su->su_badword); | |
| 344 | 3832 |
| 372 | 3833 /* Make a copy of 'spellsuggest', because the expression may change it. */ |
| 3834 sps_copy = vim_strsave(p_sps); | |
| 3835 if (sps_copy == NULL) | |
| 3836 return; | |
| 3837 | |
| 3838 /* Loop over the items in 'spellsuggest'. */ | |
| 3839 for (p = sps_copy; *p != NUL; ) | |
| 3840 { | |
| 3841 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 3842 | |
| 3843 if (STRNCMP(buf, "expr:", 5) == 0) | |
| 3844 { | |
| 3845 #ifdef FEAT_EVAL | |
| 375 | 3846 /* Evaluate an expression. Skip this when called recursively, |
| 3847 * when using spellsuggest() in the expression. */ | |
| 372 | 3848 if (!expr_busy) |
| 3849 { | |
| 3850 expr_busy = TRUE; | |
| 3851 spell_suggest_expr(su, buf + 5); | |
| 3852 expr_busy = FALSE; | |
| 3853 } | |
| 3854 #endif | |
| 3855 } | |
| 3856 else if (STRNCMP(buf, "file:", 5) == 0) | |
| 3857 /* Use list of suggestions in a file. */ | |
| 3858 spell_suggest_file(su, buf + 5); | |
| 3859 else | |
| 3860 { | |
| 3861 /* Use internal method. */ | |
| 625 | 3862 spell_suggest_intern(su, interactive); |
| 372 | 3863 if (sps_flags & SPS_DOUBLE) |
| 3864 do_combine = TRUE; | |
| 3865 } | |
| 3866 } | |
| 3867 | |
| 3868 vim_free(sps_copy); | |
| 3869 | |
| 3870 if (do_combine) | |
| 3871 /* Combine the two list of suggestions. This must be done last, | |
| 3872 * because sorting changes the order again. */ | |
| 3873 score_combine(su); | |
| 3874 } | |
| 3875 | |
| 3876 #ifdef FEAT_EVAL | |
| 3877 /* | |
| 3878 * Find suggestions by evaluating expression "expr". | |
| 3879 */ | |
| 3880 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3881 spell_suggest_expr(suginfo_T *su, char_u *expr) |
| 372 | 3882 { |
| 3883 list_T *list; | |
| 3884 listitem_T *li; | |
| 3885 int score; | |
| 3886 char_u *p; | |
| 3887 | |
| 3888 /* The work is split up in a few parts to avoid having to export | |
| 3889 * suginfo_T. | |
| 3890 * First evaluate the expression and get the resulting list. */ | |
| 3891 list = eval_spell_expr(su->su_badword, expr); | |
| 3892 if (list != NULL) | |
| 3893 { | |
| 3894 /* Loop over the items in the list. */ | |
| 3895 for (li = list->lv_first; li != NULL; li = li->li_next) | |
| 3896 if (li->li_tv.v_type == VAR_LIST) | |
| 3897 { | |
| 3898 /* Get the word and the score from the items. */ | |
| 3899 score = get_spellword(li->li_tv.vval.v_list, &p); | |
| 625 | 3900 if (score >= 0 && score <= su->su_maxscore) |
| 3901 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 3902 score, 0, TRUE, su->su_sallang, FALSE); | |
| 372 | 3903 } |
| 3904 list_unref(list); | |
| 3905 } | |
| 3906 | |
| 625 | 3907 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3908 check_suggestions(su, &su->su_ga); | |
| 372 | 3909 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 3910 } | |
| 3911 #endif | |
| 3912 | |
| 3913 /* | |
| 474 | 3914 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'. |
| 372 | 3915 */ |
| 3916 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3917 spell_suggest_file(suginfo_T *su, char_u *fname) |
| 372 | 3918 { |
| 3919 FILE *fd; | |
| 3920 char_u line[MAXWLEN * 2]; | |
| 3921 char_u *p; | |
| 3922 int len; | |
| 3923 char_u cword[MAXWLEN]; | |
| 3924 | |
| 3925 /* Open the file. */ | |
| 3926 fd = mch_fopen((char *)fname, "r"); | |
| 3927 if (fd == NULL) | |
| 3928 { | |
| 3929 EMSG2(_(e_notopen), fname); | |
| 3930 return; | |
| 3931 } | |
| 3932 | |
| 3933 /* Read it line by line. */ | |
| 3934 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int) | |
| 3935 { | |
| 3936 line_breakcheck(); | |
| 3937 | |
| 3938 p = vim_strchr(line, '/'); | |
| 3939 if (p == NULL) | |
| 3940 continue; /* No Tab found, just skip the line. */ | |
| 3941 *p++ = NUL; | |
| 3942 if (STRICMP(su->su_badword, line) == 0) | |
| 3943 { | |
| 3944 /* Match! Isolate the good word, until CR or NL. */ | |
| 3945 for (len = 0; p[len] >= ' '; ++len) | |
| 3946 ; | |
| 3947 p[len] = NUL; | |
| 3948 | |
| 3949 /* If the suggestion doesn't have specific case duplicate the case | |
| 3950 * of the bad word. */ | |
| 3951 if (captype(p, NULL) == 0) | |
| 3952 { | |
| 3953 make_case_word(p, cword, su->su_badflags); | |
| 3954 p = cword; | |
| 3955 } | |
| 3956 | |
| 3957 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 625 | 3958 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE); |
| 372 | 3959 } |
| 3960 } | |
| 3961 | |
| 3962 fclose(fd); | |
| 3963 | |
| 625 | 3964 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3965 check_suggestions(su, &su->su_ga); | |
| 372 | 3966 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 3967 } | |
| 3968 | |
| 3969 /* | |
| 3970 * Find suggestions for the internal method indicated by "sps_flags". | |
| 3971 */ | |
| 3972 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3973 spell_suggest_intern(suginfo_T *su, int interactive) |
| 625 | 3974 { |
| 3975 /* | |
| 3976 * Load the .sug file(s) that are available and not done yet. | |
| 3977 */ | |
| 3978 suggest_load_files(); | |
| 3979 | |
| 344 | 3980 /* |
| 346 | 3981 * 1. Try special cases, such as repeating a word: "the the" -> "the". |
| 344 | 3982 * |
| 3983 * Set a maximum score to limit the combination of operations that is | |
| 3984 * tried. | |
| 3985 */ | |
| 346 | 3986 suggest_try_special(su); |
| 3987 | |
| 3988 /* | |
| 3989 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries | |
| 3990 * from the .aff file and inserting a space (split the word). | |
| 3991 */ | |
| 3992 suggest_try_change(su); | |
| 344 | 3993 |
| 3994 /* For the resulting top-scorers compute the sound-a-like score. */ | |
| 3995 if (sps_flags & SPS_DOUBLE) | |
| 3996 score_comp_sal(su); | |
| 3997 | |
| 3998 /* | |
| 346 | 3999 * 3. Try finding sound-a-like words. |
| 344 | 4000 */ |
| 625 | 4001 if ((sps_flags & SPS_FAST) == 0) |
| 4002 { | |
| 4003 if (sps_flags & SPS_BEST) | |
| 4004 /* Adjust the word score for the suggestions found so far for how | |
| 4005 * they sounds like. */ | |
| 4006 rescore_suggestions(su); | |
| 4007 | |
| 4008 /* | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
4009 * While going through the soundfold tree "su_maxscore" is the score |
| 625 | 4010 * for the soundfold word, limits the changes that are being tried, |
| 4011 * and "su_sfmaxscore" the rescored score, which is set by | |
| 4012 * cleanup_suggestions(). | |
| 4013 * First find words with a small edit distance, because this is much | |
| 4014 * faster and often already finds the top-N suggestions. If we didn't | |
| 4015 * find many suggestions try again with a higher edit distance. | |
| 4016 * "sl_sounddone" is used to avoid doing the same word twice. | |
| 4017 */ | |
| 4018 suggest_try_soundalike_prep(); | |
| 4019 su->su_maxscore = SCORE_SFMAX1; | |
| 4020 su->su_sfmaxscore = SCORE_MAXINIT * 3; | |
| 346 | 4021 suggest_try_soundalike(su); |
| 625 | 4022 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) |
| 4023 { | |
| 4024 /* We didn't find enough matches, try again, allowing more | |
| 4025 * changes to the soundfold word. */ | |
| 4026 su->su_maxscore = SCORE_SFMAX2; | |
| 4027 suggest_try_soundalike(su); | |
| 4028 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) | |
| 4029 { | |
| 4030 /* Still didn't find enough matches, try again, allowing even | |
| 4031 * more changes to the soundfold word. */ | |
| 4032 su->su_maxscore = SCORE_SFMAX3; | |
| 4033 suggest_try_soundalike(su); | |
| 4034 } | |
| 4035 } | |
| 4036 su->su_maxscore = su->su_sfmaxscore; | |
| 4037 suggest_try_soundalike_finish(); | |
| 4038 } | |
| 4039 | |
| 4040 /* When CTRL-C was hit while searching do show the results. Only clear | |
| 4041 * got_int when using a command, not for spellsuggest(). */ | |
| 344 | 4042 ui_breakcheck(); |
| 625 | 4043 if (interactive && got_int) |
| 344 | 4044 { |
| 4045 (void)vgetc(); | |
| 4046 got_int = FALSE; | |
| 4047 } | |
| 4048 | |
| 372 | 4049 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0) |
| 344 | 4050 { |
| 4051 if (sps_flags & SPS_BEST) | |
| 4052 /* Adjust the word score for how it sounds like. */ | |
| 4053 rescore_suggestions(su); | |
| 4054 | |
| 625 | 4055 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 4056 check_suggestions(su, &su->su_ga); | |
| 372 | 4057 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 344 | 4058 } |
| 4059 } | |
| 4060 | |
| 4061 /* | |
| 4062 * Free the info put in "*su" by spell_find_suggest(). | |
| 4063 */ | |
| 4064 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4065 spell_find_cleanup(suginfo_T *su) |
| 344 | 4066 { |
| 4067 int i; | |
| 4068 | |
| 4069 /* Free the suggestions. */ | |
| 4070 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 4071 vim_free(SUG(su->su_ga, i).st_word); | |
| 4072 ga_clear(&su->su_ga); | |
| 4073 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 4074 vim_free(SUG(su->su_sga, i).st_word); | |
| 4075 ga_clear(&su->su_sga); | |
| 323 | 4076 |
| 4077 /* Free the banned words. */ | |
| 625 | 4078 hash_clear_all(&su->su_banned, 0); |
| 323 | 4079 } |
| 4080 | |
| 4081 /* | |
| 324 | 4082 * Make a copy of "word", with the first letter upper or lower cased, to |
| 4083 * "wcopy[MAXWLEN]". "word" must not be empty. | |
| 4084 * The result is NUL terminated. | |
| 323 | 4085 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
4086 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4087 onecap_copy( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4088 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4089 char_u *wcopy, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4090 int upper) /* TRUE: first letter made upper case */ |
| 323 | 4091 { |
| 4092 char_u *p; | |
| 4093 int c; | |
| 4094 int l; | |
| 4095 | |
| 4096 p = word; | |
| 4097 #ifdef FEAT_MBYTE | |
| 4098 if (has_mbyte) | |
| 474 | 4099 c = mb_cptr2char_adv(&p); |
| 323 | 4100 else |
| 4101 #endif | |
| 4102 c = *p++; | |
| 4103 if (upper) | |
| 324 | 4104 c = SPELL_TOUPPER(c); |
| 323 | 4105 else |
| 324 | 4106 c = SPELL_TOFOLD(c); |
| 323 | 4107 #ifdef FEAT_MBYTE |
| 4108 if (has_mbyte) | |
| 4109 l = mb_char2bytes(c, wcopy); | |
| 4110 else | |
| 4111 #endif | |
| 4112 { | |
| 4113 l = 1; | |
| 4114 wcopy[0] = c; | |
| 4115 } | |
| 376 | 4116 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1); |
| 323 | 4117 } |
| 4118 | |
| 4119 /* | |
| 324 | 4120 * Make a copy of "word" with all the letters upper cased into |
| 4121 * "wcopy[MAXWLEN]". The result is NUL terminated. | |
| 323 | 4122 */ |
| 4123 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4124 allcap_copy(char_u *word, char_u *wcopy) |
| 323 | 4125 { |
| 4126 char_u *s; | |
| 4127 char_u *d; | |
| 4128 int c; | |
| 4129 | |
| 4130 d = wcopy; | |
| 4131 for (s = word; *s != NUL; ) | |
| 4132 { | |
| 4133 #ifdef FEAT_MBYTE | |
| 4134 if (has_mbyte) | |
| 474 | 4135 c = mb_cptr2char_adv(&s); |
| 323 | 4136 else |
| 4137 #endif | |
| 4138 c = *s++; | |
| 492 | 4139 |
| 4140 #ifdef FEAT_MBYTE | |
| 3042 | 4141 /* We only change 0xdf to SS when we are certain latin1 is used. It |
| 492 | 4142 * would cause weird errors in other 8-bit encodings. */ |
| 4143 if (enc_latin1like && c == 0xdf) | |
| 4144 { | |
| 4145 c = 'S'; | |
| 4146 if (d - wcopy >= MAXWLEN - 1) | |
| 4147 break; | |
| 4148 *d++ = c; | |
| 4149 } | |
| 4150 else | |
| 4151 #endif | |
| 4152 c = SPELL_TOUPPER(c); | |
| 323 | 4153 |
| 4154 #ifdef FEAT_MBYTE | |
| 4155 if (has_mbyte) | |
| 4156 { | |
| 4157 if (d - wcopy >= MAXWLEN - MB_MAXBYTES) | |
| 4158 break; | |
| 4159 d += mb_char2bytes(c, d); | |
| 4160 } | |
| 4161 else | |
| 4162 #endif | |
| 4163 { | |
| 4164 if (d - wcopy >= MAXWLEN - 1) | |
| 4165 break; | |
| 4166 *d++ = c; | |
| 4167 } | |
| 4168 } | |
| 4169 *d = NUL; | |
| 4170 } | |
| 4171 | |
| 4172 /* | |
| 346 | 4173 * Try finding suggestions by recognizing specific situations. |
| 4174 */ | |
| 4175 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4176 suggest_try_special(suginfo_T *su) |
| 346 | 4177 { |
| 4178 char_u *p; | |
| 376 | 4179 size_t len; |
| 346 | 4180 int c; |
| 4181 char_u word[MAXWLEN]; | |
| 4182 | |
| 4183 /* | |
| 4184 * Recognize a word that is repeated: "the the". | |
| 4185 */ | |
| 4186 p = skiptowhite(su->su_fbadword); | |
| 4187 len = p - su->su_fbadword; | |
| 4188 p = skipwhite(p); | |
| 4189 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0) | |
| 4190 { | |
| 4191 /* Include badflags: if the badword is onecap or allcap | |
| 4192 * use that for the goodword too: "The the" -> "The". */ | |
| 4193 c = su->su_fbadword[len]; | |
| 4194 su->su_fbadword[len] = NUL; | |
| 4195 make_case_word(su->su_fbadword, word, su->su_badflags); | |
| 4196 su->su_fbadword[len] = c; | |
| 534 | 4197 |
| 4198 /* Give a soundalike score of 0, compute the score as if deleting one | |
| 4199 * character. */ | |
| 4200 add_suggestion(su, &su->su_ga, word, su->su_badlen, | |
| 625 | 4201 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE); |
| 346 | 4202 } |
| 4203 } | |
| 4204 | |
| 4205 /* | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4206 * 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
|
4207 * Output is dumped in "suggestprof". |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4208 */ |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4209 #if 0 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4210 # define SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4211 proftime_T current; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4212 proftime_T total; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4213 proftime_T times[STATE_FINAL + 1]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4214 long counts[STATE_FINAL + 1]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4215 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4216 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4217 prof_init(void) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4218 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4219 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
|
4220 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4221 profile_zero(×[i]); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4222 counts[i] = 0; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4223 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4224 profile_start(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4225 profile_start(&total); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4226 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4227 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4228 /* call before changing state */ |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4229 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4230 prof_store(state_T state) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4231 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4232 profile_end(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4233 profile_add(×[state], ¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4234 ++counts[state]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4235 profile_start(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4236 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4237 # 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
|
4238 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4239 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4240 prof_report(char *name) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4241 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4242 FILE *fd = fopen("suggestprof", "a"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4243 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4244 profile_end(&total); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4245 fprintf(fd, "-----------------------\n"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4246 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
|
4247 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
|
4248 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
|
4249 fclose(fd); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4250 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4251 #else |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4252 # define PROF_STORE(state) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4253 #endif |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4254 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4255 /* |
| 323 | 4256 * Try finding suggestions by adding/removing/swapping letters. |
| 4257 */ | |
| 4258 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4259 suggest_try_change(suginfo_T *su) |
| 323 | 4260 { |
| 4261 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */ | |
| 625 | 4262 int n; |
| 4263 char_u *p; | |
| 4264 int lpi; | |
| 323 | 4265 langp_T *lp; |
| 4266 | |
| 4267 /* We make a copy of the case-folded bad word, so that we can modify it | |
| 346 | 4268 * to find matches (esp. REP items). Append some more text, changing |
| 4269 * chars after the bad word may help. */ | |
| 323 | 4270 STRCPY(fword, su->su_fbadword); |
| 835 | 4271 n = (int)STRLEN(fword); |
| 346 | 4272 p = su->su_badptr + su->su_badlen; |
| 835 | 4273 (void)spell_casefold(p, (int)STRLEN(p), fword + n, MAXWLEN - n); |
| 323 | 4274 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4275 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
|
4276 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4277 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 482 | 4278 |
| 500 | 4279 /* If reloading a spell file fails it's still in the list but |
| 4280 * everything has been cleared. */ | |
| 625 | 4281 if (lp->lp_slang->sl_fbyts == NULL) |
| 500 | 4282 continue; |
| 4283 | |
| 625 | 4284 /* 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
|
4285 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4286 prof_init(); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4287 #endif |
| 625 | 4288 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
|
4289 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4290 prof_report("try_change"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4291 #endif |
| 625 | 4292 } |
| 4293 } | |
| 4294 | |
| 4295 /* Check the maximum score, if we go over it we won't try this change. */ | |
| 4296 #define TRY_DEEPER(su, stack, depth, add) \ | |
| 4297 (stack[depth].ts_score + (add) < su->su_maxscore) | |
| 4298 | |
| 4299 /* | |
| 4300 * Try finding suggestions by adding/removing/swapping letters. | |
| 4301 * | |
| 4302 * This uses a state machine. At each node in the tree we try various | |
| 4303 * operations. When trying if an operation works "depth" is increased and the | |
| 4304 * stack[] is used to store info. This allows combinations, thus insert one | |
| 4305 * character, replace one and delete another. The number of changes is | |
| 4306 * limited by su->su_maxscore. | |
| 4307 * | |
| 4308 * After implementing this I noticed an article by Kemal Oflazer that | |
| 4309 * describes something similar: "Error-tolerant Finite State Recognition with | |
| 4310 * Applications to Morphological Analysis and Spelling Correction" (1996). | |
| 4311 * The implementation in the article is simplified and requires a stack of | |
| 4312 * unknown depth. The implementation here only needs a stack depth equal to | |
| 4313 * the length of the word. | |
| 4314 * | |
| 4315 * This is also used for the sound-folded word, "soundfold" is TRUE then. | |
| 4316 * The mechanism is the same, but we find a match with a sound-folded word | |
| 4317 * that comes from one or more original words. Each of these words may be | |
| 4318 * added, this is done by add_sound_suggest(). | |
| 4319 * Don't use: | |
| 4320 * the prefix tree or the keep-case tree | |
| 4321 * "su->su_badlen" | |
| 4322 * anything to do with upper and lower case | |
| 4323 * anything to do with word or non-word characters ("spell_iswordp()") | |
| 4324 * banned words | |
| 4325 * word flags (rare, region, compounding) | |
| 4326 * word splitting for now | |
| 4327 * "similar_chars()" | |
| 4328 * use "slang->sl_repsal" instead of "lp->lp_replang->sl_rep" | |
| 4329 */ | |
| 4330 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4331 suggest_trie_walk( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4332 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4333 langp_T *lp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4334 char_u *fword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4335 int soundfold) |
| 625 | 4336 { |
| 4337 char_u tword[MAXWLEN]; /* good word collected so far */ | |
| 4338 trystate_T stack[MAXWLEN]; | |
| 4339 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
|
4340 * concatenation of prefix compound |
| 625 | 4341 * words and split word. NUL terminated |
| 4342 * when going deeper but not when coming | |
| 4343 * back. */ | |
| 4344 char_u compflags[MAXWLEN]; /* compound flags, one for each word */ | |
| 4345 trystate_T *sp; | |
| 4346 int newscore; | |
| 4347 int score; | |
| 4348 char_u *byts, *fbyts, *pbyts; | |
| 4349 idx_T *idxs, *fidxs, *pidxs; | |
| 4350 int depth; | |
| 4351 int c, c2, c3; | |
| 4352 int n = 0; | |
| 4353 int flags; | |
| 4354 garray_T *gap; | |
| 4355 idx_T arridx; | |
| 4356 int len; | |
| 4357 char_u *p; | |
| 4358 fromto_T *ftp; | |
| 4359 int fl = 0, tl; | |
| 4360 int repextra = 0; /* extra bytes in fword[] from REP item */ | |
| 4361 slang_T *slang = lp->lp_slang; | |
| 4362 int fword_ends; | |
| 4363 int goodword_ends; | |
| 4364 #ifdef DEBUG_TRIEWALK | |
| 4365 /* Stores the name of the change made at each level. */ | |
| 4366 char_u changename[MAXWLEN][80]; | |
| 4367 #endif | |
| 4368 int breakcheckcount = 1000; | |
| 4369 int compound_ok; | |
| 4370 | |
| 4371 /* | |
| 4372 * Go through the whole case-fold tree, try changes at each node. | |
| 4373 * "tword[]" contains the word collected from nodes in the tree. | |
| 4374 * "fword[]" the word we are trying to match with (initially the bad | |
| 4375 * word). | |
| 4376 */ | |
| 4377 depth = 0; | |
| 4378 sp = &stack[0]; | |
| 4379 vim_memset(sp, 0, sizeof(trystate_T)); | |
| 4380 sp->ts_curi = 1; | |
| 4381 | |
| 4382 if (soundfold) | |
| 4383 { | |
| 4384 /* Going through the soundfold tree. */ | |
| 4385 byts = fbyts = slang->sl_sbyts; | |
| 4386 idxs = fidxs = slang->sl_sidxs; | |
| 4387 pbyts = NULL; | |
| 4388 pidxs = NULL; | |
| 4389 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 4390 sp->ts_state = STATE_START; | |
| 4391 } | |
| 4392 else | |
| 4393 { | |
| 330 | 4394 /* |
| 375 | 4395 * When there are postponed prefixes we need to use these first. At |
| 4396 * the end of the prefix we continue in the case-fold tree. | |
| 4397 */ | |
| 482 | 4398 fbyts = slang->sl_fbyts; |
| 4399 fidxs = slang->sl_fidxs; | |
| 4400 pbyts = slang->sl_pbyts; | |
| 4401 pidxs = slang->sl_pidxs; | |
| 375 | 4402 if (pbyts != NULL) |
| 4403 { | |
| 4404 byts = pbyts; | |
| 4405 idxs = pidxs; | |
| 482 | 4406 sp->ts_prefixdepth = PFD_PREFIXTREE; |
| 375 | 4407 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */ |
| 4408 } | |
| 4409 else | |
| 4410 { | |
| 4411 byts = fbyts; | |
| 4412 idxs = fidxs; | |
| 482 | 4413 sp->ts_prefixdepth = PFD_NOPREFIX; |
| 485 | 4414 sp->ts_state = STATE_START; |
| 375 | 4415 } |
| 625 | 4416 } |
| 4417 | |
| 4418 /* | |
| 4419 * Loop to find all suggestions. At each round we either: | |
| 4420 * - For the current state try one operation, advance "ts_curi", | |
| 4421 * increase "depth". | |
| 4422 * - When a state is done go to the next, set "ts_state". | |
| 4423 * - When all states are tried decrease "depth". | |
| 4424 */ | |
| 4425 while (depth >= 0 && !got_int) | |
| 4426 { | |
| 4427 sp = &stack[depth]; | |
| 4428 switch (sp->ts_state) | |
| 4429 { | |
| 4430 case STATE_START: | |
| 4431 case STATE_NOPREFIX: | |
| 4432 /* | |
| 4433 * Start of node: Deal with NUL bytes, which means | |
| 4434 * tword[] may end here. | |
| 4435 */ | |
| 4436 arridx = sp->ts_arridx; /* current node in the tree */ | |
| 4437 len = byts[arridx]; /* bytes in this node */ | |
| 4438 arridx += sp->ts_curi; /* index of current byte */ | |
| 4439 | |
| 4440 if (sp->ts_prefixdepth == PFD_PREFIXTREE) | |
| 4441 { | |
| 4442 /* Skip over the NUL bytes, we use them later. */ | |
| 4443 for (n = 0; n < len && byts[arridx + n] == 0; ++n) | |
| 4444 ; | |
| 4445 sp->ts_curi += n; | |
| 4446 | |
| 4447 /* Always past NUL bytes now. */ | |
| 4448 n = (int)sp->ts_state; | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4449 PROF_STORE(sp->ts_state) |
| 625 | 4450 sp->ts_state = STATE_ENDNUL; |
| 4451 sp->ts_save_badflags = su->su_badflags; | |
| 4452 | |
| 4453 /* At end of a prefix or at start of prefixtree: check for | |
| 4454 * following word. */ | |
| 4455 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX) | |
| 4456 { | |
| 4457 /* Set su->su_badflags to the caps type at this position. | |
| 4458 * Use the caps type until here for the prefix itself. */ | |
| 4459 #ifdef FEAT_MBYTE | |
| 4460 if (has_mbyte) | |
| 4461 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); | |
| 4462 else | |
| 4463 #endif | |
| 4464 n = sp->ts_fidx; | |
| 4465 flags = badword_captype(su->su_badptr, su->su_badptr + n); | |
| 4466 su->su_badflags = badword_captype(su->su_badptr + n, | |
| 455 | 4467 su->su_badptr + su->su_badlen); |
| 625 | 4468 #ifdef DEBUG_TRIEWALK |
| 4469 sprintf(changename[depth], "prefix"); | |
| 4470 #endif | |
| 4471 go_deeper(stack, depth, 0); | |
| 4472 ++depth; | |
| 4473 sp = &stack[depth]; | |
| 4474 sp->ts_prefixdepth = depth - 1; | |
| 4475 byts = fbyts; | |
| 4476 idxs = fidxs; | |
| 4477 sp->ts_arridx = 0; | |
| 4478 | |
| 4479 /* Move the prefix to preword[] with the right case | |
| 4480 * and make find_keepcap_word() works. */ | |
| 4481 tword[sp->ts_twordlen] = NUL; | |
| 4482 make_case_word(tword + sp->ts_splitoff, | |
| 4483 preword + sp->ts_prewordlen, flags); | |
| 835 | 4484 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 625 | 4485 sp->ts_splitoff = sp->ts_twordlen; |
| 4486 } | |
| 4487 break; | |
| 4488 } | |
| 4489 | |
| 4490 if (sp->ts_curi > len || byts[arridx] != 0) | |
| 4491 { | |
| 4492 /* 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
|
4493 PROF_STORE(sp->ts_state) |
| 625 | 4494 sp->ts_state = STATE_ENDNUL; |
| 4495 sp->ts_save_badflags = su->su_badflags; | |
| 4496 break; | |
| 4497 } | |
| 4498 | |
| 4499 /* | |
| 4500 * End of word in tree. | |
| 4501 */ | |
| 4502 ++sp->ts_curi; /* eat one NUL byte */ | |
| 4503 | |
| 4504 flags = (int)idxs[arridx]; | |
| 698 | 4505 |
| 4506 /* Skip words with the NOSUGGEST flag. */ | |
| 4507 if (flags & WF_NOSUGGEST) | |
| 4508 break; | |
| 4509 | |
| 625 | 4510 fword_ends = (fword[sp->ts_fidx] == NUL |
| 4511 || (soundfold | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
4512 ? VIM_ISWHITE(fword[sp->ts_fidx]) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4513 : !spell_iswordp(fword + sp->ts_fidx, curwin))); |
| 625 | 4514 tword[sp->ts_twordlen] = NUL; |
| 4515 | |
| 4516 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL | |
| 485 | 4517 && (sp->ts_flags & TSF_PREFIXOK) == 0) |
| 625 | 4518 { |
| 4519 /* There was a prefix before the word. Check that the prefix | |
| 4520 * can be used with this word. */ | |
| 4521 /* Count the length of the NULs in the prefix. If there are | |
| 4522 * none this must be the first try without a prefix. */ | |
| 4523 n = stack[sp->ts_prefixdepth].ts_arridx; | |
| 4524 len = pbyts[n++]; | |
| 4525 for (c = 0; c < len && pbyts[n + c] == 0; ++c) | |
| 4526 ; | |
| 4527 if (c > 0) | |
| 4528 { | |
| 4529 c = valid_word_prefix(c, n, flags, | |
| 483 | 4530 tword + sp->ts_splitoff, slang, FALSE); |
| 625 | 4531 if (c == 0) |
| 4532 break; | |
| 4533 | |
| 4534 /* Use the WF_RARE flag for a rare prefix. */ | |
| 4535 if (c & WF_RAREPFX) | |
| 4536 flags |= WF_RARE; | |
| 4537 | |
| 4538 /* Tricky: when checking for both prefix and compounding | |
| 4539 * we run into the prefix flag first. | |
| 4540 * Remember that it's OK, so that we accept the prefix | |
| 4541 * when arriving at a compound flag. */ | |
| 4542 sp->ts_flags |= TSF_PREFIXOK; | |
| 4543 } | |
| 4544 } | |
| 4545 | |
| 4546 /* Check NEEDCOMPOUND: can't use word without compounding. Do try | |
| 4547 * appending another compound word below. */ | |
| 4548 if (sp->ts_complen == sp->ts_compsplit && fword_ends | |
| 500 | 4549 && (flags & WF_NEEDCOMP)) |
| 625 | 4550 goodword_ends = FALSE; |
| 4551 else | |
| 4552 goodword_ends = TRUE; | |
| 4553 | |
| 4554 p = NULL; | |
| 4555 compound_ok = TRUE; | |
| 4556 if (sp->ts_complen > sp->ts_compsplit) | |
| 4557 { | |
| 4558 if (slang->sl_nobreak) | |
| 4559 { | |
| 4560 /* There was a word before this word. When there was no | |
| 4561 * change in this word (it was correct) add the first word | |
| 4562 * as a suggestion. If this word was corrected too, we | |
| 4563 * need to check if a correct word follows. */ | |
| 4564 if (sp->ts_fidx - sp->ts_splitfidx | |
| 4565 == sp->ts_twordlen - sp->ts_splitoff | |
| 4566 && STRNCMP(fword + sp->ts_splitfidx, | |
| 4567 tword + sp->ts_splitoff, | |
| 4568 sp->ts_fidx - sp->ts_splitfidx) == 0) | |
| 492 | 4569 { |
| 625 | 4570 preword[sp->ts_prewordlen] = NUL; |
| 4571 newscore = score_wordcount_adj(slang, sp->ts_score, | |
| 4572 preword + sp->ts_prewordlen, | |
| 4573 sp->ts_prewordlen > 0); | |
| 4574 /* Add the suggestion if the score isn't too bad. */ | |
| 4575 if (newscore <= su->su_maxscore) | |
| 492 | 4576 add_suggestion(su, &su->su_ga, preword, |
| 503 | 4577 sp->ts_splitfidx - repextra, |
| 625 | 4578 newscore, 0, FALSE, |
| 4579 lp->lp_sallang, FALSE); | |
| 4580 break; | |
| 492 | 4581 } |
| 490 | 4582 } |
| 4583 else | |
| 625 | 4584 { |
| 4585 /* There was a compound word before this word. If this | |
| 4586 * word does not support compounding then give up | |
| 4587 * (splitting is tried for the word without compound | |
| 4588 * flag). */ | |
| 4589 if (((unsigned)flags >> 24) == 0 | |
| 4590 || sp->ts_twordlen - sp->ts_splitoff | |
| 4591 < slang->sl_compminlen) | |
| 4592 break; | |
| 4593 #ifdef FEAT_MBYTE | |
| 4594 /* For multi-byte chars check character length against | |
| 4595 * COMPOUNDMIN. */ | |
| 4596 if (has_mbyte | |
| 4597 && slang->sl_compminlen > 0 | |
| 4598 && mb_charlen(tword + sp->ts_splitoff) | |
| 4599 < slang->sl_compminlen) | |
| 4600 break; | |
| 4601 #endif | |
| 4602 | |
| 4603 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 4604 compflags[sp->ts_complen + 1] = NUL; | |
| 4605 vim_strncpy(preword + sp->ts_prewordlen, | |
| 4606 tword + sp->ts_splitoff, | |
| 4607 sp->ts_twordlen - sp->ts_splitoff); | |
| 1762 | 4608 |
| 4609 /* Verify CHECKCOMPOUNDPATTERN rules. */ | |
| 4610 if (match_checkcompoundpattern(preword, sp->ts_prewordlen, | |
| 4611 &slang->sl_comppat)) | |
| 4612 compound_ok = FALSE; | |
| 4613 | |
| 4614 if (compound_ok) | |
| 4615 { | |
| 4616 p = preword; | |
| 4617 while (*skiptowhite(p) != NUL) | |
| 4618 p = skipwhite(skiptowhite(p)); | |
| 4619 if (fword_ends && !can_compound(slang, p, | |
| 625 | 4620 compflags + sp->ts_compsplit)) |
| 1762 | 4621 /* Compound is not allowed. But it may still be |
| 4622 * possible if we add another (short) word. */ | |
| 4623 compound_ok = FALSE; | |
| 4624 } | |
| 625 | 4625 |
| 4626 /* Get pointer to last char of previous word. */ | |
| 4627 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
|
4628 MB_PTR_BACK(preword, p); |
| 625 | 4629 } |
| 4630 } | |
| 4631 | |
| 4632 /* | |
| 4633 * Form the word with proper case in preword. | |
| 4634 * If there is a word from a previous split, append. | |
| 4635 * For the soundfold tree don't change the case, simply append. | |
| 4636 */ | |
| 4637 if (soundfold) | |
| 4638 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff); | |
| 4639 else if (flags & WF_KEEPCAP) | |
| 4640 /* Must find the word in the keep-case tree. */ | |
| 4641 find_keepcap_word(slang, tword + sp->ts_splitoff, | |
| 483 | 4642 preword + sp->ts_prewordlen); |
| 625 | 4643 else |
| 4644 { | |
| 4645 /* Include badflags: If the badword is onecap or allcap | |
| 4646 * use that for the goodword too. But if the badword is | |
| 4647 * allcap and it's only one char long use onecap. */ | |
| 4648 c = su->su_badflags; | |
| 4649 if ((c & WF_ALLCAP) | |
| 4650 #ifdef FEAT_MBYTE | |
| 4651 && su->su_badlen == (*mb_ptr2len)(su->su_badptr) | |
| 346 | 4652 #else |
| 625 | 4653 && su->su_badlen == 1 |
| 4654 #endif | |
| 4655 ) | |
| 4656 c = WF_ONECAP; | |
| 4657 c |= flags; | |
| 4658 | |
| 4659 /* When appending a compound word after a word character don't | |
| 4660 * use Onecap. */ | |
| 5477 | 4661 if (p != NULL && spell_iswordp_nmw(p, curwin)) |
| 625 | 4662 c &= ~WF_ONECAP; |
| 4663 make_case_word(tword + sp->ts_splitoff, | |
| 490 | 4664 preword + sp->ts_prewordlen, c); |
| 625 | 4665 } |
| 4666 | |
| 4667 if (!soundfold) | |
| 4668 { | |
| 323 | 4669 /* Don't use a banned word. It may appear again as a good |
| 4670 * word, thus remember it. */ | |
| 4671 if (flags & WF_BANNED) | |
| 4672 { | |
| 483 | 4673 add_banned(su, preword + sp->ts_prewordlen); |
| 323 | 4674 break; |
| 4675 } | |
| 534 | 4676 if ((sp->ts_complen == sp->ts_compsplit |
| 625 | 4677 && WAS_BANNED(su, preword + sp->ts_prewordlen)) |
| 4678 || WAS_BANNED(su, preword)) | |
| 534 | 4679 { |
| 4680 if (slang->sl_compprog == NULL) | |
| 4681 break; | |
| 4682 /* the word so far was banned but we may try compounding */ | |
| 4683 goodword_ends = FALSE; | |
| 4684 } | |
| 625 | 4685 } |
| 4686 | |
| 4687 newscore = 0; | |
| 4688 if (!soundfold) /* soundfold words don't have flags */ | |
| 4689 { | |
| 323 | 4690 if ((flags & WF_REGION) |
| 390 | 4691 && (((unsigned)flags >> 16) & lp->lp_region) == 0) |
| 323 | 4692 newscore += SCORE_REGION; |
| 4693 if (flags & WF_RARE) | |
| 4694 newscore += SCORE_RARE; | |
| 4695 | |
| 346 | 4696 if (!spell_valid_case(su->su_badflags, |
| 483 | 4697 captype(preword + sp->ts_prewordlen, NULL))) |
| 323 | 4698 newscore += SCORE_ICASE; |
| 625 | 4699 } |
| 4700 | |
| 4701 /* TODO: how about splitting in the soundfold tree? */ | |
| 4702 if (fword_ends | |
| 4703 && goodword_ends | |
| 4704 && sp->ts_fidx >= sp->ts_fidxtry | |
| 4705 && compound_ok) | |
| 4706 { | |
| 4707 /* The badword also ends: add suggestions. */ | |
| 4708 #ifdef DEBUG_TRIEWALK | |
| 4709 if (soundfold && STRCMP(preword, "smwrd") == 0) | |
| 4710 { | |
| 4711 int j; | |
| 4712 | |
| 4713 /* print the stack of changes that brought us here */ | |
| 4714 smsg("------ %s -------", fword); | |
| 4715 for (j = 0; j < depth; ++j) | |
| 4716 smsg("%s", changename[j]); | |
| 4717 } | |
| 4718 #endif | |
| 4719 if (soundfold) | |
| 4720 { | |
| 4721 /* For soundfolded words we need to find the original | |
| 1188 | 4722 * words, the edit distance and then add them. */ |
| 625 | 4723 add_sound_suggest(su, preword, sp->ts_score, lp); |
| 4724 } | |
|
2457
6c1d10a85398
Avoid illegal memory access in spell suggestion. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2455
diff
changeset
|
4725 else if (sp->ts_fidx > 0) |
| 625 | 4726 { |
| 4727 /* Give a penalty when changing non-word char to word | |
| 4728 * char, e.g., "thes," -> "these". */ | |
| 366 | 4729 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
|
4730 MB_PTR_BACK(fword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4731 if (!spell_iswordp(p, curwin)) |
| 366 | 4732 { |
| 4733 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
|
4734 MB_PTR_BACK(preword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4735 if (spell_iswordp(p, curwin)) |
| 366 | 4736 newscore += SCORE_NONWORD; |
| 4737 } | |
| 4738 | |
| 625 | 4739 /* Give a bonus to words seen before. */ |
| 4740 score = score_wordcount_adj(slang, | |
| 4741 sp->ts_score + newscore, | |
| 4742 preword + sp->ts_prewordlen, | |
| 4743 sp->ts_prewordlen > 0); | |
| 4744 | |
| 4745 /* Add the suggestion if the score isn't too bad. */ | |
| 4746 if (score <= su->su_maxscore) | |
| 633 | 4747 { |
| 625 | 4748 add_suggestion(su, &su->su_ga, preword, |
| 4749 sp->ts_fidx - repextra, | |
| 4750 score, 0, FALSE, lp->lp_sallang, FALSE); | |
| 633 | 4751 |
| 4752 if (su->su_badflags & WF_MIXCAP) | |
| 4753 { | |
| 4754 /* We really don't know if the word should be | |
| 4755 * upper or lower case, add both. */ | |
| 4756 c = captype(preword, NULL); | |
| 4757 if (c == 0 || c == WF_ALLCAP) | |
| 4758 { | |
| 4759 make_case_word(tword + sp->ts_splitoff, | |
| 4760 preword + sp->ts_prewordlen, | |
| 4761 c == 0 ? WF_ALLCAP : 0); | |
| 4762 | |
| 4763 add_suggestion(su, &su->su_ga, preword, | |
| 4764 sp->ts_fidx - repextra, | |
| 4765 score + SCORE_ICASE, 0, FALSE, | |
| 4766 lp->lp_sallang, FALSE); | |
| 4767 } | |
| 4768 } | |
| 4769 } | |
| 625 | 4770 } |
| 4771 } | |
| 4772 | |
| 4773 /* | |
| 4774 * Try word split and/or compounding. | |
| 4775 */ | |
| 4776 if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends) | |
| 4777 #ifdef FEAT_MBYTE | |
| 4778 /* Don't split halfway a character. */ | |
| 4779 && (!has_mbyte || sp->ts_tcharlen == 0) | |
| 4780 #endif | |
| 4781 ) | |
| 4782 { | |
| 4783 int try_compound; | |
| 4784 int try_split; | |
| 4785 | |
| 4786 /* If past the end of the bad word don't try a split. | |
| 4787 * Otherwise try changing the next word. E.g., find | |
| 4788 * suggestions for "the the" where the second "the" is | |
| 4789 * different. It's done like a split. | |
| 4790 * TODO: word split for soundfold words */ | |
| 4791 try_split = (sp->ts_fidx - repextra < su->su_badlen) | |
| 4792 && !soundfold; | |
| 4793 | |
| 4794 /* Get here in several situations: | |
| 4795 * 1. The word in the tree ends: | |
| 4796 * If the word allows compounding try that. Otherwise try | |
| 4797 * a split by inserting a space. For both check that a | |
| 4798 * valid words starts at fword[sp->ts_fidx]. | |
| 4799 * For NOBREAK do like compounding to be able to check if | |
| 4800 * the next word is valid. | |
| 4801 * 2. The badword does end, but it was due to a change (e.g., | |
| 4802 * a swap). No need to split, but do check that the | |
| 4803 * following word is valid. | |
| 4804 * 3. The badword and the word in the tree end. It may still | |
| 4805 * be possible to compound another (short) word. | |
| 4806 */ | |
| 4807 try_compound = FALSE; | |
| 4808 if (!soundfold | |
|
7526
08c1f73efcde
commit https://github.com/vim/vim/commit/7b877b360532713dc21a0ff3d55a76ac02eaf573
Christian Brabandt <cb@256bit.org>
parents:
7509
diff
changeset
|
4809 && !slang->sl_nocompoundsugs |
| 625 | 4810 && slang->sl_compprog != NULL |
| 4811 && ((unsigned)flags >> 24) != 0 | |
| 4812 && sp->ts_twordlen - sp->ts_splitoff | |
| 4813 >= slang->sl_compminlen | |
| 4814 #ifdef FEAT_MBYTE | |
| 4815 && (!has_mbyte | |
| 4816 || slang->sl_compminlen == 0 | |
| 4817 || mb_charlen(tword + sp->ts_splitoff) | |
| 4818 >= slang->sl_compminlen) | |
| 4819 #endif | |
| 4820 && (slang->sl_compsylmax < MAXWLEN | |
| 4821 || sp->ts_complen + 1 - sp->ts_compsplit | |
| 4822 < slang->sl_compmax) | |
| 1762 | 4823 && (can_be_compound(sp, slang, |
| 4824 compflags, ((unsigned)flags >> 24)))) | |
| 4825 | |
| 625 | 4826 { |
| 4827 try_compound = TRUE; | |
| 4828 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 4829 compflags[sp->ts_complen + 1] = NUL; | |
| 4830 } | |
| 4831 | |
| 4832 /* For NOBREAK we never try splitting, it won't make any word | |
| 4833 * valid. */ | |
|
7526
08c1f73efcde
commit https://github.com/vim/vim/commit/7b877b360532713dc21a0ff3d55a76ac02eaf573
Christian Brabandt <cb@256bit.org>
parents:
7509
diff
changeset
|
4834 if (slang->sl_nobreak && !slang->sl_nocompoundsugs) |
| 625 | 4835 try_compound = TRUE; |
| 4836 | |
| 4837 /* If we could add a compound word, and it's also possible to | |
| 4838 * split at this point, do the split first and set | |
| 4839 * TSF_DIDSPLIT to avoid doing it again. */ | |
| 4840 else if (!fword_ends | |
| 4841 && try_compound | |
| 4842 && (sp->ts_flags & TSF_DIDSPLIT) == 0) | |
| 4843 { | |
| 485 | 4844 try_compound = FALSE; |
| 625 | 4845 sp->ts_flags |= TSF_DIDSPLIT; |
| 4846 --sp->ts_curi; /* do the same NUL again */ | |
| 4847 compflags[sp->ts_complen] = NUL; | |
| 4848 } | |
| 4849 else | |
| 4850 sp->ts_flags &= ~TSF_DIDSPLIT; | |
| 4851 | |
| 4852 if (try_split || try_compound) | |
| 4853 { | |
| 534 | 4854 if (!try_compound && (!fword_ends || !goodword_ends)) |
| 485 | 4855 { |
| 4856 /* If we're going to split need to check that the | |
| 501 | 4857 * words so far are valid for compounding. If there |
| 4858 * is only one word it must not have the NEEDCOMPOUND | |
| 4859 * flag. */ | |
| 4860 if (sp->ts_complen == sp->ts_compsplit | |
| 4861 && (flags & WF_NEEDCOMP)) | |
| 4862 break; | |
| 490 | 4863 p = preword; |
| 4864 while (*skiptowhite(p) != NUL) | |
| 4865 p = skipwhite(skiptowhite(p)); | |
| 485 | 4866 if (sp->ts_complen > sp->ts_compsplit |
| 490 | 4867 && !can_compound(slang, p, |
| 485 | 4868 compflags + sp->ts_compsplit)) |
| 4869 break; | |
| 698 | 4870 |
| 4871 if (slang->sl_nosplitsugs) | |
| 4872 newscore += SCORE_SPLIT_NO; | |
| 4873 else | |
| 4874 newscore += SCORE_SPLIT; | |
| 625 | 4875 |
| 4876 /* Give a bonus to words seen before. */ | |
| 4877 newscore = score_wordcount_adj(slang, newscore, | |
| 4878 preword + sp->ts_prewordlen, TRUE); | |
| 482 | 4879 } |
| 4880 | |
| 625 | 4881 if (TRY_DEEPER(su, stack, depth, newscore)) |
| 323 | 4882 { |
| 625 | 4883 go_deeper(stack, depth, newscore); |
| 4884 #ifdef DEBUG_TRIEWALK | |
| 4885 if (!try_compound && !fword_ends) | |
| 4886 sprintf(changename[depth], "%.*s-%s: split", | |
| 4887 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 4888 else | |
| 4889 sprintf(changename[depth], "%.*s-%s: compound", | |
| 4890 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 4891 #endif | |
| 323 | 4892 /* Save things to be restored at STATE_SPLITUNDO. */ |
| 346 | 4893 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
|
4894 PROF_STORE(sp->ts_state) |
| 376 | 4895 sp->ts_state = STATE_SPLITUNDO; |
| 4896 | |
| 4897 ++depth; | |
| 4898 sp = &stack[depth]; | |
| 323 | 4899 |
| 482 | 4900 /* Append a space to preword when splitting. */ |
| 4901 if (!try_compound && !fword_ends) | |
| 4902 STRCAT(preword, " "); | |
| 835 | 4903 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 483 | 4904 sp->ts_splitoff = sp->ts_twordlen; |
| 492 | 4905 sp->ts_splitfidx = sp->ts_fidx; |
| 376 | 4906 |
| 4907 /* If the badword has a non-word character at this | |
| 4908 * position skip it. That means replacing the | |
| 482 | 4909 * non-word character with a space. Always skip a |
| 534 | 4910 * character when the word ends. But only when the |
| 4911 * good word can end. */ | |
| 625 | 4912 if (((!try_compound && !spell_iswordp_nmw(fword |
| 5477 | 4913 + sp->ts_fidx, |
| 4914 curwin)) | |
| 625 | 4915 || fword_ends) |
| 4916 && fword[sp->ts_fidx] != NUL | |
| 4917 && goodword_ends) | |
| 376 | 4918 { |
| 482 | 4919 int l; |
| 4920 | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
4921 l = MB_PTR2LEN(fword + sp->ts_fidx); |
| 482 | 4922 if (fword_ends) |
| 4923 { | |
| 4924 /* Copy the skipped character to preword. */ | |
| 483 | 4925 mch_memmove(preword + sp->ts_prewordlen, |
| 482 | 4926 fword + sp->ts_fidx, l); |
| 483 | 4927 sp->ts_prewordlen += l; |
| 4928 preword[sp->ts_prewordlen] = NUL; | |
| 482 | 4929 } |
| 4930 else | |
| 4931 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST; | |
| 4932 sp->ts_fidx += l; | |
| 376 | 4933 } |
| 455 | 4934 |
| 485 | 4935 /* When compounding include compound flag in |
| 4936 * compflags[] (already set above). When splitting we | |
| 4937 * may start compounding over again. */ | |
| 482 | 4938 if (try_compound) |
| 483 | 4939 ++sp->ts_complen; |
| 482 | 4940 else |
| 485 | 4941 sp->ts_compsplit = sp->ts_complen; |
| 4942 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 482 | 4943 |
| 455 | 4944 /* set su->su_badflags to the caps type at this |
| 4945 * position */ | |
| 324 | 4946 #ifdef FEAT_MBYTE |
| 4947 if (has_mbyte) | |
| 455 | 4948 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); |
| 324 | 4949 else |
| 4950 #endif | |
| 455 | 4951 n = sp->ts_fidx; |
| 474 | 4952 su->su_badflags = badword_captype(su->su_badptr + n, |
| 455 | 4953 su->su_badptr + su->su_badlen); |
| 323 | 4954 |
| 4955 /* Restart at top of the tree. */ | |
| 376 | 4956 sp->ts_arridx = 0; |
| 485 | 4957 |
| 4958 /* If there are postponed prefixes, try these too. */ | |
| 4959 if (pbyts != NULL) | |
| 4960 { | |
| 4961 byts = pbyts; | |
| 4962 idxs = pidxs; | |
| 4963 sp->ts_prefixdepth = PFD_PREFIXTREE; | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4964 PROF_STORE(sp->ts_state) |
| 485 | 4965 sp->ts_state = STATE_NOPREFIX; |
| 4966 } | |
| 323 | 4967 } |
| 4968 } | |
| 625 | 4969 } |
| 4970 break; | |
| 4971 | |
| 4972 case STATE_SPLITUNDO: | |
| 4973 /* Undo the changes done for word split or compound word. */ | |
| 4974 su->su_badflags = sp->ts_save_badflags; | |
| 4975 | |
| 4976 /* Continue looking for NUL bytes. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4977 PROF_STORE(sp->ts_state) |
| 625 | 4978 sp->ts_state = STATE_START; |
| 4979 | |
| 4980 /* In case we went into the prefix tree. */ | |
| 4981 byts = fbyts; | |
| 4982 idxs = fidxs; | |
| 4983 break; | |
| 4984 | |
| 4985 case STATE_ENDNUL: | |
| 4986 /* Past the NUL bytes in the node. */ | |
| 4987 su->su_badflags = sp->ts_save_badflags; | |
| 4988 if (fword[sp->ts_fidx] == NUL | |
| 4989 #ifdef FEAT_MBYTE | |
| 4990 && sp->ts_tcharlen == 0 | |
| 4991 #endif | |
| 4992 ) | |
| 4993 { | |
| 4994 /* 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
|
4995 PROF_STORE(sp->ts_state) |
| 625 | 4996 sp->ts_state = STATE_DEL; |
| 323 | 4997 break; |
| 625 | 4998 } |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4999 PROF_STORE(sp->ts_state) |
| 625 | 5000 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
|
5001 /* FALLTHROUGH */ |
| 625 | 5002 |
| 5003 case STATE_PLAIN: | |
| 5004 /* | |
| 5005 * Go over all possible bytes at this node, add each to tword[] | |
| 5006 * and use child node. "ts_curi" is the index. | |
| 5007 */ | |
| 5008 arridx = sp->ts_arridx; | |
| 5009 if (sp->ts_curi > byts[arridx]) | |
| 5010 { | |
| 5011 /* Done all bytes at this node, do next state. When still at | |
| 5012 * 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
|
5013 PROF_STORE(sp->ts_state) |
| 625 | 5014 if (sp->ts_fidx >= sp->ts_fidxtry) |
| 323 | 5015 sp->ts_state = STATE_DEL; |
| 5016 else | |
| 625 | 5017 sp->ts_state = STATE_FINAL; |
| 5018 } | |
| 5019 else | |
| 5020 { | |
| 5021 arridx += sp->ts_curi++; | |
| 5022 c = byts[arridx]; | |
| 5023 | |
| 5024 /* Normal byte, go one level deeper. If it's not equal to the | |
| 5025 * byte in the bad word adjust the score. But don't even try | |
| 5026 * when the byte was already changed. And don't try when we | |
| 5695 | 5027 * just deleted this byte, accepting it is always cheaper than |
| 625 | 5028 * delete + substitute. */ |
| 5029 if (c == fword[sp->ts_fidx] | |
| 5030 #ifdef FEAT_MBYTE | |
| 5031 || (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE) | |
| 5032 #endif | |
| 5033 ) | |
| 5034 newscore = 0; | |
| 5035 else | |
| 5036 newscore = SCORE_SUBST; | |
| 5037 if ((newscore == 0 | |
| 5038 || (sp->ts_fidx >= sp->ts_fidxtry | |
| 5039 && ((sp->ts_flags & TSF_DIDDEL) == 0 | |
| 5040 || c != fword[sp->ts_delidx]))) | |
| 5041 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5042 { | |
| 5043 go_deeper(stack, depth, newscore); | |
| 5044 #ifdef DEBUG_TRIEWALK | |
| 5045 if (newscore > 0) | |
| 5046 sprintf(changename[depth], "%.*s-%s: subst %c to %c", | |
| 5047 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5048 fword[sp->ts_fidx], c); | |
| 323 | 5049 else |
| 625 | 5050 sprintf(changename[depth], "%.*s-%s: accept %c", |
| 5051 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5052 fword[sp->ts_fidx]); | |
| 5053 #endif | |
| 323 | 5054 ++depth; |
| 625 | 5055 sp = &stack[depth]; |
| 5056 ++sp->ts_fidx; | |
| 5057 tword[sp->ts_twordlen++] = c; | |
| 5058 sp->ts_arridx = idxs[arridx]; | |
| 5059 #ifdef FEAT_MBYTE | |
| 5060 if (newscore == SCORE_SUBST) | |
| 5061 sp->ts_isdiff = DIFF_YES; | |
| 330 | 5062 if (has_mbyte) |
| 358 | 5063 { |
| 625 | 5064 /* Multi-byte characters are a bit complicated to |
| 5065 * handle: They differ when any of the bytes differ | |
| 5066 * and then their length may also differ. */ | |
| 5067 if (sp->ts_tcharlen == 0) | |
| 5068 { | |
| 5069 /* First byte. */ | |
| 5070 sp->ts_tcharidx = 0; | |
| 5071 sp->ts_tcharlen = MB_BYTE2LEN(c); | |
| 5072 sp->ts_fcharstart = sp->ts_fidx - 1; | |
| 5073 sp->ts_isdiff = (newscore != 0) | |
| 5074 ? DIFF_YES : DIFF_NONE; | |
| 5075 } | |
| 5076 else if (sp->ts_isdiff == DIFF_INSERT) | |
| 5077 /* When inserting trail bytes don't advance in the | |
| 5078 * bad word. */ | |
| 5079 --sp->ts_fidx; | |
| 5080 if (++sp->ts_tcharidx == sp->ts_tcharlen) | |
| 5081 { | |
| 5082 /* Last byte of character. */ | |
| 5083 if (sp->ts_isdiff == DIFF_YES) | |
| 5084 { | |
| 5085 /* Correct ts_fidx for the byte length of the | |
| 5086 * character (we didn't check that before). */ | |
| 5087 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
|
5088 + MB_PTR2LEN( |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5089 fword + sp->ts_fcharstart); |
| 625 | 5090 /* For changing a composing character adjust |
| 5091 * the score from SCORE_SUBST to | |
| 5092 * SCORE_SUBCOMP. */ | |
| 5093 if (enc_utf8 | |
| 5094 && 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
|
5095 utf_ptr2char(tword |
| 625 | 5096 + sp->ts_twordlen |
| 5097 - sp->ts_tcharlen)) | |
| 5098 && 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
|
5099 utf_ptr2char(fword |
| 625 | 5100 + sp->ts_fcharstart))) |
| 5101 sp->ts_score -= | |
| 5102 SCORE_SUBST - SCORE_SUBCOMP; | |
| 5103 | |
| 5104 /* For a similar character adjust score from | |
| 5105 * SCORE_SUBST to SCORE_SIMILAR. */ | |
| 5106 else if (!soundfold | |
| 5107 && slang->sl_has_map | |
| 5108 && similar_chars(slang, | |
| 5109 mb_ptr2char(tword | |
| 5110 + sp->ts_twordlen | |
| 5111 - sp->ts_tcharlen), | |
| 5112 mb_ptr2char(fword | |
| 5113 + sp->ts_fcharstart))) | |
| 5114 sp->ts_score -= | |
| 5115 SCORE_SUBST - SCORE_SIMILAR; | |
| 5116 } | |
| 5117 else if (sp->ts_isdiff == DIFF_INSERT | |
| 5118 && sp->ts_twordlen > sp->ts_tcharlen) | |
| 5119 { | |
| 5120 p = tword + sp->ts_twordlen - sp->ts_tcharlen; | |
| 5121 c = mb_ptr2char(p); | |
| 5122 if (enc_utf8 && utf_iscomposing(c)) | |
| 5123 { | |
| 5124 /* Inserting a composing char doesn't | |
| 5125 * count that much. */ | |
| 5126 sp->ts_score -= SCORE_INS - SCORE_INSCOMP; | |
| 5127 } | |
| 5128 else | |
| 5129 { | |
| 5130 /* If the previous character was the same, | |
| 5131 * thus doubling a character, give a bonus | |
| 5132 * to the score. Also for the soundfold | |
| 5133 * tree (might seem illogical but does | |
| 5134 * give better scores). */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5135 MB_PTR_BACK(tword, p); |
| 625 | 5136 if (c == mb_ptr2char(p)) |
| 5137 sp->ts_score -= SCORE_INS | |
| 5138 - SCORE_INSDUP; | |
| 5139 } | |
| 5140 } | |
| 5141 | |
| 5142 /* Starting a new char, reset the length. */ | |
| 5143 sp->ts_tcharlen = 0; | |
| 5144 } | |
| 358 | 5145 } |
| 330 | 5146 else |
| 5147 #endif | |
| 358 | 5148 { |
| 625 | 5149 /* If we found a similar char adjust the score. |
| 5150 * We do this after calling go_deeper() because | |
| 5151 * it's slow. */ | |
| 5152 if (newscore != 0 | |
| 5153 && !soundfold | |
| 5154 && slang->sl_has_map | |
| 5155 && similar_chars(slang, | |
| 5156 c, fword[sp->ts_fidx - 1])) | |
| 5157 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR; | |
| 358 | 5158 } |
| 625 | 5159 } |
| 5160 } | |
| 5161 break; | |
| 5162 | |
| 5163 case STATE_DEL: | |
| 5164 #ifdef FEAT_MBYTE | |
| 5165 /* When past the first byte of a multi-byte char don't try | |
| 5166 * delete/insert/swap a character. */ | |
| 5167 if (has_mbyte && sp->ts_tcharlen > 0) | |
| 5168 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5169 PROF_STORE(sp->ts_state) |
| 625 | 5170 sp->ts_state = STATE_FINAL; |
| 5171 break; | |
| 5172 } | |
| 5173 #endif | |
| 5174 /* | |
| 5175 * Try skipping one character in the bad word (delete it). | |
| 5176 */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5177 PROF_STORE(sp->ts_state) |
| 625 | 5178 sp->ts_state = STATE_INS_PREP; |
| 5179 sp->ts_curi = 1; | |
| 5180 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*') | |
| 5181 /* Deleting a vowel at the start of a word counts less, see | |
| 5182 * soundalike_score(). */ | |
| 5183 newscore = 2 * SCORE_DEL / 3; | |
| 5184 else | |
| 5185 newscore = SCORE_DEL; | |
| 5186 if (fword[sp->ts_fidx] != NUL | |
| 5187 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5188 { | |
| 5189 go_deeper(stack, depth, newscore); | |
| 5190 #ifdef DEBUG_TRIEWALK | |
| 5191 sprintf(changename[depth], "%.*s-%s: delete %c", | |
| 5192 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5193 fword[sp->ts_fidx]); | |
| 5194 #endif | |
| 5195 ++depth; | |
| 5196 | |
| 5197 /* Remember what character we deleted, so that we can avoid | |
| 5198 * inserting it again. */ | |
| 5199 stack[depth].ts_flags |= TSF_DIDDEL; | |
| 5200 stack[depth].ts_delidx = sp->ts_fidx; | |
| 5201 | |
| 5202 /* Advance over the character in fword[]. Give a bonus to the | |
| 5203 * score if the same character is following "nn" -> "n". It's | |
| 5204 * a bit illogical for soundfold tree but it does give better | |
| 5205 * results. */ | |
| 5206 #ifdef FEAT_MBYTE | |
| 5207 if (has_mbyte) | |
| 5208 { | |
| 5209 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
|
5210 stack[depth].ts_fidx += MB_PTR2LEN(fword + sp->ts_fidx); |
| 625 | 5211 if (enc_utf8 && utf_iscomposing(c)) |
| 5212 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP; | |
| 5213 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx)) | |
| 5214 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 5215 } |
| 5216 else | |
| 625 | 5217 #endif |
| 5218 { | |
| 5219 ++stack[depth].ts_fidx; | |
| 5220 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1]) | |
| 5221 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 5222 } |
| 5223 break; | |
| 625 | 5224 } |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5225 /* FALLTHROUGH */ |
| 625 | 5226 |
| 5227 case STATE_INS_PREP: | |
| 5228 if (sp->ts_flags & TSF_DIDDEL) | |
| 5229 { | |
| 5230 /* If we just deleted a byte then inserting won't make sense, | |
| 5231 * a substitute is always cheaper. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5232 PROF_STORE(sp->ts_state) |
| 625 | 5233 sp->ts_state = STATE_SWAP; |
| 5234 break; | |
| 5235 } | |
| 5236 | |
| 5237 /* skip over NUL bytes */ | |
| 5238 n = sp->ts_arridx; | |
| 5239 for (;;) | |
| 5240 { | |
| 5241 if (sp->ts_curi > byts[n]) | |
| 5242 { | |
| 5243 /* 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
|
5244 PROF_STORE(sp->ts_state) |
| 625 | 5245 sp->ts_state = STATE_SWAP; |
| 5246 break; | |
| 5247 } | |
| 5248 if (byts[n + sp->ts_curi] != NUL) | |
| 5249 { | |
| 5250 /* Found a byte to insert. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5251 PROF_STORE(sp->ts_state) |
| 625 | 5252 sp->ts_state = STATE_INS; |
| 330 | 5253 break; |
| 5254 } | |
| 625 | 5255 ++sp->ts_curi; |
| 5256 } | |
| 5257 break; | |
| 5258 | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5259 /* FALLTHROUGH */ |
| 625 | 5260 |
| 5261 case STATE_INS: | |
| 5262 /* Insert one byte. Repeat this for each possible byte at this | |
| 5263 * node. */ | |
| 5264 n = sp->ts_arridx; | |
| 5265 if (sp->ts_curi > byts[n]) | |
| 5266 { | |
| 5267 /* 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
|
5268 PROF_STORE(sp->ts_state) |
| 625 | 5269 sp->ts_state = STATE_SWAP; |
| 5270 break; | |
| 5271 } | |
| 5272 | |
| 5273 /* Do one more byte at this node, but: | |
| 5274 * - Skip NUL bytes. | |
| 5275 * - Skip the byte if it's equal to the byte in the word, | |
| 5276 * accepting that byte is always better. | |
| 5277 */ | |
| 5278 n += sp->ts_curi++; | |
| 5279 c = byts[n]; | |
| 5280 if (soundfold && sp->ts_twordlen == 0 && c == '*') | |
| 5281 /* Inserting a vowel at the start of a word counts less, | |
| 5282 * see soundalike_score(). */ | |
| 5283 newscore = 2 * SCORE_INS / 3; | |
| 5284 else | |
| 5285 newscore = SCORE_INS; | |
| 5286 if (c != fword[sp->ts_fidx] | |
| 5287 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5288 { | |
| 5289 go_deeper(stack, depth, newscore); | |
| 5290 #ifdef DEBUG_TRIEWALK | |
| 5291 sprintf(changename[depth], "%.*s-%s: insert %c", | |
| 5292 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5293 c); | |
| 5294 #endif | |
| 5295 ++depth; | |
| 5296 sp = &stack[depth]; | |
| 5297 tword[sp->ts_twordlen++] = c; | |
| 5298 sp->ts_arridx = idxs[n]; | |
| 330 | 5299 #ifdef FEAT_MBYTE |
| 5300 if (has_mbyte) | |
| 5301 { | |
| 625 | 5302 fl = MB_BYTE2LEN(c); |
| 5303 if (fl > 1) | |
| 330 | 5304 { |
| 625 | 5305 /* There are following bytes for the same character. |
| 5306 * We must find all bytes before trying | |
| 5307 * delete/insert/swap/etc. */ | |
| 5308 sp->ts_tcharlen = fl; | |
| 5309 sp->ts_tcharidx = 1; | |
| 5310 sp->ts_isdiff = DIFF_INSERT; | |
| 330 | 5311 } |
| 323 | 5312 } |
| 5313 else | |
| 625 | 5314 fl = 1; |
| 5315 if (fl == 1) | |
| 5316 #endif | |
| 5317 { | |
| 5318 /* If the previous character was the same, thus doubling a | |
| 5319 * character, give a bonus to the score. Also for | |
| 5320 * soundfold words (illogical but does give a better | |
| 5321 * score). */ | |
| 5322 if (sp->ts_twordlen >= 2 | |
| 5323 && tword[sp->ts_twordlen - 2] == c) | |
| 5324 sp->ts_score -= SCORE_INS - SCORE_INSDUP; | |
| 5325 } | |
| 5326 } | |
| 5327 break; | |
| 5328 | |
| 5329 case STATE_SWAP: | |
| 5330 /* | |
| 5331 * Swap two bytes in the bad word: "12" -> "21". | |
| 5332 * We change "fword" here, it's changed back afterwards at | |
| 5333 * STATE_UNSWAP. | |
| 5334 */ | |
| 5335 p = fword + sp->ts_fidx; | |
| 5336 c = *p; | |
| 5337 if (c == NUL) | |
| 5338 { | |
| 5339 /* 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
|
5340 PROF_STORE(sp->ts_state) |
| 625 | 5341 sp->ts_state = STATE_FINAL; |
| 5342 break; | |
| 5343 } | |
| 5344 | |
| 5345 /* Don't swap if the first character is not a word character. | |
| 5346 * 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
|
5347 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 5348 { |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5349 PROF_STORE(sp->ts_state) |
| 625 | 5350 sp->ts_state = STATE_REP_INI; |
| 323 | 5351 break; |
| 625 | 5352 } |
| 5353 | |
| 5354 #ifdef FEAT_MBYTE | |
| 5355 if (has_mbyte) | |
| 5356 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5357 n = MB_CPTR2LEN(p); |
| 625 | 5358 c = mb_ptr2char(p); |
| 1337 | 5359 if (p[n] == NUL) |
| 5360 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5361 else if (!soundfold && !spell_iswordp(p + n, curwin)) |
| 625 | 5362 c2 = c; /* don't swap non-word char */ |
| 5363 else | |
| 5364 c2 = mb_ptr2char(p + n); | |
| 5365 } | |
| 5366 else | |
| 5367 #endif | |
| 5368 { | |
| 1337 | 5369 if (p[1] == NUL) |
| 5370 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5371 else if (!soundfold && !spell_iswordp(p + 1, curwin)) |
| 625 | 5372 c2 = c; /* don't swap non-word char */ |
| 5373 else | |
| 5374 c2 = p[1]; | |
| 5375 } | |
| 5376 | |
| 1337 | 5377 /* When the second character is NUL we can't swap. */ |
| 5378 if (c2 == NUL) | |
| 5379 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5380 PROF_STORE(sp->ts_state) |
| 1337 | 5381 sp->ts_state = STATE_REP_INI; |
| 5382 break; | |
| 5383 } | |
| 5384 | |
| 625 | 5385 /* When characters are identical, swap won't do anything. |
| 5386 * Also get here if the second char is not a word character. */ | |
| 5387 if (c == c2) | |
| 5388 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5389 PROF_STORE(sp->ts_state) |
| 625 | 5390 sp->ts_state = STATE_SWAP3; |
| 5391 break; | |
| 5392 } | |
| 5393 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP)) | |
| 5394 { | |
| 5395 go_deeper(stack, depth, SCORE_SWAP); | |
| 5396 #ifdef DEBUG_TRIEWALK | |
| 5397 sprintf(changename[depth], "%.*s-%s: swap %c and %c", | |
| 5398 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5399 c, c2); | |
| 5400 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5401 PROF_STORE(sp->ts_state) |
| 625 | 5402 sp->ts_state = STATE_UNSWAP; |
| 5403 ++depth; | |
| 330 | 5404 #ifdef FEAT_MBYTE |
| 5405 if (has_mbyte) | |
| 5406 { | |
| 625 | 5407 fl = mb_char2len(c2); |
| 5408 mch_memmove(p, p + n, fl); | |
| 5409 mb_char2bytes(c, p + fl); | |
| 5410 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 5411 } |
| 5412 else | |
| 5413 #endif | |
| 5414 { | |
| 625 | 5415 p[0] = c2; |
| 330 | 5416 p[1] = c; |
| 625 | 5417 stack[depth].ts_fidxtry = sp->ts_fidx + 2; |
| 5418 } | |
| 5419 } | |
| 5420 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5421 { |
| 625 | 5422 /* 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
|
5423 PROF_STORE(sp->ts_state) |
| 625 | 5424 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
|
5425 } |
| 625 | 5426 break; |
| 5427 | |
| 5428 case STATE_UNSWAP: | |
| 5429 /* Undo the STATE_SWAP swap: "21" -> "12". */ | |
| 5430 p = fword + sp->ts_fidx; | |
| 5431 #ifdef FEAT_MBYTE | |
| 5432 if (has_mbyte) | |
| 5433 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5434 n = MB_PTR2LEN(p); |
| 625 | 5435 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
|
5436 mch_memmove(p + MB_PTR2LEN(p + n), p, n); |
| 625 | 5437 mb_char2bytes(c, p); |
| 5438 } | |
| 5439 else | |
| 5440 #endif | |
| 5441 { | |
| 5442 c = *p; | |
| 5443 *p = p[1]; | |
| 5444 p[1] = c; | |
| 5445 } | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5446 /* FALLTHROUGH */ |
| 625 | 5447 |
| 5448 case STATE_SWAP3: | |
| 5449 /* Swap two bytes, skipping one: "123" -> "321". We change | |
| 5450 * "fword" here, it's changed back afterwards at STATE_UNSWAP3. */ | |
| 5451 p = fword + sp->ts_fidx; | |
| 5452 #ifdef FEAT_MBYTE | |
| 5453 if (has_mbyte) | |
| 5454 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5455 n = MB_CPTR2LEN(p); |
| 625 | 5456 c = mb_ptr2char(p); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5457 fl = MB_CPTR2LEN(p + n); |
| 625 | 5458 c2 = mb_ptr2char(p + n); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5459 if (!soundfold && !spell_iswordp(p + n + fl, curwin)) |
| 625 | 5460 c3 = c; /* don't swap non-word char */ |
| 5461 else | |
| 5462 c3 = mb_ptr2char(p + n + fl); | |
| 5463 } | |
| 5464 else | |
| 5465 #endif | |
| 5466 { | |
| 5467 c = *p; | |
| 5468 c2 = p[1]; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5469 if (!soundfold && !spell_iswordp(p + 2, curwin)) |
| 625 | 5470 c3 = c; /* don't swap non-word char */ |
| 5471 else | |
| 5472 c3 = p[2]; | |
| 5473 } | |
| 5474 | |
| 5475 /* When characters are identical: "121" then SWAP3 result is | |
| 5476 * identical, ROT3L result is same as SWAP: "211", ROT3L result is | |
| 5477 * same as SWAP on next char: "112". Thus skip all swapping. | |
| 5478 * Also skip when c3 is NUL. | |
| 5479 * Also get here when the third character is not a word character. | |
| 5480 * Second character may any char: "a.b" -> "b.a" */ | |
| 5481 if (c == c3 || c3 == NUL) | |
| 5482 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5483 PROF_STORE(sp->ts_state) |
| 625 | 5484 sp->ts_state = STATE_REP_INI; |
| 5485 break; | |
| 5486 } | |
| 5487 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5488 { | |
| 5489 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5490 #ifdef DEBUG_TRIEWALK | |
| 5491 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c", | |
| 5492 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5493 c, c3); | |
| 5494 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5495 PROF_STORE(sp->ts_state) |
| 625 | 5496 sp->ts_state = STATE_UNSWAP3; |
| 5497 ++depth; | |
| 5498 #ifdef FEAT_MBYTE | |
| 5499 if (has_mbyte) | |
| 5500 { | |
| 5501 tl = mb_char2len(c3); | |
| 5502 mch_memmove(p, p + n + fl, tl); | |
| 5503 mb_char2bytes(c2, p + tl); | |
| 5504 mb_char2bytes(c, p + fl + tl); | |
| 5505 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl; | |
| 5506 } | |
| 5507 else | |
| 5508 #endif | |
| 5509 { | |
| 5510 p[0] = p[2]; | |
| 5511 p[2] = c; | |
| 5512 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 5513 } | |
| 5514 } | |
| 5515 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5516 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5517 PROF_STORE(sp->ts_state) |
| 625 | 5518 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
|
5519 } |
| 625 | 5520 break; |
| 5521 | |
| 5522 case STATE_UNSWAP3: | |
| 5523 /* Undo STATE_SWAP3: "321" -> "123" */ | |
| 5524 p = fword + sp->ts_fidx; | |
| 5525 #ifdef FEAT_MBYTE | |
| 5526 if (has_mbyte) | |
| 5527 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5528 n = MB_PTR2LEN(p); |
| 625 | 5529 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
|
5530 fl = MB_PTR2LEN(p + n); |
| 625 | 5531 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
|
5532 tl = MB_PTR2LEN(p + n + fl); |
| 625 | 5533 mch_memmove(p + fl + tl, p, n); |
| 5534 mb_char2bytes(c, p); | |
| 5535 mb_char2bytes(c2, p + tl); | |
| 5536 p = p + tl; | |
| 5537 } | |
| 5538 else | |
| 5539 #endif | |
| 5540 { | |
| 5541 c = *p; | |
| 5542 *p = p[2]; | |
| 5543 p[2] = c; | |
| 5544 ++p; | |
| 5545 } | |
| 5546 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5547 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 5548 { |
| 5549 /* Middle char is not a word char, skip the rotate. First and | |
| 5550 * 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
|
5551 PROF_STORE(sp->ts_state) |
| 625 | 5552 sp->ts_state = STATE_REP_INI; |
| 5553 break; | |
| 5554 } | |
| 5555 | |
| 5556 /* Rotate three characters left: "123" -> "231". We change | |
| 5557 * "fword" here, it's changed back afterwards at STATE_UNROT3L. */ | |
| 5558 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5559 { | |
| 5560 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5561 #ifdef DEBUG_TRIEWALK | |
| 5562 p = fword + sp->ts_fidx; | |
| 5563 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c", | |
| 5564 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5565 p[0], p[1], p[2]); | |
| 5566 #endif | |
|
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_UNROT3L; |
| 5569 ++depth; | |
| 330 | 5570 p = fword + sp->ts_fidx; |
| 5571 #ifdef FEAT_MBYTE | |
| 5572 if (has_mbyte) | |
| 5573 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5574 n = MB_CPTR2LEN(p); |
| 330 | 5575 c = mb_ptr2char(p); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5576 fl = MB_CPTR2LEN(p + n); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5577 fl += MB_CPTR2LEN(p + n + fl); |
| 625 | 5578 mch_memmove(p, p + n, fl); |
| 5579 mb_char2bytes(c, p + fl); | |
| 5580 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 5581 } |
| 5582 else | |
| 5583 #endif | |
| 323 | 5584 { |
| 330 | 5585 c = *p; |
| 625 | 5586 *p = p[1]; |
| 5587 p[1] = p[2]; | |
| 5588 p[2] = c; | |
| 5589 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 5590 } | |
| 5591 } | |
| 5592 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5593 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5594 PROF_STORE(sp->ts_state) |
| 625 | 5595 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
|
5596 } |
| 625 | 5597 break; |
| 5598 | |
| 5599 case STATE_UNROT3L: | |
| 5600 /* Undo ROT3L: "231" -> "123" */ | |
| 5601 p = fword + sp->ts_fidx; | |
| 5602 #ifdef FEAT_MBYTE | |
| 5603 if (has_mbyte) | |
| 5604 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5605 n = MB_PTR2LEN(p); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5606 n += MB_PTR2LEN(p + n); |
| 625 | 5607 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
|
5608 tl = MB_PTR2LEN(p + n); |
| 625 | 5609 mch_memmove(p + tl, p, n); |
| 5610 mb_char2bytes(c, p); | |
| 5611 } | |
| 5612 else | |
| 5613 #endif | |
| 5614 { | |
| 5615 c = p[2]; | |
| 5616 p[2] = p[1]; | |
| 5617 p[1] = *p; | |
| 5618 *p = c; | |
| 5619 } | |
| 5620 | |
| 5621 /* Rotate three bytes right: "123" -> "312". We change "fword" | |
| 5622 * here, it's changed back afterwards at STATE_UNROT3R. */ | |
| 5623 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5624 { | |
| 5625 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5626 #ifdef DEBUG_TRIEWALK | |
| 5627 p = fword + sp->ts_fidx; | |
| 5628 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c", | |
| 5629 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5630 p[0], p[1], p[2]); | |
| 5631 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5632 PROF_STORE(sp->ts_state) |
| 625 | 5633 sp->ts_state = STATE_UNROT3R; |
| 5634 ++depth; | |
| 330 | 5635 p = fword + sp->ts_fidx; |
| 5636 #ifdef FEAT_MBYTE | |
| 5637 if (has_mbyte) | |
| 5638 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5639 n = MB_CPTR2LEN(p); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5640 n += MB_CPTR2LEN(p + n); |
| 330 | 5641 c = mb_ptr2char(p + n); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5642 tl = MB_CPTR2LEN(p + n); |
| 330 | 5643 mch_memmove(p + tl, p, n); |
| 5644 mb_char2bytes(c, p); | |
| 625 | 5645 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl; |
| 330 | 5646 } |
| 5647 else | |
| 5648 #endif | |
| 5649 { | |
| 5650 c = p[2]; | |
| 5651 p[2] = p[1]; | |
| 5652 p[1] = *p; | |
| 5653 *p = c; | |
| 625 | 5654 stack[depth].ts_fidxtry = sp->ts_fidx + 3; |
| 5655 } | |
| 5656 } | |
| 5657 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5658 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5659 PROF_STORE(sp->ts_state) |
| 625 | 5660 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
|
5661 } |
| 625 | 5662 break; |
| 5663 | |
| 5664 case STATE_UNROT3R: | |
| 5665 /* Undo ROT3R: "312" -> "123" */ | |
| 5666 p = fword + sp->ts_fidx; | |
| 5667 #ifdef FEAT_MBYTE | |
| 5668 if (has_mbyte) | |
| 5669 { | |
| 5670 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
|
5671 tl = MB_PTR2LEN(p); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5672 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
|
5673 n += MB_PTR2LEN(p + tl + n); |
| 625 | 5674 mch_memmove(p, p + tl, n); |
| 5675 mb_char2bytes(c, p + n); | |
| 5676 } | |
| 5677 else | |
| 5678 #endif | |
| 5679 { | |
| 5680 c = *p; | |
| 5681 *p = p[1]; | |
| 5682 p[1] = p[2]; | |
| 5683 p[2] = c; | |
| 5684 } | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5685 /* FALLTHROUGH */ |
| 625 | 5686 |
| 5687 case STATE_REP_INI: | |
| 5688 /* Check if matching with REP items from the .aff file would work. | |
| 5689 * Quickly skip if: | |
| 5690 * - there are no REP items and we are not in the soundfold trie | |
| 5691 * - the score is going to be too high anyway | |
| 5692 * - already applied a REP item or swapped here */ | |
| 5693 if ((lp->lp_replang == NULL && !soundfold) | |
| 5694 || sp->ts_score + SCORE_REP >= su->su_maxscore | |
| 5695 || sp->ts_fidx < sp->ts_fidxtry) | |
| 5696 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5697 PROF_STORE(sp->ts_state) |
| 625 | 5698 sp->ts_state = STATE_FINAL; |
| 323 | 5699 break; |
| 625 | 5700 } |
| 5701 | |
| 5702 /* Use the first byte to quickly find the first entry that may | |
| 5703 * match. If the index is -1 there is none. */ | |
| 5704 if (soundfold) | |
| 5705 sp->ts_curi = slang->sl_repsal_first[fword[sp->ts_fidx]]; | |
| 5706 else | |
| 5707 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]]; | |
| 5708 | |
| 5709 if (sp->ts_curi < 0) | |
| 5710 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5711 PROF_STORE(sp->ts_state) |
| 625 | 5712 sp->ts_state = STATE_FINAL; |
| 5713 break; | |
| 5714 } | |
| 5715 | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5716 PROF_STORE(sp->ts_state) |
| 625 | 5717 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
|
5718 /* FALLTHROUGH */ |
| 625 | 5719 |
| 5720 case STATE_REP: | |
| 5721 /* Try matching with REP items from the .aff file. For each match | |
| 5722 * replace the characters and check if the resulting word is | |
| 5723 * valid. */ | |
| 5724 p = fword + sp->ts_fidx; | |
| 5725 | |
| 5726 if (soundfold) | |
| 5727 gap = &slang->sl_repsal; | |
| 5728 else | |
| 503 | 5729 gap = &lp->lp_replang->sl_rep; |
| 625 | 5730 while (sp->ts_curi < gap->ga_len) |
| 5731 { | |
| 5732 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++; | |
| 5733 if (*ftp->ft_from != *p) | |
| 5734 { | |
| 5735 /* past possible matching entries */ | |
| 5736 sp->ts_curi = gap->ga_len; | |
| 323 | 5737 break; |
| 5738 } | |
| 625 | 5739 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0 |
| 5740 && TRY_DEEPER(su, stack, depth, SCORE_REP)) | |
| 5741 { | |
| 5742 go_deeper(stack, depth, SCORE_REP); | |
| 5743 #ifdef DEBUG_TRIEWALK | |
| 5744 sprintf(changename[depth], "%.*s-%s: replace %s with %s", | |
| 5745 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5746 ftp->ft_from, ftp->ft_to); | |
| 5747 #endif | |
| 5748 /* Need to undo this afterwards. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5749 PROF_STORE(sp->ts_state) |
| 625 | 5750 sp->ts_state = STATE_REP_UNDO; |
| 5751 | |
| 5752 /* Change the "from" to the "to" string. */ | |
| 5753 ++depth; | |
| 835 | 5754 fl = (int)STRLEN(ftp->ft_from); |
| 5755 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 5756 if (fl != tl) |
| 323 | 5757 { |
| 1619 | 5758 STRMOVE(p + tl, p + fl); |
| 625 | 5759 repextra += tl - fl; |
| 323 | 5760 } |
| 625 | 5761 mch_memmove(p, ftp->ft_to, tl); |
| 5762 stack[depth].ts_fidxtry = sp->ts_fidx + tl; | |
| 5763 #ifdef FEAT_MBYTE | |
| 5764 stack[depth].ts_tcharlen = 0; | |
| 5765 #endif | |
| 5766 break; | |
| 5767 } | |
| 5768 } | |
| 5769 | |
| 5770 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
|
5771 { |
| 625 | 5772 /* No (more) matches. */ |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5773 PROF_STORE(sp->ts_state) |
| 625 | 5774 sp->ts_state = STATE_FINAL; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5775 } |
| 625 | 5776 |
| 5777 break; | |
| 5778 | |
| 5779 case STATE_REP_UNDO: | |
| 5780 /* Undo a REP replacement and continue with the next one. */ | |
| 5781 if (soundfold) | |
| 5782 gap = &slang->sl_repsal; | |
| 5783 else | |
| 5784 gap = &lp->lp_replang->sl_rep; | |
| 5785 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1; | |
| 835 | 5786 fl = (int)STRLEN(ftp->ft_from); |
| 5787 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 5788 p = fword + sp->ts_fidx; |
| 5789 if (fl != tl) | |
| 5790 { | |
| 1619 | 5791 STRMOVE(p + fl, p + tl); |
| 625 | 5792 repextra -= tl - fl; |
| 5793 } | |
| 5794 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
|
5795 PROF_STORE(sp->ts_state) |
| 625 | 5796 sp->ts_state = STATE_REP; |
| 5797 break; | |
| 5798 | |
| 5799 default: | |
| 5800 /* Did all possible states at this level, go up one level. */ | |
| 5801 --depth; | |
| 5802 | |
| 5803 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE) | |
| 5804 { | |
| 5805 /* Continue in or go back to the prefix tree. */ | |
| 5806 byts = pbyts; | |
| 5807 idxs = pidxs; | |
| 5808 } | |
| 5809 | |
| 5810 /* Don't check for CTRL-C too often, it takes time. */ | |
| 5811 if (--breakcheckcount == 0) | |
| 5812 { | |
| 5813 ui_breakcheck(); | |
| 5814 breakcheckcount = 1000; | |
| 5815 } | |
| 5816 } | |
| 5817 } | |
| 5818 } | |
| 5819 | |
| 5820 | |
| 5821 /* | |
| 5822 * Go one level deeper in the tree. | |
| 5823 */ | |
| 5824 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5825 go_deeper(trystate_T *stack, int depth, int score_add) |
| 323 | 5826 { |
| 330 | 5827 stack[depth + 1] = stack[depth]; |
| 323 | 5828 stack[depth + 1].ts_state = STATE_START; |
| 625 | 5829 stack[depth + 1].ts_score = stack[depth].ts_score + score_add; |
| 323 | 5830 stack[depth + 1].ts_curi = 1; /* start just after length byte */ |
| 485 | 5831 stack[depth + 1].ts_flags = 0; |
| 323 | 5832 } |
| 5833 | |
| 455 | 5834 #ifdef FEAT_MBYTE |
| 5835 /* | |
| 5836 * Case-folding may change the number of bytes: Count nr of chars in | |
| 5837 * fword[flen] and return the byte length of that many chars in "word". | |
| 5838 */ | |
| 5839 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5840 nofold_len(char_u *fword, int flen, char_u *word) |
| 455 | 5841 { |
| 5842 char_u *p; | |
| 5843 int i = 0; | |
| 5844 | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5845 for (p = fword; p < fword + flen; MB_PTR_ADV(p)) |
| 455 | 5846 ++i; |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5847 for (p = word; i > 0; MB_PTR_ADV(p)) |
| 455 | 5848 --i; |
| 5849 return (int)(p - word); | |
| 5850 } | |
| 5851 #endif | |
| 5852 | |
| 323 | 5853 /* |
| 5854 * "fword" is a good word with case folded. Find the matching keep-case | |
| 5855 * words and put it in "kword". | |
| 5856 * Theoretically there could be several keep-case words that result in the | |
| 5857 * same case-folded word, but we only find one... | |
| 5858 */ | |
| 5859 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5860 find_keepcap_word(slang_T *slang, char_u *fword, char_u *kword) |
| 323 | 5861 { |
| 5862 char_u uword[MAXWLEN]; /* "fword" in upper-case */ | |
| 5863 int depth; | |
| 324 | 5864 idx_T tryidx; |
| 323 | 5865 |
| 5866 /* The following arrays are used at each depth in the tree. */ | |
| 324 | 5867 idx_T arridx[MAXWLEN]; |
| 323 | 5868 int round[MAXWLEN]; |
| 5869 int fwordidx[MAXWLEN]; | |
| 5870 int uwordidx[MAXWLEN]; | |
| 5871 int kwordlen[MAXWLEN]; | |
| 5872 | |
| 5873 int flen, ulen; | |
| 5874 int l; | |
| 5875 int len; | |
| 5876 int c; | |
| 324 | 5877 idx_T lo, hi, m; |
| 323 | 5878 char_u *p; |
| 5879 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */ | |
| 324 | 5880 idx_T *idxs = slang->sl_kidxs; /* array with indexes */ |
| 323 | 5881 |
| 5882 if (byts == NULL) | |
| 5883 { | |
| 5884 /* array is empty: "cannot happen" */ | |
| 5885 *kword = NUL; | |
| 5886 return; | |
| 5887 } | |
| 5888 | |
| 5889 /* Make an all-cap version of "fword". */ | |
| 5890 allcap_copy(fword, uword); | |
| 5891 | |
| 5892 /* | |
| 5893 * Each character needs to be tried both case-folded and upper-case. | |
| 5894 * All this gets very complicated if we keep in mind that changing case | |
| 5895 * may change the byte length of a multi-byte character... | |
| 5896 */ | |
| 5897 depth = 0; | |
| 5898 arridx[0] = 0; | |
| 5899 round[0] = 0; | |
| 5900 fwordidx[0] = 0; | |
| 5901 uwordidx[0] = 0; | |
| 5902 kwordlen[0] = 0; | |
| 5903 while (depth >= 0) | |
| 5904 { | |
| 5905 if (fword[fwordidx[depth]] == NUL) | |
| 5906 { | |
| 5907 /* We are at the end of "fword". If the tree allows a word to end | |
| 5908 * here we have found a match. */ | |
| 5909 if (byts[arridx[depth] + 1] == 0) | |
| 5910 { | |
| 5911 kword[kwordlen[depth]] = NUL; | |
| 5912 return; | |
| 5913 } | |
| 5914 | |
| 5915 /* kword is getting too long, continue one level up */ | |
| 5916 --depth; | |
| 5917 } | |
| 5918 else if (++round[depth] > 2) | |
| 5919 { | |
| 5920 /* tried both fold-case and upper-case character, continue one | |
| 5921 * level up */ | |
| 5922 --depth; | |
| 5923 } | |
| 5924 else | |
| 5925 { | |
| 5926 /* | |
| 5927 * round[depth] == 1: Try using the folded-case character. | |
| 5928 * round[depth] == 2: Try using the upper-case character. | |
| 5929 */ | |
| 5930 #ifdef FEAT_MBYTE | |
| 5931 if (has_mbyte) | |
| 5932 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5933 flen = MB_CPTR2LEN(fword + fwordidx[depth]); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5934 ulen = MB_CPTR2LEN(uword + uwordidx[depth]); |
| 323 | 5935 } |
| 5936 else | |
| 5937 #endif | |
| 5938 ulen = flen = 1; | |
| 5939 if (round[depth] == 1) | |
| 5940 { | |
| 5941 p = fword + fwordidx[depth]; | |
| 5942 l = flen; | |
| 5943 } | |
| 5944 else | |
| 5945 { | |
| 5946 p = uword + uwordidx[depth]; | |
| 5947 l = ulen; | |
| 5948 } | |
| 5949 | |
| 5950 for (tryidx = arridx[depth]; l > 0; --l) | |
| 5951 { | |
| 5952 /* Perform a binary search in the list of accepted bytes. */ | |
| 5953 len = byts[tryidx++]; | |
| 5954 c = *p++; | |
| 5955 lo = tryidx; | |
| 5956 hi = tryidx + len - 1; | |
| 5957 while (lo < hi) | |
| 5958 { | |
| 5959 m = (lo + hi) / 2; | |
| 5960 if (byts[m] > c) | |
| 5961 hi = m - 1; | |
| 5962 else if (byts[m] < c) | |
| 5963 lo = m + 1; | |
| 5964 else | |
| 5965 { | |
| 5966 lo = hi = m; | |
| 5967 break; | |
| 5968 } | |
| 5969 } | |
| 5970 | |
| 5971 /* Stop if there is no matching byte. */ | |
| 5972 if (hi < lo || byts[lo] != c) | |
| 5973 break; | |
| 5974 | |
| 5975 /* Continue at the child (if there is one). */ | |
| 5976 tryidx = idxs[lo]; | |
| 5977 } | |
| 5978 | |
| 5979 if (l == 0) | |
| 5980 { | |
| 5981 /* | |
| 5982 * Found the matching char. Copy it to "kword" and go a | |
| 5983 * level deeper. | |
| 5984 */ | |
| 5985 if (round[depth] == 1) | |
| 5986 { | |
| 5987 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth], | |
| 5988 flen); | |
| 5989 kwordlen[depth + 1] = kwordlen[depth] + flen; | |
| 5990 } | |
| 5991 else | |
| 5992 { | |
| 5993 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth], | |
| 5994 ulen); | |
| 5995 kwordlen[depth + 1] = kwordlen[depth] + ulen; | |
| 5996 } | |
| 5997 fwordidx[depth + 1] = fwordidx[depth] + flen; | |
| 5998 uwordidx[depth + 1] = uwordidx[depth] + ulen; | |
| 5999 | |
| 6000 ++depth; | |
| 6001 arridx[depth] = tryidx; | |
| 6002 round[depth] = 0; | |
| 6003 } | |
| 6004 } | |
| 6005 } | |
| 6006 | |
| 6007 /* Didn't find it: "cannot happen". */ | |
| 6008 *kword = NUL; | |
| 6009 } | |
| 6010 | |
| 6011 /* | |
| 344 | 6012 * Compute the sound-a-like score for suggestions in su->su_ga and add them to |
| 6013 * su->su_sga. | |
| 6014 */ | |
| 6015 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6016 score_comp_sal(suginfo_T *su) |
| 344 | 6017 { |
| 6018 langp_T *lp; | |
| 6019 char_u badsound[MAXWLEN]; | |
| 6020 int i; | |
| 6021 suggest_T *stp; | |
| 6022 suggest_T *sstp; | |
| 6023 int score; | |
| 500 | 6024 int lpi; |
| 344 | 6025 |
| 6026 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL) | |
| 6027 return; | |
| 6028 | |
| 6029 /* 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
|
6030 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
|
6031 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6032 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 6033 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6034 { | |
| 6035 /* soundfold the bad word */ | |
| 375 | 6036 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound); |
| 344 | 6037 |
| 6038 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 6039 { | |
| 6040 stp = &SUG(su->su_ga, i); | |
| 6041 | |
| 351 | 6042 /* Case-fold the suggested word, sound-fold it and compute the |
| 6043 * sound-a-like score. */ | |
| 6044 score = stp_sal_score(stp, su, lp->lp_slang, badsound); | |
| 344 | 6045 if (score < SCORE_MAXMAX) |
| 6046 { | |
| 6047 /* Add the suggestion. */ | |
| 6048 sstp = &SUG(su->su_sga, su->su_sga.ga_len); | |
| 6049 sstp->st_word = vim_strsave(stp->st_word); | |
| 6050 if (sstp->st_word != NULL) | |
| 6051 { | |
| 625 | 6052 sstp->st_wordlen = stp->st_wordlen; |
| 344 | 6053 sstp->st_score = score; |
| 6054 sstp->st_altscore = 0; | |
| 6055 sstp->st_orglen = stp->st_orglen; | |
| 6056 ++su->su_sga.ga_len; | |
| 6057 } | |
| 6058 } | |
| 6059 } | |
| 6060 break; | |
| 6061 } | |
| 500 | 6062 } |
| 344 | 6063 } |
| 6064 | |
| 6065 /* | |
| 6066 * Combine the list of suggestions in su->su_ga and su->su_sga. | |
| 4352 | 6067 * They are entwined. |
| 344 | 6068 */ |
| 6069 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6070 score_combine(suginfo_T *su) |
| 344 | 6071 { |
| 6072 int i; | |
| 6073 int j; | |
| 6074 garray_T ga; | |
| 6075 garray_T *gap; | |
| 6076 langp_T *lp; | |
| 6077 suggest_T *stp; | |
| 6078 char_u *p; | |
| 6079 char_u badsound[MAXWLEN]; | |
| 6080 int round; | |
| 500 | 6081 int lpi; |
| 625 | 6082 slang_T *slang = NULL; |
| 344 | 6083 |
| 6084 /* 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
|
6085 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
|
6086 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6087 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 6088 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6089 { | |
| 6090 /* soundfold the bad word */ | |
| 625 | 6091 slang = lp->lp_slang; |
| 6092 spell_soundfold(slang, su->su_fbadword, TRUE, badsound); | |
| 344 | 6093 |
| 6094 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 6095 { | |
| 6096 stp = &SUG(su->su_ga, i); | |
| 625 | 6097 stp->st_altscore = stp_sal_score(stp, su, slang, badsound); |
| 344 | 6098 if (stp->st_altscore == SCORE_MAXMAX) |
| 6099 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4; | |
| 6100 else | |
| 6101 stp->st_score = (stp->st_score * 3 | |
| 6102 + stp->st_altscore) / 4; | |
| 6103 stp->st_salscore = FALSE; | |
| 6104 } | |
| 6105 break; | |
| 6106 } | |
| 6107 } | |
| 6108 | |
| 844 | 6109 if (slang == NULL) /* Using "double" without sound folding. */ |
| 6110 { | |
| 6111 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, | |
| 6112 su->su_maxcount); | |
| 625 | 6113 return; |
| 844 | 6114 } |
| 625 | 6115 |
| 344 | 6116 /* Add the alternate score to su_sga. */ |
| 6117 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 6118 { | |
| 6119 stp = &SUG(su->su_sga, i); | |
| 625 | 6120 stp->st_altscore = spell_edit_score(slang, |
| 6121 su->su_badword, stp->st_word); | |
| 344 | 6122 if (stp->st_score == SCORE_MAXMAX) |
| 6123 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8; | |
| 6124 else | |
| 6125 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8; | |
| 6126 stp->st_salscore = TRUE; | |
| 6127 } | |
| 6128 | |
| 625 | 6129 /* Remove bad suggestions, sort the suggestions and truncate at "maxcount" |
| 6130 * for both lists. */ | |
| 6131 check_suggestions(su, &su->su_ga); | |
| 344 | 6132 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 625 | 6133 check_suggestions(su, &su->su_sga); |
| 344 | 6134 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount); |
| 6135 | |
| 6136 ga_init2(&ga, (int)sizeof(suginfo_T), 1); | |
| 6137 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL) | |
| 6138 return; | |
| 6139 | |
| 6140 stp = &SUG(ga, 0); | |
| 6141 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i) | |
| 6142 { | |
| 6143 /* round 1: get a suggestion from su_ga | |
| 6144 * round 2: get a suggestion from su_sga */ | |
| 6145 for (round = 1; round <= 2; ++round) | |
| 6146 { | |
| 6147 gap = round == 1 ? &su->su_ga : &su->su_sga; | |
| 6148 if (i < gap->ga_len) | |
| 6149 { | |
| 6150 /* Don't add a word if it's already there. */ | |
| 6151 p = SUG(*gap, i).st_word; | |
| 6152 for (j = 0; j < ga.ga_len; ++j) | |
| 6153 if (STRCMP(stp[j].st_word, p) == 0) | |
| 6154 break; | |
| 6155 if (j == ga.ga_len) | |
| 6156 stp[ga.ga_len++] = SUG(*gap, i); | |
| 6157 else | |
| 6158 vim_free(p); | |
| 6159 } | |
| 6160 } | |
| 6161 } | |
| 6162 | |
| 6163 ga_clear(&su->su_ga); | |
| 6164 ga_clear(&su->su_sga); | |
| 6165 | |
| 6166 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 6167 if (ga.ga_len > su->su_maxcount) | |
| 6168 { | |
| 6169 for (i = su->su_maxcount; i < ga.ga_len; ++i) | |
| 6170 vim_free(stp[i].st_word); | |
| 6171 ga.ga_len = su->su_maxcount; | |
| 6172 } | |
| 6173 | |
| 6174 su->su_ga = ga; | |
| 6175 } | |
| 6176 | |
| 6177 /* | |
| 351 | 6178 * For the goodword in "stp" compute the soundalike score compared to the |
| 6179 * badword. | |
| 6180 */ | |
| 6181 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6182 stp_sal_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6183 suggest_T *stp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6184 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6185 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6186 char_u *badsound) /* sound-folded badword */ |
| 351 | 6187 { |
| 6188 char_u *p; | |
| 534 | 6189 char_u *pbad; |
| 6190 char_u *pgood; | |
| 351 | 6191 char_u badsound2[MAXWLEN]; |
| 6192 char_u fword[MAXWLEN]; | |
| 6193 char_u goodsound[MAXWLEN]; | |
| 534 | 6194 char_u goodword[MAXWLEN]; |
| 6195 int lendiff; | |
| 6196 | |
| 6197 lendiff = (int)(su->su_badlen - stp->st_orglen); | |
| 6198 if (lendiff >= 0) | |
| 6199 pbad = badsound; | |
| 351 | 6200 else |
| 6201 { | |
| 6202 /* soundfold the bad word with more characters following */ | |
| 6203 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN); | |
| 6204 | |
| 6205 /* When joining two words the sound often changes a lot. E.g., "t he" | |
| 6206 * sounds like "t h" while "the" sounds like "@". Avoid that by | |
| 6207 * removing the space. Don't do it when the good word also contains a | |
| 6208 * space. */ | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
6209 if (VIM_ISWHITE(su->su_badptr[su->su_badlen]) |
| 351 | 6210 && *skiptowhite(stp->st_word) == NUL) |
| 6211 for (p = fword; *(p = skiptowhite(p)) != NUL; ) | |
| 1619 | 6212 STRMOVE(p, p + 1); |
| 351 | 6213 |
| 375 | 6214 spell_soundfold(slang, fword, TRUE, badsound2); |
| 534 | 6215 pbad = badsound2; |
| 6216 } | |
| 6217 | |
| 2768 | 6218 if (lendiff > 0 && stp->st_wordlen + lendiff < MAXWLEN) |
| 534 | 6219 { |
| 6220 /* Add part of the bad word to the good word, so that we soundfold | |
| 6221 * what replaces the bad word. */ | |
| 6222 STRCPY(goodword, stp->st_word); | |
| 625 | 6223 vim_strncpy(goodword + stp->st_wordlen, |
| 6224 su->su_badptr + su->su_badlen - lendiff, lendiff); | |
| 534 | 6225 pgood = goodword; |
| 6226 } | |
| 6227 else | |
| 6228 pgood = stp->st_word; | |
| 351 | 6229 |
| 375 | 6230 /* Sound-fold the word and compute the score for the difference. */ |
| 534 | 6231 spell_soundfold(slang, pgood, FALSE, goodsound); |
| 6232 | |
| 6233 return soundalike_score(goodsound, pbad); | |
| 351 | 6234 } |
| 6235 | |
| 625 | 6236 /* structure used to store soundfolded words that add_sound_suggest() has |
| 6237 * handled already. */ | |
| 6238 typedef struct | |
| 6239 { | |
| 6240 short sft_score; /* lowest score used */ | |
| 6241 char_u sft_word[1]; /* soundfolded word, actually longer */ | |
| 6242 } sftword_T; | |
| 6243 | |
| 6244 static sftword_T dumsft; | |
| 6245 #define HIKEY2SFT(p) ((sftword_T *)(p - (dumsft.sft_word - (char_u *)&dumsft))) | |
| 6246 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key) | |
| 6247 | |
| 6248 /* | |
| 6249 * Prepare for calling suggest_try_soundalike(). | |
| 6250 */ | |
| 6251 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6252 suggest_try_soundalike_prep(void) |
| 625 | 6253 { |
| 6254 langp_T *lp; | |
| 6255 int lpi; | |
| 6256 slang_T *slang; | |
| 6257 | |
| 6258 /* Do this for all languages that support sound folding and for which a | |
| 6259 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6260 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
|
6261 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6262 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 6263 slang = lp->lp_slang; |
| 6264 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 6265 /* prepare the hashtable used by add_sound_suggest() */ | |
| 6266 hash_init(&slang->sl_sounddone); | |
| 6267 } | |
| 6268 } | |
| 6269 | |
| 351 | 6270 /* |
| 323 | 6271 * Find suggestions by comparing the word in a sound-a-like form. |
| 503 | 6272 * Note: This doesn't support postponed prefixes. |
| 323 | 6273 */ |
| 6274 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6275 suggest_try_soundalike(suginfo_T *su) |
| 323 | 6276 { |
| 6277 char_u salword[MAXWLEN]; | |
| 6278 langp_T *lp; | |
| 500 | 6279 int lpi; |
| 503 | 6280 slang_T *slang; |
| 344 | 6281 |
| 625 | 6282 /* Do this for all languages that support sound folding and for which a |
| 6283 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6284 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
|
6285 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6286 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 6287 slang = lp->lp_slang; |
| 6288 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 6289 { | |
| 6290 /* soundfold the bad word */ | |
| 6291 spell_soundfold(slang, su->su_fbadword, TRUE, salword); | |
| 6292 | |
| 6293 /* try all kinds of inserts/deletes/swaps/etc. */ | |
| 6294 /* TODO: also soundfold the next words, so that we can try joining | |
| 6295 * and splitting */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6296 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6297 prof_init(); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6298 #endif |
| 625 | 6299 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
|
6300 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6301 prof_report("soundalike"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6302 #endif |
| 625 | 6303 } |
| 6304 } | |
| 6305 } | |
| 6306 | |
| 6307 /* | |
| 6308 * Finish up after calling suggest_try_soundalike(). | |
| 6309 */ | |
| 6310 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6311 suggest_try_soundalike_finish(void) |
| 625 | 6312 { |
| 6313 langp_T *lp; | |
| 6314 int lpi; | |
| 6315 slang_T *slang; | |
| 6316 int todo; | |
| 6317 hashitem_T *hi; | |
| 6318 | |
| 6319 /* Do this for all languages that support sound folding and for which a | |
| 6320 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6321 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
|
6322 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6323 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 503 | 6324 slang = lp->lp_slang; |
| 625 | 6325 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) |
| 6326 { | |
| 6327 /* Free the info about handled words. */ | |
| 835 | 6328 todo = (int)slang->sl_sounddone.ht_used; |
| 625 | 6329 for (hi = slang->sl_sounddone.ht_array; todo > 0; ++hi) |
| 6330 if (!HASHITEM_EMPTY(hi)) | |
| 6331 { | |
| 6332 vim_free(HI2SFT(hi)); | |
| 6333 --todo; | |
| 6334 } | |
| 1087 | 6335 |
| 6336 /* Clear the hashtable, it may also be used by another region. */ | |
| 625 | 6337 hash_clear(&slang->sl_sounddone); |
| 1087 | 6338 hash_init(&slang->sl_sounddone); |
| 625 | 6339 } |
| 6340 } | |
| 6341 } | |
| 6342 | |
| 6343 /* | |
| 6344 * A match with a soundfolded word is found. Add the good word(s) that | |
| 6345 * produce this soundfolded word. | |
| 6346 */ | |
| 6347 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6348 add_sound_suggest( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6349 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6350 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6351 int score, /* soundfold score */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6352 langp_T *lp) |
| 625 | 6353 { |
| 6354 slang_T *slang = lp->lp_slang; /* language for sound folding */ | |
| 6355 int sfwordnr; | |
| 6356 char_u *nrline; | |
| 6357 int orgnr; | |
| 6358 char_u theword[MAXWLEN]; | |
| 6359 int i; | |
| 6360 int wlen; | |
| 6361 char_u *byts; | |
| 6362 idx_T *idxs; | |
| 6363 int n; | |
| 6364 int wordcount; | |
| 6365 int wc; | |
| 6366 int goodscore; | |
| 6367 hash_T hash; | |
| 6368 hashitem_T *hi; | |
| 6369 sftword_T *sft; | |
| 6370 int bc, gc; | |
| 6371 int limit; | |
| 6372 | |
| 6373 /* | |
| 6374 * It's very well possible that the same soundfold word is found several | |
| 6375 * times with different scores. Since the following is quite slow only do | |
| 6376 * the words that have a better score than before. Use a hashtable to | |
| 6377 * remember the words that have been done. | |
| 6378 */ | |
| 6379 hash = hash_hash(goodword); | |
| 6380 hi = hash_lookup(&slang->sl_sounddone, goodword, hash); | |
| 6381 if (HASHITEM_EMPTY(hi)) | |
| 6382 { | |
| 844 | 6383 sft = (sftword_T *)alloc((unsigned)(sizeof(sftword_T) |
| 6384 + STRLEN(goodword))); | |
| 625 | 6385 if (sft != NULL) |
| 6386 { | |
| 6387 sft->sft_score = score; | |
| 6388 STRCPY(sft->sft_word, goodword); | |
| 6389 hash_add_item(&slang->sl_sounddone, hi, sft->sft_word, hash); | |
| 6390 } | |
| 6391 } | |
| 6392 else | |
| 6393 { | |
| 6394 sft = HI2SFT(hi); | |
| 6395 if (score >= sft->sft_score) | |
| 6396 return; | |
| 6397 sft->sft_score = score; | |
| 6398 } | |
| 6399 | |
| 6400 /* | |
| 6401 * Find the word nr in the soundfold tree. | |
| 6402 */ | |
| 6403 sfwordnr = soundfold_find(slang, goodword); | |
| 6404 if (sfwordnr < 0) | |
| 6405 { | |
|
10359
66f1b5bf3fa6
commit https://github.com/vim/vim/commit/95f096030ed1a8afea028f2ea295d6f6a70f466f
Christian Brabandt <cb@256bit.org>
parents:
10277
diff
changeset
|
6406 internal_error("add_sound_suggest()"); |
| 625 | 6407 return; |
| 6408 } | |
| 6409 | |
| 6410 /* | |
| 6411 * go over the list of good words that produce this soundfold word | |
| 6412 */ | |
| 6413 nrline = ml_get_buf(slang->sl_sugbuf, (linenr_T)(sfwordnr + 1), FALSE); | |
| 6414 orgnr = 0; | |
| 6415 while (*nrline != NUL) | |
| 6416 { | |
| 6417 /* The wordnr was stored in a minimal nr of bytes as an offset to the | |
| 6418 * previous wordnr. */ | |
| 6419 orgnr += bytes2offset(&nrline); | |
| 6420 | |
| 6421 byts = slang->sl_fbyts; | |
| 6422 idxs = slang->sl_fidxs; | |
| 6423 | |
| 6424 /* Lookup the word "orgnr" one of the two tries. */ | |
| 6425 n = 0; | |
| 6426 wordcount = 0; | |
| 5506 | 6427 for (wlen = 0; wlen < MAXWLEN - 3; ++wlen) |
| 625 | 6428 { |
| 6429 i = 1; | |
| 6430 if (wordcount == orgnr && byts[n + 1] == NUL) | |
| 6431 break; /* found end of word */ | |
| 6432 | |
| 6433 if (byts[n + 1] == NUL) | |
| 6434 ++wordcount; | |
| 6435 | |
| 6436 /* skip over the NUL bytes */ | |
| 6437 for ( ; byts[n + i] == NUL; ++i) | |
| 6438 if (i > byts[n]) /* safety check */ | |
| 6439 { | |
| 6440 STRCPY(theword + wlen, "BAD"); | |
| 5506 | 6441 wlen += 3; |
| 625 | 6442 goto badword; |
| 6443 } | |
| 6444 | |
| 6445 /* One of the siblings must have the word. */ | |
| 6446 for ( ; i < byts[n]; ++i) | |
| 6447 { | |
| 6448 wc = idxs[idxs[n + i]]; /* nr of words under this byte */ | |
| 6449 if (wordcount + wc > orgnr) | |
| 6450 break; | |
| 6451 wordcount += wc; | |
| 6452 } | |
| 6453 | |
| 5506 | 6454 theword[wlen] = byts[n + i]; |
| 625 | 6455 n = idxs[n + i]; |
| 6456 } | |
| 6457 badword: | |
| 6458 theword[wlen] = NUL; | |
| 6459 | |
| 6460 /* Go over the possible flags and regions. */ | |
| 6461 for (; i <= byts[n] && byts[n + i] == NUL; ++i) | |
| 6462 { | |
| 6463 char_u cword[MAXWLEN]; | |
| 6464 char_u *p; | |
| 6465 int flags = (int)idxs[n + i]; | |
| 6466 | |
| 698 | 6467 /* Skip words with the NOSUGGEST flag */ |
| 6468 if (flags & WF_NOSUGGEST) | |
| 6469 continue; | |
| 6470 | |
| 625 | 6471 if (flags & WF_KEEPCAP) |
| 6472 { | |
| 6473 /* Must find the word in the keep-case tree. */ | |
| 6474 find_keepcap_word(slang, theword, cword); | |
| 6475 p = cword; | |
| 6476 } | |
| 6477 else | |
| 6478 { | |
| 6479 flags |= su->su_badflags; | |
| 6480 if ((flags & WF_CAPMASK) != 0) | |
| 6481 { | |
| 6482 /* Need to fix case according to "flags". */ | |
| 6483 make_case_word(theword, cword, flags); | |
| 6484 p = cword; | |
| 323 | 6485 } |
| 6486 else | |
| 625 | 6487 p = theword; |
| 6488 } | |
| 6489 | |
| 6490 /* Add the suggestion. */ | |
| 6491 if (sps_flags & SPS_DOUBLE) | |
| 6492 { | |
| 6493 /* Add the suggestion if the score isn't too bad. */ | |
| 6494 if (score <= su->su_maxscore) | |
| 6495 add_suggestion(su, &su->su_sga, p, su->su_badlen, | |
| 6496 score, 0, FALSE, slang, FALSE); | |
| 6497 } | |
| 6498 else | |
| 6499 { | |
| 6500 /* Add a penalty for words in another region. */ | |
| 6501 if ((flags & WF_REGION) | |
| 6502 && (((unsigned)flags >> 16) & lp->lp_region) == 0) | |
| 6503 goodscore = SCORE_REGION; | |
| 6504 else | |
| 6505 goodscore = 0; | |
| 6506 | |
| 6507 /* Add a small penalty for changing the first letter from | |
| 6508 * lower to upper case. Helps for "tath" -> "Kath", which is | |
| 4352 | 6509 * less common than "tath" -> "path". Don't do it when the |
| 625 | 6510 * letter is the same, that has already been counted. */ |
| 6511 gc = PTR2CHAR(p); | |
| 6512 if (SPELL_ISUPPER(gc)) | |
| 6513 { | |
| 6514 bc = PTR2CHAR(su->su_badword); | |
| 6515 if (!SPELL_ISUPPER(bc) | |
| 6516 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc)) | |
| 6517 goodscore += SCORE_ICASE / 2; | |
| 6518 } | |
| 6519 | |
| 6520 /* Compute the score for the good word. This only does letter | |
| 6521 * insert/delete/swap/replace. REP items are not considered, | |
| 6522 * which may make the score a bit higher. | |
| 6523 * Use a limit for the score to make it work faster. Use | |
| 6524 * MAXSCORE(), because RESCORE() will change the score. | |
| 6525 * If the limit is very high then the iterative method is | |
| 6526 * inefficient, using an array is quicker. */ | |
| 6527 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score); | |
| 6528 if (limit > SCORE_LIMITMAX) | |
| 6529 goodscore += spell_edit_score(slang, su->su_badword, p); | |
| 6530 else | |
| 6531 goodscore += spell_edit_score_limit(slang, su->su_badword, | |
| 6532 p, limit); | |
| 6533 | |
| 6534 /* When going over the limit don't bother to do the rest. */ | |
| 6535 if (goodscore < SCORE_MAXMAX) | |
| 6536 { | |
| 6537 /* Give a bonus to words seen before. */ | |
| 6538 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE); | |
| 6539 | |
| 6540 /* Add the suggestion if the score isn't too bad. */ | |
| 6541 goodscore = RESCORE(goodscore, score); | |
| 6542 if (goodscore <= su->su_sfmaxscore) | |
| 6543 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 6544 goodscore, score, TRUE, slang, TRUE); | |
| 6545 } | |
| 6546 } | |
| 6547 } | |
| 6548 /* smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr); */ | |
| 6549 } | |
| 6550 } | |
| 6551 | |
| 6552 /* | |
| 6553 * Find word "word" in fold-case tree for "slang" and return the word number. | |
| 6554 */ | |
| 6555 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6556 soundfold_find(slang_T *slang, char_u *word) |
| 625 | 6557 { |
| 6558 idx_T arridx = 0; | |
| 6559 int len; | |
| 6560 int wlen = 0; | |
| 6561 int c; | |
| 6562 char_u *ptr = word; | |
| 6563 char_u *byts; | |
| 6564 idx_T *idxs; | |
| 6565 int wordnr = 0; | |
| 6566 | |
| 6567 byts = slang->sl_sbyts; | |
| 6568 idxs = slang->sl_sidxs; | |
| 6569 | |
| 6570 for (;;) | |
| 6571 { | |
| 6572 /* First byte is the number of possible bytes. */ | |
| 6573 len = byts[arridx++]; | |
| 6574 | |
| 6575 /* If the first possible byte is a zero the word could end here. | |
| 6576 * If the word ends we found the word. If not skip the NUL bytes. */ | |
| 6577 c = ptr[wlen]; | |
| 6578 if (byts[arridx] == NUL) | |
| 6579 { | |
| 6580 if (c == NUL) | |
| 6581 break; | |
| 6582 | |
| 6583 /* Skip over the zeros, there can be several. */ | |
| 6584 while (len > 0 && byts[arridx] == NUL) | |
| 6585 { | |
| 6586 ++arridx; | |
| 6587 --len; | |
| 6588 } | |
| 6589 if (len == 0) | |
| 6590 return -1; /* no children, word should have ended here */ | |
| 6591 ++wordnr; | |
| 6592 } | |
| 6593 | |
| 6594 /* If the word ends we didn't find it. */ | |
| 6595 if (c == NUL) | |
| 6596 return -1; | |
| 6597 | |
| 6598 /* Perform a binary search in the list of accepted bytes. */ | |
| 6599 if (c == TAB) /* <Tab> is handled like <Space> */ | |
| 6600 c = ' '; | |
| 6601 while (byts[arridx] < c) | |
| 6602 { | |
| 6603 /* The word count is in the first idxs[] entry of the child. */ | |
| 6604 wordnr += idxs[idxs[arridx]]; | |
| 6605 ++arridx; | |
| 6606 if (--len == 0) /* end of the bytes, didn't find it */ | |
| 6607 return -1; | |
| 6608 } | |
| 6609 if (byts[arridx] != c) /* didn't find the byte */ | |
| 6610 return -1; | |
| 6611 | |
| 6612 /* Continue at the child (if there is one). */ | |
| 6613 arridx = idxs[arridx]; | |
| 6614 ++wlen; | |
| 6615 | |
| 6616 /* One space in the good word may stand for several spaces in the | |
| 6617 * checked word. */ | |
| 6618 if (c == ' ') | |
| 6619 while (ptr[wlen] == ' ' || ptr[wlen] == TAB) | |
| 6620 ++wlen; | |
| 6621 } | |
| 6622 | |
| 6623 return wordnr; | |
| 323 | 6624 } |
| 6625 | |
| 6626 /* | |
| 324 | 6627 * Copy "fword" to "cword", fixing case according to "flags". |
| 323 | 6628 */ |
| 6629 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6630 make_case_word(char_u *fword, char_u *cword, int flags) |
| 323 | 6631 { |
| 6632 if (flags & WF_ALLCAP) | |
| 6633 /* Make it all upper-case */ | |
| 6634 allcap_copy(fword, cword); | |
| 6635 else if (flags & WF_ONECAP) | |
| 6636 /* Make the first letter upper-case */ | |
| 324 | 6637 onecap_copy(fword, cword, TRUE); |
| 323 | 6638 else |
| 6639 /* Use goodword as-is. */ | |
| 6640 STRCPY(cword, fword); | |
| 6641 } | |
| 6642 | |
| 330 | 6643 |
| 323 | 6644 /* |
| 6645 * Return TRUE if "c1" and "c2" are similar characters according to the MAP | |
| 6646 * lines in the .aff file. | |
| 6647 */ | |
| 6648 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6649 similar_chars(slang_T *slang, int c1, int c2) |
| 323 | 6650 { |
| 330 | 6651 int m1, m2; |
| 6652 #ifdef FEAT_MBYTE | |
| 3549 | 6653 char_u buf[MB_MAXBYTES + 1]; |
| 330 | 6654 hashitem_T *hi; |
| 6655 | |
| 6656 if (c1 >= 256) | |
| 6657 { | |
| 6658 buf[mb_char2bytes(c1, buf)] = 0; | |
| 6659 hi = hash_find(&slang->sl_map_hash, buf); | |
| 6660 if (HASHITEM_EMPTY(hi)) | |
| 6661 m1 = 0; | |
| 6662 else | |
| 6663 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 6664 } | |
| 6665 else | |
| 6666 #endif | |
| 6667 m1 = slang->sl_map_array[c1]; | |
| 6668 if (m1 == 0) | |
| 323 | 6669 return FALSE; |
| 330 | 6670 |
| 6671 | |
| 6672 #ifdef FEAT_MBYTE | |
| 6673 if (c2 >= 256) | |
| 6674 { | |
| 6675 buf[mb_char2bytes(c2, buf)] = 0; | |
| 6676 hi = hash_find(&slang->sl_map_hash, buf); | |
| 6677 if (HASHITEM_EMPTY(hi)) | |
| 6678 m2 = 0; | |
| 6679 else | |
| 6680 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 6681 } | |
| 6682 else | |
| 6683 #endif | |
| 6684 m2 = slang->sl_map_array[c2]; | |
| 6685 | |
| 6686 return m1 == m2; | |
| 323 | 6687 } |
| 6688 | |
| 6689 /* | |
| 6690 * Add a suggestion to the list of suggestions. | |
| 625 | 6691 * For a suggestion that is already in the list the lowest score is remembered. |
| 6692 */ | |
| 6693 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6694 add_suggestion( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6695 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6696 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
|
6697 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6698 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
|
6699 int score, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6700 int altscore, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6701 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
|
6702 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
|
6703 int maxsf) /* su_maxscore applies to soundfold score, |
| 625 | 6704 su_sfmaxscore to the total score. */ |
| 6705 { | |
| 6706 int goodlen; /* len of goodword changed */ | |
| 6707 int badlen; /* len of bad word changed */ | |
| 323 | 6708 suggest_T *stp; |
| 534 | 6709 suggest_T new_sug; |
| 323 | 6710 int i; |
| 534 | 6711 char_u *pgood, *pbad; |
| 532 | 6712 |
| 534 | 6713 /* Minimize "badlen" for consistency. Avoids that changing "the the" to |
| 6714 * "thee the" is added next to changing the first "the" the "thee". */ | |
| 6715 pgood = goodword + STRLEN(goodword); | |
| 625 | 6716 pbad = su->su_badptr + badlenarg; |
| 6717 for (;;) | |
| 6718 { | |
| 835 | 6719 goodlen = (int)(pgood - goodword); |
| 6720 badlen = (int)(pbad - su->su_badptr); | |
| 625 | 6721 if (goodlen <= 0 || badlen <= 0) |
| 6722 break; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
6723 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
|
6724 MB_PTR_BACK(su->su_badptr, pbad); |
| 534 | 6725 #ifdef FEAT_MBYTE |
| 6726 if (has_mbyte) | |
| 6727 { | |
| 6728 if (mb_ptr2char(pgood) != mb_ptr2char(pbad)) | |
| 6729 break; | |
| 346 | 6730 } |
| 6731 else | |
| 534 | 6732 #endif |
| 6733 if (*pgood != *pbad) | |
| 6734 break; | |
| 625 | 6735 } |
| 6736 | |
| 534 | 6737 if (badlen == 0 && goodlen == 0) |
| 6738 /* goodword doesn't change anything; may happen for "the the" changing | |
| 6739 * the first "the" to itself. */ | |
| 6740 return; | |
| 346 | 6741 |
| 944 | 6742 if (gap->ga_len == 0) |
| 6743 i = -1; | |
| 6744 else | |
| 6745 { | |
| 6746 /* Check if the word is already there. Also check the length that is | |
| 6747 * being replaced "thes," -> "these" is a different suggestion from | |
| 6748 * "thes" -> "these". */ | |
| 6749 stp = &SUG(*gap, 0); | |
| 6750 for (i = gap->ga_len; --i >= 0; ++stp) | |
| 6751 if (stp->st_wordlen == goodlen | |
| 6752 && stp->st_orglen == badlen | |
| 6753 && STRNCMP(stp->st_word, goodword, goodlen) == 0) | |
| 6754 { | |
| 6755 /* | |
| 6756 * Found it. Remember the word with the lowest score. | |
| 6757 */ | |
| 6758 if (stp->st_slang == NULL) | |
| 6759 stp->st_slang = slang; | |
| 6760 | |
| 6761 new_sug.st_score = score; | |
| 6762 new_sug.st_altscore = altscore; | |
| 6763 new_sug.st_had_bonus = had_bonus; | |
| 6764 | |
| 6765 if (stp->st_had_bonus != had_bonus) | |
| 6766 { | |
| 6767 /* Only one of the two had the soundalike score computed. | |
| 6768 * Need to do that for the other one now, otherwise the | |
| 6769 * scores can't be compared. This happens because | |
| 6770 * suggest_try_change() doesn't compute the soundalike | |
| 6771 * word to keep it fast, while some special methods set | |
| 6772 * the soundalike score to zero. */ | |
| 6773 if (had_bonus) | |
| 6774 rescore_one(su, stp); | |
| 6775 else | |
| 6776 { | |
| 6777 new_sug.st_word = stp->st_word; | |
| 6778 new_sug.st_wordlen = stp->st_wordlen; | |
| 6779 new_sug.st_slang = stp->st_slang; | |
| 6780 new_sug.st_orglen = badlen; | |
| 6781 rescore_one(su, &new_sug); | |
| 6782 } | |
| 6783 } | |
| 6784 | |
| 6785 if (stp->st_score > new_sug.st_score) | |
| 6786 { | |
| 6787 stp->st_score = new_sug.st_score; | |
| 6788 stp->st_altscore = new_sug.st_altscore; | |
| 6789 stp->st_had_bonus = new_sug.st_had_bonus; | |
| 6790 } | |
| 6791 break; | |
| 6792 } | |
| 6793 } | |
| 625 | 6794 |
| 6795 if (i < 0 && ga_grow(gap, 1) == OK) | |
| 6796 { | |
| 6797 /* Add a suggestion. */ | |
| 6798 stp = &SUG(*gap, gap->ga_len); | |
| 6799 stp->st_word = vim_strnsave(goodword, goodlen); | |
| 6800 if (stp->st_word != NULL) | |
| 6801 { | |
| 6802 stp->st_wordlen = goodlen; | |
| 6803 stp->st_score = score; | |
| 6804 stp->st_altscore = altscore; | |
| 6805 stp->st_had_bonus = had_bonus; | |
| 6806 stp->st_orglen = badlen; | |
| 6807 stp->st_slang = slang; | |
| 6808 ++gap->ga_len; | |
| 6809 | |
| 6810 /* If we have too many suggestions now, sort the list and keep | |
| 6811 * the best suggestions. */ | |
| 6812 if (gap->ga_len > SUG_MAX_COUNT(su)) | |
| 6813 { | |
| 6814 if (maxsf) | |
| 6815 su->su_sfmaxscore = cleanup_suggestions(gap, | |
| 6816 su->su_sfmaxscore, SUG_CLEAN_COUNT(su)); | |
| 6817 else | |
| 6818 su->su_maxscore = cleanup_suggestions(gap, | |
| 6819 su->su_maxscore, SUG_CLEAN_COUNT(su)); | |
| 6820 } | |
| 6821 } | |
| 6822 } | |
| 6823 } | |
| 6824 | |
| 6825 /* | |
| 6826 * Suggestions may in fact be flagged as errors. Esp. for banned words and | |
| 6827 * for split words, such as "the the". Remove these from the list here. | |
| 6828 */ | |
| 6829 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6830 check_suggestions( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6831 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6832 garray_T *gap) /* either su_ga or su_sga */ |
| 625 | 6833 { |
| 6834 suggest_T *stp; | |
| 6835 int i; | |
| 6836 char_u longword[MAXWLEN + 1]; | |
| 6837 int len; | |
| 6838 hlf_T attr; | |
| 6839 | |
| 6840 stp = &SUG(*gap, 0); | |
| 6841 for (i = gap->ga_len - 1; i >= 0; --i) | |
| 6842 { | |
| 6843 /* Need to append what follows to check for "the the". */ | |
| 2768 | 6844 vim_strncpy(longword, stp[i].st_word, MAXWLEN); |
| 625 | 6845 len = stp[i].st_wordlen; |
| 6846 vim_strncpy(longword + len, su->su_badptr + stp[i].st_orglen, | |
| 6847 MAXWLEN - len); | |
| 6848 attr = HLF_COUNT; | |
| 6849 (void)spell_check(curwin, longword, &attr, NULL, FALSE); | |
| 6850 if (attr != HLF_COUNT) | |
| 6851 { | |
| 6852 /* Remove this entry. */ | |
| 6853 vim_free(stp[i].st_word); | |
| 6854 --gap->ga_len; | |
| 6855 if (i < gap->ga_len) | |
| 6856 mch_memmove(stp + i, stp + i + 1, | |
| 6857 sizeof(suggest_T) * (gap->ga_len - i)); | |
| 6858 } | |
| 6859 } | |
| 6860 } | |
| 6861 | |
| 323 | 6862 |
| 6863 /* | |
| 6864 * Add a word to be banned. | |
| 6865 */ | |
| 6866 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6867 add_banned( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6868 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6869 char_u *word) |
| 323 | 6870 { |
| 626 | 6871 char_u *s; |
| 323 | 6872 hash_T hash; |
| 6873 hashitem_T *hi; | |
| 6874 | |
| 625 | 6875 hash = hash_hash(word); |
| 6876 hi = hash_lookup(&su->su_banned, word, hash); | |
| 6877 if (HASHITEM_EMPTY(hi)) | |
| 6878 { | |
| 6879 s = vim_strsave(word); | |
| 6880 if (s != NULL) | |
| 323 | 6881 hash_add_item(&su->su_banned, hi, s, hash); |
| 625 | 6882 } |
| 323 | 6883 } |
| 6884 | |
| 324 | 6885 /* |
| 534 | 6886 * Recompute the score for all suggestions if sound-folding is possible. This |
| 6887 * is slow, thus only done for the final results. | |
| 324 | 6888 */ |
| 6889 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6890 rescore_suggestions(suginfo_T *su) |
| 324 | 6891 { |
| 6892 int i; | |
| 534 | 6893 |
| 6894 if (su->su_sallang != NULL) | |
| 503 | 6895 for (i = 0; i < su->su_ga.ga_len; ++i) |
| 534 | 6896 rescore_one(su, &SUG(su->su_ga, i)); |
| 6897 } | |
| 6898 | |
| 6899 /* | |
| 6900 * Recompute the score for one suggestion if sound-folding is possible. | |
| 6901 */ | |
| 6902 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6903 rescore_one(suginfo_T *su, suggest_T *stp) |
| 534 | 6904 { |
| 6905 slang_T *slang = stp->st_slang; | |
| 6906 char_u sal_badword[MAXWLEN]; | |
| 535 | 6907 char_u *p; |
| 534 | 6908 |
| 6909 /* Only rescore suggestions that have no sal score yet and do have a | |
| 6910 * language. */ | |
| 6911 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus) | |
| 6912 { | |
| 6913 if (slang == su->su_sallang) | |
| 535 | 6914 p = su->su_sal_badword; |
| 534 | 6915 else |
| 6916 { | |
| 6917 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword); | |
| 535 | 6918 p = sal_badword; |
| 6919 } | |
| 6920 | |
| 6921 stp->st_altscore = stp_sal_score(stp, su, slang, p); | |
| 534 | 6922 if (stp->st_altscore == SCORE_MAXMAX) |
| 6923 stp->st_altscore = SCORE_BIG; | |
| 6924 stp->st_score = RESCORE(stp->st_score, stp->st_altscore); | |
| 6925 stp->st_had_bonus = TRUE; | |
| 324 | 6926 } |
| 6927 } | |
| 6928 | |
| 323 | 6929 static int |
| 6930 #ifdef __BORLANDC__ | |
| 6931 _RTLENTRYF | |
| 6932 #endif | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
6933 sug_compare(const void *s1, const void *s2); |
| 323 | 6934 |
| 6935 /* | |
| 6936 * Function given to qsort() to sort the suggestions on st_score. | |
| 529 | 6937 * First on "st_score", then "st_altscore" then alphabetically. |
| 323 | 6938 */ |
| 6939 static int | |
| 6940 #ifdef __BORLANDC__ | |
| 6941 _RTLENTRYF | |
| 6942 #endif | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6943 sug_compare(const void *s1, const void *s2) |
| 323 | 6944 { |
| 6945 suggest_T *p1 = (suggest_T *)s1; | |
| 6946 suggest_T *p2 = (suggest_T *)s2; | |
| 344 | 6947 int n = p1->st_score - p2->st_score; |
| 6948 | |
| 6949 if (n == 0) | |
| 529 | 6950 { |
| 6951 n = p1->st_altscore - p2->st_altscore; | |
| 6952 if (n == 0) | |
| 6953 n = STRICMP(p1->st_word, p2->st_word); | |
| 6954 } | |
| 344 | 6955 return n; |
| 323 | 6956 } |
| 6957 | |
| 6958 /* | |
| 6959 * Cleanup the suggestions: | |
| 6960 * - Sort on score. | |
| 6961 * - Remove words that won't be displayed. | |
| 344 | 6962 * Returns the maximum score in the list or "maxscore" unmodified. |
| 323 | 6963 */ |
| 344 | 6964 static int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6965 cleanup_suggestions( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6966 garray_T *gap, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6967 int maxscore, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6968 int keep) /* nr of suggestions to keep */ |
| 323 | 6969 { |
| 344 | 6970 suggest_T *stp = &SUG(*gap, 0); |
| 323 | 6971 int i; |
| 6972 | |
| 6973 /* Sort the list. */ | |
| 344 | 6974 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare); |
| 323 | 6975 |
| 6976 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 344 | 6977 if (gap->ga_len > keep) |
| 323 | 6978 { |
| 344 | 6979 for (i = keep; i < gap->ga_len; ++i) |
| 323 | 6980 vim_free(stp[i].st_word); |
| 344 | 6981 gap->ga_len = keep; |
| 6982 return stp[keep - 1].st_score; | |
| 323 | 6983 } |
| 344 | 6984 return maxscore; |
| 323 | 6985 } |
| 6986 | |
| 372 | 6987 #if defined(FEAT_EVAL) || defined(PROTO) |
| 6988 /* | |
| 6989 * Soundfold a string, for soundfold(). | |
| 6990 * Result is in allocated memory, NULL for an error. | |
| 6991 */ | |
| 6992 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6993 eval_soundfold(char_u *word) |
| 372 | 6994 { |
| 6995 langp_T *lp; | |
| 6996 char_u sound[MAXWLEN]; | |
| 500 | 6997 int lpi; |
| 372 | 6998 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6999 if (curwin->w_p_spell && *curwin->w_s->b_p_spl != NUL) |
| 372 | 7000 /* 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
|
7001 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
|
7002 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7003 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 372 | 7004 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 7005 { | |
| 7006 /* soundfold the word */ | |
| 375 | 7007 spell_soundfold(lp->lp_slang, word, FALSE, sound); |
| 372 | 7008 return vim_strsave(sound); |
| 7009 } | |
| 500 | 7010 } |
| 372 | 7011 |
| 7012 /* No language with sound folding, return word as-is. */ | |
| 7013 return vim_strsave(word); | |
| 7014 } | |
| 7015 #endif | |
| 7016 | |
| 323 | 7017 /* |
| 7018 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 485 | 7019 * |
| 7020 * There are many ways to turn a word into a sound-a-like representation. The | |
| 7021 * oldest is Soundex (1918!). A nice overview can be found in "Approximate | |
| 7022 * swedish name matching - survey and test of different algorithms" by Klas | |
| 7023 * Erikson. | |
| 7024 * | |
| 7025 * We support two methods: | |
| 7026 * 1. SOFOFROM/SOFOTO do a simple character mapping. | |
| 7027 * 2. SAL items define a more advanced sound-folding (and much slower). | |
| 323 | 7028 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
7029 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7030 spell_soundfold( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7031 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7032 char_u *inword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7033 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
|
7034 char_u *res) |
| 375 | 7035 { |
| 7036 char_u fword[MAXWLEN]; | |
| 7037 char_u *word; | |
| 7038 | |
| 7039 if (slang->sl_sofo) | |
| 7040 /* SOFOFROM and SOFOTO used */ | |
| 7041 spell_soundfold_sofo(slang, inword, res); | |
| 7042 else | |
| 7043 { | |
| 7044 /* SAL items used. Requires the word to be case-folded. */ | |
| 7045 if (folded) | |
| 7046 word = inword; | |
| 7047 else | |
| 7048 { | |
| 835 | 7049 (void)spell_casefold(inword, (int)STRLEN(inword), fword, MAXWLEN); |
| 375 | 7050 word = fword; |
| 7051 } | |
| 7052 | |
| 7053 #ifdef FEAT_MBYTE | |
| 7054 if (has_mbyte) | |
| 7055 spell_soundfold_wsal(slang, word, res); | |
| 7056 else | |
| 7057 #endif | |
| 7058 spell_soundfold_sal(slang, word, res); | |
| 7059 } | |
| 7060 } | |
| 7061 | |
| 7062 /* | |
| 7063 * Perform sound folding of "inword" into "res" according to SOFOFROM and | |
| 7064 * SOFOTO lines. | |
| 7065 */ | |
| 7066 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7067 spell_soundfold_sofo(slang_T *slang, char_u *inword, char_u *res) |
| 375 | 7068 { |
| 7069 char_u *s; | |
| 7070 int ri = 0; | |
| 7071 int c; | |
| 7072 | |
| 7073 #ifdef FEAT_MBYTE | |
| 7074 if (has_mbyte) | |
| 7075 { | |
| 7076 int prevc = 0; | |
| 7077 int *ip; | |
| 7078 | |
| 7079 /* The sl_sal_first[] table contains the translation for chars up to | |
| 7080 * 255, sl_sal the rest. */ | |
| 7081 for (s = inword; *s != NUL; ) | |
| 7082 { | |
| 474 | 7083 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
|
7084 if (enc_utf8 ? utf_class(c) == 0 : VIM_ISWHITE(c)) |
| 375 | 7085 c = ' '; |
| 7086 else if (c < 256) | |
| 7087 c = slang->sl_sal_first[c]; | |
| 7088 else | |
| 7089 { | |
| 7090 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff]; | |
| 7091 if (ip == NULL) /* empty list, can't match */ | |
| 7092 c = NUL; | |
| 7093 else | |
| 7094 for (;;) /* find "c" in the list */ | |
| 7095 { | |
| 7096 if (*ip == 0) /* not found */ | |
| 7097 { | |
| 7098 c = NUL; | |
| 7099 break; | |
| 7100 } | |
| 7101 if (*ip == c) /* match! */ | |
| 7102 { | |
| 7103 c = ip[1]; | |
| 7104 break; | |
| 7105 } | |
| 7106 ip += 2; | |
| 7107 } | |
| 7108 } | |
| 7109 | |
| 7110 if (c != NUL && c != prevc) | |
| 7111 { | |
| 7112 ri += mb_char2bytes(c, res + ri); | |
| 7113 if (ri + MB_MAXBYTES > MAXWLEN) | |
| 7114 break; | |
| 7115 prevc = c; | |
| 7116 } | |
| 7117 } | |
| 7118 } | |
| 7119 else | |
| 7120 #endif | |
| 7121 { | |
| 7122 /* The sl_sal_first[] table contains the translation. */ | |
| 7123 for (s = inword; (c = *s) != NUL; ++s) | |
| 7124 { | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7125 if (VIM_ISWHITE(c)) |
| 375 | 7126 c = ' '; |
| 7127 else | |
| 7128 c = slang->sl_sal_first[c]; | |
| 7129 if (c != NUL && (ri == 0 || res[ri - 1] != c)) | |
| 7130 res[ri++] = c; | |
| 7131 } | |
| 7132 } | |
| 7133 | |
| 7134 res[ri] = NUL; | |
| 7135 } | |
| 7136 | |
| 7137 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7138 spell_soundfold_sal(slang_T *slang, char_u *inword, char_u *res) |
| 323 | 7139 { |
| 344 | 7140 salitem_T *smp; |
| 323 | 7141 char_u word[MAXWLEN]; |
| 375 | 7142 char_u *s = inword; |
| 323 | 7143 char_u *t; |
| 344 | 7144 char_u *pf; |
| 323 | 7145 int i, j, z; |
| 344 | 7146 int reslen; |
| 323 | 7147 int n, k = 0; |
| 7148 int z0; | |
| 7149 int k0; | |
| 7150 int n0; | |
| 7151 int c; | |
| 7152 int pri; | |
| 7153 int p0 = -333; | |
| 7154 int c0; | |
| 7155 | |
| 324 | 7156 /* Remove accents, if wanted. We actually remove all non-word characters. |
| 375 | 7157 * But keep white space. We need a copy, the word may be changed here. */ |
| 323 | 7158 if (slang->sl_rem_accents) |
| 7159 { | |
| 7160 t = word; | |
| 375 | 7161 while (*s != NUL) |
| 323 | 7162 { |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7163 if (VIM_ISWHITE(*s)) |
| 344 | 7164 { |
| 7165 *t++ = ' '; | |
| 7166 s = skipwhite(s); | |
| 7167 } | |
| 323 | 7168 else |
| 7169 { | |
| 5477 | 7170 if (spell_iswordp_nmw(s, curwin)) |
| 323 | 7171 *t++ = *s; |
| 7172 ++s; | |
| 7173 } | |
| 7174 } | |
| 7175 *t = NUL; | |
| 7176 } | |
| 7177 else | |
| 2768 | 7178 vim_strncpy(word, s, MAXWLEN - 1); |
| 323 | 7179 |
| 344 | 7180 smp = (salitem_T *)slang->sl_sal.ga_data; |
| 323 | 7181 |
| 7182 /* | |
| 7183 * This comes from Aspell phonet.cpp. Converted from C++ to C. | |
| 324 | 7184 * Changed to keep spaces. |
| 323 | 7185 */ |
| 344 | 7186 i = reslen = z = 0; |
| 323 | 7187 while ((c = word[i]) != NUL) |
| 7188 { | |
| 344 | 7189 /* Start with the first rule that has the character in the word. */ |
| 323 | 7190 n = slang->sl_sal_first[c]; |
| 7191 z0 = 0; | |
| 7192 | |
| 7193 if (n >= 0) | |
| 7194 { | |
| 7195 /* check all rules for the same letter */ | |
| 344 | 7196 for (; (s = smp[n].sm_lead)[0] == c; ++n) |
| 323 | 7197 { |
| 344 | 7198 /* Quickly skip entries that don't match the word. Most |
| 7199 * entries are less then three chars, optimize for that. */ | |
| 7200 k = smp[n].sm_leadlen; | |
| 7201 if (k > 1) | |
| 323 | 7202 { |
| 344 | 7203 if (word[i + 1] != s[1]) |
| 7204 continue; | |
| 7205 if (k > 2) | |
| 7206 { | |
| 7207 for (j = 2; j < k; ++j) | |
| 7208 if (word[i + j] != s[j]) | |
| 7209 break; | |
| 7210 if (j < k) | |
| 7211 continue; | |
| 7212 } | |
| 323 | 7213 } |
| 7214 | |
| 375 | 7215 if ((pf = smp[n].sm_oneof) != NULL) |
| 323 | 7216 { |
| 375 | 7217 /* Check for match with one of the chars in "sm_oneof". */ |
| 344 | 7218 while (*pf != NUL && *pf != word[i + k]) |
| 7219 ++pf; | |
| 7220 if (*pf == NUL) | |
| 7221 continue; | |
| 7222 ++k; | |
| 323 | 7223 } |
| 344 | 7224 s = smp[n].sm_rules; |
| 7225 pri = 5; /* default priority */ | |
| 323 | 7226 |
| 7227 p0 = *s; | |
| 7228 k0 = k; | |
| 7229 while (*s == '-' && k > 1) | |
| 7230 { | |
| 7231 k--; | |
| 7232 s++; | |
| 7233 } | |
| 7234 if (*s == '<') | |
| 7235 s++; | |
| 344 | 7236 if (VIM_ISDIGIT(*s)) |
| 323 | 7237 { |
| 7238 /* determine priority */ | |
| 7239 pri = *s - '0'; | |
| 7240 s++; | |
| 7241 } | |
| 7242 if (*s == '^' && *(s + 1) == '^') | |
| 7243 s++; | |
| 7244 | |
| 7245 if (*s == NUL | |
| 7246 || (*s == '^' | |
| 324 | 7247 && (i == 0 || !(word[i - 1] == ' ' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7248 || spell_iswordp(word + i - 1, curwin))) |
| 323 | 7249 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7250 || (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 7251 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7252 && spell_iswordp(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7253 && (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 7254 { |
| 7255 /* search for followup rules, if: */ | |
| 7256 /* followup and k > 1 and NO '-' in searchstring */ | |
| 7257 c0 = word[i + k - 1]; | |
| 7258 n0 = slang->sl_sal_first[c0]; | |
| 7259 | |
| 7260 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 344 | 7261 && p0 != '-' && word[i + k] != NUL) |
| 323 | 7262 { |
| 7263 /* test follow-up rule for "word[i + k]" */ | |
| 344 | 7264 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0) |
| 323 | 7265 { |
| 344 | 7266 /* Quickly skip entries that don't match the word. |
| 7267 * */ | |
| 7268 k0 = smp[n0].sm_leadlen; | |
| 7269 if (k0 > 1) | |
| 323 | 7270 { |
| 344 | 7271 if (word[i + k] != s[1]) |
| 7272 continue; | |
| 7273 if (k0 > 2) | |
| 7274 { | |
| 7275 pf = word + i + k + 1; | |
| 7276 for (j = 2; j < k0; ++j) | |
| 7277 if (*pf++ != s[j]) | |
| 7278 break; | |
| 7279 if (j < k0) | |
| 7280 continue; | |
| 7281 } | |
| 323 | 7282 } |
| 344 | 7283 k0 += k - 1; |
| 7284 | |
| 375 | 7285 if ((pf = smp[n0].sm_oneof) != NULL) |
| 323 | 7286 { |
| 344 | 7287 /* Check for match with one of the chars in |
| 375 | 7288 * "sm_oneof". */ |
| 344 | 7289 while (*pf != NUL && *pf != word[i + k0]) |
| 7290 ++pf; | |
| 7291 if (*pf == NUL) | |
| 7292 continue; | |
| 7293 ++k0; | |
| 323 | 7294 } |
| 344 | 7295 |
| 7296 p0 = 5; | |
| 7297 s = smp[n0].sm_rules; | |
| 323 | 7298 while (*s == '-') |
| 7299 { | |
| 344 | 7300 /* "k0" gets NOT reduced because |
| 7301 * "if (k0 == k)" */ | |
| 323 | 7302 s++; |
| 7303 } | |
| 7304 if (*s == '<') | |
| 7305 s++; | |
| 344 | 7306 if (VIM_ISDIGIT(*s)) |
| 323 | 7307 { |
| 7308 p0 = *s - '0'; | |
| 7309 s++; | |
| 7310 } | |
| 7311 | |
| 7312 if (*s == NUL | |
| 7313 /* *s == '^' cuts */ | |
| 7314 || (*s == '$' | |
| 376 | 7315 && !spell_iswordp(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7316 curwin))) |
| 323 | 7317 { |
| 7318 if (k0 == k) | |
| 7319 /* this is just a piece of the string */ | |
| 7320 continue; | |
| 7321 | |
| 7322 if (p0 < pri) | |
| 7323 /* priority too low */ | |
| 7324 continue; | |
| 7325 /* rule fits; stop search */ | |
| 7326 break; | |
| 7327 } | |
| 7328 } | |
| 7329 | |
| 344 | 7330 if (p0 >= pri && smp[n0].sm_lead[0] == c0) |
| 323 | 7331 continue; |
| 7332 } | |
| 7333 | |
| 7334 /* replace string */ | |
| 344 | 7335 s = smp[n].sm_to; |
| 389 | 7336 if (s == NULL) |
| 7337 s = (char_u *)""; | |
| 344 | 7338 pf = smp[n].sm_rules; |
| 7339 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0; | |
| 323 | 7340 if (p0 == 1 && z == 0) |
| 7341 { | |
| 7342 /* rule with '<' is used */ | |
| 344 | 7343 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c |
| 7344 || res[reslen - 1] == *s)) | |
| 7345 reslen--; | |
| 323 | 7346 z0 = 1; |
| 7347 z = 1; | |
| 7348 k0 = 0; | |
| 372 | 7349 while (*s != NUL && word[i + k0] != NUL) |
| 323 | 7350 { |
| 7351 word[i + k0] = *s; | |
| 7352 k0++; | |
| 7353 s++; | |
| 7354 } | |
| 7355 if (k > k0) | |
| 1619 | 7356 STRMOVE(word + i + k0, word + i + k); |
| 323 | 7357 |
| 7358 /* new "actual letter" */ | |
| 7359 c = word[i]; | |
| 7360 } | |
| 7361 else | |
| 7362 { | |
| 7363 /* no '<' rule used */ | |
| 7364 i += k - 1; | |
| 7365 z = 0; | |
| 344 | 7366 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN) |
| 323 | 7367 { |
| 344 | 7368 if (reslen == 0 || res[reslen - 1] != *s) |
| 372 | 7369 res[reslen++] = *s; |
| 323 | 7370 s++; |
| 7371 } | |
| 7372 /* new "actual letter" */ | |
| 7373 c = *s; | |
| 344 | 7374 if (strstr((char *)pf, "^^") != NULL) |
| 323 | 7375 { |
| 7376 if (c != NUL) | |
| 372 | 7377 res[reslen++] = c; |
| 1619 | 7378 STRMOVE(word, word + i + 1); |
| 323 | 7379 i = 0; |
| 7380 z0 = 1; | |
| 7381 } | |
| 7382 } | |
| 7383 break; | |
| 7384 } | |
| 7385 } | |
| 7386 } | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7387 else if (VIM_ISWHITE(c)) |
| 324 | 7388 { |
| 7389 c = ' '; | |
| 7390 k = 1; | |
| 7391 } | |
| 323 | 7392 |
| 7393 if (z0 == 0) | |
| 7394 { | |
| 344 | 7395 if (k && !p0 && reslen < MAXWLEN && c != NUL |
| 7396 && (!slang->sl_collapse || reslen == 0 | |
| 7397 || res[reslen - 1] != c)) | |
| 323 | 7398 /* condense only double letters */ |
| 372 | 7399 res[reslen++] = c; |
| 323 | 7400 |
| 7401 i++; | |
| 7402 z = 0; | |
| 7403 k = 0; | |
| 7404 } | |
| 7405 } | |
| 7406 | |
| 344 | 7407 res[reslen] = NUL; |
| 323 | 7408 } |
| 7409 | |
| 372 | 7410 #ifdef FEAT_MBYTE |
| 7411 /* | |
| 7412 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 7413 * Multi-byte version of spell_soundfold(). | |
| 7414 */ | |
| 7415 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7416 spell_soundfold_wsal(slang_T *slang, char_u *inword, char_u *res) |
| 372 | 7417 { |
| 375 | 7418 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data; |
| 372 | 7419 int word[MAXWLEN]; |
| 7420 int wres[MAXWLEN]; | |
| 7421 int l; | |
| 7422 char_u *s; | |
| 7423 int *ws; | |
| 7424 char_u *t; | |
| 7425 int *pf; | |
| 7426 int i, j, z; | |
| 7427 int reslen; | |
| 7428 int n, k = 0; | |
| 7429 int z0; | |
| 7430 int k0; | |
| 7431 int n0; | |
| 7432 int c; | |
| 7433 int pri; | |
| 7434 int p0 = -333; | |
| 7435 int c0; | |
| 7436 int did_white = FALSE; | |
| 3520 | 7437 int wordlen; |
| 7438 | |
| 372 | 7439 |
| 7440 /* | |
| 7441 * Convert the multi-byte string to a wide-character string. | |
| 7442 * Remove accents, if wanted. We actually remove all non-word characters. | |
| 7443 * But keep white space. | |
| 7444 */ | |
| 3520 | 7445 wordlen = 0; |
| 372 | 7446 for (s = inword; *s != NUL; ) |
| 7447 { | |
| 7448 t = s; | |
| 474 | 7449 c = mb_cptr2char_adv(&s); |
| 372 | 7450 if (slang->sl_rem_accents) |
| 7451 { | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7452 if (enc_utf8 ? utf_class(c) == 0 : VIM_ISWHITE(c)) |
| 372 | 7453 { |
| 7454 if (did_white) | |
| 7455 continue; | |
| 7456 c = ' '; | |
| 7457 did_white = TRUE; | |
| 7458 } | |
| 7459 else | |
| 7460 { | |
| 7461 did_white = FALSE; | |
| 5477 | 7462 if (!spell_iswordp_nmw(t, curwin)) |
| 372 | 7463 continue; |
| 7464 } | |
| 7465 } | |
| 3520 | 7466 word[wordlen++] = c; |
| 7467 } | |
| 7468 word[wordlen] = NUL; | |
| 372 | 7469 |
| 7470 /* | |
| 3520 | 7471 * This algorithm comes from Aspell phonet.cpp. |
| 372 | 7472 * Converted from C++ to C. Added support for multi-byte chars. |
| 7473 * Changed to keep spaces. | |
| 7474 */ | |
| 7475 i = reslen = z = 0; | |
| 7476 while ((c = word[i]) != NUL) | |
| 7477 { | |
| 7478 /* Start with the first rule that has the character in the word. */ | |
| 7479 n = slang->sl_sal_first[c & 0xff]; | |
| 7480 z0 = 0; | |
| 7481 | |
| 7482 if (n >= 0) | |
| 7483 { | |
|
2455
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
7484 /* 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
|
7485 * 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
|
7486 * (c & 0xff) is NUL. */ |
|
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
7487 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
|
7488 && ws[0] != NUL; ++n) |
| 372 | 7489 { |
| 7490 /* Quickly skip entries that don't match the word. Most | |
| 7491 * entries are less then three chars, optimize for that. */ | |
| 375 | 7492 if (c != ws[0]) |
| 7493 continue; | |
| 372 | 7494 k = smp[n].sm_leadlen; |
| 7495 if (k > 1) | |
| 7496 { | |
| 7497 if (word[i + 1] != ws[1]) | |
| 7498 continue; | |
| 7499 if (k > 2) | |
| 7500 { | |
| 7501 for (j = 2; j < k; ++j) | |
| 7502 if (word[i + j] != ws[j]) | |
| 7503 break; | |
| 7504 if (j < k) | |
| 7505 continue; | |
| 7506 } | |
| 7507 } | |
| 7508 | |
| 375 | 7509 if ((pf = smp[n].sm_oneof_w) != NULL) |
| 372 | 7510 { |
| 375 | 7511 /* Check for match with one of the chars in "sm_oneof". */ |
| 372 | 7512 while (*pf != NUL && *pf != word[i + k]) |
| 7513 ++pf; | |
| 7514 if (*pf == NUL) | |
| 7515 continue; | |
| 7516 ++k; | |
| 7517 } | |
| 7518 s = smp[n].sm_rules; | |
| 7519 pri = 5; /* default priority */ | |
| 7520 | |
| 7521 p0 = *s; | |
| 7522 k0 = k; | |
| 7523 while (*s == '-' && k > 1) | |
| 7524 { | |
| 7525 k--; | |
| 7526 s++; | |
| 7527 } | |
| 7528 if (*s == '<') | |
| 7529 s++; | |
| 7530 if (VIM_ISDIGIT(*s)) | |
| 7531 { | |
| 7532 /* determine priority */ | |
| 7533 pri = *s - '0'; | |
| 7534 s++; | |
| 7535 } | |
| 7536 if (*s == '^' && *(s + 1) == '^') | |
| 7537 s++; | |
| 7538 | |
| 7539 if (*s == NUL | |
| 7540 || (*s == '^' | |
| 7541 && (i == 0 || !(word[i - 1] == ' ' | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7542 || spell_iswordp_w(word + i - 1, curwin))) |
| 372 | 7543 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7544 || (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 7545 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7546 && spell_iswordp_w(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7547 && (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 7548 { |
| 7549 /* search for followup rules, if: */ | |
| 7550 /* followup and k > 1 and NO '-' in searchstring */ | |
| 7551 c0 = word[i + k - 1]; | |
| 7552 n0 = slang->sl_sal_first[c0 & 0xff]; | |
| 7553 | |
| 7554 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 7555 && p0 != '-' && word[i + k] != NUL) | |
| 7556 { | |
| 375 | 7557 /* Test follow-up rule for "word[i + k]"; loop over |
| 7558 * all entries with the same index byte. */ | |
| 372 | 7559 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff) |
| 7560 == (c0 & 0xff); ++n0) | |
| 7561 { | |
| 7562 /* Quickly skip entries that don't match the word. | |
| 375 | 7563 */ |
| 7564 if (c0 != ws[0]) | |
| 7565 continue; | |
| 372 | 7566 k0 = smp[n0].sm_leadlen; |
| 7567 if (k0 > 1) | |
| 7568 { | |
| 7569 if (word[i + k] != ws[1]) | |
| 7570 continue; | |
| 7571 if (k0 > 2) | |
| 7572 { | |
| 7573 pf = word + i + k + 1; | |
| 7574 for (j = 2; j < k0; ++j) | |
| 7575 if (*pf++ != ws[j]) | |
| 7576 break; | |
| 7577 if (j < k0) | |
| 7578 continue; | |
| 7579 } | |
| 7580 } | |
| 7581 k0 += k - 1; | |
| 7582 | |
| 375 | 7583 if ((pf = smp[n0].sm_oneof_w) != NULL) |
| 372 | 7584 { |
| 7585 /* Check for match with one of the chars in | |
| 375 | 7586 * "sm_oneof". */ |
| 372 | 7587 while (*pf != NUL && *pf != word[i + k0]) |
| 7588 ++pf; | |
| 7589 if (*pf == NUL) | |
| 7590 continue; | |
| 7591 ++k0; | |
| 7592 } | |
| 7593 | |
| 7594 p0 = 5; | |
| 7595 s = smp[n0].sm_rules; | |
| 7596 while (*s == '-') | |
| 7597 { | |
| 7598 /* "k0" gets NOT reduced because | |
| 7599 * "if (k0 == k)" */ | |
| 7600 s++; | |
| 7601 } | |
| 7602 if (*s == '<') | |
| 7603 s++; | |
| 7604 if (VIM_ISDIGIT(*s)) | |
| 7605 { | |
| 7606 p0 = *s - '0'; | |
| 7607 s++; | |
| 7608 } | |
| 7609 | |
| 7610 if (*s == NUL | |
| 7611 /* *s == '^' cuts */ | |
| 7612 || (*s == '$' | |
| 376 | 7613 && !spell_iswordp_w(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7614 curwin))) |
| 372 | 7615 { |
| 7616 if (k0 == k) | |
| 7617 /* this is just a piece of the string */ | |
| 7618 continue; | |
| 7619 | |
| 7620 if (p0 < pri) | |
| 7621 /* priority too low */ | |
| 7622 continue; | |
| 7623 /* rule fits; stop search */ | |
| 7624 break; | |
| 7625 } | |
| 7626 } | |
| 7627 | |
| 7628 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff) | |
| 7629 == (c0 & 0xff)) | |
| 7630 continue; | |
| 7631 } | |
| 7632 | |
| 7633 /* replace string */ | |
| 7634 ws = smp[n].sm_to_w; | |
| 7635 s = smp[n].sm_rules; | |
| 7636 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0; | |
| 7637 if (p0 == 1 && z == 0) | |
| 7638 { | |
| 7639 /* rule with '<' is used */ | |
| 389 | 7640 if (reslen > 0 && ws != NULL && *ws != NUL |
| 7641 && (wres[reslen - 1] == c | |
| 372 | 7642 || wres[reslen - 1] == *ws)) |
| 7643 reslen--; | |
| 7644 z0 = 1; | |
| 7645 z = 1; | |
| 7646 k0 = 0; | |
| 389 | 7647 if (ws != NULL) |
| 7648 while (*ws != NUL && word[i + k0] != NUL) | |
| 7649 { | |
| 7650 word[i + k0] = *ws; | |
| 7651 k0++; | |
| 7652 ws++; | |
| 7653 } | |
| 372 | 7654 if (k > k0) |
| 7655 mch_memmove(word + i + k0, word + i + k, | |
| 3520 | 7656 sizeof(int) * (wordlen - (i + k) + 1)); |
| 372 | 7657 |
| 7658 /* new "actual letter" */ | |
| 7659 c = word[i]; | |
| 7660 } | |
| 7661 else | |
| 7662 { | |
| 7663 /* no '<' rule used */ | |
| 7664 i += k - 1; | |
| 7665 z = 0; | |
| 389 | 7666 if (ws != NULL) |
| 7667 while (*ws != NUL && ws[1] != NUL | |
| 7668 && reslen < MAXWLEN) | |
| 7669 { | |
| 7670 if (reslen == 0 || wres[reslen - 1] != *ws) | |
| 7671 wres[reslen++] = *ws; | |
| 7672 ws++; | |
| 7673 } | |
| 372 | 7674 /* new "actual letter" */ |
| 389 | 7675 if (ws == NULL) |
| 7676 c = NUL; | |
| 7677 else | |
| 7678 c = *ws; | |
| 372 | 7679 if (strstr((char *)s, "^^") != NULL) |
| 7680 { | |
| 7681 if (c != NUL) | |
| 7682 wres[reslen++] = c; | |
| 7683 mch_memmove(word, word + i + 1, | |
| 3520 | 7684 sizeof(int) * (wordlen - (i + 1) + 1)); |
| 372 | 7685 i = 0; |
| 7686 z0 = 1; | |
| 7687 } | |
| 7688 } | |
| 7689 break; | |
| 7690 } | |
| 7691 } | |
| 7692 } | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7693 else if (VIM_ISWHITE(c)) |
| 372 | 7694 { |
| 7695 c = ' '; | |
| 7696 k = 1; | |
| 7697 } | |
| 7698 | |
| 7699 if (z0 == 0) | |
| 7700 { | |
| 7701 if (k && !p0 && reslen < MAXWLEN && c != NUL | |
| 7702 && (!slang->sl_collapse || reslen == 0 | |
| 7703 || wres[reslen - 1] != c)) | |
| 7704 /* condense only double letters */ | |
| 7705 wres[reslen++] = c; | |
| 7706 | |
| 7707 i++; | |
| 7708 z = 0; | |
| 7709 k = 0; | |
| 7710 } | |
| 7711 } | |
| 7712 | |
| 7713 /* Convert wide characters in "wres" to a multi-byte string in "res". */ | |
| 7714 l = 0; | |
| 7715 for (n = 0; n < reslen; ++n) | |
| 7716 { | |
| 7717 l += mb_char2bytes(wres[n], res + l); | |
| 7718 if (l + MB_MAXBYTES > MAXWLEN) | |
| 7719 break; | |
| 7720 } | |
| 7721 res[l] = NUL; | |
| 7722 } | |
| 7723 #endif | |
| 7724 | |
| 324 | 7725 /* |
| 344 | 7726 * Compute a score for two sound-a-like words. |
| 7727 * This permits up to two inserts/deletes/swaps/etc. to keep things fast. | |
| 7728 * Instead of a generic loop we write out the code. That keeps it fast by | |
| 7729 * avoiding checks that will not be possible. | |
| 7730 */ | |
| 7731 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7732 soundalike_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7733 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
|
7734 char_u *badstart) /* sound-folded bad word */ |
| 344 | 7735 { |
| 351 | 7736 char_u *goodsound = goodstart; |
| 7737 char_u *badsound = badstart; | |
| 7738 int goodlen; | |
| 7739 int badlen; | |
| 344 | 7740 int n; |
| 7741 char_u *pl, *ps; | |
| 7742 char_u *pl2, *ps2; | |
| 351 | 7743 int score = 0; |
| 7744 | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7745 /* Adding/inserting "*" at the start (word starts with vowel) shouldn't be |
| 351 | 7746 * counted so much, vowels halfway the word aren't counted at all. */ |
| 7747 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound) | |
| 7748 { | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7749 if ((badsound[0] == NUL && goodsound[1] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7750 || (goodsound[0] == NUL && badsound[1] == NUL)) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7751 /* 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
|
7752 return SCORE_DEL; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7753 if (badsound[0] == NUL || goodsound[0] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7754 /* more than two changes */ |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7755 return SCORE_MAXMAX; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7756 |
| 625 | 7757 if (badsound[1] == goodsound[1] |
| 7758 || (badsound[1] != NUL | |
| 7759 && goodsound[1] != NUL | |
| 7760 && badsound[2] == goodsound[2])) | |
| 7761 { | |
| 7762 /* handle like a substitute */ | |
| 7763 } | |
| 351 | 7764 else |
| 625 | 7765 { |
| 7766 score = 2 * SCORE_DEL / 3; | |
| 7767 if (*badsound == '*') | |
| 7768 ++badsound; | |
| 7769 else | |
| 7770 ++goodsound; | |
| 7771 } | |
| 351 | 7772 } |
| 7773 | |
| 835 | 7774 goodlen = (int)STRLEN(goodsound); |
| 7775 badlen = (int)STRLEN(badsound); | |
| 344 | 7776 |
| 1188 | 7777 /* Return quickly if the lengths are too different to be fixed by two |
| 344 | 7778 * changes. */ |
| 7779 n = goodlen - badlen; | |
| 7780 if (n < -2 || n > 2) | |
| 7781 return SCORE_MAXMAX; | |
| 7782 | |
| 7783 if (n > 0) | |
| 7784 { | |
| 351 | 7785 pl = goodsound; /* goodsound is longest */ |
| 344 | 7786 ps = badsound; |
| 7787 } | |
| 7788 else | |
| 7789 { | |
| 351 | 7790 pl = badsound; /* badsound is longest */ |
| 344 | 7791 ps = goodsound; |
| 7792 } | |
| 7793 | |
| 7794 /* Skip over the identical part. */ | |
| 7795 while (*pl == *ps && *pl != NUL) | |
| 7796 { | |
| 7797 ++pl; | |
| 7798 ++ps; | |
| 7799 } | |
| 7800 | |
| 7801 switch (n) | |
| 7802 { | |
| 7803 case -2: | |
| 7804 case 2: | |
| 7805 /* | |
| 7806 * Must delete two characters from "pl". | |
| 7807 */ | |
| 7808 ++pl; /* first delete */ | |
| 7809 while (*pl == *ps) | |
| 7810 { | |
| 7811 ++pl; | |
| 7812 ++ps; | |
| 7813 } | |
| 7814 /* strings must be equal after second delete */ | |
| 7815 if (STRCMP(pl + 1, ps) == 0) | |
| 351 | 7816 return score + SCORE_DEL * 2; |
| 344 | 7817 |
| 7818 /* Failed to compare. */ | |
| 7819 break; | |
| 7820 | |
| 7821 case -1: | |
| 7822 case 1: | |
| 7823 /* | |
| 7824 * Minimal one delete from "pl" required. | |
| 7825 */ | |
| 7826 | |
| 7827 /* 1: delete */ | |
| 7828 pl2 = pl + 1; | |
| 7829 ps2 = ps; | |
| 7830 while (*pl2 == *ps2) | |
| 7831 { | |
| 7832 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7833 return score + SCORE_DEL; |
| 344 | 7834 ++pl2; |
| 7835 ++ps2; | |
| 7836 } | |
| 7837 | |
| 7838 /* 2: delete then swap, then rest must be equal */ | |
| 7839 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7840 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7841 return score + SCORE_DEL + SCORE_SWAP; |
| 344 | 7842 |
| 7843 /* 3: delete then substitute, then the rest must be equal */ | |
| 7844 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7845 return score + SCORE_DEL + SCORE_SUBST; |
| 344 | 7846 |
| 7847 /* 4: first swap then delete */ | |
| 7848 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 7849 { | |
| 7850 pl2 = pl + 2; /* swap, skip two chars */ | |
| 7851 ps2 = ps + 2; | |
| 7852 while (*pl2 == *ps2) | |
| 7853 { | |
| 7854 ++pl2; | |
| 7855 ++ps2; | |
| 7856 } | |
| 7857 /* delete a char and then strings must be equal */ | |
| 7858 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7859 return score + SCORE_SWAP + SCORE_DEL; |
| 344 | 7860 } |
| 7861 | |
| 7862 /* 5: first substitute then delete */ | |
| 7863 pl2 = pl + 1; /* substitute, skip one char */ | |
| 7864 ps2 = ps + 1; | |
| 7865 while (*pl2 == *ps2) | |
| 7866 { | |
| 7867 ++pl2; | |
| 7868 ++ps2; | |
| 7869 } | |
| 7870 /* delete a char and then strings must be equal */ | |
| 7871 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7872 return score + SCORE_SUBST + SCORE_DEL; |
| 344 | 7873 |
| 7874 /* Failed to compare. */ | |
| 7875 break; | |
| 7876 | |
| 7877 case 0: | |
| 7878 /* | |
| 1808 | 7879 * Lengths are equal, thus changes must result in same length: An |
| 344 | 7880 * insert is only possible in combination with a delete. |
| 7881 * 1: check if for identical strings | |
| 7882 */ | |
| 7883 if (*pl == NUL) | |
| 351 | 7884 return score; |
| 344 | 7885 |
| 7886 /* 2: swap */ | |
| 7887 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 7888 { | |
| 7889 pl2 = pl + 2; /* swap, skip two chars */ | |
| 7890 ps2 = ps + 2; | |
| 7891 while (*pl2 == *ps2) | |
| 7892 { | |
| 7893 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7894 return score + SCORE_SWAP; |
| 344 | 7895 ++pl2; |
| 7896 ++ps2; | |
| 7897 } | |
| 7898 /* 3: swap and swap again */ | |
| 7899 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7900 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7901 return score + SCORE_SWAP + SCORE_SWAP; |
| 344 | 7902 |
| 7903 /* 4: swap and substitute */ | |
| 7904 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7905 return score + SCORE_SWAP + SCORE_SUBST; |
| 344 | 7906 } |
| 7907 | |
| 7908 /* 5: substitute */ | |
| 7909 pl2 = pl + 1; | |
| 7910 ps2 = ps + 1; | |
| 7911 while (*pl2 == *ps2) | |
| 7912 { | |
| 7913 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7914 return score + SCORE_SUBST; |
| 344 | 7915 ++pl2; |
| 7916 ++ps2; | |
| 7917 } | |
| 7918 | |
| 7919 /* 6: substitute and swap */ | |
| 7920 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7921 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7922 return score + SCORE_SUBST + SCORE_SWAP; |
| 344 | 7923 |
| 7924 /* 7: substitute and substitute */ | |
| 7925 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7926 return score + SCORE_SUBST + SCORE_SUBST; |
| 344 | 7927 |
| 7928 /* 8: insert then delete */ | |
| 7929 pl2 = pl; | |
| 7930 ps2 = ps + 1; | |
| 7931 while (*pl2 == *ps2) | |
| 7932 { | |
| 7933 ++pl2; | |
| 7934 ++ps2; | |
| 7935 } | |
| 7936 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7937 return score + SCORE_INS + SCORE_DEL; |
| 344 | 7938 |
| 7939 /* 9: delete then insert */ | |
| 7940 pl2 = pl + 1; | |
| 7941 ps2 = ps; | |
| 7942 while (*pl2 == *ps2) | |
| 7943 { | |
| 7944 ++pl2; | |
| 7945 ++ps2; | |
| 7946 } | |
| 7947 if (STRCMP(pl2, ps2 + 1) == 0) | |
| 351 | 7948 return score + SCORE_INS + SCORE_DEL; |
| 344 | 7949 |
| 7950 /* Failed to compare. */ | |
| 7951 break; | |
| 7952 } | |
| 7953 | |
| 7954 return SCORE_MAXMAX; | |
| 7955 } | |
| 324 | 7956 |
| 323 | 7957 /* |
| 7958 * Compute the "edit distance" to turn "badword" into "goodword". The less | |
| 344 | 7959 * deletes/inserts/substitutes/swaps are required the lower the score. |
| 324 | 7960 * |
| 485 | 7961 * The algorithm is described by Du and Chang, 1992. |
| 7962 * The implementation of the algorithm comes from Aspell editdist.cpp, | |
| 7963 * edit_distance(). It has been converted from C++ to C and modified to | |
| 7964 * support multi-byte characters. | |
| 323 | 7965 */ |
| 7966 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7967 spell_edit_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7968 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7969 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7970 char_u *goodword) |
| 323 | 7971 { |
| 7972 int *cnt; | |
| 1188 | 7973 int badlen, goodlen; /* lengths including NUL */ |
| 323 | 7974 int j, i; |
| 7975 int t; | |
| 7976 int bc, gc; | |
| 324 | 7977 int pbc, pgc; |
| 7978 #ifdef FEAT_MBYTE | |
| 7979 char_u *p; | |
| 7980 int wbadword[MAXWLEN]; | |
| 7981 int wgoodword[MAXWLEN]; | |
| 7982 | |
| 7983 if (has_mbyte) | |
| 7984 { | |
| 7985 /* Get the characters from the multi-byte strings and put them in an | |
| 7986 * int array for easy access. */ | |
| 7987 for (p = badword, badlen = 0; *p != NUL; ) | |
| 474 | 7988 wbadword[badlen++] = mb_cptr2char_adv(&p); |
| 398 | 7989 wbadword[badlen++] = 0; |
| 324 | 7990 for (p = goodword, goodlen = 0; *p != NUL; ) |
| 474 | 7991 wgoodword[goodlen++] = mb_cptr2char_adv(&p); |
| 398 | 7992 wgoodword[goodlen++] = 0; |
| 324 | 7993 } |
| 7994 else | |
| 7995 #endif | |
| 7996 { | |
| 835 | 7997 badlen = (int)STRLEN(badword) + 1; |
| 7998 goodlen = (int)STRLEN(goodword) + 1; | |
| 324 | 7999 } |
| 323 | 8000 |
| 8001 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */ | |
| 8002 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)] | |
| 8003 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)), | |
| 8004 TRUE); | |
| 324 | 8005 if (cnt == NULL) |
| 8006 return 0; /* out of memory */ | |
| 323 | 8007 |
| 8008 CNT(0, 0) = 0; | |
| 8009 for (j = 1; j <= goodlen; ++j) | |
| 625 | 8010 CNT(0, j) = CNT(0, j - 1) + SCORE_INS; |
| 323 | 8011 |
| 8012 for (i = 1; i <= badlen; ++i) | |
| 8013 { | |
| 625 | 8014 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL; |
| 323 | 8015 for (j = 1; j <= goodlen; ++j) |
| 8016 { | |
| 324 | 8017 #ifdef FEAT_MBYTE |
| 8018 if (has_mbyte) | |
| 8019 { | |
| 8020 bc = wbadword[i - 1]; | |
| 8021 gc = wgoodword[j - 1]; | |
| 8022 } | |
| 8023 else | |
| 8024 #endif | |
| 8025 { | |
| 8026 bc = badword[i - 1]; | |
| 8027 gc = goodword[j - 1]; | |
| 8028 } | |
| 323 | 8029 if (bc == gc) |
| 8030 CNT(i, j) = CNT(i - 1, j - 1); | |
| 8031 else | |
| 8032 { | |
| 8033 /* Use a better score when there is only a case difference. */ | |
| 324 | 8034 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) |
| 323 | 8035 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1); |
| 8036 else | |
| 625 | 8037 { |
| 8038 /* For a similar character use SCORE_SIMILAR. */ | |
| 8039 if (slang != NULL | |
| 8040 && slang->sl_has_map | |
| 8041 && similar_chars(slang, gc, bc)) | |
| 8042 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1); | |
| 8043 else | |
| 8044 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1); | |
| 8045 } | |
| 323 | 8046 |
| 324 | 8047 if (i > 1 && j > 1) |
| 323 | 8048 { |
| 324 | 8049 #ifdef FEAT_MBYTE |
| 8050 if (has_mbyte) | |
| 8051 { | |
| 8052 pbc = wbadword[i - 2]; | |
| 8053 pgc = wgoodword[j - 2]; | |
| 8054 } | |
| 8055 else | |
| 8056 #endif | |
| 8057 { | |
| 8058 pbc = badword[i - 2]; | |
| 8059 pgc = goodword[j - 2]; | |
| 8060 } | |
| 8061 if (bc == pgc && pbc == gc) | |
| 8062 { | |
| 8063 t = SCORE_SWAP + CNT(i - 2, j - 2); | |
| 8064 if (t < CNT(i, j)) | |
| 8065 CNT(i, j) = t; | |
| 8066 } | |
| 323 | 8067 } |
| 8068 t = SCORE_DEL + CNT(i - 1, j); | |
| 8069 if (t < CNT(i, j)) | |
| 8070 CNT(i, j) = t; | |
| 8071 t = SCORE_INS + CNT(i, j - 1); | |
| 8072 if (t < CNT(i, j)) | |
| 8073 CNT(i, j) = t; | |
| 8074 } | |
| 8075 } | |
| 8076 } | |
| 344 | 8077 |
| 8078 i = CNT(badlen - 1, goodlen - 1); | |
| 8079 vim_free(cnt); | |
| 8080 return i; | |
| 323 | 8081 } |
| 307 | 8082 |
| 625 | 8083 typedef struct |
| 8084 { | |
| 8085 int badi; | |
| 8086 int goodi; | |
| 8087 int score; | |
| 8088 } limitscore_T; | |
| 8089 | |
| 8090 /* | |
| 8091 * Like spell_edit_score(), but with a limit on the score to make it faster. | |
| 8092 * May return SCORE_MAXMAX when the score is higher than "limit". | |
| 8093 * | |
| 8094 * This uses a stack for the edits still to be tried. | |
| 8095 * The idea comes from Aspell leditdist.cpp. Rewritten in C and added support | |
| 8096 * for multi-byte characters. | |
| 8097 */ | |
| 8098 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8099 spell_edit_score_limit( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8100 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8101 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8102 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8103 int limit) |
| 625 | 8104 { |
| 8105 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 8106 int stackidx; | |
| 8107 int bi, gi; | |
| 8108 int bi2, gi2; | |
| 8109 int bc, gc; | |
| 8110 int score; | |
| 8111 int score_off; | |
| 8112 int minscore; | |
| 8113 int round; | |
| 8114 | |
| 8115 #ifdef FEAT_MBYTE | |
| 8116 /* Multi-byte characters require a bit more work, use a different function | |
| 8117 * to avoid testing "has_mbyte" quite often. */ | |
| 8118 if (has_mbyte) | |
| 8119 return spell_edit_score_limit_w(slang, badword, goodword, limit); | |
| 8120 #endif | |
| 8121 | |
| 8122 /* | |
| 8123 * The idea is to go from start to end over the words. So long as | |
| 8124 * characters are equal just continue, this always gives the lowest score. | |
| 8125 * When there is a difference try several alternatives. Each alternative | |
| 8126 * increases "score" for the edit distance. Some of the alternatives are | |
| 8127 * pushed unto a stack and tried later, some are tried right away. At the | |
| 8128 * end of the word the score for one alternative is known. The lowest | |
| 8129 * possible score is stored in "minscore". | |
| 8130 */ | |
| 8131 stackidx = 0; | |
| 8132 bi = 0; | |
| 8133 gi = 0; | |
| 8134 score = 0; | |
| 8135 minscore = limit + 1; | |
| 8136 | |
| 8137 for (;;) | |
| 8138 { | |
| 8139 /* Skip over an equal part, score remains the same. */ | |
| 8140 for (;;) | |
| 8141 { | |
| 8142 bc = badword[bi]; | |
| 8143 gc = goodword[gi]; | |
| 8144 if (bc != gc) /* stop at a char that's different */ | |
| 8145 break; | |
| 8146 if (bc == NUL) /* both words end */ | |
| 8147 { | |
| 8148 if (score < minscore) | |
| 8149 minscore = score; | |
| 8150 goto pop; /* do next alternative */ | |
| 8151 } | |
| 8152 ++bi; | |
| 8153 ++gi; | |
| 8154 } | |
| 8155 | |
| 8156 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 8157 { | |
| 8158 do | |
| 8159 { | |
| 8160 if ((score += SCORE_DEL) >= minscore) | |
| 8161 goto pop; /* do next alternative */ | |
| 8162 } while (badword[++bi] != NUL); | |
| 8163 minscore = score; | |
| 8164 } | |
| 8165 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 8166 { | |
| 8167 do | |
| 8168 { | |
| 8169 if ((score += SCORE_INS) >= minscore) | |
| 8170 goto pop; /* do next alternative */ | |
| 8171 } while (goodword[++gi] != NUL); | |
| 8172 minscore = score; | |
| 8173 } | |
| 8174 else /* both words continue */ | |
| 8175 { | |
| 8176 /* If not close to the limit, perform a change. Only try changes | |
| 8177 * that may lead to a lower score than "minscore". | |
| 8178 * round 0: try deleting a char from badword | |
| 8179 * round 1: try inserting a char in badword */ | |
| 8180 for (round = 0; round <= 1; ++round) | |
| 8181 { | |
| 8182 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 8183 if (score_off < minscore) | |
| 8184 { | |
| 8185 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 8186 { | |
| 8187 /* Near the limit, rest of the words must match. We | |
| 8188 * can check that right now, no need to push an item | |
| 8189 * onto the stack. */ | |
| 8190 bi2 = bi + 1 - round; | |
| 8191 gi2 = gi + round; | |
| 8192 while (goodword[gi2] == badword[bi2]) | |
| 8193 { | |
| 8194 if (goodword[gi2] == NUL) | |
| 8195 { | |
| 8196 minscore = score_off; | |
| 8197 break; | |
| 8198 } | |
| 8199 ++bi2; | |
| 8200 ++gi2; | |
| 8201 } | |
| 8202 } | |
| 8203 else | |
| 8204 { | |
| 8205 /* try deleting/inserting a character later */ | |
| 8206 stack[stackidx].badi = bi + 1 - round; | |
| 8207 stack[stackidx].goodi = gi + round; | |
| 8208 stack[stackidx].score = score_off; | |
| 8209 ++stackidx; | |
| 8210 } | |
| 8211 } | |
| 8212 } | |
| 8213 | |
| 8214 if (score + SCORE_SWAP < minscore) | |
| 8215 { | |
| 8216 /* If swapping two characters makes a match then the | |
| 8217 * substitution is more expensive, thus there is no need to | |
| 8218 * try both. */ | |
| 8219 if (gc == badword[bi + 1] && bc == goodword[gi + 1]) | |
| 8220 { | |
| 8221 /* Swap two characters, that is: skip them. */ | |
| 8222 gi += 2; | |
| 8223 bi += 2; | |
| 8224 score += SCORE_SWAP; | |
| 8225 continue; | |
| 8226 } | |
| 8227 } | |
| 8228 | |
| 8229 /* Substitute one character for another which is the same | |
| 8230 * thing as deleting a character from both goodword and badword. | |
| 8231 * Use a better score when there is only a case difference. */ | |
| 8232 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 8233 score += SCORE_ICASE; | |
| 8234 else | |
| 8235 { | |
| 8236 /* For a similar character use SCORE_SIMILAR. */ | |
| 8237 if (slang != NULL | |
| 8238 && slang->sl_has_map | |
| 8239 && similar_chars(slang, gc, bc)) | |
| 8240 score += SCORE_SIMILAR; | |
| 8241 else | |
| 8242 score += SCORE_SUBST; | |
| 8243 } | |
| 8244 | |
| 8245 if (score < minscore) | |
| 8246 { | |
| 8247 /* Do the substitution. */ | |
| 8248 ++gi; | |
| 8249 ++bi; | |
| 8250 continue; | |
| 8251 } | |
| 8252 } | |
| 8253 pop: | |
| 8254 /* | |
| 8255 * Get here to try the next alternative, pop it from the stack. | |
| 8256 */ | |
| 8257 if (stackidx == 0) /* stack is empty, finished */ | |
| 8258 break; | |
| 8259 | |
| 8260 /* pop an item from the stack */ | |
| 8261 --stackidx; | |
| 8262 gi = stack[stackidx].goodi; | |
| 8263 bi = stack[stackidx].badi; | |
| 8264 score = stack[stackidx].score; | |
| 8265 } | |
| 8266 | |
| 8267 /* When the score goes over "limit" it may actually be much higher. | |
| 8268 * Return a very large number to avoid going below the limit when giving a | |
| 8269 * bonus. */ | |
| 8270 if (minscore > limit) | |
| 8271 return SCORE_MAXMAX; | |
| 8272 return minscore; | |
| 8273 } | |
| 8274 | |
| 8275 #ifdef FEAT_MBYTE | |
| 8276 /* | |
| 8277 * Multi-byte version of spell_edit_score_limit(). | |
| 8278 * Keep it in sync with the above! | |
| 8279 */ | |
| 8280 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8281 spell_edit_score_limit_w( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8282 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8283 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8284 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8285 int limit) |
| 625 | 8286 { |
| 8287 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 8288 int stackidx; | |
| 8289 int bi, gi; | |
| 8290 int bi2, gi2; | |
| 8291 int bc, gc; | |
| 8292 int score; | |
| 8293 int score_off; | |
| 8294 int minscore; | |
| 8295 int round; | |
| 8296 char_u *p; | |
| 8297 int wbadword[MAXWLEN]; | |
| 8298 int wgoodword[MAXWLEN]; | |
| 8299 | |
| 8300 /* Get the characters from the multi-byte strings and put them in an | |
| 8301 * int array for easy access. */ | |
| 8302 bi = 0; | |
| 8303 for (p = badword; *p != NUL; ) | |
| 8304 wbadword[bi++] = mb_cptr2char_adv(&p); | |
| 8305 wbadword[bi++] = 0; | |
| 8306 gi = 0; | |
| 8307 for (p = goodword; *p != NUL; ) | |
| 8308 wgoodword[gi++] = mb_cptr2char_adv(&p); | |
| 8309 wgoodword[gi++] = 0; | |
| 8310 | |
| 8311 /* | |
| 8312 * The idea is to go from start to end over the words. So long as | |
| 8313 * characters are equal just continue, this always gives the lowest score. | |
| 8314 * When there is a difference try several alternatives. Each alternative | |
| 8315 * increases "score" for the edit distance. Some of the alternatives are | |
| 8316 * pushed unto a stack and tried later, some are tried right away. At the | |
| 8317 * end of the word the score for one alternative is known. The lowest | |
| 8318 * possible score is stored in "minscore". | |
| 8319 */ | |
| 8320 stackidx = 0; | |
| 8321 bi = 0; | |
| 8322 gi = 0; | |
| 8323 score = 0; | |
| 8324 minscore = limit + 1; | |
| 8325 | |
| 8326 for (;;) | |
| 8327 { | |
| 8328 /* Skip over an equal part, score remains the same. */ | |
| 8329 for (;;) | |
| 8330 { | |
| 8331 bc = wbadword[bi]; | |
| 8332 gc = wgoodword[gi]; | |
| 8333 | |
| 8334 if (bc != gc) /* stop at a char that's different */ | |
| 8335 break; | |
| 8336 if (bc == NUL) /* both words end */ | |
| 8337 { | |
| 8338 if (score < minscore) | |
| 8339 minscore = score; | |
| 8340 goto pop; /* do next alternative */ | |
| 8341 } | |
| 8342 ++bi; | |
| 8343 ++gi; | |
| 8344 } | |
| 8345 | |
| 8346 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 8347 { | |
| 8348 do | |
| 8349 { | |
| 8350 if ((score += SCORE_DEL) >= minscore) | |
| 8351 goto pop; /* do next alternative */ | |
| 8352 } while (wbadword[++bi] != NUL); | |
| 8353 minscore = score; | |
| 8354 } | |
| 8355 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 8356 { | |
| 8357 do | |
| 8358 { | |
| 8359 if ((score += SCORE_INS) >= minscore) | |
| 8360 goto pop; /* do next alternative */ | |
| 8361 } while (wgoodword[++gi] != NUL); | |
| 8362 minscore = score; | |
| 8363 } | |
| 8364 else /* both words continue */ | |
| 8365 { | |
| 8366 /* If not close to the limit, perform a change. Only try changes | |
| 8367 * that may lead to a lower score than "minscore". | |
| 8368 * round 0: try deleting a char from badword | |
| 8369 * round 1: try inserting a char in badword */ | |
| 8370 for (round = 0; round <= 1; ++round) | |
| 8371 { | |
| 8372 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 8373 if (score_off < minscore) | |
| 8374 { | |
| 8375 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 8376 { | |
| 8377 /* Near the limit, rest of the words must match. We | |
| 8378 * can check that right now, no need to push an item | |
| 8379 * onto the stack. */ | |
| 8380 bi2 = bi + 1 - round; | |
| 8381 gi2 = gi + round; | |
| 8382 while (wgoodword[gi2] == wbadword[bi2]) | |
| 8383 { | |
| 8384 if (wgoodword[gi2] == NUL) | |
| 8385 { | |
| 8386 minscore = score_off; | |
| 8387 break; | |
| 8388 } | |
| 8389 ++bi2; | |
| 8390 ++gi2; | |
| 8391 } | |
| 8392 } | |
| 8393 else | |
| 8394 { | |
| 8395 /* try deleting a character from badword later */ | |
| 8396 stack[stackidx].badi = bi + 1 - round; | |
| 8397 stack[stackidx].goodi = gi + round; | |
| 8398 stack[stackidx].score = score_off; | |
| 8399 ++stackidx; | |
| 8400 } | |
| 8401 } | |
| 8402 } | |
| 8403 | |
| 8404 if (score + SCORE_SWAP < minscore) | |
| 8405 { | |
| 8406 /* If swapping two characters makes a match then the | |
| 8407 * substitution is more expensive, thus there is no need to | |
| 8408 * try both. */ | |
| 8409 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1]) | |
| 8410 { | |
| 8411 /* Swap two characters, that is: skip them. */ | |
| 8412 gi += 2; | |
| 8413 bi += 2; | |
| 8414 score += SCORE_SWAP; | |
| 8415 continue; | |
| 8416 } | |
| 8417 } | |
| 8418 | |
| 8419 /* Substitute one character for another which is the same | |
| 8420 * thing as deleting a character from both goodword and badword. | |
| 8421 * Use a better score when there is only a case difference. */ | |
| 8422 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 8423 score += SCORE_ICASE; | |
| 8424 else | |
| 8425 { | |
| 8426 /* For a similar character use SCORE_SIMILAR. */ | |
| 8427 if (slang != NULL | |
| 8428 && slang->sl_has_map | |
| 8429 && similar_chars(slang, gc, bc)) | |
| 8430 score += SCORE_SIMILAR; | |
| 8431 else | |
| 8432 score += SCORE_SUBST; | |
| 8433 } | |
| 8434 | |
| 8435 if (score < minscore) | |
| 8436 { | |
| 8437 /* Do the substitution. */ | |
| 8438 ++gi; | |
| 8439 ++bi; | |
| 8440 continue; | |
| 8441 } | |
| 8442 } | |
| 8443 pop: | |
| 8444 /* | |
| 8445 * Get here to try the next alternative, pop it from the stack. | |
| 8446 */ | |
| 8447 if (stackidx == 0) /* stack is empty, finished */ | |
| 8448 break; | |
| 8449 | |
| 8450 /* pop an item from the stack */ | |
| 8451 --stackidx; | |
| 8452 gi = stack[stackidx].goodi; | |
| 8453 bi = stack[stackidx].badi; | |
| 8454 score = stack[stackidx].score; | |
| 8455 } | |
| 8456 | |
| 8457 /* When the score goes over "limit" it may actually be much higher. | |
| 8458 * Return a very large number to avoid going below the limit when giving a | |
| 8459 * bonus. */ | |
| 8460 if (minscore > limit) | |
| 8461 return SCORE_MAXMAX; | |
| 8462 return minscore; | |
| 8463 } | |
| 8464 #endif | |
| 8465 | |
| 714 | 8466 /* |
| 8467 * ":spellinfo" | |
| 8468 */ | |
| 8469 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8470 ex_spellinfo(exarg_T *eap UNUSED) |
| 714 | 8471 { |
| 8472 int lpi; | |
| 8473 langp_T *lp; | |
| 8474 char_u *p; | |
| 8475 | |
| 8476 if (no_spell_checking(curwin)) | |
| 8477 return; | |
| 8478 | |
| 8479 msg_start(); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8480 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
|
8481 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8482 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 714 | 8483 msg_puts((char_u *)"file: "); |
| 8484 msg_puts(lp->lp_slang->sl_fname); | |
| 8485 msg_putchar('\n'); | |
| 8486 p = lp->lp_slang->sl_info; | |
| 8487 if (p != NULL) | |
| 8488 { | |
| 8489 msg_puts(p); | |
| 8490 msg_putchar('\n'); | |
| 8491 } | |
| 8492 } | |
| 8493 msg_end(); | |
| 8494 } | |
| 8495 | |
| 625 | 8496 #define DUMPFLAG_KEEPCASE 1 /* round 2: keep-case tree */ |
| 8497 #define DUMPFLAG_COUNT 2 /* include word count */ | |
| 701 | 8498 #define DUMPFLAG_ICASE 4 /* ignore case when finding matches */ |
| 710 | 8499 #define DUMPFLAG_ONECAP 8 /* pattern starts with capital */ |
| 8500 #define DUMPFLAG_ALLCAP 16 /* pattern is all capitals */ | |
| 625 | 8501 |
| 351 | 8502 /* |
| 8503 * ":spelldump" | |
| 8504 */ | |
| 8505 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8506 ex_spelldump(exarg_T *eap) |
| 351 | 8507 { |
| 5382 | 8508 char_u *spl; |
| 8509 long dummy; | |
| 8510 | |
| 701 | 8511 if (no_spell_checking(curwin)) |
| 8512 return; | |
| 5382 | 8513 get_option_value((char_u*)"spl", &dummy, &spl, OPT_LOCAL); |
| 8514 | |
| 8515 /* Create a new empty buffer in a new window. */ | |
| 701 | 8516 do_cmdline_cmd((char_u *)"new"); |
| 5382 | 8517 |
| 8518 /* enable spelling locally in the new window */ | |
| 8519 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
|
8520 set_option_value((char_u*)"spl", dummy, spl, OPT_LOCAL); |
| 5382 | 8521 vim_free(spl); |
| 8522 | |
|
11121
778c10516955
patch 8.0.0448: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
10950
diff
changeset
|
8523 if (!BUFEMPTY()) |
| 701 | 8524 return; |
| 8525 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8526 spell_dump_compl(NULL, 0, NULL, eap->forceit ? DUMPFLAG_COUNT : 0); |
| 701 | 8527 |
| 8528 /* Delete the empty line that we started with. */ | |
| 8529 if (curbuf->b_ml.ml_line_count > 1) | |
| 8530 ml_delete(curbuf->b_ml.ml_line_count, FALSE); | |
| 8531 | |
| 8532 redraw_later(NOT_VALID); | |
| 8533 } | |
| 8534 | |
| 8535 /* | |
| 8536 * Go through all possible words and: | |
| 8537 * 1. When "pat" is NULL: dump a list of all words in the current buffer. | |
| 8538 * "ic" and "dir" are not used. | |
| 8539 * 2. When "pat" is not NULL: add matching words to insert mode completion. | |
| 8540 */ | |
| 8541 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8542 spell_dump_compl( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8543 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
|
8544 int ic, /* ignore case */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8545 int *dir, /* direction for adding matches */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8546 int dumpflags_arg) /* DUMPFLAG_* */ |
| 701 | 8547 { |
| 351 | 8548 langp_T *lp; |
| 8549 slang_T *slang; | |
| 8550 idx_T arridx[MAXWLEN]; | |
| 8551 int curi[MAXWLEN]; | |
| 8552 char_u word[MAXWLEN]; | |
| 8553 int c; | |
| 8554 char_u *byts; | |
| 8555 idx_T *idxs; | |
| 8556 linenr_T lnum = 0; | |
| 8557 int round; | |
| 8558 int depth; | |
| 8559 int n; | |
| 8560 int flags; | |
| 381 | 8561 char_u *region_names = NULL; /* region names being used */ |
| 8562 int do_region = TRUE; /* dump region names and numbers */ | |
| 8563 char_u *p; | |
| 500 | 8564 int lpi; |
| 701 | 8565 int dumpflags = dumpflags_arg; |
| 8566 int patlen; | |
| 8567 | |
| 710 | 8568 /* When ignoring case or when the pattern starts with capital pass this on |
| 8569 * to dump_word(). */ | |
| 8570 if (pat != NULL) | |
| 8571 { | |
| 8572 if (ic) | |
| 8573 dumpflags |= DUMPFLAG_ICASE; | |
| 8574 else | |
| 8575 { | |
| 8576 n = captype(pat, NULL); | |
| 8577 if (n == WF_ONECAP) | |
| 8578 dumpflags |= DUMPFLAG_ONECAP; | |
| 8579 else if (n == WF_ALLCAP | |
| 8580 #ifdef FEAT_MBYTE | |
| 714 | 8581 && (int)STRLEN(pat) > mb_ptr2len(pat) |
| 710 | 8582 #else |
| 714 | 8583 && (int)STRLEN(pat) > 1 |
| 710 | 8584 #endif |
| 8585 ) | |
| 8586 dumpflags |= DUMPFLAG_ALLCAP; | |
| 8587 } | |
| 8588 } | |
| 351 | 8589 |
| 381 | 8590 /* Find out if we can support regions: All languages must support the same |
| 8591 * regions or none at all. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8592 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
|
8593 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8594 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 381 | 8595 p = lp->lp_slang->sl_regions; |
| 8596 if (p[0] != 0) | |
| 8597 { | |
| 8598 if (region_names == NULL) /* first language with regions */ | |
| 8599 region_names = p; | |
| 8600 else if (STRCMP(region_names, p) != 0) | |
| 8601 { | |
| 8602 do_region = FALSE; /* region names are different */ | |
| 8603 break; | |
| 8604 } | |
| 8605 } | |
| 8606 } | |
| 8607 | |
| 8608 if (do_region && region_names != NULL) | |
| 8609 { | |
| 701 | 8610 if (pat == NULL) |
| 8611 { | |
| 8612 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names); | |
| 8613 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 8614 } | |
| 381 | 8615 } |
| 8616 else | |
| 8617 do_region = FALSE; | |
| 8618 | |
| 8619 /* | |
| 8620 * Loop over all files loaded for the entries in 'spelllang'. | |
| 8621 */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8622 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
|
8623 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8624 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 351 | 8625 slang = lp->lp_slang; |
| 500 | 8626 if (slang->sl_fbyts == NULL) /* reloading failed */ |
| 8627 continue; | |
| 351 | 8628 |
| 701 | 8629 if (pat == NULL) |
| 8630 { | |
| 8631 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname); | |
| 8632 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 8633 } | |
| 8634 | |
| 8635 /* When matching with a pattern and there are no prefixes only use | |
| 8636 * parts of the tree that match "pat". */ | |
| 8637 if (pat != NULL && slang->sl_pbyts == NULL) | |
| 835 | 8638 patlen = (int)STRLEN(pat); |
| 701 | 8639 else |
| 840 | 8640 patlen = -1; |
| 351 | 8641 |
| 8642 /* round 1: case-folded tree | |
| 8643 * round 2: keep-case tree */ | |
| 8644 for (round = 1; round <= 2; ++round) | |
| 8645 { | |
| 8646 if (round == 1) | |
| 8647 { | |
| 701 | 8648 dumpflags &= ~DUMPFLAG_KEEPCASE; |
| 351 | 8649 byts = slang->sl_fbyts; |
| 8650 idxs = slang->sl_fidxs; | |
| 8651 } | |
| 8652 else | |
| 8653 { | |
| 701 | 8654 dumpflags |= DUMPFLAG_KEEPCASE; |
| 351 | 8655 byts = slang->sl_kbyts; |
| 8656 idxs = slang->sl_kidxs; | |
| 8657 } | |
| 8658 if (byts == NULL) | |
| 8659 continue; /* array is empty */ | |
| 8660 | |
| 8661 depth = 0; | |
| 8662 arridx[0] = 0; | |
| 8663 curi[0] = 1; | |
| 701 | 8664 while (depth >= 0 && !got_int |
| 8665 && (pat == NULL || !compl_interrupted)) | |
| 351 | 8666 { |
| 8667 if (curi[depth] > byts[arridx[depth]]) | |
| 8668 { | |
| 8669 /* Done all bytes at this node, go up one level. */ | |
| 8670 --depth; | |
| 8671 line_breakcheck(); | |
|
10277
154d5a2e7395
commit https://github.com/vim/vim/commit/472e85970ee3a80abd824bef510df12e9cfe9e96
Christian Brabandt <cb@256bit.org>
parents:
10042
diff
changeset
|
8672 ins_compl_check_keys(50, FALSE); |
| 351 | 8673 } |
| 8674 else | |
| 8675 { | |
| 8676 /* Do one more byte at this node. */ | |
| 8677 n = arridx[depth] + curi[depth]; | |
| 8678 ++curi[depth]; | |
| 8679 c = byts[n]; | |
| 8680 if (c == 0) | |
| 8681 { | |
| 8682 /* End of word, deal with the word. | |
| 8683 * Don't use keep-case words in the fold-case tree, | |
| 8684 * they will appear in the keep-case tree. | |
| 8685 * Only use the word when the region matches. */ | |
| 8686 flags = (int)idxs[n]; | |
| 8687 if ((round == 2 || (flags & WF_KEEPCAP) == 0) | |
| 500 | 8688 && (flags & WF_NEEDCOMP) == 0 |
| 381 | 8689 && (do_region |
| 8690 || (flags & WF_REGION) == 0 | |
| 390 | 8691 || (((unsigned)flags >> 16) |
| 351 | 8692 & lp->lp_region) != 0)) |
| 8693 { | |
| 8694 word[depth] = NUL; | |
| 381 | 8695 if (!do_region) |
| 8696 flags &= ~WF_REGION; | |
| 355 | 8697 |
| 8698 /* Dump the basic word if there is no prefix or | |
| 8699 * when it's the first one. */ | |
| 390 | 8700 c = (unsigned)flags >> 24; |
| 355 | 8701 if (c == 0 || curi[depth] == 2) |
| 701 | 8702 { |
| 8703 dump_word(slang, word, pat, dir, | |
| 8704 dumpflags, flags, lnum); | |
| 8705 if (pat == NULL) | |
| 8706 ++lnum; | |
| 8707 } | |
| 351 | 8708 |
| 8709 /* Apply the prefix, if there is one. */ | |
| 355 | 8710 if (c != 0) |
| 701 | 8711 lnum = dump_prefixes(slang, word, pat, dir, |
| 8712 dumpflags, flags, lnum); | |
| 351 | 8713 } |
| 8714 } | |
| 8715 else | |
| 8716 { | |
| 8717 /* Normal char, go one level deeper. */ | |
| 8718 word[depth++] = c; | |
| 8719 arridx[depth] = idxs[n]; | |
| 8720 curi[depth] = 1; | |
| 701 | 8721 |
| 8722 /* Check if this characters matches with the pattern. | |
| 8723 * If not skip the whole tree below it. | |
| 710 | 8724 * Always ignore case here, dump_word() will check |
| 8725 * proper case later. This isn't exactly right when | |
| 8726 * length changes for multi-byte characters with | |
| 8727 * ignore case... */ | |
| 8728 if (depth <= patlen | |
| 8729 && MB_STRNICMP(word, pat, depth) != 0) | |
| 701 | 8730 --depth; |
| 351 | 8731 } |
| 8732 } | |
| 8733 } | |
| 8734 } | |
| 8735 } | |
| 8736 } | |
| 8737 | |
| 8738 /* | |
| 8739 * Dump one word: apply case modifications and append a line to the buffer. | |
| 701 | 8740 * When "lnum" is zero add insert mode completion. |
| 8741 */ | |
| 8742 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8743 dump_word( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8744 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8745 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8746 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8747 int *dir, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8748 int dumpflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8749 int wordflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8750 linenr_T lnum) |
| 351 | 8751 { |
| 8752 int keepcap = FALSE; | |
| 8753 char_u *p; | |
| 625 | 8754 char_u *tw; |
| 351 | 8755 char_u cword[MAXWLEN]; |
| 381 | 8756 char_u badword[MAXWLEN + 10]; |
| 8757 int i; | |
| 710 | 8758 int flags = wordflags; |
| 8759 | |
| 8760 if (dumpflags & DUMPFLAG_ONECAP) | |
| 8761 flags |= WF_ONECAP; | |
| 8762 if (dumpflags & DUMPFLAG_ALLCAP) | |
| 8763 flags |= WF_ALLCAP; | |
| 351 | 8764 |
| 625 | 8765 if ((dumpflags & DUMPFLAG_KEEPCASE) == 0 && (flags & WF_CAPMASK) != 0) |
| 351 | 8766 { |
| 8767 /* Need to fix case according to "flags". */ | |
| 8768 make_case_word(word, cword, flags); | |
| 8769 p = cword; | |
| 8770 } | |
| 8771 else | |
| 8772 { | |
| 8773 p = word; | |
| 625 | 8774 if ((dumpflags & DUMPFLAG_KEEPCASE) |
| 8775 && ((captype(word, NULL) & WF_KEEPCAP) == 0 | |
| 389 | 8776 || (flags & WF_FIXCAP) != 0)) |
| 351 | 8777 keepcap = TRUE; |
| 8778 } | |
| 625 | 8779 tw = p; |
| 351 | 8780 |
| 701 | 8781 if (pat == NULL) |
| 8782 { | |
| 8783 /* Add flags and regions after a slash. */ | |
| 8784 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap) | |
| 8785 { | |
| 8786 STRCPY(badword, p); | |
| 8787 STRCAT(badword, "/"); | |
| 8788 if (keepcap) | |
| 8789 STRCAT(badword, "="); | |
| 8790 if (flags & WF_BANNED) | |
| 8791 STRCAT(badword, "!"); | |
| 8792 else if (flags & WF_RARE) | |
| 8793 STRCAT(badword, "?"); | |
| 8794 if (flags & WF_REGION) | |
| 8795 for (i = 0; i < 7; ++i) | |
| 8796 if (flags & (0x10000 << i)) | |
| 8797 sprintf((char *)badword + STRLEN(badword), "%d", i + 1); | |
| 8798 p = badword; | |
| 8799 } | |
| 8800 | |
| 8801 if (dumpflags & DUMPFLAG_COUNT) | |
| 8802 { | |
| 8803 hashitem_T *hi; | |
| 8804 | |
| 8805 /* Include the word count for ":spelldump!". */ | |
| 8806 hi = hash_find(&slang->sl_wordcount, tw); | |
| 8807 if (!HASHITEM_EMPTY(hi)) | |
| 8808 { | |
| 8809 vim_snprintf((char *)IObuff, IOSIZE, "%s\t%d", | |
| 625 | 8810 tw, HI2WC(hi)->wc_count); |
| 701 | 8811 p = IObuff; |
| 8812 } | |
| 8813 } | |
| 8814 | |
| 8815 ml_append(lnum, p, (colnr_T)0, FALSE); | |
| 8816 } | |
| 710 | 8817 else if (((dumpflags & DUMPFLAG_ICASE) |
| 8818 ? MB_STRNICMP(p, pat, STRLEN(pat)) == 0 | |
| 8819 : STRNCMP(p, pat, STRLEN(pat)) == 0) | |
| 701 | 8820 && ins_compl_add_infercase(p, (int)STRLEN(p), |
| 942 | 8821 p_ic, NULL, *dir, 0) == OK) |
| 710 | 8822 /* if dir was BACKWARD then honor it just once */ |
| 8823 *dir = FORWARD; | |
| 351 | 8824 } |
| 8825 | |
| 8826 /* | |
| 372 | 8827 * For ":spelldump": Find matching prefixes for "word". Prepend each to |
| 8828 * "word" and append a line to the buffer. | |
| 701 | 8829 * When "lnum" is zero add insert mode completion. |
| 351 | 8830 * Return the updated line number. |
| 8831 */ | |
| 8832 static linenr_T | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8833 dump_prefixes( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8834 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8835 char_u *word, /* case-folded word */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8836 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8837 int *dir, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8838 int dumpflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8839 int flags, /* flags with prefix ID */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8840 linenr_T startlnum) |
| 351 | 8841 { |
| 8842 idx_T arridx[MAXWLEN]; | |
| 8843 int curi[MAXWLEN]; | |
| 8844 char_u prefix[MAXWLEN]; | |
| 455 | 8845 char_u word_up[MAXWLEN]; |
| 8846 int has_word_up = FALSE; | |
| 351 | 8847 int c; |
| 8848 char_u *byts; | |
| 8849 idx_T *idxs; | |
| 8850 linenr_T lnum = startlnum; | |
| 8851 int depth; | |
| 8852 int n; | |
| 8853 int len; | |
| 8854 int i; | |
| 8855 | |
| 626 | 8856 /* If the word starts with a lower-case letter make the word with an |
| 455 | 8857 * upper-case letter in word_up[]. */ |
| 8858 c = PTR2CHAR(word); | |
| 8859 if (SPELL_TOUPPER(c) != c) | |
| 8860 { | |
| 8861 onecap_copy(word, word_up, TRUE); | |
| 8862 has_word_up = TRUE; | |
| 8863 } | |
| 8864 | |
| 351 | 8865 byts = slang->sl_pbyts; |
| 8866 idxs = slang->sl_pidxs; | |
| 8867 if (byts != NULL) /* array not is empty */ | |
| 8868 { | |
| 8869 /* | |
| 8870 * Loop over all prefixes, building them byte-by-byte in prefix[]. | |
| 390 | 8871 * When at the end of a prefix check that it supports "flags". |
| 351 | 8872 */ |
| 8873 depth = 0; | |
| 8874 arridx[0] = 0; | |
| 8875 curi[0] = 1; | |
| 8876 while (depth >= 0 && !got_int) | |
| 8877 { | |
| 390 | 8878 n = arridx[depth]; |
| 8879 len = byts[n]; | |
| 8880 if (curi[depth] > len) | |
| 351 | 8881 { |
| 8882 /* Done all bytes at this node, go up one level. */ | |
| 8883 --depth; | |
| 8884 line_breakcheck(); | |
| 8885 } | |
| 8886 else | |
| 8887 { | |
| 8888 /* Do one more byte at this node. */ | |
| 390 | 8889 n += curi[depth]; |
| 351 | 8890 ++curi[depth]; |
| 8891 c = byts[n]; | |
| 8892 if (c == 0) | |
| 8893 { | |
| 8894 /* End of prefix, find out how many IDs there are. */ | |
| 8895 for (i = 1; i < len; ++i) | |
| 8896 if (byts[n + i] != 0) | |
| 8897 break; | |
| 8898 curi[depth] += i - 1; | |
| 8899 | |
| 455 | 8900 c = valid_word_prefix(i, n, flags, word, slang, FALSE); |
| 8901 if (c != 0) | |
| 351 | 8902 { |
| 376 | 8903 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1); |
| 701 | 8904 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 8905 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 8906 : flags, lnum); |
| 8907 if (lnum != 0) | |
| 8908 ++lnum; | |
| 351 | 8909 } |
| 455 | 8910 |
| 8911 /* Check for prefix that matches the word when the | |
| 8912 * first letter is upper-case, but only if the prefix has | |
| 8913 * a condition. */ | |
| 8914 if (has_word_up) | |
| 8915 { | |
| 8916 c = valid_word_prefix(i, n, flags, word_up, slang, | |
| 8917 TRUE); | |
| 8918 if (c != 0) | |
| 8919 { | |
| 8920 vim_strncpy(prefix + depth, word_up, | |
| 8921 MAXWLEN - depth - 1); | |
| 701 | 8922 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 8923 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 8924 : flags, lnum); |
| 8925 if (lnum != 0) | |
| 8926 ++lnum; | |
| 455 | 8927 } |
| 8928 } | |
| 351 | 8929 } |
| 8930 else | |
| 8931 { | |
| 8932 /* Normal char, go one level deeper. */ | |
| 8933 prefix[depth++] = c; | |
| 8934 arridx[depth] = idxs[n]; | |
| 8935 curi[depth] = 1; | |
| 8936 } | |
| 8937 } | |
| 8938 } | |
| 8939 } | |
| 8940 | |
| 8941 return lnum; | |
| 8942 } | |
| 8943 | |
| 498 | 8944 /* |
| 626 | 8945 * Move "p" to the end of word "start". |
| 8946 * Uses the spell-checking word characters. | |
| 498 | 8947 */ |
| 8948 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8949 spell_to_word_end(char_u *start, win_T *win) |
| 498 | 8950 { |
| 8951 char_u *p = start; | |
| 8952 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8953 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
|
8954 MB_PTR_ADV(p); |
| 498 | 8955 return p; |
| 8956 } | |
| 8957 | |
| 475 | 8958 #if defined(FEAT_INS_EXPAND) || defined(PROTO) |
| 8959 /* | |
| 626 | 8960 * For Insert mode completion CTRL-X s: |
| 8961 * Find start of the word in front of column "startcol". | |
| 8962 * We don't check if it is badly spelled, with completion we can only change | |
| 8963 * the word in front of the cursor. | |
| 475 | 8964 * Returns the column number of the word. |
| 8965 */ | |
| 8966 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8967 spell_word_start(int startcol) |
| 475 | 8968 { |
| 8969 char_u *line; | |
| 8970 char_u *p; | |
| 8971 int col = 0; | |
| 8972 | |
| 498 | 8973 if (no_spell_checking(curwin)) |
| 475 | 8974 return startcol; |
| 8975 | |
| 8976 /* Find a word character before "startcol". */ | |
| 8977 line = ml_get_curline(); | |
| 8978 for (p = line + startcol; p > line; ) | |
| 8979 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
8980 MB_PTR_BACK(line, p); |
| 5477 | 8981 if (spell_iswordp_nmw(p, curwin)) |
| 475 | 8982 break; |
| 8983 } | |
| 8984 | |
| 8985 /* Go back to start of the word. */ | |
| 8986 while (p > line) | |
| 8987 { | |
| 835 | 8988 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
|
8989 MB_PTR_BACK(line, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8990 if (!spell_iswordp(p, curwin)) |
| 475 | 8991 break; |
| 8992 col = 0; | |
| 8993 } | |
| 8994 | |
| 535 | 8995 return col; |
| 8996 } | |
| 8997 | |
| 8998 /* | |
| 8999 * Need to check for 'spellcapcheck' now, the word is removed before | |
| 9000 * expand_spelling() is called. Therefore the ugly global variable. | |
| 9001 */ | |
| 9002 static int spell_expand_need_cap; | |
| 9003 | |
| 9004 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
9005 spell_expand_check_cap(colnr_T col) |
| 535 | 9006 { |
| 475 | 9007 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col); |
| 9008 } | |
| 9009 | |
| 9010 /* | |
| 9011 * Get list of spelling suggestions. | |
| 9012 * Used for Insert mode completion CTRL-X ?. | |
| 9013 * Returns the number of matches. The matches are in "matchp[]", array of | |
| 9014 * allocated strings. | |
| 9015 */ | |
| 9016 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
9017 expand_spelling( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
9018 linenr_T lnum UNUSED, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
9019 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
9020 char_u ***matchp) |
| 475 | 9021 { |
| 9022 garray_T ga; | |
| 9023 | |
| 625 | 9024 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap, TRUE); |
| 475 | 9025 *matchp = ga.ga_data; |
| 9026 return ga.ga_len; | |
| 9027 } | |
| 9028 #endif | |
| 9029 | |
| 737 | 9030 #endif /* FEAT_SPELL */ |