Mercurial > vim
annotate src/spell.c @ 13380:69517d67421f v8.0.1564
patch 8.0.1564: too many #ifdefs
commit https://github.com/vim/vim/commit/f2bd8ef2b4507d02c6043affff8f7e85e3414d5f
Author: Bram Moolenaar <Bram@vim.org>
Date: Sun Mar 4 18:08:14 2018 +0100
patch 8.0.1564: too many #ifdefs
Problem: Too many #ifdefs.
Solution: Graduate the +autocmd feature. Takes away 450 #ifdefs and
increases code size of tiny Vim by only 40 Kbyte.
| author | Christian Brabandt <cb@256bit.org> |
|---|---|
| date | Sun, 04 Mar 2018 18:15:08 +0100 |
| parents | ae312df8d0ab |
| children | 27b9a84395b5 |
| 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 int round; |
| 307 | 1853 |
| 310 | 1854 /* Copy the language name to pass it to spell_load_cb() as a cookie. |
| 307 | 1855 * It's truncated when an error is detected. */ |
| 501 | 1856 STRCPY(sl.sl_lang, lang); |
| 1857 sl.sl_slang = NULL; | |
| 1858 sl.sl_nobreak = FALSE; | |
| 307 | 1859 |
| 649 | 1860 /* We may retry when no spell file is found for the language, an |
| 1861 * autocommand may load it then. */ | |
| 1862 for (round = 1; round <= 2; ++round) | |
| 1863 { | |
| 1864 /* | |
| 1865 * Find the first spell file for "lang" in 'runtimepath' and load it. | |
| 1866 */ | |
| 1867 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 1868 #ifdef VMS |
| 1869 "spell/%s_%s.spl", | |
| 1870 #else | |
| 1871 "spell/%s.%s.spl", | |
| 1872 #endif | |
| 1873 lang, spell_enc()); | |
|
8524
2f57bbe870ea
commit https://github.com/vim/vim/commit/7f8989dd8a627af2185df381195351a913f3777f
Christian Brabandt <cb@256bit.org>
parents:
7835
diff
changeset
|
1874 r = do_in_runtimepath(fname_enc, 0, spell_load_cb, &sl); |
| 649 | 1875 |
| 1876 if (r == FAIL && *sl.sl_lang != NUL) | |
| 1877 { | |
| 1878 /* Try loading the ASCII version. */ | |
| 1879 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 1880 #ifdef VMS |
| 1881 "spell/%s_ascii.spl", | |
| 1882 #else | |
| 1883 "spell/%s.ascii.spl", | |
| 1884 #endif | |
| 1885 lang); | |
|
8524
2f57bbe870ea
commit https://github.com/vim/vim/commit/7f8989dd8a627af2185df381195351a913f3777f
Christian Brabandt <cb@256bit.org>
parents:
7835
diff
changeset
|
1886 r = do_in_runtimepath(fname_enc, 0, spell_load_cb, &sl); |
| 649 | 1887 |
| 1888 if (r == FAIL && *sl.sl_lang != NUL && round == 1 | |
| 1889 && apply_autocmds(EVENT_SPELLFILEMISSING, lang, | |
| 1890 curbuf->b_fname, FALSE, curbuf)) | |
| 1891 continue; | |
| 1892 break; | |
|
13380
69517d67421f
patch 8.0.1564: too many #ifdefs
Christian Brabandt <cb@256bit.org>
parents:
13308
diff
changeset
|
1893 } |
| 714 | 1894 break; |
| 307 | 1895 } |
| 1896 | |
| 1897 if (r == FAIL) | |
| 649 | 1898 { |
| 2660 | 1899 smsg((char_u *) |
| 1900 #ifdef VMS | |
| 1901 _("Warning: Cannot find word list \"%s_%s.spl\" or \"%s_ascii.spl\""), | |
| 1902 #else | |
| 1903 _("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""), | |
| 1904 #endif | |
| 483 | 1905 lang, spell_enc(), lang); |
| 649 | 1906 } |
| 501 | 1907 else if (sl.sl_slang != NULL) |
| 1908 { | |
| 626 | 1909 /* At least one file was loaded, now load ALL the additions. */ |
| 310 | 1910 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
|
1911 do_in_runtimepath(fname_enc, DIP_ALL, spell_load_cb, &sl); |
| 310 | 1912 } |
| 1913 } | |
| 1914 | |
| 1915 /* | |
| 1916 * Return the encoding used for spell checking: Use 'encoding', except that we | |
| 1917 * use "latin1" for "latin9". And limit to 60 characters (just in case). | |
| 1918 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1919 char_u * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1920 spell_enc(void) |
| 310 | 1921 { |
| 1922 | |
| 1923 #ifdef FEAT_MBYTE | |
| 1924 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0) | |
| 1925 return p_enc; | |
| 1926 #endif | |
| 1927 return (char_u *)"latin1"; | |
| 236 | 1928 } |
| 1929 | |
| 1930 /* | |
| 385 | 1931 * Get the name of the .spl file for the internal wordlist into |
| 1932 * "fname[MAXPATHL]". | |
| 1933 */ | |
| 1934 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1935 int_wordlist_spl(char_u *fname) |
| 385 | 1936 { |
| 2660 | 1937 vim_snprintf((char *)fname, MAXPATHL, SPL_FNAME_TMPL, |
| 385 | 1938 int_wordlist, spell_enc()); |
| 1939 } | |
| 1940 | |
| 1941 /* | |
| 625 | 1942 * Allocate a new slang_T for language "lang". "lang" can be NULL. |
| 236 | 1943 * Caller must fill "sl_next". |
| 1944 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1945 slang_T * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1946 slang_alloc(char_u *lang) |
| 236 | 1947 { |
| 1948 slang_T *lp; | |
| 1949 | |
| 300 | 1950 lp = (slang_T *)alloc_clear(sizeof(slang_T)); |
| 236 | 1951 if (lp != NULL) |
| 1952 { | |
| 625 | 1953 if (lang != NULL) |
| 1954 lp->sl_name = vim_strsave(lang); | |
| 323 | 1955 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10); |
| 625 | 1956 ga_init2(&lp->sl_repsal, sizeof(fromto_T), 10); |
| 483 | 1957 lp->sl_compmax = MAXWLEN; |
| 1958 lp->sl_compsylmax = MAXWLEN; | |
| 625 | 1959 hash_init(&lp->sl_wordcount); |
| 1960 } | |
| 1961 | |
| 236 | 1962 return lp; |
| 1963 } | |
| 1964 | |
| 1965 /* | |
| 1966 * Free the contents of an slang_T and the structure itself. | |
| 1967 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1968 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1969 slang_free(slang_T *lp) |
| 236 | 1970 { |
| 1971 vim_free(lp->sl_name); | |
| 310 | 1972 vim_free(lp->sl_fname); |
| 1973 slang_clear(lp); | |
| 1974 vim_free(lp); | |
| 1975 } | |
| 1976 | |
| 1977 /* | |
| 1978 * Clear an slang_T so that the file can be reloaded. | |
| 1979 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1980 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1981 slang_clear(slang_T *lp) |
| 310 | 1982 { |
| 339 | 1983 garray_T *gap; |
| 1984 fromto_T *ftp; | |
| 344 | 1985 salitem_T *smp; |
| 339 | 1986 int i; |
| 625 | 1987 int round; |
| 323 | 1988 |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1989 VIM_CLEAR(lp->sl_fbyts); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1990 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
|
1991 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
|
1992 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1993 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
|
1994 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
|
1995 VIM_CLEAR(lp->sl_pidxs); |
| 323 | 1996 |
| 625 | 1997 for (round = 1; round <= 2; ++round) |
| 1998 { | |
| 1999 gap = round == 1 ? &lp->sl_rep : &lp->sl_repsal; | |
| 2000 while (gap->ga_len > 0) | |
| 2001 { | |
| 2002 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len]; | |
| 2003 vim_free(ftp->ft_from); | |
| 2004 vim_free(ftp->ft_to); | |
| 2005 } | |
| 2006 ga_clear(gap); | |
| 2007 } | |
| 344 | 2008 |
| 2009 gap = &lp->sl_sal; | |
| 375 | 2010 if (lp->sl_sofo) |
| 376 | 2011 { |
| 2012 /* "ga_len" is set to 1 without adding an item for latin1 */ | |
| 2013 if (gap->ga_data != NULL) | |
| 2014 /* SOFOFROM and SOFOTO items: free lists of wide characters. */ | |
| 2015 for (i = 0; i < gap->ga_len; ++i) | |
| 2016 vim_free(((int **)gap->ga_data)[i]); | |
| 2017 } | |
| 375 | 2018 else |
| 2019 /* SAL items: free salitem_T items */ | |
| 2020 while (gap->ga_len > 0) | |
| 2021 { | |
| 2022 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len]; | |
| 2023 vim_free(smp->sm_lead); | |
| 2024 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */ | |
| 2025 vim_free(smp->sm_to); | |
| 2026 #ifdef FEAT_MBYTE | |
| 2027 vim_free(smp->sm_lead_w); | |
| 2028 vim_free(smp->sm_oneof_w); | |
| 2029 vim_free(smp->sm_to_w); | |
| 2030 #endif | |
| 2031 } | |
| 344 | 2032 ga_clear(gap); |
| 323 | 2033 |
| 339 | 2034 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
|
2035 vim_regfree(lp->sl_prefprog[i]); |
| 376 | 2036 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
|
2037 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
|
2038 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2039 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
|
2040 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2041 VIM_CLEAR(lp->sl_midword); |
| 339 | 2042 |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4352
diff
changeset
|
2043 vim_regfree(lp->sl_compprog); |
| 483 | 2044 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
|
2045 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
|
2046 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
|
2047 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
|
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_syllable); |
| 483 | 2050 ga_clear(&lp->sl_syl_items); |
| 481 | 2051 |
| 809 | 2052 ga_clear_strings(&lp->sl_comppat); |
| 2053 | |
| 625 | 2054 hash_clear_all(&lp->sl_wordcount, WC_KEY_OFF); |
| 2055 hash_init(&lp->sl_wordcount); | |
| 2056 | |
| 2057 #ifdef FEAT_MBYTE | |
| 2058 hash_clear_all(&lp->sl_map_hash, 0); | |
| 2059 #endif | |
| 2060 | |
| 2061 /* Clear info from .sug file. */ | |
| 2062 slang_clear_sug(lp); | |
| 483 | 2063 |
| 2064 lp->sl_compmax = MAXWLEN; | |
| 501 | 2065 lp->sl_compminlen = 0; |
| 483 | 2066 lp->sl_compsylmax = MAXWLEN; |
| 2067 lp->sl_regions[0] = NUL; | |
| 236 | 2068 } |
| 2069 | |
| 2070 /* | |
| 625 | 2071 * Clear the info from the .sug file in "lp". |
| 2072 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2073 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2074 slang_clear_sug(slang_T *lp) |
| 625 | 2075 { |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2076 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
|
2077 VIM_CLEAR(lp->sl_sidxs); |
| 625 | 2078 close_spellbuf(lp->sl_sugbuf); |
| 2079 lp->sl_sugbuf = NULL; | |
| 2080 lp->sl_sugloaded = FALSE; | |
| 2081 lp->sl_sugtime = 0; | |
| 2082 } | |
| 2083 | |
| 2084 /* | |
| 307 | 2085 * Load one spell file and store the info into a slang_T. |
| 236 | 2086 * Invoked through do_in_runtimepath(). |
| 2087 */ | |
| 2088 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2089 spell_load_cb(char_u *fname, void *cookie) |
| 501 | 2090 { |
| 2091 spelload_T *slp = (spelload_T *)cookie; | |
| 2092 slang_T *slang; | |
| 2093 | |
| 2094 slang = spell_load_file(fname, slp->sl_lang, NULL, FALSE); | |
| 2095 if (slang != NULL) | |
| 2096 { | |
| 2097 /* When a previously loaded file has NOBREAK also use it for the | |
| 2098 * ".add" files. */ | |
| 2099 if (slp->sl_nobreak && slang->sl_add) | |
| 2100 slang->sl_nobreak = TRUE; | |
| 2101 else if (slang->sl_nobreak) | |
| 2102 slp->sl_nobreak = TRUE; | |
| 2103 | |
| 2104 slp->sl_slang = slang; | |
| 2105 } | |
| 310 | 2106 } |
| 2107 | |
| 625 | 2108 |
| 2109 /* | |
| 2110 * Add a word to the hashtable of common words. | |
| 2111 * If it's already there then the counter is increased. | |
| 2112 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2113 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2114 count_common_word( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2115 slang_T *lp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2116 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2117 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
|
2118 int count) /* 1 to count once, 10 to init */ |
| 625 | 2119 { |
| 2120 hash_T hash; | |
| 2121 hashitem_T *hi; | |
| 2122 wordcount_T *wc; | |
| 2123 char_u buf[MAXWLEN]; | |
| 2124 char_u *p; | |
| 2125 | |
| 2126 if (len == -1) | |
| 2127 p = word; | |
| 2128 else | |
| 2129 { | |
| 2130 vim_strncpy(buf, word, len); | |
| 2131 p = buf; | |
| 2132 } | |
| 2133 | |
| 2134 hash = hash_hash(p); | |
| 2135 hi = hash_lookup(&lp->sl_wordcount, p, hash); | |
| 2136 if (HASHITEM_EMPTY(hi)) | |
| 2137 { | |
| 835 | 2138 wc = (wordcount_T *)alloc((unsigned)(sizeof(wordcount_T) + STRLEN(p))); |
| 625 | 2139 if (wc == NULL) |
| 2140 return; | |
| 2141 STRCPY(wc->wc_word, p); | |
| 2142 wc->wc_count = count; | |
| 2143 hash_add_item(&lp->sl_wordcount, hi, wc->wc_word, hash); | |
| 2144 } | |
| 2145 else | |
| 2146 { | |
| 2147 wc = HI2WC(hi); | |
| 2148 if ((wc->wc_count += count) < (unsigned)count) /* check for overflow */ | |
| 2149 wc->wc_count = MAXWORDCOUNT; | |
| 2150 } | |
| 2151 } | |
| 2152 | |
| 2153 /* | |
| 2154 * Adjust the score of common words. | |
| 2155 */ | |
| 2156 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2157 score_wordcount_adj( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2158 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2159 int score, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2160 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2161 int split) /* word was split, less bonus */ |
| 625 | 2162 { |
| 2163 hashitem_T *hi; | |
| 2164 wordcount_T *wc; | |
| 2165 int bonus; | |
| 2166 int newscore; | |
| 2167 | |
| 2168 hi = hash_find(&slang->sl_wordcount, word); | |
| 2169 if (!HASHITEM_EMPTY(hi)) | |
| 2170 { | |
| 2171 wc = HI2WC(hi); | |
| 2172 if (wc->wc_count < SCORE_THRES2) | |
| 2173 bonus = SCORE_COMMON1; | |
| 2174 else if (wc->wc_count < SCORE_THRES3) | |
| 2175 bonus = SCORE_COMMON2; | |
| 2176 else | |
| 2177 bonus = SCORE_COMMON3; | |
| 2178 if (split) | |
| 2179 newscore = score - bonus / 2; | |
| 2180 else | |
| 2181 newscore = score - bonus; | |
| 2182 if (newscore < 0) | |
| 2183 return 0; | |
| 2184 return newscore; | |
| 2185 } | |
| 2186 return score; | |
| 2187 } | |
| 2188 | |
| 483 | 2189 |
| 495 | 2190 /* |
| 498 | 2191 * Return TRUE if byte "n" appears in "str". |
| 495 | 2192 * Like strchr() but independent of locale. |
| 2193 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2194 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2195 byte_in_str(char_u *str, int n) |
| 495 | 2196 { |
| 2197 char_u *p; | |
| 2198 | |
| 2199 for (p = str; *p != NUL; ++p) | |
| 498 | 2200 if (*p == n) |
| 495 | 2201 return TRUE; |
| 2202 return FALSE; | |
| 2203 } | |
| 2204 | |
| 483 | 2205 #define SY_MAXLEN 30 |
| 2206 typedef struct syl_item_S | |
| 2207 { | |
| 2208 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */ | |
| 2209 int sy_len; | |
| 2210 } syl_item_T; | |
| 2211 | |
| 2212 /* | |
| 2213 * Truncate "slang->sl_syllable" at the first slash and put the following items | |
| 2214 * in "slang->sl_syl_items". | |
| 2215 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2216 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2217 init_syl_tab(slang_T *slang) |
| 483 | 2218 { |
| 2219 char_u *p; | |
| 2220 char_u *s; | |
| 2221 int l; | |
| 2222 syl_item_T *syl; | |
| 2223 | |
| 2224 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4); | |
| 2225 p = vim_strchr(slang->sl_syllable, '/'); | |
| 2226 while (p != NULL) | |
| 2227 { | |
| 2228 *p++ = NUL; | |
| 495 | 2229 if (*p == NUL) /* trailing slash */ |
| 483 | 2230 break; |
| 2231 s = p; | |
| 2232 p = vim_strchr(p, '/'); | |
| 2233 if (p == NULL) | |
| 835 | 2234 l = (int)STRLEN(s); |
| 483 | 2235 else |
| 835 | 2236 l = (int)(p - s); |
| 483 | 2237 if (l >= SY_MAXLEN) |
| 2238 return SP_FORMERROR; | |
| 2239 if (ga_grow(&slang->sl_syl_items, 1) == FAIL) | |
| 495 | 2240 return SP_OTHERERROR; |
| 483 | 2241 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) |
| 2242 + slang->sl_syl_items.ga_len++; | |
| 2243 vim_strncpy(syl->sy_chars, s, l); | |
| 2244 syl->sy_len = l; | |
| 2245 } | |
| 2246 return OK; | |
| 2247 } | |
| 2248 | |
| 2249 /* | |
| 2250 * Count the number of syllables in "word". | |
| 2251 * When "word" contains spaces the syllables after the last space are counted. | |
| 2252 * Returns zero if syllables are not defines. | |
| 2253 */ | |
| 2254 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2255 count_syllables(slang_T *slang, char_u *word) |
| 483 | 2256 { |
| 2257 int cnt = 0; | |
| 2258 int skip = FALSE; | |
| 2259 char_u *p; | |
| 2260 int len; | |
| 2261 int i; | |
| 2262 syl_item_T *syl; | |
| 2263 int c; | |
| 2264 | |
| 2265 if (slang->sl_syllable == NULL) | |
| 2266 return 0; | |
| 2267 | |
| 2268 for (p = word; *p != NUL; p += len) | |
| 2269 { | |
| 2270 /* When running into a space reset counter. */ | |
| 2271 if (*p == ' ') | |
| 2272 { | |
| 2273 len = 1; | |
| 2274 cnt = 0; | |
| 2275 continue; | |
| 2276 } | |
| 2277 | |
| 2278 /* Find longest match of syllable items. */ | |
| 2279 len = 0; | |
| 2280 for (i = 0; i < slang->sl_syl_items.ga_len; ++i) | |
| 2281 { | |
| 2282 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i; | |
| 2283 if (syl->sy_len > len | |
| 2284 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0) | |
| 2285 len = syl->sy_len; | |
| 2286 } | |
| 2287 if (len != 0) /* found a match, count syllable */ | |
| 2288 { | |
| 2289 ++cnt; | |
| 2290 skip = FALSE; | |
| 2291 } | |
| 2292 else | |
| 2293 { | |
| 2294 /* No recognized syllable item, at least a syllable char then? */ | |
| 2295 #ifdef FEAT_MBYTE | |
| 2296 c = mb_ptr2char(p); | |
| 2297 len = (*mb_ptr2len)(p); | |
| 2298 #else | |
| 2299 c = *p; | |
| 2300 len = 1; | |
| 2301 #endif | |
| 2302 if (vim_strchr(slang->sl_syllable, c) == NULL) | |
| 2303 skip = FALSE; /* No, search for next syllable */ | |
| 2304 else if (!skip) | |
| 2305 { | |
| 2306 ++cnt; /* Yes, count it */ | |
| 2307 skip = TRUE; /* don't count following syllable chars */ | |
| 2308 } | |
| 2309 } | |
| 2310 } | |
| 2311 return cnt; | |
| 376 | 2312 } |
| 2313 | |
| 381 | 2314 /* |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2315 * Parse 'spelllang' and set w_s->b_langp accordingly. |
| 351 | 2316 * Returns NULL if it's OK, an error message otherwise. |
| 236 | 2317 */ |
| 2318 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2319 did_set_spelllang(win_T *wp) |
| 236 | 2320 { |
| 2321 garray_T ga; | |
| 351 | 2322 char_u *splp; |
| 236 | 2323 char_u *region; |
| 416 | 2324 char_u region_cp[3]; |
| 355 | 2325 int filename; |
| 236 | 2326 int region_mask; |
| 503 | 2327 slang_T *slang; |
| 236 | 2328 int c; |
| 351 | 2329 char_u lang[MAXWLEN + 1]; |
| 323 | 2330 char_u spf_name[MAXPATHL]; |
| 351 | 2331 int len; |
| 2332 char_u *p; | |
| 381 | 2333 int round; |
| 385 | 2334 char_u *spf; |
| 389 | 2335 char_u *use_region = NULL; |
| 2336 int dont_use_region = FALSE; | |
| 501 | 2337 int nobreak = FALSE; |
| 503 | 2338 int i, j; |
| 2339 langp_T *lp, *lp2; | |
| 1185 | 2340 static int recursive = FALSE; |
| 2341 char_u *ret_msg = NULL; | |
| 2342 char_u *spl_copy; | |
|
9487
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2343 bufref_T bufref; |
|
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2344 |
|
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2345 set_bufref(&bufref, wp->w_buffer); |
| 1185 | 2346 |
| 2347 /* We don't want to do this recursively. May happen when a language is | |
| 2348 * not available and the SpellFileMissing autocommand opens a new buffer | |
| 2349 * in which 'spell' is set. */ | |
| 2350 if (recursive) | |
| 2351 return NULL; | |
| 2352 recursive = TRUE; | |
| 236 | 2353 |
| 2354 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
|
2355 clear_midword(wp); |
| 236 | 2356 |
| 4352 | 2357 /* Make a copy of 'spelllang', the SpellFileMissing autocommands may change |
| 1185 | 2358 * it under our fingers. */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2359 spl_copy = vim_strsave(wp->w_s->b_p_spl); |
| 1185 | 2360 if (spl_copy == NULL) |
| 2361 goto theend; | |
| 2362 | |
| 5485 | 2363 #ifdef FEAT_MBYTE |
| 5477 | 2364 wp->w_s->b_cjk = 0; |
| 5485 | 2365 #endif |
| 5477 | 2366 |
| 2367 /* Loop over comma separated language names. */ | |
| 1185 | 2368 for (splp = spl_copy; *splp != NUL; ) |
| 351 | 2369 { |
| 2370 /* Get one language name. */ | |
| 2371 copy_option_part(&splp, lang, MAXWLEN, ","); | |
| 240 | 2372 region = NULL; |
| 835 | 2373 len = (int)STRLEN(lang); |
| 355 | 2374 |
| 5477 | 2375 if (STRCMP(lang, "cjk") == 0) |
| 2376 { | |
| 5485 | 2377 #ifdef FEAT_MBYTE |
| 5477 | 2378 wp->w_s->b_cjk = 1; |
| 5485 | 2379 #endif |
| 5477 | 2380 continue; |
| 2381 } | |
| 2382 | |
| 355 | 2383 /* If the name ends in ".spl" use it as the name of the spell file. |
| 2384 * If there is a region name let "region" point to it and remove it | |
| 2385 * from the name. */ | |
| 2386 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0) | |
| 236 | 2387 { |
| 355 | 2388 filename = TRUE; |
| 2389 | |
| 416 | 2390 /* Locate a region and remove it from the file name. */ |
| 2391 p = vim_strchr(gettail(lang), '_'); | |
| 2392 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2]) | |
| 2393 && !ASCII_ISALPHA(p[3])) | |
| 2394 { | |
| 2395 vim_strncpy(region_cp, p + 1, 2); | |
| 2396 mch_memmove(p, p + 3, len - (p - lang) - 2); | |
| 2397 region = region_cp; | |
| 2398 } | |
| 2399 else | |
| 2400 dont_use_region = TRUE; | |
| 2401 | |
| 355 | 2402 /* Check if we loaded this language before. */ |
| 503 | 2403 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2404 if (fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME) | |
| 355 | 2405 break; |
| 236 | 2406 } |
| 355 | 2407 else |
| 2408 { | |
| 2409 filename = FALSE; | |
| 2410 if (len > 3 && lang[len - 3] == '_') | |
| 2411 { | |
| 2412 region = lang + len - 2; | |
| 2413 len -= 3; | |
| 2414 lang[len] = NUL; | |
| 389 | 2415 } |
| 2416 else | |
| 2417 dont_use_region = TRUE; | |
| 355 | 2418 |
| 2419 /* Check if we loaded this language before. */ | |
| 503 | 2420 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2421 if (STRICMP(lang, slang->sl_name) == 0) | |
| 355 | 2422 break; |
| 2423 } | |
| 236 | 2424 |
| 416 | 2425 if (region != NULL) |
| 2426 { | |
| 2427 /* If the region differs from what was used before then don't | |
| 2428 * use it for 'spellfile'. */ | |
| 2429 if (use_region != NULL && STRCMP(region, use_region) != 0) | |
| 2430 dont_use_region = TRUE; | |
| 2431 use_region = region; | |
| 2432 } | |
| 2433 | |
| 351 | 2434 /* If not found try loading the language now. */ |
| 503 | 2435 if (slang == NULL) |
| 355 | 2436 { |
| 2437 if (filename) | |
| 2438 (void)spell_load_file(lang, lang, NULL, FALSE); | |
| 2439 else | |
| 1185 | 2440 { |
| 355 | 2441 spell_load_lang(lang); |
| 1185 | 2442 /* SpellFileMissing autocommands may do anything, including |
| 2443 * 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
|
2444 if (!bufref_valid(&bufref)) |
| 1185 | 2445 { |
|
9953
9560a2eb7968
commit https://github.com/vim/vim/commit/5b30291785e6b9be1a607504c14bd03c601b59a6
Christian Brabandt <cb@256bit.org>
parents:
9898
diff
changeset
|
2446 ret_msg = (char_u *)N_("E797: SpellFileMissing autocommand deleted buffer"); |
| 1185 | 2447 goto theend; |
| 2448 } | |
| 2449 } | |
| 355 | 2450 } |
| 236 | 2451 |
| 307 | 2452 /* |
| 351 | 2453 * Loop over the languages, there can be several files for "lang". |
| 307 | 2454 */ |
| 503 | 2455 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2456 if (filename ? fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME | |
| 2457 : STRICMP(lang, slang->sl_name) == 0) | |
| 236 | 2458 { |
| 316 | 2459 region_mask = REGION_ALL; |
| 355 | 2460 if (!filename && region != NULL) |
| 236 | 2461 { |
| 307 | 2462 /* find region in sl_regions */ |
| 503 | 2463 c = find_region(slang->sl_regions, region); |
| 307 | 2464 if (c == REGION_ALL) |
| 2465 { | |
| 503 | 2466 if (slang->sl_add) |
| 389 | 2467 { |
| 503 | 2468 if (*slang->sl_regions != NUL) |
| 389 | 2469 /* This addition file is for other regions. */ |
| 2470 region_mask = 0; | |
| 2471 } | |
| 2472 else | |
| 2473 /* This is probably an error. Give a warning and | |
| 2474 * accept the words anyway. */ | |
| 351 | 2475 smsg((char_u *) |
| 2476 _("Warning: region %s not supported"), | |
| 2477 region); | |
| 307 | 2478 } |
| 2479 else | |
| 2480 region_mask = 1 << c; | |
| 236 | 2481 } |
| 307 | 2482 |
| 389 | 2483 if (region_mask != 0) |
| 307 | 2484 { |
| 389 | 2485 if (ga_grow(&ga, 1) == FAIL) |
| 2486 { | |
| 2487 ga_clear(&ga); | |
| 1185 | 2488 ret_msg = e_outofmem; |
| 2489 goto theend; | |
| 389 | 2490 } |
| 503 | 2491 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 389 | 2492 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 2493 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2494 use_midword(slang, wp); |
| 503 | 2495 if (slang->sl_nobreak) |
| 501 | 2496 nobreak = TRUE; |
| 307 | 2497 } |
| 385 | 2498 } |
| 2499 } | |
| 2500 | |
| 2501 /* round 0: load int_wordlist, if possible. | |
| 2502 * round 1: load first name in 'spellfile'. | |
| 2503 * round 2: load second name in 'spellfile. | |
| 2504 * etc. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2505 spf = curwin->w_s->b_p_spf; |
| 385 | 2506 for (round = 0; round == 0 || *spf != NUL; ++round) |
| 2507 { | |
| 2508 if (round == 0) | |
| 2509 { | |
| 2510 /* Internal wordlist, if there is one. */ | |
| 2511 if (int_wordlist == NULL) | |
| 381 | 2512 continue; |
| 385 | 2513 int_wordlist_spl(spf_name); |
| 381 | 2514 } |
| 2515 else | |
| 2516 { | |
| 385 | 2517 /* One entry in 'spellfile'. */ |
| 2518 copy_option_part(&spf, spf_name, MAXPATHL - 5, ","); | |
| 2519 STRCAT(spf_name, ".spl"); | |
| 2520 | |
| 2521 /* If it was already found above then skip it. */ | |
| 2522 for (c = 0; c < ga.ga_len; ++c) | |
| 500 | 2523 { |
| 2524 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname; | |
| 2525 if (p != NULL && fullpathcmp(spf_name, p, FALSE) == FPC_SAME) | |
| 385 | 2526 break; |
| 500 | 2527 } |
| 385 | 2528 if (c < ga.ga_len) |
| 381 | 2529 continue; |
| 2530 } | |
| 2531 | |
| 351 | 2532 /* Check if it was loaded already. */ |
| 503 | 2533 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2534 if (fullpathcmp(spf_name, slang->sl_fname, FALSE) == FPC_SAME) | |
| 323 | 2535 break; |
| 503 | 2536 if (slang == NULL) |
| 323 | 2537 { |
| 351 | 2538 /* Not loaded, try loading it now. The language name includes the |
| 385 | 2539 * region name, the region is ignored otherwise. for int_wordlist |
| 2540 * use an arbitrary name. */ | |
| 2541 if (round == 0) | |
| 2542 STRCPY(lang, "internal wordlist"); | |
| 2543 else | |
| 2544 { | |
| 2545 vim_strncpy(lang, gettail(spf_name), MAXWLEN); | |
| 381 | 2546 p = vim_strchr(lang, '.'); |
| 2547 if (p != NULL) | |
| 2548 *p = NUL; /* truncate at ".encoding.add" */ | |
| 2549 } | |
| 503 | 2550 slang = spell_load_file(spf_name, lang, NULL, TRUE); |
| 501 | 2551 |
| 2552 /* If one of the languages has NOBREAK we assume the addition | |
| 2553 * files also have this. */ | |
| 503 | 2554 if (slang != NULL && nobreak) |
| 2555 slang->sl_nobreak = TRUE; | |
| 2556 } | |
| 2557 if (slang != NULL && ga_grow(&ga, 1) == OK) | |
| 323 | 2558 { |
| 389 | 2559 region_mask = REGION_ALL; |
| 2560 if (use_region != NULL && !dont_use_region) | |
| 2561 { | |
| 2562 /* find region in sl_regions */ | |
| 503 | 2563 c = find_region(slang->sl_regions, use_region); |
| 389 | 2564 if (c != REGION_ALL) |
| 2565 region_mask = 1 << c; | |
| 503 | 2566 else if (*slang->sl_regions != NUL) |
| 389 | 2567 /* This spell file is for other regions. */ |
| 2568 region_mask = 0; | |
| 2569 } | |
| 2570 | |
| 2571 if (region_mask != 0) | |
| 2572 { | |
| 503 | 2573 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 2574 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL; | |
| 2575 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL; | |
| 389 | 2576 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 2577 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2578 use_midword(slang, wp); |
| 389 | 2579 } |
| 323 | 2580 } |
| 2581 } | |
| 2582 | |
| 236 | 2583 /* 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
|
2584 ga_clear(&wp->w_s->b_langp); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2585 wp->w_s->b_langp = ga; |
| 236 | 2586 |
| 503 | 2587 /* For each language figure out what language to use for sound folding and |
| 2588 * REP items. If the language doesn't support it itself use another one | |
| 2589 * with the same name. E.g. for "en-math" use "en". */ | |
| 2590 for (i = 0; i < ga.ga_len; ++i) | |
| 2591 { | |
| 2592 lp = LANGP_ENTRY(ga, i); | |
| 2593 | |
| 2594 /* sound folding */ | |
| 2595 if (lp->lp_slang->sl_sal.ga_len > 0) | |
| 2596 /* language does sound folding itself */ | |
| 2597 lp->lp_sallang = lp->lp_slang; | |
| 2598 else | |
| 2599 /* find first similar language that does sound folding */ | |
| 2600 for (j = 0; j < ga.ga_len; ++j) | |
| 2601 { | |
| 2602 lp2 = LANGP_ENTRY(ga, j); | |
| 2603 if (lp2->lp_slang->sl_sal.ga_len > 0 | |
| 2604 && STRNCMP(lp->lp_slang->sl_name, | |
| 2605 lp2->lp_slang->sl_name, 2) == 0) | |
| 2606 { | |
| 2607 lp->lp_sallang = lp2->lp_slang; | |
| 2608 break; | |
| 2609 } | |
| 2610 } | |
| 2611 | |
| 2612 /* REP items */ | |
| 2613 if (lp->lp_slang->sl_rep.ga_len > 0) | |
| 2614 /* language has REP items itself */ | |
| 2615 lp->lp_replang = lp->lp_slang; | |
| 2616 else | |
| 625 | 2617 /* find first similar language that has REP items */ |
| 503 | 2618 for (j = 0; j < ga.ga_len; ++j) |
| 2619 { | |
| 2620 lp2 = LANGP_ENTRY(ga, j); | |
| 2621 if (lp2->lp_slang->sl_rep.ga_len > 0 | |
| 2622 && STRNCMP(lp->lp_slang->sl_name, | |
| 2623 lp2->lp_slang->sl_name, 2) == 0) | |
| 2624 { | |
| 2625 lp->lp_replang = lp2->lp_slang; | |
| 2626 break; | |
| 2627 } | |
| 2628 } | |
| 2629 } | |
| 2630 | |
| 1185 | 2631 theend: |
| 2632 vim_free(spl_copy); | |
| 2633 recursive = FALSE; | |
| 5891 | 2634 redraw_win_later(wp, NOT_VALID); |
| 1185 | 2635 return ret_msg; |
| 236 | 2636 } |
| 2637 | |
| 2638 /* | |
| 376 | 2639 * Clear the midword characters for buffer "buf". |
| 2640 */ | |
| 2641 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2642 clear_midword(win_T *wp) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2643 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2644 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
|
2645 #ifdef FEAT_MBYTE |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2646 VIM_CLEAR(wp->w_s->b_spell_ismw_mb); |
| 376 | 2647 #endif |
| 2648 } | |
| 2649 | |
| 2650 /* | |
| 2651 * Use the "sl_midword" field of language "lp" for buffer "buf". | |
| 2652 * They add up to any currently used midword characters. | |
| 2653 */ | |
| 2654 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2655 use_midword(slang_T *lp, win_T *wp) |
| 376 | 2656 { |
| 2657 char_u *p; | |
| 2658 | |
| 389 | 2659 if (lp->sl_midword == NULL) /* there aren't any */ |
| 2660 return; | |
| 2661 | |
| 376 | 2662 for (p = lp->sl_midword; *p != NUL; ) |
| 2663 #ifdef FEAT_MBYTE | |
| 2664 if (has_mbyte) | |
| 2665 { | |
| 2666 int c, l, n; | |
| 2667 char_u *bp; | |
| 2668 | |
| 2669 c = mb_ptr2char(p); | |
| 474 | 2670 l = (*mb_ptr2len)(p); |
| 2671 if (c < 256 && l <= 2) | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2672 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
|
2673 else if (wp->w_s->b_spell_ismw_mb == NULL) |
| 376 | 2674 /* 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
|
2675 wp->w_s->b_spell_ismw_mb = vim_strnsave(p, l); |
| 376 | 2676 else |
| 2677 { | |
| 2678 /* 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
|
2679 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
|
2680 bp = vim_strnsave(wp->w_s->b_spell_ismw_mb, n + l); |
| 376 | 2681 if (bp != NULL) |
| 2682 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2683 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
|
2684 wp->w_s->b_spell_ismw_mb = bp; |
| 376 | 2685 vim_strncpy(bp + n, p, l); |
| 2686 } | |
| 2687 } | |
| 2688 p += l; | |
| 2689 } | |
| 2690 else | |
| 2691 #endif | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2692 wp->w_s->b_spell_ismw[*p++] = TRUE; |
| 376 | 2693 } |
| 2694 | |
| 2695 /* | |
| 236 | 2696 * Find the region "region[2]" in "rp" (points to "sl_regions"). |
| 2697 * Each region is simply stored as the two characters of it's name. | |
| 381 | 2698 * Returns the index if found (first is 0), REGION_ALL if not found. |
| 236 | 2699 */ |
| 2700 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2701 find_region(char_u *rp, char_u *region) |
| 236 | 2702 { |
| 2703 int i; | |
| 2704 | |
| 2705 for (i = 0; ; i += 2) | |
| 2706 { | |
| 2707 if (rp[i] == NUL) | |
| 2708 return REGION_ALL; | |
| 2709 if (rp[i] == region[0] && rp[i + 1] == region[1]) | |
| 2710 break; | |
| 2711 } | |
| 2712 return i / 2; | |
| 2713 } | |
| 2714 | |
| 2715 /* | |
| 323 | 2716 * Return case type of word: |
| 236 | 2717 * w word 0 |
| 300 | 2718 * Word WF_ONECAP |
| 2719 * W WORD WF_ALLCAP | |
| 2720 * WoRd wOrd WF_KEEPCAP | |
| 236 | 2721 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2722 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2723 captype( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2724 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2725 char_u *end) /* When NULL use up to NUL byte. */ |
| 236 | 2726 { |
| 2727 char_u *p; | |
| 2728 int c; | |
| 2729 int firstcap; | |
| 2730 int allcap; | |
| 2731 int past_second = FALSE; /* past second word char */ | |
| 2732 | |
| 2733 /* find first letter */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2734 for (p = word; !spell_iswordp_nmw(p, curwin); MB_PTR_ADV(p)) |
| 323 | 2735 if (end == NULL ? *p == NUL : p >= end) |
| 236 | 2736 return 0; /* only non-word characters, illegal word */ |
| 2737 #ifdef FEAT_MBYTE | |
| 310 | 2738 if (has_mbyte) |
| 2739 c = mb_ptr2char_adv(&p); | |
| 2740 else | |
| 236 | 2741 #endif |
| 310 | 2742 c = *p++; |
| 324 | 2743 firstcap = allcap = SPELL_ISUPPER(c); |
| 236 | 2744 |
| 2745 /* | |
| 2746 * Need to check all letters to find a word with mixed upper/lower. | |
| 2747 * But a word with an upper char only at start is a ONECAP. | |
| 2748 */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2749 for ( ; end == NULL ? *p != NUL : p < end; MB_PTR_ADV(p)) |
| 5477 | 2750 if (spell_iswordp_nmw(p, curwin)) |
| 236 | 2751 { |
| 455 | 2752 c = PTR2CHAR(p); |
| 324 | 2753 if (!SPELL_ISUPPER(c)) |
| 236 | 2754 { |
| 2755 /* UUl -> KEEPCAP */ | |
| 2756 if (past_second && allcap) | |
| 300 | 2757 return WF_KEEPCAP; |
| 236 | 2758 allcap = FALSE; |
| 2759 } | |
| 2760 else if (!allcap) | |
| 2761 /* UlU -> KEEPCAP */ | |
| 300 | 2762 return WF_KEEPCAP; |
| 236 | 2763 past_second = TRUE; |
| 2764 } | |
| 2765 | |
| 2766 if (allcap) | |
| 300 | 2767 return WF_ALLCAP; |
| 236 | 2768 if (firstcap) |
| 300 | 2769 return WF_ONECAP; |
| 236 | 2770 return 0; |
| 2771 } | |
| 2772 | |
| 474 | 2773 /* |
| 2774 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a | |
| 2775 * capital. So that make_case_word() can turn WOrd into Word. | |
| 2776 * Add ALLCAP for "WOrD". | |
| 2777 */ | |
| 2778 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2779 badword_captype(char_u *word, char_u *end) |
| 474 | 2780 { |
| 2781 int flags = captype(word, end); | |
| 475 | 2782 int c; |
| 474 | 2783 int l, u; |
| 2784 int first; | |
| 2785 char_u *p; | |
| 2786 | |
| 2787 if (flags & WF_KEEPCAP) | |
| 2788 { | |
| 2789 /* Count the number of UPPER and lower case letters. */ | |
| 2790 l = u = 0; | |
| 2791 first = FALSE; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2792 for (p = word; p < end; MB_PTR_ADV(p)) |
| 474 | 2793 { |
| 475 | 2794 c = PTR2CHAR(p); |
| 2795 if (SPELL_ISUPPER(c)) | |
| 474 | 2796 { |
| 2797 ++u; | |
| 2798 if (p == word) | |
| 2799 first = TRUE; | |
| 2800 } | |
| 2801 else | |
| 2802 ++l; | |
| 2803 } | |
| 2804 | |
| 2805 /* If there are more UPPER than lower case letters suggest an | |
| 2806 * ALLCAP word. Otherwise, if the first letter is UPPER then | |
| 2807 * suggest ONECAP. Exception: "ALl" most likely should be "All", | |
| 2808 * require three upper case letters. */ | |
| 2809 if (u > l && u > 2) | |
| 2810 flags |= WF_ALLCAP; | |
| 2811 else if (first) | |
| 2812 flags |= WF_ONECAP; | |
| 633 | 2813 |
| 2814 if (u >= 2 && l >= 2) /* maCARONI maCAroni */ | |
| 2815 flags |= WF_MIXCAP; | |
| 474 | 2816 } |
| 2817 return flags; | |
| 2818 } | |
| 2819 | |
| 5519 | 2820 /* |
| 2821 * Delete the internal wordlist and its .spl file. | |
| 2822 */ | |
| 2823 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2824 spell_delete_wordlist(void) |
| 5519 | 2825 { |
| 2826 char_u fname[MAXPATHL]; | |
| 2827 | |
| 2828 if (int_wordlist != NULL) | |
| 2829 { | |
| 2830 mch_remove(int_wordlist); | |
| 2831 int_wordlist_spl(fname); | |
| 2832 mch_remove(fname); | |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2833 VIM_CLEAR(int_wordlist); |
| 5519 | 2834 } |
| 2835 } | |
| 2836 | |
| 2837 #if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO) | |
| 355 | 2838 /* |
| 2839 * Free all languages. | |
| 2840 */ | |
| 2841 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2842 spell_free_all(void) |
| 355 | 2843 { |
| 503 | 2844 slang_T *slang; |
| 355 | 2845 buf_T *buf; |
| 2846 | |
| 2580 | 2847 /* 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
|
2848 FOR_ALL_BUFFERS(buf) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2849 ga_clear(&buf->b_s.b_langp); |
| 355 | 2850 |
| 2851 while (first_lang != NULL) | |
| 2852 { | |
| 503 | 2853 slang = first_lang; |
| 2854 first_lang = slang->sl_next; | |
| 2855 slang_free(slang); | |
| 355 | 2856 } |
| 366 | 2857 |
| 5519 | 2858 spell_delete_wordlist(); |
| 381 | 2859 |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2860 VIM_CLEAR(repl_to); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2861 VIM_CLEAR(repl_from); |
| 355 | 2862 } |
| 5519 | 2863 #endif |
| 2864 | |
| 2865 #if defined(FEAT_MBYTE) || defined(PROTO) | |
| 236 | 2866 /* |
| 2867 * Clear all spelling tables and reload them. | |
| 307 | 2868 * Used after 'encoding' is set and when ":mkspell" was used. |
| 236 | 2869 */ |
| 2870 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2871 spell_reload(void) |
| 236 | 2872 { |
| 316 | 2873 win_T *wp; |
| 236 | 2874 |
| 358 | 2875 /* Initialize the table for spell_iswordp(). */ |
| 236 | 2876 init_spell_chartab(); |
| 2877 | |
| 2878 /* Unload all allocated memory. */ | |
| 355 | 2879 spell_free_all(); |
| 236 | 2880 |
| 2881 /* 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
|
2882 FOR_ALL_WINDOWS(wp) |
| 236 | 2883 { |
| 316 | 2884 /* Only load the wordlists when 'spelllang' is set and there is a |
| 2885 * 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
|
2886 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
|
2887 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2888 if (wp->w_p_spell) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2889 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2890 (void)did_set_spelllang(wp); |
| 316 | 2891 break; |
| 2892 } | |
| 2893 } | |
| 236 | 2894 } |
| 2895 } | |
| 5519 | 2896 #endif |
| 236 | 2897 |
| 310 | 2898 /* |
| 625 | 2899 * Opposite of offset2bytes(). |
| 2900 * "pp" points to the bytes and is advanced over it. | |
| 2901 * Returns the offset. | |
| 2902 */ | |
| 2903 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2904 bytes2offset(char_u **pp) |
| 625 | 2905 { |
| 2906 char_u *p = *pp; | |
| 2907 int nr; | |
| 2908 int c; | |
| 2909 | |
| 2910 c = *p++; | |
| 2911 if ((c & 0x80) == 0x00) /* 1 byte */ | |
| 2912 { | |
| 2913 nr = c - 1; | |
| 2914 } | |
| 2915 else if ((c & 0xc0) == 0x80) /* 2 bytes */ | |
| 2916 { | |
| 2917 nr = (c & 0x3f) - 1; | |
| 2918 nr = nr * 255 + (*p++ - 1); | |
| 2919 } | |
| 2920 else if ((c & 0xe0) == 0xc0) /* 3 bytes */ | |
| 2921 { | |
| 2922 nr = (c & 0x1f) - 1; | |
| 2923 nr = nr * 255 + (*p++ - 1); | |
| 2924 nr = nr * 255 + (*p++ - 1); | |
| 2925 } | |
| 2926 else /* 4 bytes */ | |
| 2927 { | |
| 2928 nr = (c & 0x0f) - 1; | |
| 2929 nr = nr * 255 + (*p++ - 1); | |
| 2930 nr = nr * 255 + (*p++ - 1); | |
| 2931 nr = nr * 255 + (*p++ - 1); | |
| 2932 } | |
| 2933 | |
| 2934 *pp = p; | |
| 2935 return nr; | |
| 2936 } | |
| 2937 | |
| 2938 | |
| 2939 /* | |
| 2940 * Open a spell buffer. This is a nameless buffer that is not in the buffer | |
| 2941 * list and only contains text lines. Can use a swapfile to reduce memory | |
| 2942 * use. | |
| 2943 * Most other fields are invalid! Esp. watch out for string options being | |
| 2944 * NULL and there is no undo info. | |
| 2945 * Returns NULL when out of memory. | |
| 2946 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2947 buf_T * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2948 open_spellbuf(void) |
| 625 | 2949 { |
| 2950 buf_T *buf; | |
| 2951 | |
| 2952 buf = (buf_T *)alloc_clear(sizeof(buf_T)); | |
| 2953 if (buf != NULL) | |
| 2954 { | |
| 2955 buf->b_spell = TRUE; | |
| 2956 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
|
2957 #ifdef FEAT_CRYPT |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
2958 buf->b_p_key = empty_option; |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
2959 #endif |
| 625 | 2960 ml_open(buf); |
| 2961 ml_open_file(buf); /* create swap file now */ | |
| 2962 } | |
| 2963 return buf; | |
| 2964 } | |
| 2965 | |
| 2966 /* | |
| 2967 * Close the buffer used for spell info. | |
| 2968 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2969 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2970 close_spellbuf(buf_T *buf) |
| 625 | 2971 { |
| 2972 if (buf != NULL) | |
| 2973 { | |
| 2974 ml_close(buf, TRUE); | |
| 2975 vim_free(buf); | |
| 2976 } | |
| 2977 } | |
| 2978 | |
| 307 | 2979 /* |
| 2980 * Init the chartab used for spelling for ASCII. | |
| 2981 * EBCDIC is not supported! | |
| 2982 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2983 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2984 clear_spell_chartab(spelltab_T *sp) |
| 307 | 2985 { |
| 324 | 2986 int i; |
| 307 | 2987 |
| 2988 /* Init everything to FALSE. */ | |
| 2989 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw)); | |
| 2990 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu)); | |
| 2991 for (i = 0; i < 256; ++i) | |
| 324 | 2992 { |
| 307 | 2993 sp->st_fold[i] = i; |
| 324 | 2994 sp->st_upper[i] = i; |
| 2995 } | |
| 307 | 2996 |
| 2997 /* We include digits. A word shouldn't start with a digit, but handling | |
| 2998 * that is done separately. */ | |
| 2999 for (i = '0'; i <= '9'; ++i) | |
| 3000 sp->st_isw[i] = TRUE; | |
| 3001 for (i = 'A'; i <= 'Z'; ++i) | |
| 3002 { | |
| 3003 sp->st_isw[i] = TRUE; | |
| 3004 sp->st_isu[i] = TRUE; | |
| 3005 sp->st_fold[i] = i + 0x20; | |
| 3006 } | |
| 3007 for (i = 'a'; i <= 'z'; ++i) | |
| 324 | 3008 { |
| 307 | 3009 sp->st_isw[i] = TRUE; |
| 324 | 3010 sp->st_upper[i] = i - 0x20; |
| 3011 } | |
| 307 | 3012 } |
| 3013 | |
| 3014 /* | |
| 3015 * Init the chartab used for spelling. Only depends on 'encoding'. | |
| 3016 * Called once while starting up and when 'encoding' changes. | |
| 3017 * The default is to use isalpha(), but the spell file should define the word | |
| 3018 * characters to make it possible that 'encoding' differs from the current | |
| 390 | 3019 * locale. For utf-8 we don't use isalpha() but our own functions. |
| 307 | 3020 */ |
| 3021 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3022 init_spell_chartab(void) |
| 307 | 3023 { |
| 3024 int i; | |
| 3025 | |
| 3026 did_set_spelltab = FALSE; | |
| 3027 clear_spell_chartab(&spelltab); | |
| 3028 #ifdef FEAT_MBYTE | |
| 3029 if (enc_dbcs) | |
| 3030 { | |
| 3031 /* DBCS: assume double-wide characters are word characters. */ | |
| 3032 for (i = 128; i <= 255; ++i) | |
| 3033 if (MB_BYTE2LEN(i) == 2) | |
| 3034 spelltab.st_isw[i] = TRUE; | |
| 3035 } | |
| 324 | 3036 else if (enc_utf8) |
| 3037 { | |
| 3038 for (i = 128; i < 256; ++i) | |
| 3039 { | |
|
2140
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3040 int f = utf_fold(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3041 int u = utf_toupper(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3042 |
| 324 | 3043 spelltab.st_isu[i] = utf_isupper(i); |
| 3044 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
|
3045 /* 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
|
3046 * 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
|
3047 * value for utf-8 to avoid this. */ |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3048 spelltab.st_fold[i] = (f < 256) ? f : i; |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3049 spelltab.st_upper[i] = (u < 256) ? u : i; |
| 324 | 3050 } |
| 3051 } | |
| 307 | 3052 else |
| 3053 #endif | |
| 3054 { | |
| 324 | 3055 /* Rough guess: use locale-dependent library functions. */ |
| 307 | 3056 for (i = 128; i < 256; ++i) |
| 3057 { | |
| 3058 if (MB_ISUPPER(i)) | |
| 3059 { | |
| 324 | 3060 spelltab.st_isw[i] = TRUE; |
| 307 | 3061 spelltab.st_isu[i] = TRUE; |
| 3062 spelltab.st_fold[i] = MB_TOLOWER(i); | |
| 3063 } | |
| 324 | 3064 else if (MB_ISLOWER(i)) |
| 3065 { | |
| 3066 spelltab.st_isw[i] = TRUE; | |
| 3067 spelltab.st_upper[i] = MB_TOUPPER(i); | |
| 3068 } | |
| 307 | 3069 } |
| 3070 } | |
| 3071 } | |
| 3072 | |
| 3073 | |
| 3074 /* | |
| 358 | 3075 * Return TRUE if "p" points to a word character. |
| 366 | 3076 * As a special case we see "midword" characters as word character when it is |
| 358 | 3077 * followed by a word character. This finds they'there but not 'they there'. |
| 366 | 3078 * Thus this only works properly when past the first character of the word. |
| 358 | 3079 */ |
| 3080 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3081 spell_iswordp( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3082 char_u *p, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3083 win_T *wp) /* buffer used */ |
| 358 | 3084 { |
| 366 | 3085 #ifdef FEAT_MBYTE |
| 358 | 3086 char_u *s; |
| 366 | 3087 int l; |
| 3088 int c; | |
| 3089 | |
| 3090 if (has_mbyte) | |
| 3091 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
3092 l = MB_PTR2LEN(p); |
| 358 | 3093 s = p; |
| 366 | 3094 if (l == 1) |
| 3095 { | |
| 3096 /* be quick for ASCII */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3097 if (wp->w_s->b_spell_ismw[*p]) |
| 366 | 3098 s = p + 1; /* skip a mid-word character */ |
| 3099 } | |
| 3100 else | |
| 3101 { | |
| 3102 c = mb_ptr2char(p); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3103 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
|
3104 : (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
|
3105 && vim_strchr(wp->w_s->b_spell_ismw_mb, c) != NULL)) |
| 366 | 3106 s = p + l; |
| 3107 } | |
| 3108 | |
| 390 | 3109 c = mb_ptr2char(s); |
| 3110 if (c > 255) | |
| 5477 | 3111 return spell_mb_isword_class(mb_get_class(s), wp); |
| 390 | 3112 return spelltab.st_isw[c]; |
| 366 | 3113 } |
| 358 | 3114 #endif |
| 366 | 3115 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3116 return spelltab.st_isw[wp->w_s->b_spell_ismw[*p] ? p[1] : p[0]]; |
| 376 | 3117 } |
| 3118 | |
| 3119 /* | |
| 3120 * Return TRUE if "p" points to a word character. | |
| 3121 * Unlike spell_iswordp() this doesn't check for "midword" characters. | |
| 3122 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3123 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3124 spell_iswordp_nmw(char_u *p, win_T *wp) |
| 376 | 3125 { |
| 3126 #ifdef FEAT_MBYTE | |
| 390 | 3127 int c; |
| 3128 | |
| 3129 if (has_mbyte) | |
| 3130 { | |
| 3131 c = mb_ptr2char(p); | |
| 3132 if (c > 255) | |
| 5477 | 3133 return spell_mb_isword_class(mb_get_class(p), wp); |
| 390 | 3134 return spelltab.st_isw[c]; |
| 3135 } | |
| 376 | 3136 #endif |
| 3137 return spelltab.st_isw[*p]; | |
| 358 | 3138 } |
| 3139 | |
| 372 | 3140 #ifdef FEAT_MBYTE |
| 3141 /* | |
| 1580 | 3142 * Return TRUE if word class indicates a word character. |
| 3143 * Only for characters above 255. | |
| 3144 * Unicode subscript and superscript are not considered word characters. | |
| 5477 | 3145 * See also dbcs_class() and utf_class() in mbyte.c. |
| 3146 */ | |
| 3147 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3148 spell_mb_isword_class(int cl, win_T *wp) |
| 5477 | 3149 { |
| 3150 if (wp->w_s->b_cjk) | |
| 3151 /* East Asian characters are not considered word characters. */ | |
| 3152 return cl == 2 || cl == 0x2800; | |
| 1580 | 3153 return cl >= 2 && cl != 0x2070 && cl != 0x2080; |
| 3154 } | |
| 3155 | |
| 3156 /* | |
| 372 | 3157 * Return TRUE if "p" points to a word character. |
| 3158 * Wide version of spell_iswordp(). | |
| 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_iswordp_w(int *p, win_T *wp) |
| 372 | 3162 { |
| 3163 int *s; | |
| 3164 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3165 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
|
3166 : (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
|
3167 && vim_strchr(wp->w_s->b_spell_ismw_mb, *p) != NULL)) |
| 372 | 3168 s = p + 1; |
| 3169 else | |
| 3170 s = p; | |
| 3171 | |
| 390 | 3172 if (*s > 255) |
| 372 | 3173 { |
| 3174 if (enc_utf8) | |
| 5477 | 3175 return spell_mb_isword_class(utf_class(*s), wp); |
| 372 | 3176 if (enc_dbcs) |
| 5477 | 3177 return spell_mb_isword_class( |
| 3178 dbcs_class((unsigned)*s >> 8, *s & 0xff), wp); | |
| 372 | 3179 return 0; |
| 3180 } | |
| 3181 return spelltab.st_isw[*s]; | |
| 3182 } | |
| 3183 #endif | |
| 3184 | |
| 358 | 3185 /* |
| 324 | 3186 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated. |
| 3187 * Uses the character definitions from the .spl file. | |
| 307 | 3188 * When using a multi-byte 'encoding' the length may change! |
| 3189 * Returns FAIL when something wrong. | |
| 3190 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3191 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3192 spell_casefold( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3193 char_u *str, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3194 int len, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3195 char_u *buf, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3196 int buflen) |
| 307 | 3197 { |
| 3198 int i; | |
| 3199 | |
| 3200 if (len >= buflen) | |
| 3201 { | |
| 3202 buf[0] = NUL; | |
| 3203 return FAIL; /* result will not fit */ | |
| 3204 } | |
| 3205 | |
| 3206 #ifdef FEAT_MBYTE | |
| 3207 if (has_mbyte) | |
| 3208 { | |
| 324 | 3209 int outi = 0; |
| 3210 char_u *p; | |
| 307 | 3211 int c; |
| 3212 | |
| 3213 /* Fold one character at a time. */ | |
| 324 | 3214 for (p = str; p < str + len; ) |
| 307 | 3215 { |
| 3216 if (outi + MB_MAXBYTES > buflen) | |
| 3217 { | |
| 3218 buf[outi] = NUL; | |
| 3219 return FAIL; | |
| 3220 } | |
| 474 | 3221 c = mb_cptr2char_adv(&p); |
| 324 | 3222 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi); |
| 307 | 3223 } |
| 3224 buf[outi] = NUL; | |
| 3225 } | |
| 3226 else | |
| 3227 #endif | |
| 3228 { | |
| 3229 /* Be quick for non-multibyte encodings. */ | |
| 3230 for (i = 0; i < len; ++i) | |
| 324 | 3231 buf[i] = spelltab.st_fold[str[i]]; |
| 307 | 3232 buf[i] = NUL; |
| 3233 } | |
| 3234 | |
| 3235 return OK; | |
| 3236 } | |
| 3237 | |
| 625 | 3238 /* values for sps_flags */ |
| 372 | 3239 #define SPS_BEST 1 |
| 3240 #define SPS_FAST 2 | |
| 3241 #define SPS_DOUBLE 4 | |
| 3242 | |
| 625 | 3243 static int sps_flags = SPS_BEST; /* flags from 'spellsuggest' */ |
| 3244 static int sps_limit = 9999; /* max nr of suggestions given */ | |
| 372 | 3245 |
| 3246 /* | |
| 3247 * Check the 'spellsuggest' option. Return FAIL if it's wrong. | |
| 483 | 3248 * Sets "sps_flags" and "sps_limit". |
| 372 | 3249 */ |
| 3250 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3251 spell_check_sps(void) |
| 372 | 3252 { |
| 3253 char_u *p; | |
| 483 | 3254 char_u *s; |
| 372 | 3255 char_u buf[MAXPATHL]; |
| 3256 int f; | |
| 3257 | |
| 3258 sps_flags = 0; | |
| 483 | 3259 sps_limit = 9999; |
| 372 | 3260 |
| 3261 for (p = p_sps; *p != NUL; ) | |
| 3262 { | |
| 3263 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 3264 | |
| 3265 f = 0; | |
| 483 | 3266 if (VIM_ISDIGIT(*buf)) |
| 3267 { | |
| 3268 s = buf; | |
| 3269 sps_limit = getdigits(&s); | |
| 3270 if (*s != NUL && !VIM_ISDIGIT(*s)) | |
| 3271 f = -1; | |
| 3272 } | |
| 3273 else if (STRCMP(buf, "best") == 0) | |
| 372 | 3274 f = SPS_BEST; |
| 3275 else if (STRCMP(buf, "fast") == 0) | |
| 3276 f = SPS_FAST; | |
| 3277 else if (STRCMP(buf, "double") == 0) | |
| 3278 f = SPS_DOUBLE; | |
| 3279 else if (STRNCMP(buf, "expr:", 5) != 0 | |
| 3280 && STRNCMP(buf, "file:", 5) != 0) | |
| 3281 f = -1; | |
| 3282 | |
| 3283 if (f == -1 || (sps_flags != 0 && f != 0)) | |
| 3284 { | |
| 3285 sps_flags = SPS_BEST; | |
| 483 | 3286 sps_limit = 9999; |
| 372 | 3287 return FAIL; |
| 3288 } | |
| 3289 if (f != 0) | |
| 3290 sps_flags = f; | |
| 3291 } | |
| 3292 | |
| 3293 if (sps_flags == 0) | |
| 3294 sps_flags = SPS_BEST; | |
| 3295 | |
| 3296 return OK; | |
| 3297 } | |
| 3298 | |
| 323 | 3299 /* |
| 5374 | 3300 * "z=": Find badly spelled word under or after the cursor. |
| 323 | 3301 * Give suggestions for the properly spelled word. |
| 638 | 3302 * In Visual mode use the highlighted word as the bad word. |
| 485 | 3303 * When "count" is non-zero use that suggestion. |
| 323 | 3304 */ |
| 3305 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3306 spell_suggest(int count) |
| 323 | 3307 { |
| 3308 char_u *line; | |
| 3309 pos_T prev_cursor = curwin->w_cursor; | |
| 3310 char_u wcopy[MAXWLEN + 2]; | |
| 3311 char_u *p; | |
| 3312 int i; | |
| 3313 int c; | |
| 3314 suginfo_T sug; | |
| 3315 suggest_T *stp; | |
| 372 | 3316 int mouse_used; |
| 388 | 3317 int need_cap; |
| 483 | 3318 int limit; |
| 485 | 3319 int selected = count; |
| 638 | 3320 int badlen = 0; |
| 1938 | 3321 int msg_scroll_save = msg_scroll; |
| 638 | 3322 |
| 3323 if (no_spell_checking(curwin)) | |
| 3324 return; | |
| 3325 | |
| 3326 if (VIsual_active) | |
| 3327 { | |
| 3328 /* Use the Visually selected text as the bad word. But reject | |
| 3329 * a multi-line selection. */ | |
| 3330 if (curwin->w_cursor.lnum != VIsual.lnum) | |
| 3331 { | |
| 6949 | 3332 vim_beep(BO_SPELL); |
| 638 | 3333 return; |
| 3334 } | |
| 3335 badlen = (int)curwin->w_cursor.col - (int)VIsual.col; | |
| 3336 if (badlen < 0) | |
| 3337 badlen = -badlen; | |
| 3338 else | |
| 3339 curwin->w_cursor.col = VIsual.col; | |
| 3340 ++badlen; | |
| 3341 end_visual_mode(); | |
| 3342 } | |
| 5735 | 3343 /* Find the start of the badly spelled word. */ |
| 3344 else if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0 | |
| 346 | 3345 || curwin->w_cursor.col > prev_cursor.col) |
| 3346 { | |
| 3347 /* No bad word or it starts after the cursor: use the word under the | |
| 3348 * cursor. */ | |
| 3349 curwin->w_cursor = prev_cursor; | |
| 3350 line = ml_get_curline(); | |
| 3351 p = line + curwin->w_cursor.col; | |
| 3352 /* Backup to before start of word. */ | |
| 5477 | 3353 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
|
3354 MB_PTR_BACK(line, p); |
| 346 | 3355 /* Forward to start of word. */ |
| 5477 | 3356 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
|
3357 MB_PTR_ADV(p); |
| 346 | 3358 |
| 5477 | 3359 if (!spell_iswordp_nmw(p, curwin)) /* No word found. */ |
| 346 | 3360 { |
| 3361 beep_flush(); | |
| 3362 return; | |
| 3363 } | |
| 835 | 3364 curwin->w_cursor.col = (colnr_T)(p - line); |
| 323 | 3365 } |
| 3366 | |
| 344 | 3367 /* Get the word and its length. */ |
| 3368 | |
| 388 | 3369 /* Figure out if the word should be capitalised. */ |
| 475 | 3370 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col); |
| 3371 | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3372 /* 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
|
3373 line = vim_strsave(ml_get_curline()); |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3374 if (line == NULL) |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3375 goto skip; |
| 388 | 3376 |
| 483 | 3377 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in |
| 3378 * 'spellsuggest', whatever is smaller. */ | |
| 3379 if (sps_limit > (int)Rows - 2) | |
| 3380 limit = (int)Rows - 2; | |
| 3381 else | |
| 3382 limit = sps_limit; | |
| 638 | 3383 spell_find_suggest(line + curwin->w_cursor.col, badlen, &sug, limit, |
| 625 | 3384 TRUE, need_cap, TRUE); |
| 323 | 3385 |
| 3386 if (sug.su_ga.ga_len == 0) | |
| 3387 MSG(_("Sorry, no suggestions")); | |
| 485 | 3388 else if (count > 0) |
| 3389 { | |
| 3390 if (count > sug.su_ga.ga_len) | |
| 3391 smsg((char_u *)_("Sorry, only %ld suggestions"), | |
| 3392 (long)sug.su_ga.ga_len); | |
| 3393 } | |
| 323 | 3394 else |
| 3395 { | |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
3396 VIM_CLEAR(repl_from); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
3397 VIM_CLEAR(repl_to); |
| 372 | 3398 |
| 474 | 3399 #ifdef FEAT_RIGHTLEFT |
| 3400 /* When 'rightleft' is set the list is drawn right-left. */ | |
| 3401 cmdmsg_rl = curwin->w_p_rl; | |
| 3402 if (cmdmsg_rl) | |
| 3403 msg_col = Columns - 1; | |
| 3404 #endif | |
| 3405 | |
| 323 | 3406 /* List the suggestions. */ |
| 3407 msg_start(); | |
| 914 | 3408 msg_row = Rows - 1; /* for when 'cmdheight' > 1 */ |
| 372 | 3409 lines_left = Rows; /* avoid more prompt */ |
| 323 | 3410 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"), |
| 3411 sug.su_badlen, sug.su_badptr); | |
| 474 | 3412 #ifdef FEAT_RIGHTLEFT |
| 3413 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0) | |
| 3414 { | |
| 3415 /* And now the rabbit from the high hat: Avoid showing the | |
| 3416 * untranslated message rightleft. */ | |
| 3417 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC", | |
| 3418 sug.su_badlen, sug.su_badptr); | |
| 3419 } | |
| 3420 #endif | |
| 323 | 3421 msg_puts(IObuff); |
| 3422 msg_clr_eos(); | |
| 3423 msg_putchar('\n'); | |
| 346 | 3424 |
| 323 | 3425 msg_scroll = TRUE; |
| 3426 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 3427 { | |
| 344 | 3428 stp = &SUG(sug.su_ga, i); |
| 323 | 3429 |
| 3430 /* The suggested word may replace only part of the bad word, add | |
| 3431 * the not replaced part. */ | |
| 2768 | 3432 vim_strncpy(wcopy, stp->st_word, MAXWLEN); |
| 323 | 3433 if (sug.su_badlen > stp->st_orglen) |
| 625 | 3434 vim_strncpy(wcopy + stp->st_wordlen, |
| 323 | 3435 sug.su_badptr + stp->st_orglen, |
| 3436 sug.su_badlen - stp->st_orglen); | |
| 474 | 3437 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1); |
| 3438 #ifdef FEAT_RIGHTLEFT | |
| 3439 if (cmdmsg_rl) | |
| 3440 rl_mirror(IObuff); | |
| 3441 #endif | |
| 3442 msg_puts(IObuff); | |
| 3443 | |
| 3444 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy); | |
| 346 | 3445 msg_puts(IObuff); |
| 3446 | |
| 3447 /* The word may replace more than "su_badlen". */ | |
| 3448 if (sug.su_badlen < stp->st_orglen) | |
| 3449 { | |
| 3450 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""), | |
| 3451 stp->st_orglen, sug.su_badptr); | |
| 3452 msg_puts(IObuff); | |
| 3453 } | |
| 3454 | |
| 324 | 3455 if (p_verbose > 0) |
| 344 | 3456 { |
| 346 | 3457 /* Add the score. */ |
| 351 | 3458 if (sps_flags & (SPS_DOUBLE | SPS_BEST)) |
| 474 | 3459 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)", |
| 344 | 3460 stp->st_salscore ? "s " : "", |
| 3461 stp->st_score, stp->st_altscore); | |
| 3462 else | |
| 474 | 3463 vim_snprintf((char *)IObuff, IOSIZE, " (%d)", |
| 346 | 3464 stp->st_score); |
| 474 | 3465 #ifdef FEAT_RIGHTLEFT |
| 3466 if (cmdmsg_rl) | |
| 3467 /* Mirror the numbers, but keep the leading space. */ | |
| 3468 rl_mirror(IObuff + 1); | |
| 3469 #endif | |
| 346 | 3470 msg_advance(30); |
| 3471 msg_puts(IObuff); | |
| 344 | 3472 } |
| 323 | 3473 msg_putchar('\n'); |
| 3474 } | |
| 3475 | |
| 474 | 3476 #ifdef FEAT_RIGHTLEFT |
| 3477 cmdmsg_rl = FALSE; | |
| 3478 msg_col = 0; | |
| 3479 #endif | |
| 323 | 3480 /* Ask for choice. */ |
| 485 | 3481 selected = prompt_for_number(&mouse_used); |
| 372 | 3482 if (mouse_used) |
| 485 | 3483 selected -= lines_left; |
| 1938 | 3484 lines_left = Rows; /* avoid more prompt */ |
| 3485 /* don't delay for 'smd' in normal_cmd() */ | |
| 3486 msg_scroll = msg_scroll_save; | |
| 485 | 3487 } |
| 3488 | |
| 3489 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK) | |
| 3490 { | |
| 3491 /* Save the from and to text for :spellrepall. */ | |
| 3492 stp = &SUG(sug.su_ga, selected - 1); | |
| 548 | 3493 if (sug.su_badlen > stp->st_orglen) |
| 3494 { | |
| 3495 /* Replacing less than "su_badlen", append the remainder to | |
| 3496 * repl_to. */ | |
| 3497 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen); | |
| 3498 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word, | |
| 3499 sug.su_badlen - stp->st_orglen, | |
| 3500 sug.su_badptr + stp->st_orglen); | |
| 3501 repl_to = vim_strsave(IObuff); | |
| 3502 } | |
| 3503 else | |
| 3504 { | |
| 3505 /* Replacing su_badlen or more, use the whole word. */ | |
| 3506 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen); | |
| 3507 repl_to = vim_strsave(stp->st_word); | |
| 3508 } | |
| 485 | 3509 |
| 3510 /* Replace the word. */ | |
| 1938 | 3511 p = alloc((unsigned)STRLEN(line) - stp->st_orglen |
| 3512 + stp->st_wordlen + 1); | |
| 485 | 3513 if (p != NULL) |
| 3514 { | |
| 835 | 3515 c = (int)(sug.su_badptr - line); |
| 485 | 3516 mch_memmove(p, line, c); |
| 3517 STRCPY(p + c, stp->st_word); | |
| 3518 STRCAT(p, sug.su_badptr + stp->st_orglen); | |
| 3519 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 3520 curwin->w_cursor.col = c; | |
| 3521 | |
| 3522 /* For redo we use a change-word command. */ | |
| 3523 ResetRedobuff(); | |
| 3524 AppendToRedobuff((char_u *)"ciw"); | |
| 620 | 3525 AppendToRedobuffLit(p + c, |
| 625 | 3526 stp->st_wordlen + sug.su_badlen - stp->st_orglen); |
| 485 | 3527 AppendCharToRedobuff(ESC); |
| 819 | 3528 |
| 3529 /* After this "p" may be invalid. */ | |
| 3530 changed_bytes(curwin->w_cursor.lnum, c); | |
| 485 | 3531 } |
| 3532 } | |
| 3533 else | |
| 3534 curwin->w_cursor = prev_cursor; | |
| 323 | 3535 |
| 344 | 3536 spell_find_cleanup(&sug); |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3537 skip: |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3538 vim_free(line); |
| 344 | 3539 } |
| 3540 | |
| 3541 /* | |
| 475 | 3542 * Check if the word at line "lnum" column "col" is required to start with a |
| 3543 * capital. This uses 'spellcapcheck' of the current buffer. | |
| 3544 */ | |
| 3545 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3546 check_need_cap(linenr_T lnum, colnr_T col) |
| 475 | 3547 { |
| 3548 int need_cap = FALSE; | |
| 3549 char_u *line; | |
| 3550 char_u *line_copy = NULL; | |
| 3551 char_u *p; | |
| 3552 colnr_T endcol; | |
| 3553 regmatch_T regmatch; | |
| 3554 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3555 if (curwin->w_s->b_cap_prog == NULL) |
| 475 | 3556 return FALSE; |
| 3557 | |
| 3558 line = ml_get_curline(); | |
| 3559 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
|
3560 if (getwhitecols(line) >= (int)col) |
| 475 | 3561 { |
| 3562 /* At start of line, check if previous line is empty or sentence | |
| 3563 * ends there. */ | |
| 3564 if (lnum == 1) | |
| 3565 need_cap = TRUE; | |
| 3566 else | |
| 3567 { | |
| 3568 line = ml_get(lnum - 1); | |
| 3569 if (*skipwhite(line) == NUL) | |
| 3570 need_cap = TRUE; | |
| 3571 else | |
| 3572 { | |
| 3573 /* Append a space in place of the line break. */ | |
| 3574 line_copy = concat_str(line, (char_u *)" "); | |
| 3575 line = line_copy; | |
| 835 | 3576 endcol = (colnr_T)STRLEN(line); |
| 475 | 3577 } |
| 3578 } | |
| 3579 } | |
| 3580 else | |
| 3581 endcol = col; | |
| 3582 | |
| 3583 if (endcol > 0) | |
| 3584 { | |
| 3585 /* 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
|
3586 regmatch.regprog = curwin->w_s->b_cap_prog; |
| 475 | 3587 regmatch.rm_ic = FALSE; |
| 3588 p = line + endcol; | |
| 3589 for (;;) | |
| 3590 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
3591 MB_PTR_BACK(line, p); |
| 5477 | 3592 if (p == line || spell_iswordp_nmw(p, curwin)) |
| 475 | 3593 break; |
| 3594 if (vim_regexec(®match, p, 0) | |
| 3595 && regmatch.endp[0] == line + endcol) | |
| 3596 { | |
| 3597 need_cap = TRUE; | |
| 3598 break; | |
| 3599 } | |
| 3600 } | |
| 6375 | 3601 curwin->w_s->b_cap_prog = regmatch.regprog; |
| 475 | 3602 } |
| 3603 | |
| 3604 vim_free(line_copy); | |
| 3605 | |
| 3606 return need_cap; | |
| 3607 } | |
| 3608 | |
| 3609 | |
| 3610 /* | |
| 372 | 3611 * ":spellrepall" |
| 3612 */ | |
| 3613 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3614 ex_spellrepall(exarg_T *eap UNUSED) |
| 372 | 3615 { |
| 3616 pos_T pos = curwin->w_cursor; | |
| 3617 char_u *frompat; | |
| 3618 int addlen; | |
| 3619 char_u *line; | |
| 3620 char_u *p; | |
| 3621 int save_ws = p_ws; | |
| 483 | 3622 linenr_T prev_lnum = 0; |
| 372 | 3623 |
| 3624 if (repl_from == NULL || repl_to == NULL) | |
| 3625 { | |
| 3626 EMSG(_("E752: No previous spell replacement")); | |
| 3627 return; | |
| 3628 } | |
| 835 | 3629 addlen = (int)(STRLEN(repl_to) - STRLEN(repl_from)); |
| 3630 | |
| 3631 frompat = alloc((unsigned)STRLEN(repl_from) + 7); | |
| 372 | 3632 if (frompat == NULL) |
| 3633 return; | |
| 3634 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from); | |
| 3635 p_ws = FALSE; | |
| 3636 | |
| 483 | 3637 sub_nsubs = 0; |
| 3638 sub_nlines = 0; | |
| 372 | 3639 curwin->w_cursor.lnum = 0; |
| 3640 while (!got_int) | |
| 3641 { | |
|
11521
578df034735d
patch 8.0.0643: when a pattern search is slow Vim becomes unusable
Christian Brabandt <cb@256bit.org>
parents:
11394
diff
changeset
|
3642 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP, NULL, NULL) == 0 |
| 372 | 3643 || u_save_cursor() == FAIL) |
| 3644 break; | |
| 3645 | |
| 3646 /* Only replace when the right word isn't there yet. This happens | |
| 3647 * when changing "etc" to "etc.". */ | |
| 3648 line = ml_get_curline(); | |
| 3649 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col, | |
| 3650 repl_to, STRLEN(repl_to)) != 0) | |
| 3651 { | |
| 835 | 3652 p = alloc((unsigned)STRLEN(line) + addlen + 1); |
| 372 | 3653 if (p == NULL) |
| 3654 break; | |
| 3655 mch_memmove(p, line, curwin->w_cursor.col); | |
| 3656 STRCPY(p + curwin->w_cursor.col, repl_to); | |
| 3657 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from)); | |
| 3658 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 3659 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col); | |
| 483 | 3660 |
| 3661 if (curwin->w_cursor.lnum != prev_lnum) | |
| 3662 { | |
| 3663 ++sub_nlines; | |
| 3664 prev_lnum = curwin->w_cursor.lnum; | |
| 3665 } | |
| 3666 ++sub_nsubs; | |
| 372 | 3667 } |
| 835 | 3668 curwin->w_cursor.col += (colnr_T)STRLEN(repl_to); |
| 372 | 3669 } |
| 3670 | |
| 3671 p_ws = save_ws; | |
| 3672 curwin->w_cursor = pos; | |
| 3673 vim_free(frompat); | |
| 3674 | |
| 483 | 3675 if (sub_nsubs == 0) |
| 372 | 3676 EMSG2(_("E753: Not found: %s"), repl_from); |
| 483 | 3677 else |
| 3678 do_sub_msg(FALSE); | |
| 372 | 3679 } |
| 3680 | |
| 3681 /* | |
| 344 | 3682 * Find spell suggestions for "word". Return them in the growarray "*gap" as |
| 3683 * a list of allocated strings. | |
| 3684 */ | |
| 3685 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3686 spell_suggest_list( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3687 garray_T *gap, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3688 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3689 int maxcount, /* maximum nr of suggestions */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3690 int need_cap, /* 'spellcapcheck' matched */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3691 int interactive) |
| 344 | 3692 { |
| 3693 suginfo_T sug; | |
| 3694 int i; | |
| 3695 suggest_T *stp; | |
| 3696 char_u *wcopy; | |
| 3697 | |
| 638 | 3698 spell_find_suggest(word, 0, &sug, maxcount, FALSE, need_cap, interactive); |
| 344 | 3699 |
| 3700 /* Make room in "gap". */ | |
| 3701 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1); | |
| 626 | 3702 if (ga_grow(gap, sug.su_ga.ga_len) == OK) |
| 3703 { | |
| 3704 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 3705 { | |
| 3706 stp = &SUG(sug.su_ga, i); | |
| 3707 | |
| 3708 /* The suggested word may replace only part of "word", add the not | |
| 3709 * replaced part. */ | |
| 3710 wcopy = alloc(stp->st_wordlen | |
| 835 | 3711 + (unsigned)STRLEN(sug.su_badptr + stp->st_orglen) + 1); |
| 626 | 3712 if (wcopy == NULL) |
| 3713 break; | |
| 3714 STRCPY(wcopy, stp->st_word); | |
| 3715 STRCPY(wcopy + stp->st_wordlen, sug.su_badptr + stp->st_orglen); | |
| 3716 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy; | |
| 3717 } | |
| 344 | 3718 } |
| 3719 | |
| 3720 spell_find_cleanup(&sug); | |
| 3721 } | |
| 3722 | |
| 3723 /* | |
| 3724 * Find spell suggestions for the word at the start of "badptr". | |
| 3725 * Return the suggestions in "su->su_ga". | |
| 3726 * The maximum number of suggestions is "maxcount". | |
| 3727 * Note: does use info for the current window. | |
| 3728 * This is based on the mechanisms of Aspell, but completely reimplemented. | |
| 3729 */ | |
| 3730 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3731 spell_find_suggest( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3732 char_u *badptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3733 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
|
3734 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3735 int maxcount, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3736 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
|
3737 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
|
3738 int interactive) |
| 344 | 3739 { |
| 534 | 3740 hlf_T attr = HLF_COUNT; |
| 372 | 3741 char_u buf[MAXPATHL]; |
| 3742 char_u *p; | |
| 3743 int do_combine = FALSE; | |
| 3744 char_u *sps_copy; | |
| 3745 #ifdef FEAT_EVAL | |
| 3746 static int expr_busy = FALSE; | |
| 3747 #endif | |
| 385 | 3748 int c; |
| 503 | 3749 int i; |
| 3750 langp_T *lp; | |
| 344 | 3751 |
| 3752 /* | |
| 3753 * Set the info in "*su". | |
| 3754 */ | |
| 3755 vim_memset(su, 0, sizeof(suginfo_T)); | |
| 3756 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10); | |
| 3757 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10); | |
| 355 | 3758 if (*badptr == NUL) |
| 3759 return; | |
| 344 | 3760 hash_init(&su->su_banned); |
| 3761 | |
| 3762 su->su_badptr = badptr; | |
| 638 | 3763 if (badlen != 0) |
| 3764 su->su_badlen = badlen; | |
| 3765 else | |
| 3766 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL, FALSE); | |
| 344 | 3767 su->su_maxcount = maxcount; |
| 372 | 3768 su->su_maxscore = SCORE_MAXINIT; |
| 344 | 3769 |
| 3770 if (su->su_badlen >= MAXWLEN) | |
| 3771 su->su_badlen = MAXWLEN - 1; /* just in case */ | |
| 3772 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen); | |
| 3773 (void)spell_casefold(su->su_badptr, su->su_badlen, | |
| 3774 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
|
3775 /* 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
|
3776 * 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
|
3777 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
|
3778 |
| 346 | 3779 /* get caps flags for bad word */ |
| 474 | 3780 su->su_badflags = badword_captype(su->su_badptr, |
| 3781 su->su_badptr + su->su_badlen); | |
| 388 | 3782 if (need_cap) |
| 3783 su->su_badflags |= WF_ONECAP; | |
| 344 | 3784 |
| 503 | 3785 /* Find the default language for sound folding. We simply use the first |
| 3786 * one in 'spelllang' that supports sound folding. That's good for when | |
| 3787 * using multiple files for one language, it's not that bad when mixing | |
| 3788 * languages (e.g., "pl,en"). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3789 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
|
3790 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3791 lp = LANGP_ENTRY(curbuf->b_s.b_langp, i); |
| 503 | 3792 if (lp->lp_sallang != NULL) |
| 3793 { | |
| 3794 su->su_sallang = lp->lp_sallang; | |
| 3795 break; | |
| 3796 } | |
| 3797 } | |
| 3798 | |
| 534 | 3799 /* Soundfold the bad word with the default sound folding, so that we don't |
| 3800 * have to do this many times. */ | |
| 3801 if (su->su_sallang != NULL) | |
| 3802 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE, | |
| 3803 su->su_sal_badword); | |
| 3804 | |
| 385 | 3805 /* If the word is not capitalised and spell_check() doesn't consider the |
| 3806 * word to be bad then it might need to be capitalised. Add a suggestion | |
| 3807 * for that. */ | |
| 455 | 3808 c = PTR2CHAR(su->su_badptr); |
| 534 | 3809 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT) |
| 385 | 3810 { |
| 3811 make_case_word(su->su_badword, buf, WF_ONECAP); | |
| 3812 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE, | |
| 625 | 3813 0, TRUE, su->su_sallang, FALSE); |
| 385 | 3814 } |
| 3815 | |
| 344 | 3816 /* Ban the bad word itself. It may appear in another region. */ |
| 358 | 3817 if (banbadword) |
| 3818 add_banned(su, su->su_badword); | |
| 344 | 3819 |
| 372 | 3820 /* Make a copy of 'spellsuggest', because the expression may change it. */ |
| 3821 sps_copy = vim_strsave(p_sps); | |
| 3822 if (sps_copy == NULL) | |
| 3823 return; | |
| 3824 | |
| 3825 /* Loop over the items in 'spellsuggest'. */ | |
| 3826 for (p = sps_copy; *p != NUL; ) | |
| 3827 { | |
| 3828 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 3829 | |
| 3830 if (STRNCMP(buf, "expr:", 5) == 0) | |
| 3831 { | |
| 3832 #ifdef FEAT_EVAL | |
| 375 | 3833 /* Evaluate an expression. Skip this when called recursively, |
| 3834 * when using spellsuggest() in the expression. */ | |
| 372 | 3835 if (!expr_busy) |
| 3836 { | |
| 3837 expr_busy = TRUE; | |
| 3838 spell_suggest_expr(su, buf + 5); | |
| 3839 expr_busy = FALSE; | |
| 3840 } | |
| 3841 #endif | |
| 3842 } | |
| 3843 else if (STRNCMP(buf, "file:", 5) == 0) | |
| 3844 /* Use list of suggestions in a file. */ | |
| 3845 spell_suggest_file(su, buf + 5); | |
| 3846 else | |
| 3847 { | |
| 3848 /* Use internal method. */ | |
| 625 | 3849 spell_suggest_intern(su, interactive); |
| 372 | 3850 if (sps_flags & SPS_DOUBLE) |
| 3851 do_combine = TRUE; | |
| 3852 } | |
| 3853 } | |
| 3854 | |
| 3855 vim_free(sps_copy); | |
| 3856 | |
| 3857 if (do_combine) | |
| 3858 /* Combine the two list of suggestions. This must be done last, | |
| 3859 * because sorting changes the order again. */ | |
| 3860 score_combine(su); | |
| 3861 } | |
| 3862 | |
| 3863 #ifdef FEAT_EVAL | |
| 3864 /* | |
| 3865 * Find suggestions by evaluating expression "expr". | |
| 3866 */ | |
| 3867 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3868 spell_suggest_expr(suginfo_T *su, char_u *expr) |
| 372 | 3869 { |
| 3870 list_T *list; | |
| 3871 listitem_T *li; | |
| 3872 int score; | |
| 3873 char_u *p; | |
| 3874 | |
| 3875 /* The work is split up in a few parts to avoid having to export | |
| 3876 * suginfo_T. | |
| 3877 * First evaluate the expression and get the resulting list. */ | |
| 3878 list = eval_spell_expr(su->su_badword, expr); | |
| 3879 if (list != NULL) | |
| 3880 { | |
| 3881 /* Loop over the items in the list. */ | |
| 3882 for (li = list->lv_first; li != NULL; li = li->li_next) | |
| 3883 if (li->li_tv.v_type == VAR_LIST) | |
| 3884 { | |
| 3885 /* Get the word and the score from the items. */ | |
| 3886 score = get_spellword(li->li_tv.vval.v_list, &p); | |
| 625 | 3887 if (score >= 0 && score <= su->su_maxscore) |
| 3888 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 3889 score, 0, TRUE, su->su_sallang, FALSE); | |
| 372 | 3890 } |
| 3891 list_unref(list); | |
| 3892 } | |
| 3893 | |
| 625 | 3894 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3895 check_suggestions(su, &su->su_ga); | |
| 372 | 3896 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 3897 } | |
| 3898 #endif | |
| 3899 | |
| 3900 /* | |
| 474 | 3901 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'. |
| 372 | 3902 */ |
| 3903 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3904 spell_suggest_file(suginfo_T *su, char_u *fname) |
| 372 | 3905 { |
| 3906 FILE *fd; | |
| 3907 char_u line[MAXWLEN * 2]; | |
| 3908 char_u *p; | |
| 3909 int len; | |
| 3910 char_u cword[MAXWLEN]; | |
| 3911 | |
| 3912 /* Open the file. */ | |
| 3913 fd = mch_fopen((char *)fname, "r"); | |
| 3914 if (fd == NULL) | |
| 3915 { | |
| 3916 EMSG2(_(e_notopen), fname); | |
| 3917 return; | |
| 3918 } | |
| 3919 | |
| 3920 /* Read it line by line. */ | |
| 3921 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int) | |
| 3922 { | |
| 3923 line_breakcheck(); | |
| 3924 | |
| 3925 p = vim_strchr(line, '/'); | |
| 3926 if (p == NULL) | |
| 3927 continue; /* No Tab found, just skip the line. */ | |
| 3928 *p++ = NUL; | |
| 3929 if (STRICMP(su->su_badword, line) == 0) | |
| 3930 { | |
| 3931 /* Match! Isolate the good word, until CR or NL. */ | |
| 3932 for (len = 0; p[len] >= ' '; ++len) | |
| 3933 ; | |
| 3934 p[len] = NUL; | |
| 3935 | |
| 3936 /* If the suggestion doesn't have specific case duplicate the case | |
| 3937 * of the bad word. */ | |
| 3938 if (captype(p, NULL) == 0) | |
| 3939 { | |
| 3940 make_case_word(p, cword, su->su_badflags); | |
| 3941 p = cword; | |
| 3942 } | |
| 3943 | |
| 3944 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 625 | 3945 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE); |
| 372 | 3946 } |
| 3947 } | |
| 3948 | |
| 3949 fclose(fd); | |
| 3950 | |
| 625 | 3951 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3952 check_suggestions(su, &su->su_ga); | |
| 372 | 3953 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 3954 } | |
| 3955 | |
| 3956 /* | |
| 3957 * Find suggestions for the internal method indicated by "sps_flags". | |
| 3958 */ | |
| 3959 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3960 spell_suggest_intern(suginfo_T *su, int interactive) |
| 625 | 3961 { |
| 3962 /* | |
| 3963 * Load the .sug file(s) that are available and not done yet. | |
| 3964 */ | |
| 3965 suggest_load_files(); | |
| 3966 | |
| 344 | 3967 /* |
| 346 | 3968 * 1. Try special cases, such as repeating a word: "the the" -> "the". |
| 344 | 3969 * |
| 3970 * Set a maximum score to limit the combination of operations that is | |
| 3971 * tried. | |
| 3972 */ | |
| 346 | 3973 suggest_try_special(su); |
| 3974 | |
| 3975 /* | |
| 3976 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries | |
| 3977 * from the .aff file and inserting a space (split the word). | |
| 3978 */ | |
| 3979 suggest_try_change(su); | |
| 344 | 3980 |
| 3981 /* For the resulting top-scorers compute the sound-a-like score. */ | |
| 3982 if (sps_flags & SPS_DOUBLE) | |
| 3983 score_comp_sal(su); | |
| 3984 | |
| 3985 /* | |
| 346 | 3986 * 3. Try finding sound-a-like words. |
| 344 | 3987 */ |
| 625 | 3988 if ((sps_flags & SPS_FAST) == 0) |
| 3989 { | |
| 3990 if (sps_flags & SPS_BEST) | |
| 3991 /* Adjust the word score for the suggestions found so far for how | |
| 3992 * they sounds like. */ | |
| 3993 rescore_suggestions(su); | |
| 3994 | |
| 3995 /* | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3996 * While going through the soundfold tree "su_maxscore" is the score |
| 625 | 3997 * for the soundfold word, limits the changes that are being tried, |
| 3998 * and "su_sfmaxscore" the rescored score, which is set by | |
| 3999 * cleanup_suggestions(). | |
| 4000 * First find words with a small edit distance, because this is much | |
| 4001 * faster and often already finds the top-N suggestions. If we didn't | |
| 4002 * find many suggestions try again with a higher edit distance. | |
| 4003 * "sl_sounddone" is used to avoid doing the same word twice. | |
| 4004 */ | |
| 4005 suggest_try_soundalike_prep(); | |
| 4006 su->su_maxscore = SCORE_SFMAX1; | |
| 4007 su->su_sfmaxscore = SCORE_MAXINIT * 3; | |
| 346 | 4008 suggest_try_soundalike(su); |
| 625 | 4009 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) |
| 4010 { | |
| 4011 /* We didn't find enough matches, try again, allowing more | |
| 4012 * changes to the soundfold word. */ | |
| 4013 su->su_maxscore = SCORE_SFMAX2; | |
| 4014 suggest_try_soundalike(su); | |
| 4015 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) | |
| 4016 { | |
| 4017 /* Still didn't find enough matches, try again, allowing even | |
| 4018 * more changes to the soundfold word. */ | |
| 4019 su->su_maxscore = SCORE_SFMAX3; | |
| 4020 suggest_try_soundalike(su); | |
| 4021 } | |
| 4022 } | |
| 4023 su->su_maxscore = su->su_sfmaxscore; | |
| 4024 suggest_try_soundalike_finish(); | |
| 4025 } | |
| 4026 | |
| 4027 /* When CTRL-C was hit while searching do show the results. Only clear | |
| 4028 * got_int when using a command, not for spellsuggest(). */ | |
| 344 | 4029 ui_breakcheck(); |
| 625 | 4030 if (interactive && got_int) |
| 344 | 4031 { |
| 4032 (void)vgetc(); | |
| 4033 got_int = FALSE; | |
| 4034 } | |
| 4035 | |
| 372 | 4036 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0) |
| 344 | 4037 { |
| 4038 if (sps_flags & SPS_BEST) | |
| 4039 /* Adjust the word score for how it sounds like. */ | |
| 4040 rescore_suggestions(su); | |
| 4041 | |
| 625 | 4042 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 4043 check_suggestions(su, &su->su_ga); | |
| 372 | 4044 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 344 | 4045 } |
| 4046 } | |
| 4047 | |
| 4048 /* | |
| 4049 * Free the info put in "*su" by spell_find_suggest(). | |
| 4050 */ | |
| 4051 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4052 spell_find_cleanup(suginfo_T *su) |
| 344 | 4053 { |
| 4054 int i; | |
| 4055 | |
| 4056 /* Free the suggestions. */ | |
| 4057 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 4058 vim_free(SUG(su->su_ga, i).st_word); | |
| 4059 ga_clear(&su->su_ga); | |
| 4060 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 4061 vim_free(SUG(su->su_sga, i).st_word); | |
| 4062 ga_clear(&su->su_sga); | |
| 323 | 4063 |
| 4064 /* Free the banned words. */ | |
| 625 | 4065 hash_clear_all(&su->su_banned, 0); |
| 323 | 4066 } |
| 4067 | |
| 4068 /* | |
| 324 | 4069 * Make a copy of "word", with the first letter upper or lower cased, to |
| 4070 * "wcopy[MAXWLEN]". "word" must not be empty. | |
| 4071 * The result is NUL terminated. | |
| 323 | 4072 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
4073 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4074 onecap_copy( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4075 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4076 char_u *wcopy, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4077 int upper) /* TRUE: first letter made upper case */ |
| 323 | 4078 { |
| 4079 char_u *p; | |
| 4080 int c; | |
| 4081 int l; | |
| 4082 | |
| 4083 p = word; | |
| 4084 #ifdef FEAT_MBYTE | |
| 4085 if (has_mbyte) | |
| 474 | 4086 c = mb_cptr2char_adv(&p); |
| 323 | 4087 else |
| 4088 #endif | |
| 4089 c = *p++; | |
| 4090 if (upper) | |
| 324 | 4091 c = SPELL_TOUPPER(c); |
| 323 | 4092 else |
| 324 | 4093 c = SPELL_TOFOLD(c); |
| 323 | 4094 #ifdef FEAT_MBYTE |
| 4095 if (has_mbyte) | |
| 4096 l = mb_char2bytes(c, wcopy); | |
| 4097 else | |
| 4098 #endif | |
| 4099 { | |
| 4100 l = 1; | |
| 4101 wcopy[0] = c; | |
| 4102 } | |
| 376 | 4103 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1); |
| 323 | 4104 } |
| 4105 | |
| 4106 /* | |
| 324 | 4107 * Make a copy of "word" with all the letters upper cased into |
| 4108 * "wcopy[MAXWLEN]". The result is NUL terminated. | |
| 323 | 4109 */ |
| 4110 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4111 allcap_copy(char_u *word, char_u *wcopy) |
| 323 | 4112 { |
| 4113 char_u *s; | |
| 4114 char_u *d; | |
| 4115 int c; | |
| 4116 | |
| 4117 d = wcopy; | |
| 4118 for (s = word; *s != NUL; ) | |
| 4119 { | |
| 4120 #ifdef FEAT_MBYTE | |
| 4121 if (has_mbyte) | |
| 474 | 4122 c = mb_cptr2char_adv(&s); |
| 323 | 4123 else |
| 4124 #endif | |
| 4125 c = *s++; | |
| 492 | 4126 |
| 4127 #ifdef FEAT_MBYTE | |
| 3042 | 4128 /* We only change 0xdf to SS when we are certain latin1 is used. It |
| 492 | 4129 * would cause weird errors in other 8-bit encodings. */ |
| 4130 if (enc_latin1like && c == 0xdf) | |
| 4131 { | |
| 4132 c = 'S'; | |
| 4133 if (d - wcopy >= MAXWLEN - 1) | |
| 4134 break; | |
| 4135 *d++ = c; | |
| 4136 } | |
| 4137 else | |
| 4138 #endif | |
| 4139 c = SPELL_TOUPPER(c); | |
| 323 | 4140 |
| 4141 #ifdef FEAT_MBYTE | |
| 4142 if (has_mbyte) | |
| 4143 { | |
| 4144 if (d - wcopy >= MAXWLEN - MB_MAXBYTES) | |
| 4145 break; | |
| 4146 d += mb_char2bytes(c, d); | |
| 4147 } | |
| 4148 else | |
| 4149 #endif | |
| 4150 { | |
| 4151 if (d - wcopy >= MAXWLEN - 1) | |
| 4152 break; | |
| 4153 *d++ = c; | |
| 4154 } | |
| 4155 } | |
| 4156 *d = NUL; | |
| 4157 } | |
| 4158 | |
| 4159 /* | |
| 346 | 4160 * Try finding suggestions by recognizing specific situations. |
| 4161 */ | |
| 4162 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4163 suggest_try_special(suginfo_T *su) |
| 346 | 4164 { |
| 4165 char_u *p; | |
| 376 | 4166 size_t len; |
| 346 | 4167 int c; |
| 4168 char_u word[MAXWLEN]; | |
| 4169 | |
| 4170 /* | |
| 4171 * Recognize a word that is repeated: "the the". | |
| 4172 */ | |
| 4173 p = skiptowhite(su->su_fbadword); | |
| 4174 len = p - su->su_fbadword; | |
| 4175 p = skipwhite(p); | |
| 4176 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0) | |
| 4177 { | |
| 4178 /* Include badflags: if the badword is onecap or allcap | |
| 4179 * use that for the goodword too: "The the" -> "The". */ | |
| 4180 c = su->su_fbadword[len]; | |
| 4181 su->su_fbadword[len] = NUL; | |
| 4182 make_case_word(su->su_fbadword, word, su->su_badflags); | |
| 4183 su->su_fbadword[len] = c; | |
| 534 | 4184 |
| 4185 /* Give a soundalike score of 0, compute the score as if deleting one | |
| 4186 * character. */ | |
| 4187 add_suggestion(su, &su->su_ga, word, su->su_badlen, | |
| 625 | 4188 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE); |
| 346 | 4189 } |
| 4190 } | |
| 4191 | |
| 4192 /* | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4193 * 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
|
4194 * Output is dumped in "suggestprof". |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4195 */ |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4196 #if 0 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4197 # define SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4198 proftime_T current; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4199 proftime_T total; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4200 proftime_T times[STATE_FINAL + 1]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4201 long counts[STATE_FINAL + 1]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4202 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4203 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4204 prof_init(void) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4205 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4206 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
|
4207 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4208 profile_zero(×[i]); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4209 counts[i] = 0; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4210 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4211 profile_start(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4212 profile_start(&total); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4213 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4214 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4215 /* call before changing state */ |
|
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_store(state_T state) |
|
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 profile_end(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4220 profile_add(×[state], ¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4221 ++counts[state]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4222 profile_start(¤t); |
|
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 # 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
|
4225 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4226 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4227 prof_report(char *name) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4228 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4229 FILE *fd = fopen("suggestprof", "a"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4230 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4231 profile_end(&total); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4232 fprintf(fd, "-----------------------\n"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4233 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
|
4234 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
|
4235 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
|
4236 fclose(fd); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4237 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4238 #else |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4239 # define PROF_STORE(state) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4240 #endif |
|
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 /* |
| 323 | 4243 * Try finding suggestions by adding/removing/swapping letters. |
| 4244 */ | |
| 4245 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4246 suggest_try_change(suginfo_T *su) |
| 323 | 4247 { |
| 4248 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */ | |
| 625 | 4249 int n; |
| 4250 char_u *p; | |
| 4251 int lpi; | |
| 323 | 4252 langp_T *lp; |
| 4253 | |
| 4254 /* We make a copy of the case-folded bad word, so that we can modify it | |
| 346 | 4255 * to find matches (esp. REP items). Append some more text, changing |
| 4256 * chars after the bad word may help. */ | |
| 323 | 4257 STRCPY(fword, su->su_fbadword); |
| 835 | 4258 n = (int)STRLEN(fword); |
| 346 | 4259 p = su->su_badptr + su->su_badlen; |
| 835 | 4260 (void)spell_casefold(p, (int)STRLEN(p), fword + n, MAXWLEN - n); |
| 323 | 4261 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4262 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
|
4263 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4264 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 482 | 4265 |
| 500 | 4266 /* If reloading a spell file fails it's still in the list but |
| 4267 * everything has been cleared. */ | |
| 625 | 4268 if (lp->lp_slang->sl_fbyts == NULL) |
| 500 | 4269 continue; |
| 4270 | |
| 625 | 4271 /* 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
|
4272 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4273 prof_init(); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4274 #endif |
| 625 | 4275 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
|
4276 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4277 prof_report("try_change"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4278 #endif |
| 625 | 4279 } |
| 4280 } | |
| 4281 | |
| 4282 /* Check the maximum score, if we go over it we won't try this change. */ | |
| 4283 #define TRY_DEEPER(su, stack, depth, add) \ | |
| 4284 (stack[depth].ts_score + (add) < su->su_maxscore) | |
| 4285 | |
| 4286 /* | |
| 4287 * Try finding suggestions by adding/removing/swapping letters. | |
| 4288 * | |
| 4289 * This uses a state machine. At each node in the tree we try various | |
| 4290 * operations. When trying if an operation works "depth" is increased and the | |
| 4291 * stack[] is used to store info. This allows combinations, thus insert one | |
| 4292 * character, replace one and delete another. The number of changes is | |
| 4293 * limited by su->su_maxscore. | |
| 4294 * | |
| 4295 * After implementing this I noticed an article by Kemal Oflazer that | |
| 4296 * describes something similar: "Error-tolerant Finite State Recognition with | |
| 4297 * Applications to Morphological Analysis and Spelling Correction" (1996). | |
| 4298 * The implementation in the article is simplified and requires a stack of | |
| 4299 * unknown depth. The implementation here only needs a stack depth equal to | |
| 4300 * the length of the word. | |
| 4301 * | |
| 4302 * This is also used for the sound-folded word, "soundfold" is TRUE then. | |
| 4303 * The mechanism is the same, but we find a match with a sound-folded word | |
| 4304 * that comes from one or more original words. Each of these words may be | |
| 4305 * added, this is done by add_sound_suggest(). | |
| 4306 * Don't use: | |
| 4307 * the prefix tree or the keep-case tree | |
| 4308 * "su->su_badlen" | |
| 4309 * anything to do with upper and lower case | |
| 4310 * anything to do with word or non-word characters ("spell_iswordp()") | |
| 4311 * banned words | |
| 4312 * word flags (rare, region, compounding) | |
| 4313 * word splitting for now | |
| 4314 * "similar_chars()" | |
| 4315 * use "slang->sl_repsal" instead of "lp->lp_replang->sl_rep" | |
| 4316 */ | |
| 4317 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4318 suggest_trie_walk( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4319 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4320 langp_T *lp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4321 char_u *fword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4322 int soundfold) |
| 625 | 4323 { |
| 4324 char_u tword[MAXWLEN]; /* good word collected so far */ | |
| 4325 trystate_T stack[MAXWLEN]; | |
| 4326 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
|
4327 * concatenation of prefix compound |
| 625 | 4328 * words and split word. NUL terminated |
| 4329 * when going deeper but not when coming | |
| 4330 * back. */ | |
| 4331 char_u compflags[MAXWLEN]; /* compound flags, one for each word */ | |
| 4332 trystate_T *sp; | |
| 4333 int newscore; | |
| 4334 int score; | |
| 4335 char_u *byts, *fbyts, *pbyts; | |
| 4336 idx_T *idxs, *fidxs, *pidxs; | |
| 4337 int depth; | |
| 4338 int c, c2, c3; | |
| 4339 int n = 0; | |
| 4340 int flags; | |
| 4341 garray_T *gap; | |
| 4342 idx_T arridx; | |
| 4343 int len; | |
| 4344 char_u *p; | |
| 4345 fromto_T *ftp; | |
| 4346 int fl = 0, tl; | |
| 4347 int repextra = 0; /* extra bytes in fword[] from REP item */ | |
| 4348 slang_T *slang = lp->lp_slang; | |
| 4349 int fword_ends; | |
| 4350 int goodword_ends; | |
| 4351 #ifdef DEBUG_TRIEWALK | |
| 4352 /* Stores the name of the change made at each level. */ | |
| 4353 char_u changename[MAXWLEN][80]; | |
| 4354 #endif | |
| 4355 int breakcheckcount = 1000; | |
| 4356 int compound_ok; | |
| 4357 | |
| 4358 /* | |
| 4359 * Go through the whole case-fold tree, try changes at each node. | |
| 4360 * "tword[]" contains the word collected from nodes in the tree. | |
| 4361 * "fword[]" the word we are trying to match with (initially the bad | |
| 4362 * word). | |
| 4363 */ | |
| 4364 depth = 0; | |
| 4365 sp = &stack[0]; | |
| 4366 vim_memset(sp, 0, sizeof(trystate_T)); | |
| 4367 sp->ts_curi = 1; | |
| 4368 | |
| 4369 if (soundfold) | |
| 4370 { | |
| 4371 /* Going through the soundfold tree. */ | |
| 4372 byts = fbyts = slang->sl_sbyts; | |
| 4373 idxs = fidxs = slang->sl_sidxs; | |
| 4374 pbyts = NULL; | |
| 4375 pidxs = NULL; | |
| 4376 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 4377 sp->ts_state = STATE_START; | |
| 4378 } | |
| 4379 else | |
| 4380 { | |
| 330 | 4381 /* |
| 375 | 4382 * When there are postponed prefixes we need to use these first. At |
| 4383 * the end of the prefix we continue in the case-fold tree. | |
| 4384 */ | |
| 482 | 4385 fbyts = slang->sl_fbyts; |
| 4386 fidxs = slang->sl_fidxs; | |
| 4387 pbyts = slang->sl_pbyts; | |
| 4388 pidxs = slang->sl_pidxs; | |
| 375 | 4389 if (pbyts != NULL) |
| 4390 { | |
| 4391 byts = pbyts; | |
| 4392 idxs = pidxs; | |
| 482 | 4393 sp->ts_prefixdepth = PFD_PREFIXTREE; |
| 375 | 4394 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */ |
| 4395 } | |
| 4396 else | |
| 4397 { | |
| 4398 byts = fbyts; | |
| 4399 idxs = fidxs; | |
| 482 | 4400 sp->ts_prefixdepth = PFD_NOPREFIX; |
| 485 | 4401 sp->ts_state = STATE_START; |
| 375 | 4402 } |
| 625 | 4403 } |
| 4404 | |
| 4405 /* | |
| 4406 * Loop to find all suggestions. At each round we either: | |
| 4407 * - For the current state try one operation, advance "ts_curi", | |
| 4408 * increase "depth". | |
| 4409 * - When a state is done go to the next, set "ts_state". | |
| 4410 * - When all states are tried decrease "depth". | |
| 4411 */ | |
| 4412 while (depth >= 0 && !got_int) | |
| 4413 { | |
| 4414 sp = &stack[depth]; | |
| 4415 switch (sp->ts_state) | |
| 4416 { | |
| 4417 case STATE_START: | |
| 4418 case STATE_NOPREFIX: | |
| 4419 /* | |
| 4420 * Start of node: Deal with NUL bytes, which means | |
| 4421 * tword[] may end here. | |
| 4422 */ | |
| 4423 arridx = sp->ts_arridx; /* current node in the tree */ | |
| 4424 len = byts[arridx]; /* bytes in this node */ | |
| 4425 arridx += sp->ts_curi; /* index of current byte */ | |
| 4426 | |
| 4427 if (sp->ts_prefixdepth == PFD_PREFIXTREE) | |
| 4428 { | |
| 4429 /* Skip over the NUL bytes, we use them later. */ | |
| 4430 for (n = 0; n < len && byts[arridx + n] == 0; ++n) | |
| 4431 ; | |
| 4432 sp->ts_curi += n; | |
| 4433 | |
| 4434 /* Always past NUL bytes now. */ | |
| 4435 n = (int)sp->ts_state; | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4436 PROF_STORE(sp->ts_state) |
| 625 | 4437 sp->ts_state = STATE_ENDNUL; |
| 4438 sp->ts_save_badflags = su->su_badflags; | |
| 4439 | |
| 4440 /* At end of a prefix or at start of prefixtree: check for | |
| 4441 * following word. */ | |
| 4442 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX) | |
| 4443 { | |
| 4444 /* Set su->su_badflags to the caps type at this position. | |
| 4445 * Use the caps type until here for the prefix itself. */ | |
| 4446 #ifdef FEAT_MBYTE | |
| 4447 if (has_mbyte) | |
| 4448 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); | |
| 4449 else | |
| 4450 #endif | |
| 4451 n = sp->ts_fidx; | |
| 4452 flags = badword_captype(su->su_badptr, su->su_badptr + n); | |
| 4453 su->su_badflags = badword_captype(su->su_badptr + n, | |
| 455 | 4454 su->su_badptr + su->su_badlen); |
| 625 | 4455 #ifdef DEBUG_TRIEWALK |
| 4456 sprintf(changename[depth], "prefix"); | |
| 4457 #endif | |
| 4458 go_deeper(stack, depth, 0); | |
| 4459 ++depth; | |
| 4460 sp = &stack[depth]; | |
| 4461 sp->ts_prefixdepth = depth - 1; | |
| 4462 byts = fbyts; | |
| 4463 idxs = fidxs; | |
| 4464 sp->ts_arridx = 0; | |
| 4465 | |
| 4466 /* Move the prefix to preword[] with the right case | |
| 4467 * and make find_keepcap_word() works. */ | |
| 4468 tword[sp->ts_twordlen] = NUL; | |
| 4469 make_case_word(tword + sp->ts_splitoff, | |
| 4470 preword + sp->ts_prewordlen, flags); | |
| 835 | 4471 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 625 | 4472 sp->ts_splitoff = sp->ts_twordlen; |
| 4473 } | |
| 4474 break; | |
| 4475 } | |
| 4476 | |
| 4477 if (sp->ts_curi > len || byts[arridx] != 0) | |
| 4478 { | |
| 4479 /* 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
|
4480 PROF_STORE(sp->ts_state) |
| 625 | 4481 sp->ts_state = STATE_ENDNUL; |
| 4482 sp->ts_save_badflags = su->su_badflags; | |
| 4483 break; | |
| 4484 } | |
| 4485 | |
| 4486 /* | |
| 4487 * End of word in tree. | |
| 4488 */ | |
| 4489 ++sp->ts_curi; /* eat one NUL byte */ | |
| 4490 | |
| 4491 flags = (int)idxs[arridx]; | |
| 698 | 4492 |
| 4493 /* Skip words with the NOSUGGEST flag. */ | |
| 4494 if (flags & WF_NOSUGGEST) | |
| 4495 break; | |
| 4496 | |
| 625 | 4497 fword_ends = (fword[sp->ts_fidx] == NUL |
| 4498 || (soundfold | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
4499 ? VIM_ISWHITE(fword[sp->ts_fidx]) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4500 : !spell_iswordp(fword + sp->ts_fidx, curwin))); |
| 625 | 4501 tword[sp->ts_twordlen] = NUL; |
| 4502 | |
| 4503 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL | |
| 485 | 4504 && (sp->ts_flags & TSF_PREFIXOK) == 0) |
| 625 | 4505 { |
| 4506 /* There was a prefix before the word. Check that the prefix | |
| 4507 * can be used with this word. */ | |
| 4508 /* Count the length of the NULs in the prefix. If there are | |
| 4509 * none this must be the first try without a prefix. */ | |
| 4510 n = stack[sp->ts_prefixdepth].ts_arridx; | |
| 4511 len = pbyts[n++]; | |
| 4512 for (c = 0; c < len && pbyts[n + c] == 0; ++c) | |
| 4513 ; | |
| 4514 if (c > 0) | |
| 4515 { | |
| 4516 c = valid_word_prefix(c, n, flags, | |
| 483 | 4517 tword + sp->ts_splitoff, slang, FALSE); |
| 625 | 4518 if (c == 0) |
| 4519 break; | |
| 4520 | |
| 4521 /* Use the WF_RARE flag for a rare prefix. */ | |
| 4522 if (c & WF_RAREPFX) | |
| 4523 flags |= WF_RARE; | |
| 4524 | |
| 4525 /* Tricky: when checking for both prefix and compounding | |
| 4526 * we run into the prefix flag first. | |
| 4527 * Remember that it's OK, so that we accept the prefix | |
| 4528 * when arriving at a compound flag. */ | |
| 4529 sp->ts_flags |= TSF_PREFIXOK; | |
| 4530 } | |
| 4531 } | |
| 4532 | |
| 4533 /* Check NEEDCOMPOUND: can't use word without compounding. Do try | |
| 4534 * appending another compound word below. */ | |
| 4535 if (sp->ts_complen == sp->ts_compsplit && fword_ends | |
| 500 | 4536 && (flags & WF_NEEDCOMP)) |
| 625 | 4537 goodword_ends = FALSE; |
| 4538 else | |
| 4539 goodword_ends = TRUE; | |
| 4540 | |
| 4541 p = NULL; | |
| 4542 compound_ok = TRUE; | |
| 4543 if (sp->ts_complen > sp->ts_compsplit) | |
| 4544 { | |
| 4545 if (slang->sl_nobreak) | |
| 4546 { | |
| 4547 /* There was a word before this word. When there was no | |
| 4548 * change in this word (it was correct) add the first word | |
| 4549 * as a suggestion. If this word was corrected too, we | |
| 4550 * need to check if a correct word follows. */ | |
| 4551 if (sp->ts_fidx - sp->ts_splitfidx | |
| 4552 == sp->ts_twordlen - sp->ts_splitoff | |
| 4553 && STRNCMP(fword + sp->ts_splitfidx, | |
| 4554 tword + sp->ts_splitoff, | |
| 4555 sp->ts_fidx - sp->ts_splitfidx) == 0) | |
| 492 | 4556 { |
| 625 | 4557 preword[sp->ts_prewordlen] = NUL; |
| 4558 newscore = score_wordcount_adj(slang, sp->ts_score, | |
| 4559 preword + sp->ts_prewordlen, | |
| 4560 sp->ts_prewordlen > 0); | |
| 4561 /* Add the suggestion if the score isn't too bad. */ | |
| 4562 if (newscore <= su->su_maxscore) | |
| 492 | 4563 add_suggestion(su, &su->su_ga, preword, |
| 503 | 4564 sp->ts_splitfidx - repextra, |
| 625 | 4565 newscore, 0, FALSE, |
| 4566 lp->lp_sallang, FALSE); | |
| 4567 break; | |
| 492 | 4568 } |
| 490 | 4569 } |
| 4570 else | |
| 625 | 4571 { |
| 4572 /* There was a compound word before this word. If this | |
| 4573 * word does not support compounding then give up | |
| 4574 * (splitting is tried for the word without compound | |
| 4575 * flag). */ | |
| 4576 if (((unsigned)flags >> 24) == 0 | |
| 4577 || sp->ts_twordlen - sp->ts_splitoff | |
| 4578 < slang->sl_compminlen) | |
| 4579 break; | |
| 4580 #ifdef FEAT_MBYTE | |
| 4581 /* For multi-byte chars check character length against | |
| 4582 * COMPOUNDMIN. */ | |
| 4583 if (has_mbyte | |
| 4584 && slang->sl_compminlen > 0 | |
| 4585 && mb_charlen(tword + sp->ts_splitoff) | |
| 4586 < slang->sl_compminlen) | |
| 4587 break; | |
| 4588 #endif | |
| 4589 | |
| 4590 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 4591 compflags[sp->ts_complen + 1] = NUL; | |
| 4592 vim_strncpy(preword + sp->ts_prewordlen, | |
| 4593 tword + sp->ts_splitoff, | |
| 4594 sp->ts_twordlen - sp->ts_splitoff); | |
| 1762 | 4595 |
| 4596 /* Verify CHECKCOMPOUNDPATTERN rules. */ | |
| 4597 if (match_checkcompoundpattern(preword, sp->ts_prewordlen, | |
| 4598 &slang->sl_comppat)) | |
| 4599 compound_ok = FALSE; | |
| 4600 | |
| 4601 if (compound_ok) | |
| 4602 { | |
| 4603 p = preword; | |
| 4604 while (*skiptowhite(p) != NUL) | |
| 4605 p = skipwhite(skiptowhite(p)); | |
| 4606 if (fword_ends && !can_compound(slang, p, | |
| 625 | 4607 compflags + sp->ts_compsplit)) |
| 1762 | 4608 /* Compound is not allowed. But it may still be |
| 4609 * possible if we add another (short) word. */ | |
| 4610 compound_ok = FALSE; | |
| 4611 } | |
| 625 | 4612 |
| 4613 /* Get pointer to last char of previous word. */ | |
| 4614 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
|
4615 MB_PTR_BACK(preword, p); |
| 625 | 4616 } |
| 4617 } | |
| 4618 | |
| 4619 /* | |
| 4620 * Form the word with proper case in preword. | |
| 4621 * If there is a word from a previous split, append. | |
| 4622 * For the soundfold tree don't change the case, simply append. | |
| 4623 */ | |
| 4624 if (soundfold) | |
| 4625 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff); | |
| 4626 else if (flags & WF_KEEPCAP) | |
| 4627 /* Must find the word in the keep-case tree. */ | |
| 4628 find_keepcap_word(slang, tword + sp->ts_splitoff, | |
| 483 | 4629 preword + sp->ts_prewordlen); |
| 625 | 4630 else |
| 4631 { | |
| 4632 /* Include badflags: If the badword is onecap or allcap | |
| 4633 * use that for the goodword too. But if the badword is | |
| 4634 * allcap and it's only one char long use onecap. */ | |
| 4635 c = su->su_badflags; | |
| 4636 if ((c & WF_ALLCAP) | |
| 4637 #ifdef FEAT_MBYTE | |
| 4638 && su->su_badlen == (*mb_ptr2len)(su->su_badptr) | |
| 346 | 4639 #else |
| 625 | 4640 && su->su_badlen == 1 |
| 4641 #endif | |
| 4642 ) | |
| 4643 c = WF_ONECAP; | |
| 4644 c |= flags; | |
| 4645 | |
| 4646 /* When appending a compound word after a word character don't | |
| 4647 * use Onecap. */ | |
| 5477 | 4648 if (p != NULL && spell_iswordp_nmw(p, curwin)) |
| 625 | 4649 c &= ~WF_ONECAP; |
| 4650 make_case_word(tword + sp->ts_splitoff, | |
| 490 | 4651 preword + sp->ts_prewordlen, c); |
| 625 | 4652 } |
| 4653 | |
| 4654 if (!soundfold) | |
| 4655 { | |
| 323 | 4656 /* Don't use a banned word. It may appear again as a good |
| 4657 * word, thus remember it. */ | |
| 4658 if (flags & WF_BANNED) | |
| 4659 { | |
| 483 | 4660 add_banned(su, preword + sp->ts_prewordlen); |
| 323 | 4661 break; |
| 4662 } | |
| 534 | 4663 if ((sp->ts_complen == sp->ts_compsplit |
| 625 | 4664 && WAS_BANNED(su, preword + sp->ts_prewordlen)) |
| 4665 || WAS_BANNED(su, preword)) | |
| 534 | 4666 { |
| 4667 if (slang->sl_compprog == NULL) | |
| 4668 break; | |
| 4669 /* the word so far was banned but we may try compounding */ | |
| 4670 goodword_ends = FALSE; | |
| 4671 } | |
| 625 | 4672 } |
| 4673 | |
| 4674 newscore = 0; | |
| 4675 if (!soundfold) /* soundfold words don't have flags */ | |
| 4676 { | |
| 323 | 4677 if ((flags & WF_REGION) |
| 390 | 4678 && (((unsigned)flags >> 16) & lp->lp_region) == 0) |
| 323 | 4679 newscore += SCORE_REGION; |
| 4680 if (flags & WF_RARE) | |
| 4681 newscore += SCORE_RARE; | |
| 4682 | |
| 346 | 4683 if (!spell_valid_case(su->su_badflags, |
| 483 | 4684 captype(preword + sp->ts_prewordlen, NULL))) |
| 323 | 4685 newscore += SCORE_ICASE; |
| 625 | 4686 } |
| 4687 | |
| 4688 /* TODO: how about splitting in the soundfold tree? */ | |
| 4689 if (fword_ends | |
| 4690 && goodword_ends | |
| 4691 && sp->ts_fidx >= sp->ts_fidxtry | |
| 4692 && compound_ok) | |
| 4693 { | |
| 4694 /* The badword also ends: add suggestions. */ | |
| 4695 #ifdef DEBUG_TRIEWALK | |
| 4696 if (soundfold && STRCMP(preword, "smwrd") == 0) | |
| 4697 { | |
| 4698 int j; | |
| 4699 | |
| 4700 /* print the stack of changes that brought us here */ | |
| 4701 smsg("------ %s -------", fword); | |
| 4702 for (j = 0; j < depth; ++j) | |
| 4703 smsg("%s", changename[j]); | |
| 4704 } | |
| 4705 #endif | |
| 4706 if (soundfold) | |
| 4707 { | |
| 4708 /* For soundfolded words we need to find the original | |
| 1188 | 4709 * words, the edit distance and then add them. */ |
| 625 | 4710 add_sound_suggest(su, preword, sp->ts_score, lp); |
| 4711 } | |
|
2457
6c1d10a85398
Avoid illegal memory access in spell suggestion. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2455
diff
changeset
|
4712 else if (sp->ts_fidx > 0) |
| 625 | 4713 { |
| 4714 /* Give a penalty when changing non-word char to word | |
| 4715 * char, e.g., "thes," -> "these". */ | |
| 366 | 4716 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
|
4717 MB_PTR_BACK(fword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4718 if (!spell_iswordp(p, curwin)) |
| 366 | 4719 { |
| 4720 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
|
4721 MB_PTR_BACK(preword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4722 if (spell_iswordp(p, curwin)) |
| 366 | 4723 newscore += SCORE_NONWORD; |
| 4724 } | |
| 4725 | |
| 625 | 4726 /* Give a bonus to words seen before. */ |
| 4727 score = score_wordcount_adj(slang, | |
| 4728 sp->ts_score + newscore, | |
| 4729 preword + sp->ts_prewordlen, | |
| 4730 sp->ts_prewordlen > 0); | |
| 4731 | |
| 4732 /* Add the suggestion if the score isn't too bad. */ | |
| 4733 if (score <= su->su_maxscore) | |
| 633 | 4734 { |
| 625 | 4735 add_suggestion(su, &su->su_ga, preword, |
| 4736 sp->ts_fidx - repextra, | |
| 4737 score, 0, FALSE, lp->lp_sallang, FALSE); | |
| 633 | 4738 |
| 4739 if (su->su_badflags & WF_MIXCAP) | |
| 4740 { | |
| 4741 /* We really don't know if the word should be | |
| 4742 * upper or lower case, add both. */ | |
| 4743 c = captype(preword, NULL); | |
| 4744 if (c == 0 || c == WF_ALLCAP) | |
| 4745 { | |
| 4746 make_case_word(tword + sp->ts_splitoff, | |
| 4747 preword + sp->ts_prewordlen, | |
| 4748 c == 0 ? WF_ALLCAP : 0); | |
| 4749 | |
| 4750 add_suggestion(su, &su->su_ga, preword, | |
| 4751 sp->ts_fidx - repextra, | |
| 4752 score + SCORE_ICASE, 0, FALSE, | |
| 4753 lp->lp_sallang, FALSE); | |
| 4754 } | |
| 4755 } | |
| 4756 } | |
| 625 | 4757 } |
| 4758 } | |
| 4759 | |
| 4760 /* | |
| 4761 * Try word split and/or compounding. | |
| 4762 */ | |
| 4763 if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends) | |
| 4764 #ifdef FEAT_MBYTE | |
| 4765 /* Don't split halfway a character. */ | |
| 4766 && (!has_mbyte || sp->ts_tcharlen == 0) | |
| 4767 #endif | |
| 4768 ) | |
| 4769 { | |
| 4770 int try_compound; | |
| 4771 int try_split; | |
| 4772 | |
| 4773 /* If past the end of the bad word don't try a split. | |
| 4774 * Otherwise try changing the next word. E.g., find | |
| 4775 * suggestions for "the the" where the second "the" is | |
| 4776 * different. It's done like a split. | |
| 4777 * TODO: word split for soundfold words */ | |
| 4778 try_split = (sp->ts_fidx - repextra < su->su_badlen) | |
| 4779 && !soundfold; | |
| 4780 | |
| 4781 /* Get here in several situations: | |
| 4782 * 1. The word in the tree ends: | |
| 4783 * If the word allows compounding try that. Otherwise try | |
| 4784 * a split by inserting a space. For both check that a | |
| 4785 * valid words starts at fword[sp->ts_fidx]. | |
| 4786 * For NOBREAK do like compounding to be able to check if | |
| 4787 * the next word is valid. | |
| 4788 * 2. The badword does end, but it was due to a change (e.g., | |
| 4789 * a swap). No need to split, but do check that the | |
| 4790 * following word is valid. | |
| 4791 * 3. The badword and the word in the tree end. It may still | |
| 4792 * be possible to compound another (short) word. | |
| 4793 */ | |
| 4794 try_compound = FALSE; | |
| 4795 if (!soundfold | |
|
7526
08c1f73efcde
commit https://github.com/vim/vim/commit/7b877b360532713dc21a0ff3d55a76ac02eaf573
Christian Brabandt <cb@256bit.org>
parents:
7509
diff
changeset
|
4796 && !slang->sl_nocompoundsugs |
| 625 | 4797 && slang->sl_compprog != NULL |
| 4798 && ((unsigned)flags >> 24) != 0 | |
| 4799 && sp->ts_twordlen - sp->ts_splitoff | |
| 4800 >= slang->sl_compminlen | |
| 4801 #ifdef FEAT_MBYTE | |
| 4802 && (!has_mbyte | |
| 4803 || slang->sl_compminlen == 0 | |
| 4804 || mb_charlen(tword + sp->ts_splitoff) | |
| 4805 >= slang->sl_compminlen) | |
| 4806 #endif | |
| 4807 && (slang->sl_compsylmax < MAXWLEN | |
| 4808 || sp->ts_complen + 1 - sp->ts_compsplit | |
| 4809 < slang->sl_compmax) | |
| 1762 | 4810 && (can_be_compound(sp, slang, |
| 4811 compflags, ((unsigned)flags >> 24)))) | |
| 4812 | |
| 625 | 4813 { |
| 4814 try_compound = TRUE; | |
| 4815 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 4816 compflags[sp->ts_complen + 1] = NUL; | |
| 4817 } | |
| 4818 | |
| 4819 /* For NOBREAK we never try splitting, it won't make any word | |
| 4820 * valid. */ | |
|
7526
08c1f73efcde
commit https://github.com/vim/vim/commit/7b877b360532713dc21a0ff3d55a76ac02eaf573
Christian Brabandt <cb@256bit.org>
parents:
7509
diff
changeset
|
4821 if (slang->sl_nobreak && !slang->sl_nocompoundsugs) |
| 625 | 4822 try_compound = TRUE; |
| 4823 | |
| 4824 /* If we could add a compound word, and it's also possible to | |
| 4825 * split at this point, do the split first and set | |
| 4826 * TSF_DIDSPLIT to avoid doing it again. */ | |
| 4827 else if (!fword_ends | |
| 4828 && try_compound | |
| 4829 && (sp->ts_flags & TSF_DIDSPLIT) == 0) | |
| 4830 { | |
| 485 | 4831 try_compound = FALSE; |
| 625 | 4832 sp->ts_flags |= TSF_DIDSPLIT; |
| 4833 --sp->ts_curi; /* do the same NUL again */ | |
| 4834 compflags[sp->ts_complen] = NUL; | |
| 4835 } | |
| 4836 else | |
| 4837 sp->ts_flags &= ~TSF_DIDSPLIT; | |
| 4838 | |
| 4839 if (try_split || try_compound) | |
| 4840 { | |
| 534 | 4841 if (!try_compound && (!fword_ends || !goodword_ends)) |
| 485 | 4842 { |
| 4843 /* If we're going to split need to check that the | |
| 501 | 4844 * words so far are valid for compounding. If there |
| 4845 * is only one word it must not have the NEEDCOMPOUND | |
| 4846 * flag. */ | |
| 4847 if (sp->ts_complen == sp->ts_compsplit | |
| 4848 && (flags & WF_NEEDCOMP)) | |
| 4849 break; | |
| 490 | 4850 p = preword; |
| 4851 while (*skiptowhite(p) != NUL) | |
| 4852 p = skipwhite(skiptowhite(p)); | |
| 485 | 4853 if (sp->ts_complen > sp->ts_compsplit |
| 490 | 4854 && !can_compound(slang, p, |
| 485 | 4855 compflags + sp->ts_compsplit)) |
| 4856 break; | |
| 698 | 4857 |
| 4858 if (slang->sl_nosplitsugs) | |
| 4859 newscore += SCORE_SPLIT_NO; | |
| 4860 else | |
| 4861 newscore += SCORE_SPLIT; | |
| 625 | 4862 |
| 4863 /* Give a bonus to words seen before. */ | |
| 4864 newscore = score_wordcount_adj(slang, newscore, | |
| 4865 preword + sp->ts_prewordlen, TRUE); | |
| 482 | 4866 } |
| 4867 | |
| 625 | 4868 if (TRY_DEEPER(su, stack, depth, newscore)) |
| 323 | 4869 { |
| 625 | 4870 go_deeper(stack, depth, newscore); |
| 4871 #ifdef DEBUG_TRIEWALK | |
| 4872 if (!try_compound && !fword_ends) | |
| 4873 sprintf(changename[depth], "%.*s-%s: split", | |
| 4874 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 4875 else | |
| 4876 sprintf(changename[depth], "%.*s-%s: compound", | |
| 4877 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 4878 #endif | |
| 323 | 4879 /* Save things to be restored at STATE_SPLITUNDO. */ |
| 346 | 4880 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
|
4881 PROF_STORE(sp->ts_state) |
| 376 | 4882 sp->ts_state = STATE_SPLITUNDO; |
| 4883 | |
| 4884 ++depth; | |
| 4885 sp = &stack[depth]; | |
| 323 | 4886 |
| 482 | 4887 /* Append a space to preword when splitting. */ |
| 4888 if (!try_compound && !fword_ends) | |
| 4889 STRCAT(preword, " "); | |
| 835 | 4890 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 483 | 4891 sp->ts_splitoff = sp->ts_twordlen; |
| 492 | 4892 sp->ts_splitfidx = sp->ts_fidx; |
| 376 | 4893 |
| 4894 /* If the badword has a non-word character at this | |
| 4895 * position skip it. That means replacing the | |
| 482 | 4896 * non-word character with a space. Always skip a |
| 534 | 4897 * character when the word ends. But only when the |
| 4898 * good word can end. */ | |
| 625 | 4899 if (((!try_compound && !spell_iswordp_nmw(fword |
| 5477 | 4900 + sp->ts_fidx, |
| 4901 curwin)) | |
| 625 | 4902 || fword_ends) |
| 4903 && fword[sp->ts_fidx] != NUL | |
| 4904 && goodword_ends) | |
| 376 | 4905 { |
| 482 | 4906 int l; |
| 4907 | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
4908 l = MB_PTR2LEN(fword + sp->ts_fidx); |
| 482 | 4909 if (fword_ends) |
| 4910 { | |
| 4911 /* Copy the skipped character to preword. */ | |
| 483 | 4912 mch_memmove(preword + sp->ts_prewordlen, |
| 482 | 4913 fword + sp->ts_fidx, l); |
| 483 | 4914 sp->ts_prewordlen += l; |
| 4915 preword[sp->ts_prewordlen] = NUL; | |
| 482 | 4916 } |
| 4917 else | |
| 4918 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST; | |
| 4919 sp->ts_fidx += l; | |
| 376 | 4920 } |
| 455 | 4921 |
| 485 | 4922 /* When compounding include compound flag in |
| 4923 * compflags[] (already set above). When splitting we | |
| 4924 * may start compounding over again. */ | |
| 482 | 4925 if (try_compound) |
| 483 | 4926 ++sp->ts_complen; |
| 482 | 4927 else |
| 485 | 4928 sp->ts_compsplit = sp->ts_complen; |
| 4929 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 482 | 4930 |
| 455 | 4931 /* set su->su_badflags to the caps type at this |
| 4932 * position */ | |
| 324 | 4933 #ifdef FEAT_MBYTE |
| 4934 if (has_mbyte) | |
| 455 | 4935 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); |
| 324 | 4936 else |
| 4937 #endif | |
| 455 | 4938 n = sp->ts_fidx; |
| 474 | 4939 su->su_badflags = badword_captype(su->su_badptr + n, |
| 455 | 4940 su->su_badptr + su->su_badlen); |
| 323 | 4941 |
| 4942 /* Restart at top of the tree. */ | |
| 376 | 4943 sp->ts_arridx = 0; |
| 485 | 4944 |
| 4945 /* If there are postponed prefixes, try these too. */ | |
| 4946 if (pbyts != NULL) | |
| 4947 { | |
| 4948 byts = pbyts; | |
| 4949 idxs = pidxs; | |
| 4950 sp->ts_prefixdepth = PFD_PREFIXTREE; | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4951 PROF_STORE(sp->ts_state) |
| 485 | 4952 sp->ts_state = STATE_NOPREFIX; |
| 4953 } | |
| 323 | 4954 } |
| 4955 } | |
| 625 | 4956 } |
| 4957 break; | |
| 4958 | |
| 4959 case STATE_SPLITUNDO: | |
| 4960 /* Undo the changes done for word split or compound word. */ | |
| 4961 su->su_badflags = sp->ts_save_badflags; | |
| 4962 | |
| 4963 /* Continue looking for NUL bytes. */ | |
|
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) |
| 625 | 4965 sp->ts_state = STATE_START; |
| 4966 | |
| 4967 /* In case we went into the prefix tree. */ | |
| 4968 byts = fbyts; | |
| 4969 idxs = fidxs; | |
| 4970 break; | |
| 4971 | |
| 4972 case STATE_ENDNUL: | |
| 4973 /* Past the NUL bytes in the node. */ | |
| 4974 su->su_badflags = sp->ts_save_badflags; | |
| 4975 if (fword[sp->ts_fidx] == NUL | |
| 4976 #ifdef FEAT_MBYTE | |
| 4977 && sp->ts_tcharlen == 0 | |
| 4978 #endif | |
| 4979 ) | |
| 4980 { | |
| 4981 /* 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
|
4982 PROF_STORE(sp->ts_state) |
| 625 | 4983 sp->ts_state = STATE_DEL; |
| 323 | 4984 break; |
| 625 | 4985 } |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4986 PROF_STORE(sp->ts_state) |
| 625 | 4987 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
|
4988 /* FALLTHROUGH */ |
| 625 | 4989 |
| 4990 case STATE_PLAIN: | |
| 4991 /* | |
| 4992 * Go over all possible bytes at this node, add each to tword[] | |
| 4993 * and use child node. "ts_curi" is the index. | |
| 4994 */ | |
| 4995 arridx = sp->ts_arridx; | |
| 4996 if (sp->ts_curi > byts[arridx]) | |
| 4997 { | |
| 4998 /* Done all bytes at this node, do next state. When still at | |
| 4999 * 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
|
5000 PROF_STORE(sp->ts_state) |
| 625 | 5001 if (sp->ts_fidx >= sp->ts_fidxtry) |
| 323 | 5002 sp->ts_state = STATE_DEL; |
| 5003 else | |
| 625 | 5004 sp->ts_state = STATE_FINAL; |
| 5005 } | |
| 5006 else | |
| 5007 { | |
| 5008 arridx += sp->ts_curi++; | |
| 5009 c = byts[arridx]; | |
| 5010 | |
| 5011 /* Normal byte, go one level deeper. If it's not equal to the | |
| 5012 * byte in the bad word adjust the score. But don't even try | |
| 5013 * when the byte was already changed. And don't try when we | |
| 5695 | 5014 * just deleted this byte, accepting it is always cheaper than |
| 625 | 5015 * delete + substitute. */ |
| 5016 if (c == fword[sp->ts_fidx] | |
| 5017 #ifdef FEAT_MBYTE | |
| 5018 || (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE) | |
| 5019 #endif | |
| 5020 ) | |
| 5021 newscore = 0; | |
| 5022 else | |
| 5023 newscore = SCORE_SUBST; | |
| 5024 if ((newscore == 0 | |
| 5025 || (sp->ts_fidx >= sp->ts_fidxtry | |
| 5026 && ((sp->ts_flags & TSF_DIDDEL) == 0 | |
| 5027 || c != fword[sp->ts_delidx]))) | |
| 5028 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5029 { | |
| 5030 go_deeper(stack, depth, newscore); | |
| 5031 #ifdef DEBUG_TRIEWALK | |
| 5032 if (newscore > 0) | |
| 5033 sprintf(changename[depth], "%.*s-%s: subst %c to %c", | |
| 5034 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5035 fword[sp->ts_fidx], c); | |
| 323 | 5036 else |
| 625 | 5037 sprintf(changename[depth], "%.*s-%s: accept %c", |
| 5038 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5039 fword[sp->ts_fidx]); | |
| 5040 #endif | |
| 323 | 5041 ++depth; |
| 625 | 5042 sp = &stack[depth]; |
| 5043 ++sp->ts_fidx; | |
| 5044 tword[sp->ts_twordlen++] = c; | |
| 5045 sp->ts_arridx = idxs[arridx]; | |
| 5046 #ifdef FEAT_MBYTE | |
| 5047 if (newscore == SCORE_SUBST) | |
| 5048 sp->ts_isdiff = DIFF_YES; | |
| 330 | 5049 if (has_mbyte) |
| 358 | 5050 { |
| 625 | 5051 /* Multi-byte characters are a bit complicated to |
| 5052 * handle: They differ when any of the bytes differ | |
| 5053 * and then their length may also differ. */ | |
| 5054 if (sp->ts_tcharlen == 0) | |
| 5055 { | |
| 5056 /* First byte. */ | |
| 5057 sp->ts_tcharidx = 0; | |
| 5058 sp->ts_tcharlen = MB_BYTE2LEN(c); | |
| 5059 sp->ts_fcharstart = sp->ts_fidx - 1; | |
| 5060 sp->ts_isdiff = (newscore != 0) | |
| 5061 ? DIFF_YES : DIFF_NONE; | |
| 5062 } | |
| 5063 else if (sp->ts_isdiff == DIFF_INSERT) | |
| 5064 /* When inserting trail bytes don't advance in the | |
| 5065 * bad word. */ | |
| 5066 --sp->ts_fidx; | |
| 5067 if (++sp->ts_tcharidx == sp->ts_tcharlen) | |
| 5068 { | |
| 5069 /* Last byte of character. */ | |
| 5070 if (sp->ts_isdiff == DIFF_YES) | |
| 5071 { | |
| 5072 /* Correct ts_fidx for the byte length of the | |
| 5073 * character (we didn't check that before). */ | |
| 5074 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
|
5075 + MB_PTR2LEN( |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5076 fword + sp->ts_fcharstart); |
| 625 | 5077 /* For changing a composing character adjust |
| 5078 * the score from SCORE_SUBST to | |
| 5079 * SCORE_SUBCOMP. */ | |
| 5080 if (enc_utf8 | |
| 5081 && 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
|
5082 utf_ptr2char(tword |
| 625 | 5083 + sp->ts_twordlen |
| 5084 - sp->ts_tcharlen)) | |
| 5085 && 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
|
5086 utf_ptr2char(fword |
| 625 | 5087 + sp->ts_fcharstart))) |
| 5088 sp->ts_score -= | |
| 5089 SCORE_SUBST - SCORE_SUBCOMP; | |
| 5090 | |
| 5091 /* For a similar character adjust score from | |
| 5092 * SCORE_SUBST to SCORE_SIMILAR. */ | |
| 5093 else if (!soundfold | |
| 5094 && slang->sl_has_map | |
| 5095 && similar_chars(slang, | |
| 5096 mb_ptr2char(tword | |
| 5097 + sp->ts_twordlen | |
| 5098 - sp->ts_tcharlen), | |
| 5099 mb_ptr2char(fword | |
| 5100 + sp->ts_fcharstart))) | |
| 5101 sp->ts_score -= | |
| 5102 SCORE_SUBST - SCORE_SIMILAR; | |
| 5103 } | |
| 5104 else if (sp->ts_isdiff == DIFF_INSERT | |
| 5105 && sp->ts_twordlen > sp->ts_tcharlen) | |
| 5106 { | |
| 5107 p = tword + sp->ts_twordlen - sp->ts_tcharlen; | |
| 5108 c = mb_ptr2char(p); | |
| 5109 if (enc_utf8 && utf_iscomposing(c)) | |
| 5110 { | |
| 5111 /* Inserting a composing char doesn't | |
| 5112 * count that much. */ | |
| 5113 sp->ts_score -= SCORE_INS - SCORE_INSCOMP; | |
| 5114 } | |
| 5115 else | |
| 5116 { | |
| 5117 /* If the previous character was the same, | |
| 5118 * thus doubling a character, give a bonus | |
| 5119 * to the score. Also for the soundfold | |
| 5120 * tree (might seem illogical but does | |
| 5121 * give better scores). */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5122 MB_PTR_BACK(tword, p); |
| 625 | 5123 if (c == mb_ptr2char(p)) |
| 5124 sp->ts_score -= SCORE_INS | |
| 5125 - SCORE_INSDUP; | |
| 5126 } | |
| 5127 } | |
| 5128 | |
| 5129 /* Starting a new char, reset the length. */ | |
| 5130 sp->ts_tcharlen = 0; | |
| 5131 } | |
| 358 | 5132 } |
| 330 | 5133 else |
| 5134 #endif | |
| 358 | 5135 { |
| 625 | 5136 /* If we found a similar char adjust the score. |
| 5137 * We do this after calling go_deeper() because | |
| 5138 * it's slow. */ | |
| 5139 if (newscore != 0 | |
| 5140 && !soundfold | |
| 5141 && slang->sl_has_map | |
| 5142 && similar_chars(slang, | |
| 5143 c, fword[sp->ts_fidx - 1])) | |
| 5144 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR; | |
| 358 | 5145 } |
| 625 | 5146 } |
| 5147 } | |
| 5148 break; | |
| 5149 | |
| 5150 case STATE_DEL: | |
| 5151 #ifdef FEAT_MBYTE | |
| 5152 /* When past the first byte of a multi-byte char don't try | |
| 5153 * delete/insert/swap a character. */ | |
| 5154 if (has_mbyte && sp->ts_tcharlen > 0) | |
| 5155 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5156 PROF_STORE(sp->ts_state) |
| 625 | 5157 sp->ts_state = STATE_FINAL; |
| 5158 break; | |
| 5159 } | |
| 5160 #endif | |
| 5161 /* | |
| 5162 * Try skipping one character in the bad word (delete it). | |
| 5163 */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5164 PROF_STORE(sp->ts_state) |
| 625 | 5165 sp->ts_state = STATE_INS_PREP; |
| 5166 sp->ts_curi = 1; | |
| 5167 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*') | |
| 5168 /* Deleting a vowel at the start of a word counts less, see | |
| 5169 * soundalike_score(). */ | |
| 5170 newscore = 2 * SCORE_DEL / 3; | |
| 5171 else | |
| 5172 newscore = SCORE_DEL; | |
| 5173 if (fword[sp->ts_fidx] != NUL | |
| 5174 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5175 { | |
| 5176 go_deeper(stack, depth, newscore); | |
| 5177 #ifdef DEBUG_TRIEWALK | |
| 5178 sprintf(changename[depth], "%.*s-%s: delete %c", | |
| 5179 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5180 fword[sp->ts_fidx]); | |
| 5181 #endif | |
| 5182 ++depth; | |
| 5183 | |
| 5184 /* Remember what character we deleted, so that we can avoid | |
| 5185 * inserting it again. */ | |
| 5186 stack[depth].ts_flags |= TSF_DIDDEL; | |
| 5187 stack[depth].ts_delidx = sp->ts_fidx; | |
| 5188 | |
| 5189 /* Advance over the character in fword[]. Give a bonus to the | |
| 5190 * score if the same character is following "nn" -> "n". It's | |
| 5191 * a bit illogical for soundfold tree but it does give better | |
| 5192 * results. */ | |
| 5193 #ifdef FEAT_MBYTE | |
| 5194 if (has_mbyte) | |
| 5195 { | |
| 5196 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
|
5197 stack[depth].ts_fidx += MB_PTR2LEN(fword + sp->ts_fidx); |
| 625 | 5198 if (enc_utf8 && utf_iscomposing(c)) |
| 5199 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP; | |
| 5200 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx)) | |
| 5201 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 5202 } |
| 5203 else | |
| 625 | 5204 #endif |
| 5205 { | |
| 5206 ++stack[depth].ts_fidx; | |
| 5207 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1]) | |
| 5208 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 5209 } |
| 5210 break; | |
| 625 | 5211 } |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5212 /* FALLTHROUGH */ |
| 625 | 5213 |
| 5214 case STATE_INS_PREP: | |
| 5215 if (sp->ts_flags & TSF_DIDDEL) | |
| 5216 { | |
| 5217 /* If we just deleted a byte then inserting won't make sense, | |
| 5218 * a substitute is always cheaper. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5219 PROF_STORE(sp->ts_state) |
| 625 | 5220 sp->ts_state = STATE_SWAP; |
| 5221 break; | |
| 5222 } | |
| 5223 | |
| 5224 /* skip over NUL bytes */ | |
| 5225 n = sp->ts_arridx; | |
| 5226 for (;;) | |
| 5227 { | |
| 5228 if (sp->ts_curi > byts[n]) | |
| 5229 { | |
| 5230 /* 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
|
5231 PROF_STORE(sp->ts_state) |
| 625 | 5232 sp->ts_state = STATE_SWAP; |
| 5233 break; | |
| 5234 } | |
| 5235 if (byts[n + sp->ts_curi] != NUL) | |
| 5236 { | |
| 5237 /* Found a byte to insert. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5238 PROF_STORE(sp->ts_state) |
| 625 | 5239 sp->ts_state = STATE_INS; |
| 330 | 5240 break; |
| 5241 } | |
| 625 | 5242 ++sp->ts_curi; |
| 5243 } | |
| 5244 break; | |
| 5245 | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5246 /* FALLTHROUGH */ |
| 625 | 5247 |
| 5248 case STATE_INS: | |
| 5249 /* Insert one byte. Repeat this for each possible byte at this | |
| 5250 * node. */ | |
| 5251 n = sp->ts_arridx; | |
| 5252 if (sp->ts_curi > byts[n]) | |
| 5253 { | |
| 5254 /* 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
|
5255 PROF_STORE(sp->ts_state) |
| 625 | 5256 sp->ts_state = STATE_SWAP; |
| 5257 break; | |
| 5258 } | |
| 5259 | |
| 5260 /* Do one more byte at this node, but: | |
| 5261 * - Skip NUL bytes. | |
| 5262 * - Skip the byte if it's equal to the byte in the word, | |
| 5263 * accepting that byte is always better. | |
| 5264 */ | |
| 5265 n += sp->ts_curi++; | |
| 5266 c = byts[n]; | |
| 5267 if (soundfold && sp->ts_twordlen == 0 && c == '*') | |
| 5268 /* Inserting a vowel at the start of a word counts less, | |
| 5269 * see soundalike_score(). */ | |
| 5270 newscore = 2 * SCORE_INS / 3; | |
| 5271 else | |
| 5272 newscore = SCORE_INS; | |
| 5273 if (c != fword[sp->ts_fidx] | |
| 5274 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5275 { | |
| 5276 go_deeper(stack, depth, newscore); | |
| 5277 #ifdef DEBUG_TRIEWALK | |
| 5278 sprintf(changename[depth], "%.*s-%s: insert %c", | |
| 5279 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5280 c); | |
| 5281 #endif | |
| 5282 ++depth; | |
| 5283 sp = &stack[depth]; | |
| 5284 tword[sp->ts_twordlen++] = c; | |
| 5285 sp->ts_arridx = idxs[n]; | |
| 330 | 5286 #ifdef FEAT_MBYTE |
| 5287 if (has_mbyte) | |
| 5288 { | |
| 625 | 5289 fl = MB_BYTE2LEN(c); |
| 5290 if (fl > 1) | |
| 330 | 5291 { |
| 625 | 5292 /* There are following bytes for the same character. |
| 5293 * We must find all bytes before trying | |
| 5294 * delete/insert/swap/etc. */ | |
| 5295 sp->ts_tcharlen = fl; | |
| 5296 sp->ts_tcharidx = 1; | |
| 5297 sp->ts_isdiff = DIFF_INSERT; | |
| 330 | 5298 } |
| 323 | 5299 } |
| 5300 else | |
| 625 | 5301 fl = 1; |
| 5302 if (fl == 1) | |
| 5303 #endif | |
| 5304 { | |
| 5305 /* If the previous character was the same, thus doubling a | |
| 5306 * character, give a bonus to the score. Also for | |
| 5307 * soundfold words (illogical but does give a better | |
| 5308 * score). */ | |
| 5309 if (sp->ts_twordlen >= 2 | |
| 5310 && tword[sp->ts_twordlen - 2] == c) | |
| 5311 sp->ts_score -= SCORE_INS - SCORE_INSDUP; | |
| 5312 } | |
| 5313 } | |
| 5314 break; | |
| 5315 | |
| 5316 case STATE_SWAP: | |
| 5317 /* | |
| 5318 * Swap two bytes in the bad word: "12" -> "21". | |
| 5319 * We change "fword" here, it's changed back afterwards at | |
| 5320 * STATE_UNSWAP. | |
| 5321 */ | |
| 5322 p = fword + sp->ts_fidx; | |
| 5323 c = *p; | |
| 5324 if (c == NUL) | |
| 5325 { | |
| 5326 /* 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
|
5327 PROF_STORE(sp->ts_state) |
| 625 | 5328 sp->ts_state = STATE_FINAL; |
| 5329 break; | |
| 5330 } | |
| 5331 | |
| 5332 /* Don't swap if the first character is not a word character. | |
| 5333 * 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
|
5334 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 5335 { |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5336 PROF_STORE(sp->ts_state) |
| 625 | 5337 sp->ts_state = STATE_REP_INI; |
| 323 | 5338 break; |
| 625 | 5339 } |
| 5340 | |
| 5341 #ifdef FEAT_MBYTE | |
| 5342 if (has_mbyte) | |
| 5343 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5344 n = MB_CPTR2LEN(p); |
| 625 | 5345 c = mb_ptr2char(p); |
| 1337 | 5346 if (p[n] == NUL) |
| 5347 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5348 else if (!soundfold && !spell_iswordp(p + n, curwin)) |
| 625 | 5349 c2 = c; /* don't swap non-word char */ |
| 5350 else | |
| 5351 c2 = mb_ptr2char(p + n); | |
| 5352 } | |
| 5353 else | |
| 5354 #endif | |
| 5355 { | |
| 1337 | 5356 if (p[1] == NUL) |
| 5357 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5358 else if (!soundfold && !spell_iswordp(p + 1, curwin)) |
| 625 | 5359 c2 = c; /* don't swap non-word char */ |
| 5360 else | |
| 5361 c2 = p[1]; | |
| 5362 } | |
| 5363 | |
| 1337 | 5364 /* When the second character is NUL we can't swap. */ |
| 5365 if (c2 == NUL) | |
| 5366 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5367 PROF_STORE(sp->ts_state) |
| 1337 | 5368 sp->ts_state = STATE_REP_INI; |
| 5369 break; | |
| 5370 } | |
| 5371 | |
| 625 | 5372 /* When characters are identical, swap won't do anything. |
| 5373 * Also get here if the second char is not a word character. */ | |
| 5374 if (c == c2) | |
| 5375 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5376 PROF_STORE(sp->ts_state) |
| 625 | 5377 sp->ts_state = STATE_SWAP3; |
| 5378 break; | |
| 5379 } | |
| 5380 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP)) | |
| 5381 { | |
| 5382 go_deeper(stack, depth, SCORE_SWAP); | |
| 5383 #ifdef DEBUG_TRIEWALK | |
| 5384 sprintf(changename[depth], "%.*s-%s: swap %c and %c", | |
| 5385 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5386 c, c2); | |
| 5387 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5388 PROF_STORE(sp->ts_state) |
| 625 | 5389 sp->ts_state = STATE_UNSWAP; |
| 5390 ++depth; | |
| 330 | 5391 #ifdef FEAT_MBYTE |
| 5392 if (has_mbyte) | |
| 5393 { | |
| 625 | 5394 fl = mb_char2len(c2); |
| 5395 mch_memmove(p, p + n, fl); | |
| 5396 mb_char2bytes(c, p + fl); | |
| 5397 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 5398 } |
| 5399 else | |
| 5400 #endif | |
| 5401 { | |
| 625 | 5402 p[0] = c2; |
| 330 | 5403 p[1] = c; |
| 625 | 5404 stack[depth].ts_fidxtry = sp->ts_fidx + 2; |
| 5405 } | |
| 5406 } | |
| 5407 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5408 { |
| 625 | 5409 /* 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
|
5410 PROF_STORE(sp->ts_state) |
| 625 | 5411 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
|
5412 } |
| 625 | 5413 break; |
| 5414 | |
| 5415 case STATE_UNSWAP: | |
| 5416 /* Undo the STATE_SWAP swap: "21" -> "12". */ | |
| 5417 p = fword + sp->ts_fidx; | |
| 5418 #ifdef FEAT_MBYTE | |
| 5419 if (has_mbyte) | |
| 5420 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5421 n = MB_PTR2LEN(p); |
| 625 | 5422 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
|
5423 mch_memmove(p + MB_PTR2LEN(p + n), p, n); |
| 625 | 5424 mb_char2bytes(c, p); |
| 5425 } | |
| 5426 else | |
| 5427 #endif | |
| 5428 { | |
| 5429 c = *p; | |
| 5430 *p = p[1]; | |
| 5431 p[1] = c; | |
| 5432 } | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5433 /* FALLTHROUGH */ |
| 625 | 5434 |
| 5435 case STATE_SWAP3: | |
| 5436 /* Swap two bytes, skipping one: "123" -> "321". We change | |
| 5437 * "fword" here, it's changed back afterwards at STATE_UNSWAP3. */ | |
| 5438 p = fword + sp->ts_fidx; | |
| 5439 #ifdef FEAT_MBYTE | |
| 5440 if (has_mbyte) | |
| 5441 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5442 n = MB_CPTR2LEN(p); |
| 625 | 5443 c = mb_ptr2char(p); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5444 fl = MB_CPTR2LEN(p + n); |
| 625 | 5445 c2 = mb_ptr2char(p + n); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5446 if (!soundfold && !spell_iswordp(p + n + fl, curwin)) |
| 625 | 5447 c3 = c; /* don't swap non-word char */ |
| 5448 else | |
| 5449 c3 = mb_ptr2char(p + n + fl); | |
| 5450 } | |
| 5451 else | |
| 5452 #endif | |
| 5453 { | |
| 5454 c = *p; | |
| 5455 c2 = p[1]; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5456 if (!soundfold && !spell_iswordp(p + 2, curwin)) |
| 625 | 5457 c3 = c; /* don't swap non-word char */ |
| 5458 else | |
| 5459 c3 = p[2]; | |
| 5460 } | |
| 5461 | |
| 5462 /* When characters are identical: "121" then SWAP3 result is | |
| 5463 * identical, ROT3L result is same as SWAP: "211", ROT3L result is | |
| 5464 * same as SWAP on next char: "112". Thus skip all swapping. | |
| 5465 * Also skip when c3 is NUL. | |
| 5466 * Also get here when the third character is not a word character. | |
| 5467 * Second character may any char: "a.b" -> "b.a" */ | |
| 5468 if (c == c3 || c3 == NUL) | |
| 5469 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5470 PROF_STORE(sp->ts_state) |
| 625 | 5471 sp->ts_state = STATE_REP_INI; |
| 5472 break; | |
| 5473 } | |
| 5474 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5475 { | |
| 5476 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5477 #ifdef DEBUG_TRIEWALK | |
| 5478 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c", | |
| 5479 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5480 c, c3); | |
| 5481 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5482 PROF_STORE(sp->ts_state) |
| 625 | 5483 sp->ts_state = STATE_UNSWAP3; |
| 5484 ++depth; | |
| 5485 #ifdef FEAT_MBYTE | |
| 5486 if (has_mbyte) | |
| 5487 { | |
| 5488 tl = mb_char2len(c3); | |
| 5489 mch_memmove(p, p + n + fl, tl); | |
| 5490 mb_char2bytes(c2, p + tl); | |
| 5491 mb_char2bytes(c, p + fl + tl); | |
| 5492 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl; | |
| 5493 } | |
| 5494 else | |
| 5495 #endif | |
| 5496 { | |
| 5497 p[0] = p[2]; | |
| 5498 p[2] = c; | |
| 5499 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 5500 } | |
| 5501 } | |
| 5502 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5503 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5504 PROF_STORE(sp->ts_state) |
| 625 | 5505 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
|
5506 } |
| 625 | 5507 break; |
| 5508 | |
| 5509 case STATE_UNSWAP3: | |
| 5510 /* Undo STATE_SWAP3: "321" -> "123" */ | |
| 5511 p = fword + sp->ts_fidx; | |
| 5512 #ifdef FEAT_MBYTE | |
| 5513 if (has_mbyte) | |
| 5514 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5515 n = MB_PTR2LEN(p); |
| 625 | 5516 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
|
5517 fl = MB_PTR2LEN(p + n); |
| 625 | 5518 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
|
5519 tl = MB_PTR2LEN(p + n + fl); |
| 625 | 5520 mch_memmove(p + fl + tl, p, n); |
| 5521 mb_char2bytes(c, p); | |
| 5522 mb_char2bytes(c2, p + tl); | |
| 5523 p = p + tl; | |
| 5524 } | |
| 5525 else | |
| 5526 #endif | |
| 5527 { | |
| 5528 c = *p; | |
| 5529 *p = p[2]; | |
| 5530 p[2] = c; | |
| 5531 ++p; | |
| 5532 } | |
| 5533 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5534 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 5535 { |
| 5536 /* Middle char is not a word char, skip the rotate. First and | |
| 5537 * 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
|
5538 PROF_STORE(sp->ts_state) |
| 625 | 5539 sp->ts_state = STATE_REP_INI; |
| 5540 break; | |
| 5541 } | |
| 5542 | |
| 5543 /* Rotate three characters left: "123" -> "231". We change | |
| 5544 * "fword" here, it's changed back afterwards at STATE_UNROT3L. */ | |
| 5545 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5546 { | |
| 5547 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5548 #ifdef DEBUG_TRIEWALK | |
| 5549 p = fword + sp->ts_fidx; | |
| 5550 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c", | |
| 5551 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5552 p[0], p[1], p[2]); | |
| 5553 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5554 PROF_STORE(sp->ts_state) |
| 625 | 5555 sp->ts_state = STATE_UNROT3L; |
| 5556 ++depth; | |
| 330 | 5557 p = fword + sp->ts_fidx; |
| 5558 #ifdef FEAT_MBYTE | |
| 5559 if (has_mbyte) | |
| 5560 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5561 n = MB_CPTR2LEN(p); |
| 330 | 5562 c = mb_ptr2char(p); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5563 fl = MB_CPTR2LEN(p + n); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5564 fl += MB_CPTR2LEN(p + n + fl); |
| 625 | 5565 mch_memmove(p, p + n, fl); |
| 5566 mb_char2bytes(c, p + fl); | |
| 5567 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 5568 } |
| 5569 else | |
| 5570 #endif | |
| 323 | 5571 { |
| 330 | 5572 c = *p; |
| 625 | 5573 *p = p[1]; |
| 5574 p[1] = p[2]; | |
| 5575 p[2] = c; | |
| 5576 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 5577 } | |
| 5578 } | |
| 5579 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5580 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5581 PROF_STORE(sp->ts_state) |
| 625 | 5582 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
|
5583 } |
| 625 | 5584 break; |
| 5585 | |
| 5586 case STATE_UNROT3L: | |
| 5587 /* Undo ROT3L: "231" -> "123" */ | |
| 5588 p = fword + sp->ts_fidx; | |
| 5589 #ifdef FEAT_MBYTE | |
| 5590 if (has_mbyte) | |
| 5591 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5592 n = MB_PTR2LEN(p); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5593 n += MB_PTR2LEN(p + n); |
| 625 | 5594 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
|
5595 tl = MB_PTR2LEN(p + n); |
| 625 | 5596 mch_memmove(p + tl, p, n); |
| 5597 mb_char2bytes(c, p); | |
| 5598 } | |
| 5599 else | |
| 5600 #endif | |
| 5601 { | |
| 5602 c = p[2]; | |
| 5603 p[2] = p[1]; | |
| 5604 p[1] = *p; | |
| 5605 *p = c; | |
| 5606 } | |
| 5607 | |
| 5608 /* Rotate three bytes right: "123" -> "312". We change "fword" | |
| 5609 * here, it's changed back afterwards at STATE_UNROT3R. */ | |
| 5610 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5611 { | |
| 5612 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5613 #ifdef DEBUG_TRIEWALK | |
| 5614 p = fword + sp->ts_fidx; | |
| 5615 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c", | |
| 5616 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5617 p[0], p[1], p[2]); | |
| 5618 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5619 PROF_STORE(sp->ts_state) |
| 625 | 5620 sp->ts_state = STATE_UNROT3R; |
| 5621 ++depth; | |
| 330 | 5622 p = fword + sp->ts_fidx; |
| 5623 #ifdef FEAT_MBYTE | |
| 5624 if (has_mbyte) | |
| 5625 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5626 n = MB_CPTR2LEN(p); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5627 n += MB_CPTR2LEN(p + n); |
| 330 | 5628 c = mb_ptr2char(p + n); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5629 tl = MB_CPTR2LEN(p + n); |
| 330 | 5630 mch_memmove(p + tl, p, n); |
| 5631 mb_char2bytes(c, p); | |
| 625 | 5632 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl; |
| 330 | 5633 } |
| 5634 else | |
| 5635 #endif | |
| 5636 { | |
| 5637 c = p[2]; | |
| 5638 p[2] = p[1]; | |
| 5639 p[1] = *p; | |
| 5640 *p = c; | |
| 625 | 5641 stack[depth].ts_fidxtry = sp->ts_fidx + 3; |
| 5642 } | |
| 5643 } | |
| 5644 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5645 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5646 PROF_STORE(sp->ts_state) |
| 625 | 5647 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
|
5648 } |
| 625 | 5649 break; |
| 5650 | |
| 5651 case STATE_UNROT3R: | |
| 5652 /* Undo ROT3R: "312" -> "123" */ | |
| 5653 p = fword + sp->ts_fidx; | |
| 5654 #ifdef FEAT_MBYTE | |
| 5655 if (has_mbyte) | |
| 5656 { | |
| 5657 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
|
5658 tl = MB_PTR2LEN(p); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5659 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
|
5660 n += MB_PTR2LEN(p + tl + n); |
| 625 | 5661 mch_memmove(p, p + tl, n); |
| 5662 mb_char2bytes(c, p + n); | |
| 5663 } | |
| 5664 else | |
| 5665 #endif | |
| 5666 { | |
| 5667 c = *p; | |
| 5668 *p = p[1]; | |
| 5669 p[1] = p[2]; | |
| 5670 p[2] = c; | |
| 5671 } | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5672 /* FALLTHROUGH */ |
| 625 | 5673 |
| 5674 case STATE_REP_INI: | |
| 5675 /* Check if matching with REP items from the .aff file would work. | |
| 5676 * Quickly skip if: | |
| 5677 * - there are no REP items and we are not in the soundfold trie | |
| 5678 * - the score is going to be too high anyway | |
| 5679 * - already applied a REP item or swapped here */ | |
| 5680 if ((lp->lp_replang == NULL && !soundfold) | |
| 5681 || sp->ts_score + SCORE_REP >= su->su_maxscore | |
| 5682 || sp->ts_fidx < sp->ts_fidxtry) | |
| 5683 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5684 PROF_STORE(sp->ts_state) |
| 625 | 5685 sp->ts_state = STATE_FINAL; |
| 323 | 5686 break; |
| 625 | 5687 } |
| 5688 | |
| 5689 /* Use the first byte to quickly find the first entry that may | |
| 5690 * match. If the index is -1 there is none. */ | |
| 5691 if (soundfold) | |
| 5692 sp->ts_curi = slang->sl_repsal_first[fword[sp->ts_fidx]]; | |
| 5693 else | |
| 5694 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]]; | |
| 5695 | |
| 5696 if (sp->ts_curi < 0) | |
| 5697 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5698 PROF_STORE(sp->ts_state) |
| 625 | 5699 sp->ts_state = STATE_FINAL; |
| 5700 break; | |
| 5701 } | |
| 5702 | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5703 PROF_STORE(sp->ts_state) |
| 625 | 5704 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
|
5705 /* FALLTHROUGH */ |
| 625 | 5706 |
| 5707 case STATE_REP: | |
| 5708 /* Try matching with REP items from the .aff file. For each match | |
| 5709 * replace the characters and check if the resulting word is | |
| 5710 * valid. */ | |
| 5711 p = fword + sp->ts_fidx; | |
| 5712 | |
| 5713 if (soundfold) | |
| 5714 gap = &slang->sl_repsal; | |
| 5715 else | |
| 503 | 5716 gap = &lp->lp_replang->sl_rep; |
| 625 | 5717 while (sp->ts_curi < gap->ga_len) |
| 5718 { | |
| 5719 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++; | |
| 5720 if (*ftp->ft_from != *p) | |
| 5721 { | |
| 5722 /* past possible matching entries */ | |
| 5723 sp->ts_curi = gap->ga_len; | |
| 323 | 5724 break; |
| 5725 } | |
| 625 | 5726 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0 |
| 5727 && TRY_DEEPER(su, stack, depth, SCORE_REP)) | |
| 5728 { | |
| 5729 go_deeper(stack, depth, SCORE_REP); | |
| 5730 #ifdef DEBUG_TRIEWALK | |
| 5731 sprintf(changename[depth], "%.*s-%s: replace %s with %s", | |
| 5732 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5733 ftp->ft_from, ftp->ft_to); | |
| 5734 #endif | |
| 5735 /* Need to undo this afterwards. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5736 PROF_STORE(sp->ts_state) |
| 625 | 5737 sp->ts_state = STATE_REP_UNDO; |
| 5738 | |
| 5739 /* Change the "from" to the "to" string. */ | |
| 5740 ++depth; | |
| 835 | 5741 fl = (int)STRLEN(ftp->ft_from); |
| 5742 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 5743 if (fl != tl) |
| 323 | 5744 { |
| 1619 | 5745 STRMOVE(p + tl, p + fl); |
| 625 | 5746 repextra += tl - fl; |
| 323 | 5747 } |
| 625 | 5748 mch_memmove(p, ftp->ft_to, tl); |
| 5749 stack[depth].ts_fidxtry = sp->ts_fidx + tl; | |
| 5750 #ifdef FEAT_MBYTE | |
| 5751 stack[depth].ts_tcharlen = 0; | |
| 5752 #endif | |
| 5753 break; | |
| 5754 } | |
| 5755 } | |
| 5756 | |
| 5757 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
|
5758 { |
| 625 | 5759 /* No (more) matches. */ |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5760 PROF_STORE(sp->ts_state) |
| 625 | 5761 sp->ts_state = STATE_FINAL; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5762 } |
| 625 | 5763 |
| 5764 break; | |
| 5765 | |
| 5766 case STATE_REP_UNDO: | |
| 5767 /* Undo a REP replacement and continue with the next one. */ | |
| 5768 if (soundfold) | |
| 5769 gap = &slang->sl_repsal; | |
| 5770 else | |
| 5771 gap = &lp->lp_replang->sl_rep; | |
| 5772 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1; | |
| 835 | 5773 fl = (int)STRLEN(ftp->ft_from); |
| 5774 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 5775 p = fword + sp->ts_fidx; |
| 5776 if (fl != tl) | |
| 5777 { | |
| 1619 | 5778 STRMOVE(p + fl, p + tl); |
| 625 | 5779 repextra -= tl - fl; |
| 5780 } | |
| 5781 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
|
5782 PROF_STORE(sp->ts_state) |
| 625 | 5783 sp->ts_state = STATE_REP; |
| 5784 break; | |
| 5785 | |
| 5786 default: | |
| 5787 /* Did all possible states at this level, go up one level. */ | |
| 5788 --depth; | |
| 5789 | |
| 5790 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE) | |
| 5791 { | |
| 5792 /* Continue in or go back to the prefix tree. */ | |
| 5793 byts = pbyts; | |
| 5794 idxs = pidxs; | |
| 5795 } | |
| 5796 | |
| 5797 /* Don't check for CTRL-C too often, it takes time. */ | |
| 5798 if (--breakcheckcount == 0) | |
| 5799 { | |
| 5800 ui_breakcheck(); | |
| 5801 breakcheckcount = 1000; | |
| 5802 } | |
| 5803 } | |
| 5804 } | |
| 5805 } | |
| 5806 | |
| 5807 | |
| 5808 /* | |
| 5809 * Go one level deeper in the tree. | |
| 5810 */ | |
| 5811 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5812 go_deeper(trystate_T *stack, int depth, int score_add) |
| 323 | 5813 { |
| 330 | 5814 stack[depth + 1] = stack[depth]; |
| 323 | 5815 stack[depth + 1].ts_state = STATE_START; |
| 625 | 5816 stack[depth + 1].ts_score = stack[depth].ts_score + score_add; |
| 323 | 5817 stack[depth + 1].ts_curi = 1; /* start just after length byte */ |
| 485 | 5818 stack[depth + 1].ts_flags = 0; |
| 323 | 5819 } |
| 5820 | |
| 455 | 5821 #ifdef FEAT_MBYTE |
| 5822 /* | |
| 5823 * Case-folding may change the number of bytes: Count nr of chars in | |
| 5824 * fword[flen] and return the byte length of that many chars in "word". | |
| 5825 */ | |
| 5826 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5827 nofold_len(char_u *fword, int flen, char_u *word) |
| 455 | 5828 { |
| 5829 char_u *p; | |
| 5830 int i = 0; | |
| 5831 | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5832 for (p = fword; p < fword + flen; MB_PTR_ADV(p)) |
| 455 | 5833 ++i; |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5834 for (p = word; i > 0; MB_PTR_ADV(p)) |
| 455 | 5835 --i; |
| 5836 return (int)(p - word); | |
| 5837 } | |
| 5838 #endif | |
| 5839 | |
| 323 | 5840 /* |
| 5841 * "fword" is a good word with case folded. Find the matching keep-case | |
| 5842 * words and put it in "kword". | |
| 5843 * Theoretically there could be several keep-case words that result in the | |
| 5844 * same case-folded word, but we only find one... | |
| 5845 */ | |
| 5846 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5847 find_keepcap_word(slang_T *slang, char_u *fword, char_u *kword) |
| 323 | 5848 { |
| 5849 char_u uword[MAXWLEN]; /* "fword" in upper-case */ | |
| 5850 int depth; | |
| 324 | 5851 idx_T tryidx; |
| 323 | 5852 |
| 5853 /* The following arrays are used at each depth in the tree. */ | |
| 324 | 5854 idx_T arridx[MAXWLEN]; |
| 323 | 5855 int round[MAXWLEN]; |
| 5856 int fwordidx[MAXWLEN]; | |
| 5857 int uwordidx[MAXWLEN]; | |
| 5858 int kwordlen[MAXWLEN]; | |
| 5859 | |
| 5860 int flen, ulen; | |
| 5861 int l; | |
| 5862 int len; | |
| 5863 int c; | |
| 324 | 5864 idx_T lo, hi, m; |
| 323 | 5865 char_u *p; |
| 5866 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */ | |
| 324 | 5867 idx_T *idxs = slang->sl_kidxs; /* array with indexes */ |
| 323 | 5868 |
| 5869 if (byts == NULL) | |
| 5870 { | |
| 5871 /* array is empty: "cannot happen" */ | |
| 5872 *kword = NUL; | |
| 5873 return; | |
| 5874 } | |
| 5875 | |
| 5876 /* Make an all-cap version of "fword". */ | |
| 5877 allcap_copy(fword, uword); | |
| 5878 | |
| 5879 /* | |
| 5880 * Each character needs to be tried both case-folded and upper-case. | |
| 5881 * All this gets very complicated if we keep in mind that changing case | |
| 5882 * may change the byte length of a multi-byte character... | |
| 5883 */ | |
| 5884 depth = 0; | |
| 5885 arridx[0] = 0; | |
| 5886 round[0] = 0; | |
| 5887 fwordidx[0] = 0; | |
| 5888 uwordidx[0] = 0; | |
| 5889 kwordlen[0] = 0; | |
| 5890 while (depth >= 0) | |
| 5891 { | |
| 5892 if (fword[fwordidx[depth]] == NUL) | |
| 5893 { | |
| 5894 /* We are at the end of "fword". If the tree allows a word to end | |
| 5895 * here we have found a match. */ | |
| 5896 if (byts[arridx[depth] + 1] == 0) | |
| 5897 { | |
| 5898 kword[kwordlen[depth]] = NUL; | |
| 5899 return; | |
| 5900 } | |
| 5901 | |
| 5902 /* kword is getting too long, continue one level up */ | |
| 5903 --depth; | |
| 5904 } | |
| 5905 else if (++round[depth] > 2) | |
| 5906 { | |
| 5907 /* tried both fold-case and upper-case character, continue one | |
| 5908 * level up */ | |
| 5909 --depth; | |
| 5910 } | |
| 5911 else | |
| 5912 { | |
| 5913 /* | |
| 5914 * round[depth] == 1: Try using the folded-case character. | |
| 5915 * round[depth] == 2: Try using the upper-case character. | |
| 5916 */ | |
| 5917 #ifdef FEAT_MBYTE | |
| 5918 if (has_mbyte) | |
| 5919 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5920 flen = MB_CPTR2LEN(fword + fwordidx[depth]); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5921 ulen = MB_CPTR2LEN(uword + uwordidx[depth]); |
| 323 | 5922 } |
| 5923 else | |
| 5924 #endif | |
| 5925 ulen = flen = 1; | |
| 5926 if (round[depth] == 1) | |
| 5927 { | |
| 5928 p = fword + fwordidx[depth]; | |
| 5929 l = flen; | |
| 5930 } | |
| 5931 else | |
| 5932 { | |
| 5933 p = uword + uwordidx[depth]; | |
| 5934 l = ulen; | |
| 5935 } | |
| 5936 | |
| 5937 for (tryidx = arridx[depth]; l > 0; --l) | |
| 5938 { | |
| 5939 /* Perform a binary search in the list of accepted bytes. */ | |
| 5940 len = byts[tryidx++]; | |
| 5941 c = *p++; | |
| 5942 lo = tryidx; | |
| 5943 hi = tryidx + len - 1; | |
| 5944 while (lo < hi) | |
| 5945 { | |
| 5946 m = (lo + hi) / 2; | |
| 5947 if (byts[m] > c) | |
| 5948 hi = m - 1; | |
| 5949 else if (byts[m] < c) | |
| 5950 lo = m + 1; | |
| 5951 else | |
| 5952 { | |
| 5953 lo = hi = m; | |
| 5954 break; | |
| 5955 } | |
| 5956 } | |
| 5957 | |
| 5958 /* Stop if there is no matching byte. */ | |
| 5959 if (hi < lo || byts[lo] != c) | |
| 5960 break; | |
| 5961 | |
| 5962 /* Continue at the child (if there is one). */ | |
| 5963 tryidx = idxs[lo]; | |
| 5964 } | |
| 5965 | |
| 5966 if (l == 0) | |
| 5967 { | |
| 5968 /* | |
| 5969 * Found the matching char. Copy it to "kword" and go a | |
| 5970 * level deeper. | |
| 5971 */ | |
| 5972 if (round[depth] == 1) | |
| 5973 { | |
| 5974 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth], | |
| 5975 flen); | |
| 5976 kwordlen[depth + 1] = kwordlen[depth] + flen; | |
| 5977 } | |
| 5978 else | |
| 5979 { | |
| 5980 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth], | |
| 5981 ulen); | |
| 5982 kwordlen[depth + 1] = kwordlen[depth] + ulen; | |
| 5983 } | |
| 5984 fwordidx[depth + 1] = fwordidx[depth] + flen; | |
| 5985 uwordidx[depth + 1] = uwordidx[depth] + ulen; | |
| 5986 | |
| 5987 ++depth; | |
| 5988 arridx[depth] = tryidx; | |
| 5989 round[depth] = 0; | |
| 5990 } | |
| 5991 } | |
| 5992 } | |
| 5993 | |
| 5994 /* Didn't find it: "cannot happen". */ | |
| 5995 *kword = NUL; | |
| 5996 } | |
| 5997 | |
| 5998 /* | |
| 344 | 5999 * Compute the sound-a-like score for suggestions in su->su_ga and add them to |
| 6000 * su->su_sga. | |
| 6001 */ | |
| 6002 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6003 score_comp_sal(suginfo_T *su) |
| 344 | 6004 { |
| 6005 langp_T *lp; | |
| 6006 char_u badsound[MAXWLEN]; | |
| 6007 int i; | |
| 6008 suggest_T *stp; | |
| 6009 suggest_T *sstp; | |
| 6010 int score; | |
| 500 | 6011 int lpi; |
| 344 | 6012 |
| 6013 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL) | |
| 6014 return; | |
| 6015 | |
| 6016 /* 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
|
6017 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
|
6018 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6019 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 6020 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6021 { | |
| 6022 /* soundfold the bad word */ | |
| 375 | 6023 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound); |
| 344 | 6024 |
| 6025 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 6026 { | |
| 6027 stp = &SUG(su->su_ga, i); | |
| 6028 | |
| 351 | 6029 /* Case-fold the suggested word, sound-fold it and compute the |
| 6030 * sound-a-like score. */ | |
| 6031 score = stp_sal_score(stp, su, lp->lp_slang, badsound); | |
| 344 | 6032 if (score < SCORE_MAXMAX) |
| 6033 { | |
| 6034 /* Add the suggestion. */ | |
| 6035 sstp = &SUG(su->su_sga, su->su_sga.ga_len); | |
| 6036 sstp->st_word = vim_strsave(stp->st_word); | |
| 6037 if (sstp->st_word != NULL) | |
| 6038 { | |
| 625 | 6039 sstp->st_wordlen = stp->st_wordlen; |
| 344 | 6040 sstp->st_score = score; |
| 6041 sstp->st_altscore = 0; | |
| 6042 sstp->st_orglen = stp->st_orglen; | |
| 6043 ++su->su_sga.ga_len; | |
| 6044 } | |
| 6045 } | |
| 6046 } | |
| 6047 break; | |
| 6048 } | |
| 500 | 6049 } |
| 344 | 6050 } |
| 6051 | |
| 6052 /* | |
| 6053 * Combine the list of suggestions in su->su_ga and su->su_sga. | |
| 4352 | 6054 * They are entwined. |
| 344 | 6055 */ |
| 6056 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6057 score_combine(suginfo_T *su) |
| 344 | 6058 { |
| 6059 int i; | |
| 6060 int j; | |
| 6061 garray_T ga; | |
| 6062 garray_T *gap; | |
| 6063 langp_T *lp; | |
| 6064 suggest_T *stp; | |
| 6065 char_u *p; | |
| 6066 char_u badsound[MAXWLEN]; | |
| 6067 int round; | |
| 500 | 6068 int lpi; |
| 625 | 6069 slang_T *slang = NULL; |
| 344 | 6070 |
| 6071 /* 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
|
6072 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
|
6073 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6074 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 6075 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6076 { | |
| 6077 /* soundfold the bad word */ | |
| 625 | 6078 slang = lp->lp_slang; |
| 6079 spell_soundfold(slang, su->su_fbadword, TRUE, badsound); | |
| 344 | 6080 |
| 6081 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 6082 { | |
| 6083 stp = &SUG(su->su_ga, i); | |
| 625 | 6084 stp->st_altscore = stp_sal_score(stp, su, slang, badsound); |
| 344 | 6085 if (stp->st_altscore == SCORE_MAXMAX) |
| 6086 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4; | |
| 6087 else | |
| 6088 stp->st_score = (stp->st_score * 3 | |
| 6089 + stp->st_altscore) / 4; | |
| 6090 stp->st_salscore = FALSE; | |
| 6091 } | |
| 6092 break; | |
| 6093 } | |
| 6094 } | |
| 6095 | |
| 844 | 6096 if (slang == NULL) /* Using "double" without sound folding. */ |
| 6097 { | |
| 6098 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, | |
| 6099 su->su_maxcount); | |
| 625 | 6100 return; |
| 844 | 6101 } |
| 625 | 6102 |
| 344 | 6103 /* Add the alternate score to su_sga. */ |
| 6104 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 6105 { | |
| 6106 stp = &SUG(su->su_sga, i); | |
| 625 | 6107 stp->st_altscore = spell_edit_score(slang, |
| 6108 su->su_badword, stp->st_word); | |
| 344 | 6109 if (stp->st_score == SCORE_MAXMAX) |
| 6110 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8; | |
| 6111 else | |
| 6112 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8; | |
| 6113 stp->st_salscore = TRUE; | |
| 6114 } | |
| 6115 | |
| 625 | 6116 /* Remove bad suggestions, sort the suggestions and truncate at "maxcount" |
| 6117 * for both lists. */ | |
| 6118 check_suggestions(su, &su->su_ga); | |
| 344 | 6119 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 625 | 6120 check_suggestions(su, &su->su_sga); |
| 344 | 6121 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount); |
| 6122 | |
| 6123 ga_init2(&ga, (int)sizeof(suginfo_T), 1); | |
| 6124 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL) | |
| 6125 return; | |
| 6126 | |
| 6127 stp = &SUG(ga, 0); | |
| 6128 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i) | |
| 6129 { | |
| 6130 /* round 1: get a suggestion from su_ga | |
| 6131 * round 2: get a suggestion from su_sga */ | |
| 6132 for (round = 1; round <= 2; ++round) | |
| 6133 { | |
| 6134 gap = round == 1 ? &su->su_ga : &su->su_sga; | |
| 6135 if (i < gap->ga_len) | |
| 6136 { | |
| 6137 /* Don't add a word if it's already there. */ | |
| 6138 p = SUG(*gap, i).st_word; | |
| 6139 for (j = 0; j < ga.ga_len; ++j) | |
| 6140 if (STRCMP(stp[j].st_word, p) == 0) | |
| 6141 break; | |
| 6142 if (j == ga.ga_len) | |
| 6143 stp[ga.ga_len++] = SUG(*gap, i); | |
| 6144 else | |
| 6145 vim_free(p); | |
| 6146 } | |
| 6147 } | |
| 6148 } | |
| 6149 | |
| 6150 ga_clear(&su->su_ga); | |
| 6151 ga_clear(&su->su_sga); | |
| 6152 | |
| 6153 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 6154 if (ga.ga_len > su->su_maxcount) | |
| 6155 { | |
| 6156 for (i = su->su_maxcount; i < ga.ga_len; ++i) | |
| 6157 vim_free(stp[i].st_word); | |
| 6158 ga.ga_len = su->su_maxcount; | |
| 6159 } | |
| 6160 | |
| 6161 su->su_ga = ga; | |
| 6162 } | |
| 6163 | |
| 6164 /* | |
| 351 | 6165 * For the goodword in "stp" compute the soundalike score compared to the |
| 6166 * badword. | |
| 6167 */ | |
| 6168 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6169 stp_sal_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6170 suggest_T *stp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6171 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6172 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6173 char_u *badsound) /* sound-folded badword */ |
| 351 | 6174 { |
| 6175 char_u *p; | |
| 534 | 6176 char_u *pbad; |
| 6177 char_u *pgood; | |
| 351 | 6178 char_u badsound2[MAXWLEN]; |
| 6179 char_u fword[MAXWLEN]; | |
| 6180 char_u goodsound[MAXWLEN]; | |
| 534 | 6181 char_u goodword[MAXWLEN]; |
| 6182 int lendiff; | |
| 6183 | |
| 6184 lendiff = (int)(su->su_badlen - stp->st_orglen); | |
| 6185 if (lendiff >= 0) | |
| 6186 pbad = badsound; | |
| 351 | 6187 else |
| 6188 { | |
| 6189 /* soundfold the bad word with more characters following */ | |
| 6190 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN); | |
| 6191 | |
| 6192 /* When joining two words the sound often changes a lot. E.g., "t he" | |
| 6193 * sounds like "t h" while "the" sounds like "@". Avoid that by | |
| 6194 * removing the space. Don't do it when the good word also contains a | |
| 6195 * space. */ | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
6196 if (VIM_ISWHITE(su->su_badptr[su->su_badlen]) |
| 351 | 6197 && *skiptowhite(stp->st_word) == NUL) |
| 6198 for (p = fword; *(p = skiptowhite(p)) != NUL; ) | |
| 1619 | 6199 STRMOVE(p, p + 1); |
| 351 | 6200 |
| 375 | 6201 spell_soundfold(slang, fword, TRUE, badsound2); |
| 534 | 6202 pbad = badsound2; |
| 6203 } | |
| 6204 | |
| 2768 | 6205 if (lendiff > 0 && stp->st_wordlen + lendiff < MAXWLEN) |
| 534 | 6206 { |
| 6207 /* Add part of the bad word to the good word, so that we soundfold | |
| 6208 * what replaces the bad word. */ | |
| 6209 STRCPY(goodword, stp->st_word); | |
| 625 | 6210 vim_strncpy(goodword + stp->st_wordlen, |
| 6211 su->su_badptr + su->su_badlen - lendiff, lendiff); | |
| 534 | 6212 pgood = goodword; |
| 6213 } | |
| 6214 else | |
| 6215 pgood = stp->st_word; | |
| 351 | 6216 |
| 375 | 6217 /* Sound-fold the word and compute the score for the difference. */ |
| 534 | 6218 spell_soundfold(slang, pgood, FALSE, goodsound); |
| 6219 | |
| 6220 return soundalike_score(goodsound, pbad); | |
| 351 | 6221 } |
| 6222 | |
| 625 | 6223 /* structure used to store soundfolded words that add_sound_suggest() has |
| 6224 * handled already. */ | |
| 6225 typedef struct | |
| 6226 { | |
| 6227 short sft_score; /* lowest score used */ | |
| 6228 char_u sft_word[1]; /* soundfolded word, actually longer */ | |
| 6229 } sftword_T; | |
| 6230 | |
| 6231 static sftword_T dumsft; | |
| 6232 #define HIKEY2SFT(p) ((sftword_T *)(p - (dumsft.sft_word - (char_u *)&dumsft))) | |
| 6233 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key) | |
| 6234 | |
| 6235 /* | |
| 6236 * Prepare for calling suggest_try_soundalike(). | |
| 6237 */ | |
| 6238 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6239 suggest_try_soundalike_prep(void) |
| 625 | 6240 { |
| 6241 langp_T *lp; | |
| 6242 int lpi; | |
| 6243 slang_T *slang; | |
| 6244 | |
| 6245 /* Do this for all languages that support sound folding and for which a | |
| 6246 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6247 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
|
6248 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6249 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 6250 slang = lp->lp_slang; |
| 6251 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 6252 /* prepare the hashtable used by add_sound_suggest() */ | |
| 6253 hash_init(&slang->sl_sounddone); | |
| 6254 } | |
| 6255 } | |
| 6256 | |
| 351 | 6257 /* |
| 323 | 6258 * Find suggestions by comparing the word in a sound-a-like form. |
| 503 | 6259 * Note: This doesn't support postponed prefixes. |
| 323 | 6260 */ |
| 6261 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6262 suggest_try_soundalike(suginfo_T *su) |
| 323 | 6263 { |
| 6264 char_u salword[MAXWLEN]; | |
| 6265 langp_T *lp; | |
| 500 | 6266 int lpi; |
| 503 | 6267 slang_T *slang; |
| 344 | 6268 |
| 625 | 6269 /* Do this for all languages that support sound folding and for which a |
| 6270 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6271 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
|
6272 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6273 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 6274 slang = lp->lp_slang; |
| 6275 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 6276 { | |
| 6277 /* soundfold the bad word */ | |
| 6278 spell_soundfold(slang, su->su_fbadword, TRUE, salword); | |
| 6279 | |
| 6280 /* try all kinds of inserts/deletes/swaps/etc. */ | |
| 6281 /* TODO: also soundfold the next words, so that we can try joining | |
| 6282 * and splitting */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6283 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6284 prof_init(); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6285 #endif |
| 625 | 6286 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
|
6287 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6288 prof_report("soundalike"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6289 #endif |
| 625 | 6290 } |
| 6291 } | |
| 6292 } | |
| 6293 | |
| 6294 /* | |
| 6295 * Finish up after calling suggest_try_soundalike(). | |
| 6296 */ | |
| 6297 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6298 suggest_try_soundalike_finish(void) |
| 625 | 6299 { |
| 6300 langp_T *lp; | |
| 6301 int lpi; | |
| 6302 slang_T *slang; | |
| 6303 int todo; | |
| 6304 hashitem_T *hi; | |
| 6305 | |
| 6306 /* Do this for all languages that support sound folding and for which a | |
| 6307 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6308 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
|
6309 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6310 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 503 | 6311 slang = lp->lp_slang; |
| 625 | 6312 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) |
| 6313 { | |
| 6314 /* Free the info about handled words. */ | |
| 835 | 6315 todo = (int)slang->sl_sounddone.ht_used; |
| 625 | 6316 for (hi = slang->sl_sounddone.ht_array; todo > 0; ++hi) |
| 6317 if (!HASHITEM_EMPTY(hi)) | |
| 6318 { | |
| 6319 vim_free(HI2SFT(hi)); | |
| 6320 --todo; | |
| 6321 } | |
| 1087 | 6322 |
| 6323 /* Clear the hashtable, it may also be used by another region. */ | |
| 625 | 6324 hash_clear(&slang->sl_sounddone); |
| 1087 | 6325 hash_init(&slang->sl_sounddone); |
| 625 | 6326 } |
| 6327 } | |
| 6328 } | |
| 6329 | |
| 6330 /* | |
| 6331 * A match with a soundfolded word is found. Add the good word(s) that | |
| 6332 * produce this soundfolded word. | |
| 6333 */ | |
| 6334 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6335 add_sound_suggest( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6336 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6337 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6338 int score, /* soundfold score */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6339 langp_T *lp) |
| 625 | 6340 { |
| 6341 slang_T *slang = lp->lp_slang; /* language for sound folding */ | |
| 6342 int sfwordnr; | |
| 6343 char_u *nrline; | |
| 6344 int orgnr; | |
| 6345 char_u theword[MAXWLEN]; | |
| 6346 int i; | |
| 6347 int wlen; | |
| 6348 char_u *byts; | |
| 6349 idx_T *idxs; | |
| 6350 int n; | |
| 6351 int wordcount; | |
| 6352 int wc; | |
| 6353 int goodscore; | |
| 6354 hash_T hash; | |
| 6355 hashitem_T *hi; | |
| 6356 sftword_T *sft; | |
| 6357 int bc, gc; | |
| 6358 int limit; | |
| 6359 | |
| 6360 /* | |
| 6361 * It's very well possible that the same soundfold word is found several | |
| 6362 * times with different scores. Since the following is quite slow only do | |
| 6363 * the words that have a better score than before. Use a hashtable to | |
| 6364 * remember the words that have been done. | |
| 6365 */ | |
| 6366 hash = hash_hash(goodword); | |
| 6367 hi = hash_lookup(&slang->sl_sounddone, goodword, hash); | |
| 6368 if (HASHITEM_EMPTY(hi)) | |
| 6369 { | |
| 844 | 6370 sft = (sftword_T *)alloc((unsigned)(sizeof(sftword_T) |
| 6371 + STRLEN(goodword))); | |
| 625 | 6372 if (sft != NULL) |
| 6373 { | |
| 6374 sft->sft_score = score; | |
| 6375 STRCPY(sft->sft_word, goodword); | |
| 6376 hash_add_item(&slang->sl_sounddone, hi, sft->sft_word, hash); | |
| 6377 } | |
| 6378 } | |
| 6379 else | |
| 6380 { | |
| 6381 sft = HI2SFT(hi); | |
| 6382 if (score >= sft->sft_score) | |
| 6383 return; | |
| 6384 sft->sft_score = score; | |
| 6385 } | |
| 6386 | |
| 6387 /* | |
| 6388 * Find the word nr in the soundfold tree. | |
| 6389 */ | |
| 6390 sfwordnr = soundfold_find(slang, goodword); | |
| 6391 if (sfwordnr < 0) | |
| 6392 { | |
|
10359
66f1b5bf3fa6
commit https://github.com/vim/vim/commit/95f096030ed1a8afea028f2ea295d6f6a70f466f
Christian Brabandt <cb@256bit.org>
parents:
10277
diff
changeset
|
6393 internal_error("add_sound_suggest()"); |
| 625 | 6394 return; |
| 6395 } | |
| 6396 | |
| 6397 /* | |
| 6398 * go over the list of good words that produce this soundfold word | |
| 6399 */ | |
| 6400 nrline = ml_get_buf(slang->sl_sugbuf, (linenr_T)(sfwordnr + 1), FALSE); | |
| 6401 orgnr = 0; | |
| 6402 while (*nrline != NUL) | |
| 6403 { | |
| 6404 /* The wordnr was stored in a minimal nr of bytes as an offset to the | |
| 6405 * previous wordnr. */ | |
| 6406 orgnr += bytes2offset(&nrline); | |
| 6407 | |
| 6408 byts = slang->sl_fbyts; | |
| 6409 idxs = slang->sl_fidxs; | |
| 6410 | |
| 6411 /* Lookup the word "orgnr" one of the two tries. */ | |
| 6412 n = 0; | |
| 6413 wordcount = 0; | |
| 5506 | 6414 for (wlen = 0; wlen < MAXWLEN - 3; ++wlen) |
| 625 | 6415 { |
| 6416 i = 1; | |
| 6417 if (wordcount == orgnr && byts[n + 1] == NUL) | |
| 6418 break; /* found end of word */ | |
| 6419 | |
| 6420 if (byts[n + 1] == NUL) | |
| 6421 ++wordcount; | |
| 6422 | |
| 6423 /* skip over the NUL bytes */ | |
| 6424 for ( ; byts[n + i] == NUL; ++i) | |
| 6425 if (i > byts[n]) /* safety check */ | |
| 6426 { | |
| 6427 STRCPY(theword + wlen, "BAD"); | |
| 5506 | 6428 wlen += 3; |
| 625 | 6429 goto badword; |
| 6430 } | |
| 6431 | |
| 6432 /* One of the siblings must have the word. */ | |
| 6433 for ( ; i < byts[n]; ++i) | |
| 6434 { | |
| 6435 wc = idxs[idxs[n + i]]; /* nr of words under this byte */ | |
| 6436 if (wordcount + wc > orgnr) | |
| 6437 break; | |
| 6438 wordcount += wc; | |
| 6439 } | |
| 6440 | |
| 5506 | 6441 theword[wlen] = byts[n + i]; |
| 625 | 6442 n = idxs[n + i]; |
| 6443 } | |
| 6444 badword: | |
| 6445 theword[wlen] = NUL; | |
| 6446 | |
| 6447 /* Go over the possible flags and regions. */ | |
| 6448 for (; i <= byts[n] && byts[n + i] == NUL; ++i) | |
| 6449 { | |
| 6450 char_u cword[MAXWLEN]; | |
| 6451 char_u *p; | |
| 6452 int flags = (int)idxs[n + i]; | |
| 6453 | |
| 698 | 6454 /* Skip words with the NOSUGGEST flag */ |
| 6455 if (flags & WF_NOSUGGEST) | |
| 6456 continue; | |
| 6457 | |
| 625 | 6458 if (flags & WF_KEEPCAP) |
| 6459 { | |
| 6460 /* Must find the word in the keep-case tree. */ | |
| 6461 find_keepcap_word(slang, theword, cword); | |
| 6462 p = cword; | |
| 6463 } | |
| 6464 else | |
| 6465 { | |
| 6466 flags |= su->su_badflags; | |
| 6467 if ((flags & WF_CAPMASK) != 0) | |
| 6468 { | |
| 6469 /* Need to fix case according to "flags". */ | |
| 6470 make_case_word(theword, cword, flags); | |
| 6471 p = cword; | |
| 323 | 6472 } |
| 6473 else | |
| 625 | 6474 p = theword; |
| 6475 } | |
| 6476 | |
| 6477 /* Add the suggestion. */ | |
| 6478 if (sps_flags & SPS_DOUBLE) | |
| 6479 { | |
| 6480 /* Add the suggestion if the score isn't too bad. */ | |
| 6481 if (score <= su->su_maxscore) | |
| 6482 add_suggestion(su, &su->su_sga, p, su->su_badlen, | |
| 6483 score, 0, FALSE, slang, FALSE); | |
| 6484 } | |
| 6485 else | |
| 6486 { | |
| 6487 /* Add a penalty for words in another region. */ | |
| 6488 if ((flags & WF_REGION) | |
| 6489 && (((unsigned)flags >> 16) & lp->lp_region) == 0) | |
| 6490 goodscore = SCORE_REGION; | |
| 6491 else | |
| 6492 goodscore = 0; | |
| 6493 | |
| 6494 /* Add a small penalty for changing the first letter from | |
| 6495 * lower to upper case. Helps for "tath" -> "Kath", which is | |
| 4352 | 6496 * less common than "tath" -> "path". Don't do it when the |
| 625 | 6497 * letter is the same, that has already been counted. */ |
| 6498 gc = PTR2CHAR(p); | |
| 6499 if (SPELL_ISUPPER(gc)) | |
| 6500 { | |
| 6501 bc = PTR2CHAR(su->su_badword); | |
| 6502 if (!SPELL_ISUPPER(bc) | |
| 6503 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc)) | |
| 6504 goodscore += SCORE_ICASE / 2; | |
| 6505 } | |
| 6506 | |
| 6507 /* Compute the score for the good word. This only does letter | |
| 6508 * insert/delete/swap/replace. REP items are not considered, | |
| 6509 * which may make the score a bit higher. | |
| 6510 * Use a limit for the score to make it work faster. Use | |
| 6511 * MAXSCORE(), because RESCORE() will change the score. | |
| 6512 * If the limit is very high then the iterative method is | |
| 6513 * inefficient, using an array is quicker. */ | |
| 6514 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score); | |
| 6515 if (limit > SCORE_LIMITMAX) | |
| 6516 goodscore += spell_edit_score(slang, su->su_badword, p); | |
| 6517 else | |
| 6518 goodscore += spell_edit_score_limit(slang, su->su_badword, | |
| 6519 p, limit); | |
| 6520 | |
| 6521 /* When going over the limit don't bother to do the rest. */ | |
| 6522 if (goodscore < SCORE_MAXMAX) | |
| 6523 { | |
| 6524 /* Give a bonus to words seen before. */ | |
| 6525 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE); | |
| 6526 | |
| 6527 /* Add the suggestion if the score isn't too bad. */ | |
| 6528 goodscore = RESCORE(goodscore, score); | |
| 6529 if (goodscore <= su->su_sfmaxscore) | |
| 6530 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 6531 goodscore, score, TRUE, slang, TRUE); | |
| 6532 } | |
| 6533 } | |
| 6534 } | |
| 6535 /* smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr); */ | |
| 6536 } | |
| 6537 } | |
| 6538 | |
| 6539 /* | |
| 6540 * Find word "word" in fold-case tree for "slang" and return the word number. | |
| 6541 */ | |
| 6542 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6543 soundfold_find(slang_T *slang, char_u *word) |
| 625 | 6544 { |
| 6545 idx_T arridx = 0; | |
| 6546 int len; | |
| 6547 int wlen = 0; | |
| 6548 int c; | |
| 6549 char_u *ptr = word; | |
| 6550 char_u *byts; | |
| 6551 idx_T *idxs; | |
| 6552 int wordnr = 0; | |
| 6553 | |
| 6554 byts = slang->sl_sbyts; | |
| 6555 idxs = slang->sl_sidxs; | |
| 6556 | |
| 6557 for (;;) | |
| 6558 { | |
| 6559 /* First byte is the number of possible bytes. */ | |
| 6560 len = byts[arridx++]; | |
| 6561 | |
| 6562 /* If the first possible byte is a zero the word could end here. | |
| 6563 * If the word ends we found the word. If not skip the NUL bytes. */ | |
| 6564 c = ptr[wlen]; | |
| 6565 if (byts[arridx] == NUL) | |
| 6566 { | |
| 6567 if (c == NUL) | |
| 6568 break; | |
| 6569 | |
| 6570 /* Skip over the zeros, there can be several. */ | |
| 6571 while (len > 0 && byts[arridx] == NUL) | |
| 6572 { | |
| 6573 ++arridx; | |
| 6574 --len; | |
| 6575 } | |
| 6576 if (len == 0) | |
| 6577 return -1; /* no children, word should have ended here */ | |
| 6578 ++wordnr; | |
| 6579 } | |
| 6580 | |
| 6581 /* If the word ends we didn't find it. */ | |
| 6582 if (c == NUL) | |
| 6583 return -1; | |
| 6584 | |
| 6585 /* Perform a binary search in the list of accepted bytes. */ | |
| 6586 if (c == TAB) /* <Tab> is handled like <Space> */ | |
| 6587 c = ' '; | |
| 6588 while (byts[arridx] < c) | |
| 6589 { | |
| 6590 /* The word count is in the first idxs[] entry of the child. */ | |
| 6591 wordnr += idxs[idxs[arridx]]; | |
| 6592 ++arridx; | |
| 6593 if (--len == 0) /* end of the bytes, didn't find it */ | |
| 6594 return -1; | |
| 6595 } | |
| 6596 if (byts[arridx] != c) /* didn't find the byte */ | |
| 6597 return -1; | |
| 6598 | |
| 6599 /* Continue at the child (if there is one). */ | |
| 6600 arridx = idxs[arridx]; | |
| 6601 ++wlen; | |
| 6602 | |
| 6603 /* One space in the good word may stand for several spaces in the | |
| 6604 * checked word. */ | |
| 6605 if (c == ' ') | |
| 6606 while (ptr[wlen] == ' ' || ptr[wlen] == TAB) | |
| 6607 ++wlen; | |
| 6608 } | |
| 6609 | |
| 6610 return wordnr; | |
| 323 | 6611 } |
| 6612 | |
| 6613 /* | |
| 324 | 6614 * Copy "fword" to "cword", fixing case according to "flags". |
| 323 | 6615 */ |
| 6616 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6617 make_case_word(char_u *fword, char_u *cword, int flags) |
| 323 | 6618 { |
| 6619 if (flags & WF_ALLCAP) | |
| 6620 /* Make it all upper-case */ | |
| 6621 allcap_copy(fword, cword); | |
| 6622 else if (flags & WF_ONECAP) | |
| 6623 /* Make the first letter upper-case */ | |
| 324 | 6624 onecap_copy(fword, cword, TRUE); |
| 323 | 6625 else |
| 6626 /* Use goodword as-is. */ | |
| 6627 STRCPY(cword, fword); | |
| 6628 } | |
| 6629 | |
| 330 | 6630 |
| 323 | 6631 /* |
| 6632 * Return TRUE if "c1" and "c2" are similar characters according to the MAP | |
| 6633 * lines in the .aff file. | |
| 6634 */ | |
| 6635 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6636 similar_chars(slang_T *slang, int c1, int c2) |
| 323 | 6637 { |
| 330 | 6638 int m1, m2; |
| 6639 #ifdef FEAT_MBYTE | |
| 3549 | 6640 char_u buf[MB_MAXBYTES + 1]; |
| 330 | 6641 hashitem_T *hi; |
| 6642 | |
| 6643 if (c1 >= 256) | |
| 6644 { | |
| 6645 buf[mb_char2bytes(c1, buf)] = 0; | |
| 6646 hi = hash_find(&slang->sl_map_hash, buf); | |
| 6647 if (HASHITEM_EMPTY(hi)) | |
| 6648 m1 = 0; | |
| 6649 else | |
| 6650 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 6651 } | |
| 6652 else | |
| 6653 #endif | |
| 6654 m1 = slang->sl_map_array[c1]; | |
| 6655 if (m1 == 0) | |
| 323 | 6656 return FALSE; |
| 330 | 6657 |
| 6658 | |
| 6659 #ifdef FEAT_MBYTE | |
| 6660 if (c2 >= 256) | |
| 6661 { | |
| 6662 buf[mb_char2bytes(c2, buf)] = 0; | |
| 6663 hi = hash_find(&slang->sl_map_hash, buf); | |
| 6664 if (HASHITEM_EMPTY(hi)) | |
| 6665 m2 = 0; | |
| 6666 else | |
| 6667 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 6668 } | |
| 6669 else | |
| 6670 #endif | |
| 6671 m2 = slang->sl_map_array[c2]; | |
| 6672 | |
| 6673 return m1 == m2; | |
| 323 | 6674 } |
| 6675 | |
| 6676 /* | |
| 6677 * Add a suggestion to the list of suggestions. | |
| 625 | 6678 * For a suggestion that is already in the list the lowest score is remembered. |
| 6679 */ | |
| 6680 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6681 add_suggestion( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6682 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6683 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
|
6684 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6685 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
|
6686 int score, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6687 int altscore, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6688 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
|
6689 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
|
6690 int maxsf) /* su_maxscore applies to soundfold score, |
| 625 | 6691 su_sfmaxscore to the total score. */ |
| 6692 { | |
| 6693 int goodlen; /* len of goodword changed */ | |
| 6694 int badlen; /* len of bad word changed */ | |
| 323 | 6695 suggest_T *stp; |
| 534 | 6696 suggest_T new_sug; |
| 323 | 6697 int i; |
| 534 | 6698 char_u *pgood, *pbad; |
| 532 | 6699 |
| 534 | 6700 /* Minimize "badlen" for consistency. Avoids that changing "the the" to |
| 6701 * "thee the" is added next to changing the first "the" the "thee". */ | |
| 6702 pgood = goodword + STRLEN(goodword); | |
| 625 | 6703 pbad = su->su_badptr + badlenarg; |
| 6704 for (;;) | |
| 6705 { | |
| 835 | 6706 goodlen = (int)(pgood - goodword); |
| 6707 badlen = (int)(pbad - su->su_badptr); | |
| 625 | 6708 if (goodlen <= 0 || badlen <= 0) |
| 6709 break; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
6710 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
|
6711 MB_PTR_BACK(su->su_badptr, pbad); |
| 534 | 6712 #ifdef FEAT_MBYTE |
| 6713 if (has_mbyte) | |
| 6714 { | |
| 6715 if (mb_ptr2char(pgood) != mb_ptr2char(pbad)) | |
| 6716 break; | |
| 346 | 6717 } |
| 6718 else | |
| 534 | 6719 #endif |
| 6720 if (*pgood != *pbad) | |
| 6721 break; | |
| 625 | 6722 } |
| 6723 | |
| 534 | 6724 if (badlen == 0 && goodlen == 0) |
| 6725 /* goodword doesn't change anything; may happen for "the the" changing | |
| 6726 * the first "the" to itself. */ | |
| 6727 return; | |
| 346 | 6728 |
| 944 | 6729 if (gap->ga_len == 0) |
| 6730 i = -1; | |
| 6731 else | |
| 6732 { | |
| 6733 /* Check if the word is already there. Also check the length that is | |
| 6734 * being replaced "thes," -> "these" is a different suggestion from | |
| 6735 * "thes" -> "these". */ | |
| 6736 stp = &SUG(*gap, 0); | |
| 6737 for (i = gap->ga_len; --i >= 0; ++stp) | |
| 6738 if (stp->st_wordlen == goodlen | |
| 6739 && stp->st_orglen == badlen | |
| 6740 && STRNCMP(stp->st_word, goodword, goodlen) == 0) | |
| 6741 { | |
| 6742 /* | |
| 6743 * Found it. Remember the word with the lowest score. | |
| 6744 */ | |
| 6745 if (stp->st_slang == NULL) | |
| 6746 stp->st_slang = slang; | |
| 6747 | |
| 6748 new_sug.st_score = score; | |
| 6749 new_sug.st_altscore = altscore; | |
| 6750 new_sug.st_had_bonus = had_bonus; | |
| 6751 | |
| 6752 if (stp->st_had_bonus != had_bonus) | |
| 6753 { | |
| 6754 /* Only one of the two had the soundalike score computed. | |
| 6755 * Need to do that for the other one now, otherwise the | |
| 6756 * scores can't be compared. This happens because | |
| 6757 * suggest_try_change() doesn't compute the soundalike | |
| 6758 * word to keep it fast, while some special methods set | |
| 6759 * the soundalike score to zero. */ | |
| 6760 if (had_bonus) | |
| 6761 rescore_one(su, stp); | |
| 6762 else | |
| 6763 { | |
| 6764 new_sug.st_word = stp->st_word; | |
| 6765 new_sug.st_wordlen = stp->st_wordlen; | |
| 6766 new_sug.st_slang = stp->st_slang; | |
| 6767 new_sug.st_orglen = badlen; | |
| 6768 rescore_one(su, &new_sug); | |
| 6769 } | |
| 6770 } | |
| 6771 | |
| 6772 if (stp->st_score > new_sug.st_score) | |
| 6773 { | |
| 6774 stp->st_score = new_sug.st_score; | |
| 6775 stp->st_altscore = new_sug.st_altscore; | |
| 6776 stp->st_had_bonus = new_sug.st_had_bonus; | |
| 6777 } | |
| 6778 break; | |
| 6779 } | |
| 6780 } | |
| 625 | 6781 |
| 6782 if (i < 0 && ga_grow(gap, 1) == OK) | |
| 6783 { | |
| 6784 /* Add a suggestion. */ | |
| 6785 stp = &SUG(*gap, gap->ga_len); | |
| 6786 stp->st_word = vim_strnsave(goodword, goodlen); | |
| 6787 if (stp->st_word != NULL) | |
| 6788 { | |
| 6789 stp->st_wordlen = goodlen; | |
| 6790 stp->st_score = score; | |
| 6791 stp->st_altscore = altscore; | |
| 6792 stp->st_had_bonus = had_bonus; | |
| 6793 stp->st_orglen = badlen; | |
| 6794 stp->st_slang = slang; | |
| 6795 ++gap->ga_len; | |
| 6796 | |
| 6797 /* If we have too many suggestions now, sort the list and keep | |
| 6798 * the best suggestions. */ | |
| 6799 if (gap->ga_len > SUG_MAX_COUNT(su)) | |
| 6800 { | |
| 6801 if (maxsf) | |
| 6802 su->su_sfmaxscore = cleanup_suggestions(gap, | |
| 6803 su->su_sfmaxscore, SUG_CLEAN_COUNT(su)); | |
| 6804 else | |
| 6805 su->su_maxscore = cleanup_suggestions(gap, | |
| 6806 su->su_maxscore, SUG_CLEAN_COUNT(su)); | |
| 6807 } | |
| 6808 } | |
| 6809 } | |
| 6810 } | |
| 6811 | |
| 6812 /* | |
| 6813 * Suggestions may in fact be flagged as errors. Esp. for banned words and | |
| 6814 * for split words, such as "the the". Remove these from the list here. | |
| 6815 */ | |
| 6816 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6817 check_suggestions( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6818 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6819 garray_T *gap) /* either su_ga or su_sga */ |
| 625 | 6820 { |
| 6821 suggest_T *stp; | |
| 6822 int i; | |
| 6823 char_u longword[MAXWLEN + 1]; | |
| 6824 int len; | |
| 6825 hlf_T attr; | |
| 6826 | |
| 6827 stp = &SUG(*gap, 0); | |
| 6828 for (i = gap->ga_len - 1; i >= 0; --i) | |
| 6829 { | |
| 6830 /* Need to append what follows to check for "the the". */ | |
| 2768 | 6831 vim_strncpy(longword, stp[i].st_word, MAXWLEN); |
| 625 | 6832 len = stp[i].st_wordlen; |
| 6833 vim_strncpy(longword + len, su->su_badptr + stp[i].st_orglen, | |
| 6834 MAXWLEN - len); | |
| 6835 attr = HLF_COUNT; | |
| 6836 (void)spell_check(curwin, longword, &attr, NULL, FALSE); | |
| 6837 if (attr != HLF_COUNT) | |
| 6838 { | |
| 6839 /* Remove this entry. */ | |
| 6840 vim_free(stp[i].st_word); | |
| 6841 --gap->ga_len; | |
| 6842 if (i < gap->ga_len) | |
| 6843 mch_memmove(stp + i, stp + i + 1, | |
| 6844 sizeof(suggest_T) * (gap->ga_len - i)); | |
| 6845 } | |
| 6846 } | |
| 6847 } | |
| 6848 | |
| 323 | 6849 |
| 6850 /* | |
| 6851 * Add a word to be banned. | |
| 6852 */ | |
| 6853 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6854 add_banned( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6855 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6856 char_u *word) |
| 323 | 6857 { |
| 626 | 6858 char_u *s; |
| 323 | 6859 hash_T hash; |
| 6860 hashitem_T *hi; | |
| 6861 | |
| 625 | 6862 hash = hash_hash(word); |
| 6863 hi = hash_lookup(&su->su_banned, word, hash); | |
| 6864 if (HASHITEM_EMPTY(hi)) | |
| 6865 { | |
| 6866 s = vim_strsave(word); | |
| 6867 if (s != NULL) | |
| 323 | 6868 hash_add_item(&su->su_banned, hi, s, hash); |
| 625 | 6869 } |
| 323 | 6870 } |
| 6871 | |
| 324 | 6872 /* |
| 534 | 6873 * Recompute the score for all suggestions if sound-folding is possible. This |
| 6874 * is slow, thus only done for the final results. | |
| 324 | 6875 */ |
| 6876 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6877 rescore_suggestions(suginfo_T *su) |
| 324 | 6878 { |
| 6879 int i; | |
| 534 | 6880 |
| 6881 if (su->su_sallang != NULL) | |
| 503 | 6882 for (i = 0; i < su->su_ga.ga_len; ++i) |
| 534 | 6883 rescore_one(su, &SUG(su->su_ga, i)); |
| 6884 } | |
| 6885 | |
| 6886 /* | |
| 6887 * Recompute the score for one suggestion if sound-folding is possible. | |
| 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_one(suginfo_T *su, suggest_T *stp) |
| 534 | 6891 { |
| 6892 slang_T *slang = stp->st_slang; | |
| 6893 char_u sal_badword[MAXWLEN]; | |
| 535 | 6894 char_u *p; |
| 534 | 6895 |
| 6896 /* Only rescore suggestions that have no sal score yet and do have a | |
| 6897 * language. */ | |
| 6898 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus) | |
| 6899 { | |
| 6900 if (slang == su->su_sallang) | |
| 535 | 6901 p = su->su_sal_badword; |
| 534 | 6902 else |
| 6903 { | |
| 6904 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword); | |
| 535 | 6905 p = sal_badword; |
| 6906 } | |
| 6907 | |
| 6908 stp->st_altscore = stp_sal_score(stp, su, slang, p); | |
| 534 | 6909 if (stp->st_altscore == SCORE_MAXMAX) |
| 6910 stp->st_altscore = SCORE_BIG; | |
| 6911 stp->st_score = RESCORE(stp->st_score, stp->st_altscore); | |
| 6912 stp->st_had_bonus = TRUE; | |
| 324 | 6913 } |
| 6914 } | |
| 6915 | |
| 323 | 6916 static int |
| 6917 #ifdef __BORLANDC__ | |
| 6918 _RTLENTRYF | |
| 6919 #endif | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
6920 sug_compare(const void *s1, const void *s2); |
| 323 | 6921 |
| 6922 /* | |
| 6923 * Function given to qsort() to sort the suggestions on st_score. | |
| 529 | 6924 * First on "st_score", then "st_altscore" then alphabetically. |
| 323 | 6925 */ |
| 6926 static int | |
| 6927 #ifdef __BORLANDC__ | |
| 6928 _RTLENTRYF | |
| 6929 #endif | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6930 sug_compare(const void *s1, const void *s2) |
| 323 | 6931 { |
| 6932 suggest_T *p1 = (suggest_T *)s1; | |
| 6933 suggest_T *p2 = (suggest_T *)s2; | |
| 344 | 6934 int n = p1->st_score - p2->st_score; |
| 6935 | |
| 6936 if (n == 0) | |
| 529 | 6937 { |
| 6938 n = p1->st_altscore - p2->st_altscore; | |
| 6939 if (n == 0) | |
| 6940 n = STRICMP(p1->st_word, p2->st_word); | |
| 6941 } | |
| 344 | 6942 return n; |
| 323 | 6943 } |
| 6944 | |
| 6945 /* | |
| 6946 * Cleanup the suggestions: | |
| 6947 * - Sort on score. | |
| 6948 * - Remove words that won't be displayed. | |
| 344 | 6949 * Returns the maximum score in the list or "maxscore" unmodified. |
| 323 | 6950 */ |
| 344 | 6951 static int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6952 cleanup_suggestions( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6953 garray_T *gap, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6954 int maxscore, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6955 int keep) /* nr of suggestions to keep */ |
| 323 | 6956 { |
| 344 | 6957 suggest_T *stp = &SUG(*gap, 0); |
| 323 | 6958 int i; |
| 6959 | |
| 6960 /* Sort the list. */ | |
| 344 | 6961 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare); |
| 323 | 6962 |
| 6963 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 344 | 6964 if (gap->ga_len > keep) |
| 323 | 6965 { |
| 344 | 6966 for (i = keep; i < gap->ga_len; ++i) |
| 323 | 6967 vim_free(stp[i].st_word); |
| 344 | 6968 gap->ga_len = keep; |
| 6969 return stp[keep - 1].st_score; | |
| 323 | 6970 } |
| 344 | 6971 return maxscore; |
| 323 | 6972 } |
| 6973 | |
| 372 | 6974 #if defined(FEAT_EVAL) || defined(PROTO) |
| 6975 /* | |
| 6976 * Soundfold a string, for soundfold(). | |
| 6977 * Result is in allocated memory, NULL for an error. | |
| 6978 */ | |
| 6979 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6980 eval_soundfold(char_u *word) |
| 372 | 6981 { |
| 6982 langp_T *lp; | |
| 6983 char_u sound[MAXWLEN]; | |
| 500 | 6984 int lpi; |
| 372 | 6985 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6986 if (curwin->w_p_spell && *curwin->w_s->b_p_spl != NUL) |
| 372 | 6987 /* 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
|
6988 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
|
6989 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6990 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 372 | 6991 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6992 { | |
| 6993 /* soundfold the word */ | |
| 375 | 6994 spell_soundfold(lp->lp_slang, word, FALSE, sound); |
| 372 | 6995 return vim_strsave(sound); |
| 6996 } | |
| 500 | 6997 } |
| 372 | 6998 |
| 6999 /* No language with sound folding, return word as-is. */ | |
| 7000 return vim_strsave(word); | |
| 7001 } | |
| 7002 #endif | |
| 7003 | |
| 323 | 7004 /* |
| 7005 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 485 | 7006 * |
| 7007 * There are many ways to turn a word into a sound-a-like representation. The | |
| 7008 * oldest is Soundex (1918!). A nice overview can be found in "Approximate | |
| 7009 * swedish name matching - survey and test of different algorithms" by Klas | |
| 7010 * Erikson. | |
| 7011 * | |
| 7012 * We support two methods: | |
| 7013 * 1. SOFOFROM/SOFOTO do a simple character mapping. | |
| 7014 * 2. SAL items define a more advanced sound-folding (and much slower). | |
| 323 | 7015 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
7016 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7017 spell_soundfold( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7018 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7019 char_u *inword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7020 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
|
7021 char_u *res) |
| 375 | 7022 { |
| 7023 char_u fword[MAXWLEN]; | |
| 7024 char_u *word; | |
| 7025 | |
| 7026 if (slang->sl_sofo) | |
| 7027 /* SOFOFROM and SOFOTO used */ | |
| 7028 spell_soundfold_sofo(slang, inword, res); | |
| 7029 else | |
| 7030 { | |
| 7031 /* SAL items used. Requires the word to be case-folded. */ | |
| 7032 if (folded) | |
| 7033 word = inword; | |
| 7034 else | |
| 7035 { | |
| 835 | 7036 (void)spell_casefold(inword, (int)STRLEN(inword), fword, MAXWLEN); |
| 375 | 7037 word = fword; |
| 7038 } | |
| 7039 | |
| 7040 #ifdef FEAT_MBYTE | |
| 7041 if (has_mbyte) | |
| 7042 spell_soundfold_wsal(slang, word, res); | |
| 7043 else | |
| 7044 #endif | |
| 7045 spell_soundfold_sal(slang, word, res); | |
| 7046 } | |
| 7047 } | |
| 7048 | |
| 7049 /* | |
| 7050 * Perform sound folding of "inword" into "res" according to SOFOFROM and | |
| 7051 * SOFOTO lines. | |
| 7052 */ | |
| 7053 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7054 spell_soundfold_sofo(slang_T *slang, char_u *inword, char_u *res) |
| 375 | 7055 { |
| 7056 char_u *s; | |
| 7057 int ri = 0; | |
| 7058 int c; | |
| 7059 | |
| 7060 #ifdef FEAT_MBYTE | |
| 7061 if (has_mbyte) | |
| 7062 { | |
| 7063 int prevc = 0; | |
| 7064 int *ip; | |
| 7065 | |
| 7066 /* The sl_sal_first[] table contains the translation for chars up to | |
| 7067 * 255, sl_sal the rest. */ | |
| 7068 for (s = inword; *s != NUL; ) | |
| 7069 { | |
| 474 | 7070 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
|
7071 if (enc_utf8 ? utf_class(c) == 0 : VIM_ISWHITE(c)) |
| 375 | 7072 c = ' '; |
| 7073 else if (c < 256) | |
| 7074 c = slang->sl_sal_first[c]; | |
| 7075 else | |
| 7076 { | |
| 7077 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff]; | |
| 7078 if (ip == NULL) /* empty list, can't match */ | |
| 7079 c = NUL; | |
| 7080 else | |
| 7081 for (;;) /* find "c" in the list */ | |
| 7082 { | |
| 7083 if (*ip == 0) /* not found */ | |
| 7084 { | |
| 7085 c = NUL; | |
| 7086 break; | |
| 7087 } | |
| 7088 if (*ip == c) /* match! */ | |
| 7089 { | |
| 7090 c = ip[1]; | |
| 7091 break; | |
| 7092 } | |
| 7093 ip += 2; | |
| 7094 } | |
| 7095 } | |
| 7096 | |
| 7097 if (c != NUL && c != prevc) | |
| 7098 { | |
| 7099 ri += mb_char2bytes(c, res + ri); | |
| 7100 if (ri + MB_MAXBYTES > MAXWLEN) | |
| 7101 break; | |
| 7102 prevc = c; | |
| 7103 } | |
| 7104 } | |
| 7105 } | |
| 7106 else | |
| 7107 #endif | |
| 7108 { | |
| 7109 /* The sl_sal_first[] table contains the translation. */ | |
| 7110 for (s = inword; (c = *s) != NUL; ++s) | |
| 7111 { | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7112 if (VIM_ISWHITE(c)) |
| 375 | 7113 c = ' '; |
| 7114 else | |
| 7115 c = slang->sl_sal_first[c]; | |
| 7116 if (c != NUL && (ri == 0 || res[ri - 1] != c)) | |
| 7117 res[ri++] = c; | |
| 7118 } | |
| 7119 } | |
| 7120 | |
| 7121 res[ri] = NUL; | |
| 7122 } | |
| 7123 | |
| 7124 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7125 spell_soundfold_sal(slang_T *slang, char_u *inword, char_u *res) |
| 323 | 7126 { |
| 344 | 7127 salitem_T *smp; |
| 323 | 7128 char_u word[MAXWLEN]; |
| 375 | 7129 char_u *s = inword; |
| 323 | 7130 char_u *t; |
| 344 | 7131 char_u *pf; |
| 323 | 7132 int i, j, z; |
| 344 | 7133 int reslen; |
| 323 | 7134 int n, k = 0; |
| 7135 int z0; | |
| 7136 int k0; | |
| 7137 int n0; | |
| 7138 int c; | |
| 7139 int pri; | |
| 7140 int p0 = -333; | |
| 7141 int c0; | |
| 7142 | |
| 324 | 7143 /* Remove accents, if wanted. We actually remove all non-word characters. |
| 375 | 7144 * But keep white space. We need a copy, the word may be changed here. */ |
| 323 | 7145 if (slang->sl_rem_accents) |
| 7146 { | |
| 7147 t = word; | |
| 375 | 7148 while (*s != NUL) |
| 323 | 7149 { |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7150 if (VIM_ISWHITE(*s)) |
| 344 | 7151 { |
| 7152 *t++ = ' '; | |
| 7153 s = skipwhite(s); | |
| 7154 } | |
| 323 | 7155 else |
| 7156 { | |
| 5477 | 7157 if (spell_iswordp_nmw(s, curwin)) |
| 323 | 7158 *t++ = *s; |
| 7159 ++s; | |
| 7160 } | |
| 7161 } | |
| 7162 *t = NUL; | |
| 7163 } | |
| 7164 else | |
| 2768 | 7165 vim_strncpy(word, s, MAXWLEN - 1); |
| 323 | 7166 |
| 344 | 7167 smp = (salitem_T *)slang->sl_sal.ga_data; |
| 323 | 7168 |
| 7169 /* | |
| 7170 * This comes from Aspell phonet.cpp. Converted from C++ to C. | |
| 324 | 7171 * Changed to keep spaces. |
| 323 | 7172 */ |
| 344 | 7173 i = reslen = z = 0; |
| 323 | 7174 while ((c = word[i]) != NUL) |
| 7175 { | |
| 344 | 7176 /* Start with the first rule that has the character in the word. */ |
| 323 | 7177 n = slang->sl_sal_first[c]; |
| 7178 z0 = 0; | |
| 7179 | |
| 7180 if (n >= 0) | |
| 7181 { | |
| 7182 /* check all rules for the same letter */ | |
| 344 | 7183 for (; (s = smp[n].sm_lead)[0] == c; ++n) |
| 323 | 7184 { |
| 344 | 7185 /* Quickly skip entries that don't match the word. Most |
| 7186 * entries are less then three chars, optimize for that. */ | |
| 7187 k = smp[n].sm_leadlen; | |
| 7188 if (k > 1) | |
| 323 | 7189 { |
| 344 | 7190 if (word[i + 1] != s[1]) |
| 7191 continue; | |
| 7192 if (k > 2) | |
| 7193 { | |
| 7194 for (j = 2; j < k; ++j) | |
| 7195 if (word[i + j] != s[j]) | |
| 7196 break; | |
| 7197 if (j < k) | |
| 7198 continue; | |
| 7199 } | |
| 323 | 7200 } |
| 7201 | |
| 375 | 7202 if ((pf = smp[n].sm_oneof) != NULL) |
| 323 | 7203 { |
| 375 | 7204 /* Check for match with one of the chars in "sm_oneof". */ |
| 344 | 7205 while (*pf != NUL && *pf != word[i + k]) |
| 7206 ++pf; | |
| 7207 if (*pf == NUL) | |
| 7208 continue; | |
| 7209 ++k; | |
| 323 | 7210 } |
| 344 | 7211 s = smp[n].sm_rules; |
| 7212 pri = 5; /* default priority */ | |
| 323 | 7213 |
| 7214 p0 = *s; | |
| 7215 k0 = k; | |
| 7216 while (*s == '-' && k > 1) | |
| 7217 { | |
| 7218 k--; | |
| 7219 s++; | |
| 7220 } | |
| 7221 if (*s == '<') | |
| 7222 s++; | |
| 344 | 7223 if (VIM_ISDIGIT(*s)) |
| 323 | 7224 { |
| 7225 /* determine priority */ | |
| 7226 pri = *s - '0'; | |
| 7227 s++; | |
| 7228 } | |
| 7229 if (*s == '^' && *(s + 1) == '^') | |
| 7230 s++; | |
| 7231 | |
| 7232 if (*s == NUL | |
| 7233 || (*s == '^' | |
| 324 | 7234 && (i == 0 || !(word[i - 1] == ' ' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7235 || spell_iswordp(word + i - 1, curwin))) |
| 323 | 7236 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7237 || (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 7238 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7239 && spell_iswordp(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7240 && (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 7241 { |
| 7242 /* search for followup rules, if: */ | |
| 7243 /* followup and k > 1 and NO '-' in searchstring */ | |
| 7244 c0 = word[i + k - 1]; | |
| 7245 n0 = slang->sl_sal_first[c0]; | |
| 7246 | |
| 7247 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 344 | 7248 && p0 != '-' && word[i + k] != NUL) |
| 323 | 7249 { |
| 7250 /* test follow-up rule for "word[i + k]" */ | |
| 344 | 7251 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0) |
| 323 | 7252 { |
| 344 | 7253 /* Quickly skip entries that don't match the word. |
| 7254 * */ | |
| 7255 k0 = smp[n0].sm_leadlen; | |
| 7256 if (k0 > 1) | |
| 323 | 7257 { |
| 344 | 7258 if (word[i + k] != s[1]) |
| 7259 continue; | |
| 7260 if (k0 > 2) | |
| 7261 { | |
| 7262 pf = word + i + k + 1; | |
| 7263 for (j = 2; j < k0; ++j) | |
| 7264 if (*pf++ != s[j]) | |
| 7265 break; | |
| 7266 if (j < k0) | |
| 7267 continue; | |
| 7268 } | |
| 323 | 7269 } |
| 344 | 7270 k0 += k - 1; |
| 7271 | |
| 375 | 7272 if ((pf = smp[n0].sm_oneof) != NULL) |
| 323 | 7273 { |
| 344 | 7274 /* Check for match with one of the chars in |
| 375 | 7275 * "sm_oneof". */ |
| 344 | 7276 while (*pf != NUL && *pf != word[i + k0]) |
| 7277 ++pf; | |
| 7278 if (*pf == NUL) | |
| 7279 continue; | |
| 7280 ++k0; | |
| 323 | 7281 } |
| 344 | 7282 |
| 7283 p0 = 5; | |
| 7284 s = smp[n0].sm_rules; | |
| 323 | 7285 while (*s == '-') |
| 7286 { | |
| 344 | 7287 /* "k0" gets NOT reduced because |
| 7288 * "if (k0 == k)" */ | |
| 323 | 7289 s++; |
| 7290 } | |
| 7291 if (*s == '<') | |
| 7292 s++; | |
| 344 | 7293 if (VIM_ISDIGIT(*s)) |
| 323 | 7294 { |
| 7295 p0 = *s - '0'; | |
| 7296 s++; | |
| 7297 } | |
| 7298 | |
| 7299 if (*s == NUL | |
| 7300 /* *s == '^' cuts */ | |
| 7301 || (*s == '$' | |
| 376 | 7302 && !spell_iswordp(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7303 curwin))) |
| 323 | 7304 { |
| 7305 if (k0 == k) | |
| 7306 /* this is just a piece of the string */ | |
| 7307 continue; | |
| 7308 | |
| 7309 if (p0 < pri) | |
| 7310 /* priority too low */ | |
| 7311 continue; | |
| 7312 /* rule fits; stop search */ | |
| 7313 break; | |
| 7314 } | |
| 7315 } | |
| 7316 | |
| 344 | 7317 if (p0 >= pri && smp[n0].sm_lead[0] == c0) |
| 323 | 7318 continue; |
| 7319 } | |
| 7320 | |
| 7321 /* replace string */ | |
| 344 | 7322 s = smp[n].sm_to; |
| 389 | 7323 if (s == NULL) |
| 7324 s = (char_u *)""; | |
| 344 | 7325 pf = smp[n].sm_rules; |
| 7326 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0; | |
| 323 | 7327 if (p0 == 1 && z == 0) |
| 7328 { | |
| 7329 /* rule with '<' is used */ | |
| 344 | 7330 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c |
| 7331 || res[reslen - 1] == *s)) | |
| 7332 reslen--; | |
| 323 | 7333 z0 = 1; |
| 7334 z = 1; | |
| 7335 k0 = 0; | |
| 372 | 7336 while (*s != NUL && word[i + k0] != NUL) |
| 323 | 7337 { |
| 7338 word[i + k0] = *s; | |
| 7339 k0++; | |
| 7340 s++; | |
| 7341 } | |
| 7342 if (k > k0) | |
| 1619 | 7343 STRMOVE(word + i + k0, word + i + k); |
| 323 | 7344 |
| 7345 /* new "actual letter" */ | |
| 7346 c = word[i]; | |
| 7347 } | |
| 7348 else | |
| 7349 { | |
| 7350 /* no '<' rule used */ | |
| 7351 i += k - 1; | |
| 7352 z = 0; | |
| 344 | 7353 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN) |
| 323 | 7354 { |
| 344 | 7355 if (reslen == 0 || res[reslen - 1] != *s) |
| 372 | 7356 res[reslen++] = *s; |
| 323 | 7357 s++; |
| 7358 } | |
| 7359 /* new "actual letter" */ | |
| 7360 c = *s; | |
| 344 | 7361 if (strstr((char *)pf, "^^") != NULL) |
| 323 | 7362 { |
| 7363 if (c != NUL) | |
| 372 | 7364 res[reslen++] = c; |
| 1619 | 7365 STRMOVE(word, word + i + 1); |
| 323 | 7366 i = 0; |
| 7367 z0 = 1; | |
| 7368 } | |
| 7369 } | |
| 7370 break; | |
| 7371 } | |
| 7372 } | |
| 7373 } | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7374 else if (VIM_ISWHITE(c)) |
| 324 | 7375 { |
| 7376 c = ' '; | |
| 7377 k = 1; | |
| 7378 } | |
| 323 | 7379 |
| 7380 if (z0 == 0) | |
| 7381 { | |
| 344 | 7382 if (k && !p0 && reslen < MAXWLEN && c != NUL |
| 7383 && (!slang->sl_collapse || reslen == 0 | |
| 7384 || res[reslen - 1] != c)) | |
| 323 | 7385 /* condense only double letters */ |
| 372 | 7386 res[reslen++] = c; |
| 323 | 7387 |
| 7388 i++; | |
| 7389 z = 0; | |
| 7390 k = 0; | |
| 7391 } | |
| 7392 } | |
| 7393 | |
| 344 | 7394 res[reslen] = NUL; |
| 323 | 7395 } |
| 7396 | |
| 372 | 7397 #ifdef FEAT_MBYTE |
| 7398 /* | |
| 7399 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 7400 * Multi-byte version of spell_soundfold(). | |
| 7401 */ | |
| 7402 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7403 spell_soundfold_wsal(slang_T *slang, char_u *inword, char_u *res) |
| 372 | 7404 { |
| 375 | 7405 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data; |
| 372 | 7406 int word[MAXWLEN]; |
| 7407 int wres[MAXWLEN]; | |
| 7408 int l; | |
| 7409 char_u *s; | |
| 7410 int *ws; | |
| 7411 char_u *t; | |
| 7412 int *pf; | |
| 7413 int i, j, z; | |
| 7414 int reslen; | |
| 7415 int n, k = 0; | |
| 7416 int z0; | |
| 7417 int k0; | |
| 7418 int n0; | |
| 7419 int c; | |
| 7420 int pri; | |
| 7421 int p0 = -333; | |
| 7422 int c0; | |
| 7423 int did_white = FALSE; | |
| 3520 | 7424 int wordlen; |
| 7425 | |
| 372 | 7426 |
| 7427 /* | |
| 7428 * Convert the multi-byte string to a wide-character string. | |
| 7429 * Remove accents, if wanted. We actually remove all non-word characters. | |
| 7430 * But keep white space. | |
| 7431 */ | |
| 3520 | 7432 wordlen = 0; |
| 372 | 7433 for (s = inword; *s != NUL; ) |
| 7434 { | |
| 7435 t = s; | |
| 474 | 7436 c = mb_cptr2char_adv(&s); |
| 372 | 7437 if (slang->sl_rem_accents) |
| 7438 { | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7439 if (enc_utf8 ? utf_class(c) == 0 : VIM_ISWHITE(c)) |
| 372 | 7440 { |
| 7441 if (did_white) | |
| 7442 continue; | |
| 7443 c = ' '; | |
| 7444 did_white = TRUE; | |
| 7445 } | |
| 7446 else | |
| 7447 { | |
| 7448 did_white = FALSE; | |
| 5477 | 7449 if (!spell_iswordp_nmw(t, curwin)) |
| 372 | 7450 continue; |
| 7451 } | |
| 7452 } | |
| 3520 | 7453 word[wordlen++] = c; |
| 7454 } | |
| 7455 word[wordlen] = NUL; | |
| 372 | 7456 |
| 7457 /* | |
| 3520 | 7458 * This algorithm comes from Aspell phonet.cpp. |
| 372 | 7459 * Converted from C++ to C. Added support for multi-byte chars. |
| 7460 * Changed to keep spaces. | |
| 7461 */ | |
| 7462 i = reslen = z = 0; | |
| 7463 while ((c = word[i]) != NUL) | |
| 7464 { | |
| 7465 /* Start with the first rule that has the character in the word. */ | |
| 7466 n = slang->sl_sal_first[c & 0xff]; | |
| 7467 z0 = 0; | |
| 7468 | |
| 7469 if (n >= 0) | |
| 7470 { | |
|
2455
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
7471 /* 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
|
7472 * 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
|
7473 * (c & 0xff) is NUL. */ |
|
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
7474 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
|
7475 && ws[0] != NUL; ++n) |
| 372 | 7476 { |
| 7477 /* Quickly skip entries that don't match the word. Most | |
| 7478 * entries are less then three chars, optimize for that. */ | |
| 375 | 7479 if (c != ws[0]) |
| 7480 continue; | |
| 372 | 7481 k = smp[n].sm_leadlen; |
| 7482 if (k > 1) | |
| 7483 { | |
| 7484 if (word[i + 1] != ws[1]) | |
| 7485 continue; | |
| 7486 if (k > 2) | |
| 7487 { | |
| 7488 for (j = 2; j < k; ++j) | |
| 7489 if (word[i + j] != ws[j]) | |
| 7490 break; | |
| 7491 if (j < k) | |
| 7492 continue; | |
| 7493 } | |
| 7494 } | |
| 7495 | |
| 375 | 7496 if ((pf = smp[n].sm_oneof_w) != NULL) |
| 372 | 7497 { |
| 375 | 7498 /* Check for match with one of the chars in "sm_oneof". */ |
| 372 | 7499 while (*pf != NUL && *pf != word[i + k]) |
| 7500 ++pf; | |
| 7501 if (*pf == NUL) | |
| 7502 continue; | |
| 7503 ++k; | |
| 7504 } | |
| 7505 s = smp[n].sm_rules; | |
| 7506 pri = 5; /* default priority */ | |
| 7507 | |
| 7508 p0 = *s; | |
| 7509 k0 = k; | |
| 7510 while (*s == '-' && k > 1) | |
| 7511 { | |
| 7512 k--; | |
| 7513 s++; | |
| 7514 } | |
| 7515 if (*s == '<') | |
| 7516 s++; | |
| 7517 if (VIM_ISDIGIT(*s)) | |
| 7518 { | |
| 7519 /* determine priority */ | |
| 7520 pri = *s - '0'; | |
| 7521 s++; | |
| 7522 } | |
| 7523 if (*s == '^' && *(s + 1) == '^') | |
| 7524 s++; | |
| 7525 | |
| 7526 if (*s == NUL | |
| 7527 || (*s == '^' | |
| 7528 && (i == 0 || !(word[i - 1] == ' ' | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7529 || spell_iswordp_w(word + i - 1, curwin))) |
| 372 | 7530 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7531 || (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 7532 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7533 && spell_iswordp_w(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7534 && (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 7535 { |
| 7536 /* search for followup rules, if: */ | |
| 7537 /* followup and k > 1 and NO '-' in searchstring */ | |
| 7538 c0 = word[i + k - 1]; | |
| 7539 n0 = slang->sl_sal_first[c0 & 0xff]; | |
| 7540 | |
| 7541 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 7542 && p0 != '-' && word[i + k] != NUL) | |
| 7543 { | |
| 375 | 7544 /* Test follow-up rule for "word[i + k]"; loop over |
| 7545 * all entries with the same index byte. */ | |
| 372 | 7546 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff) |
| 7547 == (c0 & 0xff); ++n0) | |
| 7548 { | |
| 7549 /* Quickly skip entries that don't match the word. | |
| 375 | 7550 */ |
| 7551 if (c0 != ws[0]) | |
| 7552 continue; | |
| 372 | 7553 k0 = smp[n0].sm_leadlen; |
| 7554 if (k0 > 1) | |
| 7555 { | |
| 7556 if (word[i + k] != ws[1]) | |
| 7557 continue; | |
| 7558 if (k0 > 2) | |
| 7559 { | |
| 7560 pf = word + i + k + 1; | |
| 7561 for (j = 2; j < k0; ++j) | |
| 7562 if (*pf++ != ws[j]) | |
| 7563 break; | |
| 7564 if (j < k0) | |
| 7565 continue; | |
| 7566 } | |
| 7567 } | |
| 7568 k0 += k - 1; | |
| 7569 | |
| 375 | 7570 if ((pf = smp[n0].sm_oneof_w) != NULL) |
| 372 | 7571 { |
| 7572 /* Check for match with one of the chars in | |
| 375 | 7573 * "sm_oneof". */ |
| 372 | 7574 while (*pf != NUL && *pf != word[i + k0]) |
| 7575 ++pf; | |
| 7576 if (*pf == NUL) | |
| 7577 continue; | |
| 7578 ++k0; | |
| 7579 } | |
| 7580 | |
| 7581 p0 = 5; | |
| 7582 s = smp[n0].sm_rules; | |
| 7583 while (*s == '-') | |
| 7584 { | |
| 7585 /* "k0" gets NOT reduced because | |
| 7586 * "if (k0 == k)" */ | |
| 7587 s++; | |
| 7588 } | |
| 7589 if (*s == '<') | |
| 7590 s++; | |
| 7591 if (VIM_ISDIGIT(*s)) | |
| 7592 { | |
| 7593 p0 = *s - '0'; | |
| 7594 s++; | |
| 7595 } | |
| 7596 | |
| 7597 if (*s == NUL | |
| 7598 /* *s == '^' cuts */ | |
| 7599 || (*s == '$' | |
| 376 | 7600 && !spell_iswordp_w(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7601 curwin))) |
| 372 | 7602 { |
| 7603 if (k0 == k) | |
| 7604 /* this is just a piece of the string */ | |
| 7605 continue; | |
| 7606 | |
| 7607 if (p0 < pri) | |
| 7608 /* priority too low */ | |
| 7609 continue; | |
| 7610 /* rule fits; stop search */ | |
| 7611 break; | |
| 7612 } | |
| 7613 } | |
| 7614 | |
| 7615 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff) | |
| 7616 == (c0 & 0xff)) | |
| 7617 continue; | |
| 7618 } | |
| 7619 | |
| 7620 /* replace string */ | |
| 7621 ws = smp[n].sm_to_w; | |
| 7622 s = smp[n].sm_rules; | |
| 7623 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0; | |
| 7624 if (p0 == 1 && z == 0) | |
| 7625 { | |
| 7626 /* rule with '<' is used */ | |
| 389 | 7627 if (reslen > 0 && ws != NULL && *ws != NUL |
| 7628 && (wres[reslen - 1] == c | |
| 372 | 7629 || wres[reslen - 1] == *ws)) |
| 7630 reslen--; | |
| 7631 z0 = 1; | |
| 7632 z = 1; | |
| 7633 k0 = 0; | |
| 389 | 7634 if (ws != NULL) |
| 7635 while (*ws != NUL && word[i + k0] != NUL) | |
| 7636 { | |
| 7637 word[i + k0] = *ws; | |
| 7638 k0++; | |
| 7639 ws++; | |
| 7640 } | |
| 372 | 7641 if (k > k0) |
| 7642 mch_memmove(word + i + k0, word + i + k, | |
| 3520 | 7643 sizeof(int) * (wordlen - (i + k) + 1)); |
| 372 | 7644 |
| 7645 /* new "actual letter" */ | |
| 7646 c = word[i]; | |
| 7647 } | |
| 7648 else | |
| 7649 { | |
| 7650 /* no '<' rule used */ | |
| 7651 i += k - 1; | |
| 7652 z = 0; | |
| 389 | 7653 if (ws != NULL) |
| 7654 while (*ws != NUL && ws[1] != NUL | |
| 7655 && reslen < MAXWLEN) | |
| 7656 { | |
| 7657 if (reslen == 0 || wres[reslen - 1] != *ws) | |
| 7658 wres[reslen++] = *ws; | |
| 7659 ws++; | |
| 7660 } | |
| 372 | 7661 /* new "actual letter" */ |
| 389 | 7662 if (ws == NULL) |
| 7663 c = NUL; | |
| 7664 else | |
| 7665 c = *ws; | |
| 372 | 7666 if (strstr((char *)s, "^^") != NULL) |
| 7667 { | |
| 7668 if (c != NUL) | |
| 7669 wres[reslen++] = c; | |
| 7670 mch_memmove(word, word + i + 1, | |
| 3520 | 7671 sizeof(int) * (wordlen - (i + 1) + 1)); |
| 372 | 7672 i = 0; |
| 7673 z0 = 1; | |
| 7674 } | |
| 7675 } | |
| 7676 break; | |
| 7677 } | |
| 7678 } | |
| 7679 } | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7680 else if (VIM_ISWHITE(c)) |
| 372 | 7681 { |
| 7682 c = ' '; | |
| 7683 k = 1; | |
| 7684 } | |
| 7685 | |
| 7686 if (z0 == 0) | |
| 7687 { | |
| 7688 if (k && !p0 && reslen < MAXWLEN && c != NUL | |
| 7689 && (!slang->sl_collapse || reslen == 0 | |
| 7690 || wres[reslen - 1] != c)) | |
| 7691 /* condense only double letters */ | |
| 7692 wres[reslen++] = c; | |
| 7693 | |
| 7694 i++; | |
| 7695 z = 0; | |
| 7696 k = 0; | |
| 7697 } | |
| 7698 } | |
| 7699 | |
| 7700 /* Convert wide characters in "wres" to a multi-byte string in "res". */ | |
| 7701 l = 0; | |
| 7702 for (n = 0; n < reslen; ++n) | |
| 7703 { | |
| 7704 l += mb_char2bytes(wres[n], res + l); | |
| 7705 if (l + MB_MAXBYTES > MAXWLEN) | |
| 7706 break; | |
| 7707 } | |
| 7708 res[l] = NUL; | |
| 7709 } | |
| 7710 #endif | |
| 7711 | |
| 324 | 7712 /* |
| 344 | 7713 * Compute a score for two sound-a-like words. |
| 7714 * This permits up to two inserts/deletes/swaps/etc. to keep things fast. | |
| 7715 * Instead of a generic loop we write out the code. That keeps it fast by | |
| 7716 * avoiding checks that will not be possible. | |
| 7717 */ | |
| 7718 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7719 soundalike_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7720 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
|
7721 char_u *badstart) /* sound-folded bad word */ |
| 344 | 7722 { |
| 351 | 7723 char_u *goodsound = goodstart; |
| 7724 char_u *badsound = badstart; | |
| 7725 int goodlen; | |
| 7726 int badlen; | |
| 344 | 7727 int n; |
| 7728 char_u *pl, *ps; | |
| 7729 char_u *pl2, *ps2; | |
| 351 | 7730 int score = 0; |
| 7731 | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7732 /* Adding/inserting "*" at the start (word starts with vowel) shouldn't be |
| 351 | 7733 * counted so much, vowels halfway the word aren't counted at all. */ |
| 7734 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound) | |
| 7735 { | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7736 if ((badsound[0] == NUL && goodsound[1] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7737 || (goodsound[0] == NUL && badsound[1] == NUL)) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7738 /* 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
|
7739 return SCORE_DEL; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7740 if (badsound[0] == NUL || goodsound[0] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7741 /* more than two changes */ |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7742 return SCORE_MAXMAX; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7743 |
| 625 | 7744 if (badsound[1] == goodsound[1] |
| 7745 || (badsound[1] != NUL | |
| 7746 && goodsound[1] != NUL | |
| 7747 && badsound[2] == goodsound[2])) | |
| 7748 { | |
| 7749 /* handle like a substitute */ | |
| 7750 } | |
| 351 | 7751 else |
| 625 | 7752 { |
| 7753 score = 2 * SCORE_DEL / 3; | |
| 7754 if (*badsound == '*') | |
| 7755 ++badsound; | |
| 7756 else | |
| 7757 ++goodsound; | |
| 7758 } | |
| 351 | 7759 } |
| 7760 | |
| 835 | 7761 goodlen = (int)STRLEN(goodsound); |
| 7762 badlen = (int)STRLEN(badsound); | |
| 344 | 7763 |
| 1188 | 7764 /* Return quickly if the lengths are too different to be fixed by two |
| 344 | 7765 * changes. */ |
| 7766 n = goodlen - badlen; | |
| 7767 if (n < -2 || n > 2) | |
| 7768 return SCORE_MAXMAX; | |
| 7769 | |
| 7770 if (n > 0) | |
| 7771 { | |
| 351 | 7772 pl = goodsound; /* goodsound is longest */ |
| 344 | 7773 ps = badsound; |
| 7774 } | |
| 7775 else | |
| 7776 { | |
| 351 | 7777 pl = badsound; /* badsound is longest */ |
| 344 | 7778 ps = goodsound; |
| 7779 } | |
| 7780 | |
| 7781 /* Skip over the identical part. */ | |
| 7782 while (*pl == *ps && *pl != NUL) | |
| 7783 { | |
| 7784 ++pl; | |
| 7785 ++ps; | |
| 7786 } | |
| 7787 | |
| 7788 switch (n) | |
| 7789 { | |
| 7790 case -2: | |
| 7791 case 2: | |
| 7792 /* | |
| 7793 * Must delete two characters from "pl". | |
| 7794 */ | |
| 7795 ++pl; /* first delete */ | |
| 7796 while (*pl == *ps) | |
| 7797 { | |
| 7798 ++pl; | |
| 7799 ++ps; | |
| 7800 } | |
| 7801 /* strings must be equal after second delete */ | |
| 7802 if (STRCMP(pl + 1, ps) == 0) | |
| 351 | 7803 return score + SCORE_DEL * 2; |
| 344 | 7804 |
| 7805 /* Failed to compare. */ | |
| 7806 break; | |
| 7807 | |
| 7808 case -1: | |
| 7809 case 1: | |
| 7810 /* | |
| 7811 * Minimal one delete from "pl" required. | |
| 7812 */ | |
| 7813 | |
| 7814 /* 1: delete */ | |
| 7815 pl2 = pl + 1; | |
| 7816 ps2 = ps; | |
| 7817 while (*pl2 == *ps2) | |
| 7818 { | |
| 7819 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7820 return score + SCORE_DEL; |
| 344 | 7821 ++pl2; |
| 7822 ++ps2; | |
| 7823 } | |
| 7824 | |
| 7825 /* 2: delete then swap, then rest must be equal */ | |
| 7826 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7827 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7828 return score + SCORE_DEL + SCORE_SWAP; |
| 344 | 7829 |
| 7830 /* 3: delete then substitute, then the rest must be equal */ | |
| 7831 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7832 return score + SCORE_DEL + SCORE_SUBST; |
| 344 | 7833 |
| 7834 /* 4: first swap then delete */ | |
| 7835 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 7836 { | |
| 7837 pl2 = pl + 2; /* swap, skip two chars */ | |
| 7838 ps2 = ps + 2; | |
| 7839 while (*pl2 == *ps2) | |
| 7840 { | |
| 7841 ++pl2; | |
| 7842 ++ps2; | |
| 7843 } | |
| 7844 /* delete a char and then strings must be equal */ | |
| 7845 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7846 return score + SCORE_SWAP + SCORE_DEL; |
| 344 | 7847 } |
| 7848 | |
| 7849 /* 5: first substitute then delete */ | |
| 7850 pl2 = pl + 1; /* substitute, skip one char */ | |
| 7851 ps2 = ps + 1; | |
| 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_SUBST + SCORE_DEL; |
| 344 | 7860 |
| 7861 /* Failed to compare. */ | |
| 7862 break; | |
| 7863 | |
| 7864 case 0: | |
| 7865 /* | |
| 1808 | 7866 * Lengths are equal, thus changes must result in same length: An |
| 344 | 7867 * insert is only possible in combination with a delete. |
| 7868 * 1: check if for identical strings | |
| 7869 */ | |
| 7870 if (*pl == NUL) | |
| 351 | 7871 return score; |
| 344 | 7872 |
| 7873 /* 2: swap */ | |
| 7874 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 7875 { | |
| 7876 pl2 = pl + 2; /* swap, skip two chars */ | |
| 7877 ps2 = ps + 2; | |
| 7878 while (*pl2 == *ps2) | |
| 7879 { | |
| 7880 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7881 return score + SCORE_SWAP; |
| 344 | 7882 ++pl2; |
| 7883 ++ps2; | |
| 7884 } | |
| 7885 /* 3: swap and swap again */ | |
| 7886 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7887 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7888 return score + SCORE_SWAP + SCORE_SWAP; |
| 344 | 7889 |
| 7890 /* 4: swap and substitute */ | |
| 7891 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7892 return score + SCORE_SWAP + SCORE_SUBST; |
| 344 | 7893 } |
| 7894 | |
| 7895 /* 5: substitute */ | |
| 7896 pl2 = pl + 1; | |
| 7897 ps2 = ps + 1; | |
| 7898 while (*pl2 == *ps2) | |
| 7899 { | |
| 7900 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7901 return score + SCORE_SUBST; |
| 344 | 7902 ++pl2; |
| 7903 ++ps2; | |
| 7904 } | |
| 7905 | |
| 7906 /* 6: substitute and swap */ | |
| 7907 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7908 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7909 return score + SCORE_SUBST + SCORE_SWAP; |
| 344 | 7910 |
| 7911 /* 7: substitute and substitute */ | |
| 7912 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7913 return score + SCORE_SUBST + SCORE_SUBST; |
| 344 | 7914 |
| 7915 /* 8: insert then delete */ | |
| 7916 pl2 = pl; | |
| 7917 ps2 = ps + 1; | |
| 7918 while (*pl2 == *ps2) | |
| 7919 { | |
| 7920 ++pl2; | |
| 7921 ++ps2; | |
| 7922 } | |
| 7923 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7924 return score + SCORE_INS + SCORE_DEL; |
| 344 | 7925 |
| 7926 /* 9: delete then insert */ | |
| 7927 pl2 = pl + 1; | |
| 7928 ps2 = ps; | |
| 7929 while (*pl2 == *ps2) | |
| 7930 { | |
| 7931 ++pl2; | |
| 7932 ++ps2; | |
| 7933 } | |
| 7934 if (STRCMP(pl2, ps2 + 1) == 0) | |
| 351 | 7935 return score + SCORE_INS + SCORE_DEL; |
| 344 | 7936 |
| 7937 /* Failed to compare. */ | |
| 7938 break; | |
| 7939 } | |
| 7940 | |
| 7941 return SCORE_MAXMAX; | |
| 7942 } | |
| 324 | 7943 |
| 323 | 7944 /* |
| 7945 * Compute the "edit distance" to turn "badword" into "goodword". The less | |
| 344 | 7946 * deletes/inserts/substitutes/swaps are required the lower the score. |
| 324 | 7947 * |
| 485 | 7948 * The algorithm is described by Du and Chang, 1992. |
| 7949 * The implementation of the algorithm comes from Aspell editdist.cpp, | |
| 7950 * edit_distance(). It has been converted from C++ to C and modified to | |
| 7951 * support multi-byte characters. | |
| 323 | 7952 */ |
| 7953 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7954 spell_edit_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7955 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7956 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7957 char_u *goodword) |
| 323 | 7958 { |
| 7959 int *cnt; | |
| 1188 | 7960 int badlen, goodlen; /* lengths including NUL */ |
| 323 | 7961 int j, i; |
| 7962 int t; | |
| 7963 int bc, gc; | |
| 324 | 7964 int pbc, pgc; |
| 7965 #ifdef FEAT_MBYTE | |
| 7966 char_u *p; | |
| 7967 int wbadword[MAXWLEN]; | |
| 7968 int wgoodword[MAXWLEN]; | |
| 7969 | |
| 7970 if (has_mbyte) | |
| 7971 { | |
| 7972 /* Get the characters from the multi-byte strings and put them in an | |
| 7973 * int array for easy access. */ | |
| 7974 for (p = badword, badlen = 0; *p != NUL; ) | |
| 474 | 7975 wbadword[badlen++] = mb_cptr2char_adv(&p); |
| 398 | 7976 wbadword[badlen++] = 0; |
| 324 | 7977 for (p = goodword, goodlen = 0; *p != NUL; ) |
| 474 | 7978 wgoodword[goodlen++] = mb_cptr2char_adv(&p); |
| 398 | 7979 wgoodword[goodlen++] = 0; |
| 324 | 7980 } |
| 7981 else | |
| 7982 #endif | |
| 7983 { | |
| 835 | 7984 badlen = (int)STRLEN(badword) + 1; |
| 7985 goodlen = (int)STRLEN(goodword) + 1; | |
| 324 | 7986 } |
| 323 | 7987 |
| 7988 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */ | |
| 7989 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)] | |
| 7990 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)), | |
| 7991 TRUE); | |
| 324 | 7992 if (cnt == NULL) |
| 7993 return 0; /* out of memory */ | |
| 323 | 7994 |
| 7995 CNT(0, 0) = 0; | |
| 7996 for (j = 1; j <= goodlen; ++j) | |
| 625 | 7997 CNT(0, j) = CNT(0, j - 1) + SCORE_INS; |
| 323 | 7998 |
| 7999 for (i = 1; i <= badlen; ++i) | |
| 8000 { | |
| 625 | 8001 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL; |
| 323 | 8002 for (j = 1; j <= goodlen; ++j) |
| 8003 { | |
| 324 | 8004 #ifdef FEAT_MBYTE |
| 8005 if (has_mbyte) | |
| 8006 { | |
| 8007 bc = wbadword[i - 1]; | |
| 8008 gc = wgoodword[j - 1]; | |
| 8009 } | |
| 8010 else | |
| 8011 #endif | |
| 8012 { | |
| 8013 bc = badword[i - 1]; | |
| 8014 gc = goodword[j - 1]; | |
| 8015 } | |
| 323 | 8016 if (bc == gc) |
| 8017 CNT(i, j) = CNT(i - 1, j - 1); | |
| 8018 else | |
| 8019 { | |
| 8020 /* Use a better score when there is only a case difference. */ | |
| 324 | 8021 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) |
| 323 | 8022 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1); |
| 8023 else | |
| 625 | 8024 { |
| 8025 /* For a similar character use SCORE_SIMILAR. */ | |
| 8026 if (slang != NULL | |
| 8027 && slang->sl_has_map | |
| 8028 && similar_chars(slang, gc, bc)) | |
| 8029 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1); | |
| 8030 else | |
| 8031 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1); | |
| 8032 } | |
| 323 | 8033 |
| 324 | 8034 if (i > 1 && j > 1) |
| 323 | 8035 { |
| 324 | 8036 #ifdef FEAT_MBYTE |
| 8037 if (has_mbyte) | |
| 8038 { | |
| 8039 pbc = wbadword[i - 2]; | |
| 8040 pgc = wgoodword[j - 2]; | |
| 8041 } | |
| 8042 else | |
| 8043 #endif | |
| 8044 { | |
| 8045 pbc = badword[i - 2]; | |
| 8046 pgc = goodword[j - 2]; | |
| 8047 } | |
| 8048 if (bc == pgc && pbc == gc) | |
| 8049 { | |
| 8050 t = SCORE_SWAP + CNT(i - 2, j - 2); | |
| 8051 if (t < CNT(i, j)) | |
| 8052 CNT(i, j) = t; | |
| 8053 } | |
| 323 | 8054 } |
| 8055 t = SCORE_DEL + CNT(i - 1, j); | |
| 8056 if (t < CNT(i, j)) | |
| 8057 CNT(i, j) = t; | |
| 8058 t = SCORE_INS + CNT(i, j - 1); | |
| 8059 if (t < CNT(i, j)) | |
| 8060 CNT(i, j) = t; | |
| 8061 } | |
| 8062 } | |
| 8063 } | |
| 344 | 8064 |
| 8065 i = CNT(badlen - 1, goodlen - 1); | |
| 8066 vim_free(cnt); | |
| 8067 return i; | |
| 323 | 8068 } |
| 307 | 8069 |
| 625 | 8070 typedef struct |
| 8071 { | |
| 8072 int badi; | |
| 8073 int goodi; | |
| 8074 int score; | |
| 8075 } limitscore_T; | |
| 8076 | |
| 8077 /* | |
| 8078 * Like spell_edit_score(), but with a limit on the score to make it faster. | |
| 8079 * May return SCORE_MAXMAX when the score is higher than "limit". | |
| 8080 * | |
| 8081 * This uses a stack for the edits still to be tried. | |
| 8082 * The idea comes from Aspell leditdist.cpp. Rewritten in C and added support | |
| 8083 * for multi-byte characters. | |
| 8084 */ | |
| 8085 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8086 spell_edit_score_limit( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8087 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8088 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8089 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8090 int limit) |
| 625 | 8091 { |
| 8092 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 8093 int stackidx; | |
| 8094 int bi, gi; | |
| 8095 int bi2, gi2; | |
| 8096 int bc, gc; | |
| 8097 int score; | |
| 8098 int score_off; | |
| 8099 int minscore; | |
| 8100 int round; | |
| 8101 | |
| 8102 #ifdef FEAT_MBYTE | |
| 8103 /* Multi-byte characters require a bit more work, use a different function | |
| 8104 * to avoid testing "has_mbyte" quite often. */ | |
| 8105 if (has_mbyte) | |
| 8106 return spell_edit_score_limit_w(slang, badword, goodword, limit); | |
| 8107 #endif | |
| 8108 | |
| 8109 /* | |
| 8110 * The idea is to go from start to end over the words. So long as | |
| 8111 * characters are equal just continue, this always gives the lowest score. | |
| 8112 * When there is a difference try several alternatives. Each alternative | |
| 8113 * increases "score" for the edit distance. Some of the alternatives are | |
| 8114 * pushed unto a stack and tried later, some are tried right away. At the | |
| 8115 * end of the word the score for one alternative is known. The lowest | |
| 8116 * possible score is stored in "minscore". | |
| 8117 */ | |
| 8118 stackidx = 0; | |
| 8119 bi = 0; | |
| 8120 gi = 0; | |
| 8121 score = 0; | |
| 8122 minscore = limit + 1; | |
| 8123 | |
| 8124 for (;;) | |
| 8125 { | |
| 8126 /* Skip over an equal part, score remains the same. */ | |
| 8127 for (;;) | |
| 8128 { | |
| 8129 bc = badword[bi]; | |
| 8130 gc = goodword[gi]; | |
| 8131 if (bc != gc) /* stop at a char that's different */ | |
| 8132 break; | |
| 8133 if (bc == NUL) /* both words end */ | |
| 8134 { | |
| 8135 if (score < minscore) | |
| 8136 minscore = score; | |
| 8137 goto pop; /* do next alternative */ | |
| 8138 } | |
| 8139 ++bi; | |
| 8140 ++gi; | |
| 8141 } | |
| 8142 | |
| 8143 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 8144 { | |
| 8145 do | |
| 8146 { | |
| 8147 if ((score += SCORE_DEL) >= minscore) | |
| 8148 goto pop; /* do next alternative */ | |
| 8149 } while (badword[++bi] != NUL); | |
| 8150 minscore = score; | |
| 8151 } | |
| 8152 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 8153 { | |
| 8154 do | |
| 8155 { | |
| 8156 if ((score += SCORE_INS) >= minscore) | |
| 8157 goto pop; /* do next alternative */ | |
| 8158 } while (goodword[++gi] != NUL); | |
| 8159 minscore = score; | |
| 8160 } | |
| 8161 else /* both words continue */ | |
| 8162 { | |
| 8163 /* If not close to the limit, perform a change. Only try changes | |
| 8164 * that may lead to a lower score than "minscore". | |
| 8165 * round 0: try deleting a char from badword | |
| 8166 * round 1: try inserting a char in badword */ | |
| 8167 for (round = 0; round <= 1; ++round) | |
| 8168 { | |
| 8169 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 8170 if (score_off < minscore) | |
| 8171 { | |
| 8172 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 8173 { | |
| 8174 /* Near the limit, rest of the words must match. We | |
| 8175 * can check that right now, no need to push an item | |
| 8176 * onto the stack. */ | |
| 8177 bi2 = bi + 1 - round; | |
| 8178 gi2 = gi + round; | |
| 8179 while (goodword[gi2] == badword[bi2]) | |
| 8180 { | |
| 8181 if (goodword[gi2] == NUL) | |
| 8182 { | |
| 8183 minscore = score_off; | |
| 8184 break; | |
| 8185 } | |
| 8186 ++bi2; | |
| 8187 ++gi2; | |
| 8188 } | |
| 8189 } | |
| 8190 else | |
| 8191 { | |
| 8192 /* try deleting/inserting a character later */ | |
| 8193 stack[stackidx].badi = bi + 1 - round; | |
| 8194 stack[stackidx].goodi = gi + round; | |
| 8195 stack[stackidx].score = score_off; | |
| 8196 ++stackidx; | |
| 8197 } | |
| 8198 } | |
| 8199 } | |
| 8200 | |
| 8201 if (score + SCORE_SWAP < minscore) | |
| 8202 { | |
| 8203 /* If swapping two characters makes a match then the | |
| 8204 * substitution is more expensive, thus there is no need to | |
| 8205 * try both. */ | |
| 8206 if (gc == badword[bi + 1] && bc == goodword[gi + 1]) | |
| 8207 { | |
| 8208 /* Swap two characters, that is: skip them. */ | |
| 8209 gi += 2; | |
| 8210 bi += 2; | |
| 8211 score += SCORE_SWAP; | |
| 8212 continue; | |
| 8213 } | |
| 8214 } | |
| 8215 | |
| 8216 /* Substitute one character for another which is the same | |
| 8217 * thing as deleting a character from both goodword and badword. | |
| 8218 * Use a better score when there is only a case difference. */ | |
| 8219 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 8220 score += SCORE_ICASE; | |
| 8221 else | |
| 8222 { | |
| 8223 /* For a similar character use SCORE_SIMILAR. */ | |
| 8224 if (slang != NULL | |
| 8225 && slang->sl_has_map | |
| 8226 && similar_chars(slang, gc, bc)) | |
| 8227 score += SCORE_SIMILAR; | |
| 8228 else | |
| 8229 score += SCORE_SUBST; | |
| 8230 } | |
| 8231 | |
| 8232 if (score < minscore) | |
| 8233 { | |
| 8234 /* Do the substitution. */ | |
| 8235 ++gi; | |
| 8236 ++bi; | |
| 8237 continue; | |
| 8238 } | |
| 8239 } | |
| 8240 pop: | |
| 8241 /* | |
| 8242 * Get here to try the next alternative, pop it from the stack. | |
| 8243 */ | |
| 8244 if (stackidx == 0) /* stack is empty, finished */ | |
| 8245 break; | |
| 8246 | |
| 8247 /* pop an item from the stack */ | |
| 8248 --stackidx; | |
| 8249 gi = stack[stackidx].goodi; | |
| 8250 bi = stack[stackidx].badi; | |
| 8251 score = stack[stackidx].score; | |
| 8252 } | |
| 8253 | |
| 8254 /* When the score goes over "limit" it may actually be much higher. | |
| 8255 * Return a very large number to avoid going below the limit when giving a | |
| 8256 * bonus. */ | |
| 8257 if (minscore > limit) | |
| 8258 return SCORE_MAXMAX; | |
| 8259 return minscore; | |
| 8260 } | |
| 8261 | |
| 8262 #ifdef FEAT_MBYTE | |
| 8263 /* | |
| 8264 * Multi-byte version of spell_edit_score_limit(). | |
| 8265 * Keep it in sync with the above! | |
| 8266 */ | |
| 8267 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8268 spell_edit_score_limit_w( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8269 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8270 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8271 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8272 int limit) |
| 625 | 8273 { |
| 8274 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 8275 int stackidx; | |
| 8276 int bi, gi; | |
| 8277 int bi2, gi2; | |
| 8278 int bc, gc; | |
| 8279 int score; | |
| 8280 int score_off; | |
| 8281 int minscore; | |
| 8282 int round; | |
| 8283 char_u *p; | |
| 8284 int wbadword[MAXWLEN]; | |
| 8285 int wgoodword[MAXWLEN]; | |
| 8286 | |
| 8287 /* Get the characters from the multi-byte strings and put them in an | |
| 8288 * int array for easy access. */ | |
| 8289 bi = 0; | |
| 8290 for (p = badword; *p != NUL; ) | |
| 8291 wbadword[bi++] = mb_cptr2char_adv(&p); | |
| 8292 wbadword[bi++] = 0; | |
| 8293 gi = 0; | |
| 8294 for (p = goodword; *p != NUL; ) | |
| 8295 wgoodword[gi++] = mb_cptr2char_adv(&p); | |
| 8296 wgoodword[gi++] = 0; | |
| 8297 | |
| 8298 /* | |
| 8299 * The idea is to go from start to end over the words. So long as | |
| 8300 * characters are equal just continue, this always gives the lowest score. | |
| 8301 * When there is a difference try several alternatives. Each alternative | |
| 8302 * increases "score" for the edit distance. Some of the alternatives are | |
| 8303 * pushed unto a stack and tried later, some are tried right away. At the | |
| 8304 * end of the word the score for one alternative is known. The lowest | |
| 8305 * possible score is stored in "minscore". | |
| 8306 */ | |
| 8307 stackidx = 0; | |
| 8308 bi = 0; | |
| 8309 gi = 0; | |
| 8310 score = 0; | |
| 8311 minscore = limit + 1; | |
| 8312 | |
| 8313 for (;;) | |
| 8314 { | |
| 8315 /* Skip over an equal part, score remains the same. */ | |
| 8316 for (;;) | |
| 8317 { | |
| 8318 bc = wbadword[bi]; | |
| 8319 gc = wgoodword[gi]; | |
| 8320 | |
| 8321 if (bc != gc) /* stop at a char that's different */ | |
| 8322 break; | |
| 8323 if (bc == NUL) /* both words end */ | |
| 8324 { | |
| 8325 if (score < minscore) | |
| 8326 minscore = score; | |
| 8327 goto pop; /* do next alternative */ | |
| 8328 } | |
| 8329 ++bi; | |
| 8330 ++gi; | |
| 8331 } | |
| 8332 | |
| 8333 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 8334 { | |
| 8335 do | |
| 8336 { | |
| 8337 if ((score += SCORE_DEL) >= minscore) | |
| 8338 goto pop; /* do next alternative */ | |
| 8339 } while (wbadword[++bi] != NUL); | |
| 8340 minscore = score; | |
| 8341 } | |
| 8342 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 8343 { | |
| 8344 do | |
| 8345 { | |
| 8346 if ((score += SCORE_INS) >= minscore) | |
| 8347 goto pop; /* do next alternative */ | |
| 8348 } while (wgoodword[++gi] != NUL); | |
| 8349 minscore = score; | |
| 8350 } | |
| 8351 else /* both words continue */ | |
| 8352 { | |
| 8353 /* If not close to the limit, perform a change. Only try changes | |
| 8354 * that may lead to a lower score than "minscore". | |
| 8355 * round 0: try deleting a char from badword | |
| 8356 * round 1: try inserting a char in badword */ | |
| 8357 for (round = 0; round <= 1; ++round) | |
| 8358 { | |
| 8359 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 8360 if (score_off < minscore) | |
| 8361 { | |
| 8362 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 8363 { | |
| 8364 /* Near the limit, rest of the words must match. We | |
| 8365 * can check that right now, no need to push an item | |
| 8366 * onto the stack. */ | |
| 8367 bi2 = bi + 1 - round; | |
| 8368 gi2 = gi + round; | |
| 8369 while (wgoodword[gi2] == wbadword[bi2]) | |
| 8370 { | |
| 8371 if (wgoodword[gi2] == NUL) | |
| 8372 { | |
| 8373 minscore = score_off; | |
| 8374 break; | |
| 8375 } | |
| 8376 ++bi2; | |
| 8377 ++gi2; | |
| 8378 } | |
| 8379 } | |
| 8380 else | |
| 8381 { | |
| 8382 /* try deleting a character from badword later */ | |
| 8383 stack[stackidx].badi = bi + 1 - round; | |
| 8384 stack[stackidx].goodi = gi + round; | |
| 8385 stack[stackidx].score = score_off; | |
| 8386 ++stackidx; | |
| 8387 } | |
| 8388 } | |
| 8389 } | |
| 8390 | |
| 8391 if (score + SCORE_SWAP < minscore) | |
| 8392 { | |
| 8393 /* If swapping two characters makes a match then the | |
| 8394 * substitution is more expensive, thus there is no need to | |
| 8395 * try both. */ | |
| 8396 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1]) | |
| 8397 { | |
| 8398 /* Swap two characters, that is: skip them. */ | |
| 8399 gi += 2; | |
| 8400 bi += 2; | |
| 8401 score += SCORE_SWAP; | |
| 8402 continue; | |
| 8403 } | |
| 8404 } | |
| 8405 | |
| 8406 /* Substitute one character for another which is the same | |
| 8407 * thing as deleting a character from both goodword and badword. | |
| 8408 * Use a better score when there is only a case difference. */ | |
| 8409 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 8410 score += SCORE_ICASE; | |
| 8411 else | |
| 8412 { | |
| 8413 /* For a similar character use SCORE_SIMILAR. */ | |
| 8414 if (slang != NULL | |
| 8415 && slang->sl_has_map | |
| 8416 && similar_chars(slang, gc, bc)) | |
| 8417 score += SCORE_SIMILAR; | |
| 8418 else | |
| 8419 score += SCORE_SUBST; | |
| 8420 } | |
| 8421 | |
| 8422 if (score < minscore) | |
| 8423 { | |
| 8424 /* Do the substitution. */ | |
| 8425 ++gi; | |
| 8426 ++bi; | |
| 8427 continue; | |
| 8428 } | |
| 8429 } | |
| 8430 pop: | |
| 8431 /* | |
| 8432 * Get here to try the next alternative, pop it from the stack. | |
| 8433 */ | |
| 8434 if (stackidx == 0) /* stack is empty, finished */ | |
| 8435 break; | |
| 8436 | |
| 8437 /* pop an item from the stack */ | |
| 8438 --stackidx; | |
| 8439 gi = stack[stackidx].goodi; | |
| 8440 bi = stack[stackidx].badi; | |
| 8441 score = stack[stackidx].score; | |
| 8442 } | |
| 8443 | |
| 8444 /* When the score goes over "limit" it may actually be much higher. | |
| 8445 * Return a very large number to avoid going below the limit when giving a | |
| 8446 * bonus. */ | |
| 8447 if (minscore > limit) | |
| 8448 return SCORE_MAXMAX; | |
| 8449 return minscore; | |
| 8450 } | |
| 8451 #endif | |
| 8452 | |
| 714 | 8453 /* |
| 8454 * ":spellinfo" | |
| 8455 */ | |
| 8456 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8457 ex_spellinfo(exarg_T *eap UNUSED) |
| 714 | 8458 { |
| 8459 int lpi; | |
| 8460 langp_T *lp; | |
| 8461 char_u *p; | |
| 8462 | |
| 8463 if (no_spell_checking(curwin)) | |
| 8464 return; | |
| 8465 | |
| 8466 msg_start(); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8467 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
|
8468 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8469 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 714 | 8470 msg_puts((char_u *)"file: "); |
| 8471 msg_puts(lp->lp_slang->sl_fname); | |
| 8472 msg_putchar('\n'); | |
| 8473 p = lp->lp_slang->sl_info; | |
| 8474 if (p != NULL) | |
| 8475 { | |
| 8476 msg_puts(p); | |
| 8477 msg_putchar('\n'); | |
| 8478 } | |
| 8479 } | |
| 8480 msg_end(); | |
| 8481 } | |
| 8482 | |
| 625 | 8483 #define DUMPFLAG_KEEPCASE 1 /* round 2: keep-case tree */ |
| 8484 #define DUMPFLAG_COUNT 2 /* include word count */ | |
| 701 | 8485 #define DUMPFLAG_ICASE 4 /* ignore case when finding matches */ |
| 710 | 8486 #define DUMPFLAG_ONECAP 8 /* pattern starts with capital */ |
| 8487 #define DUMPFLAG_ALLCAP 16 /* pattern is all capitals */ | |
| 625 | 8488 |
| 351 | 8489 /* |
| 8490 * ":spelldump" | |
| 8491 */ | |
| 8492 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8493 ex_spelldump(exarg_T *eap) |
| 351 | 8494 { |
| 5382 | 8495 char_u *spl; |
| 8496 long dummy; | |
| 8497 | |
| 701 | 8498 if (no_spell_checking(curwin)) |
| 8499 return; | |
| 5382 | 8500 get_option_value((char_u*)"spl", &dummy, &spl, OPT_LOCAL); |
| 8501 | |
| 8502 /* Create a new empty buffer in a new window. */ | |
| 701 | 8503 do_cmdline_cmd((char_u *)"new"); |
| 5382 | 8504 |
| 8505 /* enable spelling locally in the new window */ | |
| 8506 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
|
8507 set_option_value((char_u*)"spl", dummy, spl, OPT_LOCAL); |
| 5382 | 8508 vim_free(spl); |
| 8509 | |
|
11121
778c10516955
patch 8.0.0448: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
10950
diff
changeset
|
8510 if (!BUFEMPTY()) |
| 701 | 8511 return; |
| 8512 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8513 spell_dump_compl(NULL, 0, NULL, eap->forceit ? DUMPFLAG_COUNT : 0); |
| 701 | 8514 |
| 8515 /* Delete the empty line that we started with. */ | |
| 8516 if (curbuf->b_ml.ml_line_count > 1) | |
| 8517 ml_delete(curbuf->b_ml.ml_line_count, FALSE); | |
| 8518 | |
| 8519 redraw_later(NOT_VALID); | |
| 8520 } | |
| 8521 | |
| 8522 /* | |
| 8523 * Go through all possible words and: | |
| 8524 * 1. When "pat" is NULL: dump a list of all words in the current buffer. | |
| 8525 * "ic" and "dir" are not used. | |
| 8526 * 2. When "pat" is not NULL: add matching words to insert mode completion. | |
| 8527 */ | |
| 8528 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8529 spell_dump_compl( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8530 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
|
8531 int ic, /* ignore case */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8532 int *dir, /* direction for adding matches */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8533 int dumpflags_arg) /* DUMPFLAG_* */ |
| 701 | 8534 { |
| 351 | 8535 langp_T *lp; |
| 8536 slang_T *slang; | |
| 8537 idx_T arridx[MAXWLEN]; | |
| 8538 int curi[MAXWLEN]; | |
| 8539 char_u word[MAXWLEN]; | |
| 8540 int c; | |
| 8541 char_u *byts; | |
| 8542 idx_T *idxs; | |
| 8543 linenr_T lnum = 0; | |
| 8544 int round; | |
| 8545 int depth; | |
| 8546 int n; | |
| 8547 int flags; | |
| 381 | 8548 char_u *region_names = NULL; /* region names being used */ |
| 8549 int do_region = TRUE; /* dump region names and numbers */ | |
| 8550 char_u *p; | |
| 500 | 8551 int lpi; |
| 701 | 8552 int dumpflags = dumpflags_arg; |
| 8553 int patlen; | |
| 8554 | |
| 710 | 8555 /* When ignoring case or when the pattern starts with capital pass this on |
| 8556 * to dump_word(). */ | |
| 8557 if (pat != NULL) | |
| 8558 { | |
| 8559 if (ic) | |
| 8560 dumpflags |= DUMPFLAG_ICASE; | |
| 8561 else | |
| 8562 { | |
| 8563 n = captype(pat, NULL); | |
| 8564 if (n == WF_ONECAP) | |
| 8565 dumpflags |= DUMPFLAG_ONECAP; | |
| 8566 else if (n == WF_ALLCAP | |
| 8567 #ifdef FEAT_MBYTE | |
| 714 | 8568 && (int)STRLEN(pat) > mb_ptr2len(pat) |
| 710 | 8569 #else |
| 714 | 8570 && (int)STRLEN(pat) > 1 |
| 710 | 8571 #endif |
| 8572 ) | |
| 8573 dumpflags |= DUMPFLAG_ALLCAP; | |
| 8574 } | |
| 8575 } | |
| 351 | 8576 |
| 381 | 8577 /* Find out if we can support regions: All languages must support the same |
| 8578 * regions or none at all. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8579 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
|
8580 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8581 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 381 | 8582 p = lp->lp_slang->sl_regions; |
| 8583 if (p[0] != 0) | |
| 8584 { | |
| 8585 if (region_names == NULL) /* first language with regions */ | |
| 8586 region_names = p; | |
| 8587 else if (STRCMP(region_names, p) != 0) | |
| 8588 { | |
| 8589 do_region = FALSE; /* region names are different */ | |
| 8590 break; | |
| 8591 } | |
| 8592 } | |
| 8593 } | |
| 8594 | |
| 8595 if (do_region && region_names != NULL) | |
| 8596 { | |
| 701 | 8597 if (pat == NULL) |
| 8598 { | |
| 8599 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names); | |
| 8600 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 8601 } | |
| 381 | 8602 } |
| 8603 else | |
| 8604 do_region = FALSE; | |
| 8605 | |
| 8606 /* | |
| 8607 * Loop over all files loaded for the entries in 'spelllang'. | |
| 8608 */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8609 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
|
8610 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8611 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 351 | 8612 slang = lp->lp_slang; |
| 500 | 8613 if (slang->sl_fbyts == NULL) /* reloading failed */ |
| 8614 continue; | |
| 351 | 8615 |
| 701 | 8616 if (pat == NULL) |
| 8617 { | |
| 8618 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname); | |
| 8619 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 8620 } | |
| 8621 | |
| 8622 /* When matching with a pattern and there are no prefixes only use | |
| 8623 * parts of the tree that match "pat". */ | |
| 8624 if (pat != NULL && slang->sl_pbyts == NULL) | |
| 835 | 8625 patlen = (int)STRLEN(pat); |
| 701 | 8626 else |
| 840 | 8627 patlen = -1; |
| 351 | 8628 |
| 8629 /* round 1: case-folded tree | |
| 8630 * round 2: keep-case tree */ | |
| 8631 for (round = 1; round <= 2; ++round) | |
| 8632 { | |
| 8633 if (round == 1) | |
| 8634 { | |
| 701 | 8635 dumpflags &= ~DUMPFLAG_KEEPCASE; |
| 351 | 8636 byts = slang->sl_fbyts; |
| 8637 idxs = slang->sl_fidxs; | |
| 8638 } | |
| 8639 else | |
| 8640 { | |
| 701 | 8641 dumpflags |= DUMPFLAG_KEEPCASE; |
| 351 | 8642 byts = slang->sl_kbyts; |
| 8643 idxs = slang->sl_kidxs; | |
| 8644 } | |
| 8645 if (byts == NULL) | |
| 8646 continue; /* array is empty */ | |
| 8647 | |
| 8648 depth = 0; | |
| 8649 arridx[0] = 0; | |
| 8650 curi[0] = 1; | |
| 701 | 8651 while (depth >= 0 && !got_int |
| 8652 && (pat == NULL || !compl_interrupted)) | |
| 351 | 8653 { |
| 8654 if (curi[depth] > byts[arridx[depth]]) | |
| 8655 { | |
| 8656 /* Done all bytes at this node, go up one level. */ | |
| 8657 --depth; | |
| 8658 line_breakcheck(); | |
|
10277
154d5a2e7395
commit https://github.com/vim/vim/commit/472e85970ee3a80abd824bef510df12e9cfe9e96
Christian Brabandt <cb@256bit.org>
parents:
10042
diff
changeset
|
8659 ins_compl_check_keys(50, FALSE); |
| 351 | 8660 } |
| 8661 else | |
| 8662 { | |
| 8663 /* Do one more byte at this node. */ | |
| 8664 n = arridx[depth] + curi[depth]; | |
| 8665 ++curi[depth]; | |
| 8666 c = byts[n]; | |
| 8667 if (c == 0) | |
| 8668 { | |
| 8669 /* End of word, deal with the word. | |
| 8670 * Don't use keep-case words in the fold-case tree, | |
| 8671 * they will appear in the keep-case tree. | |
| 8672 * Only use the word when the region matches. */ | |
| 8673 flags = (int)idxs[n]; | |
| 8674 if ((round == 2 || (flags & WF_KEEPCAP) == 0) | |
| 500 | 8675 && (flags & WF_NEEDCOMP) == 0 |
| 381 | 8676 && (do_region |
| 8677 || (flags & WF_REGION) == 0 | |
| 390 | 8678 || (((unsigned)flags >> 16) |
| 351 | 8679 & lp->lp_region) != 0)) |
| 8680 { | |
| 8681 word[depth] = NUL; | |
| 381 | 8682 if (!do_region) |
| 8683 flags &= ~WF_REGION; | |
| 355 | 8684 |
| 8685 /* Dump the basic word if there is no prefix or | |
| 8686 * when it's the first one. */ | |
| 390 | 8687 c = (unsigned)flags >> 24; |
| 355 | 8688 if (c == 0 || curi[depth] == 2) |
| 701 | 8689 { |
| 8690 dump_word(slang, word, pat, dir, | |
| 8691 dumpflags, flags, lnum); | |
| 8692 if (pat == NULL) | |
| 8693 ++lnum; | |
| 8694 } | |
| 351 | 8695 |
| 8696 /* Apply the prefix, if there is one. */ | |
| 355 | 8697 if (c != 0) |
| 701 | 8698 lnum = dump_prefixes(slang, word, pat, dir, |
| 8699 dumpflags, flags, lnum); | |
| 351 | 8700 } |
| 8701 } | |
| 8702 else | |
| 8703 { | |
| 8704 /* Normal char, go one level deeper. */ | |
| 8705 word[depth++] = c; | |
| 8706 arridx[depth] = idxs[n]; | |
| 8707 curi[depth] = 1; | |
| 701 | 8708 |
| 8709 /* Check if this characters matches with the pattern. | |
| 8710 * If not skip the whole tree below it. | |
| 710 | 8711 * Always ignore case here, dump_word() will check |
| 8712 * proper case later. This isn't exactly right when | |
| 8713 * length changes for multi-byte characters with | |
| 8714 * ignore case... */ | |
| 8715 if (depth <= patlen | |
| 8716 && MB_STRNICMP(word, pat, depth) != 0) | |
| 701 | 8717 --depth; |
| 351 | 8718 } |
| 8719 } | |
| 8720 } | |
| 8721 } | |
| 8722 } | |
| 8723 } | |
| 8724 | |
| 8725 /* | |
| 8726 * Dump one word: apply case modifications and append a line to the buffer. | |
| 701 | 8727 * When "lnum" is zero add insert mode completion. |
| 8728 */ | |
| 8729 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8730 dump_word( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8731 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8732 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8733 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8734 int *dir, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8735 int dumpflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8736 int wordflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8737 linenr_T lnum) |
| 351 | 8738 { |
| 8739 int keepcap = FALSE; | |
| 8740 char_u *p; | |
| 625 | 8741 char_u *tw; |
| 351 | 8742 char_u cword[MAXWLEN]; |
| 381 | 8743 char_u badword[MAXWLEN + 10]; |
| 8744 int i; | |
| 710 | 8745 int flags = wordflags; |
| 8746 | |
| 8747 if (dumpflags & DUMPFLAG_ONECAP) | |
| 8748 flags |= WF_ONECAP; | |
| 8749 if (dumpflags & DUMPFLAG_ALLCAP) | |
| 8750 flags |= WF_ALLCAP; | |
| 351 | 8751 |
| 625 | 8752 if ((dumpflags & DUMPFLAG_KEEPCASE) == 0 && (flags & WF_CAPMASK) != 0) |
| 351 | 8753 { |
| 8754 /* Need to fix case according to "flags". */ | |
| 8755 make_case_word(word, cword, flags); | |
| 8756 p = cword; | |
| 8757 } | |
| 8758 else | |
| 8759 { | |
| 8760 p = word; | |
| 625 | 8761 if ((dumpflags & DUMPFLAG_KEEPCASE) |
| 8762 && ((captype(word, NULL) & WF_KEEPCAP) == 0 | |
| 389 | 8763 || (flags & WF_FIXCAP) != 0)) |
| 351 | 8764 keepcap = TRUE; |
| 8765 } | |
| 625 | 8766 tw = p; |
| 351 | 8767 |
| 701 | 8768 if (pat == NULL) |
| 8769 { | |
| 8770 /* Add flags and regions after a slash. */ | |
| 8771 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap) | |
| 8772 { | |
| 8773 STRCPY(badword, p); | |
| 8774 STRCAT(badword, "/"); | |
| 8775 if (keepcap) | |
| 8776 STRCAT(badword, "="); | |
| 8777 if (flags & WF_BANNED) | |
| 8778 STRCAT(badword, "!"); | |
| 8779 else if (flags & WF_RARE) | |
| 8780 STRCAT(badword, "?"); | |
| 8781 if (flags & WF_REGION) | |
| 8782 for (i = 0; i < 7; ++i) | |
| 8783 if (flags & (0x10000 << i)) | |
| 8784 sprintf((char *)badword + STRLEN(badword), "%d", i + 1); | |
| 8785 p = badword; | |
| 8786 } | |
| 8787 | |
| 8788 if (dumpflags & DUMPFLAG_COUNT) | |
| 8789 { | |
| 8790 hashitem_T *hi; | |
| 8791 | |
| 8792 /* Include the word count for ":spelldump!". */ | |
| 8793 hi = hash_find(&slang->sl_wordcount, tw); | |
| 8794 if (!HASHITEM_EMPTY(hi)) | |
| 8795 { | |
| 8796 vim_snprintf((char *)IObuff, IOSIZE, "%s\t%d", | |
| 625 | 8797 tw, HI2WC(hi)->wc_count); |
| 701 | 8798 p = IObuff; |
| 8799 } | |
| 8800 } | |
| 8801 | |
| 8802 ml_append(lnum, p, (colnr_T)0, FALSE); | |
| 8803 } | |
| 710 | 8804 else if (((dumpflags & DUMPFLAG_ICASE) |
| 8805 ? MB_STRNICMP(p, pat, STRLEN(pat)) == 0 | |
| 8806 : STRNCMP(p, pat, STRLEN(pat)) == 0) | |
| 701 | 8807 && ins_compl_add_infercase(p, (int)STRLEN(p), |
| 942 | 8808 p_ic, NULL, *dir, 0) == OK) |
| 710 | 8809 /* if dir was BACKWARD then honor it just once */ |
| 8810 *dir = FORWARD; | |
| 351 | 8811 } |
| 8812 | |
| 8813 /* | |
| 372 | 8814 * For ":spelldump": Find matching prefixes for "word". Prepend each to |
| 8815 * "word" and append a line to the buffer. | |
| 701 | 8816 * When "lnum" is zero add insert mode completion. |
| 351 | 8817 * Return the updated line number. |
| 8818 */ | |
| 8819 static linenr_T | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8820 dump_prefixes( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8821 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8822 char_u *word, /* case-folded word */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8823 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8824 int *dir, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8825 int dumpflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8826 int flags, /* flags with prefix ID */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8827 linenr_T startlnum) |
| 351 | 8828 { |
| 8829 idx_T arridx[MAXWLEN]; | |
| 8830 int curi[MAXWLEN]; | |
| 8831 char_u prefix[MAXWLEN]; | |
| 455 | 8832 char_u word_up[MAXWLEN]; |
| 8833 int has_word_up = FALSE; | |
| 351 | 8834 int c; |
| 8835 char_u *byts; | |
| 8836 idx_T *idxs; | |
| 8837 linenr_T lnum = startlnum; | |
| 8838 int depth; | |
| 8839 int n; | |
| 8840 int len; | |
| 8841 int i; | |
| 8842 | |
| 626 | 8843 /* If the word starts with a lower-case letter make the word with an |
| 455 | 8844 * upper-case letter in word_up[]. */ |
| 8845 c = PTR2CHAR(word); | |
| 8846 if (SPELL_TOUPPER(c) != c) | |
| 8847 { | |
| 8848 onecap_copy(word, word_up, TRUE); | |
| 8849 has_word_up = TRUE; | |
| 8850 } | |
| 8851 | |
| 351 | 8852 byts = slang->sl_pbyts; |
| 8853 idxs = slang->sl_pidxs; | |
| 8854 if (byts != NULL) /* array not is empty */ | |
| 8855 { | |
| 8856 /* | |
| 8857 * Loop over all prefixes, building them byte-by-byte in prefix[]. | |
| 390 | 8858 * When at the end of a prefix check that it supports "flags". |
| 351 | 8859 */ |
| 8860 depth = 0; | |
| 8861 arridx[0] = 0; | |
| 8862 curi[0] = 1; | |
| 8863 while (depth >= 0 && !got_int) | |
| 8864 { | |
| 390 | 8865 n = arridx[depth]; |
| 8866 len = byts[n]; | |
| 8867 if (curi[depth] > len) | |
| 351 | 8868 { |
| 8869 /* Done all bytes at this node, go up one level. */ | |
| 8870 --depth; | |
| 8871 line_breakcheck(); | |
| 8872 } | |
| 8873 else | |
| 8874 { | |
| 8875 /* Do one more byte at this node. */ | |
| 390 | 8876 n += curi[depth]; |
| 351 | 8877 ++curi[depth]; |
| 8878 c = byts[n]; | |
| 8879 if (c == 0) | |
| 8880 { | |
| 8881 /* End of prefix, find out how many IDs there are. */ | |
| 8882 for (i = 1; i < len; ++i) | |
| 8883 if (byts[n + i] != 0) | |
| 8884 break; | |
| 8885 curi[depth] += i - 1; | |
| 8886 | |
| 455 | 8887 c = valid_word_prefix(i, n, flags, word, slang, FALSE); |
| 8888 if (c != 0) | |
| 351 | 8889 { |
| 376 | 8890 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1); |
| 701 | 8891 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 8892 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 8893 : flags, lnum); |
| 8894 if (lnum != 0) | |
| 8895 ++lnum; | |
| 351 | 8896 } |
| 455 | 8897 |
| 8898 /* Check for prefix that matches the word when the | |
| 8899 * first letter is upper-case, but only if the prefix has | |
| 8900 * a condition. */ | |
| 8901 if (has_word_up) | |
| 8902 { | |
| 8903 c = valid_word_prefix(i, n, flags, word_up, slang, | |
| 8904 TRUE); | |
| 8905 if (c != 0) | |
| 8906 { | |
| 8907 vim_strncpy(prefix + depth, word_up, | |
| 8908 MAXWLEN - depth - 1); | |
| 701 | 8909 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 8910 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 8911 : flags, lnum); |
| 8912 if (lnum != 0) | |
| 8913 ++lnum; | |
| 455 | 8914 } |
| 8915 } | |
| 351 | 8916 } |
| 8917 else | |
| 8918 { | |
| 8919 /* Normal char, go one level deeper. */ | |
| 8920 prefix[depth++] = c; | |
| 8921 arridx[depth] = idxs[n]; | |
| 8922 curi[depth] = 1; | |
| 8923 } | |
| 8924 } | |
| 8925 } | |
| 8926 } | |
| 8927 | |
| 8928 return lnum; | |
| 8929 } | |
| 8930 | |
| 498 | 8931 /* |
| 626 | 8932 * Move "p" to the end of word "start". |
| 8933 * Uses the spell-checking word characters. | |
| 498 | 8934 */ |
| 8935 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8936 spell_to_word_end(char_u *start, win_T *win) |
| 498 | 8937 { |
| 8938 char_u *p = start; | |
| 8939 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8940 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
|
8941 MB_PTR_ADV(p); |
| 498 | 8942 return p; |
| 8943 } | |
| 8944 | |
| 475 | 8945 #if defined(FEAT_INS_EXPAND) || defined(PROTO) |
| 8946 /* | |
| 626 | 8947 * For Insert mode completion CTRL-X s: |
| 8948 * Find start of the word in front of column "startcol". | |
| 8949 * We don't check if it is badly spelled, with completion we can only change | |
| 8950 * the word in front of the cursor. | |
| 475 | 8951 * Returns the column number of the word. |
| 8952 */ | |
| 8953 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8954 spell_word_start(int startcol) |
| 475 | 8955 { |
| 8956 char_u *line; | |
| 8957 char_u *p; | |
| 8958 int col = 0; | |
| 8959 | |
| 498 | 8960 if (no_spell_checking(curwin)) |
| 475 | 8961 return startcol; |
| 8962 | |
| 8963 /* Find a word character before "startcol". */ | |
| 8964 line = ml_get_curline(); | |
| 8965 for (p = line + startcol; p > line; ) | |
| 8966 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
8967 MB_PTR_BACK(line, p); |
| 5477 | 8968 if (spell_iswordp_nmw(p, curwin)) |
| 475 | 8969 break; |
| 8970 } | |
| 8971 | |
| 8972 /* Go back to start of the word. */ | |
| 8973 while (p > line) | |
| 8974 { | |
| 835 | 8975 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
|
8976 MB_PTR_BACK(line, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8977 if (!spell_iswordp(p, curwin)) |
| 475 | 8978 break; |
| 8979 col = 0; | |
| 8980 } | |
| 8981 | |
| 535 | 8982 return col; |
| 8983 } | |
| 8984 | |
| 8985 /* | |
| 8986 * Need to check for 'spellcapcheck' now, the word is removed before | |
| 8987 * expand_spelling() is called. Therefore the ugly global variable. | |
| 8988 */ | |
| 8989 static int spell_expand_need_cap; | |
| 8990 | |
| 8991 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8992 spell_expand_check_cap(colnr_T col) |
| 535 | 8993 { |
| 475 | 8994 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col); |
| 8995 } | |
| 8996 | |
| 8997 /* | |
| 8998 * Get list of spelling suggestions. | |
| 8999 * Used for Insert mode completion CTRL-X ?. | |
| 9000 * Returns the number of matches. The matches are in "matchp[]", array of | |
| 9001 * allocated strings. | |
| 9002 */ | |
| 9003 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
9004 expand_spelling( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
9005 linenr_T lnum UNUSED, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
9006 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
9007 char_u ***matchp) |
| 475 | 9008 { |
| 9009 garray_T ga; | |
| 9010 | |
| 625 | 9011 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap, TRUE); |
| 475 | 9012 *matchp = ga.ga_data; |
| 9013 return ga.ga_len; | |
| 9014 } | |
| 9015 #endif | |
| 9016 | |
| 737 | 9017 #endif /* FEAT_SPELL */ |