Mercurial > vim
annotate src/spell.c @ 15521:6d949e552e99 v8.1.0768
patch 8.1.0768: updating completions may cause the popup menu to flicker
commit https://github.com/vim/vim/commit/ae654385dfb2ae4c1d70789d1dce3676dba4dfbc
Author: Bram Moolenaar <Bram@vim.org>
Date: Thu Jan 17 21:09:05 2019 +0100
patch 8.1.0768: updating completions may cause the popup menu to flicker
Problem: Updating completions may cause the popup menu to flicker.
Solution: Avoid updating the text below the popup menu before drawing the
popup menu.
| author | Bram Moolenaar <Bram@vim.org> |
|---|---|
| date | Thu, 17 Jan 2019 21:15:06 +0100 |
| parents | 55ccc2d353bd |
| children | dd725a8ab112 |
| 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 #endif |
| 307 | 255 |
| 256 /* | |
| 344 | 257 * For finding suggestions: At each node in the tree these states are tried: |
| 330 | 258 */ |
| 259 typedef enum | |
| 260 { | |
| 344 | 261 STATE_START = 0, /* At start of node check for NUL bytes (goodword |
| 262 * ends); if badword ends there is a match, otherwise | |
| 263 * try splitting word. */ | |
| 375 | 264 STATE_NOPREFIX, /* try without prefix */ |
| 344 | 265 STATE_SPLITUNDO, /* Undo splitting. */ |
| 330 | 266 STATE_ENDNUL, /* Past NUL bytes at start of the node. */ |
| 267 STATE_PLAIN, /* Use each byte of the node. */ | |
| 268 STATE_DEL, /* Delete a byte from the bad word. */ | |
| 625 | 269 STATE_INS_PREP, /* Prepare for inserting bytes. */ |
| 330 | 270 STATE_INS, /* Insert a byte in the bad word. */ |
| 271 STATE_SWAP, /* Swap two bytes. */ | |
| 344 | 272 STATE_UNSWAP, /* Undo swap two characters. */ |
| 273 STATE_SWAP3, /* Swap two characters over three. */ | |
| 274 STATE_UNSWAP3, /* Undo Swap two characters over three. */ | |
| 275 STATE_UNROT3L, /* Undo rotate three characters left */ | |
| 276 STATE_UNROT3R, /* Undo rotate three characters right */ | |
| 330 | 277 STATE_REP_INI, /* Prepare for using REP items. */ |
| 278 STATE_REP, /* Use matching REP items from the .aff file. */ | |
| 279 STATE_REP_UNDO, /* Undo a REP item replacement. */ | |
| 280 STATE_FINAL /* End of this node. */ | |
| 281 } state_T; | |
| 282 | |
| 283 /* | |
| 346 | 284 * Struct to keep the state at each level in suggest_try_change(). |
| 323 | 285 */ |
| 286 typedef struct trystate_S | |
| 287 { | |
| 330 | 288 state_T ts_state; /* state at this level, STATE_ */ |
| 323 | 289 int ts_score; /* score */ |
| 344 | 290 idx_T ts_arridx; /* index in tree array, start of node */ |
| 330 | 291 short ts_curi; /* index in list of child nodes */ |
| 292 char_u ts_fidx; /* index in fword[], case-folded bad word */ | |
| 293 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */ | |
| 294 char_u ts_twordlen; /* valid length of tword[] */ | |
| 482 | 295 char_u ts_prefixdepth; /* stack depth for end of prefix or |
| 485 | 296 * PFD_PREFIXTREE or PFD_NOPREFIX */ |
| 297 char_u ts_flags; /* TSF_ flags */ | |
| 330 | 298 #ifdef FEAT_MBYTE |
| 299 char_u ts_tcharlen; /* number of bytes in tword character */ | |
| 300 char_u ts_tcharidx; /* current byte index in tword character */ | |
| 301 char_u ts_isdiff; /* DIFF_ values */ | |
| 302 char_u ts_fcharstart; /* index in fword where badword char started */ | |
| 303 #endif | |
| 483 | 304 char_u ts_prewordlen; /* length of word in "preword[]" */ |
| 305 char_u ts_splitoff; /* index in "tword" after last split */ | |
| 492 | 306 char_u ts_splitfidx; /* "ts_fidx" at word split */ |
| 483 | 307 char_u ts_complen; /* nr of compound words used */ |
| 485 | 308 char_u ts_compsplit; /* index for "compflags" where word was spit */ |
| 346 | 309 char_u ts_save_badflags; /* su_badflags saved here */ |
| 625 | 310 char_u ts_delidx; /* index in fword for char that was deleted, |
| 311 valid when "ts_flags" has TSF_DIDDEL */ | |
| 323 | 312 } trystate_T; |
| 313 | |
| 330 | 314 /* values for ts_isdiff */ |
| 315 #define DIFF_NONE 0 /* no different byte (yet) */ | |
| 316 #define DIFF_YES 1 /* different byte found */ | |
| 317 #define DIFF_INSERT 2 /* inserting character */ | |
| 318 | |
| 485 | 319 /* values for ts_flags */ |
| 320 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */ | |
| 321 #define TSF_DIDSPLIT 2 /* tried split at this point */ | |
| 625 | 322 #define TSF_DIDDEL 4 /* did a delete, "ts_delidx" has index */ |
| 485 | 323 |
| 375 | 324 /* special values ts_prefixdepth */ |
| 485 | 325 #define PFD_NOPREFIX 0xff /* not using prefixes */ |
| 482 | 326 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */ |
| 625 | 327 #define PFD_NOTSPECIAL 0xfd /* highest value that's not special */ |
| 375 | 328 |
| 339 | 329 /* mode values for find_word */ |
| 481 | 330 #define FIND_FOLDWORD 0 /* find word case-folded */ |
| 331 #define FIND_KEEPWORD 1 /* find keep-case word */ | |
| 332 #define FIND_PREFIX 2 /* find word after prefix */ | |
| 333 #define FIND_COMPOUND 3 /* find case-folded compound word */ | |
| 334 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */ | |
| 339 | 335 |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
336 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
|
337 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
|
338 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
|
339 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
|
340 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
|
341 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
|
342 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
|
343 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
|
344 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
|
345 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
|
346 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
|
347 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
|
348 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
|
349 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
|
350 static void spell_find_suggest(char_u *badptr, int badlen, suginfo_T *su, int maxcount, int banbadword, int need_cap, int interactive); |
| 372 | 351 #ifdef FEAT_EVAL |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
352 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
|
353 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
354 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
|
355 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
|
356 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
|
357 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
|
358 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
|
359 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
|
360 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
|
361 #ifdef FEAT_MBYTE |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
362 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
|
363 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
364 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
|
365 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
|
366 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
|
367 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
|
368 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
|
369 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
|
370 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
|
371 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
|
372 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
|
373 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
|
374 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
|
375 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
|
376 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
|
377 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
|
378 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
|
379 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
|
380 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
|
381 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
|
382 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
|
383 #ifdef FEAT_MBYTE |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
384 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
|
385 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
386 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
|
387 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
|
388 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
|
389 #ifdef FEAT_MBYTE |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
390 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
|
391 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
392 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
|
393 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
|
394 |
| 236 | 395 |
| 626 | 396 /* Remember what "z?" replaced. */ |
| 397 static char_u *repl_from = NULL; | |
| 398 static char_u *repl_to = NULL; | |
| 399 | |
| 236 | 400 /* |
| 401 * Main spell-checking function. | |
| 300 | 402 * "ptr" points to a character that could be the start of a word. |
| 534 | 403 * "*attrp" is set to the highlight index for a badly spelled word. For a |
| 404 * non-word or when it's OK it remains unchanged. | |
| 236 | 405 * This must only be called when 'spelllang' is not empty. |
| 323 | 406 * |
| 385 | 407 * "capcol" is used to check for a Capitalised word after the end of a |
| 408 * sentence. If it's zero then perform the check. Return the column where to | |
| 409 * check next, or -1 when no sentence end was found. If it's NULL then don't | |
| 410 * worry. | |
| 323 | 411 * |
| 236 | 412 * Returns the length of the word in bytes, also when it's OK, so that the |
| 413 * caller can skip over the word. | |
| 414 */ | |
| 415 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
416 spell_check( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
417 win_T *wp, /* current window */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
418 char_u *ptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
419 hlf_T *attrp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
420 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
|
421 int docount) /* count good words */ |
| 236 | 422 { |
| 423 matchinf_T mi; /* Most things are put in "mi" so that it can | |
| 424 be passed to functions quickly. */ | |
| 344 | 425 int nrlen = 0; /* found a number first */ |
| 385 | 426 int c; |
| 483 | 427 int wrongcaplen = 0; |
| 500 | 428 int lpi; |
| 625 | 429 int count_word = docount; |
| 236 | 430 |
| 307 | 431 /* A word never starts at a space or a control character. Return quickly |
| 432 * then, skipping over the character. */ | |
| 433 if (*ptr <= ' ') | |
| 434 return 1; | |
| 690 | 435 |
| 436 /* 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
|
437 if (wp->w_s->b_langp.ga_len == 0) |
| 690 | 438 return 1; |
| 439 | |
| 483 | 440 vim_memset(&mi, 0, sizeof(matchinf_T)); |
| 236 | 441 |
| 344 | 442 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and |
| 584 | 443 * 0X99FF. But always do check spelling to find "3GPP" and "11 |
| 444 * julifeest". */ | |
| 300 | 445 if (*ptr >= '0' && *ptr <= '9') |
| 446 { | |
|
7447
ad432f8f68fb
commit https://github.com/vim/vim/commit/887c1fea4a114e7170091942d0446c8882701b5b
Christian Brabandt <cb@256bit.org>
parents:
6949
diff
changeset
|
447 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
|
448 mi.mi_end = skipbin(ptr + 2); |
|
ad432f8f68fb
commit https://github.com/vim/vim/commit/887c1fea4a114e7170091942d0446c8882701b5b
Christian Brabandt <cb@256bit.org>
parents:
6949
diff
changeset
|
449 else if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X')) |
| 316 | 450 mi.mi_end = skiphex(ptr + 2); |
| 300 | 451 else |
| 344 | 452 mi.mi_end = skipdigits(ptr); |
| 835 | 453 nrlen = (int)(mi.mi_end - ptr); |
| 584 | 454 } |
| 346 | 455 |
| 456 /* Find the normal end of the word (until the next non-word character). */ | |
| 344 | 457 mi.mi_word = ptr; |
| 584 | 458 mi.mi_fend = ptr; |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
459 if (spell_iswordp(mi.mi_fend, wp)) |
| 344 | 460 { |
| 461 do | |
| 462 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
463 MB_PTR_ADV(mi.mi_fend); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
464 } 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
|
465 |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
466 if (capcol != NULL && *capcol == 0 && wp->w_s->b_cap_prog != NULL) |
| 385 | 467 { |
| 468 /* Check word starting with capital letter. */ | |
| 455 | 469 c = PTR2CHAR(ptr); |
| 385 | 470 if (!SPELL_ISUPPER(c)) |
| 483 | 471 wrongcaplen = (int)(mi.mi_fend - ptr); |
| 385 | 472 } |
| 473 } | |
| 474 if (capcol != NULL) | |
| 475 *capcol = -1; | |
| 344 | 476 |
| 477 /* We always use the characters up to the next non-word character, | |
| 478 * also for bad words. */ | |
| 479 mi.mi_end = mi.mi_fend; | |
| 480 | |
| 481 /* Check caps type later. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
482 mi.mi_capflags = 0; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
483 mi.mi_cend = NULL; |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
484 mi.mi_win = wp; |
| 344 | 485 |
| 483 | 486 /* case-fold the word with one non-word character, so that we can check |
| 487 * for the word end. */ | |
| 344 | 488 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
|
489 MB_PTR_ADV(mi.mi_fend); |
| 344 | 490 |
| 491 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword, | |
| 492 MAXWLEN + 1); | |
| 835 | 493 mi.mi_fwordlen = (int)STRLEN(mi.mi_fword); |
| 344 | 494 |
| 495 /* The word is bad unless we recognize it. */ | |
| 496 mi.mi_result = SP_BAD; | |
| 492 | 497 mi.mi_result2 = SP_BAD; |
| 344 | 498 |
| 499 /* | |
| 500 * Loop over the languages specified in 'spelllang'. | |
| 625 | 501 * We check them all, because a word may be matched longer in another |
| 502 * language. | |
| 344 | 503 */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
504 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
|
505 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
506 mi.mi_lp = LANGP_ENTRY(wp->w_s->b_langp, lpi); |
| 500 | 507 |
| 508 /* If reloading fails the language is still in the list but everything | |
| 509 * has been cleared. */ | |
| 510 if (mi.mi_lp->lp_slang->sl_fidxs == NULL) | |
| 511 continue; | |
| 512 | |
| 344 | 513 /* Check for a matching word in case-folded words. */ |
| 514 find_word(&mi, FIND_FOLDWORD); | |
| 515 | |
| 516 /* Check for a matching word in keep-case words. */ | |
| 517 find_word(&mi, FIND_KEEPWORD); | |
| 518 | |
| 519 /* Check for matching prefixes. */ | |
| 485 | 520 find_prefix(&mi, FIND_FOLDWORD); |
| 492 | 521 |
| 522 /* For a NOBREAK language, may want to use a word without a following | |
| 523 * word as a backup. */ | |
| 524 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD | |
| 525 && mi.mi_result2 != SP_BAD) | |
| 526 { | |
| 527 mi.mi_result = mi.mi_result2; | |
| 528 mi.mi_end = mi.mi_end2; | |
| 529 } | |
| 625 | 530 |
| 531 /* Count the word in the first language where it's found to be OK. */ | |
| 532 if (count_word && mi.mi_result == SP_OK) | |
| 533 { | |
| 534 count_common_word(mi.mi_lp->lp_slang, ptr, | |
| 535 (int)(mi.mi_end - ptr), 1); | |
| 536 count_word = FALSE; | |
| 537 } | |
| 344 | 538 } |
| 539 | |
| 540 if (mi.mi_result != SP_OK) | |
| 541 { | |
| 346 | 542 /* If we found a number skip over it. Allows for "42nd". Do flag |
| 543 * rare and local words, e.g., "3GPP". */ | |
| 344 | 544 if (nrlen > 0) |
| 346 | 545 { |
| 546 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED) | |
| 547 return nrlen; | |
| 548 } | |
| 344 | 549 |
| 550 /* When we are at a non-word character there is no error, just | |
| 551 * skip over the character (try looking for a word after it). */ | |
| 5477 | 552 else if (!spell_iswordp_nmw(ptr, wp)) |
| 243 | 553 { |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
554 if (capcol != NULL && wp->w_s->b_cap_prog != NULL) |
| 385 | 555 { |
| 556 regmatch_T regmatch; | |
| 6375 | 557 int r; |
| 385 | 558 |
| 559 /* Check for end of sentence. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
560 regmatch.regprog = wp->w_s->b_cap_prog; |
| 385 | 561 regmatch.rm_ic = FALSE; |
| 6375 | 562 r = vim_regexec(®match, ptr, 0); |
| 563 wp->w_s->b_cap_prog = regmatch.regprog; | |
| 564 if (r) | |
| 385 | 565 *capcol = (int)(regmatch.endp[0] - ptr); |
| 566 } | |
| 567 | |
| 344 | 568 #ifdef FEAT_MBYTE |
| 569 if (has_mbyte) | |
| 474 | 570 return (*mb_ptr2len)(ptr); |
| 344 | 571 #endif |
| 572 return 1; | |
| 300 | 573 } |
| 483 | 574 else if (mi.mi_end == ptr) |
| 575 /* Always include at least one character. Required for when there | |
| 576 * 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
|
577 MB_PTR_ADV(mi.mi_end); |
| 492 | 578 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
|
579 && LANGP_ENTRY(wp->w_s->b_langp, 0)->lp_slang->sl_nobreak) |
| 492 | 580 { |
| 581 char_u *p, *fp; | |
| 582 int save_result = mi.mi_result; | |
| 583 | |
| 584 /* First language in 'spelllang' is NOBREAK. Find first position | |
| 585 * 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
|
586 mi.mi_lp = LANGP_ENTRY(wp->w_s->b_langp, 0); |
| 500 | 587 if (mi.mi_lp->lp_slang->sl_fidxs != NULL) |
| 588 { | |
| 589 p = mi.mi_word; | |
| 590 fp = mi.mi_fword; | |
| 591 for (;;) | |
| 592 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
593 MB_PTR_ADV(p); |
|
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
594 MB_PTR_ADV(fp); |
| 500 | 595 if (p >= mi.mi_end) |
| 596 break; | |
| 835 | 597 mi.mi_compoff = (int)(fp - mi.mi_fword); |
| 500 | 598 find_word(&mi, FIND_COMPOUND); |
| 599 if (mi.mi_result != SP_BAD) | |
| 600 { | |
| 601 mi.mi_end = p; | |
| 602 break; | |
| 603 } | |
| 604 } | |
| 605 mi.mi_result = save_result; | |
| 606 } | |
| 492 | 607 } |
| 243 | 608 |
| 344 | 609 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED) |
| 534 | 610 *attrp = HLF_SPB; |
| 344 | 611 else if (mi.mi_result == SP_RARE) |
| 534 | 612 *attrp = HLF_SPR; |
| 344 | 613 else |
| 534 | 614 *attrp = HLF_SPL; |
| 243 | 615 } |
| 616 | |
| 483 | 617 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE)) |
| 618 { | |
| 619 /* Report SpellCap only when the word isn't badly spelled. */ | |
| 534 | 620 *attrp = HLF_SPC; |
| 483 | 621 return wrongcaplen; |
| 622 } | |
| 623 | |
| 300 | 624 return (int)(mi.mi_end - ptr); |
| 236 | 625 } |
| 626 | |
| 627 /* | |
| 300 | 628 * Check if the word at "mip->mi_word" is in the tree. |
| 339 | 629 * When "mode" is FIND_FOLDWORD check in fold-case word tree. |
| 630 * When "mode" is FIND_KEEPWORD check in keep-case word tree. | |
| 631 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word | |
| 632 * tree. | |
| 300 | 633 * |
| 634 * For a match mip->mi_result is updated. | |
| 243 | 635 */ |
| 636 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
637 find_word(matchinf_T *mip, int mode) |
| 243 | 638 { |
| 324 | 639 idx_T arridx = 0; |
| 300 | 640 int endlen[MAXWLEN]; /* length at possible word endings */ |
| 324 | 641 idx_T endidx[MAXWLEN]; /* possible word endings */ |
| 300 | 642 int endidxcnt = 0; |
| 643 int len; | |
| 644 int wlen = 0; | |
| 645 int flen; | |
| 646 int c; | |
| 647 char_u *ptr; | |
| 324 | 648 idx_T lo, hi, m; |
| 243 | 649 #ifdef FEAT_MBYTE |
| 300 | 650 char_u *s; |
| 490 | 651 #endif |
| 339 | 652 char_u *p; |
| 307 | 653 int res = SP_BAD; |
| 300 | 654 slang_T *slang = mip->mi_lp->lp_slang; |
| 655 unsigned flags; | |
| 656 char_u *byts; | |
| 324 | 657 idx_T *idxs; |
| 481 | 658 int word_ends; |
| 485 | 659 int prefix_found; |
| 492 | 660 int nobreak_result; |
| 481 | 661 |
| 662 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND) | |
| 236 | 663 { |
| 300 | 664 /* Check for word with matching case in keep-case tree. */ |
| 665 ptr = mip->mi_word; | |
| 666 flen = 9999; /* no case folding, always enough bytes */ | |
| 667 byts = slang->sl_kbyts; | |
| 668 idxs = slang->sl_kidxs; | |
| 481 | 669 |
| 670 if (mode == FIND_KEEPCOMPOUND) | |
| 671 /* Skip over the previously found word(s). */ | |
| 672 wlen += mip->mi_compoff; | |
| 236 | 673 } |
| 674 else | |
| 675 { | |
| 300 | 676 /* Check for case-folded in case-folded tree. */ |
| 677 ptr = mip->mi_fword; | |
| 678 flen = mip->mi_fwordlen; /* available case-folded bytes */ | |
| 679 byts = slang->sl_fbyts; | |
| 680 idxs = slang->sl_fidxs; | |
| 339 | 681 |
| 682 if (mode == FIND_PREFIX) | |
| 683 { | |
| 684 /* Skip over the prefix. */ | |
| 685 wlen = mip->mi_prefixlen; | |
| 686 flen -= mip->mi_prefixlen; | |
| 687 } | |
| 481 | 688 else if (mode == FIND_COMPOUND) |
| 689 { | |
| 690 /* Skip over the previously found word(s). */ | |
| 691 wlen = mip->mi_compoff; | |
| 692 flen -= mip->mi_compoff; | |
| 693 } | |
| 694 | |
| 243 | 695 } |
| 696 | |
| 300 | 697 if (byts == NULL) |
| 698 return; /* array is empty */ | |
| 236 | 699 |
| 700 /* | |
| 307 | 701 * Repeat advancing in the tree until: |
| 702 * - there is a byte that doesn't match, | |
| 703 * - we reach the end of the tree, | |
| 704 * - or we reach the end of the line. | |
| 236 | 705 */ |
| 300 | 706 for (;;) |
| 236 | 707 { |
| 346 | 708 if (flen <= 0 && *mip->mi_fend != NUL) |
| 339 | 709 flen = fold_more(mip); |
| 300 | 710 |
| 711 len = byts[arridx++]; | |
| 712 | |
| 713 /* If the first possible byte is a zero the word could end here. | |
| 714 * Remember this index, we first check for the longest word. */ | |
| 715 if (byts[arridx] == 0) | |
| 716 { | |
| 307 | 717 if (endidxcnt == MAXWLEN) |
| 718 { | |
| 719 /* Must be a corrupted spell file. */ | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
720 emsg(_(e_format)); |
| 307 | 721 return; |
| 722 } | |
| 300 | 723 endlen[endidxcnt] = wlen; |
| 724 endidx[endidxcnt++] = arridx++; | |
| 725 --len; | |
| 726 | |
| 727 /* Skip over the zeros, there can be several flag/region | |
| 728 * combinations. */ | |
| 729 while (len > 0 && byts[arridx] == 0) | |
| 730 { | |
| 731 ++arridx; | |
| 732 --len; | |
| 733 } | |
| 734 if (len == 0) | |
| 735 break; /* no children, word must end here */ | |
| 736 } | |
| 737 | |
| 738 /* Stop looking at end of the line. */ | |
| 739 if (ptr[wlen] == NUL) | |
| 740 break; | |
| 741 | |
| 742 /* Perform a binary search in the list of accepted bytes. */ | |
| 743 c = ptr[wlen]; | |
| 346 | 744 if (c == TAB) /* <Tab> is handled like <Space> */ |
| 745 c = ' '; | |
| 300 | 746 lo = arridx; |
| 747 hi = arridx + len - 1; | |
| 748 while (lo < hi) | |
| 749 { | |
| 750 m = (lo + hi) / 2; | |
| 751 if (byts[m] > c) | |
| 752 hi = m - 1; | |
| 753 else if (byts[m] < c) | |
| 754 lo = m + 1; | |
| 755 else | |
| 756 { | |
| 757 lo = hi = m; | |
| 758 break; | |
| 236 | 759 } |
| 760 } | |
| 300 | 761 |
| 762 /* Stop if there is no matching byte. */ | |
| 763 if (hi < lo || byts[lo] != c) | |
| 764 break; | |
| 765 | |
| 766 /* Continue at the child (if there is one). */ | |
| 767 arridx = idxs[lo]; | |
| 768 ++wlen; | |
| 769 --flen; | |
| 346 | 770 |
| 771 /* One space in the good word may stand for several spaces in the | |
| 772 * checked word. */ | |
| 773 if (c == ' ') | |
| 774 { | |
| 775 for (;;) | |
| 776 { | |
| 777 if (flen <= 0 && *mip->mi_fend != NUL) | |
| 778 flen = fold_more(mip); | |
| 779 if (ptr[wlen] != ' ' && ptr[wlen] != TAB) | |
| 780 break; | |
| 781 ++wlen; | |
| 782 --flen; | |
| 783 } | |
| 784 } | |
| 236 | 785 } |
| 786 | |
| 300 | 787 /* |
| 788 * Verify that one of the possible endings is valid. Try the longest | |
| 789 * first. | |
| 790 */ | |
| 791 while (endidxcnt > 0) | |
| 792 { | |
| 793 --endidxcnt; | |
| 794 arridx = endidx[endidxcnt]; | |
| 795 wlen = endlen[endidxcnt]; | |
| 236 | 796 |
| 300 | 797 #ifdef FEAT_MBYTE |
| 798 if ((*mb_head_off)(ptr, ptr + wlen) > 0) | |
| 799 continue; /* not at first byte of character */ | |
| 800 #endif | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
801 if (spell_iswordp(ptr + wlen, mip->mi_win)) |
| 481 | 802 { |
| 492 | 803 if (slang->sl_compprog == NULL && !slang->sl_nobreak) |
| 481 | 804 continue; /* next char is a word character */ |
| 805 word_ends = FALSE; | |
| 806 } | |
| 807 else | |
| 808 word_ends = TRUE; | |
| 485 | 809 /* The prefix flag is before compound flags. Once a valid prefix flag |
| 810 * has been found we try compound flags. */ | |
| 811 prefix_found = FALSE; | |
| 300 | 812 |
| 813 #ifdef FEAT_MBYTE | |
| 339 | 814 if (mode != FIND_KEEPWORD && has_mbyte) |
| 300 | 815 { |
| 816 /* Compute byte length in original word, length may change | |
| 339 | 817 * when folding case. This can be slow, take a shortcut when the |
| 818 * case-folded word is equal to the keep-case word. */ | |
| 300 | 819 p = mip->mi_word; |
| 339 | 820 if (STRNCMP(ptr, p, wlen) != 0) |
| 821 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
822 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
|
823 MB_PTR_ADV(p); |
| 835 | 824 wlen = (int)(p - mip->mi_word); |
| 339 | 825 } |
| 300 | 826 } |
| 827 #endif | |
| 236 | 828 |
| 339 | 829 /* Check flags and region. For FIND_PREFIX check the condition and |
| 830 * prefix ID. | |
| 831 * Repeat this if there are more flags/region alternatives until there | |
| 832 * is a match. */ | |
| 833 res = SP_BAD; | |
| 834 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0; | |
| 835 --len, ++arridx) | |
| 300 | 836 { |
| 837 flags = idxs[arridx]; | |
| 324 | 838 |
| 339 | 839 /* For the fold-case tree check that the case of the checked word |
| 840 * matches with what the word in the tree requires. | |
| 841 * For keep-case tree the case is always right. For prefixes we | |
| 842 * don't bother to check. */ | |
| 843 if (mode == FIND_FOLDWORD) | |
| 300 | 844 { |
| 845 if (mip->mi_cend != mip->mi_word + wlen) | |
| 846 { | |
| 323 | 847 /* mi_capflags was set for a different word length, need |
| 848 * to do it again. */ | |
| 300 | 849 mip->mi_cend = mip->mi_word + wlen; |
| 323 | 850 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend); |
| 300 | 851 } |
| 852 | |
| 346 | 853 if (mip->mi_capflags == WF_KEEPCAP |
| 854 || !spell_valid_case(mip->mi_capflags, flags)) | |
| 339 | 855 continue; |
| 300 | 856 } |
| 236 | 857 |
| 339 | 858 /* When mode is FIND_PREFIX the word must support the prefix: |
| 859 * check the prefix ID and the condition. Do that for the list at | |
| 366 | 860 * mip->mi_prefarridx that find_prefix() filled. */ |
| 485 | 861 else if (mode == FIND_PREFIX && !prefix_found) |
| 481 | 862 { |
| 366 | 863 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx, |
| 390 | 864 flags, |
| 455 | 865 mip->mi_word + mip->mi_cprefixlen, slang, |
| 866 FALSE); | |
| 366 | 867 if (c == 0) |
| 339 | 868 continue; |
| 366 | 869 |
| 870 /* Use the WF_RARE flag for a rare prefix. */ | |
| 871 if (c & WF_RAREPFX) | |
| 872 flags |= WF_RARE; | |
| 485 | 873 prefix_found = TRUE; |
| 339 | 874 } |
| 875 | |
| 492 | 876 if (slang->sl_nobreak) |
| 877 { | |
| 878 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND) | |
| 879 && (flags & WF_BANNED) == 0) | |
| 880 { | |
| 881 /* NOBREAK: found a valid following word. That's all we | |
| 882 * need to know, so return. */ | |
| 883 mip->mi_result = SP_OK; | |
| 884 break; | |
| 885 } | |
| 886 } | |
| 887 | |
| 888 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND | |
| 889 || !word_ends)) | |
| 481 | 890 { |
| 974 | 891 /* If there is no compound flag or the word is shorter than |
| 483 | 892 * COMPOUNDMIN reject it quickly. |
| 893 * Makes you wonder why someone puts a compound flag on a word | |
| 481 | 894 * that's too short... Myspell compatibility requires this |
| 895 * anyway. */ | |
| 490 | 896 if (((unsigned)flags >> 24) == 0 |
| 897 || wlen - mip->mi_compoff < slang->sl_compminlen) | |
| 483 | 898 continue; |
| 500 | 899 #ifdef FEAT_MBYTE |
| 900 /* For multi-byte chars check character length against | |
| 901 * COMPOUNDMIN. */ | |
| 902 if (has_mbyte | |
| 501 | 903 && slang->sl_compminlen > 0 |
| 500 | 904 && mb_charlen_len(mip->mi_word + mip->mi_compoff, |
| 905 wlen - mip->mi_compoff) < slang->sl_compminlen) | |
| 906 continue; | |
| 907 #endif | |
| 483 | 908 |
| 809 | 909 /* Limit the number of compound words to COMPOUNDWORDMAX if no |
| 490 | 910 * maximum for syllables is specified. */ |
| 809 | 911 if (!word_ends && mip->mi_complen + mip->mi_compextra + 2 |
| 912 > slang->sl_compmax | |
| 490 | 913 && slang->sl_compsylmax == MAXWLEN) |
| 483 | 914 continue; |
| 915 | |
| 819 | 916 /* Don't allow compounding on a side where an affix was added, |
| 917 * unless COMPOUNDPERMITFLAG was used. */ | |
| 918 if (mip->mi_complen > 0 && (flags & WF_NOCOMPBEF)) | |
| 919 continue; | |
| 920 if (!word_ends && (flags & WF_NOCOMPAFT)) | |
| 921 continue; | |
| 922 | |
| 485 | 923 /* Quickly check if compounding is possible with this flag. */ |
| 495 | 924 if (!byte_in_str(mip->mi_complen == 0 |
| 485 | 925 ? slang->sl_compstartflags |
| 926 : slang->sl_compallflags, | |
| 495 | 927 ((unsigned)flags >> 24))) |
| 481 | 928 continue; |
| 929 | |
| 1762 | 930 /* If there is a match with a CHECKCOMPOUNDPATTERN rule |
| 931 * discard the compound word. */ | |
| 932 if (match_checkcompoundpattern(ptr, wlen, &slang->sl_comppat)) | |
| 933 continue; | |
| 934 | |
| 490 | 935 if (mode == FIND_COMPOUND) |
| 936 { | |
| 937 int capflags; | |
| 938 | |
| 939 /* Need to check the caps type of the appended compound | |
| 940 * word. */ | |
| 941 #ifdef FEAT_MBYTE | |
| 942 if (has_mbyte && STRNCMP(ptr, mip->mi_word, | |
| 943 mip->mi_compoff) != 0) | |
| 944 { | |
| 945 /* case folding may have changed the length */ | |
| 946 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
|
947 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
|
948 MB_PTR_ADV(p); |
| 490 | 949 } |
| 950 else | |
| 951 #endif | |
| 952 p = mip->mi_word + mip->mi_compoff; | |
| 953 capflags = captype(p, mip->mi_word + wlen); | |
| 954 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP | |
| 955 && (flags & WF_FIXCAP) != 0)) | |
| 956 continue; | |
| 957 | |
| 958 if (capflags != WF_ALLCAP) | |
| 959 { | |
| 960 /* When the character before the word is a word | |
| 961 * character we do not accept a Onecap word. We do | |
| 962 * accept a no-caps word, even when the dictionary | |
| 963 * word specifies ONECAP. */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
964 MB_PTR_BACK(mip->mi_word, p); |
| 5477 | 965 if (spell_iswordp_nmw(p, mip->mi_win) |
| 490 | 966 ? capflags == WF_ONECAP |
| 967 : (flags & WF_ONECAP) != 0 | |
| 968 && capflags != WF_ONECAP) | |
| 969 continue; | |
| 970 } | |
| 971 } | |
| 972 | |
| 483 | 973 /* If the word ends the sequence of compound flags of the |
| 714 | 974 * words must match with one of the COMPOUNDRULE items and |
| 483 | 975 * the number of syllables must not be too large. */ |
| 976 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24); | |
| 977 mip->mi_compflags[mip->mi_complen + 1] = NUL; | |
| 978 if (word_ends) | |
| 979 { | |
| 980 char_u fword[MAXWLEN]; | |
| 981 | |
| 982 if (slang->sl_compsylmax < MAXWLEN) | |
| 983 { | |
| 984 /* "fword" is only needed for checking syllables. */ | |
| 985 if (ptr == mip->mi_word) | |
| 986 (void)spell_casefold(ptr, wlen, fword, MAXWLEN); | |
| 987 else | |
| 988 vim_strncpy(fword, ptr, endlen[endidxcnt]); | |
| 989 } | |
| 990 if (!can_compound(slang, fword, mip->mi_compflags)) | |
| 991 continue; | |
| 992 } | |
| 1762 | 993 else if (slang->sl_comprules != NULL |
| 994 && !match_compoundrule(slang, mip->mi_compflags)) | |
| 995 /* The compound flags collected so far do not match any | |
| 996 * COMPOUNDRULE, discard the compounded word. */ | |
| 997 continue; | |
| 481 | 998 } |
| 999 | |
| 500 | 1000 /* Check NEEDCOMPOUND: can't use word without compounding. */ |
| 1001 else if (flags & WF_NEEDCOMP) | |
| 1002 continue; | |
| 1003 | |
| 492 | 1004 nobreak_result = SP_OK; |
| 1005 | |
| 481 | 1006 if (!word_ends) |
| 1007 { | |
| 492 | 1008 int save_result = mip->mi_result; |
| 1009 char_u *save_end = mip->mi_end; | |
| 501 | 1010 langp_T *save_lp = mip->mi_lp; |
| 1011 int lpi; | |
| 492 | 1012 |
| 1013 /* Check that a valid word follows. If there is one and we | |
| 1014 * are compounding, it will set "mi_result", thus we are | |
| 1015 * always finished here. For NOBREAK we only check that a | |
| 1016 * valid word follows. | |
| 481 | 1017 * Recursive! */ |
| 492 | 1018 if (slang->sl_nobreak) |
| 1019 mip->mi_result = SP_BAD; | |
| 481 | 1020 |
| 1021 /* Find following word in case-folded tree. */ | |
| 1022 mip->mi_compoff = endlen[endidxcnt]; | |
| 1023 #ifdef FEAT_MBYTE | |
| 1024 if (has_mbyte && mode == FIND_KEEPWORD) | |
| 1025 { | |
| 1026 /* Compute byte length in case-folded word from "wlen": | |
| 1027 * byte length in keep-case word. Length may change when | |
| 1028 * folding case. This can be slow, take a shortcut when | |
| 1029 * the case-folded word is equal to the keep-case word. */ | |
| 1030 p = mip->mi_fword; | |
| 1031 if (STRNCMP(ptr, p, wlen) != 0) | |
| 1032 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
1033 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
|
1034 MB_PTR_ADV(p); |
| 835 | 1035 mip->mi_compoff = (int)(p - mip->mi_fword); |
| 481 | 1036 } |
| 1037 } | |
| 1038 #endif | |
|
8956
d9e671c5afe6
commit https://github.com/vim/vim/commit/ba53435144f46eaaa53c63a62e748b3feee9742c
Christian Brabandt <cb@256bit.org>
parents:
8524
diff
changeset
|
1039 #if 0 /* Disabled, see below */ |
| 485 | 1040 c = mip->mi_compoff; |
|
8956
d9e671c5afe6
commit https://github.com/vim/vim/commit/ba53435144f46eaaa53c63a62e748b3feee9742c
Christian Brabandt <cb@256bit.org>
parents:
8524
diff
changeset
|
1041 #endif |
| 483 | 1042 ++mip->mi_complen; |
| 809 | 1043 if (flags & WF_COMPROOT) |
| 1044 ++mip->mi_compextra; | |
| 501 | 1045 |
| 1046 /* For NOBREAK we need to try all NOBREAK languages, at least | |
| 1047 * to find the ".add" file(s). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1048 for (lpi = 0; lpi < mip->mi_win->w_s->b_langp.ga_len; ++lpi) |
| 501 | 1049 { |
| 1050 if (slang->sl_nobreak) | |
| 1051 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1052 mip->mi_lp = LANGP_ENTRY(mip->mi_win->w_s->b_langp, lpi); |
| 501 | 1053 if (mip->mi_lp->lp_slang->sl_fidxs == NULL |
| 1054 || !mip->mi_lp->lp_slang->sl_nobreak) | |
| 1055 continue; | |
| 1056 } | |
| 1057 | |
| 1058 find_word(mip, FIND_COMPOUND); | |
| 1059 | |
| 1060 /* When NOBREAK any word that matches is OK. Otherwise we | |
| 1061 * need to find the longest match, thus try with keep-case | |
| 1062 * and prefix too. */ | |
| 492 | 1063 if (!slang->sl_nobreak || mip->mi_result == SP_BAD) |
| 1064 { | |
| 501 | 1065 /* Find following word in keep-case tree. */ |
| 1066 mip->mi_compoff = wlen; | |
| 1067 find_word(mip, FIND_KEEPCOMPOUND); | |
| 1068 | |
| 819 | 1069 #if 0 /* Disabled, a prefix must not appear halfway a compound word, |
| 1070 unless the COMPOUNDPERMITFLAG is used and then it can't be a | |
| 1071 postponed prefix. */ | |
| 501 | 1072 if (!slang->sl_nobreak || mip->mi_result == SP_BAD) |
| 1073 { | |
| 1074 /* Check for following word with prefix. */ | |
| 1075 mip->mi_compoff = c; | |
| 1076 find_prefix(mip, FIND_COMPOUND); | |
| 1077 } | |
| 819 | 1078 #endif |
| 492 | 1079 } |
| 501 | 1080 |
| 1081 if (!slang->sl_nobreak) | |
| 1082 break; | |
| 492 | 1083 } |
| 483 | 1084 --mip->mi_complen; |
| 809 | 1085 if (flags & WF_COMPROOT) |
| 1086 --mip->mi_compextra; | |
| 501 | 1087 mip->mi_lp = save_lp; |
| 485 | 1088 |
| 492 | 1089 if (slang->sl_nobreak) |
| 1090 { | |
| 1091 nobreak_result = mip->mi_result; | |
| 1092 mip->mi_result = save_result; | |
| 1093 mip->mi_end = save_end; | |
| 1094 } | |
| 1095 else | |
| 1096 { | |
| 1097 if (mip->mi_result == SP_OK) | |
| 1098 break; | |
| 1099 continue; | |
| 1100 } | |
| 481 | 1101 } |
| 1102 | |
| 339 | 1103 if (flags & WF_BANNED) |
| 1104 res = SP_BANNED; | |
| 1105 else if (flags & WF_REGION) | |
| 1106 { | |
| 1107 /* Check region. */ | |
| 390 | 1108 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0) |
| 300 | 1109 res = SP_OK; |
| 339 | 1110 else |
| 1111 res = SP_LOCAL; | |
| 300 | 1112 } |
| 339 | 1113 else if (flags & WF_RARE) |
| 1114 res = SP_RARE; | |
| 307 | 1115 else |
| 339 | 1116 res = SP_OK; |
| 1117 | |
| 492 | 1118 /* Always use the longest match and the best result. For NOBREAK |
| 1119 * we separately keep the longest match without a following good | |
| 1120 * word as a fall-back. */ | |
| 1121 if (nobreak_result == SP_BAD) | |
| 1122 { | |
| 1123 if (mip->mi_result2 > res) | |
| 1124 { | |
| 1125 mip->mi_result2 = res; | |
| 1126 mip->mi_end2 = mip->mi_word + wlen; | |
| 1127 } | |
| 1128 else if (mip->mi_result2 == res | |
| 1129 && mip->mi_end2 < mip->mi_word + wlen) | |
| 1130 mip->mi_end2 = mip->mi_word + wlen; | |
| 1131 } | |
| 1132 else if (mip->mi_result > res) | |
| 339 | 1133 { |
| 1134 mip->mi_result = res; | |
| 1135 mip->mi_end = mip->mi_word + wlen; | |
| 1136 } | |
| 351 | 1137 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen) |
| 339 | 1138 mip->mi_end = mip->mi_word + wlen; |
| 1139 | |
| 492 | 1140 if (mip->mi_result == SP_OK) |
| 339 | 1141 break; |
| 300 | 1142 } |
| 1143 | |
| 492 | 1144 if (mip->mi_result == SP_OK) |
| 300 | 1145 break; |
| 1146 } | |
| 236 | 1147 } |
| 1148 | |
| 323 | 1149 /* |
| 1762 | 1150 * Return TRUE if there is a match between the word ptr[wlen] and |
| 1151 * CHECKCOMPOUNDPATTERN rules, assuming that we will concatenate with another | |
| 1152 * word. | |
| 1153 * A match means that the first part of CHECKCOMPOUNDPATTERN matches at the | |
| 1154 * end of ptr[wlen] and the second part matches after it. | |
| 1155 */ | |
| 1156 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1157 match_checkcompoundpattern( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1158 char_u *ptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1159 int wlen, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1160 garray_T *gap) /* &sl_comppat */ |
| 1762 | 1161 { |
| 1162 int i; | |
| 1163 char_u *p; | |
| 1164 int len; | |
| 1165 | |
| 1166 for (i = 0; i + 1 < gap->ga_len; i += 2) | |
| 1167 { | |
| 1168 p = ((char_u **)gap->ga_data)[i + 1]; | |
| 1169 if (STRNCMP(ptr + wlen, p, STRLEN(p)) == 0) | |
| 1170 { | |
| 1171 /* Second part matches at start of following compound word, now | |
| 1172 * check if first part matches at end of previous word. */ | |
| 1173 p = ((char_u **)gap->ga_data)[i]; | |
| 1771 | 1174 len = (int)STRLEN(p); |
| 1762 | 1175 if (len <= wlen && STRNCMP(ptr + wlen - len, p, len) == 0) |
| 1176 return TRUE; | |
| 1177 } | |
| 1178 } | |
| 1179 return FALSE; | |
| 1180 } | |
| 1181 | |
| 1182 /* | |
| 626 | 1183 * Return TRUE if "flags" is a valid sequence of compound flags and "word" |
| 1184 * does not have too many syllables. | |
| 482 | 1185 */ |
| 1186 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1187 can_compound(slang_T *slang, char_u *word, char_u *flags) |
| 483 | 1188 { |
| 495 | 1189 #ifdef FEAT_MBYTE |
| 1190 char_u uflags[MAXWLEN * 2]; | |
| 1191 int i; | |
| 1192 #endif | |
| 1193 char_u *p; | |
| 483 | 1194 |
| 1195 if (slang->sl_compprog == NULL) | |
| 1196 return FALSE; | |
| 495 | 1197 #ifdef FEAT_MBYTE |
| 1198 if (enc_utf8) | |
| 1199 { | |
| 1200 /* Need to convert the single byte flags to utf8 characters. */ | |
| 1201 p = uflags; | |
| 1202 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
|
1203 p += utf_char2bytes(flags[i], p); |
| 495 | 1204 *p = NUL; |
| 1205 p = uflags; | |
| 1206 } | |
| 1207 else | |
| 1208 #endif | |
| 1209 p = flags; | |
| 6375 | 1210 if (!vim_regexec_prog(&slang->sl_compprog, FALSE, p, 0)) |
| 483 | 1211 return FALSE; |
| 1212 | |
| 490 | 1213 /* Count the number of syllables. This may be slow, do it last. If there |
| 1214 * are too many syllables AND the number of compound words is above | |
| 809 | 1215 * COMPOUNDWORDMAX then compounding is not allowed. */ |
| 483 | 1216 if (slang->sl_compsylmax < MAXWLEN |
| 1217 && count_syllables(slang, word) > slang->sl_compsylmax) | |
| 495 | 1218 return (int)STRLEN(flags) < slang->sl_compmax; |
| 483 | 1219 return TRUE; |
| 482 | 1220 } |
| 1221 | |
| 1222 /* | |
| 1762 | 1223 * Return TRUE when the sequence of flags in "compflags" plus "flag" can |
| 1224 * possibly form a valid compounded word. This also checks the COMPOUNDRULE | |
| 1225 * lines if they don't contain wildcards. | |
| 1226 */ | |
| 1227 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1228 can_be_compound( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1229 trystate_T *sp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1230 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1231 char_u *compflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1232 int flag) |
| 1762 | 1233 { |
| 1234 /* If the flag doesn't appear in sl_compstartflags or sl_compallflags | |
| 1235 * then it can't possibly compound. */ | |
| 1236 if (!byte_in_str(sp->ts_complen == sp->ts_compsplit | |
| 1237 ? slang->sl_compstartflags : slang->sl_compallflags, flag)) | |
| 1238 return FALSE; | |
| 1239 | |
| 1240 /* If there are no wildcards, we can check if the flags collected so far | |
| 1241 * possibly can form a match with COMPOUNDRULE patterns. This only | |
| 1242 * makes sense when we have two or more words. */ | |
| 1243 if (slang->sl_comprules != NULL && sp->ts_complen > sp->ts_compsplit) | |
| 1244 { | |
| 1245 int v; | |
| 1246 | |
| 1247 compflags[sp->ts_complen] = flag; | |
| 1248 compflags[sp->ts_complen + 1] = NUL; | |
| 1249 v = match_compoundrule(slang, compflags + sp->ts_compsplit); | |
| 1250 compflags[sp->ts_complen] = NUL; | |
| 1251 return v; | |
| 1252 } | |
| 1253 | |
| 1254 return TRUE; | |
| 1255 } | |
| 1256 | |
| 1257 | |
| 1258 /* | |
| 1259 * Return TRUE if the compound flags in compflags[] match the start of any | |
| 1260 * compound rule. This is used to stop trying a compound if the flags | |
| 1261 * collected so far can't possibly match any compound rule. | |
| 1262 * Caller must check that slang->sl_comprules is not NULL. | |
| 1263 */ | |
| 1264 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1265 match_compoundrule(slang_T *slang, char_u *compflags) |
| 1762 | 1266 { |
| 1267 char_u *p; | |
| 1268 int i; | |
| 1269 int c; | |
| 1270 | |
| 1271 /* loop over all the COMPOUNDRULE entries */ | |
| 1272 for (p = slang->sl_comprules; *p != NUL; ++p) | |
| 1273 { | |
| 1274 /* loop over the flags in the compound word we have made, match | |
| 1275 * them against the current rule entry */ | |
| 1276 for (i = 0; ; ++i) | |
| 1277 { | |
| 1278 c = compflags[i]; | |
| 1279 if (c == NUL) | |
| 1280 /* found a rule that matches for the flags we have so far */ | |
| 1281 return TRUE; | |
| 1282 if (*p == '/' || *p == NUL) | |
| 1283 break; /* end of rule, it's too short */ | |
| 1284 if (*p == '[') | |
| 1285 { | |
| 1286 int match = FALSE; | |
| 1287 | |
| 1288 /* compare against all the flags in [] */ | |
| 1289 ++p; | |
| 1290 while (*p != ']' && *p != NUL) | |
| 1291 if (*p++ == c) | |
| 1292 match = TRUE; | |
| 1293 if (!match) | |
| 1294 break; /* none matches */ | |
| 1295 } | |
| 1296 else if (*p != c) | |
| 1297 break; /* flag of word doesn't match flag in pattern */ | |
| 1298 ++p; | |
| 1299 } | |
| 1300 | |
| 1301 /* Skip to the next "/", where the next pattern starts. */ | |
| 1302 p = vim_strchr(p, '/'); | |
| 1303 if (p == NULL) | |
| 1304 break; | |
| 1305 } | |
| 1306 | |
| 1307 /* Checked all the rules and none of them match the flags, so there | |
| 1308 * can't possibly be a compound starting with these flags. */ | |
| 1309 return FALSE; | |
| 1310 } | |
| 1311 | |
| 1312 /* | |
| 390 | 1313 * Return non-zero if the prefix indicated by "arridx" matches with the prefix |
| 1314 * ID in "flags" for the word "word". | |
| 366 | 1315 * The WF_RAREPFX flag is included in the return value for a rare prefix. |
| 351 | 1316 */ |
| 1317 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1318 valid_word_prefix( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1319 int totprefcnt, /* nr of prefix IDs */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1320 int arridx, /* idx in sl_pidxs[] */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1321 int flags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1322 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1323 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1324 int cond_req) /* only use prefixes with a condition */ |
| 351 | 1325 { |
| 1326 int prefcnt; | |
| 1327 int pidx; | |
| 6375 | 1328 regprog_T **rp; |
| 390 | 1329 int prefid; |
| 1330 | |
| 1331 prefid = (unsigned)flags >> 24; | |
| 351 | 1332 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt) |
| 1333 { | |
| 1334 pidx = slang->sl_pidxs[arridx + prefcnt]; | |
| 1335 | |
| 1336 /* Check the prefix ID. */ | |
| 1337 if (prefid != (pidx & 0xff)) | |
| 1338 continue; | |
| 1339 | |
| 390 | 1340 /* Check if the prefix doesn't combine and the word already has a |
| 1341 * suffix. */ | |
| 1342 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC)) | |
| 1343 continue; | |
| 1344 | |
| 351 | 1345 /* Check the condition, if there is one. The condition index is |
| 366 | 1346 * stored in the two bytes above the prefix ID byte. */ |
| 6375 | 1347 rp = &slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff]; |
| 1348 if (*rp != NULL) | |
| 1349 { | |
| 1350 if (!vim_regexec_prog(rp, FALSE, word, 0)) | |
| 351 | 1351 continue; |
| 1352 } | |
| 455 | 1353 else if (cond_req) |
| 1354 continue; | |
| 1355 | |
| 1356 /* It's a match! Return the WF_ flags. */ | |
| 366 | 1357 return pidx; |
| 1358 } | |
| 1359 return 0; | |
| 351 | 1360 } |
| 1361 | |
| 1362 /* | |
| 339 | 1363 * Check if the word at "mip->mi_word" has a matching prefix. |
| 1364 * If it does, then check the following word. | |
| 1365 * | |
| 485 | 1366 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a |
| 1367 * prefix in a compound word. | |
| 1368 * | |
| 339 | 1369 * For a match mip->mi_result is updated. |
| 1370 */ | |
| 1371 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1372 find_prefix(matchinf_T *mip, int mode) |
| 339 | 1373 { |
| 1374 idx_T arridx = 0; | |
| 1375 int len; | |
| 1376 int wlen = 0; | |
| 1377 int flen; | |
| 1378 int c; | |
| 1379 char_u *ptr; | |
| 1380 idx_T lo, hi, m; | |
| 1381 slang_T *slang = mip->mi_lp->lp_slang; | |
| 1382 char_u *byts; | |
| 1383 idx_T *idxs; | |
| 1384 | |
| 375 | 1385 byts = slang->sl_pbyts; |
| 1386 if (byts == NULL) | |
| 1387 return; /* array is empty */ | |
| 1388 | |
| 339 | 1389 /* We use the case-folded word here, since prefixes are always |
| 1390 * case-folded. */ | |
| 1391 ptr = mip->mi_fword; | |
| 1392 flen = mip->mi_fwordlen; /* available case-folded bytes */ | |
| 485 | 1393 if (mode == FIND_COMPOUND) |
| 1394 { | |
| 1395 /* Skip over the previously found word(s). */ | |
| 1396 ptr += mip->mi_compoff; | |
| 1397 flen -= mip->mi_compoff; | |
| 1398 } | |
| 339 | 1399 idxs = slang->sl_pidxs; |
| 1400 | |
| 1401 /* | |
| 1402 * Repeat advancing in the tree until: | |
| 1403 * - there is a byte that doesn't match, | |
| 1404 * - we reach the end of the tree, | |
| 1405 * - or we reach the end of the line. | |
| 1406 */ | |
| 1407 for (;;) | |
| 1408 { | |
| 1409 if (flen == 0 && *mip->mi_fend != NUL) | |
| 1410 flen = fold_more(mip); | |
| 1411 | |
| 1412 len = byts[arridx++]; | |
| 1413 | |
| 1414 /* If the first possible byte is a zero the prefix could end here. | |
| 1415 * Check if the following word matches and supports the prefix. */ | |
| 1416 if (byts[arridx] == 0) | |
| 1417 { | |
| 1418 /* There can be several prefixes with different conditions. We | |
| 1419 * try them all, since we don't know which one will give the | |
| 1420 * longest match. The word is the same each time, pass the list | |
| 1421 * of possible prefixes to find_word(). */ | |
| 1422 mip->mi_prefarridx = arridx; | |
| 1423 mip->mi_prefcnt = len; | |
| 1424 while (len > 0 && byts[arridx] == 0) | |
| 1425 { | |
| 1426 ++arridx; | |
| 1427 --len; | |
| 1428 } | |
| 1429 mip->mi_prefcnt -= len; | |
| 1430 | |
| 1431 /* Find the word that comes after the prefix. */ | |
| 1432 mip->mi_prefixlen = wlen; | |
| 485 | 1433 if (mode == FIND_COMPOUND) |
| 1434 /* Skip over the previously found word(s). */ | |
| 1435 mip->mi_prefixlen += mip->mi_compoff; | |
| 1436 | |
| 455 | 1437 #ifdef FEAT_MBYTE |
| 1438 if (has_mbyte) | |
| 1439 { | |
| 1440 /* Case-folded length may differ from original length. */ | |
| 485 | 1441 mip->mi_cprefixlen = nofold_len(mip->mi_fword, |
| 1442 mip->mi_prefixlen, mip->mi_word); | |
| 455 | 1443 } |
| 1444 else | |
| 485 | 1445 mip->mi_cprefixlen = mip->mi_prefixlen; |
| 455 | 1446 #endif |
| 339 | 1447 find_word(mip, FIND_PREFIX); |
| 1448 | |
| 1449 | |
| 1450 if (len == 0) | |
| 1451 break; /* no children, word must end here */ | |
| 1452 } | |
| 1453 | |
| 1454 /* Stop looking at end of the line. */ | |
| 1455 if (ptr[wlen] == NUL) | |
| 1456 break; | |
| 1457 | |
| 1458 /* Perform a binary search in the list of accepted bytes. */ | |
| 1459 c = ptr[wlen]; | |
| 1460 lo = arridx; | |
| 1461 hi = arridx + len - 1; | |
| 1462 while (lo < hi) | |
| 1463 { | |
| 1464 m = (lo + hi) / 2; | |
| 1465 if (byts[m] > c) | |
| 1466 hi = m - 1; | |
| 1467 else if (byts[m] < c) | |
| 1468 lo = m + 1; | |
| 1469 else | |
| 1470 { | |
| 1471 lo = hi = m; | |
| 1472 break; | |
| 1473 } | |
| 1474 } | |
| 1475 | |
| 1476 /* Stop if there is no matching byte. */ | |
| 1477 if (hi < lo || byts[lo] != c) | |
| 1478 break; | |
| 1479 | |
| 1480 /* Continue at the child (if there is one). */ | |
| 1481 arridx = idxs[lo]; | |
| 1482 ++wlen; | |
| 1483 --flen; | |
| 1484 } | |
| 1485 } | |
| 1486 | |
| 1487 /* | |
| 1488 * Need to fold at least one more character. Do until next non-word character | |
| 626 | 1489 * for efficiency. Include the non-word character too. |
| 339 | 1490 * Return the length of the folded chars in bytes. |
| 1491 */ | |
| 1492 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1493 fold_more(matchinf_T *mip) |
| 339 | 1494 { |
| 1495 int flen; | |
| 1496 char_u *p; | |
| 1497 | |
| 1498 p = mip->mi_fend; | |
| 1499 do | |
| 1500 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
1501 MB_PTR_ADV(mip->mi_fend); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1502 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_win)); |
| 339 | 1503 |
| 626 | 1504 /* Include the non-word character so that we can check for the word end. */ |
| 339 | 1505 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
|
1506 MB_PTR_ADV(mip->mi_fend); |
| 339 | 1507 |
| 1508 (void)spell_casefold(p, (int)(mip->mi_fend - p), | |
| 1509 mip->mi_fword + mip->mi_fwordlen, | |
| 1510 MAXWLEN - mip->mi_fwordlen); | |
| 835 | 1511 flen = (int)STRLEN(mip->mi_fword + mip->mi_fwordlen); |
| 339 | 1512 mip->mi_fwordlen += flen; |
| 1513 return flen; | |
| 1514 } | |
| 1515 | |
| 1516 /* | |
| 323 | 1517 * Check case flags for a word. Return TRUE if the word has the requested |
| 1518 * case. | |
| 1519 */ | |
| 1520 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1521 spell_valid_case( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1522 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
|
1523 int treeflags) /* flags for the word in the spell tree */ |
| 323 | 1524 { |
| 389 | 1525 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0) |
| 323 | 1526 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0 |
| 474 | 1527 && ((treeflags & WF_ONECAP) == 0 |
| 1528 || (wordflags & WF_ONECAP) != 0))); | |
| 323 | 1529 } |
| 1530 | |
| 351 | 1531 /* |
| 1532 * Return TRUE if spell checking is not enabled. | |
| 1533 */ | |
| 1534 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1535 no_spell_checking(win_T *wp) |
| 498 | 1536 { |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1537 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
|
1538 || wp->w_s->b_langp.ga_len == 0) |
| 351 | 1539 { |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
1540 emsg(_("E756: Spell checking is not enabled")); |
| 351 | 1541 return TRUE; |
| 1542 } | |
| 1543 return FALSE; | |
| 1544 } | |
| 300 | 1545 |
| 236 | 1546 /* |
| 1547 * Move to next spell error. | |
| 500 | 1548 * "curline" is FALSE for "[s", "]s", "[S" and "]S". |
| 1549 * "curline" is TRUE to find word under/after cursor in the same line. | |
| 483 | 1550 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move |
| 1551 * to after badly spelled word before the cursor. | |
| 495 | 1552 * Return 0 if not found, length of the badly spelled word otherwise. |
| 236 | 1553 */ |
| 1554 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1555 spell_move_to( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1556 win_T *wp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1557 int dir, /* FORWARD or BACKWARD */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1558 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
|
1559 int curline, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1560 hlf_T *attrp) /* return: attributes of bad word or NULL |
| 534 | 1561 (only when "dir" is FORWARD) */ |
| 236 | 1562 { |
| 249 | 1563 linenr_T lnum; |
| 1564 pos_T found_pos; | |
| 495 | 1565 int found_len = 0; |
| 236 | 1566 char_u *line; |
| 1567 char_u *p; | |
| 346 | 1568 char_u *endp; |
| 534 | 1569 hlf_T attr; |
| 236 | 1570 int len; |
| 5519 | 1571 #ifdef FEAT_SYN_HL |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
1572 int has_syntax = syntax_present(wp); |
| 5519 | 1573 #endif |
| 249 | 1574 int col; |
| 1575 int can_spell; | |
| 346 | 1576 char_u *buf = NULL; |
| 1577 int buflen = 0; | |
| 1578 int skip = 0; | |
| 385 | 1579 int capcol = -1; |
| 500 | 1580 int found_one = FALSE; |
| 1581 int wrapped = FALSE; | |
| 236 | 1582 |
| 498 | 1583 if (no_spell_checking(wp)) |
| 495 | 1584 return 0; |
| 236 | 1585 |
| 249 | 1586 /* |
| 1587 * Start looking for bad word at the start of the line, because we can't | |
| 817 | 1588 * start halfway a word, we don't know where it starts or ends. |
| 249 | 1589 * |
| 1590 * When searching backwards, we continue in the line to find the last | |
| 1591 * bad word (in the cursor line: before the cursor). | |
| 346 | 1592 * |
| 1593 * We concatenate the start of the next line, so that wrapped words work | |
| 1594 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards | |
| 1595 * though... | |
| 249 | 1596 */ |
| 498 | 1597 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
|
1598 CLEAR_POS(&found_pos); |
| 236 | 1599 |
| 1600 while (!got_int) | |
| 1601 { | |
| 498 | 1602 line = ml_get_buf(wp->w_buffer, lnum, FALSE); |
| 346 | 1603 |
| 835 | 1604 len = (int)STRLEN(line); |
| 346 | 1605 if (buflen < len + MAXWLEN + 2) |
| 1606 { | |
| 1607 vim_free(buf); | |
| 1608 buflen = len + MAXWLEN + 2; | |
| 1609 buf = alloc(buflen); | |
| 1610 if (buf == NULL) | |
| 1611 break; | |
| 1612 } | |
| 1613 | |
| 385 | 1614 /* In first line check first word for Capital. */ |
| 1615 if (lnum == 1) | |
| 1616 capcol = 0; | |
| 1617 | |
| 1618 /* For checking first word with a capital skip white space. */ | |
| 1619 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
|
1620 capcol = getwhitecols(line); |
| 835 | 1621 else if (curline && wp == curwin) |
| 1622 { | |
| 1623 /* 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
|
1624 col = getwhitecols(line); |
| 835 | 1625 if (check_need_cap(lnum, col)) |
| 1626 capcol = col; | |
| 1627 | |
| 1628 /* Need to get the line again, may have looked at the previous | |
| 1629 * one. */ | |
| 1630 line = ml_get_buf(wp->w_buffer, lnum, FALSE); | |
| 1631 } | |
| 385 | 1632 |
| 346 | 1633 /* Copy the line into "buf" and append the start of the next line if |
| 1634 * possible. */ | |
| 1635 STRCPY(buf, line); | |
| 498 | 1636 if (lnum < wp->w_buffer->b_ml.ml_line_count) |
| 883 | 1637 spell_cat_line(buf + STRLEN(buf), |
| 1638 ml_get_buf(wp->w_buffer, lnum + 1, FALSE), MAXWLEN); | |
| 346 | 1639 |
| 1640 p = buf + skip; | |
| 1641 endp = buf + len; | |
| 1642 while (p < endp) | |
| 236 | 1643 { |
| 500 | 1644 /* When searching backward don't search after the cursor. Unless |
| 1645 * we wrapped around the end of the buffer. */ | |
| 300 | 1646 if (dir == BACKWARD |
| 498 | 1647 && lnum == wp->w_cursor.lnum |
| 500 | 1648 && !wrapped |
| 498 | 1649 && (colnr_T)(p - buf) >= wp->w_cursor.col) |
| 300 | 1650 break; |
| 249 | 1651 |
| 300 | 1652 /* start of word */ |
| 534 | 1653 attr = HLF_COUNT; |
| 625 | 1654 len = spell_check(wp, p, &attr, &capcol, FALSE); |
| 249 | 1655 |
| 534 | 1656 if (attr != HLF_COUNT) |
| 300 | 1657 { |
| 1658 /* We found a bad word. Check the attribute. */ | |
| 534 | 1659 if (allwords || attr == HLF_SPB) |
| 236 | 1660 { |
| 300 | 1661 /* When searching forward only accept a bad word after |
| 1662 * the cursor. */ | |
| 1663 if (dir == BACKWARD | |
| 500 | 1664 || lnum != wp->w_cursor.lnum |
| 498 | 1665 || (lnum == wp->w_cursor.lnum |
| 500 | 1666 && (wrapped |
| 1667 || (colnr_T)(curline ? p - buf + len | |
| 346 | 1668 : p - buf) |
| 500 | 1669 > wp->w_cursor.col))) |
| 236 | 1670 { |
| 5519 | 1671 #ifdef FEAT_SYN_HL |
| 300 | 1672 if (has_syntax) |
| 249 | 1673 { |
| 835 | 1674 col = (int)(p - buf); |
| 498 | 1675 (void)syn_get_id(wp, lnum, (colnr_T)col, |
| 1504 | 1676 FALSE, &can_spell, FALSE); |
| 857 | 1677 if (!can_spell) |
| 1678 attr = HLF_COUNT; | |
| 300 | 1679 } |
| 1680 else | |
| 737 | 1681 #endif |
| 300 | 1682 can_spell = TRUE; |
| 249 | 1683 |
| 300 | 1684 if (can_spell) |
| 1685 { | |
| 857 | 1686 found_one = TRUE; |
| 300 | 1687 found_pos.lnum = lnum; |
| 835 | 1688 found_pos.col = (int)(p - buf); |
| 249 | 1689 #ifdef FEAT_VIRTUALEDIT |
| 300 | 1690 found_pos.coladd = 0; |
| 249 | 1691 #endif |
| 300 | 1692 if (dir == FORWARD) |
| 1693 { | |
| 1694 /* No need to search further. */ | |
| 498 | 1695 wp->w_cursor = found_pos; |
| 346 | 1696 vim_free(buf); |
| 498 | 1697 if (attrp != NULL) |
| 1698 *attrp = attr; | |
| 495 | 1699 return len; |
| 249 | 1700 } |
| 483 | 1701 else if (curline) |
| 1702 /* Insert mode completion: put cursor after | |
| 1703 * the bad word. */ | |
| 1704 found_pos.col += len; | |
| 495 | 1705 found_len = len; |
| 249 | 1706 } |
| 236 | 1707 } |
| 857 | 1708 else |
| 1709 found_one = TRUE; | |
| 236 | 1710 } |
| 1711 } | |
| 1712 | |
| 300 | 1713 /* advance to character after the word */ |
| 1714 p += len; | |
| 385 | 1715 capcol -= len; |
| 236 | 1716 } |
| 1717 | |
| 483 | 1718 if (dir == BACKWARD && found_pos.lnum != 0) |
| 1719 { | |
| 500 | 1720 /* Use the last match in the line (before the cursor). */ |
| 498 | 1721 wp->w_cursor = found_pos; |
| 483 | 1722 vim_free(buf); |
| 495 | 1723 return found_len; |
| 483 | 1724 } |
| 1725 | |
| 323 | 1726 if (curline) |
| 346 | 1727 break; /* only check cursor line */ |
| 323 | 1728 |
|
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
|
1729 /* 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
|
1730 * 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
|
1731 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
|
1732 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
|
1733 |
| 236 | 1734 /* Advance to next line. */ |
| 249 | 1735 if (dir == BACKWARD) |
| 1736 { | |
| 500 | 1737 if (lnum > 1) |
| 1738 --lnum; | |
| 1739 else if (!p_ws) | |
| 1740 break; /* at first line and 'nowrapscan' */ | |
| 1741 else | |
| 1742 { | |
| 1743 /* Wrap around to the end of the buffer. May search the | |
| 1744 * starting line again and accept the last match. */ | |
| 1745 lnum = wp->w_buffer->b_ml.ml_line_count; | |
| 1746 wrapped = TRUE; | |
| 503 | 1747 if (!shortmess(SHM_SEARCH)) |
| 1748 give_warning((char_u *)_(top_bot_msg), TRUE); | |
| 500 | 1749 } |
| 385 | 1750 capcol = -1; |
| 249 | 1751 } |
| 1752 else | |
| 1753 { | |
| 500 | 1754 if (lnum < wp->w_buffer->b_ml.ml_line_count) |
| 1755 ++lnum; | |
| 1756 else if (!p_ws) | |
| 1757 break; /* at first line and 'nowrapscan' */ | |
| 1758 else | |
| 1759 { | |
| 1760 /* Wrap around to the start of the buffer. May search the | |
| 1761 * starting line again and accept the first match. */ | |
| 1762 lnum = 1; | |
| 1763 wrapped = TRUE; | |
| 503 | 1764 if (!shortmess(SHM_SEARCH)) |
| 1765 give_warning((char_u *)_(bot_top_msg), TRUE); | |
| 500 | 1766 } |
| 1767 | |
| 1768 /* If we are back at the starting line and there is no match then | |
| 1769 * 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
|
1770 if (lnum == wp->w_cursor.lnum && !found_one) |
| 346 | 1771 break; |
| 1772 | |
| 1773 /* Skip the characters at the start of the next line that were | |
| 1774 * included in a match crossing line boundaries. */ | |
| 534 | 1775 if (attr == HLF_COUNT) |
| 835 | 1776 skip = (int)(p - endp); |
| 346 | 1777 else |
| 1778 skip = 0; | |
| 385 | 1779 |
| 835 | 1780 /* Capcol skips over the inserted space. */ |
| 385 | 1781 --capcol; |
| 1782 | |
| 1783 /* But after empty line check first word in next line */ | |
| 1784 if (*skipwhite(line) == NUL) | |
| 1785 capcol = 0; | |
| 249 | 1786 } |
| 236 | 1787 |
| 1788 line_breakcheck(); | |
| 1789 } | |
| 1790 | |
| 346 | 1791 vim_free(buf); |
| 495 | 1792 return 0; |
| 346 | 1793 } |
| 1794 | |
| 1795 /* | |
| 1796 * For spell checking: concatenate the start of the following line "line" into | |
| 1797 * "buf", blanking-out special characters. Copy less then "maxlen" bytes. | |
| 1577 | 1798 * Keep the blanks at the start of the next line, this is used in win_line() |
| 1799 * to skip those bytes if the word was OK. | |
| 346 | 1800 */ |
| 1801 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1802 spell_cat_line(char_u *buf, char_u *line, int maxlen) |
| 346 | 1803 { |
| 1804 char_u *p; | |
| 1805 int n; | |
| 1806 | |
| 1807 p = skipwhite(line); | |
| 1808 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL) | |
| 1809 p = skipwhite(p + 1); | |
| 1810 | |
| 1811 if (*p != NUL) | |
| 1812 { | |
| 1577 | 1813 /* Only worth concatenating if there is something else than spaces to |
| 1814 * concatenate. */ | |
| 1815 n = (int)(p - line) + 1; | |
| 1816 if (n < maxlen - 1) | |
| 1817 { | |
| 1818 vim_memset(buf, ' ', n); | |
| 1819 vim_strncpy(buf + n, p, maxlen - 1 - n); | |
| 1820 } | |
| 346 | 1821 } |
| 236 | 1822 } |
| 1823 | |
| 626 | 1824 /* |
| 1825 * Structure used for the cookie argument of do_in_runtimepath(). | |
| 1826 */ | |
| 501 | 1827 typedef struct spelload_S |
| 1828 { | |
| 1829 char_u sl_lang[MAXWLEN + 1]; /* language name */ | |
| 1830 slang_T *sl_slang; /* resulting slang_T struct */ | |
| 1831 int sl_nobreak; /* NOBREAK language found */ | |
| 1832 } spelload_T; | |
| 1833 | |
| 236 | 1834 /* |
| 307 | 1835 * Load word list(s) for "lang" from Vim spell file(s). |
| 310 | 1836 * "lang" must be the language without the region: e.g., "en". |
| 236 | 1837 */ |
| 307 | 1838 static void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1839 spell_load_lang(char_u *lang) |
| 236 | 1840 { |
| 310 | 1841 char_u fname_enc[85]; |
| 236 | 1842 int r; |
| 501 | 1843 spelload_T sl; |
| 649 | 1844 int round; |
| 307 | 1845 |
| 310 | 1846 /* Copy the language name to pass it to spell_load_cb() as a cookie. |
| 307 | 1847 * It's truncated when an error is detected. */ |
| 501 | 1848 STRCPY(sl.sl_lang, lang); |
| 1849 sl.sl_slang = NULL; | |
| 1850 sl.sl_nobreak = FALSE; | |
| 307 | 1851 |
| 649 | 1852 /* We may retry when no spell file is found for the language, an |
| 1853 * autocommand may load it then. */ | |
| 1854 for (round = 1; round <= 2; ++round) | |
| 1855 { | |
| 1856 /* | |
| 1857 * Find the first spell file for "lang" in 'runtimepath' and load it. | |
| 1858 */ | |
| 1859 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 1860 #ifdef VMS |
| 1861 "spell/%s_%s.spl", | |
| 1862 #else | |
| 1863 "spell/%s.%s.spl", | |
| 1864 #endif | |
| 1865 lang, spell_enc()); | |
|
8524
2f57bbe870ea
commit https://github.com/vim/vim/commit/7f8989dd8a627af2185df381195351a913f3777f
Christian Brabandt <cb@256bit.org>
parents:
7835
diff
changeset
|
1866 r = do_in_runtimepath(fname_enc, 0, spell_load_cb, &sl); |
| 649 | 1867 |
| 1868 if (r == FAIL && *sl.sl_lang != NUL) | |
| 1869 { | |
| 1870 /* Try loading the ASCII version. */ | |
| 1871 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5, | |
| 2660 | 1872 #ifdef VMS |
| 1873 "spell/%s_ascii.spl", | |
| 1874 #else | |
| 1875 "spell/%s.ascii.spl", | |
| 1876 #endif | |
| 1877 lang); | |
|
8524
2f57bbe870ea
commit https://github.com/vim/vim/commit/7f8989dd8a627af2185df381195351a913f3777f
Christian Brabandt <cb@256bit.org>
parents:
7835
diff
changeset
|
1878 r = do_in_runtimepath(fname_enc, 0, spell_load_cb, &sl); |
| 649 | 1879 |
| 1880 if (r == FAIL && *sl.sl_lang != NUL && round == 1 | |
| 1881 && apply_autocmds(EVENT_SPELLFILEMISSING, lang, | |
| 1882 curbuf->b_fname, FALSE, curbuf)) | |
| 1883 continue; | |
| 1884 break; | |
|
13380
69517d67421f
patch 8.0.1564: too many #ifdefs
Christian Brabandt <cb@256bit.org>
parents:
13308
diff
changeset
|
1885 } |
| 714 | 1886 break; |
| 307 | 1887 } |
| 1888 | |
| 1889 if (r == FAIL) | |
| 649 | 1890 { |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
1891 smsg( |
| 2660 | 1892 #ifdef VMS |
| 1893 _("Warning: Cannot find word list \"%s_%s.spl\" or \"%s_ascii.spl\""), | |
| 1894 #else | |
| 1895 _("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""), | |
| 1896 #endif | |
| 483 | 1897 lang, spell_enc(), lang); |
| 649 | 1898 } |
| 501 | 1899 else if (sl.sl_slang != NULL) |
| 1900 { | |
| 626 | 1901 /* At least one file was loaded, now load ALL the additions. */ |
| 310 | 1902 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
|
1903 do_in_runtimepath(fname_enc, DIP_ALL, spell_load_cb, &sl); |
| 310 | 1904 } |
| 1905 } | |
| 1906 | |
| 1907 /* | |
| 1908 * Return the encoding used for spell checking: Use 'encoding', except that we | |
| 1909 * use "latin1" for "latin9". And limit to 60 characters (just in case). | |
| 1910 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1911 char_u * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1912 spell_enc(void) |
| 310 | 1913 { |
| 1914 | |
| 1915 #ifdef FEAT_MBYTE | |
| 1916 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0) | |
| 1917 return p_enc; | |
| 1918 #endif | |
| 1919 return (char_u *)"latin1"; | |
| 236 | 1920 } |
| 1921 | |
| 1922 /* | |
| 385 | 1923 * Get the name of the .spl file for the internal wordlist into |
| 1924 * "fname[MAXPATHL]". | |
| 1925 */ | |
| 1926 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1927 int_wordlist_spl(char_u *fname) |
| 385 | 1928 { |
| 2660 | 1929 vim_snprintf((char *)fname, MAXPATHL, SPL_FNAME_TMPL, |
| 385 | 1930 int_wordlist, spell_enc()); |
| 1931 } | |
| 1932 | |
| 1933 /* | |
| 625 | 1934 * Allocate a new slang_T for language "lang". "lang" can be NULL. |
| 236 | 1935 * Caller must fill "sl_next". |
| 1936 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1937 slang_T * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1938 slang_alloc(char_u *lang) |
| 236 | 1939 { |
| 1940 slang_T *lp; | |
| 1941 | |
| 300 | 1942 lp = (slang_T *)alloc_clear(sizeof(slang_T)); |
| 236 | 1943 if (lp != NULL) |
| 1944 { | |
| 625 | 1945 if (lang != NULL) |
| 1946 lp->sl_name = vim_strsave(lang); | |
| 323 | 1947 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10); |
| 625 | 1948 ga_init2(&lp->sl_repsal, sizeof(fromto_T), 10); |
| 483 | 1949 lp->sl_compmax = MAXWLEN; |
| 1950 lp->sl_compsylmax = MAXWLEN; | |
| 625 | 1951 hash_init(&lp->sl_wordcount); |
| 1952 } | |
| 1953 | |
| 236 | 1954 return lp; |
| 1955 } | |
| 1956 | |
| 1957 /* | |
| 1958 * Free the contents of an slang_T and the structure itself. | |
| 1959 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1960 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1961 slang_free(slang_T *lp) |
| 236 | 1962 { |
| 1963 vim_free(lp->sl_name); | |
| 310 | 1964 vim_free(lp->sl_fname); |
| 1965 slang_clear(lp); | |
| 1966 vim_free(lp); | |
| 1967 } | |
| 1968 | |
| 1969 /* | |
| 1970 * Clear an slang_T so that the file can be reloaded. | |
| 1971 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
1972 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
1973 slang_clear(slang_T *lp) |
| 310 | 1974 { |
| 339 | 1975 garray_T *gap; |
| 1976 fromto_T *ftp; | |
| 344 | 1977 salitem_T *smp; |
| 339 | 1978 int i; |
| 625 | 1979 int round; |
| 323 | 1980 |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1981 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
|
1982 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
|
1983 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
|
1984 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
1985 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
|
1986 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
|
1987 VIM_CLEAR(lp->sl_pidxs); |
| 323 | 1988 |
| 625 | 1989 for (round = 1; round <= 2; ++round) |
| 1990 { | |
| 1991 gap = round == 1 ? &lp->sl_rep : &lp->sl_repsal; | |
| 1992 while (gap->ga_len > 0) | |
| 1993 { | |
| 1994 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len]; | |
| 1995 vim_free(ftp->ft_from); | |
| 1996 vim_free(ftp->ft_to); | |
| 1997 } | |
| 1998 ga_clear(gap); | |
| 1999 } | |
| 344 | 2000 |
| 2001 gap = &lp->sl_sal; | |
| 375 | 2002 if (lp->sl_sofo) |
| 376 | 2003 { |
| 2004 /* "ga_len" is set to 1 without adding an item for latin1 */ | |
| 2005 if (gap->ga_data != NULL) | |
| 2006 /* SOFOFROM and SOFOTO items: free lists of wide characters. */ | |
| 2007 for (i = 0; i < gap->ga_len; ++i) | |
| 2008 vim_free(((int **)gap->ga_data)[i]); | |
| 2009 } | |
| 375 | 2010 else |
| 2011 /* SAL items: free salitem_T items */ | |
| 2012 while (gap->ga_len > 0) | |
| 2013 { | |
| 2014 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len]; | |
| 2015 vim_free(smp->sm_lead); | |
| 2016 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */ | |
| 2017 vim_free(smp->sm_to); | |
| 2018 #ifdef FEAT_MBYTE | |
| 2019 vim_free(smp->sm_lead_w); | |
| 2020 vim_free(smp->sm_oneof_w); | |
| 2021 vim_free(smp->sm_to_w); | |
| 2022 #endif | |
| 2023 } | |
| 344 | 2024 ga_clear(gap); |
| 323 | 2025 |
| 339 | 2026 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
|
2027 vim_regfree(lp->sl_prefprog[i]); |
| 376 | 2028 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
|
2029 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
|
2030 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2031 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
|
2032 |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2033 VIM_CLEAR(lp->sl_midword); |
| 339 | 2034 |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4352
diff
changeset
|
2035 vim_regfree(lp->sl_compprog); |
| 483 | 2036 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
|
2037 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
|
2038 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
|
2039 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
|
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_syllable); |
| 483 | 2042 ga_clear(&lp->sl_syl_items); |
| 481 | 2043 |
| 809 | 2044 ga_clear_strings(&lp->sl_comppat); |
| 2045 | |
| 625 | 2046 hash_clear_all(&lp->sl_wordcount, WC_KEY_OFF); |
| 2047 hash_init(&lp->sl_wordcount); | |
| 2048 | |
| 2049 #ifdef FEAT_MBYTE | |
| 2050 hash_clear_all(&lp->sl_map_hash, 0); | |
| 2051 #endif | |
| 2052 | |
| 2053 /* Clear info from .sug file. */ | |
| 2054 slang_clear_sug(lp); | |
| 483 | 2055 |
| 2056 lp->sl_compmax = MAXWLEN; | |
| 501 | 2057 lp->sl_compminlen = 0; |
| 483 | 2058 lp->sl_compsylmax = MAXWLEN; |
| 2059 lp->sl_regions[0] = NUL; | |
| 236 | 2060 } |
| 2061 | |
| 2062 /* | |
| 625 | 2063 * Clear the info from the .sug file in "lp". |
| 2064 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2065 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2066 slang_clear_sug(slang_T *lp) |
| 625 | 2067 { |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2068 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
|
2069 VIM_CLEAR(lp->sl_sidxs); |
| 625 | 2070 close_spellbuf(lp->sl_sugbuf); |
| 2071 lp->sl_sugbuf = NULL; | |
| 2072 lp->sl_sugloaded = FALSE; | |
| 2073 lp->sl_sugtime = 0; | |
| 2074 } | |
| 2075 | |
| 2076 /* | |
| 307 | 2077 * Load one spell file and store the info into a slang_T. |
| 236 | 2078 * Invoked through do_in_runtimepath(). |
| 2079 */ | |
| 2080 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2081 spell_load_cb(char_u *fname, void *cookie) |
| 501 | 2082 { |
| 2083 spelload_T *slp = (spelload_T *)cookie; | |
| 2084 slang_T *slang; | |
| 2085 | |
| 2086 slang = spell_load_file(fname, slp->sl_lang, NULL, FALSE); | |
| 2087 if (slang != NULL) | |
| 2088 { | |
| 2089 /* When a previously loaded file has NOBREAK also use it for the | |
| 2090 * ".add" files. */ | |
| 2091 if (slp->sl_nobreak && slang->sl_add) | |
| 2092 slang->sl_nobreak = TRUE; | |
| 2093 else if (slang->sl_nobreak) | |
| 2094 slp->sl_nobreak = TRUE; | |
| 2095 | |
| 2096 slp->sl_slang = slang; | |
| 2097 } | |
| 310 | 2098 } |
| 2099 | |
| 625 | 2100 |
| 2101 /* | |
| 2102 * Add a word to the hashtable of common words. | |
| 2103 * If it's already there then the counter is increased. | |
| 2104 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2105 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2106 count_common_word( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2107 slang_T *lp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2108 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2109 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
|
2110 int count) /* 1 to count once, 10 to init */ |
| 625 | 2111 { |
| 2112 hash_T hash; | |
| 2113 hashitem_T *hi; | |
| 2114 wordcount_T *wc; | |
| 2115 char_u buf[MAXWLEN]; | |
| 2116 char_u *p; | |
| 2117 | |
| 2118 if (len == -1) | |
| 2119 p = word; | |
| 2120 else | |
| 2121 { | |
| 2122 vim_strncpy(buf, word, len); | |
| 2123 p = buf; | |
| 2124 } | |
| 2125 | |
| 2126 hash = hash_hash(p); | |
| 2127 hi = hash_lookup(&lp->sl_wordcount, p, hash); | |
| 2128 if (HASHITEM_EMPTY(hi)) | |
| 2129 { | |
| 835 | 2130 wc = (wordcount_T *)alloc((unsigned)(sizeof(wordcount_T) + STRLEN(p))); |
| 625 | 2131 if (wc == NULL) |
| 2132 return; | |
| 2133 STRCPY(wc->wc_word, p); | |
| 2134 wc->wc_count = count; | |
| 2135 hash_add_item(&lp->sl_wordcount, hi, wc->wc_word, hash); | |
| 2136 } | |
| 2137 else | |
| 2138 { | |
| 2139 wc = HI2WC(hi); | |
| 2140 if ((wc->wc_count += count) < (unsigned)count) /* check for overflow */ | |
| 2141 wc->wc_count = MAXWORDCOUNT; | |
| 2142 } | |
| 2143 } | |
| 2144 | |
| 2145 /* | |
| 2146 * Adjust the score of common words. | |
| 2147 */ | |
| 2148 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2149 score_wordcount_adj( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2150 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2151 int score, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2152 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2153 int split) /* word was split, less bonus */ |
| 625 | 2154 { |
| 2155 hashitem_T *hi; | |
| 2156 wordcount_T *wc; | |
| 2157 int bonus; | |
| 2158 int newscore; | |
| 2159 | |
| 2160 hi = hash_find(&slang->sl_wordcount, word); | |
| 2161 if (!HASHITEM_EMPTY(hi)) | |
| 2162 { | |
| 2163 wc = HI2WC(hi); | |
| 2164 if (wc->wc_count < SCORE_THRES2) | |
| 2165 bonus = SCORE_COMMON1; | |
| 2166 else if (wc->wc_count < SCORE_THRES3) | |
| 2167 bonus = SCORE_COMMON2; | |
| 2168 else | |
| 2169 bonus = SCORE_COMMON3; | |
| 2170 if (split) | |
| 2171 newscore = score - bonus / 2; | |
| 2172 else | |
| 2173 newscore = score - bonus; | |
| 2174 if (newscore < 0) | |
| 2175 return 0; | |
| 2176 return newscore; | |
| 2177 } | |
| 2178 return score; | |
| 2179 } | |
| 2180 | |
| 483 | 2181 |
| 495 | 2182 /* |
| 498 | 2183 * Return TRUE if byte "n" appears in "str". |
| 495 | 2184 * Like strchr() but independent of locale. |
| 2185 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2186 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2187 byte_in_str(char_u *str, int n) |
| 495 | 2188 { |
| 2189 char_u *p; | |
| 2190 | |
| 2191 for (p = str; *p != NUL; ++p) | |
| 498 | 2192 if (*p == n) |
| 495 | 2193 return TRUE; |
| 2194 return FALSE; | |
| 2195 } | |
| 2196 | |
| 483 | 2197 #define SY_MAXLEN 30 |
| 2198 typedef struct syl_item_S | |
| 2199 { | |
| 2200 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */ | |
| 2201 int sy_len; | |
| 2202 } syl_item_T; | |
| 2203 | |
| 2204 /* | |
| 2205 * Truncate "slang->sl_syllable" at the first slash and put the following items | |
| 2206 * in "slang->sl_syl_items". | |
| 2207 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2208 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2209 init_syl_tab(slang_T *slang) |
| 483 | 2210 { |
| 2211 char_u *p; | |
| 2212 char_u *s; | |
| 2213 int l; | |
| 2214 syl_item_T *syl; | |
| 2215 | |
| 2216 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4); | |
| 2217 p = vim_strchr(slang->sl_syllable, '/'); | |
| 2218 while (p != NULL) | |
| 2219 { | |
| 2220 *p++ = NUL; | |
| 495 | 2221 if (*p == NUL) /* trailing slash */ |
| 483 | 2222 break; |
| 2223 s = p; | |
| 2224 p = vim_strchr(p, '/'); | |
| 2225 if (p == NULL) | |
| 835 | 2226 l = (int)STRLEN(s); |
| 483 | 2227 else |
| 835 | 2228 l = (int)(p - s); |
| 483 | 2229 if (l >= SY_MAXLEN) |
| 2230 return SP_FORMERROR; | |
| 2231 if (ga_grow(&slang->sl_syl_items, 1) == FAIL) | |
| 495 | 2232 return SP_OTHERERROR; |
| 483 | 2233 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) |
| 2234 + slang->sl_syl_items.ga_len++; | |
| 2235 vim_strncpy(syl->sy_chars, s, l); | |
| 2236 syl->sy_len = l; | |
| 2237 } | |
| 2238 return OK; | |
| 2239 } | |
| 2240 | |
| 2241 /* | |
| 2242 * Count the number of syllables in "word". | |
| 2243 * When "word" contains spaces the syllables after the last space are counted. | |
| 2244 * Returns zero if syllables are not defines. | |
| 2245 */ | |
| 2246 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2247 count_syllables(slang_T *slang, char_u *word) |
| 483 | 2248 { |
| 2249 int cnt = 0; | |
| 2250 int skip = FALSE; | |
| 2251 char_u *p; | |
| 2252 int len; | |
| 2253 int i; | |
| 2254 syl_item_T *syl; | |
| 2255 int c; | |
| 2256 | |
| 2257 if (slang->sl_syllable == NULL) | |
| 2258 return 0; | |
| 2259 | |
| 2260 for (p = word; *p != NUL; p += len) | |
| 2261 { | |
| 2262 /* When running into a space reset counter. */ | |
| 2263 if (*p == ' ') | |
| 2264 { | |
| 2265 len = 1; | |
| 2266 cnt = 0; | |
| 2267 continue; | |
| 2268 } | |
| 2269 | |
| 2270 /* Find longest match of syllable items. */ | |
| 2271 len = 0; | |
| 2272 for (i = 0; i < slang->sl_syl_items.ga_len; ++i) | |
| 2273 { | |
| 2274 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i; | |
| 2275 if (syl->sy_len > len | |
| 2276 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0) | |
| 2277 len = syl->sy_len; | |
| 2278 } | |
| 2279 if (len != 0) /* found a match, count syllable */ | |
| 2280 { | |
| 2281 ++cnt; | |
| 2282 skip = FALSE; | |
| 2283 } | |
| 2284 else | |
| 2285 { | |
| 2286 /* No recognized syllable item, at least a syllable char then? */ | |
| 2287 #ifdef FEAT_MBYTE | |
| 2288 c = mb_ptr2char(p); | |
| 2289 len = (*mb_ptr2len)(p); | |
| 2290 #else | |
| 2291 c = *p; | |
| 2292 len = 1; | |
| 2293 #endif | |
| 2294 if (vim_strchr(slang->sl_syllable, c) == NULL) | |
| 2295 skip = FALSE; /* No, search for next syllable */ | |
| 2296 else if (!skip) | |
| 2297 { | |
| 2298 ++cnt; /* Yes, count it */ | |
| 2299 skip = TRUE; /* don't count following syllable chars */ | |
| 2300 } | |
| 2301 } | |
| 2302 } | |
| 2303 return cnt; | |
| 376 | 2304 } |
| 2305 | |
| 381 | 2306 /* |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2307 * Parse 'spelllang' and set w_s->b_langp accordingly. |
| 351 | 2308 * Returns NULL if it's OK, an error message otherwise. |
| 236 | 2309 */ |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
2310 char * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2311 did_set_spelllang(win_T *wp) |
| 236 | 2312 { |
| 2313 garray_T ga; | |
| 351 | 2314 char_u *splp; |
| 236 | 2315 char_u *region; |
| 416 | 2316 char_u region_cp[3]; |
| 355 | 2317 int filename; |
| 236 | 2318 int region_mask; |
| 503 | 2319 slang_T *slang; |
| 236 | 2320 int c; |
| 351 | 2321 char_u lang[MAXWLEN + 1]; |
| 323 | 2322 char_u spf_name[MAXPATHL]; |
| 351 | 2323 int len; |
| 2324 char_u *p; | |
| 381 | 2325 int round; |
| 385 | 2326 char_u *spf; |
| 389 | 2327 char_u *use_region = NULL; |
| 2328 int dont_use_region = FALSE; | |
| 501 | 2329 int nobreak = FALSE; |
| 503 | 2330 int i, j; |
| 2331 langp_T *lp, *lp2; | |
| 1185 | 2332 static int recursive = FALSE; |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
2333 char *ret_msg = NULL; |
| 1185 | 2334 char_u *spl_copy; |
|
9487
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2335 bufref_T bufref; |
|
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2336 |
|
69ed2c9d34a6
commit https://github.com/vim/vim/commit/7c0a2f367f2507669560b1a66423155c70d2e75b
Christian Brabandt <cb@256bit.org>
parents:
9387
diff
changeset
|
2337 set_bufref(&bufref, wp->w_buffer); |
| 1185 | 2338 |
| 2339 /* We don't want to do this recursively. May happen when a language is | |
| 2340 * not available and the SpellFileMissing autocommand opens a new buffer | |
| 2341 * in which 'spell' is set. */ | |
| 2342 if (recursive) | |
| 2343 return NULL; | |
| 2344 recursive = TRUE; | |
| 236 | 2345 |
| 2346 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
|
2347 clear_midword(wp); |
| 236 | 2348 |
| 4352 | 2349 /* Make a copy of 'spelllang', the SpellFileMissing autocommands may change |
| 1185 | 2350 * it under our fingers. */ |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2351 spl_copy = vim_strsave(wp->w_s->b_p_spl); |
| 1185 | 2352 if (spl_copy == NULL) |
| 2353 goto theend; | |
| 2354 | |
| 5485 | 2355 #ifdef FEAT_MBYTE |
| 5477 | 2356 wp->w_s->b_cjk = 0; |
| 5485 | 2357 #endif |
| 5477 | 2358 |
| 2359 /* Loop over comma separated language names. */ | |
| 1185 | 2360 for (splp = spl_copy; *splp != NUL; ) |
| 351 | 2361 { |
| 2362 /* Get one language name. */ | |
| 2363 copy_option_part(&splp, lang, MAXWLEN, ","); | |
| 240 | 2364 region = NULL; |
| 835 | 2365 len = (int)STRLEN(lang); |
| 355 | 2366 |
| 5477 | 2367 if (STRCMP(lang, "cjk") == 0) |
| 2368 { | |
| 5485 | 2369 #ifdef FEAT_MBYTE |
| 5477 | 2370 wp->w_s->b_cjk = 1; |
| 5485 | 2371 #endif |
| 5477 | 2372 continue; |
| 2373 } | |
| 2374 | |
| 355 | 2375 /* If the name ends in ".spl" use it as the name of the spell file. |
| 2376 * If there is a region name let "region" point to it and remove it | |
| 2377 * from the name. */ | |
| 2378 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0) | |
| 236 | 2379 { |
| 355 | 2380 filename = TRUE; |
| 2381 | |
| 416 | 2382 /* Locate a region and remove it from the file name. */ |
| 2383 p = vim_strchr(gettail(lang), '_'); | |
| 2384 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2]) | |
| 2385 && !ASCII_ISALPHA(p[3])) | |
| 2386 { | |
| 2387 vim_strncpy(region_cp, p + 1, 2); | |
| 2388 mch_memmove(p, p + 3, len - (p - lang) - 2); | |
| 2389 region = region_cp; | |
| 2390 } | |
| 2391 else | |
| 2392 dont_use_region = TRUE; | |
| 2393 | |
| 355 | 2394 /* Check if we loaded this language before. */ |
| 503 | 2395 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2396 if (fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME) | |
| 355 | 2397 break; |
| 236 | 2398 } |
| 355 | 2399 else |
| 2400 { | |
| 2401 filename = FALSE; | |
| 2402 if (len > 3 && lang[len - 3] == '_') | |
| 2403 { | |
| 2404 region = lang + len - 2; | |
| 2405 len -= 3; | |
| 2406 lang[len] = NUL; | |
| 389 | 2407 } |
| 2408 else | |
| 2409 dont_use_region = TRUE; | |
| 355 | 2410 |
| 2411 /* Check if we loaded this language before. */ | |
| 503 | 2412 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2413 if (STRICMP(lang, slang->sl_name) == 0) | |
| 355 | 2414 break; |
| 2415 } | |
| 236 | 2416 |
| 416 | 2417 if (region != NULL) |
| 2418 { | |
| 2419 /* If the region differs from what was used before then don't | |
| 2420 * use it for 'spellfile'. */ | |
| 2421 if (use_region != NULL && STRCMP(region, use_region) != 0) | |
| 2422 dont_use_region = TRUE; | |
| 2423 use_region = region; | |
| 2424 } | |
| 2425 | |
| 351 | 2426 /* If not found try loading the language now. */ |
| 503 | 2427 if (slang == NULL) |
| 355 | 2428 { |
| 2429 if (filename) | |
| 2430 (void)spell_load_file(lang, lang, NULL, FALSE); | |
| 2431 else | |
| 1185 | 2432 { |
| 355 | 2433 spell_load_lang(lang); |
| 1185 | 2434 /* SpellFileMissing autocommands may do anything, including |
| 2435 * 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
|
2436 if (!bufref_valid(&bufref)) |
| 1185 | 2437 { |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
2438 ret_msg = N_("E797: SpellFileMissing autocommand deleted buffer"); |
| 1185 | 2439 goto theend; |
| 2440 } | |
| 2441 } | |
| 355 | 2442 } |
| 236 | 2443 |
| 307 | 2444 /* |
| 351 | 2445 * Loop over the languages, there can be several files for "lang". |
| 307 | 2446 */ |
| 503 | 2447 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2448 if (filename ? fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME | |
| 2449 : STRICMP(lang, slang->sl_name) == 0) | |
| 236 | 2450 { |
| 316 | 2451 region_mask = REGION_ALL; |
| 355 | 2452 if (!filename && region != NULL) |
| 236 | 2453 { |
| 307 | 2454 /* find region in sl_regions */ |
| 503 | 2455 c = find_region(slang->sl_regions, region); |
| 307 | 2456 if (c == REGION_ALL) |
| 2457 { | |
| 503 | 2458 if (slang->sl_add) |
| 389 | 2459 { |
| 503 | 2460 if (*slang->sl_regions != NUL) |
| 389 | 2461 /* This addition file is for other regions. */ |
| 2462 region_mask = 0; | |
| 2463 } | |
| 2464 else | |
| 2465 /* This is probably an error. Give a warning and | |
| 2466 * accept the words anyway. */ | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
2467 smsg(_("Warning: region %s not supported"), |
| 351 | 2468 region); |
| 307 | 2469 } |
| 2470 else | |
| 2471 region_mask = 1 << c; | |
| 236 | 2472 } |
| 307 | 2473 |
| 389 | 2474 if (region_mask != 0) |
| 307 | 2475 { |
| 389 | 2476 if (ga_grow(&ga, 1) == FAIL) |
| 2477 { | |
| 2478 ga_clear(&ga); | |
| 1185 | 2479 ret_msg = e_outofmem; |
| 2480 goto theend; | |
| 389 | 2481 } |
| 503 | 2482 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 389 | 2483 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 2484 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2485 use_midword(slang, wp); |
| 503 | 2486 if (slang->sl_nobreak) |
| 501 | 2487 nobreak = TRUE; |
| 307 | 2488 } |
| 385 | 2489 } |
| 2490 } | |
| 2491 | |
| 2492 /* round 0: load int_wordlist, if possible. | |
| 2493 * round 1: load first name in 'spellfile'. | |
| 2494 * round 2: load second name in 'spellfile. | |
| 2495 * etc. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2496 spf = curwin->w_s->b_p_spf; |
| 385 | 2497 for (round = 0; round == 0 || *spf != NUL; ++round) |
| 2498 { | |
| 2499 if (round == 0) | |
| 2500 { | |
| 2501 /* Internal wordlist, if there is one. */ | |
| 2502 if (int_wordlist == NULL) | |
| 381 | 2503 continue; |
| 385 | 2504 int_wordlist_spl(spf_name); |
| 381 | 2505 } |
| 2506 else | |
| 2507 { | |
| 385 | 2508 /* One entry in 'spellfile'. */ |
| 2509 copy_option_part(&spf, spf_name, MAXPATHL - 5, ","); | |
| 2510 STRCAT(spf_name, ".spl"); | |
| 2511 | |
| 2512 /* If it was already found above then skip it. */ | |
| 2513 for (c = 0; c < ga.ga_len; ++c) | |
| 500 | 2514 { |
| 2515 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname; | |
| 2516 if (p != NULL && fullpathcmp(spf_name, p, FALSE) == FPC_SAME) | |
| 385 | 2517 break; |
| 500 | 2518 } |
| 385 | 2519 if (c < ga.ga_len) |
| 381 | 2520 continue; |
| 2521 } | |
| 2522 | |
| 351 | 2523 /* Check if it was loaded already. */ |
| 503 | 2524 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2525 if (fullpathcmp(spf_name, slang->sl_fname, FALSE) == FPC_SAME) | |
| 323 | 2526 break; |
| 503 | 2527 if (slang == NULL) |
| 323 | 2528 { |
| 351 | 2529 /* Not loaded, try loading it now. The language name includes the |
| 385 | 2530 * region name, the region is ignored otherwise. for int_wordlist |
| 2531 * use an arbitrary name. */ | |
| 2532 if (round == 0) | |
| 2533 STRCPY(lang, "internal wordlist"); | |
| 2534 else | |
| 2535 { | |
| 2536 vim_strncpy(lang, gettail(spf_name), MAXWLEN); | |
| 381 | 2537 p = vim_strchr(lang, '.'); |
| 2538 if (p != NULL) | |
| 2539 *p = NUL; /* truncate at ".encoding.add" */ | |
| 2540 } | |
| 503 | 2541 slang = spell_load_file(spf_name, lang, NULL, TRUE); |
| 501 | 2542 |
| 2543 /* If one of the languages has NOBREAK we assume the addition | |
| 2544 * files also have this. */ | |
| 503 | 2545 if (slang != NULL && nobreak) |
| 2546 slang->sl_nobreak = TRUE; | |
| 2547 } | |
| 2548 if (slang != NULL && ga_grow(&ga, 1) == OK) | |
| 323 | 2549 { |
| 389 | 2550 region_mask = REGION_ALL; |
| 2551 if (use_region != NULL && !dont_use_region) | |
| 2552 { | |
| 2553 /* find region in sl_regions */ | |
| 503 | 2554 c = find_region(slang->sl_regions, use_region); |
| 389 | 2555 if (c != REGION_ALL) |
| 2556 region_mask = 1 << c; | |
| 503 | 2557 else if (*slang->sl_regions != NUL) |
| 389 | 2558 /* This spell file is for other regions. */ |
| 2559 region_mask = 0; | |
| 2560 } | |
| 2561 | |
| 2562 if (region_mask != 0) | |
| 2563 { | |
| 503 | 2564 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 2565 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL; | |
| 2566 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL; | |
| 389 | 2567 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 2568 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2569 use_midword(slang, wp); |
| 389 | 2570 } |
| 323 | 2571 } |
| 2572 } | |
| 2573 | |
| 236 | 2574 /* 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
|
2575 ga_clear(&wp->w_s->b_langp); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2576 wp->w_s->b_langp = ga; |
| 236 | 2577 |
| 503 | 2578 /* For each language figure out what language to use for sound folding and |
| 2579 * REP items. If the language doesn't support it itself use another one | |
| 2580 * with the same name. E.g. for "en-math" use "en". */ | |
| 2581 for (i = 0; i < ga.ga_len; ++i) | |
| 2582 { | |
| 2583 lp = LANGP_ENTRY(ga, i); | |
| 2584 | |
| 2585 /* sound folding */ | |
| 2586 if (lp->lp_slang->sl_sal.ga_len > 0) | |
| 2587 /* language does sound folding itself */ | |
| 2588 lp->lp_sallang = lp->lp_slang; | |
| 2589 else | |
| 2590 /* find first similar language that does sound folding */ | |
| 2591 for (j = 0; j < ga.ga_len; ++j) | |
| 2592 { | |
| 2593 lp2 = LANGP_ENTRY(ga, j); | |
| 2594 if (lp2->lp_slang->sl_sal.ga_len > 0 | |
| 2595 && STRNCMP(lp->lp_slang->sl_name, | |
| 2596 lp2->lp_slang->sl_name, 2) == 0) | |
| 2597 { | |
| 2598 lp->lp_sallang = lp2->lp_slang; | |
| 2599 break; | |
| 2600 } | |
| 2601 } | |
| 2602 | |
| 2603 /* REP items */ | |
| 2604 if (lp->lp_slang->sl_rep.ga_len > 0) | |
| 2605 /* language has REP items itself */ | |
| 2606 lp->lp_replang = lp->lp_slang; | |
| 2607 else | |
| 625 | 2608 /* find first similar language that has REP items */ |
| 503 | 2609 for (j = 0; j < ga.ga_len; ++j) |
| 2610 { | |
| 2611 lp2 = LANGP_ENTRY(ga, j); | |
| 2612 if (lp2->lp_slang->sl_rep.ga_len > 0 | |
| 2613 && STRNCMP(lp->lp_slang->sl_name, | |
| 2614 lp2->lp_slang->sl_name, 2) == 0) | |
| 2615 { | |
| 2616 lp->lp_replang = lp2->lp_slang; | |
| 2617 break; | |
| 2618 } | |
| 2619 } | |
| 2620 } | |
| 2621 | |
| 1185 | 2622 theend: |
| 2623 vim_free(spl_copy); | |
| 2624 recursive = FALSE; | |
| 5891 | 2625 redraw_win_later(wp, NOT_VALID); |
| 1185 | 2626 return ret_msg; |
| 236 | 2627 } |
| 2628 | |
| 2629 /* | |
| 376 | 2630 * Clear the midword characters for buffer "buf". |
| 2631 */ | |
| 2632 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2633 clear_midword(win_T *wp) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2634 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2635 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
|
2636 #ifdef FEAT_MBYTE |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2637 VIM_CLEAR(wp->w_s->b_spell_ismw_mb); |
| 376 | 2638 #endif |
| 2639 } | |
| 2640 | |
| 2641 /* | |
| 2642 * Use the "sl_midword" field of language "lp" for buffer "buf". | |
| 2643 * They add up to any currently used midword characters. | |
| 2644 */ | |
| 2645 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2646 use_midword(slang_T *lp, win_T *wp) |
| 376 | 2647 { |
| 2648 char_u *p; | |
| 2649 | |
| 389 | 2650 if (lp->sl_midword == NULL) /* there aren't any */ |
| 2651 return; | |
| 2652 | |
| 376 | 2653 for (p = lp->sl_midword; *p != NUL; ) |
| 2654 #ifdef FEAT_MBYTE | |
| 2655 if (has_mbyte) | |
| 2656 { | |
| 2657 int c, l, n; | |
| 2658 char_u *bp; | |
| 2659 | |
| 2660 c = mb_ptr2char(p); | |
| 474 | 2661 l = (*mb_ptr2len)(p); |
| 2662 if (c < 256 && l <= 2) | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2663 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
|
2664 else if (wp->w_s->b_spell_ismw_mb == NULL) |
| 376 | 2665 /* 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
|
2666 wp->w_s->b_spell_ismw_mb = vim_strnsave(p, l); |
| 376 | 2667 else |
| 2668 { | |
| 2669 /* 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
|
2670 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
|
2671 bp = vim_strnsave(wp->w_s->b_spell_ismw_mb, n + l); |
| 376 | 2672 if (bp != NULL) |
| 2673 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2674 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
|
2675 wp->w_s->b_spell_ismw_mb = bp; |
| 376 | 2676 vim_strncpy(bp + n, p, l); |
| 2677 } | |
| 2678 } | |
| 2679 p += l; | |
| 2680 } | |
| 2681 else | |
| 2682 #endif | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2683 wp->w_s->b_spell_ismw[*p++] = TRUE; |
| 376 | 2684 } |
| 2685 | |
| 2686 /* | |
| 236 | 2687 * Find the region "region[2]" in "rp" (points to "sl_regions"). |
|
15034
6e4e0d43b20b
patch 8.1.0528: various typos in comments
Bram Moolenaar <Bram@vim.org>
parents:
14862
diff
changeset
|
2688 * Each region is simply stored as the two characters of its name. |
| 381 | 2689 * Returns the index if found (first is 0), REGION_ALL if not found. |
| 236 | 2690 */ |
| 2691 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2692 find_region(char_u *rp, char_u *region) |
| 236 | 2693 { |
| 2694 int i; | |
| 2695 | |
| 2696 for (i = 0; ; i += 2) | |
| 2697 { | |
| 2698 if (rp[i] == NUL) | |
| 2699 return REGION_ALL; | |
| 2700 if (rp[i] == region[0] && rp[i + 1] == region[1]) | |
| 2701 break; | |
| 2702 } | |
| 2703 return i / 2; | |
| 2704 } | |
| 2705 | |
| 2706 /* | |
| 323 | 2707 * Return case type of word: |
| 236 | 2708 * w word 0 |
| 300 | 2709 * Word WF_ONECAP |
| 2710 * W WORD WF_ALLCAP | |
| 2711 * WoRd wOrd WF_KEEPCAP | |
| 236 | 2712 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2713 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2714 captype( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2715 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2716 char_u *end) /* When NULL use up to NUL byte. */ |
| 236 | 2717 { |
| 2718 char_u *p; | |
| 2719 int c; | |
| 2720 int firstcap; | |
| 2721 int allcap; | |
| 2722 int past_second = FALSE; /* past second word char */ | |
| 2723 | |
| 2724 /* find first letter */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2725 for (p = word; !spell_iswordp_nmw(p, curwin); MB_PTR_ADV(p)) |
| 323 | 2726 if (end == NULL ? *p == NUL : p >= end) |
| 236 | 2727 return 0; /* only non-word characters, illegal word */ |
| 2728 #ifdef FEAT_MBYTE | |
| 310 | 2729 if (has_mbyte) |
| 2730 c = mb_ptr2char_adv(&p); | |
| 2731 else | |
| 236 | 2732 #endif |
| 310 | 2733 c = *p++; |
| 324 | 2734 firstcap = allcap = SPELL_ISUPPER(c); |
| 236 | 2735 |
| 2736 /* | |
| 2737 * Need to check all letters to find a word with mixed upper/lower. | |
| 2738 * But a word with an upper char only at start is a ONECAP. | |
| 2739 */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2740 for ( ; end == NULL ? *p != NUL : p < end; MB_PTR_ADV(p)) |
| 5477 | 2741 if (spell_iswordp_nmw(p, curwin)) |
| 236 | 2742 { |
| 455 | 2743 c = PTR2CHAR(p); |
| 324 | 2744 if (!SPELL_ISUPPER(c)) |
| 236 | 2745 { |
| 2746 /* UUl -> KEEPCAP */ | |
| 2747 if (past_second && allcap) | |
| 300 | 2748 return WF_KEEPCAP; |
| 236 | 2749 allcap = FALSE; |
| 2750 } | |
| 2751 else if (!allcap) | |
| 2752 /* UlU -> KEEPCAP */ | |
| 300 | 2753 return WF_KEEPCAP; |
| 236 | 2754 past_second = TRUE; |
| 2755 } | |
| 2756 | |
| 2757 if (allcap) | |
| 300 | 2758 return WF_ALLCAP; |
| 236 | 2759 if (firstcap) |
| 300 | 2760 return WF_ONECAP; |
| 236 | 2761 return 0; |
| 2762 } | |
| 2763 | |
| 474 | 2764 /* |
| 2765 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a | |
| 2766 * capital. So that make_case_word() can turn WOrd into Word. | |
| 2767 * Add ALLCAP for "WOrD". | |
| 2768 */ | |
| 2769 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2770 badword_captype(char_u *word, char_u *end) |
| 474 | 2771 { |
| 2772 int flags = captype(word, end); | |
| 475 | 2773 int c; |
| 474 | 2774 int l, u; |
| 2775 int first; | |
| 2776 char_u *p; | |
| 2777 | |
| 2778 if (flags & WF_KEEPCAP) | |
| 2779 { | |
| 2780 /* Count the number of UPPER and lower case letters. */ | |
| 2781 l = u = 0; | |
| 2782 first = FALSE; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2783 for (p = word; p < end; MB_PTR_ADV(p)) |
| 474 | 2784 { |
| 475 | 2785 c = PTR2CHAR(p); |
| 2786 if (SPELL_ISUPPER(c)) | |
| 474 | 2787 { |
| 2788 ++u; | |
| 2789 if (p == word) | |
| 2790 first = TRUE; | |
| 2791 } | |
| 2792 else | |
| 2793 ++l; | |
| 2794 } | |
| 2795 | |
| 2796 /* If there are more UPPER than lower case letters suggest an | |
| 2797 * ALLCAP word. Otherwise, if the first letter is UPPER then | |
| 2798 * suggest ONECAP. Exception: "ALl" most likely should be "All", | |
| 2799 * require three upper case letters. */ | |
| 2800 if (u > l && u > 2) | |
| 2801 flags |= WF_ALLCAP; | |
| 2802 else if (first) | |
| 2803 flags |= WF_ONECAP; | |
| 633 | 2804 |
| 2805 if (u >= 2 && l >= 2) /* maCARONI maCAroni */ | |
| 2806 flags |= WF_MIXCAP; | |
| 474 | 2807 } |
| 2808 return flags; | |
| 2809 } | |
| 2810 | |
| 5519 | 2811 /* |
| 2812 * Delete the internal wordlist and its .spl file. | |
| 2813 */ | |
| 2814 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2815 spell_delete_wordlist(void) |
| 5519 | 2816 { |
| 2817 char_u fname[MAXPATHL]; | |
| 2818 | |
| 2819 if (int_wordlist != NULL) | |
| 2820 { | |
| 2821 mch_remove(int_wordlist); | |
| 2822 int_wordlist_spl(fname); | |
| 2823 mch_remove(fname); | |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2824 VIM_CLEAR(int_wordlist); |
| 5519 | 2825 } |
| 2826 } | |
| 2827 | |
| 2828 #if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO) | |
| 355 | 2829 /* |
| 2830 * Free all languages. | |
| 2831 */ | |
| 2832 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2833 spell_free_all(void) |
| 355 | 2834 { |
| 503 | 2835 slang_T *slang; |
| 355 | 2836 buf_T *buf; |
| 2837 | |
| 2580 | 2838 /* 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
|
2839 FOR_ALL_BUFFERS(buf) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2840 ga_clear(&buf->b_s.b_langp); |
| 355 | 2841 |
| 2842 while (first_lang != NULL) | |
| 2843 { | |
| 503 | 2844 slang = first_lang; |
| 2845 first_lang = slang->sl_next; | |
| 2846 slang_free(slang); | |
| 355 | 2847 } |
| 366 | 2848 |
| 5519 | 2849 spell_delete_wordlist(); |
| 381 | 2850 |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2851 VIM_CLEAR(repl_to); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2852 VIM_CLEAR(repl_from); |
| 355 | 2853 } |
| 5519 | 2854 #endif |
| 2855 | |
| 2856 #if defined(FEAT_MBYTE) || defined(PROTO) | |
| 236 | 2857 /* |
| 2858 * Clear all spelling tables and reload them. | |
| 307 | 2859 * Used after 'encoding' is set and when ":mkspell" was used. |
| 236 | 2860 */ |
| 2861 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2862 spell_reload(void) |
| 236 | 2863 { |
| 316 | 2864 win_T *wp; |
| 236 | 2865 |
| 358 | 2866 /* Initialize the table for spell_iswordp(). */ |
| 236 | 2867 init_spell_chartab(); |
| 2868 | |
| 2869 /* Unload all allocated memory. */ | |
| 355 | 2870 spell_free_all(); |
| 236 | 2871 |
| 2872 /* 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
|
2873 FOR_ALL_WINDOWS(wp) |
| 236 | 2874 { |
| 316 | 2875 /* Only load the wordlists when 'spelllang' is set and there is a |
| 2876 * 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
|
2877 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
|
2878 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2879 if (wp->w_p_spell) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2880 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2881 (void)did_set_spelllang(wp); |
| 316 | 2882 break; |
| 2883 } | |
| 2884 } | |
| 236 | 2885 } |
| 2886 } | |
| 5519 | 2887 #endif |
| 236 | 2888 |
| 310 | 2889 /* |
| 625 | 2890 * Opposite of offset2bytes(). |
| 2891 * "pp" points to the bytes and is advanced over it. | |
| 2892 * Returns the offset. | |
| 2893 */ | |
| 2894 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2895 bytes2offset(char_u **pp) |
| 625 | 2896 { |
| 2897 char_u *p = *pp; | |
| 2898 int nr; | |
| 2899 int c; | |
| 2900 | |
| 2901 c = *p++; | |
| 2902 if ((c & 0x80) == 0x00) /* 1 byte */ | |
| 2903 { | |
| 2904 nr = c - 1; | |
| 2905 } | |
| 2906 else if ((c & 0xc0) == 0x80) /* 2 bytes */ | |
| 2907 { | |
| 2908 nr = (c & 0x3f) - 1; | |
| 2909 nr = nr * 255 + (*p++ - 1); | |
| 2910 } | |
| 2911 else if ((c & 0xe0) == 0xc0) /* 3 bytes */ | |
| 2912 { | |
| 2913 nr = (c & 0x1f) - 1; | |
| 2914 nr = nr * 255 + (*p++ - 1); | |
| 2915 nr = nr * 255 + (*p++ - 1); | |
| 2916 } | |
| 2917 else /* 4 bytes */ | |
| 2918 { | |
| 2919 nr = (c & 0x0f) - 1; | |
| 2920 nr = nr * 255 + (*p++ - 1); | |
| 2921 nr = nr * 255 + (*p++ - 1); | |
| 2922 nr = nr * 255 + (*p++ - 1); | |
| 2923 } | |
| 2924 | |
| 2925 *pp = p; | |
| 2926 return nr; | |
| 2927 } | |
| 2928 | |
| 2929 | |
| 2930 /* | |
| 2931 * Open a spell buffer. This is a nameless buffer that is not in the buffer | |
| 2932 * list and only contains text lines. Can use a swapfile to reduce memory | |
| 2933 * use. | |
| 2934 * Most other fields are invalid! Esp. watch out for string options being | |
| 2935 * NULL and there is no undo info. | |
| 2936 * Returns NULL when out of memory. | |
| 2937 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2938 buf_T * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2939 open_spellbuf(void) |
| 625 | 2940 { |
| 2941 buf_T *buf; | |
| 2942 | |
| 2943 buf = (buf_T *)alloc_clear(sizeof(buf_T)); | |
| 2944 if (buf != NULL) | |
| 2945 { | |
| 2946 buf->b_spell = TRUE; | |
| 2947 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
|
2948 #ifdef FEAT_CRYPT |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
2949 buf->b_p_key = empty_option; |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
2950 #endif |
| 625 | 2951 ml_open(buf); |
| 2952 ml_open_file(buf); /* create swap file now */ | |
| 2953 } | |
| 2954 return buf; | |
| 2955 } | |
| 2956 | |
| 2957 /* | |
| 2958 * Close the buffer used for spell info. | |
| 2959 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2960 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2961 close_spellbuf(buf_T *buf) |
| 625 | 2962 { |
| 2963 if (buf != NULL) | |
| 2964 { | |
| 2965 ml_close(buf, TRUE); | |
| 2966 vim_free(buf); | |
| 2967 } | |
| 2968 } | |
| 2969 | |
| 307 | 2970 /* |
| 2971 * Init the chartab used for spelling for ASCII. | |
| 2972 * EBCDIC is not supported! | |
| 2973 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2974 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2975 clear_spell_chartab(spelltab_T *sp) |
| 307 | 2976 { |
| 324 | 2977 int i; |
| 307 | 2978 |
| 2979 /* Init everything to FALSE. */ | |
| 2980 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw)); | |
| 2981 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu)); | |
| 2982 for (i = 0; i < 256; ++i) | |
| 324 | 2983 { |
| 307 | 2984 sp->st_fold[i] = i; |
| 324 | 2985 sp->st_upper[i] = i; |
| 2986 } | |
| 307 | 2987 |
| 2988 /* We include digits. A word shouldn't start with a digit, but handling | |
| 2989 * that is done separately. */ | |
| 2990 for (i = '0'; i <= '9'; ++i) | |
| 2991 sp->st_isw[i] = TRUE; | |
| 2992 for (i = 'A'; i <= 'Z'; ++i) | |
| 2993 { | |
| 2994 sp->st_isw[i] = TRUE; | |
| 2995 sp->st_isu[i] = TRUE; | |
| 2996 sp->st_fold[i] = i + 0x20; | |
| 2997 } | |
| 2998 for (i = 'a'; i <= 'z'; ++i) | |
| 324 | 2999 { |
| 307 | 3000 sp->st_isw[i] = TRUE; |
| 324 | 3001 sp->st_upper[i] = i - 0x20; |
| 3002 } | |
| 307 | 3003 } |
| 3004 | |
| 3005 /* | |
| 3006 * Init the chartab used for spelling. Only depends on 'encoding'. | |
| 3007 * Called once while starting up and when 'encoding' changes. | |
| 3008 * The default is to use isalpha(), but the spell file should define the word | |
| 3009 * characters to make it possible that 'encoding' differs from the current | |
| 390 | 3010 * locale. For utf-8 we don't use isalpha() but our own functions. |
| 307 | 3011 */ |
| 3012 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3013 init_spell_chartab(void) |
| 307 | 3014 { |
| 3015 int i; | |
| 3016 | |
| 3017 did_set_spelltab = FALSE; | |
| 3018 clear_spell_chartab(&spelltab); | |
| 3019 #ifdef FEAT_MBYTE | |
| 3020 if (enc_dbcs) | |
| 3021 { | |
| 3022 /* DBCS: assume double-wide characters are word characters. */ | |
| 3023 for (i = 128; i <= 255; ++i) | |
| 3024 if (MB_BYTE2LEN(i) == 2) | |
| 3025 spelltab.st_isw[i] = TRUE; | |
| 3026 } | |
| 324 | 3027 else if (enc_utf8) |
| 3028 { | |
| 3029 for (i = 128; i < 256; ++i) | |
| 3030 { | |
|
2140
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3031 int f = utf_fold(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3032 int u = utf_toupper(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3033 |
| 324 | 3034 spelltab.st_isu[i] = utf_isupper(i); |
| 3035 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
|
3036 /* 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
|
3037 * 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
|
3038 * value for utf-8 to avoid this. */ |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3039 spelltab.st_fold[i] = (f < 256) ? f : i; |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3040 spelltab.st_upper[i] = (u < 256) ? u : i; |
| 324 | 3041 } |
| 3042 } | |
| 307 | 3043 else |
| 3044 #endif | |
| 3045 { | |
| 324 | 3046 /* Rough guess: use locale-dependent library functions. */ |
| 307 | 3047 for (i = 128; i < 256; ++i) |
| 3048 { | |
| 3049 if (MB_ISUPPER(i)) | |
| 3050 { | |
| 324 | 3051 spelltab.st_isw[i] = TRUE; |
| 307 | 3052 spelltab.st_isu[i] = TRUE; |
| 3053 spelltab.st_fold[i] = MB_TOLOWER(i); | |
| 3054 } | |
| 324 | 3055 else if (MB_ISLOWER(i)) |
| 3056 { | |
| 3057 spelltab.st_isw[i] = TRUE; | |
| 3058 spelltab.st_upper[i] = MB_TOUPPER(i); | |
| 3059 } | |
| 307 | 3060 } |
| 3061 } | |
| 3062 } | |
| 3063 | |
| 3064 | |
| 3065 /* | |
| 358 | 3066 * Return TRUE if "p" points to a word character. |
| 366 | 3067 * As a special case we see "midword" characters as word character when it is |
| 358 | 3068 * followed by a word character. This finds they'there but not 'they there'. |
| 366 | 3069 * Thus this only works properly when past the first character of the word. |
| 358 | 3070 */ |
| 3071 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3072 spell_iswordp( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3073 char_u *p, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3074 win_T *wp) /* buffer used */ |
| 358 | 3075 { |
| 366 | 3076 #ifdef FEAT_MBYTE |
| 358 | 3077 char_u *s; |
| 366 | 3078 int l; |
| 3079 int c; | |
| 3080 | |
| 3081 if (has_mbyte) | |
| 3082 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
3083 l = MB_PTR2LEN(p); |
| 358 | 3084 s = p; |
| 366 | 3085 if (l == 1) |
| 3086 { | |
| 3087 /* be quick for ASCII */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3088 if (wp->w_s->b_spell_ismw[*p]) |
| 366 | 3089 s = p + 1; /* skip a mid-word character */ |
| 3090 } | |
| 3091 else | |
| 3092 { | |
| 3093 c = mb_ptr2char(p); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3094 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
|
3095 : (wp->w_s->b_spell_ismw_mb != NULL |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3096 && vim_strchr(wp->w_s->b_spell_ismw_mb, c) != NULL)) |
| 366 | 3097 s = p + l; |
| 3098 } | |
| 3099 | |
| 390 | 3100 c = mb_ptr2char(s); |
| 3101 if (c > 255) | |
| 5477 | 3102 return spell_mb_isword_class(mb_get_class(s), wp); |
| 390 | 3103 return spelltab.st_isw[c]; |
| 366 | 3104 } |
| 358 | 3105 #endif |
| 366 | 3106 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3107 return spelltab.st_isw[wp->w_s->b_spell_ismw[*p] ? p[1] : p[0]]; |
| 376 | 3108 } |
| 3109 | |
| 3110 /* | |
| 3111 * Return TRUE if "p" points to a word character. | |
| 3112 * Unlike spell_iswordp() this doesn't check for "midword" characters. | |
| 3113 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3114 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3115 spell_iswordp_nmw(char_u *p, win_T *wp) |
| 376 | 3116 { |
| 3117 #ifdef FEAT_MBYTE | |
| 390 | 3118 int c; |
| 3119 | |
| 3120 if (has_mbyte) | |
| 3121 { | |
| 3122 c = mb_ptr2char(p); | |
| 3123 if (c > 255) | |
| 5477 | 3124 return spell_mb_isword_class(mb_get_class(p), wp); |
| 390 | 3125 return spelltab.st_isw[c]; |
| 3126 } | |
| 376 | 3127 #endif |
| 3128 return spelltab.st_isw[*p]; | |
| 358 | 3129 } |
| 3130 | |
| 372 | 3131 #ifdef FEAT_MBYTE |
| 3132 /* | |
| 1580 | 3133 * Return TRUE if word class indicates a word character. |
| 3134 * Only for characters above 255. | |
| 3135 * Unicode subscript and superscript are not considered word characters. | |
| 5477 | 3136 * See also dbcs_class() and utf_class() in mbyte.c. |
| 3137 */ | |
| 3138 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3139 spell_mb_isword_class(int cl, win_T *wp) |
| 5477 | 3140 { |
| 3141 if (wp->w_s->b_cjk) | |
| 3142 /* East Asian characters are not considered word characters. */ | |
| 3143 return cl == 2 || cl == 0x2800; | |
| 1580 | 3144 return cl >= 2 && cl != 0x2070 && cl != 0x2080; |
| 3145 } | |
| 3146 | |
| 3147 /* | |
| 372 | 3148 * Return TRUE if "p" points to a word character. |
| 3149 * Wide version of spell_iswordp(). | |
| 3150 */ | |
| 3151 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3152 spell_iswordp_w(int *p, win_T *wp) |
| 372 | 3153 { |
| 3154 int *s; | |
| 3155 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3156 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
|
3157 : (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
|
3158 && vim_strchr(wp->w_s->b_spell_ismw_mb, *p) != NULL)) |
| 372 | 3159 s = p + 1; |
| 3160 else | |
| 3161 s = p; | |
| 3162 | |
| 390 | 3163 if (*s > 255) |
| 372 | 3164 { |
| 3165 if (enc_utf8) | |
| 5477 | 3166 return spell_mb_isword_class(utf_class(*s), wp); |
| 372 | 3167 if (enc_dbcs) |
| 5477 | 3168 return spell_mb_isword_class( |
| 3169 dbcs_class((unsigned)*s >> 8, *s & 0xff), wp); | |
| 372 | 3170 return 0; |
| 3171 } | |
| 3172 return spelltab.st_isw[*s]; | |
| 3173 } | |
| 3174 #endif | |
| 3175 | |
| 358 | 3176 /* |
| 324 | 3177 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated. |
| 3178 * Uses the character definitions from the .spl file. | |
| 307 | 3179 * When using a multi-byte 'encoding' the length may change! |
| 3180 * Returns FAIL when something wrong. | |
| 3181 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3182 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3183 spell_casefold( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3184 char_u *str, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3185 int len, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3186 char_u *buf, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3187 int buflen) |
| 307 | 3188 { |
| 3189 int i; | |
| 3190 | |
| 3191 if (len >= buflen) | |
| 3192 { | |
| 3193 buf[0] = NUL; | |
| 3194 return FAIL; /* result will not fit */ | |
| 3195 } | |
| 3196 | |
| 3197 #ifdef FEAT_MBYTE | |
| 3198 if (has_mbyte) | |
| 3199 { | |
| 324 | 3200 int outi = 0; |
| 3201 char_u *p; | |
| 307 | 3202 int c; |
| 3203 | |
| 3204 /* Fold one character at a time. */ | |
| 324 | 3205 for (p = str; p < str + len; ) |
| 307 | 3206 { |
| 3207 if (outi + MB_MAXBYTES > buflen) | |
| 3208 { | |
| 3209 buf[outi] = NUL; | |
| 3210 return FAIL; | |
| 3211 } | |
| 474 | 3212 c = mb_cptr2char_adv(&p); |
| 324 | 3213 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi); |
| 307 | 3214 } |
| 3215 buf[outi] = NUL; | |
| 3216 } | |
| 3217 else | |
| 3218 #endif | |
| 3219 { | |
| 3220 /* Be quick for non-multibyte encodings. */ | |
| 3221 for (i = 0; i < len; ++i) | |
| 324 | 3222 buf[i] = spelltab.st_fold[str[i]]; |
| 307 | 3223 buf[i] = NUL; |
| 3224 } | |
| 3225 | |
| 3226 return OK; | |
| 3227 } | |
| 3228 | |
| 625 | 3229 /* values for sps_flags */ |
| 372 | 3230 #define SPS_BEST 1 |
| 3231 #define SPS_FAST 2 | |
| 3232 #define SPS_DOUBLE 4 | |
| 3233 | |
| 625 | 3234 static int sps_flags = SPS_BEST; /* flags from 'spellsuggest' */ |
| 3235 static int sps_limit = 9999; /* max nr of suggestions given */ | |
| 372 | 3236 |
| 3237 /* | |
| 3238 * Check the 'spellsuggest' option. Return FAIL if it's wrong. | |
| 483 | 3239 * Sets "sps_flags" and "sps_limit". |
| 372 | 3240 */ |
| 3241 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3242 spell_check_sps(void) |
| 372 | 3243 { |
| 3244 char_u *p; | |
| 483 | 3245 char_u *s; |
| 372 | 3246 char_u buf[MAXPATHL]; |
| 3247 int f; | |
| 3248 | |
| 3249 sps_flags = 0; | |
| 483 | 3250 sps_limit = 9999; |
| 372 | 3251 |
| 3252 for (p = p_sps; *p != NUL; ) | |
| 3253 { | |
| 3254 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 3255 | |
| 3256 f = 0; | |
| 483 | 3257 if (VIM_ISDIGIT(*buf)) |
| 3258 { | |
| 3259 s = buf; | |
| 3260 sps_limit = getdigits(&s); | |
| 3261 if (*s != NUL && !VIM_ISDIGIT(*s)) | |
| 3262 f = -1; | |
| 3263 } | |
| 3264 else if (STRCMP(buf, "best") == 0) | |
| 372 | 3265 f = SPS_BEST; |
| 3266 else if (STRCMP(buf, "fast") == 0) | |
| 3267 f = SPS_FAST; | |
| 3268 else if (STRCMP(buf, "double") == 0) | |
| 3269 f = SPS_DOUBLE; | |
| 3270 else if (STRNCMP(buf, "expr:", 5) != 0 | |
| 3271 && STRNCMP(buf, "file:", 5) != 0) | |
| 3272 f = -1; | |
| 3273 | |
| 3274 if (f == -1 || (sps_flags != 0 && f != 0)) | |
| 3275 { | |
| 3276 sps_flags = SPS_BEST; | |
| 483 | 3277 sps_limit = 9999; |
| 372 | 3278 return FAIL; |
| 3279 } | |
| 3280 if (f != 0) | |
| 3281 sps_flags = f; | |
| 3282 } | |
| 3283 | |
| 3284 if (sps_flags == 0) | |
| 3285 sps_flags = SPS_BEST; | |
| 3286 | |
| 3287 return OK; | |
| 3288 } | |
| 3289 | |
| 323 | 3290 /* |
| 5374 | 3291 * "z=": Find badly spelled word under or after the cursor. |
| 323 | 3292 * Give suggestions for the properly spelled word. |
| 638 | 3293 * In Visual mode use the highlighted word as the bad word. |
| 485 | 3294 * When "count" is non-zero use that suggestion. |
| 323 | 3295 */ |
| 3296 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3297 spell_suggest(int count) |
| 323 | 3298 { |
| 3299 char_u *line; | |
| 3300 pos_T prev_cursor = curwin->w_cursor; | |
| 3301 char_u wcopy[MAXWLEN + 2]; | |
| 3302 char_u *p; | |
| 3303 int i; | |
| 3304 int c; | |
| 3305 suginfo_T sug; | |
| 3306 suggest_T *stp; | |
| 372 | 3307 int mouse_used; |
| 388 | 3308 int need_cap; |
| 483 | 3309 int limit; |
| 485 | 3310 int selected = count; |
| 638 | 3311 int badlen = 0; |
| 1938 | 3312 int msg_scroll_save = msg_scroll; |
| 638 | 3313 |
| 3314 if (no_spell_checking(curwin)) | |
| 3315 return; | |
| 3316 | |
| 3317 if (VIsual_active) | |
| 3318 { | |
| 3319 /* Use the Visually selected text as the bad word. But reject | |
| 3320 * a multi-line selection. */ | |
| 3321 if (curwin->w_cursor.lnum != VIsual.lnum) | |
| 3322 { | |
| 6949 | 3323 vim_beep(BO_SPELL); |
| 638 | 3324 return; |
| 3325 } | |
| 3326 badlen = (int)curwin->w_cursor.col - (int)VIsual.col; | |
| 3327 if (badlen < 0) | |
| 3328 badlen = -badlen; | |
| 3329 else | |
| 3330 curwin->w_cursor.col = VIsual.col; | |
| 3331 ++badlen; | |
| 3332 end_visual_mode(); | |
| 3333 } | |
| 5735 | 3334 /* Find the start of the badly spelled word. */ |
| 3335 else if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0 | |
| 346 | 3336 || curwin->w_cursor.col > prev_cursor.col) |
| 3337 { | |
| 3338 /* No bad word or it starts after the cursor: use the word under the | |
| 3339 * cursor. */ | |
| 3340 curwin->w_cursor = prev_cursor; | |
| 3341 line = ml_get_curline(); | |
| 3342 p = line + curwin->w_cursor.col; | |
| 3343 /* Backup to before start of word. */ | |
| 5477 | 3344 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
|
3345 MB_PTR_BACK(line, p); |
| 346 | 3346 /* Forward to start of word. */ |
| 5477 | 3347 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
|
3348 MB_PTR_ADV(p); |
| 346 | 3349 |
| 5477 | 3350 if (!spell_iswordp_nmw(p, curwin)) /* No word found. */ |
| 346 | 3351 { |
| 3352 beep_flush(); | |
| 3353 return; | |
| 3354 } | |
| 835 | 3355 curwin->w_cursor.col = (colnr_T)(p - line); |
| 323 | 3356 } |
| 3357 | |
| 344 | 3358 /* Get the word and its length. */ |
| 3359 | |
| 388 | 3360 /* Figure out if the word should be capitalised. */ |
| 475 | 3361 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col); |
| 3362 | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3363 /* 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
|
3364 line = vim_strsave(ml_get_curline()); |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3365 if (line == NULL) |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3366 goto skip; |
| 388 | 3367 |
| 483 | 3368 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in |
| 3369 * 'spellsuggest', whatever is smaller. */ | |
| 3370 if (sps_limit > (int)Rows - 2) | |
| 3371 limit = (int)Rows - 2; | |
| 3372 else | |
| 3373 limit = sps_limit; | |
| 638 | 3374 spell_find_suggest(line + curwin->w_cursor.col, badlen, &sug, limit, |
| 625 | 3375 TRUE, need_cap, TRUE); |
| 323 | 3376 |
| 3377 if (sug.su_ga.ga_len == 0) | |
| 3378 MSG(_("Sorry, no suggestions")); | |
| 485 | 3379 else if (count > 0) |
| 3380 { | |
| 3381 if (count > sug.su_ga.ga_len) | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
3382 smsg(_("Sorry, only %ld suggestions"), |
| 485 | 3383 (long)sug.su_ga.ga_len); |
| 3384 } | |
| 323 | 3385 else |
| 3386 { | |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
3387 VIM_CLEAR(repl_from); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
3388 VIM_CLEAR(repl_to); |
| 372 | 3389 |
| 474 | 3390 #ifdef FEAT_RIGHTLEFT |
| 3391 /* When 'rightleft' is set the list is drawn right-left. */ | |
| 3392 cmdmsg_rl = curwin->w_p_rl; | |
| 3393 if (cmdmsg_rl) | |
| 3394 msg_col = Columns - 1; | |
| 3395 #endif | |
| 3396 | |
| 323 | 3397 /* List the suggestions. */ |
| 3398 msg_start(); | |
| 914 | 3399 msg_row = Rows - 1; /* for when 'cmdheight' > 1 */ |
| 372 | 3400 lines_left = Rows; /* avoid more prompt */ |
| 323 | 3401 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"), |
| 3402 sug.su_badlen, sug.su_badptr); | |
| 474 | 3403 #ifdef FEAT_RIGHTLEFT |
| 3404 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0) | |
| 3405 { | |
| 3406 /* And now the rabbit from the high hat: Avoid showing the | |
| 3407 * untranslated message rightleft. */ | |
| 3408 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC", | |
| 3409 sug.su_badlen, sug.su_badptr); | |
| 3410 } | |
| 3411 #endif | |
| 323 | 3412 msg_puts(IObuff); |
| 3413 msg_clr_eos(); | |
| 3414 msg_putchar('\n'); | |
| 346 | 3415 |
| 323 | 3416 msg_scroll = TRUE; |
| 3417 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 3418 { | |
| 344 | 3419 stp = &SUG(sug.su_ga, i); |
| 323 | 3420 |
| 3421 /* The suggested word may replace only part of the bad word, add | |
| 3422 * the not replaced part. */ | |
| 2768 | 3423 vim_strncpy(wcopy, stp->st_word, MAXWLEN); |
| 323 | 3424 if (sug.su_badlen > stp->st_orglen) |
| 625 | 3425 vim_strncpy(wcopy + stp->st_wordlen, |
| 323 | 3426 sug.su_badptr + stp->st_orglen, |
| 3427 sug.su_badlen - stp->st_orglen); | |
| 474 | 3428 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1); |
| 3429 #ifdef FEAT_RIGHTLEFT | |
| 3430 if (cmdmsg_rl) | |
| 3431 rl_mirror(IObuff); | |
| 3432 #endif | |
| 3433 msg_puts(IObuff); | |
| 3434 | |
| 3435 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy); | |
| 346 | 3436 msg_puts(IObuff); |
| 3437 | |
| 3438 /* The word may replace more than "su_badlen". */ | |
| 3439 if (sug.su_badlen < stp->st_orglen) | |
| 3440 { | |
| 3441 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""), | |
| 3442 stp->st_orglen, sug.su_badptr); | |
| 3443 msg_puts(IObuff); | |
| 3444 } | |
| 3445 | |
| 324 | 3446 if (p_verbose > 0) |
| 344 | 3447 { |
| 346 | 3448 /* Add the score. */ |
| 351 | 3449 if (sps_flags & (SPS_DOUBLE | SPS_BEST)) |
| 474 | 3450 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)", |
| 344 | 3451 stp->st_salscore ? "s " : "", |
| 3452 stp->st_score, stp->st_altscore); | |
| 3453 else | |
| 474 | 3454 vim_snprintf((char *)IObuff, IOSIZE, " (%d)", |
| 346 | 3455 stp->st_score); |
| 474 | 3456 #ifdef FEAT_RIGHTLEFT |
| 3457 if (cmdmsg_rl) | |
| 3458 /* Mirror the numbers, but keep the leading space. */ | |
| 3459 rl_mirror(IObuff + 1); | |
| 3460 #endif | |
| 346 | 3461 msg_advance(30); |
| 3462 msg_puts(IObuff); | |
| 344 | 3463 } |
| 323 | 3464 msg_putchar('\n'); |
| 3465 } | |
| 3466 | |
| 474 | 3467 #ifdef FEAT_RIGHTLEFT |
| 3468 cmdmsg_rl = FALSE; | |
| 3469 msg_col = 0; | |
| 3470 #endif | |
| 323 | 3471 /* Ask for choice. */ |
| 485 | 3472 selected = prompt_for_number(&mouse_used); |
| 372 | 3473 if (mouse_used) |
| 485 | 3474 selected -= lines_left; |
| 1938 | 3475 lines_left = Rows; /* avoid more prompt */ |
| 3476 /* don't delay for 'smd' in normal_cmd() */ | |
| 3477 msg_scroll = msg_scroll_save; | |
| 485 | 3478 } |
| 3479 | |
| 3480 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK) | |
| 3481 { | |
| 3482 /* Save the from and to text for :spellrepall. */ | |
| 3483 stp = &SUG(sug.su_ga, selected - 1); | |
| 548 | 3484 if (sug.su_badlen > stp->st_orglen) |
| 3485 { | |
| 3486 /* Replacing less than "su_badlen", append the remainder to | |
| 3487 * repl_to. */ | |
| 3488 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen); | |
| 3489 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word, | |
| 3490 sug.su_badlen - stp->st_orglen, | |
| 3491 sug.su_badptr + stp->st_orglen); | |
| 3492 repl_to = vim_strsave(IObuff); | |
| 3493 } | |
| 3494 else | |
| 3495 { | |
| 3496 /* Replacing su_badlen or more, use the whole word. */ | |
| 3497 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen); | |
| 3498 repl_to = vim_strsave(stp->st_word); | |
| 3499 } | |
| 485 | 3500 |
| 3501 /* Replace the word. */ | |
| 1938 | 3502 p = alloc((unsigned)STRLEN(line) - stp->st_orglen |
| 3503 + stp->st_wordlen + 1); | |
| 485 | 3504 if (p != NULL) |
| 3505 { | |
| 835 | 3506 c = (int)(sug.su_badptr - line); |
| 485 | 3507 mch_memmove(p, line, c); |
| 3508 STRCPY(p + c, stp->st_word); | |
| 3509 STRCAT(p, sug.su_badptr + stp->st_orglen); | |
| 3510 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 3511 curwin->w_cursor.col = c; | |
| 3512 | |
| 3513 /* For redo we use a change-word command. */ | |
| 3514 ResetRedobuff(); | |
| 3515 AppendToRedobuff((char_u *)"ciw"); | |
| 620 | 3516 AppendToRedobuffLit(p + c, |
| 625 | 3517 stp->st_wordlen + sug.su_badlen - stp->st_orglen); |
| 485 | 3518 AppendCharToRedobuff(ESC); |
| 819 | 3519 |
| 3520 /* After this "p" may be invalid. */ | |
| 3521 changed_bytes(curwin->w_cursor.lnum, c); | |
| 485 | 3522 } |
| 3523 } | |
| 3524 else | |
| 3525 curwin->w_cursor = prev_cursor; | |
| 323 | 3526 |
| 344 | 3527 spell_find_cleanup(&sug); |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3528 skip: |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3529 vim_free(line); |
| 344 | 3530 } |
| 3531 | |
| 3532 /* | |
| 475 | 3533 * Check if the word at line "lnum" column "col" is required to start with a |
| 3534 * capital. This uses 'spellcapcheck' of the current buffer. | |
| 3535 */ | |
| 3536 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3537 check_need_cap(linenr_T lnum, colnr_T col) |
| 475 | 3538 { |
| 3539 int need_cap = FALSE; | |
| 3540 char_u *line; | |
| 3541 char_u *line_copy = NULL; | |
| 3542 char_u *p; | |
| 3543 colnr_T endcol; | |
| 3544 regmatch_T regmatch; | |
| 3545 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3546 if (curwin->w_s->b_cap_prog == NULL) |
| 475 | 3547 return FALSE; |
| 3548 | |
| 3549 line = ml_get_curline(); | |
| 3550 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
|
3551 if (getwhitecols(line) >= (int)col) |
| 475 | 3552 { |
| 3553 /* At start of line, check if previous line is empty or sentence | |
| 3554 * ends there. */ | |
| 3555 if (lnum == 1) | |
| 3556 need_cap = TRUE; | |
| 3557 else | |
| 3558 { | |
| 3559 line = ml_get(lnum - 1); | |
| 3560 if (*skipwhite(line) == NUL) | |
| 3561 need_cap = TRUE; | |
| 3562 else | |
| 3563 { | |
| 3564 /* Append a space in place of the line break. */ | |
| 3565 line_copy = concat_str(line, (char_u *)" "); | |
| 3566 line = line_copy; | |
| 835 | 3567 endcol = (colnr_T)STRLEN(line); |
| 475 | 3568 } |
| 3569 } | |
| 3570 } | |
| 3571 else | |
| 3572 endcol = col; | |
| 3573 | |
| 3574 if (endcol > 0) | |
| 3575 { | |
| 3576 /* 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
|
3577 regmatch.regprog = curwin->w_s->b_cap_prog; |
| 475 | 3578 regmatch.rm_ic = FALSE; |
| 3579 p = line + endcol; | |
| 3580 for (;;) | |
| 3581 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
3582 MB_PTR_BACK(line, p); |
| 5477 | 3583 if (p == line || spell_iswordp_nmw(p, curwin)) |
| 475 | 3584 break; |
| 3585 if (vim_regexec(®match, p, 0) | |
| 3586 && regmatch.endp[0] == line + endcol) | |
| 3587 { | |
| 3588 need_cap = TRUE; | |
| 3589 break; | |
| 3590 } | |
| 3591 } | |
| 6375 | 3592 curwin->w_s->b_cap_prog = regmatch.regprog; |
| 475 | 3593 } |
| 3594 | |
| 3595 vim_free(line_copy); | |
| 3596 | |
| 3597 return need_cap; | |
| 3598 } | |
| 3599 | |
| 3600 | |
| 3601 /* | |
| 372 | 3602 * ":spellrepall" |
| 3603 */ | |
| 3604 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3605 ex_spellrepall(exarg_T *eap UNUSED) |
| 372 | 3606 { |
| 3607 pos_T pos = curwin->w_cursor; | |
| 3608 char_u *frompat; | |
| 3609 int addlen; | |
| 3610 char_u *line; | |
| 3611 char_u *p; | |
| 3612 int save_ws = p_ws; | |
| 483 | 3613 linenr_T prev_lnum = 0; |
| 372 | 3614 |
| 3615 if (repl_from == NULL || repl_to == NULL) | |
| 3616 { | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
3617 emsg(_("E752: No previous spell replacement")); |
| 372 | 3618 return; |
| 3619 } | |
| 835 | 3620 addlen = (int)(STRLEN(repl_to) - STRLEN(repl_from)); |
| 3621 | |
| 3622 frompat = alloc((unsigned)STRLEN(repl_from) + 7); | |
| 372 | 3623 if (frompat == NULL) |
| 3624 return; | |
| 3625 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from); | |
| 3626 p_ws = FALSE; | |
| 3627 | |
| 483 | 3628 sub_nsubs = 0; |
| 3629 sub_nlines = 0; | |
| 372 | 3630 curwin->w_cursor.lnum = 0; |
| 3631 while (!got_int) | |
| 3632 { | |
|
11521
578df034735d
patch 8.0.0643: when a pattern search is slow Vim becomes unusable
Christian Brabandt <cb@256bit.org>
parents:
11394
diff
changeset
|
3633 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP, NULL, NULL) == 0 |
| 372 | 3634 || u_save_cursor() == FAIL) |
| 3635 break; | |
| 3636 | |
| 3637 /* Only replace when the right word isn't there yet. This happens | |
| 3638 * when changing "etc" to "etc.". */ | |
| 3639 line = ml_get_curline(); | |
| 3640 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col, | |
| 3641 repl_to, STRLEN(repl_to)) != 0) | |
| 3642 { | |
| 835 | 3643 p = alloc((unsigned)STRLEN(line) + addlen + 1); |
| 372 | 3644 if (p == NULL) |
| 3645 break; | |
| 3646 mch_memmove(p, line, curwin->w_cursor.col); | |
| 3647 STRCPY(p + curwin->w_cursor.col, repl_to); | |
| 3648 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from)); | |
| 3649 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 3650 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col); | |
| 483 | 3651 |
| 3652 if (curwin->w_cursor.lnum != prev_lnum) | |
| 3653 { | |
| 3654 ++sub_nlines; | |
| 3655 prev_lnum = curwin->w_cursor.lnum; | |
| 3656 } | |
| 3657 ++sub_nsubs; | |
| 372 | 3658 } |
| 835 | 3659 curwin->w_cursor.col += (colnr_T)STRLEN(repl_to); |
| 372 | 3660 } |
| 3661 | |
| 3662 p_ws = save_ws; | |
| 3663 curwin->w_cursor = pos; | |
| 3664 vim_free(frompat); | |
| 3665 | |
| 483 | 3666 if (sub_nsubs == 0) |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
3667 semsg(_("E753: Not found: %s"), repl_from); |
| 483 | 3668 else |
| 3669 do_sub_msg(FALSE); | |
| 372 | 3670 } |
| 3671 | |
| 3672 /* | |
| 344 | 3673 * Find spell suggestions for "word". Return them in the growarray "*gap" as |
| 3674 * a list of allocated strings. | |
| 3675 */ | |
| 3676 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3677 spell_suggest_list( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3678 garray_T *gap, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3679 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3680 int maxcount, /* maximum nr of suggestions */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3681 int need_cap, /* 'spellcapcheck' matched */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3682 int interactive) |
| 344 | 3683 { |
| 3684 suginfo_T sug; | |
| 3685 int i; | |
| 3686 suggest_T *stp; | |
| 3687 char_u *wcopy; | |
| 3688 | |
| 638 | 3689 spell_find_suggest(word, 0, &sug, maxcount, FALSE, need_cap, interactive); |
| 344 | 3690 |
| 3691 /* Make room in "gap". */ | |
| 3692 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1); | |
| 626 | 3693 if (ga_grow(gap, sug.su_ga.ga_len) == OK) |
| 3694 { | |
| 3695 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 3696 { | |
| 3697 stp = &SUG(sug.su_ga, i); | |
| 3698 | |
| 3699 /* The suggested word may replace only part of "word", add the not | |
| 3700 * replaced part. */ | |
| 3701 wcopy = alloc(stp->st_wordlen | |
| 835 | 3702 + (unsigned)STRLEN(sug.su_badptr + stp->st_orglen) + 1); |
| 626 | 3703 if (wcopy == NULL) |
| 3704 break; | |
| 3705 STRCPY(wcopy, stp->st_word); | |
| 3706 STRCPY(wcopy + stp->st_wordlen, sug.su_badptr + stp->st_orglen); | |
| 3707 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy; | |
| 3708 } | |
| 344 | 3709 } |
| 3710 | |
| 3711 spell_find_cleanup(&sug); | |
| 3712 } | |
| 3713 | |
| 3714 /* | |
| 3715 * Find spell suggestions for the word at the start of "badptr". | |
| 3716 * Return the suggestions in "su->su_ga". | |
| 3717 * The maximum number of suggestions is "maxcount". | |
| 3718 * Note: does use info for the current window. | |
| 3719 * This is based on the mechanisms of Aspell, but completely reimplemented. | |
| 3720 */ | |
| 3721 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3722 spell_find_suggest( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3723 char_u *badptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3724 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
|
3725 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3726 int maxcount, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3727 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
|
3728 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
|
3729 int interactive) |
| 344 | 3730 { |
| 534 | 3731 hlf_T attr = HLF_COUNT; |
| 372 | 3732 char_u buf[MAXPATHL]; |
| 3733 char_u *p; | |
| 3734 int do_combine = FALSE; | |
| 3735 char_u *sps_copy; | |
| 3736 #ifdef FEAT_EVAL | |
| 3737 static int expr_busy = FALSE; | |
| 3738 #endif | |
| 385 | 3739 int c; |
| 503 | 3740 int i; |
| 3741 langp_T *lp; | |
| 344 | 3742 |
| 3743 /* | |
| 3744 * Set the info in "*su". | |
| 3745 */ | |
| 3746 vim_memset(su, 0, sizeof(suginfo_T)); | |
| 3747 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10); | |
| 3748 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10); | |
| 355 | 3749 if (*badptr == NUL) |
| 3750 return; | |
| 344 | 3751 hash_init(&su->su_banned); |
| 3752 | |
| 3753 su->su_badptr = badptr; | |
| 638 | 3754 if (badlen != 0) |
| 3755 su->su_badlen = badlen; | |
| 3756 else | |
| 3757 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL, FALSE); | |
| 344 | 3758 su->su_maxcount = maxcount; |
| 372 | 3759 su->su_maxscore = SCORE_MAXINIT; |
| 344 | 3760 |
| 3761 if (su->su_badlen >= MAXWLEN) | |
| 3762 su->su_badlen = MAXWLEN - 1; /* just in case */ | |
| 3763 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen); | |
| 3764 (void)spell_casefold(su->su_badptr, su->su_badlen, | |
| 3765 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
|
3766 /* 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
|
3767 * 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
|
3768 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
|
3769 |
| 346 | 3770 /* get caps flags for bad word */ |
| 474 | 3771 su->su_badflags = badword_captype(su->su_badptr, |
| 3772 su->su_badptr + su->su_badlen); | |
| 388 | 3773 if (need_cap) |
| 3774 su->su_badflags |= WF_ONECAP; | |
| 344 | 3775 |
| 503 | 3776 /* Find the default language for sound folding. We simply use the first |
| 3777 * one in 'spelllang' that supports sound folding. That's good for when | |
| 3778 * using multiple files for one language, it's not that bad when mixing | |
| 3779 * languages (e.g., "pl,en"). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3780 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
|
3781 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3782 lp = LANGP_ENTRY(curbuf->b_s.b_langp, i); |
| 503 | 3783 if (lp->lp_sallang != NULL) |
| 3784 { | |
| 3785 su->su_sallang = lp->lp_sallang; | |
| 3786 break; | |
| 3787 } | |
| 3788 } | |
| 3789 | |
| 534 | 3790 /* Soundfold the bad word with the default sound folding, so that we don't |
| 3791 * have to do this many times. */ | |
| 3792 if (su->su_sallang != NULL) | |
| 3793 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE, | |
| 3794 su->su_sal_badword); | |
| 3795 | |
| 385 | 3796 /* If the word is not capitalised and spell_check() doesn't consider the |
| 3797 * word to be bad then it might need to be capitalised. Add a suggestion | |
| 3798 * for that. */ | |
| 455 | 3799 c = PTR2CHAR(su->su_badptr); |
| 534 | 3800 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT) |
| 385 | 3801 { |
| 3802 make_case_word(su->su_badword, buf, WF_ONECAP); | |
| 3803 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE, | |
| 625 | 3804 0, TRUE, su->su_sallang, FALSE); |
| 385 | 3805 } |
| 3806 | |
| 344 | 3807 /* Ban the bad word itself. It may appear in another region. */ |
| 358 | 3808 if (banbadword) |
| 3809 add_banned(su, su->su_badword); | |
| 344 | 3810 |
| 372 | 3811 /* Make a copy of 'spellsuggest', because the expression may change it. */ |
| 3812 sps_copy = vim_strsave(p_sps); | |
| 3813 if (sps_copy == NULL) | |
| 3814 return; | |
| 3815 | |
| 3816 /* Loop over the items in 'spellsuggest'. */ | |
| 3817 for (p = sps_copy; *p != NUL; ) | |
| 3818 { | |
| 3819 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 3820 | |
| 3821 if (STRNCMP(buf, "expr:", 5) == 0) | |
| 3822 { | |
| 3823 #ifdef FEAT_EVAL | |
| 375 | 3824 /* Evaluate an expression. Skip this when called recursively, |
| 3825 * when using spellsuggest() in the expression. */ | |
| 372 | 3826 if (!expr_busy) |
| 3827 { | |
| 3828 expr_busy = TRUE; | |
| 3829 spell_suggest_expr(su, buf + 5); | |
| 3830 expr_busy = FALSE; | |
| 3831 } | |
| 3832 #endif | |
| 3833 } | |
| 3834 else if (STRNCMP(buf, "file:", 5) == 0) | |
| 3835 /* Use list of suggestions in a file. */ | |
| 3836 spell_suggest_file(su, buf + 5); | |
| 3837 else | |
| 3838 { | |
| 3839 /* Use internal method. */ | |
| 625 | 3840 spell_suggest_intern(su, interactive); |
| 372 | 3841 if (sps_flags & SPS_DOUBLE) |
| 3842 do_combine = TRUE; | |
| 3843 } | |
| 3844 } | |
| 3845 | |
| 3846 vim_free(sps_copy); | |
| 3847 | |
| 3848 if (do_combine) | |
| 3849 /* Combine the two list of suggestions. This must be done last, | |
| 3850 * because sorting changes the order again. */ | |
| 3851 score_combine(su); | |
| 3852 } | |
| 3853 | |
| 3854 #ifdef FEAT_EVAL | |
| 3855 /* | |
| 3856 * Find suggestions by evaluating expression "expr". | |
| 3857 */ | |
| 3858 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3859 spell_suggest_expr(suginfo_T *su, char_u *expr) |
| 372 | 3860 { |
| 3861 list_T *list; | |
| 3862 listitem_T *li; | |
| 3863 int score; | |
| 3864 char_u *p; | |
| 3865 | |
| 3866 /* The work is split up in a few parts to avoid having to export | |
| 3867 * suginfo_T. | |
| 3868 * First evaluate the expression and get the resulting list. */ | |
| 3869 list = eval_spell_expr(su->su_badword, expr); | |
| 3870 if (list != NULL) | |
| 3871 { | |
| 3872 /* Loop over the items in the list. */ | |
| 3873 for (li = list->lv_first; li != NULL; li = li->li_next) | |
| 3874 if (li->li_tv.v_type == VAR_LIST) | |
| 3875 { | |
| 3876 /* Get the word and the score from the items. */ | |
| 3877 score = get_spellword(li->li_tv.vval.v_list, &p); | |
| 625 | 3878 if (score >= 0 && score <= su->su_maxscore) |
| 3879 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 3880 score, 0, TRUE, su->su_sallang, FALSE); | |
| 372 | 3881 } |
| 3882 list_unref(list); | |
| 3883 } | |
| 3884 | |
| 625 | 3885 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3886 check_suggestions(su, &su->su_ga); | |
| 372 | 3887 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 3888 } | |
| 3889 #endif | |
| 3890 | |
| 3891 /* | |
| 474 | 3892 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'. |
| 372 | 3893 */ |
| 3894 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3895 spell_suggest_file(suginfo_T *su, char_u *fname) |
| 372 | 3896 { |
| 3897 FILE *fd; | |
| 3898 char_u line[MAXWLEN * 2]; | |
| 3899 char_u *p; | |
| 3900 int len; | |
| 3901 char_u cword[MAXWLEN]; | |
| 3902 | |
| 3903 /* Open the file. */ | |
| 3904 fd = mch_fopen((char *)fname, "r"); | |
| 3905 if (fd == NULL) | |
| 3906 { | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
15034
diff
changeset
|
3907 semsg(_(e_notopen), fname); |
| 372 | 3908 return; |
| 3909 } | |
| 3910 | |
| 3911 /* Read it line by line. */ | |
| 3912 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int) | |
| 3913 { | |
| 3914 line_breakcheck(); | |
| 3915 | |
| 3916 p = vim_strchr(line, '/'); | |
| 3917 if (p == NULL) | |
| 3918 continue; /* No Tab found, just skip the line. */ | |
| 3919 *p++ = NUL; | |
| 3920 if (STRICMP(su->su_badword, line) == 0) | |
| 3921 { | |
| 3922 /* Match! Isolate the good word, until CR or NL. */ | |
| 3923 for (len = 0; p[len] >= ' '; ++len) | |
| 3924 ; | |
| 3925 p[len] = NUL; | |
| 3926 | |
| 3927 /* If the suggestion doesn't have specific case duplicate the case | |
| 3928 * of the bad word. */ | |
| 3929 if (captype(p, NULL) == 0) | |
| 3930 { | |
| 3931 make_case_word(p, cword, su->su_badflags); | |
| 3932 p = cword; | |
| 3933 } | |
| 3934 | |
| 3935 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 625 | 3936 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE); |
| 372 | 3937 } |
| 3938 } | |
| 3939 | |
| 3940 fclose(fd); | |
| 3941 | |
| 625 | 3942 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3943 check_suggestions(su, &su->su_ga); | |
| 372 | 3944 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 3945 } | |
| 3946 | |
| 3947 /* | |
| 3948 * Find suggestions for the internal method indicated by "sps_flags". | |
| 3949 */ | |
| 3950 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3951 spell_suggest_intern(suginfo_T *su, int interactive) |
| 625 | 3952 { |
| 3953 /* | |
| 3954 * Load the .sug file(s) that are available and not done yet. | |
| 3955 */ | |
| 3956 suggest_load_files(); | |
| 3957 | |
| 344 | 3958 /* |
| 346 | 3959 * 1. Try special cases, such as repeating a word: "the the" -> "the". |
| 344 | 3960 * |
| 3961 * Set a maximum score to limit the combination of operations that is | |
| 3962 * tried. | |
| 3963 */ | |
| 346 | 3964 suggest_try_special(su); |
| 3965 | |
| 3966 /* | |
| 3967 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries | |
| 3968 * from the .aff file and inserting a space (split the word). | |
| 3969 */ | |
| 3970 suggest_try_change(su); | |
| 344 | 3971 |
| 3972 /* For the resulting top-scorers compute the sound-a-like score. */ | |
| 3973 if (sps_flags & SPS_DOUBLE) | |
| 3974 score_comp_sal(su); | |
| 3975 | |
| 3976 /* | |
| 346 | 3977 * 3. Try finding sound-a-like words. |
| 344 | 3978 */ |
| 625 | 3979 if ((sps_flags & SPS_FAST) == 0) |
| 3980 { | |
| 3981 if (sps_flags & SPS_BEST) | |
| 3982 /* Adjust the word score for the suggestions found so far for how | |
| 3983 * they sounds like. */ | |
| 3984 rescore_suggestions(su); | |
| 3985 | |
| 3986 /* | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3987 * While going through the soundfold tree "su_maxscore" is the score |
| 625 | 3988 * for the soundfold word, limits the changes that are being tried, |
| 3989 * and "su_sfmaxscore" the rescored score, which is set by | |
| 3990 * cleanup_suggestions(). | |
| 3991 * First find words with a small edit distance, because this is much | |
| 3992 * faster and often already finds the top-N suggestions. If we didn't | |
| 3993 * find many suggestions try again with a higher edit distance. | |
| 3994 * "sl_sounddone" is used to avoid doing the same word twice. | |
| 3995 */ | |
| 3996 suggest_try_soundalike_prep(); | |
| 3997 su->su_maxscore = SCORE_SFMAX1; | |
| 3998 su->su_sfmaxscore = SCORE_MAXINIT * 3; | |
| 346 | 3999 suggest_try_soundalike(su); |
| 625 | 4000 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) |
| 4001 { | |
| 4002 /* We didn't find enough matches, try again, allowing more | |
| 4003 * changes to the soundfold word. */ | |
| 4004 su->su_maxscore = SCORE_SFMAX2; | |
| 4005 suggest_try_soundalike(su); | |
| 4006 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) | |
| 4007 { | |
| 4008 /* Still didn't find enough matches, try again, allowing even | |
| 4009 * more changes to the soundfold word. */ | |
| 4010 su->su_maxscore = SCORE_SFMAX3; | |
| 4011 suggest_try_soundalike(su); | |
| 4012 } | |
| 4013 } | |
| 4014 su->su_maxscore = su->su_sfmaxscore; | |
| 4015 suggest_try_soundalike_finish(); | |
| 4016 } | |
| 4017 | |
| 4018 /* When CTRL-C was hit while searching do show the results. Only clear | |
| 4019 * got_int when using a command, not for spellsuggest(). */ | |
| 344 | 4020 ui_breakcheck(); |
| 625 | 4021 if (interactive && got_int) |
| 344 | 4022 { |
| 4023 (void)vgetc(); | |
| 4024 got_int = FALSE; | |
| 4025 } | |
| 4026 | |
| 372 | 4027 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0) |
| 344 | 4028 { |
| 4029 if (sps_flags & SPS_BEST) | |
| 4030 /* Adjust the word score for how it sounds like. */ | |
| 4031 rescore_suggestions(su); | |
| 4032 | |
| 625 | 4033 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 4034 check_suggestions(su, &su->su_ga); | |
| 372 | 4035 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 344 | 4036 } |
| 4037 } | |
| 4038 | |
| 4039 /* | |
| 4040 * Free the info put in "*su" by spell_find_suggest(). | |
| 4041 */ | |
| 4042 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4043 spell_find_cleanup(suginfo_T *su) |
| 344 | 4044 { |
| 4045 int i; | |
| 4046 | |
| 4047 /* Free the suggestions. */ | |
| 4048 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 4049 vim_free(SUG(su->su_ga, i).st_word); | |
| 4050 ga_clear(&su->su_ga); | |
| 4051 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 4052 vim_free(SUG(su->su_sga, i).st_word); | |
| 4053 ga_clear(&su->su_sga); | |
| 323 | 4054 |
| 4055 /* Free the banned words. */ | |
| 625 | 4056 hash_clear_all(&su->su_banned, 0); |
| 323 | 4057 } |
| 4058 | |
| 4059 /* | |
| 324 | 4060 * Make a copy of "word", with the first letter upper or lower cased, to |
| 4061 * "wcopy[MAXWLEN]". "word" must not be empty. | |
| 4062 * The result is NUL terminated. | |
| 323 | 4063 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
4064 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4065 onecap_copy( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4066 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4067 char_u *wcopy, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4068 int upper) /* TRUE: first letter made upper case */ |
| 323 | 4069 { |
| 4070 char_u *p; | |
| 4071 int c; | |
| 4072 int l; | |
| 4073 | |
| 4074 p = word; | |
| 4075 #ifdef FEAT_MBYTE | |
| 4076 if (has_mbyte) | |
| 474 | 4077 c = mb_cptr2char_adv(&p); |
| 323 | 4078 else |
| 4079 #endif | |
| 4080 c = *p++; | |
| 4081 if (upper) | |
| 324 | 4082 c = SPELL_TOUPPER(c); |
| 323 | 4083 else |
| 324 | 4084 c = SPELL_TOFOLD(c); |
| 323 | 4085 #ifdef FEAT_MBYTE |
| 4086 if (has_mbyte) | |
| 4087 l = mb_char2bytes(c, wcopy); | |
| 4088 else | |
| 4089 #endif | |
| 4090 { | |
| 4091 l = 1; | |
| 4092 wcopy[0] = c; | |
| 4093 } | |
| 376 | 4094 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1); |
| 323 | 4095 } |
| 4096 | |
| 4097 /* | |
| 324 | 4098 * Make a copy of "word" with all the letters upper cased into |
| 4099 * "wcopy[MAXWLEN]". The result is NUL terminated. | |
| 323 | 4100 */ |
| 4101 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4102 allcap_copy(char_u *word, char_u *wcopy) |
| 323 | 4103 { |
| 4104 char_u *s; | |
| 4105 char_u *d; | |
| 4106 int c; | |
| 4107 | |
| 4108 d = wcopy; | |
| 4109 for (s = word; *s != NUL; ) | |
| 4110 { | |
| 4111 #ifdef FEAT_MBYTE | |
| 4112 if (has_mbyte) | |
| 474 | 4113 c = mb_cptr2char_adv(&s); |
| 323 | 4114 else |
| 4115 #endif | |
| 4116 c = *s++; | |
| 492 | 4117 |
| 4118 #ifdef FEAT_MBYTE | |
| 3042 | 4119 /* We only change 0xdf to SS when we are certain latin1 is used. It |
| 492 | 4120 * would cause weird errors in other 8-bit encodings. */ |
| 4121 if (enc_latin1like && c == 0xdf) | |
| 4122 { | |
| 4123 c = 'S'; | |
| 4124 if (d - wcopy >= MAXWLEN - 1) | |
| 4125 break; | |
| 4126 *d++ = c; | |
| 4127 } | |
| 4128 else | |
| 4129 #endif | |
| 4130 c = SPELL_TOUPPER(c); | |
| 323 | 4131 |
| 4132 #ifdef FEAT_MBYTE | |
| 4133 if (has_mbyte) | |
| 4134 { | |
| 4135 if (d - wcopy >= MAXWLEN - MB_MAXBYTES) | |
| 4136 break; | |
| 4137 d += mb_char2bytes(c, d); | |
| 4138 } | |
| 4139 else | |
| 4140 #endif | |
| 4141 { | |
| 4142 if (d - wcopy >= MAXWLEN - 1) | |
| 4143 break; | |
| 4144 *d++ = c; | |
| 4145 } | |
| 4146 } | |
| 4147 *d = NUL; | |
| 4148 } | |
| 4149 | |
| 4150 /* | |
| 346 | 4151 * Try finding suggestions by recognizing specific situations. |
| 4152 */ | |
| 4153 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4154 suggest_try_special(suginfo_T *su) |
| 346 | 4155 { |
| 4156 char_u *p; | |
| 376 | 4157 size_t len; |
| 346 | 4158 int c; |
| 4159 char_u word[MAXWLEN]; | |
| 4160 | |
| 4161 /* | |
| 4162 * Recognize a word that is repeated: "the the". | |
| 4163 */ | |
| 4164 p = skiptowhite(su->su_fbadword); | |
| 4165 len = p - su->su_fbadword; | |
| 4166 p = skipwhite(p); | |
| 4167 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0) | |
| 4168 { | |
| 4169 /* Include badflags: if the badword is onecap or allcap | |
| 4170 * use that for the goodword too: "The the" -> "The". */ | |
| 4171 c = su->su_fbadword[len]; | |
| 4172 su->su_fbadword[len] = NUL; | |
| 4173 make_case_word(su->su_fbadword, word, su->su_badflags); | |
| 4174 su->su_fbadword[len] = c; | |
| 534 | 4175 |
| 4176 /* Give a soundalike score of 0, compute the score as if deleting one | |
| 4177 * character. */ | |
| 4178 add_suggestion(su, &su->su_ga, word, su->su_badlen, | |
| 625 | 4179 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE); |
| 346 | 4180 } |
| 4181 } | |
| 4182 | |
| 4183 /* | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4184 * 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
|
4185 * Output is dumped in "suggestprof". |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4186 */ |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4187 #if 0 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4188 # define SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4189 proftime_T current; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4190 proftime_T total; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4191 proftime_T times[STATE_FINAL + 1]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4192 long counts[STATE_FINAL + 1]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4193 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4194 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4195 prof_init(void) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4196 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4197 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
|
4198 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4199 profile_zero(×[i]); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4200 counts[i] = 0; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4201 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4202 profile_start(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4203 profile_start(&total); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4204 } |
|
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 /* call before changing state */ |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4207 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4208 prof_store(state_T state) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4209 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4210 profile_end(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4211 profile_add(×[state], ¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4212 ++counts[state]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4213 profile_start(¤t); |
|
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 # 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
|
4216 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4217 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4218 prof_report(char *name) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4219 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4220 FILE *fd = fopen("suggestprof", "a"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4221 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4222 profile_end(&total); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4223 fprintf(fd, "-----------------------\n"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4224 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
|
4225 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
|
4226 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
|
4227 fclose(fd); |
|
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 #else |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4230 # define PROF_STORE(state) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4231 #endif |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4232 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4233 /* |
| 323 | 4234 * Try finding suggestions by adding/removing/swapping letters. |
| 4235 */ | |
| 4236 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4237 suggest_try_change(suginfo_T *su) |
| 323 | 4238 { |
| 4239 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */ | |
| 625 | 4240 int n; |
| 4241 char_u *p; | |
| 4242 int lpi; | |
| 323 | 4243 langp_T *lp; |
| 4244 | |
| 4245 /* We make a copy of the case-folded bad word, so that we can modify it | |
| 346 | 4246 * to find matches (esp. REP items). Append some more text, changing |
| 4247 * chars after the bad word may help. */ | |
| 323 | 4248 STRCPY(fword, su->su_fbadword); |
| 835 | 4249 n = (int)STRLEN(fword); |
| 346 | 4250 p = su->su_badptr + su->su_badlen; |
| 835 | 4251 (void)spell_casefold(p, (int)STRLEN(p), fword + n, MAXWLEN - n); |
| 323 | 4252 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4253 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
|
4254 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4255 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 482 | 4256 |
| 500 | 4257 /* If reloading a spell file fails it's still in the list but |
| 4258 * everything has been cleared. */ | |
| 625 | 4259 if (lp->lp_slang->sl_fbyts == NULL) |
| 500 | 4260 continue; |
| 4261 | |
| 625 | 4262 /* 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
|
4263 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4264 prof_init(); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4265 #endif |
| 625 | 4266 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
|
4267 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4268 prof_report("try_change"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4269 #endif |
| 625 | 4270 } |
| 4271 } | |
| 4272 | |
| 4273 /* Check the maximum score, if we go over it we won't try this change. */ | |
| 4274 #define TRY_DEEPER(su, stack, depth, add) \ | |
| 4275 (stack[depth].ts_score + (add) < su->su_maxscore) | |
| 4276 | |
| 4277 /* | |
| 4278 * Try finding suggestions by adding/removing/swapping letters. | |
| 4279 * | |
| 4280 * This uses a state machine. At each node in the tree we try various | |
| 4281 * operations. When trying if an operation works "depth" is increased and the | |
| 4282 * stack[] is used to store info. This allows combinations, thus insert one | |
| 4283 * character, replace one and delete another. The number of changes is | |
| 4284 * limited by su->su_maxscore. | |
| 4285 * | |
| 4286 * After implementing this I noticed an article by Kemal Oflazer that | |
| 4287 * describes something similar: "Error-tolerant Finite State Recognition with | |
| 4288 * Applications to Morphological Analysis and Spelling Correction" (1996). | |
| 4289 * The implementation in the article is simplified and requires a stack of | |
| 4290 * unknown depth. The implementation here only needs a stack depth equal to | |
| 4291 * the length of the word. | |
| 4292 * | |
| 4293 * This is also used for the sound-folded word, "soundfold" is TRUE then. | |
| 4294 * The mechanism is the same, but we find a match with a sound-folded word | |
| 4295 * that comes from one or more original words. Each of these words may be | |
| 4296 * added, this is done by add_sound_suggest(). | |
| 4297 * Don't use: | |
| 4298 * the prefix tree or the keep-case tree | |
| 4299 * "su->su_badlen" | |
| 4300 * anything to do with upper and lower case | |
| 4301 * anything to do with word or non-word characters ("spell_iswordp()") | |
| 4302 * banned words | |
| 4303 * word flags (rare, region, compounding) | |
| 4304 * word splitting for now | |
| 4305 * "similar_chars()" | |
| 4306 * use "slang->sl_repsal" instead of "lp->lp_replang->sl_rep" | |
| 4307 */ | |
| 4308 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4309 suggest_trie_walk( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4310 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4311 langp_T *lp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4312 char_u *fword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4313 int soundfold) |
| 625 | 4314 { |
| 4315 char_u tword[MAXWLEN]; /* good word collected so far */ | |
| 4316 trystate_T stack[MAXWLEN]; | |
| 4317 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
|
4318 * concatenation of prefix compound |
| 625 | 4319 * words and split word. NUL terminated |
| 4320 * when going deeper but not when coming | |
| 4321 * back. */ | |
| 4322 char_u compflags[MAXWLEN]; /* compound flags, one for each word */ | |
| 4323 trystate_T *sp; | |
| 4324 int newscore; | |
| 4325 int score; | |
| 4326 char_u *byts, *fbyts, *pbyts; | |
| 4327 idx_T *idxs, *fidxs, *pidxs; | |
| 4328 int depth; | |
| 4329 int c, c2, c3; | |
| 4330 int n = 0; | |
| 4331 int flags; | |
| 4332 garray_T *gap; | |
| 4333 idx_T arridx; | |
| 4334 int len; | |
| 4335 char_u *p; | |
| 4336 fromto_T *ftp; | |
| 4337 int fl = 0, tl; | |
| 4338 int repextra = 0; /* extra bytes in fword[] from REP item */ | |
| 4339 slang_T *slang = lp->lp_slang; | |
| 4340 int fword_ends; | |
| 4341 int goodword_ends; | |
| 4342 #ifdef DEBUG_TRIEWALK | |
| 4343 /* Stores the name of the change made at each level. */ | |
| 4344 char_u changename[MAXWLEN][80]; | |
| 4345 #endif | |
| 4346 int breakcheckcount = 1000; | |
| 4347 int compound_ok; | |
| 4348 | |
| 4349 /* | |
| 4350 * Go through the whole case-fold tree, try changes at each node. | |
| 4351 * "tword[]" contains the word collected from nodes in the tree. | |
| 4352 * "fword[]" the word we are trying to match with (initially the bad | |
| 4353 * word). | |
| 4354 */ | |
| 4355 depth = 0; | |
| 4356 sp = &stack[0]; | |
| 4357 vim_memset(sp, 0, sizeof(trystate_T)); | |
| 4358 sp->ts_curi = 1; | |
| 4359 | |
| 4360 if (soundfold) | |
| 4361 { | |
| 4362 /* Going through the soundfold tree. */ | |
| 4363 byts = fbyts = slang->sl_sbyts; | |
| 4364 idxs = fidxs = slang->sl_sidxs; | |
| 4365 pbyts = NULL; | |
| 4366 pidxs = NULL; | |
| 4367 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 4368 sp->ts_state = STATE_START; | |
| 4369 } | |
| 4370 else | |
| 4371 { | |
| 330 | 4372 /* |
| 375 | 4373 * When there are postponed prefixes we need to use these first. At |
| 4374 * the end of the prefix we continue in the case-fold tree. | |
| 4375 */ | |
| 482 | 4376 fbyts = slang->sl_fbyts; |
| 4377 fidxs = slang->sl_fidxs; | |
| 4378 pbyts = slang->sl_pbyts; | |
| 4379 pidxs = slang->sl_pidxs; | |
| 375 | 4380 if (pbyts != NULL) |
| 4381 { | |
| 4382 byts = pbyts; | |
| 4383 idxs = pidxs; | |
| 482 | 4384 sp->ts_prefixdepth = PFD_PREFIXTREE; |
| 375 | 4385 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */ |
| 4386 } | |
| 4387 else | |
| 4388 { | |
| 4389 byts = fbyts; | |
| 4390 idxs = fidxs; | |
| 482 | 4391 sp->ts_prefixdepth = PFD_NOPREFIX; |
| 485 | 4392 sp->ts_state = STATE_START; |
| 375 | 4393 } |
| 625 | 4394 } |
| 4395 | |
| 4396 /* | |
| 4397 * Loop to find all suggestions. At each round we either: | |
| 4398 * - For the current state try one operation, advance "ts_curi", | |
| 4399 * increase "depth". | |
| 4400 * - When a state is done go to the next, set "ts_state". | |
| 4401 * - When all states are tried decrease "depth". | |
| 4402 */ | |
| 4403 while (depth >= 0 && !got_int) | |
| 4404 { | |
| 4405 sp = &stack[depth]; | |
| 4406 switch (sp->ts_state) | |
| 4407 { | |
| 4408 case STATE_START: | |
| 4409 case STATE_NOPREFIX: | |
| 4410 /* | |
| 4411 * Start of node: Deal with NUL bytes, which means | |
| 4412 * tword[] may end here. | |
| 4413 */ | |
| 4414 arridx = sp->ts_arridx; /* current node in the tree */ | |
| 4415 len = byts[arridx]; /* bytes in this node */ | |
| 4416 arridx += sp->ts_curi; /* index of current byte */ | |
| 4417 | |
| 4418 if (sp->ts_prefixdepth == PFD_PREFIXTREE) | |
| 4419 { | |
| 4420 /* Skip over the NUL bytes, we use them later. */ | |
| 4421 for (n = 0; n < len && byts[arridx + n] == 0; ++n) | |
| 4422 ; | |
| 4423 sp->ts_curi += n; | |
| 4424 | |
| 4425 /* Always past NUL bytes now. */ | |
| 4426 n = (int)sp->ts_state; | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4427 PROF_STORE(sp->ts_state) |
| 625 | 4428 sp->ts_state = STATE_ENDNUL; |
| 4429 sp->ts_save_badflags = su->su_badflags; | |
| 4430 | |
| 4431 /* At end of a prefix or at start of prefixtree: check for | |
| 4432 * following word. */ | |
| 4433 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX) | |
| 4434 { | |
| 4435 /* Set su->su_badflags to the caps type at this position. | |
| 4436 * Use the caps type until here for the prefix itself. */ | |
| 4437 #ifdef FEAT_MBYTE | |
| 4438 if (has_mbyte) | |
| 4439 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); | |
| 4440 else | |
| 4441 #endif | |
| 4442 n = sp->ts_fidx; | |
| 4443 flags = badword_captype(su->su_badptr, su->su_badptr + n); | |
| 4444 su->su_badflags = badword_captype(su->su_badptr + n, | |
| 455 | 4445 su->su_badptr + su->su_badlen); |
| 625 | 4446 #ifdef DEBUG_TRIEWALK |
| 4447 sprintf(changename[depth], "prefix"); | |
| 4448 #endif | |
| 4449 go_deeper(stack, depth, 0); | |
| 4450 ++depth; | |
| 4451 sp = &stack[depth]; | |
| 4452 sp->ts_prefixdepth = depth - 1; | |
| 4453 byts = fbyts; | |
| 4454 idxs = fidxs; | |
| 4455 sp->ts_arridx = 0; | |
| 4456 | |
| 4457 /* Move the prefix to preword[] with the right case | |
| 4458 * and make find_keepcap_word() works. */ | |
| 4459 tword[sp->ts_twordlen] = NUL; | |
| 4460 make_case_word(tword + sp->ts_splitoff, | |
| 4461 preword + sp->ts_prewordlen, flags); | |
| 835 | 4462 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 625 | 4463 sp->ts_splitoff = sp->ts_twordlen; |
| 4464 } | |
| 4465 break; | |
| 4466 } | |
| 4467 | |
| 4468 if (sp->ts_curi > len || byts[arridx] != 0) | |
| 4469 { | |
| 4470 /* 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
|
4471 PROF_STORE(sp->ts_state) |
| 625 | 4472 sp->ts_state = STATE_ENDNUL; |
| 4473 sp->ts_save_badflags = su->su_badflags; | |
| 4474 break; | |
| 4475 } | |
| 4476 | |
| 4477 /* | |
| 4478 * End of word in tree. | |
| 4479 */ | |
| 4480 ++sp->ts_curi; /* eat one NUL byte */ | |
| 4481 | |
| 4482 flags = (int)idxs[arridx]; | |
| 698 | 4483 |
| 4484 /* Skip words with the NOSUGGEST flag. */ | |
| 4485 if (flags & WF_NOSUGGEST) | |
| 4486 break; | |
| 4487 | |
| 625 | 4488 fword_ends = (fword[sp->ts_fidx] == NUL |
| 4489 || (soundfold | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
4490 ? VIM_ISWHITE(fword[sp->ts_fidx]) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4491 : !spell_iswordp(fword + sp->ts_fidx, curwin))); |
| 625 | 4492 tword[sp->ts_twordlen] = NUL; |
| 4493 | |
| 4494 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL | |
| 485 | 4495 && (sp->ts_flags & TSF_PREFIXOK) == 0) |
| 625 | 4496 { |
| 4497 /* There was a prefix before the word. Check that the prefix | |
| 4498 * can be used with this word. */ | |
| 4499 /* Count the length of the NULs in the prefix. If there are | |
| 4500 * none this must be the first try without a prefix. */ | |
| 4501 n = stack[sp->ts_prefixdepth].ts_arridx; | |
| 4502 len = pbyts[n++]; | |
| 4503 for (c = 0; c < len && pbyts[n + c] == 0; ++c) | |
| 4504 ; | |
| 4505 if (c > 0) | |
| 4506 { | |
| 4507 c = valid_word_prefix(c, n, flags, | |
| 483 | 4508 tword + sp->ts_splitoff, slang, FALSE); |
| 625 | 4509 if (c == 0) |
| 4510 break; | |
| 4511 | |
| 4512 /* Use the WF_RARE flag for a rare prefix. */ | |
| 4513 if (c & WF_RAREPFX) | |
| 4514 flags |= WF_RARE; | |
| 4515 | |
| 4516 /* Tricky: when checking for both prefix and compounding | |
| 4517 * we run into the prefix flag first. | |
| 4518 * Remember that it's OK, so that we accept the prefix | |
| 4519 * when arriving at a compound flag. */ | |
| 4520 sp->ts_flags |= TSF_PREFIXOK; | |
| 4521 } | |
| 4522 } | |
| 4523 | |
| 4524 /* Check NEEDCOMPOUND: can't use word without compounding. Do try | |
| 4525 * appending another compound word below. */ | |
| 4526 if (sp->ts_complen == sp->ts_compsplit && fword_ends | |
| 500 | 4527 && (flags & WF_NEEDCOMP)) |
| 625 | 4528 goodword_ends = FALSE; |
| 4529 else | |
| 4530 goodword_ends = TRUE; | |
| 4531 | |
| 4532 p = NULL; | |
| 4533 compound_ok = TRUE; | |
| 4534 if (sp->ts_complen > sp->ts_compsplit) | |
| 4535 { | |
| 4536 if (slang->sl_nobreak) | |
| 4537 { | |
| 4538 /* There was a word before this word. When there was no | |
| 4539 * change in this word (it was correct) add the first word | |
| 4540 * as a suggestion. If this word was corrected too, we | |
| 4541 * need to check if a correct word follows. */ | |
| 4542 if (sp->ts_fidx - sp->ts_splitfidx | |
| 4543 == sp->ts_twordlen - sp->ts_splitoff | |
| 4544 && STRNCMP(fword + sp->ts_splitfidx, | |
| 4545 tword + sp->ts_splitoff, | |
| 4546 sp->ts_fidx - sp->ts_splitfidx) == 0) | |
| 492 | 4547 { |
| 625 | 4548 preword[sp->ts_prewordlen] = NUL; |
| 4549 newscore = score_wordcount_adj(slang, sp->ts_score, | |
| 4550 preword + sp->ts_prewordlen, | |
| 4551 sp->ts_prewordlen > 0); | |
| 4552 /* Add the suggestion if the score isn't too bad. */ | |
| 4553 if (newscore <= su->su_maxscore) | |
| 492 | 4554 add_suggestion(su, &su->su_ga, preword, |
| 503 | 4555 sp->ts_splitfidx - repextra, |
| 625 | 4556 newscore, 0, FALSE, |
| 4557 lp->lp_sallang, FALSE); | |
| 4558 break; | |
| 492 | 4559 } |
| 490 | 4560 } |
| 4561 else | |
| 625 | 4562 { |
| 4563 /* There was a compound word before this word. If this | |
| 4564 * word does not support compounding then give up | |
| 4565 * (splitting is tried for the word without compound | |
| 4566 * flag). */ | |
| 4567 if (((unsigned)flags >> 24) == 0 | |
| 4568 || sp->ts_twordlen - sp->ts_splitoff | |
| 4569 < slang->sl_compminlen) | |
| 4570 break; | |
| 4571 #ifdef FEAT_MBYTE | |
| 4572 /* For multi-byte chars check character length against | |
| 4573 * COMPOUNDMIN. */ | |
| 4574 if (has_mbyte | |
| 4575 && slang->sl_compminlen > 0 | |
| 4576 && mb_charlen(tword + sp->ts_splitoff) | |
| 4577 < slang->sl_compminlen) | |
| 4578 break; | |
| 4579 #endif | |
| 4580 | |
| 4581 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 4582 compflags[sp->ts_complen + 1] = NUL; | |
| 4583 vim_strncpy(preword + sp->ts_prewordlen, | |
| 4584 tword + sp->ts_splitoff, | |
| 4585 sp->ts_twordlen - sp->ts_splitoff); | |
| 1762 | 4586 |
| 4587 /* Verify CHECKCOMPOUNDPATTERN rules. */ | |
| 4588 if (match_checkcompoundpattern(preword, sp->ts_prewordlen, | |
| 4589 &slang->sl_comppat)) | |
| 4590 compound_ok = FALSE; | |
| 4591 | |
| 4592 if (compound_ok) | |
| 4593 { | |
| 4594 p = preword; | |
| 4595 while (*skiptowhite(p) != NUL) | |
| 4596 p = skipwhite(skiptowhite(p)); | |
| 4597 if (fword_ends && !can_compound(slang, p, | |
| 625 | 4598 compflags + sp->ts_compsplit)) |
| 1762 | 4599 /* Compound is not allowed. But it may still be |
| 4600 * possible if we add another (short) word. */ | |
| 4601 compound_ok = FALSE; | |
| 4602 } | |
| 625 | 4603 |
| 4604 /* Get pointer to last char of previous word. */ | |
| 4605 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
|
4606 MB_PTR_BACK(preword, p); |
| 625 | 4607 } |
| 4608 } | |
| 4609 | |
| 4610 /* | |
| 4611 * Form the word with proper case in preword. | |
| 4612 * If there is a word from a previous split, append. | |
| 4613 * For the soundfold tree don't change the case, simply append. | |
| 4614 */ | |
| 4615 if (soundfold) | |
| 4616 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff); | |
| 4617 else if (flags & WF_KEEPCAP) | |
| 4618 /* Must find the word in the keep-case tree. */ | |
| 4619 find_keepcap_word(slang, tword + sp->ts_splitoff, | |
| 483 | 4620 preword + sp->ts_prewordlen); |
| 625 | 4621 else |
| 4622 { | |
| 4623 /* Include badflags: If the badword is onecap or allcap | |
| 4624 * use that for the goodword too. But if the badword is | |
| 4625 * allcap and it's only one char long use onecap. */ | |
| 4626 c = su->su_badflags; | |
| 4627 if ((c & WF_ALLCAP) | |
| 4628 #ifdef FEAT_MBYTE | |
| 4629 && su->su_badlen == (*mb_ptr2len)(su->su_badptr) | |
| 346 | 4630 #else |
| 625 | 4631 && su->su_badlen == 1 |
| 4632 #endif | |
| 4633 ) | |
| 4634 c = WF_ONECAP; | |
| 4635 c |= flags; | |
| 4636 | |
| 4637 /* When appending a compound word after a word character don't | |
| 4638 * use Onecap. */ | |
| 5477 | 4639 if (p != NULL && spell_iswordp_nmw(p, curwin)) |
| 625 | 4640 c &= ~WF_ONECAP; |
| 4641 make_case_word(tword + sp->ts_splitoff, | |
| 490 | 4642 preword + sp->ts_prewordlen, c); |
| 625 | 4643 } |
| 4644 | |
| 4645 if (!soundfold) | |
| 4646 { | |
| 323 | 4647 /* Don't use a banned word. It may appear again as a good |
| 4648 * word, thus remember it. */ | |
| 4649 if (flags & WF_BANNED) | |
| 4650 { | |
| 483 | 4651 add_banned(su, preword + sp->ts_prewordlen); |
| 323 | 4652 break; |
| 4653 } | |
| 534 | 4654 if ((sp->ts_complen == sp->ts_compsplit |
| 625 | 4655 && WAS_BANNED(su, preword + sp->ts_prewordlen)) |
| 4656 || WAS_BANNED(su, preword)) | |
| 534 | 4657 { |
| 4658 if (slang->sl_compprog == NULL) | |
| 4659 break; | |
| 4660 /* the word so far was banned but we may try compounding */ | |
| 4661 goodword_ends = FALSE; | |
| 4662 } | |
| 625 | 4663 } |
| 4664 | |
| 4665 newscore = 0; | |
| 4666 if (!soundfold) /* soundfold words don't have flags */ | |
| 4667 { | |
| 323 | 4668 if ((flags & WF_REGION) |
| 390 | 4669 && (((unsigned)flags >> 16) & lp->lp_region) == 0) |
| 323 | 4670 newscore += SCORE_REGION; |
| 4671 if (flags & WF_RARE) | |
| 4672 newscore += SCORE_RARE; | |
| 4673 | |
| 346 | 4674 if (!spell_valid_case(su->su_badflags, |
| 483 | 4675 captype(preword + sp->ts_prewordlen, NULL))) |
| 323 | 4676 newscore += SCORE_ICASE; |
| 625 | 4677 } |
| 4678 | |
| 4679 /* TODO: how about splitting in the soundfold tree? */ | |
| 4680 if (fword_ends | |
| 4681 && goodword_ends | |
| 4682 && sp->ts_fidx >= sp->ts_fidxtry | |
| 4683 && compound_ok) | |
| 4684 { | |
| 4685 /* The badword also ends: add suggestions. */ | |
| 4686 #ifdef DEBUG_TRIEWALK | |
| 4687 if (soundfold && STRCMP(preword, "smwrd") == 0) | |
| 4688 { | |
| 4689 int j; | |
| 4690 | |
| 4691 /* print the stack of changes that brought us here */ | |
| 4692 smsg("------ %s -------", fword); | |
| 4693 for (j = 0; j < depth; ++j) | |
| 4694 smsg("%s", changename[j]); | |
| 4695 } | |
| 4696 #endif | |
| 4697 if (soundfold) | |
| 4698 { | |
| 4699 /* For soundfolded words we need to find the original | |
| 1188 | 4700 * words, the edit distance and then add them. */ |
| 625 | 4701 add_sound_suggest(su, preword, sp->ts_score, lp); |
| 4702 } | |
|
2457
6c1d10a85398
Avoid illegal memory access in spell suggestion. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2455
diff
changeset
|
4703 else if (sp->ts_fidx > 0) |
| 625 | 4704 { |
| 4705 /* Give a penalty when changing non-word char to word | |
| 4706 * char, e.g., "thes," -> "these". */ | |
| 366 | 4707 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
|
4708 MB_PTR_BACK(fword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4709 if (!spell_iswordp(p, curwin)) |
| 366 | 4710 { |
| 4711 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
|
4712 MB_PTR_BACK(preword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4713 if (spell_iswordp(p, curwin)) |
| 366 | 4714 newscore += SCORE_NONWORD; |
| 4715 } | |
| 4716 | |
| 625 | 4717 /* Give a bonus to words seen before. */ |
| 4718 score = score_wordcount_adj(slang, | |
| 4719 sp->ts_score + newscore, | |
| 4720 preword + sp->ts_prewordlen, | |
| 4721 sp->ts_prewordlen > 0); | |
| 4722 | |
| 4723 /* Add the suggestion if the score isn't too bad. */ | |
| 4724 if (score <= su->su_maxscore) | |
| 633 | 4725 { |
| 625 | 4726 add_suggestion(su, &su->su_ga, preword, |
| 4727 sp->ts_fidx - repextra, | |
| 4728 score, 0, FALSE, lp->lp_sallang, FALSE); | |
| 633 | 4729 |
| 4730 if (su->su_badflags & WF_MIXCAP) | |
| 4731 { | |
| 4732 /* We really don't know if the word should be | |
| 4733 * upper or lower case, add both. */ | |
| 4734 c = captype(preword, NULL); | |
| 4735 if (c == 0 || c == WF_ALLCAP) | |
| 4736 { | |
| 4737 make_case_word(tword + sp->ts_splitoff, | |
| 4738 preword + sp->ts_prewordlen, | |
| 4739 c == 0 ? WF_ALLCAP : 0); | |
| 4740 | |
| 4741 add_suggestion(su, &su->su_ga, preword, | |
| 4742 sp->ts_fidx - repextra, | |
| 4743 score + SCORE_ICASE, 0, FALSE, | |
| 4744 lp->lp_sallang, FALSE); | |
| 4745 } | |
| 4746 } | |
| 4747 } | |
| 625 | 4748 } |
| 4749 } | |
| 4750 | |
| 4751 /* | |
| 4752 * Try word split and/or compounding. | |
| 4753 */ | |
| 4754 if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends) | |
| 4755 #ifdef FEAT_MBYTE | |
| 4756 /* Don't split halfway a character. */ | |
| 4757 && (!has_mbyte || sp->ts_tcharlen == 0) | |
| 4758 #endif | |
| 4759 ) | |
| 4760 { | |
| 4761 int try_compound; | |
| 4762 int try_split; | |
| 4763 | |
| 4764 /* If past the end of the bad word don't try a split. | |
| 4765 * Otherwise try changing the next word. E.g., find | |
| 4766 * suggestions for "the the" where the second "the" is | |
| 4767 * different. It's done like a split. | |
| 4768 * TODO: word split for soundfold words */ | |
| 4769 try_split = (sp->ts_fidx - repextra < su->su_badlen) | |
| 4770 && !soundfold; | |
| 4771 | |
| 4772 /* Get here in several situations: | |
| 4773 * 1. The word in the tree ends: | |
| 4774 * If the word allows compounding try that. Otherwise try | |
| 4775 * a split by inserting a space. For both check that a | |
| 4776 * valid words starts at fword[sp->ts_fidx]. | |
| 4777 * For NOBREAK do like compounding to be able to check if | |
| 4778 * the next word is valid. | |
| 4779 * 2. The badword does end, but it was due to a change (e.g., | |
| 4780 * a swap). No need to split, but do check that the | |
| 4781 * following word is valid. | |
| 4782 * 3. The badword and the word in the tree end. It may still | |
| 4783 * be possible to compound another (short) word. | |
| 4784 */ | |
| 4785 try_compound = FALSE; | |
| 4786 if (!soundfold | |
|
7526
08c1f73efcde
commit https://github.com/vim/vim/commit/7b877b360532713dc21a0ff3d55a76ac02eaf573
Christian Brabandt <cb@256bit.org>
parents:
7509
diff
changeset
|
4787 && !slang->sl_nocompoundsugs |
| 625 | 4788 && slang->sl_compprog != NULL |
| 4789 && ((unsigned)flags >> 24) != 0 | |
| 4790 && sp->ts_twordlen - sp->ts_splitoff | |
| 4791 >= slang->sl_compminlen | |
| 4792 #ifdef FEAT_MBYTE | |
| 4793 && (!has_mbyte | |
| 4794 || slang->sl_compminlen == 0 | |
| 4795 || mb_charlen(tword + sp->ts_splitoff) | |
| 4796 >= slang->sl_compminlen) | |
| 4797 #endif | |
| 4798 && (slang->sl_compsylmax < MAXWLEN | |
| 4799 || sp->ts_complen + 1 - sp->ts_compsplit | |
| 4800 < slang->sl_compmax) | |
| 1762 | 4801 && (can_be_compound(sp, slang, |
| 4802 compflags, ((unsigned)flags >> 24)))) | |
| 4803 | |
| 625 | 4804 { |
| 4805 try_compound = TRUE; | |
| 4806 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 4807 compflags[sp->ts_complen + 1] = NUL; | |
| 4808 } | |
| 4809 | |
| 4810 /* For NOBREAK we never try splitting, it won't make any word | |
| 4811 * valid. */ | |
|
7526
08c1f73efcde
commit https://github.com/vim/vim/commit/7b877b360532713dc21a0ff3d55a76ac02eaf573
Christian Brabandt <cb@256bit.org>
parents:
7509
diff
changeset
|
4812 if (slang->sl_nobreak && !slang->sl_nocompoundsugs) |
| 625 | 4813 try_compound = TRUE; |
| 4814 | |
| 4815 /* If we could add a compound word, and it's also possible to | |
| 4816 * split at this point, do the split first and set | |
| 4817 * TSF_DIDSPLIT to avoid doing it again. */ | |
| 4818 else if (!fword_ends | |
| 4819 && try_compound | |
| 4820 && (sp->ts_flags & TSF_DIDSPLIT) == 0) | |
| 4821 { | |
| 485 | 4822 try_compound = FALSE; |
| 625 | 4823 sp->ts_flags |= TSF_DIDSPLIT; |
| 4824 --sp->ts_curi; /* do the same NUL again */ | |
| 4825 compflags[sp->ts_complen] = NUL; | |
| 4826 } | |
| 4827 else | |
| 4828 sp->ts_flags &= ~TSF_DIDSPLIT; | |
| 4829 | |
| 4830 if (try_split || try_compound) | |
| 4831 { | |
| 534 | 4832 if (!try_compound && (!fword_ends || !goodword_ends)) |
| 485 | 4833 { |
| 4834 /* If we're going to split need to check that the | |
| 501 | 4835 * words so far are valid for compounding. If there |
| 4836 * is only one word it must not have the NEEDCOMPOUND | |
| 4837 * flag. */ | |
| 4838 if (sp->ts_complen == sp->ts_compsplit | |
| 4839 && (flags & WF_NEEDCOMP)) | |
| 4840 break; | |
| 490 | 4841 p = preword; |
| 4842 while (*skiptowhite(p) != NUL) | |
| 4843 p = skipwhite(skiptowhite(p)); | |
| 485 | 4844 if (sp->ts_complen > sp->ts_compsplit |
| 490 | 4845 && !can_compound(slang, p, |
| 485 | 4846 compflags + sp->ts_compsplit)) |
| 4847 break; | |
| 698 | 4848 |
| 4849 if (slang->sl_nosplitsugs) | |
| 4850 newscore += SCORE_SPLIT_NO; | |
| 4851 else | |
| 4852 newscore += SCORE_SPLIT; | |
| 625 | 4853 |
| 4854 /* Give a bonus to words seen before. */ | |
| 4855 newscore = score_wordcount_adj(slang, newscore, | |
| 4856 preword + sp->ts_prewordlen, TRUE); | |
| 482 | 4857 } |
| 4858 | |
| 625 | 4859 if (TRY_DEEPER(su, stack, depth, newscore)) |
| 323 | 4860 { |
| 625 | 4861 go_deeper(stack, depth, newscore); |
| 4862 #ifdef DEBUG_TRIEWALK | |
| 4863 if (!try_compound && !fword_ends) | |
| 4864 sprintf(changename[depth], "%.*s-%s: split", | |
| 4865 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 4866 else | |
| 4867 sprintf(changename[depth], "%.*s-%s: compound", | |
| 4868 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 4869 #endif | |
| 323 | 4870 /* Save things to be restored at STATE_SPLITUNDO. */ |
| 346 | 4871 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
|
4872 PROF_STORE(sp->ts_state) |
| 376 | 4873 sp->ts_state = STATE_SPLITUNDO; |
| 4874 | |
| 4875 ++depth; | |
| 4876 sp = &stack[depth]; | |
| 323 | 4877 |
| 482 | 4878 /* Append a space to preword when splitting. */ |
| 4879 if (!try_compound && !fword_ends) | |
| 4880 STRCAT(preword, " "); | |
| 835 | 4881 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 483 | 4882 sp->ts_splitoff = sp->ts_twordlen; |
| 492 | 4883 sp->ts_splitfidx = sp->ts_fidx; |
| 376 | 4884 |
| 4885 /* If the badword has a non-word character at this | |
| 4886 * position skip it. That means replacing the | |
| 482 | 4887 * non-word character with a space. Always skip a |
| 534 | 4888 * character when the word ends. But only when the |
| 4889 * good word can end. */ | |
| 625 | 4890 if (((!try_compound && !spell_iswordp_nmw(fword |
| 5477 | 4891 + sp->ts_fidx, |
| 4892 curwin)) | |
| 625 | 4893 || fword_ends) |
| 4894 && fword[sp->ts_fidx] != NUL | |
| 4895 && goodword_ends) | |
| 376 | 4896 { |
| 482 | 4897 int l; |
| 4898 | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
4899 l = MB_PTR2LEN(fword + sp->ts_fidx); |
| 482 | 4900 if (fword_ends) |
| 4901 { | |
| 4902 /* Copy the skipped character to preword. */ | |
| 483 | 4903 mch_memmove(preword + sp->ts_prewordlen, |
| 482 | 4904 fword + sp->ts_fidx, l); |
| 483 | 4905 sp->ts_prewordlen += l; |
| 4906 preword[sp->ts_prewordlen] = NUL; | |
| 482 | 4907 } |
| 4908 else | |
| 4909 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST; | |
| 4910 sp->ts_fidx += l; | |
| 376 | 4911 } |
| 455 | 4912 |
| 485 | 4913 /* When compounding include compound flag in |
| 4914 * compflags[] (already set above). When splitting we | |
| 4915 * may start compounding over again. */ | |
| 482 | 4916 if (try_compound) |
| 483 | 4917 ++sp->ts_complen; |
| 482 | 4918 else |
| 485 | 4919 sp->ts_compsplit = sp->ts_complen; |
| 4920 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 482 | 4921 |
| 455 | 4922 /* set su->su_badflags to the caps type at this |
| 4923 * position */ | |
| 324 | 4924 #ifdef FEAT_MBYTE |
| 4925 if (has_mbyte) | |
| 455 | 4926 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); |
| 324 | 4927 else |
| 4928 #endif | |
| 455 | 4929 n = sp->ts_fidx; |
| 474 | 4930 su->su_badflags = badword_captype(su->su_badptr + n, |
| 455 | 4931 su->su_badptr + su->su_badlen); |
| 323 | 4932 |
| 4933 /* Restart at top of the tree. */ | |
| 376 | 4934 sp->ts_arridx = 0; |
| 485 | 4935 |
| 4936 /* If there are postponed prefixes, try these too. */ | |
| 4937 if (pbyts != NULL) | |
| 4938 { | |
| 4939 byts = pbyts; | |
| 4940 idxs = pidxs; | |
| 4941 sp->ts_prefixdepth = PFD_PREFIXTREE; | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4942 PROF_STORE(sp->ts_state) |
| 485 | 4943 sp->ts_state = STATE_NOPREFIX; |
| 4944 } | |
| 323 | 4945 } |
| 4946 } | |
| 625 | 4947 } |
| 4948 break; | |
| 4949 | |
| 4950 case STATE_SPLITUNDO: | |
| 4951 /* Undo the changes done for word split or compound word. */ | |
| 4952 su->su_badflags = sp->ts_save_badflags; | |
| 4953 | |
| 4954 /* Continue looking for NUL bytes. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4955 PROF_STORE(sp->ts_state) |
| 625 | 4956 sp->ts_state = STATE_START; |
| 4957 | |
| 4958 /* In case we went into the prefix tree. */ | |
| 4959 byts = fbyts; | |
| 4960 idxs = fidxs; | |
| 4961 break; | |
| 4962 | |
| 4963 case STATE_ENDNUL: | |
| 4964 /* Past the NUL bytes in the node. */ | |
| 4965 su->su_badflags = sp->ts_save_badflags; | |
| 4966 if (fword[sp->ts_fidx] == NUL | |
| 4967 #ifdef FEAT_MBYTE | |
| 4968 && sp->ts_tcharlen == 0 | |
| 4969 #endif | |
| 4970 ) | |
| 4971 { | |
| 4972 /* 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
|
4973 PROF_STORE(sp->ts_state) |
| 625 | 4974 sp->ts_state = STATE_DEL; |
| 323 | 4975 break; |
| 625 | 4976 } |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4977 PROF_STORE(sp->ts_state) |
| 625 | 4978 sp->ts_state = STATE_PLAIN; |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
4979 /* FALLTHROUGH */ |
| 625 | 4980 |
| 4981 case STATE_PLAIN: | |
| 4982 /* | |
| 4983 * Go over all possible bytes at this node, add each to tword[] | |
| 4984 * and use child node. "ts_curi" is the index. | |
| 4985 */ | |
| 4986 arridx = sp->ts_arridx; | |
| 4987 if (sp->ts_curi > byts[arridx]) | |
| 4988 { | |
| 4989 /* Done all bytes at this node, do next state. When still at | |
| 4990 * 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
|
4991 PROF_STORE(sp->ts_state) |
| 625 | 4992 if (sp->ts_fidx >= sp->ts_fidxtry) |
| 323 | 4993 sp->ts_state = STATE_DEL; |
| 4994 else | |
| 625 | 4995 sp->ts_state = STATE_FINAL; |
| 4996 } | |
| 4997 else | |
| 4998 { | |
| 4999 arridx += sp->ts_curi++; | |
| 5000 c = byts[arridx]; | |
| 5001 | |
| 5002 /* Normal byte, go one level deeper. If it's not equal to the | |
| 5003 * byte in the bad word adjust the score. But don't even try | |
| 5004 * when the byte was already changed. And don't try when we | |
| 5695 | 5005 * just deleted this byte, accepting it is always cheaper than |
| 625 | 5006 * delete + substitute. */ |
| 5007 if (c == fword[sp->ts_fidx] | |
| 5008 #ifdef FEAT_MBYTE | |
| 5009 || (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE) | |
| 5010 #endif | |
| 5011 ) | |
| 5012 newscore = 0; | |
| 5013 else | |
| 5014 newscore = SCORE_SUBST; | |
| 5015 if ((newscore == 0 | |
| 5016 || (sp->ts_fidx >= sp->ts_fidxtry | |
| 5017 && ((sp->ts_flags & TSF_DIDDEL) == 0 | |
| 5018 || c != fword[sp->ts_delidx]))) | |
| 5019 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5020 { | |
| 5021 go_deeper(stack, depth, newscore); | |
| 5022 #ifdef DEBUG_TRIEWALK | |
| 5023 if (newscore > 0) | |
| 5024 sprintf(changename[depth], "%.*s-%s: subst %c to %c", | |
| 5025 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5026 fword[sp->ts_fidx], c); | |
| 323 | 5027 else |
| 625 | 5028 sprintf(changename[depth], "%.*s-%s: accept %c", |
| 5029 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5030 fword[sp->ts_fidx]); | |
| 5031 #endif | |
| 323 | 5032 ++depth; |
| 625 | 5033 sp = &stack[depth]; |
| 5034 ++sp->ts_fidx; | |
| 5035 tword[sp->ts_twordlen++] = c; | |
| 5036 sp->ts_arridx = idxs[arridx]; | |
| 5037 #ifdef FEAT_MBYTE | |
| 5038 if (newscore == SCORE_SUBST) | |
| 5039 sp->ts_isdiff = DIFF_YES; | |
| 330 | 5040 if (has_mbyte) |
| 358 | 5041 { |
| 625 | 5042 /* Multi-byte characters are a bit complicated to |
| 5043 * handle: They differ when any of the bytes differ | |
| 5044 * and then their length may also differ. */ | |
| 5045 if (sp->ts_tcharlen == 0) | |
| 5046 { | |
| 5047 /* First byte. */ | |
| 5048 sp->ts_tcharidx = 0; | |
| 5049 sp->ts_tcharlen = MB_BYTE2LEN(c); | |
| 5050 sp->ts_fcharstart = sp->ts_fidx - 1; | |
| 5051 sp->ts_isdiff = (newscore != 0) | |
| 5052 ? DIFF_YES : DIFF_NONE; | |
| 5053 } | |
| 5054 else if (sp->ts_isdiff == DIFF_INSERT) | |
| 5055 /* When inserting trail bytes don't advance in the | |
| 5056 * bad word. */ | |
| 5057 --sp->ts_fidx; | |
| 5058 if (++sp->ts_tcharidx == sp->ts_tcharlen) | |
| 5059 { | |
| 5060 /* Last byte of character. */ | |
| 5061 if (sp->ts_isdiff == DIFF_YES) | |
| 5062 { | |
| 5063 /* Correct ts_fidx for the byte length of the | |
| 5064 * character (we didn't check that before). */ | |
| 5065 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
|
5066 + MB_PTR2LEN( |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5067 fword + sp->ts_fcharstart); |
| 625 | 5068 /* For changing a composing character adjust |
| 5069 * the score from SCORE_SUBST to | |
| 5070 * SCORE_SUBCOMP. */ | |
| 5071 if (enc_utf8 | |
| 5072 && 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
|
5073 utf_ptr2char(tword |
| 625 | 5074 + sp->ts_twordlen |
| 5075 - sp->ts_tcharlen)) | |
| 5076 && 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
|
5077 utf_ptr2char(fword |
| 625 | 5078 + sp->ts_fcharstart))) |
| 5079 sp->ts_score -= | |
| 5080 SCORE_SUBST - SCORE_SUBCOMP; | |
| 5081 | |
| 5082 /* For a similar character adjust score from | |
| 5083 * SCORE_SUBST to SCORE_SIMILAR. */ | |
| 5084 else if (!soundfold | |
| 5085 && slang->sl_has_map | |
| 5086 && similar_chars(slang, | |
| 5087 mb_ptr2char(tword | |
| 5088 + sp->ts_twordlen | |
| 5089 - sp->ts_tcharlen), | |
| 5090 mb_ptr2char(fword | |
| 5091 + sp->ts_fcharstart))) | |
| 5092 sp->ts_score -= | |
| 5093 SCORE_SUBST - SCORE_SIMILAR; | |
| 5094 } | |
| 5095 else if (sp->ts_isdiff == DIFF_INSERT | |
| 5096 && sp->ts_twordlen > sp->ts_tcharlen) | |
| 5097 { | |
| 5098 p = tword + sp->ts_twordlen - sp->ts_tcharlen; | |
| 5099 c = mb_ptr2char(p); | |
| 5100 if (enc_utf8 && utf_iscomposing(c)) | |
| 5101 { | |
| 5102 /* Inserting a composing char doesn't | |
| 5103 * count that much. */ | |
| 5104 sp->ts_score -= SCORE_INS - SCORE_INSCOMP; | |
| 5105 } | |
| 5106 else | |
| 5107 { | |
| 5108 /* If the previous character was the same, | |
| 5109 * thus doubling a character, give a bonus | |
| 5110 * to the score. Also for the soundfold | |
| 5111 * tree (might seem illogical but does | |
| 5112 * give better scores). */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5113 MB_PTR_BACK(tword, p); |
| 625 | 5114 if (c == mb_ptr2char(p)) |
| 5115 sp->ts_score -= SCORE_INS | |
| 5116 - SCORE_INSDUP; | |
| 5117 } | |
| 5118 } | |
| 5119 | |
| 5120 /* Starting a new char, reset the length. */ | |
| 5121 sp->ts_tcharlen = 0; | |
| 5122 } | |
| 358 | 5123 } |
| 330 | 5124 else |
| 5125 #endif | |
| 358 | 5126 { |
| 625 | 5127 /* If we found a similar char adjust the score. |
| 5128 * We do this after calling go_deeper() because | |
| 5129 * it's slow. */ | |
| 5130 if (newscore != 0 | |
| 5131 && !soundfold | |
| 5132 && slang->sl_has_map | |
| 5133 && similar_chars(slang, | |
| 5134 c, fword[sp->ts_fidx - 1])) | |
| 5135 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR; | |
| 358 | 5136 } |
| 625 | 5137 } |
| 5138 } | |
| 5139 break; | |
| 5140 | |
| 5141 case STATE_DEL: | |
| 5142 #ifdef FEAT_MBYTE | |
| 5143 /* When past the first byte of a multi-byte char don't try | |
| 5144 * delete/insert/swap a character. */ | |
| 5145 if (has_mbyte && sp->ts_tcharlen > 0) | |
| 5146 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5147 PROF_STORE(sp->ts_state) |
| 625 | 5148 sp->ts_state = STATE_FINAL; |
| 5149 break; | |
| 5150 } | |
| 5151 #endif | |
| 5152 /* | |
| 5153 * Try skipping one character in the bad word (delete it). | |
| 5154 */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5155 PROF_STORE(sp->ts_state) |
| 625 | 5156 sp->ts_state = STATE_INS_PREP; |
| 5157 sp->ts_curi = 1; | |
| 5158 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*') | |
| 5159 /* Deleting a vowel at the start of a word counts less, see | |
| 5160 * soundalike_score(). */ | |
| 5161 newscore = 2 * SCORE_DEL / 3; | |
| 5162 else | |
| 5163 newscore = SCORE_DEL; | |
| 5164 if (fword[sp->ts_fidx] != NUL | |
| 5165 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5166 { | |
| 5167 go_deeper(stack, depth, newscore); | |
| 5168 #ifdef DEBUG_TRIEWALK | |
| 5169 sprintf(changename[depth], "%.*s-%s: delete %c", | |
| 5170 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5171 fword[sp->ts_fidx]); | |
| 5172 #endif | |
| 5173 ++depth; | |
| 5174 | |
| 5175 /* Remember what character we deleted, so that we can avoid | |
| 5176 * inserting it again. */ | |
| 5177 stack[depth].ts_flags |= TSF_DIDDEL; | |
| 5178 stack[depth].ts_delidx = sp->ts_fidx; | |
| 5179 | |
| 5180 /* Advance over the character in fword[]. Give a bonus to the | |
| 5181 * score if the same character is following "nn" -> "n". It's | |
| 5182 * a bit illogical for soundfold tree but it does give better | |
| 5183 * results. */ | |
| 5184 #ifdef FEAT_MBYTE | |
| 5185 if (has_mbyte) | |
| 5186 { | |
| 5187 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
|
5188 stack[depth].ts_fidx += MB_PTR2LEN(fword + sp->ts_fidx); |
| 625 | 5189 if (enc_utf8 && utf_iscomposing(c)) |
| 5190 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP; | |
| 5191 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx)) | |
| 5192 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 5193 } |
| 5194 else | |
| 625 | 5195 #endif |
| 5196 { | |
| 5197 ++stack[depth].ts_fidx; | |
| 5198 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1]) | |
| 5199 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 5200 } |
| 5201 break; | |
| 625 | 5202 } |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5203 /* FALLTHROUGH */ |
| 625 | 5204 |
| 5205 case STATE_INS_PREP: | |
| 5206 if (sp->ts_flags & TSF_DIDDEL) | |
| 5207 { | |
| 5208 /* If we just deleted a byte then inserting won't make sense, | |
| 5209 * a substitute is always cheaper. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5210 PROF_STORE(sp->ts_state) |
| 625 | 5211 sp->ts_state = STATE_SWAP; |
| 5212 break; | |
| 5213 } | |
| 5214 | |
| 5215 /* skip over NUL bytes */ | |
| 5216 n = sp->ts_arridx; | |
| 5217 for (;;) | |
| 5218 { | |
| 5219 if (sp->ts_curi > byts[n]) | |
| 5220 { | |
| 5221 /* 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
|
5222 PROF_STORE(sp->ts_state) |
| 625 | 5223 sp->ts_state = STATE_SWAP; |
| 5224 break; | |
| 5225 } | |
| 5226 if (byts[n + sp->ts_curi] != NUL) | |
| 5227 { | |
| 5228 /* Found a byte to insert. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5229 PROF_STORE(sp->ts_state) |
| 625 | 5230 sp->ts_state = STATE_INS; |
| 330 | 5231 break; |
| 5232 } | |
| 625 | 5233 ++sp->ts_curi; |
| 5234 } | |
| 5235 break; | |
| 5236 | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5237 /* FALLTHROUGH */ |
| 625 | 5238 |
| 5239 case STATE_INS: | |
| 5240 /* Insert one byte. Repeat this for each possible byte at this | |
| 5241 * node. */ | |
| 5242 n = sp->ts_arridx; | |
| 5243 if (sp->ts_curi > byts[n]) | |
| 5244 { | |
| 5245 /* 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
|
5246 PROF_STORE(sp->ts_state) |
| 625 | 5247 sp->ts_state = STATE_SWAP; |
| 5248 break; | |
| 5249 } | |
| 5250 | |
| 5251 /* Do one more byte at this node, but: | |
| 5252 * - Skip NUL bytes. | |
| 5253 * - Skip the byte if it's equal to the byte in the word, | |
| 5254 * accepting that byte is always better. | |
| 5255 */ | |
| 5256 n += sp->ts_curi++; | |
| 5257 c = byts[n]; | |
| 5258 if (soundfold && sp->ts_twordlen == 0 && c == '*') | |
| 5259 /* Inserting a vowel at the start of a word counts less, | |
| 5260 * see soundalike_score(). */ | |
| 5261 newscore = 2 * SCORE_INS / 3; | |
| 5262 else | |
| 5263 newscore = SCORE_INS; | |
| 5264 if (c != fword[sp->ts_fidx] | |
| 5265 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5266 { | |
| 5267 go_deeper(stack, depth, newscore); | |
| 5268 #ifdef DEBUG_TRIEWALK | |
| 5269 sprintf(changename[depth], "%.*s-%s: insert %c", | |
| 5270 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5271 c); | |
| 5272 #endif | |
| 5273 ++depth; | |
| 5274 sp = &stack[depth]; | |
| 5275 tword[sp->ts_twordlen++] = c; | |
| 5276 sp->ts_arridx = idxs[n]; | |
| 330 | 5277 #ifdef FEAT_MBYTE |
| 5278 if (has_mbyte) | |
| 5279 { | |
| 625 | 5280 fl = MB_BYTE2LEN(c); |
| 5281 if (fl > 1) | |
| 330 | 5282 { |
| 625 | 5283 /* There are following bytes for the same character. |
| 5284 * We must find all bytes before trying | |
| 5285 * delete/insert/swap/etc. */ | |
| 5286 sp->ts_tcharlen = fl; | |
| 5287 sp->ts_tcharidx = 1; | |
| 5288 sp->ts_isdiff = DIFF_INSERT; | |
| 330 | 5289 } |
| 323 | 5290 } |
| 5291 else | |
| 625 | 5292 fl = 1; |
| 5293 if (fl == 1) | |
| 5294 #endif | |
| 5295 { | |
| 5296 /* If the previous character was the same, thus doubling a | |
| 5297 * character, give a bonus to the score. Also for | |
| 5298 * soundfold words (illogical but does give a better | |
| 5299 * score). */ | |
| 5300 if (sp->ts_twordlen >= 2 | |
| 5301 && tword[sp->ts_twordlen - 2] == c) | |
| 5302 sp->ts_score -= SCORE_INS - SCORE_INSDUP; | |
| 5303 } | |
| 5304 } | |
| 5305 break; | |
| 5306 | |
| 5307 case STATE_SWAP: | |
| 5308 /* | |
| 5309 * Swap two bytes in the bad word: "12" -> "21". | |
| 5310 * We change "fword" here, it's changed back afterwards at | |
| 5311 * STATE_UNSWAP. | |
| 5312 */ | |
| 5313 p = fword + sp->ts_fidx; | |
| 5314 c = *p; | |
| 5315 if (c == NUL) | |
| 5316 { | |
| 5317 /* 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
|
5318 PROF_STORE(sp->ts_state) |
| 625 | 5319 sp->ts_state = STATE_FINAL; |
| 5320 break; | |
| 5321 } | |
| 5322 | |
| 5323 /* Don't swap if the first character is not a word character. | |
| 5324 * 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
|
5325 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 5326 { |
|
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_REP_INI; |
| 323 | 5329 break; |
| 625 | 5330 } |
| 5331 | |
| 5332 #ifdef FEAT_MBYTE | |
| 5333 if (has_mbyte) | |
| 5334 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5335 n = MB_CPTR2LEN(p); |
| 625 | 5336 c = mb_ptr2char(p); |
| 1337 | 5337 if (p[n] == NUL) |
| 5338 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5339 else if (!soundfold && !spell_iswordp(p + n, curwin)) |
| 625 | 5340 c2 = c; /* don't swap non-word char */ |
| 5341 else | |
| 5342 c2 = mb_ptr2char(p + n); | |
| 5343 } | |
| 5344 else | |
| 5345 #endif | |
| 5346 { | |
| 1337 | 5347 if (p[1] == NUL) |
| 5348 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5349 else if (!soundfold && !spell_iswordp(p + 1, curwin)) |
| 625 | 5350 c2 = c; /* don't swap non-word char */ |
| 5351 else | |
| 5352 c2 = p[1]; | |
| 5353 } | |
| 5354 | |
| 1337 | 5355 /* When the second character is NUL we can't swap. */ |
| 5356 if (c2 == NUL) | |
| 5357 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5358 PROF_STORE(sp->ts_state) |
| 1337 | 5359 sp->ts_state = STATE_REP_INI; |
| 5360 break; | |
| 5361 } | |
| 5362 | |
| 625 | 5363 /* When characters are identical, swap won't do anything. |
| 5364 * Also get here if the second char is not a word character. */ | |
| 5365 if (c == c2) | |
| 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) |
| 625 | 5368 sp->ts_state = STATE_SWAP3; |
| 5369 break; | |
| 5370 } | |
| 5371 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP)) | |
| 5372 { | |
| 5373 go_deeper(stack, depth, SCORE_SWAP); | |
| 5374 #ifdef DEBUG_TRIEWALK | |
| 5375 sprintf(changename[depth], "%.*s-%s: swap %c and %c", | |
| 5376 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5377 c, c2); | |
| 5378 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5379 PROF_STORE(sp->ts_state) |
| 625 | 5380 sp->ts_state = STATE_UNSWAP; |
| 5381 ++depth; | |
| 330 | 5382 #ifdef FEAT_MBYTE |
| 5383 if (has_mbyte) | |
| 5384 { | |
| 625 | 5385 fl = mb_char2len(c2); |
| 5386 mch_memmove(p, p + n, fl); | |
| 5387 mb_char2bytes(c, p + fl); | |
| 5388 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 5389 } |
| 5390 else | |
| 5391 #endif | |
| 5392 { | |
| 625 | 5393 p[0] = c2; |
| 330 | 5394 p[1] = c; |
| 625 | 5395 stack[depth].ts_fidxtry = sp->ts_fidx + 2; |
| 5396 } | |
| 5397 } | |
| 5398 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5399 { |
| 625 | 5400 /* 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
|
5401 PROF_STORE(sp->ts_state) |
| 625 | 5402 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
|
5403 } |
| 625 | 5404 break; |
| 5405 | |
| 5406 case STATE_UNSWAP: | |
| 5407 /* Undo the STATE_SWAP swap: "21" -> "12". */ | |
| 5408 p = fword + sp->ts_fidx; | |
| 5409 #ifdef FEAT_MBYTE | |
| 5410 if (has_mbyte) | |
| 5411 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5412 n = MB_PTR2LEN(p); |
| 625 | 5413 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
|
5414 mch_memmove(p + MB_PTR2LEN(p + n), p, n); |
| 625 | 5415 mb_char2bytes(c, p); |
| 5416 } | |
| 5417 else | |
| 5418 #endif | |
| 5419 { | |
| 5420 c = *p; | |
| 5421 *p = p[1]; | |
| 5422 p[1] = c; | |
| 5423 } | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5424 /* FALLTHROUGH */ |
| 625 | 5425 |
| 5426 case STATE_SWAP3: | |
| 5427 /* Swap two bytes, skipping one: "123" -> "321". We change | |
| 5428 * "fword" here, it's changed back afterwards at STATE_UNSWAP3. */ | |
| 5429 p = fword + sp->ts_fidx; | |
| 5430 #ifdef FEAT_MBYTE | |
| 5431 if (has_mbyte) | |
| 5432 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5433 n = MB_CPTR2LEN(p); |
| 625 | 5434 c = mb_ptr2char(p); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5435 fl = MB_CPTR2LEN(p + n); |
| 625 | 5436 c2 = mb_ptr2char(p + n); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5437 if (!soundfold && !spell_iswordp(p + n + fl, curwin)) |
| 625 | 5438 c3 = c; /* don't swap non-word char */ |
| 5439 else | |
| 5440 c3 = mb_ptr2char(p + n + fl); | |
| 5441 } | |
| 5442 else | |
| 5443 #endif | |
| 5444 { | |
| 5445 c = *p; | |
| 5446 c2 = p[1]; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5447 if (!soundfold && !spell_iswordp(p + 2, curwin)) |
| 625 | 5448 c3 = c; /* don't swap non-word char */ |
| 5449 else | |
| 5450 c3 = p[2]; | |
| 5451 } | |
| 5452 | |
| 5453 /* When characters are identical: "121" then SWAP3 result is | |
| 5454 * identical, ROT3L result is same as SWAP: "211", ROT3L result is | |
| 5455 * same as SWAP on next char: "112". Thus skip all swapping. | |
| 5456 * Also skip when c3 is NUL. | |
| 5457 * Also get here when the third character is not a word character. | |
| 5458 * Second character may any char: "a.b" -> "b.a" */ | |
| 5459 if (c == c3 || c3 == NUL) | |
| 5460 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5461 PROF_STORE(sp->ts_state) |
| 625 | 5462 sp->ts_state = STATE_REP_INI; |
| 5463 break; | |
| 5464 } | |
| 5465 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5466 { | |
| 5467 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5468 #ifdef DEBUG_TRIEWALK | |
| 5469 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c", | |
| 5470 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5471 c, c3); | |
| 5472 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5473 PROF_STORE(sp->ts_state) |
| 625 | 5474 sp->ts_state = STATE_UNSWAP3; |
| 5475 ++depth; | |
| 5476 #ifdef FEAT_MBYTE | |
| 5477 if (has_mbyte) | |
| 5478 { | |
| 5479 tl = mb_char2len(c3); | |
| 5480 mch_memmove(p, p + n + fl, tl); | |
| 5481 mb_char2bytes(c2, p + tl); | |
| 5482 mb_char2bytes(c, p + fl + tl); | |
| 5483 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl; | |
| 5484 } | |
| 5485 else | |
| 5486 #endif | |
| 5487 { | |
| 5488 p[0] = p[2]; | |
| 5489 p[2] = c; | |
| 5490 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 5491 } | |
| 5492 } | |
| 5493 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5494 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5495 PROF_STORE(sp->ts_state) |
| 625 | 5496 sp->ts_state = STATE_REP_INI; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5497 } |
| 625 | 5498 break; |
| 5499 | |
| 5500 case STATE_UNSWAP3: | |
| 5501 /* Undo STATE_SWAP3: "321" -> "123" */ | |
| 5502 p = fword + sp->ts_fidx; | |
| 5503 #ifdef FEAT_MBYTE | |
| 5504 if (has_mbyte) | |
| 5505 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5506 n = MB_PTR2LEN(p); |
| 625 | 5507 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
|
5508 fl = MB_PTR2LEN(p + n); |
| 625 | 5509 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
|
5510 tl = MB_PTR2LEN(p + n + fl); |
| 625 | 5511 mch_memmove(p + fl + tl, p, n); |
| 5512 mb_char2bytes(c, p); | |
| 5513 mb_char2bytes(c2, p + tl); | |
| 5514 p = p + tl; | |
| 5515 } | |
| 5516 else | |
| 5517 #endif | |
| 5518 { | |
| 5519 c = *p; | |
| 5520 *p = p[2]; | |
| 5521 p[2] = c; | |
| 5522 ++p; | |
| 5523 } | |
| 5524 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5525 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 5526 { |
| 5527 /* Middle char is not a word char, skip the rotate. First and | |
| 5528 * 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
|
5529 PROF_STORE(sp->ts_state) |
| 625 | 5530 sp->ts_state = STATE_REP_INI; |
| 5531 break; | |
| 5532 } | |
| 5533 | |
| 5534 /* Rotate three characters left: "123" -> "231". We change | |
| 5535 * "fword" here, it's changed back afterwards at STATE_UNROT3L. */ | |
| 5536 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5537 { | |
| 5538 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5539 #ifdef DEBUG_TRIEWALK | |
| 5540 p = fword + sp->ts_fidx; | |
| 5541 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c", | |
| 5542 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5543 p[0], p[1], p[2]); | |
| 5544 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5545 PROF_STORE(sp->ts_state) |
| 625 | 5546 sp->ts_state = STATE_UNROT3L; |
| 5547 ++depth; | |
| 330 | 5548 p = fword + sp->ts_fidx; |
| 5549 #ifdef FEAT_MBYTE | |
| 5550 if (has_mbyte) | |
| 5551 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5552 n = MB_CPTR2LEN(p); |
| 330 | 5553 c = mb_ptr2char(p); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5554 fl = MB_CPTR2LEN(p + n); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5555 fl += MB_CPTR2LEN(p + n + fl); |
| 625 | 5556 mch_memmove(p, p + n, fl); |
| 5557 mb_char2bytes(c, p + fl); | |
| 5558 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 5559 } |
| 5560 else | |
| 5561 #endif | |
| 323 | 5562 { |
| 330 | 5563 c = *p; |
| 625 | 5564 *p = p[1]; |
| 5565 p[1] = p[2]; | |
| 5566 p[2] = c; | |
| 5567 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 5568 } | |
| 5569 } | |
| 5570 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5571 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5572 PROF_STORE(sp->ts_state) |
| 625 | 5573 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
|
5574 } |
| 625 | 5575 break; |
| 5576 | |
| 5577 case STATE_UNROT3L: | |
| 5578 /* Undo ROT3L: "231" -> "123" */ | |
| 5579 p = fword + sp->ts_fidx; | |
| 5580 #ifdef FEAT_MBYTE | |
| 5581 if (has_mbyte) | |
| 5582 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5583 n = MB_PTR2LEN(p); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5584 n += MB_PTR2LEN(p + n); |
| 625 | 5585 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
|
5586 tl = MB_PTR2LEN(p + n); |
| 625 | 5587 mch_memmove(p + tl, p, n); |
| 5588 mb_char2bytes(c, p); | |
| 5589 } | |
| 5590 else | |
| 5591 #endif | |
| 5592 { | |
| 5593 c = p[2]; | |
| 5594 p[2] = p[1]; | |
| 5595 p[1] = *p; | |
| 5596 *p = c; | |
| 5597 } | |
| 5598 | |
| 5599 /* Rotate three bytes right: "123" -> "312". We change "fword" | |
| 5600 * here, it's changed back afterwards at STATE_UNROT3R. */ | |
| 5601 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5602 { | |
| 5603 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5604 #ifdef DEBUG_TRIEWALK | |
| 5605 p = fword + sp->ts_fidx; | |
| 5606 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c", | |
| 5607 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5608 p[0], p[1], p[2]); | |
| 5609 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5610 PROF_STORE(sp->ts_state) |
| 625 | 5611 sp->ts_state = STATE_UNROT3R; |
| 5612 ++depth; | |
| 330 | 5613 p = fword + sp->ts_fidx; |
| 5614 #ifdef FEAT_MBYTE | |
| 5615 if (has_mbyte) | |
| 5616 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5617 n = MB_CPTR2LEN(p); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5618 n += MB_CPTR2LEN(p + n); |
| 330 | 5619 c = mb_ptr2char(p + n); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5620 tl = MB_CPTR2LEN(p + n); |
| 330 | 5621 mch_memmove(p + tl, p, n); |
| 5622 mb_char2bytes(c, p); | |
| 625 | 5623 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl; |
| 330 | 5624 } |
| 5625 else | |
| 5626 #endif | |
| 5627 { | |
| 5628 c = p[2]; | |
| 5629 p[2] = p[1]; | |
| 5630 p[1] = *p; | |
| 5631 *p = c; | |
| 625 | 5632 stack[depth].ts_fidxtry = sp->ts_fidx + 3; |
| 5633 } | |
| 5634 } | |
| 5635 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5636 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5637 PROF_STORE(sp->ts_state) |
| 625 | 5638 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
|
5639 } |
| 625 | 5640 break; |
| 5641 | |
| 5642 case STATE_UNROT3R: | |
| 5643 /* Undo ROT3R: "312" -> "123" */ | |
| 5644 p = fword + sp->ts_fidx; | |
| 5645 #ifdef FEAT_MBYTE | |
| 5646 if (has_mbyte) | |
| 5647 { | |
| 5648 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
|
5649 tl = MB_PTR2LEN(p); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5650 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
|
5651 n += MB_PTR2LEN(p + tl + n); |
| 625 | 5652 mch_memmove(p, p + tl, n); |
| 5653 mb_char2bytes(c, p + n); | |
| 5654 } | |
| 5655 else | |
| 5656 #endif | |
| 5657 { | |
| 5658 c = *p; | |
| 5659 *p = p[1]; | |
| 5660 p[1] = p[2]; | |
| 5661 p[2] = c; | |
| 5662 } | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5663 /* FALLTHROUGH */ |
| 625 | 5664 |
| 5665 case STATE_REP_INI: | |
| 5666 /* Check if matching with REP items from the .aff file would work. | |
| 5667 * Quickly skip if: | |
| 5668 * - there are no REP items and we are not in the soundfold trie | |
| 5669 * - the score is going to be too high anyway | |
| 5670 * - already applied a REP item or swapped here */ | |
| 5671 if ((lp->lp_replang == NULL && !soundfold) | |
| 5672 || sp->ts_score + SCORE_REP >= su->su_maxscore | |
| 5673 || sp->ts_fidx < sp->ts_fidxtry) | |
| 5674 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5675 PROF_STORE(sp->ts_state) |
| 625 | 5676 sp->ts_state = STATE_FINAL; |
| 323 | 5677 break; |
| 625 | 5678 } |
| 5679 | |
| 5680 /* Use the first byte to quickly find the first entry that may | |
| 5681 * match. If the index is -1 there is none. */ | |
| 5682 if (soundfold) | |
| 5683 sp->ts_curi = slang->sl_repsal_first[fword[sp->ts_fidx]]; | |
| 5684 else | |
| 5685 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]]; | |
| 5686 | |
| 5687 if (sp->ts_curi < 0) | |
| 5688 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5689 PROF_STORE(sp->ts_state) |
| 625 | 5690 sp->ts_state = STATE_FINAL; |
| 5691 break; | |
| 5692 } | |
| 5693 | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5694 PROF_STORE(sp->ts_state) |
| 625 | 5695 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
|
5696 /* FALLTHROUGH */ |
| 625 | 5697 |
| 5698 case STATE_REP: | |
| 5699 /* Try matching with REP items from the .aff file. For each match | |
| 5700 * replace the characters and check if the resulting word is | |
| 5701 * valid. */ | |
| 5702 p = fword + sp->ts_fidx; | |
| 5703 | |
| 5704 if (soundfold) | |
| 5705 gap = &slang->sl_repsal; | |
| 5706 else | |
| 503 | 5707 gap = &lp->lp_replang->sl_rep; |
| 625 | 5708 while (sp->ts_curi < gap->ga_len) |
| 5709 { | |
| 5710 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++; | |
| 5711 if (*ftp->ft_from != *p) | |
| 5712 { | |
| 5713 /* past possible matching entries */ | |
| 5714 sp->ts_curi = gap->ga_len; | |
| 323 | 5715 break; |
| 5716 } | |
| 625 | 5717 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0 |
| 5718 && TRY_DEEPER(su, stack, depth, SCORE_REP)) | |
| 5719 { | |
| 5720 go_deeper(stack, depth, SCORE_REP); | |
| 5721 #ifdef DEBUG_TRIEWALK | |
| 5722 sprintf(changename[depth], "%.*s-%s: replace %s with %s", | |
| 5723 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5724 ftp->ft_from, ftp->ft_to); | |
| 5725 #endif | |
| 5726 /* Need to undo this afterwards. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5727 PROF_STORE(sp->ts_state) |
| 625 | 5728 sp->ts_state = STATE_REP_UNDO; |
| 5729 | |
| 5730 /* Change the "from" to the "to" string. */ | |
| 5731 ++depth; | |
| 835 | 5732 fl = (int)STRLEN(ftp->ft_from); |
| 5733 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 5734 if (fl != tl) |
| 323 | 5735 { |
| 1619 | 5736 STRMOVE(p + tl, p + fl); |
| 625 | 5737 repextra += tl - fl; |
| 323 | 5738 } |
| 625 | 5739 mch_memmove(p, ftp->ft_to, tl); |
| 5740 stack[depth].ts_fidxtry = sp->ts_fidx + tl; | |
| 5741 #ifdef FEAT_MBYTE | |
| 5742 stack[depth].ts_tcharlen = 0; | |
| 5743 #endif | |
| 5744 break; | |
| 5745 } | |
| 5746 } | |
| 5747 | |
| 5748 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
|
5749 { |
| 625 | 5750 /* No (more) matches. */ |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5751 PROF_STORE(sp->ts_state) |
| 625 | 5752 sp->ts_state = STATE_FINAL; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5753 } |
| 625 | 5754 |
| 5755 break; | |
| 5756 | |
| 5757 case STATE_REP_UNDO: | |
| 5758 /* Undo a REP replacement and continue with the next one. */ | |
| 5759 if (soundfold) | |
| 5760 gap = &slang->sl_repsal; | |
| 5761 else | |
| 5762 gap = &lp->lp_replang->sl_rep; | |
| 5763 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1; | |
| 835 | 5764 fl = (int)STRLEN(ftp->ft_from); |
| 5765 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 5766 p = fword + sp->ts_fidx; |
| 5767 if (fl != tl) | |
| 5768 { | |
| 1619 | 5769 STRMOVE(p + fl, p + tl); |
| 625 | 5770 repextra -= tl - fl; |
| 5771 } | |
| 5772 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
|
5773 PROF_STORE(sp->ts_state) |
| 625 | 5774 sp->ts_state = STATE_REP; |
| 5775 break; | |
| 5776 | |
| 5777 default: | |
| 5778 /* Did all possible states at this level, go up one level. */ | |
| 5779 --depth; | |
| 5780 | |
| 5781 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE) | |
| 5782 { | |
| 5783 /* Continue in or go back to the prefix tree. */ | |
| 5784 byts = pbyts; | |
| 5785 idxs = pidxs; | |
| 5786 } | |
| 5787 | |
| 5788 /* Don't check for CTRL-C too often, it takes time. */ | |
| 5789 if (--breakcheckcount == 0) | |
| 5790 { | |
| 5791 ui_breakcheck(); | |
| 5792 breakcheckcount = 1000; | |
| 5793 } | |
| 5794 } | |
| 5795 } | |
| 5796 } | |
| 5797 | |
| 5798 | |
| 5799 /* | |
| 5800 * Go one level deeper in the tree. | |
| 5801 */ | |
| 5802 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5803 go_deeper(trystate_T *stack, int depth, int score_add) |
| 323 | 5804 { |
| 330 | 5805 stack[depth + 1] = stack[depth]; |
| 323 | 5806 stack[depth + 1].ts_state = STATE_START; |
| 625 | 5807 stack[depth + 1].ts_score = stack[depth].ts_score + score_add; |
| 323 | 5808 stack[depth + 1].ts_curi = 1; /* start just after length byte */ |
| 485 | 5809 stack[depth + 1].ts_flags = 0; |
| 323 | 5810 } |
| 5811 | |
| 455 | 5812 #ifdef FEAT_MBYTE |
| 5813 /* | |
| 5814 * Case-folding may change the number of bytes: Count nr of chars in | |
| 5815 * fword[flen] and return the byte length of that many chars in "word". | |
| 5816 */ | |
| 5817 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5818 nofold_len(char_u *fword, int flen, char_u *word) |
| 455 | 5819 { |
| 5820 char_u *p; | |
| 5821 int i = 0; | |
| 5822 | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5823 for (p = fword; p < fword + flen; MB_PTR_ADV(p)) |
| 455 | 5824 ++i; |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5825 for (p = word; i > 0; MB_PTR_ADV(p)) |
| 455 | 5826 --i; |
| 5827 return (int)(p - word); | |
| 5828 } | |
| 5829 #endif | |
| 5830 | |
| 323 | 5831 /* |
| 5832 * "fword" is a good word with case folded. Find the matching keep-case | |
| 5833 * words and put it in "kword". | |
| 5834 * Theoretically there could be several keep-case words that result in the | |
| 5835 * same case-folded word, but we only find one... | |
| 5836 */ | |
| 5837 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5838 find_keepcap_word(slang_T *slang, char_u *fword, char_u *kword) |
| 323 | 5839 { |
| 5840 char_u uword[MAXWLEN]; /* "fword" in upper-case */ | |
| 5841 int depth; | |
| 324 | 5842 idx_T tryidx; |
| 323 | 5843 |
| 5844 /* The following arrays are used at each depth in the tree. */ | |
| 324 | 5845 idx_T arridx[MAXWLEN]; |
| 323 | 5846 int round[MAXWLEN]; |
| 5847 int fwordidx[MAXWLEN]; | |
| 5848 int uwordidx[MAXWLEN]; | |
| 5849 int kwordlen[MAXWLEN]; | |
| 5850 | |
| 5851 int flen, ulen; | |
| 5852 int l; | |
| 5853 int len; | |
| 5854 int c; | |
| 324 | 5855 idx_T lo, hi, m; |
| 323 | 5856 char_u *p; |
| 5857 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */ | |
| 324 | 5858 idx_T *idxs = slang->sl_kidxs; /* array with indexes */ |
| 323 | 5859 |
| 5860 if (byts == NULL) | |
| 5861 { | |
| 5862 /* array is empty: "cannot happen" */ | |
| 5863 *kword = NUL; | |
| 5864 return; | |
| 5865 } | |
| 5866 | |
| 5867 /* Make an all-cap version of "fword". */ | |
| 5868 allcap_copy(fword, uword); | |
| 5869 | |
| 5870 /* | |
| 5871 * Each character needs to be tried both case-folded and upper-case. | |
| 5872 * All this gets very complicated if we keep in mind that changing case | |
| 5873 * may change the byte length of a multi-byte character... | |
| 5874 */ | |
| 5875 depth = 0; | |
| 5876 arridx[0] = 0; | |
| 5877 round[0] = 0; | |
| 5878 fwordidx[0] = 0; | |
| 5879 uwordidx[0] = 0; | |
| 5880 kwordlen[0] = 0; | |
| 5881 while (depth >= 0) | |
| 5882 { | |
| 5883 if (fword[fwordidx[depth]] == NUL) | |
| 5884 { | |
| 5885 /* We are at the end of "fword". If the tree allows a word to end | |
| 5886 * here we have found a match. */ | |
| 5887 if (byts[arridx[depth] + 1] == 0) | |
| 5888 { | |
| 5889 kword[kwordlen[depth]] = NUL; | |
| 5890 return; | |
| 5891 } | |
| 5892 | |
| 5893 /* kword is getting too long, continue one level up */ | |
| 5894 --depth; | |
| 5895 } | |
| 5896 else if (++round[depth] > 2) | |
| 5897 { | |
| 5898 /* tried both fold-case and upper-case character, continue one | |
| 5899 * level up */ | |
| 5900 --depth; | |
| 5901 } | |
| 5902 else | |
| 5903 { | |
| 5904 /* | |
| 5905 * round[depth] == 1: Try using the folded-case character. | |
| 5906 * round[depth] == 2: Try using the upper-case character. | |
| 5907 */ | |
| 5908 #ifdef FEAT_MBYTE | |
| 5909 if (has_mbyte) | |
| 5910 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5911 flen = MB_CPTR2LEN(fword + fwordidx[depth]); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5912 ulen = MB_CPTR2LEN(uword + uwordidx[depth]); |
| 323 | 5913 } |
| 5914 else | |
| 5915 #endif | |
| 5916 ulen = flen = 1; | |
| 5917 if (round[depth] == 1) | |
| 5918 { | |
| 5919 p = fword + fwordidx[depth]; | |
| 5920 l = flen; | |
| 5921 } | |
| 5922 else | |
| 5923 { | |
| 5924 p = uword + uwordidx[depth]; | |
| 5925 l = ulen; | |
| 5926 } | |
| 5927 | |
| 5928 for (tryidx = arridx[depth]; l > 0; --l) | |
| 5929 { | |
| 5930 /* Perform a binary search in the list of accepted bytes. */ | |
| 5931 len = byts[tryidx++]; | |
| 5932 c = *p++; | |
| 5933 lo = tryidx; | |
| 5934 hi = tryidx + len - 1; | |
| 5935 while (lo < hi) | |
| 5936 { | |
| 5937 m = (lo + hi) / 2; | |
| 5938 if (byts[m] > c) | |
| 5939 hi = m - 1; | |
| 5940 else if (byts[m] < c) | |
| 5941 lo = m + 1; | |
| 5942 else | |
| 5943 { | |
| 5944 lo = hi = m; | |
| 5945 break; | |
| 5946 } | |
| 5947 } | |
| 5948 | |
| 5949 /* Stop if there is no matching byte. */ | |
| 5950 if (hi < lo || byts[lo] != c) | |
| 5951 break; | |
| 5952 | |
| 5953 /* Continue at the child (if there is one). */ | |
| 5954 tryidx = idxs[lo]; | |
| 5955 } | |
| 5956 | |
| 5957 if (l == 0) | |
| 5958 { | |
| 5959 /* | |
| 5960 * Found the matching char. Copy it to "kword" and go a | |
| 5961 * level deeper. | |
| 5962 */ | |
| 5963 if (round[depth] == 1) | |
| 5964 { | |
| 5965 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth], | |
| 5966 flen); | |
| 5967 kwordlen[depth + 1] = kwordlen[depth] + flen; | |
| 5968 } | |
| 5969 else | |
| 5970 { | |
| 5971 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth], | |
| 5972 ulen); | |
| 5973 kwordlen[depth + 1] = kwordlen[depth] + ulen; | |
| 5974 } | |
| 5975 fwordidx[depth + 1] = fwordidx[depth] + flen; | |
| 5976 uwordidx[depth + 1] = uwordidx[depth] + ulen; | |
| 5977 | |
| 5978 ++depth; | |
| 5979 arridx[depth] = tryidx; | |
| 5980 round[depth] = 0; | |
| 5981 } | |
| 5982 } | |
| 5983 } | |
| 5984 | |
| 5985 /* Didn't find it: "cannot happen". */ | |
| 5986 *kword = NUL; | |
| 5987 } | |
| 5988 | |
| 5989 /* | |
| 344 | 5990 * Compute the sound-a-like score for suggestions in su->su_ga and add them to |
| 5991 * su->su_sga. | |
| 5992 */ | |
| 5993 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5994 score_comp_sal(suginfo_T *su) |
| 344 | 5995 { |
| 5996 langp_T *lp; | |
| 5997 char_u badsound[MAXWLEN]; | |
| 5998 int i; | |
| 5999 suggest_T *stp; | |
| 6000 suggest_T *sstp; | |
| 6001 int score; | |
| 500 | 6002 int lpi; |
| 344 | 6003 |
| 6004 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL) | |
| 6005 return; | |
| 6006 | |
| 6007 /* 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
|
6008 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
|
6009 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6010 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 6011 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6012 { | |
| 6013 /* soundfold the bad word */ | |
| 375 | 6014 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound); |
| 344 | 6015 |
| 6016 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 6017 { | |
| 6018 stp = &SUG(su->su_ga, i); | |
| 6019 | |
| 351 | 6020 /* Case-fold the suggested word, sound-fold it and compute the |
| 6021 * sound-a-like score. */ | |
| 6022 score = stp_sal_score(stp, su, lp->lp_slang, badsound); | |
| 344 | 6023 if (score < SCORE_MAXMAX) |
| 6024 { | |
| 6025 /* Add the suggestion. */ | |
| 6026 sstp = &SUG(su->su_sga, su->su_sga.ga_len); | |
| 6027 sstp->st_word = vim_strsave(stp->st_word); | |
| 6028 if (sstp->st_word != NULL) | |
| 6029 { | |
| 625 | 6030 sstp->st_wordlen = stp->st_wordlen; |
| 344 | 6031 sstp->st_score = score; |
| 6032 sstp->st_altscore = 0; | |
| 6033 sstp->st_orglen = stp->st_orglen; | |
| 6034 ++su->su_sga.ga_len; | |
| 6035 } | |
| 6036 } | |
| 6037 } | |
| 6038 break; | |
| 6039 } | |
| 500 | 6040 } |
| 344 | 6041 } |
| 6042 | |
| 6043 /* | |
| 6044 * Combine the list of suggestions in su->su_ga and su->su_sga. | |
| 4352 | 6045 * They are entwined. |
| 344 | 6046 */ |
| 6047 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6048 score_combine(suginfo_T *su) |
| 344 | 6049 { |
| 6050 int i; | |
| 6051 int j; | |
| 6052 garray_T ga; | |
| 6053 garray_T *gap; | |
| 6054 langp_T *lp; | |
| 6055 suggest_T *stp; | |
| 6056 char_u *p; | |
| 6057 char_u badsound[MAXWLEN]; | |
| 6058 int round; | |
| 500 | 6059 int lpi; |
| 625 | 6060 slang_T *slang = NULL; |
| 344 | 6061 |
| 6062 /* 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
|
6063 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
|
6064 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6065 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 6066 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6067 { | |
| 6068 /* soundfold the bad word */ | |
| 625 | 6069 slang = lp->lp_slang; |
| 6070 spell_soundfold(slang, su->su_fbadword, TRUE, badsound); | |
| 344 | 6071 |
| 6072 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 6073 { | |
| 6074 stp = &SUG(su->su_ga, i); | |
| 625 | 6075 stp->st_altscore = stp_sal_score(stp, su, slang, badsound); |
| 344 | 6076 if (stp->st_altscore == SCORE_MAXMAX) |
| 6077 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4; | |
| 6078 else | |
| 6079 stp->st_score = (stp->st_score * 3 | |
| 6080 + stp->st_altscore) / 4; | |
| 6081 stp->st_salscore = FALSE; | |
| 6082 } | |
| 6083 break; | |
| 6084 } | |
| 6085 } | |
| 6086 | |
| 844 | 6087 if (slang == NULL) /* Using "double" without sound folding. */ |
| 6088 { | |
| 6089 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, | |
| 6090 su->su_maxcount); | |
| 625 | 6091 return; |
| 844 | 6092 } |
| 625 | 6093 |
| 344 | 6094 /* Add the alternate score to su_sga. */ |
| 6095 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 6096 { | |
| 6097 stp = &SUG(su->su_sga, i); | |
| 625 | 6098 stp->st_altscore = spell_edit_score(slang, |
| 6099 su->su_badword, stp->st_word); | |
| 344 | 6100 if (stp->st_score == SCORE_MAXMAX) |
| 6101 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8; | |
| 6102 else | |
| 6103 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8; | |
| 6104 stp->st_salscore = TRUE; | |
| 6105 } | |
| 6106 | |
| 625 | 6107 /* Remove bad suggestions, sort the suggestions and truncate at "maxcount" |
| 6108 * for both lists. */ | |
| 6109 check_suggestions(su, &su->su_ga); | |
| 344 | 6110 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 625 | 6111 check_suggestions(su, &su->su_sga); |
| 344 | 6112 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount); |
| 6113 | |
| 6114 ga_init2(&ga, (int)sizeof(suginfo_T), 1); | |
| 6115 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL) | |
| 6116 return; | |
| 6117 | |
| 6118 stp = &SUG(ga, 0); | |
| 6119 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i) | |
| 6120 { | |
| 6121 /* round 1: get a suggestion from su_ga | |
| 6122 * round 2: get a suggestion from su_sga */ | |
| 6123 for (round = 1; round <= 2; ++round) | |
| 6124 { | |
| 6125 gap = round == 1 ? &su->su_ga : &su->su_sga; | |
| 6126 if (i < gap->ga_len) | |
| 6127 { | |
| 6128 /* Don't add a word if it's already there. */ | |
| 6129 p = SUG(*gap, i).st_word; | |
| 6130 for (j = 0; j < ga.ga_len; ++j) | |
| 6131 if (STRCMP(stp[j].st_word, p) == 0) | |
| 6132 break; | |
| 6133 if (j == ga.ga_len) | |
| 6134 stp[ga.ga_len++] = SUG(*gap, i); | |
| 6135 else | |
| 6136 vim_free(p); | |
| 6137 } | |
| 6138 } | |
| 6139 } | |
| 6140 | |
| 6141 ga_clear(&su->su_ga); | |
| 6142 ga_clear(&su->su_sga); | |
| 6143 | |
| 6144 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 6145 if (ga.ga_len > su->su_maxcount) | |
| 6146 { | |
| 6147 for (i = su->su_maxcount; i < ga.ga_len; ++i) | |
| 6148 vim_free(stp[i].st_word); | |
| 6149 ga.ga_len = su->su_maxcount; | |
| 6150 } | |
| 6151 | |
| 6152 su->su_ga = ga; | |
| 6153 } | |
| 6154 | |
| 6155 /* | |
| 351 | 6156 * For the goodword in "stp" compute the soundalike score compared to the |
| 6157 * badword. | |
| 6158 */ | |
| 6159 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6160 stp_sal_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6161 suggest_T *stp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6162 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6163 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6164 char_u *badsound) /* sound-folded badword */ |
| 351 | 6165 { |
| 6166 char_u *p; | |
| 534 | 6167 char_u *pbad; |
| 6168 char_u *pgood; | |
| 351 | 6169 char_u badsound2[MAXWLEN]; |
| 6170 char_u fword[MAXWLEN]; | |
| 6171 char_u goodsound[MAXWLEN]; | |
| 534 | 6172 char_u goodword[MAXWLEN]; |
| 6173 int lendiff; | |
| 6174 | |
| 6175 lendiff = (int)(su->su_badlen - stp->st_orglen); | |
| 6176 if (lendiff >= 0) | |
| 6177 pbad = badsound; | |
| 351 | 6178 else |
| 6179 { | |
| 6180 /* soundfold the bad word with more characters following */ | |
| 6181 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN); | |
| 6182 | |
| 6183 /* When joining two words the sound often changes a lot. E.g., "t he" | |
| 6184 * sounds like "t h" while "the" sounds like "@". Avoid that by | |
| 6185 * removing the space. Don't do it when the good word also contains a | |
| 6186 * space. */ | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
6187 if (VIM_ISWHITE(su->su_badptr[su->su_badlen]) |
| 351 | 6188 && *skiptowhite(stp->st_word) == NUL) |
| 6189 for (p = fword; *(p = skiptowhite(p)) != NUL; ) | |
| 1619 | 6190 STRMOVE(p, p + 1); |
| 351 | 6191 |
| 375 | 6192 spell_soundfold(slang, fword, TRUE, badsound2); |
| 534 | 6193 pbad = badsound2; |
| 6194 } | |
| 6195 | |
| 2768 | 6196 if (lendiff > 0 && stp->st_wordlen + lendiff < MAXWLEN) |
| 534 | 6197 { |
| 6198 /* Add part of the bad word to the good word, so that we soundfold | |
| 6199 * what replaces the bad word. */ | |
| 6200 STRCPY(goodword, stp->st_word); | |
| 625 | 6201 vim_strncpy(goodword + stp->st_wordlen, |
| 6202 su->su_badptr + su->su_badlen - lendiff, lendiff); | |
| 534 | 6203 pgood = goodword; |
| 6204 } | |
| 6205 else | |
| 6206 pgood = stp->st_word; | |
| 351 | 6207 |
| 375 | 6208 /* Sound-fold the word and compute the score for the difference. */ |
| 534 | 6209 spell_soundfold(slang, pgood, FALSE, goodsound); |
| 6210 | |
| 6211 return soundalike_score(goodsound, pbad); | |
| 351 | 6212 } |
| 6213 | |
| 625 | 6214 /* structure used to store soundfolded words that add_sound_suggest() has |
| 6215 * handled already. */ | |
| 6216 typedef struct | |
| 6217 { | |
| 6218 short sft_score; /* lowest score used */ | |
| 6219 char_u sft_word[1]; /* soundfolded word, actually longer */ | |
| 6220 } sftword_T; | |
| 6221 | |
| 6222 static sftword_T dumsft; | |
| 6223 #define HIKEY2SFT(p) ((sftword_T *)(p - (dumsft.sft_word - (char_u *)&dumsft))) | |
| 6224 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key) | |
| 6225 | |
| 6226 /* | |
| 6227 * Prepare for calling suggest_try_soundalike(). | |
| 6228 */ | |
| 6229 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6230 suggest_try_soundalike_prep(void) |
| 625 | 6231 { |
| 6232 langp_T *lp; | |
| 6233 int lpi; | |
| 6234 slang_T *slang; | |
| 6235 | |
| 6236 /* Do this for all languages that support sound folding and for which a | |
| 6237 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6238 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
|
6239 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6240 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 6241 slang = lp->lp_slang; |
| 6242 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 6243 /* prepare the hashtable used by add_sound_suggest() */ | |
| 6244 hash_init(&slang->sl_sounddone); | |
| 6245 } | |
| 6246 } | |
| 6247 | |
| 351 | 6248 /* |
| 323 | 6249 * Find suggestions by comparing the word in a sound-a-like form. |
| 503 | 6250 * Note: This doesn't support postponed prefixes. |
| 323 | 6251 */ |
| 6252 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6253 suggest_try_soundalike(suginfo_T *su) |
| 323 | 6254 { |
| 6255 char_u salword[MAXWLEN]; | |
| 6256 langp_T *lp; | |
| 500 | 6257 int lpi; |
| 503 | 6258 slang_T *slang; |
| 344 | 6259 |
| 625 | 6260 /* Do this for all languages that support sound folding and for which a |
| 6261 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6262 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
|
6263 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6264 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 6265 slang = lp->lp_slang; |
| 6266 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 6267 { | |
| 6268 /* soundfold the bad word */ | |
| 6269 spell_soundfold(slang, su->su_fbadword, TRUE, salword); | |
| 6270 | |
| 6271 /* try all kinds of inserts/deletes/swaps/etc. */ | |
| 6272 /* TODO: also soundfold the next words, so that we can try joining | |
| 6273 * and splitting */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6274 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6275 prof_init(); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6276 #endif |
| 625 | 6277 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
|
6278 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6279 prof_report("soundalike"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6280 #endif |
| 625 | 6281 } |
| 6282 } | |
| 6283 } | |
| 6284 | |
| 6285 /* | |
| 6286 * Finish up after calling suggest_try_soundalike(). | |
| 6287 */ | |
| 6288 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6289 suggest_try_soundalike_finish(void) |
| 625 | 6290 { |
| 6291 langp_T *lp; | |
| 6292 int lpi; | |
| 6293 slang_T *slang; | |
| 6294 int todo; | |
| 6295 hashitem_T *hi; | |
| 6296 | |
| 6297 /* Do this for all languages that support sound folding and for which a | |
| 6298 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6299 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
|
6300 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6301 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 503 | 6302 slang = lp->lp_slang; |
| 625 | 6303 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) |
| 6304 { | |
| 6305 /* Free the info about handled words. */ | |
| 835 | 6306 todo = (int)slang->sl_sounddone.ht_used; |
| 625 | 6307 for (hi = slang->sl_sounddone.ht_array; todo > 0; ++hi) |
| 6308 if (!HASHITEM_EMPTY(hi)) | |
| 6309 { | |
| 6310 vim_free(HI2SFT(hi)); | |
| 6311 --todo; | |
| 6312 } | |
| 1087 | 6313 |
| 6314 /* Clear the hashtable, it may also be used by another region. */ | |
| 625 | 6315 hash_clear(&slang->sl_sounddone); |
| 1087 | 6316 hash_init(&slang->sl_sounddone); |
| 625 | 6317 } |
| 6318 } | |
| 6319 } | |
| 6320 | |
| 6321 /* | |
| 6322 * A match with a soundfolded word is found. Add the good word(s) that | |
| 6323 * produce this soundfolded word. | |
| 6324 */ | |
| 6325 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6326 add_sound_suggest( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6327 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6328 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6329 int score, /* soundfold score */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6330 langp_T *lp) |
| 625 | 6331 { |
| 6332 slang_T *slang = lp->lp_slang; /* language for sound folding */ | |
| 6333 int sfwordnr; | |
| 6334 char_u *nrline; | |
| 6335 int orgnr; | |
| 6336 char_u theword[MAXWLEN]; | |
| 6337 int i; | |
| 6338 int wlen; | |
| 6339 char_u *byts; | |
| 6340 idx_T *idxs; | |
| 6341 int n; | |
| 6342 int wordcount; | |
| 6343 int wc; | |
| 6344 int goodscore; | |
| 6345 hash_T hash; | |
| 6346 hashitem_T *hi; | |
| 6347 sftword_T *sft; | |
| 6348 int bc, gc; | |
| 6349 int limit; | |
| 6350 | |
| 6351 /* | |
| 6352 * It's very well possible that the same soundfold word is found several | |
| 6353 * times with different scores. Since the following is quite slow only do | |
| 6354 * the words that have a better score than before. Use a hashtable to | |
| 6355 * remember the words that have been done. | |
| 6356 */ | |
| 6357 hash = hash_hash(goodword); | |
| 6358 hi = hash_lookup(&slang->sl_sounddone, goodword, hash); | |
| 6359 if (HASHITEM_EMPTY(hi)) | |
| 6360 { | |
| 844 | 6361 sft = (sftword_T *)alloc((unsigned)(sizeof(sftword_T) |
| 6362 + STRLEN(goodword))); | |
| 625 | 6363 if (sft != NULL) |
| 6364 { | |
| 6365 sft->sft_score = score; | |
| 6366 STRCPY(sft->sft_word, goodword); | |
| 6367 hash_add_item(&slang->sl_sounddone, hi, sft->sft_word, hash); | |
| 6368 } | |
| 6369 } | |
| 6370 else | |
| 6371 { | |
| 6372 sft = HI2SFT(hi); | |
| 6373 if (score >= sft->sft_score) | |
| 6374 return; | |
| 6375 sft->sft_score = score; | |
| 6376 } | |
| 6377 | |
| 6378 /* | |
| 6379 * Find the word nr in the soundfold tree. | |
| 6380 */ | |
| 6381 sfwordnr = soundfold_find(slang, goodword); | |
| 6382 if (sfwordnr < 0) | |
| 6383 { | |
|
10359
66f1b5bf3fa6
commit https://github.com/vim/vim/commit/95f096030ed1a8afea028f2ea295d6f6a70f466f
Christian Brabandt <cb@256bit.org>
parents:
10277
diff
changeset
|
6384 internal_error("add_sound_suggest()"); |
| 625 | 6385 return; |
| 6386 } | |
| 6387 | |
| 6388 /* | |
| 6389 * go over the list of good words that produce this soundfold word | |
| 6390 */ | |
| 6391 nrline = ml_get_buf(slang->sl_sugbuf, (linenr_T)(sfwordnr + 1), FALSE); | |
| 6392 orgnr = 0; | |
| 6393 while (*nrline != NUL) | |
| 6394 { | |
| 6395 /* The wordnr was stored in a minimal nr of bytes as an offset to the | |
| 6396 * previous wordnr. */ | |
| 6397 orgnr += bytes2offset(&nrline); | |
| 6398 | |
| 6399 byts = slang->sl_fbyts; | |
| 6400 idxs = slang->sl_fidxs; | |
| 6401 | |
| 6402 /* Lookup the word "orgnr" one of the two tries. */ | |
| 6403 n = 0; | |
| 6404 wordcount = 0; | |
| 5506 | 6405 for (wlen = 0; wlen < MAXWLEN - 3; ++wlen) |
| 625 | 6406 { |
| 6407 i = 1; | |
| 6408 if (wordcount == orgnr && byts[n + 1] == NUL) | |
| 6409 break; /* found end of word */ | |
| 6410 | |
| 6411 if (byts[n + 1] == NUL) | |
| 6412 ++wordcount; | |
| 6413 | |
| 6414 /* skip over the NUL bytes */ | |
| 6415 for ( ; byts[n + i] == NUL; ++i) | |
| 6416 if (i > byts[n]) /* safety check */ | |
| 6417 { | |
| 6418 STRCPY(theword + wlen, "BAD"); | |
| 5506 | 6419 wlen += 3; |
| 625 | 6420 goto badword; |
| 6421 } | |
| 6422 | |
| 6423 /* One of the siblings must have the word. */ | |
| 6424 for ( ; i < byts[n]; ++i) | |
| 6425 { | |
| 6426 wc = idxs[idxs[n + i]]; /* nr of words under this byte */ | |
| 6427 if (wordcount + wc > orgnr) | |
| 6428 break; | |
| 6429 wordcount += wc; | |
| 6430 } | |
| 6431 | |
| 5506 | 6432 theword[wlen] = byts[n + i]; |
| 625 | 6433 n = idxs[n + i]; |
| 6434 } | |
| 6435 badword: | |
| 6436 theword[wlen] = NUL; | |
| 6437 | |
| 6438 /* Go over the possible flags and regions. */ | |
| 6439 for (; i <= byts[n] && byts[n + i] == NUL; ++i) | |
| 6440 { | |
| 6441 char_u cword[MAXWLEN]; | |
| 6442 char_u *p; | |
| 6443 int flags = (int)idxs[n + i]; | |
| 6444 | |
| 698 | 6445 /* Skip words with the NOSUGGEST flag */ |
| 6446 if (flags & WF_NOSUGGEST) | |
| 6447 continue; | |
| 6448 | |
| 625 | 6449 if (flags & WF_KEEPCAP) |
| 6450 { | |
| 6451 /* Must find the word in the keep-case tree. */ | |
| 6452 find_keepcap_word(slang, theword, cword); | |
| 6453 p = cword; | |
| 6454 } | |
| 6455 else | |
| 6456 { | |
| 6457 flags |= su->su_badflags; | |
| 6458 if ((flags & WF_CAPMASK) != 0) | |
| 6459 { | |
| 6460 /* Need to fix case according to "flags". */ | |
| 6461 make_case_word(theword, cword, flags); | |
| 6462 p = cword; | |
| 323 | 6463 } |
| 6464 else | |
| 625 | 6465 p = theword; |
| 6466 } | |
| 6467 | |
| 6468 /* Add the suggestion. */ | |
| 6469 if (sps_flags & SPS_DOUBLE) | |
| 6470 { | |
| 6471 /* Add the suggestion if the score isn't too bad. */ | |
| 6472 if (score <= su->su_maxscore) | |
| 6473 add_suggestion(su, &su->su_sga, p, su->su_badlen, | |
| 6474 score, 0, FALSE, slang, FALSE); | |
| 6475 } | |
| 6476 else | |
| 6477 { | |
| 6478 /* Add a penalty for words in another region. */ | |
| 6479 if ((flags & WF_REGION) | |
| 6480 && (((unsigned)flags >> 16) & lp->lp_region) == 0) | |
| 6481 goodscore = SCORE_REGION; | |
| 6482 else | |
| 6483 goodscore = 0; | |
| 6484 | |
| 6485 /* Add a small penalty for changing the first letter from | |
| 6486 * lower to upper case. Helps for "tath" -> "Kath", which is | |
| 4352 | 6487 * less common than "tath" -> "path". Don't do it when the |
| 625 | 6488 * letter is the same, that has already been counted. */ |
| 6489 gc = PTR2CHAR(p); | |
| 6490 if (SPELL_ISUPPER(gc)) | |
| 6491 { | |
| 6492 bc = PTR2CHAR(su->su_badword); | |
| 6493 if (!SPELL_ISUPPER(bc) | |
| 6494 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc)) | |
| 6495 goodscore += SCORE_ICASE / 2; | |
| 6496 } | |
| 6497 | |
| 6498 /* Compute the score for the good word. This only does letter | |
| 6499 * insert/delete/swap/replace. REP items are not considered, | |
| 6500 * which may make the score a bit higher. | |
| 6501 * Use a limit for the score to make it work faster. Use | |
| 6502 * MAXSCORE(), because RESCORE() will change the score. | |
| 6503 * If the limit is very high then the iterative method is | |
| 6504 * inefficient, using an array is quicker. */ | |
| 6505 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score); | |
| 6506 if (limit > SCORE_LIMITMAX) | |
| 6507 goodscore += spell_edit_score(slang, su->su_badword, p); | |
| 6508 else | |
| 6509 goodscore += spell_edit_score_limit(slang, su->su_badword, | |
| 6510 p, limit); | |
| 6511 | |
| 6512 /* When going over the limit don't bother to do the rest. */ | |
| 6513 if (goodscore < SCORE_MAXMAX) | |
| 6514 { | |
| 6515 /* Give a bonus to words seen before. */ | |
| 6516 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE); | |
| 6517 | |
| 6518 /* Add the suggestion if the score isn't too bad. */ | |
| 6519 goodscore = RESCORE(goodscore, score); | |
| 6520 if (goodscore <= su->su_sfmaxscore) | |
| 6521 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 6522 goodscore, score, TRUE, slang, TRUE); | |
| 6523 } | |
| 6524 } | |
| 6525 } | |
| 6526 /* smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr); */ | |
| 6527 } | |
| 6528 } | |
| 6529 | |
| 6530 /* | |
| 6531 * Find word "word" in fold-case tree for "slang" and return the word number. | |
| 6532 */ | |
| 6533 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6534 soundfold_find(slang_T *slang, char_u *word) |
| 625 | 6535 { |
| 6536 idx_T arridx = 0; | |
| 6537 int len; | |
| 6538 int wlen = 0; | |
| 6539 int c; | |
| 6540 char_u *ptr = word; | |
| 6541 char_u *byts; | |
| 6542 idx_T *idxs; | |
| 6543 int wordnr = 0; | |
| 6544 | |
| 6545 byts = slang->sl_sbyts; | |
| 6546 idxs = slang->sl_sidxs; | |
| 6547 | |
| 6548 for (;;) | |
| 6549 { | |
| 6550 /* First byte is the number of possible bytes. */ | |
| 6551 len = byts[arridx++]; | |
| 6552 | |
| 6553 /* If the first possible byte is a zero the word could end here. | |
| 6554 * If the word ends we found the word. If not skip the NUL bytes. */ | |
| 6555 c = ptr[wlen]; | |
| 6556 if (byts[arridx] == NUL) | |
| 6557 { | |
| 6558 if (c == NUL) | |
| 6559 break; | |
| 6560 | |
| 6561 /* Skip over the zeros, there can be several. */ | |
| 6562 while (len > 0 && byts[arridx] == NUL) | |
| 6563 { | |
| 6564 ++arridx; | |
| 6565 --len; | |
| 6566 } | |
| 6567 if (len == 0) | |
| 6568 return -1; /* no children, word should have ended here */ | |
| 6569 ++wordnr; | |
| 6570 } | |
| 6571 | |
| 6572 /* If the word ends we didn't find it. */ | |
| 6573 if (c == NUL) | |
| 6574 return -1; | |
| 6575 | |
| 6576 /* Perform a binary search in the list of accepted bytes. */ | |
| 6577 if (c == TAB) /* <Tab> is handled like <Space> */ | |
| 6578 c = ' '; | |
| 6579 while (byts[arridx] < c) | |
| 6580 { | |
| 6581 /* The word count is in the first idxs[] entry of the child. */ | |
| 6582 wordnr += idxs[idxs[arridx]]; | |
| 6583 ++arridx; | |
| 6584 if (--len == 0) /* end of the bytes, didn't find it */ | |
| 6585 return -1; | |
| 6586 } | |
| 6587 if (byts[arridx] != c) /* didn't find the byte */ | |
| 6588 return -1; | |
| 6589 | |
| 6590 /* Continue at the child (if there is one). */ | |
| 6591 arridx = idxs[arridx]; | |
| 6592 ++wlen; | |
| 6593 | |
| 6594 /* One space in the good word may stand for several spaces in the | |
| 6595 * checked word. */ | |
| 6596 if (c == ' ') | |
| 6597 while (ptr[wlen] == ' ' || ptr[wlen] == TAB) | |
| 6598 ++wlen; | |
| 6599 } | |
| 6600 | |
| 6601 return wordnr; | |
| 323 | 6602 } |
| 6603 | |
| 6604 /* | |
| 324 | 6605 * Copy "fword" to "cword", fixing case according to "flags". |
| 323 | 6606 */ |
| 6607 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6608 make_case_word(char_u *fword, char_u *cword, int flags) |
| 323 | 6609 { |
| 6610 if (flags & WF_ALLCAP) | |
| 6611 /* Make it all upper-case */ | |
| 6612 allcap_copy(fword, cword); | |
| 6613 else if (flags & WF_ONECAP) | |
| 6614 /* Make the first letter upper-case */ | |
| 324 | 6615 onecap_copy(fword, cword, TRUE); |
| 323 | 6616 else |
| 6617 /* Use goodword as-is. */ | |
| 6618 STRCPY(cword, fword); | |
| 6619 } | |
| 6620 | |
| 330 | 6621 |
| 323 | 6622 /* |
| 6623 * Return TRUE if "c1" and "c2" are similar characters according to the MAP | |
| 6624 * lines in the .aff file. | |
| 6625 */ | |
| 6626 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6627 similar_chars(slang_T *slang, int c1, int c2) |
| 323 | 6628 { |
| 330 | 6629 int m1, m2; |
| 6630 #ifdef FEAT_MBYTE | |
| 3549 | 6631 char_u buf[MB_MAXBYTES + 1]; |
| 330 | 6632 hashitem_T *hi; |
| 6633 | |
| 6634 if (c1 >= 256) | |
| 6635 { | |
| 6636 buf[mb_char2bytes(c1, buf)] = 0; | |
| 6637 hi = hash_find(&slang->sl_map_hash, buf); | |
| 6638 if (HASHITEM_EMPTY(hi)) | |
| 6639 m1 = 0; | |
| 6640 else | |
| 6641 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 6642 } | |
| 6643 else | |
| 6644 #endif | |
| 6645 m1 = slang->sl_map_array[c1]; | |
| 6646 if (m1 == 0) | |
| 323 | 6647 return FALSE; |
| 330 | 6648 |
| 6649 | |
| 6650 #ifdef FEAT_MBYTE | |
| 6651 if (c2 >= 256) | |
| 6652 { | |
| 6653 buf[mb_char2bytes(c2, buf)] = 0; | |
| 6654 hi = hash_find(&slang->sl_map_hash, buf); | |
| 6655 if (HASHITEM_EMPTY(hi)) | |
| 6656 m2 = 0; | |
| 6657 else | |
| 6658 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 6659 } | |
| 6660 else | |
| 6661 #endif | |
| 6662 m2 = slang->sl_map_array[c2]; | |
| 6663 | |
| 6664 return m1 == m2; | |
| 323 | 6665 } |
| 6666 | |
| 6667 /* | |
| 6668 * Add a suggestion to the list of suggestions. | |
| 625 | 6669 * For a suggestion that is already in the list the lowest score is remembered. |
| 6670 */ | |
| 6671 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6672 add_suggestion( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6673 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6674 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
|
6675 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6676 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
|
6677 int score, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6678 int altscore, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6679 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
|
6680 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
|
6681 int maxsf) /* su_maxscore applies to soundfold score, |
| 625 | 6682 su_sfmaxscore to the total score. */ |
| 6683 { | |
| 6684 int goodlen; /* len of goodword changed */ | |
| 6685 int badlen; /* len of bad word changed */ | |
| 323 | 6686 suggest_T *stp; |
| 534 | 6687 suggest_T new_sug; |
| 323 | 6688 int i; |
| 534 | 6689 char_u *pgood, *pbad; |
| 532 | 6690 |
| 534 | 6691 /* Minimize "badlen" for consistency. Avoids that changing "the the" to |
| 6692 * "thee the" is added next to changing the first "the" the "thee". */ | |
| 6693 pgood = goodword + STRLEN(goodword); | |
| 625 | 6694 pbad = su->su_badptr + badlenarg; |
| 6695 for (;;) | |
| 6696 { | |
| 835 | 6697 goodlen = (int)(pgood - goodword); |
| 6698 badlen = (int)(pbad - su->su_badptr); | |
| 625 | 6699 if (goodlen <= 0 || badlen <= 0) |
| 6700 break; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
6701 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
|
6702 MB_PTR_BACK(su->su_badptr, pbad); |
| 534 | 6703 #ifdef FEAT_MBYTE |
| 6704 if (has_mbyte) | |
| 6705 { | |
| 6706 if (mb_ptr2char(pgood) != mb_ptr2char(pbad)) | |
| 6707 break; | |
| 346 | 6708 } |
| 6709 else | |
| 534 | 6710 #endif |
| 6711 if (*pgood != *pbad) | |
| 6712 break; | |
| 625 | 6713 } |
| 6714 | |
| 534 | 6715 if (badlen == 0 && goodlen == 0) |
| 6716 /* goodword doesn't change anything; may happen for "the the" changing | |
| 6717 * the first "the" to itself. */ | |
| 6718 return; | |
| 346 | 6719 |
| 944 | 6720 if (gap->ga_len == 0) |
| 6721 i = -1; | |
| 6722 else | |
| 6723 { | |
| 6724 /* Check if the word is already there. Also check the length that is | |
| 6725 * being replaced "thes," -> "these" is a different suggestion from | |
| 6726 * "thes" -> "these". */ | |
| 6727 stp = &SUG(*gap, 0); | |
| 6728 for (i = gap->ga_len; --i >= 0; ++stp) | |
| 6729 if (stp->st_wordlen == goodlen | |
| 6730 && stp->st_orglen == badlen | |
| 6731 && STRNCMP(stp->st_word, goodword, goodlen) == 0) | |
| 6732 { | |
| 6733 /* | |
| 6734 * Found it. Remember the word with the lowest score. | |
| 6735 */ | |
| 6736 if (stp->st_slang == NULL) | |
| 6737 stp->st_slang = slang; | |
| 6738 | |
| 6739 new_sug.st_score = score; | |
| 6740 new_sug.st_altscore = altscore; | |
| 6741 new_sug.st_had_bonus = had_bonus; | |
| 6742 | |
| 6743 if (stp->st_had_bonus != had_bonus) | |
| 6744 { | |
| 6745 /* Only one of the two had the soundalike score computed. | |
| 6746 * Need to do that for the other one now, otherwise the | |
| 6747 * scores can't be compared. This happens because | |
| 6748 * suggest_try_change() doesn't compute the soundalike | |
| 6749 * word to keep it fast, while some special methods set | |
| 6750 * the soundalike score to zero. */ | |
| 6751 if (had_bonus) | |
| 6752 rescore_one(su, stp); | |
| 6753 else | |
| 6754 { | |
| 6755 new_sug.st_word = stp->st_word; | |
| 6756 new_sug.st_wordlen = stp->st_wordlen; | |
| 6757 new_sug.st_slang = stp->st_slang; | |
| 6758 new_sug.st_orglen = badlen; | |
| 6759 rescore_one(su, &new_sug); | |
| 6760 } | |
| 6761 } | |
| 6762 | |
| 6763 if (stp->st_score > new_sug.st_score) | |
| 6764 { | |
| 6765 stp->st_score = new_sug.st_score; | |
| 6766 stp->st_altscore = new_sug.st_altscore; | |
| 6767 stp->st_had_bonus = new_sug.st_had_bonus; | |
| 6768 } | |
| 6769 break; | |
| 6770 } | |
| 6771 } | |
| 625 | 6772 |
| 6773 if (i < 0 && ga_grow(gap, 1) == OK) | |
| 6774 { | |
| 6775 /* Add a suggestion. */ | |
| 6776 stp = &SUG(*gap, gap->ga_len); | |
| 6777 stp->st_word = vim_strnsave(goodword, goodlen); | |
| 6778 if (stp->st_word != NULL) | |
| 6779 { | |
| 6780 stp->st_wordlen = goodlen; | |
| 6781 stp->st_score = score; | |
| 6782 stp->st_altscore = altscore; | |
| 6783 stp->st_had_bonus = had_bonus; | |
| 6784 stp->st_orglen = badlen; | |
| 6785 stp->st_slang = slang; | |
| 6786 ++gap->ga_len; | |
| 6787 | |
| 6788 /* If we have too many suggestions now, sort the list and keep | |
| 6789 * the best suggestions. */ | |
| 6790 if (gap->ga_len > SUG_MAX_COUNT(su)) | |
| 6791 { | |
| 6792 if (maxsf) | |
| 6793 su->su_sfmaxscore = cleanup_suggestions(gap, | |
| 6794 su->su_sfmaxscore, SUG_CLEAN_COUNT(su)); | |
| 6795 else | |
| 6796 su->su_maxscore = cleanup_suggestions(gap, | |
| 6797 su->su_maxscore, SUG_CLEAN_COUNT(su)); | |
| 6798 } | |
| 6799 } | |
| 6800 } | |
| 6801 } | |
| 6802 | |
| 6803 /* | |
| 6804 * Suggestions may in fact be flagged as errors. Esp. for banned words and | |
| 6805 * for split words, such as "the the". Remove these from the list here. | |
| 6806 */ | |
| 6807 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6808 check_suggestions( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6809 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6810 garray_T *gap) /* either su_ga or su_sga */ |
| 625 | 6811 { |
| 6812 suggest_T *stp; | |
| 6813 int i; | |
| 6814 char_u longword[MAXWLEN + 1]; | |
| 6815 int len; | |
| 6816 hlf_T attr; | |
| 6817 | |
| 6818 stp = &SUG(*gap, 0); | |
| 6819 for (i = gap->ga_len - 1; i >= 0; --i) | |
| 6820 { | |
| 6821 /* Need to append what follows to check for "the the". */ | |
| 2768 | 6822 vim_strncpy(longword, stp[i].st_word, MAXWLEN); |
| 625 | 6823 len = stp[i].st_wordlen; |
| 6824 vim_strncpy(longword + len, su->su_badptr + stp[i].st_orglen, | |
| 6825 MAXWLEN - len); | |
| 6826 attr = HLF_COUNT; | |
| 6827 (void)spell_check(curwin, longword, &attr, NULL, FALSE); | |
| 6828 if (attr != HLF_COUNT) | |
| 6829 { | |
| 6830 /* Remove this entry. */ | |
| 6831 vim_free(stp[i].st_word); | |
| 6832 --gap->ga_len; | |
| 6833 if (i < gap->ga_len) | |
| 6834 mch_memmove(stp + i, stp + i + 1, | |
| 6835 sizeof(suggest_T) * (gap->ga_len - i)); | |
| 6836 } | |
| 6837 } | |
| 6838 } | |
| 6839 | |
| 323 | 6840 |
| 6841 /* | |
| 6842 * Add a word to be banned. | |
| 6843 */ | |
| 6844 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6845 add_banned( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6846 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6847 char_u *word) |
| 323 | 6848 { |
| 626 | 6849 char_u *s; |
| 323 | 6850 hash_T hash; |
| 6851 hashitem_T *hi; | |
| 6852 | |
| 625 | 6853 hash = hash_hash(word); |
| 6854 hi = hash_lookup(&su->su_banned, word, hash); | |
| 6855 if (HASHITEM_EMPTY(hi)) | |
| 6856 { | |
| 6857 s = vim_strsave(word); | |
| 6858 if (s != NULL) | |
| 323 | 6859 hash_add_item(&su->su_banned, hi, s, hash); |
| 625 | 6860 } |
| 323 | 6861 } |
| 6862 | |
| 324 | 6863 /* |
| 534 | 6864 * Recompute the score for all suggestions if sound-folding is possible. This |
| 6865 * is slow, thus only done for the final results. | |
| 324 | 6866 */ |
| 6867 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6868 rescore_suggestions(suginfo_T *su) |
| 324 | 6869 { |
| 6870 int i; | |
| 534 | 6871 |
| 6872 if (su->su_sallang != NULL) | |
| 503 | 6873 for (i = 0; i < su->su_ga.ga_len; ++i) |
| 534 | 6874 rescore_one(su, &SUG(su->su_ga, i)); |
| 6875 } | |
| 6876 | |
| 6877 /* | |
| 6878 * Recompute the score for one suggestion if sound-folding is possible. | |
| 6879 */ | |
| 6880 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6881 rescore_one(suginfo_T *su, suggest_T *stp) |
| 534 | 6882 { |
| 6883 slang_T *slang = stp->st_slang; | |
| 6884 char_u sal_badword[MAXWLEN]; | |
| 535 | 6885 char_u *p; |
| 534 | 6886 |
| 6887 /* Only rescore suggestions that have no sal score yet and do have a | |
| 6888 * language. */ | |
| 6889 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus) | |
| 6890 { | |
| 6891 if (slang == su->su_sallang) | |
| 535 | 6892 p = su->su_sal_badword; |
| 534 | 6893 else |
| 6894 { | |
| 6895 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword); | |
| 535 | 6896 p = sal_badword; |
| 6897 } | |
| 6898 | |
| 6899 stp->st_altscore = stp_sal_score(stp, su, slang, p); | |
| 534 | 6900 if (stp->st_altscore == SCORE_MAXMAX) |
| 6901 stp->st_altscore = SCORE_BIG; | |
| 6902 stp->st_score = RESCORE(stp->st_score, stp->st_altscore); | |
| 6903 stp->st_had_bonus = TRUE; | |
| 324 | 6904 } |
| 6905 } | |
| 6906 | |
| 323 | 6907 static int |
| 6908 #ifdef __BORLANDC__ | |
| 6909 _RTLENTRYF | |
| 6910 #endif | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
6911 sug_compare(const void *s1, const void *s2); |
| 323 | 6912 |
| 6913 /* | |
| 6914 * Function given to qsort() to sort the suggestions on st_score. | |
| 529 | 6915 * First on "st_score", then "st_altscore" then alphabetically. |
| 323 | 6916 */ |
| 6917 static int | |
| 6918 #ifdef __BORLANDC__ | |
| 6919 _RTLENTRYF | |
| 6920 #endif | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6921 sug_compare(const void *s1, const void *s2) |
| 323 | 6922 { |
| 6923 suggest_T *p1 = (suggest_T *)s1; | |
| 6924 suggest_T *p2 = (suggest_T *)s2; | |
| 344 | 6925 int n = p1->st_score - p2->st_score; |
| 6926 | |
| 6927 if (n == 0) | |
| 529 | 6928 { |
| 6929 n = p1->st_altscore - p2->st_altscore; | |
| 6930 if (n == 0) | |
| 6931 n = STRICMP(p1->st_word, p2->st_word); | |
| 6932 } | |
| 344 | 6933 return n; |
| 323 | 6934 } |
| 6935 | |
| 6936 /* | |
| 6937 * Cleanup the suggestions: | |
| 6938 * - Sort on score. | |
| 6939 * - Remove words that won't be displayed. | |
| 344 | 6940 * Returns the maximum score in the list or "maxscore" unmodified. |
| 323 | 6941 */ |
| 344 | 6942 static int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6943 cleanup_suggestions( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6944 garray_T *gap, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6945 int maxscore, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6946 int keep) /* nr of suggestions to keep */ |
| 323 | 6947 { |
| 344 | 6948 suggest_T *stp = &SUG(*gap, 0); |
| 323 | 6949 int i; |
| 6950 | |
| 6951 /* Sort the list. */ | |
| 344 | 6952 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare); |
| 323 | 6953 |
| 6954 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 344 | 6955 if (gap->ga_len > keep) |
| 323 | 6956 { |
| 344 | 6957 for (i = keep; i < gap->ga_len; ++i) |
| 323 | 6958 vim_free(stp[i].st_word); |
| 344 | 6959 gap->ga_len = keep; |
| 6960 return stp[keep - 1].st_score; | |
| 323 | 6961 } |
| 344 | 6962 return maxscore; |
| 323 | 6963 } |
| 6964 | |
| 372 | 6965 #if defined(FEAT_EVAL) || defined(PROTO) |
| 6966 /* | |
| 6967 * Soundfold a string, for soundfold(). | |
| 6968 * Result is in allocated memory, NULL for an error. | |
| 6969 */ | |
| 6970 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6971 eval_soundfold(char_u *word) |
| 372 | 6972 { |
| 6973 langp_T *lp; | |
| 6974 char_u sound[MAXWLEN]; | |
| 500 | 6975 int lpi; |
| 372 | 6976 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6977 if (curwin->w_p_spell && *curwin->w_s->b_p_spl != NUL) |
| 372 | 6978 /* 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
|
6979 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
|
6980 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6981 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 372 | 6982 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6983 { | |
| 6984 /* soundfold the word */ | |
| 375 | 6985 spell_soundfold(lp->lp_slang, word, FALSE, sound); |
| 372 | 6986 return vim_strsave(sound); |
| 6987 } | |
| 500 | 6988 } |
| 372 | 6989 |
| 6990 /* No language with sound folding, return word as-is. */ | |
| 6991 return vim_strsave(word); | |
| 6992 } | |
| 6993 #endif | |
| 6994 | |
| 323 | 6995 /* |
| 6996 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 485 | 6997 * |
| 6998 * There are many ways to turn a word into a sound-a-like representation. The | |
| 6999 * oldest is Soundex (1918!). A nice overview can be found in "Approximate | |
| 7000 * swedish name matching - survey and test of different algorithms" by Klas | |
| 7001 * Erikson. | |
| 7002 * | |
| 7003 * We support two methods: | |
| 7004 * 1. SOFOFROM/SOFOTO do a simple character mapping. | |
| 7005 * 2. SAL items define a more advanced sound-folding (and much slower). | |
| 323 | 7006 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
7007 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7008 spell_soundfold( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7009 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7010 char_u *inword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7011 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
|
7012 char_u *res) |
| 375 | 7013 { |
| 7014 char_u fword[MAXWLEN]; | |
| 7015 char_u *word; | |
| 7016 | |
| 7017 if (slang->sl_sofo) | |
| 7018 /* SOFOFROM and SOFOTO used */ | |
| 7019 spell_soundfold_sofo(slang, inword, res); | |
| 7020 else | |
| 7021 { | |
| 7022 /* SAL items used. Requires the word to be case-folded. */ | |
| 7023 if (folded) | |
| 7024 word = inword; | |
| 7025 else | |
| 7026 { | |
| 835 | 7027 (void)spell_casefold(inword, (int)STRLEN(inword), fword, MAXWLEN); |
| 375 | 7028 word = fword; |
| 7029 } | |
| 7030 | |
| 7031 #ifdef FEAT_MBYTE | |
| 7032 if (has_mbyte) | |
| 7033 spell_soundfold_wsal(slang, word, res); | |
| 7034 else | |
| 7035 #endif | |
| 7036 spell_soundfold_sal(slang, word, res); | |
| 7037 } | |
| 7038 } | |
| 7039 | |
| 7040 /* | |
| 7041 * Perform sound folding of "inword" into "res" according to SOFOFROM and | |
| 7042 * SOFOTO lines. | |
| 7043 */ | |
| 7044 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7045 spell_soundfold_sofo(slang_T *slang, char_u *inword, char_u *res) |
| 375 | 7046 { |
| 7047 char_u *s; | |
| 7048 int ri = 0; | |
| 7049 int c; | |
| 7050 | |
| 7051 #ifdef FEAT_MBYTE | |
| 7052 if (has_mbyte) | |
| 7053 { | |
| 7054 int prevc = 0; | |
| 7055 int *ip; | |
| 7056 | |
| 7057 /* The sl_sal_first[] table contains the translation for chars up to | |
| 7058 * 255, sl_sal the rest. */ | |
| 7059 for (s = inword; *s != NUL; ) | |
| 7060 { | |
| 474 | 7061 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
|
7062 if (enc_utf8 ? utf_class(c) == 0 : VIM_ISWHITE(c)) |
| 375 | 7063 c = ' '; |
| 7064 else if (c < 256) | |
| 7065 c = slang->sl_sal_first[c]; | |
| 7066 else | |
| 7067 { | |
| 7068 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff]; | |
| 7069 if (ip == NULL) /* empty list, can't match */ | |
| 7070 c = NUL; | |
| 7071 else | |
| 7072 for (;;) /* find "c" in the list */ | |
| 7073 { | |
| 7074 if (*ip == 0) /* not found */ | |
| 7075 { | |
| 7076 c = NUL; | |
| 7077 break; | |
| 7078 } | |
| 7079 if (*ip == c) /* match! */ | |
| 7080 { | |
| 7081 c = ip[1]; | |
| 7082 break; | |
| 7083 } | |
| 7084 ip += 2; | |
| 7085 } | |
| 7086 } | |
| 7087 | |
| 7088 if (c != NUL && c != prevc) | |
| 7089 { | |
| 7090 ri += mb_char2bytes(c, res + ri); | |
| 7091 if (ri + MB_MAXBYTES > MAXWLEN) | |
| 7092 break; | |
| 7093 prevc = c; | |
| 7094 } | |
| 7095 } | |
| 7096 } | |
| 7097 else | |
| 7098 #endif | |
| 7099 { | |
| 7100 /* The sl_sal_first[] table contains the translation. */ | |
| 7101 for (s = inword; (c = *s) != NUL; ++s) | |
| 7102 { | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7103 if (VIM_ISWHITE(c)) |
| 375 | 7104 c = ' '; |
| 7105 else | |
| 7106 c = slang->sl_sal_first[c]; | |
| 7107 if (c != NUL && (ri == 0 || res[ri - 1] != c)) | |
| 7108 res[ri++] = c; | |
| 7109 } | |
| 7110 } | |
| 7111 | |
| 7112 res[ri] = NUL; | |
| 7113 } | |
| 7114 | |
| 7115 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7116 spell_soundfold_sal(slang_T *slang, char_u *inword, char_u *res) |
| 323 | 7117 { |
| 344 | 7118 salitem_T *smp; |
| 323 | 7119 char_u word[MAXWLEN]; |
| 375 | 7120 char_u *s = inword; |
| 323 | 7121 char_u *t; |
| 344 | 7122 char_u *pf; |
| 323 | 7123 int i, j, z; |
| 344 | 7124 int reslen; |
| 323 | 7125 int n, k = 0; |
| 7126 int z0; | |
| 7127 int k0; | |
| 7128 int n0; | |
| 7129 int c; | |
| 7130 int pri; | |
| 7131 int p0 = -333; | |
| 7132 int c0; | |
| 7133 | |
| 324 | 7134 /* Remove accents, if wanted. We actually remove all non-word characters. |
| 375 | 7135 * But keep white space. We need a copy, the word may be changed here. */ |
| 323 | 7136 if (slang->sl_rem_accents) |
| 7137 { | |
| 7138 t = word; | |
| 375 | 7139 while (*s != NUL) |
| 323 | 7140 { |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7141 if (VIM_ISWHITE(*s)) |
| 344 | 7142 { |
| 7143 *t++ = ' '; | |
| 7144 s = skipwhite(s); | |
| 7145 } | |
| 323 | 7146 else |
| 7147 { | |
| 5477 | 7148 if (spell_iswordp_nmw(s, curwin)) |
| 323 | 7149 *t++ = *s; |
| 7150 ++s; | |
| 7151 } | |
| 7152 } | |
| 7153 *t = NUL; | |
| 7154 } | |
| 7155 else | |
| 2768 | 7156 vim_strncpy(word, s, MAXWLEN - 1); |
| 323 | 7157 |
| 344 | 7158 smp = (salitem_T *)slang->sl_sal.ga_data; |
| 323 | 7159 |
| 7160 /* | |
| 7161 * This comes from Aspell phonet.cpp. Converted from C++ to C. | |
| 324 | 7162 * Changed to keep spaces. |
| 323 | 7163 */ |
| 344 | 7164 i = reslen = z = 0; |
| 323 | 7165 while ((c = word[i]) != NUL) |
| 7166 { | |
| 344 | 7167 /* Start with the first rule that has the character in the word. */ |
| 323 | 7168 n = slang->sl_sal_first[c]; |
| 7169 z0 = 0; | |
| 7170 | |
| 7171 if (n >= 0) | |
| 7172 { | |
| 7173 /* check all rules for the same letter */ | |
| 344 | 7174 for (; (s = smp[n].sm_lead)[0] == c; ++n) |
| 323 | 7175 { |
| 344 | 7176 /* Quickly skip entries that don't match the word. Most |
| 7177 * entries are less then three chars, optimize for that. */ | |
| 7178 k = smp[n].sm_leadlen; | |
| 7179 if (k > 1) | |
| 323 | 7180 { |
| 344 | 7181 if (word[i + 1] != s[1]) |
| 7182 continue; | |
| 7183 if (k > 2) | |
| 7184 { | |
| 7185 for (j = 2; j < k; ++j) | |
| 7186 if (word[i + j] != s[j]) | |
| 7187 break; | |
| 7188 if (j < k) | |
| 7189 continue; | |
| 7190 } | |
| 323 | 7191 } |
| 7192 | |
| 375 | 7193 if ((pf = smp[n].sm_oneof) != NULL) |
| 323 | 7194 { |
| 375 | 7195 /* Check for match with one of the chars in "sm_oneof". */ |
| 344 | 7196 while (*pf != NUL && *pf != word[i + k]) |
| 7197 ++pf; | |
| 7198 if (*pf == NUL) | |
| 7199 continue; | |
| 7200 ++k; | |
| 323 | 7201 } |
| 344 | 7202 s = smp[n].sm_rules; |
| 7203 pri = 5; /* default priority */ | |
| 323 | 7204 |
| 7205 p0 = *s; | |
| 7206 k0 = k; | |
| 7207 while (*s == '-' && k > 1) | |
| 7208 { | |
| 7209 k--; | |
| 7210 s++; | |
| 7211 } | |
| 7212 if (*s == '<') | |
| 7213 s++; | |
| 344 | 7214 if (VIM_ISDIGIT(*s)) |
| 323 | 7215 { |
| 7216 /* determine priority */ | |
| 7217 pri = *s - '0'; | |
| 7218 s++; | |
| 7219 } | |
| 7220 if (*s == '^' && *(s + 1) == '^') | |
| 7221 s++; | |
| 7222 | |
| 7223 if (*s == NUL | |
| 7224 || (*s == '^' | |
| 324 | 7225 && (i == 0 || !(word[i - 1] == ' ' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7226 || spell_iswordp(word + i - 1, curwin))) |
| 323 | 7227 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7228 || (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 7229 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7230 && spell_iswordp(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7231 && (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 7232 { |
| 7233 /* search for followup rules, if: */ | |
| 7234 /* followup and k > 1 and NO '-' in searchstring */ | |
| 7235 c0 = word[i + k - 1]; | |
| 7236 n0 = slang->sl_sal_first[c0]; | |
| 7237 | |
| 7238 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 344 | 7239 && p0 != '-' && word[i + k] != NUL) |
| 323 | 7240 { |
| 7241 /* test follow-up rule for "word[i + k]" */ | |
| 344 | 7242 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0) |
| 323 | 7243 { |
| 344 | 7244 /* Quickly skip entries that don't match the word. |
| 7245 * */ | |
| 7246 k0 = smp[n0].sm_leadlen; | |
| 7247 if (k0 > 1) | |
| 323 | 7248 { |
| 344 | 7249 if (word[i + k] != s[1]) |
| 7250 continue; | |
| 7251 if (k0 > 2) | |
| 7252 { | |
| 7253 pf = word + i + k + 1; | |
| 7254 for (j = 2; j < k0; ++j) | |
| 7255 if (*pf++ != s[j]) | |
| 7256 break; | |
| 7257 if (j < k0) | |
| 7258 continue; | |
| 7259 } | |
| 323 | 7260 } |
| 344 | 7261 k0 += k - 1; |
| 7262 | |
| 375 | 7263 if ((pf = smp[n0].sm_oneof) != NULL) |
| 323 | 7264 { |
| 344 | 7265 /* Check for match with one of the chars in |
| 375 | 7266 * "sm_oneof". */ |
| 344 | 7267 while (*pf != NUL && *pf != word[i + k0]) |
| 7268 ++pf; | |
| 7269 if (*pf == NUL) | |
| 7270 continue; | |
| 7271 ++k0; | |
| 323 | 7272 } |
| 344 | 7273 |
| 7274 p0 = 5; | |
| 7275 s = smp[n0].sm_rules; | |
| 323 | 7276 while (*s == '-') |
| 7277 { | |
| 344 | 7278 /* "k0" gets NOT reduced because |
| 7279 * "if (k0 == k)" */ | |
| 323 | 7280 s++; |
| 7281 } | |
| 7282 if (*s == '<') | |
| 7283 s++; | |
| 344 | 7284 if (VIM_ISDIGIT(*s)) |
| 323 | 7285 { |
| 7286 p0 = *s - '0'; | |
| 7287 s++; | |
| 7288 } | |
| 7289 | |
| 7290 if (*s == NUL | |
| 7291 /* *s == '^' cuts */ | |
| 7292 || (*s == '$' | |
| 376 | 7293 && !spell_iswordp(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7294 curwin))) |
| 323 | 7295 { |
| 7296 if (k0 == k) | |
| 7297 /* this is just a piece of the string */ | |
| 7298 continue; | |
| 7299 | |
| 7300 if (p0 < pri) | |
| 7301 /* priority too low */ | |
| 7302 continue; | |
| 7303 /* rule fits; stop search */ | |
| 7304 break; | |
| 7305 } | |
| 7306 } | |
| 7307 | |
| 344 | 7308 if (p0 >= pri && smp[n0].sm_lead[0] == c0) |
| 323 | 7309 continue; |
| 7310 } | |
| 7311 | |
| 7312 /* replace string */ | |
| 344 | 7313 s = smp[n].sm_to; |
| 389 | 7314 if (s == NULL) |
| 7315 s = (char_u *)""; | |
| 344 | 7316 pf = smp[n].sm_rules; |
| 7317 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0; | |
| 323 | 7318 if (p0 == 1 && z == 0) |
| 7319 { | |
| 7320 /* rule with '<' is used */ | |
| 344 | 7321 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c |
| 7322 || res[reslen - 1] == *s)) | |
| 7323 reslen--; | |
| 323 | 7324 z0 = 1; |
| 7325 z = 1; | |
| 7326 k0 = 0; | |
| 372 | 7327 while (*s != NUL && word[i + k0] != NUL) |
| 323 | 7328 { |
| 7329 word[i + k0] = *s; | |
| 7330 k0++; | |
| 7331 s++; | |
| 7332 } | |
| 7333 if (k > k0) | |
| 1619 | 7334 STRMOVE(word + i + k0, word + i + k); |
| 323 | 7335 |
| 7336 /* new "actual letter" */ | |
| 7337 c = word[i]; | |
| 7338 } | |
| 7339 else | |
| 7340 { | |
| 7341 /* no '<' rule used */ | |
| 7342 i += k - 1; | |
| 7343 z = 0; | |
| 344 | 7344 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN) |
| 323 | 7345 { |
| 344 | 7346 if (reslen == 0 || res[reslen - 1] != *s) |
| 372 | 7347 res[reslen++] = *s; |
| 323 | 7348 s++; |
| 7349 } | |
| 7350 /* new "actual letter" */ | |
| 7351 c = *s; | |
| 344 | 7352 if (strstr((char *)pf, "^^") != NULL) |
| 323 | 7353 { |
| 7354 if (c != NUL) | |
| 372 | 7355 res[reslen++] = c; |
| 1619 | 7356 STRMOVE(word, word + i + 1); |
| 323 | 7357 i = 0; |
| 7358 z0 = 1; | |
| 7359 } | |
| 7360 } | |
| 7361 break; | |
| 7362 } | |
| 7363 } | |
| 7364 } | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7365 else if (VIM_ISWHITE(c)) |
| 324 | 7366 { |
| 7367 c = ' '; | |
| 7368 k = 1; | |
| 7369 } | |
| 323 | 7370 |
| 7371 if (z0 == 0) | |
| 7372 { | |
| 344 | 7373 if (k && !p0 && reslen < MAXWLEN && c != NUL |
| 7374 && (!slang->sl_collapse || reslen == 0 | |
| 7375 || res[reslen - 1] != c)) | |
| 323 | 7376 /* condense only double letters */ |
| 372 | 7377 res[reslen++] = c; |
| 323 | 7378 |
| 7379 i++; | |
| 7380 z = 0; | |
| 7381 k = 0; | |
| 7382 } | |
| 7383 } | |
| 7384 | |
| 344 | 7385 res[reslen] = NUL; |
| 323 | 7386 } |
| 7387 | |
| 372 | 7388 #ifdef FEAT_MBYTE |
| 7389 /* | |
| 7390 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 7391 * Multi-byte version of spell_soundfold(). | |
| 7392 */ | |
| 7393 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7394 spell_soundfold_wsal(slang_T *slang, char_u *inword, char_u *res) |
| 372 | 7395 { |
| 375 | 7396 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data; |
| 372 | 7397 int word[MAXWLEN]; |
| 7398 int wres[MAXWLEN]; | |
| 7399 int l; | |
| 7400 char_u *s; | |
| 7401 int *ws; | |
| 7402 char_u *t; | |
| 7403 int *pf; | |
| 7404 int i, j, z; | |
| 7405 int reslen; | |
| 7406 int n, k = 0; | |
| 7407 int z0; | |
| 7408 int k0; | |
| 7409 int n0; | |
| 7410 int c; | |
| 7411 int pri; | |
| 7412 int p0 = -333; | |
| 7413 int c0; | |
| 7414 int did_white = FALSE; | |
| 3520 | 7415 int wordlen; |
| 7416 | |
| 372 | 7417 |
| 7418 /* | |
| 7419 * Convert the multi-byte string to a wide-character string. | |
| 7420 * Remove accents, if wanted. We actually remove all non-word characters. | |
| 7421 * But keep white space. | |
| 7422 */ | |
| 3520 | 7423 wordlen = 0; |
| 372 | 7424 for (s = inword; *s != NUL; ) |
| 7425 { | |
| 7426 t = s; | |
| 474 | 7427 c = mb_cptr2char_adv(&s); |
| 372 | 7428 if (slang->sl_rem_accents) |
| 7429 { | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7430 if (enc_utf8 ? utf_class(c) == 0 : VIM_ISWHITE(c)) |
| 372 | 7431 { |
| 7432 if (did_white) | |
| 7433 continue; | |
| 7434 c = ' '; | |
| 7435 did_white = TRUE; | |
| 7436 } | |
| 7437 else | |
| 7438 { | |
| 7439 did_white = FALSE; | |
| 5477 | 7440 if (!spell_iswordp_nmw(t, curwin)) |
| 372 | 7441 continue; |
| 7442 } | |
| 7443 } | |
| 3520 | 7444 word[wordlen++] = c; |
| 7445 } | |
| 7446 word[wordlen] = NUL; | |
| 372 | 7447 |
| 7448 /* | |
| 3520 | 7449 * This algorithm comes from Aspell phonet.cpp. |
| 372 | 7450 * Converted from C++ to C. Added support for multi-byte chars. |
| 7451 * Changed to keep spaces. | |
| 7452 */ | |
| 7453 i = reslen = z = 0; | |
| 7454 while ((c = word[i]) != NUL) | |
| 7455 { | |
| 7456 /* Start with the first rule that has the character in the word. */ | |
| 7457 n = slang->sl_sal_first[c & 0xff]; | |
| 7458 z0 = 0; | |
| 7459 | |
| 7460 if (n >= 0) | |
| 7461 { | |
|
2455
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
7462 /* 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
|
7463 * 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
|
7464 * (c & 0xff) is NUL. */ |
|
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
7465 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
|
7466 && ws[0] != NUL; ++n) |
| 372 | 7467 { |
| 7468 /* Quickly skip entries that don't match the word. Most | |
| 7469 * entries are less then three chars, optimize for that. */ | |
| 375 | 7470 if (c != ws[0]) |
| 7471 continue; | |
| 372 | 7472 k = smp[n].sm_leadlen; |
| 7473 if (k > 1) | |
| 7474 { | |
| 7475 if (word[i + 1] != ws[1]) | |
| 7476 continue; | |
| 7477 if (k > 2) | |
| 7478 { | |
| 7479 for (j = 2; j < k; ++j) | |
| 7480 if (word[i + j] != ws[j]) | |
| 7481 break; | |
| 7482 if (j < k) | |
| 7483 continue; | |
| 7484 } | |
| 7485 } | |
| 7486 | |
| 375 | 7487 if ((pf = smp[n].sm_oneof_w) != NULL) |
| 372 | 7488 { |
| 375 | 7489 /* Check for match with one of the chars in "sm_oneof". */ |
| 372 | 7490 while (*pf != NUL && *pf != word[i + k]) |
| 7491 ++pf; | |
| 7492 if (*pf == NUL) | |
| 7493 continue; | |
| 7494 ++k; | |
| 7495 } | |
| 7496 s = smp[n].sm_rules; | |
| 7497 pri = 5; /* default priority */ | |
| 7498 | |
| 7499 p0 = *s; | |
| 7500 k0 = k; | |
| 7501 while (*s == '-' && k > 1) | |
| 7502 { | |
| 7503 k--; | |
| 7504 s++; | |
| 7505 } | |
| 7506 if (*s == '<') | |
| 7507 s++; | |
| 7508 if (VIM_ISDIGIT(*s)) | |
| 7509 { | |
| 7510 /* determine priority */ | |
| 7511 pri = *s - '0'; | |
| 7512 s++; | |
| 7513 } | |
| 7514 if (*s == '^' && *(s + 1) == '^') | |
| 7515 s++; | |
| 7516 | |
| 7517 if (*s == NUL | |
| 7518 || (*s == '^' | |
| 7519 && (i == 0 || !(word[i - 1] == ' ' | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7520 || spell_iswordp_w(word + i - 1, curwin))) |
| 372 | 7521 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7522 || (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 7523 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7524 && spell_iswordp_w(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7525 && (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 7526 { |
| 7527 /* search for followup rules, if: */ | |
| 7528 /* followup and k > 1 and NO '-' in searchstring */ | |
| 7529 c0 = word[i + k - 1]; | |
| 7530 n0 = slang->sl_sal_first[c0 & 0xff]; | |
| 7531 | |
| 7532 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 7533 && p0 != '-' && word[i + k] != NUL) | |
| 7534 { | |
| 375 | 7535 /* Test follow-up rule for "word[i + k]"; loop over |
| 7536 * all entries with the same index byte. */ | |
| 372 | 7537 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff) |
| 7538 == (c0 & 0xff); ++n0) | |
| 7539 { | |
| 7540 /* Quickly skip entries that don't match the word. | |
| 375 | 7541 */ |
| 7542 if (c0 != ws[0]) | |
| 7543 continue; | |
| 372 | 7544 k0 = smp[n0].sm_leadlen; |
| 7545 if (k0 > 1) | |
| 7546 { | |
| 7547 if (word[i + k] != ws[1]) | |
| 7548 continue; | |
| 7549 if (k0 > 2) | |
| 7550 { | |
| 7551 pf = word + i + k + 1; | |
| 7552 for (j = 2; j < k0; ++j) | |
| 7553 if (*pf++ != ws[j]) | |
| 7554 break; | |
| 7555 if (j < k0) | |
| 7556 continue; | |
| 7557 } | |
| 7558 } | |
| 7559 k0 += k - 1; | |
| 7560 | |
| 375 | 7561 if ((pf = smp[n0].sm_oneof_w) != NULL) |
| 372 | 7562 { |
| 7563 /* Check for match with one of the chars in | |
| 375 | 7564 * "sm_oneof". */ |
| 372 | 7565 while (*pf != NUL && *pf != word[i + k0]) |
| 7566 ++pf; | |
| 7567 if (*pf == NUL) | |
| 7568 continue; | |
| 7569 ++k0; | |
| 7570 } | |
| 7571 | |
| 7572 p0 = 5; | |
| 7573 s = smp[n0].sm_rules; | |
| 7574 while (*s == '-') | |
| 7575 { | |
| 7576 /* "k0" gets NOT reduced because | |
| 7577 * "if (k0 == k)" */ | |
| 7578 s++; | |
| 7579 } | |
| 7580 if (*s == '<') | |
| 7581 s++; | |
| 7582 if (VIM_ISDIGIT(*s)) | |
| 7583 { | |
| 7584 p0 = *s - '0'; | |
| 7585 s++; | |
| 7586 } | |
| 7587 | |
| 7588 if (*s == NUL | |
| 7589 /* *s == '^' cuts */ | |
| 7590 || (*s == '$' | |
| 376 | 7591 && !spell_iswordp_w(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7592 curwin))) |
| 372 | 7593 { |
| 7594 if (k0 == k) | |
| 7595 /* this is just a piece of the string */ | |
| 7596 continue; | |
| 7597 | |
| 7598 if (p0 < pri) | |
| 7599 /* priority too low */ | |
| 7600 continue; | |
| 7601 /* rule fits; stop search */ | |
| 7602 break; | |
| 7603 } | |
| 7604 } | |
| 7605 | |
| 7606 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff) | |
| 7607 == (c0 & 0xff)) | |
| 7608 continue; | |
| 7609 } | |
| 7610 | |
| 7611 /* replace string */ | |
| 7612 ws = smp[n].sm_to_w; | |
| 7613 s = smp[n].sm_rules; | |
| 7614 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0; | |
| 7615 if (p0 == 1 && z == 0) | |
| 7616 { | |
| 7617 /* rule with '<' is used */ | |
| 389 | 7618 if (reslen > 0 && ws != NULL && *ws != NUL |
| 7619 && (wres[reslen - 1] == c | |
| 372 | 7620 || wres[reslen - 1] == *ws)) |
| 7621 reslen--; | |
| 7622 z0 = 1; | |
| 7623 z = 1; | |
| 7624 k0 = 0; | |
| 389 | 7625 if (ws != NULL) |
| 7626 while (*ws != NUL && word[i + k0] != NUL) | |
| 7627 { | |
| 7628 word[i + k0] = *ws; | |
| 7629 k0++; | |
| 7630 ws++; | |
| 7631 } | |
| 372 | 7632 if (k > k0) |
| 7633 mch_memmove(word + i + k0, word + i + k, | |
| 3520 | 7634 sizeof(int) * (wordlen - (i + k) + 1)); |
| 372 | 7635 |
| 7636 /* new "actual letter" */ | |
| 7637 c = word[i]; | |
| 7638 } | |
| 7639 else | |
| 7640 { | |
| 7641 /* no '<' rule used */ | |
| 7642 i += k - 1; | |
| 7643 z = 0; | |
| 389 | 7644 if (ws != NULL) |
| 7645 while (*ws != NUL && ws[1] != NUL | |
| 7646 && reslen < MAXWLEN) | |
| 7647 { | |
| 7648 if (reslen == 0 || wres[reslen - 1] != *ws) | |
| 7649 wres[reslen++] = *ws; | |
| 7650 ws++; | |
| 7651 } | |
| 372 | 7652 /* new "actual letter" */ |
| 389 | 7653 if (ws == NULL) |
| 7654 c = NUL; | |
| 7655 else | |
| 7656 c = *ws; | |
| 372 | 7657 if (strstr((char *)s, "^^") != NULL) |
| 7658 { | |
| 7659 if (c != NUL) | |
| 7660 wres[reslen++] = c; | |
| 7661 mch_memmove(word, word + i + 1, | |
| 3520 | 7662 sizeof(int) * (wordlen - (i + 1) + 1)); |
| 372 | 7663 i = 0; |
| 7664 z0 = 1; | |
| 7665 } | |
| 7666 } | |
| 7667 break; | |
| 7668 } | |
| 7669 } | |
| 7670 } | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7671 else if (VIM_ISWHITE(c)) |
| 372 | 7672 { |
| 7673 c = ' '; | |
| 7674 k = 1; | |
| 7675 } | |
| 7676 | |
| 7677 if (z0 == 0) | |
| 7678 { | |
| 7679 if (k && !p0 && reslen < MAXWLEN && c != NUL | |
| 7680 && (!slang->sl_collapse || reslen == 0 | |
| 7681 || wres[reslen - 1] != c)) | |
| 7682 /* condense only double letters */ | |
| 7683 wres[reslen++] = c; | |
| 7684 | |
| 7685 i++; | |
| 7686 z = 0; | |
| 7687 k = 0; | |
| 7688 } | |
| 7689 } | |
| 7690 | |
| 7691 /* Convert wide characters in "wres" to a multi-byte string in "res". */ | |
| 7692 l = 0; | |
| 7693 for (n = 0; n < reslen; ++n) | |
| 7694 { | |
| 7695 l += mb_char2bytes(wres[n], res + l); | |
| 7696 if (l + MB_MAXBYTES > MAXWLEN) | |
| 7697 break; | |
| 7698 } | |
| 7699 res[l] = NUL; | |
| 7700 } | |
| 7701 #endif | |
| 7702 | |
| 324 | 7703 /* |
| 344 | 7704 * Compute a score for two sound-a-like words. |
| 7705 * This permits up to two inserts/deletes/swaps/etc. to keep things fast. | |
| 7706 * Instead of a generic loop we write out the code. That keeps it fast by | |
| 7707 * avoiding checks that will not be possible. | |
| 7708 */ | |
| 7709 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7710 soundalike_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7711 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
|
7712 char_u *badstart) /* sound-folded bad word */ |
| 344 | 7713 { |
| 351 | 7714 char_u *goodsound = goodstart; |
| 7715 char_u *badsound = badstart; | |
| 7716 int goodlen; | |
| 7717 int badlen; | |
| 344 | 7718 int n; |
| 7719 char_u *pl, *ps; | |
| 7720 char_u *pl2, *ps2; | |
| 351 | 7721 int score = 0; |
| 7722 | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7723 /* Adding/inserting "*" at the start (word starts with vowel) shouldn't be |
| 351 | 7724 * counted so much, vowels halfway the word aren't counted at all. */ |
| 7725 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound) | |
| 7726 { | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7727 if ((badsound[0] == NUL && goodsound[1] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7728 || (goodsound[0] == NUL && badsound[1] == NUL)) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7729 /* 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
|
7730 return SCORE_DEL; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7731 if (badsound[0] == NUL || goodsound[0] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7732 /* more than two changes */ |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7733 return SCORE_MAXMAX; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7734 |
| 625 | 7735 if (badsound[1] == goodsound[1] |
| 7736 || (badsound[1] != NUL | |
| 7737 && goodsound[1] != NUL | |
| 7738 && badsound[2] == goodsound[2])) | |
| 7739 { | |
| 7740 /* handle like a substitute */ | |
| 7741 } | |
| 351 | 7742 else |
| 625 | 7743 { |
| 7744 score = 2 * SCORE_DEL / 3; | |
| 7745 if (*badsound == '*') | |
| 7746 ++badsound; | |
| 7747 else | |
| 7748 ++goodsound; | |
| 7749 } | |
| 351 | 7750 } |
| 7751 | |
| 835 | 7752 goodlen = (int)STRLEN(goodsound); |
| 7753 badlen = (int)STRLEN(badsound); | |
| 344 | 7754 |
| 1188 | 7755 /* Return quickly if the lengths are too different to be fixed by two |
| 344 | 7756 * changes. */ |
| 7757 n = goodlen - badlen; | |
| 7758 if (n < -2 || n > 2) | |
| 7759 return SCORE_MAXMAX; | |
| 7760 | |
| 7761 if (n > 0) | |
| 7762 { | |
| 351 | 7763 pl = goodsound; /* goodsound is longest */ |
| 344 | 7764 ps = badsound; |
| 7765 } | |
| 7766 else | |
| 7767 { | |
| 351 | 7768 pl = badsound; /* badsound is longest */ |
| 344 | 7769 ps = goodsound; |
| 7770 } | |
| 7771 | |
| 7772 /* Skip over the identical part. */ | |
| 7773 while (*pl == *ps && *pl != NUL) | |
| 7774 { | |
| 7775 ++pl; | |
| 7776 ++ps; | |
| 7777 } | |
| 7778 | |
| 7779 switch (n) | |
| 7780 { | |
| 7781 case -2: | |
| 7782 case 2: | |
| 7783 /* | |
| 7784 * Must delete two characters from "pl". | |
| 7785 */ | |
| 7786 ++pl; /* first delete */ | |
| 7787 while (*pl == *ps) | |
| 7788 { | |
| 7789 ++pl; | |
| 7790 ++ps; | |
| 7791 } | |
| 7792 /* strings must be equal after second delete */ | |
| 7793 if (STRCMP(pl + 1, ps) == 0) | |
| 351 | 7794 return score + SCORE_DEL * 2; |
| 344 | 7795 |
| 7796 /* Failed to compare. */ | |
| 7797 break; | |
| 7798 | |
| 7799 case -1: | |
| 7800 case 1: | |
| 7801 /* | |
| 7802 * Minimal one delete from "pl" required. | |
| 7803 */ | |
| 7804 | |
| 7805 /* 1: delete */ | |
| 7806 pl2 = pl + 1; | |
| 7807 ps2 = ps; | |
| 7808 while (*pl2 == *ps2) | |
| 7809 { | |
| 7810 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7811 return score + SCORE_DEL; |
| 344 | 7812 ++pl2; |
| 7813 ++ps2; | |
| 7814 } | |
| 7815 | |
| 7816 /* 2: delete then swap, then rest must be equal */ | |
| 7817 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7818 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7819 return score + SCORE_DEL + SCORE_SWAP; |
| 344 | 7820 |
| 7821 /* 3: delete then substitute, then the rest must be equal */ | |
| 7822 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7823 return score + SCORE_DEL + SCORE_SUBST; |
| 344 | 7824 |
| 7825 /* 4: first swap then delete */ | |
| 7826 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 7827 { | |
| 7828 pl2 = pl + 2; /* swap, skip two chars */ | |
| 7829 ps2 = ps + 2; | |
| 7830 while (*pl2 == *ps2) | |
| 7831 { | |
| 7832 ++pl2; | |
| 7833 ++ps2; | |
| 7834 } | |
| 7835 /* delete a char and then strings must be equal */ | |
| 7836 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7837 return score + SCORE_SWAP + SCORE_DEL; |
| 344 | 7838 } |
| 7839 | |
| 7840 /* 5: first substitute then delete */ | |
| 7841 pl2 = pl + 1; /* substitute, skip one char */ | |
| 7842 ps2 = ps + 1; | |
| 7843 while (*pl2 == *ps2) | |
| 7844 { | |
| 7845 ++pl2; | |
| 7846 ++ps2; | |
| 7847 } | |
| 7848 /* delete a char and then strings must be equal */ | |
| 7849 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7850 return score + SCORE_SUBST + SCORE_DEL; |
| 344 | 7851 |
| 7852 /* Failed to compare. */ | |
| 7853 break; | |
| 7854 | |
| 7855 case 0: | |
| 7856 /* | |
| 1808 | 7857 * Lengths are equal, thus changes must result in same length: An |
| 344 | 7858 * insert is only possible in combination with a delete. |
| 7859 * 1: check if for identical strings | |
| 7860 */ | |
| 7861 if (*pl == NUL) | |
| 351 | 7862 return score; |
| 344 | 7863 |
| 7864 /* 2: swap */ | |
| 7865 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 7866 { | |
| 7867 pl2 = pl + 2; /* swap, skip two chars */ | |
| 7868 ps2 = ps + 2; | |
| 7869 while (*pl2 == *ps2) | |
| 7870 { | |
| 7871 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7872 return score + SCORE_SWAP; |
| 344 | 7873 ++pl2; |
| 7874 ++ps2; | |
| 7875 } | |
| 7876 /* 3: swap and swap again */ | |
| 7877 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7878 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7879 return score + SCORE_SWAP + SCORE_SWAP; |
| 344 | 7880 |
| 7881 /* 4: swap and substitute */ | |
| 7882 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7883 return score + SCORE_SWAP + SCORE_SUBST; |
| 344 | 7884 } |
| 7885 | |
| 7886 /* 5: substitute */ | |
| 7887 pl2 = pl + 1; | |
| 7888 ps2 = ps + 1; | |
| 7889 while (*pl2 == *ps2) | |
| 7890 { | |
| 7891 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7892 return score + SCORE_SUBST; |
| 344 | 7893 ++pl2; |
| 7894 ++ps2; | |
| 7895 } | |
| 7896 | |
| 7897 /* 6: substitute and swap */ | |
| 7898 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7899 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7900 return score + SCORE_SUBST + SCORE_SWAP; |
| 344 | 7901 |
| 7902 /* 7: substitute and substitute */ | |
| 7903 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7904 return score + SCORE_SUBST + SCORE_SUBST; |
| 344 | 7905 |
| 7906 /* 8: insert then delete */ | |
| 7907 pl2 = pl; | |
| 7908 ps2 = ps + 1; | |
| 7909 while (*pl2 == *ps2) | |
| 7910 { | |
| 7911 ++pl2; | |
| 7912 ++ps2; | |
| 7913 } | |
| 7914 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7915 return score + SCORE_INS + SCORE_DEL; |
| 344 | 7916 |
| 7917 /* 9: delete then insert */ | |
| 7918 pl2 = pl + 1; | |
| 7919 ps2 = ps; | |
| 7920 while (*pl2 == *ps2) | |
| 7921 { | |
| 7922 ++pl2; | |
| 7923 ++ps2; | |
| 7924 } | |
| 7925 if (STRCMP(pl2, ps2 + 1) == 0) | |
| 351 | 7926 return score + SCORE_INS + SCORE_DEL; |
| 344 | 7927 |
| 7928 /* Failed to compare. */ | |
| 7929 break; | |
| 7930 } | |
| 7931 | |
| 7932 return SCORE_MAXMAX; | |
| 7933 } | |
| 324 | 7934 |
| 323 | 7935 /* |
| 7936 * Compute the "edit distance" to turn "badword" into "goodword". The less | |
| 344 | 7937 * deletes/inserts/substitutes/swaps are required the lower the score. |
| 324 | 7938 * |
| 485 | 7939 * The algorithm is described by Du and Chang, 1992. |
| 7940 * The implementation of the algorithm comes from Aspell editdist.cpp, | |
| 7941 * edit_distance(). It has been converted from C++ to C and modified to | |
| 7942 * support multi-byte characters. | |
| 323 | 7943 */ |
| 7944 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7945 spell_edit_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7946 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7947 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7948 char_u *goodword) |
| 323 | 7949 { |
| 7950 int *cnt; | |
| 1188 | 7951 int badlen, goodlen; /* lengths including NUL */ |
| 323 | 7952 int j, i; |
| 7953 int t; | |
| 7954 int bc, gc; | |
| 324 | 7955 int pbc, pgc; |
| 7956 #ifdef FEAT_MBYTE | |
| 7957 char_u *p; | |
| 7958 int wbadword[MAXWLEN]; | |
| 7959 int wgoodword[MAXWLEN]; | |
| 7960 | |
| 7961 if (has_mbyte) | |
| 7962 { | |
| 7963 /* Get the characters from the multi-byte strings and put them in an | |
| 7964 * int array for easy access. */ | |
| 7965 for (p = badword, badlen = 0; *p != NUL; ) | |
| 474 | 7966 wbadword[badlen++] = mb_cptr2char_adv(&p); |
| 398 | 7967 wbadword[badlen++] = 0; |
| 324 | 7968 for (p = goodword, goodlen = 0; *p != NUL; ) |
| 474 | 7969 wgoodword[goodlen++] = mb_cptr2char_adv(&p); |
| 398 | 7970 wgoodword[goodlen++] = 0; |
| 324 | 7971 } |
| 7972 else | |
| 7973 #endif | |
| 7974 { | |
| 835 | 7975 badlen = (int)STRLEN(badword) + 1; |
| 7976 goodlen = (int)STRLEN(goodword) + 1; | |
| 324 | 7977 } |
| 323 | 7978 |
| 7979 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */ | |
| 7980 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)] | |
| 7981 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)), | |
| 7982 TRUE); | |
| 324 | 7983 if (cnt == NULL) |
| 7984 return 0; /* out of memory */ | |
| 323 | 7985 |
| 7986 CNT(0, 0) = 0; | |
| 7987 for (j = 1; j <= goodlen; ++j) | |
| 625 | 7988 CNT(0, j) = CNT(0, j - 1) + SCORE_INS; |
| 323 | 7989 |
| 7990 for (i = 1; i <= badlen; ++i) | |
| 7991 { | |
| 625 | 7992 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL; |
| 323 | 7993 for (j = 1; j <= goodlen; ++j) |
| 7994 { | |
| 324 | 7995 #ifdef FEAT_MBYTE |
| 7996 if (has_mbyte) | |
| 7997 { | |
| 7998 bc = wbadword[i - 1]; | |
| 7999 gc = wgoodword[j - 1]; | |
| 8000 } | |
| 8001 else | |
| 8002 #endif | |
| 8003 { | |
| 8004 bc = badword[i - 1]; | |
| 8005 gc = goodword[j - 1]; | |
| 8006 } | |
| 323 | 8007 if (bc == gc) |
| 8008 CNT(i, j) = CNT(i - 1, j - 1); | |
| 8009 else | |
| 8010 { | |
| 8011 /* Use a better score when there is only a case difference. */ | |
| 324 | 8012 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) |
| 323 | 8013 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1); |
| 8014 else | |
| 625 | 8015 { |
| 8016 /* For a similar character use SCORE_SIMILAR. */ | |
| 8017 if (slang != NULL | |
| 8018 && slang->sl_has_map | |
| 8019 && similar_chars(slang, gc, bc)) | |
| 8020 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1); | |
| 8021 else | |
| 8022 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1); | |
| 8023 } | |
| 323 | 8024 |
| 324 | 8025 if (i > 1 && j > 1) |
| 323 | 8026 { |
| 324 | 8027 #ifdef FEAT_MBYTE |
| 8028 if (has_mbyte) | |
| 8029 { | |
| 8030 pbc = wbadword[i - 2]; | |
| 8031 pgc = wgoodword[j - 2]; | |
| 8032 } | |
| 8033 else | |
| 8034 #endif | |
| 8035 { | |
| 8036 pbc = badword[i - 2]; | |
| 8037 pgc = goodword[j - 2]; | |
| 8038 } | |
| 8039 if (bc == pgc && pbc == gc) | |
| 8040 { | |
| 8041 t = SCORE_SWAP + CNT(i - 2, j - 2); | |
| 8042 if (t < CNT(i, j)) | |
| 8043 CNT(i, j) = t; | |
| 8044 } | |
| 323 | 8045 } |
| 8046 t = SCORE_DEL + CNT(i - 1, j); | |
| 8047 if (t < CNT(i, j)) | |
| 8048 CNT(i, j) = t; | |
| 8049 t = SCORE_INS + CNT(i, j - 1); | |
| 8050 if (t < CNT(i, j)) | |
| 8051 CNT(i, j) = t; | |
| 8052 } | |
| 8053 } | |
| 8054 } | |
| 344 | 8055 |
| 8056 i = CNT(badlen - 1, goodlen - 1); | |
| 8057 vim_free(cnt); | |
| 8058 return i; | |
| 323 | 8059 } |
| 307 | 8060 |
| 625 | 8061 typedef struct |
| 8062 { | |
| 8063 int badi; | |
| 8064 int goodi; | |
| 8065 int score; | |
| 8066 } limitscore_T; | |
| 8067 | |
| 8068 /* | |
| 8069 * Like spell_edit_score(), but with a limit on the score to make it faster. | |
| 8070 * May return SCORE_MAXMAX when the score is higher than "limit". | |
| 8071 * | |
| 8072 * This uses a stack for the edits still to be tried. | |
| 8073 * The idea comes from Aspell leditdist.cpp. Rewritten in C and added support | |
| 8074 * for multi-byte characters. | |
| 8075 */ | |
| 8076 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8077 spell_edit_score_limit( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8078 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8079 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8080 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8081 int limit) |
| 625 | 8082 { |
| 8083 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 8084 int stackidx; | |
| 8085 int bi, gi; | |
| 8086 int bi2, gi2; | |
| 8087 int bc, gc; | |
| 8088 int score; | |
| 8089 int score_off; | |
| 8090 int minscore; | |
| 8091 int round; | |
| 8092 | |
| 8093 #ifdef FEAT_MBYTE | |
| 8094 /* Multi-byte characters require a bit more work, use a different function | |
| 8095 * to avoid testing "has_mbyte" quite often. */ | |
| 8096 if (has_mbyte) | |
| 8097 return spell_edit_score_limit_w(slang, badword, goodword, limit); | |
| 8098 #endif | |
| 8099 | |
| 8100 /* | |
| 8101 * The idea is to go from start to end over the words. So long as | |
| 8102 * characters are equal just continue, this always gives the lowest score. | |
| 8103 * When there is a difference try several alternatives. Each alternative | |
| 8104 * increases "score" for the edit distance. Some of the alternatives are | |
| 8105 * pushed unto a stack and tried later, some are tried right away. At the | |
| 8106 * end of the word the score for one alternative is known. The lowest | |
| 8107 * possible score is stored in "minscore". | |
| 8108 */ | |
| 8109 stackidx = 0; | |
| 8110 bi = 0; | |
| 8111 gi = 0; | |
| 8112 score = 0; | |
| 8113 minscore = limit + 1; | |
| 8114 | |
| 8115 for (;;) | |
| 8116 { | |
| 8117 /* Skip over an equal part, score remains the same. */ | |
| 8118 for (;;) | |
| 8119 { | |
| 8120 bc = badword[bi]; | |
| 8121 gc = goodword[gi]; | |
| 8122 if (bc != gc) /* stop at a char that's different */ | |
| 8123 break; | |
| 8124 if (bc == NUL) /* both words end */ | |
| 8125 { | |
| 8126 if (score < minscore) | |
| 8127 minscore = score; | |
| 8128 goto pop; /* do next alternative */ | |
| 8129 } | |
| 8130 ++bi; | |
| 8131 ++gi; | |
| 8132 } | |
| 8133 | |
| 8134 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 8135 { | |
| 8136 do | |
| 8137 { | |
| 8138 if ((score += SCORE_DEL) >= minscore) | |
| 8139 goto pop; /* do next alternative */ | |
| 8140 } while (badword[++bi] != NUL); | |
| 8141 minscore = score; | |
| 8142 } | |
| 8143 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 8144 { | |
| 8145 do | |
| 8146 { | |
| 8147 if ((score += SCORE_INS) >= minscore) | |
| 8148 goto pop; /* do next alternative */ | |
| 8149 } while (goodword[++gi] != NUL); | |
| 8150 minscore = score; | |
| 8151 } | |
| 8152 else /* both words continue */ | |
| 8153 { | |
| 8154 /* If not close to the limit, perform a change. Only try changes | |
| 8155 * that may lead to a lower score than "minscore". | |
| 8156 * round 0: try deleting a char from badword | |
| 8157 * round 1: try inserting a char in badword */ | |
| 8158 for (round = 0; round <= 1; ++round) | |
| 8159 { | |
| 8160 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 8161 if (score_off < minscore) | |
| 8162 { | |
| 8163 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 8164 { | |
| 8165 /* Near the limit, rest of the words must match. We | |
| 8166 * can check that right now, no need to push an item | |
| 8167 * onto the stack. */ | |
| 8168 bi2 = bi + 1 - round; | |
| 8169 gi2 = gi + round; | |
| 8170 while (goodword[gi2] == badword[bi2]) | |
| 8171 { | |
| 8172 if (goodword[gi2] == NUL) | |
| 8173 { | |
| 8174 minscore = score_off; | |
| 8175 break; | |
| 8176 } | |
| 8177 ++bi2; | |
| 8178 ++gi2; | |
| 8179 } | |
| 8180 } | |
| 8181 else | |
| 8182 { | |
| 8183 /* try deleting/inserting a character later */ | |
| 8184 stack[stackidx].badi = bi + 1 - round; | |
| 8185 stack[stackidx].goodi = gi + round; | |
| 8186 stack[stackidx].score = score_off; | |
| 8187 ++stackidx; | |
| 8188 } | |
| 8189 } | |
| 8190 } | |
| 8191 | |
| 8192 if (score + SCORE_SWAP < minscore) | |
| 8193 { | |
| 8194 /* If swapping two characters makes a match then the | |
| 8195 * substitution is more expensive, thus there is no need to | |
| 8196 * try both. */ | |
| 8197 if (gc == badword[bi + 1] && bc == goodword[gi + 1]) | |
| 8198 { | |
| 8199 /* Swap two characters, that is: skip them. */ | |
| 8200 gi += 2; | |
| 8201 bi += 2; | |
| 8202 score += SCORE_SWAP; | |
| 8203 continue; | |
| 8204 } | |
| 8205 } | |
| 8206 | |
| 8207 /* Substitute one character for another which is the same | |
| 8208 * thing as deleting a character from both goodword and badword. | |
| 8209 * Use a better score when there is only a case difference. */ | |
| 8210 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 8211 score += SCORE_ICASE; | |
| 8212 else | |
| 8213 { | |
| 8214 /* For a similar character use SCORE_SIMILAR. */ | |
| 8215 if (slang != NULL | |
| 8216 && slang->sl_has_map | |
| 8217 && similar_chars(slang, gc, bc)) | |
| 8218 score += SCORE_SIMILAR; | |
| 8219 else | |
| 8220 score += SCORE_SUBST; | |
| 8221 } | |
| 8222 | |
| 8223 if (score < minscore) | |
| 8224 { | |
| 8225 /* Do the substitution. */ | |
| 8226 ++gi; | |
| 8227 ++bi; | |
| 8228 continue; | |
| 8229 } | |
| 8230 } | |
| 8231 pop: | |
| 8232 /* | |
| 8233 * Get here to try the next alternative, pop it from the stack. | |
| 8234 */ | |
| 8235 if (stackidx == 0) /* stack is empty, finished */ | |
| 8236 break; | |
| 8237 | |
| 8238 /* pop an item from the stack */ | |
| 8239 --stackidx; | |
| 8240 gi = stack[stackidx].goodi; | |
| 8241 bi = stack[stackidx].badi; | |
| 8242 score = stack[stackidx].score; | |
| 8243 } | |
| 8244 | |
| 8245 /* When the score goes over "limit" it may actually be much higher. | |
| 8246 * Return a very large number to avoid going below the limit when giving a | |
| 8247 * bonus. */ | |
| 8248 if (minscore > limit) | |
| 8249 return SCORE_MAXMAX; | |
| 8250 return minscore; | |
| 8251 } | |
| 8252 | |
| 8253 #ifdef FEAT_MBYTE | |
| 8254 /* | |
| 8255 * Multi-byte version of spell_edit_score_limit(). | |
| 8256 * Keep it in sync with the above! | |
| 8257 */ | |
| 8258 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8259 spell_edit_score_limit_w( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8260 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8261 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8262 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8263 int limit) |
| 625 | 8264 { |
| 8265 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 8266 int stackidx; | |
| 8267 int bi, gi; | |
| 8268 int bi2, gi2; | |
| 8269 int bc, gc; | |
| 8270 int score; | |
| 8271 int score_off; | |
| 8272 int minscore; | |
| 8273 int round; | |
| 8274 char_u *p; | |
| 8275 int wbadword[MAXWLEN]; | |
| 8276 int wgoodword[MAXWLEN]; | |
| 8277 | |
| 8278 /* Get the characters from the multi-byte strings and put them in an | |
| 8279 * int array for easy access. */ | |
| 8280 bi = 0; | |
| 8281 for (p = badword; *p != NUL; ) | |
| 8282 wbadword[bi++] = mb_cptr2char_adv(&p); | |
| 8283 wbadword[bi++] = 0; | |
| 8284 gi = 0; | |
| 8285 for (p = goodword; *p != NUL; ) | |
| 8286 wgoodword[gi++] = mb_cptr2char_adv(&p); | |
| 8287 wgoodword[gi++] = 0; | |
| 8288 | |
| 8289 /* | |
| 8290 * The idea is to go from start to end over the words. So long as | |
| 8291 * characters are equal just continue, this always gives the lowest score. | |
| 8292 * When there is a difference try several alternatives. Each alternative | |
| 8293 * increases "score" for the edit distance. Some of the alternatives are | |
| 8294 * pushed unto a stack and tried later, some are tried right away. At the | |
| 8295 * end of the word the score for one alternative is known. The lowest | |
| 8296 * possible score is stored in "minscore". | |
| 8297 */ | |
| 8298 stackidx = 0; | |
| 8299 bi = 0; | |
| 8300 gi = 0; | |
| 8301 score = 0; | |
| 8302 minscore = limit + 1; | |
| 8303 | |
| 8304 for (;;) | |
| 8305 { | |
| 8306 /* Skip over an equal part, score remains the same. */ | |
| 8307 for (;;) | |
| 8308 { | |
| 8309 bc = wbadword[bi]; | |
| 8310 gc = wgoodword[gi]; | |
| 8311 | |
| 8312 if (bc != gc) /* stop at a char that's different */ | |
| 8313 break; | |
| 8314 if (bc == NUL) /* both words end */ | |
| 8315 { | |
| 8316 if (score < minscore) | |
| 8317 minscore = score; | |
| 8318 goto pop; /* do next alternative */ | |
| 8319 } | |
| 8320 ++bi; | |
| 8321 ++gi; | |
| 8322 } | |
| 8323 | |
| 8324 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 8325 { | |
| 8326 do | |
| 8327 { | |
| 8328 if ((score += SCORE_DEL) >= minscore) | |
| 8329 goto pop; /* do next alternative */ | |
| 8330 } while (wbadword[++bi] != NUL); | |
| 8331 minscore = score; | |
| 8332 } | |
| 8333 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 8334 { | |
| 8335 do | |
| 8336 { | |
| 8337 if ((score += SCORE_INS) >= minscore) | |
| 8338 goto pop; /* do next alternative */ | |
| 8339 } while (wgoodword[++gi] != NUL); | |
| 8340 minscore = score; | |
| 8341 } | |
| 8342 else /* both words continue */ | |
| 8343 { | |
| 8344 /* If not close to the limit, perform a change. Only try changes | |
| 8345 * that may lead to a lower score than "minscore". | |
| 8346 * round 0: try deleting a char from badword | |
| 8347 * round 1: try inserting a char in badword */ | |
| 8348 for (round = 0; round <= 1; ++round) | |
| 8349 { | |
| 8350 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 8351 if (score_off < minscore) | |
| 8352 { | |
| 8353 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 8354 { | |
| 8355 /* Near the limit, rest of the words must match. We | |
| 8356 * can check that right now, no need to push an item | |
| 8357 * onto the stack. */ | |
| 8358 bi2 = bi + 1 - round; | |
| 8359 gi2 = gi + round; | |
| 8360 while (wgoodword[gi2] == wbadword[bi2]) | |
| 8361 { | |
| 8362 if (wgoodword[gi2] == NUL) | |
| 8363 { | |
| 8364 minscore = score_off; | |
| 8365 break; | |
| 8366 } | |
| 8367 ++bi2; | |
| 8368 ++gi2; | |
| 8369 } | |
| 8370 } | |
| 8371 else | |
| 8372 { | |
| 8373 /* try deleting a character from badword later */ | |
| 8374 stack[stackidx].badi = bi + 1 - round; | |
| 8375 stack[stackidx].goodi = gi + round; | |
| 8376 stack[stackidx].score = score_off; | |
| 8377 ++stackidx; | |
| 8378 } | |
| 8379 } | |
| 8380 } | |
| 8381 | |
| 8382 if (score + SCORE_SWAP < minscore) | |
| 8383 { | |
| 8384 /* If swapping two characters makes a match then the | |
| 8385 * substitution is more expensive, thus there is no need to | |
| 8386 * try both. */ | |
| 8387 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1]) | |
| 8388 { | |
| 8389 /* Swap two characters, that is: skip them. */ | |
| 8390 gi += 2; | |
| 8391 bi += 2; | |
| 8392 score += SCORE_SWAP; | |
| 8393 continue; | |
| 8394 } | |
| 8395 } | |
| 8396 | |
| 8397 /* Substitute one character for another which is the same | |
| 8398 * thing as deleting a character from both goodword and badword. | |
| 8399 * Use a better score when there is only a case difference. */ | |
| 8400 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 8401 score += SCORE_ICASE; | |
| 8402 else | |
| 8403 { | |
| 8404 /* For a similar character use SCORE_SIMILAR. */ | |
| 8405 if (slang != NULL | |
| 8406 && slang->sl_has_map | |
| 8407 && similar_chars(slang, gc, bc)) | |
| 8408 score += SCORE_SIMILAR; | |
| 8409 else | |
| 8410 score += SCORE_SUBST; | |
| 8411 } | |
| 8412 | |
| 8413 if (score < minscore) | |
| 8414 { | |
| 8415 /* Do the substitution. */ | |
| 8416 ++gi; | |
| 8417 ++bi; | |
| 8418 continue; | |
| 8419 } | |
| 8420 } | |
| 8421 pop: | |
| 8422 /* | |
| 8423 * Get here to try the next alternative, pop it from the stack. | |
| 8424 */ | |
| 8425 if (stackidx == 0) /* stack is empty, finished */ | |
| 8426 break; | |
| 8427 | |
| 8428 /* pop an item from the stack */ | |
| 8429 --stackidx; | |
| 8430 gi = stack[stackidx].goodi; | |
| 8431 bi = stack[stackidx].badi; | |
| 8432 score = stack[stackidx].score; | |
| 8433 } | |
| 8434 | |
| 8435 /* When the score goes over "limit" it may actually be much higher. | |
| 8436 * Return a very large number to avoid going below the limit when giving a | |
| 8437 * bonus. */ | |
| 8438 if (minscore > limit) | |
| 8439 return SCORE_MAXMAX; | |
| 8440 return minscore; | |
| 8441 } | |
| 8442 #endif | |
| 8443 | |
| 714 | 8444 /* |
| 8445 * ":spellinfo" | |
| 8446 */ | |
| 8447 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8448 ex_spellinfo(exarg_T *eap UNUSED) |
| 714 | 8449 { |
| 8450 int lpi; | |
| 8451 langp_T *lp; | |
| 8452 char_u *p; | |
| 8453 | |
| 8454 if (no_spell_checking(curwin)) | |
| 8455 return; | |
| 8456 | |
| 8457 msg_start(); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8458 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
|
8459 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8460 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 714 | 8461 msg_puts((char_u *)"file: "); |
| 8462 msg_puts(lp->lp_slang->sl_fname); | |
| 8463 msg_putchar('\n'); | |
| 8464 p = lp->lp_slang->sl_info; | |
| 8465 if (p != NULL) | |
| 8466 { | |
| 8467 msg_puts(p); | |
| 8468 msg_putchar('\n'); | |
| 8469 } | |
| 8470 } | |
| 8471 msg_end(); | |
| 8472 } | |
| 8473 | |
| 625 | 8474 #define DUMPFLAG_KEEPCASE 1 /* round 2: keep-case tree */ |
| 8475 #define DUMPFLAG_COUNT 2 /* include word count */ | |
| 701 | 8476 #define DUMPFLAG_ICASE 4 /* ignore case when finding matches */ |
| 710 | 8477 #define DUMPFLAG_ONECAP 8 /* pattern starts with capital */ |
| 8478 #define DUMPFLAG_ALLCAP 16 /* pattern is all capitals */ | |
| 625 | 8479 |
| 351 | 8480 /* |
| 8481 * ":spelldump" | |
| 8482 */ | |
| 8483 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8484 ex_spelldump(exarg_T *eap) |
| 351 | 8485 { |
| 5382 | 8486 char_u *spl; |
| 8487 long dummy; | |
| 8488 | |
| 701 | 8489 if (no_spell_checking(curwin)) |
| 8490 return; | |
| 5382 | 8491 get_option_value((char_u*)"spl", &dummy, &spl, OPT_LOCAL); |
| 8492 | |
| 8493 /* Create a new empty buffer in a new window. */ | |
| 701 | 8494 do_cmdline_cmd((char_u *)"new"); |
| 5382 | 8495 |
| 8496 /* enable spelling locally in the new window */ | |
| 8497 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
|
8498 set_option_value((char_u*)"spl", dummy, spl, OPT_LOCAL); |
| 5382 | 8499 vim_free(spl); |
| 8500 | |
|
11121
778c10516955
patch 8.0.0448: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
10950
diff
changeset
|
8501 if (!BUFEMPTY()) |
| 701 | 8502 return; |
| 8503 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8504 spell_dump_compl(NULL, 0, NULL, eap->forceit ? DUMPFLAG_COUNT : 0); |
| 701 | 8505 |
| 8506 /* Delete the empty line that we started with. */ | |
| 8507 if (curbuf->b_ml.ml_line_count > 1) | |
| 8508 ml_delete(curbuf->b_ml.ml_line_count, FALSE); | |
| 8509 | |
| 8510 redraw_later(NOT_VALID); | |
| 8511 } | |
| 8512 | |
| 8513 /* | |
| 8514 * Go through all possible words and: | |
| 8515 * 1. When "pat" is NULL: dump a list of all words in the current buffer. | |
| 8516 * "ic" and "dir" are not used. | |
| 8517 * 2. When "pat" is not NULL: add matching words to insert mode completion. | |
| 8518 */ | |
| 8519 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8520 spell_dump_compl( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8521 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
|
8522 int ic, /* ignore case */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8523 int *dir, /* direction for adding matches */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8524 int dumpflags_arg) /* DUMPFLAG_* */ |
| 701 | 8525 { |
| 351 | 8526 langp_T *lp; |
| 8527 slang_T *slang; | |
| 8528 idx_T arridx[MAXWLEN]; | |
| 8529 int curi[MAXWLEN]; | |
| 8530 char_u word[MAXWLEN]; | |
| 8531 int c; | |
| 8532 char_u *byts; | |
| 8533 idx_T *idxs; | |
| 8534 linenr_T lnum = 0; | |
| 8535 int round; | |
| 8536 int depth; | |
| 8537 int n; | |
| 8538 int flags; | |
| 381 | 8539 char_u *region_names = NULL; /* region names being used */ |
| 8540 int do_region = TRUE; /* dump region names and numbers */ | |
| 8541 char_u *p; | |
| 500 | 8542 int lpi; |
| 701 | 8543 int dumpflags = dumpflags_arg; |
| 8544 int patlen; | |
| 8545 | |
| 710 | 8546 /* When ignoring case or when the pattern starts with capital pass this on |
| 8547 * to dump_word(). */ | |
| 8548 if (pat != NULL) | |
| 8549 { | |
| 8550 if (ic) | |
| 8551 dumpflags |= DUMPFLAG_ICASE; | |
| 8552 else | |
| 8553 { | |
| 8554 n = captype(pat, NULL); | |
| 8555 if (n == WF_ONECAP) | |
| 8556 dumpflags |= DUMPFLAG_ONECAP; | |
| 8557 else if (n == WF_ALLCAP | |
| 8558 #ifdef FEAT_MBYTE | |
| 714 | 8559 && (int)STRLEN(pat) > mb_ptr2len(pat) |
| 710 | 8560 #else |
| 714 | 8561 && (int)STRLEN(pat) > 1 |
| 710 | 8562 #endif |
| 8563 ) | |
| 8564 dumpflags |= DUMPFLAG_ALLCAP; | |
| 8565 } | |
| 8566 } | |
| 351 | 8567 |
| 381 | 8568 /* Find out if we can support regions: All languages must support the same |
| 8569 * regions or none at all. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8570 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
|
8571 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8572 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 381 | 8573 p = lp->lp_slang->sl_regions; |
| 8574 if (p[0] != 0) | |
| 8575 { | |
| 8576 if (region_names == NULL) /* first language with regions */ | |
| 8577 region_names = p; | |
| 8578 else if (STRCMP(region_names, p) != 0) | |
| 8579 { | |
| 8580 do_region = FALSE; /* region names are different */ | |
| 8581 break; | |
| 8582 } | |
| 8583 } | |
| 8584 } | |
| 8585 | |
| 8586 if (do_region && region_names != NULL) | |
| 8587 { | |
| 701 | 8588 if (pat == NULL) |
| 8589 { | |
| 8590 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names); | |
| 8591 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 8592 } | |
| 381 | 8593 } |
| 8594 else | |
| 8595 do_region = FALSE; | |
| 8596 | |
| 8597 /* | |
| 8598 * Loop over all files loaded for the entries in 'spelllang'. | |
| 8599 */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8600 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
|
8601 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8602 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 351 | 8603 slang = lp->lp_slang; |
| 500 | 8604 if (slang->sl_fbyts == NULL) /* reloading failed */ |
| 8605 continue; | |
| 351 | 8606 |
| 701 | 8607 if (pat == NULL) |
| 8608 { | |
| 8609 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname); | |
| 8610 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 8611 } | |
| 8612 | |
| 8613 /* When matching with a pattern and there are no prefixes only use | |
| 8614 * parts of the tree that match "pat". */ | |
| 8615 if (pat != NULL && slang->sl_pbyts == NULL) | |
| 835 | 8616 patlen = (int)STRLEN(pat); |
| 701 | 8617 else |
| 840 | 8618 patlen = -1; |
| 351 | 8619 |
| 8620 /* round 1: case-folded tree | |
| 8621 * round 2: keep-case tree */ | |
| 8622 for (round = 1; round <= 2; ++round) | |
| 8623 { | |
| 8624 if (round == 1) | |
| 8625 { | |
| 701 | 8626 dumpflags &= ~DUMPFLAG_KEEPCASE; |
| 351 | 8627 byts = slang->sl_fbyts; |
| 8628 idxs = slang->sl_fidxs; | |
| 8629 } | |
| 8630 else | |
| 8631 { | |
| 701 | 8632 dumpflags |= DUMPFLAG_KEEPCASE; |
| 351 | 8633 byts = slang->sl_kbyts; |
| 8634 idxs = slang->sl_kidxs; | |
| 8635 } | |
| 8636 if (byts == NULL) | |
| 8637 continue; /* array is empty */ | |
| 8638 | |
| 8639 depth = 0; | |
| 8640 arridx[0] = 0; | |
| 8641 curi[0] = 1; | |
| 701 | 8642 while (depth >= 0 && !got_int |
| 8643 && (pat == NULL || !compl_interrupted)) | |
| 351 | 8644 { |
| 8645 if (curi[depth] > byts[arridx[depth]]) | |
| 8646 { | |
| 8647 /* Done all bytes at this node, go up one level. */ | |
| 8648 --depth; | |
| 8649 line_breakcheck(); | |
|
10277
154d5a2e7395
commit https://github.com/vim/vim/commit/472e85970ee3a80abd824bef510df12e9cfe9e96
Christian Brabandt <cb@256bit.org>
parents:
10042
diff
changeset
|
8650 ins_compl_check_keys(50, FALSE); |
| 351 | 8651 } |
| 8652 else | |
| 8653 { | |
| 8654 /* Do one more byte at this node. */ | |
| 8655 n = arridx[depth] + curi[depth]; | |
| 8656 ++curi[depth]; | |
| 8657 c = byts[n]; | |
| 8658 if (c == 0) | |
| 8659 { | |
| 8660 /* End of word, deal with the word. | |
| 8661 * Don't use keep-case words in the fold-case tree, | |
| 8662 * they will appear in the keep-case tree. | |
| 8663 * Only use the word when the region matches. */ | |
| 8664 flags = (int)idxs[n]; | |
| 8665 if ((round == 2 || (flags & WF_KEEPCAP) == 0) | |
| 500 | 8666 && (flags & WF_NEEDCOMP) == 0 |
| 381 | 8667 && (do_region |
| 8668 || (flags & WF_REGION) == 0 | |
| 390 | 8669 || (((unsigned)flags >> 16) |
| 351 | 8670 & lp->lp_region) != 0)) |
| 8671 { | |
| 8672 word[depth] = NUL; | |
| 381 | 8673 if (!do_region) |
| 8674 flags &= ~WF_REGION; | |
| 355 | 8675 |
| 8676 /* Dump the basic word if there is no prefix or | |
| 8677 * when it's the first one. */ | |
| 390 | 8678 c = (unsigned)flags >> 24; |
| 355 | 8679 if (c == 0 || curi[depth] == 2) |
| 701 | 8680 { |
| 8681 dump_word(slang, word, pat, dir, | |
| 8682 dumpflags, flags, lnum); | |
| 8683 if (pat == NULL) | |
| 8684 ++lnum; | |
| 8685 } | |
| 351 | 8686 |
| 8687 /* Apply the prefix, if there is one. */ | |
| 355 | 8688 if (c != 0) |
| 701 | 8689 lnum = dump_prefixes(slang, word, pat, dir, |
| 8690 dumpflags, flags, lnum); | |
| 351 | 8691 } |
| 8692 } | |
| 8693 else | |
| 8694 { | |
| 8695 /* Normal char, go one level deeper. */ | |
| 8696 word[depth++] = c; | |
| 8697 arridx[depth] = idxs[n]; | |
| 8698 curi[depth] = 1; | |
| 701 | 8699 |
| 8700 /* Check if this characters matches with the pattern. | |
| 8701 * If not skip the whole tree below it. | |
| 710 | 8702 * Always ignore case here, dump_word() will check |
| 8703 * proper case later. This isn't exactly right when | |
| 8704 * length changes for multi-byte characters with | |
| 8705 * ignore case... */ | |
| 8706 if (depth <= patlen | |
| 8707 && MB_STRNICMP(word, pat, depth) != 0) | |
| 701 | 8708 --depth; |
| 351 | 8709 } |
| 8710 } | |
| 8711 } | |
| 8712 } | |
| 8713 } | |
| 8714 } | |
| 8715 | |
| 8716 /* | |
| 8717 * Dump one word: apply case modifications and append a line to the buffer. | |
| 701 | 8718 * When "lnum" is zero add insert mode completion. |
| 8719 */ | |
| 8720 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8721 dump_word( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8722 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8723 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8724 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8725 int *dir, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8726 int dumpflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8727 int wordflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8728 linenr_T lnum) |
| 351 | 8729 { |
| 8730 int keepcap = FALSE; | |
| 8731 char_u *p; | |
| 625 | 8732 char_u *tw; |
| 351 | 8733 char_u cword[MAXWLEN]; |
| 381 | 8734 char_u badword[MAXWLEN + 10]; |
| 8735 int i; | |
| 710 | 8736 int flags = wordflags; |
| 8737 | |
| 8738 if (dumpflags & DUMPFLAG_ONECAP) | |
| 8739 flags |= WF_ONECAP; | |
| 8740 if (dumpflags & DUMPFLAG_ALLCAP) | |
| 8741 flags |= WF_ALLCAP; | |
| 351 | 8742 |
| 625 | 8743 if ((dumpflags & DUMPFLAG_KEEPCASE) == 0 && (flags & WF_CAPMASK) != 0) |
| 351 | 8744 { |
| 8745 /* Need to fix case according to "flags". */ | |
| 8746 make_case_word(word, cword, flags); | |
| 8747 p = cword; | |
| 8748 } | |
| 8749 else | |
| 8750 { | |
| 8751 p = word; | |
| 625 | 8752 if ((dumpflags & DUMPFLAG_KEEPCASE) |
| 8753 && ((captype(word, NULL) & WF_KEEPCAP) == 0 | |
| 389 | 8754 || (flags & WF_FIXCAP) != 0)) |
| 351 | 8755 keepcap = TRUE; |
| 8756 } | |
| 625 | 8757 tw = p; |
| 351 | 8758 |
| 701 | 8759 if (pat == NULL) |
| 8760 { | |
| 8761 /* Add flags and regions after a slash. */ | |
| 8762 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap) | |
| 8763 { | |
| 8764 STRCPY(badword, p); | |
| 8765 STRCAT(badword, "/"); | |
| 8766 if (keepcap) | |
| 8767 STRCAT(badword, "="); | |
| 8768 if (flags & WF_BANNED) | |
| 8769 STRCAT(badword, "!"); | |
| 8770 else if (flags & WF_RARE) | |
| 8771 STRCAT(badword, "?"); | |
| 8772 if (flags & WF_REGION) | |
| 8773 for (i = 0; i < 7; ++i) | |
| 8774 if (flags & (0x10000 << i)) | |
| 8775 sprintf((char *)badword + STRLEN(badword), "%d", i + 1); | |
| 8776 p = badword; | |
| 8777 } | |
| 8778 | |
| 8779 if (dumpflags & DUMPFLAG_COUNT) | |
| 8780 { | |
| 8781 hashitem_T *hi; | |
| 8782 | |
| 8783 /* Include the word count for ":spelldump!". */ | |
| 8784 hi = hash_find(&slang->sl_wordcount, tw); | |
| 8785 if (!HASHITEM_EMPTY(hi)) | |
| 8786 { | |
| 8787 vim_snprintf((char *)IObuff, IOSIZE, "%s\t%d", | |
| 625 | 8788 tw, HI2WC(hi)->wc_count); |
| 701 | 8789 p = IObuff; |
| 8790 } | |
| 8791 } | |
| 8792 | |
| 8793 ml_append(lnum, p, (colnr_T)0, FALSE); | |
| 8794 } | |
| 710 | 8795 else if (((dumpflags & DUMPFLAG_ICASE) |
| 8796 ? MB_STRNICMP(p, pat, STRLEN(pat)) == 0 | |
| 8797 : STRNCMP(p, pat, STRLEN(pat)) == 0) | |
| 701 | 8798 && ins_compl_add_infercase(p, (int)STRLEN(p), |
| 942 | 8799 p_ic, NULL, *dir, 0) == OK) |
| 710 | 8800 /* if dir was BACKWARD then honor it just once */ |
| 8801 *dir = FORWARD; | |
| 351 | 8802 } |
| 8803 | |
| 8804 /* | |
| 372 | 8805 * For ":spelldump": Find matching prefixes for "word". Prepend each to |
| 8806 * "word" and append a line to the buffer. | |
| 701 | 8807 * When "lnum" is zero add insert mode completion. |
| 351 | 8808 * Return the updated line number. |
| 8809 */ | |
| 8810 static linenr_T | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8811 dump_prefixes( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8812 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8813 char_u *word, /* case-folded word */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8814 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8815 int *dir, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8816 int dumpflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8817 int flags, /* flags with prefix ID */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8818 linenr_T startlnum) |
| 351 | 8819 { |
| 8820 idx_T arridx[MAXWLEN]; | |
| 8821 int curi[MAXWLEN]; | |
| 8822 char_u prefix[MAXWLEN]; | |
| 455 | 8823 char_u word_up[MAXWLEN]; |
| 8824 int has_word_up = FALSE; | |
| 351 | 8825 int c; |
| 8826 char_u *byts; | |
| 8827 idx_T *idxs; | |
| 8828 linenr_T lnum = startlnum; | |
| 8829 int depth; | |
| 8830 int n; | |
| 8831 int len; | |
| 8832 int i; | |
| 8833 | |
| 626 | 8834 /* If the word starts with a lower-case letter make the word with an |
| 455 | 8835 * upper-case letter in word_up[]. */ |
| 8836 c = PTR2CHAR(word); | |
| 8837 if (SPELL_TOUPPER(c) != c) | |
| 8838 { | |
| 8839 onecap_copy(word, word_up, TRUE); | |
| 8840 has_word_up = TRUE; | |
| 8841 } | |
| 8842 | |
| 351 | 8843 byts = slang->sl_pbyts; |
| 8844 idxs = slang->sl_pidxs; | |
| 8845 if (byts != NULL) /* array not is empty */ | |
| 8846 { | |
| 8847 /* | |
| 8848 * Loop over all prefixes, building them byte-by-byte in prefix[]. | |
| 390 | 8849 * When at the end of a prefix check that it supports "flags". |
| 351 | 8850 */ |
| 8851 depth = 0; | |
| 8852 arridx[0] = 0; | |
| 8853 curi[0] = 1; | |
| 8854 while (depth >= 0 && !got_int) | |
| 8855 { | |
| 390 | 8856 n = arridx[depth]; |
| 8857 len = byts[n]; | |
| 8858 if (curi[depth] > len) | |
| 351 | 8859 { |
| 8860 /* Done all bytes at this node, go up one level. */ | |
| 8861 --depth; | |
| 8862 line_breakcheck(); | |
| 8863 } | |
| 8864 else | |
| 8865 { | |
| 8866 /* Do one more byte at this node. */ | |
| 390 | 8867 n += curi[depth]; |
| 351 | 8868 ++curi[depth]; |
| 8869 c = byts[n]; | |
| 8870 if (c == 0) | |
| 8871 { | |
| 8872 /* End of prefix, find out how many IDs there are. */ | |
| 8873 for (i = 1; i < len; ++i) | |
| 8874 if (byts[n + i] != 0) | |
| 8875 break; | |
| 8876 curi[depth] += i - 1; | |
| 8877 | |
| 455 | 8878 c = valid_word_prefix(i, n, flags, word, slang, FALSE); |
| 8879 if (c != 0) | |
| 351 | 8880 { |
| 376 | 8881 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1); |
| 701 | 8882 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 8883 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 8884 : flags, lnum); |
| 8885 if (lnum != 0) | |
| 8886 ++lnum; | |
| 351 | 8887 } |
| 455 | 8888 |
| 8889 /* Check for prefix that matches the word when the | |
| 8890 * first letter is upper-case, but only if the prefix has | |
| 8891 * a condition. */ | |
| 8892 if (has_word_up) | |
| 8893 { | |
| 8894 c = valid_word_prefix(i, n, flags, word_up, slang, | |
| 8895 TRUE); | |
| 8896 if (c != 0) | |
| 8897 { | |
| 8898 vim_strncpy(prefix + depth, word_up, | |
| 8899 MAXWLEN - depth - 1); | |
| 701 | 8900 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 8901 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 8902 : flags, lnum); |
| 8903 if (lnum != 0) | |
| 8904 ++lnum; | |
| 455 | 8905 } |
| 8906 } | |
| 351 | 8907 } |
| 8908 else | |
| 8909 { | |
| 8910 /* Normal char, go one level deeper. */ | |
| 8911 prefix[depth++] = c; | |
| 8912 arridx[depth] = idxs[n]; | |
| 8913 curi[depth] = 1; | |
| 8914 } | |
| 8915 } | |
| 8916 } | |
| 8917 } | |
| 8918 | |
| 8919 return lnum; | |
| 8920 } | |
| 8921 | |
| 498 | 8922 /* |
| 626 | 8923 * Move "p" to the end of word "start". |
| 8924 * Uses the spell-checking word characters. | |
| 498 | 8925 */ |
| 8926 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8927 spell_to_word_end(char_u *start, win_T *win) |
| 498 | 8928 { |
| 8929 char_u *p = start; | |
| 8930 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8931 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
|
8932 MB_PTR_ADV(p); |
| 498 | 8933 return p; |
| 8934 } | |
| 8935 | |
| 475 | 8936 #if defined(FEAT_INS_EXPAND) || defined(PROTO) |
| 8937 /* | |
| 626 | 8938 * For Insert mode completion CTRL-X s: |
| 8939 * Find start of the word in front of column "startcol". | |
| 8940 * We don't check if it is badly spelled, with completion we can only change | |
| 8941 * the word in front of the cursor. | |
| 475 | 8942 * Returns the column number of the word. |
| 8943 */ | |
| 8944 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8945 spell_word_start(int startcol) |
| 475 | 8946 { |
| 8947 char_u *line; | |
| 8948 char_u *p; | |
| 8949 int col = 0; | |
| 8950 | |
| 498 | 8951 if (no_spell_checking(curwin)) |
| 475 | 8952 return startcol; |
| 8953 | |
| 8954 /* Find a word character before "startcol". */ | |
| 8955 line = ml_get_curline(); | |
| 8956 for (p = line + startcol; p > line; ) | |
| 8957 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
8958 MB_PTR_BACK(line, p); |
| 5477 | 8959 if (spell_iswordp_nmw(p, curwin)) |
| 475 | 8960 break; |
| 8961 } | |
| 8962 | |
| 8963 /* Go back to start of the word. */ | |
| 8964 while (p > line) | |
| 8965 { | |
| 835 | 8966 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
|
8967 MB_PTR_BACK(line, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8968 if (!spell_iswordp(p, curwin)) |
| 475 | 8969 break; |
| 8970 col = 0; | |
| 8971 } | |
| 8972 | |
| 535 | 8973 return col; |
| 8974 } | |
| 8975 | |
| 8976 /* | |
| 8977 * Need to check for 'spellcapcheck' now, the word is removed before | |
| 8978 * expand_spelling() is called. Therefore the ugly global variable. | |
| 8979 */ | |
| 8980 static int spell_expand_need_cap; | |
| 8981 | |
| 8982 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8983 spell_expand_check_cap(colnr_T col) |
| 535 | 8984 { |
| 475 | 8985 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col); |
| 8986 } | |
| 8987 | |
| 8988 /* | |
| 8989 * Get list of spelling suggestions. | |
| 8990 * Used for Insert mode completion CTRL-X ?. | |
| 8991 * Returns the number of matches. The matches are in "matchp[]", array of | |
| 8992 * allocated strings. | |
| 8993 */ | |
| 8994 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8995 expand_spelling( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8996 linenr_T lnum UNUSED, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8997 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8998 char_u ***matchp) |
| 475 | 8999 { |
| 9000 garray_T ga; | |
| 9001 | |
| 625 | 9002 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap, TRUE); |
| 475 | 9003 *matchp = ga.ga_data; |
| 9004 return ga.ga_len; | |
| 9005 } | |
| 9006 #endif | |
| 9007 | |
| 737 | 9008 #endif /* FEAT_SPELL */ |