Mercurial > vim
annotate src/spell.c @ 14862:27b9a84395b5 v8.1.0443
patch 8.1.0443: unnecessary static function prototypes
commit https://github.com/vim/vim/commit/6dff58f15cede9139b2fcfc64c9064326ea3d3b0
Author: Bram Moolenaar <Bram@vim.org>
Date: Sun Sep 30 21:43:26 2018 +0200
patch 8.1.0443: unnecessary static function prototypes
Problem: Unnecessary static function prototypes.
Solution: Remove unnecessary prototypes.
| author | Christian Brabandt <cb@256bit.org> |
|---|---|
| date | Sun, 30 Sep 2018 21:45:07 +0200 |
| parents | 69517d67421f |
| children | 6e4e0d43b20b |
| 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. */ | |
| 720 EMSG(_(e_format)); | |
| 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 { |
| 1540 EMSG(_("E756: Spell checking is not enabled")); | |
| 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 { |
| 2660 | 1891 smsg((char_u *) |
| 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 */ |
| 2310 char_u * | |
|
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; |
| 2333 char_u *ret_msg = NULL; | |
| 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 { |
|
9953
9560a2eb7968
commit https://github.com/vim/vim/commit/5b30291785e6b9be1a607504c14bd03c601b59a6
Christian Brabandt <cb@256bit.org>
parents:
9898
diff
changeset
|
2438 ret_msg = (char_u *)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. */ | |
| 351 | 2467 smsg((char_u *) |
| 2468 _("Warning: region %s not supported"), | |
| 2469 region); | |
| 307 | 2470 } |
| 2471 else | |
| 2472 region_mask = 1 << c; | |
| 236 | 2473 } |
| 307 | 2474 |
| 389 | 2475 if (region_mask != 0) |
| 307 | 2476 { |
| 389 | 2477 if (ga_grow(&ga, 1) == FAIL) |
| 2478 { | |
| 2479 ga_clear(&ga); | |
| 1185 | 2480 ret_msg = e_outofmem; |
| 2481 goto theend; | |
| 389 | 2482 } |
| 503 | 2483 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 389 | 2484 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 2485 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2486 use_midword(slang, wp); |
| 503 | 2487 if (slang->sl_nobreak) |
| 501 | 2488 nobreak = TRUE; |
| 307 | 2489 } |
| 385 | 2490 } |
| 2491 } | |
| 2492 | |
| 2493 /* round 0: load int_wordlist, if possible. | |
| 2494 * round 1: load first name in 'spellfile'. | |
| 2495 * round 2: load second name in 'spellfile. | |
| 2496 * etc. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2497 spf = curwin->w_s->b_p_spf; |
| 385 | 2498 for (round = 0; round == 0 || *spf != NUL; ++round) |
| 2499 { | |
| 2500 if (round == 0) | |
| 2501 { | |
| 2502 /* Internal wordlist, if there is one. */ | |
| 2503 if (int_wordlist == NULL) | |
| 381 | 2504 continue; |
| 385 | 2505 int_wordlist_spl(spf_name); |
| 381 | 2506 } |
| 2507 else | |
| 2508 { | |
| 385 | 2509 /* One entry in 'spellfile'. */ |
| 2510 copy_option_part(&spf, spf_name, MAXPATHL - 5, ","); | |
| 2511 STRCAT(spf_name, ".spl"); | |
| 2512 | |
| 2513 /* If it was already found above then skip it. */ | |
| 2514 for (c = 0; c < ga.ga_len; ++c) | |
| 500 | 2515 { |
| 2516 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname; | |
| 2517 if (p != NULL && fullpathcmp(spf_name, p, FALSE) == FPC_SAME) | |
| 385 | 2518 break; |
| 500 | 2519 } |
| 385 | 2520 if (c < ga.ga_len) |
| 381 | 2521 continue; |
| 2522 } | |
| 2523 | |
| 351 | 2524 /* Check if it was loaded already. */ |
| 503 | 2525 for (slang = first_lang; slang != NULL; slang = slang->sl_next) |
| 2526 if (fullpathcmp(spf_name, slang->sl_fname, FALSE) == FPC_SAME) | |
| 323 | 2527 break; |
| 503 | 2528 if (slang == NULL) |
| 323 | 2529 { |
| 351 | 2530 /* Not loaded, try loading it now. The language name includes the |
| 385 | 2531 * region name, the region is ignored otherwise. for int_wordlist |
| 2532 * use an arbitrary name. */ | |
| 2533 if (round == 0) | |
| 2534 STRCPY(lang, "internal wordlist"); | |
| 2535 else | |
| 2536 { | |
| 2537 vim_strncpy(lang, gettail(spf_name), MAXWLEN); | |
| 381 | 2538 p = vim_strchr(lang, '.'); |
| 2539 if (p != NULL) | |
| 2540 *p = NUL; /* truncate at ".encoding.add" */ | |
| 2541 } | |
| 503 | 2542 slang = spell_load_file(spf_name, lang, NULL, TRUE); |
| 501 | 2543 |
| 2544 /* If one of the languages has NOBREAK we assume the addition | |
| 2545 * files also have this. */ | |
| 503 | 2546 if (slang != NULL && nobreak) |
| 2547 slang->sl_nobreak = TRUE; | |
| 2548 } | |
| 2549 if (slang != NULL && ga_grow(&ga, 1) == OK) | |
| 323 | 2550 { |
| 389 | 2551 region_mask = REGION_ALL; |
| 2552 if (use_region != NULL && !dont_use_region) | |
| 2553 { | |
| 2554 /* find region in sl_regions */ | |
| 503 | 2555 c = find_region(slang->sl_regions, use_region); |
| 389 | 2556 if (c != REGION_ALL) |
| 2557 region_mask = 1 << c; | |
| 503 | 2558 else if (*slang->sl_regions != NUL) |
| 389 | 2559 /* This spell file is for other regions. */ |
| 2560 region_mask = 0; | |
| 2561 } | |
| 2562 | |
| 2563 if (region_mask != 0) | |
| 2564 { | |
| 503 | 2565 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang; |
| 2566 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL; | |
| 2567 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL; | |
| 389 | 2568 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask; |
| 2569 ++ga.ga_len; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2570 use_midword(slang, wp); |
| 389 | 2571 } |
| 323 | 2572 } |
| 2573 } | |
| 2574 | |
| 236 | 2575 /* 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
|
2576 ga_clear(&wp->w_s->b_langp); |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2577 wp->w_s->b_langp = ga; |
| 236 | 2578 |
| 503 | 2579 /* For each language figure out what language to use for sound folding and |
| 2580 * REP items. If the language doesn't support it itself use another one | |
| 2581 * with the same name. E.g. for "en-math" use "en". */ | |
| 2582 for (i = 0; i < ga.ga_len; ++i) | |
| 2583 { | |
| 2584 lp = LANGP_ENTRY(ga, i); | |
| 2585 | |
| 2586 /* sound folding */ | |
| 2587 if (lp->lp_slang->sl_sal.ga_len > 0) | |
| 2588 /* language does sound folding itself */ | |
| 2589 lp->lp_sallang = lp->lp_slang; | |
| 2590 else | |
| 2591 /* find first similar language that does sound folding */ | |
| 2592 for (j = 0; j < ga.ga_len; ++j) | |
| 2593 { | |
| 2594 lp2 = LANGP_ENTRY(ga, j); | |
| 2595 if (lp2->lp_slang->sl_sal.ga_len > 0 | |
| 2596 && STRNCMP(lp->lp_slang->sl_name, | |
| 2597 lp2->lp_slang->sl_name, 2) == 0) | |
| 2598 { | |
| 2599 lp->lp_sallang = lp2->lp_slang; | |
| 2600 break; | |
| 2601 } | |
| 2602 } | |
| 2603 | |
| 2604 /* REP items */ | |
| 2605 if (lp->lp_slang->sl_rep.ga_len > 0) | |
| 2606 /* language has REP items itself */ | |
| 2607 lp->lp_replang = lp->lp_slang; | |
| 2608 else | |
| 625 | 2609 /* find first similar language that has REP items */ |
| 503 | 2610 for (j = 0; j < ga.ga_len; ++j) |
| 2611 { | |
| 2612 lp2 = LANGP_ENTRY(ga, j); | |
| 2613 if (lp2->lp_slang->sl_rep.ga_len > 0 | |
| 2614 && STRNCMP(lp->lp_slang->sl_name, | |
| 2615 lp2->lp_slang->sl_name, 2) == 0) | |
| 2616 { | |
| 2617 lp->lp_replang = lp2->lp_slang; | |
| 2618 break; | |
| 2619 } | |
| 2620 } | |
| 2621 } | |
| 2622 | |
| 1185 | 2623 theend: |
| 2624 vim_free(spl_copy); | |
| 2625 recursive = FALSE; | |
| 5891 | 2626 redraw_win_later(wp, NOT_VALID); |
| 1185 | 2627 return ret_msg; |
| 236 | 2628 } |
| 2629 | |
| 2630 /* | |
| 376 | 2631 * Clear the midword characters for buffer "buf". |
| 2632 */ | |
| 2633 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2634 clear_midword(win_T *wp) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2635 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2636 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
|
2637 #ifdef FEAT_MBYTE |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2638 VIM_CLEAR(wp->w_s->b_spell_ismw_mb); |
| 376 | 2639 #endif |
| 2640 } | |
| 2641 | |
| 2642 /* | |
| 2643 * Use the "sl_midword" field of language "lp" for buffer "buf". | |
| 2644 * They add up to any currently used midword characters. | |
| 2645 */ | |
| 2646 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2647 use_midword(slang_T *lp, win_T *wp) |
| 376 | 2648 { |
| 2649 char_u *p; | |
| 2650 | |
| 389 | 2651 if (lp->sl_midword == NULL) /* there aren't any */ |
| 2652 return; | |
| 2653 | |
| 376 | 2654 for (p = lp->sl_midword; *p != NUL; ) |
| 2655 #ifdef FEAT_MBYTE | |
| 2656 if (has_mbyte) | |
| 2657 { | |
| 2658 int c, l, n; | |
| 2659 char_u *bp; | |
| 2660 | |
| 2661 c = mb_ptr2char(p); | |
| 474 | 2662 l = (*mb_ptr2len)(p); |
| 2663 if (c < 256 && l <= 2) | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2664 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
|
2665 else if (wp->w_s->b_spell_ismw_mb == NULL) |
| 376 | 2666 /* 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
|
2667 wp->w_s->b_spell_ismw_mb = vim_strnsave(p, l); |
| 376 | 2668 else |
| 2669 { | |
| 2670 /* 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
|
2671 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
|
2672 bp = vim_strnsave(wp->w_s->b_spell_ismw_mb, n + l); |
| 376 | 2673 if (bp != NULL) |
| 2674 { | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2675 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
|
2676 wp->w_s->b_spell_ismw_mb = bp; |
| 376 | 2677 vim_strncpy(bp + n, p, l); |
| 2678 } | |
| 2679 } | |
| 2680 p += l; | |
| 2681 } | |
| 2682 else | |
| 2683 #endif | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2684 wp->w_s->b_spell_ismw[*p++] = TRUE; |
| 376 | 2685 } |
| 2686 | |
| 2687 /* | |
| 236 | 2688 * Find the region "region[2]" in "rp" (points to "sl_regions"). |
| 2689 * Each region is simply stored as the two characters of it's name. | |
| 381 | 2690 * Returns the index if found (first is 0), REGION_ALL if not found. |
| 236 | 2691 */ |
| 2692 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2693 find_region(char_u *rp, char_u *region) |
| 236 | 2694 { |
| 2695 int i; | |
| 2696 | |
| 2697 for (i = 0; ; i += 2) | |
| 2698 { | |
| 2699 if (rp[i] == NUL) | |
| 2700 return REGION_ALL; | |
| 2701 if (rp[i] == region[0] && rp[i + 1] == region[1]) | |
| 2702 break; | |
| 2703 } | |
| 2704 return i / 2; | |
| 2705 } | |
| 2706 | |
| 2707 /* | |
| 323 | 2708 * Return case type of word: |
| 236 | 2709 * w word 0 |
| 300 | 2710 * Word WF_ONECAP |
| 2711 * W WORD WF_ALLCAP | |
| 2712 * WoRd wOrd WF_KEEPCAP | |
| 236 | 2713 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2714 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2715 captype( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2716 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2717 char_u *end) /* When NULL use up to NUL byte. */ |
| 236 | 2718 { |
| 2719 char_u *p; | |
| 2720 int c; | |
| 2721 int firstcap; | |
| 2722 int allcap; | |
| 2723 int past_second = FALSE; /* past second word char */ | |
| 2724 | |
| 2725 /* find first letter */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2726 for (p = word; !spell_iswordp_nmw(p, curwin); MB_PTR_ADV(p)) |
| 323 | 2727 if (end == NULL ? *p == NUL : p >= end) |
| 236 | 2728 return 0; /* only non-word characters, illegal word */ |
| 2729 #ifdef FEAT_MBYTE | |
| 310 | 2730 if (has_mbyte) |
| 2731 c = mb_ptr2char_adv(&p); | |
| 2732 else | |
| 236 | 2733 #endif |
| 310 | 2734 c = *p++; |
| 324 | 2735 firstcap = allcap = SPELL_ISUPPER(c); |
| 236 | 2736 |
| 2737 /* | |
| 2738 * Need to check all letters to find a word with mixed upper/lower. | |
| 2739 * But a word with an upper char only at start is a ONECAP. | |
| 2740 */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2741 for ( ; end == NULL ? *p != NUL : p < end; MB_PTR_ADV(p)) |
| 5477 | 2742 if (spell_iswordp_nmw(p, curwin)) |
| 236 | 2743 { |
| 455 | 2744 c = PTR2CHAR(p); |
| 324 | 2745 if (!SPELL_ISUPPER(c)) |
| 236 | 2746 { |
| 2747 /* UUl -> KEEPCAP */ | |
| 2748 if (past_second && allcap) | |
| 300 | 2749 return WF_KEEPCAP; |
| 236 | 2750 allcap = FALSE; |
| 2751 } | |
| 2752 else if (!allcap) | |
| 2753 /* UlU -> KEEPCAP */ | |
| 300 | 2754 return WF_KEEPCAP; |
| 236 | 2755 past_second = TRUE; |
| 2756 } | |
| 2757 | |
| 2758 if (allcap) | |
| 300 | 2759 return WF_ALLCAP; |
| 236 | 2760 if (firstcap) |
| 300 | 2761 return WF_ONECAP; |
| 236 | 2762 return 0; |
| 2763 } | |
| 2764 | |
| 474 | 2765 /* |
| 2766 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a | |
| 2767 * capital. So that make_case_word() can turn WOrd into Word. | |
| 2768 * Add ALLCAP for "WOrD". | |
| 2769 */ | |
| 2770 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2771 badword_captype(char_u *word, char_u *end) |
| 474 | 2772 { |
| 2773 int flags = captype(word, end); | |
| 475 | 2774 int c; |
| 474 | 2775 int l, u; |
| 2776 int first; | |
| 2777 char_u *p; | |
| 2778 | |
| 2779 if (flags & WF_KEEPCAP) | |
| 2780 { | |
| 2781 /* Count the number of UPPER and lower case letters. */ | |
| 2782 l = u = 0; | |
| 2783 first = FALSE; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
2784 for (p = word; p < end; MB_PTR_ADV(p)) |
| 474 | 2785 { |
| 475 | 2786 c = PTR2CHAR(p); |
| 2787 if (SPELL_ISUPPER(c)) | |
| 474 | 2788 { |
| 2789 ++u; | |
| 2790 if (p == word) | |
| 2791 first = TRUE; | |
| 2792 } | |
| 2793 else | |
| 2794 ++l; | |
| 2795 } | |
| 2796 | |
| 2797 /* If there are more UPPER than lower case letters suggest an | |
| 2798 * ALLCAP word. Otherwise, if the first letter is UPPER then | |
| 2799 * suggest ONECAP. Exception: "ALl" most likely should be "All", | |
| 2800 * require three upper case letters. */ | |
| 2801 if (u > l && u > 2) | |
| 2802 flags |= WF_ALLCAP; | |
| 2803 else if (first) | |
| 2804 flags |= WF_ONECAP; | |
| 633 | 2805 |
| 2806 if (u >= 2 && l >= 2) /* maCARONI maCAroni */ | |
| 2807 flags |= WF_MIXCAP; | |
| 474 | 2808 } |
| 2809 return flags; | |
| 2810 } | |
| 2811 | |
| 5519 | 2812 /* |
| 2813 * Delete the internal wordlist and its .spl file. | |
| 2814 */ | |
| 2815 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2816 spell_delete_wordlist(void) |
| 5519 | 2817 { |
| 2818 char_u fname[MAXPATHL]; | |
| 2819 | |
| 2820 if (int_wordlist != NULL) | |
| 2821 { | |
| 2822 mch_remove(int_wordlist); | |
| 2823 int_wordlist_spl(fname); | |
| 2824 mch_remove(fname); | |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2825 VIM_CLEAR(int_wordlist); |
| 5519 | 2826 } |
| 2827 } | |
| 2828 | |
| 2829 #if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO) | |
| 355 | 2830 /* |
| 2831 * Free all languages. | |
| 2832 */ | |
| 2833 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2834 spell_free_all(void) |
| 355 | 2835 { |
| 503 | 2836 slang_T *slang; |
| 355 | 2837 buf_T *buf; |
| 2838 | |
| 2580 | 2839 /* 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
|
2840 FOR_ALL_BUFFERS(buf) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2841 ga_clear(&buf->b_s.b_langp); |
| 355 | 2842 |
| 2843 while (first_lang != NULL) | |
| 2844 { | |
| 503 | 2845 slang = first_lang; |
| 2846 first_lang = slang->sl_next; | |
| 2847 slang_free(slang); | |
| 355 | 2848 } |
| 366 | 2849 |
| 5519 | 2850 spell_delete_wordlist(); |
| 381 | 2851 |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2852 VIM_CLEAR(repl_to); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
2853 VIM_CLEAR(repl_from); |
| 355 | 2854 } |
| 5519 | 2855 #endif |
| 2856 | |
| 2857 #if defined(FEAT_MBYTE) || defined(PROTO) | |
| 236 | 2858 /* |
| 2859 * Clear all spelling tables and reload them. | |
| 307 | 2860 * Used after 'encoding' is set and when ":mkspell" was used. |
| 236 | 2861 */ |
| 2862 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2863 spell_reload(void) |
| 236 | 2864 { |
| 316 | 2865 win_T *wp; |
| 236 | 2866 |
| 358 | 2867 /* Initialize the table for spell_iswordp(). */ |
| 236 | 2868 init_spell_chartab(); |
| 2869 | |
| 2870 /* Unload all allocated memory. */ | |
| 355 | 2871 spell_free_all(); |
| 236 | 2872 |
| 2873 /* 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
|
2874 FOR_ALL_WINDOWS(wp) |
| 236 | 2875 { |
| 316 | 2876 /* Only load the wordlists when 'spelllang' is set and there is a |
| 2877 * 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
|
2878 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
|
2879 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2880 if (wp->w_p_spell) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2881 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
2882 (void)did_set_spelllang(wp); |
| 316 | 2883 break; |
| 2884 } | |
| 2885 } | |
| 236 | 2886 } |
| 2887 } | |
| 5519 | 2888 #endif |
| 236 | 2889 |
| 310 | 2890 /* |
| 625 | 2891 * Opposite of offset2bytes(). |
| 2892 * "pp" points to the bytes and is advanced over it. | |
| 2893 * Returns the offset. | |
| 2894 */ | |
| 2895 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2896 bytes2offset(char_u **pp) |
| 625 | 2897 { |
| 2898 char_u *p = *pp; | |
| 2899 int nr; | |
| 2900 int c; | |
| 2901 | |
| 2902 c = *p++; | |
| 2903 if ((c & 0x80) == 0x00) /* 1 byte */ | |
| 2904 { | |
| 2905 nr = c - 1; | |
| 2906 } | |
| 2907 else if ((c & 0xc0) == 0x80) /* 2 bytes */ | |
| 2908 { | |
| 2909 nr = (c & 0x3f) - 1; | |
| 2910 nr = nr * 255 + (*p++ - 1); | |
| 2911 } | |
| 2912 else if ((c & 0xe0) == 0xc0) /* 3 bytes */ | |
| 2913 { | |
| 2914 nr = (c & 0x1f) - 1; | |
| 2915 nr = nr * 255 + (*p++ - 1); | |
| 2916 nr = nr * 255 + (*p++ - 1); | |
| 2917 } | |
| 2918 else /* 4 bytes */ | |
| 2919 { | |
| 2920 nr = (c & 0x0f) - 1; | |
| 2921 nr = nr * 255 + (*p++ - 1); | |
| 2922 nr = nr * 255 + (*p++ - 1); | |
| 2923 nr = nr * 255 + (*p++ - 1); | |
| 2924 } | |
| 2925 | |
| 2926 *pp = p; | |
| 2927 return nr; | |
| 2928 } | |
| 2929 | |
| 2930 | |
| 2931 /* | |
| 2932 * Open a spell buffer. This is a nameless buffer that is not in the buffer | |
| 2933 * list and only contains text lines. Can use a swapfile to reduce memory | |
| 2934 * use. | |
| 2935 * Most other fields are invalid! Esp. watch out for string options being | |
| 2936 * NULL and there is no undo info. | |
| 2937 * Returns NULL when out of memory. | |
| 2938 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2939 buf_T * |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2940 open_spellbuf(void) |
| 625 | 2941 { |
| 2942 buf_T *buf; | |
| 2943 | |
| 2944 buf = (buf_T *)alloc_clear(sizeof(buf_T)); | |
| 2945 if (buf != NULL) | |
| 2946 { | |
| 2947 buf->b_spell = TRUE; | |
| 2948 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
|
2949 #ifdef FEAT_CRYPT |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
2950 buf->b_p_key = empty_option; |
|
7aca84c0cd37
updated for version 7.4a.028
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
2951 #endif |
| 625 | 2952 ml_open(buf); |
| 2953 ml_open_file(buf); /* create swap file now */ | |
| 2954 } | |
| 2955 return buf; | |
| 2956 } | |
| 2957 | |
| 2958 /* | |
| 2959 * Close the buffer used for spell info. | |
| 2960 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2961 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2962 close_spellbuf(buf_T *buf) |
| 625 | 2963 { |
| 2964 if (buf != NULL) | |
| 2965 { | |
| 2966 ml_close(buf, TRUE); | |
| 2967 vim_free(buf); | |
| 2968 } | |
| 2969 } | |
| 2970 | |
| 307 | 2971 /* |
| 2972 * Init the chartab used for spelling for ASCII. | |
| 2973 * EBCDIC is not supported! | |
| 2974 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
2975 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
2976 clear_spell_chartab(spelltab_T *sp) |
| 307 | 2977 { |
| 324 | 2978 int i; |
| 307 | 2979 |
| 2980 /* Init everything to FALSE. */ | |
| 2981 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw)); | |
| 2982 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu)); | |
| 2983 for (i = 0; i < 256; ++i) | |
| 324 | 2984 { |
| 307 | 2985 sp->st_fold[i] = i; |
| 324 | 2986 sp->st_upper[i] = i; |
| 2987 } | |
| 307 | 2988 |
| 2989 /* We include digits. A word shouldn't start with a digit, but handling | |
| 2990 * that is done separately. */ | |
| 2991 for (i = '0'; i <= '9'; ++i) | |
| 2992 sp->st_isw[i] = TRUE; | |
| 2993 for (i = 'A'; i <= 'Z'; ++i) | |
| 2994 { | |
| 2995 sp->st_isw[i] = TRUE; | |
| 2996 sp->st_isu[i] = TRUE; | |
| 2997 sp->st_fold[i] = i + 0x20; | |
| 2998 } | |
| 2999 for (i = 'a'; i <= 'z'; ++i) | |
| 324 | 3000 { |
| 307 | 3001 sp->st_isw[i] = TRUE; |
| 324 | 3002 sp->st_upper[i] = i - 0x20; |
| 3003 } | |
| 307 | 3004 } |
| 3005 | |
| 3006 /* | |
| 3007 * Init the chartab used for spelling. Only depends on 'encoding'. | |
| 3008 * Called once while starting up and when 'encoding' changes. | |
| 3009 * The default is to use isalpha(), but the spell file should define the word | |
| 3010 * characters to make it possible that 'encoding' differs from the current | |
| 390 | 3011 * locale. For utf-8 we don't use isalpha() but our own functions. |
| 307 | 3012 */ |
| 3013 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3014 init_spell_chartab(void) |
| 307 | 3015 { |
| 3016 int i; | |
| 3017 | |
| 3018 did_set_spelltab = FALSE; | |
| 3019 clear_spell_chartab(&spelltab); | |
| 3020 #ifdef FEAT_MBYTE | |
| 3021 if (enc_dbcs) | |
| 3022 { | |
| 3023 /* DBCS: assume double-wide characters are word characters. */ | |
| 3024 for (i = 128; i <= 255; ++i) | |
| 3025 if (MB_BYTE2LEN(i) == 2) | |
| 3026 spelltab.st_isw[i] = TRUE; | |
| 3027 } | |
| 324 | 3028 else if (enc_utf8) |
| 3029 { | |
| 3030 for (i = 128; i < 256; ++i) | |
| 3031 { | |
|
2140
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3032 int f = utf_fold(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3033 int u = utf_toupper(i); |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3034 |
| 324 | 3035 spelltab.st_isu[i] = utf_isupper(i); |
| 3036 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
|
3037 /* 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
|
3038 * 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
|
3039 * value for utf-8 to avoid this. */ |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3040 spelltab.st_fold[i] = (f < 256) ? f : i; |
|
12aba62fa7c6
updated for version 7.2.422
Bram Moolenaar <bram@zimbu.org>
parents:
2046
diff
changeset
|
3041 spelltab.st_upper[i] = (u < 256) ? u : i; |
| 324 | 3042 } |
| 3043 } | |
| 307 | 3044 else |
| 3045 #endif | |
| 3046 { | |
| 324 | 3047 /* Rough guess: use locale-dependent library functions. */ |
| 307 | 3048 for (i = 128; i < 256; ++i) |
| 3049 { | |
| 3050 if (MB_ISUPPER(i)) | |
| 3051 { | |
| 324 | 3052 spelltab.st_isw[i] = TRUE; |
| 307 | 3053 spelltab.st_isu[i] = TRUE; |
| 3054 spelltab.st_fold[i] = MB_TOLOWER(i); | |
| 3055 } | |
| 324 | 3056 else if (MB_ISLOWER(i)) |
| 3057 { | |
| 3058 spelltab.st_isw[i] = TRUE; | |
| 3059 spelltab.st_upper[i] = MB_TOUPPER(i); | |
| 3060 } | |
| 307 | 3061 } |
| 3062 } | |
| 3063 } | |
| 3064 | |
| 3065 | |
| 3066 /* | |
| 358 | 3067 * Return TRUE if "p" points to a word character. |
| 366 | 3068 * As a special case we see "midword" characters as word character when it is |
| 358 | 3069 * followed by a word character. This finds they'there but not 'they there'. |
| 366 | 3070 * Thus this only works properly when past the first character of the word. |
| 358 | 3071 */ |
| 3072 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3073 spell_iswordp( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3074 char_u *p, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3075 win_T *wp) /* buffer used */ |
| 358 | 3076 { |
| 366 | 3077 #ifdef FEAT_MBYTE |
| 358 | 3078 char_u *s; |
| 366 | 3079 int l; |
| 3080 int c; | |
| 3081 | |
| 3082 if (has_mbyte) | |
| 3083 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
3084 l = MB_PTR2LEN(p); |
| 358 | 3085 s = p; |
| 366 | 3086 if (l == 1) |
| 3087 { | |
| 3088 /* be quick for ASCII */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3089 if (wp->w_s->b_spell_ismw[*p]) |
| 366 | 3090 s = p + 1; /* skip a mid-word character */ |
| 3091 } | |
| 3092 else | |
| 3093 { | |
| 3094 c = mb_ptr2char(p); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3095 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
|
3096 : (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
|
3097 && vim_strchr(wp->w_s->b_spell_ismw_mb, c) != NULL)) |
| 366 | 3098 s = p + l; |
| 3099 } | |
| 3100 | |
| 390 | 3101 c = mb_ptr2char(s); |
| 3102 if (c > 255) | |
| 5477 | 3103 return spell_mb_isword_class(mb_get_class(s), wp); |
| 390 | 3104 return spelltab.st_isw[c]; |
| 366 | 3105 } |
| 358 | 3106 #endif |
| 366 | 3107 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3108 return spelltab.st_isw[wp->w_s->b_spell_ismw[*p] ? p[1] : p[0]]; |
| 376 | 3109 } |
| 3110 | |
| 3111 /* | |
| 3112 * Return TRUE if "p" points to a word character. | |
| 3113 * Unlike spell_iswordp() this doesn't check for "midword" characters. | |
| 3114 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3115 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3116 spell_iswordp_nmw(char_u *p, win_T *wp) |
| 376 | 3117 { |
| 3118 #ifdef FEAT_MBYTE | |
| 390 | 3119 int c; |
| 3120 | |
| 3121 if (has_mbyte) | |
| 3122 { | |
| 3123 c = mb_ptr2char(p); | |
| 3124 if (c > 255) | |
| 5477 | 3125 return spell_mb_isword_class(mb_get_class(p), wp); |
| 390 | 3126 return spelltab.st_isw[c]; |
| 3127 } | |
| 376 | 3128 #endif |
| 3129 return spelltab.st_isw[*p]; | |
| 358 | 3130 } |
| 3131 | |
| 372 | 3132 #ifdef FEAT_MBYTE |
| 3133 /* | |
| 1580 | 3134 * Return TRUE if word class indicates a word character. |
| 3135 * Only for characters above 255. | |
| 3136 * Unicode subscript and superscript are not considered word characters. | |
| 5477 | 3137 * See also dbcs_class() and utf_class() in mbyte.c. |
| 3138 */ | |
| 3139 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3140 spell_mb_isword_class(int cl, win_T *wp) |
| 5477 | 3141 { |
| 3142 if (wp->w_s->b_cjk) | |
| 3143 /* East Asian characters are not considered word characters. */ | |
| 3144 return cl == 2 || cl == 0x2800; | |
| 1580 | 3145 return cl >= 2 && cl != 0x2070 && cl != 0x2080; |
| 3146 } | |
| 3147 | |
| 3148 /* | |
| 372 | 3149 * Return TRUE if "p" points to a word character. |
| 3150 * Wide version of spell_iswordp(). | |
| 3151 */ | |
| 3152 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3153 spell_iswordp_w(int *p, win_T *wp) |
| 372 | 3154 { |
| 3155 int *s; | |
| 3156 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3157 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
|
3158 : (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
|
3159 && vim_strchr(wp->w_s->b_spell_ismw_mb, *p) != NULL)) |
| 372 | 3160 s = p + 1; |
| 3161 else | |
| 3162 s = p; | |
| 3163 | |
| 390 | 3164 if (*s > 255) |
| 372 | 3165 { |
| 3166 if (enc_utf8) | |
| 5477 | 3167 return spell_mb_isword_class(utf_class(*s), wp); |
| 372 | 3168 if (enc_dbcs) |
| 5477 | 3169 return spell_mb_isword_class( |
| 3170 dbcs_class((unsigned)*s >> 8, *s & 0xff), wp); | |
| 372 | 3171 return 0; |
| 3172 } | |
| 3173 return spelltab.st_isw[*s]; | |
| 3174 } | |
| 3175 #endif | |
| 3176 | |
| 358 | 3177 /* |
| 324 | 3178 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated. |
| 3179 * Uses the character definitions from the .spl file. | |
| 307 | 3180 * When using a multi-byte 'encoding' the length may change! |
| 3181 * Returns FAIL when something wrong. | |
| 3182 */ | |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
3183 int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3184 spell_casefold( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3185 char_u *str, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3186 int len, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3187 char_u *buf, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3188 int buflen) |
| 307 | 3189 { |
| 3190 int i; | |
| 3191 | |
| 3192 if (len >= buflen) | |
| 3193 { | |
| 3194 buf[0] = NUL; | |
| 3195 return FAIL; /* result will not fit */ | |
| 3196 } | |
| 3197 | |
| 3198 #ifdef FEAT_MBYTE | |
| 3199 if (has_mbyte) | |
| 3200 { | |
| 324 | 3201 int outi = 0; |
| 3202 char_u *p; | |
| 307 | 3203 int c; |
| 3204 | |
| 3205 /* Fold one character at a time. */ | |
| 324 | 3206 for (p = str; p < str + len; ) |
| 307 | 3207 { |
| 3208 if (outi + MB_MAXBYTES > buflen) | |
| 3209 { | |
| 3210 buf[outi] = NUL; | |
| 3211 return FAIL; | |
| 3212 } | |
| 474 | 3213 c = mb_cptr2char_adv(&p); |
| 324 | 3214 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi); |
| 307 | 3215 } |
| 3216 buf[outi] = NUL; | |
| 3217 } | |
| 3218 else | |
| 3219 #endif | |
| 3220 { | |
| 3221 /* Be quick for non-multibyte encodings. */ | |
| 3222 for (i = 0; i < len; ++i) | |
| 324 | 3223 buf[i] = spelltab.st_fold[str[i]]; |
| 307 | 3224 buf[i] = NUL; |
| 3225 } | |
| 3226 | |
| 3227 return OK; | |
| 3228 } | |
| 3229 | |
| 625 | 3230 /* values for sps_flags */ |
| 372 | 3231 #define SPS_BEST 1 |
| 3232 #define SPS_FAST 2 | |
| 3233 #define SPS_DOUBLE 4 | |
| 3234 | |
| 625 | 3235 static int sps_flags = SPS_BEST; /* flags from 'spellsuggest' */ |
| 3236 static int sps_limit = 9999; /* max nr of suggestions given */ | |
| 372 | 3237 |
| 3238 /* | |
| 3239 * Check the 'spellsuggest' option. Return FAIL if it's wrong. | |
| 483 | 3240 * Sets "sps_flags" and "sps_limit". |
| 372 | 3241 */ |
| 3242 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3243 spell_check_sps(void) |
| 372 | 3244 { |
| 3245 char_u *p; | |
| 483 | 3246 char_u *s; |
| 372 | 3247 char_u buf[MAXPATHL]; |
| 3248 int f; | |
| 3249 | |
| 3250 sps_flags = 0; | |
| 483 | 3251 sps_limit = 9999; |
| 372 | 3252 |
| 3253 for (p = p_sps; *p != NUL; ) | |
| 3254 { | |
| 3255 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 3256 | |
| 3257 f = 0; | |
| 483 | 3258 if (VIM_ISDIGIT(*buf)) |
| 3259 { | |
| 3260 s = buf; | |
| 3261 sps_limit = getdigits(&s); | |
| 3262 if (*s != NUL && !VIM_ISDIGIT(*s)) | |
| 3263 f = -1; | |
| 3264 } | |
| 3265 else if (STRCMP(buf, "best") == 0) | |
| 372 | 3266 f = SPS_BEST; |
| 3267 else if (STRCMP(buf, "fast") == 0) | |
| 3268 f = SPS_FAST; | |
| 3269 else if (STRCMP(buf, "double") == 0) | |
| 3270 f = SPS_DOUBLE; | |
| 3271 else if (STRNCMP(buf, "expr:", 5) != 0 | |
| 3272 && STRNCMP(buf, "file:", 5) != 0) | |
| 3273 f = -1; | |
| 3274 | |
| 3275 if (f == -1 || (sps_flags != 0 && f != 0)) | |
| 3276 { | |
| 3277 sps_flags = SPS_BEST; | |
| 483 | 3278 sps_limit = 9999; |
| 372 | 3279 return FAIL; |
| 3280 } | |
| 3281 if (f != 0) | |
| 3282 sps_flags = f; | |
| 3283 } | |
| 3284 | |
| 3285 if (sps_flags == 0) | |
| 3286 sps_flags = SPS_BEST; | |
| 3287 | |
| 3288 return OK; | |
| 3289 } | |
| 3290 | |
| 323 | 3291 /* |
| 5374 | 3292 * "z=": Find badly spelled word under or after the cursor. |
| 323 | 3293 * Give suggestions for the properly spelled word. |
| 638 | 3294 * In Visual mode use the highlighted word as the bad word. |
| 485 | 3295 * When "count" is non-zero use that suggestion. |
| 323 | 3296 */ |
| 3297 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3298 spell_suggest(int count) |
| 323 | 3299 { |
| 3300 char_u *line; | |
| 3301 pos_T prev_cursor = curwin->w_cursor; | |
| 3302 char_u wcopy[MAXWLEN + 2]; | |
| 3303 char_u *p; | |
| 3304 int i; | |
| 3305 int c; | |
| 3306 suginfo_T sug; | |
| 3307 suggest_T *stp; | |
| 372 | 3308 int mouse_used; |
| 388 | 3309 int need_cap; |
| 483 | 3310 int limit; |
| 485 | 3311 int selected = count; |
| 638 | 3312 int badlen = 0; |
| 1938 | 3313 int msg_scroll_save = msg_scroll; |
| 638 | 3314 |
| 3315 if (no_spell_checking(curwin)) | |
| 3316 return; | |
| 3317 | |
| 3318 if (VIsual_active) | |
| 3319 { | |
| 3320 /* Use the Visually selected text as the bad word. But reject | |
| 3321 * a multi-line selection. */ | |
| 3322 if (curwin->w_cursor.lnum != VIsual.lnum) | |
| 3323 { | |
| 6949 | 3324 vim_beep(BO_SPELL); |
| 638 | 3325 return; |
| 3326 } | |
| 3327 badlen = (int)curwin->w_cursor.col - (int)VIsual.col; | |
| 3328 if (badlen < 0) | |
| 3329 badlen = -badlen; | |
| 3330 else | |
| 3331 curwin->w_cursor.col = VIsual.col; | |
| 3332 ++badlen; | |
| 3333 end_visual_mode(); | |
| 3334 } | |
| 5735 | 3335 /* Find the start of the badly spelled word. */ |
| 3336 else if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0 | |
| 346 | 3337 || curwin->w_cursor.col > prev_cursor.col) |
| 3338 { | |
| 3339 /* No bad word or it starts after the cursor: use the word under the | |
| 3340 * cursor. */ | |
| 3341 curwin->w_cursor = prev_cursor; | |
| 3342 line = ml_get_curline(); | |
| 3343 p = line + curwin->w_cursor.col; | |
| 3344 /* Backup to before start of word. */ | |
| 5477 | 3345 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
|
3346 MB_PTR_BACK(line, p); |
| 346 | 3347 /* Forward to start of word. */ |
| 5477 | 3348 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
|
3349 MB_PTR_ADV(p); |
| 346 | 3350 |
| 5477 | 3351 if (!spell_iswordp_nmw(p, curwin)) /* No word found. */ |
| 346 | 3352 { |
| 3353 beep_flush(); | |
| 3354 return; | |
| 3355 } | |
| 835 | 3356 curwin->w_cursor.col = (colnr_T)(p - line); |
| 323 | 3357 } |
| 3358 | |
| 344 | 3359 /* Get the word and its length. */ |
| 3360 | |
| 388 | 3361 /* Figure out if the word should be capitalised. */ |
| 475 | 3362 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col); |
| 3363 | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3364 /* 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
|
3365 line = vim_strsave(ml_get_curline()); |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3366 if (line == NULL) |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3367 goto skip; |
| 388 | 3368 |
| 483 | 3369 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in |
| 3370 * 'spellsuggest', whatever is smaller. */ | |
| 3371 if (sps_limit > (int)Rows - 2) | |
| 3372 limit = (int)Rows - 2; | |
| 3373 else | |
| 3374 limit = sps_limit; | |
| 638 | 3375 spell_find_suggest(line + curwin->w_cursor.col, badlen, &sug, limit, |
| 625 | 3376 TRUE, need_cap, TRUE); |
| 323 | 3377 |
| 3378 if (sug.su_ga.ga_len == 0) | |
| 3379 MSG(_("Sorry, no suggestions")); | |
| 485 | 3380 else if (count > 0) |
| 3381 { | |
| 3382 if (count > sug.su_ga.ga_len) | |
| 3383 smsg((char_u *)_("Sorry, only %ld suggestions"), | |
| 3384 (long)sug.su_ga.ga_len); | |
| 3385 } | |
| 323 | 3386 else |
| 3387 { | |
|
13244
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
3388 VIM_CLEAR(repl_from); |
|
ac42c4b11dbc
patch 8.0.1496: clearing a pointer takes two lines
Christian Brabandt <cb@256bit.org>
parents:
12674
diff
changeset
|
3389 VIM_CLEAR(repl_to); |
| 372 | 3390 |
| 474 | 3391 #ifdef FEAT_RIGHTLEFT |
| 3392 /* When 'rightleft' is set the list is drawn right-left. */ | |
| 3393 cmdmsg_rl = curwin->w_p_rl; | |
| 3394 if (cmdmsg_rl) | |
| 3395 msg_col = Columns - 1; | |
| 3396 #endif | |
| 3397 | |
| 323 | 3398 /* List the suggestions. */ |
| 3399 msg_start(); | |
| 914 | 3400 msg_row = Rows - 1; /* for when 'cmdheight' > 1 */ |
| 372 | 3401 lines_left = Rows; /* avoid more prompt */ |
| 323 | 3402 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"), |
| 3403 sug.su_badlen, sug.su_badptr); | |
| 474 | 3404 #ifdef FEAT_RIGHTLEFT |
| 3405 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0) | |
| 3406 { | |
| 3407 /* And now the rabbit from the high hat: Avoid showing the | |
| 3408 * untranslated message rightleft. */ | |
| 3409 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC", | |
| 3410 sug.su_badlen, sug.su_badptr); | |
| 3411 } | |
| 3412 #endif | |
| 323 | 3413 msg_puts(IObuff); |
| 3414 msg_clr_eos(); | |
| 3415 msg_putchar('\n'); | |
| 346 | 3416 |
| 323 | 3417 msg_scroll = TRUE; |
| 3418 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 3419 { | |
| 344 | 3420 stp = &SUG(sug.su_ga, i); |
| 323 | 3421 |
| 3422 /* The suggested word may replace only part of the bad word, add | |
| 3423 * the not replaced part. */ | |
| 2768 | 3424 vim_strncpy(wcopy, stp->st_word, MAXWLEN); |
| 323 | 3425 if (sug.su_badlen > stp->st_orglen) |
| 625 | 3426 vim_strncpy(wcopy + stp->st_wordlen, |
| 323 | 3427 sug.su_badptr + stp->st_orglen, |
| 3428 sug.su_badlen - stp->st_orglen); | |
| 474 | 3429 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1); |
| 3430 #ifdef FEAT_RIGHTLEFT | |
| 3431 if (cmdmsg_rl) | |
| 3432 rl_mirror(IObuff); | |
| 3433 #endif | |
| 3434 msg_puts(IObuff); | |
| 3435 | |
| 3436 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy); | |
| 346 | 3437 msg_puts(IObuff); |
| 3438 | |
| 3439 /* The word may replace more than "su_badlen". */ | |
| 3440 if (sug.su_badlen < stp->st_orglen) | |
| 3441 { | |
| 3442 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""), | |
| 3443 stp->st_orglen, sug.su_badptr); | |
| 3444 msg_puts(IObuff); | |
| 3445 } | |
| 3446 | |
| 324 | 3447 if (p_verbose > 0) |
| 344 | 3448 { |
| 346 | 3449 /* Add the score. */ |
| 351 | 3450 if (sps_flags & (SPS_DOUBLE | SPS_BEST)) |
| 474 | 3451 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)", |
| 344 | 3452 stp->st_salscore ? "s " : "", |
| 3453 stp->st_score, stp->st_altscore); | |
| 3454 else | |
| 474 | 3455 vim_snprintf((char *)IObuff, IOSIZE, " (%d)", |
| 346 | 3456 stp->st_score); |
| 474 | 3457 #ifdef FEAT_RIGHTLEFT |
| 3458 if (cmdmsg_rl) | |
| 3459 /* Mirror the numbers, but keep the leading space. */ | |
| 3460 rl_mirror(IObuff + 1); | |
| 3461 #endif | |
| 346 | 3462 msg_advance(30); |
| 3463 msg_puts(IObuff); | |
| 344 | 3464 } |
| 323 | 3465 msg_putchar('\n'); |
| 3466 } | |
| 3467 | |
| 474 | 3468 #ifdef FEAT_RIGHTLEFT |
| 3469 cmdmsg_rl = FALSE; | |
| 3470 msg_col = 0; | |
| 3471 #endif | |
| 323 | 3472 /* Ask for choice. */ |
| 485 | 3473 selected = prompt_for_number(&mouse_used); |
| 372 | 3474 if (mouse_used) |
| 485 | 3475 selected -= lines_left; |
| 1938 | 3476 lines_left = Rows; /* avoid more prompt */ |
| 3477 /* don't delay for 'smd' in normal_cmd() */ | |
| 3478 msg_scroll = msg_scroll_save; | |
| 485 | 3479 } |
| 3480 | |
| 3481 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK) | |
| 3482 { | |
| 3483 /* Save the from and to text for :spellrepall. */ | |
| 3484 stp = &SUG(sug.su_ga, selected - 1); | |
| 548 | 3485 if (sug.su_badlen > stp->st_orglen) |
| 3486 { | |
| 3487 /* Replacing less than "su_badlen", append the remainder to | |
| 3488 * repl_to. */ | |
| 3489 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen); | |
| 3490 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word, | |
| 3491 sug.su_badlen - stp->st_orglen, | |
| 3492 sug.su_badptr + stp->st_orglen); | |
| 3493 repl_to = vim_strsave(IObuff); | |
| 3494 } | |
| 3495 else | |
| 3496 { | |
| 3497 /* Replacing su_badlen or more, use the whole word. */ | |
| 3498 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen); | |
| 3499 repl_to = vim_strsave(stp->st_word); | |
| 3500 } | |
| 485 | 3501 |
| 3502 /* Replace the word. */ | |
| 1938 | 3503 p = alloc((unsigned)STRLEN(line) - stp->st_orglen |
| 3504 + stp->st_wordlen + 1); | |
| 485 | 3505 if (p != NULL) |
| 3506 { | |
| 835 | 3507 c = (int)(sug.su_badptr - line); |
| 485 | 3508 mch_memmove(p, line, c); |
| 3509 STRCPY(p + c, stp->st_word); | |
| 3510 STRCAT(p, sug.su_badptr + stp->st_orglen); | |
| 3511 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 3512 curwin->w_cursor.col = c; | |
| 3513 | |
| 3514 /* For redo we use a change-word command. */ | |
| 3515 ResetRedobuff(); | |
| 3516 AppendToRedobuff((char_u *)"ciw"); | |
| 620 | 3517 AppendToRedobuffLit(p + c, |
| 625 | 3518 stp->st_wordlen + sug.su_badlen - stp->st_orglen); |
| 485 | 3519 AppendCharToRedobuff(ESC); |
| 819 | 3520 |
| 3521 /* After this "p" may be invalid. */ | |
| 3522 changed_bytes(curwin->w_cursor.lnum, c); | |
| 485 | 3523 } |
| 3524 } | |
| 3525 else | |
| 3526 curwin->w_cursor = prev_cursor; | |
| 323 | 3527 |
| 344 | 3528 spell_find_cleanup(&sug); |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3529 skip: |
|
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3530 vim_free(line); |
| 344 | 3531 } |
| 3532 | |
| 3533 /* | |
| 475 | 3534 * Check if the word at line "lnum" column "col" is required to start with a |
| 3535 * capital. This uses 'spellcapcheck' of the current buffer. | |
| 3536 */ | |
| 3537 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3538 check_need_cap(linenr_T lnum, colnr_T col) |
| 475 | 3539 { |
| 3540 int need_cap = FALSE; | |
| 3541 char_u *line; | |
| 3542 char_u *line_copy = NULL; | |
| 3543 char_u *p; | |
| 3544 colnr_T endcol; | |
| 3545 regmatch_T regmatch; | |
| 3546 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3547 if (curwin->w_s->b_cap_prog == NULL) |
| 475 | 3548 return FALSE; |
| 3549 | |
| 3550 line = ml_get_curline(); | |
| 3551 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
|
3552 if (getwhitecols(line) >= (int)col) |
| 475 | 3553 { |
| 3554 /* At start of line, check if previous line is empty or sentence | |
| 3555 * ends there. */ | |
| 3556 if (lnum == 1) | |
| 3557 need_cap = TRUE; | |
| 3558 else | |
| 3559 { | |
| 3560 line = ml_get(lnum - 1); | |
| 3561 if (*skipwhite(line) == NUL) | |
| 3562 need_cap = TRUE; | |
| 3563 else | |
| 3564 { | |
| 3565 /* Append a space in place of the line break. */ | |
| 3566 line_copy = concat_str(line, (char_u *)" "); | |
| 3567 line = line_copy; | |
| 835 | 3568 endcol = (colnr_T)STRLEN(line); |
| 475 | 3569 } |
| 3570 } | |
| 3571 } | |
| 3572 else | |
| 3573 endcol = col; | |
| 3574 | |
| 3575 if (endcol > 0) | |
| 3576 { | |
| 3577 /* 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
|
3578 regmatch.regprog = curwin->w_s->b_cap_prog; |
| 475 | 3579 regmatch.rm_ic = FALSE; |
| 3580 p = line + endcol; | |
| 3581 for (;;) | |
| 3582 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
3583 MB_PTR_BACK(line, p); |
| 5477 | 3584 if (p == line || spell_iswordp_nmw(p, curwin)) |
| 475 | 3585 break; |
| 3586 if (vim_regexec(®match, p, 0) | |
| 3587 && regmatch.endp[0] == line + endcol) | |
| 3588 { | |
| 3589 need_cap = TRUE; | |
| 3590 break; | |
| 3591 } | |
| 3592 } | |
| 6375 | 3593 curwin->w_s->b_cap_prog = regmatch.regprog; |
| 475 | 3594 } |
| 3595 | |
| 3596 vim_free(line_copy); | |
| 3597 | |
| 3598 return need_cap; | |
| 3599 } | |
| 3600 | |
| 3601 | |
| 3602 /* | |
| 372 | 3603 * ":spellrepall" |
| 3604 */ | |
| 3605 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3606 ex_spellrepall(exarg_T *eap UNUSED) |
| 372 | 3607 { |
| 3608 pos_T pos = curwin->w_cursor; | |
| 3609 char_u *frompat; | |
| 3610 int addlen; | |
| 3611 char_u *line; | |
| 3612 char_u *p; | |
| 3613 int save_ws = p_ws; | |
| 483 | 3614 linenr_T prev_lnum = 0; |
| 372 | 3615 |
| 3616 if (repl_from == NULL || repl_to == NULL) | |
| 3617 { | |
| 3618 EMSG(_("E752: No previous spell replacement")); | |
| 3619 return; | |
| 3620 } | |
| 835 | 3621 addlen = (int)(STRLEN(repl_to) - STRLEN(repl_from)); |
| 3622 | |
| 3623 frompat = alloc((unsigned)STRLEN(repl_from) + 7); | |
| 372 | 3624 if (frompat == NULL) |
| 3625 return; | |
| 3626 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from); | |
| 3627 p_ws = FALSE; | |
| 3628 | |
| 483 | 3629 sub_nsubs = 0; |
| 3630 sub_nlines = 0; | |
| 372 | 3631 curwin->w_cursor.lnum = 0; |
| 3632 while (!got_int) | |
| 3633 { | |
|
11521
578df034735d
patch 8.0.0643: when a pattern search is slow Vim becomes unusable
Christian Brabandt <cb@256bit.org>
parents:
11394
diff
changeset
|
3634 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP, NULL, NULL) == 0 |
| 372 | 3635 || u_save_cursor() == FAIL) |
| 3636 break; | |
| 3637 | |
| 3638 /* Only replace when the right word isn't there yet. This happens | |
| 3639 * when changing "etc" to "etc.". */ | |
| 3640 line = ml_get_curline(); | |
| 3641 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col, | |
| 3642 repl_to, STRLEN(repl_to)) != 0) | |
| 3643 { | |
| 835 | 3644 p = alloc((unsigned)STRLEN(line) + addlen + 1); |
| 372 | 3645 if (p == NULL) |
| 3646 break; | |
| 3647 mch_memmove(p, line, curwin->w_cursor.col); | |
| 3648 STRCPY(p + curwin->w_cursor.col, repl_to); | |
| 3649 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from)); | |
| 3650 ml_replace(curwin->w_cursor.lnum, p, FALSE); | |
| 3651 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col); | |
| 483 | 3652 |
| 3653 if (curwin->w_cursor.lnum != prev_lnum) | |
| 3654 { | |
| 3655 ++sub_nlines; | |
| 3656 prev_lnum = curwin->w_cursor.lnum; | |
| 3657 } | |
| 3658 ++sub_nsubs; | |
| 372 | 3659 } |
| 835 | 3660 curwin->w_cursor.col += (colnr_T)STRLEN(repl_to); |
| 372 | 3661 } |
| 3662 | |
| 3663 p_ws = save_ws; | |
| 3664 curwin->w_cursor = pos; | |
| 3665 vim_free(frompat); | |
| 3666 | |
| 483 | 3667 if (sub_nsubs == 0) |
| 372 | 3668 EMSG2(_("E753: Not found: %s"), repl_from); |
| 483 | 3669 else |
| 3670 do_sub_msg(FALSE); | |
| 372 | 3671 } |
| 3672 | |
| 3673 /* | |
| 344 | 3674 * Find spell suggestions for "word". Return them in the growarray "*gap" as |
| 3675 * a list of allocated strings. | |
| 3676 */ | |
| 3677 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3678 spell_suggest_list( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3679 garray_T *gap, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3680 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3681 int maxcount, /* maximum nr of suggestions */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3682 int need_cap, /* 'spellcapcheck' matched */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3683 int interactive) |
| 344 | 3684 { |
| 3685 suginfo_T sug; | |
| 3686 int i; | |
| 3687 suggest_T *stp; | |
| 3688 char_u *wcopy; | |
| 3689 | |
| 638 | 3690 spell_find_suggest(word, 0, &sug, maxcount, FALSE, need_cap, interactive); |
| 344 | 3691 |
| 3692 /* Make room in "gap". */ | |
| 3693 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1); | |
| 626 | 3694 if (ga_grow(gap, sug.su_ga.ga_len) == OK) |
| 3695 { | |
| 3696 for (i = 0; i < sug.su_ga.ga_len; ++i) | |
| 3697 { | |
| 3698 stp = &SUG(sug.su_ga, i); | |
| 3699 | |
| 3700 /* The suggested word may replace only part of "word", add the not | |
| 3701 * replaced part. */ | |
| 3702 wcopy = alloc(stp->st_wordlen | |
| 835 | 3703 + (unsigned)STRLEN(sug.su_badptr + stp->st_orglen) + 1); |
| 626 | 3704 if (wcopy == NULL) |
| 3705 break; | |
| 3706 STRCPY(wcopy, stp->st_word); | |
| 3707 STRCPY(wcopy + stp->st_wordlen, sug.su_badptr + stp->st_orglen); | |
| 3708 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy; | |
| 3709 } | |
| 344 | 3710 } |
| 3711 | |
| 3712 spell_find_cleanup(&sug); | |
| 3713 } | |
| 3714 | |
| 3715 /* | |
| 3716 * Find spell suggestions for the word at the start of "badptr". | |
| 3717 * Return the suggestions in "su->su_ga". | |
| 3718 * The maximum number of suggestions is "maxcount". | |
| 3719 * Note: does use info for the current window. | |
| 3720 * This is based on the mechanisms of Aspell, but completely reimplemented. | |
| 3721 */ | |
| 3722 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3723 spell_find_suggest( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3724 char_u *badptr, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3725 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
|
3726 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3727 int maxcount, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3728 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
|
3729 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
|
3730 int interactive) |
| 344 | 3731 { |
| 534 | 3732 hlf_T attr = HLF_COUNT; |
| 372 | 3733 char_u buf[MAXPATHL]; |
| 3734 char_u *p; | |
| 3735 int do_combine = FALSE; | |
| 3736 char_u *sps_copy; | |
| 3737 #ifdef FEAT_EVAL | |
| 3738 static int expr_busy = FALSE; | |
| 3739 #endif | |
| 385 | 3740 int c; |
| 503 | 3741 int i; |
| 3742 langp_T *lp; | |
| 344 | 3743 |
| 3744 /* | |
| 3745 * Set the info in "*su". | |
| 3746 */ | |
| 3747 vim_memset(su, 0, sizeof(suginfo_T)); | |
| 3748 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10); | |
| 3749 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10); | |
| 355 | 3750 if (*badptr == NUL) |
| 3751 return; | |
| 344 | 3752 hash_init(&su->su_banned); |
| 3753 | |
| 3754 su->su_badptr = badptr; | |
| 638 | 3755 if (badlen != 0) |
| 3756 su->su_badlen = badlen; | |
| 3757 else | |
| 3758 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL, FALSE); | |
| 344 | 3759 su->su_maxcount = maxcount; |
| 372 | 3760 su->su_maxscore = SCORE_MAXINIT; |
| 344 | 3761 |
| 3762 if (su->su_badlen >= MAXWLEN) | |
| 3763 su->su_badlen = MAXWLEN - 1; /* just in case */ | |
| 3764 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen); | |
| 3765 (void)spell_casefold(su->su_badptr, su->su_badlen, | |
| 3766 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
|
3767 /* 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
|
3768 * 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
|
3769 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
|
3770 |
| 346 | 3771 /* get caps flags for bad word */ |
| 474 | 3772 su->su_badflags = badword_captype(su->su_badptr, |
| 3773 su->su_badptr + su->su_badlen); | |
| 388 | 3774 if (need_cap) |
| 3775 su->su_badflags |= WF_ONECAP; | |
| 344 | 3776 |
| 503 | 3777 /* Find the default language for sound folding. We simply use the first |
| 3778 * one in 'spelllang' that supports sound folding. That's good for when | |
| 3779 * using multiple files for one language, it's not that bad when mixing | |
| 3780 * languages (e.g., "pl,en"). */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3781 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
|
3782 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
3783 lp = LANGP_ENTRY(curbuf->b_s.b_langp, i); |
| 503 | 3784 if (lp->lp_sallang != NULL) |
| 3785 { | |
| 3786 su->su_sallang = lp->lp_sallang; | |
| 3787 break; | |
| 3788 } | |
| 3789 } | |
| 3790 | |
| 534 | 3791 /* Soundfold the bad word with the default sound folding, so that we don't |
| 3792 * have to do this many times. */ | |
| 3793 if (su->su_sallang != NULL) | |
| 3794 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE, | |
| 3795 su->su_sal_badword); | |
| 3796 | |
| 385 | 3797 /* If the word is not capitalised and spell_check() doesn't consider the |
| 3798 * word to be bad then it might need to be capitalised. Add a suggestion | |
| 3799 * for that. */ | |
| 455 | 3800 c = PTR2CHAR(su->su_badptr); |
| 534 | 3801 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT) |
| 385 | 3802 { |
| 3803 make_case_word(su->su_badword, buf, WF_ONECAP); | |
| 3804 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE, | |
| 625 | 3805 0, TRUE, su->su_sallang, FALSE); |
| 385 | 3806 } |
| 3807 | |
| 344 | 3808 /* Ban the bad word itself. It may appear in another region. */ |
| 358 | 3809 if (banbadword) |
| 3810 add_banned(su, su->su_badword); | |
| 344 | 3811 |
| 372 | 3812 /* Make a copy of 'spellsuggest', because the expression may change it. */ |
| 3813 sps_copy = vim_strsave(p_sps); | |
| 3814 if (sps_copy == NULL) | |
| 3815 return; | |
| 3816 | |
| 3817 /* Loop over the items in 'spellsuggest'. */ | |
| 3818 for (p = sps_copy; *p != NUL; ) | |
| 3819 { | |
| 3820 copy_option_part(&p, buf, MAXPATHL, ","); | |
| 3821 | |
| 3822 if (STRNCMP(buf, "expr:", 5) == 0) | |
| 3823 { | |
| 3824 #ifdef FEAT_EVAL | |
| 375 | 3825 /* Evaluate an expression. Skip this when called recursively, |
| 3826 * when using spellsuggest() in the expression. */ | |
| 372 | 3827 if (!expr_busy) |
| 3828 { | |
| 3829 expr_busy = TRUE; | |
| 3830 spell_suggest_expr(su, buf + 5); | |
| 3831 expr_busy = FALSE; | |
| 3832 } | |
| 3833 #endif | |
| 3834 } | |
| 3835 else if (STRNCMP(buf, "file:", 5) == 0) | |
| 3836 /* Use list of suggestions in a file. */ | |
| 3837 spell_suggest_file(su, buf + 5); | |
| 3838 else | |
| 3839 { | |
| 3840 /* Use internal method. */ | |
| 625 | 3841 spell_suggest_intern(su, interactive); |
| 372 | 3842 if (sps_flags & SPS_DOUBLE) |
| 3843 do_combine = TRUE; | |
| 3844 } | |
| 3845 } | |
| 3846 | |
| 3847 vim_free(sps_copy); | |
| 3848 | |
| 3849 if (do_combine) | |
| 3850 /* Combine the two list of suggestions. This must be done last, | |
| 3851 * because sorting changes the order again. */ | |
| 3852 score_combine(su); | |
| 3853 } | |
| 3854 | |
| 3855 #ifdef FEAT_EVAL | |
| 3856 /* | |
| 3857 * Find suggestions by evaluating expression "expr". | |
| 3858 */ | |
| 3859 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3860 spell_suggest_expr(suginfo_T *su, char_u *expr) |
| 372 | 3861 { |
| 3862 list_T *list; | |
| 3863 listitem_T *li; | |
| 3864 int score; | |
| 3865 char_u *p; | |
| 3866 | |
| 3867 /* The work is split up in a few parts to avoid having to export | |
| 3868 * suginfo_T. | |
| 3869 * First evaluate the expression and get the resulting list. */ | |
| 3870 list = eval_spell_expr(su->su_badword, expr); | |
| 3871 if (list != NULL) | |
| 3872 { | |
| 3873 /* Loop over the items in the list. */ | |
| 3874 for (li = list->lv_first; li != NULL; li = li->li_next) | |
| 3875 if (li->li_tv.v_type == VAR_LIST) | |
| 3876 { | |
| 3877 /* Get the word and the score from the items. */ | |
| 3878 score = get_spellword(li->li_tv.vval.v_list, &p); | |
| 625 | 3879 if (score >= 0 && score <= su->su_maxscore) |
| 3880 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 3881 score, 0, TRUE, su->su_sallang, FALSE); | |
| 372 | 3882 } |
| 3883 list_unref(list); | |
| 3884 } | |
| 3885 | |
| 625 | 3886 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3887 check_suggestions(su, &su->su_ga); | |
| 372 | 3888 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 3889 } | |
| 3890 #endif | |
| 3891 | |
| 3892 /* | |
| 474 | 3893 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'. |
| 372 | 3894 */ |
| 3895 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3896 spell_suggest_file(suginfo_T *su, char_u *fname) |
| 372 | 3897 { |
| 3898 FILE *fd; | |
| 3899 char_u line[MAXWLEN * 2]; | |
| 3900 char_u *p; | |
| 3901 int len; | |
| 3902 char_u cword[MAXWLEN]; | |
| 3903 | |
| 3904 /* Open the file. */ | |
| 3905 fd = mch_fopen((char *)fname, "r"); | |
| 3906 if (fd == NULL) | |
| 3907 { | |
| 3908 EMSG2(_(e_notopen), fname); | |
| 3909 return; | |
| 3910 } | |
| 3911 | |
| 3912 /* Read it line by line. */ | |
| 3913 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int) | |
| 3914 { | |
| 3915 line_breakcheck(); | |
| 3916 | |
| 3917 p = vim_strchr(line, '/'); | |
| 3918 if (p == NULL) | |
| 3919 continue; /* No Tab found, just skip the line. */ | |
| 3920 *p++ = NUL; | |
| 3921 if (STRICMP(su->su_badword, line) == 0) | |
| 3922 { | |
| 3923 /* Match! Isolate the good word, until CR or NL. */ | |
| 3924 for (len = 0; p[len] >= ' '; ++len) | |
| 3925 ; | |
| 3926 p[len] = NUL; | |
| 3927 | |
| 3928 /* If the suggestion doesn't have specific case duplicate the case | |
| 3929 * of the bad word. */ | |
| 3930 if (captype(p, NULL) == 0) | |
| 3931 { | |
| 3932 make_case_word(p, cword, su->su_badflags); | |
| 3933 p = cword; | |
| 3934 } | |
| 3935 | |
| 3936 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 625 | 3937 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE); |
| 372 | 3938 } |
| 3939 } | |
| 3940 | |
| 3941 fclose(fd); | |
| 3942 | |
| 625 | 3943 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 3944 check_suggestions(su, &su->su_ga); | |
| 372 | 3945 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 3946 } | |
| 3947 | |
| 3948 /* | |
| 3949 * Find suggestions for the internal method indicated by "sps_flags". | |
| 3950 */ | |
| 3951 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
3952 spell_suggest_intern(suginfo_T *su, int interactive) |
| 625 | 3953 { |
| 3954 /* | |
| 3955 * Load the .sug file(s) that are available and not done yet. | |
| 3956 */ | |
| 3957 suggest_load_files(); | |
| 3958 | |
| 344 | 3959 /* |
| 346 | 3960 * 1. Try special cases, such as repeating a word: "the the" -> "the". |
| 344 | 3961 * |
| 3962 * Set a maximum score to limit the combination of operations that is | |
| 3963 * tried. | |
| 3964 */ | |
| 346 | 3965 suggest_try_special(su); |
| 3966 | |
| 3967 /* | |
| 3968 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries | |
| 3969 * from the .aff file and inserting a space (split the word). | |
| 3970 */ | |
| 3971 suggest_try_change(su); | |
| 344 | 3972 |
| 3973 /* For the resulting top-scorers compute the sound-a-like score. */ | |
| 3974 if (sps_flags & SPS_DOUBLE) | |
| 3975 score_comp_sal(su); | |
| 3976 | |
| 3977 /* | |
| 346 | 3978 * 3. Try finding sound-a-like words. |
| 344 | 3979 */ |
| 625 | 3980 if ((sps_flags & SPS_FAST) == 0) |
| 3981 { | |
| 3982 if (sps_flags & SPS_BEST) | |
| 3983 /* Adjust the word score for the suggestions found so far for how | |
| 3984 * they sounds like. */ | |
| 3985 rescore_suggestions(su); | |
| 3986 | |
| 3987 /* | |
|
2046
d433f43849ba
updated for version 7.2.332
Bram Moolenaar <bram@zimbu.org>
parents:
1938
diff
changeset
|
3988 * While going through the soundfold tree "su_maxscore" is the score |
| 625 | 3989 * for the soundfold word, limits the changes that are being tried, |
| 3990 * and "su_sfmaxscore" the rescored score, which is set by | |
| 3991 * cleanup_suggestions(). | |
| 3992 * First find words with a small edit distance, because this is much | |
| 3993 * faster and often already finds the top-N suggestions. If we didn't | |
| 3994 * find many suggestions try again with a higher edit distance. | |
| 3995 * "sl_sounddone" is used to avoid doing the same word twice. | |
| 3996 */ | |
| 3997 suggest_try_soundalike_prep(); | |
| 3998 su->su_maxscore = SCORE_SFMAX1; | |
| 3999 su->su_sfmaxscore = SCORE_MAXINIT * 3; | |
| 346 | 4000 suggest_try_soundalike(su); |
| 625 | 4001 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) |
| 4002 { | |
| 4003 /* We didn't find enough matches, try again, allowing more | |
| 4004 * changes to the soundfold word. */ | |
| 4005 su->su_maxscore = SCORE_SFMAX2; | |
| 4006 suggest_try_soundalike(su); | |
| 4007 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su)) | |
| 4008 { | |
| 4009 /* Still didn't find enough matches, try again, allowing even | |
| 4010 * more changes to the soundfold word. */ | |
| 4011 su->su_maxscore = SCORE_SFMAX3; | |
| 4012 suggest_try_soundalike(su); | |
| 4013 } | |
| 4014 } | |
| 4015 su->su_maxscore = su->su_sfmaxscore; | |
| 4016 suggest_try_soundalike_finish(); | |
| 4017 } | |
| 4018 | |
| 4019 /* When CTRL-C was hit while searching do show the results. Only clear | |
| 4020 * got_int when using a command, not for spellsuggest(). */ | |
| 344 | 4021 ui_breakcheck(); |
| 625 | 4022 if (interactive && got_int) |
| 344 | 4023 { |
| 4024 (void)vgetc(); | |
| 4025 got_int = FALSE; | |
| 4026 } | |
| 4027 | |
| 372 | 4028 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0) |
| 344 | 4029 { |
| 4030 if (sps_flags & SPS_BEST) | |
| 4031 /* Adjust the word score for how it sounds like. */ | |
| 4032 rescore_suggestions(su); | |
| 4033 | |
| 625 | 4034 /* Remove bogus suggestions, sort and truncate at "maxcount". */ |
| 4035 check_suggestions(su, &su->su_ga); | |
| 372 | 4036 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 344 | 4037 } |
| 4038 } | |
| 4039 | |
| 4040 /* | |
| 4041 * Free the info put in "*su" by spell_find_suggest(). | |
| 4042 */ | |
| 4043 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4044 spell_find_cleanup(suginfo_T *su) |
| 344 | 4045 { |
| 4046 int i; | |
| 4047 | |
| 4048 /* Free the suggestions. */ | |
| 4049 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 4050 vim_free(SUG(su->su_ga, i).st_word); | |
| 4051 ga_clear(&su->su_ga); | |
| 4052 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 4053 vim_free(SUG(su->su_sga, i).st_word); | |
| 4054 ga_clear(&su->su_sga); | |
| 323 | 4055 |
| 4056 /* Free the banned words. */ | |
| 625 | 4057 hash_clear_all(&su->su_banned, 0); |
| 323 | 4058 } |
| 4059 | |
| 4060 /* | |
| 324 | 4061 * Make a copy of "word", with the first letter upper or lower cased, to |
| 4062 * "wcopy[MAXWLEN]". "word" must not be empty. | |
| 4063 * The result is NUL terminated. | |
| 323 | 4064 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
4065 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4066 onecap_copy( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4067 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4068 char_u *wcopy, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4069 int upper) /* TRUE: first letter made upper case */ |
| 323 | 4070 { |
| 4071 char_u *p; | |
| 4072 int c; | |
| 4073 int l; | |
| 4074 | |
| 4075 p = word; | |
| 4076 #ifdef FEAT_MBYTE | |
| 4077 if (has_mbyte) | |
| 474 | 4078 c = mb_cptr2char_adv(&p); |
| 323 | 4079 else |
| 4080 #endif | |
| 4081 c = *p++; | |
| 4082 if (upper) | |
| 324 | 4083 c = SPELL_TOUPPER(c); |
| 323 | 4084 else |
| 324 | 4085 c = SPELL_TOFOLD(c); |
| 323 | 4086 #ifdef FEAT_MBYTE |
| 4087 if (has_mbyte) | |
| 4088 l = mb_char2bytes(c, wcopy); | |
| 4089 else | |
| 4090 #endif | |
| 4091 { | |
| 4092 l = 1; | |
| 4093 wcopy[0] = c; | |
| 4094 } | |
| 376 | 4095 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1); |
| 323 | 4096 } |
| 4097 | |
| 4098 /* | |
| 324 | 4099 * Make a copy of "word" with all the letters upper cased into |
| 4100 * "wcopy[MAXWLEN]". The result is NUL terminated. | |
| 323 | 4101 */ |
| 4102 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4103 allcap_copy(char_u *word, char_u *wcopy) |
| 323 | 4104 { |
| 4105 char_u *s; | |
| 4106 char_u *d; | |
| 4107 int c; | |
| 4108 | |
| 4109 d = wcopy; | |
| 4110 for (s = word; *s != NUL; ) | |
| 4111 { | |
| 4112 #ifdef FEAT_MBYTE | |
| 4113 if (has_mbyte) | |
| 474 | 4114 c = mb_cptr2char_adv(&s); |
| 323 | 4115 else |
| 4116 #endif | |
| 4117 c = *s++; | |
| 492 | 4118 |
| 4119 #ifdef FEAT_MBYTE | |
| 3042 | 4120 /* We only change 0xdf to SS when we are certain latin1 is used. It |
| 492 | 4121 * would cause weird errors in other 8-bit encodings. */ |
| 4122 if (enc_latin1like && c == 0xdf) | |
| 4123 { | |
| 4124 c = 'S'; | |
| 4125 if (d - wcopy >= MAXWLEN - 1) | |
| 4126 break; | |
| 4127 *d++ = c; | |
| 4128 } | |
| 4129 else | |
| 4130 #endif | |
| 4131 c = SPELL_TOUPPER(c); | |
| 323 | 4132 |
| 4133 #ifdef FEAT_MBYTE | |
| 4134 if (has_mbyte) | |
| 4135 { | |
| 4136 if (d - wcopy >= MAXWLEN - MB_MAXBYTES) | |
| 4137 break; | |
| 4138 d += mb_char2bytes(c, d); | |
| 4139 } | |
| 4140 else | |
| 4141 #endif | |
| 4142 { | |
| 4143 if (d - wcopy >= MAXWLEN - 1) | |
| 4144 break; | |
| 4145 *d++ = c; | |
| 4146 } | |
| 4147 } | |
| 4148 *d = NUL; | |
| 4149 } | |
| 4150 | |
| 4151 /* | |
| 346 | 4152 * Try finding suggestions by recognizing specific situations. |
| 4153 */ | |
| 4154 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4155 suggest_try_special(suginfo_T *su) |
| 346 | 4156 { |
| 4157 char_u *p; | |
| 376 | 4158 size_t len; |
| 346 | 4159 int c; |
| 4160 char_u word[MAXWLEN]; | |
| 4161 | |
| 4162 /* | |
| 4163 * Recognize a word that is repeated: "the the". | |
| 4164 */ | |
| 4165 p = skiptowhite(su->su_fbadword); | |
| 4166 len = p - su->su_fbadword; | |
| 4167 p = skipwhite(p); | |
| 4168 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0) | |
| 4169 { | |
| 4170 /* Include badflags: if the badword is onecap or allcap | |
| 4171 * use that for the goodword too: "The the" -> "The". */ | |
| 4172 c = su->su_fbadword[len]; | |
| 4173 su->su_fbadword[len] = NUL; | |
| 4174 make_case_word(su->su_fbadword, word, su->su_badflags); | |
| 4175 su->su_fbadword[len] = c; | |
| 534 | 4176 |
| 4177 /* Give a soundalike score of 0, compute the score as if deleting one | |
| 4178 * character. */ | |
| 4179 add_suggestion(su, &su->su_ga, word, su->su_badlen, | |
| 625 | 4180 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE); |
| 346 | 4181 } |
| 4182 } | |
| 4183 | |
| 4184 /* | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4185 * 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
|
4186 * Output is dumped in "suggestprof". |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4187 */ |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4188 #if 0 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4189 # define SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4190 proftime_T current; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4191 proftime_T total; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4192 proftime_T times[STATE_FINAL + 1]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4193 long counts[STATE_FINAL + 1]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4194 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4195 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4196 prof_init(void) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4197 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4198 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
|
4199 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4200 profile_zero(×[i]); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4201 counts[i] = 0; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4202 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4203 profile_start(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4204 profile_start(&total); |
|
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 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4207 /* call before changing state */ |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4208 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4209 prof_store(state_T state) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4210 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4211 profile_end(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4212 profile_add(×[state], ¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4213 ++counts[state]; |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4214 profile_start(¤t); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4215 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4216 # 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
|
4217 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4218 static void |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4219 prof_report(char *name) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4220 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4221 FILE *fd = fopen("suggestprof", "a"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4222 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4223 profile_end(&total); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4224 fprintf(fd, "-----------------------\n"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4225 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
|
4226 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
|
4227 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
|
4228 fclose(fd); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4229 } |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4230 #else |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4231 # define PROF_STORE(state) |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4232 #endif |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4233 |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4234 /* |
| 323 | 4235 * Try finding suggestions by adding/removing/swapping letters. |
| 4236 */ | |
| 4237 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4238 suggest_try_change(suginfo_T *su) |
| 323 | 4239 { |
| 4240 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */ | |
| 625 | 4241 int n; |
| 4242 char_u *p; | |
| 4243 int lpi; | |
| 323 | 4244 langp_T *lp; |
| 4245 | |
| 4246 /* We make a copy of the case-folded bad word, so that we can modify it | |
| 346 | 4247 * to find matches (esp. REP items). Append some more text, changing |
| 4248 * chars after the bad word may help. */ | |
| 323 | 4249 STRCPY(fword, su->su_fbadword); |
| 835 | 4250 n = (int)STRLEN(fword); |
| 346 | 4251 p = su->su_badptr + su->su_badlen; |
| 835 | 4252 (void)spell_casefold(p, (int)STRLEN(p), fword + n, MAXWLEN - n); |
| 323 | 4253 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4254 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
|
4255 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4256 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 482 | 4257 |
| 500 | 4258 /* If reloading a spell file fails it's still in the list but |
| 4259 * everything has been cleared. */ | |
| 625 | 4260 if (lp->lp_slang->sl_fbyts == NULL) |
| 500 | 4261 continue; |
| 4262 | |
| 625 | 4263 /* 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
|
4264 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4265 prof_init(); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4266 #endif |
| 625 | 4267 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
|
4268 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4269 prof_report("try_change"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4270 #endif |
| 625 | 4271 } |
| 4272 } | |
| 4273 | |
| 4274 /* Check the maximum score, if we go over it we won't try this change. */ | |
| 4275 #define TRY_DEEPER(su, stack, depth, add) \ | |
| 4276 (stack[depth].ts_score + (add) < su->su_maxscore) | |
| 4277 | |
| 4278 /* | |
| 4279 * Try finding suggestions by adding/removing/swapping letters. | |
| 4280 * | |
| 4281 * This uses a state machine. At each node in the tree we try various | |
| 4282 * operations. When trying if an operation works "depth" is increased and the | |
| 4283 * stack[] is used to store info. This allows combinations, thus insert one | |
| 4284 * character, replace one and delete another. The number of changes is | |
| 4285 * limited by su->su_maxscore. | |
| 4286 * | |
| 4287 * After implementing this I noticed an article by Kemal Oflazer that | |
| 4288 * describes something similar: "Error-tolerant Finite State Recognition with | |
| 4289 * Applications to Morphological Analysis and Spelling Correction" (1996). | |
| 4290 * The implementation in the article is simplified and requires a stack of | |
| 4291 * unknown depth. The implementation here only needs a stack depth equal to | |
| 4292 * the length of the word. | |
| 4293 * | |
| 4294 * This is also used for the sound-folded word, "soundfold" is TRUE then. | |
| 4295 * The mechanism is the same, but we find a match with a sound-folded word | |
| 4296 * that comes from one or more original words. Each of these words may be | |
| 4297 * added, this is done by add_sound_suggest(). | |
| 4298 * Don't use: | |
| 4299 * the prefix tree or the keep-case tree | |
| 4300 * "su->su_badlen" | |
| 4301 * anything to do with upper and lower case | |
| 4302 * anything to do with word or non-word characters ("spell_iswordp()") | |
| 4303 * banned words | |
| 4304 * word flags (rare, region, compounding) | |
| 4305 * word splitting for now | |
| 4306 * "similar_chars()" | |
| 4307 * use "slang->sl_repsal" instead of "lp->lp_replang->sl_rep" | |
| 4308 */ | |
| 4309 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4310 suggest_trie_walk( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4311 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4312 langp_T *lp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4313 char_u *fword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
4314 int soundfold) |
| 625 | 4315 { |
| 4316 char_u tword[MAXWLEN]; /* good word collected so far */ | |
| 4317 trystate_T stack[MAXWLEN]; | |
| 4318 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
|
4319 * concatenation of prefix compound |
| 625 | 4320 * words and split word. NUL terminated |
| 4321 * when going deeper but not when coming | |
| 4322 * back. */ | |
| 4323 char_u compflags[MAXWLEN]; /* compound flags, one for each word */ | |
| 4324 trystate_T *sp; | |
| 4325 int newscore; | |
| 4326 int score; | |
| 4327 char_u *byts, *fbyts, *pbyts; | |
| 4328 idx_T *idxs, *fidxs, *pidxs; | |
| 4329 int depth; | |
| 4330 int c, c2, c3; | |
| 4331 int n = 0; | |
| 4332 int flags; | |
| 4333 garray_T *gap; | |
| 4334 idx_T arridx; | |
| 4335 int len; | |
| 4336 char_u *p; | |
| 4337 fromto_T *ftp; | |
| 4338 int fl = 0, tl; | |
| 4339 int repextra = 0; /* extra bytes in fword[] from REP item */ | |
| 4340 slang_T *slang = lp->lp_slang; | |
| 4341 int fword_ends; | |
| 4342 int goodword_ends; | |
| 4343 #ifdef DEBUG_TRIEWALK | |
| 4344 /* Stores the name of the change made at each level. */ | |
| 4345 char_u changename[MAXWLEN][80]; | |
| 4346 #endif | |
| 4347 int breakcheckcount = 1000; | |
| 4348 int compound_ok; | |
| 4349 | |
| 4350 /* | |
| 4351 * Go through the whole case-fold tree, try changes at each node. | |
| 4352 * "tword[]" contains the word collected from nodes in the tree. | |
| 4353 * "fword[]" the word we are trying to match with (initially the bad | |
| 4354 * word). | |
| 4355 */ | |
| 4356 depth = 0; | |
| 4357 sp = &stack[0]; | |
| 4358 vim_memset(sp, 0, sizeof(trystate_T)); | |
| 4359 sp->ts_curi = 1; | |
| 4360 | |
| 4361 if (soundfold) | |
| 4362 { | |
| 4363 /* Going through the soundfold tree. */ | |
| 4364 byts = fbyts = slang->sl_sbyts; | |
| 4365 idxs = fidxs = slang->sl_sidxs; | |
| 4366 pbyts = NULL; | |
| 4367 pidxs = NULL; | |
| 4368 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 4369 sp->ts_state = STATE_START; | |
| 4370 } | |
| 4371 else | |
| 4372 { | |
| 330 | 4373 /* |
| 375 | 4374 * When there are postponed prefixes we need to use these first. At |
| 4375 * the end of the prefix we continue in the case-fold tree. | |
| 4376 */ | |
| 482 | 4377 fbyts = slang->sl_fbyts; |
| 4378 fidxs = slang->sl_fidxs; | |
| 4379 pbyts = slang->sl_pbyts; | |
| 4380 pidxs = slang->sl_pidxs; | |
| 375 | 4381 if (pbyts != NULL) |
| 4382 { | |
| 4383 byts = pbyts; | |
| 4384 idxs = pidxs; | |
| 482 | 4385 sp->ts_prefixdepth = PFD_PREFIXTREE; |
| 375 | 4386 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */ |
| 4387 } | |
| 4388 else | |
| 4389 { | |
| 4390 byts = fbyts; | |
| 4391 idxs = fidxs; | |
| 482 | 4392 sp->ts_prefixdepth = PFD_NOPREFIX; |
| 485 | 4393 sp->ts_state = STATE_START; |
| 375 | 4394 } |
| 625 | 4395 } |
| 4396 | |
| 4397 /* | |
| 4398 * Loop to find all suggestions. At each round we either: | |
| 4399 * - For the current state try one operation, advance "ts_curi", | |
| 4400 * increase "depth". | |
| 4401 * - When a state is done go to the next, set "ts_state". | |
| 4402 * - When all states are tried decrease "depth". | |
| 4403 */ | |
| 4404 while (depth >= 0 && !got_int) | |
| 4405 { | |
| 4406 sp = &stack[depth]; | |
| 4407 switch (sp->ts_state) | |
| 4408 { | |
| 4409 case STATE_START: | |
| 4410 case STATE_NOPREFIX: | |
| 4411 /* | |
| 4412 * Start of node: Deal with NUL bytes, which means | |
| 4413 * tword[] may end here. | |
| 4414 */ | |
| 4415 arridx = sp->ts_arridx; /* current node in the tree */ | |
| 4416 len = byts[arridx]; /* bytes in this node */ | |
| 4417 arridx += sp->ts_curi; /* index of current byte */ | |
| 4418 | |
| 4419 if (sp->ts_prefixdepth == PFD_PREFIXTREE) | |
| 4420 { | |
| 4421 /* Skip over the NUL bytes, we use them later. */ | |
| 4422 for (n = 0; n < len && byts[arridx + n] == 0; ++n) | |
| 4423 ; | |
| 4424 sp->ts_curi += n; | |
| 4425 | |
| 4426 /* Always past NUL bytes now. */ | |
| 4427 n = (int)sp->ts_state; | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4428 PROF_STORE(sp->ts_state) |
| 625 | 4429 sp->ts_state = STATE_ENDNUL; |
| 4430 sp->ts_save_badflags = su->su_badflags; | |
| 4431 | |
| 4432 /* At end of a prefix or at start of prefixtree: check for | |
| 4433 * following word. */ | |
| 4434 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX) | |
| 4435 { | |
| 4436 /* Set su->su_badflags to the caps type at this position. | |
| 4437 * Use the caps type until here for the prefix itself. */ | |
| 4438 #ifdef FEAT_MBYTE | |
| 4439 if (has_mbyte) | |
| 4440 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); | |
| 4441 else | |
| 4442 #endif | |
| 4443 n = sp->ts_fidx; | |
| 4444 flags = badword_captype(su->su_badptr, su->su_badptr + n); | |
| 4445 su->su_badflags = badword_captype(su->su_badptr + n, | |
| 455 | 4446 su->su_badptr + su->su_badlen); |
| 625 | 4447 #ifdef DEBUG_TRIEWALK |
| 4448 sprintf(changename[depth], "prefix"); | |
| 4449 #endif | |
| 4450 go_deeper(stack, depth, 0); | |
| 4451 ++depth; | |
| 4452 sp = &stack[depth]; | |
| 4453 sp->ts_prefixdepth = depth - 1; | |
| 4454 byts = fbyts; | |
| 4455 idxs = fidxs; | |
| 4456 sp->ts_arridx = 0; | |
| 4457 | |
| 4458 /* Move the prefix to preword[] with the right case | |
| 4459 * and make find_keepcap_word() works. */ | |
| 4460 tword[sp->ts_twordlen] = NUL; | |
| 4461 make_case_word(tword + sp->ts_splitoff, | |
| 4462 preword + sp->ts_prewordlen, flags); | |
| 835 | 4463 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 625 | 4464 sp->ts_splitoff = sp->ts_twordlen; |
| 4465 } | |
| 4466 break; | |
| 4467 } | |
| 4468 | |
| 4469 if (sp->ts_curi > len || byts[arridx] != 0) | |
| 4470 { | |
| 4471 /* 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
|
4472 PROF_STORE(sp->ts_state) |
| 625 | 4473 sp->ts_state = STATE_ENDNUL; |
| 4474 sp->ts_save_badflags = su->su_badflags; | |
| 4475 break; | |
| 4476 } | |
| 4477 | |
| 4478 /* | |
| 4479 * End of word in tree. | |
| 4480 */ | |
| 4481 ++sp->ts_curi; /* eat one NUL byte */ | |
| 4482 | |
| 4483 flags = (int)idxs[arridx]; | |
| 698 | 4484 |
| 4485 /* Skip words with the NOSUGGEST flag. */ | |
| 4486 if (flags & WF_NOSUGGEST) | |
| 4487 break; | |
| 4488 | |
| 625 | 4489 fword_ends = (fword[sp->ts_fidx] == NUL |
| 4490 || (soundfold | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
4491 ? VIM_ISWHITE(fword[sp->ts_fidx]) |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4492 : !spell_iswordp(fword + sp->ts_fidx, curwin))); |
| 625 | 4493 tword[sp->ts_twordlen] = NUL; |
| 4494 | |
| 4495 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL | |
| 485 | 4496 && (sp->ts_flags & TSF_PREFIXOK) == 0) |
| 625 | 4497 { |
| 4498 /* There was a prefix before the word. Check that the prefix | |
| 4499 * can be used with this word. */ | |
| 4500 /* Count the length of the NULs in the prefix. If there are | |
| 4501 * none this must be the first try without a prefix. */ | |
| 4502 n = stack[sp->ts_prefixdepth].ts_arridx; | |
| 4503 len = pbyts[n++]; | |
| 4504 for (c = 0; c < len && pbyts[n + c] == 0; ++c) | |
| 4505 ; | |
| 4506 if (c > 0) | |
| 4507 { | |
| 4508 c = valid_word_prefix(c, n, flags, | |
| 483 | 4509 tword + sp->ts_splitoff, slang, FALSE); |
| 625 | 4510 if (c == 0) |
| 4511 break; | |
| 4512 | |
| 4513 /* Use the WF_RARE flag for a rare prefix. */ | |
| 4514 if (c & WF_RAREPFX) | |
| 4515 flags |= WF_RARE; | |
| 4516 | |
| 4517 /* Tricky: when checking for both prefix and compounding | |
| 4518 * we run into the prefix flag first. | |
| 4519 * Remember that it's OK, so that we accept the prefix | |
| 4520 * when arriving at a compound flag. */ | |
| 4521 sp->ts_flags |= TSF_PREFIXOK; | |
| 4522 } | |
| 4523 } | |
| 4524 | |
| 4525 /* Check NEEDCOMPOUND: can't use word without compounding. Do try | |
| 4526 * appending another compound word below. */ | |
| 4527 if (sp->ts_complen == sp->ts_compsplit && fword_ends | |
| 500 | 4528 && (flags & WF_NEEDCOMP)) |
| 625 | 4529 goodword_ends = FALSE; |
| 4530 else | |
| 4531 goodword_ends = TRUE; | |
| 4532 | |
| 4533 p = NULL; | |
| 4534 compound_ok = TRUE; | |
| 4535 if (sp->ts_complen > sp->ts_compsplit) | |
| 4536 { | |
| 4537 if (slang->sl_nobreak) | |
| 4538 { | |
| 4539 /* There was a word before this word. When there was no | |
| 4540 * change in this word (it was correct) add the first word | |
| 4541 * as a suggestion. If this word was corrected too, we | |
| 4542 * need to check if a correct word follows. */ | |
| 4543 if (sp->ts_fidx - sp->ts_splitfidx | |
| 4544 == sp->ts_twordlen - sp->ts_splitoff | |
| 4545 && STRNCMP(fword + sp->ts_splitfidx, | |
| 4546 tword + sp->ts_splitoff, | |
| 4547 sp->ts_fidx - sp->ts_splitfidx) == 0) | |
| 492 | 4548 { |
| 625 | 4549 preword[sp->ts_prewordlen] = NUL; |
| 4550 newscore = score_wordcount_adj(slang, sp->ts_score, | |
| 4551 preword + sp->ts_prewordlen, | |
| 4552 sp->ts_prewordlen > 0); | |
| 4553 /* Add the suggestion if the score isn't too bad. */ | |
| 4554 if (newscore <= su->su_maxscore) | |
| 492 | 4555 add_suggestion(su, &su->su_ga, preword, |
| 503 | 4556 sp->ts_splitfidx - repextra, |
| 625 | 4557 newscore, 0, FALSE, |
| 4558 lp->lp_sallang, FALSE); | |
| 4559 break; | |
| 492 | 4560 } |
| 490 | 4561 } |
| 4562 else | |
| 625 | 4563 { |
| 4564 /* There was a compound word before this word. If this | |
| 4565 * word does not support compounding then give up | |
| 4566 * (splitting is tried for the word without compound | |
| 4567 * flag). */ | |
| 4568 if (((unsigned)flags >> 24) == 0 | |
| 4569 || sp->ts_twordlen - sp->ts_splitoff | |
| 4570 < slang->sl_compminlen) | |
| 4571 break; | |
| 4572 #ifdef FEAT_MBYTE | |
| 4573 /* For multi-byte chars check character length against | |
| 4574 * COMPOUNDMIN. */ | |
| 4575 if (has_mbyte | |
| 4576 && slang->sl_compminlen > 0 | |
| 4577 && mb_charlen(tword + sp->ts_splitoff) | |
| 4578 < slang->sl_compminlen) | |
| 4579 break; | |
| 4580 #endif | |
| 4581 | |
| 4582 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 4583 compflags[sp->ts_complen + 1] = NUL; | |
| 4584 vim_strncpy(preword + sp->ts_prewordlen, | |
| 4585 tword + sp->ts_splitoff, | |
| 4586 sp->ts_twordlen - sp->ts_splitoff); | |
| 1762 | 4587 |
| 4588 /* Verify CHECKCOMPOUNDPATTERN rules. */ | |
| 4589 if (match_checkcompoundpattern(preword, sp->ts_prewordlen, | |
| 4590 &slang->sl_comppat)) | |
| 4591 compound_ok = FALSE; | |
| 4592 | |
| 4593 if (compound_ok) | |
| 4594 { | |
| 4595 p = preword; | |
| 4596 while (*skiptowhite(p) != NUL) | |
| 4597 p = skipwhite(skiptowhite(p)); | |
| 4598 if (fword_ends && !can_compound(slang, p, | |
| 625 | 4599 compflags + sp->ts_compsplit)) |
| 1762 | 4600 /* Compound is not allowed. But it may still be |
| 4601 * possible if we add another (short) word. */ | |
| 4602 compound_ok = FALSE; | |
| 4603 } | |
| 625 | 4604 |
| 4605 /* Get pointer to last char of previous word. */ | |
| 4606 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
|
4607 MB_PTR_BACK(preword, p); |
| 625 | 4608 } |
| 4609 } | |
| 4610 | |
| 4611 /* | |
| 4612 * Form the word with proper case in preword. | |
| 4613 * If there is a word from a previous split, append. | |
| 4614 * For the soundfold tree don't change the case, simply append. | |
| 4615 */ | |
| 4616 if (soundfold) | |
| 4617 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff); | |
| 4618 else if (flags & WF_KEEPCAP) | |
| 4619 /* Must find the word in the keep-case tree. */ | |
| 4620 find_keepcap_word(slang, tword + sp->ts_splitoff, | |
| 483 | 4621 preword + sp->ts_prewordlen); |
| 625 | 4622 else |
| 4623 { | |
| 4624 /* Include badflags: If the badword is onecap or allcap | |
| 4625 * use that for the goodword too. But if the badword is | |
| 4626 * allcap and it's only one char long use onecap. */ | |
| 4627 c = su->su_badflags; | |
| 4628 if ((c & WF_ALLCAP) | |
| 4629 #ifdef FEAT_MBYTE | |
| 4630 && su->su_badlen == (*mb_ptr2len)(su->su_badptr) | |
| 346 | 4631 #else |
| 625 | 4632 && su->su_badlen == 1 |
| 4633 #endif | |
| 4634 ) | |
| 4635 c = WF_ONECAP; | |
| 4636 c |= flags; | |
| 4637 | |
| 4638 /* When appending a compound word after a word character don't | |
| 4639 * use Onecap. */ | |
| 5477 | 4640 if (p != NULL && spell_iswordp_nmw(p, curwin)) |
| 625 | 4641 c &= ~WF_ONECAP; |
| 4642 make_case_word(tword + sp->ts_splitoff, | |
| 490 | 4643 preword + sp->ts_prewordlen, c); |
| 625 | 4644 } |
| 4645 | |
| 4646 if (!soundfold) | |
| 4647 { | |
| 323 | 4648 /* Don't use a banned word. It may appear again as a good |
| 4649 * word, thus remember it. */ | |
| 4650 if (flags & WF_BANNED) | |
| 4651 { | |
| 483 | 4652 add_banned(su, preword + sp->ts_prewordlen); |
| 323 | 4653 break; |
| 4654 } | |
| 534 | 4655 if ((sp->ts_complen == sp->ts_compsplit |
| 625 | 4656 && WAS_BANNED(su, preword + sp->ts_prewordlen)) |
| 4657 || WAS_BANNED(su, preword)) | |
| 534 | 4658 { |
| 4659 if (slang->sl_compprog == NULL) | |
| 4660 break; | |
| 4661 /* the word so far was banned but we may try compounding */ | |
| 4662 goodword_ends = FALSE; | |
| 4663 } | |
| 625 | 4664 } |
| 4665 | |
| 4666 newscore = 0; | |
| 4667 if (!soundfold) /* soundfold words don't have flags */ | |
| 4668 { | |
| 323 | 4669 if ((flags & WF_REGION) |
| 390 | 4670 && (((unsigned)flags >> 16) & lp->lp_region) == 0) |
| 323 | 4671 newscore += SCORE_REGION; |
| 4672 if (flags & WF_RARE) | |
| 4673 newscore += SCORE_RARE; | |
| 4674 | |
| 346 | 4675 if (!spell_valid_case(su->su_badflags, |
| 483 | 4676 captype(preword + sp->ts_prewordlen, NULL))) |
| 323 | 4677 newscore += SCORE_ICASE; |
| 625 | 4678 } |
| 4679 | |
| 4680 /* TODO: how about splitting in the soundfold tree? */ | |
| 4681 if (fword_ends | |
| 4682 && goodword_ends | |
| 4683 && sp->ts_fidx >= sp->ts_fidxtry | |
| 4684 && compound_ok) | |
| 4685 { | |
| 4686 /* The badword also ends: add suggestions. */ | |
| 4687 #ifdef DEBUG_TRIEWALK | |
| 4688 if (soundfold && STRCMP(preword, "smwrd") == 0) | |
| 4689 { | |
| 4690 int j; | |
| 4691 | |
| 4692 /* print the stack of changes that brought us here */ | |
| 4693 smsg("------ %s -------", fword); | |
| 4694 for (j = 0; j < depth; ++j) | |
| 4695 smsg("%s", changename[j]); | |
| 4696 } | |
| 4697 #endif | |
| 4698 if (soundfold) | |
| 4699 { | |
| 4700 /* For soundfolded words we need to find the original | |
| 1188 | 4701 * words, the edit distance and then add them. */ |
| 625 | 4702 add_sound_suggest(su, preword, sp->ts_score, lp); |
| 4703 } | |
|
2457
6c1d10a85398
Avoid illegal memory access in spell suggestion. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2455
diff
changeset
|
4704 else if (sp->ts_fidx > 0) |
| 625 | 4705 { |
| 4706 /* Give a penalty when changing non-word char to word | |
| 4707 * char, e.g., "thes," -> "these". */ | |
| 366 | 4708 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
|
4709 MB_PTR_BACK(fword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4710 if (!spell_iswordp(p, curwin)) |
| 366 | 4711 { |
| 4712 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
|
4713 MB_PTR_BACK(preword, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
4714 if (spell_iswordp(p, curwin)) |
| 366 | 4715 newscore += SCORE_NONWORD; |
| 4716 } | |
| 4717 | |
| 625 | 4718 /* Give a bonus to words seen before. */ |
| 4719 score = score_wordcount_adj(slang, | |
| 4720 sp->ts_score + newscore, | |
| 4721 preword + sp->ts_prewordlen, | |
| 4722 sp->ts_prewordlen > 0); | |
| 4723 | |
| 4724 /* Add the suggestion if the score isn't too bad. */ | |
| 4725 if (score <= su->su_maxscore) | |
| 633 | 4726 { |
| 625 | 4727 add_suggestion(su, &su->su_ga, preword, |
| 4728 sp->ts_fidx - repextra, | |
| 4729 score, 0, FALSE, lp->lp_sallang, FALSE); | |
| 633 | 4730 |
| 4731 if (su->su_badflags & WF_MIXCAP) | |
| 4732 { | |
| 4733 /* We really don't know if the word should be | |
| 4734 * upper or lower case, add both. */ | |
| 4735 c = captype(preword, NULL); | |
| 4736 if (c == 0 || c == WF_ALLCAP) | |
| 4737 { | |
| 4738 make_case_word(tword + sp->ts_splitoff, | |
| 4739 preword + sp->ts_prewordlen, | |
| 4740 c == 0 ? WF_ALLCAP : 0); | |
| 4741 | |
| 4742 add_suggestion(su, &su->su_ga, preword, | |
| 4743 sp->ts_fidx - repextra, | |
| 4744 score + SCORE_ICASE, 0, FALSE, | |
| 4745 lp->lp_sallang, FALSE); | |
| 4746 } | |
| 4747 } | |
| 4748 } | |
| 625 | 4749 } |
| 4750 } | |
| 4751 | |
| 4752 /* | |
| 4753 * Try word split and/or compounding. | |
| 4754 */ | |
| 4755 if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends) | |
| 4756 #ifdef FEAT_MBYTE | |
| 4757 /* Don't split halfway a character. */ | |
| 4758 && (!has_mbyte || sp->ts_tcharlen == 0) | |
| 4759 #endif | |
| 4760 ) | |
| 4761 { | |
| 4762 int try_compound; | |
| 4763 int try_split; | |
| 4764 | |
| 4765 /* If past the end of the bad word don't try a split. | |
| 4766 * Otherwise try changing the next word. E.g., find | |
| 4767 * suggestions for "the the" where the second "the" is | |
| 4768 * different. It's done like a split. | |
| 4769 * TODO: word split for soundfold words */ | |
| 4770 try_split = (sp->ts_fidx - repextra < su->su_badlen) | |
| 4771 && !soundfold; | |
| 4772 | |
| 4773 /* Get here in several situations: | |
| 4774 * 1. The word in the tree ends: | |
| 4775 * If the word allows compounding try that. Otherwise try | |
| 4776 * a split by inserting a space. For both check that a | |
| 4777 * valid words starts at fword[sp->ts_fidx]. | |
| 4778 * For NOBREAK do like compounding to be able to check if | |
| 4779 * the next word is valid. | |
| 4780 * 2. The badword does end, but it was due to a change (e.g., | |
| 4781 * a swap). No need to split, but do check that the | |
| 4782 * following word is valid. | |
| 4783 * 3. The badword and the word in the tree end. It may still | |
| 4784 * be possible to compound another (short) word. | |
| 4785 */ | |
| 4786 try_compound = FALSE; | |
| 4787 if (!soundfold | |
|
7526
08c1f73efcde
commit https://github.com/vim/vim/commit/7b877b360532713dc21a0ff3d55a76ac02eaf573
Christian Brabandt <cb@256bit.org>
parents:
7509
diff
changeset
|
4788 && !slang->sl_nocompoundsugs |
| 625 | 4789 && slang->sl_compprog != NULL |
| 4790 && ((unsigned)flags >> 24) != 0 | |
| 4791 && sp->ts_twordlen - sp->ts_splitoff | |
| 4792 >= slang->sl_compminlen | |
| 4793 #ifdef FEAT_MBYTE | |
| 4794 && (!has_mbyte | |
| 4795 || slang->sl_compminlen == 0 | |
| 4796 || mb_charlen(tword + sp->ts_splitoff) | |
| 4797 >= slang->sl_compminlen) | |
| 4798 #endif | |
| 4799 && (slang->sl_compsylmax < MAXWLEN | |
| 4800 || sp->ts_complen + 1 - sp->ts_compsplit | |
| 4801 < slang->sl_compmax) | |
| 1762 | 4802 && (can_be_compound(sp, slang, |
| 4803 compflags, ((unsigned)flags >> 24)))) | |
| 4804 | |
| 625 | 4805 { |
| 4806 try_compound = TRUE; | |
| 4807 compflags[sp->ts_complen] = ((unsigned)flags >> 24); | |
| 4808 compflags[sp->ts_complen + 1] = NUL; | |
| 4809 } | |
| 4810 | |
| 4811 /* For NOBREAK we never try splitting, it won't make any word | |
| 4812 * valid. */ | |
|
7526
08c1f73efcde
commit https://github.com/vim/vim/commit/7b877b360532713dc21a0ff3d55a76ac02eaf573
Christian Brabandt <cb@256bit.org>
parents:
7509
diff
changeset
|
4813 if (slang->sl_nobreak && !slang->sl_nocompoundsugs) |
| 625 | 4814 try_compound = TRUE; |
| 4815 | |
| 4816 /* If we could add a compound word, and it's also possible to | |
| 4817 * split at this point, do the split first and set | |
| 4818 * TSF_DIDSPLIT to avoid doing it again. */ | |
| 4819 else if (!fword_ends | |
| 4820 && try_compound | |
| 4821 && (sp->ts_flags & TSF_DIDSPLIT) == 0) | |
| 4822 { | |
| 485 | 4823 try_compound = FALSE; |
| 625 | 4824 sp->ts_flags |= TSF_DIDSPLIT; |
| 4825 --sp->ts_curi; /* do the same NUL again */ | |
| 4826 compflags[sp->ts_complen] = NUL; | |
| 4827 } | |
| 4828 else | |
| 4829 sp->ts_flags &= ~TSF_DIDSPLIT; | |
| 4830 | |
| 4831 if (try_split || try_compound) | |
| 4832 { | |
| 534 | 4833 if (!try_compound && (!fword_ends || !goodword_ends)) |
| 485 | 4834 { |
| 4835 /* If we're going to split need to check that the | |
| 501 | 4836 * words so far are valid for compounding. If there |
| 4837 * is only one word it must not have the NEEDCOMPOUND | |
| 4838 * flag. */ | |
| 4839 if (sp->ts_complen == sp->ts_compsplit | |
| 4840 && (flags & WF_NEEDCOMP)) | |
| 4841 break; | |
| 490 | 4842 p = preword; |
| 4843 while (*skiptowhite(p) != NUL) | |
| 4844 p = skipwhite(skiptowhite(p)); | |
| 485 | 4845 if (sp->ts_complen > sp->ts_compsplit |
| 490 | 4846 && !can_compound(slang, p, |
| 485 | 4847 compflags + sp->ts_compsplit)) |
| 4848 break; | |
| 698 | 4849 |
| 4850 if (slang->sl_nosplitsugs) | |
| 4851 newscore += SCORE_SPLIT_NO; | |
| 4852 else | |
| 4853 newscore += SCORE_SPLIT; | |
| 625 | 4854 |
| 4855 /* Give a bonus to words seen before. */ | |
| 4856 newscore = score_wordcount_adj(slang, newscore, | |
| 4857 preword + sp->ts_prewordlen, TRUE); | |
| 482 | 4858 } |
| 4859 | |
| 625 | 4860 if (TRY_DEEPER(su, stack, depth, newscore)) |
| 323 | 4861 { |
| 625 | 4862 go_deeper(stack, depth, newscore); |
| 4863 #ifdef DEBUG_TRIEWALK | |
| 4864 if (!try_compound && !fword_ends) | |
| 4865 sprintf(changename[depth], "%.*s-%s: split", | |
| 4866 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 4867 else | |
| 4868 sprintf(changename[depth], "%.*s-%s: compound", | |
| 4869 sp->ts_twordlen, tword, fword + sp->ts_fidx); | |
| 4870 #endif | |
| 323 | 4871 /* Save things to be restored at STATE_SPLITUNDO. */ |
| 346 | 4872 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
|
4873 PROF_STORE(sp->ts_state) |
| 376 | 4874 sp->ts_state = STATE_SPLITUNDO; |
| 4875 | |
| 4876 ++depth; | |
| 4877 sp = &stack[depth]; | |
| 323 | 4878 |
| 482 | 4879 /* Append a space to preword when splitting. */ |
| 4880 if (!try_compound && !fword_ends) | |
| 4881 STRCAT(preword, " "); | |
| 835 | 4882 sp->ts_prewordlen = (char_u)STRLEN(preword); |
| 483 | 4883 sp->ts_splitoff = sp->ts_twordlen; |
| 492 | 4884 sp->ts_splitfidx = sp->ts_fidx; |
| 376 | 4885 |
| 4886 /* If the badword has a non-word character at this | |
| 4887 * position skip it. That means replacing the | |
| 482 | 4888 * non-word character with a space. Always skip a |
| 534 | 4889 * character when the word ends. But only when the |
| 4890 * good word can end. */ | |
| 625 | 4891 if (((!try_compound && !spell_iswordp_nmw(fword |
| 5477 | 4892 + sp->ts_fidx, |
| 4893 curwin)) | |
| 625 | 4894 || fword_ends) |
| 4895 && fword[sp->ts_fidx] != NUL | |
| 4896 && goodword_ends) | |
| 376 | 4897 { |
| 482 | 4898 int l; |
| 4899 | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
4900 l = MB_PTR2LEN(fword + sp->ts_fidx); |
| 482 | 4901 if (fword_ends) |
| 4902 { | |
| 4903 /* Copy the skipped character to preword. */ | |
| 483 | 4904 mch_memmove(preword + sp->ts_prewordlen, |
| 482 | 4905 fword + sp->ts_fidx, l); |
| 483 | 4906 sp->ts_prewordlen += l; |
| 4907 preword[sp->ts_prewordlen] = NUL; | |
| 482 | 4908 } |
| 4909 else | |
| 4910 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST; | |
| 4911 sp->ts_fidx += l; | |
| 376 | 4912 } |
| 455 | 4913 |
| 485 | 4914 /* When compounding include compound flag in |
| 4915 * compflags[] (already set above). When splitting we | |
| 4916 * may start compounding over again. */ | |
| 482 | 4917 if (try_compound) |
| 483 | 4918 ++sp->ts_complen; |
| 482 | 4919 else |
| 485 | 4920 sp->ts_compsplit = sp->ts_complen; |
| 4921 sp->ts_prefixdepth = PFD_NOPREFIX; | |
| 482 | 4922 |
| 455 | 4923 /* set su->su_badflags to the caps type at this |
| 4924 * position */ | |
| 324 | 4925 #ifdef FEAT_MBYTE |
| 4926 if (has_mbyte) | |
| 455 | 4927 n = nofold_len(fword, sp->ts_fidx, su->su_badptr); |
| 324 | 4928 else |
| 4929 #endif | |
| 455 | 4930 n = sp->ts_fidx; |
| 474 | 4931 su->su_badflags = badword_captype(su->su_badptr + n, |
| 455 | 4932 su->su_badptr + su->su_badlen); |
| 323 | 4933 |
| 4934 /* Restart at top of the tree. */ | |
| 376 | 4935 sp->ts_arridx = 0; |
| 485 | 4936 |
| 4937 /* If there are postponed prefixes, try these too. */ | |
| 4938 if (pbyts != NULL) | |
| 4939 { | |
| 4940 byts = pbyts; | |
| 4941 idxs = pidxs; | |
| 4942 sp->ts_prefixdepth = PFD_PREFIXTREE; | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4943 PROF_STORE(sp->ts_state) |
| 485 | 4944 sp->ts_state = STATE_NOPREFIX; |
| 4945 } | |
| 323 | 4946 } |
| 4947 } | |
| 625 | 4948 } |
| 4949 break; | |
| 4950 | |
| 4951 case STATE_SPLITUNDO: | |
| 4952 /* Undo the changes done for word split or compound word. */ | |
| 4953 su->su_badflags = sp->ts_save_badflags; | |
| 4954 | |
| 4955 /* Continue looking for NUL bytes. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4956 PROF_STORE(sp->ts_state) |
| 625 | 4957 sp->ts_state = STATE_START; |
| 4958 | |
| 4959 /* In case we went into the prefix tree. */ | |
| 4960 byts = fbyts; | |
| 4961 idxs = fidxs; | |
| 4962 break; | |
| 4963 | |
| 4964 case STATE_ENDNUL: | |
| 4965 /* Past the NUL bytes in the node. */ | |
| 4966 su->su_badflags = sp->ts_save_badflags; | |
| 4967 if (fword[sp->ts_fidx] == NUL | |
| 4968 #ifdef FEAT_MBYTE | |
| 4969 && sp->ts_tcharlen == 0 | |
| 4970 #endif | |
| 4971 ) | |
| 4972 { | |
| 4973 /* 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
|
4974 PROF_STORE(sp->ts_state) |
| 625 | 4975 sp->ts_state = STATE_DEL; |
| 323 | 4976 break; |
| 625 | 4977 } |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
4978 PROF_STORE(sp->ts_state) |
| 625 | 4979 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
|
4980 /* FALLTHROUGH */ |
| 625 | 4981 |
| 4982 case STATE_PLAIN: | |
| 4983 /* | |
| 4984 * Go over all possible bytes at this node, add each to tword[] | |
| 4985 * and use child node. "ts_curi" is the index. | |
| 4986 */ | |
| 4987 arridx = sp->ts_arridx; | |
| 4988 if (sp->ts_curi > byts[arridx]) | |
| 4989 { | |
| 4990 /* Done all bytes at this node, do next state. When still at | |
| 4991 * 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
|
4992 PROF_STORE(sp->ts_state) |
| 625 | 4993 if (sp->ts_fidx >= sp->ts_fidxtry) |
| 323 | 4994 sp->ts_state = STATE_DEL; |
| 4995 else | |
| 625 | 4996 sp->ts_state = STATE_FINAL; |
| 4997 } | |
| 4998 else | |
| 4999 { | |
| 5000 arridx += sp->ts_curi++; | |
| 5001 c = byts[arridx]; | |
| 5002 | |
| 5003 /* Normal byte, go one level deeper. If it's not equal to the | |
| 5004 * byte in the bad word adjust the score. But don't even try | |
| 5005 * when the byte was already changed. And don't try when we | |
| 5695 | 5006 * just deleted this byte, accepting it is always cheaper than |
| 625 | 5007 * delete + substitute. */ |
| 5008 if (c == fword[sp->ts_fidx] | |
| 5009 #ifdef FEAT_MBYTE | |
| 5010 || (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE) | |
| 5011 #endif | |
| 5012 ) | |
| 5013 newscore = 0; | |
| 5014 else | |
| 5015 newscore = SCORE_SUBST; | |
| 5016 if ((newscore == 0 | |
| 5017 || (sp->ts_fidx >= sp->ts_fidxtry | |
| 5018 && ((sp->ts_flags & TSF_DIDDEL) == 0 | |
| 5019 || c != fword[sp->ts_delidx]))) | |
| 5020 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5021 { | |
| 5022 go_deeper(stack, depth, newscore); | |
| 5023 #ifdef DEBUG_TRIEWALK | |
| 5024 if (newscore > 0) | |
| 5025 sprintf(changename[depth], "%.*s-%s: subst %c to %c", | |
| 5026 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5027 fword[sp->ts_fidx], c); | |
| 323 | 5028 else |
| 625 | 5029 sprintf(changename[depth], "%.*s-%s: accept %c", |
| 5030 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5031 fword[sp->ts_fidx]); | |
| 5032 #endif | |
| 323 | 5033 ++depth; |
| 625 | 5034 sp = &stack[depth]; |
| 5035 ++sp->ts_fidx; | |
| 5036 tword[sp->ts_twordlen++] = c; | |
| 5037 sp->ts_arridx = idxs[arridx]; | |
| 5038 #ifdef FEAT_MBYTE | |
| 5039 if (newscore == SCORE_SUBST) | |
| 5040 sp->ts_isdiff = DIFF_YES; | |
| 330 | 5041 if (has_mbyte) |
| 358 | 5042 { |
| 625 | 5043 /* Multi-byte characters are a bit complicated to |
| 5044 * handle: They differ when any of the bytes differ | |
| 5045 * and then their length may also differ. */ | |
| 5046 if (sp->ts_tcharlen == 0) | |
| 5047 { | |
| 5048 /* First byte. */ | |
| 5049 sp->ts_tcharidx = 0; | |
| 5050 sp->ts_tcharlen = MB_BYTE2LEN(c); | |
| 5051 sp->ts_fcharstart = sp->ts_fidx - 1; | |
| 5052 sp->ts_isdiff = (newscore != 0) | |
| 5053 ? DIFF_YES : DIFF_NONE; | |
| 5054 } | |
| 5055 else if (sp->ts_isdiff == DIFF_INSERT) | |
| 5056 /* When inserting trail bytes don't advance in the | |
| 5057 * bad word. */ | |
| 5058 --sp->ts_fidx; | |
| 5059 if (++sp->ts_tcharidx == sp->ts_tcharlen) | |
| 5060 { | |
| 5061 /* Last byte of character. */ | |
| 5062 if (sp->ts_isdiff == DIFF_YES) | |
| 5063 { | |
| 5064 /* Correct ts_fidx for the byte length of the | |
| 5065 * character (we didn't check that before). */ | |
| 5066 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
|
5067 + MB_PTR2LEN( |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5068 fword + sp->ts_fcharstart); |
| 625 | 5069 /* For changing a composing character adjust |
| 5070 * the score from SCORE_SUBST to | |
| 5071 * SCORE_SUBCOMP. */ | |
| 5072 if (enc_utf8 | |
| 5073 && 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
|
5074 utf_ptr2char(tword |
| 625 | 5075 + sp->ts_twordlen |
| 5076 - sp->ts_tcharlen)) | |
| 5077 && 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
|
5078 utf_ptr2char(fword |
| 625 | 5079 + sp->ts_fcharstart))) |
| 5080 sp->ts_score -= | |
| 5081 SCORE_SUBST - SCORE_SUBCOMP; | |
| 5082 | |
| 5083 /* For a similar character adjust score from | |
| 5084 * SCORE_SUBST to SCORE_SIMILAR. */ | |
| 5085 else if (!soundfold | |
| 5086 && slang->sl_has_map | |
| 5087 && similar_chars(slang, | |
| 5088 mb_ptr2char(tword | |
| 5089 + sp->ts_twordlen | |
| 5090 - sp->ts_tcharlen), | |
| 5091 mb_ptr2char(fword | |
| 5092 + sp->ts_fcharstart))) | |
| 5093 sp->ts_score -= | |
| 5094 SCORE_SUBST - SCORE_SIMILAR; | |
| 5095 } | |
| 5096 else if (sp->ts_isdiff == DIFF_INSERT | |
| 5097 && sp->ts_twordlen > sp->ts_tcharlen) | |
| 5098 { | |
| 5099 p = tword + sp->ts_twordlen - sp->ts_tcharlen; | |
| 5100 c = mb_ptr2char(p); | |
| 5101 if (enc_utf8 && utf_iscomposing(c)) | |
| 5102 { | |
| 5103 /* Inserting a composing char doesn't | |
| 5104 * count that much. */ | |
| 5105 sp->ts_score -= SCORE_INS - SCORE_INSCOMP; | |
| 5106 } | |
| 5107 else | |
| 5108 { | |
| 5109 /* If the previous character was the same, | |
| 5110 * thus doubling a character, give a bonus | |
| 5111 * to the score. Also for the soundfold | |
| 5112 * tree (might seem illogical but does | |
| 5113 * give better scores). */ | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5114 MB_PTR_BACK(tword, p); |
| 625 | 5115 if (c == mb_ptr2char(p)) |
| 5116 sp->ts_score -= SCORE_INS | |
| 5117 - SCORE_INSDUP; | |
| 5118 } | |
| 5119 } | |
| 5120 | |
| 5121 /* Starting a new char, reset the length. */ | |
| 5122 sp->ts_tcharlen = 0; | |
| 5123 } | |
| 358 | 5124 } |
| 330 | 5125 else |
| 5126 #endif | |
| 358 | 5127 { |
| 625 | 5128 /* If we found a similar char adjust the score. |
| 5129 * We do this after calling go_deeper() because | |
| 5130 * it's slow. */ | |
| 5131 if (newscore != 0 | |
| 5132 && !soundfold | |
| 5133 && slang->sl_has_map | |
| 5134 && similar_chars(slang, | |
| 5135 c, fword[sp->ts_fidx - 1])) | |
| 5136 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR; | |
| 358 | 5137 } |
| 625 | 5138 } |
| 5139 } | |
| 5140 break; | |
| 5141 | |
| 5142 case STATE_DEL: | |
| 5143 #ifdef FEAT_MBYTE | |
| 5144 /* When past the first byte of a multi-byte char don't try | |
| 5145 * delete/insert/swap a character. */ | |
| 5146 if (has_mbyte && sp->ts_tcharlen > 0) | |
| 5147 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5148 PROF_STORE(sp->ts_state) |
| 625 | 5149 sp->ts_state = STATE_FINAL; |
| 5150 break; | |
| 5151 } | |
| 5152 #endif | |
| 5153 /* | |
| 5154 * Try skipping one character in the bad word (delete it). | |
| 5155 */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5156 PROF_STORE(sp->ts_state) |
| 625 | 5157 sp->ts_state = STATE_INS_PREP; |
| 5158 sp->ts_curi = 1; | |
| 5159 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*') | |
| 5160 /* Deleting a vowel at the start of a word counts less, see | |
| 5161 * soundalike_score(). */ | |
| 5162 newscore = 2 * SCORE_DEL / 3; | |
| 5163 else | |
| 5164 newscore = SCORE_DEL; | |
| 5165 if (fword[sp->ts_fidx] != NUL | |
| 5166 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5167 { | |
| 5168 go_deeper(stack, depth, newscore); | |
| 5169 #ifdef DEBUG_TRIEWALK | |
| 5170 sprintf(changename[depth], "%.*s-%s: delete %c", | |
| 5171 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5172 fword[sp->ts_fidx]); | |
| 5173 #endif | |
| 5174 ++depth; | |
| 5175 | |
| 5176 /* Remember what character we deleted, so that we can avoid | |
| 5177 * inserting it again. */ | |
| 5178 stack[depth].ts_flags |= TSF_DIDDEL; | |
| 5179 stack[depth].ts_delidx = sp->ts_fidx; | |
| 5180 | |
| 5181 /* Advance over the character in fword[]. Give a bonus to the | |
| 5182 * score if the same character is following "nn" -> "n". It's | |
| 5183 * a bit illogical for soundfold tree but it does give better | |
| 5184 * results. */ | |
| 5185 #ifdef FEAT_MBYTE | |
| 5186 if (has_mbyte) | |
| 5187 { | |
| 5188 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
|
5189 stack[depth].ts_fidx += MB_PTR2LEN(fword + sp->ts_fidx); |
| 625 | 5190 if (enc_utf8 && utf_iscomposing(c)) |
| 5191 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP; | |
| 5192 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx)) | |
| 5193 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 5194 } |
| 5195 else | |
| 625 | 5196 #endif |
| 5197 { | |
| 5198 ++stack[depth].ts_fidx; | |
| 5199 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1]) | |
| 5200 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP; | |
| 323 | 5201 } |
| 5202 break; | |
| 625 | 5203 } |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5204 /* FALLTHROUGH */ |
| 625 | 5205 |
| 5206 case STATE_INS_PREP: | |
| 5207 if (sp->ts_flags & TSF_DIDDEL) | |
| 5208 { | |
| 5209 /* If we just deleted a byte then inserting won't make sense, | |
| 5210 * a substitute is always cheaper. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5211 PROF_STORE(sp->ts_state) |
| 625 | 5212 sp->ts_state = STATE_SWAP; |
| 5213 break; | |
| 5214 } | |
| 5215 | |
| 5216 /* skip over NUL bytes */ | |
| 5217 n = sp->ts_arridx; | |
| 5218 for (;;) | |
| 5219 { | |
| 5220 if (sp->ts_curi > byts[n]) | |
| 5221 { | |
| 5222 /* 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
|
5223 PROF_STORE(sp->ts_state) |
| 625 | 5224 sp->ts_state = STATE_SWAP; |
| 5225 break; | |
| 5226 } | |
| 5227 if (byts[n + sp->ts_curi] != NUL) | |
| 5228 { | |
| 5229 /* Found a byte to insert. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5230 PROF_STORE(sp->ts_state) |
| 625 | 5231 sp->ts_state = STATE_INS; |
| 330 | 5232 break; |
| 5233 } | |
| 625 | 5234 ++sp->ts_curi; |
| 5235 } | |
| 5236 break; | |
| 5237 | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5238 /* FALLTHROUGH */ |
| 625 | 5239 |
| 5240 case STATE_INS: | |
| 5241 /* Insert one byte. Repeat this for each possible byte at this | |
| 5242 * node. */ | |
| 5243 n = sp->ts_arridx; | |
| 5244 if (sp->ts_curi > byts[n]) | |
| 5245 { | |
| 5246 /* 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
|
5247 PROF_STORE(sp->ts_state) |
| 625 | 5248 sp->ts_state = STATE_SWAP; |
| 5249 break; | |
| 5250 } | |
| 5251 | |
| 5252 /* Do one more byte at this node, but: | |
| 5253 * - Skip NUL bytes. | |
| 5254 * - Skip the byte if it's equal to the byte in the word, | |
| 5255 * accepting that byte is always better. | |
| 5256 */ | |
| 5257 n += sp->ts_curi++; | |
| 5258 c = byts[n]; | |
| 5259 if (soundfold && sp->ts_twordlen == 0 && c == '*') | |
| 5260 /* Inserting a vowel at the start of a word counts less, | |
| 5261 * see soundalike_score(). */ | |
| 5262 newscore = 2 * SCORE_INS / 3; | |
| 5263 else | |
| 5264 newscore = SCORE_INS; | |
| 5265 if (c != fword[sp->ts_fidx] | |
| 5266 && TRY_DEEPER(su, stack, depth, newscore)) | |
| 5267 { | |
| 5268 go_deeper(stack, depth, newscore); | |
| 5269 #ifdef DEBUG_TRIEWALK | |
| 5270 sprintf(changename[depth], "%.*s-%s: insert %c", | |
| 5271 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5272 c); | |
| 5273 #endif | |
| 5274 ++depth; | |
| 5275 sp = &stack[depth]; | |
| 5276 tword[sp->ts_twordlen++] = c; | |
| 5277 sp->ts_arridx = idxs[n]; | |
| 330 | 5278 #ifdef FEAT_MBYTE |
| 5279 if (has_mbyte) | |
| 5280 { | |
| 625 | 5281 fl = MB_BYTE2LEN(c); |
| 5282 if (fl > 1) | |
| 330 | 5283 { |
| 625 | 5284 /* There are following bytes for the same character. |
| 5285 * We must find all bytes before trying | |
| 5286 * delete/insert/swap/etc. */ | |
| 5287 sp->ts_tcharlen = fl; | |
| 5288 sp->ts_tcharidx = 1; | |
| 5289 sp->ts_isdiff = DIFF_INSERT; | |
| 330 | 5290 } |
| 323 | 5291 } |
| 5292 else | |
| 625 | 5293 fl = 1; |
| 5294 if (fl == 1) | |
| 5295 #endif | |
| 5296 { | |
| 5297 /* If the previous character was the same, thus doubling a | |
| 5298 * character, give a bonus to the score. Also for | |
| 5299 * soundfold words (illogical but does give a better | |
| 5300 * score). */ | |
| 5301 if (sp->ts_twordlen >= 2 | |
| 5302 && tword[sp->ts_twordlen - 2] == c) | |
| 5303 sp->ts_score -= SCORE_INS - SCORE_INSDUP; | |
| 5304 } | |
| 5305 } | |
| 5306 break; | |
| 5307 | |
| 5308 case STATE_SWAP: | |
| 5309 /* | |
| 5310 * Swap two bytes in the bad word: "12" -> "21". | |
| 5311 * We change "fword" here, it's changed back afterwards at | |
| 5312 * STATE_UNSWAP. | |
| 5313 */ | |
| 5314 p = fword + sp->ts_fidx; | |
| 5315 c = *p; | |
| 5316 if (c == NUL) | |
| 5317 { | |
| 5318 /* 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
|
5319 PROF_STORE(sp->ts_state) |
| 625 | 5320 sp->ts_state = STATE_FINAL; |
| 5321 break; | |
| 5322 } | |
| 5323 | |
| 5324 /* Don't swap if the first character is not a word character. | |
| 5325 * 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
|
5326 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 5327 { |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5328 PROF_STORE(sp->ts_state) |
| 625 | 5329 sp->ts_state = STATE_REP_INI; |
| 323 | 5330 break; |
| 625 | 5331 } |
| 5332 | |
| 5333 #ifdef FEAT_MBYTE | |
| 5334 if (has_mbyte) | |
| 5335 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5336 n = MB_CPTR2LEN(p); |
| 625 | 5337 c = mb_ptr2char(p); |
| 1337 | 5338 if (p[n] == NUL) |
| 5339 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5340 else if (!soundfold && !spell_iswordp(p + n, curwin)) |
| 625 | 5341 c2 = c; /* don't swap non-word char */ |
| 5342 else | |
| 5343 c2 = mb_ptr2char(p + n); | |
| 5344 } | |
| 5345 else | |
| 5346 #endif | |
| 5347 { | |
| 1337 | 5348 if (p[1] == NUL) |
| 5349 c2 = NUL; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5350 else if (!soundfold && !spell_iswordp(p + 1, curwin)) |
| 625 | 5351 c2 = c; /* don't swap non-word char */ |
| 5352 else | |
| 5353 c2 = p[1]; | |
| 5354 } | |
| 5355 | |
| 1337 | 5356 /* When the second character is NUL we can't swap. */ |
| 5357 if (c2 == NUL) | |
| 5358 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5359 PROF_STORE(sp->ts_state) |
| 1337 | 5360 sp->ts_state = STATE_REP_INI; |
| 5361 break; | |
| 5362 } | |
| 5363 | |
| 625 | 5364 /* When characters are identical, swap won't do anything. |
| 5365 * Also get here if the second char is not a word character. */ | |
| 5366 if (c == c2) | |
| 5367 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5368 PROF_STORE(sp->ts_state) |
| 625 | 5369 sp->ts_state = STATE_SWAP3; |
| 5370 break; | |
| 5371 } | |
| 5372 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP)) | |
| 5373 { | |
| 5374 go_deeper(stack, depth, SCORE_SWAP); | |
| 5375 #ifdef DEBUG_TRIEWALK | |
| 5376 sprintf(changename[depth], "%.*s-%s: swap %c and %c", | |
| 5377 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5378 c, c2); | |
| 5379 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5380 PROF_STORE(sp->ts_state) |
| 625 | 5381 sp->ts_state = STATE_UNSWAP; |
| 5382 ++depth; | |
| 330 | 5383 #ifdef FEAT_MBYTE |
| 5384 if (has_mbyte) | |
| 5385 { | |
| 625 | 5386 fl = mb_char2len(c2); |
| 5387 mch_memmove(p, p + n, fl); | |
| 5388 mb_char2bytes(c, p + fl); | |
| 5389 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 5390 } |
| 5391 else | |
| 5392 #endif | |
| 5393 { | |
| 625 | 5394 p[0] = c2; |
| 330 | 5395 p[1] = c; |
| 625 | 5396 stack[depth].ts_fidxtry = sp->ts_fidx + 2; |
| 5397 } | |
| 5398 } | |
| 5399 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5400 { |
| 625 | 5401 /* 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
|
5402 PROF_STORE(sp->ts_state) |
| 625 | 5403 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
|
5404 } |
| 625 | 5405 break; |
| 5406 | |
| 5407 case STATE_UNSWAP: | |
| 5408 /* Undo the STATE_SWAP swap: "21" -> "12". */ | |
| 5409 p = fword + sp->ts_fidx; | |
| 5410 #ifdef FEAT_MBYTE | |
| 5411 if (has_mbyte) | |
| 5412 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5413 n = MB_PTR2LEN(p); |
| 625 | 5414 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
|
5415 mch_memmove(p + MB_PTR2LEN(p + n), p, n); |
| 625 | 5416 mb_char2bytes(c, p); |
| 5417 } | |
| 5418 else | |
| 5419 #endif | |
| 5420 { | |
| 5421 c = *p; | |
| 5422 *p = p[1]; | |
| 5423 p[1] = c; | |
| 5424 } | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5425 /* FALLTHROUGH */ |
| 625 | 5426 |
| 5427 case STATE_SWAP3: | |
| 5428 /* Swap two bytes, skipping one: "123" -> "321". We change | |
| 5429 * "fword" here, it's changed back afterwards at STATE_UNSWAP3. */ | |
| 5430 p = fword + sp->ts_fidx; | |
| 5431 #ifdef FEAT_MBYTE | |
| 5432 if (has_mbyte) | |
| 5433 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5434 n = MB_CPTR2LEN(p); |
| 625 | 5435 c = mb_ptr2char(p); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5436 fl = MB_CPTR2LEN(p + n); |
| 625 | 5437 c2 = mb_ptr2char(p + n); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5438 if (!soundfold && !spell_iswordp(p + n + fl, curwin)) |
| 625 | 5439 c3 = c; /* don't swap non-word char */ |
| 5440 else | |
| 5441 c3 = mb_ptr2char(p + n + fl); | |
| 5442 } | |
| 5443 else | |
| 5444 #endif | |
| 5445 { | |
| 5446 c = *p; | |
| 5447 c2 = p[1]; | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5448 if (!soundfold && !spell_iswordp(p + 2, curwin)) |
| 625 | 5449 c3 = c; /* don't swap non-word char */ |
| 5450 else | |
| 5451 c3 = p[2]; | |
| 5452 } | |
| 5453 | |
| 5454 /* When characters are identical: "121" then SWAP3 result is | |
| 5455 * identical, ROT3L result is same as SWAP: "211", ROT3L result is | |
| 5456 * same as SWAP on next char: "112". Thus skip all swapping. | |
| 5457 * Also skip when c3 is NUL. | |
| 5458 * Also get here when the third character is not a word character. | |
| 5459 * Second character may any char: "a.b" -> "b.a" */ | |
| 5460 if (c == c3 || c3 == NUL) | |
| 5461 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5462 PROF_STORE(sp->ts_state) |
| 625 | 5463 sp->ts_state = STATE_REP_INI; |
| 5464 break; | |
| 5465 } | |
| 5466 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5467 { | |
| 5468 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5469 #ifdef DEBUG_TRIEWALK | |
| 5470 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c", | |
| 5471 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5472 c, c3); | |
| 5473 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5474 PROF_STORE(sp->ts_state) |
| 625 | 5475 sp->ts_state = STATE_UNSWAP3; |
| 5476 ++depth; | |
| 5477 #ifdef FEAT_MBYTE | |
| 5478 if (has_mbyte) | |
| 5479 { | |
| 5480 tl = mb_char2len(c3); | |
| 5481 mch_memmove(p, p + n + fl, tl); | |
| 5482 mb_char2bytes(c2, p + tl); | |
| 5483 mb_char2bytes(c, p + fl + tl); | |
| 5484 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl; | |
| 5485 } | |
| 5486 else | |
| 5487 #endif | |
| 5488 { | |
| 5489 p[0] = p[2]; | |
| 5490 p[2] = c; | |
| 5491 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 5492 } | |
| 5493 } | |
| 5494 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5495 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5496 PROF_STORE(sp->ts_state) |
| 625 | 5497 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
|
5498 } |
| 625 | 5499 break; |
| 5500 | |
| 5501 case STATE_UNSWAP3: | |
| 5502 /* Undo STATE_SWAP3: "321" -> "123" */ | |
| 5503 p = fword + sp->ts_fidx; | |
| 5504 #ifdef FEAT_MBYTE | |
| 5505 if (has_mbyte) | |
| 5506 { | |
|
11394
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5507 n = MB_PTR2LEN(p); |
| 625 | 5508 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
|
5509 fl = MB_PTR2LEN(p + n); |
| 625 | 5510 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
|
5511 tl = MB_PTR2LEN(p + n + fl); |
| 625 | 5512 mch_memmove(p + fl + tl, p, n); |
| 5513 mb_char2bytes(c, p); | |
| 5514 mb_char2bytes(c2, p + tl); | |
| 5515 p = p + tl; | |
| 5516 } | |
| 5517 else | |
| 5518 #endif | |
| 5519 { | |
| 5520 c = *p; | |
| 5521 *p = p[2]; | |
| 5522 p[2] = c; | |
| 5523 ++p; | |
| 5524 } | |
| 5525 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
5526 if (!soundfold && !spell_iswordp(p, curwin)) |
| 625 | 5527 { |
| 5528 /* Middle char is not a word char, skip the rotate. First and | |
| 5529 * 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
|
5530 PROF_STORE(sp->ts_state) |
| 625 | 5531 sp->ts_state = STATE_REP_INI; |
| 5532 break; | |
| 5533 } | |
| 5534 | |
| 5535 /* Rotate three characters left: "123" -> "231". We change | |
| 5536 * "fword" here, it's changed back afterwards at STATE_UNROT3L. */ | |
| 5537 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5538 { | |
| 5539 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5540 #ifdef DEBUG_TRIEWALK | |
| 5541 p = fword + sp->ts_fidx; | |
| 5542 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c", | |
| 5543 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5544 p[0], p[1], p[2]); | |
| 5545 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5546 PROF_STORE(sp->ts_state) |
| 625 | 5547 sp->ts_state = STATE_UNROT3L; |
| 5548 ++depth; | |
| 330 | 5549 p = fword + sp->ts_fidx; |
| 5550 #ifdef FEAT_MBYTE | |
| 5551 if (has_mbyte) | |
| 5552 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5553 n = MB_CPTR2LEN(p); |
| 330 | 5554 c = mb_ptr2char(p); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5555 fl = MB_CPTR2LEN(p + n); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5556 fl += MB_CPTR2LEN(p + n + fl); |
| 625 | 5557 mch_memmove(p, p + n, fl); |
| 5558 mb_char2bytes(c, p + fl); | |
| 5559 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl; | |
| 330 | 5560 } |
| 5561 else | |
| 5562 #endif | |
| 323 | 5563 { |
| 330 | 5564 c = *p; |
| 625 | 5565 *p = p[1]; |
| 5566 p[1] = p[2]; | |
| 5567 p[2] = c; | |
| 5568 stack[depth].ts_fidxtry = sp->ts_fidx + 3; | |
| 5569 } | |
| 5570 } | |
| 5571 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5572 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5573 PROF_STORE(sp->ts_state) |
| 625 | 5574 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
|
5575 } |
| 625 | 5576 break; |
| 5577 | |
| 5578 case STATE_UNROT3L: | |
| 5579 /* Undo ROT3L: "231" -> "123" */ | |
| 5580 p = fword + sp->ts_fidx; | |
| 5581 #ifdef FEAT_MBYTE | |
| 5582 if (has_mbyte) | |
| 5583 { | |
|
11394
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); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5585 n += MB_PTR2LEN(p + n); |
| 625 | 5586 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
|
5587 tl = MB_PTR2LEN(p + n); |
| 625 | 5588 mch_memmove(p + tl, p, n); |
| 5589 mb_char2bytes(c, p); | |
| 5590 } | |
| 5591 else | |
| 5592 #endif | |
| 5593 { | |
| 5594 c = p[2]; | |
| 5595 p[2] = p[1]; | |
| 5596 p[1] = *p; | |
| 5597 *p = c; | |
| 5598 } | |
| 5599 | |
| 5600 /* Rotate three bytes right: "123" -> "312". We change "fword" | |
| 5601 * here, it's changed back afterwards at STATE_UNROT3R. */ | |
| 5602 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3)) | |
| 5603 { | |
| 5604 go_deeper(stack, depth, SCORE_SWAP3); | |
| 5605 #ifdef DEBUG_TRIEWALK | |
| 5606 p = fword + sp->ts_fidx; | |
| 5607 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c", | |
| 5608 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5609 p[0], p[1], p[2]); | |
| 5610 #endif | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5611 PROF_STORE(sp->ts_state) |
| 625 | 5612 sp->ts_state = STATE_UNROT3R; |
| 5613 ++depth; | |
| 330 | 5614 p = fword + sp->ts_fidx; |
| 5615 #ifdef FEAT_MBYTE | |
| 5616 if (has_mbyte) | |
| 5617 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5618 n = MB_CPTR2LEN(p); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5619 n += MB_CPTR2LEN(p + n); |
| 330 | 5620 c = mb_ptr2char(p + n); |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5621 tl = MB_CPTR2LEN(p + n); |
| 330 | 5622 mch_memmove(p + tl, p, n); |
| 5623 mb_char2bytes(c, p); | |
| 625 | 5624 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl; |
| 330 | 5625 } |
| 5626 else | |
| 5627 #endif | |
| 5628 { | |
| 5629 c = p[2]; | |
| 5630 p[2] = p[1]; | |
| 5631 p[1] = *p; | |
| 5632 *p = c; | |
| 625 | 5633 stack[depth].ts_fidxtry = sp->ts_fidx + 3; |
| 5634 } | |
| 5635 } | |
| 5636 else | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5637 { |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5638 PROF_STORE(sp->ts_state) |
| 625 | 5639 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
|
5640 } |
| 625 | 5641 break; |
| 5642 | |
| 5643 case STATE_UNROT3R: | |
| 5644 /* Undo ROT3R: "312" -> "123" */ | |
| 5645 p = fword + sp->ts_fidx; | |
| 5646 #ifdef FEAT_MBYTE | |
| 5647 if (has_mbyte) | |
| 5648 { | |
| 5649 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
|
5650 tl = MB_PTR2LEN(p); |
|
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); |
|
a2124e9bbb6a
patch 8.0.0582: illegal memory access with z= command
Christian Brabandt <cb@256bit.org>
parents:
11269
diff
changeset
|
5652 n += MB_PTR2LEN(p + tl + n); |
| 625 | 5653 mch_memmove(p, p + tl, n); |
| 5654 mb_char2bytes(c, p + n); | |
| 5655 } | |
| 5656 else | |
| 5657 #endif | |
| 5658 { | |
| 5659 c = *p; | |
| 5660 *p = p[1]; | |
| 5661 p[1] = p[2]; | |
| 5662 p[2] = c; | |
| 5663 } | |
|
12674
e769c912fcd9
patch 8.0.1215: newer gcc warns for implicit fallthrough
Christian Brabandt <cb@256bit.org>
parents:
12477
diff
changeset
|
5664 /* FALLTHROUGH */ |
| 625 | 5665 |
| 5666 case STATE_REP_INI: | |
| 5667 /* Check if matching with REP items from the .aff file would work. | |
| 5668 * Quickly skip if: | |
| 5669 * - there are no REP items and we are not in the soundfold trie | |
| 5670 * - the score is going to be too high anyway | |
| 5671 * - already applied a REP item or swapped here */ | |
| 5672 if ((lp->lp_replang == NULL && !soundfold) | |
| 5673 || sp->ts_score + SCORE_REP >= su->su_maxscore | |
| 5674 || sp->ts_fidx < sp->ts_fidxtry) | |
| 5675 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5676 PROF_STORE(sp->ts_state) |
| 625 | 5677 sp->ts_state = STATE_FINAL; |
| 323 | 5678 break; |
| 625 | 5679 } |
| 5680 | |
| 5681 /* Use the first byte to quickly find the first entry that may | |
| 5682 * match. If the index is -1 there is none. */ | |
| 5683 if (soundfold) | |
| 5684 sp->ts_curi = slang->sl_repsal_first[fword[sp->ts_fidx]]; | |
| 5685 else | |
| 5686 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]]; | |
| 5687 | |
| 5688 if (sp->ts_curi < 0) | |
| 5689 { | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5690 PROF_STORE(sp->ts_state) |
| 625 | 5691 sp->ts_state = STATE_FINAL; |
| 5692 break; | |
| 5693 } | |
| 5694 | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5695 PROF_STORE(sp->ts_state) |
| 625 | 5696 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
|
5697 /* FALLTHROUGH */ |
| 625 | 5698 |
| 5699 case STATE_REP: | |
| 5700 /* Try matching with REP items from the .aff file. For each match | |
| 5701 * replace the characters and check if the resulting word is | |
| 5702 * valid. */ | |
| 5703 p = fword + sp->ts_fidx; | |
| 5704 | |
| 5705 if (soundfold) | |
| 5706 gap = &slang->sl_repsal; | |
| 5707 else | |
| 503 | 5708 gap = &lp->lp_replang->sl_rep; |
| 625 | 5709 while (sp->ts_curi < gap->ga_len) |
| 5710 { | |
| 5711 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++; | |
| 5712 if (*ftp->ft_from != *p) | |
| 5713 { | |
| 5714 /* past possible matching entries */ | |
| 5715 sp->ts_curi = gap->ga_len; | |
| 323 | 5716 break; |
| 5717 } | |
| 625 | 5718 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0 |
| 5719 && TRY_DEEPER(su, stack, depth, SCORE_REP)) | |
| 5720 { | |
| 5721 go_deeper(stack, depth, SCORE_REP); | |
| 5722 #ifdef DEBUG_TRIEWALK | |
| 5723 sprintf(changename[depth], "%.*s-%s: replace %s with %s", | |
| 5724 sp->ts_twordlen, tword, fword + sp->ts_fidx, | |
| 5725 ftp->ft_from, ftp->ft_to); | |
| 5726 #endif | |
| 5727 /* Need to undo this afterwards. */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5728 PROF_STORE(sp->ts_state) |
| 625 | 5729 sp->ts_state = STATE_REP_UNDO; |
| 5730 | |
| 5731 /* Change the "from" to the "to" string. */ | |
| 5732 ++depth; | |
| 835 | 5733 fl = (int)STRLEN(ftp->ft_from); |
| 5734 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 5735 if (fl != tl) |
| 323 | 5736 { |
| 1619 | 5737 STRMOVE(p + tl, p + fl); |
| 625 | 5738 repextra += tl - fl; |
| 323 | 5739 } |
| 625 | 5740 mch_memmove(p, ftp->ft_to, tl); |
| 5741 stack[depth].ts_fidxtry = sp->ts_fidx + tl; | |
| 5742 #ifdef FEAT_MBYTE | |
| 5743 stack[depth].ts_tcharlen = 0; | |
| 5744 #endif | |
| 5745 break; | |
| 5746 } | |
| 5747 } | |
| 5748 | |
| 5749 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
|
5750 { |
| 625 | 5751 /* No (more) matches. */ |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5752 PROF_STORE(sp->ts_state) |
| 625 | 5753 sp->ts_state = STATE_FINAL; |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
5754 } |
| 625 | 5755 |
| 5756 break; | |
| 5757 | |
| 5758 case STATE_REP_UNDO: | |
| 5759 /* Undo a REP replacement and continue with the next one. */ | |
| 5760 if (soundfold) | |
| 5761 gap = &slang->sl_repsal; | |
| 5762 else | |
| 5763 gap = &lp->lp_replang->sl_rep; | |
| 5764 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1; | |
| 835 | 5765 fl = (int)STRLEN(ftp->ft_from); |
| 5766 tl = (int)STRLEN(ftp->ft_to); | |
| 625 | 5767 p = fword + sp->ts_fidx; |
| 5768 if (fl != tl) | |
| 5769 { | |
| 1619 | 5770 STRMOVE(p + fl, p + tl); |
| 625 | 5771 repextra -= tl - fl; |
| 5772 } | |
| 5773 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
|
5774 PROF_STORE(sp->ts_state) |
| 625 | 5775 sp->ts_state = STATE_REP; |
| 5776 break; | |
| 5777 | |
| 5778 default: | |
| 5779 /* Did all possible states at this level, go up one level. */ | |
| 5780 --depth; | |
| 5781 | |
| 5782 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE) | |
| 5783 { | |
| 5784 /* Continue in or go back to the prefix tree. */ | |
| 5785 byts = pbyts; | |
| 5786 idxs = pidxs; | |
| 5787 } | |
| 5788 | |
| 5789 /* Don't check for CTRL-C too often, it takes time. */ | |
| 5790 if (--breakcheckcount == 0) | |
| 5791 { | |
| 5792 ui_breakcheck(); | |
| 5793 breakcheckcount = 1000; | |
| 5794 } | |
| 5795 } | |
| 5796 } | |
| 5797 } | |
| 5798 | |
| 5799 | |
| 5800 /* | |
| 5801 * Go one level deeper in the tree. | |
| 5802 */ | |
| 5803 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5804 go_deeper(trystate_T *stack, int depth, int score_add) |
| 323 | 5805 { |
| 330 | 5806 stack[depth + 1] = stack[depth]; |
| 323 | 5807 stack[depth + 1].ts_state = STATE_START; |
| 625 | 5808 stack[depth + 1].ts_score = stack[depth].ts_score + score_add; |
| 323 | 5809 stack[depth + 1].ts_curi = 1; /* start just after length byte */ |
| 485 | 5810 stack[depth + 1].ts_flags = 0; |
| 323 | 5811 } |
| 5812 | |
| 455 | 5813 #ifdef FEAT_MBYTE |
| 5814 /* | |
| 5815 * Case-folding may change the number of bytes: Count nr of chars in | |
| 5816 * fword[flen] and return the byte length of that many chars in "word". | |
| 5817 */ | |
| 5818 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5819 nofold_len(char_u *fword, int flen, char_u *word) |
| 455 | 5820 { |
| 5821 char_u *p; | |
| 5822 int i = 0; | |
| 5823 | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5824 for (p = fword; p < fword + flen; MB_PTR_ADV(p)) |
| 455 | 5825 ++i; |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
5826 for (p = word; i > 0; MB_PTR_ADV(p)) |
| 455 | 5827 --i; |
| 5828 return (int)(p - word); | |
| 5829 } | |
| 5830 #endif | |
| 5831 | |
| 323 | 5832 /* |
| 5833 * "fword" is a good word with case folded. Find the matching keep-case | |
| 5834 * words and put it in "kword". | |
| 5835 * Theoretically there could be several keep-case words that result in the | |
| 5836 * same case-folded word, but we only find one... | |
| 5837 */ | |
| 5838 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5839 find_keepcap_word(slang_T *slang, char_u *fword, char_u *kword) |
| 323 | 5840 { |
| 5841 char_u uword[MAXWLEN]; /* "fword" in upper-case */ | |
| 5842 int depth; | |
| 324 | 5843 idx_T tryidx; |
| 323 | 5844 |
| 5845 /* The following arrays are used at each depth in the tree. */ | |
| 324 | 5846 idx_T arridx[MAXWLEN]; |
| 323 | 5847 int round[MAXWLEN]; |
| 5848 int fwordidx[MAXWLEN]; | |
| 5849 int uwordidx[MAXWLEN]; | |
| 5850 int kwordlen[MAXWLEN]; | |
| 5851 | |
| 5852 int flen, ulen; | |
| 5853 int l; | |
| 5854 int len; | |
| 5855 int c; | |
| 324 | 5856 idx_T lo, hi, m; |
| 323 | 5857 char_u *p; |
| 5858 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */ | |
| 324 | 5859 idx_T *idxs = slang->sl_kidxs; /* array with indexes */ |
| 323 | 5860 |
| 5861 if (byts == NULL) | |
| 5862 { | |
| 5863 /* array is empty: "cannot happen" */ | |
| 5864 *kword = NUL; | |
| 5865 return; | |
| 5866 } | |
| 5867 | |
| 5868 /* Make an all-cap version of "fword". */ | |
| 5869 allcap_copy(fword, uword); | |
| 5870 | |
| 5871 /* | |
| 5872 * Each character needs to be tried both case-folded and upper-case. | |
| 5873 * All this gets very complicated if we keep in mind that changing case | |
| 5874 * may change the byte length of a multi-byte character... | |
| 5875 */ | |
| 5876 depth = 0; | |
| 5877 arridx[0] = 0; | |
| 5878 round[0] = 0; | |
| 5879 fwordidx[0] = 0; | |
| 5880 uwordidx[0] = 0; | |
| 5881 kwordlen[0] = 0; | |
| 5882 while (depth >= 0) | |
| 5883 { | |
| 5884 if (fword[fwordidx[depth]] == NUL) | |
| 5885 { | |
| 5886 /* We are at the end of "fword". If the tree allows a word to end | |
| 5887 * here we have found a match. */ | |
| 5888 if (byts[arridx[depth] + 1] == 0) | |
| 5889 { | |
| 5890 kword[kwordlen[depth]] = NUL; | |
| 5891 return; | |
| 5892 } | |
| 5893 | |
| 5894 /* kword is getting too long, continue one level up */ | |
| 5895 --depth; | |
| 5896 } | |
| 5897 else if (++round[depth] > 2) | |
| 5898 { | |
| 5899 /* tried both fold-case and upper-case character, continue one | |
| 5900 * level up */ | |
| 5901 --depth; | |
| 5902 } | |
| 5903 else | |
| 5904 { | |
| 5905 /* | |
| 5906 * round[depth] == 1: Try using the folded-case character. | |
| 5907 * round[depth] == 2: Try using the upper-case character. | |
| 5908 */ | |
| 5909 #ifdef FEAT_MBYTE | |
| 5910 if (has_mbyte) | |
| 5911 { | |
|
9898
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5912 flen = MB_CPTR2LEN(fword + fwordidx[depth]); |
|
bff8a09016a5
commit https://github.com/vim/vim/commit/d3c907b5d2b352482b580a0cf687cbbea4c19ea1
Christian Brabandt <cb@256bit.org>
parents:
9649
diff
changeset
|
5913 ulen = MB_CPTR2LEN(uword + uwordidx[depth]); |
| 323 | 5914 } |
| 5915 else | |
| 5916 #endif | |
| 5917 ulen = flen = 1; | |
| 5918 if (round[depth] == 1) | |
| 5919 { | |
| 5920 p = fword + fwordidx[depth]; | |
| 5921 l = flen; | |
| 5922 } | |
| 5923 else | |
| 5924 { | |
| 5925 p = uword + uwordidx[depth]; | |
| 5926 l = ulen; | |
| 5927 } | |
| 5928 | |
| 5929 for (tryidx = arridx[depth]; l > 0; --l) | |
| 5930 { | |
| 5931 /* Perform a binary search in the list of accepted bytes. */ | |
| 5932 len = byts[tryidx++]; | |
| 5933 c = *p++; | |
| 5934 lo = tryidx; | |
| 5935 hi = tryidx + len - 1; | |
| 5936 while (lo < hi) | |
| 5937 { | |
| 5938 m = (lo + hi) / 2; | |
| 5939 if (byts[m] > c) | |
| 5940 hi = m - 1; | |
| 5941 else if (byts[m] < c) | |
| 5942 lo = m + 1; | |
| 5943 else | |
| 5944 { | |
| 5945 lo = hi = m; | |
| 5946 break; | |
| 5947 } | |
| 5948 } | |
| 5949 | |
| 5950 /* Stop if there is no matching byte. */ | |
| 5951 if (hi < lo || byts[lo] != c) | |
| 5952 break; | |
| 5953 | |
| 5954 /* Continue at the child (if there is one). */ | |
| 5955 tryidx = idxs[lo]; | |
| 5956 } | |
| 5957 | |
| 5958 if (l == 0) | |
| 5959 { | |
| 5960 /* | |
| 5961 * Found the matching char. Copy it to "kword" and go a | |
| 5962 * level deeper. | |
| 5963 */ | |
| 5964 if (round[depth] == 1) | |
| 5965 { | |
| 5966 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth], | |
| 5967 flen); | |
| 5968 kwordlen[depth + 1] = kwordlen[depth] + flen; | |
| 5969 } | |
| 5970 else | |
| 5971 { | |
| 5972 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth], | |
| 5973 ulen); | |
| 5974 kwordlen[depth + 1] = kwordlen[depth] + ulen; | |
| 5975 } | |
| 5976 fwordidx[depth + 1] = fwordidx[depth] + flen; | |
| 5977 uwordidx[depth + 1] = uwordidx[depth] + ulen; | |
| 5978 | |
| 5979 ++depth; | |
| 5980 arridx[depth] = tryidx; | |
| 5981 round[depth] = 0; | |
| 5982 } | |
| 5983 } | |
| 5984 } | |
| 5985 | |
| 5986 /* Didn't find it: "cannot happen". */ | |
| 5987 *kword = NUL; | |
| 5988 } | |
| 5989 | |
| 5990 /* | |
| 344 | 5991 * Compute the sound-a-like score for suggestions in su->su_ga and add them to |
| 5992 * su->su_sga. | |
| 5993 */ | |
| 5994 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
5995 score_comp_sal(suginfo_T *su) |
| 344 | 5996 { |
| 5997 langp_T *lp; | |
| 5998 char_u badsound[MAXWLEN]; | |
| 5999 int i; | |
| 6000 suggest_T *stp; | |
| 6001 suggest_T *sstp; | |
| 6002 int score; | |
| 500 | 6003 int lpi; |
| 344 | 6004 |
| 6005 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL) | |
| 6006 return; | |
| 6007 | |
| 6008 /* 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
|
6009 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
|
6010 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6011 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 6012 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6013 { | |
| 6014 /* soundfold the bad word */ | |
| 375 | 6015 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound); |
| 344 | 6016 |
| 6017 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 6018 { | |
| 6019 stp = &SUG(su->su_ga, i); | |
| 6020 | |
| 351 | 6021 /* Case-fold the suggested word, sound-fold it and compute the |
| 6022 * sound-a-like score. */ | |
| 6023 score = stp_sal_score(stp, su, lp->lp_slang, badsound); | |
| 344 | 6024 if (score < SCORE_MAXMAX) |
| 6025 { | |
| 6026 /* Add the suggestion. */ | |
| 6027 sstp = &SUG(su->su_sga, su->su_sga.ga_len); | |
| 6028 sstp->st_word = vim_strsave(stp->st_word); | |
| 6029 if (sstp->st_word != NULL) | |
| 6030 { | |
| 625 | 6031 sstp->st_wordlen = stp->st_wordlen; |
| 344 | 6032 sstp->st_score = score; |
| 6033 sstp->st_altscore = 0; | |
| 6034 sstp->st_orglen = stp->st_orglen; | |
| 6035 ++su->su_sga.ga_len; | |
| 6036 } | |
| 6037 } | |
| 6038 } | |
| 6039 break; | |
| 6040 } | |
| 500 | 6041 } |
| 344 | 6042 } |
| 6043 | |
| 6044 /* | |
| 6045 * Combine the list of suggestions in su->su_ga and su->su_sga. | |
| 4352 | 6046 * They are entwined. |
| 344 | 6047 */ |
| 6048 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6049 score_combine(suginfo_T *su) |
| 344 | 6050 { |
| 6051 int i; | |
| 6052 int j; | |
| 6053 garray_T ga; | |
| 6054 garray_T *gap; | |
| 6055 langp_T *lp; | |
| 6056 suggest_T *stp; | |
| 6057 char_u *p; | |
| 6058 char_u badsound[MAXWLEN]; | |
| 6059 int round; | |
| 500 | 6060 int lpi; |
| 625 | 6061 slang_T *slang = NULL; |
| 344 | 6062 |
| 6063 /* 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
|
6064 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
|
6065 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6066 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 344 | 6067 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6068 { | |
| 6069 /* soundfold the bad word */ | |
| 625 | 6070 slang = lp->lp_slang; |
| 6071 spell_soundfold(slang, su->su_fbadword, TRUE, badsound); | |
| 344 | 6072 |
| 6073 for (i = 0; i < su->su_ga.ga_len; ++i) | |
| 6074 { | |
| 6075 stp = &SUG(su->su_ga, i); | |
| 625 | 6076 stp->st_altscore = stp_sal_score(stp, su, slang, badsound); |
| 344 | 6077 if (stp->st_altscore == SCORE_MAXMAX) |
| 6078 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4; | |
| 6079 else | |
| 6080 stp->st_score = (stp->st_score * 3 | |
| 6081 + stp->st_altscore) / 4; | |
| 6082 stp->st_salscore = FALSE; | |
| 6083 } | |
| 6084 break; | |
| 6085 } | |
| 6086 } | |
| 6087 | |
| 844 | 6088 if (slang == NULL) /* Using "double" without sound folding. */ |
| 6089 { | |
| 6090 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, | |
| 6091 su->su_maxcount); | |
| 625 | 6092 return; |
| 844 | 6093 } |
| 625 | 6094 |
| 344 | 6095 /* Add the alternate score to su_sga. */ |
| 6096 for (i = 0; i < su->su_sga.ga_len; ++i) | |
| 6097 { | |
| 6098 stp = &SUG(su->su_sga, i); | |
| 625 | 6099 stp->st_altscore = spell_edit_score(slang, |
| 6100 su->su_badword, stp->st_word); | |
| 344 | 6101 if (stp->st_score == SCORE_MAXMAX) |
| 6102 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8; | |
| 6103 else | |
| 6104 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8; | |
| 6105 stp->st_salscore = TRUE; | |
| 6106 } | |
| 6107 | |
| 625 | 6108 /* Remove bad suggestions, sort the suggestions and truncate at "maxcount" |
| 6109 * for both lists. */ | |
| 6110 check_suggestions(su, &su->su_ga); | |
| 344 | 6111 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount); |
| 625 | 6112 check_suggestions(su, &su->su_sga); |
| 344 | 6113 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount); |
| 6114 | |
| 6115 ga_init2(&ga, (int)sizeof(suginfo_T), 1); | |
| 6116 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL) | |
| 6117 return; | |
| 6118 | |
| 6119 stp = &SUG(ga, 0); | |
| 6120 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i) | |
| 6121 { | |
| 6122 /* round 1: get a suggestion from su_ga | |
| 6123 * round 2: get a suggestion from su_sga */ | |
| 6124 for (round = 1; round <= 2; ++round) | |
| 6125 { | |
| 6126 gap = round == 1 ? &su->su_ga : &su->su_sga; | |
| 6127 if (i < gap->ga_len) | |
| 6128 { | |
| 6129 /* Don't add a word if it's already there. */ | |
| 6130 p = SUG(*gap, i).st_word; | |
| 6131 for (j = 0; j < ga.ga_len; ++j) | |
| 6132 if (STRCMP(stp[j].st_word, p) == 0) | |
| 6133 break; | |
| 6134 if (j == ga.ga_len) | |
| 6135 stp[ga.ga_len++] = SUG(*gap, i); | |
| 6136 else | |
| 6137 vim_free(p); | |
| 6138 } | |
| 6139 } | |
| 6140 } | |
| 6141 | |
| 6142 ga_clear(&su->su_ga); | |
| 6143 ga_clear(&su->su_sga); | |
| 6144 | |
| 6145 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 6146 if (ga.ga_len > su->su_maxcount) | |
| 6147 { | |
| 6148 for (i = su->su_maxcount; i < ga.ga_len; ++i) | |
| 6149 vim_free(stp[i].st_word); | |
| 6150 ga.ga_len = su->su_maxcount; | |
| 6151 } | |
| 6152 | |
| 6153 su->su_ga = ga; | |
| 6154 } | |
| 6155 | |
| 6156 /* | |
| 351 | 6157 * For the goodword in "stp" compute the soundalike score compared to the |
| 6158 * badword. | |
| 6159 */ | |
| 6160 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6161 stp_sal_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6162 suggest_T *stp, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6163 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6164 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6165 char_u *badsound) /* sound-folded badword */ |
| 351 | 6166 { |
| 6167 char_u *p; | |
| 534 | 6168 char_u *pbad; |
| 6169 char_u *pgood; | |
| 351 | 6170 char_u badsound2[MAXWLEN]; |
| 6171 char_u fword[MAXWLEN]; | |
| 6172 char_u goodsound[MAXWLEN]; | |
| 534 | 6173 char_u goodword[MAXWLEN]; |
| 6174 int lendiff; | |
| 6175 | |
| 6176 lendiff = (int)(su->su_badlen - stp->st_orglen); | |
| 6177 if (lendiff >= 0) | |
| 6178 pbad = badsound; | |
| 351 | 6179 else |
| 6180 { | |
| 6181 /* soundfold the bad word with more characters following */ | |
| 6182 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN); | |
| 6183 | |
| 6184 /* When joining two words the sound often changes a lot. E.g., "t he" | |
| 6185 * sounds like "t h" while "the" sounds like "@". Avoid that by | |
| 6186 * removing the space. Don't do it when the good word also contains a | |
| 6187 * space. */ | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
6188 if (VIM_ISWHITE(su->su_badptr[su->su_badlen]) |
| 351 | 6189 && *skiptowhite(stp->st_word) == NUL) |
| 6190 for (p = fword; *(p = skiptowhite(p)) != NUL; ) | |
| 1619 | 6191 STRMOVE(p, p + 1); |
| 351 | 6192 |
| 375 | 6193 spell_soundfold(slang, fword, TRUE, badsound2); |
| 534 | 6194 pbad = badsound2; |
| 6195 } | |
| 6196 | |
| 2768 | 6197 if (lendiff > 0 && stp->st_wordlen + lendiff < MAXWLEN) |
| 534 | 6198 { |
| 6199 /* Add part of the bad word to the good word, so that we soundfold | |
| 6200 * what replaces the bad word. */ | |
| 6201 STRCPY(goodword, stp->st_word); | |
| 625 | 6202 vim_strncpy(goodword + stp->st_wordlen, |
| 6203 su->su_badptr + su->su_badlen - lendiff, lendiff); | |
| 534 | 6204 pgood = goodword; |
| 6205 } | |
| 6206 else | |
| 6207 pgood = stp->st_word; | |
| 351 | 6208 |
| 375 | 6209 /* Sound-fold the word and compute the score for the difference. */ |
| 534 | 6210 spell_soundfold(slang, pgood, FALSE, goodsound); |
| 6211 | |
| 6212 return soundalike_score(goodsound, pbad); | |
| 351 | 6213 } |
| 6214 | |
| 625 | 6215 /* structure used to store soundfolded words that add_sound_suggest() has |
| 6216 * handled already. */ | |
| 6217 typedef struct | |
| 6218 { | |
| 6219 short sft_score; /* lowest score used */ | |
| 6220 char_u sft_word[1]; /* soundfolded word, actually longer */ | |
| 6221 } sftword_T; | |
| 6222 | |
| 6223 static sftword_T dumsft; | |
| 6224 #define HIKEY2SFT(p) ((sftword_T *)(p - (dumsft.sft_word - (char_u *)&dumsft))) | |
| 6225 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key) | |
| 6226 | |
| 6227 /* | |
| 6228 * Prepare for calling suggest_try_soundalike(). | |
| 6229 */ | |
| 6230 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6231 suggest_try_soundalike_prep(void) |
| 625 | 6232 { |
| 6233 langp_T *lp; | |
| 6234 int lpi; | |
| 6235 slang_T *slang; | |
| 6236 | |
| 6237 /* Do this for all languages that support sound folding and for which a | |
| 6238 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6239 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
|
6240 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6241 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 6242 slang = lp->lp_slang; |
| 6243 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 6244 /* prepare the hashtable used by add_sound_suggest() */ | |
| 6245 hash_init(&slang->sl_sounddone); | |
| 6246 } | |
| 6247 } | |
| 6248 | |
| 351 | 6249 /* |
| 323 | 6250 * Find suggestions by comparing the word in a sound-a-like form. |
| 503 | 6251 * Note: This doesn't support postponed prefixes. |
| 323 | 6252 */ |
| 6253 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6254 suggest_try_soundalike(suginfo_T *su) |
| 323 | 6255 { |
| 6256 char_u salword[MAXWLEN]; | |
| 6257 langp_T *lp; | |
| 500 | 6258 int lpi; |
| 503 | 6259 slang_T *slang; |
| 344 | 6260 |
| 625 | 6261 /* Do this for all languages that support sound folding and for which a |
| 6262 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6263 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
|
6264 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6265 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 625 | 6266 slang = lp->lp_slang; |
| 6267 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) | |
| 6268 { | |
| 6269 /* soundfold the bad word */ | |
| 6270 spell_soundfold(slang, su->su_fbadword, TRUE, salword); | |
| 6271 | |
| 6272 /* try all kinds of inserts/deletes/swaps/etc. */ | |
| 6273 /* TODO: also soundfold the next words, so that we can try joining | |
| 6274 * and splitting */ | |
|
7509
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6275 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6276 prof_init(); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6277 #endif |
| 625 | 6278 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
|
6279 #ifdef SUGGEST_PROFILE |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6280 prof_report("soundalike"); |
|
a64793340689
commit https://github.com/vim/vim/commit/ca1fe985175385c609f8e06672a1014729aba05c
Christian Brabandt <cb@256bit.org>
parents:
7447
diff
changeset
|
6281 #endif |
| 625 | 6282 } |
| 6283 } | |
| 6284 } | |
| 6285 | |
| 6286 /* | |
| 6287 * Finish up after calling suggest_try_soundalike(). | |
| 6288 */ | |
| 6289 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6290 suggest_try_soundalike_finish(void) |
| 625 | 6291 { |
| 6292 langp_T *lp; | |
| 6293 int lpi; | |
| 6294 slang_T *slang; | |
| 6295 int todo; | |
| 6296 hashitem_T *hi; | |
| 6297 | |
| 6298 /* Do this for all languages that support sound folding and for which a | |
| 6299 * .sug file has been loaded. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6300 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
|
6301 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6302 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 503 | 6303 slang = lp->lp_slang; |
| 625 | 6304 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL) |
| 6305 { | |
| 6306 /* Free the info about handled words. */ | |
| 835 | 6307 todo = (int)slang->sl_sounddone.ht_used; |
| 625 | 6308 for (hi = slang->sl_sounddone.ht_array; todo > 0; ++hi) |
| 6309 if (!HASHITEM_EMPTY(hi)) | |
| 6310 { | |
| 6311 vim_free(HI2SFT(hi)); | |
| 6312 --todo; | |
| 6313 } | |
| 1087 | 6314 |
| 6315 /* Clear the hashtable, it may also be used by another region. */ | |
| 625 | 6316 hash_clear(&slang->sl_sounddone); |
| 1087 | 6317 hash_init(&slang->sl_sounddone); |
| 625 | 6318 } |
| 6319 } | |
| 6320 } | |
| 6321 | |
| 6322 /* | |
| 6323 * A match with a soundfolded word is found. Add the good word(s) that | |
| 6324 * produce this soundfolded word. | |
| 6325 */ | |
| 6326 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6327 add_sound_suggest( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6328 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6329 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6330 int score, /* soundfold score */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6331 langp_T *lp) |
| 625 | 6332 { |
| 6333 slang_T *slang = lp->lp_slang; /* language for sound folding */ | |
| 6334 int sfwordnr; | |
| 6335 char_u *nrline; | |
| 6336 int orgnr; | |
| 6337 char_u theword[MAXWLEN]; | |
| 6338 int i; | |
| 6339 int wlen; | |
| 6340 char_u *byts; | |
| 6341 idx_T *idxs; | |
| 6342 int n; | |
| 6343 int wordcount; | |
| 6344 int wc; | |
| 6345 int goodscore; | |
| 6346 hash_T hash; | |
| 6347 hashitem_T *hi; | |
| 6348 sftword_T *sft; | |
| 6349 int bc, gc; | |
| 6350 int limit; | |
| 6351 | |
| 6352 /* | |
| 6353 * It's very well possible that the same soundfold word is found several | |
| 6354 * times with different scores. Since the following is quite slow only do | |
| 6355 * the words that have a better score than before. Use a hashtable to | |
| 6356 * remember the words that have been done. | |
| 6357 */ | |
| 6358 hash = hash_hash(goodword); | |
| 6359 hi = hash_lookup(&slang->sl_sounddone, goodword, hash); | |
| 6360 if (HASHITEM_EMPTY(hi)) | |
| 6361 { | |
| 844 | 6362 sft = (sftword_T *)alloc((unsigned)(sizeof(sftword_T) |
| 6363 + STRLEN(goodword))); | |
| 625 | 6364 if (sft != NULL) |
| 6365 { | |
| 6366 sft->sft_score = score; | |
| 6367 STRCPY(sft->sft_word, goodword); | |
| 6368 hash_add_item(&slang->sl_sounddone, hi, sft->sft_word, hash); | |
| 6369 } | |
| 6370 } | |
| 6371 else | |
| 6372 { | |
| 6373 sft = HI2SFT(hi); | |
| 6374 if (score >= sft->sft_score) | |
| 6375 return; | |
| 6376 sft->sft_score = score; | |
| 6377 } | |
| 6378 | |
| 6379 /* | |
| 6380 * Find the word nr in the soundfold tree. | |
| 6381 */ | |
| 6382 sfwordnr = soundfold_find(slang, goodword); | |
| 6383 if (sfwordnr < 0) | |
| 6384 { | |
|
10359
66f1b5bf3fa6
commit https://github.com/vim/vim/commit/95f096030ed1a8afea028f2ea295d6f6a70f466f
Christian Brabandt <cb@256bit.org>
parents:
10277
diff
changeset
|
6385 internal_error("add_sound_suggest()"); |
| 625 | 6386 return; |
| 6387 } | |
| 6388 | |
| 6389 /* | |
| 6390 * go over the list of good words that produce this soundfold word | |
| 6391 */ | |
| 6392 nrline = ml_get_buf(slang->sl_sugbuf, (linenr_T)(sfwordnr + 1), FALSE); | |
| 6393 orgnr = 0; | |
| 6394 while (*nrline != NUL) | |
| 6395 { | |
| 6396 /* The wordnr was stored in a minimal nr of bytes as an offset to the | |
| 6397 * previous wordnr. */ | |
| 6398 orgnr += bytes2offset(&nrline); | |
| 6399 | |
| 6400 byts = slang->sl_fbyts; | |
| 6401 idxs = slang->sl_fidxs; | |
| 6402 | |
| 6403 /* Lookup the word "orgnr" one of the two tries. */ | |
| 6404 n = 0; | |
| 6405 wordcount = 0; | |
| 5506 | 6406 for (wlen = 0; wlen < MAXWLEN - 3; ++wlen) |
| 625 | 6407 { |
| 6408 i = 1; | |
| 6409 if (wordcount == orgnr && byts[n + 1] == NUL) | |
| 6410 break; /* found end of word */ | |
| 6411 | |
| 6412 if (byts[n + 1] == NUL) | |
| 6413 ++wordcount; | |
| 6414 | |
| 6415 /* skip over the NUL bytes */ | |
| 6416 for ( ; byts[n + i] == NUL; ++i) | |
| 6417 if (i > byts[n]) /* safety check */ | |
| 6418 { | |
| 6419 STRCPY(theword + wlen, "BAD"); | |
| 5506 | 6420 wlen += 3; |
| 625 | 6421 goto badword; |
| 6422 } | |
| 6423 | |
| 6424 /* One of the siblings must have the word. */ | |
| 6425 for ( ; i < byts[n]; ++i) | |
| 6426 { | |
| 6427 wc = idxs[idxs[n + i]]; /* nr of words under this byte */ | |
| 6428 if (wordcount + wc > orgnr) | |
| 6429 break; | |
| 6430 wordcount += wc; | |
| 6431 } | |
| 6432 | |
| 5506 | 6433 theword[wlen] = byts[n + i]; |
| 625 | 6434 n = idxs[n + i]; |
| 6435 } | |
| 6436 badword: | |
| 6437 theword[wlen] = NUL; | |
| 6438 | |
| 6439 /* Go over the possible flags and regions. */ | |
| 6440 for (; i <= byts[n] && byts[n + i] == NUL; ++i) | |
| 6441 { | |
| 6442 char_u cword[MAXWLEN]; | |
| 6443 char_u *p; | |
| 6444 int flags = (int)idxs[n + i]; | |
| 6445 | |
| 698 | 6446 /* Skip words with the NOSUGGEST flag */ |
| 6447 if (flags & WF_NOSUGGEST) | |
| 6448 continue; | |
| 6449 | |
| 625 | 6450 if (flags & WF_KEEPCAP) |
| 6451 { | |
| 6452 /* Must find the word in the keep-case tree. */ | |
| 6453 find_keepcap_word(slang, theword, cword); | |
| 6454 p = cword; | |
| 6455 } | |
| 6456 else | |
| 6457 { | |
| 6458 flags |= su->su_badflags; | |
| 6459 if ((flags & WF_CAPMASK) != 0) | |
| 6460 { | |
| 6461 /* Need to fix case according to "flags". */ | |
| 6462 make_case_word(theword, cword, flags); | |
| 6463 p = cword; | |
| 323 | 6464 } |
| 6465 else | |
| 625 | 6466 p = theword; |
| 6467 } | |
| 6468 | |
| 6469 /* Add the suggestion. */ | |
| 6470 if (sps_flags & SPS_DOUBLE) | |
| 6471 { | |
| 6472 /* Add the suggestion if the score isn't too bad. */ | |
| 6473 if (score <= su->su_maxscore) | |
| 6474 add_suggestion(su, &su->su_sga, p, su->su_badlen, | |
| 6475 score, 0, FALSE, slang, FALSE); | |
| 6476 } | |
| 6477 else | |
| 6478 { | |
| 6479 /* Add a penalty for words in another region. */ | |
| 6480 if ((flags & WF_REGION) | |
| 6481 && (((unsigned)flags >> 16) & lp->lp_region) == 0) | |
| 6482 goodscore = SCORE_REGION; | |
| 6483 else | |
| 6484 goodscore = 0; | |
| 6485 | |
| 6486 /* Add a small penalty for changing the first letter from | |
| 6487 * lower to upper case. Helps for "tath" -> "Kath", which is | |
| 4352 | 6488 * less common than "tath" -> "path". Don't do it when the |
| 625 | 6489 * letter is the same, that has already been counted. */ |
| 6490 gc = PTR2CHAR(p); | |
| 6491 if (SPELL_ISUPPER(gc)) | |
| 6492 { | |
| 6493 bc = PTR2CHAR(su->su_badword); | |
| 6494 if (!SPELL_ISUPPER(bc) | |
| 6495 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc)) | |
| 6496 goodscore += SCORE_ICASE / 2; | |
| 6497 } | |
| 6498 | |
| 6499 /* Compute the score for the good word. This only does letter | |
| 6500 * insert/delete/swap/replace. REP items are not considered, | |
| 6501 * which may make the score a bit higher. | |
| 6502 * Use a limit for the score to make it work faster. Use | |
| 6503 * MAXSCORE(), because RESCORE() will change the score. | |
| 6504 * If the limit is very high then the iterative method is | |
| 6505 * inefficient, using an array is quicker. */ | |
| 6506 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score); | |
| 6507 if (limit > SCORE_LIMITMAX) | |
| 6508 goodscore += spell_edit_score(slang, su->su_badword, p); | |
| 6509 else | |
| 6510 goodscore += spell_edit_score_limit(slang, su->su_badword, | |
| 6511 p, limit); | |
| 6512 | |
| 6513 /* When going over the limit don't bother to do the rest. */ | |
| 6514 if (goodscore < SCORE_MAXMAX) | |
| 6515 { | |
| 6516 /* Give a bonus to words seen before. */ | |
| 6517 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE); | |
| 6518 | |
| 6519 /* Add the suggestion if the score isn't too bad. */ | |
| 6520 goodscore = RESCORE(goodscore, score); | |
| 6521 if (goodscore <= su->su_sfmaxscore) | |
| 6522 add_suggestion(su, &su->su_ga, p, su->su_badlen, | |
| 6523 goodscore, score, TRUE, slang, TRUE); | |
| 6524 } | |
| 6525 } | |
| 6526 } | |
| 6527 /* smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr); */ | |
| 6528 } | |
| 6529 } | |
| 6530 | |
| 6531 /* | |
| 6532 * Find word "word" in fold-case tree for "slang" and return the word number. | |
| 6533 */ | |
| 6534 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6535 soundfold_find(slang_T *slang, char_u *word) |
| 625 | 6536 { |
| 6537 idx_T arridx = 0; | |
| 6538 int len; | |
| 6539 int wlen = 0; | |
| 6540 int c; | |
| 6541 char_u *ptr = word; | |
| 6542 char_u *byts; | |
| 6543 idx_T *idxs; | |
| 6544 int wordnr = 0; | |
| 6545 | |
| 6546 byts = slang->sl_sbyts; | |
| 6547 idxs = slang->sl_sidxs; | |
| 6548 | |
| 6549 for (;;) | |
| 6550 { | |
| 6551 /* First byte is the number of possible bytes. */ | |
| 6552 len = byts[arridx++]; | |
| 6553 | |
| 6554 /* If the first possible byte is a zero the word could end here. | |
| 6555 * If the word ends we found the word. If not skip the NUL bytes. */ | |
| 6556 c = ptr[wlen]; | |
| 6557 if (byts[arridx] == NUL) | |
| 6558 { | |
| 6559 if (c == NUL) | |
| 6560 break; | |
| 6561 | |
| 6562 /* Skip over the zeros, there can be several. */ | |
| 6563 while (len > 0 && byts[arridx] == NUL) | |
| 6564 { | |
| 6565 ++arridx; | |
| 6566 --len; | |
| 6567 } | |
| 6568 if (len == 0) | |
| 6569 return -1; /* no children, word should have ended here */ | |
| 6570 ++wordnr; | |
| 6571 } | |
| 6572 | |
| 6573 /* If the word ends we didn't find it. */ | |
| 6574 if (c == NUL) | |
| 6575 return -1; | |
| 6576 | |
| 6577 /* Perform a binary search in the list of accepted bytes. */ | |
| 6578 if (c == TAB) /* <Tab> is handled like <Space> */ | |
| 6579 c = ' '; | |
| 6580 while (byts[arridx] < c) | |
| 6581 { | |
| 6582 /* The word count is in the first idxs[] entry of the child. */ | |
| 6583 wordnr += idxs[idxs[arridx]]; | |
| 6584 ++arridx; | |
| 6585 if (--len == 0) /* end of the bytes, didn't find it */ | |
| 6586 return -1; | |
| 6587 } | |
| 6588 if (byts[arridx] != c) /* didn't find the byte */ | |
| 6589 return -1; | |
| 6590 | |
| 6591 /* Continue at the child (if there is one). */ | |
| 6592 arridx = idxs[arridx]; | |
| 6593 ++wlen; | |
| 6594 | |
| 6595 /* One space in the good word may stand for several spaces in the | |
| 6596 * checked word. */ | |
| 6597 if (c == ' ') | |
| 6598 while (ptr[wlen] == ' ' || ptr[wlen] == TAB) | |
| 6599 ++wlen; | |
| 6600 } | |
| 6601 | |
| 6602 return wordnr; | |
| 323 | 6603 } |
| 6604 | |
| 6605 /* | |
| 324 | 6606 * Copy "fword" to "cword", fixing case according to "flags". |
| 323 | 6607 */ |
| 6608 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6609 make_case_word(char_u *fword, char_u *cword, int flags) |
| 323 | 6610 { |
| 6611 if (flags & WF_ALLCAP) | |
| 6612 /* Make it all upper-case */ | |
| 6613 allcap_copy(fword, cword); | |
| 6614 else if (flags & WF_ONECAP) | |
| 6615 /* Make the first letter upper-case */ | |
| 324 | 6616 onecap_copy(fword, cword, TRUE); |
| 323 | 6617 else |
| 6618 /* Use goodword as-is. */ | |
| 6619 STRCPY(cword, fword); | |
| 6620 } | |
| 6621 | |
| 330 | 6622 |
| 323 | 6623 /* |
| 6624 * Return TRUE if "c1" and "c2" are similar characters according to the MAP | |
| 6625 * lines in the .aff file. | |
| 6626 */ | |
| 6627 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6628 similar_chars(slang_T *slang, int c1, int c2) |
| 323 | 6629 { |
| 330 | 6630 int m1, m2; |
| 6631 #ifdef FEAT_MBYTE | |
| 3549 | 6632 char_u buf[MB_MAXBYTES + 1]; |
| 330 | 6633 hashitem_T *hi; |
| 6634 | |
| 6635 if (c1 >= 256) | |
| 6636 { | |
| 6637 buf[mb_char2bytes(c1, buf)] = 0; | |
| 6638 hi = hash_find(&slang->sl_map_hash, buf); | |
| 6639 if (HASHITEM_EMPTY(hi)) | |
| 6640 m1 = 0; | |
| 6641 else | |
| 6642 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 6643 } | |
| 6644 else | |
| 6645 #endif | |
| 6646 m1 = slang->sl_map_array[c1]; | |
| 6647 if (m1 == 0) | |
| 323 | 6648 return FALSE; |
| 330 | 6649 |
| 6650 | |
| 6651 #ifdef FEAT_MBYTE | |
| 6652 if (c2 >= 256) | |
| 6653 { | |
| 6654 buf[mb_char2bytes(c2, buf)] = 0; | |
| 6655 hi = hash_find(&slang->sl_map_hash, buf); | |
| 6656 if (HASHITEM_EMPTY(hi)) | |
| 6657 m2 = 0; | |
| 6658 else | |
| 6659 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1); | |
| 6660 } | |
| 6661 else | |
| 6662 #endif | |
| 6663 m2 = slang->sl_map_array[c2]; | |
| 6664 | |
| 6665 return m1 == m2; | |
| 323 | 6666 } |
| 6667 | |
| 6668 /* | |
| 6669 * Add a suggestion to the list of suggestions. | |
| 625 | 6670 * For a suggestion that is already in the list the lowest score is remembered. |
| 6671 */ | |
| 6672 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6673 add_suggestion( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6674 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6675 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
|
6676 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6677 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
|
6678 int score, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6679 int altscore, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6680 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
|
6681 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
|
6682 int maxsf) /* su_maxscore applies to soundfold score, |
| 625 | 6683 su_sfmaxscore to the total score. */ |
| 6684 { | |
| 6685 int goodlen; /* len of goodword changed */ | |
| 6686 int badlen; /* len of bad word changed */ | |
| 323 | 6687 suggest_T *stp; |
| 534 | 6688 suggest_T new_sug; |
| 323 | 6689 int i; |
| 534 | 6690 char_u *pgood, *pbad; |
| 532 | 6691 |
| 534 | 6692 /* Minimize "badlen" for consistency. Avoids that changing "the the" to |
| 6693 * "thee the" is added next to changing the first "the" the "thee". */ | |
| 6694 pgood = goodword + STRLEN(goodword); | |
| 625 | 6695 pbad = su->su_badptr + badlenarg; |
| 6696 for (;;) | |
| 6697 { | |
| 835 | 6698 goodlen = (int)(pgood - goodword); |
| 6699 badlen = (int)(pbad - su->su_badptr); | |
| 625 | 6700 if (goodlen <= 0 || badlen <= 0) |
| 6701 break; | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
6702 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
|
6703 MB_PTR_BACK(su->su_badptr, pbad); |
| 534 | 6704 #ifdef FEAT_MBYTE |
| 6705 if (has_mbyte) | |
| 6706 { | |
| 6707 if (mb_ptr2char(pgood) != mb_ptr2char(pbad)) | |
| 6708 break; | |
| 346 | 6709 } |
| 6710 else | |
| 534 | 6711 #endif |
| 6712 if (*pgood != *pbad) | |
| 6713 break; | |
| 625 | 6714 } |
| 6715 | |
| 534 | 6716 if (badlen == 0 && goodlen == 0) |
| 6717 /* goodword doesn't change anything; may happen for "the the" changing | |
| 6718 * the first "the" to itself. */ | |
| 6719 return; | |
| 346 | 6720 |
| 944 | 6721 if (gap->ga_len == 0) |
| 6722 i = -1; | |
| 6723 else | |
| 6724 { | |
| 6725 /* Check if the word is already there. Also check the length that is | |
| 6726 * being replaced "thes," -> "these" is a different suggestion from | |
| 6727 * "thes" -> "these". */ | |
| 6728 stp = &SUG(*gap, 0); | |
| 6729 for (i = gap->ga_len; --i >= 0; ++stp) | |
| 6730 if (stp->st_wordlen == goodlen | |
| 6731 && stp->st_orglen == badlen | |
| 6732 && STRNCMP(stp->st_word, goodword, goodlen) == 0) | |
| 6733 { | |
| 6734 /* | |
| 6735 * Found it. Remember the word with the lowest score. | |
| 6736 */ | |
| 6737 if (stp->st_slang == NULL) | |
| 6738 stp->st_slang = slang; | |
| 6739 | |
| 6740 new_sug.st_score = score; | |
| 6741 new_sug.st_altscore = altscore; | |
| 6742 new_sug.st_had_bonus = had_bonus; | |
| 6743 | |
| 6744 if (stp->st_had_bonus != had_bonus) | |
| 6745 { | |
| 6746 /* Only one of the two had the soundalike score computed. | |
| 6747 * Need to do that for the other one now, otherwise the | |
| 6748 * scores can't be compared. This happens because | |
| 6749 * suggest_try_change() doesn't compute the soundalike | |
| 6750 * word to keep it fast, while some special methods set | |
| 6751 * the soundalike score to zero. */ | |
| 6752 if (had_bonus) | |
| 6753 rescore_one(su, stp); | |
| 6754 else | |
| 6755 { | |
| 6756 new_sug.st_word = stp->st_word; | |
| 6757 new_sug.st_wordlen = stp->st_wordlen; | |
| 6758 new_sug.st_slang = stp->st_slang; | |
| 6759 new_sug.st_orglen = badlen; | |
| 6760 rescore_one(su, &new_sug); | |
| 6761 } | |
| 6762 } | |
| 6763 | |
| 6764 if (stp->st_score > new_sug.st_score) | |
| 6765 { | |
| 6766 stp->st_score = new_sug.st_score; | |
| 6767 stp->st_altscore = new_sug.st_altscore; | |
| 6768 stp->st_had_bonus = new_sug.st_had_bonus; | |
| 6769 } | |
| 6770 break; | |
| 6771 } | |
| 6772 } | |
| 625 | 6773 |
| 6774 if (i < 0 && ga_grow(gap, 1) == OK) | |
| 6775 { | |
| 6776 /* Add a suggestion. */ | |
| 6777 stp = &SUG(*gap, gap->ga_len); | |
| 6778 stp->st_word = vim_strnsave(goodword, goodlen); | |
| 6779 if (stp->st_word != NULL) | |
| 6780 { | |
| 6781 stp->st_wordlen = goodlen; | |
| 6782 stp->st_score = score; | |
| 6783 stp->st_altscore = altscore; | |
| 6784 stp->st_had_bonus = had_bonus; | |
| 6785 stp->st_orglen = badlen; | |
| 6786 stp->st_slang = slang; | |
| 6787 ++gap->ga_len; | |
| 6788 | |
| 6789 /* If we have too many suggestions now, sort the list and keep | |
| 6790 * the best suggestions. */ | |
| 6791 if (gap->ga_len > SUG_MAX_COUNT(su)) | |
| 6792 { | |
| 6793 if (maxsf) | |
| 6794 su->su_sfmaxscore = cleanup_suggestions(gap, | |
| 6795 su->su_sfmaxscore, SUG_CLEAN_COUNT(su)); | |
| 6796 else | |
| 6797 su->su_maxscore = cleanup_suggestions(gap, | |
| 6798 su->su_maxscore, SUG_CLEAN_COUNT(su)); | |
| 6799 } | |
| 6800 } | |
| 6801 } | |
| 6802 } | |
| 6803 | |
| 6804 /* | |
| 6805 * Suggestions may in fact be flagged as errors. Esp. for banned words and | |
| 6806 * for split words, such as "the the". Remove these from the list here. | |
| 6807 */ | |
| 6808 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6809 check_suggestions( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6810 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6811 garray_T *gap) /* either su_ga or su_sga */ |
| 625 | 6812 { |
| 6813 suggest_T *stp; | |
| 6814 int i; | |
| 6815 char_u longword[MAXWLEN + 1]; | |
| 6816 int len; | |
| 6817 hlf_T attr; | |
| 6818 | |
| 6819 stp = &SUG(*gap, 0); | |
| 6820 for (i = gap->ga_len - 1; i >= 0; --i) | |
| 6821 { | |
| 6822 /* Need to append what follows to check for "the the". */ | |
| 2768 | 6823 vim_strncpy(longword, stp[i].st_word, MAXWLEN); |
| 625 | 6824 len = stp[i].st_wordlen; |
| 6825 vim_strncpy(longword + len, su->su_badptr + stp[i].st_orglen, | |
| 6826 MAXWLEN - len); | |
| 6827 attr = HLF_COUNT; | |
| 6828 (void)spell_check(curwin, longword, &attr, NULL, FALSE); | |
| 6829 if (attr != HLF_COUNT) | |
| 6830 { | |
| 6831 /* Remove this entry. */ | |
| 6832 vim_free(stp[i].st_word); | |
| 6833 --gap->ga_len; | |
| 6834 if (i < gap->ga_len) | |
| 6835 mch_memmove(stp + i, stp + i + 1, | |
| 6836 sizeof(suggest_T) * (gap->ga_len - i)); | |
| 6837 } | |
| 6838 } | |
| 6839 } | |
| 6840 | |
| 323 | 6841 |
| 6842 /* | |
| 6843 * Add a word to be banned. | |
| 6844 */ | |
| 6845 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6846 add_banned( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6847 suginfo_T *su, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6848 char_u *word) |
| 323 | 6849 { |
| 626 | 6850 char_u *s; |
| 323 | 6851 hash_T hash; |
| 6852 hashitem_T *hi; | |
| 6853 | |
| 625 | 6854 hash = hash_hash(word); |
| 6855 hi = hash_lookup(&su->su_banned, word, hash); | |
| 6856 if (HASHITEM_EMPTY(hi)) | |
| 6857 { | |
| 6858 s = vim_strsave(word); | |
| 6859 if (s != NULL) | |
| 323 | 6860 hash_add_item(&su->su_banned, hi, s, hash); |
| 625 | 6861 } |
| 323 | 6862 } |
| 6863 | |
| 324 | 6864 /* |
| 534 | 6865 * Recompute the score for all suggestions if sound-folding is possible. This |
| 6866 * is slow, thus only done for the final results. | |
| 324 | 6867 */ |
| 6868 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6869 rescore_suggestions(suginfo_T *su) |
| 324 | 6870 { |
| 6871 int i; | |
| 534 | 6872 |
| 6873 if (su->su_sallang != NULL) | |
| 503 | 6874 for (i = 0; i < su->su_ga.ga_len; ++i) |
| 534 | 6875 rescore_one(su, &SUG(su->su_ga, i)); |
| 6876 } | |
| 6877 | |
| 6878 /* | |
| 6879 * Recompute the score for one suggestion if sound-folding is possible. | |
| 6880 */ | |
| 6881 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6882 rescore_one(suginfo_T *su, suggest_T *stp) |
| 534 | 6883 { |
| 6884 slang_T *slang = stp->st_slang; | |
| 6885 char_u sal_badword[MAXWLEN]; | |
| 535 | 6886 char_u *p; |
| 534 | 6887 |
| 6888 /* Only rescore suggestions that have no sal score yet and do have a | |
| 6889 * language. */ | |
| 6890 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus) | |
| 6891 { | |
| 6892 if (slang == su->su_sallang) | |
| 535 | 6893 p = su->su_sal_badword; |
| 534 | 6894 else |
| 6895 { | |
| 6896 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword); | |
| 535 | 6897 p = sal_badword; |
| 6898 } | |
| 6899 | |
| 6900 stp->st_altscore = stp_sal_score(stp, su, slang, p); | |
| 534 | 6901 if (stp->st_altscore == SCORE_MAXMAX) |
| 6902 stp->st_altscore = SCORE_BIG; | |
| 6903 stp->st_score = RESCORE(stp->st_score, stp->st_altscore); | |
| 6904 stp->st_had_bonus = TRUE; | |
| 324 | 6905 } |
| 6906 } | |
| 6907 | |
| 323 | 6908 static int |
| 6909 #ifdef __BORLANDC__ | |
| 6910 _RTLENTRYF | |
| 6911 #endif | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7526
diff
changeset
|
6912 sug_compare(const void *s1, const void *s2); |
| 323 | 6913 |
| 6914 /* | |
| 6915 * Function given to qsort() to sort the suggestions on st_score. | |
| 529 | 6916 * First on "st_score", then "st_altscore" then alphabetically. |
| 323 | 6917 */ |
| 6918 static int | |
| 6919 #ifdef __BORLANDC__ | |
| 6920 _RTLENTRYF | |
| 6921 #endif | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6922 sug_compare(const void *s1, const void *s2) |
| 323 | 6923 { |
| 6924 suggest_T *p1 = (suggest_T *)s1; | |
| 6925 suggest_T *p2 = (suggest_T *)s2; | |
| 344 | 6926 int n = p1->st_score - p2->st_score; |
| 6927 | |
| 6928 if (n == 0) | |
| 529 | 6929 { |
| 6930 n = p1->st_altscore - p2->st_altscore; | |
| 6931 if (n == 0) | |
| 6932 n = STRICMP(p1->st_word, p2->st_word); | |
| 6933 } | |
| 344 | 6934 return n; |
| 323 | 6935 } |
| 6936 | |
| 6937 /* | |
| 6938 * Cleanup the suggestions: | |
| 6939 * - Sort on score. | |
| 6940 * - Remove words that won't be displayed. | |
| 344 | 6941 * Returns the maximum score in the list or "maxscore" unmodified. |
| 323 | 6942 */ |
| 344 | 6943 static int |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6944 cleanup_suggestions( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6945 garray_T *gap, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6946 int maxscore, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6947 int keep) /* nr of suggestions to keep */ |
| 323 | 6948 { |
| 344 | 6949 suggest_T *stp = &SUG(*gap, 0); |
| 323 | 6950 int i; |
| 6951 | |
| 6952 /* Sort the list. */ | |
| 344 | 6953 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare); |
| 323 | 6954 |
| 6955 /* Truncate the list to the number of suggestions that will be displayed. */ | |
| 344 | 6956 if (gap->ga_len > keep) |
| 323 | 6957 { |
| 344 | 6958 for (i = keep; i < gap->ga_len; ++i) |
| 323 | 6959 vim_free(stp[i].st_word); |
| 344 | 6960 gap->ga_len = keep; |
| 6961 return stp[keep - 1].st_score; | |
| 323 | 6962 } |
| 344 | 6963 return maxscore; |
| 323 | 6964 } |
| 6965 | |
| 372 | 6966 #if defined(FEAT_EVAL) || defined(PROTO) |
| 6967 /* | |
| 6968 * Soundfold a string, for soundfold(). | |
| 6969 * Result is in allocated memory, NULL for an error. | |
| 6970 */ | |
| 6971 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
6972 eval_soundfold(char_u *word) |
| 372 | 6973 { |
| 6974 langp_T *lp; | |
| 6975 char_u sound[MAXWLEN]; | |
| 500 | 6976 int lpi; |
| 372 | 6977 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6978 if (curwin->w_p_spell && *curwin->w_s->b_p_spl != NUL) |
| 372 | 6979 /* 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
|
6980 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
|
6981 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
6982 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 372 | 6983 if (lp->lp_slang->sl_sal.ga_len > 0) |
| 6984 { | |
| 6985 /* soundfold the word */ | |
| 375 | 6986 spell_soundfold(lp->lp_slang, word, FALSE, sound); |
| 372 | 6987 return vim_strsave(sound); |
| 6988 } | |
| 500 | 6989 } |
| 372 | 6990 |
| 6991 /* No language with sound folding, return word as-is. */ | |
| 6992 return vim_strsave(word); | |
| 6993 } | |
| 6994 #endif | |
| 6995 | |
| 323 | 6996 /* |
| 6997 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 485 | 6998 * |
| 6999 * There are many ways to turn a word into a sound-a-like representation. The | |
| 7000 * oldest is Soundex (1918!). A nice overview can be found in "Approximate | |
| 7001 * swedish name matching - survey and test of different algorithms" by Klas | |
| 7002 * Erikson. | |
| 7003 * | |
| 7004 * We support two methods: | |
| 7005 * 1. SOFOFROM/SOFOTO do a simple character mapping. | |
| 7006 * 2. SAL items define a more advanced sound-folding (and much slower). | |
| 323 | 7007 */ |
|
9583
b0c7061d6439
commit https://github.com/vim/vim/commit/9ccfebddc3ff2a3c2853cf706fd4c26f639bf381
Christian Brabandt <cb@256bit.org>
parents:
9570
diff
changeset
|
7008 void |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7009 spell_soundfold( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7010 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7011 char_u *inword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7012 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
|
7013 char_u *res) |
| 375 | 7014 { |
| 7015 char_u fword[MAXWLEN]; | |
| 7016 char_u *word; | |
| 7017 | |
| 7018 if (slang->sl_sofo) | |
| 7019 /* SOFOFROM and SOFOTO used */ | |
| 7020 spell_soundfold_sofo(slang, inword, res); | |
| 7021 else | |
| 7022 { | |
| 7023 /* SAL items used. Requires the word to be case-folded. */ | |
| 7024 if (folded) | |
| 7025 word = inword; | |
| 7026 else | |
| 7027 { | |
| 835 | 7028 (void)spell_casefold(inword, (int)STRLEN(inword), fword, MAXWLEN); |
| 375 | 7029 word = fword; |
| 7030 } | |
| 7031 | |
| 7032 #ifdef FEAT_MBYTE | |
| 7033 if (has_mbyte) | |
| 7034 spell_soundfold_wsal(slang, word, res); | |
| 7035 else | |
| 7036 #endif | |
| 7037 spell_soundfold_sal(slang, word, res); | |
| 7038 } | |
| 7039 } | |
| 7040 | |
| 7041 /* | |
| 7042 * Perform sound folding of "inword" into "res" according to SOFOFROM and | |
| 7043 * SOFOTO lines. | |
| 7044 */ | |
| 7045 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7046 spell_soundfold_sofo(slang_T *slang, char_u *inword, char_u *res) |
| 375 | 7047 { |
| 7048 char_u *s; | |
| 7049 int ri = 0; | |
| 7050 int c; | |
| 7051 | |
| 7052 #ifdef FEAT_MBYTE | |
| 7053 if (has_mbyte) | |
| 7054 { | |
| 7055 int prevc = 0; | |
| 7056 int *ip; | |
| 7057 | |
| 7058 /* The sl_sal_first[] table contains the translation for chars up to | |
| 7059 * 255, sl_sal the rest. */ | |
| 7060 for (s = inword; *s != NUL; ) | |
| 7061 { | |
| 474 | 7062 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
|
7063 if (enc_utf8 ? utf_class(c) == 0 : VIM_ISWHITE(c)) |
| 375 | 7064 c = ' '; |
| 7065 else if (c < 256) | |
| 7066 c = slang->sl_sal_first[c]; | |
| 7067 else | |
| 7068 { | |
| 7069 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff]; | |
| 7070 if (ip == NULL) /* empty list, can't match */ | |
| 7071 c = NUL; | |
| 7072 else | |
| 7073 for (;;) /* find "c" in the list */ | |
| 7074 { | |
| 7075 if (*ip == 0) /* not found */ | |
| 7076 { | |
| 7077 c = NUL; | |
| 7078 break; | |
| 7079 } | |
| 7080 if (*ip == c) /* match! */ | |
| 7081 { | |
| 7082 c = ip[1]; | |
| 7083 break; | |
| 7084 } | |
| 7085 ip += 2; | |
| 7086 } | |
| 7087 } | |
| 7088 | |
| 7089 if (c != NUL && c != prevc) | |
| 7090 { | |
| 7091 ri += mb_char2bytes(c, res + ri); | |
| 7092 if (ri + MB_MAXBYTES > MAXWLEN) | |
| 7093 break; | |
| 7094 prevc = c; | |
| 7095 } | |
| 7096 } | |
| 7097 } | |
| 7098 else | |
| 7099 #endif | |
| 7100 { | |
| 7101 /* The sl_sal_first[] table contains the translation. */ | |
| 7102 for (s = inword; (c = *s) != NUL; ++s) | |
| 7103 { | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7104 if (VIM_ISWHITE(c)) |
| 375 | 7105 c = ' '; |
| 7106 else | |
| 7107 c = slang->sl_sal_first[c]; | |
| 7108 if (c != NUL && (ri == 0 || res[ri - 1] != c)) | |
| 7109 res[ri++] = c; | |
| 7110 } | |
| 7111 } | |
| 7112 | |
| 7113 res[ri] = NUL; | |
| 7114 } | |
| 7115 | |
| 7116 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7117 spell_soundfold_sal(slang_T *slang, char_u *inword, char_u *res) |
| 323 | 7118 { |
| 344 | 7119 salitem_T *smp; |
| 323 | 7120 char_u word[MAXWLEN]; |
| 375 | 7121 char_u *s = inword; |
| 323 | 7122 char_u *t; |
| 344 | 7123 char_u *pf; |
| 323 | 7124 int i, j, z; |
| 344 | 7125 int reslen; |
| 323 | 7126 int n, k = 0; |
| 7127 int z0; | |
| 7128 int k0; | |
| 7129 int n0; | |
| 7130 int c; | |
| 7131 int pri; | |
| 7132 int p0 = -333; | |
| 7133 int c0; | |
| 7134 | |
| 324 | 7135 /* Remove accents, if wanted. We actually remove all non-word characters. |
| 375 | 7136 * But keep white space. We need a copy, the word may be changed here. */ |
| 323 | 7137 if (slang->sl_rem_accents) |
| 7138 { | |
| 7139 t = word; | |
| 375 | 7140 while (*s != NUL) |
| 323 | 7141 { |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7142 if (VIM_ISWHITE(*s)) |
| 344 | 7143 { |
| 7144 *t++ = ' '; | |
| 7145 s = skipwhite(s); | |
| 7146 } | |
| 323 | 7147 else |
| 7148 { | |
| 5477 | 7149 if (spell_iswordp_nmw(s, curwin)) |
| 323 | 7150 *t++ = *s; |
| 7151 ++s; | |
| 7152 } | |
| 7153 } | |
| 7154 *t = NUL; | |
| 7155 } | |
| 7156 else | |
| 2768 | 7157 vim_strncpy(word, s, MAXWLEN - 1); |
| 323 | 7158 |
| 344 | 7159 smp = (salitem_T *)slang->sl_sal.ga_data; |
| 323 | 7160 |
| 7161 /* | |
| 7162 * This comes from Aspell phonet.cpp. Converted from C++ to C. | |
| 324 | 7163 * Changed to keep spaces. |
| 323 | 7164 */ |
| 344 | 7165 i = reslen = z = 0; |
| 323 | 7166 while ((c = word[i]) != NUL) |
| 7167 { | |
| 344 | 7168 /* Start with the first rule that has the character in the word. */ |
| 323 | 7169 n = slang->sl_sal_first[c]; |
| 7170 z0 = 0; | |
| 7171 | |
| 7172 if (n >= 0) | |
| 7173 { | |
| 7174 /* check all rules for the same letter */ | |
| 344 | 7175 for (; (s = smp[n].sm_lead)[0] == c; ++n) |
| 323 | 7176 { |
| 344 | 7177 /* Quickly skip entries that don't match the word. Most |
| 7178 * entries are less then three chars, optimize for that. */ | |
| 7179 k = smp[n].sm_leadlen; | |
| 7180 if (k > 1) | |
| 323 | 7181 { |
| 344 | 7182 if (word[i + 1] != s[1]) |
| 7183 continue; | |
| 7184 if (k > 2) | |
| 7185 { | |
| 7186 for (j = 2; j < k; ++j) | |
| 7187 if (word[i + j] != s[j]) | |
| 7188 break; | |
| 7189 if (j < k) | |
| 7190 continue; | |
| 7191 } | |
| 323 | 7192 } |
| 7193 | |
| 375 | 7194 if ((pf = smp[n].sm_oneof) != NULL) |
| 323 | 7195 { |
| 375 | 7196 /* Check for match with one of the chars in "sm_oneof". */ |
| 344 | 7197 while (*pf != NUL && *pf != word[i + k]) |
| 7198 ++pf; | |
| 7199 if (*pf == NUL) | |
| 7200 continue; | |
| 7201 ++k; | |
| 323 | 7202 } |
| 344 | 7203 s = smp[n].sm_rules; |
| 7204 pri = 5; /* default priority */ | |
| 323 | 7205 |
| 7206 p0 = *s; | |
| 7207 k0 = k; | |
| 7208 while (*s == '-' && k > 1) | |
| 7209 { | |
| 7210 k--; | |
| 7211 s++; | |
| 7212 } | |
| 7213 if (*s == '<') | |
| 7214 s++; | |
| 344 | 7215 if (VIM_ISDIGIT(*s)) |
| 323 | 7216 { |
| 7217 /* determine priority */ | |
| 7218 pri = *s - '0'; | |
| 7219 s++; | |
| 7220 } | |
| 7221 if (*s == '^' && *(s + 1) == '^') | |
| 7222 s++; | |
| 7223 | |
| 7224 if (*s == NUL | |
| 7225 || (*s == '^' | |
| 324 | 7226 && (i == 0 || !(word[i - 1] == ' ' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7227 || spell_iswordp(word + i - 1, curwin))) |
| 323 | 7228 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7229 || (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 7230 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7231 && spell_iswordp(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7232 && (!spell_iswordp(word + i + k0, curwin)))) |
| 323 | 7233 { |
| 7234 /* search for followup rules, if: */ | |
| 7235 /* followup and k > 1 and NO '-' in searchstring */ | |
| 7236 c0 = word[i + k - 1]; | |
| 7237 n0 = slang->sl_sal_first[c0]; | |
| 7238 | |
| 7239 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 344 | 7240 && p0 != '-' && word[i + k] != NUL) |
| 323 | 7241 { |
| 7242 /* test follow-up rule for "word[i + k]" */ | |
| 344 | 7243 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0) |
| 323 | 7244 { |
| 344 | 7245 /* Quickly skip entries that don't match the word. |
| 7246 * */ | |
| 7247 k0 = smp[n0].sm_leadlen; | |
| 7248 if (k0 > 1) | |
| 323 | 7249 { |
| 344 | 7250 if (word[i + k] != s[1]) |
| 7251 continue; | |
| 7252 if (k0 > 2) | |
| 7253 { | |
| 7254 pf = word + i + k + 1; | |
| 7255 for (j = 2; j < k0; ++j) | |
| 7256 if (*pf++ != s[j]) | |
| 7257 break; | |
| 7258 if (j < k0) | |
| 7259 continue; | |
| 7260 } | |
| 323 | 7261 } |
| 344 | 7262 k0 += k - 1; |
| 7263 | |
| 375 | 7264 if ((pf = smp[n0].sm_oneof) != NULL) |
| 323 | 7265 { |
| 344 | 7266 /* Check for match with one of the chars in |
| 375 | 7267 * "sm_oneof". */ |
| 344 | 7268 while (*pf != NUL && *pf != word[i + k0]) |
| 7269 ++pf; | |
| 7270 if (*pf == NUL) | |
| 7271 continue; | |
| 7272 ++k0; | |
| 323 | 7273 } |
| 344 | 7274 |
| 7275 p0 = 5; | |
| 7276 s = smp[n0].sm_rules; | |
| 323 | 7277 while (*s == '-') |
| 7278 { | |
| 344 | 7279 /* "k0" gets NOT reduced because |
| 7280 * "if (k0 == k)" */ | |
| 323 | 7281 s++; |
| 7282 } | |
| 7283 if (*s == '<') | |
| 7284 s++; | |
| 344 | 7285 if (VIM_ISDIGIT(*s)) |
| 323 | 7286 { |
| 7287 p0 = *s - '0'; | |
| 7288 s++; | |
| 7289 } | |
| 7290 | |
| 7291 if (*s == NUL | |
| 7292 /* *s == '^' cuts */ | |
| 7293 || (*s == '$' | |
| 376 | 7294 && !spell_iswordp(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7295 curwin))) |
| 323 | 7296 { |
| 7297 if (k0 == k) | |
| 7298 /* this is just a piece of the string */ | |
| 7299 continue; | |
| 7300 | |
| 7301 if (p0 < pri) | |
| 7302 /* priority too low */ | |
| 7303 continue; | |
| 7304 /* rule fits; stop search */ | |
| 7305 break; | |
| 7306 } | |
| 7307 } | |
| 7308 | |
| 344 | 7309 if (p0 >= pri && smp[n0].sm_lead[0] == c0) |
| 323 | 7310 continue; |
| 7311 } | |
| 7312 | |
| 7313 /* replace string */ | |
| 344 | 7314 s = smp[n].sm_to; |
| 389 | 7315 if (s == NULL) |
| 7316 s = (char_u *)""; | |
| 344 | 7317 pf = smp[n].sm_rules; |
| 7318 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0; | |
| 323 | 7319 if (p0 == 1 && z == 0) |
| 7320 { | |
| 7321 /* rule with '<' is used */ | |
| 344 | 7322 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c |
| 7323 || res[reslen - 1] == *s)) | |
| 7324 reslen--; | |
| 323 | 7325 z0 = 1; |
| 7326 z = 1; | |
| 7327 k0 = 0; | |
| 372 | 7328 while (*s != NUL && word[i + k0] != NUL) |
| 323 | 7329 { |
| 7330 word[i + k0] = *s; | |
| 7331 k0++; | |
| 7332 s++; | |
| 7333 } | |
| 7334 if (k > k0) | |
| 1619 | 7335 STRMOVE(word + i + k0, word + i + k); |
| 323 | 7336 |
| 7337 /* new "actual letter" */ | |
| 7338 c = word[i]; | |
| 7339 } | |
| 7340 else | |
| 7341 { | |
| 7342 /* no '<' rule used */ | |
| 7343 i += k - 1; | |
| 7344 z = 0; | |
| 344 | 7345 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN) |
| 323 | 7346 { |
| 344 | 7347 if (reslen == 0 || res[reslen - 1] != *s) |
| 372 | 7348 res[reslen++] = *s; |
| 323 | 7349 s++; |
| 7350 } | |
| 7351 /* new "actual letter" */ | |
| 7352 c = *s; | |
| 344 | 7353 if (strstr((char *)pf, "^^") != NULL) |
| 323 | 7354 { |
| 7355 if (c != NUL) | |
| 372 | 7356 res[reslen++] = c; |
| 1619 | 7357 STRMOVE(word, word + i + 1); |
| 323 | 7358 i = 0; |
| 7359 z0 = 1; | |
| 7360 } | |
| 7361 } | |
| 7362 break; | |
| 7363 } | |
| 7364 } | |
| 7365 } | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7366 else if (VIM_ISWHITE(c)) |
| 324 | 7367 { |
| 7368 c = ' '; | |
| 7369 k = 1; | |
| 7370 } | |
| 323 | 7371 |
| 7372 if (z0 == 0) | |
| 7373 { | |
| 344 | 7374 if (k && !p0 && reslen < MAXWLEN && c != NUL |
| 7375 && (!slang->sl_collapse || reslen == 0 | |
| 7376 || res[reslen - 1] != c)) | |
| 323 | 7377 /* condense only double letters */ |
| 372 | 7378 res[reslen++] = c; |
| 323 | 7379 |
| 7380 i++; | |
| 7381 z = 0; | |
| 7382 k = 0; | |
| 7383 } | |
| 7384 } | |
| 7385 | |
| 344 | 7386 res[reslen] = NUL; |
| 323 | 7387 } |
| 7388 | |
| 372 | 7389 #ifdef FEAT_MBYTE |
| 7390 /* | |
| 7391 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]". | |
| 7392 * Multi-byte version of spell_soundfold(). | |
| 7393 */ | |
| 7394 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7395 spell_soundfold_wsal(slang_T *slang, char_u *inword, char_u *res) |
| 372 | 7396 { |
| 375 | 7397 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data; |
| 372 | 7398 int word[MAXWLEN]; |
| 7399 int wres[MAXWLEN]; | |
| 7400 int l; | |
| 7401 char_u *s; | |
| 7402 int *ws; | |
| 7403 char_u *t; | |
| 7404 int *pf; | |
| 7405 int i, j, z; | |
| 7406 int reslen; | |
| 7407 int n, k = 0; | |
| 7408 int z0; | |
| 7409 int k0; | |
| 7410 int n0; | |
| 7411 int c; | |
| 7412 int pri; | |
| 7413 int p0 = -333; | |
| 7414 int c0; | |
| 7415 int did_white = FALSE; | |
| 3520 | 7416 int wordlen; |
| 7417 | |
| 372 | 7418 |
| 7419 /* | |
| 7420 * Convert the multi-byte string to a wide-character string. | |
| 7421 * Remove accents, if wanted. We actually remove all non-word characters. | |
| 7422 * But keep white space. | |
| 7423 */ | |
| 3520 | 7424 wordlen = 0; |
| 372 | 7425 for (s = inword; *s != NUL; ) |
| 7426 { | |
| 7427 t = s; | |
| 474 | 7428 c = mb_cptr2char_adv(&s); |
| 372 | 7429 if (slang->sl_rem_accents) |
| 7430 { | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7431 if (enc_utf8 ? utf_class(c) == 0 : VIM_ISWHITE(c)) |
| 372 | 7432 { |
| 7433 if (did_white) | |
| 7434 continue; | |
| 7435 c = ' '; | |
| 7436 did_white = TRUE; | |
| 7437 } | |
| 7438 else | |
| 7439 { | |
| 7440 did_white = FALSE; | |
| 5477 | 7441 if (!spell_iswordp_nmw(t, curwin)) |
| 372 | 7442 continue; |
| 7443 } | |
| 7444 } | |
| 3520 | 7445 word[wordlen++] = c; |
| 7446 } | |
| 7447 word[wordlen] = NUL; | |
| 372 | 7448 |
| 7449 /* | |
| 3520 | 7450 * This algorithm comes from Aspell phonet.cpp. |
| 372 | 7451 * Converted from C++ to C. Added support for multi-byte chars. |
| 7452 * Changed to keep spaces. | |
| 7453 */ | |
| 7454 i = reslen = z = 0; | |
| 7455 while ((c = word[i]) != NUL) | |
| 7456 { | |
| 7457 /* Start with the first rule that has the character in the word. */ | |
| 7458 n = slang->sl_sal_first[c & 0xff]; | |
| 7459 z0 = 0; | |
| 7460 | |
| 7461 if (n >= 0) | |
| 7462 { | |
|
2455
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
7463 /* 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
|
7464 * 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
|
7465 * (c & 0xff) is NUL. */ |
|
9367de3e2e1b
Fix: crash in spell checking with a 0x300 character.
Bram Moolenaar <bram@vim.org>
parents:
2454
diff
changeset
|
7466 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
|
7467 && ws[0] != NUL; ++n) |
| 372 | 7468 { |
| 7469 /* Quickly skip entries that don't match the word. Most | |
| 7470 * entries are less then three chars, optimize for that. */ | |
| 375 | 7471 if (c != ws[0]) |
| 7472 continue; | |
| 372 | 7473 k = smp[n].sm_leadlen; |
| 7474 if (k > 1) | |
| 7475 { | |
| 7476 if (word[i + 1] != ws[1]) | |
| 7477 continue; | |
| 7478 if (k > 2) | |
| 7479 { | |
| 7480 for (j = 2; j < k; ++j) | |
| 7481 if (word[i + j] != ws[j]) | |
| 7482 break; | |
| 7483 if (j < k) | |
| 7484 continue; | |
| 7485 } | |
| 7486 } | |
| 7487 | |
| 375 | 7488 if ((pf = smp[n].sm_oneof_w) != NULL) |
| 372 | 7489 { |
| 375 | 7490 /* Check for match with one of the chars in "sm_oneof". */ |
| 372 | 7491 while (*pf != NUL && *pf != word[i + k]) |
| 7492 ++pf; | |
| 7493 if (*pf == NUL) | |
| 7494 continue; | |
| 7495 ++k; | |
| 7496 } | |
| 7497 s = smp[n].sm_rules; | |
| 7498 pri = 5; /* default priority */ | |
| 7499 | |
| 7500 p0 = *s; | |
| 7501 k0 = k; | |
| 7502 while (*s == '-' && k > 1) | |
| 7503 { | |
| 7504 k--; | |
| 7505 s++; | |
| 7506 } | |
| 7507 if (*s == '<') | |
| 7508 s++; | |
| 7509 if (VIM_ISDIGIT(*s)) | |
| 7510 { | |
| 7511 /* determine priority */ | |
| 7512 pri = *s - '0'; | |
| 7513 s++; | |
| 7514 } | |
| 7515 if (*s == '^' && *(s + 1) == '^') | |
| 7516 s++; | |
| 7517 | |
| 7518 if (*s == NUL | |
| 7519 || (*s == '^' | |
| 7520 && (i == 0 || !(word[i - 1] == ' ' | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7521 || spell_iswordp_w(word + i - 1, curwin))) |
| 372 | 7522 && (*(s + 1) != '$' |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7523 || (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 7524 || (*s == '$' && i > 0 |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7525 && spell_iswordp_w(word + i - 1, curwin) |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7526 && (!spell_iswordp_w(word + i + k0, curwin)))) |
| 372 | 7527 { |
| 7528 /* search for followup rules, if: */ | |
| 7529 /* followup and k > 1 and NO '-' in searchstring */ | |
| 7530 c0 = word[i + k - 1]; | |
| 7531 n0 = slang->sl_sal_first[c0 & 0xff]; | |
| 7532 | |
| 7533 if (slang->sl_followup && k > 1 && n0 >= 0 | |
| 7534 && p0 != '-' && word[i + k] != NUL) | |
| 7535 { | |
| 375 | 7536 /* Test follow-up rule for "word[i + k]"; loop over |
| 7537 * all entries with the same index byte. */ | |
| 372 | 7538 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff) |
| 7539 == (c0 & 0xff); ++n0) | |
| 7540 { | |
| 7541 /* Quickly skip entries that don't match the word. | |
| 375 | 7542 */ |
| 7543 if (c0 != ws[0]) | |
| 7544 continue; | |
| 372 | 7545 k0 = smp[n0].sm_leadlen; |
| 7546 if (k0 > 1) | |
| 7547 { | |
| 7548 if (word[i + k] != ws[1]) | |
| 7549 continue; | |
| 7550 if (k0 > 2) | |
| 7551 { | |
| 7552 pf = word + i + k + 1; | |
| 7553 for (j = 2; j < k0; ++j) | |
| 7554 if (*pf++ != ws[j]) | |
| 7555 break; | |
| 7556 if (j < k0) | |
| 7557 continue; | |
| 7558 } | |
| 7559 } | |
| 7560 k0 += k - 1; | |
| 7561 | |
| 375 | 7562 if ((pf = smp[n0].sm_oneof_w) != NULL) |
| 372 | 7563 { |
| 7564 /* Check for match with one of the chars in | |
| 375 | 7565 * "sm_oneof". */ |
| 372 | 7566 while (*pf != NUL && *pf != word[i + k0]) |
| 7567 ++pf; | |
| 7568 if (*pf == NUL) | |
| 7569 continue; | |
| 7570 ++k0; | |
| 7571 } | |
| 7572 | |
| 7573 p0 = 5; | |
| 7574 s = smp[n0].sm_rules; | |
| 7575 while (*s == '-') | |
| 7576 { | |
| 7577 /* "k0" gets NOT reduced because | |
| 7578 * "if (k0 == k)" */ | |
| 7579 s++; | |
| 7580 } | |
| 7581 if (*s == '<') | |
| 7582 s++; | |
| 7583 if (VIM_ISDIGIT(*s)) | |
| 7584 { | |
| 7585 p0 = *s - '0'; | |
| 7586 s++; | |
| 7587 } | |
| 7588 | |
| 7589 if (*s == NUL | |
| 7590 /* *s == '^' cuts */ | |
| 7591 || (*s == '$' | |
| 376 | 7592 && !spell_iswordp_w(word + i + k0, |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
7593 curwin))) |
| 372 | 7594 { |
| 7595 if (k0 == k) | |
| 7596 /* this is just a piece of the string */ | |
| 7597 continue; | |
| 7598 | |
| 7599 if (p0 < pri) | |
| 7600 /* priority too low */ | |
| 7601 continue; | |
| 7602 /* rule fits; stop search */ | |
| 7603 break; | |
| 7604 } | |
| 7605 } | |
| 7606 | |
| 7607 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff) | |
| 7608 == (c0 & 0xff)) | |
| 7609 continue; | |
| 7610 } | |
| 7611 | |
| 7612 /* replace string */ | |
| 7613 ws = smp[n].sm_to_w; | |
| 7614 s = smp[n].sm_rules; | |
| 7615 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0; | |
| 7616 if (p0 == 1 && z == 0) | |
| 7617 { | |
| 7618 /* rule with '<' is used */ | |
| 389 | 7619 if (reslen > 0 && ws != NULL && *ws != NUL |
| 7620 && (wres[reslen - 1] == c | |
| 372 | 7621 || wres[reslen - 1] == *ws)) |
| 7622 reslen--; | |
| 7623 z0 = 1; | |
| 7624 z = 1; | |
| 7625 k0 = 0; | |
| 389 | 7626 if (ws != NULL) |
| 7627 while (*ws != NUL && word[i + k0] != NUL) | |
| 7628 { | |
| 7629 word[i + k0] = *ws; | |
| 7630 k0++; | |
| 7631 ws++; | |
| 7632 } | |
| 372 | 7633 if (k > k0) |
| 7634 mch_memmove(word + i + k0, word + i + k, | |
| 3520 | 7635 sizeof(int) * (wordlen - (i + k) + 1)); |
| 372 | 7636 |
| 7637 /* new "actual letter" */ | |
| 7638 c = word[i]; | |
| 7639 } | |
| 7640 else | |
| 7641 { | |
| 7642 /* no '<' rule used */ | |
| 7643 i += k - 1; | |
| 7644 z = 0; | |
| 389 | 7645 if (ws != NULL) |
| 7646 while (*ws != NUL && ws[1] != NUL | |
| 7647 && reslen < MAXWLEN) | |
| 7648 { | |
| 7649 if (reslen == 0 || wres[reslen - 1] != *ws) | |
| 7650 wres[reslen++] = *ws; | |
| 7651 ws++; | |
| 7652 } | |
| 372 | 7653 /* new "actual letter" */ |
| 389 | 7654 if (ws == NULL) |
| 7655 c = NUL; | |
| 7656 else | |
| 7657 c = *ws; | |
| 372 | 7658 if (strstr((char *)s, "^^") != NULL) |
| 7659 { | |
| 7660 if (c != NUL) | |
| 7661 wres[reslen++] = c; | |
| 7662 mch_memmove(word, word + i + 1, | |
| 3520 | 7663 sizeof(int) * (wordlen - (i + 1) + 1)); |
| 372 | 7664 i = 0; |
| 7665 z0 = 1; | |
| 7666 } | |
| 7667 } | |
| 7668 break; | |
| 7669 } | |
| 7670 } | |
| 7671 } | |
|
11129
f4ea50924c6d
patch 8.0.0452: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11127
diff
changeset
|
7672 else if (VIM_ISWHITE(c)) |
| 372 | 7673 { |
| 7674 c = ' '; | |
| 7675 k = 1; | |
| 7676 } | |
| 7677 | |
| 7678 if (z0 == 0) | |
| 7679 { | |
| 7680 if (k && !p0 && reslen < MAXWLEN && c != NUL | |
| 7681 && (!slang->sl_collapse || reslen == 0 | |
| 7682 || wres[reslen - 1] != c)) | |
| 7683 /* condense only double letters */ | |
| 7684 wres[reslen++] = c; | |
| 7685 | |
| 7686 i++; | |
| 7687 z = 0; | |
| 7688 k = 0; | |
| 7689 } | |
| 7690 } | |
| 7691 | |
| 7692 /* Convert wide characters in "wres" to a multi-byte string in "res". */ | |
| 7693 l = 0; | |
| 7694 for (n = 0; n < reslen; ++n) | |
| 7695 { | |
| 7696 l += mb_char2bytes(wres[n], res + l); | |
| 7697 if (l + MB_MAXBYTES > MAXWLEN) | |
| 7698 break; | |
| 7699 } | |
| 7700 res[l] = NUL; | |
| 7701 } | |
| 7702 #endif | |
| 7703 | |
| 324 | 7704 /* |
| 344 | 7705 * Compute a score for two sound-a-like words. |
| 7706 * This permits up to two inserts/deletes/swaps/etc. to keep things fast. | |
| 7707 * Instead of a generic loop we write out the code. That keeps it fast by | |
| 7708 * avoiding checks that will not be possible. | |
| 7709 */ | |
| 7710 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7711 soundalike_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7712 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
|
7713 char_u *badstart) /* sound-folded bad word */ |
| 344 | 7714 { |
| 351 | 7715 char_u *goodsound = goodstart; |
| 7716 char_u *badsound = badstart; | |
| 7717 int goodlen; | |
| 7718 int badlen; | |
| 344 | 7719 int n; |
| 7720 char_u *pl, *ps; | |
| 7721 char_u *pl2, *ps2; | |
| 351 | 7722 int score = 0; |
| 7723 | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7724 /* Adding/inserting "*" at the start (word starts with vowel) shouldn't be |
| 351 | 7725 * counted so much, vowels halfway the word aren't counted at all. */ |
| 7726 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound) | |
| 7727 { | |
|
2454
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7728 if ((badsound[0] == NUL && goodsound[1] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7729 || (goodsound[0] == NUL && badsound[1] == NUL)) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7730 /* 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
|
7731 return SCORE_DEL; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7732 if (badsound[0] == NUL || goodsound[0] == NUL) |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7733 /* more than two changes */ |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7734 return SCORE_MAXMAX; |
|
91fec91a522c
Fix: uninitialized memory access. (Dominique Pelle)
Bram Moolenaar <bram@vim.org>
parents:
2445
diff
changeset
|
7735 |
| 625 | 7736 if (badsound[1] == goodsound[1] |
| 7737 || (badsound[1] != NUL | |
| 7738 && goodsound[1] != NUL | |
| 7739 && badsound[2] == goodsound[2])) | |
| 7740 { | |
| 7741 /* handle like a substitute */ | |
| 7742 } | |
| 351 | 7743 else |
| 625 | 7744 { |
| 7745 score = 2 * SCORE_DEL / 3; | |
| 7746 if (*badsound == '*') | |
| 7747 ++badsound; | |
| 7748 else | |
| 7749 ++goodsound; | |
| 7750 } | |
| 351 | 7751 } |
| 7752 | |
| 835 | 7753 goodlen = (int)STRLEN(goodsound); |
| 7754 badlen = (int)STRLEN(badsound); | |
| 344 | 7755 |
| 1188 | 7756 /* Return quickly if the lengths are too different to be fixed by two |
| 344 | 7757 * changes. */ |
| 7758 n = goodlen - badlen; | |
| 7759 if (n < -2 || n > 2) | |
| 7760 return SCORE_MAXMAX; | |
| 7761 | |
| 7762 if (n > 0) | |
| 7763 { | |
| 351 | 7764 pl = goodsound; /* goodsound is longest */ |
| 344 | 7765 ps = badsound; |
| 7766 } | |
| 7767 else | |
| 7768 { | |
| 351 | 7769 pl = badsound; /* badsound is longest */ |
| 344 | 7770 ps = goodsound; |
| 7771 } | |
| 7772 | |
| 7773 /* Skip over the identical part. */ | |
| 7774 while (*pl == *ps && *pl != NUL) | |
| 7775 { | |
| 7776 ++pl; | |
| 7777 ++ps; | |
| 7778 } | |
| 7779 | |
| 7780 switch (n) | |
| 7781 { | |
| 7782 case -2: | |
| 7783 case 2: | |
| 7784 /* | |
| 7785 * Must delete two characters from "pl". | |
| 7786 */ | |
| 7787 ++pl; /* first delete */ | |
| 7788 while (*pl == *ps) | |
| 7789 { | |
| 7790 ++pl; | |
| 7791 ++ps; | |
| 7792 } | |
| 7793 /* strings must be equal after second delete */ | |
| 7794 if (STRCMP(pl + 1, ps) == 0) | |
| 351 | 7795 return score + SCORE_DEL * 2; |
| 344 | 7796 |
| 7797 /* Failed to compare. */ | |
| 7798 break; | |
| 7799 | |
| 7800 case -1: | |
| 7801 case 1: | |
| 7802 /* | |
| 7803 * Minimal one delete from "pl" required. | |
| 7804 */ | |
| 7805 | |
| 7806 /* 1: delete */ | |
| 7807 pl2 = pl + 1; | |
| 7808 ps2 = ps; | |
| 7809 while (*pl2 == *ps2) | |
| 7810 { | |
| 7811 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7812 return score + SCORE_DEL; |
| 344 | 7813 ++pl2; |
| 7814 ++ps2; | |
| 7815 } | |
| 7816 | |
| 7817 /* 2: delete then swap, then rest must be equal */ | |
| 7818 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7819 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7820 return score + SCORE_DEL + SCORE_SWAP; |
| 344 | 7821 |
| 7822 /* 3: delete then substitute, then the rest must be equal */ | |
| 7823 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7824 return score + SCORE_DEL + SCORE_SUBST; |
| 344 | 7825 |
| 7826 /* 4: first swap then delete */ | |
| 7827 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 7828 { | |
| 7829 pl2 = pl + 2; /* swap, skip two chars */ | |
| 7830 ps2 = ps + 2; | |
| 7831 while (*pl2 == *ps2) | |
| 7832 { | |
| 7833 ++pl2; | |
| 7834 ++ps2; | |
| 7835 } | |
| 7836 /* delete a char and then strings must be equal */ | |
| 7837 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7838 return score + SCORE_SWAP + SCORE_DEL; |
| 344 | 7839 } |
| 7840 | |
| 7841 /* 5: first substitute then delete */ | |
| 7842 pl2 = pl + 1; /* substitute, skip one char */ | |
| 7843 ps2 = ps + 1; | |
| 7844 while (*pl2 == *ps2) | |
| 7845 { | |
| 7846 ++pl2; | |
| 7847 ++ps2; | |
| 7848 } | |
| 7849 /* delete a char and then strings must be equal */ | |
| 7850 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7851 return score + SCORE_SUBST + SCORE_DEL; |
| 344 | 7852 |
| 7853 /* Failed to compare. */ | |
| 7854 break; | |
| 7855 | |
| 7856 case 0: | |
| 7857 /* | |
| 1808 | 7858 * Lengths are equal, thus changes must result in same length: An |
| 344 | 7859 * insert is only possible in combination with a delete. |
| 7860 * 1: check if for identical strings | |
| 7861 */ | |
| 7862 if (*pl == NUL) | |
| 351 | 7863 return score; |
| 344 | 7864 |
| 7865 /* 2: swap */ | |
| 7866 if (pl[0] == ps[1] && pl[1] == ps[0]) | |
| 7867 { | |
| 7868 pl2 = pl + 2; /* swap, skip two chars */ | |
| 7869 ps2 = ps + 2; | |
| 7870 while (*pl2 == *ps2) | |
| 7871 { | |
| 7872 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7873 return score + SCORE_SWAP; |
| 344 | 7874 ++pl2; |
| 7875 ++ps2; | |
| 7876 } | |
| 7877 /* 3: swap and swap again */ | |
| 7878 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7879 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7880 return score + SCORE_SWAP + SCORE_SWAP; |
| 344 | 7881 |
| 7882 /* 4: swap and substitute */ | |
| 7883 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7884 return score + SCORE_SWAP + SCORE_SUBST; |
| 344 | 7885 } |
| 7886 | |
| 7887 /* 5: substitute */ | |
| 7888 pl2 = pl + 1; | |
| 7889 ps2 = ps + 1; | |
| 7890 while (*pl2 == *ps2) | |
| 7891 { | |
| 7892 if (*pl2 == NUL) /* reached the end */ | |
| 351 | 7893 return score + SCORE_SUBST; |
| 344 | 7894 ++pl2; |
| 7895 ++ps2; | |
| 7896 } | |
| 7897 | |
| 7898 /* 6: substitute and swap */ | |
| 7899 if (pl2[0] == ps2[1] && pl2[1] == ps2[0] | |
| 7900 && STRCMP(pl2 + 2, ps2 + 2) == 0) | |
| 351 | 7901 return score + SCORE_SUBST + SCORE_SWAP; |
| 344 | 7902 |
| 7903 /* 7: substitute and substitute */ | |
| 7904 if (STRCMP(pl2 + 1, ps2 + 1) == 0) | |
| 351 | 7905 return score + SCORE_SUBST + SCORE_SUBST; |
| 344 | 7906 |
| 7907 /* 8: insert then delete */ | |
| 7908 pl2 = pl; | |
| 7909 ps2 = ps + 1; | |
| 7910 while (*pl2 == *ps2) | |
| 7911 { | |
| 7912 ++pl2; | |
| 7913 ++ps2; | |
| 7914 } | |
| 7915 if (STRCMP(pl2 + 1, ps2) == 0) | |
| 351 | 7916 return score + SCORE_INS + SCORE_DEL; |
| 344 | 7917 |
| 7918 /* 9: delete then insert */ | |
| 7919 pl2 = pl + 1; | |
| 7920 ps2 = ps; | |
| 7921 while (*pl2 == *ps2) | |
| 7922 { | |
| 7923 ++pl2; | |
| 7924 ++ps2; | |
| 7925 } | |
| 7926 if (STRCMP(pl2, ps2 + 1) == 0) | |
| 351 | 7927 return score + SCORE_INS + SCORE_DEL; |
| 344 | 7928 |
| 7929 /* Failed to compare. */ | |
| 7930 break; | |
| 7931 } | |
| 7932 | |
| 7933 return SCORE_MAXMAX; | |
| 7934 } | |
| 324 | 7935 |
| 323 | 7936 /* |
| 7937 * Compute the "edit distance" to turn "badword" into "goodword". The less | |
| 344 | 7938 * deletes/inserts/substitutes/swaps are required the lower the score. |
| 324 | 7939 * |
| 485 | 7940 * The algorithm is described by Du and Chang, 1992. |
| 7941 * The implementation of the algorithm comes from Aspell editdist.cpp, | |
| 7942 * edit_distance(). It has been converted from C++ to C and modified to | |
| 7943 * support multi-byte characters. | |
| 323 | 7944 */ |
| 7945 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7946 spell_edit_score( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7947 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7948 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
7949 char_u *goodword) |
| 323 | 7950 { |
| 7951 int *cnt; | |
| 1188 | 7952 int badlen, goodlen; /* lengths including NUL */ |
| 323 | 7953 int j, i; |
| 7954 int t; | |
| 7955 int bc, gc; | |
| 324 | 7956 int pbc, pgc; |
| 7957 #ifdef FEAT_MBYTE | |
| 7958 char_u *p; | |
| 7959 int wbadword[MAXWLEN]; | |
| 7960 int wgoodword[MAXWLEN]; | |
| 7961 | |
| 7962 if (has_mbyte) | |
| 7963 { | |
| 7964 /* Get the characters from the multi-byte strings and put them in an | |
| 7965 * int array for easy access. */ | |
| 7966 for (p = badword, badlen = 0; *p != NUL; ) | |
| 474 | 7967 wbadword[badlen++] = mb_cptr2char_adv(&p); |
| 398 | 7968 wbadword[badlen++] = 0; |
| 324 | 7969 for (p = goodword, goodlen = 0; *p != NUL; ) |
| 474 | 7970 wgoodword[goodlen++] = mb_cptr2char_adv(&p); |
| 398 | 7971 wgoodword[goodlen++] = 0; |
| 324 | 7972 } |
| 7973 else | |
| 7974 #endif | |
| 7975 { | |
| 835 | 7976 badlen = (int)STRLEN(badword) + 1; |
| 7977 goodlen = (int)STRLEN(goodword) + 1; | |
| 324 | 7978 } |
| 323 | 7979 |
| 7980 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */ | |
| 7981 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)] | |
| 7982 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)), | |
| 7983 TRUE); | |
| 324 | 7984 if (cnt == NULL) |
| 7985 return 0; /* out of memory */ | |
| 323 | 7986 |
| 7987 CNT(0, 0) = 0; | |
| 7988 for (j = 1; j <= goodlen; ++j) | |
| 625 | 7989 CNT(0, j) = CNT(0, j - 1) + SCORE_INS; |
| 323 | 7990 |
| 7991 for (i = 1; i <= badlen; ++i) | |
| 7992 { | |
| 625 | 7993 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL; |
| 323 | 7994 for (j = 1; j <= goodlen; ++j) |
| 7995 { | |
| 324 | 7996 #ifdef FEAT_MBYTE |
| 7997 if (has_mbyte) | |
| 7998 { | |
| 7999 bc = wbadword[i - 1]; | |
| 8000 gc = wgoodword[j - 1]; | |
| 8001 } | |
| 8002 else | |
| 8003 #endif | |
| 8004 { | |
| 8005 bc = badword[i - 1]; | |
| 8006 gc = goodword[j - 1]; | |
| 8007 } | |
| 323 | 8008 if (bc == gc) |
| 8009 CNT(i, j) = CNT(i - 1, j - 1); | |
| 8010 else | |
| 8011 { | |
| 8012 /* Use a better score when there is only a case difference. */ | |
| 324 | 8013 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) |
| 323 | 8014 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1); |
| 8015 else | |
| 625 | 8016 { |
| 8017 /* For a similar character use SCORE_SIMILAR. */ | |
| 8018 if (slang != NULL | |
| 8019 && slang->sl_has_map | |
| 8020 && similar_chars(slang, gc, bc)) | |
| 8021 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1); | |
| 8022 else | |
| 8023 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1); | |
| 8024 } | |
| 323 | 8025 |
| 324 | 8026 if (i > 1 && j > 1) |
| 323 | 8027 { |
| 324 | 8028 #ifdef FEAT_MBYTE |
| 8029 if (has_mbyte) | |
| 8030 { | |
| 8031 pbc = wbadword[i - 2]; | |
| 8032 pgc = wgoodword[j - 2]; | |
| 8033 } | |
| 8034 else | |
| 8035 #endif | |
| 8036 { | |
| 8037 pbc = badword[i - 2]; | |
| 8038 pgc = goodword[j - 2]; | |
| 8039 } | |
| 8040 if (bc == pgc && pbc == gc) | |
| 8041 { | |
| 8042 t = SCORE_SWAP + CNT(i - 2, j - 2); | |
| 8043 if (t < CNT(i, j)) | |
| 8044 CNT(i, j) = t; | |
| 8045 } | |
| 323 | 8046 } |
| 8047 t = SCORE_DEL + CNT(i - 1, j); | |
| 8048 if (t < CNT(i, j)) | |
| 8049 CNT(i, j) = t; | |
| 8050 t = SCORE_INS + CNT(i, j - 1); | |
| 8051 if (t < CNT(i, j)) | |
| 8052 CNT(i, j) = t; | |
| 8053 } | |
| 8054 } | |
| 8055 } | |
| 344 | 8056 |
| 8057 i = CNT(badlen - 1, goodlen - 1); | |
| 8058 vim_free(cnt); | |
| 8059 return i; | |
| 323 | 8060 } |
| 307 | 8061 |
| 625 | 8062 typedef struct |
| 8063 { | |
| 8064 int badi; | |
| 8065 int goodi; | |
| 8066 int score; | |
| 8067 } limitscore_T; | |
| 8068 | |
| 8069 /* | |
| 8070 * Like spell_edit_score(), but with a limit on the score to make it faster. | |
| 8071 * May return SCORE_MAXMAX when the score is higher than "limit". | |
| 8072 * | |
| 8073 * This uses a stack for the edits still to be tried. | |
| 8074 * The idea comes from Aspell leditdist.cpp. Rewritten in C and added support | |
| 8075 * for multi-byte characters. | |
| 8076 */ | |
| 8077 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8078 spell_edit_score_limit( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8079 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8080 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8081 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8082 int limit) |
| 625 | 8083 { |
| 8084 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 8085 int stackidx; | |
| 8086 int bi, gi; | |
| 8087 int bi2, gi2; | |
| 8088 int bc, gc; | |
| 8089 int score; | |
| 8090 int score_off; | |
| 8091 int minscore; | |
| 8092 int round; | |
| 8093 | |
| 8094 #ifdef FEAT_MBYTE | |
| 8095 /* Multi-byte characters require a bit more work, use a different function | |
| 8096 * to avoid testing "has_mbyte" quite often. */ | |
| 8097 if (has_mbyte) | |
| 8098 return spell_edit_score_limit_w(slang, badword, goodword, limit); | |
| 8099 #endif | |
| 8100 | |
| 8101 /* | |
| 8102 * The idea is to go from start to end over the words. So long as | |
| 8103 * characters are equal just continue, this always gives the lowest score. | |
| 8104 * When there is a difference try several alternatives. Each alternative | |
| 8105 * increases "score" for the edit distance. Some of the alternatives are | |
| 8106 * pushed unto a stack and tried later, some are tried right away. At the | |
| 8107 * end of the word the score for one alternative is known. The lowest | |
| 8108 * possible score is stored in "minscore". | |
| 8109 */ | |
| 8110 stackidx = 0; | |
| 8111 bi = 0; | |
| 8112 gi = 0; | |
| 8113 score = 0; | |
| 8114 minscore = limit + 1; | |
| 8115 | |
| 8116 for (;;) | |
| 8117 { | |
| 8118 /* Skip over an equal part, score remains the same. */ | |
| 8119 for (;;) | |
| 8120 { | |
| 8121 bc = badword[bi]; | |
| 8122 gc = goodword[gi]; | |
| 8123 if (bc != gc) /* stop at a char that's different */ | |
| 8124 break; | |
| 8125 if (bc == NUL) /* both words end */ | |
| 8126 { | |
| 8127 if (score < minscore) | |
| 8128 minscore = score; | |
| 8129 goto pop; /* do next alternative */ | |
| 8130 } | |
| 8131 ++bi; | |
| 8132 ++gi; | |
| 8133 } | |
| 8134 | |
| 8135 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 8136 { | |
| 8137 do | |
| 8138 { | |
| 8139 if ((score += SCORE_DEL) >= minscore) | |
| 8140 goto pop; /* do next alternative */ | |
| 8141 } while (badword[++bi] != NUL); | |
| 8142 minscore = score; | |
| 8143 } | |
| 8144 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 8145 { | |
| 8146 do | |
| 8147 { | |
| 8148 if ((score += SCORE_INS) >= minscore) | |
| 8149 goto pop; /* do next alternative */ | |
| 8150 } while (goodword[++gi] != NUL); | |
| 8151 minscore = score; | |
| 8152 } | |
| 8153 else /* both words continue */ | |
| 8154 { | |
| 8155 /* If not close to the limit, perform a change. Only try changes | |
| 8156 * that may lead to a lower score than "minscore". | |
| 8157 * round 0: try deleting a char from badword | |
| 8158 * round 1: try inserting a char in badword */ | |
| 8159 for (round = 0; round <= 1; ++round) | |
| 8160 { | |
| 8161 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 8162 if (score_off < minscore) | |
| 8163 { | |
| 8164 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 8165 { | |
| 8166 /* Near the limit, rest of the words must match. We | |
| 8167 * can check that right now, no need to push an item | |
| 8168 * onto the stack. */ | |
| 8169 bi2 = bi + 1 - round; | |
| 8170 gi2 = gi + round; | |
| 8171 while (goodword[gi2] == badword[bi2]) | |
| 8172 { | |
| 8173 if (goodword[gi2] == NUL) | |
| 8174 { | |
| 8175 minscore = score_off; | |
| 8176 break; | |
| 8177 } | |
| 8178 ++bi2; | |
| 8179 ++gi2; | |
| 8180 } | |
| 8181 } | |
| 8182 else | |
| 8183 { | |
| 8184 /* try deleting/inserting a character later */ | |
| 8185 stack[stackidx].badi = bi + 1 - round; | |
| 8186 stack[stackidx].goodi = gi + round; | |
| 8187 stack[stackidx].score = score_off; | |
| 8188 ++stackidx; | |
| 8189 } | |
| 8190 } | |
| 8191 } | |
| 8192 | |
| 8193 if (score + SCORE_SWAP < minscore) | |
| 8194 { | |
| 8195 /* If swapping two characters makes a match then the | |
| 8196 * substitution is more expensive, thus there is no need to | |
| 8197 * try both. */ | |
| 8198 if (gc == badword[bi + 1] && bc == goodword[gi + 1]) | |
| 8199 { | |
| 8200 /* Swap two characters, that is: skip them. */ | |
| 8201 gi += 2; | |
| 8202 bi += 2; | |
| 8203 score += SCORE_SWAP; | |
| 8204 continue; | |
| 8205 } | |
| 8206 } | |
| 8207 | |
| 8208 /* Substitute one character for another which is the same | |
| 8209 * thing as deleting a character from both goodword and badword. | |
| 8210 * Use a better score when there is only a case difference. */ | |
| 8211 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 8212 score += SCORE_ICASE; | |
| 8213 else | |
| 8214 { | |
| 8215 /* For a similar character use SCORE_SIMILAR. */ | |
| 8216 if (slang != NULL | |
| 8217 && slang->sl_has_map | |
| 8218 && similar_chars(slang, gc, bc)) | |
| 8219 score += SCORE_SIMILAR; | |
| 8220 else | |
| 8221 score += SCORE_SUBST; | |
| 8222 } | |
| 8223 | |
| 8224 if (score < minscore) | |
| 8225 { | |
| 8226 /* Do the substitution. */ | |
| 8227 ++gi; | |
| 8228 ++bi; | |
| 8229 continue; | |
| 8230 } | |
| 8231 } | |
| 8232 pop: | |
| 8233 /* | |
| 8234 * Get here to try the next alternative, pop it from the stack. | |
| 8235 */ | |
| 8236 if (stackidx == 0) /* stack is empty, finished */ | |
| 8237 break; | |
| 8238 | |
| 8239 /* pop an item from the stack */ | |
| 8240 --stackidx; | |
| 8241 gi = stack[stackidx].goodi; | |
| 8242 bi = stack[stackidx].badi; | |
| 8243 score = stack[stackidx].score; | |
| 8244 } | |
| 8245 | |
| 8246 /* When the score goes over "limit" it may actually be much higher. | |
| 8247 * Return a very large number to avoid going below the limit when giving a | |
| 8248 * bonus. */ | |
| 8249 if (minscore > limit) | |
| 8250 return SCORE_MAXMAX; | |
| 8251 return minscore; | |
| 8252 } | |
| 8253 | |
| 8254 #ifdef FEAT_MBYTE | |
| 8255 /* | |
| 8256 * Multi-byte version of spell_edit_score_limit(). | |
| 8257 * Keep it in sync with the above! | |
| 8258 */ | |
| 8259 static int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8260 spell_edit_score_limit_w( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8261 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8262 char_u *badword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8263 char_u *goodword, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8264 int limit) |
| 625 | 8265 { |
| 8266 limitscore_T stack[10]; /* allow for over 3 * 2 edits */ | |
| 8267 int stackidx; | |
| 8268 int bi, gi; | |
| 8269 int bi2, gi2; | |
| 8270 int bc, gc; | |
| 8271 int score; | |
| 8272 int score_off; | |
| 8273 int minscore; | |
| 8274 int round; | |
| 8275 char_u *p; | |
| 8276 int wbadword[MAXWLEN]; | |
| 8277 int wgoodword[MAXWLEN]; | |
| 8278 | |
| 8279 /* Get the characters from the multi-byte strings and put them in an | |
| 8280 * int array for easy access. */ | |
| 8281 bi = 0; | |
| 8282 for (p = badword; *p != NUL; ) | |
| 8283 wbadword[bi++] = mb_cptr2char_adv(&p); | |
| 8284 wbadword[bi++] = 0; | |
| 8285 gi = 0; | |
| 8286 for (p = goodword; *p != NUL; ) | |
| 8287 wgoodword[gi++] = mb_cptr2char_adv(&p); | |
| 8288 wgoodword[gi++] = 0; | |
| 8289 | |
| 8290 /* | |
| 8291 * The idea is to go from start to end over the words. So long as | |
| 8292 * characters are equal just continue, this always gives the lowest score. | |
| 8293 * When there is a difference try several alternatives. Each alternative | |
| 8294 * increases "score" for the edit distance. Some of the alternatives are | |
| 8295 * pushed unto a stack and tried later, some are tried right away. At the | |
| 8296 * end of the word the score for one alternative is known. The lowest | |
| 8297 * possible score is stored in "minscore". | |
| 8298 */ | |
| 8299 stackidx = 0; | |
| 8300 bi = 0; | |
| 8301 gi = 0; | |
| 8302 score = 0; | |
| 8303 minscore = limit + 1; | |
| 8304 | |
| 8305 for (;;) | |
| 8306 { | |
| 8307 /* Skip over an equal part, score remains the same. */ | |
| 8308 for (;;) | |
| 8309 { | |
| 8310 bc = wbadword[bi]; | |
| 8311 gc = wgoodword[gi]; | |
| 8312 | |
| 8313 if (bc != gc) /* stop at a char that's different */ | |
| 8314 break; | |
| 8315 if (bc == NUL) /* both words end */ | |
| 8316 { | |
| 8317 if (score < minscore) | |
| 8318 minscore = score; | |
| 8319 goto pop; /* do next alternative */ | |
| 8320 } | |
| 8321 ++bi; | |
| 8322 ++gi; | |
| 8323 } | |
| 8324 | |
| 8325 if (gc == NUL) /* goodword ends, delete badword chars */ | |
| 8326 { | |
| 8327 do | |
| 8328 { | |
| 8329 if ((score += SCORE_DEL) >= minscore) | |
| 8330 goto pop; /* do next alternative */ | |
| 8331 } while (wbadword[++bi] != NUL); | |
| 8332 minscore = score; | |
| 8333 } | |
| 8334 else if (bc == NUL) /* badword ends, insert badword chars */ | |
| 8335 { | |
| 8336 do | |
| 8337 { | |
| 8338 if ((score += SCORE_INS) >= minscore) | |
| 8339 goto pop; /* do next alternative */ | |
| 8340 } while (wgoodword[++gi] != NUL); | |
| 8341 minscore = score; | |
| 8342 } | |
| 8343 else /* both words continue */ | |
| 8344 { | |
| 8345 /* If not close to the limit, perform a change. Only try changes | |
| 8346 * that may lead to a lower score than "minscore". | |
| 8347 * round 0: try deleting a char from badword | |
| 8348 * round 1: try inserting a char in badword */ | |
| 8349 for (round = 0; round <= 1; ++round) | |
| 8350 { | |
| 8351 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS); | |
| 8352 if (score_off < minscore) | |
| 8353 { | |
| 8354 if (score_off + SCORE_EDIT_MIN >= minscore) | |
| 8355 { | |
| 8356 /* Near the limit, rest of the words must match. We | |
| 8357 * can check that right now, no need to push an item | |
| 8358 * onto the stack. */ | |
| 8359 bi2 = bi + 1 - round; | |
| 8360 gi2 = gi + round; | |
| 8361 while (wgoodword[gi2] == wbadword[bi2]) | |
| 8362 { | |
| 8363 if (wgoodword[gi2] == NUL) | |
| 8364 { | |
| 8365 minscore = score_off; | |
| 8366 break; | |
| 8367 } | |
| 8368 ++bi2; | |
| 8369 ++gi2; | |
| 8370 } | |
| 8371 } | |
| 8372 else | |
| 8373 { | |
| 8374 /* try deleting a character from badword later */ | |
| 8375 stack[stackidx].badi = bi + 1 - round; | |
| 8376 stack[stackidx].goodi = gi + round; | |
| 8377 stack[stackidx].score = score_off; | |
| 8378 ++stackidx; | |
| 8379 } | |
| 8380 } | |
| 8381 } | |
| 8382 | |
| 8383 if (score + SCORE_SWAP < minscore) | |
| 8384 { | |
| 8385 /* If swapping two characters makes a match then the | |
| 8386 * substitution is more expensive, thus there is no need to | |
| 8387 * try both. */ | |
| 8388 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1]) | |
| 8389 { | |
| 8390 /* Swap two characters, that is: skip them. */ | |
| 8391 gi += 2; | |
| 8392 bi += 2; | |
| 8393 score += SCORE_SWAP; | |
| 8394 continue; | |
| 8395 } | |
| 8396 } | |
| 8397 | |
| 8398 /* Substitute one character for another which is the same | |
| 8399 * thing as deleting a character from both goodword and badword. | |
| 8400 * Use a better score when there is only a case difference. */ | |
| 8401 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc)) | |
| 8402 score += SCORE_ICASE; | |
| 8403 else | |
| 8404 { | |
| 8405 /* For a similar character use SCORE_SIMILAR. */ | |
| 8406 if (slang != NULL | |
| 8407 && slang->sl_has_map | |
| 8408 && similar_chars(slang, gc, bc)) | |
| 8409 score += SCORE_SIMILAR; | |
| 8410 else | |
| 8411 score += SCORE_SUBST; | |
| 8412 } | |
| 8413 | |
| 8414 if (score < minscore) | |
| 8415 { | |
| 8416 /* Do the substitution. */ | |
| 8417 ++gi; | |
| 8418 ++bi; | |
| 8419 continue; | |
| 8420 } | |
| 8421 } | |
| 8422 pop: | |
| 8423 /* | |
| 8424 * Get here to try the next alternative, pop it from the stack. | |
| 8425 */ | |
| 8426 if (stackidx == 0) /* stack is empty, finished */ | |
| 8427 break; | |
| 8428 | |
| 8429 /* pop an item from the stack */ | |
| 8430 --stackidx; | |
| 8431 gi = stack[stackidx].goodi; | |
| 8432 bi = stack[stackidx].badi; | |
| 8433 score = stack[stackidx].score; | |
| 8434 } | |
| 8435 | |
| 8436 /* When the score goes over "limit" it may actually be much higher. | |
| 8437 * Return a very large number to avoid going below the limit when giving a | |
| 8438 * bonus. */ | |
| 8439 if (minscore > limit) | |
| 8440 return SCORE_MAXMAX; | |
| 8441 return minscore; | |
| 8442 } | |
| 8443 #endif | |
| 8444 | |
| 714 | 8445 /* |
| 8446 * ":spellinfo" | |
| 8447 */ | |
| 8448 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8449 ex_spellinfo(exarg_T *eap UNUSED) |
| 714 | 8450 { |
| 8451 int lpi; | |
| 8452 langp_T *lp; | |
| 8453 char_u *p; | |
| 8454 | |
| 8455 if (no_spell_checking(curwin)) | |
| 8456 return; | |
| 8457 | |
| 8458 msg_start(); | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8459 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
|
8460 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8461 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 714 | 8462 msg_puts((char_u *)"file: "); |
| 8463 msg_puts(lp->lp_slang->sl_fname); | |
| 8464 msg_putchar('\n'); | |
| 8465 p = lp->lp_slang->sl_info; | |
| 8466 if (p != NULL) | |
| 8467 { | |
| 8468 msg_puts(p); | |
| 8469 msg_putchar('\n'); | |
| 8470 } | |
| 8471 } | |
| 8472 msg_end(); | |
| 8473 } | |
| 8474 | |
| 625 | 8475 #define DUMPFLAG_KEEPCASE 1 /* round 2: keep-case tree */ |
| 8476 #define DUMPFLAG_COUNT 2 /* include word count */ | |
| 701 | 8477 #define DUMPFLAG_ICASE 4 /* ignore case when finding matches */ |
| 710 | 8478 #define DUMPFLAG_ONECAP 8 /* pattern starts with capital */ |
| 8479 #define DUMPFLAG_ALLCAP 16 /* pattern is all capitals */ | |
| 625 | 8480 |
| 351 | 8481 /* |
| 8482 * ":spelldump" | |
| 8483 */ | |
| 8484 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8485 ex_spelldump(exarg_T *eap) |
| 351 | 8486 { |
| 5382 | 8487 char_u *spl; |
| 8488 long dummy; | |
| 8489 | |
| 701 | 8490 if (no_spell_checking(curwin)) |
| 8491 return; | |
| 5382 | 8492 get_option_value((char_u*)"spl", &dummy, &spl, OPT_LOCAL); |
| 8493 | |
| 8494 /* Create a new empty buffer in a new window. */ | |
| 701 | 8495 do_cmdline_cmd((char_u *)"new"); |
| 5382 | 8496 |
| 8497 /* enable spelling locally in the new window */ | |
| 8498 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
|
8499 set_option_value((char_u*)"spl", dummy, spl, OPT_LOCAL); |
| 5382 | 8500 vim_free(spl); |
| 8501 | |
|
11121
778c10516955
patch 8.0.0448: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
10950
diff
changeset
|
8502 if (!BUFEMPTY()) |
| 701 | 8503 return; |
| 8504 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8505 spell_dump_compl(NULL, 0, NULL, eap->forceit ? DUMPFLAG_COUNT : 0); |
| 701 | 8506 |
| 8507 /* Delete the empty line that we started with. */ | |
| 8508 if (curbuf->b_ml.ml_line_count > 1) | |
| 8509 ml_delete(curbuf->b_ml.ml_line_count, FALSE); | |
| 8510 | |
| 8511 redraw_later(NOT_VALID); | |
| 8512 } | |
| 8513 | |
| 8514 /* | |
| 8515 * Go through all possible words and: | |
| 8516 * 1. When "pat" is NULL: dump a list of all words in the current buffer. | |
| 8517 * "ic" and "dir" are not used. | |
| 8518 * 2. When "pat" is not NULL: add matching words to insert mode completion. | |
| 8519 */ | |
| 8520 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8521 spell_dump_compl( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8522 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
|
8523 int ic, /* ignore case */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8524 int *dir, /* direction for adding matches */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8525 int dumpflags_arg) /* DUMPFLAG_* */ |
| 701 | 8526 { |
| 351 | 8527 langp_T *lp; |
| 8528 slang_T *slang; | |
| 8529 idx_T arridx[MAXWLEN]; | |
| 8530 int curi[MAXWLEN]; | |
| 8531 char_u word[MAXWLEN]; | |
| 8532 int c; | |
| 8533 char_u *byts; | |
| 8534 idx_T *idxs; | |
| 8535 linenr_T lnum = 0; | |
| 8536 int round; | |
| 8537 int depth; | |
| 8538 int n; | |
| 8539 int flags; | |
| 381 | 8540 char_u *region_names = NULL; /* region names being used */ |
| 8541 int do_region = TRUE; /* dump region names and numbers */ | |
| 8542 char_u *p; | |
| 500 | 8543 int lpi; |
| 701 | 8544 int dumpflags = dumpflags_arg; |
| 8545 int patlen; | |
| 8546 | |
| 710 | 8547 /* When ignoring case or when the pattern starts with capital pass this on |
| 8548 * to dump_word(). */ | |
| 8549 if (pat != NULL) | |
| 8550 { | |
| 8551 if (ic) | |
| 8552 dumpflags |= DUMPFLAG_ICASE; | |
| 8553 else | |
| 8554 { | |
| 8555 n = captype(pat, NULL); | |
| 8556 if (n == WF_ONECAP) | |
| 8557 dumpflags |= DUMPFLAG_ONECAP; | |
| 8558 else if (n == WF_ALLCAP | |
| 8559 #ifdef FEAT_MBYTE | |
| 714 | 8560 && (int)STRLEN(pat) > mb_ptr2len(pat) |
| 710 | 8561 #else |
| 714 | 8562 && (int)STRLEN(pat) > 1 |
| 710 | 8563 #endif |
| 8564 ) | |
| 8565 dumpflags |= DUMPFLAG_ALLCAP; | |
| 8566 } | |
| 8567 } | |
| 351 | 8568 |
| 381 | 8569 /* Find out if we can support regions: All languages must support the same |
| 8570 * regions or none at all. */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8571 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
|
8572 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8573 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 381 | 8574 p = lp->lp_slang->sl_regions; |
| 8575 if (p[0] != 0) | |
| 8576 { | |
| 8577 if (region_names == NULL) /* first language with regions */ | |
| 8578 region_names = p; | |
| 8579 else if (STRCMP(region_names, p) != 0) | |
| 8580 { | |
| 8581 do_region = FALSE; /* region names are different */ | |
| 8582 break; | |
| 8583 } | |
| 8584 } | |
| 8585 } | |
| 8586 | |
| 8587 if (do_region && region_names != NULL) | |
| 8588 { | |
| 701 | 8589 if (pat == NULL) |
| 8590 { | |
| 8591 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names); | |
| 8592 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 8593 } | |
| 381 | 8594 } |
| 8595 else | |
| 8596 do_region = FALSE; | |
| 8597 | |
| 8598 /* | |
| 8599 * Loop over all files loaded for the entries in 'spelllang'. | |
| 8600 */ | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8601 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
|
8602 { |
|
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8603 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi); |
| 351 | 8604 slang = lp->lp_slang; |
| 500 | 8605 if (slang->sl_fbyts == NULL) /* reloading failed */ |
| 8606 continue; | |
| 351 | 8607 |
| 701 | 8608 if (pat == NULL) |
| 8609 { | |
| 8610 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname); | |
| 8611 ml_append(lnum++, IObuff, (colnr_T)0, FALSE); | |
| 8612 } | |
| 8613 | |
| 8614 /* When matching with a pattern and there are no prefixes only use | |
| 8615 * parts of the tree that match "pat". */ | |
| 8616 if (pat != NULL && slang->sl_pbyts == NULL) | |
| 835 | 8617 patlen = (int)STRLEN(pat); |
| 701 | 8618 else |
| 840 | 8619 patlen = -1; |
| 351 | 8620 |
| 8621 /* round 1: case-folded tree | |
| 8622 * round 2: keep-case tree */ | |
| 8623 for (round = 1; round <= 2; ++round) | |
| 8624 { | |
| 8625 if (round == 1) | |
| 8626 { | |
| 701 | 8627 dumpflags &= ~DUMPFLAG_KEEPCASE; |
| 351 | 8628 byts = slang->sl_fbyts; |
| 8629 idxs = slang->sl_fidxs; | |
| 8630 } | |
| 8631 else | |
| 8632 { | |
| 701 | 8633 dumpflags |= DUMPFLAG_KEEPCASE; |
| 351 | 8634 byts = slang->sl_kbyts; |
| 8635 idxs = slang->sl_kidxs; | |
| 8636 } | |
| 8637 if (byts == NULL) | |
| 8638 continue; /* array is empty */ | |
| 8639 | |
| 8640 depth = 0; | |
| 8641 arridx[0] = 0; | |
| 8642 curi[0] = 1; | |
| 701 | 8643 while (depth >= 0 && !got_int |
| 8644 && (pat == NULL || !compl_interrupted)) | |
| 351 | 8645 { |
| 8646 if (curi[depth] > byts[arridx[depth]]) | |
| 8647 { | |
| 8648 /* Done all bytes at this node, go up one level. */ | |
| 8649 --depth; | |
| 8650 line_breakcheck(); | |
|
10277
154d5a2e7395
commit https://github.com/vim/vim/commit/472e85970ee3a80abd824bef510df12e9cfe9e96
Christian Brabandt <cb@256bit.org>
parents:
10042
diff
changeset
|
8651 ins_compl_check_keys(50, FALSE); |
| 351 | 8652 } |
| 8653 else | |
| 8654 { | |
| 8655 /* Do one more byte at this node. */ | |
| 8656 n = arridx[depth] + curi[depth]; | |
| 8657 ++curi[depth]; | |
| 8658 c = byts[n]; | |
| 8659 if (c == 0) | |
| 8660 { | |
| 8661 /* End of word, deal with the word. | |
| 8662 * Don't use keep-case words in the fold-case tree, | |
| 8663 * they will appear in the keep-case tree. | |
| 8664 * Only use the word when the region matches. */ | |
| 8665 flags = (int)idxs[n]; | |
| 8666 if ((round == 2 || (flags & WF_KEEPCAP) == 0) | |
| 500 | 8667 && (flags & WF_NEEDCOMP) == 0 |
| 381 | 8668 && (do_region |
| 8669 || (flags & WF_REGION) == 0 | |
| 390 | 8670 || (((unsigned)flags >> 16) |
| 351 | 8671 & lp->lp_region) != 0)) |
| 8672 { | |
| 8673 word[depth] = NUL; | |
| 381 | 8674 if (!do_region) |
| 8675 flags &= ~WF_REGION; | |
| 355 | 8676 |
| 8677 /* Dump the basic word if there is no prefix or | |
| 8678 * when it's the first one. */ | |
| 390 | 8679 c = (unsigned)flags >> 24; |
| 355 | 8680 if (c == 0 || curi[depth] == 2) |
| 701 | 8681 { |
| 8682 dump_word(slang, word, pat, dir, | |
| 8683 dumpflags, flags, lnum); | |
| 8684 if (pat == NULL) | |
| 8685 ++lnum; | |
| 8686 } | |
| 351 | 8687 |
| 8688 /* Apply the prefix, if there is one. */ | |
| 355 | 8689 if (c != 0) |
| 701 | 8690 lnum = dump_prefixes(slang, word, pat, dir, |
| 8691 dumpflags, flags, lnum); | |
| 351 | 8692 } |
| 8693 } | |
| 8694 else | |
| 8695 { | |
| 8696 /* Normal char, go one level deeper. */ | |
| 8697 word[depth++] = c; | |
| 8698 arridx[depth] = idxs[n]; | |
| 8699 curi[depth] = 1; | |
| 701 | 8700 |
| 8701 /* Check if this characters matches with the pattern. | |
| 8702 * If not skip the whole tree below it. | |
| 710 | 8703 * Always ignore case here, dump_word() will check |
| 8704 * proper case later. This isn't exactly right when | |
| 8705 * length changes for multi-byte characters with | |
| 8706 * ignore case... */ | |
| 8707 if (depth <= patlen | |
| 8708 && MB_STRNICMP(word, pat, depth) != 0) | |
| 701 | 8709 --depth; |
| 351 | 8710 } |
| 8711 } | |
| 8712 } | |
| 8713 } | |
| 8714 } | |
| 8715 } | |
| 8716 | |
| 8717 /* | |
| 8718 * Dump one word: apply case modifications and append a line to the buffer. | |
| 701 | 8719 * When "lnum" is zero add insert mode completion. |
| 8720 */ | |
| 8721 static void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8722 dump_word( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8723 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8724 char_u *word, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8725 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8726 int *dir, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8727 int dumpflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8728 int wordflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8729 linenr_T lnum) |
| 351 | 8730 { |
| 8731 int keepcap = FALSE; | |
| 8732 char_u *p; | |
| 625 | 8733 char_u *tw; |
| 351 | 8734 char_u cword[MAXWLEN]; |
| 381 | 8735 char_u badword[MAXWLEN + 10]; |
| 8736 int i; | |
| 710 | 8737 int flags = wordflags; |
| 8738 | |
| 8739 if (dumpflags & DUMPFLAG_ONECAP) | |
| 8740 flags |= WF_ONECAP; | |
| 8741 if (dumpflags & DUMPFLAG_ALLCAP) | |
| 8742 flags |= WF_ALLCAP; | |
| 351 | 8743 |
| 625 | 8744 if ((dumpflags & DUMPFLAG_KEEPCASE) == 0 && (flags & WF_CAPMASK) != 0) |
| 351 | 8745 { |
| 8746 /* Need to fix case according to "flags". */ | |
| 8747 make_case_word(word, cword, flags); | |
| 8748 p = cword; | |
| 8749 } | |
| 8750 else | |
| 8751 { | |
| 8752 p = word; | |
| 625 | 8753 if ((dumpflags & DUMPFLAG_KEEPCASE) |
| 8754 && ((captype(word, NULL) & WF_KEEPCAP) == 0 | |
| 389 | 8755 || (flags & WF_FIXCAP) != 0)) |
| 351 | 8756 keepcap = TRUE; |
| 8757 } | |
| 625 | 8758 tw = p; |
| 351 | 8759 |
| 701 | 8760 if (pat == NULL) |
| 8761 { | |
| 8762 /* Add flags and regions after a slash. */ | |
| 8763 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap) | |
| 8764 { | |
| 8765 STRCPY(badword, p); | |
| 8766 STRCAT(badword, "/"); | |
| 8767 if (keepcap) | |
| 8768 STRCAT(badword, "="); | |
| 8769 if (flags & WF_BANNED) | |
| 8770 STRCAT(badword, "!"); | |
| 8771 else if (flags & WF_RARE) | |
| 8772 STRCAT(badword, "?"); | |
| 8773 if (flags & WF_REGION) | |
| 8774 for (i = 0; i < 7; ++i) | |
| 8775 if (flags & (0x10000 << i)) | |
| 8776 sprintf((char *)badword + STRLEN(badword), "%d", i + 1); | |
| 8777 p = badword; | |
| 8778 } | |
| 8779 | |
| 8780 if (dumpflags & DUMPFLAG_COUNT) | |
| 8781 { | |
| 8782 hashitem_T *hi; | |
| 8783 | |
| 8784 /* Include the word count for ":spelldump!". */ | |
| 8785 hi = hash_find(&slang->sl_wordcount, tw); | |
| 8786 if (!HASHITEM_EMPTY(hi)) | |
| 8787 { | |
| 8788 vim_snprintf((char *)IObuff, IOSIZE, "%s\t%d", | |
| 625 | 8789 tw, HI2WC(hi)->wc_count); |
| 701 | 8790 p = IObuff; |
| 8791 } | |
| 8792 } | |
| 8793 | |
| 8794 ml_append(lnum, p, (colnr_T)0, FALSE); | |
| 8795 } | |
| 710 | 8796 else if (((dumpflags & DUMPFLAG_ICASE) |
| 8797 ? MB_STRNICMP(p, pat, STRLEN(pat)) == 0 | |
| 8798 : STRNCMP(p, pat, STRLEN(pat)) == 0) | |
| 701 | 8799 && ins_compl_add_infercase(p, (int)STRLEN(p), |
| 942 | 8800 p_ic, NULL, *dir, 0) == OK) |
| 710 | 8801 /* if dir was BACKWARD then honor it just once */ |
| 8802 *dir = FORWARD; | |
| 351 | 8803 } |
| 8804 | |
| 8805 /* | |
| 372 | 8806 * For ":spelldump": Find matching prefixes for "word". Prepend each to |
| 8807 * "word" and append a line to the buffer. | |
| 701 | 8808 * When "lnum" is zero add insert mode completion. |
| 351 | 8809 * Return the updated line number. |
| 8810 */ | |
| 8811 static linenr_T | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8812 dump_prefixes( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8813 slang_T *slang, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8814 char_u *word, /* case-folded word */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8815 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8816 int *dir, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8817 int dumpflags, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8818 int flags, /* flags with prefix ID */ |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8819 linenr_T startlnum) |
| 351 | 8820 { |
| 8821 idx_T arridx[MAXWLEN]; | |
| 8822 int curi[MAXWLEN]; | |
| 8823 char_u prefix[MAXWLEN]; | |
| 455 | 8824 char_u word_up[MAXWLEN]; |
| 8825 int has_word_up = FALSE; | |
| 351 | 8826 int c; |
| 8827 char_u *byts; | |
| 8828 idx_T *idxs; | |
| 8829 linenr_T lnum = startlnum; | |
| 8830 int depth; | |
| 8831 int n; | |
| 8832 int len; | |
| 8833 int i; | |
| 8834 | |
| 626 | 8835 /* If the word starts with a lower-case letter make the word with an |
| 455 | 8836 * upper-case letter in word_up[]. */ |
| 8837 c = PTR2CHAR(word); | |
| 8838 if (SPELL_TOUPPER(c) != c) | |
| 8839 { | |
| 8840 onecap_copy(word, word_up, TRUE); | |
| 8841 has_word_up = TRUE; | |
| 8842 } | |
| 8843 | |
| 351 | 8844 byts = slang->sl_pbyts; |
| 8845 idxs = slang->sl_pidxs; | |
| 8846 if (byts != NULL) /* array not is empty */ | |
| 8847 { | |
| 8848 /* | |
| 8849 * Loop over all prefixes, building them byte-by-byte in prefix[]. | |
| 390 | 8850 * When at the end of a prefix check that it supports "flags". |
| 351 | 8851 */ |
| 8852 depth = 0; | |
| 8853 arridx[0] = 0; | |
| 8854 curi[0] = 1; | |
| 8855 while (depth >= 0 && !got_int) | |
| 8856 { | |
| 390 | 8857 n = arridx[depth]; |
| 8858 len = byts[n]; | |
| 8859 if (curi[depth] > len) | |
| 351 | 8860 { |
| 8861 /* Done all bytes at this node, go up one level. */ | |
| 8862 --depth; | |
| 8863 line_breakcheck(); | |
| 8864 } | |
| 8865 else | |
| 8866 { | |
| 8867 /* Do one more byte at this node. */ | |
| 390 | 8868 n += curi[depth]; |
| 351 | 8869 ++curi[depth]; |
| 8870 c = byts[n]; | |
| 8871 if (c == 0) | |
| 8872 { | |
| 8873 /* End of prefix, find out how many IDs there are. */ | |
| 8874 for (i = 1; i < len; ++i) | |
| 8875 if (byts[n + i] != 0) | |
| 8876 break; | |
| 8877 curi[depth] += i - 1; | |
| 8878 | |
| 455 | 8879 c = valid_word_prefix(i, n, flags, word, slang, FALSE); |
| 8880 if (c != 0) | |
| 351 | 8881 { |
| 376 | 8882 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1); |
| 701 | 8883 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 8884 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 8885 : flags, lnum); |
| 8886 if (lnum != 0) | |
| 8887 ++lnum; | |
| 351 | 8888 } |
| 455 | 8889 |
| 8890 /* Check for prefix that matches the word when the | |
| 8891 * first letter is upper-case, but only if the prefix has | |
| 8892 * a condition. */ | |
| 8893 if (has_word_up) | |
| 8894 { | |
| 8895 c = valid_word_prefix(i, n, flags, word_up, slang, | |
| 8896 TRUE); | |
| 8897 if (c != 0) | |
| 8898 { | |
| 8899 vim_strncpy(prefix + depth, word_up, | |
| 8900 MAXWLEN - depth - 1); | |
| 701 | 8901 dump_word(slang, prefix, pat, dir, dumpflags, |
| 455 | 8902 (c & WF_RAREPFX) ? (flags | WF_RARE) |
| 701 | 8903 : flags, lnum); |
| 8904 if (lnum != 0) | |
| 8905 ++lnum; | |
| 455 | 8906 } |
| 8907 } | |
| 351 | 8908 } |
| 8909 else | |
| 8910 { | |
| 8911 /* Normal char, go one level deeper. */ | |
| 8912 prefix[depth++] = c; | |
| 8913 arridx[depth] = idxs[n]; | |
| 8914 curi[depth] = 1; | |
| 8915 } | |
| 8916 } | |
| 8917 } | |
| 8918 } | |
| 8919 | |
| 8920 return lnum; | |
| 8921 } | |
| 8922 | |
| 498 | 8923 /* |
| 626 | 8924 * Move "p" to the end of word "start". |
| 8925 * Uses the spell-checking word characters. | |
| 498 | 8926 */ |
| 8927 char_u * | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8928 spell_to_word_end(char_u *start, win_T *win) |
| 498 | 8929 { |
| 8930 char_u *p = start; | |
| 8931 | |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8932 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
|
8933 MB_PTR_ADV(p); |
| 498 | 8934 return p; |
| 8935 } | |
| 8936 | |
| 475 | 8937 #if defined(FEAT_INS_EXPAND) || defined(PROTO) |
| 8938 /* | |
| 626 | 8939 * For Insert mode completion CTRL-X s: |
| 8940 * Find start of the word in front of column "startcol". | |
| 8941 * We don't check if it is badly spelled, with completion we can only change | |
| 8942 * the word in front of the cursor. | |
| 475 | 8943 * Returns the column number of the word. |
| 8944 */ | |
| 8945 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8946 spell_word_start(int startcol) |
| 475 | 8947 { |
| 8948 char_u *line; | |
| 8949 char_u *p; | |
| 8950 int col = 0; | |
| 8951 | |
| 498 | 8952 if (no_spell_checking(curwin)) |
| 475 | 8953 return startcol; |
| 8954 | |
| 8955 /* Find a word character before "startcol". */ | |
| 8956 line = ml_get_curline(); | |
| 8957 for (p = line + startcol; p > line; ) | |
| 8958 { | |
|
11127
506f5d8b7d8b
patch 8.0.0451: some macros are in lower case
Christian Brabandt <cb@256bit.org>
parents:
11121
diff
changeset
|
8959 MB_PTR_BACK(line, p); |
| 5477 | 8960 if (spell_iswordp_nmw(p, curwin)) |
| 475 | 8961 break; |
| 8962 } | |
| 8963 | |
| 8964 /* Go back to start of the word. */ | |
| 8965 while (p > line) | |
| 8966 { | |
| 835 | 8967 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
|
8968 MB_PTR_BACK(line, p); |
|
2250
1bac28a53fae
Add the conceal patch from Vince Negri.
Bram Moolenaar <bram@vim.org>
parents:
2229
diff
changeset
|
8969 if (!spell_iswordp(p, curwin)) |
| 475 | 8970 break; |
| 8971 col = 0; | |
| 8972 } | |
| 8973 | |
| 535 | 8974 return col; |
| 8975 } | |
| 8976 | |
| 8977 /* | |
| 8978 * Need to check for 'spellcapcheck' now, the word is removed before | |
| 8979 * expand_spelling() is called. Therefore the ugly global variable. | |
| 8980 */ | |
| 8981 static int spell_expand_need_cap; | |
| 8982 | |
| 8983 void | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8984 spell_expand_check_cap(colnr_T col) |
| 535 | 8985 { |
| 475 | 8986 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col); |
| 8987 } | |
| 8988 | |
| 8989 /* | |
| 8990 * Get list of spelling suggestions. | |
| 8991 * Used for Insert mode completion CTRL-X ?. | |
| 8992 * Returns the number of matches. The matches are in "matchp[]", array of | |
| 8993 * allocated strings. | |
| 8994 */ | |
| 8995 int | |
|
7835
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8996 expand_spelling( |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8997 linenr_T lnum UNUSED, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8998 char_u *pat, |
|
4d7ce6c03fda
commit https://github.com/vim/vim/commit/764b23c8fd3369cb05ae9122abf3ca16fec539d7
Christian Brabandt <cb@256bit.org>
parents:
7805
diff
changeset
|
8999 char_u ***matchp) |
| 475 | 9000 { |
| 9001 garray_T ga; | |
| 9002 | |
| 625 | 9003 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap, TRUE); |
| 475 | 9004 *matchp = ga.ga_data; |
| 9005 return ga.ga_len; | |
| 9006 } | |
| 9007 #endif | |
| 9008 | |
| 737 | 9009 #endif /* FEAT_SPELL */ |