|
223
|
1 /* vi:set ts=8 sts=4 sw=4:
|
|
|
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 *
|
|
300
|
13 * The spell checking mechanism uses a tree (aka trie). Each node in the tree
|
|
|
14 * has a list of bytes that can appear (siblings). For each byte there is a
|
|
|
15 * pointer to the node with the byte that follows in the word (child).
|
|
324
|
16 *
|
|
|
17 * A NUL byte is used where the word may end. The bytes are sorted, so that
|
|
|
18 * binary searching can be used and the NUL bytes are at the start. The
|
|
|
19 * number of possible bytes is stored before the list of bytes.
|
|
|
20 *
|
|
|
21 * The tree uses two arrays: "byts" stores the characters, "idxs" stores
|
|
|
22 * either the next index or flags. The tree starts at index 0. For example,
|
|
|
23 * to lookup "vi" this sequence is followed:
|
|
|
24 * i = 0
|
|
|
25 * len = byts[i]
|
|
|
26 * n = where "v" appears in byts[i + 1] to byts[i + len]
|
|
|
27 * i = idxs[n]
|
|
|
28 * len = byts[i]
|
|
|
29 * n = where "i" appears in byts[i + 1] to byts[i + len]
|
|
|
30 * i = idxs[n]
|
|
|
31 * len = byts[i]
|
|
|
32 * find that byts[i + 1] is 0, idxs[i + 1] has flags for "vi".
|
|
300
|
33 *
|
|
339
|
34 * There are two word trees: one with case-folded words and one with words in
|
|
300
|
35 * original case. The second one is only used for keep-case words and is
|
|
|
36 * usually small.
|
|
|
37 *
|
|
339
|
38 * There is one additional tree for when prefixes are not applied when
|
|
|
39 * generating the .spl file. This tree stores all the possible prefixes, as
|
|
|
40 * if they were words. At each word (prefix) end the prefix nr is stored, the
|
|
|
41 * following word must support this prefix nr. And the condition nr is
|
|
|
42 * stored, used to lookup the condition that the word must match with.
|
|
|
43 *
|
|
300
|
44 * Thanks to Olaf Seibert for providing an example implementation of this tree
|
|
|
45 * and the compression mechanism.
|
|
243
|
46 *
|
|
|
47 * Matching involves checking the caps type: Onecap ALLCAP KeepCap.
|
|
|
48 *
|
|
236
|
49 * Why doesn't Vim use aspell/ispell/myspell/etc.?
|
|
|
50 * See ":help develop-spell".
|
|
|
51 */
|
|
|
52
|
|
300
|
53 /*
|
|
324
|
54 * Use this to adjust the score after finding suggestions, based on the
|
|
|
55 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
56 * it for every possible suggestion.
|
|
|
57 * Disadvantage: When "the" is typed as "hte" it sounds different and goes
|
|
|
58 * down in the list.
|
|
344
|
59 * Used when 'spellsuggest' is set to "best".
|
|
|
60 */
|
|
|
61 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
62
|
|
|
63 /*
|
|
|
64 * The double scoring mechanism is based on the principle that there are two
|
|
|
65 * kinds of spelling mistakes:
|
|
|
66 * 1. You know how to spell the word, but mistype something. This results in
|
|
|
67 * a small editing distance (character swapped/omitted/inserted) and
|
|
|
68 * possibly a word that sounds completely different.
|
|
|
69 * 2. You don't know how to spell the word and type something that sounds
|
|
|
70 * right. The edit distance can be big but the word is similar after
|
|
|
71 * sound-folding.
|
|
|
72 * Since scores for these two mistakes will be very different we use a list
|
|
|
73 * for each.
|
|
|
74 * The sound-folding is slow, only do double scoring when 'spellsuggest' is
|
|
|
75 * "double".
|
|
323
|
76 */
|
|
|
77
|
|
|
78 /*
|
|
339
|
79 * Vim spell file format: <HEADER>
|
|
|
80 * <SUGGEST>
|
|
|
81 * <LWORDTREE>
|
|
|
82 * <KWORDTREE>
|
|
|
83 * <PREFIXTREE>
|
|
300
|
84 *
|
|
339
|
85 * <HEADER>: <fileID>
|
|
|
86 * <regioncnt> <regionname> ...
|
|
|
87 * <charflagslen> <charflags>
|
|
|
88 * <fcharslen> <fchars>
|
|
366
|
89 * <midwordlen> <midword>
|
|
339
|
90 * <prefcondcnt> <prefcond> ...
|
|
300
|
91 *
|
|
366
|
92 * <fileID> 10 bytes "VIMspell08"
|
|
300
|
93 * <regioncnt> 1 byte number of regions following (8 supported)
|
|
307
|
94 * <regionname> 2 bytes Region name: ca, au, etc. Lower case.
|
|
300
|
95 * First <regionname> is region 1.
|
|
|
96 *
|
|
|
97 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
98 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
99 * 0x01 word character CF_WORD
|
|
|
100 * 0x02 upper-case character CF_UPPER
|
|
300
|
101 * <fcharslen> 2 bytes Number of bytes in <fchars>.
|
|
|
102 * <fchars> N bytes Folded characters, first one is for character 128.
|
|
|
103 *
|
|
366
|
104 * <midwordlen> 2 bytes Number of bytes in <midword>.
|
|
|
105 * <midword> N bytes Characters that are word characters only when used
|
|
|
106 * in the middle of a word.
|
|
|
107 *
|
|
339
|
108 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
109 *
|
|
|
110 * <prefcond> : <condlen> <condstr>
|
|
|
111 *
|
|
|
112 * <condlen> 1 byte Length of <condstr>.
|
|
|
113 *
|
|
|
114 * <condstr> N bytes Condition for the prefix.
|
|
|
115 *
|
|
300
|
116 *
|
|
323
|
117 * <SUGGEST> : <repcount> <rep> ...
|
|
|
118 * <salflags> <salcount> <sal> ...
|
|
|
119 * <maplen> <mapstr>
|
|
|
120 *
|
|
|
121 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
|
122 *
|
|
|
123 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
|
124 *
|
|
|
125 * <repfromlen> 1 byte length of <repfrom>
|
|
|
126 *
|
|
|
127 * <repfrom> N bytes "from" part of replacement
|
|
|
128 *
|
|
|
129 * <reptolen> 1 byte length of <repto>
|
|
|
130 *
|
|
|
131 * <repto> N bytes "to" part of replacement
|
|
300
|
132 *
|
|
323
|
133 * <salflags> 1 byte flags for soundsalike conversion:
|
|
|
134 * SAL_F0LLOWUP
|
|
|
135 * SAL_COLLAPSE
|
|
|
136 * SAL_REM_ACCENTS
|
|
375
|
137 * SAL_SOFO: SOFOFROM and SOFOTO used instead of SAL
|
|
|
138 *
|
|
|
139 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
140 *
|
|
|
141 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
|
142 *
|
|
375
|
143 * <salfromlen> 1-2 bytes length of <salfrom> (2 bytes for SAL_SOFO)
|
|
323
|
144 *
|
|
|
145 * <salfrom> N bytes "from" part of soundsalike
|
|
|
146 *
|
|
375
|
147 * <saltolen> 1-2 bytes length of <salto> (2 bytes for SAL_SOFO)
|
|
323
|
148 *
|
|
|
149 * <salto> N bytes "to" part of soundsalike
|
|
|
150 *
|
|
|
151 * <maplen> 2 bytes length of <mapstr>, MSB first
|
|
|
152 *
|
|
|
153 * <mapstr> N bytes String with sequences of similar characters,
|
|
|
154 * separated by slashes.
|
|
300
|
155 *
|
|
|
156 *
|
|
|
157 * <LWORDTREE>: <wordtree>
|
|
|
158 *
|
|
339
|
159 * <KWORDTREE>: <wordtree>
|
|
|
160 *
|
|
|
161 * <PREFIXTREE>: <wordtree>
|
|
|
162 *
|
|
|
163 *
|
|
300
|
164 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
165 *
|
|
|
166 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
167 *
|
|
|
168 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
169 *
|
|
|
170 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
171 * follow in sorted order.
|
|
|
172 *
|
|
339
|
173 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
|
174 * | <flags> [<region>] [<prefixID>]
|
|
|
175 * | <prefixID> <prefcondnr> ]
|
|
300
|
176 *
|
|
|
177 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
178 * BY_NOFLAGS: End of word without flags and for all
|
|
|
179 * regions.
|
|
366
|
180 * For PREFIXTREE <prefixID> and
|
|
|
181 * <prefcondnr> follow.
|
|
|
182 * BY_FLAGS: End of word, <flags> follow.
|
|
|
183 * For PREFIXTREE <prefixID> and
|
|
|
184 * <prefcondnr> follow for rare prefix.
|
|
300
|
185 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
|
186 * and <xbyte> follow.
|
|
|
187 *
|
|
|
188 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
189 *
|
|
|
190 * <xbyte> 1 byte byte value of the sibling.
|
|
|
191 *
|
|
|
192 * <flags> 1 byte bitmask of:
|
|
|
193 * WF_ALLCAP word must have only capitals
|
|
|
194 * WF_ONECAP first char of word must be capital
|
|
|
195 * WF_RARE rare word
|
|
|
196 * WF_REGION <region> follows
|
|
339
|
197 * WF_PFX <prefixID> follows
|
|
300
|
198 *
|
|
|
199 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
200 * omitted it's valid in all regions.
|
|
|
201 * Lowest bit is for region 1.
|
|
|
202 *
|
|
339
|
203 * <prefixID> 1 byte ID of prefix that can be used with this word. For
|
|
|
204 * PREFIXTREE used for the required prefix ID.
|
|
|
205 *
|
|
|
206 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
207 * from HEADER.
|
|
300
|
208 *
|
|
|
209 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
210 */
|
|
|
211
|
|
223
|
212 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
213 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
214 #endif
|
|
|
215
|
|
|
216 #include "vim.h"
|
|
|
217
|
|
|
218 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
219
|
|
|
220 #ifdef HAVE_FCNTL_H
|
|
|
221 # include <fcntl.h>
|
|
|
222 #endif
|
|
|
223
|
|
323
|
224 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
225 Some places assume a word length fits in a
|
|
|
226 byte, thus it can't be above 255. */
|
|
226
|
227
|
|
324
|
228 /* Type used for indexes in the word tree need to be at least 3 bytes. If int
|
|
|
229 * is 8 bytes we could use something smaller, but what? */
|
|
|
230 #if SIZEOF_INT > 2
|
|
|
231 typedef int idx_T;
|
|
|
232 #else
|
|
|
233 typedef long idx_T;
|
|
|
234 #endif
|
|
|
235
|
|
|
236 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
237 * comes above it. */
|
|
300
|
238 #define WF_REGION 0x01 /* region byte follows */
|
|
|
239 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
240 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
241 #define WF_RARE 0x08 /* rare word */
|
|
307
|
242 #define WF_BANNED 0x10 /* bad word */
|
|
339
|
243 #define WF_PFX 0x20 /* prefix ID list follows */
|
|
323
|
244 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
245
|
|
|
246 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP)
|
|
300
|
247
|
|
366
|
248 #define WF_RAREPFX 0x1000000 /* in sl_pidxs: flag for rare postponed
|
|
|
249 prefix; must be above prefixID (one byte)
|
|
|
250 and prefcondnr (two bytes) */
|
|
|
251
|
|
300
|
252 #define BY_NOFLAGS 0 /* end of word without flags or region */
|
|
|
253 #define BY_FLAGS 1 /* end of word, flag byte follows */
|
|
|
254 #define BY_INDEX 2 /* child is shared, index follows */
|
|
|
255 #define BY_SPECIAL BY_INDEX /* hightest special byte value */
|
|
236
|
256
|
|
323
|
257 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
258 * and si_sal. Not for sl_sal!
|
|
323
|
259 * One replacement: from "ft_from" to "ft_to". */
|
|
|
260 typedef struct fromto_S
|
|
236
|
261 {
|
|
323
|
262 char_u *ft_from;
|
|
|
263 char_u *ft_to;
|
|
|
264 } fromto_T;
|
|
236
|
265
|
|
344
|
266 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
267 * The info is split for quick processing by spell_soundfold().
|
|
|
268 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
269 typedef struct salitem_S
|
|
|
270 {
|
|
|
271 char_u *sm_lead; /* leading letters */
|
|
|
272 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
273 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
274 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
275 char_u *sm_to; /* replacement. */
|
|
372
|
276 #ifdef FEAT_MBYTE
|
|
|
277 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
278 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
279 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
280 #endif
|
|
344
|
281 } salitem_T;
|
|
|
282
|
|
375
|
283 #ifdef FEAT_MBYTE
|
|
|
284 typedef int salfirst_T;
|
|
|
285 #else
|
|
|
286 typedef short salfirst_T;
|
|
|
287 #endif
|
|
|
288
|
|
236
|
289 /*
|
|
243
|
290 * Structure used to store words and other info for one language, loaded from
|
|
|
291 * a .spl file.
|
|
300
|
292 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
293 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
294 *
|
|
|
295 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
296 * the number of possible bytes, sorted on byte value:
|
|
|
297 * <len> <byte1> <byte2> ...
|
|
|
298 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
299 * byte in "byts".
|
|
|
300 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
|
301 * the flags and region for the word. There may be several zeros in sequence
|
|
|
302 * for alternative flag/region combinations.
|
|
236
|
303 */
|
|
|
304 typedef struct slang_S slang_T;
|
|
|
305 struct slang_S
|
|
|
306 {
|
|
|
307 slang_T *sl_next; /* next language */
|
|
|
308 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
309 char_u *sl_fname; /* name of .spl file */
|
|
323
|
310 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
311
|
|
300
|
312 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
313 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
314 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
315 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
316 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
317 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
318
|
|
236
|
319 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
320
|
|
376
|
321 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
322
|
|
339
|
323 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
324 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
325
|
|
323
|
326 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
327 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
328 there is none */
|
|
344
|
329 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
330 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
331 there is none */
|
|
375
|
332 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
333 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
334 * "sl_sal" is a list of wide char lists. */
|
|
323
|
335 int sl_followup; /* SAL followup */
|
|
|
336 int sl_collapse; /* SAL collapse_result */
|
|
|
337 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
338 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
339 #ifdef FEAT_MBYTE
|
|
|
340 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
341 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
342 #else
|
|
|
343 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
344 #endif
|
|
236
|
345 };
|
|
|
346
|
|
243
|
347 /* First language that is loaded, start of the linked list of loaded
|
|
|
348 * languages. */
|
|
236
|
349 static slang_T *first_lang = NULL;
|
|
|
350
|
|
323
|
351 /* Flags used in .spl file for soundsalike flags. */
|
|
|
352 #define SAL_F0LLOWUP 1
|
|
|
353 #define SAL_COLLAPSE 2
|
|
|
354 #define SAL_REM_ACCENTS 4
|
|
375
|
355 #define SAL_SOFO 8 /* SOFOFROM and SOFOTO instead of SAL */
|
|
323
|
356
|
|
236
|
357 /*
|
|
|
358 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
359 */
|
|
|
360 typedef struct langp_S
|
|
|
361 {
|
|
|
362 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
363 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
364 } langp_T;
|
|
|
365
|
|
|
366 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
367
|
|
307
|
368 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
369
|
|
|
370 /* Result values. Lower number is accepted over higher one. */
|
|
|
371 #define SP_BANNED -1
|
|
236
|
372 #define SP_OK 0
|
|
307
|
373 #define SP_RARE 1
|
|
|
374 #define SP_LOCAL 2
|
|
|
375 #define SP_BAD 3
|
|
236
|
376
|
|
366
|
377 #define VIMSPELLMAGIC "VIMspell08" /* string at start of Vim spell file */
|
|
236
|
378 #define VIMSPELLMAGICL 10
|
|
|
379
|
|
381
|
380 /* file used for "zG" and "zW" */
|
|
|
381 static char_u *temp_wordlist = NULL;
|
|
|
382
|
|
236
|
383 /*
|
|
323
|
384 * Information used when looking for suggestions.
|
|
|
385 */
|
|
|
386 typedef struct suginfo_S
|
|
|
387 {
|
|
|
388 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
389 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
390 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
391 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
392 char_u *su_badptr; /* start of bad word in line */
|
|
|
393 int su_badlen; /* length of detected bad word in line */
|
|
346
|
394 int su_badflags; /* caps flags for bad word */
|
|
323
|
395 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
396 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
397 hashtab_T su_banned; /* table with banned words */
|
|
|
398 } suginfo_T;
|
|
|
399
|
|
|
400 /* One word suggestion. Used in "si_ga". */
|
|
|
401 typedef struct suggest_S
|
|
|
402 {
|
|
|
403 char_u *st_word; /* suggested word, allocated string */
|
|
|
404 int st_orglen; /* length of replaced text */
|
|
|
405 int st_score; /* lower is better */
|
|
344
|
406 int st_altscore; /* used when st_score compares equal */
|
|
|
407 int st_salscore; /* st_score is for soundalike */
|
|
324
|
408 int st_had_bonus; /* bonus already included in score */
|
|
323
|
409 } suggest_T;
|
|
|
410
|
|
344
|
411 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
412
|
|
324
|
413 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
414 * called the score may change, thus we need to keep more than what is
|
|
|
415 * displayed. */
|
|
351
|
416 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
417
|
|
|
418 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
419 * of suggestions that are not going to be displayed. */
|
|
344
|
420 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
421
|
|
|
422 /* score for various changes */
|
|
344
|
423 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
424 #define SCORE_ICASE 52 /* slightly different case */
|
|
|
425 #define SCORE_REGION 70 /* word is for different region */
|
|
|
426 #define SCORE_RARE 180 /* rare word */
|
|
|
427 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
428 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
429 #define SCORE_REP 87 /* REP replacement */
|
|
|
430 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
431 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
324
|
432 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
433 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
324
|
434 #define SCORE_INS 96 /* insert a character */
|
|
358
|
435 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
366
|
436 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
437
|
|
372
|
438 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
439 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
440 * 350 allows for about three changes. */
|
|
344
|
441
|
|
|
442 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
443 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
444
|
|
|
445 /*
|
|
236
|
446 * Structure to store info for word matching.
|
|
|
447 */
|
|
|
448 typedef struct matchinf_S
|
|
|
449 {
|
|
|
450 langp_T *mi_lp; /* info for language and region */
|
|
243
|
451
|
|
|
452 /* pointers to original text to be checked */
|
|
236
|
453 char_u *mi_word; /* start of word being checked */
|
|
339
|
454 char_u *mi_end; /* end of matching word so far */
|
|
243
|
455 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
456 char_u *mi_cend; /* char after what was used for
|
|
|
457 mi_capflags */
|
|
243
|
458
|
|
|
459 /* case-folded text */
|
|
|
460 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
461 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
462
|
|
339
|
463 /* for when checking word after a prefix */
|
|
|
464 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
|
465 prefixID/condition */
|
|
|
466 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
467 int mi_prefixlen; /* byte length of prefix */
|
|
|
468
|
|
243
|
469 /* others */
|
|
236
|
470 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
471 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
472 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
473 } matchinf_T;
|
|
|
474
|
|
307
|
475 /*
|
|
|
476 * The tables used for recognizing word characters according to spelling.
|
|
|
477 * These are only used for the first 256 characters of 'encoding'.
|
|
|
478 */
|
|
|
479 typedef struct spelltab_S
|
|
|
480 {
|
|
|
481 char_u st_isw[256]; /* flags: is word char */
|
|
|
482 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
483 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
484 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
485 } spelltab_T;
|
|
|
486
|
|
|
487 static spelltab_T spelltab;
|
|
|
488 static int did_set_spelltab;
|
|
|
489
|
|
324
|
490 #define CF_WORD 0x01
|
|
|
491 #define CF_UPPER 0x02
|
|
307
|
492
|
|
|
493 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
494 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
495 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
496 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
497 #ifdef FEAT_MBYTE
|
|
|
498 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
499 #endif
|
|
339
|
500 static void write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
501
|
|
|
502 /*
|
|
358
|
503 * Return TRUE if "p" points to a word character. Like spell_iswordp() but
|
|
|
504 * without the special handling of a single quote.
|
|
307
|
505 * Checking for a word character is done very often, avoid the function call
|
|
|
506 * overhead.
|
|
|
507 */
|
|
|
508 #ifdef FEAT_MBYTE
|
|
|
509 # define SPELL_ISWORDP(p) ((has_mbyte && MB_BYTE2LEN(*(p)) > 1) \
|
|
|
510 ? (mb_get_class(p) >= 2) : spelltab.st_isw[*(p)])
|
|
|
511 #else
|
|
|
512 # define SPELL_ISWORDP(p) (spelltab.st_isw[*(p)])
|
|
|
513 #endif
|
|
|
514
|
|
323
|
515 /*
|
|
344
|
516 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
517 */
|
|
|
518 typedef enum
|
|
|
519 {
|
|
344
|
520 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
521 * ends); if badword ends there is a match, otherwise
|
|
|
522 * try splitting word. */
|
|
375
|
523 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
524 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
525 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
526 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
527 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
528 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
529 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
530 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
531 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
532 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
533 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
534 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
535 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
536 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
537 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
538 STATE_FINAL /* End of this node. */
|
|
|
539 } state_T;
|
|
|
540
|
|
|
541 /*
|
|
346
|
542 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
543 */
|
|
|
544 typedef struct trystate_S
|
|
|
545 {
|
|
330
|
546 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
547 int ts_score; /* score */
|
|
344
|
548 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
549 short ts_curi; /* index in list of child nodes */
|
|
|
550 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
551 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
552 char_u ts_twordlen; /* valid length of tword[] */
|
|
375
|
553 char_u ts_prefixdepth; /* stack depth for end of prefix or PREFIXTREE
|
|
|
554 * or NOPREFIX */
|
|
330
|
555 #ifdef FEAT_MBYTE
|
|
|
556 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
557 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
558 char_u ts_isdiff; /* DIFF_ values */
|
|
|
559 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
560 #endif
|
|
323
|
561 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
330
|
562 char_u ts_save_splitoff; /* su_splitoff saved here */
|
|
346
|
563 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
564 } trystate_T;
|
|
|
565
|
|
330
|
566 /* values for ts_isdiff */
|
|
|
567 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
568 #define DIFF_YES 1 /* different byte found */
|
|
|
569 #define DIFF_INSERT 2 /* inserting character */
|
|
|
570
|
|
375
|
571 /* special values ts_prefixdepth */
|
|
|
572 #define PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
573 #define NOPREFIX 0xff /* not using prefixes */
|
|
|
574
|
|
339
|
575 /* mode values for find_word */
|
|
|
576 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
577 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
578 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
579
|
|
236
|
580 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
581 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
582 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
583 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
351
|
584 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int prefid, char_u *word, slang_T *slang));
|
|
339
|
585 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
586 static int fold_more __ARGS((matchinf_T *mip));
|
|
323
|
587 static int spell_valid_case __ARGS((int origflags, int treeflags));
|
|
351
|
588 static int no_spell_checking __ARGS((void));
|
|
307
|
589 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
590 static char_u *spell_enc __ARGS((void));
|
|
|
591 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
592 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
376
|
593 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *errp));
|
|
381
|
594 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
595 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
596 #ifdef FEAT_MBYTE
|
|
|
597 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
598 #endif
|
|
339
|
599 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
600 static void clear_midword __ARGS((buf_T *buf));
|
|
|
601 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
602 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
603 static int captype __ARGS((char_u *word, char_u *end));
|
|
323
|
604 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
381
|
605 static int set_spell_charflags __ARGS((char_u *flags, char_u *upp));
|
|
307
|
606 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
607 static void write_spell_chartab __ARGS((FILE *fd));
|
|
|
608 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
358
|
609 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword));
|
|
372
|
610 #ifdef FEAT_EVAL
|
|
|
611 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
612 #endif
|
|
|
613 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
614 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
615 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
616 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
617 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
618 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
619 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
620 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
|
621 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
622 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
623 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
624 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
625 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
626 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
627 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
628 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
629 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
630 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
631 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
632 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
633 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
634 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
635 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
636 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
637 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
638 #ifdef FEAT_MBYTE
|
|
375
|
639 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
640 #endif
|
|
344
|
641 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
642 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
643 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
644 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
645
|
|
324
|
646 /*
|
|
|
647 * Use our own character-case definitions, because the current locale may
|
|
|
648 * differ from what the .spl file uses.
|
|
|
649 * These must not be called with negative number!
|
|
|
650 */
|
|
|
651 #ifndef FEAT_MBYTE
|
|
|
652 /* Non-multi-byte implementation. */
|
|
|
653 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
654 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
655 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
656 #else
|
|
|
657 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
658 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
659 * the "w" library function for characters above 255 if available. */
|
|
|
660 # ifdef HAVE_TOWLOWER
|
|
|
661 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
662 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
663 # else
|
|
|
664 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
665 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
666 # endif
|
|
|
667
|
|
|
668 # ifdef HAVE_TOWUPPER
|
|
|
669 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
670 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
671 # else
|
|
|
672 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
673 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
674 # endif
|
|
|
675
|
|
|
676 # ifdef HAVE_ISWUPPER
|
|
|
677 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
678 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
679 # else
|
|
|
680 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
681 : (c) < 256 ? spelltab.st_isu[c] : (c))
|
|
|
682 # endif
|
|
|
683 #endif
|
|
|
684
|
|
307
|
685
|
|
|
686 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
687 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
236
|
688
|
|
|
689 /*
|
|
|
690 * Main spell-checking function.
|
|
300
|
691 * "ptr" points to a character that could be the start of a word.
|
|
236
|
692 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
693 * or when it's OK it remains unchanged.
|
|
|
694 * This must only be called when 'spelllang' is not empty.
|
|
323
|
695 *
|
|
|
696 * "sug" is normally NULL. When looking for suggestions it points to
|
|
|
697 * suginfo_T. It's passed as a void pointer to keep the struct local.
|
|
|
698 *
|
|
236
|
699 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
700 * caller can skip over the word.
|
|
|
701 */
|
|
|
702 int
|
|
300
|
703 spell_check(wp, ptr, attrp)
|
|
236
|
704 win_T *wp; /* current window */
|
|
|
705 char_u *ptr;
|
|
|
706 int *attrp;
|
|
|
707 {
|
|
|
708 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
709 be passed to functions quickly. */
|
|
344
|
710 int nrlen = 0; /* found a number first */
|
|
236
|
711
|
|
307
|
712 /* A word never starts at a space or a control character. Return quickly
|
|
|
713 * then, skipping over the character. */
|
|
|
714 if (*ptr <= ' ')
|
|
|
715 return 1;
|
|
236
|
716
|
|
344
|
717 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
718 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
719 * "3GPP". */
|
|
300
|
720 if (*ptr >= '0' && *ptr <= '9')
|
|
|
721 {
|
|
316
|
722 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
723 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
724 else
|
|
|
725 {
|
|
344
|
726 mi.mi_end = skipdigits(ptr);
|
|
|
727 nrlen = mi.mi_end - ptr;
|
|
300
|
728 }
|
|
376
|
729 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
730 return (int)(mi.mi_end - ptr);
|
|
346
|
731
|
|
|
732 /* Try including the digits in the word. */
|
|
|
733 mi.mi_fend = ptr + nrlen;
|
|
|
734 }
|
|
|
735 else
|
|
|
736 mi.mi_fend = ptr;
|
|
|
737
|
|
|
738 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
739 mi.mi_word = ptr;
|
|
376
|
740 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
741 {
|
|
|
742 do
|
|
|
743 {
|
|
307
|
744 mb_ptr_adv(mi.mi_fend);
|
|
376
|
745 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
344
|
746 }
|
|
|
747
|
|
|
748 /* We always use the characters up to the next non-word character,
|
|
|
749 * also for bad words. */
|
|
|
750 mi.mi_end = mi.mi_fend;
|
|
|
751
|
|
|
752 /* Check caps type later. */
|
|
|
753 mi.mi_capflags = 0;
|
|
|
754 mi.mi_cend = NULL;
|
|
376
|
755 mi.mi_buf = wp->w_buffer;
|
|
344
|
756
|
|
|
757 /* Include one non-word character so that we can check for the
|
|
|
758 * word end. */
|
|
|
759 if (*mi.mi_fend != NUL)
|
|
|
760 mb_ptr_adv(mi.mi_fend);
|
|
|
761
|
|
|
762 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
763 MAXWLEN + 1);
|
|
|
764 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
765
|
|
|
766 /* The word is bad unless we recognize it. */
|
|
|
767 mi.mi_result = SP_BAD;
|
|
|
768
|
|
|
769 /*
|
|
|
770 * Loop over the languages specified in 'spelllang'.
|
|
|
771 * We check them all, because a matching word may be longer than an
|
|
|
772 * already found matching word.
|
|
|
773 */
|
|
|
774 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
775 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
776 {
|
|
|
777 /* Check for a matching word in case-folded words. */
|
|
|
778 find_word(&mi, FIND_FOLDWORD);
|
|
|
779
|
|
|
780 /* Check for a matching word in keep-case words. */
|
|
|
781 find_word(&mi, FIND_KEEPWORD);
|
|
|
782
|
|
|
783 /* Check for matching prefixes. */
|
|
|
784 find_prefix(&mi);
|
|
|
785 }
|
|
|
786
|
|
|
787 if (mi.mi_result != SP_OK)
|
|
|
788 {
|
|
346
|
789 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
790 * rare and local words, e.g., "3GPP". */
|
|
344
|
791 if (nrlen > 0)
|
|
346
|
792 {
|
|
|
793 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
794 return nrlen;
|
|
|
795 }
|
|
344
|
796
|
|
|
797 /* When we are at a non-word character there is no error, just
|
|
|
798 * skip over the character (try looking for a word after it). */
|
|
346
|
799 else if (!SPELL_ISWORDP(ptr))
|
|
243
|
800 {
|
|
344
|
801 #ifdef FEAT_MBYTE
|
|
|
802 if (has_mbyte)
|
|
|
803 return mb_ptr2len_check(ptr);
|
|
|
804 #endif
|
|
|
805 return 1;
|
|
300
|
806 }
|
|
243
|
807
|
|
344
|
808 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
809 *attrp = highlight_attr[HLF_SPB];
|
|
|
810 else if (mi.mi_result == SP_RARE)
|
|
|
811 *attrp = highlight_attr[HLF_SPR];
|
|
|
812 else
|
|
|
813 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
814 }
|
|
|
815
|
|
300
|
816 return (int)(mi.mi_end - ptr);
|
|
236
|
817 }
|
|
|
818
|
|
|
819 /*
|
|
300
|
820 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
821 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
822 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
823 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
824 * tree.
|
|
300
|
825 *
|
|
|
826 * For a match mip->mi_result is updated.
|
|
243
|
827 */
|
|
|
828 static void
|
|
339
|
829 find_word(mip, mode)
|
|
243
|
830 matchinf_T *mip;
|
|
339
|
831 int mode;
|
|
243
|
832 {
|
|
324
|
833 idx_T arridx = 0;
|
|
300
|
834 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
835 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
836 int endidxcnt = 0;
|
|
|
837 int len;
|
|
|
838 int wlen = 0;
|
|
|
839 int flen;
|
|
|
840 int c;
|
|
|
841 char_u *ptr;
|
|
324
|
842 idx_T lo, hi, m;
|
|
243
|
843 #ifdef FEAT_MBYTE
|
|
300
|
844 char_u *s;
|
|
339
|
845 char_u *p;
|
|
307
|
846 #endif
|
|
|
847 int res = SP_BAD;
|
|
300
|
848 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
849 unsigned flags;
|
|
|
850 char_u *byts;
|
|
324
|
851 idx_T *idxs;
|
|
339
|
852 int prefid;
|
|
|
853
|
|
|
854 if (mode == FIND_KEEPWORD)
|
|
236
|
855 {
|
|
300
|
856 /* Check for word with matching case in keep-case tree. */
|
|
|
857 ptr = mip->mi_word;
|
|
|
858 flen = 9999; /* no case folding, always enough bytes */
|
|
|
859 byts = slang->sl_kbyts;
|
|
|
860 idxs = slang->sl_kidxs;
|
|
236
|
861 }
|
|
|
862 else
|
|
|
863 {
|
|
300
|
864 /* Check for case-folded in case-folded tree. */
|
|
|
865 ptr = mip->mi_fword;
|
|
|
866 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
867 byts = slang->sl_fbyts;
|
|
|
868 idxs = slang->sl_fidxs;
|
|
339
|
869
|
|
|
870 if (mode == FIND_PREFIX)
|
|
|
871 {
|
|
|
872 /* Skip over the prefix. */
|
|
|
873 wlen = mip->mi_prefixlen;
|
|
|
874 flen -= mip->mi_prefixlen;
|
|
|
875 }
|
|
243
|
876 }
|
|
|
877
|
|
300
|
878 if (byts == NULL)
|
|
|
879 return; /* array is empty */
|
|
236
|
880
|
|
|
881 /*
|
|
307
|
882 * Repeat advancing in the tree until:
|
|
|
883 * - there is a byte that doesn't match,
|
|
|
884 * - we reach the end of the tree,
|
|
|
885 * - or we reach the end of the line.
|
|
236
|
886 */
|
|
300
|
887 for (;;)
|
|
236
|
888 {
|
|
346
|
889 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
890 flen = fold_more(mip);
|
|
300
|
891
|
|
|
892 len = byts[arridx++];
|
|
|
893
|
|
|
894 /* If the first possible byte is a zero the word could end here.
|
|
|
895 * Remember this index, we first check for the longest word. */
|
|
|
896 if (byts[arridx] == 0)
|
|
|
897 {
|
|
307
|
898 if (endidxcnt == MAXWLEN)
|
|
|
899 {
|
|
|
900 /* Must be a corrupted spell file. */
|
|
|
901 EMSG(_(e_format));
|
|
|
902 return;
|
|
|
903 }
|
|
300
|
904 endlen[endidxcnt] = wlen;
|
|
|
905 endidx[endidxcnt++] = arridx++;
|
|
|
906 --len;
|
|
|
907
|
|
|
908 /* Skip over the zeros, there can be several flag/region
|
|
|
909 * combinations. */
|
|
|
910 while (len > 0 && byts[arridx] == 0)
|
|
|
911 {
|
|
|
912 ++arridx;
|
|
|
913 --len;
|
|
|
914 }
|
|
|
915 if (len == 0)
|
|
|
916 break; /* no children, word must end here */
|
|
|
917 }
|
|
|
918
|
|
|
919 /* Stop looking at end of the line. */
|
|
|
920 if (ptr[wlen] == NUL)
|
|
|
921 break;
|
|
|
922
|
|
|
923 /* Perform a binary search in the list of accepted bytes. */
|
|
|
924 c = ptr[wlen];
|
|
346
|
925 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
926 c = ' ';
|
|
300
|
927 lo = arridx;
|
|
|
928 hi = arridx + len - 1;
|
|
|
929 while (lo < hi)
|
|
|
930 {
|
|
|
931 m = (lo + hi) / 2;
|
|
|
932 if (byts[m] > c)
|
|
|
933 hi = m - 1;
|
|
|
934 else if (byts[m] < c)
|
|
|
935 lo = m + 1;
|
|
|
936 else
|
|
|
937 {
|
|
|
938 lo = hi = m;
|
|
|
939 break;
|
|
236
|
940 }
|
|
|
941 }
|
|
300
|
942
|
|
|
943 /* Stop if there is no matching byte. */
|
|
|
944 if (hi < lo || byts[lo] != c)
|
|
|
945 break;
|
|
|
946
|
|
|
947 /* Continue at the child (if there is one). */
|
|
|
948 arridx = idxs[lo];
|
|
|
949 ++wlen;
|
|
|
950 --flen;
|
|
346
|
951
|
|
|
952 /* One space in the good word may stand for several spaces in the
|
|
|
953 * checked word. */
|
|
|
954 if (c == ' ')
|
|
|
955 {
|
|
|
956 for (;;)
|
|
|
957 {
|
|
|
958 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
959 flen = fold_more(mip);
|
|
|
960 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
961 break;
|
|
|
962 ++wlen;
|
|
|
963 --flen;
|
|
|
964 }
|
|
|
965 }
|
|
236
|
966 }
|
|
|
967
|
|
300
|
968 /*
|
|
|
969 * Verify that one of the possible endings is valid. Try the longest
|
|
|
970 * first.
|
|
|
971 */
|
|
|
972 while (endidxcnt > 0)
|
|
|
973 {
|
|
|
974 --endidxcnt;
|
|
|
975 arridx = endidx[endidxcnt];
|
|
|
976 wlen = endlen[endidxcnt];
|
|
236
|
977
|
|
300
|
978 #ifdef FEAT_MBYTE
|
|
|
979 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
980 continue; /* not at first byte of character */
|
|
|
981 #endif
|
|
376
|
982 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
300
|
983 continue; /* next char is a word character */
|
|
|
984
|
|
|
985 #ifdef FEAT_MBYTE
|
|
339
|
986 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
987 {
|
|
|
988 /* Compute byte length in original word, length may change
|
|
339
|
989 * when folding case. This can be slow, take a shortcut when the
|
|
|
990 * case-folded word is equal to the keep-case word. */
|
|
300
|
991 p = mip->mi_word;
|
|
339
|
992 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
993 {
|
|
|
994 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
995 mb_ptr_adv(p);
|
|
|
996 wlen = p - mip->mi_word;
|
|
|
997 }
|
|
300
|
998 }
|
|
|
999 #endif
|
|
236
|
1000
|
|
339
|
1001 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1002 * prefix ID.
|
|
|
1003 * Repeat this if there are more flags/region alternatives until there
|
|
|
1004 * is a match. */
|
|
|
1005 res = SP_BAD;
|
|
|
1006 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1007 --len, ++arridx)
|
|
300
|
1008 {
|
|
|
1009 flags = idxs[arridx];
|
|
324
|
1010
|
|
339
|
1011 /* For the fold-case tree check that the case of the checked word
|
|
|
1012 * matches with what the word in the tree requires.
|
|
|
1013 * For keep-case tree the case is always right. For prefixes we
|
|
|
1014 * don't bother to check. */
|
|
|
1015 if (mode == FIND_FOLDWORD)
|
|
300
|
1016 {
|
|
|
1017 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1018 {
|
|
323
|
1019 /* mi_capflags was set for a different word length, need
|
|
|
1020 * to do it again. */
|
|
300
|
1021 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1022 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1023 }
|
|
|
1024
|
|
346
|
1025 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1026 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1027 continue;
|
|
300
|
1028 }
|
|
236
|
1029
|
|
339
|
1030 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1031 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1032 * mip->mi_prefarridx that find_prefix() filled. */
|
|
339
|
1033 if (mode == FIND_PREFIX)
|
|
300
|
1034 {
|
|
339
|
1035 /* The prefix ID is stored two bytes above the flags. */
|
|
|
1036 prefid = (unsigned)flags >> 16;
|
|
366
|
1037 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
351
|
1038 prefid, mip->mi_fword + mip->mi_prefixlen,
|
|
366
|
1039 slang);
|
|
|
1040 if (c == 0)
|
|
339
|
1041 continue;
|
|
366
|
1042
|
|
|
1043 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1044 if (c & WF_RAREPFX)
|
|
|
1045 flags |= WF_RARE;
|
|
339
|
1046 }
|
|
|
1047
|
|
|
1048 if (flags & WF_BANNED)
|
|
|
1049 res = SP_BANNED;
|
|
|
1050 else if (flags & WF_REGION)
|
|
|
1051 {
|
|
|
1052 /* Check region. */
|
|
|
1053 if ((mip->mi_lp->lp_region & (flags >> 8)) != 0)
|
|
300
|
1054 res = SP_OK;
|
|
339
|
1055 else
|
|
|
1056 res = SP_LOCAL;
|
|
300
|
1057 }
|
|
339
|
1058 else if (flags & WF_RARE)
|
|
|
1059 res = SP_RARE;
|
|
307
|
1060 else
|
|
339
|
1061 res = SP_OK;
|
|
|
1062
|
|
|
1063 /* Always use the longest match and the best result. */
|
|
|
1064 if (mip->mi_result > res)
|
|
|
1065 {
|
|
|
1066 mip->mi_result = res;
|
|
|
1067 mip->mi_end = mip->mi_word + wlen;
|
|
|
1068 }
|
|
351
|
1069 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1070 mip->mi_end = mip->mi_word + wlen;
|
|
|
1071
|
|
|
1072 if (res == SP_OK)
|
|
|
1073 break;
|
|
300
|
1074 }
|
|
|
1075
|
|
307
|
1076 if (res == SP_OK)
|
|
300
|
1077 break;
|
|
|
1078 }
|
|
236
|
1079 }
|
|
|
1080
|
|
323
|
1081 /*
|
|
366
|
1082 * Return non-zero if the prefix indicated by "mip->mi_prefarridx" matches
|
|
|
1083 * with the prefix ID "prefid" for the word "word".
|
|
|
1084 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1085 */
|
|
|
1086 static int
|
|
|
1087 valid_word_prefix(totprefcnt, arridx, prefid, word, slang)
|
|
|
1088 int totprefcnt; /* nr of prefix IDs */
|
|
|
1089 int arridx; /* idx in sl_pidxs[] */
|
|
|
1090 int prefid;
|
|
|
1091 char_u *word;
|
|
|
1092 slang_T *slang;
|
|
|
1093 {
|
|
|
1094 int prefcnt;
|
|
|
1095 int pidx;
|
|
|
1096 regprog_T *rp;
|
|
|
1097 regmatch_T regmatch;
|
|
|
1098
|
|
|
1099 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1100 {
|
|
|
1101 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1102
|
|
|
1103 /* Check the prefix ID. */
|
|
|
1104 if (prefid != (pidx & 0xff))
|
|
|
1105 continue;
|
|
|
1106
|
|
|
1107 /* Check the condition, if there is one. The condition index is
|
|
366
|
1108 * stored in the two bytes above the prefix ID byte. */
|
|
|
1109 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1110 if (rp != NULL)
|
|
|
1111 {
|
|
|
1112 regmatch.regprog = rp;
|
|
|
1113 regmatch.rm_ic = FALSE;
|
|
|
1114 if (!vim_regexec(®match, word, 0))
|
|
|
1115 continue;
|
|
|
1116 }
|
|
|
1117
|
|
366
|
1118 /* It's a match! Return the WF_RAREPFX flag. */
|
|
|
1119 return pidx;
|
|
|
1120 }
|
|
|
1121 return 0;
|
|
351
|
1122 }
|
|
|
1123
|
|
|
1124 /*
|
|
339
|
1125 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1126 * If it does, then check the following word.
|
|
|
1127 *
|
|
|
1128 * For a match mip->mi_result is updated.
|
|
|
1129 */
|
|
|
1130 static void
|
|
|
1131 find_prefix(mip)
|
|
|
1132 matchinf_T *mip;
|
|
|
1133 {
|
|
|
1134 idx_T arridx = 0;
|
|
|
1135 int len;
|
|
|
1136 int wlen = 0;
|
|
|
1137 int flen;
|
|
|
1138 int c;
|
|
|
1139 char_u *ptr;
|
|
|
1140 idx_T lo, hi, m;
|
|
|
1141 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1142 char_u *byts;
|
|
|
1143 idx_T *idxs;
|
|
|
1144
|
|
375
|
1145 byts = slang->sl_pbyts;
|
|
|
1146 if (byts == NULL)
|
|
|
1147 return; /* array is empty */
|
|
|
1148
|
|
339
|
1149 /* We use the case-folded word here, since prefixes are always
|
|
|
1150 * case-folded. */
|
|
|
1151 ptr = mip->mi_fword;
|
|
|
1152 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1153 idxs = slang->sl_pidxs;
|
|
|
1154
|
|
|
1155 /*
|
|
|
1156 * Repeat advancing in the tree until:
|
|
|
1157 * - there is a byte that doesn't match,
|
|
|
1158 * - we reach the end of the tree,
|
|
|
1159 * - or we reach the end of the line.
|
|
|
1160 */
|
|
|
1161 for (;;)
|
|
|
1162 {
|
|
|
1163 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1164 flen = fold_more(mip);
|
|
|
1165
|
|
|
1166 len = byts[arridx++];
|
|
|
1167
|
|
|
1168 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1169 * Check if the following word matches and supports the prefix. */
|
|
|
1170 if (byts[arridx] == 0)
|
|
|
1171 {
|
|
|
1172 /* There can be several prefixes with different conditions. We
|
|
|
1173 * try them all, since we don't know which one will give the
|
|
|
1174 * longest match. The word is the same each time, pass the list
|
|
|
1175 * of possible prefixes to find_word(). */
|
|
|
1176 mip->mi_prefarridx = arridx;
|
|
|
1177 mip->mi_prefcnt = len;
|
|
|
1178 while (len > 0 && byts[arridx] == 0)
|
|
|
1179 {
|
|
|
1180 ++arridx;
|
|
|
1181 --len;
|
|
|
1182 }
|
|
|
1183 mip->mi_prefcnt -= len;
|
|
|
1184
|
|
|
1185 /* Find the word that comes after the prefix. */
|
|
|
1186 mip->mi_prefixlen = wlen;
|
|
|
1187 find_word(mip, FIND_PREFIX);
|
|
|
1188
|
|
|
1189
|
|
|
1190 if (len == 0)
|
|
|
1191 break; /* no children, word must end here */
|
|
|
1192 }
|
|
|
1193
|
|
|
1194 /* Stop looking at end of the line. */
|
|
|
1195 if (ptr[wlen] == NUL)
|
|
|
1196 break;
|
|
|
1197
|
|
|
1198 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1199 c = ptr[wlen];
|
|
|
1200 lo = arridx;
|
|
|
1201 hi = arridx + len - 1;
|
|
|
1202 while (lo < hi)
|
|
|
1203 {
|
|
|
1204 m = (lo + hi) / 2;
|
|
|
1205 if (byts[m] > c)
|
|
|
1206 hi = m - 1;
|
|
|
1207 else if (byts[m] < c)
|
|
|
1208 lo = m + 1;
|
|
|
1209 else
|
|
|
1210 {
|
|
|
1211 lo = hi = m;
|
|
|
1212 break;
|
|
|
1213 }
|
|
|
1214 }
|
|
|
1215
|
|
|
1216 /* Stop if there is no matching byte. */
|
|
|
1217 if (hi < lo || byts[lo] != c)
|
|
|
1218 break;
|
|
|
1219
|
|
|
1220 /* Continue at the child (if there is one). */
|
|
|
1221 arridx = idxs[lo];
|
|
|
1222 ++wlen;
|
|
|
1223 --flen;
|
|
|
1224 }
|
|
|
1225 }
|
|
|
1226
|
|
|
1227 /*
|
|
|
1228 * Need to fold at least one more character. Do until next non-word character
|
|
|
1229 * for efficiency.
|
|
|
1230 * Return the length of the folded chars in bytes.
|
|
|
1231 */
|
|
|
1232 static int
|
|
|
1233 fold_more(mip)
|
|
|
1234 matchinf_T *mip;
|
|
|
1235 {
|
|
|
1236 int flen;
|
|
|
1237 char_u *p;
|
|
|
1238
|
|
|
1239 p = mip->mi_fend;
|
|
|
1240 do
|
|
|
1241 {
|
|
|
1242 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1243 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1244
|
|
|
1245 /* Include the non-word character so that we can check for the
|
|
|
1246 * word end. */
|
|
|
1247 if (*mip->mi_fend != NUL)
|
|
|
1248 mb_ptr_adv(mip->mi_fend);
|
|
|
1249
|
|
|
1250 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1251 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1252 MAXWLEN - mip->mi_fwordlen);
|
|
|
1253 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1254 mip->mi_fwordlen += flen;
|
|
|
1255 return flen;
|
|
|
1256 }
|
|
|
1257
|
|
|
1258 /*
|
|
323
|
1259 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1260 * case.
|
|
|
1261 */
|
|
|
1262 static int
|
|
|
1263 spell_valid_case(origflags, treeflags)
|
|
|
1264 int origflags; /* flags for the checked word. */
|
|
|
1265 int treeflags; /* flags for the word in the spell tree */
|
|
|
1266 {
|
|
|
1267 return (origflags == WF_ALLCAP
|
|
|
1268 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
|
1269 && ((treeflags & WF_ONECAP) == 0 || origflags == WF_ONECAP)));
|
|
|
1270 }
|
|
|
1271
|
|
351
|
1272 /*
|
|
|
1273 * Return TRUE if spell checking is not enabled.
|
|
|
1274 */
|
|
|
1275 static int
|
|
|
1276 no_spell_checking()
|
|
|
1277 {
|
|
|
1278 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1279 {
|
|
|
1280 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1281 return TRUE;
|
|
|
1282 }
|
|
|
1283 return FALSE;
|
|
|
1284 }
|
|
300
|
1285
|
|
236
|
1286 /*
|
|
|
1287 * Move to next spell error.
|
|
323
|
1288 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
236
|
1289 * Return OK if found, FAIL otherwise.
|
|
|
1290 */
|
|
|
1291 int
|
|
323
|
1292 spell_move_to(dir, allwords, curline)
|
|
236
|
1293 int dir; /* FORWARD or BACKWARD */
|
|
|
1294 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1295 int curline;
|
|
236
|
1296 {
|
|
249
|
1297 linenr_T lnum;
|
|
|
1298 pos_T found_pos;
|
|
236
|
1299 char_u *line;
|
|
|
1300 char_u *p;
|
|
346
|
1301 char_u *endp;
|
|
|
1302 int attr;
|
|
236
|
1303 int len;
|
|
249
|
1304 int has_syntax = syntax_present(curbuf);
|
|
|
1305 int col;
|
|
|
1306 int can_spell;
|
|
346
|
1307 char_u *buf = NULL;
|
|
|
1308 int buflen = 0;
|
|
|
1309 int skip = 0;
|
|
236
|
1310
|
|
351
|
1311 if (no_spell_checking())
|
|
236
|
1312 return FAIL;
|
|
|
1313
|
|
249
|
1314 /*
|
|
|
1315 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1316 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1317 *
|
|
|
1318 * When searching backwards, we continue in the line to find the last
|
|
|
1319 * bad word (in the cursor line: before the cursor).
|
|
346
|
1320 *
|
|
|
1321 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1322 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1323 * though...
|
|
249
|
1324 */
|
|
|
1325 lnum = curwin->w_cursor.lnum;
|
|
|
1326 found_pos.lnum = 0;
|
|
236
|
1327
|
|
|
1328 while (!got_int)
|
|
|
1329 {
|
|
249
|
1330 line = ml_get(lnum);
|
|
346
|
1331
|
|
|
1332 len = STRLEN(line);
|
|
|
1333 if (buflen < len + MAXWLEN + 2)
|
|
|
1334 {
|
|
|
1335 vim_free(buf);
|
|
|
1336 buflen = len + MAXWLEN + 2;
|
|
|
1337 buf = alloc(buflen);
|
|
|
1338 if (buf == NULL)
|
|
|
1339 break;
|
|
|
1340 }
|
|
|
1341
|
|
|
1342 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1343 * possible. */
|
|
|
1344 STRCPY(buf, line);
|
|
|
1345 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1346 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1347
|
|
|
1348 p = buf + skip;
|
|
|
1349 endp = buf + len;
|
|
|
1350 while (p < endp)
|
|
236
|
1351 {
|
|
300
|
1352 /* When searching backward don't search after the cursor. */
|
|
|
1353 if (dir == BACKWARD
|
|
|
1354 && lnum == curwin->w_cursor.lnum
|
|
346
|
1355 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1356 break;
|
|
249
|
1357
|
|
300
|
1358 /* start of word */
|
|
346
|
1359 attr = 0;
|
|
300
|
1360 len = spell_check(curwin, p, &attr);
|
|
249
|
1361
|
|
300
|
1362 if (attr != 0)
|
|
|
1363 {
|
|
|
1364 /* We found a bad word. Check the attribute. */
|
|
|
1365 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1366 {
|
|
300
|
1367 /* When searching forward only accept a bad word after
|
|
|
1368 * the cursor. */
|
|
|
1369 if (dir == BACKWARD
|
|
|
1370 || lnum > curwin->w_cursor.lnum
|
|
|
1371 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1372 && (colnr_T)(curline ? p - buf + len
|
|
|
1373 : p - buf)
|
|
300
|
1374 > curwin->w_cursor.col))
|
|
236
|
1375 {
|
|
300
|
1376 if (has_syntax)
|
|
249
|
1377 {
|
|
346
|
1378 col = p - buf;
|
|
300
|
1379 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1380 FALSE, &can_spell);
|
|
|
1381 }
|
|
|
1382 else
|
|
|
1383 can_spell = TRUE;
|
|
249
|
1384
|
|
300
|
1385 if (can_spell)
|
|
|
1386 {
|
|
|
1387 found_pos.lnum = lnum;
|
|
346
|
1388 found_pos.col = p - buf;
|
|
249
|
1389 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1390 found_pos.coladd = 0;
|
|
249
|
1391 #endif
|
|
300
|
1392 if (dir == FORWARD)
|
|
|
1393 {
|
|
|
1394 /* No need to search further. */
|
|
|
1395 curwin->w_cursor = found_pos;
|
|
346
|
1396 vim_free(buf);
|
|
300
|
1397 return OK;
|
|
249
|
1398 }
|
|
|
1399 }
|
|
236
|
1400 }
|
|
|
1401 }
|
|
|
1402 }
|
|
|
1403
|
|
300
|
1404 /* advance to character after the word */
|
|
|
1405 p += len;
|
|
236
|
1406 }
|
|
|
1407
|
|
323
|
1408 if (curline)
|
|
346
|
1409 break; /* only check cursor line */
|
|
323
|
1410
|
|
236
|
1411 /* Advance to next line. */
|
|
249
|
1412 if (dir == BACKWARD)
|
|
|
1413 {
|
|
|
1414 if (found_pos.lnum != 0)
|
|
|
1415 {
|
|
|
1416 /* Use the last match in the line. */
|
|
|
1417 curwin->w_cursor = found_pos;
|
|
346
|
1418 vim_free(buf);
|
|
249
|
1419 return OK;
|
|
|
1420 }
|
|
|
1421 if (lnum == 1)
|
|
346
|
1422 break;
|
|
249
|
1423 --lnum;
|
|
|
1424 }
|
|
|
1425 else
|
|
|
1426 {
|
|
|
1427 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1428 break;
|
|
249
|
1429 ++lnum;
|
|
346
|
1430
|
|
|
1431 /* Skip the characters at the start of the next line that were
|
|
|
1432 * included in a match crossing line boundaries. */
|
|
|
1433 if (attr == 0)
|
|
|
1434 skip = p - endp;
|
|
|
1435 else
|
|
|
1436 skip = 0;
|
|
249
|
1437 }
|
|
236
|
1438
|
|
|
1439 line_breakcheck();
|
|
|
1440 }
|
|
|
1441
|
|
346
|
1442 vim_free(buf);
|
|
|
1443 return FAIL;
|
|
|
1444 }
|
|
|
1445
|
|
|
1446 /*
|
|
|
1447 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1448 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1449 */
|
|
|
1450 void
|
|
|
1451 spell_cat_line(buf, line, maxlen)
|
|
|
1452 char_u *buf;
|
|
|
1453 char_u *line;
|
|
|
1454 int maxlen;
|
|
|
1455 {
|
|
|
1456 char_u *p;
|
|
|
1457 int n;
|
|
|
1458
|
|
|
1459 p = skipwhite(line);
|
|
|
1460 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1461 p = skipwhite(p + 1);
|
|
|
1462
|
|
|
1463 if (*p != NUL)
|
|
|
1464 {
|
|
|
1465 *buf = ' ';
|
|
376
|
1466 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1467 n = p - line;
|
|
|
1468 if (n >= maxlen)
|
|
|
1469 n = maxlen - 1;
|
|
|
1470 vim_memset(buf + 1, ' ', n);
|
|
|
1471 }
|
|
236
|
1472 }
|
|
|
1473
|
|
|
1474 /*
|
|
307
|
1475 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1476 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1477 */
|
|
307
|
1478 static void
|
|
236
|
1479 spell_load_lang(lang)
|
|
|
1480 char_u *lang;
|
|
|
1481 {
|
|
310
|
1482 char_u fname_enc[85];
|
|
236
|
1483 int r;
|
|
307
|
1484 char_u langcp[MAXWLEN + 1];
|
|
|
1485
|
|
310
|
1486 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1487 * It's truncated when an error is detected. */
|
|
|
1488 STRCPY(langcp, lang);
|
|
|
1489
|
|
310
|
1490 /*
|
|
|
1491 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1492 */
|
|
|
1493 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1494 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1495 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1496
|
|
|
1497 if (r == FAIL && *langcp != NUL)
|
|
|
1498 {
|
|
|
1499 /* Try loading the ASCII version. */
|
|
310
|
1500 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1501 "spell/%s.ascii.spl", lang);
|
|
310
|
1502 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1503 }
|
|
|
1504
|
|
|
1505 if (r == FAIL)
|
|
|
1506 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
|
236
|
1507 fname_enc + 6);
|
|
310
|
1508 else if (*langcp != NUL)
|
|
|
1509 {
|
|
|
1510 /* Load all the additions. */
|
|
|
1511 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1512 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1513 }
|
|
|
1514 }
|
|
|
1515
|
|
|
1516 /*
|
|
|
1517 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1518 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1519 */
|
|
|
1520 static char_u *
|
|
|
1521 spell_enc()
|
|
|
1522 {
|
|
|
1523
|
|
|
1524 #ifdef FEAT_MBYTE
|
|
|
1525 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1526 return p_enc;
|
|
|
1527 #endif
|
|
|
1528 return (char_u *)"latin1";
|
|
236
|
1529 }
|
|
|
1530
|
|
|
1531 /*
|
|
|
1532 * Allocate a new slang_T.
|
|
|
1533 * Caller must fill "sl_next".
|
|
|
1534 */
|
|
|
1535 static slang_T *
|
|
|
1536 slang_alloc(lang)
|
|
|
1537 char_u *lang;
|
|
|
1538 {
|
|
|
1539 slang_T *lp;
|
|
|
1540
|
|
300
|
1541 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1542 if (lp != NULL)
|
|
|
1543 {
|
|
|
1544 lp->sl_name = vim_strsave(lang);
|
|
323
|
1545 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
344
|
1546 ga_init2(&lp->sl_sal, sizeof(salitem_T), 10);
|
|
236
|
1547 }
|
|
|
1548 return lp;
|
|
|
1549 }
|
|
|
1550
|
|
|
1551 /*
|
|
|
1552 * Free the contents of an slang_T and the structure itself.
|
|
|
1553 */
|
|
|
1554 static void
|
|
|
1555 slang_free(lp)
|
|
|
1556 slang_T *lp;
|
|
|
1557 {
|
|
|
1558 vim_free(lp->sl_name);
|
|
310
|
1559 vim_free(lp->sl_fname);
|
|
|
1560 slang_clear(lp);
|
|
|
1561 vim_free(lp);
|
|
|
1562 }
|
|
|
1563
|
|
|
1564 /*
|
|
|
1565 * Clear an slang_T so that the file can be reloaded.
|
|
|
1566 */
|
|
|
1567 static void
|
|
|
1568 slang_clear(lp)
|
|
|
1569 slang_T *lp;
|
|
|
1570 {
|
|
339
|
1571 garray_T *gap;
|
|
|
1572 fromto_T *ftp;
|
|
344
|
1573 salitem_T *smp;
|
|
339
|
1574 int i;
|
|
323
|
1575
|
|
300
|
1576 vim_free(lp->sl_fbyts);
|
|
310
|
1577 lp->sl_fbyts = NULL;
|
|
300
|
1578 vim_free(lp->sl_kbyts);
|
|
310
|
1579 lp->sl_kbyts = NULL;
|
|
339
|
1580 vim_free(lp->sl_pbyts);
|
|
|
1581 lp->sl_pbyts = NULL;
|
|
|
1582
|
|
300
|
1583 vim_free(lp->sl_fidxs);
|
|
310
|
1584 lp->sl_fidxs = NULL;
|
|
300
|
1585 vim_free(lp->sl_kidxs);
|
|
310
|
1586 lp->sl_kidxs = NULL;
|
|
339
|
1587 vim_free(lp->sl_pidxs);
|
|
|
1588 lp->sl_pidxs = NULL;
|
|
323
|
1589
|
|
344
|
1590 gap = &lp->sl_rep;
|
|
|
1591 while (gap->ga_len > 0)
|
|
323
|
1592 {
|
|
344
|
1593 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1594 vim_free(ftp->ft_from);
|
|
|
1595 vim_free(ftp->ft_to);
|
|
323
|
1596 }
|
|
344
|
1597 ga_clear(gap);
|
|
|
1598
|
|
|
1599 gap = &lp->sl_sal;
|
|
375
|
1600 if (lp->sl_sofo)
|
|
376
|
1601 {
|
|
|
1602 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1603 if (gap->ga_data != NULL)
|
|
|
1604 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1605 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1606 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1607 }
|
|
375
|
1608 else
|
|
|
1609 /* SAL items: free salitem_T items */
|
|
|
1610 while (gap->ga_len > 0)
|
|
|
1611 {
|
|
|
1612 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1613 vim_free(smp->sm_lead);
|
|
|
1614 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1615 vim_free(smp->sm_to);
|
|
|
1616 #ifdef FEAT_MBYTE
|
|
|
1617 vim_free(smp->sm_lead_w);
|
|
|
1618 vim_free(smp->sm_oneof_w);
|
|
|
1619 vim_free(smp->sm_to_w);
|
|
|
1620 #endif
|
|
|
1621 }
|
|
344
|
1622 ga_clear(gap);
|
|
323
|
1623
|
|
339
|
1624 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1625 vim_free(lp->sl_prefprog[i]);
|
|
376
|
1626 lp->sl_prefixcnt = 0;
|
|
339
|
1627 vim_free(lp->sl_prefprog);
|
|
376
|
1628 lp->sl_prefprog = NULL;
|
|
|
1629
|
|
|
1630 vim_free(lp->sl_midword);
|
|
|
1631 lp->sl_midword = NULL;
|
|
339
|
1632
|
|
330
|
1633 #ifdef FEAT_MBYTE
|
|
|
1634 {
|
|
|
1635 int todo = lp->sl_map_hash.ht_used;
|
|
|
1636 hashitem_T *hi;
|
|
|
1637
|
|
|
1638 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1639 if (!HASHITEM_EMPTY(hi))
|
|
|
1640 {
|
|
|
1641 --todo;
|
|
|
1642 vim_free(hi->hi_key);
|
|
|
1643 }
|
|
|
1644 }
|
|
|
1645 hash_clear(&lp->sl_map_hash);
|
|
|
1646 #endif
|
|
236
|
1647 }
|
|
|
1648
|
|
|
1649 /*
|
|
307
|
1650 * Load one spell file and store the info into a slang_T.
|
|
236
|
1651 * Invoked through do_in_runtimepath().
|
|
|
1652 */
|
|
|
1653 static void
|
|
310
|
1654 spell_load_cb(fname, cookie)
|
|
236
|
1655 char_u *fname;
|
|
307
|
1656 void *cookie; /* points to the language name */
|
|
236
|
1657 {
|
|
323
|
1658 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1659 }
|
|
|
1660
|
|
|
1661 /*
|
|
|
1662 * Load one spell file and store the info into a slang_T.
|
|
|
1663 *
|
|
|
1664 * This is invoked in two ways:
|
|
|
1665 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
1666 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
1667 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
1668 * points to the existing slang_T.
|
|
323
|
1669 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
1670 */
|
|
323
|
1671 static slang_T *
|
|
|
1672 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
1673 char_u *fname;
|
|
|
1674 char_u *lang;
|
|
|
1675 slang_T *old_lp;
|
|
323
|
1676 int silent; /* no error if file doesn't exist */
|
|
310
|
1677 {
|
|
236
|
1678 FILE *fd;
|
|
|
1679 char_u buf[MAXWLEN + 1];
|
|
|
1680 char_u *p;
|
|
339
|
1681 char_u *bp;
|
|
|
1682 idx_T *ip;
|
|
236
|
1683 int i;
|
|
339
|
1684 int n;
|
|
300
|
1685 int len;
|
|
236
|
1686 int round;
|
|
|
1687 char_u *save_sourcing_name = sourcing_name;
|
|
|
1688 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
255
|
1689 int cnt, ccnt;
|
|
|
1690 char_u *fol;
|
|
307
|
1691 slang_T *lp = NULL;
|
|
323
|
1692 garray_T *gap;
|
|
|
1693 fromto_T *ftp;
|
|
344
|
1694 salitem_T *smp;
|
|
323
|
1695 short *first;
|
|
324
|
1696 idx_T idx;
|
|
344
|
1697 int c = 0;
|
|
236
|
1698
|
|
310
|
1699 fd = mch_fopen((char *)fname, "r");
|
|
236
|
1700 if (fd == NULL)
|
|
|
1701 {
|
|
323
|
1702 if (!silent)
|
|
|
1703 EMSG2(_(e_notopen), fname);
|
|
|
1704 else if (p_verbose > 2)
|
|
|
1705 {
|
|
|
1706 verbose_enter();
|
|
|
1707 smsg((char_u *)e_notopen, fname);
|
|
|
1708 verbose_leave();
|
|
|
1709 }
|
|
255
|
1710 goto endFAIL;
|
|
236
|
1711 }
|
|
310
|
1712 if (p_verbose > 2)
|
|
|
1713 {
|
|
|
1714 verbose_enter();
|
|
|
1715 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
1716 verbose_leave();
|
|
|
1717 }
|
|
|
1718
|
|
|
1719 if (old_lp == NULL)
|
|
|
1720 {
|
|
|
1721 lp = slang_alloc(lang);
|
|
|
1722 if (lp == NULL)
|
|
|
1723 goto endFAIL;
|
|
|
1724
|
|
|
1725 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
1726 lp->sl_fname = vim_strsave(fname);
|
|
|
1727 if (lp->sl_fname == NULL)
|
|
|
1728 goto endFAIL;
|
|
|
1729
|
|
|
1730 /* Check for .add.spl. */
|
|
|
1731 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
1732 }
|
|
|
1733 else
|
|
|
1734 lp = old_lp;
|
|
307
|
1735
|
|
236
|
1736 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
1737 sourcing_name = fname;
|
|
|
1738 sourcing_lnum = 0;
|
|
|
1739
|
|
339
|
1740 /* <HEADER>: <fileID>
|
|
|
1741 * <regioncnt> <regionname> ...
|
|
|
1742 * <charflagslen> <charflags>
|
|
|
1743 * <fcharslen> <fchars>
|
|
366
|
1744 * <midwordlen> <midword>
|
|
339
|
1745 * <prefcondcnt> <prefcond> ...
|
|
|
1746 */
|
|
236
|
1747 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
1748 buf[i] = getc(fd); /* <fileID> */
|
|
|
1749 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
1750 {
|
|
|
1751 EMSG(_("E757: Wrong file ID in spell file"));
|
|
255
|
1752 goto endFAIL;
|
|
236
|
1753 }
|
|
|
1754
|
|
|
1755 cnt = getc(fd); /* <regioncnt> */
|
|
255
|
1756 if (cnt < 0)
|
|
236
|
1757 {
|
|
|
1758 truncerr:
|
|
381
|
1759 EMSG(_(e_spell_trunc));
|
|
255
|
1760 goto endFAIL;
|
|
236
|
1761 }
|
|
|
1762 if (cnt > 8)
|
|
|
1763 {
|
|
|
1764 formerr:
|
|
307
|
1765 EMSG(_(e_format));
|
|
255
|
1766 goto endFAIL;
|
|
236
|
1767 }
|
|
|
1768 for (i = 0; i < cnt; ++i)
|
|
|
1769 {
|
|
|
1770 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
|
1771 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
|
1772 }
|
|
|
1773 lp->sl_regions[cnt * 2] = NUL;
|
|
|
1774
|
|
381
|
1775 /* <charflagslen> <charflags> */
|
|
|
1776 p = read_cnt_string(fd, 1, &cnt);
|
|
|
1777 if (cnt == FAIL)
|
|
|
1778 goto endFAIL;
|
|
|
1779
|
|
|
1780 /* <fcharslen> <fchars> */
|
|
|
1781 fol = read_cnt_string(fd, 2, &cnt);
|
|
|
1782 if (cnt == FAIL)
|
|
|
1783 {
|
|
300
|
1784 vim_free(p);
|
|
381
|
1785 goto endFAIL;
|
|
|
1786 }
|
|
|
1787
|
|
|
1788 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
1789 if (p != NULL && fol != NULL)
|
|
|
1790 i = set_spell_charflags(p, fol);
|
|
|
1791
|
|
|
1792 vim_free(p);
|
|
|
1793 vim_free(fol);
|
|
|
1794
|
|
|
1795 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
1796 if ((p == NULL) != (fol == NULL))
|
|
|
1797 goto formerr;
|
|
255
|
1798
|
|
366
|
1799 /* <midwordlen> <midword> */
|
|
376
|
1800 lp->sl_midword = read_cnt_string(fd, 2, &cnt);
|
|
381
|
1801 if (cnt == FAIL)
|
|
376
|
1802 goto endFAIL;
|
|
366
|
1803
|
|
339
|
1804 /* <prefcondcnt> <prefcond> ... */
|
|
|
1805 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
1806 if (cnt > 0)
|
|
|
1807 {
|
|
|
1808 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
1809 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
1810 if (lp->sl_prefprog == NULL)
|
|
|
1811 goto endFAIL;
|
|
|
1812 lp->sl_prefixcnt = cnt;
|
|
|
1813
|
|
|
1814 for (i = 0; i < cnt; ++i)
|
|
|
1815 {
|
|
|
1816 /* <prefcond> : <condlen> <condstr> */
|
|
|
1817 n = getc(fd); /* <condlen> */
|
|
|
1818 if (n < 0)
|
|
|
1819 goto formerr;
|
|
|
1820 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
1821 * compile the regexp program used to check for the condition. */
|
|
|
1822 if (n > 0)
|
|
|
1823 {
|
|
344
|
1824 buf[0] = '^'; /* always match at one position only */
|
|
|
1825 p = buf + 1;
|
|
339
|
1826 while (n-- > 0)
|
|
|
1827 *p++ = getc(fd); /* <condstr> */
|
|
|
1828 *p = NUL;
|
|
|
1829 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
1830 }
|
|
|
1831 }
|
|
|
1832 }
|
|
|
1833
|
|
|
1834
|
|
323
|
1835 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
1836 * <salflags> <salcount> <sal> ...
|
|
|
1837 * <maplen> <mapstr> */
|
|
339
|
1838
|
|
344
|
1839 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
1840 if (cnt < 0)
|
|
|
1841 goto formerr;
|
|
|
1842
|
|
|
1843 gap = &lp->sl_rep;
|
|
|
1844 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1845 goto endFAIL;
|
|
|
1846
|
|
|
1847 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
1848 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
1849 {
|
|
|
1850 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
381
|
1851 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
1852 if (i == FAIL)
|
|
|
1853 goto endFAIL;
|
|
|
1854 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
1855 if (i == FAIL)
|
|
|
1856 {
|
|
|
1857 vim_free(ftp->ft_from);
|
|
|
1858 goto endFAIL;
|
|
344
|
1859 }
|
|
|
1860 }
|
|
|
1861
|
|
|
1862 /* Fill the first-index table. */
|
|
|
1863 first = lp->sl_rep_first;
|
|
|
1864 for (i = 0; i < 256; ++i)
|
|
|
1865 first[i] = -1;
|
|
|
1866 for (i = 0; i < gap->ga_len; ++i)
|
|
323
|
1867 {
|
|
344
|
1868 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
1869 if (first[*ftp->ft_from] == -1)
|
|
|
1870 first[*ftp->ft_from] = i;
|
|
|
1871 }
|
|
|
1872
|
|
|
1873 i = getc(fd); /* <salflags> */
|
|
|
1874 if (i & SAL_F0LLOWUP)
|
|
|
1875 lp->sl_followup = TRUE;
|
|
|
1876 if (i & SAL_COLLAPSE)
|
|
|
1877 lp->sl_collapse = TRUE;
|
|
|
1878 if (i & SAL_REM_ACCENTS)
|
|
|
1879 lp->sl_rem_accents = TRUE;
|
|
375
|
1880 if (i & SAL_SOFO)
|
|
|
1881 lp->sl_sofo = TRUE;
|
|
344
|
1882
|
|
|
1883 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
1884 if (cnt < 0)
|
|
|
1885 goto formerr;
|
|
|
1886
|
|
375
|
1887 if (lp->sl_sofo)
|
|
|
1888 {
|
|
|
1889 /*
|
|
|
1890 * SOFOFROM and SOFOTO items come in one <salfrom> and <salto>
|
|
|
1891 */
|
|
|
1892 if (cnt != 1)
|
|
|
1893 goto formerr;
|
|
|
1894
|
|
381
|
1895 /* <salfromlen> <salfrom> */
|
|
|
1896 bp = read_cnt_string(fd, 2, &cnt);
|
|
|
1897 if (cnt == FAIL)
|
|
375
|
1898 goto endFAIL;
|
|
381
|
1899
|
|
|
1900 /* <saltolen> <salto> */
|
|
|
1901 fol = read_cnt_string(fd, 2, &cnt);
|
|
|
1902 if (cnt == FAIL)
|
|
375
|
1903 {
|
|
|
1904 vim_free(bp);
|
|
344
|
1905 goto endFAIL;
|
|
375
|
1906 }
|
|
381
|
1907
|
|
|
1908 /* Store the info in lp->sl_sal and/or lp->sl_sal_first. */
|
|
|
1909 i = set_sofo(lp, bp, fol);
|
|
|
1910
|
|
375
|
1911 vim_free(bp);
|
|
|
1912 vim_free(fol);
|
|
381
|
1913 if (i == FAIL)
|
|
|
1914 goto formerr;
|
|
375
|
1915 }
|
|
|
1916 else
|
|
|
1917 {
|
|
|
1918 /*
|
|
|
1919 * SAL items
|
|
|
1920 */
|
|
|
1921 gap = &lp->sl_sal;
|
|
|
1922 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1923 goto endFAIL;
|
|
|
1924
|
|
|
1925 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
1926 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
1927 {
|
|
|
1928 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
1929 ccnt = getc(fd); /* <salfromlen> */
|
|
|
1930 if (ccnt < 0)
|
|
|
1931 goto formerr;
|
|
|
1932 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
1933 goto endFAIL;
|
|
|
1934 smp->sm_lead = p;
|
|
|
1935
|
|
|
1936 /* Read up to the first special char into sm_lead. */
|
|
|
1937 for (i = 0; i < ccnt; ++i)
|
|
|
1938 {
|
|
|
1939 c = getc(fd); /* <salfrom> */
|
|
|
1940 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
1941 break;
|
|
|
1942 *p++ = c;
|
|
|
1943 }
|
|
|
1944 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
1945 *p++ = NUL;
|
|
|
1946
|
|
|
1947 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
1948 if (c == '(')
|
|
|
1949 {
|
|
|
1950 smp->sm_oneof = p;
|
|
|
1951 for (++i; i < ccnt; ++i)
|
|
|
1952 {
|
|
|
1953 c = getc(fd); /* <salfrom> */
|
|
|
1954 if (c == ')')
|
|
|
1955 break;
|
|
|
1956 *p++ = c;
|
|
|
1957 }
|
|
|
1958 *p++ = NUL;
|
|
|
1959 if (++i < ccnt)
|
|
|
1960 c = getc(fd);
|
|
|
1961 }
|
|
372
|
1962 else
|
|
375
|
1963 smp->sm_oneof = NULL;
|
|
|
1964
|
|
|
1965 /* Any following chars go in sm_rules. */
|
|
|
1966 smp->sm_rules = p;
|
|
|
1967 if (i < ccnt)
|
|
|
1968 /* store the char we got while checking for end of sm_lead */
|
|
|
1969 *p++ = c;
|
|
|
1970 for (++i; i < ccnt; ++i)
|
|
|
1971 *p++ = getc(fd); /* <salfrom> */
|
|
|
1972 *p++ = NUL;
|
|
|
1973
|
|
381
|
1974 /* <saltolen> <salto> */
|
|
|
1975 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
1976 if (ccnt == FAIL)
|
|
372
|
1977 {
|
|
|
1978 vim_free(smp->sm_lead);
|
|
375
|
1979 goto formerr;
|
|
|
1980 }
|
|
|
1981
|
|
372
|
1982 #ifdef FEAT_MBYTE
|
|
|
1983 if (has_mbyte)
|
|
|
1984 {
|
|
375
|
1985 /* convert the multi-byte strings to wide char strings */
|
|
|
1986 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
1987 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
1988 if (smp->sm_oneof == NULL)
|
|
|
1989 smp->sm_oneof_w = NULL;
|
|
|
1990 else
|
|
|
1991 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
1992 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
1993 if (smp->sm_lead_w == NULL
|
|
|
1994 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
1995 || smp->sm_to_w == NULL)
|
|
372
|
1996 {
|
|
375
|
1997 vim_free(smp->sm_lead);
|
|
|
1998 vim_free(smp->sm_to);
|
|
|
1999 vim_free(smp->sm_lead_w);
|
|
|
2000 vim_free(smp->sm_oneof_w);
|
|
|
2001 vim_free(smp->sm_to_w);
|
|
|
2002 goto endFAIL;
|
|
372
|
2003 }
|
|
|
2004 }
|
|
|
2005 #endif
|
|
|
2006 }
|
|
375
|
2007
|
|
|
2008 /* Fill the first-index table. */
|
|
381
|
2009 set_sal_first(lp);
|
|
|
2010 }
|
|
|
2011
|
|
|
2012 /* <maplen> <mapstr> */
|
|
|
2013 p = read_cnt_string(fd, 2, &cnt);
|
|
|
2014 if (cnt == FAIL)
|
|
323
|
2015 goto endFAIL;
|
|
330
|
2016 set_map_str(lp, p);
|
|
|
2017 vim_free(p);
|
|
323
|
2018
|
|
236
|
2019
|
|
300
|
2020 /* round 1: <LWORDTREE>
|
|
339
|
2021 * round 2: <KWORDTREE>
|
|
|
2022 * round 3: <PREFIXTREE> */
|
|
|
2023 for (round = 1; round <= 3; ++round)
|
|
236
|
2024 {
|
|
300
|
2025 /* The tree size was computed when writing the file, so that we can
|
|
|
2026 * allocate it as one long block. <nodecount> */
|
|
|
2027 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2028 if (len < 0)
|
|
|
2029 goto truncerr;
|
|
|
2030 if (len > 0)
|
|
236
|
2031 {
|
|
300
|
2032 /* Allocate the byte array. */
|
|
339
|
2033 bp = lalloc((long_u)len, TRUE);
|
|
|
2034 if (bp == NULL)
|
|
300
|
2035 goto endFAIL;
|
|
|
2036 if (round == 1)
|
|
339
|
2037 lp->sl_fbyts = bp;
|
|
|
2038 else if (round == 2)
|
|
|
2039 lp->sl_kbyts = bp;
|
|
300
|
2040 else
|
|
339
|
2041 lp->sl_pbyts = bp;
|
|
236
|
2042
|
|
300
|
2043 /* Allocate the index array. */
|
|
339
|
2044 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2045 if (ip == NULL)
|
|
300
|
2046 goto endFAIL;
|
|
|
2047 if (round == 1)
|
|
339
|
2048 lp->sl_fidxs = ip;
|
|
|
2049 else if (round == 2)
|
|
|
2050 lp->sl_kidxs = ip;
|
|
300
|
2051 else
|
|
339
|
2052 lp->sl_pidxs = ip;
|
|
300
|
2053
|
|
|
2054 /* Read the tree and store it in the array. */
|
|
339
|
2055 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2056 if (idx == -1)
|
|
300
|
2057 goto truncerr;
|
|
324
|
2058 if (idx < 0)
|
|
236
|
2059 goto formerr;
|
|
|
2060 }
|
|
300
|
2061 }
|
|
243
|
2062
|
|
310
|
2063 /* For a new file link it in the list of spell files. */
|
|
|
2064 if (old_lp == NULL)
|
|
|
2065 {
|
|
|
2066 lp->sl_next = first_lang;
|
|
|
2067 first_lang = lp;
|
|
|
2068 }
|
|
307
|
2069
|
|
255
|
2070 goto endOK;
|
|
|
2071
|
|
|
2072 endFAIL:
|
|
310
|
2073 if (lang != NULL)
|
|
|
2074 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2075 *lang = NUL;
|
|
|
2076 if (lp != NULL && old_lp == NULL)
|
|
323
|
2077 {
|
|
307
|
2078 slang_free(lp);
|
|
323
|
2079 lp = NULL;
|
|
|
2080 }
|
|
255
|
2081
|
|
|
2082 endOK:
|
|
236
|
2083 if (fd != NULL)
|
|
|
2084 fclose(fd);
|
|
|
2085 sourcing_name = save_sourcing_name;
|
|
|
2086 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2087
|
|
|
2088 return lp;
|
|
236
|
2089 }
|
|
|
2090
|
|
376
|
2091 /*
|
|
|
2092 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2093 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2094 * Returns NULL when the count is zero.
|
|
381
|
2095 * Sets "*errp" to FAIL when there is an error, OK otherwise.
|
|
376
|
2096 */
|
|
|
2097 static char_u *
|
|
|
2098 read_cnt_string(fd, cnt_bytes, errp)
|
|
|
2099 FILE *fd;
|
|
|
2100 int cnt_bytes;
|
|
|
2101 int *errp;
|
|
|
2102 {
|
|
|
2103 int cnt = 0;
|
|
|
2104 int i;
|
|
|
2105 char_u *str;
|
|
|
2106
|
|
|
2107 /* read the length bytes, MSB first */
|
|
|
2108 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2109 cnt = (cnt << 8) + getc(fd);
|
|
|
2110 if (cnt < 0)
|
|
|
2111 {
|
|
381
|
2112 EMSG(_(e_spell_trunc));
|
|
|
2113 *errp = FAIL;
|
|
376
|
2114 return NULL;
|
|
|
2115 }
|
|
|
2116
|
|
|
2117 /* allocate memory */
|
|
|
2118 str = alloc((unsigned)cnt + 1);
|
|
|
2119 if (str == NULL)
|
|
|
2120 {
|
|
381
|
2121 *errp = FAIL;
|
|
376
|
2122 return NULL;
|
|
|
2123 }
|
|
381
|
2124 *errp = OK;
|
|
376
|
2125
|
|
|
2126 /* Read the string. Doesn't check for truncated file. */
|
|
|
2127 for (i = 0; i < cnt; ++i)
|
|
|
2128 str[i] = getc(fd);
|
|
|
2129 str[i] = NUL;
|
|
|
2130
|
|
|
2131 return str;
|
|
|
2132 }
|
|
|
2133
|
|
381
|
2134 /*
|
|
|
2135 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
|
2136 * Returns FAIL when there is something wrong.
|
|
|
2137 */
|
|
|
2138 static int
|
|
|
2139 set_sofo(lp, from, to)
|
|
|
2140 slang_T *lp;
|
|
|
2141 char_u *from;
|
|
|
2142 char_u *to;
|
|
|
2143 {
|
|
|
2144 int i;
|
|
|
2145
|
|
|
2146 #ifdef FEAT_MBYTE
|
|
|
2147 garray_T *gap;
|
|
|
2148 char_u *s;
|
|
|
2149 char_u *p;
|
|
|
2150 int c;
|
|
|
2151 int *inp;
|
|
|
2152
|
|
|
2153 if (has_mbyte)
|
|
|
2154 {
|
|
|
2155 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2156 * characters. The index is the low byte of the character.
|
|
|
2157 * The list contains from-to pairs with a terminating NUL.
|
|
|
2158 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2159 gap = &lp->sl_sal;
|
|
|
2160 ga_init2(gap, sizeof(int *), 1);
|
|
|
2161 if (ga_grow(gap, 256) == FAIL)
|
|
|
2162 return FAIL;
|
|
|
2163 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2164 gap->ga_len = 256;
|
|
|
2165
|
|
|
2166 /* First count the number of items for each list. Temporarily use
|
|
|
2167 * sl_sal_first[] for this. */
|
|
|
2168 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2169 {
|
|
|
2170 c = mb_ptr2char_adv(&p);
|
|
|
2171 mb_ptr_adv(s);
|
|
|
2172 if (c >= 256)
|
|
|
2173 ++lp->sl_sal_first[c & 0xff];
|
|
|
2174 }
|
|
|
2175 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
|
2176 return FAIL;
|
|
|
2177
|
|
|
2178 /* Allocate the lists. */
|
|
|
2179 for (i = 0; i < 256; ++i)
|
|
|
2180 if (lp->sl_sal_first[i] > 0)
|
|
|
2181 {
|
|
|
2182 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2183 if (p == NULL)
|
|
|
2184 return FAIL;
|
|
|
2185 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2186 *(int *)p = 0;
|
|
|
2187 }
|
|
|
2188
|
|
|
2189 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
2190 * list. */
|
|
|
2191 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2192 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2193 {
|
|
|
2194 c = mb_ptr2char_adv(&p);
|
|
|
2195 i = mb_ptr2char_adv(&s);
|
|
|
2196 if (c >= 256)
|
|
|
2197 {
|
|
|
2198 /* Append the from-to chars at the end of the list with
|
|
|
2199 * the low byte. */
|
|
|
2200 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2201 while (*inp != 0)
|
|
|
2202 ++inp;
|
|
|
2203 *inp++ = c; /* from char */
|
|
|
2204 *inp++ = i; /* to char */
|
|
|
2205 *inp++ = NUL; /* NUL at the end */
|
|
|
2206 }
|
|
|
2207 else
|
|
|
2208 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2209 lp->sl_sal_first[c] = i;
|
|
|
2210 }
|
|
|
2211 }
|
|
|
2212 else
|
|
|
2213 #endif
|
|
|
2214 {
|
|
|
2215 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2216 if (STRLEN(from) != STRLEN(to))
|
|
|
2217 return FAIL;
|
|
|
2218
|
|
|
2219 for (i = 0; to[i] != NUL; ++i)
|
|
|
2220 lp->sl_sal_first[from[i]] = to[i];
|
|
|
2221 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2222 }
|
|
|
2223
|
|
|
2224 return OK;
|
|
|
2225 }
|
|
|
2226
|
|
|
2227 /*
|
|
|
2228 * Fill the first-index table for "lp".
|
|
|
2229 */
|
|
|
2230 static void
|
|
|
2231 set_sal_first(lp)
|
|
|
2232 slang_T *lp;
|
|
|
2233 {
|
|
|
2234 salfirst_T *sfirst;
|
|
|
2235 int i;
|
|
|
2236 salitem_T *smp;
|
|
|
2237 int c;
|
|
|
2238 garray_T *gap = &lp->sl_sal;
|
|
|
2239
|
|
|
2240 sfirst = lp->sl_sal_first;
|
|
|
2241 for (i = 0; i < 256; ++i)
|
|
|
2242 sfirst[i] = -1;
|
|
|
2243 smp = (salitem_T *)gap->ga_data;
|
|
|
2244 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2245 {
|
|
|
2246 #ifdef FEAT_MBYTE
|
|
|
2247 if (has_mbyte)
|
|
|
2248 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2249 * the character, for other encodings it should differ for most
|
|
|
2250 * characters. */
|
|
|
2251 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2252 else
|
|
|
2253 #endif
|
|
|
2254 c = *smp[i].sm_lead;
|
|
|
2255 if (sfirst[c] == -1)
|
|
|
2256 {
|
|
|
2257 sfirst[c] = i;
|
|
|
2258 #ifdef FEAT_MBYTE
|
|
|
2259 if (has_mbyte)
|
|
|
2260 {
|
|
|
2261 int n;
|
|
|
2262
|
|
|
2263 /* Make sure all entries with this byte are following each
|
|
|
2264 * other. Move the ones that are in the wrong position. Do
|
|
|
2265 * keep the same ordering! */
|
|
|
2266 while (i + 1 < gap->ga_len
|
|
|
2267 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
2268 /* Skip over entry with same index byte. */
|
|
|
2269 ++i;
|
|
|
2270
|
|
|
2271 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
2272 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
2273 {
|
|
|
2274 salitem_T tsal;
|
|
|
2275
|
|
|
2276 /* Move entry with same index byte after the entries
|
|
|
2277 * we already found. */
|
|
|
2278 ++i;
|
|
|
2279 --n;
|
|
|
2280 tsal = smp[i + n];
|
|
|
2281 mch_memmove(smp + i + 1, smp + i,
|
|
|
2282 sizeof(salitem_T) * n);
|
|
|
2283 smp[i] = tsal;
|
|
|
2284 }
|
|
|
2285 }
|
|
|
2286 #endif
|
|
|
2287 }
|
|
|
2288 }
|
|
|
2289 }
|
|
376
|
2290
|
|
372
|
2291 #ifdef FEAT_MBYTE
|
|
|
2292 /*
|
|
|
2293 * Turn a multi-byte string into a wide character string.
|
|
|
2294 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
2295 */
|
|
|
2296 static int *
|
|
|
2297 mb_str2wide(s)
|
|
|
2298 char_u *s;
|
|
|
2299 {
|
|
|
2300 int *res;
|
|
|
2301 char_u *p;
|
|
|
2302 int i = 0;
|
|
|
2303
|
|
|
2304 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
2305 if (res != NULL)
|
|
|
2306 {
|
|
|
2307 for (p = s; *p != NUL; )
|
|
|
2308 res[i++] = mb_ptr2char_adv(&p);
|
|
|
2309 res[i] = NUL;
|
|
|
2310 }
|
|
|
2311 return res;
|
|
|
2312 }
|
|
|
2313 #endif
|
|
|
2314
|
|
236
|
2315 /*
|
|
300
|
2316 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
2317 * "byts" and index array "idxs". Recursively read the children.
|
|
|
2318 *
|
|
346
|
2319 * NOTE: The code here must match put_node().
|
|
300
|
2320 *
|
|
|
2321 * Returns the index follosing the siblings.
|
|
|
2322 * Returns -1 if the file is shorter than expected.
|
|
|
2323 * Returns -2 if there is a format error.
|
|
236
|
2324 */
|
|
324
|
2325 static idx_T
|
|
339
|
2326 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
2327 FILE *fd;
|
|
|
2328 char_u *byts;
|
|
324
|
2329 idx_T *idxs;
|
|
300
|
2330 int maxidx; /* size of arrays */
|
|
324
|
2331 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
2332 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
2333 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
2334 {
|
|
300
|
2335 int len;
|
|
|
2336 int i;
|
|
|
2337 int n;
|
|
324
|
2338 idx_T idx = startidx;
|
|
300
|
2339 int c;
|
|
366
|
2340 int c2;
|
|
300
|
2341 #define SHARED_MASK 0x8000000
|
|
236
|
2342
|
|
300
|
2343 len = getc(fd); /* <siblingcount> */
|
|
|
2344 if (len <= 0)
|
|
|
2345 return -1;
|
|
|
2346
|
|
|
2347 if (startidx + len >= maxidx)
|
|
|
2348 return -2;
|
|
|
2349 byts[idx++] = len;
|
|
|
2350
|
|
|
2351 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
2352 for (i = 1; i <= len; ++i)
|
|
236
|
2353 {
|
|
300
|
2354 c = getc(fd); /* <byte> */
|
|
|
2355 if (c < 0)
|
|
|
2356 return -1;
|
|
|
2357 if (c <= BY_SPECIAL)
|
|
|
2358 {
|
|
366
|
2359 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
2360 {
|
|
|
2361 /* No flags, all regions. */
|
|
|
2362 idxs[idx] = 0;
|
|
|
2363 c = 0;
|
|
|
2364 }
|
|
366
|
2365 else if (c == BY_FLAGS || c == BY_NOFLAGS)
|
|
300
|
2366 {
|
|
339
|
2367 if (prefixtree)
|
|
|
2368 {
|
|
|
2369 /* Read the prefix ID and the condition nr. In idxs[]
|
|
|
2370 * store the prefix ID in the low byte, the condition
|
|
|
2371 * index shifted up 8 bits. */
|
|
366
|
2372 c2 = getc(fd); /* <prefixID> */
|
|
339
|
2373 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
2374 if (n >= maxprefcondnr)
|
|
|
2375 return -2;
|
|
366
|
2376 c2 += (n << 8);
|
|
|
2377 if (c == BY_NOFLAGS)
|
|
|
2378 c = c2;
|
|
|
2379 else
|
|
|
2380 c = c2 | WF_RAREPFX;
|
|
339
|
2381 }
|
|
|
2382 else
|
|
|
2383 {
|
|
|
2384 /* Read flags and optional region and prefix ID. In
|
|
|
2385 * idxs[] the flags go in the low byte, region above that
|
|
|
2386 * and prefix ID above the region. */
|
|
|
2387 c = getc(fd); /* <flags> */
|
|
|
2388 if (c & WF_REGION)
|
|
|
2389 c = (getc(fd) << 8) + c; /* <region> */
|
|
|
2390 if (c & WF_PFX)
|
|
|
2391 c = (getc(fd) << 16) + c; /* <prefixID> */
|
|
|
2392 }
|
|
|
2393
|
|
300
|
2394 idxs[idx] = c;
|
|
|
2395 c = 0;
|
|
|
2396 }
|
|
|
2397 else /* c == BY_INDEX */
|
|
|
2398 {
|
|
|
2399 /* <nodeidx> */
|
|
|
2400 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2401 if (n < 0 || n >= maxidx)
|
|
|
2402 return -2;
|
|
|
2403 idxs[idx] = n + SHARED_MASK;
|
|
|
2404 c = getc(fd); /* <xbyte> */
|
|
|
2405 }
|
|
|
2406 }
|
|
|
2407 byts[idx++] = c;
|
|
236
|
2408 }
|
|
|
2409
|
|
300
|
2410 /* Recursively read the children for non-shared siblings.
|
|
|
2411 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
2412 * remove SHARED_MASK) */
|
|
|
2413 for (i = 1; i <= len; ++i)
|
|
|
2414 if (byts[startidx + i] != 0)
|
|
|
2415 {
|
|
|
2416 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
2417 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
2418 else
|
|
|
2419 {
|
|
|
2420 idxs[startidx + i] = idx;
|
|
339
|
2421 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
2422 prefixtree, maxprefcondnr);
|
|
300
|
2423 if (idx < 0)
|
|
|
2424 break;
|
|
|
2425 }
|
|
|
2426 }
|
|
236
|
2427
|
|
300
|
2428 return idx;
|
|
236
|
2429 }
|
|
|
2430
|
|
|
2431 /*
|
|
|
2432 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
2433 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
2434 */
|
|
|
2435 char_u *
|
|
|
2436 did_set_spelllang(buf)
|
|
|
2437 buf_T *buf;
|
|
|
2438 {
|
|
|
2439 garray_T ga;
|
|
351
|
2440 char_u *splp;
|
|
236
|
2441 char_u *region;
|
|
355
|
2442 int filename;
|
|
236
|
2443 int region_mask;
|
|
|
2444 slang_T *lp;
|
|
|
2445 int c;
|
|
351
|
2446 char_u lang[MAXWLEN + 1];
|
|
323
|
2447 char_u spf_name[MAXPATHL];
|
|
351
|
2448 int load_spf;
|
|
|
2449 int len;
|
|
|
2450 char_u *p;
|
|
381
|
2451 int round;
|
|
236
|
2452
|
|
|
2453 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
2454 clear_midword(buf);
|
|
236
|
2455
|
|
351
|
2456 /* Make the name of the .spl file associated with 'spellfile'. */
|
|
323
|
2457 if (*buf->b_p_spf == NUL)
|
|
351
|
2458 load_spf = FALSE;
|
|
323
|
2459 else
|
|
351
|
2460 {
|
|
323
|
2461 vim_snprintf((char *)spf_name, sizeof(spf_name), "%s.spl",
|
|
|
2462 buf->b_p_spf);
|
|
351
|
2463 load_spf = TRUE;
|
|
|
2464 }
|
|
|
2465
|
|
|
2466 /* loop over comma separated language names. */
|
|
|
2467 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
2468 {
|
|
|
2469 /* Get one language name. */
|
|
|
2470 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
2471
|
|
240
|
2472 region = NULL;
|
|
351
|
2473 len = STRLEN(lang);
|
|
355
|
2474
|
|
|
2475 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
2476 * If there is a region name let "region" point to it and remove it
|
|
|
2477 * from the name. */
|
|
|
2478 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
2479 {
|
|
355
|
2480 filename = TRUE;
|
|
|
2481
|
|
|
2482 /* Check if we loaded this language before. */
|
|
|
2483 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2484 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2485 break;
|
|
236
|
2486 }
|
|
355
|
2487 else
|
|
|
2488 {
|
|
|
2489 filename = FALSE;
|
|
|
2490 if (len > 3 && lang[len - 3] == '_')
|
|
|
2491 {
|
|
|
2492 region = lang + len - 2;
|
|
|
2493 len -= 3;
|
|
|
2494 lang[len] = NUL;
|
|
|
2495 }
|
|
|
2496
|
|
|
2497 /* Check if we loaded this language before. */
|
|
|
2498 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2499 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
2500 break;
|
|
|
2501 }
|
|
236
|
2502
|
|
351
|
2503 /* If not found try loading the language now. */
|
|
236
|
2504 if (lp == NULL)
|
|
355
|
2505 {
|
|
|
2506 if (filename)
|
|
|
2507 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
2508 else
|
|
|
2509 spell_load_lang(lang);
|
|
|
2510 }
|
|
236
|
2511
|
|
307
|
2512 /*
|
|
351
|
2513 * Loop over the languages, there can be several files for "lang".
|
|
307
|
2514 */
|
|
|
2515 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
2516 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
2517 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
2518 {
|
|
316
|
2519 region_mask = REGION_ALL;
|
|
355
|
2520 if (!filename && region != NULL)
|
|
236
|
2521 {
|
|
307
|
2522 /* find region in sl_regions */
|
|
|
2523 c = find_region(lp->sl_regions, region);
|
|
|
2524 if (c == REGION_ALL)
|
|
|
2525 {
|
|
316
|
2526 if (!lp->sl_add)
|
|
351
|
2527 smsg((char_u *)
|
|
|
2528 _("Warning: region %s not supported"),
|
|
|
2529 region);
|
|
307
|
2530 }
|
|
|
2531 else
|
|
|
2532 region_mask = 1 << c;
|
|
236
|
2533 }
|
|
307
|
2534
|
|
|
2535 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2536 {
|
|
|
2537 ga_clear(&ga);
|
|
|
2538 return e_outofmem;
|
|
|
2539 }
|
|
|
2540 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2541 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2542 ++ga.ga_len;
|
|
376
|
2543 use_midword(lp, buf);
|
|
323
|
2544
|
|
351
|
2545 /* Check if this is the spell file related to 'spellfile'. */
|
|
|
2546 if (load_spf && fullpathcmp(spf_name, lp->sl_fname, FALSE)
|
|
|
2547 == FPC_SAME)
|
|
|
2548 load_spf = FALSE;
|
|
236
|
2549 }
|
|
|
2550 }
|
|
|
2551
|
|
381
|
2552 /* round 1: load 'spellfile', if needed.
|
|
|
2553 * round 2: load temp_wordlist, if possible. */
|
|
|
2554 for (round = 1; round <= 2; ++round)
|
|
|
2555 {
|
|
|
2556 if (round == 1)
|
|
|
2557 {
|
|
|
2558 /* Make sure the 'spellfile' file is loaded. It may be in
|
|
|
2559 * 'runtimepath', then it's probably loaded above already.
|
|
|
2560 * Otherwise load it here. */
|
|
|
2561 if (!load_spf)
|
|
|
2562 continue;
|
|
|
2563 }
|
|
|
2564 else
|
|
|
2565 {
|
|
|
2566 if (temp_wordlist == NULL)
|
|
|
2567 continue;
|
|
|
2568 vim_snprintf((char *)spf_name, sizeof(spf_name), "%s.%s.spl",
|
|
|
2569 temp_wordlist, spell_enc());
|
|
|
2570 }
|
|
|
2571
|
|
351
|
2572 /* Check if it was loaded already. */
|
|
323
|
2573 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2574 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2575 break;
|
|
|
2576 if (lp == NULL)
|
|
|
2577 {
|
|
351
|
2578 /* Not loaded, try loading it now. The language name includes the
|
|
381
|
2579 * region name, the region is ignored otherwise. for
|
|
|
2580 * temp_wordlist use an arbitrary name. */
|
|
|
2581 if (round == 1)
|
|
|
2582 {
|
|
|
2583 vim_strncpy(lang, gettail(buf->b_p_spf), MAXWLEN);
|
|
|
2584 p = vim_strchr(lang, '.');
|
|
|
2585 if (p != NULL)
|
|
|
2586 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
2587 }
|
|
|
2588 else
|
|
|
2589 STRCPY(lang, "temp wordlist");
|
|
351
|
2590 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
2591 }
|
|
|
2592 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
2593 {
|
|
|
2594 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2595 LANGP_ENTRY(ga, ga.ga_len)->lp_region = REGION_ALL;
|
|
|
2596 ++ga.ga_len;
|
|
376
|
2597 use_midword(lp, buf);
|
|
323
|
2598 }
|
|
|
2599 }
|
|
|
2600
|
|
236
|
2601 /* Add a NULL entry to mark the end of the list. */
|
|
|
2602 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2603 {
|
|
|
2604 ga_clear(&ga);
|
|
|
2605 return e_outofmem;
|
|
|
2606 }
|
|
|
2607 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
2608 ++ga.ga_len;
|
|
|
2609
|
|
|
2610 /* Everything is fine, store the new b_langp value. */
|
|
|
2611 ga_clear(&buf->b_langp);
|
|
|
2612 buf->b_langp = ga;
|
|
|
2613
|
|
|
2614 return NULL;
|
|
|
2615 }
|
|
|
2616
|
|
|
2617 /*
|
|
376
|
2618 * Clear the midword characters for buffer "buf".
|
|
|
2619 */
|
|
|
2620 static void
|
|
|
2621 clear_midword(buf)
|
|
|
2622 buf_T *buf;
|
|
|
2623 {
|
|
|
2624 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
2625 #ifdef FEAT_MBYTE
|
|
|
2626 vim_free(buf->b_spell_ismw_mb);
|
|
|
2627 buf->b_spell_ismw_mb = NULL;
|
|
|
2628 #endif
|
|
|
2629 }
|
|
|
2630
|
|
|
2631 /*
|
|
|
2632 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
2633 * They add up to any currently used midword characters.
|
|
|
2634 */
|
|
|
2635 static void
|
|
|
2636 use_midword(lp, buf)
|
|
|
2637 slang_T *lp;
|
|
|
2638 buf_T *buf;
|
|
|
2639 {
|
|
|
2640 char_u *p;
|
|
|
2641
|
|
|
2642 for (p = lp->sl_midword; *p != NUL; )
|
|
|
2643 #ifdef FEAT_MBYTE
|
|
|
2644 if (has_mbyte)
|
|
|
2645 {
|
|
|
2646 int c, l, n;
|
|
|
2647 char_u *bp;
|
|
|
2648
|
|
|
2649 c = mb_ptr2char(p);
|
|
|
2650 l = mb_ptr2len_check(p);
|
|
|
2651 if (c < 256)
|
|
|
2652 buf->b_spell_ismw[c] = TRUE;
|
|
|
2653 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
2654 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
2655 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
2656 else
|
|
|
2657 {
|
|
|
2658 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
2659 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
2660 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
2661 if (bp != NULL)
|
|
|
2662 {
|
|
|
2663 vim_free(buf->b_spell_ismw_mb);
|
|
|
2664 buf->b_spell_ismw_mb = bp;
|
|
|
2665 vim_strncpy(bp + n, p, l);
|
|
|
2666 }
|
|
|
2667 }
|
|
|
2668 p += l;
|
|
|
2669 }
|
|
|
2670 else
|
|
|
2671 #endif
|
|
|
2672 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
2673 }
|
|
|
2674
|
|
|
2675 /*
|
|
236
|
2676 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
2677 * Each region is simply stored as the two characters of it's name.
|
|
381
|
2678 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
2679 */
|
|
|
2680 static int
|
|
|
2681 find_region(rp, region)
|
|
|
2682 char_u *rp;
|
|
|
2683 char_u *region;
|
|
|
2684 {
|
|
|
2685 int i;
|
|
|
2686
|
|
|
2687 for (i = 0; ; i += 2)
|
|
|
2688 {
|
|
|
2689 if (rp[i] == NUL)
|
|
|
2690 return REGION_ALL;
|
|
|
2691 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
2692 break;
|
|
|
2693 }
|
|
|
2694 return i / 2;
|
|
|
2695 }
|
|
|
2696
|
|
|
2697 /*
|
|
323
|
2698 * Return case type of word:
|
|
236
|
2699 * w word 0
|
|
300
|
2700 * Word WF_ONECAP
|
|
|
2701 * W WORD WF_ALLCAP
|
|
|
2702 * WoRd wOrd WF_KEEPCAP
|
|
236
|
2703 */
|
|
|
2704 static int
|
|
|
2705 captype(word, end)
|
|
|
2706 char_u *word;
|
|
323
|
2707 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
2708 {
|
|
|
2709 char_u *p;
|
|
|
2710 int c;
|
|
|
2711 int firstcap;
|
|
|
2712 int allcap;
|
|
|
2713 int past_second = FALSE; /* past second word char */
|
|
|
2714
|
|
|
2715 /* find first letter */
|
|
376
|
2716 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
2717 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
2718 return 0; /* only non-word characters, illegal word */
|
|
|
2719 #ifdef FEAT_MBYTE
|
|
310
|
2720 if (has_mbyte)
|
|
|
2721 c = mb_ptr2char_adv(&p);
|
|
|
2722 else
|
|
236
|
2723 #endif
|
|
310
|
2724 c = *p++;
|
|
324
|
2725 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
2726
|
|
|
2727 /*
|
|
|
2728 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
2729 * But a word with an upper char only at start is a ONECAP.
|
|
|
2730 */
|
|
323
|
2731 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
2732 if (spell_iswordp_nmw(p))
|
|
236
|
2733 {
|
|
|
2734 #ifdef FEAT_MBYTE
|
|
|
2735 c = mb_ptr2char(p);
|
|
|
2736 #else
|
|
|
2737 c = *p;
|
|
|
2738 #endif
|
|
324
|
2739 if (!SPELL_ISUPPER(c))
|
|
236
|
2740 {
|
|
|
2741 /* UUl -> KEEPCAP */
|
|
|
2742 if (past_second && allcap)
|
|
300
|
2743 return WF_KEEPCAP;
|
|
236
|
2744 allcap = FALSE;
|
|
|
2745 }
|
|
|
2746 else if (!allcap)
|
|
|
2747 /* UlU -> KEEPCAP */
|
|
300
|
2748 return WF_KEEPCAP;
|
|
236
|
2749 past_second = TRUE;
|
|
|
2750 }
|
|
|
2751
|
|
|
2752 if (allcap)
|
|
300
|
2753 return WF_ALLCAP;
|
|
236
|
2754 if (firstcap)
|
|
300
|
2755 return WF_ONECAP;
|
|
236
|
2756 return 0;
|
|
|
2757 }
|
|
|
2758
|
|
355
|
2759 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
2760 /*
|
|
|
2761 * Free all languages.
|
|
|
2762 */
|
|
|
2763 void
|
|
|
2764 spell_free_all()
|
|
|
2765 {
|
|
|
2766 slang_T *lp;
|
|
|
2767 buf_T *buf;
|
|
|
2768
|
|
|
2769 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2770 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2771 ga_clear(&buf->b_langp);
|
|
|
2772
|
|
|
2773 while (first_lang != NULL)
|
|
|
2774 {
|
|
|
2775 lp = first_lang;
|
|
|
2776 first_lang = lp->sl_next;
|
|
|
2777 slang_free(lp);
|
|
|
2778 }
|
|
366
|
2779
|
|
381
|
2780 if (temp_wordlist != NULL)
|
|
|
2781 {
|
|
|
2782 mch_remove(temp_wordlist);
|
|
|
2783 vim_free(temp_wordlist);
|
|
|
2784 temp_wordlist = NULL;
|
|
|
2785 }
|
|
|
2786
|
|
366
|
2787 init_spell_chartab();
|
|
355
|
2788 }
|
|
|
2789 # endif
|
|
|
2790
|
|
236
|
2791 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
2792 /*
|
|
|
2793 * Clear all spelling tables and reload them.
|
|
307
|
2794 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
2795 */
|
|
|
2796 void
|
|
|
2797 spell_reload()
|
|
|
2798 {
|
|
|
2799 buf_T *buf;
|
|
316
|
2800 win_T *wp;
|
|
236
|
2801
|
|
358
|
2802 /* Initialize the table for spell_iswordp(). */
|
|
236
|
2803 init_spell_chartab();
|
|
|
2804
|
|
|
2805 /* Unload all allocated memory. */
|
|
355
|
2806 spell_free_all();
|
|
236
|
2807
|
|
|
2808 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2809 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2810 {
|
|
316
|
2811 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
2812 * window for this buffer in which 'spell' is set. */
|
|
236
|
2813 if (*buf->b_p_spl != NUL)
|
|
316
|
2814 {
|
|
|
2815 FOR_ALL_WINDOWS(wp)
|
|
|
2816 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
2817 {
|
|
|
2818 (void)did_set_spelllang(buf);
|
|
|
2819 # ifdef FEAT_WINDOWS
|
|
|
2820 break;
|
|
|
2821 # endif
|
|
|
2822 }
|
|
|
2823 }
|
|
236
|
2824 }
|
|
|
2825 }
|
|
|
2826 # endif
|
|
|
2827
|
|
310
|
2828 /*
|
|
|
2829 * Reload the spell file "fname" if it's loaded.
|
|
|
2830 */
|
|
|
2831 static void
|
|
323
|
2832 spell_reload_one(fname, added_word)
|
|
310
|
2833 char_u *fname;
|
|
323
|
2834 int added_word; /* invoked through "zg" */
|
|
310
|
2835 {
|
|
|
2836 slang_T *lp;
|
|
323
|
2837 int didit = FALSE;
|
|
310
|
2838
|
|
|
2839 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2840 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2841 {
|
|
|
2842 slang_clear(lp);
|
|
323
|
2843 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
2844 redraw_all_later(NOT_VALID);
|
|
323
|
2845 didit = TRUE;
|
|
310
|
2846 }
|
|
323
|
2847
|
|
|
2848 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
2849 * 'spelllang' to get it loaded. */
|
|
|
2850 if (added_word && !didit)
|
|
|
2851 did_set_spelllang(curbuf);
|
|
310
|
2852 }
|
|
|
2853
|
|
|
2854
|
|
236
|
2855 /*
|
|
|
2856 * Functions for ":mkspell".
|
|
|
2857 */
|
|
|
2858
|
|
300
|
2859 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
2860 and .dic file. */
|
|
|
2861 /*
|
|
|
2862 * Main structure to store the contents of a ".aff" file.
|
|
|
2863 */
|
|
|
2864 typedef struct afffile_S
|
|
|
2865 {
|
|
|
2866 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
310
|
2867 int af_rar; /* RAR ID for rare word */
|
|
|
2868 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
2869 int af_bad; /* BAD ID for banned word */
|
|
339
|
2870 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
2871 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
2872 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
2873 } afffile_T;
|
|
|
2874
|
|
|
2875 typedef struct affentry_S affentry_T;
|
|
|
2876 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
2877 struct affentry_S
|
|
|
2878 {
|
|
|
2879 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
2880 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
2881 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
2882 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
2883 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
366
|
2884 int ae_rare; /* rare affix */
|
|
300
|
2885 };
|
|
|
2886
|
|
|
2887 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
2888 typedef struct affheader_S
|
|
|
2889 {
|
|
|
2890 char_u ah_key[2]; /* key for hashtable == name of affix entry */
|
|
339
|
2891 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
2892 int ah_combine; /* suffix may combine with prefix */
|
|
|
2893 affentry_T *ah_first; /* first affix entry */
|
|
|
2894 } affheader_T;
|
|
|
2895
|
|
|
2896 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
2897
|
|
|
2898 /*
|
|
|
2899 * Structure that is used to store the items in the word tree. This avoids
|
|
|
2900 * the need to keep track of each allocated thing, it's freed all at once
|
|
|
2901 * after ":mkspell" is done.
|
|
|
2902 */
|
|
|
2903 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
2904 typedef struct sblock_S sblock_T;
|
|
|
2905 struct sblock_S
|
|
|
2906 {
|
|
|
2907 sblock_T *sb_next; /* next block in list */
|
|
|
2908 int sb_used; /* nr of bytes already in use */
|
|
|
2909 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
2910 };
|
|
|
2911
|
|
|
2912 /*
|
|
300
|
2913 * A node in the tree.
|
|
236
|
2914 */
|
|
300
|
2915 typedef struct wordnode_S wordnode_T;
|
|
|
2916 struct wordnode_S
|
|
236
|
2917 {
|
|
346
|
2918 union /* shared to save space */
|
|
|
2919 {
|
|
|
2920 char_u hashkey[6]; /* room for the hash key */
|
|
|
2921 int index; /* index in written nodes (valid after first
|
|
|
2922 round) */
|
|
|
2923 } wn_u1;
|
|
|
2924 union /* shared to save space */
|
|
|
2925 {
|
|
|
2926 wordnode_T *next; /* next node with same hash key */
|
|
|
2927 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
2928 } wn_u2;
|
|
300
|
2929 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
2930 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
2931 always sorted) */
|
|
|
2932 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
|
2933 char_u wn_flags; /* when wn_byte is NUL: WF_ flags */
|
|
339
|
2934 short wn_region; /* when wn_byte is NUL: region mask; for
|
|
|
2935 PREFIXTREE it's the prefcondnr */
|
|
|
2936 char_u wn_prefixID; /* supported/required prefix ID or 0 */
|
|
236
|
2937 };
|
|
|
2938
|
|
300
|
2939 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
2940
|
|
300
|
2941 /*
|
|
|
2942 * Info used while reading the spell files.
|
|
|
2943 */
|
|
|
2944 typedef struct spellinfo_S
|
|
249
|
2945 {
|
|
300
|
2946 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
2947 long si_foldwcount; /* nr of words in si_foldroot */
|
|
300
|
2948 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
2949 long si_keepwcount; /* nr of words in si_keeproot */
|
|
339
|
2950 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
300
|
2951 sblock_T *si_blocks; /* memory blocks used */
|
|
|
2952 int si_ascii; /* handling only ASCII words */
|
|
310
|
2953 int si_add; /* addition file */
|
|
351
|
2954 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
2955 int si_region; /* region mask */
|
|
|
2956 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
2957 int si_memtot; /* runtime memory used */
|
|
310
|
2958 int si_verbose; /* verbose messages */
|
|
316
|
2959 int si_region_count; /* number of regions supported (1 when there
|
|
|
2960 are no regions) */
|
|
|
2961 char_u si_region_name[16]; /* region names (if count > 1) */
|
|
323
|
2962
|
|
|
2963 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
2964 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
2965 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
2966 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
2967 int si_followup; /* soundsalike: ? */
|
|
|
2968 int si_collapse; /* soundsalike: ? */
|
|
|
2969 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
2970 garray_T si_map; /* MAP info concatenated */
|
|
366
|
2971 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
339
|
2972 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
2973 * prefixes, each stored as a string */
|
|
|
2974 int si_newID; /* current value for ah_newID */
|
|
300
|
2975 } spellinfo_T;
|
|
249
|
2976
|
|
300
|
2977 static afffile_T *spell_read_aff __ARGS((char_u *fname, spellinfo_T *spin));
|
|
339
|
2978 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
2979 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
2980 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
2981 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
2982 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
|
2983 static int spell_read_dic __ARGS((char_u *fname, spellinfo_T *spin, afffile_T *affile));
|
|
339
|
2984 static char_u *get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, sblock_T **blp));
|
|
|
2985 static int store_aff_word __ARGS((char_u *word, spellinfo_T *spin, char_u *afflist, afffile_T *affile, hashtab_T *ht, hashtab_T *xht, int comb, int flags, char_u *pfxlist));
|
|
300
|
2986 static int spell_read_wordfile __ARGS((char_u *fname, spellinfo_T *spin));
|
|
|
2987 static void *getroom __ARGS((sblock_T **blp, size_t len));
|
|
|
2988 static char_u *getroom_save __ARGS((sblock_T **blp, char_u *s));
|
|
|
2989 static void free_blocks __ARGS((sblock_T *bl));
|
|
|
2990 static wordnode_T *wordtree_alloc __ARGS((sblock_T **blp));
|
|
339
|
2991 static int store_word __ARGS((char_u *word, spellinfo_T *spin, int flags, int region, char_u *pfxlist));
|
|
|
2992 static int tree_add_word __ARGS((char_u *word, wordnode_T *tree, int flags, int region, int prefixID, sblock_T **blp));
|
|
310
|
2993 static void wordtree_compress __ARGS((wordnode_T *root, spellinfo_T *spin));
|
|
300
|
2994 static int node_compress __ARGS((wordnode_T *node, hashtab_T *ht, int *tot));
|
|
|
2995 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
316
|
2996 static void write_vim_spell __ARGS((char_u *fname, spellinfo_T *spin));
|
|
346
|
2997 static void clear_node __ARGS((wordnode_T *node));
|
|
|
2998 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
2999 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3000 static void init_spellfile __ARGS((void));
|
|
236
|
3001
|
|
|
3002 /*
|
|
323
|
3003 * Read the affix file "fname".
|
|
316
|
3004 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
3005 */
|
|
|
3006 static afffile_T *
|
|
300
|
3007 spell_read_aff(fname, spin)
|
|
236
|
3008 char_u *fname;
|
|
300
|
3009 spellinfo_T *spin;
|
|
236
|
3010 {
|
|
|
3011 FILE *fd;
|
|
|
3012 afffile_T *aff;
|
|
|
3013 char_u rline[MAXLINELEN];
|
|
|
3014 char_u *line;
|
|
|
3015 char_u *pc = NULL;
|
|
334
|
3016 #define MAXITEMCNT 7
|
|
|
3017 char_u *(items[MAXITEMCNT]);
|
|
236
|
3018 int itemcnt;
|
|
|
3019 char_u *p;
|
|
|
3020 int lnum = 0;
|
|
|
3021 affheader_T *cur_aff = NULL;
|
|
|
3022 int aff_todo = 0;
|
|
|
3023 hashtab_T *tp;
|
|
255
|
3024 char_u *low = NULL;
|
|
|
3025 char_u *fol = NULL;
|
|
|
3026 char_u *upp = NULL;
|
|
307
|
3027 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
3028 int do_rep;
|
|
|
3029 int do_sal;
|
|
|
3030 int do_map;
|
|
366
|
3031 int do_midword;
|
|
375
|
3032 int do_sofo;
|
|
323
|
3033 int found_map = FALSE;
|
|
324
|
3034 hashitem_T *hi;
|
|
236
|
3035
|
|
300
|
3036 /*
|
|
|
3037 * Open the file.
|
|
|
3038 */
|
|
310
|
3039 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3040 if (fd == NULL)
|
|
|
3041 {
|
|
|
3042 EMSG2(_(e_notopen), fname);
|
|
|
3043 return NULL;
|
|
|
3044 }
|
|
|
3045
|
|
310
|
3046 if (spin->si_verbose || p_verbose > 2)
|
|
|
3047 {
|
|
|
3048 if (!spin->si_verbose)
|
|
|
3049 verbose_enter();
|
|
366
|
3050 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
3051 out_flush();
|
|
|
3052 if (!spin->si_verbose)
|
|
|
3053 verbose_leave();
|
|
|
3054 }
|
|
236
|
3055
|
|
323
|
3056 /* Only do REP lines when not done in another .aff file already. */
|
|
|
3057 do_rep = spin->si_rep.ga_len == 0;
|
|
|
3058
|
|
|
3059 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
3060 do_sal = spin->si_sal.ga_len == 0;
|
|
|
3061
|
|
|
3062 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
3063 do_map = spin->si_map.ga_len == 0;
|
|
|
3064
|
|
366
|
3065 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
3066 do_midword = spin->si_midword == NULL;
|
|
|
3067
|
|
375
|
3068 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
3069 do_sofo = spin->si_sofofr == NULL;
|
|
|
3070
|
|
300
|
3071 /*
|
|
|
3072 * Allocate and init the afffile_T structure.
|
|
|
3073 */
|
|
|
3074 aff = (afffile_T *)getroom(&spin->si_blocks, sizeof(afffile_T));
|
|
236
|
3075 if (aff == NULL)
|
|
|
3076 return NULL;
|
|
|
3077 hash_init(&aff->af_pref);
|
|
|
3078 hash_init(&aff->af_suff);
|
|
|
3079
|
|
|
3080 /*
|
|
|
3081 * Read all the lines in the file one by one.
|
|
|
3082 */
|
|
255
|
3083 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
3084 {
|
|
255
|
3085 line_breakcheck();
|
|
236
|
3086 ++lnum;
|
|
|
3087
|
|
|
3088 /* Skip comment lines. */
|
|
|
3089 if (*rline == '#')
|
|
|
3090 continue;
|
|
|
3091
|
|
|
3092 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
3093 vim_free(pc);
|
|
310
|
3094 #ifdef FEAT_MBYTE
|
|
300
|
3095 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3096 {
|
|
300
|
3097 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
3098 if (pc == NULL)
|
|
|
3099 {
|
|
|
3100 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3101 fname, lnum, rline);
|
|
|
3102 continue;
|
|
|
3103 }
|
|
236
|
3104 line = pc;
|
|
|
3105 }
|
|
|
3106 else
|
|
310
|
3107 #endif
|
|
236
|
3108 {
|
|
|
3109 pc = NULL;
|
|
|
3110 line = rline;
|
|
|
3111 }
|
|
|
3112
|
|
|
3113 /* Split the line up in white separated items. Put a NUL after each
|
|
|
3114 * item. */
|
|
|
3115 itemcnt = 0;
|
|
|
3116 for (p = line; ; )
|
|
|
3117 {
|
|
|
3118 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
3119 ++p;
|
|
|
3120 if (*p == NUL)
|
|
|
3121 break;
|
|
334
|
3122 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
3123 break;
|
|
236
|
3124 items[itemcnt++] = p;
|
|
300
|
3125 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
3126 ++p;
|
|
|
3127 if (*p == NUL)
|
|
|
3128 break;
|
|
|
3129 *p++ = NUL;
|
|
|
3130 }
|
|
|
3131
|
|
|
3132 /* Handle non-empty lines. */
|
|
|
3133 if (itemcnt > 0)
|
|
|
3134 {
|
|
|
3135 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
3136 && aff->af_enc == NULL)
|
|
|
3137 {
|
|
310
|
3138 #ifdef FEAT_MBYTE
|
|
300
|
3139 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
3140 aff->af_enc = enc_canonize(items[1]);
|
|
|
3141 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
3142 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
3143 p_enc) == FAIL)
|
|
|
3144 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
3145 fname, aff->af_enc, p_enc);
|
|
375
|
3146 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3147 #else
|
|
|
3148 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3149 #endif
|
|
236
|
3150 }
|
|
366
|
3151 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
3152 {
|
|
|
3153 if (do_midword)
|
|
|
3154 spin->si_midword = vim_strsave(items[1]);
|
|
|
3155 }
|
|
302
|
3156 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
3157 {
|
|
323
|
3158 /* ignored, we always split */
|
|
302
|
3159 }
|
|
323
|
3160 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
3161 {
|
|
323
|
3162 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
3163 }
|
|
307
|
3164 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
3165 && aff->af_rar == 0)
|
|
|
3166 {
|
|
|
3167 aff->af_rar = items[1][0];
|
|
|
3168 if (items[1][1] != NUL)
|
|
|
3169 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3170 }
|
|
310
|
3171 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
3172 && aff->af_kep == 0)
|
|
307
|
3173 {
|
|
310
|
3174 aff->af_kep = items[1][0];
|
|
307
|
3175 if (items[1][1] != NUL)
|
|
|
3176 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3177 }
|
|
346
|
3178 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
3179 && aff->af_bad == 0)
|
|
|
3180 {
|
|
|
3181 aff->af_bad = items[1][0];
|
|
|
3182 if (items[1][1] != NUL)
|
|
|
3183 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3184 }
|
|
339
|
3185 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
3186 {
|
|
|
3187 aff->af_pfxpostpone = TRUE;
|
|
|
3188 }
|
|
236
|
3189 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3190 || STRCMP(items[0], "SFX") == 0)
|
|
|
3191 && aff_todo == 0
|
|
334
|
3192 && itemcnt >= 4)
|
|
236
|
3193 {
|
|
334
|
3194 /* Myspell allows extra text after the item, but that might
|
|
|
3195 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
3196 if (itemcnt > 4 && *items[4] != '#')
|
|
|
3197 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
3198 fname, lnum, items[4]);
|
|
|
3199
|
|
236
|
3200 /* New affix letter. */
|
|
300
|
3201 cur_aff = (affheader_T *)getroom(&spin->si_blocks,
|
|
|
3202 sizeof(affheader_T));
|
|
236
|
3203 if (cur_aff == NULL)
|
|
|
3204 break;
|
|
339
|
3205 cur_aff->ah_key[0] = *items[1]; /* TODO: multi-byte? */
|
|
236
|
3206 cur_aff->ah_key[1] = NUL;
|
|
|
3207 if (items[1][1] != NUL)
|
|
307
|
3208 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
3209 if (*items[2] == 'Y')
|
|
|
3210 cur_aff->ah_combine = TRUE;
|
|
300
|
3211 else if (*items[2] != 'N')
|
|
236
|
3212 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
3213 fname, lnum, items[2]);
|
|
339
|
3214
|
|
236
|
3215 if (*items[0] == 'P')
|
|
339
|
3216 {
|
|
236
|
3217 tp = &aff->af_pref;
|
|
339
|
3218 /* Use a new number in the .spl file later, to be able to
|
|
|
3219 * handle multiple .aff files. */
|
|
|
3220 if (aff->af_pfxpostpone)
|
|
344
|
3221 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
3222 }
|
|
236
|
3223 else
|
|
|
3224 tp = &aff->af_suff;
|
|
300
|
3225 aff_todo = atoi((char *)items[3]);
|
|
324
|
3226 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
3227 if (!HASHITEM_EMPTY(hi))
|
|
300
|
3228 {
|
|
236
|
3229 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
3230 fname, lnum, items[1]);
|
|
300
|
3231 aff_todo = 0;
|
|
|
3232 }
|
|
236
|
3233 else
|
|
|
3234 hash_add(tp, cur_aff->ah_key);
|
|
|
3235 }
|
|
|
3236 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3237 || STRCMP(items[0], "SFX") == 0)
|
|
|
3238 && aff_todo > 0
|
|
|
3239 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
3240 && itemcnt >= 5)
|
|
236
|
3241 {
|
|
|
3242 affentry_T *aff_entry;
|
|
366
|
3243 int rare = FALSE;
|
|
|
3244 int lasti = 5;
|
|
|
3245
|
|
|
3246 /* Check for "rare" after the other info. */
|
|
|
3247 if (itemcnt > 5 && STRICMP(items[5], "rare") == 0)
|
|
|
3248 {
|
|
|
3249 rare = TRUE;
|
|
|
3250 lasti = 6;
|
|
|
3251 }
|
|
236
|
3252
|
|
334
|
3253 /* Myspell allows extra text after the item, but that might
|
|
|
3254 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
3255 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
3256 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
366
|
3257 fname, lnum, items[lasti]);
|
|
334
|
3258
|
|
236
|
3259 /* New item for an affix letter. */
|
|
|
3260 --aff_todo;
|
|
300
|
3261 aff_entry = (affentry_T *)getroom(&spin->si_blocks,
|
|
|
3262 sizeof(affentry_T));
|
|
236
|
3263 if (aff_entry == NULL)
|
|
|
3264 break;
|
|
366
|
3265 aff_entry->ae_rare = rare;
|
|
240
|
3266
|
|
236
|
3267 if (STRCMP(items[2], "0") != 0)
|
|
300
|
3268 aff_entry->ae_chop = getroom_save(&spin->si_blocks,
|
|
|
3269 items[2]);
|
|
236
|
3270 if (STRCMP(items[3], "0") != 0)
|
|
300
|
3271 aff_entry->ae_add = getroom_save(&spin->si_blocks,
|
|
|
3272 items[3]);
|
|
236
|
3273
|
|
300
|
3274 /* Don't use an affix entry with non-ASCII characters when
|
|
|
3275 * "spin->si_ascii" is TRUE. */
|
|
|
3276 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
3277 || has_non_ascii(aff_entry->ae_add)))
|
|
|
3278 {
|
|
|
3279 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
3280 cur_aff->ah_first = aff_entry;
|
|
300
|
3281
|
|
|
3282 if (STRCMP(items[4], ".") != 0)
|
|
|
3283 {
|
|
|
3284 char_u buf[MAXLINELEN];
|
|
|
3285
|
|
|
3286 aff_entry->ae_cond = getroom_save(&spin->si_blocks,
|
|
|
3287 items[4]);
|
|
|
3288 if (*items[0] == 'P')
|
|
|
3289 sprintf((char *)buf, "^%s", items[4]);
|
|
|
3290 else
|
|
|
3291 sprintf((char *)buf, "%s$", items[4]);
|
|
|
3292 aff_entry->ae_prog = vim_regcomp(buf,
|
|
|
3293 RE_MAGIC + RE_STRING);
|
|
|
3294 }
|
|
339
|
3295
|
|
|
3296 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
3297 * for the condition. Use an existing one if possible. */
|
|
|
3298 if (*items[0] == 'P' && aff->af_pfxpostpone
|
|
|
3299 && aff_entry->ae_chop == NULL)
|
|
|
3300 {
|
|
|
3301 int idx;
|
|
|
3302 char_u **pp;
|
|
|
3303
|
|
|
3304 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
3305 --idx)
|
|
|
3306 {
|
|
|
3307 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
3308 if (str_equal(p, aff_entry->ae_cond))
|
|
|
3309 break;
|
|
|
3310 }
|
|
|
3311 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
3312 {
|
|
|
3313 /* Not found, add a new condition. */
|
|
|
3314 idx = spin->si_prefcond.ga_len++;
|
|
|
3315 pp = ((char_u **)spin->si_prefcond.ga_data) + idx;
|
|
|
3316 if (aff_entry->ae_cond == NULL)
|
|
|
3317 *pp = NULL;
|
|
|
3318 else
|
|
|
3319 *pp = getroom_save(&spin->si_blocks,
|
|
|
3320 aff_entry->ae_cond);
|
|
|
3321 }
|
|
|
3322
|
|
|
3323 /* Add the prefix to the prefix tree. */
|
|
|
3324 if (aff_entry->ae_add == NULL)
|
|
|
3325 p = (char_u *)"";
|
|
|
3326 else
|
|
|
3327 p = aff_entry->ae_add;
|
|
366
|
3328 tree_add_word(p, spin->si_prefroot, rare ? -2 : -1,
|
|
|
3329 idx, cur_aff->ah_newID, &spin->si_blocks);
|
|
339
|
3330 }
|
|
240
|
3331 }
|
|
236
|
3332 }
|
|
255
|
3333 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
3334 {
|
|
|
3335 if (fol != NULL)
|
|
|
3336 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
3337 fname, lnum);
|
|
|
3338 else
|
|
|
3339 fol = vim_strsave(items[1]);
|
|
|
3340 }
|
|
|
3341 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
3342 {
|
|
|
3343 if (low != NULL)
|
|
|
3344 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
3345 fname, lnum);
|
|
|
3346 else
|
|
|
3347 low = vim_strsave(items[1]);
|
|
|
3348 }
|
|
|
3349 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
3350 {
|
|
|
3351 if (upp != NULL)
|
|
|
3352 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
3353 fname, lnum);
|
|
|
3354 else
|
|
|
3355 upp = vim_strsave(items[1]);
|
|
|
3356 }
|
|
236
|
3357 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
3358 {
|
|
236
|
3359 /* Ignore REP count */;
|
|
323
|
3360 if (!isdigit(*items[1]))
|
|
|
3361 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
3362 fname, lnum);
|
|
|
3363 }
|
|
236
|
3364 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 3)
|
|
|
3365 {
|
|
|
3366 /* REP item */
|
|
323
|
3367 if (do_rep)
|
|
|
3368 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
3369 }
|
|
|
3370 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
3371 {
|
|
|
3372 /* MAP item or count */
|
|
|
3373 if (!found_map)
|
|
|
3374 {
|
|
|
3375 /* First line contains the count. */
|
|
|
3376 found_map = TRUE;
|
|
|
3377 if (!isdigit(*items[1]))
|
|
|
3378 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
3379 fname, lnum);
|
|
|
3380 }
|
|
|
3381 else if (do_map)
|
|
|
3382 {
|
|
346
|
3383 int c;
|
|
|
3384
|
|
|
3385 /* Check that every character appears only once. */
|
|
|
3386 for (p = items[1]; *p != NUL; )
|
|
|
3387 {
|
|
|
3388 #ifdef FEAT_MBYTE
|
|
|
3389 c = mb_ptr2char_adv(&p);
|
|
|
3390 #else
|
|
|
3391 c = *p++;
|
|
|
3392 #endif
|
|
|
3393 if ((spin->si_map.ga_len > 0
|
|
|
3394 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
3395 != NULL)
|
|
|
3396 || vim_strchr(p, c) != NULL)
|
|
|
3397 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
3398 fname, lnum);
|
|
|
3399 }
|
|
|
3400
|
|
323
|
3401 /* We simply concatenate all the MAP strings, separated by
|
|
|
3402 * slashes. */
|
|
|
3403 ga_concat(&spin->si_map, items[1]);
|
|
|
3404 ga_append(&spin->si_map, '/');
|
|
|
3405 }
|
|
|
3406 }
|
|
|
3407 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
3408 {
|
|
|
3409 if (do_sal)
|
|
|
3410 {
|
|
|
3411 /* SAL item (sounds-a-like)
|
|
|
3412 * Either one of the known keys or a from-to pair. */
|
|
|
3413 if (STRCMP(items[1], "followup") == 0)
|
|
|
3414 spin->si_followup = sal_to_bool(items[2]);
|
|
|
3415 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
3416 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
3417 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
3418 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
3419 else
|
|
|
3420 /* when "to" is "_" it means empty */
|
|
|
3421 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
3422 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
3423 : items[2]);
|
|
|
3424 }
|
|
236
|
3425 }
|
|
375
|
3426 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
3427 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
3428 {
|
|
|
3429 if (do_sofo)
|
|
|
3430 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
3431 }
|
|
|
3432 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
3433 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
3434 {
|
|
|
3435 if (do_sofo)
|
|
|
3436 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
3437 }
|
|
300
|
3438 else
|
|
236
|
3439 smsg((char_u *)_("Unrecognized item in %s line %d: %s"),
|
|
|
3440 fname, lnum, items[0]);
|
|
|
3441 }
|
|
|
3442 }
|
|
|
3443
|
|
375
|
3444 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
3445 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
3446 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
3447 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
3448 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
3449
|
|
255
|
3450 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
3451 {
|
|
351
|
3452 if (spin->si_clear_chartab)
|
|
|
3453 {
|
|
|
3454 /* Clear the char type tables, don't want to use any of the
|
|
|
3455 * currently used spell properties. */
|
|
|
3456 init_spell_chartab();
|
|
|
3457 spin->si_clear_chartab = FALSE;
|
|
|
3458 }
|
|
|
3459
|
|
316
|
3460 /*
|
|
|
3461 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
3462 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
3463 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
3464 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
3465 */
|
|
|
3466 if (!spin->si_ascii
|
|
|
3467 #ifdef FEAT_MBYTE
|
|
|
3468 && !enc_utf8
|
|
|
3469 #endif
|
|
|
3470 )
|
|
260
|
3471 {
|
|
|
3472 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
3473 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
3474 else
|
|
316
|
3475 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
3476 }
|
|
255
|
3477
|
|
|
3478 vim_free(fol);
|
|
|
3479 vim_free(low);
|
|
|
3480 vim_free(upp);
|
|
|
3481 }
|
|
|
3482
|
|
236
|
3483 vim_free(pc);
|
|
|
3484 fclose(fd);
|
|
|
3485 return aff;
|
|
|
3486 }
|
|
|
3487
|
|
|
3488 /*
|
|
339
|
3489 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
3490 * NULL as equal.
|
|
|
3491 */
|
|
|
3492 static int
|
|
|
3493 str_equal(s1, s2)
|
|
|
3494 char_u *s1;
|
|
|
3495 char_u *s2;
|
|
|
3496 {
|
|
|
3497 if (s1 == NULL || s2 == NULL)
|
|
|
3498 return s1 == s2;
|
|
|
3499 return STRCMP(s1, s2) == 0;
|
|
|
3500 }
|
|
|
3501
|
|
|
3502 /*
|
|
323
|
3503 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
3504 * They are stored case-folded.
|
|
|
3505 */
|
|
|
3506 static void
|
|
|
3507 add_fromto(spin, gap, from, to)
|
|
|
3508 spellinfo_T *spin;
|
|
|
3509 garray_T *gap;
|
|
|
3510 char_u *from;
|
|
|
3511 char_u *to;
|
|
|
3512 {
|
|
|
3513 fromto_T *ftp;
|
|
|
3514 char_u word[MAXWLEN];
|
|
|
3515
|
|
|
3516 if (ga_grow(gap, 1) == OK)
|
|
|
3517 {
|
|
|
3518 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
3519 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
|
3520 ftp->ft_from = getroom_save(&spin->si_blocks, word);
|
|
|
3521 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
|
3522 ftp->ft_to = getroom_save(&spin->si_blocks, word);
|
|
|
3523 ++gap->ga_len;
|
|
|
3524 }
|
|
|
3525 }
|
|
|
3526
|
|
|
3527 /*
|
|
|
3528 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
3529 */
|
|
|
3530 static int
|
|
|
3531 sal_to_bool(s)
|
|
|
3532 char_u *s;
|
|
|
3533 {
|
|
|
3534 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
3535 }
|
|
|
3536
|
|
|
3537 /*
|
|
240
|
3538 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
3539 * When "s" is NULL FALSE is returned.
|
|
|
3540 */
|
|
|
3541 static int
|
|
|
3542 has_non_ascii(s)
|
|
|
3543 char_u *s;
|
|
|
3544 {
|
|
|
3545 char_u *p;
|
|
|
3546
|
|
|
3547 if (s != NULL)
|
|
|
3548 for (p = s; *p != NUL; ++p)
|
|
|
3549 if (*p >= 128)
|
|
|
3550 return TRUE;
|
|
|
3551 return FALSE;
|
|
|
3552 }
|
|
|
3553
|
|
|
3554 /*
|
|
236
|
3555 * Free the structure filled by spell_read_aff().
|
|
|
3556 */
|
|
|
3557 static void
|
|
|
3558 spell_free_aff(aff)
|
|
|
3559 afffile_T *aff;
|
|
|
3560 {
|
|
|
3561 hashtab_T *ht;
|
|
|
3562 hashitem_T *hi;
|
|
|
3563 int todo;
|
|
|
3564 affheader_T *ah;
|
|
300
|
3565 affentry_T *ae;
|
|
236
|
3566
|
|
|
3567 vim_free(aff->af_enc);
|
|
|
3568
|
|
339
|
3569 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
3570 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
3571 {
|
|
|
3572 todo = ht->ht_used;
|
|
|
3573 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
3574 {
|
|
|
3575 if (!HASHITEM_EMPTY(hi))
|
|
|
3576 {
|
|
|
3577 --todo;
|
|
|
3578 ah = HI2AH(hi);
|
|
300
|
3579 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
3580 vim_free(ae->ae_prog);
|
|
236
|
3581 }
|
|
|
3582 }
|
|
|
3583 if (ht == &aff->af_suff)
|
|
|
3584 break;
|
|
|
3585 }
|
|
300
|
3586
|
|
236
|
3587 hash_clear(&aff->af_pref);
|
|
|
3588 hash_clear(&aff->af_suff);
|
|
|
3589 }
|
|
|
3590
|
|
|
3591 /*
|
|
300
|
3592 * Read dictionary file "fname".
|
|
236
|
3593 * Returns OK or FAIL;
|
|
|
3594 */
|
|
|
3595 static int
|
|
300
|
3596 spell_read_dic(fname, spin, affile)
|
|
236
|
3597 char_u *fname;
|
|
300
|
3598 spellinfo_T *spin;
|
|
|
3599 afffile_T *affile;
|
|
236
|
3600 {
|
|
300
|
3601 hashtab_T ht;
|
|
236
|
3602 char_u line[MAXLINELEN];
|
|
300
|
3603 char_u *afflist;
|
|
339
|
3604 char_u *pfxlist;
|
|
300
|
3605 char_u *dw;
|
|
236
|
3606 char_u *pc;
|
|
|
3607 char_u *w;
|
|
|
3608 int l;
|
|
|
3609 hash_T hash;
|
|
|
3610 hashitem_T *hi;
|
|
|
3611 FILE *fd;
|
|
|
3612 int lnum = 1;
|
|
300
|
3613 int non_ascii = 0;
|
|
|
3614 int retval = OK;
|
|
|
3615 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
3616 int flags;
|
|
236
|
3617
|
|
300
|
3618 /*
|
|
|
3619 * Open the file.
|
|
|
3620 */
|
|
310
|
3621 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3622 if (fd == NULL)
|
|
|
3623 {
|
|
|
3624 EMSG2(_(e_notopen), fname);
|
|
|
3625 return FAIL;
|
|
|
3626 }
|
|
|
3627
|
|
300
|
3628 /* The hashtable is only used to detect duplicated words. */
|
|
|
3629 hash_init(&ht);
|
|
|
3630
|
|
334
|
3631 spin->si_foldwcount = 0;
|
|
|
3632 spin->si_keepwcount = 0;
|
|
|
3633
|
|
310
|
3634 if (spin->si_verbose || p_verbose > 2)
|
|
|
3635 {
|
|
|
3636 if (!spin->si_verbose)
|
|
|
3637 verbose_enter();
|
|
366
|
3638 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
3639 out_flush();
|
|
|
3640 if (!spin->si_verbose)
|
|
|
3641 verbose_leave();
|
|
|
3642 }
|
|
236
|
3643
|
|
|
3644 /* Read and ignore the first line: word count. */
|
|
|
3645 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
3646 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
3647 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
3648
|
|
|
3649 /*
|
|
|
3650 * Read all the lines in the file one by one.
|
|
|
3651 * The words are converted to 'encoding' here, before being added to
|
|
|
3652 * the hashtable.
|
|
|
3653 */
|
|
255
|
3654 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
3655 {
|
|
255
|
3656 line_breakcheck();
|
|
236
|
3657 ++lnum;
|
|
351
|
3658 if (line[0] == '#')
|
|
|
3659 continue; /* comment line */
|
|
236
|
3660
|
|
300
|
3661 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
3662 * the word is kept to allow e.g., "et al.". */
|
|
236
|
3663 l = STRLEN(line);
|
|
|
3664 while (l > 0 && line[l - 1] <= ' ')
|
|
|
3665 --l;
|
|
|
3666 if (l == 0)
|
|
|
3667 continue; /* empty line */
|
|
|
3668 line[l] = NUL;
|
|
|
3669
|
|
|
3670 /* Find the optional affix names. */
|
|
300
|
3671 afflist = vim_strchr(line, '/');
|
|
|
3672 if (afflist != NULL)
|
|
|
3673 *afflist++ = NUL;
|
|
236
|
3674
|
|
300
|
3675 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
3676 if (spin->si_ascii && has_non_ascii(line))
|
|
|
3677 {
|
|
|
3678 ++non_ascii;
|
|
240
|
3679 continue;
|
|
300
|
3680 }
|
|
240
|
3681
|
|
310
|
3682 #ifdef FEAT_MBYTE
|
|
236
|
3683 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
3684 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3685 {
|
|
300
|
3686 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
3687 if (pc == NULL)
|
|
|
3688 {
|
|
|
3689 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3690 fname, lnum, line);
|
|
|
3691 continue;
|
|
|
3692 }
|
|
236
|
3693 w = pc;
|
|
|
3694 }
|
|
|
3695 else
|
|
310
|
3696 #endif
|
|
236
|
3697 {
|
|
|
3698 pc = NULL;
|
|
|
3699 w = line;
|
|
|
3700 }
|
|
|
3701
|
|
339
|
3702 /* This takes time, print a message now and then. */
|
|
|
3703 if (spin->si_verbose && (lnum & 0x3ff) == 0)
|
|
|
3704 {
|
|
|
3705 vim_snprintf((char *)message, sizeof(message),
|
|
|
3706 _("line %6d, word %6d - %s"),
|
|
|
3707 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
3708 msg_start();
|
|
|
3709 msg_puts_long_attr(message, 0);
|
|
|
3710 msg_clr_eos();
|
|
|
3711 msg_didout = FALSE;
|
|
|
3712 msg_col = 0;
|
|
|
3713 out_flush();
|
|
|
3714 }
|
|
|
3715
|
|
300
|
3716 /* Store the word in the hashtable to be able to find duplicates. */
|
|
|
3717 dw = (char_u *)getroom_save(&spin->si_blocks, w);
|
|
236
|
3718 if (dw == NULL)
|
|
300
|
3719 retval = FAIL;
|
|
|
3720 vim_free(pc);
|
|
|
3721 if (retval == FAIL)
|
|
236
|
3722 break;
|
|
|
3723
|
|
300
|
3724 hash = hash_hash(dw);
|
|
|
3725 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
3726 if (!HASHITEM_EMPTY(hi))
|
|
|
3727 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
3728 fname, lnum, dw);
|
|
236
|
3729 else
|
|
300
|
3730 hash_add_item(&ht, hi, dw, hash);
|
|
|
3731
|
|
307
|
3732 flags = 0;
|
|
339
|
3733 pfxlist = NULL;
|
|
307
|
3734 if (afflist != NULL)
|
|
|
3735 {
|
|
|
3736 /* Check for affix name that stands for keep-case word and stands
|
|
|
3737 * for rare word (if defined). */
|
|
310
|
3738 if (affile->af_kep != NUL
|
|
|
3739 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
307
|
3740 flags |= WF_KEEPCAP;
|
|
|
3741 if (affile->af_rar != NUL
|
|
|
3742 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
3743 flags |= WF_RARE;
|
|
346
|
3744 if (affile->af_bad != NUL
|
|
|
3745 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
3746 flags |= WF_BANNED;
|
|
339
|
3747
|
|
|
3748 if (affile->af_pfxpostpone)
|
|
|
3749 /* Need to store the list of prefix IDs with the word. */
|
|
|
3750 pfxlist = get_pfxlist(affile, afflist, &spin->si_blocks);
|
|
307
|
3751 }
|
|
|
3752
|
|
300
|
3753 /* Add the word to the word tree(s). */
|
|
339
|
3754 if (store_word(dw, spin, flags, spin->si_region, pfxlist) == FAIL)
|
|
300
|
3755 retval = FAIL;
|
|
236
|
3756
|
|
300
|
3757 if (afflist != NULL)
|
|
|
3758 {
|
|
|
3759 /* Find all matching suffixes and add the resulting words.
|
|
|
3760 * Additionally do matching prefixes that combine. */
|
|
339
|
3761 if (store_aff_word(dw, spin, afflist, affile,
|
|
307
|
3762 &affile->af_suff, &affile->af_pref,
|
|
339
|
3763 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
3764 retval = FAIL;
|
|
|
3765
|
|
|
3766 /* Find all matching prefixes and add the resulting words. */
|
|
339
|
3767 if (store_aff_word(dw, spin, afflist, affile,
|
|
|
3768 &affile->af_pref, NULL,
|
|
|
3769 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
3770 retval = FAIL;
|
|
|
3771 }
|
|
236
|
3772 }
|
|
|
3773
|
|
300
|
3774 if (spin->si_ascii && non_ascii > 0)
|
|
|
3775 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
3776 non_ascii);
|
|
|
3777 hash_clear(&ht);
|
|
|
3778
|
|
236
|
3779 fclose(fd);
|
|
300
|
3780 return retval;
|
|
236
|
3781 }
|
|
|
3782
|
|
|
3783 /*
|
|
339
|
3784 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
3785 * Used for PFXPOSTPONE.
|
|
|
3786 * Returns a string allocated with getroom(). NULL when there are no prefixes
|
|
|
3787 * or when out of memory.
|
|
|
3788 */
|
|
|
3789 static char_u *
|
|
|
3790 get_pfxlist(affile, afflist, blp)
|
|
|
3791 afffile_T *affile;
|
|
|
3792 char_u *afflist;
|
|
|
3793 sblock_T **blp;
|
|
|
3794 {
|
|
|
3795 char_u *p;
|
|
|
3796 int cnt;
|
|
|
3797 int round;
|
|
|
3798 char_u *res = NULL;
|
|
|
3799 char_u key[2];
|
|
|
3800 hashitem_T *hi;
|
|
|
3801
|
|
|
3802 key[1] = NUL;
|
|
|
3803
|
|
|
3804 /* round 1: count the number of prefix IDs.
|
|
|
3805 * round 2: move prefix IDs to "res" */
|
|
|
3806 for (round = 1; round <= 2; ++round)
|
|
|
3807 {
|
|
|
3808 cnt = 0;
|
|
|
3809 for (p = afflist; *p != NUL; ++p)
|
|
|
3810 {
|
|
|
3811 key[0] = *p;
|
|
|
3812 hi = hash_find(&affile->af_pref, key);
|
|
|
3813 if (!HASHITEM_EMPTY(hi))
|
|
|
3814 {
|
|
|
3815 /* This is a prefix ID, use the new number. */
|
|
|
3816 if (round == 2)
|
|
|
3817 res[cnt] = HI2AH(hi)->ah_newID;
|
|
|
3818 ++cnt;
|
|
|
3819 }
|
|
|
3820 }
|
|
|
3821 if (round == 1 && cnt > 0)
|
|
|
3822 res = getroom(blp, cnt + 1);
|
|
|
3823 if (res == NULL)
|
|
|
3824 break;
|
|
|
3825 }
|
|
|
3826
|
|
|
3827 if (res != NULL)
|
|
|
3828 res[cnt] = NUL;
|
|
|
3829 return res;
|
|
|
3830 }
|
|
|
3831
|
|
|
3832 /*
|
|
300
|
3833 * Apply affixes to a word and store the resulting words.
|
|
|
3834 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
3835 * prefixes or suffixes.
|
|
|
3836 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
3837 * the resulting words for combining affixes.
|
|
|
3838 *
|
|
|
3839 * Returns FAIL when out of memory.
|
|
236
|
3840 */
|
|
300
|
3841 static int
|
|
339
|
3842 store_aff_word(word, spin, afflist, affile, ht, xht, comb, flags, pfxlist)
|
|
300
|
3843 char_u *word; /* basic word start */
|
|
|
3844 spellinfo_T *spin; /* spell info */
|
|
|
3845 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
3846 afffile_T *affile;
|
|
300
|
3847 hashtab_T *ht;
|
|
|
3848 hashtab_T *xht;
|
|
|
3849 int comb; /* only use affixes that combine */
|
|
307
|
3850 int flags; /* flags for the word */
|
|
339
|
3851 char_u *pfxlist; /* list of prefix IDs */
|
|
236
|
3852 {
|
|
|
3853 int todo;
|
|
|
3854 hashitem_T *hi;
|
|
300
|
3855 affheader_T *ah;
|
|
|
3856 affentry_T *ae;
|
|
|
3857 regmatch_T regmatch;
|
|
|
3858 char_u newword[MAXWLEN];
|
|
|
3859 int retval = OK;
|
|
|
3860 int i;
|
|
|
3861 char_u *p;
|
|
366
|
3862 int use_flags;
|
|
236
|
3863
|
|
300
|
3864 todo = ht->ht_used;
|
|
|
3865 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
3866 {
|
|
|
3867 if (!HASHITEM_EMPTY(hi))
|
|
|
3868 {
|
|
|
3869 --todo;
|
|
300
|
3870 ah = HI2AH(hi);
|
|
236
|
3871
|
|
300
|
3872 /* Check that the affix combines, if required, and that the word
|
|
|
3873 * supports this affix. */
|
|
|
3874 if ((!comb || ah->ah_combine)
|
|
|
3875 && vim_strchr(afflist, *ah->ah_key) != NULL)
|
|
236
|
3876 {
|
|
300
|
3877 /* Loop over all affix entries with this name. */
|
|
|
3878 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
3879 {
|
|
300
|
3880 /* Check the condition. It's not logical to match case
|
|
|
3881 * here, but it is required for compatibility with
|
|
339
|
3882 * Myspell.
|
|
|
3883 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
3884 * prefixes with a chop string. */
|
|
300
|
3885 regmatch.regprog = ae->ae_prog;
|
|
|
3886 regmatch.rm_ic = FALSE;
|
|
339
|
3887 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
3888 || ae->ae_chop != NULL)
|
|
|
3889 && (ae->ae_prog == NULL
|
|
|
3890 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
3891 {
|
|
|
3892 /* Match. Remove the chop and add the affix. */
|
|
|
3893 if (xht == NULL)
|
|
240
|
3894 {
|
|
300
|
3895 /* prefix: chop/add at the start of the word */
|
|
|
3896 if (ae->ae_add == NULL)
|
|
|
3897 *newword = NUL;
|
|
|
3898 else
|
|
|
3899 STRCPY(newword, ae->ae_add);
|
|
|
3900 p = word;
|
|
|
3901 if (ae->ae_chop != NULL)
|
|
310
|
3902 {
|
|
300
|
3903 /* Skip chop string. */
|
|
310
|
3904 #ifdef FEAT_MBYTE
|
|
|
3905 if (has_mbyte)
|
|
324
|
3906 {
|
|
310
|
3907 i = mb_charlen(ae->ae_chop);
|
|
324
|
3908 for ( ; i > 0; --i)
|
|
|
3909 mb_ptr_adv(p);
|
|
|
3910 }
|
|
310
|
3911 else
|
|
|
3912 #endif
|
|
324
|
3913 p += STRLEN(ae->ae_chop);
|
|
310
|
3914 }
|
|
300
|
3915 STRCAT(newword, p);
|
|
|
3916 }
|
|
|
3917 else
|
|
|
3918 {
|
|
|
3919 /* suffix: chop/add at the end of the word */
|
|
|
3920 STRCPY(newword, word);
|
|
|
3921 if (ae->ae_chop != NULL)
|
|
|
3922 {
|
|
|
3923 /* Remove chop string. */
|
|
|
3924 p = newword + STRLEN(newword);
|
|
376
|
3925 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
3926 for ( ; i > 0; --i)
|
|
300
|
3927 mb_ptr_back(newword, p);
|
|
|
3928 *p = NUL;
|
|
|
3929 }
|
|
|
3930 if (ae->ae_add != NULL)
|
|
|
3931 STRCAT(newword, ae->ae_add);
|
|
240
|
3932 }
|
|
|
3933
|
|
366
|
3934 /* Obey the "rare" flag of the affix. */
|
|
|
3935 if (ae->ae_rare)
|
|
|
3936 use_flags = flags | WF_RARE;
|
|
|
3937 else
|
|
|
3938 use_flags = flags;
|
|
|
3939
|
|
300
|
3940 /* Store the modified word. */
|
|
366
|
3941 if (store_word(newword, spin, use_flags,
|
|
|
3942 spin->si_region, pfxlist) == FAIL)
|
|
300
|
3943 retval = FAIL;
|
|
236
|
3944
|
|
300
|
3945 /* When added a suffix and combining is allowed also
|
|
|
3946 * try adding prefixes additionally. */
|
|
|
3947 if (xht != NULL && ah->ah_combine)
|
|
339
|
3948 if (store_aff_word(newword, spin, afflist, affile,
|
|
366
|
3949 xht, NULL, TRUE, use_flags, pfxlist)
|
|
|
3950 == FAIL)
|
|
300
|
3951 retval = FAIL;
|
|
236
|
3952 }
|
|
|
3953 }
|
|
|
3954 }
|
|
|
3955 }
|
|
|
3956 }
|
|
|
3957
|
|
|
3958 return retval;
|
|
|
3959 }
|
|
|
3960
|
|
|
3961 /*
|
|
300
|
3962 * Read a file with a list of words.
|
|
236
|
3963 */
|
|
|
3964 static int
|
|
300
|
3965 spell_read_wordfile(fname, spin)
|
|
|
3966 char_u *fname;
|
|
|
3967 spellinfo_T *spin;
|
|
236
|
3968 {
|
|
300
|
3969 FILE *fd;
|
|
|
3970 long lnum = 0;
|
|
|
3971 char_u rline[MAXLINELEN];
|
|
|
3972 char_u *line;
|
|
|
3973 char_u *pc = NULL;
|
|
381
|
3974 char_u *p;
|
|
300
|
3975 int l;
|
|
|
3976 int retval = OK;
|
|
|
3977 int did_word = FALSE;
|
|
|
3978 int non_ascii = 0;
|
|
307
|
3979 int flags;
|
|
316
|
3980 int regionmask;
|
|
236
|
3981
|
|
300
|
3982 /*
|
|
|
3983 * Open the file.
|
|
|
3984 */
|
|
310
|
3985 fd = mch_fopen((char *)fname, "r");
|
|
300
|
3986 if (fd == NULL)
|
|
236
|
3987 {
|
|
300
|
3988 EMSG2(_(e_notopen), fname);
|
|
|
3989 return FAIL;
|
|
236
|
3990 }
|
|
|
3991
|
|
310
|
3992 if (spin->si_verbose || p_verbose > 2)
|
|
|
3993 {
|
|
|
3994 if (!spin->si_verbose)
|
|
|
3995 verbose_enter();
|
|
366
|
3996 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
3997 out_flush();
|
|
|
3998 if (!spin->si_verbose)
|
|
|
3999 verbose_leave();
|
|
|
4000 }
|
|
300
|
4001
|
|
|
4002 /*
|
|
|
4003 * Read all the lines in the file one by one.
|
|
|
4004 */
|
|
|
4005 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
4006 {
|
|
|
4007 line_breakcheck();
|
|
|
4008 ++lnum;
|
|
|
4009
|
|
|
4010 /* Skip comment lines. */
|
|
|
4011 if (*rline == '#')
|
|
|
4012 continue;
|
|
|
4013
|
|
|
4014 /* Remove CR, LF and white space from the end. */
|
|
|
4015 l = STRLEN(rline);
|
|
|
4016 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
4017 --l;
|
|
|
4018 if (l == 0)
|
|
|
4019 continue; /* empty or blank line */
|
|
|
4020 rline[l] = NUL;
|
|
|
4021
|
|
|
4022 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
4023 vim_free(pc);
|
|
310
|
4024 #ifdef FEAT_MBYTE
|
|
300
|
4025 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
4026 {
|
|
|
4027 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
4028 if (pc == NULL)
|
|
|
4029 {
|
|
|
4030 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4031 fname, lnum, rline);
|
|
|
4032 continue;
|
|
|
4033 }
|
|
|
4034 line = pc;
|
|
|
4035 }
|
|
|
4036 else
|
|
310
|
4037 #endif
|
|
300
|
4038 {
|
|
|
4039 pc = NULL;
|
|
|
4040 line = rline;
|
|
|
4041 }
|
|
|
4042
|
|
307
|
4043 if (*line == '/')
|
|
300
|
4044 {
|
|
307
|
4045 ++line;
|
|
|
4046 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
4047 {
|
|
|
4048 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
4049 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
4050 fname, lnum, line - 1);
|
|
300
|
4051 else if (did_word)
|
|
316
|
4052 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
4053 fname, lnum, line - 1);
|
|
300
|
4054 else
|
|
|
4055 {
|
|
310
|
4056 #ifdef FEAT_MBYTE
|
|
|
4057 char_u *enc;
|
|
|
4058
|
|
300
|
4059 /* Setup for conversion to 'encoding'. */
|
|
316
|
4060 line += 10;
|
|
|
4061 enc = enc_canonize(line);
|
|
300
|
4062 if (enc != NULL && !spin->si_ascii
|
|
|
4063 && convert_setup(&spin->si_conv, enc,
|
|
|
4064 p_enc) == FAIL)
|
|
|
4065 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
4066 fname, line, p_enc);
|
|
300
|
4067 vim_free(enc);
|
|
375
|
4068 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4069 #else
|
|
|
4070 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4071 #endif
|
|
300
|
4072 }
|
|
307
|
4073 continue;
|
|
300
|
4074 }
|
|
307
|
4075
|
|
316
|
4076 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
4077 {
|
|
|
4078 if (spin->si_region_count > 1)
|
|
|
4079 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
4080 fname, lnum, line);
|
|
|
4081 else
|
|
|
4082 {
|
|
|
4083 line += 8;
|
|
|
4084 if (STRLEN(line) > 16)
|
|
|
4085 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
4086 fname, lnum, line);
|
|
|
4087 else
|
|
|
4088 {
|
|
|
4089 spin->si_region_count = STRLEN(line) / 2;
|
|
|
4090 STRCPY(spin->si_region_name, line);
|
|
|
4091 }
|
|
|
4092 }
|
|
|
4093 continue;
|
|
|
4094 }
|
|
|
4095
|
|
381
|
4096 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
4097 fname, lnum, line - 1);
|
|
|
4098 continue;
|
|
|
4099 }
|
|
|
4100
|
|
|
4101 flags = 0;
|
|
|
4102 regionmask = spin->si_region;
|
|
|
4103
|
|
|
4104 /* Check for flags and region after a slash. */
|
|
|
4105 p = vim_strchr(line, '/');
|
|
|
4106 if (p != NULL)
|
|
|
4107 {
|
|
|
4108 *p++ = NUL;
|
|
|
4109 while (*p != NUL)
|
|
|
4110 {
|
|
|
4111 if (*p == '=') /* keep-case word */
|
|
|
4112 flags |= WF_KEEPCAP;
|
|
|
4113 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
4114 flags |= WF_BANNED;
|
|
|
4115 else if (*p == '?') /* Rare word. */
|
|
|
4116 flags |= WF_RARE;
|
|
|
4117 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
4118 {
|
|
381
|
4119 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
4120 regionmask = 0;
|
|
|
4121 flags |= WF_REGION;
|
|
|
4122
|
|
|
4123 l = *p - '0';
|
|
316
|
4124 if (l > spin->si_region_count)
|
|
|
4125 {
|
|
|
4126 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
4127 fname, lnum, p);
|
|
316
|
4128 break;
|
|
|
4129 }
|
|
|
4130 regionmask |= 1 << (l - 1);
|
|
|
4131 }
|
|
381
|
4132 else
|
|
|
4133 {
|
|
|
4134 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
4135 fname, lnum, p);
|
|
|
4136 break;
|
|
|
4137 }
|
|
|
4138 ++p;
|
|
307
|
4139 }
|
|
300
|
4140 }
|
|
|
4141
|
|
|
4142 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4143 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4144 {
|
|
|
4145 ++non_ascii;
|
|
|
4146 continue;
|
|
|
4147 }
|
|
|
4148
|
|
|
4149 /* Normal word: store it. */
|
|
339
|
4150 if (store_word(line, spin, flags, regionmask, NULL) == FAIL)
|
|
300
|
4151 {
|
|
|
4152 retval = FAIL;
|
|
|
4153 break;
|
|
|
4154 }
|
|
|
4155 did_word = TRUE;
|
|
|
4156 }
|
|
|
4157
|
|
|
4158 vim_free(pc);
|
|
|
4159 fclose(fd);
|
|
|
4160
|
|
310
|
4161 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
4162 {
|
|
|
4163 if (p_verbose > 2)
|
|
|
4164 verbose_enter();
|
|
300
|
4165 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4166 non_ascii);
|
|
310
|
4167 if (p_verbose > 2)
|
|
|
4168 verbose_leave();
|
|
|
4169 }
|
|
300
|
4170 return retval;
|
|
236
|
4171 }
|
|
|
4172
|
|
|
4173 /*
|
|
300
|
4174 * Get part of an sblock_T, "len" bytes long.
|
|
|
4175 * This avoids calling free() for every little struct we use.
|
|
|
4176 * The memory is cleared to all zeros.
|
|
|
4177 * Returns NULL when out of memory.
|
|
|
4178 */
|
|
|
4179 static void *
|
|
|
4180 getroom(blp, len)
|
|
|
4181 sblock_T **blp;
|
|
|
4182 size_t len; /* length needed */
|
|
|
4183 {
|
|
|
4184 char_u *p;
|
|
|
4185 sblock_T *bl = *blp;
|
|
|
4186
|
|
|
4187 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
4188 {
|
|
|
4189 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
4190 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
4191 if (bl == NULL)
|
|
|
4192 return NULL;
|
|
|
4193 bl->sb_next = *blp;
|
|
|
4194 *blp = bl;
|
|
|
4195 bl->sb_used = 0;
|
|
|
4196 }
|
|
|
4197
|
|
|
4198 p = bl->sb_data + bl->sb_used;
|
|
|
4199 bl->sb_used += len;
|
|
|
4200
|
|
|
4201 return p;
|
|
|
4202 }
|
|
|
4203
|
|
|
4204 /*
|
|
|
4205 * Make a copy of a string into memory allocated with getroom().
|
|
|
4206 */
|
|
|
4207 static char_u *
|
|
|
4208 getroom_save(blp, s)
|
|
|
4209 sblock_T **blp;
|
|
|
4210 char_u *s;
|
|
|
4211 {
|
|
|
4212 char_u *sc;
|
|
|
4213
|
|
|
4214 sc = (char_u *)getroom(blp, STRLEN(s) + 1);
|
|
|
4215 if (sc != NULL)
|
|
|
4216 STRCPY(sc, s);
|
|
|
4217 return sc;
|
|
|
4218 }
|
|
|
4219
|
|
|
4220
|
|
|
4221 /*
|
|
|
4222 * Free the list of allocated sblock_T.
|
|
236
|
4223 */
|
|
|
4224 static void
|
|
300
|
4225 free_blocks(bl)
|
|
|
4226 sblock_T *bl;
|
|
236
|
4227 {
|
|
300
|
4228 sblock_T *next;
|
|
236
|
4229
|
|
300
|
4230 while (bl != NULL)
|
|
236
|
4231 {
|
|
300
|
4232 next = bl->sb_next;
|
|
|
4233 vim_free(bl);
|
|
|
4234 bl = next;
|
|
236
|
4235 }
|
|
|
4236 }
|
|
|
4237
|
|
|
4238 /*
|
|
300
|
4239 * Allocate the root of a word tree.
|
|
236
|
4240 */
|
|
300
|
4241 static wordnode_T *
|
|
|
4242 wordtree_alloc(blp)
|
|
|
4243 sblock_T **blp;
|
|
236
|
4244 {
|
|
300
|
4245 return (wordnode_T *)getroom(blp, sizeof(wordnode_T));
|
|
236
|
4246 }
|
|
|
4247
|
|
|
4248 /*
|
|
300
|
4249 * Store a word in the tree(s).
|
|
307
|
4250 * Always store it in the case-folded tree. A keep-case word can also be used
|
|
|
4251 * with all caps.
|
|
300
|
4252 * For a keep-case word also store it in the keep-case tree.
|
|
339
|
4253 * When "pfxlist" is not NULL store the word for each prefix ID.
|
|
236
|
4254 */
|
|
|
4255 static int
|
|
339
|
4256 store_word(word, spin, flags, region, pfxlist)
|
|
300
|
4257 char_u *word;
|
|
|
4258 spellinfo_T *spin;
|
|
307
|
4259 int flags; /* extra flags, WF_BANNED */
|
|
316
|
4260 int region; /* supported region(s) */
|
|
339
|
4261 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
236
|
4262 {
|
|
300
|
4263 int len = STRLEN(word);
|
|
|
4264 int ct = captype(word, word + len);
|
|
|
4265 char_u foldword[MAXWLEN];
|
|
339
|
4266 int res = OK;
|
|
|
4267 char_u *p;
|
|
236
|
4268
|
|
323
|
4269 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
4270 for (p = pfxlist; res == OK; ++p)
|
|
|
4271 {
|
|
|
4272 res = tree_add_word(foldword, spin->si_foldroot, ct | flags,
|
|
|
4273 region, p == NULL ? 0 : *p, &spin->si_blocks);
|
|
|
4274 if (p == NULL || *p == NUL)
|
|
|
4275 break;
|
|
|
4276 }
|
|
334
|
4277 ++spin->si_foldwcount;
|
|
307
|
4278
|
|
|
4279 if (res == OK && (ct == WF_KEEPCAP || flags & WF_KEEPCAP))
|
|
334
|
4280 {
|
|
339
|
4281 for (p = pfxlist; res == OK; ++p)
|
|
|
4282 {
|
|
|
4283 res = tree_add_word(word, spin->si_keeproot, flags,
|
|
|
4284 region, p == NULL ? 0 : *p, &spin->si_blocks);
|
|
|
4285 if (p == NULL || *p == NUL)
|
|
|
4286 break;
|
|
|
4287 }
|
|
334
|
4288 ++spin->si_keepwcount;
|
|
|
4289 }
|
|
300
|
4290 return res;
|
|
236
|
4291 }
|
|
|
4292
|
|
|
4293 /*
|
|
300
|
4294 * Add word "word" to a word tree at "root".
|
|
366
|
4295 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
4296 * "rare" and "region" is the condition nr.
|
|
255
|
4297 * Returns FAIL when out of memory.
|
|
236
|
4298 */
|
|
255
|
4299 static int
|
|
339
|
4300 tree_add_word(word, root, flags, region, prefixID, blp)
|
|
300
|
4301 char_u *word;
|
|
|
4302 wordnode_T *root;
|
|
|
4303 int flags;
|
|
|
4304 int region;
|
|
339
|
4305 int prefixID;
|
|
300
|
4306 sblock_T **blp;
|
|
236
|
4307 {
|
|
300
|
4308 wordnode_T *node = root;
|
|
|
4309 wordnode_T *np;
|
|
|
4310 wordnode_T **prev = NULL;
|
|
|
4311 int i;
|
|
255
|
4312
|
|
300
|
4313 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
4314 for (i = 0; ; ++i)
|
|
255
|
4315 {
|
|
300
|
4316 /* Look for the sibling that has the same character. They are sorted
|
|
|
4317 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
4318 * higher byte value. For zero bytes (end of word) the sorting is
|
|
|
4319 * done on flags and then on prefixID
|
|
300
|
4320 */
|
|
339
|
4321 while (node != NULL
|
|
|
4322 && (node->wn_byte < word[i]
|
|
|
4323 || (node->wn_byte == NUL
|
|
|
4324 && (flags < 0
|
|
|
4325 ? node->wn_prefixID < prefixID
|
|
|
4326 : node->wn_flags < (flags & 0xff)
|
|
|
4327 || (node->wn_flags == (flags & 0xff)
|
|
|
4328 && node->wn_prefixID < prefixID)))))
|
|
236
|
4329 {
|
|
300
|
4330 prev = &node->wn_sibling;
|
|
|
4331 node = *prev;
|
|
236
|
4332 }
|
|
339
|
4333 if (node == NULL
|
|
|
4334 || node->wn_byte != word[i]
|
|
|
4335 || (word[i] == NUL
|
|
|
4336 && (flags < 0
|
|
|
4337 || node->wn_flags != (flags & 0xff)
|
|
|
4338 || node->wn_prefixID != prefixID)))
|
|
255
|
4339 {
|
|
300
|
4340 /* Allocate a new node. */
|
|
|
4341 np = (wordnode_T *)getroom(blp, sizeof(wordnode_T));
|
|
|
4342 if (np == NULL)
|
|
|
4343 return FAIL;
|
|
|
4344 np->wn_byte = word[i];
|
|
|
4345 *prev = np;
|
|
|
4346 np->wn_sibling = node;
|
|
|
4347 node = np;
|
|
255
|
4348 }
|
|
300
|
4349
|
|
|
4350 if (word[i] == NUL)
|
|
|
4351 {
|
|
|
4352 node->wn_flags = flags;
|
|
|
4353 node->wn_region |= region;
|
|
339
|
4354 node->wn_prefixID = prefixID;
|
|
300
|
4355 break;
|
|
|
4356 }
|
|
|
4357 prev = &node->wn_child;
|
|
|
4358 node = *prev;
|
|
255
|
4359 }
|
|
|
4360
|
|
|
4361 return OK;
|
|
236
|
4362 }
|
|
|
4363
|
|
|
4364 /*
|
|
300
|
4365 * Compress a tree: find tails that are identical and can be shared.
|
|
|
4366 */
|
|
|
4367 static void
|
|
310
|
4368 wordtree_compress(root, spin)
|
|
300
|
4369 wordnode_T *root;
|
|
310
|
4370 spellinfo_T *spin;
|
|
300
|
4371 {
|
|
|
4372 hashtab_T ht;
|
|
|
4373 int n;
|
|
|
4374 int tot = 0;
|
|
|
4375
|
|
|
4376 if (root != NULL)
|
|
|
4377 {
|
|
|
4378 hash_init(&ht);
|
|
|
4379 n = node_compress(root, &ht, &tot);
|
|
310
|
4380 if (spin->si_verbose || p_verbose > 2)
|
|
|
4381 {
|
|
|
4382 if (!spin->si_verbose)
|
|
|
4383 verbose_enter();
|
|
|
4384 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
300
|
4385 n, tot, (tot - n) * 100 / tot);
|
|
310
|
4386 if (p_verbose > 2)
|
|
|
4387 verbose_leave();
|
|
|
4388 }
|
|
300
|
4389 hash_clear(&ht);
|
|
|
4390 }
|
|
|
4391 }
|
|
|
4392
|
|
|
4393 /*
|
|
|
4394 * Compress a node, its siblings and its children, depth first.
|
|
|
4395 * Returns the number of compressed nodes.
|
|
236
|
4396 */
|
|
255
|
4397 static int
|
|
300
|
4398 node_compress(node, ht, tot)
|
|
|
4399 wordnode_T *node;
|
|
|
4400 hashtab_T *ht;
|
|
|
4401 int *tot; /* total count of nodes before compressing,
|
|
|
4402 incremented while going through the tree */
|
|
236
|
4403 {
|
|
300
|
4404 wordnode_T *np;
|
|
|
4405 wordnode_T *tp;
|
|
|
4406 wordnode_T *child;
|
|
|
4407 hash_T hash;
|
|
236
|
4408 hashitem_T *hi;
|
|
300
|
4409 int len = 0;
|
|
|
4410 unsigned nr, n;
|
|
|
4411 int compressed = 0;
|
|
236
|
4412
|
|
300
|
4413 /*
|
|
|
4414 * Go through the list of siblings. Compress each child and then try
|
|
|
4415 * finding an identical child to replace it.
|
|
|
4416 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
4417 * but the whole list of siblings, of which the node is the first.
|
|
|
4418 */
|
|
|
4419 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4420 {
|
|
300
|
4421 ++len;
|
|
|
4422 if ((child = np->wn_child) != NULL)
|
|
|
4423 {
|
|
346
|
4424 /* Compress the child. This fills hashkey. */
|
|
300
|
4425 compressed += node_compress(child, ht, tot);
|
|
|
4426
|
|
|
4427 /* Try to find an identical child. */
|
|
346
|
4428 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
4429 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
4430 tp = NULL;
|
|
|
4431 if (!HASHITEM_EMPTY(hi))
|
|
|
4432 {
|
|
|
4433 /* There are children with an identical hash value. Now check
|
|
|
4434 * if there is one that is really identical. */
|
|
346
|
4435 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
4436 if (node_equal(child, tp))
|
|
|
4437 {
|
|
|
4438 /* Found one! Now use that child in place of the
|
|
|
4439 * current one. This means the current child is
|
|
|
4440 * dropped from the tree. */
|
|
|
4441 np->wn_child = tp;
|
|
|
4442 ++compressed;
|
|
|
4443 break;
|
|
|
4444 }
|
|
|
4445 if (tp == NULL)
|
|
|
4446 {
|
|
|
4447 /* No other child with this hash value equals the child of
|
|
|
4448 * the node, add it to the linked list after the first
|
|
|
4449 * item. */
|
|
|
4450 tp = HI2WN(hi);
|
|
346
|
4451 child->wn_u2.next = tp->wn_u2.next;
|
|
|
4452 tp->wn_u2.next = child;
|
|
300
|
4453 }
|
|
|
4454 }
|
|
|
4455 else
|
|
|
4456 /* No other child has this hash value, add it to the
|
|
|
4457 * hashtable. */
|
|
346
|
4458 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
4459 }
|
|
236
|
4460 }
|
|
300
|
4461 *tot += len;
|
|
|
4462
|
|
|
4463 /*
|
|
|
4464 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
4465 * find a lookalike node. This must be done after compressing the sibling
|
|
|
4466 * list, otherwise the hash key would become invalid by the compression.
|
|
|
4467 */
|
|
346
|
4468 node->wn_u1.hashkey[0] = len;
|
|
300
|
4469 nr = 0;
|
|
|
4470 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4471 {
|
|
300
|
4472 if (np->wn_byte == NUL)
|
|
339
|
4473 /* end node: use wn_flags, wn_region and wn_prefixID */
|
|
|
4474 n = np->wn_flags + (np->wn_region << 8) + (np->wn_prefixID << 16);
|
|
300
|
4475 else
|
|
|
4476 /* byte node: use the byte value and the child pointer */
|
|
|
4477 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
4478 nr = nr * 101 + n;
|
|
236
|
4479 }
|
|
300
|
4480
|
|
|
4481 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
4482 n = nr & 0xff;
|
|
346
|
4483 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
4484 n = (nr >> 8) & 0xff;
|
|
346
|
4485 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
4486 n = (nr >> 16) & 0xff;
|
|
346
|
4487 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
4488 n = (nr >> 24) & 0xff;
|
|
346
|
4489 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
4490 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
4491
|
|
|
4492 return compressed;
|
|
|
4493 }
|
|
|
4494
|
|
|
4495 /*
|
|
|
4496 * Return TRUE when two nodes have identical siblings and children.
|
|
|
4497 */
|
|
|
4498 static int
|
|
|
4499 node_equal(n1, n2)
|
|
|
4500 wordnode_T *n1;
|
|
|
4501 wordnode_T *n2;
|
|
|
4502 {
|
|
|
4503 wordnode_T *p1;
|
|
|
4504 wordnode_T *p2;
|
|
|
4505
|
|
|
4506 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
4507 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
4508 if (p1->wn_byte != p2->wn_byte
|
|
|
4509 || (p1->wn_byte == NUL
|
|
|
4510 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
4511 || p1->wn_region != p2->wn_region
|
|
|
4512 || p1->wn_prefixID != p2->wn_prefixID)
|
|
300
|
4513 : (p1->wn_child != p2->wn_child)))
|
|
|
4514 break;
|
|
|
4515
|
|
|
4516 return p1 == NULL && p2 == NULL;
|
|
236
|
4517 }
|
|
|
4518
|
|
|
4519 /*
|
|
|
4520 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
4521 */
|
|
255
|
4522 void
|
|
236
|
4523 put_bytes(fd, nr, len)
|
|
|
4524 FILE *fd;
|
|
|
4525 long_u nr;
|
|
|
4526 int len;
|
|
|
4527 {
|
|
|
4528 int i;
|
|
|
4529
|
|
|
4530 for (i = len - 1; i >= 0; --i)
|
|
|
4531 putc((int)(nr >> (i * 8)), fd);
|
|
|
4532 }
|
|
|
4533
|
|
323
|
4534 static int
|
|
|
4535 #ifdef __BORLANDC__
|
|
|
4536 _RTLENTRYF
|
|
|
4537 #endif
|
|
|
4538 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
4539
|
|
|
4540 /*
|
|
|
4541 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
4542 */
|
|
|
4543 static int
|
|
|
4544 #ifdef __BORLANDC__
|
|
|
4545 _RTLENTRYF
|
|
|
4546 #endif
|
|
|
4547 rep_compare(s1, s2)
|
|
|
4548 const void *s1;
|
|
|
4549 const void *s2;
|
|
|
4550 {
|
|
|
4551 fromto_T *p1 = (fromto_T *)s1;
|
|
|
4552 fromto_T *p2 = (fromto_T *)s2;
|
|
|
4553
|
|
|
4554 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
4555 }
|
|
|
4556
|
|
236
|
4557 /*
|
|
|
4558 * Write the Vim spell file "fname".
|
|
|
4559 */
|
|
|
4560 static void
|
|
316
|
4561 write_vim_spell(fname, spin)
|
|
236
|
4562 char_u *fname;
|
|
300
|
4563 spellinfo_T *spin;
|
|
236
|
4564 {
|
|
300
|
4565 FILE *fd;
|
|
|
4566 int regionmask;
|
|
236
|
4567 int round;
|
|
300
|
4568 wordnode_T *tree;
|
|
|
4569 int nodecount;
|
|
323
|
4570 int i;
|
|
|
4571 int l;
|
|
|
4572 garray_T *gap;
|
|
|
4573 fromto_T *ftp;
|
|
|
4574 char_u *p;
|
|
|
4575 int rr;
|
|
236
|
4576
|
|
310
|
4577 fd = mch_fopen((char *)fname, "w");
|
|
300
|
4578 if (fd == NULL)
|
|
236
|
4579 {
|
|
|
4580 EMSG2(_(e_notopen), fname);
|
|
|
4581 return;
|
|
|
4582 }
|
|
|
4583
|
|
255
|
4584 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
339
|
4585 * <charflagslen> <charflags>
|
|
|
4586 * <fcharslen> <fchars>
|
|
366
|
4587 * <midwordlen> <midword>
|
|
339
|
4588 * <prefcondcnt> <prefcond> ... */
|
|
300
|
4589
|
|
|
4590 /* <fileID> */
|
|
|
4591 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
4592 EMSG(_(e_write));
|
|
236
|
4593
|
|
|
4594 /* write the region names if there is more than one */
|
|
316
|
4595 if (spin->si_region_count > 1)
|
|
236
|
4596 {
|
|
316
|
4597 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
|
4598 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
|
4599 (size_t)1, fd);
|
|
|
4600 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
4601 }
|
|
|
4602 else
|
|
|
4603 {
|
|
300
|
4604 putc(0, fd);
|
|
|
4605 regionmask = 0;
|
|
236
|
4606 }
|
|
|
4607
|
|
323
|
4608 /*
|
|
|
4609 * Write the table with character flags and table for case folding.
|
|
260
|
4610 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
|
4611 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
4612 * 'encoding'.
|
|
|
4613 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
4614 * the table (avoids that it conflicts). File is shorter too.
|
|
|
4615 */
|
|
|
4616 if (spin->si_ascii || spin->si_add)
|
|
260
|
4617 {
|
|
300
|
4618 putc(0, fd);
|
|
|
4619 putc(0, fd);
|
|
|
4620 putc(0, fd);
|
|
260
|
4621 }
|
|
|
4622 else
|
|
300
|
4623 write_spell_chartab(fd);
|
|
255
|
4624
|
|
366
|
4625
|
|
|
4626 if (spin->si_midword == NULL)
|
|
|
4627 put_bytes(fd, 0L, 2); /* <midwordlen> */
|
|
|
4628 else
|
|
|
4629 {
|
|
|
4630 i = STRLEN(spin->si_midword);
|
|
|
4631 put_bytes(fd, (long_u)i, 2); /* <midwordlen> */
|
|
|
4632 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
4633 }
|
|
|
4634
|
|
|
4635
|
|
339
|
4636 /* Write the prefix conditions. */
|
|
|
4637 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
4638
|
|
366
|
4639 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
4640 * <salflags> <salcount> <sal> ...
|
|
|
4641 * <maplen> <mapstr> */
|
|
|
4642
|
|
323
|
4643 /* Sort the REP items. */
|
|
|
4644 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
4645 sizeof(fromto_T), rep_compare);
|
|
|
4646
|
|
375
|
4647 /* round 1: REP items
|
|
|
4648 * round 2: SAL items (unless SOFO is used) */
|
|
323
|
4649 for (round = 1; round <= 2; ++round)
|
|
|
4650 {
|
|
|
4651 if (round == 1)
|
|
|
4652 gap = &spin->si_rep;
|
|
|
4653 else
|
|
|
4654 {
|
|
|
4655 gap = &spin->si_sal;
|
|
|
4656
|
|
|
4657 i = 0;
|
|
|
4658 if (spin->si_followup)
|
|
|
4659 i |= SAL_F0LLOWUP;
|
|
|
4660 if (spin->si_collapse)
|
|
|
4661 i |= SAL_COLLAPSE;
|
|
|
4662 if (spin->si_rem_accents)
|
|
|
4663 i |= SAL_REM_ACCENTS;
|
|
375
|
4664 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
4665 i |= SAL_SOFO;
|
|
323
|
4666 putc(i, fd); /* <salflags> */
|
|
375
|
4667 if (i & SAL_SOFO)
|
|
|
4668 break;
|
|
323
|
4669 }
|
|
|
4670
|
|
|
4671 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
4672 for (i = 0; i < gap->ga_len; ++i)
|
|
|
4673 {
|
|
|
4674 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
4675 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
4676 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
4677 for (rr = 1; rr <= 2; ++rr)
|
|
|
4678 {
|
|
|
4679 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
4680 l = STRLEN(p);
|
|
|
4681 putc(l, fd);
|
|
|
4682 fwrite(p, l, (size_t)1, fd);
|
|
|
4683 }
|
|
|
4684 }
|
|
|
4685 }
|
|
|
4686
|
|
375
|
4687 /* SOFOFROM and SOFOTO */
|
|
|
4688 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
4689 {
|
|
|
4690 put_bytes(fd, 1L, 2); /* <salcount> */
|
|
|
4691
|
|
|
4692 l = STRLEN(spin->si_sofofr);
|
|
|
4693 put_bytes(fd, (long_u)l, 2); /* <salfromlen> */
|
|
|
4694 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <salfrom> */
|
|
|
4695
|
|
|
4696 l = STRLEN(spin->si_sofoto);
|
|
|
4697 put_bytes(fd, (long_u)l, 2); /* <saltolen> */
|
|
|
4698 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <salto> */
|
|
|
4699 }
|
|
|
4700
|
|
323
|
4701 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
|
4702 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
|
4703 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
|
4704 (size_t)1, fd);
|
|
302
|
4705
|
|
236
|
4706 /*
|
|
339
|
4707 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
4708 */
|
|
323
|
4709 spin->si_memtot = 0;
|
|
339
|
4710 for (round = 1; round <= 3; ++round)
|
|
236
|
4711 {
|
|
339
|
4712 if (round == 1)
|
|
|
4713 tree = spin->si_foldroot;
|
|
|
4714 else if (round == 2)
|
|
|
4715 tree = spin->si_keeproot;
|
|
|
4716 else
|
|
|
4717 tree = spin->si_prefroot;
|
|
236
|
4718
|
|
346
|
4719 /* Clear the index and wnode fields in the tree. */
|
|
|
4720 clear_node(tree);
|
|
|
4721
|
|
300
|
4722 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
4723 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
4724 * nodes. */
|
|
346
|
4725 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
4726
|
|
300
|
4727 /* number of nodes in 4 bytes */
|
|
|
4728 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
4729 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
4730
|
|
300
|
4731 /* Write the nodes. */
|
|
346
|
4732 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
4733 }
|
|
|
4734
|
|
300
|
4735 fclose(fd);
|
|
236
|
4736 }
|
|
|
4737
|
|
|
4738 /*
|
|
346
|
4739 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
4740 * children. This is needed because they are a union with other items to save
|
|
|
4741 * space.
|
|
|
4742 */
|
|
|
4743 static void
|
|
|
4744 clear_node(node)
|
|
|
4745 wordnode_T *node;
|
|
|
4746 {
|
|
|
4747 wordnode_T *np;
|
|
|
4748
|
|
|
4749 if (node != NULL)
|
|
|
4750 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4751 {
|
|
|
4752 np->wn_u1.index = 0;
|
|
|
4753 np->wn_u2.wnode = NULL;
|
|
|
4754
|
|
|
4755 if (np->wn_byte != NUL)
|
|
|
4756 clear_node(np->wn_child);
|
|
|
4757 }
|
|
|
4758 }
|
|
|
4759
|
|
|
4760
|
|
|
4761 /*
|
|
300
|
4762 * Dump a word tree at node "node".
|
|
|
4763 *
|
|
|
4764 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
4765 * byte recursively write the children.
|
|
|
4766 *
|
|
|
4767 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
4768 * are made about the indexes (so that we don't have to write them in the
|
|
|
4769 * file).
|
|
236
|
4770 *
|
|
300
|
4771 * Returns the number of nodes used.
|
|
236
|
4772 */
|
|
300
|
4773 static int
|
|
346
|
4774 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
4775 FILE *fd; /* NULL when only counting */
|
|
300
|
4776 wordnode_T *node;
|
|
|
4777 int index;
|
|
|
4778 int regionmask;
|
|
339
|
4779 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
4780 {
|
|
300
|
4781 int newindex = index;
|
|
|
4782 int siblingcount = 0;
|
|
|
4783 wordnode_T *np;
|
|
236
|
4784 int flags;
|
|
300
|
4785
|
|
|
4786 /* If "node" is zero the tree is empty. */
|
|
|
4787 if (node == NULL)
|
|
|
4788 return 0;
|
|
|
4789
|
|
|
4790 /* Store the index where this node is written. */
|
|
346
|
4791 node->wn_u1.index = index;
|
|
236
|
4792
|
|
300
|
4793 /* Count the number of siblings. */
|
|
|
4794 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4795 ++siblingcount;
|
|
236
|
4796
|
|
300
|
4797 /* Write the sibling count. */
|
|
|
4798 if (fd != NULL)
|
|
|
4799 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
4800
|
|
300
|
4801 /* Write each sibling byte and optionally extra info. */
|
|
|
4802 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4803 {
|
|
300
|
4804 if (np->wn_byte == 0)
|
|
|
4805 {
|
|
|
4806 if (fd != NULL)
|
|
|
4807 {
|
|
339
|
4808 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
4809 if (prefixtree)
|
|
300
|
4810 {
|
|
339
|
4811 /* In PREFIXTREE write the required prefixID and the
|
|
|
4812 * associated condition nr (stored in wn_region). */
|
|
366
|
4813 if (np->wn_flags == (char_u)-2)
|
|
|
4814 putc(BY_FLAGS, fd); /* <byte> rare */
|
|
|
4815 else
|
|
|
4816 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
339
|
4817 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
4818 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
4819 }
|
|
|
4820 else
|
|
|
4821 {
|
|
339
|
4822 /* For word trees we write the flag/region items. */
|
|
|
4823 flags = np->wn_flags;
|
|
|
4824 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
4825 flags |= WF_REGION;
|
|
|
4826 if (np->wn_prefixID != 0)
|
|
|
4827 flags |= WF_PFX;
|
|
|
4828 if (flags == 0)
|
|
|
4829 {
|
|
|
4830 /* word without flags or region */
|
|
|
4831 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
4832 }
|
|
|
4833 else
|
|
|
4834 {
|
|
|
4835 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
4836 putc(flags, fd); /* <flags> */
|
|
|
4837 if (flags & WF_REGION)
|
|
|
4838 putc(np->wn_region, fd); /* <region> */
|
|
|
4839 if (flags & WF_PFX)
|
|
|
4840 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
4841 }
|
|
300
|
4842 }
|
|
|
4843 }
|
|
|
4844 }
|
|
|
4845 else
|
|
|
4846 {
|
|
346
|
4847 if (np->wn_child->wn_u1.index != 0
|
|
|
4848 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
4849 {
|
|
|
4850 /* The child is written elsewhere, write the reference. */
|
|
|
4851 if (fd != NULL)
|
|
|
4852 {
|
|
|
4853 putc(BY_INDEX, fd); /* <byte> */
|
|
|
4854 /* <nodeidx> */
|
|
346
|
4855 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
4856 }
|
|
|
4857 }
|
|
346
|
4858 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
4859 /* We will write the child below and give it an index. */
|
|
346
|
4860 np->wn_child->wn_u2.wnode = node;
|
|
236
|
4861
|
|
300
|
4862 if (fd != NULL)
|
|
|
4863 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
4864 {
|
|
|
4865 EMSG(_(e_write));
|
|
|
4866 return 0;
|
|
|
4867 }
|
|
|
4868 }
|
|
236
|
4869 }
|
|
|
4870
|
|
300
|
4871 /* Space used in the array when reading: one for each sibling and one for
|
|
|
4872 * the count. */
|
|
|
4873 newindex += siblingcount + 1;
|
|
249
|
4874
|
|
300
|
4875 /* Recursively dump the children of each sibling. */
|
|
|
4876 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
4877 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
4878 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
4879 prefixtree);
|
|
249
|
4880
|
|
300
|
4881 return newindex;
|
|
236
|
4882 }
|
|
|
4883
|
|
|
4884
|
|
|
4885 /*
|
|
310
|
4886 * ":mkspell [-ascii] outfile infile ..."
|
|
|
4887 * ":mkspell [-ascii] addfile"
|
|
236
|
4888 */
|
|
|
4889 void
|
|
|
4890 ex_mkspell(eap)
|
|
|
4891 exarg_T *eap;
|
|
|
4892 {
|
|
|
4893 int fcount;
|
|
|
4894 char_u **fnames;
|
|
310
|
4895 char_u *arg = eap->arg;
|
|
|
4896 int ascii = FALSE;
|
|
|
4897
|
|
|
4898 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
4899 {
|
|
|
4900 ascii = TRUE;
|
|
|
4901 arg = skipwhite(arg + 6);
|
|
|
4902 }
|
|
|
4903
|
|
|
4904 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
4905 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
4906 {
|
|
323
|
4907 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
4908 FreeWild(fcount, fnames);
|
|
|
4909 }
|
|
|
4910 }
|
|
|
4911
|
|
|
4912 /*
|
|
|
4913 * Create a Vim spell file from one or more word lists.
|
|
|
4914 * "fnames[0]" is the output file name.
|
|
|
4915 * "fnames[fcount - 1]" is the last input file name.
|
|
|
4916 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
4917 * and ".spl" is appended to make the output file name.
|
|
|
4918 */
|
|
|
4919 static void
|
|
323
|
4920 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
4921 int fcount;
|
|
|
4922 char_u **fnames;
|
|
|
4923 int ascii; /* -ascii argument given */
|
|
|
4924 int overwrite; /* overwrite existing output file */
|
|
323
|
4925 int added_word; /* invoked through "zg" */
|
|
310
|
4926 {
|
|
236
|
4927 char_u fname[MAXPATHL];
|
|
|
4928 char_u wfname[MAXPATHL];
|
|
310
|
4929 char_u **innames;
|
|
|
4930 int incount;
|
|
236
|
4931 afffile_T *(afile[8]);
|
|
|
4932 int i;
|
|
|
4933 int len;
|
|
|
4934 struct stat st;
|
|
255
|
4935 int error = FALSE;
|
|
300
|
4936 spellinfo_T spin;
|
|
|
4937
|
|
|
4938 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
4939 spin.si_verbose = !added_word;
|
|
310
|
4940 spin.si_ascii = ascii;
|
|
323
|
4941 spin.si_followup = TRUE;
|
|
|
4942 spin.si_rem_accents = TRUE;
|
|
|
4943 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
4944 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
4945 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
4946 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
4947
|
|
|
4948 /* default: fnames[0] is output file, following are input files */
|
|
|
4949 innames = &fnames[1];
|
|
|
4950 incount = fcount - 1;
|
|
|
4951
|
|
|
4952 if (fcount >= 1)
|
|
240
|
4953 {
|
|
310
|
4954 len = STRLEN(fnames[0]);
|
|
|
4955 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
4956 {
|
|
|
4957 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
4958 * "path/en.latin1.add.spl". */
|
|
|
4959 innames = &fnames[0];
|
|
|
4960 incount = 1;
|
|
|
4961 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
4962 }
|
|
366
|
4963 else if (fcount == 1)
|
|
|
4964 {
|
|
|
4965 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
4966 innames = &fnames[0];
|
|
|
4967 incount = 1;
|
|
|
4968 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
4969 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
4970 }
|
|
310
|
4971 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
4972 {
|
|
|
4973 /* Name ends in ".spl", use as the file name. */
|
|
323
|
4974 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
4975 }
|
|
|
4976 else
|
|
|
4977 /* Name should be language, make the file name from it. */
|
|
|
4978 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
4979 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
4980
|
|
|
4981 /* Check for .ascii.spl. */
|
|
|
4982 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
4983 spin.si_ascii = TRUE;
|
|
|
4984
|
|
|
4985 /* Check for .add.spl. */
|
|
|
4986 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
4987 spin.si_add = TRUE;
|
|
240
|
4988 }
|
|
|
4989
|
|
310
|
4990 if (incount <= 0)
|
|
236
|
4991 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
4992 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
4993 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
4994 else if (incount > 8)
|
|
236
|
4995 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
4996 else
|
|
|
4997 {
|
|
|
4998 /* Check for overwriting before doing things that may take a lot of
|
|
|
4999 * time. */
|
|
310
|
5000 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
5001 {
|
|
|
5002 EMSG(_(e_exists));
|
|
310
|
5003 return;
|
|
236
|
5004 }
|
|
310
|
5005 if (mch_isdir(wfname))
|
|
236
|
5006 {
|
|
310
|
5007 EMSG2(_(e_isadir2), wfname);
|
|
|
5008 return;
|
|
236
|
5009 }
|
|
|
5010
|
|
|
5011 /*
|
|
|
5012 * Init the aff and dic pointers.
|
|
|
5013 * Get the region names if there are more than 2 arguments.
|
|
|
5014 */
|
|
310
|
5015 for (i = 0; i < incount; ++i)
|
|
236
|
5016 {
|
|
310
|
5017 afile[i] = NULL;
|
|
300
|
5018
|
|
316
|
5019 if (incount > 1)
|
|
236
|
5020 {
|
|
310
|
5021 len = STRLEN(innames[i]);
|
|
|
5022 if (STRLEN(gettail(innames[i])) < 5
|
|
|
5023 || innames[i][len - 3] != '_')
|
|
236
|
5024 {
|
|
310
|
5025 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
5026 return;
|
|
236
|
5027 }
|
|
316
|
5028 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
5029 spin.si_region_name[i * 2 + 1] =
|
|
|
5030 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
5031 }
|
|
|
5032 }
|
|
316
|
5033 spin.si_region_count = incount;
|
|
236
|
5034
|
|
300
|
5035 spin.si_foldroot = wordtree_alloc(&spin.si_blocks);
|
|
|
5036 spin.si_keeproot = wordtree_alloc(&spin.si_blocks);
|
|
339
|
5037 spin.si_prefroot = wordtree_alloc(&spin.si_blocks);
|
|
|
5038 if (spin.si_foldroot == NULL
|
|
|
5039 || spin.si_keeproot == NULL
|
|
|
5040 || spin.si_prefroot == NULL)
|
|
300
|
5041 {
|
|
|
5042 error = TRUE;
|
|
310
|
5043 return;
|
|
300
|
5044 }
|
|
|
5045
|
|
351
|
5046 /* When not producing a .add.spl file clear the character table when
|
|
|
5047 * we encounter one in the .aff file. This means we dump the current
|
|
|
5048 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
5049 * better than guessing the contents, the table will match a
|
|
|
5050 * previously loaded spell file. */
|
|
|
5051 if (!spin.si_add)
|
|
|
5052 spin.si_clear_chartab = TRUE;
|
|
|
5053
|
|
236
|
5054 /*
|
|
|
5055 * Read all the .aff and .dic files.
|
|
|
5056 * Text is converted to 'encoding'.
|
|
300
|
5057 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
5058 */
|
|
310
|
5059 for (i = 0; i < incount && !error; ++i)
|
|
236
|
5060 {
|
|
300
|
5061 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
5062 spin.si_region = 1 << i;
|
|
|
5063
|
|
|
5064 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
5065 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
5066 {
|
|
|
5067 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
5068 * "SET" line. */
|
|
310
|
5069 afile[i] = spell_read_aff(fname, &spin);
|
|
|
5070 if (afile[i] == NULL)
|
|
300
|
5071 error = TRUE;
|
|
|
5072 else
|
|
|
5073 {
|
|
|
5074 /* Read the .dic file and store the words in the trees. */
|
|
|
5075 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
5076 innames[i]);
|
|
|
5077 if (spell_read_dic(fname, &spin, afile[i]) == FAIL)
|
|
300
|
5078 error = TRUE;
|
|
|
5079 }
|
|
|
5080 }
|
|
|
5081 else
|
|
|
5082 {
|
|
|
5083 /* No .aff file, try reading the file as a word list. Store
|
|
|
5084 * the words in the trees. */
|
|
310
|
5085 if (spell_read_wordfile(innames[i], &spin) == FAIL)
|
|
300
|
5086 error = TRUE;
|
|
|
5087 }
|
|
236
|
5088
|
|
310
|
5089 #ifdef FEAT_MBYTE
|
|
236
|
5090 /* Free any conversion stuff. */
|
|
300
|
5091 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
5092 #endif
|
|
236
|
5093 }
|
|
|
5094
|
|
300
|
5095 if (!error)
|
|
236
|
5096 {
|
|
|
5097 /*
|
|
300
|
5098 * Remove the dummy NUL from the start of the tree root.
|
|
236
|
5099 */
|
|
300
|
5100 spin.si_foldroot = spin.si_foldroot->wn_sibling;
|
|
|
5101 spin.si_keeproot = spin.si_keeproot->wn_sibling;
|
|
339
|
5102 spin.si_prefroot = spin.si_prefroot->wn_sibling;
|
|
236
|
5103
|
|
|
5104 /*
|
|
300
|
5105 * Combine tails in the tree.
|
|
236
|
5106 */
|
|
323
|
5107 if (!added_word || p_verbose > 2)
|
|
310
|
5108 {
|
|
323
|
5109 if (added_word)
|
|
310
|
5110 verbose_enter();
|
|
|
5111 MSG(_("Compressing word tree..."));
|
|
|
5112 out_flush();
|
|
323
|
5113 if (added_word)
|
|
310
|
5114 verbose_leave();
|
|
|
5115 }
|
|
|
5116 wordtree_compress(spin.si_foldroot, &spin);
|
|
|
5117 wordtree_compress(spin.si_keeproot, &spin);
|
|
339
|
5118 wordtree_compress(spin.si_prefroot, &spin);
|
|
236
|
5119 }
|
|
|
5120
|
|
300
|
5121 if (!error)
|
|
|
5122 {
|
|
|
5123 /*
|
|
|
5124 * Write the info in the spell file.
|
|
|
5125 */
|
|
323
|
5126 if (!added_word || p_verbose > 2)
|
|
310
|
5127 {
|
|
323
|
5128 if (added_word)
|
|
310
|
5129 verbose_enter();
|
|
366
|
5130 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
5131 out_flush();
|
|
323
|
5132 if (added_word)
|
|
310
|
5133 verbose_leave();
|
|
|
5134 }
|
|
|
5135
|
|
316
|
5136 write_vim_spell(wfname, &spin);
|
|
310
|
5137
|
|
323
|
5138 if (!added_word || p_verbose > 2)
|
|
310
|
5139 {
|
|
323
|
5140 if (added_word)
|
|
310
|
5141 verbose_enter();
|
|
|
5142 MSG(_("Done!"));
|
|
|
5143 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
5144 spin.si_memtot);
|
|
310
|
5145 out_flush();
|
|
323
|
5146 if (added_word)
|
|
310
|
5147 verbose_leave();
|
|
|
5148 }
|
|
|
5149
|
|
|
5150 /* If the file is loaded need to reload it. */
|
|
323
|
5151 spell_reload_one(wfname, added_word);
|
|
300
|
5152 }
|
|
|
5153
|
|
|
5154 /* Free the allocated memory. */
|
|
323
|
5155 ga_clear(&spin.si_rep);
|
|
|
5156 ga_clear(&spin.si_sal);
|
|
|
5157 ga_clear(&spin.si_map);
|
|
339
|
5158 ga_clear(&spin.si_prefcond);
|
|
366
|
5159 vim_free(spin.si_midword);
|
|
375
|
5160 vim_free(spin.si_sofofr);
|
|
|
5161 vim_free(spin.si_sofoto);
|
|
300
|
5162
|
|
|
5163 /* Free the .aff file structures. */
|
|
310
|
5164 for (i = 0; i < incount; ++i)
|
|
|
5165 if (afile[i] != NULL)
|
|
|
5166 spell_free_aff(afile[i]);
|
|
339
|
5167
|
|
|
5168 /* Free all the bits and pieces at once. */
|
|
|
5169 free_blocks(spin.si_blocks);
|
|
236
|
5170 }
|
|
310
|
5171 }
|
|
|
5172
|
|
|
5173
|
|
|
5174 /*
|
|
|
5175 * ":spellgood {word}"
|
|
|
5176 * ":spellwrong {word}"
|
|
|
5177 */
|
|
|
5178 void
|
|
|
5179 ex_spell(eap)
|
|
|
5180 exarg_T *eap;
|
|
|
5181 {
|
|
381
|
5182 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
|
5183 eap->forceit);
|
|
236
|
5184 }
|
|
|
5185
|
|
310
|
5186 /*
|
|
|
5187 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
5188 */
|
|
|
5189 void
|
|
381
|
5190 spell_add_word(word, len, bad, temp)
|
|
310
|
5191 char_u *word;
|
|
|
5192 int len;
|
|
|
5193 int bad;
|
|
381
|
5194 int temp; /* "zG" and "zW": use temp word list */
|
|
310
|
5195 {
|
|
|
5196 FILE *fd;
|
|
|
5197 buf_T *buf;
|
|
351
|
5198 int new_spf = FALSE;
|
|
|
5199 struct stat st;
|
|
381
|
5200 char_u *fname;
|
|
|
5201
|
|
|
5202 if (temp) /* use temp word list */
|
|
|
5203 {
|
|
|
5204 if (temp_wordlist == NULL)
|
|
|
5205 {
|
|
|
5206 temp_wordlist = vim_tempname('s');
|
|
|
5207 if (temp_wordlist == NULL)
|
|
|
5208 return;
|
|
|
5209 }
|
|
|
5210 fname = temp_wordlist;
|
|
|
5211 }
|
|
310
|
5212 else
|
|
|
5213 {
|
|
381
|
5214 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
5215 if (*curbuf->b_p_spf == NUL)
|
|
|
5216 {
|
|
|
5217 init_spellfile();
|
|
|
5218 new_spf = TRUE;
|
|
|
5219 }
|
|
|
5220
|
|
|
5221 if (*curbuf->b_p_spf == NUL)
|
|
|
5222 {
|
|
|
5223 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
5224 return;
|
|
|
5225 }
|
|
|
5226
|
|
310
|
5227 /* Check that the user isn't editing the .add file somewhere. */
|
|
|
5228 buf = buflist_findname_exp(curbuf->b_p_spf);
|
|
|
5229 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
5230 buf = NULL;
|
|
|
5231 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
5232 {
|
|
310
|
5233 EMSG(_(e_bufloaded));
|
|
381
|
5234 return;
|
|
|
5235 }
|
|
|
5236
|
|
|
5237 fname = curbuf->b_p_spf;
|
|
|
5238 }
|
|
|
5239
|
|
|
5240 fd = mch_fopen((char *)fname, "a");
|
|
|
5241 if (fd == NULL && new_spf)
|
|
|
5242 {
|
|
|
5243 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
5244 * We may need to create the "spell" directory first. We already
|
|
|
5245 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
5246 STRCPY(NameBuff, fname);
|
|
|
5247 *gettail_sep(NameBuff) = NUL;
|
|
|
5248 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
5249 {
|
|
|
5250 /* The directory doesn't exist. Try creating it and opening the
|
|
|
5251 * file again. */
|
|
|
5252 vim_mkdir(NameBuff, 0755);
|
|
|
5253 fd = mch_fopen((char *)fname, "a");
|
|
|
5254 }
|
|
|
5255 }
|
|
|
5256
|
|
|
5257 if (fd == NULL)
|
|
|
5258 EMSG2(_(e_notopen), fname);
|
|
|
5259 else
|
|
|
5260 {
|
|
|
5261 if (bad)
|
|
|
5262 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
5263 else
|
|
381
|
5264 fprintf(fd, "%.*s\n", len, word);
|
|
|
5265 fclose(fd);
|
|
|
5266
|
|
|
5267 /* Update the .add.spl file. */
|
|
|
5268 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
5269
|
|
|
5270 /* If the .add file is edited somewhere, reload it. */
|
|
|
5271 if (buf != NULL)
|
|
|
5272 buf_reload(buf);
|
|
|
5273
|
|
|
5274 redraw_all_later(NOT_VALID);
|
|
310
|
5275 }
|
|
|
5276 }
|
|
|
5277
|
|
|
5278 /*
|
|
|
5279 * Initialize 'spellfile' for the current buffer.
|
|
|
5280 */
|
|
|
5281 static void
|
|
|
5282 init_spellfile()
|
|
|
5283 {
|
|
|
5284 char_u buf[MAXPATHL];
|
|
|
5285 int l;
|
|
|
5286 slang_T *sl;
|
|
|
5287 char_u *rtp;
|
|
351
|
5288 char_u *lend;
|
|
310
|
5289
|
|
|
5290 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
5291 {
|
|
351
|
5292 /* Find the end of the language name. Exclude the region. */
|
|
|
5293 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
5294 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
5295 ;
|
|
|
5296
|
|
|
5297 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
5298 * are allowed to write. */
|
|
310
|
5299 rtp = p_rtp;
|
|
|
5300 while (*rtp != NUL)
|
|
|
5301 {
|
|
|
5302 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
5303 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
5304 if (filewritable(buf) == 2)
|
|
|
5305 {
|
|
316
|
5306 /* Use the first language name from 'spelllang' and the
|
|
|
5307 * encoding used in the first loaded .spl file. */
|
|
310
|
5308 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
5309 l = STRLEN(buf);
|
|
|
5310 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
5311 "/spell/%.*s.%s.add",
|
|
351
|
5312 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
5313 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
5314 ? (char_u *)"ascii" : spell_enc());
|
|
|
5315 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
5316 break;
|
|
|
5317 }
|
|
|
5318 }
|
|
|
5319 }
|
|
|
5320 }
|
|
236
|
5321
|
|
300
|
5322
|
|
307
|
5323 /*
|
|
|
5324 * Init the chartab used for spelling for ASCII.
|
|
|
5325 * EBCDIC is not supported!
|
|
|
5326 */
|
|
|
5327 static void
|
|
|
5328 clear_spell_chartab(sp)
|
|
|
5329 spelltab_T *sp;
|
|
|
5330 {
|
|
324
|
5331 int i;
|
|
307
|
5332
|
|
|
5333 /* Init everything to FALSE. */
|
|
|
5334 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
5335 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
5336 for (i = 0; i < 256; ++i)
|
|
324
|
5337 {
|
|
307
|
5338 sp->st_fold[i] = i;
|
|
324
|
5339 sp->st_upper[i] = i;
|
|
|
5340 }
|
|
307
|
5341
|
|
|
5342 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
5343 * that is done separately. */
|
|
|
5344 for (i = '0'; i <= '9'; ++i)
|
|
|
5345 sp->st_isw[i] = TRUE;
|
|
|
5346 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
5347 {
|
|
|
5348 sp->st_isw[i] = TRUE;
|
|
|
5349 sp->st_isu[i] = TRUE;
|
|
|
5350 sp->st_fold[i] = i + 0x20;
|
|
|
5351 }
|
|
|
5352 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
5353 {
|
|
307
|
5354 sp->st_isw[i] = TRUE;
|
|
324
|
5355 sp->st_upper[i] = i - 0x20;
|
|
|
5356 }
|
|
307
|
5357 }
|
|
|
5358
|
|
|
5359 /*
|
|
|
5360 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
5361 * Called once while starting up and when 'encoding' changes.
|
|
|
5362 * The default is to use isalpha(), but the spell file should define the word
|
|
|
5363 * characters to make it possible that 'encoding' differs from the current
|
|
|
5364 * locale.
|
|
|
5365 */
|
|
|
5366 void
|
|
|
5367 init_spell_chartab()
|
|
|
5368 {
|
|
|
5369 int i;
|
|
|
5370
|
|
|
5371 did_set_spelltab = FALSE;
|
|
|
5372 clear_spell_chartab(&spelltab);
|
|
|
5373 #ifdef FEAT_MBYTE
|
|
|
5374 if (enc_dbcs)
|
|
|
5375 {
|
|
|
5376 /* DBCS: assume double-wide characters are word characters. */
|
|
|
5377 for (i = 128; i <= 255; ++i)
|
|
|
5378 if (MB_BYTE2LEN(i) == 2)
|
|
|
5379 spelltab.st_isw[i] = TRUE;
|
|
|
5380 }
|
|
324
|
5381 else if (enc_utf8)
|
|
|
5382 {
|
|
|
5383 for (i = 128; i < 256; ++i)
|
|
|
5384 {
|
|
|
5385 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
5386 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
5387 spelltab.st_fold[i] = utf_fold(i);
|
|
|
5388 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
5389 }
|
|
|
5390 }
|
|
307
|
5391 else
|
|
|
5392 #endif
|
|
|
5393 {
|
|
324
|
5394 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
5395 for (i = 128; i < 256; ++i)
|
|
|
5396 {
|
|
|
5397 if (MB_ISUPPER(i))
|
|
|
5398 {
|
|
324
|
5399 spelltab.st_isw[i] = TRUE;
|
|
307
|
5400 spelltab.st_isu[i] = TRUE;
|
|
|
5401 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
5402 }
|
|
324
|
5403 else if (MB_ISLOWER(i))
|
|
|
5404 {
|
|
|
5405 spelltab.st_isw[i] = TRUE;
|
|
|
5406 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
5407 }
|
|
307
|
5408 }
|
|
|
5409 }
|
|
|
5410 }
|
|
|
5411
|
|
|
5412 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
5413 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
5414
|
|
|
5415 /*
|
|
|
5416 * Set the spell character tables from strings in the affix file.
|
|
|
5417 */
|
|
|
5418 static int
|
|
|
5419 set_spell_chartab(fol, low, upp)
|
|
|
5420 char_u *fol;
|
|
|
5421 char_u *low;
|
|
|
5422 char_u *upp;
|
|
|
5423 {
|
|
|
5424 /* We build the new tables here first, so that we can compare with the
|
|
|
5425 * previous one. */
|
|
|
5426 spelltab_T new_st;
|
|
|
5427 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
5428 int f, l, u;
|
|
|
5429
|
|
|
5430 clear_spell_chartab(&new_st);
|
|
|
5431
|
|
|
5432 while (*pf != NUL)
|
|
|
5433 {
|
|
|
5434 if (*pl == NUL || *pu == NUL)
|
|
|
5435 {
|
|
|
5436 EMSG(_(e_affform));
|
|
|
5437 return FAIL;
|
|
|
5438 }
|
|
|
5439 #ifdef FEAT_MBYTE
|
|
|
5440 f = mb_ptr2char_adv(&pf);
|
|
|
5441 l = mb_ptr2char_adv(&pl);
|
|
|
5442 u = mb_ptr2char_adv(&pu);
|
|
|
5443 #else
|
|
|
5444 f = *pf++;
|
|
|
5445 l = *pl++;
|
|
|
5446 u = *pu++;
|
|
|
5447 #endif
|
|
|
5448 /* Every character that appears is a word character. */
|
|
|
5449 if (f < 256)
|
|
|
5450 new_st.st_isw[f] = TRUE;
|
|
|
5451 if (l < 256)
|
|
|
5452 new_st.st_isw[l] = TRUE;
|
|
|
5453 if (u < 256)
|
|
|
5454 new_st.st_isw[u] = TRUE;
|
|
|
5455
|
|
|
5456 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
5457 * case-folding */
|
|
|
5458 if (l < 256 && l != f)
|
|
|
5459 {
|
|
|
5460 if (f >= 256)
|
|
|
5461 {
|
|
|
5462 EMSG(_(e_affrange));
|
|
|
5463 return FAIL;
|
|
|
5464 }
|
|
|
5465 new_st.st_fold[l] = f;
|
|
|
5466 }
|
|
|
5467
|
|
|
5468 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
5469 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
5470 * "FOL" . */
|
|
307
|
5471 if (u < 256 && u != f)
|
|
|
5472 {
|
|
|
5473 if (f >= 256)
|
|
|
5474 {
|
|
|
5475 EMSG(_(e_affrange));
|
|
|
5476 return FAIL;
|
|
|
5477 }
|
|
|
5478 new_st.st_fold[u] = f;
|
|
|
5479 new_st.st_isu[u] = TRUE;
|
|
324
|
5480 new_st.st_upper[f] = u;
|
|
307
|
5481 }
|
|
|
5482 }
|
|
|
5483
|
|
|
5484 if (*pl != NUL || *pu != NUL)
|
|
|
5485 {
|
|
|
5486 EMSG(_(e_affform));
|
|
|
5487 return FAIL;
|
|
|
5488 }
|
|
|
5489
|
|
|
5490 return set_spell_finish(&new_st);
|
|
|
5491 }
|
|
|
5492
|
|
|
5493 /*
|
|
|
5494 * Set the spell character tables from strings in the .spl file.
|
|
|
5495 */
|
|
|
5496 static int
|
|
381
|
5497 set_spell_charflags(flags, upp)
|
|
307
|
5498 char_u *flags;
|
|
|
5499 char_u *upp;
|
|
|
5500 {
|
|
|
5501 /* We build the new tables here first, so that we can compare with the
|
|
|
5502 * previous one. */
|
|
|
5503 spelltab_T new_st;
|
|
|
5504 int i;
|
|
|
5505 char_u *p = upp;
|
|
324
|
5506 int c;
|
|
307
|
5507
|
|
|
5508 clear_spell_chartab(&new_st);
|
|
|
5509
|
|
381
|
5510 for (i = 0; flags[i] != NUL; ++i)
|
|
307
|
5511 {
|
|
324
|
5512 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
5513 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
307
|
5514
|
|
|
5515 if (*p == NUL)
|
|
|
5516 return FAIL;
|
|
|
5517 #ifdef FEAT_MBYTE
|
|
324
|
5518 c = mb_ptr2char_adv(&p);
|
|
307
|
5519 #else
|
|
324
|
5520 c = *p++;
|
|
307
|
5521 #endif
|
|
324
|
5522 new_st.st_fold[i + 128] = c;
|
|
|
5523 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
5524 new_st.st_upper[c] = i + 128;
|
|
307
|
5525 }
|
|
|
5526
|
|
|
5527 return set_spell_finish(&new_st);
|
|
|
5528 }
|
|
|
5529
|
|
|
5530 static int
|
|
|
5531 set_spell_finish(new_st)
|
|
|
5532 spelltab_T *new_st;
|
|
|
5533 {
|
|
|
5534 int i;
|
|
|
5535
|
|
|
5536 if (did_set_spelltab)
|
|
|
5537 {
|
|
|
5538 /* check that it's the same table */
|
|
|
5539 for (i = 0; i < 256; ++i)
|
|
|
5540 {
|
|
|
5541 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
5542 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
5543 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
5544 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
5545 {
|
|
|
5546 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
5547 return FAIL;
|
|
|
5548 }
|
|
|
5549 }
|
|
|
5550 }
|
|
|
5551 else
|
|
|
5552 {
|
|
|
5553 /* copy the new spelltab into the one being used */
|
|
|
5554 spelltab = *new_st;
|
|
|
5555 did_set_spelltab = TRUE;
|
|
|
5556 }
|
|
|
5557
|
|
|
5558 return OK;
|
|
|
5559 }
|
|
|
5560
|
|
|
5561 /*
|
|
358
|
5562 * Return TRUE if "p" points to a word character.
|
|
366
|
5563 * As a special case we see "midword" characters as word character when it is
|
|
358
|
5564 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
5565 * Thus this only works properly when past the first character of the word.
|
|
358
|
5566 */
|
|
|
5567 static int
|
|
376
|
5568 spell_iswordp(p, buf)
|
|
358
|
5569 char_u *p;
|
|
376
|
5570 buf_T *buf; /* buffer used */
|
|
358
|
5571 {
|
|
366
|
5572 #ifdef FEAT_MBYTE
|
|
358
|
5573 char_u *s;
|
|
366
|
5574 int l;
|
|
|
5575 int c;
|
|
|
5576
|
|
|
5577 if (has_mbyte)
|
|
|
5578 {
|
|
|
5579 l = MB_BYTE2LEN(*p);
|
|
358
|
5580 s = p;
|
|
366
|
5581 if (l == 1)
|
|
|
5582 {
|
|
|
5583 /* be quick for ASCII */
|
|
376
|
5584 if (buf->b_spell_ismw[*p])
|
|
366
|
5585 {
|
|
|
5586 s = p + 1; /* skip a mid-word character */
|
|
|
5587 l = MB_BYTE2LEN(*s);
|
|
|
5588 }
|
|
|
5589 }
|
|
|
5590 else
|
|
|
5591 {
|
|
|
5592 c = mb_ptr2char(p);
|
|
376
|
5593 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
5594 : (buf->b_spell_ismw_mb != NULL
|
|
|
5595 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
5596 {
|
|
|
5597 s = p + l;
|
|
|
5598 l = MB_BYTE2LEN(*s);
|
|
|
5599 }
|
|
|
5600 }
|
|
|
5601
|
|
|
5602 if (l > 1)
|
|
|
5603 return mb_get_class(s) >= 2;
|
|
|
5604 return spelltab.st_isw[*s];
|
|
|
5605 }
|
|
358
|
5606 #endif
|
|
366
|
5607
|
|
376
|
5608 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
5609 }
|
|
|
5610
|
|
|
5611 /*
|
|
|
5612 * Return TRUE if "p" points to a word character.
|
|
|
5613 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
5614 */
|
|
|
5615 static int
|
|
|
5616 spell_iswordp_nmw(p)
|
|
|
5617 char_u *p;
|
|
|
5618 {
|
|
|
5619 #ifdef FEAT_MBYTE
|
|
|
5620 if (has_mbyte && MB_BYTE2LEN(*p) > 1)
|
|
|
5621 return mb_get_class(p) >= 2;
|
|
|
5622 #endif
|
|
|
5623
|
|
|
5624 return spelltab.st_isw[*p];
|
|
358
|
5625 }
|
|
|
5626
|
|
372
|
5627 #ifdef FEAT_MBYTE
|
|
|
5628 /*
|
|
|
5629 * Return TRUE if "p" points to a word character.
|
|
|
5630 * Wide version of spell_iswordp().
|
|
|
5631 */
|
|
|
5632 static int
|
|
376
|
5633 spell_iswordp_w(p, buf)
|
|
372
|
5634 int *p;
|
|
376
|
5635 buf_T *buf;
|
|
372
|
5636 {
|
|
|
5637 int *s;
|
|
|
5638
|
|
376
|
5639 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
5640 : (buf->b_spell_ismw_mb != NULL
|
|
|
5641 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
5642 s = p + 1;
|
|
|
5643 else
|
|
|
5644 s = p;
|
|
|
5645
|
|
|
5646 if (mb_char2len(*s) > 1)
|
|
|
5647 {
|
|
|
5648 if (enc_utf8)
|
|
|
5649 return utf_class(*s) >= 2;
|
|
|
5650 if (enc_dbcs)
|
|
|
5651 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
5652 return 0;
|
|
|
5653 }
|
|
|
5654 return spelltab.st_isw[*s];
|
|
|
5655 }
|
|
|
5656 #endif
|
|
|
5657
|
|
358
|
5658 /*
|
|
339
|
5659 * Write the table with prefix conditions to the .spl file.
|
|
|
5660 */
|
|
|
5661 static void
|
|
|
5662 write_spell_prefcond(fd, gap)
|
|
|
5663 FILE *fd;
|
|
|
5664 garray_T *gap;
|
|
|
5665 {
|
|
|
5666 int i;
|
|
|
5667 char_u *p;
|
|
|
5668 int len;
|
|
|
5669
|
|
|
5670 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
5671
|
|
|
5672 for (i = 0; i < gap->ga_len; ++i)
|
|
|
5673 {
|
|
|
5674 /* <prefcond> : <condlen> <condstr> */
|
|
|
5675 p = ((char_u **)gap->ga_data)[i];
|
|
|
5676 if (p == NULL)
|
|
|
5677 fputc(0, fd);
|
|
|
5678 else
|
|
|
5679 {
|
|
|
5680 len = STRLEN(p);
|
|
|
5681 fputc(len, fd);
|
|
|
5682 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
5683 }
|
|
|
5684 }
|
|
|
5685 }
|
|
|
5686
|
|
|
5687 /*
|
|
307
|
5688 * Write the current tables into the .spl file.
|
|
|
5689 * This makes sure the same characters are recognized as word characters when
|
|
|
5690 * generating an when using a spell file.
|
|
|
5691 */
|
|
|
5692 static void
|
|
|
5693 write_spell_chartab(fd)
|
|
|
5694 FILE *fd;
|
|
|
5695 {
|
|
|
5696 char_u charbuf[256 * 4];
|
|
|
5697 int len = 0;
|
|
|
5698 int flags;
|
|
|
5699 int i;
|
|
|
5700
|
|
|
5701 fputc(128, fd); /* <charflagslen> */
|
|
|
5702 for (i = 128; i < 256; ++i)
|
|
|
5703 {
|
|
|
5704 flags = 0;
|
|
|
5705 if (spelltab.st_isw[i])
|
|
324
|
5706 flags |= CF_WORD;
|
|
307
|
5707 if (spelltab.st_isu[i])
|
|
324
|
5708 flags |= CF_UPPER;
|
|
307
|
5709 fputc(flags, fd); /* <charflags> */
|
|
|
5710
|
|
310
|
5711 #ifdef FEAT_MBYTE
|
|
|
5712 if (has_mbyte)
|
|
|
5713 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
|
5714 else
|
|
|
5715 #endif
|
|
|
5716 charbuf[len++] = spelltab.st_fold[i];
|
|
307
|
5717 }
|
|
|
5718
|
|
|
5719 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
|
5720 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
|
5721 }
|
|
|
5722
|
|
|
5723 /*
|
|
324
|
5724 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
5725 * Uses the character definitions from the .spl file.
|
|
307
|
5726 * When using a multi-byte 'encoding' the length may change!
|
|
|
5727 * Returns FAIL when something wrong.
|
|
|
5728 */
|
|
|
5729 static int
|
|
324
|
5730 spell_casefold(str, len, buf, buflen)
|
|
|
5731 char_u *str;
|
|
307
|
5732 int len;
|
|
|
5733 char_u *buf;
|
|
|
5734 int buflen;
|
|
|
5735 {
|
|
|
5736 int i;
|
|
|
5737
|
|
|
5738 if (len >= buflen)
|
|
|
5739 {
|
|
|
5740 buf[0] = NUL;
|
|
|
5741 return FAIL; /* result will not fit */
|
|
|
5742 }
|
|
|
5743
|
|
|
5744 #ifdef FEAT_MBYTE
|
|
|
5745 if (has_mbyte)
|
|
|
5746 {
|
|
324
|
5747 int outi = 0;
|
|
|
5748 char_u *p;
|
|
307
|
5749 int c;
|
|
|
5750
|
|
|
5751 /* Fold one character at a time. */
|
|
324
|
5752 for (p = str; p < str + len; )
|
|
307
|
5753 {
|
|
|
5754 if (outi + MB_MAXBYTES > buflen)
|
|
|
5755 {
|
|
|
5756 buf[outi] = NUL;
|
|
|
5757 return FAIL;
|
|
|
5758 }
|
|
324
|
5759 c = mb_ptr2char_adv(&p);
|
|
|
5760 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
5761 }
|
|
|
5762 buf[outi] = NUL;
|
|
|
5763 }
|
|
|
5764 else
|
|
|
5765 #endif
|
|
|
5766 {
|
|
|
5767 /* Be quick for non-multibyte encodings. */
|
|
|
5768 for (i = 0; i < len; ++i)
|
|
324
|
5769 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
5770 buf[i] = NUL;
|
|
|
5771 }
|
|
|
5772
|
|
|
5773 return OK;
|
|
|
5774 }
|
|
|
5775
|
|
372
|
5776 #define SPS_BEST 1
|
|
|
5777 #define SPS_FAST 2
|
|
|
5778 #define SPS_DOUBLE 4
|
|
|
5779
|
|
|
5780 static int sps_flags = SPS_BEST;
|
|
|
5781
|
|
|
5782 /*
|
|
|
5783 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
|
5784 * Sets "sps_flags".
|
|
|
5785 */
|
|
|
5786 int
|
|
|
5787 spell_check_sps()
|
|
|
5788 {
|
|
|
5789 char_u *p;
|
|
|
5790 char_u buf[MAXPATHL];
|
|
|
5791 int f;
|
|
|
5792
|
|
|
5793 sps_flags = 0;
|
|
|
5794
|
|
|
5795 for (p = p_sps; *p != NUL; )
|
|
|
5796 {
|
|
|
5797 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
5798
|
|
|
5799 f = 0;
|
|
|
5800 if (STRCMP(buf, "best") == 0)
|
|
|
5801 f = SPS_BEST;
|
|
|
5802 else if (STRCMP(buf, "fast") == 0)
|
|
|
5803 f = SPS_FAST;
|
|
|
5804 else if (STRCMP(buf, "double") == 0)
|
|
|
5805 f = SPS_DOUBLE;
|
|
|
5806 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
5807 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
5808 f = -1;
|
|
|
5809
|
|
|
5810 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
5811 {
|
|
|
5812 sps_flags = SPS_BEST;
|
|
|
5813 return FAIL;
|
|
|
5814 }
|
|
|
5815 if (f != 0)
|
|
|
5816 sps_flags = f;
|
|
|
5817 }
|
|
|
5818
|
|
|
5819 if (sps_flags == 0)
|
|
|
5820 sps_flags = SPS_BEST;
|
|
|
5821
|
|
|
5822 return OK;
|
|
|
5823 }
|
|
|
5824
|
|
|
5825 /* Remember what "z?" replaced. */
|
|
|
5826 static char_u *repl_from = NULL;
|
|
|
5827 static char_u *repl_to = NULL;
|
|
|
5828
|
|
323
|
5829 /*
|
|
|
5830 * "z?": Find badly spelled word under or after the cursor.
|
|
|
5831 * Give suggestions for the properly spelled word.
|
|
|
5832 */
|
|
|
5833 void
|
|
|
5834 spell_suggest()
|
|
|
5835 {
|
|
|
5836 char_u *line;
|
|
|
5837 pos_T prev_cursor = curwin->w_cursor;
|
|
|
5838 char_u wcopy[MAXWLEN + 2];
|
|
|
5839 char_u *p;
|
|
|
5840 int i;
|
|
|
5841 int c;
|
|
|
5842 suginfo_T sug;
|
|
|
5843 suggest_T *stp;
|
|
372
|
5844 int mouse_used;
|
|
323
|
5845
|
|
344
|
5846 /* Find the start of the badly spelled word. */
|
|
346
|
5847 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
5848 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
5849 {
|
|
|
5850 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
5851 return;
|
|
|
5852
|
|
|
5853 /* No bad word or it starts after the cursor: use the word under the
|
|
|
5854 * cursor. */
|
|
|
5855 curwin->w_cursor = prev_cursor;
|
|
|
5856 line = ml_get_curline();
|
|
|
5857 p = line + curwin->w_cursor.col;
|
|
|
5858 /* Backup to before start of word. */
|
|
|
5859 while (p > line && SPELL_ISWORDP(p))
|
|
|
5860 mb_ptr_back(line, p);
|
|
|
5861 /* Forward to start of word. */
|
|
375
|
5862 while (*p != NUL && !SPELL_ISWORDP(p))
|
|
346
|
5863 mb_ptr_adv(p);
|
|
|
5864
|
|
|
5865 if (!SPELL_ISWORDP(p)) /* No word found. */
|
|
|
5866 {
|
|
|
5867 beep_flush();
|
|
|
5868 return;
|
|
|
5869 }
|
|
|
5870 curwin->w_cursor.col = p - line;
|
|
323
|
5871 }
|
|
|
5872
|
|
344
|
5873 /* Get the word and its length. */
|
|
323
|
5874 line = ml_get_curline();
|
|
344
|
5875
|
|
|
5876 /* Get the list of suggestions */
|
|
358
|
5877 spell_find_suggest(line + curwin->w_cursor.col, &sug, (int)Rows - 2, TRUE);
|
|
323
|
5878
|
|
|
5879 if (sug.su_ga.ga_len == 0)
|
|
|
5880 MSG(_("Sorry, no suggestions"));
|
|
|
5881 else
|
|
|
5882 {
|
|
372
|
5883 vim_free(repl_from);
|
|
|
5884 repl_from = NULL;
|
|
|
5885 vim_free(repl_to);
|
|
|
5886 repl_to = NULL;
|
|
|
5887
|
|
323
|
5888 /* List the suggestions. */
|
|
|
5889 msg_start();
|
|
372
|
5890 lines_left = Rows; /* avoid more prompt */
|
|
323
|
5891 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
5892 sug.su_badlen, sug.su_badptr);
|
|
|
5893 msg_puts(IObuff);
|
|
|
5894 msg_clr_eos();
|
|
|
5895 msg_putchar('\n');
|
|
346
|
5896
|
|
323
|
5897 msg_scroll = TRUE;
|
|
|
5898 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
5899 {
|
|
344
|
5900 stp = &SUG(sug.su_ga, i);
|
|
323
|
5901
|
|
|
5902 /* The suggested word may replace only part of the bad word, add
|
|
|
5903 * the not replaced part. */
|
|
|
5904 STRCPY(wcopy, stp->st_word);
|
|
|
5905 if (sug.su_badlen > stp->st_orglen)
|
|
|
5906 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
5907 sug.su_badptr + stp->st_orglen,
|
|
|
5908 sug.su_badlen - stp->st_orglen);
|
|
346
|
5909 vim_snprintf((char *)IObuff, IOSIZE, _("%2d \"%s\""), i + 1, wcopy);
|
|
|
5910 msg_puts(IObuff);
|
|
|
5911
|
|
|
5912 /* The word may replace more than "su_badlen". */
|
|
|
5913 if (sug.su_badlen < stp->st_orglen)
|
|
|
5914 {
|
|
|
5915 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
5916 stp->st_orglen, sug.su_badptr);
|
|
|
5917 msg_puts(IObuff);
|
|
|
5918 }
|
|
|
5919
|
|
324
|
5920 if (p_verbose > 0)
|
|
344
|
5921 {
|
|
346
|
5922 /* Add the score. */
|
|
351
|
5923 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
346
|
5924 vim_snprintf((char *)IObuff, IOSIZE, _(" (%s%d - %d)"),
|
|
344
|
5925 stp->st_salscore ? "s " : "",
|
|
|
5926 stp->st_score, stp->st_altscore);
|
|
|
5927 else
|
|
346
|
5928 vim_snprintf((char *)IObuff, IOSIZE, _(" (%d)"),
|
|
|
5929 stp->st_score);
|
|
|
5930 msg_advance(30);
|
|
|
5931 msg_puts(IObuff);
|
|
344
|
5932 }
|
|
323
|
5933 msg_putchar('\n');
|
|
|
5934 }
|
|
|
5935
|
|
|
5936 /* Ask for choice. */
|
|
372
|
5937 i = prompt_for_number(&mouse_used);
|
|
|
5938 if (mouse_used)
|
|
|
5939 i -= lines_left;
|
|
|
5940
|
|
344
|
5941 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
5942 {
|
|
372
|
5943 /* Save the from and to text for :spellrepall. */
|
|
|
5944 stp = &SUG(sug.su_ga, i - 1);
|
|
|
5945 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
5946 repl_to = vim_strsave(stp->st_word);
|
|
|
5947
|
|
323
|
5948 /* Replace the word. */
|
|
|
5949 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
5950 if (p != NULL)
|
|
|
5951 {
|
|
|
5952 c = sug.su_badptr - line;
|
|
|
5953 mch_memmove(p, line, c);
|
|
|
5954 STRCPY(p + c, stp->st_word);
|
|
|
5955 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
5956 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
5957 curwin->w_cursor.col = c;
|
|
|
5958 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
5959
|
|
|
5960 /* For redo we use a change-word command. */
|
|
|
5961 ResetRedobuff();
|
|
|
5962 AppendToRedobuff((char_u *)"ciw");
|
|
|
5963 AppendToRedobuff(stp->st_word);
|
|
|
5964 AppendCharToRedobuff(ESC);
|
|
323
|
5965 }
|
|
|
5966 }
|
|
|
5967 else
|
|
|
5968 curwin->w_cursor = prev_cursor;
|
|
|
5969 }
|
|
|
5970
|
|
344
|
5971 spell_find_cleanup(&sug);
|
|
|
5972 }
|
|
|
5973
|
|
|
5974 /*
|
|
372
|
5975 * ":spellrepall"
|
|
|
5976 */
|
|
|
5977 /*ARGSUSED*/
|
|
|
5978 void
|
|
|
5979 ex_spellrepall(eap)
|
|
|
5980 exarg_T *eap;
|
|
|
5981 {
|
|
|
5982 pos_T pos = curwin->w_cursor;
|
|
|
5983 char_u *frompat;
|
|
|
5984 int addlen;
|
|
|
5985 char_u *line;
|
|
|
5986 char_u *p;
|
|
|
5987 int didone = FALSE;
|
|
|
5988 int save_ws = p_ws;
|
|
|
5989
|
|
|
5990 if (repl_from == NULL || repl_to == NULL)
|
|
|
5991 {
|
|
|
5992 EMSG(_("E752: No previous spell replacement"));
|
|
|
5993 return;
|
|
|
5994 }
|
|
|
5995 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
5996
|
|
|
5997 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
5998 if (frompat == NULL)
|
|
|
5999 return;
|
|
|
6000 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
6001 p_ws = FALSE;
|
|
|
6002
|
|
|
6003 curwin->w_cursor.lnum = 0;
|
|
|
6004 while (!got_int)
|
|
|
6005 {
|
|
|
6006 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
6007 || u_save_cursor() == FAIL)
|
|
|
6008 break;
|
|
|
6009
|
|
|
6010 /* Only replace when the right word isn't there yet. This happens
|
|
|
6011 * when changing "etc" to "etc.". */
|
|
|
6012 line = ml_get_curline();
|
|
|
6013 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
6014 repl_to, STRLEN(repl_to)) != 0)
|
|
|
6015 {
|
|
|
6016 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
6017 if (p == NULL)
|
|
|
6018 break;
|
|
|
6019 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
6020 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
6021 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
6022 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6023 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
6024 didone = TRUE;
|
|
|
6025 }
|
|
|
6026 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
6027 }
|
|
|
6028
|
|
|
6029 p_ws = save_ws;
|
|
|
6030 curwin->w_cursor = pos;
|
|
|
6031 vim_free(frompat);
|
|
|
6032
|
|
|
6033 if (!didone)
|
|
|
6034 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
|
6035 }
|
|
|
6036
|
|
|
6037 /*
|
|
344
|
6038 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
6039 * a list of allocated strings.
|
|
|
6040 */
|
|
|
6041 void
|
|
|
6042 spell_suggest_list(gap, word, maxcount)
|
|
|
6043 garray_T *gap;
|
|
|
6044 char_u *word;
|
|
|
6045 int maxcount; /* maximum nr of suggestions */
|
|
|
6046 {
|
|
|
6047 suginfo_T sug;
|
|
|
6048 int i;
|
|
|
6049 suggest_T *stp;
|
|
|
6050 char_u *wcopy;
|
|
|
6051
|
|
358
|
6052 spell_find_suggest(word, &sug, maxcount, FALSE);
|
|
344
|
6053
|
|
|
6054 /* Make room in "gap". */
|
|
|
6055 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
6056 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
6057 return;
|
|
|
6058
|
|
323
|
6059 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
6060 {
|
|
|
6061 stp = &SUG(sug.su_ga, i);
|
|
|
6062
|
|
|
6063 /* The suggested word may replace only part of "word", add the not
|
|
|
6064 * replaced part. */
|
|
|
6065 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
6066 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
6067 if (wcopy == NULL)
|
|
|
6068 break;
|
|
|
6069 STRCPY(wcopy, stp->st_word);
|
|
|
6070 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
6071 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
6072 }
|
|
|
6073
|
|
|
6074 spell_find_cleanup(&sug);
|
|
|
6075 }
|
|
|
6076
|
|
|
6077 /*
|
|
|
6078 * Find spell suggestions for the word at the start of "badptr".
|
|
|
6079 * Return the suggestions in "su->su_ga".
|
|
|
6080 * The maximum number of suggestions is "maxcount".
|
|
|
6081 * Note: does use info for the current window.
|
|
|
6082 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
6083 */
|
|
|
6084 static void
|
|
358
|
6085 spell_find_suggest(badptr, su, maxcount, banbadword)
|
|
344
|
6086 char_u *badptr;
|
|
|
6087 suginfo_T *su;
|
|
|
6088 int maxcount;
|
|
358
|
6089 int banbadword; /* don't include badword in suggestions */
|
|
344
|
6090 {
|
|
|
6091 int attr;
|
|
372
|
6092 char_u buf[MAXPATHL];
|
|
|
6093 char_u *p;
|
|
|
6094 int do_combine = FALSE;
|
|
|
6095 char_u *sps_copy;
|
|
|
6096 #ifdef FEAT_EVAL
|
|
|
6097 static int expr_busy = FALSE;
|
|
|
6098 #endif
|
|
344
|
6099
|
|
|
6100 /*
|
|
|
6101 * Set the info in "*su".
|
|
|
6102 */
|
|
|
6103 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
6104 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
6105 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
6106 if (*badptr == NUL)
|
|
|
6107 return;
|
|
344
|
6108 hash_init(&su->su_banned);
|
|
|
6109
|
|
|
6110 su->su_badptr = badptr;
|
|
|
6111 su->su_badlen = spell_check(curwin, su->su_badptr, &attr);
|
|
|
6112 su->su_maxcount = maxcount;
|
|
372
|
6113 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
6114
|
|
|
6115 if (su->su_badlen >= MAXWLEN)
|
|
|
6116 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
6117 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
6118 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
6119 su->su_fbadword, MAXWLEN);
|
|
346
|
6120 /* get caps flags for bad word */
|
|
|
6121 su->su_badflags = captype(su->su_badptr, su->su_badptr + su->su_badlen);
|
|
344
|
6122
|
|
|
6123 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
6124 if (banbadword)
|
|
|
6125 add_banned(su, su->su_badword);
|
|
344
|
6126
|
|
372
|
6127 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
6128 sps_copy = vim_strsave(p_sps);
|
|
|
6129 if (sps_copy == NULL)
|
|
|
6130 return;
|
|
|
6131
|
|
|
6132 /* Loop over the items in 'spellsuggest'. */
|
|
|
6133 for (p = sps_copy; *p != NUL; )
|
|
|
6134 {
|
|
|
6135 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6136
|
|
|
6137 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
6138 {
|
|
|
6139 #ifdef FEAT_EVAL
|
|
375
|
6140 /* Evaluate an expression. Skip this when called recursively,
|
|
|
6141 * when using spellsuggest() in the expression. */
|
|
372
|
6142 if (!expr_busy)
|
|
|
6143 {
|
|
|
6144 expr_busy = TRUE;
|
|
|
6145 spell_suggest_expr(su, buf + 5);
|
|
|
6146 expr_busy = FALSE;
|
|
|
6147 }
|
|
|
6148 #endif
|
|
|
6149 }
|
|
|
6150 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
6151 /* Use list of suggestions in a file. */
|
|
|
6152 spell_suggest_file(su, buf + 5);
|
|
|
6153 else
|
|
|
6154 {
|
|
|
6155 /* Use internal method. */
|
|
|
6156 spell_suggest_intern(su);
|
|
|
6157 if (sps_flags & SPS_DOUBLE)
|
|
|
6158 do_combine = TRUE;
|
|
|
6159 }
|
|
|
6160 }
|
|
|
6161
|
|
|
6162 vim_free(sps_copy);
|
|
|
6163
|
|
|
6164 if (do_combine)
|
|
|
6165 /* Combine the two list of suggestions. This must be done last,
|
|
|
6166 * because sorting changes the order again. */
|
|
|
6167 score_combine(su);
|
|
|
6168 }
|
|
|
6169
|
|
|
6170 #ifdef FEAT_EVAL
|
|
|
6171 /*
|
|
|
6172 * Find suggestions by evaluating expression "expr".
|
|
|
6173 */
|
|
|
6174 static void
|
|
|
6175 spell_suggest_expr(su, expr)
|
|
|
6176 suginfo_T *su;
|
|
|
6177 char_u *expr;
|
|
|
6178 {
|
|
|
6179 list_T *list;
|
|
|
6180 listitem_T *li;
|
|
|
6181 int score;
|
|
|
6182 char_u *p;
|
|
|
6183
|
|
|
6184 /* The work is split up in a few parts to avoid having to export
|
|
|
6185 * suginfo_T.
|
|
|
6186 * First evaluate the expression and get the resulting list. */
|
|
|
6187 list = eval_spell_expr(su->su_badword, expr);
|
|
|
6188 if (list != NULL)
|
|
|
6189 {
|
|
|
6190 /* Loop over the items in the list. */
|
|
|
6191 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
6192 if (li->li_tv.v_type == VAR_LIST)
|
|
|
6193 {
|
|
|
6194 /* Get the word and the score from the items. */
|
|
|
6195 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
6196 if (score >= 0)
|
|
|
6197 add_suggestion(su, &su->su_ga, p,
|
|
|
6198 su->su_badlen, score, 0, TRUE);
|
|
|
6199 }
|
|
|
6200 list_unref(list);
|
|
|
6201 }
|
|
|
6202
|
|
|
6203 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6204 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6205 }
|
|
|
6206 #endif
|
|
|
6207
|
|
|
6208 /*
|
|
|
6209 * Find suggestions a file "fname".
|
|
|
6210 */
|
|
|
6211 static void
|
|
|
6212 spell_suggest_file(su, fname)
|
|
|
6213 suginfo_T *su;
|
|
|
6214 char_u *fname;
|
|
|
6215 {
|
|
|
6216 FILE *fd;
|
|
|
6217 char_u line[MAXWLEN * 2];
|
|
|
6218 char_u *p;
|
|
|
6219 int len;
|
|
|
6220 char_u cword[MAXWLEN];
|
|
|
6221
|
|
|
6222 /* Open the file. */
|
|
|
6223 fd = mch_fopen((char *)fname, "r");
|
|
|
6224 if (fd == NULL)
|
|
|
6225 {
|
|
|
6226 EMSG2(_(e_notopen), fname);
|
|
|
6227 return;
|
|
|
6228 }
|
|
|
6229
|
|
|
6230 /* Read it line by line. */
|
|
|
6231 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
6232 {
|
|
|
6233 line_breakcheck();
|
|
|
6234
|
|
|
6235 p = vim_strchr(line, '/');
|
|
|
6236 if (p == NULL)
|
|
|
6237 continue; /* No Tab found, just skip the line. */
|
|
|
6238 *p++ = NUL;
|
|
|
6239 if (STRICMP(su->su_badword, line) == 0)
|
|
|
6240 {
|
|
|
6241 /* Match! Isolate the good word, until CR or NL. */
|
|
|
6242 for (len = 0; p[len] >= ' '; ++len)
|
|
|
6243 ;
|
|
|
6244 p[len] = NUL;
|
|
|
6245
|
|
|
6246 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
6247 * of the bad word. */
|
|
|
6248 if (captype(p, NULL) == 0)
|
|
|
6249 {
|
|
|
6250 make_case_word(p, cword, su->su_badflags);
|
|
|
6251 p = cword;
|
|
|
6252 }
|
|
|
6253
|
|
|
6254 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
6255 SCORE_FILE, 0, TRUE);
|
|
|
6256 }
|
|
|
6257 }
|
|
|
6258
|
|
|
6259 fclose(fd);
|
|
|
6260
|
|
|
6261 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6262 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6263 }
|
|
|
6264
|
|
|
6265 /*
|
|
|
6266 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
6267 */
|
|
|
6268 static void
|
|
|
6269 spell_suggest_intern(su)
|
|
|
6270 suginfo_T *su;
|
|
|
6271 {
|
|
344
|
6272 /*
|
|
346
|
6273 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
6274 *
|
|
|
6275 * Set a maximum score to limit the combination of operations that is
|
|
|
6276 * tried.
|
|
|
6277 */
|
|
346
|
6278 suggest_try_special(su);
|
|
|
6279
|
|
|
6280 /*
|
|
|
6281 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
6282 * from the .aff file and inserting a space (split the word).
|
|
|
6283 */
|
|
|
6284 suggest_try_change(su);
|
|
344
|
6285
|
|
|
6286 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
6287 if (sps_flags & SPS_DOUBLE)
|
|
|
6288 score_comp_sal(su);
|
|
|
6289
|
|
|
6290 /*
|
|
346
|
6291 * 3. Try finding sound-a-like words.
|
|
344
|
6292 *
|
|
|
6293 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
6294 * very slow and often doesn't find new suggestions.
|
|
|
6295 */
|
|
|
6296 if ((sps_flags & SPS_DOUBLE)
|
|
|
6297 || (!(sps_flags & SPS_FAST)
|
|
|
6298 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
6299 {
|
|
|
6300 /* Allow a higher score now. */
|
|
|
6301 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
6302 suggest_try_soundalike(su);
|
|
344
|
6303 }
|
|
|
6304
|
|
|
6305 /* When CTRL-C was hit while searching do show the results. */
|
|
|
6306 ui_breakcheck();
|
|
|
6307 if (got_int)
|
|
|
6308 {
|
|
|
6309 (void)vgetc();
|
|
|
6310 got_int = FALSE;
|
|
|
6311 }
|
|
|
6312
|
|
372
|
6313 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
6314 {
|
|
|
6315 if (sps_flags & SPS_BEST)
|
|
|
6316 /* Adjust the word score for how it sounds like. */
|
|
|
6317 rescore_suggestions(su);
|
|
|
6318
|
|
|
6319 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
6320 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
6321 }
|
|
|
6322 }
|
|
|
6323
|
|
|
6324 /*
|
|
|
6325 * Free the info put in "*su" by spell_find_suggest().
|
|
|
6326 */
|
|
|
6327 static void
|
|
|
6328 spell_find_cleanup(su)
|
|
|
6329 suginfo_T *su;
|
|
|
6330 {
|
|
|
6331 int i;
|
|
|
6332
|
|
|
6333 /* Free the suggestions. */
|
|
|
6334 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
6335 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
6336 ga_clear(&su->su_ga);
|
|
|
6337 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
6338 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
6339 ga_clear(&su->su_sga);
|
|
323
|
6340
|
|
|
6341 /* Free the banned words. */
|
|
344
|
6342 free_banned(su);
|
|
323
|
6343 }
|
|
|
6344
|
|
|
6345 /*
|
|
324
|
6346 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
6347 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
6348 * The result is NUL terminated.
|
|
323
|
6349 */
|
|
|
6350 static void
|
|
324
|
6351 onecap_copy(word, wcopy, upper)
|
|
323
|
6352 char_u *word;
|
|
|
6353 char_u *wcopy;
|
|
|
6354 int upper; /* TRUE: first letter made upper case */
|
|
|
6355 {
|
|
|
6356 char_u *p;
|
|
|
6357 int c;
|
|
|
6358 int l;
|
|
|
6359
|
|
|
6360 p = word;
|
|
|
6361 #ifdef FEAT_MBYTE
|
|
|
6362 if (has_mbyte)
|
|
|
6363 c = mb_ptr2char_adv(&p);
|
|
|
6364 else
|
|
|
6365 #endif
|
|
|
6366 c = *p++;
|
|
|
6367 if (upper)
|
|
324
|
6368 c = SPELL_TOUPPER(c);
|
|
323
|
6369 else
|
|
324
|
6370 c = SPELL_TOFOLD(c);
|
|
323
|
6371 #ifdef FEAT_MBYTE
|
|
|
6372 if (has_mbyte)
|
|
|
6373 l = mb_char2bytes(c, wcopy);
|
|
|
6374 else
|
|
|
6375 #endif
|
|
|
6376 {
|
|
|
6377 l = 1;
|
|
|
6378 wcopy[0] = c;
|
|
|
6379 }
|
|
376
|
6380 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
6381 }
|
|
|
6382
|
|
|
6383 /*
|
|
324
|
6384 * Make a copy of "word" with all the letters upper cased into
|
|
|
6385 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
6386 */
|
|
|
6387 static void
|
|
|
6388 allcap_copy(word, wcopy)
|
|
|
6389 char_u *word;
|
|
|
6390 char_u *wcopy;
|
|
|
6391 {
|
|
|
6392 char_u *s;
|
|
|
6393 char_u *d;
|
|
|
6394 int c;
|
|
|
6395
|
|
|
6396 d = wcopy;
|
|
|
6397 for (s = word; *s != NUL; )
|
|
|
6398 {
|
|
|
6399 #ifdef FEAT_MBYTE
|
|
|
6400 if (has_mbyte)
|
|
|
6401 c = mb_ptr2char_adv(&s);
|
|
|
6402 else
|
|
|
6403 #endif
|
|
|
6404 c = *s++;
|
|
324
|
6405 c = SPELL_TOUPPER(c);
|
|
323
|
6406
|
|
|
6407 #ifdef FEAT_MBYTE
|
|
|
6408 if (has_mbyte)
|
|
|
6409 {
|
|
|
6410 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
6411 break;
|
|
|
6412 d += mb_char2bytes(c, d);
|
|
|
6413 }
|
|
|
6414 else
|
|
|
6415 #endif
|
|
|
6416 {
|
|
|
6417 if (d - wcopy >= MAXWLEN - 1)
|
|
|
6418 break;
|
|
|
6419 *d++ = c;
|
|
|
6420 }
|
|
|
6421 }
|
|
|
6422 *d = NUL;
|
|
|
6423 }
|
|
|
6424
|
|
|
6425 /*
|
|
346
|
6426 * Try finding suggestions by recognizing specific situations.
|
|
|
6427 */
|
|
|
6428 static void
|
|
|
6429 suggest_try_special(su)
|
|
|
6430 suginfo_T *su;
|
|
|
6431 {
|
|
|
6432 char_u *p;
|
|
376
|
6433 size_t len;
|
|
346
|
6434 int c;
|
|
|
6435 char_u word[MAXWLEN];
|
|
|
6436
|
|
|
6437 /*
|
|
|
6438 * Recognize a word that is repeated: "the the".
|
|
|
6439 */
|
|
|
6440 p = skiptowhite(su->su_fbadword);
|
|
|
6441 len = p - su->su_fbadword;
|
|
|
6442 p = skipwhite(p);
|
|
|
6443 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
6444 {
|
|
|
6445 /* Include badflags: if the badword is onecap or allcap
|
|
|
6446 * use that for the goodword too: "The the" -> "The". */
|
|
|
6447 c = su->su_fbadword[len];
|
|
|
6448 su->su_fbadword[len] = NUL;
|
|
|
6449 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
6450 su->su_fbadword[len] = c;
|
|
351
|
6451 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
6452 }
|
|
|
6453 }
|
|
|
6454
|
|
|
6455 /*
|
|
323
|
6456 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
6457 *
|
|
|
6458 * This uses a state machine. At each node in the tree we try various
|
|
|
6459 * operations. When trying if an operation work "depth" is increased and the
|
|
|
6460 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
6461 * character, replace one and delete another. The number of changes is
|
|
|
6462 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
6463 */
|
|
|
6464 static void
|
|
346
|
6465 suggest_try_change(su)
|
|
323
|
6466 suginfo_T *su;
|
|
|
6467 {
|
|
|
6468 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
6469 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
6470 trystate_T stack[MAXWLEN];
|
|
|
6471 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
|
6472 * to for word split) */
|
|
|
6473 char_u prewordlen = 0; /* length of word in "preword" */
|
|
|
6474 int splitoff = 0; /* index in tword after last split */
|
|
|
6475 trystate_T *sp;
|
|
|
6476 int newscore;
|
|
|
6477 langp_T *lp;
|
|
375
|
6478 char_u *byts, *fbyts, *pbyts;
|
|
|
6479 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
6480 int depth;
|
|
330
|
6481 int c, c2, c3;
|
|
|
6482 int n = 0;
|
|
323
|
6483 int flags;
|
|
|
6484 garray_T *gap;
|
|
324
|
6485 idx_T arridx;
|
|
323
|
6486 int len;
|
|
|
6487 char_u *p;
|
|
|
6488 fromto_T *ftp;
|
|
330
|
6489 int fl = 0, tl;
|
|
346
|
6490 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
323
|
6491
|
|
|
6492 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
6493 * to find matches (esp. REP items). Append some more text, changing
|
|
|
6494 * chars after the bad word may help. */
|
|
323
|
6495 STRCPY(fword, su->su_fbadword);
|
|
346
|
6496 n = STRLEN(fword);
|
|
|
6497 p = su->su_badptr + su->su_badlen;
|
|
|
6498 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
6499
|
|
|
6500 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
6501 lp->lp_slang != NULL; ++lp)
|
|
|
6502 {
|
|
|
6503 /*
|
|
|
6504 * Go through the whole case-fold tree, try changes at each node.
|
|
|
6505 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
6506 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
6507 * word).
|
|
|
6508 */
|
|
|
6509 depth = 0;
|
|
375
|
6510 sp = &stack[0];
|
|
|
6511 sp->ts_state = STATE_START;
|
|
|
6512 sp->ts_score = 0;
|
|
|
6513 sp->ts_curi = 1;
|
|
|
6514 sp->ts_fidx = 0;
|
|
|
6515 sp->ts_fidxtry = 0;
|
|
|
6516 sp->ts_twordlen = 0;
|
|
|
6517 sp->ts_arridx = 0;
|
|
330
|
6518 #ifdef FEAT_MBYTE
|
|
375
|
6519 sp->ts_tcharlen = 0;
|
|
330
|
6520 #endif
|
|
|
6521
|
|
|
6522 /*
|
|
375
|
6523 * When there are postponed prefixes we need to use these first. At
|
|
|
6524 * the end of the prefix we continue in the case-fold tree.
|
|
|
6525 */
|
|
|
6526 fbyts = lp->lp_slang->sl_fbyts;
|
|
|
6527 fidxs = lp->lp_slang->sl_fidxs;
|
|
|
6528 pbyts = lp->lp_slang->sl_pbyts;
|
|
|
6529 pidxs = lp->lp_slang->sl_pidxs;
|
|
|
6530 if (pbyts != NULL)
|
|
|
6531 {
|
|
|
6532 byts = pbyts;
|
|
|
6533 idxs = pidxs;
|
|
|
6534 sp->ts_prefixdepth = PREFIXTREE;
|
|
|
6535 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
6536 }
|
|
|
6537 else
|
|
|
6538 {
|
|
|
6539 byts = fbyts;
|
|
|
6540 idxs = fidxs;
|
|
|
6541 sp->ts_prefixdepth = NOPREFIX;
|
|
|
6542 }
|
|
|
6543
|
|
|
6544 /*
|
|
330
|
6545 * Loop to find all suggestions. At each round we either:
|
|
|
6546 * - For the current state try one operation, advance "ts_curi",
|
|
|
6547 * increase "depth".
|
|
|
6548 * - When a state is done go to the next, set "ts_state".
|
|
|
6549 * - When all states are tried decrease "depth".
|
|
|
6550 */
|
|
323
|
6551 while (depth >= 0 && !got_int)
|
|
|
6552 {
|
|
|
6553 sp = &stack[depth];
|
|
|
6554 switch (sp->ts_state)
|
|
|
6555 {
|
|
|
6556 case STATE_START:
|
|
375
|
6557 case STATE_NOPREFIX:
|
|
323
|
6558 /*
|
|
|
6559 * Start of node: Deal with NUL bytes, which means
|
|
|
6560 * tword[] may end here.
|
|
|
6561 */
|
|
|
6562 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
6563 len = byts[arridx]; /* bytes in this node */
|
|
|
6564 arridx += sp->ts_curi; /* index of current byte */
|
|
|
6565
|
|
375
|
6566 if (sp->ts_prefixdepth == PREFIXTREE)
|
|
|
6567 {
|
|
|
6568 /* Skip over the NUL bytes, we use them later. */
|
|
|
6569 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
6570 ;
|
|
|
6571 sp->ts_curi += n;
|
|
|
6572
|
|
|
6573 /* At end of a prefix or at start of prefixtree: check for
|
|
|
6574 * following word. */
|
|
|
6575 if (byts[arridx] == 0 || sp->ts_state == STATE_NOPREFIX)
|
|
|
6576 {
|
|
|
6577 sp->ts_state = STATE_START;
|
|
|
6578 ++depth;
|
|
|
6579 stack[depth] = stack[depth - 1];
|
|
|
6580 sp = &stack[depth];
|
|
|
6581 sp->ts_prefixdepth = depth - 1;
|
|
|
6582 byts = fbyts;
|
|
|
6583 idxs = fidxs;
|
|
|
6584 sp->ts_state = STATE_START;
|
|
|
6585 sp->ts_curi = 1; /* start just after length byte */
|
|
|
6586 sp->ts_arridx = 0;
|
|
|
6587
|
|
|
6588 /* Move the prefix to preword[] so that
|
|
|
6589 * find_keepcap_word() works. */
|
|
|
6590 prewordlen = splitoff = sp->ts_twordlen;
|
|
|
6591 mch_memmove(preword, tword, splitoff);
|
|
|
6592 break;
|
|
|
6593 }
|
|
|
6594
|
|
|
6595 /* Always past NUL bytes now. */
|
|
|
6596 sp->ts_state = STATE_ENDNUL;
|
|
|
6597 break;
|
|
|
6598 }
|
|
|
6599
|
|
346
|
6600 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
6601 {
|
|
|
6602 /* Past bytes in node and/or past NUL bytes. */
|
|
|
6603 sp->ts_state = STATE_ENDNUL;
|
|
|
6604 break;
|
|
|
6605 }
|
|
|
6606
|
|
|
6607 /*
|
|
|
6608 * End of word in tree.
|
|
|
6609 */
|
|
|
6610 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
6611
|
|
324
|
6612 flags = (int)idxs[arridx];
|
|
323
|
6613
|
|
375
|
6614 if (sp->ts_prefixdepth < MAXWLEN)
|
|
|
6615 {
|
|
|
6616 /* There was a prefix before the word. Check that the
|
|
|
6617 * prefix can be used with this word. */
|
|
|
6618 /* Count the length of the NULs in the prefix. If there
|
|
|
6619 * are none this must be the first try without a prefix.
|
|
|
6620 */
|
|
|
6621 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
6622 len = pbyts[n++];
|
|
|
6623 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
6624 ;
|
|
|
6625 if (c > 0)
|
|
|
6626 {
|
|
|
6627 /* The prefix ID is stored two bytes above the flags. */
|
|
|
6628 c = valid_word_prefix(c, n, (unsigned)flags >> 16,
|
|
|
6629 tword + splitoff, lp->lp_slang);
|
|
|
6630 if (c == 0)
|
|
|
6631 break;
|
|
|
6632
|
|
|
6633 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
6634 if (c & WF_RAREPFX)
|
|
|
6635 flags |= WF_RARE;
|
|
|
6636 }
|
|
|
6637 }
|
|
|
6638
|
|
323
|
6639 /*
|
|
|
6640 * Form the word with proper case in preword.
|
|
|
6641 * If there is a word from a previous split, append.
|
|
|
6642 */
|
|
|
6643 tword[sp->ts_twordlen] = NUL;
|
|
|
6644 if (flags & WF_KEEPCAP)
|
|
|
6645 /* Must find the word in the keep-case tree. */
|
|
|
6646 find_keepcap_word(lp->lp_slang, tword + splitoff,
|
|
|
6647 preword + prewordlen);
|
|
|
6648 else
|
|
346
|
6649 {
|
|
323
|
6650 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
6651 * use that for the goodword too. But if the badword is
|
|
|
6652 * allcap and it's only one char long use onecap. */
|
|
|
6653 c = su->su_badflags;
|
|
|
6654 if ((c & WF_ALLCAP)
|
|
|
6655 #ifdef FEAT_MBYTE
|
|
|
6656 && su->su_badlen == mb_ptr2len_check(su->su_badptr)
|
|
|
6657 #else
|
|
|
6658 && su->su_badlen == 1
|
|
|
6659 #endif
|
|
|
6660 )
|
|
|
6661 c = WF_ONECAP;
|
|
323
|
6662 make_case_word(tword + splitoff,
|
|
346
|
6663 preword + prewordlen, flags | c);
|
|
|
6664 }
|
|
323
|
6665
|
|
|
6666 /* Don't use a banned word. It may appear again as a good
|
|
|
6667 * word, thus remember it. */
|
|
|
6668 if (flags & WF_BANNED)
|
|
|
6669 {
|
|
|
6670 add_banned(su, preword + prewordlen);
|
|
|
6671 break;
|
|
|
6672 }
|
|
372
|
6673 if (was_banned(su, preword + prewordlen)
|
|
|
6674 || was_banned(su, preword))
|
|
323
|
6675 break;
|
|
|
6676
|
|
|
6677 newscore = 0;
|
|
|
6678 if ((flags & WF_REGION)
|
|
|
6679 && (((unsigned)flags >> 8) & lp->lp_region) == 0)
|
|
|
6680 newscore += SCORE_REGION;
|
|
|
6681 if (flags & WF_RARE)
|
|
|
6682 newscore += SCORE_RARE;
|
|
|
6683
|
|
346
|
6684 if (!spell_valid_case(su->su_badflags,
|
|
323
|
6685 captype(preword + prewordlen, NULL)))
|
|
|
6686 newscore += SCORE_ICASE;
|
|
|
6687
|
|
346
|
6688 if ((fword[sp->ts_fidx] == NUL
|
|
376
|
6689 || !spell_iswordp(fword + sp->ts_fidx, curbuf))
|
|
346
|
6690 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
6691 {
|
|
366
|
6692 /* The badword also ends: add suggestions. Give a penalty
|
|
|
6693 * when changing non-word char to word char, e.g., "thes,"
|
|
|
6694 * -> "these". */
|
|
|
6695 p = fword + sp->ts_fidx;
|
|
|
6696 #ifdef FEAT_MBYTE
|
|
|
6697 if (has_mbyte)
|
|
|
6698 mb_ptr_back(fword, p);
|
|
|
6699 else
|
|
|
6700 #endif
|
|
|
6701 --p;
|
|
376
|
6702 if (!spell_iswordp(p, curbuf))
|
|
366
|
6703 {
|
|
|
6704 p = preword + STRLEN(preword);
|
|
|
6705 #ifdef FEAT_MBYTE
|
|
|
6706 if (has_mbyte)
|
|
|
6707 mb_ptr_back(preword, p);
|
|
|
6708 else
|
|
|
6709 #endif
|
|
|
6710 --p;
|
|
376
|
6711 if (spell_iswordp(p, curbuf))
|
|
366
|
6712 newscore += SCORE_NONWORD;
|
|
|
6713 }
|
|
|
6714
|
|
344
|
6715 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
6716 sp->ts_fidx - repextra,
|
|
351
|
6717 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
6718 }
|
|
330
|
6719 else if (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
6720 #ifdef FEAT_MBYTE
|
|
|
6721 /* Don't split halfway a character. */
|
|
|
6722 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
6723 #endif
|
|
|
6724 )
|
|
323
|
6725 {
|
|
|
6726 /* The word in the tree ends but the badword
|
|
|
6727 * continues: try inserting a space and check that a valid
|
|
|
6728 * words starts at fword[sp->ts_fidx]. */
|
|
|
6729 if (try_deeper(su, stack, depth, newscore + SCORE_SPLIT))
|
|
|
6730 {
|
|
|
6731 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
|
6732 sp->ts_save_prewordlen = prewordlen;
|
|
346
|
6733 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
6734 sp->ts_save_splitoff = splitoff;
|
|
376
|
6735 sp->ts_state = STATE_SPLITUNDO;
|
|
|
6736
|
|
|
6737 ++depth;
|
|
|
6738 sp = &stack[depth];
|
|
323
|
6739
|
|
|
6740 /* Append a space to preword. */
|
|
|
6741 STRCAT(preword, " ");
|
|
|
6742 prewordlen = STRLEN(preword);
|
|
|
6743 splitoff = sp->ts_twordlen;
|
|
376
|
6744
|
|
|
6745 /* If the badword has a non-word character at this
|
|
|
6746 * position skip it. That means replacing the
|
|
|
6747 * non-word character with a space. */
|
|
|
6748 if (!spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
6749 {
|
|
|
6750 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
6751 #ifdef FEAT_MBYTE
|
|
|
6752 if (has_mbyte)
|
|
|
6753 sp->ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
|
6754 else
|
|
|
6755 #endif
|
|
|
6756 ++sp->ts_fidx;
|
|
|
6757 }
|
|
324
|
6758 #ifdef FEAT_MBYTE
|
|
|
6759 if (has_mbyte)
|
|
|
6760 {
|
|
|
6761 int i = 0;
|
|
|
6762
|
|
|
6763 /* Case-folding may change the number of bytes:
|
|
376
|
6764 * Count nr of chars in fword[ts_fidx] and
|
|
324
|
6765 * advance that many chars in su->su_badptr. */
|
|
|
6766 for (p = fword; p < fword + sp->ts_fidx;
|
|
|
6767 mb_ptr_adv(p))
|
|
|
6768 ++i;
|
|
|
6769 for (p = su->su_badptr; i > 0; mb_ptr_adv(p))
|
|
|
6770 --i;
|
|
|
6771 }
|
|
|
6772 else
|
|
|
6773 #endif
|
|
|
6774 p = su->su_badptr + sp->ts_fidx;
|
|
346
|
6775 su->su_badflags = captype(p, su->su_badptr
|
|
|
6776 + su->su_badlen);
|
|
323
|
6777
|
|
|
6778 /* Restart at top of the tree. */
|
|
376
|
6779 sp->ts_arridx = 0;
|
|
323
|
6780 }
|
|
|
6781 }
|
|
|
6782 break;
|
|
|
6783
|
|
|
6784 case STATE_SPLITUNDO:
|
|
346
|
6785 /* Undo the changes done for word split. */
|
|
|
6786 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
6787 splitoff = sp->ts_save_splitoff;
|
|
|
6788 prewordlen = sp->ts_save_prewordlen;
|
|
|
6789
|
|
|
6790 /* Continue looking for NUL bytes. */
|
|
|
6791 sp->ts_state = STATE_START;
|
|
|
6792 break;
|
|
|
6793
|
|
|
6794 case STATE_ENDNUL:
|
|
|
6795 /* Past the NUL bytes in the node. */
|
|
346
|
6796 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
6797 {
|
|
|
6798 /* The badword ends, can't use the bytes in this node. */
|
|
|
6799 sp->ts_state = STATE_DEL;
|
|
|
6800 break;
|
|
|
6801 }
|
|
|
6802 sp->ts_state = STATE_PLAIN;
|
|
|
6803 /*FALLTHROUGH*/
|
|
|
6804
|
|
|
6805 case STATE_PLAIN:
|
|
|
6806 /*
|
|
|
6807 * Go over all possible bytes at this node, add each to
|
|
|
6808 * tword[] and use child node. "ts_curi" is the index.
|
|
|
6809 */
|
|
|
6810 arridx = sp->ts_arridx;
|
|
|
6811 if (sp->ts_curi > byts[arridx])
|
|
|
6812 {
|
|
|
6813 /* Done all bytes at this node, do next state. When still
|
|
|
6814 * at already changed bytes skip the other tricks. */
|
|
|
6815 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
6816 sp->ts_state = STATE_DEL;
|
|
|
6817 else
|
|
|
6818 sp->ts_state = STATE_FINAL;
|
|
|
6819 }
|
|
|
6820 else
|
|
|
6821 {
|
|
|
6822 arridx += sp->ts_curi++;
|
|
|
6823 c = byts[arridx];
|
|
|
6824
|
|
|
6825 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
6826 * the byte in the bad word adjust the score. But don't
|
|
|
6827 * even try when the byte was already changed. */
|
|
330
|
6828 if (c == fword[sp->ts_fidx]
|
|
|
6829 #ifdef FEAT_MBYTE
|
|
|
6830 || (sp->ts_tcharlen > 0
|
|
|
6831 && sp->ts_isdiff != DIFF_NONE)
|
|
|
6832 #endif
|
|
|
6833 )
|
|
323
|
6834 newscore = 0;
|
|
|
6835 else
|
|
|
6836 newscore = SCORE_SUBST;
|
|
|
6837 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
6838 && try_deeper(su, stack, depth, newscore))
|
|
|
6839 {
|
|
|
6840 ++depth;
|
|
330
|
6841 sp = &stack[depth];
|
|
|
6842 ++sp->ts_fidx;
|
|
|
6843 tword[sp->ts_twordlen++] = c;
|
|
|
6844 sp->ts_arridx = idxs[arridx];
|
|
|
6845 #ifdef FEAT_MBYTE
|
|
|
6846 if (newscore == SCORE_SUBST)
|
|
|
6847 sp->ts_isdiff = DIFF_YES;
|
|
|
6848 if (has_mbyte)
|
|
|
6849 {
|
|
|
6850 /* Multi-byte characters are a bit complicated to
|
|
|
6851 * handle: They differ when any of the bytes
|
|
|
6852 * differ and then their length may also differ. */
|
|
|
6853 if (sp->ts_tcharlen == 0)
|
|
|
6854 {
|
|
|
6855 /* First byte. */
|
|
|
6856 sp->ts_tcharidx = 0;
|
|
|
6857 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
6858 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
6859 sp->ts_isdiff = (newscore != 0)
|
|
|
6860 ? DIFF_YES : DIFF_NONE;
|
|
|
6861 }
|
|
|
6862 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
6863 /* When inserting trail bytes don't advance in
|
|
|
6864 * the bad word. */
|
|
|
6865 --sp->ts_fidx;
|
|
|
6866 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
6867 {
|
|
|
6868 /* Last byte of character. */
|
|
|
6869 if (sp->ts_isdiff == DIFF_YES)
|
|
|
6870 {
|
|
|
6871 /* Correct ts_fidx for the byte length of
|
|
|
6872 * the character (we didn't check that
|
|
|
6873 * before). */
|
|
|
6874 sp->ts_fidx = sp->ts_fcharstart
|
|
|
6875 + MB_BYTE2LEN(
|
|
|
6876 fword[sp->ts_fcharstart]);
|
|
|
6877
|
|
|
6878 /* For a similar character adjust score
|
|
|
6879 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
|
6880 if (lp->lp_slang->sl_has_map
|
|
|
6881 && similar_chars(lp->lp_slang,
|
|
|
6882 mb_ptr2char(tword
|
|
|
6883 + sp->ts_twordlen
|
|
|
6884 - sp->ts_tcharlen),
|
|
|
6885 mb_ptr2char(fword
|
|
|
6886 + sp->ts_fcharstart)))
|
|
|
6887 sp->ts_score -=
|
|
|
6888 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
6889 }
|
|
358
|
6890 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
6891 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
6892 {
|
|
|
6893 /* If the previous character was the same,
|
|
|
6894 * thus doubling a character, give a bonus
|
|
|
6895 * to the score. */
|
|
|
6896 p = tword + sp->ts_twordlen
|
|
|
6897 - sp->ts_tcharlen;
|
|
|
6898 c = mb_ptr2char(p);
|
|
|
6899 mb_ptr_back(tword, p);
|
|
|
6900 if (c == mb_ptr2char(p))
|
|
|
6901 sp->ts_score -= SCORE_INS
|
|
|
6902 - SCORE_INSDUP;
|
|
|
6903 }
|
|
330
|
6904
|
|
|
6905 /* Starting a new char, reset the length. */
|
|
|
6906 sp->ts_tcharlen = 0;
|
|
|
6907 }
|
|
|
6908 }
|
|
|
6909 else
|
|
|
6910 #endif
|
|
|
6911 {
|
|
|
6912 /* If we found a similar char adjust the score.
|
|
|
6913 * We do this after calling try_deeper() because
|
|
|
6914 * it's slow. */
|
|
|
6915 if (newscore != 0
|
|
|
6916 && lp->lp_slang->sl_has_map
|
|
|
6917 && similar_chars(lp->lp_slang,
|
|
|
6918 c, fword[sp->ts_fidx - 1]))
|
|
|
6919 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
6920 }
|
|
323
|
6921 }
|
|
|
6922 }
|
|
|
6923 break;
|
|
|
6924
|
|
|
6925 case STATE_DEL:
|
|
330
|
6926 #ifdef FEAT_MBYTE
|
|
|
6927 /* When past the first byte of a multi-byte char don't try
|
|
|
6928 * delete/insert/swap a character. */
|
|
|
6929 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
6930 {
|
|
|
6931 sp->ts_state = STATE_FINAL;
|
|
|
6932 break;
|
|
|
6933 }
|
|
|
6934 #endif
|
|
|
6935 /*
|
|
|
6936 * Try skipping one character in the bad word (delete it).
|
|
|
6937 */
|
|
323
|
6938 sp->ts_state = STATE_INS;
|
|
|
6939 sp->ts_curi = 1;
|
|
|
6940 if (fword[sp->ts_fidx] != NUL
|
|
|
6941 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
6942 {
|
|
|
6943 ++depth;
|
|
358
|
6944
|
|
|
6945 /* Advance over the character in fword[]. Give a bonus to
|
|
|
6946 * the score if the same character is following "nn" ->
|
|
|
6947 * "n". */
|
|
330
|
6948 #ifdef FEAT_MBYTE
|
|
|
6949 if (has_mbyte)
|
|
358
|
6950 {
|
|
|
6951 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
6952 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
358
|
6953 if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
|
6954 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
6955 }
|
|
330
|
6956 else
|
|
|
6957 #endif
|
|
358
|
6958 {
|
|
330
|
6959 ++stack[depth].ts_fidx;
|
|
358
|
6960 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
6961 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
6962 }
|
|
323
|
6963 break;
|
|
|
6964 }
|
|
|
6965 /*FALLTHROUGH*/
|
|
|
6966
|
|
|
6967 case STATE_INS:
|
|
330
|
6968 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
6969 * node. */
|
|
|
6970 n = sp->ts_arridx;
|
|
|
6971 if (sp->ts_curi > byts[n])
|
|
|
6972 {
|
|
|
6973 /* Done all bytes at this node, do next state. */
|
|
|
6974 sp->ts_state = STATE_SWAP;
|
|
|
6975 }
|
|
|
6976 else
|
|
|
6977 {
|
|
330
|
6978 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
6979 n += sp->ts_curi++;
|
|
|
6980 c = byts[n];
|
|
|
6981 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
6982 {
|
|
|
6983 ++depth;
|
|
330
|
6984 sp = &stack[depth];
|
|
|
6985 tword[sp->ts_twordlen++] = c;
|
|
|
6986 sp->ts_arridx = idxs[n];
|
|
|
6987 #ifdef FEAT_MBYTE
|
|
|
6988 if (has_mbyte)
|
|
|
6989 {
|
|
|
6990 fl = MB_BYTE2LEN(c);
|
|
|
6991 if (fl > 1)
|
|
|
6992 {
|
|
|
6993 /* There are following bytes for the same
|
|
|
6994 * character. We must find all bytes before
|
|
|
6995 * trying delete/insert/swap/etc. */
|
|
|
6996 sp->ts_tcharlen = fl;
|
|
|
6997 sp->ts_tcharidx = 1;
|
|
|
6998 sp->ts_isdiff = DIFF_INSERT;
|
|
|
6999 }
|
|
|
7000 }
|
|
358
|
7001 else
|
|
|
7002 fl = 1;
|
|
|
7003 if (fl == 1)
|
|
330
|
7004 #endif
|
|
358
|
7005 {
|
|
|
7006 /* If the previous character was the same, thus
|
|
|
7007 * doubling a character, give a bonus to the
|
|
|
7008 * score. */
|
|
|
7009 if (sp->ts_twordlen >= 2
|
|
|
7010 && tword[sp->ts_twordlen - 2] == c)
|
|
|
7011 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
7012 }
|
|
323
|
7013 }
|
|
|
7014 }
|
|
|
7015 break;
|
|
|
7016
|
|
|
7017 case STATE_SWAP:
|
|
330
|
7018 /*
|
|
|
7019 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
7020 * We change "fword" here, it's changed back afterwards.
|
|
|
7021 */
|
|
|
7022 p = fword + sp->ts_fidx;
|
|
|
7023 c = *p;
|
|
|
7024 if (c == NUL)
|
|
|
7025 {
|
|
|
7026 /* End of word, can't swap or replace. */
|
|
|
7027 sp->ts_state = STATE_FINAL;
|
|
|
7028 break;
|
|
|
7029 }
|
|
|
7030 #ifdef FEAT_MBYTE
|
|
|
7031 if (has_mbyte)
|
|
|
7032 {
|
|
|
7033 n = mb_ptr2len_check(p);
|
|
|
7034 c = mb_ptr2char(p);
|
|
|
7035 c2 = mb_ptr2char(p + n);
|
|
|
7036 }
|
|
|
7037 else
|
|
|
7038 #endif
|
|
|
7039 c2 = p[1];
|
|
|
7040 if (c == c2)
|
|
323
|
7041 {
|
|
330
|
7042 /* Characters are identical, swap won't do anything. */
|
|
|
7043 sp->ts_state = STATE_SWAP3;
|
|
|
7044 break;
|
|
|
7045 }
|
|
|
7046 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
7047 {
|
|
|
7048 sp->ts_state = STATE_UNSWAP;
|
|
323
|
7049 ++depth;
|
|
330
|
7050 #ifdef FEAT_MBYTE
|
|
|
7051 if (has_mbyte)
|
|
|
7052 {
|
|
|
7053 fl = mb_char2len(c2);
|
|
|
7054 mch_memmove(p, p + n, fl);
|
|
|
7055 mb_char2bytes(c, p + fl);
|
|
|
7056 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7057 }
|
|
|
7058 else
|
|
|
7059 #endif
|
|
|
7060 {
|
|
|
7061 p[0] = c2;
|
|
|
7062 p[1] = c;
|
|
|
7063 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
7064 }
|
|
323
|
7065 }
|
|
|
7066 else
|
|
|
7067 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
7068 sp->ts_state = STATE_REP_INI;
|
|
|
7069 break;
|
|
|
7070
|
|
330
|
7071 case STATE_UNSWAP:
|
|
|
7072 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
7073 p = fword + sp->ts_fidx;
|
|
|
7074 #ifdef FEAT_MBYTE
|
|
|
7075 if (has_mbyte)
|
|
|
7076 {
|
|
|
7077 n = MB_BYTE2LEN(*p);
|
|
|
7078 c = mb_ptr2char(p + n);
|
|
|
7079 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
7080 mb_char2bytes(c, p);
|
|
|
7081 }
|
|
|
7082 else
|
|
|
7083 #endif
|
|
|
7084 {
|
|
|
7085 c = *p;
|
|
|
7086 *p = p[1];
|
|
|
7087 p[1] = c;
|
|
|
7088 }
|
|
|
7089 /*FALLTHROUGH*/
|
|
|
7090
|
|
|
7091 case STATE_SWAP3:
|
|
323
|
7092 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
7093 * "fword" here, it's changed back afterwards. */
|
|
|
7094 p = fword + sp->ts_fidx;
|
|
|
7095 #ifdef FEAT_MBYTE
|
|
|
7096 if (has_mbyte)
|
|
|
7097 {
|
|
|
7098 n = mb_ptr2len_check(p);
|
|
|
7099 c = mb_ptr2char(p);
|
|
|
7100 fl = mb_ptr2len_check(p + n);
|
|
|
7101 c2 = mb_ptr2char(p + n);
|
|
|
7102 c3 = mb_ptr2char(p + n + fl);
|
|
|
7103 }
|
|
|
7104 else
|
|
|
7105 #endif
|
|
323
|
7106 {
|
|
330
|
7107 c = *p;
|
|
|
7108 c2 = p[1];
|
|
|
7109 c3 = p[2];
|
|
|
7110 }
|
|
|
7111
|
|
|
7112 /* When characters are identical: "121" then SWAP3 result is
|
|
|
7113 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
7114 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
7115 * swapping. Also skip when c3 is NUL. */
|
|
|
7116 if (c == c3 || c3 == NUL)
|
|
|
7117 {
|
|
|
7118 sp->ts_state = STATE_REP_INI;
|
|
|
7119 break;
|
|
|
7120 }
|
|
|
7121 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7122 {
|
|
|
7123 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
7124 ++depth;
|
|
330
|
7125 #ifdef FEAT_MBYTE
|
|
|
7126 if (has_mbyte)
|
|
|
7127 {
|
|
|
7128 tl = mb_char2len(c3);
|
|
|
7129 mch_memmove(p, p + n + fl, tl);
|
|
|
7130 mb_char2bytes(c2, p + tl);
|
|
|
7131 mb_char2bytes(c, p + fl + tl);
|
|
|
7132 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
7133 }
|
|
|
7134 else
|
|
|
7135 #endif
|
|
|
7136 {
|
|
|
7137 p[0] = p[2];
|
|
|
7138 p[2] = c;
|
|
|
7139 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7140 }
|
|
323
|
7141 }
|
|
|
7142 else
|
|
|
7143 sp->ts_state = STATE_REP_INI;
|
|
|
7144 break;
|
|
|
7145
|
|
330
|
7146 case STATE_UNSWAP3:
|
|
|
7147 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
7148 p = fword + sp->ts_fidx;
|
|
|
7149 #ifdef FEAT_MBYTE
|
|
|
7150 if (has_mbyte)
|
|
|
7151 {
|
|
|
7152 n = MB_BYTE2LEN(*p);
|
|
|
7153 c2 = mb_ptr2char(p + n);
|
|
|
7154 fl = MB_BYTE2LEN(p[n]);
|
|
|
7155 c = mb_ptr2char(p + n + fl);
|
|
|
7156 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
7157 mch_memmove(p + fl + tl, p, n);
|
|
|
7158 mb_char2bytes(c, p);
|
|
|
7159 mb_char2bytes(c2, p + tl);
|
|
|
7160 }
|
|
|
7161 else
|
|
|
7162 #endif
|
|
|
7163 {
|
|
|
7164 c = *p;
|
|
|
7165 *p = p[2];
|
|
|
7166 p[2] = c;
|
|
|
7167 }
|
|
346
|
7168
|
|
330
|
7169 /* Rotate three characters left: "123" -> "231". We change
|
|
|
7170 * "fword" here, it's changed back afterwards. */
|
|
323
|
7171 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7172 {
|
|
330
|
7173 sp->ts_state = STATE_UNROT3L;
|
|
323
|
7174 ++depth;
|
|
330
|
7175 p = fword + sp->ts_fidx;
|
|
|
7176 #ifdef FEAT_MBYTE
|
|
|
7177 if (has_mbyte)
|
|
|
7178 {
|
|
|
7179 n = mb_ptr2len_check(p);
|
|
|
7180 c = mb_ptr2char(p);
|
|
|
7181 fl = mb_ptr2len_check(p + n);
|
|
|
7182 fl += mb_ptr2len_check(p + n + fl);
|
|
|
7183 mch_memmove(p, p + n, fl);
|
|
|
7184 mb_char2bytes(c, p + fl);
|
|
|
7185 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7186 }
|
|
|
7187 else
|
|
|
7188 #endif
|
|
|
7189 {
|
|
|
7190 c = *p;
|
|
|
7191 *p = p[1];
|
|
|
7192 p[1] = p[2];
|
|
|
7193 p[2] = c;
|
|
|
7194 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7195 }
|
|
323
|
7196 }
|
|
|
7197 else
|
|
|
7198 sp->ts_state = STATE_REP_INI;
|
|
|
7199 break;
|
|
|
7200
|
|
330
|
7201 case STATE_UNROT3L:
|
|
346
|
7202 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
7203 p = fword + sp->ts_fidx;
|
|
|
7204 #ifdef FEAT_MBYTE
|
|
|
7205 if (has_mbyte)
|
|
|
7206 {
|
|
|
7207 n = MB_BYTE2LEN(*p);
|
|
|
7208 n += MB_BYTE2LEN(p[n]);
|
|
|
7209 c = mb_ptr2char(p + n);
|
|
|
7210 tl = MB_BYTE2LEN(p[n]);
|
|
|
7211 mch_memmove(p + tl, p, n);
|
|
|
7212 mb_char2bytes(c, p);
|
|
|
7213 }
|
|
|
7214 else
|
|
|
7215 #endif
|
|
|
7216 {
|
|
|
7217 c = p[2];
|
|
|
7218 p[2] = p[1];
|
|
|
7219 p[1] = *p;
|
|
|
7220 *p = c;
|
|
|
7221 }
|
|
346
|
7222
|
|
323
|
7223 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
7224 * "fword" here, it's changed back afterwards. */
|
|
323
|
7225 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7226 {
|
|
330
|
7227 sp->ts_state = STATE_UNROT3R;
|
|
323
|
7228 ++depth;
|
|
330
|
7229 p = fword + sp->ts_fidx;
|
|
|
7230 #ifdef FEAT_MBYTE
|
|
|
7231 if (has_mbyte)
|
|
|
7232 {
|
|
|
7233 n = mb_ptr2len_check(p);
|
|
|
7234 n += mb_ptr2len_check(p + n);
|
|
|
7235 c = mb_ptr2char(p + n);
|
|
|
7236 tl = mb_ptr2len_check(p + n);
|
|
|
7237 mch_memmove(p + tl, p, n);
|
|
|
7238 mb_char2bytes(c, p);
|
|
|
7239 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
7240 }
|
|
|
7241 else
|
|
|
7242 #endif
|
|
|
7243 {
|
|
|
7244 c = p[2];
|
|
|
7245 p[2] = p[1];
|
|
|
7246 p[1] = *p;
|
|
|
7247 *p = c;
|
|
|
7248 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7249 }
|
|
323
|
7250 }
|
|
|
7251 else
|
|
|
7252 sp->ts_state = STATE_REP_INI;
|
|
|
7253 break;
|
|
|
7254
|
|
330
|
7255 case STATE_UNROT3R:
|
|
346
|
7256 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
7257 p = fword + sp->ts_fidx;
|
|
|
7258 #ifdef FEAT_MBYTE
|
|
|
7259 if (has_mbyte)
|
|
|
7260 {
|
|
|
7261 c = mb_ptr2char(p);
|
|
|
7262 tl = MB_BYTE2LEN(*p);
|
|
|
7263 n = MB_BYTE2LEN(p[tl]);
|
|
|
7264 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
7265 mch_memmove(p, p + tl, n);
|
|
|
7266 mb_char2bytes(c, p + n);
|
|
|
7267 }
|
|
|
7268 else
|
|
|
7269 #endif
|
|
|
7270 {
|
|
|
7271 c = *p;
|
|
|
7272 *p = p[1];
|
|
|
7273 p[1] = p[2];
|
|
|
7274 p[2] = c;
|
|
|
7275 }
|
|
323
|
7276 /*FALLTHROUGH*/
|
|
|
7277
|
|
|
7278 case STATE_REP_INI:
|
|
|
7279 /* Check if matching with REP items from the .aff file would
|
|
|
7280 * work. Quickly skip if there are no REP items or the score
|
|
|
7281 * is going to be too high anyway. */
|
|
|
7282 gap = &lp->lp_slang->sl_rep;
|
|
|
7283 if (gap->ga_len == 0
|
|
|
7284 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
7285 {
|
|
|
7286 sp->ts_state = STATE_FINAL;
|
|
|
7287 break;
|
|
|
7288 }
|
|
|
7289
|
|
|
7290 /* Use the first byte to quickly find the first entry that
|
|
330
|
7291 * may match. If the index is -1 there is none. */
|
|
323
|
7292 sp->ts_curi = lp->lp_slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
7293 if (sp->ts_curi < 0)
|
|
|
7294 {
|
|
|
7295 sp->ts_state = STATE_FINAL;
|
|
|
7296 break;
|
|
|
7297 }
|
|
|
7298
|
|
|
7299 sp->ts_state = STATE_REP;
|
|
|
7300 /*FALLTHROUGH*/
|
|
|
7301
|
|
|
7302 case STATE_REP:
|
|
|
7303 /* Try matching with REP items from the .aff file. For each
|
|
330
|
7304 * match replace the characters and check if the resulting
|
|
|
7305 * word is valid. */
|
|
323
|
7306 p = fword + sp->ts_fidx;
|
|
|
7307
|
|
|
7308 gap = &lp->lp_slang->sl_rep;
|
|
|
7309 while (sp->ts_curi < gap->ga_len)
|
|
|
7310 {
|
|
|
7311 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
7312 if (*ftp->ft_from != *p)
|
|
|
7313 {
|
|
|
7314 /* past possible matching entries */
|
|
|
7315 sp->ts_curi = gap->ga_len;
|
|
|
7316 break;
|
|
|
7317 }
|
|
|
7318 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
7319 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
7320 {
|
|
|
7321 /* Need to undo this afterwards. */
|
|
|
7322 sp->ts_state = STATE_REP_UNDO;
|
|
|
7323
|
|
|
7324 /* Change the "from" to the "to" string. */
|
|
|
7325 ++depth;
|
|
|
7326 fl = STRLEN(ftp->ft_from);
|
|
|
7327 tl = STRLEN(ftp->ft_to);
|
|
|
7328 if (fl != tl)
|
|
346
|
7329 {
|
|
323
|
7330 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
7331 repextra += tl - fl;
|
|
|
7332 }
|
|
323
|
7333 mch_memmove(p, ftp->ft_to, tl);
|
|
|
7334 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
7335 #ifdef FEAT_MBYTE
|
|
|
7336 stack[depth].ts_tcharlen = 0;
|
|
|
7337 #endif
|
|
323
|
7338 break;
|
|
|
7339 }
|
|
|
7340 }
|
|
|
7341
|
|
|
7342 if (sp->ts_curi >= gap->ga_len)
|
|
|
7343 /* No (more) matches. */
|
|
|
7344 sp->ts_state = STATE_FINAL;
|
|
|
7345
|
|
|
7346 break;
|
|
|
7347
|
|
|
7348 case STATE_REP_UNDO:
|
|
|
7349 /* Undo a REP replacement and continue with the next one. */
|
|
|
7350 ftp = (fromto_T *)lp->lp_slang->sl_rep.ga_data
|
|
|
7351 + sp->ts_curi - 1;
|
|
|
7352 fl = STRLEN(ftp->ft_from);
|
|
|
7353 tl = STRLEN(ftp->ft_to);
|
|
|
7354 p = fword + sp->ts_fidx;
|
|
|
7355 if (fl != tl)
|
|
346
|
7356 {
|
|
323
|
7357 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
7358 repextra -= tl - fl;
|
|
|
7359 }
|
|
323
|
7360 mch_memmove(p, ftp->ft_from, fl);
|
|
|
7361 sp->ts_state = STATE_REP;
|
|
|
7362 break;
|
|
|
7363
|
|
|
7364 default:
|
|
|
7365 /* Did all possible states at this level, go up one level. */
|
|
|
7366 --depth;
|
|
344
|
7367
|
|
375
|
7368 if (depth >= 0 && stack[depth].ts_prefixdepth == PREFIXTREE)
|
|
|
7369 {
|
|
|
7370 /* Continue in or go back to the prefix tree. */
|
|
|
7371 byts = pbyts;
|
|
|
7372 idxs = pidxs;
|
|
|
7373 splitoff = 0;
|
|
|
7374 }
|
|
|
7375
|
|
344
|
7376 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
7377 line_breakcheck();
|
|
323
|
7378 }
|
|
|
7379 }
|
|
|
7380 }
|
|
|
7381 }
|
|
|
7382
|
|
|
7383 /*
|
|
|
7384 * Try going one level deeper in the tree.
|
|
|
7385 */
|
|
|
7386 static int
|
|
|
7387 try_deeper(su, stack, depth, score_add)
|
|
|
7388 suginfo_T *su;
|
|
|
7389 trystate_T *stack;
|
|
|
7390 int depth;
|
|
|
7391 int score_add;
|
|
|
7392 {
|
|
|
7393 int newscore;
|
|
|
7394
|
|
|
7395 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
7396 newscore = stack[depth].ts_score + score_add;
|
|
|
7397 if (newscore >= su->su_maxscore)
|
|
|
7398 return FALSE;
|
|
|
7399
|
|
330
|
7400 stack[depth + 1] = stack[depth];
|
|
323
|
7401 stack[depth + 1].ts_state = STATE_START;
|
|
|
7402 stack[depth + 1].ts_score = newscore;
|
|
|
7403 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
7404 return TRUE;
|
|
|
7405 }
|
|
|
7406
|
|
|
7407 /*
|
|
|
7408 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
7409 * words and put it in "kword".
|
|
|
7410 * Theoretically there could be several keep-case words that result in the
|
|
|
7411 * same case-folded word, but we only find one...
|
|
|
7412 */
|
|
|
7413 static void
|
|
|
7414 find_keepcap_word(slang, fword, kword)
|
|
|
7415 slang_T *slang;
|
|
|
7416 char_u *fword;
|
|
|
7417 char_u *kword;
|
|
|
7418 {
|
|
|
7419 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
7420 int depth;
|
|
324
|
7421 idx_T tryidx;
|
|
323
|
7422
|
|
|
7423 /* The following arrays are used at each depth in the tree. */
|
|
324
|
7424 idx_T arridx[MAXWLEN];
|
|
323
|
7425 int round[MAXWLEN];
|
|
|
7426 int fwordidx[MAXWLEN];
|
|
|
7427 int uwordidx[MAXWLEN];
|
|
|
7428 int kwordlen[MAXWLEN];
|
|
|
7429
|
|
|
7430 int flen, ulen;
|
|
|
7431 int l;
|
|
|
7432 int len;
|
|
|
7433 int c;
|
|
324
|
7434 idx_T lo, hi, m;
|
|
323
|
7435 char_u *p;
|
|
|
7436 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
7437 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
7438
|
|
|
7439 if (byts == NULL)
|
|
|
7440 {
|
|
|
7441 /* array is empty: "cannot happen" */
|
|
|
7442 *kword = NUL;
|
|
|
7443 return;
|
|
|
7444 }
|
|
|
7445
|
|
|
7446 /* Make an all-cap version of "fword". */
|
|
|
7447 allcap_copy(fword, uword);
|
|
|
7448
|
|
|
7449 /*
|
|
|
7450 * Each character needs to be tried both case-folded and upper-case.
|
|
|
7451 * All this gets very complicated if we keep in mind that changing case
|
|
|
7452 * may change the byte length of a multi-byte character...
|
|
|
7453 */
|
|
|
7454 depth = 0;
|
|
|
7455 arridx[0] = 0;
|
|
|
7456 round[0] = 0;
|
|
|
7457 fwordidx[0] = 0;
|
|
|
7458 uwordidx[0] = 0;
|
|
|
7459 kwordlen[0] = 0;
|
|
|
7460 while (depth >= 0)
|
|
|
7461 {
|
|
|
7462 if (fword[fwordidx[depth]] == NUL)
|
|
|
7463 {
|
|
|
7464 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
7465 * here we have found a match. */
|
|
|
7466 if (byts[arridx[depth] + 1] == 0)
|
|
|
7467 {
|
|
|
7468 kword[kwordlen[depth]] = NUL;
|
|
|
7469 return;
|
|
|
7470 }
|
|
|
7471
|
|
|
7472 /* kword is getting too long, continue one level up */
|
|
|
7473 --depth;
|
|
|
7474 }
|
|
|
7475 else if (++round[depth] > 2)
|
|
|
7476 {
|
|
|
7477 /* tried both fold-case and upper-case character, continue one
|
|
|
7478 * level up */
|
|
|
7479 --depth;
|
|
|
7480 }
|
|
|
7481 else
|
|
|
7482 {
|
|
|
7483 /*
|
|
|
7484 * round[depth] == 1: Try using the folded-case character.
|
|
|
7485 * round[depth] == 2: Try using the upper-case character.
|
|
|
7486 */
|
|
|
7487 #ifdef FEAT_MBYTE
|
|
|
7488 if (has_mbyte)
|
|
|
7489 {
|
|
|
7490 flen = mb_ptr2len_check(fword + fwordidx[depth]);
|
|
|
7491 ulen = mb_ptr2len_check(uword + uwordidx[depth]);
|
|
|
7492 }
|
|
|
7493 else
|
|
|
7494 #endif
|
|
|
7495 ulen = flen = 1;
|
|
|
7496 if (round[depth] == 1)
|
|
|
7497 {
|
|
|
7498 p = fword + fwordidx[depth];
|
|
|
7499 l = flen;
|
|
|
7500 }
|
|
|
7501 else
|
|
|
7502 {
|
|
|
7503 p = uword + uwordidx[depth];
|
|
|
7504 l = ulen;
|
|
|
7505 }
|
|
|
7506
|
|
|
7507 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
7508 {
|
|
|
7509 /* Perform a binary search in the list of accepted bytes. */
|
|
|
7510 len = byts[tryidx++];
|
|
|
7511 c = *p++;
|
|
|
7512 lo = tryidx;
|
|
|
7513 hi = tryidx + len - 1;
|
|
|
7514 while (lo < hi)
|
|
|
7515 {
|
|
|
7516 m = (lo + hi) / 2;
|
|
|
7517 if (byts[m] > c)
|
|
|
7518 hi = m - 1;
|
|
|
7519 else if (byts[m] < c)
|
|
|
7520 lo = m + 1;
|
|
|
7521 else
|
|
|
7522 {
|
|
|
7523 lo = hi = m;
|
|
|
7524 break;
|
|
|
7525 }
|
|
|
7526 }
|
|
|
7527
|
|
|
7528 /* Stop if there is no matching byte. */
|
|
|
7529 if (hi < lo || byts[lo] != c)
|
|
|
7530 break;
|
|
|
7531
|
|
|
7532 /* Continue at the child (if there is one). */
|
|
|
7533 tryidx = idxs[lo];
|
|
|
7534 }
|
|
|
7535
|
|
|
7536 if (l == 0)
|
|
|
7537 {
|
|
|
7538 /*
|
|
|
7539 * Found the matching char. Copy it to "kword" and go a
|
|
|
7540 * level deeper.
|
|
|
7541 */
|
|
|
7542 if (round[depth] == 1)
|
|
|
7543 {
|
|
|
7544 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
7545 flen);
|
|
|
7546 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
7547 }
|
|
|
7548 else
|
|
|
7549 {
|
|
|
7550 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
7551 ulen);
|
|
|
7552 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
7553 }
|
|
|
7554 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
7555 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
7556
|
|
|
7557 ++depth;
|
|
|
7558 arridx[depth] = tryidx;
|
|
|
7559 round[depth] = 0;
|
|
|
7560 }
|
|
|
7561 }
|
|
|
7562 }
|
|
|
7563
|
|
|
7564 /* Didn't find it: "cannot happen". */
|
|
|
7565 *kword = NUL;
|
|
|
7566 }
|
|
|
7567
|
|
|
7568 /*
|
|
344
|
7569 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
7570 * su->su_sga.
|
|
|
7571 */
|
|
|
7572 static void
|
|
|
7573 score_comp_sal(su)
|
|
|
7574 suginfo_T *su;
|
|
|
7575 {
|
|
|
7576 langp_T *lp;
|
|
|
7577 char_u badsound[MAXWLEN];
|
|
|
7578 int i;
|
|
|
7579 suggest_T *stp;
|
|
|
7580 suggest_T *sstp;
|
|
|
7581 int score;
|
|
|
7582
|
|
|
7583 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
7584 return;
|
|
|
7585
|
|
|
7586 /* Use the sound-folding of the first language that supports it. */
|
|
|
7587 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7588 lp->lp_slang != NULL; ++lp)
|
|
|
7589 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7590 {
|
|
|
7591 /* soundfold the bad word */
|
|
375
|
7592 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
7593
|
|
|
7594 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7595 {
|
|
|
7596 stp = &SUG(su->su_ga, i);
|
|
|
7597
|
|
351
|
7598 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
7599 * sound-a-like score. */
|
|
|
7600 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
7601 if (score < SCORE_MAXMAX)
|
|
|
7602 {
|
|
|
7603 /* Add the suggestion. */
|
|
|
7604 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
7605 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
7606 if (sstp->st_word != NULL)
|
|
|
7607 {
|
|
|
7608 sstp->st_score = score;
|
|
|
7609 sstp->st_altscore = 0;
|
|
|
7610 sstp->st_orglen = stp->st_orglen;
|
|
|
7611 ++su->su_sga.ga_len;
|
|
|
7612 }
|
|
|
7613 }
|
|
|
7614 }
|
|
|
7615 break;
|
|
|
7616 }
|
|
|
7617 }
|
|
|
7618
|
|
|
7619 /*
|
|
|
7620 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
7621 * They are intwined.
|
|
|
7622 */
|
|
|
7623 static void
|
|
|
7624 score_combine(su)
|
|
|
7625 suginfo_T *su;
|
|
|
7626 {
|
|
|
7627 int i;
|
|
|
7628 int j;
|
|
|
7629 garray_T ga;
|
|
|
7630 garray_T *gap;
|
|
|
7631 langp_T *lp;
|
|
|
7632 suggest_T *stp;
|
|
|
7633 char_u *p;
|
|
|
7634 char_u badsound[MAXWLEN];
|
|
|
7635 int round;
|
|
|
7636
|
|
|
7637 /* Add the alternate score to su_ga. */
|
|
|
7638 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7639 lp->lp_slang != NULL; ++lp)
|
|
|
7640 {
|
|
|
7641 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7642 {
|
|
|
7643 /* soundfold the bad word */
|
|
375
|
7644 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
7645
|
|
|
7646 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7647 {
|
|
|
7648 stp = &SUG(su->su_ga, i);
|
|
351
|
7649 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
7650 badsound);
|
|
344
|
7651 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
7652 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
7653 else
|
|
|
7654 stp->st_score = (stp->st_score * 3
|
|
|
7655 + stp->st_altscore) / 4;
|
|
|
7656 stp->st_salscore = FALSE;
|
|
|
7657 }
|
|
|
7658 break;
|
|
|
7659 }
|
|
|
7660 }
|
|
|
7661
|
|
|
7662 /* Add the alternate score to su_sga. */
|
|
|
7663 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
7664 {
|
|
|
7665 stp = &SUG(su->su_sga, i);
|
|
|
7666 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
7667 if (stp->st_score == SCORE_MAXMAX)
|
|
|
7668 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
7669 else
|
|
|
7670 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
7671 stp->st_salscore = TRUE;
|
|
|
7672 }
|
|
|
7673
|
|
|
7674 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
7675 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7676 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
7677
|
|
|
7678 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
7679 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
7680 return;
|
|
|
7681
|
|
|
7682 stp = &SUG(ga, 0);
|
|
|
7683 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
7684 {
|
|
|
7685 /* round 1: get a suggestion from su_ga
|
|
|
7686 * round 2: get a suggestion from su_sga */
|
|
|
7687 for (round = 1; round <= 2; ++round)
|
|
|
7688 {
|
|
|
7689 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
7690 if (i < gap->ga_len)
|
|
|
7691 {
|
|
|
7692 /* Don't add a word if it's already there. */
|
|
|
7693 p = SUG(*gap, i).st_word;
|
|
|
7694 for (j = 0; j < ga.ga_len; ++j)
|
|
|
7695 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
7696 break;
|
|
|
7697 if (j == ga.ga_len)
|
|
|
7698 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
7699 else
|
|
|
7700 vim_free(p);
|
|
|
7701 }
|
|
|
7702 }
|
|
|
7703 }
|
|
|
7704
|
|
|
7705 ga_clear(&su->su_ga);
|
|
|
7706 ga_clear(&su->su_sga);
|
|
|
7707
|
|
|
7708 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
7709 if (ga.ga_len > su->su_maxcount)
|
|
|
7710 {
|
|
|
7711 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
7712 vim_free(stp[i].st_word);
|
|
|
7713 ga.ga_len = su->su_maxcount;
|
|
|
7714 }
|
|
|
7715
|
|
|
7716 su->su_ga = ga;
|
|
|
7717 }
|
|
|
7718
|
|
|
7719 /*
|
|
351
|
7720 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
7721 * badword.
|
|
|
7722 */
|
|
|
7723 static int
|
|
|
7724 stp_sal_score(stp, su, slang, badsound)
|
|
|
7725 suggest_T *stp;
|
|
|
7726 suginfo_T *su;
|
|
|
7727 slang_T *slang;
|
|
|
7728 char_u *badsound; /* sound-folded badword */
|
|
|
7729 {
|
|
|
7730 char_u *p;
|
|
|
7731 char_u badsound2[MAXWLEN];
|
|
|
7732 char_u fword[MAXWLEN];
|
|
|
7733 char_u goodsound[MAXWLEN];
|
|
|
7734
|
|
|
7735 if (stp->st_orglen <= su->su_badlen)
|
|
|
7736 p = badsound;
|
|
|
7737 else
|
|
|
7738 {
|
|
|
7739 /* soundfold the bad word with more characters following */
|
|
|
7740 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
7741
|
|
|
7742 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
7743 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
7744 * removing the space. Don't do it when the good word also contains a
|
|
|
7745 * space. */
|
|
|
7746 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
7747 && *skiptowhite(stp->st_word) == NUL)
|
|
|
7748 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
7749 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
7750
|
|
375
|
7751 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
7752 p = badsound2;
|
|
|
7753 }
|
|
|
7754
|
|
375
|
7755 /* Sound-fold the word and compute the score for the difference. */
|
|
|
7756 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
7757
|
|
|
7758 return soundalike_score(goodsound, p);
|
|
|
7759 }
|
|
|
7760
|
|
|
7761 /*
|
|
323
|
7762 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
7763 */
|
|
|
7764 static void
|
|
346
|
7765 suggest_try_soundalike(su)
|
|
323
|
7766 suginfo_T *su;
|
|
|
7767 {
|
|
|
7768 char_u salword[MAXWLEN];
|
|
|
7769 char_u tword[MAXWLEN];
|
|
|
7770 char_u tsalword[MAXWLEN];
|
|
324
|
7771 idx_T arridx[MAXWLEN];
|
|
323
|
7772 int curi[MAXWLEN];
|
|
|
7773 langp_T *lp;
|
|
|
7774 char_u *byts;
|
|
324
|
7775 idx_T *idxs;
|
|
323
|
7776 int depth;
|
|
|
7777 int c;
|
|
324
|
7778 idx_T n;
|
|
323
|
7779 int round;
|
|
|
7780 int flags;
|
|
344
|
7781 int sound_score;
|
|
|
7782
|
|
|
7783 /* Do this for all languages that support sound folding. */
|
|
323
|
7784 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7785 lp->lp_slang != NULL; ++lp)
|
|
|
7786 {
|
|
|
7787 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7788 {
|
|
|
7789 /* soundfold the bad word */
|
|
375
|
7790 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
7791
|
|
|
7792 /*
|
|
|
7793 * Go through the whole tree, soundfold each word and compare.
|
|
|
7794 * round 1: use the case-folded tree.
|
|
|
7795 * round 2: use the keep-case tree.
|
|
|
7796 */
|
|
|
7797 for (round = 1; round <= 2; ++round)
|
|
|
7798 {
|
|
|
7799 if (round == 1)
|
|
|
7800 {
|
|
|
7801 byts = lp->lp_slang->sl_fbyts;
|
|
|
7802 idxs = lp->lp_slang->sl_fidxs;
|
|
|
7803 }
|
|
|
7804 else
|
|
|
7805 {
|
|
|
7806 byts = lp->lp_slang->sl_kbyts;
|
|
|
7807 idxs = lp->lp_slang->sl_kidxs;
|
|
|
7808 }
|
|
|
7809
|
|
|
7810 depth = 0;
|
|
|
7811 arridx[0] = 0;
|
|
|
7812 curi[0] = 1;
|
|
|
7813 while (depth >= 0 && !got_int)
|
|
|
7814 {
|
|
|
7815 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
7816 {
|
|
323
|
7817 /* Done all bytes at this node, go up one level. */
|
|
|
7818 --depth;
|
|
351
|
7819 line_breakcheck();
|
|
|
7820 }
|
|
323
|
7821 else
|
|
|
7822 {
|
|
|
7823 /* Do one more byte at this node. */
|
|
|
7824 n = arridx[depth] + curi[depth];
|
|
|
7825 ++curi[depth];
|
|
|
7826 c = byts[n];
|
|
|
7827 if (c == 0)
|
|
|
7828 {
|
|
|
7829 /* End of word, deal with the word. */
|
|
324
|
7830 flags = (int)idxs[n];
|
|
323
|
7831 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
7832 {
|
|
|
7833 tword[depth] = NUL;
|
|
375
|
7834 /* Sound-fold. Only in keep-case tree need to
|
|
|
7835 * case-fold the word. */
|
|
|
7836 spell_soundfold(lp->lp_slang, tword,
|
|
|
7837 round == 1, tsalword);
|
|
323
|
7838
|
|
344
|
7839 /* Compute the edit distance between the
|
|
|
7840 * sound-a-like words. */
|
|
|
7841 sound_score = soundalike_score(salword,
|
|
|
7842 tsalword);
|
|
324
|
7843 if (sound_score < SCORE_MAXMAX)
|
|
323
|
7844 {
|
|
324
|
7845 char_u cword[MAXWLEN];
|
|
|
7846 char_u *p;
|
|
344
|
7847 int score;
|
|
324
|
7848
|
|
351
|
7849 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
7850 {
|
|
324
|
7851 /* Need to fix case according to
|
|
|
7852 * "flags". */
|
|
323
|
7853 make_case_word(tword, cword, flags);
|
|
324
|
7854 p = cword;
|
|
323
|
7855 }
|
|
|
7856 else
|
|
324
|
7857 p = tword;
|
|
|
7858
|
|
344
|
7859 if (sps_flags & SPS_DOUBLE)
|
|
|
7860 add_suggestion(su, &su->su_sga, p,
|
|
346
|
7861 su->su_badlen,
|
|
351
|
7862 sound_score, 0, FALSE);
|
|
344
|
7863 else
|
|
|
7864 {
|
|
|
7865 /* Compute the score. */
|
|
|
7866 score = spell_edit_score(
|
|
|
7867 su->su_badword, p);
|
|
|
7868 if (sps_flags & SPS_BEST)
|
|
|
7869 /* give a bonus for the good word
|
|
|
7870 * sounding the same as the bad
|
|
|
7871 * word */
|
|
|
7872 add_suggestion(su, &su->su_ga, p,
|
|
346
|
7873 su->su_badlen,
|
|
344
|
7874 RESCORE(score, sound_score),
|
|
351
|
7875 sound_score, TRUE);
|
|
344
|
7876 else
|
|
|
7877 add_suggestion(su, &su->su_ga, p,
|
|
346
|
7878 su->su_badlen,
|
|
351
|
7879 score + sound_score, 0, FALSE);
|
|
344
|
7880 }
|
|
323
|
7881 }
|
|
|
7882 }
|
|
|
7883
|
|
|
7884 /* Skip over other NUL bytes. */
|
|
|
7885 while (byts[n + 1] == 0)
|
|
|
7886 {
|
|
|
7887 ++n;
|
|
|
7888 ++curi[depth];
|
|
|
7889 }
|
|
|
7890 }
|
|
|
7891 else
|
|
|
7892 {
|
|
|
7893 /* Normal char, go one level deeper. */
|
|
|
7894 tword[depth++] = c;
|
|
|
7895 arridx[depth] = idxs[n];
|
|
|
7896 curi[depth] = 1;
|
|
|
7897 }
|
|
|
7898 }
|
|
|
7899 }
|
|
|
7900 }
|
|
|
7901 }
|
|
|
7902 }
|
|
|
7903 }
|
|
|
7904
|
|
|
7905 /*
|
|
324
|
7906 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
7907 */
|
|
|
7908 static void
|
|
|
7909 make_case_word(fword, cword, flags)
|
|
|
7910 char_u *fword;
|
|
|
7911 char_u *cword;
|
|
|
7912 int flags;
|
|
|
7913 {
|
|
|
7914 if (flags & WF_ALLCAP)
|
|
|
7915 /* Make it all upper-case */
|
|
|
7916 allcap_copy(fword, cword);
|
|
|
7917 else if (flags & WF_ONECAP)
|
|
|
7918 /* Make the first letter upper-case */
|
|
324
|
7919 onecap_copy(fword, cword, TRUE);
|
|
323
|
7920 else
|
|
|
7921 /* Use goodword as-is. */
|
|
|
7922 STRCPY(cword, fword);
|
|
|
7923 }
|
|
|
7924
|
|
330
|
7925 /*
|
|
|
7926 * Use map string "map" for languages "lp".
|
|
|
7927 */
|
|
|
7928 static void
|
|
|
7929 set_map_str(lp, map)
|
|
|
7930 slang_T *lp;
|
|
|
7931 char_u *map;
|
|
|
7932 {
|
|
|
7933 char_u *p;
|
|
|
7934 int headc = 0;
|
|
|
7935 int c;
|
|
|
7936 int i;
|
|
|
7937
|
|
|
7938 if (*map == NUL)
|
|
|
7939 {
|
|
|
7940 lp->sl_has_map = FALSE;
|
|
|
7941 return;
|
|
|
7942 }
|
|
|
7943 lp->sl_has_map = TRUE;
|
|
|
7944
|
|
|
7945 /* Init the array and hash table empty. */
|
|
|
7946 for (i = 0; i < 256; ++i)
|
|
|
7947 lp->sl_map_array[i] = 0;
|
|
|
7948 #ifdef FEAT_MBYTE
|
|
|
7949 hash_init(&lp->sl_map_hash);
|
|
|
7950 #endif
|
|
|
7951
|
|
|
7952 /*
|
|
|
7953 * The similar characters are stored separated with slashes:
|
|
|
7954 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
7955 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
7956 */
|
|
|
7957 for (p = map; *p != NUL; )
|
|
|
7958 {
|
|
|
7959 #ifdef FEAT_MBYTE
|
|
|
7960 c = mb_ptr2char_adv(&p);
|
|
|
7961 #else
|
|
|
7962 c = *p++;
|
|
|
7963 #endif
|
|
|
7964 if (c == '/')
|
|
|
7965 headc = 0;
|
|
|
7966 else
|
|
|
7967 {
|
|
|
7968 if (headc == 0)
|
|
|
7969 headc = c;
|
|
|
7970
|
|
|
7971 #ifdef FEAT_MBYTE
|
|
|
7972 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
7973 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
7974 * a NUL. */
|
|
|
7975 if (c >= 256)
|
|
|
7976 {
|
|
|
7977 int cl = mb_char2len(c);
|
|
|
7978 int headcl = mb_char2len(headc);
|
|
|
7979 char_u *b;
|
|
|
7980 hash_T hash;
|
|
|
7981 hashitem_T *hi;
|
|
|
7982
|
|
|
7983 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
7984 if (b == NULL)
|
|
|
7985 return;
|
|
|
7986 mb_char2bytes(c, b);
|
|
|
7987 b[cl] = NUL;
|
|
|
7988 mb_char2bytes(headc, b + cl + 1);
|
|
|
7989 b[cl + 1 + headcl] = NUL;
|
|
|
7990 hash = hash_hash(b);
|
|
|
7991 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
7992 if (HASHITEM_EMPTY(hi))
|
|
|
7993 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
7994 else
|
|
|
7995 {
|
|
|
7996 /* This should have been checked when generating the .spl
|
|
|
7997 * file. */
|
|
|
7998 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
7999 vim_free(b);
|
|
|
8000 }
|
|
|
8001 }
|
|
|
8002 else
|
|
|
8003 #endif
|
|
|
8004 lp->sl_map_array[c] = headc;
|
|
|
8005 }
|
|
|
8006 }
|
|
|
8007 }
|
|
|
8008
|
|
323
|
8009 /*
|
|
|
8010 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
8011 * lines in the .aff file.
|
|
|
8012 */
|
|
|
8013 static int
|
|
|
8014 similar_chars(slang, c1, c2)
|
|
|
8015 slang_T *slang;
|
|
|
8016 int c1;
|
|
|
8017 int c2;
|
|
|
8018 {
|
|
330
|
8019 int m1, m2;
|
|
|
8020 #ifdef FEAT_MBYTE
|
|
|
8021 char_u buf[MB_MAXBYTES];
|
|
|
8022 hashitem_T *hi;
|
|
|
8023
|
|
|
8024 if (c1 >= 256)
|
|
|
8025 {
|
|
|
8026 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
8027 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8028 if (HASHITEM_EMPTY(hi))
|
|
|
8029 m1 = 0;
|
|
|
8030 else
|
|
|
8031 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8032 }
|
|
|
8033 else
|
|
|
8034 #endif
|
|
|
8035 m1 = slang->sl_map_array[c1];
|
|
|
8036 if (m1 == 0)
|
|
323
|
8037 return FALSE;
|
|
330
|
8038
|
|
|
8039
|
|
|
8040 #ifdef FEAT_MBYTE
|
|
|
8041 if (c2 >= 256)
|
|
|
8042 {
|
|
|
8043 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
8044 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8045 if (HASHITEM_EMPTY(hi))
|
|
|
8046 m2 = 0;
|
|
|
8047 else
|
|
|
8048 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8049 }
|
|
|
8050 else
|
|
|
8051 #endif
|
|
|
8052 m2 = slang->sl_map_array[c2];
|
|
|
8053
|
|
|
8054 return m1 == m2;
|
|
323
|
8055 }
|
|
|
8056
|
|
|
8057 /*
|
|
|
8058 * Add a suggestion to the list of suggestions.
|
|
|
8059 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
8060 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
8061 * with spell_edit_score().
|
|
|
8062 */
|
|
|
8063 static void
|
|
351
|
8064 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
8065 suginfo_T *su;
|
|
344
|
8066 garray_T *gap;
|
|
323
|
8067 char_u *goodword;
|
|
346
|
8068 int badlen; /* length of bad word used */
|
|
324
|
8069 int score;
|
|
351
|
8070 int altscore;
|
|
344
|
8071 int had_bonus; /* value for st_had_bonus */
|
|
323
|
8072 {
|
|
|
8073 suggest_T *stp;
|
|
|
8074 int i;
|
|
346
|
8075 char_u *p = NULL;
|
|
|
8076 int c = 0;
|
|
323
|
8077
|
|
|
8078 /* Check that the word wasn't banned. */
|
|
|
8079 if (was_banned(su, goodword))
|
|
|
8080 return;
|
|
|
8081
|
|
346
|
8082 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
8083 * Remove the common part from "goodword". */
|
|
|
8084 i = badlen - su->su_badlen;
|
|
|
8085 if (i > 0)
|
|
|
8086 {
|
|
|
8087 /* This assumes there was no case folding or it didn't change the
|
|
|
8088 * length... */
|
|
|
8089 p = goodword + STRLEN(goodword) - i;
|
|
|
8090 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
8091 {
|
|
|
8092 badlen = su->su_badlen;
|
|
|
8093 c = *p;
|
|
|
8094 *p = NUL;
|
|
|
8095 }
|
|
|
8096 else
|
|
|
8097 p = NULL;
|
|
|
8098 }
|
|
372
|
8099 else if (i < 0)
|
|
|
8100 {
|
|
|
8101 /* When replacing part of the word check that we actually change
|
|
|
8102 * something. For "the the" a suggestion can be replacing the first
|
|
|
8103 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
8104 if (badlen == (int)STRLEN(goodword)
|
|
372
|
8105 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
8106 return;
|
|
|
8107 }
|
|
|
8108
|
|
346
|
8109
|
|
323
|
8110 if (score <= su->su_maxscore)
|
|
|
8111 {
|
|
366
|
8112 /* Check if the word is already there. Also check the length that is
|
|
|
8113 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
8114 * "thes" -> "these". */
|
|
344
|
8115 stp = &SUG(*gap, 0);
|
|
|
8116 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
8117 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
8118 && stp[i].st_orglen == badlen)
|
|
323
|
8119 {
|
|
|
8120 /* Found it. Remember the lowest score. */
|
|
|
8121 if (stp[i].st_score > score)
|
|
324
|
8122 {
|
|
323
|
8123 stp[i].st_score = score;
|
|
376
|
8124 stp[i].st_altscore = altscore;
|
|
324
|
8125 stp[i].st_had_bonus = had_bonus;
|
|
|
8126 }
|
|
323
|
8127 break;
|
|
|
8128 }
|
|
|
8129
|
|
344
|
8130 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
8131 {
|
|
|
8132 /* Add a suggestion. */
|
|
344
|
8133 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
8134 stp->st_word = vim_strsave(goodword);
|
|
|
8135 if (stp->st_word != NULL)
|
|
|
8136 {
|
|
|
8137 stp->st_score = score;
|
|
351
|
8138 stp->st_altscore = altscore;
|
|
324
|
8139 stp->st_had_bonus = had_bonus;
|
|
346
|
8140 stp->st_orglen = badlen;
|
|
344
|
8141 ++gap->ga_len;
|
|
323
|
8142
|
|
|
8143 /* If we have too many suggestions now, sort the list and keep
|
|
|
8144 * the best suggestions. */
|
|
344
|
8145 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
8146 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
8147 SUG_CLEAN_COUNT(su));
|
|
323
|
8148 }
|
|
|
8149 }
|
|
|
8150 }
|
|
346
|
8151
|
|
|
8152 if (p != NULL)
|
|
|
8153 *p = c; /* restore "goodword" */
|
|
323
|
8154 }
|
|
|
8155
|
|
|
8156 /*
|
|
|
8157 * Add a word to be banned.
|
|
|
8158 */
|
|
|
8159 static void
|
|
|
8160 add_banned(su, word)
|
|
|
8161 suginfo_T *su;
|
|
|
8162 char_u *word;
|
|
|
8163 {
|
|
|
8164 char_u *s = vim_strsave(word);
|
|
|
8165 hash_T hash;
|
|
|
8166 hashitem_T *hi;
|
|
|
8167
|
|
|
8168 if (s != NULL)
|
|
|
8169 {
|
|
|
8170 hash = hash_hash(s);
|
|
|
8171 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
8172 if (HASHITEM_EMPTY(hi))
|
|
|
8173 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
8174 else
|
|
|
8175 vim_free(s);
|
|
323
|
8176 }
|
|
|
8177 }
|
|
|
8178
|
|
|
8179 /*
|
|
|
8180 * Return TRUE if a word appears in the list of banned words.
|
|
|
8181 */
|
|
|
8182 static int
|
|
|
8183 was_banned(su, word)
|
|
|
8184 suginfo_T *su;
|
|
|
8185 char_u *word;
|
|
|
8186 {
|
|
324
|
8187 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
8188
|
|
|
8189 return !HASHITEM_EMPTY(hi);
|
|
323
|
8190 }
|
|
|
8191
|
|
|
8192 /*
|
|
|
8193 * Free the banned words in "su".
|
|
|
8194 */
|
|
|
8195 static void
|
|
|
8196 free_banned(su)
|
|
|
8197 suginfo_T *su;
|
|
|
8198 {
|
|
|
8199 int todo;
|
|
|
8200 hashitem_T *hi;
|
|
|
8201
|
|
|
8202 todo = su->su_banned.ht_used;
|
|
|
8203 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
8204 {
|
|
|
8205 if (!HASHITEM_EMPTY(hi))
|
|
|
8206 {
|
|
|
8207 vim_free(hi->hi_key);
|
|
|
8208 --todo;
|
|
|
8209 }
|
|
|
8210 }
|
|
|
8211 hash_clear(&su->su_banned);
|
|
|
8212 }
|
|
|
8213
|
|
324
|
8214 /*
|
|
|
8215 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
8216 * thus only done for the final results.
|
|
|
8217 */
|
|
|
8218 static void
|
|
|
8219 rescore_suggestions(su)
|
|
|
8220 suginfo_T *su;
|
|
|
8221 {
|
|
|
8222 langp_T *lp;
|
|
|
8223 suggest_T *stp;
|
|
|
8224 char_u sal_badword[MAXWLEN];
|
|
|
8225 int i;
|
|
|
8226
|
|
|
8227 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8228 lp->lp_slang != NULL; ++lp)
|
|
|
8229 {
|
|
|
8230 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8231 {
|
|
|
8232 /* soundfold the bad word */
|
|
375
|
8233 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
8234
|
|
|
8235 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8236 {
|
|
344
|
8237 stp = &SUG(su->su_ga, i);
|
|
324
|
8238 if (!stp->st_had_bonus)
|
|
|
8239 {
|
|
351
|
8240 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
8241 lp->lp_slang, sal_badword);
|
|
|
8242 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8243 stp->st_altscore = SCORE_BIG;
|
|
|
8244 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
8245 }
|
|
|
8246 }
|
|
|
8247 break;
|
|
|
8248 }
|
|
|
8249 }
|
|
|
8250 }
|
|
|
8251
|
|
323
|
8252 static int
|
|
|
8253 #ifdef __BORLANDC__
|
|
|
8254 _RTLENTRYF
|
|
|
8255 #endif
|
|
|
8256 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
8257
|
|
|
8258 /*
|
|
|
8259 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
8260 */
|
|
|
8261 static int
|
|
|
8262 #ifdef __BORLANDC__
|
|
|
8263 _RTLENTRYF
|
|
|
8264 #endif
|
|
|
8265 sug_compare(s1, s2)
|
|
|
8266 const void *s1;
|
|
|
8267 const void *s2;
|
|
|
8268 {
|
|
|
8269 suggest_T *p1 = (suggest_T *)s1;
|
|
|
8270 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
8271 int n = p1->st_score - p2->st_score;
|
|
|
8272
|
|
|
8273 if (n == 0)
|
|
|
8274 return p1->st_altscore - p2->st_altscore;
|
|
|
8275 return n;
|
|
323
|
8276 }
|
|
|
8277
|
|
|
8278 /*
|
|
|
8279 * Cleanup the suggestions:
|
|
|
8280 * - Sort on score.
|
|
|
8281 * - Remove words that won't be displayed.
|
|
344
|
8282 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
8283 */
|
|
344
|
8284 static int
|
|
|
8285 cleanup_suggestions(gap, maxscore, keep)
|
|
|
8286 garray_T *gap;
|
|
|
8287 int maxscore;
|
|
324
|
8288 int keep; /* nr of suggestions to keep */
|
|
323
|
8289 {
|
|
344
|
8290 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
8291 int i;
|
|
|
8292
|
|
|
8293 /* Sort the list. */
|
|
344
|
8294 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
8295
|
|
|
8296 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
8297 if (gap->ga_len > keep)
|
|
323
|
8298 {
|
|
344
|
8299 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
8300 vim_free(stp[i].st_word);
|
|
344
|
8301 gap->ga_len = keep;
|
|
|
8302 return stp[keep - 1].st_score;
|
|
323
|
8303 }
|
|
344
|
8304 return maxscore;
|
|
323
|
8305 }
|
|
|
8306
|
|
372
|
8307 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
8308 /*
|
|
|
8309 * Soundfold a string, for soundfold().
|
|
|
8310 * Result is in allocated memory, NULL for an error.
|
|
|
8311 */
|
|
|
8312 char_u *
|
|
|
8313 eval_soundfold(word)
|
|
|
8314 char_u *word;
|
|
|
8315 {
|
|
|
8316 langp_T *lp;
|
|
|
8317 char_u sound[MAXWLEN];
|
|
|
8318
|
|
|
8319 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
8320 /* Use the sound-folding of the first language that supports it. */
|
|
|
8321 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8322 lp->lp_slang != NULL; ++lp)
|
|
|
8323 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8324 {
|
|
|
8325 /* soundfold the word */
|
|
375
|
8326 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
8327 return vim_strsave(sound);
|
|
|
8328 }
|
|
|
8329
|
|
|
8330 /* No language with sound folding, return word as-is. */
|
|
|
8331 return vim_strsave(word);
|
|
|
8332 }
|
|
|
8333 #endif
|
|
|
8334
|
|
323
|
8335 /*
|
|
|
8336 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
8337 */
|
|
|
8338 static void
|
|
375
|
8339 spell_soundfold(slang, inword, folded, res)
|
|
|
8340 slang_T *slang;
|
|
|
8341 char_u *inword;
|
|
|
8342 int folded; /* "inword" is already case-folded */
|
|
|
8343 char_u *res;
|
|
|
8344 {
|
|
|
8345 char_u fword[MAXWLEN];
|
|
|
8346 char_u *word;
|
|
|
8347
|
|
|
8348 if (slang->sl_sofo)
|
|
|
8349 /* SOFOFROM and SOFOTO used */
|
|
|
8350 spell_soundfold_sofo(slang, inword, res);
|
|
|
8351 else
|
|
|
8352 {
|
|
|
8353 /* SAL items used. Requires the word to be case-folded. */
|
|
|
8354 if (folded)
|
|
|
8355 word = inword;
|
|
|
8356 else
|
|
|
8357 {
|
|
|
8358 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
8359 word = fword;
|
|
|
8360 }
|
|
|
8361
|
|
|
8362 #ifdef FEAT_MBYTE
|
|
|
8363 if (has_mbyte)
|
|
|
8364 spell_soundfold_wsal(slang, word, res);
|
|
|
8365 else
|
|
|
8366 #endif
|
|
|
8367 spell_soundfold_sal(slang, word, res);
|
|
|
8368 }
|
|
|
8369 }
|
|
|
8370
|
|
|
8371 /*
|
|
|
8372 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
8373 * SOFOTO lines.
|
|
|
8374 */
|
|
|
8375 static void
|
|
|
8376 spell_soundfold_sofo(slang, inword, res)
|
|
|
8377 slang_T *slang;
|
|
|
8378 char_u *inword;
|
|
|
8379 char_u *res;
|
|
|
8380 {
|
|
|
8381 char_u *s;
|
|
|
8382 int ri = 0;
|
|
|
8383 int c;
|
|
|
8384
|
|
|
8385 #ifdef FEAT_MBYTE
|
|
|
8386 if (has_mbyte)
|
|
|
8387 {
|
|
|
8388 int prevc = 0;
|
|
|
8389 int *ip;
|
|
|
8390
|
|
|
8391 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
8392 * 255, sl_sal the rest. */
|
|
|
8393 for (s = inword; *s != NUL; )
|
|
|
8394 {
|
|
|
8395 c = mb_ptr2char_adv(&s);
|
|
|
8396 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
8397 c = ' ';
|
|
|
8398 else if (c < 256)
|
|
|
8399 c = slang->sl_sal_first[c];
|
|
|
8400 else
|
|
|
8401 {
|
|
|
8402 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
8403 if (ip == NULL) /* empty list, can't match */
|
|
|
8404 c = NUL;
|
|
|
8405 else
|
|
|
8406 for (;;) /* find "c" in the list */
|
|
|
8407 {
|
|
|
8408 if (*ip == 0) /* not found */
|
|
|
8409 {
|
|
|
8410 c = NUL;
|
|
|
8411 break;
|
|
|
8412 }
|
|
|
8413 if (*ip == c) /* match! */
|
|
|
8414 {
|
|
|
8415 c = ip[1];
|
|
|
8416 break;
|
|
|
8417 }
|
|
|
8418 ip += 2;
|
|
|
8419 }
|
|
|
8420 }
|
|
|
8421
|
|
|
8422 if (c != NUL && c != prevc)
|
|
|
8423 {
|
|
|
8424 ri += mb_char2bytes(c, res + ri);
|
|
|
8425 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
8426 break;
|
|
|
8427 prevc = c;
|
|
|
8428 }
|
|
|
8429 }
|
|
|
8430 }
|
|
|
8431 else
|
|
|
8432 #endif
|
|
|
8433 {
|
|
|
8434 /* The sl_sal_first[] table contains the translation. */
|
|
|
8435 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
8436 {
|
|
|
8437 if (vim_iswhite(c))
|
|
|
8438 c = ' ';
|
|
|
8439 else
|
|
|
8440 c = slang->sl_sal_first[c];
|
|
|
8441 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
8442 res[ri++] = c;
|
|
|
8443 }
|
|
|
8444 }
|
|
|
8445
|
|
|
8446 res[ri] = NUL;
|
|
|
8447 }
|
|
|
8448
|
|
|
8449 static void
|
|
|
8450 spell_soundfold_sal(slang, inword, res)
|
|
323
|
8451 slang_T *slang;
|
|
|
8452 char_u *inword;
|
|
|
8453 char_u *res;
|
|
|
8454 {
|
|
344
|
8455 salitem_T *smp;
|
|
323
|
8456 char_u word[MAXWLEN];
|
|
375
|
8457 char_u *s = inword;
|
|
323
|
8458 char_u *t;
|
|
344
|
8459 char_u *pf;
|
|
323
|
8460 int i, j, z;
|
|
344
|
8461 int reslen;
|
|
323
|
8462 int n, k = 0;
|
|
|
8463 int z0;
|
|
|
8464 int k0;
|
|
|
8465 int n0;
|
|
|
8466 int c;
|
|
|
8467 int pri;
|
|
|
8468 int p0 = -333;
|
|
|
8469 int c0;
|
|
|
8470
|
|
324
|
8471 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
8472 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
8473 if (slang->sl_rem_accents)
|
|
|
8474 {
|
|
|
8475 t = word;
|
|
375
|
8476 while (*s != NUL)
|
|
323
|
8477 {
|
|
324
|
8478 if (vim_iswhite(*s))
|
|
344
|
8479 {
|
|
|
8480 *t++ = ' ';
|
|
|
8481 s = skipwhite(s);
|
|
|
8482 }
|
|
323
|
8483 else
|
|
|
8484 {
|
|
376
|
8485 if (spell_iswordp_nmw(s))
|
|
323
|
8486 *t++ = *s;
|
|
|
8487 ++s;
|
|
|
8488 }
|
|
|
8489 }
|
|
|
8490 *t = NUL;
|
|
|
8491 }
|
|
|
8492 else
|
|
375
|
8493 STRCPY(word, s);
|
|
323
|
8494
|
|
344
|
8495 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
8496
|
|
|
8497 /*
|
|
|
8498 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
8499 * Changed to keep spaces.
|
|
323
|
8500 */
|
|
344
|
8501 i = reslen = z = 0;
|
|
323
|
8502 while ((c = word[i]) != NUL)
|
|
|
8503 {
|
|
344
|
8504 /* Start with the first rule that has the character in the word. */
|
|
323
|
8505 n = slang->sl_sal_first[c];
|
|
|
8506 z0 = 0;
|
|
|
8507
|
|
|
8508 if (n >= 0)
|
|
|
8509 {
|
|
|
8510 /* check all rules for the same letter */
|
|
344
|
8511 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
8512 {
|
|
344
|
8513 /* Quickly skip entries that don't match the word. Most
|
|
|
8514 * entries are less then three chars, optimize for that. */
|
|
|
8515 k = smp[n].sm_leadlen;
|
|
|
8516 if (k > 1)
|
|
323
|
8517 {
|
|
344
|
8518 if (word[i + 1] != s[1])
|
|
|
8519 continue;
|
|
|
8520 if (k > 2)
|
|
|
8521 {
|
|
|
8522 for (j = 2; j < k; ++j)
|
|
|
8523 if (word[i + j] != s[j])
|
|
|
8524 break;
|
|
|
8525 if (j < k)
|
|
|
8526 continue;
|
|
|
8527 }
|
|
323
|
8528 }
|
|
|
8529
|
|
375
|
8530 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
8531 {
|
|
375
|
8532 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
8533 while (*pf != NUL && *pf != word[i + k])
|
|
|
8534 ++pf;
|
|
|
8535 if (*pf == NUL)
|
|
|
8536 continue;
|
|
|
8537 ++k;
|
|
323
|
8538 }
|
|
344
|
8539 s = smp[n].sm_rules;
|
|
|
8540 pri = 5; /* default priority */
|
|
323
|
8541
|
|
|
8542 p0 = *s;
|
|
|
8543 k0 = k;
|
|
|
8544 while (*s == '-' && k > 1)
|
|
|
8545 {
|
|
|
8546 k--;
|
|
|
8547 s++;
|
|
|
8548 }
|
|
|
8549 if (*s == '<')
|
|
|
8550 s++;
|
|
344
|
8551 if (VIM_ISDIGIT(*s))
|
|
323
|
8552 {
|
|
|
8553 /* determine priority */
|
|
|
8554 pri = *s - '0';
|
|
|
8555 s++;
|
|
|
8556 }
|
|
|
8557 if (*s == '^' && *(s + 1) == '^')
|
|
|
8558 s++;
|
|
|
8559
|
|
|
8560 if (*s == NUL
|
|
|
8561 || (*s == '^'
|
|
324
|
8562 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
8563 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
8564 && (*(s + 1) != '$'
|
|
376
|
8565 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
8566 || (*s == '$' && i > 0
|
|
376
|
8567 && spell_iswordp(word + i - 1, curbuf)
|
|
|
8568 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
8569 {
|
|
|
8570 /* search for followup rules, if: */
|
|
|
8571 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
8572 c0 = word[i + k - 1];
|
|
|
8573 n0 = slang->sl_sal_first[c0];
|
|
|
8574
|
|
|
8575 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
8576 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
8577 {
|
|
|
8578 /* test follow-up rule for "word[i + k]" */
|
|
344
|
8579 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
8580 {
|
|
344
|
8581 /* Quickly skip entries that don't match the word.
|
|
|
8582 * */
|
|
|
8583 k0 = smp[n0].sm_leadlen;
|
|
|
8584 if (k0 > 1)
|
|
323
|
8585 {
|
|
344
|
8586 if (word[i + k] != s[1])
|
|
|
8587 continue;
|
|
|
8588 if (k0 > 2)
|
|
|
8589 {
|
|
|
8590 pf = word + i + k + 1;
|
|
|
8591 for (j = 2; j < k0; ++j)
|
|
|
8592 if (*pf++ != s[j])
|
|
|
8593 break;
|
|
|
8594 if (j < k0)
|
|
|
8595 continue;
|
|
|
8596 }
|
|
323
|
8597 }
|
|
344
|
8598 k0 += k - 1;
|
|
|
8599
|
|
375
|
8600 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
8601 {
|
|
344
|
8602 /* Check for match with one of the chars in
|
|
375
|
8603 * "sm_oneof". */
|
|
344
|
8604 while (*pf != NUL && *pf != word[i + k0])
|
|
|
8605 ++pf;
|
|
|
8606 if (*pf == NUL)
|
|
|
8607 continue;
|
|
|
8608 ++k0;
|
|
323
|
8609 }
|
|
344
|
8610
|
|
|
8611 p0 = 5;
|
|
|
8612 s = smp[n0].sm_rules;
|
|
323
|
8613 while (*s == '-')
|
|
|
8614 {
|
|
344
|
8615 /* "k0" gets NOT reduced because
|
|
|
8616 * "if (k0 == k)" */
|
|
323
|
8617 s++;
|
|
|
8618 }
|
|
|
8619 if (*s == '<')
|
|
|
8620 s++;
|
|
344
|
8621 if (VIM_ISDIGIT(*s))
|
|
323
|
8622 {
|
|
|
8623 p0 = *s - '0';
|
|
|
8624 s++;
|
|
|
8625 }
|
|
|
8626
|
|
|
8627 if (*s == NUL
|
|
|
8628 /* *s == '^' cuts */
|
|
|
8629 || (*s == '$'
|
|
376
|
8630 && !spell_iswordp(word + i + k0,
|
|
|
8631 curbuf)))
|
|
323
|
8632 {
|
|
|
8633 if (k0 == k)
|
|
|
8634 /* this is just a piece of the string */
|
|
|
8635 continue;
|
|
|
8636
|
|
|
8637 if (p0 < pri)
|
|
|
8638 /* priority too low */
|
|
|
8639 continue;
|
|
|
8640 /* rule fits; stop search */
|
|
|
8641 break;
|
|
|
8642 }
|
|
|
8643 }
|
|
|
8644
|
|
344
|
8645 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
8646 continue;
|
|
|
8647 }
|
|
|
8648
|
|
|
8649 /* replace string */
|
|
344
|
8650 s = smp[n].sm_to;
|
|
|
8651 pf = smp[n].sm_rules;
|
|
|
8652 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
8653 if (p0 == 1 && z == 0)
|
|
|
8654 {
|
|
|
8655 /* rule with '<' is used */
|
|
344
|
8656 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
8657 || res[reslen - 1] == *s))
|
|
|
8658 reslen--;
|
|
323
|
8659 z0 = 1;
|
|
|
8660 z = 1;
|
|
|
8661 k0 = 0;
|
|
372
|
8662 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
8663 {
|
|
|
8664 word[i + k0] = *s;
|
|
|
8665 k0++;
|
|
|
8666 s++;
|
|
|
8667 }
|
|
|
8668 if (k > k0)
|
|
|
8669 mch_memmove(word + i + k0, word + i + k,
|
|
|
8670 STRLEN(word + i + k) + 1);
|
|
|
8671
|
|
|
8672 /* new "actual letter" */
|
|
|
8673 c = word[i];
|
|
|
8674 }
|
|
|
8675 else
|
|
|
8676 {
|
|
|
8677 /* no '<' rule used */
|
|
|
8678 i += k - 1;
|
|
|
8679 z = 0;
|
|
344
|
8680 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
8681 {
|
|
344
|
8682 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
8683 res[reslen++] = *s;
|
|
323
|
8684 s++;
|
|
|
8685 }
|
|
|
8686 /* new "actual letter" */
|
|
|
8687 c = *s;
|
|
344
|
8688 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
8689 {
|
|
|
8690 if (c != NUL)
|
|
372
|
8691 res[reslen++] = c;
|
|
323
|
8692 mch_memmove(word, word + i + 1,
|
|
|
8693 STRLEN(word + i + 1) + 1);
|
|
|
8694 i = 0;
|
|
|
8695 z0 = 1;
|
|
|
8696 }
|
|
|
8697 }
|
|
|
8698 break;
|
|
|
8699 }
|
|
|
8700 }
|
|
|
8701 }
|
|
324
|
8702 else if (vim_iswhite(c))
|
|
|
8703 {
|
|
|
8704 c = ' ';
|
|
|
8705 k = 1;
|
|
|
8706 }
|
|
323
|
8707
|
|
|
8708 if (z0 == 0)
|
|
|
8709 {
|
|
344
|
8710 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
8711 && (!slang->sl_collapse || reslen == 0
|
|
|
8712 || res[reslen - 1] != c))
|
|
323
|
8713 /* condense only double letters */
|
|
372
|
8714 res[reslen++] = c;
|
|
323
|
8715
|
|
|
8716 i++;
|
|
|
8717 z = 0;
|
|
|
8718 k = 0;
|
|
|
8719 }
|
|
|
8720 }
|
|
|
8721
|
|
344
|
8722 res[reslen] = NUL;
|
|
323
|
8723 }
|
|
|
8724
|
|
372
|
8725 #ifdef FEAT_MBYTE
|
|
|
8726 /*
|
|
|
8727 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
8728 * Multi-byte version of spell_soundfold().
|
|
|
8729 */
|
|
|
8730 static void
|
|
375
|
8731 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
8732 slang_T *slang;
|
|
|
8733 char_u *inword;
|
|
|
8734 char_u *res;
|
|
|
8735 {
|
|
375
|
8736 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
8737 int word[MAXWLEN];
|
|
|
8738 int wres[MAXWLEN];
|
|
|
8739 int l;
|
|
|
8740 char_u *s;
|
|
|
8741 int *ws;
|
|
|
8742 char_u *t;
|
|
|
8743 int *pf;
|
|
|
8744 int i, j, z;
|
|
|
8745 int reslen;
|
|
|
8746 int n, k = 0;
|
|
|
8747 int z0;
|
|
|
8748 int k0;
|
|
|
8749 int n0;
|
|
|
8750 int c;
|
|
|
8751 int pri;
|
|
|
8752 int p0 = -333;
|
|
|
8753 int c0;
|
|
|
8754 int did_white = FALSE;
|
|
|
8755
|
|
|
8756 /*
|
|
|
8757 * Convert the multi-byte string to a wide-character string.
|
|
|
8758 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
8759 * But keep white space.
|
|
|
8760 */
|
|
|
8761 n = 0;
|
|
|
8762 for (s = inword; *s != NUL; )
|
|
|
8763 {
|
|
|
8764 t = s;
|
|
|
8765 c = mb_ptr2char_adv(&s);
|
|
|
8766 if (slang->sl_rem_accents)
|
|
|
8767 {
|
|
|
8768 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
8769 {
|
|
|
8770 if (did_white)
|
|
|
8771 continue;
|
|
|
8772 c = ' ';
|
|
|
8773 did_white = TRUE;
|
|
|
8774 }
|
|
|
8775 else
|
|
|
8776 {
|
|
|
8777 did_white = FALSE;
|
|
376
|
8778 if (!spell_iswordp_nmw(t))
|
|
372
|
8779 continue;
|
|
|
8780 }
|
|
|
8781 }
|
|
|
8782 word[n++] = c;
|
|
|
8783 }
|
|
|
8784 word[n] = NUL;
|
|
|
8785
|
|
|
8786 /*
|
|
|
8787 * This comes from Aspell phonet.cpp.
|
|
|
8788 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
8789 * Changed to keep spaces.
|
|
|
8790 */
|
|
|
8791 i = reslen = z = 0;
|
|
|
8792 while ((c = word[i]) != NUL)
|
|
|
8793 {
|
|
|
8794 /* Start with the first rule that has the character in the word. */
|
|
|
8795 n = slang->sl_sal_first[c & 0xff];
|
|
|
8796 z0 = 0;
|
|
|
8797
|
|
|
8798 if (n >= 0)
|
|
|
8799 {
|
|
375
|
8800 /* check all rules for the same index byte */
|
|
372
|
8801 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
8802 {
|
|
|
8803 /* Quickly skip entries that don't match the word. Most
|
|
|
8804 * entries are less then three chars, optimize for that. */
|
|
375
|
8805 if (c != ws[0])
|
|
|
8806 continue;
|
|
372
|
8807 k = smp[n].sm_leadlen;
|
|
|
8808 if (k > 1)
|
|
|
8809 {
|
|
|
8810 if (word[i + 1] != ws[1])
|
|
|
8811 continue;
|
|
|
8812 if (k > 2)
|
|
|
8813 {
|
|
|
8814 for (j = 2; j < k; ++j)
|
|
|
8815 if (word[i + j] != ws[j])
|
|
|
8816 break;
|
|
|
8817 if (j < k)
|
|
|
8818 continue;
|
|
|
8819 }
|
|
|
8820 }
|
|
|
8821
|
|
375
|
8822 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
8823 {
|
|
375
|
8824 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
8825 while (*pf != NUL && *pf != word[i + k])
|
|
|
8826 ++pf;
|
|
|
8827 if (*pf == NUL)
|
|
|
8828 continue;
|
|
|
8829 ++k;
|
|
|
8830 }
|
|
|
8831 s = smp[n].sm_rules;
|
|
|
8832 pri = 5; /* default priority */
|
|
|
8833
|
|
|
8834 p0 = *s;
|
|
|
8835 k0 = k;
|
|
|
8836 while (*s == '-' && k > 1)
|
|
|
8837 {
|
|
|
8838 k--;
|
|
|
8839 s++;
|
|
|
8840 }
|
|
|
8841 if (*s == '<')
|
|
|
8842 s++;
|
|
|
8843 if (VIM_ISDIGIT(*s))
|
|
|
8844 {
|
|
|
8845 /* determine priority */
|
|
|
8846 pri = *s - '0';
|
|
|
8847 s++;
|
|
|
8848 }
|
|
|
8849 if (*s == '^' && *(s + 1) == '^')
|
|
|
8850 s++;
|
|
|
8851
|
|
|
8852 if (*s == NUL
|
|
|
8853 || (*s == '^'
|
|
|
8854 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
8855 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
8856 && (*(s + 1) != '$'
|
|
376
|
8857 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
8858 || (*s == '$' && i > 0
|
|
376
|
8859 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
8860 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
8861 {
|
|
|
8862 /* search for followup rules, if: */
|
|
|
8863 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
8864 c0 = word[i + k - 1];
|
|
|
8865 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
8866
|
|
|
8867 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
8868 && p0 != '-' && word[i + k] != NUL)
|
|
|
8869 {
|
|
375
|
8870 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
8871 * all entries with the same index byte. */
|
|
372
|
8872 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
8873 == (c0 & 0xff); ++n0)
|
|
|
8874 {
|
|
|
8875 /* Quickly skip entries that don't match the word.
|
|
375
|
8876 */
|
|
|
8877 if (c0 != ws[0])
|
|
|
8878 continue;
|
|
372
|
8879 k0 = smp[n0].sm_leadlen;
|
|
|
8880 if (k0 > 1)
|
|
|
8881 {
|
|
|
8882 if (word[i + k] != ws[1])
|
|
|
8883 continue;
|
|
|
8884 if (k0 > 2)
|
|
|
8885 {
|
|
|
8886 pf = word + i + k + 1;
|
|
|
8887 for (j = 2; j < k0; ++j)
|
|
|
8888 if (*pf++ != ws[j])
|
|
|
8889 break;
|
|
|
8890 if (j < k0)
|
|
|
8891 continue;
|
|
|
8892 }
|
|
|
8893 }
|
|
|
8894 k0 += k - 1;
|
|
|
8895
|
|
375
|
8896 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
8897 {
|
|
|
8898 /* Check for match with one of the chars in
|
|
375
|
8899 * "sm_oneof". */
|
|
372
|
8900 while (*pf != NUL && *pf != word[i + k0])
|
|
|
8901 ++pf;
|
|
|
8902 if (*pf == NUL)
|
|
|
8903 continue;
|
|
|
8904 ++k0;
|
|
|
8905 }
|
|
|
8906
|
|
|
8907 p0 = 5;
|
|
|
8908 s = smp[n0].sm_rules;
|
|
|
8909 while (*s == '-')
|
|
|
8910 {
|
|
|
8911 /* "k0" gets NOT reduced because
|
|
|
8912 * "if (k0 == k)" */
|
|
|
8913 s++;
|
|
|
8914 }
|
|
|
8915 if (*s == '<')
|
|
|
8916 s++;
|
|
|
8917 if (VIM_ISDIGIT(*s))
|
|
|
8918 {
|
|
|
8919 p0 = *s - '0';
|
|
|
8920 s++;
|
|
|
8921 }
|
|
|
8922
|
|
|
8923 if (*s == NUL
|
|
|
8924 /* *s == '^' cuts */
|
|
|
8925 || (*s == '$'
|
|
376
|
8926 && !spell_iswordp_w(word + i + k0,
|
|
|
8927 curbuf)))
|
|
372
|
8928 {
|
|
|
8929 if (k0 == k)
|
|
|
8930 /* this is just a piece of the string */
|
|
|
8931 continue;
|
|
|
8932
|
|
|
8933 if (p0 < pri)
|
|
|
8934 /* priority too low */
|
|
|
8935 continue;
|
|
|
8936 /* rule fits; stop search */
|
|
|
8937 break;
|
|
|
8938 }
|
|
|
8939 }
|
|
|
8940
|
|
|
8941 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
8942 == (c0 & 0xff))
|
|
|
8943 continue;
|
|
|
8944 }
|
|
|
8945
|
|
|
8946 /* replace string */
|
|
|
8947 ws = smp[n].sm_to_w;
|
|
|
8948 s = smp[n].sm_rules;
|
|
|
8949 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
8950 if (p0 == 1 && z == 0)
|
|
|
8951 {
|
|
|
8952 /* rule with '<' is used */
|
|
|
8953 if (reslen > 0 && *ws != NUL && (wres[reslen - 1] == c
|
|
|
8954 || wres[reslen - 1] == *ws))
|
|
|
8955 reslen--;
|
|
|
8956 z0 = 1;
|
|
|
8957 z = 1;
|
|
|
8958 k0 = 0;
|
|
|
8959 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
8960 {
|
|
|
8961 word[i + k0] = *ws;
|
|
|
8962 k0++;
|
|
|
8963 ws++;
|
|
|
8964 }
|
|
|
8965 if (k > k0)
|
|
|
8966 mch_memmove(word + i + k0, word + i + k,
|
|
|
8967 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
8968
|
|
|
8969 /* new "actual letter" */
|
|
|
8970 c = word[i];
|
|
|
8971 }
|
|
|
8972 else
|
|
|
8973 {
|
|
|
8974 /* no '<' rule used */
|
|
|
8975 i += k - 1;
|
|
|
8976 z = 0;
|
|
|
8977 while (*ws != NUL && ws[1] != NUL && reslen < MAXWLEN)
|
|
|
8978 {
|
|
|
8979 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
8980 wres[reslen++] = *ws;
|
|
|
8981 ws++;
|
|
|
8982 }
|
|
|
8983 /* new "actual letter" */
|
|
|
8984 c = *ws;
|
|
|
8985 if (strstr((char *)s, "^^") != NULL)
|
|
|
8986 {
|
|
|
8987 if (c != NUL)
|
|
|
8988 wres[reslen++] = c;
|
|
|
8989 mch_memmove(word, word + i + 1,
|
|
|
8990 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
8991 i = 0;
|
|
|
8992 z0 = 1;
|
|
|
8993 }
|
|
|
8994 }
|
|
|
8995 break;
|
|
|
8996 }
|
|
|
8997 }
|
|
|
8998 }
|
|
|
8999 else if (vim_iswhite(c))
|
|
|
9000 {
|
|
|
9001 c = ' ';
|
|
|
9002 k = 1;
|
|
|
9003 }
|
|
|
9004
|
|
|
9005 if (z0 == 0)
|
|
|
9006 {
|
|
|
9007 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9008 && (!slang->sl_collapse || reslen == 0
|
|
|
9009 || wres[reslen - 1] != c))
|
|
|
9010 /* condense only double letters */
|
|
|
9011 wres[reslen++] = c;
|
|
|
9012
|
|
|
9013 i++;
|
|
|
9014 z = 0;
|
|
|
9015 k = 0;
|
|
|
9016 }
|
|
|
9017 }
|
|
|
9018
|
|
|
9019 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
9020 l = 0;
|
|
|
9021 for (n = 0; n < reslen; ++n)
|
|
|
9022 {
|
|
|
9023 l += mb_char2bytes(wres[n], res + l);
|
|
|
9024 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
9025 break;
|
|
|
9026 }
|
|
|
9027 res[l] = NUL;
|
|
|
9028 }
|
|
|
9029 #endif
|
|
|
9030
|
|
324
|
9031 /*
|
|
344
|
9032 * Compute a score for two sound-a-like words.
|
|
|
9033 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
9034 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
9035 * avoiding checks that will not be possible.
|
|
|
9036 */
|
|
|
9037 static int
|
|
351
|
9038 soundalike_score(goodstart, badstart)
|
|
|
9039 char_u *goodstart; /* sound-folded good word */
|
|
|
9040 char_u *badstart; /* sound-folded bad word */
|
|
344
|
9041 {
|
|
351
|
9042 char_u *goodsound = goodstart;
|
|
|
9043 char_u *badsound = badstart;
|
|
|
9044 int goodlen;
|
|
|
9045 int badlen;
|
|
344
|
9046 int n;
|
|
|
9047 char_u *pl, *ps;
|
|
|
9048 char_u *pl2, *ps2;
|
|
351
|
9049 int score = 0;
|
|
|
9050
|
|
|
9051 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
9052 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
9053 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
9054 {
|
|
|
9055 score = SCORE_DEL / 2;
|
|
|
9056 if (*badsound == '*')
|
|
|
9057 ++badsound;
|
|
|
9058 else
|
|
|
9059 ++goodsound;
|
|
|
9060 }
|
|
|
9061
|
|
|
9062 goodlen = STRLEN(goodsound);
|
|
|
9063 badlen = STRLEN(badsound);
|
|
344
|
9064
|
|
|
9065 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
9066 * changes. */
|
|
|
9067 n = goodlen - badlen;
|
|
|
9068 if (n < -2 || n > 2)
|
|
|
9069 return SCORE_MAXMAX;
|
|
|
9070
|
|
|
9071 if (n > 0)
|
|
|
9072 {
|
|
351
|
9073 pl = goodsound; /* goodsound is longest */
|
|
344
|
9074 ps = badsound;
|
|
|
9075 }
|
|
|
9076 else
|
|
|
9077 {
|
|
351
|
9078 pl = badsound; /* badsound is longest */
|
|
344
|
9079 ps = goodsound;
|
|
|
9080 }
|
|
|
9081
|
|
|
9082 /* Skip over the identical part. */
|
|
|
9083 while (*pl == *ps && *pl != NUL)
|
|
|
9084 {
|
|
|
9085 ++pl;
|
|
|
9086 ++ps;
|
|
|
9087 }
|
|
|
9088
|
|
|
9089 switch (n)
|
|
|
9090 {
|
|
|
9091 case -2:
|
|
|
9092 case 2:
|
|
|
9093 /*
|
|
|
9094 * Must delete two characters from "pl".
|
|
|
9095 */
|
|
|
9096 ++pl; /* first delete */
|
|
|
9097 while (*pl == *ps)
|
|
|
9098 {
|
|
|
9099 ++pl;
|
|
|
9100 ++ps;
|
|
|
9101 }
|
|
|
9102 /* strings must be equal after second delete */
|
|
|
9103 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
9104 return score + SCORE_DEL * 2;
|
|
344
|
9105
|
|
|
9106 /* Failed to compare. */
|
|
|
9107 break;
|
|
|
9108
|
|
|
9109 case -1:
|
|
|
9110 case 1:
|
|
|
9111 /*
|
|
|
9112 * Minimal one delete from "pl" required.
|
|
|
9113 */
|
|
|
9114
|
|
|
9115 /* 1: delete */
|
|
|
9116 pl2 = pl + 1;
|
|
|
9117 ps2 = ps;
|
|
|
9118 while (*pl2 == *ps2)
|
|
|
9119 {
|
|
|
9120 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9121 return score + SCORE_DEL;
|
|
344
|
9122 ++pl2;
|
|
|
9123 ++ps2;
|
|
|
9124 }
|
|
|
9125
|
|
|
9126 /* 2: delete then swap, then rest must be equal */
|
|
|
9127 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9128 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9129 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
9130
|
|
|
9131 /* 3: delete then substitute, then the rest must be equal */
|
|
|
9132 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9133 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
9134
|
|
|
9135 /* 4: first swap then delete */
|
|
|
9136 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9137 {
|
|
|
9138 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9139 ps2 = ps + 2;
|
|
|
9140 while (*pl2 == *ps2)
|
|
|
9141 {
|
|
|
9142 ++pl2;
|
|
|
9143 ++ps2;
|
|
|
9144 }
|
|
|
9145 /* delete a char and then strings must be equal */
|
|
|
9146 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9147 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
9148 }
|
|
|
9149
|
|
|
9150 /* 5: first substitute then delete */
|
|
|
9151 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
9152 ps2 = ps + 1;
|
|
|
9153 while (*pl2 == *ps2)
|
|
|
9154 {
|
|
|
9155 ++pl2;
|
|
|
9156 ++ps2;
|
|
|
9157 }
|
|
|
9158 /* delete a char and then strings must be equal */
|
|
|
9159 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9160 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
9161
|
|
|
9162 /* Failed to compare. */
|
|
|
9163 break;
|
|
|
9164
|
|
|
9165 case 0:
|
|
|
9166 /*
|
|
|
9167 * Lenghts are equal, thus changes must result in same length: An
|
|
|
9168 * insert is only possible in combination with a delete.
|
|
|
9169 * 1: check if for identical strings
|
|
|
9170 */
|
|
|
9171 if (*pl == NUL)
|
|
351
|
9172 return score;
|
|
344
|
9173
|
|
|
9174 /* 2: swap */
|
|
|
9175 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9176 {
|
|
|
9177 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9178 ps2 = ps + 2;
|
|
|
9179 while (*pl2 == *ps2)
|
|
|
9180 {
|
|
|
9181 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9182 return score + SCORE_SWAP;
|
|
344
|
9183 ++pl2;
|
|
|
9184 ++ps2;
|
|
|
9185 }
|
|
|
9186 /* 3: swap and swap again */
|
|
|
9187 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9188 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9189 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
9190
|
|
|
9191 /* 4: swap and substitute */
|
|
|
9192 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9193 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
9194 }
|
|
|
9195
|
|
|
9196 /* 5: substitute */
|
|
|
9197 pl2 = pl + 1;
|
|
|
9198 ps2 = ps + 1;
|
|
|
9199 while (*pl2 == *ps2)
|
|
|
9200 {
|
|
|
9201 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9202 return score + SCORE_SUBST;
|
|
344
|
9203 ++pl2;
|
|
|
9204 ++ps2;
|
|
|
9205 }
|
|
|
9206
|
|
|
9207 /* 6: substitute and swap */
|
|
|
9208 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9209 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9210 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
9211
|
|
|
9212 /* 7: substitute and substitute */
|
|
|
9213 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9214 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
9215
|
|
|
9216 /* 8: insert then delete */
|
|
|
9217 pl2 = pl;
|
|
|
9218 ps2 = ps + 1;
|
|
|
9219 while (*pl2 == *ps2)
|
|
|
9220 {
|
|
|
9221 ++pl2;
|
|
|
9222 ++ps2;
|
|
|
9223 }
|
|
|
9224 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9225 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9226
|
|
|
9227 /* 9: delete then insert */
|
|
|
9228 pl2 = pl + 1;
|
|
|
9229 ps2 = ps;
|
|
|
9230 while (*pl2 == *ps2)
|
|
|
9231 {
|
|
|
9232 ++pl2;
|
|
|
9233 ++ps2;
|
|
|
9234 }
|
|
|
9235 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
9236 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9237
|
|
|
9238 /* Failed to compare. */
|
|
|
9239 break;
|
|
|
9240 }
|
|
|
9241
|
|
|
9242 return SCORE_MAXMAX;
|
|
|
9243 }
|
|
324
|
9244
|
|
323
|
9245 /*
|
|
|
9246 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
9247 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
9248 *
|
|
323
|
9249 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
9250 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
9251 * characters.
|
|
323
|
9252 */
|
|
|
9253 static int
|
|
|
9254 spell_edit_score(badword, goodword)
|
|
|
9255 char_u *badword;
|
|
|
9256 char_u *goodword;
|
|
|
9257 {
|
|
|
9258 int *cnt;
|
|
372
|
9259 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
9260 int j, i;
|
|
|
9261 int t;
|
|
|
9262 int bc, gc;
|
|
324
|
9263 int pbc, pgc;
|
|
|
9264 #ifdef FEAT_MBYTE
|
|
|
9265 char_u *p;
|
|
|
9266 int wbadword[MAXWLEN];
|
|
|
9267 int wgoodword[MAXWLEN];
|
|
|
9268
|
|
|
9269 if (has_mbyte)
|
|
|
9270 {
|
|
|
9271 /* Get the characters from the multi-byte strings and put them in an
|
|
|
9272 * int array for easy access. */
|
|
|
9273 for (p = badword, badlen = 0; *p != NUL; )
|
|
|
9274 wbadword[badlen++] = mb_ptr2char_adv(&p);
|
|
|
9275 ++badlen;
|
|
|
9276 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
|
9277 wgoodword[goodlen++] = mb_ptr2char_adv(&p);
|
|
|
9278 ++goodlen;
|
|
|
9279 }
|
|
|
9280 else
|
|
|
9281 #endif
|
|
|
9282 {
|
|
|
9283 badlen = STRLEN(badword) + 1;
|
|
|
9284 goodlen = STRLEN(goodword) + 1;
|
|
|
9285 }
|
|
323
|
9286
|
|
|
9287 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
9288 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
9289 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
9290 TRUE);
|
|
324
|
9291 if (cnt == NULL)
|
|
|
9292 return 0; /* out of memory */
|
|
323
|
9293
|
|
|
9294 CNT(0, 0) = 0;
|
|
|
9295 for (j = 1; j <= goodlen; ++j)
|
|
|
9296 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
9297
|
|
|
9298 for (i = 1; i <= badlen; ++i)
|
|
|
9299 {
|
|
|
9300 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
9301 for (j = 1; j <= goodlen; ++j)
|
|
|
9302 {
|
|
324
|
9303 #ifdef FEAT_MBYTE
|
|
|
9304 if (has_mbyte)
|
|
|
9305 {
|
|
|
9306 bc = wbadword[i - 1];
|
|
|
9307 gc = wgoodword[j - 1];
|
|
|
9308 }
|
|
|
9309 else
|
|
|
9310 #endif
|
|
|
9311 {
|
|
|
9312 bc = badword[i - 1];
|
|
|
9313 gc = goodword[j - 1];
|
|
|
9314 }
|
|
323
|
9315 if (bc == gc)
|
|
|
9316 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
9317 else
|
|
|
9318 {
|
|
|
9319 /* Use a better score when there is only a case difference. */
|
|
324
|
9320 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
9321 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
9322 else
|
|
|
9323 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
9324
|
|
324
|
9325 if (i > 1 && j > 1)
|
|
323
|
9326 {
|
|
324
|
9327 #ifdef FEAT_MBYTE
|
|
|
9328 if (has_mbyte)
|
|
|
9329 {
|
|
|
9330 pbc = wbadword[i - 2];
|
|
|
9331 pgc = wgoodword[j - 2];
|
|
|
9332 }
|
|
|
9333 else
|
|
|
9334 #endif
|
|
|
9335 {
|
|
|
9336 pbc = badword[i - 2];
|
|
|
9337 pgc = goodword[j - 2];
|
|
|
9338 }
|
|
|
9339 if (bc == pgc && pbc == gc)
|
|
|
9340 {
|
|
|
9341 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
9342 if (t < CNT(i, j))
|
|
|
9343 CNT(i, j) = t;
|
|
|
9344 }
|
|
323
|
9345 }
|
|
|
9346 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
9347 if (t < CNT(i, j))
|
|
|
9348 CNT(i, j) = t;
|
|
|
9349 t = SCORE_INS + CNT(i, j - 1);
|
|
|
9350 if (t < CNT(i, j))
|
|
|
9351 CNT(i, j) = t;
|
|
|
9352 }
|
|
|
9353 }
|
|
|
9354 }
|
|
344
|
9355
|
|
|
9356 i = CNT(badlen - 1, goodlen - 1);
|
|
|
9357 vim_free(cnt);
|
|
|
9358 return i;
|
|
323
|
9359 }
|
|
307
|
9360
|
|
351
|
9361 /*
|
|
|
9362 * ":spelldump"
|
|
|
9363 */
|
|
|
9364 /*ARGSUSED*/
|
|
|
9365 void
|
|
|
9366 ex_spelldump(eap)
|
|
|
9367 exarg_T *eap;
|
|
|
9368 {
|
|
|
9369 buf_T *buf = curbuf;
|
|
|
9370 langp_T *lp;
|
|
|
9371 slang_T *slang;
|
|
|
9372 idx_T arridx[MAXWLEN];
|
|
|
9373 int curi[MAXWLEN];
|
|
|
9374 char_u word[MAXWLEN];
|
|
|
9375 int c;
|
|
|
9376 char_u *byts;
|
|
|
9377 idx_T *idxs;
|
|
|
9378 linenr_T lnum = 0;
|
|
|
9379 int round;
|
|
|
9380 int depth;
|
|
|
9381 int n;
|
|
|
9382 int flags;
|
|
381
|
9383 char_u *region_names = NULL; /* region names being used */
|
|
|
9384 int do_region = TRUE; /* dump region names and numbers */
|
|
|
9385 char_u *p;
|
|
351
|
9386
|
|
|
9387 if (no_spell_checking())
|
|
|
9388 return;
|
|
|
9389
|
|
|
9390 /* Create a new empty buffer by splitting the window. */
|
|
|
9391 do_cmdline_cmd((char_u *)"new");
|
|
|
9392 if (!bufempty() || !buf_valid(buf))
|
|
|
9393 return;
|
|
|
9394
|
|
381
|
9395 /* Find out if we can support regions: All languages must support the same
|
|
|
9396 * regions or none at all. */
|
|
|
9397 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
9398 {
|
|
|
9399 p = lp->lp_slang->sl_regions;
|
|
|
9400 if (p[0] != 0)
|
|
|
9401 {
|
|
|
9402 if (region_names == NULL) /* first language with regions */
|
|
|
9403 region_names = p;
|
|
|
9404 else if (STRCMP(region_names, p) != 0)
|
|
|
9405 {
|
|
|
9406 do_region = FALSE; /* region names are different */
|
|
|
9407 break;
|
|
|
9408 }
|
|
|
9409 }
|
|
|
9410 }
|
|
|
9411
|
|
|
9412 if (do_region && region_names != NULL)
|
|
|
9413 {
|
|
|
9414 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
9415 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
9416 }
|
|
|
9417 else
|
|
|
9418 do_region = FALSE;
|
|
|
9419
|
|
|
9420 /*
|
|
|
9421 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
9422 */
|
|
351
|
9423 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
9424 {
|
|
|
9425 slang = lp->lp_slang;
|
|
|
9426
|
|
|
9427 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
9428 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
9429
|
|
|
9430 /* round 1: case-folded tree
|
|
|
9431 * round 2: keep-case tree */
|
|
|
9432 for (round = 1; round <= 2; ++round)
|
|
|
9433 {
|
|
|
9434 if (round == 1)
|
|
|
9435 {
|
|
|
9436 byts = slang->sl_fbyts;
|
|
|
9437 idxs = slang->sl_fidxs;
|
|
|
9438 }
|
|
|
9439 else
|
|
|
9440 {
|
|
|
9441 byts = slang->sl_kbyts;
|
|
|
9442 idxs = slang->sl_kidxs;
|
|
|
9443 }
|
|
|
9444 if (byts == NULL)
|
|
|
9445 continue; /* array is empty */
|
|
|
9446
|
|
|
9447 depth = 0;
|
|
|
9448 arridx[0] = 0;
|
|
|
9449 curi[0] = 1;
|
|
|
9450 while (depth >= 0 && !got_int)
|
|
|
9451 {
|
|
|
9452 if (curi[depth] > byts[arridx[depth]])
|
|
|
9453 {
|
|
|
9454 /* Done all bytes at this node, go up one level. */
|
|
|
9455 --depth;
|
|
|
9456 line_breakcheck();
|
|
|
9457 }
|
|
|
9458 else
|
|
|
9459 {
|
|
|
9460 /* Do one more byte at this node. */
|
|
|
9461 n = arridx[depth] + curi[depth];
|
|
|
9462 ++curi[depth];
|
|
|
9463 c = byts[n];
|
|
|
9464 if (c == 0)
|
|
|
9465 {
|
|
|
9466 /* End of word, deal with the word.
|
|
|
9467 * Don't use keep-case words in the fold-case tree,
|
|
|
9468 * they will appear in the keep-case tree.
|
|
|
9469 * Only use the word when the region matches. */
|
|
|
9470 flags = (int)idxs[n];
|
|
|
9471 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
9472 && (do_region
|
|
|
9473 || (flags & WF_REGION) == 0
|
|
|
9474 || (((unsigned)flags >> 8)
|
|
351
|
9475 & lp->lp_region) != 0))
|
|
|
9476 {
|
|
|
9477 word[depth] = NUL;
|
|
381
|
9478 if (!do_region)
|
|
|
9479 flags &= ~WF_REGION;
|
|
355
|
9480
|
|
|
9481 /* Dump the basic word if there is no prefix or
|
|
|
9482 * when it's the first one. */
|
|
|
9483 c = (unsigned)flags >> 16;
|
|
|
9484 if (c == 0 || curi[depth] == 2)
|
|
|
9485 dump_word(word, round, flags, lnum++);
|
|
351
|
9486
|
|
|
9487 /* Apply the prefix, if there is one. */
|
|
355
|
9488 if (c != 0)
|
|
351
|
9489 lnum = apply_prefixes(slang, word, round,
|
|
|
9490 flags, lnum);
|
|
|
9491 }
|
|
|
9492 }
|
|
|
9493 else
|
|
|
9494 {
|
|
|
9495 /* Normal char, go one level deeper. */
|
|
|
9496 word[depth++] = c;
|
|
|
9497 arridx[depth] = idxs[n];
|
|
|
9498 curi[depth] = 1;
|
|
|
9499 }
|
|
|
9500 }
|
|
|
9501 }
|
|
|
9502 }
|
|
|
9503 }
|
|
|
9504
|
|
|
9505 /* Delete the empty line that we started with. */
|
|
|
9506 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
9507 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
9508
|
|
|
9509 redraw_later(NOT_VALID);
|
|
|
9510 }
|
|
|
9511
|
|
|
9512 /*
|
|
|
9513 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
9514 */
|
|
|
9515 static void
|
|
|
9516 dump_word(word, round, flags, lnum)
|
|
|
9517 char_u *word;
|
|
|
9518 int round;
|
|
|
9519 int flags;
|
|
|
9520 linenr_T lnum;
|
|
|
9521 {
|
|
|
9522 int keepcap = FALSE;
|
|
|
9523 char_u *p;
|
|
|
9524 char_u cword[MAXWLEN];
|
|
381
|
9525 char_u badword[MAXWLEN + 10];
|
|
|
9526 int i;
|
|
351
|
9527
|
|
|
9528 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
9529 {
|
|
|
9530 /* Need to fix case according to "flags". */
|
|
|
9531 make_case_word(word, cword, flags);
|
|
|
9532 p = cword;
|
|
|
9533 }
|
|
|
9534 else
|
|
|
9535 {
|
|
|
9536 p = word;
|
|
|
9537 if (round == 2 && (captype(word, NULL) & WF_KEEPCAP) == 0)
|
|
|
9538 keepcap = TRUE;
|
|
|
9539 }
|
|
|
9540
|
|
381
|
9541 /* Add flags and regions after a slash. */
|
|
|
9542 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
9543 {
|
|
|
9544 STRCPY(badword, p);
|
|
|
9545 STRCAT(badword, "/");
|
|
351
|
9546 if (keepcap)
|
|
|
9547 STRCAT(badword, "=");
|
|
|
9548 if (flags & WF_BANNED)
|
|
|
9549 STRCAT(badword, "!");
|
|
|
9550 else if (flags & WF_RARE)
|
|
|
9551 STRCAT(badword, "?");
|
|
381
|
9552 if (flags & WF_REGION)
|
|
|
9553 for (i = 0; i < 7; ++i)
|
|
|
9554 if (flags & (0x100 << i))
|
|
|
9555 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
9556 p = badword;
|
|
|
9557 }
|
|
|
9558
|
|
|
9559 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
9560 }
|
|
|
9561
|
|
|
9562 /*
|
|
372
|
9563 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
9564 * "word" and append a line to the buffer.
|
|
351
|
9565 * Return the updated line number.
|
|
|
9566 */
|
|
|
9567 static linenr_T
|
|
|
9568 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
9569 slang_T *slang;
|
|
|
9570 char_u *word; /* case-folded word */
|
|
|
9571 int round;
|
|
|
9572 int flags; /* flags with prefix ID */
|
|
|
9573 linenr_T startlnum;
|
|
|
9574 {
|
|
|
9575 idx_T arridx[MAXWLEN];
|
|
|
9576 int curi[MAXWLEN];
|
|
|
9577 char_u prefix[MAXWLEN];
|
|
|
9578 int c;
|
|
|
9579 char_u *byts;
|
|
|
9580 idx_T *idxs;
|
|
|
9581 linenr_T lnum = startlnum;
|
|
|
9582 int depth;
|
|
|
9583 int n;
|
|
|
9584 int len;
|
|
|
9585 int prefid = (unsigned)flags >> 16;
|
|
|
9586 int i;
|
|
|
9587
|
|
|
9588 byts = slang->sl_pbyts;
|
|
|
9589 idxs = slang->sl_pidxs;
|
|
|
9590 if (byts != NULL) /* array not is empty */
|
|
|
9591 {
|
|
|
9592 /*
|
|
|
9593 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
|
9594 * When at the end of a prefix check that it supports "prefid".
|
|
|
9595 */
|
|
|
9596 depth = 0;
|
|
|
9597 arridx[0] = 0;
|
|
|
9598 curi[0] = 1;
|
|
|
9599 while (depth >= 0 && !got_int)
|
|
|
9600 {
|
|
|
9601 len = arridx[depth];
|
|
|
9602 if (curi[depth] > byts[len])
|
|
|
9603 {
|
|
|
9604 /* Done all bytes at this node, go up one level. */
|
|
|
9605 --depth;
|
|
|
9606 line_breakcheck();
|
|
|
9607 }
|
|
|
9608 else
|
|
|
9609 {
|
|
|
9610 /* Do one more byte at this node. */
|
|
|
9611 n = len + curi[depth];
|
|
|
9612 ++curi[depth];
|
|
|
9613 c = byts[n];
|
|
|
9614 if (c == 0)
|
|
|
9615 {
|
|
|
9616 /* End of prefix, find out how many IDs there are. */
|
|
|
9617 for (i = 1; i < len; ++i)
|
|
|
9618 if (byts[n + i] != 0)
|
|
|
9619 break;
|
|
|
9620 curi[depth] += i - 1;
|
|
|
9621
|
|
366
|
9622 i = valid_word_prefix(i, n, prefid, word, slang);
|
|
|
9623 if (i != 0)
|
|
351
|
9624 {
|
|
376
|
9625 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
9626 dump_word(prefix, round,
|
|
|
9627 (i & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
9628 : flags, lnum++);
|
|
351
|
9629 }
|
|
|
9630 }
|
|
|
9631 else
|
|
|
9632 {
|
|
|
9633 /* Normal char, go one level deeper. */
|
|
|
9634 prefix[depth++] = c;
|
|
|
9635 arridx[depth] = idxs[n];
|
|
|
9636 curi[depth] = 1;
|
|
|
9637 }
|
|
|
9638 }
|
|
|
9639 }
|
|
|
9640 }
|
|
|
9641
|
|
|
9642 return lnum;
|
|
|
9643 }
|
|
|
9644
|
|
236
|
9645 #endif /* FEAT_SYN_HL */
|
