|
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
|
|
339
|
321 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
322 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
323
|
|
323
|
324 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
325 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
326 there is none */
|
|
344
|
327 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
328 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
329 there is none */
|
|
375
|
330 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
331 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
332 * "sl_sal" is a list of wide char lists. */
|
|
323
|
333 int sl_followup; /* SAL followup */
|
|
|
334 int sl_collapse; /* SAL collapse_result */
|
|
|
335 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
336 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
337 #ifdef FEAT_MBYTE
|
|
|
338 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
339 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
340 #else
|
|
|
341 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
342 #endif
|
|
236
|
343 };
|
|
|
344
|
|
243
|
345 /* First language that is loaded, start of the linked list of loaded
|
|
|
346 * languages. */
|
|
236
|
347 static slang_T *first_lang = NULL;
|
|
|
348
|
|
323
|
349 /* Flags used in .spl file for soundsalike flags. */
|
|
|
350 #define SAL_F0LLOWUP 1
|
|
|
351 #define SAL_COLLAPSE 2
|
|
|
352 #define SAL_REM_ACCENTS 4
|
|
375
|
353 #define SAL_SOFO 8 /* SOFOFROM and SOFOTO instead of SAL */
|
|
323
|
354
|
|
236
|
355 /*
|
|
|
356 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
357 */
|
|
|
358 typedef struct langp_S
|
|
|
359 {
|
|
|
360 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
361 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
362 } langp_T;
|
|
|
363
|
|
|
364 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
365
|
|
307
|
366 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
367
|
|
|
368 /* Result values. Lower number is accepted over higher one. */
|
|
|
369 #define SP_BANNED -1
|
|
236
|
370 #define SP_OK 0
|
|
307
|
371 #define SP_RARE 1
|
|
|
372 #define SP_LOCAL 2
|
|
|
373 #define SP_BAD 3
|
|
236
|
374
|
|
366
|
375 #define VIMSPELLMAGIC "VIMspell08" /* string at start of Vim spell file */
|
|
236
|
376 #define VIMSPELLMAGICL 10
|
|
|
377
|
|
|
378 /*
|
|
323
|
379 * Information used when looking for suggestions.
|
|
|
380 */
|
|
|
381 typedef struct suginfo_S
|
|
|
382 {
|
|
|
383 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
384 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
385 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
386 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
387 char_u *su_badptr; /* start of bad word in line */
|
|
|
388 int su_badlen; /* length of detected bad word in line */
|
|
346
|
389 int su_badflags; /* caps flags for bad word */
|
|
323
|
390 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
391 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
392 hashtab_T su_banned; /* table with banned words */
|
|
|
393 } suginfo_T;
|
|
|
394
|
|
|
395 /* One word suggestion. Used in "si_ga". */
|
|
|
396 typedef struct suggest_S
|
|
|
397 {
|
|
|
398 char_u *st_word; /* suggested word, allocated string */
|
|
|
399 int st_orglen; /* length of replaced text */
|
|
|
400 int st_score; /* lower is better */
|
|
344
|
401 int st_altscore; /* used when st_score compares equal */
|
|
|
402 int st_salscore; /* st_score is for soundalike */
|
|
324
|
403 int st_had_bonus; /* bonus already included in score */
|
|
323
|
404 } suggest_T;
|
|
|
405
|
|
344
|
406 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
407
|
|
324
|
408 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
409 * called the score may change, thus we need to keep more than what is
|
|
|
410 * displayed. */
|
|
351
|
411 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
412
|
|
|
413 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
414 * of suggestions that are not going to be displayed. */
|
|
344
|
415 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
416
|
|
|
417 /* score for various changes */
|
|
344
|
418 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
419 #define SCORE_ICASE 52 /* slightly different case */
|
|
|
420 #define SCORE_REGION 70 /* word is for different region */
|
|
|
421 #define SCORE_RARE 180 /* rare word */
|
|
|
422 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
423 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
424 #define SCORE_REP 87 /* REP replacement */
|
|
|
425 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
426 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
324
|
427 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
428 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
324
|
429 #define SCORE_INS 96 /* insert a character */
|
|
358
|
430 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
366
|
431 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
432
|
|
372
|
433 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
434 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
435 * 350 allows for about three changes. */
|
|
344
|
436
|
|
|
437 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
438 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
439
|
|
|
440 /*
|
|
236
|
441 * Structure to store info for word matching.
|
|
|
442 */
|
|
|
443 typedef struct matchinf_S
|
|
|
444 {
|
|
|
445 langp_T *mi_lp; /* info for language and region */
|
|
243
|
446
|
|
|
447 /* pointers to original text to be checked */
|
|
236
|
448 char_u *mi_word; /* start of word being checked */
|
|
339
|
449 char_u *mi_end; /* end of matching word so far */
|
|
243
|
450 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
451 char_u *mi_cend; /* char after what was used for
|
|
|
452 mi_capflags */
|
|
243
|
453
|
|
|
454 /* case-folded text */
|
|
|
455 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
456 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
457
|
|
339
|
458 /* for when checking word after a prefix */
|
|
|
459 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
|
460 prefixID/condition */
|
|
|
461 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
462 int mi_prefixlen; /* byte length of prefix */
|
|
|
463
|
|
243
|
464 /* others */
|
|
236
|
465 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
466 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
236
|
467 } matchinf_T;
|
|
|
468
|
|
307
|
469 /*
|
|
|
470 * The tables used for recognizing word characters according to spelling.
|
|
|
471 * These are only used for the first 256 characters of 'encoding'.
|
|
|
472 */
|
|
|
473 typedef struct spelltab_S
|
|
|
474 {
|
|
|
475 char_u st_isw[256]; /* flags: is word char */
|
|
|
476 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
477 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
478 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
479 } spelltab_T;
|
|
|
480
|
|
|
481 static spelltab_T spelltab;
|
|
|
482 static int did_set_spelltab;
|
|
366
|
483 static char_u spell_ismw[256]; /* flags: is midword char */
|
|
|
484 #ifdef FEAT_MBYTE
|
|
|
485 static char_u *spell_ismw_mb = NULL; /* multi-byte midword chars */
|
|
|
486 #endif
|
|
307
|
487
|
|
324
|
488 #define CF_WORD 0x01
|
|
|
489 #define CF_UPPER 0x02
|
|
307
|
490
|
|
|
491 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
492 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
358
|
493 static int spell_iswordp __ARGS((char_u *p));
|
|
339
|
494 static void write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
495
|
|
|
496 /*
|
|
358
|
497 * Return TRUE if "p" points to a word character. Like spell_iswordp() but
|
|
|
498 * without the special handling of a single quote.
|
|
307
|
499 * Checking for a word character is done very often, avoid the function call
|
|
|
500 * overhead.
|
|
|
501 */
|
|
|
502 #ifdef FEAT_MBYTE
|
|
|
503 # define SPELL_ISWORDP(p) ((has_mbyte && MB_BYTE2LEN(*(p)) > 1) \
|
|
|
504 ? (mb_get_class(p) >= 2) : spelltab.st_isw[*(p)])
|
|
|
505 #else
|
|
|
506 # define SPELL_ISWORDP(p) (spelltab.st_isw[*(p)])
|
|
|
507 #endif
|
|
|
508
|
|
323
|
509 /*
|
|
344
|
510 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
511 */
|
|
|
512 typedef enum
|
|
|
513 {
|
|
344
|
514 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
515 * ends); if badword ends there is a match, otherwise
|
|
|
516 * try splitting word. */
|
|
375
|
517 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
518 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
519 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
520 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
521 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
522 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
523 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
524 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
525 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
526 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
527 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
528 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
529 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
530 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
531 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
532 STATE_FINAL /* End of this node. */
|
|
|
533 } state_T;
|
|
|
534
|
|
|
535 /*
|
|
346
|
536 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
537 */
|
|
|
538 typedef struct trystate_S
|
|
|
539 {
|
|
330
|
540 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
541 int ts_score; /* score */
|
|
344
|
542 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
543 short ts_curi; /* index in list of child nodes */
|
|
|
544 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
545 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
546 char_u ts_twordlen; /* valid length of tword[] */
|
|
375
|
547 char_u ts_prefixdepth; /* stack depth for end of prefix or PREFIXTREE
|
|
|
548 * or NOPREFIX */
|
|
330
|
549 #ifdef FEAT_MBYTE
|
|
|
550 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
551 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
552 char_u ts_isdiff; /* DIFF_ values */
|
|
|
553 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
554 #endif
|
|
323
|
555 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
330
|
556 char_u ts_save_splitoff; /* su_splitoff saved here */
|
|
346
|
557 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
558 } trystate_T;
|
|
|
559
|
|
330
|
560 /* values for ts_isdiff */
|
|
|
561 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
562 #define DIFF_YES 1 /* different byte found */
|
|
|
563 #define DIFF_INSERT 2 /* inserting character */
|
|
|
564
|
|
375
|
565 /* special values ts_prefixdepth */
|
|
|
566 #define PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
567 #define NOPREFIX 0xff /* not using prefixes */
|
|
|
568
|
|
339
|
569 /* mode values for find_word */
|
|
|
570 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
571 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
572 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
573
|
|
236
|
574 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
575 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
576 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
577 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
351
|
578 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int prefid, char_u *word, slang_T *slang));
|
|
339
|
579 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
580 static int fold_more __ARGS((matchinf_T *mip));
|
|
323
|
581 static int spell_valid_case __ARGS((int origflags, int treeflags));
|
|
351
|
582 static int no_spell_checking __ARGS((void));
|
|
307
|
583 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
584 static char_u *spell_enc __ARGS((void));
|
|
|
585 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
586 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
372
|
587 #ifdef FEAT_MBYTE
|
|
|
588 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
589 #endif
|
|
339
|
590 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
236
|
591 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
592 static int captype __ARGS((char_u *word, char_u *end));
|
|
323
|
593 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
307
|
594 static int set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
|
595 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
596 static void write_spell_chartab __ARGS((FILE *fd));
|
|
|
597 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
358
|
598 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword));
|
|
372
|
599 #ifdef FEAT_EVAL
|
|
|
600 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
601 #endif
|
|
|
602 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
603 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
604 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
605 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
606 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
607 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
608 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
609 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
|
610 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
611 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
612 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
613 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
614 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
615 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
616 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
617 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
618 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
619 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
620 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
621 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
622 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
623 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
624 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
625 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
626 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
627 #ifdef FEAT_MBYTE
|
|
375
|
628 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
629 #endif
|
|
344
|
630 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
631 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
632 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
633 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
634
|
|
324
|
635 /*
|
|
|
636 * Use our own character-case definitions, because the current locale may
|
|
|
637 * differ from what the .spl file uses.
|
|
|
638 * These must not be called with negative number!
|
|
|
639 */
|
|
|
640 #ifndef FEAT_MBYTE
|
|
|
641 /* Non-multi-byte implementation. */
|
|
|
642 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
643 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
644 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
645 #else
|
|
|
646 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
647 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
648 * the "w" library function for characters above 255 if available. */
|
|
|
649 # ifdef HAVE_TOWLOWER
|
|
|
650 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
651 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
652 # else
|
|
|
653 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
654 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
655 # endif
|
|
|
656
|
|
|
657 # ifdef HAVE_TOWUPPER
|
|
|
658 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
659 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
660 # else
|
|
|
661 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
662 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
663 # endif
|
|
|
664
|
|
|
665 # ifdef HAVE_ISWUPPER
|
|
|
666 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
667 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
668 # else
|
|
|
669 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
670 : (c) < 256 ? spelltab.st_isu[c] : (c))
|
|
|
671 # endif
|
|
|
672 #endif
|
|
|
673
|
|
307
|
674
|
|
|
675 static char *e_format = N_("E759: Format error in spell file");
|
|
236
|
676
|
|
|
677 /*
|
|
|
678 * Main spell-checking function.
|
|
300
|
679 * "ptr" points to a character that could be the start of a word.
|
|
236
|
680 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
681 * or when it's OK it remains unchanged.
|
|
|
682 * This must only be called when 'spelllang' is not empty.
|
|
323
|
683 *
|
|
|
684 * "sug" is normally NULL. When looking for suggestions it points to
|
|
|
685 * suginfo_T. It's passed as a void pointer to keep the struct local.
|
|
|
686 *
|
|
236
|
687 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
688 * caller can skip over the word.
|
|
|
689 */
|
|
|
690 int
|
|
300
|
691 spell_check(wp, ptr, attrp)
|
|
236
|
692 win_T *wp; /* current window */
|
|
|
693 char_u *ptr;
|
|
|
694 int *attrp;
|
|
|
695 {
|
|
|
696 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
697 be passed to functions quickly. */
|
|
344
|
698 int nrlen = 0; /* found a number first */
|
|
236
|
699
|
|
307
|
700 /* A word never starts at a space or a control character. Return quickly
|
|
|
701 * then, skipping over the character. */
|
|
|
702 if (*ptr <= ' ')
|
|
|
703 return 1;
|
|
236
|
704
|
|
344
|
705 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
706 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
707 * "3GPP". */
|
|
300
|
708 if (*ptr >= '0' && *ptr <= '9')
|
|
|
709 {
|
|
316
|
710 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
711 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
712 else
|
|
|
713 {
|
|
344
|
714 mi.mi_end = skipdigits(ptr);
|
|
|
715 nrlen = mi.mi_end - ptr;
|
|
300
|
716 }
|
|
358
|
717 if (!spell_iswordp(mi.mi_end))
|
|
344
|
718 return (int)(mi.mi_end - ptr);
|
|
346
|
719
|
|
|
720 /* Try including the digits in the word. */
|
|
|
721 mi.mi_fend = ptr + nrlen;
|
|
|
722 }
|
|
|
723 else
|
|
|
724 mi.mi_fend = ptr;
|
|
|
725
|
|
|
726 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
727 mi.mi_word = ptr;
|
|
358
|
728 if (spell_iswordp(mi.mi_fend))
|
|
344
|
729 {
|
|
|
730 do
|
|
|
731 {
|
|
307
|
732 mb_ptr_adv(mi.mi_fend);
|
|
358
|
733 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend));
|
|
344
|
734 }
|
|
|
735
|
|
|
736 /* We always use the characters up to the next non-word character,
|
|
|
737 * also for bad words. */
|
|
|
738 mi.mi_end = mi.mi_fend;
|
|
|
739
|
|
|
740 /* Check caps type later. */
|
|
|
741 mi.mi_capflags = 0;
|
|
|
742 mi.mi_cend = NULL;
|
|
|
743
|
|
|
744 /* Include one non-word character so that we can check for the
|
|
|
745 * word end. */
|
|
|
746 if (*mi.mi_fend != NUL)
|
|
|
747 mb_ptr_adv(mi.mi_fend);
|
|
|
748
|
|
|
749 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
750 MAXWLEN + 1);
|
|
|
751 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
752
|
|
|
753 /* The word is bad unless we recognize it. */
|
|
|
754 mi.mi_result = SP_BAD;
|
|
|
755
|
|
|
756 /*
|
|
|
757 * Loop over the languages specified in 'spelllang'.
|
|
|
758 * We check them all, because a matching word may be longer than an
|
|
|
759 * already found matching word.
|
|
|
760 */
|
|
|
761 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
762 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
763 {
|
|
|
764 /* Check for a matching word in case-folded words. */
|
|
|
765 find_word(&mi, FIND_FOLDWORD);
|
|
|
766
|
|
|
767 /* Check for a matching word in keep-case words. */
|
|
|
768 find_word(&mi, FIND_KEEPWORD);
|
|
|
769
|
|
|
770 /* Check for matching prefixes. */
|
|
|
771 find_prefix(&mi);
|
|
|
772 }
|
|
|
773
|
|
|
774 if (mi.mi_result != SP_OK)
|
|
|
775 {
|
|
346
|
776 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
777 * rare and local words, e.g., "3GPP". */
|
|
344
|
778 if (nrlen > 0)
|
|
346
|
779 {
|
|
|
780 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
781 return nrlen;
|
|
|
782 }
|
|
344
|
783
|
|
|
784 /* When we are at a non-word character there is no error, just
|
|
|
785 * skip over the character (try looking for a word after it). */
|
|
346
|
786 else if (!SPELL_ISWORDP(ptr))
|
|
243
|
787 {
|
|
344
|
788 #ifdef FEAT_MBYTE
|
|
|
789 if (has_mbyte)
|
|
|
790 return mb_ptr2len_check(ptr);
|
|
|
791 #endif
|
|
|
792 return 1;
|
|
300
|
793 }
|
|
243
|
794
|
|
344
|
795 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
796 *attrp = highlight_attr[HLF_SPB];
|
|
|
797 else if (mi.mi_result == SP_RARE)
|
|
|
798 *attrp = highlight_attr[HLF_SPR];
|
|
|
799 else
|
|
|
800 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
801 }
|
|
|
802
|
|
300
|
803 return (int)(mi.mi_end - ptr);
|
|
236
|
804 }
|
|
|
805
|
|
|
806 /*
|
|
300
|
807 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
808 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
809 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
810 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
811 * tree.
|
|
300
|
812 *
|
|
|
813 * For a match mip->mi_result is updated.
|
|
243
|
814 */
|
|
|
815 static void
|
|
339
|
816 find_word(mip, mode)
|
|
243
|
817 matchinf_T *mip;
|
|
339
|
818 int mode;
|
|
243
|
819 {
|
|
324
|
820 idx_T arridx = 0;
|
|
300
|
821 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
822 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
823 int endidxcnt = 0;
|
|
|
824 int len;
|
|
|
825 int wlen = 0;
|
|
|
826 int flen;
|
|
|
827 int c;
|
|
|
828 char_u *ptr;
|
|
324
|
829 idx_T lo, hi, m;
|
|
243
|
830 #ifdef FEAT_MBYTE
|
|
300
|
831 char_u *s;
|
|
339
|
832 char_u *p;
|
|
307
|
833 #endif
|
|
|
834 int res = SP_BAD;
|
|
300
|
835 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
836 unsigned flags;
|
|
|
837 char_u *byts;
|
|
324
|
838 idx_T *idxs;
|
|
339
|
839 int prefid;
|
|
|
840
|
|
|
841 if (mode == FIND_KEEPWORD)
|
|
236
|
842 {
|
|
300
|
843 /* Check for word with matching case in keep-case tree. */
|
|
|
844 ptr = mip->mi_word;
|
|
|
845 flen = 9999; /* no case folding, always enough bytes */
|
|
|
846 byts = slang->sl_kbyts;
|
|
|
847 idxs = slang->sl_kidxs;
|
|
236
|
848 }
|
|
|
849 else
|
|
|
850 {
|
|
300
|
851 /* Check for case-folded in case-folded tree. */
|
|
|
852 ptr = mip->mi_fword;
|
|
|
853 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
854 byts = slang->sl_fbyts;
|
|
|
855 idxs = slang->sl_fidxs;
|
|
339
|
856
|
|
|
857 if (mode == FIND_PREFIX)
|
|
|
858 {
|
|
|
859 /* Skip over the prefix. */
|
|
|
860 wlen = mip->mi_prefixlen;
|
|
|
861 flen -= mip->mi_prefixlen;
|
|
|
862 }
|
|
243
|
863 }
|
|
|
864
|
|
300
|
865 if (byts == NULL)
|
|
|
866 return; /* array is empty */
|
|
236
|
867
|
|
|
868 /*
|
|
307
|
869 * Repeat advancing in the tree until:
|
|
|
870 * - there is a byte that doesn't match,
|
|
|
871 * - we reach the end of the tree,
|
|
|
872 * - or we reach the end of the line.
|
|
236
|
873 */
|
|
300
|
874 for (;;)
|
|
236
|
875 {
|
|
346
|
876 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
877 flen = fold_more(mip);
|
|
300
|
878
|
|
|
879 len = byts[arridx++];
|
|
|
880
|
|
|
881 /* If the first possible byte is a zero the word could end here.
|
|
|
882 * Remember this index, we first check for the longest word. */
|
|
|
883 if (byts[arridx] == 0)
|
|
|
884 {
|
|
307
|
885 if (endidxcnt == MAXWLEN)
|
|
|
886 {
|
|
|
887 /* Must be a corrupted spell file. */
|
|
|
888 EMSG(_(e_format));
|
|
|
889 return;
|
|
|
890 }
|
|
300
|
891 endlen[endidxcnt] = wlen;
|
|
|
892 endidx[endidxcnt++] = arridx++;
|
|
|
893 --len;
|
|
|
894
|
|
|
895 /* Skip over the zeros, there can be several flag/region
|
|
|
896 * combinations. */
|
|
|
897 while (len > 0 && byts[arridx] == 0)
|
|
|
898 {
|
|
|
899 ++arridx;
|
|
|
900 --len;
|
|
|
901 }
|
|
|
902 if (len == 0)
|
|
|
903 break; /* no children, word must end here */
|
|
|
904 }
|
|
|
905
|
|
|
906 /* Stop looking at end of the line. */
|
|
|
907 if (ptr[wlen] == NUL)
|
|
|
908 break;
|
|
|
909
|
|
|
910 /* Perform a binary search in the list of accepted bytes. */
|
|
|
911 c = ptr[wlen];
|
|
346
|
912 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
913 c = ' ';
|
|
300
|
914 lo = arridx;
|
|
|
915 hi = arridx + len - 1;
|
|
|
916 while (lo < hi)
|
|
|
917 {
|
|
|
918 m = (lo + hi) / 2;
|
|
|
919 if (byts[m] > c)
|
|
|
920 hi = m - 1;
|
|
|
921 else if (byts[m] < c)
|
|
|
922 lo = m + 1;
|
|
|
923 else
|
|
|
924 {
|
|
|
925 lo = hi = m;
|
|
|
926 break;
|
|
236
|
927 }
|
|
|
928 }
|
|
300
|
929
|
|
|
930 /* Stop if there is no matching byte. */
|
|
|
931 if (hi < lo || byts[lo] != c)
|
|
|
932 break;
|
|
|
933
|
|
|
934 /* Continue at the child (if there is one). */
|
|
|
935 arridx = idxs[lo];
|
|
|
936 ++wlen;
|
|
|
937 --flen;
|
|
346
|
938
|
|
|
939 /* One space in the good word may stand for several spaces in the
|
|
|
940 * checked word. */
|
|
|
941 if (c == ' ')
|
|
|
942 {
|
|
|
943 for (;;)
|
|
|
944 {
|
|
|
945 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
946 flen = fold_more(mip);
|
|
|
947 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
948 break;
|
|
|
949 ++wlen;
|
|
|
950 --flen;
|
|
|
951 }
|
|
|
952 }
|
|
236
|
953 }
|
|
|
954
|
|
300
|
955 /*
|
|
|
956 * Verify that one of the possible endings is valid. Try the longest
|
|
|
957 * first.
|
|
|
958 */
|
|
|
959 while (endidxcnt > 0)
|
|
|
960 {
|
|
|
961 --endidxcnt;
|
|
|
962 arridx = endidx[endidxcnt];
|
|
|
963 wlen = endlen[endidxcnt];
|
|
236
|
964
|
|
300
|
965 #ifdef FEAT_MBYTE
|
|
|
966 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
967 continue; /* not at first byte of character */
|
|
|
968 #endif
|
|
358
|
969 if (spell_iswordp(ptr + wlen))
|
|
300
|
970 continue; /* next char is a word character */
|
|
|
971
|
|
|
972 #ifdef FEAT_MBYTE
|
|
339
|
973 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
974 {
|
|
|
975 /* Compute byte length in original word, length may change
|
|
339
|
976 * when folding case. This can be slow, take a shortcut when the
|
|
|
977 * case-folded word is equal to the keep-case word. */
|
|
300
|
978 p = mip->mi_word;
|
|
339
|
979 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
980 {
|
|
|
981 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
982 mb_ptr_adv(p);
|
|
|
983 wlen = p - mip->mi_word;
|
|
|
984 }
|
|
300
|
985 }
|
|
|
986 #endif
|
|
236
|
987
|
|
339
|
988 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
989 * prefix ID.
|
|
|
990 * Repeat this if there are more flags/region alternatives until there
|
|
|
991 * is a match. */
|
|
|
992 res = SP_BAD;
|
|
|
993 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
994 --len, ++arridx)
|
|
300
|
995 {
|
|
|
996 flags = idxs[arridx];
|
|
324
|
997
|
|
339
|
998 /* For the fold-case tree check that the case of the checked word
|
|
|
999 * matches with what the word in the tree requires.
|
|
|
1000 * For keep-case tree the case is always right. For prefixes we
|
|
|
1001 * don't bother to check. */
|
|
|
1002 if (mode == FIND_FOLDWORD)
|
|
300
|
1003 {
|
|
|
1004 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1005 {
|
|
323
|
1006 /* mi_capflags was set for a different word length, need
|
|
|
1007 * to do it again. */
|
|
300
|
1008 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1009 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1010 }
|
|
|
1011
|
|
346
|
1012 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1013 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1014 continue;
|
|
300
|
1015 }
|
|
236
|
1016
|
|
339
|
1017 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1018 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1019 * mip->mi_prefarridx that find_prefix() filled. */
|
|
339
|
1020 if (mode == FIND_PREFIX)
|
|
300
|
1021 {
|
|
339
|
1022 /* The prefix ID is stored two bytes above the flags. */
|
|
|
1023 prefid = (unsigned)flags >> 16;
|
|
366
|
1024 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
351
|
1025 prefid, mip->mi_fword + mip->mi_prefixlen,
|
|
366
|
1026 slang);
|
|
|
1027 if (c == 0)
|
|
339
|
1028 continue;
|
|
366
|
1029
|
|
|
1030 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1031 if (c & WF_RAREPFX)
|
|
|
1032 flags |= WF_RARE;
|
|
339
|
1033 }
|
|
|
1034
|
|
|
1035 if (flags & WF_BANNED)
|
|
|
1036 res = SP_BANNED;
|
|
|
1037 else if (flags & WF_REGION)
|
|
|
1038 {
|
|
|
1039 /* Check region. */
|
|
|
1040 if ((mip->mi_lp->lp_region & (flags >> 8)) != 0)
|
|
300
|
1041 res = SP_OK;
|
|
339
|
1042 else
|
|
|
1043 res = SP_LOCAL;
|
|
300
|
1044 }
|
|
339
|
1045 else if (flags & WF_RARE)
|
|
|
1046 res = SP_RARE;
|
|
307
|
1047 else
|
|
339
|
1048 res = SP_OK;
|
|
|
1049
|
|
|
1050 /* Always use the longest match and the best result. */
|
|
|
1051 if (mip->mi_result > res)
|
|
|
1052 {
|
|
|
1053 mip->mi_result = res;
|
|
|
1054 mip->mi_end = mip->mi_word + wlen;
|
|
|
1055 }
|
|
351
|
1056 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1057 mip->mi_end = mip->mi_word + wlen;
|
|
|
1058
|
|
|
1059 if (res == SP_OK)
|
|
|
1060 break;
|
|
300
|
1061 }
|
|
|
1062
|
|
307
|
1063 if (res == SP_OK)
|
|
300
|
1064 break;
|
|
|
1065 }
|
|
236
|
1066 }
|
|
|
1067
|
|
323
|
1068 /*
|
|
366
|
1069 * Return non-zero if the prefix indicated by "mip->mi_prefarridx" matches
|
|
|
1070 * with the prefix ID "prefid" for the word "word".
|
|
|
1071 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1072 */
|
|
|
1073 static int
|
|
|
1074 valid_word_prefix(totprefcnt, arridx, prefid, word, slang)
|
|
|
1075 int totprefcnt; /* nr of prefix IDs */
|
|
|
1076 int arridx; /* idx in sl_pidxs[] */
|
|
|
1077 int prefid;
|
|
|
1078 char_u *word;
|
|
|
1079 slang_T *slang;
|
|
|
1080 {
|
|
|
1081 int prefcnt;
|
|
|
1082 int pidx;
|
|
|
1083 regprog_T *rp;
|
|
|
1084 regmatch_T regmatch;
|
|
|
1085
|
|
|
1086 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1087 {
|
|
|
1088 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1089
|
|
|
1090 /* Check the prefix ID. */
|
|
|
1091 if (prefid != (pidx & 0xff))
|
|
|
1092 continue;
|
|
|
1093
|
|
|
1094 /* Check the condition, if there is one. The condition index is
|
|
366
|
1095 * stored in the two bytes above the prefix ID byte. */
|
|
|
1096 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1097 if (rp != NULL)
|
|
|
1098 {
|
|
|
1099 regmatch.regprog = rp;
|
|
|
1100 regmatch.rm_ic = FALSE;
|
|
|
1101 if (!vim_regexec(®match, word, 0))
|
|
|
1102 continue;
|
|
|
1103 }
|
|
|
1104
|
|
366
|
1105 /* It's a match! Return the WF_RAREPFX flag. */
|
|
|
1106 return pidx;
|
|
|
1107 }
|
|
|
1108 return 0;
|
|
351
|
1109 }
|
|
|
1110
|
|
|
1111 /*
|
|
339
|
1112 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1113 * If it does, then check the following word.
|
|
|
1114 *
|
|
|
1115 * For a match mip->mi_result is updated.
|
|
|
1116 */
|
|
|
1117 static void
|
|
|
1118 find_prefix(mip)
|
|
|
1119 matchinf_T *mip;
|
|
|
1120 {
|
|
|
1121 idx_T arridx = 0;
|
|
|
1122 int len;
|
|
|
1123 int wlen = 0;
|
|
|
1124 int flen;
|
|
|
1125 int c;
|
|
|
1126 char_u *ptr;
|
|
|
1127 idx_T lo, hi, m;
|
|
|
1128 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1129 char_u *byts;
|
|
|
1130 idx_T *idxs;
|
|
|
1131
|
|
375
|
1132 byts = slang->sl_pbyts;
|
|
|
1133 if (byts == NULL)
|
|
|
1134 return; /* array is empty */
|
|
|
1135
|
|
339
|
1136 /* We use the case-folded word here, since prefixes are always
|
|
|
1137 * case-folded. */
|
|
|
1138 ptr = mip->mi_fword;
|
|
|
1139 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1140 idxs = slang->sl_pidxs;
|
|
|
1141
|
|
|
1142 /*
|
|
|
1143 * Repeat advancing in the tree until:
|
|
|
1144 * - there is a byte that doesn't match,
|
|
|
1145 * - we reach the end of the tree,
|
|
|
1146 * - or we reach the end of the line.
|
|
|
1147 */
|
|
|
1148 for (;;)
|
|
|
1149 {
|
|
|
1150 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1151 flen = fold_more(mip);
|
|
|
1152
|
|
|
1153 len = byts[arridx++];
|
|
|
1154
|
|
|
1155 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1156 * Check if the following word matches and supports the prefix. */
|
|
|
1157 if (byts[arridx] == 0)
|
|
|
1158 {
|
|
|
1159 /* There can be several prefixes with different conditions. We
|
|
|
1160 * try them all, since we don't know which one will give the
|
|
|
1161 * longest match. The word is the same each time, pass the list
|
|
|
1162 * of possible prefixes to find_word(). */
|
|
|
1163 mip->mi_prefarridx = arridx;
|
|
|
1164 mip->mi_prefcnt = len;
|
|
|
1165 while (len > 0 && byts[arridx] == 0)
|
|
|
1166 {
|
|
|
1167 ++arridx;
|
|
|
1168 --len;
|
|
|
1169 }
|
|
|
1170 mip->mi_prefcnt -= len;
|
|
|
1171
|
|
|
1172 /* Find the word that comes after the prefix. */
|
|
|
1173 mip->mi_prefixlen = wlen;
|
|
|
1174 find_word(mip, FIND_PREFIX);
|
|
|
1175
|
|
|
1176
|
|
|
1177 if (len == 0)
|
|
|
1178 break; /* no children, word must end here */
|
|
|
1179 }
|
|
|
1180
|
|
|
1181 /* Stop looking at end of the line. */
|
|
|
1182 if (ptr[wlen] == NUL)
|
|
|
1183 break;
|
|
|
1184
|
|
|
1185 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1186 c = ptr[wlen];
|
|
|
1187 lo = arridx;
|
|
|
1188 hi = arridx + len - 1;
|
|
|
1189 while (lo < hi)
|
|
|
1190 {
|
|
|
1191 m = (lo + hi) / 2;
|
|
|
1192 if (byts[m] > c)
|
|
|
1193 hi = m - 1;
|
|
|
1194 else if (byts[m] < c)
|
|
|
1195 lo = m + 1;
|
|
|
1196 else
|
|
|
1197 {
|
|
|
1198 lo = hi = m;
|
|
|
1199 break;
|
|
|
1200 }
|
|
|
1201 }
|
|
|
1202
|
|
|
1203 /* Stop if there is no matching byte. */
|
|
|
1204 if (hi < lo || byts[lo] != c)
|
|
|
1205 break;
|
|
|
1206
|
|
|
1207 /* Continue at the child (if there is one). */
|
|
|
1208 arridx = idxs[lo];
|
|
|
1209 ++wlen;
|
|
|
1210 --flen;
|
|
|
1211 }
|
|
|
1212 }
|
|
|
1213
|
|
|
1214 /*
|
|
|
1215 * Need to fold at least one more character. Do until next non-word character
|
|
|
1216 * for efficiency.
|
|
|
1217 * Return the length of the folded chars in bytes.
|
|
|
1218 */
|
|
|
1219 static int
|
|
|
1220 fold_more(mip)
|
|
|
1221 matchinf_T *mip;
|
|
|
1222 {
|
|
|
1223 int flen;
|
|
|
1224 char_u *p;
|
|
|
1225
|
|
|
1226 p = mip->mi_fend;
|
|
|
1227 do
|
|
|
1228 {
|
|
|
1229 mb_ptr_adv(mip->mi_fend);
|
|
358
|
1230 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend));
|
|
339
|
1231
|
|
|
1232 /* Include the non-word character so that we can check for the
|
|
|
1233 * word end. */
|
|
|
1234 if (*mip->mi_fend != NUL)
|
|
|
1235 mb_ptr_adv(mip->mi_fend);
|
|
|
1236
|
|
|
1237 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1238 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1239 MAXWLEN - mip->mi_fwordlen);
|
|
|
1240 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1241 mip->mi_fwordlen += flen;
|
|
|
1242 return flen;
|
|
|
1243 }
|
|
|
1244
|
|
|
1245 /*
|
|
323
|
1246 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1247 * case.
|
|
|
1248 */
|
|
|
1249 static int
|
|
|
1250 spell_valid_case(origflags, treeflags)
|
|
|
1251 int origflags; /* flags for the checked word. */
|
|
|
1252 int treeflags; /* flags for the word in the spell tree */
|
|
|
1253 {
|
|
|
1254 return (origflags == WF_ALLCAP
|
|
|
1255 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
|
1256 && ((treeflags & WF_ONECAP) == 0 || origflags == WF_ONECAP)));
|
|
|
1257 }
|
|
|
1258
|
|
351
|
1259 /*
|
|
|
1260 * Return TRUE if spell checking is not enabled.
|
|
|
1261 */
|
|
|
1262 static int
|
|
|
1263 no_spell_checking()
|
|
|
1264 {
|
|
|
1265 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1266 {
|
|
|
1267 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1268 return TRUE;
|
|
|
1269 }
|
|
|
1270 return FALSE;
|
|
|
1271 }
|
|
300
|
1272
|
|
236
|
1273 /*
|
|
|
1274 * Move to next spell error.
|
|
323
|
1275 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
236
|
1276 * Return OK if found, FAIL otherwise.
|
|
|
1277 */
|
|
|
1278 int
|
|
323
|
1279 spell_move_to(dir, allwords, curline)
|
|
236
|
1280 int dir; /* FORWARD or BACKWARD */
|
|
|
1281 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1282 int curline;
|
|
236
|
1283 {
|
|
249
|
1284 linenr_T lnum;
|
|
|
1285 pos_T found_pos;
|
|
236
|
1286 char_u *line;
|
|
|
1287 char_u *p;
|
|
346
|
1288 char_u *endp;
|
|
|
1289 int attr;
|
|
236
|
1290 int len;
|
|
249
|
1291 int has_syntax = syntax_present(curbuf);
|
|
|
1292 int col;
|
|
|
1293 int can_spell;
|
|
346
|
1294 char_u *buf = NULL;
|
|
|
1295 int buflen = 0;
|
|
|
1296 int skip = 0;
|
|
236
|
1297
|
|
351
|
1298 if (no_spell_checking())
|
|
236
|
1299 return FAIL;
|
|
|
1300
|
|
249
|
1301 /*
|
|
|
1302 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1303 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1304 *
|
|
|
1305 * When searching backwards, we continue in the line to find the last
|
|
|
1306 * bad word (in the cursor line: before the cursor).
|
|
346
|
1307 *
|
|
|
1308 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1309 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1310 * though...
|
|
249
|
1311 */
|
|
|
1312 lnum = curwin->w_cursor.lnum;
|
|
|
1313 found_pos.lnum = 0;
|
|
236
|
1314
|
|
|
1315 while (!got_int)
|
|
|
1316 {
|
|
249
|
1317 line = ml_get(lnum);
|
|
346
|
1318
|
|
|
1319 len = STRLEN(line);
|
|
|
1320 if (buflen < len + MAXWLEN + 2)
|
|
|
1321 {
|
|
|
1322 vim_free(buf);
|
|
|
1323 buflen = len + MAXWLEN + 2;
|
|
|
1324 buf = alloc(buflen);
|
|
|
1325 if (buf == NULL)
|
|
|
1326 break;
|
|
|
1327 }
|
|
|
1328
|
|
|
1329 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1330 * possible. */
|
|
|
1331 STRCPY(buf, line);
|
|
|
1332 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1333 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1334
|
|
|
1335 p = buf + skip;
|
|
|
1336 endp = buf + len;
|
|
|
1337 while (p < endp)
|
|
236
|
1338 {
|
|
300
|
1339 /* When searching backward don't search after the cursor. */
|
|
|
1340 if (dir == BACKWARD
|
|
|
1341 && lnum == curwin->w_cursor.lnum
|
|
346
|
1342 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1343 break;
|
|
249
|
1344
|
|
300
|
1345 /* start of word */
|
|
346
|
1346 attr = 0;
|
|
300
|
1347 len = spell_check(curwin, p, &attr);
|
|
249
|
1348
|
|
300
|
1349 if (attr != 0)
|
|
|
1350 {
|
|
|
1351 /* We found a bad word. Check the attribute. */
|
|
|
1352 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1353 {
|
|
300
|
1354 /* When searching forward only accept a bad word after
|
|
|
1355 * the cursor. */
|
|
|
1356 if (dir == BACKWARD
|
|
|
1357 || lnum > curwin->w_cursor.lnum
|
|
|
1358 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1359 && (colnr_T)(curline ? p - buf + len
|
|
|
1360 : p - buf)
|
|
300
|
1361 > curwin->w_cursor.col))
|
|
236
|
1362 {
|
|
300
|
1363 if (has_syntax)
|
|
249
|
1364 {
|
|
346
|
1365 col = p - buf;
|
|
300
|
1366 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1367 FALSE, &can_spell);
|
|
|
1368 }
|
|
|
1369 else
|
|
|
1370 can_spell = TRUE;
|
|
249
|
1371
|
|
300
|
1372 if (can_spell)
|
|
|
1373 {
|
|
|
1374 found_pos.lnum = lnum;
|
|
346
|
1375 found_pos.col = p - buf;
|
|
249
|
1376 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1377 found_pos.coladd = 0;
|
|
249
|
1378 #endif
|
|
300
|
1379 if (dir == FORWARD)
|
|
|
1380 {
|
|
|
1381 /* No need to search further. */
|
|
|
1382 curwin->w_cursor = found_pos;
|
|
346
|
1383 vim_free(buf);
|
|
300
|
1384 return OK;
|
|
249
|
1385 }
|
|
|
1386 }
|
|
236
|
1387 }
|
|
|
1388 }
|
|
|
1389 }
|
|
|
1390
|
|
300
|
1391 /* advance to character after the word */
|
|
|
1392 p += len;
|
|
236
|
1393 }
|
|
|
1394
|
|
323
|
1395 if (curline)
|
|
346
|
1396 break; /* only check cursor line */
|
|
323
|
1397
|
|
236
|
1398 /* Advance to next line. */
|
|
249
|
1399 if (dir == BACKWARD)
|
|
|
1400 {
|
|
|
1401 if (found_pos.lnum != 0)
|
|
|
1402 {
|
|
|
1403 /* Use the last match in the line. */
|
|
|
1404 curwin->w_cursor = found_pos;
|
|
346
|
1405 vim_free(buf);
|
|
249
|
1406 return OK;
|
|
|
1407 }
|
|
|
1408 if (lnum == 1)
|
|
346
|
1409 break;
|
|
249
|
1410 --lnum;
|
|
|
1411 }
|
|
|
1412 else
|
|
|
1413 {
|
|
|
1414 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1415 break;
|
|
249
|
1416 ++lnum;
|
|
346
|
1417
|
|
|
1418 /* Skip the characters at the start of the next line that were
|
|
|
1419 * included in a match crossing line boundaries. */
|
|
|
1420 if (attr == 0)
|
|
|
1421 skip = p - endp;
|
|
|
1422 else
|
|
|
1423 skip = 0;
|
|
249
|
1424 }
|
|
236
|
1425
|
|
|
1426 line_breakcheck();
|
|
|
1427 }
|
|
|
1428
|
|
346
|
1429 vim_free(buf);
|
|
|
1430 return FAIL;
|
|
|
1431 }
|
|
|
1432
|
|
|
1433 /*
|
|
|
1434 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1435 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1436 */
|
|
|
1437 void
|
|
|
1438 spell_cat_line(buf, line, maxlen)
|
|
|
1439 char_u *buf;
|
|
|
1440 char_u *line;
|
|
|
1441 int maxlen;
|
|
|
1442 {
|
|
|
1443 char_u *p;
|
|
|
1444 int n;
|
|
|
1445
|
|
|
1446 p = skipwhite(line);
|
|
|
1447 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1448 p = skipwhite(p + 1);
|
|
|
1449
|
|
|
1450 if (*p != NUL)
|
|
|
1451 {
|
|
|
1452 *buf = ' ';
|
|
|
1453 vim_strncpy(buf + 1, line, maxlen - 1);
|
|
|
1454 n = p - line;
|
|
|
1455 if (n >= maxlen)
|
|
|
1456 n = maxlen - 1;
|
|
|
1457 vim_memset(buf + 1, ' ', n);
|
|
|
1458 }
|
|
236
|
1459 }
|
|
|
1460
|
|
|
1461 /*
|
|
307
|
1462 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1463 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1464 */
|
|
307
|
1465 static void
|
|
236
|
1466 spell_load_lang(lang)
|
|
|
1467 char_u *lang;
|
|
|
1468 {
|
|
310
|
1469 char_u fname_enc[85];
|
|
236
|
1470 int r;
|
|
307
|
1471 char_u langcp[MAXWLEN + 1];
|
|
|
1472
|
|
310
|
1473 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1474 * It's truncated when an error is detected. */
|
|
|
1475 STRCPY(langcp, lang);
|
|
|
1476
|
|
310
|
1477 /*
|
|
|
1478 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1479 */
|
|
|
1480 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1481 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1482 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1483
|
|
|
1484 if (r == FAIL && *langcp != NUL)
|
|
|
1485 {
|
|
|
1486 /* Try loading the ASCII version. */
|
|
310
|
1487 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1488 "spell/%s.ascii.spl", lang);
|
|
310
|
1489 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1490 }
|
|
|
1491
|
|
|
1492 if (r == FAIL)
|
|
|
1493 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
|
236
|
1494 fname_enc + 6);
|
|
310
|
1495 else if (*langcp != NUL)
|
|
|
1496 {
|
|
|
1497 /* Load all the additions. */
|
|
|
1498 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1499 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1500 }
|
|
|
1501 }
|
|
|
1502
|
|
|
1503 /*
|
|
|
1504 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1505 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1506 */
|
|
|
1507 static char_u *
|
|
|
1508 spell_enc()
|
|
|
1509 {
|
|
|
1510
|
|
|
1511 #ifdef FEAT_MBYTE
|
|
|
1512 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1513 return p_enc;
|
|
|
1514 #endif
|
|
|
1515 return (char_u *)"latin1";
|
|
236
|
1516 }
|
|
|
1517
|
|
|
1518 /*
|
|
|
1519 * Allocate a new slang_T.
|
|
|
1520 * Caller must fill "sl_next".
|
|
|
1521 */
|
|
|
1522 static slang_T *
|
|
|
1523 slang_alloc(lang)
|
|
|
1524 char_u *lang;
|
|
|
1525 {
|
|
|
1526 slang_T *lp;
|
|
|
1527
|
|
300
|
1528 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1529 if (lp != NULL)
|
|
|
1530 {
|
|
|
1531 lp->sl_name = vim_strsave(lang);
|
|
323
|
1532 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
344
|
1533 ga_init2(&lp->sl_sal, sizeof(salitem_T), 10);
|
|
236
|
1534 }
|
|
|
1535 return lp;
|
|
|
1536 }
|
|
|
1537
|
|
|
1538 /*
|
|
|
1539 * Free the contents of an slang_T and the structure itself.
|
|
|
1540 */
|
|
|
1541 static void
|
|
|
1542 slang_free(lp)
|
|
|
1543 slang_T *lp;
|
|
|
1544 {
|
|
|
1545 vim_free(lp->sl_name);
|
|
310
|
1546 vim_free(lp->sl_fname);
|
|
|
1547 slang_clear(lp);
|
|
|
1548 vim_free(lp);
|
|
|
1549 }
|
|
|
1550
|
|
|
1551 /*
|
|
|
1552 * Clear an slang_T so that the file can be reloaded.
|
|
|
1553 */
|
|
|
1554 static void
|
|
|
1555 slang_clear(lp)
|
|
|
1556 slang_T *lp;
|
|
|
1557 {
|
|
339
|
1558 garray_T *gap;
|
|
|
1559 fromto_T *ftp;
|
|
344
|
1560 salitem_T *smp;
|
|
339
|
1561 int i;
|
|
323
|
1562
|
|
300
|
1563 vim_free(lp->sl_fbyts);
|
|
310
|
1564 lp->sl_fbyts = NULL;
|
|
300
|
1565 vim_free(lp->sl_kbyts);
|
|
310
|
1566 lp->sl_kbyts = NULL;
|
|
339
|
1567 vim_free(lp->sl_pbyts);
|
|
|
1568 lp->sl_pbyts = NULL;
|
|
|
1569
|
|
300
|
1570 vim_free(lp->sl_fidxs);
|
|
310
|
1571 lp->sl_fidxs = NULL;
|
|
300
|
1572 vim_free(lp->sl_kidxs);
|
|
310
|
1573 lp->sl_kidxs = NULL;
|
|
339
|
1574 vim_free(lp->sl_pidxs);
|
|
|
1575 lp->sl_pidxs = NULL;
|
|
323
|
1576
|
|
344
|
1577 gap = &lp->sl_rep;
|
|
|
1578 while (gap->ga_len > 0)
|
|
323
|
1579 {
|
|
344
|
1580 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1581 vim_free(ftp->ft_from);
|
|
|
1582 vim_free(ftp->ft_to);
|
|
323
|
1583 }
|
|
344
|
1584 ga_clear(gap);
|
|
|
1585
|
|
|
1586 gap = &lp->sl_sal;
|
|
375
|
1587 if (lp->sl_sofo)
|
|
|
1588 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1589 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1590 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1591 else
|
|
|
1592 /* SAL items: free salitem_T items */
|
|
|
1593 while (gap->ga_len > 0)
|
|
|
1594 {
|
|
|
1595 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1596 vim_free(smp->sm_lead);
|
|
|
1597 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1598 vim_free(smp->sm_to);
|
|
|
1599 #ifdef FEAT_MBYTE
|
|
|
1600 vim_free(smp->sm_lead_w);
|
|
|
1601 vim_free(smp->sm_oneof_w);
|
|
|
1602 vim_free(smp->sm_to_w);
|
|
|
1603 #endif
|
|
|
1604 }
|
|
344
|
1605 ga_clear(gap);
|
|
323
|
1606
|
|
339
|
1607 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1608 vim_free(lp->sl_prefprog[i]);
|
|
|
1609 vim_free(lp->sl_prefprog);
|
|
|
1610
|
|
330
|
1611 #ifdef FEAT_MBYTE
|
|
|
1612 {
|
|
|
1613 int todo = lp->sl_map_hash.ht_used;
|
|
|
1614 hashitem_T *hi;
|
|
|
1615
|
|
|
1616 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1617 if (!HASHITEM_EMPTY(hi))
|
|
|
1618 {
|
|
|
1619 --todo;
|
|
|
1620 vim_free(hi->hi_key);
|
|
|
1621 }
|
|
|
1622 }
|
|
|
1623 hash_clear(&lp->sl_map_hash);
|
|
|
1624 #endif
|
|
236
|
1625 }
|
|
|
1626
|
|
|
1627 /*
|
|
307
|
1628 * Load one spell file and store the info into a slang_T.
|
|
236
|
1629 * Invoked through do_in_runtimepath().
|
|
|
1630 */
|
|
|
1631 static void
|
|
310
|
1632 spell_load_cb(fname, cookie)
|
|
236
|
1633 char_u *fname;
|
|
307
|
1634 void *cookie; /* points to the language name */
|
|
236
|
1635 {
|
|
323
|
1636 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1637 }
|
|
|
1638
|
|
|
1639 /*
|
|
|
1640 * Load one spell file and store the info into a slang_T.
|
|
|
1641 *
|
|
|
1642 * This is invoked in two ways:
|
|
|
1643 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
1644 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
1645 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
1646 * points to the existing slang_T.
|
|
323
|
1647 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
1648 */
|
|
323
|
1649 static slang_T *
|
|
|
1650 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
1651 char_u *fname;
|
|
|
1652 char_u *lang;
|
|
|
1653 slang_T *old_lp;
|
|
323
|
1654 int silent; /* no error if file doesn't exist */
|
|
310
|
1655 {
|
|
236
|
1656 FILE *fd;
|
|
|
1657 char_u buf[MAXWLEN + 1];
|
|
|
1658 char_u *p;
|
|
339
|
1659 char_u *bp;
|
|
|
1660 idx_T *ip;
|
|
236
|
1661 int i;
|
|
339
|
1662 int n;
|
|
300
|
1663 int len;
|
|
236
|
1664 int round;
|
|
|
1665 char_u *save_sourcing_name = sourcing_name;
|
|
|
1666 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
255
|
1667 int cnt, ccnt;
|
|
|
1668 char_u *fol;
|
|
307
|
1669 slang_T *lp = NULL;
|
|
323
|
1670 garray_T *gap;
|
|
|
1671 fromto_T *ftp;
|
|
344
|
1672 salitem_T *smp;
|
|
323
|
1673 int rr;
|
|
|
1674 short *first;
|
|
375
|
1675 salfirst_T *sfirst;
|
|
324
|
1676 idx_T idx;
|
|
344
|
1677 int c = 0;
|
|
236
|
1678
|
|
310
|
1679 fd = mch_fopen((char *)fname, "r");
|
|
236
|
1680 if (fd == NULL)
|
|
|
1681 {
|
|
323
|
1682 if (!silent)
|
|
|
1683 EMSG2(_(e_notopen), fname);
|
|
|
1684 else if (p_verbose > 2)
|
|
|
1685 {
|
|
|
1686 verbose_enter();
|
|
|
1687 smsg((char_u *)e_notopen, fname);
|
|
|
1688 verbose_leave();
|
|
|
1689 }
|
|
255
|
1690 goto endFAIL;
|
|
236
|
1691 }
|
|
310
|
1692 if (p_verbose > 2)
|
|
|
1693 {
|
|
|
1694 verbose_enter();
|
|
|
1695 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
1696 verbose_leave();
|
|
|
1697 }
|
|
|
1698
|
|
|
1699 if (old_lp == NULL)
|
|
|
1700 {
|
|
|
1701 lp = slang_alloc(lang);
|
|
|
1702 if (lp == NULL)
|
|
|
1703 goto endFAIL;
|
|
|
1704
|
|
|
1705 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
1706 lp->sl_fname = vim_strsave(fname);
|
|
|
1707 if (lp->sl_fname == NULL)
|
|
|
1708 goto endFAIL;
|
|
|
1709
|
|
|
1710 /* Check for .add.spl. */
|
|
|
1711 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
1712 }
|
|
|
1713 else
|
|
|
1714 lp = old_lp;
|
|
307
|
1715
|
|
236
|
1716 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
1717 sourcing_name = fname;
|
|
|
1718 sourcing_lnum = 0;
|
|
|
1719
|
|
339
|
1720 /* <HEADER>: <fileID>
|
|
|
1721 * <regioncnt> <regionname> ...
|
|
|
1722 * <charflagslen> <charflags>
|
|
|
1723 * <fcharslen> <fchars>
|
|
366
|
1724 * <midwordlen> <midword>
|
|
339
|
1725 * <prefcondcnt> <prefcond> ...
|
|
|
1726 */
|
|
236
|
1727 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
1728 buf[i] = getc(fd); /* <fileID> */
|
|
|
1729 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
1730 {
|
|
|
1731 EMSG(_("E757: Wrong file ID in spell file"));
|
|
255
|
1732 goto endFAIL;
|
|
236
|
1733 }
|
|
|
1734
|
|
|
1735 cnt = getc(fd); /* <regioncnt> */
|
|
255
|
1736 if (cnt < 0)
|
|
236
|
1737 {
|
|
|
1738 truncerr:
|
|
|
1739 EMSG(_("E758: Truncated spell file"));
|
|
255
|
1740 goto endFAIL;
|
|
236
|
1741 }
|
|
|
1742 if (cnt > 8)
|
|
|
1743 {
|
|
|
1744 formerr:
|
|
307
|
1745 EMSG(_(e_format));
|
|
255
|
1746 goto endFAIL;
|
|
236
|
1747 }
|
|
|
1748 for (i = 0; i < cnt; ++i)
|
|
|
1749 {
|
|
|
1750 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
|
1751 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
|
1752 }
|
|
|
1753 lp->sl_regions[cnt * 2] = NUL;
|
|
|
1754
|
|
255
|
1755 cnt = getc(fd); /* <charflagslen> */
|
|
|
1756 if (cnt > 0)
|
|
|
1757 {
|
|
300
|
1758 p = alloc((unsigned)cnt);
|
|
255
|
1759 if (p == NULL)
|
|
|
1760 goto endFAIL;
|
|
|
1761 for (i = 0; i < cnt; ++i)
|
|
|
1762 p[i] = getc(fd); /* <charflags> */
|
|
|
1763
|
|
|
1764 ccnt = (getc(fd) << 8) + getc(fd); /* <fcharslen> */
|
|
|
1765 if (ccnt <= 0)
|
|
300
|
1766 {
|
|
|
1767 vim_free(p);
|
|
255
|
1768 goto formerr;
|
|
300
|
1769 }
|
|
|
1770 fol = alloc((unsigned)ccnt + 1);
|
|
255
|
1771 if (fol == NULL)
|
|
300
|
1772 {
|
|
|
1773 vim_free(p);
|
|
255
|
1774 goto endFAIL;
|
|
300
|
1775 }
|
|
255
|
1776 for (i = 0; i < ccnt; ++i)
|
|
|
1777 fol[i] = getc(fd); /* <fchars> */
|
|
|
1778 fol[i] = NUL;
|
|
|
1779
|
|
324
|
1780 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
300
|
1781 i = set_spell_charflags(p, cnt, fol);
|
|
|
1782 vim_free(p);
|
|
|
1783 vim_free(fol);
|
|
351
|
1784 #if 0 /* tolerate the differences */
|
|
300
|
1785 if (i == FAIL)
|
|
255
|
1786 goto formerr;
|
|
351
|
1787 #endif
|
|
255
|
1788 }
|
|
|
1789 else
|
|
|
1790 {
|
|
|
1791 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
1792 cnt = (getc(fd) << 8) + getc(fd);
|
|
|
1793 if (cnt != 0)
|
|
|
1794 goto formerr;
|
|
|
1795 }
|
|
|
1796
|
|
366
|
1797 /* <midwordlen> <midword> */
|
|
|
1798 cnt = (getc(fd) << 8) + getc(fd);
|
|
|
1799 if (cnt < 0)
|
|
|
1800 goto truncerr;
|
|
|
1801 if (cnt > 0)
|
|
|
1802 {
|
|
|
1803 for (i = 0; i < cnt; ++i)
|
|
|
1804 if (i < MAXWLEN) /* truncate at reasonable length */
|
|
|
1805 buf[i] = getc(fd);
|
|
|
1806 if (i < MAXWLEN)
|
|
|
1807 buf[i] = NUL;
|
|
|
1808 else
|
|
|
1809 buf[MAXWLEN] = NUL;
|
|
|
1810
|
|
|
1811 /* The midword characters add up to any midword characters from other
|
|
|
1812 * .spel files. */
|
|
|
1813 for (p = buf; *p != NUL; )
|
|
|
1814 #ifdef FEAT_MBYTE
|
|
|
1815 if (has_mbyte)
|
|
|
1816 {
|
|
|
1817 c = mb_ptr2char(p);
|
|
|
1818 i = mb_ptr2len_check(p);
|
|
|
1819 if (c < 256)
|
|
|
1820 spell_ismw[c] = TRUE;
|
|
|
1821 else if (spell_ismw_mb == NULL)
|
|
|
1822 /* First multi-byte char in "spell_ismw_mb". */
|
|
|
1823 spell_ismw_mb = vim_strnsave(p, i);
|
|
|
1824 else
|
|
|
1825 {
|
|
|
1826 /* Append multi-byte chars to "spell_ismw_mb". */
|
|
|
1827 n = STRLEN(spell_ismw_mb);
|
|
|
1828 bp = vim_strnsave(spell_ismw_mb, n + i);
|
|
|
1829 if (bp != NULL)
|
|
|
1830 {
|
|
|
1831 vim_free(spell_ismw_mb);
|
|
|
1832 spell_ismw_mb = bp;
|
|
|
1833 vim_strncpy(bp + n, p, i);
|
|
|
1834 }
|
|
|
1835 }
|
|
|
1836 p += i;
|
|
|
1837 }
|
|
|
1838 else
|
|
|
1839 #endif
|
|
|
1840 spell_ismw[*p++] = TRUE;
|
|
|
1841 }
|
|
|
1842
|
|
339
|
1843 /* <prefcondcnt> <prefcond> ... */
|
|
|
1844 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
1845 if (cnt > 0)
|
|
|
1846 {
|
|
|
1847 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
1848 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
1849 if (lp->sl_prefprog == NULL)
|
|
|
1850 goto endFAIL;
|
|
|
1851 lp->sl_prefixcnt = cnt;
|
|
|
1852
|
|
|
1853 for (i = 0; i < cnt; ++i)
|
|
|
1854 {
|
|
|
1855 /* <prefcond> : <condlen> <condstr> */
|
|
|
1856 n = getc(fd); /* <condlen> */
|
|
|
1857 if (n < 0)
|
|
|
1858 goto formerr;
|
|
|
1859 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
1860 * compile the regexp program used to check for the condition. */
|
|
|
1861 if (n > 0)
|
|
|
1862 {
|
|
344
|
1863 buf[0] = '^'; /* always match at one position only */
|
|
|
1864 p = buf + 1;
|
|
339
|
1865 while (n-- > 0)
|
|
|
1866 *p++ = getc(fd); /* <condstr> */
|
|
|
1867 *p = NUL;
|
|
|
1868 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
1869 }
|
|
|
1870 }
|
|
|
1871 }
|
|
|
1872
|
|
|
1873
|
|
323
|
1874 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
1875 * <salflags> <salcount> <sal> ...
|
|
|
1876 * <maplen> <mapstr> */
|
|
339
|
1877
|
|
344
|
1878 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
1879 if (cnt < 0)
|
|
|
1880 goto formerr;
|
|
|
1881
|
|
|
1882 gap = &lp->sl_rep;
|
|
|
1883 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1884 goto endFAIL;
|
|
|
1885
|
|
|
1886 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
1887 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
1888 {
|
|
|
1889 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
|
1890 for (rr = 1; rr <= 2; ++rr)
|
|
|
1891 {
|
|
|
1892 ccnt = getc(fd);
|
|
|
1893 if (ccnt < 0)
|
|
|
1894 {
|
|
|
1895 if (rr == 2)
|
|
|
1896 vim_free(ftp->ft_from);
|
|
|
1897 goto formerr;
|
|
|
1898 }
|
|
|
1899 if ((p = alloc(ccnt + 1)) == NULL)
|
|
|
1900 {
|
|
|
1901 if (rr == 2)
|
|
|
1902 vim_free(ftp->ft_from);
|
|
|
1903 goto endFAIL;
|
|
|
1904 }
|
|
|
1905 for (i = 0; i < ccnt; ++i)
|
|
|
1906 p[i] = getc(fd); /* <repfrom> or <repto> */
|
|
|
1907 p[i] = NUL;
|
|
|
1908 if (rr == 1)
|
|
|
1909 ftp->ft_from = p;
|
|
|
1910 else
|
|
|
1911 ftp->ft_to = p;
|
|
|
1912 }
|
|
|
1913 }
|
|
|
1914
|
|
|
1915 /* Fill the first-index table. */
|
|
|
1916 first = lp->sl_rep_first;
|
|
|
1917 for (i = 0; i < 256; ++i)
|
|
|
1918 first[i] = -1;
|
|
|
1919 for (i = 0; i < gap->ga_len; ++i)
|
|
323
|
1920 {
|
|
344
|
1921 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
1922 if (first[*ftp->ft_from] == -1)
|
|
|
1923 first[*ftp->ft_from] = i;
|
|
|
1924 }
|
|
|
1925
|
|
|
1926 i = getc(fd); /* <salflags> */
|
|
|
1927 if (i & SAL_F0LLOWUP)
|
|
|
1928 lp->sl_followup = TRUE;
|
|
|
1929 if (i & SAL_COLLAPSE)
|
|
|
1930 lp->sl_collapse = TRUE;
|
|
|
1931 if (i & SAL_REM_ACCENTS)
|
|
|
1932 lp->sl_rem_accents = TRUE;
|
|
375
|
1933 if (i & SAL_SOFO)
|
|
|
1934 lp->sl_sofo = TRUE;
|
|
344
|
1935
|
|
|
1936 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
1937 if (cnt < 0)
|
|
|
1938 goto formerr;
|
|
|
1939
|
|
375
|
1940 if (lp->sl_sofo)
|
|
|
1941 {
|
|
|
1942 /*
|
|
|
1943 * SOFOFROM and SOFOTO items come in one <salfrom> and <salto>
|
|
|
1944 */
|
|
|
1945 if (cnt != 1)
|
|
|
1946 goto formerr;
|
|
|
1947
|
|
|
1948 cnt = (getc(fd) << 8) + getc(fd); /* <salfromlen> */
|
|
|
1949 if (cnt < 0)
|
|
|
1950 goto formerr;
|
|
|
1951 if ((bp = alloc(cnt + 1)) == NULL)
|
|
|
1952 goto endFAIL;
|
|
|
1953 for (i = 0; i < cnt; ++i)
|
|
|
1954 bp[i] = getc(fd); /* <salfrom> */
|
|
|
1955 bp[i] = NUL;
|
|
|
1956
|
|
|
1957 ccnt = (getc(fd) << 8) + getc(fd); /* <saltolen> */
|
|
344
|
1958 if (ccnt < 0)
|
|
375
|
1959 {
|
|
|
1960 vim_free(bp);
|
|
344
|
1961 goto formerr;
|
|
375
|
1962 }
|
|
|
1963 if ((fol = alloc(ccnt + 1)) == NULL)
|
|
|
1964 {
|
|
|
1965 vim_free(bp);
|
|
344
|
1966 goto endFAIL;
|
|
375
|
1967 }
|
|
344
|
1968 for (i = 0; i < ccnt; ++i)
|
|
375
|
1969 fol[i] = getc(fd); /* <salto> */
|
|
|
1970 fol[i] = NUL;
|
|
372
|
1971
|
|
|
1972 #ifdef FEAT_MBYTE
|
|
|
1973 if (has_mbyte)
|
|
|
1974 {
|
|
375
|
1975 char_u *s;
|
|
|
1976
|
|
|
1977 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
1978 * characters. The index is the low byte of the character.
|
|
|
1979 * The list contains from-to pairs with a terminating NUL.
|
|
|
1980 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
1981 gap = &lp->sl_sal;
|
|
|
1982 ga_init2(gap, sizeof(int *), 1);
|
|
|
1983 if (ga_grow(gap, 256) == FAIL)
|
|
|
1984 {
|
|
|
1985 sofoFAIL:
|
|
|
1986 vim_free(bp);
|
|
|
1987 vim_free(fol);
|
|
|
1988 goto endFAIL;
|
|
|
1989 }
|
|
|
1990 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
1991 gap->ga_len = 256;
|
|
|
1992
|
|
|
1993 /* First count the number of items for each list. Temporarily use
|
|
|
1994 * sl_sal_first[] for this. */
|
|
|
1995 for (p = bp, s = fol; *p != NUL && *s != NUL; )
|
|
|
1996 {
|
|
|
1997 c = mb_ptr2char_adv(&p);
|
|
|
1998 mb_ptr_adv(s);
|
|
|
1999 if (c >= 256)
|
|
|
2000 ++lp->sl_sal_first[c & 0xff];
|
|
|
2001 }
|
|
|
2002 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
|
2003 goto sofoerr;
|
|
|
2004
|
|
|
2005 /* Allocate the lists. */
|
|
|
2006 for (i = 0; i < 256; ++i)
|
|
|
2007 if (lp->sl_sal_first[i] > 0)
|
|
|
2008 {
|
|
|
2009 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2010 if (p == NULL)
|
|
|
2011 goto sofoFAIL;
|
|
|
2012 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2013 *(int *)p = 0;
|
|
|
2014 }
|
|
|
2015
|
|
|
2016 /* Put the characters in sl_sal_first[] or a sl_sal list. */
|
|
|
2017 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2018 for (p = bp, s = fol; *p != NUL && *s != NUL; )
|
|
|
2019 {
|
|
|
2020 c = mb_ptr2char_adv(&p);
|
|
|
2021 i = mb_ptr2char_adv(&s);
|
|
|
2022 if (c >= 256)
|
|
|
2023 {
|
|
|
2024 int *inp;
|
|
|
2025
|
|
|
2026 /* Append the from-to chars at the end of the list with
|
|
|
2027 * the low byte. */
|
|
|
2028 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2029 while (*inp != 0)
|
|
|
2030 ++inp;
|
|
|
2031 *inp++ = c; /* from char */
|
|
|
2032 *inp++ = i; /* to char */
|
|
|
2033 *inp++ = NUL; /* NUL at the end */
|
|
|
2034 }
|
|
|
2035 else
|
|
|
2036 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2037 lp->sl_sal_first[c] = i;
|
|
|
2038 }
|
|
|
2039 }
|
|
|
2040 else
|
|
|
2041 #endif
|
|
|
2042 {
|
|
|
2043 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2044 if (cnt != ccnt)
|
|
|
2045 {
|
|
|
2046 #ifdef FEAT_MBYTE
|
|
|
2047 sofoerr:
|
|
|
2048 #endif
|
|
|
2049 vim_free(bp);
|
|
|
2050 vim_free(fol);
|
|
|
2051 goto formerr;
|
|
|
2052 }
|
|
|
2053 for (i = 0; i < cnt; ++i)
|
|
|
2054 lp->sl_sal_first[bp[i]] = fol[i];
|
|
|
2055 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2056 }
|
|
|
2057 vim_free(bp);
|
|
|
2058 vim_free(fol);
|
|
|
2059 }
|
|
|
2060 else
|
|
|
2061 {
|
|
|
2062 /*
|
|
|
2063 * SAL items
|
|
|
2064 */
|
|
|
2065 gap = &lp->sl_sal;
|
|
|
2066 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2067 goto endFAIL;
|
|
|
2068
|
|
|
2069 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2070 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2071 {
|
|
|
2072 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2073 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2074 if (ccnt < 0)
|
|
|
2075 goto formerr;
|
|
|
2076 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2077 goto endFAIL;
|
|
|
2078 smp->sm_lead = p;
|
|
|
2079
|
|
|
2080 /* Read up to the first special char into sm_lead. */
|
|
|
2081 for (i = 0; i < ccnt; ++i)
|
|
|
2082 {
|
|
|
2083 c = getc(fd); /* <salfrom> */
|
|
|
2084 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2085 break;
|
|
|
2086 *p++ = c;
|
|
|
2087 }
|
|
|
2088 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2089 *p++ = NUL;
|
|
|
2090
|
|
|
2091 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2092 if (c == '(')
|
|
|
2093 {
|
|
|
2094 smp->sm_oneof = p;
|
|
|
2095 for (++i; i < ccnt; ++i)
|
|
|
2096 {
|
|
|
2097 c = getc(fd); /* <salfrom> */
|
|
|
2098 if (c == ')')
|
|
|
2099 break;
|
|
|
2100 *p++ = c;
|
|
|
2101 }
|
|
|
2102 *p++ = NUL;
|
|
|
2103 if (++i < ccnt)
|
|
|
2104 c = getc(fd);
|
|
|
2105 }
|
|
372
|
2106 else
|
|
375
|
2107 smp->sm_oneof = NULL;
|
|
|
2108
|
|
|
2109 /* Any following chars go in sm_rules. */
|
|
|
2110 smp->sm_rules = p;
|
|
|
2111 if (i < ccnt)
|
|
|
2112 /* store the char we got while checking for end of sm_lead */
|
|
|
2113 *p++ = c;
|
|
|
2114 for (++i; i < ccnt; ++i)
|
|
|
2115 *p++ = getc(fd); /* <salfrom> */
|
|
|
2116 *p++ = NUL;
|
|
|
2117
|
|
|
2118 ccnt = getc(fd); /* <saltolen> */
|
|
|
2119 if (ccnt < 0)
|
|
372
|
2120 {
|
|
|
2121 vim_free(smp->sm_lead);
|
|
375
|
2122 goto formerr;
|
|
|
2123 }
|
|
|
2124 if ((p = alloc(ccnt + 1)) == NULL)
|
|
|
2125 {
|
|
|
2126 vim_free(smp->sm_lead);
|
|
372
|
2127 goto endFAIL;
|
|
|
2128 }
|
|
375
|
2129 smp->sm_to = p;
|
|
|
2130
|
|
|
2131 for (i = 0; i < ccnt; ++i)
|
|
|
2132 *p++ = getc(fd); /* <salto> */
|
|
|
2133 *p++ = NUL;
|
|
|
2134
|
|
372
|
2135 #ifdef FEAT_MBYTE
|
|
|
2136 if (has_mbyte)
|
|
|
2137 {
|
|
375
|
2138 /* convert the multi-byte strings to wide char strings */
|
|
|
2139 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2140 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2141 if (smp->sm_oneof == NULL)
|
|
|
2142 smp->sm_oneof_w = NULL;
|
|
|
2143 else
|
|
|
2144 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
2145 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
2146 if (smp->sm_lead_w == NULL
|
|
|
2147 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
2148 || smp->sm_to_w == NULL)
|
|
372
|
2149 {
|
|
375
|
2150 vim_free(smp->sm_lead);
|
|
|
2151 vim_free(smp->sm_to);
|
|
|
2152 vim_free(smp->sm_lead_w);
|
|
|
2153 vim_free(smp->sm_oneof_w);
|
|
|
2154 vim_free(smp->sm_to_w);
|
|
|
2155 goto endFAIL;
|
|
372
|
2156 }
|
|
|
2157 }
|
|
|
2158 #endif
|
|
|
2159 }
|
|
375
|
2160
|
|
|
2161 /* Fill the first-index table. */
|
|
|
2162 sfirst = lp->sl_sal_first;
|
|
|
2163 for (i = 0; i < 256; ++i)
|
|
|
2164 sfirst[i] = -1;
|
|
|
2165 smp = (salitem_T *)gap->ga_data;
|
|
|
2166 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2167 {
|
|
|
2168 #ifdef FEAT_MBYTE
|
|
|
2169 if (has_mbyte)
|
|
|
2170 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2171 * the character, for other encodings it should differ for most
|
|
|
2172 * characters. */
|
|
|
2173 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2174 else
|
|
|
2175 #endif
|
|
|
2176 c = *smp[i].sm_lead;
|
|
|
2177 if (sfirst[c] == -1)
|
|
|
2178 {
|
|
|
2179 sfirst[c] = i;
|
|
|
2180 #ifdef FEAT_MBYTE
|
|
|
2181 if (has_mbyte)
|
|
|
2182 {
|
|
|
2183 /* Make sure all entries with this byte are following each
|
|
|
2184 * other. Move the ones that are in the wrong position. Do
|
|
|
2185 * keep the same ordering! */
|
|
|
2186 while (i + 1 < gap->ga_len
|
|
|
2187 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
2188 /* Skip over entry with same index byte. */
|
|
|
2189 ++i;
|
|
|
2190
|
|
|
2191 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
2192 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
2193 {
|
|
|
2194 salitem_T tsal;
|
|
|
2195
|
|
|
2196 /* Move entry with same index byte after the entries
|
|
|
2197 * we already found. */
|
|
|
2198 ++i;
|
|
|
2199 --n;
|
|
|
2200 tsal = smp[i + n];
|
|
|
2201 mch_memmove(smp + i + 1, smp + i,
|
|
|
2202 sizeof(salitem_T) * n);
|
|
|
2203 smp[i] = tsal;
|
|
|
2204 }
|
|
|
2205 }
|
|
|
2206 #endif
|
|
|
2207 }
|
|
|
2208 }
|
|
323
|
2209 }
|
|
|
2210
|
|
|
2211 cnt = (getc(fd) << 8) + getc(fd); /* <maplen> */
|
|
|
2212 if (cnt < 0)
|
|
|
2213 goto formerr;
|
|
|
2214 p = alloc(cnt + 1);
|
|
|
2215 if (p == NULL)
|
|
|
2216 goto endFAIL;
|
|
|
2217 for (i = 0; i < cnt; ++i)
|
|
|
2218 p[i] = getc(fd); /* <mapstr> */
|
|
|
2219 p[i] = NUL;
|
|
330
|
2220 set_map_str(lp, p);
|
|
|
2221 vim_free(p);
|
|
323
|
2222
|
|
236
|
2223
|
|
300
|
2224 /* round 1: <LWORDTREE>
|
|
339
|
2225 * round 2: <KWORDTREE>
|
|
|
2226 * round 3: <PREFIXTREE> */
|
|
|
2227 for (round = 1; round <= 3; ++round)
|
|
236
|
2228 {
|
|
300
|
2229 /* The tree size was computed when writing the file, so that we can
|
|
|
2230 * allocate it as one long block. <nodecount> */
|
|
|
2231 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2232 if (len < 0)
|
|
|
2233 goto truncerr;
|
|
|
2234 if (len > 0)
|
|
236
|
2235 {
|
|
300
|
2236 /* Allocate the byte array. */
|
|
339
|
2237 bp = lalloc((long_u)len, TRUE);
|
|
|
2238 if (bp == NULL)
|
|
300
|
2239 goto endFAIL;
|
|
|
2240 if (round == 1)
|
|
339
|
2241 lp->sl_fbyts = bp;
|
|
|
2242 else if (round == 2)
|
|
|
2243 lp->sl_kbyts = bp;
|
|
300
|
2244 else
|
|
339
|
2245 lp->sl_pbyts = bp;
|
|
236
|
2246
|
|
300
|
2247 /* Allocate the index array. */
|
|
339
|
2248 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2249 if (ip == NULL)
|
|
300
|
2250 goto endFAIL;
|
|
|
2251 if (round == 1)
|
|
339
|
2252 lp->sl_fidxs = ip;
|
|
|
2253 else if (round == 2)
|
|
|
2254 lp->sl_kidxs = ip;
|
|
300
|
2255 else
|
|
339
|
2256 lp->sl_pidxs = ip;
|
|
300
|
2257
|
|
|
2258 /* Read the tree and store it in the array. */
|
|
339
|
2259 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2260 if (idx == -1)
|
|
300
|
2261 goto truncerr;
|
|
324
|
2262 if (idx < 0)
|
|
236
|
2263 goto formerr;
|
|
|
2264 }
|
|
300
|
2265 }
|
|
243
|
2266
|
|
310
|
2267 /* For a new file link it in the list of spell files. */
|
|
|
2268 if (old_lp == NULL)
|
|
|
2269 {
|
|
|
2270 lp->sl_next = first_lang;
|
|
|
2271 first_lang = lp;
|
|
|
2272 }
|
|
307
|
2273
|
|
255
|
2274 goto endOK;
|
|
|
2275
|
|
|
2276 endFAIL:
|
|
310
|
2277 if (lang != NULL)
|
|
|
2278 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2279 *lang = NUL;
|
|
|
2280 if (lp != NULL && old_lp == NULL)
|
|
323
|
2281 {
|
|
307
|
2282 slang_free(lp);
|
|
323
|
2283 lp = NULL;
|
|
|
2284 }
|
|
255
|
2285
|
|
|
2286 endOK:
|
|
236
|
2287 if (fd != NULL)
|
|
|
2288 fclose(fd);
|
|
|
2289 sourcing_name = save_sourcing_name;
|
|
|
2290 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2291
|
|
|
2292 return lp;
|
|
236
|
2293 }
|
|
|
2294
|
|
372
|
2295 #ifdef FEAT_MBYTE
|
|
|
2296 /*
|
|
|
2297 * Turn a multi-byte string into a wide character string.
|
|
|
2298 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
2299 */
|
|
|
2300 static int *
|
|
|
2301 mb_str2wide(s)
|
|
|
2302 char_u *s;
|
|
|
2303 {
|
|
|
2304 int *res;
|
|
|
2305 char_u *p;
|
|
|
2306 int i = 0;
|
|
|
2307
|
|
|
2308 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
2309 if (res != NULL)
|
|
|
2310 {
|
|
|
2311 for (p = s; *p != NUL; )
|
|
|
2312 res[i++] = mb_ptr2char_adv(&p);
|
|
|
2313 res[i] = NUL;
|
|
|
2314 }
|
|
|
2315 return res;
|
|
|
2316 }
|
|
|
2317 #endif
|
|
|
2318
|
|
236
|
2319 /*
|
|
300
|
2320 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
2321 * "byts" and index array "idxs". Recursively read the children.
|
|
|
2322 *
|
|
346
|
2323 * NOTE: The code here must match put_node().
|
|
300
|
2324 *
|
|
|
2325 * Returns the index follosing the siblings.
|
|
|
2326 * Returns -1 if the file is shorter than expected.
|
|
|
2327 * Returns -2 if there is a format error.
|
|
236
|
2328 */
|
|
324
|
2329 static idx_T
|
|
339
|
2330 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
2331 FILE *fd;
|
|
|
2332 char_u *byts;
|
|
324
|
2333 idx_T *idxs;
|
|
300
|
2334 int maxidx; /* size of arrays */
|
|
324
|
2335 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
2336 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
2337 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
2338 {
|
|
300
|
2339 int len;
|
|
|
2340 int i;
|
|
|
2341 int n;
|
|
324
|
2342 idx_T idx = startidx;
|
|
300
|
2343 int c;
|
|
366
|
2344 int c2;
|
|
300
|
2345 #define SHARED_MASK 0x8000000
|
|
236
|
2346
|
|
300
|
2347 len = getc(fd); /* <siblingcount> */
|
|
|
2348 if (len <= 0)
|
|
|
2349 return -1;
|
|
|
2350
|
|
|
2351 if (startidx + len >= maxidx)
|
|
|
2352 return -2;
|
|
|
2353 byts[idx++] = len;
|
|
|
2354
|
|
|
2355 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
2356 for (i = 1; i <= len; ++i)
|
|
236
|
2357 {
|
|
300
|
2358 c = getc(fd); /* <byte> */
|
|
|
2359 if (c < 0)
|
|
|
2360 return -1;
|
|
|
2361 if (c <= BY_SPECIAL)
|
|
|
2362 {
|
|
366
|
2363 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
2364 {
|
|
|
2365 /* No flags, all regions. */
|
|
|
2366 idxs[idx] = 0;
|
|
|
2367 c = 0;
|
|
|
2368 }
|
|
366
|
2369 else if (c == BY_FLAGS || c == BY_NOFLAGS)
|
|
300
|
2370 {
|
|
339
|
2371 if (prefixtree)
|
|
|
2372 {
|
|
|
2373 /* Read the prefix ID and the condition nr. In idxs[]
|
|
|
2374 * store the prefix ID in the low byte, the condition
|
|
|
2375 * index shifted up 8 bits. */
|
|
366
|
2376 c2 = getc(fd); /* <prefixID> */
|
|
339
|
2377 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
2378 if (n >= maxprefcondnr)
|
|
|
2379 return -2;
|
|
366
|
2380 c2 += (n << 8);
|
|
|
2381 if (c == BY_NOFLAGS)
|
|
|
2382 c = c2;
|
|
|
2383 else
|
|
|
2384 c = c2 | WF_RAREPFX;
|
|
339
|
2385 }
|
|
|
2386 else
|
|
|
2387 {
|
|
|
2388 /* Read flags and optional region and prefix ID. In
|
|
|
2389 * idxs[] the flags go in the low byte, region above that
|
|
|
2390 * and prefix ID above the region. */
|
|
|
2391 c = getc(fd); /* <flags> */
|
|
|
2392 if (c & WF_REGION)
|
|
|
2393 c = (getc(fd) << 8) + c; /* <region> */
|
|
|
2394 if (c & WF_PFX)
|
|
|
2395 c = (getc(fd) << 16) + c; /* <prefixID> */
|
|
|
2396 }
|
|
|
2397
|
|
300
|
2398 idxs[idx] = c;
|
|
|
2399 c = 0;
|
|
|
2400 }
|
|
|
2401 else /* c == BY_INDEX */
|
|
|
2402 {
|
|
|
2403 /* <nodeidx> */
|
|
|
2404 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2405 if (n < 0 || n >= maxidx)
|
|
|
2406 return -2;
|
|
|
2407 idxs[idx] = n + SHARED_MASK;
|
|
|
2408 c = getc(fd); /* <xbyte> */
|
|
|
2409 }
|
|
|
2410 }
|
|
|
2411 byts[idx++] = c;
|
|
236
|
2412 }
|
|
|
2413
|
|
300
|
2414 /* Recursively read the children for non-shared siblings.
|
|
|
2415 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
2416 * remove SHARED_MASK) */
|
|
|
2417 for (i = 1; i <= len; ++i)
|
|
|
2418 if (byts[startidx + i] != 0)
|
|
|
2419 {
|
|
|
2420 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
2421 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
2422 else
|
|
|
2423 {
|
|
|
2424 idxs[startidx + i] = idx;
|
|
339
|
2425 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
2426 prefixtree, maxprefcondnr);
|
|
300
|
2427 if (idx < 0)
|
|
|
2428 break;
|
|
|
2429 }
|
|
|
2430 }
|
|
236
|
2431
|
|
300
|
2432 return idx;
|
|
236
|
2433 }
|
|
|
2434
|
|
|
2435 /*
|
|
|
2436 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
2437 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
2438 */
|
|
|
2439 char_u *
|
|
|
2440 did_set_spelllang(buf)
|
|
|
2441 buf_T *buf;
|
|
|
2442 {
|
|
|
2443 garray_T ga;
|
|
351
|
2444 char_u *splp;
|
|
236
|
2445 char_u *region;
|
|
355
|
2446 int filename;
|
|
236
|
2447 int region_mask;
|
|
|
2448 slang_T *lp;
|
|
|
2449 int c;
|
|
351
|
2450 char_u lang[MAXWLEN + 1];
|
|
323
|
2451 char_u spf_name[MAXPATHL];
|
|
351
|
2452 int load_spf;
|
|
|
2453 int len;
|
|
|
2454 char_u *p;
|
|
236
|
2455
|
|
|
2456 ga_init2(&ga, sizeof(langp_T), 2);
|
|
|
2457
|
|
351
|
2458 /* Make the name of the .spl file associated with 'spellfile'. */
|
|
323
|
2459 if (*buf->b_p_spf == NUL)
|
|
351
|
2460 load_spf = FALSE;
|
|
323
|
2461 else
|
|
351
|
2462 {
|
|
323
|
2463 vim_snprintf((char *)spf_name, sizeof(spf_name), "%s.spl",
|
|
|
2464 buf->b_p_spf);
|
|
351
|
2465 load_spf = TRUE;
|
|
|
2466 }
|
|
|
2467
|
|
|
2468 /* loop over comma separated language names. */
|
|
|
2469 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
2470 {
|
|
|
2471 /* Get one language name. */
|
|
|
2472 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
2473
|
|
240
|
2474 region = NULL;
|
|
351
|
2475 len = STRLEN(lang);
|
|
355
|
2476
|
|
|
2477 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
2478 * If there is a region name let "region" point to it and remove it
|
|
|
2479 * from the name. */
|
|
|
2480 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
2481 {
|
|
355
|
2482 filename = TRUE;
|
|
|
2483
|
|
|
2484 /* Check if we loaded this language before. */
|
|
|
2485 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2486 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2487 break;
|
|
236
|
2488 }
|
|
355
|
2489 else
|
|
|
2490 {
|
|
|
2491 filename = FALSE;
|
|
|
2492 if (len > 3 && lang[len - 3] == '_')
|
|
|
2493 {
|
|
|
2494 region = lang + len - 2;
|
|
|
2495 len -= 3;
|
|
|
2496 lang[len] = NUL;
|
|
|
2497 }
|
|
|
2498
|
|
|
2499 /* Check if we loaded this language before. */
|
|
|
2500 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2501 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
2502 break;
|
|
|
2503 }
|
|
236
|
2504
|
|
351
|
2505 /* If not found try loading the language now. */
|
|
236
|
2506 if (lp == NULL)
|
|
355
|
2507 {
|
|
|
2508 if (filename)
|
|
|
2509 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
2510 else
|
|
|
2511 spell_load_lang(lang);
|
|
|
2512 }
|
|
236
|
2513
|
|
307
|
2514 /*
|
|
351
|
2515 * Loop over the languages, there can be several files for "lang".
|
|
307
|
2516 */
|
|
|
2517 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
2518 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
2519 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
2520 {
|
|
316
|
2521 region_mask = REGION_ALL;
|
|
355
|
2522 if (!filename && region != NULL)
|
|
236
|
2523 {
|
|
307
|
2524 /* find region in sl_regions */
|
|
|
2525 c = find_region(lp->sl_regions, region);
|
|
|
2526 if (c == REGION_ALL)
|
|
|
2527 {
|
|
316
|
2528 if (!lp->sl_add)
|
|
351
|
2529 smsg((char_u *)
|
|
|
2530 _("Warning: region %s not supported"),
|
|
|
2531 region);
|
|
307
|
2532 }
|
|
|
2533 else
|
|
|
2534 region_mask = 1 << c;
|
|
236
|
2535 }
|
|
307
|
2536
|
|
|
2537 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2538 {
|
|
|
2539 ga_clear(&ga);
|
|
|
2540 return e_outofmem;
|
|
|
2541 }
|
|
|
2542 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2543 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2544 ++ga.ga_len;
|
|
323
|
2545
|
|
351
|
2546 /* Check if this is the spell file related to 'spellfile'. */
|
|
|
2547 if (load_spf && fullpathcmp(spf_name, lp->sl_fname, FALSE)
|
|
|
2548 == FPC_SAME)
|
|
|
2549 load_spf = FALSE;
|
|
236
|
2550 }
|
|
|
2551 }
|
|
|
2552
|
|
323
|
2553 /*
|
|
|
2554 * Make sure the 'spellfile' file is loaded. It may be in 'runtimepath',
|
|
|
2555 * then it's probably loaded above already. Otherwise load it here.
|
|
|
2556 */
|
|
351
|
2557 if (load_spf)
|
|
|
2558 {
|
|
|
2559 /* Check if it was loaded already. */
|
|
323
|
2560 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2561 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2562 break;
|
|
|
2563 if (lp == NULL)
|
|
|
2564 {
|
|
351
|
2565 /* Not loaded, try loading it now. The language name includes the
|
|
|
2566 * region name, the region is ignored otherwise. */
|
|
|
2567 vim_strncpy(lang, gettail(buf->b_p_spf), MAXWLEN);
|
|
|
2568 p = vim_strchr(lang, '.');
|
|
|
2569 if (p != NULL)
|
|
|
2570 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
2571 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
2572 }
|
|
|
2573 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
2574 {
|
|
|
2575 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2576 LANGP_ENTRY(ga, ga.ga_len)->lp_region = REGION_ALL;
|
|
|
2577 ++ga.ga_len;
|
|
|
2578 }
|
|
|
2579 }
|
|
|
2580
|
|
236
|
2581 /* Add a NULL entry to mark the end of the list. */
|
|
|
2582 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2583 {
|
|
|
2584 ga_clear(&ga);
|
|
|
2585 return e_outofmem;
|
|
|
2586 }
|
|
|
2587 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
2588 ++ga.ga_len;
|
|
|
2589
|
|
|
2590 /* Everything is fine, store the new b_langp value. */
|
|
|
2591 ga_clear(&buf->b_langp);
|
|
|
2592 buf->b_langp = ga;
|
|
|
2593
|
|
|
2594 return NULL;
|
|
|
2595 }
|
|
|
2596
|
|
|
2597 /*
|
|
|
2598 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
2599 * Each region is simply stored as the two characters of it's name.
|
|
|
2600 * Returns the index if found, REGION_ALL if not found.
|
|
|
2601 */
|
|
|
2602 static int
|
|
|
2603 find_region(rp, region)
|
|
|
2604 char_u *rp;
|
|
|
2605 char_u *region;
|
|
|
2606 {
|
|
|
2607 int i;
|
|
|
2608
|
|
|
2609 for (i = 0; ; i += 2)
|
|
|
2610 {
|
|
|
2611 if (rp[i] == NUL)
|
|
|
2612 return REGION_ALL;
|
|
|
2613 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
2614 break;
|
|
|
2615 }
|
|
|
2616 return i / 2;
|
|
|
2617 }
|
|
|
2618
|
|
|
2619 /*
|
|
323
|
2620 * Return case type of word:
|
|
236
|
2621 * w word 0
|
|
300
|
2622 * Word WF_ONECAP
|
|
|
2623 * W WORD WF_ALLCAP
|
|
|
2624 * WoRd wOrd WF_KEEPCAP
|
|
236
|
2625 */
|
|
|
2626 static int
|
|
|
2627 captype(word, end)
|
|
|
2628 char_u *word;
|
|
323
|
2629 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
2630 {
|
|
|
2631 char_u *p;
|
|
|
2632 int c;
|
|
|
2633 int firstcap;
|
|
|
2634 int allcap;
|
|
|
2635 int past_second = FALSE; /* past second word char */
|
|
|
2636
|
|
|
2637 /* find first letter */
|
|
358
|
2638 for (p = word; !spell_iswordp(p); mb_ptr_adv(p))
|
|
323
|
2639 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
2640 return 0; /* only non-word characters, illegal word */
|
|
|
2641 #ifdef FEAT_MBYTE
|
|
310
|
2642 if (has_mbyte)
|
|
|
2643 c = mb_ptr2char_adv(&p);
|
|
|
2644 else
|
|
236
|
2645 #endif
|
|
310
|
2646 c = *p++;
|
|
324
|
2647 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
2648
|
|
|
2649 /*
|
|
|
2650 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
2651 * But a word with an upper char only at start is a ONECAP.
|
|
|
2652 */
|
|
323
|
2653 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
358
|
2654 if (spell_iswordp(p))
|
|
236
|
2655 {
|
|
|
2656 #ifdef FEAT_MBYTE
|
|
|
2657 c = mb_ptr2char(p);
|
|
|
2658 #else
|
|
|
2659 c = *p;
|
|
|
2660 #endif
|
|
324
|
2661 if (!SPELL_ISUPPER(c))
|
|
236
|
2662 {
|
|
|
2663 /* UUl -> KEEPCAP */
|
|
|
2664 if (past_second && allcap)
|
|
300
|
2665 return WF_KEEPCAP;
|
|
236
|
2666 allcap = FALSE;
|
|
|
2667 }
|
|
|
2668 else if (!allcap)
|
|
|
2669 /* UlU -> KEEPCAP */
|
|
300
|
2670 return WF_KEEPCAP;
|
|
236
|
2671 past_second = TRUE;
|
|
|
2672 }
|
|
|
2673
|
|
|
2674 if (allcap)
|
|
300
|
2675 return WF_ALLCAP;
|
|
236
|
2676 if (firstcap)
|
|
300
|
2677 return WF_ONECAP;
|
|
236
|
2678 return 0;
|
|
|
2679 }
|
|
|
2680
|
|
355
|
2681 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
2682 /*
|
|
|
2683 * Free all languages.
|
|
|
2684 */
|
|
|
2685 void
|
|
|
2686 spell_free_all()
|
|
|
2687 {
|
|
|
2688 slang_T *lp;
|
|
|
2689 buf_T *buf;
|
|
|
2690
|
|
|
2691 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2692 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2693 ga_clear(&buf->b_langp);
|
|
|
2694
|
|
|
2695 while (first_lang != NULL)
|
|
|
2696 {
|
|
|
2697 lp = first_lang;
|
|
|
2698 first_lang = lp->sl_next;
|
|
|
2699 slang_free(lp);
|
|
|
2700 }
|
|
366
|
2701
|
|
|
2702 init_spell_chartab();
|
|
355
|
2703 }
|
|
|
2704 # endif
|
|
|
2705
|
|
236
|
2706 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
2707 /*
|
|
|
2708 * Clear all spelling tables and reload them.
|
|
307
|
2709 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
2710 */
|
|
|
2711 void
|
|
|
2712 spell_reload()
|
|
|
2713 {
|
|
|
2714 buf_T *buf;
|
|
316
|
2715 win_T *wp;
|
|
236
|
2716
|
|
358
|
2717 /* Initialize the table for spell_iswordp(). */
|
|
236
|
2718 init_spell_chartab();
|
|
|
2719
|
|
|
2720 /* Unload all allocated memory. */
|
|
355
|
2721 spell_free_all();
|
|
236
|
2722
|
|
|
2723 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2724 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2725 {
|
|
316
|
2726 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
2727 * window for this buffer in which 'spell' is set. */
|
|
236
|
2728 if (*buf->b_p_spl != NUL)
|
|
316
|
2729 {
|
|
|
2730 FOR_ALL_WINDOWS(wp)
|
|
|
2731 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
2732 {
|
|
|
2733 (void)did_set_spelllang(buf);
|
|
|
2734 # ifdef FEAT_WINDOWS
|
|
|
2735 break;
|
|
|
2736 # endif
|
|
|
2737 }
|
|
|
2738 }
|
|
236
|
2739 }
|
|
|
2740 }
|
|
|
2741 # endif
|
|
|
2742
|
|
310
|
2743 /*
|
|
|
2744 * Reload the spell file "fname" if it's loaded.
|
|
|
2745 */
|
|
|
2746 static void
|
|
323
|
2747 spell_reload_one(fname, added_word)
|
|
310
|
2748 char_u *fname;
|
|
323
|
2749 int added_word; /* invoked through "zg" */
|
|
310
|
2750 {
|
|
|
2751 slang_T *lp;
|
|
323
|
2752 int didit = FALSE;
|
|
310
|
2753
|
|
|
2754 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2755 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2756 {
|
|
|
2757 slang_clear(lp);
|
|
323
|
2758 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
2759 redraw_all_later(NOT_VALID);
|
|
323
|
2760 didit = TRUE;
|
|
310
|
2761 }
|
|
323
|
2762
|
|
|
2763 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
2764 * 'spelllang' to get it loaded. */
|
|
|
2765 if (added_word && !didit)
|
|
|
2766 did_set_spelllang(curbuf);
|
|
310
|
2767 }
|
|
|
2768
|
|
|
2769
|
|
236
|
2770 /*
|
|
|
2771 * Functions for ":mkspell".
|
|
|
2772 */
|
|
|
2773
|
|
300
|
2774 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
2775 and .dic file. */
|
|
|
2776 /*
|
|
|
2777 * Main structure to store the contents of a ".aff" file.
|
|
|
2778 */
|
|
|
2779 typedef struct afffile_S
|
|
|
2780 {
|
|
|
2781 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
310
|
2782 int af_rar; /* RAR ID for rare word */
|
|
|
2783 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
2784 int af_bad; /* BAD ID for banned word */
|
|
339
|
2785 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
2786 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
2787 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
2788 } afffile_T;
|
|
|
2789
|
|
|
2790 typedef struct affentry_S affentry_T;
|
|
|
2791 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
2792 struct affentry_S
|
|
|
2793 {
|
|
|
2794 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
2795 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
2796 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
2797 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
2798 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
366
|
2799 int ae_rare; /* rare affix */
|
|
300
|
2800 };
|
|
|
2801
|
|
|
2802 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
2803 typedef struct affheader_S
|
|
|
2804 {
|
|
|
2805 char_u ah_key[2]; /* key for hashtable == name of affix entry */
|
|
339
|
2806 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
2807 int ah_combine; /* suffix may combine with prefix */
|
|
|
2808 affentry_T *ah_first; /* first affix entry */
|
|
|
2809 } affheader_T;
|
|
|
2810
|
|
|
2811 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
2812
|
|
|
2813 /*
|
|
|
2814 * Structure that is used to store the items in the word tree. This avoids
|
|
|
2815 * the need to keep track of each allocated thing, it's freed all at once
|
|
|
2816 * after ":mkspell" is done.
|
|
|
2817 */
|
|
|
2818 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
2819 typedef struct sblock_S sblock_T;
|
|
|
2820 struct sblock_S
|
|
|
2821 {
|
|
|
2822 sblock_T *sb_next; /* next block in list */
|
|
|
2823 int sb_used; /* nr of bytes already in use */
|
|
|
2824 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
2825 };
|
|
|
2826
|
|
|
2827 /*
|
|
300
|
2828 * A node in the tree.
|
|
236
|
2829 */
|
|
300
|
2830 typedef struct wordnode_S wordnode_T;
|
|
|
2831 struct wordnode_S
|
|
236
|
2832 {
|
|
346
|
2833 union /* shared to save space */
|
|
|
2834 {
|
|
|
2835 char_u hashkey[6]; /* room for the hash key */
|
|
|
2836 int index; /* index in written nodes (valid after first
|
|
|
2837 round) */
|
|
|
2838 } wn_u1;
|
|
|
2839 union /* shared to save space */
|
|
|
2840 {
|
|
|
2841 wordnode_T *next; /* next node with same hash key */
|
|
|
2842 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
2843 } wn_u2;
|
|
300
|
2844 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
2845 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
2846 always sorted) */
|
|
|
2847 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
|
2848 char_u wn_flags; /* when wn_byte is NUL: WF_ flags */
|
|
339
|
2849 short wn_region; /* when wn_byte is NUL: region mask; for
|
|
|
2850 PREFIXTREE it's the prefcondnr */
|
|
|
2851 char_u wn_prefixID; /* supported/required prefix ID or 0 */
|
|
236
|
2852 };
|
|
|
2853
|
|
300
|
2854 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
2855
|
|
300
|
2856 /*
|
|
|
2857 * Info used while reading the spell files.
|
|
|
2858 */
|
|
|
2859 typedef struct spellinfo_S
|
|
249
|
2860 {
|
|
300
|
2861 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
2862 long si_foldwcount; /* nr of words in si_foldroot */
|
|
300
|
2863 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
2864 long si_keepwcount; /* nr of words in si_keeproot */
|
|
339
|
2865 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
300
|
2866 sblock_T *si_blocks; /* memory blocks used */
|
|
|
2867 int si_ascii; /* handling only ASCII words */
|
|
310
|
2868 int si_add; /* addition file */
|
|
351
|
2869 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
2870 int si_region; /* region mask */
|
|
|
2871 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
2872 int si_memtot; /* runtime memory used */
|
|
310
|
2873 int si_verbose; /* verbose messages */
|
|
316
|
2874 int si_region_count; /* number of regions supported (1 when there
|
|
|
2875 are no regions) */
|
|
|
2876 char_u si_region_name[16]; /* region names (if count > 1) */
|
|
323
|
2877
|
|
|
2878 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
2879 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
2880 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
2881 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
2882 int si_followup; /* soundsalike: ? */
|
|
|
2883 int si_collapse; /* soundsalike: ? */
|
|
|
2884 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
2885 garray_T si_map; /* MAP info concatenated */
|
|
366
|
2886 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
339
|
2887 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
2888 * prefixes, each stored as a string */
|
|
|
2889 int si_newID; /* current value for ah_newID */
|
|
300
|
2890 } spellinfo_T;
|
|
249
|
2891
|
|
300
|
2892 static afffile_T *spell_read_aff __ARGS((char_u *fname, spellinfo_T *spin));
|
|
339
|
2893 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
2894 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
2895 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
2896 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
2897 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
|
2898 static int spell_read_dic __ARGS((char_u *fname, spellinfo_T *spin, afffile_T *affile));
|
|
339
|
2899 static char_u *get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, sblock_T **blp));
|
|
|
2900 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
|
2901 static int spell_read_wordfile __ARGS((char_u *fname, spellinfo_T *spin));
|
|
|
2902 static void *getroom __ARGS((sblock_T **blp, size_t len));
|
|
|
2903 static char_u *getroom_save __ARGS((sblock_T **blp, char_u *s));
|
|
|
2904 static void free_blocks __ARGS((sblock_T *bl));
|
|
|
2905 static wordnode_T *wordtree_alloc __ARGS((sblock_T **blp));
|
|
339
|
2906 static int store_word __ARGS((char_u *word, spellinfo_T *spin, int flags, int region, char_u *pfxlist));
|
|
|
2907 static int tree_add_word __ARGS((char_u *word, wordnode_T *tree, int flags, int region, int prefixID, sblock_T **blp));
|
|
310
|
2908 static void wordtree_compress __ARGS((wordnode_T *root, spellinfo_T *spin));
|
|
300
|
2909 static int node_compress __ARGS((wordnode_T *node, hashtab_T *ht, int *tot));
|
|
|
2910 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
316
|
2911 static void write_vim_spell __ARGS((char_u *fname, spellinfo_T *spin));
|
|
346
|
2912 static void clear_node __ARGS((wordnode_T *node));
|
|
|
2913 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
2914 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
2915 static void init_spellfile __ARGS((void));
|
|
236
|
2916
|
|
|
2917 /*
|
|
323
|
2918 * Read the affix file "fname".
|
|
316
|
2919 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
2920 */
|
|
|
2921 static afffile_T *
|
|
300
|
2922 spell_read_aff(fname, spin)
|
|
236
|
2923 char_u *fname;
|
|
300
|
2924 spellinfo_T *spin;
|
|
236
|
2925 {
|
|
|
2926 FILE *fd;
|
|
|
2927 afffile_T *aff;
|
|
|
2928 char_u rline[MAXLINELEN];
|
|
|
2929 char_u *line;
|
|
|
2930 char_u *pc = NULL;
|
|
334
|
2931 #define MAXITEMCNT 7
|
|
|
2932 char_u *(items[MAXITEMCNT]);
|
|
236
|
2933 int itemcnt;
|
|
|
2934 char_u *p;
|
|
|
2935 int lnum = 0;
|
|
|
2936 affheader_T *cur_aff = NULL;
|
|
|
2937 int aff_todo = 0;
|
|
|
2938 hashtab_T *tp;
|
|
255
|
2939 char_u *low = NULL;
|
|
|
2940 char_u *fol = NULL;
|
|
|
2941 char_u *upp = NULL;
|
|
307
|
2942 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
2943 int do_rep;
|
|
|
2944 int do_sal;
|
|
|
2945 int do_map;
|
|
366
|
2946 int do_midword;
|
|
375
|
2947 int do_sofo;
|
|
323
|
2948 int found_map = FALSE;
|
|
324
|
2949 hashitem_T *hi;
|
|
236
|
2950
|
|
300
|
2951 /*
|
|
|
2952 * Open the file.
|
|
|
2953 */
|
|
310
|
2954 fd = mch_fopen((char *)fname, "r");
|
|
236
|
2955 if (fd == NULL)
|
|
|
2956 {
|
|
|
2957 EMSG2(_(e_notopen), fname);
|
|
|
2958 return NULL;
|
|
|
2959 }
|
|
|
2960
|
|
310
|
2961 if (spin->si_verbose || p_verbose > 2)
|
|
|
2962 {
|
|
|
2963 if (!spin->si_verbose)
|
|
|
2964 verbose_enter();
|
|
366
|
2965 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
2966 out_flush();
|
|
|
2967 if (!spin->si_verbose)
|
|
|
2968 verbose_leave();
|
|
|
2969 }
|
|
236
|
2970
|
|
323
|
2971 /* Only do REP lines when not done in another .aff file already. */
|
|
|
2972 do_rep = spin->si_rep.ga_len == 0;
|
|
|
2973
|
|
|
2974 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
2975 do_sal = spin->si_sal.ga_len == 0;
|
|
|
2976
|
|
|
2977 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
2978 do_map = spin->si_map.ga_len == 0;
|
|
|
2979
|
|
366
|
2980 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
2981 do_midword = spin->si_midword == NULL;
|
|
|
2982
|
|
375
|
2983 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
2984 do_sofo = spin->si_sofofr == NULL;
|
|
|
2985
|
|
300
|
2986 /*
|
|
|
2987 * Allocate and init the afffile_T structure.
|
|
|
2988 */
|
|
|
2989 aff = (afffile_T *)getroom(&spin->si_blocks, sizeof(afffile_T));
|
|
236
|
2990 if (aff == NULL)
|
|
|
2991 return NULL;
|
|
|
2992 hash_init(&aff->af_pref);
|
|
|
2993 hash_init(&aff->af_suff);
|
|
|
2994
|
|
|
2995 /*
|
|
|
2996 * Read all the lines in the file one by one.
|
|
|
2997 */
|
|
255
|
2998 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
2999 {
|
|
255
|
3000 line_breakcheck();
|
|
236
|
3001 ++lnum;
|
|
|
3002
|
|
|
3003 /* Skip comment lines. */
|
|
|
3004 if (*rline == '#')
|
|
|
3005 continue;
|
|
|
3006
|
|
|
3007 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
3008 vim_free(pc);
|
|
310
|
3009 #ifdef FEAT_MBYTE
|
|
300
|
3010 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3011 {
|
|
300
|
3012 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
3013 if (pc == NULL)
|
|
|
3014 {
|
|
|
3015 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3016 fname, lnum, rline);
|
|
|
3017 continue;
|
|
|
3018 }
|
|
236
|
3019 line = pc;
|
|
|
3020 }
|
|
|
3021 else
|
|
310
|
3022 #endif
|
|
236
|
3023 {
|
|
|
3024 pc = NULL;
|
|
|
3025 line = rline;
|
|
|
3026 }
|
|
|
3027
|
|
|
3028 /* Split the line up in white separated items. Put a NUL after each
|
|
|
3029 * item. */
|
|
|
3030 itemcnt = 0;
|
|
|
3031 for (p = line; ; )
|
|
|
3032 {
|
|
|
3033 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
3034 ++p;
|
|
|
3035 if (*p == NUL)
|
|
|
3036 break;
|
|
334
|
3037 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
3038 break;
|
|
236
|
3039 items[itemcnt++] = p;
|
|
300
|
3040 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
3041 ++p;
|
|
|
3042 if (*p == NUL)
|
|
|
3043 break;
|
|
|
3044 *p++ = NUL;
|
|
|
3045 }
|
|
|
3046
|
|
|
3047 /* Handle non-empty lines. */
|
|
|
3048 if (itemcnt > 0)
|
|
|
3049 {
|
|
|
3050 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
3051 && aff->af_enc == NULL)
|
|
|
3052 {
|
|
310
|
3053 #ifdef FEAT_MBYTE
|
|
300
|
3054 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
3055 aff->af_enc = enc_canonize(items[1]);
|
|
|
3056 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
3057 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
3058 p_enc) == FAIL)
|
|
|
3059 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
3060 fname, aff->af_enc, p_enc);
|
|
375
|
3061 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3062 #else
|
|
|
3063 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3064 #endif
|
|
236
|
3065 }
|
|
366
|
3066 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
3067 {
|
|
|
3068 if (do_midword)
|
|
|
3069 spin->si_midword = vim_strsave(items[1]);
|
|
|
3070 }
|
|
302
|
3071 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
3072 {
|
|
323
|
3073 /* ignored, we always split */
|
|
302
|
3074 }
|
|
323
|
3075 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
3076 {
|
|
323
|
3077 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
3078 }
|
|
307
|
3079 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
3080 && aff->af_rar == 0)
|
|
|
3081 {
|
|
|
3082 aff->af_rar = items[1][0];
|
|
|
3083 if (items[1][1] != NUL)
|
|
|
3084 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3085 }
|
|
310
|
3086 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
3087 && aff->af_kep == 0)
|
|
307
|
3088 {
|
|
310
|
3089 aff->af_kep = items[1][0];
|
|
307
|
3090 if (items[1][1] != NUL)
|
|
|
3091 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3092 }
|
|
346
|
3093 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
3094 && aff->af_bad == 0)
|
|
|
3095 {
|
|
|
3096 aff->af_bad = items[1][0];
|
|
|
3097 if (items[1][1] != NUL)
|
|
|
3098 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3099 }
|
|
339
|
3100 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
3101 {
|
|
|
3102 aff->af_pfxpostpone = TRUE;
|
|
|
3103 }
|
|
236
|
3104 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3105 || STRCMP(items[0], "SFX") == 0)
|
|
|
3106 && aff_todo == 0
|
|
334
|
3107 && itemcnt >= 4)
|
|
236
|
3108 {
|
|
334
|
3109 /* Myspell allows extra text after the item, but that might
|
|
|
3110 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
3111 if (itemcnt > 4 && *items[4] != '#')
|
|
|
3112 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
3113 fname, lnum, items[4]);
|
|
|
3114
|
|
236
|
3115 /* New affix letter. */
|
|
300
|
3116 cur_aff = (affheader_T *)getroom(&spin->si_blocks,
|
|
|
3117 sizeof(affheader_T));
|
|
236
|
3118 if (cur_aff == NULL)
|
|
|
3119 break;
|
|
339
|
3120 cur_aff->ah_key[0] = *items[1]; /* TODO: multi-byte? */
|
|
236
|
3121 cur_aff->ah_key[1] = NUL;
|
|
|
3122 if (items[1][1] != NUL)
|
|
307
|
3123 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
3124 if (*items[2] == 'Y')
|
|
|
3125 cur_aff->ah_combine = TRUE;
|
|
300
|
3126 else if (*items[2] != 'N')
|
|
236
|
3127 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
3128 fname, lnum, items[2]);
|
|
339
|
3129
|
|
236
|
3130 if (*items[0] == 'P')
|
|
339
|
3131 {
|
|
236
|
3132 tp = &aff->af_pref;
|
|
339
|
3133 /* Use a new number in the .spl file later, to be able to
|
|
|
3134 * handle multiple .aff files. */
|
|
|
3135 if (aff->af_pfxpostpone)
|
|
344
|
3136 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
3137 }
|
|
236
|
3138 else
|
|
|
3139 tp = &aff->af_suff;
|
|
300
|
3140 aff_todo = atoi((char *)items[3]);
|
|
324
|
3141 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
3142 if (!HASHITEM_EMPTY(hi))
|
|
300
|
3143 {
|
|
236
|
3144 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
3145 fname, lnum, items[1]);
|
|
300
|
3146 aff_todo = 0;
|
|
|
3147 }
|
|
236
|
3148 else
|
|
|
3149 hash_add(tp, cur_aff->ah_key);
|
|
|
3150 }
|
|
|
3151 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3152 || STRCMP(items[0], "SFX") == 0)
|
|
|
3153 && aff_todo > 0
|
|
|
3154 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
3155 && itemcnt >= 5)
|
|
236
|
3156 {
|
|
|
3157 affentry_T *aff_entry;
|
|
366
|
3158 int rare = FALSE;
|
|
|
3159 int lasti = 5;
|
|
|
3160
|
|
|
3161 /* Check for "rare" after the other info. */
|
|
|
3162 if (itemcnt > 5 && STRICMP(items[5], "rare") == 0)
|
|
|
3163 {
|
|
|
3164 rare = TRUE;
|
|
|
3165 lasti = 6;
|
|
|
3166 }
|
|
236
|
3167
|
|
334
|
3168 /* Myspell allows extra text after the item, but that might
|
|
|
3169 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
3170 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
3171 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
366
|
3172 fname, lnum, items[lasti]);
|
|
334
|
3173
|
|
236
|
3174 /* New item for an affix letter. */
|
|
|
3175 --aff_todo;
|
|
300
|
3176 aff_entry = (affentry_T *)getroom(&spin->si_blocks,
|
|
|
3177 sizeof(affentry_T));
|
|
236
|
3178 if (aff_entry == NULL)
|
|
|
3179 break;
|
|
366
|
3180 aff_entry->ae_rare = rare;
|
|
240
|
3181
|
|
236
|
3182 if (STRCMP(items[2], "0") != 0)
|
|
300
|
3183 aff_entry->ae_chop = getroom_save(&spin->si_blocks,
|
|
|
3184 items[2]);
|
|
236
|
3185 if (STRCMP(items[3], "0") != 0)
|
|
300
|
3186 aff_entry->ae_add = getroom_save(&spin->si_blocks,
|
|
|
3187 items[3]);
|
|
236
|
3188
|
|
300
|
3189 /* Don't use an affix entry with non-ASCII characters when
|
|
|
3190 * "spin->si_ascii" is TRUE. */
|
|
|
3191 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
3192 || has_non_ascii(aff_entry->ae_add)))
|
|
|
3193 {
|
|
|
3194 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
3195 cur_aff->ah_first = aff_entry;
|
|
300
|
3196
|
|
|
3197 if (STRCMP(items[4], ".") != 0)
|
|
|
3198 {
|
|
|
3199 char_u buf[MAXLINELEN];
|
|
|
3200
|
|
|
3201 aff_entry->ae_cond = getroom_save(&spin->si_blocks,
|
|
|
3202 items[4]);
|
|
|
3203 if (*items[0] == 'P')
|
|
|
3204 sprintf((char *)buf, "^%s", items[4]);
|
|
|
3205 else
|
|
|
3206 sprintf((char *)buf, "%s$", items[4]);
|
|
|
3207 aff_entry->ae_prog = vim_regcomp(buf,
|
|
|
3208 RE_MAGIC + RE_STRING);
|
|
|
3209 }
|
|
339
|
3210
|
|
|
3211 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
3212 * for the condition. Use an existing one if possible. */
|
|
|
3213 if (*items[0] == 'P' && aff->af_pfxpostpone
|
|
|
3214 && aff_entry->ae_chop == NULL)
|
|
|
3215 {
|
|
|
3216 int idx;
|
|
|
3217 char_u **pp;
|
|
|
3218
|
|
|
3219 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
3220 --idx)
|
|
|
3221 {
|
|
|
3222 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
3223 if (str_equal(p, aff_entry->ae_cond))
|
|
|
3224 break;
|
|
|
3225 }
|
|
|
3226 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
3227 {
|
|
|
3228 /* Not found, add a new condition. */
|
|
|
3229 idx = spin->si_prefcond.ga_len++;
|
|
|
3230 pp = ((char_u **)spin->si_prefcond.ga_data) + idx;
|
|
|
3231 if (aff_entry->ae_cond == NULL)
|
|
|
3232 *pp = NULL;
|
|
|
3233 else
|
|
|
3234 *pp = getroom_save(&spin->si_blocks,
|
|
|
3235 aff_entry->ae_cond);
|
|
|
3236 }
|
|
|
3237
|
|
|
3238 /* Add the prefix to the prefix tree. */
|
|
|
3239 if (aff_entry->ae_add == NULL)
|
|
|
3240 p = (char_u *)"";
|
|
|
3241 else
|
|
|
3242 p = aff_entry->ae_add;
|
|
366
|
3243 tree_add_word(p, spin->si_prefroot, rare ? -2 : -1,
|
|
|
3244 idx, cur_aff->ah_newID, &spin->si_blocks);
|
|
339
|
3245 }
|
|
240
|
3246 }
|
|
236
|
3247 }
|
|
255
|
3248 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
3249 {
|
|
|
3250 if (fol != NULL)
|
|
|
3251 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
3252 fname, lnum);
|
|
|
3253 else
|
|
|
3254 fol = vim_strsave(items[1]);
|
|
|
3255 }
|
|
|
3256 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
3257 {
|
|
|
3258 if (low != NULL)
|
|
|
3259 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
3260 fname, lnum);
|
|
|
3261 else
|
|
|
3262 low = vim_strsave(items[1]);
|
|
|
3263 }
|
|
|
3264 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
3265 {
|
|
|
3266 if (upp != NULL)
|
|
|
3267 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
3268 fname, lnum);
|
|
|
3269 else
|
|
|
3270 upp = vim_strsave(items[1]);
|
|
|
3271 }
|
|
236
|
3272 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
3273 {
|
|
236
|
3274 /* Ignore REP count */;
|
|
323
|
3275 if (!isdigit(*items[1]))
|
|
|
3276 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
3277 fname, lnum);
|
|
|
3278 }
|
|
236
|
3279 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 3)
|
|
|
3280 {
|
|
|
3281 /* REP item */
|
|
323
|
3282 if (do_rep)
|
|
|
3283 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
3284 }
|
|
|
3285 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
3286 {
|
|
|
3287 /* MAP item or count */
|
|
|
3288 if (!found_map)
|
|
|
3289 {
|
|
|
3290 /* First line contains the count. */
|
|
|
3291 found_map = TRUE;
|
|
|
3292 if (!isdigit(*items[1]))
|
|
|
3293 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
3294 fname, lnum);
|
|
|
3295 }
|
|
|
3296 else if (do_map)
|
|
|
3297 {
|
|
346
|
3298 int c;
|
|
|
3299
|
|
|
3300 /* Check that every character appears only once. */
|
|
|
3301 for (p = items[1]; *p != NUL; )
|
|
|
3302 {
|
|
|
3303 #ifdef FEAT_MBYTE
|
|
|
3304 c = mb_ptr2char_adv(&p);
|
|
|
3305 #else
|
|
|
3306 c = *p++;
|
|
|
3307 #endif
|
|
|
3308 if ((spin->si_map.ga_len > 0
|
|
|
3309 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
3310 != NULL)
|
|
|
3311 || vim_strchr(p, c) != NULL)
|
|
|
3312 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
3313 fname, lnum);
|
|
|
3314 }
|
|
|
3315
|
|
323
|
3316 /* We simply concatenate all the MAP strings, separated by
|
|
|
3317 * slashes. */
|
|
|
3318 ga_concat(&spin->si_map, items[1]);
|
|
|
3319 ga_append(&spin->si_map, '/');
|
|
|
3320 }
|
|
|
3321 }
|
|
|
3322 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
3323 {
|
|
|
3324 if (do_sal)
|
|
|
3325 {
|
|
|
3326 /* SAL item (sounds-a-like)
|
|
|
3327 * Either one of the known keys or a from-to pair. */
|
|
|
3328 if (STRCMP(items[1], "followup") == 0)
|
|
|
3329 spin->si_followup = sal_to_bool(items[2]);
|
|
|
3330 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
3331 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
3332 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
3333 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
3334 else
|
|
|
3335 /* when "to" is "_" it means empty */
|
|
|
3336 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
3337 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
3338 : items[2]);
|
|
|
3339 }
|
|
236
|
3340 }
|
|
375
|
3341 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
3342 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
3343 {
|
|
|
3344 if (do_sofo)
|
|
|
3345 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
3346 }
|
|
|
3347 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
3348 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
3349 {
|
|
|
3350 if (do_sofo)
|
|
|
3351 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
3352 }
|
|
300
|
3353 else
|
|
236
|
3354 smsg((char_u *)_("Unrecognized item in %s line %d: %s"),
|
|
|
3355 fname, lnum, items[0]);
|
|
|
3356 }
|
|
|
3357 }
|
|
|
3358
|
|
375
|
3359 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
3360 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
3361 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
3362 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
3363 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
3364
|
|
255
|
3365 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
3366 {
|
|
351
|
3367 if (spin->si_clear_chartab)
|
|
|
3368 {
|
|
|
3369 /* Clear the char type tables, don't want to use any of the
|
|
|
3370 * currently used spell properties. */
|
|
|
3371 init_spell_chartab();
|
|
|
3372 spin->si_clear_chartab = FALSE;
|
|
|
3373 }
|
|
|
3374
|
|
316
|
3375 /*
|
|
|
3376 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
3377 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
3378 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
3379 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
3380 */
|
|
|
3381 if (!spin->si_ascii
|
|
|
3382 #ifdef FEAT_MBYTE
|
|
|
3383 && !enc_utf8
|
|
|
3384 #endif
|
|
|
3385 )
|
|
260
|
3386 {
|
|
|
3387 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
3388 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
3389 else
|
|
316
|
3390 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
3391 }
|
|
255
|
3392
|
|
|
3393 vim_free(fol);
|
|
|
3394 vim_free(low);
|
|
|
3395 vim_free(upp);
|
|
|
3396 }
|
|
|
3397
|
|
236
|
3398 vim_free(pc);
|
|
|
3399 fclose(fd);
|
|
|
3400 return aff;
|
|
|
3401 }
|
|
|
3402
|
|
|
3403 /*
|
|
339
|
3404 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
3405 * NULL as equal.
|
|
|
3406 */
|
|
|
3407 static int
|
|
|
3408 str_equal(s1, s2)
|
|
|
3409 char_u *s1;
|
|
|
3410 char_u *s2;
|
|
|
3411 {
|
|
|
3412 if (s1 == NULL || s2 == NULL)
|
|
|
3413 return s1 == s2;
|
|
|
3414 return STRCMP(s1, s2) == 0;
|
|
|
3415 }
|
|
|
3416
|
|
|
3417 /*
|
|
323
|
3418 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
3419 * They are stored case-folded.
|
|
|
3420 */
|
|
|
3421 static void
|
|
|
3422 add_fromto(spin, gap, from, to)
|
|
|
3423 spellinfo_T *spin;
|
|
|
3424 garray_T *gap;
|
|
|
3425 char_u *from;
|
|
|
3426 char_u *to;
|
|
|
3427 {
|
|
|
3428 fromto_T *ftp;
|
|
|
3429 char_u word[MAXWLEN];
|
|
|
3430
|
|
|
3431 if (ga_grow(gap, 1) == OK)
|
|
|
3432 {
|
|
|
3433 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
3434 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
|
3435 ftp->ft_from = getroom_save(&spin->si_blocks, word);
|
|
|
3436 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
|
3437 ftp->ft_to = getroom_save(&spin->si_blocks, word);
|
|
|
3438 ++gap->ga_len;
|
|
|
3439 }
|
|
|
3440 }
|
|
|
3441
|
|
|
3442 /*
|
|
|
3443 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
3444 */
|
|
|
3445 static int
|
|
|
3446 sal_to_bool(s)
|
|
|
3447 char_u *s;
|
|
|
3448 {
|
|
|
3449 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
3450 }
|
|
|
3451
|
|
|
3452 /*
|
|
240
|
3453 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
3454 * When "s" is NULL FALSE is returned.
|
|
|
3455 */
|
|
|
3456 static int
|
|
|
3457 has_non_ascii(s)
|
|
|
3458 char_u *s;
|
|
|
3459 {
|
|
|
3460 char_u *p;
|
|
|
3461
|
|
|
3462 if (s != NULL)
|
|
|
3463 for (p = s; *p != NUL; ++p)
|
|
|
3464 if (*p >= 128)
|
|
|
3465 return TRUE;
|
|
|
3466 return FALSE;
|
|
|
3467 }
|
|
|
3468
|
|
|
3469 /*
|
|
236
|
3470 * Free the structure filled by spell_read_aff().
|
|
|
3471 */
|
|
|
3472 static void
|
|
|
3473 spell_free_aff(aff)
|
|
|
3474 afffile_T *aff;
|
|
|
3475 {
|
|
|
3476 hashtab_T *ht;
|
|
|
3477 hashitem_T *hi;
|
|
|
3478 int todo;
|
|
|
3479 affheader_T *ah;
|
|
300
|
3480 affentry_T *ae;
|
|
236
|
3481
|
|
|
3482 vim_free(aff->af_enc);
|
|
|
3483
|
|
339
|
3484 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
3485 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
3486 {
|
|
|
3487 todo = ht->ht_used;
|
|
|
3488 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
3489 {
|
|
|
3490 if (!HASHITEM_EMPTY(hi))
|
|
|
3491 {
|
|
|
3492 --todo;
|
|
|
3493 ah = HI2AH(hi);
|
|
300
|
3494 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
3495 vim_free(ae->ae_prog);
|
|
236
|
3496 }
|
|
|
3497 }
|
|
|
3498 if (ht == &aff->af_suff)
|
|
|
3499 break;
|
|
|
3500 }
|
|
300
|
3501
|
|
236
|
3502 hash_clear(&aff->af_pref);
|
|
|
3503 hash_clear(&aff->af_suff);
|
|
|
3504 }
|
|
|
3505
|
|
|
3506 /*
|
|
300
|
3507 * Read dictionary file "fname".
|
|
236
|
3508 * Returns OK or FAIL;
|
|
|
3509 */
|
|
|
3510 static int
|
|
300
|
3511 spell_read_dic(fname, spin, affile)
|
|
236
|
3512 char_u *fname;
|
|
300
|
3513 spellinfo_T *spin;
|
|
|
3514 afffile_T *affile;
|
|
236
|
3515 {
|
|
300
|
3516 hashtab_T ht;
|
|
236
|
3517 char_u line[MAXLINELEN];
|
|
300
|
3518 char_u *afflist;
|
|
339
|
3519 char_u *pfxlist;
|
|
300
|
3520 char_u *dw;
|
|
236
|
3521 char_u *pc;
|
|
|
3522 char_u *w;
|
|
|
3523 int l;
|
|
|
3524 hash_T hash;
|
|
|
3525 hashitem_T *hi;
|
|
|
3526 FILE *fd;
|
|
|
3527 int lnum = 1;
|
|
300
|
3528 int non_ascii = 0;
|
|
|
3529 int retval = OK;
|
|
|
3530 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
3531 int flags;
|
|
236
|
3532
|
|
300
|
3533 /*
|
|
|
3534 * Open the file.
|
|
|
3535 */
|
|
310
|
3536 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3537 if (fd == NULL)
|
|
|
3538 {
|
|
|
3539 EMSG2(_(e_notopen), fname);
|
|
|
3540 return FAIL;
|
|
|
3541 }
|
|
|
3542
|
|
300
|
3543 /* The hashtable is only used to detect duplicated words. */
|
|
|
3544 hash_init(&ht);
|
|
|
3545
|
|
334
|
3546 spin->si_foldwcount = 0;
|
|
|
3547 spin->si_keepwcount = 0;
|
|
|
3548
|
|
310
|
3549 if (spin->si_verbose || p_verbose > 2)
|
|
|
3550 {
|
|
|
3551 if (!spin->si_verbose)
|
|
|
3552 verbose_enter();
|
|
366
|
3553 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
3554 out_flush();
|
|
|
3555 if (!spin->si_verbose)
|
|
|
3556 verbose_leave();
|
|
|
3557 }
|
|
236
|
3558
|
|
|
3559 /* Read and ignore the first line: word count. */
|
|
|
3560 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
3561 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
3562 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
3563
|
|
|
3564 /*
|
|
|
3565 * Read all the lines in the file one by one.
|
|
|
3566 * The words are converted to 'encoding' here, before being added to
|
|
|
3567 * the hashtable.
|
|
|
3568 */
|
|
255
|
3569 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
3570 {
|
|
255
|
3571 line_breakcheck();
|
|
236
|
3572 ++lnum;
|
|
351
|
3573 if (line[0] == '#')
|
|
|
3574 continue; /* comment line */
|
|
236
|
3575
|
|
300
|
3576 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
3577 * the word is kept to allow e.g., "et al.". */
|
|
236
|
3578 l = STRLEN(line);
|
|
|
3579 while (l > 0 && line[l - 1] <= ' ')
|
|
|
3580 --l;
|
|
|
3581 if (l == 0)
|
|
|
3582 continue; /* empty line */
|
|
|
3583 line[l] = NUL;
|
|
|
3584
|
|
|
3585 /* Find the optional affix names. */
|
|
300
|
3586 afflist = vim_strchr(line, '/');
|
|
|
3587 if (afflist != NULL)
|
|
|
3588 *afflist++ = NUL;
|
|
236
|
3589
|
|
300
|
3590 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
3591 if (spin->si_ascii && has_non_ascii(line))
|
|
|
3592 {
|
|
|
3593 ++non_ascii;
|
|
240
|
3594 continue;
|
|
300
|
3595 }
|
|
240
|
3596
|
|
310
|
3597 #ifdef FEAT_MBYTE
|
|
236
|
3598 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
3599 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3600 {
|
|
300
|
3601 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
3602 if (pc == NULL)
|
|
|
3603 {
|
|
|
3604 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3605 fname, lnum, line);
|
|
|
3606 continue;
|
|
|
3607 }
|
|
236
|
3608 w = pc;
|
|
|
3609 }
|
|
|
3610 else
|
|
310
|
3611 #endif
|
|
236
|
3612 {
|
|
|
3613 pc = NULL;
|
|
|
3614 w = line;
|
|
|
3615 }
|
|
|
3616
|
|
339
|
3617 /* This takes time, print a message now and then. */
|
|
|
3618 if (spin->si_verbose && (lnum & 0x3ff) == 0)
|
|
|
3619 {
|
|
|
3620 vim_snprintf((char *)message, sizeof(message),
|
|
|
3621 _("line %6d, word %6d - %s"),
|
|
|
3622 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
3623 msg_start();
|
|
|
3624 msg_puts_long_attr(message, 0);
|
|
|
3625 msg_clr_eos();
|
|
|
3626 msg_didout = FALSE;
|
|
|
3627 msg_col = 0;
|
|
|
3628 out_flush();
|
|
|
3629 }
|
|
|
3630
|
|
300
|
3631 /* Store the word in the hashtable to be able to find duplicates. */
|
|
|
3632 dw = (char_u *)getroom_save(&spin->si_blocks, w);
|
|
236
|
3633 if (dw == NULL)
|
|
300
|
3634 retval = FAIL;
|
|
|
3635 vim_free(pc);
|
|
|
3636 if (retval == FAIL)
|
|
236
|
3637 break;
|
|
|
3638
|
|
300
|
3639 hash = hash_hash(dw);
|
|
|
3640 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
3641 if (!HASHITEM_EMPTY(hi))
|
|
|
3642 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
3643 fname, lnum, dw);
|
|
236
|
3644 else
|
|
300
|
3645 hash_add_item(&ht, hi, dw, hash);
|
|
|
3646
|
|
307
|
3647 flags = 0;
|
|
339
|
3648 pfxlist = NULL;
|
|
307
|
3649 if (afflist != NULL)
|
|
|
3650 {
|
|
|
3651 /* Check for affix name that stands for keep-case word and stands
|
|
|
3652 * for rare word (if defined). */
|
|
310
|
3653 if (affile->af_kep != NUL
|
|
|
3654 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
307
|
3655 flags |= WF_KEEPCAP;
|
|
|
3656 if (affile->af_rar != NUL
|
|
|
3657 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
3658 flags |= WF_RARE;
|
|
346
|
3659 if (affile->af_bad != NUL
|
|
|
3660 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
3661 flags |= WF_BANNED;
|
|
339
|
3662
|
|
|
3663 if (affile->af_pfxpostpone)
|
|
|
3664 /* Need to store the list of prefix IDs with the word. */
|
|
|
3665 pfxlist = get_pfxlist(affile, afflist, &spin->si_blocks);
|
|
307
|
3666 }
|
|
|
3667
|
|
300
|
3668 /* Add the word to the word tree(s). */
|
|
339
|
3669 if (store_word(dw, spin, flags, spin->si_region, pfxlist) == FAIL)
|
|
300
|
3670 retval = FAIL;
|
|
236
|
3671
|
|
300
|
3672 if (afflist != NULL)
|
|
|
3673 {
|
|
|
3674 /* Find all matching suffixes and add the resulting words.
|
|
|
3675 * Additionally do matching prefixes that combine. */
|
|
339
|
3676 if (store_aff_word(dw, spin, afflist, affile,
|
|
307
|
3677 &affile->af_suff, &affile->af_pref,
|
|
339
|
3678 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
3679 retval = FAIL;
|
|
|
3680
|
|
|
3681 /* Find all matching prefixes and add the resulting words. */
|
|
339
|
3682 if (store_aff_word(dw, spin, afflist, affile,
|
|
|
3683 &affile->af_pref, NULL,
|
|
|
3684 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
3685 retval = FAIL;
|
|
|
3686 }
|
|
236
|
3687 }
|
|
|
3688
|
|
300
|
3689 if (spin->si_ascii && non_ascii > 0)
|
|
|
3690 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
3691 non_ascii);
|
|
|
3692 hash_clear(&ht);
|
|
|
3693
|
|
236
|
3694 fclose(fd);
|
|
300
|
3695 return retval;
|
|
236
|
3696 }
|
|
|
3697
|
|
|
3698 /*
|
|
339
|
3699 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
3700 * Used for PFXPOSTPONE.
|
|
|
3701 * Returns a string allocated with getroom(). NULL when there are no prefixes
|
|
|
3702 * or when out of memory.
|
|
|
3703 */
|
|
|
3704 static char_u *
|
|
|
3705 get_pfxlist(affile, afflist, blp)
|
|
|
3706 afffile_T *affile;
|
|
|
3707 char_u *afflist;
|
|
|
3708 sblock_T **blp;
|
|
|
3709 {
|
|
|
3710 char_u *p;
|
|
|
3711 int cnt;
|
|
|
3712 int round;
|
|
|
3713 char_u *res = NULL;
|
|
|
3714 char_u key[2];
|
|
|
3715 hashitem_T *hi;
|
|
|
3716
|
|
|
3717 key[1] = NUL;
|
|
|
3718
|
|
|
3719 /* round 1: count the number of prefix IDs.
|
|
|
3720 * round 2: move prefix IDs to "res" */
|
|
|
3721 for (round = 1; round <= 2; ++round)
|
|
|
3722 {
|
|
|
3723 cnt = 0;
|
|
|
3724 for (p = afflist; *p != NUL; ++p)
|
|
|
3725 {
|
|
|
3726 key[0] = *p;
|
|
|
3727 hi = hash_find(&affile->af_pref, key);
|
|
|
3728 if (!HASHITEM_EMPTY(hi))
|
|
|
3729 {
|
|
|
3730 /* This is a prefix ID, use the new number. */
|
|
|
3731 if (round == 2)
|
|
|
3732 res[cnt] = HI2AH(hi)->ah_newID;
|
|
|
3733 ++cnt;
|
|
|
3734 }
|
|
|
3735 }
|
|
|
3736 if (round == 1 && cnt > 0)
|
|
|
3737 res = getroom(blp, cnt + 1);
|
|
|
3738 if (res == NULL)
|
|
|
3739 break;
|
|
|
3740 }
|
|
|
3741
|
|
|
3742 if (res != NULL)
|
|
|
3743 res[cnt] = NUL;
|
|
|
3744 return res;
|
|
|
3745 }
|
|
|
3746
|
|
|
3747 /*
|
|
300
|
3748 * Apply affixes to a word and store the resulting words.
|
|
|
3749 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
3750 * prefixes or suffixes.
|
|
|
3751 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
3752 * the resulting words for combining affixes.
|
|
|
3753 *
|
|
|
3754 * Returns FAIL when out of memory.
|
|
236
|
3755 */
|
|
300
|
3756 static int
|
|
339
|
3757 store_aff_word(word, spin, afflist, affile, ht, xht, comb, flags, pfxlist)
|
|
300
|
3758 char_u *word; /* basic word start */
|
|
|
3759 spellinfo_T *spin; /* spell info */
|
|
|
3760 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
3761 afffile_T *affile;
|
|
300
|
3762 hashtab_T *ht;
|
|
|
3763 hashtab_T *xht;
|
|
|
3764 int comb; /* only use affixes that combine */
|
|
307
|
3765 int flags; /* flags for the word */
|
|
339
|
3766 char_u *pfxlist; /* list of prefix IDs */
|
|
236
|
3767 {
|
|
|
3768 int todo;
|
|
|
3769 hashitem_T *hi;
|
|
300
|
3770 affheader_T *ah;
|
|
|
3771 affentry_T *ae;
|
|
|
3772 regmatch_T regmatch;
|
|
|
3773 char_u newword[MAXWLEN];
|
|
|
3774 int retval = OK;
|
|
|
3775 int i;
|
|
|
3776 char_u *p;
|
|
366
|
3777 int use_flags;
|
|
236
|
3778
|
|
300
|
3779 todo = ht->ht_used;
|
|
|
3780 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
3781 {
|
|
|
3782 if (!HASHITEM_EMPTY(hi))
|
|
|
3783 {
|
|
|
3784 --todo;
|
|
300
|
3785 ah = HI2AH(hi);
|
|
236
|
3786
|
|
300
|
3787 /* Check that the affix combines, if required, and that the word
|
|
|
3788 * supports this affix. */
|
|
|
3789 if ((!comb || ah->ah_combine)
|
|
|
3790 && vim_strchr(afflist, *ah->ah_key) != NULL)
|
|
236
|
3791 {
|
|
300
|
3792 /* Loop over all affix entries with this name. */
|
|
|
3793 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
3794 {
|
|
300
|
3795 /* Check the condition. It's not logical to match case
|
|
|
3796 * here, but it is required for compatibility with
|
|
339
|
3797 * Myspell.
|
|
|
3798 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
3799 * prefixes with a chop string. */
|
|
300
|
3800 regmatch.regprog = ae->ae_prog;
|
|
|
3801 regmatch.rm_ic = FALSE;
|
|
339
|
3802 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
3803 || ae->ae_chop != NULL)
|
|
|
3804 && (ae->ae_prog == NULL
|
|
|
3805 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
3806 {
|
|
|
3807 /* Match. Remove the chop and add the affix. */
|
|
|
3808 if (xht == NULL)
|
|
240
|
3809 {
|
|
300
|
3810 /* prefix: chop/add at the start of the word */
|
|
|
3811 if (ae->ae_add == NULL)
|
|
|
3812 *newword = NUL;
|
|
|
3813 else
|
|
|
3814 STRCPY(newword, ae->ae_add);
|
|
|
3815 p = word;
|
|
|
3816 if (ae->ae_chop != NULL)
|
|
310
|
3817 {
|
|
300
|
3818 /* Skip chop string. */
|
|
310
|
3819 #ifdef FEAT_MBYTE
|
|
|
3820 if (has_mbyte)
|
|
324
|
3821 {
|
|
310
|
3822 i = mb_charlen(ae->ae_chop);
|
|
324
|
3823 for ( ; i > 0; --i)
|
|
|
3824 mb_ptr_adv(p);
|
|
|
3825 }
|
|
310
|
3826 else
|
|
|
3827 #endif
|
|
324
|
3828 p += STRLEN(ae->ae_chop);
|
|
310
|
3829 }
|
|
300
|
3830 STRCAT(newword, p);
|
|
|
3831 }
|
|
|
3832 else
|
|
|
3833 {
|
|
|
3834 /* suffix: chop/add at the end of the word */
|
|
|
3835 STRCPY(newword, word);
|
|
|
3836 if (ae->ae_chop != NULL)
|
|
|
3837 {
|
|
|
3838 /* Remove chop string. */
|
|
|
3839 p = newword + STRLEN(newword);
|
|
310
|
3840 #ifdef FEAT_MBYTE
|
|
|
3841 if (has_mbyte)
|
|
|
3842 i = mb_charlen(ae->ae_chop);
|
|
|
3843 else
|
|
|
3844 #endif
|
|
|
3845 i = STRLEN(ae->ae_chop);
|
|
|
3846 for ( ; i > 0; --i)
|
|
300
|
3847 mb_ptr_back(newword, p);
|
|
|
3848 *p = NUL;
|
|
|
3849 }
|
|
|
3850 if (ae->ae_add != NULL)
|
|
|
3851 STRCAT(newword, ae->ae_add);
|
|
240
|
3852 }
|
|
|
3853
|
|
366
|
3854 /* Obey the "rare" flag of the affix. */
|
|
|
3855 if (ae->ae_rare)
|
|
|
3856 use_flags = flags | WF_RARE;
|
|
|
3857 else
|
|
|
3858 use_flags = flags;
|
|
|
3859
|
|
300
|
3860 /* Store the modified word. */
|
|
366
|
3861 if (store_word(newword, spin, use_flags,
|
|
|
3862 spin->si_region, pfxlist) == FAIL)
|
|
300
|
3863 retval = FAIL;
|
|
236
|
3864
|
|
300
|
3865 /* When added a suffix and combining is allowed also
|
|
|
3866 * try adding prefixes additionally. */
|
|
|
3867 if (xht != NULL && ah->ah_combine)
|
|
339
|
3868 if (store_aff_word(newword, spin, afflist, affile,
|
|
366
|
3869 xht, NULL, TRUE, use_flags, pfxlist)
|
|
|
3870 == FAIL)
|
|
300
|
3871 retval = FAIL;
|
|
236
|
3872 }
|
|
|
3873 }
|
|
|
3874 }
|
|
|
3875 }
|
|
|
3876 }
|
|
|
3877
|
|
|
3878 return retval;
|
|
|
3879 }
|
|
|
3880
|
|
|
3881 /*
|
|
300
|
3882 * Read a file with a list of words.
|
|
236
|
3883 */
|
|
|
3884 static int
|
|
300
|
3885 spell_read_wordfile(fname, spin)
|
|
|
3886 char_u *fname;
|
|
|
3887 spellinfo_T *spin;
|
|
236
|
3888 {
|
|
300
|
3889 FILE *fd;
|
|
|
3890 long lnum = 0;
|
|
|
3891 char_u rline[MAXLINELEN];
|
|
|
3892 char_u *line;
|
|
|
3893 char_u *pc = NULL;
|
|
|
3894 int l;
|
|
|
3895 int retval = OK;
|
|
|
3896 int did_word = FALSE;
|
|
|
3897 int non_ascii = 0;
|
|
307
|
3898 int flags;
|
|
316
|
3899 int regionmask;
|
|
236
|
3900
|
|
300
|
3901 /*
|
|
|
3902 * Open the file.
|
|
|
3903 */
|
|
310
|
3904 fd = mch_fopen((char *)fname, "r");
|
|
300
|
3905 if (fd == NULL)
|
|
236
|
3906 {
|
|
300
|
3907 EMSG2(_(e_notopen), fname);
|
|
|
3908 return FAIL;
|
|
236
|
3909 }
|
|
|
3910
|
|
310
|
3911 if (spin->si_verbose || p_verbose > 2)
|
|
|
3912 {
|
|
|
3913 if (!spin->si_verbose)
|
|
|
3914 verbose_enter();
|
|
366
|
3915 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
3916 out_flush();
|
|
|
3917 if (!spin->si_verbose)
|
|
|
3918 verbose_leave();
|
|
|
3919 }
|
|
300
|
3920
|
|
|
3921 /*
|
|
|
3922 * Read all the lines in the file one by one.
|
|
|
3923 */
|
|
|
3924 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
3925 {
|
|
|
3926 line_breakcheck();
|
|
|
3927 ++lnum;
|
|
|
3928
|
|
|
3929 /* Skip comment lines. */
|
|
|
3930 if (*rline == '#')
|
|
|
3931 continue;
|
|
|
3932
|
|
|
3933 /* Remove CR, LF and white space from the end. */
|
|
|
3934 l = STRLEN(rline);
|
|
|
3935 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
3936 --l;
|
|
|
3937 if (l == 0)
|
|
|
3938 continue; /* empty or blank line */
|
|
|
3939 rline[l] = NUL;
|
|
|
3940
|
|
|
3941 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
3942 vim_free(pc);
|
|
310
|
3943 #ifdef FEAT_MBYTE
|
|
300
|
3944 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
3945 {
|
|
|
3946 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
3947 if (pc == NULL)
|
|
|
3948 {
|
|
|
3949 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3950 fname, lnum, rline);
|
|
|
3951 continue;
|
|
|
3952 }
|
|
|
3953 line = pc;
|
|
|
3954 }
|
|
|
3955 else
|
|
310
|
3956 #endif
|
|
300
|
3957 {
|
|
|
3958 pc = NULL;
|
|
|
3959 line = rline;
|
|
|
3960 }
|
|
|
3961
|
|
307
|
3962 flags = 0;
|
|
316
|
3963 regionmask = spin->si_region;
|
|
307
|
3964
|
|
|
3965 if (*line == '/')
|
|
300
|
3966 {
|
|
307
|
3967 ++line;
|
|
316
|
3968
|
|
307
|
3969 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
3970 {
|
|
|
3971 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
3972 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
3973 fname, lnum, line - 1);
|
|
300
|
3974 else if (did_word)
|
|
316
|
3975 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
3976 fname, lnum, line - 1);
|
|
300
|
3977 else
|
|
|
3978 {
|
|
310
|
3979 #ifdef FEAT_MBYTE
|
|
|
3980 char_u *enc;
|
|
|
3981
|
|
300
|
3982 /* Setup for conversion to 'encoding'. */
|
|
316
|
3983 line += 10;
|
|
|
3984 enc = enc_canonize(line);
|
|
300
|
3985 if (enc != NULL && !spin->si_ascii
|
|
|
3986 && convert_setup(&spin->si_conv, enc,
|
|
|
3987 p_enc) == FAIL)
|
|
|
3988 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
3989 fname, line, p_enc);
|
|
300
|
3990 vim_free(enc);
|
|
375
|
3991 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3992 #else
|
|
|
3993 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3994 #endif
|
|
300
|
3995 }
|
|
307
|
3996 continue;
|
|
300
|
3997 }
|
|
307
|
3998
|
|
316
|
3999 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
4000 {
|
|
|
4001 if (spin->si_region_count > 1)
|
|
|
4002 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
4003 fname, lnum, line);
|
|
|
4004 else
|
|
|
4005 {
|
|
|
4006 line += 8;
|
|
|
4007 if (STRLEN(line) > 16)
|
|
|
4008 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
4009 fname, lnum, line);
|
|
|
4010 else
|
|
|
4011 {
|
|
|
4012 spin->si_region_count = STRLEN(line) / 2;
|
|
|
4013 STRCPY(spin->si_region_name, line);
|
|
|
4014 }
|
|
|
4015 }
|
|
|
4016 continue;
|
|
|
4017 }
|
|
|
4018
|
|
307
|
4019 if (*line == '=')
|
|
|
4020 {
|
|
|
4021 /* keep-case word */
|
|
|
4022 flags |= WF_KEEPCAP;
|
|
|
4023 ++line;
|
|
|
4024 }
|
|
|
4025
|
|
|
4026 if (*line == '!')
|
|
|
4027 {
|
|
|
4028 /* Bad, bad, wicked word. */
|
|
|
4029 flags |= WF_BANNED;
|
|
|
4030 ++line;
|
|
|
4031 }
|
|
|
4032 else if (*line == '?')
|
|
|
4033 {
|
|
|
4034 /* Rare word. */
|
|
|
4035 flags |= WF_RARE;
|
|
|
4036 ++line;
|
|
|
4037 }
|
|
|
4038
|
|
316
|
4039 if (VIM_ISDIGIT(*line))
|
|
|
4040 {
|
|
|
4041 /* region number(s) */
|
|
|
4042 regionmask = 0;
|
|
|
4043 while (VIM_ISDIGIT(*line))
|
|
|
4044 {
|
|
|
4045 l = *line - '0';
|
|
|
4046 if (l > spin->si_region_count)
|
|
|
4047 {
|
|
|
4048 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
|
4049 fname, lnum, line);
|
|
|
4050 break;
|
|
|
4051 }
|
|
|
4052 regionmask |= 1 << (l - 1);
|
|
|
4053 ++line;
|
|
|
4054 }
|
|
|
4055 flags |= WF_REGION;
|
|
|
4056 }
|
|
|
4057
|
|
307
|
4058 if (flags == 0)
|
|
|
4059 {
|
|
|
4060 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
300
|
4061 fname, lnum, line);
|
|
307
|
4062 continue;
|
|
|
4063 }
|
|
300
|
4064 }
|
|
|
4065
|
|
|
4066 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4067 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4068 {
|
|
|
4069 ++non_ascii;
|
|
|
4070 continue;
|
|
|
4071 }
|
|
|
4072
|
|
|
4073 /* Normal word: store it. */
|
|
339
|
4074 if (store_word(line, spin, flags, regionmask, NULL) == FAIL)
|
|
300
|
4075 {
|
|
|
4076 retval = FAIL;
|
|
|
4077 break;
|
|
|
4078 }
|
|
|
4079 did_word = TRUE;
|
|
|
4080 }
|
|
|
4081
|
|
|
4082 vim_free(pc);
|
|
|
4083 fclose(fd);
|
|
|
4084
|
|
310
|
4085 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
4086 {
|
|
|
4087 if (p_verbose > 2)
|
|
|
4088 verbose_enter();
|
|
300
|
4089 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4090 non_ascii);
|
|
310
|
4091 if (p_verbose > 2)
|
|
|
4092 verbose_leave();
|
|
|
4093 }
|
|
300
|
4094 return retval;
|
|
236
|
4095 }
|
|
|
4096
|
|
|
4097 /*
|
|
300
|
4098 * Get part of an sblock_T, "len" bytes long.
|
|
|
4099 * This avoids calling free() for every little struct we use.
|
|
|
4100 * The memory is cleared to all zeros.
|
|
|
4101 * Returns NULL when out of memory.
|
|
|
4102 */
|
|
|
4103 static void *
|
|
|
4104 getroom(blp, len)
|
|
|
4105 sblock_T **blp;
|
|
|
4106 size_t len; /* length needed */
|
|
|
4107 {
|
|
|
4108 char_u *p;
|
|
|
4109 sblock_T *bl = *blp;
|
|
|
4110
|
|
|
4111 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
4112 {
|
|
|
4113 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
4114 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
4115 if (bl == NULL)
|
|
|
4116 return NULL;
|
|
|
4117 bl->sb_next = *blp;
|
|
|
4118 *blp = bl;
|
|
|
4119 bl->sb_used = 0;
|
|
|
4120 }
|
|
|
4121
|
|
|
4122 p = bl->sb_data + bl->sb_used;
|
|
|
4123 bl->sb_used += len;
|
|
|
4124
|
|
|
4125 return p;
|
|
|
4126 }
|
|
|
4127
|
|
|
4128 /*
|
|
|
4129 * Make a copy of a string into memory allocated with getroom().
|
|
|
4130 */
|
|
|
4131 static char_u *
|
|
|
4132 getroom_save(blp, s)
|
|
|
4133 sblock_T **blp;
|
|
|
4134 char_u *s;
|
|
|
4135 {
|
|
|
4136 char_u *sc;
|
|
|
4137
|
|
|
4138 sc = (char_u *)getroom(blp, STRLEN(s) + 1);
|
|
|
4139 if (sc != NULL)
|
|
|
4140 STRCPY(sc, s);
|
|
|
4141 return sc;
|
|
|
4142 }
|
|
|
4143
|
|
|
4144
|
|
|
4145 /*
|
|
|
4146 * Free the list of allocated sblock_T.
|
|
236
|
4147 */
|
|
|
4148 static void
|
|
300
|
4149 free_blocks(bl)
|
|
|
4150 sblock_T *bl;
|
|
236
|
4151 {
|
|
300
|
4152 sblock_T *next;
|
|
236
|
4153
|
|
300
|
4154 while (bl != NULL)
|
|
236
|
4155 {
|
|
300
|
4156 next = bl->sb_next;
|
|
|
4157 vim_free(bl);
|
|
|
4158 bl = next;
|
|
236
|
4159 }
|
|
|
4160 }
|
|
|
4161
|
|
|
4162 /*
|
|
300
|
4163 * Allocate the root of a word tree.
|
|
236
|
4164 */
|
|
300
|
4165 static wordnode_T *
|
|
|
4166 wordtree_alloc(blp)
|
|
|
4167 sblock_T **blp;
|
|
236
|
4168 {
|
|
300
|
4169 return (wordnode_T *)getroom(blp, sizeof(wordnode_T));
|
|
236
|
4170 }
|
|
|
4171
|
|
|
4172 /*
|
|
300
|
4173 * Store a word in the tree(s).
|
|
307
|
4174 * Always store it in the case-folded tree. A keep-case word can also be used
|
|
|
4175 * with all caps.
|
|
300
|
4176 * For a keep-case word also store it in the keep-case tree.
|
|
339
|
4177 * When "pfxlist" is not NULL store the word for each prefix ID.
|
|
236
|
4178 */
|
|
|
4179 static int
|
|
339
|
4180 store_word(word, spin, flags, region, pfxlist)
|
|
300
|
4181 char_u *word;
|
|
|
4182 spellinfo_T *spin;
|
|
307
|
4183 int flags; /* extra flags, WF_BANNED */
|
|
316
|
4184 int region; /* supported region(s) */
|
|
339
|
4185 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
236
|
4186 {
|
|
300
|
4187 int len = STRLEN(word);
|
|
|
4188 int ct = captype(word, word + len);
|
|
|
4189 char_u foldword[MAXWLEN];
|
|
339
|
4190 int res = OK;
|
|
|
4191 char_u *p;
|
|
236
|
4192
|
|
323
|
4193 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
4194 for (p = pfxlist; res == OK; ++p)
|
|
|
4195 {
|
|
|
4196 res = tree_add_word(foldword, spin->si_foldroot, ct | flags,
|
|
|
4197 region, p == NULL ? 0 : *p, &spin->si_blocks);
|
|
|
4198 if (p == NULL || *p == NUL)
|
|
|
4199 break;
|
|
|
4200 }
|
|
334
|
4201 ++spin->si_foldwcount;
|
|
307
|
4202
|
|
|
4203 if (res == OK && (ct == WF_KEEPCAP || flags & WF_KEEPCAP))
|
|
334
|
4204 {
|
|
339
|
4205 for (p = pfxlist; res == OK; ++p)
|
|
|
4206 {
|
|
|
4207 res = tree_add_word(word, spin->si_keeproot, flags,
|
|
|
4208 region, p == NULL ? 0 : *p, &spin->si_blocks);
|
|
|
4209 if (p == NULL || *p == NUL)
|
|
|
4210 break;
|
|
|
4211 }
|
|
334
|
4212 ++spin->si_keepwcount;
|
|
|
4213 }
|
|
300
|
4214 return res;
|
|
236
|
4215 }
|
|
|
4216
|
|
|
4217 /*
|
|
300
|
4218 * Add word "word" to a word tree at "root".
|
|
366
|
4219 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
4220 * "rare" and "region" is the condition nr.
|
|
255
|
4221 * Returns FAIL when out of memory.
|
|
236
|
4222 */
|
|
255
|
4223 static int
|
|
339
|
4224 tree_add_word(word, root, flags, region, prefixID, blp)
|
|
300
|
4225 char_u *word;
|
|
|
4226 wordnode_T *root;
|
|
|
4227 int flags;
|
|
|
4228 int region;
|
|
339
|
4229 int prefixID;
|
|
300
|
4230 sblock_T **blp;
|
|
236
|
4231 {
|
|
300
|
4232 wordnode_T *node = root;
|
|
|
4233 wordnode_T *np;
|
|
|
4234 wordnode_T **prev = NULL;
|
|
|
4235 int i;
|
|
255
|
4236
|
|
300
|
4237 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
4238 for (i = 0; ; ++i)
|
|
255
|
4239 {
|
|
300
|
4240 /* Look for the sibling that has the same character. They are sorted
|
|
|
4241 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
4242 * higher byte value. For zero bytes (end of word) the sorting is
|
|
|
4243 * done on flags and then on prefixID
|
|
300
|
4244 */
|
|
339
|
4245 while (node != NULL
|
|
|
4246 && (node->wn_byte < word[i]
|
|
|
4247 || (node->wn_byte == NUL
|
|
|
4248 && (flags < 0
|
|
|
4249 ? node->wn_prefixID < prefixID
|
|
|
4250 : node->wn_flags < (flags & 0xff)
|
|
|
4251 || (node->wn_flags == (flags & 0xff)
|
|
|
4252 && node->wn_prefixID < prefixID)))))
|
|
236
|
4253 {
|
|
300
|
4254 prev = &node->wn_sibling;
|
|
|
4255 node = *prev;
|
|
236
|
4256 }
|
|
339
|
4257 if (node == NULL
|
|
|
4258 || node->wn_byte != word[i]
|
|
|
4259 || (word[i] == NUL
|
|
|
4260 && (flags < 0
|
|
|
4261 || node->wn_flags != (flags & 0xff)
|
|
|
4262 || node->wn_prefixID != prefixID)))
|
|
255
|
4263 {
|
|
300
|
4264 /* Allocate a new node. */
|
|
|
4265 np = (wordnode_T *)getroom(blp, sizeof(wordnode_T));
|
|
|
4266 if (np == NULL)
|
|
|
4267 return FAIL;
|
|
|
4268 np->wn_byte = word[i];
|
|
|
4269 *prev = np;
|
|
|
4270 np->wn_sibling = node;
|
|
|
4271 node = np;
|
|
255
|
4272 }
|
|
300
|
4273
|
|
|
4274 if (word[i] == NUL)
|
|
|
4275 {
|
|
|
4276 node->wn_flags = flags;
|
|
|
4277 node->wn_region |= region;
|
|
339
|
4278 node->wn_prefixID = prefixID;
|
|
300
|
4279 break;
|
|
|
4280 }
|
|
|
4281 prev = &node->wn_child;
|
|
|
4282 node = *prev;
|
|
255
|
4283 }
|
|
|
4284
|
|
|
4285 return OK;
|
|
236
|
4286 }
|
|
|
4287
|
|
|
4288 /*
|
|
300
|
4289 * Compress a tree: find tails that are identical and can be shared.
|
|
|
4290 */
|
|
|
4291 static void
|
|
310
|
4292 wordtree_compress(root, spin)
|
|
300
|
4293 wordnode_T *root;
|
|
310
|
4294 spellinfo_T *spin;
|
|
300
|
4295 {
|
|
|
4296 hashtab_T ht;
|
|
|
4297 int n;
|
|
|
4298 int tot = 0;
|
|
|
4299
|
|
|
4300 if (root != NULL)
|
|
|
4301 {
|
|
|
4302 hash_init(&ht);
|
|
|
4303 n = node_compress(root, &ht, &tot);
|
|
310
|
4304 if (spin->si_verbose || p_verbose > 2)
|
|
|
4305 {
|
|
|
4306 if (!spin->si_verbose)
|
|
|
4307 verbose_enter();
|
|
|
4308 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
300
|
4309 n, tot, (tot - n) * 100 / tot);
|
|
310
|
4310 if (p_verbose > 2)
|
|
|
4311 verbose_leave();
|
|
|
4312 }
|
|
300
|
4313 hash_clear(&ht);
|
|
|
4314 }
|
|
|
4315 }
|
|
|
4316
|
|
|
4317 /*
|
|
|
4318 * Compress a node, its siblings and its children, depth first.
|
|
|
4319 * Returns the number of compressed nodes.
|
|
236
|
4320 */
|
|
255
|
4321 static int
|
|
300
|
4322 node_compress(node, ht, tot)
|
|
|
4323 wordnode_T *node;
|
|
|
4324 hashtab_T *ht;
|
|
|
4325 int *tot; /* total count of nodes before compressing,
|
|
|
4326 incremented while going through the tree */
|
|
236
|
4327 {
|
|
300
|
4328 wordnode_T *np;
|
|
|
4329 wordnode_T *tp;
|
|
|
4330 wordnode_T *child;
|
|
|
4331 hash_T hash;
|
|
236
|
4332 hashitem_T *hi;
|
|
300
|
4333 int len = 0;
|
|
|
4334 unsigned nr, n;
|
|
|
4335 int compressed = 0;
|
|
236
|
4336
|
|
300
|
4337 /*
|
|
|
4338 * Go through the list of siblings. Compress each child and then try
|
|
|
4339 * finding an identical child to replace it.
|
|
|
4340 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
4341 * but the whole list of siblings, of which the node is the first.
|
|
|
4342 */
|
|
|
4343 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4344 {
|
|
300
|
4345 ++len;
|
|
|
4346 if ((child = np->wn_child) != NULL)
|
|
|
4347 {
|
|
346
|
4348 /* Compress the child. This fills hashkey. */
|
|
300
|
4349 compressed += node_compress(child, ht, tot);
|
|
|
4350
|
|
|
4351 /* Try to find an identical child. */
|
|
346
|
4352 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
4353 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
4354 tp = NULL;
|
|
|
4355 if (!HASHITEM_EMPTY(hi))
|
|
|
4356 {
|
|
|
4357 /* There are children with an identical hash value. Now check
|
|
|
4358 * if there is one that is really identical. */
|
|
346
|
4359 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
4360 if (node_equal(child, tp))
|
|
|
4361 {
|
|
|
4362 /* Found one! Now use that child in place of the
|
|
|
4363 * current one. This means the current child is
|
|
|
4364 * dropped from the tree. */
|
|
|
4365 np->wn_child = tp;
|
|
|
4366 ++compressed;
|
|
|
4367 break;
|
|
|
4368 }
|
|
|
4369 if (tp == NULL)
|
|
|
4370 {
|
|
|
4371 /* No other child with this hash value equals the child of
|
|
|
4372 * the node, add it to the linked list after the first
|
|
|
4373 * item. */
|
|
|
4374 tp = HI2WN(hi);
|
|
346
|
4375 child->wn_u2.next = tp->wn_u2.next;
|
|
|
4376 tp->wn_u2.next = child;
|
|
300
|
4377 }
|
|
|
4378 }
|
|
|
4379 else
|
|
|
4380 /* No other child has this hash value, add it to the
|
|
|
4381 * hashtable. */
|
|
346
|
4382 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
4383 }
|
|
236
|
4384 }
|
|
300
|
4385 *tot += len;
|
|
|
4386
|
|
|
4387 /*
|
|
|
4388 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
4389 * find a lookalike node. This must be done after compressing the sibling
|
|
|
4390 * list, otherwise the hash key would become invalid by the compression.
|
|
|
4391 */
|
|
346
|
4392 node->wn_u1.hashkey[0] = len;
|
|
300
|
4393 nr = 0;
|
|
|
4394 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4395 {
|
|
300
|
4396 if (np->wn_byte == NUL)
|
|
339
|
4397 /* end node: use wn_flags, wn_region and wn_prefixID */
|
|
|
4398 n = np->wn_flags + (np->wn_region << 8) + (np->wn_prefixID << 16);
|
|
300
|
4399 else
|
|
|
4400 /* byte node: use the byte value and the child pointer */
|
|
|
4401 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
4402 nr = nr * 101 + n;
|
|
236
|
4403 }
|
|
300
|
4404
|
|
|
4405 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
4406 n = nr & 0xff;
|
|
346
|
4407 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
4408 n = (nr >> 8) & 0xff;
|
|
346
|
4409 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
4410 n = (nr >> 16) & 0xff;
|
|
346
|
4411 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
4412 n = (nr >> 24) & 0xff;
|
|
346
|
4413 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
4414 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
4415
|
|
|
4416 return compressed;
|
|
|
4417 }
|
|
|
4418
|
|
|
4419 /*
|
|
|
4420 * Return TRUE when two nodes have identical siblings and children.
|
|
|
4421 */
|
|
|
4422 static int
|
|
|
4423 node_equal(n1, n2)
|
|
|
4424 wordnode_T *n1;
|
|
|
4425 wordnode_T *n2;
|
|
|
4426 {
|
|
|
4427 wordnode_T *p1;
|
|
|
4428 wordnode_T *p2;
|
|
|
4429
|
|
|
4430 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
4431 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
4432 if (p1->wn_byte != p2->wn_byte
|
|
|
4433 || (p1->wn_byte == NUL
|
|
|
4434 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
4435 || p1->wn_region != p2->wn_region
|
|
|
4436 || p1->wn_prefixID != p2->wn_prefixID)
|
|
300
|
4437 : (p1->wn_child != p2->wn_child)))
|
|
|
4438 break;
|
|
|
4439
|
|
|
4440 return p1 == NULL && p2 == NULL;
|
|
236
|
4441 }
|
|
|
4442
|
|
|
4443 /*
|
|
|
4444 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
4445 */
|
|
255
|
4446 void
|
|
236
|
4447 put_bytes(fd, nr, len)
|
|
|
4448 FILE *fd;
|
|
|
4449 long_u nr;
|
|
|
4450 int len;
|
|
|
4451 {
|
|
|
4452 int i;
|
|
|
4453
|
|
|
4454 for (i = len - 1; i >= 0; --i)
|
|
|
4455 putc((int)(nr >> (i * 8)), fd);
|
|
|
4456 }
|
|
|
4457
|
|
323
|
4458 static int
|
|
|
4459 #ifdef __BORLANDC__
|
|
|
4460 _RTLENTRYF
|
|
|
4461 #endif
|
|
|
4462 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
4463
|
|
|
4464 /*
|
|
|
4465 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
4466 */
|
|
|
4467 static int
|
|
|
4468 #ifdef __BORLANDC__
|
|
|
4469 _RTLENTRYF
|
|
|
4470 #endif
|
|
|
4471 rep_compare(s1, s2)
|
|
|
4472 const void *s1;
|
|
|
4473 const void *s2;
|
|
|
4474 {
|
|
|
4475 fromto_T *p1 = (fromto_T *)s1;
|
|
|
4476 fromto_T *p2 = (fromto_T *)s2;
|
|
|
4477
|
|
|
4478 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
4479 }
|
|
|
4480
|
|
236
|
4481 /*
|
|
|
4482 * Write the Vim spell file "fname".
|
|
|
4483 */
|
|
|
4484 static void
|
|
316
|
4485 write_vim_spell(fname, spin)
|
|
236
|
4486 char_u *fname;
|
|
300
|
4487 spellinfo_T *spin;
|
|
236
|
4488 {
|
|
300
|
4489 FILE *fd;
|
|
|
4490 int regionmask;
|
|
236
|
4491 int round;
|
|
300
|
4492 wordnode_T *tree;
|
|
|
4493 int nodecount;
|
|
323
|
4494 int i;
|
|
|
4495 int l;
|
|
|
4496 garray_T *gap;
|
|
|
4497 fromto_T *ftp;
|
|
|
4498 char_u *p;
|
|
|
4499 int rr;
|
|
236
|
4500
|
|
310
|
4501 fd = mch_fopen((char *)fname, "w");
|
|
300
|
4502 if (fd == NULL)
|
|
236
|
4503 {
|
|
|
4504 EMSG2(_(e_notopen), fname);
|
|
|
4505 return;
|
|
|
4506 }
|
|
|
4507
|
|
255
|
4508 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
339
|
4509 * <charflagslen> <charflags>
|
|
|
4510 * <fcharslen> <fchars>
|
|
366
|
4511 * <midwordlen> <midword>
|
|
339
|
4512 * <prefcondcnt> <prefcond> ... */
|
|
300
|
4513
|
|
|
4514 /* <fileID> */
|
|
|
4515 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
4516 EMSG(_(e_write));
|
|
236
|
4517
|
|
|
4518 /* write the region names if there is more than one */
|
|
316
|
4519 if (spin->si_region_count > 1)
|
|
236
|
4520 {
|
|
316
|
4521 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
|
4522 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
|
4523 (size_t)1, fd);
|
|
|
4524 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
4525 }
|
|
|
4526 else
|
|
|
4527 {
|
|
300
|
4528 putc(0, fd);
|
|
|
4529 regionmask = 0;
|
|
236
|
4530 }
|
|
|
4531
|
|
323
|
4532 /*
|
|
|
4533 * Write the table with character flags and table for case folding.
|
|
260
|
4534 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
|
4535 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
4536 * 'encoding'.
|
|
|
4537 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
4538 * the table (avoids that it conflicts). File is shorter too.
|
|
|
4539 */
|
|
|
4540 if (spin->si_ascii || spin->si_add)
|
|
260
|
4541 {
|
|
300
|
4542 putc(0, fd);
|
|
|
4543 putc(0, fd);
|
|
|
4544 putc(0, fd);
|
|
260
|
4545 }
|
|
|
4546 else
|
|
300
|
4547 write_spell_chartab(fd);
|
|
255
|
4548
|
|
366
|
4549
|
|
|
4550 if (spin->si_midword == NULL)
|
|
|
4551 put_bytes(fd, 0L, 2); /* <midwordlen> */
|
|
|
4552 else
|
|
|
4553 {
|
|
|
4554 i = STRLEN(spin->si_midword);
|
|
|
4555 put_bytes(fd, (long_u)i, 2); /* <midwordlen> */
|
|
|
4556 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
4557 }
|
|
|
4558
|
|
|
4559
|
|
339
|
4560 /* Write the prefix conditions. */
|
|
|
4561 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
4562
|
|
366
|
4563 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
4564 * <salflags> <salcount> <sal> ...
|
|
|
4565 * <maplen> <mapstr> */
|
|
|
4566
|
|
323
|
4567 /* Sort the REP items. */
|
|
|
4568 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
4569 sizeof(fromto_T), rep_compare);
|
|
|
4570
|
|
375
|
4571 /* round 1: REP items
|
|
|
4572 * round 2: SAL items (unless SOFO is used) */
|
|
323
|
4573 for (round = 1; round <= 2; ++round)
|
|
|
4574 {
|
|
|
4575 if (round == 1)
|
|
|
4576 gap = &spin->si_rep;
|
|
|
4577 else
|
|
|
4578 {
|
|
|
4579 gap = &spin->si_sal;
|
|
|
4580
|
|
|
4581 i = 0;
|
|
|
4582 if (spin->si_followup)
|
|
|
4583 i |= SAL_F0LLOWUP;
|
|
|
4584 if (spin->si_collapse)
|
|
|
4585 i |= SAL_COLLAPSE;
|
|
|
4586 if (spin->si_rem_accents)
|
|
|
4587 i |= SAL_REM_ACCENTS;
|
|
375
|
4588 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
4589 i |= SAL_SOFO;
|
|
323
|
4590 putc(i, fd); /* <salflags> */
|
|
375
|
4591 if (i & SAL_SOFO)
|
|
|
4592 break;
|
|
323
|
4593 }
|
|
|
4594
|
|
|
4595 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
4596 for (i = 0; i < gap->ga_len; ++i)
|
|
|
4597 {
|
|
|
4598 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
4599 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
4600 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
4601 for (rr = 1; rr <= 2; ++rr)
|
|
|
4602 {
|
|
|
4603 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
4604 l = STRLEN(p);
|
|
|
4605 putc(l, fd);
|
|
|
4606 fwrite(p, l, (size_t)1, fd);
|
|
|
4607 }
|
|
|
4608 }
|
|
|
4609 }
|
|
|
4610
|
|
375
|
4611 /* SOFOFROM and SOFOTO */
|
|
|
4612 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
4613 {
|
|
|
4614 put_bytes(fd, 1L, 2); /* <salcount> */
|
|
|
4615
|
|
|
4616 l = STRLEN(spin->si_sofofr);
|
|
|
4617 put_bytes(fd, (long_u)l, 2); /* <salfromlen> */
|
|
|
4618 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <salfrom> */
|
|
|
4619
|
|
|
4620 l = STRLEN(spin->si_sofoto);
|
|
|
4621 put_bytes(fd, (long_u)l, 2); /* <saltolen> */
|
|
|
4622 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <salto> */
|
|
|
4623 }
|
|
|
4624
|
|
323
|
4625 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
|
4626 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
|
4627 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
|
4628 (size_t)1, fd);
|
|
302
|
4629
|
|
236
|
4630 /*
|
|
339
|
4631 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
4632 */
|
|
323
|
4633 spin->si_memtot = 0;
|
|
339
|
4634 for (round = 1; round <= 3; ++round)
|
|
236
|
4635 {
|
|
339
|
4636 if (round == 1)
|
|
|
4637 tree = spin->si_foldroot;
|
|
|
4638 else if (round == 2)
|
|
|
4639 tree = spin->si_keeproot;
|
|
|
4640 else
|
|
|
4641 tree = spin->si_prefroot;
|
|
236
|
4642
|
|
346
|
4643 /* Clear the index and wnode fields in the tree. */
|
|
|
4644 clear_node(tree);
|
|
|
4645
|
|
300
|
4646 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
4647 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
4648 * nodes. */
|
|
346
|
4649 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
4650
|
|
300
|
4651 /* number of nodes in 4 bytes */
|
|
|
4652 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
4653 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
4654
|
|
300
|
4655 /* Write the nodes. */
|
|
346
|
4656 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
4657 }
|
|
|
4658
|
|
300
|
4659 fclose(fd);
|
|
236
|
4660 }
|
|
|
4661
|
|
|
4662 /*
|
|
346
|
4663 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
4664 * children. This is needed because they are a union with other items to save
|
|
|
4665 * space.
|
|
|
4666 */
|
|
|
4667 static void
|
|
|
4668 clear_node(node)
|
|
|
4669 wordnode_T *node;
|
|
|
4670 {
|
|
|
4671 wordnode_T *np;
|
|
|
4672
|
|
|
4673 if (node != NULL)
|
|
|
4674 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4675 {
|
|
|
4676 np->wn_u1.index = 0;
|
|
|
4677 np->wn_u2.wnode = NULL;
|
|
|
4678
|
|
|
4679 if (np->wn_byte != NUL)
|
|
|
4680 clear_node(np->wn_child);
|
|
|
4681 }
|
|
|
4682 }
|
|
|
4683
|
|
|
4684
|
|
|
4685 /*
|
|
300
|
4686 * Dump a word tree at node "node".
|
|
|
4687 *
|
|
|
4688 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
4689 * byte recursively write the children.
|
|
|
4690 *
|
|
|
4691 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
4692 * are made about the indexes (so that we don't have to write them in the
|
|
|
4693 * file).
|
|
236
|
4694 *
|
|
300
|
4695 * Returns the number of nodes used.
|
|
236
|
4696 */
|
|
300
|
4697 static int
|
|
346
|
4698 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
4699 FILE *fd; /* NULL when only counting */
|
|
300
|
4700 wordnode_T *node;
|
|
|
4701 int index;
|
|
|
4702 int regionmask;
|
|
339
|
4703 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
4704 {
|
|
300
|
4705 int newindex = index;
|
|
|
4706 int siblingcount = 0;
|
|
|
4707 wordnode_T *np;
|
|
236
|
4708 int flags;
|
|
300
|
4709
|
|
|
4710 /* If "node" is zero the tree is empty. */
|
|
|
4711 if (node == NULL)
|
|
|
4712 return 0;
|
|
|
4713
|
|
|
4714 /* Store the index where this node is written. */
|
|
346
|
4715 node->wn_u1.index = index;
|
|
236
|
4716
|
|
300
|
4717 /* Count the number of siblings. */
|
|
|
4718 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4719 ++siblingcount;
|
|
236
|
4720
|
|
300
|
4721 /* Write the sibling count. */
|
|
|
4722 if (fd != NULL)
|
|
|
4723 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
4724
|
|
300
|
4725 /* Write each sibling byte and optionally extra info. */
|
|
|
4726 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4727 {
|
|
300
|
4728 if (np->wn_byte == 0)
|
|
|
4729 {
|
|
|
4730 if (fd != NULL)
|
|
|
4731 {
|
|
339
|
4732 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
4733 if (prefixtree)
|
|
300
|
4734 {
|
|
339
|
4735 /* In PREFIXTREE write the required prefixID and the
|
|
|
4736 * associated condition nr (stored in wn_region). */
|
|
366
|
4737 if (np->wn_flags == (char_u)-2)
|
|
|
4738 putc(BY_FLAGS, fd); /* <byte> rare */
|
|
|
4739 else
|
|
|
4740 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
339
|
4741 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
4742 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
4743 }
|
|
|
4744 else
|
|
|
4745 {
|
|
339
|
4746 /* For word trees we write the flag/region items. */
|
|
|
4747 flags = np->wn_flags;
|
|
|
4748 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
4749 flags |= WF_REGION;
|
|
|
4750 if (np->wn_prefixID != 0)
|
|
|
4751 flags |= WF_PFX;
|
|
|
4752 if (flags == 0)
|
|
|
4753 {
|
|
|
4754 /* word without flags or region */
|
|
|
4755 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
4756 }
|
|
|
4757 else
|
|
|
4758 {
|
|
|
4759 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
4760 putc(flags, fd); /* <flags> */
|
|
|
4761 if (flags & WF_REGION)
|
|
|
4762 putc(np->wn_region, fd); /* <region> */
|
|
|
4763 if (flags & WF_PFX)
|
|
|
4764 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
4765 }
|
|
300
|
4766 }
|
|
|
4767 }
|
|
|
4768 }
|
|
|
4769 else
|
|
|
4770 {
|
|
346
|
4771 if (np->wn_child->wn_u1.index != 0
|
|
|
4772 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
4773 {
|
|
|
4774 /* The child is written elsewhere, write the reference. */
|
|
|
4775 if (fd != NULL)
|
|
|
4776 {
|
|
|
4777 putc(BY_INDEX, fd); /* <byte> */
|
|
|
4778 /* <nodeidx> */
|
|
346
|
4779 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
4780 }
|
|
|
4781 }
|
|
346
|
4782 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
4783 /* We will write the child below and give it an index. */
|
|
346
|
4784 np->wn_child->wn_u2.wnode = node;
|
|
236
|
4785
|
|
300
|
4786 if (fd != NULL)
|
|
|
4787 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
4788 {
|
|
|
4789 EMSG(_(e_write));
|
|
|
4790 return 0;
|
|
|
4791 }
|
|
|
4792 }
|
|
236
|
4793 }
|
|
|
4794
|
|
300
|
4795 /* Space used in the array when reading: one for each sibling and one for
|
|
|
4796 * the count. */
|
|
|
4797 newindex += siblingcount + 1;
|
|
249
|
4798
|
|
300
|
4799 /* Recursively dump the children of each sibling. */
|
|
|
4800 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
4801 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
4802 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
4803 prefixtree);
|
|
249
|
4804
|
|
300
|
4805 return newindex;
|
|
236
|
4806 }
|
|
|
4807
|
|
|
4808
|
|
|
4809 /*
|
|
310
|
4810 * ":mkspell [-ascii] outfile infile ..."
|
|
|
4811 * ":mkspell [-ascii] addfile"
|
|
236
|
4812 */
|
|
|
4813 void
|
|
|
4814 ex_mkspell(eap)
|
|
|
4815 exarg_T *eap;
|
|
|
4816 {
|
|
|
4817 int fcount;
|
|
|
4818 char_u **fnames;
|
|
310
|
4819 char_u *arg = eap->arg;
|
|
|
4820 int ascii = FALSE;
|
|
|
4821
|
|
|
4822 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
4823 {
|
|
|
4824 ascii = TRUE;
|
|
|
4825 arg = skipwhite(arg + 6);
|
|
|
4826 }
|
|
|
4827
|
|
|
4828 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
4829 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
4830 {
|
|
323
|
4831 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
4832 FreeWild(fcount, fnames);
|
|
|
4833 }
|
|
|
4834 }
|
|
|
4835
|
|
|
4836 /*
|
|
|
4837 * Create a Vim spell file from one or more word lists.
|
|
|
4838 * "fnames[0]" is the output file name.
|
|
|
4839 * "fnames[fcount - 1]" is the last input file name.
|
|
|
4840 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
4841 * and ".spl" is appended to make the output file name.
|
|
|
4842 */
|
|
|
4843 static void
|
|
323
|
4844 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
4845 int fcount;
|
|
|
4846 char_u **fnames;
|
|
|
4847 int ascii; /* -ascii argument given */
|
|
|
4848 int overwrite; /* overwrite existing output file */
|
|
323
|
4849 int added_word; /* invoked through "zg" */
|
|
310
|
4850 {
|
|
236
|
4851 char_u fname[MAXPATHL];
|
|
|
4852 char_u wfname[MAXPATHL];
|
|
310
|
4853 char_u **innames;
|
|
|
4854 int incount;
|
|
236
|
4855 afffile_T *(afile[8]);
|
|
|
4856 int i;
|
|
|
4857 int len;
|
|
|
4858 struct stat st;
|
|
255
|
4859 int error = FALSE;
|
|
300
|
4860 spellinfo_T spin;
|
|
|
4861
|
|
|
4862 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
4863 spin.si_verbose = !added_word;
|
|
310
|
4864 spin.si_ascii = ascii;
|
|
323
|
4865 spin.si_followup = TRUE;
|
|
|
4866 spin.si_rem_accents = TRUE;
|
|
|
4867 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
4868 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
4869 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
4870 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
4871
|
|
|
4872 /* default: fnames[0] is output file, following are input files */
|
|
|
4873 innames = &fnames[1];
|
|
|
4874 incount = fcount - 1;
|
|
|
4875
|
|
|
4876 if (fcount >= 1)
|
|
240
|
4877 {
|
|
310
|
4878 len = STRLEN(fnames[0]);
|
|
|
4879 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
4880 {
|
|
|
4881 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
4882 * "path/en.latin1.add.spl". */
|
|
|
4883 innames = &fnames[0];
|
|
|
4884 incount = 1;
|
|
|
4885 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
4886 }
|
|
366
|
4887 else if (fcount == 1)
|
|
|
4888 {
|
|
|
4889 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
4890 innames = &fnames[0];
|
|
|
4891 incount = 1;
|
|
|
4892 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
4893 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
4894 }
|
|
310
|
4895 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
4896 {
|
|
|
4897 /* Name ends in ".spl", use as the file name. */
|
|
323
|
4898 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
4899 }
|
|
|
4900 else
|
|
|
4901 /* Name should be language, make the file name from it. */
|
|
|
4902 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
4903 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
4904
|
|
|
4905 /* Check for .ascii.spl. */
|
|
|
4906 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
4907 spin.si_ascii = TRUE;
|
|
|
4908
|
|
|
4909 /* Check for .add.spl. */
|
|
|
4910 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
4911 spin.si_add = TRUE;
|
|
240
|
4912 }
|
|
|
4913
|
|
310
|
4914 if (incount <= 0)
|
|
236
|
4915 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
4916 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
4917 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
4918 else if (incount > 8)
|
|
236
|
4919 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
4920 else
|
|
|
4921 {
|
|
|
4922 /* Check for overwriting before doing things that may take a lot of
|
|
|
4923 * time. */
|
|
310
|
4924 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
4925 {
|
|
|
4926 EMSG(_(e_exists));
|
|
310
|
4927 return;
|
|
236
|
4928 }
|
|
310
|
4929 if (mch_isdir(wfname))
|
|
236
|
4930 {
|
|
310
|
4931 EMSG2(_(e_isadir2), wfname);
|
|
|
4932 return;
|
|
236
|
4933 }
|
|
|
4934
|
|
|
4935 /*
|
|
|
4936 * Init the aff and dic pointers.
|
|
|
4937 * Get the region names if there are more than 2 arguments.
|
|
|
4938 */
|
|
310
|
4939 for (i = 0; i < incount; ++i)
|
|
236
|
4940 {
|
|
310
|
4941 afile[i] = NULL;
|
|
300
|
4942
|
|
316
|
4943 if (incount > 1)
|
|
236
|
4944 {
|
|
310
|
4945 len = STRLEN(innames[i]);
|
|
|
4946 if (STRLEN(gettail(innames[i])) < 5
|
|
|
4947 || innames[i][len - 3] != '_')
|
|
236
|
4948 {
|
|
310
|
4949 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
4950 return;
|
|
236
|
4951 }
|
|
316
|
4952 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
4953 spin.si_region_name[i * 2 + 1] =
|
|
|
4954 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
4955 }
|
|
|
4956 }
|
|
316
|
4957 spin.si_region_count = incount;
|
|
236
|
4958
|
|
300
|
4959 spin.si_foldroot = wordtree_alloc(&spin.si_blocks);
|
|
|
4960 spin.si_keeproot = wordtree_alloc(&spin.si_blocks);
|
|
339
|
4961 spin.si_prefroot = wordtree_alloc(&spin.si_blocks);
|
|
|
4962 if (spin.si_foldroot == NULL
|
|
|
4963 || spin.si_keeproot == NULL
|
|
|
4964 || spin.si_prefroot == NULL)
|
|
300
|
4965 {
|
|
|
4966 error = TRUE;
|
|
310
|
4967 return;
|
|
300
|
4968 }
|
|
|
4969
|
|
351
|
4970 /* When not producing a .add.spl file clear the character table when
|
|
|
4971 * we encounter one in the .aff file. This means we dump the current
|
|
|
4972 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
4973 * better than guessing the contents, the table will match a
|
|
|
4974 * previously loaded spell file. */
|
|
|
4975 if (!spin.si_add)
|
|
|
4976 spin.si_clear_chartab = TRUE;
|
|
|
4977
|
|
236
|
4978 /*
|
|
|
4979 * Read all the .aff and .dic files.
|
|
|
4980 * Text is converted to 'encoding'.
|
|
300
|
4981 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
4982 */
|
|
310
|
4983 for (i = 0; i < incount && !error; ++i)
|
|
236
|
4984 {
|
|
300
|
4985 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
4986 spin.si_region = 1 << i;
|
|
|
4987
|
|
|
4988 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
4989 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
4990 {
|
|
|
4991 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
4992 * "SET" line. */
|
|
310
|
4993 afile[i] = spell_read_aff(fname, &spin);
|
|
|
4994 if (afile[i] == NULL)
|
|
300
|
4995 error = TRUE;
|
|
|
4996 else
|
|
|
4997 {
|
|
|
4998 /* Read the .dic file and store the words in the trees. */
|
|
|
4999 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
5000 innames[i]);
|
|
|
5001 if (spell_read_dic(fname, &spin, afile[i]) == FAIL)
|
|
300
|
5002 error = TRUE;
|
|
|
5003 }
|
|
|
5004 }
|
|
|
5005 else
|
|
|
5006 {
|
|
|
5007 /* No .aff file, try reading the file as a word list. Store
|
|
|
5008 * the words in the trees. */
|
|
310
|
5009 if (spell_read_wordfile(innames[i], &spin) == FAIL)
|
|
300
|
5010 error = TRUE;
|
|
|
5011 }
|
|
236
|
5012
|
|
310
|
5013 #ifdef FEAT_MBYTE
|
|
236
|
5014 /* Free any conversion stuff. */
|
|
300
|
5015 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
5016 #endif
|
|
236
|
5017 }
|
|
|
5018
|
|
300
|
5019 if (!error)
|
|
236
|
5020 {
|
|
|
5021 /*
|
|
300
|
5022 * Remove the dummy NUL from the start of the tree root.
|
|
236
|
5023 */
|
|
300
|
5024 spin.si_foldroot = spin.si_foldroot->wn_sibling;
|
|
|
5025 spin.si_keeproot = spin.si_keeproot->wn_sibling;
|
|
339
|
5026 spin.si_prefroot = spin.si_prefroot->wn_sibling;
|
|
236
|
5027
|
|
|
5028 /*
|
|
300
|
5029 * Combine tails in the tree.
|
|
236
|
5030 */
|
|
323
|
5031 if (!added_word || p_verbose > 2)
|
|
310
|
5032 {
|
|
323
|
5033 if (added_word)
|
|
310
|
5034 verbose_enter();
|
|
|
5035 MSG(_("Compressing word tree..."));
|
|
|
5036 out_flush();
|
|
323
|
5037 if (added_word)
|
|
310
|
5038 verbose_leave();
|
|
|
5039 }
|
|
|
5040 wordtree_compress(spin.si_foldroot, &spin);
|
|
|
5041 wordtree_compress(spin.si_keeproot, &spin);
|
|
339
|
5042 wordtree_compress(spin.si_prefroot, &spin);
|
|
236
|
5043 }
|
|
|
5044
|
|
300
|
5045 if (!error)
|
|
|
5046 {
|
|
|
5047 /*
|
|
|
5048 * Write the info in the spell file.
|
|
|
5049 */
|
|
323
|
5050 if (!added_word || p_verbose > 2)
|
|
310
|
5051 {
|
|
323
|
5052 if (added_word)
|
|
310
|
5053 verbose_enter();
|
|
366
|
5054 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
5055 out_flush();
|
|
323
|
5056 if (added_word)
|
|
310
|
5057 verbose_leave();
|
|
|
5058 }
|
|
|
5059
|
|
316
|
5060 write_vim_spell(wfname, &spin);
|
|
310
|
5061
|
|
323
|
5062 if (!added_word || p_verbose > 2)
|
|
310
|
5063 {
|
|
323
|
5064 if (added_word)
|
|
310
|
5065 verbose_enter();
|
|
|
5066 MSG(_("Done!"));
|
|
|
5067 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
5068 spin.si_memtot);
|
|
310
|
5069 out_flush();
|
|
323
|
5070 if (added_word)
|
|
310
|
5071 verbose_leave();
|
|
|
5072 }
|
|
|
5073
|
|
|
5074 /* If the file is loaded need to reload it. */
|
|
323
|
5075 spell_reload_one(wfname, added_word);
|
|
300
|
5076 }
|
|
|
5077
|
|
|
5078 /* Free the allocated memory. */
|
|
323
|
5079 ga_clear(&spin.si_rep);
|
|
|
5080 ga_clear(&spin.si_sal);
|
|
|
5081 ga_clear(&spin.si_map);
|
|
339
|
5082 ga_clear(&spin.si_prefcond);
|
|
366
|
5083 vim_free(spin.si_midword);
|
|
375
|
5084 vim_free(spin.si_sofofr);
|
|
|
5085 vim_free(spin.si_sofoto);
|
|
300
|
5086
|
|
|
5087 /* Free the .aff file structures. */
|
|
310
|
5088 for (i = 0; i < incount; ++i)
|
|
|
5089 if (afile[i] != NULL)
|
|
|
5090 spell_free_aff(afile[i]);
|
|
339
|
5091
|
|
|
5092 /* Free all the bits and pieces at once. */
|
|
|
5093 free_blocks(spin.si_blocks);
|
|
236
|
5094 }
|
|
310
|
5095 }
|
|
|
5096
|
|
|
5097
|
|
|
5098 /*
|
|
|
5099 * ":spellgood {word}"
|
|
|
5100 * ":spellwrong {word}"
|
|
|
5101 */
|
|
|
5102 void
|
|
|
5103 ex_spell(eap)
|
|
|
5104 exarg_T *eap;
|
|
|
5105 {
|
|
|
5106 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong);
|
|
236
|
5107 }
|
|
|
5108
|
|
310
|
5109 /*
|
|
|
5110 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
5111 */
|
|
|
5112 void
|
|
|
5113 spell_add_word(word, len, bad)
|
|
|
5114 char_u *word;
|
|
|
5115 int len;
|
|
|
5116 int bad;
|
|
|
5117 {
|
|
|
5118 FILE *fd;
|
|
|
5119 buf_T *buf;
|
|
351
|
5120 int new_spf = FALSE;
|
|
|
5121 struct stat st;
|
|
|
5122
|
|
|
5123 /* If 'spellfile' isn't set figure out a good default value. */
|
|
310
|
5124 if (*curbuf->b_p_spf == NUL)
|
|
351
|
5125 {
|
|
310
|
5126 init_spellfile();
|
|
351
|
5127 new_spf = TRUE;
|
|
|
5128 }
|
|
|
5129
|
|
310
|
5130 if (*curbuf->b_p_spf == NUL)
|
|
323
|
5131 EMSG(_("E764: 'spellfile' is not set"));
|
|
310
|
5132 else
|
|
|
5133 {
|
|
|
5134 /* Check that the user isn't editing the .add file somewhere. */
|
|
|
5135 buf = buflist_findname_exp(curbuf->b_p_spf);
|
|
|
5136 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
5137 buf = NULL;
|
|
|
5138 if (buf != NULL && bufIsChanged(buf))
|
|
|
5139 EMSG(_(e_bufloaded));
|
|
|
5140 else
|
|
|
5141 {
|
|
|
5142 fd = mch_fopen((char *)curbuf->b_p_spf, "a");
|
|
351
|
5143 if (fd == NULL && new_spf)
|
|
|
5144 {
|
|
|
5145 /* We just initialized the 'spellfile' option and can't open
|
|
|
5146 * the file. We may need to create the "spell" directory
|
|
|
5147 * first. We already checked the runtime directory is
|
|
|
5148 * writable in init_spellfile(). */
|
|
|
5149 STRCPY(NameBuff, curbuf->b_p_spf);
|
|
|
5150 *gettail_sep(NameBuff) = NUL;
|
|
|
5151 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
5152 {
|
|
|
5153 /* The directory doesn't exist. Try creating it and
|
|
|
5154 * opening the file again. */
|
|
|
5155 vim_mkdir(NameBuff, 0755);
|
|
|
5156 fd = mch_fopen((char *)curbuf->b_p_spf, "a");
|
|
|
5157 }
|
|
|
5158 }
|
|
|
5159
|
|
310
|
5160 if (fd == NULL)
|
|
|
5161 EMSG2(_(e_notopen), curbuf->b_p_spf);
|
|
|
5162 else
|
|
|
5163 {
|
|
|
5164 if (bad)
|
|
|
5165 fprintf(fd, "/!%.*s\n", len, word);
|
|
|
5166 else
|
|
|
5167 fprintf(fd, "%.*s\n", len, word);
|
|
|
5168 fclose(fd);
|
|
|
5169
|
|
|
5170 /* Update the .add.spl file. */
|
|
323
|
5171 mkspell(1, &curbuf->b_p_spf, FALSE, TRUE, TRUE);
|
|
310
|
5172
|
|
|
5173 /* If the .add file is edited somewhere, reload it. */
|
|
|
5174 if (buf != NULL)
|
|
|
5175 buf_reload(buf);
|
|
323
|
5176
|
|
|
5177 redraw_all_later(NOT_VALID);
|
|
310
|
5178 }
|
|
|
5179 }
|
|
|
5180 }
|
|
|
5181 }
|
|
|
5182
|
|
|
5183 /*
|
|
|
5184 * Initialize 'spellfile' for the current buffer.
|
|
|
5185 */
|
|
|
5186 static void
|
|
|
5187 init_spellfile()
|
|
|
5188 {
|
|
|
5189 char_u buf[MAXPATHL];
|
|
|
5190 int l;
|
|
|
5191 slang_T *sl;
|
|
|
5192 char_u *rtp;
|
|
351
|
5193 char_u *lend;
|
|
310
|
5194
|
|
|
5195 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
5196 {
|
|
351
|
5197 /* Find the end of the language name. Exclude the region. */
|
|
|
5198 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
5199 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
5200 ;
|
|
|
5201
|
|
|
5202 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
5203 * are allowed to write. */
|
|
310
|
5204 rtp = p_rtp;
|
|
|
5205 while (*rtp != NUL)
|
|
|
5206 {
|
|
|
5207 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
5208 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
5209 if (filewritable(buf) == 2)
|
|
|
5210 {
|
|
316
|
5211 /* Use the first language name from 'spelllang' and the
|
|
|
5212 * encoding used in the first loaded .spl file. */
|
|
310
|
5213 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
5214 l = STRLEN(buf);
|
|
|
5215 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
5216 "/spell/%.*s.%s.add",
|
|
351
|
5217 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
5218 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
5219 ? (char_u *)"ascii" : spell_enc());
|
|
|
5220 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
5221 break;
|
|
|
5222 }
|
|
|
5223 }
|
|
|
5224 }
|
|
|
5225 }
|
|
236
|
5226
|
|
300
|
5227
|
|
307
|
5228 /*
|
|
|
5229 * Init the chartab used for spelling for ASCII.
|
|
|
5230 * EBCDIC is not supported!
|
|
|
5231 */
|
|
|
5232 static void
|
|
|
5233 clear_spell_chartab(sp)
|
|
|
5234 spelltab_T *sp;
|
|
|
5235 {
|
|
324
|
5236 int i;
|
|
307
|
5237
|
|
|
5238 /* Init everything to FALSE. */
|
|
|
5239 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
5240 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
5241 for (i = 0; i < 256; ++i)
|
|
324
|
5242 {
|
|
307
|
5243 sp->st_fold[i] = i;
|
|
324
|
5244 sp->st_upper[i] = i;
|
|
|
5245 }
|
|
307
|
5246
|
|
|
5247 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
5248 * that is done separately. */
|
|
|
5249 for (i = '0'; i <= '9'; ++i)
|
|
|
5250 sp->st_isw[i] = TRUE;
|
|
|
5251 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
5252 {
|
|
|
5253 sp->st_isw[i] = TRUE;
|
|
|
5254 sp->st_isu[i] = TRUE;
|
|
|
5255 sp->st_fold[i] = i + 0x20;
|
|
|
5256 }
|
|
|
5257 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
5258 {
|
|
307
|
5259 sp->st_isw[i] = TRUE;
|
|
324
|
5260 sp->st_upper[i] = i - 0x20;
|
|
|
5261 }
|
|
307
|
5262 }
|
|
|
5263
|
|
|
5264 /*
|
|
|
5265 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
5266 * Called once while starting up and when 'encoding' changes.
|
|
|
5267 * The default is to use isalpha(), but the spell file should define the word
|
|
|
5268 * characters to make it possible that 'encoding' differs from the current
|
|
|
5269 * locale.
|
|
|
5270 */
|
|
|
5271 void
|
|
|
5272 init_spell_chartab()
|
|
|
5273 {
|
|
|
5274 int i;
|
|
|
5275
|
|
|
5276 did_set_spelltab = FALSE;
|
|
|
5277 clear_spell_chartab(&spelltab);
|
|
366
|
5278 vim_memset(spell_ismw, FALSE, sizeof(spell_ismw));
|
|
307
|
5279 #ifdef FEAT_MBYTE
|
|
366
|
5280 vim_free(spell_ismw_mb);
|
|
|
5281 spell_ismw_mb = NULL;
|
|
|
5282
|
|
307
|
5283 if (enc_dbcs)
|
|
|
5284 {
|
|
|
5285 /* DBCS: assume double-wide characters are word characters. */
|
|
|
5286 for (i = 128; i <= 255; ++i)
|
|
|
5287 if (MB_BYTE2LEN(i) == 2)
|
|
|
5288 spelltab.st_isw[i] = TRUE;
|
|
|
5289 }
|
|
324
|
5290 else if (enc_utf8)
|
|
|
5291 {
|
|
|
5292 for (i = 128; i < 256; ++i)
|
|
|
5293 {
|
|
|
5294 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
5295 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
5296 spelltab.st_fold[i] = utf_fold(i);
|
|
|
5297 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
5298 }
|
|
|
5299 }
|
|
307
|
5300 else
|
|
|
5301 #endif
|
|
|
5302 {
|
|
324
|
5303 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
5304 for (i = 128; i < 256; ++i)
|
|
|
5305 {
|
|
|
5306 if (MB_ISUPPER(i))
|
|
|
5307 {
|
|
324
|
5308 spelltab.st_isw[i] = TRUE;
|
|
307
|
5309 spelltab.st_isu[i] = TRUE;
|
|
|
5310 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
5311 }
|
|
324
|
5312 else if (MB_ISLOWER(i))
|
|
|
5313 {
|
|
|
5314 spelltab.st_isw[i] = TRUE;
|
|
|
5315 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
5316 }
|
|
307
|
5317 }
|
|
|
5318 }
|
|
|
5319 }
|
|
|
5320
|
|
|
5321 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
5322 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
5323
|
|
|
5324 /*
|
|
|
5325 * Set the spell character tables from strings in the affix file.
|
|
|
5326 */
|
|
|
5327 static int
|
|
|
5328 set_spell_chartab(fol, low, upp)
|
|
|
5329 char_u *fol;
|
|
|
5330 char_u *low;
|
|
|
5331 char_u *upp;
|
|
|
5332 {
|
|
|
5333 /* We build the new tables here first, so that we can compare with the
|
|
|
5334 * previous one. */
|
|
|
5335 spelltab_T new_st;
|
|
|
5336 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
5337 int f, l, u;
|
|
|
5338
|
|
|
5339 clear_spell_chartab(&new_st);
|
|
|
5340
|
|
|
5341 while (*pf != NUL)
|
|
|
5342 {
|
|
|
5343 if (*pl == NUL || *pu == NUL)
|
|
|
5344 {
|
|
|
5345 EMSG(_(e_affform));
|
|
|
5346 return FAIL;
|
|
|
5347 }
|
|
|
5348 #ifdef FEAT_MBYTE
|
|
|
5349 f = mb_ptr2char_adv(&pf);
|
|
|
5350 l = mb_ptr2char_adv(&pl);
|
|
|
5351 u = mb_ptr2char_adv(&pu);
|
|
|
5352 #else
|
|
|
5353 f = *pf++;
|
|
|
5354 l = *pl++;
|
|
|
5355 u = *pu++;
|
|
|
5356 #endif
|
|
|
5357 /* Every character that appears is a word character. */
|
|
|
5358 if (f < 256)
|
|
|
5359 new_st.st_isw[f] = TRUE;
|
|
|
5360 if (l < 256)
|
|
|
5361 new_st.st_isw[l] = TRUE;
|
|
|
5362 if (u < 256)
|
|
|
5363 new_st.st_isw[u] = TRUE;
|
|
|
5364
|
|
|
5365 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
5366 * case-folding */
|
|
|
5367 if (l < 256 && l != f)
|
|
|
5368 {
|
|
|
5369 if (f >= 256)
|
|
|
5370 {
|
|
|
5371 EMSG(_(e_affrange));
|
|
|
5372 return FAIL;
|
|
|
5373 }
|
|
|
5374 new_st.st_fold[l] = f;
|
|
|
5375 }
|
|
|
5376
|
|
|
5377 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
5378 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
5379 * "FOL" . */
|
|
307
|
5380 if (u < 256 && u != f)
|
|
|
5381 {
|
|
|
5382 if (f >= 256)
|
|
|
5383 {
|
|
|
5384 EMSG(_(e_affrange));
|
|
|
5385 return FAIL;
|
|
|
5386 }
|
|
|
5387 new_st.st_fold[u] = f;
|
|
|
5388 new_st.st_isu[u] = TRUE;
|
|
324
|
5389 new_st.st_upper[f] = u;
|
|
307
|
5390 }
|
|
|
5391 }
|
|
|
5392
|
|
|
5393 if (*pl != NUL || *pu != NUL)
|
|
|
5394 {
|
|
|
5395 EMSG(_(e_affform));
|
|
|
5396 return FAIL;
|
|
|
5397 }
|
|
|
5398
|
|
|
5399 return set_spell_finish(&new_st);
|
|
|
5400 }
|
|
|
5401
|
|
|
5402 /*
|
|
|
5403 * Set the spell character tables from strings in the .spl file.
|
|
|
5404 */
|
|
|
5405 static int
|
|
|
5406 set_spell_charflags(flags, cnt, upp)
|
|
|
5407 char_u *flags;
|
|
|
5408 int cnt;
|
|
|
5409 char_u *upp;
|
|
|
5410 {
|
|
|
5411 /* We build the new tables here first, so that we can compare with the
|
|
|
5412 * previous one. */
|
|
|
5413 spelltab_T new_st;
|
|
|
5414 int i;
|
|
|
5415 char_u *p = upp;
|
|
324
|
5416 int c;
|
|
307
|
5417
|
|
|
5418 clear_spell_chartab(&new_st);
|
|
|
5419
|
|
|
5420 for (i = 0; i < cnt; ++i)
|
|
|
5421 {
|
|
324
|
5422 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
5423 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
307
|
5424
|
|
|
5425 if (*p == NUL)
|
|
|
5426 return FAIL;
|
|
|
5427 #ifdef FEAT_MBYTE
|
|
324
|
5428 c = mb_ptr2char_adv(&p);
|
|
307
|
5429 #else
|
|
324
|
5430 c = *p++;
|
|
307
|
5431 #endif
|
|
324
|
5432 new_st.st_fold[i + 128] = c;
|
|
|
5433 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
5434 new_st.st_upper[c] = i + 128;
|
|
307
|
5435 }
|
|
|
5436
|
|
|
5437 return set_spell_finish(&new_st);
|
|
|
5438 }
|
|
|
5439
|
|
|
5440 static int
|
|
|
5441 set_spell_finish(new_st)
|
|
|
5442 spelltab_T *new_st;
|
|
|
5443 {
|
|
|
5444 int i;
|
|
|
5445
|
|
|
5446 if (did_set_spelltab)
|
|
|
5447 {
|
|
|
5448 /* check that it's the same table */
|
|
|
5449 for (i = 0; i < 256; ++i)
|
|
|
5450 {
|
|
|
5451 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
5452 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
5453 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
5454 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
5455 {
|
|
|
5456 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
5457 return FAIL;
|
|
|
5458 }
|
|
|
5459 }
|
|
|
5460 }
|
|
|
5461 else
|
|
|
5462 {
|
|
|
5463 /* copy the new spelltab into the one being used */
|
|
|
5464 spelltab = *new_st;
|
|
|
5465 did_set_spelltab = TRUE;
|
|
|
5466 }
|
|
|
5467
|
|
|
5468 return OK;
|
|
|
5469 }
|
|
|
5470
|
|
|
5471 /*
|
|
358
|
5472 * Return TRUE if "p" points to a word character.
|
|
366
|
5473 * As a special case we see "midword" characters as word character when it is
|
|
358
|
5474 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
5475 * Thus this only works properly when past the first character of the word.
|
|
358
|
5476 */
|
|
|
5477 static int
|
|
|
5478 spell_iswordp(p)
|
|
|
5479 char_u *p;
|
|
|
5480 {
|
|
366
|
5481 #ifdef FEAT_MBYTE
|
|
358
|
5482 char_u *s;
|
|
366
|
5483 int l;
|
|
|
5484 int c;
|
|
|
5485
|
|
|
5486 if (has_mbyte)
|
|
|
5487 {
|
|
|
5488 l = MB_BYTE2LEN(*p);
|
|
358
|
5489 s = p;
|
|
366
|
5490 if (l == 1)
|
|
|
5491 {
|
|
|
5492 /* be quick for ASCII */
|
|
|
5493 if (spell_ismw[*p])
|
|
|
5494 {
|
|
|
5495 s = p + 1; /* skip a mid-word character */
|
|
|
5496 l = MB_BYTE2LEN(*s);
|
|
|
5497 }
|
|
|
5498 }
|
|
|
5499 else
|
|
|
5500 {
|
|
|
5501 c = mb_ptr2char(p);
|
|
|
5502 if (c < 256 ? spell_ismw[c] : (spell_ismw_mb != NULL
|
|
|
5503 && vim_strchr(spell_ismw_mb, c) != NULL))
|
|
|
5504 {
|
|
|
5505 s = p + l;
|
|
|
5506 l = MB_BYTE2LEN(*s);
|
|
|
5507 }
|
|
|
5508 }
|
|
|
5509
|
|
|
5510 if (l > 1)
|
|
|
5511 return mb_get_class(s) >= 2;
|
|
|
5512 return spelltab.st_isw[*s];
|
|
|
5513 }
|
|
358
|
5514 #endif
|
|
366
|
5515
|
|
|
5516 return spelltab.st_isw[spell_ismw[*p] ? p[1] : p[0]];
|
|
358
|
5517 }
|
|
|
5518
|
|
372
|
5519 #ifdef FEAT_MBYTE
|
|
|
5520 /*
|
|
|
5521 * Return TRUE if "p" points to a word character.
|
|
|
5522 * Wide version of spell_iswordp().
|
|
|
5523 */
|
|
|
5524 static int
|
|
|
5525 spell_iswordp_w(p)
|
|
|
5526 int *p;
|
|
|
5527 {
|
|
|
5528 int *s;
|
|
|
5529
|
|
|
5530 if (*p < 256 ? spell_ismw[*p] : (spell_ismw_mb != NULL
|
|
|
5531 && vim_strchr(spell_ismw_mb, *p) != NULL))
|
|
|
5532 s = p + 1;
|
|
|
5533 else
|
|
|
5534 s = p;
|
|
|
5535
|
|
|
5536 if (mb_char2len(*s) > 1)
|
|
|
5537 {
|
|
|
5538 if (enc_utf8)
|
|
|
5539 return utf_class(*s) >= 2;
|
|
|
5540 if (enc_dbcs)
|
|
|
5541 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
5542 return 0;
|
|
|
5543 }
|
|
|
5544 return spelltab.st_isw[*s];
|
|
|
5545 }
|
|
|
5546 #endif
|
|
|
5547
|
|
358
|
5548 /*
|
|
339
|
5549 * Write the table with prefix conditions to the .spl file.
|
|
|
5550 */
|
|
|
5551 static void
|
|
|
5552 write_spell_prefcond(fd, gap)
|
|
|
5553 FILE *fd;
|
|
|
5554 garray_T *gap;
|
|
|
5555 {
|
|
|
5556 int i;
|
|
|
5557 char_u *p;
|
|
|
5558 int len;
|
|
|
5559
|
|
|
5560 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
5561
|
|
|
5562 for (i = 0; i < gap->ga_len; ++i)
|
|
|
5563 {
|
|
|
5564 /* <prefcond> : <condlen> <condstr> */
|
|
|
5565 p = ((char_u **)gap->ga_data)[i];
|
|
|
5566 if (p == NULL)
|
|
|
5567 fputc(0, fd);
|
|
|
5568 else
|
|
|
5569 {
|
|
|
5570 len = STRLEN(p);
|
|
|
5571 fputc(len, fd);
|
|
|
5572 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
5573 }
|
|
|
5574 }
|
|
|
5575 }
|
|
|
5576
|
|
|
5577 /*
|
|
307
|
5578 * Write the current tables into the .spl file.
|
|
|
5579 * This makes sure the same characters are recognized as word characters when
|
|
|
5580 * generating an when using a spell file.
|
|
|
5581 */
|
|
|
5582 static void
|
|
|
5583 write_spell_chartab(fd)
|
|
|
5584 FILE *fd;
|
|
|
5585 {
|
|
|
5586 char_u charbuf[256 * 4];
|
|
|
5587 int len = 0;
|
|
|
5588 int flags;
|
|
|
5589 int i;
|
|
|
5590
|
|
|
5591 fputc(128, fd); /* <charflagslen> */
|
|
|
5592 for (i = 128; i < 256; ++i)
|
|
|
5593 {
|
|
|
5594 flags = 0;
|
|
|
5595 if (spelltab.st_isw[i])
|
|
324
|
5596 flags |= CF_WORD;
|
|
307
|
5597 if (spelltab.st_isu[i])
|
|
324
|
5598 flags |= CF_UPPER;
|
|
307
|
5599 fputc(flags, fd); /* <charflags> */
|
|
|
5600
|
|
310
|
5601 #ifdef FEAT_MBYTE
|
|
|
5602 if (has_mbyte)
|
|
|
5603 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
|
5604 else
|
|
|
5605 #endif
|
|
|
5606 charbuf[len++] = spelltab.st_fold[i];
|
|
307
|
5607 }
|
|
|
5608
|
|
|
5609 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
|
5610 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
|
5611 }
|
|
|
5612
|
|
|
5613 /*
|
|
324
|
5614 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
5615 * Uses the character definitions from the .spl file.
|
|
307
|
5616 * When using a multi-byte 'encoding' the length may change!
|
|
|
5617 * Returns FAIL when something wrong.
|
|
|
5618 */
|
|
|
5619 static int
|
|
324
|
5620 spell_casefold(str, len, buf, buflen)
|
|
|
5621 char_u *str;
|
|
307
|
5622 int len;
|
|
|
5623 char_u *buf;
|
|
|
5624 int buflen;
|
|
|
5625 {
|
|
|
5626 int i;
|
|
|
5627
|
|
|
5628 if (len >= buflen)
|
|
|
5629 {
|
|
|
5630 buf[0] = NUL;
|
|
|
5631 return FAIL; /* result will not fit */
|
|
|
5632 }
|
|
|
5633
|
|
|
5634 #ifdef FEAT_MBYTE
|
|
|
5635 if (has_mbyte)
|
|
|
5636 {
|
|
324
|
5637 int outi = 0;
|
|
|
5638 char_u *p;
|
|
307
|
5639 int c;
|
|
|
5640
|
|
|
5641 /* Fold one character at a time. */
|
|
324
|
5642 for (p = str; p < str + len; )
|
|
307
|
5643 {
|
|
|
5644 if (outi + MB_MAXBYTES > buflen)
|
|
|
5645 {
|
|
|
5646 buf[outi] = NUL;
|
|
|
5647 return FAIL;
|
|
|
5648 }
|
|
324
|
5649 c = mb_ptr2char_adv(&p);
|
|
|
5650 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
5651 }
|
|
|
5652 buf[outi] = NUL;
|
|
|
5653 }
|
|
|
5654 else
|
|
|
5655 #endif
|
|
|
5656 {
|
|
|
5657 /* Be quick for non-multibyte encodings. */
|
|
|
5658 for (i = 0; i < len; ++i)
|
|
324
|
5659 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
5660 buf[i] = NUL;
|
|
|
5661 }
|
|
|
5662
|
|
|
5663 return OK;
|
|
|
5664 }
|
|
|
5665
|
|
372
|
5666 #define SPS_BEST 1
|
|
|
5667 #define SPS_FAST 2
|
|
|
5668 #define SPS_DOUBLE 4
|
|
|
5669
|
|
|
5670 static int sps_flags = SPS_BEST;
|
|
|
5671
|
|
|
5672 /*
|
|
|
5673 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
|
5674 * Sets "sps_flags".
|
|
|
5675 */
|
|
|
5676 int
|
|
|
5677 spell_check_sps()
|
|
|
5678 {
|
|
|
5679 char_u *p;
|
|
|
5680 char_u buf[MAXPATHL];
|
|
|
5681 int f;
|
|
|
5682
|
|
|
5683 sps_flags = 0;
|
|
|
5684
|
|
|
5685 for (p = p_sps; *p != NUL; )
|
|
|
5686 {
|
|
|
5687 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
5688
|
|
|
5689 f = 0;
|
|
|
5690 if (STRCMP(buf, "best") == 0)
|
|
|
5691 f = SPS_BEST;
|
|
|
5692 else if (STRCMP(buf, "fast") == 0)
|
|
|
5693 f = SPS_FAST;
|
|
|
5694 else if (STRCMP(buf, "double") == 0)
|
|
|
5695 f = SPS_DOUBLE;
|
|
|
5696 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
5697 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
5698 f = -1;
|
|
|
5699
|
|
|
5700 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
5701 {
|
|
|
5702 sps_flags = SPS_BEST;
|
|
|
5703 return FAIL;
|
|
|
5704 }
|
|
|
5705 if (f != 0)
|
|
|
5706 sps_flags = f;
|
|
|
5707 }
|
|
|
5708
|
|
|
5709 if (sps_flags == 0)
|
|
|
5710 sps_flags = SPS_BEST;
|
|
|
5711
|
|
|
5712 return OK;
|
|
|
5713 }
|
|
|
5714
|
|
|
5715 /* Remember what "z?" replaced. */
|
|
|
5716 static char_u *repl_from = NULL;
|
|
|
5717 static char_u *repl_to = NULL;
|
|
|
5718
|
|
323
|
5719 /*
|
|
|
5720 * "z?": Find badly spelled word under or after the cursor.
|
|
|
5721 * Give suggestions for the properly spelled word.
|
|
|
5722 */
|
|
|
5723 void
|
|
|
5724 spell_suggest()
|
|
|
5725 {
|
|
|
5726 char_u *line;
|
|
|
5727 pos_T prev_cursor = curwin->w_cursor;
|
|
|
5728 char_u wcopy[MAXWLEN + 2];
|
|
|
5729 char_u *p;
|
|
|
5730 int i;
|
|
|
5731 int c;
|
|
|
5732 suginfo_T sug;
|
|
|
5733 suggest_T *stp;
|
|
372
|
5734 int mouse_used;
|
|
323
|
5735
|
|
344
|
5736 /* Find the start of the badly spelled word. */
|
|
346
|
5737 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
5738 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
5739 {
|
|
|
5740 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
5741 return;
|
|
|
5742
|
|
|
5743 /* No bad word or it starts after the cursor: use the word under the
|
|
|
5744 * cursor. */
|
|
|
5745 curwin->w_cursor = prev_cursor;
|
|
|
5746 line = ml_get_curline();
|
|
|
5747 p = line + curwin->w_cursor.col;
|
|
|
5748 /* Backup to before start of word. */
|
|
|
5749 while (p > line && SPELL_ISWORDP(p))
|
|
|
5750 mb_ptr_back(line, p);
|
|
|
5751 /* Forward to start of word. */
|
|
375
|
5752 while (*p != NUL && !SPELL_ISWORDP(p))
|
|
346
|
5753 mb_ptr_adv(p);
|
|
|
5754
|
|
|
5755 if (!SPELL_ISWORDP(p)) /* No word found. */
|
|
|
5756 {
|
|
|
5757 beep_flush();
|
|
|
5758 return;
|
|
|
5759 }
|
|
|
5760 curwin->w_cursor.col = p - line;
|
|
323
|
5761 }
|
|
|
5762
|
|
344
|
5763 /* Get the word and its length. */
|
|
323
|
5764 line = ml_get_curline();
|
|
344
|
5765
|
|
|
5766 /* Get the list of suggestions */
|
|
358
|
5767 spell_find_suggest(line + curwin->w_cursor.col, &sug, (int)Rows - 2, TRUE);
|
|
323
|
5768
|
|
|
5769 if (sug.su_ga.ga_len == 0)
|
|
|
5770 MSG(_("Sorry, no suggestions"));
|
|
|
5771 else
|
|
|
5772 {
|
|
372
|
5773 vim_free(repl_from);
|
|
|
5774 repl_from = NULL;
|
|
|
5775 vim_free(repl_to);
|
|
|
5776 repl_to = NULL;
|
|
|
5777
|
|
323
|
5778 /* List the suggestions. */
|
|
|
5779 msg_start();
|
|
372
|
5780 lines_left = Rows; /* avoid more prompt */
|
|
323
|
5781 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
5782 sug.su_badlen, sug.su_badptr);
|
|
|
5783 msg_puts(IObuff);
|
|
|
5784 msg_clr_eos();
|
|
|
5785 msg_putchar('\n');
|
|
346
|
5786
|
|
323
|
5787 msg_scroll = TRUE;
|
|
|
5788 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
5789 {
|
|
344
|
5790 stp = &SUG(sug.su_ga, i);
|
|
323
|
5791
|
|
|
5792 /* The suggested word may replace only part of the bad word, add
|
|
|
5793 * the not replaced part. */
|
|
|
5794 STRCPY(wcopy, stp->st_word);
|
|
|
5795 if (sug.su_badlen > stp->st_orglen)
|
|
|
5796 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
5797 sug.su_badptr + stp->st_orglen,
|
|
|
5798 sug.su_badlen - stp->st_orglen);
|
|
346
|
5799 vim_snprintf((char *)IObuff, IOSIZE, _("%2d \"%s\""), i + 1, wcopy);
|
|
|
5800 msg_puts(IObuff);
|
|
|
5801
|
|
|
5802 /* The word may replace more than "su_badlen". */
|
|
|
5803 if (sug.su_badlen < stp->st_orglen)
|
|
|
5804 {
|
|
|
5805 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
5806 stp->st_orglen, sug.su_badptr);
|
|
|
5807 msg_puts(IObuff);
|
|
|
5808 }
|
|
|
5809
|
|
324
|
5810 if (p_verbose > 0)
|
|
344
|
5811 {
|
|
346
|
5812 /* Add the score. */
|
|
351
|
5813 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
346
|
5814 vim_snprintf((char *)IObuff, IOSIZE, _(" (%s%d - %d)"),
|
|
344
|
5815 stp->st_salscore ? "s " : "",
|
|
|
5816 stp->st_score, stp->st_altscore);
|
|
|
5817 else
|
|
346
|
5818 vim_snprintf((char *)IObuff, IOSIZE, _(" (%d)"),
|
|
|
5819 stp->st_score);
|
|
|
5820 msg_advance(30);
|
|
|
5821 msg_puts(IObuff);
|
|
344
|
5822 }
|
|
323
|
5823 msg_putchar('\n');
|
|
|
5824 }
|
|
|
5825
|
|
|
5826 /* Ask for choice. */
|
|
372
|
5827 i = prompt_for_number(&mouse_used);
|
|
|
5828 if (mouse_used)
|
|
|
5829 i -= lines_left;
|
|
|
5830
|
|
344
|
5831 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
5832 {
|
|
372
|
5833 /* Save the from and to text for :spellrepall. */
|
|
|
5834 stp = &SUG(sug.su_ga, i - 1);
|
|
|
5835 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
5836 repl_to = vim_strsave(stp->st_word);
|
|
|
5837
|
|
323
|
5838 /* Replace the word. */
|
|
|
5839 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
5840 if (p != NULL)
|
|
|
5841 {
|
|
|
5842 c = sug.su_badptr - line;
|
|
|
5843 mch_memmove(p, line, c);
|
|
|
5844 STRCPY(p + c, stp->st_word);
|
|
|
5845 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
5846 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
5847 curwin->w_cursor.col = c;
|
|
|
5848 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
5849
|
|
|
5850 /* For redo we use a change-word command. */
|
|
|
5851 ResetRedobuff();
|
|
|
5852 AppendToRedobuff((char_u *)"ciw");
|
|
|
5853 AppendToRedobuff(stp->st_word);
|
|
|
5854 AppendCharToRedobuff(ESC);
|
|
323
|
5855 }
|
|
|
5856 }
|
|
|
5857 else
|
|
|
5858 curwin->w_cursor = prev_cursor;
|
|
|
5859 }
|
|
|
5860
|
|
344
|
5861 spell_find_cleanup(&sug);
|
|
|
5862 }
|
|
|
5863
|
|
|
5864 /*
|
|
372
|
5865 * ":spellrepall"
|
|
|
5866 */
|
|
|
5867 /*ARGSUSED*/
|
|
|
5868 void
|
|
|
5869 ex_spellrepall(eap)
|
|
|
5870 exarg_T *eap;
|
|
|
5871 {
|
|
|
5872 pos_T pos = curwin->w_cursor;
|
|
|
5873 char_u *frompat;
|
|
|
5874 int addlen;
|
|
|
5875 char_u *line;
|
|
|
5876 char_u *p;
|
|
|
5877 int didone = FALSE;
|
|
|
5878 int save_ws = p_ws;
|
|
|
5879
|
|
|
5880 if (repl_from == NULL || repl_to == NULL)
|
|
|
5881 {
|
|
|
5882 EMSG(_("E752: No previous spell replacement"));
|
|
|
5883 return;
|
|
|
5884 }
|
|
|
5885 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
5886
|
|
|
5887 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
5888 if (frompat == NULL)
|
|
|
5889 return;
|
|
|
5890 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
5891 p_ws = FALSE;
|
|
|
5892
|
|
|
5893 curwin->w_cursor.lnum = 0;
|
|
|
5894 while (!got_int)
|
|
|
5895 {
|
|
|
5896 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
5897 || u_save_cursor() == FAIL)
|
|
|
5898 break;
|
|
|
5899
|
|
|
5900 /* Only replace when the right word isn't there yet. This happens
|
|
|
5901 * when changing "etc" to "etc.". */
|
|
|
5902 line = ml_get_curline();
|
|
|
5903 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
5904 repl_to, STRLEN(repl_to)) != 0)
|
|
|
5905 {
|
|
|
5906 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
5907 if (p == NULL)
|
|
|
5908 break;
|
|
|
5909 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
5910 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
5911 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
5912 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
5913 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
5914 didone = TRUE;
|
|
|
5915 }
|
|
|
5916 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
5917 }
|
|
|
5918
|
|
|
5919 p_ws = save_ws;
|
|
|
5920 curwin->w_cursor = pos;
|
|
|
5921 vim_free(frompat);
|
|
|
5922
|
|
|
5923 if (!didone)
|
|
|
5924 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
|
5925 }
|
|
|
5926
|
|
|
5927 /*
|
|
344
|
5928 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
5929 * a list of allocated strings.
|
|
|
5930 */
|
|
|
5931 void
|
|
|
5932 spell_suggest_list(gap, word, maxcount)
|
|
|
5933 garray_T *gap;
|
|
|
5934 char_u *word;
|
|
|
5935 int maxcount; /* maximum nr of suggestions */
|
|
|
5936 {
|
|
|
5937 suginfo_T sug;
|
|
|
5938 int i;
|
|
|
5939 suggest_T *stp;
|
|
|
5940 char_u *wcopy;
|
|
|
5941
|
|
358
|
5942 spell_find_suggest(word, &sug, maxcount, FALSE);
|
|
344
|
5943
|
|
|
5944 /* Make room in "gap". */
|
|
|
5945 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
5946 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
5947 return;
|
|
|
5948
|
|
323
|
5949 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
5950 {
|
|
|
5951 stp = &SUG(sug.su_ga, i);
|
|
|
5952
|
|
|
5953 /* The suggested word may replace only part of "word", add the not
|
|
|
5954 * replaced part. */
|
|
|
5955 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
5956 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
5957 if (wcopy == NULL)
|
|
|
5958 break;
|
|
|
5959 STRCPY(wcopy, stp->st_word);
|
|
|
5960 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
5961 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
5962 }
|
|
|
5963
|
|
|
5964 spell_find_cleanup(&sug);
|
|
|
5965 }
|
|
|
5966
|
|
|
5967 /*
|
|
|
5968 * Find spell suggestions for the word at the start of "badptr".
|
|
|
5969 * Return the suggestions in "su->su_ga".
|
|
|
5970 * The maximum number of suggestions is "maxcount".
|
|
|
5971 * Note: does use info for the current window.
|
|
|
5972 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
5973 */
|
|
|
5974 static void
|
|
358
|
5975 spell_find_suggest(badptr, su, maxcount, banbadword)
|
|
344
|
5976 char_u *badptr;
|
|
|
5977 suginfo_T *su;
|
|
|
5978 int maxcount;
|
|
358
|
5979 int banbadword; /* don't include badword in suggestions */
|
|
344
|
5980 {
|
|
|
5981 int attr;
|
|
372
|
5982 char_u buf[MAXPATHL];
|
|
|
5983 char_u *p;
|
|
|
5984 int do_combine = FALSE;
|
|
|
5985 char_u *sps_copy;
|
|
|
5986 #ifdef FEAT_EVAL
|
|
|
5987 static int expr_busy = FALSE;
|
|
|
5988 #endif
|
|
344
|
5989
|
|
|
5990 /*
|
|
|
5991 * Set the info in "*su".
|
|
|
5992 */
|
|
|
5993 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
5994 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
5995 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
5996 if (*badptr == NUL)
|
|
|
5997 return;
|
|
344
|
5998 hash_init(&su->su_banned);
|
|
|
5999
|
|
|
6000 su->su_badptr = badptr;
|
|
|
6001 su->su_badlen = spell_check(curwin, su->su_badptr, &attr);
|
|
|
6002 su->su_maxcount = maxcount;
|
|
372
|
6003 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
6004
|
|
|
6005 if (su->su_badlen >= MAXWLEN)
|
|
|
6006 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
6007 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
6008 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
6009 su->su_fbadword, MAXWLEN);
|
|
346
|
6010 /* get caps flags for bad word */
|
|
|
6011 su->su_badflags = captype(su->su_badptr, su->su_badptr + su->su_badlen);
|
|
344
|
6012
|
|
|
6013 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
6014 if (banbadword)
|
|
|
6015 add_banned(su, su->su_badword);
|
|
344
|
6016
|
|
372
|
6017 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
6018 sps_copy = vim_strsave(p_sps);
|
|
|
6019 if (sps_copy == NULL)
|
|
|
6020 return;
|
|
|
6021
|
|
|
6022 /* Loop over the items in 'spellsuggest'. */
|
|
|
6023 for (p = sps_copy; *p != NUL; )
|
|
|
6024 {
|
|
|
6025 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6026
|
|
|
6027 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
6028 {
|
|
|
6029 #ifdef FEAT_EVAL
|
|
375
|
6030 /* Evaluate an expression. Skip this when called recursively,
|
|
|
6031 * when using spellsuggest() in the expression. */
|
|
372
|
6032 if (!expr_busy)
|
|
|
6033 {
|
|
|
6034 expr_busy = TRUE;
|
|
|
6035 spell_suggest_expr(su, buf + 5);
|
|
|
6036 expr_busy = FALSE;
|
|
|
6037 }
|
|
|
6038 #endif
|
|
|
6039 }
|
|
|
6040 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
6041 /* Use list of suggestions in a file. */
|
|
|
6042 spell_suggest_file(su, buf + 5);
|
|
|
6043 else
|
|
|
6044 {
|
|
|
6045 /* Use internal method. */
|
|
|
6046 spell_suggest_intern(su);
|
|
|
6047 if (sps_flags & SPS_DOUBLE)
|
|
|
6048 do_combine = TRUE;
|
|
|
6049 }
|
|
|
6050 }
|
|
|
6051
|
|
|
6052 vim_free(sps_copy);
|
|
|
6053
|
|
|
6054 if (do_combine)
|
|
|
6055 /* Combine the two list of suggestions. This must be done last,
|
|
|
6056 * because sorting changes the order again. */
|
|
|
6057 score_combine(su);
|
|
|
6058 }
|
|
|
6059
|
|
|
6060 #ifdef FEAT_EVAL
|
|
|
6061 /*
|
|
|
6062 * Find suggestions by evaluating expression "expr".
|
|
|
6063 */
|
|
|
6064 static void
|
|
|
6065 spell_suggest_expr(su, expr)
|
|
|
6066 suginfo_T *su;
|
|
|
6067 char_u *expr;
|
|
|
6068 {
|
|
|
6069 list_T *list;
|
|
|
6070 listitem_T *li;
|
|
|
6071 int score;
|
|
|
6072 char_u *p;
|
|
|
6073
|
|
|
6074 /* The work is split up in a few parts to avoid having to export
|
|
|
6075 * suginfo_T.
|
|
|
6076 * First evaluate the expression and get the resulting list. */
|
|
|
6077 list = eval_spell_expr(su->su_badword, expr);
|
|
|
6078 if (list != NULL)
|
|
|
6079 {
|
|
|
6080 /* Loop over the items in the list. */
|
|
|
6081 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
6082 if (li->li_tv.v_type == VAR_LIST)
|
|
|
6083 {
|
|
|
6084 /* Get the word and the score from the items. */
|
|
|
6085 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
6086 if (score >= 0)
|
|
|
6087 add_suggestion(su, &su->su_ga, p,
|
|
|
6088 su->su_badlen, score, 0, TRUE);
|
|
|
6089 }
|
|
|
6090 list_unref(list);
|
|
|
6091 }
|
|
|
6092
|
|
|
6093 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6094 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6095 }
|
|
|
6096 #endif
|
|
|
6097
|
|
|
6098 /*
|
|
|
6099 * Find suggestions a file "fname".
|
|
|
6100 */
|
|
|
6101 static void
|
|
|
6102 spell_suggest_file(su, fname)
|
|
|
6103 suginfo_T *su;
|
|
|
6104 char_u *fname;
|
|
|
6105 {
|
|
|
6106 FILE *fd;
|
|
|
6107 char_u line[MAXWLEN * 2];
|
|
|
6108 char_u *p;
|
|
|
6109 int len;
|
|
|
6110 char_u cword[MAXWLEN];
|
|
|
6111
|
|
|
6112 /* Open the file. */
|
|
|
6113 fd = mch_fopen((char *)fname, "r");
|
|
|
6114 if (fd == NULL)
|
|
|
6115 {
|
|
|
6116 EMSG2(_(e_notopen), fname);
|
|
|
6117 return;
|
|
|
6118 }
|
|
|
6119
|
|
|
6120 /* Read it line by line. */
|
|
|
6121 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
6122 {
|
|
|
6123 line_breakcheck();
|
|
|
6124
|
|
|
6125 p = vim_strchr(line, '/');
|
|
|
6126 if (p == NULL)
|
|
|
6127 continue; /* No Tab found, just skip the line. */
|
|
|
6128 *p++ = NUL;
|
|
|
6129 if (STRICMP(su->su_badword, line) == 0)
|
|
|
6130 {
|
|
|
6131 /* Match! Isolate the good word, until CR or NL. */
|
|
|
6132 for (len = 0; p[len] >= ' '; ++len)
|
|
|
6133 ;
|
|
|
6134 p[len] = NUL;
|
|
|
6135
|
|
|
6136 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
6137 * of the bad word. */
|
|
|
6138 if (captype(p, NULL) == 0)
|
|
|
6139 {
|
|
|
6140 make_case_word(p, cword, su->su_badflags);
|
|
|
6141 p = cword;
|
|
|
6142 }
|
|
|
6143
|
|
|
6144 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
6145 SCORE_FILE, 0, TRUE);
|
|
|
6146 }
|
|
|
6147 }
|
|
|
6148
|
|
|
6149 fclose(fd);
|
|
|
6150
|
|
|
6151 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6152 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6153 }
|
|
|
6154
|
|
|
6155 /*
|
|
|
6156 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
6157 */
|
|
|
6158 static void
|
|
|
6159 spell_suggest_intern(su)
|
|
|
6160 suginfo_T *su;
|
|
|
6161 {
|
|
344
|
6162 /*
|
|
346
|
6163 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
6164 *
|
|
|
6165 * Set a maximum score to limit the combination of operations that is
|
|
|
6166 * tried.
|
|
|
6167 */
|
|
346
|
6168 suggest_try_special(su);
|
|
|
6169
|
|
|
6170 /*
|
|
|
6171 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
6172 * from the .aff file and inserting a space (split the word).
|
|
|
6173 */
|
|
|
6174 suggest_try_change(su);
|
|
344
|
6175
|
|
|
6176 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
6177 if (sps_flags & SPS_DOUBLE)
|
|
|
6178 score_comp_sal(su);
|
|
|
6179
|
|
|
6180 /*
|
|
346
|
6181 * 3. Try finding sound-a-like words.
|
|
344
|
6182 *
|
|
|
6183 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
6184 * very slow and often doesn't find new suggestions.
|
|
|
6185 */
|
|
|
6186 if ((sps_flags & SPS_DOUBLE)
|
|
|
6187 || (!(sps_flags & SPS_FAST)
|
|
|
6188 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
6189 {
|
|
|
6190 /* Allow a higher score now. */
|
|
|
6191 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
6192 suggest_try_soundalike(su);
|
|
344
|
6193 }
|
|
|
6194
|
|
|
6195 /* When CTRL-C was hit while searching do show the results. */
|
|
|
6196 ui_breakcheck();
|
|
|
6197 if (got_int)
|
|
|
6198 {
|
|
|
6199 (void)vgetc();
|
|
|
6200 got_int = FALSE;
|
|
|
6201 }
|
|
|
6202
|
|
372
|
6203 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
6204 {
|
|
|
6205 if (sps_flags & SPS_BEST)
|
|
|
6206 /* Adjust the word score for how it sounds like. */
|
|
|
6207 rescore_suggestions(su);
|
|
|
6208
|
|
|
6209 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
6210 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
6211 }
|
|
|
6212 }
|
|
|
6213
|
|
|
6214 /*
|
|
|
6215 * Free the info put in "*su" by spell_find_suggest().
|
|
|
6216 */
|
|
|
6217 static void
|
|
|
6218 spell_find_cleanup(su)
|
|
|
6219 suginfo_T *su;
|
|
|
6220 {
|
|
|
6221 int i;
|
|
|
6222
|
|
|
6223 /* Free the suggestions. */
|
|
|
6224 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
6225 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
6226 ga_clear(&su->su_ga);
|
|
|
6227 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
6228 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
6229 ga_clear(&su->su_sga);
|
|
323
|
6230
|
|
|
6231 /* Free the banned words. */
|
|
344
|
6232 free_banned(su);
|
|
323
|
6233 }
|
|
|
6234
|
|
|
6235 /*
|
|
324
|
6236 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
6237 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
6238 * The result is NUL terminated.
|
|
323
|
6239 */
|
|
|
6240 static void
|
|
324
|
6241 onecap_copy(word, wcopy, upper)
|
|
323
|
6242 char_u *word;
|
|
|
6243 char_u *wcopy;
|
|
|
6244 int upper; /* TRUE: first letter made upper case */
|
|
|
6245 {
|
|
|
6246 char_u *p;
|
|
|
6247 int c;
|
|
|
6248 int l;
|
|
|
6249
|
|
|
6250 p = word;
|
|
|
6251 #ifdef FEAT_MBYTE
|
|
|
6252 if (has_mbyte)
|
|
|
6253 c = mb_ptr2char_adv(&p);
|
|
|
6254 else
|
|
|
6255 #endif
|
|
|
6256 c = *p++;
|
|
|
6257 if (upper)
|
|
324
|
6258 c = SPELL_TOUPPER(c);
|
|
323
|
6259 else
|
|
324
|
6260 c = SPELL_TOFOLD(c);
|
|
323
|
6261 #ifdef FEAT_MBYTE
|
|
|
6262 if (has_mbyte)
|
|
|
6263 l = mb_char2bytes(c, wcopy);
|
|
|
6264 else
|
|
|
6265 #endif
|
|
|
6266 {
|
|
|
6267 l = 1;
|
|
|
6268 wcopy[0] = c;
|
|
|
6269 }
|
|
324
|
6270 vim_strncpy(wcopy + l, p, MAXWLEN - l);
|
|
323
|
6271 }
|
|
|
6272
|
|
|
6273 /*
|
|
324
|
6274 * Make a copy of "word" with all the letters upper cased into
|
|
|
6275 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
6276 */
|
|
|
6277 static void
|
|
|
6278 allcap_copy(word, wcopy)
|
|
|
6279 char_u *word;
|
|
|
6280 char_u *wcopy;
|
|
|
6281 {
|
|
|
6282 char_u *s;
|
|
|
6283 char_u *d;
|
|
|
6284 int c;
|
|
|
6285
|
|
|
6286 d = wcopy;
|
|
|
6287 for (s = word; *s != NUL; )
|
|
|
6288 {
|
|
|
6289 #ifdef FEAT_MBYTE
|
|
|
6290 if (has_mbyte)
|
|
|
6291 c = mb_ptr2char_adv(&s);
|
|
|
6292 else
|
|
|
6293 #endif
|
|
|
6294 c = *s++;
|
|
324
|
6295 c = SPELL_TOUPPER(c);
|
|
323
|
6296
|
|
|
6297 #ifdef FEAT_MBYTE
|
|
|
6298 if (has_mbyte)
|
|
|
6299 {
|
|
|
6300 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
6301 break;
|
|
|
6302 d += mb_char2bytes(c, d);
|
|
|
6303 }
|
|
|
6304 else
|
|
|
6305 #endif
|
|
|
6306 {
|
|
|
6307 if (d - wcopy >= MAXWLEN - 1)
|
|
|
6308 break;
|
|
|
6309 *d++ = c;
|
|
|
6310 }
|
|
|
6311 }
|
|
|
6312 *d = NUL;
|
|
|
6313 }
|
|
|
6314
|
|
|
6315 /*
|
|
346
|
6316 * Try finding suggestions by recognizing specific situations.
|
|
|
6317 */
|
|
|
6318 static void
|
|
|
6319 suggest_try_special(su)
|
|
|
6320 suginfo_T *su;
|
|
|
6321 {
|
|
|
6322 char_u *p;
|
|
|
6323 int len;
|
|
|
6324 int c;
|
|
|
6325 char_u word[MAXWLEN];
|
|
|
6326
|
|
|
6327 /*
|
|
|
6328 * Recognize a word that is repeated: "the the".
|
|
|
6329 */
|
|
|
6330 p = skiptowhite(su->su_fbadword);
|
|
|
6331 len = p - su->su_fbadword;
|
|
|
6332 p = skipwhite(p);
|
|
|
6333 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
6334 {
|
|
|
6335 /* Include badflags: if the badword is onecap or allcap
|
|
|
6336 * use that for the goodword too: "The the" -> "The". */
|
|
|
6337 c = su->su_fbadword[len];
|
|
|
6338 su->su_fbadword[len] = NUL;
|
|
|
6339 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
6340 su->su_fbadword[len] = c;
|
|
351
|
6341 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
6342 }
|
|
|
6343 }
|
|
|
6344
|
|
|
6345 /*
|
|
323
|
6346 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
6347 *
|
|
|
6348 * This uses a state machine. At each node in the tree we try various
|
|
|
6349 * operations. When trying if an operation work "depth" is increased and the
|
|
|
6350 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
6351 * character, replace one and delete another. The number of changes is
|
|
|
6352 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
6353 */
|
|
|
6354 static void
|
|
346
|
6355 suggest_try_change(su)
|
|
323
|
6356 suginfo_T *su;
|
|
|
6357 {
|
|
|
6358 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
6359 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
6360 trystate_T stack[MAXWLEN];
|
|
|
6361 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
|
6362 * to for word split) */
|
|
|
6363 char_u prewordlen = 0; /* length of word in "preword" */
|
|
|
6364 int splitoff = 0; /* index in tword after last split */
|
|
|
6365 trystate_T *sp;
|
|
|
6366 int newscore;
|
|
|
6367 langp_T *lp;
|
|
375
|
6368 char_u *byts, *fbyts, *pbyts;
|
|
|
6369 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
6370 int depth;
|
|
330
|
6371 int c, c2, c3;
|
|
|
6372 int n = 0;
|
|
323
|
6373 int flags;
|
|
|
6374 garray_T *gap;
|
|
324
|
6375 idx_T arridx;
|
|
323
|
6376 int len;
|
|
|
6377 char_u *p;
|
|
|
6378 fromto_T *ftp;
|
|
330
|
6379 int fl = 0, tl;
|
|
346
|
6380 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
323
|
6381
|
|
|
6382 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
6383 * to find matches (esp. REP items). Append some more text, changing
|
|
|
6384 * chars after the bad word may help. */
|
|
323
|
6385 STRCPY(fword, su->su_fbadword);
|
|
346
|
6386 n = STRLEN(fword);
|
|
|
6387 p = su->su_badptr + su->su_badlen;
|
|
|
6388 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
6389
|
|
|
6390 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
6391 lp->lp_slang != NULL; ++lp)
|
|
|
6392 {
|
|
|
6393 /*
|
|
|
6394 * Go through the whole case-fold tree, try changes at each node.
|
|
|
6395 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
6396 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
6397 * word).
|
|
|
6398 */
|
|
|
6399 depth = 0;
|
|
375
|
6400 sp = &stack[0];
|
|
|
6401 sp->ts_state = STATE_START;
|
|
|
6402 sp->ts_score = 0;
|
|
|
6403 sp->ts_curi = 1;
|
|
|
6404 sp->ts_fidx = 0;
|
|
|
6405 sp->ts_fidxtry = 0;
|
|
|
6406 sp->ts_twordlen = 0;
|
|
|
6407 sp->ts_arridx = 0;
|
|
330
|
6408 #ifdef FEAT_MBYTE
|
|
375
|
6409 sp->ts_tcharlen = 0;
|
|
330
|
6410 #endif
|
|
|
6411
|
|
|
6412 /*
|
|
375
|
6413 * When there are postponed prefixes we need to use these first. At
|
|
|
6414 * the end of the prefix we continue in the case-fold tree.
|
|
|
6415 */
|
|
|
6416 fbyts = lp->lp_slang->sl_fbyts;
|
|
|
6417 fidxs = lp->lp_slang->sl_fidxs;
|
|
|
6418 pbyts = lp->lp_slang->sl_pbyts;
|
|
|
6419 pidxs = lp->lp_slang->sl_pidxs;
|
|
|
6420 if (pbyts != NULL)
|
|
|
6421 {
|
|
|
6422 byts = pbyts;
|
|
|
6423 idxs = pidxs;
|
|
|
6424 sp->ts_prefixdepth = PREFIXTREE;
|
|
|
6425 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
6426 }
|
|
|
6427 else
|
|
|
6428 {
|
|
|
6429 byts = fbyts;
|
|
|
6430 idxs = fidxs;
|
|
|
6431 sp->ts_prefixdepth = NOPREFIX;
|
|
|
6432 }
|
|
|
6433
|
|
|
6434 /*
|
|
330
|
6435 * Loop to find all suggestions. At each round we either:
|
|
|
6436 * - For the current state try one operation, advance "ts_curi",
|
|
|
6437 * increase "depth".
|
|
|
6438 * - When a state is done go to the next, set "ts_state".
|
|
|
6439 * - When all states are tried decrease "depth".
|
|
|
6440 */
|
|
323
|
6441 while (depth >= 0 && !got_int)
|
|
|
6442 {
|
|
|
6443 sp = &stack[depth];
|
|
|
6444 switch (sp->ts_state)
|
|
|
6445 {
|
|
|
6446 case STATE_START:
|
|
375
|
6447 case STATE_NOPREFIX:
|
|
323
|
6448 /*
|
|
|
6449 * Start of node: Deal with NUL bytes, which means
|
|
|
6450 * tword[] may end here.
|
|
|
6451 */
|
|
|
6452 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
6453 len = byts[arridx]; /* bytes in this node */
|
|
|
6454 arridx += sp->ts_curi; /* index of current byte */
|
|
|
6455
|
|
375
|
6456 if (sp->ts_prefixdepth == PREFIXTREE)
|
|
|
6457 {
|
|
|
6458 /* Skip over the NUL bytes, we use them later. */
|
|
|
6459 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
6460 ;
|
|
|
6461 sp->ts_curi += n;
|
|
|
6462
|
|
|
6463 /* At end of a prefix or at start of prefixtree: check for
|
|
|
6464 * following word. */
|
|
|
6465 if (byts[arridx] == 0 || sp->ts_state == STATE_NOPREFIX)
|
|
|
6466 {
|
|
|
6467 sp->ts_state = STATE_START;
|
|
|
6468 ++depth;
|
|
|
6469 stack[depth] = stack[depth - 1];
|
|
|
6470 sp = &stack[depth];
|
|
|
6471 sp->ts_prefixdepth = depth - 1;
|
|
|
6472 byts = fbyts;
|
|
|
6473 idxs = fidxs;
|
|
|
6474 sp->ts_state = STATE_START;
|
|
|
6475 sp->ts_curi = 1; /* start just after length byte */
|
|
|
6476 sp->ts_arridx = 0;
|
|
|
6477
|
|
|
6478 /* Move the prefix to preword[] so that
|
|
|
6479 * find_keepcap_word() works. */
|
|
|
6480 prewordlen = splitoff = sp->ts_twordlen;
|
|
|
6481 mch_memmove(preword, tword, splitoff);
|
|
|
6482 break;
|
|
|
6483 }
|
|
|
6484
|
|
|
6485 /* Always past NUL bytes now. */
|
|
|
6486 sp->ts_state = STATE_ENDNUL;
|
|
|
6487 break;
|
|
|
6488 }
|
|
|
6489
|
|
346
|
6490 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
6491 {
|
|
|
6492 /* Past bytes in node and/or past NUL bytes. */
|
|
|
6493 sp->ts_state = STATE_ENDNUL;
|
|
|
6494 break;
|
|
|
6495 }
|
|
|
6496
|
|
|
6497 /*
|
|
|
6498 * End of word in tree.
|
|
|
6499 */
|
|
|
6500 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
6501
|
|
324
|
6502 flags = (int)idxs[arridx];
|
|
323
|
6503
|
|
375
|
6504 if (sp->ts_prefixdepth < MAXWLEN)
|
|
|
6505 {
|
|
|
6506 /* There was a prefix before the word. Check that the
|
|
|
6507 * prefix can be used with this word. */
|
|
|
6508 /* Count the length of the NULs in the prefix. If there
|
|
|
6509 * are none this must be the first try without a prefix.
|
|
|
6510 */
|
|
|
6511 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
6512 len = pbyts[n++];
|
|
|
6513 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
6514 ;
|
|
|
6515 if (c > 0)
|
|
|
6516 {
|
|
|
6517 /* The prefix ID is stored two bytes above the flags. */
|
|
|
6518 c = valid_word_prefix(c, n, (unsigned)flags >> 16,
|
|
|
6519 tword + splitoff, lp->lp_slang);
|
|
|
6520 if (c == 0)
|
|
|
6521 break;
|
|
|
6522
|
|
|
6523 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
6524 if (c & WF_RAREPFX)
|
|
|
6525 flags |= WF_RARE;
|
|
|
6526 }
|
|
|
6527 }
|
|
|
6528
|
|
323
|
6529 /*
|
|
|
6530 * Form the word with proper case in preword.
|
|
|
6531 * If there is a word from a previous split, append.
|
|
|
6532 */
|
|
|
6533 tword[sp->ts_twordlen] = NUL;
|
|
|
6534 if (flags & WF_KEEPCAP)
|
|
|
6535 /* Must find the word in the keep-case tree. */
|
|
|
6536 find_keepcap_word(lp->lp_slang, tword + splitoff,
|
|
|
6537 preword + prewordlen);
|
|
|
6538 else
|
|
346
|
6539 {
|
|
323
|
6540 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
6541 * use that for the goodword too. But if the badword is
|
|
|
6542 * allcap and it's only one char long use onecap. */
|
|
|
6543 c = su->su_badflags;
|
|
|
6544 if ((c & WF_ALLCAP)
|
|
|
6545 #ifdef FEAT_MBYTE
|
|
|
6546 && su->su_badlen == mb_ptr2len_check(su->su_badptr)
|
|
|
6547 #else
|
|
|
6548 && su->su_badlen == 1
|
|
|
6549 #endif
|
|
|
6550 )
|
|
|
6551 c = WF_ONECAP;
|
|
323
|
6552 make_case_word(tword + splitoff,
|
|
346
|
6553 preword + prewordlen, flags | c);
|
|
|
6554 }
|
|
323
|
6555
|
|
|
6556 /* Don't use a banned word. It may appear again as a good
|
|
|
6557 * word, thus remember it. */
|
|
|
6558 if (flags & WF_BANNED)
|
|
|
6559 {
|
|
|
6560 add_banned(su, preword + prewordlen);
|
|
|
6561 break;
|
|
|
6562 }
|
|
372
|
6563 if (was_banned(su, preword + prewordlen)
|
|
|
6564 || was_banned(su, preword))
|
|
323
|
6565 break;
|
|
|
6566
|
|
|
6567 newscore = 0;
|
|
|
6568 if ((flags & WF_REGION)
|
|
|
6569 && (((unsigned)flags >> 8) & lp->lp_region) == 0)
|
|
|
6570 newscore += SCORE_REGION;
|
|
|
6571 if (flags & WF_RARE)
|
|
|
6572 newscore += SCORE_RARE;
|
|
|
6573
|
|
346
|
6574 if (!spell_valid_case(su->su_badflags,
|
|
323
|
6575 captype(preword + prewordlen, NULL)))
|
|
|
6576 newscore += SCORE_ICASE;
|
|
|
6577
|
|
346
|
6578 if ((fword[sp->ts_fidx] == NUL
|
|
358
|
6579 || !spell_iswordp(fword + sp->ts_fidx))
|
|
346
|
6580 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
6581 {
|
|
366
|
6582 /* The badword also ends: add suggestions. Give a penalty
|
|
|
6583 * when changing non-word char to word char, e.g., "thes,"
|
|
|
6584 * -> "these". */
|
|
|
6585 p = fword + sp->ts_fidx;
|
|
|
6586 #ifdef FEAT_MBYTE
|
|
|
6587 if (has_mbyte)
|
|
|
6588 mb_ptr_back(fword, p);
|
|
|
6589 else
|
|
|
6590 #endif
|
|
|
6591 --p;
|
|
|
6592 if (!spell_iswordp(p))
|
|
|
6593 {
|
|
|
6594 p = preword + STRLEN(preword);
|
|
|
6595 #ifdef FEAT_MBYTE
|
|
|
6596 if (has_mbyte)
|
|
|
6597 mb_ptr_back(preword, p);
|
|
|
6598 else
|
|
|
6599 #endif
|
|
|
6600 --p;
|
|
|
6601 if (spell_iswordp(p))
|
|
|
6602 newscore += SCORE_NONWORD;
|
|
|
6603 }
|
|
|
6604
|
|
344
|
6605 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
6606 sp->ts_fidx - repextra,
|
|
351
|
6607 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
6608 }
|
|
330
|
6609 else if (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
6610 #ifdef FEAT_MBYTE
|
|
|
6611 /* Don't split halfway a character. */
|
|
|
6612 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
6613 #endif
|
|
|
6614 )
|
|
323
|
6615 {
|
|
|
6616 /* The word in the tree ends but the badword
|
|
|
6617 * continues: try inserting a space and check that a valid
|
|
|
6618 * words starts at fword[sp->ts_fidx]. */
|
|
|
6619 if (try_deeper(su, stack, depth, newscore + SCORE_SPLIT))
|
|
|
6620 {
|
|
|
6621 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
|
6622 sp->ts_save_prewordlen = prewordlen;
|
|
346
|
6623 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
6624 sp->ts_save_splitoff = splitoff;
|
|
|
6625
|
|
|
6626 /* Append a space to preword. */
|
|
|
6627 STRCAT(preword, " ");
|
|
|
6628 prewordlen = STRLEN(preword);
|
|
|
6629 splitoff = sp->ts_twordlen;
|
|
324
|
6630 #ifdef FEAT_MBYTE
|
|
|
6631 if (has_mbyte)
|
|
|
6632 {
|
|
|
6633 int i = 0;
|
|
|
6634
|
|
|
6635 /* Case-folding may change the number of bytes:
|
|
|
6636 * Count nr of chars in fword[sp->ts_fidx] and
|
|
|
6637 * advance that many chars in su->su_badptr. */
|
|
|
6638 for (p = fword; p < fword + sp->ts_fidx;
|
|
|
6639 mb_ptr_adv(p))
|
|
|
6640 ++i;
|
|
|
6641 for (p = su->su_badptr; i > 0; mb_ptr_adv(p))
|
|
|
6642 --i;
|
|
|
6643 }
|
|
|
6644 else
|
|
|
6645 #endif
|
|
|
6646 p = su->su_badptr + sp->ts_fidx;
|
|
346
|
6647 su->su_badflags = captype(p, su->su_badptr
|
|
|
6648 + su->su_badlen);
|
|
323
|
6649
|
|
|
6650 sp->ts_state = STATE_SPLITUNDO;
|
|
|
6651 ++depth;
|
|
|
6652 /* Restart at top of the tree. */
|
|
|
6653 stack[depth].ts_arridx = 0;
|
|
|
6654 }
|
|
|
6655 }
|
|
|
6656 break;
|
|
|
6657
|
|
|
6658 case STATE_SPLITUNDO:
|
|
346
|
6659 /* Undo the changes done for word split. */
|
|
|
6660 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
6661 splitoff = sp->ts_save_splitoff;
|
|
|
6662 prewordlen = sp->ts_save_prewordlen;
|
|
|
6663
|
|
|
6664 /* Continue looking for NUL bytes. */
|
|
|
6665 sp->ts_state = STATE_START;
|
|
|
6666 break;
|
|
|
6667
|
|
|
6668 case STATE_ENDNUL:
|
|
|
6669 /* Past the NUL bytes in the node. */
|
|
346
|
6670 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
6671 {
|
|
|
6672 /* The badword ends, can't use the bytes in this node. */
|
|
|
6673 sp->ts_state = STATE_DEL;
|
|
|
6674 break;
|
|
|
6675 }
|
|
|
6676 sp->ts_state = STATE_PLAIN;
|
|
|
6677 /*FALLTHROUGH*/
|
|
|
6678
|
|
|
6679 case STATE_PLAIN:
|
|
|
6680 /*
|
|
|
6681 * Go over all possible bytes at this node, add each to
|
|
|
6682 * tword[] and use child node. "ts_curi" is the index.
|
|
|
6683 */
|
|
|
6684 arridx = sp->ts_arridx;
|
|
|
6685 if (sp->ts_curi > byts[arridx])
|
|
|
6686 {
|
|
|
6687 /* Done all bytes at this node, do next state. When still
|
|
|
6688 * at already changed bytes skip the other tricks. */
|
|
|
6689 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
6690 sp->ts_state = STATE_DEL;
|
|
|
6691 else
|
|
|
6692 sp->ts_state = STATE_FINAL;
|
|
|
6693 }
|
|
|
6694 else
|
|
|
6695 {
|
|
|
6696 arridx += sp->ts_curi++;
|
|
|
6697 c = byts[arridx];
|
|
|
6698
|
|
|
6699 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
6700 * the byte in the bad word adjust the score. But don't
|
|
|
6701 * even try when the byte was already changed. */
|
|
330
|
6702 if (c == fword[sp->ts_fidx]
|
|
|
6703 #ifdef FEAT_MBYTE
|
|
|
6704 || (sp->ts_tcharlen > 0
|
|
|
6705 && sp->ts_isdiff != DIFF_NONE)
|
|
|
6706 #endif
|
|
|
6707 )
|
|
323
|
6708 newscore = 0;
|
|
|
6709 else
|
|
|
6710 newscore = SCORE_SUBST;
|
|
|
6711 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
6712 && try_deeper(su, stack, depth, newscore))
|
|
|
6713 {
|
|
|
6714 ++depth;
|
|
330
|
6715 sp = &stack[depth];
|
|
|
6716 ++sp->ts_fidx;
|
|
|
6717 tword[sp->ts_twordlen++] = c;
|
|
|
6718 sp->ts_arridx = idxs[arridx];
|
|
|
6719 #ifdef FEAT_MBYTE
|
|
|
6720 if (newscore == SCORE_SUBST)
|
|
|
6721 sp->ts_isdiff = DIFF_YES;
|
|
|
6722 if (has_mbyte)
|
|
|
6723 {
|
|
|
6724 /* Multi-byte characters are a bit complicated to
|
|
|
6725 * handle: They differ when any of the bytes
|
|
|
6726 * differ and then their length may also differ. */
|
|
|
6727 if (sp->ts_tcharlen == 0)
|
|
|
6728 {
|
|
|
6729 /* First byte. */
|
|
|
6730 sp->ts_tcharidx = 0;
|
|
|
6731 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
6732 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
6733 sp->ts_isdiff = (newscore != 0)
|
|
|
6734 ? DIFF_YES : DIFF_NONE;
|
|
|
6735 }
|
|
|
6736 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
6737 /* When inserting trail bytes don't advance in
|
|
|
6738 * the bad word. */
|
|
|
6739 --sp->ts_fidx;
|
|
|
6740 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
6741 {
|
|
|
6742 /* Last byte of character. */
|
|
|
6743 if (sp->ts_isdiff == DIFF_YES)
|
|
|
6744 {
|
|
|
6745 /* Correct ts_fidx for the byte length of
|
|
|
6746 * the character (we didn't check that
|
|
|
6747 * before). */
|
|
|
6748 sp->ts_fidx = sp->ts_fcharstart
|
|
|
6749 + MB_BYTE2LEN(
|
|
|
6750 fword[sp->ts_fcharstart]);
|
|
|
6751
|
|
|
6752 /* For a similar character adjust score
|
|
|
6753 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
|
6754 if (lp->lp_slang->sl_has_map
|
|
|
6755 && similar_chars(lp->lp_slang,
|
|
|
6756 mb_ptr2char(tword
|
|
|
6757 + sp->ts_twordlen
|
|
|
6758 - sp->ts_tcharlen),
|
|
|
6759 mb_ptr2char(fword
|
|
|
6760 + sp->ts_fcharstart)))
|
|
|
6761 sp->ts_score -=
|
|
|
6762 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
6763 }
|
|
358
|
6764 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
6765 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
6766 {
|
|
|
6767 /* If the previous character was the same,
|
|
|
6768 * thus doubling a character, give a bonus
|
|
|
6769 * to the score. */
|
|
|
6770 p = tword + sp->ts_twordlen
|
|
|
6771 - sp->ts_tcharlen;
|
|
|
6772 c = mb_ptr2char(p);
|
|
|
6773 mb_ptr_back(tword, p);
|
|
|
6774 if (c == mb_ptr2char(p))
|
|
|
6775 sp->ts_score -= SCORE_INS
|
|
|
6776 - SCORE_INSDUP;
|
|
|
6777 }
|
|
330
|
6778
|
|
|
6779 /* Starting a new char, reset the length. */
|
|
|
6780 sp->ts_tcharlen = 0;
|
|
|
6781 }
|
|
|
6782 }
|
|
|
6783 else
|
|
|
6784 #endif
|
|
|
6785 {
|
|
|
6786 /* If we found a similar char adjust the score.
|
|
|
6787 * We do this after calling try_deeper() because
|
|
|
6788 * it's slow. */
|
|
|
6789 if (newscore != 0
|
|
|
6790 && lp->lp_slang->sl_has_map
|
|
|
6791 && similar_chars(lp->lp_slang,
|
|
|
6792 c, fword[sp->ts_fidx - 1]))
|
|
|
6793 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
6794 }
|
|
323
|
6795 }
|
|
|
6796 }
|
|
|
6797 break;
|
|
|
6798
|
|
|
6799 case STATE_DEL:
|
|
330
|
6800 #ifdef FEAT_MBYTE
|
|
|
6801 /* When past the first byte of a multi-byte char don't try
|
|
|
6802 * delete/insert/swap a character. */
|
|
|
6803 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
6804 {
|
|
|
6805 sp->ts_state = STATE_FINAL;
|
|
|
6806 break;
|
|
|
6807 }
|
|
|
6808 #endif
|
|
|
6809 /*
|
|
|
6810 * Try skipping one character in the bad word (delete it).
|
|
|
6811 */
|
|
323
|
6812 sp->ts_state = STATE_INS;
|
|
|
6813 sp->ts_curi = 1;
|
|
|
6814 if (fword[sp->ts_fidx] != NUL
|
|
|
6815 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
6816 {
|
|
|
6817 ++depth;
|
|
358
|
6818
|
|
|
6819 /* Advance over the character in fword[]. Give a bonus to
|
|
|
6820 * the score if the same character is following "nn" ->
|
|
|
6821 * "n". */
|
|
330
|
6822 #ifdef FEAT_MBYTE
|
|
|
6823 if (has_mbyte)
|
|
358
|
6824 {
|
|
|
6825 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
6826 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
358
|
6827 if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
|
6828 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
6829 }
|
|
330
|
6830 else
|
|
|
6831 #endif
|
|
358
|
6832 {
|
|
330
|
6833 ++stack[depth].ts_fidx;
|
|
358
|
6834 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
6835 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
6836 }
|
|
323
|
6837 break;
|
|
|
6838 }
|
|
|
6839 /*FALLTHROUGH*/
|
|
|
6840
|
|
|
6841 case STATE_INS:
|
|
330
|
6842 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
6843 * node. */
|
|
|
6844 n = sp->ts_arridx;
|
|
|
6845 if (sp->ts_curi > byts[n])
|
|
|
6846 {
|
|
|
6847 /* Done all bytes at this node, do next state. */
|
|
|
6848 sp->ts_state = STATE_SWAP;
|
|
|
6849 }
|
|
|
6850 else
|
|
|
6851 {
|
|
330
|
6852 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
6853 n += sp->ts_curi++;
|
|
|
6854 c = byts[n];
|
|
|
6855 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
6856 {
|
|
|
6857 ++depth;
|
|
330
|
6858 sp = &stack[depth];
|
|
|
6859 tword[sp->ts_twordlen++] = c;
|
|
|
6860 sp->ts_arridx = idxs[n];
|
|
|
6861 #ifdef FEAT_MBYTE
|
|
|
6862 if (has_mbyte)
|
|
|
6863 {
|
|
|
6864 fl = MB_BYTE2LEN(c);
|
|
|
6865 if (fl > 1)
|
|
|
6866 {
|
|
|
6867 /* There are following bytes for the same
|
|
|
6868 * character. We must find all bytes before
|
|
|
6869 * trying delete/insert/swap/etc. */
|
|
|
6870 sp->ts_tcharlen = fl;
|
|
|
6871 sp->ts_tcharidx = 1;
|
|
|
6872 sp->ts_isdiff = DIFF_INSERT;
|
|
|
6873 }
|
|
|
6874 }
|
|
358
|
6875 else
|
|
|
6876 fl = 1;
|
|
|
6877 if (fl == 1)
|
|
330
|
6878 #endif
|
|
358
|
6879 {
|
|
|
6880 /* If the previous character was the same, thus
|
|
|
6881 * doubling a character, give a bonus to the
|
|
|
6882 * score. */
|
|
|
6883 if (sp->ts_twordlen >= 2
|
|
|
6884 && tword[sp->ts_twordlen - 2] == c)
|
|
|
6885 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
6886 }
|
|
323
|
6887 }
|
|
|
6888 }
|
|
|
6889 break;
|
|
|
6890
|
|
|
6891 case STATE_SWAP:
|
|
330
|
6892 /*
|
|
|
6893 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
6894 * We change "fword" here, it's changed back afterwards.
|
|
|
6895 */
|
|
|
6896 p = fword + sp->ts_fidx;
|
|
|
6897 c = *p;
|
|
|
6898 if (c == NUL)
|
|
|
6899 {
|
|
|
6900 /* End of word, can't swap or replace. */
|
|
|
6901 sp->ts_state = STATE_FINAL;
|
|
|
6902 break;
|
|
|
6903 }
|
|
|
6904 #ifdef FEAT_MBYTE
|
|
|
6905 if (has_mbyte)
|
|
|
6906 {
|
|
|
6907 n = mb_ptr2len_check(p);
|
|
|
6908 c = mb_ptr2char(p);
|
|
|
6909 c2 = mb_ptr2char(p + n);
|
|
|
6910 }
|
|
|
6911 else
|
|
|
6912 #endif
|
|
|
6913 c2 = p[1];
|
|
|
6914 if (c == c2)
|
|
323
|
6915 {
|
|
330
|
6916 /* Characters are identical, swap won't do anything. */
|
|
|
6917 sp->ts_state = STATE_SWAP3;
|
|
|
6918 break;
|
|
|
6919 }
|
|
|
6920 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
6921 {
|
|
|
6922 sp->ts_state = STATE_UNSWAP;
|
|
323
|
6923 ++depth;
|
|
330
|
6924 #ifdef FEAT_MBYTE
|
|
|
6925 if (has_mbyte)
|
|
|
6926 {
|
|
|
6927 fl = mb_char2len(c2);
|
|
|
6928 mch_memmove(p, p + n, fl);
|
|
|
6929 mb_char2bytes(c, p + fl);
|
|
|
6930 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
6931 }
|
|
|
6932 else
|
|
|
6933 #endif
|
|
|
6934 {
|
|
|
6935 p[0] = c2;
|
|
|
6936 p[1] = c;
|
|
|
6937 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
6938 }
|
|
323
|
6939 }
|
|
|
6940 else
|
|
|
6941 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
6942 sp->ts_state = STATE_REP_INI;
|
|
|
6943 break;
|
|
|
6944
|
|
330
|
6945 case STATE_UNSWAP:
|
|
|
6946 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
6947 p = fword + sp->ts_fidx;
|
|
|
6948 #ifdef FEAT_MBYTE
|
|
|
6949 if (has_mbyte)
|
|
|
6950 {
|
|
|
6951 n = MB_BYTE2LEN(*p);
|
|
|
6952 c = mb_ptr2char(p + n);
|
|
|
6953 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
6954 mb_char2bytes(c, p);
|
|
|
6955 }
|
|
|
6956 else
|
|
|
6957 #endif
|
|
|
6958 {
|
|
|
6959 c = *p;
|
|
|
6960 *p = p[1];
|
|
|
6961 p[1] = c;
|
|
|
6962 }
|
|
|
6963 /*FALLTHROUGH*/
|
|
|
6964
|
|
|
6965 case STATE_SWAP3:
|
|
323
|
6966 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
6967 * "fword" here, it's changed back afterwards. */
|
|
|
6968 p = fword + sp->ts_fidx;
|
|
|
6969 #ifdef FEAT_MBYTE
|
|
|
6970 if (has_mbyte)
|
|
|
6971 {
|
|
|
6972 n = mb_ptr2len_check(p);
|
|
|
6973 c = mb_ptr2char(p);
|
|
|
6974 fl = mb_ptr2len_check(p + n);
|
|
|
6975 c2 = mb_ptr2char(p + n);
|
|
|
6976 c3 = mb_ptr2char(p + n + fl);
|
|
|
6977 }
|
|
|
6978 else
|
|
|
6979 #endif
|
|
323
|
6980 {
|
|
330
|
6981 c = *p;
|
|
|
6982 c2 = p[1];
|
|
|
6983 c3 = p[2];
|
|
|
6984 }
|
|
|
6985
|
|
|
6986 /* When characters are identical: "121" then SWAP3 result is
|
|
|
6987 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
6988 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
6989 * swapping. Also skip when c3 is NUL. */
|
|
|
6990 if (c == c3 || c3 == NUL)
|
|
|
6991 {
|
|
|
6992 sp->ts_state = STATE_REP_INI;
|
|
|
6993 break;
|
|
|
6994 }
|
|
|
6995 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
6996 {
|
|
|
6997 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
6998 ++depth;
|
|
330
|
6999 #ifdef FEAT_MBYTE
|
|
|
7000 if (has_mbyte)
|
|
|
7001 {
|
|
|
7002 tl = mb_char2len(c3);
|
|
|
7003 mch_memmove(p, p + n + fl, tl);
|
|
|
7004 mb_char2bytes(c2, p + tl);
|
|
|
7005 mb_char2bytes(c, p + fl + tl);
|
|
|
7006 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
7007 }
|
|
|
7008 else
|
|
|
7009 #endif
|
|
|
7010 {
|
|
|
7011 p[0] = p[2];
|
|
|
7012 p[2] = c;
|
|
|
7013 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7014 }
|
|
323
|
7015 }
|
|
|
7016 else
|
|
|
7017 sp->ts_state = STATE_REP_INI;
|
|
|
7018 break;
|
|
|
7019
|
|
330
|
7020 case STATE_UNSWAP3:
|
|
|
7021 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
7022 p = fword + sp->ts_fidx;
|
|
|
7023 #ifdef FEAT_MBYTE
|
|
|
7024 if (has_mbyte)
|
|
|
7025 {
|
|
|
7026 n = MB_BYTE2LEN(*p);
|
|
|
7027 c2 = mb_ptr2char(p + n);
|
|
|
7028 fl = MB_BYTE2LEN(p[n]);
|
|
|
7029 c = mb_ptr2char(p + n + fl);
|
|
|
7030 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
7031 mch_memmove(p + fl + tl, p, n);
|
|
|
7032 mb_char2bytes(c, p);
|
|
|
7033 mb_char2bytes(c2, p + tl);
|
|
|
7034 }
|
|
|
7035 else
|
|
|
7036 #endif
|
|
|
7037 {
|
|
|
7038 c = *p;
|
|
|
7039 *p = p[2];
|
|
|
7040 p[2] = c;
|
|
|
7041 }
|
|
346
|
7042
|
|
330
|
7043 /* Rotate three characters left: "123" -> "231". We change
|
|
|
7044 * "fword" here, it's changed back afterwards. */
|
|
323
|
7045 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7046 {
|
|
330
|
7047 sp->ts_state = STATE_UNROT3L;
|
|
323
|
7048 ++depth;
|
|
330
|
7049 p = fword + sp->ts_fidx;
|
|
|
7050 #ifdef FEAT_MBYTE
|
|
|
7051 if (has_mbyte)
|
|
|
7052 {
|
|
|
7053 n = mb_ptr2len_check(p);
|
|
|
7054 c = mb_ptr2char(p);
|
|
|
7055 fl = mb_ptr2len_check(p + n);
|
|
|
7056 fl += mb_ptr2len_check(p + n + fl);
|
|
|
7057 mch_memmove(p, p + n, fl);
|
|
|
7058 mb_char2bytes(c, p + fl);
|
|
|
7059 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7060 }
|
|
|
7061 else
|
|
|
7062 #endif
|
|
|
7063 {
|
|
|
7064 c = *p;
|
|
|
7065 *p = p[1];
|
|
|
7066 p[1] = p[2];
|
|
|
7067 p[2] = c;
|
|
|
7068 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7069 }
|
|
323
|
7070 }
|
|
|
7071 else
|
|
|
7072 sp->ts_state = STATE_REP_INI;
|
|
|
7073 break;
|
|
|
7074
|
|
330
|
7075 case STATE_UNROT3L:
|
|
346
|
7076 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
7077 p = fword + sp->ts_fidx;
|
|
|
7078 #ifdef FEAT_MBYTE
|
|
|
7079 if (has_mbyte)
|
|
|
7080 {
|
|
|
7081 n = MB_BYTE2LEN(*p);
|
|
|
7082 n += MB_BYTE2LEN(p[n]);
|
|
|
7083 c = mb_ptr2char(p + n);
|
|
|
7084 tl = MB_BYTE2LEN(p[n]);
|
|
|
7085 mch_memmove(p + tl, p, n);
|
|
|
7086 mb_char2bytes(c, p);
|
|
|
7087 }
|
|
|
7088 else
|
|
|
7089 #endif
|
|
|
7090 {
|
|
|
7091 c = p[2];
|
|
|
7092 p[2] = p[1];
|
|
|
7093 p[1] = *p;
|
|
|
7094 *p = c;
|
|
|
7095 }
|
|
346
|
7096
|
|
323
|
7097 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
7098 * "fword" here, it's changed back afterwards. */
|
|
323
|
7099 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7100 {
|
|
330
|
7101 sp->ts_state = STATE_UNROT3R;
|
|
323
|
7102 ++depth;
|
|
330
|
7103 p = fword + sp->ts_fidx;
|
|
|
7104 #ifdef FEAT_MBYTE
|
|
|
7105 if (has_mbyte)
|
|
|
7106 {
|
|
|
7107 n = mb_ptr2len_check(p);
|
|
|
7108 n += mb_ptr2len_check(p + n);
|
|
|
7109 c = mb_ptr2char(p + n);
|
|
|
7110 tl = mb_ptr2len_check(p + n);
|
|
|
7111 mch_memmove(p + tl, p, n);
|
|
|
7112 mb_char2bytes(c, p);
|
|
|
7113 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
7114 }
|
|
|
7115 else
|
|
|
7116 #endif
|
|
|
7117 {
|
|
|
7118 c = p[2];
|
|
|
7119 p[2] = p[1];
|
|
|
7120 p[1] = *p;
|
|
|
7121 *p = c;
|
|
|
7122 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7123 }
|
|
323
|
7124 }
|
|
|
7125 else
|
|
|
7126 sp->ts_state = STATE_REP_INI;
|
|
|
7127 break;
|
|
|
7128
|
|
330
|
7129 case STATE_UNROT3R:
|
|
346
|
7130 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
7131 p = fword + sp->ts_fidx;
|
|
|
7132 #ifdef FEAT_MBYTE
|
|
|
7133 if (has_mbyte)
|
|
|
7134 {
|
|
|
7135 c = mb_ptr2char(p);
|
|
|
7136 tl = MB_BYTE2LEN(*p);
|
|
|
7137 n = MB_BYTE2LEN(p[tl]);
|
|
|
7138 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
7139 mch_memmove(p, p + tl, n);
|
|
|
7140 mb_char2bytes(c, p + n);
|
|
|
7141 }
|
|
|
7142 else
|
|
|
7143 #endif
|
|
|
7144 {
|
|
|
7145 c = *p;
|
|
|
7146 *p = p[1];
|
|
|
7147 p[1] = p[2];
|
|
|
7148 p[2] = c;
|
|
|
7149 }
|
|
323
|
7150 /*FALLTHROUGH*/
|
|
|
7151
|
|
|
7152 case STATE_REP_INI:
|
|
|
7153 /* Check if matching with REP items from the .aff file would
|
|
|
7154 * work. Quickly skip if there are no REP items or the score
|
|
|
7155 * is going to be too high anyway. */
|
|
|
7156 gap = &lp->lp_slang->sl_rep;
|
|
|
7157 if (gap->ga_len == 0
|
|
|
7158 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
7159 {
|
|
|
7160 sp->ts_state = STATE_FINAL;
|
|
|
7161 break;
|
|
|
7162 }
|
|
|
7163
|
|
|
7164 /* Use the first byte to quickly find the first entry that
|
|
330
|
7165 * may match. If the index is -1 there is none. */
|
|
323
|
7166 sp->ts_curi = lp->lp_slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
7167 if (sp->ts_curi < 0)
|
|
|
7168 {
|
|
|
7169 sp->ts_state = STATE_FINAL;
|
|
|
7170 break;
|
|
|
7171 }
|
|
|
7172
|
|
|
7173 sp->ts_state = STATE_REP;
|
|
|
7174 /*FALLTHROUGH*/
|
|
|
7175
|
|
|
7176 case STATE_REP:
|
|
|
7177 /* Try matching with REP items from the .aff file. For each
|
|
330
|
7178 * match replace the characters and check if the resulting
|
|
|
7179 * word is valid. */
|
|
323
|
7180 p = fword + sp->ts_fidx;
|
|
|
7181
|
|
|
7182 gap = &lp->lp_slang->sl_rep;
|
|
|
7183 while (sp->ts_curi < gap->ga_len)
|
|
|
7184 {
|
|
|
7185 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
7186 if (*ftp->ft_from != *p)
|
|
|
7187 {
|
|
|
7188 /* past possible matching entries */
|
|
|
7189 sp->ts_curi = gap->ga_len;
|
|
|
7190 break;
|
|
|
7191 }
|
|
|
7192 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
7193 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
7194 {
|
|
|
7195 /* Need to undo this afterwards. */
|
|
|
7196 sp->ts_state = STATE_REP_UNDO;
|
|
|
7197
|
|
|
7198 /* Change the "from" to the "to" string. */
|
|
|
7199 ++depth;
|
|
|
7200 fl = STRLEN(ftp->ft_from);
|
|
|
7201 tl = STRLEN(ftp->ft_to);
|
|
|
7202 if (fl != tl)
|
|
346
|
7203 {
|
|
323
|
7204 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
7205 repextra += tl - fl;
|
|
|
7206 }
|
|
323
|
7207 mch_memmove(p, ftp->ft_to, tl);
|
|
|
7208 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
7209 #ifdef FEAT_MBYTE
|
|
|
7210 stack[depth].ts_tcharlen = 0;
|
|
|
7211 #endif
|
|
323
|
7212 break;
|
|
|
7213 }
|
|
|
7214 }
|
|
|
7215
|
|
|
7216 if (sp->ts_curi >= gap->ga_len)
|
|
|
7217 /* No (more) matches. */
|
|
|
7218 sp->ts_state = STATE_FINAL;
|
|
|
7219
|
|
|
7220 break;
|
|
|
7221
|
|
|
7222 case STATE_REP_UNDO:
|
|
|
7223 /* Undo a REP replacement and continue with the next one. */
|
|
|
7224 ftp = (fromto_T *)lp->lp_slang->sl_rep.ga_data
|
|
|
7225 + sp->ts_curi - 1;
|
|
|
7226 fl = STRLEN(ftp->ft_from);
|
|
|
7227 tl = STRLEN(ftp->ft_to);
|
|
|
7228 p = fword + sp->ts_fidx;
|
|
|
7229 if (fl != tl)
|
|
346
|
7230 {
|
|
323
|
7231 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
7232 repextra -= tl - fl;
|
|
|
7233 }
|
|
323
|
7234 mch_memmove(p, ftp->ft_from, fl);
|
|
|
7235 sp->ts_state = STATE_REP;
|
|
|
7236 break;
|
|
|
7237
|
|
|
7238 default:
|
|
|
7239 /* Did all possible states at this level, go up one level. */
|
|
|
7240 --depth;
|
|
344
|
7241
|
|
375
|
7242 if (depth >= 0 && stack[depth].ts_prefixdepth == PREFIXTREE)
|
|
|
7243 {
|
|
|
7244 /* Continue in or go back to the prefix tree. */
|
|
|
7245 byts = pbyts;
|
|
|
7246 idxs = pidxs;
|
|
|
7247 splitoff = 0;
|
|
|
7248 }
|
|
|
7249
|
|
344
|
7250 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
7251 line_breakcheck();
|
|
323
|
7252 }
|
|
|
7253 }
|
|
|
7254 }
|
|
|
7255 }
|
|
|
7256
|
|
|
7257 /*
|
|
|
7258 * Try going one level deeper in the tree.
|
|
|
7259 */
|
|
|
7260 static int
|
|
|
7261 try_deeper(su, stack, depth, score_add)
|
|
|
7262 suginfo_T *su;
|
|
|
7263 trystate_T *stack;
|
|
|
7264 int depth;
|
|
|
7265 int score_add;
|
|
|
7266 {
|
|
|
7267 int newscore;
|
|
|
7268
|
|
|
7269 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
7270 newscore = stack[depth].ts_score + score_add;
|
|
|
7271 if (newscore >= su->su_maxscore)
|
|
|
7272 return FALSE;
|
|
|
7273
|
|
330
|
7274 stack[depth + 1] = stack[depth];
|
|
323
|
7275 stack[depth + 1].ts_state = STATE_START;
|
|
|
7276 stack[depth + 1].ts_score = newscore;
|
|
|
7277 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
7278 return TRUE;
|
|
|
7279 }
|
|
|
7280
|
|
|
7281 /*
|
|
|
7282 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
7283 * words and put it in "kword".
|
|
|
7284 * Theoretically there could be several keep-case words that result in the
|
|
|
7285 * same case-folded word, but we only find one...
|
|
|
7286 */
|
|
|
7287 static void
|
|
|
7288 find_keepcap_word(slang, fword, kword)
|
|
|
7289 slang_T *slang;
|
|
|
7290 char_u *fword;
|
|
|
7291 char_u *kword;
|
|
|
7292 {
|
|
|
7293 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
7294 int depth;
|
|
324
|
7295 idx_T tryidx;
|
|
323
|
7296
|
|
|
7297 /* The following arrays are used at each depth in the tree. */
|
|
324
|
7298 idx_T arridx[MAXWLEN];
|
|
323
|
7299 int round[MAXWLEN];
|
|
|
7300 int fwordidx[MAXWLEN];
|
|
|
7301 int uwordidx[MAXWLEN];
|
|
|
7302 int kwordlen[MAXWLEN];
|
|
|
7303
|
|
|
7304 int flen, ulen;
|
|
|
7305 int l;
|
|
|
7306 int len;
|
|
|
7307 int c;
|
|
324
|
7308 idx_T lo, hi, m;
|
|
323
|
7309 char_u *p;
|
|
|
7310 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
7311 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
7312
|
|
|
7313 if (byts == NULL)
|
|
|
7314 {
|
|
|
7315 /* array is empty: "cannot happen" */
|
|
|
7316 *kword = NUL;
|
|
|
7317 return;
|
|
|
7318 }
|
|
|
7319
|
|
|
7320 /* Make an all-cap version of "fword". */
|
|
|
7321 allcap_copy(fword, uword);
|
|
|
7322
|
|
|
7323 /*
|
|
|
7324 * Each character needs to be tried both case-folded and upper-case.
|
|
|
7325 * All this gets very complicated if we keep in mind that changing case
|
|
|
7326 * may change the byte length of a multi-byte character...
|
|
|
7327 */
|
|
|
7328 depth = 0;
|
|
|
7329 arridx[0] = 0;
|
|
|
7330 round[0] = 0;
|
|
|
7331 fwordidx[0] = 0;
|
|
|
7332 uwordidx[0] = 0;
|
|
|
7333 kwordlen[0] = 0;
|
|
|
7334 while (depth >= 0)
|
|
|
7335 {
|
|
|
7336 if (fword[fwordidx[depth]] == NUL)
|
|
|
7337 {
|
|
|
7338 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
7339 * here we have found a match. */
|
|
|
7340 if (byts[arridx[depth] + 1] == 0)
|
|
|
7341 {
|
|
|
7342 kword[kwordlen[depth]] = NUL;
|
|
|
7343 return;
|
|
|
7344 }
|
|
|
7345
|
|
|
7346 /* kword is getting too long, continue one level up */
|
|
|
7347 --depth;
|
|
|
7348 }
|
|
|
7349 else if (++round[depth] > 2)
|
|
|
7350 {
|
|
|
7351 /* tried both fold-case and upper-case character, continue one
|
|
|
7352 * level up */
|
|
|
7353 --depth;
|
|
|
7354 }
|
|
|
7355 else
|
|
|
7356 {
|
|
|
7357 /*
|
|
|
7358 * round[depth] == 1: Try using the folded-case character.
|
|
|
7359 * round[depth] == 2: Try using the upper-case character.
|
|
|
7360 */
|
|
|
7361 #ifdef FEAT_MBYTE
|
|
|
7362 if (has_mbyte)
|
|
|
7363 {
|
|
|
7364 flen = mb_ptr2len_check(fword + fwordidx[depth]);
|
|
|
7365 ulen = mb_ptr2len_check(uword + uwordidx[depth]);
|
|
|
7366 }
|
|
|
7367 else
|
|
|
7368 #endif
|
|
|
7369 ulen = flen = 1;
|
|
|
7370 if (round[depth] == 1)
|
|
|
7371 {
|
|
|
7372 p = fword + fwordidx[depth];
|
|
|
7373 l = flen;
|
|
|
7374 }
|
|
|
7375 else
|
|
|
7376 {
|
|
|
7377 p = uword + uwordidx[depth];
|
|
|
7378 l = ulen;
|
|
|
7379 }
|
|
|
7380
|
|
|
7381 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
7382 {
|
|
|
7383 /* Perform a binary search in the list of accepted bytes. */
|
|
|
7384 len = byts[tryidx++];
|
|
|
7385 c = *p++;
|
|
|
7386 lo = tryidx;
|
|
|
7387 hi = tryidx + len - 1;
|
|
|
7388 while (lo < hi)
|
|
|
7389 {
|
|
|
7390 m = (lo + hi) / 2;
|
|
|
7391 if (byts[m] > c)
|
|
|
7392 hi = m - 1;
|
|
|
7393 else if (byts[m] < c)
|
|
|
7394 lo = m + 1;
|
|
|
7395 else
|
|
|
7396 {
|
|
|
7397 lo = hi = m;
|
|
|
7398 break;
|
|
|
7399 }
|
|
|
7400 }
|
|
|
7401
|
|
|
7402 /* Stop if there is no matching byte. */
|
|
|
7403 if (hi < lo || byts[lo] != c)
|
|
|
7404 break;
|
|
|
7405
|
|
|
7406 /* Continue at the child (if there is one). */
|
|
|
7407 tryidx = idxs[lo];
|
|
|
7408 }
|
|
|
7409
|
|
|
7410 if (l == 0)
|
|
|
7411 {
|
|
|
7412 /*
|
|
|
7413 * Found the matching char. Copy it to "kword" and go a
|
|
|
7414 * level deeper.
|
|
|
7415 */
|
|
|
7416 if (round[depth] == 1)
|
|
|
7417 {
|
|
|
7418 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
7419 flen);
|
|
|
7420 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
7421 }
|
|
|
7422 else
|
|
|
7423 {
|
|
|
7424 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
7425 ulen);
|
|
|
7426 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
7427 }
|
|
|
7428 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
7429 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
7430
|
|
|
7431 ++depth;
|
|
|
7432 arridx[depth] = tryidx;
|
|
|
7433 round[depth] = 0;
|
|
|
7434 }
|
|
|
7435 }
|
|
|
7436 }
|
|
|
7437
|
|
|
7438 /* Didn't find it: "cannot happen". */
|
|
|
7439 *kword = NUL;
|
|
|
7440 }
|
|
|
7441
|
|
|
7442 /*
|
|
344
|
7443 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
7444 * su->su_sga.
|
|
|
7445 */
|
|
|
7446 static void
|
|
|
7447 score_comp_sal(su)
|
|
|
7448 suginfo_T *su;
|
|
|
7449 {
|
|
|
7450 langp_T *lp;
|
|
|
7451 char_u badsound[MAXWLEN];
|
|
|
7452 int i;
|
|
|
7453 suggest_T *stp;
|
|
|
7454 suggest_T *sstp;
|
|
|
7455 int score;
|
|
|
7456
|
|
|
7457 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
7458 return;
|
|
|
7459
|
|
|
7460 /* Use the sound-folding of the first language that supports it. */
|
|
|
7461 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7462 lp->lp_slang != NULL; ++lp)
|
|
|
7463 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7464 {
|
|
|
7465 /* soundfold the bad word */
|
|
375
|
7466 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
7467
|
|
|
7468 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7469 {
|
|
|
7470 stp = &SUG(su->su_ga, i);
|
|
|
7471
|
|
351
|
7472 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
7473 * sound-a-like score. */
|
|
|
7474 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
7475 if (score < SCORE_MAXMAX)
|
|
|
7476 {
|
|
|
7477 /* Add the suggestion. */
|
|
|
7478 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
7479 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
7480 if (sstp->st_word != NULL)
|
|
|
7481 {
|
|
|
7482 sstp->st_score = score;
|
|
|
7483 sstp->st_altscore = 0;
|
|
|
7484 sstp->st_orglen = stp->st_orglen;
|
|
|
7485 ++su->su_sga.ga_len;
|
|
|
7486 }
|
|
|
7487 }
|
|
|
7488 }
|
|
|
7489 break;
|
|
|
7490 }
|
|
|
7491 }
|
|
|
7492
|
|
|
7493 /*
|
|
|
7494 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
7495 * They are intwined.
|
|
|
7496 */
|
|
|
7497 static void
|
|
|
7498 score_combine(su)
|
|
|
7499 suginfo_T *su;
|
|
|
7500 {
|
|
|
7501 int i;
|
|
|
7502 int j;
|
|
|
7503 garray_T ga;
|
|
|
7504 garray_T *gap;
|
|
|
7505 langp_T *lp;
|
|
|
7506 suggest_T *stp;
|
|
|
7507 char_u *p;
|
|
|
7508 char_u badsound[MAXWLEN];
|
|
|
7509 int round;
|
|
|
7510
|
|
|
7511 /* Add the alternate score to su_ga. */
|
|
|
7512 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7513 lp->lp_slang != NULL; ++lp)
|
|
|
7514 {
|
|
|
7515 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7516 {
|
|
|
7517 /* soundfold the bad word */
|
|
375
|
7518 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
7519
|
|
|
7520 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7521 {
|
|
|
7522 stp = &SUG(su->su_ga, i);
|
|
351
|
7523 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
7524 badsound);
|
|
344
|
7525 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
7526 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
7527 else
|
|
|
7528 stp->st_score = (stp->st_score * 3
|
|
|
7529 + stp->st_altscore) / 4;
|
|
|
7530 stp->st_salscore = FALSE;
|
|
|
7531 }
|
|
|
7532 break;
|
|
|
7533 }
|
|
|
7534 }
|
|
|
7535
|
|
|
7536 /* Add the alternate score to su_sga. */
|
|
|
7537 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
7538 {
|
|
|
7539 stp = &SUG(su->su_sga, i);
|
|
|
7540 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
7541 if (stp->st_score == SCORE_MAXMAX)
|
|
|
7542 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
7543 else
|
|
|
7544 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
7545 stp->st_salscore = TRUE;
|
|
|
7546 }
|
|
|
7547
|
|
|
7548 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
7549 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7550 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
7551
|
|
|
7552 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
7553 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
7554 return;
|
|
|
7555
|
|
|
7556 stp = &SUG(ga, 0);
|
|
|
7557 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
7558 {
|
|
|
7559 /* round 1: get a suggestion from su_ga
|
|
|
7560 * round 2: get a suggestion from su_sga */
|
|
|
7561 for (round = 1; round <= 2; ++round)
|
|
|
7562 {
|
|
|
7563 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
7564 if (i < gap->ga_len)
|
|
|
7565 {
|
|
|
7566 /* Don't add a word if it's already there. */
|
|
|
7567 p = SUG(*gap, i).st_word;
|
|
|
7568 for (j = 0; j < ga.ga_len; ++j)
|
|
|
7569 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
7570 break;
|
|
|
7571 if (j == ga.ga_len)
|
|
|
7572 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
7573 else
|
|
|
7574 vim_free(p);
|
|
|
7575 }
|
|
|
7576 }
|
|
|
7577 }
|
|
|
7578
|
|
|
7579 ga_clear(&su->su_ga);
|
|
|
7580 ga_clear(&su->su_sga);
|
|
|
7581
|
|
|
7582 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
7583 if (ga.ga_len > su->su_maxcount)
|
|
|
7584 {
|
|
|
7585 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
7586 vim_free(stp[i].st_word);
|
|
|
7587 ga.ga_len = su->su_maxcount;
|
|
|
7588 }
|
|
|
7589
|
|
|
7590 su->su_ga = ga;
|
|
|
7591 }
|
|
|
7592
|
|
|
7593 /*
|
|
351
|
7594 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
7595 * badword.
|
|
|
7596 */
|
|
|
7597 static int
|
|
|
7598 stp_sal_score(stp, su, slang, badsound)
|
|
|
7599 suggest_T *stp;
|
|
|
7600 suginfo_T *su;
|
|
|
7601 slang_T *slang;
|
|
|
7602 char_u *badsound; /* sound-folded badword */
|
|
|
7603 {
|
|
|
7604 char_u *p;
|
|
|
7605 char_u badsound2[MAXWLEN];
|
|
|
7606 char_u fword[MAXWLEN];
|
|
|
7607 char_u goodsound[MAXWLEN];
|
|
|
7608
|
|
|
7609 if (stp->st_orglen <= su->su_badlen)
|
|
|
7610 p = badsound;
|
|
|
7611 else
|
|
|
7612 {
|
|
|
7613 /* soundfold the bad word with more characters following */
|
|
|
7614 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
7615
|
|
|
7616 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
7617 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
7618 * removing the space. Don't do it when the good word also contains a
|
|
|
7619 * space. */
|
|
|
7620 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
7621 && *skiptowhite(stp->st_word) == NUL)
|
|
|
7622 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
7623 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
7624
|
|
375
|
7625 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
7626 p = badsound2;
|
|
|
7627 }
|
|
|
7628
|
|
375
|
7629 /* Sound-fold the word and compute the score for the difference. */
|
|
|
7630 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
7631
|
|
|
7632 return soundalike_score(goodsound, p);
|
|
|
7633 }
|
|
|
7634
|
|
|
7635 /*
|
|
323
|
7636 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
7637 */
|
|
|
7638 static void
|
|
346
|
7639 suggest_try_soundalike(su)
|
|
323
|
7640 suginfo_T *su;
|
|
|
7641 {
|
|
|
7642 char_u salword[MAXWLEN];
|
|
|
7643 char_u tword[MAXWLEN];
|
|
|
7644 char_u tsalword[MAXWLEN];
|
|
324
|
7645 idx_T arridx[MAXWLEN];
|
|
323
|
7646 int curi[MAXWLEN];
|
|
|
7647 langp_T *lp;
|
|
|
7648 char_u *byts;
|
|
324
|
7649 idx_T *idxs;
|
|
323
|
7650 int depth;
|
|
|
7651 int c;
|
|
324
|
7652 idx_T n;
|
|
323
|
7653 int round;
|
|
|
7654 int flags;
|
|
344
|
7655 int sound_score;
|
|
|
7656
|
|
|
7657 /* Do this for all languages that support sound folding. */
|
|
323
|
7658 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7659 lp->lp_slang != NULL; ++lp)
|
|
|
7660 {
|
|
|
7661 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7662 {
|
|
|
7663 /* soundfold the bad word */
|
|
375
|
7664 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
7665
|
|
|
7666 /*
|
|
|
7667 * Go through the whole tree, soundfold each word and compare.
|
|
|
7668 * round 1: use the case-folded tree.
|
|
|
7669 * round 2: use the keep-case tree.
|
|
|
7670 */
|
|
|
7671 for (round = 1; round <= 2; ++round)
|
|
|
7672 {
|
|
|
7673 if (round == 1)
|
|
|
7674 {
|
|
|
7675 byts = lp->lp_slang->sl_fbyts;
|
|
|
7676 idxs = lp->lp_slang->sl_fidxs;
|
|
|
7677 }
|
|
|
7678 else
|
|
|
7679 {
|
|
|
7680 byts = lp->lp_slang->sl_kbyts;
|
|
|
7681 idxs = lp->lp_slang->sl_kidxs;
|
|
|
7682 }
|
|
|
7683
|
|
|
7684 depth = 0;
|
|
|
7685 arridx[0] = 0;
|
|
|
7686 curi[0] = 1;
|
|
|
7687 while (depth >= 0 && !got_int)
|
|
|
7688 {
|
|
|
7689 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
7690 {
|
|
323
|
7691 /* Done all bytes at this node, go up one level. */
|
|
|
7692 --depth;
|
|
351
|
7693 line_breakcheck();
|
|
|
7694 }
|
|
323
|
7695 else
|
|
|
7696 {
|
|
|
7697 /* Do one more byte at this node. */
|
|
|
7698 n = arridx[depth] + curi[depth];
|
|
|
7699 ++curi[depth];
|
|
|
7700 c = byts[n];
|
|
|
7701 if (c == 0)
|
|
|
7702 {
|
|
|
7703 /* End of word, deal with the word. */
|
|
324
|
7704 flags = (int)idxs[n];
|
|
323
|
7705 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
7706 {
|
|
|
7707 tword[depth] = NUL;
|
|
375
|
7708 /* Sound-fold. Only in keep-case tree need to
|
|
|
7709 * case-fold the word. */
|
|
|
7710 spell_soundfold(lp->lp_slang, tword,
|
|
|
7711 round == 1, tsalword);
|
|
323
|
7712
|
|
344
|
7713 /* Compute the edit distance between the
|
|
|
7714 * sound-a-like words. */
|
|
|
7715 sound_score = soundalike_score(salword,
|
|
|
7716 tsalword);
|
|
324
|
7717 if (sound_score < SCORE_MAXMAX)
|
|
323
|
7718 {
|
|
324
|
7719 char_u cword[MAXWLEN];
|
|
|
7720 char_u *p;
|
|
344
|
7721 int score;
|
|
324
|
7722
|
|
351
|
7723 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
7724 {
|
|
324
|
7725 /* Need to fix case according to
|
|
|
7726 * "flags". */
|
|
323
|
7727 make_case_word(tword, cword, flags);
|
|
324
|
7728 p = cword;
|
|
323
|
7729 }
|
|
|
7730 else
|
|
324
|
7731 p = tword;
|
|
|
7732
|
|
344
|
7733 if (sps_flags & SPS_DOUBLE)
|
|
|
7734 add_suggestion(su, &su->su_sga, p,
|
|
346
|
7735 su->su_badlen,
|
|
351
|
7736 sound_score, 0, FALSE);
|
|
344
|
7737 else
|
|
|
7738 {
|
|
|
7739 /* Compute the score. */
|
|
|
7740 score = spell_edit_score(
|
|
|
7741 su->su_badword, p);
|
|
|
7742 if (sps_flags & SPS_BEST)
|
|
|
7743 /* give a bonus for the good word
|
|
|
7744 * sounding the same as the bad
|
|
|
7745 * word */
|
|
|
7746 add_suggestion(su, &su->su_ga, p,
|
|
346
|
7747 su->su_badlen,
|
|
344
|
7748 RESCORE(score, sound_score),
|
|
351
|
7749 sound_score, TRUE);
|
|
344
|
7750 else
|
|
|
7751 add_suggestion(su, &su->su_ga, p,
|
|
346
|
7752 su->su_badlen,
|
|
351
|
7753 score + sound_score, 0, FALSE);
|
|
344
|
7754 }
|
|
323
|
7755 }
|
|
|
7756 }
|
|
|
7757
|
|
|
7758 /* Skip over other NUL bytes. */
|
|
|
7759 while (byts[n + 1] == 0)
|
|
|
7760 {
|
|
|
7761 ++n;
|
|
|
7762 ++curi[depth];
|
|
|
7763 }
|
|
|
7764 }
|
|
|
7765 else
|
|
|
7766 {
|
|
|
7767 /* Normal char, go one level deeper. */
|
|
|
7768 tword[depth++] = c;
|
|
|
7769 arridx[depth] = idxs[n];
|
|
|
7770 curi[depth] = 1;
|
|
|
7771 }
|
|
|
7772 }
|
|
|
7773 }
|
|
|
7774 }
|
|
|
7775 }
|
|
|
7776 }
|
|
|
7777 }
|
|
|
7778
|
|
|
7779 /*
|
|
324
|
7780 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
7781 */
|
|
|
7782 static void
|
|
|
7783 make_case_word(fword, cword, flags)
|
|
|
7784 char_u *fword;
|
|
|
7785 char_u *cword;
|
|
|
7786 int flags;
|
|
|
7787 {
|
|
|
7788 if (flags & WF_ALLCAP)
|
|
|
7789 /* Make it all upper-case */
|
|
|
7790 allcap_copy(fword, cword);
|
|
|
7791 else if (flags & WF_ONECAP)
|
|
|
7792 /* Make the first letter upper-case */
|
|
324
|
7793 onecap_copy(fword, cword, TRUE);
|
|
323
|
7794 else
|
|
|
7795 /* Use goodword as-is. */
|
|
|
7796 STRCPY(cword, fword);
|
|
|
7797 }
|
|
|
7798
|
|
330
|
7799 /*
|
|
|
7800 * Use map string "map" for languages "lp".
|
|
|
7801 */
|
|
|
7802 static void
|
|
|
7803 set_map_str(lp, map)
|
|
|
7804 slang_T *lp;
|
|
|
7805 char_u *map;
|
|
|
7806 {
|
|
|
7807 char_u *p;
|
|
|
7808 int headc = 0;
|
|
|
7809 int c;
|
|
|
7810 int i;
|
|
|
7811
|
|
|
7812 if (*map == NUL)
|
|
|
7813 {
|
|
|
7814 lp->sl_has_map = FALSE;
|
|
|
7815 return;
|
|
|
7816 }
|
|
|
7817 lp->sl_has_map = TRUE;
|
|
|
7818
|
|
|
7819 /* Init the array and hash table empty. */
|
|
|
7820 for (i = 0; i < 256; ++i)
|
|
|
7821 lp->sl_map_array[i] = 0;
|
|
|
7822 #ifdef FEAT_MBYTE
|
|
|
7823 hash_init(&lp->sl_map_hash);
|
|
|
7824 #endif
|
|
|
7825
|
|
|
7826 /*
|
|
|
7827 * The similar characters are stored separated with slashes:
|
|
|
7828 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
7829 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
7830 */
|
|
|
7831 for (p = map; *p != NUL; )
|
|
|
7832 {
|
|
|
7833 #ifdef FEAT_MBYTE
|
|
|
7834 c = mb_ptr2char_adv(&p);
|
|
|
7835 #else
|
|
|
7836 c = *p++;
|
|
|
7837 #endif
|
|
|
7838 if (c == '/')
|
|
|
7839 headc = 0;
|
|
|
7840 else
|
|
|
7841 {
|
|
|
7842 if (headc == 0)
|
|
|
7843 headc = c;
|
|
|
7844
|
|
|
7845 #ifdef FEAT_MBYTE
|
|
|
7846 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
7847 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
7848 * a NUL. */
|
|
|
7849 if (c >= 256)
|
|
|
7850 {
|
|
|
7851 int cl = mb_char2len(c);
|
|
|
7852 int headcl = mb_char2len(headc);
|
|
|
7853 char_u *b;
|
|
|
7854 hash_T hash;
|
|
|
7855 hashitem_T *hi;
|
|
|
7856
|
|
|
7857 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
7858 if (b == NULL)
|
|
|
7859 return;
|
|
|
7860 mb_char2bytes(c, b);
|
|
|
7861 b[cl] = NUL;
|
|
|
7862 mb_char2bytes(headc, b + cl + 1);
|
|
|
7863 b[cl + 1 + headcl] = NUL;
|
|
|
7864 hash = hash_hash(b);
|
|
|
7865 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
7866 if (HASHITEM_EMPTY(hi))
|
|
|
7867 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
7868 else
|
|
|
7869 {
|
|
|
7870 /* This should have been checked when generating the .spl
|
|
|
7871 * file. */
|
|
|
7872 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
7873 vim_free(b);
|
|
|
7874 }
|
|
|
7875 }
|
|
|
7876 else
|
|
|
7877 #endif
|
|
|
7878 lp->sl_map_array[c] = headc;
|
|
|
7879 }
|
|
|
7880 }
|
|
|
7881 }
|
|
|
7882
|
|
323
|
7883 /*
|
|
|
7884 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
7885 * lines in the .aff file.
|
|
|
7886 */
|
|
|
7887 static int
|
|
|
7888 similar_chars(slang, c1, c2)
|
|
|
7889 slang_T *slang;
|
|
|
7890 int c1;
|
|
|
7891 int c2;
|
|
|
7892 {
|
|
330
|
7893 int m1, m2;
|
|
|
7894 #ifdef FEAT_MBYTE
|
|
|
7895 char_u buf[MB_MAXBYTES];
|
|
|
7896 hashitem_T *hi;
|
|
|
7897
|
|
|
7898 if (c1 >= 256)
|
|
|
7899 {
|
|
|
7900 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
7901 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
7902 if (HASHITEM_EMPTY(hi))
|
|
|
7903 m1 = 0;
|
|
|
7904 else
|
|
|
7905 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
7906 }
|
|
|
7907 else
|
|
|
7908 #endif
|
|
|
7909 m1 = slang->sl_map_array[c1];
|
|
|
7910 if (m1 == 0)
|
|
323
|
7911 return FALSE;
|
|
330
|
7912
|
|
|
7913
|
|
|
7914 #ifdef FEAT_MBYTE
|
|
|
7915 if (c2 >= 256)
|
|
|
7916 {
|
|
|
7917 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
7918 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
7919 if (HASHITEM_EMPTY(hi))
|
|
|
7920 m2 = 0;
|
|
|
7921 else
|
|
|
7922 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
7923 }
|
|
|
7924 else
|
|
|
7925 #endif
|
|
|
7926 m2 = slang->sl_map_array[c2];
|
|
|
7927
|
|
|
7928 return m1 == m2;
|
|
323
|
7929 }
|
|
|
7930
|
|
|
7931 /*
|
|
|
7932 * Add a suggestion to the list of suggestions.
|
|
|
7933 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
7934 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
7935 * with spell_edit_score().
|
|
|
7936 */
|
|
|
7937 static void
|
|
351
|
7938 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
7939 suginfo_T *su;
|
|
344
|
7940 garray_T *gap;
|
|
323
|
7941 char_u *goodword;
|
|
346
|
7942 int badlen; /* length of bad word used */
|
|
324
|
7943 int score;
|
|
351
|
7944 int altscore;
|
|
344
|
7945 int had_bonus; /* value for st_had_bonus */
|
|
323
|
7946 {
|
|
|
7947 suggest_T *stp;
|
|
|
7948 int i;
|
|
346
|
7949 char_u *p = NULL;
|
|
|
7950 int c = 0;
|
|
323
|
7951
|
|
|
7952 /* Check that the word wasn't banned. */
|
|
|
7953 if (was_banned(su, goodword))
|
|
|
7954 return;
|
|
|
7955
|
|
346
|
7956 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
7957 * Remove the common part from "goodword". */
|
|
|
7958 i = badlen - su->su_badlen;
|
|
|
7959 if (i > 0)
|
|
|
7960 {
|
|
|
7961 /* This assumes there was no case folding or it didn't change the
|
|
|
7962 * length... */
|
|
|
7963 p = goodword + STRLEN(goodword) - i;
|
|
|
7964 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
7965 {
|
|
|
7966 badlen = su->su_badlen;
|
|
|
7967 c = *p;
|
|
|
7968 *p = NUL;
|
|
|
7969 }
|
|
|
7970 else
|
|
|
7971 p = NULL;
|
|
|
7972 }
|
|
372
|
7973 else if (i < 0)
|
|
|
7974 {
|
|
|
7975 /* When replacing part of the word check that we actually change
|
|
|
7976 * something. For "the the" a suggestion can be replacing the first
|
|
|
7977 * "the" with itself, since "the" wasn't banned. */
|
|
|
7978 if (badlen == STRLEN(goodword)
|
|
|
7979 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
7980 return;
|
|
|
7981 }
|
|
|
7982
|
|
346
|
7983
|
|
323
|
7984 if (score <= su->su_maxscore)
|
|
|
7985 {
|
|
366
|
7986 /* Check if the word is already there. Also check the length that is
|
|
|
7987 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
7988 * "thes" -> "these". */
|
|
344
|
7989 stp = &SUG(*gap, 0);
|
|
|
7990 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
7991 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
7992 && stp[i].st_orglen == badlen)
|
|
323
|
7993 {
|
|
|
7994 /* Found it. Remember the lowest score. */
|
|
|
7995 if (stp[i].st_score > score)
|
|
324
|
7996 {
|
|
323
|
7997 stp[i].st_score = score;
|
|
324
|
7998 stp[i].st_had_bonus = had_bonus;
|
|
|
7999 }
|
|
323
|
8000 break;
|
|
|
8001 }
|
|
|
8002
|
|
344
|
8003 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
8004 {
|
|
|
8005 /* Add a suggestion. */
|
|
344
|
8006 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
8007 stp->st_word = vim_strsave(goodword);
|
|
|
8008 if (stp->st_word != NULL)
|
|
|
8009 {
|
|
|
8010 stp->st_score = score;
|
|
351
|
8011 stp->st_altscore = altscore;
|
|
324
|
8012 stp->st_had_bonus = had_bonus;
|
|
346
|
8013 stp->st_orglen = badlen;
|
|
344
|
8014 ++gap->ga_len;
|
|
323
|
8015
|
|
|
8016 /* If we have too many suggestions now, sort the list and keep
|
|
|
8017 * the best suggestions. */
|
|
344
|
8018 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
8019 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
8020 SUG_CLEAN_COUNT(su));
|
|
323
|
8021 }
|
|
|
8022 }
|
|
|
8023 }
|
|
346
|
8024
|
|
|
8025 if (p != NULL)
|
|
|
8026 *p = c; /* restore "goodword" */
|
|
323
|
8027 }
|
|
|
8028
|
|
|
8029 /*
|
|
|
8030 * Add a word to be banned.
|
|
|
8031 */
|
|
|
8032 static void
|
|
|
8033 add_banned(su, word)
|
|
|
8034 suginfo_T *su;
|
|
|
8035 char_u *word;
|
|
|
8036 {
|
|
|
8037 char_u *s = vim_strsave(word);
|
|
|
8038 hash_T hash;
|
|
|
8039 hashitem_T *hi;
|
|
|
8040
|
|
|
8041 if (s != NULL)
|
|
|
8042 {
|
|
|
8043 hash = hash_hash(s);
|
|
|
8044 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
8045 if (HASHITEM_EMPTY(hi))
|
|
|
8046 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
8047 else
|
|
|
8048 vim_free(s);
|
|
323
|
8049 }
|
|
|
8050 }
|
|
|
8051
|
|
|
8052 /*
|
|
|
8053 * Return TRUE if a word appears in the list of banned words.
|
|
|
8054 */
|
|
|
8055 static int
|
|
|
8056 was_banned(su, word)
|
|
|
8057 suginfo_T *su;
|
|
|
8058 char_u *word;
|
|
|
8059 {
|
|
324
|
8060 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
8061
|
|
|
8062 return !HASHITEM_EMPTY(hi);
|
|
323
|
8063 }
|
|
|
8064
|
|
|
8065 /*
|
|
|
8066 * Free the banned words in "su".
|
|
|
8067 */
|
|
|
8068 static void
|
|
|
8069 free_banned(su)
|
|
|
8070 suginfo_T *su;
|
|
|
8071 {
|
|
|
8072 int todo;
|
|
|
8073 hashitem_T *hi;
|
|
|
8074
|
|
|
8075 todo = su->su_banned.ht_used;
|
|
|
8076 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
8077 {
|
|
|
8078 if (!HASHITEM_EMPTY(hi))
|
|
|
8079 {
|
|
|
8080 vim_free(hi->hi_key);
|
|
|
8081 --todo;
|
|
|
8082 }
|
|
|
8083 }
|
|
|
8084 hash_clear(&su->su_banned);
|
|
|
8085 }
|
|
|
8086
|
|
324
|
8087 /*
|
|
|
8088 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
8089 * thus only done for the final results.
|
|
|
8090 */
|
|
|
8091 static void
|
|
|
8092 rescore_suggestions(su)
|
|
|
8093 suginfo_T *su;
|
|
|
8094 {
|
|
|
8095 langp_T *lp;
|
|
|
8096 suggest_T *stp;
|
|
|
8097 char_u sal_badword[MAXWLEN];
|
|
|
8098 int i;
|
|
|
8099
|
|
|
8100 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8101 lp->lp_slang != NULL; ++lp)
|
|
|
8102 {
|
|
|
8103 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8104 {
|
|
|
8105 /* soundfold the bad word */
|
|
375
|
8106 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
8107
|
|
|
8108 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8109 {
|
|
344
|
8110 stp = &SUG(su->su_ga, i);
|
|
324
|
8111 if (!stp->st_had_bonus)
|
|
|
8112 {
|
|
351
|
8113 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
8114 lp->lp_slang, sal_badword);
|
|
|
8115 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8116 stp->st_altscore = SCORE_BIG;
|
|
|
8117 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
8118 }
|
|
|
8119 }
|
|
|
8120 break;
|
|
|
8121 }
|
|
|
8122 }
|
|
|
8123 }
|
|
|
8124
|
|
323
|
8125 static int
|
|
|
8126 #ifdef __BORLANDC__
|
|
|
8127 _RTLENTRYF
|
|
|
8128 #endif
|
|
|
8129 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
8130
|
|
|
8131 /*
|
|
|
8132 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
8133 */
|
|
|
8134 static int
|
|
|
8135 #ifdef __BORLANDC__
|
|
|
8136 _RTLENTRYF
|
|
|
8137 #endif
|
|
|
8138 sug_compare(s1, s2)
|
|
|
8139 const void *s1;
|
|
|
8140 const void *s2;
|
|
|
8141 {
|
|
|
8142 suggest_T *p1 = (suggest_T *)s1;
|
|
|
8143 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
8144 int n = p1->st_score - p2->st_score;
|
|
|
8145
|
|
|
8146 if (n == 0)
|
|
|
8147 return p1->st_altscore - p2->st_altscore;
|
|
|
8148 return n;
|
|
323
|
8149 }
|
|
|
8150
|
|
|
8151 /*
|
|
|
8152 * Cleanup the suggestions:
|
|
|
8153 * - Sort on score.
|
|
|
8154 * - Remove words that won't be displayed.
|
|
344
|
8155 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
8156 */
|
|
344
|
8157 static int
|
|
|
8158 cleanup_suggestions(gap, maxscore, keep)
|
|
|
8159 garray_T *gap;
|
|
|
8160 int maxscore;
|
|
324
|
8161 int keep; /* nr of suggestions to keep */
|
|
323
|
8162 {
|
|
344
|
8163 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
8164 int i;
|
|
|
8165
|
|
|
8166 /* Sort the list. */
|
|
344
|
8167 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
8168
|
|
|
8169 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
8170 if (gap->ga_len > keep)
|
|
323
|
8171 {
|
|
344
|
8172 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
8173 vim_free(stp[i].st_word);
|
|
344
|
8174 gap->ga_len = keep;
|
|
|
8175 return stp[keep - 1].st_score;
|
|
323
|
8176 }
|
|
344
|
8177 return maxscore;
|
|
323
|
8178 }
|
|
|
8179
|
|
372
|
8180 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
8181 /*
|
|
|
8182 * Soundfold a string, for soundfold().
|
|
|
8183 * Result is in allocated memory, NULL for an error.
|
|
|
8184 */
|
|
|
8185 char_u *
|
|
|
8186 eval_soundfold(word)
|
|
|
8187 char_u *word;
|
|
|
8188 {
|
|
|
8189 langp_T *lp;
|
|
|
8190 char_u sound[MAXWLEN];
|
|
|
8191
|
|
|
8192 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
8193 /* Use the sound-folding of the first language that supports it. */
|
|
|
8194 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8195 lp->lp_slang != NULL; ++lp)
|
|
|
8196 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8197 {
|
|
|
8198 /* soundfold the word */
|
|
375
|
8199 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
8200 return vim_strsave(sound);
|
|
|
8201 }
|
|
|
8202
|
|
|
8203 /* No language with sound folding, return word as-is. */
|
|
|
8204 return vim_strsave(word);
|
|
|
8205 }
|
|
|
8206 #endif
|
|
|
8207
|
|
323
|
8208 /*
|
|
|
8209 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
8210 */
|
|
|
8211 static void
|
|
375
|
8212 spell_soundfold(slang, inword, folded, res)
|
|
|
8213 slang_T *slang;
|
|
|
8214 char_u *inword;
|
|
|
8215 int folded; /* "inword" is already case-folded */
|
|
|
8216 char_u *res;
|
|
|
8217 {
|
|
|
8218 char_u fword[MAXWLEN];
|
|
|
8219 char_u *word;
|
|
|
8220
|
|
|
8221 if (slang->sl_sofo)
|
|
|
8222 /* SOFOFROM and SOFOTO used */
|
|
|
8223 spell_soundfold_sofo(slang, inword, res);
|
|
|
8224 else
|
|
|
8225 {
|
|
|
8226 /* SAL items used. Requires the word to be case-folded. */
|
|
|
8227 if (folded)
|
|
|
8228 word = inword;
|
|
|
8229 else
|
|
|
8230 {
|
|
|
8231 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
8232 word = fword;
|
|
|
8233 }
|
|
|
8234
|
|
|
8235 #ifdef FEAT_MBYTE
|
|
|
8236 if (has_mbyte)
|
|
|
8237 spell_soundfold_wsal(slang, word, res);
|
|
|
8238 else
|
|
|
8239 #endif
|
|
|
8240 spell_soundfold_sal(slang, word, res);
|
|
|
8241 }
|
|
|
8242 }
|
|
|
8243
|
|
|
8244 /*
|
|
|
8245 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
8246 * SOFOTO lines.
|
|
|
8247 */
|
|
|
8248 static void
|
|
|
8249 spell_soundfold_sofo(slang, inword, res)
|
|
|
8250 slang_T *slang;
|
|
|
8251 char_u *inword;
|
|
|
8252 char_u *res;
|
|
|
8253 {
|
|
|
8254 char_u *s;
|
|
|
8255 int ri = 0;
|
|
|
8256 int c;
|
|
|
8257
|
|
|
8258 #ifdef FEAT_MBYTE
|
|
|
8259 if (has_mbyte)
|
|
|
8260 {
|
|
|
8261 int prevc = 0;
|
|
|
8262 int *ip;
|
|
|
8263
|
|
|
8264 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
8265 * 255, sl_sal the rest. */
|
|
|
8266 for (s = inword; *s != NUL; )
|
|
|
8267 {
|
|
|
8268 c = mb_ptr2char_adv(&s);
|
|
|
8269 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
8270 c = ' ';
|
|
|
8271 else if (c < 256)
|
|
|
8272 c = slang->sl_sal_first[c];
|
|
|
8273 else
|
|
|
8274 {
|
|
|
8275 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
8276 if (ip == NULL) /* empty list, can't match */
|
|
|
8277 c = NUL;
|
|
|
8278 else
|
|
|
8279 for (;;) /* find "c" in the list */
|
|
|
8280 {
|
|
|
8281 if (*ip == 0) /* not found */
|
|
|
8282 {
|
|
|
8283 c = NUL;
|
|
|
8284 break;
|
|
|
8285 }
|
|
|
8286 if (*ip == c) /* match! */
|
|
|
8287 {
|
|
|
8288 c = ip[1];
|
|
|
8289 break;
|
|
|
8290 }
|
|
|
8291 ip += 2;
|
|
|
8292 }
|
|
|
8293 }
|
|
|
8294
|
|
|
8295 if (c != NUL && c != prevc)
|
|
|
8296 {
|
|
|
8297 ri += mb_char2bytes(c, res + ri);
|
|
|
8298 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
8299 break;
|
|
|
8300 prevc = c;
|
|
|
8301 }
|
|
|
8302 }
|
|
|
8303 }
|
|
|
8304 else
|
|
|
8305 #endif
|
|
|
8306 {
|
|
|
8307 /* The sl_sal_first[] table contains the translation. */
|
|
|
8308 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
8309 {
|
|
|
8310 if (vim_iswhite(c))
|
|
|
8311 c = ' ';
|
|
|
8312 else
|
|
|
8313 c = slang->sl_sal_first[c];
|
|
|
8314 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
8315 res[ri++] = c;
|
|
|
8316 }
|
|
|
8317 }
|
|
|
8318
|
|
|
8319 res[ri] = NUL;
|
|
|
8320 }
|
|
|
8321
|
|
|
8322 static void
|
|
|
8323 spell_soundfold_sal(slang, inword, res)
|
|
323
|
8324 slang_T *slang;
|
|
|
8325 char_u *inword;
|
|
|
8326 char_u *res;
|
|
|
8327 {
|
|
344
|
8328 salitem_T *smp;
|
|
323
|
8329 char_u word[MAXWLEN];
|
|
375
|
8330 char_u *s = inword;
|
|
323
|
8331 char_u *t;
|
|
344
|
8332 char_u *pf;
|
|
323
|
8333 int i, j, z;
|
|
344
|
8334 int reslen;
|
|
323
|
8335 int n, k = 0;
|
|
|
8336 int z0;
|
|
|
8337 int k0;
|
|
|
8338 int n0;
|
|
|
8339 int c;
|
|
|
8340 int pri;
|
|
|
8341 int p0 = -333;
|
|
|
8342 int c0;
|
|
|
8343
|
|
324
|
8344 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
8345 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
8346 if (slang->sl_rem_accents)
|
|
|
8347 {
|
|
|
8348 t = word;
|
|
375
|
8349 while (*s != NUL)
|
|
323
|
8350 {
|
|
324
|
8351 if (vim_iswhite(*s))
|
|
344
|
8352 {
|
|
|
8353 *t++ = ' ';
|
|
|
8354 s = skipwhite(s);
|
|
|
8355 }
|
|
323
|
8356 else
|
|
|
8357 {
|
|
358
|
8358 if (spell_iswordp(s))
|
|
323
|
8359 *t++ = *s;
|
|
|
8360 ++s;
|
|
|
8361 }
|
|
|
8362 }
|
|
|
8363 *t = NUL;
|
|
|
8364 }
|
|
|
8365 else
|
|
375
|
8366 STRCPY(word, s);
|
|
323
|
8367
|
|
344
|
8368 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
8369
|
|
|
8370 /*
|
|
|
8371 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
8372 * Changed to keep spaces.
|
|
323
|
8373 */
|
|
344
|
8374 i = reslen = z = 0;
|
|
323
|
8375 while ((c = word[i]) != NUL)
|
|
|
8376 {
|
|
344
|
8377 /* Start with the first rule that has the character in the word. */
|
|
323
|
8378 n = slang->sl_sal_first[c];
|
|
|
8379 z0 = 0;
|
|
|
8380
|
|
|
8381 if (n >= 0)
|
|
|
8382 {
|
|
|
8383 /* check all rules for the same letter */
|
|
344
|
8384 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
8385 {
|
|
344
|
8386 /* Quickly skip entries that don't match the word. Most
|
|
|
8387 * entries are less then three chars, optimize for that. */
|
|
|
8388 k = smp[n].sm_leadlen;
|
|
|
8389 if (k > 1)
|
|
323
|
8390 {
|
|
344
|
8391 if (word[i + 1] != s[1])
|
|
|
8392 continue;
|
|
|
8393 if (k > 2)
|
|
|
8394 {
|
|
|
8395 for (j = 2; j < k; ++j)
|
|
|
8396 if (word[i + j] != s[j])
|
|
|
8397 break;
|
|
|
8398 if (j < k)
|
|
|
8399 continue;
|
|
|
8400 }
|
|
323
|
8401 }
|
|
|
8402
|
|
375
|
8403 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
8404 {
|
|
375
|
8405 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
8406 while (*pf != NUL && *pf != word[i + k])
|
|
|
8407 ++pf;
|
|
|
8408 if (*pf == NUL)
|
|
|
8409 continue;
|
|
|
8410 ++k;
|
|
323
|
8411 }
|
|
344
|
8412 s = smp[n].sm_rules;
|
|
|
8413 pri = 5; /* default priority */
|
|
323
|
8414
|
|
|
8415 p0 = *s;
|
|
|
8416 k0 = k;
|
|
|
8417 while (*s == '-' && k > 1)
|
|
|
8418 {
|
|
|
8419 k--;
|
|
|
8420 s++;
|
|
|
8421 }
|
|
|
8422 if (*s == '<')
|
|
|
8423 s++;
|
|
344
|
8424 if (VIM_ISDIGIT(*s))
|
|
323
|
8425 {
|
|
|
8426 /* determine priority */
|
|
|
8427 pri = *s - '0';
|
|
|
8428 s++;
|
|
|
8429 }
|
|
|
8430 if (*s == '^' && *(s + 1) == '^')
|
|
|
8431 s++;
|
|
|
8432
|
|
|
8433 if (*s == NUL
|
|
|
8434 || (*s == '^'
|
|
324
|
8435 && (i == 0 || !(word[i - 1] == ' '
|
|
358
|
8436 || spell_iswordp(word + i - 1)))
|
|
323
|
8437 && (*(s + 1) != '$'
|
|
358
|
8438 || (!spell_iswordp(word + i + k0))))
|
|
323
|
8439 || (*s == '$' && i > 0
|
|
358
|
8440 && spell_iswordp(word + i - 1)
|
|
|
8441 && (!spell_iswordp(word + i + k0))))
|
|
323
|
8442 {
|
|
|
8443 /* search for followup rules, if: */
|
|
|
8444 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
8445 c0 = word[i + k - 1];
|
|
|
8446 n0 = slang->sl_sal_first[c0];
|
|
|
8447
|
|
|
8448 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
8449 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
8450 {
|
|
|
8451 /* test follow-up rule for "word[i + k]" */
|
|
344
|
8452 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
8453 {
|
|
344
|
8454 /* Quickly skip entries that don't match the word.
|
|
|
8455 * */
|
|
|
8456 k0 = smp[n0].sm_leadlen;
|
|
|
8457 if (k0 > 1)
|
|
323
|
8458 {
|
|
344
|
8459 if (word[i + k] != s[1])
|
|
|
8460 continue;
|
|
|
8461 if (k0 > 2)
|
|
|
8462 {
|
|
|
8463 pf = word + i + k + 1;
|
|
|
8464 for (j = 2; j < k0; ++j)
|
|
|
8465 if (*pf++ != s[j])
|
|
|
8466 break;
|
|
|
8467 if (j < k0)
|
|
|
8468 continue;
|
|
|
8469 }
|
|
323
|
8470 }
|
|
344
|
8471 k0 += k - 1;
|
|
|
8472
|
|
375
|
8473 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
8474 {
|
|
344
|
8475 /* Check for match with one of the chars in
|
|
375
|
8476 * "sm_oneof". */
|
|
344
|
8477 while (*pf != NUL && *pf != word[i + k0])
|
|
|
8478 ++pf;
|
|
|
8479 if (*pf == NUL)
|
|
|
8480 continue;
|
|
|
8481 ++k0;
|
|
323
|
8482 }
|
|
344
|
8483
|
|
|
8484 p0 = 5;
|
|
|
8485 s = smp[n0].sm_rules;
|
|
323
|
8486 while (*s == '-')
|
|
|
8487 {
|
|
344
|
8488 /* "k0" gets NOT reduced because
|
|
|
8489 * "if (k0 == k)" */
|
|
323
|
8490 s++;
|
|
|
8491 }
|
|
|
8492 if (*s == '<')
|
|
|
8493 s++;
|
|
344
|
8494 if (VIM_ISDIGIT(*s))
|
|
323
|
8495 {
|
|
|
8496 p0 = *s - '0';
|
|
|
8497 s++;
|
|
|
8498 }
|
|
|
8499
|
|
|
8500 if (*s == NUL
|
|
|
8501 /* *s == '^' cuts */
|
|
|
8502 || (*s == '$'
|
|
358
|
8503 && !spell_iswordp(word + i + k0)))
|
|
323
|
8504 {
|
|
|
8505 if (k0 == k)
|
|
|
8506 /* this is just a piece of the string */
|
|
|
8507 continue;
|
|
|
8508
|
|
|
8509 if (p0 < pri)
|
|
|
8510 /* priority too low */
|
|
|
8511 continue;
|
|
|
8512 /* rule fits; stop search */
|
|
|
8513 break;
|
|
|
8514 }
|
|
|
8515 }
|
|
|
8516
|
|
344
|
8517 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
8518 continue;
|
|
|
8519 }
|
|
|
8520
|
|
|
8521 /* replace string */
|
|
344
|
8522 s = smp[n].sm_to;
|
|
|
8523 pf = smp[n].sm_rules;
|
|
|
8524 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
8525 if (p0 == 1 && z == 0)
|
|
|
8526 {
|
|
|
8527 /* rule with '<' is used */
|
|
344
|
8528 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
8529 || res[reslen - 1] == *s))
|
|
|
8530 reslen--;
|
|
323
|
8531 z0 = 1;
|
|
|
8532 z = 1;
|
|
|
8533 k0 = 0;
|
|
372
|
8534 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
8535 {
|
|
|
8536 word[i + k0] = *s;
|
|
|
8537 k0++;
|
|
|
8538 s++;
|
|
|
8539 }
|
|
|
8540 if (k > k0)
|
|
|
8541 mch_memmove(word + i + k0, word + i + k,
|
|
|
8542 STRLEN(word + i + k) + 1);
|
|
|
8543
|
|
|
8544 /* new "actual letter" */
|
|
|
8545 c = word[i];
|
|
|
8546 }
|
|
|
8547 else
|
|
|
8548 {
|
|
|
8549 /* no '<' rule used */
|
|
|
8550 i += k - 1;
|
|
|
8551 z = 0;
|
|
344
|
8552 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
8553 {
|
|
344
|
8554 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
8555 res[reslen++] = *s;
|
|
323
|
8556 s++;
|
|
|
8557 }
|
|
|
8558 /* new "actual letter" */
|
|
|
8559 c = *s;
|
|
344
|
8560 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
8561 {
|
|
|
8562 if (c != NUL)
|
|
372
|
8563 res[reslen++] = c;
|
|
323
|
8564 mch_memmove(word, word + i + 1,
|
|
|
8565 STRLEN(word + i + 1) + 1);
|
|
|
8566 i = 0;
|
|
|
8567 z0 = 1;
|
|
|
8568 }
|
|
|
8569 }
|
|
|
8570 break;
|
|
|
8571 }
|
|
|
8572 }
|
|
|
8573 }
|
|
324
|
8574 else if (vim_iswhite(c))
|
|
|
8575 {
|
|
|
8576 c = ' ';
|
|
|
8577 k = 1;
|
|
|
8578 }
|
|
323
|
8579
|
|
|
8580 if (z0 == 0)
|
|
|
8581 {
|
|
344
|
8582 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
8583 && (!slang->sl_collapse || reslen == 0
|
|
|
8584 || res[reslen - 1] != c))
|
|
323
|
8585 /* condense only double letters */
|
|
372
|
8586 res[reslen++] = c;
|
|
323
|
8587
|
|
|
8588 i++;
|
|
|
8589 z = 0;
|
|
|
8590 k = 0;
|
|
|
8591 }
|
|
|
8592 }
|
|
|
8593
|
|
344
|
8594 res[reslen] = NUL;
|
|
323
|
8595 }
|
|
|
8596
|
|
372
|
8597 #ifdef FEAT_MBYTE
|
|
|
8598 /*
|
|
|
8599 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
8600 * Multi-byte version of spell_soundfold().
|
|
|
8601 */
|
|
|
8602 static void
|
|
375
|
8603 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
8604 slang_T *slang;
|
|
|
8605 char_u *inword;
|
|
|
8606 char_u *res;
|
|
|
8607 {
|
|
375
|
8608 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
8609 int word[MAXWLEN];
|
|
|
8610 int wres[MAXWLEN];
|
|
|
8611 int l;
|
|
|
8612 char_u *s;
|
|
|
8613 int *ws;
|
|
|
8614 char_u *t;
|
|
|
8615 int *pf;
|
|
|
8616 int i, j, z;
|
|
|
8617 int reslen;
|
|
|
8618 int n, k = 0;
|
|
|
8619 int z0;
|
|
|
8620 int k0;
|
|
|
8621 int n0;
|
|
|
8622 int c;
|
|
|
8623 int pri;
|
|
|
8624 int p0 = -333;
|
|
|
8625 int c0;
|
|
|
8626 int did_white = FALSE;
|
|
|
8627
|
|
|
8628 /*
|
|
|
8629 * Convert the multi-byte string to a wide-character string.
|
|
|
8630 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
8631 * But keep white space.
|
|
|
8632 */
|
|
|
8633 n = 0;
|
|
|
8634 for (s = inword; *s != NUL; )
|
|
|
8635 {
|
|
|
8636 t = s;
|
|
|
8637 c = mb_ptr2char_adv(&s);
|
|
|
8638 if (slang->sl_rem_accents)
|
|
|
8639 {
|
|
|
8640 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
8641 {
|
|
|
8642 if (did_white)
|
|
|
8643 continue;
|
|
|
8644 c = ' ';
|
|
|
8645 did_white = TRUE;
|
|
|
8646 }
|
|
|
8647 else
|
|
|
8648 {
|
|
|
8649 did_white = FALSE;
|
|
|
8650 if (!spell_iswordp(t))
|
|
|
8651 continue;
|
|
|
8652 }
|
|
|
8653 }
|
|
|
8654 word[n++] = c;
|
|
|
8655 }
|
|
|
8656 word[n] = NUL;
|
|
|
8657
|
|
|
8658 /*
|
|
|
8659 * This comes from Aspell phonet.cpp.
|
|
|
8660 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
8661 * Changed to keep spaces.
|
|
|
8662 */
|
|
|
8663 i = reslen = z = 0;
|
|
|
8664 while ((c = word[i]) != NUL)
|
|
|
8665 {
|
|
|
8666 /* Start with the first rule that has the character in the word. */
|
|
|
8667 n = slang->sl_sal_first[c & 0xff];
|
|
|
8668 z0 = 0;
|
|
|
8669
|
|
|
8670 if (n >= 0)
|
|
|
8671 {
|
|
375
|
8672 /* check all rules for the same index byte */
|
|
372
|
8673 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
8674 {
|
|
|
8675 /* Quickly skip entries that don't match the word. Most
|
|
|
8676 * entries are less then three chars, optimize for that. */
|
|
375
|
8677 if (c != ws[0])
|
|
|
8678 continue;
|
|
372
|
8679 k = smp[n].sm_leadlen;
|
|
|
8680 if (k > 1)
|
|
|
8681 {
|
|
|
8682 if (word[i + 1] != ws[1])
|
|
|
8683 continue;
|
|
|
8684 if (k > 2)
|
|
|
8685 {
|
|
|
8686 for (j = 2; j < k; ++j)
|
|
|
8687 if (word[i + j] != ws[j])
|
|
|
8688 break;
|
|
|
8689 if (j < k)
|
|
|
8690 continue;
|
|
|
8691 }
|
|
|
8692 }
|
|
|
8693
|
|
375
|
8694 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
8695 {
|
|
375
|
8696 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
8697 while (*pf != NUL && *pf != word[i + k])
|
|
|
8698 ++pf;
|
|
|
8699 if (*pf == NUL)
|
|
|
8700 continue;
|
|
|
8701 ++k;
|
|
|
8702 }
|
|
|
8703 s = smp[n].sm_rules;
|
|
|
8704 pri = 5; /* default priority */
|
|
|
8705
|
|
|
8706 p0 = *s;
|
|
|
8707 k0 = k;
|
|
|
8708 while (*s == '-' && k > 1)
|
|
|
8709 {
|
|
|
8710 k--;
|
|
|
8711 s++;
|
|
|
8712 }
|
|
|
8713 if (*s == '<')
|
|
|
8714 s++;
|
|
|
8715 if (VIM_ISDIGIT(*s))
|
|
|
8716 {
|
|
|
8717 /* determine priority */
|
|
|
8718 pri = *s - '0';
|
|
|
8719 s++;
|
|
|
8720 }
|
|
|
8721 if (*s == '^' && *(s + 1) == '^')
|
|
|
8722 s++;
|
|
|
8723
|
|
|
8724 if (*s == NUL
|
|
|
8725 || (*s == '^'
|
|
|
8726 && (i == 0 || !(word[i - 1] == ' '
|
|
|
8727 || spell_iswordp_w(word + i - 1)))
|
|
|
8728 && (*(s + 1) != '$'
|
|
|
8729 || (!spell_iswordp_w(word + i + k0))))
|
|
|
8730 || (*s == '$' && i > 0
|
|
|
8731 && spell_iswordp_w(word + i - 1)
|
|
|
8732 && (!spell_iswordp_w(word + i + k0))))
|
|
|
8733 {
|
|
|
8734 /* search for followup rules, if: */
|
|
|
8735 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
8736 c0 = word[i + k - 1];
|
|
|
8737 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
8738
|
|
|
8739 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
8740 && p0 != '-' && word[i + k] != NUL)
|
|
|
8741 {
|
|
375
|
8742 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
8743 * all entries with the same index byte. */
|
|
372
|
8744 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
8745 == (c0 & 0xff); ++n0)
|
|
|
8746 {
|
|
|
8747 /* Quickly skip entries that don't match the word.
|
|
375
|
8748 */
|
|
|
8749 if (c0 != ws[0])
|
|
|
8750 continue;
|
|
372
|
8751 k0 = smp[n0].sm_leadlen;
|
|
|
8752 if (k0 > 1)
|
|
|
8753 {
|
|
|
8754 if (word[i + k] != ws[1])
|
|
|
8755 continue;
|
|
|
8756 if (k0 > 2)
|
|
|
8757 {
|
|
|
8758 pf = word + i + k + 1;
|
|
|
8759 for (j = 2; j < k0; ++j)
|
|
|
8760 if (*pf++ != ws[j])
|
|
|
8761 break;
|
|
|
8762 if (j < k0)
|
|
|
8763 continue;
|
|
|
8764 }
|
|
|
8765 }
|
|
|
8766 k0 += k - 1;
|
|
|
8767
|
|
375
|
8768 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
8769 {
|
|
|
8770 /* Check for match with one of the chars in
|
|
375
|
8771 * "sm_oneof". */
|
|
372
|
8772 while (*pf != NUL && *pf != word[i + k0])
|
|
|
8773 ++pf;
|
|
|
8774 if (*pf == NUL)
|
|
|
8775 continue;
|
|
|
8776 ++k0;
|
|
|
8777 }
|
|
|
8778
|
|
|
8779 p0 = 5;
|
|
|
8780 s = smp[n0].sm_rules;
|
|
|
8781 while (*s == '-')
|
|
|
8782 {
|
|
|
8783 /* "k0" gets NOT reduced because
|
|
|
8784 * "if (k0 == k)" */
|
|
|
8785 s++;
|
|
|
8786 }
|
|
|
8787 if (*s == '<')
|
|
|
8788 s++;
|
|
|
8789 if (VIM_ISDIGIT(*s))
|
|
|
8790 {
|
|
|
8791 p0 = *s - '0';
|
|
|
8792 s++;
|
|
|
8793 }
|
|
|
8794
|
|
|
8795 if (*s == NUL
|
|
|
8796 /* *s == '^' cuts */
|
|
|
8797 || (*s == '$'
|
|
|
8798 && !spell_iswordp_w(word + i + k0)))
|
|
|
8799 {
|
|
|
8800 if (k0 == k)
|
|
|
8801 /* this is just a piece of the string */
|
|
|
8802 continue;
|
|
|
8803
|
|
|
8804 if (p0 < pri)
|
|
|
8805 /* priority too low */
|
|
|
8806 continue;
|
|
|
8807 /* rule fits; stop search */
|
|
|
8808 break;
|
|
|
8809 }
|
|
|
8810 }
|
|
|
8811
|
|
|
8812 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
8813 == (c0 & 0xff))
|
|
|
8814 continue;
|
|
|
8815 }
|
|
|
8816
|
|
|
8817 /* replace string */
|
|
|
8818 ws = smp[n].sm_to_w;
|
|
|
8819 s = smp[n].sm_rules;
|
|
|
8820 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
8821 if (p0 == 1 && z == 0)
|
|
|
8822 {
|
|
|
8823 /* rule with '<' is used */
|
|
|
8824 if (reslen > 0 && *ws != NUL && (wres[reslen - 1] == c
|
|
|
8825 || wres[reslen - 1] == *ws))
|
|
|
8826 reslen--;
|
|
|
8827 z0 = 1;
|
|
|
8828 z = 1;
|
|
|
8829 k0 = 0;
|
|
|
8830 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
8831 {
|
|
|
8832 word[i + k0] = *ws;
|
|
|
8833 k0++;
|
|
|
8834 ws++;
|
|
|
8835 }
|
|
|
8836 if (k > k0)
|
|
|
8837 mch_memmove(word + i + k0, word + i + k,
|
|
|
8838 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
8839
|
|
|
8840 /* new "actual letter" */
|
|
|
8841 c = word[i];
|
|
|
8842 }
|
|
|
8843 else
|
|
|
8844 {
|
|
|
8845 /* no '<' rule used */
|
|
|
8846 i += k - 1;
|
|
|
8847 z = 0;
|
|
|
8848 while (*ws != NUL && ws[1] != NUL && reslen < MAXWLEN)
|
|
|
8849 {
|
|
|
8850 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
8851 wres[reslen++] = *ws;
|
|
|
8852 ws++;
|
|
|
8853 }
|
|
|
8854 /* new "actual letter" */
|
|
|
8855 c = *ws;
|
|
|
8856 if (strstr((char *)s, "^^") != NULL)
|
|
|
8857 {
|
|
|
8858 if (c != NUL)
|
|
|
8859 wres[reslen++] = c;
|
|
|
8860 mch_memmove(word, word + i + 1,
|
|
|
8861 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
8862 i = 0;
|
|
|
8863 z0 = 1;
|
|
|
8864 }
|
|
|
8865 }
|
|
|
8866 break;
|
|
|
8867 }
|
|
|
8868 }
|
|
|
8869 }
|
|
|
8870 else if (vim_iswhite(c))
|
|
|
8871 {
|
|
|
8872 c = ' ';
|
|
|
8873 k = 1;
|
|
|
8874 }
|
|
|
8875
|
|
|
8876 if (z0 == 0)
|
|
|
8877 {
|
|
|
8878 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
8879 && (!slang->sl_collapse || reslen == 0
|
|
|
8880 || wres[reslen - 1] != c))
|
|
|
8881 /* condense only double letters */
|
|
|
8882 wres[reslen++] = c;
|
|
|
8883
|
|
|
8884 i++;
|
|
|
8885 z = 0;
|
|
|
8886 k = 0;
|
|
|
8887 }
|
|
|
8888 }
|
|
|
8889
|
|
|
8890 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
8891 l = 0;
|
|
|
8892 for (n = 0; n < reslen; ++n)
|
|
|
8893 {
|
|
|
8894 l += mb_char2bytes(wres[n], res + l);
|
|
|
8895 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
8896 break;
|
|
|
8897 }
|
|
|
8898 res[l] = NUL;
|
|
|
8899 }
|
|
|
8900 #endif
|
|
|
8901
|
|
324
|
8902 /*
|
|
344
|
8903 * Compute a score for two sound-a-like words.
|
|
|
8904 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
8905 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
8906 * avoiding checks that will not be possible.
|
|
|
8907 */
|
|
|
8908 static int
|
|
351
|
8909 soundalike_score(goodstart, badstart)
|
|
|
8910 char_u *goodstart; /* sound-folded good word */
|
|
|
8911 char_u *badstart; /* sound-folded bad word */
|
|
344
|
8912 {
|
|
351
|
8913 char_u *goodsound = goodstart;
|
|
|
8914 char_u *badsound = badstart;
|
|
|
8915 int goodlen;
|
|
|
8916 int badlen;
|
|
344
|
8917 int n;
|
|
|
8918 char_u *pl, *ps;
|
|
|
8919 char_u *pl2, *ps2;
|
|
351
|
8920 int score = 0;
|
|
|
8921
|
|
|
8922 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
8923 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
8924 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
8925 {
|
|
|
8926 score = SCORE_DEL / 2;
|
|
|
8927 if (*badsound == '*')
|
|
|
8928 ++badsound;
|
|
|
8929 else
|
|
|
8930 ++goodsound;
|
|
|
8931 }
|
|
|
8932
|
|
|
8933 goodlen = STRLEN(goodsound);
|
|
|
8934 badlen = STRLEN(badsound);
|
|
344
|
8935
|
|
|
8936 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
8937 * changes. */
|
|
|
8938 n = goodlen - badlen;
|
|
|
8939 if (n < -2 || n > 2)
|
|
|
8940 return SCORE_MAXMAX;
|
|
|
8941
|
|
|
8942 if (n > 0)
|
|
|
8943 {
|
|
351
|
8944 pl = goodsound; /* goodsound is longest */
|
|
344
|
8945 ps = badsound;
|
|
|
8946 }
|
|
|
8947 else
|
|
|
8948 {
|
|
351
|
8949 pl = badsound; /* badsound is longest */
|
|
344
|
8950 ps = goodsound;
|
|
|
8951 }
|
|
|
8952
|
|
|
8953 /* Skip over the identical part. */
|
|
|
8954 while (*pl == *ps && *pl != NUL)
|
|
|
8955 {
|
|
|
8956 ++pl;
|
|
|
8957 ++ps;
|
|
|
8958 }
|
|
|
8959
|
|
|
8960 switch (n)
|
|
|
8961 {
|
|
|
8962 case -2:
|
|
|
8963 case 2:
|
|
|
8964 /*
|
|
|
8965 * Must delete two characters from "pl".
|
|
|
8966 */
|
|
|
8967 ++pl; /* first delete */
|
|
|
8968 while (*pl == *ps)
|
|
|
8969 {
|
|
|
8970 ++pl;
|
|
|
8971 ++ps;
|
|
|
8972 }
|
|
|
8973 /* strings must be equal after second delete */
|
|
|
8974 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
8975 return score + SCORE_DEL * 2;
|
|
344
|
8976
|
|
|
8977 /* Failed to compare. */
|
|
|
8978 break;
|
|
|
8979
|
|
|
8980 case -1:
|
|
|
8981 case 1:
|
|
|
8982 /*
|
|
|
8983 * Minimal one delete from "pl" required.
|
|
|
8984 */
|
|
|
8985
|
|
|
8986 /* 1: delete */
|
|
|
8987 pl2 = pl + 1;
|
|
|
8988 ps2 = ps;
|
|
|
8989 while (*pl2 == *ps2)
|
|
|
8990 {
|
|
|
8991 if (*pl2 == NUL) /* reached the end */
|
|
351
|
8992 return score + SCORE_DEL;
|
|
344
|
8993 ++pl2;
|
|
|
8994 ++ps2;
|
|
|
8995 }
|
|
|
8996
|
|
|
8997 /* 2: delete then swap, then rest must be equal */
|
|
|
8998 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
8999 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9000 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
9001
|
|
|
9002 /* 3: delete then substitute, then the rest must be equal */
|
|
|
9003 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9004 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
9005
|
|
|
9006 /* 4: first swap then delete */
|
|
|
9007 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9008 {
|
|
|
9009 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9010 ps2 = ps + 2;
|
|
|
9011 while (*pl2 == *ps2)
|
|
|
9012 {
|
|
|
9013 ++pl2;
|
|
|
9014 ++ps2;
|
|
|
9015 }
|
|
|
9016 /* delete a char and then strings must be equal */
|
|
|
9017 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9018 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
9019 }
|
|
|
9020
|
|
|
9021 /* 5: first substitute then delete */
|
|
|
9022 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
9023 ps2 = ps + 1;
|
|
|
9024 while (*pl2 == *ps2)
|
|
|
9025 {
|
|
|
9026 ++pl2;
|
|
|
9027 ++ps2;
|
|
|
9028 }
|
|
|
9029 /* delete a char and then strings must be equal */
|
|
|
9030 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9031 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
9032
|
|
|
9033 /* Failed to compare. */
|
|
|
9034 break;
|
|
|
9035
|
|
|
9036 case 0:
|
|
|
9037 /*
|
|
|
9038 * Lenghts are equal, thus changes must result in same length: An
|
|
|
9039 * insert is only possible in combination with a delete.
|
|
|
9040 * 1: check if for identical strings
|
|
|
9041 */
|
|
|
9042 if (*pl == NUL)
|
|
351
|
9043 return score;
|
|
344
|
9044
|
|
|
9045 /* 2: swap */
|
|
|
9046 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9047 {
|
|
|
9048 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9049 ps2 = ps + 2;
|
|
|
9050 while (*pl2 == *ps2)
|
|
|
9051 {
|
|
|
9052 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9053 return score + SCORE_SWAP;
|
|
344
|
9054 ++pl2;
|
|
|
9055 ++ps2;
|
|
|
9056 }
|
|
|
9057 /* 3: swap and swap again */
|
|
|
9058 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9059 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9060 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
9061
|
|
|
9062 /* 4: swap and substitute */
|
|
|
9063 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9064 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
9065 }
|
|
|
9066
|
|
|
9067 /* 5: substitute */
|
|
|
9068 pl2 = pl + 1;
|
|
|
9069 ps2 = ps + 1;
|
|
|
9070 while (*pl2 == *ps2)
|
|
|
9071 {
|
|
|
9072 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9073 return score + SCORE_SUBST;
|
|
344
|
9074 ++pl2;
|
|
|
9075 ++ps2;
|
|
|
9076 }
|
|
|
9077
|
|
|
9078 /* 6: substitute and swap */
|
|
|
9079 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9080 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9081 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
9082
|
|
|
9083 /* 7: substitute and substitute */
|
|
|
9084 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9085 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
9086
|
|
|
9087 /* 8: insert then delete */
|
|
|
9088 pl2 = pl;
|
|
|
9089 ps2 = ps + 1;
|
|
|
9090 while (*pl2 == *ps2)
|
|
|
9091 {
|
|
|
9092 ++pl2;
|
|
|
9093 ++ps2;
|
|
|
9094 }
|
|
|
9095 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9096 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9097
|
|
|
9098 /* 9: delete then insert */
|
|
|
9099 pl2 = pl + 1;
|
|
|
9100 ps2 = ps;
|
|
|
9101 while (*pl2 == *ps2)
|
|
|
9102 {
|
|
|
9103 ++pl2;
|
|
|
9104 ++ps2;
|
|
|
9105 }
|
|
|
9106 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
9107 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9108
|
|
|
9109 /* Failed to compare. */
|
|
|
9110 break;
|
|
|
9111 }
|
|
|
9112
|
|
|
9113 return SCORE_MAXMAX;
|
|
|
9114 }
|
|
324
|
9115
|
|
323
|
9116 /*
|
|
|
9117 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
9118 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
9119 *
|
|
323
|
9120 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
9121 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
9122 * characters.
|
|
323
|
9123 */
|
|
|
9124 static int
|
|
|
9125 spell_edit_score(badword, goodword)
|
|
|
9126 char_u *badword;
|
|
|
9127 char_u *goodword;
|
|
|
9128 {
|
|
|
9129 int *cnt;
|
|
372
|
9130 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
9131 int j, i;
|
|
|
9132 int t;
|
|
|
9133 int bc, gc;
|
|
324
|
9134 int pbc, pgc;
|
|
|
9135 #ifdef FEAT_MBYTE
|
|
|
9136 char_u *p;
|
|
|
9137 int wbadword[MAXWLEN];
|
|
|
9138 int wgoodword[MAXWLEN];
|
|
|
9139
|
|
|
9140 if (has_mbyte)
|
|
|
9141 {
|
|
|
9142 /* Get the characters from the multi-byte strings and put them in an
|
|
|
9143 * int array for easy access. */
|
|
|
9144 for (p = badword, badlen = 0; *p != NUL; )
|
|
|
9145 wbadword[badlen++] = mb_ptr2char_adv(&p);
|
|
|
9146 ++badlen;
|
|
|
9147 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
|
9148 wgoodword[goodlen++] = mb_ptr2char_adv(&p);
|
|
|
9149 ++goodlen;
|
|
|
9150 }
|
|
|
9151 else
|
|
|
9152 #endif
|
|
|
9153 {
|
|
|
9154 badlen = STRLEN(badword) + 1;
|
|
|
9155 goodlen = STRLEN(goodword) + 1;
|
|
|
9156 }
|
|
323
|
9157
|
|
|
9158 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
9159 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
9160 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
9161 TRUE);
|
|
324
|
9162 if (cnt == NULL)
|
|
|
9163 return 0; /* out of memory */
|
|
323
|
9164
|
|
|
9165 CNT(0, 0) = 0;
|
|
|
9166 for (j = 1; j <= goodlen; ++j)
|
|
|
9167 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
9168
|
|
|
9169 for (i = 1; i <= badlen; ++i)
|
|
|
9170 {
|
|
|
9171 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
9172 for (j = 1; j <= goodlen; ++j)
|
|
|
9173 {
|
|
324
|
9174 #ifdef FEAT_MBYTE
|
|
|
9175 if (has_mbyte)
|
|
|
9176 {
|
|
|
9177 bc = wbadword[i - 1];
|
|
|
9178 gc = wgoodword[j - 1];
|
|
|
9179 }
|
|
|
9180 else
|
|
|
9181 #endif
|
|
|
9182 {
|
|
|
9183 bc = badword[i - 1];
|
|
|
9184 gc = goodword[j - 1];
|
|
|
9185 }
|
|
323
|
9186 if (bc == gc)
|
|
|
9187 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
9188 else
|
|
|
9189 {
|
|
|
9190 /* Use a better score when there is only a case difference. */
|
|
324
|
9191 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
9192 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
9193 else
|
|
|
9194 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
9195
|
|
324
|
9196 if (i > 1 && j > 1)
|
|
323
|
9197 {
|
|
324
|
9198 #ifdef FEAT_MBYTE
|
|
|
9199 if (has_mbyte)
|
|
|
9200 {
|
|
|
9201 pbc = wbadword[i - 2];
|
|
|
9202 pgc = wgoodword[j - 2];
|
|
|
9203 }
|
|
|
9204 else
|
|
|
9205 #endif
|
|
|
9206 {
|
|
|
9207 pbc = badword[i - 2];
|
|
|
9208 pgc = goodword[j - 2];
|
|
|
9209 }
|
|
|
9210 if (bc == pgc && pbc == gc)
|
|
|
9211 {
|
|
|
9212 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
9213 if (t < CNT(i, j))
|
|
|
9214 CNT(i, j) = t;
|
|
|
9215 }
|
|
323
|
9216 }
|
|
|
9217 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
9218 if (t < CNT(i, j))
|
|
|
9219 CNT(i, j) = t;
|
|
|
9220 t = SCORE_INS + CNT(i, j - 1);
|
|
|
9221 if (t < CNT(i, j))
|
|
|
9222 CNT(i, j) = t;
|
|
|
9223 }
|
|
|
9224 }
|
|
|
9225 }
|
|
344
|
9226
|
|
|
9227 i = CNT(badlen - 1, goodlen - 1);
|
|
|
9228 vim_free(cnt);
|
|
|
9229 return i;
|
|
323
|
9230 }
|
|
307
|
9231
|
|
351
|
9232 /*
|
|
|
9233 * ":spelldump"
|
|
|
9234 */
|
|
|
9235 /*ARGSUSED*/
|
|
|
9236 void
|
|
|
9237 ex_spelldump(eap)
|
|
|
9238 exarg_T *eap;
|
|
|
9239 {
|
|
|
9240 buf_T *buf = curbuf;
|
|
|
9241 langp_T *lp;
|
|
|
9242 slang_T *slang;
|
|
|
9243 idx_T arridx[MAXWLEN];
|
|
|
9244 int curi[MAXWLEN];
|
|
|
9245 char_u word[MAXWLEN];
|
|
|
9246 int c;
|
|
|
9247 char_u *byts;
|
|
|
9248 idx_T *idxs;
|
|
|
9249 linenr_T lnum = 0;
|
|
|
9250 int round;
|
|
|
9251 int depth;
|
|
|
9252 int n;
|
|
|
9253 int flags;
|
|
|
9254
|
|
|
9255 if (no_spell_checking())
|
|
|
9256 return;
|
|
|
9257
|
|
|
9258 /* Create a new empty buffer by splitting the window. */
|
|
|
9259 do_cmdline_cmd((char_u *)"new");
|
|
|
9260 if (!bufempty() || !buf_valid(buf))
|
|
|
9261 return;
|
|
|
9262
|
|
|
9263 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
9264 {
|
|
|
9265 slang = lp->lp_slang;
|
|
|
9266
|
|
|
9267 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
9268 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
9269
|
|
|
9270 /* round 1: case-folded tree
|
|
|
9271 * round 2: keep-case tree */
|
|
|
9272 for (round = 1; round <= 2; ++round)
|
|
|
9273 {
|
|
|
9274 if (round == 1)
|
|
|
9275 {
|
|
|
9276 byts = slang->sl_fbyts;
|
|
|
9277 idxs = slang->sl_fidxs;
|
|
|
9278 }
|
|
|
9279 else
|
|
|
9280 {
|
|
|
9281 byts = slang->sl_kbyts;
|
|
|
9282 idxs = slang->sl_kidxs;
|
|
|
9283 }
|
|
|
9284 if (byts == NULL)
|
|
|
9285 continue; /* array is empty */
|
|
|
9286
|
|
|
9287 depth = 0;
|
|
|
9288 arridx[0] = 0;
|
|
|
9289 curi[0] = 1;
|
|
|
9290 while (depth >= 0 && !got_int)
|
|
|
9291 {
|
|
|
9292 if (curi[depth] > byts[arridx[depth]])
|
|
|
9293 {
|
|
|
9294 /* Done all bytes at this node, go up one level. */
|
|
|
9295 --depth;
|
|
|
9296 line_breakcheck();
|
|
|
9297 }
|
|
|
9298 else
|
|
|
9299 {
|
|
|
9300 /* Do one more byte at this node. */
|
|
|
9301 n = arridx[depth] + curi[depth];
|
|
|
9302 ++curi[depth];
|
|
|
9303 c = byts[n];
|
|
|
9304 if (c == 0)
|
|
|
9305 {
|
|
|
9306 /* End of word, deal with the word.
|
|
|
9307 * Don't use keep-case words in the fold-case tree,
|
|
|
9308 * they will appear in the keep-case tree.
|
|
|
9309 * Only use the word when the region matches. */
|
|
|
9310 flags = (int)idxs[n];
|
|
|
9311 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
9312 && ((flags & WF_REGION) == 0
|
|
|
9313 || (((unsigned)flags >> 8)
|
|
|
9314 & lp->lp_region) != 0))
|
|
|
9315 {
|
|
|
9316 word[depth] = NUL;
|
|
355
|
9317
|
|
|
9318 /* Dump the basic word if there is no prefix or
|
|
|
9319 * when it's the first one. */
|
|
|
9320 c = (unsigned)flags >> 16;
|
|
|
9321 if (c == 0 || curi[depth] == 2)
|
|
|
9322 dump_word(word, round, flags, lnum++);
|
|
351
|
9323
|
|
|
9324 /* Apply the prefix, if there is one. */
|
|
355
|
9325 if (c != 0)
|
|
351
|
9326 lnum = apply_prefixes(slang, word, round,
|
|
|
9327 flags, lnum);
|
|
|
9328 }
|
|
|
9329 }
|
|
|
9330 else
|
|
|
9331 {
|
|
|
9332 /* Normal char, go one level deeper. */
|
|
|
9333 word[depth++] = c;
|
|
|
9334 arridx[depth] = idxs[n];
|
|
|
9335 curi[depth] = 1;
|
|
|
9336 }
|
|
|
9337 }
|
|
|
9338 }
|
|
|
9339 }
|
|
|
9340 }
|
|
|
9341
|
|
|
9342 /* Delete the empty line that we started with. */
|
|
|
9343 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
9344 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
9345
|
|
|
9346 redraw_later(NOT_VALID);
|
|
|
9347 }
|
|
|
9348
|
|
|
9349 /*
|
|
|
9350 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
9351 */
|
|
|
9352 static void
|
|
|
9353 dump_word(word, round, flags, lnum)
|
|
|
9354 char_u *word;
|
|
|
9355 int round;
|
|
|
9356 int flags;
|
|
|
9357 linenr_T lnum;
|
|
|
9358 {
|
|
|
9359 int keepcap = FALSE;
|
|
|
9360 char_u *p;
|
|
|
9361 char_u cword[MAXWLEN];
|
|
|
9362 char_u badword[MAXWLEN + 3];
|
|
|
9363
|
|
|
9364 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
9365 {
|
|
|
9366 /* Need to fix case according to "flags". */
|
|
|
9367 make_case_word(word, cword, flags);
|
|
|
9368 p = cword;
|
|
|
9369 }
|
|
|
9370 else
|
|
|
9371 {
|
|
|
9372 p = word;
|
|
|
9373 if (round == 2 && (captype(word, NULL) & WF_KEEPCAP) == 0)
|
|
|
9374 keepcap = TRUE;
|
|
|
9375 }
|
|
|
9376
|
|
|
9377 /* Bad word is preceded by "/!" and some other
|
|
|
9378 * flags. */
|
|
|
9379 if ((flags & (WF_BANNED | WF_RARE)) || keepcap)
|
|
|
9380 {
|
|
|
9381 STRCPY(badword, "/");
|
|
|
9382 if (keepcap)
|
|
|
9383 STRCAT(badword, "=");
|
|
|
9384 if (flags & WF_BANNED)
|
|
|
9385 STRCAT(badword, "!");
|
|
|
9386 else if (flags & WF_RARE)
|
|
|
9387 STRCAT(badword, "?");
|
|
|
9388 STRCAT(badword, p);
|
|
|
9389 p = badword;
|
|
|
9390 }
|
|
|
9391
|
|
|
9392 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
9393 }
|
|
|
9394
|
|
|
9395 /*
|
|
372
|
9396 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
9397 * "word" and append a line to the buffer.
|
|
351
|
9398 * Return the updated line number.
|
|
|
9399 */
|
|
|
9400 static linenr_T
|
|
|
9401 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
9402 slang_T *slang;
|
|
|
9403 char_u *word; /* case-folded word */
|
|
|
9404 int round;
|
|
|
9405 int flags; /* flags with prefix ID */
|
|
|
9406 linenr_T startlnum;
|
|
|
9407 {
|
|
|
9408 idx_T arridx[MAXWLEN];
|
|
|
9409 int curi[MAXWLEN];
|
|
|
9410 char_u prefix[MAXWLEN];
|
|
|
9411 int c;
|
|
|
9412 char_u *byts;
|
|
|
9413 idx_T *idxs;
|
|
|
9414 linenr_T lnum = startlnum;
|
|
|
9415 int depth;
|
|
|
9416 int n;
|
|
|
9417 int len;
|
|
|
9418 int prefid = (unsigned)flags >> 16;
|
|
|
9419 int i;
|
|
|
9420
|
|
|
9421 byts = slang->sl_pbyts;
|
|
|
9422 idxs = slang->sl_pidxs;
|
|
|
9423 if (byts != NULL) /* array not is empty */
|
|
|
9424 {
|
|
|
9425 /*
|
|
|
9426 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
|
9427 * When at the end of a prefix check that it supports "prefid".
|
|
|
9428 */
|
|
|
9429 depth = 0;
|
|
|
9430 arridx[0] = 0;
|
|
|
9431 curi[0] = 1;
|
|
|
9432 while (depth >= 0 && !got_int)
|
|
|
9433 {
|
|
|
9434 len = arridx[depth];
|
|
|
9435 if (curi[depth] > byts[len])
|
|
|
9436 {
|
|
|
9437 /* Done all bytes at this node, go up one level. */
|
|
|
9438 --depth;
|
|
|
9439 line_breakcheck();
|
|
|
9440 }
|
|
|
9441 else
|
|
|
9442 {
|
|
|
9443 /* Do one more byte at this node. */
|
|
|
9444 n = len + curi[depth];
|
|
|
9445 ++curi[depth];
|
|
|
9446 c = byts[n];
|
|
|
9447 if (c == 0)
|
|
|
9448 {
|
|
|
9449 /* End of prefix, find out how many IDs there are. */
|
|
|
9450 for (i = 1; i < len; ++i)
|
|
|
9451 if (byts[n + i] != 0)
|
|
|
9452 break;
|
|
|
9453 curi[depth] += i - 1;
|
|
|
9454
|
|
366
|
9455 i = valid_word_prefix(i, n, prefid, word, slang);
|
|
|
9456 if (i != 0)
|
|
351
|
9457 {
|
|
|
9458 vim_strncpy(prefix + depth, word, MAXWLEN - depth);
|
|
366
|
9459 dump_word(prefix, round,
|
|
|
9460 (i & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
9461 : flags, lnum++);
|
|
351
|
9462 }
|
|
|
9463 }
|
|
|
9464 else
|
|
|
9465 {
|
|
|
9466 /* Normal char, go one level deeper. */
|
|
|
9467 prefix[depth++] = c;
|
|
|
9468 arridx[depth] = idxs[n];
|
|
|
9469 curi[depth] = 1;
|
|
|
9470 }
|
|
|
9471 }
|
|
|
9472 }
|
|
|
9473 }
|
|
|
9474
|
|
|
9475 return lnum;
|
|
|
9476 }
|
|
|
9477
|
|
236
|
9478 #endif /* FEAT_SYN_HL */
|
