|
223
|
1 /* vi:set ts=8 sts=4 sw=4:
|
|
|
2 *
|
|
|
3 * VIM - Vi IMproved by Bram Moolenaar
|
|
|
4 *
|
|
|
5 * Do ":help uganda" in Vim to read copying and usage conditions.
|
|
|
6 * Do ":help credits" in Vim to see a list of people who contributed.
|
|
|
7 * See README.txt for an overview of the Vim source code.
|
|
|
8 */
|
|
|
9
|
|
|
10 /*
|
|
|
11 * spell.c: code for spell checking
|
|
226
|
12 *
|
|
300
|
13 * The spell checking mechanism uses a tree (aka trie). Each node in the tree
|
|
|
14 * has a list of bytes that can appear (siblings). For each byte there is a
|
|
|
15 * pointer to the node with the byte that follows in the word (child).
|
|
324
|
16 *
|
|
|
17 * A NUL byte is used where the word may end. The bytes are sorted, so that
|
|
|
18 * binary searching can be used and the NUL bytes are at the start. The
|
|
|
19 * number of possible bytes is stored before the list of bytes.
|
|
|
20 *
|
|
|
21 * The tree uses two arrays: "byts" stores the characters, "idxs" stores
|
|
|
22 * either the next index or flags. The tree starts at index 0. For example,
|
|
|
23 * to lookup "vi" this sequence is followed:
|
|
|
24 * i = 0
|
|
|
25 * len = byts[i]
|
|
|
26 * n = where "v" appears in byts[i + 1] to byts[i + len]
|
|
|
27 * i = idxs[n]
|
|
|
28 * len = byts[i]
|
|
|
29 * n = where "i" appears in byts[i + 1] to byts[i + len]
|
|
|
30 * i = idxs[n]
|
|
|
31 * len = byts[i]
|
|
|
32 * find that byts[i + 1] is 0, idxs[i + 1] has flags for "vi".
|
|
300
|
33 *
|
|
339
|
34 * There are two word trees: one with case-folded words and one with words in
|
|
300
|
35 * original case. The second one is only used for keep-case words and is
|
|
|
36 * usually small.
|
|
|
37 *
|
|
339
|
38 * There is one additional tree for when prefixes are not applied when
|
|
|
39 * generating the .spl file. This tree stores all the possible prefixes, as
|
|
|
40 * if they were words. At each word (prefix) end the prefix nr is stored, the
|
|
|
41 * following word must support this prefix nr. And the condition nr is
|
|
|
42 * stored, used to lookup the condition that the word must match with.
|
|
|
43 *
|
|
300
|
44 * Thanks to Olaf Seibert for providing an example implementation of this tree
|
|
|
45 * and the compression mechanism.
|
|
243
|
46 *
|
|
|
47 * Matching involves checking the caps type: Onecap ALLCAP KeepCap.
|
|
|
48 *
|
|
236
|
49 * Why doesn't Vim use aspell/ispell/myspell/etc.?
|
|
|
50 * See ":help develop-spell".
|
|
|
51 */
|
|
|
52
|
|
300
|
53 /*
|
|
324
|
54 * Use this to adjust the score after finding suggestions, based on the
|
|
|
55 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
56 * it for every possible suggestion.
|
|
|
57 * Disadvantage: When "the" is typed as "hte" it sounds different and goes
|
|
|
58 * down in the list.
|
|
344
|
59 * Used when 'spellsuggest' is set to "best".
|
|
|
60 */
|
|
|
61 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
62
|
|
|
63 /*
|
|
|
64 * The double scoring mechanism is based on the principle that there are two
|
|
|
65 * kinds of spelling mistakes:
|
|
|
66 * 1. You know how to spell the word, but mistype something. This results in
|
|
|
67 * a small editing distance (character swapped/omitted/inserted) and
|
|
|
68 * possibly a word that sounds completely different.
|
|
|
69 * 2. You don't know how to spell the word and type something that sounds
|
|
|
70 * right. The edit distance can be big but the word is similar after
|
|
|
71 * sound-folding.
|
|
|
72 * Since scores for these two mistakes will be very different we use a list
|
|
|
73 * for each.
|
|
|
74 * The sound-folding is slow, only do double scoring when 'spellsuggest' is
|
|
|
75 * "double".
|
|
323
|
76 */
|
|
|
77
|
|
|
78 /*
|
|
339
|
79 * Vim spell file format: <HEADER>
|
|
|
80 * <SUGGEST>
|
|
|
81 * <LWORDTREE>
|
|
|
82 * <KWORDTREE>
|
|
|
83 * <PREFIXTREE>
|
|
300
|
84 *
|
|
339
|
85 * <HEADER>: <fileID>
|
|
|
86 * <regioncnt> <regionname> ...
|
|
|
87 * <charflagslen> <charflags>
|
|
|
88 * <fcharslen> <fchars>
|
|
366
|
89 * <midwordlen> <midword>
|
|
339
|
90 * <prefcondcnt> <prefcond> ...
|
|
300
|
91 *
|
|
366
|
92 * <fileID> 10 bytes "VIMspell08"
|
|
300
|
93 * <regioncnt> 1 byte number of regions following (8 supported)
|
|
307
|
94 * <regionname> 2 bytes Region name: ca, au, etc. Lower case.
|
|
300
|
95 * First <regionname> is region 1.
|
|
|
96 *
|
|
|
97 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
98 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
99 * 0x01 word character CF_WORD
|
|
|
100 * 0x02 upper-case character CF_UPPER
|
|
300
|
101 * <fcharslen> 2 bytes Number of bytes in <fchars>.
|
|
|
102 * <fchars> N bytes Folded characters, first one is for character 128.
|
|
|
103 *
|
|
366
|
104 * <midwordlen> 2 bytes Number of bytes in <midword>.
|
|
|
105 * <midword> N bytes Characters that are word characters only when used
|
|
|
106 * in the middle of a word.
|
|
|
107 *
|
|
339
|
108 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
109 *
|
|
|
110 * <prefcond> : <condlen> <condstr>
|
|
|
111 *
|
|
|
112 * <condlen> 1 byte Length of <condstr>.
|
|
|
113 *
|
|
|
114 * <condstr> N bytes Condition for the prefix.
|
|
|
115 *
|
|
300
|
116 *
|
|
323
|
117 * <SUGGEST> : <repcount> <rep> ...
|
|
|
118 * <salflags> <salcount> <sal> ...
|
|
|
119 * <maplen> <mapstr>
|
|
|
120 *
|
|
|
121 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
|
122 *
|
|
|
123 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
|
124 *
|
|
|
125 * <repfromlen> 1 byte length of <repfrom>
|
|
|
126 *
|
|
|
127 * <repfrom> N bytes "from" part of replacement
|
|
|
128 *
|
|
|
129 * <reptolen> 1 byte length of <repto>
|
|
|
130 *
|
|
|
131 * <repto> N bytes "to" part of replacement
|
|
300
|
132 *
|
|
323
|
133 * <salflags> 1 byte flags for soundsalike conversion:
|
|
|
134 * SAL_F0LLOWUP
|
|
|
135 * SAL_COLLAPSE
|
|
|
136 * SAL_REM_ACCENTS
|
|
375
|
137 * SAL_SOFO: SOFOFROM and SOFOTO used instead of SAL
|
|
|
138 *
|
|
|
139 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
140 *
|
|
|
141 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
|
142 *
|
|
375
|
143 * <salfromlen> 1-2 bytes length of <salfrom> (2 bytes for SAL_SOFO)
|
|
323
|
144 *
|
|
|
145 * <salfrom> N bytes "from" part of soundsalike
|
|
|
146 *
|
|
375
|
147 * <saltolen> 1-2 bytes length of <salto> (2 bytes for SAL_SOFO)
|
|
323
|
148 *
|
|
|
149 * <salto> N bytes "to" part of soundsalike
|
|
|
150 *
|
|
|
151 * <maplen> 2 bytes length of <mapstr>, MSB first
|
|
|
152 *
|
|
|
153 * <mapstr> N bytes String with sequences of similar characters,
|
|
|
154 * separated by slashes.
|
|
300
|
155 *
|
|
|
156 *
|
|
|
157 * <LWORDTREE>: <wordtree>
|
|
|
158 *
|
|
339
|
159 * <KWORDTREE>: <wordtree>
|
|
|
160 *
|
|
|
161 * <PREFIXTREE>: <wordtree>
|
|
|
162 *
|
|
|
163 *
|
|
300
|
164 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
165 *
|
|
|
166 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
167 *
|
|
|
168 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
169 *
|
|
|
170 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
171 * follow in sorted order.
|
|
|
172 *
|
|
339
|
173 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
|
174 * | <flags> [<region>] [<prefixID>]
|
|
|
175 * | <prefixID> <prefcondnr> ]
|
|
300
|
176 *
|
|
|
177 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
178 * BY_NOFLAGS: End of word without flags and for all
|
|
|
179 * regions.
|
|
366
|
180 * For PREFIXTREE <prefixID> and
|
|
|
181 * <prefcondnr> follow.
|
|
|
182 * BY_FLAGS: End of word, <flags> follow.
|
|
|
183 * For PREFIXTREE <prefixID> and
|
|
|
184 * <prefcondnr> follow for rare prefix.
|
|
300
|
185 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
|
186 * and <xbyte> follow.
|
|
|
187 *
|
|
|
188 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
189 *
|
|
|
190 * <xbyte> 1 byte byte value of the sibling.
|
|
|
191 *
|
|
|
192 * <flags> 1 byte bitmask of:
|
|
|
193 * WF_ALLCAP word must have only capitals
|
|
|
194 * WF_ONECAP first char of word must be capital
|
|
|
195 * WF_RARE rare word
|
|
|
196 * WF_REGION <region> follows
|
|
339
|
197 * WF_PFX <prefixID> follows
|
|
300
|
198 *
|
|
|
199 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
200 * omitted it's valid in all regions.
|
|
|
201 * Lowest bit is for region 1.
|
|
|
202 *
|
|
339
|
203 * <prefixID> 1 byte ID of prefix that can be used with this word. For
|
|
|
204 * PREFIXTREE used for the required prefix ID.
|
|
|
205 *
|
|
|
206 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
207 * from HEADER.
|
|
300
|
208 *
|
|
|
209 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
210 */
|
|
|
211
|
|
223
|
212 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
213 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
214 #endif
|
|
|
215
|
|
|
216 #include "vim.h"
|
|
|
217
|
|
|
218 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
219
|
|
|
220 #ifdef HAVE_FCNTL_H
|
|
|
221 # include <fcntl.h>
|
|
|
222 #endif
|
|
|
223
|
|
323
|
224 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
225 Some places assume a word length fits in a
|
|
|
226 byte, thus it can't be above 255. */
|
|
226
|
227
|
|
324
|
228 /* Type used for indexes in the word tree need to be at least 3 bytes. If int
|
|
|
229 * is 8 bytes we could use something smaller, but what? */
|
|
|
230 #if SIZEOF_INT > 2
|
|
|
231 typedef int idx_T;
|
|
|
232 #else
|
|
|
233 typedef long idx_T;
|
|
|
234 #endif
|
|
|
235
|
|
|
236 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
237 * comes above it. */
|
|
300
|
238 #define WF_REGION 0x01 /* region byte follows */
|
|
|
239 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
240 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
241 #define WF_RARE 0x08 /* rare word */
|
|
307
|
242 #define WF_BANNED 0x10 /* bad word */
|
|
339
|
243 #define WF_PFX 0x20 /* prefix ID list follows */
|
|
323
|
244 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
245
|
|
|
246 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP)
|
|
300
|
247
|
|
366
|
248 #define WF_RAREPFX 0x1000000 /* in sl_pidxs: flag for rare postponed
|
|
|
249 prefix; must be above prefixID (one byte)
|
|
|
250 and prefcondnr (two bytes) */
|
|
|
251
|
|
300
|
252 #define BY_NOFLAGS 0 /* end of word without flags or region */
|
|
|
253 #define BY_FLAGS 1 /* end of word, flag byte follows */
|
|
|
254 #define BY_INDEX 2 /* child is shared, index follows */
|
|
|
255 #define BY_SPECIAL BY_INDEX /* hightest special byte value */
|
|
236
|
256
|
|
323
|
257 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
258 * and si_sal. Not for sl_sal!
|
|
323
|
259 * One replacement: from "ft_from" to "ft_to". */
|
|
|
260 typedef struct fromto_S
|
|
236
|
261 {
|
|
323
|
262 char_u *ft_from;
|
|
|
263 char_u *ft_to;
|
|
|
264 } fromto_T;
|
|
236
|
265
|
|
344
|
266 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
267 * The info is split for quick processing by spell_soundfold().
|
|
|
268 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
269 typedef struct salitem_S
|
|
|
270 {
|
|
|
271 char_u *sm_lead; /* leading letters */
|
|
|
272 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
273 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
274 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
275 char_u *sm_to; /* replacement. */
|
|
372
|
276 #ifdef FEAT_MBYTE
|
|
|
277 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
278 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
279 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
280 #endif
|
|
344
|
281 } salitem_T;
|
|
|
282
|
|
375
|
283 #ifdef FEAT_MBYTE
|
|
|
284 typedef int salfirst_T;
|
|
|
285 #else
|
|
|
286 typedef short salfirst_T;
|
|
|
287 #endif
|
|
|
288
|
|
236
|
289 /*
|
|
243
|
290 * Structure used to store words and other info for one language, loaded from
|
|
|
291 * a .spl file.
|
|
300
|
292 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
293 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
294 *
|
|
|
295 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
296 * the number of possible bytes, sorted on byte value:
|
|
|
297 * <len> <byte1> <byte2> ...
|
|
|
298 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
299 * byte in "byts".
|
|
|
300 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
|
301 * the flags and region for the word. There may be several zeros in sequence
|
|
|
302 * for alternative flag/region combinations.
|
|
236
|
303 */
|
|
|
304 typedef struct slang_S slang_T;
|
|
|
305 struct slang_S
|
|
|
306 {
|
|
|
307 slang_T *sl_next; /* next language */
|
|
|
308 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
309 char_u *sl_fname; /* name of .spl file */
|
|
323
|
310 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
311
|
|
300
|
312 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
313 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
314 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
315 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
316 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
317 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
318
|
|
236
|
319 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
320
|
|
376
|
321 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
322
|
|
339
|
323 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
324 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
325
|
|
323
|
326 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
327 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
328 there is none */
|
|
344
|
329 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
330 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
331 there is none */
|
|
375
|
332 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
333 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
334 * "sl_sal" is a list of wide char lists. */
|
|
323
|
335 int sl_followup; /* SAL followup */
|
|
|
336 int sl_collapse; /* SAL collapse_result */
|
|
|
337 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
338 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
339 #ifdef FEAT_MBYTE
|
|
|
340 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
341 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
342 #else
|
|
|
343 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
344 #endif
|
|
236
|
345 };
|
|
|
346
|
|
243
|
347 /* First language that is loaded, start of the linked list of loaded
|
|
|
348 * languages. */
|
|
236
|
349 static slang_T *first_lang = NULL;
|
|
|
350
|
|
323
|
351 /* Flags used in .spl file for soundsalike flags. */
|
|
|
352 #define SAL_F0LLOWUP 1
|
|
|
353 #define SAL_COLLAPSE 2
|
|
|
354 #define SAL_REM_ACCENTS 4
|
|
375
|
355 #define SAL_SOFO 8 /* SOFOFROM and SOFOTO instead of SAL */
|
|
323
|
356
|
|
236
|
357 /*
|
|
|
358 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
359 */
|
|
|
360 typedef struct langp_S
|
|
|
361 {
|
|
|
362 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
363 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
364 } langp_T;
|
|
|
365
|
|
|
366 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
367
|
|
307
|
368 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
369
|
|
|
370 /* Result values. Lower number is accepted over higher one. */
|
|
|
371 #define SP_BANNED -1
|
|
236
|
372 #define SP_OK 0
|
|
307
|
373 #define SP_RARE 1
|
|
|
374 #define SP_LOCAL 2
|
|
|
375 #define SP_BAD 3
|
|
236
|
376
|
|
366
|
377 #define VIMSPELLMAGIC "VIMspell08" /* string at start of Vim spell file */
|
|
236
|
378 #define VIMSPELLMAGICL 10
|
|
|
379
|
|
381
|
380 /* file used for "zG" and "zW" */
|
|
385
|
381 static char_u *int_wordlist = NULL;
|
|
381
|
382
|
|
236
|
383 /*
|
|
323
|
384 * Information used when looking for suggestions.
|
|
|
385 */
|
|
|
386 typedef struct suginfo_S
|
|
|
387 {
|
|
|
388 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
389 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
390 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
391 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
392 char_u *su_badptr; /* start of bad word in line */
|
|
|
393 int su_badlen; /* length of detected bad word in line */
|
|
346
|
394 int su_badflags; /* caps flags for bad word */
|
|
323
|
395 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
396 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
397 hashtab_T su_banned; /* table with banned words */
|
|
|
398 } suginfo_T;
|
|
|
399
|
|
|
400 /* One word suggestion. Used in "si_ga". */
|
|
|
401 typedef struct suggest_S
|
|
|
402 {
|
|
|
403 char_u *st_word; /* suggested word, allocated string */
|
|
|
404 int st_orglen; /* length of replaced text */
|
|
|
405 int st_score; /* lower is better */
|
|
344
|
406 int st_altscore; /* used when st_score compares equal */
|
|
|
407 int st_salscore; /* st_score is for soundalike */
|
|
324
|
408 int st_had_bonus; /* bonus already included in score */
|
|
323
|
409 } suggest_T;
|
|
|
410
|
|
344
|
411 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
412
|
|
324
|
413 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
414 * called the score may change, thus we need to keep more than what is
|
|
|
415 * displayed. */
|
|
351
|
416 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
417
|
|
|
418 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
419 * of suggestions that are not going to be displayed. */
|
|
344
|
420 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
421
|
|
|
422 /* score for various changes */
|
|
344
|
423 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
424 #define SCORE_ICASE 52 /* slightly different case */
|
|
|
425 #define SCORE_REGION 70 /* word is for different region */
|
|
|
426 #define SCORE_RARE 180 /* rare word */
|
|
|
427 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
428 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
429 #define SCORE_REP 87 /* REP replacement */
|
|
|
430 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
431 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
324
|
432 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
433 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
324
|
434 #define SCORE_INS 96 /* insert a character */
|
|
358
|
435 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
366
|
436 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
437
|
|
372
|
438 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
439 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
440 * 350 allows for about three changes. */
|
|
344
|
441
|
|
|
442 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
443 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
444
|
|
|
445 /*
|
|
236
|
446 * Structure to store info for word matching.
|
|
|
447 */
|
|
|
448 typedef struct matchinf_S
|
|
|
449 {
|
|
|
450 langp_T *mi_lp; /* info for language and region */
|
|
243
|
451
|
|
|
452 /* pointers to original text to be checked */
|
|
236
|
453 char_u *mi_word; /* start of word being checked */
|
|
339
|
454 char_u *mi_end; /* end of matching word so far */
|
|
243
|
455 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
456 char_u *mi_cend; /* char after what was used for
|
|
|
457 mi_capflags */
|
|
243
|
458
|
|
|
459 /* case-folded text */
|
|
|
460 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
461 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
462
|
|
339
|
463 /* for when checking word after a prefix */
|
|
|
464 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
|
465 prefixID/condition */
|
|
|
466 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
467 int mi_prefixlen; /* byte length of prefix */
|
|
|
468
|
|
243
|
469 /* others */
|
|
236
|
470 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
471 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
472 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
473 } matchinf_T;
|
|
|
474
|
|
307
|
475 /*
|
|
|
476 * The tables used for recognizing word characters according to spelling.
|
|
|
477 * These are only used for the first 256 characters of 'encoding'.
|
|
|
478 */
|
|
|
479 typedef struct spelltab_S
|
|
|
480 {
|
|
|
481 char_u st_isw[256]; /* flags: is word char */
|
|
|
482 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
483 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
484 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
485 } spelltab_T;
|
|
|
486
|
|
|
487 static spelltab_T spelltab;
|
|
|
488 static int did_set_spelltab;
|
|
|
489
|
|
324
|
490 #define CF_WORD 0x01
|
|
|
491 #define CF_UPPER 0x02
|
|
307
|
492
|
|
|
493 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
494 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
495 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
496 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
497 #ifdef FEAT_MBYTE
|
|
|
498 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
499 #endif
|
|
339
|
500 static void write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
501
|
|
|
502 /*
|
|
358
|
503 * Return TRUE if "p" points to a word character. Like spell_iswordp() but
|
|
|
504 * without the special handling of a single quote.
|
|
307
|
505 * Checking for a word character is done very often, avoid the function call
|
|
|
506 * overhead.
|
|
|
507 */
|
|
|
508 #ifdef FEAT_MBYTE
|
|
|
509 # define SPELL_ISWORDP(p) ((has_mbyte && MB_BYTE2LEN(*(p)) > 1) \
|
|
|
510 ? (mb_get_class(p) >= 2) : spelltab.st_isw[*(p)])
|
|
|
511 #else
|
|
|
512 # define SPELL_ISWORDP(p) (spelltab.st_isw[*(p)])
|
|
|
513 #endif
|
|
|
514
|
|
323
|
515 /*
|
|
344
|
516 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
517 */
|
|
|
518 typedef enum
|
|
|
519 {
|
|
344
|
520 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
521 * ends); if badword ends there is a match, otherwise
|
|
|
522 * try splitting word. */
|
|
375
|
523 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
524 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
525 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
526 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
527 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
528 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
529 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
530 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
531 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
532 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
533 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
534 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
535 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
536 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
537 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
538 STATE_FINAL /* End of this node. */
|
|
|
539 } state_T;
|
|
|
540
|
|
|
541 /*
|
|
346
|
542 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
543 */
|
|
|
544 typedef struct trystate_S
|
|
|
545 {
|
|
330
|
546 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
547 int ts_score; /* score */
|
|
344
|
548 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
549 short ts_curi; /* index in list of child nodes */
|
|
|
550 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
551 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
552 char_u ts_twordlen; /* valid length of tword[] */
|
|
375
|
553 char_u ts_prefixdepth; /* stack depth for end of prefix or PREFIXTREE
|
|
|
554 * or NOPREFIX */
|
|
330
|
555 #ifdef FEAT_MBYTE
|
|
|
556 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
557 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
558 char_u ts_isdiff; /* DIFF_ values */
|
|
|
559 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
560 #endif
|
|
323
|
561 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
330
|
562 char_u ts_save_splitoff; /* su_splitoff saved here */
|
|
346
|
563 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
564 } trystate_T;
|
|
|
565
|
|
330
|
566 /* values for ts_isdiff */
|
|
|
567 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
568 #define DIFF_YES 1 /* different byte found */
|
|
|
569 #define DIFF_INSERT 2 /* inserting character */
|
|
|
570
|
|
375
|
571 /* special values ts_prefixdepth */
|
|
|
572 #define PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
573 #define NOPREFIX 0xff /* not using prefixes */
|
|
|
574
|
|
339
|
575 /* mode values for find_word */
|
|
|
576 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
577 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
578 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
579
|
|
236
|
580 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
581 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
582 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
583 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
351
|
584 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int prefid, char_u *word, slang_T *slang));
|
|
339
|
585 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
586 static int fold_more __ARGS((matchinf_T *mip));
|
|
323
|
587 static int spell_valid_case __ARGS((int origflags, int treeflags));
|
|
351
|
588 static int no_spell_checking __ARGS((void));
|
|
307
|
589 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
590 static char_u *spell_enc __ARGS((void));
|
|
385
|
591 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
592 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
593 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
376
|
594 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *errp));
|
|
381
|
595 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
596 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
597 #ifdef FEAT_MBYTE
|
|
|
598 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
599 #endif
|
|
339
|
600 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
601 static void clear_midword __ARGS((buf_T *buf));
|
|
|
602 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
603 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
604 static int captype __ARGS((char_u *word, char_u *end));
|
|
323
|
605 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
381
|
606 static int set_spell_charflags __ARGS((char_u *flags, char_u *upp));
|
|
307
|
607 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
608 static void write_spell_chartab __ARGS((FILE *fd));
|
|
|
609 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
388
|
610 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
611 #ifdef FEAT_EVAL
|
|
|
612 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
613 #endif
|
|
|
614 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
615 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
616 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
617 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
618 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
619 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
620 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
621 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
|
622 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
623 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
624 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
625 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
626 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
627 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
628 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
629 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
630 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
631 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
632 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
633 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
634 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
635 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
636 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
637 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
638 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
639 #ifdef FEAT_MBYTE
|
|
375
|
640 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
641 #endif
|
|
344
|
642 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
643 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
644 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
645 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
646
|
|
324
|
647 /*
|
|
|
648 * Use our own character-case definitions, because the current locale may
|
|
|
649 * differ from what the .spl file uses.
|
|
|
650 * These must not be called with negative number!
|
|
|
651 */
|
|
|
652 #ifndef FEAT_MBYTE
|
|
|
653 /* Non-multi-byte implementation. */
|
|
|
654 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
655 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
656 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
657 #else
|
|
|
658 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
659 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
660 * the "w" library function for characters above 255 if available. */
|
|
|
661 # ifdef HAVE_TOWLOWER
|
|
|
662 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
663 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
664 # else
|
|
|
665 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
666 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
667 # endif
|
|
|
668
|
|
|
669 # ifdef HAVE_TOWUPPER
|
|
|
670 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
671 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
672 # else
|
|
|
673 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
674 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
675 # endif
|
|
|
676
|
|
|
677 # ifdef HAVE_ISWUPPER
|
|
|
678 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
679 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
680 # else
|
|
|
681 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
682 : (c) < 256 ? spelltab.st_isu[c] : (c))
|
|
|
683 # endif
|
|
|
684 #endif
|
|
|
685
|
|
307
|
686
|
|
|
687 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
688 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
236
|
689
|
|
|
690 /*
|
|
|
691 * Main spell-checking function.
|
|
300
|
692 * "ptr" points to a character that could be the start of a word.
|
|
236
|
693 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
694 * or when it's OK it remains unchanged.
|
|
|
695 * This must only be called when 'spelllang' is not empty.
|
|
323
|
696 *
|
|
385
|
697 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
698 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
699 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
700 * worry.
|
|
323
|
701 *
|
|
236
|
702 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
703 * caller can skip over the word.
|
|
|
704 */
|
|
|
705 int
|
|
385
|
706 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
707 win_T *wp; /* current window */
|
|
|
708 char_u *ptr;
|
|
|
709 int *attrp;
|
|
385
|
710 int *capcol; /* column to check for Capital */
|
|
236
|
711 {
|
|
|
712 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
713 be passed to functions quickly. */
|
|
344
|
714 int nrlen = 0; /* found a number first */
|
|
385
|
715 int c;
|
|
236
|
716
|
|
307
|
717 /* A word never starts at a space or a control character. Return quickly
|
|
|
718 * then, skipping over the character. */
|
|
|
719 if (*ptr <= ' ')
|
|
|
720 return 1;
|
|
236
|
721
|
|
344
|
722 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
723 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
724 * "3GPP". */
|
|
300
|
725 if (*ptr >= '0' && *ptr <= '9')
|
|
|
726 {
|
|
316
|
727 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
728 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
729 else
|
|
|
730 {
|
|
344
|
731 mi.mi_end = skipdigits(ptr);
|
|
|
732 nrlen = mi.mi_end - ptr;
|
|
300
|
733 }
|
|
376
|
734 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
735 return (int)(mi.mi_end - ptr);
|
|
346
|
736
|
|
|
737 /* Try including the digits in the word. */
|
|
|
738 mi.mi_fend = ptr + nrlen;
|
|
|
739 }
|
|
|
740 else
|
|
|
741 mi.mi_fend = ptr;
|
|
|
742
|
|
|
743 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
744 mi.mi_word = ptr;
|
|
376
|
745 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
746 {
|
|
|
747 do
|
|
|
748 {
|
|
307
|
749 mb_ptr_adv(mi.mi_fend);
|
|
376
|
750 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
751
|
|
|
752 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
753 {
|
|
|
754 /* Check word starting with capital letter. */
|
|
|
755 #ifdef FEAT_MBYTE
|
|
|
756 c = mb_ptr2char(ptr);
|
|
|
757 #else
|
|
|
758 c = *ptr;
|
|
|
759 #endif
|
|
|
760 if (!SPELL_ISUPPER(c))
|
|
|
761 {
|
|
|
762 *attrp = highlight_attr[HLF_SPC];
|
|
|
763 return (int)(mi.mi_fend - ptr);
|
|
|
764 }
|
|
|
765 }
|
|
|
766 }
|
|
|
767 if (capcol != NULL)
|
|
|
768 *capcol = -1;
|
|
344
|
769
|
|
|
770 /* We always use the characters up to the next non-word character,
|
|
|
771 * also for bad words. */
|
|
|
772 mi.mi_end = mi.mi_fend;
|
|
|
773
|
|
|
774 /* Check caps type later. */
|
|
|
775 mi.mi_capflags = 0;
|
|
|
776 mi.mi_cend = NULL;
|
|
376
|
777 mi.mi_buf = wp->w_buffer;
|
|
344
|
778
|
|
|
779 /* Include one non-word character so that we can check for the
|
|
|
780 * word end. */
|
|
|
781 if (*mi.mi_fend != NUL)
|
|
|
782 mb_ptr_adv(mi.mi_fend);
|
|
|
783
|
|
|
784 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
785 MAXWLEN + 1);
|
|
|
786 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
787
|
|
|
788 /* The word is bad unless we recognize it. */
|
|
|
789 mi.mi_result = SP_BAD;
|
|
|
790
|
|
|
791 /*
|
|
|
792 * Loop over the languages specified in 'spelllang'.
|
|
|
793 * We check them all, because a matching word may be longer than an
|
|
|
794 * already found matching word.
|
|
|
795 */
|
|
|
796 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
797 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
798 {
|
|
|
799 /* Check for a matching word in case-folded words. */
|
|
|
800 find_word(&mi, FIND_FOLDWORD);
|
|
|
801
|
|
|
802 /* Check for a matching word in keep-case words. */
|
|
|
803 find_word(&mi, FIND_KEEPWORD);
|
|
|
804
|
|
|
805 /* Check for matching prefixes. */
|
|
|
806 find_prefix(&mi);
|
|
|
807 }
|
|
|
808
|
|
|
809 if (mi.mi_result != SP_OK)
|
|
|
810 {
|
|
346
|
811 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
812 * rare and local words, e.g., "3GPP". */
|
|
344
|
813 if (nrlen > 0)
|
|
346
|
814 {
|
|
|
815 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
816 return nrlen;
|
|
|
817 }
|
|
344
|
818
|
|
|
819 /* When we are at a non-word character there is no error, just
|
|
|
820 * skip over the character (try looking for a word after it). */
|
|
346
|
821 else if (!SPELL_ISWORDP(ptr))
|
|
243
|
822 {
|
|
385
|
823 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
824 {
|
|
|
825 regmatch_T regmatch;
|
|
|
826
|
|
|
827 /* Check for end of sentence. */
|
|
|
828 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
829 regmatch.rm_ic = FALSE;
|
|
|
830 if (vim_regexec(®match, ptr, 0))
|
|
|
831 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
832 }
|
|
|
833
|
|
344
|
834 #ifdef FEAT_MBYTE
|
|
|
835 if (has_mbyte)
|
|
|
836 return mb_ptr2len_check(ptr);
|
|
|
837 #endif
|
|
|
838 return 1;
|
|
300
|
839 }
|
|
243
|
840
|
|
344
|
841 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
842 *attrp = highlight_attr[HLF_SPB];
|
|
|
843 else if (mi.mi_result == SP_RARE)
|
|
|
844 *attrp = highlight_attr[HLF_SPR];
|
|
|
845 else
|
|
|
846 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
847 }
|
|
|
848
|
|
300
|
849 return (int)(mi.mi_end - ptr);
|
|
236
|
850 }
|
|
|
851
|
|
|
852 /*
|
|
300
|
853 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
854 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
855 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
856 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
857 * tree.
|
|
300
|
858 *
|
|
|
859 * For a match mip->mi_result is updated.
|
|
243
|
860 */
|
|
|
861 static void
|
|
339
|
862 find_word(mip, mode)
|
|
243
|
863 matchinf_T *mip;
|
|
339
|
864 int mode;
|
|
243
|
865 {
|
|
324
|
866 idx_T arridx = 0;
|
|
300
|
867 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
868 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
869 int endidxcnt = 0;
|
|
|
870 int len;
|
|
|
871 int wlen = 0;
|
|
|
872 int flen;
|
|
|
873 int c;
|
|
|
874 char_u *ptr;
|
|
324
|
875 idx_T lo, hi, m;
|
|
243
|
876 #ifdef FEAT_MBYTE
|
|
300
|
877 char_u *s;
|
|
339
|
878 char_u *p;
|
|
307
|
879 #endif
|
|
|
880 int res = SP_BAD;
|
|
300
|
881 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
882 unsigned flags;
|
|
|
883 char_u *byts;
|
|
324
|
884 idx_T *idxs;
|
|
339
|
885 int prefid;
|
|
|
886
|
|
|
887 if (mode == FIND_KEEPWORD)
|
|
236
|
888 {
|
|
300
|
889 /* Check for word with matching case in keep-case tree. */
|
|
|
890 ptr = mip->mi_word;
|
|
|
891 flen = 9999; /* no case folding, always enough bytes */
|
|
|
892 byts = slang->sl_kbyts;
|
|
|
893 idxs = slang->sl_kidxs;
|
|
236
|
894 }
|
|
|
895 else
|
|
|
896 {
|
|
300
|
897 /* Check for case-folded in case-folded tree. */
|
|
|
898 ptr = mip->mi_fword;
|
|
|
899 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
900 byts = slang->sl_fbyts;
|
|
|
901 idxs = slang->sl_fidxs;
|
|
339
|
902
|
|
|
903 if (mode == FIND_PREFIX)
|
|
|
904 {
|
|
|
905 /* Skip over the prefix. */
|
|
|
906 wlen = mip->mi_prefixlen;
|
|
|
907 flen -= mip->mi_prefixlen;
|
|
|
908 }
|
|
243
|
909 }
|
|
|
910
|
|
300
|
911 if (byts == NULL)
|
|
|
912 return; /* array is empty */
|
|
236
|
913
|
|
|
914 /*
|
|
307
|
915 * Repeat advancing in the tree until:
|
|
|
916 * - there is a byte that doesn't match,
|
|
|
917 * - we reach the end of the tree,
|
|
|
918 * - or we reach the end of the line.
|
|
236
|
919 */
|
|
300
|
920 for (;;)
|
|
236
|
921 {
|
|
346
|
922 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
923 flen = fold_more(mip);
|
|
300
|
924
|
|
|
925 len = byts[arridx++];
|
|
|
926
|
|
|
927 /* If the first possible byte is a zero the word could end here.
|
|
|
928 * Remember this index, we first check for the longest word. */
|
|
|
929 if (byts[arridx] == 0)
|
|
|
930 {
|
|
307
|
931 if (endidxcnt == MAXWLEN)
|
|
|
932 {
|
|
|
933 /* Must be a corrupted spell file. */
|
|
|
934 EMSG(_(e_format));
|
|
|
935 return;
|
|
|
936 }
|
|
300
|
937 endlen[endidxcnt] = wlen;
|
|
|
938 endidx[endidxcnt++] = arridx++;
|
|
|
939 --len;
|
|
|
940
|
|
|
941 /* Skip over the zeros, there can be several flag/region
|
|
|
942 * combinations. */
|
|
|
943 while (len > 0 && byts[arridx] == 0)
|
|
|
944 {
|
|
|
945 ++arridx;
|
|
|
946 --len;
|
|
|
947 }
|
|
|
948 if (len == 0)
|
|
|
949 break; /* no children, word must end here */
|
|
|
950 }
|
|
|
951
|
|
|
952 /* Stop looking at end of the line. */
|
|
|
953 if (ptr[wlen] == NUL)
|
|
|
954 break;
|
|
|
955
|
|
|
956 /* Perform a binary search in the list of accepted bytes. */
|
|
|
957 c = ptr[wlen];
|
|
346
|
958 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
959 c = ' ';
|
|
300
|
960 lo = arridx;
|
|
|
961 hi = arridx + len - 1;
|
|
|
962 while (lo < hi)
|
|
|
963 {
|
|
|
964 m = (lo + hi) / 2;
|
|
|
965 if (byts[m] > c)
|
|
|
966 hi = m - 1;
|
|
|
967 else if (byts[m] < c)
|
|
|
968 lo = m + 1;
|
|
|
969 else
|
|
|
970 {
|
|
|
971 lo = hi = m;
|
|
|
972 break;
|
|
236
|
973 }
|
|
|
974 }
|
|
300
|
975
|
|
|
976 /* Stop if there is no matching byte. */
|
|
|
977 if (hi < lo || byts[lo] != c)
|
|
|
978 break;
|
|
|
979
|
|
|
980 /* Continue at the child (if there is one). */
|
|
|
981 arridx = idxs[lo];
|
|
|
982 ++wlen;
|
|
|
983 --flen;
|
|
346
|
984
|
|
|
985 /* One space in the good word may stand for several spaces in the
|
|
|
986 * checked word. */
|
|
|
987 if (c == ' ')
|
|
|
988 {
|
|
|
989 for (;;)
|
|
|
990 {
|
|
|
991 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
992 flen = fold_more(mip);
|
|
|
993 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
994 break;
|
|
|
995 ++wlen;
|
|
|
996 --flen;
|
|
|
997 }
|
|
|
998 }
|
|
236
|
999 }
|
|
|
1000
|
|
300
|
1001 /*
|
|
|
1002 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1003 * first.
|
|
|
1004 */
|
|
|
1005 while (endidxcnt > 0)
|
|
|
1006 {
|
|
|
1007 --endidxcnt;
|
|
|
1008 arridx = endidx[endidxcnt];
|
|
|
1009 wlen = endlen[endidxcnt];
|
|
236
|
1010
|
|
300
|
1011 #ifdef FEAT_MBYTE
|
|
|
1012 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1013 continue; /* not at first byte of character */
|
|
|
1014 #endif
|
|
376
|
1015 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
300
|
1016 continue; /* next char is a word character */
|
|
|
1017
|
|
|
1018 #ifdef FEAT_MBYTE
|
|
339
|
1019 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1020 {
|
|
|
1021 /* Compute byte length in original word, length may change
|
|
339
|
1022 * when folding case. This can be slow, take a shortcut when the
|
|
|
1023 * case-folded word is equal to the keep-case word. */
|
|
300
|
1024 p = mip->mi_word;
|
|
339
|
1025 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1026 {
|
|
|
1027 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1028 mb_ptr_adv(p);
|
|
|
1029 wlen = p - mip->mi_word;
|
|
|
1030 }
|
|
300
|
1031 }
|
|
|
1032 #endif
|
|
236
|
1033
|
|
339
|
1034 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1035 * prefix ID.
|
|
|
1036 * Repeat this if there are more flags/region alternatives until there
|
|
|
1037 * is a match. */
|
|
|
1038 res = SP_BAD;
|
|
|
1039 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1040 --len, ++arridx)
|
|
300
|
1041 {
|
|
|
1042 flags = idxs[arridx];
|
|
324
|
1043
|
|
339
|
1044 /* For the fold-case tree check that the case of the checked word
|
|
|
1045 * matches with what the word in the tree requires.
|
|
|
1046 * For keep-case tree the case is always right. For prefixes we
|
|
|
1047 * don't bother to check. */
|
|
|
1048 if (mode == FIND_FOLDWORD)
|
|
300
|
1049 {
|
|
|
1050 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1051 {
|
|
323
|
1052 /* mi_capflags was set for a different word length, need
|
|
|
1053 * to do it again. */
|
|
300
|
1054 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1055 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1056 }
|
|
|
1057
|
|
346
|
1058 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1059 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1060 continue;
|
|
300
|
1061 }
|
|
236
|
1062
|
|
339
|
1063 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1064 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1065 * mip->mi_prefarridx that find_prefix() filled. */
|
|
339
|
1066 if (mode == FIND_PREFIX)
|
|
300
|
1067 {
|
|
339
|
1068 /* The prefix ID is stored two bytes above the flags. */
|
|
|
1069 prefid = (unsigned)flags >> 16;
|
|
366
|
1070 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
351
|
1071 prefid, mip->mi_fword + mip->mi_prefixlen,
|
|
366
|
1072 slang);
|
|
|
1073 if (c == 0)
|
|
339
|
1074 continue;
|
|
366
|
1075
|
|
|
1076 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1077 if (c & WF_RAREPFX)
|
|
|
1078 flags |= WF_RARE;
|
|
339
|
1079 }
|
|
|
1080
|
|
|
1081 if (flags & WF_BANNED)
|
|
|
1082 res = SP_BANNED;
|
|
|
1083 else if (flags & WF_REGION)
|
|
|
1084 {
|
|
|
1085 /* Check region. */
|
|
|
1086 if ((mip->mi_lp->lp_region & (flags >> 8)) != 0)
|
|
300
|
1087 res = SP_OK;
|
|
339
|
1088 else
|
|
|
1089 res = SP_LOCAL;
|
|
300
|
1090 }
|
|
339
|
1091 else if (flags & WF_RARE)
|
|
|
1092 res = SP_RARE;
|
|
307
|
1093 else
|
|
339
|
1094 res = SP_OK;
|
|
|
1095
|
|
|
1096 /* Always use the longest match and the best result. */
|
|
|
1097 if (mip->mi_result > res)
|
|
|
1098 {
|
|
|
1099 mip->mi_result = res;
|
|
|
1100 mip->mi_end = mip->mi_word + wlen;
|
|
|
1101 }
|
|
351
|
1102 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1103 mip->mi_end = mip->mi_word + wlen;
|
|
|
1104
|
|
|
1105 if (res == SP_OK)
|
|
|
1106 break;
|
|
300
|
1107 }
|
|
|
1108
|
|
307
|
1109 if (res == SP_OK)
|
|
300
|
1110 break;
|
|
|
1111 }
|
|
236
|
1112 }
|
|
|
1113
|
|
323
|
1114 /*
|
|
366
|
1115 * Return non-zero if the prefix indicated by "mip->mi_prefarridx" matches
|
|
|
1116 * with the prefix ID "prefid" for the word "word".
|
|
|
1117 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1118 */
|
|
|
1119 static int
|
|
|
1120 valid_word_prefix(totprefcnt, arridx, prefid, word, slang)
|
|
|
1121 int totprefcnt; /* nr of prefix IDs */
|
|
|
1122 int arridx; /* idx in sl_pidxs[] */
|
|
|
1123 int prefid;
|
|
|
1124 char_u *word;
|
|
|
1125 slang_T *slang;
|
|
|
1126 {
|
|
|
1127 int prefcnt;
|
|
|
1128 int pidx;
|
|
|
1129 regprog_T *rp;
|
|
|
1130 regmatch_T regmatch;
|
|
|
1131
|
|
|
1132 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1133 {
|
|
|
1134 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1135
|
|
|
1136 /* Check the prefix ID. */
|
|
|
1137 if (prefid != (pidx & 0xff))
|
|
|
1138 continue;
|
|
|
1139
|
|
|
1140 /* Check the condition, if there is one. The condition index is
|
|
366
|
1141 * stored in the two bytes above the prefix ID byte. */
|
|
|
1142 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1143 if (rp != NULL)
|
|
|
1144 {
|
|
|
1145 regmatch.regprog = rp;
|
|
|
1146 regmatch.rm_ic = FALSE;
|
|
|
1147 if (!vim_regexec(®match, word, 0))
|
|
|
1148 continue;
|
|
|
1149 }
|
|
|
1150
|
|
366
|
1151 /* It's a match! Return the WF_RAREPFX flag. */
|
|
|
1152 return pidx;
|
|
|
1153 }
|
|
|
1154 return 0;
|
|
351
|
1155 }
|
|
|
1156
|
|
|
1157 /*
|
|
339
|
1158 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1159 * If it does, then check the following word.
|
|
|
1160 *
|
|
|
1161 * For a match mip->mi_result is updated.
|
|
|
1162 */
|
|
|
1163 static void
|
|
|
1164 find_prefix(mip)
|
|
|
1165 matchinf_T *mip;
|
|
|
1166 {
|
|
|
1167 idx_T arridx = 0;
|
|
|
1168 int len;
|
|
|
1169 int wlen = 0;
|
|
|
1170 int flen;
|
|
|
1171 int c;
|
|
|
1172 char_u *ptr;
|
|
|
1173 idx_T lo, hi, m;
|
|
|
1174 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1175 char_u *byts;
|
|
|
1176 idx_T *idxs;
|
|
|
1177
|
|
375
|
1178 byts = slang->sl_pbyts;
|
|
|
1179 if (byts == NULL)
|
|
|
1180 return; /* array is empty */
|
|
|
1181
|
|
339
|
1182 /* We use the case-folded word here, since prefixes are always
|
|
|
1183 * case-folded. */
|
|
|
1184 ptr = mip->mi_fword;
|
|
|
1185 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1186 idxs = slang->sl_pidxs;
|
|
|
1187
|
|
|
1188 /*
|
|
|
1189 * Repeat advancing in the tree until:
|
|
|
1190 * - there is a byte that doesn't match,
|
|
|
1191 * - we reach the end of the tree,
|
|
|
1192 * - or we reach the end of the line.
|
|
|
1193 */
|
|
|
1194 for (;;)
|
|
|
1195 {
|
|
|
1196 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1197 flen = fold_more(mip);
|
|
|
1198
|
|
|
1199 len = byts[arridx++];
|
|
|
1200
|
|
|
1201 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1202 * Check if the following word matches and supports the prefix. */
|
|
|
1203 if (byts[arridx] == 0)
|
|
|
1204 {
|
|
|
1205 /* There can be several prefixes with different conditions. We
|
|
|
1206 * try them all, since we don't know which one will give the
|
|
|
1207 * longest match. The word is the same each time, pass the list
|
|
|
1208 * of possible prefixes to find_word(). */
|
|
|
1209 mip->mi_prefarridx = arridx;
|
|
|
1210 mip->mi_prefcnt = len;
|
|
|
1211 while (len > 0 && byts[arridx] == 0)
|
|
|
1212 {
|
|
|
1213 ++arridx;
|
|
|
1214 --len;
|
|
|
1215 }
|
|
|
1216 mip->mi_prefcnt -= len;
|
|
|
1217
|
|
|
1218 /* Find the word that comes after the prefix. */
|
|
|
1219 mip->mi_prefixlen = wlen;
|
|
|
1220 find_word(mip, FIND_PREFIX);
|
|
|
1221
|
|
|
1222
|
|
|
1223 if (len == 0)
|
|
|
1224 break; /* no children, word must end here */
|
|
|
1225 }
|
|
|
1226
|
|
|
1227 /* Stop looking at end of the line. */
|
|
|
1228 if (ptr[wlen] == NUL)
|
|
|
1229 break;
|
|
|
1230
|
|
|
1231 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1232 c = ptr[wlen];
|
|
|
1233 lo = arridx;
|
|
|
1234 hi = arridx + len - 1;
|
|
|
1235 while (lo < hi)
|
|
|
1236 {
|
|
|
1237 m = (lo + hi) / 2;
|
|
|
1238 if (byts[m] > c)
|
|
|
1239 hi = m - 1;
|
|
|
1240 else if (byts[m] < c)
|
|
|
1241 lo = m + 1;
|
|
|
1242 else
|
|
|
1243 {
|
|
|
1244 lo = hi = m;
|
|
|
1245 break;
|
|
|
1246 }
|
|
|
1247 }
|
|
|
1248
|
|
|
1249 /* Stop if there is no matching byte. */
|
|
|
1250 if (hi < lo || byts[lo] != c)
|
|
|
1251 break;
|
|
|
1252
|
|
|
1253 /* Continue at the child (if there is one). */
|
|
|
1254 arridx = idxs[lo];
|
|
|
1255 ++wlen;
|
|
|
1256 --flen;
|
|
|
1257 }
|
|
|
1258 }
|
|
|
1259
|
|
|
1260 /*
|
|
|
1261 * Need to fold at least one more character. Do until next non-word character
|
|
|
1262 * for efficiency.
|
|
|
1263 * Return the length of the folded chars in bytes.
|
|
|
1264 */
|
|
|
1265 static int
|
|
|
1266 fold_more(mip)
|
|
|
1267 matchinf_T *mip;
|
|
|
1268 {
|
|
|
1269 int flen;
|
|
|
1270 char_u *p;
|
|
|
1271
|
|
|
1272 p = mip->mi_fend;
|
|
|
1273 do
|
|
|
1274 {
|
|
|
1275 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1276 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1277
|
|
|
1278 /* Include the non-word character so that we can check for the
|
|
|
1279 * word end. */
|
|
|
1280 if (*mip->mi_fend != NUL)
|
|
|
1281 mb_ptr_adv(mip->mi_fend);
|
|
|
1282
|
|
|
1283 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1284 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1285 MAXWLEN - mip->mi_fwordlen);
|
|
|
1286 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1287 mip->mi_fwordlen += flen;
|
|
|
1288 return flen;
|
|
|
1289 }
|
|
|
1290
|
|
|
1291 /*
|
|
323
|
1292 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1293 * case.
|
|
|
1294 */
|
|
|
1295 static int
|
|
|
1296 spell_valid_case(origflags, treeflags)
|
|
|
1297 int origflags; /* flags for the checked word. */
|
|
|
1298 int treeflags; /* flags for the word in the spell tree */
|
|
|
1299 {
|
|
|
1300 return (origflags == WF_ALLCAP
|
|
|
1301 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
|
1302 && ((treeflags & WF_ONECAP) == 0 || origflags == WF_ONECAP)));
|
|
|
1303 }
|
|
|
1304
|
|
351
|
1305 /*
|
|
|
1306 * Return TRUE if spell checking is not enabled.
|
|
|
1307 */
|
|
|
1308 static int
|
|
|
1309 no_spell_checking()
|
|
|
1310 {
|
|
|
1311 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1312 {
|
|
|
1313 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1314 return TRUE;
|
|
|
1315 }
|
|
|
1316 return FALSE;
|
|
|
1317 }
|
|
300
|
1318
|
|
236
|
1319 /*
|
|
|
1320 * Move to next spell error.
|
|
323
|
1321 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
236
|
1322 * Return OK if found, FAIL otherwise.
|
|
|
1323 */
|
|
|
1324 int
|
|
323
|
1325 spell_move_to(dir, allwords, curline)
|
|
236
|
1326 int dir; /* FORWARD or BACKWARD */
|
|
|
1327 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1328 int curline;
|
|
236
|
1329 {
|
|
249
|
1330 linenr_T lnum;
|
|
|
1331 pos_T found_pos;
|
|
236
|
1332 char_u *line;
|
|
|
1333 char_u *p;
|
|
346
|
1334 char_u *endp;
|
|
|
1335 int attr;
|
|
236
|
1336 int len;
|
|
249
|
1337 int has_syntax = syntax_present(curbuf);
|
|
|
1338 int col;
|
|
|
1339 int can_spell;
|
|
346
|
1340 char_u *buf = NULL;
|
|
|
1341 int buflen = 0;
|
|
|
1342 int skip = 0;
|
|
385
|
1343 int capcol = -1;
|
|
236
|
1344
|
|
351
|
1345 if (no_spell_checking())
|
|
236
|
1346 return FAIL;
|
|
|
1347
|
|
249
|
1348 /*
|
|
|
1349 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1350 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1351 *
|
|
|
1352 * When searching backwards, we continue in the line to find the last
|
|
|
1353 * bad word (in the cursor line: before the cursor).
|
|
346
|
1354 *
|
|
|
1355 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1356 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1357 * though...
|
|
249
|
1358 */
|
|
|
1359 lnum = curwin->w_cursor.lnum;
|
|
|
1360 found_pos.lnum = 0;
|
|
236
|
1361
|
|
|
1362 while (!got_int)
|
|
|
1363 {
|
|
249
|
1364 line = ml_get(lnum);
|
|
346
|
1365
|
|
|
1366 len = STRLEN(line);
|
|
|
1367 if (buflen < len + MAXWLEN + 2)
|
|
|
1368 {
|
|
|
1369 vim_free(buf);
|
|
|
1370 buflen = len + MAXWLEN + 2;
|
|
|
1371 buf = alloc(buflen);
|
|
|
1372 if (buf == NULL)
|
|
|
1373 break;
|
|
|
1374 }
|
|
|
1375
|
|
385
|
1376 /* In first line check first word for Capital. */
|
|
|
1377 if (lnum == 1)
|
|
|
1378 capcol = 0;
|
|
|
1379
|
|
|
1380 /* For checking first word with a capital skip white space. */
|
|
|
1381 if (capcol == 0)
|
|
|
1382 capcol = skipwhite(line) - line;
|
|
|
1383
|
|
346
|
1384 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1385 * possible. */
|
|
|
1386 STRCPY(buf, line);
|
|
|
1387 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1388 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1389
|
|
|
1390 p = buf + skip;
|
|
|
1391 endp = buf + len;
|
|
|
1392 while (p < endp)
|
|
236
|
1393 {
|
|
300
|
1394 /* When searching backward don't search after the cursor. */
|
|
|
1395 if (dir == BACKWARD
|
|
|
1396 && lnum == curwin->w_cursor.lnum
|
|
346
|
1397 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1398 break;
|
|
249
|
1399
|
|
300
|
1400 /* start of word */
|
|
346
|
1401 attr = 0;
|
|
385
|
1402 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1403
|
|
300
|
1404 if (attr != 0)
|
|
|
1405 {
|
|
|
1406 /* We found a bad word. Check the attribute. */
|
|
|
1407 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1408 {
|
|
300
|
1409 /* When searching forward only accept a bad word after
|
|
|
1410 * the cursor. */
|
|
|
1411 if (dir == BACKWARD
|
|
|
1412 || lnum > curwin->w_cursor.lnum
|
|
|
1413 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1414 && (colnr_T)(curline ? p - buf + len
|
|
|
1415 : p - buf)
|
|
300
|
1416 > curwin->w_cursor.col))
|
|
236
|
1417 {
|
|
300
|
1418 if (has_syntax)
|
|
249
|
1419 {
|
|
346
|
1420 col = p - buf;
|
|
300
|
1421 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1422 FALSE, &can_spell);
|
|
|
1423 }
|
|
|
1424 else
|
|
|
1425 can_spell = TRUE;
|
|
249
|
1426
|
|
300
|
1427 if (can_spell)
|
|
|
1428 {
|
|
|
1429 found_pos.lnum = lnum;
|
|
346
|
1430 found_pos.col = p - buf;
|
|
249
|
1431 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1432 found_pos.coladd = 0;
|
|
249
|
1433 #endif
|
|
300
|
1434 if (dir == FORWARD)
|
|
|
1435 {
|
|
|
1436 /* No need to search further. */
|
|
|
1437 curwin->w_cursor = found_pos;
|
|
346
|
1438 vim_free(buf);
|
|
300
|
1439 return OK;
|
|
249
|
1440 }
|
|
|
1441 }
|
|
236
|
1442 }
|
|
|
1443 }
|
|
|
1444 }
|
|
|
1445
|
|
300
|
1446 /* advance to character after the word */
|
|
|
1447 p += len;
|
|
385
|
1448 capcol -= len;
|
|
236
|
1449 }
|
|
|
1450
|
|
323
|
1451 if (curline)
|
|
346
|
1452 break; /* only check cursor line */
|
|
323
|
1453
|
|
236
|
1454 /* Advance to next line. */
|
|
249
|
1455 if (dir == BACKWARD)
|
|
|
1456 {
|
|
|
1457 if (found_pos.lnum != 0)
|
|
|
1458 {
|
|
|
1459 /* Use the last match in the line. */
|
|
|
1460 curwin->w_cursor = found_pos;
|
|
346
|
1461 vim_free(buf);
|
|
249
|
1462 return OK;
|
|
|
1463 }
|
|
|
1464 if (lnum == 1)
|
|
346
|
1465 break;
|
|
249
|
1466 --lnum;
|
|
385
|
1467 capcol = -1;
|
|
249
|
1468 }
|
|
|
1469 else
|
|
|
1470 {
|
|
|
1471 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1472 break;
|
|
249
|
1473 ++lnum;
|
|
346
|
1474
|
|
|
1475 /* Skip the characters at the start of the next line that were
|
|
|
1476 * included in a match crossing line boundaries. */
|
|
|
1477 if (attr == 0)
|
|
|
1478 skip = p - endp;
|
|
|
1479 else
|
|
|
1480 skip = 0;
|
|
385
|
1481
|
|
|
1482 /* Capscol skips over the inserted space. */
|
|
|
1483 --capcol;
|
|
|
1484
|
|
|
1485 /* But after empty line check first word in next line */
|
|
|
1486 if (*skipwhite(line) == NUL)
|
|
|
1487 capcol = 0;
|
|
249
|
1488 }
|
|
236
|
1489
|
|
|
1490 line_breakcheck();
|
|
|
1491 }
|
|
|
1492
|
|
346
|
1493 vim_free(buf);
|
|
|
1494 return FAIL;
|
|
|
1495 }
|
|
|
1496
|
|
|
1497 /*
|
|
|
1498 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1499 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1500 */
|
|
|
1501 void
|
|
|
1502 spell_cat_line(buf, line, maxlen)
|
|
|
1503 char_u *buf;
|
|
|
1504 char_u *line;
|
|
|
1505 int maxlen;
|
|
|
1506 {
|
|
|
1507 char_u *p;
|
|
|
1508 int n;
|
|
|
1509
|
|
|
1510 p = skipwhite(line);
|
|
|
1511 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1512 p = skipwhite(p + 1);
|
|
|
1513
|
|
|
1514 if (*p != NUL)
|
|
|
1515 {
|
|
|
1516 *buf = ' ';
|
|
376
|
1517 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1518 n = p - line;
|
|
|
1519 if (n >= maxlen)
|
|
|
1520 n = maxlen - 1;
|
|
|
1521 vim_memset(buf + 1, ' ', n);
|
|
|
1522 }
|
|
236
|
1523 }
|
|
|
1524
|
|
|
1525 /*
|
|
307
|
1526 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1527 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1528 */
|
|
307
|
1529 static void
|
|
236
|
1530 spell_load_lang(lang)
|
|
|
1531 char_u *lang;
|
|
|
1532 {
|
|
310
|
1533 char_u fname_enc[85];
|
|
236
|
1534 int r;
|
|
307
|
1535 char_u langcp[MAXWLEN + 1];
|
|
|
1536
|
|
310
|
1537 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1538 * It's truncated when an error is detected. */
|
|
|
1539 STRCPY(langcp, lang);
|
|
|
1540
|
|
310
|
1541 /*
|
|
|
1542 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1543 */
|
|
|
1544 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1545 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1546 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1547
|
|
|
1548 if (r == FAIL && *langcp != NUL)
|
|
|
1549 {
|
|
|
1550 /* Try loading the ASCII version. */
|
|
310
|
1551 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1552 "spell/%s.ascii.spl", lang);
|
|
310
|
1553 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1554 }
|
|
|
1555
|
|
|
1556 if (r == FAIL)
|
|
|
1557 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
|
236
|
1558 fname_enc + 6);
|
|
310
|
1559 else if (*langcp != NUL)
|
|
|
1560 {
|
|
|
1561 /* Load all the additions. */
|
|
|
1562 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1563 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1564 }
|
|
|
1565 }
|
|
|
1566
|
|
|
1567 /*
|
|
|
1568 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1569 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1570 */
|
|
|
1571 static char_u *
|
|
|
1572 spell_enc()
|
|
|
1573 {
|
|
|
1574
|
|
|
1575 #ifdef FEAT_MBYTE
|
|
|
1576 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1577 return p_enc;
|
|
|
1578 #endif
|
|
|
1579 return (char_u *)"latin1";
|
|
236
|
1580 }
|
|
|
1581
|
|
|
1582 /*
|
|
385
|
1583 * Get the name of the .spl file for the internal wordlist into
|
|
|
1584 * "fname[MAXPATHL]".
|
|
|
1585 */
|
|
|
1586 static void
|
|
|
1587 int_wordlist_spl(fname)
|
|
|
1588 char_u *fname;
|
|
|
1589 {
|
|
|
1590 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
1591 int_wordlist, spell_enc());
|
|
|
1592 }
|
|
|
1593
|
|
|
1594 /*
|
|
236
|
1595 * Allocate a new slang_T.
|
|
|
1596 * Caller must fill "sl_next".
|
|
|
1597 */
|
|
|
1598 static slang_T *
|
|
|
1599 slang_alloc(lang)
|
|
|
1600 char_u *lang;
|
|
|
1601 {
|
|
|
1602 slang_T *lp;
|
|
|
1603
|
|
300
|
1604 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1605 if (lp != NULL)
|
|
|
1606 {
|
|
|
1607 lp->sl_name = vim_strsave(lang);
|
|
323
|
1608 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
344
|
1609 ga_init2(&lp->sl_sal, sizeof(salitem_T), 10);
|
|
236
|
1610 }
|
|
|
1611 return lp;
|
|
|
1612 }
|
|
|
1613
|
|
|
1614 /*
|
|
|
1615 * Free the contents of an slang_T and the structure itself.
|
|
|
1616 */
|
|
|
1617 static void
|
|
|
1618 slang_free(lp)
|
|
|
1619 slang_T *lp;
|
|
|
1620 {
|
|
|
1621 vim_free(lp->sl_name);
|
|
310
|
1622 vim_free(lp->sl_fname);
|
|
|
1623 slang_clear(lp);
|
|
|
1624 vim_free(lp);
|
|
|
1625 }
|
|
|
1626
|
|
|
1627 /*
|
|
|
1628 * Clear an slang_T so that the file can be reloaded.
|
|
|
1629 */
|
|
|
1630 static void
|
|
|
1631 slang_clear(lp)
|
|
|
1632 slang_T *lp;
|
|
|
1633 {
|
|
339
|
1634 garray_T *gap;
|
|
|
1635 fromto_T *ftp;
|
|
344
|
1636 salitem_T *smp;
|
|
339
|
1637 int i;
|
|
323
|
1638
|
|
300
|
1639 vim_free(lp->sl_fbyts);
|
|
310
|
1640 lp->sl_fbyts = NULL;
|
|
300
|
1641 vim_free(lp->sl_kbyts);
|
|
310
|
1642 lp->sl_kbyts = NULL;
|
|
339
|
1643 vim_free(lp->sl_pbyts);
|
|
|
1644 lp->sl_pbyts = NULL;
|
|
|
1645
|
|
300
|
1646 vim_free(lp->sl_fidxs);
|
|
310
|
1647 lp->sl_fidxs = NULL;
|
|
300
|
1648 vim_free(lp->sl_kidxs);
|
|
310
|
1649 lp->sl_kidxs = NULL;
|
|
339
|
1650 vim_free(lp->sl_pidxs);
|
|
|
1651 lp->sl_pidxs = NULL;
|
|
323
|
1652
|
|
344
|
1653 gap = &lp->sl_rep;
|
|
|
1654 while (gap->ga_len > 0)
|
|
323
|
1655 {
|
|
344
|
1656 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1657 vim_free(ftp->ft_from);
|
|
|
1658 vim_free(ftp->ft_to);
|
|
323
|
1659 }
|
|
344
|
1660 ga_clear(gap);
|
|
|
1661
|
|
|
1662 gap = &lp->sl_sal;
|
|
375
|
1663 if (lp->sl_sofo)
|
|
376
|
1664 {
|
|
|
1665 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1666 if (gap->ga_data != NULL)
|
|
|
1667 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1668 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1669 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1670 }
|
|
375
|
1671 else
|
|
|
1672 /* SAL items: free salitem_T items */
|
|
|
1673 while (gap->ga_len > 0)
|
|
|
1674 {
|
|
|
1675 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1676 vim_free(smp->sm_lead);
|
|
|
1677 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1678 vim_free(smp->sm_to);
|
|
|
1679 #ifdef FEAT_MBYTE
|
|
|
1680 vim_free(smp->sm_lead_w);
|
|
|
1681 vim_free(smp->sm_oneof_w);
|
|
|
1682 vim_free(smp->sm_to_w);
|
|
|
1683 #endif
|
|
|
1684 }
|
|
344
|
1685 ga_clear(gap);
|
|
323
|
1686
|
|
339
|
1687 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1688 vim_free(lp->sl_prefprog[i]);
|
|
376
|
1689 lp->sl_prefixcnt = 0;
|
|
339
|
1690 vim_free(lp->sl_prefprog);
|
|
376
|
1691 lp->sl_prefprog = NULL;
|
|
|
1692
|
|
|
1693 vim_free(lp->sl_midword);
|
|
|
1694 lp->sl_midword = NULL;
|
|
339
|
1695
|
|
330
|
1696 #ifdef FEAT_MBYTE
|
|
|
1697 {
|
|
|
1698 int todo = lp->sl_map_hash.ht_used;
|
|
|
1699 hashitem_T *hi;
|
|
|
1700
|
|
|
1701 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1702 if (!HASHITEM_EMPTY(hi))
|
|
|
1703 {
|
|
|
1704 --todo;
|
|
|
1705 vim_free(hi->hi_key);
|
|
|
1706 }
|
|
|
1707 }
|
|
|
1708 hash_clear(&lp->sl_map_hash);
|
|
|
1709 #endif
|
|
236
|
1710 }
|
|
|
1711
|
|
|
1712 /*
|
|
307
|
1713 * Load one spell file and store the info into a slang_T.
|
|
236
|
1714 * Invoked through do_in_runtimepath().
|
|
|
1715 */
|
|
|
1716 static void
|
|
310
|
1717 spell_load_cb(fname, cookie)
|
|
236
|
1718 char_u *fname;
|
|
307
|
1719 void *cookie; /* points to the language name */
|
|
236
|
1720 {
|
|
323
|
1721 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1722 }
|
|
|
1723
|
|
|
1724 /*
|
|
|
1725 * Load one spell file and store the info into a slang_T.
|
|
|
1726 *
|
|
|
1727 * This is invoked in two ways:
|
|
|
1728 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
1729 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
1730 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
1731 * points to the existing slang_T.
|
|
323
|
1732 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
1733 */
|
|
323
|
1734 static slang_T *
|
|
|
1735 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
1736 char_u *fname;
|
|
|
1737 char_u *lang;
|
|
|
1738 slang_T *old_lp;
|
|
323
|
1739 int silent; /* no error if file doesn't exist */
|
|
310
|
1740 {
|
|
236
|
1741 FILE *fd;
|
|
|
1742 char_u buf[MAXWLEN + 1];
|
|
|
1743 char_u *p;
|
|
339
|
1744 char_u *bp;
|
|
|
1745 idx_T *ip;
|
|
236
|
1746 int i;
|
|
339
|
1747 int n;
|
|
300
|
1748 int len;
|
|
236
|
1749 int round;
|
|
|
1750 char_u *save_sourcing_name = sourcing_name;
|
|
|
1751 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
255
|
1752 int cnt, ccnt;
|
|
|
1753 char_u *fol;
|
|
307
|
1754 slang_T *lp = NULL;
|
|
323
|
1755 garray_T *gap;
|
|
|
1756 fromto_T *ftp;
|
|
344
|
1757 salitem_T *smp;
|
|
323
|
1758 short *first;
|
|
324
|
1759 idx_T idx;
|
|
344
|
1760 int c = 0;
|
|
236
|
1761
|
|
310
|
1762 fd = mch_fopen((char *)fname, "r");
|
|
236
|
1763 if (fd == NULL)
|
|
|
1764 {
|
|
323
|
1765 if (!silent)
|
|
|
1766 EMSG2(_(e_notopen), fname);
|
|
|
1767 else if (p_verbose > 2)
|
|
|
1768 {
|
|
|
1769 verbose_enter();
|
|
|
1770 smsg((char_u *)e_notopen, fname);
|
|
|
1771 verbose_leave();
|
|
|
1772 }
|
|
255
|
1773 goto endFAIL;
|
|
236
|
1774 }
|
|
310
|
1775 if (p_verbose > 2)
|
|
|
1776 {
|
|
|
1777 verbose_enter();
|
|
|
1778 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
1779 verbose_leave();
|
|
|
1780 }
|
|
|
1781
|
|
|
1782 if (old_lp == NULL)
|
|
|
1783 {
|
|
|
1784 lp = slang_alloc(lang);
|
|
|
1785 if (lp == NULL)
|
|
|
1786 goto endFAIL;
|
|
|
1787
|
|
|
1788 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
1789 lp->sl_fname = vim_strsave(fname);
|
|
|
1790 if (lp->sl_fname == NULL)
|
|
|
1791 goto endFAIL;
|
|
|
1792
|
|
|
1793 /* Check for .add.spl. */
|
|
|
1794 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
1795 }
|
|
|
1796 else
|
|
|
1797 lp = old_lp;
|
|
307
|
1798
|
|
236
|
1799 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
1800 sourcing_name = fname;
|
|
|
1801 sourcing_lnum = 0;
|
|
|
1802
|
|
339
|
1803 /* <HEADER>: <fileID>
|
|
|
1804 * <regioncnt> <regionname> ...
|
|
|
1805 * <charflagslen> <charflags>
|
|
|
1806 * <fcharslen> <fchars>
|
|
366
|
1807 * <midwordlen> <midword>
|
|
339
|
1808 * <prefcondcnt> <prefcond> ...
|
|
|
1809 */
|
|
236
|
1810 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
1811 buf[i] = getc(fd); /* <fileID> */
|
|
|
1812 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
1813 {
|
|
|
1814 EMSG(_("E757: Wrong file ID in spell file"));
|
|
255
|
1815 goto endFAIL;
|
|
236
|
1816 }
|
|
|
1817
|
|
|
1818 cnt = getc(fd); /* <regioncnt> */
|
|
255
|
1819 if (cnt < 0)
|
|
236
|
1820 {
|
|
|
1821 truncerr:
|
|
381
|
1822 EMSG(_(e_spell_trunc));
|
|
255
|
1823 goto endFAIL;
|
|
236
|
1824 }
|
|
|
1825 if (cnt > 8)
|
|
|
1826 {
|
|
|
1827 formerr:
|
|
307
|
1828 EMSG(_(e_format));
|
|
255
|
1829 goto endFAIL;
|
|
236
|
1830 }
|
|
|
1831 for (i = 0; i < cnt; ++i)
|
|
|
1832 {
|
|
|
1833 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
|
1834 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
|
1835 }
|
|
|
1836 lp->sl_regions[cnt * 2] = NUL;
|
|
|
1837
|
|
381
|
1838 /* <charflagslen> <charflags> */
|
|
|
1839 p = read_cnt_string(fd, 1, &cnt);
|
|
|
1840 if (cnt == FAIL)
|
|
|
1841 goto endFAIL;
|
|
|
1842
|
|
|
1843 /* <fcharslen> <fchars> */
|
|
|
1844 fol = read_cnt_string(fd, 2, &cnt);
|
|
|
1845 if (cnt == FAIL)
|
|
|
1846 {
|
|
300
|
1847 vim_free(p);
|
|
381
|
1848 goto endFAIL;
|
|
|
1849 }
|
|
|
1850
|
|
|
1851 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
1852 if (p != NULL && fol != NULL)
|
|
|
1853 i = set_spell_charflags(p, fol);
|
|
|
1854
|
|
|
1855 vim_free(p);
|
|
|
1856 vim_free(fol);
|
|
|
1857
|
|
|
1858 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
1859 if ((p == NULL) != (fol == NULL))
|
|
|
1860 goto formerr;
|
|
255
|
1861
|
|
366
|
1862 /* <midwordlen> <midword> */
|
|
376
|
1863 lp->sl_midword = read_cnt_string(fd, 2, &cnt);
|
|
381
|
1864 if (cnt == FAIL)
|
|
376
|
1865 goto endFAIL;
|
|
366
|
1866
|
|
339
|
1867 /* <prefcondcnt> <prefcond> ... */
|
|
|
1868 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
1869 if (cnt > 0)
|
|
|
1870 {
|
|
|
1871 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
1872 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
1873 if (lp->sl_prefprog == NULL)
|
|
|
1874 goto endFAIL;
|
|
|
1875 lp->sl_prefixcnt = cnt;
|
|
|
1876
|
|
|
1877 for (i = 0; i < cnt; ++i)
|
|
|
1878 {
|
|
|
1879 /* <prefcond> : <condlen> <condstr> */
|
|
|
1880 n = getc(fd); /* <condlen> */
|
|
|
1881 if (n < 0)
|
|
|
1882 goto formerr;
|
|
|
1883 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
1884 * compile the regexp program used to check for the condition. */
|
|
|
1885 if (n > 0)
|
|
|
1886 {
|
|
344
|
1887 buf[0] = '^'; /* always match at one position only */
|
|
|
1888 p = buf + 1;
|
|
339
|
1889 while (n-- > 0)
|
|
|
1890 *p++ = getc(fd); /* <condstr> */
|
|
|
1891 *p = NUL;
|
|
|
1892 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
1893 }
|
|
|
1894 }
|
|
|
1895 }
|
|
|
1896
|
|
|
1897
|
|
323
|
1898 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
1899 * <salflags> <salcount> <sal> ...
|
|
|
1900 * <maplen> <mapstr> */
|
|
339
|
1901
|
|
344
|
1902 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
1903 if (cnt < 0)
|
|
|
1904 goto formerr;
|
|
|
1905
|
|
|
1906 gap = &lp->sl_rep;
|
|
|
1907 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1908 goto endFAIL;
|
|
|
1909
|
|
|
1910 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
1911 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
1912 {
|
|
|
1913 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
381
|
1914 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
1915 if (i == FAIL)
|
|
|
1916 goto endFAIL;
|
|
|
1917 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
1918 if (i == FAIL)
|
|
|
1919 {
|
|
|
1920 vim_free(ftp->ft_from);
|
|
|
1921 goto endFAIL;
|
|
344
|
1922 }
|
|
|
1923 }
|
|
|
1924
|
|
|
1925 /* Fill the first-index table. */
|
|
|
1926 first = lp->sl_rep_first;
|
|
|
1927 for (i = 0; i < 256; ++i)
|
|
|
1928 first[i] = -1;
|
|
|
1929 for (i = 0; i < gap->ga_len; ++i)
|
|
323
|
1930 {
|
|
344
|
1931 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
1932 if (first[*ftp->ft_from] == -1)
|
|
|
1933 first[*ftp->ft_from] = i;
|
|
|
1934 }
|
|
|
1935
|
|
|
1936 i = getc(fd); /* <salflags> */
|
|
|
1937 if (i & SAL_F0LLOWUP)
|
|
|
1938 lp->sl_followup = TRUE;
|
|
|
1939 if (i & SAL_COLLAPSE)
|
|
|
1940 lp->sl_collapse = TRUE;
|
|
|
1941 if (i & SAL_REM_ACCENTS)
|
|
|
1942 lp->sl_rem_accents = TRUE;
|
|
375
|
1943 if (i & SAL_SOFO)
|
|
|
1944 lp->sl_sofo = TRUE;
|
|
344
|
1945
|
|
|
1946 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
1947 if (cnt < 0)
|
|
|
1948 goto formerr;
|
|
|
1949
|
|
375
|
1950 if (lp->sl_sofo)
|
|
|
1951 {
|
|
|
1952 /*
|
|
|
1953 * SOFOFROM and SOFOTO items come in one <salfrom> and <salto>
|
|
|
1954 */
|
|
|
1955 if (cnt != 1)
|
|
|
1956 goto formerr;
|
|
|
1957
|
|
381
|
1958 /* <salfromlen> <salfrom> */
|
|
|
1959 bp = read_cnt_string(fd, 2, &cnt);
|
|
|
1960 if (cnt == FAIL)
|
|
375
|
1961 goto endFAIL;
|
|
381
|
1962
|
|
|
1963 /* <saltolen> <salto> */
|
|
|
1964 fol = read_cnt_string(fd, 2, &cnt);
|
|
|
1965 if (cnt == FAIL)
|
|
375
|
1966 {
|
|
|
1967 vim_free(bp);
|
|
344
|
1968 goto endFAIL;
|
|
375
|
1969 }
|
|
381
|
1970
|
|
|
1971 /* Store the info in lp->sl_sal and/or lp->sl_sal_first. */
|
|
|
1972 i = set_sofo(lp, bp, fol);
|
|
|
1973
|
|
375
|
1974 vim_free(bp);
|
|
|
1975 vim_free(fol);
|
|
381
|
1976 if (i == FAIL)
|
|
|
1977 goto formerr;
|
|
375
|
1978 }
|
|
|
1979 else
|
|
|
1980 {
|
|
|
1981 /*
|
|
|
1982 * SAL items
|
|
|
1983 */
|
|
|
1984 gap = &lp->sl_sal;
|
|
|
1985 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1986 goto endFAIL;
|
|
|
1987
|
|
|
1988 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
1989 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
1990 {
|
|
|
1991 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
1992 ccnt = getc(fd); /* <salfromlen> */
|
|
|
1993 if (ccnt < 0)
|
|
|
1994 goto formerr;
|
|
|
1995 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
1996 goto endFAIL;
|
|
|
1997 smp->sm_lead = p;
|
|
|
1998
|
|
|
1999 /* Read up to the first special char into sm_lead. */
|
|
|
2000 for (i = 0; i < ccnt; ++i)
|
|
|
2001 {
|
|
|
2002 c = getc(fd); /* <salfrom> */
|
|
|
2003 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2004 break;
|
|
|
2005 *p++ = c;
|
|
|
2006 }
|
|
|
2007 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2008 *p++ = NUL;
|
|
|
2009
|
|
|
2010 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2011 if (c == '(')
|
|
|
2012 {
|
|
|
2013 smp->sm_oneof = p;
|
|
|
2014 for (++i; i < ccnt; ++i)
|
|
|
2015 {
|
|
|
2016 c = getc(fd); /* <salfrom> */
|
|
|
2017 if (c == ')')
|
|
|
2018 break;
|
|
|
2019 *p++ = c;
|
|
|
2020 }
|
|
|
2021 *p++ = NUL;
|
|
|
2022 if (++i < ccnt)
|
|
|
2023 c = getc(fd);
|
|
|
2024 }
|
|
372
|
2025 else
|
|
375
|
2026 smp->sm_oneof = NULL;
|
|
|
2027
|
|
|
2028 /* Any following chars go in sm_rules. */
|
|
|
2029 smp->sm_rules = p;
|
|
|
2030 if (i < ccnt)
|
|
|
2031 /* store the char we got while checking for end of sm_lead */
|
|
|
2032 *p++ = c;
|
|
|
2033 for (++i; i < ccnt; ++i)
|
|
|
2034 *p++ = getc(fd); /* <salfrom> */
|
|
|
2035 *p++ = NUL;
|
|
|
2036
|
|
381
|
2037 /* <saltolen> <salto> */
|
|
|
2038 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
2039 if (ccnt == FAIL)
|
|
372
|
2040 {
|
|
|
2041 vim_free(smp->sm_lead);
|
|
375
|
2042 goto formerr;
|
|
|
2043 }
|
|
|
2044
|
|
372
|
2045 #ifdef FEAT_MBYTE
|
|
|
2046 if (has_mbyte)
|
|
|
2047 {
|
|
375
|
2048 /* convert the multi-byte strings to wide char strings */
|
|
|
2049 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2050 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2051 if (smp->sm_oneof == NULL)
|
|
|
2052 smp->sm_oneof_w = NULL;
|
|
|
2053 else
|
|
|
2054 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
2055 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
2056 if (smp->sm_lead_w == NULL
|
|
|
2057 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
2058 || smp->sm_to_w == NULL)
|
|
372
|
2059 {
|
|
375
|
2060 vim_free(smp->sm_lead);
|
|
|
2061 vim_free(smp->sm_to);
|
|
|
2062 vim_free(smp->sm_lead_w);
|
|
|
2063 vim_free(smp->sm_oneof_w);
|
|
|
2064 vim_free(smp->sm_to_w);
|
|
|
2065 goto endFAIL;
|
|
372
|
2066 }
|
|
|
2067 }
|
|
|
2068 #endif
|
|
|
2069 }
|
|
375
|
2070
|
|
|
2071 /* Fill the first-index table. */
|
|
381
|
2072 set_sal_first(lp);
|
|
|
2073 }
|
|
|
2074
|
|
|
2075 /* <maplen> <mapstr> */
|
|
|
2076 p = read_cnt_string(fd, 2, &cnt);
|
|
|
2077 if (cnt == FAIL)
|
|
323
|
2078 goto endFAIL;
|
|
330
|
2079 set_map_str(lp, p);
|
|
|
2080 vim_free(p);
|
|
323
|
2081
|
|
236
|
2082
|
|
300
|
2083 /* round 1: <LWORDTREE>
|
|
339
|
2084 * round 2: <KWORDTREE>
|
|
|
2085 * round 3: <PREFIXTREE> */
|
|
|
2086 for (round = 1; round <= 3; ++round)
|
|
236
|
2087 {
|
|
300
|
2088 /* The tree size was computed when writing the file, so that we can
|
|
|
2089 * allocate it as one long block. <nodecount> */
|
|
|
2090 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2091 if (len < 0)
|
|
|
2092 goto truncerr;
|
|
|
2093 if (len > 0)
|
|
236
|
2094 {
|
|
300
|
2095 /* Allocate the byte array. */
|
|
339
|
2096 bp = lalloc((long_u)len, TRUE);
|
|
|
2097 if (bp == NULL)
|
|
300
|
2098 goto endFAIL;
|
|
|
2099 if (round == 1)
|
|
339
|
2100 lp->sl_fbyts = bp;
|
|
|
2101 else if (round == 2)
|
|
|
2102 lp->sl_kbyts = bp;
|
|
300
|
2103 else
|
|
339
|
2104 lp->sl_pbyts = bp;
|
|
236
|
2105
|
|
300
|
2106 /* Allocate the index array. */
|
|
339
|
2107 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2108 if (ip == NULL)
|
|
300
|
2109 goto endFAIL;
|
|
|
2110 if (round == 1)
|
|
339
|
2111 lp->sl_fidxs = ip;
|
|
|
2112 else if (round == 2)
|
|
|
2113 lp->sl_kidxs = ip;
|
|
300
|
2114 else
|
|
339
|
2115 lp->sl_pidxs = ip;
|
|
300
|
2116
|
|
|
2117 /* Read the tree and store it in the array. */
|
|
339
|
2118 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2119 if (idx == -1)
|
|
300
|
2120 goto truncerr;
|
|
324
|
2121 if (idx < 0)
|
|
236
|
2122 goto formerr;
|
|
|
2123 }
|
|
300
|
2124 }
|
|
243
|
2125
|
|
310
|
2126 /* For a new file link it in the list of spell files. */
|
|
|
2127 if (old_lp == NULL)
|
|
|
2128 {
|
|
|
2129 lp->sl_next = first_lang;
|
|
|
2130 first_lang = lp;
|
|
|
2131 }
|
|
307
|
2132
|
|
255
|
2133 goto endOK;
|
|
|
2134
|
|
|
2135 endFAIL:
|
|
310
|
2136 if (lang != NULL)
|
|
|
2137 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2138 *lang = NUL;
|
|
|
2139 if (lp != NULL && old_lp == NULL)
|
|
323
|
2140 {
|
|
307
|
2141 slang_free(lp);
|
|
323
|
2142 lp = NULL;
|
|
|
2143 }
|
|
255
|
2144
|
|
|
2145 endOK:
|
|
236
|
2146 if (fd != NULL)
|
|
|
2147 fclose(fd);
|
|
|
2148 sourcing_name = save_sourcing_name;
|
|
|
2149 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2150
|
|
|
2151 return lp;
|
|
236
|
2152 }
|
|
|
2153
|
|
376
|
2154 /*
|
|
|
2155 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2156 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2157 * Returns NULL when the count is zero.
|
|
381
|
2158 * Sets "*errp" to FAIL when there is an error, OK otherwise.
|
|
376
|
2159 */
|
|
|
2160 static char_u *
|
|
|
2161 read_cnt_string(fd, cnt_bytes, errp)
|
|
|
2162 FILE *fd;
|
|
|
2163 int cnt_bytes;
|
|
|
2164 int *errp;
|
|
|
2165 {
|
|
|
2166 int cnt = 0;
|
|
|
2167 int i;
|
|
|
2168 char_u *str;
|
|
|
2169
|
|
|
2170 /* read the length bytes, MSB first */
|
|
|
2171 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2172 cnt = (cnt << 8) + getc(fd);
|
|
|
2173 if (cnt < 0)
|
|
|
2174 {
|
|
381
|
2175 EMSG(_(e_spell_trunc));
|
|
|
2176 *errp = FAIL;
|
|
376
|
2177 return NULL;
|
|
|
2178 }
|
|
|
2179
|
|
|
2180 /* allocate memory */
|
|
|
2181 str = alloc((unsigned)cnt + 1);
|
|
|
2182 if (str == NULL)
|
|
|
2183 {
|
|
381
|
2184 *errp = FAIL;
|
|
376
|
2185 return NULL;
|
|
|
2186 }
|
|
381
|
2187 *errp = OK;
|
|
376
|
2188
|
|
|
2189 /* Read the string. Doesn't check for truncated file. */
|
|
|
2190 for (i = 0; i < cnt; ++i)
|
|
|
2191 str[i] = getc(fd);
|
|
|
2192 str[i] = NUL;
|
|
|
2193
|
|
|
2194 return str;
|
|
|
2195 }
|
|
|
2196
|
|
381
|
2197 /*
|
|
|
2198 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
|
2199 * Returns FAIL when there is something wrong.
|
|
|
2200 */
|
|
|
2201 static int
|
|
|
2202 set_sofo(lp, from, to)
|
|
|
2203 slang_T *lp;
|
|
|
2204 char_u *from;
|
|
|
2205 char_u *to;
|
|
|
2206 {
|
|
|
2207 int i;
|
|
|
2208
|
|
|
2209 #ifdef FEAT_MBYTE
|
|
|
2210 garray_T *gap;
|
|
|
2211 char_u *s;
|
|
|
2212 char_u *p;
|
|
|
2213 int c;
|
|
|
2214 int *inp;
|
|
|
2215
|
|
|
2216 if (has_mbyte)
|
|
|
2217 {
|
|
|
2218 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2219 * characters. The index is the low byte of the character.
|
|
|
2220 * The list contains from-to pairs with a terminating NUL.
|
|
|
2221 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2222 gap = &lp->sl_sal;
|
|
|
2223 ga_init2(gap, sizeof(int *), 1);
|
|
|
2224 if (ga_grow(gap, 256) == FAIL)
|
|
|
2225 return FAIL;
|
|
|
2226 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2227 gap->ga_len = 256;
|
|
|
2228
|
|
|
2229 /* First count the number of items for each list. Temporarily use
|
|
|
2230 * sl_sal_first[] for this. */
|
|
|
2231 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2232 {
|
|
|
2233 c = mb_ptr2char_adv(&p);
|
|
|
2234 mb_ptr_adv(s);
|
|
|
2235 if (c >= 256)
|
|
|
2236 ++lp->sl_sal_first[c & 0xff];
|
|
|
2237 }
|
|
|
2238 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
|
2239 return FAIL;
|
|
|
2240
|
|
|
2241 /* Allocate the lists. */
|
|
|
2242 for (i = 0; i < 256; ++i)
|
|
|
2243 if (lp->sl_sal_first[i] > 0)
|
|
|
2244 {
|
|
|
2245 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2246 if (p == NULL)
|
|
|
2247 return FAIL;
|
|
|
2248 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2249 *(int *)p = 0;
|
|
|
2250 }
|
|
|
2251
|
|
|
2252 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
2253 * list. */
|
|
|
2254 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2255 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2256 {
|
|
|
2257 c = mb_ptr2char_adv(&p);
|
|
|
2258 i = mb_ptr2char_adv(&s);
|
|
|
2259 if (c >= 256)
|
|
|
2260 {
|
|
|
2261 /* Append the from-to chars at the end of the list with
|
|
|
2262 * the low byte. */
|
|
|
2263 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2264 while (*inp != 0)
|
|
|
2265 ++inp;
|
|
|
2266 *inp++ = c; /* from char */
|
|
|
2267 *inp++ = i; /* to char */
|
|
|
2268 *inp++ = NUL; /* NUL at the end */
|
|
|
2269 }
|
|
|
2270 else
|
|
|
2271 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2272 lp->sl_sal_first[c] = i;
|
|
|
2273 }
|
|
|
2274 }
|
|
|
2275 else
|
|
|
2276 #endif
|
|
|
2277 {
|
|
|
2278 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2279 if (STRLEN(from) != STRLEN(to))
|
|
|
2280 return FAIL;
|
|
|
2281
|
|
|
2282 for (i = 0; to[i] != NUL; ++i)
|
|
|
2283 lp->sl_sal_first[from[i]] = to[i];
|
|
|
2284 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2285 }
|
|
|
2286
|
|
|
2287 return OK;
|
|
|
2288 }
|
|
|
2289
|
|
|
2290 /*
|
|
|
2291 * Fill the first-index table for "lp".
|
|
|
2292 */
|
|
|
2293 static void
|
|
|
2294 set_sal_first(lp)
|
|
|
2295 slang_T *lp;
|
|
|
2296 {
|
|
|
2297 salfirst_T *sfirst;
|
|
|
2298 int i;
|
|
|
2299 salitem_T *smp;
|
|
|
2300 int c;
|
|
|
2301 garray_T *gap = &lp->sl_sal;
|
|
|
2302
|
|
|
2303 sfirst = lp->sl_sal_first;
|
|
|
2304 for (i = 0; i < 256; ++i)
|
|
|
2305 sfirst[i] = -1;
|
|
|
2306 smp = (salitem_T *)gap->ga_data;
|
|
|
2307 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2308 {
|
|
|
2309 #ifdef FEAT_MBYTE
|
|
|
2310 if (has_mbyte)
|
|
|
2311 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2312 * the character, for other encodings it should differ for most
|
|
|
2313 * characters. */
|
|
|
2314 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2315 else
|
|
|
2316 #endif
|
|
|
2317 c = *smp[i].sm_lead;
|
|
|
2318 if (sfirst[c] == -1)
|
|
|
2319 {
|
|
|
2320 sfirst[c] = i;
|
|
|
2321 #ifdef FEAT_MBYTE
|
|
|
2322 if (has_mbyte)
|
|
|
2323 {
|
|
|
2324 int n;
|
|
|
2325
|
|
|
2326 /* Make sure all entries with this byte are following each
|
|
|
2327 * other. Move the ones that are in the wrong position. Do
|
|
|
2328 * keep the same ordering! */
|
|
|
2329 while (i + 1 < gap->ga_len
|
|
|
2330 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
2331 /* Skip over entry with same index byte. */
|
|
|
2332 ++i;
|
|
|
2333
|
|
|
2334 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
2335 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
2336 {
|
|
|
2337 salitem_T tsal;
|
|
|
2338
|
|
|
2339 /* Move entry with same index byte after the entries
|
|
|
2340 * we already found. */
|
|
|
2341 ++i;
|
|
|
2342 --n;
|
|
|
2343 tsal = smp[i + n];
|
|
|
2344 mch_memmove(smp + i + 1, smp + i,
|
|
|
2345 sizeof(salitem_T) * n);
|
|
|
2346 smp[i] = tsal;
|
|
|
2347 }
|
|
|
2348 }
|
|
|
2349 #endif
|
|
|
2350 }
|
|
|
2351 }
|
|
|
2352 }
|
|
376
|
2353
|
|
372
|
2354 #ifdef FEAT_MBYTE
|
|
|
2355 /*
|
|
|
2356 * Turn a multi-byte string into a wide character string.
|
|
|
2357 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
2358 */
|
|
|
2359 static int *
|
|
|
2360 mb_str2wide(s)
|
|
|
2361 char_u *s;
|
|
|
2362 {
|
|
|
2363 int *res;
|
|
|
2364 char_u *p;
|
|
|
2365 int i = 0;
|
|
|
2366
|
|
|
2367 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
2368 if (res != NULL)
|
|
|
2369 {
|
|
|
2370 for (p = s; *p != NUL; )
|
|
|
2371 res[i++] = mb_ptr2char_adv(&p);
|
|
|
2372 res[i] = NUL;
|
|
|
2373 }
|
|
|
2374 return res;
|
|
|
2375 }
|
|
|
2376 #endif
|
|
|
2377
|
|
236
|
2378 /*
|
|
300
|
2379 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
2380 * "byts" and index array "idxs". Recursively read the children.
|
|
|
2381 *
|
|
346
|
2382 * NOTE: The code here must match put_node().
|
|
300
|
2383 *
|
|
|
2384 * Returns the index follosing the siblings.
|
|
|
2385 * Returns -1 if the file is shorter than expected.
|
|
|
2386 * Returns -2 if there is a format error.
|
|
236
|
2387 */
|
|
324
|
2388 static idx_T
|
|
339
|
2389 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
2390 FILE *fd;
|
|
|
2391 char_u *byts;
|
|
324
|
2392 idx_T *idxs;
|
|
300
|
2393 int maxidx; /* size of arrays */
|
|
324
|
2394 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
2395 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
2396 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
2397 {
|
|
300
|
2398 int len;
|
|
|
2399 int i;
|
|
|
2400 int n;
|
|
324
|
2401 idx_T idx = startidx;
|
|
300
|
2402 int c;
|
|
366
|
2403 int c2;
|
|
300
|
2404 #define SHARED_MASK 0x8000000
|
|
236
|
2405
|
|
300
|
2406 len = getc(fd); /* <siblingcount> */
|
|
|
2407 if (len <= 0)
|
|
|
2408 return -1;
|
|
|
2409
|
|
|
2410 if (startidx + len >= maxidx)
|
|
|
2411 return -2;
|
|
|
2412 byts[idx++] = len;
|
|
|
2413
|
|
|
2414 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
2415 for (i = 1; i <= len; ++i)
|
|
236
|
2416 {
|
|
300
|
2417 c = getc(fd); /* <byte> */
|
|
|
2418 if (c < 0)
|
|
|
2419 return -1;
|
|
|
2420 if (c <= BY_SPECIAL)
|
|
|
2421 {
|
|
366
|
2422 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
2423 {
|
|
|
2424 /* No flags, all regions. */
|
|
|
2425 idxs[idx] = 0;
|
|
|
2426 c = 0;
|
|
|
2427 }
|
|
366
|
2428 else if (c == BY_FLAGS || c == BY_NOFLAGS)
|
|
300
|
2429 {
|
|
339
|
2430 if (prefixtree)
|
|
|
2431 {
|
|
|
2432 /* Read the prefix ID and the condition nr. In idxs[]
|
|
|
2433 * store the prefix ID in the low byte, the condition
|
|
|
2434 * index shifted up 8 bits. */
|
|
366
|
2435 c2 = getc(fd); /* <prefixID> */
|
|
339
|
2436 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
2437 if (n >= maxprefcondnr)
|
|
|
2438 return -2;
|
|
366
|
2439 c2 += (n << 8);
|
|
|
2440 if (c == BY_NOFLAGS)
|
|
|
2441 c = c2;
|
|
|
2442 else
|
|
|
2443 c = c2 | WF_RAREPFX;
|
|
339
|
2444 }
|
|
|
2445 else
|
|
|
2446 {
|
|
|
2447 /* Read flags and optional region and prefix ID. In
|
|
|
2448 * idxs[] the flags go in the low byte, region above that
|
|
|
2449 * and prefix ID above the region. */
|
|
|
2450 c = getc(fd); /* <flags> */
|
|
|
2451 if (c & WF_REGION)
|
|
|
2452 c = (getc(fd) << 8) + c; /* <region> */
|
|
|
2453 if (c & WF_PFX)
|
|
|
2454 c = (getc(fd) << 16) + c; /* <prefixID> */
|
|
|
2455 }
|
|
|
2456
|
|
300
|
2457 idxs[idx] = c;
|
|
|
2458 c = 0;
|
|
|
2459 }
|
|
|
2460 else /* c == BY_INDEX */
|
|
|
2461 {
|
|
|
2462 /* <nodeidx> */
|
|
|
2463 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2464 if (n < 0 || n >= maxidx)
|
|
|
2465 return -2;
|
|
|
2466 idxs[idx] = n + SHARED_MASK;
|
|
|
2467 c = getc(fd); /* <xbyte> */
|
|
|
2468 }
|
|
|
2469 }
|
|
|
2470 byts[idx++] = c;
|
|
236
|
2471 }
|
|
|
2472
|
|
300
|
2473 /* Recursively read the children for non-shared siblings.
|
|
|
2474 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
2475 * remove SHARED_MASK) */
|
|
|
2476 for (i = 1; i <= len; ++i)
|
|
|
2477 if (byts[startidx + i] != 0)
|
|
|
2478 {
|
|
|
2479 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
2480 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
2481 else
|
|
|
2482 {
|
|
|
2483 idxs[startidx + i] = idx;
|
|
339
|
2484 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
2485 prefixtree, maxprefcondnr);
|
|
300
|
2486 if (idx < 0)
|
|
|
2487 break;
|
|
|
2488 }
|
|
|
2489 }
|
|
236
|
2490
|
|
300
|
2491 return idx;
|
|
236
|
2492 }
|
|
|
2493
|
|
|
2494 /*
|
|
|
2495 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
2496 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
2497 */
|
|
|
2498 char_u *
|
|
|
2499 did_set_spelllang(buf)
|
|
|
2500 buf_T *buf;
|
|
|
2501 {
|
|
|
2502 garray_T ga;
|
|
351
|
2503 char_u *splp;
|
|
236
|
2504 char_u *region;
|
|
355
|
2505 int filename;
|
|
236
|
2506 int region_mask;
|
|
|
2507 slang_T *lp;
|
|
|
2508 int c;
|
|
351
|
2509 char_u lang[MAXWLEN + 1];
|
|
323
|
2510 char_u spf_name[MAXPATHL];
|
|
351
|
2511 int len;
|
|
|
2512 char_u *p;
|
|
381
|
2513 int round;
|
|
385
|
2514 char_u *spf;
|
|
236
|
2515
|
|
|
2516 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
2517 clear_midword(buf);
|
|
236
|
2518
|
|
351
|
2519 /* loop over comma separated language names. */
|
|
|
2520 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
2521 {
|
|
|
2522 /* Get one language name. */
|
|
|
2523 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
2524
|
|
240
|
2525 region = NULL;
|
|
351
|
2526 len = STRLEN(lang);
|
|
355
|
2527
|
|
|
2528 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
2529 * If there is a region name let "region" point to it and remove it
|
|
|
2530 * from the name. */
|
|
|
2531 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
2532 {
|
|
355
|
2533 filename = TRUE;
|
|
|
2534
|
|
|
2535 /* Check if we loaded this language before. */
|
|
|
2536 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2537 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2538 break;
|
|
236
|
2539 }
|
|
355
|
2540 else
|
|
|
2541 {
|
|
|
2542 filename = FALSE;
|
|
|
2543 if (len > 3 && lang[len - 3] == '_')
|
|
|
2544 {
|
|
|
2545 region = lang + len - 2;
|
|
|
2546 len -= 3;
|
|
|
2547 lang[len] = NUL;
|
|
|
2548 }
|
|
|
2549
|
|
|
2550 /* Check if we loaded this language before. */
|
|
|
2551 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2552 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
2553 break;
|
|
|
2554 }
|
|
236
|
2555
|
|
351
|
2556 /* If not found try loading the language now. */
|
|
236
|
2557 if (lp == NULL)
|
|
355
|
2558 {
|
|
|
2559 if (filename)
|
|
|
2560 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
2561 else
|
|
|
2562 spell_load_lang(lang);
|
|
|
2563 }
|
|
236
|
2564
|
|
307
|
2565 /*
|
|
351
|
2566 * Loop over the languages, there can be several files for "lang".
|
|
307
|
2567 */
|
|
|
2568 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
2569 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
2570 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
2571 {
|
|
316
|
2572 region_mask = REGION_ALL;
|
|
355
|
2573 if (!filename && region != NULL)
|
|
236
|
2574 {
|
|
307
|
2575 /* find region in sl_regions */
|
|
|
2576 c = find_region(lp->sl_regions, region);
|
|
|
2577 if (c == REGION_ALL)
|
|
|
2578 {
|
|
316
|
2579 if (!lp->sl_add)
|
|
351
|
2580 smsg((char_u *)
|
|
|
2581 _("Warning: region %s not supported"),
|
|
|
2582 region);
|
|
307
|
2583 }
|
|
|
2584 else
|
|
|
2585 region_mask = 1 << c;
|
|
236
|
2586 }
|
|
307
|
2587
|
|
|
2588 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2589 {
|
|
|
2590 ga_clear(&ga);
|
|
|
2591 return e_outofmem;
|
|
|
2592 }
|
|
|
2593 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2594 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2595 ++ga.ga_len;
|
|
376
|
2596 use_midword(lp, buf);
|
|
385
|
2597 }
|
|
|
2598 }
|
|
|
2599
|
|
|
2600 /* round 0: load int_wordlist, if possible.
|
|
|
2601 * round 1: load first name in 'spellfile'.
|
|
|
2602 * round 2: load second name in 'spellfile.
|
|
|
2603 * etc. */
|
|
|
2604 spf = curbuf->b_p_spf;
|
|
|
2605 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
2606 {
|
|
|
2607 if (round == 0)
|
|
|
2608 {
|
|
|
2609 /* Internal wordlist, if there is one. */
|
|
|
2610 if (int_wordlist == NULL)
|
|
381
|
2611 continue;
|
|
385
|
2612 int_wordlist_spl(spf_name);
|
|
381
|
2613 }
|
|
|
2614 else
|
|
|
2615 {
|
|
385
|
2616 /* One entry in 'spellfile'. */
|
|
|
2617 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
2618 STRCAT(spf_name, ".spl");
|
|
|
2619
|
|
|
2620 /* If it was already found above then skip it. */
|
|
|
2621 for (c = 0; c < ga.ga_len; ++c)
|
|
|
2622 if (fullpathcmp(spf_name,
|
|
|
2623 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
2624 FALSE) == FPC_SAME)
|
|
|
2625 break;
|
|
|
2626 if (c < ga.ga_len)
|
|
381
|
2627 continue;
|
|
|
2628 }
|
|
|
2629
|
|
351
|
2630 /* Check if it was loaded already. */
|
|
323
|
2631 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2632 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2633 break;
|
|
|
2634 if (lp == NULL)
|
|
|
2635 {
|
|
351
|
2636 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
2637 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
2638 * use an arbitrary name. */
|
|
|
2639 if (round == 0)
|
|
|
2640 STRCPY(lang, "internal wordlist");
|
|
|
2641 else
|
|
|
2642 {
|
|
|
2643 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
2644 p = vim_strchr(lang, '.');
|
|
|
2645 if (p != NULL)
|
|
|
2646 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
2647 }
|
|
351
|
2648 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
2649 }
|
|
|
2650 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
2651 {
|
|
|
2652 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2653 LANGP_ENTRY(ga, ga.ga_len)->lp_region = REGION_ALL;
|
|
|
2654 ++ga.ga_len;
|
|
376
|
2655 use_midword(lp, buf);
|
|
323
|
2656 }
|
|
|
2657 }
|
|
|
2658
|
|
236
|
2659 /* Add a NULL entry to mark the end of the list. */
|
|
|
2660 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2661 {
|
|
|
2662 ga_clear(&ga);
|
|
|
2663 return e_outofmem;
|
|
|
2664 }
|
|
|
2665 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
2666 ++ga.ga_len;
|
|
|
2667
|
|
|
2668 /* Everything is fine, store the new b_langp value. */
|
|
|
2669 ga_clear(&buf->b_langp);
|
|
|
2670 buf->b_langp = ga;
|
|
|
2671
|
|
|
2672 return NULL;
|
|
|
2673 }
|
|
|
2674
|
|
|
2675 /*
|
|
376
|
2676 * Clear the midword characters for buffer "buf".
|
|
|
2677 */
|
|
|
2678 static void
|
|
|
2679 clear_midword(buf)
|
|
|
2680 buf_T *buf;
|
|
|
2681 {
|
|
|
2682 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
2683 #ifdef FEAT_MBYTE
|
|
|
2684 vim_free(buf->b_spell_ismw_mb);
|
|
|
2685 buf->b_spell_ismw_mb = NULL;
|
|
|
2686 #endif
|
|
|
2687 }
|
|
|
2688
|
|
|
2689 /*
|
|
|
2690 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
2691 * They add up to any currently used midword characters.
|
|
|
2692 */
|
|
|
2693 static void
|
|
|
2694 use_midword(lp, buf)
|
|
|
2695 slang_T *lp;
|
|
|
2696 buf_T *buf;
|
|
|
2697 {
|
|
|
2698 char_u *p;
|
|
|
2699
|
|
|
2700 for (p = lp->sl_midword; *p != NUL; )
|
|
|
2701 #ifdef FEAT_MBYTE
|
|
|
2702 if (has_mbyte)
|
|
|
2703 {
|
|
|
2704 int c, l, n;
|
|
|
2705 char_u *bp;
|
|
|
2706
|
|
|
2707 c = mb_ptr2char(p);
|
|
|
2708 l = mb_ptr2len_check(p);
|
|
|
2709 if (c < 256)
|
|
|
2710 buf->b_spell_ismw[c] = TRUE;
|
|
|
2711 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
2712 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
2713 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
2714 else
|
|
|
2715 {
|
|
|
2716 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
2717 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
2718 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
2719 if (bp != NULL)
|
|
|
2720 {
|
|
|
2721 vim_free(buf->b_spell_ismw_mb);
|
|
|
2722 buf->b_spell_ismw_mb = bp;
|
|
|
2723 vim_strncpy(bp + n, p, l);
|
|
|
2724 }
|
|
|
2725 }
|
|
|
2726 p += l;
|
|
|
2727 }
|
|
|
2728 else
|
|
|
2729 #endif
|
|
|
2730 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
2731 }
|
|
|
2732
|
|
|
2733 /*
|
|
236
|
2734 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
2735 * Each region is simply stored as the two characters of it's name.
|
|
381
|
2736 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
2737 */
|
|
|
2738 static int
|
|
|
2739 find_region(rp, region)
|
|
|
2740 char_u *rp;
|
|
|
2741 char_u *region;
|
|
|
2742 {
|
|
|
2743 int i;
|
|
|
2744
|
|
|
2745 for (i = 0; ; i += 2)
|
|
|
2746 {
|
|
|
2747 if (rp[i] == NUL)
|
|
|
2748 return REGION_ALL;
|
|
|
2749 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
2750 break;
|
|
|
2751 }
|
|
|
2752 return i / 2;
|
|
|
2753 }
|
|
|
2754
|
|
|
2755 /*
|
|
323
|
2756 * Return case type of word:
|
|
236
|
2757 * w word 0
|
|
300
|
2758 * Word WF_ONECAP
|
|
|
2759 * W WORD WF_ALLCAP
|
|
|
2760 * WoRd wOrd WF_KEEPCAP
|
|
236
|
2761 */
|
|
|
2762 static int
|
|
|
2763 captype(word, end)
|
|
|
2764 char_u *word;
|
|
323
|
2765 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
2766 {
|
|
|
2767 char_u *p;
|
|
|
2768 int c;
|
|
|
2769 int firstcap;
|
|
|
2770 int allcap;
|
|
|
2771 int past_second = FALSE; /* past second word char */
|
|
|
2772
|
|
|
2773 /* find first letter */
|
|
376
|
2774 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
2775 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
2776 return 0; /* only non-word characters, illegal word */
|
|
|
2777 #ifdef FEAT_MBYTE
|
|
310
|
2778 if (has_mbyte)
|
|
|
2779 c = mb_ptr2char_adv(&p);
|
|
|
2780 else
|
|
236
|
2781 #endif
|
|
310
|
2782 c = *p++;
|
|
324
|
2783 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
2784
|
|
|
2785 /*
|
|
|
2786 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
2787 * But a word with an upper char only at start is a ONECAP.
|
|
|
2788 */
|
|
323
|
2789 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
2790 if (spell_iswordp_nmw(p))
|
|
236
|
2791 {
|
|
|
2792 #ifdef FEAT_MBYTE
|
|
|
2793 c = mb_ptr2char(p);
|
|
|
2794 #else
|
|
|
2795 c = *p;
|
|
|
2796 #endif
|
|
324
|
2797 if (!SPELL_ISUPPER(c))
|
|
236
|
2798 {
|
|
|
2799 /* UUl -> KEEPCAP */
|
|
|
2800 if (past_second && allcap)
|
|
300
|
2801 return WF_KEEPCAP;
|
|
236
|
2802 allcap = FALSE;
|
|
|
2803 }
|
|
|
2804 else if (!allcap)
|
|
|
2805 /* UlU -> KEEPCAP */
|
|
300
|
2806 return WF_KEEPCAP;
|
|
236
|
2807 past_second = TRUE;
|
|
|
2808 }
|
|
|
2809
|
|
|
2810 if (allcap)
|
|
300
|
2811 return WF_ALLCAP;
|
|
236
|
2812 if (firstcap)
|
|
300
|
2813 return WF_ONECAP;
|
|
236
|
2814 return 0;
|
|
|
2815 }
|
|
|
2816
|
|
355
|
2817 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
2818 /*
|
|
|
2819 * Free all languages.
|
|
|
2820 */
|
|
|
2821 void
|
|
|
2822 spell_free_all()
|
|
|
2823 {
|
|
|
2824 slang_T *lp;
|
|
|
2825 buf_T *buf;
|
|
385
|
2826 char_u fname[MAXPATHL];
|
|
355
|
2827
|
|
|
2828 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2829 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2830 ga_clear(&buf->b_langp);
|
|
|
2831
|
|
|
2832 while (first_lang != NULL)
|
|
|
2833 {
|
|
|
2834 lp = first_lang;
|
|
|
2835 first_lang = lp->sl_next;
|
|
|
2836 slang_free(lp);
|
|
|
2837 }
|
|
366
|
2838
|
|
385
|
2839 if (int_wordlist != NULL)
|
|
|
2840 {
|
|
|
2841 /* Delete the internal wordlist and its .spl file */
|
|
|
2842 mch_remove(int_wordlist);
|
|
|
2843 int_wordlist_spl(fname);
|
|
|
2844 mch_remove(fname);
|
|
|
2845 vim_free(int_wordlist);
|
|
|
2846 int_wordlist = NULL;
|
|
381
|
2847 }
|
|
|
2848
|
|
366
|
2849 init_spell_chartab();
|
|
355
|
2850 }
|
|
|
2851 # endif
|
|
|
2852
|
|
236
|
2853 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
2854 /*
|
|
|
2855 * Clear all spelling tables and reload them.
|
|
307
|
2856 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
2857 */
|
|
|
2858 void
|
|
|
2859 spell_reload()
|
|
|
2860 {
|
|
|
2861 buf_T *buf;
|
|
316
|
2862 win_T *wp;
|
|
236
|
2863
|
|
358
|
2864 /* Initialize the table for spell_iswordp(). */
|
|
236
|
2865 init_spell_chartab();
|
|
|
2866
|
|
|
2867 /* Unload all allocated memory. */
|
|
355
|
2868 spell_free_all();
|
|
236
|
2869
|
|
|
2870 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2871 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2872 {
|
|
316
|
2873 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
2874 * window for this buffer in which 'spell' is set. */
|
|
236
|
2875 if (*buf->b_p_spl != NUL)
|
|
316
|
2876 {
|
|
|
2877 FOR_ALL_WINDOWS(wp)
|
|
|
2878 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
2879 {
|
|
|
2880 (void)did_set_spelllang(buf);
|
|
|
2881 # ifdef FEAT_WINDOWS
|
|
|
2882 break;
|
|
|
2883 # endif
|
|
|
2884 }
|
|
|
2885 }
|
|
236
|
2886 }
|
|
|
2887 }
|
|
|
2888 # endif
|
|
|
2889
|
|
310
|
2890 /*
|
|
|
2891 * Reload the spell file "fname" if it's loaded.
|
|
|
2892 */
|
|
|
2893 static void
|
|
323
|
2894 spell_reload_one(fname, added_word)
|
|
310
|
2895 char_u *fname;
|
|
323
|
2896 int added_word; /* invoked through "zg" */
|
|
310
|
2897 {
|
|
|
2898 slang_T *lp;
|
|
323
|
2899 int didit = FALSE;
|
|
310
|
2900
|
|
|
2901 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2902 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2903 {
|
|
|
2904 slang_clear(lp);
|
|
323
|
2905 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
2906 redraw_all_later(NOT_VALID);
|
|
323
|
2907 didit = TRUE;
|
|
310
|
2908 }
|
|
323
|
2909
|
|
|
2910 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
2911 * 'spelllang' to get it loaded. */
|
|
|
2912 if (added_word && !didit)
|
|
|
2913 did_set_spelllang(curbuf);
|
|
310
|
2914 }
|
|
|
2915
|
|
|
2916
|
|
236
|
2917 /*
|
|
|
2918 * Functions for ":mkspell".
|
|
|
2919 */
|
|
|
2920
|
|
300
|
2921 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
2922 and .dic file. */
|
|
|
2923 /*
|
|
|
2924 * Main structure to store the contents of a ".aff" file.
|
|
|
2925 */
|
|
|
2926 typedef struct afffile_S
|
|
|
2927 {
|
|
|
2928 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
310
|
2929 int af_rar; /* RAR ID for rare word */
|
|
|
2930 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
2931 int af_bad; /* BAD ID for banned word */
|
|
339
|
2932 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
2933 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
2934 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
2935 } afffile_T;
|
|
|
2936
|
|
|
2937 typedef struct affentry_S affentry_T;
|
|
|
2938 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
2939 struct affentry_S
|
|
|
2940 {
|
|
|
2941 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
2942 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
2943 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
2944 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
2945 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
366
|
2946 int ae_rare; /* rare affix */
|
|
300
|
2947 };
|
|
|
2948
|
|
|
2949 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
2950 typedef struct affheader_S
|
|
|
2951 {
|
|
|
2952 char_u ah_key[2]; /* key for hashtable == name of affix entry */
|
|
339
|
2953 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
2954 int ah_combine; /* suffix may combine with prefix */
|
|
|
2955 affentry_T *ah_first; /* first affix entry */
|
|
|
2956 } affheader_T;
|
|
|
2957
|
|
|
2958 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
2959
|
|
|
2960 /*
|
|
|
2961 * Structure that is used to store the items in the word tree. This avoids
|
|
|
2962 * the need to keep track of each allocated thing, it's freed all at once
|
|
|
2963 * after ":mkspell" is done.
|
|
|
2964 */
|
|
|
2965 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
2966 typedef struct sblock_S sblock_T;
|
|
|
2967 struct sblock_S
|
|
|
2968 {
|
|
|
2969 sblock_T *sb_next; /* next block in list */
|
|
|
2970 int sb_used; /* nr of bytes already in use */
|
|
|
2971 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
2972 };
|
|
|
2973
|
|
|
2974 /*
|
|
300
|
2975 * A node in the tree.
|
|
236
|
2976 */
|
|
300
|
2977 typedef struct wordnode_S wordnode_T;
|
|
|
2978 struct wordnode_S
|
|
236
|
2979 {
|
|
346
|
2980 union /* shared to save space */
|
|
|
2981 {
|
|
|
2982 char_u hashkey[6]; /* room for the hash key */
|
|
|
2983 int index; /* index in written nodes (valid after first
|
|
|
2984 round) */
|
|
|
2985 } wn_u1;
|
|
|
2986 union /* shared to save space */
|
|
|
2987 {
|
|
|
2988 wordnode_T *next; /* next node with same hash key */
|
|
|
2989 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
2990 } wn_u2;
|
|
300
|
2991 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
2992 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
2993 always sorted) */
|
|
|
2994 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
|
2995 char_u wn_flags; /* when wn_byte is NUL: WF_ flags */
|
|
339
|
2996 short wn_region; /* when wn_byte is NUL: region mask; for
|
|
|
2997 PREFIXTREE it's the prefcondnr */
|
|
|
2998 char_u wn_prefixID; /* supported/required prefix ID or 0 */
|
|
236
|
2999 };
|
|
|
3000
|
|
300
|
3001 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
3002
|
|
300
|
3003 /*
|
|
|
3004 * Info used while reading the spell files.
|
|
|
3005 */
|
|
|
3006 typedef struct spellinfo_S
|
|
249
|
3007 {
|
|
300
|
3008 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
3009 long si_foldwcount; /* nr of words in si_foldroot */
|
|
300
|
3010 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
3011 long si_keepwcount; /* nr of words in si_keeproot */
|
|
339
|
3012 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
300
|
3013 sblock_T *si_blocks; /* memory blocks used */
|
|
|
3014 int si_ascii; /* handling only ASCII words */
|
|
310
|
3015 int si_add; /* addition file */
|
|
351
|
3016 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
3017 int si_region; /* region mask */
|
|
|
3018 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
3019 int si_memtot; /* runtime memory used */
|
|
310
|
3020 int si_verbose; /* verbose messages */
|
|
316
|
3021 int si_region_count; /* number of regions supported (1 when there
|
|
|
3022 are no regions) */
|
|
|
3023 char_u si_region_name[16]; /* region names (if count > 1) */
|
|
323
|
3024
|
|
|
3025 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
3026 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
3027 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
3028 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
3029 int si_followup; /* soundsalike: ? */
|
|
|
3030 int si_collapse; /* soundsalike: ? */
|
|
|
3031 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
3032 garray_T si_map; /* MAP info concatenated */
|
|
366
|
3033 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
339
|
3034 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
3035 * prefixes, each stored as a string */
|
|
|
3036 int si_newID; /* current value for ah_newID */
|
|
300
|
3037 } spellinfo_T;
|
|
249
|
3038
|
|
300
|
3039 static afffile_T *spell_read_aff __ARGS((char_u *fname, spellinfo_T *spin));
|
|
339
|
3040 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
3041 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
3042 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
3043 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
3044 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
|
3045 static int spell_read_dic __ARGS((char_u *fname, spellinfo_T *spin, afffile_T *affile));
|
|
339
|
3046 static char_u *get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, sblock_T **blp));
|
|
|
3047 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
|
3048 static int spell_read_wordfile __ARGS((char_u *fname, spellinfo_T *spin));
|
|
|
3049 static void *getroom __ARGS((sblock_T **blp, size_t len));
|
|
|
3050 static char_u *getroom_save __ARGS((sblock_T **blp, char_u *s));
|
|
|
3051 static void free_blocks __ARGS((sblock_T *bl));
|
|
|
3052 static wordnode_T *wordtree_alloc __ARGS((sblock_T **blp));
|
|
339
|
3053 static int store_word __ARGS((char_u *word, spellinfo_T *spin, int flags, int region, char_u *pfxlist));
|
|
|
3054 static int tree_add_word __ARGS((char_u *word, wordnode_T *tree, int flags, int region, int prefixID, sblock_T **blp));
|
|
310
|
3055 static void wordtree_compress __ARGS((wordnode_T *root, spellinfo_T *spin));
|
|
300
|
3056 static int node_compress __ARGS((wordnode_T *node, hashtab_T *ht, int *tot));
|
|
|
3057 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
316
|
3058 static void write_vim_spell __ARGS((char_u *fname, spellinfo_T *spin));
|
|
346
|
3059 static void clear_node __ARGS((wordnode_T *node));
|
|
|
3060 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
3061 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3062 static void init_spellfile __ARGS((void));
|
|
236
|
3063
|
|
|
3064 /*
|
|
323
|
3065 * Read the affix file "fname".
|
|
316
|
3066 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
3067 */
|
|
|
3068 static afffile_T *
|
|
300
|
3069 spell_read_aff(fname, spin)
|
|
236
|
3070 char_u *fname;
|
|
300
|
3071 spellinfo_T *spin;
|
|
236
|
3072 {
|
|
|
3073 FILE *fd;
|
|
|
3074 afffile_T *aff;
|
|
|
3075 char_u rline[MAXLINELEN];
|
|
|
3076 char_u *line;
|
|
|
3077 char_u *pc = NULL;
|
|
334
|
3078 #define MAXITEMCNT 7
|
|
|
3079 char_u *(items[MAXITEMCNT]);
|
|
236
|
3080 int itemcnt;
|
|
|
3081 char_u *p;
|
|
|
3082 int lnum = 0;
|
|
|
3083 affheader_T *cur_aff = NULL;
|
|
|
3084 int aff_todo = 0;
|
|
|
3085 hashtab_T *tp;
|
|
255
|
3086 char_u *low = NULL;
|
|
|
3087 char_u *fol = NULL;
|
|
|
3088 char_u *upp = NULL;
|
|
307
|
3089 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
3090 int do_rep;
|
|
|
3091 int do_sal;
|
|
|
3092 int do_map;
|
|
366
|
3093 int do_midword;
|
|
375
|
3094 int do_sofo;
|
|
323
|
3095 int found_map = FALSE;
|
|
324
|
3096 hashitem_T *hi;
|
|
236
|
3097
|
|
300
|
3098 /*
|
|
|
3099 * Open the file.
|
|
|
3100 */
|
|
310
|
3101 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3102 if (fd == NULL)
|
|
|
3103 {
|
|
|
3104 EMSG2(_(e_notopen), fname);
|
|
|
3105 return NULL;
|
|
|
3106 }
|
|
|
3107
|
|
310
|
3108 if (spin->si_verbose || p_verbose > 2)
|
|
|
3109 {
|
|
|
3110 if (!spin->si_verbose)
|
|
|
3111 verbose_enter();
|
|
366
|
3112 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
3113 out_flush();
|
|
|
3114 if (!spin->si_verbose)
|
|
|
3115 verbose_leave();
|
|
|
3116 }
|
|
236
|
3117
|
|
323
|
3118 /* Only do REP lines when not done in another .aff file already. */
|
|
|
3119 do_rep = spin->si_rep.ga_len == 0;
|
|
|
3120
|
|
|
3121 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
3122 do_sal = spin->si_sal.ga_len == 0;
|
|
|
3123
|
|
|
3124 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
3125 do_map = spin->si_map.ga_len == 0;
|
|
|
3126
|
|
366
|
3127 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
3128 do_midword = spin->si_midword == NULL;
|
|
|
3129
|
|
375
|
3130 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
3131 do_sofo = spin->si_sofofr == NULL;
|
|
|
3132
|
|
300
|
3133 /*
|
|
|
3134 * Allocate and init the afffile_T structure.
|
|
|
3135 */
|
|
|
3136 aff = (afffile_T *)getroom(&spin->si_blocks, sizeof(afffile_T));
|
|
236
|
3137 if (aff == NULL)
|
|
|
3138 return NULL;
|
|
|
3139 hash_init(&aff->af_pref);
|
|
|
3140 hash_init(&aff->af_suff);
|
|
|
3141
|
|
|
3142 /*
|
|
|
3143 * Read all the lines in the file one by one.
|
|
|
3144 */
|
|
255
|
3145 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
3146 {
|
|
255
|
3147 line_breakcheck();
|
|
236
|
3148 ++lnum;
|
|
|
3149
|
|
|
3150 /* Skip comment lines. */
|
|
|
3151 if (*rline == '#')
|
|
|
3152 continue;
|
|
|
3153
|
|
|
3154 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
3155 vim_free(pc);
|
|
310
|
3156 #ifdef FEAT_MBYTE
|
|
300
|
3157 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3158 {
|
|
300
|
3159 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
3160 if (pc == NULL)
|
|
|
3161 {
|
|
|
3162 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3163 fname, lnum, rline);
|
|
|
3164 continue;
|
|
|
3165 }
|
|
236
|
3166 line = pc;
|
|
|
3167 }
|
|
|
3168 else
|
|
310
|
3169 #endif
|
|
236
|
3170 {
|
|
|
3171 pc = NULL;
|
|
|
3172 line = rline;
|
|
|
3173 }
|
|
|
3174
|
|
|
3175 /* Split the line up in white separated items. Put a NUL after each
|
|
|
3176 * item. */
|
|
|
3177 itemcnt = 0;
|
|
|
3178 for (p = line; ; )
|
|
|
3179 {
|
|
|
3180 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
3181 ++p;
|
|
|
3182 if (*p == NUL)
|
|
|
3183 break;
|
|
334
|
3184 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
3185 break;
|
|
236
|
3186 items[itemcnt++] = p;
|
|
300
|
3187 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
3188 ++p;
|
|
|
3189 if (*p == NUL)
|
|
|
3190 break;
|
|
|
3191 *p++ = NUL;
|
|
|
3192 }
|
|
|
3193
|
|
|
3194 /* Handle non-empty lines. */
|
|
|
3195 if (itemcnt > 0)
|
|
|
3196 {
|
|
|
3197 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
3198 && aff->af_enc == NULL)
|
|
|
3199 {
|
|
310
|
3200 #ifdef FEAT_MBYTE
|
|
300
|
3201 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
3202 aff->af_enc = enc_canonize(items[1]);
|
|
|
3203 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
3204 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
3205 p_enc) == FAIL)
|
|
|
3206 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
3207 fname, aff->af_enc, p_enc);
|
|
375
|
3208 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3209 #else
|
|
|
3210 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3211 #endif
|
|
236
|
3212 }
|
|
366
|
3213 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
3214 {
|
|
|
3215 if (do_midword)
|
|
|
3216 spin->si_midword = vim_strsave(items[1]);
|
|
|
3217 }
|
|
302
|
3218 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
3219 {
|
|
323
|
3220 /* ignored, we always split */
|
|
302
|
3221 }
|
|
323
|
3222 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
3223 {
|
|
323
|
3224 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
3225 }
|
|
307
|
3226 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
3227 && aff->af_rar == 0)
|
|
|
3228 {
|
|
|
3229 aff->af_rar = items[1][0];
|
|
|
3230 if (items[1][1] != NUL)
|
|
|
3231 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3232 }
|
|
310
|
3233 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
3234 && aff->af_kep == 0)
|
|
307
|
3235 {
|
|
310
|
3236 aff->af_kep = items[1][0];
|
|
307
|
3237 if (items[1][1] != NUL)
|
|
|
3238 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3239 }
|
|
346
|
3240 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
3241 && aff->af_bad == 0)
|
|
|
3242 {
|
|
|
3243 aff->af_bad = items[1][0];
|
|
|
3244 if (items[1][1] != NUL)
|
|
|
3245 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3246 }
|
|
339
|
3247 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
3248 {
|
|
|
3249 aff->af_pfxpostpone = TRUE;
|
|
|
3250 }
|
|
236
|
3251 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3252 || STRCMP(items[0], "SFX") == 0)
|
|
|
3253 && aff_todo == 0
|
|
334
|
3254 && itemcnt >= 4)
|
|
236
|
3255 {
|
|
334
|
3256 /* Myspell allows extra text after the item, but that might
|
|
|
3257 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
3258 if (itemcnt > 4 && *items[4] != '#')
|
|
|
3259 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
3260 fname, lnum, items[4]);
|
|
|
3261
|
|
236
|
3262 /* New affix letter. */
|
|
300
|
3263 cur_aff = (affheader_T *)getroom(&spin->si_blocks,
|
|
|
3264 sizeof(affheader_T));
|
|
236
|
3265 if (cur_aff == NULL)
|
|
|
3266 break;
|
|
339
|
3267 cur_aff->ah_key[0] = *items[1]; /* TODO: multi-byte? */
|
|
236
|
3268 cur_aff->ah_key[1] = NUL;
|
|
|
3269 if (items[1][1] != NUL)
|
|
307
|
3270 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
3271 if (*items[2] == 'Y')
|
|
|
3272 cur_aff->ah_combine = TRUE;
|
|
300
|
3273 else if (*items[2] != 'N')
|
|
236
|
3274 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
3275 fname, lnum, items[2]);
|
|
339
|
3276
|
|
236
|
3277 if (*items[0] == 'P')
|
|
339
|
3278 {
|
|
236
|
3279 tp = &aff->af_pref;
|
|
339
|
3280 /* Use a new number in the .spl file later, to be able to
|
|
|
3281 * handle multiple .aff files. */
|
|
|
3282 if (aff->af_pfxpostpone)
|
|
344
|
3283 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
3284 }
|
|
236
|
3285 else
|
|
|
3286 tp = &aff->af_suff;
|
|
300
|
3287 aff_todo = atoi((char *)items[3]);
|
|
324
|
3288 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
3289 if (!HASHITEM_EMPTY(hi))
|
|
300
|
3290 {
|
|
236
|
3291 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
3292 fname, lnum, items[1]);
|
|
300
|
3293 aff_todo = 0;
|
|
|
3294 }
|
|
236
|
3295 else
|
|
|
3296 hash_add(tp, cur_aff->ah_key);
|
|
|
3297 }
|
|
|
3298 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3299 || STRCMP(items[0], "SFX") == 0)
|
|
|
3300 && aff_todo > 0
|
|
|
3301 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
3302 && itemcnt >= 5)
|
|
236
|
3303 {
|
|
|
3304 affentry_T *aff_entry;
|
|
366
|
3305 int rare = FALSE;
|
|
|
3306 int lasti = 5;
|
|
|
3307
|
|
|
3308 /* Check for "rare" after the other info. */
|
|
|
3309 if (itemcnt > 5 && STRICMP(items[5], "rare") == 0)
|
|
|
3310 {
|
|
|
3311 rare = TRUE;
|
|
|
3312 lasti = 6;
|
|
|
3313 }
|
|
236
|
3314
|
|
334
|
3315 /* Myspell allows extra text after the item, but that might
|
|
|
3316 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
3317 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
3318 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
366
|
3319 fname, lnum, items[lasti]);
|
|
334
|
3320
|
|
236
|
3321 /* New item for an affix letter. */
|
|
|
3322 --aff_todo;
|
|
300
|
3323 aff_entry = (affentry_T *)getroom(&spin->si_blocks,
|
|
|
3324 sizeof(affentry_T));
|
|
236
|
3325 if (aff_entry == NULL)
|
|
|
3326 break;
|
|
366
|
3327 aff_entry->ae_rare = rare;
|
|
240
|
3328
|
|
236
|
3329 if (STRCMP(items[2], "0") != 0)
|
|
300
|
3330 aff_entry->ae_chop = getroom_save(&spin->si_blocks,
|
|
|
3331 items[2]);
|
|
236
|
3332 if (STRCMP(items[3], "0") != 0)
|
|
300
|
3333 aff_entry->ae_add = getroom_save(&spin->si_blocks,
|
|
|
3334 items[3]);
|
|
236
|
3335
|
|
300
|
3336 /* Don't use an affix entry with non-ASCII characters when
|
|
|
3337 * "spin->si_ascii" is TRUE. */
|
|
|
3338 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
3339 || has_non_ascii(aff_entry->ae_add)))
|
|
|
3340 {
|
|
|
3341 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
3342 cur_aff->ah_first = aff_entry;
|
|
300
|
3343
|
|
|
3344 if (STRCMP(items[4], ".") != 0)
|
|
|
3345 {
|
|
|
3346 char_u buf[MAXLINELEN];
|
|
|
3347
|
|
|
3348 aff_entry->ae_cond = getroom_save(&spin->si_blocks,
|
|
|
3349 items[4]);
|
|
|
3350 if (*items[0] == 'P')
|
|
|
3351 sprintf((char *)buf, "^%s", items[4]);
|
|
|
3352 else
|
|
|
3353 sprintf((char *)buf, "%s$", items[4]);
|
|
|
3354 aff_entry->ae_prog = vim_regcomp(buf,
|
|
|
3355 RE_MAGIC + RE_STRING);
|
|
|
3356 }
|
|
339
|
3357
|
|
|
3358 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
3359 * for the condition. Use an existing one if possible. */
|
|
|
3360 if (*items[0] == 'P' && aff->af_pfxpostpone
|
|
|
3361 && aff_entry->ae_chop == NULL)
|
|
|
3362 {
|
|
|
3363 int idx;
|
|
|
3364 char_u **pp;
|
|
|
3365
|
|
|
3366 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
3367 --idx)
|
|
|
3368 {
|
|
|
3369 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
3370 if (str_equal(p, aff_entry->ae_cond))
|
|
|
3371 break;
|
|
|
3372 }
|
|
|
3373 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
3374 {
|
|
|
3375 /* Not found, add a new condition. */
|
|
|
3376 idx = spin->si_prefcond.ga_len++;
|
|
|
3377 pp = ((char_u **)spin->si_prefcond.ga_data) + idx;
|
|
|
3378 if (aff_entry->ae_cond == NULL)
|
|
|
3379 *pp = NULL;
|
|
|
3380 else
|
|
|
3381 *pp = getroom_save(&spin->si_blocks,
|
|
|
3382 aff_entry->ae_cond);
|
|
|
3383 }
|
|
|
3384
|
|
|
3385 /* Add the prefix to the prefix tree. */
|
|
|
3386 if (aff_entry->ae_add == NULL)
|
|
|
3387 p = (char_u *)"";
|
|
|
3388 else
|
|
|
3389 p = aff_entry->ae_add;
|
|
366
|
3390 tree_add_word(p, spin->si_prefroot, rare ? -2 : -1,
|
|
|
3391 idx, cur_aff->ah_newID, &spin->si_blocks);
|
|
339
|
3392 }
|
|
240
|
3393 }
|
|
236
|
3394 }
|
|
255
|
3395 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
3396 {
|
|
|
3397 if (fol != NULL)
|
|
|
3398 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
3399 fname, lnum);
|
|
|
3400 else
|
|
|
3401 fol = vim_strsave(items[1]);
|
|
|
3402 }
|
|
|
3403 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
3404 {
|
|
|
3405 if (low != NULL)
|
|
|
3406 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
3407 fname, lnum);
|
|
|
3408 else
|
|
|
3409 low = vim_strsave(items[1]);
|
|
|
3410 }
|
|
|
3411 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
3412 {
|
|
|
3413 if (upp != NULL)
|
|
|
3414 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
3415 fname, lnum);
|
|
|
3416 else
|
|
|
3417 upp = vim_strsave(items[1]);
|
|
|
3418 }
|
|
236
|
3419 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
3420 {
|
|
236
|
3421 /* Ignore REP count */;
|
|
323
|
3422 if (!isdigit(*items[1]))
|
|
|
3423 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
3424 fname, lnum);
|
|
|
3425 }
|
|
236
|
3426 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 3)
|
|
|
3427 {
|
|
|
3428 /* REP item */
|
|
323
|
3429 if (do_rep)
|
|
|
3430 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
3431 }
|
|
|
3432 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
3433 {
|
|
|
3434 /* MAP item or count */
|
|
|
3435 if (!found_map)
|
|
|
3436 {
|
|
|
3437 /* First line contains the count. */
|
|
|
3438 found_map = TRUE;
|
|
|
3439 if (!isdigit(*items[1]))
|
|
|
3440 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
3441 fname, lnum);
|
|
|
3442 }
|
|
|
3443 else if (do_map)
|
|
|
3444 {
|
|
346
|
3445 int c;
|
|
|
3446
|
|
|
3447 /* Check that every character appears only once. */
|
|
|
3448 for (p = items[1]; *p != NUL; )
|
|
|
3449 {
|
|
|
3450 #ifdef FEAT_MBYTE
|
|
|
3451 c = mb_ptr2char_adv(&p);
|
|
|
3452 #else
|
|
|
3453 c = *p++;
|
|
|
3454 #endif
|
|
|
3455 if ((spin->si_map.ga_len > 0
|
|
|
3456 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
3457 != NULL)
|
|
|
3458 || vim_strchr(p, c) != NULL)
|
|
|
3459 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
3460 fname, lnum);
|
|
|
3461 }
|
|
|
3462
|
|
323
|
3463 /* We simply concatenate all the MAP strings, separated by
|
|
|
3464 * slashes. */
|
|
|
3465 ga_concat(&spin->si_map, items[1]);
|
|
|
3466 ga_append(&spin->si_map, '/');
|
|
|
3467 }
|
|
|
3468 }
|
|
|
3469 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
3470 {
|
|
|
3471 if (do_sal)
|
|
|
3472 {
|
|
|
3473 /* SAL item (sounds-a-like)
|
|
|
3474 * Either one of the known keys or a from-to pair. */
|
|
|
3475 if (STRCMP(items[1], "followup") == 0)
|
|
|
3476 spin->si_followup = sal_to_bool(items[2]);
|
|
|
3477 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
3478 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
3479 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
3480 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
3481 else
|
|
|
3482 /* when "to" is "_" it means empty */
|
|
|
3483 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
3484 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
3485 : items[2]);
|
|
|
3486 }
|
|
236
|
3487 }
|
|
375
|
3488 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
3489 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
3490 {
|
|
|
3491 if (do_sofo)
|
|
|
3492 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
3493 }
|
|
|
3494 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
3495 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
3496 {
|
|
|
3497 if (do_sofo)
|
|
|
3498 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
3499 }
|
|
300
|
3500 else
|
|
236
|
3501 smsg((char_u *)_("Unrecognized item in %s line %d: %s"),
|
|
|
3502 fname, lnum, items[0]);
|
|
|
3503 }
|
|
|
3504 }
|
|
|
3505
|
|
375
|
3506 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
3507 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
3508 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
3509 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
3510 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
3511
|
|
255
|
3512 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
3513 {
|
|
351
|
3514 if (spin->si_clear_chartab)
|
|
|
3515 {
|
|
|
3516 /* Clear the char type tables, don't want to use any of the
|
|
|
3517 * currently used spell properties. */
|
|
|
3518 init_spell_chartab();
|
|
|
3519 spin->si_clear_chartab = FALSE;
|
|
|
3520 }
|
|
|
3521
|
|
316
|
3522 /*
|
|
|
3523 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
3524 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
3525 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
3526 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
3527 */
|
|
|
3528 if (!spin->si_ascii
|
|
|
3529 #ifdef FEAT_MBYTE
|
|
|
3530 && !enc_utf8
|
|
|
3531 #endif
|
|
|
3532 )
|
|
260
|
3533 {
|
|
|
3534 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
3535 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
3536 else
|
|
316
|
3537 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
3538 }
|
|
255
|
3539
|
|
|
3540 vim_free(fol);
|
|
|
3541 vim_free(low);
|
|
|
3542 vim_free(upp);
|
|
|
3543 }
|
|
|
3544
|
|
236
|
3545 vim_free(pc);
|
|
|
3546 fclose(fd);
|
|
|
3547 return aff;
|
|
|
3548 }
|
|
|
3549
|
|
|
3550 /*
|
|
339
|
3551 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
3552 * NULL as equal.
|
|
|
3553 */
|
|
|
3554 static int
|
|
|
3555 str_equal(s1, s2)
|
|
|
3556 char_u *s1;
|
|
|
3557 char_u *s2;
|
|
|
3558 {
|
|
|
3559 if (s1 == NULL || s2 == NULL)
|
|
|
3560 return s1 == s2;
|
|
|
3561 return STRCMP(s1, s2) == 0;
|
|
|
3562 }
|
|
|
3563
|
|
|
3564 /*
|
|
323
|
3565 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
3566 * They are stored case-folded.
|
|
|
3567 */
|
|
|
3568 static void
|
|
|
3569 add_fromto(spin, gap, from, to)
|
|
|
3570 spellinfo_T *spin;
|
|
|
3571 garray_T *gap;
|
|
|
3572 char_u *from;
|
|
|
3573 char_u *to;
|
|
|
3574 {
|
|
|
3575 fromto_T *ftp;
|
|
|
3576 char_u word[MAXWLEN];
|
|
|
3577
|
|
|
3578 if (ga_grow(gap, 1) == OK)
|
|
|
3579 {
|
|
|
3580 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
3581 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
|
3582 ftp->ft_from = getroom_save(&spin->si_blocks, word);
|
|
|
3583 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
|
3584 ftp->ft_to = getroom_save(&spin->si_blocks, word);
|
|
|
3585 ++gap->ga_len;
|
|
|
3586 }
|
|
|
3587 }
|
|
|
3588
|
|
|
3589 /*
|
|
|
3590 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
3591 */
|
|
|
3592 static int
|
|
|
3593 sal_to_bool(s)
|
|
|
3594 char_u *s;
|
|
|
3595 {
|
|
|
3596 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
3597 }
|
|
|
3598
|
|
|
3599 /*
|
|
240
|
3600 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
3601 * When "s" is NULL FALSE is returned.
|
|
|
3602 */
|
|
|
3603 static int
|
|
|
3604 has_non_ascii(s)
|
|
|
3605 char_u *s;
|
|
|
3606 {
|
|
|
3607 char_u *p;
|
|
|
3608
|
|
|
3609 if (s != NULL)
|
|
|
3610 for (p = s; *p != NUL; ++p)
|
|
|
3611 if (*p >= 128)
|
|
|
3612 return TRUE;
|
|
|
3613 return FALSE;
|
|
|
3614 }
|
|
|
3615
|
|
|
3616 /*
|
|
236
|
3617 * Free the structure filled by spell_read_aff().
|
|
|
3618 */
|
|
|
3619 static void
|
|
|
3620 spell_free_aff(aff)
|
|
|
3621 afffile_T *aff;
|
|
|
3622 {
|
|
|
3623 hashtab_T *ht;
|
|
|
3624 hashitem_T *hi;
|
|
|
3625 int todo;
|
|
|
3626 affheader_T *ah;
|
|
300
|
3627 affentry_T *ae;
|
|
236
|
3628
|
|
|
3629 vim_free(aff->af_enc);
|
|
|
3630
|
|
339
|
3631 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
3632 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
3633 {
|
|
|
3634 todo = ht->ht_used;
|
|
|
3635 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
3636 {
|
|
|
3637 if (!HASHITEM_EMPTY(hi))
|
|
|
3638 {
|
|
|
3639 --todo;
|
|
|
3640 ah = HI2AH(hi);
|
|
300
|
3641 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
3642 vim_free(ae->ae_prog);
|
|
236
|
3643 }
|
|
|
3644 }
|
|
|
3645 if (ht == &aff->af_suff)
|
|
|
3646 break;
|
|
|
3647 }
|
|
300
|
3648
|
|
236
|
3649 hash_clear(&aff->af_pref);
|
|
|
3650 hash_clear(&aff->af_suff);
|
|
|
3651 }
|
|
|
3652
|
|
|
3653 /*
|
|
300
|
3654 * Read dictionary file "fname".
|
|
236
|
3655 * Returns OK or FAIL;
|
|
|
3656 */
|
|
|
3657 static int
|
|
300
|
3658 spell_read_dic(fname, spin, affile)
|
|
236
|
3659 char_u *fname;
|
|
300
|
3660 spellinfo_T *spin;
|
|
|
3661 afffile_T *affile;
|
|
236
|
3662 {
|
|
300
|
3663 hashtab_T ht;
|
|
236
|
3664 char_u line[MAXLINELEN];
|
|
300
|
3665 char_u *afflist;
|
|
339
|
3666 char_u *pfxlist;
|
|
300
|
3667 char_u *dw;
|
|
236
|
3668 char_u *pc;
|
|
|
3669 char_u *w;
|
|
|
3670 int l;
|
|
|
3671 hash_T hash;
|
|
|
3672 hashitem_T *hi;
|
|
|
3673 FILE *fd;
|
|
|
3674 int lnum = 1;
|
|
300
|
3675 int non_ascii = 0;
|
|
|
3676 int retval = OK;
|
|
|
3677 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
3678 int flags;
|
|
236
|
3679
|
|
300
|
3680 /*
|
|
|
3681 * Open the file.
|
|
|
3682 */
|
|
310
|
3683 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3684 if (fd == NULL)
|
|
|
3685 {
|
|
|
3686 EMSG2(_(e_notopen), fname);
|
|
|
3687 return FAIL;
|
|
|
3688 }
|
|
|
3689
|
|
300
|
3690 /* The hashtable is only used to detect duplicated words. */
|
|
|
3691 hash_init(&ht);
|
|
|
3692
|
|
334
|
3693 spin->si_foldwcount = 0;
|
|
|
3694 spin->si_keepwcount = 0;
|
|
|
3695
|
|
310
|
3696 if (spin->si_verbose || p_verbose > 2)
|
|
|
3697 {
|
|
|
3698 if (!spin->si_verbose)
|
|
|
3699 verbose_enter();
|
|
366
|
3700 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
3701 out_flush();
|
|
|
3702 if (!spin->si_verbose)
|
|
|
3703 verbose_leave();
|
|
|
3704 }
|
|
236
|
3705
|
|
|
3706 /* Read and ignore the first line: word count. */
|
|
|
3707 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
3708 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
3709 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
3710
|
|
|
3711 /*
|
|
|
3712 * Read all the lines in the file one by one.
|
|
|
3713 * The words are converted to 'encoding' here, before being added to
|
|
|
3714 * the hashtable.
|
|
|
3715 */
|
|
255
|
3716 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
3717 {
|
|
255
|
3718 line_breakcheck();
|
|
236
|
3719 ++lnum;
|
|
351
|
3720 if (line[0] == '#')
|
|
|
3721 continue; /* comment line */
|
|
236
|
3722
|
|
300
|
3723 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
3724 * the word is kept to allow e.g., "et al.". */
|
|
236
|
3725 l = STRLEN(line);
|
|
|
3726 while (l > 0 && line[l - 1] <= ' ')
|
|
|
3727 --l;
|
|
|
3728 if (l == 0)
|
|
|
3729 continue; /* empty line */
|
|
|
3730 line[l] = NUL;
|
|
|
3731
|
|
|
3732 /* Find the optional affix names. */
|
|
300
|
3733 afflist = vim_strchr(line, '/');
|
|
|
3734 if (afflist != NULL)
|
|
|
3735 *afflist++ = NUL;
|
|
236
|
3736
|
|
300
|
3737 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
3738 if (spin->si_ascii && has_non_ascii(line))
|
|
|
3739 {
|
|
|
3740 ++non_ascii;
|
|
240
|
3741 continue;
|
|
300
|
3742 }
|
|
240
|
3743
|
|
310
|
3744 #ifdef FEAT_MBYTE
|
|
236
|
3745 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
3746 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3747 {
|
|
300
|
3748 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
3749 if (pc == NULL)
|
|
|
3750 {
|
|
|
3751 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3752 fname, lnum, line);
|
|
|
3753 continue;
|
|
|
3754 }
|
|
236
|
3755 w = pc;
|
|
|
3756 }
|
|
|
3757 else
|
|
310
|
3758 #endif
|
|
236
|
3759 {
|
|
|
3760 pc = NULL;
|
|
|
3761 w = line;
|
|
|
3762 }
|
|
|
3763
|
|
339
|
3764 /* This takes time, print a message now and then. */
|
|
|
3765 if (spin->si_verbose && (lnum & 0x3ff) == 0)
|
|
|
3766 {
|
|
|
3767 vim_snprintf((char *)message, sizeof(message),
|
|
|
3768 _("line %6d, word %6d - %s"),
|
|
|
3769 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
3770 msg_start();
|
|
|
3771 msg_puts_long_attr(message, 0);
|
|
|
3772 msg_clr_eos();
|
|
|
3773 msg_didout = FALSE;
|
|
|
3774 msg_col = 0;
|
|
|
3775 out_flush();
|
|
|
3776 }
|
|
|
3777
|
|
300
|
3778 /* Store the word in the hashtable to be able to find duplicates. */
|
|
|
3779 dw = (char_u *)getroom_save(&spin->si_blocks, w);
|
|
236
|
3780 if (dw == NULL)
|
|
300
|
3781 retval = FAIL;
|
|
|
3782 vim_free(pc);
|
|
|
3783 if (retval == FAIL)
|
|
236
|
3784 break;
|
|
|
3785
|
|
300
|
3786 hash = hash_hash(dw);
|
|
|
3787 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
3788 if (!HASHITEM_EMPTY(hi))
|
|
|
3789 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
3790 fname, lnum, dw);
|
|
236
|
3791 else
|
|
300
|
3792 hash_add_item(&ht, hi, dw, hash);
|
|
|
3793
|
|
307
|
3794 flags = 0;
|
|
339
|
3795 pfxlist = NULL;
|
|
307
|
3796 if (afflist != NULL)
|
|
|
3797 {
|
|
|
3798 /* Check for affix name that stands for keep-case word and stands
|
|
|
3799 * for rare word (if defined). */
|
|
310
|
3800 if (affile->af_kep != NUL
|
|
|
3801 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
307
|
3802 flags |= WF_KEEPCAP;
|
|
|
3803 if (affile->af_rar != NUL
|
|
|
3804 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
3805 flags |= WF_RARE;
|
|
346
|
3806 if (affile->af_bad != NUL
|
|
|
3807 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
3808 flags |= WF_BANNED;
|
|
339
|
3809
|
|
|
3810 if (affile->af_pfxpostpone)
|
|
|
3811 /* Need to store the list of prefix IDs with the word. */
|
|
|
3812 pfxlist = get_pfxlist(affile, afflist, &spin->si_blocks);
|
|
307
|
3813 }
|
|
|
3814
|
|
300
|
3815 /* Add the word to the word tree(s). */
|
|
339
|
3816 if (store_word(dw, spin, flags, spin->si_region, pfxlist) == FAIL)
|
|
300
|
3817 retval = FAIL;
|
|
236
|
3818
|
|
300
|
3819 if (afflist != NULL)
|
|
|
3820 {
|
|
|
3821 /* Find all matching suffixes and add the resulting words.
|
|
|
3822 * Additionally do matching prefixes that combine. */
|
|
339
|
3823 if (store_aff_word(dw, spin, afflist, affile,
|
|
307
|
3824 &affile->af_suff, &affile->af_pref,
|
|
339
|
3825 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
3826 retval = FAIL;
|
|
|
3827
|
|
|
3828 /* Find all matching prefixes and add the resulting words. */
|
|
339
|
3829 if (store_aff_word(dw, spin, afflist, affile,
|
|
|
3830 &affile->af_pref, NULL,
|
|
|
3831 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
3832 retval = FAIL;
|
|
|
3833 }
|
|
236
|
3834 }
|
|
|
3835
|
|
300
|
3836 if (spin->si_ascii && non_ascii > 0)
|
|
|
3837 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
3838 non_ascii);
|
|
|
3839 hash_clear(&ht);
|
|
|
3840
|
|
236
|
3841 fclose(fd);
|
|
300
|
3842 return retval;
|
|
236
|
3843 }
|
|
|
3844
|
|
|
3845 /*
|
|
339
|
3846 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
3847 * Used for PFXPOSTPONE.
|
|
|
3848 * Returns a string allocated with getroom(). NULL when there are no prefixes
|
|
|
3849 * or when out of memory.
|
|
|
3850 */
|
|
|
3851 static char_u *
|
|
|
3852 get_pfxlist(affile, afflist, blp)
|
|
|
3853 afffile_T *affile;
|
|
|
3854 char_u *afflist;
|
|
|
3855 sblock_T **blp;
|
|
|
3856 {
|
|
|
3857 char_u *p;
|
|
|
3858 int cnt;
|
|
|
3859 int round;
|
|
|
3860 char_u *res = NULL;
|
|
|
3861 char_u key[2];
|
|
|
3862 hashitem_T *hi;
|
|
|
3863
|
|
|
3864 key[1] = NUL;
|
|
|
3865
|
|
|
3866 /* round 1: count the number of prefix IDs.
|
|
|
3867 * round 2: move prefix IDs to "res" */
|
|
|
3868 for (round = 1; round <= 2; ++round)
|
|
|
3869 {
|
|
|
3870 cnt = 0;
|
|
|
3871 for (p = afflist; *p != NUL; ++p)
|
|
|
3872 {
|
|
|
3873 key[0] = *p;
|
|
|
3874 hi = hash_find(&affile->af_pref, key);
|
|
|
3875 if (!HASHITEM_EMPTY(hi))
|
|
|
3876 {
|
|
|
3877 /* This is a prefix ID, use the new number. */
|
|
|
3878 if (round == 2)
|
|
|
3879 res[cnt] = HI2AH(hi)->ah_newID;
|
|
|
3880 ++cnt;
|
|
|
3881 }
|
|
|
3882 }
|
|
|
3883 if (round == 1 && cnt > 0)
|
|
|
3884 res = getroom(blp, cnt + 1);
|
|
|
3885 if (res == NULL)
|
|
|
3886 break;
|
|
|
3887 }
|
|
|
3888
|
|
|
3889 if (res != NULL)
|
|
|
3890 res[cnt] = NUL;
|
|
|
3891 return res;
|
|
|
3892 }
|
|
|
3893
|
|
|
3894 /*
|
|
300
|
3895 * Apply affixes to a word and store the resulting words.
|
|
|
3896 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
3897 * prefixes or suffixes.
|
|
|
3898 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
3899 * the resulting words for combining affixes.
|
|
|
3900 *
|
|
|
3901 * Returns FAIL when out of memory.
|
|
236
|
3902 */
|
|
300
|
3903 static int
|
|
339
|
3904 store_aff_word(word, spin, afflist, affile, ht, xht, comb, flags, pfxlist)
|
|
300
|
3905 char_u *word; /* basic word start */
|
|
|
3906 spellinfo_T *spin; /* spell info */
|
|
|
3907 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
3908 afffile_T *affile;
|
|
300
|
3909 hashtab_T *ht;
|
|
|
3910 hashtab_T *xht;
|
|
|
3911 int comb; /* only use affixes that combine */
|
|
307
|
3912 int flags; /* flags for the word */
|
|
339
|
3913 char_u *pfxlist; /* list of prefix IDs */
|
|
236
|
3914 {
|
|
|
3915 int todo;
|
|
|
3916 hashitem_T *hi;
|
|
300
|
3917 affheader_T *ah;
|
|
|
3918 affentry_T *ae;
|
|
|
3919 regmatch_T regmatch;
|
|
|
3920 char_u newword[MAXWLEN];
|
|
|
3921 int retval = OK;
|
|
|
3922 int i;
|
|
|
3923 char_u *p;
|
|
366
|
3924 int use_flags;
|
|
236
|
3925
|
|
300
|
3926 todo = ht->ht_used;
|
|
|
3927 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
3928 {
|
|
|
3929 if (!HASHITEM_EMPTY(hi))
|
|
|
3930 {
|
|
|
3931 --todo;
|
|
300
|
3932 ah = HI2AH(hi);
|
|
236
|
3933
|
|
300
|
3934 /* Check that the affix combines, if required, and that the word
|
|
|
3935 * supports this affix. */
|
|
|
3936 if ((!comb || ah->ah_combine)
|
|
|
3937 && vim_strchr(afflist, *ah->ah_key) != NULL)
|
|
236
|
3938 {
|
|
300
|
3939 /* Loop over all affix entries with this name. */
|
|
|
3940 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
3941 {
|
|
300
|
3942 /* Check the condition. It's not logical to match case
|
|
|
3943 * here, but it is required for compatibility with
|
|
339
|
3944 * Myspell.
|
|
|
3945 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
3946 * prefixes with a chop string. */
|
|
300
|
3947 regmatch.regprog = ae->ae_prog;
|
|
|
3948 regmatch.rm_ic = FALSE;
|
|
339
|
3949 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
3950 || ae->ae_chop != NULL)
|
|
|
3951 && (ae->ae_prog == NULL
|
|
|
3952 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
3953 {
|
|
|
3954 /* Match. Remove the chop and add the affix. */
|
|
|
3955 if (xht == NULL)
|
|
240
|
3956 {
|
|
300
|
3957 /* prefix: chop/add at the start of the word */
|
|
|
3958 if (ae->ae_add == NULL)
|
|
|
3959 *newword = NUL;
|
|
|
3960 else
|
|
|
3961 STRCPY(newword, ae->ae_add);
|
|
|
3962 p = word;
|
|
|
3963 if (ae->ae_chop != NULL)
|
|
310
|
3964 {
|
|
300
|
3965 /* Skip chop string. */
|
|
310
|
3966 #ifdef FEAT_MBYTE
|
|
|
3967 if (has_mbyte)
|
|
324
|
3968 {
|
|
310
|
3969 i = mb_charlen(ae->ae_chop);
|
|
324
|
3970 for ( ; i > 0; --i)
|
|
|
3971 mb_ptr_adv(p);
|
|
|
3972 }
|
|
310
|
3973 else
|
|
|
3974 #endif
|
|
324
|
3975 p += STRLEN(ae->ae_chop);
|
|
310
|
3976 }
|
|
300
|
3977 STRCAT(newword, p);
|
|
|
3978 }
|
|
|
3979 else
|
|
|
3980 {
|
|
|
3981 /* suffix: chop/add at the end of the word */
|
|
|
3982 STRCPY(newword, word);
|
|
|
3983 if (ae->ae_chop != NULL)
|
|
|
3984 {
|
|
|
3985 /* Remove chop string. */
|
|
|
3986 p = newword + STRLEN(newword);
|
|
376
|
3987 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
3988 for ( ; i > 0; --i)
|
|
300
|
3989 mb_ptr_back(newword, p);
|
|
|
3990 *p = NUL;
|
|
|
3991 }
|
|
|
3992 if (ae->ae_add != NULL)
|
|
|
3993 STRCAT(newword, ae->ae_add);
|
|
240
|
3994 }
|
|
|
3995
|
|
366
|
3996 /* Obey the "rare" flag of the affix. */
|
|
|
3997 if (ae->ae_rare)
|
|
|
3998 use_flags = flags | WF_RARE;
|
|
|
3999 else
|
|
|
4000 use_flags = flags;
|
|
|
4001
|
|
300
|
4002 /* Store the modified word. */
|
|
366
|
4003 if (store_word(newword, spin, use_flags,
|
|
|
4004 spin->si_region, pfxlist) == FAIL)
|
|
300
|
4005 retval = FAIL;
|
|
236
|
4006
|
|
300
|
4007 /* When added a suffix and combining is allowed also
|
|
|
4008 * try adding prefixes additionally. */
|
|
|
4009 if (xht != NULL && ah->ah_combine)
|
|
339
|
4010 if (store_aff_word(newword, spin, afflist, affile,
|
|
366
|
4011 xht, NULL, TRUE, use_flags, pfxlist)
|
|
|
4012 == FAIL)
|
|
300
|
4013 retval = FAIL;
|
|
236
|
4014 }
|
|
|
4015 }
|
|
|
4016 }
|
|
|
4017 }
|
|
|
4018 }
|
|
|
4019
|
|
|
4020 return retval;
|
|
|
4021 }
|
|
|
4022
|
|
|
4023 /*
|
|
300
|
4024 * Read a file with a list of words.
|
|
236
|
4025 */
|
|
|
4026 static int
|
|
300
|
4027 spell_read_wordfile(fname, spin)
|
|
|
4028 char_u *fname;
|
|
|
4029 spellinfo_T *spin;
|
|
236
|
4030 {
|
|
300
|
4031 FILE *fd;
|
|
|
4032 long lnum = 0;
|
|
|
4033 char_u rline[MAXLINELEN];
|
|
|
4034 char_u *line;
|
|
|
4035 char_u *pc = NULL;
|
|
381
|
4036 char_u *p;
|
|
300
|
4037 int l;
|
|
|
4038 int retval = OK;
|
|
|
4039 int did_word = FALSE;
|
|
|
4040 int non_ascii = 0;
|
|
307
|
4041 int flags;
|
|
316
|
4042 int regionmask;
|
|
236
|
4043
|
|
300
|
4044 /*
|
|
|
4045 * Open the file.
|
|
|
4046 */
|
|
310
|
4047 fd = mch_fopen((char *)fname, "r");
|
|
300
|
4048 if (fd == NULL)
|
|
236
|
4049 {
|
|
300
|
4050 EMSG2(_(e_notopen), fname);
|
|
|
4051 return FAIL;
|
|
236
|
4052 }
|
|
|
4053
|
|
310
|
4054 if (spin->si_verbose || p_verbose > 2)
|
|
|
4055 {
|
|
|
4056 if (!spin->si_verbose)
|
|
|
4057 verbose_enter();
|
|
366
|
4058 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
4059 out_flush();
|
|
|
4060 if (!spin->si_verbose)
|
|
|
4061 verbose_leave();
|
|
|
4062 }
|
|
300
|
4063
|
|
|
4064 /*
|
|
|
4065 * Read all the lines in the file one by one.
|
|
|
4066 */
|
|
|
4067 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
4068 {
|
|
|
4069 line_breakcheck();
|
|
|
4070 ++lnum;
|
|
|
4071
|
|
|
4072 /* Skip comment lines. */
|
|
|
4073 if (*rline == '#')
|
|
|
4074 continue;
|
|
|
4075
|
|
|
4076 /* Remove CR, LF and white space from the end. */
|
|
|
4077 l = STRLEN(rline);
|
|
|
4078 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
4079 --l;
|
|
|
4080 if (l == 0)
|
|
|
4081 continue; /* empty or blank line */
|
|
|
4082 rline[l] = NUL;
|
|
|
4083
|
|
|
4084 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
4085 vim_free(pc);
|
|
310
|
4086 #ifdef FEAT_MBYTE
|
|
300
|
4087 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
4088 {
|
|
|
4089 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
4090 if (pc == NULL)
|
|
|
4091 {
|
|
|
4092 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4093 fname, lnum, rline);
|
|
|
4094 continue;
|
|
|
4095 }
|
|
|
4096 line = pc;
|
|
|
4097 }
|
|
|
4098 else
|
|
310
|
4099 #endif
|
|
300
|
4100 {
|
|
|
4101 pc = NULL;
|
|
|
4102 line = rline;
|
|
|
4103 }
|
|
|
4104
|
|
307
|
4105 if (*line == '/')
|
|
300
|
4106 {
|
|
307
|
4107 ++line;
|
|
|
4108 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
4109 {
|
|
|
4110 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
4111 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
4112 fname, lnum, line - 1);
|
|
300
|
4113 else if (did_word)
|
|
316
|
4114 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
4115 fname, lnum, line - 1);
|
|
300
|
4116 else
|
|
|
4117 {
|
|
310
|
4118 #ifdef FEAT_MBYTE
|
|
|
4119 char_u *enc;
|
|
|
4120
|
|
300
|
4121 /* Setup for conversion to 'encoding'. */
|
|
316
|
4122 line += 10;
|
|
|
4123 enc = enc_canonize(line);
|
|
300
|
4124 if (enc != NULL && !spin->si_ascii
|
|
|
4125 && convert_setup(&spin->si_conv, enc,
|
|
|
4126 p_enc) == FAIL)
|
|
|
4127 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
4128 fname, line, p_enc);
|
|
300
|
4129 vim_free(enc);
|
|
375
|
4130 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4131 #else
|
|
|
4132 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4133 #endif
|
|
300
|
4134 }
|
|
307
|
4135 continue;
|
|
300
|
4136 }
|
|
307
|
4137
|
|
316
|
4138 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
4139 {
|
|
|
4140 if (spin->si_region_count > 1)
|
|
|
4141 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
4142 fname, lnum, line);
|
|
|
4143 else
|
|
|
4144 {
|
|
|
4145 line += 8;
|
|
|
4146 if (STRLEN(line) > 16)
|
|
|
4147 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
4148 fname, lnum, line);
|
|
|
4149 else
|
|
|
4150 {
|
|
|
4151 spin->si_region_count = STRLEN(line) / 2;
|
|
|
4152 STRCPY(spin->si_region_name, line);
|
|
|
4153 }
|
|
|
4154 }
|
|
|
4155 continue;
|
|
|
4156 }
|
|
|
4157
|
|
381
|
4158 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
4159 fname, lnum, line - 1);
|
|
|
4160 continue;
|
|
|
4161 }
|
|
|
4162
|
|
|
4163 flags = 0;
|
|
|
4164 regionmask = spin->si_region;
|
|
|
4165
|
|
|
4166 /* Check for flags and region after a slash. */
|
|
|
4167 p = vim_strchr(line, '/');
|
|
|
4168 if (p != NULL)
|
|
|
4169 {
|
|
|
4170 *p++ = NUL;
|
|
|
4171 while (*p != NUL)
|
|
|
4172 {
|
|
|
4173 if (*p == '=') /* keep-case word */
|
|
|
4174 flags |= WF_KEEPCAP;
|
|
|
4175 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
4176 flags |= WF_BANNED;
|
|
|
4177 else if (*p == '?') /* Rare word. */
|
|
|
4178 flags |= WF_RARE;
|
|
|
4179 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
4180 {
|
|
381
|
4181 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
4182 regionmask = 0;
|
|
|
4183 flags |= WF_REGION;
|
|
|
4184
|
|
|
4185 l = *p - '0';
|
|
316
|
4186 if (l > spin->si_region_count)
|
|
|
4187 {
|
|
|
4188 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
4189 fname, lnum, p);
|
|
316
|
4190 break;
|
|
|
4191 }
|
|
|
4192 regionmask |= 1 << (l - 1);
|
|
|
4193 }
|
|
381
|
4194 else
|
|
|
4195 {
|
|
|
4196 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
4197 fname, lnum, p);
|
|
|
4198 break;
|
|
|
4199 }
|
|
|
4200 ++p;
|
|
307
|
4201 }
|
|
300
|
4202 }
|
|
|
4203
|
|
|
4204 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4205 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4206 {
|
|
|
4207 ++non_ascii;
|
|
|
4208 continue;
|
|
|
4209 }
|
|
|
4210
|
|
|
4211 /* Normal word: store it. */
|
|
339
|
4212 if (store_word(line, spin, flags, regionmask, NULL) == FAIL)
|
|
300
|
4213 {
|
|
|
4214 retval = FAIL;
|
|
|
4215 break;
|
|
|
4216 }
|
|
|
4217 did_word = TRUE;
|
|
|
4218 }
|
|
|
4219
|
|
|
4220 vim_free(pc);
|
|
|
4221 fclose(fd);
|
|
|
4222
|
|
310
|
4223 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
4224 {
|
|
|
4225 if (p_verbose > 2)
|
|
|
4226 verbose_enter();
|
|
300
|
4227 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4228 non_ascii);
|
|
310
|
4229 if (p_verbose > 2)
|
|
|
4230 verbose_leave();
|
|
|
4231 }
|
|
300
|
4232 return retval;
|
|
236
|
4233 }
|
|
|
4234
|
|
|
4235 /*
|
|
300
|
4236 * Get part of an sblock_T, "len" bytes long.
|
|
|
4237 * This avoids calling free() for every little struct we use.
|
|
|
4238 * The memory is cleared to all zeros.
|
|
|
4239 * Returns NULL when out of memory.
|
|
|
4240 */
|
|
|
4241 static void *
|
|
|
4242 getroom(blp, len)
|
|
|
4243 sblock_T **blp;
|
|
|
4244 size_t len; /* length needed */
|
|
|
4245 {
|
|
|
4246 char_u *p;
|
|
|
4247 sblock_T *bl = *blp;
|
|
|
4248
|
|
|
4249 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
4250 {
|
|
|
4251 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
4252 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
4253 if (bl == NULL)
|
|
|
4254 return NULL;
|
|
|
4255 bl->sb_next = *blp;
|
|
|
4256 *blp = bl;
|
|
|
4257 bl->sb_used = 0;
|
|
|
4258 }
|
|
|
4259
|
|
|
4260 p = bl->sb_data + bl->sb_used;
|
|
|
4261 bl->sb_used += len;
|
|
|
4262
|
|
|
4263 return p;
|
|
|
4264 }
|
|
|
4265
|
|
|
4266 /*
|
|
|
4267 * Make a copy of a string into memory allocated with getroom().
|
|
|
4268 */
|
|
|
4269 static char_u *
|
|
|
4270 getroom_save(blp, s)
|
|
|
4271 sblock_T **blp;
|
|
|
4272 char_u *s;
|
|
|
4273 {
|
|
|
4274 char_u *sc;
|
|
|
4275
|
|
|
4276 sc = (char_u *)getroom(blp, STRLEN(s) + 1);
|
|
|
4277 if (sc != NULL)
|
|
|
4278 STRCPY(sc, s);
|
|
|
4279 return sc;
|
|
|
4280 }
|
|
|
4281
|
|
|
4282
|
|
|
4283 /*
|
|
|
4284 * Free the list of allocated sblock_T.
|
|
236
|
4285 */
|
|
|
4286 static void
|
|
300
|
4287 free_blocks(bl)
|
|
|
4288 sblock_T *bl;
|
|
236
|
4289 {
|
|
300
|
4290 sblock_T *next;
|
|
236
|
4291
|
|
300
|
4292 while (bl != NULL)
|
|
236
|
4293 {
|
|
300
|
4294 next = bl->sb_next;
|
|
|
4295 vim_free(bl);
|
|
|
4296 bl = next;
|
|
236
|
4297 }
|
|
|
4298 }
|
|
|
4299
|
|
|
4300 /*
|
|
300
|
4301 * Allocate the root of a word tree.
|
|
236
|
4302 */
|
|
300
|
4303 static wordnode_T *
|
|
|
4304 wordtree_alloc(blp)
|
|
|
4305 sblock_T **blp;
|
|
236
|
4306 {
|
|
300
|
4307 return (wordnode_T *)getroom(blp, sizeof(wordnode_T));
|
|
236
|
4308 }
|
|
|
4309
|
|
|
4310 /*
|
|
300
|
4311 * Store a word in the tree(s).
|
|
307
|
4312 * Always store it in the case-folded tree. A keep-case word can also be used
|
|
|
4313 * with all caps.
|
|
300
|
4314 * For a keep-case word also store it in the keep-case tree.
|
|
339
|
4315 * When "pfxlist" is not NULL store the word for each prefix ID.
|
|
236
|
4316 */
|
|
|
4317 static int
|
|
339
|
4318 store_word(word, spin, flags, region, pfxlist)
|
|
300
|
4319 char_u *word;
|
|
|
4320 spellinfo_T *spin;
|
|
307
|
4321 int flags; /* extra flags, WF_BANNED */
|
|
316
|
4322 int region; /* supported region(s) */
|
|
339
|
4323 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
236
|
4324 {
|
|
300
|
4325 int len = STRLEN(word);
|
|
|
4326 int ct = captype(word, word + len);
|
|
|
4327 char_u foldword[MAXWLEN];
|
|
339
|
4328 int res = OK;
|
|
|
4329 char_u *p;
|
|
236
|
4330
|
|
323
|
4331 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
4332 for (p = pfxlist; res == OK; ++p)
|
|
|
4333 {
|
|
|
4334 res = tree_add_word(foldword, spin->si_foldroot, ct | flags,
|
|
|
4335 region, p == NULL ? 0 : *p, &spin->si_blocks);
|
|
|
4336 if (p == NULL || *p == NUL)
|
|
|
4337 break;
|
|
|
4338 }
|
|
334
|
4339 ++spin->si_foldwcount;
|
|
307
|
4340
|
|
|
4341 if (res == OK && (ct == WF_KEEPCAP || flags & WF_KEEPCAP))
|
|
334
|
4342 {
|
|
339
|
4343 for (p = pfxlist; res == OK; ++p)
|
|
|
4344 {
|
|
|
4345 res = tree_add_word(word, spin->si_keeproot, flags,
|
|
|
4346 region, p == NULL ? 0 : *p, &spin->si_blocks);
|
|
|
4347 if (p == NULL || *p == NUL)
|
|
|
4348 break;
|
|
|
4349 }
|
|
334
|
4350 ++spin->si_keepwcount;
|
|
|
4351 }
|
|
300
|
4352 return res;
|
|
236
|
4353 }
|
|
|
4354
|
|
|
4355 /*
|
|
300
|
4356 * Add word "word" to a word tree at "root".
|
|
366
|
4357 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
4358 * "rare" and "region" is the condition nr.
|
|
255
|
4359 * Returns FAIL when out of memory.
|
|
236
|
4360 */
|
|
255
|
4361 static int
|
|
339
|
4362 tree_add_word(word, root, flags, region, prefixID, blp)
|
|
300
|
4363 char_u *word;
|
|
|
4364 wordnode_T *root;
|
|
|
4365 int flags;
|
|
|
4366 int region;
|
|
339
|
4367 int prefixID;
|
|
300
|
4368 sblock_T **blp;
|
|
236
|
4369 {
|
|
300
|
4370 wordnode_T *node = root;
|
|
|
4371 wordnode_T *np;
|
|
|
4372 wordnode_T **prev = NULL;
|
|
|
4373 int i;
|
|
255
|
4374
|
|
300
|
4375 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
4376 for (i = 0; ; ++i)
|
|
255
|
4377 {
|
|
300
|
4378 /* Look for the sibling that has the same character. They are sorted
|
|
|
4379 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
4380 * higher byte value. For zero bytes (end of word) the sorting is
|
|
|
4381 * done on flags and then on prefixID
|
|
300
|
4382 */
|
|
339
|
4383 while (node != NULL
|
|
|
4384 && (node->wn_byte < word[i]
|
|
|
4385 || (node->wn_byte == NUL
|
|
|
4386 && (flags < 0
|
|
|
4387 ? node->wn_prefixID < prefixID
|
|
|
4388 : node->wn_flags < (flags & 0xff)
|
|
|
4389 || (node->wn_flags == (flags & 0xff)
|
|
|
4390 && node->wn_prefixID < prefixID)))))
|
|
236
|
4391 {
|
|
300
|
4392 prev = &node->wn_sibling;
|
|
|
4393 node = *prev;
|
|
236
|
4394 }
|
|
339
|
4395 if (node == NULL
|
|
|
4396 || node->wn_byte != word[i]
|
|
|
4397 || (word[i] == NUL
|
|
|
4398 && (flags < 0
|
|
|
4399 || node->wn_flags != (flags & 0xff)
|
|
|
4400 || node->wn_prefixID != prefixID)))
|
|
255
|
4401 {
|
|
300
|
4402 /* Allocate a new node. */
|
|
|
4403 np = (wordnode_T *)getroom(blp, sizeof(wordnode_T));
|
|
|
4404 if (np == NULL)
|
|
|
4405 return FAIL;
|
|
|
4406 np->wn_byte = word[i];
|
|
|
4407 *prev = np;
|
|
|
4408 np->wn_sibling = node;
|
|
|
4409 node = np;
|
|
255
|
4410 }
|
|
300
|
4411
|
|
|
4412 if (word[i] == NUL)
|
|
|
4413 {
|
|
|
4414 node->wn_flags = flags;
|
|
|
4415 node->wn_region |= region;
|
|
339
|
4416 node->wn_prefixID = prefixID;
|
|
300
|
4417 break;
|
|
|
4418 }
|
|
|
4419 prev = &node->wn_child;
|
|
|
4420 node = *prev;
|
|
255
|
4421 }
|
|
|
4422
|
|
|
4423 return OK;
|
|
236
|
4424 }
|
|
|
4425
|
|
|
4426 /*
|
|
300
|
4427 * Compress a tree: find tails that are identical and can be shared.
|
|
|
4428 */
|
|
|
4429 static void
|
|
310
|
4430 wordtree_compress(root, spin)
|
|
300
|
4431 wordnode_T *root;
|
|
310
|
4432 spellinfo_T *spin;
|
|
300
|
4433 {
|
|
|
4434 hashtab_T ht;
|
|
|
4435 int n;
|
|
|
4436 int tot = 0;
|
|
|
4437
|
|
|
4438 if (root != NULL)
|
|
|
4439 {
|
|
|
4440 hash_init(&ht);
|
|
|
4441 n = node_compress(root, &ht, &tot);
|
|
310
|
4442 if (spin->si_verbose || p_verbose > 2)
|
|
|
4443 {
|
|
|
4444 if (!spin->si_verbose)
|
|
|
4445 verbose_enter();
|
|
|
4446 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
300
|
4447 n, tot, (tot - n) * 100 / tot);
|
|
310
|
4448 if (p_verbose > 2)
|
|
|
4449 verbose_leave();
|
|
|
4450 }
|
|
300
|
4451 hash_clear(&ht);
|
|
|
4452 }
|
|
|
4453 }
|
|
|
4454
|
|
|
4455 /*
|
|
|
4456 * Compress a node, its siblings and its children, depth first.
|
|
|
4457 * Returns the number of compressed nodes.
|
|
236
|
4458 */
|
|
255
|
4459 static int
|
|
300
|
4460 node_compress(node, ht, tot)
|
|
|
4461 wordnode_T *node;
|
|
|
4462 hashtab_T *ht;
|
|
|
4463 int *tot; /* total count of nodes before compressing,
|
|
|
4464 incremented while going through the tree */
|
|
236
|
4465 {
|
|
300
|
4466 wordnode_T *np;
|
|
|
4467 wordnode_T *tp;
|
|
|
4468 wordnode_T *child;
|
|
|
4469 hash_T hash;
|
|
236
|
4470 hashitem_T *hi;
|
|
300
|
4471 int len = 0;
|
|
|
4472 unsigned nr, n;
|
|
|
4473 int compressed = 0;
|
|
236
|
4474
|
|
300
|
4475 /*
|
|
|
4476 * Go through the list of siblings. Compress each child and then try
|
|
|
4477 * finding an identical child to replace it.
|
|
|
4478 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
4479 * but the whole list of siblings, of which the node is the first.
|
|
|
4480 */
|
|
|
4481 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4482 {
|
|
300
|
4483 ++len;
|
|
|
4484 if ((child = np->wn_child) != NULL)
|
|
|
4485 {
|
|
346
|
4486 /* Compress the child. This fills hashkey. */
|
|
300
|
4487 compressed += node_compress(child, ht, tot);
|
|
|
4488
|
|
|
4489 /* Try to find an identical child. */
|
|
346
|
4490 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
4491 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
4492 tp = NULL;
|
|
|
4493 if (!HASHITEM_EMPTY(hi))
|
|
|
4494 {
|
|
|
4495 /* There are children with an identical hash value. Now check
|
|
|
4496 * if there is one that is really identical. */
|
|
346
|
4497 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
4498 if (node_equal(child, tp))
|
|
|
4499 {
|
|
|
4500 /* Found one! Now use that child in place of the
|
|
|
4501 * current one. This means the current child is
|
|
|
4502 * dropped from the tree. */
|
|
|
4503 np->wn_child = tp;
|
|
|
4504 ++compressed;
|
|
|
4505 break;
|
|
|
4506 }
|
|
|
4507 if (tp == NULL)
|
|
|
4508 {
|
|
|
4509 /* No other child with this hash value equals the child of
|
|
|
4510 * the node, add it to the linked list after the first
|
|
|
4511 * item. */
|
|
|
4512 tp = HI2WN(hi);
|
|
346
|
4513 child->wn_u2.next = tp->wn_u2.next;
|
|
|
4514 tp->wn_u2.next = child;
|
|
300
|
4515 }
|
|
|
4516 }
|
|
|
4517 else
|
|
|
4518 /* No other child has this hash value, add it to the
|
|
|
4519 * hashtable. */
|
|
346
|
4520 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
4521 }
|
|
236
|
4522 }
|
|
300
|
4523 *tot += len;
|
|
|
4524
|
|
|
4525 /*
|
|
|
4526 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
4527 * find a lookalike node. This must be done after compressing the sibling
|
|
|
4528 * list, otherwise the hash key would become invalid by the compression.
|
|
|
4529 */
|
|
346
|
4530 node->wn_u1.hashkey[0] = len;
|
|
300
|
4531 nr = 0;
|
|
|
4532 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4533 {
|
|
300
|
4534 if (np->wn_byte == NUL)
|
|
339
|
4535 /* end node: use wn_flags, wn_region and wn_prefixID */
|
|
|
4536 n = np->wn_flags + (np->wn_region << 8) + (np->wn_prefixID << 16);
|
|
300
|
4537 else
|
|
|
4538 /* byte node: use the byte value and the child pointer */
|
|
|
4539 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
4540 nr = nr * 101 + n;
|
|
236
|
4541 }
|
|
300
|
4542
|
|
|
4543 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
4544 n = nr & 0xff;
|
|
346
|
4545 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
4546 n = (nr >> 8) & 0xff;
|
|
346
|
4547 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
4548 n = (nr >> 16) & 0xff;
|
|
346
|
4549 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
4550 n = (nr >> 24) & 0xff;
|
|
346
|
4551 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
4552 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
4553
|
|
|
4554 return compressed;
|
|
|
4555 }
|
|
|
4556
|
|
|
4557 /*
|
|
|
4558 * Return TRUE when two nodes have identical siblings and children.
|
|
|
4559 */
|
|
|
4560 static int
|
|
|
4561 node_equal(n1, n2)
|
|
|
4562 wordnode_T *n1;
|
|
|
4563 wordnode_T *n2;
|
|
|
4564 {
|
|
|
4565 wordnode_T *p1;
|
|
|
4566 wordnode_T *p2;
|
|
|
4567
|
|
|
4568 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
4569 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
4570 if (p1->wn_byte != p2->wn_byte
|
|
|
4571 || (p1->wn_byte == NUL
|
|
|
4572 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
4573 || p1->wn_region != p2->wn_region
|
|
|
4574 || p1->wn_prefixID != p2->wn_prefixID)
|
|
300
|
4575 : (p1->wn_child != p2->wn_child)))
|
|
|
4576 break;
|
|
|
4577
|
|
|
4578 return p1 == NULL && p2 == NULL;
|
|
236
|
4579 }
|
|
|
4580
|
|
|
4581 /*
|
|
|
4582 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
4583 */
|
|
255
|
4584 void
|
|
236
|
4585 put_bytes(fd, nr, len)
|
|
|
4586 FILE *fd;
|
|
|
4587 long_u nr;
|
|
|
4588 int len;
|
|
|
4589 {
|
|
|
4590 int i;
|
|
|
4591
|
|
|
4592 for (i = len - 1; i >= 0; --i)
|
|
|
4593 putc((int)(nr >> (i * 8)), fd);
|
|
|
4594 }
|
|
|
4595
|
|
323
|
4596 static int
|
|
|
4597 #ifdef __BORLANDC__
|
|
|
4598 _RTLENTRYF
|
|
|
4599 #endif
|
|
|
4600 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
4601
|
|
|
4602 /*
|
|
|
4603 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
4604 */
|
|
|
4605 static int
|
|
|
4606 #ifdef __BORLANDC__
|
|
|
4607 _RTLENTRYF
|
|
|
4608 #endif
|
|
|
4609 rep_compare(s1, s2)
|
|
|
4610 const void *s1;
|
|
|
4611 const void *s2;
|
|
|
4612 {
|
|
|
4613 fromto_T *p1 = (fromto_T *)s1;
|
|
|
4614 fromto_T *p2 = (fromto_T *)s2;
|
|
|
4615
|
|
|
4616 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
4617 }
|
|
|
4618
|
|
236
|
4619 /*
|
|
|
4620 * Write the Vim spell file "fname".
|
|
|
4621 */
|
|
|
4622 static void
|
|
316
|
4623 write_vim_spell(fname, spin)
|
|
236
|
4624 char_u *fname;
|
|
300
|
4625 spellinfo_T *spin;
|
|
236
|
4626 {
|
|
300
|
4627 FILE *fd;
|
|
|
4628 int regionmask;
|
|
236
|
4629 int round;
|
|
300
|
4630 wordnode_T *tree;
|
|
|
4631 int nodecount;
|
|
323
|
4632 int i;
|
|
|
4633 int l;
|
|
|
4634 garray_T *gap;
|
|
|
4635 fromto_T *ftp;
|
|
|
4636 char_u *p;
|
|
|
4637 int rr;
|
|
236
|
4638
|
|
310
|
4639 fd = mch_fopen((char *)fname, "w");
|
|
300
|
4640 if (fd == NULL)
|
|
236
|
4641 {
|
|
|
4642 EMSG2(_(e_notopen), fname);
|
|
|
4643 return;
|
|
|
4644 }
|
|
|
4645
|
|
255
|
4646 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
339
|
4647 * <charflagslen> <charflags>
|
|
|
4648 * <fcharslen> <fchars>
|
|
366
|
4649 * <midwordlen> <midword>
|
|
339
|
4650 * <prefcondcnt> <prefcond> ... */
|
|
300
|
4651
|
|
|
4652 /* <fileID> */
|
|
|
4653 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
4654 EMSG(_(e_write));
|
|
236
|
4655
|
|
|
4656 /* write the region names if there is more than one */
|
|
316
|
4657 if (spin->si_region_count > 1)
|
|
236
|
4658 {
|
|
316
|
4659 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
|
4660 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
|
4661 (size_t)1, fd);
|
|
|
4662 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
4663 }
|
|
|
4664 else
|
|
|
4665 {
|
|
300
|
4666 putc(0, fd);
|
|
|
4667 regionmask = 0;
|
|
236
|
4668 }
|
|
|
4669
|
|
323
|
4670 /*
|
|
|
4671 * Write the table with character flags and table for case folding.
|
|
260
|
4672 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
|
4673 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
4674 * 'encoding'.
|
|
|
4675 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
4676 * the table (avoids that it conflicts). File is shorter too.
|
|
|
4677 */
|
|
|
4678 if (spin->si_ascii || spin->si_add)
|
|
260
|
4679 {
|
|
300
|
4680 putc(0, fd);
|
|
|
4681 putc(0, fd);
|
|
|
4682 putc(0, fd);
|
|
260
|
4683 }
|
|
|
4684 else
|
|
300
|
4685 write_spell_chartab(fd);
|
|
255
|
4686
|
|
366
|
4687
|
|
|
4688 if (spin->si_midword == NULL)
|
|
|
4689 put_bytes(fd, 0L, 2); /* <midwordlen> */
|
|
|
4690 else
|
|
|
4691 {
|
|
|
4692 i = STRLEN(spin->si_midword);
|
|
|
4693 put_bytes(fd, (long_u)i, 2); /* <midwordlen> */
|
|
|
4694 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
4695 }
|
|
|
4696
|
|
|
4697
|
|
339
|
4698 /* Write the prefix conditions. */
|
|
|
4699 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
4700
|
|
366
|
4701 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
4702 * <salflags> <salcount> <sal> ...
|
|
|
4703 * <maplen> <mapstr> */
|
|
|
4704
|
|
323
|
4705 /* Sort the REP items. */
|
|
|
4706 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
4707 sizeof(fromto_T), rep_compare);
|
|
|
4708
|
|
375
|
4709 /* round 1: REP items
|
|
|
4710 * round 2: SAL items (unless SOFO is used) */
|
|
323
|
4711 for (round = 1; round <= 2; ++round)
|
|
|
4712 {
|
|
|
4713 if (round == 1)
|
|
|
4714 gap = &spin->si_rep;
|
|
|
4715 else
|
|
|
4716 {
|
|
|
4717 gap = &spin->si_sal;
|
|
|
4718
|
|
|
4719 i = 0;
|
|
|
4720 if (spin->si_followup)
|
|
|
4721 i |= SAL_F0LLOWUP;
|
|
|
4722 if (spin->si_collapse)
|
|
|
4723 i |= SAL_COLLAPSE;
|
|
|
4724 if (spin->si_rem_accents)
|
|
|
4725 i |= SAL_REM_ACCENTS;
|
|
375
|
4726 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
4727 i |= SAL_SOFO;
|
|
323
|
4728 putc(i, fd); /* <salflags> */
|
|
375
|
4729 if (i & SAL_SOFO)
|
|
|
4730 break;
|
|
323
|
4731 }
|
|
|
4732
|
|
|
4733 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
4734 for (i = 0; i < gap->ga_len; ++i)
|
|
|
4735 {
|
|
|
4736 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
4737 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
4738 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
4739 for (rr = 1; rr <= 2; ++rr)
|
|
|
4740 {
|
|
|
4741 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
4742 l = STRLEN(p);
|
|
|
4743 putc(l, fd);
|
|
|
4744 fwrite(p, l, (size_t)1, fd);
|
|
|
4745 }
|
|
|
4746 }
|
|
|
4747 }
|
|
|
4748
|
|
375
|
4749 /* SOFOFROM and SOFOTO */
|
|
|
4750 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
4751 {
|
|
|
4752 put_bytes(fd, 1L, 2); /* <salcount> */
|
|
|
4753
|
|
|
4754 l = STRLEN(spin->si_sofofr);
|
|
|
4755 put_bytes(fd, (long_u)l, 2); /* <salfromlen> */
|
|
|
4756 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <salfrom> */
|
|
|
4757
|
|
|
4758 l = STRLEN(spin->si_sofoto);
|
|
|
4759 put_bytes(fd, (long_u)l, 2); /* <saltolen> */
|
|
|
4760 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <salto> */
|
|
|
4761 }
|
|
|
4762
|
|
323
|
4763 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
|
4764 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
|
4765 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
|
4766 (size_t)1, fd);
|
|
302
|
4767
|
|
236
|
4768 /*
|
|
339
|
4769 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
4770 */
|
|
323
|
4771 spin->si_memtot = 0;
|
|
339
|
4772 for (round = 1; round <= 3; ++round)
|
|
236
|
4773 {
|
|
339
|
4774 if (round == 1)
|
|
|
4775 tree = spin->si_foldroot;
|
|
|
4776 else if (round == 2)
|
|
|
4777 tree = spin->si_keeproot;
|
|
|
4778 else
|
|
|
4779 tree = spin->si_prefroot;
|
|
236
|
4780
|
|
346
|
4781 /* Clear the index and wnode fields in the tree. */
|
|
|
4782 clear_node(tree);
|
|
|
4783
|
|
300
|
4784 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
4785 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
4786 * nodes. */
|
|
346
|
4787 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
4788
|
|
300
|
4789 /* number of nodes in 4 bytes */
|
|
|
4790 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
4791 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
4792
|
|
300
|
4793 /* Write the nodes. */
|
|
346
|
4794 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
4795 }
|
|
|
4796
|
|
300
|
4797 fclose(fd);
|
|
236
|
4798 }
|
|
|
4799
|
|
|
4800 /*
|
|
346
|
4801 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
4802 * children. This is needed because they are a union with other items to save
|
|
|
4803 * space.
|
|
|
4804 */
|
|
|
4805 static void
|
|
|
4806 clear_node(node)
|
|
|
4807 wordnode_T *node;
|
|
|
4808 {
|
|
|
4809 wordnode_T *np;
|
|
|
4810
|
|
|
4811 if (node != NULL)
|
|
|
4812 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4813 {
|
|
|
4814 np->wn_u1.index = 0;
|
|
|
4815 np->wn_u2.wnode = NULL;
|
|
|
4816
|
|
|
4817 if (np->wn_byte != NUL)
|
|
|
4818 clear_node(np->wn_child);
|
|
|
4819 }
|
|
|
4820 }
|
|
|
4821
|
|
|
4822
|
|
|
4823 /*
|
|
300
|
4824 * Dump a word tree at node "node".
|
|
|
4825 *
|
|
|
4826 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
4827 * byte recursively write the children.
|
|
|
4828 *
|
|
|
4829 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
4830 * are made about the indexes (so that we don't have to write them in the
|
|
|
4831 * file).
|
|
236
|
4832 *
|
|
300
|
4833 * Returns the number of nodes used.
|
|
236
|
4834 */
|
|
300
|
4835 static int
|
|
346
|
4836 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
4837 FILE *fd; /* NULL when only counting */
|
|
300
|
4838 wordnode_T *node;
|
|
|
4839 int index;
|
|
|
4840 int regionmask;
|
|
339
|
4841 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
4842 {
|
|
300
|
4843 int newindex = index;
|
|
|
4844 int siblingcount = 0;
|
|
|
4845 wordnode_T *np;
|
|
236
|
4846 int flags;
|
|
300
|
4847
|
|
|
4848 /* If "node" is zero the tree is empty. */
|
|
|
4849 if (node == NULL)
|
|
|
4850 return 0;
|
|
|
4851
|
|
|
4852 /* Store the index where this node is written. */
|
|
346
|
4853 node->wn_u1.index = index;
|
|
236
|
4854
|
|
300
|
4855 /* Count the number of siblings. */
|
|
|
4856 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4857 ++siblingcount;
|
|
236
|
4858
|
|
300
|
4859 /* Write the sibling count. */
|
|
|
4860 if (fd != NULL)
|
|
|
4861 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
4862
|
|
300
|
4863 /* Write each sibling byte and optionally extra info. */
|
|
|
4864 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4865 {
|
|
300
|
4866 if (np->wn_byte == 0)
|
|
|
4867 {
|
|
|
4868 if (fd != NULL)
|
|
|
4869 {
|
|
339
|
4870 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
4871 if (prefixtree)
|
|
300
|
4872 {
|
|
339
|
4873 /* In PREFIXTREE write the required prefixID and the
|
|
|
4874 * associated condition nr (stored in wn_region). */
|
|
366
|
4875 if (np->wn_flags == (char_u)-2)
|
|
|
4876 putc(BY_FLAGS, fd); /* <byte> rare */
|
|
|
4877 else
|
|
|
4878 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
339
|
4879 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
4880 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
4881 }
|
|
|
4882 else
|
|
|
4883 {
|
|
339
|
4884 /* For word trees we write the flag/region items. */
|
|
|
4885 flags = np->wn_flags;
|
|
|
4886 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
4887 flags |= WF_REGION;
|
|
|
4888 if (np->wn_prefixID != 0)
|
|
|
4889 flags |= WF_PFX;
|
|
|
4890 if (flags == 0)
|
|
|
4891 {
|
|
|
4892 /* word without flags or region */
|
|
|
4893 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
4894 }
|
|
|
4895 else
|
|
|
4896 {
|
|
|
4897 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
4898 putc(flags, fd); /* <flags> */
|
|
|
4899 if (flags & WF_REGION)
|
|
|
4900 putc(np->wn_region, fd); /* <region> */
|
|
|
4901 if (flags & WF_PFX)
|
|
|
4902 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
4903 }
|
|
300
|
4904 }
|
|
|
4905 }
|
|
|
4906 }
|
|
|
4907 else
|
|
|
4908 {
|
|
346
|
4909 if (np->wn_child->wn_u1.index != 0
|
|
|
4910 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
4911 {
|
|
|
4912 /* The child is written elsewhere, write the reference. */
|
|
|
4913 if (fd != NULL)
|
|
|
4914 {
|
|
|
4915 putc(BY_INDEX, fd); /* <byte> */
|
|
|
4916 /* <nodeidx> */
|
|
346
|
4917 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
4918 }
|
|
|
4919 }
|
|
346
|
4920 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
4921 /* We will write the child below and give it an index. */
|
|
346
|
4922 np->wn_child->wn_u2.wnode = node;
|
|
236
|
4923
|
|
300
|
4924 if (fd != NULL)
|
|
|
4925 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
4926 {
|
|
|
4927 EMSG(_(e_write));
|
|
|
4928 return 0;
|
|
|
4929 }
|
|
|
4930 }
|
|
236
|
4931 }
|
|
|
4932
|
|
300
|
4933 /* Space used in the array when reading: one for each sibling and one for
|
|
|
4934 * the count. */
|
|
|
4935 newindex += siblingcount + 1;
|
|
249
|
4936
|
|
300
|
4937 /* Recursively dump the children of each sibling. */
|
|
|
4938 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
4939 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
4940 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
4941 prefixtree);
|
|
249
|
4942
|
|
300
|
4943 return newindex;
|
|
236
|
4944 }
|
|
|
4945
|
|
|
4946
|
|
|
4947 /*
|
|
310
|
4948 * ":mkspell [-ascii] outfile infile ..."
|
|
|
4949 * ":mkspell [-ascii] addfile"
|
|
236
|
4950 */
|
|
|
4951 void
|
|
|
4952 ex_mkspell(eap)
|
|
|
4953 exarg_T *eap;
|
|
|
4954 {
|
|
|
4955 int fcount;
|
|
|
4956 char_u **fnames;
|
|
310
|
4957 char_u *arg = eap->arg;
|
|
|
4958 int ascii = FALSE;
|
|
|
4959
|
|
|
4960 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
4961 {
|
|
|
4962 ascii = TRUE;
|
|
|
4963 arg = skipwhite(arg + 6);
|
|
|
4964 }
|
|
|
4965
|
|
|
4966 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
4967 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
4968 {
|
|
323
|
4969 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
4970 FreeWild(fcount, fnames);
|
|
|
4971 }
|
|
|
4972 }
|
|
|
4973
|
|
|
4974 /*
|
|
|
4975 * Create a Vim spell file from one or more word lists.
|
|
|
4976 * "fnames[0]" is the output file name.
|
|
|
4977 * "fnames[fcount - 1]" is the last input file name.
|
|
|
4978 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
4979 * and ".spl" is appended to make the output file name.
|
|
|
4980 */
|
|
|
4981 static void
|
|
323
|
4982 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
4983 int fcount;
|
|
|
4984 char_u **fnames;
|
|
|
4985 int ascii; /* -ascii argument given */
|
|
|
4986 int overwrite; /* overwrite existing output file */
|
|
323
|
4987 int added_word; /* invoked through "zg" */
|
|
310
|
4988 {
|
|
236
|
4989 char_u fname[MAXPATHL];
|
|
|
4990 char_u wfname[MAXPATHL];
|
|
310
|
4991 char_u **innames;
|
|
|
4992 int incount;
|
|
236
|
4993 afffile_T *(afile[8]);
|
|
|
4994 int i;
|
|
|
4995 int len;
|
|
|
4996 struct stat st;
|
|
255
|
4997 int error = FALSE;
|
|
300
|
4998 spellinfo_T spin;
|
|
|
4999
|
|
|
5000 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
5001 spin.si_verbose = !added_word;
|
|
310
|
5002 spin.si_ascii = ascii;
|
|
323
|
5003 spin.si_followup = TRUE;
|
|
|
5004 spin.si_rem_accents = TRUE;
|
|
|
5005 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
5006 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
5007 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
5008 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
5009
|
|
|
5010 /* default: fnames[0] is output file, following are input files */
|
|
|
5011 innames = &fnames[1];
|
|
|
5012 incount = fcount - 1;
|
|
|
5013
|
|
|
5014 if (fcount >= 1)
|
|
240
|
5015 {
|
|
310
|
5016 len = STRLEN(fnames[0]);
|
|
|
5017 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
5018 {
|
|
|
5019 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
5020 * "path/en.latin1.add.spl". */
|
|
|
5021 innames = &fnames[0];
|
|
|
5022 incount = 1;
|
|
|
5023 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
5024 }
|
|
366
|
5025 else if (fcount == 1)
|
|
|
5026 {
|
|
|
5027 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
5028 innames = &fnames[0];
|
|
|
5029 incount = 1;
|
|
|
5030 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5031 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5032 }
|
|
310
|
5033 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
5034 {
|
|
|
5035 /* Name ends in ".spl", use as the file name. */
|
|
323
|
5036 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
5037 }
|
|
|
5038 else
|
|
|
5039 /* Name should be language, make the file name from it. */
|
|
|
5040 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5041 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5042
|
|
|
5043 /* Check for .ascii.spl. */
|
|
|
5044 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
5045 spin.si_ascii = TRUE;
|
|
|
5046
|
|
|
5047 /* Check for .add.spl. */
|
|
|
5048 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
5049 spin.si_add = TRUE;
|
|
240
|
5050 }
|
|
|
5051
|
|
310
|
5052 if (incount <= 0)
|
|
236
|
5053 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
5054 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
5055 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
5056 else if (incount > 8)
|
|
236
|
5057 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
5058 else
|
|
|
5059 {
|
|
|
5060 /* Check for overwriting before doing things that may take a lot of
|
|
|
5061 * time. */
|
|
310
|
5062 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
5063 {
|
|
|
5064 EMSG(_(e_exists));
|
|
310
|
5065 return;
|
|
236
|
5066 }
|
|
310
|
5067 if (mch_isdir(wfname))
|
|
236
|
5068 {
|
|
310
|
5069 EMSG2(_(e_isadir2), wfname);
|
|
|
5070 return;
|
|
236
|
5071 }
|
|
|
5072
|
|
|
5073 /*
|
|
|
5074 * Init the aff and dic pointers.
|
|
|
5075 * Get the region names if there are more than 2 arguments.
|
|
|
5076 */
|
|
310
|
5077 for (i = 0; i < incount; ++i)
|
|
236
|
5078 {
|
|
310
|
5079 afile[i] = NULL;
|
|
300
|
5080
|
|
316
|
5081 if (incount > 1)
|
|
236
|
5082 {
|
|
310
|
5083 len = STRLEN(innames[i]);
|
|
|
5084 if (STRLEN(gettail(innames[i])) < 5
|
|
|
5085 || innames[i][len - 3] != '_')
|
|
236
|
5086 {
|
|
310
|
5087 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
5088 return;
|
|
236
|
5089 }
|
|
316
|
5090 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
5091 spin.si_region_name[i * 2 + 1] =
|
|
|
5092 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
5093 }
|
|
|
5094 }
|
|
316
|
5095 spin.si_region_count = incount;
|
|
236
|
5096
|
|
300
|
5097 spin.si_foldroot = wordtree_alloc(&spin.si_blocks);
|
|
|
5098 spin.si_keeproot = wordtree_alloc(&spin.si_blocks);
|
|
339
|
5099 spin.si_prefroot = wordtree_alloc(&spin.si_blocks);
|
|
|
5100 if (spin.si_foldroot == NULL
|
|
|
5101 || spin.si_keeproot == NULL
|
|
|
5102 || spin.si_prefroot == NULL)
|
|
300
|
5103 {
|
|
|
5104 error = TRUE;
|
|
310
|
5105 return;
|
|
300
|
5106 }
|
|
|
5107
|
|
351
|
5108 /* When not producing a .add.spl file clear the character table when
|
|
|
5109 * we encounter one in the .aff file. This means we dump the current
|
|
|
5110 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
5111 * better than guessing the contents, the table will match a
|
|
|
5112 * previously loaded spell file. */
|
|
|
5113 if (!spin.si_add)
|
|
|
5114 spin.si_clear_chartab = TRUE;
|
|
|
5115
|
|
236
|
5116 /*
|
|
|
5117 * Read all the .aff and .dic files.
|
|
|
5118 * Text is converted to 'encoding'.
|
|
300
|
5119 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
5120 */
|
|
310
|
5121 for (i = 0; i < incount && !error; ++i)
|
|
236
|
5122 {
|
|
300
|
5123 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
5124 spin.si_region = 1 << i;
|
|
|
5125
|
|
|
5126 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
5127 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
5128 {
|
|
|
5129 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
5130 * "SET" line. */
|
|
310
|
5131 afile[i] = spell_read_aff(fname, &spin);
|
|
|
5132 if (afile[i] == NULL)
|
|
300
|
5133 error = TRUE;
|
|
|
5134 else
|
|
|
5135 {
|
|
|
5136 /* Read the .dic file and store the words in the trees. */
|
|
|
5137 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
5138 innames[i]);
|
|
|
5139 if (spell_read_dic(fname, &spin, afile[i]) == FAIL)
|
|
300
|
5140 error = TRUE;
|
|
|
5141 }
|
|
|
5142 }
|
|
|
5143 else
|
|
|
5144 {
|
|
|
5145 /* No .aff file, try reading the file as a word list. Store
|
|
|
5146 * the words in the trees. */
|
|
310
|
5147 if (spell_read_wordfile(innames[i], &spin) == FAIL)
|
|
300
|
5148 error = TRUE;
|
|
|
5149 }
|
|
236
|
5150
|
|
310
|
5151 #ifdef FEAT_MBYTE
|
|
236
|
5152 /* Free any conversion stuff. */
|
|
300
|
5153 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
5154 #endif
|
|
236
|
5155 }
|
|
|
5156
|
|
300
|
5157 if (!error)
|
|
236
|
5158 {
|
|
|
5159 /*
|
|
300
|
5160 * Remove the dummy NUL from the start of the tree root.
|
|
236
|
5161 */
|
|
300
|
5162 spin.si_foldroot = spin.si_foldroot->wn_sibling;
|
|
|
5163 spin.si_keeproot = spin.si_keeproot->wn_sibling;
|
|
339
|
5164 spin.si_prefroot = spin.si_prefroot->wn_sibling;
|
|
236
|
5165
|
|
|
5166 /*
|
|
300
|
5167 * Combine tails in the tree.
|
|
236
|
5168 */
|
|
323
|
5169 if (!added_word || p_verbose > 2)
|
|
310
|
5170 {
|
|
323
|
5171 if (added_word)
|
|
310
|
5172 verbose_enter();
|
|
|
5173 MSG(_("Compressing word tree..."));
|
|
|
5174 out_flush();
|
|
323
|
5175 if (added_word)
|
|
310
|
5176 verbose_leave();
|
|
|
5177 }
|
|
|
5178 wordtree_compress(spin.si_foldroot, &spin);
|
|
|
5179 wordtree_compress(spin.si_keeproot, &spin);
|
|
339
|
5180 wordtree_compress(spin.si_prefroot, &spin);
|
|
236
|
5181 }
|
|
|
5182
|
|
300
|
5183 if (!error)
|
|
|
5184 {
|
|
|
5185 /*
|
|
|
5186 * Write the info in the spell file.
|
|
|
5187 */
|
|
323
|
5188 if (!added_word || p_verbose > 2)
|
|
310
|
5189 {
|
|
323
|
5190 if (added_word)
|
|
310
|
5191 verbose_enter();
|
|
366
|
5192 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
5193 out_flush();
|
|
323
|
5194 if (added_word)
|
|
310
|
5195 verbose_leave();
|
|
|
5196 }
|
|
|
5197
|
|
316
|
5198 write_vim_spell(wfname, &spin);
|
|
310
|
5199
|
|
323
|
5200 if (!added_word || p_verbose > 2)
|
|
310
|
5201 {
|
|
323
|
5202 if (added_word)
|
|
310
|
5203 verbose_enter();
|
|
|
5204 MSG(_("Done!"));
|
|
|
5205 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
5206 spin.si_memtot);
|
|
310
|
5207 out_flush();
|
|
323
|
5208 if (added_word)
|
|
310
|
5209 verbose_leave();
|
|
|
5210 }
|
|
|
5211
|
|
|
5212 /* If the file is loaded need to reload it. */
|
|
323
|
5213 spell_reload_one(wfname, added_word);
|
|
300
|
5214 }
|
|
|
5215
|
|
|
5216 /* Free the allocated memory. */
|
|
323
|
5217 ga_clear(&spin.si_rep);
|
|
|
5218 ga_clear(&spin.si_sal);
|
|
|
5219 ga_clear(&spin.si_map);
|
|
339
|
5220 ga_clear(&spin.si_prefcond);
|
|
366
|
5221 vim_free(spin.si_midword);
|
|
375
|
5222 vim_free(spin.si_sofofr);
|
|
|
5223 vim_free(spin.si_sofoto);
|
|
300
|
5224
|
|
|
5225 /* Free the .aff file structures. */
|
|
310
|
5226 for (i = 0; i < incount; ++i)
|
|
|
5227 if (afile[i] != NULL)
|
|
|
5228 spell_free_aff(afile[i]);
|
|
339
|
5229
|
|
|
5230 /* Free all the bits and pieces at once. */
|
|
|
5231 free_blocks(spin.si_blocks);
|
|
236
|
5232 }
|
|
310
|
5233 }
|
|
|
5234
|
|
|
5235
|
|
|
5236 /*
|
|
385
|
5237 * ":[count]spellgood {word}"
|
|
|
5238 * ":[count]spellwrong {word}"
|
|
310
|
5239 */
|
|
|
5240 void
|
|
|
5241 ex_spell(eap)
|
|
|
5242 exarg_T *eap;
|
|
|
5243 {
|
|
381
|
5244 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
5245 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
5246 }
|
|
|
5247
|
|
310
|
5248 /*
|
|
|
5249 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
5250 */
|
|
|
5251 void
|
|
385
|
5252 spell_add_word(word, len, bad, index)
|
|
310
|
5253 char_u *word;
|
|
|
5254 int len;
|
|
|
5255 int bad;
|
|
385
|
5256 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
5257 'spellfile' */
|
|
310
|
5258 {
|
|
|
5259 FILE *fd;
|
|
385
|
5260 buf_T *buf = NULL;
|
|
351
|
5261 int new_spf = FALSE;
|
|
|
5262 struct stat st;
|
|
381
|
5263 char_u *fname;
|
|
385
|
5264 char_u fnamebuf[MAXPATHL];
|
|
|
5265 char_u line[MAXWLEN * 2];
|
|
|
5266 long fpos, fpos_next = 0;
|
|
|
5267 int i;
|
|
|
5268 char_u *spf;
|
|
|
5269
|
|
|
5270 if (index == 0) /* use internal wordlist */
|
|
|
5271 {
|
|
|
5272 if (int_wordlist == NULL)
|
|
|
5273 {
|
|
|
5274 int_wordlist = vim_tempname('s');
|
|
|
5275 if (int_wordlist == NULL)
|
|
381
|
5276 return;
|
|
|
5277 }
|
|
385
|
5278 fname = int_wordlist;
|
|
381
|
5279 }
|
|
310
|
5280 else
|
|
|
5281 {
|
|
381
|
5282 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
5283 if (*curbuf->b_p_spf == NUL)
|
|
|
5284 {
|
|
|
5285 init_spellfile();
|
|
|
5286 new_spf = TRUE;
|
|
|
5287 }
|
|
|
5288
|
|
|
5289 if (*curbuf->b_p_spf == NUL)
|
|
|
5290 {
|
|
|
5291 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
5292 return;
|
|
|
5293 }
|
|
|
5294
|
|
385
|
5295 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
5296 {
|
|
|
5297 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
5298 if (i == index)
|
|
|
5299 break;
|
|
|
5300 if (*spf == NUL)
|
|
|
5301 {
|
|
|
5302 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
5303 return;
|
|
|
5304 }
|
|
|
5305 }
|
|
|
5306
|
|
310
|
5307 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
5308 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
5309 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
5310 buf = NULL;
|
|
|
5311 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
5312 {
|
|
310
|
5313 EMSG(_(e_bufloaded));
|
|
381
|
5314 return;
|
|
|
5315 }
|
|
|
5316
|
|
385
|
5317 fname = fnamebuf;
|
|
|
5318 }
|
|
|
5319
|
|
|
5320 if (bad)
|
|
|
5321 {
|
|
|
5322 /* When the word also appears as good word we need to remove that one,
|
|
|
5323 * since its flags sort before the one with WF_BANNED. */
|
|
|
5324 fd = mch_fopen((char *)fname, "r");
|
|
|
5325 if (fd != NULL)
|
|
|
5326 {
|
|
|
5327 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
5328 {
|
|
|
5329 fpos = fpos_next;
|
|
|
5330 fpos_next = ftell(fd);
|
|
|
5331 if (STRNCMP(word, line, len) == 0
|
|
|
5332 && (line[len] == '/' || line[len] < ' '))
|
|
|
5333 {
|
|
|
5334 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
5335 * the start of the line. Mixing reading and writing
|
|
|
5336 * doesn't work for all systems, close the file first. */
|
|
|
5337 fclose(fd);
|
|
|
5338 fd = mch_fopen((char *)fname, "r+");
|
|
|
5339 if (fd == NULL)
|
|
|
5340 break;
|
|
|
5341 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
5342 fputc('#', fd);
|
|
|
5343 fseek(fd, fpos_next, SEEK_SET);
|
|
|
5344 }
|
|
|
5345 }
|
|
|
5346 fclose(fd);
|
|
|
5347 }
|
|
381
|
5348 }
|
|
|
5349
|
|
|
5350 fd = mch_fopen((char *)fname, "a");
|
|
|
5351 if (fd == NULL && new_spf)
|
|
|
5352 {
|
|
|
5353 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
5354 * We may need to create the "spell" directory first. We already
|
|
|
5355 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
5356 STRCPY(NameBuff, fname);
|
|
|
5357 *gettail_sep(NameBuff) = NUL;
|
|
|
5358 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
5359 {
|
|
|
5360 /* The directory doesn't exist. Try creating it and opening the
|
|
|
5361 * file again. */
|
|
|
5362 vim_mkdir(NameBuff, 0755);
|
|
|
5363 fd = mch_fopen((char *)fname, "a");
|
|
|
5364 }
|
|
|
5365 }
|
|
|
5366
|
|
|
5367 if (fd == NULL)
|
|
|
5368 EMSG2(_(e_notopen), fname);
|
|
|
5369 else
|
|
|
5370 {
|
|
|
5371 if (bad)
|
|
|
5372 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
5373 else
|
|
381
|
5374 fprintf(fd, "%.*s\n", len, word);
|
|
|
5375 fclose(fd);
|
|
|
5376
|
|
|
5377 /* Update the .add.spl file. */
|
|
|
5378 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
5379
|
|
|
5380 /* If the .add file is edited somewhere, reload it. */
|
|
|
5381 if (buf != NULL)
|
|
|
5382 buf_reload(buf);
|
|
|
5383
|
|
|
5384 redraw_all_later(NOT_VALID);
|
|
310
|
5385 }
|
|
|
5386 }
|
|
|
5387
|
|
|
5388 /*
|
|
|
5389 * Initialize 'spellfile' for the current buffer.
|
|
|
5390 */
|
|
|
5391 static void
|
|
|
5392 init_spellfile()
|
|
|
5393 {
|
|
|
5394 char_u buf[MAXPATHL];
|
|
|
5395 int l;
|
|
|
5396 slang_T *sl;
|
|
|
5397 char_u *rtp;
|
|
351
|
5398 char_u *lend;
|
|
310
|
5399
|
|
|
5400 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
5401 {
|
|
351
|
5402 /* Find the end of the language name. Exclude the region. */
|
|
|
5403 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
5404 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
5405 ;
|
|
|
5406
|
|
|
5407 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
5408 * are allowed to write. */
|
|
310
|
5409 rtp = p_rtp;
|
|
|
5410 while (*rtp != NUL)
|
|
|
5411 {
|
|
|
5412 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
5413 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
5414 if (filewritable(buf) == 2)
|
|
|
5415 {
|
|
316
|
5416 /* Use the first language name from 'spelllang' and the
|
|
|
5417 * encoding used in the first loaded .spl file. */
|
|
310
|
5418 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
5419 l = STRLEN(buf);
|
|
|
5420 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
5421 "/spell/%.*s.%s.add",
|
|
351
|
5422 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
5423 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
5424 ? (char_u *)"ascii" : spell_enc());
|
|
|
5425 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
5426 break;
|
|
|
5427 }
|
|
|
5428 }
|
|
|
5429 }
|
|
|
5430 }
|
|
236
|
5431
|
|
300
|
5432
|
|
307
|
5433 /*
|
|
|
5434 * Init the chartab used for spelling for ASCII.
|
|
|
5435 * EBCDIC is not supported!
|
|
|
5436 */
|
|
|
5437 static void
|
|
|
5438 clear_spell_chartab(sp)
|
|
|
5439 spelltab_T *sp;
|
|
|
5440 {
|
|
324
|
5441 int i;
|
|
307
|
5442
|
|
|
5443 /* Init everything to FALSE. */
|
|
|
5444 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
5445 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
5446 for (i = 0; i < 256; ++i)
|
|
324
|
5447 {
|
|
307
|
5448 sp->st_fold[i] = i;
|
|
324
|
5449 sp->st_upper[i] = i;
|
|
|
5450 }
|
|
307
|
5451
|
|
|
5452 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
5453 * that is done separately. */
|
|
|
5454 for (i = '0'; i <= '9'; ++i)
|
|
|
5455 sp->st_isw[i] = TRUE;
|
|
|
5456 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
5457 {
|
|
|
5458 sp->st_isw[i] = TRUE;
|
|
|
5459 sp->st_isu[i] = TRUE;
|
|
|
5460 sp->st_fold[i] = i + 0x20;
|
|
|
5461 }
|
|
|
5462 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
5463 {
|
|
307
|
5464 sp->st_isw[i] = TRUE;
|
|
324
|
5465 sp->st_upper[i] = i - 0x20;
|
|
|
5466 }
|
|
307
|
5467 }
|
|
|
5468
|
|
|
5469 /*
|
|
|
5470 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
5471 * Called once while starting up and when 'encoding' changes.
|
|
|
5472 * The default is to use isalpha(), but the spell file should define the word
|
|
|
5473 * characters to make it possible that 'encoding' differs from the current
|
|
|
5474 * locale.
|
|
|
5475 */
|
|
|
5476 void
|
|
|
5477 init_spell_chartab()
|
|
|
5478 {
|
|
|
5479 int i;
|
|
|
5480
|
|
|
5481 did_set_spelltab = FALSE;
|
|
|
5482 clear_spell_chartab(&spelltab);
|
|
|
5483 #ifdef FEAT_MBYTE
|
|
|
5484 if (enc_dbcs)
|
|
|
5485 {
|
|
|
5486 /* DBCS: assume double-wide characters are word characters. */
|
|
|
5487 for (i = 128; i <= 255; ++i)
|
|
|
5488 if (MB_BYTE2LEN(i) == 2)
|
|
|
5489 spelltab.st_isw[i] = TRUE;
|
|
|
5490 }
|
|
324
|
5491 else if (enc_utf8)
|
|
|
5492 {
|
|
|
5493 for (i = 128; i < 256; ++i)
|
|
|
5494 {
|
|
|
5495 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
5496 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
5497 spelltab.st_fold[i] = utf_fold(i);
|
|
|
5498 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
5499 }
|
|
|
5500 }
|
|
307
|
5501 else
|
|
|
5502 #endif
|
|
|
5503 {
|
|
324
|
5504 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
5505 for (i = 128; i < 256; ++i)
|
|
|
5506 {
|
|
|
5507 if (MB_ISUPPER(i))
|
|
|
5508 {
|
|
324
|
5509 spelltab.st_isw[i] = TRUE;
|
|
307
|
5510 spelltab.st_isu[i] = TRUE;
|
|
|
5511 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
5512 }
|
|
324
|
5513 else if (MB_ISLOWER(i))
|
|
|
5514 {
|
|
|
5515 spelltab.st_isw[i] = TRUE;
|
|
|
5516 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
5517 }
|
|
307
|
5518 }
|
|
|
5519 }
|
|
|
5520 }
|
|
|
5521
|
|
|
5522 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
5523 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
5524
|
|
|
5525 /*
|
|
|
5526 * Set the spell character tables from strings in the affix file.
|
|
|
5527 */
|
|
|
5528 static int
|
|
|
5529 set_spell_chartab(fol, low, upp)
|
|
|
5530 char_u *fol;
|
|
|
5531 char_u *low;
|
|
|
5532 char_u *upp;
|
|
|
5533 {
|
|
|
5534 /* We build the new tables here first, so that we can compare with the
|
|
|
5535 * previous one. */
|
|
|
5536 spelltab_T new_st;
|
|
|
5537 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
5538 int f, l, u;
|
|
|
5539
|
|
|
5540 clear_spell_chartab(&new_st);
|
|
|
5541
|
|
|
5542 while (*pf != NUL)
|
|
|
5543 {
|
|
|
5544 if (*pl == NUL || *pu == NUL)
|
|
|
5545 {
|
|
|
5546 EMSG(_(e_affform));
|
|
|
5547 return FAIL;
|
|
|
5548 }
|
|
|
5549 #ifdef FEAT_MBYTE
|
|
|
5550 f = mb_ptr2char_adv(&pf);
|
|
|
5551 l = mb_ptr2char_adv(&pl);
|
|
|
5552 u = mb_ptr2char_adv(&pu);
|
|
|
5553 #else
|
|
|
5554 f = *pf++;
|
|
|
5555 l = *pl++;
|
|
|
5556 u = *pu++;
|
|
|
5557 #endif
|
|
|
5558 /* Every character that appears is a word character. */
|
|
|
5559 if (f < 256)
|
|
|
5560 new_st.st_isw[f] = TRUE;
|
|
|
5561 if (l < 256)
|
|
|
5562 new_st.st_isw[l] = TRUE;
|
|
|
5563 if (u < 256)
|
|
|
5564 new_st.st_isw[u] = TRUE;
|
|
|
5565
|
|
|
5566 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
5567 * case-folding */
|
|
|
5568 if (l < 256 && l != f)
|
|
|
5569 {
|
|
|
5570 if (f >= 256)
|
|
|
5571 {
|
|
|
5572 EMSG(_(e_affrange));
|
|
|
5573 return FAIL;
|
|
|
5574 }
|
|
|
5575 new_st.st_fold[l] = f;
|
|
|
5576 }
|
|
|
5577
|
|
|
5578 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
5579 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
5580 * "FOL" . */
|
|
307
|
5581 if (u < 256 && u != f)
|
|
|
5582 {
|
|
|
5583 if (f >= 256)
|
|
|
5584 {
|
|
|
5585 EMSG(_(e_affrange));
|
|
|
5586 return FAIL;
|
|
|
5587 }
|
|
|
5588 new_st.st_fold[u] = f;
|
|
|
5589 new_st.st_isu[u] = TRUE;
|
|
324
|
5590 new_st.st_upper[f] = u;
|
|
307
|
5591 }
|
|
|
5592 }
|
|
|
5593
|
|
|
5594 if (*pl != NUL || *pu != NUL)
|
|
|
5595 {
|
|
|
5596 EMSG(_(e_affform));
|
|
|
5597 return FAIL;
|
|
|
5598 }
|
|
|
5599
|
|
|
5600 return set_spell_finish(&new_st);
|
|
|
5601 }
|
|
|
5602
|
|
|
5603 /*
|
|
|
5604 * Set the spell character tables from strings in the .spl file.
|
|
|
5605 */
|
|
|
5606 static int
|
|
381
|
5607 set_spell_charflags(flags, upp)
|
|
307
|
5608 char_u *flags;
|
|
|
5609 char_u *upp;
|
|
|
5610 {
|
|
|
5611 /* We build the new tables here first, so that we can compare with the
|
|
|
5612 * previous one. */
|
|
|
5613 spelltab_T new_st;
|
|
|
5614 int i;
|
|
|
5615 char_u *p = upp;
|
|
324
|
5616 int c;
|
|
307
|
5617
|
|
|
5618 clear_spell_chartab(&new_st);
|
|
|
5619
|
|
381
|
5620 for (i = 0; flags[i] != NUL; ++i)
|
|
307
|
5621 {
|
|
324
|
5622 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
5623 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
307
|
5624
|
|
|
5625 if (*p == NUL)
|
|
|
5626 return FAIL;
|
|
|
5627 #ifdef FEAT_MBYTE
|
|
324
|
5628 c = mb_ptr2char_adv(&p);
|
|
307
|
5629 #else
|
|
324
|
5630 c = *p++;
|
|
307
|
5631 #endif
|
|
324
|
5632 new_st.st_fold[i + 128] = c;
|
|
|
5633 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
5634 new_st.st_upper[c] = i + 128;
|
|
307
|
5635 }
|
|
|
5636
|
|
|
5637 return set_spell_finish(&new_st);
|
|
|
5638 }
|
|
|
5639
|
|
|
5640 static int
|
|
|
5641 set_spell_finish(new_st)
|
|
|
5642 spelltab_T *new_st;
|
|
|
5643 {
|
|
|
5644 int i;
|
|
|
5645
|
|
|
5646 if (did_set_spelltab)
|
|
|
5647 {
|
|
|
5648 /* check that it's the same table */
|
|
|
5649 for (i = 0; i < 256; ++i)
|
|
|
5650 {
|
|
|
5651 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
5652 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
5653 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
5654 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
5655 {
|
|
|
5656 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
5657 return FAIL;
|
|
|
5658 }
|
|
|
5659 }
|
|
|
5660 }
|
|
|
5661 else
|
|
|
5662 {
|
|
|
5663 /* copy the new spelltab into the one being used */
|
|
|
5664 spelltab = *new_st;
|
|
|
5665 did_set_spelltab = TRUE;
|
|
|
5666 }
|
|
|
5667
|
|
|
5668 return OK;
|
|
|
5669 }
|
|
|
5670
|
|
|
5671 /*
|
|
358
|
5672 * Return TRUE if "p" points to a word character.
|
|
366
|
5673 * As a special case we see "midword" characters as word character when it is
|
|
358
|
5674 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
5675 * Thus this only works properly when past the first character of the word.
|
|
358
|
5676 */
|
|
|
5677 static int
|
|
376
|
5678 spell_iswordp(p, buf)
|
|
358
|
5679 char_u *p;
|
|
376
|
5680 buf_T *buf; /* buffer used */
|
|
358
|
5681 {
|
|
366
|
5682 #ifdef FEAT_MBYTE
|
|
358
|
5683 char_u *s;
|
|
366
|
5684 int l;
|
|
|
5685 int c;
|
|
|
5686
|
|
|
5687 if (has_mbyte)
|
|
|
5688 {
|
|
|
5689 l = MB_BYTE2LEN(*p);
|
|
358
|
5690 s = p;
|
|
366
|
5691 if (l == 1)
|
|
|
5692 {
|
|
|
5693 /* be quick for ASCII */
|
|
376
|
5694 if (buf->b_spell_ismw[*p])
|
|
366
|
5695 {
|
|
|
5696 s = p + 1; /* skip a mid-word character */
|
|
|
5697 l = MB_BYTE2LEN(*s);
|
|
|
5698 }
|
|
|
5699 }
|
|
|
5700 else
|
|
|
5701 {
|
|
|
5702 c = mb_ptr2char(p);
|
|
376
|
5703 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
5704 : (buf->b_spell_ismw_mb != NULL
|
|
|
5705 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
5706 {
|
|
|
5707 s = p + l;
|
|
|
5708 l = MB_BYTE2LEN(*s);
|
|
|
5709 }
|
|
|
5710 }
|
|
|
5711
|
|
|
5712 if (l > 1)
|
|
|
5713 return mb_get_class(s) >= 2;
|
|
|
5714 return spelltab.st_isw[*s];
|
|
|
5715 }
|
|
358
|
5716 #endif
|
|
366
|
5717
|
|
376
|
5718 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
5719 }
|
|
|
5720
|
|
|
5721 /*
|
|
|
5722 * Return TRUE if "p" points to a word character.
|
|
|
5723 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
5724 */
|
|
|
5725 static int
|
|
|
5726 spell_iswordp_nmw(p)
|
|
|
5727 char_u *p;
|
|
|
5728 {
|
|
|
5729 #ifdef FEAT_MBYTE
|
|
|
5730 if (has_mbyte && MB_BYTE2LEN(*p) > 1)
|
|
|
5731 return mb_get_class(p) >= 2;
|
|
|
5732 #endif
|
|
|
5733
|
|
|
5734 return spelltab.st_isw[*p];
|
|
358
|
5735 }
|
|
|
5736
|
|
372
|
5737 #ifdef FEAT_MBYTE
|
|
|
5738 /*
|
|
|
5739 * Return TRUE if "p" points to a word character.
|
|
|
5740 * Wide version of spell_iswordp().
|
|
|
5741 */
|
|
|
5742 static int
|
|
376
|
5743 spell_iswordp_w(p, buf)
|
|
372
|
5744 int *p;
|
|
376
|
5745 buf_T *buf;
|
|
372
|
5746 {
|
|
|
5747 int *s;
|
|
|
5748
|
|
376
|
5749 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
5750 : (buf->b_spell_ismw_mb != NULL
|
|
|
5751 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
5752 s = p + 1;
|
|
|
5753 else
|
|
|
5754 s = p;
|
|
|
5755
|
|
|
5756 if (mb_char2len(*s) > 1)
|
|
|
5757 {
|
|
|
5758 if (enc_utf8)
|
|
|
5759 return utf_class(*s) >= 2;
|
|
|
5760 if (enc_dbcs)
|
|
|
5761 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
5762 return 0;
|
|
|
5763 }
|
|
|
5764 return spelltab.st_isw[*s];
|
|
|
5765 }
|
|
|
5766 #endif
|
|
|
5767
|
|
358
|
5768 /*
|
|
339
|
5769 * Write the table with prefix conditions to the .spl file.
|
|
|
5770 */
|
|
|
5771 static void
|
|
|
5772 write_spell_prefcond(fd, gap)
|
|
|
5773 FILE *fd;
|
|
|
5774 garray_T *gap;
|
|
|
5775 {
|
|
|
5776 int i;
|
|
|
5777 char_u *p;
|
|
|
5778 int len;
|
|
|
5779
|
|
|
5780 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
5781
|
|
|
5782 for (i = 0; i < gap->ga_len; ++i)
|
|
|
5783 {
|
|
|
5784 /* <prefcond> : <condlen> <condstr> */
|
|
|
5785 p = ((char_u **)gap->ga_data)[i];
|
|
|
5786 if (p == NULL)
|
|
|
5787 fputc(0, fd);
|
|
|
5788 else
|
|
|
5789 {
|
|
|
5790 len = STRLEN(p);
|
|
|
5791 fputc(len, fd);
|
|
|
5792 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
5793 }
|
|
|
5794 }
|
|
|
5795 }
|
|
|
5796
|
|
|
5797 /*
|
|
307
|
5798 * Write the current tables into the .spl file.
|
|
|
5799 * This makes sure the same characters are recognized as word characters when
|
|
|
5800 * generating an when using a spell file.
|
|
|
5801 */
|
|
|
5802 static void
|
|
|
5803 write_spell_chartab(fd)
|
|
|
5804 FILE *fd;
|
|
|
5805 {
|
|
|
5806 char_u charbuf[256 * 4];
|
|
|
5807 int len = 0;
|
|
|
5808 int flags;
|
|
|
5809 int i;
|
|
|
5810
|
|
|
5811 fputc(128, fd); /* <charflagslen> */
|
|
|
5812 for (i = 128; i < 256; ++i)
|
|
|
5813 {
|
|
|
5814 flags = 0;
|
|
|
5815 if (spelltab.st_isw[i])
|
|
324
|
5816 flags |= CF_WORD;
|
|
307
|
5817 if (spelltab.st_isu[i])
|
|
324
|
5818 flags |= CF_UPPER;
|
|
307
|
5819 fputc(flags, fd); /* <charflags> */
|
|
|
5820
|
|
310
|
5821 #ifdef FEAT_MBYTE
|
|
|
5822 if (has_mbyte)
|
|
|
5823 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
|
5824 else
|
|
|
5825 #endif
|
|
|
5826 charbuf[len++] = spelltab.st_fold[i];
|
|
307
|
5827 }
|
|
|
5828
|
|
|
5829 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
|
5830 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
|
5831 }
|
|
|
5832
|
|
|
5833 /*
|
|
324
|
5834 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
5835 * Uses the character definitions from the .spl file.
|
|
307
|
5836 * When using a multi-byte 'encoding' the length may change!
|
|
|
5837 * Returns FAIL when something wrong.
|
|
|
5838 */
|
|
|
5839 static int
|
|
324
|
5840 spell_casefold(str, len, buf, buflen)
|
|
|
5841 char_u *str;
|
|
307
|
5842 int len;
|
|
|
5843 char_u *buf;
|
|
|
5844 int buflen;
|
|
|
5845 {
|
|
|
5846 int i;
|
|
|
5847
|
|
|
5848 if (len >= buflen)
|
|
|
5849 {
|
|
|
5850 buf[0] = NUL;
|
|
|
5851 return FAIL; /* result will not fit */
|
|
|
5852 }
|
|
|
5853
|
|
|
5854 #ifdef FEAT_MBYTE
|
|
|
5855 if (has_mbyte)
|
|
|
5856 {
|
|
324
|
5857 int outi = 0;
|
|
|
5858 char_u *p;
|
|
307
|
5859 int c;
|
|
|
5860
|
|
|
5861 /* Fold one character at a time. */
|
|
324
|
5862 for (p = str; p < str + len; )
|
|
307
|
5863 {
|
|
|
5864 if (outi + MB_MAXBYTES > buflen)
|
|
|
5865 {
|
|
|
5866 buf[outi] = NUL;
|
|
|
5867 return FAIL;
|
|
|
5868 }
|
|
324
|
5869 c = mb_ptr2char_adv(&p);
|
|
|
5870 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
5871 }
|
|
|
5872 buf[outi] = NUL;
|
|
|
5873 }
|
|
|
5874 else
|
|
|
5875 #endif
|
|
|
5876 {
|
|
|
5877 /* Be quick for non-multibyte encodings. */
|
|
|
5878 for (i = 0; i < len; ++i)
|
|
324
|
5879 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
5880 buf[i] = NUL;
|
|
|
5881 }
|
|
|
5882
|
|
|
5883 return OK;
|
|
|
5884 }
|
|
|
5885
|
|
372
|
5886 #define SPS_BEST 1
|
|
|
5887 #define SPS_FAST 2
|
|
|
5888 #define SPS_DOUBLE 4
|
|
|
5889
|
|
|
5890 static int sps_flags = SPS_BEST;
|
|
|
5891
|
|
|
5892 /*
|
|
|
5893 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
|
5894 * Sets "sps_flags".
|
|
|
5895 */
|
|
|
5896 int
|
|
|
5897 spell_check_sps()
|
|
|
5898 {
|
|
|
5899 char_u *p;
|
|
|
5900 char_u buf[MAXPATHL];
|
|
|
5901 int f;
|
|
|
5902
|
|
|
5903 sps_flags = 0;
|
|
|
5904
|
|
|
5905 for (p = p_sps; *p != NUL; )
|
|
|
5906 {
|
|
|
5907 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
5908
|
|
|
5909 f = 0;
|
|
|
5910 if (STRCMP(buf, "best") == 0)
|
|
|
5911 f = SPS_BEST;
|
|
|
5912 else if (STRCMP(buf, "fast") == 0)
|
|
|
5913 f = SPS_FAST;
|
|
|
5914 else if (STRCMP(buf, "double") == 0)
|
|
|
5915 f = SPS_DOUBLE;
|
|
|
5916 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
5917 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
5918 f = -1;
|
|
|
5919
|
|
|
5920 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
5921 {
|
|
|
5922 sps_flags = SPS_BEST;
|
|
|
5923 return FAIL;
|
|
|
5924 }
|
|
|
5925 if (f != 0)
|
|
|
5926 sps_flags = f;
|
|
|
5927 }
|
|
|
5928
|
|
|
5929 if (sps_flags == 0)
|
|
|
5930 sps_flags = SPS_BEST;
|
|
|
5931
|
|
|
5932 return OK;
|
|
|
5933 }
|
|
|
5934
|
|
|
5935 /* Remember what "z?" replaced. */
|
|
|
5936 static char_u *repl_from = NULL;
|
|
|
5937 static char_u *repl_to = NULL;
|
|
|
5938
|
|
323
|
5939 /*
|
|
|
5940 * "z?": Find badly spelled word under or after the cursor.
|
|
|
5941 * Give suggestions for the properly spelled word.
|
|
|
5942 */
|
|
|
5943 void
|
|
|
5944 spell_suggest()
|
|
|
5945 {
|
|
|
5946 char_u *line;
|
|
|
5947 pos_T prev_cursor = curwin->w_cursor;
|
|
|
5948 char_u wcopy[MAXWLEN + 2];
|
|
|
5949 char_u *p;
|
|
|
5950 int i;
|
|
|
5951 int c;
|
|
|
5952 suginfo_T sug;
|
|
|
5953 suggest_T *stp;
|
|
372
|
5954 int mouse_used;
|
|
388
|
5955 int need_cap;
|
|
|
5956 regmatch_T regmatch;
|
|
|
5957 int endcol;
|
|
|
5958 char_u *line_copy = NULL;
|
|
323
|
5959
|
|
344
|
5960 /* Find the start of the badly spelled word. */
|
|
346
|
5961 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
5962 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
5963 {
|
|
|
5964 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
5965 return;
|
|
|
5966
|
|
|
5967 /* No bad word or it starts after the cursor: use the word under the
|
|
|
5968 * cursor. */
|
|
|
5969 curwin->w_cursor = prev_cursor;
|
|
|
5970 line = ml_get_curline();
|
|
|
5971 p = line + curwin->w_cursor.col;
|
|
|
5972 /* Backup to before start of word. */
|
|
|
5973 while (p > line && SPELL_ISWORDP(p))
|
|
|
5974 mb_ptr_back(line, p);
|
|
|
5975 /* Forward to start of word. */
|
|
375
|
5976 while (*p != NUL && !SPELL_ISWORDP(p))
|
|
346
|
5977 mb_ptr_adv(p);
|
|
|
5978
|
|
|
5979 if (!SPELL_ISWORDP(p)) /* No word found. */
|
|
|
5980 {
|
|
|
5981 beep_flush();
|
|
|
5982 return;
|
|
|
5983 }
|
|
|
5984 curwin->w_cursor.col = p - line;
|
|
323
|
5985 }
|
|
|
5986
|
|
344
|
5987 /* Get the word and its length. */
|
|
323
|
5988 line = ml_get_curline();
|
|
344
|
5989
|
|
388
|
5990 /* Figure out if the word should be capitalised. */
|
|
|
5991 need_cap = FALSE;
|
|
|
5992 if (curbuf->b_cap_prog != NULL)
|
|
|
5993 {
|
|
|
5994 endcol = 0;
|
|
|
5995 if (skipwhite(line) - line == curwin->w_cursor.col)
|
|
|
5996 {
|
|
|
5997 /* At start of line, check if previous line is empty or sentence
|
|
|
5998 * ends there. */
|
|
|
5999 if (curwin->w_cursor.lnum == 1)
|
|
|
6000 need_cap = TRUE;
|
|
|
6001 else
|
|
|
6002 {
|
|
|
6003 line = ml_get(curwin->w_cursor.lnum - 1);
|
|
|
6004 if (*skipwhite(line) == NUL)
|
|
|
6005 need_cap = TRUE;
|
|
|
6006 else
|
|
|
6007 {
|
|
|
6008 /* Append a space in place of the line break. */
|
|
|
6009 line_copy = concat_str(line, (char_u *)" ");
|
|
|
6010 line = line_copy;
|
|
|
6011 endcol = STRLEN(line);
|
|
|
6012 }
|
|
|
6013 }
|
|
|
6014 }
|
|
|
6015 else
|
|
|
6016 endcol = curwin->w_cursor.col;
|
|
|
6017
|
|
|
6018 if (endcol > 0)
|
|
|
6019 {
|
|
|
6020 /* Check if sentence ends before the bad word. */
|
|
|
6021 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
6022 regmatch.rm_ic = FALSE;
|
|
|
6023 p = line + endcol;
|
|
|
6024 for (;;)
|
|
|
6025 {
|
|
|
6026 mb_ptr_back(line, p);
|
|
|
6027 if (p == line || SPELL_ISWORDP(p))
|
|
|
6028 break;
|
|
|
6029 if (vim_regexec(®match, p, 0)
|
|
|
6030 && regmatch.endp[0] == line + endcol)
|
|
|
6031 {
|
|
|
6032 need_cap = TRUE;
|
|
|
6033 break;
|
|
|
6034 }
|
|
|
6035 }
|
|
|
6036 }
|
|
|
6037
|
|
|
6038 /* get the line again, we may have been using the previous one */
|
|
|
6039 line = ml_get_curline();
|
|
|
6040 vim_free(line_copy);
|
|
|
6041 }
|
|
|
6042
|
|
344
|
6043 /* Get the list of suggestions */
|
|
388
|
6044 spell_find_suggest(line + curwin->w_cursor.col, &sug, (int)Rows - 2,
|
|
|
6045 TRUE, need_cap);
|
|
323
|
6046
|
|
|
6047 if (sug.su_ga.ga_len == 0)
|
|
|
6048 MSG(_("Sorry, no suggestions"));
|
|
|
6049 else
|
|
|
6050 {
|
|
372
|
6051 vim_free(repl_from);
|
|
|
6052 repl_from = NULL;
|
|
|
6053 vim_free(repl_to);
|
|
|
6054 repl_to = NULL;
|
|
|
6055
|
|
323
|
6056 /* List the suggestions. */
|
|
|
6057 msg_start();
|
|
372
|
6058 lines_left = Rows; /* avoid more prompt */
|
|
323
|
6059 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
6060 sug.su_badlen, sug.su_badptr);
|
|
|
6061 msg_puts(IObuff);
|
|
|
6062 msg_clr_eos();
|
|
|
6063 msg_putchar('\n');
|
|
346
|
6064
|
|
323
|
6065 msg_scroll = TRUE;
|
|
|
6066 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
6067 {
|
|
344
|
6068 stp = &SUG(sug.su_ga, i);
|
|
323
|
6069
|
|
|
6070 /* The suggested word may replace only part of the bad word, add
|
|
|
6071 * the not replaced part. */
|
|
|
6072 STRCPY(wcopy, stp->st_word);
|
|
|
6073 if (sug.su_badlen > stp->st_orglen)
|
|
|
6074 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
6075 sug.su_badptr + stp->st_orglen,
|
|
|
6076 sug.su_badlen - stp->st_orglen);
|
|
346
|
6077 vim_snprintf((char *)IObuff, IOSIZE, _("%2d \"%s\""), i + 1, wcopy);
|
|
|
6078 msg_puts(IObuff);
|
|
|
6079
|
|
|
6080 /* The word may replace more than "su_badlen". */
|
|
|
6081 if (sug.su_badlen < stp->st_orglen)
|
|
|
6082 {
|
|
|
6083 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
6084 stp->st_orglen, sug.su_badptr);
|
|
|
6085 msg_puts(IObuff);
|
|
|
6086 }
|
|
|
6087
|
|
324
|
6088 if (p_verbose > 0)
|
|
344
|
6089 {
|
|
346
|
6090 /* Add the score. */
|
|
351
|
6091 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
346
|
6092 vim_snprintf((char *)IObuff, IOSIZE, _(" (%s%d - %d)"),
|
|
344
|
6093 stp->st_salscore ? "s " : "",
|
|
|
6094 stp->st_score, stp->st_altscore);
|
|
|
6095 else
|
|
346
|
6096 vim_snprintf((char *)IObuff, IOSIZE, _(" (%d)"),
|
|
|
6097 stp->st_score);
|
|
|
6098 msg_advance(30);
|
|
|
6099 msg_puts(IObuff);
|
|
344
|
6100 }
|
|
323
|
6101 msg_putchar('\n');
|
|
|
6102 }
|
|
|
6103
|
|
|
6104 /* Ask for choice. */
|
|
372
|
6105 i = prompt_for_number(&mouse_used);
|
|
|
6106 if (mouse_used)
|
|
|
6107 i -= lines_left;
|
|
|
6108
|
|
344
|
6109 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
6110 {
|
|
372
|
6111 /* Save the from and to text for :spellrepall. */
|
|
|
6112 stp = &SUG(sug.su_ga, i - 1);
|
|
|
6113 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
6114 repl_to = vim_strsave(stp->st_word);
|
|
|
6115
|
|
323
|
6116 /* Replace the word. */
|
|
|
6117 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
6118 if (p != NULL)
|
|
|
6119 {
|
|
|
6120 c = sug.su_badptr - line;
|
|
|
6121 mch_memmove(p, line, c);
|
|
|
6122 STRCPY(p + c, stp->st_word);
|
|
|
6123 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
6124 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6125 curwin->w_cursor.col = c;
|
|
|
6126 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
6127
|
|
|
6128 /* For redo we use a change-word command. */
|
|
|
6129 ResetRedobuff();
|
|
|
6130 AppendToRedobuff((char_u *)"ciw");
|
|
|
6131 AppendToRedobuff(stp->st_word);
|
|
|
6132 AppendCharToRedobuff(ESC);
|
|
323
|
6133 }
|
|
|
6134 }
|
|
|
6135 else
|
|
|
6136 curwin->w_cursor = prev_cursor;
|
|
|
6137 }
|
|
|
6138
|
|
344
|
6139 spell_find_cleanup(&sug);
|
|
|
6140 }
|
|
|
6141
|
|
|
6142 /*
|
|
372
|
6143 * ":spellrepall"
|
|
|
6144 */
|
|
|
6145 /*ARGSUSED*/
|
|
|
6146 void
|
|
|
6147 ex_spellrepall(eap)
|
|
|
6148 exarg_T *eap;
|
|
|
6149 {
|
|
|
6150 pos_T pos = curwin->w_cursor;
|
|
|
6151 char_u *frompat;
|
|
|
6152 int addlen;
|
|
|
6153 char_u *line;
|
|
|
6154 char_u *p;
|
|
|
6155 int didone = FALSE;
|
|
|
6156 int save_ws = p_ws;
|
|
|
6157
|
|
|
6158 if (repl_from == NULL || repl_to == NULL)
|
|
|
6159 {
|
|
|
6160 EMSG(_("E752: No previous spell replacement"));
|
|
|
6161 return;
|
|
|
6162 }
|
|
|
6163 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
6164
|
|
|
6165 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
6166 if (frompat == NULL)
|
|
|
6167 return;
|
|
|
6168 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
6169 p_ws = FALSE;
|
|
|
6170
|
|
|
6171 curwin->w_cursor.lnum = 0;
|
|
|
6172 while (!got_int)
|
|
|
6173 {
|
|
|
6174 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
6175 || u_save_cursor() == FAIL)
|
|
|
6176 break;
|
|
|
6177
|
|
|
6178 /* Only replace when the right word isn't there yet. This happens
|
|
|
6179 * when changing "etc" to "etc.". */
|
|
|
6180 line = ml_get_curline();
|
|
|
6181 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
6182 repl_to, STRLEN(repl_to)) != 0)
|
|
|
6183 {
|
|
|
6184 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
6185 if (p == NULL)
|
|
|
6186 break;
|
|
|
6187 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
6188 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
6189 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
6190 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6191 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
6192 didone = TRUE;
|
|
|
6193 }
|
|
|
6194 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
6195 }
|
|
|
6196
|
|
|
6197 p_ws = save_ws;
|
|
|
6198 curwin->w_cursor = pos;
|
|
|
6199 vim_free(frompat);
|
|
|
6200
|
|
|
6201 if (!didone)
|
|
|
6202 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
|
6203 }
|
|
|
6204
|
|
|
6205 /*
|
|
344
|
6206 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
6207 * a list of allocated strings.
|
|
|
6208 */
|
|
|
6209 void
|
|
|
6210 spell_suggest_list(gap, word, maxcount)
|
|
|
6211 garray_T *gap;
|
|
|
6212 char_u *word;
|
|
|
6213 int maxcount; /* maximum nr of suggestions */
|
|
|
6214 {
|
|
|
6215 suginfo_T sug;
|
|
|
6216 int i;
|
|
|
6217 suggest_T *stp;
|
|
|
6218 char_u *wcopy;
|
|
|
6219
|
|
388
|
6220 spell_find_suggest(word, &sug, maxcount, FALSE, FALSE);
|
|
344
|
6221
|
|
|
6222 /* Make room in "gap". */
|
|
|
6223 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
6224 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
6225 return;
|
|
|
6226
|
|
323
|
6227 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
6228 {
|
|
|
6229 stp = &SUG(sug.su_ga, i);
|
|
|
6230
|
|
|
6231 /* The suggested word may replace only part of "word", add the not
|
|
|
6232 * replaced part. */
|
|
|
6233 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
6234 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
6235 if (wcopy == NULL)
|
|
|
6236 break;
|
|
|
6237 STRCPY(wcopy, stp->st_word);
|
|
|
6238 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
6239 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
6240 }
|
|
|
6241
|
|
|
6242 spell_find_cleanup(&sug);
|
|
|
6243 }
|
|
|
6244
|
|
|
6245 /*
|
|
|
6246 * Find spell suggestions for the word at the start of "badptr".
|
|
|
6247 * Return the suggestions in "su->su_ga".
|
|
|
6248 * The maximum number of suggestions is "maxcount".
|
|
|
6249 * Note: does use info for the current window.
|
|
|
6250 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
6251 */
|
|
|
6252 static void
|
|
388
|
6253 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
6254 char_u *badptr;
|
|
|
6255 suginfo_T *su;
|
|
|
6256 int maxcount;
|
|
358
|
6257 int banbadword; /* don't include badword in suggestions */
|
|
388
|
6258 int need_cap; /* word should start with capital */
|
|
344
|
6259 {
|
|
385
|
6260 int attr = 0;
|
|
372
|
6261 char_u buf[MAXPATHL];
|
|
|
6262 char_u *p;
|
|
|
6263 int do_combine = FALSE;
|
|
|
6264 char_u *sps_copy;
|
|
|
6265 #ifdef FEAT_EVAL
|
|
|
6266 static int expr_busy = FALSE;
|
|
|
6267 #endif
|
|
385
|
6268 int c;
|
|
344
|
6269
|
|
|
6270 /*
|
|
|
6271 * Set the info in "*su".
|
|
|
6272 */
|
|
|
6273 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
6274 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
6275 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
6276 if (*badptr == NUL)
|
|
|
6277 return;
|
|
344
|
6278 hash_init(&su->su_banned);
|
|
|
6279
|
|
|
6280 su->su_badptr = badptr;
|
|
385
|
6281 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
6282 su->su_maxcount = maxcount;
|
|
372
|
6283 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
6284
|
|
|
6285 if (su->su_badlen >= MAXWLEN)
|
|
|
6286 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
6287 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
6288 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
6289 su->su_fbadword, MAXWLEN);
|
|
346
|
6290 /* get caps flags for bad word */
|
|
|
6291 su->su_badflags = captype(su->su_badptr, su->su_badptr + su->su_badlen);
|
|
388
|
6292 if (need_cap)
|
|
|
6293 su->su_badflags |= WF_ONECAP;
|
|
344
|
6294
|
|
385
|
6295 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
6296 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
6297 * for that. */
|
|
|
6298 #ifdef FEAT_MBYTE
|
|
|
6299 c = mb_ptr2char(su->su_badptr);
|
|
|
6300 #else
|
|
388
|
6301 c = *su->su_badptr;
|
|
385
|
6302 #endif
|
|
|
6303 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
6304 {
|
|
|
6305 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
6306 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
6307 0, TRUE);
|
|
|
6308 }
|
|
|
6309
|
|
344
|
6310 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
6311 if (banbadword)
|
|
|
6312 add_banned(su, su->su_badword);
|
|
344
|
6313
|
|
372
|
6314 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
6315 sps_copy = vim_strsave(p_sps);
|
|
|
6316 if (sps_copy == NULL)
|
|
|
6317 return;
|
|
|
6318
|
|
|
6319 /* Loop over the items in 'spellsuggest'. */
|
|
|
6320 for (p = sps_copy; *p != NUL; )
|
|
|
6321 {
|
|
|
6322 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6323
|
|
|
6324 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
6325 {
|
|
|
6326 #ifdef FEAT_EVAL
|
|
375
|
6327 /* Evaluate an expression. Skip this when called recursively,
|
|
|
6328 * when using spellsuggest() in the expression. */
|
|
372
|
6329 if (!expr_busy)
|
|
|
6330 {
|
|
|
6331 expr_busy = TRUE;
|
|
|
6332 spell_suggest_expr(su, buf + 5);
|
|
|
6333 expr_busy = FALSE;
|
|
|
6334 }
|
|
|
6335 #endif
|
|
|
6336 }
|
|
|
6337 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
6338 /* Use list of suggestions in a file. */
|
|
|
6339 spell_suggest_file(su, buf + 5);
|
|
|
6340 else
|
|
|
6341 {
|
|
|
6342 /* Use internal method. */
|
|
|
6343 spell_suggest_intern(su);
|
|
|
6344 if (sps_flags & SPS_DOUBLE)
|
|
|
6345 do_combine = TRUE;
|
|
|
6346 }
|
|
|
6347 }
|
|
|
6348
|
|
|
6349 vim_free(sps_copy);
|
|
|
6350
|
|
|
6351 if (do_combine)
|
|
|
6352 /* Combine the two list of suggestions. This must be done last,
|
|
|
6353 * because sorting changes the order again. */
|
|
|
6354 score_combine(su);
|
|
|
6355 }
|
|
|
6356
|
|
|
6357 #ifdef FEAT_EVAL
|
|
|
6358 /*
|
|
|
6359 * Find suggestions by evaluating expression "expr".
|
|
|
6360 */
|
|
|
6361 static void
|
|
|
6362 spell_suggest_expr(su, expr)
|
|
|
6363 suginfo_T *su;
|
|
|
6364 char_u *expr;
|
|
|
6365 {
|
|
|
6366 list_T *list;
|
|
|
6367 listitem_T *li;
|
|
|
6368 int score;
|
|
|
6369 char_u *p;
|
|
|
6370
|
|
|
6371 /* The work is split up in a few parts to avoid having to export
|
|
|
6372 * suginfo_T.
|
|
|
6373 * First evaluate the expression and get the resulting list. */
|
|
|
6374 list = eval_spell_expr(su->su_badword, expr);
|
|
|
6375 if (list != NULL)
|
|
|
6376 {
|
|
|
6377 /* Loop over the items in the list. */
|
|
|
6378 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
6379 if (li->li_tv.v_type == VAR_LIST)
|
|
|
6380 {
|
|
|
6381 /* Get the word and the score from the items. */
|
|
|
6382 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
6383 if (score >= 0)
|
|
|
6384 add_suggestion(su, &su->su_ga, p,
|
|
|
6385 su->su_badlen, score, 0, TRUE);
|
|
|
6386 }
|
|
|
6387 list_unref(list);
|
|
|
6388 }
|
|
|
6389
|
|
|
6390 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6391 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6392 }
|
|
|
6393 #endif
|
|
|
6394
|
|
|
6395 /*
|
|
|
6396 * Find suggestions a file "fname".
|
|
|
6397 */
|
|
|
6398 static void
|
|
|
6399 spell_suggest_file(su, fname)
|
|
|
6400 suginfo_T *su;
|
|
|
6401 char_u *fname;
|
|
|
6402 {
|
|
|
6403 FILE *fd;
|
|
|
6404 char_u line[MAXWLEN * 2];
|
|
|
6405 char_u *p;
|
|
|
6406 int len;
|
|
|
6407 char_u cword[MAXWLEN];
|
|
|
6408
|
|
|
6409 /* Open the file. */
|
|
|
6410 fd = mch_fopen((char *)fname, "r");
|
|
|
6411 if (fd == NULL)
|
|
|
6412 {
|
|
|
6413 EMSG2(_(e_notopen), fname);
|
|
|
6414 return;
|
|
|
6415 }
|
|
|
6416
|
|
|
6417 /* Read it line by line. */
|
|
|
6418 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
6419 {
|
|
|
6420 line_breakcheck();
|
|
|
6421
|
|
|
6422 p = vim_strchr(line, '/');
|
|
|
6423 if (p == NULL)
|
|
|
6424 continue; /* No Tab found, just skip the line. */
|
|
|
6425 *p++ = NUL;
|
|
|
6426 if (STRICMP(su->su_badword, line) == 0)
|
|
|
6427 {
|
|
|
6428 /* Match! Isolate the good word, until CR or NL. */
|
|
|
6429 for (len = 0; p[len] >= ' '; ++len)
|
|
|
6430 ;
|
|
|
6431 p[len] = NUL;
|
|
|
6432
|
|
|
6433 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
6434 * of the bad word. */
|
|
|
6435 if (captype(p, NULL) == 0)
|
|
|
6436 {
|
|
|
6437 make_case_word(p, cword, su->su_badflags);
|
|
|
6438 p = cword;
|
|
|
6439 }
|
|
|
6440
|
|
|
6441 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
6442 SCORE_FILE, 0, TRUE);
|
|
|
6443 }
|
|
|
6444 }
|
|
|
6445
|
|
|
6446 fclose(fd);
|
|
|
6447
|
|
|
6448 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6449 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6450 }
|
|
|
6451
|
|
|
6452 /*
|
|
|
6453 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
6454 */
|
|
|
6455 static void
|
|
|
6456 spell_suggest_intern(su)
|
|
|
6457 suginfo_T *su;
|
|
|
6458 {
|
|
344
|
6459 /*
|
|
346
|
6460 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
6461 *
|
|
|
6462 * Set a maximum score to limit the combination of operations that is
|
|
|
6463 * tried.
|
|
|
6464 */
|
|
346
|
6465 suggest_try_special(su);
|
|
|
6466
|
|
|
6467 /*
|
|
|
6468 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
6469 * from the .aff file and inserting a space (split the word).
|
|
|
6470 */
|
|
|
6471 suggest_try_change(su);
|
|
344
|
6472
|
|
|
6473 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
6474 if (sps_flags & SPS_DOUBLE)
|
|
|
6475 score_comp_sal(su);
|
|
|
6476
|
|
|
6477 /*
|
|
346
|
6478 * 3. Try finding sound-a-like words.
|
|
344
|
6479 *
|
|
|
6480 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
6481 * very slow and often doesn't find new suggestions.
|
|
|
6482 */
|
|
|
6483 if ((sps_flags & SPS_DOUBLE)
|
|
|
6484 || (!(sps_flags & SPS_FAST)
|
|
|
6485 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
6486 {
|
|
|
6487 /* Allow a higher score now. */
|
|
|
6488 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
6489 suggest_try_soundalike(su);
|
|
344
|
6490 }
|
|
|
6491
|
|
|
6492 /* When CTRL-C was hit while searching do show the results. */
|
|
|
6493 ui_breakcheck();
|
|
|
6494 if (got_int)
|
|
|
6495 {
|
|
|
6496 (void)vgetc();
|
|
|
6497 got_int = FALSE;
|
|
|
6498 }
|
|
|
6499
|
|
372
|
6500 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
6501 {
|
|
|
6502 if (sps_flags & SPS_BEST)
|
|
|
6503 /* Adjust the word score for how it sounds like. */
|
|
|
6504 rescore_suggestions(su);
|
|
|
6505
|
|
|
6506 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
6507 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
6508 }
|
|
|
6509 }
|
|
|
6510
|
|
|
6511 /*
|
|
|
6512 * Free the info put in "*su" by spell_find_suggest().
|
|
|
6513 */
|
|
|
6514 static void
|
|
|
6515 spell_find_cleanup(su)
|
|
|
6516 suginfo_T *su;
|
|
|
6517 {
|
|
|
6518 int i;
|
|
|
6519
|
|
|
6520 /* Free the suggestions. */
|
|
|
6521 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
6522 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
6523 ga_clear(&su->su_ga);
|
|
|
6524 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
6525 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
6526 ga_clear(&su->su_sga);
|
|
323
|
6527
|
|
|
6528 /* Free the banned words. */
|
|
344
|
6529 free_banned(su);
|
|
323
|
6530 }
|
|
|
6531
|
|
|
6532 /*
|
|
324
|
6533 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
6534 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
6535 * The result is NUL terminated.
|
|
323
|
6536 */
|
|
|
6537 static void
|
|
324
|
6538 onecap_copy(word, wcopy, upper)
|
|
323
|
6539 char_u *word;
|
|
|
6540 char_u *wcopy;
|
|
|
6541 int upper; /* TRUE: first letter made upper case */
|
|
|
6542 {
|
|
|
6543 char_u *p;
|
|
|
6544 int c;
|
|
|
6545 int l;
|
|
|
6546
|
|
|
6547 p = word;
|
|
|
6548 #ifdef FEAT_MBYTE
|
|
|
6549 if (has_mbyte)
|
|
|
6550 c = mb_ptr2char_adv(&p);
|
|
|
6551 else
|
|
|
6552 #endif
|
|
|
6553 c = *p++;
|
|
|
6554 if (upper)
|
|
324
|
6555 c = SPELL_TOUPPER(c);
|
|
323
|
6556 else
|
|
324
|
6557 c = SPELL_TOFOLD(c);
|
|
323
|
6558 #ifdef FEAT_MBYTE
|
|
|
6559 if (has_mbyte)
|
|
|
6560 l = mb_char2bytes(c, wcopy);
|
|
|
6561 else
|
|
|
6562 #endif
|
|
|
6563 {
|
|
|
6564 l = 1;
|
|
|
6565 wcopy[0] = c;
|
|
|
6566 }
|
|
376
|
6567 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
6568 }
|
|
|
6569
|
|
|
6570 /*
|
|
324
|
6571 * Make a copy of "word" with all the letters upper cased into
|
|
|
6572 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
6573 */
|
|
|
6574 static void
|
|
|
6575 allcap_copy(word, wcopy)
|
|
|
6576 char_u *word;
|
|
|
6577 char_u *wcopy;
|
|
|
6578 {
|
|
|
6579 char_u *s;
|
|
|
6580 char_u *d;
|
|
|
6581 int c;
|
|
|
6582
|
|
|
6583 d = wcopy;
|
|
|
6584 for (s = word; *s != NUL; )
|
|
|
6585 {
|
|
|
6586 #ifdef FEAT_MBYTE
|
|
|
6587 if (has_mbyte)
|
|
|
6588 c = mb_ptr2char_adv(&s);
|
|
|
6589 else
|
|
|
6590 #endif
|
|
|
6591 c = *s++;
|
|
324
|
6592 c = SPELL_TOUPPER(c);
|
|
323
|
6593
|
|
|
6594 #ifdef FEAT_MBYTE
|
|
|
6595 if (has_mbyte)
|
|
|
6596 {
|
|
|
6597 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
6598 break;
|
|
|
6599 d += mb_char2bytes(c, d);
|
|
|
6600 }
|
|
|
6601 else
|
|
|
6602 #endif
|
|
|
6603 {
|
|
|
6604 if (d - wcopy >= MAXWLEN - 1)
|
|
|
6605 break;
|
|
|
6606 *d++ = c;
|
|
|
6607 }
|
|
|
6608 }
|
|
|
6609 *d = NUL;
|
|
|
6610 }
|
|
|
6611
|
|
|
6612 /*
|
|
346
|
6613 * Try finding suggestions by recognizing specific situations.
|
|
|
6614 */
|
|
|
6615 static void
|
|
|
6616 suggest_try_special(su)
|
|
|
6617 suginfo_T *su;
|
|
|
6618 {
|
|
|
6619 char_u *p;
|
|
376
|
6620 size_t len;
|
|
346
|
6621 int c;
|
|
|
6622 char_u word[MAXWLEN];
|
|
|
6623
|
|
|
6624 /*
|
|
|
6625 * Recognize a word that is repeated: "the the".
|
|
|
6626 */
|
|
|
6627 p = skiptowhite(su->su_fbadword);
|
|
|
6628 len = p - su->su_fbadword;
|
|
|
6629 p = skipwhite(p);
|
|
|
6630 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
6631 {
|
|
|
6632 /* Include badflags: if the badword is onecap or allcap
|
|
|
6633 * use that for the goodword too: "The the" -> "The". */
|
|
|
6634 c = su->su_fbadword[len];
|
|
|
6635 su->su_fbadword[len] = NUL;
|
|
|
6636 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
6637 su->su_fbadword[len] = c;
|
|
351
|
6638 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
6639 }
|
|
|
6640 }
|
|
|
6641
|
|
|
6642 /*
|
|
323
|
6643 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
6644 *
|
|
|
6645 * This uses a state machine. At each node in the tree we try various
|
|
|
6646 * operations. When trying if an operation work "depth" is increased and the
|
|
|
6647 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
6648 * character, replace one and delete another. The number of changes is
|
|
|
6649 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
6650 */
|
|
|
6651 static void
|
|
346
|
6652 suggest_try_change(su)
|
|
323
|
6653 suginfo_T *su;
|
|
|
6654 {
|
|
|
6655 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
6656 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
6657 trystate_T stack[MAXWLEN];
|
|
|
6658 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
|
6659 * to for word split) */
|
|
|
6660 char_u prewordlen = 0; /* length of word in "preword" */
|
|
|
6661 int splitoff = 0; /* index in tword after last split */
|
|
|
6662 trystate_T *sp;
|
|
|
6663 int newscore;
|
|
|
6664 langp_T *lp;
|
|
375
|
6665 char_u *byts, *fbyts, *pbyts;
|
|
|
6666 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
6667 int depth;
|
|
330
|
6668 int c, c2, c3;
|
|
|
6669 int n = 0;
|
|
323
|
6670 int flags;
|
|
|
6671 garray_T *gap;
|
|
324
|
6672 idx_T arridx;
|
|
323
|
6673 int len;
|
|
|
6674 char_u *p;
|
|
|
6675 fromto_T *ftp;
|
|
330
|
6676 int fl = 0, tl;
|
|
346
|
6677 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
323
|
6678
|
|
|
6679 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
6680 * to find matches (esp. REP items). Append some more text, changing
|
|
|
6681 * chars after the bad word may help. */
|
|
323
|
6682 STRCPY(fword, su->su_fbadword);
|
|
346
|
6683 n = STRLEN(fword);
|
|
|
6684 p = su->su_badptr + su->su_badlen;
|
|
|
6685 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
6686
|
|
|
6687 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
6688 lp->lp_slang != NULL; ++lp)
|
|
|
6689 {
|
|
|
6690 /*
|
|
|
6691 * Go through the whole case-fold tree, try changes at each node.
|
|
|
6692 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
6693 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
6694 * word).
|
|
|
6695 */
|
|
|
6696 depth = 0;
|
|
375
|
6697 sp = &stack[0];
|
|
|
6698 sp->ts_state = STATE_START;
|
|
|
6699 sp->ts_score = 0;
|
|
|
6700 sp->ts_curi = 1;
|
|
|
6701 sp->ts_fidx = 0;
|
|
|
6702 sp->ts_fidxtry = 0;
|
|
|
6703 sp->ts_twordlen = 0;
|
|
|
6704 sp->ts_arridx = 0;
|
|
330
|
6705 #ifdef FEAT_MBYTE
|
|
375
|
6706 sp->ts_tcharlen = 0;
|
|
330
|
6707 #endif
|
|
|
6708
|
|
|
6709 /*
|
|
375
|
6710 * When there are postponed prefixes we need to use these first. At
|
|
|
6711 * the end of the prefix we continue in the case-fold tree.
|
|
|
6712 */
|
|
|
6713 fbyts = lp->lp_slang->sl_fbyts;
|
|
|
6714 fidxs = lp->lp_slang->sl_fidxs;
|
|
|
6715 pbyts = lp->lp_slang->sl_pbyts;
|
|
|
6716 pidxs = lp->lp_slang->sl_pidxs;
|
|
|
6717 if (pbyts != NULL)
|
|
|
6718 {
|
|
|
6719 byts = pbyts;
|
|
|
6720 idxs = pidxs;
|
|
|
6721 sp->ts_prefixdepth = PREFIXTREE;
|
|
|
6722 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
6723 }
|
|
|
6724 else
|
|
|
6725 {
|
|
|
6726 byts = fbyts;
|
|
|
6727 idxs = fidxs;
|
|
|
6728 sp->ts_prefixdepth = NOPREFIX;
|
|
|
6729 }
|
|
|
6730
|
|
|
6731 /*
|
|
330
|
6732 * Loop to find all suggestions. At each round we either:
|
|
|
6733 * - For the current state try one operation, advance "ts_curi",
|
|
|
6734 * increase "depth".
|
|
|
6735 * - When a state is done go to the next, set "ts_state".
|
|
|
6736 * - When all states are tried decrease "depth".
|
|
|
6737 */
|
|
323
|
6738 while (depth >= 0 && !got_int)
|
|
|
6739 {
|
|
|
6740 sp = &stack[depth];
|
|
|
6741 switch (sp->ts_state)
|
|
|
6742 {
|
|
|
6743 case STATE_START:
|
|
375
|
6744 case STATE_NOPREFIX:
|
|
323
|
6745 /*
|
|
|
6746 * Start of node: Deal with NUL bytes, which means
|
|
|
6747 * tword[] may end here.
|
|
|
6748 */
|
|
|
6749 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
6750 len = byts[arridx]; /* bytes in this node */
|
|
|
6751 arridx += sp->ts_curi; /* index of current byte */
|
|
|
6752
|
|
375
|
6753 if (sp->ts_prefixdepth == PREFIXTREE)
|
|
|
6754 {
|
|
|
6755 /* Skip over the NUL bytes, we use them later. */
|
|
|
6756 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
6757 ;
|
|
|
6758 sp->ts_curi += n;
|
|
|
6759
|
|
|
6760 /* At end of a prefix or at start of prefixtree: check for
|
|
|
6761 * following word. */
|
|
|
6762 if (byts[arridx] == 0 || sp->ts_state == STATE_NOPREFIX)
|
|
|
6763 {
|
|
|
6764 sp->ts_state = STATE_START;
|
|
|
6765 ++depth;
|
|
|
6766 stack[depth] = stack[depth - 1];
|
|
|
6767 sp = &stack[depth];
|
|
|
6768 sp->ts_prefixdepth = depth - 1;
|
|
|
6769 byts = fbyts;
|
|
|
6770 idxs = fidxs;
|
|
|
6771 sp->ts_state = STATE_START;
|
|
|
6772 sp->ts_curi = 1; /* start just after length byte */
|
|
|
6773 sp->ts_arridx = 0;
|
|
|
6774
|
|
|
6775 /* Move the prefix to preword[] so that
|
|
|
6776 * find_keepcap_word() works. */
|
|
|
6777 prewordlen = splitoff = sp->ts_twordlen;
|
|
|
6778 mch_memmove(preword, tword, splitoff);
|
|
|
6779 break;
|
|
|
6780 }
|
|
|
6781
|
|
|
6782 /* Always past NUL bytes now. */
|
|
|
6783 sp->ts_state = STATE_ENDNUL;
|
|
|
6784 break;
|
|
|
6785 }
|
|
|
6786
|
|
346
|
6787 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
6788 {
|
|
|
6789 /* Past bytes in node and/or past NUL bytes. */
|
|
|
6790 sp->ts_state = STATE_ENDNUL;
|
|
|
6791 break;
|
|
|
6792 }
|
|
|
6793
|
|
|
6794 /*
|
|
|
6795 * End of word in tree.
|
|
|
6796 */
|
|
|
6797 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
6798
|
|
324
|
6799 flags = (int)idxs[arridx];
|
|
323
|
6800
|
|
375
|
6801 if (sp->ts_prefixdepth < MAXWLEN)
|
|
|
6802 {
|
|
|
6803 /* There was a prefix before the word. Check that the
|
|
|
6804 * prefix can be used with this word. */
|
|
|
6805 /* Count the length of the NULs in the prefix. If there
|
|
|
6806 * are none this must be the first try without a prefix.
|
|
|
6807 */
|
|
|
6808 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
6809 len = pbyts[n++];
|
|
|
6810 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
6811 ;
|
|
|
6812 if (c > 0)
|
|
|
6813 {
|
|
|
6814 /* The prefix ID is stored two bytes above the flags. */
|
|
|
6815 c = valid_word_prefix(c, n, (unsigned)flags >> 16,
|
|
|
6816 tword + splitoff, lp->lp_slang);
|
|
|
6817 if (c == 0)
|
|
|
6818 break;
|
|
|
6819
|
|
|
6820 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
6821 if (c & WF_RAREPFX)
|
|
|
6822 flags |= WF_RARE;
|
|
|
6823 }
|
|
|
6824 }
|
|
|
6825
|
|
323
|
6826 /*
|
|
|
6827 * Form the word with proper case in preword.
|
|
|
6828 * If there is a word from a previous split, append.
|
|
|
6829 */
|
|
|
6830 tword[sp->ts_twordlen] = NUL;
|
|
|
6831 if (flags & WF_KEEPCAP)
|
|
|
6832 /* Must find the word in the keep-case tree. */
|
|
|
6833 find_keepcap_word(lp->lp_slang, tword + splitoff,
|
|
|
6834 preword + prewordlen);
|
|
|
6835 else
|
|
346
|
6836 {
|
|
323
|
6837 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
6838 * use that for the goodword too. But if the badword is
|
|
|
6839 * allcap and it's only one char long use onecap. */
|
|
|
6840 c = su->su_badflags;
|
|
|
6841 if ((c & WF_ALLCAP)
|
|
|
6842 #ifdef FEAT_MBYTE
|
|
|
6843 && su->su_badlen == mb_ptr2len_check(su->su_badptr)
|
|
|
6844 #else
|
|
|
6845 && su->su_badlen == 1
|
|
|
6846 #endif
|
|
|
6847 )
|
|
|
6848 c = WF_ONECAP;
|
|
323
|
6849 make_case_word(tword + splitoff,
|
|
346
|
6850 preword + prewordlen, flags | c);
|
|
|
6851 }
|
|
323
|
6852
|
|
|
6853 /* Don't use a banned word. It may appear again as a good
|
|
|
6854 * word, thus remember it. */
|
|
|
6855 if (flags & WF_BANNED)
|
|
|
6856 {
|
|
|
6857 add_banned(su, preword + prewordlen);
|
|
|
6858 break;
|
|
|
6859 }
|
|
372
|
6860 if (was_banned(su, preword + prewordlen)
|
|
|
6861 || was_banned(su, preword))
|
|
323
|
6862 break;
|
|
|
6863
|
|
|
6864 newscore = 0;
|
|
|
6865 if ((flags & WF_REGION)
|
|
|
6866 && (((unsigned)flags >> 8) & lp->lp_region) == 0)
|
|
|
6867 newscore += SCORE_REGION;
|
|
|
6868 if (flags & WF_RARE)
|
|
|
6869 newscore += SCORE_RARE;
|
|
|
6870
|
|
346
|
6871 if (!spell_valid_case(su->su_badflags,
|
|
323
|
6872 captype(preword + prewordlen, NULL)))
|
|
|
6873 newscore += SCORE_ICASE;
|
|
|
6874
|
|
346
|
6875 if ((fword[sp->ts_fidx] == NUL
|
|
376
|
6876 || !spell_iswordp(fword + sp->ts_fidx, curbuf))
|
|
346
|
6877 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
6878 {
|
|
366
|
6879 /* The badword also ends: add suggestions. Give a penalty
|
|
|
6880 * when changing non-word char to word char, e.g., "thes,"
|
|
|
6881 * -> "these". */
|
|
|
6882 p = fword + sp->ts_fidx;
|
|
|
6883 #ifdef FEAT_MBYTE
|
|
|
6884 if (has_mbyte)
|
|
|
6885 mb_ptr_back(fword, p);
|
|
|
6886 else
|
|
|
6887 #endif
|
|
|
6888 --p;
|
|
376
|
6889 if (!spell_iswordp(p, curbuf))
|
|
366
|
6890 {
|
|
|
6891 p = preword + STRLEN(preword);
|
|
|
6892 #ifdef FEAT_MBYTE
|
|
|
6893 if (has_mbyte)
|
|
|
6894 mb_ptr_back(preword, p);
|
|
|
6895 else
|
|
|
6896 #endif
|
|
|
6897 --p;
|
|
376
|
6898 if (spell_iswordp(p, curbuf))
|
|
366
|
6899 newscore += SCORE_NONWORD;
|
|
|
6900 }
|
|
|
6901
|
|
344
|
6902 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
6903 sp->ts_fidx - repextra,
|
|
351
|
6904 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
6905 }
|
|
330
|
6906 else if (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
6907 #ifdef FEAT_MBYTE
|
|
|
6908 /* Don't split halfway a character. */
|
|
|
6909 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
6910 #endif
|
|
|
6911 )
|
|
323
|
6912 {
|
|
|
6913 /* The word in the tree ends but the badword
|
|
|
6914 * continues: try inserting a space and check that a valid
|
|
|
6915 * words starts at fword[sp->ts_fidx]. */
|
|
|
6916 if (try_deeper(su, stack, depth, newscore + SCORE_SPLIT))
|
|
|
6917 {
|
|
|
6918 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
|
6919 sp->ts_save_prewordlen = prewordlen;
|
|
346
|
6920 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
6921 sp->ts_save_splitoff = splitoff;
|
|
376
|
6922 sp->ts_state = STATE_SPLITUNDO;
|
|
|
6923
|
|
|
6924 ++depth;
|
|
|
6925 sp = &stack[depth];
|
|
323
|
6926
|
|
|
6927 /* Append a space to preword. */
|
|
|
6928 STRCAT(preword, " ");
|
|
|
6929 prewordlen = STRLEN(preword);
|
|
|
6930 splitoff = sp->ts_twordlen;
|
|
376
|
6931
|
|
|
6932 /* If the badword has a non-word character at this
|
|
|
6933 * position skip it. That means replacing the
|
|
|
6934 * non-word character with a space. */
|
|
|
6935 if (!spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
6936 {
|
|
|
6937 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
6938 #ifdef FEAT_MBYTE
|
|
|
6939 if (has_mbyte)
|
|
|
6940 sp->ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
|
6941 else
|
|
|
6942 #endif
|
|
|
6943 ++sp->ts_fidx;
|
|
|
6944 }
|
|
324
|
6945 #ifdef FEAT_MBYTE
|
|
|
6946 if (has_mbyte)
|
|
|
6947 {
|
|
|
6948 int i = 0;
|
|
|
6949
|
|
|
6950 /* Case-folding may change the number of bytes:
|
|
376
|
6951 * Count nr of chars in fword[ts_fidx] and
|
|
324
|
6952 * advance that many chars in su->su_badptr. */
|
|
|
6953 for (p = fword; p < fword + sp->ts_fidx;
|
|
|
6954 mb_ptr_adv(p))
|
|
|
6955 ++i;
|
|
|
6956 for (p = su->su_badptr; i > 0; mb_ptr_adv(p))
|
|
|
6957 --i;
|
|
|
6958 }
|
|
|
6959 else
|
|
|
6960 #endif
|
|
|
6961 p = su->su_badptr + sp->ts_fidx;
|
|
346
|
6962 su->su_badflags = captype(p, su->su_badptr
|
|
|
6963 + su->su_badlen);
|
|
323
|
6964
|
|
|
6965 /* Restart at top of the tree. */
|
|
376
|
6966 sp->ts_arridx = 0;
|
|
323
|
6967 }
|
|
|
6968 }
|
|
|
6969 break;
|
|
|
6970
|
|
|
6971 case STATE_SPLITUNDO:
|
|
346
|
6972 /* Undo the changes done for word split. */
|
|
|
6973 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
6974 splitoff = sp->ts_save_splitoff;
|
|
|
6975 prewordlen = sp->ts_save_prewordlen;
|
|
|
6976
|
|
|
6977 /* Continue looking for NUL bytes. */
|
|
|
6978 sp->ts_state = STATE_START;
|
|
|
6979 break;
|
|
|
6980
|
|
|
6981 case STATE_ENDNUL:
|
|
|
6982 /* Past the NUL bytes in the node. */
|
|
346
|
6983 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
6984 {
|
|
|
6985 /* The badword ends, can't use the bytes in this node. */
|
|
|
6986 sp->ts_state = STATE_DEL;
|
|
|
6987 break;
|
|
|
6988 }
|
|
|
6989 sp->ts_state = STATE_PLAIN;
|
|
|
6990 /*FALLTHROUGH*/
|
|
|
6991
|
|
|
6992 case STATE_PLAIN:
|
|
|
6993 /*
|
|
|
6994 * Go over all possible bytes at this node, add each to
|
|
|
6995 * tword[] and use child node. "ts_curi" is the index.
|
|
|
6996 */
|
|
|
6997 arridx = sp->ts_arridx;
|
|
|
6998 if (sp->ts_curi > byts[arridx])
|
|
|
6999 {
|
|
|
7000 /* Done all bytes at this node, do next state. When still
|
|
|
7001 * at already changed bytes skip the other tricks. */
|
|
|
7002 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7003 sp->ts_state = STATE_DEL;
|
|
|
7004 else
|
|
|
7005 sp->ts_state = STATE_FINAL;
|
|
|
7006 }
|
|
|
7007 else
|
|
|
7008 {
|
|
|
7009 arridx += sp->ts_curi++;
|
|
|
7010 c = byts[arridx];
|
|
|
7011
|
|
|
7012 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
7013 * the byte in the bad word adjust the score. But don't
|
|
|
7014 * even try when the byte was already changed. */
|
|
330
|
7015 if (c == fword[sp->ts_fidx]
|
|
|
7016 #ifdef FEAT_MBYTE
|
|
|
7017 || (sp->ts_tcharlen > 0
|
|
|
7018 && sp->ts_isdiff != DIFF_NONE)
|
|
|
7019 #endif
|
|
|
7020 )
|
|
323
|
7021 newscore = 0;
|
|
|
7022 else
|
|
|
7023 newscore = SCORE_SUBST;
|
|
|
7024 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7025 && try_deeper(su, stack, depth, newscore))
|
|
|
7026 {
|
|
|
7027 ++depth;
|
|
330
|
7028 sp = &stack[depth];
|
|
|
7029 ++sp->ts_fidx;
|
|
|
7030 tword[sp->ts_twordlen++] = c;
|
|
|
7031 sp->ts_arridx = idxs[arridx];
|
|
|
7032 #ifdef FEAT_MBYTE
|
|
|
7033 if (newscore == SCORE_SUBST)
|
|
|
7034 sp->ts_isdiff = DIFF_YES;
|
|
|
7035 if (has_mbyte)
|
|
|
7036 {
|
|
|
7037 /* Multi-byte characters are a bit complicated to
|
|
|
7038 * handle: They differ when any of the bytes
|
|
|
7039 * differ and then their length may also differ. */
|
|
|
7040 if (sp->ts_tcharlen == 0)
|
|
|
7041 {
|
|
|
7042 /* First byte. */
|
|
|
7043 sp->ts_tcharidx = 0;
|
|
|
7044 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
7045 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
7046 sp->ts_isdiff = (newscore != 0)
|
|
|
7047 ? DIFF_YES : DIFF_NONE;
|
|
|
7048 }
|
|
|
7049 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
7050 /* When inserting trail bytes don't advance in
|
|
|
7051 * the bad word. */
|
|
|
7052 --sp->ts_fidx;
|
|
|
7053 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
7054 {
|
|
|
7055 /* Last byte of character. */
|
|
|
7056 if (sp->ts_isdiff == DIFF_YES)
|
|
|
7057 {
|
|
|
7058 /* Correct ts_fidx for the byte length of
|
|
|
7059 * the character (we didn't check that
|
|
|
7060 * before). */
|
|
|
7061 sp->ts_fidx = sp->ts_fcharstart
|
|
|
7062 + MB_BYTE2LEN(
|
|
|
7063 fword[sp->ts_fcharstart]);
|
|
|
7064
|
|
|
7065 /* For a similar character adjust score
|
|
|
7066 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
|
7067 if (lp->lp_slang->sl_has_map
|
|
|
7068 && similar_chars(lp->lp_slang,
|
|
|
7069 mb_ptr2char(tword
|
|
|
7070 + sp->ts_twordlen
|
|
|
7071 - sp->ts_tcharlen),
|
|
|
7072 mb_ptr2char(fword
|
|
|
7073 + sp->ts_fcharstart)))
|
|
|
7074 sp->ts_score -=
|
|
|
7075 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7076 }
|
|
358
|
7077 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
7078 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
7079 {
|
|
|
7080 /* If the previous character was the same,
|
|
|
7081 * thus doubling a character, give a bonus
|
|
|
7082 * to the score. */
|
|
|
7083 p = tword + sp->ts_twordlen
|
|
|
7084 - sp->ts_tcharlen;
|
|
|
7085 c = mb_ptr2char(p);
|
|
|
7086 mb_ptr_back(tword, p);
|
|
|
7087 if (c == mb_ptr2char(p))
|
|
|
7088 sp->ts_score -= SCORE_INS
|
|
|
7089 - SCORE_INSDUP;
|
|
|
7090 }
|
|
330
|
7091
|
|
|
7092 /* Starting a new char, reset the length. */
|
|
|
7093 sp->ts_tcharlen = 0;
|
|
|
7094 }
|
|
|
7095 }
|
|
|
7096 else
|
|
|
7097 #endif
|
|
|
7098 {
|
|
|
7099 /* If we found a similar char adjust the score.
|
|
|
7100 * We do this after calling try_deeper() because
|
|
|
7101 * it's slow. */
|
|
|
7102 if (newscore != 0
|
|
|
7103 && lp->lp_slang->sl_has_map
|
|
|
7104 && similar_chars(lp->lp_slang,
|
|
|
7105 c, fword[sp->ts_fidx - 1]))
|
|
|
7106 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7107 }
|
|
323
|
7108 }
|
|
|
7109 }
|
|
|
7110 break;
|
|
|
7111
|
|
|
7112 case STATE_DEL:
|
|
330
|
7113 #ifdef FEAT_MBYTE
|
|
|
7114 /* When past the first byte of a multi-byte char don't try
|
|
|
7115 * delete/insert/swap a character. */
|
|
|
7116 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
7117 {
|
|
|
7118 sp->ts_state = STATE_FINAL;
|
|
|
7119 break;
|
|
|
7120 }
|
|
|
7121 #endif
|
|
|
7122 /*
|
|
|
7123 * Try skipping one character in the bad word (delete it).
|
|
|
7124 */
|
|
323
|
7125 sp->ts_state = STATE_INS;
|
|
|
7126 sp->ts_curi = 1;
|
|
|
7127 if (fword[sp->ts_fidx] != NUL
|
|
|
7128 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
7129 {
|
|
|
7130 ++depth;
|
|
358
|
7131
|
|
|
7132 /* Advance over the character in fword[]. Give a bonus to
|
|
|
7133 * the score if the same character is following "nn" ->
|
|
|
7134 * "n". */
|
|
330
|
7135 #ifdef FEAT_MBYTE
|
|
|
7136 if (has_mbyte)
|
|
358
|
7137 {
|
|
|
7138 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
7139 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
358
|
7140 if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
|
7141 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7142 }
|
|
330
|
7143 else
|
|
|
7144 #endif
|
|
358
|
7145 {
|
|
330
|
7146 ++stack[depth].ts_fidx;
|
|
358
|
7147 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
7148 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7149 }
|
|
323
|
7150 break;
|
|
|
7151 }
|
|
|
7152 /*FALLTHROUGH*/
|
|
|
7153
|
|
|
7154 case STATE_INS:
|
|
330
|
7155 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
7156 * node. */
|
|
|
7157 n = sp->ts_arridx;
|
|
|
7158 if (sp->ts_curi > byts[n])
|
|
|
7159 {
|
|
|
7160 /* Done all bytes at this node, do next state. */
|
|
|
7161 sp->ts_state = STATE_SWAP;
|
|
|
7162 }
|
|
|
7163 else
|
|
|
7164 {
|
|
330
|
7165 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
7166 n += sp->ts_curi++;
|
|
|
7167 c = byts[n];
|
|
|
7168 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
7169 {
|
|
|
7170 ++depth;
|
|
330
|
7171 sp = &stack[depth];
|
|
|
7172 tword[sp->ts_twordlen++] = c;
|
|
|
7173 sp->ts_arridx = idxs[n];
|
|
|
7174 #ifdef FEAT_MBYTE
|
|
|
7175 if (has_mbyte)
|
|
|
7176 {
|
|
|
7177 fl = MB_BYTE2LEN(c);
|
|
|
7178 if (fl > 1)
|
|
|
7179 {
|
|
|
7180 /* There are following bytes for the same
|
|
|
7181 * character. We must find all bytes before
|
|
|
7182 * trying delete/insert/swap/etc. */
|
|
|
7183 sp->ts_tcharlen = fl;
|
|
|
7184 sp->ts_tcharidx = 1;
|
|
|
7185 sp->ts_isdiff = DIFF_INSERT;
|
|
|
7186 }
|
|
|
7187 }
|
|
358
|
7188 else
|
|
|
7189 fl = 1;
|
|
|
7190 if (fl == 1)
|
|
330
|
7191 #endif
|
|
358
|
7192 {
|
|
|
7193 /* If the previous character was the same, thus
|
|
|
7194 * doubling a character, give a bonus to the
|
|
|
7195 * score. */
|
|
|
7196 if (sp->ts_twordlen >= 2
|
|
|
7197 && tword[sp->ts_twordlen - 2] == c)
|
|
|
7198 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
7199 }
|
|
323
|
7200 }
|
|
|
7201 }
|
|
|
7202 break;
|
|
|
7203
|
|
|
7204 case STATE_SWAP:
|
|
330
|
7205 /*
|
|
|
7206 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
7207 * We change "fword" here, it's changed back afterwards.
|
|
|
7208 */
|
|
|
7209 p = fword + sp->ts_fidx;
|
|
|
7210 c = *p;
|
|
|
7211 if (c == NUL)
|
|
|
7212 {
|
|
|
7213 /* End of word, can't swap or replace. */
|
|
|
7214 sp->ts_state = STATE_FINAL;
|
|
|
7215 break;
|
|
|
7216 }
|
|
|
7217 #ifdef FEAT_MBYTE
|
|
|
7218 if (has_mbyte)
|
|
|
7219 {
|
|
|
7220 n = mb_ptr2len_check(p);
|
|
|
7221 c = mb_ptr2char(p);
|
|
|
7222 c2 = mb_ptr2char(p + n);
|
|
|
7223 }
|
|
|
7224 else
|
|
|
7225 #endif
|
|
|
7226 c2 = p[1];
|
|
|
7227 if (c == c2)
|
|
323
|
7228 {
|
|
330
|
7229 /* Characters are identical, swap won't do anything. */
|
|
|
7230 sp->ts_state = STATE_SWAP3;
|
|
|
7231 break;
|
|
|
7232 }
|
|
|
7233 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
7234 {
|
|
|
7235 sp->ts_state = STATE_UNSWAP;
|
|
323
|
7236 ++depth;
|
|
330
|
7237 #ifdef FEAT_MBYTE
|
|
|
7238 if (has_mbyte)
|
|
|
7239 {
|
|
|
7240 fl = mb_char2len(c2);
|
|
|
7241 mch_memmove(p, p + n, fl);
|
|
|
7242 mb_char2bytes(c, p + fl);
|
|
|
7243 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7244 }
|
|
|
7245 else
|
|
|
7246 #endif
|
|
|
7247 {
|
|
|
7248 p[0] = c2;
|
|
|
7249 p[1] = c;
|
|
|
7250 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
7251 }
|
|
323
|
7252 }
|
|
|
7253 else
|
|
|
7254 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
7255 sp->ts_state = STATE_REP_INI;
|
|
|
7256 break;
|
|
|
7257
|
|
330
|
7258 case STATE_UNSWAP:
|
|
|
7259 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
7260 p = fword + sp->ts_fidx;
|
|
|
7261 #ifdef FEAT_MBYTE
|
|
|
7262 if (has_mbyte)
|
|
|
7263 {
|
|
|
7264 n = MB_BYTE2LEN(*p);
|
|
|
7265 c = mb_ptr2char(p + n);
|
|
|
7266 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
7267 mb_char2bytes(c, p);
|
|
|
7268 }
|
|
|
7269 else
|
|
|
7270 #endif
|
|
|
7271 {
|
|
|
7272 c = *p;
|
|
|
7273 *p = p[1];
|
|
|
7274 p[1] = c;
|
|
|
7275 }
|
|
|
7276 /*FALLTHROUGH*/
|
|
|
7277
|
|
|
7278 case STATE_SWAP3:
|
|
323
|
7279 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
7280 * "fword" here, it's changed back afterwards. */
|
|
|
7281 p = fword + sp->ts_fidx;
|
|
|
7282 #ifdef FEAT_MBYTE
|
|
|
7283 if (has_mbyte)
|
|
|
7284 {
|
|
|
7285 n = mb_ptr2len_check(p);
|
|
|
7286 c = mb_ptr2char(p);
|
|
|
7287 fl = mb_ptr2len_check(p + n);
|
|
|
7288 c2 = mb_ptr2char(p + n);
|
|
|
7289 c3 = mb_ptr2char(p + n + fl);
|
|
|
7290 }
|
|
|
7291 else
|
|
|
7292 #endif
|
|
323
|
7293 {
|
|
330
|
7294 c = *p;
|
|
|
7295 c2 = p[1];
|
|
|
7296 c3 = p[2];
|
|
|
7297 }
|
|
|
7298
|
|
|
7299 /* When characters are identical: "121" then SWAP3 result is
|
|
|
7300 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
7301 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
7302 * swapping. Also skip when c3 is NUL. */
|
|
|
7303 if (c == c3 || c3 == NUL)
|
|
|
7304 {
|
|
|
7305 sp->ts_state = STATE_REP_INI;
|
|
|
7306 break;
|
|
|
7307 }
|
|
|
7308 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7309 {
|
|
|
7310 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
7311 ++depth;
|
|
330
|
7312 #ifdef FEAT_MBYTE
|
|
|
7313 if (has_mbyte)
|
|
|
7314 {
|
|
|
7315 tl = mb_char2len(c3);
|
|
|
7316 mch_memmove(p, p + n + fl, tl);
|
|
|
7317 mb_char2bytes(c2, p + tl);
|
|
|
7318 mb_char2bytes(c, p + fl + tl);
|
|
|
7319 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
7320 }
|
|
|
7321 else
|
|
|
7322 #endif
|
|
|
7323 {
|
|
|
7324 p[0] = p[2];
|
|
|
7325 p[2] = c;
|
|
|
7326 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7327 }
|
|
323
|
7328 }
|
|
|
7329 else
|
|
|
7330 sp->ts_state = STATE_REP_INI;
|
|
|
7331 break;
|
|
|
7332
|
|
330
|
7333 case STATE_UNSWAP3:
|
|
|
7334 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
7335 p = fword + sp->ts_fidx;
|
|
|
7336 #ifdef FEAT_MBYTE
|
|
|
7337 if (has_mbyte)
|
|
|
7338 {
|
|
|
7339 n = MB_BYTE2LEN(*p);
|
|
|
7340 c2 = mb_ptr2char(p + n);
|
|
|
7341 fl = MB_BYTE2LEN(p[n]);
|
|
|
7342 c = mb_ptr2char(p + n + fl);
|
|
|
7343 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
7344 mch_memmove(p + fl + tl, p, n);
|
|
|
7345 mb_char2bytes(c, p);
|
|
|
7346 mb_char2bytes(c2, p + tl);
|
|
|
7347 }
|
|
|
7348 else
|
|
|
7349 #endif
|
|
|
7350 {
|
|
|
7351 c = *p;
|
|
|
7352 *p = p[2];
|
|
|
7353 p[2] = c;
|
|
|
7354 }
|
|
346
|
7355
|
|
330
|
7356 /* Rotate three characters left: "123" -> "231". We change
|
|
|
7357 * "fword" here, it's changed back afterwards. */
|
|
323
|
7358 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7359 {
|
|
330
|
7360 sp->ts_state = STATE_UNROT3L;
|
|
323
|
7361 ++depth;
|
|
330
|
7362 p = fword + sp->ts_fidx;
|
|
|
7363 #ifdef FEAT_MBYTE
|
|
|
7364 if (has_mbyte)
|
|
|
7365 {
|
|
|
7366 n = mb_ptr2len_check(p);
|
|
|
7367 c = mb_ptr2char(p);
|
|
|
7368 fl = mb_ptr2len_check(p + n);
|
|
|
7369 fl += mb_ptr2len_check(p + n + fl);
|
|
|
7370 mch_memmove(p, p + n, fl);
|
|
|
7371 mb_char2bytes(c, p + fl);
|
|
|
7372 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7373 }
|
|
|
7374 else
|
|
|
7375 #endif
|
|
|
7376 {
|
|
|
7377 c = *p;
|
|
|
7378 *p = p[1];
|
|
|
7379 p[1] = p[2];
|
|
|
7380 p[2] = c;
|
|
|
7381 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7382 }
|
|
323
|
7383 }
|
|
|
7384 else
|
|
|
7385 sp->ts_state = STATE_REP_INI;
|
|
|
7386 break;
|
|
|
7387
|
|
330
|
7388 case STATE_UNROT3L:
|
|
346
|
7389 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
7390 p = fword + sp->ts_fidx;
|
|
|
7391 #ifdef FEAT_MBYTE
|
|
|
7392 if (has_mbyte)
|
|
|
7393 {
|
|
|
7394 n = MB_BYTE2LEN(*p);
|
|
|
7395 n += MB_BYTE2LEN(p[n]);
|
|
|
7396 c = mb_ptr2char(p + n);
|
|
|
7397 tl = MB_BYTE2LEN(p[n]);
|
|
|
7398 mch_memmove(p + tl, p, n);
|
|
|
7399 mb_char2bytes(c, p);
|
|
|
7400 }
|
|
|
7401 else
|
|
|
7402 #endif
|
|
|
7403 {
|
|
|
7404 c = p[2];
|
|
|
7405 p[2] = p[1];
|
|
|
7406 p[1] = *p;
|
|
|
7407 *p = c;
|
|
|
7408 }
|
|
346
|
7409
|
|
323
|
7410 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
7411 * "fword" here, it's changed back afterwards. */
|
|
323
|
7412 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7413 {
|
|
330
|
7414 sp->ts_state = STATE_UNROT3R;
|
|
323
|
7415 ++depth;
|
|
330
|
7416 p = fword + sp->ts_fidx;
|
|
|
7417 #ifdef FEAT_MBYTE
|
|
|
7418 if (has_mbyte)
|
|
|
7419 {
|
|
|
7420 n = mb_ptr2len_check(p);
|
|
|
7421 n += mb_ptr2len_check(p + n);
|
|
|
7422 c = mb_ptr2char(p + n);
|
|
|
7423 tl = mb_ptr2len_check(p + n);
|
|
|
7424 mch_memmove(p + tl, p, n);
|
|
|
7425 mb_char2bytes(c, p);
|
|
|
7426 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
7427 }
|
|
|
7428 else
|
|
|
7429 #endif
|
|
|
7430 {
|
|
|
7431 c = p[2];
|
|
|
7432 p[2] = p[1];
|
|
|
7433 p[1] = *p;
|
|
|
7434 *p = c;
|
|
|
7435 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7436 }
|
|
323
|
7437 }
|
|
|
7438 else
|
|
|
7439 sp->ts_state = STATE_REP_INI;
|
|
|
7440 break;
|
|
|
7441
|
|
330
|
7442 case STATE_UNROT3R:
|
|
346
|
7443 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
7444 p = fword + sp->ts_fidx;
|
|
|
7445 #ifdef FEAT_MBYTE
|
|
|
7446 if (has_mbyte)
|
|
|
7447 {
|
|
|
7448 c = mb_ptr2char(p);
|
|
|
7449 tl = MB_BYTE2LEN(*p);
|
|
|
7450 n = MB_BYTE2LEN(p[tl]);
|
|
|
7451 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
7452 mch_memmove(p, p + tl, n);
|
|
|
7453 mb_char2bytes(c, p + n);
|
|
|
7454 }
|
|
|
7455 else
|
|
|
7456 #endif
|
|
|
7457 {
|
|
|
7458 c = *p;
|
|
|
7459 *p = p[1];
|
|
|
7460 p[1] = p[2];
|
|
|
7461 p[2] = c;
|
|
|
7462 }
|
|
323
|
7463 /*FALLTHROUGH*/
|
|
|
7464
|
|
|
7465 case STATE_REP_INI:
|
|
|
7466 /* Check if matching with REP items from the .aff file would
|
|
|
7467 * work. Quickly skip if there are no REP items or the score
|
|
|
7468 * is going to be too high anyway. */
|
|
|
7469 gap = &lp->lp_slang->sl_rep;
|
|
|
7470 if (gap->ga_len == 0
|
|
|
7471 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
7472 {
|
|
|
7473 sp->ts_state = STATE_FINAL;
|
|
|
7474 break;
|
|
|
7475 }
|
|
|
7476
|
|
|
7477 /* Use the first byte to quickly find the first entry that
|
|
330
|
7478 * may match. If the index is -1 there is none. */
|
|
323
|
7479 sp->ts_curi = lp->lp_slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
7480 if (sp->ts_curi < 0)
|
|
|
7481 {
|
|
|
7482 sp->ts_state = STATE_FINAL;
|
|
|
7483 break;
|
|
|
7484 }
|
|
|
7485
|
|
|
7486 sp->ts_state = STATE_REP;
|
|
|
7487 /*FALLTHROUGH*/
|
|
|
7488
|
|
|
7489 case STATE_REP:
|
|
|
7490 /* Try matching with REP items from the .aff file. For each
|
|
330
|
7491 * match replace the characters and check if the resulting
|
|
|
7492 * word is valid. */
|
|
323
|
7493 p = fword + sp->ts_fidx;
|
|
|
7494
|
|
|
7495 gap = &lp->lp_slang->sl_rep;
|
|
|
7496 while (sp->ts_curi < gap->ga_len)
|
|
|
7497 {
|
|
|
7498 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
7499 if (*ftp->ft_from != *p)
|
|
|
7500 {
|
|
|
7501 /* past possible matching entries */
|
|
|
7502 sp->ts_curi = gap->ga_len;
|
|
|
7503 break;
|
|
|
7504 }
|
|
|
7505 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
7506 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
7507 {
|
|
|
7508 /* Need to undo this afterwards. */
|
|
|
7509 sp->ts_state = STATE_REP_UNDO;
|
|
|
7510
|
|
|
7511 /* Change the "from" to the "to" string. */
|
|
|
7512 ++depth;
|
|
|
7513 fl = STRLEN(ftp->ft_from);
|
|
|
7514 tl = STRLEN(ftp->ft_to);
|
|
|
7515 if (fl != tl)
|
|
346
|
7516 {
|
|
323
|
7517 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
7518 repextra += tl - fl;
|
|
|
7519 }
|
|
323
|
7520 mch_memmove(p, ftp->ft_to, tl);
|
|
|
7521 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
7522 #ifdef FEAT_MBYTE
|
|
|
7523 stack[depth].ts_tcharlen = 0;
|
|
|
7524 #endif
|
|
323
|
7525 break;
|
|
|
7526 }
|
|
|
7527 }
|
|
|
7528
|
|
|
7529 if (sp->ts_curi >= gap->ga_len)
|
|
|
7530 /* No (more) matches. */
|
|
|
7531 sp->ts_state = STATE_FINAL;
|
|
|
7532
|
|
|
7533 break;
|
|
|
7534
|
|
|
7535 case STATE_REP_UNDO:
|
|
|
7536 /* Undo a REP replacement and continue with the next one. */
|
|
|
7537 ftp = (fromto_T *)lp->lp_slang->sl_rep.ga_data
|
|
|
7538 + sp->ts_curi - 1;
|
|
|
7539 fl = STRLEN(ftp->ft_from);
|
|
|
7540 tl = STRLEN(ftp->ft_to);
|
|
|
7541 p = fword + sp->ts_fidx;
|
|
|
7542 if (fl != tl)
|
|
346
|
7543 {
|
|
323
|
7544 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
7545 repextra -= tl - fl;
|
|
|
7546 }
|
|
323
|
7547 mch_memmove(p, ftp->ft_from, fl);
|
|
|
7548 sp->ts_state = STATE_REP;
|
|
|
7549 break;
|
|
|
7550
|
|
|
7551 default:
|
|
|
7552 /* Did all possible states at this level, go up one level. */
|
|
|
7553 --depth;
|
|
344
|
7554
|
|
375
|
7555 if (depth >= 0 && stack[depth].ts_prefixdepth == PREFIXTREE)
|
|
|
7556 {
|
|
|
7557 /* Continue in or go back to the prefix tree. */
|
|
|
7558 byts = pbyts;
|
|
|
7559 idxs = pidxs;
|
|
|
7560 splitoff = 0;
|
|
|
7561 }
|
|
|
7562
|
|
344
|
7563 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
7564 line_breakcheck();
|
|
323
|
7565 }
|
|
|
7566 }
|
|
|
7567 }
|
|
|
7568 }
|
|
|
7569
|
|
|
7570 /*
|
|
|
7571 * Try going one level deeper in the tree.
|
|
|
7572 */
|
|
|
7573 static int
|
|
|
7574 try_deeper(su, stack, depth, score_add)
|
|
|
7575 suginfo_T *su;
|
|
|
7576 trystate_T *stack;
|
|
|
7577 int depth;
|
|
|
7578 int score_add;
|
|
|
7579 {
|
|
|
7580 int newscore;
|
|
|
7581
|
|
|
7582 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
7583 newscore = stack[depth].ts_score + score_add;
|
|
|
7584 if (newscore >= su->su_maxscore)
|
|
|
7585 return FALSE;
|
|
|
7586
|
|
330
|
7587 stack[depth + 1] = stack[depth];
|
|
323
|
7588 stack[depth + 1].ts_state = STATE_START;
|
|
|
7589 stack[depth + 1].ts_score = newscore;
|
|
|
7590 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
7591 return TRUE;
|
|
|
7592 }
|
|
|
7593
|
|
|
7594 /*
|
|
|
7595 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
7596 * words and put it in "kword".
|
|
|
7597 * Theoretically there could be several keep-case words that result in the
|
|
|
7598 * same case-folded word, but we only find one...
|
|
|
7599 */
|
|
|
7600 static void
|
|
|
7601 find_keepcap_word(slang, fword, kword)
|
|
|
7602 slang_T *slang;
|
|
|
7603 char_u *fword;
|
|
|
7604 char_u *kword;
|
|
|
7605 {
|
|
|
7606 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
7607 int depth;
|
|
324
|
7608 idx_T tryidx;
|
|
323
|
7609
|
|
|
7610 /* The following arrays are used at each depth in the tree. */
|
|
324
|
7611 idx_T arridx[MAXWLEN];
|
|
323
|
7612 int round[MAXWLEN];
|
|
|
7613 int fwordidx[MAXWLEN];
|
|
|
7614 int uwordidx[MAXWLEN];
|
|
|
7615 int kwordlen[MAXWLEN];
|
|
|
7616
|
|
|
7617 int flen, ulen;
|
|
|
7618 int l;
|
|
|
7619 int len;
|
|
|
7620 int c;
|
|
324
|
7621 idx_T lo, hi, m;
|
|
323
|
7622 char_u *p;
|
|
|
7623 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
7624 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
7625
|
|
|
7626 if (byts == NULL)
|
|
|
7627 {
|
|
|
7628 /* array is empty: "cannot happen" */
|
|
|
7629 *kword = NUL;
|
|
|
7630 return;
|
|
|
7631 }
|
|
|
7632
|
|
|
7633 /* Make an all-cap version of "fword". */
|
|
|
7634 allcap_copy(fword, uword);
|
|
|
7635
|
|
|
7636 /*
|
|
|
7637 * Each character needs to be tried both case-folded and upper-case.
|
|
|
7638 * All this gets very complicated if we keep in mind that changing case
|
|
|
7639 * may change the byte length of a multi-byte character...
|
|
|
7640 */
|
|
|
7641 depth = 0;
|
|
|
7642 arridx[0] = 0;
|
|
|
7643 round[0] = 0;
|
|
|
7644 fwordidx[0] = 0;
|
|
|
7645 uwordidx[0] = 0;
|
|
|
7646 kwordlen[0] = 0;
|
|
|
7647 while (depth >= 0)
|
|
|
7648 {
|
|
|
7649 if (fword[fwordidx[depth]] == NUL)
|
|
|
7650 {
|
|
|
7651 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
7652 * here we have found a match. */
|
|
|
7653 if (byts[arridx[depth] + 1] == 0)
|
|
|
7654 {
|
|
|
7655 kword[kwordlen[depth]] = NUL;
|
|
|
7656 return;
|
|
|
7657 }
|
|
|
7658
|
|
|
7659 /* kword is getting too long, continue one level up */
|
|
|
7660 --depth;
|
|
|
7661 }
|
|
|
7662 else if (++round[depth] > 2)
|
|
|
7663 {
|
|
|
7664 /* tried both fold-case and upper-case character, continue one
|
|
|
7665 * level up */
|
|
|
7666 --depth;
|
|
|
7667 }
|
|
|
7668 else
|
|
|
7669 {
|
|
|
7670 /*
|
|
|
7671 * round[depth] == 1: Try using the folded-case character.
|
|
|
7672 * round[depth] == 2: Try using the upper-case character.
|
|
|
7673 */
|
|
|
7674 #ifdef FEAT_MBYTE
|
|
|
7675 if (has_mbyte)
|
|
|
7676 {
|
|
|
7677 flen = mb_ptr2len_check(fword + fwordidx[depth]);
|
|
|
7678 ulen = mb_ptr2len_check(uword + uwordidx[depth]);
|
|
|
7679 }
|
|
|
7680 else
|
|
|
7681 #endif
|
|
|
7682 ulen = flen = 1;
|
|
|
7683 if (round[depth] == 1)
|
|
|
7684 {
|
|
|
7685 p = fword + fwordidx[depth];
|
|
|
7686 l = flen;
|
|
|
7687 }
|
|
|
7688 else
|
|
|
7689 {
|
|
|
7690 p = uword + uwordidx[depth];
|
|
|
7691 l = ulen;
|
|
|
7692 }
|
|
|
7693
|
|
|
7694 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
7695 {
|
|
|
7696 /* Perform a binary search in the list of accepted bytes. */
|
|
|
7697 len = byts[tryidx++];
|
|
|
7698 c = *p++;
|
|
|
7699 lo = tryidx;
|
|
|
7700 hi = tryidx + len - 1;
|
|
|
7701 while (lo < hi)
|
|
|
7702 {
|
|
|
7703 m = (lo + hi) / 2;
|
|
|
7704 if (byts[m] > c)
|
|
|
7705 hi = m - 1;
|
|
|
7706 else if (byts[m] < c)
|
|
|
7707 lo = m + 1;
|
|
|
7708 else
|
|
|
7709 {
|
|
|
7710 lo = hi = m;
|
|
|
7711 break;
|
|
|
7712 }
|
|
|
7713 }
|
|
|
7714
|
|
|
7715 /* Stop if there is no matching byte. */
|
|
|
7716 if (hi < lo || byts[lo] != c)
|
|
|
7717 break;
|
|
|
7718
|
|
|
7719 /* Continue at the child (if there is one). */
|
|
|
7720 tryidx = idxs[lo];
|
|
|
7721 }
|
|
|
7722
|
|
|
7723 if (l == 0)
|
|
|
7724 {
|
|
|
7725 /*
|
|
|
7726 * Found the matching char. Copy it to "kword" and go a
|
|
|
7727 * level deeper.
|
|
|
7728 */
|
|
|
7729 if (round[depth] == 1)
|
|
|
7730 {
|
|
|
7731 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
7732 flen);
|
|
|
7733 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
7734 }
|
|
|
7735 else
|
|
|
7736 {
|
|
|
7737 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
7738 ulen);
|
|
|
7739 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
7740 }
|
|
|
7741 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
7742 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
7743
|
|
|
7744 ++depth;
|
|
|
7745 arridx[depth] = tryidx;
|
|
|
7746 round[depth] = 0;
|
|
|
7747 }
|
|
|
7748 }
|
|
|
7749 }
|
|
|
7750
|
|
|
7751 /* Didn't find it: "cannot happen". */
|
|
|
7752 *kword = NUL;
|
|
|
7753 }
|
|
|
7754
|
|
|
7755 /*
|
|
344
|
7756 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
7757 * su->su_sga.
|
|
|
7758 */
|
|
|
7759 static void
|
|
|
7760 score_comp_sal(su)
|
|
|
7761 suginfo_T *su;
|
|
|
7762 {
|
|
|
7763 langp_T *lp;
|
|
|
7764 char_u badsound[MAXWLEN];
|
|
|
7765 int i;
|
|
|
7766 suggest_T *stp;
|
|
|
7767 suggest_T *sstp;
|
|
|
7768 int score;
|
|
|
7769
|
|
|
7770 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
7771 return;
|
|
|
7772
|
|
|
7773 /* Use the sound-folding of the first language that supports it. */
|
|
|
7774 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7775 lp->lp_slang != NULL; ++lp)
|
|
|
7776 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7777 {
|
|
|
7778 /* soundfold the bad word */
|
|
375
|
7779 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
7780
|
|
|
7781 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7782 {
|
|
|
7783 stp = &SUG(su->su_ga, i);
|
|
|
7784
|
|
351
|
7785 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
7786 * sound-a-like score. */
|
|
|
7787 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
7788 if (score < SCORE_MAXMAX)
|
|
|
7789 {
|
|
|
7790 /* Add the suggestion. */
|
|
|
7791 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
7792 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
7793 if (sstp->st_word != NULL)
|
|
|
7794 {
|
|
|
7795 sstp->st_score = score;
|
|
|
7796 sstp->st_altscore = 0;
|
|
|
7797 sstp->st_orglen = stp->st_orglen;
|
|
|
7798 ++su->su_sga.ga_len;
|
|
|
7799 }
|
|
|
7800 }
|
|
|
7801 }
|
|
|
7802 break;
|
|
|
7803 }
|
|
|
7804 }
|
|
|
7805
|
|
|
7806 /*
|
|
|
7807 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
7808 * They are intwined.
|
|
|
7809 */
|
|
|
7810 static void
|
|
|
7811 score_combine(su)
|
|
|
7812 suginfo_T *su;
|
|
|
7813 {
|
|
|
7814 int i;
|
|
|
7815 int j;
|
|
|
7816 garray_T ga;
|
|
|
7817 garray_T *gap;
|
|
|
7818 langp_T *lp;
|
|
|
7819 suggest_T *stp;
|
|
|
7820 char_u *p;
|
|
|
7821 char_u badsound[MAXWLEN];
|
|
|
7822 int round;
|
|
|
7823
|
|
|
7824 /* Add the alternate score to su_ga. */
|
|
|
7825 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7826 lp->lp_slang != NULL; ++lp)
|
|
|
7827 {
|
|
|
7828 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7829 {
|
|
|
7830 /* soundfold the bad word */
|
|
375
|
7831 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
7832
|
|
|
7833 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7834 {
|
|
|
7835 stp = &SUG(su->su_ga, i);
|
|
351
|
7836 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
7837 badsound);
|
|
344
|
7838 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
7839 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
7840 else
|
|
|
7841 stp->st_score = (stp->st_score * 3
|
|
|
7842 + stp->st_altscore) / 4;
|
|
|
7843 stp->st_salscore = FALSE;
|
|
|
7844 }
|
|
|
7845 break;
|
|
|
7846 }
|
|
|
7847 }
|
|
|
7848
|
|
|
7849 /* Add the alternate score to su_sga. */
|
|
|
7850 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
7851 {
|
|
|
7852 stp = &SUG(su->su_sga, i);
|
|
|
7853 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
7854 if (stp->st_score == SCORE_MAXMAX)
|
|
|
7855 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
7856 else
|
|
|
7857 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
7858 stp->st_salscore = TRUE;
|
|
|
7859 }
|
|
|
7860
|
|
|
7861 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
7862 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7863 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
7864
|
|
|
7865 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
7866 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
7867 return;
|
|
|
7868
|
|
|
7869 stp = &SUG(ga, 0);
|
|
|
7870 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
7871 {
|
|
|
7872 /* round 1: get a suggestion from su_ga
|
|
|
7873 * round 2: get a suggestion from su_sga */
|
|
|
7874 for (round = 1; round <= 2; ++round)
|
|
|
7875 {
|
|
|
7876 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
7877 if (i < gap->ga_len)
|
|
|
7878 {
|
|
|
7879 /* Don't add a word if it's already there. */
|
|
|
7880 p = SUG(*gap, i).st_word;
|
|
|
7881 for (j = 0; j < ga.ga_len; ++j)
|
|
|
7882 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
7883 break;
|
|
|
7884 if (j == ga.ga_len)
|
|
|
7885 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
7886 else
|
|
|
7887 vim_free(p);
|
|
|
7888 }
|
|
|
7889 }
|
|
|
7890 }
|
|
|
7891
|
|
|
7892 ga_clear(&su->su_ga);
|
|
|
7893 ga_clear(&su->su_sga);
|
|
|
7894
|
|
|
7895 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
7896 if (ga.ga_len > su->su_maxcount)
|
|
|
7897 {
|
|
|
7898 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
7899 vim_free(stp[i].st_word);
|
|
|
7900 ga.ga_len = su->su_maxcount;
|
|
|
7901 }
|
|
|
7902
|
|
|
7903 su->su_ga = ga;
|
|
|
7904 }
|
|
|
7905
|
|
|
7906 /*
|
|
351
|
7907 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
7908 * badword.
|
|
|
7909 */
|
|
|
7910 static int
|
|
|
7911 stp_sal_score(stp, su, slang, badsound)
|
|
|
7912 suggest_T *stp;
|
|
|
7913 suginfo_T *su;
|
|
|
7914 slang_T *slang;
|
|
|
7915 char_u *badsound; /* sound-folded badword */
|
|
|
7916 {
|
|
|
7917 char_u *p;
|
|
|
7918 char_u badsound2[MAXWLEN];
|
|
|
7919 char_u fword[MAXWLEN];
|
|
|
7920 char_u goodsound[MAXWLEN];
|
|
|
7921
|
|
|
7922 if (stp->st_orglen <= su->su_badlen)
|
|
|
7923 p = badsound;
|
|
|
7924 else
|
|
|
7925 {
|
|
|
7926 /* soundfold the bad word with more characters following */
|
|
|
7927 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
7928
|
|
|
7929 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
7930 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
7931 * removing the space. Don't do it when the good word also contains a
|
|
|
7932 * space. */
|
|
|
7933 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
7934 && *skiptowhite(stp->st_word) == NUL)
|
|
|
7935 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
7936 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
7937
|
|
375
|
7938 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
7939 p = badsound2;
|
|
|
7940 }
|
|
|
7941
|
|
375
|
7942 /* Sound-fold the word and compute the score for the difference. */
|
|
|
7943 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
7944
|
|
|
7945 return soundalike_score(goodsound, p);
|
|
|
7946 }
|
|
|
7947
|
|
|
7948 /*
|
|
323
|
7949 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
7950 */
|
|
|
7951 static void
|
|
346
|
7952 suggest_try_soundalike(su)
|
|
323
|
7953 suginfo_T *su;
|
|
|
7954 {
|
|
|
7955 char_u salword[MAXWLEN];
|
|
|
7956 char_u tword[MAXWLEN];
|
|
|
7957 char_u tsalword[MAXWLEN];
|
|
324
|
7958 idx_T arridx[MAXWLEN];
|
|
323
|
7959 int curi[MAXWLEN];
|
|
|
7960 langp_T *lp;
|
|
|
7961 char_u *byts;
|
|
324
|
7962 idx_T *idxs;
|
|
323
|
7963 int depth;
|
|
|
7964 int c;
|
|
324
|
7965 idx_T n;
|
|
323
|
7966 int round;
|
|
|
7967 int flags;
|
|
344
|
7968 int sound_score;
|
|
|
7969
|
|
|
7970 /* Do this for all languages that support sound folding. */
|
|
323
|
7971 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7972 lp->lp_slang != NULL; ++lp)
|
|
|
7973 {
|
|
|
7974 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7975 {
|
|
|
7976 /* soundfold the bad word */
|
|
375
|
7977 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
7978
|
|
|
7979 /*
|
|
|
7980 * Go through the whole tree, soundfold each word and compare.
|
|
|
7981 * round 1: use the case-folded tree.
|
|
|
7982 * round 2: use the keep-case tree.
|
|
|
7983 */
|
|
|
7984 for (round = 1; round <= 2; ++round)
|
|
|
7985 {
|
|
|
7986 if (round == 1)
|
|
|
7987 {
|
|
|
7988 byts = lp->lp_slang->sl_fbyts;
|
|
|
7989 idxs = lp->lp_slang->sl_fidxs;
|
|
|
7990 }
|
|
|
7991 else
|
|
|
7992 {
|
|
|
7993 byts = lp->lp_slang->sl_kbyts;
|
|
|
7994 idxs = lp->lp_slang->sl_kidxs;
|
|
|
7995 }
|
|
|
7996
|
|
|
7997 depth = 0;
|
|
|
7998 arridx[0] = 0;
|
|
|
7999 curi[0] = 1;
|
|
|
8000 while (depth >= 0 && !got_int)
|
|
|
8001 {
|
|
|
8002 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
8003 {
|
|
323
|
8004 /* Done all bytes at this node, go up one level. */
|
|
|
8005 --depth;
|
|
351
|
8006 line_breakcheck();
|
|
|
8007 }
|
|
323
|
8008 else
|
|
|
8009 {
|
|
|
8010 /* Do one more byte at this node. */
|
|
|
8011 n = arridx[depth] + curi[depth];
|
|
|
8012 ++curi[depth];
|
|
|
8013 c = byts[n];
|
|
|
8014 if (c == 0)
|
|
|
8015 {
|
|
|
8016 /* End of word, deal with the word. */
|
|
324
|
8017 flags = (int)idxs[n];
|
|
323
|
8018 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
8019 {
|
|
|
8020 tword[depth] = NUL;
|
|
375
|
8021 /* Sound-fold. Only in keep-case tree need to
|
|
|
8022 * case-fold the word. */
|
|
|
8023 spell_soundfold(lp->lp_slang, tword,
|
|
|
8024 round == 1, tsalword);
|
|
323
|
8025
|
|
344
|
8026 /* Compute the edit distance between the
|
|
|
8027 * sound-a-like words. */
|
|
|
8028 sound_score = soundalike_score(salword,
|
|
|
8029 tsalword);
|
|
324
|
8030 if (sound_score < SCORE_MAXMAX)
|
|
323
|
8031 {
|
|
324
|
8032 char_u cword[MAXWLEN];
|
|
|
8033 char_u *p;
|
|
344
|
8034 int score;
|
|
324
|
8035
|
|
388
|
8036 flags |= su->su_badflags;
|
|
351
|
8037 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
8038 {
|
|
324
|
8039 /* Need to fix case according to
|
|
|
8040 * "flags". */
|
|
323
|
8041 make_case_word(tword, cword, flags);
|
|
324
|
8042 p = cword;
|
|
323
|
8043 }
|
|
|
8044 else
|
|
324
|
8045 p = tword;
|
|
|
8046
|
|
344
|
8047 if (sps_flags & SPS_DOUBLE)
|
|
|
8048 add_suggestion(su, &su->su_sga, p,
|
|
346
|
8049 su->su_badlen,
|
|
351
|
8050 sound_score, 0, FALSE);
|
|
344
|
8051 else
|
|
|
8052 {
|
|
|
8053 /* Compute the score. */
|
|
|
8054 score = spell_edit_score(
|
|
|
8055 su->su_badword, p);
|
|
|
8056 if (sps_flags & SPS_BEST)
|
|
|
8057 /* give a bonus for the good word
|
|
|
8058 * sounding the same as the bad
|
|
|
8059 * word */
|
|
|
8060 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8061 su->su_badlen,
|
|
344
|
8062 RESCORE(score, sound_score),
|
|
351
|
8063 sound_score, TRUE);
|
|
344
|
8064 else
|
|
|
8065 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8066 su->su_badlen,
|
|
351
|
8067 score + sound_score, 0, FALSE);
|
|
344
|
8068 }
|
|
323
|
8069 }
|
|
|
8070 }
|
|
|
8071
|
|
|
8072 /* Skip over other NUL bytes. */
|
|
|
8073 while (byts[n + 1] == 0)
|
|
|
8074 {
|
|
|
8075 ++n;
|
|
|
8076 ++curi[depth];
|
|
|
8077 }
|
|
|
8078 }
|
|
|
8079 else
|
|
|
8080 {
|
|
|
8081 /* Normal char, go one level deeper. */
|
|
|
8082 tword[depth++] = c;
|
|
|
8083 arridx[depth] = idxs[n];
|
|
|
8084 curi[depth] = 1;
|
|
|
8085 }
|
|
|
8086 }
|
|
|
8087 }
|
|
|
8088 }
|
|
|
8089 }
|
|
|
8090 }
|
|
|
8091 }
|
|
|
8092
|
|
|
8093 /*
|
|
324
|
8094 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
8095 */
|
|
|
8096 static void
|
|
|
8097 make_case_word(fword, cword, flags)
|
|
|
8098 char_u *fword;
|
|
|
8099 char_u *cword;
|
|
|
8100 int flags;
|
|
|
8101 {
|
|
|
8102 if (flags & WF_ALLCAP)
|
|
|
8103 /* Make it all upper-case */
|
|
|
8104 allcap_copy(fword, cword);
|
|
|
8105 else if (flags & WF_ONECAP)
|
|
|
8106 /* Make the first letter upper-case */
|
|
324
|
8107 onecap_copy(fword, cword, TRUE);
|
|
323
|
8108 else
|
|
|
8109 /* Use goodword as-is. */
|
|
|
8110 STRCPY(cword, fword);
|
|
|
8111 }
|
|
|
8112
|
|
330
|
8113 /*
|
|
|
8114 * Use map string "map" for languages "lp".
|
|
|
8115 */
|
|
|
8116 static void
|
|
|
8117 set_map_str(lp, map)
|
|
|
8118 slang_T *lp;
|
|
|
8119 char_u *map;
|
|
|
8120 {
|
|
|
8121 char_u *p;
|
|
|
8122 int headc = 0;
|
|
|
8123 int c;
|
|
|
8124 int i;
|
|
|
8125
|
|
|
8126 if (*map == NUL)
|
|
|
8127 {
|
|
|
8128 lp->sl_has_map = FALSE;
|
|
|
8129 return;
|
|
|
8130 }
|
|
|
8131 lp->sl_has_map = TRUE;
|
|
|
8132
|
|
|
8133 /* Init the array and hash table empty. */
|
|
|
8134 for (i = 0; i < 256; ++i)
|
|
|
8135 lp->sl_map_array[i] = 0;
|
|
|
8136 #ifdef FEAT_MBYTE
|
|
|
8137 hash_init(&lp->sl_map_hash);
|
|
|
8138 #endif
|
|
|
8139
|
|
|
8140 /*
|
|
|
8141 * The similar characters are stored separated with slashes:
|
|
|
8142 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
8143 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
8144 */
|
|
|
8145 for (p = map; *p != NUL; )
|
|
|
8146 {
|
|
|
8147 #ifdef FEAT_MBYTE
|
|
|
8148 c = mb_ptr2char_adv(&p);
|
|
|
8149 #else
|
|
|
8150 c = *p++;
|
|
|
8151 #endif
|
|
|
8152 if (c == '/')
|
|
|
8153 headc = 0;
|
|
|
8154 else
|
|
|
8155 {
|
|
|
8156 if (headc == 0)
|
|
|
8157 headc = c;
|
|
|
8158
|
|
|
8159 #ifdef FEAT_MBYTE
|
|
|
8160 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
8161 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
8162 * a NUL. */
|
|
|
8163 if (c >= 256)
|
|
|
8164 {
|
|
|
8165 int cl = mb_char2len(c);
|
|
|
8166 int headcl = mb_char2len(headc);
|
|
|
8167 char_u *b;
|
|
|
8168 hash_T hash;
|
|
|
8169 hashitem_T *hi;
|
|
|
8170
|
|
|
8171 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
8172 if (b == NULL)
|
|
|
8173 return;
|
|
|
8174 mb_char2bytes(c, b);
|
|
|
8175 b[cl] = NUL;
|
|
|
8176 mb_char2bytes(headc, b + cl + 1);
|
|
|
8177 b[cl + 1 + headcl] = NUL;
|
|
|
8178 hash = hash_hash(b);
|
|
|
8179 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
8180 if (HASHITEM_EMPTY(hi))
|
|
|
8181 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
8182 else
|
|
|
8183 {
|
|
|
8184 /* This should have been checked when generating the .spl
|
|
|
8185 * file. */
|
|
|
8186 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
8187 vim_free(b);
|
|
|
8188 }
|
|
|
8189 }
|
|
|
8190 else
|
|
|
8191 #endif
|
|
|
8192 lp->sl_map_array[c] = headc;
|
|
|
8193 }
|
|
|
8194 }
|
|
|
8195 }
|
|
|
8196
|
|
323
|
8197 /*
|
|
|
8198 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
8199 * lines in the .aff file.
|
|
|
8200 */
|
|
|
8201 static int
|
|
|
8202 similar_chars(slang, c1, c2)
|
|
|
8203 slang_T *slang;
|
|
|
8204 int c1;
|
|
|
8205 int c2;
|
|
|
8206 {
|
|
330
|
8207 int m1, m2;
|
|
|
8208 #ifdef FEAT_MBYTE
|
|
|
8209 char_u buf[MB_MAXBYTES];
|
|
|
8210 hashitem_T *hi;
|
|
|
8211
|
|
|
8212 if (c1 >= 256)
|
|
|
8213 {
|
|
|
8214 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
8215 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8216 if (HASHITEM_EMPTY(hi))
|
|
|
8217 m1 = 0;
|
|
|
8218 else
|
|
|
8219 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8220 }
|
|
|
8221 else
|
|
|
8222 #endif
|
|
|
8223 m1 = slang->sl_map_array[c1];
|
|
|
8224 if (m1 == 0)
|
|
323
|
8225 return FALSE;
|
|
330
|
8226
|
|
|
8227
|
|
|
8228 #ifdef FEAT_MBYTE
|
|
|
8229 if (c2 >= 256)
|
|
|
8230 {
|
|
|
8231 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
8232 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8233 if (HASHITEM_EMPTY(hi))
|
|
|
8234 m2 = 0;
|
|
|
8235 else
|
|
|
8236 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8237 }
|
|
|
8238 else
|
|
|
8239 #endif
|
|
|
8240 m2 = slang->sl_map_array[c2];
|
|
|
8241
|
|
|
8242 return m1 == m2;
|
|
323
|
8243 }
|
|
|
8244
|
|
|
8245 /*
|
|
|
8246 * Add a suggestion to the list of suggestions.
|
|
|
8247 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
8248 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
8249 * with spell_edit_score().
|
|
|
8250 */
|
|
|
8251 static void
|
|
351
|
8252 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
8253 suginfo_T *su;
|
|
344
|
8254 garray_T *gap;
|
|
323
|
8255 char_u *goodword;
|
|
346
|
8256 int badlen; /* length of bad word used */
|
|
324
|
8257 int score;
|
|
351
|
8258 int altscore;
|
|
344
|
8259 int had_bonus; /* value for st_had_bonus */
|
|
323
|
8260 {
|
|
|
8261 suggest_T *stp;
|
|
|
8262 int i;
|
|
346
|
8263 char_u *p = NULL;
|
|
|
8264 int c = 0;
|
|
323
|
8265
|
|
|
8266 /* Check that the word wasn't banned. */
|
|
|
8267 if (was_banned(su, goodword))
|
|
|
8268 return;
|
|
|
8269
|
|
346
|
8270 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
8271 * Remove the common part from "goodword". */
|
|
|
8272 i = badlen - su->su_badlen;
|
|
|
8273 if (i > 0)
|
|
|
8274 {
|
|
|
8275 /* This assumes there was no case folding or it didn't change the
|
|
|
8276 * length... */
|
|
|
8277 p = goodword + STRLEN(goodword) - i;
|
|
|
8278 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
8279 {
|
|
|
8280 badlen = su->su_badlen;
|
|
|
8281 c = *p;
|
|
|
8282 *p = NUL;
|
|
|
8283 }
|
|
|
8284 else
|
|
|
8285 p = NULL;
|
|
|
8286 }
|
|
372
|
8287 else if (i < 0)
|
|
|
8288 {
|
|
|
8289 /* When replacing part of the word check that we actually change
|
|
|
8290 * something. For "the the" a suggestion can be replacing the first
|
|
|
8291 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
8292 if (badlen == (int)STRLEN(goodword)
|
|
372
|
8293 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
8294 return;
|
|
|
8295 }
|
|
|
8296
|
|
346
|
8297
|
|
323
|
8298 if (score <= su->su_maxscore)
|
|
|
8299 {
|
|
366
|
8300 /* Check if the word is already there. Also check the length that is
|
|
|
8301 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
8302 * "thes" -> "these". */
|
|
344
|
8303 stp = &SUG(*gap, 0);
|
|
|
8304 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
8305 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
8306 && stp[i].st_orglen == badlen)
|
|
323
|
8307 {
|
|
|
8308 /* Found it. Remember the lowest score. */
|
|
|
8309 if (stp[i].st_score > score)
|
|
324
|
8310 {
|
|
323
|
8311 stp[i].st_score = score;
|
|
376
|
8312 stp[i].st_altscore = altscore;
|
|
324
|
8313 stp[i].st_had_bonus = had_bonus;
|
|
|
8314 }
|
|
323
|
8315 break;
|
|
|
8316 }
|
|
|
8317
|
|
344
|
8318 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
8319 {
|
|
|
8320 /* Add a suggestion. */
|
|
344
|
8321 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
8322 stp->st_word = vim_strsave(goodword);
|
|
|
8323 if (stp->st_word != NULL)
|
|
|
8324 {
|
|
|
8325 stp->st_score = score;
|
|
351
|
8326 stp->st_altscore = altscore;
|
|
324
|
8327 stp->st_had_bonus = had_bonus;
|
|
346
|
8328 stp->st_orglen = badlen;
|
|
344
|
8329 ++gap->ga_len;
|
|
323
|
8330
|
|
|
8331 /* If we have too many suggestions now, sort the list and keep
|
|
|
8332 * the best suggestions. */
|
|
344
|
8333 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
8334 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
8335 SUG_CLEAN_COUNT(su));
|
|
323
|
8336 }
|
|
|
8337 }
|
|
|
8338 }
|
|
346
|
8339
|
|
|
8340 if (p != NULL)
|
|
|
8341 *p = c; /* restore "goodword" */
|
|
323
|
8342 }
|
|
|
8343
|
|
|
8344 /*
|
|
|
8345 * Add a word to be banned.
|
|
|
8346 */
|
|
|
8347 static void
|
|
|
8348 add_banned(su, word)
|
|
|
8349 suginfo_T *su;
|
|
|
8350 char_u *word;
|
|
|
8351 {
|
|
|
8352 char_u *s = vim_strsave(word);
|
|
|
8353 hash_T hash;
|
|
|
8354 hashitem_T *hi;
|
|
|
8355
|
|
|
8356 if (s != NULL)
|
|
|
8357 {
|
|
|
8358 hash = hash_hash(s);
|
|
|
8359 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
8360 if (HASHITEM_EMPTY(hi))
|
|
|
8361 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
8362 else
|
|
|
8363 vim_free(s);
|
|
323
|
8364 }
|
|
|
8365 }
|
|
|
8366
|
|
|
8367 /*
|
|
|
8368 * Return TRUE if a word appears in the list of banned words.
|
|
|
8369 */
|
|
|
8370 static int
|
|
|
8371 was_banned(su, word)
|
|
|
8372 suginfo_T *su;
|
|
|
8373 char_u *word;
|
|
|
8374 {
|
|
324
|
8375 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
8376
|
|
|
8377 return !HASHITEM_EMPTY(hi);
|
|
323
|
8378 }
|
|
|
8379
|
|
|
8380 /*
|
|
|
8381 * Free the banned words in "su".
|
|
|
8382 */
|
|
|
8383 static void
|
|
|
8384 free_banned(su)
|
|
|
8385 suginfo_T *su;
|
|
|
8386 {
|
|
|
8387 int todo;
|
|
|
8388 hashitem_T *hi;
|
|
|
8389
|
|
|
8390 todo = su->su_banned.ht_used;
|
|
|
8391 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
8392 {
|
|
|
8393 if (!HASHITEM_EMPTY(hi))
|
|
|
8394 {
|
|
|
8395 vim_free(hi->hi_key);
|
|
|
8396 --todo;
|
|
|
8397 }
|
|
|
8398 }
|
|
|
8399 hash_clear(&su->su_banned);
|
|
|
8400 }
|
|
|
8401
|
|
324
|
8402 /*
|
|
|
8403 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
8404 * thus only done for the final results.
|
|
|
8405 */
|
|
|
8406 static void
|
|
|
8407 rescore_suggestions(su)
|
|
|
8408 suginfo_T *su;
|
|
|
8409 {
|
|
|
8410 langp_T *lp;
|
|
|
8411 suggest_T *stp;
|
|
|
8412 char_u sal_badword[MAXWLEN];
|
|
|
8413 int i;
|
|
|
8414
|
|
|
8415 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8416 lp->lp_slang != NULL; ++lp)
|
|
|
8417 {
|
|
|
8418 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8419 {
|
|
|
8420 /* soundfold the bad word */
|
|
375
|
8421 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
8422
|
|
|
8423 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8424 {
|
|
344
|
8425 stp = &SUG(su->su_ga, i);
|
|
324
|
8426 if (!stp->st_had_bonus)
|
|
|
8427 {
|
|
351
|
8428 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
8429 lp->lp_slang, sal_badword);
|
|
|
8430 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8431 stp->st_altscore = SCORE_BIG;
|
|
|
8432 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
8433 }
|
|
|
8434 }
|
|
|
8435 break;
|
|
|
8436 }
|
|
|
8437 }
|
|
|
8438 }
|
|
|
8439
|
|
323
|
8440 static int
|
|
|
8441 #ifdef __BORLANDC__
|
|
|
8442 _RTLENTRYF
|
|
|
8443 #endif
|
|
|
8444 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
8445
|
|
|
8446 /*
|
|
|
8447 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
8448 */
|
|
|
8449 static int
|
|
|
8450 #ifdef __BORLANDC__
|
|
|
8451 _RTLENTRYF
|
|
|
8452 #endif
|
|
|
8453 sug_compare(s1, s2)
|
|
|
8454 const void *s1;
|
|
|
8455 const void *s2;
|
|
|
8456 {
|
|
|
8457 suggest_T *p1 = (suggest_T *)s1;
|
|
|
8458 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
8459 int n = p1->st_score - p2->st_score;
|
|
|
8460
|
|
|
8461 if (n == 0)
|
|
|
8462 return p1->st_altscore - p2->st_altscore;
|
|
|
8463 return n;
|
|
323
|
8464 }
|
|
|
8465
|
|
|
8466 /*
|
|
|
8467 * Cleanup the suggestions:
|
|
|
8468 * - Sort on score.
|
|
|
8469 * - Remove words that won't be displayed.
|
|
344
|
8470 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
8471 */
|
|
344
|
8472 static int
|
|
|
8473 cleanup_suggestions(gap, maxscore, keep)
|
|
|
8474 garray_T *gap;
|
|
|
8475 int maxscore;
|
|
324
|
8476 int keep; /* nr of suggestions to keep */
|
|
323
|
8477 {
|
|
344
|
8478 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
8479 int i;
|
|
|
8480
|
|
|
8481 /* Sort the list. */
|
|
344
|
8482 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
8483
|
|
|
8484 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
8485 if (gap->ga_len > keep)
|
|
323
|
8486 {
|
|
344
|
8487 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
8488 vim_free(stp[i].st_word);
|
|
344
|
8489 gap->ga_len = keep;
|
|
|
8490 return stp[keep - 1].st_score;
|
|
323
|
8491 }
|
|
344
|
8492 return maxscore;
|
|
323
|
8493 }
|
|
|
8494
|
|
372
|
8495 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
8496 /*
|
|
|
8497 * Soundfold a string, for soundfold().
|
|
|
8498 * Result is in allocated memory, NULL for an error.
|
|
|
8499 */
|
|
|
8500 char_u *
|
|
|
8501 eval_soundfold(word)
|
|
|
8502 char_u *word;
|
|
|
8503 {
|
|
|
8504 langp_T *lp;
|
|
|
8505 char_u sound[MAXWLEN];
|
|
|
8506
|
|
|
8507 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
8508 /* Use the sound-folding of the first language that supports it. */
|
|
|
8509 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8510 lp->lp_slang != NULL; ++lp)
|
|
|
8511 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8512 {
|
|
|
8513 /* soundfold the word */
|
|
375
|
8514 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
8515 return vim_strsave(sound);
|
|
|
8516 }
|
|
|
8517
|
|
|
8518 /* No language with sound folding, return word as-is. */
|
|
|
8519 return vim_strsave(word);
|
|
|
8520 }
|
|
|
8521 #endif
|
|
|
8522
|
|
323
|
8523 /*
|
|
|
8524 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
8525 */
|
|
|
8526 static void
|
|
375
|
8527 spell_soundfold(slang, inword, folded, res)
|
|
|
8528 slang_T *slang;
|
|
|
8529 char_u *inword;
|
|
|
8530 int folded; /* "inword" is already case-folded */
|
|
|
8531 char_u *res;
|
|
|
8532 {
|
|
|
8533 char_u fword[MAXWLEN];
|
|
|
8534 char_u *word;
|
|
|
8535
|
|
|
8536 if (slang->sl_sofo)
|
|
|
8537 /* SOFOFROM and SOFOTO used */
|
|
|
8538 spell_soundfold_sofo(slang, inword, res);
|
|
|
8539 else
|
|
|
8540 {
|
|
|
8541 /* SAL items used. Requires the word to be case-folded. */
|
|
|
8542 if (folded)
|
|
|
8543 word = inword;
|
|
|
8544 else
|
|
|
8545 {
|
|
|
8546 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
8547 word = fword;
|
|
|
8548 }
|
|
|
8549
|
|
|
8550 #ifdef FEAT_MBYTE
|
|
|
8551 if (has_mbyte)
|
|
|
8552 spell_soundfold_wsal(slang, word, res);
|
|
|
8553 else
|
|
|
8554 #endif
|
|
|
8555 spell_soundfold_sal(slang, word, res);
|
|
|
8556 }
|
|
|
8557 }
|
|
|
8558
|
|
|
8559 /*
|
|
|
8560 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
8561 * SOFOTO lines.
|
|
|
8562 */
|
|
|
8563 static void
|
|
|
8564 spell_soundfold_sofo(slang, inword, res)
|
|
|
8565 slang_T *slang;
|
|
|
8566 char_u *inword;
|
|
|
8567 char_u *res;
|
|
|
8568 {
|
|
|
8569 char_u *s;
|
|
|
8570 int ri = 0;
|
|
|
8571 int c;
|
|
|
8572
|
|
|
8573 #ifdef FEAT_MBYTE
|
|
|
8574 if (has_mbyte)
|
|
|
8575 {
|
|
|
8576 int prevc = 0;
|
|
|
8577 int *ip;
|
|
|
8578
|
|
|
8579 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
8580 * 255, sl_sal the rest. */
|
|
|
8581 for (s = inword; *s != NUL; )
|
|
|
8582 {
|
|
|
8583 c = mb_ptr2char_adv(&s);
|
|
|
8584 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
8585 c = ' ';
|
|
|
8586 else if (c < 256)
|
|
|
8587 c = slang->sl_sal_first[c];
|
|
|
8588 else
|
|
|
8589 {
|
|
|
8590 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
8591 if (ip == NULL) /* empty list, can't match */
|
|
|
8592 c = NUL;
|
|
|
8593 else
|
|
|
8594 for (;;) /* find "c" in the list */
|
|
|
8595 {
|
|
|
8596 if (*ip == 0) /* not found */
|
|
|
8597 {
|
|
|
8598 c = NUL;
|
|
|
8599 break;
|
|
|
8600 }
|
|
|
8601 if (*ip == c) /* match! */
|
|
|
8602 {
|
|
|
8603 c = ip[1];
|
|
|
8604 break;
|
|
|
8605 }
|
|
|
8606 ip += 2;
|
|
|
8607 }
|
|
|
8608 }
|
|
|
8609
|
|
|
8610 if (c != NUL && c != prevc)
|
|
|
8611 {
|
|
|
8612 ri += mb_char2bytes(c, res + ri);
|
|
|
8613 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
8614 break;
|
|
|
8615 prevc = c;
|
|
|
8616 }
|
|
|
8617 }
|
|
|
8618 }
|
|
|
8619 else
|
|
|
8620 #endif
|
|
|
8621 {
|
|
|
8622 /* The sl_sal_first[] table contains the translation. */
|
|
|
8623 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
8624 {
|
|
|
8625 if (vim_iswhite(c))
|
|
|
8626 c = ' ';
|
|
|
8627 else
|
|
|
8628 c = slang->sl_sal_first[c];
|
|
|
8629 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
8630 res[ri++] = c;
|
|
|
8631 }
|
|
|
8632 }
|
|
|
8633
|
|
|
8634 res[ri] = NUL;
|
|
|
8635 }
|
|
|
8636
|
|
|
8637 static void
|
|
|
8638 spell_soundfold_sal(slang, inword, res)
|
|
323
|
8639 slang_T *slang;
|
|
|
8640 char_u *inword;
|
|
|
8641 char_u *res;
|
|
|
8642 {
|
|
344
|
8643 salitem_T *smp;
|
|
323
|
8644 char_u word[MAXWLEN];
|
|
375
|
8645 char_u *s = inword;
|
|
323
|
8646 char_u *t;
|
|
344
|
8647 char_u *pf;
|
|
323
|
8648 int i, j, z;
|
|
344
|
8649 int reslen;
|
|
323
|
8650 int n, k = 0;
|
|
|
8651 int z0;
|
|
|
8652 int k0;
|
|
|
8653 int n0;
|
|
|
8654 int c;
|
|
|
8655 int pri;
|
|
|
8656 int p0 = -333;
|
|
|
8657 int c0;
|
|
|
8658
|
|
324
|
8659 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
8660 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
8661 if (slang->sl_rem_accents)
|
|
|
8662 {
|
|
|
8663 t = word;
|
|
375
|
8664 while (*s != NUL)
|
|
323
|
8665 {
|
|
324
|
8666 if (vim_iswhite(*s))
|
|
344
|
8667 {
|
|
|
8668 *t++ = ' ';
|
|
|
8669 s = skipwhite(s);
|
|
|
8670 }
|
|
323
|
8671 else
|
|
|
8672 {
|
|
376
|
8673 if (spell_iswordp_nmw(s))
|
|
323
|
8674 *t++ = *s;
|
|
|
8675 ++s;
|
|
|
8676 }
|
|
|
8677 }
|
|
|
8678 *t = NUL;
|
|
|
8679 }
|
|
|
8680 else
|
|
375
|
8681 STRCPY(word, s);
|
|
323
|
8682
|
|
344
|
8683 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
8684
|
|
|
8685 /*
|
|
|
8686 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
8687 * Changed to keep spaces.
|
|
323
|
8688 */
|
|
344
|
8689 i = reslen = z = 0;
|
|
323
|
8690 while ((c = word[i]) != NUL)
|
|
|
8691 {
|
|
344
|
8692 /* Start with the first rule that has the character in the word. */
|
|
323
|
8693 n = slang->sl_sal_first[c];
|
|
|
8694 z0 = 0;
|
|
|
8695
|
|
|
8696 if (n >= 0)
|
|
|
8697 {
|
|
|
8698 /* check all rules for the same letter */
|
|
344
|
8699 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
8700 {
|
|
344
|
8701 /* Quickly skip entries that don't match the word. Most
|
|
|
8702 * entries are less then three chars, optimize for that. */
|
|
|
8703 k = smp[n].sm_leadlen;
|
|
|
8704 if (k > 1)
|
|
323
|
8705 {
|
|
344
|
8706 if (word[i + 1] != s[1])
|
|
|
8707 continue;
|
|
|
8708 if (k > 2)
|
|
|
8709 {
|
|
|
8710 for (j = 2; j < k; ++j)
|
|
|
8711 if (word[i + j] != s[j])
|
|
|
8712 break;
|
|
|
8713 if (j < k)
|
|
|
8714 continue;
|
|
|
8715 }
|
|
323
|
8716 }
|
|
|
8717
|
|
375
|
8718 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
8719 {
|
|
375
|
8720 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
8721 while (*pf != NUL && *pf != word[i + k])
|
|
|
8722 ++pf;
|
|
|
8723 if (*pf == NUL)
|
|
|
8724 continue;
|
|
|
8725 ++k;
|
|
323
|
8726 }
|
|
344
|
8727 s = smp[n].sm_rules;
|
|
|
8728 pri = 5; /* default priority */
|
|
323
|
8729
|
|
|
8730 p0 = *s;
|
|
|
8731 k0 = k;
|
|
|
8732 while (*s == '-' && k > 1)
|
|
|
8733 {
|
|
|
8734 k--;
|
|
|
8735 s++;
|
|
|
8736 }
|
|
|
8737 if (*s == '<')
|
|
|
8738 s++;
|
|
344
|
8739 if (VIM_ISDIGIT(*s))
|
|
323
|
8740 {
|
|
|
8741 /* determine priority */
|
|
|
8742 pri = *s - '0';
|
|
|
8743 s++;
|
|
|
8744 }
|
|
|
8745 if (*s == '^' && *(s + 1) == '^')
|
|
|
8746 s++;
|
|
|
8747
|
|
|
8748 if (*s == NUL
|
|
|
8749 || (*s == '^'
|
|
324
|
8750 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
8751 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
8752 && (*(s + 1) != '$'
|
|
376
|
8753 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
8754 || (*s == '$' && i > 0
|
|
376
|
8755 && spell_iswordp(word + i - 1, curbuf)
|
|
|
8756 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
8757 {
|
|
|
8758 /* search for followup rules, if: */
|
|
|
8759 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
8760 c0 = word[i + k - 1];
|
|
|
8761 n0 = slang->sl_sal_first[c0];
|
|
|
8762
|
|
|
8763 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
8764 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
8765 {
|
|
|
8766 /* test follow-up rule for "word[i + k]" */
|
|
344
|
8767 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
8768 {
|
|
344
|
8769 /* Quickly skip entries that don't match the word.
|
|
|
8770 * */
|
|
|
8771 k0 = smp[n0].sm_leadlen;
|
|
|
8772 if (k0 > 1)
|
|
323
|
8773 {
|
|
344
|
8774 if (word[i + k] != s[1])
|
|
|
8775 continue;
|
|
|
8776 if (k0 > 2)
|
|
|
8777 {
|
|
|
8778 pf = word + i + k + 1;
|
|
|
8779 for (j = 2; j < k0; ++j)
|
|
|
8780 if (*pf++ != s[j])
|
|
|
8781 break;
|
|
|
8782 if (j < k0)
|
|
|
8783 continue;
|
|
|
8784 }
|
|
323
|
8785 }
|
|
344
|
8786 k0 += k - 1;
|
|
|
8787
|
|
375
|
8788 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
8789 {
|
|
344
|
8790 /* Check for match with one of the chars in
|
|
375
|
8791 * "sm_oneof". */
|
|
344
|
8792 while (*pf != NUL && *pf != word[i + k0])
|
|
|
8793 ++pf;
|
|
|
8794 if (*pf == NUL)
|
|
|
8795 continue;
|
|
|
8796 ++k0;
|
|
323
|
8797 }
|
|
344
|
8798
|
|
|
8799 p0 = 5;
|
|
|
8800 s = smp[n0].sm_rules;
|
|
323
|
8801 while (*s == '-')
|
|
|
8802 {
|
|
344
|
8803 /* "k0" gets NOT reduced because
|
|
|
8804 * "if (k0 == k)" */
|
|
323
|
8805 s++;
|
|
|
8806 }
|
|
|
8807 if (*s == '<')
|
|
|
8808 s++;
|
|
344
|
8809 if (VIM_ISDIGIT(*s))
|
|
323
|
8810 {
|
|
|
8811 p0 = *s - '0';
|
|
|
8812 s++;
|
|
|
8813 }
|
|
|
8814
|
|
|
8815 if (*s == NUL
|
|
|
8816 /* *s == '^' cuts */
|
|
|
8817 || (*s == '$'
|
|
376
|
8818 && !spell_iswordp(word + i + k0,
|
|
|
8819 curbuf)))
|
|
323
|
8820 {
|
|
|
8821 if (k0 == k)
|
|
|
8822 /* this is just a piece of the string */
|
|
|
8823 continue;
|
|
|
8824
|
|
|
8825 if (p0 < pri)
|
|
|
8826 /* priority too low */
|
|
|
8827 continue;
|
|
|
8828 /* rule fits; stop search */
|
|
|
8829 break;
|
|
|
8830 }
|
|
|
8831 }
|
|
|
8832
|
|
344
|
8833 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
8834 continue;
|
|
|
8835 }
|
|
|
8836
|
|
|
8837 /* replace string */
|
|
344
|
8838 s = smp[n].sm_to;
|
|
|
8839 pf = smp[n].sm_rules;
|
|
|
8840 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
8841 if (p0 == 1 && z == 0)
|
|
|
8842 {
|
|
|
8843 /* rule with '<' is used */
|
|
344
|
8844 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
8845 || res[reslen - 1] == *s))
|
|
|
8846 reslen--;
|
|
323
|
8847 z0 = 1;
|
|
|
8848 z = 1;
|
|
|
8849 k0 = 0;
|
|
372
|
8850 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
8851 {
|
|
|
8852 word[i + k0] = *s;
|
|
|
8853 k0++;
|
|
|
8854 s++;
|
|
|
8855 }
|
|
|
8856 if (k > k0)
|
|
|
8857 mch_memmove(word + i + k0, word + i + k,
|
|
|
8858 STRLEN(word + i + k) + 1);
|
|
|
8859
|
|
|
8860 /* new "actual letter" */
|
|
|
8861 c = word[i];
|
|
|
8862 }
|
|
|
8863 else
|
|
|
8864 {
|
|
|
8865 /* no '<' rule used */
|
|
|
8866 i += k - 1;
|
|
|
8867 z = 0;
|
|
344
|
8868 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
8869 {
|
|
344
|
8870 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
8871 res[reslen++] = *s;
|
|
323
|
8872 s++;
|
|
|
8873 }
|
|
|
8874 /* new "actual letter" */
|
|
|
8875 c = *s;
|
|
344
|
8876 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
8877 {
|
|
|
8878 if (c != NUL)
|
|
372
|
8879 res[reslen++] = c;
|
|
323
|
8880 mch_memmove(word, word + i + 1,
|
|
|
8881 STRLEN(word + i + 1) + 1);
|
|
|
8882 i = 0;
|
|
|
8883 z0 = 1;
|
|
|
8884 }
|
|
|
8885 }
|
|
|
8886 break;
|
|
|
8887 }
|
|
|
8888 }
|
|
|
8889 }
|
|
324
|
8890 else if (vim_iswhite(c))
|
|
|
8891 {
|
|
|
8892 c = ' ';
|
|
|
8893 k = 1;
|
|
|
8894 }
|
|
323
|
8895
|
|
|
8896 if (z0 == 0)
|
|
|
8897 {
|
|
344
|
8898 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
8899 && (!slang->sl_collapse || reslen == 0
|
|
|
8900 || res[reslen - 1] != c))
|
|
323
|
8901 /* condense only double letters */
|
|
372
|
8902 res[reslen++] = c;
|
|
323
|
8903
|
|
|
8904 i++;
|
|
|
8905 z = 0;
|
|
|
8906 k = 0;
|
|
|
8907 }
|
|
|
8908 }
|
|
|
8909
|
|
344
|
8910 res[reslen] = NUL;
|
|
323
|
8911 }
|
|
|
8912
|
|
372
|
8913 #ifdef FEAT_MBYTE
|
|
|
8914 /*
|
|
|
8915 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
8916 * Multi-byte version of spell_soundfold().
|
|
|
8917 */
|
|
|
8918 static void
|
|
375
|
8919 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
8920 slang_T *slang;
|
|
|
8921 char_u *inword;
|
|
|
8922 char_u *res;
|
|
|
8923 {
|
|
375
|
8924 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
8925 int word[MAXWLEN];
|
|
|
8926 int wres[MAXWLEN];
|
|
|
8927 int l;
|
|
|
8928 char_u *s;
|
|
|
8929 int *ws;
|
|
|
8930 char_u *t;
|
|
|
8931 int *pf;
|
|
|
8932 int i, j, z;
|
|
|
8933 int reslen;
|
|
|
8934 int n, k = 0;
|
|
|
8935 int z0;
|
|
|
8936 int k0;
|
|
|
8937 int n0;
|
|
|
8938 int c;
|
|
|
8939 int pri;
|
|
|
8940 int p0 = -333;
|
|
|
8941 int c0;
|
|
|
8942 int did_white = FALSE;
|
|
|
8943
|
|
|
8944 /*
|
|
|
8945 * Convert the multi-byte string to a wide-character string.
|
|
|
8946 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
8947 * But keep white space.
|
|
|
8948 */
|
|
|
8949 n = 0;
|
|
|
8950 for (s = inword; *s != NUL; )
|
|
|
8951 {
|
|
|
8952 t = s;
|
|
|
8953 c = mb_ptr2char_adv(&s);
|
|
|
8954 if (slang->sl_rem_accents)
|
|
|
8955 {
|
|
|
8956 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
8957 {
|
|
|
8958 if (did_white)
|
|
|
8959 continue;
|
|
|
8960 c = ' ';
|
|
|
8961 did_white = TRUE;
|
|
|
8962 }
|
|
|
8963 else
|
|
|
8964 {
|
|
|
8965 did_white = FALSE;
|
|
376
|
8966 if (!spell_iswordp_nmw(t))
|
|
372
|
8967 continue;
|
|
|
8968 }
|
|
|
8969 }
|
|
|
8970 word[n++] = c;
|
|
|
8971 }
|
|
|
8972 word[n] = NUL;
|
|
|
8973
|
|
|
8974 /*
|
|
|
8975 * This comes from Aspell phonet.cpp.
|
|
|
8976 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
8977 * Changed to keep spaces.
|
|
|
8978 */
|
|
|
8979 i = reslen = z = 0;
|
|
|
8980 while ((c = word[i]) != NUL)
|
|
|
8981 {
|
|
|
8982 /* Start with the first rule that has the character in the word. */
|
|
|
8983 n = slang->sl_sal_first[c & 0xff];
|
|
|
8984 z0 = 0;
|
|
|
8985
|
|
|
8986 if (n >= 0)
|
|
|
8987 {
|
|
375
|
8988 /* check all rules for the same index byte */
|
|
372
|
8989 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
8990 {
|
|
|
8991 /* Quickly skip entries that don't match the word. Most
|
|
|
8992 * entries are less then three chars, optimize for that. */
|
|
375
|
8993 if (c != ws[0])
|
|
|
8994 continue;
|
|
372
|
8995 k = smp[n].sm_leadlen;
|
|
|
8996 if (k > 1)
|
|
|
8997 {
|
|
|
8998 if (word[i + 1] != ws[1])
|
|
|
8999 continue;
|
|
|
9000 if (k > 2)
|
|
|
9001 {
|
|
|
9002 for (j = 2; j < k; ++j)
|
|
|
9003 if (word[i + j] != ws[j])
|
|
|
9004 break;
|
|
|
9005 if (j < k)
|
|
|
9006 continue;
|
|
|
9007 }
|
|
|
9008 }
|
|
|
9009
|
|
375
|
9010 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
9011 {
|
|
375
|
9012 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
9013 while (*pf != NUL && *pf != word[i + k])
|
|
|
9014 ++pf;
|
|
|
9015 if (*pf == NUL)
|
|
|
9016 continue;
|
|
|
9017 ++k;
|
|
|
9018 }
|
|
|
9019 s = smp[n].sm_rules;
|
|
|
9020 pri = 5; /* default priority */
|
|
|
9021
|
|
|
9022 p0 = *s;
|
|
|
9023 k0 = k;
|
|
|
9024 while (*s == '-' && k > 1)
|
|
|
9025 {
|
|
|
9026 k--;
|
|
|
9027 s++;
|
|
|
9028 }
|
|
|
9029 if (*s == '<')
|
|
|
9030 s++;
|
|
|
9031 if (VIM_ISDIGIT(*s))
|
|
|
9032 {
|
|
|
9033 /* determine priority */
|
|
|
9034 pri = *s - '0';
|
|
|
9035 s++;
|
|
|
9036 }
|
|
|
9037 if (*s == '^' && *(s + 1) == '^')
|
|
|
9038 s++;
|
|
|
9039
|
|
|
9040 if (*s == NUL
|
|
|
9041 || (*s == '^'
|
|
|
9042 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9043 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
9044 && (*(s + 1) != '$'
|
|
376
|
9045 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9046 || (*s == '$' && i > 0
|
|
376
|
9047 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
9048 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9049 {
|
|
|
9050 /* search for followup rules, if: */
|
|
|
9051 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9052 c0 = word[i + k - 1];
|
|
|
9053 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
9054
|
|
|
9055 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
9056 && p0 != '-' && word[i + k] != NUL)
|
|
|
9057 {
|
|
375
|
9058 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
9059 * all entries with the same index byte. */
|
|
372
|
9060 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
9061 == (c0 & 0xff); ++n0)
|
|
|
9062 {
|
|
|
9063 /* Quickly skip entries that don't match the word.
|
|
375
|
9064 */
|
|
|
9065 if (c0 != ws[0])
|
|
|
9066 continue;
|
|
372
|
9067 k0 = smp[n0].sm_leadlen;
|
|
|
9068 if (k0 > 1)
|
|
|
9069 {
|
|
|
9070 if (word[i + k] != ws[1])
|
|
|
9071 continue;
|
|
|
9072 if (k0 > 2)
|
|
|
9073 {
|
|
|
9074 pf = word + i + k + 1;
|
|
|
9075 for (j = 2; j < k0; ++j)
|
|
|
9076 if (*pf++ != ws[j])
|
|
|
9077 break;
|
|
|
9078 if (j < k0)
|
|
|
9079 continue;
|
|
|
9080 }
|
|
|
9081 }
|
|
|
9082 k0 += k - 1;
|
|
|
9083
|
|
375
|
9084 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
9085 {
|
|
|
9086 /* Check for match with one of the chars in
|
|
375
|
9087 * "sm_oneof". */
|
|
372
|
9088 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9089 ++pf;
|
|
|
9090 if (*pf == NUL)
|
|
|
9091 continue;
|
|
|
9092 ++k0;
|
|
|
9093 }
|
|
|
9094
|
|
|
9095 p0 = 5;
|
|
|
9096 s = smp[n0].sm_rules;
|
|
|
9097 while (*s == '-')
|
|
|
9098 {
|
|
|
9099 /* "k0" gets NOT reduced because
|
|
|
9100 * "if (k0 == k)" */
|
|
|
9101 s++;
|
|
|
9102 }
|
|
|
9103 if (*s == '<')
|
|
|
9104 s++;
|
|
|
9105 if (VIM_ISDIGIT(*s))
|
|
|
9106 {
|
|
|
9107 p0 = *s - '0';
|
|
|
9108 s++;
|
|
|
9109 }
|
|
|
9110
|
|
|
9111 if (*s == NUL
|
|
|
9112 /* *s == '^' cuts */
|
|
|
9113 || (*s == '$'
|
|
376
|
9114 && !spell_iswordp_w(word + i + k0,
|
|
|
9115 curbuf)))
|
|
372
|
9116 {
|
|
|
9117 if (k0 == k)
|
|
|
9118 /* this is just a piece of the string */
|
|
|
9119 continue;
|
|
|
9120
|
|
|
9121 if (p0 < pri)
|
|
|
9122 /* priority too low */
|
|
|
9123 continue;
|
|
|
9124 /* rule fits; stop search */
|
|
|
9125 break;
|
|
|
9126 }
|
|
|
9127 }
|
|
|
9128
|
|
|
9129 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
9130 == (c0 & 0xff))
|
|
|
9131 continue;
|
|
|
9132 }
|
|
|
9133
|
|
|
9134 /* replace string */
|
|
|
9135 ws = smp[n].sm_to_w;
|
|
|
9136 s = smp[n].sm_rules;
|
|
|
9137 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
9138 if (p0 == 1 && z == 0)
|
|
|
9139 {
|
|
|
9140 /* rule with '<' is used */
|
|
|
9141 if (reslen > 0 && *ws != NUL && (wres[reslen - 1] == c
|
|
|
9142 || wres[reslen - 1] == *ws))
|
|
|
9143 reslen--;
|
|
|
9144 z0 = 1;
|
|
|
9145 z = 1;
|
|
|
9146 k0 = 0;
|
|
|
9147 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
9148 {
|
|
|
9149 word[i + k0] = *ws;
|
|
|
9150 k0++;
|
|
|
9151 ws++;
|
|
|
9152 }
|
|
|
9153 if (k > k0)
|
|
|
9154 mch_memmove(word + i + k0, word + i + k,
|
|
|
9155 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
9156
|
|
|
9157 /* new "actual letter" */
|
|
|
9158 c = word[i];
|
|
|
9159 }
|
|
|
9160 else
|
|
|
9161 {
|
|
|
9162 /* no '<' rule used */
|
|
|
9163 i += k - 1;
|
|
|
9164 z = 0;
|
|
|
9165 while (*ws != NUL && ws[1] != NUL && reslen < MAXWLEN)
|
|
|
9166 {
|
|
|
9167 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
9168 wres[reslen++] = *ws;
|
|
|
9169 ws++;
|
|
|
9170 }
|
|
|
9171 /* new "actual letter" */
|
|
|
9172 c = *ws;
|
|
|
9173 if (strstr((char *)s, "^^") != NULL)
|
|
|
9174 {
|
|
|
9175 if (c != NUL)
|
|
|
9176 wres[reslen++] = c;
|
|
|
9177 mch_memmove(word, word + i + 1,
|
|
|
9178 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
9179 i = 0;
|
|
|
9180 z0 = 1;
|
|
|
9181 }
|
|
|
9182 }
|
|
|
9183 break;
|
|
|
9184 }
|
|
|
9185 }
|
|
|
9186 }
|
|
|
9187 else if (vim_iswhite(c))
|
|
|
9188 {
|
|
|
9189 c = ' ';
|
|
|
9190 k = 1;
|
|
|
9191 }
|
|
|
9192
|
|
|
9193 if (z0 == 0)
|
|
|
9194 {
|
|
|
9195 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9196 && (!slang->sl_collapse || reslen == 0
|
|
|
9197 || wres[reslen - 1] != c))
|
|
|
9198 /* condense only double letters */
|
|
|
9199 wres[reslen++] = c;
|
|
|
9200
|
|
|
9201 i++;
|
|
|
9202 z = 0;
|
|
|
9203 k = 0;
|
|
|
9204 }
|
|
|
9205 }
|
|
|
9206
|
|
|
9207 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
9208 l = 0;
|
|
|
9209 for (n = 0; n < reslen; ++n)
|
|
|
9210 {
|
|
|
9211 l += mb_char2bytes(wres[n], res + l);
|
|
|
9212 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
9213 break;
|
|
|
9214 }
|
|
|
9215 res[l] = NUL;
|
|
|
9216 }
|
|
|
9217 #endif
|
|
|
9218
|
|
324
|
9219 /*
|
|
344
|
9220 * Compute a score for two sound-a-like words.
|
|
|
9221 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
9222 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
9223 * avoiding checks that will not be possible.
|
|
|
9224 */
|
|
|
9225 static int
|
|
351
|
9226 soundalike_score(goodstart, badstart)
|
|
|
9227 char_u *goodstart; /* sound-folded good word */
|
|
|
9228 char_u *badstart; /* sound-folded bad word */
|
|
344
|
9229 {
|
|
351
|
9230 char_u *goodsound = goodstart;
|
|
|
9231 char_u *badsound = badstart;
|
|
|
9232 int goodlen;
|
|
|
9233 int badlen;
|
|
344
|
9234 int n;
|
|
|
9235 char_u *pl, *ps;
|
|
|
9236 char_u *pl2, *ps2;
|
|
351
|
9237 int score = 0;
|
|
|
9238
|
|
|
9239 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
9240 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
9241 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
9242 {
|
|
|
9243 score = SCORE_DEL / 2;
|
|
|
9244 if (*badsound == '*')
|
|
|
9245 ++badsound;
|
|
|
9246 else
|
|
|
9247 ++goodsound;
|
|
|
9248 }
|
|
|
9249
|
|
|
9250 goodlen = STRLEN(goodsound);
|
|
|
9251 badlen = STRLEN(badsound);
|
|
344
|
9252
|
|
|
9253 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
9254 * changes. */
|
|
|
9255 n = goodlen - badlen;
|
|
|
9256 if (n < -2 || n > 2)
|
|
|
9257 return SCORE_MAXMAX;
|
|
|
9258
|
|
|
9259 if (n > 0)
|
|
|
9260 {
|
|
351
|
9261 pl = goodsound; /* goodsound is longest */
|
|
344
|
9262 ps = badsound;
|
|
|
9263 }
|
|
|
9264 else
|
|
|
9265 {
|
|
351
|
9266 pl = badsound; /* badsound is longest */
|
|
344
|
9267 ps = goodsound;
|
|
|
9268 }
|
|
|
9269
|
|
|
9270 /* Skip over the identical part. */
|
|
|
9271 while (*pl == *ps && *pl != NUL)
|
|
|
9272 {
|
|
|
9273 ++pl;
|
|
|
9274 ++ps;
|
|
|
9275 }
|
|
|
9276
|
|
|
9277 switch (n)
|
|
|
9278 {
|
|
|
9279 case -2:
|
|
|
9280 case 2:
|
|
|
9281 /*
|
|
|
9282 * Must delete two characters from "pl".
|
|
|
9283 */
|
|
|
9284 ++pl; /* first delete */
|
|
|
9285 while (*pl == *ps)
|
|
|
9286 {
|
|
|
9287 ++pl;
|
|
|
9288 ++ps;
|
|
|
9289 }
|
|
|
9290 /* strings must be equal after second delete */
|
|
|
9291 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
9292 return score + SCORE_DEL * 2;
|
|
344
|
9293
|
|
|
9294 /* Failed to compare. */
|
|
|
9295 break;
|
|
|
9296
|
|
|
9297 case -1:
|
|
|
9298 case 1:
|
|
|
9299 /*
|
|
|
9300 * Minimal one delete from "pl" required.
|
|
|
9301 */
|
|
|
9302
|
|
|
9303 /* 1: delete */
|
|
|
9304 pl2 = pl + 1;
|
|
|
9305 ps2 = ps;
|
|
|
9306 while (*pl2 == *ps2)
|
|
|
9307 {
|
|
|
9308 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9309 return score + SCORE_DEL;
|
|
344
|
9310 ++pl2;
|
|
|
9311 ++ps2;
|
|
|
9312 }
|
|
|
9313
|
|
|
9314 /* 2: delete then swap, then rest must be equal */
|
|
|
9315 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9316 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9317 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
9318
|
|
|
9319 /* 3: delete then substitute, then the rest must be equal */
|
|
|
9320 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9321 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
9322
|
|
|
9323 /* 4: first swap then delete */
|
|
|
9324 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9325 {
|
|
|
9326 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9327 ps2 = ps + 2;
|
|
|
9328 while (*pl2 == *ps2)
|
|
|
9329 {
|
|
|
9330 ++pl2;
|
|
|
9331 ++ps2;
|
|
|
9332 }
|
|
|
9333 /* delete a char and then strings must be equal */
|
|
|
9334 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9335 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
9336 }
|
|
|
9337
|
|
|
9338 /* 5: first substitute then delete */
|
|
|
9339 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
9340 ps2 = ps + 1;
|
|
|
9341 while (*pl2 == *ps2)
|
|
|
9342 {
|
|
|
9343 ++pl2;
|
|
|
9344 ++ps2;
|
|
|
9345 }
|
|
|
9346 /* delete a char and then strings must be equal */
|
|
|
9347 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9348 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
9349
|
|
|
9350 /* Failed to compare. */
|
|
|
9351 break;
|
|
|
9352
|
|
|
9353 case 0:
|
|
|
9354 /*
|
|
|
9355 * Lenghts are equal, thus changes must result in same length: An
|
|
|
9356 * insert is only possible in combination with a delete.
|
|
|
9357 * 1: check if for identical strings
|
|
|
9358 */
|
|
|
9359 if (*pl == NUL)
|
|
351
|
9360 return score;
|
|
344
|
9361
|
|
|
9362 /* 2: swap */
|
|
|
9363 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9364 {
|
|
|
9365 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9366 ps2 = ps + 2;
|
|
|
9367 while (*pl2 == *ps2)
|
|
|
9368 {
|
|
|
9369 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9370 return score + SCORE_SWAP;
|
|
344
|
9371 ++pl2;
|
|
|
9372 ++ps2;
|
|
|
9373 }
|
|
|
9374 /* 3: swap and swap again */
|
|
|
9375 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9376 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9377 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
9378
|
|
|
9379 /* 4: swap and substitute */
|
|
|
9380 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9381 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
9382 }
|
|
|
9383
|
|
|
9384 /* 5: substitute */
|
|
|
9385 pl2 = pl + 1;
|
|
|
9386 ps2 = ps + 1;
|
|
|
9387 while (*pl2 == *ps2)
|
|
|
9388 {
|
|
|
9389 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9390 return score + SCORE_SUBST;
|
|
344
|
9391 ++pl2;
|
|
|
9392 ++ps2;
|
|
|
9393 }
|
|
|
9394
|
|
|
9395 /* 6: substitute and swap */
|
|
|
9396 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9397 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9398 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
9399
|
|
|
9400 /* 7: substitute and substitute */
|
|
|
9401 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9402 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
9403
|
|
|
9404 /* 8: insert then delete */
|
|
|
9405 pl2 = pl;
|
|
|
9406 ps2 = ps + 1;
|
|
|
9407 while (*pl2 == *ps2)
|
|
|
9408 {
|
|
|
9409 ++pl2;
|
|
|
9410 ++ps2;
|
|
|
9411 }
|
|
|
9412 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9413 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9414
|
|
|
9415 /* 9: delete then insert */
|
|
|
9416 pl2 = pl + 1;
|
|
|
9417 ps2 = ps;
|
|
|
9418 while (*pl2 == *ps2)
|
|
|
9419 {
|
|
|
9420 ++pl2;
|
|
|
9421 ++ps2;
|
|
|
9422 }
|
|
|
9423 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
9424 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9425
|
|
|
9426 /* Failed to compare. */
|
|
|
9427 break;
|
|
|
9428 }
|
|
|
9429
|
|
|
9430 return SCORE_MAXMAX;
|
|
|
9431 }
|
|
324
|
9432
|
|
323
|
9433 /*
|
|
|
9434 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
9435 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
9436 *
|
|
323
|
9437 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
9438 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
9439 * characters.
|
|
323
|
9440 */
|
|
|
9441 static int
|
|
|
9442 spell_edit_score(badword, goodword)
|
|
|
9443 char_u *badword;
|
|
|
9444 char_u *goodword;
|
|
|
9445 {
|
|
|
9446 int *cnt;
|
|
372
|
9447 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
9448 int j, i;
|
|
|
9449 int t;
|
|
|
9450 int bc, gc;
|
|
324
|
9451 int pbc, pgc;
|
|
|
9452 #ifdef FEAT_MBYTE
|
|
|
9453 char_u *p;
|
|
|
9454 int wbadword[MAXWLEN];
|
|
|
9455 int wgoodword[MAXWLEN];
|
|
|
9456
|
|
|
9457 if (has_mbyte)
|
|
|
9458 {
|
|
|
9459 /* Get the characters from the multi-byte strings and put them in an
|
|
|
9460 * int array for easy access. */
|
|
|
9461 for (p = badword, badlen = 0; *p != NUL; )
|
|
|
9462 wbadword[badlen++] = mb_ptr2char_adv(&p);
|
|
|
9463 ++badlen;
|
|
|
9464 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
|
9465 wgoodword[goodlen++] = mb_ptr2char_adv(&p);
|
|
|
9466 ++goodlen;
|
|
|
9467 }
|
|
|
9468 else
|
|
|
9469 #endif
|
|
|
9470 {
|
|
|
9471 badlen = STRLEN(badword) + 1;
|
|
|
9472 goodlen = STRLEN(goodword) + 1;
|
|
|
9473 }
|
|
323
|
9474
|
|
|
9475 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
9476 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
9477 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
9478 TRUE);
|
|
324
|
9479 if (cnt == NULL)
|
|
|
9480 return 0; /* out of memory */
|
|
323
|
9481
|
|
|
9482 CNT(0, 0) = 0;
|
|
|
9483 for (j = 1; j <= goodlen; ++j)
|
|
|
9484 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
9485
|
|
|
9486 for (i = 1; i <= badlen; ++i)
|
|
|
9487 {
|
|
|
9488 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
9489 for (j = 1; j <= goodlen; ++j)
|
|
|
9490 {
|
|
324
|
9491 #ifdef FEAT_MBYTE
|
|
|
9492 if (has_mbyte)
|
|
|
9493 {
|
|
|
9494 bc = wbadword[i - 1];
|
|
|
9495 gc = wgoodword[j - 1];
|
|
|
9496 }
|
|
|
9497 else
|
|
|
9498 #endif
|
|
|
9499 {
|
|
|
9500 bc = badword[i - 1];
|
|
|
9501 gc = goodword[j - 1];
|
|
|
9502 }
|
|
323
|
9503 if (bc == gc)
|
|
|
9504 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
9505 else
|
|
|
9506 {
|
|
|
9507 /* Use a better score when there is only a case difference. */
|
|
324
|
9508 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
9509 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
9510 else
|
|
|
9511 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
9512
|
|
324
|
9513 if (i > 1 && j > 1)
|
|
323
|
9514 {
|
|
324
|
9515 #ifdef FEAT_MBYTE
|
|
|
9516 if (has_mbyte)
|
|
|
9517 {
|
|
|
9518 pbc = wbadword[i - 2];
|
|
|
9519 pgc = wgoodword[j - 2];
|
|
|
9520 }
|
|
|
9521 else
|
|
|
9522 #endif
|
|
|
9523 {
|
|
|
9524 pbc = badword[i - 2];
|
|
|
9525 pgc = goodword[j - 2];
|
|
|
9526 }
|
|
|
9527 if (bc == pgc && pbc == gc)
|
|
|
9528 {
|
|
|
9529 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
9530 if (t < CNT(i, j))
|
|
|
9531 CNT(i, j) = t;
|
|
|
9532 }
|
|
323
|
9533 }
|
|
|
9534 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
9535 if (t < CNT(i, j))
|
|
|
9536 CNT(i, j) = t;
|
|
|
9537 t = SCORE_INS + CNT(i, j - 1);
|
|
|
9538 if (t < CNT(i, j))
|
|
|
9539 CNT(i, j) = t;
|
|
|
9540 }
|
|
|
9541 }
|
|
|
9542 }
|
|
344
|
9543
|
|
|
9544 i = CNT(badlen - 1, goodlen - 1);
|
|
|
9545 vim_free(cnt);
|
|
|
9546 return i;
|
|
323
|
9547 }
|
|
307
|
9548
|
|
351
|
9549 /*
|
|
|
9550 * ":spelldump"
|
|
|
9551 */
|
|
|
9552 /*ARGSUSED*/
|
|
|
9553 void
|
|
|
9554 ex_spelldump(eap)
|
|
|
9555 exarg_T *eap;
|
|
|
9556 {
|
|
|
9557 buf_T *buf = curbuf;
|
|
|
9558 langp_T *lp;
|
|
|
9559 slang_T *slang;
|
|
|
9560 idx_T arridx[MAXWLEN];
|
|
|
9561 int curi[MAXWLEN];
|
|
|
9562 char_u word[MAXWLEN];
|
|
|
9563 int c;
|
|
|
9564 char_u *byts;
|
|
|
9565 idx_T *idxs;
|
|
|
9566 linenr_T lnum = 0;
|
|
|
9567 int round;
|
|
|
9568 int depth;
|
|
|
9569 int n;
|
|
|
9570 int flags;
|
|
381
|
9571 char_u *region_names = NULL; /* region names being used */
|
|
|
9572 int do_region = TRUE; /* dump region names and numbers */
|
|
|
9573 char_u *p;
|
|
351
|
9574
|
|
|
9575 if (no_spell_checking())
|
|
|
9576 return;
|
|
|
9577
|
|
|
9578 /* Create a new empty buffer by splitting the window. */
|
|
|
9579 do_cmdline_cmd((char_u *)"new");
|
|
|
9580 if (!bufempty() || !buf_valid(buf))
|
|
|
9581 return;
|
|
|
9582
|
|
381
|
9583 /* Find out if we can support regions: All languages must support the same
|
|
|
9584 * regions or none at all. */
|
|
|
9585 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
9586 {
|
|
|
9587 p = lp->lp_slang->sl_regions;
|
|
|
9588 if (p[0] != 0)
|
|
|
9589 {
|
|
|
9590 if (region_names == NULL) /* first language with regions */
|
|
|
9591 region_names = p;
|
|
|
9592 else if (STRCMP(region_names, p) != 0)
|
|
|
9593 {
|
|
|
9594 do_region = FALSE; /* region names are different */
|
|
|
9595 break;
|
|
|
9596 }
|
|
|
9597 }
|
|
|
9598 }
|
|
|
9599
|
|
|
9600 if (do_region && region_names != NULL)
|
|
|
9601 {
|
|
|
9602 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
9603 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
9604 }
|
|
|
9605 else
|
|
|
9606 do_region = FALSE;
|
|
|
9607
|
|
|
9608 /*
|
|
|
9609 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
9610 */
|
|
351
|
9611 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
9612 {
|
|
|
9613 slang = lp->lp_slang;
|
|
|
9614
|
|
|
9615 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
9616 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
9617
|
|
|
9618 /* round 1: case-folded tree
|
|
|
9619 * round 2: keep-case tree */
|
|
|
9620 for (round = 1; round <= 2; ++round)
|
|
|
9621 {
|
|
|
9622 if (round == 1)
|
|
|
9623 {
|
|
|
9624 byts = slang->sl_fbyts;
|
|
|
9625 idxs = slang->sl_fidxs;
|
|
|
9626 }
|
|
|
9627 else
|
|
|
9628 {
|
|
|
9629 byts = slang->sl_kbyts;
|
|
|
9630 idxs = slang->sl_kidxs;
|
|
|
9631 }
|
|
|
9632 if (byts == NULL)
|
|
|
9633 continue; /* array is empty */
|
|
|
9634
|
|
|
9635 depth = 0;
|
|
|
9636 arridx[0] = 0;
|
|
|
9637 curi[0] = 1;
|
|
|
9638 while (depth >= 0 && !got_int)
|
|
|
9639 {
|
|
|
9640 if (curi[depth] > byts[arridx[depth]])
|
|
|
9641 {
|
|
|
9642 /* Done all bytes at this node, go up one level. */
|
|
|
9643 --depth;
|
|
|
9644 line_breakcheck();
|
|
|
9645 }
|
|
|
9646 else
|
|
|
9647 {
|
|
|
9648 /* Do one more byte at this node. */
|
|
|
9649 n = arridx[depth] + curi[depth];
|
|
|
9650 ++curi[depth];
|
|
|
9651 c = byts[n];
|
|
|
9652 if (c == 0)
|
|
|
9653 {
|
|
|
9654 /* End of word, deal with the word.
|
|
|
9655 * Don't use keep-case words in the fold-case tree,
|
|
|
9656 * they will appear in the keep-case tree.
|
|
|
9657 * Only use the word when the region matches. */
|
|
|
9658 flags = (int)idxs[n];
|
|
|
9659 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
9660 && (do_region
|
|
|
9661 || (flags & WF_REGION) == 0
|
|
|
9662 || (((unsigned)flags >> 8)
|
|
351
|
9663 & lp->lp_region) != 0))
|
|
|
9664 {
|
|
|
9665 word[depth] = NUL;
|
|
381
|
9666 if (!do_region)
|
|
|
9667 flags &= ~WF_REGION;
|
|
355
|
9668
|
|
|
9669 /* Dump the basic word if there is no prefix or
|
|
|
9670 * when it's the first one. */
|
|
|
9671 c = (unsigned)flags >> 16;
|
|
|
9672 if (c == 0 || curi[depth] == 2)
|
|
|
9673 dump_word(word, round, flags, lnum++);
|
|
351
|
9674
|
|
|
9675 /* Apply the prefix, if there is one. */
|
|
355
|
9676 if (c != 0)
|
|
351
|
9677 lnum = apply_prefixes(slang, word, round,
|
|
|
9678 flags, lnum);
|
|
|
9679 }
|
|
|
9680 }
|
|
|
9681 else
|
|
|
9682 {
|
|
|
9683 /* Normal char, go one level deeper. */
|
|
|
9684 word[depth++] = c;
|
|
|
9685 arridx[depth] = idxs[n];
|
|
|
9686 curi[depth] = 1;
|
|
|
9687 }
|
|
|
9688 }
|
|
|
9689 }
|
|
|
9690 }
|
|
|
9691 }
|
|
|
9692
|
|
|
9693 /* Delete the empty line that we started with. */
|
|
|
9694 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
9695 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
9696
|
|
|
9697 redraw_later(NOT_VALID);
|
|
|
9698 }
|
|
|
9699
|
|
|
9700 /*
|
|
|
9701 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
9702 */
|
|
|
9703 static void
|
|
|
9704 dump_word(word, round, flags, lnum)
|
|
|
9705 char_u *word;
|
|
|
9706 int round;
|
|
|
9707 int flags;
|
|
|
9708 linenr_T lnum;
|
|
|
9709 {
|
|
|
9710 int keepcap = FALSE;
|
|
|
9711 char_u *p;
|
|
|
9712 char_u cword[MAXWLEN];
|
|
381
|
9713 char_u badword[MAXWLEN + 10];
|
|
|
9714 int i;
|
|
351
|
9715
|
|
|
9716 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
9717 {
|
|
|
9718 /* Need to fix case according to "flags". */
|
|
|
9719 make_case_word(word, cword, flags);
|
|
|
9720 p = cword;
|
|
|
9721 }
|
|
|
9722 else
|
|
|
9723 {
|
|
|
9724 p = word;
|
|
|
9725 if (round == 2 && (captype(word, NULL) & WF_KEEPCAP) == 0)
|
|
|
9726 keepcap = TRUE;
|
|
|
9727 }
|
|
|
9728
|
|
381
|
9729 /* Add flags and regions after a slash. */
|
|
|
9730 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
9731 {
|
|
|
9732 STRCPY(badword, p);
|
|
|
9733 STRCAT(badword, "/");
|
|
351
|
9734 if (keepcap)
|
|
|
9735 STRCAT(badword, "=");
|
|
|
9736 if (flags & WF_BANNED)
|
|
|
9737 STRCAT(badword, "!");
|
|
|
9738 else if (flags & WF_RARE)
|
|
|
9739 STRCAT(badword, "?");
|
|
381
|
9740 if (flags & WF_REGION)
|
|
|
9741 for (i = 0; i < 7; ++i)
|
|
|
9742 if (flags & (0x100 << i))
|
|
|
9743 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
9744 p = badword;
|
|
|
9745 }
|
|
|
9746
|
|
|
9747 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
9748 }
|
|
|
9749
|
|
|
9750 /*
|
|
372
|
9751 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
9752 * "word" and append a line to the buffer.
|
|
351
|
9753 * Return the updated line number.
|
|
|
9754 */
|
|
|
9755 static linenr_T
|
|
|
9756 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
9757 slang_T *slang;
|
|
|
9758 char_u *word; /* case-folded word */
|
|
|
9759 int round;
|
|
|
9760 int flags; /* flags with prefix ID */
|
|
|
9761 linenr_T startlnum;
|
|
|
9762 {
|
|
|
9763 idx_T arridx[MAXWLEN];
|
|
|
9764 int curi[MAXWLEN];
|
|
|
9765 char_u prefix[MAXWLEN];
|
|
|
9766 int c;
|
|
|
9767 char_u *byts;
|
|
|
9768 idx_T *idxs;
|
|
|
9769 linenr_T lnum = startlnum;
|
|
|
9770 int depth;
|
|
|
9771 int n;
|
|
|
9772 int len;
|
|
|
9773 int prefid = (unsigned)flags >> 16;
|
|
|
9774 int i;
|
|
|
9775
|
|
|
9776 byts = slang->sl_pbyts;
|
|
|
9777 idxs = slang->sl_pidxs;
|
|
|
9778 if (byts != NULL) /* array not is empty */
|
|
|
9779 {
|
|
|
9780 /*
|
|
|
9781 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
|
9782 * When at the end of a prefix check that it supports "prefid".
|
|
|
9783 */
|
|
|
9784 depth = 0;
|
|
|
9785 arridx[0] = 0;
|
|
|
9786 curi[0] = 1;
|
|
|
9787 while (depth >= 0 && !got_int)
|
|
|
9788 {
|
|
|
9789 len = arridx[depth];
|
|
|
9790 if (curi[depth] > byts[len])
|
|
|
9791 {
|
|
|
9792 /* Done all bytes at this node, go up one level. */
|
|
|
9793 --depth;
|
|
|
9794 line_breakcheck();
|
|
|
9795 }
|
|
|
9796 else
|
|
|
9797 {
|
|
|
9798 /* Do one more byte at this node. */
|
|
|
9799 n = len + curi[depth];
|
|
|
9800 ++curi[depth];
|
|
|
9801 c = byts[n];
|
|
|
9802 if (c == 0)
|
|
|
9803 {
|
|
|
9804 /* End of prefix, find out how many IDs there are. */
|
|
|
9805 for (i = 1; i < len; ++i)
|
|
|
9806 if (byts[n + i] != 0)
|
|
|
9807 break;
|
|
|
9808 curi[depth] += i - 1;
|
|
|
9809
|
|
366
|
9810 i = valid_word_prefix(i, n, prefid, word, slang);
|
|
|
9811 if (i != 0)
|
|
351
|
9812 {
|
|
376
|
9813 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
9814 dump_word(prefix, round,
|
|
|
9815 (i & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
9816 : flags, lnum++);
|
|
351
|
9817 }
|
|
|
9818 }
|
|
|
9819 else
|
|
|
9820 {
|
|
|
9821 /* Normal char, go one level deeper. */
|
|
|
9822 prefix[depth++] = c;
|
|
|
9823 arridx[depth] = idxs[n];
|
|
|
9824 curi[depth] = 1;
|
|
|
9825 }
|
|
|
9826 }
|
|
|
9827 }
|
|
|
9828 }
|
|
|
9829
|
|
|
9830 return lnum;
|
|
|
9831 }
|
|
|
9832
|
|
236
|
9833 #endif /* FEAT_SYN_HL */
|
