|
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
|
|
389
|
195 * WF_KEEPCAP keep-case word
|
|
|
196 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
197 * WF_RARE rare word
|
|
389
|
198 * WF_BANNED bad word
|
|
300
|
199 * WF_REGION <region> follows
|
|
339
|
200 * WF_PFX <prefixID> follows
|
|
300
|
201 *
|
|
|
202 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
203 * omitted it's valid in all regions.
|
|
|
204 * Lowest bit is for region 1.
|
|
|
205 *
|
|
339
|
206 * <prefixID> 1 byte ID of prefix that can be used with this word. For
|
|
|
207 * PREFIXTREE used for the required prefix ID.
|
|
|
208 *
|
|
|
209 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
210 * from HEADER.
|
|
300
|
211 *
|
|
|
212 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
213 */
|
|
|
214
|
|
223
|
215 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
216 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
217 #endif
|
|
|
218
|
|
|
219 #include "vim.h"
|
|
|
220
|
|
|
221 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
222
|
|
|
223 #ifdef HAVE_FCNTL_H
|
|
|
224 # include <fcntl.h>
|
|
|
225 #endif
|
|
|
226
|
|
323
|
227 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
228 Some places assume a word length fits in a
|
|
|
229 byte, thus it can't be above 255. */
|
|
226
|
230
|
|
324
|
231 /* Type used for indexes in the word tree need to be at least 3 bytes. If int
|
|
|
232 * is 8 bytes we could use something smaller, but what? */
|
|
|
233 #if SIZEOF_INT > 2
|
|
|
234 typedef int idx_T;
|
|
|
235 #else
|
|
|
236 typedef long idx_T;
|
|
|
237 #endif
|
|
|
238
|
|
|
239 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
240 * comes above it. */
|
|
300
|
241 #define WF_REGION 0x01 /* region byte follows */
|
|
|
242 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
243 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
244 #define WF_RARE 0x08 /* rare word */
|
|
307
|
245 #define WF_BANNED 0x10 /* bad word */
|
|
339
|
246 #define WF_PFX 0x20 /* prefix ID list follows */
|
|
389
|
247 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
248 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
249
|
|
389
|
250 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
251
|
|
366
|
252 #define WF_RAREPFX 0x1000000 /* in sl_pidxs: flag for rare postponed
|
|
|
253 prefix; must be above prefixID (one byte)
|
|
|
254 and prefcondnr (two bytes) */
|
|
|
255
|
|
300
|
256 #define BY_NOFLAGS 0 /* end of word without flags or region */
|
|
|
257 #define BY_FLAGS 1 /* end of word, flag byte follows */
|
|
|
258 #define BY_INDEX 2 /* child is shared, index follows */
|
|
|
259 #define BY_SPECIAL BY_INDEX /* hightest special byte value */
|
|
236
|
260
|
|
323
|
261 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
262 * and si_sal. Not for sl_sal!
|
|
323
|
263 * One replacement: from "ft_from" to "ft_to". */
|
|
|
264 typedef struct fromto_S
|
|
236
|
265 {
|
|
323
|
266 char_u *ft_from;
|
|
|
267 char_u *ft_to;
|
|
|
268 } fromto_T;
|
|
236
|
269
|
|
344
|
270 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
271 * The info is split for quick processing by spell_soundfold().
|
|
|
272 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
273 typedef struct salitem_S
|
|
|
274 {
|
|
|
275 char_u *sm_lead; /* leading letters */
|
|
|
276 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
277 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
278 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
279 char_u *sm_to; /* replacement. */
|
|
372
|
280 #ifdef FEAT_MBYTE
|
|
|
281 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
282 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
283 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
284 #endif
|
|
344
|
285 } salitem_T;
|
|
|
286
|
|
375
|
287 #ifdef FEAT_MBYTE
|
|
|
288 typedef int salfirst_T;
|
|
|
289 #else
|
|
|
290 typedef short salfirst_T;
|
|
|
291 #endif
|
|
|
292
|
|
236
|
293 /*
|
|
243
|
294 * Structure used to store words and other info for one language, loaded from
|
|
|
295 * a .spl file.
|
|
300
|
296 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
297 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
298 *
|
|
|
299 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
300 * the number of possible bytes, sorted on byte value:
|
|
|
301 * <len> <byte1> <byte2> ...
|
|
|
302 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
303 * byte in "byts".
|
|
|
304 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
|
305 * the flags and region for the word. There may be several zeros in sequence
|
|
|
306 * for alternative flag/region combinations.
|
|
236
|
307 */
|
|
|
308 typedef struct slang_S slang_T;
|
|
|
309 struct slang_S
|
|
|
310 {
|
|
|
311 slang_T *sl_next; /* next language */
|
|
|
312 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
313 char_u *sl_fname; /* name of .spl file */
|
|
323
|
314 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
315
|
|
300
|
316 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
317 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
318 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
319 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
320 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
321 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
322
|
|
236
|
323 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
324
|
|
376
|
325 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
326
|
|
339
|
327 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
328 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
329
|
|
323
|
330 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
331 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
332 there is none */
|
|
344
|
333 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
334 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
335 there is none */
|
|
375
|
336 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
337 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
338 * "sl_sal" is a list of wide char lists. */
|
|
323
|
339 int sl_followup; /* SAL followup */
|
|
|
340 int sl_collapse; /* SAL collapse_result */
|
|
|
341 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
342 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
343 #ifdef FEAT_MBYTE
|
|
|
344 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
345 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
346 #else
|
|
|
347 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
348 #endif
|
|
236
|
349 };
|
|
|
350
|
|
243
|
351 /* First language that is loaded, start of the linked list of loaded
|
|
|
352 * languages. */
|
|
236
|
353 static slang_T *first_lang = NULL;
|
|
|
354
|
|
323
|
355 /* Flags used in .spl file for soundsalike flags. */
|
|
|
356 #define SAL_F0LLOWUP 1
|
|
|
357 #define SAL_COLLAPSE 2
|
|
|
358 #define SAL_REM_ACCENTS 4
|
|
375
|
359 #define SAL_SOFO 8 /* SOFOFROM and SOFOTO instead of SAL */
|
|
323
|
360
|
|
236
|
361 /*
|
|
|
362 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
363 */
|
|
|
364 typedef struct langp_S
|
|
|
365 {
|
|
|
366 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
367 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
368 } langp_T;
|
|
|
369
|
|
|
370 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
371
|
|
307
|
372 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
373
|
|
|
374 /* Result values. Lower number is accepted over higher one. */
|
|
|
375 #define SP_BANNED -1
|
|
236
|
376 #define SP_OK 0
|
|
307
|
377 #define SP_RARE 1
|
|
|
378 #define SP_LOCAL 2
|
|
|
379 #define SP_BAD 3
|
|
236
|
380
|
|
366
|
381 #define VIMSPELLMAGIC "VIMspell08" /* string at start of Vim spell file */
|
|
236
|
382 #define VIMSPELLMAGICL 10
|
|
|
383
|
|
381
|
384 /* file used for "zG" and "zW" */
|
|
385
|
385 static char_u *int_wordlist = NULL;
|
|
381
|
386
|
|
236
|
387 /*
|
|
323
|
388 * Information used when looking for suggestions.
|
|
|
389 */
|
|
|
390 typedef struct suginfo_S
|
|
|
391 {
|
|
|
392 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
393 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
394 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
395 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
396 char_u *su_badptr; /* start of bad word in line */
|
|
|
397 int su_badlen; /* length of detected bad word in line */
|
|
346
|
398 int su_badflags; /* caps flags for bad word */
|
|
323
|
399 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
400 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
401 hashtab_T su_banned; /* table with banned words */
|
|
|
402 } suginfo_T;
|
|
|
403
|
|
|
404 /* One word suggestion. Used in "si_ga". */
|
|
|
405 typedef struct suggest_S
|
|
|
406 {
|
|
|
407 char_u *st_word; /* suggested word, allocated string */
|
|
|
408 int st_orglen; /* length of replaced text */
|
|
|
409 int st_score; /* lower is better */
|
|
344
|
410 int st_altscore; /* used when st_score compares equal */
|
|
|
411 int st_salscore; /* st_score is for soundalike */
|
|
324
|
412 int st_had_bonus; /* bonus already included in score */
|
|
323
|
413 } suggest_T;
|
|
|
414
|
|
344
|
415 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
416
|
|
324
|
417 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
418 * called the score may change, thus we need to keep more than what is
|
|
|
419 * displayed. */
|
|
351
|
420 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
421
|
|
|
422 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
423 * of suggestions that are not going to be displayed. */
|
|
344
|
424 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
425
|
|
|
426 /* score for various changes */
|
|
344
|
427 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
428 #define SCORE_ICASE 52 /* slightly different case */
|
|
|
429 #define SCORE_REGION 70 /* word is for different region */
|
|
|
430 #define SCORE_RARE 180 /* rare word */
|
|
|
431 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
432 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
433 #define SCORE_REP 87 /* REP replacement */
|
|
|
434 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
435 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
324
|
436 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
437 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
324
|
438 #define SCORE_INS 96 /* insert a character */
|
|
358
|
439 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
366
|
440 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
441
|
|
372
|
442 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
443 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
444 * 350 allows for about three changes. */
|
|
344
|
445
|
|
|
446 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
447 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
448
|
|
|
449 /*
|
|
236
|
450 * Structure to store info for word matching.
|
|
|
451 */
|
|
|
452 typedef struct matchinf_S
|
|
|
453 {
|
|
|
454 langp_T *mi_lp; /* info for language and region */
|
|
243
|
455
|
|
|
456 /* pointers to original text to be checked */
|
|
236
|
457 char_u *mi_word; /* start of word being checked */
|
|
339
|
458 char_u *mi_end; /* end of matching word so far */
|
|
243
|
459 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
460 char_u *mi_cend; /* char after what was used for
|
|
|
461 mi_capflags */
|
|
243
|
462
|
|
|
463 /* case-folded text */
|
|
|
464 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
465 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
466
|
|
339
|
467 /* for when checking word after a prefix */
|
|
|
468 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
|
469 prefixID/condition */
|
|
|
470 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
471 int mi_prefixlen; /* byte length of prefix */
|
|
|
472
|
|
243
|
473 /* others */
|
|
236
|
474 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
475 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
476 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
477 } matchinf_T;
|
|
|
478
|
|
307
|
479 /*
|
|
|
480 * The tables used for recognizing word characters according to spelling.
|
|
|
481 * These are only used for the first 256 characters of 'encoding'.
|
|
|
482 */
|
|
|
483 typedef struct spelltab_S
|
|
|
484 {
|
|
|
485 char_u st_isw[256]; /* flags: is word char */
|
|
|
486 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
487 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
488 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
489 } spelltab_T;
|
|
|
490
|
|
|
491 static spelltab_T spelltab;
|
|
|
492 static int did_set_spelltab;
|
|
|
493
|
|
324
|
494 #define CF_WORD 0x01
|
|
|
495 #define CF_UPPER 0x02
|
|
307
|
496
|
|
|
497 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
498 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
499 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
500 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
501 #ifdef FEAT_MBYTE
|
|
|
502 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
503 #endif
|
|
339
|
504 static void write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
505
|
|
|
506 /*
|
|
358
|
507 * Return TRUE if "p" points to a word character. Like spell_iswordp() but
|
|
|
508 * without the special handling of a single quote.
|
|
307
|
509 * Checking for a word character is done very often, avoid the function call
|
|
|
510 * overhead.
|
|
|
511 */
|
|
|
512 #ifdef FEAT_MBYTE
|
|
|
513 # define SPELL_ISWORDP(p) ((has_mbyte && MB_BYTE2LEN(*(p)) > 1) \
|
|
|
514 ? (mb_get_class(p) >= 2) : spelltab.st_isw[*(p)])
|
|
|
515 #else
|
|
|
516 # define SPELL_ISWORDP(p) (spelltab.st_isw[*(p)])
|
|
|
517 #endif
|
|
|
518
|
|
323
|
519 /*
|
|
344
|
520 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
521 */
|
|
|
522 typedef enum
|
|
|
523 {
|
|
344
|
524 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
525 * ends); if badword ends there is a match, otherwise
|
|
|
526 * try splitting word. */
|
|
375
|
527 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
528 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
529 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
530 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
531 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
532 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
533 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
534 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
535 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
536 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
537 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
538 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
539 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
540 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
541 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
542 STATE_FINAL /* End of this node. */
|
|
|
543 } state_T;
|
|
|
544
|
|
|
545 /*
|
|
346
|
546 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
547 */
|
|
|
548 typedef struct trystate_S
|
|
|
549 {
|
|
330
|
550 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
551 int ts_score; /* score */
|
|
344
|
552 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
553 short ts_curi; /* index in list of child nodes */
|
|
|
554 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
555 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
556 char_u ts_twordlen; /* valid length of tword[] */
|
|
375
|
557 char_u ts_prefixdepth; /* stack depth for end of prefix or PREFIXTREE
|
|
|
558 * or NOPREFIX */
|
|
330
|
559 #ifdef FEAT_MBYTE
|
|
|
560 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
561 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
562 char_u ts_isdiff; /* DIFF_ values */
|
|
|
563 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
564 #endif
|
|
323
|
565 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
330
|
566 char_u ts_save_splitoff; /* su_splitoff saved here */
|
|
346
|
567 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
568 } trystate_T;
|
|
|
569
|
|
330
|
570 /* values for ts_isdiff */
|
|
|
571 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
572 #define DIFF_YES 1 /* different byte found */
|
|
|
573 #define DIFF_INSERT 2 /* inserting character */
|
|
|
574
|
|
375
|
575 /* special values ts_prefixdepth */
|
|
|
576 #define PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
577 #define NOPREFIX 0xff /* not using prefixes */
|
|
|
578
|
|
339
|
579 /* mode values for find_word */
|
|
|
580 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
581 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
582 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
583
|
|
236
|
584 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
585 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
586 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
587 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
351
|
588 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int prefid, char_u *word, slang_T *slang));
|
|
339
|
589 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
590 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
591 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
592 static int no_spell_checking __ARGS((void));
|
|
307
|
593 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
594 static char_u *spell_enc __ARGS((void));
|
|
385
|
595 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
596 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
597 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
598 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
381
|
599 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
600 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
601 #ifdef FEAT_MBYTE
|
|
|
602 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
603 #endif
|
|
339
|
604 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
605 static void clear_midword __ARGS((buf_T *buf));
|
|
|
606 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
607 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
608 static int captype __ARGS((char_u *word, char_u *end));
|
|
323
|
609 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
389
|
610 static int set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
611 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
612 static void write_spell_chartab __ARGS((FILE *fd));
|
|
|
613 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
388
|
614 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
615 #ifdef FEAT_EVAL
|
|
|
616 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
617 #endif
|
|
|
618 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
619 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
620 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
621 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
622 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
623 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
624 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
625 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
|
626 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
627 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
628 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
629 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
630 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
631 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
632 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
633 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
634 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
635 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
636 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
637 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
638 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
639 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
640 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
641 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
642 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
643 #ifdef FEAT_MBYTE
|
|
375
|
644 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
645 #endif
|
|
344
|
646 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
647 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
648 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
649 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
650
|
|
324
|
651 /*
|
|
|
652 * Use our own character-case definitions, because the current locale may
|
|
|
653 * differ from what the .spl file uses.
|
|
|
654 * These must not be called with negative number!
|
|
|
655 */
|
|
|
656 #ifndef FEAT_MBYTE
|
|
|
657 /* Non-multi-byte implementation. */
|
|
|
658 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
659 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
660 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
661 #else
|
|
|
662 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
663 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
664 * the "w" library function for characters above 255 if available. */
|
|
|
665 # ifdef HAVE_TOWLOWER
|
|
|
666 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
667 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
668 # else
|
|
|
669 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
670 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
671 # endif
|
|
|
672
|
|
|
673 # ifdef HAVE_TOWUPPER
|
|
|
674 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
675 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
676 # else
|
|
|
677 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
678 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
679 # endif
|
|
|
680
|
|
|
681 # ifdef HAVE_ISWUPPER
|
|
|
682 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
683 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
684 # else
|
|
|
685 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
686 : (c) < 256 ? spelltab.st_isu[c] : (c))
|
|
|
687 # endif
|
|
|
688 #endif
|
|
|
689
|
|
307
|
690
|
|
|
691 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
692 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
236
|
693
|
|
|
694 /*
|
|
|
695 * Main spell-checking function.
|
|
300
|
696 * "ptr" points to a character that could be the start of a word.
|
|
236
|
697 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
698 * or when it's OK it remains unchanged.
|
|
|
699 * This must only be called when 'spelllang' is not empty.
|
|
323
|
700 *
|
|
385
|
701 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
702 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
703 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
704 * worry.
|
|
323
|
705 *
|
|
236
|
706 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
707 * caller can skip over the word.
|
|
|
708 */
|
|
|
709 int
|
|
385
|
710 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
711 win_T *wp; /* current window */
|
|
|
712 char_u *ptr;
|
|
|
713 int *attrp;
|
|
385
|
714 int *capcol; /* column to check for Capital */
|
|
236
|
715 {
|
|
|
716 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
717 be passed to functions quickly. */
|
|
344
|
718 int nrlen = 0; /* found a number first */
|
|
385
|
719 int c;
|
|
236
|
720
|
|
307
|
721 /* A word never starts at a space or a control character. Return quickly
|
|
|
722 * then, skipping over the character. */
|
|
|
723 if (*ptr <= ' ')
|
|
|
724 return 1;
|
|
236
|
725
|
|
344
|
726 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
727 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
728 * "3GPP". */
|
|
300
|
729 if (*ptr >= '0' && *ptr <= '9')
|
|
|
730 {
|
|
316
|
731 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
732 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
733 else
|
|
|
734 {
|
|
344
|
735 mi.mi_end = skipdigits(ptr);
|
|
|
736 nrlen = mi.mi_end - ptr;
|
|
300
|
737 }
|
|
376
|
738 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
739 return (int)(mi.mi_end - ptr);
|
|
346
|
740
|
|
|
741 /* Try including the digits in the word. */
|
|
|
742 mi.mi_fend = ptr + nrlen;
|
|
|
743 }
|
|
|
744 else
|
|
|
745 mi.mi_fend = ptr;
|
|
|
746
|
|
|
747 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
748 mi.mi_word = ptr;
|
|
376
|
749 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
750 {
|
|
|
751 do
|
|
|
752 {
|
|
307
|
753 mb_ptr_adv(mi.mi_fend);
|
|
376
|
754 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
755
|
|
|
756 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
757 {
|
|
|
758 /* Check word starting with capital letter. */
|
|
|
759 #ifdef FEAT_MBYTE
|
|
|
760 c = mb_ptr2char(ptr);
|
|
|
761 #else
|
|
|
762 c = *ptr;
|
|
|
763 #endif
|
|
|
764 if (!SPELL_ISUPPER(c))
|
|
|
765 {
|
|
|
766 *attrp = highlight_attr[HLF_SPC];
|
|
|
767 return (int)(mi.mi_fend - ptr);
|
|
|
768 }
|
|
|
769 }
|
|
|
770 }
|
|
|
771 if (capcol != NULL)
|
|
|
772 *capcol = -1;
|
|
344
|
773
|
|
|
774 /* We always use the characters up to the next non-word character,
|
|
|
775 * also for bad words. */
|
|
|
776 mi.mi_end = mi.mi_fend;
|
|
|
777
|
|
|
778 /* Check caps type later. */
|
|
|
779 mi.mi_capflags = 0;
|
|
|
780 mi.mi_cend = NULL;
|
|
376
|
781 mi.mi_buf = wp->w_buffer;
|
|
344
|
782
|
|
|
783 /* Include one non-word character so that we can check for the
|
|
|
784 * word end. */
|
|
|
785 if (*mi.mi_fend != NUL)
|
|
|
786 mb_ptr_adv(mi.mi_fend);
|
|
|
787
|
|
|
788 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
789 MAXWLEN + 1);
|
|
|
790 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
791
|
|
|
792 /* The word is bad unless we recognize it. */
|
|
|
793 mi.mi_result = SP_BAD;
|
|
|
794
|
|
|
795 /*
|
|
|
796 * Loop over the languages specified in 'spelllang'.
|
|
|
797 * We check them all, because a matching word may be longer than an
|
|
|
798 * already found matching word.
|
|
|
799 */
|
|
|
800 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
801 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
802 {
|
|
|
803 /* Check for a matching word in case-folded words. */
|
|
|
804 find_word(&mi, FIND_FOLDWORD);
|
|
|
805
|
|
|
806 /* Check for a matching word in keep-case words. */
|
|
|
807 find_word(&mi, FIND_KEEPWORD);
|
|
|
808
|
|
|
809 /* Check for matching prefixes. */
|
|
|
810 find_prefix(&mi);
|
|
|
811 }
|
|
|
812
|
|
|
813 if (mi.mi_result != SP_OK)
|
|
|
814 {
|
|
346
|
815 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
816 * rare and local words, e.g., "3GPP". */
|
|
344
|
817 if (nrlen > 0)
|
|
346
|
818 {
|
|
|
819 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
820 return nrlen;
|
|
|
821 }
|
|
344
|
822
|
|
|
823 /* When we are at a non-word character there is no error, just
|
|
|
824 * skip over the character (try looking for a word after it). */
|
|
346
|
825 else if (!SPELL_ISWORDP(ptr))
|
|
243
|
826 {
|
|
385
|
827 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
828 {
|
|
|
829 regmatch_T regmatch;
|
|
|
830
|
|
|
831 /* Check for end of sentence. */
|
|
|
832 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
833 regmatch.rm_ic = FALSE;
|
|
|
834 if (vim_regexec(®match, ptr, 0))
|
|
|
835 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
836 }
|
|
|
837
|
|
344
|
838 #ifdef FEAT_MBYTE
|
|
|
839 if (has_mbyte)
|
|
|
840 return mb_ptr2len_check(ptr);
|
|
|
841 #endif
|
|
|
842 return 1;
|
|
300
|
843 }
|
|
243
|
844
|
|
344
|
845 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
846 *attrp = highlight_attr[HLF_SPB];
|
|
|
847 else if (mi.mi_result == SP_RARE)
|
|
|
848 *attrp = highlight_attr[HLF_SPR];
|
|
|
849 else
|
|
|
850 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
851 }
|
|
|
852
|
|
300
|
853 return (int)(mi.mi_end - ptr);
|
|
236
|
854 }
|
|
|
855
|
|
|
856 /*
|
|
300
|
857 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
858 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
859 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
860 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
861 * tree.
|
|
300
|
862 *
|
|
|
863 * For a match mip->mi_result is updated.
|
|
243
|
864 */
|
|
|
865 static void
|
|
339
|
866 find_word(mip, mode)
|
|
243
|
867 matchinf_T *mip;
|
|
339
|
868 int mode;
|
|
243
|
869 {
|
|
324
|
870 idx_T arridx = 0;
|
|
300
|
871 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
872 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
873 int endidxcnt = 0;
|
|
|
874 int len;
|
|
|
875 int wlen = 0;
|
|
|
876 int flen;
|
|
|
877 int c;
|
|
|
878 char_u *ptr;
|
|
324
|
879 idx_T lo, hi, m;
|
|
243
|
880 #ifdef FEAT_MBYTE
|
|
300
|
881 char_u *s;
|
|
339
|
882 char_u *p;
|
|
307
|
883 #endif
|
|
|
884 int res = SP_BAD;
|
|
300
|
885 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
886 unsigned flags;
|
|
|
887 char_u *byts;
|
|
324
|
888 idx_T *idxs;
|
|
339
|
889 int prefid;
|
|
|
890
|
|
|
891 if (mode == FIND_KEEPWORD)
|
|
236
|
892 {
|
|
300
|
893 /* Check for word with matching case in keep-case tree. */
|
|
|
894 ptr = mip->mi_word;
|
|
|
895 flen = 9999; /* no case folding, always enough bytes */
|
|
|
896 byts = slang->sl_kbyts;
|
|
|
897 idxs = slang->sl_kidxs;
|
|
236
|
898 }
|
|
|
899 else
|
|
|
900 {
|
|
300
|
901 /* Check for case-folded in case-folded tree. */
|
|
|
902 ptr = mip->mi_fword;
|
|
|
903 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
904 byts = slang->sl_fbyts;
|
|
|
905 idxs = slang->sl_fidxs;
|
|
339
|
906
|
|
|
907 if (mode == FIND_PREFIX)
|
|
|
908 {
|
|
|
909 /* Skip over the prefix. */
|
|
|
910 wlen = mip->mi_prefixlen;
|
|
|
911 flen -= mip->mi_prefixlen;
|
|
|
912 }
|
|
243
|
913 }
|
|
|
914
|
|
300
|
915 if (byts == NULL)
|
|
|
916 return; /* array is empty */
|
|
236
|
917
|
|
|
918 /*
|
|
307
|
919 * Repeat advancing in the tree until:
|
|
|
920 * - there is a byte that doesn't match,
|
|
|
921 * - we reach the end of the tree,
|
|
|
922 * - or we reach the end of the line.
|
|
236
|
923 */
|
|
300
|
924 for (;;)
|
|
236
|
925 {
|
|
346
|
926 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
927 flen = fold_more(mip);
|
|
300
|
928
|
|
|
929 len = byts[arridx++];
|
|
|
930
|
|
|
931 /* If the first possible byte is a zero the word could end here.
|
|
|
932 * Remember this index, we first check for the longest word. */
|
|
|
933 if (byts[arridx] == 0)
|
|
|
934 {
|
|
307
|
935 if (endidxcnt == MAXWLEN)
|
|
|
936 {
|
|
|
937 /* Must be a corrupted spell file. */
|
|
|
938 EMSG(_(e_format));
|
|
|
939 return;
|
|
|
940 }
|
|
300
|
941 endlen[endidxcnt] = wlen;
|
|
|
942 endidx[endidxcnt++] = arridx++;
|
|
|
943 --len;
|
|
|
944
|
|
|
945 /* Skip over the zeros, there can be several flag/region
|
|
|
946 * combinations. */
|
|
|
947 while (len > 0 && byts[arridx] == 0)
|
|
|
948 {
|
|
|
949 ++arridx;
|
|
|
950 --len;
|
|
|
951 }
|
|
|
952 if (len == 0)
|
|
|
953 break; /* no children, word must end here */
|
|
|
954 }
|
|
|
955
|
|
|
956 /* Stop looking at end of the line. */
|
|
|
957 if (ptr[wlen] == NUL)
|
|
|
958 break;
|
|
|
959
|
|
|
960 /* Perform a binary search in the list of accepted bytes. */
|
|
|
961 c = ptr[wlen];
|
|
346
|
962 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
963 c = ' ';
|
|
300
|
964 lo = arridx;
|
|
|
965 hi = arridx + len - 1;
|
|
|
966 while (lo < hi)
|
|
|
967 {
|
|
|
968 m = (lo + hi) / 2;
|
|
|
969 if (byts[m] > c)
|
|
|
970 hi = m - 1;
|
|
|
971 else if (byts[m] < c)
|
|
|
972 lo = m + 1;
|
|
|
973 else
|
|
|
974 {
|
|
|
975 lo = hi = m;
|
|
|
976 break;
|
|
236
|
977 }
|
|
|
978 }
|
|
300
|
979
|
|
|
980 /* Stop if there is no matching byte. */
|
|
|
981 if (hi < lo || byts[lo] != c)
|
|
|
982 break;
|
|
|
983
|
|
|
984 /* Continue at the child (if there is one). */
|
|
|
985 arridx = idxs[lo];
|
|
|
986 ++wlen;
|
|
|
987 --flen;
|
|
346
|
988
|
|
|
989 /* One space in the good word may stand for several spaces in the
|
|
|
990 * checked word. */
|
|
|
991 if (c == ' ')
|
|
|
992 {
|
|
|
993 for (;;)
|
|
|
994 {
|
|
|
995 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
996 flen = fold_more(mip);
|
|
|
997 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
998 break;
|
|
|
999 ++wlen;
|
|
|
1000 --flen;
|
|
|
1001 }
|
|
|
1002 }
|
|
236
|
1003 }
|
|
|
1004
|
|
300
|
1005 /*
|
|
|
1006 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1007 * first.
|
|
|
1008 */
|
|
|
1009 while (endidxcnt > 0)
|
|
|
1010 {
|
|
|
1011 --endidxcnt;
|
|
|
1012 arridx = endidx[endidxcnt];
|
|
|
1013 wlen = endlen[endidxcnt];
|
|
236
|
1014
|
|
300
|
1015 #ifdef FEAT_MBYTE
|
|
|
1016 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1017 continue; /* not at first byte of character */
|
|
|
1018 #endif
|
|
376
|
1019 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
300
|
1020 continue; /* next char is a word character */
|
|
|
1021
|
|
|
1022 #ifdef FEAT_MBYTE
|
|
339
|
1023 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1024 {
|
|
|
1025 /* Compute byte length in original word, length may change
|
|
339
|
1026 * when folding case. This can be slow, take a shortcut when the
|
|
|
1027 * case-folded word is equal to the keep-case word. */
|
|
300
|
1028 p = mip->mi_word;
|
|
339
|
1029 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1030 {
|
|
|
1031 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1032 mb_ptr_adv(p);
|
|
|
1033 wlen = p - mip->mi_word;
|
|
|
1034 }
|
|
300
|
1035 }
|
|
|
1036 #endif
|
|
236
|
1037
|
|
339
|
1038 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1039 * prefix ID.
|
|
|
1040 * Repeat this if there are more flags/region alternatives until there
|
|
|
1041 * is a match. */
|
|
|
1042 res = SP_BAD;
|
|
|
1043 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1044 --len, ++arridx)
|
|
300
|
1045 {
|
|
|
1046 flags = idxs[arridx];
|
|
324
|
1047
|
|
339
|
1048 /* For the fold-case tree check that the case of the checked word
|
|
|
1049 * matches with what the word in the tree requires.
|
|
|
1050 * For keep-case tree the case is always right. For prefixes we
|
|
|
1051 * don't bother to check. */
|
|
|
1052 if (mode == FIND_FOLDWORD)
|
|
300
|
1053 {
|
|
|
1054 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1055 {
|
|
323
|
1056 /* mi_capflags was set for a different word length, need
|
|
|
1057 * to do it again. */
|
|
300
|
1058 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1059 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1060 }
|
|
|
1061
|
|
346
|
1062 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1063 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1064 continue;
|
|
300
|
1065 }
|
|
236
|
1066
|
|
339
|
1067 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1068 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1069 * mip->mi_prefarridx that find_prefix() filled. */
|
|
339
|
1070 if (mode == FIND_PREFIX)
|
|
300
|
1071 {
|
|
339
|
1072 /* The prefix ID is stored two bytes above the flags. */
|
|
|
1073 prefid = (unsigned)flags >> 16;
|
|
366
|
1074 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
351
|
1075 prefid, mip->mi_fword + mip->mi_prefixlen,
|
|
366
|
1076 slang);
|
|
|
1077 if (c == 0)
|
|
339
|
1078 continue;
|
|
366
|
1079
|
|
|
1080 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1081 if (c & WF_RAREPFX)
|
|
|
1082 flags |= WF_RARE;
|
|
339
|
1083 }
|
|
|
1084
|
|
|
1085 if (flags & WF_BANNED)
|
|
|
1086 res = SP_BANNED;
|
|
|
1087 else if (flags & WF_REGION)
|
|
|
1088 {
|
|
|
1089 /* Check region. */
|
|
|
1090 if ((mip->mi_lp->lp_region & (flags >> 8)) != 0)
|
|
300
|
1091 res = SP_OK;
|
|
339
|
1092 else
|
|
|
1093 res = SP_LOCAL;
|
|
300
|
1094 }
|
|
339
|
1095 else if (flags & WF_RARE)
|
|
|
1096 res = SP_RARE;
|
|
307
|
1097 else
|
|
339
|
1098 res = SP_OK;
|
|
|
1099
|
|
|
1100 /* Always use the longest match and the best result. */
|
|
|
1101 if (mip->mi_result > res)
|
|
|
1102 {
|
|
|
1103 mip->mi_result = res;
|
|
|
1104 mip->mi_end = mip->mi_word + wlen;
|
|
|
1105 }
|
|
351
|
1106 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1107 mip->mi_end = mip->mi_word + wlen;
|
|
|
1108
|
|
|
1109 if (res == SP_OK)
|
|
|
1110 break;
|
|
300
|
1111 }
|
|
|
1112
|
|
307
|
1113 if (res == SP_OK)
|
|
300
|
1114 break;
|
|
|
1115 }
|
|
236
|
1116 }
|
|
|
1117
|
|
323
|
1118 /*
|
|
366
|
1119 * Return non-zero if the prefix indicated by "mip->mi_prefarridx" matches
|
|
|
1120 * with the prefix ID "prefid" for the word "word".
|
|
|
1121 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1122 */
|
|
|
1123 static int
|
|
|
1124 valid_word_prefix(totprefcnt, arridx, prefid, word, slang)
|
|
|
1125 int totprefcnt; /* nr of prefix IDs */
|
|
|
1126 int arridx; /* idx in sl_pidxs[] */
|
|
|
1127 int prefid;
|
|
|
1128 char_u *word;
|
|
|
1129 slang_T *slang;
|
|
|
1130 {
|
|
|
1131 int prefcnt;
|
|
|
1132 int pidx;
|
|
|
1133 regprog_T *rp;
|
|
|
1134 regmatch_T regmatch;
|
|
|
1135
|
|
|
1136 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1137 {
|
|
|
1138 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1139
|
|
|
1140 /* Check the prefix ID. */
|
|
|
1141 if (prefid != (pidx & 0xff))
|
|
|
1142 continue;
|
|
|
1143
|
|
|
1144 /* Check the condition, if there is one. The condition index is
|
|
366
|
1145 * stored in the two bytes above the prefix ID byte. */
|
|
|
1146 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1147 if (rp != NULL)
|
|
|
1148 {
|
|
|
1149 regmatch.regprog = rp;
|
|
|
1150 regmatch.rm_ic = FALSE;
|
|
|
1151 if (!vim_regexec(®match, word, 0))
|
|
|
1152 continue;
|
|
|
1153 }
|
|
|
1154
|
|
366
|
1155 /* It's a match! Return the WF_RAREPFX flag. */
|
|
|
1156 return pidx;
|
|
|
1157 }
|
|
|
1158 return 0;
|
|
351
|
1159 }
|
|
|
1160
|
|
|
1161 /*
|
|
339
|
1162 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1163 * If it does, then check the following word.
|
|
|
1164 *
|
|
|
1165 * For a match mip->mi_result is updated.
|
|
|
1166 */
|
|
|
1167 static void
|
|
|
1168 find_prefix(mip)
|
|
|
1169 matchinf_T *mip;
|
|
|
1170 {
|
|
|
1171 idx_T arridx = 0;
|
|
|
1172 int len;
|
|
|
1173 int wlen = 0;
|
|
|
1174 int flen;
|
|
|
1175 int c;
|
|
|
1176 char_u *ptr;
|
|
|
1177 idx_T lo, hi, m;
|
|
|
1178 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1179 char_u *byts;
|
|
|
1180 idx_T *idxs;
|
|
|
1181
|
|
375
|
1182 byts = slang->sl_pbyts;
|
|
|
1183 if (byts == NULL)
|
|
|
1184 return; /* array is empty */
|
|
|
1185
|
|
339
|
1186 /* We use the case-folded word here, since prefixes are always
|
|
|
1187 * case-folded. */
|
|
|
1188 ptr = mip->mi_fword;
|
|
|
1189 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1190 idxs = slang->sl_pidxs;
|
|
|
1191
|
|
|
1192 /*
|
|
|
1193 * Repeat advancing in the tree until:
|
|
|
1194 * - there is a byte that doesn't match,
|
|
|
1195 * - we reach the end of the tree,
|
|
|
1196 * - or we reach the end of the line.
|
|
|
1197 */
|
|
|
1198 for (;;)
|
|
|
1199 {
|
|
|
1200 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1201 flen = fold_more(mip);
|
|
|
1202
|
|
|
1203 len = byts[arridx++];
|
|
|
1204
|
|
|
1205 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1206 * Check if the following word matches and supports the prefix. */
|
|
|
1207 if (byts[arridx] == 0)
|
|
|
1208 {
|
|
|
1209 /* There can be several prefixes with different conditions. We
|
|
|
1210 * try them all, since we don't know which one will give the
|
|
|
1211 * longest match. The word is the same each time, pass the list
|
|
|
1212 * of possible prefixes to find_word(). */
|
|
|
1213 mip->mi_prefarridx = arridx;
|
|
|
1214 mip->mi_prefcnt = len;
|
|
|
1215 while (len > 0 && byts[arridx] == 0)
|
|
|
1216 {
|
|
|
1217 ++arridx;
|
|
|
1218 --len;
|
|
|
1219 }
|
|
|
1220 mip->mi_prefcnt -= len;
|
|
|
1221
|
|
|
1222 /* Find the word that comes after the prefix. */
|
|
|
1223 mip->mi_prefixlen = wlen;
|
|
|
1224 find_word(mip, FIND_PREFIX);
|
|
|
1225
|
|
|
1226
|
|
|
1227 if (len == 0)
|
|
|
1228 break; /* no children, word must end here */
|
|
|
1229 }
|
|
|
1230
|
|
|
1231 /* Stop looking at end of the line. */
|
|
|
1232 if (ptr[wlen] == NUL)
|
|
|
1233 break;
|
|
|
1234
|
|
|
1235 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1236 c = ptr[wlen];
|
|
|
1237 lo = arridx;
|
|
|
1238 hi = arridx + len - 1;
|
|
|
1239 while (lo < hi)
|
|
|
1240 {
|
|
|
1241 m = (lo + hi) / 2;
|
|
|
1242 if (byts[m] > c)
|
|
|
1243 hi = m - 1;
|
|
|
1244 else if (byts[m] < c)
|
|
|
1245 lo = m + 1;
|
|
|
1246 else
|
|
|
1247 {
|
|
|
1248 lo = hi = m;
|
|
|
1249 break;
|
|
|
1250 }
|
|
|
1251 }
|
|
|
1252
|
|
|
1253 /* Stop if there is no matching byte. */
|
|
|
1254 if (hi < lo || byts[lo] != c)
|
|
|
1255 break;
|
|
|
1256
|
|
|
1257 /* Continue at the child (if there is one). */
|
|
|
1258 arridx = idxs[lo];
|
|
|
1259 ++wlen;
|
|
|
1260 --flen;
|
|
|
1261 }
|
|
|
1262 }
|
|
|
1263
|
|
|
1264 /*
|
|
|
1265 * Need to fold at least one more character. Do until next non-word character
|
|
|
1266 * for efficiency.
|
|
|
1267 * Return the length of the folded chars in bytes.
|
|
|
1268 */
|
|
|
1269 static int
|
|
|
1270 fold_more(mip)
|
|
|
1271 matchinf_T *mip;
|
|
|
1272 {
|
|
|
1273 int flen;
|
|
|
1274 char_u *p;
|
|
|
1275
|
|
|
1276 p = mip->mi_fend;
|
|
|
1277 do
|
|
|
1278 {
|
|
|
1279 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1280 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1281
|
|
|
1282 /* Include the non-word character so that we can check for the
|
|
|
1283 * word end. */
|
|
|
1284 if (*mip->mi_fend != NUL)
|
|
|
1285 mb_ptr_adv(mip->mi_fend);
|
|
|
1286
|
|
|
1287 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1288 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1289 MAXWLEN - mip->mi_fwordlen);
|
|
|
1290 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1291 mip->mi_fwordlen += flen;
|
|
|
1292 return flen;
|
|
|
1293 }
|
|
|
1294
|
|
|
1295 /*
|
|
323
|
1296 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1297 * case.
|
|
|
1298 */
|
|
|
1299 static int
|
|
389
|
1300 spell_valid_case(wordflags, treeflags)
|
|
|
1301 int wordflags; /* flags for the checked word. */
|
|
323
|
1302 int treeflags; /* flags for the word in the spell tree */
|
|
|
1303 {
|
|
389
|
1304 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1305 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
389
|
1306 && ((treeflags & WF_ONECAP) == 0 || wordflags == WF_ONECAP)));
|
|
323
|
1307 }
|
|
|
1308
|
|
351
|
1309 /*
|
|
|
1310 * Return TRUE if spell checking is not enabled.
|
|
|
1311 */
|
|
|
1312 static int
|
|
|
1313 no_spell_checking()
|
|
|
1314 {
|
|
|
1315 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1316 {
|
|
|
1317 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1318 return TRUE;
|
|
|
1319 }
|
|
|
1320 return FALSE;
|
|
|
1321 }
|
|
300
|
1322
|
|
236
|
1323 /*
|
|
|
1324 * Move to next spell error.
|
|
323
|
1325 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
236
|
1326 * Return OK if found, FAIL otherwise.
|
|
|
1327 */
|
|
|
1328 int
|
|
323
|
1329 spell_move_to(dir, allwords, curline)
|
|
236
|
1330 int dir; /* FORWARD or BACKWARD */
|
|
|
1331 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1332 int curline;
|
|
236
|
1333 {
|
|
249
|
1334 linenr_T lnum;
|
|
|
1335 pos_T found_pos;
|
|
236
|
1336 char_u *line;
|
|
|
1337 char_u *p;
|
|
346
|
1338 char_u *endp;
|
|
|
1339 int attr;
|
|
236
|
1340 int len;
|
|
249
|
1341 int has_syntax = syntax_present(curbuf);
|
|
|
1342 int col;
|
|
|
1343 int can_spell;
|
|
346
|
1344 char_u *buf = NULL;
|
|
|
1345 int buflen = 0;
|
|
|
1346 int skip = 0;
|
|
385
|
1347 int capcol = -1;
|
|
236
|
1348
|
|
351
|
1349 if (no_spell_checking())
|
|
236
|
1350 return FAIL;
|
|
|
1351
|
|
249
|
1352 /*
|
|
|
1353 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1354 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1355 *
|
|
|
1356 * When searching backwards, we continue in the line to find the last
|
|
|
1357 * bad word (in the cursor line: before the cursor).
|
|
346
|
1358 *
|
|
|
1359 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1360 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1361 * though...
|
|
249
|
1362 */
|
|
|
1363 lnum = curwin->w_cursor.lnum;
|
|
|
1364 found_pos.lnum = 0;
|
|
236
|
1365
|
|
|
1366 while (!got_int)
|
|
|
1367 {
|
|
249
|
1368 line = ml_get(lnum);
|
|
346
|
1369
|
|
|
1370 len = STRLEN(line);
|
|
|
1371 if (buflen < len + MAXWLEN + 2)
|
|
|
1372 {
|
|
|
1373 vim_free(buf);
|
|
|
1374 buflen = len + MAXWLEN + 2;
|
|
|
1375 buf = alloc(buflen);
|
|
|
1376 if (buf == NULL)
|
|
|
1377 break;
|
|
|
1378 }
|
|
|
1379
|
|
385
|
1380 /* In first line check first word for Capital. */
|
|
|
1381 if (lnum == 1)
|
|
|
1382 capcol = 0;
|
|
|
1383
|
|
|
1384 /* For checking first word with a capital skip white space. */
|
|
|
1385 if (capcol == 0)
|
|
|
1386 capcol = skipwhite(line) - line;
|
|
|
1387
|
|
346
|
1388 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1389 * possible. */
|
|
|
1390 STRCPY(buf, line);
|
|
|
1391 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1392 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1393
|
|
|
1394 p = buf + skip;
|
|
|
1395 endp = buf + len;
|
|
|
1396 while (p < endp)
|
|
236
|
1397 {
|
|
300
|
1398 /* When searching backward don't search after the cursor. */
|
|
|
1399 if (dir == BACKWARD
|
|
|
1400 && lnum == curwin->w_cursor.lnum
|
|
346
|
1401 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1402 break;
|
|
249
|
1403
|
|
300
|
1404 /* start of word */
|
|
346
|
1405 attr = 0;
|
|
385
|
1406 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1407
|
|
300
|
1408 if (attr != 0)
|
|
|
1409 {
|
|
|
1410 /* We found a bad word. Check the attribute. */
|
|
|
1411 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1412 {
|
|
300
|
1413 /* When searching forward only accept a bad word after
|
|
|
1414 * the cursor. */
|
|
|
1415 if (dir == BACKWARD
|
|
|
1416 || lnum > curwin->w_cursor.lnum
|
|
|
1417 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1418 && (colnr_T)(curline ? p - buf + len
|
|
|
1419 : p - buf)
|
|
300
|
1420 > curwin->w_cursor.col))
|
|
236
|
1421 {
|
|
300
|
1422 if (has_syntax)
|
|
249
|
1423 {
|
|
346
|
1424 col = p - buf;
|
|
300
|
1425 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1426 FALSE, &can_spell);
|
|
|
1427 }
|
|
|
1428 else
|
|
|
1429 can_spell = TRUE;
|
|
249
|
1430
|
|
300
|
1431 if (can_spell)
|
|
|
1432 {
|
|
|
1433 found_pos.lnum = lnum;
|
|
346
|
1434 found_pos.col = p - buf;
|
|
249
|
1435 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1436 found_pos.coladd = 0;
|
|
249
|
1437 #endif
|
|
300
|
1438 if (dir == FORWARD)
|
|
|
1439 {
|
|
|
1440 /* No need to search further. */
|
|
|
1441 curwin->w_cursor = found_pos;
|
|
346
|
1442 vim_free(buf);
|
|
300
|
1443 return OK;
|
|
249
|
1444 }
|
|
|
1445 }
|
|
236
|
1446 }
|
|
|
1447 }
|
|
|
1448 }
|
|
|
1449
|
|
300
|
1450 /* advance to character after the word */
|
|
|
1451 p += len;
|
|
385
|
1452 capcol -= len;
|
|
236
|
1453 }
|
|
|
1454
|
|
323
|
1455 if (curline)
|
|
346
|
1456 break; /* only check cursor line */
|
|
323
|
1457
|
|
236
|
1458 /* Advance to next line. */
|
|
249
|
1459 if (dir == BACKWARD)
|
|
|
1460 {
|
|
|
1461 if (found_pos.lnum != 0)
|
|
|
1462 {
|
|
|
1463 /* Use the last match in the line. */
|
|
|
1464 curwin->w_cursor = found_pos;
|
|
346
|
1465 vim_free(buf);
|
|
249
|
1466 return OK;
|
|
|
1467 }
|
|
|
1468 if (lnum == 1)
|
|
346
|
1469 break;
|
|
249
|
1470 --lnum;
|
|
385
|
1471 capcol = -1;
|
|
249
|
1472 }
|
|
|
1473 else
|
|
|
1474 {
|
|
|
1475 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1476 break;
|
|
249
|
1477 ++lnum;
|
|
346
|
1478
|
|
|
1479 /* Skip the characters at the start of the next line that were
|
|
|
1480 * included in a match crossing line boundaries. */
|
|
|
1481 if (attr == 0)
|
|
|
1482 skip = p - endp;
|
|
|
1483 else
|
|
|
1484 skip = 0;
|
|
385
|
1485
|
|
|
1486 /* Capscol skips over the inserted space. */
|
|
|
1487 --capcol;
|
|
|
1488
|
|
|
1489 /* But after empty line check first word in next line */
|
|
|
1490 if (*skipwhite(line) == NUL)
|
|
|
1491 capcol = 0;
|
|
249
|
1492 }
|
|
236
|
1493
|
|
|
1494 line_breakcheck();
|
|
|
1495 }
|
|
|
1496
|
|
346
|
1497 vim_free(buf);
|
|
|
1498 return FAIL;
|
|
|
1499 }
|
|
|
1500
|
|
|
1501 /*
|
|
|
1502 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1503 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1504 */
|
|
|
1505 void
|
|
|
1506 spell_cat_line(buf, line, maxlen)
|
|
|
1507 char_u *buf;
|
|
|
1508 char_u *line;
|
|
|
1509 int maxlen;
|
|
|
1510 {
|
|
|
1511 char_u *p;
|
|
|
1512 int n;
|
|
|
1513
|
|
|
1514 p = skipwhite(line);
|
|
|
1515 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1516 p = skipwhite(p + 1);
|
|
|
1517
|
|
|
1518 if (*p != NUL)
|
|
|
1519 {
|
|
|
1520 *buf = ' ';
|
|
376
|
1521 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1522 n = p - line;
|
|
|
1523 if (n >= maxlen)
|
|
|
1524 n = maxlen - 1;
|
|
|
1525 vim_memset(buf + 1, ' ', n);
|
|
|
1526 }
|
|
236
|
1527 }
|
|
|
1528
|
|
|
1529 /*
|
|
307
|
1530 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1531 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1532 */
|
|
307
|
1533 static void
|
|
236
|
1534 spell_load_lang(lang)
|
|
|
1535 char_u *lang;
|
|
|
1536 {
|
|
310
|
1537 char_u fname_enc[85];
|
|
236
|
1538 int r;
|
|
307
|
1539 char_u langcp[MAXWLEN + 1];
|
|
|
1540
|
|
310
|
1541 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1542 * It's truncated when an error is detected. */
|
|
|
1543 STRCPY(langcp, lang);
|
|
|
1544
|
|
310
|
1545 /*
|
|
|
1546 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1547 */
|
|
|
1548 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1549 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1550 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1551
|
|
|
1552 if (r == FAIL && *langcp != NUL)
|
|
|
1553 {
|
|
|
1554 /* Try loading the ASCII version. */
|
|
310
|
1555 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1556 "spell/%s.ascii.spl", lang);
|
|
310
|
1557 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1558 }
|
|
|
1559
|
|
|
1560 if (r == FAIL)
|
|
|
1561 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
|
236
|
1562 fname_enc + 6);
|
|
310
|
1563 else if (*langcp != NUL)
|
|
|
1564 {
|
|
|
1565 /* Load all the additions. */
|
|
|
1566 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1567 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1568 }
|
|
|
1569 }
|
|
|
1570
|
|
|
1571 /*
|
|
|
1572 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1573 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1574 */
|
|
|
1575 static char_u *
|
|
|
1576 spell_enc()
|
|
|
1577 {
|
|
|
1578
|
|
|
1579 #ifdef FEAT_MBYTE
|
|
|
1580 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1581 return p_enc;
|
|
|
1582 #endif
|
|
|
1583 return (char_u *)"latin1";
|
|
236
|
1584 }
|
|
|
1585
|
|
|
1586 /*
|
|
385
|
1587 * Get the name of the .spl file for the internal wordlist into
|
|
|
1588 * "fname[MAXPATHL]".
|
|
|
1589 */
|
|
|
1590 static void
|
|
|
1591 int_wordlist_spl(fname)
|
|
|
1592 char_u *fname;
|
|
|
1593 {
|
|
|
1594 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
1595 int_wordlist, spell_enc());
|
|
|
1596 }
|
|
|
1597
|
|
|
1598 /*
|
|
236
|
1599 * Allocate a new slang_T.
|
|
|
1600 * Caller must fill "sl_next".
|
|
|
1601 */
|
|
|
1602 static slang_T *
|
|
|
1603 slang_alloc(lang)
|
|
|
1604 char_u *lang;
|
|
|
1605 {
|
|
|
1606 slang_T *lp;
|
|
|
1607
|
|
300
|
1608 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1609 if (lp != NULL)
|
|
|
1610 {
|
|
|
1611 lp->sl_name = vim_strsave(lang);
|
|
323
|
1612 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
344
|
1613 ga_init2(&lp->sl_sal, sizeof(salitem_T), 10);
|
|
236
|
1614 }
|
|
|
1615 return lp;
|
|
|
1616 }
|
|
|
1617
|
|
|
1618 /*
|
|
|
1619 * Free the contents of an slang_T and the structure itself.
|
|
|
1620 */
|
|
|
1621 static void
|
|
|
1622 slang_free(lp)
|
|
|
1623 slang_T *lp;
|
|
|
1624 {
|
|
|
1625 vim_free(lp->sl_name);
|
|
310
|
1626 vim_free(lp->sl_fname);
|
|
|
1627 slang_clear(lp);
|
|
|
1628 vim_free(lp);
|
|
|
1629 }
|
|
|
1630
|
|
|
1631 /*
|
|
|
1632 * Clear an slang_T so that the file can be reloaded.
|
|
|
1633 */
|
|
|
1634 static void
|
|
|
1635 slang_clear(lp)
|
|
|
1636 slang_T *lp;
|
|
|
1637 {
|
|
339
|
1638 garray_T *gap;
|
|
|
1639 fromto_T *ftp;
|
|
344
|
1640 salitem_T *smp;
|
|
339
|
1641 int i;
|
|
323
|
1642
|
|
300
|
1643 vim_free(lp->sl_fbyts);
|
|
310
|
1644 lp->sl_fbyts = NULL;
|
|
300
|
1645 vim_free(lp->sl_kbyts);
|
|
310
|
1646 lp->sl_kbyts = NULL;
|
|
339
|
1647 vim_free(lp->sl_pbyts);
|
|
|
1648 lp->sl_pbyts = NULL;
|
|
|
1649
|
|
300
|
1650 vim_free(lp->sl_fidxs);
|
|
310
|
1651 lp->sl_fidxs = NULL;
|
|
300
|
1652 vim_free(lp->sl_kidxs);
|
|
310
|
1653 lp->sl_kidxs = NULL;
|
|
339
|
1654 vim_free(lp->sl_pidxs);
|
|
|
1655 lp->sl_pidxs = NULL;
|
|
323
|
1656
|
|
344
|
1657 gap = &lp->sl_rep;
|
|
|
1658 while (gap->ga_len > 0)
|
|
323
|
1659 {
|
|
344
|
1660 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1661 vim_free(ftp->ft_from);
|
|
|
1662 vim_free(ftp->ft_to);
|
|
323
|
1663 }
|
|
344
|
1664 ga_clear(gap);
|
|
|
1665
|
|
|
1666 gap = &lp->sl_sal;
|
|
375
|
1667 if (lp->sl_sofo)
|
|
376
|
1668 {
|
|
|
1669 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1670 if (gap->ga_data != NULL)
|
|
|
1671 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1672 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1673 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1674 }
|
|
375
|
1675 else
|
|
|
1676 /* SAL items: free salitem_T items */
|
|
|
1677 while (gap->ga_len > 0)
|
|
|
1678 {
|
|
|
1679 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1680 vim_free(smp->sm_lead);
|
|
|
1681 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1682 vim_free(smp->sm_to);
|
|
|
1683 #ifdef FEAT_MBYTE
|
|
|
1684 vim_free(smp->sm_lead_w);
|
|
|
1685 vim_free(smp->sm_oneof_w);
|
|
|
1686 vim_free(smp->sm_to_w);
|
|
|
1687 #endif
|
|
|
1688 }
|
|
344
|
1689 ga_clear(gap);
|
|
323
|
1690
|
|
339
|
1691 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1692 vim_free(lp->sl_prefprog[i]);
|
|
376
|
1693 lp->sl_prefixcnt = 0;
|
|
339
|
1694 vim_free(lp->sl_prefprog);
|
|
376
|
1695 lp->sl_prefprog = NULL;
|
|
|
1696
|
|
|
1697 vim_free(lp->sl_midword);
|
|
|
1698 lp->sl_midword = NULL;
|
|
339
|
1699
|
|
330
|
1700 #ifdef FEAT_MBYTE
|
|
|
1701 {
|
|
|
1702 int todo = lp->sl_map_hash.ht_used;
|
|
|
1703 hashitem_T *hi;
|
|
|
1704
|
|
|
1705 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1706 if (!HASHITEM_EMPTY(hi))
|
|
|
1707 {
|
|
|
1708 --todo;
|
|
|
1709 vim_free(hi->hi_key);
|
|
|
1710 }
|
|
|
1711 }
|
|
|
1712 hash_clear(&lp->sl_map_hash);
|
|
|
1713 #endif
|
|
236
|
1714 }
|
|
|
1715
|
|
|
1716 /*
|
|
307
|
1717 * Load one spell file and store the info into a slang_T.
|
|
236
|
1718 * Invoked through do_in_runtimepath().
|
|
|
1719 */
|
|
|
1720 static void
|
|
310
|
1721 spell_load_cb(fname, cookie)
|
|
236
|
1722 char_u *fname;
|
|
307
|
1723 void *cookie; /* points to the language name */
|
|
236
|
1724 {
|
|
323
|
1725 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1726 }
|
|
|
1727
|
|
|
1728 /*
|
|
|
1729 * Load one spell file and store the info into a slang_T.
|
|
|
1730 *
|
|
|
1731 * This is invoked in two ways:
|
|
|
1732 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
1733 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
1734 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
1735 * points to the existing slang_T.
|
|
323
|
1736 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
1737 */
|
|
323
|
1738 static slang_T *
|
|
|
1739 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
1740 char_u *fname;
|
|
|
1741 char_u *lang;
|
|
|
1742 slang_T *old_lp;
|
|
323
|
1743 int silent; /* no error if file doesn't exist */
|
|
310
|
1744 {
|
|
236
|
1745 FILE *fd;
|
|
|
1746 char_u buf[MAXWLEN + 1];
|
|
|
1747 char_u *p;
|
|
339
|
1748 char_u *bp;
|
|
|
1749 idx_T *ip;
|
|
236
|
1750 int i;
|
|
339
|
1751 int n;
|
|
300
|
1752 int len;
|
|
236
|
1753 int round;
|
|
|
1754 char_u *save_sourcing_name = sourcing_name;
|
|
|
1755 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
255
|
1756 int cnt, ccnt;
|
|
|
1757 char_u *fol;
|
|
307
|
1758 slang_T *lp = NULL;
|
|
323
|
1759 garray_T *gap;
|
|
|
1760 fromto_T *ftp;
|
|
344
|
1761 salitem_T *smp;
|
|
323
|
1762 short *first;
|
|
324
|
1763 idx_T idx;
|
|
344
|
1764 int c = 0;
|
|
236
|
1765
|
|
310
|
1766 fd = mch_fopen((char *)fname, "r");
|
|
236
|
1767 if (fd == NULL)
|
|
|
1768 {
|
|
323
|
1769 if (!silent)
|
|
|
1770 EMSG2(_(e_notopen), fname);
|
|
|
1771 else if (p_verbose > 2)
|
|
|
1772 {
|
|
|
1773 verbose_enter();
|
|
|
1774 smsg((char_u *)e_notopen, fname);
|
|
|
1775 verbose_leave();
|
|
|
1776 }
|
|
255
|
1777 goto endFAIL;
|
|
236
|
1778 }
|
|
310
|
1779 if (p_verbose > 2)
|
|
|
1780 {
|
|
|
1781 verbose_enter();
|
|
|
1782 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
1783 verbose_leave();
|
|
|
1784 }
|
|
|
1785
|
|
|
1786 if (old_lp == NULL)
|
|
|
1787 {
|
|
|
1788 lp = slang_alloc(lang);
|
|
|
1789 if (lp == NULL)
|
|
|
1790 goto endFAIL;
|
|
|
1791
|
|
|
1792 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
1793 lp->sl_fname = vim_strsave(fname);
|
|
|
1794 if (lp->sl_fname == NULL)
|
|
|
1795 goto endFAIL;
|
|
|
1796
|
|
|
1797 /* Check for .add.spl. */
|
|
|
1798 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
1799 }
|
|
|
1800 else
|
|
|
1801 lp = old_lp;
|
|
307
|
1802
|
|
236
|
1803 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
1804 sourcing_name = fname;
|
|
|
1805 sourcing_lnum = 0;
|
|
|
1806
|
|
339
|
1807 /* <HEADER>: <fileID>
|
|
|
1808 * <regioncnt> <regionname> ...
|
|
|
1809 * <charflagslen> <charflags>
|
|
|
1810 * <fcharslen> <fchars>
|
|
366
|
1811 * <midwordlen> <midword>
|
|
339
|
1812 * <prefcondcnt> <prefcond> ...
|
|
|
1813 */
|
|
236
|
1814 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
1815 buf[i] = getc(fd); /* <fileID> */
|
|
|
1816 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
1817 {
|
|
|
1818 EMSG(_("E757: Wrong file ID in spell file"));
|
|
255
|
1819 goto endFAIL;
|
|
236
|
1820 }
|
|
|
1821
|
|
|
1822 cnt = getc(fd); /* <regioncnt> */
|
|
255
|
1823 if (cnt < 0)
|
|
236
|
1824 {
|
|
|
1825 truncerr:
|
|
381
|
1826 EMSG(_(e_spell_trunc));
|
|
255
|
1827 goto endFAIL;
|
|
236
|
1828 }
|
|
|
1829 if (cnt > 8)
|
|
|
1830 {
|
|
|
1831 formerr:
|
|
307
|
1832 EMSG(_(e_format));
|
|
255
|
1833 goto endFAIL;
|
|
236
|
1834 }
|
|
|
1835 for (i = 0; i < cnt; ++i)
|
|
|
1836 {
|
|
|
1837 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
|
1838 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
|
1839 }
|
|
|
1840 lp->sl_regions[cnt * 2] = NUL;
|
|
|
1841
|
|
381
|
1842 /* <charflagslen> <charflags> */
|
|
|
1843 p = read_cnt_string(fd, 1, &cnt);
|
|
389
|
1844 if (cnt < 0)
|
|
381
|
1845 goto endFAIL;
|
|
|
1846
|
|
|
1847 /* <fcharslen> <fchars> */
|
|
389
|
1848 fol = read_cnt_string(fd, 2, &ccnt);
|
|
|
1849 if (ccnt < 0)
|
|
381
|
1850 {
|
|
300
|
1851 vim_free(p);
|
|
381
|
1852 goto endFAIL;
|
|
|
1853 }
|
|
|
1854
|
|
|
1855 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
1856 if (p != NULL && fol != NULL)
|
|
389
|
1857 i = set_spell_charflags(p, cnt, fol);
|
|
381
|
1858
|
|
|
1859 vim_free(p);
|
|
|
1860 vim_free(fol);
|
|
|
1861
|
|
|
1862 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
1863 if ((p == NULL) != (fol == NULL))
|
|
|
1864 goto formerr;
|
|
255
|
1865
|
|
366
|
1866 /* <midwordlen> <midword> */
|
|
376
|
1867 lp->sl_midword = read_cnt_string(fd, 2, &cnt);
|
|
389
|
1868 if (cnt < 0)
|
|
376
|
1869 goto endFAIL;
|
|
366
|
1870
|
|
339
|
1871 /* <prefcondcnt> <prefcond> ... */
|
|
|
1872 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
1873 if (cnt > 0)
|
|
|
1874 {
|
|
|
1875 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
1876 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
1877 if (lp->sl_prefprog == NULL)
|
|
|
1878 goto endFAIL;
|
|
|
1879 lp->sl_prefixcnt = cnt;
|
|
|
1880
|
|
|
1881 for (i = 0; i < cnt; ++i)
|
|
|
1882 {
|
|
|
1883 /* <prefcond> : <condlen> <condstr> */
|
|
|
1884 n = getc(fd); /* <condlen> */
|
|
|
1885 if (n < 0)
|
|
|
1886 goto formerr;
|
|
|
1887 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
1888 * compile the regexp program used to check for the condition. */
|
|
|
1889 if (n > 0)
|
|
|
1890 {
|
|
344
|
1891 buf[0] = '^'; /* always match at one position only */
|
|
|
1892 p = buf + 1;
|
|
339
|
1893 while (n-- > 0)
|
|
|
1894 *p++ = getc(fd); /* <condstr> */
|
|
|
1895 *p = NUL;
|
|
|
1896 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
1897 }
|
|
|
1898 }
|
|
|
1899 }
|
|
|
1900
|
|
|
1901
|
|
323
|
1902 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
1903 * <salflags> <salcount> <sal> ...
|
|
|
1904 * <maplen> <mapstr> */
|
|
339
|
1905
|
|
344
|
1906 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
1907 if (cnt < 0)
|
|
|
1908 goto formerr;
|
|
|
1909
|
|
|
1910 gap = &lp->sl_rep;
|
|
|
1911 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1912 goto endFAIL;
|
|
|
1913
|
|
|
1914 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
1915 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
1916 {
|
|
|
1917 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
381
|
1918 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
389
|
1919 if (i <= 0)
|
|
381
|
1920 goto endFAIL;
|
|
|
1921 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
389
|
1922 if (i <= 0)
|
|
381
|
1923 {
|
|
|
1924 vim_free(ftp->ft_from);
|
|
|
1925 goto endFAIL;
|
|
344
|
1926 }
|
|
|
1927 }
|
|
|
1928
|
|
|
1929 /* Fill the first-index table. */
|
|
|
1930 first = lp->sl_rep_first;
|
|
|
1931 for (i = 0; i < 256; ++i)
|
|
|
1932 first[i] = -1;
|
|
|
1933 for (i = 0; i < gap->ga_len; ++i)
|
|
323
|
1934 {
|
|
344
|
1935 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
1936 if (first[*ftp->ft_from] == -1)
|
|
|
1937 first[*ftp->ft_from] = i;
|
|
|
1938 }
|
|
|
1939
|
|
|
1940 i = getc(fd); /* <salflags> */
|
|
|
1941 if (i & SAL_F0LLOWUP)
|
|
|
1942 lp->sl_followup = TRUE;
|
|
|
1943 if (i & SAL_COLLAPSE)
|
|
|
1944 lp->sl_collapse = TRUE;
|
|
|
1945 if (i & SAL_REM_ACCENTS)
|
|
|
1946 lp->sl_rem_accents = TRUE;
|
|
375
|
1947 if (i & SAL_SOFO)
|
|
|
1948 lp->sl_sofo = TRUE;
|
|
389
|
1949 else
|
|
|
1950 lp->sl_sofo = FALSE;
|
|
344
|
1951
|
|
|
1952 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
1953 if (cnt < 0)
|
|
|
1954 goto formerr;
|
|
|
1955
|
|
375
|
1956 if (lp->sl_sofo)
|
|
|
1957 {
|
|
|
1958 /*
|
|
|
1959 * SOFOFROM and SOFOTO items come in one <salfrom> and <salto>
|
|
|
1960 */
|
|
|
1961 if (cnt != 1)
|
|
|
1962 goto formerr;
|
|
|
1963
|
|
381
|
1964 /* <salfromlen> <salfrom> */
|
|
|
1965 bp = read_cnt_string(fd, 2, &cnt);
|
|
389
|
1966 if (cnt < 0)
|
|
375
|
1967 goto endFAIL;
|
|
381
|
1968
|
|
|
1969 /* <saltolen> <salto> */
|
|
|
1970 fol = read_cnt_string(fd, 2, &cnt);
|
|
389
|
1971 if (cnt < 0)
|
|
375
|
1972 {
|
|
|
1973 vim_free(bp);
|
|
344
|
1974 goto endFAIL;
|
|
375
|
1975 }
|
|
381
|
1976
|
|
|
1977 /* Store the info in lp->sl_sal and/or lp->sl_sal_first. */
|
|
389
|
1978 if (bp != NULL && fol != NULL)
|
|
|
1979 i = set_sofo(lp, bp, fol);
|
|
|
1980 else if (bp != NULL || fol != NULL)
|
|
|
1981 i = FAIL; /* only one of two strings is an error */
|
|
|
1982 else
|
|
|
1983 i = OK;
|
|
381
|
1984
|
|
375
|
1985 vim_free(bp);
|
|
|
1986 vim_free(fol);
|
|
381
|
1987 if (i == FAIL)
|
|
|
1988 goto formerr;
|
|
375
|
1989 }
|
|
|
1990 else
|
|
|
1991 {
|
|
|
1992 /*
|
|
|
1993 * SAL items
|
|
|
1994 */
|
|
|
1995 gap = &lp->sl_sal;
|
|
|
1996 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1997 goto endFAIL;
|
|
|
1998
|
|
|
1999 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2000 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2001 {
|
|
|
2002 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2003 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2004 if (ccnt < 0)
|
|
|
2005 goto formerr;
|
|
|
2006 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2007 goto endFAIL;
|
|
|
2008 smp->sm_lead = p;
|
|
|
2009
|
|
|
2010 /* Read up to the first special char into sm_lead. */
|
|
|
2011 for (i = 0; i < ccnt; ++i)
|
|
|
2012 {
|
|
|
2013 c = getc(fd); /* <salfrom> */
|
|
|
2014 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2015 break;
|
|
|
2016 *p++ = c;
|
|
|
2017 }
|
|
|
2018 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2019 *p++ = NUL;
|
|
|
2020
|
|
|
2021 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2022 if (c == '(')
|
|
|
2023 {
|
|
|
2024 smp->sm_oneof = p;
|
|
|
2025 for (++i; i < ccnt; ++i)
|
|
|
2026 {
|
|
|
2027 c = getc(fd); /* <salfrom> */
|
|
|
2028 if (c == ')')
|
|
|
2029 break;
|
|
|
2030 *p++ = c;
|
|
|
2031 }
|
|
|
2032 *p++ = NUL;
|
|
|
2033 if (++i < ccnt)
|
|
|
2034 c = getc(fd);
|
|
|
2035 }
|
|
372
|
2036 else
|
|
375
|
2037 smp->sm_oneof = NULL;
|
|
|
2038
|
|
|
2039 /* Any following chars go in sm_rules. */
|
|
|
2040 smp->sm_rules = p;
|
|
|
2041 if (i < ccnt)
|
|
|
2042 /* store the char we got while checking for end of sm_lead */
|
|
|
2043 *p++ = c;
|
|
|
2044 for (++i; i < ccnt; ++i)
|
|
|
2045 *p++ = getc(fd); /* <salfrom> */
|
|
|
2046 *p++ = NUL;
|
|
|
2047
|
|
381
|
2048 /* <saltolen> <salto> */
|
|
|
2049 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
389
|
2050 if (ccnt < 0)
|
|
372
|
2051 {
|
|
|
2052 vim_free(smp->sm_lead);
|
|
375
|
2053 goto formerr;
|
|
|
2054 }
|
|
|
2055
|
|
372
|
2056 #ifdef FEAT_MBYTE
|
|
|
2057 if (has_mbyte)
|
|
|
2058 {
|
|
375
|
2059 /* convert the multi-byte strings to wide char strings */
|
|
|
2060 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2061 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2062 if (smp->sm_oneof == NULL)
|
|
|
2063 smp->sm_oneof_w = NULL;
|
|
|
2064 else
|
|
|
2065 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
389
|
2066 if (smp->sm_to == NULL)
|
|
|
2067 smp->sm_to_w = NULL;
|
|
|
2068 else
|
|
|
2069 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
375
|
2070 if (smp->sm_lead_w == NULL
|
|
|
2071 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
389
|
2072 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
372
|
2073 {
|
|
375
|
2074 vim_free(smp->sm_lead);
|
|
|
2075 vim_free(smp->sm_to);
|
|
|
2076 vim_free(smp->sm_lead_w);
|
|
|
2077 vim_free(smp->sm_oneof_w);
|
|
|
2078 vim_free(smp->sm_to_w);
|
|
|
2079 goto endFAIL;
|
|
372
|
2080 }
|
|
|
2081 }
|
|
|
2082 #endif
|
|
|
2083 }
|
|
375
|
2084
|
|
|
2085 /* Fill the first-index table. */
|
|
381
|
2086 set_sal_first(lp);
|
|
|
2087 }
|
|
|
2088
|
|
|
2089 /* <maplen> <mapstr> */
|
|
|
2090 p = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2091 if (cnt < 0)
|
|
323
|
2092 goto endFAIL;
|
|
389
|
2093 if (p != NULL)
|
|
|
2094 {
|
|
|
2095 set_map_str(lp, p);
|
|
|
2096 vim_free(p);
|
|
|
2097 }
|
|
236
|
2098
|
|
300
|
2099 /* round 1: <LWORDTREE>
|
|
339
|
2100 * round 2: <KWORDTREE>
|
|
|
2101 * round 3: <PREFIXTREE> */
|
|
|
2102 for (round = 1; round <= 3; ++round)
|
|
236
|
2103 {
|
|
300
|
2104 /* The tree size was computed when writing the file, so that we can
|
|
|
2105 * allocate it as one long block. <nodecount> */
|
|
|
2106 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2107 if (len < 0)
|
|
|
2108 goto truncerr;
|
|
|
2109 if (len > 0)
|
|
236
|
2110 {
|
|
300
|
2111 /* Allocate the byte array. */
|
|
339
|
2112 bp = lalloc((long_u)len, TRUE);
|
|
|
2113 if (bp == NULL)
|
|
300
|
2114 goto endFAIL;
|
|
|
2115 if (round == 1)
|
|
339
|
2116 lp->sl_fbyts = bp;
|
|
|
2117 else if (round == 2)
|
|
|
2118 lp->sl_kbyts = bp;
|
|
300
|
2119 else
|
|
339
|
2120 lp->sl_pbyts = bp;
|
|
236
|
2121
|
|
300
|
2122 /* Allocate the index array. */
|
|
339
|
2123 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2124 if (ip == NULL)
|
|
300
|
2125 goto endFAIL;
|
|
|
2126 if (round == 1)
|
|
339
|
2127 lp->sl_fidxs = ip;
|
|
|
2128 else if (round == 2)
|
|
|
2129 lp->sl_kidxs = ip;
|
|
300
|
2130 else
|
|
339
|
2131 lp->sl_pidxs = ip;
|
|
300
|
2132
|
|
|
2133 /* Read the tree and store it in the array. */
|
|
339
|
2134 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2135 if (idx == -1)
|
|
300
|
2136 goto truncerr;
|
|
324
|
2137 if (idx < 0)
|
|
236
|
2138 goto formerr;
|
|
|
2139 }
|
|
300
|
2140 }
|
|
243
|
2141
|
|
310
|
2142 /* For a new file link it in the list of spell files. */
|
|
|
2143 if (old_lp == NULL)
|
|
|
2144 {
|
|
|
2145 lp->sl_next = first_lang;
|
|
|
2146 first_lang = lp;
|
|
|
2147 }
|
|
307
|
2148
|
|
255
|
2149 goto endOK;
|
|
|
2150
|
|
|
2151 endFAIL:
|
|
310
|
2152 if (lang != NULL)
|
|
|
2153 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2154 *lang = NUL;
|
|
|
2155 if (lp != NULL && old_lp == NULL)
|
|
323
|
2156 {
|
|
307
|
2157 slang_free(lp);
|
|
323
|
2158 lp = NULL;
|
|
|
2159 }
|
|
255
|
2160
|
|
|
2161 endOK:
|
|
236
|
2162 if (fd != NULL)
|
|
|
2163 fclose(fd);
|
|
|
2164 sourcing_name = save_sourcing_name;
|
|
|
2165 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2166
|
|
|
2167 return lp;
|
|
236
|
2168 }
|
|
|
2169
|
|
376
|
2170 /*
|
|
|
2171 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2172 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2173 * Returns NULL when the count is zero.
|
|
389
|
2174 * Sets "*cntp" to -1 when there is an error, length of the result otherwise.
|
|
376
|
2175 */
|
|
|
2176 static char_u *
|
|
389
|
2177 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2178 FILE *fd;
|
|
|
2179 int cnt_bytes;
|
|
389
|
2180 int *cntp;
|
|
376
|
2181 {
|
|
|
2182 int cnt = 0;
|
|
|
2183 int i;
|
|
|
2184 char_u *str;
|
|
|
2185
|
|
|
2186 /* read the length bytes, MSB first */
|
|
|
2187 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2188 cnt = (cnt << 8) + getc(fd);
|
|
|
2189 if (cnt < 0)
|
|
|
2190 {
|
|
381
|
2191 EMSG(_(e_spell_trunc));
|
|
389
|
2192 *cntp = -1;
|
|
376
|
2193 return NULL;
|
|
|
2194 }
|
|
389
|
2195 *cntp = cnt;
|
|
|
2196 if (cnt == 0)
|
|
|
2197 return NULL; /* nothing to read, return NULL */
|
|
376
|
2198
|
|
|
2199 /* allocate memory */
|
|
|
2200 str = alloc((unsigned)cnt + 1);
|
|
|
2201 if (str == NULL)
|
|
|
2202 {
|
|
389
|
2203 *cntp = -1;
|
|
376
|
2204 return NULL;
|
|
|
2205 }
|
|
|
2206
|
|
|
2207 /* Read the string. Doesn't check for truncated file. */
|
|
|
2208 for (i = 0; i < cnt; ++i)
|
|
|
2209 str[i] = getc(fd);
|
|
|
2210 str[i] = NUL;
|
|
|
2211
|
|
|
2212 return str;
|
|
|
2213 }
|
|
|
2214
|
|
381
|
2215 /*
|
|
|
2216 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
|
2217 * Returns FAIL when there is something wrong.
|
|
|
2218 */
|
|
|
2219 static int
|
|
|
2220 set_sofo(lp, from, to)
|
|
|
2221 slang_T *lp;
|
|
|
2222 char_u *from;
|
|
|
2223 char_u *to;
|
|
|
2224 {
|
|
|
2225 int i;
|
|
|
2226
|
|
|
2227 #ifdef FEAT_MBYTE
|
|
|
2228 garray_T *gap;
|
|
|
2229 char_u *s;
|
|
|
2230 char_u *p;
|
|
|
2231 int c;
|
|
|
2232 int *inp;
|
|
|
2233
|
|
|
2234 if (has_mbyte)
|
|
|
2235 {
|
|
|
2236 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2237 * characters. The index is the low byte of the character.
|
|
|
2238 * The list contains from-to pairs with a terminating NUL.
|
|
|
2239 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2240 gap = &lp->sl_sal;
|
|
|
2241 ga_init2(gap, sizeof(int *), 1);
|
|
|
2242 if (ga_grow(gap, 256) == FAIL)
|
|
|
2243 return FAIL;
|
|
|
2244 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2245 gap->ga_len = 256;
|
|
|
2246
|
|
|
2247 /* First count the number of items for each list. Temporarily use
|
|
|
2248 * sl_sal_first[] for this. */
|
|
|
2249 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2250 {
|
|
|
2251 c = mb_ptr2char_adv(&p);
|
|
|
2252 mb_ptr_adv(s);
|
|
|
2253 if (c >= 256)
|
|
|
2254 ++lp->sl_sal_first[c & 0xff];
|
|
|
2255 }
|
|
|
2256 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
|
2257 return FAIL;
|
|
|
2258
|
|
|
2259 /* Allocate the lists. */
|
|
|
2260 for (i = 0; i < 256; ++i)
|
|
|
2261 if (lp->sl_sal_first[i] > 0)
|
|
|
2262 {
|
|
|
2263 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2264 if (p == NULL)
|
|
|
2265 return FAIL;
|
|
|
2266 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2267 *(int *)p = 0;
|
|
|
2268 }
|
|
|
2269
|
|
|
2270 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
2271 * list. */
|
|
|
2272 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2273 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2274 {
|
|
|
2275 c = mb_ptr2char_adv(&p);
|
|
|
2276 i = mb_ptr2char_adv(&s);
|
|
|
2277 if (c >= 256)
|
|
|
2278 {
|
|
|
2279 /* Append the from-to chars at the end of the list with
|
|
|
2280 * the low byte. */
|
|
|
2281 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2282 while (*inp != 0)
|
|
|
2283 ++inp;
|
|
|
2284 *inp++ = c; /* from char */
|
|
|
2285 *inp++ = i; /* to char */
|
|
|
2286 *inp++ = NUL; /* NUL at the end */
|
|
|
2287 }
|
|
|
2288 else
|
|
|
2289 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2290 lp->sl_sal_first[c] = i;
|
|
|
2291 }
|
|
|
2292 }
|
|
|
2293 else
|
|
|
2294 #endif
|
|
|
2295 {
|
|
|
2296 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2297 if (STRLEN(from) != STRLEN(to))
|
|
|
2298 return FAIL;
|
|
|
2299
|
|
|
2300 for (i = 0; to[i] != NUL; ++i)
|
|
|
2301 lp->sl_sal_first[from[i]] = to[i];
|
|
|
2302 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2303 }
|
|
|
2304
|
|
|
2305 return OK;
|
|
|
2306 }
|
|
|
2307
|
|
|
2308 /*
|
|
|
2309 * Fill the first-index table for "lp".
|
|
|
2310 */
|
|
|
2311 static void
|
|
|
2312 set_sal_first(lp)
|
|
|
2313 slang_T *lp;
|
|
|
2314 {
|
|
|
2315 salfirst_T *sfirst;
|
|
|
2316 int i;
|
|
|
2317 salitem_T *smp;
|
|
|
2318 int c;
|
|
|
2319 garray_T *gap = &lp->sl_sal;
|
|
|
2320
|
|
|
2321 sfirst = lp->sl_sal_first;
|
|
|
2322 for (i = 0; i < 256; ++i)
|
|
|
2323 sfirst[i] = -1;
|
|
|
2324 smp = (salitem_T *)gap->ga_data;
|
|
|
2325 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2326 {
|
|
|
2327 #ifdef FEAT_MBYTE
|
|
|
2328 if (has_mbyte)
|
|
|
2329 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2330 * the character, for other encodings it should differ for most
|
|
|
2331 * characters. */
|
|
|
2332 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2333 else
|
|
|
2334 #endif
|
|
|
2335 c = *smp[i].sm_lead;
|
|
|
2336 if (sfirst[c] == -1)
|
|
|
2337 {
|
|
|
2338 sfirst[c] = i;
|
|
|
2339 #ifdef FEAT_MBYTE
|
|
|
2340 if (has_mbyte)
|
|
|
2341 {
|
|
|
2342 int n;
|
|
|
2343
|
|
|
2344 /* Make sure all entries with this byte are following each
|
|
|
2345 * other. Move the ones that are in the wrong position. Do
|
|
|
2346 * keep the same ordering! */
|
|
|
2347 while (i + 1 < gap->ga_len
|
|
|
2348 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
2349 /* Skip over entry with same index byte. */
|
|
|
2350 ++i;
|
|
|
2351
|
|
|
2352 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
2353 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
2354 {
|
|
|
2355 salitem_T tsal;
|
|
|
2356
|
|
|
2357 /* Move entry with same index byte after the entries
|
|
|
2358 * we already found. */
|
|
|
2359 ++i;
|
|
|
2360 --n;
|
|
|
2361 tsal = smp[i + n];
|
|
|
2362 mch_memmove(smp + i + 1, smp + i,
|
|
|
2363 sizeof(salitem_T) * n);
|
|
|
2364 smp[i] = tsal;
|
|
|
2365 }
|
|
|
2366 }
|
|
|
2367 #endif
|
|
|
2368 }
|
|
|
2369 }
|
|
|
2370 }
|
|
376
|
2371
|
|
372
|
2372 #ifdef FEAT_MBYTE
|
|
|
2373 /*
|
|
|
2374 * Turn a multi-byte string into a wide character string.
|
|
|
2375 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
2376 */
|
|
|
2377 static int *
|
|
|
2378 mb_str2wide(s)
|
|
|
2379 char_u *s;
|
|
|
2380 {
|
|
|
2381 int *res;
|
|
|
2382 char_u *p;
|
|
|
2383 int i = 0;
|
|
|
2384
|
|
|
2385 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
2386 if (res != NULL)
|
|
|
2387 {
|
|
|
2388 for (p = s; *p != NUL; )
|
|
|
2389 res[i++] = mb_ptr2char_adv(&p);
|
|
|
2390 res[i] = NUL;
|
|
|
2391 }
|
|
|
2392 return res;
|
|
|
2393 }
|
|
|
2394 #endif
|
|
|
2395
|
|
236
|
2396 /*
|
|
300
|
2397 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
2398 * "byts" and index array "idxs". Recursively read the children.
|
|
|
2399 *
|
|
346
|
2400 * NOTE: The code here must match put_node().
|
|
300
|
2401 *
|
|
|
2402 * Returns the index follosing the siblings.
|
|
|
2403 * Returns -1 if the file is shorter than expected.
|
|
|
2404 * Returns -2 if there is a format error.
|
|
236
|
2405 */
|
|
324
|
2406 static idx_T
|
|
339
|
2407 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
2408 FILE *fd;
|
|
|
2409 char_u *byts;
|
|
324
|
2410 idx_T *idxs;
|
|
300
|
2411 int maxidx; /* size of arrays */
|
|
324
|
2412 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
2413 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
2414 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
2415 {
|
|
300
|
2416 int len;
|
|
|
2417 int i;
|
|
|
2418 int n;
|
|
324
|
2419 idx_T idx = startidx;
|
|
300
|
2420 int c;
|
|
366
|
2421 int c2;
|
|
300
|
2422 #define SHARED_MASK 0x8000000
|
|
236
|
2423
|
|
300
|
2424 len = getc(fd); /* <siblingcount> */
|
|
|
2425 if (len <= 0)
|
|
|
2426 return -1;
|
|
|
2427
|
|
|
2428 if (startidx + len >= maxidx)
|
|
|
2429 return -2;
|
|
|
2430 byts[idx++] = len;
|
|
|
2431
|
|
|
2432 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
2433 for (i = 1; i <= len; ++i)
|
|
236
|
2434 {
|
|
300
|
2435 c = getc(fd); /* <byte> */
|
|
|
2436 if (c < 0)
|
|
|
2437 return -1;
|
|
|
2438 if (c <= BY_SPECIAL)
|
|
|
2439 {
|
|
366
|
2440 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
2441 {
|
|
|
2442 /* No flags, all regions. */
|
|
|
2443 idxs[idx] = 0;
|
|
|
2444 c = 0;
|
|
|
2445 }
|
|
366
|
2446 else if (c == BY_FLAGS || c == BY_NOFLAGS)
|
|
300
|
2447 {
|
|
339
|
2448 if (prefixtree)
|
|
|
2449 {
|
|
|
2450 /* Read the prefix ID and the condition nr. In idxs[]
|
|
|
2451 * store the prefix ID in the low byte, the condition
|
|
|
2452 * index shifted up 8 bits. */
|
|
366
|
2453 c2 = getc(fd); /* <prefixID> */
|
|
339
|
2454 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
2455 if (n >= maxprefcondnr)
|
|
|
2456 return -2;
|
|
366
|
2457 c2 += (n << 8);
|
|
|
2458 if (c == BY_NOFLAGS)
|
|
|
2459 c = c2;
|
|
|
2460 else
|
|
|
2461 c = c2 | WF_RAREPFX;
|
|
339
|
2462 }
|
|
|
2463 else
|
|
|
2464 {
|
|
|
2465 /* Read flags and optional region and prefix ID. In
|
|
|
2466 * idxs[] the flags go in the low byte, region above that
|
|
|
2467 * and prefix ID above the region. */
|
|
|
2468 c = getc(fd); /* <flags> */
|
|
|
2469 if (c & WF_REGION)
|
|
|
2470 c = (getc(fd) << 8) + c; /* <region> */
|
|
|
2471 if (c & WF_PFX)
|
|
|
2472 c = (getc(fd) << 16) + c; /* <prefixID> */
|
|
|
2473 }
|
|
|
2474
|
|
300
|
2475 idxs[idx] = c;
|
|
|
2476 c = 0;
|
|
|
2477 }
|
|
|
2478 else /* c == BY_INDEX */
|
|
|
2479 {
|
|
|
2480 /* <nodeidx> */
|
|
|
2481 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2482 if (n < 0 || n >= maxidx)
|
|
|
2483 return -2;
|
|
|
2484 idxs[idx] = n + SHARED_MASK;
|
|
|
2485 c = getc(fd); /* <xbyte> */
|
|
|
2486 }
|
|
|
2487 }
|
|
|
2488 byts[idx++] = c;
|
|
236
|
2489 }
|
|
|
2490
|
|
300
|
2491 /* Recursively read the children for non-shared siblings.
|
|
|
2492 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
2493 * remove SHARED_MASK) */
|
|
|
2494 for (i = 1; i <= len; ++i)
|
|
|
2495 if (byts[startidx + i] != 0)
|
|
|
2496 {
|
|
|
2497 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
2498 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
2499 else
|
|
|
2500 {
|
|
|
2501 idxs[startidx + i] = idx;
|
|
339
|
2502 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
2503 prefixtree, maxprefcondnr);
|
|
300
|
2504 if (idx < 0)
|
|
|
2505 break;
|
|
|
2506 }
|
|
|
2507 }
|
|
236
|
2508
|
|
300
|
2509 return idx;
|
|
236
|
2510 }
|
|
|
2511
|
|
|
2512 /*
|
|
|
2513 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
2514 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
2515 */
|
|
|
2516 char_u *
|
|
|
2517 did_set_spelllang(buf)
|
|
|
2518 buf_T *buf;
|
|
|
2519 {
|
|
|
2520 garray_T ga;
|
|
351
|
2521 char_u *splp;
|
|
236
|
2522 char_u *region;
|
|
355
|
2523 int filename;
|
|
236
|
2524 int region_mask;
|
|
|
2525 slang_T *lp;
|
|
|
2526 int c;
|
|
351
|
2527 char_u lang[MAXWLEN + 1];
|
|
323
|
2528 char_u spf_name[MAXPATHL];
|
|
351
|
2529 int len;
|
|
|
2530 char_u *p;
|
|
381
|
2531 int round;
|
|
385
|
2532 char_u *spf;
|
|
389
|
2533 char_u *use_region = NULL;
|
|
|
2534 int dont_use_region = FALSE;
|
|
236
|
2535
|
|
|
2536 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
2537 clear_midword(buf);
|
|
236
|
2538
|
|
351
|
2539 /* loop over comma separated language names. */
|
|
|
2540 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
2541 {
|
|
|
2542 /* Get one language name. */
|
|
|
2543 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
2544
|
|
240
|
2545 region = NULL;
|
|
351
|
2546 len = STRLEN(lang);
|
|
355
|
2547
|
|
|
2548 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
2549 * If there is a region name let "region" point to it and remove it
|
|
|
2550 * from the name. */
|
|
|
2551 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
2552 {
|
|
355
|
2553 filename = TRUE;
|
|
|
2554
|
|
|
2555 /* Check if we loaded this language before. */
|
|
|
2556 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2557 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2558 break;
|
|
236
|
2559 }
|
|
355
|
2560 else
|
|
|
2561 {
|
|
|
2562 filename = FALSE;
|
|
|
2563 if (len > 3 && lang[len - 3] == '_')
|
|
|
2564 {
|
|
|
2565 region = lang + len - 2;
|
|
|
2566 len -= 3;
|
|
|
2567 lang[len] = NUL;
|
|
389
|
2568
|
|
|
2569 /* If the region differs from what was used before then don't
|
|
|
2570 * use it for 'spellfile'. */
|
|
|
2571 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
2572 dont_use_region = TRUE;
|
|
|
2573 use_region = region;
|
|
|
2574 }
|
|
|
2575 else
|
|
|
2576 dont_use_region = TRUE;
|
|
355
|
2577
|
|
|
2578 /* Check if we loaded this language before. */
|
|
|
2579 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2580 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
2581 break;
|
|
|
2582 }
|
|
236
|
2583
|
|
351
|
2584 /* If not found try loading the language now. */
|
|
236
|
2585 if (lp == NULL)
|
|
355
|
2586 {
|
|
|
2587 if (filename)
|
|
|
2588 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
2589 else
|
|
|
2590 spell_load_lang(lang);
|
|
|
2591 }
|
|
236
|
2592
|
|
307
|
2593 /*
|
|
351
|
2594 * Loop over the languages, there can be several files for "lang".
|
|
307
|
2595 */
|
|
|
2596 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
2597 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
2598 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
2599 {
|
|
316
|
2600 region_mask = REGION_ALL;
|
|
355
|
2601 if (!filename && region != NULL)
|
|
236
|
2602 {
|
|
307
|
2603 /* find region in sl_regions */
|
|
|
2604 c = find_region(lp->sl_regions, region);
|
|
|
2605 if (c == REGION_ALL)
|
|
|
2606 {
|
|
389
|
2607 if (lp->sl_add)
|
|
|
2608 {
|
|
|
2609 if (*lp->sl_regions != NUL)
|
|
|
2610 /* This addition file is for other regions. */
|
|
|
2611 region_mask = 0;
|
|
|
2612 }
|
|
|
2613 else
|
|
|
2614 /* This is probably an error. Give a warning and
|
|
|
2615 * accept the words anyway. */
|
|
351
|
2616 smsg((char_u *)
|
|
|
2617 _("Warning: region %s not supported"),
|
|
|
2618 region);
|
|
307
|
2619 }
|
|
|
2620 else
|
|
|
2621 region_mask = 1 << c;
|
|
236
|
2622 }
|
|
307
|
2623
|
|
389
|
2624 if (region_mask != 0)
|
|
307
|
2625 {
|
|
389
|
2626 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2627 {
|
|
|
2628 ga_clear(&ga);
|
|
|
2629 return e_outofmem;
|
|
|
2630 }
|
|
|
2631 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2632 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2633 ++ga.ga_len;
|
|
|
2634 use_midword(lp, buf);
|
|
307
|
2635 }
|
|
385
|
2636 }
|
|
|
2637 }
|
|
|
2638
|
|
|
2639 /* round 0: load int_wordlist, if possible.
|
|
|
2640 * round 1: load first name in 'spellfile'.
|
|
|
2641 * round 2: load second name in 'spellfile.
|
|
|
2642 * etc. */
|
|
|
2643 spf = curbuf->b_p_spf;
|
|
|
2644 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
2645 {
|
|
|
2646 if (round == 0)
|
|
|
2647 {
|
|
|
2648 /* Internal wordlist, if there is one. */
|
|
|
2649 if (int_wordlist == NULL)
|
|
381
|
2650 continue;
|
|
385
|
2651 int_wordlist_spl(spf_name);
|
|
381
|
2652 }
|
|
|
2653 else
|
|
|
2654 {
|
|
385
|
2655 /* One entry in 'spellfile'. */
|
|
|
2656 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
2657 STRCAT(spf_name, ".spl");
|
|
|
2658
|
|
|
2659 /* If it was already found above then skip it. */
|
|
|
2660 for (c = 0; c < ga.ga_len; ++c)
|
|
|
2661 if (fullpathcmp(spf_name,
|
|
|
2662 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
2663 FALSE) == FPC_SAME)
|
|
|
2664 break;
|
|
|
2665 if (c < ga.ga_len)
|
|
381
|
2666 continue;
|
|
|
2667 }
|
|
|
2668
|
|
351
|
2669 /* Check if it was loaded already. */
|
|
323
|
2670 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2671 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2672 break;
|
|
|
2673 if (lp == NULL)
|
|
|
2674 {
|
|
351
|
2675 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
2676 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
2677 * use an arbitrary name. */
|
|
|
2678 if (round == 0)
|
|
|
2679 STRCPY(lang, "internal wordlist");
|
|
|
2680 else
|
|
|
2681 {
|
|
|
2682 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
2683 p = vim_strchr(lang, '.');
|
|
|
2684 if (p != NULL)
|
|
|
2685 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
2686 }
|
|
351
|
2687 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
2688 }
|
|
|
2689 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
2690 {
|
|
389
|
2691 region_mask = REGION_ALL;
|
|
|
2692 if (use_region != NULL && !dont_use_region)
|
|
|
2693 {
|
|
|
2694 /* find region in sl_regions */
|
|
|
2695 c = find_region(lp->sl_regions, use_region);
|
|
|
2696 if (c != REGION_ALL)
|
|
|
2697 region_mask = 1 << c;
|
|
|
2698 else if (*lp->sl_regions != NUL)
|
|
|
2699 /* This spell file is for other regions. */
|
|
|
2700 region_mask = 0;
|
|
|
2701 }
|
|
|
2702
|
|
|
2703 if (region_mask != 0)
|
|
|
2704 {
|
|
|
2705 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2706 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2707 ++ga.ga_len;
|
|
|
2708 use_midword(lp, buf);
|
|
|
2709 }
|
|
323
|
2710 }
|
|
|
2711 }
|
|
|
2712
|
|
236
|
2713 /* Add a NULL entry to mark the end of the list. */
|
|
|
2714 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2715 {
|
|
|
2716 ga_clear(&ga);
|
|
|
2717 return e_outofmem;
|
|
|
2718 }
|
|
|
2719 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
2720 ++ga.ga_len;
|
|
|
2721
|
|
|
2722 /* Everything is fine, store the new b_langp value. */
|
|
|
2723 ga_clear(&buf->b_langp);
|
|
|
2724 buf->b_langp = ga;
|
|
|
2725
|
|
|
2726 return NULL;
|
|
|
2727 }
|
|
|
2728
|
|
|
2729 /*
|
|
376
|
2730 * Clear the midword characters for buffer "buf".
|
|
|
2731 */
|
|
|
2732 static void
|
|
|
2733 clear_midword(buf)
|
|
|
2734 buf_T *buf;
|
|
|
2735 {
|
|
|
2736 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
2737 #ifdef FEAT_MBYTE
|
|
|
2738 vim_free(buf->b_spell_ismw_mb);
|
|
|
2739 buf->b_spell_ismw_mb = NULL;
|
|
|
2740 #endif
|
|
|
2741 }
|
|
|
2742
|
|
|
2743 /*
|
|
|
2744 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
2745 * They add up to any currently used midword characters.
|
|
|
2746 */
|
|
|
2747 static void
|
|
|
2748 use_midword(lp, buf)
|
|
|
2749 slang_T *lp;
|
|
|
2750 buf_T *buf;
|
|
|
2751 {
|
|
|
2752 char_u *p;
|
|
|
2753
|
|
389
|
2754 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
2755 return;
|
|
|
2756
|
|
376
|
2757 for (p = lp->sl_midword; *p != NUL; )
|
|
|
2758 #ifdef FEAT_MBYTE
|
|
|
2759 if (has_mbyte)
|
|
|
2760 {
|
|
|
2761 int c, l, n;
|
|
|
2762 char_u *bp;
|
|
|
2763
|
|
|
2764 c = mb_ptr2char(p);
|
|
|
2765 l = mb_ptr2len_check(p);
|
|
|
2766 if (c < 256)
|
|
|
2767 buf->b_spell_ismw[c] = TRUE;
|
|
|
2768 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
2769 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
2770 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
2771 else
|
|
|
2772 {
|
|
|
2773 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
2774 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
2775 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
2776 if (bp != NULL)
|
|
|
2777 {
|
|
|
2778 vim_free(buf->b_spell_ismw_mb);
|
|
|
2779 buf->b_spell_ismw_mb = bp;
|
|
|
2780 vim_strncpy(bp + n, p, l);
|
|
|
2781 }
|
|
|
2782 }
|
|
|
2783 p += l;
|
|
|
2784 }
|
|
|
2785 else
|
|
|
2786 #endif
|
|
|
2787 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
2788 }
|
|
|
2789
|
|
|
2790 /*
|
|
236
|
2791 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
2792 * Each region is simply stored as the two characters of it's name.
|
|
381
|
2793 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
2794 */
|
|
|
2795 static int
|
|
|
2796 find_region(rp, region)
|
|
|
2797 char_u *rp;
|
|
|
2798 char_u *region;
|
|
|
2799 {
|
|
|
2800 int i;
|
|
|
2801
|
|
|
2802 for (i = 0; ; i += 2)
|
|
|
2803 {
|
|
|
2804 if (rp[i] == NUL)
|
|
|
2805 return REGION_ALL;
|
|
|
2806 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
2807 break;
|
|
|
2808 }
|
|
|
2809 return i / 2;
|
|
|
2810 }
|
|
|
2811
|
|
|
2812 /*
|
|
323
|
2813 * Return case type of word:
|
|
236
|
2814 * w word 0
|
|
300
|
2815 * Word WF_ONECAP
|
|
|
2816 * W WORD WF_ALLCAP
|
|
|
2817 * WoRd wOrd WF_KEEPCAP
|
|
236
|
2818 */
|
|
|
2819 static int
|
|
|
2820 captype(word, end)
|
|
|
2821 char_u *word;
|
|
323
|
2822 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
2823 {
|
|
|
2824 char_u *p;
|
|
|
2825 int c;
|
|
|
2826 int firstcap;
|
|
|
2827 int allcap;
|
|
|
2828 int past_second = FALSE; /* past second word char */
|
|
|
2829
|
|
|
2830 /* find first letter */
|
|
376
|
2831 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
2832 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
2833 return 0; /* only non-word characters, illegal word */
|
|
|
2834 #ifdef FEAT_MBYTE
|
|
310
|
2835 if (has_mbyte)
|
|
|
2836 c = mb_ptr2char_adv(&p);
|
|
|
2837 else
|
|
236
|
2838 #endif
|
|
310
|
2839 c = *p++;
|
|
324
|
2840 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
2841
|
|
|
2842 /*
|
|
|
2843 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
2844 * But a word with an upper char only at start is a ONECAP.
|
|
|
2845 */
|
|
323
|
2846 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
2847 if (spell_iswordp_nmw(p))
|
|
236
|
2848 {
|
|
|
2849 #ifdef FEAT_MBYTE
|
|
|
2850 c = mb_ptr2char(p);
|
|
|
2851 #else
|
|
|
2852 c = *p;
|
|
|
2853 #endif
|
|
324
|
2854 if (!SPELL_ISUPPER(c))
|
|
236
|
2855 {
|
|
|
2856 /* UUl -> KEEPCAP */
|
|
|
2857 if (past_second && allcap)
|
|
300
|
2858 return WF_KEEPCAP;
|
|
236
|
2859 allcap = FALSE;
|
|
|
2860 }
|
|
|
2861 else if (!allcap)
|
|
|
2862 /* UlU -> KEEPCAP */
|
|
300
|
2863 return WF_KEEPCAP;
|
|
236
|
2864 past_second = TRUE;
|
|
|
2865 }
|
|
|
2866
|
|
|
2867 if (allcap)
|
|
300
|
2868 return WF_ALLCAP;
|
|
236
|
2869 if (firstcap)
|
|
300
|
2870 return WF_ONECAP;
|
|
236
|
2871 return 0;
|
|
|
2872 }
|
|
|
2873
|
|
355
|
2874 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
2875 /*
|
|
|
2876 * Free all languages.
|
|
|
2877 */
|
|
|
2878 void
|
|
|
2879 spell_free_all()
|
|
|
2880 {
|
|
|
2881 slang_T *lp;
|
|
|
2882 buf_T *buf;
|
|
385
|
2883 char_u fname[MAXPATHL];
|
|
355
|
2884
|
|
|
2885 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2886 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2887 ga_clear(&buf->b_langp);
|
|
|
2888
|
|
|
2889 while (first_lang != NULL)
|
|
|
2890 {
|
|
|
2891 lp = first_lang;
|
|
|
2892 first_lang = lp->sl_next;
|
|
|
2893 slang_free(lp);
|
|
|
2894 }
|
|
366
|
2895
|
|
385
|
2896 if (int_wordlist != NULL)
|
|
|
2897 {
|
|
|
2898 /* Delete the internal wordlist and its .spl file */
|
|
|
2899 mch_remove(int_wordlist);
|
|
|
2900 int_wordlist_spl(fname);
|
|
|
2901 mch_remove(fname);
|
|
|
2902 vim_free(int_wordlist);
|
|
|
2903 int_wordlist = NULL;
|
|
381
|
2904 }
|
|
|
2905
|
|
366
|
2906 init_spell_chartab();
|
|
355
|
2907 }
|
|
|
2908 # endif
|
|
|
2909
|
|
236
|
2910 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
2911 /*
|
|
|
2912 * Clear all spelling tables and reload them.
|
|
307
|
2913 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
2914 */
|
|
|
2915 void
|
|
|
2916 spell_reload()
|
|
|
2917 {
|
|
|
2918 buf_T *buf;
|
|
316
|
2919 win_T *wp;
|
|
236
|
2920
|
|
358
|
2921 /* Initialize the table for spell_iswordp(). */
|
|
236
|
2922 init_spell_chartab();
|
|
|
2923
|
|
|
2924 /* Unload all allocated memory. */
|
|
355
|
2925 spell_free_all();
|
|
236
|
2926
|
|
|
2927 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2928 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2929 {
|
|
316
|
2930 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
2931 * window for this buffer in which 'spell' is set. */
|
|
236
|
2932 if (*buf->b_p_spl != NUL)
|
|
316
|
2933 {
|
|
|
2934 FOR_ALL_WINDOWS(wp)
|
|
|
2935 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
2936 {
|
|
|
2937 (void)did_set_spelllang(buf);
|
|
|
2938 # ifdef FEAT_WINDOWS
|
|
|
2939 break;
|
|
|
2940 # endif
|
|
|
2941 }
|
|
|
2942 }
|
|
236
|
2943 }
|
|
|
2944 }
|
|
|
2945 # endif
|
|
|
2946
|
|
310
|
2947 /*
|
|
|
2948 * Reload the spell file "fname" if it's loaded.
|
|
|
2949 */
|
|
|
2950 static void
|
|
323
|
2951 spell_reload_one(fname, added_word)
|
|
310
|
2952 char_u *fname;
|
|
323
|
2953 int added_word; /* invoked through "zg" */
|
|
310
|
2954 {
|
|
|
2955 slang_T *lp;
|
|
323
|
2956 int didit = FALSE;
|
|
310
|
2957
|
|
|
2958 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2959 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2960 {
|
|
|
2961 slang_clear(lp);
|
|
323
|
2962 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
2963 redraw_all_later(NOT_VALID);
|
|
323
|
2964 didit = TRUE;
|
|
310
|
2965 }
|
|
323
|
2966
|
|
|
2967 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
2968 * 'spelllang' to get it loaded. */
|
|
|
2969 if (added_word && !didit)
|
|
|
2970 did_set_spelllang(curbuf);
|
|
310
|
2971 }
|
|
|
2972
|
|
|
2973
|
|
236
|
2974 /*
|
|
|
2975 * Functions for ":mkspell".
|
|
|
2976 */
|
|
|
2977
|
|
300
|
2978 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
2979 and .dic file. */
|
|
|
2980 /*
|
|
|
2981 * Main structure to store the contents of a ".aff" file.
|
|
|
2982 */
|
|
|
2983 typedef struct afffile_S
|
|
|
2984 {
|
|
|
2985 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
310
|
2986 int af_rar; /* RAR ID for rare word */
|
|
|
2987 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
2988 int af_bad; /* BAD ID for banned word */
|
|
339
|
2989 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
2990 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
2991 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
2992 } afffile_T;
|
|
|
2993
|
|
|
2994 typedef struct affentry_S affentry_T;
|
|
|
2995 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
2996 struct affentry_S
|
|
|
2997 {
|
|
|
2998 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
2999 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3000 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3001 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3002 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
366
|
3003 int ae_rare; /* rare affix */
|
|
300
|
3004 };
|
|
|
3005
|
|
|
3006 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3007 typedef struct affheader_S
|
|
|
3008 {
|
|
|
3009 char_u ah_key[2]; /* key for hashtable == name of affix entry */
|
|
339
|
3010 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
3011 int ah_combine; /* suffix may combine with prefix */
|
|
|
3012 affentry_T *ah_first; /* first affix entry */
|
|
|
3013 } affheader_T;
|
|
|
3014
|
|
|
3015 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
3016
|
|
|
3017 /*
|
|
|
3018 * Structure that is used to store the items in the word tree. This avoids
|
|
|
3019 * the need to keep track of each allocated thing, it's freed all at once
|
|
|
3020 * after ":mkspell" is done.
|
|
|
3021 */
|
|
|
3022 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
3023 typedef struct sblock_S sblock_T;
|
|
|
3024 struct sblock_S
|
|
|
3025 {
|
|
|
3026 sblock_T *sb_next; /* next block in list */
|
|
|
3027 int sb_used; /* nr of bytes already in use */
|
|
|
3028 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
3029 };
|
|
|
3030
|
|
|
3031 /*
|
|
300
|
3032 * A node in the tree.
|
|
236
|
3033 */
|
|
300
|
3034 typedef struct wordnode_S wordnode_T;
|
|
|
3035 struct wordnode_S
|
|
236
|
3036 {
|
|
346
|
3037 union /* shared to save space */
|
|
|
3038 {
|
|
|
3039 char_u hashkey[6]; /* room for the hash key */
|
|
|
3040 int index; /* index in written nodes (valid after first
|
|
|
3041 round) */
|
|
|
3042 } wn_u1;
|
|
|
3043 union /* shared to save space */
|
|
|
3044 {
|
|
|
3045 wordnode_T *next; /* next node with same hash key */
|
|
|
3046 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
3047 } wn_u2;
|
|
300
|
3048 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
3049 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
3050 always sorted) */
|
|
|
3051 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
|
3052 char_u wn_flags; /* when wn_byte is NUL: WF_ flags */
|
|
339
|
3053 short wn_region; /* when wn_byte is NUL: region mask; for
|
|
|
3054 PREFIXTREE it's the prefcondnr */
|
|
|
3055 char_u wn_prefixID; /* supported/required prefix ID or 0 */
|
|
236
|
3056 };
|
|
|
3057
|
|
300
|
3058 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
3059
|
|
300
|
3060 /*
|
|
|
3061 * Info used while reading the spell files.
|
|
|
3062 */
|
|
|
3063 typedef struct spellinfo_S
|
|
249
|
3064 {
|
|
300
|
3065 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
3066 long si_foldwcount; /* nr of words in si_foldroot */
|
|
300
|
3067 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
3068 long si_keepwcount; /* nr of words in si_keeproot */
|
|
339
|
3069 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
300
|
3070 sblock_T *si_blocks; /* memory blocks used */
|
|
|
3071 int si_ascii; /* handling only ASCII words */
|
|
310
|
3072 int si_add; /* addition file */
|
|
351
|
3073 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
3074 int si_region; /* region mask */
|
|
|
3075 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
3076 int si_memtot; /* runtime memory used */
|
|
310
|
3077 int si_verbose; /* verbose messages */
|
|
316
|
3078 int si_region_count; /* number of regions supported (1 when there
|
|
|
3079 are no regions) */
|
|
|
3080 char_u si_region_name[16]; /* region names (if count > 1) */
|
|
323
|
3081
|
|
|
3082 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
3083 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
3084 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
3085 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
3086 int si_followup; /* soundsalike: ? */
|
|
|
3087 int si_collapse; /* soundsalike: ? */
|
|
|
3088 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
3089 garray_T si_map; /* MAP info concatenated */
|
|
366
|
3090 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
339
|
3091 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
3092 * prefixes, each stored as a string */
|
|
|
3093 int si_newID; /* current value for ah_newID */
|
|
300
|
3094 } spellinfo_T;
|
|
249
|
3095
|
|
300
|
3096 static afffile_T *spell_read_aff __ARGS((char_u *fname, spellinfo_T *spin));
|
|
339
|
3097 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
3098 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
3099 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
3100 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
3101 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
|
3102 static int spell_read_dic __ARGS((char_u *fname, spellinfo_T *spin, afffile_T *affile));
|
|
339
|
3103 static char_u *get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, sblock_T **blp));
|
|
|
3104 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
|
3105 static int spell_read_wordfile __ARGS((char_u *fname, spellinfo_T *spin));
|
|
|
3106 static void *getroom __ARGS((sblock_T **blp, size_t len));
|
|
|
3107 static char_u *getroom_save __ARGS((sblock_T **blp, char_u *s));
|
|
|
3108 static void free_blocks __ARGS((sblock_T *bl));
|
|
|
3109 static wordnode_T *wordtree_alloc __ARGS((sblock_T **blp));
|
|
339
|
3110 static int store_word __ARGS((char_u *word, spellinfo_T *spin, int flags, int region, char_u *pfxlist));
|
|
|
3111 static int tree_add_word __ARGS((char_u *word, wordnode_T *tree, int flags, int region, int prefixID, sblock_T **blp));
|
|
310
|
3112 static void wordtree_compress __ARGS((wordnode_T *root, spellinfo_T *spin));
|
|
300
|
3113 static int node_compress __ARGS((wordnode_T *node, hashtab_T *ht, int *tot));
|
|
|
3114 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
316
|
3115 static void write_vim_spell __ARGS((char_u *fname, spellinfo_T *spin));
|
|
346
|
3116 static void clear_node __ARGS((wordnode_T *node));
|
|
|
3117 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
3118 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3119 static void init_spellfile __ARGS((void));
|
|
236
|
3120
|
|
|
3121 /*
|
|
323
|
3122 * Read the affix file "fname".
|
|
316
|
3123 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
3124 */
|
|
|
3125 static afffile_T *
|
|
300
|
3126 spell_read_aff(fname, spin)
|
|
236
|
3127 char_u *fname;
|
|
300
|
3128 spellinfo_T *spin;
|
|
236
|
3129 {
|
|
|
3130 FILE *fd;
|
|
|
3131 afffile_T *aff;
|
|
|
3132 char_u rline[MAXLINELEN];
|
|
|
3133 char_u *line;
|
|
|
3134 char_u *pc = NULL;
|
|
334
|
3135 #define MAXITEMCNT 7
|
|
|
3136 char_u *(items[MAXITEMCNT]);
|
|
236
|
3137 int itemcnt;
|
|
|
3138 char_u *p;
|
|
|
3139 int lnum = 0;
|
|
|
3140 affheader_T *cur_aff = NULL;
|
|
|
3141 int aff_todo = 0;
|
|
|
3142 hashtab_T *tp;
|
|
255
|
3143 char_u *low = NULL;
|
|
|
3144 char_u *fol = NULL;
|
|
|
3145 char_u *upp = NULL;
|
|
307
|
3146 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
3147 int do_rep;
|
|
|
3148 int do_sal;
|
|
|
3149 int do_map;
|
|
366
|
3150 int do_midword;
|
|
375
|
3151 int do_sofo;
|
|
323
|
3152 int found_map = FALSE;
|
|
324
|
3153 hashitem_T *hi;
|
|
236
|
3154
|
|
300
|
3155 /*
|
|
|
3156 * Open the file.
|
|
|
3157 */
|
|
310
|
3158 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3159 if (fd == NULL)
|
|
|
3160 {
|
|
|
3161 EMSG2(_(e_notopen), fname);
|
|
|
3162 return NULL;
|
|
|
3163 }
|
|
|
3164
|
|
310
|
3165 if (spin->si_verbose || p_verbose > 2)
|
|
|
3166 {
|
|
|
3167 if (!spin->si_verbose)
|
|
|
3168 verbose_enter();
|
|
366
|
3169 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
3170 out_flush();
|
|
|
3171 if (!spin->si_verbose)
|
|
|
3172 verbose_leave();
|
|
|
3173 }
|
|
236
|
3174
|
|
323
|
3175 /* Only do REP lines when not done in another .aff file already. */
|
|
|
3176 do_rep = spin->si_rep.ga_len == 0;
|
|
|
3177
|
|
|
3178 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
3179 do_sal = spin->si_sal.ga_len == 0;
|
|
|
3180
|
|
|
3181 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
3182 do_map = spin->si_map.ga_len == 0;
|
|
|
3183
|
|
366
|
3184 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
3185 do_midword = spin->si_midword == NULL;
|
|
|
3186
|
|
375
|
3187 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
3188 do_sofo = spin->si_sofofr == NULL;
|
|
|
3189
|
|
300
|
3190 /*
|
|
|
3191 * Allocate and init the afffile_T structure.
|
|
|
3192 */
|
|
|
3193 aff = (afffile_T *)getroom(&spin->si_blocks, sizeof(afffile_T));
|
|
236
|
3194 if (aff == NULL)
|
|
|
3195 return NULL;
|
|
|
3196 hash_init(&aff->af_pref);
|
|
|
3197 hash_init(&aff->af_suff);
|
|
|
3198
|
|
|
3199 /*
|
|
|
3200 * Read all the lines in the file one by one.
|
|
|
3201 */
|
|
255
|
3202 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
3203 {
|
|
255
|
3204 line_breakcheck();
|
|
236
|
3205 ++lnum;
|
|
|
3206
|
|
|
3207 /* Skip comment lines. */
|
|
|
3208 if (*rline == '#')
|
|
|
3209 continue;
|
|
|
3210
|
|
|
3211 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
3212 vim_free(pc);
|
|
310
|
3213 #ifdef FEAT_MBYTE
|
|
300
|
3214 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3215 {
|
|
300
|
3216 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
3217 if (pc == NULL)
|
|
|
3218 {
|
|
|
3219 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3220 fname, lnum, rline);
|
|
|
3221 continue;
|
|
|
3222 }
|
|
236
|
3223 line = pc;
|
|
|
3224 }
|
|
|
3225 else
|
|
310
|
3226 #endif
|
|
236
|
3227 {
|
|
|
3228 pc = NULL;
|
|
|
3229 line = rline;
|
|
|
3230 }
|
|
|
3231
|
|
|
3232 /* Split the line up in white separated items. Put a NUL after each
|
|
|
3233 * item. */
|
|
|
3234 itemcnt = 0;
|
|
|
3235 for (p = line; ; )
|
|
|
3236 {
|
|
|
3237 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
3238 ++p;
|
|
|
3239 if (*p == NUL)
|
|
|
3240 break;
|
|
334
|
3241 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
3242 break;
|
|
236
|
3243 items[itemcnt++] = p;
|
|
300
|
3244 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
3245 ++p;
|
|
|
3246 if (*p == NUL)
|
|
|
3247 break;
|
|
|
3248 *p++ = NUL;
|
|
|
3249 }
|
|
|
3250
|
|
|
3251 /* Handle non-empty lines. */
|
|
|
3252 if (itemcnt > 0)
|
|
|
3253 {
|
|
|
3254 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
3255 && aff->af_enc == NULL)
|
|
|
3256 {
|
|
310
|
3257 #ifdef FEAT_MBYTE
|
|
300
|
3258 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
3259 aff->af_enc = enc_canonize(items[1]);
|
|
|
3260 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
3261 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
3262 p_enc) == FAIL)
|
|
|
3263 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
3264 fname, aff->af_enc, p_enc);
|
|
375
|
3265 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3266 #else
|
|
|
3267 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3268 #endif
|
|
236
|
3269 }
|
|
366
|
3270 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
3271 {
|
|
|
3272 if (do_midword)
|
|
|
3273 spin->si_midword = vim_strsave(items[1]);
|
|
|
3274 }
|
|
302
|
3275 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
3276 {
|
|
323
|
3277 /* ignored, we always split */
|
|
302
|
3278 }
|
|
323
|
3279 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
3280 {
|
|
323
|
3281 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
3282 }
|
|
307
|
3283 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
3284 && aff->af_rar == 0)
|
|
|
3285 {
|
|
|
3286 aff->af_rar = items[1][0];
|
|
|
3287 if (items[1][1] != NUL)
|
|
|
3288 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3289 }
|
|
310
|
3290 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
3291 && aff->af_kep == 0)
|
|
307
|
3292 {
|
|
310
|
3293 aff->af_kep = items[1][0];
|
|
307
|
3294 if (items[1][1] != NUL)
|
|
|
3295 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3296 }
|
|
346
|
3297 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
3298 && aff->af_bad == 0)
|
|
|
3299 {
|
|
|
3300 aff->af_bad = items[1][0];
|
|
|
3301 if (items[1][1] != NUL)
|
|
|
3302 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3303 }
|
|
339
|
3304 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
3305 {
|
|
|
3306 aff->af_pfxpostpone = TRUE;
|
|
|
3307 }
|
|
236
|
3308 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3309 || STRCMP(items[0], "SFX") == 0)
|
|
|
3310 && aff_todo == 0
|
|
334
|
3311 && itemcnt >= 4)
|
|
236
|
3312 {
|
|
334
|
3313 /* Myspell allows extra text after the item, but that might
|
|
|
3314 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
3315 if (itemcnt > 4 && *items[4] != '#')
|
|
|
3316 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
3317 fname, lnum, items[4]);
|
|
|
3318
|
|
236
|
3319 /* New affix letter. */
|
|
300
|
3320 cur_aff = (affheader_T *)getroom(&spin->si_blocks,
|
|
|
3321 sizeof(affheader_T));
|
|
236
|
3322 if (cur_aff == NULL)
|
|
|
3323 break;
|
|
339
|
3324 cur_aff->ah_key[0] = *items[1]; /* TODO: multi-byte? */
|
|
236
|
3325 cur_aff->ah_key[1] = NUL;
|
|
|
3326 if (items[1][1] != NUL)
|
|
307
|
3327 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
3328 if (*items[2] == 'Y')
|
|
|
3329 cur_aff->ah_combine = TRUE;
|
|
300
|
3330 else if (*items[2] != 'N')
|
|
236
|
3331 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
3332 fname, lnum, items[2]);
|
|
339
|
3333
|
|
236
|
3334 if (*items[0] == 'P')
|
|
339
|
3335 {
|
|
236
|
3336 tp = &aff->af_pref;
|
|
339
|
3337 /* Use a new number in the .spl file later, to be able to
|
|
|
3338 * handle multiple .aff files. */
|
|
|
3339 if (aff->af_pfxpostpone)
|
|
344
|
3340 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
3341 }
|
|
236
|
3342 else
|
|
|
3343 tp = &aff->af_suff;
|
|
300
|
3344 aff_todo = atoi((char *)items[3]);
|
|
324
|
3345 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
3346 if (!HASHITEM_EMPTY(hi))
|
|
300
|
3347 {
|
|
236
|
3348 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
3349 fname, lnum, items[1]);
|
|
300
|
3350 aff_todo = 0;
|
|
|
3351 }
|
|
236
|
3352 else
|
|
|
3353 hash_add(tp, cur_aff->ah_key);
|
|
|
3354 }
|
|
|
3355 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3356 || STRCMP(items[0], "SFX") == 0)
|
|
|
3357 && aff_todo > 0
|
|
|
3358 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
3359 && itemcnt >= 5)
|
|
236
|
3360 {
|
|
|
3361 affentry_T *aff_entry;
|
|
366
|
3362 int rare = FALSE;
|
|
|
3363 int lasti = 5;
|
|
|
3364
|
|
|
3365 /* Check for "rare" after the other info. */
|
|
|
3366 if (itemcnt > 5 && STRICMP(items[5], "rare") == 0)
|
|
|
3367 {
|
|
|
3368 rare = TRUE;
|
|
|
3369 lasti = 6;
|
|
|
3370 }
|
|
236
|
3371
|
|
334
|
3372 /* Myspell allows extra text after the item, but that might
|
|
|
3373 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
3374 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
3375 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
366
|
3376 fname, lnum, items[lasti]);
|
|
334
|
3377
|
|
236
|
3378 /* New item for an affix letter. */
|
|
|
3379 --aff_todo;
|
|
300
|
3380 aff_entry = (affentry_T *)getroom(&spin->si_blocks,
|
|
|
3381 sizeof(affentry_T));
|
|
236
|
3382 if (aff_entry == NULL)
|
|
|
3383 break;
|
|
366
|
3384 aff_entry->ae_rare = rare;
|
|
240
|
3385
|
|
236
|
3386 if (STRCMP(items[2], "0") != 0)
|
|
300
|
3387 aff_entry->ae_chop = getroom_save(&spin->si_blocks,
|
|
|
3388 items[2]);
|
|
236
|
3389 if (STRCMP(items[3], "0") != 0)
|
|
300
|
3390 aff_entry->ae_add = getroom_save(&spin->si_blocks,
|
|
|
3391 items[3]);
|
|
236
|
3392
|
|
300
|
3393 /* Don't use an affix entry with non-ASCII characters when
|
|
|
3394 * "spin->si_ascii" is TRUE. */
|
|
|
3395 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
3396 || has_non_ascii(aff_entry->ae_add)))
|
|
|
3397 {
|
|
|
3398 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
3399 cur_aff->ah_first = aff_entry;
|
|
300
|
3400
|
|
|
3401 if (STRCMP(items[4], ".") != 0)
|
|
|
3402 {
|
|
|
3403 char_u buf[MAXLINELEN];
|
|
|
3404
|
|
|
3405 aff_entry->ae_cond = getroom_save(&spin->si_blocks,
|
|
|
3406 items[4]);
|
|
|
3407 if (*items[0] == 'P')
|
|
|
3408 sprintf((char *)buf, "^%s", items[4]);
|
|
|
3409 else
|
|
|
3410 sprintf((char *)buf, "%s$", items[4]);
|
|
|
3411 aff_entry->ae_prog = vim_regcomp(buf,
|
|
|
3412 RE_MAGIC + RE_STRING);
|
|
|
3413 }
|
|
339
|
3414
|
|
|
3415 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
3416 * for the condition. Use an existing one if possible. */
|
|
|
3417 if (*items[0] == 'P' && aff->af_pfxpostpone
|
|
|
3418 && aff_entry->ae_chop == NULL)
|
|
|
3419 {
|
|
|
3420 int idx;
|
|
|
3421 char_u **pp;
|
|
|
3422
|
|
|
3423 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
3424 --idx)
|
|
|
3425 {
|
|
|
3426 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
3427 if (str_equal(p, aff_entry->ae_cond))
|
|
|
3428 break;
|
|
|
3429 }
|
|
|
3430 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
3431 {
|
|
|
3432 /* Not found, add a new condition. */
|
|
|
3433 idx = spin->si_prefcond.ga_len++;
|
|
|
3434 pp = ((char_u **)spin->si_prefcond.ga_data) + idx;
|
|
|
3435 if (aff_entry->ae_cond == NULL)
|
|
|
3436 *pp = NULL;
|
|
|
3437 else
|
|
|
3438 *pp = getroom_save(&spin->si_blocks,
|
|
|
3439 aff_entry->ae_cond);
|
|
|
3440 }
|
|
|
3441
|
|
|
3442 /* Add the prefix to the prefix tree. */
|
|
|
3443 if (aff_entry->ae_add == NULL)
|
|
|
3444 p = (char_u *)"";
|
|
|
3445 else
|
|
|
3446 p = aff_entry->ae_add;
|
|
366
|
3447 tree_add_word(p, spin->si_prefroot, rare ? -2 : -1,
|
|
|
3448 idx, cur_aff->ah_newID, &spin->si_blocks);
|
|
339
|
3449 }
|
|
240
|
3450 }
|
|
236
|
3451 }
|
|
255
|
3452 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
3453 {
|
|
|
3454 if (fol != NULL)
|
|
|
3455 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
3456 fname, lnum);
|
|
|
3457 else
|
|
|
3458 fol = vim_strsave(items[1]);
|
|
|
3459 }
|
|
|
3460 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
3461 {
|
|
|
3462 if (low != NULL)
|
|
|
3463 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
3464 fname, lnum);
|
|
|
3465 else
|
|
|
3466 low = vim_strsave(items[1]);
|
|
|
3467 }
|
|
|
3468 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
3469 {
|
|
|
3470 if (upp != NULL)
|
|
|
3471 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
3472 fname, lnum);
|
|
|
3473 else
|
|
|
3474 upp = vim_strsave(items[1]);
|
|
|
3475 }
|
|
236
|
3476 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
3477 {
|
|
236
|
3478 /* Ignore REP count */;
|
|
323
|
3479 if (!isdigit(*items[1]))
|
|
|
3480 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
3481 fname, lnum);
|
|
|
3482 }
|
|
236
|
3483 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 3)
|
|
|
3484 {
|
|
|
3485 /* REP item */
|
|
323
|
3486 if (do_rep)
|
|
|
3487 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
3488 }
|
|
|
3489 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
3490 {
|
|
|
3491 /* MAP item or count */
|
|
|
3492 if (!found_map)
|
|
|
3493 {
|
|
|
3494 /* First line contains the count. */
|
|
|
3495 found_map = TRUE;
|
|
|
3496 if (!isdigit(*items[1]))
|
|
|
3497 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
3498 fname, lnum);
|
|
|
3499 }
|
|
|
3500 else if (do_map)
|
|
|
3501 {
|
|
346
|
3502 int c;
|
|
|
3503
|
|
|
3504 /* Check that every character appears only once. */
|
|
|
3505 for (p = items[1]; *p != NUL; )
|
|
|
3506 {
|
|
|
3507 #ifdef FEAT_MBYTE
|
|
|
3508 c = mb_ptr2char_adv(&p);
|
|
|
3509 #else
|
|
|
3510 c = *p++;
|
|
|
3511 #endif
|
|
|
3512 if ((spin->si_map.ga_len > 0
|
|
|
3513 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
3514 != NULL)
|
|
|
3515 || vim_strchr(p, c) != NULL)
|
|
|
3516 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
3517 fname, lnum);
|
|
|
3518 }
|
|
|
3519
|
|
323
|
3520 /* We simply concatenate all the MAP strings, separated by
|
|
|
3521 * slashes. */
|
|
|
3522 ga_concat(&spin->si_map, items[1]);
|
|
|
3523 ga_append(&spin->si_map, '/');
|
|
|
3524 }
|
|
|
3525 }
|
|
|
3526 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
3527 {
|
|
|
3528 if (do_sal)
|
|
|
3529 {
|
|
|
3530 /* SAL item (sounds-a-like)
|
|
|
3531 * Either one of the known keys or a from-to pair. */
|
|
|
3532 if (STRCMP(items[1], "followup") == 0)
|
|
|
3533 spin->si_followup = sal_to_bool(items[2]);
|
|
|
3534 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
3535 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
3536 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
3537 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
3538 else
|
|
|
3539 /* when "to" is "_" it means empty */
|
|
|
3540 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
3541 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
3542 : items[2]);
|
|
|
3543 }
|
|
236
|
3544 }
|
|
375
|
3545 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
3546 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
3547 {
|
|
|
3548 if (do_sofo)
|
|
|
3549 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
3550 }
|
|
|
3551 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
3552 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
3553 {
|
|
|
3554 if (do_sofo)
|
|
|
3555 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
3556 }
|
|
300
|
3557 else
|
|
236
|
3558 smsg((char_u *)_("Unrecognized item in %s line %d: %s"),
|
|
|
3559 fname, lnum, items[0]);
|
|
|
3560 }
|
|
|
3561 }
|
|
|
3562
|
|
375
|
3563 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
3564 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
3565 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
3566 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
3567 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
3568
|
|
255
|
3569 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
3570 {
|
|
351
|
3571 if (spin->si_clear_chartab)
|
|
|
3572 {
|
|
|
3573 /* Clear the char type tables, don't want to use any of the
|
|
|
3574 * currently used spell properties. */
|
|
|
3575 init_spell_chartab();
|
|
|
3576 spin->si_clear_chartab = FALSE;
|
|
|
3577 }
|
|
|
3578
|
|
316
|
3579 /*
|
|
|
3580 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
3581 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
3582 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
3583 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
3584 */
|
|
|
3585 if (!spin->si_ascii
|
|
|
3586 #ifdef FEAT_MBYTE
|
|
|
3587 && !enc_utf8
|
|
|
3588 #endif
|
|
|
3589 )
|
|
260
|
3590 {
|
|
|
3591 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
3592 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
3593 else
|
|
316
|
3594 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
3595 }
|
|
255
|
3596
|
|
|
3597 vim_free(fol);
|
|
|
3598 vim_free(low);
|
|
|
3599 vim_free(upp);
|
|
|
3600 }
|
|
|
3601
|
|
236
|
3602 vim_free(pc);
|
|
|
3603 fclose(fd);
|
|
|
3604 return aff;
|
|
|
3605 }
|
|
|
3606
|
|
|
3607 /*
|
|
339
|
3608 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
3609 * NULL as equal.
|
|
|
3610 */
|
|
|
3611 static int
|
|
|
3612 str_equal(s1, s2)
|
|
|
3613 char_u *s1;
|
|
|
3614 char_u *s2;
|
|
|
3615 {
|
|
|
3616 if (s1 == NULL || s2 == NULL)
|
|
|
3617 return s1 == s2;
|
|
|
3618 return STRCMP(s1, s2) == 0;
|
|
|
3619 }
|
|
|
3620
|
|
|
3621 /*
|
|
323
|
3622 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
3623 * They are stored case-folded.
|
|
|
3624 */
|
|
|
3625 static void
|
|
|
3626 add_fromto(spin, gap, from, to)
|
|
|
3627 spellinfo_T *spin;
|
|
|
3628 garray_T *gap;
|
|
|
3629 char_u *from;
|
|
|
3630 char_u *to;
|
|
|
3631 {
|
|
|
3632 fromto_T *ftp;
|
|
|
3633 char_u word[MAXWLEN];
|
|
|
3634
|
|
|
3635 if (ga_grow(gap, 1) == OK)
|
|
|
3636 {
|
|
|
3637 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
3638 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
|
3639 ftp->ft_from = getroom_save(&spin->si_blocks, word);
|
|
|
3640 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
|
3641 ftp->ft_to = getroom_save(&spin->si_blocks, word);
|
|
|
3642 ++gap->ga_len;
|
|
|
3643 }
|
|
|
3644 }
|
|
|
3645
|
|
|
3646 /*
|
|
|
3647 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
3648 */
|
|
|
3649 static int
|
|
|
3650 sal_to_bool(s)
|
|
|
3651 char_u *s;
|
|
|
3652 {
|
|
|
3653 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
3654 }
|
|
|
3655
|
|
|
3656 /*
|
|
240
|
3657 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
3658 * When "s" is NULL FALSE is returned.
|
|
|
3659 */
|
|
|
3660 static int
|
|
|
3661 has_non_ascii(s)
|
|
|
3662 char_u *s;
|
|
|
3663 {
|
|
|
3664 char_u *p;
|
|
|
3665
|
|
|
3666 if (s != NULL)
|
|
|
3667 for (p = s; *p != NUL; ++p)
|
|
|
3668 if (*p >= 128)
|
|
|
3669 return TRUE;
|
|
|
3670 return FALSE;
|
|
|
3671 }
|
|
|
3672
|
|
|
3673 /*
|
|
236
|
3674 * Free the structure filled by spell_read_aff().
|
|
|
3675 */
|
|
|
3676 static void
|
|
|
3677 spell_free_aff(aff)
|
|
|
3678 afffile_T *aff;
|
|
|
3679 {
|
|
|
3680 hashtab_T *ht;
|
|
|
3681 hashitem_T *hi;
|
|
|
3682 int todo;
|
|
|
3683 affheader_T *ah;
|
|
300
|
3684 affentry_T *ae;
|
|
236
|
3685
|
|
|
3686 vim_free(aff->af_enc);
|
|
|
3687
|
|
339
|
3688 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
3689 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
3690 {
|
|
|
3691 todo = ht->ht_used;
|
|
|
3692 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
3693 {
|
|
|
3694 if (!HASHITEM_EMPTY(hi))
|
|
|
3695 {
|
|
|
3696 --todo;
|
|
|
3697 ah = HI2AH(hi);
|
|
300
|
3698 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
3699 vim_free(ae->ae_prog);
|
|
236
|
3700 }
|
|
|
3701 }
|
|
|
3702 if (ht == &aff->af_suff)
|
|
|
3703 break;
|
|
|
3704 }
|
|
300
|
3705
|
|
236
|
3706 hash_clear(&aff->af_pref);
|
|
|
3707 hash_clear(&aff->af_suff);
|
|
|
3708 }
|
|
|
3709
|
|
|
3710 /*
|
|
300
|
3711 * Read dictionary file "fname".
|
|
236
|
3712 * Returns OK or FAIL;
|
|
|
3713 */
|
|
|
3714 static int
|
|
300
|
3715 spell_read_dic(fname, spin, affile)
|
|
236
|
3716 char_u *fname;
|
|
300
|
3717 spellinfo_T *spin;
|
|
|
3718 afffile_T *affile;
|
|
236
|
3719 {
|
|
300
|
3720 hashtab_T ht;
|
|
236
|
3721 char_u line[MAXLINELEN];
|
|
300
|
3722 char_u *afflist;
|
|
339
|
3723 char_u *pfxlist;
|
|
300
|
3724 char_u *dw;
|
|
236
|
3725 char_u *pc;
|
|
|
3726 char_u *w;
|
|
|
3727 int l;
|
|
|
3728 hash_T hash;
|
|
|
3729 hashitem_T *hi;
|
|
|
3730 FILE *fd;
|
|
|
3731 int lnum = 1;
|
|
300
|
3732 int non_ascii = 0;
|
|
|
3733 int retval = OK;
|
|
|
3734 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
3735 int flags;
|
|
236
|
3736
|
|
300
|
3737 /*
|
|
|
3738 * Open the file.
|
|
|
3739 */
|
|
310
|
3740 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3741 if (fd == NULL)
|
|
|
3742 {
|
|
|
3743 EMSG2(_(e_notopen), fname);
|
|
|
3744 return FAIL;
|
|
|
3745 }
|
|
|
3746
|
|
300
|
3747 /* The hashtable is only used to detect duplicated words. */
|
|
|
3748 hash_init(&ht);
|
|
|
3749
|
|
334
|
3750 spin->si_foldwcount = 0;
|
|
|
3751 spin->si_keepwcount = 0;
|
|
|
3752
|
|
310
|
3753 if (spin->si_verbose || p_verbose > 2)
|
|
|
3754 {
|
|
|
3755 if (!spin->si_verbose)
|
|
|
3756 verbose_enter();
|
|
366
|
3757 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
3758 out_flush();
|
|
|
3759 if (!spin->si_verbose)
|
|
|
3760 verbose_leave();
|
|
|
3761 }
|
|
236
|
3762
|
|
|
3763 /* Read and ignore the first line: word count. */
|
|
|
3764 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
3765 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
3766 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
3767
|
|
|
3768 /*
|
|
|
3769 * Read all the lines in the file one by one.
|
|
|
3770 * The words are converted to 'encoding' here, before being added to
|
|
|
3771 * the hashtable.
|
|
|
3772 */
|
|
255
|
3773 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
3774 {
|
|
255
|
3775 line_breakcheck();
|
|
236
|
3776 ++lnum;
|
|
351
|
3777 if (line[0] == '#')
|
|
|
3778 continue; /* comment line */
|
|
236
|
3779
|
|
300
|
3780 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
3781 * the word is kept to allow e.g., "et al.". */
|
|
236
|
3782 l = STRLEN(line);
|
|
|
3783 while (l > 0 && line[l - 1] <= ' ')
|
|
|
3784 --l;
|
|
|
3785 if (l == 0)
|
|
|
3786 continue; /* empty line */
|
|
|
3787 line[l] = NUL;
|
|
|
3788
|
|
|
3789 /* Find the optional affix names. */
|
|
300
|
3790 afflist = vim_strchr(line, '/');
|
|
|
3791 if (afflist != NULL)
|
|
|
3792 *afflist++ = NUL;
|
|
236
|
3793
|
|
300
|
3794 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
3795 if (spin->si_ascii && has_non_ascii(line))
|
|
|
3796 {
|
|
|
3797 ++non_ascii;
|
|
240
|
3798 continue;
|
|
300
|
3799 }
|
|
240
|
3800
|
|
310
|
3801 #ifdef FEAT_MBYTE
|
|
236
|
3802 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
3803 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3804 {
|
|
300
|
3805 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
3806 if (pc == NULL)
|
|
|
3807 {
|
|
|
3808 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3809 fname, lnum, line);
|
|
|
3810 continue;
|
|
|
3811 }
|
|
236
|
3812 w = pc;
|
|
|
3813 }
|
|
|
3814 else
|
|
310
|
3815 #endif
|
|
236
|
3816 {
|
|
|
3817 pc = NULL;
|
|
|
3818 w = line;
|
|
|
3819 }
|
|
|
3820
|
|
339
|
3821 /* This takes time, print a message now and then. */
|
|
|
3822 if (spin->si_verbose && (lnum & 0x3ff) == 0)
|
|
|
3823 {
|
|
|
3824 vim_snprintf((char *)message, sizeof(message),
|
|
|
3825 _("line %6d, word %6d - %s"),
|
|
|
3826 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
3827 msg_start();
|
|
|
3828 msg_puts_long_attr(message, 0);
|
|
|
3829 msg_clr_eos();
|
|
|
3830 msg_didout = FALSE;
|
|
|
3831 msg_col = 0;
|
|
|
3832 out_flush();
|
|
|
3833 }
|
|
|
3834
|
|
300
|
3835 /* Store the word in the hashtable to be able to find duplicates. */
|
|
|
3836 dw = (char_u *)getroom_save(&spin->si_blocks, w);
|
|
236
|
3837 if (dw == NULL)
|
|
300
|
3838 retval = FAIL;
|
|
|
3839 vim_free(pc);
|
|
|
3840 if (retval == FAIL)
|
|
236
|
3841 break;
|
|
|
3842
|
|
300
|
3843 hash = hash_hash(dw);
|
|
|
3844 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
3845 if (!HASHITEM_EMPTY(hi))
|
|
|
3846 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
3847 fname, lnum, dw);
|
|
236
|
3848 else
|
|
300
|
3849 hash_add_item(&ht, hi, dw, hash);
|
|
|
3850
|
|
307
|
3851 flags = 0;
|
|
339
|
3852 pfxlist = NULL;
|
|
307
|
3853 if (afflist != NULL)
|
|
|
3854 {
|
|
|
3855 /* Check for affix name that stands for keep-case word and stands
|
|
|
3856 * for rare word (if defined). */
|
|
310
|
3857 if (affile->af_kep != NUL
|
|
|
3858 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
389
|
3859 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
307
|
3860 if (affile->af_rar != NUL
|
|
|
3861 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
3862 flags |= WF_RARE;
|
|
346
|
3863 if (affile->af_bad != NUL
|
|
|
3864 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
3865 flags |= WF_BANNED;
|
|
339
|
3866
|
|
|
3867 if (affile->af_pfxpostpone)
|
|
|
3868 /* Need to store the list of prefix IDs with the word. */
|
|
|
3869 pfxlist = get_pfxlist(affile, afflist, &spin->si_blocks);
|
|
307
|
3870 }
|
|
|
3871
|
|
300
|
3872 /* Add the word to the word tree(s). */
|
|
339
|
3873 if (store_word(dw, spin, flags, spin->si_region, pfxlist) == FAIL)
|
|
300
|
3874 retval = FAIL;
|
|
236
|
3875
|
|
300
|
3876 if (afflist != NULL)
|
|
|
3877 {
|
|
|
3878 /* Find all matching suffixes and add the resulting words.
|
|
|
3879 * Additionally do matching prefixes that combine. */
|
|
339
|
3880 if (store_aff_word(dw, spin, afflist, affile,
|
|
307
|
3881 &affile->af_suff, &affile->af_pref,
|
|
339
|
3882 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
3883 retval = FAIL;
|
|
|
3884
|
|
|
3885 /* Find all matching prefixes and add the resulting words. */
|
|
339
|
3886 if (store_aff_word(dw, spin, afflist, affile,
|
|
|
3887 &affile->af_pref, NULL,
|
|
|
3888 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
3889 retval = FAIL;
|
|
|
3890 }
|
|
236
|
3891 }
|
|
|
3892
|
|
300
|
3893 if (spin->si_ascii && non_ascii > 0)
|
|
|
3894 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
3895 non_ascii);
|
|
|
3896 hash_clear(&ht);
|
|
|
3897
|
|
236
|
3898 fclose(fd);
|
|
300
|
3899 return retval;
|
|
236
|
3900 }
|
|
|
3901
|
|
|
3902 /*
|
|
339
|
3903 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
3904 * Used for PFXPOSTPONE.
|
|
|
3905 * Returns a string allocated with getroom(). NULL when there are no prefixes
|
|
|
3906 * or when out of memory.
|
|
|
3907 */
|
|
|
3908 static char_u *
|
|
|
3909 get_pfxlist(affile, afflist, blp)
|
|
|
3910 afffile_T *affile;
|
|
|
3911 char_u *afflist;
|
|
|
3912 sblock_T **blp;
|
|
|
3913 {
|
|
|
3914 char_u *p;
|
|
|
3915 int cnt;
|
|
|
3916 int round;
|
|
|
3917 char_u *res = NULL;
|
|
|
3918 char_u key[2];
|
|
|
3919 hashitem_T *hi;
|
|
|
3920
|
|
|
3921 key[1] = NUL;
|
|
|
3922
|
|
|
3923 /* round 1: count the number of prefix IDs.
|
|
|
3924 * round 2: move prefix IDs to "res" */
|
|
|
3925 for (round = 1; round <= 2; ++round)
|
|
|
3926 {
|
|
|
3927 cnt = 0;
|
|
|
3928 for (p = afflist; *p != NUL; ++p)
|
|
|
3929 {
|
|
|
3930 key[0] = *p;
|
|
|
3931 hi = hash_find(&affile->af_pref, key);
|
|
|
3932 if (!HASHITEM_EMPTY(hi))
|
|
|
3933 {
|
|
|
3934 /* This is a prefix ID, use the new number. */
|
|
|
3935 if (round == 2)
|
|
|
3936 res[cnt] = HI2AH(hi)->ah_newID;
|
|
|
3937 ++cnt;
|
|
|
3938 }
|
|
|
3939 }
|
|
|
3940 if (round == 1 && cnt > 0)
|
|
|
3941 res = getroom(blp, cnt + 1);
|
|
|
3942 if (res == NULL)
|
|
|
3943 break;
|
|
|
3944 }
|
|
|
3945
|
|
|
3946 if (res != NULL)
|
|
|
3947 res[cnt] = NUL;
|
|
|
3948 return res;
|
|
|
3949 }
|
|
|
3950
|
|
|
3951 /*
|
|
300
|
3952 * Apply affixes to a word and store the resulting words.
|
|
|
3953 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
3954 * prefixes or suffixes.
|
|
|
3955 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
3956 * the resulting words for combining affixes.
|
|
|
3957 *
|
|
|
3958 * Returns FAIL when out of memory.
|
|
236
|
3959 */
|
|
300
|
3960 static int
|
|
339
|
3961 store_aff_word(word, spin, afflist, affile, ht, xht, comb, flags, pfxlist)
|
|
300
|
3962 char_u *word; /* basic word start */
|
|
|
3963 spellinfo_T *spin; /* spell info */
|
|
|
3964 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
3965 afffile_T *affile;
|
|
300
|
3966 hashtab_T *ht;
|
|
|
3967 hashtab_T *xht;
|
|
|
3968 int comb; /* only use affixes that combine */
|
|
307
|
3969 int flags; /* flags for the word */
|
|
339
|
3970 char_u *pfxlist; /* list of prefix IDs */
|
|
236
|
3971 {
|
|
|
3972 int todo;
|
|
|
3973 hashitem_T *hi;
|
|
300
|
3974 affheader_T *ah;
|
|
|
3975 affentry_T *ae;
|
|
|
3976 regmatch_T regmatch;
|
|
|
3977 char_u newword[MAXWLEN];
|
|
|
3978 int retval = OK;
|
|
|
3979 int i;
|
|
|
3980 char_u *p;
|
|
366
|
3981 int use_flags;
|
|
236
|
3982
|
|
300
|
3983 todo = ht->ht_used;
|
|
|
3984 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
3985 {
|
|
|
3986 if (!HASHITEM_EMPTY(hi))
|
|
|
3987 {
|
|
|
3988 --todo;
|
|
300
|
3989 ah = HI2AH(hi);
|
|
236
|
3990
|
|
300
|
3991 /* Check that the affix combines, if required, and that the word
|
|
|
3992 * supports this affix. */
|
|
|
3993 if ((!comb || ah->ah_combine)
|
|
|
3994 && vim_strchr(afflist, *ah->ah_key) != NULL)
|
|
236
|
3995 {
|
|
300
|
3996 /* Loop over all affix entries with this name. */
|
|
|
3997 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
3998 {
|
|
300
|
3999 /* Check the condition. It's not logical to match case
|
|
|
4000 * here, but it is required for compatibility with
|
|
339
|
4001 * Myspell.
|
|
|
4002 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
4003 * prefixes with a chop string. */
|
|
300
|
4004 regmatch.regprog = ae->ae_prog;
|
|
|
4005 regmatch.rm_ic = FALSE;
|
|
339
|
4006 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
4007 || ae->ae_chop != NULL)
|
|
|
4008 && (ae->ae_prog == NULL
|
|
|
4009 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
4010 {
|
|
|
4011 /* Match. Remove the chop and add the affix. */
|
|
|
4012 if (xht == NULL)
|
|
240
|
4013 {
|
|
300
|
4014 /* prefix: chop/add at the start of the word */
|
|
|
4015 if (ae->ae_add == NULL)
|
|
|
4016 *newword = NUL;
|
|
|
4017 else
|
|
|
4018 STRCPY(newword, ae->ae_add);
|
|
|
4019 p = word;
|
|
|
4020 if (ae->ae_chop != NULL)
|
|
310
|
4021 {
|
|
300
|
4022 /* Skip chop string. */
|
|
310
|
4023 #ifdef FEAT_MBYTE
|
|
|
4024 if (has_mbyte)
|
|
324
|
4025 {
|
|
310
|
4026 i = mb_charlen(ae->ae_chop);
|
|
324
|
4027 for ( ; i > 0; --i)
|
|
|
4028 mb_ptr_adv(p);
|
|
|
4029 }
|
|
310
|
4030 else
|
|
|
4031 #endif
|
|
324
|
4032 p += STRLEN(ae->ae_chop);
|
|
310
|
4033 }
|
|
300
|
4034 STRCAT(newword, p);
|
|
|
4035 }
|
|
|
4036 else
|
|
|
4037 {
|
|
|
4038 /* suffix: chop/add at the end of the word */
|
|
|
4039 STRCPY(newword, word);
|
|
|
4040 if (ae->ae_chop != NULL)
|
|
|
4041 {
|
|
|
4042 /* Remove chop string. */
|
|
|
4043 p = newword + STRLEN(newword);
|
|
376
|
4044 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
4045 for ( ; i > 0; --i)
|
|
300
|
4046 mb_ptr_back(newword, p);
|
|
|
4047 *p = NUL;
|
|
|
4048 }
|
|
|
4049 if (ae->ae_add != NULL)
|
|
|
4050 STRCAT(newword, ae->ae_add);
|
|
240
|
4051 }
|
|
|
4052
|
|
366
|
4053 /* Obey the "rare" flag of the affix. */
|
|
|
4054 if (ae->ae_rare)
|
|
|
4055 use_flags = flags | WF_RARE;
|
|
|
4056 else
|
|
|
4057 use_flags = flags;
|
|
|
4058
|
|
300
|
4059 /* Store the modified word. */
|
|
366
|
4060 if (store_word(newword, spin, use_flags,
|
|
|
4061 spin->si_region, pfxlist) == FAIL)
|
|
300
|
4062 retval = FAIL;
|
|
236
|
4063
|
|
300
|
4064 /* When added a suffix and combining is allowed also
|
|
|
4065 * try adding prefixes additionally. */
|
|
|
4066 if (xht != NULL && ah->ah_combine)
|
|
339
|
4067 if (store_aff_word(newword, spin, afflist, affile,
|
|
366
|
4068 xht, NULL, TRUE, use_flags, pfxlist)
|
|
|
4069 == FAIL)
|
|
300
|
4070 retval = FAIL;
|
|
236
|
4071 }
|
|
|
4072 }
|
|
|
4073 }
|
|
|
4074 }
|
|
|
4075 }
|
|
|
4076
|
|
|
4077 return retval;
|
|
|
4078 }
|
|
|
4079
|
|
|
4080 /*
|
|
300
|
4081 * Read a file with a list of words.
|
|
236
|
4082 */
|
|
|
4083 static int
|
|
300
|
4084 spell_read_wordfile(fname, spin)
|
|
|
4085 char_u *fname;
|
|
|
4086 spellinfo_T *spin;
|
|
236
|
4087 {
|
|
300
|
4088 FILE *fd;
|
|
|
4089 long lnum = 0;
|
|
|
4090 char_u rline[MAXLINELEN];
|
|
|
4091 char_u *line;
|
|
|
4092 char_u *pc = NULL;
|
|
381
|
4093 char_u *p;
|
|
300
|
4094 int l;
|
|
|
4095 int retval = OK;
|
|
|
4096 int did_word = FALSE;
|
|
|
4097 int non_ascii = 0;
|
|
307
|
4098 int flags;
|
|
316
|
4099 int regionmask;
|
|
236
|
4100
|
|
300
|
4101 /*
|
|
|
4102 * Open the file.
|
|
|
4103 */
|
|
310
|
4104 fd = mch_fopen((char *)fname, "r");
|
|
300
|
4105 if (fd == NULL)
|
|
236
|
4106 {
|
|
300
|
4107 EMSG2(_(e_notopen), fname);
|
|
|
4108 return FAIL;
|
|
236
|
4109 }
|
|
|
4110
|
|
310
|
4111 if (spin->si_verbose || p_verbose > 2)
|
|
|
4112 {
|
|
|
4113 if (!spin->si_verbose)
|
|
|
4114 verbose_enter();
|
|
366
|
4115 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
4116 out_flush();
|
|
|
4117 if (!spin->si_verbose)
|
|
|
4118 verbose_leave();
|
|
|
4119 }
|
|
300
|
4120
|
|
|
4121 /*
|
|
|
4122 * Read all the lines in the file one by one.
|
|
|
4123 */
|
|
|
4124 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
4125 {
|
|
|
4126 line_breakcheck();
|
|
|
4127 ++lnum;
|
|
|
4128
|
|
|
4129 /* Skip comment lines. */
|
|
|
4130 if (*rline == '#')
|
|
|
4131 continue;
|
|
|
4132
|
|
|
4133 /* Remove CR, LF and white space from the end. */
|
|
|
4134 l = STRLEN(rline);
|
|
|
4135 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
4136 --l;
|
|
|
4137 if (l == 0)
|
|
|
4138 continue; /* empty or blank line */
|
|
|
4139 rline[l] = NUL;
|
|
|
4140
|
|
|
4141 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
4142 vim_free(pc);
|
|
310
|
4143 #ifdef FEAT_MBYTE
|
|
300
|
4144 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
4145 {
|
|
|
4146 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
4147 if (pc == NULL)
|
|
|
4148 {
|
|
|
4149 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4150 fname, lnum, rline);
|
|
|
4151 continue;
|
|
|
4152 }
|
|
|
4153 line = pc;
|
|
|
4154 }
|
|
|
4155 else
|
|
310
|
4156 #endif
|
|
300
|
4157 {
|
|
|
4158 pc = NULL;
|
|
|
4159 line = rline;
|
|
|
4160 }
|
|
|
4161
|
|
307
|
4162 if (*line == '/')
|
|
300
|
4163 {
|
|
307
|
4164 ++line;
|
|
|
4165 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
4166 {
|
|
|
4167 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
4168 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
4169 fname, lnum, line - 1);
|
|
300
|
4170 else if (did_word)
|
|
316
|
4171 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
4172 fname, lnum, line - 1);
|
|
300
|
4173 else
|
|
|
4174 {
|
|
310
|
4175 #ifdef FEAT_MBYTE
|
|
|
4176 char_u *enc;
|
|
|
4177
|
|
300
|
4178 /* Setup for conversion to 'encoding'. */
|
|
316
|
4179 line += 10;
|
|
|
4180 enc = enc_canonize(line);
|
|
300
|
4181 if (enc != NULL && !spin->si_ascii
|
|
|
4182 && convert_setup(&spin->si_conv, enc,
|
|
|
4183 p_enc) == FAIL)
|
|
|
4184 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
4185 fname, line, p_enc);
|
|
300
|
4186 vim_free(enc);
|
|
375
|
4187 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4188 #else
|
|
|
4189 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4190 #endif
|
|
300
|
4191 }
|
|
307
|
4192 continue;
|
|
300
|
4193 }
|
|
307
|
4194
|
|
316
|
4195 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
4196 {
|
|
|
4197 if (spin->si_region_count > 1)
|
|
|
4198 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
4199 fname, lnum, line);
|
|
|
4200 else
|
|
|
4201 {
|
|
|
4202 line += 8;
|
|
|
4203 if (STRLEN(line) > 16)
|
|
|
4204 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
4205 fname, lnum, line);
|
|
|
4206 else
|
|
|
4207 {
|
|
|
4208 spin->si_region_count = STRLEN(line) / 2;
|
|
|
4209 STRCPY(spin->si_region_name, line);
|
|
389
|
4210
|
|
|
4211 /* Adjust the mask for a word valid in all regions. */
|
|
|
4212 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
4213 }
|
|
|
4214 }
|
|
|
4215 continue;
|
|
|
4216 }
|
|
|
4217
|
|
381
|
4218 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
4219 fname, lnum, line - 1);
|
|
|
4220 continue;
|
|
|
4221 }
|
|
|
4222
|
|
|
4223 flags = 0;
|
|
|
4224 regionmask = spin->si_region;
|
|
|
4225
|
|
|
4226 /* Check for flags and region after a slash. */
|
|
|
4227 p = vim_strchr(line, '/');
|
|
|
4228 if (p != NULL)
|
|
|
4229 {
|
|
|
4230 *p++ = NUL;
|
|
|
4231 while (*p != NUL)
|
|
|
4232 {
|
|
|
4233 if (*p == '=') /* keep-case word */
|
|
389
|
4234 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
4235 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
4236 flags |= WF_BANNED;
|
|
|
4237 else if (*p == '?') /* Rare word. */
|
|
|
4238 flags |= WF_RARE;
|
|
|
4239 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
4240 {
|
|
381
|
4241 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
4242 regionmask = 0;
|
|
|
4243 flags |= WF_REGION;
|
|
|
4244
|
|
|
4245 l = *p - '0';
|
|
316
|
4246 if (l > spin->si_region_count)
|
|
|
4247 {
|
|
|
4248 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
4249 fname, lnum, p);
|
|
316
|
4250 break;
|
|
|
4251 }
|
|
|
4252 regionmask |= 1 << (l - 1);
|
|
|
4253 }
|
|
381
|
4254 else
|
|
|
4255 {
|
|
|
4256 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
4257 fname, lnum, p);
|
|
|
4258 break;
|
|
|
4259 }
|
|
|
4260 ++p;
|
|
307
|
4261 }
|
|
300
|
4262 }
|
|
|
4263
|
|
|
4264 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4265 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4266 {
|
|
|
4267 ++non_ascii;
|
|
|
4268 continue;
|
|
|
4269 }
|
|
|
4270
|
|
|
4271 /* Normal word: store it. */
|
|
339
|
4272 if (store_word(line, spin, flags, regionmask, NULL) == FAIL)
|
|
300
|
4273 {
|
|
|
4274 retval = FAIL;
|
|
|
4275 break;
|
|
|
4276 }
|
|
|
4277 did_word = TRUE;
|
|
|
4278 }
|
|
|
4279
|
|
|
4280 vim_free(pc);
|
|
|
4281 fclose(fd);
|
|
|
4282
|
|
310
|
4283 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
4284 {
|
|
|
4285 if (p_verbose > 2)
|
|
|
4286 verbose_enter();
|
|
300
|
4287 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4288 non_ascii);
|
|
310
|
4289 if (p_verbose > 2)
|
|
|
4290 verbose_leave();
|
|
|
4291 }
|
|
300
|
4292 return retval;
|
|
236
|
4293 }
|
|
|
4294
|
|
|
4295 /*
|
|
300
|
4296 * Get part of an sblock_T, "len" bytes long.
|
|
|
4297 * This avoids calling free() for every little struct we use.
|
|
|
4298 * The memory is cleared to all zeros.
|
|
|
4299 * Returns NULL when out of memory.
|
|
|
4300 */
|
|
|
4301 static void *
|
|
|
4302 getroom(blp, len)
|
|
|
4303 sblock_T **blp;
|
|
|
4304 size_t len; /* length needed */
|
|
|
4305 {
|
|
|
4306 char_u *p;
|
|
|
4307 sblock_T *bl = *blp;
|
|
|
4308
|
|
|
4309 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
4310 {
|
|
|
4311 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
4312 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
4313 if (bl == NULL)
|
|
|
4314 return NULL;
|
|
|
4315 bl->sb_next = *blp;
|
|
|
4316 *blp = bl;
|
|
|
4317 bl->sb_used = 0;
|
|
|
4318 }
|
|
|
4319
|
|
|
4320 p = bl->sb_data + bl->sb_used;
|
|
|
4321 bl->sb_used += len;
|
|
|
4322
|
|
|
4323 return p;
|
|
|
4324 }
|
|
|
4325
|
|
|
4326 /*
|
|
|
4327 * Make a copy of a string into memory allocated with getroom().
|
|
|
4328 */
|
|
|
4329 static char_u *
|
|
|
4330 getroom_save(blp, s)
|
|
|
4331 sblock_T **blp;
|
|
|
4332 char_u *s;
|
|
|
4333 {
|
|
|
4334 char_u *sc;
|
|
|
4335
|
|
|
4336 sc = (char_u *)getroom(blp, STRLEN(s) + 1);
|
|
|
4337 if (sc != NULL)
|
|
|
4338 STRCPY(sc, s);
|
|
|
4339 return sc;
|
|
|
4340 }
|
|
|
4341
|
|
|
4342
|
|
|
4343 /*
|
|
|
4344 * Free the list of allocated sblock_T.
|
|
236
|
4345 */
|
|
|
4346 static void
|
|
300
|
4347 free_blocks(bl)
|
|
|
4348 sblock_T *bl;
|
|
236
|
4349 {
|
|
300
|
4350 sblock_T *next;
|
|
236
|
4351
|
|
300
|
4352 while (bl != NULL)
|
|
236
|
4353 {
|
|
300
|
4354 next = bl->sb_next;
|
|
|
4355 vim_free(bl);
|
|
|
4356 bl = next;
|
|
236
|
4357 }
|
|
|
4358 }
|
|
|
4359
|
|
|
4360 /*
|
|
300
|
4361 * Allocate the root of a word tree.
|
|
236
|
4362 */
|
|
300
|
4363 static wordnode_T *
|
|
|
4364 wordtree_alloc(blp)
|
|
|
4365 sblock_T **blp;
|
|
236
|
4366 {
|
|
300
|
4367 return (wordnode_T *)getroom(blp, sizeof(wordnode_T));
|
|
236
|
4368 }
|
|
|
4369
|
|
|
4370 /*
|
|
300
|
4371 * Store a word in the tree(s).
|
|
307
|
4372 * Always store it in the case-folded tree. A keep-case word can also be used
|
|
|
4373 * with all caps.
|
|
300
|
4374 * For a keep-case word also store it in the keep-case tree.
|
|
339
|
4375 * When "pfxlist" is not NULL store the word for each prefix ID.
|
|
236
|
4376 */
|
|
|
4377 static int
|
|
339
|
4378 store_word(word, spin, flags, region, pfxlist)
|
|
300
|
4379 char_u *word;
|
|
|
4380 spellinfo_T *spin;
|
|
307
|
4381 int flags; /* extra flags, WF_BANNED */
|
|
316
|
4382 int region; /* supported region(s) */
|
|
339
|
4383 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
236
|
4384 {
|
|
300
|
4385 int len = STRLEN(word);
|
|
|
4386 int ct = captype(word, word + len);
|
|
|
4387 char_u foldword[MAXWLEN];
|
|
339
|
4388 int res = OK;
|
|
|
4389 char_u *p;
|
|
236
|
4390
|
|
323
|
4391 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
4392 for (p = pfxlist; res == OK; ++p)
|
|
|
4393 {
|
|
|
4394 res = tree_add_word(foldword, spin->si_foldroot, ct | flags,
|
|
|
4395 region, p == NULL ? 0 : *p, &spin->si_blocks);
|
|
|
4396 if (p == NULL || *p == NUL)
|
|
|
4397 break;
|
|
|
4398 }
|
|
334
|
4399 ++spin->si_foldwcount;
|
|
307
|
4400
|
|
|
4401 if (res == OK && (ct == WF_KEEPCAP || flags & WF_KEEPCAP))
|
|
334
|
4402 {
|
|
339
|
4403 for (p = pfxlist; res == OK; ++p)
|
|
|
4404 {
|
|
|
4405 res = tree_add_word(word, spin->si_keeproot, flags,
|
|
|
4406 region, p == NULL ? 0 : *p, &spin->si_blocks);
|
|
|
4407 if (p == NULL || *p == NUL)
|
|
|
4408 break;
|
|
|
4409 }
|
|
334
|
4410 ++spin->si_keepwcount;
|
|
|
4411 }
|
|
300
|
4412 return res;
|
|
236
|
4413 }
|
|
|
4414
|
|
|
4415 /*
|
|
300
|
4416 * Add word "word" to a word tree at "root".
|
|
366
|
4417 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
4418 * "rare" and "region" is the condition nr.
|
|
255
|
4419 * Returns FAIL when out of memory.
|
|
236
|
4420 */
|
|
255
|
4421 static int
|
|
339
|
4422 tree_add_word(word, root, flags, region, prefixID, blp)
|
|
300
|
4423 char_u *word;
|
|
|
4424 wordnode_T *root;
|
|
|
4425 int flags;
|
|
|
4426 int region;
|
|
339
|
4427 int prefixID;
|
|
300
|
4428 sblock_T **blp;
|
|
236
|
4429 {
|
|
300
|
4430 wordnode_T *node = root;
|
|
|
4431 wordnode_T *np;
|
|
|
4432 wordnode_T **prev = NULL;
|
|
|
4433 int i;
|
|
255
|
4434
|
|
300
|
4435 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
4436 for (i = 0; ; ++i)
|
|
255
|
4437 {
|
|
300
|
4438 /* Look for the sibling that has the same character. They are sorted
|
|
|
4439 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
4440 * higher byte value. For zero bytes (end of word) the sorting is
|
|
|
4441 * done on flags and then on prefixID
|
|
300
|
4442 */
|
|
339
|
4443 while (node != NULL
|
|
|
4444 && (node->wn_byte < word[i]
|
|
|
4445 || (node->wn_byte == NUL
|
|
|
4446 && (flags < 0
|
|
|
4447 ? node->wn_prefixID < prefixID
|
|
|
4448 : node->wn_flags < (flags & 0xff)
|
|
|
4449 || (node->wn_flags == (flags & 0xff)
|
|
|
4450 && node->wn_prefixID < prefixID)))))
|
|
236
|
4451 {
|
|
300
|
4452 prev = &node->wn_sibling;
|
|
|
4453 node = *prev;
|
|
236
|
4454 }
|
|
339
|
4455 if (node == NULL
|
|
|
4456 || node->wn_byte != word[i]
|
|
|
4457 || (word[i] == NUL
|
|
|
4458 && (flags < 0
|
|
|
4459 || node->wn_flags != (flags & 0xff)
|
|
|
4460 || node->wn_prefixID != prefixID)))
|
|
255
|
4461 {
|
|
300
|
4462 /* Allocate a new node. */
|
|
|
4463 np = (wordnode_T *)getroom(blp, sizeof(wordnode_T));
|
|
|
4464 if (np == NULL)
|
|
|
4465 return FAIL;
|
|
|
4466 np->wn_byte = word[i];
|
|
|
4467 *prev = np;
|
|
|
4468 np->wn_sibling = node;
|
|
|
4469 node = np;
|
|
255
|
4470 }
|
|
300
|
4471
|
|
|
4472 if (word[i] == NUL)
|
|
|
4473 {
|
|
|
4474 node->wn_flags = flags;
|
|
|
4475 node->wn_region |= region;
|
|
339
|
4476 node->wn_prefixID = prefixID;
|
|
300
|
4477 break;
|
|
|
4478 }
|
|
|
4479 prev = &node->wn_child;
|
|
|
4480 node = *prev;
|
|
255
|
4481 }
|
|
|
4482
|
|
|
4483 return OK;
|
|
236
|
4484 }
|
|
|
4485
|
|
|
4486 /*
|
|
300
|
4487 * Compress a tree: find tails that are identical and can be shared.
|
|
|
4488 */
|
|
|
4489 static void
|
|
310
|
4490 wordtree_compress(root, spin)
|
|
300
|
4491 wordnode_T *root;
|
|
310
|
4492 spellinfo_T *spin;
|
|
300
|
4493 {
|
|
|
4494 hashtab_T ht;
|
|
|
4495 int n;
|
|
|
4496 int tot = 0;
|
|
|
4497
|
|
|
4498 if (root != NULL)
|
|
|
4499 {
|
|
|
4500 hash_init(&ht);
|
|
|
4501 n = node_compress(root, &ht, &tot);
|
|
310
|
4502 if (spin->si_verbose || p_verbose > 2)
|
|
|
4503 {
|
|
|
4504 if (!spin->si_verbose)
|
|
|
4505 verbose_enter();
|
|
|
4506 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
300
|
4507 n, tot, (tot - n) * 100 / tot);
|
|
310
|
4508 if (p_verbose > 2)
|
|
|
4509 verbose_leave();
|
|
|
4510 }
|
|
300
|
4511 hash_clear(&ht);
|
|
|
4512 }
|
|
|
4513 }
|
|
|
4514
|
|
|
4515 /*
|
|
|
4516 * Compress a node, its siblings and its children, depth first.
|
|
|
4517 * Returns the number of compressed nodes.
|
|
236
|
4518 */
|
|
255
|
4519 static int
|
|
300
|
4520 node_compress(node, ht, tot)
|
|
|
4521 wordnode_T *node;
|
|
|
4522 hashtab_T *ht;
|
|
|
4523 int *tot; /* total count of nodes before compressing,
|
|
|
4524 incremented while going through the tree */
|
|
236
|
4525 {
|
|
300
|
4526 wordnode_T *np;
|
|
|
4527 wordnode_T *tp;
|
|
|
4528 wordnode_T *child;
|
|
|
4529 hash_T hash;
|
|
236
|
4530 hashitem_T *hi;
|
|
300
|
4531 int len = 0;
|
|
|
4532 unsigned nr, n;
|
|
|
4533 int compressed = 0;
|
|
236
|
4534
|
|
300
|
4535 /*
|
|
|
4536 * Go through the list of siblings. Compress each child and then try
|
|
|
4537 * finding an identical child to replace it.
|
|
|
4538 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
4539 * but the whole list of siblings, of which the node is the first.
|
|
|
4540 */
|
|
|
4541 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4542 {
|
|
300
|
4543 ++len;
|
|
|
4544 if ((child = np->wn_child) != NULL)
|
|
|
4545 {
|
|
346
|
4546 /* Compress the child. This fills hashkey. */
|
|
300
|
4547 compressed += node_compress(child, ht, tot);
|
|
|
4548
|
|
|
4549 /* Try to find an identical child. */
|
|
346
|
4550 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
4551 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
4552 tp = NULL;
|
|
|
4553 if (!HASHITEM_EMPTY(hi))
|
|
|
4554 {
|
|
|
4555 /* There are children with an identical hash value. Now check
|
|
|
4556 * if there is one that is really identical. */
|
|
346
|
4557 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
4558 if (node_equal(child, tp))
|
|
|
4559 {
|
|
|
4560 /* Found one! Now use that child in place of the
|
|
|
4561 * current one. This means the current child is
|
|
|
4562 * dropped from the tree. */
|
|
|
4563 np->wn_child = tp;
|
|
|
4564 ++compressed;
|
|
|
4565 break;
|
|
|
4566 }
|
|
|
4567 if (tp == NULL)
|
|
|
4568 {
|
|
|
4569 /* No other child with this hash value equals the child of
|
|
|
4570 * the node, add it to the linked list after the first
|
|
|
4571 * item. */
|
|
|
4572 tp = HI2WN(hi);
|
|
346
|
4573 child->wn_u2.next = tp->wn_u2.next;
|
|
|
4574 tp->wn_u2.next = child;
|
|
300
|
4575 }
|
|
|
4576 }
|
|
|
4577 else
|
|
|
4578 /* No other child has this hash value, add it to the
|
|
|
4579 * hashtable. */
|
|
346
|
4580 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
4581 }
|
|
236
|
4582 }
|
|
300
|
4583 *tot += len;
|
|
|
4584
|
|
|
4585 /*
|
|
|
4586 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
4587 * find a lookalike node. This must be done after compressing the sibling
|
|
|
4588 * list, otherwise the hash key would become invalid by the compression.
|
|
|
4589 */
|
|
346
|
4590 node->wn_u1.hashkey[0] = len;
|
|
300
|
4591 nr = 0;
|
|
|
4592 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4593 {
|
|
300
|
4594 if (np->wn_byte == NUL)
|
|
339
|
4595 /* end node: use wn_flags, wn_region and wn_prefixID */
|
|
|
4596 n = np->wn_flags + (np->wn_region << 8) + (np->wn_prefixID << 16);
|
|
300
|
4597 else
|
|
|
4598 /* byte node: use the byte value and the child pointer */
|
|
|
4599 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
4600 nr = nr * 101 + n;
|
|
236
|
4601 }
|
|
300
|
4602
|
|
|
4603 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
4604 n = nr & 0xff;
|
|
346
|
4605 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
4606 n = (nr >> 8) & 0xff;
|
|
346
|
4607 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
4608 n = (nr >> 16) & 0xff;
|
|
346
|
4609 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
4610 n = (nr >> 24) & 0xff;
|
|
346
|
4611 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
4612 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
4613
|
|
|
4614 return compressed;
|
|
|
4615 }
|
|
|
4616
|
|
|
4617 /*
|
|
|
4618 * Return TRUE when two nodes have identical siblings and children.
|
|
|
4619 */
|
|
|
4620 static int
|
|
|
4621 node_equal(n1, n2)
|
|
|
4622 wordnode_T *n1;
|
|
|
4623 wordnode_T *n2;
|
|
|
4624 {
|
|
|
4625 wordnode_T *p1;
|
|
|
4626 wordnode_T *p2;
|
|
|
4627
|
|
|
4628 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
4629 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
4630 if (p1->wn_byte != p2->wn_byte
|
|
|
4631 || (p1->wn_byte == NUL
|
|
|
4632 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
4633 || p1->wn_region != p2->wn_region
|
|
|
4634 || p1->wn_prefixID != p2->wn_prefixID)
|
|
300
|
4635 : (p1->wn_child != p2->wn_child)))
|
|
|
4636 break;
|
|
|
4637
|
|
|
4638 return p1 == NULL && p2 == NULL;
|
|
236
|
4639 }
|
|
|
4640
|
|
|
4641 /*
|
|
|
4642 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
4643 */
|
|
255
|
4644 void
|
|
236
|
4645 put_bytes(fd, nr, len)
|
|
|
4646 FILE *fd;
|
|
|
4647 long_u nr;
|
|
|
4648 int len;
|
|
|
4649 {
|
|
|
4650 int i;
|
|
|
4651
|
|
|
4652 for (i = len - 1; i >= 0; --i)
|
|
|
4653 putc((int)(nr >> (i * 8)), fd);
|
|
|
4654 }
|
|
|
4655
|
|
323
|
4656 static int
|
|
|
4657 #ifdef __BORLANDC__
|
|
|
4658 _RTLENTRYF
|
|
|
4659 #endif
|
|
|
4660 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
4661
|
|
|
4662 /*
|
|
|
4663 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
4664 */
|
|
|
4665 static int
|
|
|
4666 #ifdef __BORLANDC__
|
|
|
4667 _RTLENTRYF
|
|
|
4668 #endif
|
|
|
4669 rep_compare(s1, s2)
|
|
|
4670 const void *s1;
|
|
|
4671 const void *s2;
|
|
|
4672 {
|
|
|
4673 fromto_T *p1 = (fromto_T *)s1;
|
|
|
4674 fromto_T *p2 = (fromto_T *)s2;
|
|
|
4675
|
|
|
4676 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
4677 }
|
|
|
4678
|
|
236
|
4679 /*
|
|
|
4680 * Write the Vim spell file "fname".
|
|
|
4681 */
|
|
|
4682 static void
|
|
316
|
4683 write_vim_spell(fname, spin)
|
|
236
|
4684 char_u *fname;
|
|
300
|
4685 spellinfo_T *spin;
|
|
236
|
4686 {
|
|
300
|
4687 FILE *fd;
|
|
|
4688 int regionmask;
|
|
236
|
4689 int round;
|
|
300
|
4690 wordnode_T *tree;
|
|
|
4691 int nodecount;
|
|
323
|
4692 int i;
|
|
|
4693 int l;
|
|
|
4694 garray_T *gap;
|
|
|
4695 fromto_T *ftp;
|
|
|
4696 char_u *p;
|
|
|
4697 int rr;
|
|
236
|
4698
|
|
310
|
4699 fd = mch_fopen((char *)fname, "w");
|
|
300
|
4700 if (fd == NULL)
|
|
236
|
4701 {
|
|
|
4702 EMSG2(_(e_notopen), fname);
|
|
|
4703 return;
|
|
|
4704 }
|
|
|
4705
|
|
255
|
4706 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
339
|
4707 * <charflagslen> <charflags>
|
|
|
4708 * <fcharslen> <fchars>
|
|
366
|
4709 * <midwordlen> <midword>
|
|
339
|
4710 * <prefcondcnt> <prefcond> ... */
|
|
300
|
4711
|
|
|
4712 /* <fileID> */
|
|
|
4713 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
4714 EMSG(_(e_write));
|
|
236
|
4715
|
|
|
4716 /* write the region names if there is more than one */
|
|
316
|
4717 if (spin->si_region_count > 1)
|
|
236
|
4718 {
|
|
316
|
4719 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
|
4720 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
|
4721 (size_t)1, fd);
|
|
|
4722 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
4723 }
|
|
|
4724 else
|
|
|
4725 {
|
|
300
|
4726 putc(0, fd);
|
|
|
4727 regionmask = 0;
|
|
236
|
4728 }
|
|
|
4729
|
|
323
|
4730 /*
|
|
|
4731 * Write the table with character flags and table for case folding.
|
|
260
|
4732 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
|
4733 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
4734 * 'encoding'.
|
|
|
4735 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
4736 * the table (avoids that it conflicts). File is shorter too.
|
|
|
4737 */
|
|
|
4738 if (spin->si_ascii || spin->si_add)
|
|
260
|
4739 {
|
|
300
|
4740 putc(0, fd);
|
|
|
4741 putc(0, fd);
|
|
|
4742 putc(0, fd);
|
|
260
|
4743 }
|
|
|
4744 else
|
|
300
|
4745 write_spell_chartab(fd);
|
|
255
|
4746
|
|
366
|
4747
|
|
|
4748 if (spin->si_midword == NULL)
|
|
|
4749 put_bytes(fd, 0L, 2); /* <midwordlen> */
|
|
|
4750 else
|
|
|
4751 {
|
|
|
4752 i = STRLEN(spin->si_midword);
|
|
|
4753 put_bytes(fd, (long_u)i, 2); /* <midwordlen> */
|
|
|
4754 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
4755 }
|
|
|
4756
|
|
|
4757
|
|
339
|
4758 /* Write the prefix conditions. */
|
|
|
4759 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
4760
|
|
366
|
4761 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
4762 * <salflags> <salcount> <sal> ...
|
|
|
4763 * <maplen> <mapstr> */
|
|
|
4764
|
|
323
|
4765 /* Sort the REP items. */
|
|
|
4766 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
4767 sizeof(fromto_T), rep_compare);
|
|
|
4768
|
|
375
|
4769 /* round 1: REP items
|
|
|
4770 * round 2: SAL items (unless SOFO is used) */
|
|
323
|
4771 for (round = 1; round <= 2; ++round)
|
|
|
4772 {
|
|
|
4773 if (round == 1)
|
|
|
4774 gap = &spin->si_rep;
|
|
|
4775 else
|
|
|
4776 {
|
|
|
4777 gap = &spin->si_sal;
|
|
|
4778
|
|
|
4779 i = 0;
|
|
|
4780 if (spin->si_followup)
|
|
|
4781 i |= SAL_F0LLOWUP;
|
|
|
4782 if (spin->si_collapse)
|
|
|
4783 i |= SAL_COLLAPSE;
|
|
|
4784 if (spin->si_rem_accents)
|
|
|
4785 i |= SAL_REM_ACCENTS;
|
|
375
|
4786 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
4787 i |= SAL_SOFO;
|
|
323
|
4788 putc(i, fd); /* <salflags> */
|
|
375
|
4789 if (i & SAL_SOFO)
|
|
|
4790 break;
|
|
323
|
4791 }
|
|
|
4792
|
|
|
4793 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
4794 for (i = 0; i < gap->ga_len; ++i)
|
|
|
4795 {
|
|
|
4796 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
4797 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
4798 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
4799 for (rr = 1; rr <= 2; ++rr)
|
|
|
4800 {
|
|
|
4801 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
4802 l = STRLEN(p);
|
|
|
4803 putc(l, fd);
|
|
|
4804 fwrite(p, l, (size_t)1, fd);
|
|
|
4805 }
|
|
|
4806 }
|
|
|
4807 }
|
|
|
4808
|
|
375
|
4809 /* SOFOFROM and SOFOTO */
|
|
|
4810 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
4811 {
|
|
|
4812 put_bytes(fd, 1L, 2); /* <salcount> */
|
|
|
4813
|
|
|
4814 l = STRLEN(spin->si_sofofr);
|
|
|
4815 put_bytes(fd, (long_u)l, 2); /* <salfromlen> */
|
|
|
4816 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <salfrom> */
|
|
|
4817
|
|
|
4818 l = STRLEN(spin->si_sofoto);
|
|
|
4819 put_bytes(fd, (long_u)l, 2); /* <saltolen> */
|
|
|
4820 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <salto> */
|
|
|
4821 }
|
|
|
4822
|
|
323
|
4823 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
|
4824 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
|
4825 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
|
4826 (size_t)1, fd);
|
|
302
|
4827
|
|
236
|
4828 /*
|
|
339
|
4829 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
4830 */
|
|
323
|
4831 spin->si_memtot = 0;
|
|
339
|
4832 for (round = 1; round <= 3; ++round)
|
|
236
|
4833 {
|
|
339
|
4834 if (round == 1)
|
|
|
4835 tree = spin->si_foldroot;
|
|
|
4836 else if (round == 2)
|
|
|
4837 tree = spin->si_keeproot;
|
|
|
4838 else
|
|
|
4839 tree = spin->si_prefroot;
|
|
236
|
4840
|
|
346
|
4841 /* Clear the index and wnode fields in the tree. */
|
|
|
4842 clear_node(tree);
|
|
|
4843
|
|
300
|
4844 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
4845 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
4846 * nodes. */
|
|
346
|
4847 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
4848
|
|
300
|
4849 /* number of nodes in 4 bytes */
|
|
|
4850 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
4851 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
4852
|
|
300
|
4853 /* Write the nodes. */
|
|
346
|
4854 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
4855 }
|
|
|
4856
|
|
300
|
4857 fclose(fd);
|
|
236
|
4858 }
|
|
|
4859
|
|
|
4860 /*
|
|
346
|
4861 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
4862 * children. This is needed because they are a union with other items to save
|
|
|
4863 * space.
|
|
|
4864 */
|
|
|
4865 static void
|
|
|
4866 clear_node(node)
|
|
|
4867 wordnode_T *node;
|
|
|
4868 {
|
|
|
4869 wordnode_T *np;
|
|
|
4870
|
|
|
4871 if (node != NULL)
|
|
|
4872 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4873 {
|
|
|
4874 np->wn_u1.index = 0;
|
|
|
4875 np->wn_u2.wnode = NULL;
|
|
|
4876
|
|
|
4877 if (np->wn_byte != NUL)
|
|
|
4878 clear_node(np->wn_child);
|
|
|
4879 }
|
|
|
4880 }
|
|
|
4881
|
|
|
4882
|
|
|
4883 /*
|
|
300
|
4884 * Dump a word tree at node "node".
|
|
|
4885 *
|
|
|
4886 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
4887 * byte recursively write the children.
|
|
|
4888 *
|
|
|
4889 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
4890 * are made about the indexes (so that we don't have to write them in the
|
|
|
4891 * file).
|
|
236
|
4892 *
|
|
300
|
4893 * Returns the number of nodes used.
|
|
236
|
4894 */
|
|
300
|
4895 static int
|
|
346
|
4896 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
4897 FILE *fd; /* NULL when only counting */
|
|
300
|
4898 wordnode_T *node;
|
|
|
4899 int index;
|
|
|
4900 int regionmask;
|
|
339
|
4901 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
4902 {
|
|
300
|
4903 int newindex = index;
|
|
|
4904 int siblingcount = 0;
|
|
|
4905 wordnode_T *np;
|
|
236
|
4906 int flags;
|
|
300
|
4907
|
|
|
4908 /* If "node" is zero the tree is empty. */
|
|
|
4909 if (node == NULL)
|
|
|
4910 return 0;
|
|
|
4911
|
|
|
4912 /* Store the index where this node is written. */
|
|
346
|
4913 node->wn_u1.index = index;
|
|
236
|
4914
|
|
300
|
4915 /* Count the number of siblings. */
|
|
|
4916 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4917 ++siblingcount;
|
|
236
|
4918
|
|
300
|
4919 /* Write the sibling count. */
|
|
|
4920 if (fd != NULL)
|
|
|
4921 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
4922
|
|
300
|
4923 /* Write each sibling byte and optionally extra info. */
|
|
|
4924 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4925 {
|
|
300
|
4926 if (np->wn_byte == 0)
|
|
|
4927 {
|
|
|
4928 if (fd != NULL)
|
|
|
4929 {
|
|
339
|
4930 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
4931 if (prefixtree)
|
|
300
|
4932 {
|
|
339
|
4933 /* In PREFIXTREE write the required prefixID and the
|
|
|
4934 * associated condition nr (stored in wn_region). */
|
|
366
|
4935 if (np->wn_flags == (char_u)-2)
|
|
|
4936 putc(BY_FLAGS, fd); /* <byte> rare */
|
|
|
4937 else
|
|
|
4938 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
339
|
4939 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
4940 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
4941 }
|
|
|
4942 else
|
|
|
4943 {
|
|
339
|
4944 /* For word trees we write the flag/region items. */
|
|
|
4945 flags = np->wn_flags;
|
|
|
4946 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
4947 flags |= WF_REGION;
|
|
|
4948 if (np->wn_prefixID != 0)
|
|
|
4949 flags |= WF_PFX;
|
|
|
4950 if (flags == 0)
|
|
|
4951 {
|
|
|
4952 /* word without flags or region */
|
|
|
4953 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
4954 }
|
|
|
4955 else
|
|
|
4956 {
|
|
|
4957 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
4958 putc(flags, fd); /* <flags> */
|
|
|
4959 if (flags & WF_REGION)
|
|
|
4960 putc(np->wn_region, fd); /* <region> */
|
|
|
4961 if (flags & WF_PFX)
|
|
|
4962 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
4963 }
|
|
300
|
4964 }
|
|
|
4965 }
|
|
|
4966 }
|
|
|
4967 else
|
|
|
4968 {
|
|
346
|
4969 if (np->wn_child->wn_u1.index != 0
|
|
|
4970 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
4971 {
|
|
|
4972 /* The child is written elsewhere, write the reference. */
|
|
|
4973 if (fd != NULL)
|
|
|
4974 {
|
|
|
4975 putc(BY_INDEX, fd); /* <byte> */
|
|
|
4976 /* <nodeidx> */
|
|
346
|
4977 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
4978 }
|
|
|
4979 }
|
|
346
|
4980 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
4981 /* We will write the child below and give it an index. */
|
|
346
|
4982 np->wn_child->wn_u2.wnode = node;
|
|
236
|
4983
|
|
300
|
4984 if (fd != NULL)
|
|
|
4985 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
4986 {
|
|
|
4987 EMSG(_(e_write));
|
|
|
4988 return 0;
|
|
|
4989 }
|
|
|
4990 }
|
|
236
|
4991 }
|
|
|
4992
|
|
300
|
4993 /* Space used in the array when reading: one for each sibling and one for
|
|
|
4994 * the count. */
|
|
|
4995 newindex += siblingcount + 1;
|
|
249
|
4996
|
|
300
|
4997 /* Recursively dump the children of each sibling. */
|
|
|
4998 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
4999 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
5000 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
5001 prefixtree);
|
|
249
|
5002
|
|
300
|
5003 return newindex;
|
|
236
|
5004 }
|
|
|
5005
|
|
|
5006
|
|
|
5007 /*
|
|
310
|
5008 * ":mkspell [-ascii] outfile infile ..."
|
|
|
5009 * ":mkspell [-ascii] addfile"
|
|
236
|
5010 */
|
|
|
5011 void
|
|
|
5012 ex_mkspell(eap)
|
|
|
5013 exarg_T *eap;
|
|
|
5014 {
|
|
|
5015 int fcount;
|
|
|
5016 char_u **fnames;
|
|
310
|
5017 char_u *arg = eap->arg;
|
|
|
5018 int ascii = FALSE;
|
|
|
5019
|
|
|
5020 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
5021 {
|
|
|
5022 ascii = TRUE;
|
|
|
5023 arg = skipwhite(arg + 6);
|
|
|
5024 }
|
|
|
5025
|
|
|
5026 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
5027 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
5028 {
|
|
323
|
5029 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
5030 FreeWild(fcount, fnames);
|
|
|
5031 }
|
|
|
5032 }
|
|
|
5033
|
|
|
5034 /*
|
|
|
5035 * Create a Vim spell file from one or more word lists.
|
|
|
5036 * "fnames[0]" is the output file name.
|
|
|
5037 * "fnames[fcount - 1]" is the last input file name.
|
|
|
5038 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
5039 * and ".spl" is appended to make the output file name.
|
|
|
5040 */
|
|
|
5041 static void
|
|
323
|
5042 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
5043 int fcount;
|
|
|
5044 char_u **fnames;
|
|
|
5045 int ascii; /* -ascii argument given */
|
|
|
5046 int overwrite; /* overwrite existing output file */
|
|
323
|
5047 int added_word; /* invoked through "zg" */
|
|
310
|
5048 {
|
|
236
|
5049 char_u fname[MAXPATHL];
|
|
|
5050 char_u wfname[MAXPATHL];
|
|
310
|
5051 char_u **innames;
|
|
|
5052 int incount;
|
|
236
|
5053 afffile_T *(afile[8]);
|
|
|
5054 int i;
|
|
|
5055 int len;
|
|
|
5056 struct stat st;
|
|
255
|
5057 int error = FALSE;
|
|
300
|
5058 spellinfo_T spin;
|
|
|
5059
|
|
|
5060 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
5061 spin.si_verbose = !added_word;
|
|
310
|
5062 spin.si_ascii = ascii;
|
|
323
|
5063 spin.si_followup = TRUE;
|
|
|
5064 spin.si_rem_accents = TRUE;
|
|
|
5065 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
5066 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
5067 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
5068 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
5069
|
|
|
5070 /* default: fnames[0] is output file, following are input files */
|
|
|
5071 innames = &fnames[1];
|
|
|
5072 incount = fcount - 1;
|
|
|
5073
|
|
|
5074 if (fcount >= 1)
|
|
240
|
5075 {
|
|
310
|
5076 len = STRLEN(fnames[0]);
|
|
|
5077 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
5078 {
|
|
|
5079 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
5080 * "path/en.latin1.add.spl". */
|
|
|
5081 innames = &fnames[0];
|
|
|
5082 incount = 1;
|
|
|
5083 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
5084 }
|
|
366
|
5085 else if (fcount == 1)
|
|
|
5086 {
|
|
|
5087 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
5088 innames = &fnames[0];
|
|
|
5089 incount = 1;
|
|
|
5090 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5091 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5092 }
|
|
310
|
5093 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
5094 {
|
|
|
5095 /* Name ends in ".spl", use as the file name. */
|
|
323
|
5096 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
5097 }
|
|
|
5098 else
|
|
|
5099 /* Name should be language, make the file name from it. */
|
|
|
5100 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5101 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5102
|
|
|
5103 /* Check for .ascii.spl. */
|
|
|
5104 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
5105 spin.si_ascii = TRUE;
|
|
|
5106
|
|
|
5107 /* Check for .add.spl. */
|
|
|
5108 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
5109 spin.si_add = TRUE;
|
|
240
|
5110 }
|
|
|
5111
|
|
310
|
5112 if (incount <= 0)
|
|
236
|
5113 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
5114 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
5115 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
5116 else if (incount > 8)
|
|
236
|
5117 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
5118 else
|
|
|
5119 {
|
|
|
5120 /* Check for overwriting before doing things that may take a lot of
|
|
|
5121 * time. */
|
|
310
|
5122 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
5123 {
|
|
|
5124 EMSG(_(e_exists));
|
|
310
|
5125 return;
|
|
236
|
5126 }
|
|
310
|
5127 if (mch_isdir(wfname))
|
|
236
|
5128 {
|
|
310
|
5129 EMSG2(_(e_isadir2), wfname);
|
|
|
5130 return;
|
|
236
|
5131 }
|
|
|
5132
|
|
|
5133 /*
|
|
|
5134 * Init the aff and dic pointers.
|
|
|
5135 * Get the region names if there are more than 2 arguments.
|
|
|
5136 */
|
|
310
|
5137 for (i = 0; i < incount; ++i)
|
|
236
|
5138 {
|
|
310
|
5139 afile[i] = NULL;
|
|
300
|
5140
|
|
316
|
5141 if (incount > 1)
|
|
236
|
5142 {
|
|
310
|
5143 len = STRLEN(innames[i]);
|
|
|
5144 if (STRLEN(gettail(innames[i])) < 5
|
|
|
5145 || innames[i][len - 3] != '_')
|
|
236
|
5146 {
|
|
310
|
5147 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
5148 return;
|
|
236
|
5149 }
|
|
316
|
5150 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
5151 spin.si_region_name[i * 2 + 1] =
|
|
|
5152 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
5153 }
|
|
|
5154 }
|
|
316
|
5155 spin.si_region_count = incount;
|
|
236
|
5156
|
|
300
|
5157 spin.si_foldroot = wordtree_alloc(&spin.si_blocks);
|
|
|
5158 spin.si_keeproot = wordtree_alloc(&spin.si_blocks);
|
|
339
|
5159 spin.si_prefroot = wordtree_alloc(&spin.si_blocks);
|
|
|
5160 if (spin.si_foldroot == NULL
|
|
|
5161 || spin.si_keeproot == NULL
|
|
|
5162 || spin.si_prefroot == NULL)
|
|
300
|
5163 {
|
|
|
5164 error = TRUE;
|
|
310
|
5165 return;
|
|
300
|
5166 }
|
|
|
5167
|
|
351
|
5168 /* When not producing a .add.spl file clear the character table when
|
|
|
5169 * we encounter one in the .aff file. This means we dump the current
|
|
|
5170 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
5171 * better than guessing the contents, the table will match a
|
|
|
5172 * previously loaded spell file. */
|
|
|
5173 if (!spin.si_add)
|
|
|
5174 spin.si_clear_chartab = TRUE;
|
|
|
5175
|
|
236
|
5176 /*
|
|
|
5177 * Read all the .aff and .dic files.
|
|
|
5178 * Text is converted to 'encoding'.
|
|
300
|
5179 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
5180 */
|
|
310
|
5181 for (i = 0; i < incount && !error; ++i)
|
|
236
|
5182 {
|
|
300
|
5183 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
5184 spin.si_region = 1 << i;
|
|
|
5185
|
|
|
5186 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
5187 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
5188 {
|
|
|
5189 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
5190 * "SET" line. */
|
|
310
|
5191 afile[i] = spell_read_aff(fname, &spin);
|
|
|
5192 if (afile[i] == NULL)
|
|
300
|
5193 error = TRUE;
|
|
|
5194 else
|
|
|
5195 {
|
|
|
5196 /* Read the .dic file and store the words in the trees. */
|
|
|
5197 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
5198 innames[i]);
|
|
|
5199 if (spell_read_dic(fname, &spin, afile[i]) == FAIL)
|
|
300
|
5200 error = TRUE;
|
|
|
5201 }
|
|
|
5202 }
|
|
|
5203 else
|
|
|
5204 {
|
|
|
5205 /* No .aff file, try reading the file as a word list. Store
|
|
|
5206 * the words in the trees. */
|
|
310
|
5207 if (spell_read_wordfile(innames[i], &spin) == FAIL)
|
|
300
|
5208 error = TRUE;
|
|
|
5209 }
|
|
236
|
5210
|
|
310
|
5211 #ifdef FEAT_MBYTE
|
|
236
|
5212 /* Free any conversion stuff. */
|
|
300
|
5213 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
5214 #endif
|
|
236
|
5215 }
|
|
|
5216
|
|
300
|
5217 if (!error)
|
|
236
|
5218 {
|
|
|
5219 /*
|
|
300
|
5220 * Remove the dummy NUL from the start of the tree root.
|
|
236
|
5221 */
|
|
300
|
5222 spin.si_foldroot = spin.si_foldroot->wn_sibling;
|
|
|
5223 spin.si_keeproot = spin.si_keeproot->wn_sibling;
|
|
339
|
5224 spin.si_prefroot = spin.si_prefroot->wn_sibling;
|
|
236
|
5225
|
|
|
5226 /*
|
|
300
|
5227 * Combine tails in the tree.
|
|
236
|
5228 */
|
|
323
|
5229 if (!added_word || p_verbose > 2)
|
|
310
|
5230 {
|
|
323
|
5231 if (added_word)
|
|
310
|
5232 verbose_enter();
|
|
|
5233 MSG(_("Compressing word tree..."));
|
|
|
5234 out_flush();
|
|
323
|
5235 if (added_word)
|
|
310
|
5236 verbose_leave();
|
|
|
5237 }
|
|
|
5238 wordtree_compress(spin.si_foldroot, &spin);
|
|
|
5239 wordtree_compress(spin.si_keeproot, &spin);
|
|
339
|
5240 wordtree_compress(spin.si_prefroot, &spin);
|
|
236
|
5241 }
|
|
|
5242
|
|
300
|
5243 if (!error)
|
|
|
5244 {
|
|
|
5245 /*
|
|
|
5246 * Write the info in the spell file.
|
|
|
5247 */
|
|
323
|
5248 if (!added_word || p_verbose > 2)
|
|
310
|
5249 {
|
|
323
|
5250 if (added_word)
|
|
310
|
5251 verbose_enter();
|
|
366
|
5252 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
5253 out_flush();
|
|
323
|
5254 if (added_word)
|
|
310
|
5255 verbose_leave();
|
|
|
5256 }
|
|
|
5257
|
|
316
|
5258 write_vim_spell(wfname, &spin);
|
|
310
|
5259
|
|
323
|
5260 if (!added_word || p_verbose > 2)
|
|
310
|
5261 {
|
|
323
|
5262 if (added_word)
|
|
310
|
5263 verbose_enter();
|
|
|
5264 MSG(_("Done!"));
|
|
|
5265 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
5266 spin.si_memtot);
|
|
310
|
5267 out_flush();
|
|
323
|
5268 if (added_word)
|
|
310
|
5269 verbose_leave();
|
|
|
5270 }
|
|
|
5271
|
|
|
5272 /* If the file is loaded need to reload it. */
|
|
323
|
5273 spell_reload_one(wfname, added_word);
|
|
300
|
5274 }
|
|
|
5275
|
|
|
5276 /* Free the allocated memory. */
|
|
323
|
5277 ga_clear(&spin.si_rep);
|
|
|
5278 ga_clear(&spin.si_sal);
|
|
|
5279 ga_clear(&spin.si_map);
|
|
339
|
5280 ga_clear(&spin.si_prefcond);
|
|
366
|
5281 vim_free(spin.si_midword);
|
|
375
|
5282 vim_free(spin.si_sofofr);
|
|
|
5283 vim_free(spin.si_sofoto);
|
|
300
|
5284
|
|
|
5285 /* Free the .aff file structures. */
|
|
310
|
5286 for (i = 0; i < incount; ++i)
|
|
|
5287 if (afile[i] != NULL)
|
|
|
5288 spell_free_aff(afile[i]);
|
|
339
|
5289
|
|
|
5290 /* Free all the bits and pieces at once. */
|
|
|
5291 free_blocks(spin.si_blocks);
|
|
236
|
5292 }
|
|
310
|
5293 }
|
|
|
5294
|
|
|
5295
|
|
|
5296 /*
|
|
385
|
5297 * ":[count]spellgood {word}"
|
|
|
5298 * ":[count]spellwrong {word}"
|
|
310
|
5299 */
|
|
|
5300 void
|
|
|
5301 ex_spell(eap)
|
|
|
5302 exarg_T *eap;
|
|
|
5303 {
|
|
381
|
5304 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
5305 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
5306 }
|
|
|
5307
|
|
310
|
5308 /*
|
|
|
5309 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
5310 */
|
|
|
5311 void
|
|
385
|
5312 spell_add_word(word, len, bad, index)
|
|
310
|
5313 char_u *word;
|
|
|
5314 int len;
|
|
|
5315 int bad;
|
|
385
|
5316 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
5317 'spellfile' */
|
|
310
|
5318 {
|
|
|
5319 FILE *fd;
|
|
385
|
5320 buf_T *buf = NULL;
|
|
351
|
5321 int new_spf = FALSE;
|
|
|
5322 struct stat st;
|
|
381
|
5323 char_u *fname;
|
|
385
|
5324 char_u fnamebuf[MAXPATHL];
|
|
|
5325 char_u line[MAXWLEN * 2];
|
|
|
5326 long fpos, fpos_next = 0;
|
|
|
5327 int i;
|
|
|
5328 char_u *spf;
|
|
|
5329
|
|
|
5330 if (index == 0) /* use internal wordlist */
|
|
|
5331 {
|
|
|
5332 if (int_wordlist == NULL)
|
|
|
5333 {
|
|
|
5334 int_wordlist = vim_tempname('s');
|
|
|
5335 if (int_wordlist == NULL)
|
|
381
|
5336 return;
|
|
|
5337 }
|
|
385
|
5338 fname = int_wordlist;
|
|
381
|
5339 }
|
|
310
|
5340 else
|
|
|
5341 {
|
|
381
|
5342 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
5343 if (*curbuf->b_p_spf == NUL)
|
|
|
5344 {
|
|
|
5345 init_spellfile();
|
|
|
5346 new_spf = TRUE;
|
|
|
5347 }
|
|
|
5348
|
|
|
5349 if (*curbuf->b_p_spf == NUL)
|
|
|
5350 {
|
|
|
5351 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
5352 return;
|
|
|
5353 }
|
|
|
5354
|
|
385
|
5355 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
5356 {
|
|
|
5357 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
5358 if (i == index)
|
|
|
5359 break;
|
|
|
5360 if (*spf == NUL)
|
|
|
5361 {
|
|
|
5362 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
5363 return;
|
|
|
5364 }
|
|
|
5365 }
|
|
|
5366
|
|
310
|
5367 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
5368 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
5369 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
5370 buf = NULL;
|
|
|
5371 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
5372 {
|
|
310
|
5373 EMSG(_(e_bufloaded));
|
|
381
|
5374 return;
|
|
|
5375 }
|
|
|
5376
|
|
385
|
5377 fname = fnamebuf;
|
|
|
5378 }
|
|
|
5379
|
|
|
5380 if (bad)
|
|
|
5381 {
|
|
|
5382 /* When the word also appears as good word we need to remove that one,
|
|
|
5383 * since its flags sort before the one with WF_BANNED. */
|
|
|
5384 fd = mch_fopen((char *)fname, "r");
|
|
|
5385 if (fd != NULL)
|
|
|
5386 {
|
|
|
5387 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
5388 {
|
|
|
5389 fpos = fpos_next;
|
|
|
5390 fpos_next = ftell(fd);
|
|
|
5391 if (STRNCMP(word, line, len) == 0
|
|
|
5392 && (line[len] == '/' || line[len] < ' '))
|
|
|
5393 {
|
|
|
5394 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
5395 * the start of the line. Mixing reading and writing
|
|
|
5396 * doesn't work for all systems, close the file first. */
|
|
|
5397 fclose(fd);
|
|
|
5398 fd = mch_fopen((char *)fname, "r+");
|
|
|
5399 if (fd == NULL)
|
|
|
5400 break;
|
|
|
5401 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
5402 fputc('#', fd);
|
|
|
5403 fseek(fd, fpos_next, SEEK_SET);
|
|
|
5404 }
|
|
|
5405 }
|
|
|
5406 fclose(fd);
|
|
|
5407 }
|
|
381
|
5408 }
|
|
|
5409
|
|
|
5410 fd = mch_fopen((char *)fname, "a");
|
|
|
5411 if (fd == NULL && new_spf)
|
|
|
5412 {
|
|
|
5413 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
5414 * We may need to create the "spell" directory first. We already
|
|
|
5415 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
5416 STRCPY(NameBuff, fname);
|
|
|
5417 *gettail_sep(NameBuff) = NUL;
|
|
|
5418 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
5419 {
|
|
|
5420 /* The directory doesn't exist. Try creating it and opening the
|
|
|
5421 * file again. */
|
|
|
5422 vim_mkdir(NameBuff, 0755);
|
|
|
5423 fd = mch_fopen((char *)fname, "a");
|
|
|
5424 }
|
|
|
5425 }
|
|
|
5426
|
|
|
5427 if (fd == NULL)
|
|
|
5428 EMSG2(_(e_notopen), fname);
|
|
|
5429 else
|
|
|
5430 {
|
|
|
5431 if (bad)
|
|
|
5432 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
5433 else
|
|
381
|
5434 fprintf(fd, "%.*s\n", len, word);
|
|
|
5435 fclose(fd);
|
|
|
5436
|
|
|
5437 /* Update the .add.spl file. */
|
|
|
5438 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
5439
|
|
|
5440 /* If the .add file is edited somewhere, reload it. */
|
|
|
5441 if (buf != NULL)
|
|
|
5442 buf_reload(buf);
|
|
|
5443
|
|
|
5444 redraw_all_later(NOT_VALID);
|
|
310
|
5445 }
|
|
|
5446 }
|
|
|
5447
|
|
|
5448 /*
|
|
|
5449 * Initialize 'spellfile' for the current buffer.
|
|
|
5450 */
|
|
|
5451 static void
|
|
|
5452 init_spellfile()
|
|
|
5453 {
|
|
|
5454 char_u buf[MAXPATHL];
|
|
|
5455 int l;
|
|
|
5456 slang_T *sl;
|
|
|
5457 char_u *rtp;
|
|
351
|
5458 char_u *lend;
|
|
310
|
5459
|
|
|
5460 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
5461 {
|
|
351
|
5462 /* Find the end of the language name. Exclude the region. */
|
|
|
5463 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
5464 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
5465 ;
|
|
|
5466
|
|
|
5467 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
5468 * are allowed to write. */
|
|
310
|
5469 rtp = p_rtp;
|
|
|
5470 while (*rtp != NUL)
|
|
|
5471 {
|
|
|
5472 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
5473 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
5474 if (filewritable(buf) == 2)
|
|
|
5475 {
|
|
316
|
5476 /* Use the first language name from 'spelllang' and the
|
|
|
5477 * encoding used in the first loaded .spl file. */
|
|
310
|
5478 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
5479 l = STRLEN(buf);
|
|
|
5480 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
5481 "/spell/%.*s.%s.add",
|
|
351
|
5482 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
5483 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
5484 ? (char_u *)"ascii" : spell_enc());
|
|
|
5485 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
5486 break;
|
|
|
5487 }
|
|
|
5488 }
|
|
|
5489 }
|
|
|
5490 }
|
|
236
|
5491
|
|
300
|
5492
|
|
307
|
5493 /*
|
|
|
5494 * Init the chartab used for spelling for ASCII.
|
|
|
5495 * EBCDIC is not supported!
|
|
|
5496 */
|
|
|
5497 static void
|
|
|
5498 clear_spell_chartab(sp)
|
|
|
5499 spelltab_T *sp;
|
|
|
5500 {
|
|
324
|
5501 int i;
|
|
307
|
5502
|
|
|
5503 /* Init everything to FALSE. */
|
|
|
5504 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
5505 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
5506 for (i = 0; i < 256; ++i)
|
|
324
|
5507 {
|
|
307
|
5508 sp->st_fold[i] = i;
|
|
324
|
5509 sp->st_upper[i] = i;
|
|
|
5510 }
|
|
307
|
5511
|
|
|
5512 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
5513 * that is done separately. */
|
|
|
5514 for (i = '0'; i <= '9'; ++i)
|
|
|
5515 sp->st_isw[i] = TRUE;
|
|
|
5516 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
5517 {
|
|
|
5518 sp->st_isw[i] = TRUE;
|
|
|
5519 sp->st_isu[i] = TRUE;
|
|
|
5520 sp->st_fold[i] = i + 0x20;
|
|
|
5521 }
|
|
|
5522 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
5523 {
|
|
307
|
5524 sp->st_isw[i] = TRUE;
|
|
324
|
5525 sp->st_upper[i] = i - 0x20;
|
|
|
5526 }
|
|
307
|
5527 }
|
|
|
5528
|
|
|
5529 /*
|
|
|
5530 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
5531 * Called once while starting up and when 'encoding' changes.
|
|
|
5532 * The default is to use isalpha(), but the spell file should define the word
|
|
|
5533 * characters to make it possible that 'encoding' differs from the current
|
|
|
5534 * locale.
|
|
|
5535 */
|
|
|
5536 void
|
|
|
5537 init_spell_chartab()
|
|
|
5538 {
|
|
|
5539 int i;
|
|
|
5540
|
|
|
5541 did_set_spelltab = FALSE;
|
|
|
5542 clear_spell_chartab(&spelltab);
|
|
|
5543 #ifdef FEAT_MBYTE
|
|
|
5544 if (enc_dbcs)
|
|
|
5545 {
|
|
|
5546 /* DBCS: assume double-wide characters are word characters. */
|
|
|
5547 for (i = 128; i <= 255; ++i)
|
|
|
5548 if (MB_BYTE2LEN(i) == 2)
|
|
|
5549 spelltab.st_isw[i] = TRUE;
|
|
|
5550 }
|
|
324
|
5551 else if (enc_utf8)
|
|
|
5552 {
|
|
|
5553 for (i = 128; i < 256; ++i)
|
|
|
5554 {
|
|
|
5555 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
5556 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
5557 spelltab.st_fold[i] = utf_fold(i);
|
|
|
5558 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
5559 }
|
|
|
5560 }
|
|
307
|
5561 else
|
|
|
5562 #endif
|
|
|
5563 {
|
|
324
|
5564 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
5565 for (i = 128; i < 256; ++i)
|
|
|
5566 {
|
|
|
5567 if (MB_ISUPPER(i))
|
|
|
5568 {
|
|
324
|
5569 spelltab.st_isw[i] = TRUE;
|
|
307
|
5570 spelltab.st_isu[i] = TRUE;
|
|
|
5571 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
5572 }
|
|
324
|
5573 else if (MB_ISLOWER(i))
|
|
|
5574 {
|
|
|
5575 spelltab.st_isw[i] = TRUE;
|
|
|
5576 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
5577 }
|
|
307
|
5578 }
|
|
|
5579 }
|
|
|
5580 }
|
|
|
5581
|
|
|
5582 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
5583 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
5584
|
|
|
5585 /*
|
|
|
5586 * Set the spell character tables from strings in the affix file.
|
|
|
5587 */
|
|
|
5588 static int
|
|
|
5589 set_spell_chartab(fol, low, upp)
|
|
|
5590 char_u *fol;
|
|
|
5591 char_u *low;
|
|
|
5592 char_u *upp;
|
|
|
5593 {
|
|
|
5594 /* We build the new tables here first, so that we can compare with the
|
|
|
5595 * previous one. */
|
|
|
5596 spelltab_T new_st;
|
|
|
5597 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
5598 int f, l, u;
|
|
|
5599
|
|
|
5600 clear_spell_chartab(&new_st);
|
|
|
5601
|
|
|
5602 while (*pf != NUL)
|
|
|
5603 {
|
|
|
5604 if (*pl == NUL || *pu == NUL)
|
|
|
5605 {
|
|
|
5606 EMSG(_(e_affform));
|
|
|
5607 return FAIL;
|
|
|
5608 }
|
|
|
5609 #ifdef FEAT_MBYTE
|
|
|
5610 f = mb_ptr2char_adv(&pf);
|
|
|
5611 l = mb_ptr2char_adv(&pl);
|
|
|
5612 u = mb_ptr2char_adv(&pu);
|
|
|
5613 #else
|
|
|
5614 f = *pf++;
|
|
|
5615 l = *pl++;
|
|
|
5616 u = *pu++;
|
|
|
5617 #endif
|
|
|
5618 /* Every character that appears is a word character. */
|
|
|
5619 if (f < 256)
|
|
|
5620 new_st.st_isw[f] = TRUE;
|
|
|
5621 if (l < 256)
|
|
|
5622 new_st.st_isw[l] = TRUE;
|
|
|
5623 if (u < 256)
|
|
|
5624 new_st.st_isw[u] = TRUE;
|
|
|
5625
|
|
|
5626 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
5627 * case-folding */
|
|
|
5628 if (l < 256 && l != f)
|
|
|
5629 {
|
|
|
5630 if (f >= 256)
|
|
|
5631 {
|
|
|
5632 EMSG(_(e_affrange));
|
|
|
5633 return FAIL;
|
|
|
5634 }
|
|
|
5635 new_st.st_fold[l] = f;
|
|
|
5636 }
|
|
|
5637
|
|
|
5638 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
5639 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
5640 * "FOL" . */
|
|
307
|
5641 if (u < 256 && u != f)
|
|
|
5642 {
|
|
|
5643 if (f >= 256)
|
|
|
5644 {
|
|
|
5645 EMSG(_(e_affrange));
|
|
|
5646 return FAIL;
|
|
|
5647 }
|
|
|
5648 new_st.st_fold[u] = f;
|
|
|
5649 new_st.st_isu[u] = TRUE;
|
|
324
|
5650 new_st.st_upper[f] = u;
|
|
307
|
5651 }
|
|
|
5652 }
|
|
|
5653
|
|
|
5654 if (*pl != NUL || *pu != NUL)
|
|
|
5655 {
|
|
|
5656 EMSG(_(e_affform));
|
|
|
5657 return FAIL;
|
|
|
5658 }
|
|
|
5659
|
|
|
5660 return set_spell_finish(&new_st);
|
|
|
5661 }
|
|
|
5662
|
|
|
5663 /*
|
|
|
5664 * Set the spell character tables from strings in the .spl file.
|
|
|
5665 */
|
|
|
5666 static int
|
|
389
|
5667 set_spell_charflags(flags, cnt, fol)
|
|
307
|
5668 char_u *flags;
|
|
389
|
5669 int cnt; /* length of "flags" */
|
|
|
5670 char_u *fol;
|
|
307
|
5671 {
|
|
|
5672 /* We build the new tables here first, so that we can compare with the
|
|
|
5673 * previous one. */
|
|
|
5674 spelltab_T new_st;
|
|
|
5675 int i;
|
|
389
|
5676 char_u *p = fol;
|
|
324
|
5677 int c;
|
|
307
|
5678
|
|
|
5679 clear_spell_chartab(&new_st);
|
|
|
5680
|
|
389
|
5681 for (i = 0; i < 128; ++i)
|
|
|
5682 {
|
|
|
5683 if (i < cnt)
|
|
|
5684 {
|
|
|
5685 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
5686 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
5687 }
|
|
|
5688
|
|
|
5689 if (*p != NUL)
|
|
|
5690 {
|
|
307
|
5691 #ifdef FEAT_MBYTE
|
|
389
|
5692 c = mb_ptr2char_adv(&p);
|
|
307
|
5693 #else
|
|
389
|
5694 c = *p++;
|
|
307
|
5695 #endif
|
|
389
|
5696 new_st.st_fold[i + 128] = c;
|
|
|
5697 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
5698 new_st.st_upper[c] = i + 128;
|
|
|
5699 }
|
|
307
|
5700 }
|
|
|
5701
|
|
|
5702 return set_spell_finish(&new_st);
|
|
|
5703 }
|
|
|
5704
|
|
|
5705 static int
|
|
|
5706 set_spell_finish(new_st)
|
|
|
5707 spelltab_T *new_st;
|
|
|
5708 {
|
|
|
5709 int i;
|
|
|
5710
|
|
|
5711 if (did_set_spelltab)
|
|
|
5712 {
|
|
|
5713 /* check that it's the same table */
|
|
|
5714 for (i = 0; i < 256; ++i)
|
|
|
5715 {
|
|
|
5716 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
5717 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
5718 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
5719 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
5720 {
|
|
|
5721 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
5722 return FAIL;
|
|
|
5723 }
|
|
|
5724 }
|
|
|
5725 }
|
|
|
5726 else
|
|
|
5727 {
|
|
|
5728 /* copy the new spelltab into the one being used */
|
|
|
5729 spelltab = *new_st;
|
|
|
5730 did_set_spelltab = TRUE;
|
|
|
5731 }
|
|
|
5732
|
|
|
5733 return OK;
|
|
|
5734 }
|
|
|
5735
|
|
|
5736 /*
|
|
358
|
5737 * Return TRUE if "p" points to a word character.
|
|
366
|
5738 * As a special case we see "midword" characters as word character when it is
|
|
358
|
5739 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
5740 * Thus this only works properly when past the first character of the word.
|
|
358
|
5741 */
|
|
|
5742 static int
|
|
376
|
5743 spell_iswordp(p, buf)
|
|
358
|
5744 char_u *p;
|
|
376
|
5745 buf_T *buf; /* buffer used */
|
|
358
|
5746 {
|
|
366
|
5747 #ifdef FEAT_MBYTE
|
|
358
|
5748 char_u *s;
|
|
366
|
5749 int l;
|
|
|
5750 int c;
|
|
|
5751
|
|
|
5752 if (has_mbyte)
|
|
|
5753 {
|
|
|
5754 l = MB_BYTE2LEN(*p);
|
|
358
|
5755 s = p;
|
|
366
|
5756 if (l == 1)
|
|
|
5757 {
|
|
|
5758 /* be quick for ASCII */
|
|
376
|
5759 if (buf->b_spell_ismw[*p])
|
|
366
|
5760 {
|
|
|
5761 s = p + 1; /* skip a mid-word character */
|
|
|
5762 l = MB_BYTE2LEN(*s);
|
|
|
5763 }
|
|
|
5764 }
|
|
|
5765 else
|
|
|
5766 {
|
|
|
5767 c = mb_ptr2char(p);
|
|
376
|
5768 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
5769 : (buf->b_spell_ismw_mb != NULL
|
|
|
5770 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
5771 {
|
|
|
5772 s = p + l;
|
|
|
5773 l = MB_BYTE2LEN(*s);
|
|
|
5774 }
|
|
|
5775 }
|
|
|
5776
|
|
|
5777 if (l > 1)
|
|
|
5778 return mb_get_class(s) >= 2;
|
|
|
5779 return spelltab.st_isw[*s];
|
|
|
5780 }
|
|
358
|
5781 #endif
|
|
366
|
5782
|
|
376
|
5783 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
5784 }
|
|
|
5785
|
|
|
5786 /*
|
|
|
5787 * Return TRUE if "p" points to a word character.
|
|
|
5788 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
5789 */
|
|
|
5790 static int
|
|
|
5791 spell_iswordp_nmw(p)
|
|
|
5792 char_u *p;
|
|
|
5793 {
|
|
|
5794 #ifdef FEAT_MBYTE
|
|
|
5795 if (has_mbyte && MB_BYTE2LEN(*p) > 1)
|
|
|
5796 return mb_get_class(p) >= 2;
|
|
|
5797 #endif
|
|
|
5798
|
|
|
5799 return spelltab.st_isw[*p];
|
|
358
|
5800 }
|
|
|
5801
|
|
372
|
5802 #ifdef FEAT_MBYTE
|
|
|
5803 /*
|
|
|
5804 * Return TRUE if "p" points to a word character.
|
|
|
5805 * Wide version of spell_iswordp().
|
|
|
5806 */
|
|
|
5807 static int
|
|
376
|
5808 spell_iswordp_w(p, buf)
|
|
372
|
5809 int *p;
|
|
376
|
5810 buf_T *buf;
|
|
372
|
5811 {
|
|
|
5812 int *s;
|
|
|
5813
|
|
376
|
5814 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
5815 : (buf->b_spell_ismw_mb != NULL
|
|
|
5816 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
5817 s = p + 1;
|
|
|
5818 else
|
|
|
5819 s = p;
|
|
|
5820
|
|
|
5821 if (mb_char2len(*s) > 1)
|
|
|
5822 {
|
|
|
5823 if (enc_utf8)
|
|
|
5824 return utf_class(*s) >= 2;
|
|
|
5825 if (enc_dbcs)
|
|
|
5826 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
5827 return 0;
|
|
|
5828 }
|
|
|
5829 return spelltab.st_isw[*s];
|
|
|
5830 }
|
|
|
5831 #endif
|
|
|
5832
|
|
358
|
5833 /*
|
|
339
|
5834 * Write the table with prefix conditions to the .spl file.
|
|
|
5835 */
|
|
|
5836 static void
|
|
|
5837 write_spell_prefcond(fd, gap)
|
|
|
5838 FILE *fd;
|
|
|
5839 garray_T *gap;
|
|
|
5840 {
|
|
|
5841 int i;
|
|
|
5842 char_u *p;
|
|
|
5843 int len;
|
|
|
5844
|
|
|
5845 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
5846
|
|
|
5847 for (i = 0; i < gap->ga_len; ++i)
|
|
|
5848 {
|
|
|
5849 /* <prefcond> : <condlen> <condstr> */
|
|
|
5850 p = ((char_u **)gap->ga_data)[i];
|
|
|
5851 if (p == NULL)
|
|
|
5852 fputc(0, fd);
|
|
|
5853 else
|
|
|
5854 {
|
|
|
5855 len = STRLEN(p);
|
|
|
5856 fputc(len, fd);
|
|
|
5857 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
5858 }
|
|
|
5859 }
|
|
|
5860 }
|
|
|
5861
|
|
|
5862 /*
|
|
307
|
5863 * Write the current tables into the .spl file.
|
|
|
5864 * This makes sure the same characters are recognized as word characters when
|
|
|
5865 * generating an when using a spell file.
|
|
|
5866 */
|
|
|
5867 static void
|
|
|
5868 write_spell_chartab(fd)
|
|
|
5869 FILE *fd;
|
|
|
5870 {
|
|
|
5871 char_u charbuf[256 * 4];
|
|
|
5872 int len = 0;
|
|
|
5873 int flags;
|
|
|
5874 int i;
|
|
|
5875
|
|
|
5876 fputc(128, fd); /* <charflagslen> */
|
|
|
5877 for (i = 128; i < 256; ++i)
|
|
|
5878 {
|
|
|
5879 flags = 0;
|
|
|
5880 if (spelltab.st_isw[i])
|
|
324
|
5881 flags |= CF_WORD;
|
|
307
|
5882 if (spelltab.st_isu[i])
|
|
324
|
5883 flags |= CF_UPPER;
|
|
307
|
5884 fputc(flags, fd); /* <charflags> */
|
|
|
5885
|
|
310
|
5886 #ifdef FEAT_MBYTE
|
|
|
5887 if (has_mbyte)
|
|
|
5888 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
|
5889 else
|
|
|
5890 #endif
|
|
|
5891 charbuf[len++] = spelltab.st_fold[i];
|
|
307
|
5892 }
|
|
|
5893
|
|
|
5894 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
|
5895 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
|
5896 }
|
|
|
5897
|
|
|
5898 /*
|
|
324
|
5899 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
5900 * Uses the character definitions from the .spl file.
|
|
307
|
5901 * When using a multi-byte 'encoding' the length may change!
|
|
|
5902 * Returns FAIL when something wrong.
|
|
|
5903 */
|
|
|
5904 static int
|
|
324
|
5905 spell_casefold(str, len, buf, buflen)
|
|
|
5906 char_u *str;
|
|
307
|
5907 int len;
|
|
|
5908 char_u *buf;
|
|
|
5909 int buflen;
|
|
|
5910 {
|
|
|
5911 int i;
|
|
|
5912
|
|
|
5913 if (len >= buflen)
|
|
|
5914 {
|
|
|
5915 buf[0] = NUL;
|
|
|
5916 return FAIL; /* result will not fit */
|
|
|
5917 }
|
|
|
5918
|
|
|
5919 #ifdef FEAT_MBYTE
|
|
|
5920 if (has_mbyte)
|
|
|
5921 {
|
|
324
|
5922 int outi = 0;
|
|
|
5923 char_u *p;
|
|
307
|
5924 int c;
|
|
|
5925
|
|
|
5926 /* Fold one character at a time. */
|
|
324
|
5927 for (p = str; p < str + len; )
|
|
307
|
5928 {
|
|
|
5929 if (outi + MB_MAXBYTES > buflen)
|
|
|
5930 {
|
|
|
5931 buf[outi] = NUL;
|
|
|
5932 return FAIL;
|
|
|
5933 }
|
|
324
|
5934 c = mb_ptr2char_adv(&p);
|
|
|
5935 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
5936 }
|
|
|
5937 buf[outi] = NUL;
|
|
|
5938 }
|
|
|
5939 else
|
|
|
5940 #endif
|
|
|
5941 {
|
|
|
5942 /* Be quick for non-multibyte encodings. */
|
|
|
5943 for (i = 0; i < len; ++i)
|
|
324
|
5944 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
5945 buf[i] = NUL;
|
|
|
5946 }
|
|
|
5947
|
|
|
5948 return OK;
|
|
|
5949 }
|
|
|
5950
|
|
372
|
5951 #define SPS_BEST 1
|
|
|
5952 #define SPS_FAST 2
|
|
|
5953 #define SPS_DOUBLE 4
|
|
|
5954
|
|
|
5955 static int sps_flags = SPS_BEST;
|
|
|
5956
|
|
|
5957 /*
|
|
|
5958 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
|
5959 * Sets "sps_flags".
|
|
|
5960 */
|
|
|
5961 int
|
|
|
5962 spell_check_sps()
|
|
|
5963 {
|
|
|
5964 char_u *p;
|
|
|
5965 char_u buf[MAXPATHL];
|
|
|
5966 int f;
|
|
|
5967
|
|
|
5968 sps_flags = 0;
|
|
|
5969
|
|
|
5970 for (p = p_sps; *p != NUL; )
|
|
|
5971 {
|
|
|
5972 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
5973
|
|
|
5974 f = 0;
|
|
|
5975 if (STRCMP(buf, "best") == 0)
|
|
|
5976 f = SPS_BEST;
|
|
|
5977 else if (STRCMP(buf, "fast") == 0)
|
|
|
5978 f = SPS_FAST;
|
|
|
5979 else if (STRCMP(buf, "double") == 0)
|
|
|
5980 f = SPS_DOUBLE;
|
|
|
5981 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
5982 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
5983 f = -1;
|
|
|
5984
|
|
|
5985 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
5986 {
|
|
|
5987 sps_flags = SPS_BEST;
|
|
|
5988 return FAIL;
|
|
|
5989 }
|
|
|
5990 if (f != 0)
|
|
|
5991 sps_flags = f;
|
|
|
5992 }
|
|
|
5993
|
|
|
5994 if (sps_flags == 0)
|
|
|
5995 sps_flags = SPS_BEST;
|
|
|
5996
|
|
|
5997 return OK;
|
|
|
5998 }
|
|
|
5999
|
|
|
6000 /* Remember what "z?" replaced. */
|
|
|
6001 static char_u *repl_from = NULL;
|
|
|
6002 static char_u *repl_to = NULL;
|
|
|
6003
|
|
323
|
6004 /*
|
|
|
6005 * "z?": Find badly spelled word under or after the cursor.
|
|
|
6006 * Give suggestions for the properly spelled word.
|
|
|
6007 */
|
|
|
6008 void
|
|
|
6009 spell_suggest()
|
|
|
6010 {
|
|
|
6011 char_u *line;
|
|
|
6012 pos_T prev_cursor = curwin->w_cursor;
|
|
|
6013 char_u wcopy[MAXWLEN + 2];
|
|
|
6014 char_u *p;
|
|
|
6015 int i;
|
|
|
6016 int c;
|
|
|
6017 suginfo_T sug;
|
|
|
6018 suggest_T *stp;
|
|
372
|
6019 int mouse_used;
|
|
388
|
6020 int need_cap;
|
|
|
6021 regmatch_T regmatch;
|
|
|
6022 int endcol;
|
|
|
6023 char_u *line_copy = NULL;
|
|
323
|
6024
|
|
344
|
6025 /* Find the start of the badly spelled word. */
|
|
346
|
6026 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
6027 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
6028 {
|
|
|
6029 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
6030 return;
|
|
|
6031
|
|
|
6032 /* No bad word or it starts after the cursor: use the word under the
|
|
|
6033 * cursor. */
|
|
|
6034 curwin->w_cursor = prev_cursor;
|
|
|
6035 line = ml_get_curline();
|
|
|
6036 p = line + curwin->w_cursor.col;
|
|
|
6037 /* Backup to before start of word. */
|
|
|
6038 while (p > line && SPELL_ISWORDP(p))
|
|
|
6039 mb_ptr_back(line, p);
|
|
|
6040 /* Forward to start of word. */
|
|
375
|
6041 while (*p != NUL && !SPELL_ISWORDP(p))
|
|
346
|
6042 mb_ptr_adv(p);
|
|
|
6043
|
|
|
6044 if (!SPELL_ISWORDP(p)) /* No word found. */
|
|
|
6045 {
|
|
|
6046 beep_flush();
|
|
|
6047 return;
|
|
|
6048 }
|
|
|
6049 curwin->w_cursor.col = p - line;
|
|
323
|
6050 }
|
|
|
6051
|
|
344
|
6052 /* Get the word and its length. */
|
|
323
|
6053 line = ml_get_curline();
|
|
344
|
6054
|
|
388
|
6055 /* Figure out if the word should be capitalised. */
|
|
|
6056 need_cap = FALSE;
|
|
|
6057 if (curbuf->b_cap_prog != NULL)
|
|
|
6058 {
|
|
|
6059 endcol = 0;
|
|
|
6060 if (skipwhite(line) - line == curwin->w_cursor.col)
|
|
|
6061 {
|
|
|
6062 /* At start of line, check if previous line is empty or sentence
|
|
|
6063 * ends there. */
|
|
|
6064 if (curwin->w_cursor.lnum == 1)
|
|
|
6065 need_cap = TRUE;
|
|
|
6066 else
|
|
|
6067 {
|
|
|
6068 line = ml_get(curwin->w_cursor.lnum - 1);
|
|
|
6069 if (*skipwhite(line) == NUL)
|
|
|
6070 need_cap = TRUE;
|
|
|
6071 else
|
|
|
6072 {
|
|
|
6073 /* Append a space in place of the line break. */
|
|
|
6074 line_copy = concat_str(line, (char_u *)" ");
|
|
|
6075 line = line_copy;
|
|
|
6076 endcol = STRLEN(line);
|
|
|
6077 }
|
|
|
6078 }
|
|
|
6079 }
|
|
|
6080 else
|
|
|
6081 endcol = curwin->w_cursor.col;
|
|
|
6082
|
|
|
6083 if (endcol > 0)
|
|
|
6084 {
|
|
|
6085 /* Check if sentence ends before the bad word. */
|
|
|
6086 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
6087 regmatch.rm_ic = FALSE;
|
|
|
6088 p = line + endcol;
|
|
|
6089 for (;;)
|
|
|
6090 {
|
|
|
6091 mb_ptr_back(line, p);
|
|
|
6092 if (p == line || SPELL_ISWORDP(p))
|
|
|
6093 break;
|
|
|
6094 if (vim_regexec(®match, p, 0)
|
|
|
6095 && regmatch.endp[0] == line + endcol)
|
|
|
6096 {
|
|
|
6097 need_cap = TRUE;
|
|
|
6098 break;
|
|
|
6099 }
|
|
|
6100 }
|
|
|
6101 }
|
|
|
6102
|
|
|
6103 /* get the line again, we may have been using the previous one */
|
|
|
6104 line = ml_get_curline();
|
|
|
6105 vim_free(line_copy);
|
|
|
6106 }
|
|
|
6107
|
|
344
|
6108 /* Get the list of suggestions */
|
|
388
|
6109 spell_find_suggest(line + curwin->w_cursor.col, &sug, (int)Rows - 2,
|
|
|
6110 TRUE, need_cap);
|
|
323
|
6111
|
|
|
6112 if (sug.su_ga.ga_len == 0)
|
|
|
6113 MSG(_("Sorry, no suggestions"));
|
|
|
6114 else
|
|
|
6115 {
|
|
372
|
6116 vim_free(repl_from);
|
|
|
6117 repl_from = NULL;
|
|
|
6118 vim_free(repl_to);
|
|
|
6119 repl_to = NULL;
|
|
|
6120
|
|
323
|
6121 /* List the suggestions. */
|
|
|
6122 msg_start();
|
|
372
|
6123 lines_left = Rows; /* avoid more prompt */
|
|
323
|
6124 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
6125 sug.su_badlen, sug.su_badptr);
|
|
|
6126 msg_puts(IObuff);
|
|
|
6127 msg_clr_eos();
|
|
|
6128 msg_putchar('\n');
|
|
346
|
6129
|
|
323
|
6130 msg_scroll = TRUE;
|
|
|
6131 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
6132 {
|
|
344
|
6133 stp = &SUG(sug.su_ga, i);
|
|
323
|
6134
|
|
|
6135 /* The suggested word may replace only part of the bad word, add
|
|
|
6136 * the not replaced part. */
|
|
|
6137 STRCPY(wcopy, stp->st_word);
|
|
|
6138 if (sug.su_badlen > stp->st_orglen)
|
|
|
6139 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
6140 sug.su_badptr + stp->st_orglen,
|
|
|
6141 sug.su_badlen - stp->st_orglen);
|
|
346
|
6142 vim_snprintf((char *)IObuff, IOSIZE, _("%2d \"%s\""), i + 1, wcopy);
|
|
|
6143 msg_puts(IObuff);
|
|
|
6144
|
|
|
6145 /* The word may replace more than "su_badlen". */
|
|
|
6146 if (sug.su_badlen < stp->st_orglen)
|
|
|
6147 {
|
|
|
6148 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
6149 stp->st_orglen, sug.su_badptr);
|
|
|
6150 msg_puts(IObuff);
|
|
|
6151 }
|
|
|
6152
|
|
324
|
6153 if (p_verbose > 0)
|
|
344
|
6154 {
|
|
346
|
6155 /* Add the score. */
|
|
351
|
6156 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
346
|
6157 vim_snprintf((char *)IObuff, IOSIZE, _(" (%s%d - %d)"),
|
|
344
|
6158 stp->st_salscore ? "s " : "",
|
|
|
6159 stp->st_score, stp->st_altscore);
|
|
|
6160 else
|
|
346
|
6161 vim_snprintf((char *)IObuff, IOSIZE, _(" (%d)"),
|
|
|
6162 stp->st_score);
|
|
|
6163 msg_advance(30);
|
|
|
6164 msg_puts(IObuff);
|
|
344
|
6165 }
|
|
323
|
6166 msg_putchar('\n');
|
|
|
6167 }
|
|
|
6168
|
|
|
6169 /* Ask for choice. */
|
|
372
|
6170 i = prompt_for_number(&mouse_used);
|
|
|
6171 if (mouse_used)
|
|
|
6172 i -= lines_left;
|
|
|
6173
|
|
344
|
6174 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
6175 {
|
|
372
|
6176 /* Save the from and to text for :spellrepall. */
|
|
|
6177 stp = &SUG(sug.su_ga, i - 1);
|
|
|
6178 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
6179 repl_to = vim_strsave(stp->st_word);
|
|
|
6180
|
|
323
|
6181 /* Replace the word. */
|
|
|
6182 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
6183 if (p != NULL)
|
|
|
6184 {
|
|
|
6185 c = sug.su_badptr - line;
|
|
|
6186 mch_memmove(p, line, c);
|
|
|
6187 STRCPY(p + c, stp->st_word);
|
|
|
6188 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
6189 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6190 curwin->w_cursor.col = c;
|
|
|
6191 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
6192
|
|
|
6193 /* For redo we use a change-word command. */
|
|
|
6194 ResetRedobuff();
|
|
|
6195 AppendToRedobuff((char_u *)"ciw");
|
|
|
6196 AppendToRedobuff(stp->st_word);
|
|
|
6197 AppendCharToRedobuff(ESC);
|
|
323
|
6198 }
|
|
|
6199 }
|
|
|
6200 else
|
|
|
6201 curwin->w_cursor = prev_cursor;
|
|
|
6202 }
|
|
|
6203
|
|
344
|
6204 spell_find_cleanup(&sug);
|
|
|
6205 }
|
|
|
6206
|
|
|
6207 /*
|
|
372
|
6208 * ":spellrepall"
|
|
|
6209 */
|
|
|
6210 /*ARGSUSED*/
|
|
|
6211 void
|
|
|
6212 ex_spellrepall(eap)
|
|
|
6213 exarg_T *eap;
|
|
|
6214 {
|
|
|
6215 pos_T pos = curwin->w_cursor;
|
|
|
6216 char_u *frompat;
|
|
|
6217 int addlen;
|
|
|
6218 char_u *line;
|
|
|
6219 char_u *p;
|
|
|
6220 int didone = FALSE;
|
|
|
6221 int save_ws = p_ws;
|
|
|
6222
|
|
|
6223 if (repl_from == NULL || repl_to == NULL)
|
|
|
6224 {
|
|
|
6225 EMSG(_("E752: No previous spell replacement"));
|
|
|
6226 return;
|
|
|
6227 }
|
|
|
6228 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
6229
|
|
|
6230 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
6231 if (frompat == NULL)
|
|
|
6232 return;
|
|
|
6233 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
6234 p_ws = FALSE;
|
|
|
6235
|
|
|
6236 curwin->w_cursor.lnum = 0;
|
|
|
6237 while (!got_int)
|
|
|
6238 {
|
|
|
6239 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
6240 || u_save_cursor() == FAIL)
|
|
|
6241 break;
|
|
|
6242
|
|
|
6243 /* Only replace when the right word isn't there yet. This happens
|
|
|
6244 * when changing "etc" to "etc.". */
|
|
|
6245 line = ml_get_curline();
|
|
|
6246 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
6247 repl_to, STRLEN(repl_to)) != 0)
|
|
|
6248 {
|
|
|
6249 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
6250 if (p == NULL)
|
|
|
6251 break;
|
|
|
6252 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
6253 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
6254 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
6255 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6256 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
6257 didone = TRUE;
|
|
|
6258 }
|
|
|
6259 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
6260 }
|
|
|
6261
|
|
|
6262 p_ws = save_ws;
|
|
|
6263 curwin->w_cursor = pos;
|
|
|
6264 vim_free(frompat);
|
|
|
6265
|
|
|
6266 if (!didone)
|
|
|
6267 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
|
6268 }
|
|
|
6269
|
|
|
6270 /*
|
|
344
|
6271 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
6272 * a list of allocated strings.
|
|
|
6273 */
|
|
|
6274 void
|
|
|
6275 spell_suggest_list(gap, word, maxcount)
|
|
|
6276 garray_T *gap;
|
|
|
6277 char_u *word;
|
|
|
6278 int maxcount; /* maximum nr of suggestions */
|
|
|
6279 {
|
|
|
6280 suginfo_T sug;
|
|
|
6281 int i;
|
|
|
6282 suggest_T *stp;
|
|
|
6283 char_u *wcopy;
|
|
|
6284
|
|
388
|
6285 spell_find_suggest(word, &sug, maxcount, FALSE, FALSE);
|
|
344
|
6286
|
|
|
6287 /* Make room in "gap". */
|
|
|
6288 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
6289 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
6290 return;
|
|
|
6291
|
|
323
|
6292 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
6293 {
|
|
|
6294 stp = &SUG(sug.su_ga, i);
|
|
|
6295
|
|
|
6296 /* The suggested word may replace only part of "word", add the not
|
|
|
6297 * replaced part. */
|
|
|
6298 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
6299 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
6300 if (wcopy == NULL)
|
|
|
6301 break;
|
|
|
6302 STRCPY(wcopy, stp->st_word);
|
|
|
6303 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
6304 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
6305 }
|
|
|
6306
|
|
|
6307 spell_find_cleanup(&sug);
|
|
|
6308 }
|
|
|
6309
|
|
|
6310 /*
|
|
|
6311 * Find spell suggestions for the word at the start of "badptr".
|
|
|
6312 * Return the suggestions in "su->su_ga".
|
|
|
6313 * The maximum number of suggestions is "maxcount".
|
|
|
6314 * Note: does use info for the current window.
|
|
|
6315 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
6316 */
|
|
|
6317 static void
|
|
388
|
6318 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
6319 char_u *badptr;
|
|
|
6320 suginfo_T *su;
|
|
|
6321 int maxcount;
|
|
358
|
6322 int banbadword; /* don't include badword in suggestions */
|
|
388
|
6323 int need_cap; /* word should start with capital */
|
|
344
|
6324 {
|
|
385
|
6325 int attr = 0;
|
|
372
|
6326 char_u buf[MAXPATHL];
|
|
|
6327 char_u *p;
|
|
|
6328 int do_combine = FALSE;
|
|
|
6329 char_u *sps_copy;
|
|
|
6330 #ifdef FEAT_EVAL
|
|
|
6331 static int expr_busy = FALSE;
|
|
|
6332 #endif
|
|
385
|
6333 int c;
|
|
344
|
6334
|
|
|
6335 /*
|
|
|
6336 * Set the info in "*su".
|
|
|
6337 */
|
|
|
6338 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
6339 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
6340 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
6341 if (*badptr == NUL)
|
|
|
6342 return;
|
|
344
|
6343 hash_init(&su->su_banned);
|
|
|
6344
|
|
|
6345 su->su_badptr = badptr;
|
|
385
|
6346 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
6347 su->su_maxcount = maxcount;
|
|
372
|
6348 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
6349
|
|
|
6350 if (su->su_badlen >= MAXWLEN)
|
|
|
6351 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
6352 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
6353 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
6354 su->su_fbadword, MAXWLEN);
|
|
346
|
6355 /* get caps flags for bad word */
|
|
|
6356 su->su_badflags = captype(su->su_badptr, su->su_badptr + su->su_badlen);
|
|
388
|
6357 if (need_cap)
|
|
|
6358 su->su_badflags |= WF_ONECAP;
|
|
344
|
6359
|
|
385
|
6360 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
6361 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
6362 * for that. */
|
|
|
6363 #ifdef FEAT_MBYTE
|
|
|
6364 c = mb_ptr2char(su->su_badptr);
|
|
|
6365 #else
|
|
388
|
6366 c = *su->su_badptr;
|
|
385
|
6367 #endif
|
|
|
6368 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
6369 {
|
|
|
6370 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
6371 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
6372 0, TRUE);
|
|
|
6373 }
|
|
|
6374
|
|
344
|
6375 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
6376 if (banbadword)
|
|
|
6377 add_banned(su, su->su_badword);
|
|
344
|
6378
|
|
372
|
6379 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
6380 sps_copy = vim_strsave(p_sps);
|
|
|
6381 if (sps_copy == NULL)
|
|
|
6382 return;
|
|
|
6383
|
|
|
6384 /* Loop over the items in 'spellsuggest'. */
|
|
|
6385 for (p = sps_copy; *p != NUL; )
|
|
|
6386 {
|
|
|
6387 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6388
|
|
|
6389 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
6390 {
|
|
|
6391 #ifdef FEAT_EVAL
|
|
375
|
6392 /* Evaluate an expression. Skip this when called recursively,
|
|
|
6393 * when using spellsuggest() in the expression. */
|
|
372
|
6394 if (!expr_busy)
|
|
|
6395 {
|
|
|
6396 expr_busy = TRUE;
|
|
|
6397 spell_suggest_expr(su, buf + 5);
|
|
|
6398 expr_busy = FALSE;
|
|
|
6399 }
|
|
|
6400 #endif
|
|
|
6401 }
|
|
|
6402 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
6403 /* Use list of suggestions in a file. */
|
|
|
6404 spell_suggest_file(su, buf + 5);
|
|
|
6405 else
|
|
|
6406 {
|
|
|
6407 /* Use internal method. */
|
|
|
6408 spell_suggest_intern(su);
|
|
|
6409 if (sps_flags & SPS_DOUBLE)
|
|
|
6410 do_combine = TRUE;
|
|
|
6411 }
|
|
|
6412 }
|
|
|
6413
|
|
|
6414 vim_free(sps_copy);
|
|
|
6415
|
|
|
6416 if (do_combine)
|
|
|
6417 /* Combine the two list of suggestions. This must be done last,
|
|
|
6418 * because sorting changes the order again. */
|
|
|
6419 score_combine(su);
|
|
|
6420 }
|
|
|
6421
|
|
|
6422 #ifdef FEAT_EVAL
|
|
|
6423 /*
|
|
|
6424 * Find suggestions by evaluating expression "expr".
|
|
|
6425 */
|
|
|
6426 static void
|
|
|
6427 spell_suggest_expr(su, expr)
|
|
|
6428 suginfo_T *su;
|
|
|
6429 char_u *expr;
|
|
|
6430 {
|
|
|
6431 list_T *list;
|
|
|
6432 listitem_T *li;
|
|
|
6433 int score;
|
|
|
6434 char_u *p;
|
|
|
6435
|
|
|
6436 /* The work is split up in a few parts to avoid having to export
|
|
|
6437 * suginfo_T.
|
|
|
6438 * First evaluate the expression and get the resulting list. */
|
|
|
6439 list = eval_spell_expr(su->su_badword, expr);
|
|
|
6440 if (list != NULL)
|
|
|
6441 {
|
|
|
6442 /* Loop over the items in the list. */
|
|
|
6443 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
6444 if (li->li_tv.v_type == VAR_LIST)
|
|
|
6445 {
|
|
|
6446 /* Get the word and the score from the items. */
|
|
|
6447 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
6448 if (score >= 0)
|
|
|
6449 add_suggestion(su, &su->su_ga, p,
|
|
|
6450 su->su_badlen, score, 0, TRUE);
|
|
|
6451 }
|
|
|
6452 list_unref(list);
|
|
|
6453 }
|
|
|
6454
|
|
|
6455 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6456 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6457 }
|
|
|
6458 #endif
|
|
|
6459
|
|
|
6460 /*
|
|
|
6461 * Find suggestions a file "fname".
|
|
|
6462 */
|
|
|
6463 static void
|
|
|
6464 spell_suggest_file(su, fname)
|
|
|
6465 suginfo_T *su;
|
|
|
6466 char_u *fname;
|
|
|
6467 {
|
|
|
6468 FILE *fd;
|
|
|
6469 char_u line[MAXWLEN * 2];
|
|
|
6470 char_u *p;
|
|
|
6471 int len;
|
|
|
6472 char_u cword[MAXWLEN];
|
|
|
6473
|
|
|
6474 /* Open the file. */
|
|
|
6475 fd = mch_fopen((char *)fname, "r");
|
|
|
6476 if (fd == NULL)
|
|
|
6477 {
|
|
|
6478 EMSG2(_(e_notopen), fname);
|
|
|
6479 return;
|
|
|
6480 }
|
|
|
6481
|
|
|
6482 /* Read it line by line. */
|
|
|
6483 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
6484 {
|
|
|
6485 line_breakcheck();
|
|
|
6486
|
|
|
6487 p = vim_strchr(line, '/');
|
|
|
6488 if (p == NULL)
|
|
|
6489 continue; /* No Tab found, just skip the line. */
|
|
|
6490 *p++ = NUL;
|
|
|
6491 if (STRICMP(su->su_badword, line) == 0)
|
|
|
6492 {
|
|
|
6493 /* Match! Isolate the good word, until CR or NL. */
|
|
|
6494 for (len = 0; p[len] >= ' '; ++len)
|
|
|
6495 ;
|
|
|
6496 p[len] = NUL;
|
|
|
6497
|
|
|
6498 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
6499 * of the bad word. */
|
|
|
6500 if (captype(p, NULL) == 0)
|
|
|
6501 {
|
|
|
6502 make_case_word(p, cword, su->su_badflags);
|
|
|
6503 p = cword;
|
|
|
6504 }
|
|
|
6505
|
|
|
6506 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
6507 SCORE_FILE, 0, TRUE);
|
|
|
6508 }
|
|
|
6509 }
|
|
|
6510
|
|
|
6511 fclose(fd);
|
|
|
6512
|
|
|
6513 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6514 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6515 }
|
|
|
6516
|
|
|
6517 /*
|
|
|
6518 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
6519 */
|
|
|
6520 static void
|
|
|
6521 spell_suggest_intern(su)
|
|
|
6522 suginfo_T *su;
|
|
|
6523 {
|
|
344
|
6524 /*
|
|
346
|
6525 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
6526 *
|
|
|
6527 * Set a maximum score to limit the combination of operations that is
|
|
|
6528 * tried.
|
|
|
6529 */
|
|
346
|
6530 suggest_try_special(su);
|
|
|
6531
|
|
|
6532 /*
|
|
|
6533 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
6534 * from the .aff file and inserting a space (split the word).
|
|
|
6535 */
|
|
|
6536 suggest_try_change(su);
|
|
344
|
6537
|
|
|
6538 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
6539 if (sps_flags & SPS_DOUBLE)
|
|
|
6540 score_comp_sal(su);
|
|
|
6541
|
|
|
6542 /*
|
|
346
|
6543 * 3. Try finding sound-a-like words.
|
|
344
|
6544 *
|
|
|
6545 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
6546 * very slow and often doesn't find new suggestions.
|
|
|
6547 */
|
|
|
6548 if ((sps_flags & SPS_DOUBLE)
|
|
|
6549 || (!(sps_flags & SPS_FAST)
|
|
|
6550 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
6551 {
|
|
|
6552 /* Allow a higher score now. */
|
|
|
6553 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
6554 suggest_try_soundalike(su);
|
|
344
|
6555 }
|
|
|
6556
|
|
|
6557 /* When CTRL-C was hit while searching do show the results. */
|
|
|
6558 ui_breakcheck();
|
|
|
6559 if (got_int)
|
|
|
6560 {
|
|
|
6561 (void)vgetc();
|
|
|
6562 got_int = FALSE;
|
|
|
6563 }
|
|
|
6564
|
|
372
|
6565 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
6566 {
|
|
|
6567 if (sps_flags & SPS_BEST)
|
|
|
6568 /* Adjust the word score for how it sounds like. */
|
|
|
6569 rescore_suggestions(su);
|
|
|
6570
|
|
|
6571 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
6572 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
6573 }
|
|
|
6574 }
|
|
|
6575
|
|
|
6576 /*
|
|
|
6577 * Free the info put in "*su" by spell_find_suggest().
|
|
|
6578 */
|
|
|
6579 static void
|
|
|
6580 spell_find_cleanup(su)
|
|
|
6581 suginfo_T *su;
|
|
|
6582 {
|
|
|
6583 int i;
|
|
|
6584
|
|
|
6585 /* Free the suggestions. */
|
|
|
6586 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
6587 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
6588 ga_clear(&su->su_ga);
|
|
|
6589 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
6590 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
6591 ga_clear(&su->su_sga);
|
|
323
|
6592
|
|
|
6593 /* Free the banned words. */
|
|
344
|
6594 free_banned(su);
|
|
323
|
6595 }
|
|
|
6596
|
|
|
6597 /*
|
|
324
|
6598 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
6599 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
6600 * The result is NUL terminated.
|
|
323
|
6601 */
|
|
|
6602 static void
|
|
324
|
6603 onecap_copy(word, wcopy, upper)
|
|
323
|
6604 char_u *word;
|
|
|
6605 char_u *wcopy;
|
|
|
6606 int upper; /* TRUE: first letter made upper case */
|
|
|
6607 {
|
|
|
6608 char_u *p;
|
|
|
6609 int c;
|
|
|
6610 int l;
|
|
|
6611
|
|
|
6612 p = word;
|
|
|
6613 #ifdef FEAT_MBYTE
|
|
|
6614 if (has_mbyte)
|
|
|
6615 c = mb_ptr2char_adv(&p);
|
|
|
6616 else
|
|
|
6617 #endif
|
|
|
6618 c = *p++;
|
|
|
6619 if (upper)
|
|
324
|
6620 c = SPELL_TOUPPER(c);
|
|
323
|
6621 else
|
|
324
|
6622 c = SPELL_TOFOLD(c);
|
|
323
|
6623 #ifdef FEAT_MBYTE
|
|
|
6624 if (has_mbyte)
|
|
|
6625 l = mb_char2bytes(c, wcopy);
|
|
|
6626 else
|
|
|
6627 #endif
|
|
|
6628 {
|
|
|
6629 l = 1;
|
|
|
6630 wcopy[0] = c;
|
|
|
6631 }
|
|
376
|
6632 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
6633 }
|
|
|
6634
|
|
|
6635 /*
|
|
324
|
6636 * Make a copy of "word" with all the letters upper cased into
|
|
|
6637 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
6638 */
|
|
|
6639 static void
|
|
|
6640 allcap_copy(word, wcopy)
|
|
|
6641 char_u *word;
|
|
|
6642 char_u *wcopy;
|
|
|
6643 {
|
|
|
6644 char_u *s;
|
|
|
6645 char_u *d;
|
|
|
6646 int c;
|
|
|
6647
|
|
|
6648 d = wcopy;
|
|
|
6649 for (s = word; *s != NUL; )
|
|
|
6650 {
|
|
|
6651 #ifdef FEAT_MBYTE
|
|
|
6652 if (has_mbyte)
|
|
|
6653 c = mb_ptr2char_adv(&s);
|
|
|
6654 else
|
|
|
6655 #endif
|
|
|
6656 c = *s++;
|
|
324
|
6657 c = SPELL_TOUPPER(c);
|
|
323
|
6658
|
|
|
6659 #ifdef FEAT_MBYTE
|
|
|
6660 if (has_mbyte)
|
|
|
6661 {
|
|
|
6662 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
6663 break;
|
|
|
6664 d += mb_char2bytes(c, d);
|
|
|
6665 }
|
|
|
6666 else
|
|
|
6667 #endif
|
|
|
6668 {
|
|
|
6669 if (d - wcopy >= MAXWLEN - 1)
|
|
|
6670 break;
|
|
|
6671 *d++ = c;
|
|
|
6672 }
|
|
|
6673 }
|
|
|
6674 *d = NUL;
|
|
|
6675 }
|
|
|
6676
|
|
|
6677 /*
|
|
346
|
6678 * Try finding suggestions by recognizing specific situations.
|
|
|
6679 */
|
|
|
6680 static void
|
|
|
6681 suggest_try_special(su)
|
|
|
6682 suginfo_T *su;
|
|
|
6683 {
|
|
|
6684 char_u *p;
|
|
376
|
6685 size_t len;
|
|
346
|
6686 int c;
|
|
|
6687 char_u word[MAXWLEN];
|
|
|
6688
|
|
|
6689 /*
|
|
|
6690 * Recognize a word that is repeated: "the the".
|
|
|
6691 */
|
|
|
6692 p = skiptowhite(su->su_fbadword);
|
|
|
6693 len = p - su->su_fbadword;
|
|
|
6694 p = skipwhite(p);
|
|
|
6695 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
6696 {
|
|
|
6697 /* Include badflags: if the badword is onecap or allcap
|
|
|
6698 * use that for the goodword too: "The the" -> "The". */
|
|
|
6699 c = su->su_fbadword[len];
|
|
|
6700 su->su_fbadword[len] = NUL;
|
|
|
6701 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
6702 su->su_fbadword[len] = c;
|
|
351
|
6703 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
6704 }
|
|
|
6705 }
|
|
|
6706
|
|
|
6707 /*
|
|
323
|
6708 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
6709 *
|
|
|
6710 * This uses a state machine. At each node in the tree we try various
|
|
|
6711 * operations. When trying if an operation work "depth" is increased and the
|
|
|
6712 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
6713 * character, replace one and delete another. The number of changes is
|
|
|
6714 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
6715 */
|
|
|
6716 static void
|
|
346
|
6717 suggest_try_change(su)
|
|
323
|
6718 suginfo_T *su;
|
|
|
6719 {
|
|
|
6720 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
6721 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
6722 trystate_T stack[MAXWLEN];
|
|
|
6723 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
|
6724 * to for word split) */
|
|
|
6725 char_u prewordlen = 0; /* length of word in "preword" */
|
|
|
6726 int splitoff = 0; /* index in tword after last split */
|
|
|
6727 trystate_T *sp;
|
|
|
6728 int newscore;
|
|
|
6729 langp_T *lp;
|
|
375
|
6730 char_u *byts, *fbyts, *pbyts;
|
|
|
6731 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
6732 int depth;
|
|
330
|
6733 int c, c2, c3;
|
|
|
6734 int n = 0;
|
|
323
|
6735 int flags;
|
|
|
6736 garray_T *gap;
|
|
324
|
6737 idx_T arridx;
|
|
323
|
6738 int len;
|
|
|
6739 char_u *p;
|
|
|
6740 fromto_T *ftp;
|
|
330
|
6741 int fl = 0, tl;
|
|
346
|
6742 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
323
|
6743
|
|
|
6744 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
6745 * to find matches (esp. REP items). Append some more text, changing
|
|
|
6746 * chars after the bad word may help. */
|
|
323
|
6747 STRCPY(fword, su->su_fbadword);
|
|
346
|
6748 n = STRLEN(fword);
|
|
|
6749 p = su->su_badptr + su->su_badlen;
|
|
|
6750 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
6751
|
|
|
6752 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
6753 lp->lp_slang != NULL; ++lp)
|
|
|
6754 {
|
|
|
6755 /*
|
|
|
6756 * Go through the whole case-fold tree, try changes at each node.
|
|
|
6757 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
6758 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
6759 * word).
|
|
|
6760 */
|
|
|
6761 depth = 0;
|
|
375
|
6762 sp = &stack[0];
|
|
|
6763 sp->ts_state = STATE_START;
|
|
|
6764 sp->ts_score = 0;
|
|
|
6765 sp->ts_curi = 1;
|
|
|
6766 sp->ts_fidx = 0;
|
|
|
6767 sp->ts_fidxtry = 0;
|
|
|
6768 sp->ts_twordlen = 0;
|
|
|
6769 sp->ts_arridx = 0;
|
|
330
|
6770 #ifdef FEAT_MBYTE
|
|
375
|
6771 sp->ts_tcharlen = 0;
|
|
330
|
6772 #endif
|
|
|
6773
|
|
|
6774 /*
|
|
375
|
6775 * When there are postponed prefixes we need to use these first. At
|
|
|
6776 * the end of the prefix we continue in the case-fold tree.
|
|
|
6777 */
|
|
|
6778 fbyts = lp->lp_slang->sl_fbyts;
|
|
|
6779 fidxs = lp->lp_slang->sl_fidxs;
|
|
|
6780 pbyts = lp->lp_slang->sl_pbyts;
|
|
|
6781 pidxs = lp->lp_slang->sl_pidxs;
|
|
|
6782 if (pbyts != NULL)
|
|
|
6783 {
|
|
|
6784 byts = pbyts;
|
|
|
6785 idxs = pidxs;
|
|
|
6786 sp->ts_prefixdepth = PREFIXTREE;
|
|
|
6787 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
6788 }
|
|
|
6789 else
|
|
|
6790 {
|
|
|
6791 byts = fbyts;
|
|
|
6792 idxs = fidxs;
|
|
|
6793 sp->ts_prefixdepth = NOPREFIX;
|
|
|
6794 }
|
|
|
6795
|
|
|
6796 /*
|
|
330
|
6797 * Loop to find all suggestions. At each round we either:
|
|
|
6798 * - For the current state try one operation, advance "ts_curi",
|
|
|
6799 * increase "depth".
|
|
|
6800 * - When a state is done go to the next, set "ts_state".
|
|
|
6801 * - When all states are tried decrease "depth".
|
|
|
6802 */
|
|
323
|
6803 while (depth >= 0 && !got_int)
|
|
|
6804 {
|
|
|
6805 sp = &stack[depth];
|
|
|
6806 switch (sp->ts_state)
|
|
|
6807 {
|
|
|
6808 case STATE_START:
|
|
375
|
6809 case STATE_NOPREFIX:
|
|
323
|
6810 /*
|
|
|
6811 * Start of node: Deal with NUL bytes, which means
|
|
|
6812 * tword[] may end here.
|
|
|
6813 */
|
|
|
6814 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
6815 len = byts[arridx]; /* bytes in this node */
|
|
|
6816 arridx += sp->ts_curi; /* index of current byte */
|
|
|
6817
|
|
375
|
6818 if (sp->ts_prefixdepth == PREFIXTREE)
|
|
|
6819 {
|
|
|
6820 /* Skip over the NUL bytes, we use them later. */
|
|
|
6821 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
6822 ;
|
|
|
6823 sp->ts_curi += n;
|
|
|
6824
|
|
|
6825 /* At end of a prefix or at start of prefixtree: check for
|
|
|
6826 * following word. */
|
|
|
6827 if (byts[arridx] == 0 || sp->ts_state == STATE_NOPREFIX)
|
|
|
6828 {
|
|
|
6829 sp->ts_state = STATE_START;
|
|
|
6830 ++depth;
|
|
|
6831 stack[depth] = stack[depth - 1];
|
|
|
6832 sp = &stack[depth];
|
|
|
6833 sp->ts_prefixdepth = depth - 1;
|
|
|
6834 byts = fbyts;
|
|
|
6835 idxs = fidxs;
|
|
|
6836 sp->ts_state = STATE_START;
|
|
|
6837 sp->ts_curi = 1; /* start just after length byte */
|
|
|
6838 sp->ts_arridx = 0;
|
|
|
6839
|
|
|
6840 /* Move the prefix to preword[] so that
|
|
|
6841 * find_keepcap_word() works. */
|
|
|
6842 prewordlen = splitoff = sp->ts_twordlen;
|
|
|
6843 mch_memmove(preword, tword, splitoff);
|
|
|
6844 break;
|
|
|
6845 }
|
|
|
6846
|
|
|
6847 /* Always past NUL bytes now. */
|
|
|
6848 sp->ts_state = STATE_ENDNUL;
|
|
|
6849 break;
|
|
|
6850 }
|
|
|
6851
|
|
346
|
6852 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
6853 {
|
|
|
6854 /* Past bytes in node and/or past NUL bytes. */
|
|
|
6855 sp->ts_state = STATE_ENDNUL;
|
|
|
6856 break;
|
|
|
6857 }
|
|
|
6858
|
|
|
6859 /*
|
|
|
6860 * End of word in tree.
|
|
|
6861 */
|
|
|
6862 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
6863
|
|
324
|
6864 flags = (int)idxs[arridx];
|
|
323
|
6865
|
|
375
|
6866 if (sp->ts_prefixdepth < MAXWLEN)
|
|
|
6867 {
|
|
|
6868 /* There was a prefix before the word. Check that the
|
|
|
6869 * prefix can be used with this word. */
|
|
|
6870 /* Count the length of the NULs in the prefix. If there
|
|
|
6871 * are none this must be the first try without a prefix.
|
|
|
6872 */
|
|
|
6873 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
6874 len = pbyts[n++];
|
|
|
6875 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
6876 ;
|
|
|
6877 if (c > 0)
|
|
|
6878 {
|
|
|
6879 /* The prefix ID is stored two bytes above the flags. */
|
|
|
6880 c = valid_word_prefix(c, n, (unsigned)flags >> 16,
|
|
|
6881 tword + splitoff, lp->lp_slang);
|
|
|
6882 if (c == 0)
|
|
|
6883 break;
|
|
|
6884
|
|
|
6885 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
6886 if (c & WF_RAREPFX)
|
|
|
6887 flags |= WF_RARE;
|
|
|
6888 }
|
|
|
6889 }
|
|
|
6890
|
|
323
|
6891 /*
|
|
|
6892 * Form the word with proper case in preword.
|
|
|
6893 * If there is a word from a previous split, append.
|
|
|
6894 */
|
|
|
6895 tword[sp->ts_twordlen] = NUL;
|
|
|
6896 if (flags & WF_KEEPCAP)
|
|
|
6897 /* Must find the word in the keep-case tree. */
|
|
|
6898 find_keepcap_word(lp->lp_slang, tword + splitoff,
|
|
|
6899 preword + prewordlen);
|
|
|
6900 else
|
|
346
|
6901 {
|
|
323
|
6902 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
6903 * use that for the goodword too. But if the badword is
|
|
|
6904 * allcap and it's only one char long use onecap. */
|
|
|
6905 c = su->su_badflags;
|
|
|
6906 if ((c & WF_ALLCAP)
|
|
|
6907 #ifdef FEAT_MBYTE
|
|
|
6908 && su->su_badlen == mb_ptr2len_check(su->su_badptr)
|
|
|
6909 #else
|
|
|
6910 && su->su_badlen == 1
|
|
|
6911 #endif
|
|
|
6912 )
|
|
|
6913 c = WF_ONECAP;
|
|
323
|
6914 make_case_word(tword + splitoff,
|
|
346
|
6915 preword + prewordlen, flags | c);
|
|
|
6916 }
|
|
323
|
6917
|
|
|
6918 /* Don't use a banned word. It may appear again as a good
|
|
|
6919 * word, thus remember it. */
|
|
|
6920 if (flags & WF_BANNED)
|
|
|
6921 {
|
|
|
6922 add_banned(su, preword + prewordlen);
|
|
|
6923 break;
|
|
|
6924 }
|
|
372
|
6925 if (was_banned(su, preword + prewordlen)
|
|
|
6926 || was_banned(su, preword))
|
|
323
|
6927 break;
|
|
|
6928
|
|
|
6929 newscore = 0;
|
|
|
6930 if ((flags & WF_REGION)
|
|
|
6931 && (((unsigned)flags >> 8) & lp->lp_region) == 0)
|
|
|
6932 newscore += SCORE_REGION;
|
|
|
6933 if (flags & WF_RARE)
|
|
|
6934 newscore += SCORE_RARE;
|
|
|
6935
|
|
346
|
6936 if (!spell_valid_case(su->su_badflags,
|
|
323
|
6937 captype(preword + prewordlen, NULL)))
|
|
|
6938 newscore += SCORE_ICASE;
|
|
|
6939
|
|
346
|
6940 if ((fword[sp->ts_fidx] == NUL
|
|
376
|
6941 || !spell_iswordp(fword + sp->ts_fidx, curbuf))
|
|
346
|
6942 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
6943 {
|
|
366
|
6944 /* The badword also ends: add suggestions. Give a penalty
|
|
|
6945 * when changing non-word char to word char, e.g., "thes,"
|
|
|
6946 * -> "these". */
|
|
|
6947 p = fword + sp->ts_fidx;
|
|
|
6948 #ifdef FEAT_MBYTE
|
|
|
6949 if (has_mbyte)
|
|
|
6950 mb_ptr_back(fword, p);
|
|
|
6951 else
|
|
|
6952 #endif
|
|
|
6953 --p;
|
|
376
|
6954 if (!spell_iswordp(p, curbuf))
|
|
366
|
6955 {
|
|
|
6956 p = preword + STRLEN(preword);
|
|
|
6957 #ifdef FEAT_MBYTE
|
|
|
6958 if (has_mbyte)
|
|
|
6959 mb_ptr_back(preword, p);
|
|
|
6960 else
|
|
|
6961 #endif
|
|
|
6962 --p;
|
|
376
|
6963 if (spell_iswordp(p, curbuf))
|
|
366
|
6964 newscore += SCORE_NONWORD;
|
|
|
6965 }
|
|
|
6966
|
|
344
|
6967 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
6968 sp->ts_fidx - repextra,
|
|
351
|
6969 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
6970 }
|
|
330
|
6971 else if (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
6972 #ifdef FEAT_MBYTE
|
|
|
6973 /* Don't split halfway a character. */
|
|
|
6974 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
6975 #endif
|
|
|
6976 )
|
|
323
|
6977 {
|
|
|
6978 /* The word in the tree ends but the badword
|
|
|
6979 * continues: try inserting a space and check that a valid
|
|
|
6980 * words starts at fword[sp->ts_fidx]. */
|
|
|
6981 if (try_deeper(su, stack, depth, newscore + SCORE_SPLIT))
|
|
|
6982 {
|
|
|
6983 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
|
6984 sp->ts_save_prewordlen = prewordlen;
|
|
346
|
6985 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
6986 sp->ts_save_splitoff = splitoff;
|
|
376
|
6987 sp->ts_state = STATE_SPLITUNDO;
|
|
|
6988
|
|
|
6989 ++depth;
|
|
|
6990 sp = &stack[depth];
|
|
323
|
6991
|
|
|
6992 /* Append a space to preword. */
|
|
|
6993 STRCAT(preword, " ");
|
|
|
6994 prewordlen = STRLEN(preword);
|
|
|
6995 splitoff = sp->ts_twordlen;
|
|
376
|
6996
|
|
|
6997 /* If the badword has a non-word character at this
|
|
|
6998 * position skip it. That means replacing the
|
|
|
6999 * non-word character with a space. */
|
|
|
7000 if (!spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
7001 {
|
|
|
7002 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
7003 #ifdef FEAT_MBYTE
|
|
|
7004 if (has_mbyte)
|
|
|
7005 sp->ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
|
7006 else
|
|
|
7007 #endif
|
|
|
7008 ++sp->ts_fidx;
|
|
|
7009 }
|
|
324
|
7010 #ifdef FEAT_MBYTE
|
|
|
7011 if (has_mbyte)
|
|
|
7012 {
|
|
|
7013 int i = 0;
|
|
|
7014
|
|
|
7015 /* Case-folding may change the number of bytes:
|
|
376
|
7016 * Count nr of chars in fword[ts_fidx] and
|
|
324
|
7017 * advance that many chars in su->su_badptr. */
|
|
|
7018 for (p = fword; p < fword + sp->ts_fidx;
|
|
|
7019 mb_ptr_adv(p))
|
|
|
7020 ++i;
|
|
|
7021 for (p = su->su_badptr; i > 0; mb_ptr_adv(p))
|
|
|
7022 --i;
|
|
|
7023 }
|
|
|
7024 else
|
|
|
7025 #endif
|
|
|
7026 p = su->su_badptr + sp->ts_fidx;
|
|
346
|
7027 su->su_badflags = captype(p, su->su_badptr
|
|
|
7028 + su->su_badlen);
|
|
323
|
7029
|
|
|
7030 /* Restart at top of the tree. */
|
|
376
|
7031 sp->ts_arridx = 0;
|
|
323
|
7032 }
|
|
|
7033 }
|
|
|
7034 break;
|
|
|
7035
|
|
|
7036 case STATE_SPLITUNDO:
|
|
346
|
7037 /* Undo the changes done for word split. */
|
|
|
7038 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
7039 splitoff = sp->ts_save_splitoff;
|
|
|
7040 prewordlen = sp->ts_save_prewordlen;
|
|
|
7041
|
|
|
7042 /* Continue looking for NUL bytes. */
|
|
|
7043 sp->ts_state = STATE_START;
|
|
|
7044 break;
|
|
|
7045
|
|
|
7046 case STATE_ENDNUL:
|
|
|
7047 /* Past the NUL bytes in the node. */
|
|
346
|
7048 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
7049 {
|
|
|
7050 /* The badword ends, can't use the bytes in this node. */
|
|
|
7051 sp->ts_state = STATE_DEL;
|
|
|
7052 break;
|
|
|
7053 }
|
|
|
7054 sp->ts_state = STATE_PLAIN;
|
|
|
7055 /*FALLTHROUGH*/
|
|
|
7056
|
|
|
7057 case STATE_PLAIN:
|
|
|
7058 /*
|
|
|
7059 * Go over all possible bytes at this node, add each to
|
|
|
7060 * tword[] and use child node. "ts_curi" is the index.
|
|
|
7061 */
|
|
|
7062 arridx = sp->ts_arridx;
|
|
|
7063 if (sp->ts_curi > byts[arridx])
|
|
|
7064 {
|
|
|
7065 /* Done all bytes at this node, do next state. When still
|
|
|
7066 * at already changed bytes skip the other tricks. */
|
|
|
7067 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7068 sp->ts_state = STATE_DEL;
|
|
|
7069 else
|
|
|
7070 sp->ts_state = STATE_FINAL;
|
|
|
7071 }
|
|
|
7072 else
|
|
|
7073 {
|
|
|
7074 arridx += sp->ts_curi++;
|
|
|
7075 c = byts[arridx];
|
|
|
7076
|
|
|
7077 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
7078 * the byte in the bad word adjust the score. But don't
|
|
|
7079 * even try when the byte was already changed. */
|
|
330
|
7080 if (c == fword[sp->ts_fidx]
|
|
|
7081 #ifdef FEAT_MBYTE
|
|
|
7082 || (sp->ts_tcharlen > 0
|
|
|
7083 && sp->ts_isdiff != DIFF_NONE)
|
|
|
7084 #endif
|
|
|
7085 )
|
|
323
|
7086 newscore = 0;
|
|
|
7087 else
|
|
|
7088 newscore = SCORE_SUBST;
|
|
|
7089 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7090 && try_deeper(su, stack, depth, newscore))
|
|
|
7091 {
|
|
|
7092 ++depth;
|
|
330
|
7093 sp = &stack[depth];
|
|
|
7094 ++sp->ts_fidx;
|
|
|
7095 tword[sp->ts_twordlen++] = c;
|
|
|
7096 sp->ts_arridx = idxs[arridx];
|
|
|
7097 #ifdef FEAT_MBYTE
|
|
|
7098 if (newscore == SCORE_SUBST)
|
|
|
7099 sp->ts_isdiff = DIFF_YES;
|
|
|
7100 if (has_mbyte)
|
|
|
7101 {
|
|
|
7102 /* Multi-byte characters are a bit complicated to
|
|
|
7103 * handle: They differ when any of the bytes
|
|
|
7104 * differ and then their length may also differ. */
|
|
|
7105 if (sp->ts_tcharlen == 0)
|
|
|
7106 {
|
|
|
7107 /* First byte. */
|
|
|
7108 sp->ts_tcharidx = 0;
|
|
|
7109 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
7110 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
7111 sp->ts_isdiff = (newscore != 0)
|
|
|
7112 ? DIFF_YES : DIFF_NONE;
|
|
|
7113 }
|
|
|
7114 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
7115 /* When inserting trail bytes don't advance in
|
|
|
7116 * the bad word. */
|
|
|
7117 --sp->ts_fidx;
|
|
|
7118 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
7119 {
|
|
|
7120 /* Last byte of character. */
|
|
|
7121 if (sp->ts_isdiff == DIFF_YES)
|
|
|
7122 {
|
|
|
7123 /* Correct ts_fidx for the byte length of
|
|
|
7124 * the character (we didn't check that
|
|
|
7125 * before). */
|
|
|
7126 sp->ts_fidx = sp->ts_fcharstart
|
|
|
7127 + MB_BYTE2LEN(
|
|
|
7128 fword[sp->ts_fcharstart]);
|
|
|
7129
|
|
|
7130 /* For a similar character adjust score
|
|
|
7131 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
|
7132 if (lp->lp_slang->sl_has_map
|
|
|
7133 && similar_chars(lp->lp_slang,
|
|
|
7134 mb_ptr2char(tword
|
|
|
7135 + sp->ts_twordlen
|
|
|
7136 - sp->ts_tcharlen),
|
|
|
7137 mb_ptr2char(fword
|
|
|
7138 + sp->ts_fcharstart)))
|
|
|
7139 sp->ts_score -=
|
|
|
7140 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7141 }
|
|
358
|
7142 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
7143 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
7144 {
|
|
|
7145 /* If the previous character was the same,
|
|
|
7146 * thus doubling a character, give a bonus
|
|
|
7147 * to the score. */
|
|
|
7148 p = tword + sp->ts_twordlen
|
|
|
7149 - sp->ts_tcharlen;
|
|
|
7150 c = mb_ptr2char(p);
|
|
|
7151 mb_ptr_back(tword, p);
|
|
|
7152 if (c == mb_ptr2char(p))
|
|
|
7153 sp->ts_score -= SCORE_INS
|
|
|
7154 - SCORE_INSDUP;
|
|
|
7155 }
|
|
330
|
7156
|
|
|
7157 /* Starting a new char, reset the length. */
|
|
|
7158 sp->ts_tcharlen = 0;
|
|
|
7159 }
|
|
|
7160 }
|
|
|
7161 else
|
|
|
7162 #endif
|
|
|
7163 {
|
|
|
7164 /* If we found a similar char adjust the score.
|
|
|
7165 * We do this after calling try_deeper() because
|
|
|
7166 * it's slow. */
|
|
|
7167 if (newscore != 0
|
|
|
7168 && lp->lp_slang->sl_has_map
|
|
|
7169 && similar_chars(lp->lp_slang,
|
|
|
7170 c, fword[sp->ts_fidx - 1]))
|
|
|
7171 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7172 }
|
|
323
|
7173 }
|
|
|
7174 }
|
|
|
7175 break;
|
|
|
7176
|
|
|
7177 case STATE_DEL:
|
|
330
|
7178 #ifdef FEAT_MBYTE
|
|
|
7179 /* When past the first byte of a multi-byte char don't try
|
|
|
7180 * delete/insert/swap a character. */
|
|
|
7181 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
7182 {
|
|
|
7183 sp->ts_state = STATE_FINAL;
|
|
|
7184 break;
|
|
|
7185 }
|
|
|
7186 #endif
|
|
|
7187 /*
|
|
|
7188 * Try skipping one character in the bad word (delete it).
|
|
|
7189 */
|
|
323
|
7190 sp->ts_state = STATE_INS;
|
|
|
7191 sp->ts_curi = 1;
|
|
|
7192 if (fword[sp->ts_fidx] != NUL
|
|
|
7193 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
7194 {
|
|
|
7195 ++depth;
|
|
358
|
7196
|
|
|
7197 /* Advance over the character in fword[]. Give a bonus to
|
|
|
7198 * the score if the same character is following "nn" ->
|
|
|
7199 * "n". */
|
|
330
|
7200 #ifdef FEAT_MBYTE
|
|
|
7201 if (has_mbyte)
|
|
358
|
7202 {
|
|
|
7203 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
7204 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
358
|
7205 if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
|
7206 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7207 }
|
|
330
|
7208 else
|
|
|
7209 #endif
|
|
358
|
7210 {
|
|
330
|
7211 ++stack[depth].ts_fidx;
|
|
358
|
7212 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
7213 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7214 }
|
|
323
|
7215 break;
|
|
|
7216 }
|
|
|
7217 /*FALLTHROUGH*/
|
|
|
7218
|
|
|
7219 case STATE_INS:
|
|
330
|
7220 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
7221 * node. */
|
|
|
7222 n = sp->ts_arridx;
|
|
|
7223 if (sp->ts_curi > byts[n])
|
|
|
7224 {
|
|
|
7225 /* Done all bytes at this node, do next state. */
|
|
|
7226 sp->ts_state = STATE_SWAP;
|
|
|
7227 }
|
|
|
7228 else
|
|
|
7229 {
|
|
330
|
7230 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
7231 n += sp->ts_curi++;
|
|
|
7232 c = byts[n];
|
|
|
7233 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
7234 {
|
|
|
7235 ++depth;
|
|
330
|
7236 sp = &stack[depth];
|
|
|
7237 tword[sp->ts_twordlen++] = c;
|
|
|
7238 sp->ts_arridx = idxs[n];
|
|
|
7239 #ifdef FEAT_MBYTE
|
|
|
7240 if (has_mbyte)
|
|
|
7241 {
|
|
|
7242 fl = MB_BYTE2LEN(c);
|
|
|
7243 if (fl > 1)
|
|
|
7244 {
|
|
|
7245 /* There are following bytes for the same
|
|
|
7246 * character. We must find all bytes before
|
|
|
7247 * trying delete/insert/swap/etc. */
|
|
|
7248 sp->ts_tcharlen = fl;
|
|
|
7249 sp->ts_tcharidx = 1;
|
|
|
7250 sp->ts_isdiff = DIFF_INSERT;
|
|
|
7251 }
|
|
|
7252 }
|
|
358
|
7253 else
|
|
|
7254 fl = 1;
|
|
|
7255 if (fl == 1)
|
|
330
|
7256 #endif
|
|
358
|
7257 {
|
|
|
7258 /* If the previous character was the same, thus
|
|
|
7259 * doubling a character, give a bonus to the
|
|
|
7260 * score. */
|
|
|
7261 if (sp->ts_twordlen >= 2
|
|
|
7262 && tword[sp->ts_twordlen - 2] == c)
|
|
|
7263 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
7264 }
|
|
323
|
7265 }
|
|
|
7266 }
|
|
|
7267 break;
|
|
|
7268
|
|
|
7269 case STATE_SWAP:
|
|
330
|
7270 /*
|
|
|
7271 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
7272 * We change "fword" here, it's changed back afterwards.
|
|
|
7273 */
|
|
|
7274 p = fword + sp->ts_fidx;
|
|
|
7275 c = *p;
|
|
|
7276 if (c == NUL)
|
|
|
7277 {
|
|
|
7278 /* End of word, can't swap or replace. */
|
|
|
7279 sp->ts_state = STATE_FINAL;
|
|
|
7280 break;
|
|
|
7281 }
|
|
|
7282 #ifdef FEAT_MBYTE
|
|
|
7283 if (has_mbyte)
|
|
|
7284 {
|
|
|
7285 n = mb_ptr2len_check(p);
|
|
|
7286 c = mb_ptr2char(p);
|
|
|
7287 c2 = mb_ptr2char(p + n);
|
|
|
7288 }
|
|
|
7289 else
|
|
|
7290 #endif
|
|
|
7291 c2 = p[1];
|
|
|
7292 if (c == c2)
|
|
323
|
7293 {
|
|
330
|
7294 /* Characters are identical, swap won't do anything. */
|
|
|
7295 sp->ts_state = STATE_SWAP3;
|
|
|
7296 break;
|
|
|
7297 }
|
|
|
7298 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
7299 {
|
|
|
7300 sp->ts_state = STATE_UNSWAP;
|
|
323
|
7301 ++depth;
|
|
330
|
7302 #ifdef FEAT_MBYTE
|
|
|
7303 if (has_mbyte)
|
|
|
7304 {
|
|
|
7305 fl = mb_char2len(c2);
|
|
|
7306 mch_memmove(p, p + n, fl);
|
|
|
7307 mb_char2bytes(c, p + fl);
|
|
|
7308 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7309 }
|
|
|
7310 else
|
|
|
7311 #endif
|
|
|
7312 {
|
|
|
7313 p[0] = c2;
|
|
|
7314 p[1] = c;
|
|
|
7315 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
7316 }
|
|
323
|
7317 }
|
|
|
7318 else
|
|
|
7319 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
7320 sp->ts_state = STATE_REP_INI;
|
|
|
7321 break;
|
|
|
7322
|
|
330
|
7323 case STATE_UNSWAP:
|
|
|
7324 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
7325 p = fword + sp->ts_fidx;
|
|
|
7326 #ifdef FEAT_MBYTE
|
|
|
7327 if (has_mbyte)
|
|
|
7328 {
|
|
|
7329 n = MB_BYTE2LEN(*p);
|
|
|
7330 c = mb_ptr2char(p + n);
|
|
|
7331 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
7332 mb_char2bytes(c, p);
|
|
|
7333 }
|
|
|
7334 else
|
|
|
7335 #endif
|
|
|
7336 {
|
|
|
7337 c = *p;
|
|
|
7338 *p = p[1];
|
|
|
7339 p[1] = c;
|
|
|
7340 }
|
|
|
7341 /*FALLTHROUGH*/
|
|
|
7342
|
|
|
7343 case STATE_SWAP3:
|
|
323
|
7344 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
7345 * "fword" here, it's changed back afterwards. */
|
|
|
7346 p = fword + sp->ts_fidx;
|
|
|
7347 #ifdef FEAT_MBYTE
|
|
|
7348 if (has_mbyte)
|
|
|
7349 {
|
|
|
7350 n = mb_ptr2len_check(p);
|
|
|
7351 c = mb_ptr2char(p);
|
|
|
7352 fl = mb_ptr2len_check(p + n);
|
|
|
7353 c2 = mb_ptr2char(p + n);
|
|
|
7354 c3 = mb_ptr2char(p + n + fl);
|
|
|
7355 }
|
|
|
7356 else
|
|
|
7357 #endif
|
|
323
|
7358 {
|
|
330
|
7359 c = *p;
|
|
|
7360 c2 = p[1];
|
|
|
7361 c3 = p[2];
|
|
|
7362 }
|
|
|
7363
|
|
|
7364 /* When characters are identical: "121" then SWAP3 result is
|
|
|
7365 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
7366 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
7367 * swapping. Also skip when c3 is NUL. */
|
|
|
7368 if (c == c3 || c3 == NUL)
|
|
|
7369 {
|
|
|
7370 sp->ts_state = STATE_REP_INI;
|
|
|
7371 break;
|
|
|
7372 }
|
|
|
7373 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7374 {
|
|
|
7375 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
7376 ++depth;
|
|
330
|
7377 #ifdef FEAT_MBYTE
|
|
|
7378 if (has_mbyte)
|
|
|
7379 {
|
|
|
7380 tl = mb_char2len(c3);
|
|
|
7381 mch_memmove(p, p + n + fl, tl);
|
|
|
7382 mb_char2bytes(c2, p + tl);
|
|
|
7383 mb_char2bytes(c, p + fl + tl);
|
|
|
7384 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
7385 }
|
|
|
7386 else
|
|
|
7387 #endif
|
|
|
7388 {
|
|
|
7389 p[0] = p[2];
|
|
|
7390 p[2] = c;
|
|
|
7391 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7392 }
|
|
323
|
7393 }
|
|
|
7394 else
|
|
|
7395 sp->ts_state = STATE_REP_INI;
|
|
|
7396 break;
|
|
|
7397
|
|
330
|
7398 case STATE_UNSWAP3:
|
|
|
7399 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
7400 p = fword + sp->ts_fidx;
|
|
|
7401 #ifdef FEAT_MBYTE
|
|
|
7402 if (has_mbyte)
|
|
|
7403 {
|
|
|
7404 n = MB_BYTE2LEN(*p);
|
|
|
7405 c2 = mb_ptr2char(p + n);
|
|
|
7406 fl = MB_BYTE2LEN(p[n]);
|
|
|
7407 c = mb_ptr2char(p + n + fl);
|
|
|
7408 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
7409 mch_memmove(p + fl + tl, p, n);
|
|
|
7410 mb_char2bytes(c, p);
|
|
|
7411 mb_char2bytes(c2, p + tl);
|
|
|
7412 }
|
|
|
7413 else
|
|
|
7414 #endif
|
|
|
7415 {
|
|
|
7416 c = *p;
|
|
|
7417 *p = p[2];
|
|
|
7418 p[2] = c;
|
|
|
7419 }
|
|
346
|
7420
|
|
330
|
7421 /* Rotate three characters left: "123" -> "231". We change
|
|
|
7422 * "fword" here, it's changed back afterwards. */
|
|
323
|
7423 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7424 {
|
|
330
|
7425 sp->ts_state = STATE_UNROT3L;
|
|
323
|
7426 ++depth;
|
|
330
|
7427 p = fword + sp->ts_fidx;
|
|
|
7428 #ifdef FEAT_MBYTE
|
|
|
7429 if (has_mbyte)
|
|
|
7430 {
|
|
|
7431 n = mb_ptr2len_check(p);
|
|
|
7432 c = mb_ptr2char(p);
|
|
|
7433 fl = mb_ptr2len_check(p + n);
|
|
|
7434 fl += mb_ptr2len_check(p + n + fl);
|
|
|
7435 mch_memmove(p, p + n, fl);
|
|
|
7436 mb_char2bytes(c, p + fl);
|
|
|
7437 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7438 }
|
|
|
7439 else
|
|
|
7440 #endif
|
|
|
7441 {
|
|
|
7442 c = *p;
|
|
|
7443 *p = p[1];
|
|
|
7444 p[1] = p[2];
|
|
|
7445 p[2] = c;
|
|
|
7446 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7447 }
|
|
323
|
7448 }
|
|
|
7449 else
|
|
|
7450 sp->ts_state = STATE_REP_INI;
|
|
|
7451 break;
|
|
|
7452
|
|
330
|
7453 case STATE_UNROT3L:
|
|
346
|
7454 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
7455 p = fword + sp->ts_fidx;
|
|
|
7456 #ifdef FEAT_MBYTE
|
|
|
7457 if (has_mbyte)
|
|
|
7458 {
|
|
|
7459 n = MB_BYTE2LEN(*p);
|
|
|
7460 n += MB_BYTE2LEN(p[n]);
|
|
|
7461 c = mb_ptr2char(p + n);
|
|
|
7462 tl = MB_BYTE2LEN(p[n]);
|
|
|
7463 mch_memmove(p + tl, p, n);
|
|
|
7464 mb_char2bytes(c, p);
|
|
|
7465 }
|
|
|
7466 else
|
|
|
7467 #endif
|
|
|
7468 {
|
|
|
7469 c = p[2];
|
|
|
7470 p[2] = p[1];
|
|
|
7471 p[1] = *p;
|
|
|
7472 *p = c;
|
|
|
7473 }
|
|
346
|
7474
|
|
323
|
7475 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
7476 * "fword" here, it's changed back afterwards. */
|
|
323
|
7477 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7478 {
|
|
330
|
7479 sp->ts_state = STATE_UNROT3R;
|
|
323
|
7480 ++depth;
|
|
330
|
7481 p = fword + sp->ts_fidx;
|
|
|
7482 #ifdef FEAT_MBYTE
|
|
|
7483 if (has_mbyte)
|
|
|
7484 {
|
|
|
7485 n = mb_ptr2len_check(p);
|
|
|
7486 n += mb_ptr2len_check(p + n);
|
|
|
7487 c = mb_ptr2char(p + n);
|
|
|
7488 tl = mb_ptr2len_check(p + n);
|
|
|
7489 mch_memmove(p + tl, p, n);
|
|
|
7490 mb_char2bytes(c, p);
|
|
|
7491 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
7492 }
|
|
|
7493 else
|
|
|
7494 #endif
|
|
|
7495 {
|
|
|
7496 c = p[2];
|
|
|
7497 p[2] = p[1];
|
|
|
7498 p[1] = *p;
|
|
|
7499 *p = c;
|
|
|
7500 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7501 }
|
|
323
|
7502 }
|
|
|
7503 else
|
|
|
7504 sp->ts_state = STATE_REP_INI;
|
|
|
7505 break;
|
|
|
7506
|
|
330
|
7507 case STATE_UNROT3R:
|
|
346
|
7508 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
7509 p = fword + sp->ts_fidx;
|
|
|
7510 #ifdef FEAT_MBYTE
|
|
|
7511 if (has_mbyte)
|
|
|
7512 {
|
|
|
7513 c = mb_ptr2char(p);
|
|
|
7514 tl = MB_BYTE2LEN(*p);
|
|
|
7515 n = MB_BYTE2LEN(p[tl]);
|
|
|
7516 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
7517 mch_memmove(p, p + tl, n);
|
|
|
7518 mb_char2bytes(c, p + n);
|
|
|
7519 }
|
|
|
7520 else
|
|
|
7521 #endif
|
|
|
7522 {
|
|
|
7523 c = *p;
|
|
|
7524 *p = p[1];
|
|
|
7525 p[1] = p[2];
|
|
|
7526 p[2] = c;
|
|
|
7527 }
|
|
323
|
7528 /*FALLTHROUGH*/
|
|
|
7529
|
|
|
7530 case STATE_REP_INI:
|
|
|
7531 /* Check if matching with REP items from the .aff file would
|
|
|
7532 * work. Quickly skip if there are no REP items or the score
|
|
|
7533 * is going to be too high anyway. */
|
|
|
7534 gap = &lp->lp_slang->sl_rep;
|
|
|
7535 if (gap->ga_len == 0
|
|
|
7536 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
7537 {
|
|
|
7538 sp->ts_state = STATE_FINAL;
|
|
|
7539 break;
|
|
|
7540 }
|
|
|
7541
|
|
|
7542 /* Use the first byte to quickly find the first entry that
|
|
330
|
7543 * may match. If the index is -1 there is none. */
|
|
323
|
7544 sp->ts_curi = lp->lp_slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
7545 if (sp->ts_curi < 0)
|
|
|
7546 {
|
|
|
7547 sp->ts_state = STATE_FINAL;
|
|
|
7548 break;
|
|
|
7549 }
|
|
|
7550
|
|
|
7551 sp->ts_state = STATE_REP;
|
|
|
7552 /*FALLTHROUGH*/
|
|
|
7553
|
|
|
7554 case STATE_REP:
|
|
|
7555 /* Try matching with REP items from the .aff file. For each
|
|
330
|
7556 * match replace the characters and check if the resulting
|
|
|
7557 * word is valid. */
|
|
323
|
7558 p = fword + sp->ts_fidx;
|
|
|
7559
|
|
|
7560 gap = &lp->lp_slang->sl_rep;
|
|
|
7561 while (sp->ts_curi < gap->ga_len)
|
|
|
7562 {
|
|
|
7563 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
7564 if (*ftp->ft_from != *p)
|
|
|
7565 {
|
|
|
7566 /* past possible matching entries */
|
|
|
7567 sp->ts_curi = gap->ga_len;
|
|
|
7568 break;
|
|
|
7569 }
|
|
|
7570 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
7571 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
7572 {
|
|
|
7573 /* Need to undo this afterwards. */
|
|
|
7574 sp->ts_state = STATE_REP_UNDO;
|
|
|
7575
|
|
|
7576 /* Change the "from" to the "to" string. */
|
|
|
7577 ++depth;
|
|
|
7578 fl = STRLEN(ftp->ft_from);
|
|
|
7579 tl = STRLEN(ftp->ft_to);
|
|
|
7580 if (fl != tl)
|
|
346
|
7581 {
|
|
323
|
7582 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
7583 repextra += tl - fl;
|
|
|
7584 }
|
|
323
|
7585 mch_memmove(p, ftp->ft_to, tl);
|
|
|
7586 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
7587 #ifdef FEAT_MBYTE
|
|
|
7588 stack[depth].ts_tcharlen = 0;
|
|
|
7589 #endif
|
|
323
|
7590 break;
|
|
|
7591 }
|
|
|
7592 }
|
|
|
7593
|
|
|
7594 if (sp->ts_curi >= gap->ga_len)
|
|
|
7595 /* No (more) matches. */
|
|
|
7596 sp->ts_state = STATE_FINAL;
|
|
|
7597
|
|
|
7598 break;
|
|
|
7599
|
|
|
7600 case STATE_REP_UNDO:
|
|
|
7601 /* Undo a REP replacement and continue with the next one. */
|
|
|
7602 ftp = (fromto_T *)lp->lp_slang->sl_rep.ga_data
|
|
|
7603 + sp->ts_curi - 1;
|
|
|
7604 fl = STRLEN(ftp->ft_from);
|
|
|
7605 tl = STRLEN(ftp->ft_to);
|
|
|
7606 p = fword + sp->ts_fidx;
|
|
|
7607 if (fl != tl)
|
|
346
|
7608 {
|
|
323
|
7609 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
7610 repextra -= tl - fl;
|
|
|
7611 }
|
|
323
|
7612 mch_memmove(p, ftp->ft_from, fl);
|
|
|
7613 sp->ts_state = STATE_REP;
|
|
|
7614 break;
|
|
|
7615
|
|
|
7616 default:
|
|
|
7617 /* Did all possible states at this level, go up one level. */
|
|
|
7618 --depth;
|
|
344
|
7619
|
|
375
|
7620 if (depth >= 0 && stack[depth].ts_prefixdepth == PREFIXTREE)
|
|
|
7621 {
|
|
|
7622 /* Continue in or go back to the prefix tree. */
|
|
|
7623 byts = pbyts;
|
|
|
7624 idxs = pidxs;
|
|
|
7625 splitoff = 0;
|
|
|
7626 }
|
|
|
7627
|
|
344
|
7628 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
7629 line_breakcheck();
|
|
323
|
7630 }
|
|
|
7631 }
|
|
|
7632 }
|
|
|
7633 }
|
|
|
7634
|
|
|
7635 /*
|
|
|
7636 * Try going one level deeper in the tree.
|
|
|
7637 */
|
|
|
7638 static int
|
|
|
7639 try_deeper(su, stack, depth, score_add)
|
|
|
7640 suginfo_T *su;
|
|
|
7641 trystate_T *stack;
|
|
|
7642 int depth;
|
|
|
7643 int score_add;
|
|
|
7644 {
|
|
|
7645 int newscore;
|
|
|
7646
|
|
|
7647 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
7648 newscore = stack[depth].ts_score + score_add;
|
|
|
7649 if (newscore >= su->su_maxscore)
|
|
|
7650 return FALSE;
|
|
|
7651
|
|
330
|
7652 stack[depth + 1] = stack[depth];
|
|
323
|
7653 stack[depth + 1].ts_state = STATE_START;
|
|
|
7654 stack[depth + 1].ts_score = newscore;
|
|
|
7655 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
7656 return TRUE;
|
|
|
7657 }
|
|
|
7658
|
|
|
7659 /*
|
|
|
7660 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
7661 * words and put it in "kword".
|
|
|
7662 * Theoretically there could be several keep-case words that result in the
|
|
|
7663 * same case-folded word, but we only find one...
|
|
|
7664 */
|
|
|
7665 static void
|
|
|
7666 find_keepcap_word(slang, fword, kword)
|
|
|
7667 slang_T *slang;
|
|
|
7668 char_u *fword;
|
|
|
7669 char_u *kword;
|
|
|
7670 {
|
|
|
7671 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
7672 int depth;
|
|
324
|
7673 idx_T tryidx;
|
|
323
|
7674
|
|
|
7675 /* The following arrays are used at each depth in the tree. */
|
|
324
|
7676 idx_T arridx[MAXWLEN];
|
|
323
|
7677 int round[MAXWLEN];
|
|
|
7678 int fwordidx[MAXWLEN];
|
|
|
7679 int uwordidx[MAXWLEN];
|
|
|
7680 int kwordlen[MAXWLEN];
|
|
|
7681
|
|
|
7682 int flen, ulen;
|
|
|
7683 int l;
|
|
|
7684 int len;
|
|
|
7685 int c;
|
|
324
|
7686 idx_T lo, hi, m;
|
|
323
|
7687 char_u *p;
|
|
|
7688 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
7689 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
7690
|
|
|
7691 if (byts == NULL)
|
|
|
7692 {
|
|
|
7693 /* array is empty: "cannot happen" */
|
|
|
7694 *kword = NUL;
|
|
|
7695 return;
|
|
|
7696 }
|
|
|
7697
|
|
|
7698 /* Make an all-cap version of "fword". */
|
|
|
7699 allcap_copy(fword, uword);
|
|
|
7700
|
|
|
7701 /*
|
|
|
7702 * Each character needs to be tried both case-folded and upper-case.
|
|
|
7703 * All this gets very complicated if we keep in mind that changing case
|
|
|
7704 * may change the byte length of a multi-byte character...
|
|
|
7705 */
|
|
|
7706 depth = 0;
|
|
|
7707 arridx[0] = 0;
|
|
|
7708 round[0] = 0;
|
|
|
7709 fwordidx[0] = 0;
|
|
|
7710 uwordidx[0] = 0;
|
|
|
7711 kwordlen[0] = 0;
|
|
|
7712 while (depth >= 0)
|
|
|
7713 {
|
|
|
7714 if (fword[fwordidx[depth]] == NUL)
|
|
|
7715 {
|
|
|
7716 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
7717 * here we have found a match. */
|
|
|
7718 if (byts[arridx[depth] + 1] == 0)
|
|
|
7719 {
|
|
|
7720 kword[kwordlen[depth]] = NUL;
|
|
|
7721 return;
|
|
|
7722 }
|
|
|
7723
|
|
|
7724 /* kword is getting too long, continue one level up */
|
|
|
7725 --depth;
|
|
|
7726 }
|
|
|
7727 else if (++round[depth] > 2)
|
|
|
7728 {
|
|
|
7729 /* tried both fold-case and upper-case character, continue one
|
|
|
7730 * level up */
|
|
|
7731 --depth;
|
|
|
7732 }
|
|
|
7733 else
|
|
|
7734 {
|
|
|
7735 /*
|
|
|
7736 * round[depth] == 1: Try using the folded-case character.
|
|
|
7737 * round[depth] == 2: Try using the upper-case character.
|
|
|
7738 */
|
|
|
7739 #ifdef FEAT_MBYTE
|
|
|
7740 if (has_mbyte)
|
|
|
7741 {
|
|
|
7742 flen = mb_ptr2len_check(fword + fwordidx[depth]);
|
|
|
7743 ulen = mb_ptr2len_check(uword + uwordidx[depth]);
|
|
|
7744 }
|
|
|
7745 else
|
|
|
7746 #endif
|
|
|
7747 ulen = flen = 1;
|
|
|
7748 if (round[depth] == 1)
|
|
|
7749 {
|
|
|
7750 p = fword + fwordidx[depth];
|
|
|
7751 l = flen;
|
|
|
7752 }
|
|
|
7753 else
|
|
|
7754 {
|
|
|
7755 p = uword + uwordidx[depth];
|
|
|
7756 l = ulen;
|
|
|
7757 }
|
|
|
7758
|
|
|
7759 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
7760 {
|
|
|
7761 /* Perform a binary search in the list of accepted bytes. */
|
|
|
7762 len = byts[tryidx++];
|
|
|
7763 c = *p++;
|
|
|
7764 lo = tryidx;
|
|
|
7765 hi = tryidx + len - 1;
|
|
|
7766 while (lo < hi)
|
|
|
7767 {
|
|
|
7768 m = (lo + hi) / 2;
|
|
|
7769 if (byts[m] > c)
|
|
|
7770 hi = m - 1;
|
|
|
7771 else if (byts[m] < c)
|
|
|
7772 lo = m + 1;
|
|
|
7773 else
|
|
|
7774 {
|
|
|
7775 lo = hi = m;
|
|
|
7776 break;
|
|
|
7777 }
|
|
|
7778 }
|
|
|
7779
|
|
|
7780 /* Stop if there is no matching byte. */
|
|
|
7781 if (hi < lo || byts[lo] != c)
|
|
|
7782 break;
|
|
|
7783
|
|
|
7784 /* Continue at the child (if there is one). */
|
|
|
7785 tryidx = idxs[lo];
|
|
|
7786 }
|
|
|
7787
|
|
|
7788 if (l == 0)
|
|
|
7789 {
|
|
|
7790 /*
|
|
|
7791 * Found the matching char. Copy it to "kword" and go a
|
|
|
7792 * level deeper.
|
|
|
7793 */
|
|
|
7794 if (round[depth] == 1)
|
|
|
7795 {
|
|
|
7796 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
7797 flen);
|
|
|
7798 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
7799 }
|
|
|
7800 else
|
|
|
7801 {
|
|
|
7802 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
7803 ulen);
|
|
|
7804 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
7805 }
|
|
|
7806 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
7807 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
7808
|
|
|
7809 ++depth;
|
|
|
7810 arridx[depth] = tryidx;
|
|
|
7811 round[depth] = 0;
|
|
|
7812 }
|
|
|
7813 }
|
|
|
7814 }
|
|
|
7815
|
|
|
7816 /* Didn't find it: "cannot happen". */
|
|
|
7817 *kword = NUL;
|
|
|
7818 }
|
|
|
7819
|
|
|
7820 /*
|
|
344
|
7821 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
7822 * su->su_sga.
|
|
|
7823 */
|
|
|
7824 static void
|
|
|
7825 score_comp_sal(su)
|
|
|
7826 suginfo_T *su;
|
|
|
7827 {
|
|
|
7828 langp_T *lp;
|
|
|
7829 char_u badsound[MAXWLEN];
|
|
|
7830 int i;
|
|
|
7831 suggest_T *stp;
|
|
|
7832 suggest_T *sstp;
|
|
|
7833 int score;
|
|
|
7834
|
|
|
7835 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
7836 return;
|
|
|
7837
|
|
|
7838 /* Use the sound-folding of the first language that supports it. */
|
|
|
7839 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7840 lp->lp_slang != NULL; ++lp)
|
|
|
7841 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7842 {
|
|
|
7843 /* soundfold the bad word */
|
|
375
|
7844 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
7845
|
|
|
7846 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7847 {
|
|
|
7848 stp = &SUG(su->su_ga, i);
|
|
|
7849
|
|
351
|
7850 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
7851 * sound-a-like score. */
|
|
|
7852 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
7853 if (score < SCORE_MAXMAX)
|
|
|
7854 {
|
|
|
7855 /* Add the suggestion. */
|
|
|
7856 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
7857 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
7858 if (sstp->st_word != NULL)
|
|
|
7859 {
|
|
|
7860 sstp->st_score = score;
|
|
|
7861 sstp->st_altscore = 0;
|
|
|
7862 sstp->st_orglen = stp->st_orglen;
|
|
|
7863 ++su->su_sga.ga_len;
|
|
|
7864 }
|
|
|
7865 }
|
|
|
7866 }
|
|
|
7867 break;
|
|
|
7868 }
|
|
|
7869 }
|
|
|
7870
|
|
|
7871 /*
|
|
|
7872 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
7873 * They are intwined.
|
|
|
7874 */
|
|
|
7875 static void
|
|
|
7876 score_combine(su)
|
|
|
7877 suginfo_T *su;
|
|
|
7878 {
|
|
|
7879 int i;
|
|
|
7880 int j;
|
|
|
7881 garray_T ga;
|
|
|
7882 garray_T *gap;
|
|
|
7883 langp_T *lp;
|
|
|
7884 suggest_T *stp;
|
|
|
7885 char_u *p;
|
|
|
7886 char_u badsound[MAXWLEN];
|
|
|
7887 int round;
|
|
|
7888
|
|
|
7889 /* Add the alternate score to su_ga. */
|
|
|
7890 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7891 lp->lp_slang != NULL; ++lp)
|
|
|
7892 {
|
|
|
7893 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
7894 {
|
|
|
7895 /* soundfold the bad word */
|
|
375
|
7896 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
7897
|
|
|
7898 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7899 {
|
|
|
7900 stp = &SUG(su->su_ga, i);
|
|
351
|
7901 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
7902 badsound);
|
|
344
|
7903 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
7904 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
7905 else
|
|
|
7906 stp->st_score = (stp->st_score * 3
|
|
|
7907 + stp->st_altscore) / 4;
|
|
|
7908 stp->st_salscore = FALSE;
|
|
|
7909 }
|
|
|
7910 break;
|
|
|
7911 }
|
|
|
7912 }
|
|
|
7913
|
|
|
7914 /* Add the alternate score to su_sga. */
|
|
|
7915 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
7916 {
|
|
|
7917 stp = &SUG(su->su_sga, i);
|
|
|
7918 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
7919 if (stp->st_score == SCORE_MAXMAX)
|
|
|
7920 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
7921 else
|
|
|
7922 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
7923 stp->st_salscore = TRUE;
|
|
|
7924 }
|
|
|
7925
|
|
|
7926 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
7927 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7928 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
7929
|
|
|
7930 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
7931 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
7932 return;
|
|
|
7933
|
|
|
7934 stp = &SUG(ga, 0);
|
|
|
7935 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
7936 {
|
|
|
7937 /* round 1: get a suggestion from su_ga
|
|
|
7938 * round 2: get a suggestion from su_sga */
|
|
|
7939 for (round = 1; round <= 2; ++round)
|
|
|
7940 {
|
|
|
7941 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
7942 if (i < gap->ga_len)
|
|
|
7943 {
|
|
|
7944 /* Don't add a word if it's already there. */
|
|
|
7945 p = SUG(*gap, i).st_word;
|
|
|
7946 for (j = 0; j < ga.ga_len; ++j)
|
|
|
7947 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
7948 break;
|
|
|
7949 if (j == ga.ga_len)
|
|
|
7950 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
7951 else
|
|
|
7952 vim_free(p);
|
|
|
7953 }
|
|
|
7954 }
|
|
|
7955 }
|
|
|
7956
|
|
|
7957 ga_clear(&su->su_ga);
|
|
|
7958 ga_clear(&su->su_sga);
|
|
|
7959
|
|
|
7960 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
7961 if (ga.ga_len > su->su_maxcount)
|
|
|
7962 {
|
|
|
7963 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
7964 vim_free(stp[i].st_word);
|
|
|
7965 ga.ga_len = su->su_maxcount;
|
|
|
7966 }
|
|
|
7967
|
|
|
7968 su->su_ga = ga;
|
|
|
7969 }
|
|
|
7970
|
|
|
7971 /*
|
|
351
|
7972 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
7973 * badword.
|
|
|
7974 */
|
|
|
7975 static int
|
|
|
7976 stp_sal_score(stp, su, slang, badsound)
|
|
|
7977 suggest_T *stp;
|
|
|
7978 suginfo_T *su;
|
|
|
7979 slang_T *slang;
|
|
|
7980 char_u *badsound; /* sound-folded badword */
|
|
|
7981 {
|
|
|
7982 char_u *p;
|
|
|
7983 char_u badsound2[MAXWLEN];
|
|
|
7984 char_u fword[MAXWLEN];
|
|
|
7985 char_u goodsound[MAXWLEN];
|
|
|
7986
|
|
|
7987 if (stp->st_orglen <= su->su_badlen)
|
|
|
7988 p = badsound;
|
|
|
7989 else
|
|
|
7990 {
|
|
|
7991 /* soundfold the bad word with more characters following */
|
|
|
7992 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
7993
|
|
|
7994 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
7995 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
7996 * removing the space. Don't do it when the good word also contains a
|
|
|
7997 * space. */
|
|
|
7998 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
7999 && *skiptowhite(stp->st_word) == NUL)
|
|
|
8000 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
8001 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
8002
|
|
375
|
8003 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
8004 p = badsound2;
|
|
|
8005 }
|
|
|
8006
|
|
375
|
8007 /* Sound-fold the word and compute the score for the difference. */
|
|
|
8008 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
8009
|
|
|
8010 return soundalike_score(goodsound, p);
|
|
|
8011 }
|
|
|
8012
|
|
|
8013 /*
|
|
323
|
8014 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
8015 */
|
|
|
8016 static void
|
|
346
|
8017 suggest_try_soundalike(su)
|
|
323
|
8018 suginfo_T *su;
|
|
|
8019 {
|
|
|
8020 char_u salword[MAXWLEN];
|
|
|
8021 char_u tword[MAXWLEN];
|
|
|
8022 char_u tsalword[MAXWLEN];
|
|
324
|
8023 idx_T arridx[MAXWLEN];
|
|
323
|
8024 int curi[MAXWLEN];
|
|
|
8025 langp_T *lp;
|
|
|
8026 char_u *byts;
|
|
324
|
8027 idx_T *idxs;
|
|
323
|
8028 int depth;
|
|
|
8029 int c;
|
|
324
|
8030 idx_T n;
|
|
323
|
8031 int round;
|
|
|
8032 int flags;
|
|
344
|
8033 int sound_score;
|
|
|
8034
|
|
|
8035 /* Do this for all languages that support sound folding. */
|
|
323
|
8036 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8037 lp->lp_slang != NULL; ++lp)
|
|
|
8038 {
|
|
|
8039 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8040 {
|
|
|
8041 /* soundfold the bad word */
|
|
375
|
8042 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
8043
|
|
|
8044 /*
|
|
|
8045 * Go through the whole tree, soundfold each word and compare.
|
|
|
8046 * round 1: use the case-folded tree.
|
|
|
8047 * round 2: use the keep-case tree.
|
|
|
8048 */
|
|
|
8049 for (round = 1; round <= 2; ++round)
|
|
|
8050 {
|
|
|
8051 if (round == 1)
|
|
|
8052 {
|
|
|
8053 byts = lp->lp_slang->sl_fbyts;
|
|
|
8054 idxs = lp->lp_slang->sl_fidxs;
|
|
|
8055 }
|
|
|
8056 else
|
|
|
8057 {
|
|
|
8058 byts = lp->lp_slang->sl_kbyts;
|
|
|
8059 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
8060 if (byts == NULL) /* no keep-case words */
|
|
|
8061 continue;
|
|
323
|
8062 }
|
|
|
8063
|
|
|
8064 depth = 0;
|
|
|
8065 arridx[0] = 0;
|
|
|
8066 curi[0] = 1;
|
|
|
8067 while (depth >= 0 && !got_int)
|
|
|
8068 {
|
|
|
8069 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
8070 {
|
|
323
|
8071 /* Done all bytes at this node, go up one level. */
|
|
|
8072 --depth;
|
|
351
|
8073 line_breakcheck();
|
|
|
8074 }
|
|
323
|
8075 else
|
|
|
8076 {
|
|
|
8077 /* Do one more byte at this node. */
|
|
|
8078 n = arridx[depth] + curi[depth];
|
|
|
8079 ++curi[depth];
|
|
|
8080 c = byts[n];
|
|
|
8081 if (c == 0)
|
|
|
8082 {
|
|
|
8083 /* End of word, deal with the word. */
|
|
324
|
8084 flags = (int)idxs[n];
|
|
323
|
8085 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
8086 {
|
|
|
8087 tword[depth] = NUL;
|
|
375
|
8088 /* Sound-fold. Only in keep-case tree need to
|
|
|
8089 * case-fold the word. */
|
|
|
8090 spell_soundfold(lp->lp_slang, tword,
|
|
|
8091 round == 1, tsalword);
|
|
323
|
8092
|
|
344
|
8093 /* Compute the edit distance between the
|
|
|
8094 * sound-a-like words. */
|
|
|
8095 sound_score = soundalike_score(salword,
|
|
|
8096 tsalword);
|
|
324
|
8097 if (sound_score < SCORE_MAXMAX)
|
|
323
|
8098 {
|
|
324
|
8099 char_u cword[MAXWLEN];
|
|
|
8100 char_u *p;
|
|
344
|
8101 int score;
|
|
324
|
8102
|
|
388
|
8103 flags |= su->su_badflags;
|
|
351
|
8104 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
8105 {
|
|
324
|
8106 /* Need to fix case according to
|
|
|
8107 * "flags". */
|
|
323
|
8108 make_case_word(tword, cword, flags);
|
|
324
|
8109 p = cword;
|
|
323
|
8110 }
|
|
|
8111 else
|
|
324
|
8112 p = tword;
|
|
|
8113
|
|
344
|
8114 if (sps_flags & SPS_DOUBLE)
|
|
|
8115 add_suggestion(su, &su->su_sga, p,
|
|
346
|
8116 su->su_badlen,
|
|
351
|
8117 sound_score, 0, FALSE);
|
|
344
|
8118 else
|
|
|
8119 {
|
|
|
8120 /* Compute the score. */
|
|
|
8121 score = spell_edit_score(
|
|
|
8122 su->su_badword, p);
|
|
|
8123 if (sps_flags & SPS_BEST)
|
|
|
8124 /* give a bonus for the good word
|
|
|
8125 * sounding the same as the bad
|
|
|
8126 * word */
|
|
|
8127 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8128 su->su_badlen,
|
|
344
|
8129 RESCORE(score, sound_score),
|
|
351
|
8130 sound_score, TRUE);
|
|
344
|
8131 else
|
|
|
8132 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8133 su->su_badlen,
|
|
351
|
8134 score + sound_score, 0, FALSE);
|
|
344
|
8135 }
|
|
323
|
8136 }
|
|
|
8137 }
|
|
|
8138
|
|
|
8139 /* Skip over other NUL bytes. */
|
|
|
8140 while (byts[n + 1] == 0)
|
|
|
8141 {
|
|
|
8142 ++n;
|
|
|
8143 ++curi[depth];
|
|
|
8144 }
|
|
|
8145 }
|
|
|
8146 else
|
|
|
8147 {
|
|
|
8148 /* Normal char, go one level deeper. */
|
|
|
8149 tword[depth++] = c;
|
|
|
8150 arridx[depth] = idxs[n];
|
|
|
8151 curi[depth] = 1;
|
|
|
8152 }
|
|
|
8153 }
|
|
|
8154 }
|
|
|
8155 }
|
|
|
8156 }
|
|
|
8157 }
|
|
|
8158 }
|
|
|
8159
|
|
|
8160 /*
|
|
324
|
8161 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
8162 */
|
|
|
8163 static void
|
|
|
8164 make_case_word(fword, cword, flags)
|
|
|
8165 char_u *fword;
|
|
|
8166 char_u *cword;
|
|
|
8167 int flags;
|
|
|
8168 {
|
|
|
8169 if (flags & WF_ALLCAP)
|
|
|
8170 /* Make it all upper-case */
|
|
|
8171 allcap_copy(fword, cword);
|
|
|
8172 else if (flags & WF_ONECAP)
|
|
|
8173 /* Make the first letter upper-case */
|
|
324
|
8174 onecap_copy(fword, cword, TRUE);
|
|
323
|
8175 else
|
|
|
8176 /* Use goodword as-is. */
|
|
|
8177 STRCPY(cword, fword);
|
|
|
8178 }
|
|
|
8179
|
|
330
|
8180 /*
|
|
|
8181 * Use map string "map" for languages "lp".
|
|
|
8182 */
|
|
|
8183 static void
|
|
|
8184 set_map_str(lp, map)
|
|
|
8185 slang_T *lp;
|
|
|
8186 char_u *map;
|
|
|
8187 {
|
|
|
8188 char_u *p;
|
|
|
8189 int headc = 0;
|
|
|
8190 int c;
|
|
|
8191 int i;
|
|
|
8192
|
|
|
8193 if (*map == NUL)
|
|
|
8194 {
|
|
|
8195 lp->sl_has_map = FALSE;
|
|
|
8196 return;
|
|
|
8197 }
|
|
|
8198 lp->sl_has_map = TRUE;
|
|
|
8199
|
|
|
8200 /* Init the array and hash table empty. */
|
|
|
8201 for (i = 0; i < 256; ++i)
|
|
|
8202 lp->sl_map_array[i] = 0;
|
|
|
8203 #ifdef FEAT_MBYTE
|
|
|
8204 hash_init(&lp->sl_map_hash);
|
|
|
8205 #endif
|
|
|
8206
|
|
|
8207 /*
|
|
|
8208 * The similar characters are stored separated with slashes:
|
|
|
8209 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
8210 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
8211 */
|
|
|
8212 for (p = map; *p != NUL; )
|
|
|
8213 {
|
|
|
8214 #ifdef FEAT_MBYTE
|
|
|
8215 c = mb_ptr2char_adv(&p);
|
|
|
8216 #else
|
|
|
8217 c = *p++;
|
|
|
8218 #endif
|
|
|
8219 if (c == '/')
|
|
|
8220 headc = 0;
|
|
|
8221 else
|
|
|
8222 {
|
|
|
8223 if (headc == 0)
|
|
|
8224 headc = c;
|
|
|
8225
|
|
|
8226 #ifdef FEAT_MBYTE
|
|
|
8227 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
8228 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
8229 * a NUL. */
|
|
|
8230 if (c >= 256)
|
|
|
8231 {
|
|
|
8232 int cl = mb_char2len(c);
|
|
|
8233 int headcl = mb_char2len(headc);
|
|
|
8234 char_u *b;
|
|
|
8235 hash_T hash;
|
|
|
8236 hashitem_T *hi;
|
|
|
8237
|
|
|
8238 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
8239 if (b == NULL)
|
|
|
8240 return;
|
|
|
8241 mb_char2bytes(c, b);
|
|
|
8242 b[cl] = NUL;
|
|
|
8243 mb_char2bytes(headc, b + cl + 1);
|
|
|
8244 b[cl + 1 + headcl] = NUL;
|
|
|
8245 hash = hash_hash(b);
|
|
|
8246 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
8247 if (HASHITEM_EMPTY(hi))
|
|
|
8248 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
8249 else
|
|
|
8250 {
|
|
|
8251 /* This should have been checked when generating the .spl
|
|
|
8252 * file. */
|
|
|
8253 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
8254 vim_free(b);
|
|
|
8255 }
|
|
|
8256 }
|
|
|
8257 else
|
|
|
8258 #endif
|
|
|
8259 lp->sl_map_array[c] = headc;
|
|
|
8260 }
|
|
|
8261 }
|
|
|
8262 }
|
|
|
8263
|
|
323
|
8264 /*
|
|
|
8265 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
8266 * lines in the .aff file.
|
|
|
8267 */
|
|
|
8268 static int
|
|
|
8269 similar_chars(slang, c1, c2)
|
|
|
8270 slang_T *slang;
|
|
|
8271 int c1;
|
|
|
8272 int c2;
|
|
|
8273 {
|
|
330
|
8274 int m1, m2;
|
|
|
8275 #ifdef FEAT_MBYTE
|
|
|
8276 char_u buf[MB_MAXBYTES];
|
|
|
8277 hashitem_T *hi;
|
|
|
8278
|
|
|
8279 if (c1 >= 256)
|
|
|
8280 {
|
|
|
8281 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
8282 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8283 if (HASHITEM_EMPTY(hi))
|
|
|
8284 m1 = 0;
|
|
|
8285 else
|
|
|
8286 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8287 }
|
|
|
8288 else
|
|
|
8289 #endif
|
|
|
8290 m1 = slang->sl_map_array[c1];
|
|
|
8291 if (m1 == 0)
|
|
323
|
8292 return FALSE;
|
|
330
|
8293
|
|
|
8294
|
|
|
8295 #ifdef FEAT_MBYTE
|
|
|
8296 if (c2 >= 256)
|
|
|
8297 {
|
|
|
8298 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
8299 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8300 if (HASHITEM_EMPTY(hi))
|
|
|
8301 m2 = 0;
|
|
|
8302 else
|
|
|
8303 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8304 }
|
|
|
8305 else
|
|
|
8306 #endif
|
|
|
8307 m2 = slang->sl_map_array[c2];
|
|
|
8308
|
|
|
8309 return m1 == m2;
|
|
323
|
8310 }
|
|
|
8311
|
|
|
8312 /*
|
|
|
8313 * Add a suggestion to the list of suggestions.
|
|
|
8314 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
8315 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
8316 * with spell_edit_score().
|
|
|
8317 */
|
|
|
8318 static void
|
|
351
|
8319 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
8320 suginfo_T *su;
|
|
344
|
8321 garray_T *gap;
|
|
323
|
8322 char_u *goodword;
|
|
346
|
8323 int badlen; /* length of bad word used */
|
|
324
|
8324 int score;
|
|
351
|
8325 int altscore;
|
|
344
|
8326 int had_bonus; /* value for st_had_bonus */
|
|
323
|
8327 {
|
|
|
8328 suggest_T *stp;
|
|
|
8329 int i;
|
|
346
|
8330 char_u *p = NULL;
|
|
|
8331 int c = 0;
|
|
323
|
8332
|
|
|
8333 /* Check that the word wasn't banned. */
|
|
|
8334 if (was_banned(su, goodword))
|
|
|
8335 return;
|
|
|
8336
|
|
346
|
8337 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
8338 * Remove the common part from "goodword". */
|
|
|
8339 i = badlen - su->su_badlen;
|
|
|
8340 if (i > 0)
|
|
|
8341 {
|
|
|
8342 /* This assumes there was no case folding or it didn't change the
|
|
|
8343 * length... */
|
|
|
8344 p = goodword + STRLEN(goodword) - i;
|
|
|
8345 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
8346 {
|
|
|
8347 badlen = su->su_badlen;
|
|
|
8348 c = *p;
|
|
|
8349 *p = NUL;
|
|
|
8350 }
|
|
|
8351 else
|
|
|
8352 p = NULL;
|
|
|
8353 }
|
|
372
|
8354 else if (i < 0)
|
|
|
8355 {
|
|
|
8356 /* When replacing part of the word check that we actually change
|
|
|
8357 * something. For "the the" a suggestion can be replacing the first
|
|
|
8358 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
8359 if (badlen == (int)STRLEN(goodword)
|
|
372
|
8360 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
8361 return;
|
|
|
8362 }
|
|
|
8363
|
|
346
|
8364
|
|
323
|
8365 if (score <= su->su_maxscore)
|
|
|
8366 {
|
|
366
|
8367 /* Check if the word is already there. Also check the length that is
|
|
|
8368 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
8369 * "thes" -> "these". */
|
|
344
|
8370 stp = &SUG(*gap, 0);
|
|
|
8371 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
8372 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
8373 && stp[i].st_orglen == badlen)
|
|
323
|
8374 {
|
|
|
8375 /* Found it. Remember the lowest score. */
|
|
|
8376 if (stp[i].st_score > score)
|
|
324
|
8377 {
|
|
323
|
8378 stp[i].st_score = score;
|
|
376
|
8379 stp[i].st_altscore = altscore;
|
|
324
|
8380 stp[i].st_had_bonus = had_bonus;
|
|
|
8381 }
|
|
323
|
8382 break;
|
|
|
8383 }
|
|
|
8384
|
|
344
|
8385 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
8386 {
|
|
|
8387 /* Add a suggestion. */
|
|
344
|
8388 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
8389 stp->st_word = vim_strsave(goodword);
|
|
|
8390 if (stp->st_word != NULL)
|
|
|
8391 {
|
|
|
8392 stp->st_score = score;
|
|
351
|
8393 stp->st_altscore = altscore;
|
|
324
|
8394 stp->st_had_bonus = had_bonus;
|
|
346
|
8395 stp->st_orglen = badlen;
|
|
344
|
8396 ++gap->ga_len;
|
|
323
|
8397
|
|
|
8398 /* If we have too many suggestions now, sort the list and keep
|
|
|
8399 * the best suggestions. */
|
|
344
|
8400 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
8401 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
8402 SUG_CLEAN_COUNT(su));
|
|
323
|
8403 }
|
|
|
8404 }
|
|
|
8405 }
|
|
346
|
8406
|
|
|
8407 if (p != NULL)
|
|
|
8408 *p = c; /* restore "goodword" */
|
|
323
|
8409 }
|
|
|
8410
|
|
|
8411 /*
|
|
|
8412 * Add a word to be banned.
|
|
|
8413 */
|
|
|
8414 static void
|
|
|
8415 add_banned(su, word)
|
|
|
8416 suginfo_T *su;
|
|
|
8417 char_u *word;
|
|
|
8418 {
|
|
|
8419 char_u *s = vim_strsave(word);
|
|
|
8420 hash_T hash;
|
|
|
8421 hashitem_T *hi;
|
|
|
8422
|
|
|
8423 if (s != NULL)
|
|
|
8424 {
|
|
|
8425 hash = hash_hash(s);
|
|
|
8426 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
8427 if (HASHITEM_EMPTY(hi))
|
|
|
8428 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
8429 else
|
|
|
8430 vim_free(s);
|
|
323
|
8431 }
|
|
|
8432 }
|
|
|
8433
|
|
|
8434 /*
|
|
|
8435 * Return TRUE if a word appears in the list of banned words.
|
|
|
8436 */
|
|
|
8437 static int
|
|
|
8438 was_banned(su, word)
|
|
|
8439 suginfo_T *su;
|
|
|
8440 char_u *word;
|
|
|
8441 {
|
|
324
|
8442 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
8443
|
|
|
8444 return !HASHITEM_EMPTY(hi);
|
|
323
|
8445 }
|
|
|
8446
|
|
|
8447 /*
|
|
|
8448 * Free the banned words in "su".
|
|
|
8449 */
|
|
|
8450 static void
|
|
|
8451 free_banned(su)
|
|
|
8452 suginfo_T *su;
|
|
|
8453 {
|
|
|
8454 int todo;
|
|
|
8455 hashitem_T *hi;
|
|
|
8456
|
|
|
8457 todo = su->su_banned.ht_used;
|
|
|
8458 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
8459 {
|
|
|
8460 if (!HASHITEM_EMPTY(hi))
|
|
|
8461 {
|
|
|
8462 vim_free(hi->hi_key);
|
|
|
8463 --todo;
|
|
|
8464 }
|
|
|
8465 }
|
|
|
8466 hash_clear(&su->su_banned);
|
|
|
8467 }
|
|
|
8468
|
|
324
|
8469 /*
|
|
|
8470 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
8471 * thus only done for the final results.
|
|
|
8472 */
|
|
|
8473 static void
|
|
|
8474 rescore_suggestions(su)
|
|
|
8475 suginfo_T *su;
|
|
|
8476 {
|
|
|
8477 langp_T *lp;
|
|
|
8478 suggest_T *stp;
|
|
|
8479 char_u sal_badword[MAXWLEN];
|
|
|
8480 int i;
|
|
|
8481
|
|
|
8482 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8483 lp->lp_slang != NULL; ++lp)
|
|
|
8484 {
|
|
|
8485 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8486 {
|
|
|
8487 /* soundfold the bad word */
|
|
375
|
8488 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
8489
|
|
|
8490 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8491 {
|
|
344
|
8492 stp = &SUG(su->su_ga, i);
|
|
324
|
8493 if (!stp->st_had_bonus)
|
|
|
8494 {
|
|
351
|
8495 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
8496 lp->lp_slang, sal_badword);
|
|
|
8497 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8498 stp->st_altscore = SCORE_BIG;
|
|
|
8499 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
8500 }
|
|
|
8501 }
|
|
|
8502 break;
|
|
|
8503 }
|
|
|
8504 }
|
|
|
8505 }
|
|
|
8506
|
|
323
|
8507 static int
|
|
|
8508 #ifdef __BORLANDC__
|
|
|
8509 _RTLENTRYF
|
|
|
8510 #endif
|
|
|
8511 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
8512
|
|
|
8513 /*
|
|
|
8514 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
8515 */
|
|
|
8516 static int
|
|
|
8517 #ifdef __BORLANDC__
|
|
|
8518 _RTLENTRYF
|
|
|
8519 #endif
|
|
|
8520 sug_compare(s1, s2)
|
|
|
8521 const void *s1;
|
|
|
8522 const void *s2;
|
|
|
8523 {
|
|
|
8524 suggest_T *p1 = (suggest_T *)s1;
|
|
|
8525 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
8526 int n = p1->st_score - p2->st_score;
|
|
|
8527
|
|
|
8528 if (n == 0)
|
|
|
8529 return p1->st_altscore - p2->st_altscore;
|
|
|
8530 return n;
|
|
323
|
8531 }
|
|
|
8532
|
|
|
8533 /*
|
|
|
8534 * Cleanup the suggestions:
|
|
|
8535 * - Sort on score.
|
|
|
8536 * - Remove words that won't be displayed.
|
|
344
|
8537 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
8538 */
|
|
344
|
8539 static int
|
|
|
8540 cleanup_suggestions(gap, maxscore, keep)
|
|
|
8541 garray_T *gap;
|
|
|
8542 int maxscore;
|
|
324
|
8543 int keep; /* nr of suggestions to keep */
|
|
323
|
8544 {
|
|
344
|
8545 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
8546 int i;
|
|
|
8547
|
|
|
8548 /* Sort the list. */
|
|
344
|
8549 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
8550
|
|
|
8551 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
8552 if (gap->ga_len > keep)
|
|
323
|
8553 {
|
|
344
|
8554 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
8555 vim_free(stp[i].st_word);
|
|
344
|
8556 gap->ga_len = keep;
|
|
|
8557 return stp[keep - 1].st_score;
|
|
323
|
8558 }
|
|
344
|
8559 return maxscore;
|
|
323
|
8560 }
|
|
|
8561
|
|
372
|
8562 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
8563 /*
|
|
|
8564 * Soundfold a string, for soundfold().
|
|
|
8565 * Result is in allocated memory, NULL for an error.
|
|
|
8566 */
|
|
|
8567 char_u *
|
|
|
8568 eval_soundfold(word)
|
|
|
8569 char_u *word;
|
|
|
8570 {
|
|
|
8571 langp_T *lp;
|
|
|
8572 char_u sound[MAXWLEN];
|
|
|
8573
|
|
|
8574 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
8575 /* Use the sound-folding of the first language that supports it. */
|
|
|
8576 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8577 lp->lp_slang != NULL; ++lp)
|
|
|
8578 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8579 {
|
|
|
8580 /* soundfold the word */
|
|
375
|
8581 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
8582 return vim_strsave(sound);
|
|
|
8583 }
|
|
|
8584
|
|
|
8585 /* No language with sound folding, return word as-is. */
|
|
|
8586 return vim_strsave(word);
|
|
|
8587 }
|
|
|
8588 #endif
|
|
|
8589
|
|
323
|
8590 /*
|
|
|
8591 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
8592 */
|
|
|
8593 static void
|
|
375
|
8594 spell_soundfold(slang, inword, folded, res)
|
|
|
8595 slang_T *slang;
|
|
|
8596 char_u *inword;
|
|
|
8597 int folded; /* "inword" is already case-folded */
|
|
|
8598 char_u *res;
|
|
|
8599 {
|
|
|
8600 char_u fword[MAXWLEN];
|
|
|
8601 char_u *word;
|
|
|
8602
|
|
|
8603 if (slang->sl_sofo)
|
|
|
8604 /* SOFOFROM and SOFOTO used */
|
|
|
8605 spell_soundfold_sofo(slang, inword, res);
|
|
|
8606 else
|
|
|
8607 {
|
|
|
8608 /* SAL items used. Requires the word to be case-folded. */
|
|
|
8609 if (folded)
|
|
|
8610 word = inword;
|
|
|
8611 else
|
|
|
8612 {
|
|
|
8613 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
8614 word = fword;
|
|
|
8615 }
|
|
|
8616
|
|
|
8617 #ifdef FEAT_MBYTE
|
|
|
8618 if (has_mbyte)
|
|
|
8619 spell_soundfold_wsal(slang, word, res);
|
|
|
8620 else
|
|
|
8621 #endif
|
|
|
8622 spell_soundfold_sal(slang, word, res);
|
|
|
8623 }
|
|
|
8624 }
|
|
|
8625
|
|
|
8626 /*
|
|
|
8627 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
8628 * SOFOTO lines.
|
|
|
8629 */
|
|
|
8630 static void
|
|
|
8631 spell_soundfold_sofo(slang, inword, res)
|
|
|
8632 slang_T *slang;
|
|
|
8633 char_u *inword;
|
|
|
8634 char_u *res;
|
|
|
8635 {
|
|
|
8636 char_u *s;
|
|
|
8637 int ri = 0;
|
|
|
8638 int c;
|
|
|
8639
|
|
|
8640 #ifdef FEAT_MBYTE
|
|
|
8641 if (has_mbyte)
|
|
|
8642 {
|
|
|
8643 int prevc = 0;
|
|
|
8644 int *ip;
|
|
|
8645
|
|
|
8646 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
8647 * 255, sl_sal the rest. */
|
|
|
8648 for (s = inword; *s != NUL; )
|
|
|
8649 {
|
|
|
8650 c = mb_ptr2char_adv(&s);
|
|
|
8651 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
8652 c = ' ';
|
|
|
8653 else if (c < 256)
|
|
|
8654 c = slang->sl_sal_first[c];
|
|
|
8655 else
|
|
|
8656 {
|
|
|
8657 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
8658 if (ip == NULL) /* empty list, can't match */
|
|
|
8659 c = NUL;
|
|
|
8660 else
|
|
|
8661 for (;;) /* find "c" in the list */
|
|
|
8662 {
|
|
|
8663 if (*ip == 0) /* not found */
|
|
|
8664 {
|
|
|
8665 c = NUL;
|
|
|
8666 break;
|
|
|
8667 }
|
|
|
8668 if (*ip == c) /* match! */
|
|
|
8669 {
|
|
|
8670 c = ip[1];
|
|
|
8671 break;
|
|
|
8672 }
|
|
|
8673 ip += 2;
|
|
|
8674 }
|
|
|
8675 }
|
|
|
8676
|
|
|
8677 if (c != NUL && c != prevc)
|
|
|
8678 {
|
|
|
8679 ri += mb_char2bytes(c, res + ri);
|
|
|
8680 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
8681 break;
|
|
|
8682 prevc = c;
|
|
|
8683 }
|
|
|
8684 }
|
|
|
8685 }
|
|
|
8686 else
|
|
|
8687 #endif
|
|
|
8688 {
|
|
|
8689 /* The sl_sal_first[] table contains the translation. */
|
|
|
8690 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
8691 {
|
|
|
8692 if (vim_iswhite(c))
|
|
|
8693 c = ' ';
|
|
|
8694 else
|
|
|
8695 c = slang->sl_sal_first[c];
|
|
|
8696 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
8697 res[ri++] = c;
|
|
|
8698 }
|
|
|
8699 }
|
|
|
8700
|
|
|
8701 res[ri] = NUL;
|
|
|
8702 }
|
|
|
8703
|
|
|
8704 static void
|
|
|
8705 spell_soundfold_sal(slang, inword, res)
|
|
323
|
8706 slang_T *slang;
|
|
|
8707 char_u *inword;
|
|
|
8708 char_u *res;
|
|
|
8709 {
|
|
344
|
8710 salitem_T *smp;
|
|
323
|
8711 char_u word[MAXWLEN];
|
|
375
|
8712 char_u *s = inword;
|
|
323
|
8713 char_u *t;
|
|
344
|
8714 char_u *pf;
|
|
323
|
8715 int i, j, z;
|
|
344
|
8716 int reslen;
|
|
323
|
8717 int n, k = 0;
|
|
|
8718 int z0;
|
|
|
8719 int k0;
|
|
|
8720 int n0;
|
|
|
8721 int c;
|
|
|
8722 int pri;
|
|
|
8723 int p0 = -333;
|
|
|
8724 int c0;
|
|
|
8725
|
|
324
|
8726 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
8727 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
8728 if (slang->sl_rem_accents)
|
|
|
8729 {
|
|
|
8730 t = word;
|
|
375
|
8731 while (*s != NUL)
|
|
323
|
8732 {
|
|
324
|
8733 if (vim_iswhite(*s))
|
|
344
|
8734 {
|
|
|
8735 *t++ = ' ';
|
|
|
8736 s = skipwhite(s);
|
|
|
8737 }
|
|
323
|
8738 else
|
|
|
8739 {
|
|
376
|
8740 if (spell_iswordp_nmw(s))
|
|
323
|
8741 *t++ = *s;
|
|
|
8742 ++s;
|
|
|
8743 }
|
|
|
8744 }
|
|
|
8745 *t = NUL;
|
|
|
8746 }
|
|
|
8747 else
|
|
375
|
8748 STRCPY(word, s);
|
|
323
|
8749
|
|
344
|
8750 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
8751
|
|
|
8752 /*
|
|
|
8753 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
8754 * Changed to keep spaces.
|
|
323
|
8755 */
|
|
344
|
8756 i = reslen = z = 0;
|
|
323
|
8757 while ((c = word[i]) != NUL)
|
|
|
8758 {
|
|
344
|
8759 /* Start with the first rule that has the character in the word. */
|
|
323
|
8760 n = slang->sl_sal_first[c];
|
|
|
8761 z0 = 0;
|
|
|
8762
|
|
|
8763 if (n >= 0)
|
|
|
8764 {
|
|
|
8765 /* check all rules for the same letter */
|
|
344
|
8766 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
8767 {
|
|
344
|
8768 /* Quickly skip entries that don't match the word. Most
|
|
|
8769 * entries are less then three chars, optimize for that. */
|
|
|
8770 k = smp[n].sm_leadlen;
|
|
|
8771 if (k > 1)
|
|
323
|
8772 {
|
|
344
|
8773 if (word[i + 1] != s[1])
|
|
|
8774 continue;
|
|
|
8775 if (k > 2)
|
|
|
8776 {
|
|
|
8777 for (j = 2; j < k; ++j)
|
|
|
8778 if (word[i + j] != s[j])
|
|
|
8779 break;
|
|
|
8780 if (j < k)
|
|
|
8781 continue;
|
|
|
8782 }
|
|
323
|
8783 }
|
|
|
8784
|
|
375
|
8785 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
8786 {
|
|
375
|
8787 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
8788 while (*pf != NUL && *pf != word[i + k])
|
|
|
8789 ++pf;
|
|
|
8790 if (*pf == NUL)
|
|
|
8791 continue;
|
|
|
8792 ++k;
|
|
323
|
8793 }
|
|
344
|
8794 s = smp[n].sm_rules;
|
|
|
8795 pri = 5; /* default priority */
|
|
323
|
8796
|
|
|
8797 p0 = *s;
|
|
|
8798 k0 = k;
|
|
|
8799 while (*s == '-' && k > 1)
|
|
|
8800 {
|
|
|
8801 k--;
|
|
|
8802 s++;
|
|
|
8803 }
|
|
|
8804 if (*s == '<')
|
|
|
8805 s++;
|
|
344
|
8806 if (VIM_ISDIGIT(*s))
|
|
323
|
8807 {
|
|
|
8808 /* determine priority */
|
|
|
8809 pri = *s - '0';
|
|
|
8810 s++;
|
|
|
8811 }
|
|
|
8812 if (*s == '^' && *(s + 1) == '^')
|
|
|
8813 s++;
|
|
|
8814
|
|
|
8815 if (*s == NUL
|
|
|
8816 || (*s == '^'
|
|
324
|
8817 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
8818 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
8819 && (*(s + 1) != '$'
|
|
376
|
8820 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
8821 || (*s == '$' && i > 0
|
|
376
|
8822 && spell_iswordp(word + i - 1, curbuf)
|
|
|
8823 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
8824 {
|
|
|
8825 /* search for followup rules, if: */
|
|
|
8826 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
8827 c0 = word[i + k - 1];
|
|
|
8828 n0 = slang->sl_sal_first[c0];
|
|
|
8829
|
|
|
8830 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
8831 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
8832 {
|
|
|
8833 /* test follow-up rule for "word[i + k]" */
|
|
344
|
8834 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
8835 {
|
|
344
|
8836 /* Quickly skip entries that don't match the word.
|
|
|
8837 * */
|
|
|
8838 k0 = smp[n0].sm_leadlen;
|
|
|
8839 if (k0 > 1)
|
|
323
|
8840 {
|
|
344
|
8841 if (word[i + k] != s[1])
|
|
|
8842 continue;
|
|
|
8843 if (k0 > 2)
|
|
|
8844 {
|
|
|
8845 pf = word + i + k + 1;
|
|
|
8846 for (j = 2; j < k0; ++j)
|
|
|
8847 if (*pf++ != s[j])
|
|
|
8848 break;
|
|
|
8849 if (j < k0)
|
|
|
8850 continue;
|
|
|
8851 }
|
|
323
|
8852 }
|
|
344
|
8853 k0 += k - 1;
|
|
|
8854
|
|
375
|
8855 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
8856 {
|
|
344
|
8857 /* Check for match with one of the chars in
|
|
375
|
8858 * "sm_oneof". */
|
|
344
|
8859 while (*pf != NUL && *pf != word[i + k0])
|
|
|
8860 ++pf;
|
|
|
8861 if (*pf == NUL)
|
|
|
8862 continue;
|
|
|
8863 ++k0;
|
|
323
|
8864 }
|
|
344
|
8865
|
|
|
8866 p0 = 5;
|
|
|
8867 s = smp[n0].sm_rules;
|
|
323
|
8868 while (*s == '-')
|
|
|
8869 {
|
|
344
|
8870 /* "k0" gets NOT reduced because
|
|
|
8871 * "if (k0 == k)" */
|
|
323
|
8872 s++;
|
|
|
8873 }
|
|
|
8874 if (*s == '<')
|
|
|
8875 s++;
|
|
344
|
8876 if (VIM_ISDIGIT(*s))
|
|
323
|
8877 {
|
|
|
8878 p0 = *s - '0';
|
|
|
8879 s++;
|
|
|
8880 }
|
|
|
8881
|
|
|
8882 if (*s == NUL
|
|
|
8883 /* *s == '^' cuts */
|
|
|
8884 || (*s == '$'
|
|
376
|
8885 && !spell_iswordp(word + i + k0,
|
|
|
8886 curbuf)))
|
|
323
|
8887 {
|
|
|
8888 if (k0 == k)
|
|
|
8889 /* this is just a piece of the string */
|
|
|
8890 continue;
|
|
|
8891
|
|
|
8892 if (p0 < pri)
|
|
|
8893 /* priority too low */
|
|
|
8894 continue;
|
|
|
8895 /* rule fits; stop search */
|
|
|
8896 break;
|
|
|
8897 }
|
|
|
8898 }
|
|
|
8899
|
|
344
|
8900 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
8901 continue;
|
|
|
8902 }
|
|
|
8903
|
|
|
8904 /* replace string */
|
|
344
|
8905 s = smp[n].sm_to;
|
|
389
|
8906 if (s == NULL)
|
|
|
8907 s = (char_u *)"";
|
|
344
|
8908 pf = smp[n].sm_rules;
|
|
|
8909 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
8910 if (p0 == 1 && z == 0)
|
|
|
8911 {
|
|
|
8912 /* rule with '<' is used */
|
|
344
|
8913 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
8914 || res[reslen - 1] == *s))
|
|
|
8915 reslen--;
|
|
323
|
8916 z0 = 1;
|
|
|
8917 z = 1;
|
|
|
8918 k0 = 0;
|
|
372
|
8919 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
8920 {
|
|
|
8921 word[i + k0] = *s;
|
|
|
8922 k0++;
|
|
|
8923 s++;
|
|
|
8924 }
|
|
|
8925 if (k > k0)
|
|
|
8926 mch_memmove(word + i + k0, word + i + k,
|
|
|
8927 STRLEN(word + i + k) + 1);
|
|
|
8928
|
|
|
8929 /* new "actual letter" */
|
|
|
8930 c = word[i];
|
|
|
8931 }
|
|
|
8932 else
|
|
|
8933 {
|
|
|
8934 /* no '<' rule used */
|
|
|
8935 i += k - 1;
|
|
|
8936 z = 0;
|
|
344
|
8937 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
8938 {
|
|
344
|
8939 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
8940 res[reslen++] = *s;
|
|
323
|
8941 s++;
|
|
|
8942 }
|
|
|
8943 /* new "actual letter" */
|
|
|
8944 c = *s;
|
|
344
|
8945 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
8946 {
|
|
|
8947 if (c != NUL)
|
|
372
|
8948 res[reslen++] = c;
|
|
323
|
8949 mch_memmove(word, word + i + 1,
|
|
|
8950 STRLEN(word + i + 1) + 1);
|
|
|
8951 i = 0;
|
|
|
8952 z0 = 1;
|
|
|
8953 }
|
|
|
8954 }
|
|
|
8955 break;
|
|
|
8956 }
|
|
|
8957 }
|
|
|
8958 }
|
|
324
|
8959 else if (vim_iswhite(c))
|
|
|
8960 {
|
|
|
8961 c = ' ';
|
|
|
8962 k = 1;
|
|
|
8963 }
|
|
323
|
8964
|
|
|
8965 if (z0 == 0)
|
|
|
8966 {
|
|
344
|
8967 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
8968 && (!slang->sl_collapse || reslen == 0
|
|
|
8969 || res[reslen - 1] != c))
|
|
323
|
8970 /* condense only double letters */
|
|
372
|
8971 res[reslen++] = c;
|
|
323
|
8972
|
|
|
8973 i++;
|
|
|
8974 z = 0;
|
|
|
8975 k = 0;
|
|
|
8976 }
|
|
|
8977 }
|
|
|
8978
|
|
344
|
8979 res[reslen] = NUL;
|
|
323
|
8980 }
|
|
|
8981
|
|
372
|
8982 #ifdef FEAT_MBYTE
|
|
|
8983 /*
|
|
|
8984 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
8985 * Multi-byte version of spell_soundfold().
|
|
|
8986 */
|
|
|
8987 static void
|
|
375
|
8988 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
8989 slang_T *slang;
|
|
|
8990 char_u *inword;
|
|
|
8991 char_u *res;
|
|
|
8992 {
|
|
375
|
8993 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
8994 int word[MAXWLEN];
|
|
|
8995 int wres[MAXWLEN];
|
|
|
8996 int l;
|
|
|
8997 char_u *s;
|
|
|
8998 int *ws;
|
|
|
8999 char_u *t;
|
|
|
9000 int *pf;
|
|
|
9001 int i, j, z;
|
|
|
9002 int reslen;
|
|
|
9003 int n, k = 0;
|
|
|
9004 int z0;
|
|
|
9005 int k0;
|
|
|
9006 int n0;
|
|
|
9007 int c;
|
|
|
9008 int pri;
|
|
|
9009 int p0 = -333;
|
|
|
9010 int c0;
|
|
|
9011 int did_white = FALSE;
|
|
|
9012
|
|
|
9013 /*
|
|
|
9014 * Convert the multi-byte string to a wide-character string.
|
|
|
9015 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
9016 * But keep white space.
|
|
|
9017 */
|
|
|
9018 n = 0;
|
|
|
9019 for (s = inword; *s != NUL; )
|
|
|
9020 {
|
|
|
9021 t = s;
|
|
|
9022 c = mb_ptr2char_adv(&s);
|
|
|
9023 if (slang->sl_rem_accents)
|
|
|
9024 {
|
|
|
9025 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9026 {
|
|
|
9027 if (did_white)
|
|
|
9028 continue;
|
|
|
9029 c = ' ';
|
|
|
9030 did_white = TRUE;
|
|
|
9031 }
|
|
|
9032 else
|
|
|
9033 {
|
|
|
9034 did_white = FALSE;
|
|
376
|
9035 if (!spell_iswordp_nmw(t))
|
|
372
|
9036 continue;
|
|
|
9037 }
|
|
|
9038 }
|
|
|
9039 word[n++] = c;
|
|
|
9040 }
|
|
|
9041 word[n] = NUL;
|
|
|
9042
|
|
|
9043 /*
|
|
|
9044 * This comes from Aspell phonet.cpp.
|
|
|
9045 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
9046 * Changed to keep spaces.
|
|
|
9047 */
|
|
|
9048 i = reslen = z = 0;
|
|
|
9049 while ((c = word[i]) != NUL)
|
|
|
9050 {
|
|
|
9051 /* Start with the first rule that has the character in the word. */
|
|
|
9052 n = slang->sl_sal_first[c & 0xff];
|
|
|
9053 z0 = 0;
|
|
|
9054
|
|
|
9055 if (n >= 0)
|
|
|
9056 {
|
|
375
|
9057 /* check all rules for the same index byte */
|
|
372
|
9058 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
9059 {
|
|
|
9060 /* Quickly skip entries that don't match the word. Most
|
|
|
9061 * entries are less then three chars, optimize for that. */
|
|
375
|
9062 if (c != ws[0])
|
|
|
9063 continue;
|
|
372
|
9064 k = smp[n].sm_leadlen;
|
|
|
9065 if (k > 1)
|
|
|
9066 {
|
|
|
9067 if (word[i + 1] != ws[1])
|
|
|
9068 continue;
|
|
|
9069 if (k > 2)
|
|
|
9070 {
|
|
|
9071 for (j = 2; j < k; ++j)
|
|
|
9072 if (word[i + j] != ws[j])
|
|
|
9073 break;
|
|
|
9074 if (j < k)
|
|
|
9075 continue;
|
|
|
9076 }
|
|
|
9077 }
|
|
|
9078
|
|
375
|
9079 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
9080 {
|
|
375
|
9081 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
9082 while (*pf != NUL && *pf != word[i + k])
|
|
|
9083 ++pf;
|
|
|
9084 if (*pf == NUL)
|
|
|
9085 continue;
|
|
|
9086 ++k;
|
|
|
9087 }
|
|
|
9088 s = smp[n].sm_rules;
|
|
|
9089 pri = 5; /* default priority */
|
|
|
9090
|
|
|
9091 p0 = *s;
|
|
|
9092 k0 = k;
|
|
|
9093 while (*s == '-' && k > 1)
|
|
|
9094 {
|
|
|
9095 k--;
|
|
|
9096 s++;
|
|
|
9097 }
|
|
|
9098 if (*s == '<')
|
|
|
9099 s++;
|
|
|
9100 if (VIM_ISDIGIT(*s))
|
|
|
9101 {
|
|
|
9102 /* determine priority */
|
|
|
9103 pri = *s - '0';
|
|
|
9104 s++;
|
|
|
9105 }
|
|
|
9106 if (*s == '^' && *(s + 1) == '^')
|
|
|
9107 s++;
|
|
|
9108
|
|
|
9109 if (*s == NUL
|
|
|
9110 || (*s == '^'
|
|
|
9111 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9112 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
9113 && (*(s + 1) != '$'
|
|
376
|
9114 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9115 || (*s == '$' && i > 0
|
|
376
|
9116 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
9117 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9118 {
|
|
|
9119 /* search for followup rules, if: */
|
|
|
9120 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9121 c0 = word[i + k - 1];
|
|
|
9122 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
9123
|
|
|
9124 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
9125 && p0 != '-' && word[i + k] != NUL)
|
|
|
9126 {
|
|
375
|
9127 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
9128 * all entries with the same index byte. */
|
|
372
|
9129 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
9130 == (c0 & 0xff); ++n0)
|
|
|
9131 {
|
|
|
9132 /* Quickly skip entries that don't match the word.
|
|
375
|
9133 */
|
|
|
9134 if (c0 != ws[0])
|
|
|
9135 continue;
|
|
372
|
9136 k0 = smp[n0].sm_leadlen;
|
|
|
9137 if (k0 > 1)
|
|
|
9138 {
|
|
|
9139 if (word[i + k] != ws[1])
|
|
|
9140 continue;
|
|
|
9141 if (k0 > 2)
|
|
|
9142 {
|
|
|
9143 pf = word + i + k + 1;
|
|
|
9144 for (j = 2; j < k0; ++j)
|
|
|
9145 if (*pf++ != ws[j])
|
|
|
9146 break;
|
|
|
9147 if (j < k0)
|
|
|
9148 continue;
|
|
|
9149 }
|
|
|
9150 }
|
|
|
9151 k0 += k - 1;
|
|
|
9152
|
|
375
|
9153 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
9154 {
|
|
|
9155 /* Check for match with one of the chars in
|
|
375
|
9156 * "sm_oneof". */
|
|
372
|
9157 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9158 ++pf;
|
|
|
9159 if (*pf == NUL)
|
|
|
9160 continue;
|
|
|
9161 ++k0;
|
|
|
9162 }
|
|
|
9163
|
|
|
9164 p0 = 5;
|
|
|
9165 s = smp[n0].sm_rules;
|
|
|
9166 while (*s == '-')
|
|
|
9167 {
|
|
|
9168 /* "k0" gets NOT reduced because
|
|
|
9169 * "if (k0 == k)" */
|
|
|
9170 s++;
|
|
|
9171 }
|
|
|
9172 if (*s == '<')
|
|
|
9173 s++;
|
|
|
9174 if (VIM_ISDIGIT(*s))
|
|
|
9175 {
|
|
|
9176 p0 = *s - '0';
|
|
|
9177 s++;
|
|
|
9178 }
|
|
|
9179
|
|
|
9180 if (*s == NUL
|
|
|
9181 /* *s == '^' cuts */
|
|
|
9182 || (*s == '$'
|
|
376
|
9183 && !spell_iswordp_w(word + i + k0,
|
|
|
9184 curbuf)))
|
|
372
|
9185 {
|
|
|
9186 if (k0 == k)
|
|
|
9187 /* this is just a piece of the string */
|
|
|
9188 continue;
|
|
|
9189
|
|
|
9190 if (p0 < pri)
|
|
|
9191 /* priority too low */
|
|
|
9192 continue;
|
|
|
9193 /* rule fits; stop search */
|
|
|
9194 break;
|
|
|
9195 }
|
|
|
9196 }
|
|
|
9197
|
|
|
9198 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
9199 == (c0 & 0xff))
|
|
|
9200 continue;
|
|
|
9201 }
|
|
|
9202
|
|
|
9203 /* replace string */
|
|
|
9204 ws = smp[n].sm_to_w;
|
|
|
9205 s = smp[n].sm_rules;
|
|
|
9206 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
9207 if (p0 == 1 && z == 0)
|
|
|
9208 {
|
|
|
9209 /* rule with '<' is used */
|
|
389
|
9210 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
9211 && (wres[reslen - 1] == c
|
|
372
|
9212 || wres[reslen - 1] == *ws))
|
|
|
9213 reslen--;
|
|
|
9214 z0 = 1;
|
|
|
9215 z = 1;
|
|
|
9216 k0 = 0;
|
|
389
|
9217 if (ws != NULL)
|
|
|
9218 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
9219 {
|
|
|
9220 word[i + k0] = *ws;
|
|
|
9221 k0++;
|
|
|
9222 ws++;
|
|
|
9223 }
|
|
372
|
9224 if (k > k0)
|
|
|
9225 mch_memmove(word + i + k0, word + i + k,
|
|
|
9226 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
9227
|
|
|
9228 /* new "actual letter" */
|
|
|
9229 c = word[i];
|
|
|
9230 }
|
|
|
9231 else
|
|
|
9232 {
|
|
|
9233 /* no '<' rule used */
|
|
|
9234 i += k - 1;
|
|
|
9235 z = 0;
|
|
389
|
9236 if (ws != NULL)
|
|
|
9237 while (*ws != NUL && ws[1] != NUL
|
|
|
9238 && reslen < MAXWLEN)
|
|
|
9239 {
|
|
|
9240 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
9241 wres[reslen++] = *ws;
|
|
|
9242 ws++;
|
|
|
9243 }
|
|
372
|
9244 /* new "actual letter" */
|
|
389
|
9245 if (ws == NULL)
|
|
|
9246 c = NUL;
|
|
|
9247 else
|
|
|
9248 c = *ws;
|
|
372
|
9249 if (strstr((char *)s, "^^") != NULL)
|
|
|
9250 {
|
|
|
9251 if (c != NUL)
|
|
|
9252 wres[reslen++] = c;
|
|
|
9253 mch_memmove(word, word + i + 1,
|
|
|
9254 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
9255 i = 0;
|
|
|
9256 z0 = 1;
|
|
|
9257 }
|
|
|
9258 }
|
|
|
9259 break;
|
|
|
9260 }
|
|
|
9261 }
|
|
|
9262 }
|
|
|
9263 else if (vim_iswhite(c))
|
|
|
9264 {
|
|
|
9265 c = ' ';
|
|
|
9266 k = 1;
|
|
|
9267 }
|
|
|
9268
|
|
|
9269 if (z0 == 0)
|
|
|
9270 {
|
|
|
9271 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9272 && (!slang->sl_collapse || reslen == 0
|
|
|
9273 || wres[reslen - 1] != c))
|
|
|
9274 /* condense only double letters */
|
|
|
9275 wres[reslen++] = c;
|
|
|
9276
|
|
|
9277 i++;
|
|
|
9278 z = 0;
|
|
|
9279 k = 0;
|
|
|
9280 }
|
|
|
9281 }
|
|
|
9282
|
|
|
9283 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
9284 l = 0;
|
|
|
9285 for (n = 0; n < reslen; ++n)
|
|
|
9286 {
|
|
|
9287 l += mb_char2bytes(wres[n], res + l);
|
|
|
9288 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
9289 break;
|
|
|
9290 }
|
|
|
9291 res[l] = NUL;
|
|
|
9292 }
|
|
|
9293 #endif
|
|
|
9294
|
|
324
|
9295 /*
|
|
344
|
9296 * Compute a score for two sound-a-like words.
|
|
|
9297 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
9298 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
9299 * avoiding checks that will not be possible.
|
|
|
9300 */
|
|
|
9301 static int
|
|
351
|
9302 soundalike_score(goodstart, badstart)
|
|
|
9303 char_u *goodstart; /* sound-folded good word */
|
|
|
9304 char_u *badstart; /* sound-folded bad word */
|
|
344
|
9305 {
|
|
351
|
9306 char_u *goodsound = goodstart;
|
|
|
9307 char_u *badsound = badstart;
|
|
|
9308 int goodlen;
|
|
|
9309 int badlen;
|
|
344
|
9310 int n;
|
|
|
9311 char_u *pl, *ps;
|
|
|
9312 char_u *pl2, *ps2;
|
|
351
|
9313 int score = 0;
|
|
|
9314
|
|
|
9315 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
9316 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
9317 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
9318 {
|
|
|
9319 score = SCORE_DEL / 2;
|
|
|
9320 if (*badsound == '*')
|
|
|
9321 ++badsound;
|
|
|
9322 else
|
|
|
9323 ++goodsound;
|
|
|
9324 }
|
|
|
9325
|
|
|
9326 goodlen = STRLEN(goodsound);
|
|
|
9327 badlen = STRLEN(badsound);
|
|
344
|
9328
|
|
|
9329 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
9330 * changes. */
|
|
|
9331 n = goodlen - badlen;
|
|
|
9332 if (n < -2 || n > 2)
|
|
|
9333 return SCORE_MAXMAX;
|
|
|
9334
|
|
|
9335 if (n > 0)
|
|
|
9336 {
|
|
351
|
9337 pl = goodsound; /* goodsound is longest */
|
|
344
|
9338 ps = badsound;
|
|
|
9339 }
|
|
|
9340 else
|
|
|
9341 {
|
|
351
|
9342 pl = badsound; /* badsound is longest */
|
|
344
|
9343 ps = goodsound;
|
|
|
9344 }
|
|
|
9345
|
|
|
9346 /* Skip over the identical part. */
|
|
|
9347 while (*pl == *ps && *pl != NUL)
|
|
|
9348 {
|
|
|
9349 ++pl;
|
|
|
9350 ++ps;
|
|
|
9351 }
|
|
|
9352
|
|
|
9353 switch (n)
|
|
|
9354 {
|
|
|
9355 case -2:
|
|
|
9356 case 2:
|
|
|
9357 /*
|
|
|
9358 * Must delete two characters from "pl".
|
|
|
9359 */
|
|
|
9360 ++pl; /* first delete */
|
|
|
9361 while (*pl == *ps)
|
|
|
9362 {
|
|
|
9363 ++pl;
|
|
|
9364 ++ps;
|
|
|
9365 }
|
|
|
9366 /* strings must be equal after second delete */
|
|
|
9367 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
9368 return score + SCORE_DEL * 2;
|
|
344
|
9369
|
|
|
9370 /* Failed to compare. */
|
|
|
9371 break;
|
|
|
9372
|
|
|
9373 case -1:
|
|
|
9374 case 1:
|
|
|
9375 /*
|
|
|
9376 * Minimal one delete from "pl" required.
|
|
|
9377 */
|
|
|
9378
|
|
|
9379 /* 1: delete */
|
|
|
9380 pl2 = pl + 1;
|
|
|
9381 ps2 = ps;
|
|
|
9382 while (*pl2 == *ps2)
|
|
|
9383 {
|
|
|
9384 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9385 return score + SCORE_DEL;
|
|
344
|
9386 ++pl2;
|
|
|
9387 ++ps2;
|
|
|
9388 }
|
|
|
9389
|
|
|
9390 /* 2: delete then swap, then rest must be equal */
|
|
|
9391 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9392 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9393 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
9394
|
|
|
9395 /* 3: delete then substitute, then the rest must be equal */
|
|
|
9396 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9397 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
9398
|
|
|
9399 /* 4: first swap then delete */
|
|
|
9400 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9401 {
|
|
|
9402 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9403 ps2 = ps + 2;
|
|
|
9404 while (*pl2 == *ps2)
|
|
|
9405 {
|
|
|
9406 ++pl2;
|
|
|
9407 ++ps2;
|
|
|
9408 }
|
|
|
9409 /* delete a char and then strings must be equal */
|
|
|
9410 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9411 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
9412 }
|
|
|
9413
|
|
|
9414 /* 5: first substitute then delete */
|
|
|
9415 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
9416 ps2 = ps + 1;
|
|
|
9417 while (*pl2 == *ps2)
|
|
|
9418 {
|
|
|
9419 ++pl2;
|
|
|
9420 ++ps2;
|
|
|
9421 }
|
|
|
9422 /* delete a char and then strings must be equal */
|
|
|
9423 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9424 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
9425
|
|
|
9426 /* Failed to compare. */
|
|
|
9427 break;
|
|
|
9428
|
|
|
9429 case 0:
|
|
|
9430 /*
|
|
|
9431 * Lenghts are equal, thus changes must result in same length: An
|
|
|
9432 * insert is only possible in combination with a delete.
|
|
|
9433 * 1: check if for identical strings
|
|
|
9434 */
|
|
|
9435 if (*pl == NUL)
|
|
351
|
9436 return score;
|
|
344
|
9437
|
|
|
9438 /* 2: swap */
|
|
|
9439 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9440 {
|
|
|
9441 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9442 ps2 = ps + 2;
|
|
|
9443 while (*pl2 == *ps2)
|
|
|
9444 {
|
|
|
9445 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9446 return score + SCORE_SWAP;
|
|
344
|
9447 ++pl2;
|
|
|
9448 ++ps2;
|
|
|
9449 }
|
|
|
9450 /* 3: swap and swap again */
|
|
|
9451 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9452 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9453 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
9454
|
|
|
9455 /* 4: swap and substitute */
|
|
|
9456 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9457 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
9458 }
|
|
|
9459
|
|
|
9460 /* 5: substitute */
|
|
|
9461 pl2 = pl + 1;
|
|
|
9462 ps2 = ps + 1;
|
|
|
9463 while (*pl2 == *ps2)
|
|
|
9464 {
|
|
|
9465 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9466 return score + SCORE_SUBST;
|
|
344
|
9467 ++pl2;
|
|
|
9468 ++ps2;
|
|
|
9469 }
|
|
|
9470
|
|
|
9471 /* 6: substitute and swap */
|
|
|
9472 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9473 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9474 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
9475
|
|
|
9476 /* 7: substitute and substitute */
|
|
|
9477 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9478 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
9479
|
|
|
9480 /* 8: insert then delete */
|
|
|
9481 pl2 = pl;
|
|
|
9482 ps2 = ps + 1;
|
|
|
9483 while (*pl2 == *ps2)
|
|
|
9484 {
|
|
|
9485 ++pl2;
|
|
|
9486 ++ps2;
|
|
|
9487 }
|
|
|
9488 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9489 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9490
|
|
|
9491 /* 9: delete then insert */
|
|
|
9492 pl2 = pl + 1;
|
|
|
9493 ps2 = ps;
|
|
|
9494 while (*pl2 == *ps2)
|
|
|
9495 {
|
|
|
9496 ++pl2;
|
|
|
9497 ++ps2;
|
|
|
9498 }
|
|
|
9499 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
9500 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9501
|
|
|
9502 /* Failed to compare. */
|
|
|
9503 break;
|
|
|
9504 }
|
|
|
9505
|
|
|
9506 return SCORE_MAXMAX;
|
|
|
9507 }
|
|
324
|
9508
|
|
323
|
9509 /*
|
|
|
9510 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
9511 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
9512 *
|
|
323
|
9513 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
9514 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
9515 * characters.
|
|
323
|
9516 */
|
|
|
9517 static int
|
|
|
9518 spell_edit_score(badword, goodword)
|
|
|
9519 char_u *badword;
|
|
|
9520 char_u *goodword;
|
|
|
9521 {
|
|
|
9522 int *cnt;
|
|
372
|
9523 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
9524 int j, i;
|
|
|
9525 int t;
|
|
|
9526 int bc, gc;
|
|
324
|
9527 int pbc, pgc;
|
|
|
9528 #ifdef FEAT_MBYTE
|
|
|
9529 char_u *p;
|
|
|
9530 int wbadword[MAXWLEN];
|
|
|
9531 int wgoodword[MAXWLEN];
|
|
|
9532
|
|
|
9533 if (has_mbyte)
|
|
|
9534 {
|
|
|
9535 /* Get the characters from the multi-byte strings and put them in an
|
|
|
9536 * int array for easy access. */
|
|
|
9537 for (p = badword, badlen = 0; *p != NUL; )
|
|
|
9538 wbadword[badlen++] = mb_ptr2char_adv(&p);
|
|
|
9539 ++badlen;
|
|
|
9540 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
|
9541 wgoodword[goodlen++] = mb_ptr2char_adv(&p);
|
|
|
9542 ++goodlen;
|
|
|
9543 }
|
|
|
9544 else
|
|
|
9545 #endif
|
|
|
9546 {
|
|
|
9547 badlen = STRLEN(badword) + 1;
|
|
|
9548 goodlen = STRLEN(goodword) + 1;
|
|
|
9549 }
|
|
323
|
9550
|
|
|
9551 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
9552 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
9553 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
9554 TRUE);
|
|
324
|
9555 if (cnt == NULL)
|
|
|
9556 return 0; /* out of memory */
|
|
323
|
9557
|
|
|
9558 CNT(0, 0) = 0;
|
|
|
9559 for (j = 1; j <= goodlen; ++j)
|
|
|
9560 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
9561
|
|
|
9562 for (i = 1; i <= badlen; ++i)
|
|
|
9563 {
|
|
|
9564 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
9565 for (j = 1; j <= goodlen; ++j)
|
|
|
9566 {
|
|
324
|
9567 #ifdef FEAT_MBYTE
|
|
|
9568 if (has_mbyte)
|
|
|
9569 {
|
|
|
9570 bc = wbadword[i - 1];
|
|
|
9571 gc = wgoodword[j - 1];
|
|
|
9572 }
|
|
|
9573 else
|
|
|
9574 #endif
|
|
|
9575 {
|
|
|
9576 bc = badword[i - 1];
|
|
|
9577 gc = goodword[j - 1];
|
|
|
9578 }
|
|
323
|
9579 if (bc == gc)
|
|
|
9580 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
9581 else
|
|
|
9582 {
|
|
|
9583 /* Use a better score when there is only a case difference. */
|
|
324
|
9584 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
9585 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
9586 else
|
|
|
9587 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
9588
|
|
324
|
9589 if (i > 1 && j > 1)
|
|
323
|
9590 {
|
|
324
|
9591 #ifdef FEAT_MBYTE
|
|
|
9592 if (has_mbyte)
|
|
|
9593 {
|
|
|
9594 pbc = wbadword[i - 2];
|
|
|
9595 pgc = wgoodword[j - 2];
|
|
|
9596 }
|
|
|
9597 else
|
|
|
9598 #endif
|
|
|
9599 {
|
|
|
9600 pbc = badword[i - 2];
|
|
|
9601 pgc = goodword[j - 2];
|
|
|
9602 }
|
|
|
9603 if (bc == pgc && pbc == gc)
|
|
|
9604 {
|
|
|
9605 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
9606 if (t < CNT(i, j))
|
|
|
9607 CNT(i, j) = t;
|
|
|
9608 }
|
|
323
|
9609 }
|
|
|
9610 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
9611 if (t < CNT(i, j))
|
|
|
9612 CNT(i, j) = t;
|
|
|
9613 t = SCORE_INS + CNT(i, j - 1);
|
|
|
9614 if (t < CNT(i, j))
|
|
|
9615 CNT(i, j) = t;
|
|
|
9616 }
|
|
|
9617 }
|
|
|
9618 }
|
|
344
|
9619
|
|
|
9620 i = CNT(badlen - 1, goodlen - 1);
|
|
|
9621 vim_free(cnt);
|
|
|
9622 return i;
|
|
323
|
9623 }
|
|
307
|
9624
|
|
351
|
9625 /*
|
|
|
9626 * ":spelldump"
|
|
|
9627 */
|
|
|
9628 /*ARGSUSED*/
|
|
|
9629 void
|
|
|
9630 ex_spelldump(eap)
|
|
|
9631 exarg_T *eap;
|
|
|
9632 {
|
|
|
9633 buf_T *buf = curbuf;
|
|
|
9634 langp_T *lp;
|
|
|
9635 slang_T *slang;
|
|
|
9636 idx_T arridx[MAXWLEN];
|
|
|
9637 int curi[MAXWLEN];
|
|
|
9638 char_u word[MAXWLEN];
|
|
|
9639 int c;
|
|
|
9640 char_u *byts;
|
|
|
9641 idx_T *idxs;
|
|
|
9642 linenr_T lnum = 0;
|
|
|
9643 int round;
|
|
|
9644 int depth;
|
|
|
9645 int n;
|
|
|
9646 int flags;
|
|
381
|
9647 char_u *region_names = NULL; /* region names being used */
|
|
|
9648 int do_region = TRUE; /* dump region names and numbers */
|
|
|
9649 char_u *p;
|
|
351
|
9650
|
|
|
9651 if (no_spell_checking())
|
|
|
9652 return;
|
|
|
9653
|
|
|
9654 /* Create a new empty buffer by splitting the window. */
|
|
|
9655 do_cmdline_cmd((char_u *)"new");
|
|
|
9656 if (!bufempty() || !buf_valid(buf))
|
|
|
9657 return;
|
|
|
9658
|
|
381
|
9659 /* Find out if we can support regions: All languages must support the same
|
|
|
9660 * regions or none at all. */
|
|
|
9661 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
9662 {
|
|
|
9663 p = lp->lp_slang->sl_regions;
|
|
|
9664 if (p[0] != 0)
|
|
|
9665 {
|
|
|
9666 if (region_names == NULL) /* first language with regions */
|
|
|
9667 region_names = p;
|
|
|
9668 else if (STRCMP(region_names, p) != 0)
|
|
|
9669 {
|
|
|
9670 do_region = FALSE; /* region names are different */
|
|
|
9671 break;
|
|
|
9672 }
|
|
|
9673 }
|
|
|
9674 }
|
|
|
9675
|
|
|
9676 if (do_region && region_names != NULL)
|
|
|
9677 {
|
|
|
9678 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
9679 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
9680 }
|
|
|
9681 else
|
|
|
9682 do_region = FALSE;
|
|
|
9683
|
|
|
9684 /*
|
|
|
9685 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
9686 */
|
|
351
|
9687 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
9688 {
|
|
|
9689 slang = lp->lp_slang;
|
|
|
9690
|
|
|
9691 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
9692 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
9693
|
|
|
9694 /* round 1: case-folded tree
|
|
|
9695 * round 2: keep-case tree */
|
|
|
9696 for (round = 1; round <= 2; ++round)
|
|
|
9697 {
|
|
|
9698 if (round == 1)
|
|
|
9699 {
|
|
|
9700 byts = slang->sl_fbyts;
|
|
|
9701 idxs = slang->sl_fidxs;
|
|
|
9702 }
|
|
|
9703 else
|
|
|
9704 {
|
|
|
9705 byts = slang->sl_kbyts;
|
|
|
9706 idxs = slang->sl_kidxs;
|
|
|
9707 }
|
|
|
9708 if (byts == NULL)
|
|
|
9709 continue; /* array is empty */
|
|
|
9710
|
|
|
9711 depth = 0;
|
|
|
9712 arridx[0] = 0;
|
|
|
9713 curi[0] = 1;
|
|
|
9714 while (depth >= 0 && !got_int)
|
|
|
9715 {
|
|
|
9716 if (curi[depth] > byts[arridx[depth]])
|
|
|
9717 {
|
|
|
9718 /* Done all bytes at this node, go up one level. */
|
|
|
9719 --depth;
|
|
|
9720 line_breakcheck();
|
|
|
9721 }
|
|
|
9722 else
|
|
|
9723 {
|
|
|
9724 /* Do one more byte at this node. */
|
|
|
9725 n = arridx[depth] + curi[depth];
|
|
|
9726 ++curi[depth];
|
|
|
9727 c = byts[n];
|
|
|
9728 if (c == 0)
|
|
|
9729 {
|
|
|
9730 /* End of word, deal with the word.
|
|
|
9731 * Don't use keep-case words in the fold-case tree,
|
|
|
9732 * they will appear in the keep-case tree.
|
|
|
9733 * Only use the word when the region matches. */
|
|
|
9734 flags = (int)idxs[n];
|
|
|
9735 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
9736 && (do_region
|
|
|
9737 || (flags & WF_REGION) == 0
|
|
|
9738 || (((unsigned)flags >> 8)
|
|
351
|
9739 & lp->lp_region) != 0))
|
|
|
9740 {
|
|
|
9741 word[depth] = NUL;
|
|
381
|
9742 if (!do_region)
|
|
|
9743 flags &= ~WF_REGION;
|
|
355
|
9744
|
|
|
9745 /* Dump the basic word if there is no prefix or
|
|
|
9746 * when it's the first one. */
|
|
|
9747 c = (unsigned)flags >> 16;
|
|
|
9748 if (c == 0 || curi[depth] == 2)
|
|
|
9749 dump_word(word, round, flags, lnum++);
|
|
351
|
9750
|
|
|
9751 /* Apply the prefix, if there is one. */
|
|
355
|
9752 if (c != 0)
|
|
351
|
9753 lnum = apply_prefixes(slang, word, round,
|
|
|
9754 flags, lnum);
|
|
|
9755 }
|
|
|
9756 }
|
|
|
9757 else
|
|
|
9758 {
|
|
|
9759 /* Normal char, go one level deeper. */
|
|
|
9760 word[depth++] = c;
|
|
|
9761 arridx[depth] = idxs[n];
|
|
|
9762 curi[depth] = 1;
|
|
|
9763 }
|
|
|
9764 }
|
|
|
9765 }
|
|
|
9766 }
|
|
|
9767 }
|
|
|
9768
|
|
|
9769 /* Delete the empty line that we started with. */
|
|
|
9770 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
9771 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
9772
|
|
|
9773 redraw_later(NOT_VALID);
|
|
|
9774 }
|
|
|
9775
|
|
|
9776 /*
|
|
|
9777 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
9778 */
|
|
|
9779 static void
|
|
|
9780 dump_word(word, round, flags, lnum)
|
|
|
9781 char_u *word;
|
|
|
9782 int round;
|
|
|
9783 int flags;
|
|
|
9784 linenr_T lnum;
|
|
|
9785 {
|
|
|
9786 int keepcap = FALSE;
|
|
|
9787 char_u *p;
|
|
|
9788 char_u cword[MAXWLEN];
|
|
381
|
9789 char_u badword[MAXWLEN + 10];
|
|
|
9790 int i;
|
|
351
|
9791
|
|
|
9792 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
9793 {
|
|
|
9794 /* Need to fix case according to "flags". */
|
|
|
9795 make_case_word(word, cword, flags);
|
|
|
9796 p = cword;
|
|
|
9797 }
|
|
|
9798 else
|
|
|
9799 {
|
|
|
9800 p = word;
|
|
389
|
9801 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
9802 || (flags & WF_FIXCAP) != 0))
|
|
351
|
9803 keepcap = TRUE;
|
|
|
9804 }
|
|
|
9805
|
|
381
|
9806 /* Add flags and regions after a slash. */
|
|
|
9807 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
9808 {
|
|
|
9809 STRCPY(badword, p);
|
|
|
9810 STRCAT(badword, "/");
|
|
351
|
9811 if (keepcap)
|
|
|
9812 STRCAT(badword, "=");
|
|
|
9813 if (flags & WF_BANNED)
|
|
|
9814 STRCAT(badword, "!");
|
|
|
9815 else if (flags & WF_RARE)
|
|
|
9816 STRCAT(badword, "?");
|
|
381
|
9817 if (flags & WF_REGION)
|
|
|
9818 for (i = 0; i < 7; ++i)
|
|
|
9819 if (flags & (0x100 << i))
|
|
|
9820 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
9821 p = badword;
|
|
|
9822 }
|
|
|
9823
|
|
|
9824 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
9825 }
|
|
|
9826
|
|
|
9827 /*
|
|
372
|
9828 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
9829 * "word" and append a line to the buffer.
|
|
351
|
9830 * Return the updated line number.
|
|
|
9831 */
|
|
|
9832 static linenr_T
|
|
|
9833 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
9834 slang_T *slang;
|
|
|
9835 char_u *word; /* case-folded word */
|
|
|
9836 int round;
|
|
|
9837 int flags; /* flags with prefix ID */
|
|
|
9838 linenr_T startlnum;
|
|
|
9839 {
|
|
|
9840 idx_T arridx[MAXWLEN];
|
|
|
9841 int curi[MAXWLEN];
|
|
|
9842 char_u prefix[MAXWLEN];
|
|
|
9843 int c;
|
|
|
9844 char_u *byts;
|
|
|
9845 idx_T *idxs;
|
|
|
9846 linenr_T lnum = startlnum;
|
|
|
9847 int depth;
|
|
|
9848 int n;
|
|
|
9849 int len;
|
|
|
9850 int prefid = (unsigned)flags >> 16;
|
|
|
9851 int i;
|
|
|
9852
|
|
|
9853 byts = slang->sl_pbyts;
|
|
|
9854 idxs = slang->sl_pidxs;
|
|
|
9855 if (byts != NULL) /* array not is empty */
|
|
|
9856 {
|
|
|
9857 /*
|
|
|
9858 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
|
9859 * When at the end of a prefix check that it supports "prefid".
|
|
|
9860 */
|
|
|
9861 depth = 0;
|
|
|
9862 arridx[0] = 0;
|
|
|
9863 curi[0] = 1;
|
|
|
9864 while (depth >= 0 && !got_int)
|
|
|
9865 {
|
|
|
9866 len = arridx[depth];
|
|
|
9867 if (curi[depth] > byts[len])
|
|
|
9868 {
|
|
|
9869 /* Done all bytes at this node, go up one level. */
|
|
|
9870 --depth;
|
|
|
9871 line_breakcheck();
|
|
|
9872 }
|
|
|
9873 else
|
|
|
9874 {
|
|
|
9875 /* Do one more byte at this node. */
|
|
|
9876 n = len + curi[depth];
|
|
|
9877 ++curi[depth];
|
|
|
9878 c = byts[n];
|
|
|
9879 if (c == 0)
|
|
|
9880 {
|
|
|
9881 /* End of prefix, find out how many IDs there are. */
|
|
|
9882 for (i = 1; i < len; ++i)
|
|
|
9883 if (byts[n + i] != 0)
|
|
|
9884 break;
|
|
|
9885 curi[depth] += i - 1;
|
|
|
9886
|
|
366
|
9887 i = valid_word_prefix(i, n, prefid, word, slang);
|
|
|
9888 if (i != 0)
|
|
351
|
9889 {
|
|
376
|
9890 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
9891 dump_word(prefix, round,
|
|
|
9892 (i & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
9893 : flags, lnum++);
|
|
351
|
9894 }
|
|
|
9895 }
|
|
|
9896 else
|
|
|
9897 {
|
|
|
9898 /* Normal char, go one level deeper. */
|
|
|
9899 prefix[depth++] = c;
|
|
|
9900 arridx[depth] = idxs[n];
|
|
|
9901 curi[depth] = 1;
|
|
|
9902 }
|
|
|
9903 }
|
|
|
9904 }
|
|
|
9905 }
|
|
|
9906
|
|
|
9907 return lnum;
|
|
|
9908 }
|
|
|
9909
|
|
236
|
9910 #endif /* FEAT_SYN_HL */
|
