|
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
|
|
469
|
53 /* Use SPELL_PRINTTREE for debugging: dump the word tree after adding a word.
|
|
474
|
54 * Only use it for small word lists!
|
|
|
55 * SPELL_COMPRESS_CNT is in how many words we compress the tree to limit the
|
|
|
56 * amount of memory used (esp. for Italian). */
|
|
469
|
57 #if 0
|
|
|
58 # define SPELL_PRINTTREE
|
|
|
59 # define SPELL_COMPRESS_CNT 1
|
|
|
60 #else
|
|
|
61 # define SPELL_COMPRESS_CNT 1000000
|
|
|
62 #endif
|
|
|
63
|
|
300
|
64 /*
|
|
324
|
65 * Use this to adjust the score after finding suggestions, based on the
|
|
|
66 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
67 * it for every possible suggestion.
|
|
|
68 * Disadvantage: When "the" is typed as "hte" it sounds different and goes
|
|
|
69 * down in the list.
|
|
344
|
70 * Used when 'spellsuggest' is set to "best".
|
|
|
71 */
|
|
|
72 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
73
|
|
|
74 /*
|
|
|
75 * The double scoring mechanism is based on the principle that there are two
|
|
|
76 * kinds of spelling mistakes:
|
|
|
77 * 1. You know how to spell the word, but mistype something. This results in
|
|
|
78 * a small editing distance (character swapped/omitted/inserted) and
|
|
|
79 * possibly a word that sounds completely different.
|
|
|
80 * 2. You don't know how to spell the word and type something that sounds
|
|
|
81 * right. The edit distance can be big but the word is similar after
|
|
|
82 * sound-folding.
|
|
|
83 * Since scores for these two mistakes will be very different we use a list
|
|
|
84 * for each.
|
|
|
85 * The sound-folding is slow, only do double scoring when 'spellsuggest' is
|
|
|
86 * "double".
|
|
323
|
87 */
|
|
|
88
|
|
|
89 /*
|
|
339
|
90 * Vim spell file format: <HEADER>
|
|
|
91 * <SUGGEST>
|
|
|
92 * <LWORDTREE>
|
|
|
93 * <KWORDTREE>
|
|
|
94 * <PREFIXTREE>
|
|
300
|
95 *
|
|
339
|
96 * <HEADER>: <fileID>
|
|
|
97 * <regioncnt> <regionname> ...
|
|
|
98 * <charflagslen> <charflags>
|
|
|
99 * <fcharslen> <fchars>
|
|
366
|
100 * <midwordlen> <midword>
|
|
339
|
101 * <prefcondcnt> <prefcond> ...
|
|
300
|
102 *
|
|
455
|
103 * <fileID> 10 bytes "VIMspell09"
|
|
300
|
104 * <regioncnt> 1 byte number of regions following (8 supported)
|
|
307
|
105 * <regionname> 2 bytes Region name: ca, au, etc. Lower case.
|
|
300
|
106 * First <regionname> is region 1.
|
|
|
107 *
|
|
|
108 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
109 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
110 * 0x01 word character CF_WORD
|
|
|
111 * 0x02 upper-case character CF_UPPER
|
|
300
|
112 * <fcharslen> 2 bytes Number of bytes in <fchars>.
|
|
|
113 * <fchars> N bytes Folded characters, first one is for character 128.
|
|
|
114 *
|
|
366
|
115 * <midwordlen> 2 bytes Number of bytes in <midword>.
|
|
|
116 * <midword> N bytes Characters that are word characters only when used
|
|
|
117 * in the middle of a word.
|
|
|
118 *
|
|
339
|
119 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
120 *
|
|
|
121 * <prefcond> : <condlen> <condstr>
|
|
|
122 *
|
|
|
123 * <condlen> 1 byte Length of <condstr>.
|
|
|
124 *
|
|
|
125 * <condstr> N bytes Condition for the prefix.
|
|
|
126 *
|
|
300
|
127 *
|
|
323
|
128 * <SUGGEST> : <repcount> <rep> ...
|
|
|
129 * <salflags> <salcount> <sal> ...
|
|
|
130 * <maplen> <mapstr>
|
|
|
131 *
|
|
|
132 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
|
133 *
|
|
|
134 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
|
135 *
|
|
|
136 * <repfromlen> 1 byte length of <repfrom>
|
|
|
137 *
|
|
|
138 * <repfrom> N bytes "from" part of replacement
|
|
|
139 *
|
|
|
140 * <reptolen> 1 byte length of <repto>
|
|
|
141 *
|
|
|
142 * <repto> N bytes "to" part of replacement
|
|
300
|
143 *
|
|
323
|
144 * <salflags> 1 byte flags for soundsalike conversion:
|
|
|
145 * SAL_F0LLOWUP
|
|
|
146 * SAL_COLLAPSE
|
|
|
147 * SAL_REM_ACCENTS
|
|
375
|
148 * SAL_SOFO: SOFOFROM and SOFOTO used instead of SAL
|
|
|
149 *
|
|
|
150 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
151 *
|
|
|
152 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
|
153 *
|
|
375
|
154 * <salfromlen> 1-2 bytes length of <salfrom> (2 bytes for SAL_SOFO)
|
|
323
|
155 *
|
|
|
156 * <salfrom> N bytes "from" part of soundsalike
|
|
|
157 *
|
|
375
|
158 * <saltolen> 1-2 bytes length of <salto> (2 bytes for SAL_SOFO)
|
|
323
|
159 *
|
|
|
160 * <salto> N bytes "to" part of soundsalike
|
|
|
161 *
|
|
|
162 * <maplen> 2 bytes length of <mapstr>, MSB first
|
|
|
163 *
|
|
|
164 * <mapstr> N bytes String with sequences of similar characters,
|
|
|
165 * separated by slashes.
|
|
300
|
166 *
|
|
|
167 *
|
|
|
168 * <LWORDTREE>: <wordtree>
|
|
|
169 *
|
|
339
|
170 * <KWORDTREE>: <wordtree>
|
|
|
171 *
|
|
|
172 * <PREFIXTREE>: <wordtree>
|
|
|
173 *
|
|
|
174 *
|
|
300
|
175 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
176 *
|
|
|
177 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
178 *
|
|
|
179 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
180 *
|
|
|
181 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
182 * follow in sorted order.
|
|
|
183 *
|
|
339
|
184 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
455
|
185 * | <flags> [<flags2>] [<region>] [<prefixID>]
|
|
|
186 * | [<pflags>] <prefixID> <prefcondnr> ]
|
|
300
|
187 *
|
|
|
188 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
189 * BY_NOFLAGS: End of word without flags and for all
|
|
|
190 * regions.
|
|
366
|
191 * For PREFIXTREE <prefixID> and
|
|
|
192 * <prefcondnr> follow.
|
|
390
|
193 * BY_FLAGS: End of word, <flags> follow.
|
|
455
|
194 * For PREFIXTREE <pflags>, <prefixID>
|
|
|
195 * and <prefcondnr> follow.
|
|
390
|
196 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
197 * follow. Not used in PREFIXTREE.
|
|
|
198 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
199 * and <xbyte> follow.
|
|
|
200 *
|
|
|
201 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
202 *
|
|
|
203 * <xbyte> 1 byte byte value of the sibling.
|
|
|
204 *
|
|
|
205 * <flags> 1 byte bitmask of:
|
|
|
206 * WF_ALLCAP word must have only capitals
|
|
|
207 * WF_ONECAP first char of word must be capital
|
|
389
|
208 * WF_KEEPCAP keep-case word
|
|
|
209 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
210 * WF_RARE rare word
|
|
389
|
211 * WF_BANNED bad word
|
|
300
|
212 * WF_REGION <region> follows
|
|
339
|
213 * WF_PFX <prefixID> follows
|
|
300
|
214 *
|
|
390
|
215 * <flags2> 1 byte Only used when there are postponed prefixes.
|
|
|
216 * Bitmask of:
|
|
|
217 * WF_HAS_AFF >> 8 word includes affix
|
|
|
218 *
|
|
455
|
219 * <pflags> 1 byte bitmask of:
|
|
|
220 * WFP_RARE rare prefix
|
|
|
221 * WFP_NC non-combining prefix
|
|
|
222 * WFP_UP letter after prefix made upper case
|
|
|
223 *
|
|
300
|
224 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
225 * omitted it's valid in all regions.
|
|
|
226 * Lowest bit is for region 1.
|
|
|
227 *
|
|
339
|
228 * <prefixID> 1 byte ID of prefix that can be used with this word. For
|
|
|
229 * PREFIXTREE used for the required prefix ID.
|
|
|
230 *
|
|
|
231 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
232 * from HEADER.
|
|
300
|
233 *
|
|
|
234 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
235 */
|
|
|
236
|
|
223
|
237 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
238 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
239 #endif
|
|
|
240
|
|
|
241 #include "vim.h"
|
|
|
242
|
|
|
243 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
244
|
|
|
245 #ifdef HAVE_FCNTL_H
|
|
|
246 # include <fcntl.h>
|
|
|
247 #endif
|
|
|
248
|
|
323
|
249 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
250 Some places assume a word length fits in a
|
|
|
251 byte, thus it can't be above 255. */
|
|
226
|
252
|
|
324
|
253 /* Type used for indexes in the word tree need to be at least 3 bytes. If int
|
|
|
254 * is 8 bytes we could use something smaller, but what? */
|
|
|
255 #if SIZEOF_INT > 2
|
|
|
256 typedef int idx_T;
|
|
|
257 #else
|
|
|
258 typedef long idx_T;
|
|
|
259 #endif
|
|
|
260
|
|
|
261 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
262 * comes above it. */
|
|
300
|
263 #define WF_REGION 0x01 /* region byte follows */
|
|
|
264 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
265 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
266 #define WF_RARE 0x08 /* rare word */
|
|
307
|
267 #define WF_BANNED 0x10 /* bad word */
|
|
390
|
268 #define WF_PFX 0x20 /* prefix ID follows */
|
|
389
|
269 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
270 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
271
|
|
390
|
272 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
273 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
|
274
|
|
389
|
275 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
276
|
|
455
|
277 /* flags for <pflags> */
|
|
|
278 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
279 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
280 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
281
|
|
390
|
282 /* flags for postponed prefixes. Must be above prefixID (one byte)
|
|
|
283 * and prefcondnr (two bytes). */
|
|
455
|
284 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
285 * postponed prefix */
|
|
|
286 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
287 * postponed prefix */
|
|
|
288 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
289 * postponed prefix */
|
|
390
|
290
|
|
|
291 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
292 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
293 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
294 * postponed prefix: no <pflags> */
|
|
|
295 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
296 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
297 * postponed prefix: <pflags> follows */
|
|
|
298 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
299 * follow; never used in prefix tree */
|
|
|
300 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
301
|
|
323
|
302 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
303 * and si_sal. Not for sl_sal!
|
|
323
|
304 * One replacement: from "ft_from" to "ft_to". */
|
|
|
305 typedef struct fromto_S
|
|
236
|
306 {
|
|
323
|
307 char_u *ft_from;
|
|
|
308 char_u *ft_to;
|
|
|
309 } fromto_T;
|
|
236
|
310
|
|
344
|
311 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
312 * The info is split for quick processing by spell_soundfold().
|
|
|
313 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
314 typedef struct salitem_S
|
|
|
315 {
|
|
|
316 char_u *sm_lead; /* leading letters */
|
|
|
317 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
318 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
319 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
320 char_u *sm_to; /* replacement. */
|
|
372
|
321 #ifdef FEAT_MBYTE
|
|
|
322 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
323 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
324 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
325 #endif
|
|
344
|
326 } salitem_T;
|
|
|
327
|
|
375
|
328 #ifdef FEAT_MBYTE
|
|
|
329 typedef int salfirst_T;
|
|
|
330 #else
|
|
|
331 typedef short salfirst_T;
|
|
|
332 #endif
|
|
|
333
|
|
236
|
334 /*
|
|
243
|
335 * Structure used to store words and other info for one language, loaded from
|
|
|
336 * a .spl file.
|
|
300
|
337 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
338 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
339 *
|
|
|
340 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
341 * the number of possible bytes, sorted on byte value:
|
|
|
342 * <len> <byte1> <byte2> ...
|
|
|
343 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
344 * byte in "byts".
|
|
|
345 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
390
|
346 * the flags, region mask and prefixID for the word. There may be several
|
|
|
347 * zeros in sequence for alternative flag/region combinations.
|
|
236
|
348 */
|
|
|
349 typedef struct slang_S slang_T;
|
|
|
350 struct slang_S
|
|
|
351 {
|
|
|
352 slang_T *sl_next; /* next language */
|
|
|
353 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
354 char_u *sl_fname; /* name of .spl file */
|
|
323
|
355 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
356
|
|
300
|
357 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
358 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
359 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
360 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
361 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
362 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
363
|
|
236
|
364 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
365
|
|
376
|
366 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
367
|
|
339
|
368 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
369 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
370
|
|
323
|
371 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
372 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
373 there is none */
|
|
344
|
374 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
375 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
376 there is none */
|
|
375
|
377 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
378 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
379 * "sl_sal" is a list of wide char lists. */
|
|
323
|
380 int sl_followup; /* SAL followup */
|
|
|
381 int sl_collapse; /* SAL collapse_result */
|
|
|
382 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
383 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
384 #ifdef FEAT_MBYTE
|
|
|
385 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
386 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
387 #else
|
|
|
388 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
389 #endif
|
|
236
|
390 };
|
|
|
391
|
|
243
|
392 /* First language that is loaded, start of the linked list of loaded
|
|
|
393 * languages. */
|
|
236
|
394 static slang_T *first_lang = NULL;
|
|
|
395
|
|
323
|
396 /* Flags used in .spl file for soundsalike flags. */
|
|
|
397 #define SAL_F0LLOWUP 1
|
|
|
398 #define SAL_COLLAPSE 2
|
|
|
399 #define SAL_REM_ACCENTS 4
|
|
375
|
400 #define SAL_SOFO 8 /* SOFOFROM and SOFOTO instead of SAL */
|
|
323
|
401
|
|
236
|
402 /*
|
|
|
403 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
404 */
|
|
|
405 typedef struct langp_S
|
|
|
406 {
|
|
|
407 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
408 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
409 } langp_T;
|
|
|
410
|
|
|
411 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
412
|
|
307
|
413 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
414
|
|
|
415 /* Result values. Lower number is accepted over higher one. */
|
|
|
416 #define SP_BANNED -1
|
|
236
|
417 #define SP_OK 0
|
|
307
|
418 #define SP_RARE 1
|
|
|
419 #define SP_LOCAL 2
|
|
|
420 #define SP_BAD 3
|
|
236
|
421
|
|
455
|
422 #define VIMSPELLMAGIC "VIMspell09" /* string at start of Vim spell file */
|
|
236
|
423 #define VIMSPELLMAGICL 10
|
|
|
424
|
|
381
|
425 /* file used for "zG" and "zW" */
|
|
385
|
426 static char_u *int_wordlist = NULL;
|
|
381
|
427
|
|
236
|
428 /*
|
|
323
|
429 * Information used when looking for suggestions.
|
|
|
430 */
|
|
|
431 typedef struct suginfo_S
|
|
|
432 {
|
|
|
433 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
434 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
435 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
436 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
437 char_u *su_badptr; /* start of bad word in line */
|
|
|
438 int su_badlen; /* length of detected bad word in line */
|
|
346
|
439 int su_badflags; /* caps flags for bad word */
|
|
323
|
440 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
441 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
442 hashtab_T su_banned; /* table with banned words */
|
|
|
443 } suginfo_T;
|
|
|
444
|
|
|
445 /* One word suggestion. Used in "si_ga". */
|
|
|
446 typedef struct suggest_S
|
|
|
447 {
|
|
|
448 char_u *st_word; /* suggested word, allocated string */
|
|
|
449 int st_orglen; /* length of replaced text */
|
|
|
450 int st_score; /* lower is better */
|
|
344
|
451 int st_altscore; /* used when st_score compares equal */
|
|
|
452 int st_salscore; /* st_score is for soundalike */
|
|
324
|
453 int st_had_bonus; /* bonus already included in score */
|
|
323
|
454 } suggest_T;
|
|
|
455
|
|
344
|
456 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
457
|
|
324
|
458 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
459 * called the score may change, thus we need to keep more than what is
|
|
|
460 * displayed. */
|
|
351
|
461 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
462
|
|
|
463 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
464 * of suggestions that are not going to be displayed. */
|
|
344
|
465 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
466
|
|
|
467 /* score for various changes */
|
|
344
|
468 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
469 #define SCORE_ICASE 52 /* slightly different case */
|
|
|
470 #define SCORE_REGION 70 /* word is for different region */
|
|
|
471 #define SCORE_RARE 180 /* rare word */
|
|
|
472 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
473 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
474 #define SCORE_REP 87 /* REP replacement */
|
|
|
475 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
476 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
324
|
477 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
478 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
324
|
479 #define SCORE_INS 96 /* insert a character */
|
|
358
|
480 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
475
|
481 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
482 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
483
|
|
372
|
484 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
485 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
486 * 350 allows for about three changes. */
|
|
344
|
487
|
|
|
488 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
489 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
490
|
|
|
491 /*
|
|
236
|
492 * Structure to store info for word matching.
|
|
|
493 */
|
|
|
494 typedef struct matchinf_S
|
|
|
495 {
|
|
|
496 langp_T *mi_lp; /* info for language and region */
|
|
243
|
497
|
|
|
498 /* pointers to original text to be checked */
|
|
236
|
499 char_u *mi_word; /* start of word being checked */
|
|
339
|
500 char_u *mi_end; /* end of matching word so far */
|
|
243
|
501 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
502 char_u *mi_cend; /* char after what was used for
|
|
|
503 mi_capflags */
|
|
243
|
504
|
|
|
505 /* case-folded text */
|
|
|
506 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
507 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
508
|
|
339
|
509 /* for when checking word after a prefix */
|
|
|
510 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
|
511 prefixID/condition */
|
|
|
512 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
513 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
514 #ifdef FEAT_MBYTE
|
|
|
515 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
516 case */
|
|
|
517 #else
|
|
|
518 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
519 #endif
|
|
339
|
520
|
|
243
|
521 /* others */
|
|
236
|
522 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
523 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
524 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
525 } matchinf_T;
|
|
|
526
|
|
307
|
527 /*
|
|
|
528 * The tables used for recognizing word characters according to spelling.
|
|
|
529 * These are only used for the first 256 characters of 'encoding'.
|
|
|
530 */
|
|
|
531 typedef struct spelltab_S
|
|
|
532 {
|
|
|
533 char_u st_isw[256]; /* flags: is word char */
|
|
|
534 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
535 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
536 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
537 } spelltab_T;
|
|
|
538
|
|
|
539 static spelltab_T spelltab;
|
|
|
540 static int did_set_spelltab;
|
|
|
541
|
|
324
|
542 #define CF_WORD 0x01
|
|
|
543 #define CF_UPPER 0x02
|
|
307
|
544
|
|
|
545 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
546 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
547 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
548 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
549 #ifdef FEAT_MBYTE
|
|
|
550 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
551 #endif
|
|
339
|
552 static void write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
553
|
|
|
554 /*
|
|
344
|
555 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
556 */
|
|
|
557 typedef enum
|
|
|
558 {
|
|
344
|
559 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
560 * ends); if badword ends there is a match, otherwise
|
|
|
561 * try splitting word. */
|
|
375
|
562 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
563 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
564 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
565 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
566 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
567 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
568 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
569 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
570 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
571 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
572 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
573 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
574 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
575 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
576 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
577 STATE_FINAL /* End of this node. */
|
|
|
578 } state_T;
|
|
|
579
|
|
|
580 /*
|
|
346
|
581 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
582 */
|
|
|
583 typedef struct trystate_S
|
|
|
584 {
|
|
330
|
585 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
586 int ts_score; /* score */
|
|
344
|
587 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
588 short ts_curi; /* index in list of child nodes */
|
|
|
589 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
590 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
591 char_u ts_twordlen; /* valid length of tword[] */
|
|
375
|
592 char_u ts_prefixdepth; /* stack depth for end of prefix or PREFIXTREE
|
|
|
593 * or NOPREFIX */
|
|
330
|
594 #ifdef FEAT_MBYTE
|
|
|
595 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
596 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
597 char_u ts_isdiff; /* DIFF_ values */
|
|
|
598 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
599 #endif
|
|
323
|
600 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
330
|
601 char_u ts_save_splitoff; /* su_splitoff saved here */
|
|
346
|
602 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
603 } trystate_T;
|
|
|
604
|
|
330
|
605 /* values for ts_isdiff */
|
|
|
606 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
607 #define DIFF_YES 1 /* different byte found */
|
|
|
608 #define DIFF_INSERT 2 /* inserting character */
|
|
|
609
|
|
375
|
610 /* special values ts_prefixdepth */
|
|
|
611 #define PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
612 #define NOPREFIX 0xff /* not using prefixes */
|
|
|
613
|
|
339
|
614 /* mode values for find_word */
|
|
|
615 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
616 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
617 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
618
|
|
236
|
619 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
620 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
621 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
622 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
455
|
623 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
339
|
624 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
625 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
626 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
627 static int no_spell_checking __ARGS((void));
|
|
307
|
628 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
629 static char_u *spell_enc __ARGS((void));
|
|
385
|
630 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
631 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
632 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
633 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
381
|
634 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
635 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
636 #ifdef FEAT_MBYTE
|
|
|
637 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
638 #endif
|
|
339
|
639 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
640 static void clear_midword __ARGS((buf_T *buf));
|
|
|
641 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
642 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
643 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
644 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
645 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
389
|
646 static int set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
647 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
648 static void write_spell_chartab __ARGS((FILE *fd));
|
|
|
649 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
650 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
388
|
651 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
652 #ifdef FEAT_EVAL
|
|
|
653 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
654 #endif
|
|
|
655 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
656 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
657 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
658 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
659 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
660 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
661 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
662 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
663 #ifdef FEAT_MBYTE
|
|
|
664 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
665 #endif
|
|
323
|
666 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
667 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
668 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
669 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
670 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
671 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
672 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
673 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
674 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
675 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
676 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
677 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
678 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
679 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
680 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
681 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
682 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
683 #ifdef FEAT_MBYTE
|
|
375
|
684 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
685 #endif
|
|
344
|
686 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
687 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
688 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
689 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
690
|
|
324
|
691 /*
|
|
|
692 * Use our own character-case definitions, because the current locale may
|
|
|
693 * differ from what the .spl file uses.
|
|
|
694 * These must not be called with negative number!
|
|
|
695 */
|
|
|
696 #ifndef FEAT_MBYTE
|
|
|
697 /* Non-multi-byte implementation. */
|
|
|
698 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
699 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
700 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
701 #else
|
|
447
|
702 # if defined(HAVE_WCHAR_H)
|
|
|
703 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
704 # endif
|
|
324
|
705 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
706 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
707 * the "w" library function for characters above 255 if available. */
|
|
|
708 # ifdef HAVE_TOWLOWER
|
|
|
709 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
710 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
711 # else
|
|
|
712 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
713 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
714 # endif
|
|
|
715
|
|
|
716 # ifdef HAVE_TOWUPPER
|
|
|
717 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
718 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
719 # else
|
|
|
720 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
721 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
722 # endif
|
|
|
723
|
|
|
724 # ifdef HAVE_ISWUPPER
|
|
|
725 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
726 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
727 # else
|
|
|
728 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
729 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
730 # endif
|
|
|
731 #endif
|
|
|
732
|
|
307
|
733
|
|
|
734 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
735 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
469
|
736 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
737
|
|
|
738 /*
|
|
|
739 * Main spell-checking function.
|
|
300
|
740 * "ptr" points to a character that could be the start of a word.
|
|
236
|
741 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
742 * or when it's OK it remains unchanged.
|
|
|
743 * This must only be called when 'spelllang' is not empty.
|
|
323
|
744 *
|
|
385
|
745 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
746 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
747 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
748 * worry.
|
|
323
|
749 *
|
|
236
|
750 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
751 * caller can skip over the word.
|
|
|
752 */
|
|
|
753 int
|
|
385
|
754 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
755 win_T *wp; /* current window */
|
|
|
756 char_u *ptr;
|
|
|
757 int *attrp;
|
|
385
|
758 int *capcol; /* column to check for Capital */
|
|
236
|
759 {
|
|
|
760 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
761 be passed to functions quickly. */
|
|
344
|
762 int nrlen = 0; /* found a number first */
|
|
385
|
763 int c;
|
|
236
|
764
|
|
307
|
765 /* A word never starts at a space or a control character. Return quickly
|
|
|
766 * then, skipping over the character. */
|
|
|
767 if (*ptr <= ' ')
|
|
|
768 return 1;
|
|
236
|
769
|
|
344
|
770 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
771 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
772 * "3GPP". */
|
|
300
|
773 if (*ptr >= '0' && *ptr <= '9')
|
|
|
774 {
|
|
316
|
775 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
776 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
777 else
|
|
|
778 {
|
|
344
|
779 mi.mi_end = skipdigits(ptr);
|
|
|
780 nrlen = mi.mi_end - ptr;
|
|
300
|
781 }
|
|
376
|
782 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
783 return (int)(mi.mi_end - ptr);
|
|
346
|
784
|
|
|
785 /* Try including the digits in the word. */
|
|
|
786 mi.mi_fend = ptr + nrlen;
|
|
|
787 }
|
|
|
788 else
|
|
|
789 mi.mi_fend = ptr;
|
|
|
790
|
|
|
791 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
792 mi.mi_word = ptr;
|
|
376
|
793 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
794 {
|
|
|
795 do
|
|
|
796 {
|
|
307
|
797 mb_ptr_adv(mi.mi_fend);
|
|
376
|
798 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
799
|
|
|
800 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
801 {
|
|
|
802 /* Check word starting with capital letter. */
|
|
455
|
803 c = PTR2CHAR(ptr);
|
|
385
|
804 if (!SPELL_ISUPPER(c))
|
|
|
805 {
|
|
|
806 *attrp = highlight_attr[HLF_SPC];
|
|
|
807 return (int)(mi.mi_fend - ptr);
|
|
|
808 }
|
|
|
809 }
|
|
|
810 }
|
|
|
811 if (capcol != NULL)
|
|
|
812 *capcol = -1;
|
|
344
|
813
|
|
|
814 /* We always use the characters up to the next non-word character,
|
|
|
815 * also for bad words. */
|
|
|
816 mi.mi_end = mi.mi_fend;
|
|
|
817
|
|
|
818 /* Check caps type later. */
|
|
|
819 mi.mi_capflags = 0;
|
|
|
820 mi.mi_cend = NULL;
|
|
376
|
821 mi.mi_buf = wp->w_buffer;
|
|
344
|
822
|
|
|
823 /* Include one non-word character so that we can check for the
|
|
|
824 * word end. */
|
|
|
825 if (*mi.mi_fend != NUL)
|
|
|
826 mb_ptr_adv(mi.mi_fend);
|
|
|
827
|
|
|
828 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
829 MAXWLEN + 1);
|
|
|
830 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
831
|
|
|
832 /* The word is bad unless we recognize it. */
|
|
|
833 mi.mi_result = SP_BAD;
|
|
|
834
|
|
|
835 /*
|
|
|
836 * Loop over the languages specified in 'spelllang'.
|
|
|
837 * We check them all, because a matching word may be longer than an
|
|
|
838 * already found matching word.
|
|
|
839 */
|
|
|
840 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
841 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
842 {
|
|
|
843 /* Check for a matching word in case-folded words. */
|
|
|
844 find_word(&mi, FIND_FOLDWORD);
|
|
|
845
|
|
|
846 /* Check for a matching word in keep-case words. */
|
|
|
847 find_word(&mi, FIND_KEEPWORD);
|
|
|
848
|
|
|
849 /* Check for matching prefixes. */
|
|
|
850 find_prefix(&mi);
|
|
|
851 }
|
|
|
852
|
|
|
853 if (mi.mi_result != SP_OK)
|
|
|
854 {
|
|
346
|
855 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
856 * rare and local words, e.g., "3GPP". */
|
|
344
|
857 if (nrlen > 0)
|
|
346
|
858 {
|
|
|
859 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
860 return nrlen;
|
|
|
861 }
|
|
344
|
862
|
|
|
863 /* When we are at a non-word character there is no error, just
|
|
|
864 * skip over the character (try looking for a word after it). */
|
|
390
|
865 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
866 {
|
|
385
|
867 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
868 {
|
|
|
869 regmatch_T regmatch;
|
|
|
870
|
|
|
871 /* Check for end of sentence. */
|
|
|
872 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
873 regmatch.rm_ic = FALSE;
|
|
|
874 if (vim_regexec(®match, ptr, 0))
|
|
|
875 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
876 }
|
|
|
877
|
|
344
|
878 #ifdef FEAT_MBYTE
|
|
|
879 if (has_mbyte)
|
|
474
|
880 return (*mb_ptr2len)(ptr);
|
|
344
|
881 #endif
|
|
|
882 return 1;
|
|
300
|
883 }
|
|
243
|
884
|
|
344
|
885 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
886 *attrp = highlight_attr[HLF_SPB];
|
|
|
887 else if (mi.mi_result == SP_RARE)
|
|
|
888 *attrp = highlight_attr[HLF_SPR];
|
|
|
889 else
|
|
|
890 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
891 }
|
|
|
892
|
|
300
|
893 return (int)(mi.mi_end - ptr);
|
|
236
|
894 }
|
|
|
895
|
|
|
896 /*
|
|
300
|
897 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
898 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
899 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
900 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
901 * tree.
|
|
300
|
902 *
|
|
|
903 * For a match mip->mi_result is updated.
|
|
243
|
904 */
|
|
|
905 static void
|
|
339
|
906 find_word(mip, mode)
|
|
243
|
907 matchinf_T *mip;
|
|
339
|
908 int mode;
|
|
243
|
909 {
|
|
324
|
910 idx_T arridx = 0;
|
|
300
|
911 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
912 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
913 int endidxcnt = 0;
|
|
|
914 int len;
|
|
|
915 int wlen = 0;
|
|
|
916 int flen;
|
|
|
917 int c;
|
|
|
918 char_u *ptr;
|
|
324
|
919 idx_T lo, hi, m;
|
|
243
|
920 #ifdef FEAT_MBYTE
|
|
300
|
921 char_u *s;
|
|
339
|
922 char_u *p;
|
|
307
|
923 #endif
|
|
|
924 int res = SP_BAD;
|
|
300
|
925 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
926 unsigned flags;
|
|
|
927 char_u *byts;
|
|
324
|
928 idx_T *idxs;
|
|
339
|
929
|
|
|
930 if (mode == FIND_KEEPWORD)
|
|
236
|
931 {
|
|
300
|
932 /* Check for word with matching case in keep-case tree. */
|
|
|
933 ptr = mip->mi_word;
|
|
|
934 flen = 9999; /* no case folding, always enough bytes */
|
|
|
935 byts = slang->sl_kbyts;
|
|
|
936 idxs = slang->sl_kidxs;
|
|
236
|
937 }
|
|
|
938 else
|
|
|
939 {
|
|
300
|
940 /* Check for case-folded in case-folded tree. */
|
|
|
941 ptr = mip->mi_fword;
|
|
|
942 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
943 byts = slang->sl_fbyts;
|
|
|
944 idxs = slang->sl_fidxs;
|
|
339
|
945
|
|
|
946 if (mode == FIND_PREFIX)
|
|
|
947 {
|
|
|
948 /* Skip over the prefix. */
|
|
|
949 wlen = mip->mi_prefixlen;
|
|
|
950 flen -= mip->mi_prefixlen;
|
|
|
951 }
|
|
243
|
952 }
|
|
|
953
|
|
300
|
954 if (byts == NULL)
|
|
|
955 return; /* array is empty */
|
|
236
|
956
|
|
|
957 /*
|
|
307
|
958 * Repeat advancing in the tree until:
|
|
|
959 * - there is a byte that doesn't match,
|
|
|
960 * - we reach the end of the tree,
|
|
|
961 * - or we reach the end of the line.
|
|
236
|
962 */
|
|
300
|
963 for (;;)
|
|
236
|
964 {
|
|
346
|
965 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
966 flen = fold_more(mip);
|
|
300
|
967
|
|
|
968 len = byts[arridx++];
|
|
|
969
|
|
|
970 /* If the first possible byte is a zero the word could end here.
|
|
|
971 * Remember this index, we first check for the longest word. */
|
|
|
972 if (byts[arridx] == 0)
|
|
|
973 {
|
|
307
|
974 if (endidxcnt == MAXWLEN)
|
|
|
975 {
|
|
|
976 /* Must be a corrupted spell file. */
|
|
|
977 EMSG(_(e_format));
|
|
|
978 return;
|
|
|
979 }
|
|
300
|
980 endlen[endidxcnt] = wlen;
|
|
|
981 endidx[endidxcnt++] = arridx++;
|
|
|
982 --len;
|
|
|
983
|
|
|
984 /* Skip over the zeros, there can be several flag/region
|
|
|
985 * combinations. */
|
|
|
986 while (len > 0 && byts[arridx] == 0)
|
|
|
987 {
|
|
|
988 ++arridx;
|
|
|
989 --len;
|
|
|
990 }
|
|
|
991 if (len == 0)
|
|
|
992 break; /* no children, word must end here */
|
|
|
993 }
|
|
|
994
|
|
|
995 /* Stop looking at end of the line. */
|
|
|
996 if (ptr[wlen] == NUL)
|
|
|
997 break;
|
|
|
998
|
|
|
999 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1000 c = ptr[wlen];
|
|
346
|
1001 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1002 c = ' ';
|
|
300
|
1003 lo = arridx;
|
|
|
1004 hi = arridx + len - 1;
|
|
|
1005 while (lo < hi)
|
|
|
1006 {
|
|
|
1007 m = (lo + hi) / 2;
|
|
|
1008 if (byts[m] > c)
|
|
|
1009 hi = m - 1;
|
|
|
1010 else if (byts[m] < c)
|
|
|
1011 lo = m + 1;
|
|
|
1012 else
|
|
|
1013 {
|
|
|
1014 lo = hi = m;
|
|
|
1015 break;
|
|
236
|
1016 }
|
|
|
1017 }
|
|
300
|
1018
|
|
|
1019 /* Stop if there is no matching byte. */
|
|
|
1020 if (hi < lo || byts[lo] != c)
|
|
|
1021 break;
|
|
|
1022
|
|
|
1023 /* Continue at the child (if there is one). */
|
|
|
1024 arridx = idxs[lo];
|
|
|
1025 ++wlen;
|
|
|
1026 --flen;
|
|
346
|
1027
|
|
|
1028 /* One space in the good word may stand for several spaces in the
|
|
|
1029 * checked word. */
|
|
|
1030 if (c == ' ')
|
|
|
1031 {
|
|
|
1032 for (;;)
|
|
|
1033 {
|
|
|
1034 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1035 flen = fold_more(mip);
|
|
|
1036 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1037 break;
|
|
|
1038 ++wlen;
|
|
|
1039 --flen;
|
|
|
1040 }
|
|
|
1041 }
|
|
236
|
1042 }
|
|
|
1043
|
|
300
|
1044 /*
|
|
|
1045 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1046 * first.
|
|
|
1047 */
|
|
|
1048 while (endidxcnt > 0)
|
|
|
1049 {
|
|
|
1050 --endidxcnt;
|
|
|
1051 arridx = endidx[endidxcnt];
|
|
|
1052 wlen = endlen[endidxcnt];
|
|
236
|
1053
|
|
300
|
1054 #ifdef FEAT_MBYTE
|
|
|
1055 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1056 continue; /* not at first byte of character */
|
|
|
1057 #endif
|
|
376
|
1058 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
300
|
1059 continue; /* next char is a word character */
|
|
|
1060
|
|
|
1061 #ifdef FEAT_MBYTE
|
|
339
|
1062 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1063 {
|
|
|
1064 /* Compute byte length in original word, length may change
|
|
339
|
1065 * when folding case. This can be slow, take a shortcut when the
|
|
|
1066 * case-folded word is equal to the keep-case word. */
|
|
300
|
1067 p = mip->mi_word;
|
|
339
|
1068 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1069 {
|
|
|
1070 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1071 mb_ptr_adv(p);
|
|
|
1072 wlen = p - mip->mi_word;
|
|
|
1073 }
|
|
300
|
1074 }
|
|
|
1075 #endif
|
|
236
|
1076
|
|
339
|
1077 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1078 * prefix ID.
|
|
|
1079 * Repeat this if there are more flags/region alternatives until there
|
|
|
1080 * is a match. */
|
|
|
1081 res = SP_BAD;
|
|
|
1082 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1083 --len, ++arridx)
|
|
300
|
1084 {
|
|
|
1085 flags = idxs[arridx];
|
|
324
|
1086
|
|
339
|
1087 /* For the fold-case tree check that the case of the checked word
|
|
|
1088 * matches with what the word in the tree requires.
|
|
|
1089 * For keep-case tree the case is always right. For prefixes we
|
|
|
1090 * don't bother to check. */
|
|
|
1091 if (mode == FIND_FOLDWORD)
|
|
300
|
1092 {
|
|
|
1093 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1094 {
|
|
323
|
1095 /* mi_capflags was set for a different word length, need
|
|
|
1096 * to do it again. */
|
|
300
|
1097 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1098 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1099 }
|
|
|
1100
|
|
346
|
1101 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1102 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1103 continue;
|
|
300
|
1104 }
|
|
236
|
1105
|
|
339
|
1106 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1107 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1108 * mip->mi_prefarridx that find_prefix() filled. */
|
|
339
|
1109 if (mode == FIND_PREFIX)
|
|
300
|
1110 {
|
|
339
|
1111 /* The prefix ID is stored two bytes above the flags. */
|
|
366
|
1112 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1113 flags,
|
|
455
|
1114 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1115 FALSE);
|
|
366
|
1116 if (c == 0)
|
|
339
|
1117 continue;
|
|
366
|
1118
|
|
|
1119 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1120 if (c & WF_RAREPFX)
|
|
|
1121 flags |= WF_RARE;
|
|
339
|
1122 }
|
|
|
1123
|
|
|
1124 if (flags & WF_BANNED)
|
|
|
1125 res = SP_BANNED;
|
|
|
1126 else if (flags & WF_REGION)
|
|
|
1127 {
|
|
|
1128 /* Check region. */
|
|
390
|
1129 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1130 res = SP_OK;
|
|
339
|
1131 else
|
|
|
1132 res = SP_LOCAL;
|
|
300
|
1133 }
|
|
339
|
1134 else if (flags & WF_RARE)
|
|
|
1135 res = SP_RARE;
|
|
307
|
1136 else
|
|
339
|
1137 res = SP_OK;
|
|
|
1138
|
|
|
1139 /* Always use the longest match and the best result. */
|
|
|
1140 if (mip->mi_result > res)
|
|
|
1141 {
|
|
|
1142 mip->mi_result = res;
|
|
|
1143 mip->mi_end = mip->mi_word + wlen;
|
|
|
1144 }
|
|
351
|
1145 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1146 mip->mi_end = mip->mi_word + wlen;
|
|
|
1147
|
|
|
1148 if (res == SP_OK)
|
|
|
1149 break;
|
|
300
|
1150 }
|
|
|
1151
|
|
307
|
1152 if (res == SP_OK)
|
|
300
|
1153 break;
|
|
|
1154 }
|
|
236
|
1155 }
|
|
|
1156
|
|
323
|
1157 /*
|
|
390
|
1158 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1159 * ID in "flags" for the word "word".
|
|
366
|
1160 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1161 */
|
|
|
1162 static int
|
|
455
|
1163 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1164 int totprefcnt; /* nr of prefix IDs */
|
|
|
1165 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1166 int flags;
|
|
351
|
1167 char_u *word;
|
|
|
1168 slang_T *slang;
|
|
455
|
1169 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1170 {
|
|
|
1171 int prefcnt;
|
|
|
1172 int pidx;
|
|
|
1173 regprog_T *rp;
|
|
|
1174 regmatch_T regmatch;
|
|
390
|
1175 int prefid;
|
|
|
1176
|
|
|
1177 prefid = (unsigned)flags >> 24;
|
|
351
|
1178 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1179 {
|
|
|
1180 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1181
|
|
|
1182 /* Check the prefix ID. */
|
|
|
1183 if (prefid != (pidx & 0xff))
|
|
|
1184 continue;
|
|
|
1185
|
|
390
|
1186 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1187 * suffix. */
|
|
|
1188 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1189 continue;
|
|
|
1190
|
|
351
|
1191 /* Check the condition, if there is one. The condition index is
|
|
366
|
1192 * stored in the two bytes above the prefix ID byte. */
|
|
|
1193 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1194 if (rp != NULL)
|
|
|
1195 {
|
|
|
1196 regmatch.regprog = rp;
|
|
|
1197 regmatch.rm_ic = FALSE;
|
|
|
1198 if (!vim_regexec(®match, word, 0))
|
|
|
1199 continue;
|
|
|
1200 }
|
|
455
|
1201 else if (cond_req)
|
|
|
1202 continue;
|
|
|
1203
|
|
|
1204 /* It's a match! Return the WF_ flags. */
|
|
366
|
1205 return pidx;
|
|
|
1206 }
|
|
|
1207 return 0;
|
|
351
|
1208 }
|
|
|
1209
|
|
|
1210 /*
|
|
339
|
1211 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1212 * If it does, then check the following word.
|
|
|
1213 *
|
|
|
1214 * For a match mip->mi_result is updated.
|
|
|
1215 */
|
|
|
1216 static void
|
|
|
1217 find_prefix(mip)
|
|
|
1218 matchinf_T *mip;
|
|
|
1219 {
|
|
|
1220 idx_T arridx = 0;
|
|
|
1221 int len;
|
|
|
1222 int wlen = 0;
|
|
|
1223 int flen;
|
|
|
1224 int c;
|
|
|
1225 char_u *ptr;
|
|
|
1226 idx_T lo, hi, m;
|
|
|
1227 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1228 char_u *byts;
|
|
|
1229 idx_T *idxs;
|
|
|
1230
|
|
375
|
1231 byts = slang->sl_pbyts;
|
|
|
1232 if (byts == NULL)
|
|
|
1233 return; /* array is empty */
|
|
|
1234
|
|
339
|
1235 /* We use the case-folded word here, since prefixes are always
|
|
|
1236 * case-folded. */
|
|
|
1237 ptr = mip->mi_fword;
|
|
|
1238 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1239 idxs = slang->sl_pidxs;
|
|
|
1240
|
|
|
1241 /*
|
|
|
1242 * Repeat advancing in the tree until:
|
|
|
1243 * - there is a byte that doesn't match,
|
|
|
1244 * - we reach the end of the tree,
|
|
|
1245 * - or we reach the end of the line.
|
|
|
1246 */
|
|
|
1247 for (;;)
|
|
|
1248 {
|
|
|
1249 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1250 flen = fold_more(mip);
|
|
|
1251
|
|
|
1252 len = byts[arridx++];
|
|
|
1253
|
|
|
1254 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1255 * Check if the following word matches and supports the prefix. */
|
|
|
1256 if (byts[arridx] == 0)
|
|
|
1257 {
|
|
|
1258 /* There can be several prefixes with different conditions. We
|
|
|
1259 * try them all, since we don't know which one will give the
|
|
|
1260 * longest match. The word is the same each time, pass the list
|
|
|
1261 * of possible prefixes to find_word(). */
|
|
|
1262 mip->mi_prefarridx = arridx;
|
|
|
1263 mip->mi_prefcnt = len;
|
|
|
1264 while (len > 0 && byts[arridx] == 0)
|
|
|
1265 {
|
|
|
1266 ++arridx;
|
|
|
1267 --len;
|
|
|
1268 }
|
|
|
1269 mip->mi_prefcnt -= len;
|
|
|
1270
|
|
|
1271 /* Find the word that comes after the prefix. */
|
|
|
1272 mip->mi_prefixlen = wlen;
|
|
455
|
1273 #ifdef FEAT_MBYTE
|
|
|
1274 if (has_mbyte)
|
|
|
1275 {
|
|
|
1276 /* Case-folded length may differ from original length. */
|
|
|
1277 mip->mi_cprefixlen = nofold_len(mip->mi_fword, wlen,
|
|
|
1278 mip->mi_word);
|
|
|
1279 }
|
|
|
1280 else
|
|
|
1281 mip->mi_cprefixlen = wlen;
|
|
|
1282 #endif
|
|
339
|
1283 find_word(mip, FIND_PREFIX);
|
|
|
1284
|
|
|
1285
|
|
|
1286 if (len == 0)
|
|
|
1287 break; /* no children, word must end here */
|
|
|
1288 }
|
|
|
1289
|
|
|
1290 /* Stop looking at end of the line. */
|
|
|
1291 if (ptr[wlen] == NUL)
|
|
|
1292 break;
|
|
|
1293
|
|
|
1294 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1295 c = ptr[wlen];
|
|
|
1296 lo = arridx;
|
|
|
1297 hi = arridx + len - 1;
|
|
|
1298 while (lo < hi)
|
|
|
1299 {
|
|
|
1300 m = (lo + hi) / 2;
|
|
|
1301 if (byts[m] > c)
|
|
|
1302 hi = m - 1;
|
|
|
1303 else if (byts[m] < c)
|
|
|
1304 lo = m + 1;
|
|
|
1305 else
|
|
|
1306 {
|
|
|
1307 lo = hi = m;
|
|
|
1308 break;
|
|
|
1309 }
|
|
|
1310 }
|
|
|
1311
|
|
|
1312 /* Stop if there is no matching byte. */
|
|
|
1313 if (hi < lo || byts[lo] != c)
|
|
|
1314 break;
|
|
|
1315
|
|
|
1316 /* Continue at the child (if there is one). */
|
|
|
1317 arridx = idxs[lo];
|
|
|
1318 ++wlen;
|
|
|
1319 --flen;
|
|
|
1320 }
|
|
|
1321 }
|
|
|
1322
|
|
|
1323 /*
|
|
|
1324 * Need to fold at least one more character. Do until next non-word character
|
|
|
1325 * for efficiency.
|
|
|
1326 * Return the length of the folded chars in bytes.
|
|
|
1327 */
|
|
|
1328 static int
|
|
|
1329 fold_more(mip)
|
|
|
1330 matchinf_T *mip;
|
|
|
1331 {
|
|
|
1332 int flen;
|
|
|
1333 char_u *p;
|
|
|
1334
|
|
|
1335 p = mip->mi_fend;
|
|
|
1336 do
|
|
|
1337 {
|
|
|
1338 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1339 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1340
|
|
|
1341 /* Include the non-word character so that we can check for the
|
|
|
1342 * word end. */
|
|
|
1343 if (*mip->mi_fend != NUL)
|
|
|
1344 mb_ptr_adv(mip->mi_fend);
|
|
|
1345
|
|
|
1346 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1347 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1348 MAXWLEN - mip->mi_fwordlen);
|
|
|
1349 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1350 mip->mi_fwordlen += flen;
|
|
|
1351 return flen;
|
|
|
1352 }
|
|
|
1353
|
|
|
1354 /*
|
|
323
|
1355 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1356 * case.
|
|
|
1357 */
|
|
|
1358 static int
|
|
389
|
1359 spell_valid_case(wordflags, treeflags)
|
|
|
1360 int wordflags; /* flags for the checked word. */
|
|
323
|
1361 int treeflags; /* flags for the word in the spell tree */
|
|
|
1362 {
|
|
389
|
1363 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1364 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1365 && ((treeflags & WF_ONECAP) == 0
|
|
|
1366 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1367 }
|
|
|
1368
|
|
351
|
1369 /*
|
|
|
1370 * Return TRUE if spell checking is not enabled.
|
|
|
1371 */
|
|
|
1372 static int
|
|
|
1373 no_spell_checking()
|
|
|
1374 {
|
|
|
1375 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1376 {
|
|
|
1377 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1378 return TRUE;
|
|
|
1379 }
|
|
|
1380 return FALSE;
|
|
|
1381 }
|
|
300
|
1382
|
|
236
|
1383 /*
|
|
|
1384 * Move to next spell error.
|
|
323
|
1385 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
236
|
1386 * Return OK if found, FAIL otherwise.
|
|
|
1387 */
|
|
|
1388 int
|
|
323
|
1389 spell_move_to(dir, allwords, curline)
|
|
236
|
1390 int dir; /* FORWARD or BACKWARD */
|
|
|
1391 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1392 int curline;
|
|
236
|
1393 {
|
|
249
|
1394 linenr_T lnum;
|
|
|
1395 pos_T found_pos;
|
|
236
|
1396 char_u *line;
|
|
|
1397 char_u *p;
|
|
346
|
1398 char_u *endp;
|
|
|
1399 int attr;
|
|
236
|
1400 int len;
|
|
249
|
1401 int has_syntax = syntax_present(curbuf);
|
|
|
1402 int col;
|
|
|
1403 int can_spell;
|
|
346
|
1404 char_u *buf = NULL;
|
|
|
1405 int buflen = 0;
|
|
|
1406 int skip = 0;
|
|
385
|
1407 int capcol = -1;
|
|
236
|
1408
|
|
351
|
1409 if (no_spell_checking())
|
|
236
|
1410 return FAIL;
|
|
|
1411
|
|
249
|
1412 /*
|
|
|
1413 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1414 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1415 *
|
|
|
1416 * When searching backwards, we continue in the line to find the last
|
|
|
1417 * bad word (in the cursor line: before the cursor).
|
|
346
|
1418 *
|
|
|
1419 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1420 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1421 * though...
|
|
249
|
1422 */
|
|
|
1423 lnum = curwin->w_cursor.lnum;
|
|
|
1424 found_pos.lnum = 0;
|
|
236
|
1425
|
|
|
1426 while (!got_int)
|
|
|
1427 {
|
|
249
|
1428 line = ml_get(lnum);
|
|
346
|
1429
|
|
|
1430 len = STRLEN(line);
|
|
|
1431 if (buflen < len + MAXWLEN + 2)
|
|
|
1432 {
|
|
|
1433 vim_free(buf);
|
|
|
1434 buflen = len + MAXWLEN + 2;
|
|
|
1435 buf = alloc(buflen);
|
|
|
1436 if (buf == NULL)
|
|
|
1437 break;
|
|
|
1438 }
|
|
|
1439
|
|
385
|
1440 /* In first line check first word for Capital. */
|
|
|
1441 if (lnum == 1)
|
|
|
1442 capcol = 0;
|
|
|
1443
|
|
|
1444 /* For checking first word with a capital skip white space. */
|
|
|
1445 if (capcol == 0)
|
|
|
1446 capcol = skipwhite(line) - line;
|
|
|
1447
|
|
346
|
1448 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1449 * possible. */
|
|
|
1450 STRCPY(buf, line);
|
|
|
1451 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1452 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1453
|
|
|
1454 p = buf + skip;
|
|
|
1455 endp = buf + len;
|
|
|
1456 while (p < endp)
|
|
236
|
1457 {
|
|
300
|
1458 /* When searching backward don't search after the cursor. */
|
|
|
1459 if (dir == BACKWARD
|
|
|
1460 && lnum == curwin->w_cursor.lnum
|
|
346
|
1461 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1462 break;
|
|
249
|
1463
|
|
300
|
1464 /* start of word */
|
|
346
|
1465 attr = 0;
|
|
385
|
1466 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1467
|
|
300
|
1468 if (attr != 0)
|
|
|
1469 {
|
|
|
1470 /* We found a bad word. Check the attribute. */
|
|
|
1471 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1472 {
|
|
300
|
1473 /* When searching forward only accept a bad word after
|
|
|
1474 * the cursor. */
|
|
|
1475 if (dir == BACKWARD
|
|
|
1476 || lnum > curwin->w_cursor.lnum
|
|
|
1477 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1478 && (colnr_T)(curline ? p - buf + len
|
|
|
1479 : p - buf)
|
|
300
|
1480 > curwin->w_cursor.col))
|
|
236
|
1481 {
|
|
300
|
1482 if (has_syntax)
|
|
249
|
1483 {
|
|
346
|
1484 col = p - buf;
|
|
300
|
1485 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1486 FALSE, &can_spell);
|
|
|
1487 }
|
|
|
1488 else
|
|
|
1489 can_spell = TRUE;
|
|
249
|
1490
|
|
300
|
1491 if (can_spell)
|
|
|
1492 {
|
|
|
1493 found_pos.lnum = lnum;
|
|
346
|
1494 found_pos.col = p - buf;
|
|
249
|
1495 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1496 found_pos.coladd = 0;
|
|
249
|
1497 #endif
|
|
300
|
1498 if (dir == FORWARD)
|
|
|
1499 {
|
|
|
1500 /* No need to search further. */
|
|
|
1501 curwin->w_cursor = found_pos;
|
|
346
|
1502 vim_free(buf);
|
|
300
|
1503 return OK;
|
|
249
|
1504 }
|
|
|
1505 }
|
|
236
|
1506 }
|
|
|
1507 }
|
|
|
1508 }
|
|
|
1509
|
|
300
|
1510 /* advance to character after the word */
|
|
|
1511 p += len;
|
|
385
|
1512 capcol -= len;
|
|
236
|
1513 }
|
|
|
1514
|
|
323
|
1515 if (curline)
|
|
346
|
1516 break; /* only check cursor line */
|
|
323
|
1517
|
|
236
|
1518 /* Advance to next line. */
|
|
249
|
1519 if (dir == BACKWARD)
|
|
|
1520 {
|
|
|
1521 if (found_pos.lnum != 0)
|
|
|
1522 {
|
|
|
1523 /* Use the last match in the line. */
|
|
|
1524 curwin->w_cursor = found_pos;
|
|
346
|
1525 vim_free(buf);
|
|
249
|
1526 return OK;
|
|
|
1527 }
|
|
|
1528 if (lnum == 1)
|
|
346
|
1529 break;
|
|
249
|
1530 --lnum;
|
|
385
|
1531 capcol = -1;
|
|
249
|
1532 }
|
|
|
1533 else
|
|
|
1534 {
|
|
|
1535 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1536 break;
|
|
249
|
1537 ++lnum;
|
|
346
|
1538
|
|
|
1539 /* Skip the characters at the start of the next line that were
|
|
|
1540 * included in a match crossing line boundaries. */
|
|
|
1541 if (attr == 0)
|
|
|
1542 skip = p - endp;
|
|
|
1543 else
|
|
|
1544 skip = 0;
|
|
385
|
1545
|
|
|
1546 /* Capscol skips over the inserted space. */
|
|
|
1547 --capcol;
|
|
|
1548
|
|
|
1549 /* But after empty line check first word in next line */
|
|
|
1550 if (*skipwhite(line) == NUL)
|
|
|
1551 capcol = 0;
|
|
249
|
1552 }
|
|
236
|
1553
|
|
|
1554 line_breakcheck();
|
|
|
1555 }
|
|
|
1556
|
|
346
|
1557 vim_free(buf);
|
|
|
1558 return FAIL;
|
|
|
1559 }
|
|
|
1560
|
|
|
1561 /*
|
|
|
1562 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1563 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1564 */
|
|
|
1565 void
|
|
|
1566 spell_cat_line(buf, line, maxlen)
|
|
|
1567 char_u *buf;
|
|
|
1568 char_u *line;
|
|
|
1569 int maxlen;
|
|
|
1570 {
|
|
|
1571 char_u *p;
|
|
|
1572 int n;
|
|
|
1573
|
|
|
1574 p = skipwhite(line);
|
|
|
1575 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1576 p = skipwhite(p + 1);
|
|
|
1577
|
|
|
1578 if (*p != NUL)
|
|
|
1579 {
|
|
|
1580 *buf = ' ';
|
|
376
|
1581 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1582 n = p - line;
|
|
|
1583 if (n >= maxlen)
|
|
|
1584 n = maxlen - 1;
|
|
|
1585 vim_memset(buf + 1, ' ', n);
|
|
|
1586 }
|
|
236
|
1587 }
|
|
|
1588
|
|
|
1589 /*
|
|
307
|
1590 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1591 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1592 */
|
|
307
|
1593 static void
|
|
236
|
1594 spell_load_lang(lang)
|
|
|
1595 char_u *lang;
|
|
|
1596 {
|
|
310
|
1597 char_u fname_enc[85];
|
|
236
|
1598 int r;
|
|
307
|
1599 char_u langcp[MAXWLEN + 1];
|
|
|
1600
|
|
310
|
1601 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1602 * It's truncated when an error is detected. */
|
|
|
1603 STRCPY(langcp, lang);
|
|
|
1604
|
|
310
|
1605 /*
|
|
|
1606 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1607 */
|
|
|
1608 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1609 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1610 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1611
|
|
|
1612 if (r == FAIL && *langcp != NUL)
|
|
|
1613 {
|
|
|
1614 /* Try loading the ASCII version. */
|
|
310
|
1615 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1616 "spell/%s.ascii.spl", lang);
|
|
310
|
1617 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1618 }
|
|
|
1619
|
|
|
1620 if (r == FAIL)
|
|
|
1621 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
|
236
|
1622 fname_enc + 6);
|
|
310
|
1623 else if (*langcp != NUL)
|
|
|
1624 {
|
|
|
1625 /* Load all the additions. */
|
|
|
1626 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1627 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1628 }
|
|
|
1629 }
|
|
|
1630
|
|
|
1631 /*
|
|
|
1632 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1633 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1634 */
|
|
|
1635 static char_u *
|
|
|
1636 spell_enc()
|
|
|
1637 {
|
|
|
1638
|
|
|
1639 #ifdef FEAT_MBYTE
|
|
|
1640 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1641 return p_enc;
|
|
|
1642 #endif
|
|
|
1643 return (char_u *)"latin1";
|
|
236
|
1644 }
|
|
|
1645
|
|
|
1646 /*
|
|
385
|
1647 * Get the name of the .spl file for the internal wordlist into
|
|
|
1648 * "fname[MAXPATHL]".
|
|
|
1649 */
|
|
|
1650 static void
|
|
|
1651 int_wordlist_spl(fname)
|
|
|
1652 char_u *fname;
|
|
|
1653 {
|
|
|
1654 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
1655 int_wordlist, spell_enc());
|
|
|
1656 }
|
|
|
1657
|
|
|
1658 /*
|
|
236
|
1659 * Allocate a new slang_T.
|
|
|
1660 * Caller must fill "sl_next".
|
|
|
1661 */
|
|
|
1662 static slang_T *
|
|
|
1663 slang_alloc(lang)
|
|
|
1664 char_u *lang;
|
|
|
1665 {
|
|
|
1666 slang_T *lp;
|
|
|
1667
|
|
300
|
1668 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1669 if (lp != NULL)
|
|
|
1670 {
|
|
|
1671 lp->sl_name = vim_strsave(lang);
|
|
323
|
1672 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
236
|
1673 }
|
|
|
1674 return lp;
|
|
|
1675 }
|
|
|
1676
|
|
|
1677 /*
|
|
|
1678 * Free the contents of an slang_T and the structure itself.
|
|
|
1679 */
|
|
|
1680 static void
|
|
|
1681 slang_free(lp)
|
|
|
1682 slang_T *lp;
|
|
|
1683 {
|
|
|
1684 vim_free(lp->sl_name);
|
|
310
|
1685 vim_free(lp->sl_fname);
|
|
|
1686 slang_clear(lp);
|
|
|
1687 vim_free(lp);
|
|
|
1688 }
|
|
|
1689
|
|
|
1690 /*
|
|
|
1691 * Clear an slang_T so that the file can be reloaded.
|
|
|
1692 */
|
|
|
1693 static void
|
|
|
1694 slang_clear(lp)
|
|
|
1695 slang_T *lp;
|
|
|
1696 {
|
|
339
|
1697 garray_T *gap;
|
|
|
1698 fromto_T *ftp;
|
|
344
|
1699 salitem_T *smp;
|
|
339
|
1700 int i;
|
|
323
|
1701
|
|
300
|
1702 vim_free(lp->sl_fbyts);
|
|
310
|
1703 lp->sl_fbyts = NULL;
|
|
300
|
1704 vim_free(lp->sl_kbyts);
|
|
310
|
1705 lp->sl_kbyts = NULL;
|
|
339
|
1706 vim_free(lp->sl_pbyts);
|
|
|
1707 lp->sl_pbyts = NULL;
|
|
|
1708
|
|
300
|
1709 vim_free(lp->sl_fidxs);
|
|
310
|
1710 lp->sl_fidxs = NULL;
|
|
300
|
1711 vim_free(lp->sl_kidxs);
|
|
310
|
1712 lp->sl_kidxs = NULL;
|
|
339
|
1713 vim_free(lp->sl_pidxs);
|
|
|
1714 lp->sl_pidxs = NULL;
|
|
323
|
1715
|
|
344
|
1716 gap = &lp->sl_rep;
|
|
|
1717 while (gap->ga_len > 0)
|
|
323
|
1718 {
|
|
344
|
1719 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1720 vim_free(ftp->ft_from);
|
|
|
1721 vim_free(ftp->ft_to);
|
|
323
|
1722 }
|
|
344
|
1723 ga_clear(gap);
|
|
|
1724
|
|
|
1725 gap = &lp->sl_sal;
|
|
375
|
1726 if (lp->sl_sofo)
|
|
376
|
1727 {
|
|
|
1728 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1729 if (gap->ga_data != NULL)
|
|
|
1730 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1731 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1732 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1733 }
|
|
375
|
1734 else
|
|
|
1735 /* SAL items: free salitem_T items */
|
|
|
1736 while (gap->ga_len > 0)
|
|
|
1737 {
|
|
|
1738 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1739 vim_free(smp->sm_lead);
|
|
|
1740 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1741 vim_free(smp->sm_to);
|
|
|
1742 #ifdef FEAT_MBYTE
|
|
|
1743 vim_free(smp->sm_lead_w);
|
|
|
1744 vim_free(smp->sm_oneof_w);
|
|
|
1745 vim_free(smp->sm_to_w);
|
|
|
1746 #endif
|
|
|
1747 }
|
|
344
|
1748 ga_clear(gap);
|
|
323
|
1749
|
|
339
|
1750 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1751 vim_free(lp->sl_prefprog[i]);
|
|
376
|
1752 lp->sl_prefixcnt = 0;
|
|
339
|
1753 vim_free(lp->sl_prefprog);
|
|
376
|
1754 lp->sl_prefprog = NULL;
|
|
|
1755
|
|
|
1756 vim_free(lp->sl_midword);
|
|
|
1757 lp->sl_midword = NULL;
|
|
339
|
1758
|
|
330
|
1759 #ifdef FEAT_MBYTE
|
|
|
1760 {
|
|
|
1761 int todo = lp->sl_map_hash.ht_used;
|
|
|
1762 hashitem_T *hi;
|
|
|
1763
|
|
|
1764 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1765 if (!HASHITEM_EMPTY(hi))
|
|
|
1766 {
|
|
|
1767 --todo;
|
|
|
1768 vim_free(hi->hi_key);
|
|
|
1769 }
|
|
|
1770 }
|
|
|
1771 hash_clear(&lp->sl_map_hash);
|
|
|
1772 #endif
|
|
236
|
1773 }
|
|
|
1774
|
|
|
1775 /*
|
|
307
|
1776 * Load one spell file and store the info into a slang_T.
|
|
236
|
1777 * Invoked through do_in_runtimepath().
|
|
|
1778 */
|
|
|
1779 static void
|
|
310
|
1780 spell_load_cb(fname, cookie)
|
|
236
|
1781 char_u *fname;
|
|
307
|
1782 void *cookie; /* points to the language name */
|
|
236
|
1783 {
|
|
323
|
1784 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1785 }
|
|
|
1786
|
|
|
1787 /*
|
|
|
1788 * Load one spell file and store the info into a slang_T.
|
|
|
1789 *
|
|
|
1790 * This is invoked in two ways:
|
|
|
1791 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
1792 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
1793 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
1794 * points to the existing slang_T.
|
|
323
|
1795 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
1796 */
|
|
323
|
1797 static slang_T *
|
|
|
1798 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
1799 char_u *fname;
|
|
|
1800 char_u *lang;
|
|
|
1801 slang_T *old_lp;
|
|
323
|
1802 int silent; /* no error if file doesn't exist */
|
|
310
|
1803 {
|
|
236
|
1804 FILE *fd;
|
|
|
1805 char_u buf[MAXWLEN + 1];
|
|
|
1806 char_u *p;
|
|
339
|
1807 char_u *bp;
|
|
|
1808 idx_T *ip;
|
|
236
|
1809 int i;
|
|
339
|
1810 int n;
|
|
300
|
1811 int len;
|
|
236
|
1812 int round;
|
|
|
1813 char_u *save_sourcing_name = sourcing_name;
|
|
|
1814 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
255
|
1815 int cnt, ccnt;
|
|
|
1816 char_u *fol;
|
|
307
|
1817 slang_T *lp = NULL;
|
|
323
|
1818 garray_T *gap;
|
|
|
1819 fromto_T *ftp;
|
|
344
|
1820 salitem_T *smp;
|
|
323
|
1821 short *first;
|
|
324
|
1822 idx_T idx;
|
|
344
|
1823 int c = 0;
|
|
236
|
1824
|
|
310
|
1825 fd = mch_fopen((char *)fname, "r");
|
|
236
|
1826 if (fd == NULL)
|
|
|
1827 {
|
|
323
|
1828 if (!silent)
|
|
|
1829 EMSG2(_(e_notopen), fname);
|
|
|
1830 else if (p_verbose > 2)
|
|
|
1831 {
|
|
|
1832 verbose_enter();
|
|
|
1833 smsg((char_u *)e_notopen, fname);
|
|
|
1834 verbose_leave();
|
|
|
1835 }
|
|
255
|
1836 goto endFAIL;
|
|
236
|
1837 }
|
|
310
|
1838 if (p_verbose > 2)
|
|
|
1839 {
|
|
|
1840 verbose_enter();
|
|
|
1841 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
1842 verbose_leave();
|
|
|
1843 }
|
|
|
1844
|
|
|
1845 if (old_lp == NULL)
|
|
|
1846 {
|
|
|
1847 lp = slang_alloc(lang);
|
|
|
1848 if (lp == NULL)
|
|
|
1849 goto endFAIL;
|
|
|
1850
|
|
|
1851 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
1852 lp->sl_fname = vim_strsave(fname);
|
|
|
1853 if (lp->sl_fname == NULL)
|
|
|
1854 goto endFAIL;
|
|
|
1855
|
|
|
1856 /* Check for .add.spl. */
|
|
|
1857 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
1858 }
|
|
|
1859 else
|
|
|
1860 lp = old_lp;
|
|
307
|
1861
|
|
236
|
1862 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
1863 sourcing_name = fname;
|
|
|
1864 sourcing_lnum = 0;
|
|
|
1865
|
|
339
|
1866 /* <HEADER>: <fileID>
|
|
|
1867 * <regioncnt> <regionname> ...
|
|
|
1868 * <charflagslen> <charflags>
|
|
|
1869 * <fcharslen> <fchars>
|
|
366
|
1870 * <midwordlen> <midword>
|
|
339
|
1871 * <prefcondcnt> <prefcond> ...
|
|
|
1872 */
|
|
236
|
1873 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
1874 buf[i] = getc(fd); /* <fileID> */
|
|
|
1875 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
1876 {
|
|
|
1877 EMSG(_("E757: Wrong file ID in spell file"));
|
|
255
|
1878 goto endFAIL;
|
|
236
|
1879 }
|
|
|
1880
|
|
|
1881 cnt = getc(fd); /* <regioncnt> */
|
|
255
|
1882 if (cnt < 0)
|
|
236
|
1883 {
|
|
|
1884 truncerr:
|
|
381
|
1885 EMSG(_(e_spell_trunc));
|
|
255
|
1886 goto endFAIL;
|
|
236
|
1887 }
|
|
|
1888 if (cnt > 8)
|
|
|
1889 {
|
|
|
1890 formerr:
|
|
307
|
1891 EMSG(_(e_format));
|
|
255
|
1892 goto endFAIL;
|
|
236
|
1893 }
|
|
|
1894 for (i = 0; i < cnt; ++i)
|
|
|
1895 {
|
|
|
1896 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
|
1897 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
|
1898 }
|
|
|
1899 lp->sl_regions[cnt * 2] = NUL;
|
|
|
1900
|
|
381
|
1901 /* <charflagslen> <charflags> */
|
|
|
1902 p = read_cnt_string(fd, 1, &cnt);
|
|
389
|
1903 if (cnt < 0)
|
|
381
|
1904 goto endFAIL;
|
|
|
1905
|
|
|
1906 /* <fcharslen> <fchars> */
|
|
389
|
1907 fol = read_cnt_string(fd, 2, &ccnt);
|
|
|
1908 if (ccnt < 0)
|
|
381
|
1909 {
|
|
300
|
1910 vim_free(p);
|
|
381
|
1911 goto endFAIL;
|
|
|
1912 }
|
|
|
1913
|
|
|
1914 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
1915 if (p != NULL && fol != NULL)
|
|
389
|
1916 i = set_spell_charflags(p, cnt, fol);
|
|
381
|
1917
|
|
|
1918 vim_free(p);
|
|
|
1919 vim_free(fol);
|
|
|
1920
|
|
|
1921 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
1922 if ((p == NULL) != (fol == NULL))
|
|
|
1923 goto formerr;
|
|
255
|
1924
|
|
366
|
1925 /* <midwordlen> <midword> */
|
|
376
|
1926 lp->sl_midword = read_cnt_string(fd, 2, &cnt);
|
|
389
|
1927 if (cnt < 0)
|
|
376
|
1928 goto endFAIL;
|
|
366
|
1929
|
|
339
|
1930 /* <prefcondcnt> <prefcond> ... */
|
|
|
1931 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
1932 if (cnt > 0)
|
|
|
1933 {
|
|
|
1934 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
1935 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
1936 if (lp->sl_prefprog == NULL)
|
|
|
1937 goto endFAIL;
|
|
|
1938 lp->sl_prefixcnt = cnt;
|
|
|
1939
|
|
|
1940 for (i = 0; i < cnt; ++i)
|
|
|
1941 {
|
|
|
1942 /* <prefcond> : <condlen> <condstr> */
|
|
|
1943 n = getc(fd); /* <condlen> */
|
|
|
1944 if (n < 0)
|
|
|
1945 goto formerr;
|
|
|
1946 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
1947 * compile the regexp program used to check for the condition. */
|
|
|
1948 if (n > 0)
|
|
|
1949 {
|
|
344
|
1950 buf[0] = '^'; /* always match at one position only */
|
|
|
1951 p = buf + 1;
|
|
339
|
1952 while (n-- > 0)
|
|
|
1953 *p++ = getc(fd); /* <condstr> */
|
|
|
1954 *p = NUL;
|
|
|
1955 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
1956 }
|
|
|
1957 }
|
|
|
1958 }
|
|
|
1959
|
|
|
1960
|
|
323
|
1961 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
1962 * <salflags> <salcount> <sal> ...
|
|
|
1963 * <maplen> <mapstr> */
|
|
339
|
1964
|
|
344
|
1965 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
1966 if (cnt < 0)
|
|
|
1967 goto formerr;
|
|
|
1968
|
|
|
1969 gap = &lp->sl_rep;
|
|
|
1970 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1971 goto endFAIL;
|
|
|
1972
|
|
|
1973 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
1974 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
1975 {
|
|
|
1976 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
381
|
1977 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
389
|
1978 if (i <= 0)
|
|
381
|
1979 goto endFAIL;
|
|
|
1980 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
389
|
1981 if (i <= 0)
|
|
381
|
1982 {
|
|
|
1983 vim_free(ftp->ft_from);
|
|
|
1984 goto endFAIL;
|
|
344
|
1985 }
|
|
|
1986 }
|
|
|
1987
|
|
|
1988 /* Fill the first-index table. */
|
|
|
1989 first = lp->sl_rep_first;
|
|
|
1990 for (i = 0; i < 256; ++i)
|
|
|
1991 first[i] = -1;
|
|
|
1992 for (i = 0; i < gap->ga_len; ++i)
|
|
323
|
1993 {
|
|
344
|
1994 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
1995 if (first[*ftp->ft_from] == -1)
|
|
|
1996 first[*ftp->ft_from] = i;
|
|
|
1997 }
|
|
|
1998
|
|
|
1999 i = getc(fd); /* <salflags> */
|
|
|
2000 if (i & SAL_F0LLOWUP)
|
|
|
2001 lp->sl_followup = TRUE;
|
|
|
2002 if (i & SAL_COLLAPSE)
|
|
|
2003 lp->sl_collapse = TRUE;
|
|
|
2004 if (i & SAL_REM_ACCENTS)
|
|
|
2005 lp->sl_rem_accents = TRUE;
|
|
375
|
2006 if (i & SAL_SOFO)
|
|
|
2007 lp->sl_sofo = TRUE;
|
|
389
|
2008 else
|
|
|
2009 lp->sl_sofo = FALSE;
|
|
344
|
2010
|
|
|
2011 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2012 if (cnt < 0)
|
|
|
2013 goto formerr;
|
|
|
2014
|
|
375
|
2015 if (lp->sl_sofo)
|
|
|
2016 {
|
|
|
2017 /*
|
|
|
2018 * SOFOFROM and SOFOTO items come in one <salfrom> and <salto>
|
|
|
2019 */
|
|
|
2020 if (cnt != 1)
|
|
|
2021 goto formerr;
|
|
|
2022
|
|
381
|
2023 /* <salfromlen> <salfrom> */
|
|
|
2024 bp = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2025 if (cnt < 0)
|
|
375
|
2026 goto endFAIL;
|
|
381
|
2027
|
|
|
2028 /* <saltolen> <salto> */
|
|
|
2029 fol = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2030 if (cnt < 0)
|
|
375
|
2031 {
|
|
|
2032 vim_free(bp);
|
|
344
|
2033 goto endFAIL;
|
|
375
|
2034 }
|
|
381
|
2035
|
|
|
2036 /* Store the info in lp->sl_sal and/or lp->sl_sal_first. */
|
|
389
|
2037 if (bp != NULL && fol != NULL)
|
|
|
2038 i = set_sofo(lp, bp, fol);
|
|
|
2039 else if (bp != NULL || fol != NULL)
|
|
|
2040 i = FAIL; /* only one of two strings is an error */
|
|
|
2041 else
|
|
|
2042 i = OK;
|
|
381
|
2043
|
|
375
|
2044 vim_free(bp);
|
|
|
2045 vim_free(fol);
|
|
381
|
2046 if (i == FAIL)
|
|
|
2047 goto formerr;
|
|
375
|
2048 }
|
|
|
2049 else
|
|
|
2050 {
|
|
|
2051 /*
|
|
|
2052 * SAL items
|
|
|
2053 */
|
|
|
2054 gap = &lp->sl_sal;
|
|
390
|
2055 ga_init2(gap, sizeof(salitem_T), 10);
|
|
375
|
2056 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2057 goto endFAIL;
|
|
|
2058
|
|
|
2059 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2060 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2061 {
|
|
|
2062 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2063 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2064 if (ccnt < 0)
|
|
|
2065 goto formerr;
|
|
|
2066 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2067 goto endFAIL;
|
|
|
2068 smp->sm_lead = p;
|
|
|
2069
|
|
|
2070 /* Read up to the first special char into sm_lead. */
|
|
|
2071 for (i = 0; i < ccnt; ++i)
|
|
|
2072 {
|
|
|
2073 c = getc(fd); /* <salfrom> */
|
|
|
2074 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2075 break;
|
|
|
2076 *p++ = c;
|
|
|
2077 }
|
|
|
2078 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2079 *p++ = NUL;
|
|
|
2080
|
|
|
2081 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2082 if (c == '(')
|
|
|
2083 {
|
|
|
2084 smp->sm_oneof = p;
|
|
|
2085 for (++i; i < ccnt; ++i)
|
|
|
2086 {
|
|
|
2087 c = getc(fd); /* <salfrom> */
|
|
|
2088 if (c == ')')
|
|
|
2089 break;
|
|
|
2090 *p++ = c;
|
|
|
2091 }
|
|
|
2092 *p++ = NUL;
|
|
|
2093 if (++i < ccnt)
|
|
|
2094 c = getc(fd);
|
|
|
2095 }
|
|
372
|
2096 else
|
|
375
|
2097 smp->sm_oneof = NULL;
|
|
|
2098
|
|
|
2099 /* Any following chars go in sm_rules. */
|
|
|
2100 smp->sm_rules = p;
|
|
|
2101 if (i < ccnt)
|
|
|
2102 /* store the char we got while checking for end of sm_lead */
|
|
|
2103 *p++ = c;
|
|
|
2104 for (++i; i < ccnt; ++i)
|
|
|
2105 *p++ = getc(fd); /* <salfrom> */
|
|
|
2106 *p++ = NUL;
|
|
|
2107
|
|
381
|
2108 /* <saltolen> <salto> */
|
|
|
2109 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
389
|
2110 if (ccnt < 0)
|
|
372
|
2111 {
|
|
|
2112 vim_free(smp->sm_lead);
|
|
375
|
2113 goto formerr;
|
|
|
2114 }
|
|
|
2115
|
|
372
|
2116 #ifdef FEAT_MBYTE
|
|
|
2117 if (has_mbyte)
|
|
|
2118 {
|
|
375
|
2119 /* convert the multi-byte strings to wide char strings */
|
|
|
2120 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2121 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2122 if (smp->sm_oneof == NULL)
|
|
|
2123 smp->sm_oneof_w = NULL;
|
|
|
2124 else
|
|
|
2125 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
389
|
2126 if (smp->sm_to == NULL)
|
|
|
2127 smp->sm_to_w = NULL;
|
|
|
2128 else
|
|
|
2129 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
375
|
2130 if (smp->sm_lead_w == NULL
|
|
|
2131 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
389
|
2132 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
372
|
2133 {
|
|
375
|
2134 vim_free(smp->sm_lead);
|
|
|
2135 vim_free(smp->sm_to);
|
|
|
2136 vim_free(smp->sm_lead_w);
|
|
|
2137 vim_free(smp->sm_oneof_w);
|
|
|
2138 vim_free(smp->sm_to_w);
|
|
|
2139 goto endFAIL;
|
|
372
|
2140 }
|
|
|
2141 }
|
|
|
2142 #endif
|
|
|
2143 }
|
|
375
|
2144
|
|
|
2145 /* Fill the first-index table. */
|
|
381
|
2146 set_sal_first(lp);
|
|
|
2147 }
|
|
|
2148
|
|
|
2149 /* <maplen> <mapstr> */
|
|
|
2150 p = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2151 if (cnt < 0)
|
|
323
|
2152 goto endFAIL;
|
|
389
|
2153 if (p != NULL)
|
|
|
2154 {
|
|
|
2155 set_map_str(lp, p);
|
|
|
2156 vim_free(p);
|
|
|
2157 }
|
|
236
|
2158
|
|
300
|
2159 /* round 1: <LWORDTREE>
|
|
339
|
2160 * round 2: <KWORDTREE>
|
|
|
2161 * round 3: <PREFIXTREE> */
|
|
|
2162 for (round = 1; round <= 3; ++round)
|
|
236
|
2163 {
|
|
300
|
2164 /* The tree size was computed when writing the file, so that we can
|
|
|
2165 * allocate it as one long block. <nodecount> */
|
|
|
2166 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2167 if (len < 0)
|
|
|
2168 goto truncerr;
|
|
|
2169 if (len > 0)
|
|
236
|
2170 {
|
|
300
|
2171 /* Allocate the byte array. */
|
|
339
|
2172 bp = lalloc((long_u)len, TRUE);
|
|
|
2173 if (bp == NULL)
|
|
300
|
2174 goto endFAIL;
|
|
|
2175 if (round == 1)
|
|
339
|
2176 lp->sl_fbyts = bp;
|
|
|
2177 else if (round == 2)
|
|
|
2178 lp->sl_kbyts = bp;
|
|
300
|
2179 else
|
|
339
|
2180 lp->sl_pbyts = bp;
|
|
236
|
2181
|
|
300
|
2182 /* Allocate the index array. */
|
|
339
|
2183 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2184 if (ip == NULL)
|
|
300
|
2185 goto endFAIL;
|
|
|
2186 if (round == 1)
|
|
339
|
2187 lp->sl_fidxs = ip;
|
|
|
2188 else if (round == 2)
|
|
|
2189 lp->sl_kidxs = ip;
|
|
300
|
2190 else
|
|
339
|
2191 lp->sl_pidxs = ip;
|
|
300
|
2192
|
|
|
2193 /* Read the tree and store it in the array. */
|
|
339
|
2194 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2195 if (idx == -1)
|
|
300
|
2196 goto truncerr;
|
|
324
|
2197 if (idx < 0)
|
|
236
|
2198 goto formerr;
|
|
|
2199 }
|
|
300
|
2200 }
|
|
243
|
2201
|
|
310
|
2202 /* For a new file link it in the list of spell files. */
|
|
|
2203 if (old_lp == NULL)
|
|
|
2204 {
|
|
|
2205 lp->sl_next = first_lang;
|
|
|
2206 first_lang = lp;
|
|
|
2207 }
|
|
307
|
2208
|
|
255
|
2209 goto endOK;
|
|
|
2210
|
|
|
2211 endFAIL:
|
|
310
|
2212 if (lang != NULL)
|
|
|
2213 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2214 *lang = NUL;
|
|
|
2215 if (lp != NULL && old_lp == NULL)
|
|
323
|
2216 {
|
|
307
|
2217 slang_free(lp);
|
|
323
|
2218 lp = NULL;
|
|
|
2219 }
|
|
255
|
2220
|
|
|
2221 endOK:
|
|
236
|
2222 if (fd != NULL)
|
|
|
2223 fclose(fd);
|
|
|
2224 sourcing_name = save_sourcing_name;
|
|
|
2225 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2226
|
|
|
2227 return lp;
|
|
236
|
2228 }
|
|
|
2229
|
|
376
|
2230 /*
|
|
|
2231 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2232 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2233 * Returns NULL when the count is zero.
|
|
389
|
2234 * Sets "*cntp" to -1 when there is an error, length of the result otherwise.
|
|
376
|
2235 */
|
|
|
2236 static char_u *
|
|
389
|
2237 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2238 FILE *fd;
|
|
|
2239 int cnt_bytes;
|
|
389
|
2240 int *cntp;
|
|
376
|
2241 {
|
|
|
2242 int cnt = 0;
|
|
|
2243 int i;
|
|
|
2244 char_u *str;
|
|
|
2245
|
|
|
2246 /* read the length bytes, MSB first */
|
|
|
2247 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2248 cnt = (cnt << 8) + getc(fd);
|
|
|
2249 if (cnt < 0)
|
|
|
2250 {
|
|
381
|
2251 EMSG(_(e_spell_trunc));
|
|
389
|
2252 *cntp = -1;
|
|
376
|
2253 return NULL;
|
|
|
2254 }
|
|
389
|
2255 *cntp = cnt;
|
|
|
2256 if (cnt == 0)
|
|
|
2257 return NULL; /* nothing to read, return NULL */
|
|
376
|
2258
|
|
|
2259 /* allocate memory */
|
|
|
2260 str = alloc((unsigned)cnt + 1);
|
|
|
2261 if (str == NULL)
|
|
|
2262 {
|
|
389
|
2263 *cntp = -1;
|
|
376
|
2264 return NULL;
|
|
|
2265 }
|
|
|
2266
|
|
|
2267 /* Read the string. Doesn't check for truncated file. */
|
|
|
2268 for (i = 0; i < cnt; ++i)
|
|
|
2269 str[i] = getc(fd);
|
|
|
2270 str[i] = NUL;
|
|
|
2271
|
|
|
2272 return str;
|
|
|
2273 }
|
|
|
2274
|
|
381
|
2275 /*
|
|
|
2276 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
|
2277 * Returns FAIL when there is something wrong.
|
|
|
2278 */
|
|
|
2279 static int
|
|
|
2280 set_sofo(lp, from, to)
|
|
|
2281 slang_T *lp;
|
|
|
2282 char_u *from;
|
|
|
2283 char_u *to;
|
|
|
2284 {
|
|
|
2285 int i;
|
|
|
2286
|
|
|
2287 #ifdef FEAT_MBYTE
|
|
|
2288 garray_T *gap;
|
|
|
2289 char_u *s;
|
|
|
2290 char_u *p;
|
|
|
2291 int c;
|
|
|
2292 int *inp;
|
|
|
2293
|
|
|
2294 if (has_mbyte)
|
|
|
2295 {
|
|
|
2296 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2297 * characters. The index is the low byte of the character.
|
|
|
2298 * The list contains from-to pairs with a terminating NUL.
|
|
|
2299 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2300 gap = &lp->sl_sal;
|
|
|
2301 ga_init2(gap, sizeof(int *), 1);
|
|
|
2302 if (ga_grow(gap, 256) == FAIL)
|
|
|
2303 return FAIL;
|
|
|
2304 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2305 gap->ga_len = 256;
|
|
|
2306
|
|
|
2307 /* First count the number of items for each list. Temporarily use
|
|
|
2308 * sl_sal_first[] for this. */
|
|
|
2309 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2310 {
|
|
474
|
2311 c = mb_cptr2char_adv(&p);
|
|
|
2312 mb_cptr_adv(s);
|
|
381
|
2313 if (c >= 256)
|
|
|
2314 ++lp->sl_sal_first[c & 0xff];
|
|
|
2315 }
|
|
|
2316 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
|
2317 return FAIL;
|
|
|
2318
|
|
|
2319 /* Allocate the lists. */
|
|
|
2320 for (i = 0; i < 256; ++i)
|
|
|
2321 if (lp->sl_sal_first[i] > 0)
|
|
|
2322 {
|
|
|
2323 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2324 if (p == NULL)
|
|
|
2325 return FAIL;
|
|
|
2326 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2327 *(int *)p = 0;
|
|
|
2328 }
|
|
|
2329
|
|
|
2330 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
2331 * list. */
|
|
|
2332 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2333 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2334 {
|
|
474
|
2335 c = mb_cptr2char_adv(&p);
|
|
|
2336 i = mb_cptr2char_adv(&s);
|
|
381
|
2337 if (c >= 256)
|
|
|
2338 {
|
|
|
2339 /* Append the from-to chars at the end of the list with
|
|
|
2340 * the low byte. */
|
|
|
2341 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2342 while (*inp != 0)
|
|
|
2343 ++inp;
|
|
|
2344 *inp++ = c; /* from char */
|
|
|
2345 *inp++ = i; /* to char */
|
|
|
2346 *inp++ = NUL; /* NUL at the end */
|
|
|
2347 }
|
|
|
2348 else
|
|
|
2349 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2350 lp->sl_sal_first[c] = i;
|
|
|
2351 }
|
|
|
2352 }
|
|
|
2353 else
|
|
|
2354 #endif
|
|
|
2355 {
|
|
|
2356 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2357 if (STRLEN(from) != STRLEN(to))
|
|
|
2358 return FAIL;
|
|
|
2359
|
|
|
2360 for (i = 0; to[i] != NUL; ++i)
|
|
|
2361 lp->sl_sal_first[from[i]] = to[i];
|
|
|
2362 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2363 }
|
|
|
2364
|
|
|
2365 return OK;
|
|
|
2366 }
|
|
|
2367
|
|
|
2368 /*
|
|
|
2369 * Fill the first-index table for "lp".
|
|
|
2370 */
|
|
|
2371 static void
|
|
|
2372 set_sal_first(lp)
|
|
|
2373 slang_T *lp;
|
|
|
2374 {
|
|
|
2375 salfirst_T *sfirst;
|
|
|
2376 int i;
|
|
|
2377 salitem_T *smp;
|
|
|
2378 int c;
|
|
|
2379 garray_T *gap = &lp->sl_sal;
|
|
|
2380
|
|
|
2381 sfirst = lp->sl_sal_first;
|
|
|
2382 for (i = 0; i < 256; ++i)
|
|
|
2383 sfirst[i] = -1;
|
|
|
2384 smp = (salitem_T *)gap->ga_data;
|
|
|
2385 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2386 {
|
|
|
2387 #ifdef FEAT_MBYTE
|
|
|
2388 if (has_mbyte)
|
|
|
2389 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2390 * the character, for other encodings it should differ for most
|
|
|
2391 * characters. */
|
|
|
2392 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2393 else
|
|
|
2394 #endif
|
|
|
2395 c = *smp[i].sm_lead;
|
|
|
2396 if (sfirst[c] == -1)
|
|
|
2397 {
|
|
|
2398 sfirst[c] = i;
|
|
|
2399 #ifdef FEAT_MBYTE
|
|
|
2400 if (has_mbyte)
|
|
|
2401 {
|
|
|
2402 int n;
|
|
|
2403
|
|
|
2404 /* Make sure all entries with this byte are following each
|
|
|
2405 * other. Move the ones that are in the wrong position. Do
|
|
|
2406 * keep the same ordering! */
|
|
|
2407 while (i + 1 < gap->ga_len
|
|
|
2408 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
2409 /* Skip over entry with same index byte. */
|
|
|
2410 ++i;
|
|
|
2411
|
|
|
2412 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
2413 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
2414 {
|
|
|
2415 salitem_T tsal;
|
|
|
2416
|
|
|
2417 /* Move entry with same index byte after the entries
|
|
|
2418 * we already found. */
|
|
|
2419 ++i;
|
|
|
2420 --n;
|
|
|
2421 tsal = smp[i + n];
|
|
|
2422 mch_memmove(smp + i + 1, smp + i,
|
|
|
2423 sizeof(salitem_T) * n);
|
|
|
2424 smp[i] = tsal;
|
|
|
2425 }
|
|
|
2426 }
|
|
|
2427 #endif
|
|
|
2428 }
|
|
|
2429 }
|
|
|
2430 }
|
|
376
|
2431
|
|
372
|
2432 #ifdef FEAT_MBYTE
|
|
|
2433 /*
|
|
|
2434 * Turn a multi-byte string into a wide character string.
|
|
|
2435 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
2436 */
|
|
|
2437 static int *
|
|
|
2438 mb_str2wide(s)
|
|
|
2439 char_u *s;
|
|
|
2440 {
|
|
|
2441 int *res;
|
|
|
2442 char_u *p;
|
|
|
2443 int i = 0;
|
|
|
2444
|
|
|
2445 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
2446 if (res != NULL)
|
|
|
2447 {
|
|
|
2448 for (p = s; *p != NUL; )
|
|
|
2449 res[i++] = mb_ptr2char_adv(&p);
|
|
|
2450 res[i] = NUL;
|
|
|
2451 }
|
|
|
2452 return res;
|
|
|
2453 }
|
|
|
2454 #endif
|
|
|
2455
|
|
236
|
2456 /*
|
|
300
|
2457 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
2458 * "byts" and index array "idxs". Recursively read the children.
|
|
|
2459 *
|
|
346
|
2460 * NOTE: The code here must match put_node().
|
|
300
|
2461 *
|
|
|
2462 * Returns the index follosing the siblings.
|
|
|
2463 * Returns -1 if the file is shorter than expected.
|
|
|
2464 * Returns -2 if there is a format error.
|
|
236
|
2465 */
|
|
324
|
2466 static idx_T
|
|
339
|
2467 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
2468 FILE *fd;
|
|
|
2469 char_u *byts;
|
|
324
|
2470 idx_T *idxs;
|
|
300
|
2471 int maxidx; /* size of arrays */
|
|
324
|
2472 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
2473 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
2474 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
2475 {
|
|
300
|
2476 int len;
|
|
|
2477 int i;
|
|
|
2478 int n;
|
|
324
|
2479 idx_T idx = startidx;
|
|
300
|
2480 int c;
|
|
366
|
2481 int c2;
|
|
300
|
2482 #define SHARED_MASK 0x8000000
|
|
236
|
2483
|
|
300
|
2484 len = getc(fd); /* <siblingcount> */
|
|
|
2485 if (len <= 0)
|
|
|
2486 return -1;
|
|
|
2487
|
|
|
2488 if (startidx + len >= maxidx)
|
|
|
2489 return -2;
|
|
|
2490 byts[idx++] = len;
|
|
|
2491
|
|
|
2492 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
2493 for (i = 1; i <= len; ++i)
|
|
236
|
2494 {
|
|
300
|
2495 c = getc(fd); /* <byte> */
|
|
|
2496 if (c < 0)
|
|
|
2497 return -1;
|
|
|
2498 if (c <= BY_SPECIAL)
|
|
|
2499 {
|
|
366
|
2500 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
2501 {
|
|
|
2502 /* No flags, all regions. */
|
|
|
2503 idxs[idx] = 0;
|
|
|
2504 c = 0;
|
|
|
2505 }
|
|
390
|
2506 else if (c != BY_INDEX)
|
|
300
|
2507 {
|
|
339
|
2508 if (prefixtree)
|
|
|
2509 {
|
|
455
|
2510 /* Read the optional pflags byte, the prefix ID and the
|
|
|
2511 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
2512 * byte, the condition index shifted up 8 bits, the flags
|
|
|
2513 * shifted up 24 bits. */
|
|
|
2514 if (c == BY_FLAGS)
|
|
|
2515 c = getc(fd) << 24; /* <pflags> */
|
|
|
2516 else
|
|
|
2517 c = 0;
|
|
|
2518
|
|
|
2519 c |= getc(fd); /* <prefixID> */
|
|
|
2520
|
|
339
|
2521 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
2522 if (n >= maxprefcondnr)
|
|
|
2523 return -2;
|
|
455
|
2524 c |= (n << 8);
|
|
339
|
2525 }
|
|
390
|
2526 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
2527 {
|
|
|
2528 /* Read flags and optional region and prefix ID. In
|
|
390
|
2529 * idxs[] the flags go in the low two bytes, region above
|
|
|
2530 * that and prefix ID above the region. */
|
|
|
2531 c2 = c;
|
|
339
|
2532 c = getc(fd); /* <flags> */
|
|
390
|
2533 if (c2 == BY_FLAGS2)
|
|
|
2534 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
2535 if (c & WF_REGION)
|
|
390
|
2536 c = (getc(fd) << 16) + c; /* <region> */
|
|
339
|
2537 if (c & WF_PFX)
|
|
390
|
2538 c = (getc(fd) << 24) + c; /* <prefixID> */
|
|
339
|
2539 }
|
|
|
2540
|
|
300
|
2541 idxs[idx] = c;
|
|
|
2542 c = 0;
|
|
|
2543 }
|
|
|
2544 else /* c == BY_INDEX */
|
|
|
2545 {
|
|
|
2546 /* <nodeidx> */
|
|
|
2547 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2548 if (n < 0 || n >= maxidx)
|
|
|
2549 return -2;
|
|
|
2550 idxs[idx] = n + SHARED_MASK;
|
|
|
2551 c = getc(fd); /* <xbyte> */
|
|
|
2552 }
|
|
|
2553 }
|
|
|
2554 byts[idx++] = c;
|
|
236
|
2555 }
|
|
|
2556
|
|
300
|
2557 /* Recursively read the children for non-shared siblings.
|
|
|
2558 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
2559 * remove SHARED_MASK) */
|
|
|
2560 for (i = 1; i <= len; ++i)
|
|
|
2561 if (byts[startidx + i] != 0)
|
|
|
2562 {
|
|
|
2563 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
2564 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
2565 else
|
|
|
2566 {
|
|
|
2567 idxs[startidx + i] = idx;
|
|
339
|
2568 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
2569 prefixtree, maxprefcondnr);
|
|
300
|
2570 if (idx < 0)
|
|
|
2571 break;
|
|
|
2572 }
|
|
|
2573 }
|
|
236
|
2574
|
|
300
|
2575 return idx;
|
|
236
|
2576 }
|
|
|
2577
|
|
|
2578 /*
|
|
|
2579 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
2580 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
2581 */
|
|
|
2582 char_u *
|
|
|
2583 did_set_spelllang(buf)
|
|
|
2584 buf_T *buf;
|
|
|
2585 {
|
|
|
2586 garray_T ga;
|
|
351
|
2587 char_u *splp;
|
|
236
|
2588 char_u *region;
|
|
416
|
2589 char_u region_cp[3];
|
|
355
|
2590 int filename;
|
|
236
|
2591 int region_mask;
|
|
|
2592 slang_T *lp;
|
|
|
2593 int c;
|
|
351
|
2594 char_u lang[MAXWLEN + 1];
|
|
323
|
2595 char_u spf_name[MAXPATHL];
|
|
351
|
2596 int len;
|
|
|
2597 char_u *p;
|
|
381
|
2598 int round;
|
|
385
|
2599 char_u *spf;
|
|
389
|
2600 char_u *use_region = NULL;
|
|
|
2601 int dont_use_region = FALSE;
|
|
236
|
2602
|
|
|
2603 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
2604 clear_midword(buf);
|
|
236
|
2605
|
|
351
|
2606 /* loop over comma separated language names. */
|
|
|
2607 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
2608 {
|
|
|
2609 /* Get one language name. */
|
|
|
2610 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
2611
|
|
240
|
2612 region = NULL;
|
|
351
|
2613 len = STRLEN(lang);
|
|
355
|
2614
|
|
|
2615 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
2616 * If there is a region name let "region" point to it and remove it
|
|
|
2617 * from the name. */
|
|
|
2618 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
2619 {
|
|
355
|
2620 filename = TRUE;
|
|
|
2621
|
|
416
|
2622 /* Locate a region and remove it from the file name. */
|
|
|
2623 p = vim_strchr(gettail(lang), '_');
|
|
|
2624 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
2625 && !ASCII_ISALPHA(p[3]))
|
|
|
2626 {
|
|
|
2627 vim_strncpy(region_cp, p + 1, 2);
|
|
|
2628 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
2629 len -= 3;
|
|
|
2630 region = region_cp;
|
|
|
2631 }
|
|
|
2632 else
|
|
|
2633 dont_use_region = TRUE;
|
|
|
2634
|
|
355
|
2635 /* Check if we loaded this language before. */
|
|
|
2636 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2637 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2638 break;
|
|
236
|
2639 }
|
|
355
|
2640 else
|
|
|
2641 {
|
|
|
2642 filename = FALSE;
|
|
|
2643 if (len > 3 && lang[len - 3] == '_')
|
|
|
2644 {
|
|
|
2645 region = lang + len - 2;
|
|
|
2646 len -= 3;
|
|
|
2647 lang[len] = NUL;
|
|
389
|
2648 }
|
|
|
2649 else
|
|
|
2650 dont_use_region = TRUE;
|
|
355
|
2651
|
|
|
2652 /* Check if we loaded this language before. */
|
|
|
2653 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2654 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
2655 break;
|
|
|
2656 }
|
|
236
|
2657
|
|
416
|
2658 if (region != NULL)
|
|
|
2659 {
|
|
|
2660 /* If the region differs from what was used before then don't
|
|
|
2661 * use it for 'spellfile'. */
|
|
|
2662 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
2663 dont_use_region = TRUE;
|
|
|
2664 use_region = region;
|
|
|
2665 }
|
|
|
2666
|
|
351
|
2667 /* If not found try loading the language now. */
|
|
236
|
2668 if (lp == NULL)
|
|
355
|
2669 {
|
|
|
2670 if (filename)
|
|
|
2671 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
2672 else
|
|
|
2673 spell_load_lang(lang);
|
|
|
2674 }
|
|
236
|
2675
|
|
307
|
2676 /*
|
|
351
|
2677 * Loop over the languages, there can be several files for "lang".
|
|
307
|
2678 */
|
|
|
2679 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
2680 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
2681 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
2682 {
|
|
316
|
2683 region_mask = REGION_ALL;
|
|
355
|
2684 if (!filename && region != NULL)
|
|
236
|
2685 {
|
|
307
|
2686 /* find region in sl_regions */
|
|
|
2687 c = find_region(lp->sl_regions, region);
|
|
|
2688 if (c == REGION_ALL)
|
|
|
2689 {
|
|
389
|
2690 if (lp->sl_add)
|
|
|
2691 {
|
|
|
2692 if (*lp->sl_regions != NUL)
|
|
|
2693 /* This addition file is for other regions. */
|
|
|
2694 region_mask = 0;
|
|
|
2695 }
|
|
|
2696 else
|
|
|
2697 /* This is probably an error. Give a warning and
|
|
|
2698 * accept the words anyway. */
|
|
351
|
2699 smsg((char_u *)
|
|
|
2700 _("Warning: region %s not supported"),
|
|
|
2701 region);
|
|
307
|
2702 }
|
|
|
2703 else
|
|
|
2704 region_mask = 1 << c;
|
|
236
|
2705 }
|
|
307
|
2706
|
|
389
|
2707 if (region_mask != 0)
|
|
307
|
2708 {
|
|
389
|
2709 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2710 {
|
|
|
2711 ga_clear(&ga);
|
|
|
2712 return e_outofmem;
|
|
|
2713 }
|
|
|
2714 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2715 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2716 ++ga.ga_len;
|
|
|
2717 use_midword(lp, buf);
|
|
307
|
2718 }
|
|
385
|
2719 }
|
|
|
2720 }
|
|
|
2721
|
|
|
2722 /* round 0: load int_wordlist, if possible.
|
|
|
2723 * round 1: load first name in 'spellfile'.
|
|
|
2724 * round 2: load second name in 'spellfile.
|
|
|
2725 * etc. */
|
|
|
2726 spf = curbuf->b_p_spf;
|
|
|
2727 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
2728 {
|
|
|
2729 if (round == 0)
|
|
|
2730 {
|
|
|
2731 /* Internal wordlist, if there is one. */
|
|
|
2732 if (int_wordlist == NULL)
|
|
381
|
2733 continue;
|
|
385
|
2734 int_wordlist_spl(spf_name);
|
|
381
|
2735 }
|
|
|
2736 else
|
|
|
2737 {
|
|
385
|
2738 /* One entry in 'spellfile'. */
|
|
|
2739 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
2740 STRCAT(spf_name, ".spl");
|
|
|
2741
|
|
|
2742 /* If it was already found above then skip it. */
|
|
|
2743 for (c = 0; c < ga.ga_len; ++c)
|
|
|
2744 if (fullpathcmp(spf_name,
|
|
|
2745 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
2746 FALSE) == FPC_SAME)
|
|
|
2747 break;
|
|
|
2748 if (c < ga.ga_len)
|
|
381
|
2749 continue;
|
|
|
2750 }
|
|
|
2751
|
|
351
|
2752 /* Check if it was loaded already. */
|
|
323
|
2753 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2754 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2755 break;
|
|
|
2756 if (lp == NULL)
|
|
|
2757 {
|
|
351
|
2758 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
2759 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
2760 * use an arbitrary name. */
|
|
|
2761 if (round == 0)
|
|
|
2762 STRCPY(lang, "internal wordlist");
|
|
|
2763 else
|
|
|
2764 {
|
|
|
2765 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
2766 p = vim_strchr(lang, '.');
|
|
|
2767 if (p != NULL)
|
|
|
2768 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
2769 }
|
|
351
|
2770 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
2771 }
|
|
|
2772 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
2773 {
|
|
389
|
2774 region_mask = REGION_ALL;
|
|
|
2775 if (use_region != NULL && !dont_use_region)
|
|
|
2776 {
|
|
|
2777 /* find region in sl_regions */
|
|
|
2778 c = find_region(lp->sl_regions, use_region);
|
|
|
2779 if (c != REGION_ALL)
|
|
|
2780 region_mask = 1 << c;
|
|
|
2781 else if (*lp->sl_regions != NUL)
|
|
|
2782 /* This spell file is for other regions. */
|
|
|
2783 region_mask = 0;
|
|
|
2784 }
|
|
|
2785
|
|
|
2786 if (region_mask != 0)
|
|
|
2787 {
|
|
|
2788 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2789 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2790 ++ga.ga_len;
|
|
|
2791 use_midword(lp, buf);
|
|
|
2792 }
|
|
323
|
2793 }
|
|
|
2794 }
|
|
|
2795
|
|
236
|
2796 /* Add a NULL entry to mark the end of the list. */
|
|
|
2797 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2798 {
|
|
|
2799 ga_clear(&ga);
|
|
|
2800 return e_outofmem;
|
|
|
2801 }
|
|
|
2802 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
2803 ++ga.ga_len;
|
|
|
2804
|
|
|
2805 /* Everything is fine, store the new b_langp value. */
|
|
|
2806 ga_clear(&buf->b_langp);
|
|
|
2807 buf->b_langp = ga;
|
|
|
2808
|
|
|
2809 return NULL;
|
|
|
2810 }
|
|
|
2811
|
|
|
2812 /*
|
|
376
|
2813 * Clear the midword characters for buffer "buf".
|
|
|
2814 */
|
|
|
2815 static void
|
|
|
2816 clear_midword(buf)
|
|
|
2817 buf_T *buf;
|
|
|
2818 {
|
|
|
2819 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
2820 #ifdef FEAT_MBYTE
|
|
|
2821 vim_free(buf->b_spell_ismw_mb);
|
|
|
2822 buf->b_spell_ismw_mb = NULL;
|
|
|
2823 #endif
|
|
|
2824 }
|
|
|
2825
|
|
|
2826 /*
|
|
|
2827 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
2828 * They add up to any currently used midword characters.
|
|
|
2829 */
|
|
|
2830 static void
|
|
|
2831 use_midword(lp, buf)
|
|
|
2832 slang_T *lp;
|
|
|
2833 buf_T *buf;
|
|
|
2834 {
|
|
|
2835 char_u *p;
|
|
|
2836
|
|
389
|
2837 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
2838 return;
|
|
|
2839
|
|
376
|
2840 for (p = lp->sl_midword; *p != NUL; )
|
|
|
2841 #ifdef FEAT_MBYTE
|
|
|
2842 if (has_mbyte)
|
|
|
2843 {
|
|
|
2844 int c, l, n;
|
|
|
2845 char_u *bp;
|
|
|
2846
|
|
|
2847 c = mb_ptr2char(p);
|
|
474
|
2848 l = (*mb_ptr2len)(p);
|
|
|
2849 if (c < 256 && l <= 2)
|
|
376
|
2850 buf->b_spell_ismw[c] = TRUE;
|
|
|
2851 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
2852 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
2853 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
2854 else
|
|
|
2855 {
|
|
|
2856 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
2857 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
2858 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
2859 if (bp != NULL)
|
|
|
2860 {
|
|
|
2861 vim_free(buf->b_spell_ismw_mb);
|
|
|
2862 buf->b_spell_ismw_mb = bp;
|
|
|
2863 vim_strncpy(bp + n, p, l);
|
|
|
2864 }
|
|
|
2865 }
|
|
|
2866 p += l;
|
|
|
2867 }
|
|
|
2868 else
|
|
|
2869 #endif
|
|
|
2870 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
2871 }
|
|
|
2872
|
|
|
2873 /*
|
|
236
|
2874 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
2875 * Each region is simply stored as the two characters of it's name.
|
|
381
|
2876 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
2877 */
|
|
|
2878 static int
|
|
|
2879 find_region(rp, region)
|
|
|
2880 char_u *rp;
|
|
|
2881 char_u *region;
|
|
|
2882 {
|
|
|
2883 int i;
|
|
|
2884
|
|
|
2885 for (i = 0; ; i += 2)
|
|
|
2886 {
|
|
|
2887 if (rp[i] == NUL)
|
|
|
2888 return REGION_ALL;
|
|
|
2889 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
2890 break;
|
|
|
2891 }
|
|
|
2892 return i / 2;
|
|
|
2893 }
|
|
|
2894
|
|
|
2895 /*
|
|
323
|
2896 * Return case type of word:
|
|
236
|
2897 * w word 0
|
|
300
|
2898 * Word WF_ONECAP
|
|
|
2899 * W WORD WF_ALLCAP
|
|
|
2900 * WoRd wOrd WF_KEEPCAP
|
|
236
|
2901 */
|
|
|
2902 static int
|
|
|
2903 captype(word, end)
|
|
|
2904 char_u *word;
|
|
323
|
2905 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
2906 {
|
|
|
2907 char_u *p;
|
|
|
2908 int c;
|
|
|
2909 int firstcap;
|
|
|
2910 int allcap;
|
|
|
2911 int past_second = FALSE; /* past second word char */
|
|
|
2912
|
|
|
2913 /* find first letter */
|
|
376
|
2914 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
2915 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
2916 return 0; /* only non-word characters, illegal word */
|
|
|
2917 #ifdef FEAT_MBYTE
|
|
310
|
2918 if (has_mbyte)
|
|
|
2919 c = mb_ptr2char_adv(&p);
|
|
|
2920 else
|
|
236
|
2921 #endif
|
|
310
|
2922 c = *p++;
|
|
324
|
2923 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
2924
|
|
|
2925 /*
|
|
|
2926 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
2927 * But a word with an upper char only at start is a ONECAP.
|
|
|
2928 */
|
|
323
|
2929 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
2930 if (spell_iswordp_nmw(p))
|
|
236
|
2931 {
|
|
455
|
2932 c = PTR2CHAR(p);
|
|
324
|
2933 if (!SPELL_ISUPPER(c))
|
|
236
|
2934 {
|
|
|
2935 /* UUl -> KEEPCAP */
|
|
|
2936 if (past_second && allcap)
|
|
300
|
2937 return WF_KEEPCAP;
|
|
236
|
2938 allcap = FALSE;
|
|
|
2939 }
|
|
|
2940 else if (!allcap)
|
|
|
2941 /* UlU -> KEEPCAP */
|
|
300
|
2942 return WF_KEEPCAP;
|
|
236
|
2943 past_second = TRUE;
|
|
|
2944 }
|
|
|
2945
|
|
|
2946 if (allcap)
|
|
300
|
2947 return WF_ALLCAP;
|
|
236
|
2948 if (firstcap)
|
|
300
|
2949 return WF_ONECAP;
|
|
236
|
2950 return 0;
|
|
|
2951 }
|
|
|
2952
|
|
474
|
2953 /*
|
|
|
2954 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
2955 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
2956 * Add ALLCAP for "WOrD".
|
|
|
2957 */
|
|
|
2958 static int
|
|
|
2959 badword_captype(word, end)
|
|
|
2960 char_u *word;
|
|
|
2961 char_u *end;
|
|
|
2962 {
|
|
|
2963 int flags = captype(word, end);
|
|
475
|
2964 int c;
|
|
474
|
2965 int l, u;
|
|
|
2966 int first;
|
|
|
2967 char_u *p;
|
|
|
2968
|
|
|
2969 if (flags & WF_KEEPCAP)
|
|
|
2970 {
|
|
|
2971 /* Count the number of UPPER and lower case letters. */
|
|
|
2972 l = u = 0;
|
|
|
2973 first = FALSE;
|
|
|
2974 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
2975 {
|
|
475
|
2976 c = PTR2CHAR(p);
|
|
|
2977 if (SPELL_ISUPPER(c))
|
|
474
|
2978 {
|
|
|
2979 ++u;
|
|
|
2980 if (p == word)
|
|
|
2981 first = TRUE;
|
|
|
2982 }
|
|
|
2983 else
|
|
|
2984 ++l;
|
|
|
2985 }
|
|
|
2986
|
|
|
2987 /* If there are more UPPER than lower case letters suggest an
|
|
|
2988 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
2989 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
2990 * require three upper case letters. */
|
|
|
2991 if (u > l && u > 2)
|
|
|
2992 flags |= WF_ALLCAP;
|
|
|
2993 else if (first)
|
|
|
2994 flags |= WF_ONECAP;
|
|
|
2995 }
|
|
|
2996 return flags;
|
|
|
2997 }
|
|
|
2998
|
|
355
|
2999 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
3000 /*
|
|
|
3001 * Free all languages.
|
|
|
3002 */
|
|
|
3003 void
|
|
|
3004 spell_free_all()
|
|
|
3005 {
|
|
|
3006 slang_T *lp;
|
|
|
3007 buf_T *buf;
|
|
385
|
3008 char_u fname[MAXPATHL];
|
|
355
|
3009
|
|
|
3010 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3011 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3012 ga_clear(&buf->b_langp);
|
|
|
3013
|
|
|
3014 while (first_lang != NULL)
|
|
|
3015 {
|
|
|
3016 lp = first_lang;
|
|
|
3017 first_lang = lp->sl_next;
|
|
|
3018 slang_free(lp);
|
|
|
3019 }
|
|
366
|
3020
|
|
385
|
3021 if (int_wordlist != NULL)
|
|
|
3022 {
|
|
|
3023 /* Delete the internal wordlist and its .spl file */
|
|
|
3024 mch_remove(int_wordlist);
|
|
|
3025 int_wordlist_spl(fname);
|
|
|
3026 mch_remove(fname);
|
|
|
3027 vim_free(int_wordlist);
|
|
|
3028 int_wordlist = NULL;
|
|
381
|
3029 }
|
|
|
3030
|
|
366
|
3031 init_spell_chartab();
|
|
355
|
3032 }
|
|
|
3033 # endif
|
|
|
3034
|
|
236
|
3035 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
3036 /*
|
|
|
3037 * Clear all spelling tables and reload them.
|
|
307
|
3038 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
3039 */
|
|
|
3040 void
|
|
|
3041 spell_reload()
|
|
|
3042 {
|
|
|
3043 buf_T *buf;
|
|
316
|
3044 win_T *wp;
|
|
236
|
3045
|
|
358
|
3046 /* Initialize the table for spell_iswordp(). */
|
|
236
|
3047 init_spell_chartab();
|
|
|
3048
|
|
|
3049 /* Unload all allocated memory. */
|
|
355
|
3050 spell_free_all();
|
|
236
|
3051
|
|
|
3052 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3053 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3054 {
|
|
316
|
3055 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
3056 * window for this buffer in which 'spell' is set. */
|
|
236
|
3057 if (*buf->b_p_spl != NUL)
|
|
316
|
3058 {
|
|
|
3059 FOR_ALL_WINDOWS(wp)
|
|
|
3060 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
3061 {
|
|
|
3062 (void)did_set_spelllang(buf);
|
|
|
3063 # ifdef FEAT_WINDOWS
|
|
|
3064 break;
|
|
|
3065 # endif
|
|
|
3066 }
|
|
|
3067 }
|
|
236
|
3068 }
|
|
|
3069 }
|
|
|
3070 # endif
|
|
|
3071
|
|
310
|
3072 /*
|
|
|
3073 * Reload the spell file "fname" if it's loaded.
|
|
|
3074 */
|
|
|
3075 static void
|
|
323
|
3076 spell_reload_one(fname, added_word)
|
|
310
|
3077 char_u *fname;
|
|
323
|
3078 int added_word; /* invoked through "zg" */
|
|
310
|
3079 {
|
|
|
3080 slang_T *lp;
|
|
323
|
3081 int didit = FALSE;
|
|
310
|
3082
|
|
|
3083 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3084 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3085 {
|
|
|
3086 slang_clear(lp);
|
|
323
|
3087 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3088 redraw_all_later(NOT_VALID);
|
|
323
|
3089 didit = TRUE;
|
|
310
|
3090 }
|
|
323
|
3091
|
|
|
3092 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3093 * 'spelllang' to get it loaded. */
|
|
|
3094 if (added_word && !didit)
|
|
|
3095 did_set_spelllang(curbuf);
|
|
310
|
3096 }
|
|
|
3097
|
|
|
3098
|
|
236
|
3099 /*
|
|
|
3100 * Functions for ":mkspell".
|
|
|
3101 */
|
|
|
3102
|
|
300
|
3103 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3104 and .dic file. */
|
|
|
3105 /*
|
|
|
3106 * Main structure to store the contents of a ".aff" file.
|
|
|
3107 */
|
|
|
3108 typedef struct afffile_S
|
|
|
3109 {
|
|
|
3110 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
310
|
3111 int af_rar; /* RAR ID for rare word */
|
|
|
3112 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
3113 int af_bad; /* BAD ID for banned word */
|
|
339
|
3114 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3115 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3116 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
3117 } afffile_T;
|
|
|
3118
|
|
|
3119 typedef struct affentry_S affentry_T;
|
|
|
3120 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3121 struct affentry_S
|
|
|
3122 {
|
|
|
3123 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3124 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3125 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3126 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3127 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
366
|
3128 int ae_rare; /* rare affix */
|
|
300
|
3129 };
|
|
|
3130
|
|
455
|
3131 #define AH_KEY_LEN 10
|
|
|
3132
|
|
300
|
3133 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3134 typedef struct affheader_S
|
|
|
3135 {
|
|
455
|
3136 /* key for hashtable == name of affix entry */
|
|
|
3137 #ifdef FEAT_MBYTE
|
|
|
3138 char_u ah_key[AH_KEY_LEN]; /* multi-byte char plus NUL */
|
|
|
3139 #else
|
|
|
3140 char_u ah_key[2]; /* one byte char plus NUL */
|
|
|
3141 #endif
|
|
339
|
3142 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
3143 int ah_combine; /* suffix may combine with prefix */
|
|
|
3144 affentry_T *ah_first; /* first affix entry */
|
|
|
3145 } affheader_T;
|
|
|
3146
|
|
|
3147 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
3148
|
|
|
3149 /*
|
|
|
3150 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
3151 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
3152 * once after ":mkspell" is done.
|
|
300
|
3153 */
|
|
|
3154 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
3155 typedef struct sblock_S sblock_T;
|
|
|
3156 struct sblock_S
|
|
|
3157 {
|
|
|
3158 sblock_T *sb_next; /* next block in list */
|
|
|
3159 int sb_used; /* nr of bytes already in use */
|
|
|
3160 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
3161 };
|
|
|
3162
|
|
|
3163 /*
|
|
300
|
3164 * A node in the tree.
|
|
236
|
3165 */
|
|
300
|
3166 typedef struct wordnode_S wordnode_T;
|
|
|
3167 struct wordnode_S
|
|
236
|
3168 {
|
|
346
|
3169 union /* shared to save space */
|
|
|
3170 {
|
|
469
|
3171 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
3172 int index; /* index in written nodes (valid after first
|
|
|
3173 round) */
|
|
|
3174 } wn_u1;
|
|
|
3175 union /* shared to save space */
|
|
|
3176 {
|
|
|
3177 wordnode_T *next; /* next node with same hash key */
|
|
|
3178 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
3179 } wn_u2;
|
|
300
|
3180 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
3181 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
3182 always sorted) */
|
|
474
|
3183 int wn_refs; /* Nr. of references to this node. Only
|
|
|
3184 relevant for first node in a list of
|
|
|
3185 siblings, in following siblings it is
|
|
|
3186 always one. */
|
|
300
|
3187 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
469
|
3188 char_u wn_prefixID; /* when "wn_byte" is NUL: supported/required
|
|
|
3189 prefix ID or 0 */
|
|
|
3190 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
3191 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
3192 PREFIXTREE it's the prefcondnr */
|
|
469
|
3193 #ifdef SPELL_PRINTTREE
|
|
|
3194 int wn_nr; /* sequence nr for printing */
|
|
|
3195 #endif
|
|
236
|
3196 };
|
|
|
3197
|
|
390
|
3198 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
3199
|
|
300
|
3200 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
3201
|
|
300
|
3202 /*
|
|
|
3203 * Info used while reading the spell files.
|
|
|
3204 */
|
|
|
3205 typedef struct spellinfo_S
|
|
249
|
3206 {
|
|
300
|
3207 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
3208 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
3209 int si_fold_added; /* nr of words added since compressing */
|
|
|
3210
|
|
300
|
3211 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
3212 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
3213 int si_keep_added; /* nr of words added since compressing */
|
|
|
3214
|
|
339
|
3215 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
3216
|
|
300
|
3217 sblock_T *si_blocks; /* memory blocks used */
|
|
474
|
3218 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
3219 compression, linked by "wn_child" field. */
|
|
|
3220 #ifdef SPELL_PRINTTREE
|
|
|
3221 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
3222 #endif
|
|
|
3223
|
|
|
3224
|
|
300
|
3225 int si_ascii; /* handling only ASCII words */
|
|
310
|
3226 int si_add; /* addition file */
|
|
351
|
3227 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
3228 int si_region; /* region mask */
|
|
|
3229 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
3230 int si_memtot; /* runtime memory used */
|
|
310
|
3231 int si_verbose; /* verbose messages */
|
|
474
|
3232 int si_msg_count; /* number of words added since last message */
|
|
316
|
3233 int si_region_count; /* number of regions supported (1 when there
|
|
|
3234 are no regions) */
|
|
|
3235 char_u si_region_name[16]; /* region names (if count > 1) */
|
|
323
|
3236
|
|
|
3237 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
3238 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
3239 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
3240 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
3241 int si_followup; /* soundsalike: ? */
|
|
|
3242 int si_collapse; /* soundsalike: ? */
|
|
|
3243 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
3244 garray_T si_map; /* MAP info concatenated */
|
|
366
|
3245 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
339
|
3246 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
3247 * prefixes, each stored as a string */
|
|
|
3248 int si_newID; /* current value for ah_newID */
|
|
300
|
3249 } spellinfo_T;
|
|
249
|
3250
|
|
474
|
3251 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
339
|
3252 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
3253 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
3254 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
3255 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
3256 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
3257 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
|
3258 static char_u *get_pfxlist __ARGS((spellinfo_T *spin, afffile_T *affile, char_u *afflist));
|
|
|
3259 static int store_aff_word __ARGS((spellinfo_T *spin, char_u *word, char_u *afflist, afffile_T *affile, hashtab_T *ht, hashtab_T *xht, int comb, int flags, char_u *pfxlist));
|
|
|
3260 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
3261 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
3262 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
3263 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
3264 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
|
3265 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist));
|
|
|
3266 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int prefixID));
|
|
|
3267 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
3268 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
3269 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
3270 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
3271 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
3272 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
474
|
3273 static void write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
3274 static void clear_node __ARGS((wordnode_T *node));
|
|
|
3275 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
3276 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3277 static void init_spellfile __ARGS((void));
|
|
236
|
3278
|
|
455
|
3279 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
3280 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
3281 * Use a negative number with the lower 8 bits zero. */
|
|
|
3282 #define PFX_FLAGS -256
|
|
390
|
3283
|
|
469
|
3284 #ifdef SPELL_PRINTTREE
|
|
|
3285 /*
|
|
|
3286 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
3287 * readable format, so that we can see what happens when adding a word and/or
|
|
|
3288 * compressing the tree.
|
|
|
3289 * Based on code from Olaf Seibert.
|
|
|
3290 */
|
|
|
3291 #define PRINTLINESIZE 1000
|
|
|
3292 #define PRINTWIDTH 6
|
|
|
3293
|
|
|
3294 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
3295 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
3296
|
|
|
3297 static char line1[PRINTLINESIZE];
|
|
|
3298 static char line2[PRINTLINESIZE];
|
|
|
3299 static char line3[PRINTLINESIZE];
|
|
|
3300
|
|
|
3301 static void
|
|
|
3302 spell_clear_flags(wordnode_T *node)
|
|
|
3303 {
|
|
|
3304 wordnode_T *np;
|
|
|
3305
|
|
|
3306 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
3307 {
|
|
|
3308 np->wn_u1.index = FALSE;
|
|
|
3309 spell_clear_flags(np->wn_child);
|
|
|
3310 }
|
|
|
3311 }
|
|
|
3312
|
|
|
3313 static void
|
|
|
3314 spell_print_node(wordnode_T *node, int depth)
|
|
|
3315 {
|
|
|
3316 if (node->wn_u1.index)
|
|
|
3317 {
|
|
|
3318 /* Done this node before, print the reference. */
|
|
|
3319 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
3320 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
3321 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3322 msg(line1);
|
|
|
3323 msg(line2);
|
|
|
3324 msg(line3);
|
|
|
3325 }
|
|
|
3326 else
|
|
|
3327 {
|
|
|
3328 node->wn_u1.index = TRUE;
|
|
|
3329
|
|
|
3330 if (node->wn_byte != NUL)
|
|
|
3331 {
|
|
|
3332 if (node->wn_child != NULL)
|
|
|
3333 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
3334 else
|
|
|
3335 /* Cannot happen? */
|
|
|
3336 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
3337 }
|
|
|
3338 else
|
|
|
3339 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
3340
|
|
|
3341 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
3342
|
|
|
3343 if (node->wn_sibling != NULL)
|
|
|
3344 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
3345 else
|
|
|
3346 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3347
|
|
|
3348 if (node->wn_byte == NUL)
|
|
|
3349 {
|
|
|
3350 msg(line1);
|
|
|
3351 msg(line2);
|
|
|
3352 msg(line3);
|
|
|
3353 }
|
|
|
3354
|
|
|
3355 /* do the children */
|
|
|
3356 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
3357 spell_print_node(node->wn_child, depth + 1);
|
|
|
3358
|
|
|
3359 /* do the siblings */
|
|
|
3360 if (node->wn_sibling != NULL)
|
|
|
3361 {
|
|
|
3362 /* get rid of all parent details except | */
|
|
|
3363 STRCPY(line1, line3);
|
|
|
3364 STRCPY(line2, line3);
|
|
|
3365 spell_print_node(node->wn_sibling, depth);
|
|
|
3366 }
|
|
|
3367 }
|
|
|
3368 }
|
|
|
3369
|
|
|
3370 static void
|
|
|
3371 spell_print_tree(wordnode_T *root)
|
|
|
3372 {
|
|
|
3373 if (root != NULL)
|
|
|
3374 {
|
|
|
3375 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
3376 * done. */
|
|
|
3377 spell_clear_flags(root);
|
|
|
3378
|
|
|
3379 /* Recursively print the tree. */
|
|
|
3380 spell_print_node(root, 0);
|
|
|
3381 }
|
|
|
3382 }
|
|
|
3383 #endif /* SPELL_PRINTTREE */
|
|
|
3384
|
|
236
|
3385 /*
|
|
323
|
3386 * Read the affix file "fname".
|
|
316
|
3387 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
3388 */
|
|
|
3389 static afffile_T *
|
|
474
|
3390 spell_read_aff(spin, fname)
|
|
|
3391 spellinfo_T *spin;
|
|
236
|
3392 char_u *fname;
|
|
|
3393 {
|
|
|
3394 FILE *fd;
|
|
|
3395 afffile_T *aff;
|
|
|
3396 char_u rline[MAXLINELEN];
|
|
|
3397 char_u *line;
|
|
|
3398 char_u *pc = NULL;
|
|
334
|
3399 #define MAXITEMCNT 7
|
|
|
3400 char_u *(items[MAXITEMCNT]);
|
|
236
|
3401 int itemcnt;
|
|
|
3402 char_u *p;
|
|
|
3403 int lnum = 0;
|
|
|
3404 affheader_T *cur_aff = NULL;
|
|
|
3405 int aff_todo = 0;
|
|
|
3406 hashtab_T *tp;
|
|
255
|
3407 char_u *low = NULL;
|
|
|
3408 char_u *fol = NULL;
|
|
|
3409 char_u *upp = NULL;
|
|
307
|
3410 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
3411 int do_rep;
|
|
|
3412 int do_sal;
|
|
|
3413 int do_map;
|
|
366
|
3414 int do_midword;
|
|
375
|
3415 int do_sofo;
|
|
323
|
3416 int found_map = FALSE;
|
|
324
|
3417 hashitem_T *hi;
|
|
455
|
3418 int l;
|
|
236
|
3419
|
|
300
|
3420 /*
|
|
|
3421 * Open the file.
|
|
|
3422 */
|
|
310
|
3423 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3424 if (fd == NULL)
|
|
|
3425 {
|
|
|
3426 EMSG2(_(e_notopen), fname);
|
|
|
3427 return NULL;
|
|
|
3428 }
|
|
|
3429
|
|
310
|
3430 if (spin->si_verbose || p_verbose > 2)
|
|
|
3431 {
|
|
|
3432 if (!spin->si_verbose)
|
|
|
3433 verbose_enter();
|
|
366
|
3434 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
3435 out_flush();
|
|
|
3436 if (!spin->si_verbose)
|
|
|
3437 verbose_leave();
|
|
|
3438 }
|
|
236
|
3439
|
|
323
|
3440 /* Only do REP lines when not done in another .aff file already. */
|
|
|
3441 do_rep = spin->si_rep.ga_len == 0;
|
|
|
3442
|
|
|
3443 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
3444 do_sal = spin->si_sal.ga_len == 0;
|
|
|
3445
|
|
|
3446 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
3447 do_map = spin->si_map.ga_len == 0;
|
|
|
3448
|
|
366
|
3449 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
3450 do_midword = spin->si_midword == NULL;
|
|
|
3451
|
|
375
|
3452 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
3453 do_sofo = spin->si_sofofr == NULL;
|
|
|
3454
|
|
300
|
3455 /*
|
|
|
3456 * Allocate and init the afffile_T structure.
|
|
|
3457 */
|
|
474
|
3458 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
3459 if (aff == NULL)
|
|
|
3460 return NULL;
|
|
|
3461 hash_init(&aff->af_pref);
|
|
|
3462 hash_init(&aff->af_suff);
|
|
|
3463
|
|
|
3464 /*
|
|
|
3465 * Read all the lines in the file one by one.
|
|
|
3466 */
|
|
255
|
3467 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
3468 {
|
|
255
|
3469 line_breakcheck();
|
|
236
|
3470 ++lnum;
|
|
|
3471
|
|
|
3472 /* Skip comment lines. */
|
|
|
3473 if (*rline == '#')
|
|
|
3474 continue;
|
|
|
3475
|
|
|
3476 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
3477 vim_free(pc);
|
|
310
|
3478 #ifdef FEAT_MBYTE
|
|
300
|
3479 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3480 {
|
|
300
|
3481 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
3482 if (pc == NULL)
|
|
|
3483 {
|
|
|
3484 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3485 fname, lnum, rline);
|
|
|
3486 continue;
|
|
|
3487 }
|
|
236
|
3488 line = pc;
|
|
|
3489 }
|
|
|
3490 else
|
|
310
|
3491 #endif
|
|
236
|
3492 {
|
|
|
3493 pc = NULL;
|
|
|
3494 line = rline;
|
|
|
3495 }
|
|
|
3496
|
|
|
3497 /* Split the line up in white separated items. Put a NUL after each
|
|
|
3498 * item. */
|
|
|
3499 itemcnt = 0;
|
|
|
3500 for (p = line; ; )
|
|
|
3501 {
|
|
|
3502 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
3503 ++p;
|
|
|
3504 if (*p == NUL)
|
|
|
3505 break;
|
|
334
|
3506 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
3507 break;
|
|
236
|
3508 items[itemcnt++] = p;
|
|
300
|
3509 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
3510 ++p;
|
|
|
3511 if (*p == NUL)
|
|
|
3512 break;
|
|
|
3513 *p++ = NUL;
|
|
|
3514 }
|
|
|
3515
|
|
|
3516 /* Handle non-empty lines. */
|
|
|
3517 if (itemcnt > 0)
|
|
|
3518 {
|
|
|
3519 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
3520 && aff->af_enc == NULL)
|
|
|
3521 {
|
|
310
|
3522 #ifdef FEAT_MBYTE
|
|
300
|
3523 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
3524 aff->af_enc = enc_canonize(items[1]);
|
|
|
3525 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
3526 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
3527 p_enc) == FAIL)
|
|
|
3528 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
3529 fname, aff->af_enc, p_enc);
|
|
375
|
3530 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3531 #else
|
|
|
3532 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3533 #endif
|
|
236
|
3534 }
|
|
366
|
3535 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
3536 {
|
|
|
3537 if (do_midword)
|
|
|
3538 spin->si_midword = vim_strsave(items[1]);
|
|
|
3539 }
|
|
302
|
3540 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
3541 {
|
|
323
|
3542 /* ignored, we always split */
|
|
302
|
3543 }
|
|
323
|
3544 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
3545 {
|
|
323
|
3546 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
3547 }
|
|
307
|
3548 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
3549 && aff->af_rar == 0)
|
|
|
3550 {
|
|
|
3551 aff->af_rar = items[1][0];
|
|
|
3552 if (items[1][1] != NUL)
|
|
|
3553 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3554 }
|
|
310
|
3555 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
3556 && aff->af_kep == 0)
|
|
307
|
3557 {
|
|
310
|
3558 aff->af_kep = items[1][0];
|
|
307
|
3559 if (items[1][1] != NUL)
|
|
|
3560 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3561 }
|
|
346
|
3562 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
3563 && aff->af_bad == 0)
|
|
|
3564 {
|
|
|
3565 aff->af_bad = items[1][0];
|
|
|
3566 if (items[1][1] != NUL)
|
|
|
3567 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3568 }
|
|
339
|
3569 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
3570 {
|
|
|
3571 aff->af_pfxpostpone = TRUE;
|
|
|
3572 }
|
|
236
|
3573 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3574 || STRCMP(items[0], "SFX") == 0)
|
|
|
3575 && aff_todo == 0
|
|
334
|
3576 && itemcnt >= 4)
|
|
236
|
3577 {
|
|
334
|
3578 /* Myspell allows extra text after the item, but that might
|
|
|
3579 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
3580 if (itemcnt > 4 && *items[4] != '#')
|
|
|
3581 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
3582 fname, lnum, items[4]);
|
|
|
3583
|
|
236
|
3584 /* New affix letter. */
|
|
474
|
3585 cur_aff = (affheader_T *)getroom(spin,
|
|
447
|
3586 sizeof(affheader_T), TRUE);
|
|
236
|
3587 if (cur_aff == NULL)
|
|
|
3588 break;
|
|
455
|
3589 #ifdef FEAT_MBYTE
|
|
|
3590 if (has_mbyte)
|
|
|
3591 {
|
|
474
|
3592 l = (*mb_ptr2len)(items[1]);
|
|
455
|
3593 if (l >= AH_KEY_LEN)
|
|
|
3594 l = 1; /* too long, must be an overlong sequence */
|
|
|
3595 else
|
|
|
3596 mch_memmove(cur_aff->ah_key, items[1], l);
|
|
|
3597 }
|
|
|
3598 else
|
|
|
3599 #endif
|
|
|
3600 {
|
|
|
3601 *cur_aff->ah_key = *items[1];
|
|
|
3602 l = 1;
|
|
|
3603 }
|
|
|
3604 cur_aff->ah_key[l] = NUL;
|
|
|
3605 if (items[1][l] != NUL)
|
|
307
|
3606 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
3607 if (*items[2] == 'Y')
|
|
|
3608 cur_aff->ah_combine = TRUE;
|
|
300
|
3609 else if (*items[2] != 'N')
|
|
236
|
3610 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
3611 fname, lnum, items[2]);
|
|
339
|
3612
|
|
236
|
3613 if (*items[0] == 'P')
|
|
339
|
3614 {
|
|
236
|
3615 tp = &aff->af_pref;
|
|
339
|
3616 /* Use a new number in the .spl file later, to be able to
|
|
|
3617 * handle multiple .aff files. */
|
|
|
3618 if (aff->af_pfxpostpone)
|
|
344
|
3619 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
3620 }
|
|
236
|
3621 else
|
|
|
3622 tp = &aff->af_suff;
|
|
300
|
3623 aff_todo = atoi((char *)items[3]);
|
|
324
|
3624 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
3625 if (!HASHITEM_EMPTY(hi))
|
|
300
|
3626 {
|
|
236
|
3627 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
3628 fname, lnum, items[1]);
|
|
300
|
3629 aff_todo = 0;
|
|
|
3630 }
|
|
236
|
3631 else
|
|
|
3632 hash_add(tp, cur_aff->ah_key);
|
|
|
3633 }
|
|
|
3634 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3635 || STRCMP(items[0], "SFX") == 0)
|
|
|
3636 && aff_todo > 0
|
|
|
3637 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
3638 && itemcnt >= 5)
|
|
236
|
3639 {
|
|
|
3640 affentry_T *aff_entry;
|
|
366
|
3641 int rare = FALSE;
|
|
455
|
3642 int upper = FALSE;
|
|
366
|
3643 int lasti = 5;
|
|
|
3644
|
|
|
3645 /* Check for "rare" after the other info. */
|
|
|
3646 if (itemcnt > 5 && STRICMP(items[5], "rare") == 0)
|
|
|
3647 {
|
|
|
3648 rare = TRUE;
|
|
|
3649 lasti = 6;
|
|
|
3650 }
|
|
236
|
3651
|
|
334
|
3652 /* Myspell allows extra text after the item, but that might
|
|
|
3653 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
3654 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
3655 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
366
|
3656 fname, lnum, items[lasti]);
|
|
334
|
3657
|
|
236
|
3658 /* New item for an affix letter. */
|
|
|
3659 --aff_todo;
|
|
474
|
3660 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
3661 sizeof(affentry_T), TRUE);
|
|
236
|
3662 if (aff_entry == NULL)
|
|
|
3663 break;
|
|
366
|
3664 aff_entry->ae_rare = rare;
|
|
240
|
3665
|
|
236
|
3666 if (STRCMP(items[2], "0") != 0)
|
|
474
|
3667 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
3668 if (STRCMP(items[3], "0") != 0)
|
|
474
|
3669 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
3670
|
|
300
|
3671 /* Don't use an affix entry with non-ASCII characters when
|
|
|
3672 * "spin->si_ascii" is TRUE. */
|
|
|
3673 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
3674 || has_non_ascii(aff_entry->ae_add)))
|
|
|
3675 {
|
|
|
3676 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
3677 cur_aff->ah_first = aff_entry;
|
|
300
|
3678
|
|
|
3679 if (STRCMP(items[4], ".") != 0)
|
|
|
3680 {
|
|
|
3681 char_u buf[MAXLINELEN];
|
|
|
3682
|
|
474
|
3683 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
3684 if (*items[0] == 'P')
|
|
|
3685 sprintf((char *)buf, "^%s", items[4]);
|
|
|
3686 else
|
|
|
3687 sprintf((char *)buf, "%s$", items[4]);
|
|
|
3688 aff_entry->ae_prog = vim_regcomp(buf,
|
|
|
3689 RE_MAGIC + RE_STRING);
|
|
|
3690 }
|
|
339
|
3691
|
|
|
3692 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
3693 * for the condition. Use an existing one if possible. */
|
|
455
|
3694 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
3695 {
|
|
455
|
3696 /* When the chop string is one lower-case letter and
|
|
|
3697 * the add string ends in the upper-case letter we set
|
|
|
3698 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
3699 * letters from "ae_add". The condition must either
|
|
|
3700 * be empty or start with the same letter. */
|
|
|
3701 if (aff_entry->ae_chop != NULL
|
|
|
3702 && aff_entry->ae_add != NULL
|
|
|
3703 #ifdef FEAT_MBYTE
|
|
474
|
3704 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
3705 aff_entry->ae_chop)] == NUL
|
|
|
3706 #else
|
|
|
3707 && aff_entry->ae_chop[1] == NUL
|
|
|
3708 #endif
|
|
|
3709 )
|
|
339
|
3710 {
|
|
455
|
3711 int c, c_up;
|
|
|
3712
|
|
|
3713 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
3714 c_up = SPELL_TOUPPER(c);
|
|
|
3715 if (c_up != c
|
|
|
3716 && (aff_entry->ae_cond == NULL
|
|
|
3717 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
3718 {
|
|
|
3719 p = aff_entry->ae_add
|
|
|
3720 + STRLEN(aff_entry->ae_add);
|
|
|
3721 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
3722 if (PTR2CHAR(p) == c_up)
|
|
|
3723 {
|
|
|
3724 upper = TRUE;
|
|
|
3725 aff_entry->ae_chop = NULL;
|
|
|
3726 *p = NUL;
|
|
|
3727
|
|
|
3728 /* The condition is matched with the
|
|
|
3729 * actual word, thus must check for the
|
|
|
3730 * upper-case letter. */
|
|
|
3731 if (aff_entry->ae_cond != NULL)
|
|
|
3732 {
|
|
|
3733 char_u buf[MAXLINELEN];
|
|
|
3734 #ifdef FEAT_MBYTE
|
|
|
3735 if (has_mbyte)
|
|
|
3736 {
|
|
|
3737 onecap_copy(items[4], buf, TRUE);
|
|
|
3738 aff_entry->ae_cond = getroom_save(
|
|
474
|
3739 spin, buf);
|
|
455
|
3740 }
|
|
|
3741 else
|
|
|
3742 #endif
|
|
|
3743 *aff_entry->ae_cond = c_up;
|
|
|
3744 if (aff_entry->ae_cond != NULL)
|
|
|
3745 {
|
|
|
3746 sprintf((char *)buf, "^%s",
|
|
339
|
3747 aff_entry->ae_cond);
|
|
455
|
3748 vim_free(aff_entry->ae_prog);
|
|
|
3749 aff_entry->ae_prog = vim_regcomp(
|
|
|
3750 buf, RE_MAGIC + RE_STRING);
|
|
|
3751 }
|
|
|
3752 }
|
|
|
3753 }
|
|
|
3754 }
|
|
339
|
3755 }
|
|
|
3756
|
|
455
|
3757 if (aff_entry->ae_chop == NULL)
|
|
390
|
3758 {
|
|
455
|
3759 int idx;
|
|
|
3760 char_u **pp;
|
|
|
3761 int n;
|
|
|
3762
|
|
|
3763 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
3764 --idx)
|
|
|
3765 {
|
|
|
3766 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
3767 if (str_equal(p, aff_entry->ae_cond))
|
|
|
3768 break;
|
|
|
3769 }
|
|
|
3770 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
3771 {
|
|
|
3772 /* Not found, add a new condition. */
|
|
|
3773 idx = spin->si_prefcond.ga_len++;
|
|
|
3774 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
3775 + idx;
|
|
|
3776 if (aff_entry->ae_cond == NULL)
|
|
|
3777 *pp = NULL;
|
|
|
3778 else
|
|
474
|
3779 *pp = getroom_save(spin,
|
|
455
|
3780 aff_entry->ae_cond);
|
|
|
3781 }
|
|
|
3782
|
|
|
3783 /* Add the prefix to the prefix tree. */
|
|
|
3784 if (aff_entry->ae_add == NULL)
|
|
|
3785 p = (char_u *)"";
|
|
390
|
3786 else
|
|
455
|
3787 p = aff_entry->ae_add;
|
|
|
3788 /* PFX_FLAGS is a negative number, so that
|
|
|
3789 * tree_add_word() knows this is the prefix tree. */
|
|
|
3790 n = PFX_FLAGS;
|
|
|
3791 if (rare)
|
|
|
3792 n |= WFP_RARE;
|
|
|
3793 if (!cur_aff->ah_combine)
|
|
|
3794 n |= WFP_NC;
|
|
|
3795 if (upper)
|
|
|
3796 n |= WFP_UP;
|
|
474
|
3797 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
3798 idx, cur_aff->ah_newID);
|
|
390
|
3799 }
|
|
339
|
3800 }
|
|
240
|
3801 }
|
|
236
|
3802 }
|
|
255
|
3803 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
3804 {
|
|
|
3805 if (fol != NULL)
|
|
|
3806 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
3807 fname, lnum);
|
|
|
3808 else
|
|
|
3809 fol = vim_strsave(items[1]);
|
|
|
3810 }
|
|
|
3811 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
3812 {
|
|
|
3813 if (low != NULL)
|
|
|
3814 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
3815 fname, lnum);
|
|
|
3816 else
|
|
|
3817 low = vim_strsave(items[1]);
|
|
|
3818 }
|
|
|
3819 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
3820 {
|
|
|
3821 if (upp != NULL)
|
|
|
3822 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
3823 fname, lnum);
|
|
|
3824 else
|
|
|
3825 upp = vim_strsave(items[1]);
|
|
|
3826 }
|
|
236
|
3827 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
3828 {
|
|
236
|
3829 /* Ignore REP count */;
|
|
323
|
3830 if (!isdigit(*items[1]))
|
|
|
3831 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
3832 fname, lnum);
|
|
|
3833 }
|
|
236
|
3834 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 3)
|
|
|
3835 {
|
|
|
3836 /* REP item */
|
|
323
|
3837 if (do_rep)
|
|
|
3838 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
3839 }
|
|
|
3840 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
3841 {
|
|
|
3842 /* MAP item or count */
|
|
|
3843 if (!found_map)
|
|
|
3844 {
|
|
|
3845 /* First line contains the count. */
|
|
|
3846 found_map = TRUE;
|
|
|
3847 if (!isdigit(*items[1]))
|
|
|
3848 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
3849 fname, lnum);
|
|
|
3850 }
|
|
|
3851 else if (do_map)
|
|
|
3852 {
|
|
346
|
3853 int c;
|
|
|
3854
|
|
|
3855 /* Check that every character appears only once. */
|
|
|
3856 for (p = items[1]; *p != NUL; )
|
|
|
3857 {
|
|
|
3858 #ifdef FEAT_MBYTE
|
|
|
3859 c = mb_ptr2char_adv(&p);
|
|
|
3860 #else
|
|
|
3861 c = *p++;
|
|
|
3862 #endif
|
|
|
3863 if ((spin->si_map.ga_len > 0
|
|
|
3864 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
3865 != NULL)
|
|
|
3866 || vim_strchr(p, c) != NULL)
|
|
|
3867 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
3868 fname, lnum);
|
|
|
3869 }
|
|
|
3870
|
|
323
|
3871 /* We simply concatenate all the MAP strings, separated by
|
|
|
3872 * slashes. */
|
|
|
3873 ga_concat(&spin->si_map, items[1]);
|
|
|
3874 ga_append(&spin->si_map, '/');
|
|
|
3875 }
|
|
|
3876 }
|
|
|
3877 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
3878 {
|
|
|
3879 if (do_sal)
|
|
|
3880 {
|
|
|
3881 /* SAL item (sounds-a-like)
|
|
|
3882 * Either one of the known keys or a from-to pair. */
|
|
|
3883 if (STRCMP(items[1], "followup") == 0)
|
|
|
3884 spin->si_followup = sal_to_bool(items[2]);
|
|
|
3885 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
3886 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
3887 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
3888 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
3889 else
|
|
|
3890 /* when "to" is "_" it means empty */
|
|
|
3891 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
3892 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
3893 : items[2]);
|
|
|
3894 }
|
|
236
|
3895 }
|
|
375
|
3896 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
3897 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
3898 {
|
|
|
3899 if (do_sofo)
|
|
|
3900 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
3901 }
|
|
|
3902 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
3903 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
3904 {
|
|
|
3905 if (do_sofo)
|
|
|
3906 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
3907 }
|
|
300
|
3908 else
|
|
236
|
3909 smsg((char_u *)_("Unrecognized item in %s line %d: %s"),
|
|
|
3910 fname, lnum, items[0]);
|
|
|
3911 }
|
|
|
3912 }
|
|
|
3913
|
|
375
|
3914 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
3915 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
3916 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
3917 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
3918 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
3919
|
|
255
|
3920 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
3921 {
|
|
351
|
3922 if (spin->si_clear_chartab)
|
|
|
3923 {
|
|
|
3924 /* Clear the char type tables, don't want to use any of the
|
|
|
3925 * currently used spell properties. */
|
|
|
3926 init_spell_chartab();
|
|
|
3927 spin->si_clear_chartab = FALSE;
|
|
|
3928 }
|
|
|
3929
|
|
316
|
3930 /*
|
|
|
3931 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
3932 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
3933 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
3934 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
3935 */
|
|
|
3936 if (!spin->si_ascii
|
|
|
3937 #ifdef FEAT_MBYTE
|
|
|
3938 && !enc_utf8
|
|
|
3939 #endif
|
|
|
3940 )
|
|
260
|
3941 {
|
|
|
3942 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
3943 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
3944 else
|
|
316
|
3945 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
3946 }
|
|
255
|
3947
|
|
|
3948 vim_free(fol);
|
|
|
3949 vim_free(low);
|
|
|
3950 vim_free(upp);
|
|
|
3951 }
|
|
|
3952
|
|
236
|
3953 vim_free(pc);
|
|
|
3954 fclose(fd);
|
|
|
3955 return aff;
|
|
|
3956 }
|
|
|
3957
|
|
|
3958 /*
|
|
339
|
3959 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
3960 * NULL as equal.
|
|
|
3961 */
|
|
|
3962 static int
|
|
|
3963 str_equal(s1, s2)
|
|
|
3964 char_u *s1;
|
|
|
3965 char_u *s2;
|
|
|
3966 {
|
|
|
3967 if (s1 == NULL || s2 == NULL)
|
|
|
3968 return s1 == s2;
|
|
|
3969 return STRCMP(s1, s2) == 0;
|
|
|
3970 }
|
|
|
3971
|
|
|
3972 /*
|
|
323
|
3973 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
3974 * They are stored case-folded.
|
|
|
3975 */
|
|
|
3976 static void
|
|
|
3977 add_fromto(spin, gap, from, to)
|
|
|
3978 spellinfo_T *spin;
|
|
|
3979 garray_T *gap;
|
|
|
3980 char_u *from;
|
|
|
3981 char_u *to;
|
|
|
3982 {
|
|
|
3983 fromto_T *ftp;
|
|
|
3984 char_u word[MAXWLEN];
|
|
|
3985
|
|
|
3986 if (ga_grow(gap, 1) == OK)
|
|
|
3987 {
|
|
|
3988 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
3989 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
3990 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
3991 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
3992 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
3993 ++gap->ga_len;
|
|
|
3994 }
|
|
|
3995 }
|
|
|
3996
|
|
|
3997 /*
|
|
|
3998 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
3999 */
|
|
|
4000 static int
|
|
|
4001 sal_to_bool(s)
|
|
|
4002 char_u *s;
|
|
|
4003 {
|
|
|
4004 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
4005 }
|
|
|
4006
|
|
|
4007 /*
|
|
240
|
4008 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
4009 * When "s" is NULL FALSE is returned.
|
|
|
4010 */
|
|
|
4011 static int
|
|
|
4012 has_non_ascii(s)
|
|
|
4013 char_u *s;
|
|
|
4014 {
|
|
|
4015 char_u *p;
|
|
|
4016
|
|
|
4017 if (s != NULL)
|
|
|
4018 for (p = s; *p != NUL; ++p)
|
|
|
4019 if (*p >= 128)
|
|
|
4020 return TRUE;
|
|
|
4021 return FALSE;
|
|
|
4022 }
|
|
|
4023
|
|
|
4024 /*
|
|
236
|
4025 * Free the structure filled by spell_read_aff().
|
|
|
4026 */
|
|
|
4027 static void
|
|
|
4028 spell_free_aff(aff)
|
|
|
4029 afffile_T *aff;
|
|
|
4030 {
|
|
|
4031 hashtab_T *ht;
|
|
|
4032 hashitem_T *hi;
|
|
|
4033 int todo;
|
|
|
4034 affheader_T *ah;
|
|
300
|
4035 affentry_T *ae;
|
|
236
|
4036
|
|
|
4037 vim_free(aff->af_enc);
|
|
|
4038
|
|
339
|
4039 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
4040 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
4041 {
|
|
|
4042 todo = ht->ht_used;
|
|
|
4043 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
4044 {
|
|
|
4045 if (!HASHITEM_EMPTY(hi))
|
|
|
4046 {
|
|
|
4047 --todo;
|
|
|
4048 ah = HI2AH(hi);
|
|
300
|
4049 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
4050 vim_free(ae->ae_prog);
|
|
236
|
4051 }
|
|
|
4052 }
|
|
|
4053 if (ht == &aff->af_suff)
|
|
|
4054 break;
|
|
|
4055 }
|
|
300
|
4056
|
|
236
|
4057 hash_clear(&aff->af_pref);
|
|
|
4058 hash_clear(&aff->af_suff);
|
|
|
4059 }
|
|
|
4060
|
|
|
4061 /*
|
|
300
|
4062 * Read dictionary file "fname".
|
|
236
|
4063 * Returns OK or FAIL;
|
|
|
4064 */
|
|
|
4065 static int
|
|
474
|
4066 spell_read_dic(spin, fname, affile)
|
|
|
4067 spellinfo_T *spin;
|
|
236
|
4068 char_u *fname;
|
|
300
|
4069 afffile_T *affile;
|
|
236
|
4070 {
|
|
300
|
4071 hashtab_T ht;
|
|
236
|
4072 char_u line[MAXLINELEN];
|
|
300
|
4073 char_u *afflist;
|
|
339
|
4074 char_u *pfxlist;
|
|
300
|
4075 char_u *dw;
|
|
236
|
4076 char_u *pc;
|
|
|
4077 char_u *w;
|
|
|
4078 int l;
|
|
|
4079 hash_T hash;
|
|
|
4080 hashitem_T *hi;
|
|
|
4081 FILE *fd;
|
|
|
4082 int lnum = 1;
|
|
300
|
4083 int non_ascii = 0;
|
|
|
4084 int retval = OK;
|
|
|
4085 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
4086 int flags;
|
|
236
|
4087
|
|
300
|
4088 /*
|
|
|
4089 * Open the file.
|
|
|
4090 */
|
|
310
|
4091 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4092 if (fd == NULL)
|
|
|
4093 {
|
|
|
4094 EMSG2(_(e_notopen), fname);
|
|
|
4095 return FAIL;
|
|
|
4096 }
|
|
|
4097
|
|
300
|
4098 /* The hashtable is only used to detect duplicated words. */
|
|
|
4099 hash_init(&ht);
|
|
|
4100
|
|
310
|
4101 if (spin->si_verbose || p_verbose > 2)
|
|
|
4102 {
|
|
|
4103 if (!spin->si_verbose)
|
|
|
4104 verbose_enter();
|
|
366
|
4105 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
4106 out_flush();
|
|
|
4107 if (!spin->si_verbose)
|
|
|
4108 verbose_leave();
|
|
|
4109 }
|
|
236
|
4110
|
|
474
|
4111 /* start with a message for the first line */
|
|
|
4112 spin->si_msg_count = 999999;
|
|
|
4113
|
|
236
|
4114 /* Read and ignore the first line: word count. */
|
|
|
4115 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
4116 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
4117 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
4118
|
|
|
4119 /*
|
|
|
4120 * Read all the lines in the file one by one.
|
|
|
4121 * The words are converted to 'encoding' here, before being added to
|
|
|
4122 * the hashtable.
|
|
|
4123 */
|
|
255
|
4124 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
4125 {
|
|
255
|
4126 line_breakcheck();
|
|
236
|
4127 ++lnum;
|
|
455
|
4128 if (line[0] == '#' || line[0] == '/')
|
|
351
|
4129 continue; /* comment line */
|
|
236
|
4130
|
|
300
|
4131 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
4132 * the word is kept to allow e.g., "et al.". */
|
|
236
|
4133 l = STRLEN(line);
|
|
|
4134 while (l > 0 && line[l - 1] <= ' ')
|
|
|
4135 --l;
|
|
|
4136 if (l == 0)
|
|
|
4137 continue; /* empty line */
|
|
|
4138 line[l] = NUL;
|
|
|
4139
|
|
|
4140 /* Find the optional affix names. */
|
|
300
|
4141 afflist = vim_strchr(line, '/');
|
|
|
4142 if (afflist != NULL)
|
|
|
4143 *afflist++ = NUL;
|
|
236
|
4144
|
|
300
|
4145 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4146 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4147 {
|
|
|
4148 ++non_ascii;
|
|
240
|
4149 continue;
|
|
300
|
4150 }
|
|
240
|
4151
|
|
310
|
4152 #ifdef FEAT_MBYTE
|
|
236
|
4153 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
4154 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4155 {
|
|
300
|
4156 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
4157 if (pc == NULL)
|
|
|
4158 {
|
|
|
4159 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4160 fname, lnum, line);
|
|
|
4161 continue;
|
|
|
4162 }
|
|
236
|
4163 w = pc;
|
|
|
4164 }
|
|
|
4165 else
|
|
310
|
4166 #endif
|
|
236
|
4167 {
|
|
|
4168 pc = NULL;
|
|
|
4169 w = line;
|
|
|
4170 }
|
|
|
4171
|
|
474
|
4172 /* This takes time, print a message every 10000 words. */
|
|
|
4173 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
4174 {
|
|
|
4175 spin->si_msg_count = 0;
|
|
339
|
4176 vim_snprintf((char *)message, sizeof(message),
|
|
|
4177 _("line %6d, word %6d - %s"),
|
|
|
4178 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
4179 msg_start();
|
|
|
4180 msg_puts_long_attr(message, 0);
|
|
|
4181 msg_clr_eos();
|
|
|
4182 msg_didout = FALSE;
|
|
|
4183 msg_col = 0;
|
|
|
4184 out_flush();
|
|
|
4185 }
|
|
|
4186
|
|
300
|
4187 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
4188 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
4189 if (dw == NULL)
|
|
300
|
4190 retval = FAIL;
|
|
|
4191 vim_free(pc);
|
|
|
4192 if (retval == FAIL)
|
|
236
|
4193 break;
|
|
|
4194
|
|
300
|
4195 hash = hash_hash(dw);
|
|
|
4196 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
4197 if (!HASHITEM_EMPTY(hi))
|
|
|
4198 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
4199 fname, lnum, dw);
|
|
236
|
4200 else
|
|
300
|
4201 hash_add_item(&ht, hi, dw, hash);
|
|
|
4202
|
|
307
|
4203 flags = 0;
|
|
339
|
4204 pfxlist = NULL;
|
|
307
|
4205 if (afflist != NULL)
|
|
|
4206 {
|
|
|
4207 /* Check for affix name that stands for keep-case word and stands
|
|
|
4208 * for rare word (if defined). */
|
|
310
|
4209 if (affile->af_kep != NUL
|
|
|
4210 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
389
|
4211 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
307
|
4212 if (affile->af_rar != NUL
|
|
|
4213 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
4214 flags |= WF_RARE;
|
|
346
|
4215 if (affile->af_bad != NUL
|
|
|
4216 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
4217 flags |= WF_BANNED;
|
|
339
|
4218
|
|
|
4219 if (affile->af_pfxpostpone)
|
|
|
4220 /* Need to store the list of prefix IDs with the word. */
|
|
474
|
4221 pfxlist = get_pfxlist(spin, affile, afflist);
|
|
307
|
4222 }
|
|
|
4223
|
|
300
|
4224 /* Add the word to the word tree(s). */
|
|
474
|
4225 if (store_word(spin, dw, flags, spin->si_region, pfxlist) == FAIL)
|
|
300
|
4226 retval = FAIL;
|
|
236
|
4227
|
|
300
|
4228 if (afflist != NULL)
|
|
|
4229 {
|
|
|
4230 /* Find all matching suffixes and add the resulting words.
|
|
|
4231 * Additionally do matching prefixes that combine. */
|
|
474
|
4232 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
4233 &affile->af_suff, &affile->af_pref,
|
|
339
|
4234 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
4235 retval = FAIL;
|
|
|
4236
|
|
|
4237 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
4238 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
4239 &affile->af_pref, NULL,
|
|
|
4240 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
4241 retval = FAIL;
|
|
|
4242 }
|
|
236
|
4243 }
|
|
|
4244
|
|
300
|
4245 if (spin->si_ascii && non_ascii > 0)
|
|
|
4246 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4247 non_ascii);
|
|
|
4248 hash_clear(&ht);
|
|
|
4249
|
|
236
|
4250 fclose(fd);
|
|
300
|
4251 return retval;
|
|
236
|
4252 }
|
|
|
4253
|
|
|
4254 /*
|
|
339
|
4255 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
4256 * Used for PFXPOSTPONE.
|
|
|
4257 * Returns a string allocated with getroom(). NULL when there are no prefixes
|
|
|
4258 * or when out of memory.
|
|
|
4259 */
|
|
|
4260 static char_u *
|
|
474
|
4261 get_pfxlist(spin, affile, afflist)
|
|
|
4262 spellinfo_T *spin;
|
|
339
|
4263 afffile_T *affile;
|
|
|
4264 char_u *afflist;
|
|
|
4265 {
|
|
|
4266 char_u *p;
|
|
|
4267 int cnt;
|
|
|
4268 int round;
|
|
|
4269 char_u *res = NULL;
|
|
|
4270 char_u key[2];
|
|
|
4271 hashitem_T *hi;
|
|
|
4272
|
|
|
4273 key[1] = NUL;
|
|
|
4274
|
|
|
4275 /* round 1: count the number of prefix IDs.
|
|
|
4276 * round 2: move prefix IDs to "res" */
|
|
|
4277 for (round = 1; round <= 2; ++round)
|
|
|
4278 {
|
|
|
4279 cnt = 0;
|
|
|
4280 for (p = afflist; *p != NUL; ++p)
|
|
|
4281 {
|
|
|
4282 key[0] = *p;
|
|
|
4283 hi = hash_find(&affile->af_pref, key);
|
|
|
4284 if (!HASHITEM_EMPTY(hi))
|
|
|
4285 {
|
|
|
4286 /* This is a prefix ID, use the new number. */
|
|
|
4287 if (round == 2)
|
|
|
4288 res[cnt] = HI2AH(hi)->ah_newID;
|
|
|
4289 ++cnt;
|
|
|
4290 }
|
|
|
4291 }
|
|
|
4292 if (round == 1 && cnt > 0)
|
|
474
|
4293 res = getroom(spin, cnt + 1, FALSE);
|
|
339
|
4294 if (res == NULL)
|
|
|
4295 break;
|
|
|
4296 }
|
|
|
4297
|
|
|
4298 if (res != NULL)
|
|
|
4299 res[cnt] = NUL;
|
|
|
4300 return res;
|
|
|
4301 }
|
|
|
4302
|
|
|
4303 /*
|
|
300
|
4304 * Apply affixes to a word and store the resulting words.
|
|
|
4305 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
4306 * prefixes or suffixes.
|
|
|
4307 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
4308 * the resulting words for combining affixes.
|
|
|
4309 *
|
|
|
4310 * Returns FAIL when out of memory.
|
|
236
|
4311 */
|
|
300
|
4312 static int
|
|
474
|
4313 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags, pfxlist)
|
|
|
4314 spellinfo_T *spin; /* spell info */
|
|
300
|
4315 char_u *word; /* basic word start */
|
|
|
4316 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
4317 afffile_T *affile;
|
|
300
|
4318 hashtab_T *ht;
|
|
|
4319 hashtab_T *xht;
|
|
|
4320 int comb; /* only use affixes that combine */
|
|
307
|
4321 int flags; /* flags for the word */
|
|
339
|
4322 char_u *pfxlist; /* list of prefix IDs */
|
|
236
|
4323 {
|
|
|
4324 int todo;
|
|
|
4325 hashitem_T *hi;
|
|
300
|
4326 affheader_T *ah;
|
|
|
4327 affentry_T *ae;
|
|
|
4328 regmatch_T regmatch;
|
|
|
4329 char_u newword[MAXWLEN];
|
|
|
4330 int retval = OK;
|
|
|
4331 int i;
|
|
|
4332 char_u *p;
|
|
366
|
4333 int use_flags;
|
|
390
|
4334 char_u *use_pfxlist;
|
|
455
|
4335 int c;
|
|
236
|
4336
|
|
300
|
4337 todo = ht->ht_used;
|
|
|
4338 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
4339 {
|
|
|
4340 if (!HASHITEM_EMPTY(hi))
|
|
|
4341 {
|
|
|
4342 --todo;
|
|
300
|
4343 ah = HI2AH(hi);
|
|
236
|
4344
|
|
300
|
4345 /* Check that the affix combines, if required, and that the word
|
|
|
4346 * supports this affix. */
|
|
455
|
4347 c = PTR2CHAR(ah->ah_key);
|
|
|
4348 if ((!comb || ah->ah_combine) && vim_strchr(afflist, c) != NULL)
|
|
236
|
4349 {
|
|
300
|
4350 /* Loop over all affix entries with this name. */
|
|
|
4351 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
4352 {
|
|
300
|
4353 /* Check the condition. It's not logical to match case
|
|
|
4354 * here, but it is required for compatibility with
|
|
339
|
4355 * Myspell.
|
|
|
4356 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
4357 * prefixes with a chop string. */
|
|
300
|
4358 regmatch.regprog = ae->ae_prog;
|
|
|
4359 regmatch.rm_ic = FALSE;
|
|
339
|
4360 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
4361 || ae->ae_chop != NULL)
|
|
|
4362 && (ae->ae_prog == NULL
|
|
|
4363 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
4364 {
|
|
|
4365 /* Match. Remove the chop and add the affix. */
|
|
|
4366 if (xht == NULL)
|
|
240
|
4367 {
|
|
300
|
4368 /* prefix: chop/add at the start of the word */
|
|
|
4369 if (ae->ae_add == NULL)
|
|
|
4370 *newword = NUL;
|
|
|
4371 else
|
|
|
4372 STRCPY(newword, ae->ae_add);
|
|
|
4373 p = word;
|
|
|
4374 if (ae->ae_chop != NULL)
|
|
310
|
4375 {
|
|
300
|
4376 /* Skip chop string. */
|
|
310
|
4377 #ifdef FEAT_MBYTE
|
|
|
4378 if (has_mbyte)
|
|
324
|
4379 {
|
|
310
|
4380 i = mb_charlen(ae->ae_chop);
|
|
324
|
4381 for ( ; i > 0; --i)
|
|
|
4382 mb_ptr_adv(p);
|
|
|
4383 }
|
|
310
|
4384 else
|
|
|
4385 #endif
|
|
324
|
4386 p += STRLEN(ae->ae_chop);
|
|
310
|
4387 }
|
|
300
|
4388 STRCAT(newword, p);
|
|
|
4389 }
|
|
|
4390 else
|
|
|
4391 {
|
|
|
4392 /* suffix: chop/add at the end of the word */
|
|
|
4393 STRCPY(newword, word);
|
|
|
4394 if (ae->ae_chop != NULL)
|
|
|
4395 {
|
|
|
4396 /* Remove chop string. */
|
|
|
4397 p = newword + STRLEN(newword);
|
|
376
|
4398 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
4399 for ( ; i > 0; --i)
|
|
300
|
4400 mb_ptr_back(newword, p);
|
|
|
4401 *p = NUL;
|
|
|
4402 }
|
|
|
4403 if (ae->ae_add != NULL)
|
|
|
4404 STRCAT(newword, ae->ae_add);
|
|
240
|
4405 }
|
|
|
4406
|
|
366
|
4407 /* Obey the "rare" flag of the affix. */
|
|
|
4408 if (ae->ae_rare)
|
|
|
4409 use_flags = flags | WF_RARE;
|
|
|
4410 else
|
|
|
4411 use_flags = flags;
|
|
390
|
4412 use_pfxlist = pfxlist;
|
|
|
4413
|
|
|
4414 /* When there are postponed prefixes... */
|
|
393
|
4415 if (spin->si_prefroot != NULL
|
|
|
4416 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
4417 {
|
|
|
4418 /* ... add a flag to indicate an affix was used. */
|
|
|
4419 use_flags |= WF_HAS_AFF;
|
|
|
4420
|
|
|
4421 /* ... don't use a prefix list if combining
|
|
|
4422 * affixes is not allowed */
|
|
|
4423 if (!ah->ah_combine || comb)
|
|
|
4424 use_pfxlist = NULL;
|
|
|
4425 }
|
|
366
|
4426
|
|
300
|
4427 /* Store the modified word. */
|
|
474
|
4428 if (store_word(spin, newword, use_flags,
|
|
390
|
4429 spin->si_region, use_pfxlist) == FAIL)
|
|
300
|
4430 retval = FAIL;
|
|
236
|
4431
|
|
300
|
4432 /* When added a suffix and combining is allowed also
|
|
|
4433 * try adding prefixes additionally. */
|
|
|
4434 if (xht != NULL && ah->ah_combine)
|
|
474
|
4435 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
4436 xht, NULL, TRUE,
|
|
|
4437 use_flags, use_pfxlist) == FAIL)
|
|
300
|
4438 retval = FAIL;
|
|
236
|
4439 }
|
|
|
4440 }
|
|
|
4441 }
|
|
|
4442 }
|
|
|
4443 }
|
|
|
4444
|
|
|
4445 return retval;
|
|
|
4446 }
|
|
|
4447
|
|
|
4448 /*
|
|
300
|
4449 * Read a file with a list of words.
|
|
236
|
4450 */
|
|
|
4451 static int
|
|
474
|
4452 spell_read_wordfile(spin, fname)
|
|
|
4453 spellinfo_T *spin;
|
|
300
|
4454 char_u *fname;
|
|
236
|
4455 {
|
|
300
|
4456 FILE *fd;
|
|
|
4457 long lnum = 0;
|
|
|
4458 char_u rline[MAXLINELEN];
|
|
|
4459 char_u *line;
|
|
|
4460 char_u *pc = NULL;
|
|
381
|
4461 char_u *p;
|
|
300
|
4462 int l;
|
|
|
4463 int retval = OK;
|
|
|
4464 int did_word = FALSE;
|
|
|
4465 int non_ascii = 0;
|
|
307
|
4466 int flags;
|
|
316
|
4467 int regionmask;
|
|
236
|
4468
|
|
300
|
4469 /*
|
|
|
4470 * Open the file.
|
|
|
4471 */
|
|
310
|
4472 fd = mch_fopen((char *)fname, "r");
|
|
300
|
4473 if (fd == NULL)
|
|
236
|
4474 {
|
|
300
|
4475 EMSG2(_(e_notopen), fname);
|
|
|
4476 return FAIL;
|
|
236
|
4477 }
|
|
|
4478
|
|
310
|
4479 if (spin->si_verbose || p_verbose > 2)
|
|
|
4480 {
|
|
|
4481 if (!spin->si_verbose)
|
|
|
4482 verbose_enter();
|
|
366
|
4483 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
4484 out_flush();
|
|
|
4485 if (!spin->si_verbose)
|
|
|
4486 verbose_leave();
|
|
|
4487 }
|
|
300
|
4488
|
|
|
4489 /*
|
|
|
4490 * Read all the lines in the file one by one.
|
|
|
4491 */
|
|
|
4492 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
4493 {
|
|
|
4494 line_breakcheck();
|
|
|
4495 ++lnum;
|
|
|
4496
|
|
|
4497 /* Skip comment lines. */
|
|
|
4498 if (*rline == '#')
|
|
|
4499 continue;
|
|
|
4500
|
|
|
4501 /* Remove CR, LF and white space from the end. */
|
|
|
4502 l = STRLEN(rline);
|
|
|
4503 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
4504 --l;
|
|
|
4505 if (l == 0)
|
|
|
4506 continue; /* empty or blank line */
|
|
|
4507 rline[l] = NUL;
|
|
|
4508
|
|
|
4509 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
4510 vim_free(pc);
|
|
310
|
4511 #ifdef FEAT_MBYTE
|
|
300
|
4512 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
4513 {
|
|
|
4514 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
4515 if (pc == NULL)
|
|
|
4516 {
|
|
|
4517 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4518 fname, lnum, rline);
|
|
|
4519 continue;
|
|
|
4520 }
|
|
|
4521 line = pc;
|
|
|
4522 }
|
|
|
4523 else
|
|
310
|
4524 #endif
|
|
300
|
4525 {
|
|
|
4526 pc = NULL;
|
|
|
4527 line = rline;
|
|
|
4528 }
|
|
|
4529
|
|
307
|
4530 if (*line == '/')
|
|
300
|
4531 {
|
|
307
|
4532 ++line;
|
|
|
4533 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
4534 {
|
|
|
4535 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
4536 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
4537 fname, lnum, line - 1);
|
|
300
|
4538 else if (did_word)
|
|
316
|
4539 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
4540 fname, lnum, line - 1);
|
|
300
|
4541 else
|
|
|
4542 {
|
|
310
|
4543 #ifdef FEAT_MBYTE
|
|
|
4544 char_u *enc;
|
|
|
4545
|
|
300
|
4546 /* Setup for conversion to 'encoding'. */
|
|
316
|
4547 line += 10;
|
|
|
4548 enc = enc_canonize(line);
|
|
300
|
4549 if (enc != NULL && !spin->si_ascii
|
|
|
4550 && convert_setup(&spin->si_conv, enc,
|
|
|
4551 p_enc) == FAIL)
|
|
|
4552 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
4553 fname, line, p_enc);
|
|
300
|
4554 vim_free(enc);
|
|
375
|
4555 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4556 #else
|
|
|
4557 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4558 #endif
|
|
300
|
4559 }
|
|
307
|
4560 continue;
|
|
300
|
4561 }
|
|
307
|
4562
|
|
316
|
4563 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
4564 {
|
|
|
4565 if (spin->si_region_count > 1)
|
|
|
4566 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
4567 fname, lnum, line);
|
|
|
4568 else
|
|
|
4569 {
|
|
|
4570 line += 8;
|
|
|
4571 if (STRLEN(line) > 16)
|
|
|
4572 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
4573 fname, lnum, line);
|
|
|
4574 else
|
|
|
4575 {
|
|
|
4576 spin->si_region_count = STRLEN(line) / 2;
|
|
|
4577 STRCPY(spin->si_region_name, line);
|
|
389
|
4578
|
|
|
4579 /* Adjust the mask for a word valid in all regions. */
|
|
|
4580 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
4581 }
|
|
|
4582 }
|
|
|
4583 continue;
|
|
|
4584 }
|
|
|
4585
|
|
381
|
4586 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
4587 fname, lnum, line - 1);
|
|
|
4588 continue;
|
|
|
4589 }
|
|
|
4590
|
|
|
4591 flags = 0;
|
|
|
4592 regionmask = spin->si_region;
|
|
|
4593
|
|
|
4594 /* Check for flags and region after a slash. */
|
|
|
4595 p = vim_strchr(line, '/');
|
|
|
4596 if (p != NULL)
|
|
|
4597 {
|
|
|
4598 *p++ = NUL;
|
|
|
4599 while (*p != NUL)
|
|
|
4600 {
|
|
|
4601 if (*p == '=') /* keep-case word */
|
|
389
|
4602 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
4603 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
4604 flags |= WF_BANNED;
|
|
|
4605 else if (*p == '?') /* Rare word. */
|
|
|
4606 flags |= WF_RARE;
|
|
|
4607 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
4608 {
|
|
381
|
4609 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
4610 regionmask = 0;
|
|
|
4611 flags |= WF_REGION;
|
|
|
4612
|
|
|
4613 l = *p - '0';
|
|
316
|
4614 if (l > spin->si_region_count)
|
|
|
4615 {
|
|
|
4616 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
4617 fname, lnum, p);
|
|
316
|
4618 break;
|
|
|
4619 }
|
|
|
4620 regionmask |= 1 << (l - 1);
|
|
|
4621 }
|
|
381
|
4622 else
|
|
|
4623 {
|
|
|
4624 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
4625 fname, lnum, p);
|
|
|
4626 break;
|
|
|
4627 }
|
|
|
4628 ++p;
|
|
307
|
4629 }
|
|
300
|
4630 }
|
|
|
4631
|
|
|
4632 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4633 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4634 {
|
|
|
4635 ++non_ascii;
|
|
|
4636 continue;
|
|
|
4637 }
|
|
|
4638
|
|
|
4639 /* Normal word: store it. */
|
|
474
|
4640 if (store_word(spin, line, flags, regionmask, NULL) == FAIL)
|
|
300
|
4641 {
|
|
|
4642 retval = FAIL;
|
|
|
4643 break;
|
|
|
4644 }
|
|
|
4645 did_word = TRUE;
|
|
|
4646 }
|
|
|
4647
|
|
|
4648 vim_free(pc);
|
|
|
4649 fclose(fd);
|
|
|
4650
|
|
310
|
4651 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
4652 {
|
|
|
4653 if (p_verbose > 2)
|
|
|
4654 verbose_enter();
|
|
300
|
4655 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4656 non_ascii);
|
|
310
|
4657 if (p_verbose > 2)
|
|
|
4658 verbose_leave();
|
|
|
4659 }
|
|
300
|
4660 return retval;
|
|
236
|
4661 }
|
|
|
4662
|
|
|
4663 /*
|
|
300
|
4664 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
4665 * This avoids calling free() for every little struct we use (and keeping
|
|
|
4666 * track of them).
|
|
300
|
4667 * The memory is cleared to all zeros.
|
|
|
4668 * Returns NULL when out of memory.
|
|
|
4669 */
|
|
|
4670 static void *
|
|
474
|
4671 getroom(spin, len, align)
|
|
|
4672 spellinfo_T *spin;
|
|
447
|
4673 size_t len; /* length needed */
|
|
|
4674 int align; /* align for pointer */
|
|
300
|
4675 {
|
|
|
4676 char_u *p;
|
|
474
|
4677 sblock_T *bl = spin->si_blocks;
|
|
300
|
4678
|
|
447
|
4679 if (align && bl != NULL)
|
|
|
4680 /* Round size up for alignment. On some systems structures need to be
|
|
|
4681 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
4682 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
4683 & ~(sizeof(char *) - 1);
|
|
|
4684
|
|
300
|
4685 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
4686 {
|
|
|
4687 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
4688 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
4689 if (bl == NULL)
|
|
|
4690 return NULL;
|
|
474
|
4691 bl->sb_next = spin->si_blocks;
|
|
|
4692 spin->si_blocks = bl;
|
|
300
|
4693 bl->sb_used = 0;
|
|
|
4694 }
|
|
|
4695
|
|
|
4696 p = bl->sb_data + bl->sb_used;
|
|
|
4697 bl->sb_used += len;
|
|
|
4698
|
|
|
4699 return p;
|
|
|
4700 }
|
|
|
4701
|
|
|
4702 /*
|
|
|
4703 * Make a copy of a string into memory allocated with getroom().
|
|
|
4704 */
|
|
|
4705 static char_u *
|
|
474
|
4706 getroom_save(spin, s)
|
|
|
4707 spellinfo_T *spin;
|
|
300
|
4708 char_u *s;
|
|
|
4709 {
|
|
|
4710 char_u *sc;
|
|
|
4711
|
|
474
|
4712 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
4713 if (sc != NULL)
|
|
|
4714 STRCPY(sc, s);
|
|
|
4715 return sc;
|
|
|
4716 }
|
|
|
4717
|
|
|
4718
|
|
|
4719 /*
|
|
|
4720 * Free the list of allocated sblock_T.
|
|
236
|
4721 */
|
|
|
4722 static void
|
|
300
|
4723 free_blocks(bl)
|
|
|
4724 sblock_T *bl;
|
|
236
|
4725 {
|
|
300
|
4726 sblock_T *next;
|
|
236
|
4727
|
|
300
|
4728 while (bl != NULL)
|
|
236
|
4729 {
|
|
300
|
4730 next = bl->sb_next;
|
|
|
4731 vim_free(bl);
|
|
|
4732 bl = next;
|
|
236
|
4733 }
|
|
|
4734 }
|
|
|
4735
|
|
|
4736 /*
|
|
300
|
4737 * Allocate the root of a word tree.
|
|
236
|
4738 */
|
|
300
|
4739 static wordnode_T *
|
|
474
|
4740 wordtree_alloc(spin)
|
|
|
4741 spellinfo_T *spin;
|
|
|
4742 {
|
|
|
4743 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
4744 }
|
|
|
4745
|
|
|
4746 /*
|
|
300
|
4747 * Store a word in the tree(s).
|
|
390
|
4748 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
4749 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
4750 * used to find suggestions.
|
|
300
|
4751 * For a keep-case word also store it in the keep-case tree.
|
|
390
|
4752 * When "pfxlist" is not NULL store the word for each postponed prefix ID.
|
|
236
|
4753 */
|
|
|
4754 static int
|
|
474
|
4755 store_word(spin, word, flags, region, pfxlist)
|
|
|
4756 spellinfo_T *spin;
|
|
300
|
4757 char_u *word;
|
|
307
|
4758 int flags; /* extra flags, WF_BANNED */
|
|
316
|
4759 int region; /* supported region(s) */
|
|
339
|
4760 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
236
|
4761 {
|
|
300
|
4762 int len = STRLEN(word);
|
|
|
4763 int ct = captype(word, word + len);
|
|
|
4764 char_u foldword[MAXWLEN];
|
|
339
|
4765 int res = OK;
|
|
|
4766 char_u *p;
|
|
236
|
4767
|
|
323
|
4768 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
4769 for (p = pfxlist; res == OK; ++p)
|
|
|
4770 {
|
|
474
|
4771 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
|
4772 region, p == NULL ? 0 : *p);
|
|
339
|
4773 if (p == NULL || *p == NUL)
|
|
|
4774 break;
|
|
|
4775 }
|
|
334
|
4776 ++spin->si_foldwcount;
|
|
307
|
4777
|
|
474
|
4778 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
4779 {
|
|
339
|
4780 for (p = pfxlist; res == OK; ++p)
|
|
|
4781 {
|
|
474
|
4782 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
|
4783 region, p == NULL ? 0 : *p);
|
|
339
|
4784 if (p == NULL || *p == NUL)
|
|
|
4785 break;
|
|
|
4786 }
|
|
334
|
4787 ++spin->si_keepwcount;
|
|
|
4788 }
|
|
300
|
4789 return res;
|
|
236
|
4790 }
|
|
|
4791
|
|
|
4792 /*
|
|
300
|
4793 * Add word "word" to a word tree at "root".
|
|
366
|
4794 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
4795 * "rare" and "region" is the condition nr.
|
|
255
|
4796 * Returns FAIL when out of memory.
|
|
236
|
4797 */
|
|
255
|
4798 static int
|
|
474
|
4799 tree_add_word(spin, word, root, flags, region, prefixID)
|
|
|
4800 spellinfo_T *spin;
|
|
300
|
4801 char_u *word;
|
|
|
4802 wordnode_T *root;
|
|
|
4803 int flags;
|
|
|
4804 int region;
|
|
339
|
4805 int prefixID;
|
|
236
|
4806 {
|
|
300
|
4807 wordnode_T *node = root;
|
|
|
4808 wordnode_T *np;
|
|
469
|
4809 wordnode_T *copyp, **copyprev;
|
|
300
|
4810 wordnode_T **prev = NULL;
|
|
|
4811 int i;
|
|
255
|
4812
|
|
300
|
4813 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
4814 for (i = 0; ; ++i)
|
|
255
|
4815 {
|
|
469
|
4816 /* When there is more than one reference to this node we need to make
|
|
|
4817 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
4818 * (we don't optimize for a partly shared list of siblings). */
|
|
|
4819 if (node != NULL && node->wn_refs > 1)
|
|
|
4820 {
|
|
|
4821 --node->wn_refs;
|
|
|
4822 copyprev = prev;
|
|
|
4823 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
4824 {
|
|
|
4825 /* Allocate a new node and copy the info. */
|
|
474
|
4826 np = get_wordnode(spin);
|
|
469
|
4827 if (np == NULL)
|
|
|
4828 return FAIL;
|
|
|
4829 np->wn_child = copyp->wn_child;
|
|
|
4830 if (np->wn_child != NULL)
|
|
|
4831 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
4832 np->wn_byte = copyp->wn_byte;
|
|
|
4833 if (np->wn_byte == NUL)
|
|
|
4834 {
|
|
|
4835 np->wn_flags = copyp->wn_flags;
|
|
|
4836 np->wn_region = copyp->wn_region;
|
|
|
4837 np->wn_prefixID = copyp->wn_prefixID;
|
|
|
4838 }
|
|
|
4839
|
|
|
4840 /* Link the new node in the list, there will be one ref. */
|
|
|
4841 np->wn_refs = 1;
|
|
|
4842 *copyprev = np;
|
|
|
4843 copyprev = &np->wn_sibling;
|
|
|
4844
|
|
|
4845 /* Let "node" point to the head of the copied list. */
|
|
|
4846 if (copyp == node)
|
|
|
4847 node = np;
|
|
|
4848 }
|
|
|
4849 }
|
|
|
4850
|
|
300
|
4851 /* Look for the sibling that has the same character. They are sorted
|
|
|
4852 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
4853 * higher byte value. For zero bytes (end of word) the sorting is
|
|
469
|
4854 * done on flags and then on prefixID. */
|
|
339
|
4855 while (node != NULL
|
|
|
4856 && (node->wn_byte < word[i]
|
|
|
4857 || (node->wn_byte == NUL
|
|
|
4858 && (flags < 0
|
|
|
4859 ? node->wn_prefixID < prefixID
|
|
390
|
4860 : node->wn_flags < (flags & WN_MASK)
|
|
|
4861 || (node->wn_flags == (flags & WN_MASK)
|
|
339
|
4862 && node->wn_prefixID < prefixID)))))
|
|
236
|
4863 {
|
|
300
|
4864 prev = &node->wn_sibling;
|
|
|
4865 node = *prev;
|
|
236
|
4866 }
|
|
339
|
4867 if (node == NULL
|
|
|
4868 || node->wn_byte != word[i]
|
|
|
4869 || (word[i] == NUL
|
|
|
4870 && (flags < 0
|
|
390
|
4871 || node->wn_flags != (flags & WN_MASK)
|
|
339
|
4872 || node->wn_prefixID != prefixID)))
|
|
255
|
4873 {
|
|
300
|
4874 /* Allocate a new node. */
|
|
474
|
4875 np = get_wordnode(spin);
|
|
300
|
4876 if (np == NULL)
|
|
|
4877 return FAIL;
|
|
|
4878 np->wn_byte = word[i];
|
|
469
|
4879
|
|
|
4880 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
4881 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
4882 * make ref count of sibling one (matters when inserting in front
|
|
|
4883 * of the list of siblings). */
|
|
|
4884 if (node == NULL)
|
|
|
4885 np->wn_refs = 1;
|
|
|
4886 else
|
|
|
4887 {
|
|
|
4888 np->wn_refs = node->wn_refs;
|
|
|
4889 node->wn_refs = 1;
|
|
|
4890 }
|
|
300
|
4891 *prev = np;
|
|
|
4892 np->wn_sibling = node;
|
|
|
4893 node = np;
|
|
255
|
4894 }
|
|
300
|
4895
|
|
|
4896 if (word[i] == NUL)
|
|
|
4897 {
|
|
|
4898 node->wn_flags = flags;
|
|
|
4899 node->wn_region |= region;
|
|
339
|
4900 node->wn_prefixID = prefixID;
|
|
300
|
4901 break;
|
|
|
4902 }
|
|
|
4903 prev = &node->wn_child;
|
|
|
4904 node = *prev;
|
|
255
|
4905 }
|
|
469
|
4906 #ifdef SPELL_PRINTTREE
|
|
|
4907 smsg("Added \"%s\"", word);
|
|
|
4908 spell_print_tree(root->wn_sibling);
|
|
|
4909 #endif
|
|
|
4910
|
|
474
|
4911 /* count nr of words added since last message */
|
|
|
4912 ++spin->si_msg_count;
|
|
|
4913
|
|
469
|
4914 /*
|
|
|
4915 * Every so many words compress the tree, so that we don't use too much
|
|
|
4916 * memory.
|
|
|
4917 */
|
|
474
|
4918 i = FALSE;
|
|
|
4919 if (root == spin->si_foldroot)
|
|
|
4920 {
|
|
|
4921 if (++spin->si_fold_added >= SPELL_COMPRESS_CNT)
|
|
|
4922 {
|
|
|
4923 i = TRUE;
|
|
|
4924 spin->si_fold_added = 0;
|
|
|
4925 }
|
|
|
4926 }
|
|
|
4927 else if (root == spin->si_keeproot)
|
|
|
4928 {
|
|
|
4929 if (++spin->si_keep_added >= SPELL_COMPRESS_CNT)
|
|
|
4930 {
|
|
|
4931 i = TRUE;
|
|
|
4932 spin->si_keep_added = 0;
|
|
|
4933 }
|
|
|
4934 }
|
|
|
4935 if (i)
|
|
|
4936 {
|
|
|
4937 if (spin->si_verbose)
|
|
|
4938 {
|
|
|
4939 msg_start();
|
|
|
4940 msg_puts((char_u *)_(msg_compressing));
|
|
|
4941 msg_clr_eos();
|
|
|
4942 msg_didout = FALSE;
|
|
|
4943 msg_col = 0;
|
|
|
4944 out_flush();
|
|
|
4945 }
|
|
|
4946 wordtree_compress(spin, root);
|
|
469
|
4947 }
|
|
255
|
4948
|
|
|
4949 return OK;
|
|
236
|
4950 }
|
|
|
4951
|
|
469
|
4952 /*
|
|
|
4953 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
4954 * allocate a new one.
|
|
|
4955 */
|
|
|
4956 static wordnode_T *
|
|
474
|
4957 get_wordnode(spin)
|
|
|
4958 spellinfo_T *spin;
|
|
469
|
4959 {
|
|
|
4960 wordnode_T *n;
|
|
|
4961
|
|
474
|
4962 if (spin->si_first_free == NULL)
|
|
|
4963 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
4964 else
|
|
|
4965 {
|
|
474
|
4966 n = spin->si_first_free;
|
|
|
4967 spin->si_first_free = n->wn_child;
|
|
469
|
4968 vim_memset(n, 0, sizeof(wordnode_T));
|
|
|
4969 }
|
|
|
4970 #ifdef SPELL_PRINTTREE
|
|
474
|
4971 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
4972 #endif
|
|
|
4973 return n;
|
|
|
4974 }
|
|
|
4975
|
|
|
4976 /*
|
|
474
|
4977 * Decrement the reference count on a node (which is the head of a list of
|
|
|
4978 * siblings). If the reference count becomes zero free the node and its
|
|
|
4979 * siblings.
|
|
|
4980 */
|
|
|
4981 static void
|
|
|
4982 deref_wordnode(spin, node)
|
|
|
4983 spellinfo_T *spin;
|
|
|
4984 wordnode_T *node;
|
|
|
4985 {
|
|
|
4986 wordnode_T *np;
|
|
|
4987
|
|
|
4988 if (--node->wn_refs == 0)
|
|
|
4989 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4990 {
|
|
|
4991 if (np->wn_child != NULL)
|
|
|
4992 deref_wordnode(spin, np->wn_child);
|
|
|
4993 free_wordnode(spin, np);
|
|
|
4994 }
|
|
|
4995 }
|
|
|
4996
|
|
|
4997 /*
|
|
469
|
4998 * Free a wordnode_T for re-use later.
|
|
474
|
4999 * Only the "wn_child" field becomes invalid.
|
|
469
|
5000 */
|
|
|
5001 static void
|
|
474
|
5002 free_wordnode(spin, n)
|
|
|
5003 spellinfo_T *spin;
|
|
|
5004 wordnode_T *n;
|
|
|
5005 {
|
|
|
5006 n->wn_child = spin->si_first_free;
|
|
|
5007 spin->si_first_free = n;
|
|
469
|
5008 }
|
|
|
5009
|
|
236
|
5010 /*
|
|
300
|
5011 * Compress a tree: find tails that are identical and can be shared.
|
|
|
5012 */
|
|
|
5013 static void
|
|
474
|
5014 wordtree_compress(spin, root)
|
|
|
5015 spellinfo_T *spin;
|
|
300
|
5016 wordnode_T *root;
|
|
|
5017 {
|
|
|
5018 hashtab_T ht;
|
|
|
5019 int n;
|
|
|
5020 int tot = 0;
|
|
469
|
5021 int perc;
|
|
300
|
5022
|
|
474
|
5023 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
5024 * start of the tree. */
|
|
|
5025 if (root->wn_sibling != NULL)
|
|
300
|
5026 {
|
|
|
5027 hash_init(&ht);
|
|
474
|
5028 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
5029
|
|
|
5030 #ifndef SPELL_PRINTTREE
|
|
310
|
5031 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
5032 #endif
|
|
310
|
5033 {
|
|
|
5034 if (!spin->si_verbose)
|
|
|
5035 verbose_enter();
|
|
469
|
5036 if (tot > 1000000)
|
|
|
5037 perc = (tot - n) / (tot / 100);
|
|
|
5038 else
|
|
|
5039 perc = (tot - n) * 100 / tot;
|
|
310
|
5040 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
5041 n, tot, perc);
|
|
310
|
5042 if (p_verbose > 2)
|
|
|
5043 verbose_leave();
|
|
|
5044 }
|
|
469
|
5045 #ifdef SPELL_PRINTTREE
|
|
474
|
5046 spell_print_tree(root->wn_sibling);
|
|
469
|
5047 #endif
|
|
300
|
5048 hash_clear(&ht);
|
|
|
5049 }
|
|
|
5050 }
|
|
|
5051
|
|
|
5052 /*
|
|
|
5053 * Compress a node, its siblings and its children, depth first.
|
|
|
5054 * Returns the number of compressed nodes.
|
|
236
|
5055 */
|
|
255
|
5056 static int
|
|
474
|
5057 node_compress(spin, node, ht, tot)
|
|
|
5058 spellinfo_T *spin;
|
|
300
|
5059 wordnode_T *node;
|
|
|
5060 hashtab_T *ht;
|
|
|
5061 int *tot; /* total count of nodes before compressing,
|
|
|
5062 incremented while going through the tree */
|
|
236
|
5063 {
|
|
300
|
5064 wordnode_T *np;
|
|
|
5065 wordnode_T *tp;
|
|
|
5066 wordnode_T *child;
|
|
|
5067 hash_T hash;
|
|
236
|
5068 hashitem_T *hi;
|
|
300
|
5069 int len = 0;
|
|
|
5070 unsigned nr, n;
|
|
|
5071 int compressed = 0;
|
|
236
|
5072
|
|
300
|
5073 /*
|
|
|
5074 * Go through the list of siblings. Compress each child and then try
|
|
|
5075 * finding an identical child to replace it.
|
|
|
5076 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
5077 * but the whole list of siblings, of which the node is the first.
|
|
|
5078 */
|
|
|
5079 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5080 {
|
|
300
|
5081 ++len;
|
|
|
5082 if ((child = np->wn_child) != NULL)
|
|
|
5083 {
|
|
346
|
5084 /* Compress the child. This fills hashkey. */
|
|
474
|
5085 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
5086
|
|
|
5087 /* Try to find an identical child. */
|
|
346
|
5088 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
5089 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
5090 tp = NULL;
|
|
|
5091 if (!HASHITEM_EMPTY(hi))
|
|
|
5092 {
|
|
|
5093 /* There are children with an identical hash value. Now check
|
|
|
5094 * if there is one that is really identical. */
|
|
346
|
5095 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
5096 if (node_equal(child, tp))
|
|
|
5097 {
|
|
|
5098 /* Found one! Now use that child in place of the
|
|
469
|
5099 * current one. This means the current child and all
|
|
|
5100 * its siblings is unlinked from the tree. */
|
|
|
5101 ++tp->wn_refs;
|
|
474
|
5102 deref_wordnode(spin, child);
|
|
300
|
5103 np->wn_child = tp;
|
|
|
5104 ++compressed;
|
|
|
5105 break;
|
|
|
5106 }
|
|
|
5107 if (tp == NULL)
|
|
|
5108 {
|
|
|
5109 /* No other child with this hash value equals the child of
|
|
|
5110 * the node, add it to the linked list after the first
|
|
|
5111 * item. */
|
|
|
5112 tp = HI2WN(hi);
|
|
346
|
5113 child->wn_u2.next = tp->wn_u2.next;
|
|
|
5114 tp->wn_u2.next = child;
|
|
300
|
5115 }
|
|
|
5116 }
|
|
|
5117 else
|
|
|
5118 /* No other child has this hash value, add it to the
|
|
|
5119 * hashtable. */
|
|
346
|
5120 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
5121 }
|
|
236
|
5122 }
|
|
300
|
5123 *tot += len;
|
|
|
5124
|
|
|
5125 /*
|
|
|
5126 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
5127 * find a lookalike node. This must be done after compressing the sibling
|
|
|
5128 * list, otherwise the hash key would become invalid by the compression.
|
|
|
5129 */
|
|
346
|
5130 node->wn_u1.hashkey[0] = len;
|
|
300
|
5131 nr = 0;
|
|
|
5132 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5133 {
|
|
300
|
5134 if (np->wn_byte == NUL)
|
|
339
|
5135 /* end node: use wn_flags, wn_region and wn_prefixID */
|
|
|
5136 n = np->wn_flags + (np->wn_region << 8) + (np->wn_prefixID << 16);
|
|
300
|
5137 else
|
|
|
5138 /* byte node: use the byte value and the child pointer */
|
|
|
5139 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
5140 nr = nr * 101 + n;
|
|
236
|
5141 }
|
|
300
|
5142
|
|
|
5143 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
5144 n = nr & 0xff;
|
|
346
|
5145 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
5146 n = (nr >> 8) & 0xff;
|
|
346
|
5147 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
5148 n = (nr >> 16) & 0xff;
|
|
346
|
5149 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
5150 n = (nr >> 24) & 0xff;
|
|
346
|
5151 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
5152 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
5153
|
|
|
5154 return compressed;
|
|
|
5155 }
|
|
|
5156
|
|
|
5157 /*
|
|
|
5158 * Return TRUE when two nodes have identical siblings and children.
|
|
|
5159 */
|
|
|
5160 static int
|
|
|
5161 node_equal(n1, n2)
|
|
|
5162 wordnode_T *n1;
|
|
|
5163 wordnode_T *n2;
|
|
|
5164 {
|
|
|
5165 wordnode_T *p1;
|
|
|
5166 wordnode_T *p2;
|
|
|
5167
|
|
|
5168 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
5169 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
5170 if (p1->wn_byte != p2->wn_byte
|
|
|
5171 || (p1->wn_byte == NUL
|
|
|
5172 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
5173 || p1->wn_region != p2->wn_region
|
|
|
5174 || p1->wn_prefixID != p2->wn_prefixID)
|
|
300
|
5175 : (p1->wn_child != p2->wn_child)))
|
|
|
5176 break;
|
|
|
5177
|
|
|
5178 return p1 == NULL && p2 == NULL;
|
|
236
|
5179 }
|
|
|
5180
|
|
|
5181 /*
|
|
|
5182 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
5183 */
|
|
255
|
5184 void
|
|
236
|
5185 put_bytes(fd, nr, len)
|
|
|
5186 FILE *fd;
|
|
|
5187 long_u nr;
|
|
|
5188 int len;
|
|
|
5189 {
|
|
|
5190 int i;
|
|
|
5191
|
|
|
5192 for (i = len - 1; i >= 0; --i)
|
|
|
5193 putc((int)(nr >> (i * 8)), fd);
|
|
|
5194 }
|
|
|
5195
|
|
323
|
5196 static int
|
|
|
5197 #ifdef __BORLANDC__
|
|
|
5198 _RTLENTRYF
|
|
|
5199 #endif
|
|
|
5200 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
5201
|
|
|
5202 /*
|
|
|
5203 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
5204 */
|
|
|
5205 static int
|
|
|
5206 #ifdef __BORLANDC__
|
|
|
5207 _RTLENTRYF
|
|
|
5208 #endif
|
|
|
5209 rep_compare(s1, s2)
|
|
|
5210 const void *s1;
|
|
|
5211 const void *s2;
|
|
|
5212 {
|
|
|
5213 fromto_T *p1 = (fromto_T *)s1;
|
|
|
5214 fromto_T *p2 = (fromto_T *)s2;
|
|
|
5215
|
|
|
5216 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
5217 }
|
|
|
5218
|
|
236
|
5219 /*
|
|
|
5220 * Write the Vim spell file "fname".
|
|
|
5221 */
|
|
|
5222 static void
|
|
474
|
5223 write_vim_spell(spin, fname)
|
|
|
5224 spellinfo_T *spin;
|
|
236
|
5225 char_u *fname;
|
|
|
5226 {
|
|
300
|
5227 FILE *fd;
|
|
|
5228 int regionmask;
|
|
236
|
5229 int round;
|
|
300
|
5230 wordnode_T *tree;
|
|
|
5231 int nodecount;
|
|
323
|
5232 int i;
|
|
|
5233 int l;
|
|
|
5234 garray_T *gap;
|
|
|
5235 fromto_T *ftp;
|
|
|
5236 char_u *p;
|
|
|
5237 int rr;
|
|
236
|
5238
|
|
310
|
5239 fd = mch_fopen((char *)fname, "w");
|
|
300
|
5240 if (fd == NULL)
|
|
236
|
5241 {
|
|
|
5242 EMSG2(_(e_notopen), fname);
|
|
|
5243 return;
|
|
|
5244 }
|
|
|
5245
|
|
255
|
5246 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
339
|
5247 * <charflagslen> <charflags>
|
|
|
5248 * <fcharslen> <fchars>
|
|
366
|
5249 * <midwordlen> <midword>
|
|
339
|
5250 * <prefcondcnt> <prefcond> ... */
|
|
300
|
5251
|
|
|
5252 /* <fileID> */
|
|
|
5253 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
5254 EMSG(_(e_write));
|
|
236
|
5255
|
|
|
5256 /* write the region names if there is more than one */
|
|
316
|
5257 if (spin->si_region_count > 1)
|
|
236
|
5258 {
|
|
316
|
5259 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
|
5260 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
|
5261 (size_t)1, fd);
|
|
|
5262 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
5263 }
|
|
|
5264 else
|
|
|
5265 {
|
|
300
|
5266 putc(0, fd);
|
|
|
5267 regionmask = 0;
|
|
236
|
5268 }
|
|
|
5269
|
|
323
|
5270 /*
|
|
|
5271 * Write the table with character flags and table for case folding.
|
|
260
|
5272 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
|
5273 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
5274 * 'encoding'.
|
|
|
5275 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
5276 * the table (avoids that it conflicts). File is shorter too.
|
|
|
5277 */
|
|
|
5278 if (spin->si_ascii || spin->si_add)
|
|
260
|
5279 {
|
|
300
|
5280 putc(0, fd);
|
|
|
5281 putc(0, fd);
|
|
|
5282 putc(0, fd);
|
|
260
|
5283 }
|
|
|
5284 else
|
|
300
|
5285 write_spell_chartab(fd);
|
|
255
|
5286
|
|
366
|
5287
|
|
|
5288 if (spin->si_midword == NULL)
|
|
|
5289 put_bytes(fd, 0L, 2); /* <midwordlen> */
|
|
|
5290 else
|
|
|
5291 {
|
|
|
5292 i = STRLEN(spin->si_midword);
|
|
|
5293 put_bytes(fd, (long_u)i, 2); /* <midwordlen> */
|
|
|
5294 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
5295 }
|
|
|
5296
|
|
|
5297
|
|
339
|
5298 /* Write the prefix conditions. */
|
|
|
5299 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
5300
|
|
366
|
5301 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
5302 * <salflags> <salcount> <sal> ...
|
|
|
5303 * <maplen> <mapstr> */
|
|
|
5304
|
|
323
|
5305 /* Sort the REP items. */
|
|
|
5306 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
5307 sizeof(fromto_T), rep_compare);
|
|
|
5308
|
|
375
|
5309 /* round 1: REP items
|
|
|
5310 * round 2: SAL items (unless SOFO is used) */
|
|
323
|
5311 for (round = 1; round <= 2; ++round)
|
|
|
5312 {
|
|
|
5313 if (round == 1)
|
|
|
5314 gap = &spin->si_rep;
|
|
|
5315 else
|
|
|
5316 {
|
|
|
5317 gap = &spin->si_sal;
|
|
|
5318
|
|
|
5319 i = 0;
|
|
|
5320 if (spin->si_followup)
|
|
|
5321 i |= SAL_F0LLOWUP;
|
|
|
5322 if (spin->si_collapse)
|
|
|
5323 i |= SAL_COLLAPSE;
|
|
|
5324 if (spin->si_rem_accents)
|
|
|
5325 i |= SAL_REM_ACCENTS;
|
|
375
|
5326 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5327 i |= SAL_SOFO;
|
|
323
|
5328 putc(i, fd); /* <salflags> */
|
|
375
|
5329 if (i & SAL_SOFO)
|
|
|
5330 break;
|
|
323
|
5331 }
|
|
|
5332
|
|
|
5333 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
5334 for (i = 0; i < gap->ga_len; ++i)
|
|
|
5335 {
|
|
|
5336 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
5337 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
5338 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
5339 for (rr = 1; rr <= 2; ++rr)
|
|
|
5340 {
|
|
|
5341 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
5342 l = STRLEN(p);
|
|
|
5343 putc(l, fd);
|
|
|
5344 fwrite(p, l, (size_t)1, fd);
|
|
|
5345 }
|
|
|
5346 }
|
|
|
5347 }
|
|
|
5348
|
|
375
|
5349 /* SOFOFROM and SOFOTO */
|
|
|
5350 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5351 {
|
|
|
5352 put_bytes(fd, 1L, 2); /* <salcount> */
|
|
|
5353
|
|
|
5354 l = STRLEN(spin->si_sofofr);
|
|
|
5355 put_bytes(fd, (long_u)l, 2); /* <salfromlen> */
|
|
|
5356 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <salfrom> */
|
|
|
5357
|
|
|
5358 l = STRLEN(spin->si_sofoto);
|
|
|
5359 put_bytes(fd, (long_u)l, 2); /* <saltolen> */
|
|
|
5360 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <salto> */
|
|
|
5361 }
|
|
|
5362
|
|
323
|
5363 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
|
5364 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
|
5365 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
|
5366 (size_t)1, fd);
|
|
302
|
5367
|
|
236
|
5368 /*
|
|
339
|
5369 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
5370 */
|
|
323
|
5371 spin->si_memtot = 0;
|
|
339
|
5372 for (round = 1; round <= 3; ++round)
|
|
236
|
5373 {
|
|
339
|
5374 if (round == 1)
|
|
469
|
5375 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
5376 else if (round == 2)
|
|
469
|
5377 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
5378 else
|
|
469
|
5379 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
5380
|
|
346
|
5381 /* Clear the index and wnode fields in the tree. */
|
|
|
5382 clear_node(tree);
|
|
|
5383
|
|
300
|
5384 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
5385 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
5386 * nodes. */
|
|
346
|
5387 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
5388
|
|
300
|
5389 /* number of nodes in 4 bytes */
|
|
|
5390 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
5391 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
5392
|
|
300
|
5393 /* Write the nodes. */
|
|
346
|
5394 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
5395 }
|
|
|
5396
|
|
300
|
5397 fclose(fd);
|
|
236
|
5398 }
|
|
|
5399
|
|
|
5400 /*
|
|
346
|
5401 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
5402 * children. This is needed because they are a union with other items to save
|
|
|
5403 * space.
|
|
|
5404 */
|
|
|
5405 static void
|
|
|
5406 clear_node(node)
|
|
|
5407 wordnode_T *node;
|
|
|
5408 {
|
|
|
5409 wordnode_T *np;
|
|
|
5410
|
|
|
5411 if (node != NULL)
|
|
|
5412 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5413 {
|
|
|
5414 np->wn_u1.index = 0;
|
|
|
5415 np->wn_u2.wnode = NULL;
|
|
|
5416
|
|
|
5417 if (np->wn_byte != NUL)
|
|
|
5418 clear_node(np->wn_child);
|
|
|
5419 }
|
|
|
5420 }
|
|
|
5421
|
|
|
5422
|
|
|
5423 /*
|
|
300
|
5424 * Dump a word tree at node "node".
|
|
|
5425 *
|
|
|
5426 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
5427 * byte recursively write the children.
|
|
|
5428 *
|
|
|
5429 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
5430 * are made about the indexes (so that we don't have to write them in the
|
|
|
5431 * file).
|
|
236
|
5432 *
|
|
300
|
5433 * Returns the number of nodes used.
|
|
236
|
5434 */
|
|
300
|
5435 static int
|
|
346
|
5436 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
5437 FILE *fd; /* NULL when only counting */
|
|
300
|
5438 wordnode_T *node;
|
|
|
5439 int index;
|
|
|
5440 int regionmask;
|
|
339
|
5441 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
5442 {
|
|
300
|
5443 int newindex = index;
|
|
|
5444 int siblingcount = 0;
|
|
|
5445 wordnode_T *np;
|
|
236
|
5446 int flags;
|
|
300
|
5447
|
|
|
5448 /* If "node" is zero the tree is empty. */
|
|
|
5449 if (node == NULL)
|
|
|
5450 return 0;
|
|
|
5451
|
|
|
5452 /* Store the index where this node is written. */
|
|
346
|
5453 node->wn_u1.index = index;
|
|
236
|
5454
|
|
300
|
5455 /* Count the number of siblings. */
|
|
|
5456 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5457 ++siblingcount;
|
|
236
|
5458
|
|
300
|
5459 /* Write the sibling count. */
|
|
|
5460 if (fd != NULL)
|
|
|
5461 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
5462
|
|
300
|
5463 /* Write each sibling byte and optionally extra info. */
|
|
|
5464 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5465 {
|
|
300
|
5466 if (np->wn_byte == 0)
|
|
|
5467 {
|
|
|
5468 if (fd != NULL)
|
|
|
5469 {
|
|
339
|
5470 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
5471 if (prefixtree)
|
|
300
|
5472 {
|
|
339
|
5473 /* In PREFIXTREE write the required prefixID and the
|
|
390
|
5474 * associated condition nr (stored in wn_region). The
|
|
|
5475 * byte value is misused to store the "rare" and "not
|
|
|
5476 * combining" flags */
|
|
455
|
5477 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
5478 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5479 else
|
|
|
5480 {
|
|
390
|
5481 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
5482 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
5483 }
|
|
339
|
5484 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
5485 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
5486 }
|
|
|
5487 else
|
|
|
5488 {
|
|
339
|
5489 /* For word trees we write the flag/region items. */
|
|
|
5490 flags = np->wn_flags;
|
|
|
5491 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
5492 flags |= WF_REGION;
|
|
|
5493 if (np->wn_prefixID != 0)
|
|
|
5494 flags |= WF_PFX;
|
|
|
5495 if (flags == 0)
|
|
|
5496 {
|
|
|
5497 /* word without flags or region */
|
|
|
5498 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5499 }
|
|
|
5500 else
|
|
|
5501 {
|
|
390
|
5502 if (np->wn_flags >= 0x100)
|
|
|
5503 {
|
|
|
5504 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
5505 putc(flags, fd); /* <flags> */
|
|
|
5506 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
5507 }
|
|
|
5508 else
|
|
|
5509 {
|
|
|
5510 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
5511 putc(flags, fd); /* <flags> */
|
|
|
5512 }
|
|
339
|
5513 if (flags & WF_REGION)
|
|
|
5514 putc(np->wn_region, fd); /* <region> */
|
|
|
5515 if (flags & WF_PFX)
|
|
|
5516 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
5517 }
|
|
300
|
5518 }
|
|
|
5519 }
|
|
|
5520 }
|
|
|
5521 else
|
|
|
5522 {
|
|
346
|
5523 if (np->wn_child->wn_u1.index != 0
|
|
|
5524 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
5525 {
|
|
|
5526 /* The child is written elsewhere, write the reference. */
|
|
|
5527 if (fd != NULL)
|
|
|
5528 {
|
|
|
5529 putc(BY_INDEX, fd); /* <byte> */
|
|
|
5530 /* <nodeidx> */
|
|
346
|
5531 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
5532 }
|
|
|
5533 }
|
|
346
|
5534 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
5535 /* We will write the child below and give it an index. */
|
|
346
|
5536 np->wn_child->wn_u2.wnode = node;
|
|
236
|
5537
|
|
300
|
5538 if (fd != NULL)
|
|
|
5539 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
5540 {
|
|
|
5541 EMSG(_(e_write));
|
|
|
5542 return 0;
|
|
|
5543 }
|
|
|
5544 }
|
|
236
|
5545 }
|
|
|
5546
|
|
300
|
5547 /* Space used in the array when reading: one for each sibling and one for
|
|
|
5548 * the count. */
|
|
|
5549 newindex += siblingcount + 1;
|
|
249
|
5550
|
|
300
|
5551 /* Recursively dump the children of each sibling. */
|
|
|
5552 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
5553 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
5554 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
5555 prefixtree);
|
|
249
|
5556
|
|
300
|
5557 return newindex;
|
|
236
|
5558 }
|
|
|
5559
|
|
|
5560
|
|
|
5561 /*
|
|
310
|
5562 * ":mkspell [-ascii] outfile infile ..."
|
|
|
5563 * ":mkspell [-ascii] addfile"
|
|
236
|
5564 */
|
|
|
5565 void
|
|
|
5566 ex_mkspell(eap)
|
|
|
5567 exarg_T *eap;
|
|
|
5568 {
|
|
|
5569 int fcount;
|
|
|
5570 char_u **fnames;
|
|
310
|
5571 char_u *arg = eap->arg;
|
|
|
5572 int ascii = FALSE;
|
|
|
5573
|
|
|
5574 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
5575 {
|
|
|
5576 ascii = TRUE;
|
|
|
5577 arg = skipwhite(arg + 6);
|
|
|
5578 }
|
|
|
5579
|
|
|
5580 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
5581 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
5582 {
|
|
323
|
5583 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
5584 FreeWild(fcount, fnames);
|
|
|
5585 }
|
|
|
5586 }
|
|
|
5587
|
|
|
5588 /*
|
|
|
5589 * Create a Vim spell file from one or more word lists.
|
|
|
5590 * "fnames[0]" is the output file name.
|
|
|
5591 * "fnames[fcount - 1]" is the last input file name.
|
|
|
5592 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
5593 * and ".spl" is appended to make the output file name.
|
|
|
5594 */
|
|
|
5595 static void
|
|
323
|
5596 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
5597 int fcount;
|
|
|
5598 char_u **fnames;
|
|
|
5599 int ascii; /* -ascii argument given */
|
|
|
5600 int overwrite; /* overwrite existing output file */
|
|
323
|
5601 int added_word; /* invoked through "zg" */
|
|
310
|
5602 {
|
|
236
|
5603 char_u fname[MAXPATHL];
|
|
|
5604 char_u wfname[MAXPATHL];
|
|
310
|
5605 char_u **innames;
|
|
|
5606 int incount;
|
|
236
|
5607 afffile_T *(afile[8]);
|
|
|
5608 int i;
|
|
|
5609 int len;
|
|
|
5610 struct stat st;
|
|
255
|
5611 int error = FALSE;
|
|
300
|
5612 spellinfo_T spin;
|
|
|
5613
|
|
|
5614 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
5615 spin.si_verbose = !added_word;
|
|
310
|
5616 spin.si_ascii = ascii;
|
|
323
|
5617 spin.si_followup = TRUE;
|
|
|
5618 spin.si_rem_accents = TRUE;
|
|
|
5619 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
5620 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
5621 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
5622 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
5623
|
|
|
5624 /* default: fnames[0] is output file, following are input files */
|
|
|
5625 innames = &fnames[1];
|
|
|
5626 incount = fcount - 1;
|
|
|
5627
|
|
|
5628 if (fcount >= 1)
|
|
240
|
5629 {
|
|
310
|
5630 len = STRLEN(fnames[0]);
|
|
|
5631 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
5632 {
|
|
|
5633 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
5634 * "path/en.latin1.add.spl". */
|
|
|
5635 innames = &fnames[0];
|
|
|
5636 incount = 1;
|
|
|
5637 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
5638 }
|
|
366
|
5639 else if (fcount == 1)
|
|
|
5640 {
|
|
|
5641 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
5642 innames = &fnames[0];
|
|
|
5643 incount = 1;
|
|
|
5644 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5645 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5646 }
|
|
310
|
5647 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
5648 {
|
|
|
5649 /* Name ends in ".spl", use as the file name. */
|
|
323
|
5650 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
5651 }
|
|
|
5652 else
|
|
|
5653 /* Name should be language, make the file name from it. */
|
|
|
5654 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5655 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5656
|
|
|
5657 /* Check for .ascii.spl. */
|
|
|
5658 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
5659 spin.si_ascii = TRUE;
|
|
|
5660
|
|
|
5661 /* Check for .add.spl. */
|
|
|
5662 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
5663 spin.si_add = TRUE;
|
|
240
|
5664 }
|
|
|
5665
|
|
310
|
5666 if (incount <= 0)
|
|
236
|
5667 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
5668 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
5669 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
5670 else if (incount > 8)
|
|
236
|
5671 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
5672 else
|
|
|
5673 {
|
|
|
5674 /* Check for overwriting before doing things that may take a lot of
|
|
|
5675 * time. */
|
|
310
|
5676 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
5677 {
|
|
|
5678 EMSG(_(e_exists));
|
|
310
|
5679 return;
|
|
236
|
5680 }
|
|
310
|
5681 if (mch_isdir(wfname))
|
|
236
|
5682 {
|
|
310
|
5683 EMSG2(_(e_isadir2), wfname);
|
|
|
5684 return;
|
|
236
|
5685 }
|
|
|
5686
|
|
|
5687 /*
|
|
|
5688 * Init the aff and dic pointers.
|
|
|
5689 * Get the region names if there are more than 2 arguments.
|
|
|
5690 */
|
|
310
|
5691 for (i = 0; i < incount; ++i)
|
|
236
|
5692 {
|
|
310
|
5693 afile[i] = NULL;
|
|
300
|
5694
|
|
316
|
5695 if (incount > 1)
|
|
236
|
5696 {
|
|
310
|
5697 len = STRLEN(innames[i]);
|
|
|
5698 if (STRLEN(gettail(innames[i])) < 5
|
|
|
5699 || innames[i][len - 3] != '_')
|
|
236
|
5700 {
|
|
310
|
5701 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
5702 return;
|
|
236
|
5703 }
|
|
316
|
5704 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
5705 spin.si_region_name[i * 2 + 1] =
|
|
|
5706 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
5707 }
|
|
|
5708 }
|
|
316
|
5709 spin.si_region_count = incount;
|
|
236
|
5710
|
|
474
|
5711 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
5712 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
5713 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
5714 if (spin.si_foldroot == NULL
|
|
|
5715 || spin.si_keeproot == NULL
|
|
|
5716 || spin.si_prefroot == NULL)
|
|
300
|
5717 {
|
|
469
|
5718 free_blocks(spin.si_blocks);
|
|
310
|
5719 return;
|
|
300
|
5720 }
|
|
|
5721
|
|
351
|
5722 /* When not producing a .add.spl file clear the character table when
|
|
|
5723 * we encounter one in the .aff file. This means we dump the current
|
|
|
5724 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
5725 * better than guessing the contents, the table will match a
|
|
|
5726 * previously loaded spell file. */
|
|
|
5727 if (!spin.si_add)
|
|
|
5728 spin.si_clear_chartab = TRUE;
|
|
|
5729
|
|
236
|
5730 /*
|
|
|
5731 * Read all the .aff and .dic files.
|
|
|
5732 * Text is converted to 'encoding'.
|
|
300
|
5733 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
5734 */
|
|
310
|
5735 for (i = 0; i < incount && !error; ++i)
|
|
236
|
5736 {
|
|
300
|
5737 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
5738 spin.si_region = 1 << i;
|
|
|
5739
|
|
|
5740 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
5741 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
5742 {
|
|
|
5743 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
5744 * "SET" line. */
|
|
474
|
5745 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
5746 if (afile[i] == NULL)
|
|
300
|
5747 error = TRUE;
|
|
|
5748 else
|
|
|
5749 {
|
|
|
5750 /* Read the .dic file and store the words in the trees. */
|
|
|
5751 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
5752 innames[i]);
|
|
474
|
5753 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
5754 error = TRUE;
|
|
|
5755 }
|
|
|
5756 }
|
|
|
5757 else
|
|
|
5758 {
|
|
|
5759 /* No .aff file, try reading the file as a word list. Store
|
|
|
5760 * the words in the trees. */
|
|
474
|
5761 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
5762 error = TRUE;
|
|
|
5763 }
|
|
236
|
5764
|
|
310
|
5765 #ifdef FEAT_MBYTE
|
|
236
|
5766 /* Free any conversion stuff. */
|
|
300
|
5767 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
5768 #endif
|
|
236
|
5769 }
|
|
|
5770
|
|
300
|
5771 if (!error)
|
|
236
|
5772 {
|
|
|
5773 /*
|
|
300
|
5774 * Combine tails in the tree.
|
|
236
|
5775 */
|
|
474
|
5776 if (spin.si_verbose || p_verbose > 2)
|
|
|
5777 {
|
|
|
5778 if (!spin.si_verbose)
|
|
310
|
5779 verbose_enter();
|
|
469
|
5780 MSG(_(msg_compressing));
|
|
310
|
5781 out_flush();
|
|
474
|
5782 if (!spin.si_verbose)
|
|
310
|
5783 verbose_leave();
|
|
|
5784 }
|
|
474
|
5785 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
5786 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
5787 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
5788 }
|
|
|
5789
|
|
300
|
5790 if (!error)
|
|
|
5791 {
|
|
|
5792 /*
|
|
|
5793 * Write the info in the spell file.
|
|
|
5794 */
|
|
474
|
5795 if (spin.si_verbose || p_verbose > 2)
|
|
|
5796 {
|
|
|
5797 if (!spin.si_verbose)
|
|
310
|
5798 verbose_enter();
|
|
366
|
5799 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
5800 out_flush();
|
|
474
|
5801 if (!spin.si_verbose)
|
|
310
|
5802 verbose_leave();
|
|
|
5803 }
|
|
|
5804
|
|
474
|
5805 write_vim_spell(&spin, wfname);
|
|
|
5806
|
|
|
5807 if (spin.si_verbose || p_verbose > 2)
|
|
|
5808 {
|
|
|
5809 if (!spin.si_verbose)
|
|
310
|
5810 verbose_enter();
|
|
|
5811 MSG(_("Done!"));
|
|
|
5812 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
5813 spin.si_memtot);
|
|
310
|
5814 out_flush();
|
|
474
|
5815 if (!spin.si_verbose)
|
|
310
|
5816 verbose_leave();
|
|
|
5817 }
|
|
|
5818
|
|
|
5819 /* If the file is loaded need to reload it. */
|
|
323
|
5820 spell_reload_one(wfname, added_word);
|
|
300
|
5821 }
|
|
|
5822
|
|
|
5823 /* Free the allocated memory. */
|
|
323
|
5824 ga_clear(&spin.si_rep);
|
|
|
5825 ga_clear(&spin.si_sal);
|
|
|
5826 ga_clear(&spin.si_map);
|
|
339
|
5827 ga_clear(&spin.si_prefcond);
|
|
366
|
5828 vim_free(spin.si_midword);
|
|
375
|
5829 vim_free(spin.si_sofofr);
|
|
|
5830 vim_free(spin.si_sofoto);
|
|
300
|
5831
|
|
|
5832 /* Free the .aff file structures. */
|
|
310
|
5833 for (i = 0; i < incount; ++i)
|
|
|
5834 if (afile[i] != NULL)
|
|
|
5835 spell_free_aff(afile[i]);
|
|
339
|
5836
|
|
|
5837 /* Free all the bits and pieces at once. */
|
|
|
5838 free_blocks(spin.si_blocks);
|
|
236
|
5839 }
|
|
310
|
5840 }
|
|
|
5841
|
|
|
5842
|
|
|
5843 /*
|
|
385
|
5844 * ":[count]spellgood {word}"
|
|
|
5845 * ":[count]spellwrong {word}"
|
|
310
|
5846 */
|
|
|
5847 void
|
|
|
5848 ex_spell(eap)
|
|
|
5849 exarg_T *eap;
|
|
|
5850 {
|
|
381
|
5851 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
5852 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
5853 }
|
|
|
5854
|
|
310
|
5855 /*
|
|
|
5856 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
5857 */
|
|
|
5858 void
|
|
385
|
5859 spell_add_word(word, len, bad, index)
|
|
310
|
5860 char_u *word;
|
|
|
5861 int len;
|
|
|
5862 int bad;
|
|
385
|
5863 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
5864 'spellfile' */
|
|
310
|
5865 {
|
|
|
5866 FILE *fd;
|
|
385
|
5867 buf_T *buf = NULL;
|
|
351
|
5868 int new_spf = FALSE;
|
|
|
5869 struct stat st;
|
|
381
|
5870 char_u *fname;
|
|
385
|
5871 char_u fnamebuf[MAXPATHL];
|
|
|
5872 char_u line[MAXWLEN * 2];
|
|
|
5873 long fpos, fpos_next = 0;
|
|
|
5874 int i;
|
|
|
5875 char_u *spf;
|
|
|
5876
|
|
|
5877 if (index == 0) /* use internal wordlist */
|
|
|
5878 {
|
|
|
5879 if (int_wordlist == NULL)
|
|
|
5880 {
|
|
|
5881 int_wordlist = vim_tempname('s');
|
|
|
5882 if (int_wordlist == NULL)
|
|
381
|
5883 return;
|
|
|
5884 }
|
|
385
|
5885 fname = int_wordlist;
|
|
381
|
5886 }
|
|
310
|
5887 else
|
|
|
5888 {
|
|
381
|
5889 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
5890 if (*curbuf->b_p_spf == NUL)
|
|
|
5891 {
|
|
|
5892 init_spellfile();
|
|
|
5893 new_spf = TRUE;
|
|
|
5894 }
|
|
|
5895
|
|
|
5896 if (*curbuf->b_p_spf == NUL)
|
|
|
5897 {
|
|
|
5898 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
5899 return;
|
|
|
5900 }
|
|
|
5901
|
|
385
|
5902 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
5903 {
|
|
|
5904 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
5905 if (i == index)
|
|
|
5906 break;
|
|
|
5907 if (*spf == NUL)
|
|
|
5908 {
|
|
|
5909 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
5910 return;
|
|
|
5911 }
|
|
|
5912 }
|
|
|
5913
|
|
310
|
5914 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
5915 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
5916 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
5917 buf = NULL;
|
|
|
5918 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
5919 {
|
|
310
|
5920 EMSG(_(e_bufloaded));
|
|
381
|
5921 return;
|
|
|
5922 }
|
|
|
5923
|
|
385
|
5924 fname = fnamebuf;
|
|
|
5925 }
|
|
|
5926
|
|
|
5927 if (bad)
|
|
|
5928 {
|
|
|
5929 /* When the word also appears as good word we need to remove that one,
|
|
|
5930 * since its flags sort before the one with WF_BANNED. */
|
|
|
5931 fd = mch_fopen((char *)fname, "r");
|
|
|
5932 if (fd != NULL)
|
|
|
5933 {
|
|
|
5934 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
5935 {
|
|
|
5936 fpos = fpos_next;
|
|
|
5937 fpos_next = ftell(fd);
|
|
|
5938 if (STRNCMP(word, line, len) == 0
|
|
|
5939 && (line[len] == '/' || line[len] < ' '))
|
|
|
5940 {
|
|
|
5941 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
5942 * the start of the line. Mixing reading and writing
|
|
|
5943 * doesn't work for all systems, close the file first. */
|
|
|
5944 fclose(fd);
|
|
|
5945 fd = mch_fopen((char *)fname, "r+");
|
|
|
5946 if (fd == NULL)
|
|
|
5947 break;
|
|
|
5948 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
5949 fputc('#', fd);
|
|
|
5950 fseek(fd, fpos_next, SEEK_SET);
|
|
|
5951 }
|
|
|
5952 }
|
|
|
5953 fclose(fd);
|
|
|
5954 }
|
|
381
|
5955 }
|
|
|
5956
|
|
|
5957 fd = mch_fopen((char *)fname, "a");
|
|
|
5958 if (fd == NULL && new_spf)
|
|
|
5959 {
|
|
|
5960 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
5961 * We may need to create the "spell" directory first. We already
|
|
|
5962 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
5963 STRCPY(NameBuff, fname);
|
|
|
5964 *gettail_sep(NameBuff) = NUL;
|
|
|
5965 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
5966 {
|
|
|
5967 /* The directory doesn't exist. Try creating it and opening the
|
|
|
5968 * file again. */
|
|
|
5969 vim_mkdir(NameBuff, 0755);
|
|
|
5970 fd = mch_fopen((char *)fname, "a");
|
|
|
5971 }
|
|
|
5972 }
|
|
|
5973
|
|
|
5974 if (fd == NULL)
|
|
|
5975 EMSG2(_(e_notopen), fname);
|
|
|
5976 else
|
|
|
5977 {
|
|
|
5978 if (bad)
|
|
|
5979 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
5980 else
|
|
381
|
5981 fprintf(fd, "%.*s\n", len, word);
|
|
|
5982 fclose(fd);
|
|
|
5983
|
|
|
5984 /* Update the .add.spl file. */
|
|
|
5985 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
5986
|
|
|
5987 /* If the .add file is edited somewhere, reload it. */
|
|
|
5988 if (buf != NULL)
|
|
|
5989 buf_reload(buf);
|
|
|
5990
|
|
|
5991 redraw_all_later(NOT_VALID);
|
|
310
|
5992 }
|
|
|
5993 }
|
|
|
5994
|
|
|
5995 /*
|
|
|
5996 * Initialize 'spellfile' for the current buffer.
|
|
|
5997 */
|
|
|
5998 static void
|
|
|
5999 init_spellfile()
|
|
|
6000 {
|
|
|
6001 char_u buf[MAXPATHL];
|
|
|
6002 int l;
|
|
|
6003 slang_T *sl;
|
|
|
6004 char_u *rtp;
|
|
351
|
6005 char_u *lend;
|
|
310
|
6006
|
|
|
6007 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
6008 {
|
|
351
|
6009 /* Find the end of the language name. Exclude the region. */
|
|
|
6010 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
6011 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
6012 ;
|
|
|
6013
|
|
|
6014 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
6015 * are allowed to write. */
|
|
310
|
6016 rtp = p_rtp;
|
|
|
6017 while (*rtp != NUL)
|
|
|
6018 {
|
|
|
6019 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
6020 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
6021 if (filewritable(buf) == 2)
|
|
|
6022 {
|
|
316
|
6023 /* Use the first language name from 'spelllang' and the
|
|
|
6024 * encoding used in the first loaded .spl file. */
|
|
310
|
6025 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
6026 l = STRLEN(buf);
|
|
|
6027 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
6028 "/spell/%.*s.%s.add",
|
|
351
|
6029 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
6030 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
6031 ? (char_u *)"ascii" : spell_enc());
|
|
|
6032 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
6033 break;
|
|
|
6034 }
|
|
|
6035 }
|
|
|
6036 }
|
|
|
6037 }
|
|
236
|
6038
|
|
300
|
6039
|
|
307
|
6040 /*
|
|
|
6041 * Init the chartab used for spelling for ASCII.
|
|
|
6042 * EBCDIC is not supported!
|
|
|
6043 */
|
|
|
6044 static void
|
|
|
6045 clear_spell_chartab(sp)
|
|
|
6046 spelltab_T *sp;
|
|
|
6047 {
|
|
324
|
6048 int i;
|
|
307
|
6049
|
|
|
6050 /* Init everything to FALSE. */
|
|
|
6051 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
6052 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
6053 for (i = 0; i < 256; ++i)
|
|
324
|
6054 {
|
|
307
|
6055 sp->st_fold[i] = i;
|
|
324
|
6056 sp->st_upper[i] = i;
|
|
|
6057 }
|
|
307
|
6058
|
|
|
6059 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
6060 * that is done separately. */
|
|
|
6061 for (i = '0'; i <= '9'; ++i)
|
|
|
6062 sp->st_isw[i] = TRUE;
|
|
|
6063 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
6064 {
|
|
|
6065 sp->st_isw[i] = TRUE;
|
|
|
6066 sp->st_isu[i] = TRUE;
|
|
|
6067 sp->st_fold[i] = i + 0x20;
|
|
|
6068 }
|
|
|
6069 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
6070 {
|
|
307
|
6071 sp->st_isw[i] = TRUE;
|
|
324
|
6072 sp->st_upper[i] = i - 0x20;
|
|
|
6073 }
|
|
307
|
6074 }
|
|
|
6075
|
|
|
6076 /*
|
|
|
6077 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
6078 * Called once while starting up and when 'encoding' changes.
|
|
|
6079 * The default is to use isalpha(), but the spell file should define the word
|
|
|
6080 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
6081 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
6082 */
|
|
|
6083 void
|
|
|
6084 init_spell_chartab()
|
|
|
6085 {
|
|
|
6086 int i;
|
|
|
6087
|
|
|
6088 did_set_spelltab = FALSE;
|
|
|
6089 clear_spell_chartab(&spelltab);
|
|
|
6090 #ifdef FEAT_MBYTE
|
|
|
6091 if (enc_dbcs)
|
|
|
6092 {
|
|
|
6093 /* DBCS: assume double-wide characters are word characters. */
|
|
|
6094 for (i = 128; i <= 255; ++i)
|
|
|
6095 if (MB_BYTE2LEN(i) == 2)
|
|
|
6096 spelltab.st_isw[i] = TRUE;
|
|
|
6097 }
|
|
324
|
6098 else if (enc_utf8)
|
|
|
6099 {
|
|
|
6100 for (i = 128; i < 256; ++i)
|
|
|
6101 {
|
|
|
6102 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
6103 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
6104 spelltab.st_fold[i] = utf_fold(i);
|
|
|
6105 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
6106 }
|
|
|
6107 }
|
|
307
|
6108 else
|
|
|
6109 #endif
|
|
|
6110 {
|
|
324
|
6111 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
6112 for (i = 128; i < 256; ++i)
|
|
|
6113 {
|
|
|
6114 if (MB_ISUPPER(i))
|
|
|
6115 {
|
|
324
|
6116 spelltab.st_isw[i] = TRUE;
|
|
307
|
6117 spelltab.st_isu[i] = TRUE;
|
|
|
6118 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
6119 }
|
|
324
|
6120 else if (MB_ISLOWER(i))
|
|
|
6121 {
|
|
|
6122 spelltab.st_isw[i] = TRUE;
|
|
|
6123 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
6124 }
|
|
307
|
6125 }
|
|
|
6126 }
|
|
|
6127 }
|
|
|
6128
|
|
|
6129 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
6130 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
6131
|
|
|
6132 /*
|
|
|
6133 * Set the spell character tables from strings in the affix file.
|
|
|
6134 */
|
|
|
6135 static int
|
|
|
6136 set_spell_chartab(fol, low, upp)
|
|
|
6137 char_u *fol;
|
|
|
6138 char_u *low;
|
|
|
6139 char_u *upp;
|
|
|
6140 {
|
|
|
6141 /* We build the new tables here first, so that we can compare with the
|
|
|
6142 * previous one. */
|
|
|
6143 spelltab_T new_st;
|
|
|
6144 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
6145 int f, l, u;
|
|
|
6146
|
|
|
6147 clear_spell_chartab(&new_st);
|
|
|
6148
|
|
|
6149 while (*pf != NUL)
|
|
|
6150 {
|
|
|
6151 if (*pl == NUL || *pu == NUL)
|
|
|
6152 {
|
|
|
6153 EMSG(_(e_affform));
|
|
|
6154 return FAIL;
|
|
|
6155 }
|
|
|
6156 #ifdef FEAT_MBYTE
|
|
|
6157 f = mb_ptr2char_adv(&pf);
|
|
|
6158 l = mb_ptr2char_adv(&pl);
|
|
|
6159 u = mb_ptr2char_adv(&pu);
|
|
|
6160 #else
|
|
|
6161 f = *pf++;
|
|
|
6162 l = *pl++;
|
|
|
6163 u = *pu++;
|
|
|
6164 #endif
|
|
|
6165 /* Every character that appears is a word character. */
|
|
|
6166 if (f < 256)
|
|
|
6167 new_st.st_isw[f] = TRUE;
|
|
|
6168 if (l < 256)
|
|
|
6169 new_st.st_isw[l] = TRUE;
|
|
|
6170 if (u < 256)
|
|
|
6171 new_st.st_isw[u] = TRUE;
|
|
|
6172
|
|
|
6173 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
6174 * case-folding */
|
|
|
6175 if (l < 256 && l != f)
|
|
|
6176 {
|
|
|
6177 if (f >= 256)
|
|
|
6178 {
|
|
|
6179 EMSG(_(e_affrange));
|
|
|
6180 return FAIL;
|
|
|
6181 }
|
|
|
6182 new_st.st_fold[l] = f;
|
|
|
6183 }
|
|
|
6184
|
|
|
6185 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
6186 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
6187 * "FOL" . */
|
|
307
|
6188 if (u < 256 && u != f)
|
|
|
6189 {
|
|
|
6190 if (f >= 256)
|
|
|
6191 {
|
|
|
6192 EMSG(_(e_affrange));
|
|
|
6193 return FAIL;
|
|
|
6194 }
|
|
|
6195 new_st.st_fold[u] = f;
|
|
|
6196 new_st.st_isu[u] = TRUE;
|
|
324
|
6197 new_st.st_upper[f] = u;
|
|
307
|
6198 }
|
|
|
6199 }
|
|
|
6200
|
|
|
6201 if (*pl != NUL || *pu != NUL)
|
|
|
6202 {
|
|
|
6203 EMSG(_(e_affform));
|
|
|
6204 return FAIL;
|
|
|
6205 }
|
|
|
6206
|
|
|
6207 return set_spell_finish(&new_st);
|
|
|
6208 }
|
|
|
6209
|
|
|
6210 /*
|
|
|
6211 * Set the spell character tables from strings in the .spl file.
|
|
|
6212 */
|
|
|
6213 static int
|
|
389
|
6214 set_spell_charflags(flags, cnt, fol)
|
|
307
|
6215 char_u *flags;
|
|
389
|
6216 int cnt; /* length of "flags" */
|
|
|
6217 char_u *fol;
|
|
307
|
6218 {
|
|
|
6219 /* We build the new tables here first, so that we can compare with the
|
|
|
6220 * previous one. */
|
|
|
6221 spelltab_T new_st;
|
|
|
6222 int i;
|
|
389
|
6223 char_u *p = fol;
|
|
324
|
6224 int c;
|
|
307
|
6225
|
|
|
6226 clear_spell_chartab(&new_st);
|
|
|
6227
|
|
389
|
6228 for (i = 0; i < 128; ++i)
|
|
|
6229 {
|
|
|
6230 if (i < cnt)
|
|
|
6231 {
|
|
|
6232 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
6233 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
6234 }
|
|
|
6235
|
|
|
6236 if (*p != NUL)
|
|
|
6237 {
|
|
307
|
6238 #ifdef FEAT_MBYTE
|
|
389
|
6239 c = mb_ptr2char_adv(&p);
|
|
307
|
6240 #else
|
|
389
|
6241 c = *p++;
|
|
307
|
6242 #endif
|
|
389
|
6243 new_st.st_fold[i + 128] = c;
|
|
|
6244 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
6245 new_st.st_upper[c] = i + 128;
|
|
|
6246 }
|
|
307
|
6247 }
|
|
|
6248
|
|
|
6249 return set_spell_finish(&new_st);
|
|
|
6250 }
|
|
|
6251
|
|
|
6252 static int
|
|
|
6253 set_spell_finish(new_st)
|
|
|
6254 spelltab_T *new_st;
|
|
|
6255 {
|
|
|
6256 int i;
|
|
|
6257
|
|
|
6258 if (did_set_spelltab)
|
|
|
6259 {
|
|
|
6260 /* check that it's the same table */
|
|
|
6261 for (i = 0; i < 256; ++i)
|
|
|
6262 {
|
|
|
6263 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
6264 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
6265 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
6266 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
6267 {
|
|
|
6268 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
6269 return FAIL;
|
|
|
6270 }
|
|
|
6271 }
|
|
|
6272 }
|
|
|
6273 else
|
|
|
6274 {
|
|
|
6275 /* copy the new spelltab into the one being used */
|
|
|
6276 spelltab = *new_st;
|
|
|
6277 did_set_spelltab = TRUE;
|
|
|
6278 }
|
|
|
6279
|
|
|
6280 return OK;
|
|
|
6281 }
|
|
|
6282
|
|
|
6283 /*
|
|
358
|
6284 * Return TRUE if "p" points to a word character.
|
|
366
|
6285 * As a special case we see "midword" characters as word character when it is
|
|
358
|
6286 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
6287 * Thus this only works properly when past the first character of the word.
|
|
358
|
6288 */
|
|
|
6289 static int
|
|
376
|
6290 spell_iswordp(p, buf)
|
|
358
|
6291 char_u *p;
|
|
376
|
6292 buf_T *buf; /* buffer used */
|
|
358
|
6293 {
|
|
366
|
6294 #ifdef FEAT_MBYTE
|
|
358
|
6295 char_u *s;
|
|
366
|
6296 int l;
|
|
|
6297 int c;
|
|
|
6298
|
|
|
6299 if (has_mbyte)
|
|
|
6300 {
|
|
|
6301 l = MB_BYTE2LEN(*p);
|
|
358
|
6302 s = p;
|
|
366
|
6303 if (l == 1)
|
|
|
6304 {
|
|
|
6305 /* be quick for ASCII */
|
|
376
|
6306 if (buf->b_spell_ismw[*p])
|
|
366
|
6307 {
|
|
|
6308 s = p + 1; /* skip a mid-word character */
|
|
|
6309 l = MB_BYTE2LEN(*s);
|
|
|
6310 }
|
|
|
6311 }
|
|
|
6312 else
|
|
|
6313 {
|
|
|
6314 c = mb_ptr2char(p);
|
|
376
|
6315 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
6316 : (buf->b_spell_ismw_mb != NULL
|
|
|
6317 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
6318 {
|
|
|
6319 s = p + l;
|
|
|
6320 l = MB_BYTE2LEN(*s);
|
|
|
6321 }
|
|
|
6322 }
|
|
|
6323
|
|
390
|
6324 c = mb_ptr2char(s);
|
|
|
6325 if (c > 255)
|
|
366
|
6326 return mb_get_class(s) >= 2;
|
|
390
|
6327 return spelltab.st_isw[c];
|
|
366
|
6328 }
|
|
358
|
6329 #endif
|
|
366
|
6330
|
|
376
|
6331 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
6332 }
|
|
|
6333
|
|
|
6334 /*
|
|
|
6335 * Return TRUE if "p" points to a word character.
|
|
|
6336 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
6337 */
|
|
|
6338 static int
|
|
|
6339 spell_iswordp_nmw(p)
|
|
|
6340 char_u *p;
|
|
|
6341 {
|
|
|
6342 #ifdef FEAT_MBYTE
|
|
390
|
6343 int c;
|
|
|
6344
|
|
|
6345 if (has_mbyte)
|
|
|
6346 {
|
|
|
6347 c = mb_ptr2char(p);
|
|
|
6348 if (c > 255)
|
|
|
6349 return mb_get_class(p) >= 2;
|
|
|
6350 return spelltab.st_isw[c];
|
|
|
6351 }
|
|
376
|
6352 #endif
|
|
|
6353 return spelltab.st_isw[*p];
|
|
358
|
6354 }
|
|
|
6355
|
|
372
|
6356 #ifdef FEAT_MBYTE
|
|
|
6357 /*
|
|
|
6358 * Return TRUE if "p" points to a word character.
|
|
|
6359 * Wide version of spell_iswordp().
|
|
|
6360 */
|
|
|
6361 static int
|
|
376
|
6362 spell_iswordp_w(p, buf)
|
|
372
|
6363 int *p;
|
|
376
|
6364 buf_T *buf;
|
|
372
|
6365 {
|
|
|
6366 int *s;
|
|
|
6367
|
|
376
|
6368 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
6369 : (buf->b_spell_ismw_mb != NULL
|
|
|
6370 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
6371 s = p + 1;
|
|
|
6372 else
|
|
|
6373 s = p;
|
|
|
6374
|
|
390
|
6375 if (*s > 255)
|
|
372
|
6376 {
|
|
|
6377 if (enc_utf8)
|
|
|
6378 return utf_class(*s) >= 2;
|
|
|
6379 if (enc_dbcs)
|
|
|
6380 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
6381 return 0;
|
|
|
6382 }
|
|
|
6383 return spelltab.st_isw[*s];
|
|
|
6384 }
|
|
|
6385 #endif
|
|
|
6386
|
|
358
|
6387 /*
|
|
339
|
6388 * Write the table with prefix conditions to the .spl file.
|
|
|
6389 */
|
|
|
6390 static void
|
|
|
6391 write_spell_prefcond(fd, gap)
|
|
|
6392 FILE *fd;
|
|
|
6393 garray_T *gap;
|
|
|
6394 {
|
|
|
6395 int i;
|
|
|
6396 char_u *p;
|
|
|
6397 int len;
|
|
|
6398
|
|
|
6399 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
6400
|
|
|
6401 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6402 {
|
|
|
6403 /* <prefcond> : <condlen> <condstr> */
|
|
|
6404 p = ((char_u **)gap->ga_data)[i];
|
|
|
6405 if (p == NULL)
|
|
|
6406 fputc(0, fd);
|
|
|
6407 else
|
|
|
6408 {
|
|
|
6409 len = STRLEN(p);
|
|
|
6410 fputc(len, fd);
|
|
|
6411 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
6412 }
|
|
|
6413 }
|
|
|
6414 }
|
|
|
6415
|
|
|
6416 /*
|
|
307
|
6417 * Write the current tables into the .spl file.
|
|
|
6418 * This makes sure the same characters are recognized as word characters when
|
|
|
6419 * generating an when using a spell file.
|
|
|
6420 */
|
|
|
6421 static void
|
|
|
6422 write_spell_chartab(fd)
|
|
|
6423 FILE *fd;
|
|
|
6424 {
|
|
|
6425 char_u charbuf[256 * 4];
|
|
|
6426 int len = 0;
|
|
|
6427 int flags;
|
|
|
6428 int i;
|
|
|
6429
|
|
|
6430 fputc(128, fd); /* <charflagslen> */
|
|
|
6431 for (i = 128; i < 256; ++i)
|
|
|
6432 {
|
|
|
6433 flags = 0;
|
|
|
6434 if (spelltab.st_isw[i])
|
|
324
|
6435 flags |= CF_WORD;
|
|
307
|
6436 if (spelltab.st_isu[i])
|
|
324
|
6437 flags |= CF_UPPER;
|
|
307
|
6438 fputc(flags, fd); /* <charflags> */
|
|
|
6439
|
|
310
|
6440 #ifdef FEAT_MBYTE
|
|
|
6441 if (has_mbyte)
|
|
|
6442 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
|
6443 else
|
|
|
6444 #endif
|
|
|
6445 charbuf[len++] = spelltab.st_fold[i];
|
|
307
|
6446 }
|
|
|
6447
|
|
|
6448 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
|
6449 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
|
6450 }
|
|
|
6451
|
|
|
6452 /*
|
|
324
|
6453 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
6454 * Uses the character definitions from the .spl file.
|
|
307
|
6455 * When using a multi-byte 'encoding' the length may change!
|
|
|
6456 * Returns FAIL when something wrong.
|
|
|
6457 */
|
|
|
6458 static int
|
|
324
|
6459 spell_casefold(str, len, buf, buflen)
|
|
|
6460 char_u *str;
|
|
307
|
6461 int len;
|
|
|
6462 char_u *buf;
|
|
|
6463 int buflen;
|
|
|
6464 {
|
|
|
6465 int i;
|
|
|
6466
|
|
|
6467 if (len >= buflen)
|
|
|
6468 {
|
|
|
6469 buf[0] = NUL;
|
|
|
6470 return FAIL; /* result will not fit */
|
|
|
6471 }
|
|
|
6472
|
|
|
6473 #ifdef FEAT_MBYTE
|
|
|
6474 if (has_mbyte)
|
|
|
6475 {
|
|
324
|
6476 int outi = 0;
|
|
|
6477 char_u *p;
|
|
307
|
6478 int c;
|
|
|
6479
|
|
|
6480 /* Fold one character at a time. */
|
|
324
|
6481 for (p = str; p < str + len; )
|
|
307
|
6482 {
|
|
|
6483 if (outi + MB_MAXBYTES > buflen)
|
|
|
6484 {
|
|
|
6485 buf[outi] = NUL;
|
|
|
6486 return FAIL;
|
|
|
6487 }
|
|
474
|
6488 c = mb_cptr2char_adv(&p);
|
|
324
|
6489 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
6490 }
|
|
|
6491 buf[outi] = NUL;
|
|
|
6492 }
|
|
|
6493 else
|
|
|
6494 #endif
|
|
|
6495 {
|
|
|
6496 /* Be quick for non-multibyte encodings. */
|
|
|
6497 for (i = 0; i < len; ++i)
|
|
324
|
6498 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
6499 buf[i] = NUL;
|
|
|
6500 }
|
|
|
6501
|
|
|
6502 return OK;
|
|
|
6503 }
|
|
|
6504
|
|
372
|
6505 #define SPS_BEST 1
|
|
|
6506 #define SPS_FAST 2
|
|
|
6507 #define SPS_DOUBLE 4
|
|
|
6508
|
|
|
6509 static int sps_flags = SPS_BEST;
|
|
|
6510
|
|
|
6511 /*
|
|
|
6512 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
|
6513 * Sets "sps_flags".
|
|
|
6514 */
|
|
|
6515 int
|
|
|
6516 spell_check_sps()
|
|
|
6517 {
|
|
|
6518 char_u *p;
|
|
|
6519 char_u buf[MAXPATHL];
|
|
|
6520 int f;
|
|
|
6521
|
|
|
6522 sps_flags = 0;
|
|
|
6523
|
|
|
6524 for (p = p_sps; *p != NUL; )
|
|
|
6525 {
|
|
|
6526 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6527
|
|
|
6528 f = 0;
|
|
|
6529 if (STRCMP(buf, "best") == 0)
|
|
|
6530 f = SPS_BEST;
|
|
|
6531 else if (STRCMP(buf, "fast") == 0)
|
|
|
6532 f = SPS_FAST;
|
|
|
6533 else if (STRCMP(buf, "double") == 0)
|
|
|
6534 f = SPS_DOUBLE;
|
|
|
6535 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
6536 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
6537 f = -1;
|
|
|
6538
|
|
|
6539 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
6540 {
|
|
|
6541 sps_flags = SPS_BEST;
|
|
|
6542 return FAIL;
|
|
|
6543 }
|
|
|
6544 if (f != 0)
|
|
|
6545 sps_flags = f;
|
|
|
6546 }
|
|
|
6547
|
|
|
6548 if (sps_flags == 0)
|
|
|
6549 sps_flags = SPS_BEST;
|
|
|
6550
|
|
|
6551 return OK;
|
|
|
6552 }
|
|
|
6553
|
|
|
6554 /* Remember what "z?" replaced. */
|
|
|
6555 static char_u *repl_from = NULL;
|
|
|
6556 static char_u *repl_to = NULL;
|
|
|
6557
|
|
323
|
6558 /*
|
|
|
6559 * "z?": Find badly spelled word under or after the cursor.
|
|
|
6560 * Give suggestions for the properly spelled word.
|
|
|
6561 */
|
|
|
6562 void
|
|
|
6563 spell_suggest()
|
|
|
6564 {
|
|
|
6565 char_u *line;
|
|
|
6566 pos_T prev_cursor = curwin->w_cursor;
|
|
|
6567 char_u wcopy[MAXWLEN + 2];
|
|
|
6568 char_u *p;
|
|
|
6569 int i;
|
|
|
6570 int c;
|
|
|
6571 suginfo_T sug;
|
|
|
6572 suggest_T *stp;
|
|
372
|
6573 int mouse_used;
|
|
388
|
6574 int need_cap;
|
|
323
|
6575
|
|
344
|
6576 /* Find the start of the badly spelled word. */
|
|
346
|
6577 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
6578 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
6579 {
|
|
|
6580 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
6581 return;
|
|
|
6582
|
|
|
6583 /* No bad word or it starts after the cursor: use the word under the
|
|
|
6584 * cursor. */
|
|
|
6585 curwin->w_cursor = prev_cursor;
|
|
|
6586 line = ml_get_curline();
|
|
|
6587 p = line + curwin->w_cursor.col;
|
|
|
6588 /* Backup to before start of word. */
|
|
390
|
6589 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
6590 mb_ptr_back(line, p);
|
|
|
6591 /* Forward to start of word. */
|
|
390
|
6592 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
6593 mb_ptr_adv(p);
|
|
|
6594
|
|
390
|
6595 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
6596 {
|
|
|
6597 beep_flush();
|
|
|
6598 return;
|
|
|
6599 }
|
|
|
6600 curwin->w_cursor.col = p - line;
|
|
323
|
6601 }
|
|
|
6602
|
|
344
|
6603 /* Get the word and its length. */
|
|
|
6604
|
|
388
|
6605 /* Figure out if the word should be capitalised. */
|
|
475
|
6606 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
6607
|
|
|
6608 line = ml_get_curline();
|
|
388
|
6609
|
|
344
|
6610 /* Get the list of suggestions */
|
|
388
|
6611 spell_find_suggest(line + curwin->w_cursor.col, &sug, (int)Rows - 2,
|
|
|
6612 TRUE, need_cap);
|
|
323
|
6613
|
|
|
6614 if (sug.su_ga.ga_len == 0)
|
|
|
6615 MSG(_("Sorry, no suggestions"));
|
|
|
6616 else
|
|
|
6617 {
|
|
372
|
6618 vim_free(repl_from);
|
|
|
6619 repl_from = NULL;
|
|
|
6620 vim_free(repl_to);
|
|
|
6621 repl_to = NULL;
|
|
|
6622
|
|
474
|
6623 #ifdef FEAT_RIGHTLEFT
|
|
|
6624 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
6625 cmdmsg_rl = curwin->w_p_rl;
|
|
|
6626 if (cmdmsg_rl)
|
|
|
6627 msg_col = Columns - 1;
|
|
|
6628 #endif
|
|
|
6629
|
|
323
|
6630 /* List the suggestions. */
|
|
|
6631 msg_start();
|
|
372
|
6632 lines_left = Rows; /* avoid more prompt */
|
|
323
|
6633 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
6634 sug.su_badlen, sug.su_badptr);
|
|
474
|
6635 #ifdef FEAT_RIGHTLEFT
|
|
|
6636 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
6637 {
|
|
|
6638 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
6639 * untranslated message rightleft. */
|
|
|
6640 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
6641 sug.su_badlen, sug.su_badptr);
|
|
|
6642 }
|
|
|
6643 #endif
|
|
323
|
6644 msg_puts(IObuff);
|
|
|
6645 msg_clr_eos();
|
|
|
6646 msg_putchar('\n');
|
|
346
|
6647
|
|
323
|
6648 msg_scroll = TRUE;
|
|
|
6649 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
6650 {
|
|
344
|
6651 stp = &SUG(sug.su_ga, i);
|
|
323
|
6652
|
|
|
6653 /* The suggested word may replace only part of the bad word, add
|
|
|
6654 * the not replaced part. */
|
|
|
6655 STRCPY(wcopy, stp->st_word);
|
|
|
6656 if (sug.su_badlen > stp->st_orglen)
|
|
|
6657 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
6658 sug.su_badptr + stp->st_orglen,
|
|
|
6659 sug.su_badlen - stp->st_orglen);
|
|
474
|
6660 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
6661 #ifdef FEAT_RIGHTLEFT
|
|
|
6662 if (cmdmsg_rl)
|
|
|
6663 rl_mirror(IObuff);
|
|
|
6664 #endif
|
|
|
6665 msg_puts(IObuff);
|
|
|
6666
|
|
|
6667 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
6668 msg_puts(IObuff);
|
|
|
6669
|
|
|
6670 /* The word may replace more than "su_badlen". */
|
|
|
6671 if (sug.su_badlen < stp->st_orglen)
|
|
|
6672 {
|
|
|
6673 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
6674 stp->st_orglen, sug.su_badptr);
|
|
|
6675 msg_puts(IObuff);
|
|
|
6676 }
|
|
|
6677
|
|
324
|
6678 if (p_verbose > 0)
|
|
344
|
6679 {
|
|
346
|
6680 /* Add the score. */
|
|
351
|
6681 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
6682 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
6683 stp->st_salscore ? "s " : "",
|
|
|
6684 stp->st_score, stp->st_altscore);
|
|
|
6685 else
|
|
474
|
6686 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
6687 stp->st_score);
|
|
474
|
6688 #ifdef FEAT_RIGHTLEFT
|
|
|
6689 if (cmdmsg_rl)
|
|
|
6690 /* Mirror the numbers, but keep the leading space. */
|
|
|
6691 rl_mirror(IObuff + 1);
|
|
|
6692 #endif
|
|
346
|
6693 msg_advance(30);
|
|
|
6694 msg_puts(IObuff);
|
|
344
|
6695 }
|
|
323
|
6696 msg_putchar('\n');
|
|
|
6697 }
|
|
|
6698
|
|
474
|
6699 #ifdef FEAT_RIGHTLEFT
|
|
|
6700 cmdmsg_rl = FALSE;
|
|
|
6701 msg_col = 0;
|
|
|
6702 #endif
|
|
323
|
6703 /* Ask for choice. */
|
|
372
|
6704 i = prompt_for_number(&mouse_used);
|
|
|
6705 if (mouse_used)
|
|
|
6706 i -= lines_left;
|
|
|
6707
|
|
344
|
6708 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
6709 {
|
|
372
|
6710 /* Save the from and to text for :spellrepall. */
|
|
|
6711 stp = &SUG(sug.su_ga, i - 1);
|
|
|
6712 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
6713 repl_to = vim_strsave(stp->st_word);
|
|
|
6714
|
|
323
|
6715 /* Replace the word. */
|
|
|
6716 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
6717 if (p != NULL)
|
|
|
6718 {
|
|
|
6719 c = sug.su_badptr - line;
|
|
|
6720 mch_memmove(p, line, c);
|
|
|
6721 STRCPY(p + c, stp->st_word);
|
|
|
6722 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
6723 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6724 curwin->w_cursor.col = c;
|
|
|
6725 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
6726
|
|
|
6727 /* For redo we use a change-word command. */
|
|
|
6728 ResetRedobuff();
|
|
|
6729 AppendToRedobuff((char_u *)"ciw");
|
|
|
6730 AppendToRedobuff(stp->st_word);
|
|
|
6731 AppendCharToRedobuff(ESC);
|
|
323
|
6732 }
|
|
|
6733 }
|
|
|
6734 else
|
|
|
6735 curwin->w_cursor = prev_cursor;
|
|
|
6736 }
|
|
|
6737
|
|
344
|
6738 spell_find_cleanup(&sug);
|
|
|
6739 }
|
|
|
6740
|
|
|
6741 /*
|
|
475
|
6742 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
6743 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
6744 */
|
|
|
6745 static int
|
|
|
6746 check_need_cap(lnum, col)
|
|
|
6747 linenr_T lnum;
|
|
|
6748 colnr_T col;
|
|
|
6749 {
|
|
|
6750 int need_cap = FALSE;
|
|
|
6751 char_u *line;
|
|
|
6752 char_u *line_copy = NULL;
|
|
|
6753 char_u *p;
|
|
|
6754 colnr_T endcol;
|
|
|
6755 regmatch_T regmatch;
|
|
|
6756
|
|
|
6757 if (curbuf->b_cap_prog == NULL)
|
|
|
6758 return FALSE;
|
|
|
6759
|
|
|
6760 line = ml_get_curline();
|
|
|
6761 endcol = 0;
|
|
|
6762 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
6763 {
|
|
|
6764 /* At start of line, check if previous line is empty or sentence
|
|
|
6765 * ends there. */
|
|
|
6766 if (lnum == 1)
|
|
|
6767 need_cap = TRUE;
|
|
|
6768 else
|
|
|
6769 {
|
|
|
6770 line = ml_get(lnum - 1);
|
|
|
6771 if (*skipwhite(line) == NUL)
|
|
|
6772 need_cap = TRUE;
|
|
|
6773 else
|
|
|
6774 {
|
|
|
6775 /* Append a space in place of the line break. */
|
|
|
6776 line_copy = concat_str(line, (char_u *)" ");
|
|
|
6777 line = line_copy;
|
|
|
6778 endcol = STRLEN(line);
|
|
|
6779 }
|
|
|
6780 }
|
|
|
6781 }
|
|
|
6782 else
|
|
|
6783 endcol = col;
|
|
|
6784
|
|
|
6785 if (endcol > 0)
|
|
|
6786 {
|
|
|
6787 /* Check if sentence ends before the bad word. */
|
|
|
6788 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
6789 regmatch.rm_ic = FALSE;
|
|
|
6790 p = line + endcol;
|
|
|
6791 for (;;)
|
|
|
6792 {
|
|
|
6793 mb_ptr_back(line, p);
|
|
|
6794 if (p == line || spell_iswordp_nmw(p))
|
|
|
6795 break;
|
|
|
6796 if (vim_regexec(®match, p, 0)
|
|
|
6797 && regmatch.endp[0] == line + endcol)
|
|
|
6798 {
|
|
|
6799 need_cap = TRUE;
|
|
|
6800 break;
|
|
|
6801 }
|
|
|
6802 }
|
|
|
6803 }
|
|
|
6804
|
|
|
6805 vim_free(line_copy);
|
|
|
6806
|
|
|
6807 return need_cap;
|
|
|
6808 }
|
|
|
6809
|
|
|
6810
|
|
|
6811 /*
|
|
372
|
6812 * ":spellrepall"
|
|
|
6813 */
|
|
|
6814 /*ARGSUSED*/
|
|
|
6815 void
|
|
|
6816 ex_spellrepall(eap)
|
|
|
6817 exarg_T *eap;
|
|
|
6818 {
|
|
|
6819 pos_T pos = curwin->w_cursor;
|
|
|
6820 char_u *frompat;
|
|
|
6821 int addlen;
|
|
|
6822 char_u *line;
|
|
|
6823 char_u *p;
|
|
|
6824 int didone = FALSE;
|
|
|
6825 int save_ws = p_ws;
|
|
|
6826
|
|
|
6827 if (repl_from == NULL || repl_to == NULL)
|
|
|
6828 {
|
|
|
6829 EMSG(_("E752: No previous spell replacement"));
|
|
|
6830 return;
|
|
|
6831 }
|
|
|
6832 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
6833
|
|
|
6834 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
6835 if (frompat == NULL)
|
|
|
6836 return;
|
|
|
6837 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
6838 p_ws = FALSE;
|
|
|
6839
|
|
|
6840 curwin->w_cursor.lnum = 0;
|
|
|
6841 while (!got_int)
|
|
|
6842 {
|
|
|
6843 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
6844 || u_save_cursor() == FAIL)
|
|
|
6845 break;
|
|
|
6846
|
|
|
6847 /* Only replace when the right word isn't there yet. This happens
|
|
|
6848 * when changing "etc" to "etc.". */
|
|
|
6849 line = ml_get_curline();
|
|
|
6850 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
6851 repl_to, STRLEN(repl_to)) != 0)
|
|
|
6852 {
|
|
|
6853 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
6854 if (p == NULL)
|
|
|
6855 break;
|
|
|
6856 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
6857 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
6858 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
6859 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6860 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
6861 didone = TRUE;
|
|
|
6862 }
|
|
|
6863 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
6864 }
|
|
|
6865
|
|
|
6866 p_ws = save_ws;
|
|
|
6867 curwin->w_cursor = pos;
|
|
|
6868 vim_free(frompat);
|
|
|
6869
|
|
|
6870 if (!didone)
|
|
|
6871 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
|
6872 }
|
|
|
6873
|
|
|
6874 /*
|
|
344
|
6875 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
6876 * a list of allocated strings.
|
|
|
6877 */
|
|
|
6878 void
|
|
475
|
6879 spell_suggest_list(gap, word, maxcount, need_cap)
|
|
344
|
6880 garray_T *gap;
|
|
|
6881 char_u *word;
|
|
|
6882 int maxcount; /* maximum nr of suggestions */
|
|
475
|
6883 int need_cap; /* 'spellcapcheck' matched */
|
|
344
|
6884 {
|
|
|
6885 suginfo_T sug;
|
|
|
6886 int i;
|
|
|
6887 suggest_T *stp;
|
|
|
6888 char_u *wcopy;
|
|
|
6889
|
|
475
|
6890 spell_find_suggest(word, &sug, maxcount, FALSE, need_cap);
|
|
344
|
6891
|
|
|
6892 /* Make room in "gap". */
|
|
|
6893 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
6894 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
6895 return;
|
|
|
6896
|
|
323
|
6897 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
6898 {
|
|
|
6899 stp = &SUG(sug.su_ga, i);
|
|
|
6900
|
|
|
6901 /* The suggested word may replace only part of "word", add the not
|
|
|
6902 * replaced part. */
|
|
|
6903 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
6904 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
6905 if (wcopy == NULL)
|
|
|
6906 break;
|
|
|
6907 STRCPY(wcopy, stp->st_word);
|
|
|
6908 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
6909 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
6910 }
|
|
|
6911
|
|
|
6912 spell_find_cleanup(&sug);
|
|
|
6913 }
|
|
|
6914
|
|
|
6915 /*
|
|
|
6916 * Find spell suggestions for the word at the start of "badptr".
|
|
|
6917 * Return the suggestions in "su->su_ga".
|
|
|
6918 * The maximum number of suggestions is "maxcount".
|
|
|
6919 * Note: does use info for the current window.
|
|
|
6920 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
6921 */
|
|
|
6922 static void
|
|
388
|
6923 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
6924 char_u *badptr;
|
|
|
6925 suginfo_T *su;
|
|
|
6926 int maxcount;
|
|
358
|
6927 int banbadword; /* don't include badword in suggestions */
|
|
388
|
6928 int need_cap; /* word should start with capital */
|
|
344
|
6929 {
|
|
385
|
6930 int attr = 0;
|
|
372
|
6931 char_u buf[MAXPATHL];
|
|
|
6932 char_u *p;
|
|
|
6933 int do_combine = FALSE;
|
|
|
6934 char_u *sps_copy;
|
|
|
6935 #ifdef FEAT_EVAL
|
|
|
6936 static int expr_busy = FALSE;
|
|
|
6937 #endif
|
|
385
|
6938 int c;
|
|
344
|
6939
|
|
|
6940 /*
|
|
|
6941 * Set the info in "*su".
|
|
|
6942 */
|
|
|
6943 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
6944 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
6945 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
6946 if (*badptr == NUL)
|
|
|
6947 return;
|
|
344
|
6948 hash_init(&su->su_banned);
|
|
|
6949
|
|
|
6950 su->su_badptr = badptr;
|
|
385
|
6951 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
6952 su->su_maxcount = maxcount;
|
|
372
|
6953 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
6954
|
|
|
6955 if (su->su_badlen >= MAXWLEN)
|
|
|
6956 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
6957 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
6958 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
6959 su->su_fbadword, MAXWLEN);
|
|
346
|
6960 /* get caps flags for bad word */
|
|
474
|
6961 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
6962 su->su_badptr + su->su_badlen);
|
|
388
|
6963 if (need_cap)
|
|
|
6964 su->su_badflags |= WF_ONECAP;
|
|
344
|
6965
|
|
385
|
6966 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
6967 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
6968 * for that. */
|
|
455
|
6969 c = PTR2CHAR(su->su_badptr);
|
|
385
|
6970 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
6971 {
|
|
|
6972 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
6973 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
6974 0, TRUE);
|
|
|
6975 }
|
|
|
6976
|
|
344
|
6977 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
6978 if (banbadword)
|
|
|
6979 add_banned(su, su->su_badword);
|
|
344
|
6980
|
|
372
|
6981 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
6982 sps_copy = vim_strsave(p_sps);
|
|
|
6983 if (sps_copy == NULL)
|
|
|
6984 return;
|
|
|
6985
|
|
|
6986 /* Loop over the items in 'spellsuggest'. */
|
|
|
6987 for (p = sps_copy; *p != NUL; )
|
|
|
6988 {
|
|
|
6989 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6990
|
|
|
6991 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
6992 {
|
|
|
6993 #ifdef FEAT_EVAL
|
|
375
|
6994 /* Evaluate an expression. Skip this when called recursively,
|
|
|
6995 * when using spellsuggest() in the expression. */
|
|
372
|
6996 if (!expr_busy)
|
|
|
6997 {
|
|
|
6998 expr_busy = TRUE;
|
|
|
6999 spell_suggest_expr(su, buf + 5);
|
|
|
7000 expr_busy = FALSE;
|
|
|
7001 }
|
|
|
7002 #endif
|
|
|
7003 }
|
|
|
7004 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
7005 /* Use list of suggestions in a file. */
|
|
|
7006 spell_suggest_file(su, buf + 5);
|
|
|
7007 else
|
|
|
7008 {
|
|
|
7009 /* Use internal method. */
|
|
|
7010 spell_suggest_intern(su);
|
|
|
7011 if (sps_flags & SPS_DOUBLE)
|
|
|
7012 do_combine = TRUE;
|
|
|
7013 }
|
|
|
7014 }
|
|
|
7015
|
|
|
7016 vim_free(sps_copy);
|
|
|
7017
|
|
|
7018 if (do_combine)
|
|
|
7019 /* Combine the two list of suggestions. This must be done last,
|
|
|
7020 * because sorting changes the order again. */
|
|
|
7021 score_combine(su);
|
|
|
7022 }
|
|
|
7023
|
|
|
7024 #ifdef FEAT_EVAL
|
|
|
7025 /*
|
|
|
7026 * Find suggestions by evaluating expression "expr".
|
|
|
7027 */
|
|
|
7028 static void
|
|
|
7029 spell_suggest_expr(su, expr)
|
|
|
7030 suginfo_T *su;
|
|
|
7031 char_u *expr;
|
|
|
7032 {
|
|
|
7033 list_T *list;
|
|
|
7034 listitem_T *li;
|
|
|
7035 int score;
|
|
|
7036 char_u *p;
|
|
|
7037
|
|
|
7038 /* The work is split up in a few parts to avoid having to export
|
|
|
7039 * suginfo_T.
|
|
|
7040 * First evaluate the expression and get the resulting list. */
|
|
|
7041 list = eval_spell_expr(su->su_badword, expr);
|
|
|
7042 if (list != NULL)
|
|
|
7043 {
|
|
|
7044 /* Loop over the items in the list. */
|
|
|
7045 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
7046 if (li->li_tv.v_type == VAR_LIST)
|
|
|
7047 {
|
|
|
7048 /* Get the word and the score from the items. */
|
|
|
7049 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
7050 if (score >= 0)
|
|
|
7051 add_suggestion(su, &su->su_ga, p,
|
|
|
7052 su->su_badlen, score, 0, TRUE);
|
|
|
7053 }
|
|
|
7054 list_unref(list);
|
|
|
7055 }
|
|
|
7056
|
|
|
7057 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
7058 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7059 }
|
|
|
7060 #endif
|
|
|
7061
|
|
|
7062 /*
|
|
474
|
7063 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
7064 */
|
|
|
7065 static void
|
|
|
7066 spell_suggest_file(su, fname)
|
|
|
7067 suginfo_T *su;
|
|
|
7068 char_u *fname;
|
|
|
7069 {
|
|
|
7070 FILE *fd;
|
|
|
7071 char_u line[MAXWLEN * 2];
|
|
|
7072 char_u *p;
|
|
|
7073 int len;
|
|
|
7074 char_u cword[MAXWLEN];
|
|
|
7075
|
|
|
7076 /* Open the file. */
|
|
|
7077 fd = mch_fopen((char *)fname, "r");
|
|
|
7078 if (fd == NULL)
|
|
|
7079 {
|
|
|
7080 EMSG2(_(e_notopen), fname);
|
|
|
7081 return;
|
|
|
7082 }
|
|
|
7083
|
|
|
7084 /* Read it line by line. */
|
|
|
7085 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
7086 {
|
|
|
7087 line_breakcheck();
|
|
|
7088
|
|
|
7089 p = vim_strchr(line, '/');
|
|
|
7090 if (p == NULL)
|
|
|
7091 continue; /* No Tab found, just skip the line. */
|
|
|
7092 *p++ = NUL;
|
|
|
7093 if (STRICMP(su->su_badword, line) == 0)
|
|
|
7094 {
|
|
|
7095 /* Match! Isolate the good word, until CR or NL. */
|
|
|
7096 for (len = 0; p[len] >= ' '; ++len)
|
|
|
7097 ;
|
|
|
7098 p[len] = NUL;
|
|
|
7099
|
|
|
7100 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
7101 * of the bad word. */
|
|
|
7102 if (captype(p, NULL) == 0)
|
|
|
7103 {
|
|
|
7104 make_case_word(p, cword, su->su_badflags);
|
|
|
7105 p = cword;
|
|
|
7106 }
|
|
|
7107
|
|
|
7108 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
7109 SCORE_FILE, 0, TRUE);
|
|
|
7110 }
|
|
|
7111 }
|
|
|
7112
|
|
|
7113 fclose(fd);
|
|
|
7114
|
|
|
7115 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
7116 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7117 }
|
|
|
7118
|
|
|
7119 /*
|
|
|
7120 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
7121 */
|
|
|
7122 static void
|
|
|
7123 spell_suggest_intern(su)
|
|
|
7124 suginfo_T *su;
|
|
|
7125 {
|
|
344
|
7126 /*
|
|
346
|
7127 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
7128 *
|
|
|
7129 * Set a maximum score to limit the combination of operations that is
|
|
|
7130 * tried.
|
|
|
7131 */
|
|
346
|
7132 suggest_try_special(su);
|
|
|
7133
|
|
|
7134 /*
|
|
|
7135 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
7136 * from the .aff file and inserting a space (split the word).
|
|
|
7137 */
|
|
|
7138 suggest_try_change(su);
|
|
344
|
7139
|
|
|
7140 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
7141 if (sps_flags & SPS_DOUBLE)
|
|
|
7142 score_comp_sal(su);
|
|
|
7143
|
|
|
7144 /*
|
|
346
|
7145 * 3. Try finding sound-a-like words.
|
|
344
|
7146 *
|
|
|
7147 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
7148 * very slow and often doesn't find new suggestions.
|
|
|
7149 */
|
|
|
7150 if ((sps_flags & SPS_DOUBLE)
|
|
|
7151 || (!(sps_flags & SPS_FAST)
|
|
|
7152 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
7153 {
|
|
|
7154 /* Allow a higher score now. */
|
|
|
7155 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
7156 suggest_try_soundalike(su);
|
|
344
|
7157 }
|
|
|
7158
|
|
|
7159 /* When CTRL-C was hit while searching do show the results. */
|
|
|
7160 ui_breakcheck();
|
|
|
7161 if (got_int)
|
|
|
7162 {
|
|
|
7163 (void)vgetc();
|
|
|
7164 got_int = FALSE;
|
|
|
7165 }
|
|
|
7166
|
|
372
|
7167 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
7168 {
|
|
|
7169 if (sps_flags & SPS_BEST)
|
|
|
7170 /* Adjust the word score for how it sounds like. */
|
|
|
7171 rescore_suggestions(su);
|
|
|
7172
|
|
|
7173 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
7174 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
7175 }
|
|
|
7176 }
|
|
|
7177
|
|
|
7178 /*
|
|
|
7179 * Free the info put in "*su" by spell_find_suggest().
|
|
|
7180 */
|
|
|
7181 static void
|
|
|
7182 spell_find_cleanup(su)
|
|
|
7183 suginfo_T *su;
|
|
|
7184 {
|
|
|
7185 int i;
|
|
|
7186
|
|
|
7187 /* Free the suggestions. */
|
|
|
7188 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7189 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
7190 ga_clear(&su->su_ga);
|
|
|
7191 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
7192 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
7193 ga_clear(&su->su_sga);
|
|
323
|
7194
|
|
|
7195 /* Free the banned words. */
|
|
344
|
7196 free_banned(su);
|
|
323
|
7197 }
|
|
|
7198
|
|
|
7199 /*
|
|
324
|
7200 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
7201 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
7202 * The result is NUL terminated.
|
|
323
|
7203 */
|
|
|
7204 static void
|
|
324
|
7205 onecap_copy(word, wcopy, upper)
|
|
323
|
7206 char_u *word;
|
|
|
7207 char_u *wcopy;
|
|
|
7208 int upper; /* TRUE: first letter made upper case */
|
|
|
7209 {
|
|
|
7210 char_u *p;
|
|
|
7211 int c;
|
|
|
7212 int l;
|
|
|
7213
|
|
|
7214 p = word;
|
|
|
7215 #ifdef FEAT_MBYTE
|
|
|
7216 if (has_mbyte)
|
|
474
|
7217 c = mb_cptr2char_adv(&p);
|
|
323
|
7218 else
|
|
|
7219 #endif
|
|
|
7220 c = *p++;
|
|
|
7221 if (upper)
|
|
324
|
7222 c = SPELL_TOUPPER(c);
|
|
323
|
7223 else
|
|
324
|
7224 c = SPELL_TOFOLD(c);
|
|
323
|
7225 #ifdef FEAT_MBYTE
|
|
|
7226 if (has_mbyte)
|
|
|
7227 l = mb_char2bytes(c, wcopy);
|
|
|
7228 else
|
|
|
7229 #endif
|
|
|
7230 {
|
|
|
7231 l = 1;
|
|
|
7232 wcopy[0] = c;
|
|
|
7233 }
|
|
376
|
7234 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
7235 }
|
|
|
7236
|
|
|
7237 /*
|
|
324
|
7238 * Make a copy of "word" with all the letters upper cased into
|
|
|
7239 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
7240 */
|
|
|
7241 static void
|
|
|
7242 allcap_copy(word, wcopy)
|
|
|
7243 char_u *word;
|
|
|
7244 char_u *wcopy;
|
|
|
7245 {
|
|
|
7246 char_u *s;
|
|
|
7247 char_u *d;
|
|
|
7248 int c;
|
|
|
7249
|
|
|
7250 d = wcopy;
|
|
|
7251 for (s = word; *s != NUL; )
|
|
|
7252 {
|
|
|
7253 #ifdef FEAT_MBYTE
|
|
|
7254 if (has_mbyte)
|
|
474
|
7255 c = mb_cptr2char_adv(&s);
|
|
323
|
7256 else
|
|
|
7257 #endif
|
|
|
7258 c = *s++;
|
|
324
|
7259 c = SPELL_TOUPPER(c);
|
|
323
|
7260
|
|
|
7261 #ifdef FEAT_MBYTE
|
|
|
7262 if (has_mbyte)
|
|
|
7263 {
|
|
|
7264 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
7265 break;
|
|
|
7266 d += mb_char2bytes(c, d);
|
|
|
7267 }
|
|
|
7268 else
|
|
|
7269 #endif
|
|
|
7270 {
|
|
|
7271 if (d - wcopy >= MAXWLEN - 1)
|
|
|
7272 break;
|
|
|
7273 *d++ = c;
|
|
|
7274 }
|
|
|
7275 }
|
|
|
7276 *d = NUL;
|
|
|
7277 }
|
|
|
7278
|
|
|
7279 /*
|
|
346
|
7280 * Try finding suggestions by recognizing specific situations.
|
|
|
7281 */
|
|
|
7282 static void
|
|
|
7283 suggest_try_special(su)
|
|
|
7284 suginfo_T *su;
|
|
|
7285 {
|
|
|
7286 char_u *p;
|
|
376
|
7287 size_t len;
|
|
346
|
7288 int c;
|
|
|
7289 char_u word[MAXWLEN];
|
|
|
7290
|
|
|
7291 /*
|
|
|
7292 * Recognize a word that is repeated: "the the".
|
|
|
7293 */
|
|
|
7294 p = skiptowhite(su->su_fbadword);
|
|
|
7295 len = p - su->su_fbadword;
|
|
|
7296 p = skipwhite(p);
|
|
|
7297 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
7298 {
|
|
|
7299 /* Include badflags: if the badword is onecap or allcap
|
|
|
7300 * use that for the goodword too: "The the" -> "The". */
|
|
|
7301 c = su->su_fbadword[len];
|
|
|
7302 su->su_fbadword[len] = NUL;
|
|
|
7303 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
7304 su->su_fbadword[len] = c;
|
|
351
|
7305 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
7306 }
|
|
|
7307 }
|
|
|
7308
|
|
|
7309 /*
|
|
323
|
7310 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
7311 *
|
|
|
7312 * This uses a state machine. At each node in the tree we try various
|
|
|
7313 * operations. When trying if an operation work "depth" is increased and the
|
|
|
7314 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
7315 * character, replace one and delete another. The number of changes is
|
|
|
7316 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
7317 */
|
|
|
7318 static void
|
|
346
|
7319 suggest_try_change(su)
|
|
323
|
7320 suginfo_T *su;
|
|
|
7321 {
|
|
|
7322 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
7323 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
7324 trystate_T stack[MAXWLEN];
|
|
|
7325 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
|
7326 * to for word split) */
|
|
|
7327 char_u prewordlen = 0; /* length of word in "preword" */
|
|
|
7328 int splitoff = 0; /* index in tword after last split */
|
|
|
7329 trystate_T *sp;
|
|
|
7330 int newscore;
|
|
|
7331 langp_T *lp;
|
|
375
|
7332 char_u *byts, *fbyts, *pbyts;
|
|
|
7333 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
7334 int depth;
|
|
330
|
7335 int c, c2, c3;
|
|
390
|
7336 int n;
|
|
323
|
7337 int flags;
|
|
|
7338 garray_T *gap;
|
|
324
|
7339 idx_T arridx;
|
|
323
|
7340 int len;
|
|
|
7341 char_u *p;
|
|
|
7342 fromto_T *ftp;
|
|
330
|
7343 int fl = 0, tl;
|
|
346
|
7344 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
323
|
7345
|
|
|
7346 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
7347 * to find matches (esp. REP items). Append some more text, changing
|
|
|
7348 * chars after the bad word may help. */
|
|
323
|
7349 STRCPY(fword, su->su_fbadword);
|
|
346
|
7350 n = STRLEN(fword);
|
|
|
7351 p = su->su_badptr + su->su_badlen;
|
|
|
7352 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
7353
|
|
|
7354 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7355 lp->lp_slang != NULL; ++lp)
|
|
|
7356 {
|
|
|
7357 /*
|
|
|
7358 * Go through the whole case-fold tree, try changes at each node.
|
|
|
7359 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
7360 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
7361 * word).
|
|
|
7362 */
|
|
|
7363 depth = 0;
|
|
375
|
7364 sp = &stack[0];
|
|
|
7365 sp->ts_state = STATE_START;
|
|
|
7366 sp->ts_score = 0;
|
|
|
7367 sp->ts_curi = 1;
|
|
|
7368 sp->ts_fidx = 0;
|
|
|
7369 sp->ts_fidxtry = 0;
|
|
|
7370 sp->ts_twordlen = 0;
|
|
|
7371 sp->ts_arridx = 0;
|
|
330
|
7372 #ifdef FEAT_MBYTE
|
|
375
|
7373 sp->ts_tcharlen = 0;
|
|
330
|
7374 #endif
|
|
|
7375
|
|
|
7376 /*
|
|
375
|
7377 * When there are postponed prefixes we need to use these first. At
|
|
|
7378 * the end of the prefix we continue in the case-fold tree.
|
|
|
7379 */
|
|
|
7380 fbyts = lp->lp_slang->sl_fbyts;
|
|
|
7381 fidxs = lp->lp_slang->sl_fidxs;
|
|
|
7382 pbyts = lp->lp_slang->sl_pbyts;
|
|
|
7383 pidxs = lp->lp_slang->sl_pidxs;
|
|
|
7384 if (pbyts != NULL)
|
|
|
7385 {
|
|
|
7386 byts = pbyts;
|
|
|
7387 idxs = pidxs;
|
|
|
7388 sp->ts_prefixdepth = PREFIXTREE;
|
|
|
7389 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
7390 }
|
|
|
7391 else
|
|
|
7392 {
|
|
|
7393 byts = fbyts;
|
|
|
7394 idxs = fidxs;
|
|
|
7395 sp->ts_prefixdepth = NOPREFIX;
|
|
|
7396 }
|
|
|
7397
|
|
|
7398 /*
|
|
330
|
7399 * Loop to find all suggestions. At each round we either:
|
|
|
7400 * - For the current state try one operation, advance "ts_curi",
|
|
|
7401 * increase "depth".
|
|
|
7402 * - When a state is done go to the next, set "ts_state".
|
|
|
7403 * - When all states are tried decrease "depth".
|
|
|
7404 */
|
|
323
|
7405 while (depth >= 0 && !got_int)
|
|
|
7406 {
|
|
|
7407 sp = &stack[depth];
|
|
|
7408 switch (sp->ts_state)
|
|
|
7409 {
|
|
|
7410 case STATE_START:
|
|
375
|
7411 case STATE_NOPREFIX:
|
|
323
|
7412 /*
|
|
|
7413 * Start of node: Deal with NUL bytes, which means
|
|
|
7414 * tword[] may end here.
|
|
|
7415 */
|
|
|
7416 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
7417 len = byts[arridx]; /* bytes in this node */
|
|
|
7418 arridx += sp->ts_curi; /* index of current byte */
|
|
|
7419
|
|
375
|
7420 if (sp->ts_prefixdepth == PREFIXTREE)
|
|
|
7421 {
|
|
|
7422 /* Skip over the NUL bytes, we use them later. */
|
|
|
7423 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
7424 ;
|
|
|
7425 sp->ts_curi += n;
|
|
|
7426
|
|
390
|
7427 /* Always past NUL bytes now. */
|
|
|
7428 n = (int)sp->ts_state;
|
|
|
7429 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7430 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
7431
|
|
375
|
7432 /* At end of a prefix or at start of prefixtree: check for
|
|
|
7433 * following word. */
|
|
390
|
7434 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
7435 {
|
|
455
|
7436 /* Set su->su_badflags to the caps type at this
|
|
|
7437 * position. Use the caps type until here for the
|
|
|
7438 * prefix itself. */
|
|
|
7439 #ifdef FEAT_MBYTE
|
|
|
7440 if (has_mbyte)
|
|
|
7441 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
7442 else
|
|
|
7443 #endif
|
|
|
7444 n = sp->ts_fidx;
|
|
474
|
7445 flags = badword_captype(su->su_badptr,
|
|
|
7446 su->su_badptr + n);
|
|
|
7447 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
7448 su->su_badptr + su->su_badlen);
|
|
375
|
7449 ++depth;
|
|
|
7450 stack[depth] = stack[depth - 1];
|
|
|
7451 sp = &stack[depth];
|
|
|
7452 sp->ts_prefixdepth = depth - 1;
|
|
|
7453 byts = fbyts;
|
|
|
7454 idxs = fidxs;
|
|
|
7455 sp->ts_state = STATE_START;
|
|
|
7456 sp->ts_curi = 1; /* start just after length byte */
|
|
|
7457 sp->ts_arridx = 0;
|
|
|
7458
|
|
455
|
7459 /* Move the prefix to preword[] with the right case
|
|
|
7460 * and make find_keepcap_word() works. */
|
|
|
7461 splitoff = sp->ts_twordlen;
|
|
|
7462 tword[splitoff] = NUL;
|
|
|
7463 make_case_word(tword, preword, flags);
|
|
|
7464 prewordlen = STRLEN(preword);
|
|
375
|
7465 }
|
|
|
7466 break;
|
|
|
7467 }
|
|
|
7468
|
|
346
|
7469 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
7470 {
|
|
|
7471 /* Past bytes in node and/or past NUL bytes. */
|
|
|
7472 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7473 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7474 break;
|
|
|
7475 }
|
|
|
7476
|
|
|
7477 /*
|
|
|
7478 * End of word in tree.
|
|
|
7479 */
|
|
|
7480 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
7481
|
|
324
|
7482 flags = (int)idxs[arridx];
|
|
323
|
7483
|
|
375
|
7484 if (sp->ts_prefixdepth < MAXWLEN)
|
|
|
7485 {
|
|
|
7486 /* There was a prefix before the word. Check that the
|
|
|
7487 * prefix can be used with this word. */
|
|
|
7488 /* Count the length of the NULs in the prefix. If there
|
|
|
7489 * are none this must be the first try without a prefix.
|
|
|
7490 */
|
|
|
7491 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
7492 len = pbyts[n++];
|
|
|
7493 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
7494 ;
|
|
|
7495 if (c > 0)
|
|
|
7496 {
|
|
390
|
7497 /* The prefix ID is stored three bytes above the
|
|
|
7498 * flags. */
|
|
|
7499 c = valid_word_prefix(c, n, flags,
|
|
455
|
7500 tword + splitoff, lp->lp_slang, FALSE);
|
|
375
|
7501 if (c == 0)
|
|
|
7502 break;
|
|
|
7503
|
|
|
7504 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
7505 if (c & WF_RAREPFX)
|
|
|
7506 flags |= WF_RARE;
|
|
|
7507 }
|
|
|
7508 }
|
|
|
7509
|
|
323
|
7510 /*
|
|
|
7511 * Form the word with proper case in preword.
|
|
|
7512 * If there is a word from a previous split, append.
|
|
|
7513 */
|
|
|
7514 tword[sp->ts_twordlen] = NUL;
|
|
|
7515 if (flags & WF_KEEPCAP)
|
|
|
7516 /* Must find the word in the keep-case tree. */
|
|
|
7517 find_keepcap_word(lp->lp_slang, tword + splitoff,
|
|
|
7518 preword + prewordlen);
|
|
|
7519 else
|
|
346
|
7520 {
|
|
323
|
7521 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
7522 * use that for the goodword too. But if the badword is
|
|
|
7523 * allcap and it's only one char long use onecap. */
|
|
|
7524 c = su->su_badflags;
|
|
|
7525 if ((c & WF_ALLCAP)
|
|
|
7526 #ifdef FEAT_MBYTE
|
|
474
|
7527 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
7528 #else
|
|
|
7529 && su->su_badlen == 1
|
|
|
7530 #endif
|
|
|
7531 )
|
|
|
7532 c = WF_ONECAP;
|
|
323
|
7533 make_case_word(tword + splitoff,
|
|
346
|
7534 preword + prewordlen, flags | c);
|
|
|
7535 }
|
|
323
|
7536
|
|
|
7537 /* Don't use a banned word. It may appear again as a good
|
|
|
7538 * word, thus remember it. */
|
|
|
7539 if (flags & WF_BANNED)
|
|
|
7540 {
|
|
|
7541 add_banned(su, preword + prewordlen);
|
|
|
7542 break;
|
|
|
7543 }
|
|
372
|
7544 if (was_banned(su, preword + prewordlen)
|
|
|
7545 || was_banned(su, preword))
|
|
323
|
7546 break;
|
|
|
7547
|
|
|
7548 newscore = 0;
|
|
|
7549 if ((flags & WF_REGION)
|
|
390
|
7550 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
7551 newscore += SCORE_REGION;
|
|
|
7552 if (flags & WF_RARE)
|
|
|
7553 newscore += SCORE_RARE;
|
|
|
7554
|
|
346
|
7555 if (!spell_valid_case(su->su_badflags,
|
|
323
|
7556 captype(preword + prewordlen, NULL)))
|
|
|
7557 newscore += SCORE_ICASE;
|
|
|
7558
|
|
346
|
7559 if ((fword[sp->ts_fidx] == NUL
|
|
376
|
7560 || !spell_iswordp(fword + sp->ts_fidx, curbuf))
|
|
346
|
7561 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
7562 {
|
|
366
|
7563 /* The badword also ends: add suggestions. Give a penalty
|
|
|
7564 * when changing non-word char to word char, e.g., "thes,"
|
|
|
7565 * -> "these". */
|
|
|
7566 p = fword + sp->ts_fidx;
|
|
|
7567 #ifdef FEAT_MBYTE
|
|
|
7568 if (has_mbyte)
|
|
|
7569 mb_ptr_back(fword, p);
|
|
|
7570 else
|
|
|
7571 #endif
|
|
|
7572 --p;
|
|
376
|
7573 if (!spell_iswordp(p, curbuf))
|
|
366
|
7574 {
|
|
|
7575 p = preword + STRLEN(preword);
|
|
|
7576 #ifdef FEAT_MBYTE
|
|
|
7577 if (has_mbyte)
|
|
|
7578 mb_ptr_back(preword, p);
|
|
|
7579 else
|
|
|
7580 #endif
|
|
|
7581 --p;
|
|
376
|
7582 if (spell_iswordp(p, curbuf))
|
|
366
|
7583 newscore += SCORE_NONWORD;
|
|
|
7584 }
|
|
|
7585
|
|
344
|
7586 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
7587 sp->ts_fidx - repextra,
|
|
351
|
7588 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
7589 }
|
|
330
|
7590 else if (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
7591 #ifdef FEAT_MBYTE
|
|
|
7592 /* Don't split halfway a character. */
|
|
|
7593 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
7594 #endif
|
|
|
7595 )
|
|
323
|
7596 {
|
|
|
7597 /* The word in the tree ends but the badword
|
|
|
7598 * continues: try inserting a space and check that a valid
|
|
|
7599 * words starts at fword[sp->ts_fidx]. */
|
|
|
7600 if (try_deeper(su, stack, depth, newscore + SCORE_SPLIT))
|
|
|
7601 {
|
|
|
7602 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
|
7603 sp->ts_save_prewordlen = prewordlen;
|
|
346
|
7604 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7605 sp->ts_save_splitoff = splitoff;
|
|
376
|
7606 sp->ts_state = STATE_SPLITUNDO;
|
|
|
7607
|
|
|
7608 ++depth;
|
|
|
7609 sp = &stack[depth];
|
|
323
|
7610
|
|
|
7611 /* Append a space to preword. */
|
|
|
7612 STRCAT(preword, " ");
|
|
|
7613 prewordlen = STRLEN(preword);
|
|
|
7614 splitoff = sp->ts_twordlen;
|
|
376
|
7615
|
|
|
7616 /* If the badword has a non-word character at this
|
|
|
7617 * position skip it. That means replacing the
|
|
|
7618 * non-word character with a space. */
|
|
|
7619 if (!spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
7620 {
|
|
|
7621 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
7622 #ifdef FEAT_MBYTE
|
|
|
7623 if (has_mbyte)
|
|
|
7624 sp->ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
|
7625 else
|
|
|
7626 #endif
|
|
|
7627 ++sp->ts_fidx;
|
|
|
7628 }
|
|
455
|
7629
|
|
|
7630 /* set su->su_badflags to the caps type at this
|
|
|
7631 * position */
|
|
|
7632
|
|
324
|
7633 #ifdef FEAT_MBYTE
|
|
|
7634 if (has_mbyte)
|
|
455
|
7635 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
7636 else
|
|
|
7637 #endif
|
|
455
|
7638 n = sp->ts_fidx;
|
|
474
|
7639 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
7640 su->su_badptr + su->su_badlen);
|
|
323
|
7641
|
|
|
7642 /* Restart at top of the tree. */
|
|
376
|
7643 sp->ts_arridx = 0;
|
|
323
|
7644 }
|
|
|
7645 }
|
|
|
7646 break;
|
|
|
7647
|
|
|
7648 case STATE_SPLITUNDO:
|
|
346
|
7649 /* Undo the changes done for word split. */
|
|
|
7650 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
7651 splitoff = sp->ts_save_splitoff;
|
|
|
7652 prewordlen = sp->ts_save_prewordlen;
|
|
|
7653
|
|
|
7654 /* Continue looking for NUL bytes. */
|
|
|
7655 sp->ts_state = STATE_START;
|
|
|
7656 break;
|
|
|
7657
|
|
|
7658 case STATE_ENDNUL:
|
|
|
7659 /* Past the NUL bytes in the node. */
|
|
455
|
7660 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
7661 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
7662 {
|
|
|
7663 /* The badword ends, can't use the bytes in this node. */
|
|
|
7664 sp->ts_state = STATE_DEL;
|
|
|
7665 break;
|
|
|
7666 }
|
|
|
7667 sp->ts_state = STATE_PLAIN;
|
|
|
7668 /*FALLTHROUGH*/
|
|
|
7669
|
|
|
7670 case STATE_PLAIN:
|
|
|
7671 /*
|
|
|
7672 * Go over all possible bytes at this node, add each to
|
|
|
7673 * tword[] and use child node. "ts_curi" is the index.
|
|
|
7674 */
|
|
|
7675 arridx = sp->ts_arridx;
|
|
|
7676 if (sp->ts_curi > byts[arridx])
|
|
|
7677 {
|
|
|
7678 /* Done all bytes at this node, do next state. When still
|
|
|
7679 * at already changed bytes skip the other tricks. */
|
|
|
7680 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7681 sp->ts_state = STATE_DEL;
|
|
|
7682 else
|
|
|
7683 sp->ts_state = STATE_FINAL;
|
|
|
7684 }
|
|
|
7685 else
|
|
|
7686 {
|
|
|
7687 arridx += sp->ts_curi++;
|
|
|
7688 c = byts[arridx];
|
|
|
7689
|
|
|
7690 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
7691 * the byte in the bad word adjust the score. But don't
|
|
|
7692 * even try when the byte was already changed. */
|
|
330
|
7693 if (c == fword[sp->ts_fidx]
|
|
|
7694 #ifdef FEAT_MBYTE
|
|
|
7695 || (sp->ts_tcharlen > 0
|
|
|
7696 && sp->ts_isdiff != DIFF_NONE)
|
|
|
7697 #endif
|
|
|
7698 )
|
|
323
|
7699 newscore = 0;
|
|
|
7700 else
|
|
|
7701 newscore = SCORE_SUBST;
|
|
|
7702 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7703 && try_deeper(su, stack, depth, newscore))
|
|
|
7704 {
|
|
|
7705 ++depth;
|
|
330
|
7706 sp = &stack[depth];
|
|
|
7707 ++sp->ts_fidx;
|
|
|
7708 tword[sp->ts_twordlen++] = c;
|
|
|
7709 sp->ts_arridx = idxs[arridx];
|
|
|
7710 #ifdef FEAT_MBYTE
|
|
|
7711 if (newscore == SCORE_SUBST)
|
|
|
7712 sp->ts_isdiff = DIFF_YES;
|
|
|
7713 if (has_mbyte)
|
|
|
7714 {
|
|
|
7715 /* Multi-byte characters are a bit complicated to
|
|
|
7716 * handle: They differ when any of the bytes
|
|
|
7717 * differ and then their length may also differ. */
|
|
|
7718 if (sp->ts_tcharlen == 0)
|
|
|
7719 {
|
|
|
7720 /* First byte. */
|
|
|
7721 sp->ts_tcharidx = 0;
|
|
|
7722 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
7723 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
7724 sp->ts_isdiff = (newscore != 0)
|
|
|
7725 ? DIFF_YES : DIFF_NONE;
|
|
|
7726 }
|
|
|
7727 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
7728 /* When inserting trail bytes don't advance in
|
|
|
7729 * the bad word. */
|
|
|
7730 --sp->ts_fidx;
|
|
|
7731 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
7732 {
|
|
|
7733 /* Last byte of character. */
|
|
|
7734 if (sp->ts_isdiff == DIFF_YES)
|
|
|
7735 {
|
|
|
7736 /* Correct ts_fidx for the byte length of
|
|
|
7737 * the character (we didn't check that
|
|
|
7738 * before). */
|
|
|
7739 sp->ts_fidx = sp->ts_fcharstart
|
|
|
7740 + MB_BYTE2LEN(
|
|
|
7741 fword[sp->ts_fcharstart]);
|
|
|
7742
|
|
|
7743 /* For a similar character adjust score
|
|
|
7744 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
|
7745 if (lp->lp_slang->sl_has_map
|
|
|
7746 && similar_chars(lp->lp_slang,
|
|
|
7747 mb_ptr2char(tword
|
|
|
7748 + sp->ts_twordlen
|
|
|
7749 - sp->ts_tcharlen),
|
|
|
7750 mb_ptr2char(fword
|
|
|
7751 + sp->ts_fcharstart)))
|
|
|
7752 sp->ts_score -=
|
|
|
7753 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7754 }
|
|
358
|
7755 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
7756 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
7757 {
|
|
|
7758 p = tword + sp->ts_twordlen
|
|
|
7759 - sp->ts_tcharlen;
|
|
|
7760 c = mb_ptr2char(p);
|
|
475
|
7761 if (enc_utf8 && utf_iscomposing(c))
|
|
|
7762 {
|
|
|
7763 /* Inserting a composing char doesn't
|
|
|
7764 * count that much. */
|
|
358
|
7765 sp->ts_score -= SCORE_INS
|
|
475
|
7766 - SCORE_INSCOMP;
|
|
|
7767 }
|
|
|
7768 else
|
|
|
7769 {
|
|
|
7770 /* If the previous character was the
|
|
|
7771 * same, thus doubling a character,
|
|
|
7772 * give a bonus to the score. */
|
|
|
7773 mb_ptr_back(tword, p);
|
|
|
7774 if (c == mb_ptr2char(p))
|
|
|
7775 sp->ts_score -= SCORE_INS
|
|
358
|
7776 - SCORE_INSDUP;
|
|
475
|
7777 }
|
|
358
|
7778 }
|
|
330
|
7779
|
|
|
7780 /* Starting a new char, reset the length. */
|
|
|
7781 sp->ts_tcharlen = 0;
|
|
|
7782 }
|
|
|
7783 }
|
|
|
7784 else
|
|
|
7785 #endif
|
|
|
7786 {
|
|
|
7787 /* If we found a similar char adjust the score.
|
|
|
7788 * We do this after calling try_deeper() because
|
|
|
7789 * it's slow. */
|
|
|
7790 if (newscore != 0
|
|
|
7791 && lp->lp_slang->sl_has_map
|
|
|
7792 && similar_chars(lp->lp_slang,
|
|
|
7793 c, fword[sp->ts_fidx - 1]))
|
|
|
7794 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7795 }
|
|
323
|
7796 }
|
|
|
7797 }
|
|
|
7798 break;
|
|
|
7799
|
|
|
7800 case STATE_DEL:
|
|
330
|
7801 #ifdef FEAT_MBYTE
|
|
|
7802 /* When past the first byte of a multi-byte char don't try
|
|
|
7803 * delete/insert/swap a character. */
|
|
|
7804 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
7805 {
|
|
|
7806 sp->ts_state = STATE_FINAL;
|
|
|
7807 break;
|
|
|
7808 }
|
|
|
7809 #endif
|
|
|
7810 /*
|
|
|
7811 * Try skipping one character in the bad word (delete it).
|
|
|
7812 */
|
|
323
|
7813 sp->ts_state = STATE_INS;
|
|
|
7814 sp->ts_curi = 1;
|
|
|
7815 if (fword[sp->ts_fidx] != NUL
|
|
|
7816 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
7817 {
|
|
|
7818 ++depth;
|
|
358
|
7819
|
|
|
7820 /* Advance over the character in fword[]. Give a bonus to
|
|
|
7821 * the score if the same character is following "nn" ->
|
|
|
7822 * "n". */
|
|
330
|
7823 #ifdef FEAT_MBYTE
|
|
|
7824 if (has_mbyte)
|
|
358
|
7825 {
|
|
|
7826 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
7827 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
358
|
7828 if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
|
7829 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7830 }
|
|
330
|
7831 else
|
|
|
7832 #endif
|
|
358
|
7833 {
|
|
330
|
7834 ++stack[depth].ts_fidx;
|
|
358
|
7835 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
7836 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7837 }
|
|
323
|
7838 break;
|
|
|
7839 }
|
|
|
7840 /*FALLTHROUGH*/
|
|
|
7841
|
|
|
7842 case STATE_INS:
|
|
330
|
7843 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
7844 * node. */
|
|
|
7845 n = sp->ts_arridx;
|
|
|
7846 if (sp->ts_curi > byts[n])
|
|
|
7847 {
|
|
|
7848 /* Done all bytes at this node, do next state. */
|
|
|
7849 sp->ts_state = STATE_SWAP;
|
|
|
7850 }
|
|
|
7851 else
|
|
|
7852 {
|
|
330
|
7853 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
7854 n += sp->ts_curi++;
|
|
|
7855 c = byts[n];
|
|
|
7856 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
7857 {
|
|
|
7858 ++depth;
|
|
330
|
7859 sp = &stack[depth];
|
|
|
7860 tword[sp->ts_twordlen++] = c;
|
|
|
7861 sp->ts_arridx = idxs[n];
|
|
|
7862 #ifdef FEAT_MBYTE
|
|
|
7863 if (has_mbyte)
|
|
|
7864 {
|
|
|
7865 fl = MB_BYTE2LEN(c);
|
|
|
7866 if (fl > 1)
|
|
|
7867 {
|
|
|
7868 /* There are following bytes for the same
|
|
|
7869 * character. We must find all bytes before
|
|
|
7870 * trying delete/insert/swap/etc. */
|
|
|
7871 sp->ts_tcharlen = fl;
|
|
|
7872 sp->ts_tcharidx = 1;
|
|
|
7873 sp->ts_isdiff = DIFF_INSERT;
|
|
|
7874 }
|
|
|
7875 }
|
|
358
|
7876 else
|
|
|
7877 fl = 1;
|
|
|
7878 if (fl == 1)
|
|
330
|
7879 #endif
|
|
358
|
7880 {
|
|
|
7881 /* If the previous character was the same, thus
|
|
|
7882 * doubling a character, give a bonus to the
|
|
|
7883 * score. */
|
|
|
7884 if (sp->ts_twordlen >= 2
|
|
|
7885 && tword[sp->ts_twordlen - 2] == c)
|
|
|
7886 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
7887 }
|
|
323
|
7888 }
|
|
|
7889 }
|
|
|
7890 break;
|
|
|
7891
|
|
|
7892 case STATE_SWAP:
|
|
330
|
7893 /*
|
|
|
7894 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
7895 * We change "fword" here, it's changed back afterwards.
|
|
|
7896 */
|
|
|
7897 p = fword + sp->ts_fidx;
|
|
|
7898 c = *p;
|
|
|
7899 if (c == NUL)
|
|
|
7900 {
|
|
|
7901 /* End of word, can't swap or replace. */
|
|
|
7902 sp->ts_state = STATE_FINAL;
|
|
|
7903 break;
|
|
|
7904 }
|
|
|
7905 #ifdef FEAT_MBYTE
|
|
|
7906 if (has_mbyte)
|
|
|
7907 {
|
|
474
|
7908 n = mb_cptr2len(p);
|
|
330
|
7909 c = mb_ptr2char(p);
|
|
|
7910 c2 = mb_ptr2char(p + n);
|
|
|
7911 }
|
|
|
7912 else
|
|
|
7913 #endif
|
|
|
7914 c2 = p[1];
|
|
|
7915 if (c == c2)
|
|
323
|
7916 {
|
|
330
|
7917 /* Characters are identical, swap won't do anything. */
|
|
|
7918 sp->ts_state = STATE_SWAP3;
|
|
|
7919 break;
|
|
|
7920 }
|
|
|
7921 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
7922 {
|
|
|
7923 sp->ts_state = STATE_UNSWAP;
|
|
323
|
7924 ++depth;
|
|
330
|
7925 #ifdef FEAT_MBYTE
|
|
|
7926 if (has_mbyte)
|
|
|
7927 {
|
|
|
7928 fl = mb_char2len(c2);
|
|
|
7929 mch_memmove(p, p + n, fl);
|
|
|
7930 mb_char2bytes(c, p + fl);
|
|
|
7931 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7932 }
|
|
|
7933 else
|
|
|
7934 #endif
|
|
|
7935 {
|
|
|
7936 p[0] = c2;
|
|
|
7937 p[1] = c;
|
|
|
7938 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
7939 }
|
|
323
|
7940 }
|
|
|
7941 else
|
|
|
7942 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
7943 sp->ts_state = STATE_REP_INI;
|
|
|
7944 break;
|
|
|
7945
|
|
330
|
7946 case STATE_UNSWAP:
|
|
|
7947 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
7948 p = fword + sp->ts_fidx;
|
|
|
7949 #ifdef FEAT_MBYTE
|
|
|
7950 if (has_mbyte)
|
|
|
7951 {
|
|
|
7952 n = MB_BYTE2LEN(*p);
|
|
|
7953 c = mb_ptr2char(p + n);
|
|
|
7954 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
7955 mb_char2bytes(c, p);
|
|
|
7956 }
|
|
|
7957 else
|
|
|
7958 #endif
|
|
|
7959 {
|
|
|
7960 c = *p;
|
|
|
7961 *p = p[1];
|
|
|
7962 p[1] = c;
|
|
|
7963 }
|
|
|
7964 /*FALLTHROUGH*/
|
|
|
7965
|
|
|
7966 case STATE_SWAP3:
|
|
323
|
7967 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
7968 * "fword" here, it's changed back afterwards. */
|
|
|
7969 p = fword + sp->ts_fidx;
|
|
|
7970 #ifdef FEAT_MBYTE
|
|
|
7971 if (has_mbyte)
|
|
|
7972 {
|
|
474
|
7973 n = mb_cptr2len(p);
|
|
330
|
7974 c = mb_ptr2char(p);
|
|
474
|
7975 fl = mb_cptr2len(p + n);
|
|
330
|
7976 c2 = mb_ptr2char(p + n);
|
|
|
7977 c3 = mb_ptr2char(p + n + fl);
|
|
|
7978 }
|
|
|
7979 else
|
|
|
7980 #endif
|
|
323
|
7981 {
|
|
330
|
7982 c = *p;
|
|
|
7983 c2 = p[1];
|
|
|
7984 c3 = p[2];
|
|
|
7985 }
|
|
|
7986
|
|
|
7987 /* When characters are identical: "121" then SWAP3 result is
|
|
|
7988 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
7989 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
7990 * swapping. Also skip when c3 is NUL. */
|
|
|
7991 if (c == c3 || c3 == NUL)
|
|
|
7992 {
|
|
|
7993 sp->ts_state = STATE_REP_INI;
|
|
|
7994 break;
|
|
|
7995 }
|
|
|
7996 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7997 {
|
|
|
7998 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
7999 ++depth;
|
|
330
|
8000 #ifdef FEAT_MBYTE
|
|
|
8001 if (has_mbyte)
|
|
|
8002 {
|
|
|
8003 tl = mb_char2len(c3);
|
|
|
8004 mch_memmove(p, p + n + fl, tl);
|
|
|
8005 mb_char2bytes(c2, p + tl);
|
|
|
8006 mb_char2bytes(c, p + fl + tl);
|
|
|
8007 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
8008 }
|
|
|
8009 else
|
|
|
8010 #endif
|
|
|
8011 {
|
|
|
8012 p[0] = p[2];
|
|
|
8013 p[2] = c;
|
|
|
8014 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8015 }
|
|
323
|
8016 }
|
|
|
8017 else
|
|
|
8018 sp->ts_state = STATE_REP_INI;
|
|
|
8019 break;
|
|
|
8020
|
|
330
|
8021 case STATE_UNSWAP3:
|
|
|
8022 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
8023 p = fword + sp->ts_fidx;
|
|
|
8024 #ifdef FEAT_MBYTE
|
|
|
8025 if (has_mbyte)
|
|
|
8026 {
|
|
|
8027 n = MB_BYTE2LEN(*p);
|
|
|
8028 c2 = mb_ptr2char(p + n);
|
|
|
8029 fl = MB_BYTE2LEN(p[n]);
|
|
|
8030 c = mb_ptr2char(p + n + fl);
|
|
|
8031 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
8032 mch_memmove(p + fl + tl, p, n);
|
|
|
8033 mb_char2bytes(c, p);
|
|
|
8034 mb_char2bytes(c2, p + tl);
|
|
|
8035 }
|
|
|
8036 else
|
|
|
8037 #endif
|
|
|
8038 {
|
|
|
8039 c = *p;
|
|
|
8040 *p = p[2];
|
|
|
8041 p[2] = c;
|
|
|
8042 }
|
|
346
|
8043
|
|
330
|
8044 /* Rotate three characters left: "123" -> "231". We change
|
|
|
8045 * "fword" here, it's changed back afterwards. */
|
|
323
|
8046 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
8047 {
|
|
330
|
8048 sp->ts_state = STATE_UNROT3L;
|
|
323
|
8049 ++depth;
|
|
330
|
8050 p = fword + sp->ts_fidx;
|
|
|
8051 #ifdef FEAT_MBYTE
|
|
|
8052 if (has_mbyte)
|
|
|
8053 {
|
|
474
|
8054 n = mb_cptr2len(p);
|
|
330
|
8055 c = mb_ptr2char(p);
|
|
474
|
8056 fl = mb_cptr2len(p + n);
|
|
|
8057 fl += mb_cptr2len(p + n + fl);
|
|
330
|
8058 mch_memmove(p, p + n, fl);
|
|
|
8059 mb_char2bytes(c, p + fl);
|
|
|
8060 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
8061 }
|
|
|
8062 else
|
|
|
8063 #endif
|
|
|
8064 {
|
|
|
8065 c = *p;
|
|
|
8066 *p = p[1];
|
|
|
8067 p[1] = p[2];
|
|
|
8068 p[2] = c;
|
|
|
8069 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8070 }
|
|
323
|
8071 }
|
|
|
8072 else
|
|
|
8073 sp->ts_state = STATE_REP_INI;
|
|
|
8074 break;
|
|
|
8075
|
|
330
|
8076 case STATE_UNROT3L:
|
|
346
|
8077 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
8078 p = fword + sp->ts_fidx;
|
|
|
8079 #ifdef FEAT_MBYTE
|
|
|
8080 if (has_mbyte)
|
|
|
8081 {
|
|
|
8082 n = MB_BYTE2LEN(*p);
|
|
|
8083 n += MB_BYTE2LEN(p[n]);
|
|
|
8084 c = mb_ptr2char(p + n);
|
|
|
8085 tl = MB_BYTE2LEN(p[n]);
|
|
|
8086 mch_memmove(p + tl, p, n);
|
|
|
8087 mb_char2bytes(c, p);
|
|
|
8088 }
|
|
|
8089 else
|
|
|
8090 #endif
|
|
|
8091 {
|
|
|
8092 c = p[2];
|
|
|
8093 p[2] = p[1];
|
|
|
8094 p[1] = *p;
|
|
|
8095 *p = c;
|
|
|
8096 }
|
|
346
|
8097
|
|
323
|
8098 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
8099 * "fword" here, it's changed back afterwards. */
|
|
323
|
8100 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
8101 {
|
|
330
|
8102 sp->ts_state = STATE_UNROT3R;
|
|
323
|
8103 ++depth;
|
|
330
|
8104 p = fword + sp->ts_fidx;
|
|
|
8105 #ifdef FEAT_MBYTE
|
|
|
8106 if (has_mbyte)
|
|
|
8107 {
|
|
474
|
8108 n = mb_cptr2len(p);
|
|
|
8109 n += mb_cptr2len(p + n);
|
|
330
|
8110 c = mb_ptr2char(p + n);
|
|
474
|
8111 tl = mb_cptr2len(p + n);
|
|
330
|
8112 mch_memmove(p + tl, p, n);
|
|
|
8113 mb_char2bytes(c, p);
|
|
|
8114 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
8115 }
|
|
|
8116 else
|
|
|
8117 #endif
|
|
|
8118 {
|
|
|
8119 c = p[2];
|
|
|
8120 p[2] = p[1];
|
|
|
8121 p[1] = *p;
|
|
|
8122 *p = c;
|
|
|
8123 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8124 }
|
|
323
|
8125 }
|
|
|
8126 else
|
|
|
8127 sp->ts_state = STATE_REP_INI;
|
|
|
8128 break;
|
|
|
8129
|
|
330
|
8130 case STATE_UNROT3R:
|
|
346
|
8131 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
8132 p = fword + sp->ts_fidx;
|
|
|
8133 #ifdef FEAT_MBYTE
|
|
|
8134 if (has_mbyte)
|
|
|
8135 {
|
|
|
8136 c = mb_ptr2char(p);
|
|
|
8137 tl = MB_BYTE2LEN(*p);
|
|
|
8138 n = MB_BYTE2LEN(p[tl]);
|
|
|
8139 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
8140 mch_memmove(p, p + tl, n);
|
|
|
8141 mb_char2bytes(c, p + n);
|
|
|
8142 }
|
|
|
8143 else
|
|
|
8144 #endif
|
|
|
8145 {
|
|
|
8146 c = *p;
|
|
|
8147 *p = p[1];
|
|
|
8148 p[1] = p[2];
|
|
|
8149 p[2] = c;
|
|
|
8150 }
|
|
323
|
8151 /*FALLTHROUGH*/
|
|
|
8152
|
|
|
8153 case STATE_REP_INI:
|
|
|
8154 /* Check if matching with REP items from the .aff file would
|
|
|
8155 * work. Quickly skip if there are no REP items or the score
|
|
|
8156 * is going to be too high anyway. */
|
|
|
8157 gap = &lp->lp_slang->sl_rep;
|
|
|
8158 if (gap->ga_len == 0
|
|
|
8159 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
8160 {
|
|
|
8161 sp->ts_state = STATE_FINAL;
|
|
|
8162 break;
|
|
|
8163 }
|
|
|
8164
|
|
|
8165 /* Use the first byte to quickly find the first entry that
|
|
330
|
8166 * may match. If the index is -1 there is none. */
|
|
323
|
8167 sp->ts_curi = lp->lp_slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
8168 if (sp->ts_curi < 0)
|
|
|
8169 {
|
|
|
8170 sp->ts_state = STATE_FINAL;
|
|
|
8171 break;
|
|
|
8172 }
|
|
|
8173
|
|
|
8174 sp->ts_state = STATE_REP;
|
|
|
8175 /*FALLTHROUGH*/
|
|
|
8176
|
|
|
8177 case STATE_REP:
|
|
|
8178 /* Try matching with REP items from the .aff file. For each
|
|
330
|
8179 * match replace the characters and check if the resulting
|
|
|
8180 * word is valid. */
|
|
323
|
8181 p = fword + sp->ts_fidx;
|
|
|
8182
|
|
|
8183 gap = &lp->lp_slang->sl_rep;
|
|
|
8184 while (sp->ts_curi < gap->ga_len)
|
|
|
8185 {
|
|
|
8186 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
8187 if (*ftp->ft_from != *p)
|
|
|
8188 {
|
|
|
8189 /* past possible matching entries */
|
|
|
8190 sp->ts_curi = gap->ga_len;
|
|
|
8191 break;
|
|
|
8192 }
|
|
|
8193 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
8194 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
8195 {
|
|
|
8196 /* Need to undo this afterwards. */
|
|
|
8197 sp->ts_state = STATE_REP_UNDO;
|
|
|
8198
|
|
|
8199 /* Change the "from" to the "to" string. */
|
|
|
8200 ++depth;
|
|
|
8201 fl = STRLEN(ftp->ft_from);
|
|
|
8202 tl = STRLEN(ftp->ft_to);
|
|
|
8203 if (fl != tl)
|
|
346
|
8204 {
|
|
323
|
8205 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
8206 repextra += tl - fl;
|
|
|
8207 }
|
|
323
|
8208 mch_memmove(p, ftp->ft_to, tl);
|
|
|
8209 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
8210 #ifdef FEAT_MBYTE
|
|
|
8211 stack[depth].ts_tcharlen = 0;
|
|
|
8212 #endif
|
|
323
|
8213 break;
|
|
|
8214 }
|
|
|
8215 }
|
|
|
8216
|
|
|
8217 if (sp->ts_curi >= gap->ga_len)
|
|
|
8218 /* No (more) matches. */
|
|
|
8219 sp->ts_state = STATE_FINAL;
|
|
|
8220
|
|
|
8221 break;
|
|
|
8222
|
|
|
8223 case STATE_REP_UNDO:
|
|
|
8224 /* Undo a REP replacement and continue with the next one. */
|
|
|
8225 ftp = (fromto_T *)lp->lp_slang->sl_rep.ga_data
|
|
|
8226 + sp->ts_curi - 1;
|
|
|
8227 fl = STRLEN(ftp->ft_from);
|
|
|
8228 tl = STRLEN(ftp->ft_to);
|
|
|
8229 p = fword + sp->ts_fidx;
|
|
|
8230 if (fl != tl)
|
|
346
|
8231 {
|
|
323
|
8232 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
8233 repextra -= tl - fl;
|
|
|
8234 }
|
|
323
|
8235 mch_memmove(p, ftp->ft_from, fl);
|
|
|
8236 sp->ts_state = STATE_REP;
|
|
|
8237 break;
|
|
|
8238
|
|
|
8239 default:
|
|
|
8240 /* Did all possible states at this level, go up one level. */
|
|
|
8241 --depth;
|
|
344
|
8242
|
|
375
|
8243 if (depth >= 0 && stack[depth].ts_prefixdepth == PREFIXTREE)
|
|
|
8244 {
|
|
|
8245 /* Continue in or go back to the prefix tree. */
|
|
|
8246 byts = pbyts;
|
|
|
8247 idxs = pidxs;
|
|
|
8248 splitoff = 0;
|
|
|
8249 }
|
|
|
8250
|
|
344
|
8251 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
8252 line_breakcheck();
|
|
323
|
8253 }
|
|
|
8254 }
|
|
|
8255 }
|
|
|
8256 }
|
|
|
8257
|
|
|
8258 /*
|
|
|
8259 * Try going one level deeper in the tree.
|
|
|
8260 */
|
|
|
8261 static int
|
|
|
8262 try_deeper(su, stack, depth, score_add)
|
|
|
8263 suginfo_T *su;
|
|
|
8264 trystate_T *stack;
|
|
|
8265 int depth;
|
|
|
8266 int score_add;
|
|
|
8267 {
|
|
|
8268 int newscore;
|
|
|
8269
|
|
|
8270 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
8271 newscore = stack[depth].ts_score + score_add;
|
|
|
8272 if (newscore >= su->su_maxscore)
|
|
|
8273 return FALSE;
|
|
|
8274
|
|
330
|
8275 stack[depth + 1] = stack[depth];
|
|
323
|
8276 stack[depth + 1].ts_state = STATE_START;
|
|
|
8277 stack[depth + 1].ts_score = newscore;
|
|
|
8278 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
8279 return TRUE;
|
|
|
8280 }
|
|
|
8281
|
|
455
|
8282 #ifdef FEAT_MBYTE
|
|
|
8283 /*
|
|
|
8284 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
8285 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
8286 */
|
|
|
8287 static int
|
|
|
8288 nofold_len(fword, flen, word)
|
|
|
8289 char_u *fword;
|
|
|
8290 int flen;
|
|
|
8291 char_u *word;
|
|
|
8292 {
|
|
|
8293 char_u *p;
|
|
|
8294 int i = 0;
|
|
|
8295
|
|
|
8296 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
8297 ++i;
|
|
|
8298 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
8299 --i;
|
|
|
8300 return (int)(p - word);
|
|
|
8301 }
|
|
|
8302 #endif
|
|
|
8303
|
|
323
|
8304 /*
|
|
|
8305 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
8306 * words and put it in "kword".
|
|
|
8307 * Theoretically there could be several keep-case words that result in the
|
|
|
8308 * same case-folded word, but we only find one...
|
|
|
8309 */
|
|
|
8310 static void
|
|
|
8311 find_keepcap_word(slang, fword, kword)
|
|
|
8312 slang_T *slang;
|
|
|
8313 char_u *fword;
|
|
|
8314 char_u *kword;
|
|
|
8315 {
|
|
|
8316 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
8317 int depth;
|
|
324
|
8318 idx_T tryidx;
|
|
323
|
8319
|
|
|
8320 /* The following arrays are used at each depth in the tree. */
|
|
324
|
8321 idx_T arridx[MAXWLEN];
|
|
323
|
8322 int round[MAXWLEN];
|
|
|
8323 int fwordidx[MAXWLEN];
|
|
|
8324 int uwordidx[MAXWLEN];
|
|
|
8325 int kwordlen[MAXWLEN];
|
|
|
8326
|
|
|
8327 int flen, ulen;
|
|
|
8328 int l;
|
|
|
8329 int len;
|
|
|
8330 int c;
|
|
324
|
8331 idx_T lo, hi, m;
|
|
323
|
8332 char_u *p;
|
|
|
8333 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
8334 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
8335
|
|
|
8336 if (byts == NULL)
|
|
|
8337 {
|
|
|
8338 /* array is empty: "cannot happen" */
|
|
|
8339 *kword = NUL;
|
|
|
8340 return;
|
|
|
8341 }
|
|
|
8342
|
|
|
8343 /* Make an all-cap version of "fword". */
|
|
|
8344 allcap_copy(fword, uword);
|
|
|
8345
|
|
|
8346 /*
|
|
|
8347 * Each character needs to be tried both case-folded and upper-case.
|
|
|
8348 * All this gets very complicated if we keep in mind that changing case
|
|
|
8349 * may change the byte length of a multi-byte character...
|
|
|
8350 */
|
|
|
8351 depth = 0;
|
|
|
8352 arridx[0] = 0;
|
|
|
8353 round[0] = 0;
|
|
|
8354 fwordidx[0] = 0;
|
|
|
8355 uwordidx[0] = 0;
|
|
|
8356 kwordlen[0] = 0;
|
|
|
8357 while (depth >= 0)
|
|
|
8358 {
|
|
|
8359 if (fword[fwordidx[depth]] == NUL)
|
|
|
8360 {
|
|
|
8361 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
8362 * here we have found a match. */
|
|
|
8363 if (byts[arridx[depth] + 1] == 0)
|
|
|
8364 {
|
|
|
8365 kword[kwordlen[depth]] = NUL;
|
|
|
8366 return;
|
|
|
8367 }
|
|
|
8368
|
|
|
8369 /* kword is getting too long, continue one level up */
|
|
|
8370 --depth;
|
|
|
8371 }
|
|
|
8372 else if (++round[depth] > 2)
|
|
|
8373 {
|
|
|
8374 /* tried both fold-case and upper-case character, continue one
|
|
|
8375 * level up */
|
|
|
8376 --depth;
|
|
|
8377 }
|
|
|
8378 else
|
|
|
8379 {
|
|
|
8380 /*
|
|
|
8381 * round[depth] == 1: Try using the folded-case character.
|
|
|
8382 * round[depth] == 2: Try using the upper-case character.
|
|
|
8383 */
|
|
|
8384 #ifdef FEAT_MBYTE
|
|
|
8385 if (has_mbyte)
|
|
|
8386 {
|
|
474
|
8387 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
8388 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
8389 }
|
|
|
8390 else
|
|
|
8391 #endif
|
|
|
8392 ulen = flen = 1;
|
|
|
8393 if (round[depth] == 1)
|
|
|
8394 {
|
|
|
8395 p = fword + fwordidx[depth];
|
|
|
8396 l = flen;
|
|
|
8397 }
|
|
|
8398 else
|
|
|
8399 {
|
|
|
8400 p = uword + uwordidx[depth];
|
|
|
8401 l = ulen;
|
|
|
8402 }
|
|
|
8403
|
|
|
8404 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
8405 {
|
|
|
8406 /* Perform a binary search in the list of accepted bytes. */
|
|
|
8407 len = byts[tryidx++];
|
|
|
8408 c = *p++;
|
|
|
8409 lo = tryidx;
|
|
|
8410 hi = tryidx + len - 1;
|
|
|
8411 while (lo < hi)
|
|
|
8412 {
|
|
|
8413 m = (lo + hi) / 2;
|
|
|
8414 if (byts[m] > c)
|
|
|
8415 hi = m - 1;
|
|
|
8416 else if (byts[m] < c)
|
|
|
8417 lo = m + 1;
|
|
|
8418 else
|
|
|
8419 {
|
|
|
8420 lo = hi = m;
|
|
|
8421 break;
|
|
|
8422 }
|
|
|
8423 }
|
|
|
8424
|
|
|
8425 /* Stop if there is no matching byte. */
|
|
|
8426 if (hi < lo || byts[lo] != c)
|
|
|
8427 break;
|
|
|
8428
|
|
|
8429 /* Continue at the child (if there is one). */
|
|
|
8430 tryidx = idxs[lo];
|
|
|
8431 }
|
|
|
8432
|
|
|
8433 if (l == 0)
|
|
|
8434 {
|
|
|
8435 /*
|
|
|
8436 * Found the matching char. Copy it to "kword" and go a
|
|
|
8437 * level deeper.
|
|
|
8438 */
|
|
|
8439 if (round[depth] == 1)
|
|
|
8440 {
|
|
|
8441 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
8442 flen);
|
|
|
8443 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
8444 }
|
|
|
8445 else
|
|
|
8446 {
|
|
|
8447 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
8448 ulen);
|
|
|
8449 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
8450 }
|
|
|
8451 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
8452 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
8453
|
|
|
8454 ++depth;
|
|
|
8455 arridx[depth] = tryidx;
|
|
|
8456 round[depth] = 0;
|
|
|
8457 }
|
|
|
8458 }
|
|
|
8459 }
|
|
|
8460
|
|
|
8461 /* Didn't find it: "cannot happen". */
|
|
|
8462 *kword = NUL;
|
|
|
8463 }
|
|
|
8464
|
|
|
8465 /*
|
|
344
|
8466 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
8467 * su->su_sga.
|
|
|
8468 */
|
|
|
8469 static void
|
|
|
8470 score_comp_sal(su)
|
|
|
8471 suginfo_T *su;
|
|
|
8472 {
|
|
|
8473 langp_T *lp;
|
|
|
8474 char_u badsound[MAXWLEN];
|
|
|
8475 int i;
|
|
|
8476 suggest_T *stp;
|
|
|
8477 suggest_T *sstp;
|
|
|
8478 int score;
|
|
|
8479
|
|
|
8480 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
8481 return;
|
|
|
8482
|
|
|
8483 /* Use the sound-folding of the first language that supports it. */
|
|
|
8484 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8485 lp->lp_slang != NULL; ++lp)
|
|
|
8486 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8487 {
|
|
|
8488 /* soundfold the bad word */
|
|
375
|
8489 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8490
|
|
|
8491 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8492 {
|
|
|
8493 stp = &SUG(su->su_ga, i);
|
|
|
8494
|
|
351
|
8495 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
8496 * sound-a-like score. */
|
|
|
8497 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
8498 if (score < SCORE_MAXMAX)
|
|
|
8499 {
|
|
|
8500 /* Add the suggestion. */
|
|
|
8501 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
8502 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
8503 if (sstp->st_word != NULL)
|
|
|
8504 {
|
|
|
8505 sstp->st_score = score;
|
|
|
8506 sstp->st_altscore = 0;
|
|
|
8507 sstp->st_orglen = stp->st_orglen;
|
|
|
8508 ++su->su_sga.ga_len;
|
|
|
8509 }
|
|
|
8510 }
|
|
|
8511 }
|
|
|
8512 break;
|
|
|
8513 }
|
|
|
8514 }
|
|
|
8515
|
|
|
8516 /*
|
|
|
8517 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
8518 * They are intwined.
|
|
|
8519 */
|
|
|
8520 static void
|
|
|
8521 score_combine(su)
|
|
|
8522 suginfo_T *su;
|
|
|
8523 {
|
|
|
8524 int i;
|
|
|
8525 int j;
|
|
|
8526 garray_T ga;
|
|
|
8527 garray_T *gap;
|
|
|
8528 langp_T *lp;
|
|
|
8529 suggest_T *stp;
|
|
|
8530 char_u *p;
|
|
|
8531 char_u badsound[MAXWLEN];
|
|
|
8532 int round;
|
|
|
8533
|
|
|
8534 /* Add the alternate score to su_ga. */
|
|
|
8535 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8536 lp->lp_slang != NULL; ++lp)
|
|
|
8537 {
|
|
|
8538 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8539 {
|
|
|
8540 /* soundfold the bad word */
|
|
375
|
8541 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8542
|
|
|
8543 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8544 {
|
|
|
8545 stp = &SUG(su->su_ga, i);
|
|
351
|
8546 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
8547 badsound);
|
|
344
|
8548 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8549 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
8550 else
|
|
|
8551 stp->st_score = (stp->st_score * 3
|
|
|
8552 + stp->st_altscore) / 4;
|
|
|
8553 stp->st_salscore = FALSE;
|
|
|
8554 }
|
|
|
8555 break;
|
|
|
8556 }
|
|
|
8557 }
|
|
|
8558
|
|
|
8559 /* Add the alternate score to su_sga. */
|
|
|
8560 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8561 {
|
|
|
8562 stp = &SUG(su->su_sga, i);
|
|
|
8563 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
8564 if (stp->st_score == SCORE_MAXMAX)
|
|
|
8565 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
8566 else
|
|
|
8567 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
8568 stp->st_salscore = TRUE;
|
|
|
8569 }
|
|
|
8570
|
|
|
8571 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
8572 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8573 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
8574
|
|
|
8575 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
8576 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
8577 return;
|
|
|
8578
|
|
|
8579 stp = &SUG(ga, 0);
|
|
|
8580 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
8581 {
|
|
|
8582 /* round 1: get a suggestion from su_ga
|
|
|
8583 * round 2: get a suggestion from su_sga */
|
|
|
8584 for (round = 1; round <= 2; ++round)
|
|
|
8585 {
|
|
|
8586 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
8587 if (i < gap->ga_len)
|
|
|
8588 {
|
|
|
8589 /* Don't add a word if it's already there. */
|
|
|
8590 p = SUG(*gap, i).st_word;
|
|
|
8591 for (j = 0; j < ga.ga_len; ++j)
|
|
|
8592 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
8593 break;
|
|
|
8594 if (j == ga.ga_len)
|
|
|
8595 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
8596 else
|
|
|
8597 vim_free(p);
|
|
|
8598 }
|
|
|
8599 }
|
|
|
8600 }
|
|
|
8601
|
|
|
8602 ga_clear(&su->su_ga);
|
|
|
8603 ga_clear(&su->su_sga);
|
|
|
8604
|
|
|
8605 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
8606 if (ga.ga_len > su->su_maxcount)
|
|
|
8607 {
|
|
|
8608 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
8609 vim_free(stp[i].st_word);
|
|
|
8610 ga.ga_len = su->su_maxcount;
|
|
|
8611 }
|
|
|
8612
|
|
|
8613 su->su_ga = ga;
|
|
|
8614 }
|
|
|
8615
|
|
|
8616 /*
|
|
351
|
8617 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
8618 * badword.
|
|
|
8619 */
|
|
|
8620 static int
|
|
|
8621 stp_sal_score(stp, su, slang, badsound)
|
|
|
8622 suggest_T *stp;
|
|
|
8623 suginfo_T *su;
|
|
|
8624 slang_T *slang;
|
|
|
8625 char_u *badsound; /* sound-folded badword */
|
|
|
8626 {
|
|
|
8627 char_u *p;
|
|
|
8628 char_u badsound2[MAXWLEN];
|
|
|
8629 char_u fword[MAXWLEN];
|
|
|
8630 char_u goodsound[MAXWLEN];
|
|
|
8631
|
|
|
8632 if (stp->st_orglen <= su->su_badlen)
|
|
|
8633 p = badsound;
|
|
|
8634 else
|
|
|
8635 {
|
|
|
8636 /* soundfold the bad word with more characters following */
|
|
|
8637 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
8638
|
|
|
8639 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
8640 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
8641 * removing the space. Don't do it when the good word also contains a
|
|
|
8642 * space. */
|
|
|
8643 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
8644 && *skiptowhite(stp->st_word) == NUL)
|
|
|
8645 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
8646 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
8647
|
|
375
|
8648 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
8649 p = badsound2;
|
|
|
8650 }
|
|
|
8651
|
|
375
|
8652 /* Sound-fold the word and compute the score for the difference. */
|
|
|
8653 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
8654
|
|
|
8655 return soundalike_score(goodsound, p);
|
|
|
8656 }
|
|
|
8657
|
|
|
8658 /*
|
|
323
|
8659 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
8660 */
|
|
|
8661 static void
|
|
346
|
8662 suggest_try_soundalike(su)
|
|
323
|
8663 suginfo_T *su;
|
|
|
8664 {
|
|
|
8665 char_u salword[MAXWLEN];
|
|
|
8666 char_u tword[MAXWLEN];
|
|
|
8667 char_u tsalword[MAXWLEN];
|
|
324
|
8668 idx_T arridx[MAXWLEN];
|
|
323
|
8669 int curi[MAXWLEN];
|
|
|
8670 langp_T *lp;
|
|
|
8671 char_u *byts;
|
|
324
|
8672 idx_T *idxs;
|
|
323
|
8673 int depth;
|
|
|
8674 int c;
|
|
324
|
8675 idx_T n;
|
|
323
|
8676 int round;
|
|
|
8677 int flags;
|
|
344
|
8678 int sound_score;
|
|
|
8679
|
|
|
8680 /* Do this for all languages that support sound folding. */
|
|
323
|
8681 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8682 lp->lp_slang != NULL; ++lp)
|
|
|
8683 {
|
|
|
8684 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8685 {
|
|
|
8686 /* soundfold the bad word */
|
|
375
|
8687 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
8688
|
|
|
8689 /*
|
|
|
8690 * Go through the whole tree, soundfold each word and compare.
|
|
|
8691 * round 1: use the case-folded tree.
|
|
|
8692 * round 2: use the keep-case tree.
|
|
|
8693 */
|
|
|
8694 for (round = 1; round <= 2; ++round)
|
|
|
8695 {
|
|
|
8696 if (round == 1)
|
|
|
8697 {
|
|
|
8698 byts = lp->lp_slang->sl_fbyts;
|
|
|
8699 idxs = lp->lp_slang->sl_fidxs;
|
|
|
8700 }
|
|
|
8701 else
|
|
|
8702 {
|
|
|
8703 byts = lp->lp_slang->sl_kbyts;
|
|
|
8704 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
8705 if (byts == NULL) /* no keep-case words */
|
|
|
8706 continue;
|
|
323
|
8707 }
|
|
|
8708
|
|
|
8709 depth = 0;
|
|
|
8710 arridx[0] = 0;
|
|
|
8711 curi[0] = 1;
|
|
|
8712 while (depth >= 0 && !got_int)
|
|
|
8713 {
|
|
|
8714 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
8715 {
|
|
323
|
8716 /* Done all bytes at this node, go up one level. */
|
|
|
8717 --depth;
|
|
351
|
8718 line_breakcheck();
|
|
|
8719 }
|
|
323
|
8720 else
|
|
|
8721 {
|
|
|
8722 /* Do one more byte at this node. */
|
|
|
8723 n = arridx[depth] + curi[depth];
|
|
|
8724 ++curi[depth];
|
|
|
8725 c = byts[n];
|
|
|
8726 if (c == 0)
|
|
|
8727 {
|
|
|
8728 /* End of word, deal with the word. */
|
|
324
|
8729 flags = (int)idxs[n];
|
|
323
|
8730 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
8731 {
|
|
|
8732 tword[depth] = NUL;
|
|
375
|
8733 /* Sound-fold. Only in keep-case tree need to
|
|
|
8734 * case-fold the word. */
|
|
|
8735 spell_soundfold(lp->lp_slang, tword,
|
|
|
8736 round == 1, tsalword);
|
|
323
|
8737
|
|
344
|
8738 /* Compute the edit distance between the
|
|
|
8739 * sound-a-like words. */
|
|
|
8740 sound_score = soundalike_score(salword,
|
|
|
8741 tsalword);
|
|
324
|
8742 if (sound_score < SCORE_MAXMAX)
|
|
323
|
8743 {
|
|
324
|
8744 char_u cword[MAXWLEN];
|
|
|
8745 char_u *p;
|
|
344
|
8746 int score;
|
|
324
|
8747
|
|
388
|
8748 flags |= su->su_badflags;
|
|
351
|
8749 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
8750 {
|
|
324
|
8751 /* Need to fix case according to
|
|
|
8752 * "flags". */
|
|
323
|
8753 make_case_word(tword, cword, flags);
|
|
324
|
8754 p = cword;
|
|
323
|
8755 }
|
|
|
8756 else
|
|
324
|
8757 p = tword;
|
|
|
8758
|
|
344
|
8759 if (sps_flags & SPS_DOUBLE)
|
|
|
8760 add_suggestion(su, &su->su_sga, p,
|
|
346
|
8761 su->su_badlen,
|
|
351
|
8762 sound_score, 0, FALSE);
|
|
344
|
8763 else
|
|
|
8764 {
|
|
|
8765 /* Compute the score. */
|
|
|
8766 score = spell_edit_score(
|
|
|
8767 su->su_badword, p);
|
|
|
8768 if (sps_flags & SPS_BEST)
|
|
|
8769 /* give a bonus for the good word
|
|
|
8770 * sounding the same as the bad
|
|
|
8771 * word */
|
|
|
8772 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8773 su->su_badlen,
|
|
344
|
8774 RESCORE(score, sound_score),
|
|
351
|
8775 sound_score, TRUE);
|
|
344
|
8776 else
|
|
|
8777 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8778 su->su_badlen,
|
|
351
|
8779 score + sound_score, 0, FALSE);
|
|
344
|
8780 }
|
|
323
|
8781 }
|
|
|
8782 }
|
|
|
8783
|
|
|
8784 /* Skip over other NUL bytes. */
|
|
|
8785 while (byts[n + 1] == 0)
|
|
|
8786 {
|
|
|
8787 ++n;
|
|
|
8788 ++curi[depth];
|
|
|
8789 }
|
|
|
8790 }
|
|
|
8791 else
|
|
|
8792 {
|
|
|
8793 /* Normal char, go one level deeper. */
|
|
|
8794 tword[depth++] = c;
|
|
|
8795 arridx[depth] = idxs[n];
|
|
|
8796 curi[depth] = 1;
|
|
|
8797 }
|
|
|
8798 }
|
|
|
8799 }
|
|
|
8800 }
|
|
|
8801 }
|
|
|
8802 }
|
|
|
8803 }
|
|
|
8804
|
|
|
8805 /*
|
|
324
|
8806 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
8807 */
|
|
|
8808 static void
|
|
|
8809 make_case_word(fword, cword, flags)
|
|
|
8810 char_u *fword;
|
|
|
8811 char_u *cword;
|
|
|
8812 int flags;
|
|
|
8813 {
|
|
|
8814 if (flags & WF_ALLCAP)
|
|
|
8815 /* Make it all upper-case */
|
|
|
8816 allcap_copy(fword, cword);
|
|
|
8817 else if (flags & WF_ONECAP)
|
|
|
8818 /* Make the first letter upper-case */
|
|
324
|
8819 onecap_copy(fword, cword, TRUE);
|
|
323
|
8820 else
|
|
|
8821 /* Use goodword as-is. */
|
|
|
8822 STRCPY(cword, fword);
|
|
|
8823 }
|
|
|
8824
|
|
330
|
8825 /*
|
|
|
8826 * Use map string "map" for languages "lp".
|
|
|
8827 */
|
|
|
8828 static void
|
|
|
8829 set_map_str(lp, map)
|
|
|
8830 slang_T *lp;
|
|
|
8831 char_u *map;
|
|
|
8832 {
|
|
|
8833 char_u *p;
|
|
|
8834 int headc = 0;
|
|
|
8835 int c;
|
|
|
8836 int i;
|
|
|
8837
|
|
|
8838 if (*map == NUL)
|
|
|
8839 {
|
|
|
8840 lp->sl_has_map = FALSE;
|
|
|
8841 return;
|
|
|
8842 }
|
|
|
8843 lp->sl_has_map = TRUE;
|
|
|
8844
|
|
|
8845 /* Init the array and hash table empty. */
|
|
|
8846 for (i = 0; i < 256; ++i)
|
|
|
8847 lp->sl_map_array[i] = 0;
|
|
|
8848 #ifdef FEAT_MBYTE
|
|
|
8849 hash_init(&lp->sl_map_hash);
|
|
|
8850 #endif
|
|
|
8851
|
|
|
8852 /*
|
|
|
8853 * The similar characters are stored separated with slashes:
|
|
|
8854 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
8855 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
8856 */
|
|
|
8857 for (p = map; *p != NUL; )
|
|
|
8858 {
|
|
|
8859 #ifdef FEAT_MBYTE
|
|
474
|
8860 c = mb_cptr2char_adv(&p);
|
|
330
|
8861 #else
|
|
|
8862 c = *p++;
|
|
|
8863 #endif
|
|
|
8864 if (c == '/')
|
|
|
8865 headc = 0;
|
|
|
8866 else
|
|
|
8867 {
|
|
|
8868 if (headc == 0)
|
|
|
8869 headc = c;
|
|
|
8870
|
|
|
8871 #ifdef FEAT_MBYTE
|
|
|
8872 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
8873 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
8874 * a NUL. */
|
|
|
8875 if (c >= 256)
|
|
|
8876 {
|
|
|
8877 int cl = mb_char2len(c);
|
|
|
8878 int headcl = mb_char2len(headc);
|
|
|
8879 char_u *b;
|
|
|
8880 hash_T hash;
|
|
|
8881 hashitem_T *hi;
|
|
|
8882
|
|
|
8883 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
8884 if (b == NULL)
|
|
|
8885 return;
|
|
|
8886 mb_char2bytes(c, b);
|
|
|
8887 b[cl] = NUL;
|
|
|
8888 mb_char2bytes(headc, b + cl + 1);
|
|
|
8889 b[cl + 1 + headcl] = NUL;
|
|
|
8890 hash = hash_hash(b);
|
|
|
8891 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
8892 if (HASHITEM_EMPTY(hi))
|
|
|
8893 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
8894 else
|
|
|
8895 {
|
|
|
8896 /* This should have been checked when generating the .spl
|
|
|
8897 * file. */
|
|
|
8898 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
8899 vim_free(b);
|
|
|
8900 }
|
|
|
8901 }
|
|
|
8902 else
|
|
|
8903 #endif
|
|
|
8904 lp->sl_map_array[c] = headc;
|
|
|
8905 }
|
|
|
8906 }
|
|
|
8907 }
|
|
|
8908
|
|
323
|
8909 /*
|
|
|
8910 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
8911 * lines in the .aff file.
|
|
|
8912 */
|
|
|
8913 static int
|
|
|
8914 similar_chars(slang, c1, c2)
|
|
|
8915 slang_T *slang;
|
|
|
8916 int c1;
|
|
|
8917 int c2;
|
|
|
8918 {
|
|
330
|
8919 int m1, m2;
|
|
|
8920 #ifdef FEAT_MBYTE
|
|
|
8921 char_u buf[MB_MAXBYTES];
|
|
|
8922 hashitem_T *hi;
|
|
|
8923
|
|
|
8924 if (c1 >= 256)
|
|
|
8925 {
|
|
|
8926 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
8927 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8928 if (HASHITEM_EMPTY(hi))
|
|
|
8929 m1 = 0;
|
|
|
8930 else
|
|
|
8931 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8932 }
|
|
|
8933 else
|
|
|
8934 #endif
|
|
|
8935 m1 = slang->sl_map_array[c1];
|
|
|
8936 if (m1 == 0)
|
|
323
|
8937 return FALSE;
|
|
330
|
8938
|
|
|
8939
|
|
|
8940 #ifdef FEAT_MBYTE
|
|
|
8941 if (c2 >= 256)
|
|
|
8942 {
|
|
|
8943 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
8944 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8945 if (HASHITEM_EMPTY(hi))
|
|
|
8946 m2 = 0;
|
|
|
8947 else
|
|
|
8948 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8949 }
|
|
|
8950 else
|
|
|
8951 #endif
|
|
|
8952 m2 = slang->sl_map_array[c2];
|
|
|
8953
|
|
|
8954 return m1 == m2;
|
|
323
|
8955 }
|
|
|
8956
|
|
|
8957 /*
|
|
|
8958 * Add a suggestion to the list of suggestions.
|
|
|
8959 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
8960 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
8961 * with spell_edit_score().
|
|
|
8962 */
|
|
|
8963 static void
|
|
351
|
8964 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
8965 suginfo_T *su;
|
|
344
|
8966 garray_T *gap;
|
|
323
|
8967 char_u *goodword;
|
|
346
|
8968 int badlen; /* length of bad word used */
|
|
324
|
8969 int score;
|
|
351
|
8970 int altscore;
|
|
344
|
8971 int had_bonus; /* value for st_had_bonus */
|
|
323
|
8972 {
|
|
|
8973 suggest_T *stp;
|
|
|
8974 int i;
|
|
346
|
8975 char_u *p = NULL;
|
|
|
8976 int c = 0;
|
|
323
|
8977
|
|
|
8978 /* Check that the word wasn't banned. */
|
|
|
8979 if (was_banned(su, goodword))
|
|
|
8980 return;
|
|
|
8981
|
|
346
|
8982 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
8983 * Remove the common part from "goodword". */
|
|
|
8984 i = badlen - su->su_badlen;
|
|
|
8985 if (i > 0)
|
|
|
8986 {
|
|
|
8987 /* This assumes there was no case folding or it didn't change the
|
|
|
8988 * length... */
|
|
|
8989 p = goodword + STRLEN(goodword) - i;
|
|
|
8990 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
8991 {
|
|
|
8992 badlen = su->su_badlen;
|
|
|
8993 c = *p;
|
|
|
8994 *p = NUL;
|
|
|
8995 }
|
|
|
8996 else
|
|
|
8997 p = NULL;
|
|
|
8998 }
|
|
372
|
8999 else if (i < 0)
|
|
|
9000 {
|
|
|
9001 /* When replacing part of the word check that we actually change
|
|
|
9002 * something. For "the the" a suggestion can be replacing the first
|
|
|
9003 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
9004 if (badlen == (int)STRLEN(goodword)
|
|
372
|
9005 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
9006 return;
|
|
|
9007 }
|
|
|
9008
|
|
346
|
9009
|
|
323
|
9010 if (score <= su->su_maxscore)
|
|
|
9011 {
|
|
366
|
9012 /* Check if the word is already there. Also check the length that is
|
|
|
9013 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
9014 * "thes" -> "these". */
|
|
344
|
9015 stp = &SUG(*gap, 0);
|
|
|
9016 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
9017 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
9018 && stp[i].st_orglen == badlen)
|
|
323
|
9019 {
|
|
|
9020 /* Found it. Remember the lowest score. */
|
|
|
9021 if (stp[i].st_score > score)
|
|
324
|
9022 {
|
|
323
|
9023 stp[i].st_score = score;
|
|
376
|
9024 stp[i].st_altscore = altscore;
|
|
324
|
9025 stp[i].st_had_bonus = had_bonus;
|
|
|
9026 }
|
|
323
|
9027 break;
|
|
|
9028 }
|
|
|
9029
|
|
344
|
9030 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
9031 {
|
|
|
9032 /* Add a suggestion. */
|
|
344
|
9033 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
9034 stp->st_word = vim_strsave(goodword);
|
|
|
9035 if (stp->st_word != NULL)
|
|
|
9036 {
|
|
|
9037 stp->st_score = score;
|
|
351
|
9038 stp->st_altscore = altscore;
|
|
324
|
9039 stp->st_had_bonus = had_bonus;
|
|
346
|
9040 stp->st_orglen = badlen;
|
|
344
|
9041 ++gap->ga_len;
|
|
323
|
9042
|
|
|
9043 /* If we have too many suggestions now, sort the list and keep
|
|
|
9044 * the best suggestions. */
|
|
344
|
9045 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
9046 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
9047 SUG_CLEAN_COUNT(su));
|
|
323
|
9048 }
|
|
|
9049 }
|
|
|
9050 }
|
|
346
|
9051
|
|
|
9052 if (p != NULL)
|
|
|
9053 *p = c; /* restore "goodword" */
|
|
323
|
9054 }
|
|
|
9055
|
|
|
9056 /*
|
|
|
9057 * Add a word to be banned.
|
|
|
9058 */
|
|
|
9059 static void
|
|
|
9060 add_banned(su, word)
|
|
|
9061 suginfo_T *su;
|
|
|
9062 char_u *word;
|
|
|
9063 {
|
|
|
9064 char_u *s = vim_strsave(word);
|
|
|
9065 hash_T hash;
|
|
|
9066 hashitem_T *hi;
|
|
|
9067
|
|
|
9068 if (s != NULL)
|
|
|
9069 {
|
|
|
9070 hash = hash_hash(s);
|
|
|
9071 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
9072 if (HASHITEM_EMPTY(hi))
|
|
|
9073 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
9074 else
|
|
|
9075 vim_free(s);
|
|
323
|
9076 }
|
|
|
9077 }
|
|
|
9078
|
|
|
9079 /*
|
|
|
9080 * Return TRUE if a word appears in the list of banned words.
|
|
|
9081 */
|
|
|
9082 static int
|
|
|
9083 was_banned(su, word)
|
|
|
9084 suginfo_T *su;
|
|
|
9085 char_u *word;
|
|
|
9086 {
|
|
324
|
9087 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
9088
|
|
|
9089 return !HASHITEM_EMPTY(hi);
|
|
323
|
9090 }
|
|
|
9091
|
|
|
9092 /*
|
|
|
9093 * Free the banned words in "su".
|
|
|
9094 */
|
|
|
9095 static void
|
|
|
9096 free_banned(su)
|
|
|
9097 suginfo_T *su;
|
|
|
9098 {
|
|
|
9099 int todo;
|
|
|
9100 hashitem_T *hi;
|
|
|
9101
|
|
|
9102 todo = su->su_banned.ht_used;
|
|
|
9103 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
9104 {
|
|
|
9105 if (!HASHITEM_EMPTY(hi))
|
|
|
9106 {
|
|
|
9107 vim_free(hi->hi_key);
|
|
|
9108 --todo;
|
|
|
9109 }
|
|
|
9110 }
|
|
|
9111 hash_clear(&su->su_banned);
|
|
|
9112 }
|
|
|
9113
|
|
324
|
9114 /*
|
|
|
9115 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
9116 * thus only done for the final results.
|
|
|
9117 */
|
|
|
9118 static void
|
|
|
9119 rescore_suggestions(su)
|
|
|
9120 suginfo_T *su;
|
|
|
9121 {
|
|
|
9122 langp_T *lp;
|
|
|
9123 suggest_T *stp;
|
|
|
9124 char_u sal_badword[MAXWLEN];
|
|
|
9125 int i;
|
|
|
9126
|
|
|
9127 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9128 lp->lp_slang != NULL; ++lp)
|
|
|
9129 {
|
|
|
9130 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9131 {
|
|
|
9132 /* soundfold the bad word */
|
|
375
|
9133 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
9134
|
|
|
9135 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9136 {
|
|
344
|
9137 stp = &SUG(su->su_ga, i);
|
|
324
|
9138 if (!stp->st_had_bonus)
|
|
|
9139 {
|
|
351
|
9140 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
9141 lp->lp_slang, sal_badword);
|
|
|
9142 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
9143 stp->st_altscore = SCORE_BIG;
|
|
|
9144 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
9145 }
|
|
|
9146 }
|
|
|
9147 break;
|
|
|
9148 }
|
|
|
9149 }
|
|
|
9150 }
|
|
|
9151
|
|
323
|
9152 static int
|
|
|
9153 #ifdef __BORLANDC__
|
|
|
9154 _RTLENTRYF
|
|
|
9155 #endif
|
|
|
9156 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
9157
|
|
|
9158 /*
|
|
|
9159 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
9160 */
|
|
|
9161 static int
|
|
|
9162 #ifdef __BORLANDC__
|
|
|
9163 _RTLENTRYF
|
|
|
9164 #endif
|
|
|
9165 sug_compare(s1, s2)
|
|
|
9166 const void *s1;
|
|
|
9167 const void *s2;
|
|
|
9168 {
|
|
|
9169 suggest_T *p1 = (suggest_T *)s1;
|
|
|
9170 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
9171 int n = p1->st_score - p2->st_score;
|
|
|
9172
|
|
|
9173 if (n == 0)
|
|
|
9174 return p1->st_altscore - p2->st_altscore;
|
|
|
9175 return n;
|
|
323
|
9176 }
|
|
|
9177
|
|
|
9178 /*
|
|
|
9179 * Cleanup the suggestions:
|
|
|
9180 * - Sort on score.
|
|
|
9181 * - Remove words that won't be displayed.
|
|
344
|
9182 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
9183 */
|
|
344
|
9184 static int
|
|
|
9185 cleanup_suggestions(gap, maxscore, keep)
|
|
|
9186 garray_T *gap;
|
|
|
9187 int maxscore;
|
|
324
|
9188 int keep; /* nr of suggestions to keep */
|
|
323
|
9189 {
|
|
344
|
9190 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
9191 int i;
|
|
|
9192
|
|
|
9193 /* Sort the list. */
|
|
344
|
9194 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
9195
|
|
|
9196 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
9197 if (gap->ga_len > keep)
|
|
323
|
9198 {
|
|
344
|
9199 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
9200 vim_free(stp[i].st_word);
|
|
344
|
9201 gap->ga_len = keep;
|
|
|
9202 return stp[keep - 1].st_score;
|
|
323
|
9203 }
|
|
344
|
9204 return maxscore;
|
|
323
|
9205 }
|
|
|
9206
|
|
372
|
9207 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
9208 /*
|
|
|
9209 * Soundfold a string, for soundfold().
|
|
|
9210 * Result is in allocated memory, NULL for an error.
|
|
|
9211 */
|
|
|
9212 char_u *
|
|
|
9213 eval_soundfold(word)
|
|
|
9214 char_u *word;
|
|
|
9215 {
|
|
|
9216 langp_T *lp;
|
|
|
9217 char_u sound[MAXWLEN];
|
|
|
9218
|
|
|
9219 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
9220 /* Use the sound-folding of the first language that supports it. */
|
|
|
9221 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9222 lp->lp_slang != NULL; ++lp)
|
|
|
9223 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9224 {
|
|
|
9225 /* soundfold the word */
|
|
375
|
9226 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
9227 return vim_strsave(sound);
|
|
|
9228 }
|
|
|
9229
|
|
|
9230 /* No language with sound folding, return word as-is. */
|
|
|
9231 return vim_strsave(word);
|
|
|
9232 }
|
|
|
9233 #endif
|
|
|
9234
|
|
323
|
9235 /*
|
|
|
9236 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
9237 */
|
|
|
9238 static void
|
|
375
|
9239 spell_soundfold(slang, inword, folded, res)
|
|
|
9240 slang_T *slang;
|
|
|
9241 char_u *inword;
|
|
|
9242 int folded; /* "inword" is already case-folded */
|
|
|
9243 char_u *res;
|
|
|
9244 {
|
|
|
9245 char_u fword[MAXWLEN];
|
|
|
9246 char_u *word;
|
|
|
9247
|
|
|
9248 if (slang->sl_sofo)
|
|
|
9249 /* SOFOFROM and SOFOTO used */
|
|
|
9250 spell_soundfold_sofo(slang, inword, res);
|
|
|
9251 else
|
|
|
9252 {
|
|
|
9253 /* SAL items used. Requires the word to be case-folded. */
|
|
|
9254 if (folded)
|
|
|
9255 word = inword;
|
|
|
9256 else
|
|
|
9257 {
|
|
|
9258 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
9259 word = fword;
|
|
|
9260 }
|
|
|
9261
|
|
|
9262 #ifdef FEAT_MBYTE
|
|
|
9263 if (has_mbyte)
|
|
|
9264 spell_soundfold_wsal(slang, word, res);
|
|
|
9265 else
|
|
|
9266 #endif
|
|
|
9267 spell_soundfold_sal(slang, word, res);
|
|
|
9268 }
|
|
|
9269 }
|
|
|
9270
|
|
|
9271 /*
|
|
|
9272 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
9273 * SOFOTO lines.
|
|
|
9274 */
|
|
|
9275 static void
|
|
|
9276 spell_soundfold_sofo(slang, inword, res)
|
|
|
9277 slang_T *slang;
|
|
|
9278 char_u *inword;
|
|
|
9279 char_u *res;
|
|
|
9280 {
|
|
|
9281 char_u *s;
|
|
|
9282 int ri = 0;
|
|
|
9283 int c;
|
|
|
9284
|
|
|
9285 #ifdef FEAT_MBYTE
|
|
|
9286 if (has_mbyte)
|
|
|
9287 {
|
|
|
9288 int prevc = 0;
|
|
|
9289 int *ip;
|
|
|
9290
|
|
|
9291 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
9292 * 255, sl_sal the rest. */
|
|
|
9293 for (s = inword; *s != NUL; )
|
|
|
9294 {
|
|
474
|
9295 c = mb_cptr2char_adv(&s);
|
|
375
|
9296 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9297 c = ' ';
|
|
|
9298 else if (c < 256)
|
|
|
9299 c = slang->sl_sal_first[c];
|
|
|
9300 else
|
|
|
9301 {
|
|
|
9302 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
9303 if (ip == NULL) /* empty list, can't match */
|
|
|
9304 c = NUL;
|
|
|
9305 else
|
|
|
9306 for (;;) /* find "c" in the list */
|
|
|
9307 {
|
|
|
9308 if (*ip == 0) /* not found */
|
|
|
9309 {
|
|
|
9310 c = NUL;
|
|
|
9311 break;
|
|
|
9312 }
|
|
|
9313 if (*ip == c) /* match! */
|
|
|
9314 {
|
|
|
9315 c = ip[1];
|
|
|
9316 break;
|
|
|
9317 }
|
|
|
9318 ip += 2;
|
|
|
9319 }
|
|
|
9320 }
|
|
|
9321
|
|
|
9322 if (c != NUL && c != prevc)
|
|
|
9323 {
|
|
|
9324 ri += mb_char2bytes(c, res + ri);
|
|
|
9325 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
9326 break;
|
|
|
9327 prevc = c;
|
|
|
9328 }
|
|
|
9329 }
|
|
|
9330 }
|
|
|
9331 else
|
|
|
9332 #endif
|
|
|
9333 {
|
|
|
9334 /* The sl_sal_first[] table contains the translation. */
|
|
|
9335 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
9336 {
|
|
|
9337 if (vim_iswhite(c))
|
|
|
9338 c = ' ';
|
|
|
9339 else
|
|
|
9340 c = slang->sl_sal_first[c];
|
|
|
9341 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
9342 res[ri++] = c;
|
|
|
9343 }
|
|
|
9344 }
|
|
|
9345
|
|
|
9346 res[ri] = NUL;
|
|
|
9347 }
|
|
|
9348
|
|
|
9349 static void
|
|
|
9350 spell_soundfold_sal(slang, inword, res)
|
|
323
|
9351 slang_T *slang;
|
|
|
9352 char_u *inword;
|
|
|
9353 char_u *res;
|
|
|
9354 {
|
|
344
|
9355 salitem_T *smp;
|
|
323
|
9356 char_u word[MAXWLEN];
|
|
375
|
9357 char_u *s = inword;
|
|
323
|
9358 char_u *t;
|
|
344
|
9359 char_u *pf;
|
|
323
|
9360 int i, j, z;
|
|
344
|
9361 int reslen;
|
|
323
|
9362 int n, k = 0;
|
|
|
9363 int z0;
|
|
|
9364 int k0;
|
|
|
9365 int n0;
|
|
|
9366 int c;
|
|
|
9367 int pri;
|
|
|
9368 int p0 = -333;
|
|
|
9369 int c0;
|
|
|
9370
|
|
324
|
9371 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
9372 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
9373 if (slang->sl_rem_accents)
|
|
|
9374 {
|
|
|
9375 t = word;
|
|
375
|
9376 while (*s != NUL)
|
|
323
|
9377 {
|
|
324
|
9378 if (vim_iswhite(*s))
|
|
344
|
9379 {
|
|
|
9380 *t++ = ' ';
|
|
|
9381 s = skipwhite(s);
|
|
|
9382 }
|
|
323
|
9383 else
|
|
|
9384 {
|
|
376
|
9385 if (spell_iswordp_nmw(s))
|
|
323
|
9386 *t++ = *s;
|
|
|
9387 ++s;
|
|
|
9388 }
|
|
|
9389 }
|
|
|
9390 *t = NUL;
|
|
|
9391 }
|
|
|
9392 else
|
|
375
|
9393 STRCPY(word, s);
|
|
323
|
9394
|
|
344
|
9395 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
9396
|
|
|
9397 /*
|
|
|
9398 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
9399 * Changed to keep spaces.
|
|
323
|
9400 */
|
|
344
|
9401 i = reslen = z = 0;
|
|
323
|
9402 while ((c = word[i]) != NUL)
|
|
|
9403 {
|
|
344
|
9404 /* Start with the first rule that has the character in the word. */
|
|
323
|
9405 n = slang->sl_sal_first[c];
|
|
|
9406 z0 = 0;
|
|
|
9407
|
|
|
9408 if (n >= 0)
|
|
|
9409 {
|
|
|
9410 /* check all rules for the same letter */
|
|
344
|
9411 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
9412 {
|
|
344
|
9413 /* Quickly skip entries that don't match the word. Most
|
|
|
9414 * entries are less then three chars, optimize for that. */
|
|
|
9415 k = smp[n].sm_leadlen;
|
|
|
9416 if (k > 1)
|
|
323
|
9417 {
|
|
344
|
9418 if (word[i + 1] != s[1])
|
|
|
9419 continue;
|
|
|
9420 if (k > 2)
|
|
|
9421 {
|
|
|
9422 for (j = 2; j < k; ++j)
|
|
|
9423 if (word[i + j] != s[j])
|
|
|
9424 break;
|
|
|
9425 if (j < k)
|
|
|
9426 continue;
|
|
|
9427 }
|
|
323
|
9428 }
|
|
|
9429
|
|
375
|
9430 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
9431 {
|
|
375
|
9432 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
9433 while (*pf != NUL && *pf != word[i + k])
|
|
|
9434 ++pf;
|
|
|
9435 if (*pf == NUL)
|
|
|
9436 continue;
|
|
|
9437 ++k;
|
|
323
|
9438 }
|
|
344
|
9439 s = smp[n].sm_rules;
|
|
|
9440 pri = 5; /* default priority */
|
|
323
|
9441
|
|
|
9442 p0 = *s;
|
|
|
9443 k0 = k;
|
|
|
9444 while (*s == '-' && k > 1)
|
|
|
9445 {
|
|
|
9446 k--;
|
|
|
9447 s++;
|
|
|
9448 }
|
|
|
9449 if (*s == '<')
|
|
|
9450 s++;
|
|
344
|
9451 if (VIM_ISDIGIT(*s))
|
|
323
|
9452 {
|
|
|
9453 /* determine priority */
|
|
|
9454 pri = *s - '0';
|
|
|
9455 s++;
|
|
|
9456 }
|
|
|
9457 if (*s == '^' && *(s + 1) == '^')
|
|
|
9458 s++;
|
|
|
9459
|
|
|
9460 if (*s == NUL
|
|
|
9461 || (*s == '^'
|
|
324
|
9462 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9463 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
9464 && (*(s + 1) != '$'
|
|
376
|
9465 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9466 || (*s == '$' && i > 0
|
|
376
|
9467 && spell_iswordp(word + i - 1, curbuf)
|
|
|
9468 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9469 {
|
|
|
9470 /* search for followup rules, if: */
|
|
|
9471 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9472 c0 = word[i + k - 1];
|
|
|
9473 n0 = slang->sl_sal_first[c0];
|
|
|
9474
|
|
|
9475 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
9476 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
9477 {
|
|
|
9478 /* test follow-up rule for "word[i + k]" */
|
|
344
|
9479 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
9480 {
|
|
344
|
9481 /* Quickly skip entries that don't match the word.
|
|
|
9482 * */
|
|
|
9483 k0 = smp[n0].sm_leadlen;
|
|
|
9484 if (k0 > 1)
|
|
323
|
9485 {
|
|
344
|
9486 if (word[i + k] != s[1])
|
|
|
9487 continue;
|
|
|
9488 if (k0 > 2)
|
|
|
9489 {
|
|
|
9490 pf = word + i + k + 1;
|
|
|
9491 for (j = 2; j < k0; ++j)
|
|
|
9492 if (*pf++ != s[j])
|
|
|
9493 break;
|
|
|
9494 if (j < k0)
|
|
|
9495 continue;
|
|
|
9496 }
|
|
323
|
9497 }
|
|
344
|
9498 k0 += k - 1;
|
|
|
9499
|
|
375
|
9500 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
9501 {
|
|
344
|
9502 /* Check for match with one of the chars in
|
|
375
|
9503 * "sm_oneof". */
|
|
344
|
9504 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9505 ++pf;
|
|
|
9506 if (*pf == NUL)
|
|
|
9507 continue;
|
|
|
9508 ++k0;
|
|
323
|
9509 }
|
|
344
|
9510
|
|
|
9511 p0 = 5;
|
|
|
9512 s = smp[n0].sm_rules;
|
|
323
|
9513 while (*s == '-')
|
|
|
9514 {
|
|
344
|
9515 /* "k0" gets NOT reduced because
|
|
|
9516 * "if (k0 == k)" */
|
|
323
|
9517 s++;
|
|
|
9518 }
|
|
|
9519 if (*s == '<')
|
|
|
9520 s++;
|
|
344
|
9521 if (VIM_ISDIGIT(*s))
|
|
323
|
9522 {
|
|
|
9523 p0 = *s - '0';
|
|
|
9524 s++;
|
|
|
9525 }
|
|
|
9526
|
|
|
9527 if (*s == NUL
|
|
|
9528 /* *s == '^' cuts */
|
|
|
9529 || (*s == '$'
|
|
376
|
9530 && !spell_iswordp(word + i + k0,
|
|
|
9531 curbuf)))
|
|
323
|
9532 {
|
|
|
9533 if (k0 == k)
|
|
|
9534 /* this is just a piece of the string */
|
|
|
9535 continue;
|
|
|
9536
|
|
|
9537 if (p0 < pri)
|
|
|
9538 /* priority too low */
|
|
|
9539 continue;
|
|
|
9540 /* rule fits; stop search */
|
|
|
9541 break;
|
|
|
9542 }
|
|
|
9543 }
|
|
|
9544
|
|
344
|
9545 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
9546 continue;
|
|
|
9547 }
|
|
|
9548
|
|
|
9549 /* replace string */
|
|
344
|
9550 s = smp[n].sm_to;
|
|
389
|
9551 if (s == NULL)
|
|
|
9552 s = (char_u *)"";
|
|
344
|
9553 pf = smp[n].sm_rules;
|
|
|
9554 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
9555 if (p0 == 1 && z == 0)
|
|
|
9556 {
|
|
|
9557 /* rule with '<' is used */
|
|
344
|
9558 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
9559 || res[reslen - 1] == *s))
|
|
|
9560 reslen--;
|
|
323
|
9561 z0 = 1;
|
|
|
9562 z = 1;
|
|
|
9563 k0 = 0;
|
|
372
|
9564 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
9565 {
|
|
|
9566 word[i + k0] = *s;
|
|
|
9567 k0++;
|
|
|
9568 s++;
|
|
|
9569 }
|
|
|
9570 if (k > k0)
|
|
|
9571 mch_memmove(word + i + k0, word + i + k,
|
|
|
9572 STRLEN(word + i + k) + 1);
|
|
|
9573
|
|
|
9574 /* new "actual letter" */
|
|
|
9575 c = word[i];
|
|
|
9576 }
|
|
|
9577 else
|
|
|
9578 {
|
|
|
9579 /* no '<' rule used */
|
|
|
9580 i += k - 1;
|
|
|
9581 z = 0;
|
|
344
|
9582 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
9583 {
|
|
344
|
9584 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
9585 res[reslen++] = *s;
|
|
323
|
9586 s++;
|
|
|
9587 }
|
|
|
9588 /* new "actual letter" */
|
|
|
9589 c = *s;
|
|
344
|
9590 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
9591 {
|
|
|
9592 if (c != NUL)
|
|
372
|
9593 res[reslen++] = c;
|
|
323
|
9594 mch_memmove(word, word + i + 1,
|
|
|
9595 STRLEN(word + i + 1) + 1);
|
|
|
9596 i = 0;
|
|
|
9597 z0 = 1;
|
|
|
9598 }
|
|
|
9599 }
|
|
|
9600 break;
|
|
|
9601 }
|
|
|
9602 }
|
|
|
9603 }
|
|
324
|
9604 else if (vim_iswhite(c))
|
|
|
9605 {
|
|
|
9606 c = ' ';
|
|
|
9607 k = 1;
|
|
|
9608 }
|
|
323
|
9609
|
|
|
9610 if (z0 == 0)
|
|
|
9611 {
|
|
344
|
9612 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9613 && (!slang->sl_collapse || reslen == 0
|
|
|
9614 || res[reslen - 1] != c))
|
|
323
|
9615 /* condense only double letters */
|
|
372
|
9616 res[reslen++] = c;
|
|
323
|
9617
|
|
|
9618 i++;
|
|
|
9619 z = 0;
|
|
|
9620 k = 0;
|
|
|
9621 }
|
|
|
9622 }
|
|
|
9623
|
|
344
|
9624 res[reslen] = NUL;
|
|
323
|
9625 }
|
|
|
9626
|
|
372
|
9627 #ifdef FEAT_MBYTE
|
|
|
9628 /*
|
|
|
9629 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
9630 * Multi-byte version of spell_soundfold().
|
|
|
9631 */
|
|
|
9632 static void
|
|
375
|
9633 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
9634 slang_T *slang;
|
|
|
9635 char_u *inword;
|
|
|
9636 char_u *res;
|
|
|
9637 {
|
|
375
|
9638 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
9639 int word[MAXWLEN];
|
|
|
9640 int wres[MAXWLEN];
|
|
|
9641 int l;
|
|
|
9642 char_u *s;
|
|
|
9643 int *ws;
|
|
|
9644 char_u *t;
|
|
|
9645 int *pf;
|
|
|
9646 int i, j, z;
|
|
|
9647 int reslen;
|
|
|
9648 int n, k = 0;
|
|
|
9649 int z0;
|
|
|
9650 int k0;
|
|
|
9651 int n0;
|
|
|
9652 int c;
|
|
|
9653 int pri;
|
|
|
9654 int p0 = -333;
|
|
|
9655 int c0;
|
|
|
9656 int did_white = FALSE;
|
|
|
9657
|
|
|
9658 /*
|
|
|
9659 * Convert the multi-byte string to a wide-character string.
|
|
|
9660 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
9661 * But keep white space.
|
|
|
9662 */
|
|
|
9663 n = 0;
|
|
|
9664 for (s = inword; *s != NUL; )
|
|
|
9665 {
|
|
|
9666 t = s;
|
|
474
|
9667 c = mb_cptr2char_adv(&s);
|
|
372
|
9668 if (slang->sl_rem_accents)
|
|
|
9669 {
|
|
|
9670 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9671 {
|
|
|
9672 if (did_white)
|
|
|
9673 continue;
|
|
|
9674 c = ' ';
|
|
|
9675 did_white = TRUE;
|
|
|
9676 }
|
|
|
9677 else
|
|
|
9678 {
|
|
|
9679 did_white = FALSE;
|
|
376
|
9680 if (!spell_iswordp_nmw(t))
|
|
372
|
9681 continue;
|
|
|
9682 }
|
|
|
9683 }
|
|
|
9684 word[n++] = c;
|
|
|
9685 }
|
|
|
9686 word[n] = NUL;
|
|
|
9687
|
|
|
9688 /*
|
|
|
9689 * This comes from Aspell phonet.cpp.
|
|
|
9690 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
9691 * Changed to keep spaces.
|
|
|
9692 */
|
|
|
9693 i = reslen = z = 0;
|
|
|
9694 while ((c = word[i]) != NUL)
|
|
|
9695 {
|
|
|
9696 /* Start with the first rule that has the character in the word. */
|
|
|
9697 n = slang->sl_sal_first[c & 0xff];
|
|
|
9698 z0 = 0;
|
|
|
9699
|
|
|
9700 if (n >= 0)
|
|
|
9701 {
|
|
375
|
9702 /* check all rules for the same index byte */
|
|
372
|
9703 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
9704 {
|
|
|
9705 /* Quickly skip entries that don't match the word. Most
|
|
|
9706 * entries are less then three chars, optimize for that. */
|
|
375
|
9707 if (c != ws[0])
|
|
|
9708 continue;
|
|
372
|
9709 k = smp[n].sm_leadlen;
|
|
|
9710 if (k > 1)
|
|
|
9711 {
|
|
|
9712 if (word[i + 1] != ws[1])
|
|
|
9713 continue;
|
|
|
9714 if (k > 2)
|
|
|
9715 {
|
|
|
9716 for (j = 2; j < k; ++j)
|
|
|
9717 if (word[i + j] != ws[j])
|
|
|
9718 break;
|
|
|
9719 if (j < k)
|
|
|
9720 continue;
|
|
|
9721 }
|
|
|
9722 }
|
|
|
9723
|
|
375
|
9724 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
9725 {
|
|
375
|
9726 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
9727 while (*pf != NUL && *pf != word[i + k])
|
|
|
9728 ++pf;
|
|
|
9729 if (*pf == NUL)
|
|
|
9730 continue;
|
|
|
9731 ++k;
|
|
|
9732 }
|
|
|
9733 s = smp[n].sm_rules;
|
|
|
9734 pri = 5; /* default priority */
|
|
|
9735
|
|
|
9736 p0 = *s;
|
|
|
9737 k0 = k;
|
|
|
9738 while (*s == '-' && k > 1)
|
|
|
9739 {
|
|
|
9740 k--;
|
|
|
9741 s++;
|
|
|
9742 }
|
|
|
9743 if (*s == '<')
|
|
|
9744 s++;
|
|
|
9745 if (VIM_ISDIGIT(*s))
|
|
|
9746 {
|
|
|
9747 /* determine priority */
|
|
|
9748 pri = *s - '0';
|
|
|
9749 s++;
|
|
|
9750 }
|
|
|
9751 if (*s == '^' && *(s + 1) == '^')
|
|
|
9752 s++;
|
|
|
9753
|
|
|
9754 if (*s == NUL
|
|
|
9755 || (*s == '^'
|
|
|
9756 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9757 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
9758 && (*(s + 1) != '$'
|
|
376
|
9759 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9760 || (*s == '$' && i > 0
|
|
376
|
9761 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
9762 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9763 {
|
|
|
9764 /* search for followup rules, if: */
|
|
|
9765 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9766 c0 = word[i + k - 1];
|
|
|
9767 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
9768
|
|
|
9769 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
9770 && p0 != '-' && word[i + k] != NUL)
|
|
|
9771 {
|
|
375
|
9772 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
9773 * all entries with the same index byte. */
|
|
372
|
9774 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
9775 == (c0 & 0xff); ++n0)
|
|
|
9776 {
|
|
|
9777 /* Quickly skip entries that don't match the word.
|
|
375
|
9778 */
|
|
|
9779 if (c0 != ws[0])
|
|
|
9780 continue;
|
|
372
|
9781 k0 = smp[n0].sm_leadlen;
|
|
|
9782 if (k0 > 1)
|
|
|
9783 {
|
|
|
9784 if (word[i + k] != ws[1])
|
|
|
9785 continue;
|
|
|
9786 if (k0 > 2)
|
|
|
9787 {
|
|
|
9788 pf = word + i + k + 1;
|
|
|
9789 for (j = 2; j < k0; ++j)
|
|
|
9790 if (*pf++ != ws[j])
|
|
|
9791 break;
|
|
|
9792 if (j < k0)
|
|
|
9793 continue;
|
|
|
9794 }
|
|
|
9795 }
|
|
|
9796 k0 += k - 1;
|
|
|
9797
|
|
375
|
9798 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
9799 {
|
|
|
9800 /* Check for match with one of the chars in
|
|
375
|
9801 * "sm_oneof". */
|
|
372
|
9802 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9803 ++pf;
|
|
|
9804 if (*pf == NUL)
|
|
|
9805 continue;
|
|
|
9806 ++k0;
|
|
|
9807 }
|
|
|
9808
|
|
|
9809 p0 = 5;
|
|
|
9810 s = smp[n0].sm_rules;
|
|
|
9811 while (*s == '-')
|
|
|
9812 {
|
|
|
9813 /* "k0" gets NOT reduced because
|
|
|
9814 * "if (k0 == k)" */
|
|
|
9815 s++;
|
|
|
9816 }
|
|
|
9817 if (*s == '<')
|
|
|
9818 s++;
|
|
|
9819 if (VIM_ISDIGIT(*s))
|
|
|
9820 {
|
|
|
9821 p0 = *s - '0';
|
|
|
9822 s++;
|
|
|
9823 }
|
|
|
9824
|
|
|
9825 if (*s == NUL
|
|
|
9826 /* *s == '^' cuts */
|
|
|
9827 || (*s == '$'
|
|
376
|
9828 && !spell_iswordp_w(word + i + k0,
|
|
|
9829 curbuf)))
|
|
372
|
9830 {
|
|
|
9831 if (k0 == k)
|
|
|
9832 /* this is just a piece of the string */
|
|
|
9833 continue;
|
|
|
9834
|
|
|
9835 if (p0 < pri)
|
|
|
9836 /* priority too low */
|
|
|
9837 continue;
|
|
|
9838 /* rule fits; stop search */
|
|
|
9839 break;
|
|
|
9840 }
|
|
|
9841 }
|
|
|
9842
|
|
|
9843 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
9844 == (c0 & 0xff))
|
|
|
9845 continue;
|
|
|
9846 }
|
|
|
9847
|
|
|
9848 /* replace string */
|
|
|
9849 ws = smp[n].sm_to_w;
|
|
|
9850 s = smp[n].sm_rules;
|
|
|
9851 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
9852 if (p0 == 1 && z == 0)
|
|
|
9853 {
|
|
|
9854 /* rule with '<' is used */
|
|
389
|
9855 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
9856 && (wres[reslen - 1] == c
|
|
372
|
9857 || wres[reslen - 1] == *ws))
|
|
|
9858 reslen--;
|
|
|
9859 z0 = 1;
|
|
|
9860 z = 1;
|
|
|
9861 k0 = 0;
|
|
389
|
9862 if (ws != NULL)
|
|
|
9863 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
9864 {
|
|
|
9865 word[i + k0] = *ws;
|
|
|
9866 k0++;
|
|
|
9867 ws++;
|
|
|
9868 }
|
|
372
|
9869 if (k > k0)
|
|
|
9870 mch_memmove(word + i + k0, word + i + k,
|
|
|
9871 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
9872
|
|
|
9873 /* new "actual letter" */
|
|
|
9874 c = word[i];
|
|
|
9875 }
|
|
|
9876 else
|
|
|
9877 {
|
|
|
9878 /* no '<' rule used */
|
|
|
9879 i += k - 1;
|
|
|
9880 z = 0;
|
|
389
|
9881 if (ws != NULL)
|
|
|
9882 while (*ws != NUL && ws[1] != NUL
|
|
|
9883 && reslen < MAXWLEN)
|
|
|
9884 {
|
|
|
9885 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
9886 wres[reslen++] = *ws;
|
|
|
9887 ws++;
|
|
|
9888 }
|
|
372
|
9889 /* new "actual letter" */
|
|
389
|
9890 if (ws == NULL)
|
|
|
9891 c = NUL;
|
|
|
9892 else
|
|
|
9893 c = *ws;
|
|
372
|
9894 if (strstr((char *)s, "^^") != NULL)
|
|
|
9895 {
|
|
|
9896 if (c != NUL)
|
|
|
9897 wres[reslen++] = c;
|
|
|
9898 mch_memmove(word, word + i + 1,
|
|
|
9899 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
9900 i = 0;
|
|
|
9901 z0 = 1;
|
|
|
9902 }
|
|
|
9903 }
|
|
|
9904 break;
|
|
|
9905 }
|
|
|
9906 }
|
|
|
9907 }
|
|
|
9908 else if (vim_iswhite(c))
|
|
|
9909 {
|
|
|
9910 c = ' ';
|
|
|
9911 k = 1;
|
|
|
9912 }
|
|
|
9913
|
|
|
9914 if (z0 == 0)
|
|
|
9915 {
|
|
|
9916 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9917 && (!slang->sl_collapse || reslen == 0
|
|
|
9918 || wres[reslen - 1] != c))
|
|
|
9919 /* condense only double letters */
|
|
|
9920 wres[reslen++] = c;
|
|
|
9921
|
|
|
9922 i++;
|
|
|
9923 z = 0;
|
|
|
9924 k = 0;
|
|
|
9925 }
|
|
|
9926 }
|
|
|
9927
|
|
|
9928 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
9929 l = 0;
|
|
|
9930 for (n = 0; n < reslen; ++n)
|
|
|
9931 {
|
|
|
9932 l += mb_char2bytes(wres[n], res + l);
|
|
|
9933 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
9934 break;
|
|
|
9935 }
|
|
|
9936 res[l] = NUL;
|
|
|
9937 }
|
|
|
9938 #endif
|
|
|
9939
|
|
324
|
9940 /*
|
|
344
|
9941 * Compute a score for two sound-a-like words.
|
|
|
9942 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
9943 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
9944 * avoiding checks that will not be possible.
|
|
|
9945 */
|
|
|
9946 static int
|
|
351
|
9947 soundalike_score(goodstart, badstart)
|
|
|
9948 char_u *goodstart; /* sound-folded good word */
|
|
|
9949 char_u *badstart; /* sound-folded bad word */
|
|
344
|
9950 {
|
|
351
|
9951 char_u *goodsound = goodstart;
|
|
|
9952 char_u *badsound = badstart;
|
|
|
9953 int goodlen;
|
|
|
9954 int badlen;
|
|
344
|
9955 int n;
|
|
|
9956 char_u *pl, *ps;
|
|
|
9957 char_u *pl2, *ps2;
|
|
351
|
9958 int score = 0;
|
|
|
9959
|
|
|
9960 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
9961 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
9962 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
9963 {
|
|
|
9964 score = SCORE_DEL / 2;
|
|
|
9965 if (*badsound == '*')
|
|
|
9966 ++badsound;
|
|
|
9967 else
|
|
|
9968 ++goodsound;
|
|
|
9969 }
|
|
|
9970
|
|
|
9971 goodlen = STRLEN(goodsound);
|
|
|
9972 badlen = STRLEN(badsound);
|
|
344
|
9973
|
|
|
9974 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
9975 * changes. */
|
|
|
9976 n = goodlen - badlen;
|
|
|
9977 if (n < -2 || n > 2)
|
|
|
9978 return SCORE_MAXMAX;
|
|
|
9979
|
|
|
9980 if (n > 0)
|
|
|
9981 {
|
|
351
|
9982 pl = goodsound; /* goodsound is longest */
|
|
344
|
9983 ps = badsound;
|
|
|
9984 }
|
|
|
9985 else
|
|
|
9986 {
|
|
351
|
9987 pl = badsound; /* badsound is longest */
|
|
344
|
9988 ps = goodsound;
|
|
|
9989 }
|
|
|
9990
|
|
|
9991 /* Skip over the identical part. */
|
|
|
9992 while (*pl == *ps && *pl != NUL)
|
|
|
9993 {
|
|
|
9994 ++pl;
|
|
|
9995 ++ps;
|
|
|
9996 }
|
|
|
9997
|
|
|
9998 switch (n)
|
|
|
9999 {
|
|
|
10000 case -2:
|
|
|
10001 case 2:
|
|
|
10002 /*
|
|
|
10003 * Must delete two characters from "pl".
|
|
|
10004 */
|
|
|
10005 ++pl; /* first delete */
|
|
|
10006 while (*pl == *ps)
|
|
|
10007 {
|
|
|
10008 ++pl;
|
|
|
10009 ++ps;
|
|
|
10010 }
|
|
|
10011 /* strings must be equal after second delete */
|
|
|
10012 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
10013 return score + SCORE_DEL * 2;
|
|
344
|
10014
|
|
|
10015 /* Failed to compare. */
|
|
|
10016 break;
|
|
|
10017
|
|
|
10018 case -1:
|
|
|
10019 case 1:
|
|
|
10020 /*
|
|
|
10021 * Minimal one delete from "pl" required.
|
|
|
10022 */
|
|
|
10023
|
|
|
10024 /* 1: delete */
|
|
|
10025 pl2 = pl + 1;
|
|
|
10026 ps2 = ps;
|
|
|
10027 while (*pl2 == *ps2)
|
|
|
10028 {
|
|
|
10029 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10030 return score + SCORE_DEL;
|
|
344
|
10031 ++pl2;
|
|
|
10032 ++ps2;
|
|
|
10033 }
|
|
|
10034
|
|
|
10035 /* 2: delete then swap, then rest must be equal */
|
|
|
10036 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10037 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10038 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
10039
|
|
|
10040 /* 3: delete then substitute, then the rest must be equal */
|
|
|
10041 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10042 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
10043
|
|
|
10044 /* 4: first swap then delete */
|
|
|
10045 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
10046 {
|
|
|
10047 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
10048 ps2 = ps + 2;
|
|
|
10049 while (*pl2 == *ps2)
|
|
|
10050 {
|
|
|
10051 ++pl2;
|
|
|
10052 ++ps2;
|
|
|
10053 }
|
|
|
10054 /* delete a char and then strings must be equal */
|
|
|
10055 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10056 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
10057 }
|
|
|
10058
|
|
|
10059 /* 5: first substitute then delete */
|
|
|
10060 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
10061 ps2 = ps + 1;
|
|
|
10062 while (*pl2 == *ps2)
|
|
|
10063 {
|
|
|
10064 ++pl2;
|
|
|
10065 ++ps2;
|
|
|
10066 }
|
|
|
10067 /* delete a char and then strings must be equal */
|
|
|
10068 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10069 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
10070
|
|
|
10071 /* Failed to compare. */
|
|
|
10072 break;
|
|
|
10073
|
|
|
10074 case 0:
|
|
|
10075 /*
|
|
|
10076 * Lenghts are equal, thus changes must result in same length: An
|
|
|
10077 * insert is only possible in combination with a delete.
|
|
|
10078 * 1: check if for identical strings
|
|
|
10079 */
|
|
|
10080 if (*pl == NUL)
|
|
351
|
10081 return score;
|
|
344
|
10082
|
|
|
10083 /* 2: swap */
|
|
|
10084 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
10085 {
|
|
|
10086 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
10087 ps2 = ps + 2;
|
|
|
10088 while (*pl2 == *ps2)
|
|
|
10089 {
|
|
|
10090 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10091 return score + SCORE_SWAP;
|
|
344
|
10092 ++pl2;
|
|
|
10093 ++ps2;
|
|
|
10094 }
|
|
|
10095 /* 3: swap and swap again */
|
|
|
10096 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10097 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10098 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
10099
|
|
|
10100 /* 4: swap and substitute */
|
|
|
10101 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10102 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
10103 }
|
|
|
10104
|
|
|
10105 /* 5: substitute */
|
|
|
10106 pl2 = pl + 1;
|
|
|
10107 ps2 = ps + 1;
|
|
|
10108 while (*pl2 == *ps2)
|
|
|
10109 {
|
|
|
10110 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10111 return score + SCORE_SUBST;
|
|
344
|
10112 ++pl2;
|
|
|
10113 ++ps2;
|
|
|
10114 }
|
|
|
10115
|
|
|
10116 /* 6: substitute and swap */
|
|
|
10117 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10118 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10119 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
10120
|
|
|
10121 /* 7: substitute and substitute */
|
|
|
10122 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10123 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
10124
|
|
|
10125 /* 8: insert then delete */
|
|
|
10126 pl2 = pl;
|
|
|
10127 ps2 = ps + 1;
|
|
|
10128 while (*pl2 == *ps2)
|
|
|
10129 {
|
|
|
10130 ++pl2;
|
|
|
10131 ++ps2;
|
|
|
10132 }
|
|
|
10133 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10134 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
10135
|
|
|
10136 /* 9: delete then insert */
|
|
|
10137 pl2 = pl + 1;
|
|
|
10138 ps2 = ps;
|
|
|
10139 while (*pl2 == *ps2)
|
|
|
10140 {
|
|
|
10141 ++pl2;
|
|
|
10142 ++ps2;
|
|
|
10143 }
|
|
|
10144 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
10145 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
10146
|
|
|
10147 /* Failed to compare. */
|
|
|
10148 break;
|
|
|
10149 }
|
|
|
10150
|
|
|
10151 return SCORE_MAXMAX;
|
|
|
10152 }
|
|
324
|
10153
|
|
323
|
10154 /*
|
|
|
10155 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
10156 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
10157 *
|
|
323
|
10158 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
10159 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
10160 * characters.
|
|
323
|
10161 */
|
|
|
10162 static int
|
|
|
10163 spell_edit_score(badword, goodword)
|
|
|
10164 char_u *badword;
|
|
|
10165 char_u *goodword;
|
|
|
10166 {
|
|
|
10167 int *cnt;
|
|
372
|
10168 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
10169 int j, i;
|
|
|
10170 int t;
|
|
|
10171 int bc, gc;
|
|
324
|
10172 int pbc, pgc;
|
|
|
10173 #ifdef FEAT_MBYTE
|
|
|
10174 char_u *p;
|
|
|
10175 int wbadword[MAXWLEN];
|
|
|
10176 int wgoodword[MAXWLEN];
|
|
|
10177
|
|
|
10178 if (has_mbyte)
|
|
|
10179 {
|
|
|
10180 /* Get the characters from the multi-byte strings and put them in an
|
|
|
10181 * int array for easy access. */
|
|
|
10182 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
10183 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
10184 wbadword[badlen++] = 0;
|
|
324
|
10185 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
10186 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
10187 wgoodword[goodlen++] = 0;
|
|
324
|
10188 }
|
|
|
10189 else
|
|
|
10190 #endif
|
|
|
10191 {
|
|
|
10192 badlen = STRLEN(badword) + 1;
|
|
|
10193 goodlen = STRLEN(goodword) + 1;
|
|
|
10194 }
|
|
323
|
10195
|
|
|
10196 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
10197 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
10198 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
10199 TRUE);
|
|
324
|
10200 if (cnt == NULL)
|
|
|
10201 return 0; /* out of memory */
|
|
323
|
10202
|
|
|
10203 CNT(0, 0) = 0;
|
|
|
10204 for (j = 1; j <= goodlen; ++j)
|
|
|
10205 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
10206
|
|
|
10207 for (i = 1; i <= badlen; ++i)
|
|
|
10208 {
|
|
|
10209 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
10210 for (j = 1; j <= goodlen; ++j)
|
|
|
10211 {
|
|
324
|
10212 #ifdef FEAT_MBYTE
|
|
|
10213 if (has_mbyte)
|
|
|
10214 {
|
|
|
10215 bc = wbadword[i - 1];
|
|
|
10216 gc = wgoodword[j - 1];
|
|
|
10217 }
|
|
|
10218 else
|
|
|
10219 #endif
|
|
|
10220 {
|
|
|
10221 bc = badword[i - 1];
|
|
|
10222 gc = goodword[j - 1];
|
|
|
10223 }
|
|
323
|
10224 if (bc == gc)
|
|
|
10225 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
10226 else
|
|
|
10227 {
|
|
|
10228 /* Use a better score when there is only a case difference. */
|
|
324
|
10229 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
10230 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
10231 else
|
|
|
10232 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
10233
|
|
324
|
10234 if (i > 1 && j > 1)
|
|
323
|
10235 {
|
|
324
|
10236 #ifdef FEAT_MBYTE
|
|
|
10237 if (has_mbyte)
|
|
|
10238 {
|
|
|
10239 pbc = wbadword[i - 2];
|
|
|
10240 pgc = wgoodword[j - 2];
|
|
|
10241 }
|
|
|
10242 else
|
|
|
10243 #endif
|
|
|
10244 {
|
|
|
10245 pbc = badword[i - 2];
|
|
|
10246 pgc = goodword[j - 2];
|
|
|
10247 }
|
|
|
10248 if (bc == pgc && pbc == gc)
|
|
|
10249 {
|
|
|
10250 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
10251 if (t < CNT(i, j))
|
|
|
10252 CNT(i, j) = t;
|
|
|
10253 }
|
|
323
|
10254 }
|
|
|
10255 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
10256 if (t < CNT(i, j))
|
|
|
10257 CNT(i, j) = t;
|
|
|
10258 t = SCORE_INS + CNT(i, j - 1);
|
|
|
10259 if (t < CNT(i, j))
|
|
|
10260 CNT(i, j) = t;
|
|
|
10261 }
|
|
|
10262 }
|
|
|
10263 }
|
|
344
|
10264
|
|
|
10265 i = CNT(badlen - 1, goodlen - 1);
|
|
|
10266 vim_free(cnt);
|
|
|
10267 return i;
|
|
323
|
10268 }
|
|
307
|
10269
|
|
351
|
10270 /*
|
|
|
10271 * ":spelldump"
|
|
|
10272 */
|
|
|
10273 /*ARGSUSED*/
|
|
|
10274 void
|
|
|
10275 ex_spelldump(eap)
|
|
|
10276 exarg_T *eap;
|
|
|
10277 {
|
|
|
10278 buf_T *buf = curbuf;
|
|
|
10279 langp_T *lp;
|
|
|
10280 slang_T *slang;
|
|
|
10281 idx_T arridx[MAXWLEN];
|
|
|
10282 int curi[MAXWLEN];
|
|
|
10283 char_u word[MAXWLEN];
|
|
|
10284 int c;
|
|
|
10285 char_u *byts;
|
|
|
10286 idx_T *idxs;
|
|
|
10287 linenr_T lnum = 0;
|
|
|
10288 int round;
|
|
|
10289 int depth;
|
|
|
10290 int n;
|
|
|
10291 int flags;
|
|
381
|
10292 char_u *region_names = NULL; /* region names being used */
|
|
|
10293 int do_region = TRUE; /* dump region names and numbers */
|
|
|
10294 char_u *p;
|
|
351
|
10295
|
|
|
10296 if (no_spell_checking())
|
|
|
10297 return;
|
|
|
10298
|
|
|
10299 /* Create a new empty buffer by splitting the window. */
|
|
|
10300 do_cmdline_cmd((char_u *)"new");
|
|
|
10301 if (!bufempty() || !buf_valid(buf))
|
|
|
10302 return;
|
|
|
10303
|
|
381
|
10304 /* Find out if we can support regions: All languages must support the same
|
|
|
10305 * regions or none at all. */
|
|
|
10306 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
10307 {
|
|
|
10308 p = lp->lp_slang->sl_regions;
|
|
|
10309 if (p[0] != 0)
|
|
|
10310 {
|
|
|
10311 if (region_names == NULL) /* first language with regions */
|
|
|
10312 region_names = p;
|
|
|
10313 else if (STRCMP(region_names, p) != 0)
|
|
|
10314 {
|
|
|
10315 do_region = FALSE; /* region names are different */
|
|
|
10316 break;
|
|
|
10317 }
|
|
|
10318 }
|
|
|
10319 }
|
|
|
10320
|
|
|
10321 if (do_region && region_names != NULL)
|
|
|
10322 {
|
|
|
10323 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
10324 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
10325 }
|
|
|
10326 else
|
|
|
10327 do_region = FALSE;
|
|
|
10328
|
|
|
10329 /*
|
|
|
10330 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
10331 */
|
|
351
|
10332 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
10333 {
|
|
|
10334 slang = lp->lp_slang;
|
|
|
10335
|
|
|
10336 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
10337 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
10338
|
|
|
10339 /* round 1: case-folded tree
|
|
|
10340 * round 2: keep-case tree */
|
|
|
10341 for (round = 1; round <= 2; ++round)
|
|
|
10342 {
|
|
|
10343 if (round == 1)
|
|
|
10344 {
|
|
|
10345 byts = slang->sl_fbyts;
|
|
|
10346 idxs = slang->sl_fidxs;
|
|
|
10347 }
|
|
|
10348 else
|
|
|
10349 {
|
|
|
10350 byts = slang->sl_kbyts;
|
|
|
10351 idxs = slang->sl_kidxs;
|
|
|
10352 }
|
|
|
10353 if (byts == NULL)
|
|
|
10354 continue; /* array is empty */
|
|
|
10355
|
|
|
10356 depth = 0;
|
|
|
10357 arridx[0] = 0;
|
|
|
10358 curi[0] = 1;
|
|
|
10359 while (depth >= 0 && !got_int)
|
|
|
10360 {
|
|
|
10361 if (curi[depth] > byts[arridx[depth]])
|
|
|
10362 {
|
|
|
10363 /* Done all bytes at this node, go up one level. */
|
|
|
10364 --depth;
|
|
|
10365 line_breakcheck();
|
|
|
10366 }
|
|
|
10367 else
|
|
|
10368 {
|
|
|
10369 /* Do one more byte at this node. */
|
|
|
10370 n = arridx[depth] + curi[depth];
|
|
|
10371 ++curi[depth];
|
|
|
10372 c = byts[n];
|
|
|
10373 if (c == 0)
|
|
|
10374 {
|
|
|
10375 /* End of word, deal with the word.
|
|
|
10376 * Don't use keep-case words in the fold-case tree,
|
|
|
10377 * they will appear in the keep-case tree.
|
|
|
10378 * Only use the word when the region matches. */
|
|
|
10379 flags = (int)idxs[n];
|
|
|
10380 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
10381 && (do_region
|
|
|
10382 || (flags & WF_REGION) == 0
|
|
390
|
10383 || (((unsigned)flags >> 16)
|
|
351
|
10384 & lp->lp_region) != 0))
|
|
|
10385 {
|
|
|
10386 word[depth] = NUL;
|
|
381
|
10387 if (!do_region)
|
|
|
10388 flags &= ~WF_REGION;
|
|
355
|
10389
|
|
|
10390 /* Dump the basic word if there is no prefix or
|
|
|
10391 * when it's the first one. */
|
|
390
|
10392 c = (unsigned)flags >> 24;
|
|
355
|
10393 if (c == 0 || curi[depth] == 2)
|
|
|
10394 dump_word(word, round, flags, lnum++);
|
|
351
|
10395
|
|
|
10396 /* Apply the prefix, if there is one. */
|
|
355
|
10397 if (c != 0)
|
|
351
|
10398 lnum = apply_prefixes(slang, word, round,
|
|
|
10399 flags, lnum);
|
|
|
10400 }
|
|
|
10401 }
|
|
|
10402 else
|
|
|
10403 {
|
|
|
10404 /* Normal char, go one level deeper. */
|
|
|
10405 word[depth++] = c;
|
|
|
10406 arridx[depth] = idxs[n];
|
|
|
10407 curi[depth] = 1;
|
|
|
10408 }
|
|
|
10409 }
|
|
|
10410 }
|
|
|
10411 }
|
|
|
10412 }
|
|
|
10413
|
|
|
10414 /* Delete the empty line that we started with. */
|
|
|
10415 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
10416 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
10417
|
|
|
10418 redraw_later(NOT_VALID);
|
|
|
10419 }
|
|
|
10420
|
|
|
10421 /*
|
|
|
10422 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
10423 */
|
|
|
10424 static void
|
|
|
10425 dump_word(word, round, flags, lnum)
|
|
|
10426 char_u *word;
|
|
|
10427 int round;
|
|
|
10428 int flags;
|
|
|
10429 linenr_T lnum;
|
|
|
10430 {
|
|
|
10431 int keepcap = FALSE;
|
|
|
10432 char_u *p;
|
|
|
10433 char_u cword[MAXWLEN];
|
|
381
|
10434 char_u badword[MAXWLEN + 10];
|
|
|
10435 int i;
|
|
351
|
10436
|
|
|
10437 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
10438 {
|
|
|
10439 /* Need to fix case according to "flags". */
|
|
|
10440 make_case_word(word, cword, flags);
|
|
|
10441 p = cword;
|
|
|
10442 }
|
|
|
10443 else
|
|
|
10444 {
|
|
|
10445 p = word;
|
|
389
|
10446 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
10447 || (flags & WF_FIXCAP) != 0))
|
|
351
|
10448 keepcap = TRUE;
|
|
|
10449 }
|
|
|
10450
|
|
381
|
10451 /* Add flags and regions after a slash. */
|
|
|
10452 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
10453 {
|
|
|
10454 STRCPY(badword, p);
|
|
|
10455 STRCAT(badword, "/");
|
|
351
|
10456 if (keepcap)
|
|
|
10457 STRCAT(badword, "=");
|
|
|
10458 if (flags & WF_BANNED)
|
|
|
10459 STRCAT(badword, "!");
|
|
|
10460 else if (flags & WF_RARE)
|
|
|
10461 STRCAT(badword, "?");
|
|
381
|
10462 if (flags & WF_REGION)
|
|
|
10463 for (i = 0; i < 7; ++i)
|
|
390
|
10464 if (flags & (0x10000 << i))
|
|
381
|
10465 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
10466 p = badword;
|
|
|
10467 }
|
|
|
10468
|
|
|
10469 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
10470 }
|
|
|
10471
|
|
|
10472 /*
|
|
372
|
10473 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
10474 * "word" and append a line to the buffer.
|
|
351
|
10475 * Return the updated line number.
|
|
|
10476 */
|
|
|
10477 static linenr_T
|
|
|
10478 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
10479 slang_T *slang;
|
|
|
10480 char_u *word; /* case-folded word */
|
|
|
10481 int round;
|
|
|
10482 int flags; /* flags with prefix ID */
|
|
|
10483 linenr_T startlnum;
|
|
|
10484 {
|
|
|
10485 idx_T arridx[MAXWLEN];
|
|
|
10486 int curi[MAXWLEN];
|
|
|
10487 char_u prefix[MAXWLEN];
|
|
455
|
10488 char_u word_up[MAXWLEN];
|
|
|
10489 int has_word_up = FALSE;
|
|
351
|
10490 int c;
|
|
|
10491 char_u *byts;
|
|
|
10492 idx_T *idxs;
|
|
|
10493 linenr_T lnum = startlnum;
|
|
|
10494 int depth;
|
|
|
10495 int n;
|
|
|
10496 int len;
|
|
|
10497 int i;
|
|
|
10498
|
|
455
|
10499 /* if the word starts with a lower-case letter make the word with an
|
|
|
10500 * upper-case letter in word_up[]. */
|
|
|
10501 c = PTR2CHAR(word);
|
|
|
10502 if (SPELL_TOUPPER(c) != c)
|
|
|
10503 {
|
|
|
10504 onecap_copy(word, word_up, TRUE);
|
|
|
10505 has_word_up = TRUE;
|
|
|
10506 }
|
|
|
10507
|
|
351
|
10508 byts = slang->sl_pbyts;
|
|
|
10509 idxs = slang->sl_pidxs;
|
|
|
10510 if (byts != NULL) /* array not is empty */
|
|
|
10511 {
|
|
|
10512 /*
|
|
|
10513 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
10514 * When at the end of a prefix check that it supports "flags".
|
|
351
|
10515 */
|
|
|
10516 depth = 0;
|
|
|
10517 arridx[0] = 0;
|
|
|
10518 curi[0] = 1;
|
|
|
10519 while (depth >= 0 && !got_int)
|
|
|
10520 {
|
|
390
|
10521 n = arridx[depth];
|
|
|
10522 len = byts[n];
|
|
|
10523 if (curi[depth] > len)
|
|
351
|
10524 {
|
|
|
10525 /* Done all bytes at this node, go up one level. */
|
|
|
10526 --depth;
|
|
|
10527 line_breakcheck();
|
|
|
10528 }
|
|
|
10529 else
|
|
|
10530 {
|
|
|
10531 /* Do one more byte at this node. */
|
|
390
|
10532 n += curi[depth];
|
|
351
|
10533 ++curi[depth];
|
|
|
10534 c = byts[n];
|
|
|
10535 if (c == 0)
|
|
|
10536 {
|
|
|
10537 /* End of prefix, find out how many IDs there are. */
|
|
|
10538 for (i = 1; i < len; ++i)
|
|
|
10539 if (byts[n + i] != 0)
|
|
|
10540 break;
|
|
|
10541 curi[depth] += i - 1;
|
|
|
10542
|
|
455
|
10543 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
10544 if (c != 0)
|
|
351
|
10545 {
|
|
376
|
10546 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
10547 dump_word(prefix, round,
|
|
455
|
10548 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
10549 : flags, lnum++);
|
|
351
|
10550 }
|
|
455
|
10551
|
|
|
10552 /* Check for prefix that matches the word when the
|
|
|
10553 * first letter is upper-case, but only if the prefix has
|
|
|
10554 * a condition. */
|
|
|
10555 if (has_word_up)
|
|
|
10556 {
|
|
|
10557 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
10558 TRUE);
|
|
|
10559 if (c != 0)
|
|
|
10560 {
|
|
|
10561 vim_strncpy(prefix + depth, word_up,
|
|
|
10562 MAXWLEN - depth - 1);
|
|
|
10563 dump_word(prefix, round,
|
|
|
10564 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
10565 : flags, lnum++);
|
|
|
10566 }
|
|
|
10567 }
|
|
351
|
10568 }
|
|
|
10569 else
|
|
|
10570 {
|
|
|
10571 /* Normal char, go one level deeper. */
|
|
|
10572 prefix[depth++] = c;
|
|
|
10573 arridx[depth] = idxs[n];
|
|
|
10574 curi[depth] = 1;
|
|
|
10575 }
|
|
|
10576 }
|
|
|
10577 }
|
|
|
10578 }
|
|
|
10579
|
|
|
10580 return lnum;
|
|
|
10581 }
|
|
|
10582
|
|
475
|
10583 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
10584 static int spell_expand_need_cap;
|
|
|
10585
|
|
|
10586 /*
|
|
|
10587 * Find start of the word in front of the cursor. We don't check if it is
|
|
|
10588 * badly spelled, with completion we can only change the word in front of the
|
|
|
10589 * cursor.
|
|
|
10590 * Used for Insert mode completion CTRL-X ?.
|
|
|
10591 * Returns the column number of the word.
|
|
|
10592 */
|
|
|
10593 int
|
|
|
10594 spell_word_start(startcol)
|
|
|
10595 int startcol;
|
|
|
10596 {
|
|
|
10597 char_u *line;
|
|
|
10598 char_u *p;
|
|
|
10599 int col = 0;
|
|
|
10600
|
|
|
10601 if (no_spell_checking())
|
|
|
10602 return startcol;
|
|
|
10603
|
|
|
10604 /* Find a word character before "startcol". */
|
|
|
10605 line = ml_get_curline();
|
|
|
10606 for (p = line + startcol; p > line; )
|
|
|
10607 {
|
|
|
10608 mb_ptr_back(line, p);
|
|
|
10609 if (spell_iswordp_nmw(p))
|
|
|
10610 break;
|
|
|
10611 }
|
|
|
10612
|
|
|
10613 /* Go back to start of the word. */
|
|
|
10614 while (p > line)
|
|
|
10615 {
|
|
|
10616 col = p - line;
|
|
|
10617 mb_ptr_back(line, p);
|
|
|
10618 if (!spell_iswordp(p, curbuf))
|
|
|
10619 break;
|
|
|
10620 col = 0;
|
|
|
10621 }
|
|
|
10622
|
|
|
10623 /* Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
10624 * expand_spelling() is called. Therefore the ugly global variable. */
|
|
|
10625 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
10626
|
|
|
10627 return col;
|
|
|
10628 }
|
|
|
10629
|
|
|
10630 /*
|
|
|
10631 * Get list of spelling suggestions.
|
|
|
10632 * Used for Insert mode completion CTRL-X ?.
|
|
|
10633 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
10634 * allocated strings.
|
|
|
10635 */
|
|
|
10636 /*ARGSUSED*/
|
|
|
10637 int
|
|
|
10638 expand_spelling(lnum, col, pat, matchp)
|
|
|
10639 linenr_T lnum;
|
|
|
10640 int col;
|
|
|
10641 char_u *pat;
|
|
|
10642 char_u ***matchp;
|
|
|
10643 {
|
|
|
10644 garray_T ga;
|
|
|
10645
|
|
|
10646 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap);
|
|
|
10647 *matchp = ga.ga_data;
|
|
|
10648 return ga.ga_len;
|
|
|
10649 }
|
|
|
10650 #endif
|
|
|
10651
|
|
236
|
10652 #endif /* FEAT_SYN_HL */
|
