|
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 */
|
|
366
|
481 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
482
|
|
372
|
483 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
484 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
485 * 350 allows for about three changes. */
|
|
344
|
486
|
|
|
487 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
488 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
489
|
|
|
490 /*
|
|
236
|
491 * Structure to store info for word matching.
|
|
|
492 */
|
|
|
493 typedef struct matchinf_S
|
|
|
494 {
|
|
|
495 langp_T *mi_lp; /* info for language and region */
|
|
243
|
496
|
|
|
497 /* pointers to original text to be checked */
|
|
236
|
498 char_u *mi_word; /* start of word being checked */
|
|
339
|
499 char_u *mi_end; /* end of matching word so far */
|
|
243
|
500 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
501 char_u *mi_cend; /* char after what was used for
|
|
|
502 mi_capflags */
|
|
243
|
503
|
|
|
504 /* case-folded text */
|
|
|
505 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
506 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
507
|
|
339
|
508 /* for when checking word after a prefix */
|
|
|
509 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
|
510 prefixID/condition */
|
|
|
511 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
512 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
513 #ifdef FEAT_MBYTE
|
|
|
514 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
515 case */
|
|
|
516 #else
|
|
|
517 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
518 #endif
|
|
339
|
519
|
|
243
|
520 /* others */
|
|
236
|
521 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
522 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
523 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
524 } matchinf_T;
|
|
|
525
|
|
307
|
526 /*
|
|
|
527 * The tables used for recognizing word characters according to spelling.
|
|
|
528 * These are only used for the first 256 characters of 'encoding'.
|
|
|
529 */
|
|
|
530 typedef struct spelltab_S
|
|
|
531 {
|
|
|
532 char_u st_isw[256]; /* flags: is word char */
|
|
|
533 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
534 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
535 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
536 } spelltab_T;
|
|
|
537
|
|
|
538 static spelltab_T spelltab;
|
|
|
539 static int did_set_spelltab;
|
|
|
540
|
|
324
|
541 #define CF_WORD 0x01
|
|
|
542 #define CF_UPPER 0x02
|
|
307
|
543
|
|
|
544 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
545 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
546 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
547 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
548 #ifdef FEAT_MBYTE
|
|
|
549 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
550 #endif
|
|
339
|
551 static void write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
552
|
|
|
553 /*
|
|
344
|
554 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
555 */
|
|
|
556 typedef enum
|
|
|
557 {
|
|
344
|
558 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
559 * ends); if badword ends there is a match, otherwise
|
|
|
560 * try splitting word. */
|
|
375
|
561 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
562 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
563 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
564 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
565 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
566 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
567 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
568 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
569 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
570 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
571 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
572 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
573 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
574 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
575 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
576 STATE_FINAL /* End of this node. */
|
|
|
577 } state_T;
|
|
|
578
|
|
|
579 /*
|
|
346
|
580 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
581 */
|
|
|
582 typedef struct trystate_S
|
|
|
583 {
|
|
330
|
584 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
585 int ts_score; /* score */
|
|
344
|
586 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
587 short ts_curi; /* index in list of child nodes */
|
|
|
588 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
589 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
590 char_u ts_twordlen; /* valid length of tword[] */
|
|
375
|
591 char_u ts_prefixdepth; /* stack depth for end of prefix or PREFIXTREE
|
|
|
592 * or NOPREFIX */
|
|
330
|
593 #ifdef FEAT_MBYTE
|
|
|
594 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
595 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
596 char_u ts_isdiff; /* DIFF_ values */
|
|
|
597 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
598 #endif
|
|
323
|
599 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
330
|
600 char_u ts_save_splitoff; /* su_splitoff saved here */
|
|
346
|
601 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
602 } trystate_T;
|
|
|
603
|
|
330
|
604 /* values for ts_isdiff */
|
|
|
605 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
606 #define DIFF_YES 1 /* different byte found */
|
|
|
607 #define DIFF_INSERT 2 /* inserting character */
|
|
|
608
|
|
375
|
609 /* special values ts_prefixdepth */
|
|
|
610 #define PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
611 #define NOPREFIX 0xff /* not using prefixes */
|
|
|
612
|
|
339
|
613 /* mode values for find_word */
|
|
|
614 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
615 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
616 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
617
|
|
236
|
618 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
619 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
620 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
621 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
455
|
622 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
339
|
623 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
624 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
625 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
626 static int no_spell_checking __ARGS((void));
|
|
307
|
627 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
628 static char_u *spell_enc __ARGS((void));
|
|
385
|
629 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
630 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
631 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
632 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
381
|
633 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
634 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
635 #ifdef FEAT_MBYTE
|
|
|
636 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
637 #endif
|
|
339
|
638 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
639 static void clear_midword __ARGS((buf_T *buf));
|
|
|
640 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
641 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
642 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
643 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
644 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
389
|
645 static int set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
646 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
647 static void write_spell_chartab __ARGS((FILE *fd));
|
|
|
648 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
388
|
649 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
650 #ifdef FEAT_EVAL
|
|
|
651 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
652 #endif
|
|
|
653 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
654 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
655 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
656 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
657 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
658 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
659 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
660 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
661 #ifdef FEAT_MBYTE
|
|
|
662 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
663 #endif
|
|
323
|
664 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
665 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
666 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
667 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
668 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
669 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
670 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
671 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
672 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
673 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
674 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
675 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
676 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
677 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
678 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
679 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
680 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
681 #ifdef FEAT_MBYTE
|
|
375
|
682 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
683 #endif
|
|
344
|
684 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
685 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
686 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
687 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
688
|
|
324
|
689 /*
|
|
|
690 * Use our own character-case definitions, because the current locale may
|
|
|
691 * differ from what the .spl file uses.
|
|
|
692 * These must not be called with negative number!
|
|
|
693 */
|
|
|
694 #ifndef FEAT_MBYTE
|
|
|
695 /* Non-multi-byte implementation. */
|
|
|
696 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
697 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
698 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
699 #else
|
|
447
|
700 # if defined(HAVE_WCHAR_H)
|
|
|
701 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
702 # endif
|
|
324
|
703 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
704 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
705 * the "w" library function for characters above 255 if available. */
|
|
|
706 # ifdef HAVE_TOWLOWER
|
|
|
707 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
708 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
709 # else
|
|
|
710 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
711 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
712 # endif
|
|
|
713
|
|
|
714 # ifdef HAVE_TOWUPPER
|
|
|
715 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
716 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
717 # else
|
|
|
718 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
719 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
720 # endif
|
|
|
721
|
|
|
722 # ifdef HAVE_ISWUPPER
|
|
|
723 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
724 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
725 # else
|
|
|
726 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
727 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
728 # endif
|
|
|
729 #endif
|
|
|
730
|
|
307
|
731
|
|
|
732 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
733 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
469
|
734 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
735
|
|
|
736 /*
|
|
|
737 * Main spell-checking function.
|
|
300
|
738 * "ptr" points to a character that could be the start of a word.
|
|
236
|
739 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
740 * or when it's OK it remains unchanged.
|
|
|
741 * This must only be called when 'spelllang' is not empty.
|
|
323
|
742 *
|
|
385
|
743 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
744 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
745 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
746 * worry.
|
|
323
|
747 *
|
|
236
|
748 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
749 * caller can skip over the word.
|
|
|
750 */
|
|
|
751 int
|
|
385
|
752 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
753 win_T *wp; /* current window */
|
|
|
754 char_u *ptr;
|
|
|
755 int *attrp;
|
|
385
|
756 int *capcol; /* column to check for Capital */
|
|
236
|
757 {
|
|
|
758 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
759 be passed to functions quickly. */
|
|
344
|
760 int nrlen = 0; /* found a number first */
|
|
385
|
761 int c;
|
|
236
|
762
|
|
307
|
763 /* A word never starts at a space or a control character. Return quickly
|
|
|
764 * then, skipping over the character. */
|
|
|
765 if (*ptr <= ' ')
|
|
|
766 return 1;
|
|
236
|
767
|
|
344
|
768 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
769 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
770 * "3GPP". */
|
|
300
|
771 if (*ptr >= '0' && *ptr <= '9')
|
|
|
772 {
|
|
316
|
773 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
774 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
775 else
|
|
|
776 {
|
|
344
|
777 mi.mi_end = skipdigits(ptr);
|
|
|
778 nrlen = mi.mi_end - ptr;
|
|
300
|
779 }
|
|
376
|
780 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
781 return (int)(mi.mi_end - ptr);
|
|
346
|
782
|
|
|
783 /* Try including the digits in the word. */
|
|
|
784 mi.mi_fend = ptr + nrlen;
|
|
|
785 }
|
|
|
786 else
|
|
|
787 mi.mi_fend = ptr;
|
|
|
788
|
|
|
789 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
790 mi.mi_word = ptr;
|
|
376
|
791 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
792 {
|
|
|
793 do
|
|
|
794 {
|
|
307
|
795 mb_ptr_adv(mi.mi_fend);
|
|
376
|
796 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
797
|
|
|
798 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
799 {
|
|
|
800 /* Check word starting with capital letter. */
|
|
455
|
801 c = PTR2CHAR(ptr);
|
|
385
|
802 if (!SPELL_ISUPPER(c))
|
|
|
803 {
|
|
|
804 *attrp = highlight_attr[HLF_SPC];
|
|
|
805 return (int)(mi.mi_fend - ptr);
|
|
|
806 }
|
|
|
807 }
|
|
|
808 }
|
|
|
809 if (capcol != NULL)
|
|
|
810 *capcol = -1;
|
|
344
|
811
|
|
|
812 /* We always use the characters up to the next non-word character,
|
|
|
813 * also for bad words. */
|
|
|
814 mi.mi_end = mi.mi_fend;
|
|
|
815
|
|
|
816 /* Check caps type later. */
|
|
|
817 mi.mi_capflags = 0;
|
|
|
818 mi.mi_cend = NULL;
|
|
376
|
819 mi.mi_buf = wp->w_buffer;
|
|
344
|
820
|
|
|
821 /* Include one non-word character so that we can check for the
|
|
|
822 * word end. */
|
|
|
823 if (*mi.mi_fend != NUL)
|
|
|
824 mb_ptr_adv(mi.mi_fend);
|
|
|
825
|
|
|
826 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
827 MAXWLEN + 1);
|
|
|
828 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
829
|
|
|
830 /* The word is bad unless we recognize it. */
|
|
|
831 mi.mi_result = SP_BAD;
|
|
|
832
|
|
|
833 /*
|
|
|
834 * Loop over the languages specified in 'spelllang'.
|
|
|
835 * We check them all, because a matching word may be longer than an
|
|
|
836 * already found matching word.
|
|
|
837 */
|
|
|
838 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
839 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
840 {
|
|
|
841 /* Check for a matching word in case-folded words. */
|
|
|
842 find_word(&mi, FIND_FOLDWORD);
|
|
|
843
|
|
|
844 /* Check for a matching word in keep-case words. */
|
|
|
845 find_word(&mi, FIND_KEEPWORD);
|
|
|
846
|
|
|
847 /* Check for matching prefixes. */
|
|
|
848 find_prefix(&mi);
|
|
|
849 }
|
|
|
850
|
|
|
851 if (mi.mi_result != SP_OK)
|
|
|
852 {
|
|
346
|
853 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
854 * rare and local words, e.g., "3GPP". */
|
|
344
|
855 if (nrlen > 0)
|
|
346
|
856 {
|
|
|
857 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
858 return nrlen;
|
|
|
859 }
|
|
344
|
860
|
|
|
861 /* When we are at a non-word character there is no error, just
|
|
|
862 * skip over the character (try looking for a word after it). */
|
|
390
|
863 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
864 {
|
|
385
|
865 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
866 {
|
|
|
867 regmatch_T regmatch;
|
|
|
868
|
|
|
869 /* Check for end of sentence. */
|
|
|
870 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
871 regmatch.rm_ic = FALSE;
|
|
|
872 if (vim_regexec(®match, ptr, 0))
|
|
|
873 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
874 }
|
|
|
875
|
|
344
|
876 #ifdef FEAT_MBYTE
|
|
|
877 if (has_mbyte)
|
|
474
|
878 return (*mb_ptr2len)(ptr);
|
|
344
|
879 #endif
|
|
|
880 return 1;
|
|
300
|
881 }
|
|
243
|
882
|
|
344
|
883 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
884 *attrp = highlight_attr[HLF_SPB];
|
|
|
885 else if (mi.mi_result == SP_RARE)
|
|
|
886 *attrp = highlight_attr[HLF_SPR];
|
|
|
887 else
|
|
|
888 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
889 }
|
|
|
890
|
|
300
|
891 return (int)(mi.mi_end - ptr);
|
|
236
|
892 }
|
|
|
893
|
|
|
894 /*
|
|
300
|
895 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
896 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
897 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
898 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
899 * tree.
|
|
300
|
900 *
|
|
|
901 * For a match mip->mi_result is updated.
|
|
243
|
902 */
|
|
|
903 static void
|
|
339
|
904 find_word(mip, mode)
|
|
243
|
905 matchinf_T *mip;
|
|
339
|
906 int mode;
|
|
243
|
907 {
|
|
324
|
908 idx_T arridx = 0;
|
|
300
|
909 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
910 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
911 int endidxcnt = 0;
|
|
|
912 int len;
|
|
|
913 int wlen = 0;
|
|
|
914 int flen;
|
|
|
915 int c;
|
|
|
916 char_u *ptr;
|
|
324
|
917 idx_T lo, hi, m;
|
|
243
|
918 #ifdef FEAT_MBYTE
|
|
300
|
919 char_u *s;
|
|
339
|
920 char_u *p;
|
|
307
|
921 #endif
|
|
|
922 int res = SP_BAD;
|
|
300
|
923 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
924 unsigned flags;
|
|
|
925 char_u *byts;
|
|
324
|
926 idx_T *idxs;
|
|
339
|
927
|
|
|
928 if (mode == FIND_KEEPWORD)
|
|
236
|
929 {
|
|
300
|
930 /* Check for word with matching case in keep-case tree. */
|
|
|
931 ptr = mip->mi_word;
|
|
|
932 flen = 9999; /* no case folding, always enough bytes */
|
|
|
933 byts = slang->sl_kbyts;
|
|
|
934 idxs = slang->sl_kidxs;
|
|
236
|
935 }
|
|
|
936 else
|
|
|
937 {
|
|
300
|
938 /* Check for case-folded in case-folded tree. */
|
|
|
939 ptr = mip->mi_fword;
|
|
|
940 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
941 byts = slang->sl_fbyts;
|
|
|
942 idxs = slang->sl_fidxs;
|
|
339
|
943
|
|
|
944 if (mode == FIND_PREFIX)
|
|
|
945 {
|
|
|
946 /* Skip over the prefix. */
|
|
|
947 wlen = mip->mi_prefixlen;
|
|
|
948 flen -= mip->mi_prefixlen;
|
|
|
949 }
|
|
243
|
950 }
|
|
|
951
|
|
300
|
952 if (byts == NULL)
|
|
|
953 return; /* array is empty */
|
|
236
|
954
|
|
|
955 /*
|
|
307
|
956 * Repeat advancing in the tree until:
|
|
|
957 * - there is a byte that doesn't match,
|
|
|
958 * - we reach the end of the tree,
|
|
|
959 * - or we reach the end of the line.
|
|
236
|
960 */
|
|
300
|
961 for (;;)
|
|
236
|
962 {
|
|
346
|
963 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
964 flen = fold_more(mip);
|
|
300
|
965
|
|
|
966 len = byts[arridx++];
|
|
|
967
|
|
|
968 /* If the first possible byte is a zero the word could end here.
|
|
|
969 * Remember this index, we first check for the longest word. */
|
|
|
970 if (byts[arridx] == 0)
|
|
|
971 {
|
|
307
|
972 if (endidxcnt == MAXWLEN)
|
|
|
973 {
|
|
|
974 /* Must be a corrupted spell file. */
|
|
|
975 EMSG(_(e_format));
|
|
|
976 return;
|
|
|
977 }
|
|
300
|
978 endlen[endidxcnt] = wlen;
|
|
|
979 endidx[endidxcnt++] = arridx++;
|
|
|
980 --len;
|
|
|
981
|
|
|
982 /* Skip over the zeros, there can be several flag/region
|
|
|
983 * combinations. */
|
|
|
984 while (len > 0 && byts[arridx] == 0)
|
|
|
985 {
|
|
|
986 ++arridx;
|
|
|
987 --len;
|
|
|
988 }
|
|
|
989 if (len == 0)
|
|
|
990 break; /* no children, word must end here */
|
|
|
991 }
|
|
|
992
|
|
|
993 /* Stop looking at end of the line. */
|
|
|
994 if (ptr[wlen] == NUL)
|
|
|
995 break;
|
|
|
996
|
|
|
997 /* Perform a binary search in the list of accepted bytes. */
|
|
|
998 c = ptr[wlen];
|
|
346
|
999 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1000 c = ' ';
|
|
300
|
1001 lo = arridx;
|
|
|
1002 hi = arridx + len - 1;
|
|
|
1003 while (lo < hi)
|
|
|
1004 {
|
|
|
1005 m = (lo + hi) / 2;
|
|
|
1006 if (byts[m] > c)
|
|
|
1007 hi = m - 1;
|
|
|
1008 else if (byts[m] < c)
|
|
|
1009 lo = m + 1;
|
|
|
1010 else
|
|
|
1011 {
|
|
|
1012 lo = hi = m;
|
|
|
1013 break;
|
|
236
|
1014 }
|
|
|
1015 }
|
|
300
|
1016
|
|
|
1017 /* Stop if there is no matching byte. */
|
|
|
1018 if (hi < lo || byts[lo] != c)
|
|
|
1019 break;
|
|
|
1020
|
|
|
1021 /* Continue at the child (if there is one). */
|
|
|
1022 arridx = idxs[lo];
|
|
|
1023 ++wlen;
|
|
|
1024 --flen;
|
|
346
|
1025
|
|
|
1026 /* One space in the good word may stand for several spaces in the
|
|
|
1027 * checked word. */
|
|
|
1028 if (c == ' ')
|
|
|
1029 {
|
|
|
1030 for (;;)
|
|
|
1031 {
|
|
|
1032 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1033 flen = fold_more(mip);
|
|
|
1034 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1035 break;
|
|
|
1036 ++wlen;
|
|
|
1037 --flen;
|
|
|
1038 }
|
|
|
1039 }
|
|
236
|
1040 }
|
|
|
1041
|
|
300
|
1042 /*
|
|
|
1043 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1044 * first.
|
|
|
1045 */
|
|
|
1046 while (endidxcnt > 0)
|
|
|
1047 {
|
|
|
1048 --endidxcnt;
|
|
|
1049 arridx = endidx[endidxcnt];
|
|
|
1050 wlen = endlen[endidxcnt];
|
|
236
|
1051
|
|
300
|
1052 #ifdef FEAT_MBYTE
|
|
|
1053 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1054 continue; /* not at first byte of character */
|
|
|
1055 #endif
|
|
376
|
1056 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
300
|
1057 continue; /* next char is a word character */
|
|
|
1058
|
|
|
1059 #ifdef FEAT_MBYTE
|
|
339
|
1060 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1061 {
|
|
|
1062 /* Compute byte length in original word, length may change
|
|
339
|
1063 * when folding case. This can be slow, take a shortcut when the
|
|
|
1064 * case-folded word is equal to the keep-case word. */
|
|
300
|
1065 p = mip->mi_word;
|
|
339
|
1066 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1067 {
|
|
|
1068 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1069 mb_ptr_adv(p);
|
|
|
1070 wlen = p - mip->mi_word;
|
|
|
1071 }
|
|
300
|
1072 }
|
|
|
1073 #endif
|
|
236
|
1074
|
|
339
|
1075 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1076 * prefix ID.
|
|
|
1077 * Repeat this if there are more flags/region alternatives until there
|
|
|
1078 * is a match. */
|
|
|
1079 res = SP_BAD;
|
|
|
1080 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1081 --len, ++arridx)
|
|
300
|
1082 {
|
|
|
1083 flags = idxs[arridx];
|
|
324
|
1084
|
|
339
|
1085 /* For the fold-case tree check that the case of the checked word
|
|
|
1086 * matches with what the word in the tree requires.
|
|
|
1087 * For keep-case tree the case is always right. For prefixes we
|
|
|
1088 * don't bother to check. */
|
|
|
1089 if (mode == FIND_FOLDWORD)
|
|
300
|
1090 {
|
|
|
1091 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1092 {
|
|
323
|
1093 /* mi_capflags was set for a different word length, need
|
|
|
1094 * to do it again. */
|
|
300
|
1095 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1096 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1097 }
|
|
|
1098
|
|
346
|
1099 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1100 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1101 continue;
|
|
300
|
1102 }
|
|
236
|
1103
|
|
339
|
1104 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1105 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1106 * mip->mi_prefarridx that find_prefix() filled. */
|
|
339
|
1107 if (mode == FIND_PREFIX)
|
|
300
|
1108 {
|
|
339
|
1109 /* The prefix ID is stored two bytes above the flags. */
|
|
366
|
1110 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1111 flags,
|
|
455
|
1112 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1113 FALSE);
|
|
366
|
1114 if (c == 0)
|
|
339
|
1115 continue;
|
|
366
|
1116
|
|
|
1117 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1118 if (c & WF_RAREPFX)
|
|
|
1119 flags |= WF_RARE;
|
|
339
|
1120 }
|
|
|
1121
|
|
|
1122 if (flags & WF_BANNED)
|
|
|
1123 res = SP_BANNED;
|
|
|
1124 else if (flags & WF_REGION)
|
|
|
1125 {
|
|
|
1126 /* Check region. */
|
|
390
|
1127 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1128 res = SP_OK;
|
|
339
|
1129 else
|
|
|
1130 res = SP_LOCAL;
|
|
300
|
1131 }
|
|
339
|
1132 else if (flags & WF_RARE)
|
|
|
1133 res = SP_RARE;
|
|
307
|
1134 else
|
|
339
|
1135 res = SP_OK;
|
|
|
1136
|
|
|
1137 /* Always use the longest match and the best result. */
|
|
|
1138 if (mip->mi_result > res)
|
|
|
1139 {
|
|
|
1140 mip->mi_result = res;
|
|
|
1141 mip->mi_end = mip->mi_word + wlen;
|
|
|
1142 }
|
|
351
|
1143 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1144 mip->mi_end = mip->mi_word + wlen;
|
|
|
1145
|
|
|
1146 if (res == SP_OK)
|
|
|
1147 break;
|
|
300
|
1148 }
|
|
|
1149
|
|
307
|
1150 if (res == SP_OK)
|
|
300
|
1151 break;
|
|
|
1152 }
|
|
236
|
1153 }
|
|
|
1154
|
|
323
|
1155 /*
|
|
390
|
1156 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1157 * ID in "flags" for the word "word".
|
|
366
|
1158 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1159 */
|
|
|
1160 static int
|
|
455
|
1161 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1162 int totprefcnt; /* nr of prefix IDs */
|
|
|
1163 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1164 int flags;
|
|
351
|
1165 char_u *word;
|
|
|
1166 slang_T *slang;
|
|
455
|
1167 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1168 {
|
|
|
1169 int prefcnt;
|
|
|
1170 int pidx;
|
|
|
1171 regprog_T *rp;
|
|
|
1172 regmatch_T regmatch;
|
|
390
|
1173 int prefid;
|
|
|
1174
|
|
|
1175 prefid = (unsigned)flags >> 24;
|
|
351
|
1176 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1177 {
|
|
|
1178 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1179
|
|
|
1180 /* Check the prefix ID. */
|
|
|
1181 if (prefid != (pidx & 0xff))
|
|
|
1182 continue;
|
|
|
1183
|
|
390
|
1184 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1185 * suffix. */
|
|
|
1186 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1187 continue;
|
|
|
1188
|
|
351
|
1189 /* Check the condition, if there is one. The condition index is
|
|
366
|
1190 * stored in the two bytes above the prefix ID byte. */
|
|
|
1191 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1192 if (rp != NULL)
|
|
|
1193 {
|
|
|
1194 regmatch.regprog = rp;
|
|
|
1195 regmatch.rm_ic = FALSE;
|
|
|
1196 if (!vim_regexec(®match, word, 0))
|
|
|
1197 continue;
|
|
|
1198 }
|
|
455
|
1199 else if (cond_req)
|
|
|
1200 continue;
|
|
|
1201
|
|
|
1202 /* It's a match! Return the WF_ flags. */
|
|
366
|
1203 return pidx;
|
|
|
1204 }
|
|
|
1205 return 0;
|
|
351
|
1206 }
|
|
|
1207
|
|
|
1208 /*
|
|
339
|
1209 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1210 * If it does, then check the following word.
|
|
|
1211 *
|
|
|
1212 * For a match mip->mi_result is updated.
|
|
|
1213 */
|
|
|
1214 static void
|
|
|
1215 find_prefix(mip)
|
|
|
1216 matchinf_T *mip;
|
|
|
1217 {
|
|
|
1218 idx_T arridx = 0;
|
|
|
1219 int len;
|
|
|
1220 int wlen = 0;
|
|
|
1221 int flen;
|
|
|
1222 int c;
|
|
|
1223 char_u *ptr;
|
|
|
1224 idx_T lo, hi, m;
|
|
|
1225 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1226 char_u *byts;
|
|
|
1227 idx_T *idxs;
|
|
|
1228
|
|
375
|
1229 byts = slang->sl_pbyts;
|
|
|
1230 if (byts == NULL)
|
|
|
1231 return; /* array is empty */
|
|
|
1232
|
|
339
|
1233 /* We use the case-folded word here, since prefixes are always
|
|
|
1234 * case-folded. */
|
|
|
1235 ptr = mip->mi_fword;
|
|
|
1236 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1237 idxs = slang->sl_pidxs;
|
|
|
1238
|
|
|
1239 /*
|
|
|
1240 * Repeat advancing in the tree until:
|
|
|
1241 * - there is a byte that doesn't match,
|
|
|
1242 * - we reach the end of the tree,
|
|
|
1243 * - or we reach the end of the line.
|
|
|
1244 */
|
|
|
1245 for (;;)
|
|
|
1246 {
|
|
|
1247 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1248 flen = fold_more(mip);
|
|
|
1249
|
|
|
1250 len = byts[arridx++];
|
|
|
1251
|
|
|
1252 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1253 * Check if the following word matches and supports the prefix. */
|
|
|
1254 if (byts[arridx] == 0)
|
|
|
1255 {
|
|
|
1256 /* There can be several prefixes with different conditions. We
|
|
|
1257 * try them all, since we don't know which one will give the
|
|
|
1258 * longest match. The word is the same each time, pass the list
|
|
|
1259 * of possible prefixes to find_word(). */
|
|
|
1260 mip->mi_prefarridx = arridx;
|
|
|
1261 mip->mi_prefcnt = len;
|
|
|
1262 while (len > 0 && byts[arridx] == 0)
|
|
|
1263 {
|
|
|
1264 ++arridx;
|
|
|
1265 --len;
|
|
|
1266 }
|
|
|
1267 mip->mi_prefcnt -= len;
|
|
|
1268
|
|
|
1269 /* Find the word that comes after the prefix. */
|
|
|
1270 mip->mi_prefixlen = wlen;
|
|
455
|
1271 #ifdef FEAT_MBYTE
|
|
|
1272 if (has_mbyte)
|
|
|
1273 {
|
|
|
1274 /* Case-folded length may differ from original length. */
|
|
|
1275 mip->mi_cprefixlen = nofold_len(mip->mi_fword, wlen,
|
|
|
1276 mip->mi_word);
|
|
|
1277 }
|
|
|
1278 else
|
|
|
1279 mip->mi_cprefixlen = wlen;
|
|
|
1280 #endif
|
|
339
|
1281 find_word(mip, FIND_PREFIX);
|
|
|
1282
|
|
|
1283
|
|
|
1284 if (len == 0)
|
|
|
1285 break; /* no children, word must end here */
|
|
|
1286 }
|
|
|
1287
|
|
|
1288 /* Stop looking at end of the line. */
|
|
|
1289 if (ptr[wlen] == NUL)
|
|
|
1290 break;
|
|
|
1291
|
|
|
1292 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1293 c = ptr[wlen];
|
|
|
1294 lo = arridx;
|
|
|
1295 hi = arridx + len - 1;
|
|
|
1296 while (lo < hi)
|
|
|
1297 {
|
|
|
1298 m = (lo + hi) / 2;
|
|
|
1299 if (byts[m] > c)
|
|
|
1300 hi = m - 1;
|
|
|
1301 else if (byts[m] < c)
|
|
|
1302 lo = m + 1;
|
|
|
1303 else
|
|
|
1304 {
|
|
|
1305 lo = hi = m;
|
|
|
1306 break;
|
|
|
1307 }
|
|
|
1308 }
|
|
|
1309
|
|
|
1310 /* Stop if there is no matching byte. */
|
|
|
1311 if (hi < lo || byts[lo] != c)
|
|
|
1312 break;
|
|
|
1313
|
|
|
1314 /* Continue at the child (if there is one). */
|
|
|
1315 arridx = idxs[lo];
|
|
|
1316 ++wlen;
|
|
|
1317 --flen;
|
|
|
1318 }
|
|
|
1319 }
|
|
|
1320
|
|
|
1321 /*
|
|
|
1322 * Need to fold at least one more character. Do until next non-word character
|
|
|
1323 * for efficiency.
|
|
|
1324 * Return the length of the folded chars in bytes.
|
|
|
1325 */
|
|
|
1326 static int
|
|
|
1327 fold_more(mip)
|
|
|
1328 matchinf_T *mip;
|
|
|
1329 {
|
|
|
1330 int flen;
|
|
|
1331 char_u *p;
|
|
|
1332
|
|
|
1333 p = mip->mi_fend;
|
|
|
1334 do
|
|
|
1335 {
|
|
|
1336 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1337 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1338
|
|
|
1339 /* Include the non-word character so that we can check for the
|
|
|
1340 * word end. */
|
|
|
1341 if (*mip->mi_fend != NUL)
|
|
|
1342 mb_ptr_adv(mip->mi_fend);
|
|
|
1343
|
|
|
1344 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1345 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1346 MAXWLEN - mip->mi_fwordlen);
|
|
|
1347 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1348 mip->mi_fwordlen += flen;
|
|
|
1349 return flen;
|
|
|
1350 }
|
|
|
1351
|
|
|
1352 /*
|
|
323
|
1353 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1354 * case.
|
|
|
1355 */
|
|
|
1356 static int
|
|
389
|
1357 spell_valid_case(wordflags, treeflags)
|
|
|
1358 int wordflags; /* flags for the checked word. */
|
|
323
|
1359 int treeflags; /* flags for the word in the spell tree */
|
|
|
1360 {
|
|
389
|
1361 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1362 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1363 && ((treeflags & WF_ONECAP) == 0
|
|
|
1364 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1365 }
|
|
|
1366
|
|
351
|
1367 /*
|
|
|
1368 * Return TRUE if spell checking is not enabled.
|
|
|
1369 */
|
|
|
1370 static int
|
|
|
1371 no_spell_checking()
|
|
|
1372 {
|
|
|
1373 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1374 {
|
|
|
1375 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1376 return TRUE;
|
|
|
1377 }
|
|
|
1378 return FALSE;
|
|
|
1379 }
|
|
300
|
1380
|
|
236
|
1381 /*
|
|
|
1382 * Move to next spell error.
|
|
323
|
1383 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
236
|
1384 * Return OK if found, FAIL otherwise.
|
|
|
1385 */
|
|
|
1386 int
|
|
323
|
1387 spell_move_to(dir, allwords, curline)
|
|
236
|
1388 int dir; /* FORWARD or BACKWARD */
|
|
|
1389 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1390 int curline;
|
|
236
|
1391 {
|
|
249
|
1392 linenr_T lnum;
|
|
|
1393 pos_T found_pos;
|
|
236
|
1394 char_u *line;
|
|
|
1395 char_u *p;
|
|
346
|
1396 char_u *endp;
|
|
|
1397 int attr;
|
|
236
|
1398 int len;
|
|
249
|
1399 int has_syntax = syntax_present(curbuf);
|
|
|
1400 int col;
|
|
|
1401 int can_spell;
|
|
346
|
1402 char_u *buf = NULL;
|
|
|
1403 int buflen = 0;
|
|
|
1404 int skip = 0;
|
|
385
|
1405 int capcol = -1;
|
|
236
|
1406
|
|
351
|
1407 if (no_spell_checking())
|
|
236
|
1408 return FAIL;
|
|
|
1409
|
|
249
|
1410 /*
|
|
|
1411 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1412 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1413 *
|
|
|
1414 * When searching backwards, we continue in the line to find the last
|
|
|
1415 * bad word (in the cursor line: before the cursor).
|
|
346
|
1416 *
|
|
|
1417 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1418 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1419 * though...
|
|
249
|
1420 */
|
|
|
1421 lnum = curwin->w_cursor.lnum;
|
|
|
1422 found_pos.lnum = 0;
|
|
236
|
1423
|
|
|
1424 while (!got_int)
|
|
|
1425 {
|
|
249
|
1426 line = ml_get(lnum);
|
|
346
|
1427
|
|
|
1428 len = STRLEN(line);
|
|
|
1429 if (buflen < len + MAXWLEN + 2)
|
|
|
1430 {
|
|
|
1431 vim_free(buf);
|
|
|
1432 buflen = len + MAXWLEN + 2;
|
|
|
1433 buf = alloc(buflen);
|
|
|
1434 if (buf == NULL)
|
|
|
1435 break;
|
|
|
1436 }
|
|
|
1437
|
|
385
|
1438 /* In first line check first word for Capital. */
|
|
|
1439 if (lnum == 1)
|
|
|
1440 capcol = 0;
|
|
|
1441
|
|
|
1442 /* For checking first word with a capital skip white space. */
|
|
|
1443 if (capcol == 0)
|
|
|
1444 capcol = skipwhite(line) - line;
|
|
|
1445
|
|
346
|
1446 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1447 * possible. */
|
|
|
1448 STRCPY(buf, line);
|
|
|
1449 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1450 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1451
|
|
|
1452 p = buf + skip;
|
|
|
1453 endp = buf + len;
|
|
|
1454 while (p < endp)
|
|
236
|
1455 {
|
|
300
|
1456 /* When searching backward don't search after the cursor. */
|
|
|
1457 if (dir == BACKWARD
|
|
|
1458 && lnum == curwin->w_cursor.lnum
|
|
346
|
1459 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1460 break;
|
|
249
|
1461
|
|
300
|
1462 /* start of word */
|
|
346
|
1463 attr = 0;
|
|
385
|
1464 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1465
|
|
300
|
1466 if (attr != 0)
|
|
|
1467 {
|
|
|
1468 /* We found a bad word. Check the attribute. */
|
|
|
1469 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1470 {
|
|
300
|
1471 /* When searching forward only accept a bad word after
|
|
|
1472 * the cursor. */
|
|
|
1473 if (dir == BACKWARD
|
|
|
1474 || lnum > curwin->w_cursor.lnum
|
|
|
1475 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1476 && (colnr_T)(curline ? p - buf + len
|
|
|
1477 : p - buf)
|
|
300
|
1478 > curwin->w_cursor.col))
|
|
236
|
1479 {
|
|
300
|
1480 if (has_syntax)
|
|
249
|
1481 {
|
|
346
|
1482 col = p - buf;
|
|
300
|
1483 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1484 FALSE, &can_spell);
|
|
|
1485 }
|
|
|
1486 else
|
|
|
1487 can_spell = TRUE;
|
|
249
|
1488
|
|
300
|
1489 if (can_spell)
|
|
|
1490 {
|
|
|
1491 found_pos.lnum = lnum;
|
|
346
|
1492 found_pos.col = p - buf;
|
|
249
|
1493 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1494 found_pos.coladd = 0;
|
|
249
|
1495 #endif
|
|
300
|
1496 if (dir == FORWARD)
|
|
|
1497 {
|
|
|
1498 /* No need to search further. */
|
|
|
1499 curwin->w_cursor = found_pos;
|
|
346
|
1500 vim_free(buf);
|
|
300
|
1501 return OK;
|
|
249
|
1502 }
|
|
|
1503 }
|
|
236
|
1504 }
|
|
|
1505 }
|
|
|
1506 }
|
|
|
1507
|
|
300
|
1508 /* advance to character after the word */
|
|
|
1509 p += len;
|
|
385
|
1510 capcol -= len;
|
|
236
|
1511 }
|
|
|
1512
|
|
323
|
1513 if (curline)
|
|
346
|
1514 break; /* only check cursor line */
|
|
323
|
1515
|
|
236
|
1516 /* Advance to next line. */
|
|
249
|
1517 if (dir == BACKWARD)
|
|
|
1518 {
|
|
|
1519 if (found_pos.lnum != 0)
|
|
|
1520 {
|
|
|
1521 /* Use the last match in the line. */
|
|
|
1522 curwin->w_cursor = found_pos;
|
|
346
|
1523 vim_free(buf);
|
|
249
|
1524 return OK;
|
|
|
1525 }
|
|
|
1526 if (lnum == 1)
|
|
346
|
1527 break;
|
|
249
|
1528 --lnum;
|
|
385
|
1529 capcol = -1;
|
|
249
|
1530 }
|
|
|
1531 else
|
|
|
1532 {
|
|
|
1533 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1534 break;
|
|
249
|
1535 ++lnum;
|
|
346
|
1536
|
|
|
1537 /* Skip the characters at the start of the next line that were
|
|
|
1538 * included in a match crossing line boundaries. */
|
|
|
1539 if (attr == 0)
|
|
|
1540 skip = p - endp;
|
|
|
1541 else
|
|
|
1542 skip = 0;
|
|
385
|
1543
|
|
|
1544 /* Capscol skips over the inserted space. */
|
|
|
1545 --capcol;
|
|
|
1546
|
|
|
1547 /* But after empty line check first word in next line */
|
|
|
1548 if (*skipwhite(line) == NUL)
|
|
|
1549 capcol = 0;
|
|
249
|
1550 }
|
|
236
|
1551
|
|
|
1552 line_breakcheck();
|
|
|
1553 }
|
|
|
1554
|
|
346
|
1555 vim_free(buf);
|
|
|
1556 return FAIL;
|
|
|
1557 }
|
|
|
1558
|
|
|
1559 /*
|
|
|
1560 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1561 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1562 */
|
|
|
1563 void
|
|
|
1564 spell_cat_line(buf, line, maxlen)
|
|
|
1565 char_u *buf;
|
|
|
1566 char_u *line;
|
|
|
1567 int maxlen;
|
|
|
1568 {
|
|
|
1569 char_u *p;
|
|
|
1570 int n;
|
|
|
1571
|
|
|
1572 p = skipwhite(line);
|
|
|
1573 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1574 p = skipwhite(p + 1);
|
|
|
1575
|
|
|
1576 if (*p != NUL)
|
|
|
1577 {
|
|
|
1578 *buf = ' ';
|
|
376
|
1579 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1580 n = p - line;
|
|
|
1581 if (n >= maxlen)
|
|
|
1582 n = maxlen - 1;
|
|
|
1583 vim_memset(buf + 1, ' ', n);
|
|
|
1584 }
|
|
236
|
1585 }
|
|
|
1586
|
|
|
1587 /*
|
|
307
|
1588 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1589 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1590 */
|
|
307
|
1591 static void
|
|
236
|
1592 spell_load_lang(lang)
|
|
|
1593 char_u *lang;
|
|
|
1594 {
|
|
310
|
1595 char_u fname_enc[85];
|
|
236
|
1596 int r;
|
|
307
|
1597 char_u langcp[MAXWLEN + 1];
|
|
|
1598
|
|
310
|
1599 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1600 * It's truncated when an error is detected. */
|
|
|
1601 STRCPY(langcp, lang);
|
|
|
1602
|
|
310
|
1603 /*
|
|
|
1604 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1605 */
|
|
|
1606 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1607 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1608 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1609
|
|
|
1610 if (r == FAIL && *langcp != NUL)
|
|
|
1611 {
|
|
|
1612 /* Try loading the ASCII version. */
|
|
310
|
1613 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1614 "spell/%s.ascii.spl", lang);
|
|
310
|
1615 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1616 }
|
|
|
1617
|
|
|
1618 if (r == FAIL)
|
|
|
1619 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
|
236
|
1620 fname_enc + 6);
|
|
310
|
1621 else if (*langcp != NUL)
|
|
|
1622 {
|
|
|
1623 /* Load all the additions. */
|
|
|
1624 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1625 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1626 }
|
|
|
1627 }
|
|
|
1628
|
|
|
1629 /*
|
|
|
1630 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1631 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1632 */
|
|
|
1633 static char_u *
|
|
|
1634 spell_enc()
|
|
|
1635 {
|
|
|
1636
|
|
|
1637 #ifdef FEAT_MBYTE
|
|
|
1638 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1639 return p_enc;
|
|
|
1640 #endif
|
|
|
1641 return (char_u *)"latin1";
|
|
236
|
1642 }
|
|
|
1643
|
|
|
1644 /*
|
|
385
|
1645 * Get the name of the .spl file for the internal wordlist into
|
|
|
1646 * "fname[MAXPATHL]".
|
|
|
1647 */
|
|
|
1648 static void
|
|
|
1649 int_wordlist_spl(fname)
|
|
|
1650 char_u *fname;
|
|
|
1651 {
|
|
|
1652 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
1653 int_wordlist, spell_enc());
|
|
|
1654 }
|
|
|
1655
|
|
|
1656 /*
|
|
236
|
1657 * Allocate a new slang_T.
|
|
|
1658 * Caller must fill "sl_next".
|
|
|
1659 */
|
|
|
1660 static slang_T *
|
|
|
1661 slang_alloc(lang)
|
|
|
1662 char_u *lang;
|
|
|
1663 {
|
|
|
1664 slang_T *lp;
|
|
|
1665
|
|
300
|
1666 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1667 if (lp != NULL)
|
|
|
1668 {
|
|
|
1669 lp->sl_name = vim_strsave(lang);
|
|
323
|
1670 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
236
|
1671 }
|
|
|
1672 return lp;
|
|
|
1673 }
|
|
|
1674
|
|
|
1675 /*
|
|
|
1676 * Free the contents of an slang_T and the structure itself.
|
|
|
1677 */
|
|
|
1678 static void
|
|
|
1679 slang_free(lp)
|
|
|
1680 slang_T *lp;
|
|
|
1681 {
|
|
|
1682 vim_free(lp->sl_name);
|
|
310
|
1683 vim_free(lp->sl_fname);
|
|
|
1684 slang_clear(lp);
|
|
|
1685 vim_free(lp);
|
|
|
1686 }
|
|
|
1687
|
|
|
1688 /*
|
|
|
1689 * Clear an slang_T so that the file can be reloaded.
|
|
|
1690 */
|
|
|
1691 static void
|
|
|
1692 slang_clear(lp)
|
|
|
1693 slang_T *lp;
|
|
|
1694 {
|
|
339
|
1695 garray_T *gap;
|
|
|
1696 fromto_T *ftp;
|
|
344
|
1697 salitem_T *smp;
|
|
339
|
1698 int i;
|
|
323
|
1699
|
|
300
|
1700 vim_free(lp->sl_fbyts);
|
|
310
|
1701 lp->sl_fbyts = NULL;
|
|
300
|
1702 vim_free(lp->sl_kbyts);
|
|
310
|
1703 lp->sl_kbyts = NULL;
|
|
339
|
1704 vim_free(lp->sl_pbyts);
|
|
|
1705 lp->sl_pbyts = NULL;
|
|
|
1706
|
|
300
|
1707 vim_free(lp->sl_fidxs);
|
|
310
|
1708 lp->sl_fidxs = NULL;
|
|
300
|
1709 vim_free(lp->sl_kidxs);
|
|
310
|
1710 lp->sl_kidxs = NULL;
|
|
339
|
1711 vim_free(lp->sl_pidxs);
|
|
|
1712 lp->sl_pidxs = NULL;
|
|
323
|
1713
|
|
344
|
1714 gap = &lp->sl_rep;
|
|
|
1715 while (gap->ga_len > 0)
|
|
323
|
1716 {
|
|
344
|
1717 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1718 vim_free(ftp->ft_from);
|
|
|
1719 vim_free(ftp->ft_to);
|
|
323
|
1720 }
|
|
344
|
1721 ga_clear(gap);
|
|
|
1722
|
|
|
1723 gap = &lp->sl_sal;
|
|
375
|
1724 if (lp->sl_sofo)
|
|
376
|
1725 {
|
|
|
1726 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1727 if (gap->ga_data != NULL)
|
|
|
1728 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1729 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1730 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1731 }
|
|
375
|
1732 else
|
|
|
1733 /* SAL items: free salitem_T items */
|
|
|
1734 while (gap->ga_len > 0)
|
|
|
1735 {
|
|
|
1736 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1737 vim_free(smp->sm_lead);
|
|
|
1738 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1739 vim_free(smp->sm_to);
|
|
|
1740 #ifdef FEAT_MBYTE
|
|
|
1741 vim_free(smp->sm_lead_w);
|
|
|
1742 vim_free(smp->sm_oneof_w);
|
|
|
1743 vim_free(smp->sm_to_w);
|
|
|
1744 #endif
|
|
|
1745 }
|
|
344
|
1746 ga_clear(gap);
|
|
323
|
1747
|
|
339
|
1748 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1749 vim_free(lp->sl_prefprog[i]);
|
|
376
|
1750 lp->sl_prefixcnt = 0;
|
|
339
|
1751 vim_free(lp->sl_prefprog);
|
|
376
|
1752 lp->sl_prefprog = NULL;
|
|
|
1753
|
|
|
1754 vim_free(lp->sl_midword);
|
|
|
1755 lp->sl_midword = NULL;
|
|
339
|
1756
|
|
330
|
1757 #ifdef FEAT_MBYTE
|
|
|
1758 {
|
|
|
1759 int todo = lp->sl_map_hash.ht_used;
|
|
|
1760 hashitem_T *hi;
|
|
|
1761
|
|
|
1762 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1763 if (!HASHITEM_EMPTY(hi))
|
|
|
1764 {
|
|
|
1765 --todo;
|
|
|
1766 vim_free(hi->hi_key);
|
|
|
1767 }
|
|
|
1768 }
|
|
|
1769 hash_clear(&lp->sl_map_hash);
|
|
|
1770 #endif
|
|
236
|
1771 }
|
|
|
1772
|
|
|
1773 /*
|
|
307
|
1774 * Load one spell file and store the info into a slang_T.
|
|
236
|
1775 * Invoked through do_in_runtimepath().
|
|
|
1776 */
|
|
|
1777 static void
|
|
310
|
1778 spell_load_cb(fname, cookie)
|
|
236
|
1779 char_u *fname;
|
|
307
|
1780 void *cookie; /* points to the language name */
|
|
236
|
1781 {
|
|
323
|
1782 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1783 }
|
|
|
1784
|
|
|
1785 /*
|
|
|
1786 * Load one spell file and store the info into a slang_T.
|
|
|
1787 *
|
|
|
1788 * This is invoked in two ways:
|
|
|
1789 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
1790 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
1791 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
1792 * points to the existing slang_T.
|
|
323
|
1793 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
1794 */
|
|
323
|
1795 static slang_T *
|
|
|
1796 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
1797 char_u *fname;
|
|
|
1798 char_u *lang;
|
|
|
1799 slang_T *old_lp;
|
|
323
|
1800 int silent; /* no error if file doesn't exist */
|
|
310
|
1801 {
|
|
236
|
1802 FILE *fd;
|
|
|
1803 char_u buf[MAXWLEN + 1];
|
|
|
1804 char_u *p;
|
|
339
|
1805 char_u *bp;
|
|
|
1806 idx_T *ip;
|
|
236
|
1807 int i;
|
|
339
|
1808 int n;
|
|
300
|
1809 int len;
|
|
236
|
1810 int round;
|
|
|
1811 char_u *save_sourcing_name = sourcing_name;
|
|
|
1812 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
255
|
1813 int cnt, ccnt;
|
|
|
1814 char_u *fol;
|
|
307
|
1815 slang_T *lp = NULL;
|
|
323
|
1816 garray_T *gap;
|
|
|
1817 fromto_T *ftp;
|
|
344
|
1818 salitem_T *smp;
|
|
323
|
1819 short *first;
|
|
324
|
1820 idx_T idx;
|
|
344
|
1821 int c = 0;
|
|
236
|
1822
|
|
310
|
1823 fd = mch_fopen((char *)fname, "r");
|
|
236
|
1824 if (fd == NULL)
|
|
|
1825 {
|
|
323
|
1826 if (!silent)
|
|
|
1827 EMSG2(_(e_notopen), fname);
|
|
|
1828 else if (p_verbose > 2)
|
|
|
1829 {
|
|
|
1830 verbose_enter();
|
|
|
1831 smsg((char_u *)e_notopen, fname);
|
|
|
1832 verbose_leave();
|
|
|
1833 }
|
|
255
|
1834 goto endFAIL;
|
|
236
|
1835 }
|
|
310
|
1836 if (p_verbose > 2)
|
|
|
1837 {
|
|
|
1838 verbose_enter();
|
|
|
1839 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
1840 verbose_leave();
|
|
|
1841 }
|
|
|
1842
|
|
|
1843 if (old_lp == NULL)
|
|
|
1844 {
|
|
|
1845 lp = slang_alloc(lang);
|
|
|
1846 if (lp == NULL)
|
|
|
1847 goto endFAIL;
|
|
|
1848
|
|
|
1849 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
1850 lp->sl_fname = vim_strsave(fname);
|
|
|
1851 if (lp->sl_fname == NULL)
|
|
|
1852 goto endFAIL;
|
|
|
1853
|
|
|
1854 /* Check for .add.spl. */
|
|
|
1855 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
1856 }
|
|
|
1857 else
|
|
|
1858 lp = old_lp;
|
|
307
|
1859
|
|
236
|
1860 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
1861 sourcing_name = fname;
|
|
|
1862 sourcing_lnum = 0;
|
|
|
1863
|
|
339
|
1864 /* <HEADER>: <fileID>
|
|
|
1865 * <regioncnt> <regionname> ...
|
|
|
1866 * <charflagslen> <charflags>
|
|
|
1867 * <fcharslen> <fchars>
|
|
366
|
1868 * <midwordlen> <midword>
|
|
339
|
1869 * <prefcondcnt> <prefcond> ...
|
|
|
1870 */
|
|
236
|
1871 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
1872 buf[i] = getc(fd); /* <fileID> */
|
|
|
1873 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
1874 {
|
|
|
1875 EMSG(_("E757: Wrong file ID in spell file"));
|
|
255
|
1876 goto endFAIL;
|
|
236
|
1877 }
|
|
|
1878
|
|
|
1879 cnt = getc(fd); /* <regioncnt> */
|
|
255
|
1880 if (cnt < 0)
|
|
236
|
1881 {
|
|
|
1882 truncerr:
|
|
381
|
1883 EMSG(_(e_spell_trunc));
|
|
255
|
1884 goto endFAIL;
|
|
236
|
1885 }
|
|
|
1886 if (cnt > 8)
|
|
|
1887 {
|
|
|
1888 formerr:
|
|
307
|
1889 EMSG(_(e_format));
|
|
255
|
1890 goto endFAIL;
|
|
236
|
1891 }
|
|
|
1892 for (i = 0; i < cnt; ++i)
|
|
|
1893 {
|
|
|
1894 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
|
1895 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
|
1896 }
|
|
|
1897 lp->sl_regions[cnt * 2] = NUL;
|
|
|
1898
|
|
381
|
1899 /* <charflagslen> <charflags> */
|
|
|
1900 p = read_cnt_string(fd, 1, &cnt);
|
|
389
|
1901 if (cnt < 0)
|
|
381
|
1902 goto endFAIL;
|
|
|
1903
|
|
|
1904 /* <fcharslen> <fchars> */
|
|
389
|
1905 fol = read_cnt_string(fd, 2, &ccnt);
|
|
|
1906 if (ccnt < 0)
|
|
381
|
1907 {
|
|
300
|
1908 vim_free(p);
|
|
381
|
1909 goto endFAIL;
|
|
|
1910 }
|
|
|
1911
|
|
|
1912 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
1913 if (p != NULL && fol != NULL)
|
|
389
|
1914 i = set_spell_charflags(p, cnt, fol);
|
|
381
|
1915
|
|
|
1916 vim_free(p);
|
|
|
1917 vim_free(fol);
|
|
|
1918
|
|
|
1919 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
1920 if ((p == NULL) != (fol == NULL))
|
|
|
1921 goto formerr;
|
|
255
|
1922
|
|
366
|
1923 /* <midwordlen> <midword> */
|
|
376
|
1924 lp->sl_midword = read_cnt_string(fd, 2, &cnt);
|
|
389
|
1925 if (cnt < 0)
|
|
376
|
1926 goto endFAIL;
|
|
366
|
1927
|
|
339
|
1928 /* <prefcondcnt> <prefcond> ... */
|
|
|
1929 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
1930 if (cnt > 0)
|
|
|
1931 {
|
|
|
1932 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
1933 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
1934 if (lp->sl_prefprog == NULL)
|
|
|
1935 goto endFAIL;
|
|
|
1936 lp->sl_prefixcnt = cnt;
|
|
|
1937
|
|
|
1938 for (i = 0; i < cnt; ++i)
|
|
|
1939 {
|
|
|
1940 /* <prefcond> : <condlen> <condstr> */
|
|
|
1941 n = getc(fd); /* <condlen> */
|
|
|
1942 if (n < 0)
|
|
|
1943 goto formerr;
|
|
|
1944 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
1945 * compile the regexp program used to check for the condition. */
|
|
|
1946 if (n > 0)
|
|
|
1947 {
|
|
344
|
1948 buf[0] = '^'; /* always match at one position only */
|
|
|
1949 p = buf + 1;
|
|
339
|
1950 while (n-- > 0)
|
|
|
1951 *p++ = getc(fd); /* <condstr> */
|
|
|
1952 *p = NUL;
|
|
|
1953 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
1954 }
|
|
|
1955 }
|
|
|
1956 }
|
|
|
1957
|
|
|
1958
|
|
323
|
1959 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
1960 * <salflags> <salcount> <sal> ...
|
|
|
1961 * <maplen> <mapstr> */
|
|
339
|
1962
|
|
344
|
1963 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
1964 if (cnt < 0)
|
|
|
1965 goto formerr;
|
|
|
1966
|
|
|
1967 gap = &lp->sl_rep;
|
|
|
1968 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1969 goto endFAIL;
|
|
|
1970
|
|
|
1971 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
1972 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
1973 {
|
|
|
1974 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
381
|
1975 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
389
|
1976 if (i <= 0)
|
|
381
|
1977 goto endFAIL;
|
|
|
1978 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
389
|
1979 if (i <= 0)
|
|
381
|
1980 {
|
|
|
1981 vim_free(ftp->ft_from);
|
|
|
1982 goto endFAIL;
|
|
344
|
1983 }
|
|
|
1984 }
|
|
|
1985
|
|
|
1986 /* Fill the first-index table. */
|
|
|
1987 first = lp->sl_rep_first;
|
|
|
1988 for (i = 0; i < 256; ++i)
|
|
|
1989 first[i] = -1;
|
|
|
1990 for (i = 0; i < gap->ga_len; ++i)
|
|
323
|
1991 {
|
|
344
|
1992 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
1993 if (first[*ftp->ft_from] == -1)
|
|
|
1994 first[*ftp->ft_from] = i;
|
|
|
1995 }
|
|
|
1996
|
|
|
1997 i = getc(fd); /* <salflags> */
|
|
|
1998 if (i & SAL_F0LLOWUP)
|
|
|
1999 lp->sl_followup = TRUE;
|
|
|
2000 if (i & SAL_COLLAPSE)
|
|
|
2001 lp->sl_collapse = TRUE;
|
|
|
2002 if (i & SAL_REM_ACCENTS)
|
|
|
2003 lp->sl_rem_accents = TRUE;
|
|
375
|
2004 if (i & SAL_SOFO)
|
|
|
2005 lp->sl_sofo = TRUE;
|
|
389
|
2006 else
|
|
|
2007 lp->sl_sofo = FALSE;
|
|
344
|
2008
|
|
|
2009 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2010 if (cnt < 0)
|
|
|
2011 goto formerr;
|
|
|
2012
|
|
375
|
2013 if (lp->sl_sofo)
|
|
|
2014 {
|
|
|
2015 /*
|
|
|
2016 * SOFOFROM and SOFOTO items come in one <salfrom> and <salto>
|
|
|
2017 */
|
|
|
2018 if (cnt != 1)
|
|
|
2019 goto formerr;
|
|
|
2020
|
|
381
|
2021 /* <salfromlen> <salfrom> */
|
|
|
2022 bp = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2023 if (cnt < 0)
|
|
375
|
2024 goto endFAIL;
|
|
381
|
2025
|
|
|
2026 /* <saltolen> <salto> */
|
|
|
2027 fol = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2028 if (cnt < 0)
|
|
375
|
2029 {
|
|
|
2030 vim_free(bp);
|
|
344
|
2031 goto endFAIL;
|
|
375
|
2032 }
|
|
381
|
2033
|
|
|
2034 /* Store the info in lp->sl_sal and/or lp->sl_sal_first. */
|
|
389
|
2035 if (bp != NULL && fol != NULL)
|
|
|
2036 i = set_sofo(lp, bp, fol);
|
|
|
2037 else if (bp != NULL || fol != NULL)
|
|
|
2038 i = FAIL; /* only one of two strings is an error */
|
|
|
2039 else
|
|
|
2040 i = OK;
|
|
381
|
2041
|
|
375
|
2042 vim_free(bp);
|
|
|
2043 vim_free(fol);
|
|
381
|
2044 if (i == FAIL)
|
|
|
2045 goto formerr;
|
|
375
|
2046 }
|
|
|
2047 else
|
|
|
2048 {
|
|
|
2049 /*
|
|
|
2050 * SAL items
|
|
|
2051 */
|
|
|
2052 gap = &lp->sl_sal;
|
|
390
|
2053 ga_init2(gap, sizeof(salitem_T), 10);
|
|
375
|
2054 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2055 goto endFAIL;
|
|
|
2056
|
|
|
2057 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2058 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2059 {
|
|
|
2060 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2061 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2062 if (ccnt < 0)
|
|
|
2063 goto formerr;
|
|
|
2064 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2065 goto endFAIL;
|
|
|
2066 smp->sm_lead = p;
|
|
|
2067
|
|
|
2068 /* Read up to the first special char into sm_lead. */
|
|
|
2069 for (i = 0; i < ccnt; ++i)
|
|
|
2070 {
|
|
|
2071 c = getc(fd); /* <salfrom> */
|
|
|
2072 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2073 break;
|
|
|
2074 *p++ = c;
|
|
|
2075 }
|
|
|
2076 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2077 *p++ = NUL;
|
|
|
2078
|
|
|
2079 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2080 if (c == '(')
|
|
|
2081 {
|
|
|
2082 smp->sm_oneof = p;
|
|
|
2083 for (++i; i < ccnt; ++i)
|
|
|
2084 {
|
|
|
2085 c = getc(fd); /* <salfrom> */
|
|
|
2086 if (c == ')')
|
|
|
2087 break;
|
|
|
2088 *p++ = c;
|
|
|
2089 }
|
|
|
2090 *p++ = NUL;
|
|
|
2091 if (++i < ccnt)
|
|
|
2092 c = getc(fd);
|
|
|
2093 }
|
|
372
|
2094 else
|
|
375
|
2095 smp->sm_oneof = NULL;
|
|
|
2096
|
|
|
2097 /* Any following chars go in sm_rules. */
|
|
|
2098 smp->sm_rules = p;
|
|
|
2099 if (i < ccnt)
|
|
|
2100 /* store the char we got while checking for end of sm_lead */
|
|
|
2101 *p++ = c;
|
|
|
2102 for (++i; i < ccnt; ++i)
|
|
|
2103 *p++ = getc(fd); /* <salfrom> */
|
|
|
2104 *p++ = NUL;
|
|
|
2105
|
|
381
|
2106 /* <saltolen> <salto> */
|
|
|
2107 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
389
|
2108 if (ccnt < 0)
|
|
372
|
2109 {
|
|
|
2110 vim_free(smp->sm_lead);
|
|
375
|
2111 goto formerr;
|
|
|
2112 }
|
|
|
2113
|
|
372
|
2114 #ifdef FEAT_MBYTE
|
|
|
2115 if (has_mbyte)
|
|
|
2116 {
|
|
375
|
2117 /* convert the multi-byte strings to wide char strings */
|
|
|
2118 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2119 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2120 if (smp->sm_oneof == NULL)
|
|
|
2121 smp->sm_oneof_w = NULL;
|
|
|
2122 else
|
|
|
2123 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
389
|
2124 if (smp->sm_to == NULL)
|
|
|
2125 smp->sm_to_w = NULL;
|
|
|
2126 else
|
|
|
2127 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
375
|
2128 if (smp->sm_lead_w == NULL
|
|
|
2129 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
389
|
2130 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
372
|
2131 {
|
|
375
|
2132 vim_free(smp->sm_lead);
|
|
|
2133 vim_free(smp->sm_to);
|
|
|
2134 vim_free(smp->sm_lead_w);
|
|
|
2135 vim_free(smp->sm_oneof_w);
|
|
|
2136 vim_free(smp->sm_to_w);
|
|
|
2137 goto endFAIL;
|
|
372
|
2138 }
|
|
|
2139 }
|
|
|
2140 #endif
|
|
|
2141 }
|
|
375
|
2142
|
|
|
2143 /* Fill the first-index table. */
|
|
381
|
2144 set_sal_first(lp);
|
|
|
2145 }
|
|
|
2146
|
|
|
2147 /* <maplen> <mapstr> */
|
|
|
2148 p = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2149 if (cnt < 0)
|
|
323
|
2150 goto endFAIL;
|
|
389
|
2151 if (p != NULL)
|
|
|
2152 {
|
|
|
2153 set_map_str(lp, p);
|
|
|
2154 vim_free(p);
|
|
|
2155 }
|
|
236
|
2156
|
|
300
|
2157 /* round 1: <LWORDTREE>
|
|
339
|
2158 * round 2: <KWORDTREE>
|
|
|
2159 * round 3: <PREFIXTREE> */
|
|
|
2160 for (round = 1; round <= 3; ++round)
|
|
236
|
2161 {
|
|
300
|
2162 /* The tree size was computed when writing the file, so that we can
|
|
|
2163 * allocate it as one long block. <nodecount> */
|
|
|
2164 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2165 if (len < 0)
|
|
|
2166 goto truncerr;
|
|
|
2167 if (len > 0)
|
|
236
|
2168 {
|
|
300
|
2169 /* Allocate the byte array. */
|
|
339
|
2170 bp = lalloc((long_u)len, TRUE);
|
|
|
2171 if (bp == NULL)
|
|
300
|
2172 goto endFAIL;
|
|
|
2173 if (round == 1)
|
|
339
|
2174 lp->sl_fbyts = bp;
|
|
|
2175 else if (round == 2)
|
|
|
2176 lp->sl_kbyts = bp;
|
|
300
|
2177 else
|
|
339
|
2178 lp->sl_pbyts = bp;
|
|
236
|
2179
|
|
300
|
2180 /* Allocate the index array. */
|
|
339
|
2181 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2182 if (ip == NULL)
|
|
300
|
2183 goto endFAIL;
|
|
|
2184 if (round == 1)
|
|
339
|
2185 lp->sl_fidxs = ip;
|
|
|
2186 else if (round == 2)
|
|
|
2187 lp->sl_kidxs = ip;
|
|
300
|
2188 else
|
|
339
|
2189 lp->sl_pidxs = ip;
|
|
300
|
2190
|
|
|
2191 /* Read the tree and store it in the array. */
|
|
339
|
2192 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2193 if (idx == -1)
|
|
300
|
2194 goto truncerr;
|
|
324
|
2195 if (idx < 0)
|
|
236
|
2196 goto formerr;
|
|
|
2197 }
|
|
300
|
2198 }
|
|
243
|
2199
|
|
310
|
2200 /* For a new file link it in the list of spell files. */
|
|
|
2201 if (old_lp == NULL)
|
|
|
2202 {
|
|
|
2203 lp->sl_next = first_lang;
|
|
|
2204 first_lang = lp;
|
|
|
2205 }
|
|
307
|
2206
|
|
255
|
2207 goto endOK;
|
|
|
2208
|
|
|
2209 endFAIL:
|
|
310
|
2210 if (lang != NULL)
|
|
|
2211 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2212 *lang = NUL;
|
|
|
2213 if (lp != NULL && old_lp == NULL)
|
|
323
|
2214 {
|
|
307
|
2215 slang_free(lp);
|
|
323
|
2216 lp = NULL;
|
|
|
2217 }
|
|
255
|
2218
|
|
|
2219 endOK:
|
|
236
|
2220 if (fd != NULL)
|
|
|
2221 fclose(fd);
|
|
|
2222 sourcing_name = save_sourcing_name;
|
|
|
2223 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2224
|
|
|
2225 return lp;
|
|
236
|
2226 }
|
|
|
2227
|
|
376
|
2228 /*
|
|
|
2229 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2230 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2231 * Returns NULL when the count is zero.
|
|
389
|
2232 * Sets "*cntp" to -1 when there is an error, length of the result otherwise.
|
|
376
|
2233 */
|
|
|
2234 static char_u *
|
|
389
|
2235 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2236 FILE *fd;
|
|
|
2237 int cnt_bytes;
|
|
389
|
2238 int *cntp;
|
|
376
|
2239 {
|
|
|
2240 int cnt = 0;
|
|
|
2241 int i;
|
|
|
2242 char_u *str;
|
|
|
2243
|
|
|
2244 /* read the length bytes, MSB first */
|
|
|
2245 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2246 cnt = (cnt << 8) + getc(fd);
|
|
|
2247 if (cnt < 0)
|
|
|
2248 {
|
|
381
|
2249 EMSG(_(e_spell_trunc));
|
|
389
|
2250 *cntp = -1;
|
|
376
|
2251 return NULL;
|
|
|
2252 }
|
|
389
|
2253 *cntp = cnt;
|
|
|
2254 if (cnt == 0)
|
|
|
2255 return NULL; /* nothing to read, return NULL */
|
|
376
|
2256
|
|
|
2257 /* allocate memory */
|
|
|
2258 str = alloc((unsigned)cnt + 1);
|
|
|
2259 if (str == NULL)
|
|
|
2260 {
|
|
389
|
2261 *cntp = -1;
|
|
376
|
2262 return NULL;
|
|
|
2263 }
|
|
|
2264
|
|
|
2265 /* Read the string. Doesn't check for truncated file. */
|
|
|
2266 for (i = 0; i < cnt; ++i)
|
|
|
2267 str[i] = getc(fd);
|
|
|
2268 str[i] = NUL;
|
|
|
2269
|
|
|
2270 return str;
|
|
|
2271 }
|
|
|
2272
|
|
381
|
2273 /*
|
|
|
2274 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
|
2275 * Returns FAIL when there is something wrong.
|
|
|
2276 */
|
|
|
2277 static int
|
|
|
2278 set_sofo(lp, from, to)
|
|
|
2279 slang_T *lp;
|
|
|
2280 char_u *from;
|
|
|
2281 char_u *to;
|
|
|
2282 {
|
|
|
2283 int i;
|
|
|
2284
|
|
|
2285 #ifdef FEAT_MBYTE
|
|
|
2286 garray_T *gap;
|
|
|
2287 char_u *s;
|
|
|
2288 char_u *p;
|
|
|
2289 int c;
|
|
|
2290 int *inp;
|
|
|
2291
|
|
|
2292 if (has_mbyte)
|
|
|
2293 {
|
|
|
2294 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2295 * characters. The index is the low byte of the character.
|
|
|
2296 * The list contains from-to pairs with a terminating NUL.
|
|
|
2297 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2298 gap = &lp->sl_sal;
|
|
|
2299 ga_init2(gap, sizeof(int *), 1);
|
|
|
2300 if (ga_grow(gap, 256) == FAIL)
|
|
|
2301 return FAIL;
|
|
|
2302 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2303 gap->ga_len = 256;
|
|
|
2304
|
|
|
2305 /* First count the number of items for each list. Temporarily use
|
|
|
2306 * sl_sal_first[] for this. */
|
|
|
2307 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2308 {
|
|
474
|
2309 c = mb_cptr2char_adv(&p);
|
|
|
2310 mb_cptr_adv(s);
|
|
381
|
2311 if (c >= 256)
|
|
|
2312 ++lp->sl_sal_first[c & 0xff];
|
|
|
2313 }
|
|
|
2314 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
|
2315 return FAIL;
|
|
|
2316
|
|
|
2317 /* Allocate the lists. */
|
|
|
2318 for (i = 0; i < 256; ++i)
|
|
|
2319 if (lp->sl_sal_first[i] > 0)
|
|
|
2320 {
|
|
|
2321 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2322 if (p == NULL)
|
|
|
2323 return FAIL;
|
|
|
2324 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2325 *(int *)p = 0;
|
|
|
2326 }
|
|
|
2327
|
|
|
2328 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
2329 * list. */
|
|
|
2330 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2331 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2332 {
|
|
474
|
2333 c = mb_cptr2char_adv(&p);
|
|
|
2334 i = mb_cptr2char_adv(&s);
|
|
381
|
2335 if (c >= 256)
|
|
|
2336 {
|
|
|
2337 /* Append the from-to chars at the end of the list with
|
|
|
2338 * the low byte. */
|
|
|
2339 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2340 while (*inp != 0)
|
|
|
2341 ++inp;
|
|
|
2342 *inp++ = c; /* from char */
|
|
|
2343 *inp++ = i; /* to char */
|
|
|
2344 *inp++ = NUL; /* NUL at the end */
|
|
|
2345 }
|
|
|
2346 else
|
|
|
2347 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2348 lp->sl_sal_first[c] = i;
|
|
|
2349 }
|
|
|
2350 }
|
|
|
2351 else
|
|
|
2352 #endif
|
|
|
2353 {
|
|
|
2354 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2355 if (STRLEN(from) != STRLEN(to))
|
|
|
2356 return FAIL;
|
|
|
2357
|
|
|
2358 for (i = 0; to[i] != NUL; ++i)
|
|
|
2359 lp->sl_sal_first[from[i]] = to[i];
|
|
|
2360 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2361 }
|
|
|
2362
|
|
|
2363 return OK;
|
|
|
2364 }
|
|
|
2365
|
|
|
2366 /*
|
|
|
2367 * Fill the first-index table for "lp".
|
|
|
2368 */
|
|
|
2369 static void
|
|
|
2370 set_sal_first(lp)
|
|
|
2371 slang_T *lp;
|
|
|
2372 {
|
|
|
2373 salfirst_T *sfirst;
|
|
|
2374 int i;
|
|
|
2375 salitem_T *smp;
|
|
|
2376 int c;
|
|
|
2377 garray_T *gap = &lp->sl_sal;
|
|
|
2378
|
|
|
2379 sfirst = lp->sl_sal_first;
|
|
|
2380 for (i = 0; i < 256; ++i)
|
|
|
2381 sfirst[i] = -1;
|
|
|
2382 smp = (salitem_T *)gap->ga_data;
|
|
|
2383 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2384 {
|
|
|
2385 #ifdef FEAT_MBYTE
|
|
|
2386 if (has_mbyte)
|
|
|
2387 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2388 * the character, for other encodings it should differ for most
|
|
|
2389 * characters. */
|
|
|
2390 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2391 else
|
|
|
2392 #endif
|
|
|
2393 c = *smp[i].sm_lead;
|
|
|
2394 if (sfirst[c] == -1)
|
|
|
2395 {
|
|
|
2396 sfirst[c] = i;
|
|
|
2397 #ifdef FEAT_MBYTE
|
|
|
2398 if (has_mbyte)
|
|
|
2399 {
|
|
|
2400 int n;
|
|
|
2401
|
|
|
2402 /* Make sure all entries with this byte are following each
|
|
|
2403 * other. Move the ones that are in the wrong position. Do
|
|
|
2404 * keep the same ordering! */
|
|
|
2405 while (i + 1 < gap->ga_len
|
|
|
2406 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
2407 /* Skip over entry with same index byte. */
|
|
|
2408 ++i;
|
|
|
2409
|
|
|
2410 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
2411 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
2412 {
|
|
|
2413 salitem_T tsal;
|
|
|
2414
|
|
|
2415 /* Move entry with same index byte after the entries
|
|
|
2416 * we already found. */
|
|
|
2417 ++i;
|
|
|
2418 --n;
|
|
|
2419 tsal = smp[i + n];
|
|
|
2420 mch_memmove(smp + i + 1, smp + i,
|
|
|
2421 sizeof(salitem_T) * n);
|
|
|
2422 smp[i] = tsal;
|
|
|
2423 }
|
|
|
2424 }
|
|
|
2425 #endif
|
|
|
2426 }
|
|
|
2427 }
|
|
|
2428 }
|
|
376
|
2429
|
|
372
|
2430 #ifdef FEAT_MBYTE
|
|
|
2431 /*
|
|
|
2432 * Turn a multi-byte string into a wide character string.
|
|
|
2433 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
2434 */
|
|
|
2435 static int *
|
|
|
2436 mb_str2wide(s)
|
|
|
2437 char_u *s;
|
|
|
2438 {
|
|
|
2439 int *res;
|
|
|
2440 char_u *p;
|
|
|
2441 int i = 0;
|
|
|
2442
|
|
|
2443 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
2444 if (res != NULL)
|
|
|
2445 {
|
|
|
2446 for (p = s; *p != NUL; )
|
|
|
2447 res[i++] = mb_ptr2char_adv(&p);
|
|
|
2448 res[i] = NUL;
|
|
|
2449 }
|
|
|
2450 return res;
|
|
|
2451 }
|
|
|
2452 #endif
|
|
|
2453
|
|
236
|
2454 /*
|
|
300
|
2455 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
2456 * "byts" and index array "idxs". Recursively read the children.
|
|
|
2457 *
|
|
346
|
2458 * NOTE: The code here must match put_node().
|
|
300
|
2459 *
|
|
|
2460 * Returns the index follosing the siblings.
|
|
|
2461 * Returns -1 if the file is shorter than expected.
|
|
|
2462 * Returns -2 if there is a format error.
|
|
236
|
2463 */
|
|
324
|
2464 static idx_T
|
|
339
|
2465 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
2466 FILE *fd;
|
|
|
2467 char_u *byts;
|
|
324
|
2468 idx_T *idxs;
|
|
300
|
2469 int maxidx; /* size of arrays */
|
|
324
|
2470 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
2471 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
2472 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
2473 {
|
|
300
|
2474 int len;
|
|
|
2475 int i;
|
|
|
2476 int n;
|
|
324
|
2477 idx_T idx = startidx;
|
|
300
|
2478 int c;
|
|
366
|
2479 int c2;
|
|
300
|
2480 #define SHARED_MASK 0x8000000
|
|
236
|
2481
|
|
300
|
2482 len = getc(fd); /* <siblingcount> */
|
|
|
2483 if (len <= 0)
|
|
|
2484 return -1;
|
|
|
2485
|
|
|
2486 if (startidx + len >= maxidx)
|
|
|
2487 return -2;
|
|
|
2488 byts[idx++] = len;
|
|
|
2489
|
|
|
2490 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
2491 for (i = 1; i <= len; ++i)
|
|
236
|
2492 {
|
|
300
|
2493 c = getc(fd); /* <byte> */
|
|
|
2494 if (c < 0)
|
|
|
2495 return -1;
|
|
|
2496 if (c <= BY_SPECIAL)
|
|
|
2497 {
|
|
366
|
2498 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
2499 {
|
|
|
2500 /* No flags, all regions. */
|
|
|
2501 idxs[idx] = 0;
|
|
|
2502 c = 0;
|
|
|
2503 }
|
|
390
|
2504 else if (c != BY_INDEX)
|
|
300
|
2505 {
|
|
339
|
2506 if (prefixtree)
|
|
|
2507 {
|
|
455
|
2508 /* Read the optional pflags byte, the prefix ID and the
|
|
|
2509 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
2510 * byte, the condition index shifted up 8 bits, the flags
|
|
|
2511 * shifted up 24 bits. */
|
|
|
2512 if (c == BY_FLAGS)
|
|
|
2513 c = getc(fd) << 24; /* <pflags> */
|
|
|
2514 else
|
|
|
2515 c = 0;
|
|
|
2516
|
|
|
2517 c |= getc(fd); /* <prefixID> */
|
|
|
2518
|
|
339
|
2519 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
2520 if (n >= maxprefcondnr)
|
|
|
2521 return -2;
|
|
455
|
2522 c |= (n << 8);
|
|
339
|
2523 }
|
|
390
|
2524 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
2525 {
|
|
|
2526 /* Read flags and optional region and prefix ID. In
|
|
390
|
2527 * idxs[] the flags go in the low two bytes, region above
|
|
|
2528 * that and prefix ID above the region. */
|
|
|
2529 c2 = c;
|
|
339
|
2530 c = getc(fd); /* <flags> */
|
|
390
|
2531 if (c2 == BY_FLAGS2)
|
|
|
2532 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
2533 if (c & WF_REGION)
|
|
390
|
2534 c = (getc(fd) << 16) + c; /* <region> */
|
|
339
|
2535 if (c & WF_PFX)
|
|
390
|
2536 c = (getc(fd) << 24) + c; /* <prefixID> */
|
|
339
|
2537 }
|
|
|
2538
|
|
300
|
2539 idxs[idx] = c;
|
|
|
2540 c = 0;
|
|
|
2541 }
|
|
|
2542 else /* c == BY_INDEX */
|
|
|
2543 {
|
|
|
2544 /* <nodeidx> */
|
|
|
2545 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2546 if (n < 0 || n >= maxidx)
|
|
|
2547 return -2;
|
|
|
2548 idxs[idx] = n + SHARED_MASK;
|
|
|
2549 c = getc(fd); /* <xbyte> */
|
|
|
2550 }
|
|
|
2551 }
|
|
|
2552 byts[idx++] = c;
|
|
236
|
2553 }
|
|
|
2554
|
|
300
|
2555 /* Recursively read the children for non-shared siblings.
|
|
|
2556 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
2557 * remove SHARED_MASK) */
|
|
|
2558 for (i = 1; i <= len; ++i)
|
|
|
2559 if (byts[startidx + i] != 0)
|
|
|
2560 {
|
|
|
2561 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
2562 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
2563 else
|
|
|
2564 {
|
|
|
2565 idxs[startidx + i] = idx;
|
|
339
|
2566 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
2567 prefixtree, maxprefcondnr);
|
|
300
|
2568 if (idx < 0)
|
|
|
2569 break;
|
|
|
2570 }
|
|
|
2571 }
|
|
236
|
2572
|
|
300
|
2573 return idx;
|
|
236
|
2574 }
|
|
|
2575
|
|
|
2576 /*
|
|
|
2577 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
2578 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
2579 */
|
|
|
2580 char_u *
|
|
|
2581 did_set_spelllang(buf)
|
|
|
2582 buf_T *buf;
|
|
|
2583 {
|
|
|
2584 garray_T ga;
|
|
351
|
2585 char_u *splp;
|
|
236
|
2586 char_u *region;
|
|
416
|
2587 char_u region_cp[3];
|
|
355
|
2588 int filename;
|
|
236
|
2589 int region_mask;
|
|
|
2590 slang_T *lp;
|
|
|
2591 int c;
|
|
351
|
2592 char_u lang[MAXWLEN + 1];
|
|
323
|
2593 char_u spf_name[MAXPATHL];
|
|
351
|
2594 int len;
|
|
|
2595 char_u *p;
|
|
381
|
2596 int round;
|
|
385
|
2597 char_u *spf;
|
|
389
|
2598 char_u *use_region = NULL;
|
|
|
2599 int dont_use_region = FALSE;
|
|
236
|
2600
|
|
|
2601 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
2602 clear_midword(buf);
|
|
236
|
2603
|
|
351
|
2604 /* loop over comma separated language names. */
|
|
|
2605 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
2606 {
|
|
|
2607 /* Get one language name. */
|
|
|
2608 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
2609
|
|
240
|
2610 region = NULL;
|
|
351
|
2611 len = STRLEN(lang);
|
|
355
|
2612
|
|
|
2613 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
2614 * If there is a region name let "region" point to it and remove it
|
|
|
2615 * from the name. */
|
|
|
2616 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
2617 {
|
|
355
|
2618 filename = TRUE;
|
|
|
2619
|
|
416
|
2620 /* Locate a region and remove it from the file name. */
|
|
|
2621 p = vim_strchr(gettail(lang), '_');
|
|
|
2622 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
2623 && !ASCII_ISALPHA(p[3]))
|
|
|
2624 {
|
|
|
2625 vim_strncpy(region_cp, p + 1, 2);
|
|
|
2626 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
2627 len -= 3;
|
|
|
2628 region = region_cp;
|
|
|
2629 }
|
|
|
2630 else
|
|
|
2631 dont_use_region = TRUE;
|
|
|
2632
|
|
355
|
2633 /* Check if we loaded this language before. */
|
|
|
2634 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2635 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2636 break;
|
|
236
|
2637 }
|
|
355
|
2638 else
|
|
|
2639 {
|
|
|
2640 filename = FALSE;
|
|
|
2641 if (len > 3 && lang[len - 3] == '_')
|
|
|
2642 {
|
|
|
2643 region = lang + len - 2;
|
|
|
2644 len -= 3;
|
|
|
2645 lang[len] = NUL;
|
|
389
|
2646 }
|
|
|
2647 else
|
|
|
2648 dont_use_region = TRUE;
|
|
355
|
2649
|
|
|
2650 /* Check if we loaded this language before. */
|
|
|
2651 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2652 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
2653 break;
|
|
|
2654 }
|
|
236
|
2655
|
|
416
|
2656 if (region != NULL)
|
|
|
2657 {
|
|
|
2658 /* If the region differs from what was used before then don't
|
|
|
2659 * use it for 'spellfile'. */
|
|
|
2660 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
2661 dont_use_region = TRUE;
|
|
|
2662 use_region = region;
|
|
|
2663 }
|
|
|
2664
|
|
351
|
2665 /* If not found try loading the language now. */
|
|
236
|
2666 if (lp == NULL)
|
|
355
|
2667 {
|
|
|
2668 if (filename)
|
|
|
2669 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
2670 else
|
|
|
2671 spell_load_lang(lang);
|
|
|
2672 }
|
|
236
|
2673
|
|
307
|
2674 /*
|
|
351
|
2675 * Loop over the languages, there can be several files for "lang".
|
|
307
|
2676 */
|
|
|
2677 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
2678 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
2679 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
2680 {
|
|
316
|
2681 region_mask = REGION_ALL;
|
|
355
|
2682 if (!filename && region != NULL)
|
|
236
|
2683 {
|
|
307
|
2684 /* find region in sl_regions */
|
|
|
2685 c = find_region(lp->sl_regions, region);
|
|
|
2686 if (c == REGION_ALL)
|
|
|
2687 {
|
|
389
|
2688 if (lp->sl_add)
|
|
|
2689 {
|
|
|
2690 if (*lp->sl_regions != NUL)
|
|
|
2691 /* This addition file is for other regions. */
|
|
|
2692 region_mask = 0;
|
|
|
2693 }
|
|
|
2694 else
|
|
|
2695 /* This is probably an error. Give a warning and
|
|
|
2696 * accept the words anyway. */
|
|
351
|
2697 smsg((char_u *)
|
|
|
2698 _("Warning: region %s not supported"),
|
|
|
2699 region);
|
|
307
|
2700 }
|
|
|
2701 else
|
|
|
2702 region_mask = 1 << c;
|
|
236
|
2703 }
|
|
307
|
2704
|
|
389
|
2705 if (region_mask != 0)
|
|
307
|
2706 {
|
|
389
|
2707 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2708 {
|
|
|
2709 ga_clear(&ga);
|
|
|
2710 return e_outofmem;
|
|
|
2711 }
|
|
|
2712 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2713 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2714 ++ga.ga_len;
|
|
|
2715 use_midword(lp, buf);
|
|
307
|
2716 }
|
|
385
|
2717 }
|
|
|
2718 }
|
|
|
2719
|
|
|
2720 /* round 0: load int_wordlist, if possible.
|
|
|
2721 * round 1: load first name in 'spellfile'.
|
|
|
2722 * round 2: load second name in 'spellfile.
|
|
|
2723 * etc. */
|
|
|
2724 spf = curbuf->b_p_spf;
|
|
|
2725 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
2726 {
|
|
|
2727 if (round == 0)
|
|
|
2728 {
|
|
|
2729 /* Internal wordlist, if there is one. */
|
|
|
2730 if (int_wordlist == NULL)
|
|
381
|
2731 continue;
|
|
385
|
2732 int_wordlist_spl(spf_name);
|
|
381
|
2733 }
|
|
|
2734 else
|
|
|
2735 {
|
|
385
|
2736 /* One entry in 'spellfile'. */
|
|
|
2737 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
2738 STRCAT(spf_name, ".spl");
|
|
|
2739
|
|
|
2740 /* If it was already found above then skip it. */
|
|
|
2741 for (c = 0; c < ga.ga_len; ++c)
|
|
|
2742 if (fullpathcmp(spf_name,
|
|
|
2743 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
2744 FALSE) == FPC_SAME)
|
|
|
2745 break;
|
|
|
2746 if (c < ga.ga_len)
|
|
381
|
2747 continue;
|
|
|
2748 }
|
|
|
2749
|
|
351
|
2750 /* Check if it was loaded already. */
|
|
323
|
2751 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2752 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2753 break;
|
|
|
2754 if (lp == NULL)
|
|
|
2755 {
|
|
351
|
2756 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
2757 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
2758 * use an arbitrary name. */
|
|
|
2759 if (round == 0)
|
|
|
2760 STRCPY(lang, "internal wordlist");
|
|
|
2761 else
|
|
|
2762 {
|
|
|
2763 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
2764 p = vim_strchr(lang, '.');
|
|
|
2765 if (p != NULL)
|
|
|
2766 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
2767 }
|
|
351
|
2768 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
2769 }
|
|
|
2770 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
2771 {
|
|
389
|
2772 region_mask = REGION_ALL;
|
|
|
2773 if (use_region != NULL && !dont_use_region)
|
|
|
2774 {
|
|
|
2775 /* find region in sl_regions */
|
|
|
2776 c = find_region(lp->sl_regions, use_region);
|
|
|
2777 if (c != REGION_ALL)
|
|
|
2778 region_mask = 1 << c;
|
|
|
2779 else if (*lp->sl_regions != NUL)
|
|
|
2780 /* This spell file is for other regions. */
|
|
|
2781 region_mask = 0;
|
|
|
2782 }
|
|
|
2783
|
|
|
2784 if (region_mask != 0)
|
|
|
2785 {
|
|
|
2786 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2787 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2788 ++ga.ga_len;
|
|
|
2789 use_midword(lp, buf);
|
|
|
2790 }
|
|
323
|
2791 }
|
|
|
2792 }
|
|
|
2793
|
|
236
|
2794 /* Add a NULL entry to mark the end of the list. */
|
|
|
2795 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2796 {
|
|
|
2797 ga_clear(&ga);
|
|
|
2798 return e_outofmem;
|
|
|
2799 }
|
|
|
2800 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
2801 ++ga.ga_len;
|
|
|
2802
|
|
|
2803 /* Everything is fine, store the new b_langp value. */
|
|
|
2804 ga_clear(&buf->b_langp);
|
|
|
2805 buf->b_langp = ga;
|
|
|
2806
|
|
|
2807 return NULL;
|
|
|
2808 }
|
|
|
2809
|
|
|
2810 /*
|
|
376
|
2811 * Clear the midword characters for buffer "buf".
|
|
|
2812 */
|
|
|
2813 static void
|
|
|
2814 clear_midword(buf)
|
|
|
2815 buf_T *buf;
|
|
|
2816 {
|
|
|
2817 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
2818 #ifdef FEAT_MBYTE
|
|
|
2819 vim_free(buf->b_spell_ismw_mb);
|
|
|
2820 buf->b_spell_ismw_mb = NULL;
|
|
|
2821 #endif
|
|
|
2822 }
|
|
|
2823
|
|
|
2824 /*
|
|
|
2825 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
2826 * They add up to any currently used midword characters.
|
|
|
2827 */
|
|
|
2828 static void
|
|
|
2829 use_midword(lp, buf)
|
|
|
2830 slang_T *lp;
|
|
|
2831 buf_T *buf;
|
|
|
2832 {
|
|
|
2833 char_u *p;
|
|
|
2834
|
|
389
|
2835 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
2836 return;
|
|
|
2837
|
|
376
|
2838 for (p = lp->sl_midword; *p != NUL; )
|
|
|
2839 #ifdef FEAT_MBYTE
|
|
|
2840 if (has_mbyte)
|
|
|
2841 {
|
|
|
2842 int c, l, n;
|
|
|
2843 char_u *bp;
|
|
|
2844
|
|
|
2845 c = mb_ptr2char(p);
|
|
474
|
2846 l = (*mb_ptr2len)(p);
|
|
|
2847 if (c < 256 && l <= 2)
|
|
376
|
2848 buf->b_spell_ismw[c] = TRUE;
|
|
|
2849 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
2850 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
2851 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
2852 else
|
|
|
2853 {
|
|
|
2854 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
2855 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
2856 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
2857 if (bp != NULL)
|
|
|
2858 {
|
|
|
2859 vim_free(buf->b_spell_ismw_mb);
|
|
|
2860 buf->b_spell_ismw_mb = bp;
|
|
|
2861 vim_strncpy(bp + n, p, l);
|
|
|
2862 }
|
|
|
2863 }
|
|
|
2864 p += l;
|
|
|
2865 }
|
|
|
2866 else
|
|
|
2867 #endif
|
|
|
2868 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
2869 }
|
|
|
2870
|
|
|
2871 /*
|
|
236
|
2872 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
2873 * Each region is simply stored as the two characters of it's name.
|
|
381
|
2874 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
2875 */
|
|
|
2876 static int
|
|
|
2877 find_region(rp, region)
|
|
|
2878 char_u *rp;
|
|
|
2879 char_u *region;
|
|
|
2880 {
|
|
|
2881 int i;
|
|
|
2882
|
|
|
2883 for (i = 0; ; i += 2)
|
|
|
2884 {
|
|
|
2885 if (rp[i] == NUL)
|
|
|
2886 return REGION_ALL;
|
|
|
2887 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
2888 break;
|
|
|
2889 }
|
|
|
2890 return i / 2;
|
|
|
2891 }
|
|
|
2892
|
|
|
2893 /*
|
|
323
|
2894 * Return case type of word:
|
|
236
|
2895 * w word 0
|
|
300
|
2896 * Word WF_ONECAP
|
|
|
2897 * W WORD WF_ALLCAP
|
|
|
2898 * WoRd wOrd WF_KEEPCAP
|
|
236
|
2899 */
|
|
|
2900 static int
|
|
|
2901 captype(word, end)
|
|
|
2902 char_u *word;
|
|
323
|
2903 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
2904 {
|
|
|
2905 char_u *p;
|
|
|
2906 int c;
|
|
|
2907 int firstcap;
|
|
|
2908 int allcap;
|
|
|
2909 int past_second = FALSE; /* past second word char */
|
|
|
2910
|
|
|
2911 /* find first letter */
|
|
376
|
2912 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
2913 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
2914 return 0; /* only non-word characters, illegal word */
|
|
|
2915 #ifdef FEAT_MBYTE
|
|
310
|
2916 if (has_mbyte)
|
|
|
2917 c = mb_ptr2char_adv(&p);
|
|
|
2918 else
|
|
236
|
2919 #endif
|
|
310
|
2920 c = *p++;
|
|
324
|
2921 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
2922
|
|
|
2923 /*
|
|
|
2924 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
2925 * But a word with an upper char only at start is a ONECAP.
|
|
|
2926 */
|
|
323
|
2927 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
2928 if (spell_iswordp_nmw(p))
|
|
236
|
2929 {
|
|
455
|
2930 c = PTR2CHAR(p);
|
|
324
|
2931 if (!SPELL_ISUPPER(c))
|
|
236
|
2932 {
|
|
|
2933 /* UUl -> KEEPCAP */
|
|
|
2934 if (past_second && allcap)
|
|
300
|
2935 return WF_KEEPCAP;
|
|
236
|
2936 allcap = FALSE;
|
|
|
2937 }
|
|
|
2938 else if (!allcap)
|
|
|
2939 /* UlU -> KEEPCAP */
|
|
300
|
2940 return WF_KEEPCAP;
|
|
236
|
2941 past_second = TRUE;
|
|
|
2942 }
|
|
|
2943
|
|
|
2944 if (allcap)
|
|
300
|
2945 return WF_ALLCAP;
|
|
236
|
2946 if (firstcap)
|
|
300
|
2947 return WF_ONECAP;
|
|
236
|
2948 return 0;
|
|
|
2949 }
|
|
|
2950
|
|
474
|
2951 /*
|
|
|
2952 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
2953 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
2954 * Add ALLCAP for "WOrD".
|
|
|
2955 */
|
|
|
2956 static int
|
|
|
2957 badword_captype(word, end)
|
|
|
2958 char_u *word;
|
|
|
2959 char_u *end;
|
|
|
2960 {
|
|
|
2961 int flags = captype(word, end);
|
|
|
2962 int l, u;
|
|
|
2963 int first;
|
|
|
2964 char_u *p;
|
|
|
2965
|
|
|
2966 if (flags & WF_KEEPCAP)
|
|
|
2967 {
|
|
|
2968 /* Count the number of UPPER and lower case letters. */
|
|
|
2969 l = u = 0;
|
|
|
2970 first = FALSE;
|
|
|
2971 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
2972 {
|
|
|
2973 if (SPELL_ISUPPER(PTR2CHAR(p)))
|
|
|
2974 {
|
|
|
2975 ++u;
|
|
|
2976 if (p == word)
|
|
|
2977 first = TRUE;
|
|
|
2978 }
|
|
|
2979 else
|
|
|
2980 ++l;
|
|
|
2981 }
|
|
|
2982
|
|
|
2983 /* If there are more UPPER than lower case letters suggest an
|
|
|
2984 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
2985 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
2986 * require three upper case letters. */
|
|
|
2987 if (u > l && u > 2)
|
|
|
2988 flags |= WF_ALLCAP;
|
|
|
2989 else if (first)
|
|
|
2990 flags |= WF_ONECAP;
|
|
|
2991 }
|
|
|
2992 return flags;
|
|
|
2993 }
|
|
|
2994
|
|
355
|
2995 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
2996 /*
|
|
|
2997 * Free all languages.
|
|
|
2998 */
|
|
|
2999 void
|
|
|
3000 spell_free_all()
|
|
|
3001 {
|
|
|
3002 slang_T *lp;
|
|
|
3003 buf_T *buf;
|
|
385
|
3004 char_u fname[MAXPATHL];
|
|
355
|
3005
|
|
|
3006 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3007 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3008 ga_clear(&buf->b_langp);
|
|
|
3009
|
|
|
3010 while (first_lang != NULL)
|
|
|
3011 {
|
|
|
3012 lp = first_lang;
|
|
|
3013 first_lang = lp->sl_next;
|
|
|
3014 slang_free(lp);
|
|
|
3015 }
|
|
366
|
3016
|
|
385
|
3017 if (int_wordlist != NULL)
|
|
|
3018 {
|
|
|
3019 /* Delete the internal wordlist and its .spl file */
|
|
|
3020 mch_remove(int_wordlist);
|
|
|
3021 int_wordlist_spl(fname);
|
|
|
3022 mch_remove(fname);
|
|
|
3023 vim_free(int_wordlist);
|
|
|
3024 int_wordlist = NULL;
|
|
381
|
3025 }
|
|
|
3026
|
|
366
|
3027 init_spell_chartab();
|
|
355
|
3028 }
|
|
|
3029 # endif
|
|
|
3030
|
|
236
|
3031 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
3032 /*
|
|
|
3033 * Clear all spelling tables and reload them.
|
|
307
|
3034 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
3035 */
|
|
|
3036 void
|
|
|
3037 spell_reload()
|
|
|
3038 {
|
|
|
3039 buf_T *buf;
|
|
316
|
3040 win_T *wp;
|
|
236
|
3041
|
|
358
|
3042 /* Initialize the table for spell_iswordp(). */
|
|
236
|
3043 init_spell_chartab();
|
|
|
3044
|
|
|
3045 /* Unload all allocated memory. */
|
|
355
|
3046 spell_free_all();
|
|
236
|
3047
|
|
|
3048 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3049 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3050 {
|
|
316
|
3051 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
3052 * window for this buffer in which 'spell' is set. */
|
|
236
|
3053 if (*buf->b_p_spl != NUL)
|
|
316
|
3054 {
|
|
|
3055 FOR_ALL_WINDOWS(wp)
|
|
|
3056 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
3057 {
|
|
|
3058 (void)did_set_spelllang(buf);
|
|
|
3059 # ifdef FEAT_WINDOWS
|
|
|
3060 break;
|
|
|
3061 # endif
|
|
|
3062 }
|
|
|
3063 }
|
|
236
|
3064 }
|
|
|
3065 }
|
|
|
3066 # endif
|
|
|
3067
|
|
310
|
3068 /*
|
|
|
3069 * Reload the spell file "fname" if it's loaded.
|
|
|
3070 */
|
|
|
3071 static void
|
|
323
|
3072 spell_reload_one(fname, added_word)
|
|
310
|
3073 char_u *fname;
|
|
323
|
3074 int added_word; /* invoked through "zg" */
|
|
310
|
3075 {
|
|
|
3076 slang_T *lp;
|
|
323
|
3077 int didit = FALSE;
|
|
310
|
3078
|
|
|
3079 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3080 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3081 {
|
|
|
3082 slang_clear(lp);
|
|
323
|
3083 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3084 redraw_all_later(NOT_VALID);
|
|
323
|
3085 didit = TRUE;
|
|
310
|
3086 }
|
|
323
|
3087
|
|
|
3088 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3089 * 'spelllang' to get it loaded. */
|
|
|
3090 if (added_word && !didit)
|
|
|
3091 did_set_spelllang(curbuf);
|
|
310
|
3092 }
|
|
|
3093
|
|
|
3094
|
|
236
|
3095 /*
|
|
|
3096 * Functions for ":mkspell".
|
|
|
3097 */
|
|
|
3098
|
|
300
|
3099 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3100 and .dic file. */
|
|
|
3101 /*
|
|
|
3102 * Main structure to store the contents of a ".aff" file.
|
|
|
3103 */
|
|
|
3104 typedef struct afffile_S
|
|
|
3105 {
|
|
|
3106 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
310
|
3107 int af_rar; /* RAR ID for rare word */
|
|
|
3108 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
3109 int af_bad; /* BAD ID for banned word */
|
|
339
|
3110 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3111 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3112 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
3113 } afffile_T;
|
|
|
3114
|
|
|
3115 typedef struct affentry_S affentry_T;
|
|
|
3116 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3117 struct affentry_S
|
|
|
3118 {
|
|
|
3119 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3120 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3121 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3122 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3123 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
366
|
3124 int ae_rare; /* rare affix */
|
|
300
|
3125 };
|
|
|
3126
|
|
455
|
3127 #define AH_KEY_LEN 10
|
|
|
3128
|
|
300
|
3129 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3130 typedef struct affheader_S
|
|
|
3131 {
|
|
455
|
3132 /* key for hashtable == name of affix entry */
|
|
|
3133 #ifdef FEAT_MBYTE
|
|
|
3134 char_u ah_key[AH_KEY_LEN]; /* multi-byte char plus NUL */
|
|
|
3135 #else
|
|
|
3136 char_u ah_key[2]; /* one byte char plus NUL */
|
|
|
3137 #endif
|
|
339
|
3138 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
3139 int ah_combine; /* suffix may combine with prefix */
|
|
|
3140 affentry_T *ah_first; /* first affix entry */
|
|
|
3141 } affheader_T;
|
|
|
3142
|
|
|
3143 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
3144
|
|
|
3145 /*
|
|
|
3146 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
3147 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
3148 * once after ":mkspell" is done.
|
|
300
|
3149 */
|
|
|
3150 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
3151 typedef struct sblock_S sblock_T;
|
|
|
3152 struct sblock_S
|
|
|
3153 {
|
|
|
3154 sblock_T *sb_next; /* next block in list */
|
|
|
3155 int sb_used; /* nr of bytes already in use */
|
|
|
3156 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
3157 };
|
|
|
3158
|
|
|
3159 /*
|
|
300
|
3160 * A node in the tree.
|
|
236
|
3161 */
|
|
300
|
3162 typedef struct wordnode_S wordnode_T;
|
|
|
3163 struct wordnode_S
|
|
236
|
3164 {
|
|
346
|
3165 union /* shared to save space */
|
|
|
3166 {
|
|
469
|
3167 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
3168 int index; /* index in written nodes (valid after first
|
|
|
3169 round) */
|
|
|
3170 } wn_u1;
|
|
|
3171 union /* shared to save space */
|
|
|
3172 {
|
|
|
3173 wordnode_T *next; /* next node with same hash key */
|
|
|
3174 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
3175 } wn_u2;
|
|
300
|
3176 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
3177 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
3178 always sorted) */
|
|
474
|
3179 int wn_refs; /* Nr. of references to this node. Only
|
|
|
3180 relevant for first node in a list of
|
|
|
3181 siblings, in following siblings it is
|
|
|
3182 always one. */
|
|
300
|
3183 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
469
|
3184 char_u wn_prefixID; /* when "wn_byte" is NUL: supported/required
|
|
|
3185 prefix ID or 0 */
|
|
|
3186 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
3187 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
3188 PREFIXTREE it's the prefcondnr */
|
|
469
|
3189 #ifdef SPELL_PRINTTREE
|
|
|
3190 int wn_nr; /* sequence nr for printing */
|
|
|
3191 #endif
|
|
236
|
3192 };
|
|
|
3193
|
|
390
|
3194 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
3195
|
|
300
|
3196 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
3197
|
|
300
|
3198 /*
|
|
|
3199 * Info used while reading the spell files.
|
|
|
3200 */
|
|
|
3201 typedef struct spellinfo_S
|
|
249
|
3202 {
|
|
300
|
3203 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
3204 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
3205 int si_fold_added; /* nr of words added since compressing */
|
|
|
3206
|
|
300
|
3207 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
3208 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
3209 int si_keep_added; /* nr of words added since compressing */
|
|
|
3210
|
|
339
|
3211 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
3212
|
|
300
|
3213 sblock_T *si_blocks; /* memory blocks used */
|
|
474
|
3214 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
3215 compression, linked by "wn_child" field. */
|
|
|
3216 #ifdef SPELL_PRINTTREE
|
|
|
3217 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
3218 #endif
|
|
|
3219
|
|
|
3220
|
|
300
|
3221 int si_ascii; /* handling only ASCII words */
|
|
310
|
3222 int si_add; /* addition file */
|
|
351
|
3223 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
3224 int si_region; /* region mask */
|
|
|
3225 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
3226 int si_memtot; /* runtime memory used */
|
|
310
|
3227 int si_verbose; /* verbose messages */
|
|
474
|
3228 int si_msg_count; /* number of words added since last message */
|
|
316
|
3229 int si_region_count; /* number of regions supported (1 when there
|
|
|
3230 are no regions) */
|
|
|
3231 char_u si_region_name[16]; /* region names (if count > 1) */
|
|
323
|
3232
|
|
|
3233 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
3234 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
3235 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
3236 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
3237 int si_followup; /* soundsalike: ? */
|
|
|
3238 int si_collapse; /* soundsalike: ? */
|
|
|
3239 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
3240 garray_T si_map; /* MAP info concatenated */
|
|
366
|
3241 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
339
|
3242 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
3243 * prefixes, each stored as a string */
|
|
|
3244 int si_newID; /* current value for ah_newID */
|
|
300
|
3245 } spellinfo_T;
|
|
249
|
3246
|
|
474
|
3247 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
339
|
3248 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
3249 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
3250 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
3251 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
3252 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
3253 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
|
3254 static char_u *get_pfxlist __ARGS((spellinfo_T *spin, afffile_T *affile, char_u *afflist));
|
|
|
3255 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));
|
|
|
3256 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
3257 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
3258 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
3259 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
3260 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
|
3261 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist));
|
|
|
3262 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int prefixID));
|
|
|
3263 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
3264 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
3265 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
3266 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
3267 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
3268 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
474
|
3269 static void write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
3270 static void clear_node __ARGS((wordnode_T *node));
|
|
|
3271 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
3272 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3273 static void init_spellfile __ARGS((void));
|
|
236
|
3274
|
|
455
|
3275 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
3276 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
3277 * Use a negative number with the lower 8 bits zero. */
|
|
|
3278 #define PFX_FLAGS -256
|
|
390
|
3279
|
|
469
|
3280 #ifdef SPELL_PRINTTREE
|
|
|
3281 /*
|
|
|
3282 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
3283 * readable format, so that we can see what happens when adding a word and/or
|
|
|
3284 * compressing the tree.
|
|
|
3285 * Based on code from Olaf Seibert.
|
|
|
3286 */
|
|
|
3287 #define PRINTLINESIZE 1000
|
|
|
3288 #define PRINTWIDTH 6
|
|
|
3289
|
|
|
3290 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
3291 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
3292
|
|
|
3293 static char line1[PRINTLINESIZE];
|
|
|
3294 static char line2[PRINTLINESIZE];
|
|
|
3295 static char line3[PRINTLINESIZE];
|
|
|
3296
|
|
|
3297 static void
|
|
|
3298 spell_clear_flags(wordnode_T *node)
|
|
|
3299 {
|
|
|
3300 wordnode_T *np;
|
|
|
3301
|
|
|
3302 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
3303 {
|
|
|
3304 np->wn_u1.index = FALSE;
|
|
|
3305 spell_clear_flags(np->wn_child);
|
|
|
3306 }
|
|
|
3307 }
|
|
|
3308
|
|
|
3309 static void
|
|
|
3310 spell_print_node(wordnode_T *node, int depth)
|
|
|
3311 {
|
|
|
3312 if (node->wn_u1.index)
|
|
|
3313 {
|
|
|
3314 /* Done this node before, print the reference. */
|
|
|
3315 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
3316 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
3317 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3318 msg(line1);
|
|
|
3319 msg(line2);
|
|
|
3320 msg(line3);
|
|
|
3321 }
|
|
|
3322 else
|
|
|
3323 {
|
|
|
3324 node->wn_u1.index = TRUE;
|
|
|
3325
|
|
|
3326 if (node->wn_byte != NUL)
|
|
|
3327 {
|
|
|
3328 if (node->wn_child != NULL)
|
|
|
3329 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
3330 else
|
|
|
3331 /* Cannot happen? */
|
|
|
3332 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
3333 }
|
|
|
3334 else
|
|
|
3335 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
3336
|
|
|
3337 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
3338
|
|
|
3339 if (node->wn_sibling != NULL)
|
|
|
3340 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
3341 else
|
|
|
3342 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3343
|
|
|
3344 if (node->wn_byte == NUL)
|
|
|
3345 {
|
|
|
3346 msg(line1);
|
|
|
3347 msg(line2);
|
|
|
3348 msg(line3);
|
|
|
3349 }
|
|
|
3350
|
|
|
3351 /* do the children */
|
|
|
3352 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
3353 spell_print_node(node->wn_child, depth + 1);
|
|
|
3354
|
|
|
3355 /* do the siblings */
|
|
|
3356 if (node->wn_sibling != NULL)
|
|
|
3357 {
|
|
|
3358 /* get rid of all parent details except | */
|
|
|
3359 STRCPY(line1, line3);
|
|
|
3360 STRCPY(line2, line3);
|
|
|
3361 spell_print_node(node->wn_sibling, depth);
|
|
|
3362 }
|
|
|
3363 }
|
|
|
3364 }
|
|
|
3365
|
|
|
3366 static void
|
|
|
3367 spell_print_tree(wordnode_T *root)
|
|
|
3368 {
|
|
|
3369 if (root != NULL)
|
|
|
3370 {
|
|
|
3371 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
3372 * done. */
|
|
|
3373 spell_clear_flags(root);
|
|
|
3374
|
|
|
3375 /* Recursively print the tree. */
|
|
|
3376 spell_print_node(root, 0);
|
|
|
3377 }
|
|
|
3378 }
|
|
|
3379 #endif /* SPELL_PRINTTREE */
|
|
|
3380
|
|
236
|
3381 /*
|
|
323
|
3382 * Read the affix file "fname".
|
|
316
|
3383 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
3384 */
|
|
|
3385 static afffile_T *
|
|
474
|
3386 spell_read_aff(spin, fname)
|
|
|
3387 spellinfo_T *spin;
|
|
236
|
3388 char_u *fname;
|
|
|
3389 {
|
|
|
3390 FILE *fd;
|
|
|
3391 afffile_T *aff;
|
|
|
3392 char_u rline[MAXLINELEN];
|
|
|
3393 char_u *line;
|
|
|
3394 char_u *pc = NULL;
|
|
334
|
3395 #define MAXITEMCNT 7
|
|
|
3396 char_u *(items[MAXITEMCNT]);
|
|
236
|
3397 int itemcnt;
|
|
|
3398 char_u *p;
|
|
|
3399 int lnum = 0;
|
|
|
3400 affheader_T *cur_aff = NULL;
|
|
|
3401 int aff_todo = 0;
|
|
|
3402 hashtab_T *tp;
|
|
255
|
3403 char_u *low = NULL;
|
|
|
3404 char_u *fol = NULL;
|
|
|
3405 char_u *upp = NULL;
|
|
307
|
3406 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
3407 int do_rep;
|
|
|
3408 int do_sal;
|
|
|
3409 int do_map;
|
|
366
|
3410 int do_midword;
|
|
375
|
3411 int do_sofo;
|
|
323
|
3412 int found_map = FALSE;
|
|
324
|
3413 hashitem_T *hi;
|
|
455
|
3414 int l;
|
|
236
|
3415
|
|
300
|
3416 /*
|
|
|
3417 * Open the file.
|
|
|
3418 */
|
|
310
|
3419 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3420 if (fd == NULL)
|
|
|
3421 {
|
|
|
3422 EMSG2(_(e_notopen), fname);
|
|
|
3423 return NULL;
|
|
|
3424 }
|
|
|
3425
|
|
310
|
3426 if (spin->si_verbose || p_verbose > 2)
|
|
|
3427 {
|
|
|
3428 if (!spin->si_verbose)
|
|
|
3429 verbose_enter();
|
|
366
|
3430 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
3431 out_flush();
|
|
|
3432 if (!spin->si_verbose)
|
|
|
3433 verbose_leave();
|
|
|
3434 }
|
|
236
|
3435
|
|
323
|
3436 /* Only do REP lines when not done in another .aff file already. */
|
|
|
3437 do_rep = spin->si_rep.ga_len == 0;
|
|
|
3438
|
|
|
3439 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
3440 do_sal = spin->si_sal.ga_len == 0;
|
|
|
3441
|
|
|
3442 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
3443 do_map = spin->si_map.ga_len == 0;
|
|
|
3444
|
|
366
|
3445 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
3446 do_midword = spin->si_midword == NULL;
|
|
|
3447
|
|
375
|
3448 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
3449 do_sofo = spin->si_sofofr == NULL;
|
|
|
3450
|
|
300
|
3451 /*
|
|
|
3452 * Allocate and init the afffile_T structure.
|
|
|
3453 */
|
|
474
|
3454 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
3455 if (aff == NULL)
|
|
|
3456 return NULL;
|
|
|
3457 hash_init(&aff->af_pref);
|
|
|
3458 hash_init(&aff->af_suff);
|
|
|
3459
|
|
|
3460 /*
|
|
|
3461 * Read all the lines in the file one by one.
|
|
|
3462 */
|
|
255
|
3463 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
3464 {
|
|
255
|
3465 line_breakcheck();
|
|
236
|
3466 ++lnum;
|
|
|
3467
|
|
|
3468 /* Skip comment lines. */
|
|
|
3469 if (*rline == '#')
|
|
|
3470 continue;
|
|
|
3471
|
|
|
3472 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
3473 vim_free(pc);
|
|
310
|
3474 #ifdef FEAT_MBYTE
|
|
300
|
3475 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3476 {
|
|
300
|
3477 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
3478 if (pc == NULL)
|
|
|
3479 {
|
|
|
3480 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3481 fname, lnum, rline);
|
|
|
3482 continue;
|
|
|
3483 }
|
|
236
|
3484 line = pc;
|
|
|
3485 }
|
|
|
3486 else
|
|
310
|
3487 #endif
|
|
236
|
3488 {
|
|
|
3489 pc = NULL;
|
|
|
3490 line = rline;
|
|
|
3491 }
|
|
|
3492
|
|
|
3493 /* Split the line up in white separated items. Put a NUL after each
|
|
|
3494 * item. */
|
|
|
3495 itemcnt = 0;
|
|
|
3496 for (p = line; ; )
|
|
|
3497 {
|
|
|
3498 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
3499 ++p;
|
|
|
3500 if (*p == NUL)
|
|
|
3501 break;
|
|
334
|
3502 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
3503 break;
|
|
236
|
3504 items[itemcnt++] = p;
|
|
300
|
3505 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
3506 ++p;
|
|
|
3507 if (*p == NUL)
|
|
|
3508 break;
|
|
|
3509 *p++ = NUL;
|
|
|
3510 }
|
|
|
3511
|
|
|
3512 /* Handle non-empty lines. */
|
|
|
3513 if (itemcnt > 0)
|
|
|
3514 {
|
|
|
3515 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
3516 && aff->af_enc == NULL)
|
|
|
3517 {
|
|
310
|
3518 #ifdef FEAT_MBYTE
|
|
300
|
3519 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
3520 aff->af_enc = enc_canonize(items[1]);
|
|
|
3521 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
3522 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
3523 p_enc) == FAIL)
|
|
|
3524 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
3525 fname, aff->af_enc, p_enc);
|
|
375
|
3526 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3527 #else
|
|
|
3528 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3529 #endif
|
|
236
|
3530 }
|
|
366
|
3531 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
3532 {
|
|
|
3533 if (do_midword)
|
|
|
3534 spin->si_midword = vim_strsave(items[1]);
|
|
|
3535 }
|
|
302
|
3536 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
3537 {
|
|
323
|
3538 /* ignored, we always split */
|
|
302
|
3539 }
|
|
323
|
3540 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
3541 {
|
|
323
|
3542 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
3543 }
|
|
307
|
3544 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
3545 && aff->af_rar == 0)
|
|
|
3546 {
|
|
|
3547 aff->af_rar = items[1][0];
|
|
|
3548 if (items[1][1] != NUL)
|
|
|
3549 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3550 }
|
|
310
|
3551 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
3552 && aff->af_kep == 0)
|
|
307
|
3553 {
|
|
310
|
3554 aff->af_kep = items[1][0];
|
|
307
|
3555 if (items[1][1] != NUL)
|
|
|
3556 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3557 }
|
|
346
|
3558 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
3559 && aff->af_bad == 0)
|
|
|
3560 {
|
|
|
3561 aff->af_bad = items[1][0];
|
|
|
3562 if (items[1][1] != NUL)
|
|
|
3563 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3564 }
|
|
339
|
3565 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
3566 {
|
|
|
3567 aff->af_pfxpostpone = TRUE;
|
|
|
3568 }
|
|
236
|
3569 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3570 || STRCMP(items[0], "SFX") == 0)
|
|
|
3571 && aff_todo == 0
|
|
334
|
3572 && itemcnt >= 4)
|
|
236
|
3573 {
|
|
334
|
3574 /* Myspell allows extra text after the item, but that might
|
|
|
3575 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
3576 if (itemcnt > 4 && *items[4] != '#')
|
|
|
3577 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
3578 fname, lnum, items[4]);
|
|
|
3579
|
|
236
|
3580 /* New affix letter. */
|
|
474
|
3581 cur_aff = (affheader_T *)getroom(spin,
|
|
447
|
3582 sizeof(affheader_T), TRUE);
|
|
236
|
3583 if (cur_aff == NULL)
|
|
|
3584 break;
|
|
455
|
3585 #ifdef FEAT_MBYTE
|
|
|
3586 if (has_mbyte)
|
|
|
3587 {
|
|
474
|
3588 l = (*mb_ptr2len)(items[1]);
|
|
455
|
3589 if (l >= AH_KEY_LEN)
|
|
|
3590 l = 1; /* too long, must be an overlong sequence */
|
|
|
3591 else
|
|
|
3592 mch_memmove(cur_aff->ah_key, items[1], l);
|
|
|
3593 }
|
|
|
3594 else
|
|
|
3595 #endif
|
|
|
3596 {
|
|
|
3597 *cur_aff->ah_key = *items[1];
|
|
|
3598 l = 1;
|
|
|
3599 }
|
|
|
3600 cur_aff->ah_key[l] = NUL;
|
|
|
3601 if (items[1][l] != NUL)
|
|
307
|
3602 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
3603 if (*items[2] == 'Y')
|
|
|
3604 cur_aff->ah_combine = TRUE;
|
|
300
|
3605 else if (*items[2] != 'N')
|
|
236
|
3606 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
3607 fname, lnum, items[2]);
|
|
339
|
3608
|
|
236
|
3609 if (*items[0] == 'P')
|
|
339
|
3610 {
|
|
236
|
3611 tp = &aff->af_pref;
|
|
339
|
3612 /* Use a new number in the .spl file later, to be able to
|
|
|
3613 * handle multiple .aff files. */
|
|
|
3614 if (aff->af_pfxpostpone)
|
|
344
|
3615 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
3616 }
|
|
236
|
3617 else
|
|
|
3618 tp = &aff->af_suff;
|
|
300
|
3619 aff_todo = atoi((char *)items[3]);
|
|
324
|
3620 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
3621 if (!HASHITEM_EMPTY(hi))
|
|
300
|
3622 {
|
|
236
|
3623 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
3624 fname, lnum, items[1]);
|
|
300
|
3625 aff_todo = 0;
|
|
|
3626 }
|
|
236
|
3627 else
|
|
|
3628 hash_add(tp, cur_aff->ah_key);
|
|
|
3629 }
|
|
|
3630 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3631 || STRCMP(items[0], "SFX") == 0)
|
|
|
3632 && aff_todo > 0
|
|
|
3633 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
3634 && itemcnt >= 5)
|
|
236
|
3635 {
|
|
|
3636 affentry_T *aff_entry;
|
|
366
|
3637 int rare = FALSE;
|
|
455
|
3638 int upper = FALSE;
|
|
366
|
3639 int lasti = 5;
|
|
|
3640
|
|
|
3641 /* Check for "rare" after the other info. */
|
|
|
3642 if (itemcnt > 5 && STRICMP(items[5], "rare") == 0)
|
|
|
3643 {
|
|
|
3644 rare = TRUE;
|
|
|
3645 lasti = 6;
|
|
|
3646 }
|
|
236
|
3647
|
|
334
|
3648 /* Myspell allows extra text after the item, but that might
|
|
|
3649 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
3650 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
3651 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
366
|
3652 fname, lnum, items[lasti]);
|
|
334
|
3653
|
|
236
|
3654 /* New item for an affix letter. */
|
|
|
3655 --aff_todo;
|
|
474
|
3656 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
3657 sizeof(affentry_T), TRUE);
|
|
236
|
3658 if (aff_entry == NULL)
|
|
|
3659 break;
|
|
366
|
3660 aff_entry->ae_rare = rare;
|
|
240
|
3661
|
|
236
|
3662 if (STRCMP(items[2], "0") != 0)
|
|
474
|
3663 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
3664 if (STRCMP(items[3], "0") != 0)
|
|
474
|
3665 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
3666
|
|
300
|
3667 /* Don't use an affix entry with non-ASCII characters when
|
|
|
3668 * "spin->si_ascii" is TRUE. */
|
|
|
3669 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
3670 || has_non_ascii(aff_entry->ae_add)))
|
|
|
3671 {
|
|
|
3672 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
3673 cur_aff->ah_first = aff_entry;
|
|
300
|
3674
|
|
|
3675 if (STRCMP(items[4], ".") != 0)
|
|
|
3676 {
|
|
|
3677 char_u buf[MAXLINELEN];
|
|
|
3678
|
|
474
|
3679 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
3680 if (*items[0] == 'P')
|
|
|
3681 sprintf((char *)buf, "^%s", items[4]);
|
|
|
3682 else
|
|
|
3683 sprintf((char *)buf, "%s$", items[4]);
|
|
|
3684 aff_entry->ae_prog = vim_regcomp(buf,
|
|
|
3685 RE_MAGIC + RE_STRING);
|
|
|
3686 }
|
|
339
|
3687
|
|
|
3688 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
3689 * for the condition. Use an existing one if possible. */
|
|
455
|
3690 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
3691 {
|
|
455
|
3692 /* When the chop string is one lower-case letter and
|
|
|
3693 * the add string ends in the upper-case letter we set
|
|
|
3694 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
3695 * letters from "ae_add". The condition must either
|
|
|
3696 * be empty or start with the same letter. */
|
|
|
3697 if (aff_entry->ae_chop != NULL
|
|
|
3698 && aff_entry->ae_add != NULL
|
|
|
3699 #ifdef FEAT_MBYTE
|
|
474
|
3700 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
3701 aff_entry->ae_chop)] == NUL
|
|
|
3702 #else
|
|
|
3703 && aff_entry->ae_chop[1] == NUL
|
|
|
3704 #endif
|
|
|
3705 )
|
|
339
|
3706 {
|
|
455
|
3707 int c, c_up;
|
|
|
3708
|
|
|
3709 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
3710 c_up = SPELL_TOUPPER(c);
|
|
|
3711 if (c_up != c
|
|
|
3712 && (aff_entry->ae_cond == NULL
|
|
|
3713 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
3714 {
|
|
|
3715 p = aff_entry->ae_add
|
|
|
3716 + STRLEN(aff_entry->ae_add);
|
|
|
3717 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
3718 if (PTR2CHAR(p) == c_up)
|
|
|
3719 {
|
|
|
3720 upper = TRUE;
|
|
|
3721 aff_entry->ae_chop = NULL;
|
|
|
3722 *p = NUL;
|
|
|
3723
|
|
|
3724 /* The condition is matched with the
|
|
|
3725 * actual word, thus must check for the
|
|
|
3726 * upper-case letter. */
|
|
|
3727 if (aff_entry->ae_cond != NULL)
|
|
|
3728 {
|
|
|
3729 char_u buf[MAXLINELEN];
|
|
|
3730 #ifdef FEAT_MBYTE
|
|
|
3731 if (has_mbyte)
|
|
|
3732 {
|
|
|
3733 onecap_copy(items[4], buf, TRUE);
|
|
|
3734 aff_entry->ae_cond = getroom_save(
|
|
474
|
3735 spin, buf);
|
|
455
|
3736 }
|
|
|
3737 else
|
|
|
3738 #endif
|
|
|
3739 *aff_entry->ae_cond = c_up;
|
|
|
3740 if (aff_entry->ae_cond != NULL)
|
|
|
3741 {
|
|
|
3742 sprintf((char *)buf, "^%s",
|
|
339
|
3743 aff_entry->ae_cond);
|
|
455
|
3744 vim_free(aff_entry->ae_prog);
|
|
|
3745 aff_entry->ae_prog = vim_regcomp(
|
|
|
3746 buf, RE_MAGIC + RE_STRING);
|
|
|
3747 }
|
|
|
3748 }
|
|
|
3749 }
|
|
|
3750 }
|
|
339
|
3751 }
|
|
|
3752
|
|
455
|
3753 if (aff_entry->ae_chop == NULL)
|
|
390
|
3754 {
|
|
455
|
3755 int idx;
|
|
|
3756 char_u **pp;
|
|
|
3757 int n;
|
|
|
3758
|
|
|
3759 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
3760 --idx)
|
|
|
3761 {
|
|
|
3762 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
3763 if (str_equal(p, aff_entry->ae_cond))
|
|
|
3764 break;
|
|
|
3765 }
|
|
|
3766 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
3767 {
|
|
|
3768 /* Not found, add a new condition. */
|
|
|
3769 idx = spin->si_prefcond.ga_len++;
|
|
|
3770 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
3771 + idx;
|
|
|
3772 if (aff_entry->ae_cond == NULL)
|
|
|
3773 *pp = NULL;
|
|
|
3774 else
|
|
474
|
3775 *pp = getroom_save(spin,
|
|
455
|
3776 aff_entry->ae_cond);
|
|
|
3777 }
|
|
|
3778
|
|
|
3779 /* Add the prefix to the prefix tree. */
|
|
|
3780 if (aff_entry->ae_add == NULL)
|
|
|
3781 p = (char_u *)"";
|
|
390
|
3782 else
|
|
455
|
3783 p = aff_entry->ae_add;
|
|
|
3784 /* PFX_FLAGS is a negative number, so that
|
|
|
3785 * tree_add_word() knows this is the prefix tree. */
|
|
|
3786 n = PFX_FLAGS;
|
|
|
3787 if (rare)
|
|
|
3788 n |= WFP_RARE;
|
|
|
3789 if (!cur_aff->ah_combine)
|
|
|
3790 n |= WFP_NC;
|
|
|
3791 if (upper)
|
|
|
3792 n |= WFP_UP;
|
|
474
|
3793 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
3794 idx, cur_aff->ah_newID);
|
|
390
|
3795 }
|
|
339
|
3796 }
|
|
240
|
3797 }
|
|
236
|
3798 }
|
|
255
|
3799 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
3800 {
|
|
|
3801 if (fol != NULL)
|
|
|
3802 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
3803 fname, lnum);
|
|
|
3804 else
|
|
|
3805 fol = vim_strsave(items[1]);
|
|
|
3806 }
|
|
|
3807 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
3808 {
|
|
|
3809 if (low != NULL)
|
|
|
3810 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
3811 fname, lnum);
|
|
|
3812 else
|
|
|
3813 low = vim_strsave(items[1]);
|
|
|
3814 }
|
|
|
3815 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
3816 {
|
|
|
3817 if (upp != NULL)
|
|
|
3818 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
3819 fname, lnum);
|
|
|
3820 else
|
|
|
3821 upp = vim_strsave(items[1]);
|
|
|
3822 }
|
|
236
|
3823 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
3824 {
|
|
236
|
3825 /* Ignore REP count */;
|
|
323
|
3826 if (!isdigit(*items[1]))
|
|
|
3827 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
3828 fname, lnum);
|
|
|
3829 }
|
|
236
|
3830 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 3)
|
|
|
3831 {
|
|
|
3832 /* REP item */
|
|
323
|
3833 if (do_rep)
|
|
|
3834 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
3835 }
|
|
|
3836 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
3837 {
|
|
|
3838 /* MAP item or count */
|
|
|
3839 if (!found_map)
|
|
|
3840 {
|
|
|
3841 /* First line contains the count. */
|
|
|
3842 found_map = TRUE;
|
|
|
3843 if (!isdigit(*items[1]))
|
|
|
3844 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
3845 fname, lnum);
|
|
|
3846 }
|
|
|
3847 else if (do_map)
|
|
|
3848 {
|
|
346
|
3849 int c;
|
|
|
3850
|
|
|
3851 /* Check that every character appears only once. */
|
|
|
3852 for (p = items[1]; *p != NUL; )
|
|
|
3853 {
|
|
|
3854 #ifdef FEAT_MBYTE
|
|
|
3855 c = mb_ptr2char_adv(&p);
|
|
|
3856 #else
|
|
|
3857 c = *p++;
|
|
|
3858 #endif
|
|
|
3859 if ((spin->si_map.ga_len > 0
|
|
|
3860 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
3861 != NULL)
|
|
|
3862 || vim_strchr(p, c) != NULL)
|
|
|
3863 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
3864 fname, lnum);
|
|
|
3865 }
|
|
|
3866
|
|
323
|
3867 /* We simply concatenate all the MAP strings, separated by
|
|
|
3868 * slashes. */
|
|
|
3869 ga_concat(&spin->si_map, items[1]);
|
|
|
3870 ga_append(&spin->si_map, '/');
|
|
|
3871 }
|
|
|
3872 }
|
|
|
3873 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
3874 {
|
|
|
3875 if (do_sal)
|
|
|
3876 {
|
|
|
3877 /* SAL item (sounds-a-like)
|
|
|
3878 * Either one of the known keys or a from-to pair. */
|
|
|
3879 if (STRCMP(items[1], "followup") == 0)
|
|
|
3880 spin->si_followup = sal_to_bool(items[2]);
|
|
|
3881 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
3882 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
3883 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
3884 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
3885 else
|
|
|
3886 /* when "to" is "_" it means empty */
|
|
|
3887 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
3888 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
3889 : items[2]);
|
|
|
3890 }
|
|
236
|
3891 }
|
|
375
|
3892 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
3893 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
3894 {
|
|
|
3895 if (do_sofo)
|
|
|
3896 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
3897 }
|
|
|
3898 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
3899 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
3900 {
|
|
|
3901 if (do_sofo)
|
|
|
3902 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
3903 }
|
|
300
|
3904 else
|
|
236
|
3905 smsg((char_u *)_("Unrecognized item in %s line %d: %s"),
|
|
|
3906 fname, lnum, items[0]);
|
|
|
3907 }
|
|
|
3908 }
|
|
|
3909
|
|
375
|
3910 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
3911 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
3912 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
3913 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
3914 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
3915
|
|
255
|
3916 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
3917 {
|
|
351
|
3918 if (spin->si_clear_chartab)
|
|
|
3919 {
|
|
|
3920 /* Clear the char type tables, don't want to use any of the
|
|
|
3921 * currently used spell properties. */
|
|
|
3922 init_spell_chartab();
|
|
|
3923 spin->si_clear_chartab = FALSE;
|
|
|
3924 }
|
|
|
3925
|
|
316
|
3926 /*
|
|
|
3927 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
3928 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
3929 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
3930 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
3931 */
|
|
|
3932 if (!spin->si_ascii
|
|
|
3933 #ifdef FEAT_MBYTE
|
|
|
3934 && !enc_utf8
|
|
|
3935 #endif
|
|
|
3936 )
|
|
260
|
3937 {
|
|
|
3938 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
3939 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
3940 else
|
|
316
|
3941 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
3942 }
|
|
255
|
3943
|
|
|
3944 vim_free(fol);
|
|
|
3945 vim_free(low);
|
|
|
3946 vim_free(upp);
|
|
|
3947 }
|
|
|
3948
|
|
236
|
3949 vim_free(pc);
|
|
|
3950 fclose(fd);
|
|
|
3951 return aff;
|
|
|
3952 }
|
|
|
3953
|
|
|
3954 /*
|
|
339
|
3955 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
3956 * NULL as equal.
|
|
|
3957 */
|
|
|
3958 static int
|
|
|
3959 str_equal(s1, s2)
|
|
|
3960 char_u *s1;
|
|
|
3961 char_u *s2;
|
|
|
3962 {
|
|
|
3963 if (s1 == NULL || s2 == NULL)
|
|
|
3964 return s1 == s2;
|
|
|
3965 return STRCMP(s1, s2) == 0;
|
|
|
3966 }
|
|
|
3967
|
|
|
3968 /*
|
|
323
|
3969 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
3970 * They are stored case-folded.
|
|
|
3971 */
|
|
|
3972 static void
|
|
|
3973 add_fromto(spin, gap, from, to)
|
|
|
3974 spellinfo_T *spin;
|
|
|
3975 garray_T *gap;
|
|
|
3976 char_u *from;
|
|
|
3977 char_u *to;
|
|
|
3978 {
|
|
|
3979 fromto_T *ftp;
|
|
|
3980 char_u word[MAXWLEN];
|
|
|
3981
|
|
|
3982 if (ga_grow(gap, 1) == OK)
|
|
|
3983 {
|
|
|
3984 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
3985 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
3986 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
3987 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
3988 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
3989 ++gap->ga_len;
|
|
|
3990 }
|
|
|
3991 }
|
|
|
3992
|
|
|
3993 /*
|
|
|
3994 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
3995 */
|
|
|
3996 static int
|
|
|
3997 sal_to_bool(s)
|
|
|
3998 char_u *s;
|
|
|
3999 {
|
|
|
4000 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
4001 }
|
|
|
4002
|
|
|
4003 /*
|
|
240
|
4004 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
4005 * When "s" is NULL FALSE is returned.
|
|
|
4006 */
|
|
|
4007 static int
|
|
|
4008 has_non_ascii(s)
|
|
|
4009 char_u *s;
|
|
|
4010 {
|
|
|
4011 char_u *p;
|
|
|
4012
|
|
|
4013 if (s != NULL)
|
|
|
4014 for (p = s; *p != NUL; ++p)
|
|
|
4015 if (*p >= 128)
|
|
|
4016 return TRUE;
|
|
|
4017 return FALSE;
|
|
|
4018 }
|
|
|
4019
|
|
|
4020 /*
|
|
236
|
4021 * Free the structure filled by spell_read_aff().
|
|
|
4022 */
|
|
|
4023 static void
|
|
|
4024 spell_free_aff(aff)
|
|
|
4025 afffile_T *aff;
|
|
|
4026 {
|
|
|
4027 hashtab_T *ht;
|
|
|
4028 hashitem_T *hi;
|
|
|
4029 int todo;
|
|
|
4030 affheader_T *ah;
|
|
300
|
4031 affentry_T *ae;
|
|
236
|
4032
|
|
|
4033 vim_free(aff->af_enc);
|
|
|
4034
|
|
339
|
4035 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
4036 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
4037 {
|
|
|
4038 todo = ht->ht_used;
|
|
|
4039 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
4040 {
|
|
|
4041 if (!HASHITEM_EMPTY(hi))
|
|
|
4042 {
|
|
|
4043 --todo;
|
|
|
4044 ah = HI2AH(hi);
|
|
300
|
4045 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
4046 vim_free(ae->ae_prog);
|
|
236
|
4047 }
|
|
|
4048 }
|
|
|
4049 if (ht == &aff->af_suff)
|
|
|
4050 break;
|
|
|
4051 }
|
|
300
|
4052
|
|
236
|
4053 hash_clear(&aff->af_pref);
|
|
|
4054 hash_clear(&aff->af_suff);
|
|
|
4055 }
|
|
|
4056
|
|
|
4057 /*
|
|
300
|
4058 * Read dictionary file "fname".
|
|
236
|
4059 * Returns OK or FAIL;
|
|
|
4060 */
|
|
|
4061 static int
|
|
474
|
4062 spell_read_dic(spin, fname, affile)
|
|
|
4063 spellinfo_T *spin;
|
|
236
|
4064 char_u *fname;
|
|
300
|
4065 afffile_T *affile;
|
|
236
|
4066 {
|
|
300
|
4067 hashtab_T ht;
|
|
236
|
4068 char_u line[MAXLINELEN];
|
|
300
|
4069 char_u *afflist;
|
|
339
|
4070 char_u *pfxlist;
|
|
300
|
4071 char_u *dw;
|
|
236
|
4072 char_u *pc;
|
|
|
4073 char_u *w;
|
|
|
4074 int l;
|
|
|
4075 hash_T hash;
|
|
|
4076 hashitem_T *hi;
|
|
|
4077 FILE *fd;
|
|
|
4078 int lnum = 1;
|
|
300
|
4079 int non_ascii = 0;
|
|
|
4080 int retval = OK;
|
|
|
4081 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
4082 int flags;
|
|
236
|
4083
|
|
300
|
4084 /*
|
|
|
4085 * Open the file.
|
|
|
4086 */
|
|
310
|
4087 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4088 if (fd == NULL)
|
|
|
4089 {
|
|
|
4090 EMSG2(_(e_notopen), fname);
|
|
|
4091 return FAIL;
|
|
|
4092 }
|
|
|
4093
|
|
300
|
4094 /* The hashtable is only used to detect duplicated words. */
|
|
|
4095 hash_init(&ht);
|
|
|
4096
|
|
310
|
4097 if (spin->si_verbose || p_verbose > 2)
|
|
|
4098 {
|
|
|
4099 if (!spin->si_verbose)
|
|
|
4100 verbose_enter();
|
|
366
|
4101 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
4102 out_flush();
|
|
|
4103 if (!spin->si_verbose)
|
|
|
4104 verbose_leave();
|
|
|
4105 }
|
|
236
|
4106
|
|
474
|
4107 /* start with a message for the first line */
|
|
|
4108 spin->si_msg_count = 999999;
|
|
|
4109
|
|
236
|
4110 /* Read and ignore the first line: word count. */
|
|
|
4111 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
4112 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
4113 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
4114
|
|
|
4115 /*
|
|
|
4116 * Read all the lines in the file one by one.
|
|
|
4117 * The words are converted to 'encoding' here, before being added to
|
|
|
4118 * the hashtable.
|
|
|
4119 */
|
|
255
|
4120 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
4121 {
|
|
255
|
4122 line_breakcheck();
|
|
236
|
4123 ++lnum;
|
|
455
|
4124 if (line[0] == '#' || line[0] == '/')
|
|
351
|
4125 continue; /* comment line */
|
|
236
|
4126
|
|
300
|
4127 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
4128 * the word is kept to allow e.g., "et al.". */
|
|
236
|
4129 l = STRLEN(line);
|
|
|
4130 while (l > 0 && line[l - 1] <= ' ')
|
|
|
4131 --l;
|
|
|
4132 if (l == 0)
|
|
|
4133 continue; /* empty line */
|
|
|
4134 line[l] = NUL;
|
|
|
4135
|
|
|
4136 /* Find the optional affix names. */
|
|
300
|
4137 afflist = vim_strchr(line, '/');
|
|
|
4138 if (afflist != NULL)
|
|
|
4139 *afflist++ = NUL;
|
|
236
|
4140
|
|
300
|
4141 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4142 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4143 {
|
|
|
4144 ++non_ascii;
|
|
240
|
4145 continue;
|
|
300
|
4146 }
|
|
240
|
4147
|
|
310
|
4148 #ifdef FEAT_MBYTE
|
|
236
|
4149 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
4150 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4151 {
|
|
300
|
4152 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
4153 if (pc == NULL)
|
|
|
4154 {
|
|
|
4155 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4156 fname, lnum, line);
|
|
|
4157 continue;
|
|
|
4158 }
|
|
236
|
4159 w = pc;
|
|
|
4160 }
|
|
|
4161 else
|
|
310
|
4162 #endif
|
|
236
|
4163 {
|
|
|
4164 pc = NULL;
|
|
|
4165 w = line;
|
|
|
4166 }
|
|
|
4167
|
|
474
|
4168 /* This takes time, print a message every 10000 words. */
|
|
|
4169 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
4170 {
|
|
|
4171 spin->si_msg_count = 0;
|
|
339
|
4172 vim_snprintf((char *)message, sizeof(message),
|
|
|
4173 _("line %6d, word %6d - %s"),
|
|
|
4174 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
4175 msg_start();
|
|
|
4176 msg_puts_long_attr(message, 0);
|
|
|
4177 msg_clr_eos();
|
|
|
4178 msg_didout = FALSE;
|
|
|
4179 msg_col = 0;
|
|
|
4180 out_flush();
|
|
|
4181 }
|
|
|
4182
|
|
300
|
4183 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
4184 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
4185 if (dw == NULL)
|
|
300
|
4186 retval = FAIL;
|
|
|
4187 vim_free(pc);
|
|
|
4188 if (retval == FAIL)
|
|
236
|
4189 break;
|
|
|
4190
|
|
300
|
4191 hash = hash_hash(dw);
|
|
|
4192 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
4193 if (!HASHITEM_EMPTY(hi))
|
|
|
4194 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
4195 fname, lnum, dw);
|
|
236
|
4196 else
|
|
300
|
4197 hash_add_item(&ht, hi, dw, hash);
|
|
|
4198
|
|
307
|
4199 flags = 0;
|
|
339
|
4200 pfxlist = NULL;
|
|
307
|
4201 if (afflist != NULL)
|
|
|
4202 {
|
|
|
4203 /* Check for affix name that stands for keep-case word and stands
|
|
|
4204 * for rare word (if defined). */
|
|
310
|
4205 if (affile->af_kep != NUL
|
|
|
4206 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
389
|
4207 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
307
|
4208 if (affile->af_rar != NUL
|
|
|
4209 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
4210 flags |= WF_RARE;
|
|
346
|
4211 if (affile->af_bad != NUL
|
|
|
4212 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
4213 flags |= WF_BANNED;
|
|
339
|
4214
|
|
|
4215 if (affile->af_pfxpostpone)
|
|
|
4216 /* Need to store the list of prefix IDs with the word. */
|
|
474
|
4217 pfxlist = get_pfxlist(spin, affile, afflist);
|
|
307
|
4218 }
|
|
|
4219
|
|
300
|
4220 /* Add the word to the word tree(s). */
|
|
474
|
4221 if (store_word(spin, dw, flags, spin->si_region, pfxlist) == FAIL)
|
|
300
|
4222 retval = FAIL;
|
|
236
|
4223
|
|
300
|
4224 if (afflist != NULL)
|
|
|
4225 {
|
|
|
4226 /* Find all matching suffixes and add the resulting words.
|
|
|
4227 * Additionally do matching prefixes that combine. */
|
|
474
|
4228 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
4229 &affile->af_suff, &affile->af_pref,
|
|
339
|
4230 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
4231 retval = FAIL;
|
|
|
4232
|
|
|
4233 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
4234 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
4235 &affile->af_pref, NULL,
|
|
|
4236 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
4237 retval = FAIL;
|
|
|
4238 }
|
|
236
|
4239 }
|
|
|
4240
|
|
300
|
4241 if (spin->si_ascii && non_ascii > 0)
|
|
|
4242 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4243 non_ascii);
|
|
|
4244 hash_clear(&ht);
|
|
|
4245
|
|
236
|
4246 fclose(fd);
|
|
300
|
4247 return retval;
|
|
236
|
4248 }
|
|
|
4249
|
|
|
4250 /*
|
|
339
|
4251 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
4252 * Used for PFXPOSTPONE.
|
|
|
4253 * Returns a string allocated with getroom(). NULL when there are no prefixes
|
|
|
4254 * or when out of memory.
|
|
|
4255 */
|
|
|
4256 static char_u *
|
|
474
|
4257 get_pfxlist(spin, affile, afflist)
|
|
|
4258 spellinfo_T *spin;
|
|
339
|
4259 afffile_T *affile;
|
|
|
4260 char_u *afflist;
|
|
|
4261 {
|
|
|
4262 char_u *p;
|
|
|
4263 int cnt;
|
|
|
4264 int round;
|
|
|
4265 char_u *res = NULL;
|
|
|
4266 char_u key[2];
|
|
|
4267 hashitem_T *hi;
|
|
|
4268
|
|
|
4269 key[1] = NUL;
|
|
|
4270
|
|
|
4271 /* round 1: count the number of prefix IDs.
|
|
|
4272 * round 2: move prefix IDs to "res" */
|
|
|
4273 for (round = 1; round <= 2; ++round)
|
|
|
4274 {
|
|
|
4275 cnt = 0;
|
|
|
4276 for (p = afflist; *p != NUL; ++p)
|
|
|
4277 {
|
|
|
4278 key[0] = *p;
|
|
|
4279 hi = hash_find(&affile->af_pref, key);
|
|
|
4280 if (!HASHITEM_EMPTY(hi))
|
|
|
4281 {
|
|
|
4282 /* This is a prefix ID, use the new number. */
|
|
|
4283 if (round == 2)
|
|
|
4284 res[cnt] = HI2AH(hi)->ah_newID;
|
|
|
4285 ++cnt;
|
|
|
4286 }
|
|
|
4287 }
|
|
|
4288 if (round == 1 && cnt > 0)
|
|
474
|
4289 res = getroom(spin, cnt + 1, FALSE);
|
|
339
|
4290 if (res == NULL)
|
|
|
4291 break;
|
|
|
4292 }
|
|
|
4293
|
|
|
4294 if (res != NULL)
|
|
|
4295 res[cnt] = NUL;
|
|
|
4296 return res;
|
|
|
4297 }
|
|
|
4298
|
|
|
4299 /*
|
|
300
|
4300 * Apply affixes to a word and store the resulting words.
|
|
|
4301 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
4302 * prefixes or suffixes.
|
|
|
4303 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
4304 * the resulting words for combining affixes.
|
|
|
4305 *
|
|
|
4306 * Returns FAIL when out of memory.
|
|
236
|
4307 */
|
|
300
|
4308 static int
|
|
474
|
4309 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags, pfxlist)
|
|
|
4310 spellinfo_T *spin; /* spell info */
|
|
300
|
4311 char_u *word; /* basic word start */
|
|
|
4312 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
4313 afffile_T *affile;
|
|
300
|
4314 hashtab_T *ht;
|
|
|
4315 hashtab_T *xht;
|
|
|
4316 int comb; /* only use affixes that combine */
|
|
307
|
4317 int flags; /* flags for the word */
|
|
339
|
4318 char_u *pfxlist; /* list of prefix IDs */
|
|
236
|
4319 {
|
|
|
4320 int todo;
|
|
|
4321 hashitem_T *hi;
|
|
300
|
4322 affheader_T *ah;
|
|
|
4323 affentry_T *ae;
|
|
|
4324 regmatch_T regmatch;
|
|
|
4325 char_u newword[MAXWLEN];
|
|
|
4326 int retval = OK;
|
|
|
4327 int i;
|
|
|
4328 char_u *p;
|
|
366
|
4329 int use_flags;
|
|
390
|
4330 char_u *use_pfxlist;
|
|
455
|
4331 int c;
|
|
236
|
4332
|
|
300
|
4333 todo = ht->ht_used;
|
|
|
4334 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
4335 {
|
|
|
4336 if (!HASHITEM_EMPTY(hi))
|
|
|
4337 {
|
|
|
4338 --todo;
|
|
300
|
4339 ah = HI2AH(hi);
|
|
236
|
4340
|
|
300
|
4341 /* Check that the affix combines, if required, and that the word
|
|
|
4342 * supports this affix. */
|
|
455
|
4343 c = PTR2CHAR(ah->ah_key);
|
|
|
4344 if ((!comb || ah->ah_combine) && vim_strchr(afflist, c) != NULL)
|
|
236
|
4345 {
|
|
300
|
4346 /* Loop over all affix entries with this name. */
|
|
|
4347 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
4348 {
|
|
300
|
4349 /* Check the condition. It's not logical to match case
|
|
|
4350 * here, but it is required for compatibility with
|
|
339
|
4351 * Myspell.
|
|
|
4352 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
4353 * prefixes with a chop string. */
|
|
300
|
4354 regmatch.regprog = ae->ae_prog;
|
|
|
4355 regmatch.rm_ic = FALSE;
|
|
339
|
4356 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
4357 || ae->ae_chop != NULL)
|
|
|
4358 && (ae->ae_prog == NULL
|
|
|
4359 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
4360 {
|
|
|
4361 /* Match. Remove the chop and add the affix. */
|
|
|
4362 if (xht == NULL)
|
|
240
|
4363 {
|
|
300
|
4364 /* prefix: chop/add at the start of the word */
|
|
|
4365 if (ae->ae_add == NULL)
|
|
|
4366 *newword = NUL;
|
|
|
4367 else
|
|
|
4368 STRCPY(newword, ae->ae_add);
|
|
|
4369 p = word;
|
|
|
4370 if (ae->ae_chop != NULL)
|
|
310
|
4371 {
|
|
300
|
4372 /* Skip chop string. */
|
|
310
|
4373 #ifdef FEAT_MBYTE
|
|
|
4374 if (has_mbyte)
|
|
324
|
4375 {
|
|
310
|
4376 i = mb_charlen(ae->ae_chop);
|
|
324
|
4377 for ( ; i > 0; --i)
|
|
|
4378 mb_ptr_adv(p);
|
|
|
4379 }
|
|
310
|
4380 else
|
|
|
4381 #endif
|
|
324
|
4382 p += STRLEN(ae->ae_chop);
|
|
310
|
4383 }
|
|
300
|
4384 STRCAT(newword, p);
|
|
|
4385 }
|
|
|
4386 else
|
|
|
4387 {
|
|
|
4388 /* suffix: chop/add at the end of the word */
|
|
|
4389 STRCPY(newword, word);
|
|
|
4390 if (ae->ae_chop != NULL)
|
|
|
4391 {
|
|
|
4392 /* Remove chop string. */
|
|
|
4393 p = newword + STRLEN(newword);
|
|
376
|
4394 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
4395 for ( ; i > 0; --i)
|
|
300
|
4396 mb_ptr_back(newword, p);
|
|
|
4397 *p = NUL;
|
|
|
4398 }
|
|
|
4399 if (ae->ae_add != NULL)
|
|
|
4400 STRCAT(newword, ae->ae_add);
|
|
240
|
4401 }
|
|
|
4402
|
|
366
|
4403 /* Obey the "rare" flag of the affix. */
|
|
|
4404 if (ae->ae_rare)
|
|
|
4405 use_flags = flags | WF_RARE;
|
|
|
4406 else
|
|
|
4407 use_flags = flags;
|
|
390
|
4408 use_pfxlist = pfxlist;
|
|
|
4409
|
|
|
4410 /* When there are postponed prefixes... */
|
|
393
|
4411 if (spin->si_prefroot != NULL
|
|
|
4412 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
4413 {
|
|
|
4414 /* ... add a flag to indicate an affix was used. */
|
|
|
4415 use_flags |= WF_HAS_AFF;
|
|
|
4416
|
|
|
4417 /* ... don't use a prefix list if combining
|
|
|
4418 * affixes is not allowed */
|
|
|
4419 if (!ah->ah_combine || comb)
|
|
|
4420 use_pfxlist = NULL;
|
|
|
4421 }
|
|
366
|
4422
|
|
300
|
4423 /* Store the modified word. */
|
|
474
|
4424 if (store_word(spin, newword, use_flags,
|
|
390
|
4425 spin->si_region, use_pfxlist) == FAIL)
|
|
300
|
4426 retval = FAIL;
|
|
236
|
4427
|
|
300
|
4428 /* When added a suffix and combining is allowed also
|
|
|
4429 * try adding prefixes additionally. */
|
|
|
4430 if (xht != NULL && ah->ah_combine)
|
|
474
|
4431 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
4432 xht, NULL, TRUE,
|
|
|
4433 use_flags, use_pfxlist) == FAIL)
|
|
300
|
4434 retval = FAIL;
|
|
236
|
4435 }
|
|
|
4436 }
|
|
|
4437 }
|
|
|
4438 }
|
|
|
4439 }
|
|
|
4440
|
|
|
4441 return retval;
|
|
|
4442 }
|
|
|
4443
|
|
|
4444 /*
|
|
300
|
4445 * Read a file with a list of words.
|
|
236
|
4446 */
|
|
|
4447 static int
|
|
474
|
4448 spell_read_wordfile(spin, fname)
|
|
|
4449 spellinfo_T *spin;
|
|
300
|
4450 char_u *fname;
|
|
236
|
4451 {
|
|
300
|
4452 FILE *fd;
|
|
|
4453 long lnum = 0;
|
|
|
4454 char_u rline[MAXLINELEN];
|
|
|
4455 char_u *line;
|
|
|
4456 char_u *pc = NULL;
|
|
381
|
4457 char_u *p;
|
|
300
|
4458 int l;
|
|
|
4459 int retval = OK;
|
|
|
4460 int did_word = FALSE;
|
|
|
4461 int non_ascii = 0;
|
|
307
|
4462 int flags;
|
|
316
|
4463 int regionmask;
|
|
236
|
4464
|
|
300
|
4465 /*
|
|
|
4466 * Open the file.
|
|
|
4467 */
|
|
310
|
4468 fd = mch_fopen((char *)fname, "r");
|
|
300
|
4469 if (fd == NULL)
|
|
236
|
4470 {
|
|
300
|
4471 EMSG2(_(e_notopen), fname);
|
|
|
4472 return FAIL;
|
|
236
|
4473 }
|
|
|
4474
|
|
310
|
4475 if (spin->si_verbose || p_verbose > 2)
|
|
|
4476 {
|
|
|
4477 if (!spin->si_verbose)
|
|
|
4478 verbose_enter();
|
|
366
|
4479 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
4480 out_flush();
|
|
|
4481 if (!spin->si_verbose)
|
|
|
4482 verbose_leave();
|
|
|
4483 }
|
|
300
|
4484
|
|
|
4485 /*
|
|
|
4486 * Read all the lines in the file one by one.
|
|
|
4487 */
|
|
|
4488 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
4489 {
|
|
|
4490 line_breakcheck();
|
|
|
4491 ++lnum;
|
|
|
4492
|
|
|
4493 /* Skip comment lines. */
|
|
|
4494 if (*rline == '#')
|
|
|
4495 continue;
|
|
|
4496
|
|
|
4497 /* Remove CR, LF and white space from the end. */
|
|
|
4498 l = STRLEN(rline);
|
|
|
4499 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
4500 --l;
|
|
|
4501 if (l == 0)
|
|
|
4502 continue; /* empty or blank line */
|
|
|
4503 rline[l] = NUL;
|
|
|
4504
|
|
|
4505 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
4506 vim_free(pc);
|
|
310
|
4507 #ifdef FEAT_MBYTE
|
|
300
|
4508 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
4509 {
|
|
|
4510 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
4511 if (pc == NULL)
|
|
|
4512 {
|
|
|
4513 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4514 fname, lnum, rline);
|
|
|
4515 continue;
|
|
|
4516 }
|
|
|
4517 line = pc;
|
|
|
4518 }
|
|
|
4519 else
|
|
310
|
4520 #endif
|
|
300
|
4521 {
|
|
|
4522 pc = NULL;
|
|
|
4523 line = rline;
|
|
|
4524 }
|
|
|
4525
|
|
307
|
4526 if (*line == '/')
|
|
300
|
4527 {
|
|
307
|
4528 ++line;
|
|
|
4529 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
4530 {
|
|
|
4531 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
4532 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
4533 fname, lnum, line - 1);
|
|
300
|
4534 else if (did_word)
|
|
316
|
4535 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
4536 fname, lnum, line - 1);
|
|
300
|
4537 else
|
|
|
4538 {
|
|
310
|
4539 #ifdef FEAT_MBYTE
|
|
|
4540 char_u *enc;
|
|
|
4541
|
|
300
|
4542 /* Setup for conversion to 'encoding'. */
|
|
316
|
4543 line += 10;
|
|
|
4544 enc = enc_canonize(line);
|
|
300
|
4545 if (enc != NULL && !spin->si_ascii
|
|
|
4546 && convert_setup(&spin->si_conv, enc,
|
|
|
4547 p_enc) == FAIL)
|
|
|
4548 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
4549 fname, line, p_enc);
|
|
300
|
4550 vim_free(enc);
|
|
375
|
4551 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4552 #else
|
|
|
4553 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4554 #endif
|
|
300
|
4555 }
|
|
307
|
4556 continue;
|
|
300
|
4557 }
|
|
307
|
4558
|
|
316
|
4559 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
4560 {
|
|
|
4561 if (spin->si_region_count > 1)
|
|
|
4562 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
4563 fname, lnum, line);
|
|
|
4564 else
|
|
|
4565 {
|
|
|
4566 line += 8;
|
|
|
4567 if (STRLEN(line) > 16)
|
|
|
4568 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
4569 fname, lnum, line);
|
|
|
4570 else
|
|
|
4571 {
|
|
|
4572 spin->si_region_count = STRLEN(line) / 2;
|
|
|
4573 STRCPY(spin->si_region_name, line);
|
|
389
|
4574
|
|
|
4575 /* Adjust the mask for a word valid in all regions. */
|
|
|
4576 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
4577 }
|
|
|
4578 }
|
|
|
4579 continue;
|
|
|
4580 }
|
|
|
4581
|
|
381
|
4582 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
4583 fname, lnum, line - 1);
|
|
|
4584 continue;
|
|
|
4585 }
|
|
|
4586
|
|
|
4587 flags = 0;
|
|
|
4588 regionmask = spin->si_region;
|
|
|
4589
|
|
|
4590 /* Check for flags and region after a slash. */
|
|
|
4591 p = vim_strchr(line, '/');
|
|
|
4592 if (p != NULL)
|
|
|
4593 {
|
|
|
4594 *p++ = NUL;
|
|
|
4595 while (*p != NUL)
|
|
|
4596 {
|
|
|
4597 if (*p == '=') /* keep-case word */
|
|
389
|
4598 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
4599 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
4600 flags |= WF_BANNED;
|
|
|
4601 else if (*p == '?') /* Rare word. */
|
|
|
4602 flags |= WF_RARE;
|
|
|
4603 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
4604 {
|
|
381
|
4605 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
4606 regionmask = 0;
|
|
|
4607 flags |= WF_REGION;
|
|
|
4608
|
|
|
4609 l = *p - '0';
|
|
316
|
4610 if (l > spin->si_region_count)
|
|
|
4611 {
|
|
|
4612 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
4613 fname, lnum, p);
|
|
316
|
4614 break;
|
|
|
4615 }
|
|
|
4616 regionmask |= 1 << (l - 1);
|
|
|
4617 }
|
|
381
|
4618 else
|
|
|
4619 {
|
|
|
4620 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
4621 fname, lnum, p);
|
|
|
4622 break;
|
|
|
4623 }
|
|
|
4624 ++p;
|
|
307
|
4625 }
|
|
300
|
4626 }
|
|
|
4627
|
|
|
4628 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4629 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4630 {
|
|
|
4631 ++non_ascii;
|
|
|
4632 continue;
|
|
|
4633 }
|
|
|
4634
|
|
|
4635 /* Normal word: store it. */
|
|
474
|
4636 if (store_word(spin, line, flags, regionmask, NULL) == FAIL)
|
|
300
|
4637 {
|
|
|
4638 retval = FAIL;
|
|
|
4639 break;
|
|
|
4640 }
|
|
|
4641 did_word = TRUE;
|
|
|
4642 }
|
|
|
4643
|
|
|
4644 vim_free(pc);
|
|
|
4645 fclose(fd);
|
|
|
4646
|
|
310
|
4647 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
4648 {
|
|
|
4649 if (p_verbose > 2)
|
|
|
4650 verbose_enter();
|
|
300
|
4651 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4652 non_ascii);
|
|
310
|
4653 if (p_verbose > 2)
|
|
|
4654 verbose_leave();
|
|
|
4655 }
|
|
300
|
4656 return retval;
|
|
236
|
4657 }
|
|
|
4658
|
|
|
4659 /*
|
|
300
|
4660 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
4661 * This avoids calling free() for every little struct we use (and keeping
|
|
|
4662 * track of them).
|
|
300
|
4663 * The memory is cleared to all zeros.
|
|
|
4664 * Returns NULL when out of memory.
|
|
|
4665 */
|
|
|
4666 static void *
|
|
474
|
4667 getroom(spin, len, align)
|
|
|
4668 spellinfo_T *spin;
|
|
447
|
4669 size_t len; /* length needed */
|
|
|
4670 int align; /* align for pointer */
|
|
300
|
4671 {
|
|
|
4672 char_u *p;
|
|
474
|
4673 sblock_T *bl = spin->si_blocks;
|
|
300
|
4674
|
|
447
|
4675 if (align && bl != NULL)
|
|
|
4676 /* Round size up for alignment. On some systems structures need to be
|
|
|
4677 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
4678 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
4679 & ~(sizeof(char *) - 1);
|
|
|
4680
|
|
300
|
4681 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
4682 {
|
|
|
4683 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
4684 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
4685 if (bl == NULL)
|
|
|
4686 return NULL;
|
|
474
|
4687 bl->sb_next = spin->si_blocks;
|
|
|
4688 spin->si_blocks = bl;
|
|
300
|
4689 bl->sb_used = 0;
|
|
|
4690 }
|
|
|
4691
|
|
|
4692 p = bl->sb_data + bl->sb_used;
|
|
|
4693 bl->sb_used += len;
|
|
|
4694
|
|
|
4695 return p;
|
|
|
4696 }
|
|
|
4697
|
|
|
4698 /*
|
|
|
4699 * Make a copy of a string into memory allocated with getroom().
|
|
|
4700 */
|
|
|
4701 static char_u *
|
|
474
|
4702 getroom_save(spin, s)
|
|
|
4703 spellinfo_T *spin;
|
|
300
|
4704 char_u *s;
|
|
|
4705 {
|
|
|
4706 char_u *sc;
|
|
|
4707
|
|
474
|
4708 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
4709 if (sc != NULL)
|
|
|
4710 STRCPY(sc, s);
|
|
|
4711 return sc;
|
|
|
4712 }
|
|
|
4713
|
|
|
4714
|
|
|
4715 /*
|
|
|
4716 * Free the list of allocated sblock_T.
|
|
236
|
4717 */
|
|
|
4718 static void
|
|
300
|
4719 free_blocks(bl)
|
|
|
4720 sblock_T *bl;
|
|
236
|
4721 {
|
|
300
|
4722 sblock_T *next;
|
|
236
|
4723
|
|
300
|
4724 while (bl != NULL)
|
|
236
|
4725 {
|
|
300
|
4726 next = bl->sb_next;
|
|
|
4727 vim_free(bl);
|
|
|
4728 bl = next;
|
|
236
|
4729 }
|
|
|
4730 }
|
|
|
4731
|
|
|
4732 /*
|
|
300
|
4733 * Allocate the root of a word tree.
|
|
236
|
4734 */
|
|
300
|
4735 static wordnode_T *
|
|
474
|
4736 wordtree_alloc(spin)
|
|
|
4737 spellinfo_T *spin;
|
|
|
4738 {
|
|
|
4739 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
4740 }
|
|
|
4741
|
|
|
4742 /*
|
|
300
|
4743 * Store a word in the tree(s).
|
|
390
|
4744 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
4745 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
4746 * used to find suggestions.
|
|
300
|
4747 * For a keep-case word also store it in the keep-case tree.
|
|
390
|
4748 * When "pfxlist" is not NULL store the word for each postponed prefix ID.
|
|
236
|
4749 */
|
|
|
4750 static int
|
|
474
|
4751 store_word(spin, word, flags, region, pfxlist)
|
|
|
4752 spellinfo_T *spin;
|
|
300
|
4753 char_u *word;
|
|
307
|
4754 int flags; /* extra flags, WF_BANNED */
|
|
316
|
4755 int region; /* supported region(s) */
|
|
339
|
4756 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
236
|
4757 {
|
|
300
|
4758 int len = STRLEN(word);
|
|
|
4759 int ct = captype(word, word + len);
|
|
|
4760 char_u foldword[MAXWLEN];
|
|
339
|
4761 int res = OK;
|
|
|
4762 char_u *p;
|
|
236
|
4763
|
|
323
|
4764 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
4765 for (p = pfxlist; res == OK; ++p)
|
|
|
4766 {
|
|
474
|
4767 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
|
4768 region, p == NULL ? 0 : *p);
|
|
339
|
4769 if (p == NULL || *p == NUL)
|
|
|
4770 break;
|
|
|
4771 }
|
|
334
|
4772 ++spin->si_foldwcount;
|
|
307
|
4773
|
|
474
|
4774 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
4775 {
|
|
339
|
4776 for (p = pfxlist; res == OK; ++p)
|
|
|
4777 {
|
|
474
|
4778 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
|
4779 region, p == NULL ? 0 : *p);
|
|
339
|
4780 if (p == NULL || *p == NUL)
|
|
|
4781 break;
|
|
|
4782 }
|
|
334
|
4783 ++spin->si_keepwcount;
|
|
|
4784 }
|
|
300
|
4785 return res;
|
|
236
|
4786 }
|
|
|
4787
|
|
|
4788 /*
|
|
300
|
4789 * Add word "word" to a word tree at "root".
|
|
366
|
4790 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
4791 * "rare" and "region" is the condition nr.
|
|
255
|
4792 * Returns FAIL when out of memory.
|
|
236
|
4793 */
|
|
255
|
4794 static int
|
|
474
|
4795 tree_add_word(spin, word, root, flags, region, prefixID)
|
|
|
4796 spellinfo_T *spin;
|
|
300
|
4797 char_u *word;
|
|
|
4798 wordnode_T *root;
|
|
|
4799 int flags;
|
|
|
4800 int region;
|
|
339
|
4801 int prefixID;
|
|
236
|
4802 {
|
|
300
|
4803 wordnode_T *node = root;
|
|
|
4804 wordnode_T *np;
|
|
469
|
4805 wordnode_T *copyp, **copyprev;
|
|
300
|
4806 wordnode_T **prev = NULL;
|
|
|
4807 int i;
|
|
255
|
4808
|
|
300
|
4809 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
4810 for (i = 0; ; ++i)
|
|
255
|
4811 {
|
|
469
|
4812 /* When there is more than one reference to this node we need to make
|
|
|
4813 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
4814 * (we don't optimize for a partly shared list of siblings). */
|
|
|
4815 if (node != NULL && node->wn_refs > 1)
|
|
|
4816 {
|
|
|
4817 --node->wn_refs;
|
|
|
4818 copyprev = prev;
|
|
|
4819 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
4820 {
|
|
|
4821 /* Allocate a new node and copy the info. */
|
|
474
|
4822 np = get_wordnode(spin);
|
|
469
|
4823 if (np == NULL)
|
|
|
4824 return FAIL;
|
|
|
4825 np->wn_child = copyp->wn_child;
|
|
|
4826 if (np->wn_child != NULL)
|
|
|
4827 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
4828 np->wn_byte = copyp->wn_byte;
|
|
|
4829 if (np->wn_byte == NUL)
|
|
|
4830 {
|
|
|
4831 np->wn_flags = copyp->wn_flags;
|
|
|
4832 np->wn_region = copyp->wn_region;
|
|
|
4833 np->wn_prefixID = copyp->wn_prefixID;
|
|
|
4834 }
|
|
|
4835
|
|
|
4836 /* Link the new node in the list, there will be one ref. */
|
|
|
4837 np->wn_refs = 1;
|
|
|
4838 *copyprev = np;
|
|
|
4839 copyprev = &np->wn_sibling;
|
|
|
4840
|
|
|
4841 /* Let "node" point to the head of the copied list. */
|
|
|
4842 if (copyp == node)
|
|
|
4843 node = np;
|
|
|
4844 }
|
|
|
4845 }
|
|
|
4846
|
|
300
|
4847 /* Look for the sibling that has the same character. They are sorted
|
|
|
4848 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
4849 * higher byte value. For zero bytes (end of word) the sorting is
|
|
469
|
4850 * done on flags and then on prefixID. */
|
|
339
|
4851 while (node != NULL
|
|
|
4852 && (node->wn_byte < word[i]
|
|
|
4853 || (node->wn_byte == NUL
|
|
|
4854 && (flags < 0
|
|
|
4855 ? node->wn_prefixID < prefixID
|
|
390
|
4856 : node->wn_flags < (flags & WN_MASK)
|
|
|
4857 || (node->wn_flags == (flags & WN_MASK)
|
|
339
|
4858 && node->wn_prefixID < prefixID)))))
|
|
236
|
4859 {
|
|
300
|
4860 prev = &node->wn_sibling;
|
|
|
4861 node = *prev;
|
|
236
|
4862 }
|
|
339
|
4863 if (node == NULL
|
|
|
4864 || node->wn_byte != word[i]
|
|
|
4865 || (word[i] == NUL
|
|
|
4866 && (flags < 0
|
|
390
|
4867 || node->wn_flags != (flags & WN_MASK)
|
|
339
|
4868 || node->wn_prefixID != prefixID)))
|
|
255
|
4869 {
|
|
300
|
4870 /* Allocate a new node. */
|
|
474
|
4871 np = get_wordnode(spin);
|
|
300
|
4872 if (np == NULL)
|
|
|
4873 return FAIL;
|
|
|
4874 np->wn_byte = word[i];
|
|
469
|
4875
|
|
|
4876 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
4877 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
4878 * make ref count of sibling one (matters when inserting in front
|
|
|
4879 * of the list of siblings). */
|
|
|
4880 if (node == NULL)
|
|
|
4881 np->wn_refs = 1;
|
|
|
4882 else
|
|
|
4883 {
|
|
|
4884 np->wn_refs = node->wn_refs;
|
|
|
4885 node->wn_refs = 1;
|
|
|
4886 }
|
|
300
|
4887 *prev = np;
|
|
|
4888 np->wn_sibling = node;
|
|
|
4889 node = np;
|
|
255
|
4890 }
|
|
300
|
4891
|
|
|
4892 if (word[i] == NUL)
|
|
|
4893 {
|
|
|
4894 node->wn_flags = flags;
|
|
|
4895 node->wn_region |= region;
|
|
339
|
4896 node->wn_prefixID = prefixID;
|
|
300
|
4897 break;
|
|
|
4898 }
|
|
|
4899 prev = &node->wn_child;
|
|
|
4900 node = *prev;
|
|
255
|
4901 }
|
|
469
|
4902 #ifdef SPELL_PRINTTREE
|
|
|
4903 smsg("Added \"%s\"", word);
|
|
|
4904 spell_print_tree(root->wn_sibling);
|
|
|
4905 #endif
|
|
|
4906
|
|
474
|
4907 /* count nr of words added since last message */
|
|
|
4908 ++spin->si_msg_count;
|
|
|
4909
|
|
469
|
4910 /*
|
|
|
4911 * Every so many words compress the tree, so that we don't use too much
|
|
|
4912 * memory.
|
|
|
4913 */
|
|
474
|
4914 i = FALSE;
|
|
|
4915 if (root == spin->si_foldroot)
|
|
|
4916 {
|
|
|
4917 if (++spin->si_fold_added >= SPELL_COMPRESS_CNT)
|
|
|
4918 {
|
|
|
4919 i = TRUE;
|
|
|
4920 spin->si_fold_added = 0;
|
|
|
4921 }
|
|
|
4922 }
|
|
|
4923 else if (root == spin->si_keeproot)
|
|
|
4924 {
|
|
|
4925 if (++spin->si_keep_added >= SPELL_COMPRESS_CNT)
|
|
|
4926 {
|
|
|
4927 i = TRUE;
|
|
|
4928 spin->si_keep_added = 0;
|
|
|
4929 }
|
|
|
4930 }
|
|
|
4931 if (i)
|
|
|
4932 {
|
|
|
4933 if (spin->si_verbose)
|
|
|
4934 {
|
|
|
4935 msg_start();
|
|
|
4936 msg_puts((char_u *)_(msg_compressing));
|
|
|
4937 msg_clr_eos();
|
|
|
4938 msg_didout = FALSE;
|
|
|
4939 msg_col = 0;
|
|
|
4940 out_flush();
|
|
|
4941 }
|
|
|
4942 wordtree_compress(spin, root);
|
|
469
|
4943 }
|
|
255
|
4944
|
|
|
4945 return OK;
|
|
236
|
4946 }
|
|
|
4947
|
|
469
|
4948 /*
|
|
|
4949 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
4950 * allocate a new one.
|
|
|
4951 */
|
|
|
4952 static wordnode_T *
|
|
474
|
4953 get_wordnode(spin)
|
|
|
4954 spellinfo_T *spin;
|
|
469
|
4955 {
|
|
|
4956 wordnode_T *n;
|
|
|
4957
|
|
474
|
4958 if (spin->si_first_free == NULL)
|
|
|
4959 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
4960 else
|
|
|
4961 {
|
|
474
|
4962 n = spin->si_first_free;
|
|
|
4963 spin->si_first_free = n->wn_child;
|
|
469
|
4964 vim_memset(n, 0, sizeof(wordnode_T));
|
|
|
4965 }
|
|
|
4966 #ifdef SPELL_PRINTTREE
|
|
474
|
4967 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
4968 #endif
|
|
|
4969 return n;
|
|
|
4970 }
|
|
|
4971
|
|
|
4972 /*
|
|
474
|
4973 * Decrement the reference count on a node (which is the head of a list of
|
|
|
4974 * siblings). If the reference count becomes zero free the node and its
|
|
|
4975 * siblings.
|
|
|
4976 */
|
|
|
4977 static void
|
|
|
4978 deref_wordnode(spin, node)
|
|
|
4979 spellinfo_T *spin;
|
|
|
4980 wordnode_T *node;
|
|
|
4981 {
|
|
|
4982 wordnode_T *np;
|
|
|
4983
|
|
|
4984 if (--node->wn_refs == 0)
|
|
|
4985 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4986 {
|
|
|
4987 if (np->wn_child != NULL)
|
|
|
4988 deref_wordnode(spin, np->wn_child);
|
|
|
4989 free_wordnode(spin, np);
|
|
|
4990 }
|
|
|
4991 }
|
|
|
4992
|
|
|
4993 /*
|
|
469
|
4994 * Free a wordnode_T for re-use later.
|
|
474
|
4995 * Only the "wn_child" field becomes invalid.
|
|
469
|
4996 */
|
|
|
4997 static void
|
|
474
|
4998 free_wordnode(spin, n)
|
|
|
4999 spellinfo_T *spin;
|
|
|
5000 wordnode_T *n;
|
|
|
5001 {
|
|
|
5002 n->wn_child = spin->si_first_free;
|
|
|
5003 spin->si_first_free = n;
|
|
469
|
5004 }
|
|
|
5005
|
|
236
|
5006 /*
|
|
300
|
5007 * Compress a tree: find tails that are identical and can be shared.
|
|
|
5008 */
|
|
|
5009 static void
|
|
474
|
5010 wordtree_compress(spin, root)
|
|
|
5011 spellinfo_T *spin;
|
|
300
|
5012 wordnode_T *root;
|
|
|
5013 {
|
|
|
5014 hashtab_T ht;
|
|
|
5015 int n;
|
|
|
5016 int tot = 0;
|
|
469
|
5017 int perc;
|
|
300
|
5018
|
|
474
|
5019 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
5020 * start of the tree. */
|
|
|
5021 if (root->wn_sibling != NULL)
|
|
300
|
5022 {
|
|
|
5023 hash_init(&ht);
|
|
474
|
5024 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
5025
|
|
|
5026 #ifndef SPELL_PRINTTREE
|
|
310
|
5027 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
5028 #endif
|
|
310
|
5029 {
|
|
|
5030 if (!spin->si_verbose)
|
|
|
5031 verbose_enter();
|
|
469
|
5032 if (tot > 1000000)
|
|
|
5033 perc = (tot - n) / (tot / 100);
|
|
|
5034 else
|
|
|
5035 perc = (tot - n) * 100 / tot;
|
|
310
|
5036 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
5037 n, tot, perc);
|
|
310
|
5038 if (p_verbose > 2)
|
|
|
5039 verbose_leave();
|
|
|
5040 }
|
|
469
|
5041 #ifdef SPELL_PRINTTREE
|
|
474
|
5042 spell_print_tree(root->wn_sibling);
|
|
469
|
5043 #endif
|
|
300
|
5044 hash_clear(&ht);
|
|
|
5045 }
|
|
|
5046 }
|
|
|
5047
|
|
|
5048 /*
|
|
|
5049 * Compress a node, its siblings and its children, depth first.
|
|
|
5050 * Returns the number of compressed nodes.
|
|
236
|
5051 */
|
|
255
|
5052 static int
|
|
474
|
5053 node_compress(spin, node, ht, tot)
|
|
|
5054 spellinfo_T *spin;
|
|
300
|
5055 wordnode_T *node;
|
|
|
5056 hashtab_T *ht;
|
|
|
5057 int *tot; /* total count of nodes before compressing,
|
|
|
5058 incremented while going through the tree */
|
|
236
|
5059 {
|
|
300
|
5060 wordnode_T *np;
|
|
|
5061 wordnode_T *tp;
|
|
|
5062 wordnode_T *child;
|
|
|
5063 hash_T hash;
|
|
236
|
5064 hashitem_T *hi;
|
|
300
|
5065 int len = 0;
|
|
|
5066 unsigned nr, n;
|
|
|
5067 int compressed = 0;
|
|
236
|
5068
|
|
300
|
5069 /*
|
|
|
5070 * Go through the list of siblings. Compress each child and then try
|
|
|
5071 * finding an identical child to replace it.
|
|
|
5072 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
5073 * but the whole list of siblings, of which the node is the first.
|
|
|
5074 */
|
|
|
5075 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5076 {
|
|
300
|
5077 ++len;
|
|
|
5078 if ((child = np->wn_child) != NULL)
|
|
|
5079 {
|
|
346
|
5080 /* Compress the child. This fills hashkey. */
|
|
474
|
5081 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
5082
|
|
|
5083 /* Try to find an identical child. */
|
|
346
|
5084 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
5085 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
5086 tp = NULL;
|
|
|
5087 if (!HASHITEM_EMPTY(hi))
|
|
|
5088 {
|
|
|
5089 /* There are children with an identical hash value. Now check
|
|
|
5090 * if there is one that is really identical. */
|
|
346
|
5091 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
5092 if (node_equal(child, tp))
|
|
|
5093 {
|
|
|
5094 /* Found one! Now use that child in place of the
|
|
469
|
5095 * current one. This means the current child and all
|
|
|
5096 * its siblings is unlinked from the tree. */
|
|
|
5097 ++tp->wn_refs;
|
|
474
|
5098 deref_wordnode(spin, child);
|
|
300
|
5099 np->wn_child = tp;
|
|
|
5100 ++compressed;
|
|
|
5101 break;
|
|
|
5102 }
|
|
|
5103 if (tp == NULL)
|
|
|
5104 {
|
|
|
5105 /* No other child with this hash value equals the child of
|
|
|
5106 * the node, add it to the linked list after the first
|
|
|
5107 * item. */
|
|
|
5108 tp = HI2WN(hi);
|
|
346
|
5109 child->wn_u2.next = tp->wn_u2.next;
|
|
|
5110 tp->wn_u2.next = child;
|
|
300
|
5111 }
|
|
|
5112 }
|
|
|
5113 else
|
|
|
5114 /* No other child has this hash value, add it to the
|
|
|
5115 * hashtable. */
|
|
346
|
5116 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
5117 }
|
|
236
|
5118 }
|
|
300
|
5119 *tot += len;
|
|
|
5120
|
|
|
5121 /*
|
|
|
5122 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
5123 * find a lookalike node. This must be done after compressing the sibling
|
|
|
5124 * list, otherwise the hash key would become invalid by the compression.
|
|
|
5125 */
|
|
346
|
5126 node->wn_u1.hashkey[0] = len;
|
|
300
|
5127 nr = 0;
|
|
|
5128 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5129 {
|
|
300
|
5130 if (np->wn_byte == NUL)
|
|
339
|
5131 /* end node: use wn_flags, wn_region and wn_prefixID */
|
|
|
5132 n = np->wn_flags + (np->wn_region << 8) + (np->wn_prefixID << 16);
|
|
300
|
5133 else
|
|
|
5134 /* byte node: use the byte value and the child pointer */
|
|
|
5135 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
5136 nr = nr * 101 + n;
|
|
236
|
5137 }
|
|
300
|
5138
|
|
|
5139 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
5140 n = nr & 0xff;
|
|
346
|
5141 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
5142 n = (nr >> 8) & 0xff;
|
|
346
|
5143 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
5144 n = (nr >> 16) & 0xff;
|
|
346
|
5145 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
5146 n = (nr >> 24) & 0xff;
|
|
346
|
5147 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
5148 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
5149
|
|
|
5150 return compressed;
|
|
|
5151 }
|
|
|
5152
|
|
|
5153 /*
|
|
|
5154 * Return TRUE when two nodes have identical siblings and children.
|
|
|
5155 */
|
|
|
5156 static int
|
|
|
5157 node_equal(n1, n2)
|
|
|
5158 wordnode_T *n1;
|
|
|
5159 wordnode_T *n2;
|
|
|
5160 {
|
|
|
5161 wordnode_T *p1;
|
|
|
5162 wordnode_T *p2;
|
|
|
5163
|
|
|
5164 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
5165 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
5166 if (p1->wn_byte != p2->wn_byte
|
|
|
5167 || (p1->wn_byte == NUL
|
|
|
5168 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
5169 || p1->wn_region != p2->wn_region
|
|
|
5170 || p1->wn_prefixID != p2->wn_prefixID)
|
|
300
|
5171 : (p1->wn_child != p2->wn_child)))
|
|
|
5172 break;
|
|
|
5173
|
|
|
5174 return p1 == NULL && p2 == NULL;
|
|
236
|
5175 }
|
|
|
5176
|
|
|
5177 /*
|
|
|
5178 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
5179 */
|
|
255
|
5180 void
|
|
236
|
5181 put_bytes(fd, nr, len)
|
|
|
5182 FILE *fd;
|
|
|
5183 long_u nr;
|
|
|
5184 int len;
|
|
|
5185 {
|
|
|
5186 int i;
|
|
|
5187
|
|
|
5188 for (i = len - 1; i >= 0; --i)
|
|
|
5189 putc((int)(nr >> (i * 8)), fd);
|
|
|
5190 }
|
|
|
5191
|
|
323
|
5192 static int
|
|
|
5193 #ifdef __BORLANDC__
|
|
|
5194 _RTLENTRYF
|
|
|
5195 #endif
|
|
|
5196 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
5197
|
|
|
5198 /*
|
|
|
5199 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
5200 */
|
|
|
5201 static int
|
|
|
5202 #ifdef __BORLANDC__
|
|
|
5203 _RTLENTRYF
|
|
|
5204 #endif
|
|
|
5205 rep_compare(s1, s2)
|
|
|
5206 const void *s1;
|
|
|
5207 const void *s2;
|
|
|
5208 {
|
|
|
5209 fromto_T *p1 = (fromto_T *)s1;
|
|
|
5210 fromto_T *p2 = (fromto_T *)s2;
|
|
|
5211
|
|
|
5212 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
5213 }
|
|
|
5214
|
|
236
|
5215 /*
|
|
|
5216 * Write the Vim spell file "fname".
|
|
|
5217 */
|
|
|
5218 static void
|
|
474
|
5219 write_vim_spell(spin, fname)
|
|
|
5220 spellinfo_T *spin;
|
|
236
|
5221 char_u *fname;
|
|
|
5222 {
|
|
300
|
5223 FILE *fd;
|
|
|
5224 int regionmask;
|
|
236
|
5225 int round;
|
|
300
|
5226 wordnode_T *tree;
|
|
|
5227 int nodecount;
|
|
323
|
5228 int i;
|
|
|
5229 int l;
|
|
|
5230 garray_T *gap;
|
|
|
5231 fromto_T *ftp;
|
|
|
5232 char_u *p;
|
|
|
5233 int rr;
|
|
236
|
5234
|
|
310
|
5235 fd = mch_fopen((char *)fname, "w");
|
|
300
|
5236 if (fd == NULL)
|
|
236
|
5237 {
|
|
|
5238 EMSG2(_(e_notopen), fname);
|
|
|
5239 return;
|
|
|
5240 }
|
|
|
5241
|
|
255
|
5242 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
339
|
5243 * <charflagslen> <charflags>
|
|
|
5244 * <fcharslen> <fchars>
|
|
366
|
5245 * <midwordlen> <midword>
|
|
339
|
5246 * <prefcondcnt> <prefcond> ... */
|
|
300
|
5247
|
|
|
5248 /* <fileID> */
|
|
|
5249 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
5250 EMSG(_(e_write));
|
|
236
|
5251
|
|
|
5252 /* write the region names if there is more than one */
|
|
316
|
5253 if (spin->si_region_count > 1)
|
|
236
|
5254 {
|
|
316
|
5255 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
|
5256 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
|
5257 (size_t)1, fd);
|
|
|
5258 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
5259 }
|
|
|
5260 else
|
|
|
5261 {
|
|
300
|
5262 putc(0, fd);
|
|
|
5263 regionmask = 0;
|
|
236
|
5264 }
|
|
|
5265
|
|
323
|
5266 /*
|
|
|
5267 * Write the table with character flags and table for case folding.
|
|
260
|
5268 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
|
5269 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
5270 * 'encoding'.
|
|
|
5271 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
5272 * the table (avoids that it conflicts). File is shorter too.
|
|
|
5273 */
|
|
|
5274 if (spin->si_ascii || spin->si_add)
|
|
260
|
5275 {
|
|
300
|
5276 putc(0, fd);
|
|
|
5277 putc(0, fd);
|
|
|
5278 putc(0, fd);
|
|
260
|
5279 }
|
|
|
5280 else
|
|
300
|
5281 write_spell_chartab(fd);
|
|
255
|
5282
|
|
366
|
5283
|
|
|
5284 if (spin->si_midword == NULL)
|
|
|
5285 put_bytes(fd, 0L, 2); /* <midwordlen> */
|
|
|
5286 else
|
|
|
5287 {
|
|
|
5288 i = STRLEN(spin->si_midword);
|
|
|
5289 put_bytes(fd, (long_u)i, 2); /* <midwordlen> */
|
|
|
5290 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
5291 }
|
|
|
5292
|
|
|
5293
|
|
339
|
5294 /* Write the prefix conditions. */
|
|
|
5295 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
5296
|
|
366
|
5297 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
5298 * <salflags> <salcount> <sal> ...
|
|
|
5299 * <maplen> <mapstr> */
|
|
|
5300
|
|
323
|
5301 /* Sort the REP items. */
|
|
|
5302 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
5303 sizeof(fromto_T), rep_compare);
|
|
|
5304
|
|
375
|
5305 /* round 1: REP items
|
|
|
5306 * round 2: SAL items (unless SOFO is used) */
|
|
323
|
5307 for (round = 1; round <= 2; ++round)
|
|
|
5308 {
|
|
|
5309 if (round == 1)
|
|
|
5310 gap = &spin->si_rep;
|
|
|
5311 else
|
|
|
5312 {
|
|
|
5313 gap = &spin->si_sal;
|
|
|
5314
|
|
|
5315 i = 0;
|
|
|
5316 if (spin->si_followup)
|
|
|
5317 i |= SAL_F0LLOWUP;
|
|
|
5318 if (spin->si_collapse)
|
|
|
5319 i |= SAL_COLLAPSE;
|
|
|
5320 if (spin->si_rem_accents)
|
|
|
5321 i |= SAL_REM_ACCENTS;
|
|
375
|
5322 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5323 i |= SAL_SOFO;
|
|
323
|
5324 putc(i, fd); /* <salflags> */
|
|
375
|
5325 if (i & SAL_SOFO)
|
|
|
5326 break;
|
|
323
|
5327 }
|
|
|
5328
|
|
|
5329 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
5330 for (i = 0; i < gap->ga_len; ++i)
|
|
|
5331 {
|
|
|
5332 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
5333 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
5334 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
5335 for (rr = 1; rr <= 2; ++rr)
|
|
|
5336 {
|
|
|
5337 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
5338 l = STRLEN(p);
|
|
|
5339 putc(l, fd);
|
|
|
5340 fwrite(p, l, (size_t)1, fd);
|
|
|
5341 }
|
|
|
5342 }
|
|
|
5343 }
|
|
|
5344
|
|
375
|
5345 /* SOFOFROM and SOFOTO */
|
|
|
5346 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5347 {
|
|
|
5348 put_bytes(fd, 1L, 2); /* <salcount> */
|
|
|
5349
|
|
|
5350 l = STRLEN(spin->si_sofofr);
|
|
|
5351 put_bytes(fd, (long_u)l, 2); /* <salfromlen> */
|
|
|
5352 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <salfrom> */
|
|
|
5353
|
|
|
5354 l = STRLEN(spin->si_sofoto);
|
|
|
5355 put_bytes(fd, (long_u)l, 2); /* <saltolen> */
|
|
|
5356 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <salto> */
|
|
|
5357 }
|
|
|
5358
|
|
323
|
5359 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
|
5360 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
|
5361 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
|
5362 (size_t)1, fd);
|
|
302
|
5363
|
|
236
|
5364 /*
|
|
339
|
5365 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
5366 */
|
|
323
|
5367 spin->si_memtot = 0;
|
|
339
|
5368 for (round = 1; round <= 3; ++round)
|
|
236
|
5369 {
|
|
339
|
5370 if (round == 1)
|
|
469
|
5371 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
5372 else if (round == 2)
|
|
469
|
5373 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
5374 else
|
|
469
|
5375 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
5376
|
|
346
|
5377 /* Clear the index and wnode fields in the tree. */
|
|
|
5378 clear_node(tree);
|
|
|
5379
|
|
300
|
5380 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
5381 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
5382 * nodes. */
|
|
346
|
5383 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
5384
|
|
300
|
5385 /* number of nodes in 4 bytes */
|
|
|
5386 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
5387 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
5388
|
|
300
|
5389 /* Write the nodes. */
|
|
346
|
5390 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
5391 }
|
|
|
5392
|
|
300
|
5393 fclose(fd);
|
|
236
|
5394 }
|
|
|
5395
|
|
|
5396 /*
|
|
346
|
5397 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
5398 * children. This is needed because they are a union with other items to save
|
|
|
5399 * space.
|
|
|
5400 */
|
|
|
5401 static void
|
|
|
5402 clear_node(node)
|
|
|
5403 wordnode_T *node;
|
|
|
5404 {
|
|
|
5405 wordnode_T *np;
|
|
|
5406
|
|
|
5407 if (node != NULL)
|
|
|
5408 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5409 {
|
|
|
5410 np->wn_u1.index = 0;
|
|
|
5411 np->wn_u2.wnode = NULL;
|
|
|
5412
|
|
|
5413 if (np->wn_byte != NUL)
|
|
|
5414 clear_node(np->wn_child);
|
|
|
5415 }
|
|
|
5416 }
|
|
|
5417
|
|
|
5418
|
|
|
5419 /*
|
|
300
|
5420 * Dump a word tree at node "node".
|
|
|
5421 *
|
|
|
5422 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
5423 * byte recursively write the children.
|
|
|
5424 *
|
|
|
5425 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
5426 * are made about the indexes (so that we don't have to write them in the
|
|
|
5427 * file).
|
|
236
|
5428 *
|
|
300
|
5429 * Returns the number of nodes used.
|
|
236
|
5430 */
|
|
300
|
5431 static int
|
|
346
|
5432 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
5433 FILE *fd; /* NULL when only counting */
|
|
300
|
5434 wordnode_T *node;
|
|
|
5435 int index;
|
|
|
5436 int regionmask;
|
|
339
|
5437 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
5438 {
|
|
300
|
5439 int newindex = index;
|
|
|
5440 int siblingcount = 0;
|
|
|
5441 wordnode_T *np;
|
|
236
|
5442 int flags;
|
|
300
|
5443
|
|
|
5444 /* If "node" is zero the tree is empty. */
|
|
|
5445 if (node == NULL)
|
|
|
5446 return 0;
|
|
|
5447
|
|
|
5448 /* Store the index where this node is written. */
|
|
346
|
5449 node->wn_u1.index = index;
|
|
236
|
5450
|
|
300
|
5451 /* Count the number of siblings. */
|
|
|
5452 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5453 ++siblingcount;
|
|
236
|
5454
|
|
300
|
5455 /* Write the sibling count. */
|
|
|
5456 if (fd != NULL)
|
|
|
5457 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
5458
|
|
300
|
5459 /* Write each sibling byte and optionally extra info. */
|
|
|
5460 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5461 {
|
|
300
|
5462 if (np->wn_byte == 0)
|
|
|
5463 {
|
|
|
5464 if (fd != NULL)
|
|
|
5465 {
|
|
339
|
5466 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
5467 if (prefixtree)
|
|
300
|
5468 {
|
|
339
|
5469 /* In PREFIXTREE write the required prefixID and the
|
|
390
|
5470 * associated condition nr (stored in wn_region). The
|
|
|
5471 * byte value is misused to store the "rare" and "not
|
|
|
5472 * combining" flags */
|
|
455
|
5473 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
5474 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5475 else
|
|
|
5476 {
|
|
390
|
5477 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
5478 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
5479 }
|
|
339
|
5480 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
5481 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
5482 }
|
|
|
5483 else
|
|
|
5484 {
|
|
339
|
5485 /* For word trees we write the flag/region items. */
|
|
|
5486 flags = np->wn_flags;
|
|
|
5487 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
5488 flags |= WF_REGION;
|
|
|
5489 if (np->wn_prefixID != 0)
|
|
|
5490 flags |= WF_PFX;
|
|
|
5491 if (flags == 0)
|
|
|
5492 {
|
|
|
5493 /* word without flags or region */
|
|
|
5494 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5495 }
|
|
|
5496 else
|
|
|
5497 {
|
|
390
|
5498 if (np->wn_flags >= 0x100)
|
|
|
5499 {
|
|
|
5500 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
5501 putc(flags, fd); /* <flags> */
|
|
|
5502 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
5503 }
|
|
|
5504 else
|
|
|
5505 {
|
|
|
5506 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
5507 putc(flags, fd); /* <flags> */
|
|
|
5508 }
|
|
339
|
5509 if (flags & WF_REGION)
|
|
|
5510 putc(np->wn_region, fd); /* <region> */
|
|
|
5511 if (flags & WF_PFX)
|
|
|
5512 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
5513 }
|
|
300
|
5514 }
|
|
|
5515 }
|
|
|
5516 }
|
|
|
5517 else
|
|
|
5518 {
|
|
346
|
5519 if (np->wn_child->wn_u1.index != 0
|
|
|
5520 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
5521 {
|
|
|
5522 /* The child is written elsewhere, write the reference. */
|
|
|
5523 if (fd != NULL)
|
|
|
5524 {
|
|
|
5525 putc(BY_INDEX, fd); /* <byte> */
|
|
|
5526 /* <nodeidx> */
|
|
346
|
5527 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
5528 }
|
|
|
5529 }
|
|
346
|
5530 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
5531 /* We will write the child below and give it an index. */
|
|
346
|
5532 np->wn_child->wn_u2.wnode = node;
|
|
236
|
5533
|
|
300
|
5534 if (fd != NULL)
|
|
|
5535 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
5536 {
|
|
|
5537 EMSG(_(e_write));
|
|
|
5538 return 0;
|
|
|
5539 }
|
|
|
5540 }
|
|
236
|
5541 }
|
|
|
5542
|
|
300
|
5543 /* Space used in the array when reading: one for each sibling and one for
|
|
|
5544 * the count. */
|
|
|
5545 newindex += siblingcount + 1;
|
|
249
|
5546
|
|
300
|
5547 /* Recursively dump the children of each sibling. */
|
|
|
5548 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
5549 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
5550 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
5551 prefixtree);
|
|
249
|
5552
|
|
300
|
5553 return newindex;
|
|
236
|
5554 }
|
|
|
5555
|
|
|
5556
|
|
|
5557 /*
|
|
310
|
5558 * ":mkspell [-ascii] outfile infile ..."
|
|
|
5559 * ":mkspell [-ascii] addfile"
|
|
236
|
5560 */
|
|
|
5561 void
|
|
|
5562 ex_mkspell(eap)
|
|
|
5563 exarg_T *eap;
|
|
|
5564 {
|
|
|
5565 int fcount;
|
|
|
5566 char_u **fnames;
|
|
310
|
5567 char_u *arg = eap->arg;
|
|
|
5568 int ascii = FALSE;
|
|
|
5569
|
|
|
5570 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
5571 {
|
|
|
5572 ascii = TRUE;
|
|
|
5573 arg = skipwhite(arg + 6);
|
|
|
5574 }
|
|
|
5575
|
|
|
5576 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
5577 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
5578 {
|
|
323
|
5579 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
5580 FreeWild(fcount, fnames);
|
|
|
5581 }
|
|
|
5582 }
|
|
|
5583
|
|
|
5584 /*
|
|
|
5585 * Create a Vim spell file from one or more word lists.
|
|
|
5586 * "fnames[0]" is the output file name.
|
|
|
5587 * "fnames[fcount - 1]" is the last input file name.
|
|
|
5588 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
5589 * and ".spl" is appended to make the output file name.
|
|
|
5590 */
|
|
|
5591 static void
|
|
323
|
5592 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
5593 int fcount;
|
|
|
5594 char_u **fnames;
|
|
|
5595 int ascii; /* -ascii argument given */
|
|
|
5596 int overwrite; /* overwrite existing output file */
|
|
323
|
5597 int added_word; /* invoked through "zg" */
|
|
310
|
5598 {
|
|
236
|
5599 char_u fname[MAXPATHL];
|
|
|
5600 char_u wfname[MAXPATHL];
|
|
310
|
5601 char_u **innames;
|
|
|
5602 int incount;
|
|
236
|
5603 afffile_T *(afile[8]);
|
|
|
5604 int i;
|
|
|
5605 int len;
|
|
|
5606 struct stat st;
|
|
255
|
5607 int error = FALSE;
|
|
300
|
5608 spellinfo_T spin;
|
|
|
5609
|
|
|
5610 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
5611 spin.si_verbose = !added_word;
|
|
310
|
5612 spin.si_ascii = ascii;
|
|
323
|
5613 spin.si_followup = TRUE;
|
|
|
5614 spin.si_rem_accents = TRUE;
|
|
|
5615 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
5616 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
5617 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
5618 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
5619
|
|
|
5620 /* default: fnames[0] is output file, following are input files */
|
|
|
5621 innames = &fnames[1];
|
|
|
5622 incount = fcount - 1;
|
|
|
5623
|
|
|
5624 if (fcount >= 1)
|
|
240
|
5625 {
|
|
310
|
5626 len = STRLEN(fnames[0]);
|
|
|
5627 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
5628 {
|
|
|
5629 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
5630 * "path/en.latin1.add.spl". */
|
|
|
5631 innames = &fnames[0];
|
|
|
5632 incount = 1;
|
|
|
5633 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
5634 }
|
|
366
|
5635 else if (fcount == 1)
|
|
|
5636 {
|
|
|
5637 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
5638 innames = &fnames[0];
|
|
|
5639 incount = 1;
|
|
|
5640 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5641 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5642 }
|
|
310
|
5643 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
5644 {
|
|
|
5645 /* Name ends in ".spl", use as the file name. */
|
|
323
|
5646 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
5647 }
|
|
|
5648 else
|
|
|
5649 /* Name should be language, make the file name from it. */
|
|
|
5650 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5651 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5652
|
|
|
5653 /* Check for .ascii.spl. */
|
|
|
5654 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
5655 spin.si_ascii = TRUE;
|
|
|
5656
|
|
|
5657 /* Check for .add.spl. */
|
|
|
5658 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
5659 spin.si_add = TRUE;
|
|
240
|
5660 }
|
|
|
5661
|
|
310
|
5662 if (incount <= 0)
|
|
236
|
5663 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
5664 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
5665 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
5666 else if (incount > 8)
|
|
236
|
5667 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
5668 else
|
|
|
5669 {
|
|
|
5670 /* Check for overwriting before doing things that may take a lot of
|
|
|
5671 * time. */
|
|
310
|
5672 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
5673 {
|
|
|
5674 EMSG(_(e_exists));
|
|
310
|
5675 return;
|
|
236
|
5676 }
|
|
310
|
5677 if (mch_isdir(wfname))
|
|
236
|
5678 {
|
|
310
|
5679 EMSG2(_(e_isadir2), wfname);
|
|
|
5680 return;
|
|
236
|
5681 }
|
|
|
5682
|
|
|
5683 /*
|
|
|
5684 * Init the aff and dic pointers.
|
|
|
5685 * Get the region names if there are more than 2 arguments.
|
|
|
5686 */
|
|
310
|
5687 for (i = 0; i < incount; ++i)
|
|
236
|
5688 {
|
|
310
|
5689 afile[i] = NULL;
|
|
300
|
5690
|
|
316
|
5691 if (incount > 1)
|
|
236
|
5692 {
|
|
310
|
5693 len = STRLEN(innames[i]);
|
|
|
5694 if (STRLEN(gettail(innames[i])) < 5
|
|
|
5695 || innames[i][len - 3] != '_')
|
|
236
|
5696 {
|
|
310
|
5697 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
5698 return;
|
|
236
|
5699 }
|
|
316
|
5700 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
5701 spin.si_region_name[i * 2 + 1] =
|
|
|
5702 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
5703 }
|
|
|
5704 }
|
|
316
|
5705 spin.si_region_count = incount;
|
|
236
|
5706
|
|
474
|
5707 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
5708 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
5709 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
5710 if (spin.si_foldroot == NULL
|
|
|
5711 || spin.si_keeproot == NULL
|
|
|
5712 || spin.si_prefroot == NULL)
|
|
300
|
5713 {
|
|
469
|
5714 free_blocks(spin.si_blocks);
|
|
310
|
5715 return;
|
|
300
|
5716 }
|
|
|
5717
|
|
351
|
5718 /* When not producing a .add.spl file clear the character table when
|
|
|
5719 * we encounter one in the .aff file. This means we dump the current
|
|
|
5720 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
5721 * better than guessing the contents, the table will match a
|
|
|
5722 * previously loaded spell file. */
|
|
|
5723 if (!spin.si_add)
|
|
|
5724 spin.si_clear_chartab = TRUE;
|
|
|
5725
|
|
236
|
5726 /*
|
|
|
5727 * Read all the .aff and .dic files.
|
|
|
5728 * Text is converted to 'encoding'.
|
|
300
|
5729 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
5730 */
|
|
310
|
5731 for (i = 0; i < incount && !error; ++i)
|
|
236
|
5732 {
|
|
300
|
5733 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
5734 spin.si_region = 1 << i;
|
|
|
5735
|
|
|
5736 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
5737 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
5738 {
|
|
|
5739 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
5740 * "SET" line. */
|
|
474
|
5741 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
5742 if (afile[i] == NULL)
|
|
300
|
5743 error = TRUE;
|
|
|
5744 else
|
|
|
5745 {
|
|
|
5746 /* Read the .dic file and store the words in the trees. */
|
|
|
5747 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
5748 innames[i]);
|
|
474
|
5749 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
5750 error = TRUE;
|
|
|
5751 }
|
|
|
5752 }
|
|
|
5753 else
|
|
|
5754 {
|
|
|
5755 /* No .aff file, try reading the file as a word list. Store
|
|
|
5756 * the words in the trees. */
|
|
474
|
5757 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
5758 error = TRUE;
|
|
|
5759 }
|
|
236
|
5760
|
|
310
|
5761 #ifdef FEAT_MBYTE
|
|
236
|
5762 /* Free any conversion stuff. */
|
|
300
|
5763 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
5764 #endif
|
|
236
|
5765 }
|
|
|
5766
|
|
300
|
5767 if (!error)
|
|
236
|
5768 {
|
|
|
5769 /*
|
|
300
|
5770 * Combine tails in the tree.
|
|
236
|
5771 */
|
|
474
|
5772 if (spin.si_verbose || p_verbose > 2)
|
|
|
5773 {
|
|
|
5774 if (!spin.si_verbose)
|
|
310
|
5775 verbose_enter();
|
|
469
|
5776 MSG(_(msg_compressing));
|
|
310
|
5777 out_flush();
|
|
474
|
5778 if (!spin.si_verbose)
|
|
310
|
5779 verbose_leave();
|
|
|
5780 }
|
|
474
|
5781 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
5782 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
5783 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
5784 }
|
|
|
5785
|
|
300
|
5786 if (!error)
|
|
|
5787 {
|
|
|
5788 /*
|
|
|
5789 * Write the info in the spell file.
|
|
|
5790 */
|
|
474
|
5791 if (spin.si_verbose || p_verbose > 2)
|
|
|
5792 {
|
|
|
5793 if (!spin.si_verbose)
|
|
310
|
5794 verbose_enter();
|
|
366
|
5795 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
5796 out_flush();
|
|
474
|
5797 if (!spin.si_verbose)
|
|
310
|
5798 verbose_leave();
|
|
|
5799 }
|
|
|
5800
|
|
474
|
5801 write_vim_spell(&spin, wfname);
|
|
|
5802
|
|
|
5803 if (spin.si_verbose || p_verbose > 2)
|
|
|
5804 {
|
|
|
5805 if (!spin.si_verbose)
|
|
310
|
5806 verbose_enter();
|
|
|
5807 MSG(_("Done!"));
|
|
|
5808 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
5809 spin.si_memtot);
|
|
310
|
5810 out_flush();
|
|
474
|
5811 if (!spin.si_verbose)
|
|
310
|
5812 verbose_leave();
|
|
|
5813 }
|
|
|
5814
|
|
|
5815 /* If the file is loaded need to reload it. */
|
|
323
|
5816 spell_reload_one(wfname, added_word);
|
|
300
|
5817 }
|
|
|
5818
|
|
|
5819 /* Free the allocated memory. */
|
|
323
|
5820 ga_clear(&spin.si_rep);
|
|
|
5821 ga_clear(&spin.si_sal);
|
|
|
5822 ga_clear(&spin.si_map);
|
|
339
|
5823 ga_clear(&spin.si_prefcond);
|
|
366
|
5824 vim_free(spin.si_midword);
|
|
375
|
5825 vim_free(spin.si_sofofr);
|
|
|
5826 vim_free(spin.si_sofoto);
|
|
300
|
5827
|
|
|
5828 /* Free the .aff file structures. */
|
|
310
|
5829 for (i = 0; i < incount; ++i)
|
|
|
5830 if (afile[i] != NULL)
|
|
|
5831 spell_free_aff(afile[i]);
|
|
339
|
5832
|
|
|
5833 /* Free all the bits and pieces at once. */
|
|
|
5834 free_blocks(spin.si_blocks);
|
|
236
|
5835 }
|
|
310
|
5836 }
|
|
|
5837
|
|
|
5838
|
|
|
5839 /*
|
|
385
|
5840 * ":[count]spellgood {word}"
|
|
|
5841 * ":[count]spellwrong {word}"
|
|
310
|
5842 */
|
|
|
5843 void
|
|
|
5844 ex_spell(eap)
|
|
|
5845 exarg_T *eap;
|
|
|
5846 {
|
|
381
|
5847 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
5848 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
5849 }
|
|
|
5850
|
|
310
|
5851 /*
|
|
|
5852 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
5853 */
|
|
|
5854 void
|
|
385
|
5855 spell_add_word(word, len, bad, index)
|
|
310
|
5856 char_u *word;
|
|
|
5857 int len;
|
|
|
5858 int bad;
|
|
385
|
5859 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
5860 'spellfile' */
|
|
310
|
5861 {
|
|
|
5862 FILE *fd;
|
|
385
|
5863 buf_T *buf = NULL;
|
|
351
|
5864 int new_spf = FALSE;
|
|
|
5865 struct stat st;
|
|
381
|
5866 char_u *fname;
|
|
385
|
5867 char_u fnamebuf[MAXPATHL];
|
|
|
5868 char_u line[MAXWLEN * 2];
|
|
|
5869 long fpos, fpos_next = 0;
|
|
|
5870 int i;
|
|
|
5871 char_u *spf;
|
|
|
5872
|
|
|
5873 if (index == 0) /* use internal wordlist */
|
|
|
5874 {
|
|
|
5875 if (int_wordlist == NULL)
|
|
|
5876 {
|
|
|
5877 int_wordlist = vim_tempname('s');
|
|
|
5878 if (int_wordlist == NULL)
|
|
381
|
5879 return;
|
|
|
5880 }
|
|
385
|
5881 fname = int_wordlist;
|
|
381
|
5882 }
|
|
310
|
5883 else
|
|
|
5884 {
|
|
381
|
5885 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
5886 if (*curbuf->b_p_spf == NUL)
|
|
|
5887 {
|
|
|
5888 init_spellfile();
|
|
|
5889 new_spf = TRUE;
|
|
|
5890 }
|
|
|
5891
|
|
|
5892 if (*curbuf->b_p_spf == NUL)
|
|
|
5893 {
|
|
|
5894 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
5895 return;
|
|
|
5896 }
|
|
|
5897
|
|
385
|
5898 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
5899 {
|
|
|
5900 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
5901 if (i == index)
|
|
|
5902 break;
|
|
|
5903 if (*spf == NUL)
|
|
|
5904 {
|
|
|
5905 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
5906 return;
|
|
|
5907 }
|
|
|
5908 }
|
|
|
5909
|
|
310
|
5910 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
5911 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
5912 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
5913 buf = NULL;
|
|
|
5914 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
5915 {
|
|
310
|
5916 EMSG(_(e_bufloaded));
|
|
381
|
5917 return;
|
|
|
5918 }
|
|
|
5919
|
|
385
|
5920 fname = fnamebuf;
|
|
|
5921 }
|
|
|
5922
|
|
|
5923 if (bad)
|
|
|
5924 {
|
|
|
5925 /* When the word also appears as good word we need to remove that one,
|
|
|
5926 * since its flags sort before the one with WF_BANNED. */
|
|
|
5927 fd = mch_fopen((char *)fname, "r");
|
|
|
5928 if (fd != NULL)
|
|
|
5929 {
|
|
|
5930 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
5931 {
|
|
|
5932 fpos = fpos_next;
|
|
|
5933 fpos_next = ftell(fd);
|
|
|
5934 if (STRNCMP(word, line, len) == 0
|
|
|
5935 && (line[len] == '/' || line[len] < ' '))
|
|
|
5936 {
|
|
|
5937 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
5938 * the start of the line. Mixing reading and writing
|
|
|
5939 * doesn't work for all systems, close the file first. */
|
|
|
5940 fclose(fd);
|
|
|
5941 fd = mch_fopen((char *)fname, "r+");
|
|
|
5942 if (fd == NULL)
|
|
|
5943 break;
|
|
|
5944 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
5945 fputc('#', fd);
|
|
|
5946 fseek(fd, fpos_next, SEEK_SET);
|
|
|
5947 }
|
|
|
5948 }
|
|
|
5949 fclose(fd);
|
|
|
5950 }
|
|
381
|
5951 }
|
|
|
5952
|
|
|
5953 fd = mch_fopen((char *)fname, "a");
|
|
|
5954 if (fd == NULL && new_spf)
|
|
|
5955 {
|
|
|
5956 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
5957 * We may need to create the "spell" directory first. We already
|
|
|
5958 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
5959 STRCPY(NameBuff, fname);
|
|
|
5960 *gettail_sep(NameBuff) = NUL;
|
|
|
5961 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
5962 {
|
|
|
5963 /* The directory doesn't exist. Try creating it and opening the
|
|
|
5964 * file again. */
|
|
|
5965 vim_mkdir(NameBuff, 0755);
|
|
|
5966 fd = mch_fopen((char *)fname, "a");
|
|
|
5967 }
|
|
|
5968 }
|
|
|
5969
|
|
|
5970 if (fd == NULL)
|
|
|
5971 EMSG2(_(e_notopen), fname);
|
|
|
5972 else
|
|
|
5973 {
|
|
|
5974 if (bad)
|
|
|
5975 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
5976 else
|
|
381
|
5977 fprintf(fd, "%.*s\n", len, word);
|
|
|
5978 fclose(fd);
|
|
|
5979
|
|
|
5980 /* Update the .add.spl file. */
|
|
|
5981 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
5982
|
|
|
5983 /* If the .add file is edited somewhere, reload it. */
|
|
|
5984 if (buf != NULL)
|
|
|
5985 buf_reload(buf);
|
|
|
5986
|
|
|
5987 redraw_all_later(NOT_VALID);
|
|
310
|
5988 }
|
|
|
5989 }
|
|
|
5990
|
|
|
5991 /*
|
|
|
5992 * Initialize 'spellfile' for the current buffer.
|
|
|
5993 */
|
|
|
5994 static void
|
|
|
5995 init_spellfile()
|
|
|
5996 {
|
|
|
5997 char_u buf[MAXPATHL];
|
|
|
5998 int l;
|
|
|
5999 slang_T *sl;
|
|
|
6000 char_u *rtp;
|
|
351
|
6001 char_u *lend;
|
|
310
|
6002
|
|
|
6003 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
6004 {
|
|
351
|
6005 /* Find the end of the language name. Exclude the region. */
|
|
|
6006 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
6007 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
6008 ;
|
|
|
6009
|
|
|
6010 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
6011 * are allowed to write. */
|
|
310
|
6012 rtp = p_rtp;
|
|
|
6013 while (*rtp != NUL)
|
|
|
6014 {
|
|
|
6015 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
6016 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
6017 if (filewritable(buf) == 2)
|
|
|
6018 {
|
|
316
|
6019 /* Use the first language name from 'spelllang' and the
|
|
|
6020 * encoding used in the first loaded .spl file. */
|
|
310
|
6021 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
6022 l = STRLEN(buf);
|
|
|
6023 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
6024 "/spell/%.*s.%s.add",
|
|
351
|
6025 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
6026 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
6027 ? (char_u *)"ascii" : spell_enc());
|
|
|
6028 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
6029 break;
|
|
|
6030 }
|
|
|
6031 }
|
|
|
6032 }
|
|
|
6033 }
|
|
236
|
6034
|
|
300
|
6035
|
|
307
|
6036 /*
|
|
|
6037 * Init the chartab used for spelling for ASCII.
|
|
|
6038 * EBCDIC is not supported!
|
|
|
6039 */
|
|
|
6040 static void
|
|
|
6041 clear_spell_chartab(sp)
|
|
|
6042 spelltab_T *sp;
|
|
|
6043 {
|
|
324
|
6044 int i;
|
|
307
|
6045
|
|
|
6046 /* Init everything to FALSE. */
|
|
|
6047 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
6048 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
6049 for (i = 0; i < 256; ++i)
|
|
324
|
6050 {
|
|
307
|
6051 sp->st_fold[i] = i;
|
|
324
|
6052 sp->st_upper[i] = i;
|
|
|
6053 }
|
|
307
|
6054
|
|
|
6055 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
6056 * that is done separately. */
|
|
|
6057 for (i = '0'; i <= '9'; ++i)
|
|
|
6058 sp->st_isw[i] = TRUE;
|
|
|
6059 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
6060 {
|
|
|
6061 sp->st_isw[i] = TRUE;
|
|
|
6062 sp->st_isu[i] = TRUE;
|
|
|
6063 sp->st_fold[i] = i + 0x20;
|
|
|
6064 }
|
|
|
6065 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
6066 {
|
|
307
|
6067 sp->st_isw[i] = TRUE;
|
|
324
|
6068 sp->st_upper[i] = i - 0x20;
|
|
|
6069 }
|
|
307
|
6070 }
|
|
|
6071
|
|
|
6072 /*
|
|
|
6073 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
6074 * Called once while starting up and when 'encoding' changes.
|
|
|
6075 * The default is to use isalpha(), but the spell file should define the word
|
|
|
6076 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
6077 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
6078 */
|
|
|
6079 void
|
|
|
6080 init_spell_chartab()
|
|
|
6081 {
|
|
|
6082 int i;
|
|
|
6083
|
|
|
6084 did_set_spelltab = FALSE;
|
|
|
6085 clear_spell_chartab(&spelltab);
|
|
|
6086 #ifdef FEAT_MBYTE
|
|
|
6087 if (enc_dbcs)
|
|
|
6088 {
|
|
|
6089 /* DBCS: assume double-wide characters are word characters. */
|
|
|
6090 for (i = 128; i <= 255; ++i)
|
|
|
6091 if (MB_BYTE2LEN(i) == 2)
|
|
|
6092 spelltab.st_isw[i] = TRUE;
|
|
|
6093 }
|
|
324
|
6094 else if (enc_utf8)
|
|
|
6095 {
|
|
|
6096 for (i = 128; i < 256; ++i)
|
|
|
6097 {
|
|
|
6098 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
6099 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
6100 spelltab.st_fold[i] = utf_fold(i);
|
|
|
6101 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
6102 }
|
|
|
6103 }
|
|
307
|
6104 else
|
|
|
6105 #endif
|
|
|
6106 {
|
|
324
|
6107 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
6108 for (i = 128; i < 256; ++i)
|
|
|
6109 {
|
|
|
6110 if (MB_ISUPPER(i))
|
|
|
6111 {
|
|
324
|
6112 spelltab.st_isw[i] = TRUE;
|
|
307
|
6113 spelltab.st_isu[i] = TRUE;
|
|
|
6114 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
6115 }
|
|
324
|
6116 else if (MB_ISLOWER(i))
|
|
|
6117 {
|
|
|
6118 spelltab.st_isw[i] = TRUE;
|
|
|
6119 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
6120 }
|
|
307
|
6121 }
|
|
|
6122 }
|
|
|
6123 }
|
|
|
6124
|
|
|
6125 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
6126 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
6127
|
|
|
6128 /*
|
|
|
6129 * Set the spell character tables from strings in the affix file.
|
|
|
6130 */
|
|
|
6131 static int
|
|
|
6132 set_spell_chartab(fol, low, upp)
|
|
|
6133 char_u *fol;
|
|
|
6134 char_u *low;
|
|
|
6135 char_u *upp;
|
|
|
6136 {
|
|
|
6137 /* We build the new tables here first, so that we can compare with the
|
|
|
6138 * previous one. */
|
|
|
6139 spelltab_T new_st;
|
|
|
6140 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
6141 int f, l, u;
|
|
|
6142
|
|
|
6143 clear_spell_chartab(&new_st);
|
|
|
6144
|
|
|
6145 while (*pf != NUL)
|
|
|
6146 {
|
|
|
6147 if (*pl == NUL || *pu == NUL)
|
|
|
6148 {
|
|
|
6149 EMSG(_(e_affform));
|
|
|
6150 return FAIL;
|
|
|
6151 }
|
|
|
6152 #ifdef FEAT_MBYTE
|
|
|
6153 f = mb_ptr2char_adv(&pf);
|
|
|
6154 l = mb_ptr2char_adv(&pl);
|
|
|
6155 u = mb_ptr2char_adv(&pu);
|
|
|
6156 #else
|
|
|
6157 f = *pf++;
|
|
|
6158 l = *pl++;
|
|
|
6159 u = *pu++;
|
|
|
6160 #endif
|
|
|
6161 /* Every character that appears is a word character. */
|
|
|
6162 if (f < 256)
|
|
|
6163 new_st.st_isw[f] = TRUE;
|
|
|
6164 if (l < 256)
|
|
|
6165 new_st.st_isw[l] = TRUE;
|
|
|
6166 if (u < 256)
|
|
|
6167 new_st.st_isw[u] = TRUE;
|
|
|
6168
|
|
|
6169 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
6170 * case-folding */
|
|
|
6171 if (l < 256 && l != f)
|
|
|
6172 {
|
|
|
6173 if (f >= 256)
|
|
|
6174 {
|
|
|
6175 EMSG(_(e_affrange));
|
|
|
6176 return FAIL;
|
|
|
6177 }
|
|
|
6178 new_st.st_fold[l] = f;
|
|
|
6179 }
|
|
|
6180
|
|
|
6181 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
6182 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
6183 * "FOL" . */
|
|
307
|
6184 if (u < 256 && u != f)
|
|
|
6185 {
|
|
|
6186 if (f >= 256)
|
|
|
6187 {
|
|
|
6188 EMSG(_(e_affrange));
|
|
|
6189 return FAIL;
|
|
|
6190 }
|
|
|
6191 new_st.st_fold[u] = f;
|
|
|
6192 new_st.st_isu[u] = TRUE;
|
|
324
|
6193 new_st.st_upper[f] = u;
|
|
307
|
6194 }
|
|
|
6195 }
|
|
|
6196
|
|
|
6197 if (*pl != NUL || *pu != NUL)
|
|
|
6198 {
|
|
|
6199 EMSG(_(e_affform));
|
|
|
6200 return FAIL;
|
|
|
6201 }
|
|
|
6202
|
|
|
6203 return set_spell_finish(&new_st);
|
|
|
6204 }
|
|
|
6205
|
|
|
6206 /*
|
|
|
6207 * Set the spell character tables from strings in the .spl file.
|
|
|
6208 */
|
|
|
6209 static int
|
|
389
|
6210 set_spell_charflags(flags, cnt, fol)
|
|
307
|
6211 char_u *flags;
|
|
389
|
6212 int cnt; /* length of "flags" */
|
|
|
6213 char_u *fol;
|
|
307
|
6214 {
|
|
|
6215 /* We build the new tables here first, so that we can compare with the
|
|
|
6216 * previous one. */
|
|
|
6217 spelltab_T new_st;
|
|
|
6218 int i;
|
|
389
|
6219 char_u *p = fol;
|
|
324
|
6220 int c;
|
|
307
|
6221
|
|
|
6222 clear_spell_chartab(&new_st);
|
|
|
6223
|
|
389
|
6224 for (i = 0; i < 128; ++i)
|
|
|
6225 {
|
|
|
6226 if (i < cnt)
|
|
|
6227 {
|
|
|
6228 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
6229 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
6230 }
|
|
|
6231
|
|
|
6232 if (*p != NUL)
|
|
|
6233 {
|
|
307
|
6234 #ifdef FEAT_MBYTE
|
|
389
|
6235 c = mb_ptr2char_adv(&p);
|
|
307
|
6236 #else
|
|
389
|
6237 c = *p++;
|
|
307
|
6238 #endif
|
|
389
|
6239 new_st.st_fold[i + 128] = c;
|
|
|
6240 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
6241 new_st.st_upper[c] = i + 128;
|
|
|
6242 }
|
|
307
|
6243 }
|
|
|
6244
|
|
|
6245 return set_spell_finish(&new_st);
|
|
|
6246 }
|
|
|
6247
|
|
|
6248 static int
|
|
|
6249 set_spell_finish(new_st)
|
|
|
6250 spelltab_T *new_st;
|
|
|
6251 {
|
|
|
6252 int i;
|
|
|
6253
|
|
|
6254 if (did_set_spelltab)
|
|
|
6255 {
|
|
|
6256 /* check that it's the same table */
|
|
|
6257 for (i = 0; i < 256; ++i)
|
|
|
6258 {
|
|
|
6259 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
6260 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
6261 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
6262 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
6263 {
|
|
|
6264 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
6265 return FAIL;
|
|
|
6266 }
|
|
|
6267 }
|
|
|
6268 }
|
|
|
6269 else
|
|
|
6270 {
|
|
|
6271 /* copy the new spelltab into the one being used */
|
|
|
6272 spelltab = *new_st;
|
|
|
6273 did_set_spelltab = TRUE;
|
|
|
6274 }
|
|
|
6275
|
|
|
6276 return OK;
|
|
|
6277 }
|
|
|
6278
|
|
|
6279 /*
|
|
358
|
6280 * Return TRUE if "p" points to a word character.
|
|
366
|
6281 * As a special case we see "midword" characters as word character when it is
|
|
358
|
6282 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
6283 * Thus this only works properly when past the first character of the word.
|
|
358
|
6284 */
|
|
|
6285 static int
|
|
376
|
6286 spell_iswordp(p, buf)
|
|
358
|
6287 char_u *p;
|
|
376
|
6288 buf_T *buf; /* buffer used */
|
|
358
|
6289 {
|
|
366
|
6290 #ifdef FEAT_MBYTE
|
|
358
|
6291 char_u *s;
|
|
366
|
6292 int l;
|
|
|
6293 int c;
|
|
|
6294
|
|
|
6295 if (has_mbyte)
|
|
|
6296 {
|
|
|
6297 l = MB_BYTE2LEN(*p);
|
|
358
|
6298 s = p;
|
|
366
|
6299 if (l == 1)
|
|
|
6300 {
|
|
|
6301 /* be quick for ASCII */
|
|
376
|
6302 if (buf->b_spell_ismw[*p])
|
|
366
|
6303 {
|
|
|
6304 s = p + 1; /* skip a mid-word character */
|
|
|
6305 l = MB_BYTE2LEN(*s);
|
|
|
6306 }
|
|
|
6307 }
|
|
|
6308 else
|
|
|
6309 {
|
|
|
6310 c = mb_ptr2char(p);
|
|
376
|
6311 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
6312 : (buf->b_spell_ismw_mb != NULL
|
|
|
6313 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
6314 {
|
|
|
6315 s = p + l;
|
|
|
6316 l = MB_BYTE2LEN(*s);
|
|
|
6317 }
|
|
|
6318 }
|
|
|
6319
|
|
390
|
6320 c = mb_ptr2char(s);
|
|
|
6321 if (c > 255)
|
|
366
|
6322 return mb_get_class(s) >= 2;
|
|
390
|
6323 return spelltab.st_isw[c];
|
|
366
|
6324 }
|
|
358
|
6325 #endif
|
|
366
|
6326
|
|
376
|
6327 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
6328 }
|
|
|
6329
|
|
|
6330 /*
|
|
|
6331 * Return TRUE if "p" points to a word character.
|
|
|
6332 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
6333 */
|
|
|
6334 static int
|
|
|
6335 spell_iswordp_nmw(p)
|
|
|
6336 char_u *p;
|
|
|
6337 {
|
|
|
6338 #ifdef FEAT_MBYTE
|
|
390
|
6339 int c;
|
|
|
6340
|
|
|
6341 if (has_mbyte)
|
|
|
6342 {
|
|
|
6343 c = mb_ptr2char(p);
|
|
|
6344 if (c > 255)
|
|
|
6345 return mb_get_class(p) >= 2;
|
|
|
6346 return spelltab.st_isw[c];
|
|
|
6347 }
|
|
376
|
6348 #endif
|
|
|
6349 return spelltab.st_isw[*p];
|
|
358
|
6350 }
|
|
|
6351
|
|
372
|
6352 #ifdef FEAT_MBYTE
|
|
|
6353 /*
|
|
|
6354 * Return TRUE if "p" points to a word character.
|
|
|
6355 * Wide version of spell_iswordp().
|
|
|
6356 */
|
|
|
6357 static int
|
|
376
|
6358 spell_iswordp_w(p, buf)
|
|
372
|
6359 int *p;
|
|
376
|
6360 buf_T *buf;
|
|
372
|
6361 {
|
|
|
6362 int *s;
|
|
|
6363
|
|
376
|
6364 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
6365 : (buf->b_spell_ismw_mb != NULL
|
|
|
6366 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
6367 s = p + 1;
|
|
|
6368 else
|
|
|
6369 s = p;
|
|
|
6370
|
|
390
|
6371 if (*s > 255)
|
|
372
|
6372 {
|
|
|
6373 if (enc_utf8)
|
|
|
6374 return utf_class(*s) >= 2;
|
|
|
6375 if (enc_dbcs)
|
|
|
6376 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
6377 return 0;
|
|
|
6378 }
|
|
|
6379 return spelltab.st_isw[*s];
|
|
|
6380 }
|
|
|
6381 #endif
|
|
|
6382
|
|
358
|
6383 /*
|
|
339
|
6384 * Write the table with prefix conditions to the .spl file.
|
|
|
6385 */
|
|
|
6386 static void
|
|
|
6387 write_spell_prefcond(fd, gap)
|
|
|
6388 FILE *fd;
|
|
|
6389 garray_T *gap;
|
|
|
6390 {
|
|
|
6391 int i;
|
|
|
6392 char_u *p;
|
|
|
6393 int len;
|
|
|
6394
|
|
|
6395 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
6396
|
|
|
6397 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6398 {
|
|
|
6399 /* <prefcond> : <condlen> <condstr> */
|
|
|
6400 p = ((char_u **)gap->ga_data)[i];
|
|
|
6401 if (p == NULL)
|
|
|
6402 fputc(0, fd);
|
|
|
6403 else
|
|
|
6404 {
|
|
|
6405 len = STRLEN(p);
|
|
|
6406 fputc(len, fd);
|
|
|
6407 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
6408 }
|
|
|
6409 }
|
|
|
6410 }
|
|
|
6411
|
|
|
6412 /*
|
|
307
|
6413 * Write the current tables into the .spl file.
|
|
|
6414 * This makes sure the same characters are recognized as word characters when
|
|
|
6415 * generating an when using a spell file.
|
|
|
6416 */
|
|
|
6417 static void
|
|
|
6418 write_spell_chartab(fd)
|
|
|
6419 FILE *fd;
|
|
|
6420 {
|
|
|
6421 char_u charbuf[256 * 4];
|
|
|
6422 int len = 0;
|
|
|
6423 int flags;
|
|
|
6424 int i;
|
|
|
6425
|
|
|
6426 fputc(128, fd); /* <charflagslen> */
|
|
|
6427 for (i = 128; i < 256; ++i)
|
|
|
6428 {
|
|
|
6429 flags = 0;
|
|
|
6430 if (spelltab.st_isw[i])
|
|
324
|
6431 flags |= CF_WORD;
|
|
307
|
6432 if (spelltab.st_isu[i])
|
|
324
|
6433 flags |= CF_UPPER;
|
|
307
|
6434 fputc(flags, fd); /* <charflags> */
|
|
|
6435
|
|
310
|
6436 #ifdef FEAT_MBYTE
|
|
|
6437 if (has_mbyte)
|
|
|
6438 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
|
6439 else
|
|
|
6440 #endif
|
|
|
6441 charbuf[len++] = spelltab.st_fold[i];
|
|
307
|
6442 }
|
|
|
6443
|
|
|
6444 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
|
6445 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
|
6446 }
|
|
|
6447
|
|
|
6448 /*
|
|
324
|
6449 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
6450 * Uses the character definitions from the .spl file.
|
|
307
|
6451 * When using a multi-byte 'encoding' the length may change!
|
|
|
6452 * Returns FAIL when something wrong.
|
|
|
6453 */
|
|
|
6454 static int
|
|
324
|
6455 spell_casefold(str, len, buf, buflen)
|
|
|
6456 char_u *str;
|
|
307
|
6457 int len;
|
|
|
6458 char_u *buf;
|
|
|
6459 int buflen;
|
|
|
6460 {
|
|
|
6461 int i;
|
|
|
6462
|
|
|
6463 if (len >= buflen)
|
|
|
6464 {
|
|
|
6465 buf[0] = NUL;
|
|
|
6466 return FAIL; /* result will not fit */
|
|
|
6467 }
|
|
|
6468
|
|
|
6469 #ifdef FEAT_MBYTE
|
|
|
6470 if (has_mbyte)
|
|
|
6471 {
|
|
324
|
6472 int outi = 0;
|
|
|
6473 char_u *p;
|
|
307
|
6474 int c;
|
|
|
6475
|
|
|
6476 /* Fold one character at a time. */
|
|
324
|
6477 for (p = str; p < str + len; )
|
|
307
|
6478 {
|
|
|
6479 if (outi + MB_MAXBYTES > buflen)
|
|
|
6480 {
|
|
|
6481 buf[outi] = NUL;
|
|
|
6482 return FAIL;
|
|
|
6483 }
|
|
474
|
6484 c = mb_cptr2char_adv(&p);
|
|
324
|
6485 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
6486 }
|
|
|
6487 buf[outi] = NUL;
|
|
|
6488 }
|
|
|
6489 else
|
|
|
6490 #endif
|
|
|
6491 {
|
|
|
6492 /* Be quick for non-multibyte encodings. */
|
|
|
6493 for (i = 0; i < len; ++i)
|
|
324
|
6494 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
6495 buf[i] = NUL;
|
|
|
6496 }
|
|
|
6497
|
|
|
6498 return OK;
|
|
|
6499 }
|
|
|
6500
|
|
372
|
6501 #define SPS_BEST 1
|
|
|
6502 #define SPS_FAST 2
|
|
|
6503 #define SPS_DOUBLE 4
|
|
|
6504
|
|
|
6505 static int sps_flags = SPS_BEST;
|
|
|
6506
|
|
|
6507 /*
|
|
|
6508 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
|
6509 * Sets "sps_flags".
|
|
|
6510 */
|
|
|
6511 int
|
|
|
6512 spell_check_sps()
|
|
|
6513 {
|
|
|
6514 char_u *p;
|
|
|
6515 char_u buf[MAXPATHL];
|
|
|
6516 int f;
|
|
|
6517
|
|
|
6518 sps_flags = 0;
|
|
|
6519
|
|
|
6520 for (p = p_sps; *p != NUL; )
|
|
|
6521 {
|
|
|
6522 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6523
|
|
|
6524 f = 0;
|
|
|
6525 if (STRCMP(buf, "best") == 0)
|
|
|
6526 f = SPS_BEST;
|
|
|
6527 else if (STRCMP(buf, "fast") == 0)
|
|
|
6528 f = SPS_FAST;
|
|
|
6529 else if (STRCMP(buf, "double") == 0)
|
|
|
6530 f = SPS_DOUBLE;
|
|
|
6531 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
6532 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
6533 f = -1;
|
|
|
6534
|
|
|
6535 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
6536 {
|
|
|
6537 sps_flags = SPS_BEST;
|
|
|
6538 return FAIL;
|
|
|
6539 }
|
|
|
6540 if (f != 0)
|
|
|
6541 sps_flags = f;
|
|
|
6542 }
|
|
|
6543
|
|
|
6544 if (sps_flags == 0)
|
|
|
6545 sps_flags = SPS_BEST;
|
|
|
6546
|
|
|
6547 return OK;
|
|
|
6548 }
|
|
|
6549
|
|
|
6550 /* Remember what "z?" replaced. */
|
|
|
6551 static char_u *repl_from = NULL;
|
|
|
6552 static char_u *repl_to = NULL;
|
|
|
6553
|
|
323
|
6554 /*
|
|
|
6555 * "z?": Find badly spelled word under or after the cursor.
|
|
|
6556 * Give suggestions for the properly spelled word.
|
|
|
6557 */
|
|
|
6558 void
|
|
|
6559 spell_suggest()
|
|
|
6560 {
|
|
|
6561 char_u *line;
|
|
|
6562 pos_T prev_cursor = curwin->w_cursor;
|
|
|
6563 char_u wcopy[MAXWLEN + 2];
|
|
|
6564 char_u *p;
|
|
|
6565 int i;
|
|
|
6566 int c;
|
|
|
6567 suginfo_T sug;
|
|
|
6568 suggest_T *stp;
|
|
372
|
6569 int mouse_used;
|
|
388
|
6570 int need_cap;
|
|
|
6571 regmatch_T regmatch;
|
|
|
6572 int endcol;
|
|
|
6573 char_u *line_copy = NULL;
|
|
323
|
6574
|
|
344
|
6575 /* Find the start of the badly spelled word. */
|
|
346
|
6576 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
6577 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
6578 {
|
|
|
6579 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
6580 return;
|
|
|
6581
|
|
|
6582 /* No bad word or it starts after the cursor: use the word under the
|
|
|
6583 * cursor. */
|
|
|
6584 curwin->w_cursor = prev_cursor;
|
|
|
6585 line = ml_get_curline();
|
|
|
6586 p = line + curwin->w_cursor.col;
|
|
|
6587 /* Backup to before start of word. */
|
|
390
|
6588 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
6589 mb_ptr_back(line, p);
|
|
|
6590 /* Forward to start of word. */
|
|
390
|
6591 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
6592 mb_ptr_adv(p);
|
|
|
6593
|
|
390
|
6594 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
6595 {
|
|
|
6596 beep_flush();
|
|
|
6597 return;
|
|
|
6598 }
|
|
|
6599 curwin->w_cursor.col = p - line;
|
|
323
|
6600 }
|
|
|
6601
|
|
344
|
6602 /* Get the word and its length. */
|
|
323
|
6603 line = ml_get_curline();
|
|
344
|
6604
|
|
388
|
6605 /* Figure out if the word should be capitalised. */
|
|
|
6606 need_cap = FALSE;
|
|
|
6607 if (curbuf->b_cap_prog != NULL)
|
|
|
6608 {
|
|
|
6609 endcol = 0;
|
|
398
|
6610 if ((int)(skipwhite(line) - line) == (int)curwin->w_cursor.col)
|
|
388
|
6611 {
|
|
|
6612 /* At start of line, check if previous line is empty or sentence
|
|
|
6613 * ends there. */
|
|
|
6614 if (curwin->w_cursor.lnum == 1)
|
|
|
6615 need_cap = TRUE;
|
|
|
6616 else
|
|
|
6617 {
|
|
|
6618 line = ml_get(curwin->w_cursor.lnum - 1);
|
|
|
6619 if (*skipwhite(line) == NUL)
|
|
|
6620 need_cap = TRUE;
|
|
|
6621 else
|
|
|
6622 {
|
|
|
6623 /* Append a space in place of the line break. */
|
|
|
6624 line_copy = concat_str(line, (char_u *)" ");
|
|
|
6625 line = line_copy;
|
|
|
6626 endcol = STRLEN(line);
|
|
|
6627 }
|
|
|
6628 }
|
|
|
6629 }
|
|
|
6630 else
|
|
|
6631 endcol = curwin->w_cursor.col;
|
|
|
6632
|
|
|
6633 if (endcol > 0)
|
|
|
6634 {
|
|
|
6635 /* Check if sentence ends before the bad word. */
|
|
|
6636 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
6637 regmatch.rm_ic = FALSE;
|
|
|
6638 p = line + endcol;
|
|
|
6639 for (;;)
|
|
|
6640 {
|
|
|
6641 mb_ptr_back(line, p);
|
|
390
|
6642 if (p == line || spell_iswordp_nmw(p))
|
|
388
|
6643 break;
|
|
|
6644 if (vim_regexec(®match, p, 0)
|
|
|
6645 && regmatch.endp[0] == line + endcol)
|
|
|
6646 {
|
|
|
6647 need_cap = TRUE;
|
|
|
6648 break;
|
|
|
6649 }
|
|
|
6650 }
|
|
|
6651 }
|
|
|
6652
|
|
|
6653 /* get the line again, we may have been using the previous one */
|
|
|
6654 line = ml_get_curline();
|
|
|
6655 vim_free(line_copy);
|
|
|
6656 }
|
|
|
6657
|
|
344
|
6658 /* Get the list of suggestions */
|
|
388
|
6659 spell_find_suggest(line + curwin->w_cursor.col, &sug, (int)Rows - 2,
|
|
|
6660 TRUE, need_cap);
|
|
323
|
6661
|
|
|
6662 if (sug.su_ga.ga_len == 0)
|
|
|
6663 MSG(_("Sorry, no suggestions"));
|
|
|
6664 else
|
|
|
6665 {
|
|
372
|
6666 vim_free(repl_from);
|
|
|
6667 repl_from = NULL;
|
|
|
6668 vim_free(repl_to);
|
|
|
6669 repl_to = NULL;
|
|
|
6670
|
|
474
|
6671 #ifdef FEAT_RIGHTLEFT
|
|
|
6672 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
6673 cmdmsg_rl = curwin->w_p_rl;
|
|
|
6674 if (cmdmsg_rl)
|
|
|
6675 msg_col = Columns - 1;
|
|
|
6676 #endif
|
|
|
6677
|
|
323
|
6678 /* List the suggestions. */
|
|
|
6679 msg_start();
|
|
372
|
6680 lines_left = Rows; /* avoid more prompt */
|
|
323
|
6681 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
6682 sug.su_badlen, sug.su_badptr);
|
|
474
|
6683 #ifdef FEAT_RIGHTLEFT
|
|
|
6684 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
6685 {
|
|
|
6686 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
6687 * untranslated message rightleft. */
|
|
|
6688 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
6689 sug.su_badlen, sug.su_badptr);
|
|
|
6690 }
|
|
|
6691 #endif
|
|
323
|
6692 msg_puts(IObuff);
|
|
|
6693 msg_clr_eos();
|
|
|
6694 msg_putchar('\n');
|
|
346
|
6695
|
|
323
|
6696 msg_scroll = TRUE;
|
|
|
6697 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
6698 {
|
|
344
|
6699 stp = &SUG(sug.su_ga, i);
|
|
323
|
6700
|
|
|
6701 /* The suggested word may replace only part of the bad word, add
|
|
|
6702 * the not replaced part. */
|
|
|
6703 STRCPY(wcopy, stp->st_word);
|
|
|
6704 if (sug.su_badlen > stp->st_orglen)
|
|
|
6705 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
6706 sug.su_badptr + stp->st_orglen,
|
|
|
6707 sug.su_badlen - stp->st_orglen);
|
|
474
|
6708 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
6709 #ifdef FEAT_RIGHTLEFT
|
|
|
6710 if (cmdmsg_rl)
|
|
|
6711 rl_mirror(IObuff);
|
|
|
6712 #endif
|
|
|
6713 msg_puts(IObuff);
|
|
|
6714
|
|
|
6715 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
6716 msg_puts(IObuff);
|
|
|
6717
|
|
|
6718 /* The word may replace more than "su_badlen". */
|
|
|
6719 if (sug.su_badlen < stp->st_orglen)
|
|
|
6720 {
|
|
|
6721 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
6722 stp->st_orglen, sug.su_badptr);
|
|
|
6723 msg_puts(IObuff);
|
|
|
6724 }
|
|
|
6725
|
|
324
|
6726 if (p_verbose > 0)
|
|
344
|
6727 {
|
|
346
|
6728 /* Add the score. */
|
|
351
|
6729 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
6730 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
6731 stp->st_salscore ? "s " : "",
|
|
|
6732 stp->st_score, stp->st_altscore);
|
|
|
6733 else
|
|
474
|
6734 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
6735 stp->st_score);
|
|
474
|
6736 #ifdef FEAT_RIGHTLEFT
|
|
|
6737 if (cmdmsg_rl)
|
|
|
6738 /* Mirror the numbers, but keep the leading space. */
|
|
|
6739 rl_mirror(IObuff + 1);
|
|
|
6740 #endif
|
|
346
|
6741 msg_advance(30);
|
|
|
6742 msg_puts(IObuff);
|
|
344
|
6743 }
|
|
323
|
6744 msg_putchar('\n');
|
|
|
6745 }
|
|
|
6746
|
|
474
|
6747 #ifdef FEAT_RIGHTLEFT
|
|
|
6748 cmdmsg_rl = FALSE;
|
|
|
6749 msg_col = 0;
|
|
|
6750 #endif
|
|
323
|
6751 /* Ask for choice. */
|
|
372
|
6752 i = prompt_for_number(&mouse_used);
|
|
|
6753 if (mouse_used)
|
|
|
6754 i -= lines_left;
|
|
|
6755
|
|
344
|
6756 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
6757 {
|
|
372
|
6758 /* Save the from and to text for :spellrepall. */
|
|
|
6759 stp = &SUG(sug.su_ga, i - 1);
|
|
|
6760 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
6761 repl_to = vim_strsave(stp->st_word);
|
|
|
6762
|
|
323
|
6763 /* Replace the word. */
|
|
|
6764 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
6765 if (p != NULL)
|
|
|
6766 {
|
|
|
6767 c = sug.su_badptr - line;
|
|
|
6768 mch_memmove(p, line, c);
|
|
|
6769 STRCPY(p + c, stp->st_word);
|
|
|
6770 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
6771 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6772 curwin->w_cursor.col = c;
|
|
|
6773 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
6774
|
|
|
6775 /* For redo we use a change-word command. */
|
|
|
6776 ResetRedobuff();
|
|
|
6777 AppendToRedobuff((char_u *)"ciw");
|
|
|
6778 AppendToRedobuff(stp->st_word);
|
|
|
6779 AppendCharToRedobuff(ESC);
|
|
323
|
6780 }
|
|
|
6781 }
|
|
|
6782 else
|
|
|
6783 curwin->w_cursor = prev_cursor;
|
|
|
6784 }
|
|
|
6785
|
|
344
|
6786 spell_find_cleanup(&sug);
|
|
|
6787 }
|
|
|
6788
|
|
|
6789 /*
|
|
372
|
6790 * ":spellrepall"
|
|
|
6791 */
|
|
|
6792 /*ARGSUSED*/
|
|
|
6793 void
|
|
|
6794 ex_spellrepall(eap)
|
|
|
6795 exarg_T *eap;
|
|
|
6796 {
|
|
|
6797 pos_T pos = curwin->w_cursor;
|
|
|
6798 char_u *frompat;
|
|
|
6799 int addlen;
|
|
|
6800 char_u *line;
|
|
|
6801 char_u *p;
|
|
|
6802 int didone = FALSE;
|
|
|
6803 int save_ws = p_ws;
|
|
|
6804
|
|
|
6805 if (repl_from == NULL || repl_to == NULL)
|
|
|
6806 {
|
|
|
6807 EMSG(_("E752: No previous spell replacement"));
|
|
|
6808 return;
|
|
|
6809 }
|
|
|
6810 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
6811
|
|
|
6812 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
6813 if (frompat == NULL)
|
|
|
6814 return;
|
|
|
6815 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
6816 p_ws = FALSE;
|
|
|
6817
|
|
|
6818 curwin->w_cursor.lnum = 0;
|
|
|
6819 while (!got_int)
|
|
|
6820 {
|
|
|
6821 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
6822 || u_save_cursor() == FAIL)
|
|
|
6823 break;
|
|
|
6824
|
|
|
6825 /* Only replace when the right word isn't there yet. This happens
|
|
|
6826 * when changing "etc" to "etc.". */
|
|
|
6827 line = ml_get_curline();
|
|
|
6828 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
6829 repl_to, STRLEN(repl_to)) != 0)
|
|
|
6830 {
|
|
|
6831 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
6832 if (p == NULL)
|
|
|
6833 break;
|
|
|
6834 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
6835 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
6836 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
6837 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6838 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
6839 didone = TRUE;
|
|
|
6840 }
|
|
|
6841 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
6842 }
|
|
|
6843
|
|
|
6844 p_ws = save_ws;
|
|
|
6845 curwin->w_cursor = pos;
|
|
|
6846 vim_free(frompat);
|
|
|
6847
|
|
|
6848 if (!didone)
|
|
|
6849 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
|
6850 }
|
|
|
6851
|
|
|
6852 /*
|
|
344
|
6853 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
6854 * a list of allocated strings.
|
|
|
6855 */
|
|
|
6856 void
|
|
|
6857 spell_suggest_list(gap, word, maxcount)
|
|
|
6858 garray_T *gap;
|
|
|
6859 char_u *word;
|
|
|
6860 int maxcount; /* maximum nr of suggestions */
|
|
|
6861 {
|
|
|
6862 suginfo_T sug;
|
|
|
6863 int i;
|
|
|
6864 suggest_T *stp;
|
|
|
6865 char_u *wcopy;
|
|
|
6866
|
|
388
|
6867 spell_find_suggest(word, &sug, maxcount, FALSE, FALSE);
|
|
344
|
6868
|
|
|
6869 /* Make room in "gap". */
|
|
|
6870 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
6871 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
6872 return;
|
|
|
6873
|
|
323
|
6874 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
6875 {
|
|
|
6876 stp = &SUG(sug.su_ga, i);
|
|
|
6877
|
|
|
6878 /* The suggested word may replace only part of "word", add the not
|
|
|
6879 * replaced part. */
|
|
|
6880 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
6881 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
6882 if (wcopy == NULL)
|
|
|
6883 break;
|
|
|
6884 STRCPY(wcopy, stp->st_word);
|
|
|
6885 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
6886 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
6887 }
|
|
|
6888
|
|
|
6889 spell_find_cleanup(&sug);
|
|
|
6890 }
|
|
|
6891
|
|
|
6892 /*
|
|
|
6893 * Find spell suggestions for the word at the start of "badptr".
|
|
|
6894 * Return the suggestions in "su->su_ga".
|
|
|
6895 * The maximum number of suggestions is "maxcount".
|
|
|
6896 * Note: does use info for the current window.
|
|
|
6897 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
6898 */
|
|
|
6899 static void
|
|
388
|
6900 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
6901 char_u *badptr;
|
|
|
6902 suginfo_T *su;
|
|
|
6903 int maxcount;
|
|
358
|
6904 int banbadword; /* don't include badword in suggestions */
|
|
388
|
6905 int need_cap; /* word should start with capital */
|
|
344
|
6906 {
|
|
385
|
6907 int attr = 0;
|
|
372
|
6908 char_u buf[MAXPATHL];
|
|
|
6909 char_u *p;
|
|
|
6910 int do_combine = FALSE;
|
|
|
6911 char_u *sps_copy;
|
|
|
6912 #ifdef FEAT_EVAL
|
|
|
6913 static int expr_busy = FALSE;
|
|
|
6914 #endif
|
|
385
|
6915 int c;
|
|
344
|
6916
|
|
|
6917 /*
|
|
|
6918 * Set the info in "*su".
|
|
|
6919 */
|
|
|
6920 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
6921 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
6922 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
6923 if (*badptr == NUL)
|
|
|
6924 return;
|
|
344
|
6925 hash_init(&su->su_banned);
|
|
|
6926
|
|
|
6927 su->su_badptr = badptr;
|
|
385
|
6928 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
6929 su->su_maxcount = maxcount;
|
|
372
|
6930 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
6931
|
|
|
6932 if (su->su_badlen >= MAXWLEN)
|
|
|
6933 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
6934 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
6935 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
6936 su->su_fbadword, MAXWLEN);
|
|
346
|
6937 /* get caps flags for bad word */
|
|
474
|
6938 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
6939 su->su_badptr + su->su_badlen);
|
|
388
|
6940 if (need_cap)
|
|
|
6941 su->su_badflags |= WF_ONECAP;
|
|
344
|
6942
|
|
385
|
6943 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
6944 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
6945 * for that. */
|
|
455
|
6946 c = PTR2CHAR(su->su_badptr);
|
|
385
|
6947 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
6948 {
|
|
|
6949 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
6950 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
6951 0, TRUE);
|
|
|
6952 }
|
|
|
6953
|
|
344
|
6954 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
6955 if (banbadword)
|
|
|
6956 add_banned(su, su->su_badword);
|
|
344
|
6957
|
|
372
|
6958 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
6959 sps_copy = vim_strsave(p_sps);
|
|
|
6960 if (sps_copy == NULL)
|
|
|
6961 return;
|
|
|
6962
|
|
|
6963 /* Loop over the items in 'spellsuggest'. */
|
|
|
6964 for (p = sps_copy; *p != NUL; )
|
|
|
6965 {
|
|
|
6966 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6967
|
|
|
6968 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
6969 {
|
|
|
6970 #ifdef FEAT_EVAL
|
|
375
|
6971 /* Evaluate an expression. Skip this when called recursively,
|
|
|
6972 * when using spellsuggest() in the expression. */
|
|
372
|
6973 if (!expr_busy)
|
|
|
6974 {
|
|
|
6975 expr_busy = TRUE;
|
|
|
6976 spell_suggest_expr(su, buf + 5);
|
|
|
6977 expr_busy = FALSE;
|
|
|
6978 }
|
|
|
6979 #endif
|
|
|
6980 }
|
|
|
6981 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
6982 /* Use list of suggestions in a file. */
|
|
|
6983 spell_suggest_file(su, buf + 5);
|
|
|
6984 else
|
|
|
6985 {
|
|
|
6986 /* Use internal method. */
|
|
|
6987 spell_suggest_intern(su);
|
|
|
6988 if (sps_flags & SPS_DOUBLE)
|
|
|
6989 do_combine = TRUE;
|
|
|
6990 }
|
|
|
6991 }
|
|
|
6992
|
|
|
6993 vim_free(sps_copy);
|
|
|
6994
|
|
|
6995 if (do_combine)
|
|
|
6996 /* Combine the two list of suggestions. This must be done last,
|
|
|
6997 * because sorting changes the order again. */
|
|
|
6998 score_combine(su);
|
|
|
6999 }
|
|
|
7000
|
|
|
7001 #ifdef FEAT_EVAL
|
|
|
7002 /*
|
|
|
7003 * Find suggestions by evaluating expression "expr".
|
|
|
7004 */
|
|
|
7005 static void
|
|
|
7006 spell_suggest_expr(su, expr)
|
|
|
7007 suginfo_T *su;
|
|
|
7008 char_u *expr;
|
|
|
7009 {
|
|
|
7010 list_T *list;
|
|
|
7011 listitem_T *li;
|
|
|
7012 int score;
|
|
|
7013 char_u *p;
|
|
|
7014
|
|
|
7015 /* The work is split up in a few parts to avoid having to export
|
|
|
7016 * suginfo_T.
|
|
|
7017 * First evaluate the expression and get the resulting list. */
|
|
|
7018 list = eval_spell_expr(su->su_badword, expr);
|
|
|
7019 if (list != NULL)
|
|
|
7020 {
|
|
|
7021 /* Loop over the items in the list. */
|
|
|
7022 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
7023 if (li->li_tv.v_type == VAR_LIST)
|
|
|
7024 {
|
|
|
7025 /* Get the word and the score from the items. */
|
|
|
7026 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
7027 if (score >= 0)
|
|
|
7028 add_suggestion(su, &su->su_ga, p,
|
|
|
7029 su->su_badlen, score, 0, TRUE);
|
|
|
7030 }
|
|
|
7031 list_unref(list);
|
|
|
7032 }
|
|
|
7033
|
|
|
7034 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
7035 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7036 }
|
|
|
7037 #endif
|
|
|
7038
|
|
|
7039 /*
|
|
474
|
7040 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
7041 */
|
|
|
7042 static void
|
|
|
7043 spell_suggest_file(su, fname)
|
|
|
7044 suginfo_T *su;
|
|
|
7045 char_u *fname;
|
|
|
7046 {
|
|
|
7047 FILE *fd;
|
|
|
7048 char_u line[MAXWLEN * 2];
|
|
|
7049 char_u *p;
|
|
|
7050 int len;
|
|
|
7051 char_u cword[MAXWLEN];
|
|
|
7052
|
|
|
7053 /* Open the file. */
|
|
|
7054 fd = mch_fopen((char *)fname, "r");
|
|
|
7055 if (fd == NULL)
|
|
|
7056 {
|
|
|
7057 EMSG2(_(e_notopen), fname);
|
|
|
7058 return;
|
|
|
7059 }
|
|
|
7060
|
|
|
7061 /* Read it line by line. */
|
|
|
7062 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
7063 {
|
|
|
7064 line_breakcheck();
|
|
|
7065
|
|
|
7066 p = vim_strchr(line, '/');
|
|
|
7067 if (p == NULL)
|
|
|
7068 continue; /* No Tab found, just skip the line. */
|
|
|
7069 *p++ = NUL;
|
|
|
7070 if (STRICMP(su->su_badword, line) == 0)
|
|
|
7071 {
|
|
|
7072 /* Match! Isolate the good word, until CR or NL. */
|
|
|
7073 for (len = 0; p[len] >= ' '; ++len)
|
|
|
7074 ;
|
|
|
7075 p[len] = NUL;
|
|
|
7076
|
|
|
7077 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
7078 * of the bad word. */
|
|
|
7079 if (captype(p, NULL) == 0)
|
|
|
7080 {
|
|
|
7081 make_case_word(p, cword, su->su_badflags);
|
|
|
7082 p = cword;
|
|
|
7083 }
|
|
|
7084
|
|
|
7085 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
7086 SCORE_FILE, 0, TRUE);
|
|
|
7087 }
|
|
|
7088 }
|
|
|
7089
|
|
|
7090 fclose(fd);
|
|
|
7091
|
|
|
7092 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
7093 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7094 }
|
|
|
7095
|
|
|
7096 /*
|
|
|
7097 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
7098 */
|
|
|
7099 static void
|
|
|
7100 spell_suggest_intern(su)
|
|
|
7101 suginfo_T *su;
|
|
|
7102 {
|
|
344
|
7103 /*
|
|
346
|
7104 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
7105 *
|
|
|
7106 * Set a maximum score to limit the combination of operations that is
|
|
|
7107 * tried.
|
|
|
7108 */
|
|
346
|
7109 suggest_try_special(su);
|
|
|
7110
|
|
|
7111 /*
|
|
|
7112 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
7113 * from the .aff file and inserting a space (split the word).
|
|
|
7114 */
|
|
|
7115 suggest_try_change(su);
|
|
344
|
7116
|
|
|
7117 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
7118 if (sps_flags & SPS_DOUBLE)
|
|
|
7119 score_comp_sal(su);
|
|
|
7120
|
|
|
7121 /*
|
|
346
|
7122 * 3. Try finding sound-a-like words.
|
|
344
|
7123 *
|
|
|
7124 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
7125 * very slow and often doesn't find new suggestions.
|
|
|
7126 */
|
|
|
7127 if ((sps_flags & SPS_DOUBLE)
|
|
|
7128 || (!(sps_flags & SPS_FAST)
|
|
|
7129 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
7130 {
|
|
|
7131 /* Allow a higher score now. */
|
|
|
7132 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
7133 suggest_try_soundalike(su);
|
|
344
|
7134 }
|
|
|
7135
|
|
|
7136 /* When CTRL-C was hit while searching do show the results. */
|
|
|
7137 ui_breakcheck();
|
|
|
7138 if (got_int)
|
|
|
7139 {
|
|
|
7140 (void)vgetc();
|
|
|
7141 got_int = FALSE;
|
|
|
7142 }
|
|
|
7143
|
|
372
|
7144 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
7145 {
|
|
|
7146 if (sps_flags & SPS_BEST)
|
|
|
7147 /* Adjust the word score for how it sounds like. */
|
|
|
7148 rescore_suggestions(su);
|
|
|
7149
|
|
|
7150 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
7151 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
7152 }
|
|
|
7153 }
|
|
|
7154
|
|
|
7155 /*
|
|
|
7156 * Free the info put in "*su" by spell_find_suggest().
|
|
|
7157 */
|
|
|
7158 static void
|
|
|
7159 spell_find_cleanup(su)
|
|
|
7160 suginfo_T *su;
|
|
|
7161 {
|
|
|
7162 int i;
|
|
|
7163
|
|
|
7164 /* Free the suggestions. */
|
|
|
7165 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7166 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
7167 ga_clear(&su->su_ga);
|
|
|
7168 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
7169 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
7170 ga_clear(&su->su_sga);
|
|
323
|
7171
|
|
|
7172 /* Free the banned words. */
|
|
344
|
7173 free_banned(su);
|
|
323
|
7174 }
|
|
|
7175
|
|
|
7176 /*
|
|
324
|
7177 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
7178 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
7179 * The result is NUL terminated.
|
|
323
|
7180 */
|
|
|
7181 static void
|
|
324
|
7182 onecap_copy(word, wcopy, upper)
|
|
323
|
7183 char_u *word;
|
|
|
7184 char_u *wcopy;
|
|
|
7185 int upper; /* TRUE: first letter made upper case */
|
|
|
7186 {
|
|
|
7187 char_u *p;
|
|
|
7188 int c;
|
|
|
7189 int l;
|
|
|
7190
|
|
|
7191 p = word;
|
|
|
7192 #ifdef FEAT_MBYTE
|
|
|
7193 if (has_mbyte)
|
|
474
|
7194 c = mb_cptr2char_adv(&p);
|
|
323
|
7195 else
|
|
|
7196 #endif
|
|
|
7197 c = *p++;
|
|
|
7198 if (upper)
|
|
324
|
7199 c = SPELL_TOUPPER(c);
|
|
323
|
7200 else
|
|
324
|
7201 c = SPELL_TOFOLD(c);
|
|
323
|
7202 #ifdef FEAT_MBYTE
|
|
|
7203 if (has_mbyte)
|
|
|
7204 l = mb_char2bytes(c, wcopy);
|
|
|
7205 else
|
|
|
7206 #endif
|
|
|
7207 {
|
|
|
7208 l = 1;
|
|
|
7209 wcopy[0] = c;
|
|
|
7210 }
|
|
376
|
7211 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
7212 }
|
|
|
7213
|
|
|
7214 /*
|
|
324
|
7215 * Make a copy of "word" with all the letters upper cased into
|
|
|
7216 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
7217 */
|
|
|
7218 static void
|
|
|
7219 allcap_copy(word, wcopy)
|
|
|
7220 char_u *word;
|
|
|
7221 char_u *wcopy;
|
|
|
7222 {
|
|
|
7223 char_u *s;
|
|
|
7224 char_u *d;
|
|
|
7225 int c;
|
|
|
7226
|
|
|
7227 d = wcopy;
|
|
|
7228 for (s = word; *s != NUL; )
|
|
|
7229 {
|
|
|
7230 #ifdef FEAT_MBYTE
|
|
|
7231 if (has_mbyte)
|
|
474
|
7232 c = mb_cptr2char_adv(&s);
|
|
323
|
7233 else
|
|
|
7234 #endif
|
|
|
7235 c = *s++;
|
|
324
|
7236 c = SPELL_TOUPPER(c);
|
|
323
|
7237
|
|
|
7238 #ifdef FEAT_MBYTE
|
|
|
7239 if (has_mbyte)
|
|
|
7240 {
|
|
|
7241 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
7242 break;
|
|
|
7243 d += mb_char2bytes(c, d);
|
|
|
7244 }
|
|
|
7245 else
|
|
|
7246 #endif
|
|
|
7247 {
|
|
|
7248 if (d - wcopy >= MAXWLEN - 1)
|
|
|
7249 break;
|
|
|
7250 *d++ = c;
|
|
|
7251 }
|
|
|
7252 }
|
|
|
7253 *d = NUL;
|
|
|
7254 }
|
|
|
7255
|
|
|
7256 /*
|
|
346
|
7257 * Try finding suggestions by recognizing specific situations.
|
|
|
7258 */
|
|
|
7259 static void
|
|
|
7260 suggest_try_special(su)
|
|
|
7261 suginfo_T *su;
|
|
|
7262 {
|
|
|
7263 char_u *p;
|
|
376
|
7264 size_t len;
|
|
346
|
7265 int c;
|
|
|
7266 char_u word[MAXWLEN];
|
|
|
7267
|
|
|
7268 /*
|
|
|
7269 * Recognize a word that is repeated: "the the".
|
|
|
7270 */
|
|
|
7271 p = skiptowhite(su->su_fbadword);
|
|
|
7272 len = p - su->su_fbadword;
|
|
|
7273 p = skipwhite(p);
|
|
|
7274 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
7275 {
|
|
|
7276 /* Include badflags: if the badword is onecap or allcap
|
|
|
7277 * use that for the goodword too: "The the" -> "The". */
|
|
|
7278 c = su->su_fbadword[len];
|
|
|
7279 su->su_fbadword[len] = NUL;
|
|
|
7280 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
7281 su->su_fbadword[len] = c;
|
|
351
|
7282 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
7283 }
|
|
|
7284 }
|
|
|
7285
|
|
|
7286 /*
|
|
323
|
7287 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
7288 *
|
|
|
7289 * This uses a state machine. At each node in the tree we try various
|
|
|
7290 * operations. When trying if an operation work "depth" is increased and the
|
|
|
7291 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
7292 * character, replace one and delete another. The number of changes is
|
|
|
7293 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
7294 */
|
|
|
7295 static void
|
|
346
|
7296 suggest_try_change(su)
|
|
323
|
7297 suginfo_T *su;
|
|
|
7298 {
|
|
|
7299 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
7300 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
7301 trystate_T stack[MAXWLEN];
|
|
|
7302 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
|
7303 * to for word split) */
|
|
|
7304 char_u prewordlen = 0; /* length of word in "preword" */
|
|
|
7305 int splitoff = 0; /* index in tword after last split */
|
|
|
7306 trystate_T *sp;
|
|
|
7307 int newscore;
|
|
|
7308 langp_T *lp;
|
|
375
|
7309 char_u *byts, *fbyts, *pbyts;
|
|
|
7310 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
7311 int depth;
|
|
330
|
7312 int c, c2, c3;
|
|
390
|
7313 int n;
|
|
323
|
7314 int flags;
|
|
|
7315 garray_T *gap;
|
|
324
|
7316 idx_T arridx;
|
|
323
|
7317 int len;
|
|
|
7318 char_u *p;
|
|
|
7319 fromto_T *ftp;
|
|
330
|
7320 int fl = 0, tl;
|
|
346
|
7321 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
323
|
7322
|
|
|
7323 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
7324 * to find matches (esp. REP items). Append some more text, changing
|
|
|
7325 * chars after the bad word may help. */
|
|
323
|
7326 STRCPY(fword, su->su_fbadword);
|
|
346
|
7327 n = STRLEN(fword);
|
|
|
7328 p = su->su_badptr + su->su_badlen;
|
|
|
7329 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
7330
|
|
|
7331 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7332 lp->lp_slang != NULL; ++lp)
|
|
|
7333 {
|
|
|
7334 /*
|
|
|
7335 * Go through the whole case-fold tree, try changes at each node.
|
|
|
7336 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
7337 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
7338 * word).
|
|
|
7339 */
|
|
|
7340 depth = 0;
|
|
375
|
7341 sp = &stack[0];
|
|
|
7342 sp->ts_state = STATE_START;
|
|
|
7343 sp->ts_score = 0;
|
|
|
7344 sp->ts_curi = 1;
|
|
|
7345 sp->ts_fidx = 0;
|
|
|
7346 sp->ts_fidxtry = 0;
|
|
|
7347 sp->ts_twordlen = 0;
|
|
|
7348 sp->ts_arridx = 0;
|
|
330
|
7349 #ifdef FEAT_MBYTE
|
|
375
|
7350 sp->ts_tcharlen = 0;
|
|
330
|
7351 #endif
|
|
|
7352
|
|
|
7353 /*
|
|
375
|
7354 * When there are postponed prefixes we need to use these first. At
|
|
|
7355 * the end of the prefix we continue in the case-fold tree.
|
|
|
7356 */
|
|
|
7357 fbyts = lp->lp_slang->sl_fbyts;
|
|
|
7358 fidxs = lp->lp_slang->sl_fidxs;
|
|
|
7359 pbyts = lp->lp_slang->sl_pbyts;
|
|
|
7360 pidxs = lp->lp_slang->sl_pidxs;
|
|
|
7361 if (pbyts != NULL)
|
|
|
7362 {
|
|
|
7363 byts = pbyts;
|
|
|
7364 idxs = pidxs;
|
|
|
7365 sp->ts_prefixdepth = PREFIXTREE;
|
|
|
7366 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
7367 }
|
|
|
7368 else
|
|
|
7369 {
|
|
|
7370 byts = fbyts;
|
|
|
7371 idxs = fidxs;
|
|
|
7372 sp->ts_prefixdepth = NOPREFIX;
|
|
|
7373 }
|
|
|
7374
|
|
|
7375 /*
|
|
330
|
7376 * Loop to find all suggestions. At each round we either:
|
|
|
7377 * - For the current state try one operation, advance "ts_curi",
|
|
|
7378 * increase "depth".
|
|
|
7379 * - When a state is done go to the next, set "ts_state".
|
|
|
7380 * - When all states are tried decrease "depth".
|
|
|
7381 */
|
|
323
|
7382 while (depth >= 0 && !got_int)
|
|
|
7383 {
|
|
|
7384 sp = &stack[depth];
|
|
|
7385 switch (sp->ts_state)
|
|
|
7386 {
|
|
|
7387 case STATE_START:
|
|
375
|
7388 case STATE_NOPREFIX:
|
|
323
|
7389 /*
|
|
|
7390 * Start of node: Deal with NUL bytes, which means
|
|
|
7391 * tword[] may end here.
|
|
|
7392 */
|
|
|
7393 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
7394 len = byts[arridx]; /* bytes in this node */
|
|
|
7395 arridx += sp->ts_curi; /* index of current byte */
|
|
|
7396
|
|
375
|
7397 if (sp->ts_prefixdepth == PREFIXTREE)
|
|
|
7398 {
|
|
|
7399 /* Skip over the NUL bytes, we use them later. */
|
|
|
7400 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
7401 ;
|
|
|
7402 sp->ts_curi += n;
|
|
|
7403
|
|
390
|
7404 /* Always past NUL bytes now. */
|
|
|
7405 n = (int)sp->ts_state;
|
|
|
7406 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7407 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
7408
|
|
375
|
7409 /* At end of a prefix or at start of prefixtree: check for
|
|
|
7410 * following word. */
|
|
390
|
7411 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
7412 {
|
|
455
|
7413 /* Set su->su_badflags to the caps type at this
|
|
|
7414 * position. Use the caps type until here for the
|
|
|
7415 * prefix itself. */
|
|
|
7416 #ifdef FEAT_MBYTE
|
|
|
7417 if (has_mbyte)
|
|
|
7418 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
7419 else
|
|
|
7420 #endif
|
|
|
7421 n = sp->ts_fidx;
|
|
474
|
7422 flags = badword_captype(su->su_badptr,
|
|
|
7423 su->su_badptr + n);
|
|
|
7424 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
7425 su->su_badptr + su->su_badlen);
|
|
375
|
7426 ++depth;
|
|
|
7427 stack[depth] = stack[depth - 1];
|
|
|
7428 sp = &stack[depth];
|
|
|
7429 sp->ts_prefixdepth = depth - 1;
|
|
|
7430 byts = fbyts;
|
|
|
7431 idxs = fidxs;
|
|
|
7432 sp->ts_state = STATE_START;
|
|
|
7433 sp->ts_curi = 1; /* start just after length byte */
|
|
|
7434 sp->ts_arridx = 0;
|
|
|
7435
|
|
455
|
7436 /* Move the prefix to preword[] with the right case
|
|
|
7437 * and make find_keepcap_word() works. */
|
|
|
7438 splitoff = sp->ts_twordlen;
|
|
|
7439 tword[splitoff] = NUL;
|
|
|
7440 make_case_word(tword, preword, flags);
|
|
|
7441 prewordlen = STRLEN(preword);
|
|
375
|
7442 }
|
|
|
7443 break;
|
|
|
7444 }
|
|
|
7445
|
|
346
|
7446 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
7447 {
|
|
|
7448 /* Past bytes in node and/or past NUL bytes. */
|
|
|
7449 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7450 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7451 break;
|
|
|
7452 }
|
|
|
7453
|
|
|
7454 /*
|
|
|
7455 * End of word in tree.
|
|
|
7456 */
|
|
|
7457 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
7458
|
|
324
|
7459 flags = (int)idxs[arridx];
|
|
323
|
7460
|
|
375
|
7461 if (sp->ts_prefixdepth < MAXWLEN)
|
|
|
7462 {
|
|
|
7463 /* There was a prefix before the word. Check that the
|
|
|
7464 * prefix can be used with this word. */
|
|
|
7465 /* Count the length of the NULs in the prefix. If there
|
|
|
7466 * are none this must be the first try without a prefix.
|
|
|
7467 */
|
|
|
7468 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
7469 len = pbyts[n++];
|
|
|
7470 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
7471 ;
|
|
|
7472 if (c > 0)
|
|
|
7473 {
|
|
390
|
7474 /* The prefix ID is stored three bytes above the
|
|
|
7475 * flags. */
|
|
|
7476 c = valid_word_prefix(c, n, flags,
|
|
455
|
7477 tword + splitoff, lp->lp_slang, FALSE);
|
|
375
|
7478 if (c == 0)
|
|
|
7479 break;
|
|
|
7480
|
|
|
7481 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
7482 if (c & WF_RAREPFX)
|
|
|
7483 flags |= WF_RARE;
|
|
|
7484 }
|
|
|
7485 }
|
|
|
7486
|
|
323
|
7487 /*
|
|
|
7488 * Form the word with proper case in preword.
|
|
|
7489 * If there is a word from a previous split, append.
|
|
|
7490 */
|
|
|
7491 tword[sp->ts_twordlen] = NUL;
|
|
|
7492 if (flags & WF_KEEPCAP)
|
|
|
7493 /* Must find the word in the keep-case tree. */
|
|
|
7494 find_keepcap_word(lp->lp_slang, tword + splitoff,
|
|
|
7495 preword + prewordlen);
|
|
|
7496 else
|
|
346
|
7497 {
|
|
323
|
7498 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
7499 * use that for the goodword too. But if the badword is
|
|
|
7500 * allcap and it's only one char long use onecap. */
|
|
|
7501 c = su->su_badflags;
|
|
|
7502 if ((c & WF_ALLCAP)
|
|
|
7503 #ifdef FEAT_MBYTE
|
|
474
|
7504 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
7505 #else
|
|
|
7506 && su->su_badlen == 1
|
|
|
7507 #endif
|
|
|
7508 )
|
|
|
7509 c = WF_ONECAP;
|
|
323
|
7510 make_case_word(tword + splitoff,
|
|
346
|
7511 preword + prewordlen, flags | c);
|
|
|
7512 }
|
|
323
|
7513
|
|
|
7514 /* Don't use a banned word. It may appear again as a good
|
|
|
7515 * word, thus remember it. */
|
|
|
7516 if (flags & WF_BANNED)
|
|
|
7517 {
|
|
|
7518 add_banned(su, preword + prewordlen);
|
|
|
7519 break;
|
|
|
7520 }
|
|
372
|
7521 if (was_banned(su, preword + prewordlen)
|
|
|
7522 || was_banned(su, preword))
|
|
323
|
7523 break;
|
|
|
7524
|
|
|
7525 newscore = 0;
|
|
|
7526 if ((flags & WF_REGION)
|
|
390
|
7527 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
7528 newscore += SCORE_REGION;
|
|
|
7529 if (flags & WF_RARE)
|
|
|
7530 newscore += SCORE_RARE;
|
|
|
7531
|
|
346
|
7532 if (!spell_valid_case(su->su_badflags,
|
|
323
|
7533 captype(preword + prewordlen, NULL)))
|
|
|
7534 newscore += SCORE_ICASE;
|
|
|
7535
|
|
346
|
7536 if ((fword[sp->ts_fidx] == NUL
|
|
376
|
7537 || !spell_iswordp(fword + sp->ts_fidx, curbuf))
|
|
346
|
7538 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
7539 {
|
|
366
|
7540 /* The badword also ends: add suggestions. Give a penalty
|
|
|
7541 * when changing non-word char to word char, e.g., "thes,"
|
|
|
7542 * -> "these". */
|
|
|
7543 p = fword + sp->ts_fidx;
|
|
|
7544 #ifdef FEAT_MBYTE
|
|
|
7545 if (has_mbyte)
|
|
|
7546 mb_ptr_back(fword, p);
|
|
|
7547 else
|
|
|
7548 #endif
|
|
|
7549 --p;
|
|
376
|
7550 if (!spell_iswordp(p, curbuf))
|
|
366
|
7551 {
|
|
|
7552 p = preword + STRLEN(preword);
|
|
|
7553 #ifdef FEAT_MBYTE
|
|
|
7554 if (has_mbyte)
|
|
|
7555 mb_ptr_back(preword, p);
|
|
|
7556 else
|
|
|
7557 #endif
|
|
|
7558 --p;
|
|
376
|
7559 if (spell_iswordp(p, curbuf))
|
|
366
|
7560 newscore += SCORE_NONWORD;
|
|
|
7561 }
|
|
|
7562
|
|
344
|
7563 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
7564 sp->ts_fidx - repextra,
|
|
351
|
7565 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
7566 }
|
|
330
|
7567 else if (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
7568 #ifdef FEAT_MBYTE
|
|
|
7569 /* Don't split halfway a character. */
|
|
|
7570 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
7571 #endif
|
|
|
7572 )
|
|
323
|
7573 {
|
|
|
7574 /* The word in the tree ends but the badword
|
|
|
7575 * continues: try inserting a space and check that a valid
|
|
|
7576 * words starts at fword[sp->ts_fidx]. */
|
|
|
7577 if (try_deeper(su, stack, depth, newscore + SCORE_SPLIT))
|
|
|
7578 {
|
|
|
7579 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
|
7580 sp->ts_save_prewordlen = prewordlen;
|
|
346
|
7581 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7582 sp->ts_save_splitoff = splitoff;
|
|
376
|
7583 sp->ts_state = STATE_SPLITUNDO;
|
|
|
7584
|
|
|
7585 ++depth;
|
|
|
7586 sp = &stack[depth];
|
|
323
|
7587
|
|
|
7588 /* Append a space to preword. */
|
|
|
7589 STRCAT(preword, " ");
|
|
|
7590 prewordlen = STRLEN(preword);
|
|
|
7591 splitoff = sp->ts_twordlen;
|
|
376
|
7592
|
|
|
7593 /* If the badword has a non-word character at this
|
|
|
7594 * position skip it. That means replacing the
|
|
|
7595 * non-word character with a space. */
|
|
|
7596 if (!spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
7597 {
|
|
|
7598 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
7599 #ifdef FEAT_MBYTE
|
|
|
7600 if (has_mbyte)
|
|
|
7601 sp->ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
|
7602 else
|
|
|
7603 #endif
|
|
|
7604 ++sp->ts_fidx;
|
|
|
7605 }
|
|
455
|
7606
|
|
|
7607 /* set su->su_badflags to the caps type at this
|
|
|
7608 * position */
|
|
|
7609
|
|
324
|
7610 #ifdef FEAT_MBYTE
|
|
|
7611 if (has_mbyte)
|
|
455
|
7612 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
7613 else
|
|
|
7614 #endif
|
|
455
|
7615 n = sp->ts_fidx;
|
|
474
|
7616 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
7617 su->su_badptr + su->su_badlen);
|
|
323
|
7618
|
|
|
7619 /* Restart at top of the tree. */
|
|
376
|
7620 sp->ts_arridx = 0;
|
|
323
|
7621 }
|
|
|
7622 }
|
|
|
7623 break;
|
|
|
7624
|
|
|
7625 case STATE_SPLITUNDO:
|
|
346
|
7626 /* Undo the changes done for word split. */
|
|
|
7627 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
7628 splitoff = sp->ts_save_splitoff;
|
|
|
7629 prewordlen = sp->ts_save_prewordlen;
|
|
|
7630
|
|
|
7631 /* Continue looking for NUL bytes. */
|
|
|
7632 sp->ts_state = STATE_START;
|
|
|
7633 break;
|
|
|
7634
|
|
|
7635 case STATE_ENDNUL:
|
|
|
7636 /* Past the NUL bytes in the node. */
|
|
455
|
7637 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
7638 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
7639 {
|
|
|
7640 /* The badword ends, can't use the bytes in this node. */
|
|
|
7641 sp->ts_state = STATE_DEL;
|
|
|
7642 break;
|
|
|
7643 }
|
|
|
7644 sp->ts_state = STATE_PLAIN;
|
|
|
7645 /*FALLTHROUGH*/
|
|
|
7646
|
|
|
7647 case STATE_PLAIN:
|
|
|
7648 /*
|
|
|
7649 * Go over all possible bytes at this node, add each to
|
|
|
7650 * tword[] and use child node. "ts_curi" is the index.
|
|
|
7651 */
|
|
|
7652 arridx = sp->ts_arridx;
|
|
|
7653 if (sp->ts_curi > byts[arridx])
|
|
|
7654 {
|
|
|
7655 /* Done all bytes at this node, do next state. When still
|
|
|
7656 * at already changed bytes skip the other tricks. */
|
|
|
7657 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7658 sp->ts_state = STATE_DEL;
|
|
|
7659 else
|
|
|
7660 sp->ts_state = STATE_FINAL;
|
|
|
7661 }
|
|
|
7662 else
|
|
|
7663 {
|
|
|
7664 arridx += sp->ts_curi++;
|
|
|
7665 c = byts[arridx];
|
|
|
7666
|
|
|
7667 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
7668 * the byte in the bad word adjust the score. But don't
|
|
|
7669 * even try when the byte was already changed. */
|
|
330
|
7670 if (c == fword[sp->ts_fidx]
|
|
|
7671 #ifdef FEAT_MBYTE
|
|
|
7672 || (sp->ts_tcharlen > 0
|
|
|
7673 && sp->ts_isdiff != DIFF_NONE)
|
|
|
7674 #endif
|
|
|
7675 )
|
|
323
|
7676 newscore = 0;
|
|
|
7677 else
|
|
|
7678 newscore = SCORE_SUBST;
|
|
|
7679 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7680 && try_deeper(su, stack, depth, newscore))
|
|
|
7681 {
|
|
|
7682 ++depth;
|
|
330
|
7683 sp = &stack[depth];
|
|
|
7684 ++sp->ts_fidx;
|
|
|
7685 tword[sp->ts_twordlen++] = c;
|
|
|
7686 sp->ts_arridx = idxs[arridx];
|
|
|
7687 #ifdef FEAT_MBYTE
|
|
|
7688 if (newscore == SCORE_SUBST)
|
|
|
7689 sp->ts_isdiff = DIFF_YES;
|
|
|
7690 if (has_mbyte)
|
|
|
7691 {
|
|
|
7692 /* Multi-byte characters are a bit complicated to
|
|
|
7693 * handle: They differ when any of the bytes
|
|
|
7694 * differ and then their length may also differ. */
|
|
|
7695 if (sp->ts_tcharlen == 0)
|
|
|
7696 {
|
|
|
7697 /* First byte. */
|
|
|
7698 sp->ts_tcharidx = 0;
|
|
|
7699 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
7700 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
7701 sp->ts_isdiff = (newscore != 0)
|
|
|
7702 ? DIFF_YES : DIFF_NONE;
|
|
|
7703 }
|
|
|
7704 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
7705 /* When inserting trail bytes don't advance in
|
|
|
7706 * the bad word. */
|
|
|
7707 --sp->ts_fidx;
|
|
|
7708 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
7709 {
|
|
|
7710 /* Last byte of character. */
|
|
|
7711 if (sp->ts_isdiff == DIFF_YES)
|
|
|
7712 {
|
|
|
7713 /* Correct ts_fidx for the byte length of
|
|
|
7714 * the character (we didn't check that
|
|
|
7715 * before). */
|
|
|
7716 sp->ts_fidx = sp->ts_fcharstart
|
|
|
7717 + MB_BYTE2LEN(
|
|
|
7718 fword[sp->ts_fcharstart]);
|
|
|
7719
|
|
|
7720 /* For a similar character adjust score
|
|
|
7721 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
|
7722 if (lp->lp_slang->sl_has_map
|
|
|
7723 && similar_chars(lp->lp_slang,
|
|
|
7724 mb_ptr2char(tword
|
|
|
7725 + sp->ts_twordlen
|
|
|
7726 - sp->ts_tcharlen),
|
|
|
7727 mb_ptr2char(fword
|
|
|
7728 + sp->ts_fcharstart)))
|
|
|
7729 sp->ts_score -=
|
|
|
7730 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7731 }
|
|
358
|
7732 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
7733 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
7734 {
|
|
|
7735 /* If the previous character was the same,
|
|
|
7736 * thus doubling a character, give a bonus
|
|
|
7737 * to the score. */
|
|
|
7738 p = tword + sp->ts_twordlen
|
|
|
7739 - sp->ts_tcharlen;
|
|
|
7740 c = mb_ptr2char(p);
|
|
|
7741 mb_ptr_back(tword, p);
|
|
|
7742 if (c == mb_ptr2char(p))
|
|
|
7743 sp->ts_score -= SCORE_INS
|
|
|
7744 - SCORE_INSDUP;
|
|
|
7745 }
|
|
330
|
7746
|
|
|
7747 /* Starting a new char, reset the length. */
|
|
|
7748 sp->ts_tcharlen = 0;
|
|
|
7749 }
|
|
|
7750 }
|
|
|
7751 else
|
|
|
7752 #endif
|
|
|
7753 {
|
|
|
7754 /* If we found a similar char adjust the score.
|
|
|
7755 * We do this after calling try_deeper() because
|
|
|
7756 * it's slow. */
|
|
|
7757 if (newscore != 0
|
|
|
7758 && lp->lp_slang->sl_has_map
|
|
|
7759 && similar_chars(lp->lp_slang,
|
|
|
7760 c, fword[sp->ts_fidx - 1]))
|
|
|
7761 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7762 }
|
|
323
|
7763 }
|
|
|
7764 }
|
|
|
7765 break;
|
|
|
7766
|
|
|
7767 case STATE_DEL:
|
|
330
|
7768 #ifdef FEAT_MBYTE
|
|
|
7769 /* When past the first byte of a multi-byte char don't try
|
|
|
7770 * delete/insert/swap a character. */
|
|
|
7771 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
7772 {
|
|
|
7773 sp->ts_state = STATE_FINAL;
|
|
|
7774 break;
|
|
|
7775 }
|
|
|
7776 #endif
|
|
|
7777 /*
|
|
|
7778 * Try skipping one character in the bad word (delete it).
|
|
|
7779 */
|
|
323
|
7780 sp->ts_state = STATE_INS;
|
|
|
7781 sp->ts_curi = 1;
|
|
|
7782 if (fword[sp->ts_fidx] != NUL
|
|
|
7783 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
7784 {
|
|
|
7785 ++depth;
|
|
358
|
7786
|
|
|
7787 /* Advance over the character in fword[]. Give a bonus to
|
|
|
7788 * the score if the same character is following "nn" ->
|
|
|
7789 * "n". */
|
|
330
|
7790 #ifdef FEAT_MBYTE
|
|
|
7791 if (has_mbyte)
|
|
358
|
7792 {
|
|
|
7793 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
7794 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
358
|
7795 if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
|
7796 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7797 }
|
|
330
|
7798 else
|
|
|
7799 #endif
|
|
358
|
7800 {
|
|
330
|
7801 ++stack[depth].ts_fidx;
|
|
358
|
7802 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
7803 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7804 }
|
|
323
|
7805 break;
|
|
|
7806 }
|
|
|
7807 /*FALLTHROUGH*/
|
|
|
7808
|
|
|
7809 case STATE_INS:
|
|
330
|
7810 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
7811 * node. */
|
|
|
7812 n = sp->ts_arridx;
|
|
|
7813 if (sp->ts_curi > byts[n])
|
|
|
7814 {
|
|
|
7815 /* Done all bytes at this node, do next state. */
|
|
|
7816 sp->ts_state = STATE_SWAP;
|
|
|
7817 }
|
|
|
7818 else
|
|
|
7819 {
|
|
330
|
7820 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
7821 n += sp->ts_curi++;
|
|
|
7822 c = byts[n];
|
|
|
7823 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
7824 {
|
|
|
7825 ++depth;
|
|
330
|
7826 sp = &stack[depth];
|
|
|
7827 tword[sp->ts_twordlen++] = c;
|
|
|
7828 sp->ts_arridx = idxs[n];
|
|
|
7829 #ifdef FEAT_MBYTE
|
|
|
7830 if (has_mbyte)
|
|
|
7831 {
|
|
|
7832 fl = MB_BYTE2LEN(c);
|
|
|
7833 if (fl > 1)
|
|
|
7834 {
|
|
|
7835 /* There are following bytes for the same
|
|
|
7836 * character. We must find all bytes before
|
|
|
7837 * trying delete/insert/swap/etc. */
|
|
|
7838 sp->ts_tcharlen = fl;
|
|
|
7839 sp->ts_tcharidx = 1;
|
|
|
7840 sp->ts_isdiff = DIFF_INSERT;
|
|
|
7841 }
|
|
|
7842 }
|
|
358
|
7843 else
|
|
|
7844 fl = 1;
|
|
|
7845 if (fl == 1)
|
|
330
|
7846 #endif
|
|
358
|
7847 {
|
|
|
7848 /* If the previous character was the same, thus
|
|
|
7849 * doubling a character, give a bonus to the
|
|
|
7850 * score. */
|
|
|
7851 if (sp->ts_twordlen >= 2
|
|
|
7852 && tword[sp->ts_twordlen - 2] == c)
|
|
|
7853 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
7854 }
|
|
323
|
7855 }
|
|
|
7856 }
|
|
|
7857 break;
|
|
|
7858
|
|
|
7859 case STATE_SWAP:
|
|
330
|
7860 /*
|
|
|
7861 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
7862 * We change "fword" here, it's changed back afterwards.
|
|
|
7863 */
|
|
|
7864 p = fword + sp->ts_fidx;
|
|
|
7865 c = *p;
|
|
|
7866 if (c == NUL)
|
|
|
7867 {
|
|
|
7868 /* End of word, can't swap or replace. */
|
|
|
7869 sp->ts_state = STATE_FINAL;
|
|
|
7870 break;
|
|
|
7871 }
|
|
|
7872 #ifdef FEAT_MBYTE
|
|
|
7873 if (has_mbyte)
|
|
|
7874 {
|
|
474
|
7875 n = mb_cptr2len(p);
|
|
330
|
7876 c = mb_ptr2char(p);
|
|
|
7877 c2 = mb_ptr2char(p + n);
|
|
|
7878 }
|
|
|
7879 else
|
|
|
7880 #endif
|
|
|
7881 c2 = p[1];
|
|
|
7882 if (c == c2)
|
|
323
|
7883 {
|
|
330
|
7884 /* Characters are identical, swap won't do anything. */
|
|
|
7885 sp->ts_state = STATE_SWAP3;
|
|
|
7886 break;
|
|
|
7887 }
|
|
|
7888 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
7889 {
|
|
|
7890 sp->ts_state = STATE_UNSWAP;
|
|
323
|
7891 ++depth;
|
|
330
|
7892 #ifdef FEAT_MBYTE
|
|
|
7893 if (has_mbyte)
|
|
|
7894 {
|
|
|
7895 fl = mb_char2len(c2);
|
|
|
7896 mch_memmove(p, p + n, fl);
|
|
|
7897 mb_char2bytes(c, p + fl);
|
|
|
7898 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7899 }
|
|
|
7900 else
|
|
|
7901 #endif
|
|
|
7902 {
|
|
|
7903 p[0] = c2;
|
|
|
7904 p[1] = c;
|
|
|
7905 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
7906 }
|
|
323
|
7907 }
|
|
|
7908 else
|
|
|
7909 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
7910 sp->ts_state = STATE_REP_INI;
|
|
|
7911 break;
|
|
|
7912
|
|
330
|
7913 case STATE_UNSWAP:
|
|
|
7914 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
7915 p = fword + sp->ts_fidx;
|
|
|
7916 #ifdef FEAT_MBYTE
|
|
|
7917 if (has_mbyte)
|
|
|
7918 {
|
|
|
7919 n = MB_BYTE2LEN(*p);
|
|
|
7920 c = mb_ptr2char(p + n);
|
|
|
7921 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
7922 mb_char2bytes(c, p);
|
|
|
7923 }
|
|
|
7924 else
|
|
|
7925 #endif
|
|
|
7926 {
|
|
|
7927 c = *p;
|
|
|
7928 *p = p[1];
|
|
|
7929 p[1] = c;
|
|
|
7930 }
|
|
|
7931 /*FALLTHROUGH*/
|
|
|
7932
|
|
|
7933 case STATE_SWAP3:
|
|
323
|
7934 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
7935 * "fword" here, it's changed back afterwards. */
|
|
|
7936 p = fword + sp->ts_fidx;
|
|
|
7937 #ifdef FEAT_MBYTE
|
|
|
7938 if (has_mbyte)
|
|
|
7939 {
|
|
474
|
7940 n = mb_cptr2len(p);
|
|
330
|
7941 c = mb_ptr2char(p);
|
|
474
|
7942 fl = mb_cptr2len(p + n);
|
|
330
|
7943 c2 = mb_ptr2char(p + n);
|
|
|
7944 c3 = mb_ptr2char(p + n + fl);
|
|
|
7945 }
|
|
|
7946 else
|
|
|
7947 #endif
|
|
323
|
7948 {
|
|
330
|
7949 c = *p;
|
|
|
7950 c2 = p[1];
|
|
|
7951 c3 = p[2];
|
|
|
7952 }
|
|
|
7953
|
|
|
7954 /* When characters are identical: "121" then SWAP3 result is
|
|
|
7955 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
7956 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
7957 * swapping. Also skip when c3 is NUL. */
|
|
|
7958 if (c == c3 || c3 == NUL)
|
|
|
7959 {
|
|
|
7960 sp->ts_state = STATE_REP_INI;
|
|
|
7961 break;
|
|
|
7962 }
|
|
|
7963 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7964 {
|
|
|
7965 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
7966 ++depth;
|
|
330
|
7967 #ifdef FEAT_MBYTE
|
|
|
7968 if (has_mbyte)
|
|
|
7969 {
|
|
|
7970 tl = mb_char2len(c3);
|
|
|
7971 mch_memmove(p, p + n + fl, tl);
|
|
|
7972 mb_char2bytes(c2, p + tl);
|
|
|
7973 mb_char2bytes(c, p + fl + tl);
|
|
|
7974 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
7975 }
|
|
|
7976 else
|
|
|
7977 #endif
|
|
|
7978 {
|
|
|
7979 p[0] = p[2];
|
|
|
7980 p[2] = c;
|
|
|
7981 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7982 }
|
|
323
|
7983 }
|
|
|
7984 else
|
|
|
7985 sp->ts_state = STATE_REP_INI;
|
|
|
7986 break;
|
|
|
7987
|
|
330
|
7988 case STATE_UNSWAP3:
|
|
|
7989 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
7990 p = fword + sp->ts_fidx;
|
|
|
7991 #ifdef FEAT_MBYTE
|
|
|
7992 if (has_mbyte)
|
|
|
7993 {
|
|
|
7994 n = MB_BYTE2LEN(*p);
|
|
|
7995 c2 = mb_ptr2char(p + n);
|
|
|
7996 fl = MB_BYTE2LEN(p[n]);
|
|
|
7997 c = mb_ptr2char(p + n + fl);
|
|
|
7998 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
7999 mch_memmove(p + fl + tl, p, n);
|
|
|
8000 mb_char2bytes(c, p);
|
|
|
8001 mb_char2bytes(c2, p + tl);
|
|
|
8002 }
|
|
|
8003 else
|
|
|
8004 #endif
|
|
|
8005 {
|
|
|
8006 c = *p;
|
|
|
8007 *p = p[2];
|
|
|
8008 p[2] = c;
|
|
|
8009 }
|
|
346
|
8010
|
|
330
|
8011 /* Rotate three characters left: "123" -> "231". We change
|
|
|
8012 * "fword" here, it's changed back afterwards. */
|
|
323
|
8013 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
8014 {
|
|
330
|
8015 sp->ts_state = STATE_UNROT3L;
|
|
323
|
8016 ++depth;
|
|
330
|
8017 p = fword + sp->ts_fidx;
|
|
|
8018 #ifdef FEAT_MBYTE
|
|
|
8019 if (has_mbyte)
|
|
|
8020 {
|
|
474
|
8021 n = mb_cptr2len(p);
|
|
330
|
8022 c = mb_ptr2char(p);
|
|
474
|
8023 fl = mb_cptr2len(p + n);
|
|
|
8024 fl += mb_cptr2len(p + n + fl);
|
|
330
|
8025 mch_memmove(p, p + n, fl);
|
|
|
8026 mb_char2bytes(c, p + fl);
|
|
|
8027 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
8028 }
|
|
|
8029 else
|
|
|
8030 #endif
|
|
|
8031 {
|
|
|
8032 c = *p;
|
|
|
8033 *p = p[1];
|
|
|
8034 p[1] = p[2];
|
|
|
8035 p[2] = c;
|
|
|
8036 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8037 }
|
|
323
|
8038 }
|
|
|
8039 else
|
|
|
8040 sp->ts_state = STATE_REP_INI;
|
|
|
8041 break;
|
|
|
8042
|
|
330
|
8043 case STATE_UNROT3L:
|
|
346
|
8044 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
8045 p = fword + sp->ts_fidx;
|
|
|
8046 #ifdef FEAT_MBYTE
|
|
|
8047 if (has_mbyte)
|
|
|
8048 {
|
|
|
8049 n = MB_BYTE2LEN(*p);
|
|
|
8050 n += MB_BYTE2LEN(p[n]);
|
|
|
8051 c = mb_ptr2char(p + n);
|
|
|
8052 tl = MB_BYTE2LEN(p[n]);
|
|
|
8053 mch_memmove(p + tl, p, n);
|
|
|
8054 mb_char2bytes(c, p);
|
|
|
8055 }
|
|
|
8056 else
|
|
|
8057 #endif
|
|
|
8058 {
|
|
|
8059 c = p[2];
|
|
|
8060 p[2] = p[1];
|
|
|
8061 p[1] = *p;
|
|
|
8062 *p = c;
|
|
|
8063 }
|
|
346
|
8064
|
|
323
|
8065 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
8066 * "fword" here, it's changed back afterwards. */
|
|
323
|
8067 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
8068 {
|
|
330
|
8069 sp->ts_state = STATE_UNROT3R;
|
|
323
|
8070 ++depth;
|
|
330
|
8071 p = fword + sp->ts_fidx;
|
|
|
8072 #ifdef FEAT_MBYTE
|
|
|
8073 if (has_mbyte)
|
|
|
8074 {
|
|
474
|
8075 n = mb_cptr2len(p);
|
|
|
8076 n += mb_cptr2len(p + n);
|
|
330
|
8077 c = mb_ptr2char(p + n);
|
|
474
|
8078 tl = mb_cptr2len(p + n);
|
|
330
|
8079 mch_memmove(p + tl, p, n);
|
|
|
8080 mb_char2bytes(c, p);
|
|
|
8081 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
8082 }
|
|
|
8083 else
|
|
|
8084 #endif
|
|
|
8085 {
|
|
|
8086 c = p[2];
|
|
|
8087 p[2] = p[1];
|
|
|
8088 p[1] = *p;
|
|
|
8089 *p = c;
|
|
|
8090 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8091 }
|
|
323
|
8092 }
|
|
|
8093 else
|
|
|
8094 sp->ts_state = STATE_REP_INI;
|
|
|
8095 break;
|
|
|
8096
|
|
330
|
8097 case STATE_UNROT3R:
|
|
346
|
8098 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
8099 p = fword + sp->ts_fidx;
|
|
|
8100 #ifdef FEAT_MBYTE
|
|
|
8101 if (has_mbyte)
|
|
|
8102 {
|
|
|
8103 c = mb_ptr2char(p);
|
|
|
8104 tl = MB_BYTE2LEN(*p);
|
|
|
8105 n = MB_BYTE2LEN(p[tl]);
|
|
|
8106 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
8107 mch_memmove(p, p + tl, n);
|
|
|
8108 mb_char2bytes(c, p + n);
|
|
|
8109 }
|
|
|
8110 else
|
|
|
8111 #endif
|
|
|
8112 {
|
|
|
8113 c = *p;
|
|
|
8114 *p = p[1];
|
|
|
8115 p[1] = p[2];
|
|
|
8116 p[2] = c;
|
|
|
8117 }
|
|
323
|
8118 /*FALLTHROUGH*/
|
|
|
8119
|
|
|
8120 case STATE_REP_INI:
|
|
|
8121 /* Check if matching with REP items from the .aff file would
|
|
|
8122 * work. Quickly skip if there are no REP items or the score
|
|
|
8123 * is going to be too high anyway. */
|
|
|
8124 gap = &lp->lp_slang->sl_rep;
|
|
|
8125 if (gap->ga_len == 0
|
|
|
8126 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
8127 {
|
|
|
8128 sp->ts_state = STATE_FINAL;
|
|
|
8129 break;
|
|
|
8130 }
|
|
|
8131
|
|
|
8132 /* Use the first byte to quickly find the first entry that
|
|
330
|
8133 * may match. If the index is -1 there is none. */
|
|
323
|
8134 sp->ts_curi = lp->lp_slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
8135 if (sp->ts_curi < 0)
|
|
|
8136 {
|
|
|
8137 sp->ts_state = STATE_FINAL;
|
|
|
8138 break;
|
|
|
8139 }
|
|
|
8140
|
|
|
8141 sp->ts_state = STATE_REP;
|
|
|
8142 /*FALLTHROUGH*/
|
|
|
8143
|
|
|
8144 case STATE_REP:
|
|
|
8145 /* Try matching with REP items from the .aff file. For each
|
|
330
|
8146 * match replace the characters and check if the resulting
|
|
|
8147 * word is valid. */
|
|
323
|
8148 p = fword + sp->ts_fidx;
|
|
|
8149
|
|
|
8150 gap = &lp->lp_slang->sl_rep;
|
|
|
8151 while (sp->ts_curi < gap->ga_len)
|
|
|
8152 {
|
|
|
8153 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
8154 if (*ftp->ft_from != *p)
|
|
|
8155 {
|
|
|
8156 /* past possible matching entries */
|
|
|
8157 sp->ts_curi = gap->ga_len;
|
|
|
8158 break;
|
|
|
8159 }
|
|
|
8160 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
8161 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
8162 {
|
|
|
8163 /* Need to undo this afterwards. */
|
|
|
8164 sp->ts_state = STATE_REP_UNDO;
|
|
|
8165
|
|
|
8166 /* Change the "from" to the "to" string. */
|
|
|
8167 ++depth;
|
|
|
8168 fl = STRLEN(ftp->ft_from);
|
|
|
8169 tl = STRLEN(ftp->ft_to);
|
|
|
8170 if (fl != tl)
|
|
346
|
8171 {
|
|
323
|
8172 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
8173 repextra += tl - fl;
|
|
|
8174 }
|
|
323
|
8175 mch_memmove(p, ftp->ft_to, tl);
|
|
|
8176 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
8177 #ifdef FEAT_MBYTE
|
|
|
8178 stack[depth].ts_tcharlen = 0;
|
|
|
8179 #endif
|
|
323
|
8180 break;
|
|
|
8181 }
|
|
|
8182 }
|
|
|
8183
|
|
|
8184 if (sp->ts_curi >= gap->ga_len)
|
|
|
8185 /* No (more) matches. */
|
|
|
8186 sp->ts_state = STATE_FINAL;
|
|
|
8187
|
|
|
8188 break;
|
|
|
8189
|
|
|
8190 case STATE_REP_UNDO:
|
|
|
8191 /* Undo a REP replacement and continue with the next one. */
|
|
|
8192 ftp = (fromto_T *)lp->lp_slang->sl_rep.ga_data
|
|
|
8193 + sp->ts_curi - 1;
|
|
|
8194 fl = STRLEN(ftp->ft_from);
|
|
|
8195 tl = STRLEN(ftp->ft_to);
|
|
|
8196 p = fword + sp->ts_fidx;
|
|
|
8197 if (fl != tl)
|
|
346
|
8198 {
|
|
323
|
8199 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
8200 repextra -= tl - fl;
|
|
|
8201 }
|
|
323
|
8202 mch_memmove(p, ftp->ft_from, fl);
|
|
|
8203 sp->ts_state = STATE_REP;
|
|
|
8204 break;
|
|
|
8205
|
|
|
8206 default:
|
|
|
8207 /* Did all possible states at this level, go up one level. */
|
|
|
8208 --depth;
|
|
344
|
8209
|
|
375
|
8210 if (depth >= 0 && stack[depth].ts_prefixdepth == PREFIXTREE)
|
|
|
8211 {
|
|
|
8212 /* Continue in or go back to the prefix tree. */
|
|
|
8213 byts = pbyts;
|
|
|
8214 idxs = pidxs;
|
|
|
8215 splitoff = 0;
|
|
|
8216 }
|
|
|
8217
|
|
344
|
8218 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
8219 line_breakcheck();
|
|
323
|
8220 }
|
|
|
8221 }
|
|
|
8222 }
|
|
|
8223 }
|
|
|
8224
|
|
|
8225 /*
|
|
|
8226 * Try going one level deeper in the tree.
|
|
|
8227 */
|
|
|
8228 static int
|
|
|
8229 try_deeper(su, stack, depth, score_add)
|
|
|
8230 suginfo_T *su;
|
|
|
8231 trystate_T *stack;
|
|
|
8232 int depth;
|
|
|
8233 int score_add;
|
|
|
8234 {
|
|
|
8235 int newscore;
|
|
|
8236
|
|
|
8237 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
8238 newscore = stack[depth].ts_score + score_add;
|
|
|
8239 if (newscore >= su->su_maxscore)
|
|
|
8240 return FALSE;
|
|
|
8241
|
|
330
|
8242 stack[depth + 1] = stack[depth];
|
|
323
|
8243 stack[depth + 1].ts_state = STATE_START;
|
|
|
8244 stack[depth + 1].ts_score = newscore;
|
|
|
8245 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
8246 return TRUE;
|
|
|
8247 }
|
|
|
8248
|
|
455
|
8249 #ifdef FEAT_MBYTE
|
|
|
8250 /*
|
|
|
8251 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
8252 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
8253 */
|
|
|
8254 static int
|
|
|
8255 nofold_len(fword, flen, word)
|
|
|
8256 char_u *fword;
|
|
|
8257 int flen;
|
|
|
8258 char_u *word;
|
|
|
8259 {
|
|
|
8260 char_u *p;
|
|
|
8261 int i = 0;
|
|
|
8262
|
|
|
8263 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
8264 ++i;
|
|
|
8265 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
8266 --i;
|
|
|
8267 return (int)(p - word);
|
|
|
8268 }
|
|
|
8269 #endif
|
|
|
8270
|
|
323
|
8271 /*
|
|
|
8272 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
8273 * words and put it in "kword".
|
|
|
8274 * Theoretically there could be several keep-case words that result in the
|
|
|
8275 * same case-folded word, but we only find one...
|
|
|
8276 */
|
|
|
8277 static void
|
|
|
8278 find_keepcap_word(slang, fword, kword)
|
|
|
8279 slang_T *slang;
|
|
|
8280 char_u *fword;
|
|
|
8281 char_u *kword;
|
|
|
8282 {
|
|
|
8283 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
8284 int depth;
|
|
324
|
8285 idx_T tryidx;
|
|
323
|
8286
|
|
|
8287 /* The following arrays are used at each depth in the tree. */
|
|
324
|
8288 idx_T arridx[MAXWLEN];
|
|
323
|
8289 int round[MAXWLEN];
|
|
|
8290 int fwordidx[MAXWLEN];
|
|
|
8291 int uwordidx[MAXWLEN];
|
|
|
8292 int kwordlen[MAXWLEN];
|
|
|
8293
|
|
|
8294 int flen, ulen;
|
|
|
8295 int l;
|
|
|
8296 int len;
|
|
|
8297 int c;
|
|
324
|
8298 idx_T lo, hi, m;
|
|
323
|
8299 char_u *p;
|
|
|
8300 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
8301 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
8302
|
|
|
8303 if (byts == NULL)
|
|
|
8304 {
|
|
|
8305 /* array is empty: "cannot happen" */
|
|
|
8306 *kword = NUL;
|
|
|
8307 return;
|
|
|
8308 }
|
|
|
8309
|
|
|
8310 /* Make an all-cap version of "fword". */
|
|
|
8311 allcap_copy(fword, uword);
|
|
|
8312
|
|
|
8313 /*
|
|
|
8314 * Each character needs to be tried both case-folded and upper-case.
|
|
|
8315 * All this gets very complicated if we keep in mind that changing case
|
|
|
8316 * may change the byte length of a multi-byte character...
|
|
|
8317 */
|
|
|
8318 depth = 0;
|
|
|
8319 arridx[0] = 0;
|
|
|
8320 round[0] = 0;
|
|
|
8321 fwordidx[0] = 0;
|
|
|
8322 uwordidx[0] = 0;
|
|
|
8323 kwordlen[0] = 0;
|
|
|
8324 while (depth >= 0)
|
|
|
8325 {
|
|
|
8326 if (fword[fwordidx[depth]] == NUL)
|
|
|
8327 {
|
|
|
8328 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
8329 * here we have found a match. */
|
|
|
8330 if (byts[arridx[depth] + 1] == 0)
|
|
|
8331 {
|
|
|
8332 kword[kwordlen[depth]] = NUL;
|
|
|
8333 return;
|
|
|
8334 }
|
|
|
8335
|
|
|
8336 /* kword is getting too long, continue one level up */
|
|
|
8337 --depth;
|
|
|
8338 }
|
|
|
8339 else if (++round[depth] > 2)
|
|
|
8340 {
|
|
|
8341 /* tried both fold-case and upper-case character, continue one
|
|
|
8342 * level up */
|
|
|
8343 --depth;
|
|
|
8344 }
|
|
|
8345 else
|
|
|
8346 {
|
|
|
8347 /*
|
|
|
8348 * round[depth] == 1: Try using the folded-case character.
|
|
|
8349 * round[depth] == 2: Try using the upper-case character.
|
|
|
8350 */
|
|
|
8351 #ifdef FEAT_MBYTE
|
|
|
8352 if (has_mbyte)
|
|
|
8353 {
|
|
474
|
8354 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
8355 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
8356 }
|
|
|
8357 else
|
|
|
8358 #endif
|
|
|
8359 ulen = flen = 1;
|
|
|
8360 if (round[depth] == 1)
|
|
|
8361 {
|
|
|
8362 p = fword + fwordidx[depth];
|
|
|
8363 l = flen;
|
|
|
8364 }
|
|
|
8365 else
|
|
|
8366 {
|
|
|
8367 p = uword + uwordidx[depth];
|
|
|
8368 l = ulen;
|
|
|
8369 }
|
|
|
8370
|
|
|
8371 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
8372 {
|
|
|
8373 /* Perform a binary search in the list of accepted bytes. */
|
|
|
8374 len = byts[tryidx++];
|
|
|
8375 c = *p++;
|
|
|
8376 lo = tryidx;
|
|
|
8377 hi = tryidx + len - 1;
|
|
|
8378 while (lo < hi)
|
|
|
8379 {
|
|
|
8380 m = (lo + hi) / 2;
|
|
|
8381 if (byts[m] > c)
|
|
|
8382 hi = m - 1;
|
|
|
8383 else if (byts[m] < c)
|
|
|
8384 lo = m + 1;
|
|
|
8385 else
|
|
|
8386 {
|
|
|
8387 lo = hi = m;
|
|
|
8388 break;
|
|
|
8389 }
|
|
|
8390 }
|
|
|
8391
|
|
|
8392 /* Stop if there is no matching byte. */
|
|
|
8393 if (hi < lo || byts[lo] != c)
|
|
|
8394 break;
|
|
|
8395
|
|
|
8396 /* Continue at the child (if there is one). */
|
|
|
8397 tryidx = idxs[lo];
|
|
|
8398 }
|
|
|
8399
|
|
|
8400 if (l == 0)
|
|
|
8401 {
|
|
|
8402 /*
|
|
|
8403 * Found the matching char. Copy it to "kword" and go a
|
|
|
8404 * level deeper.
|
|
|
8405 */
|
|
|
8406 if (round[depth] == 1)
|
|
|
8407 {
|
|
|
8408 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
8409 flen);
|
|
|
8410 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
8411 }
|
|
|
8412 else
|
|
|
8413 {
|
|
|
8414 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
8415 ulen);
|
|
|
8416 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
8417 }
|
|
|
8418 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
8419 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
8420
|
|
|
8421 ++depth;
|
|
|
8422 arridx[depth] = tryidx;
|
|
|
8423 round[depth] = 0;
|
|
|
8424 }
|
|
|
8425 }
|
|
|
8426 }
|
|
|
8427
|
|
|
8428 /* Didn't find it: "cannot happen". */
|
|
|
8429 *kword = NUL;
|
|
|
8430 }
|
|
|
8431
|
|
|
8432 /*
|
|
344
|
8433 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
8434 * su->su_sga.
|
|
|
8435 */
|
|
|
8436 static void
|
|
|
8437 score_comp_sal(su)
|
|
|
8438 suginfo_T *su;
|
|
|
8439 {
|
|
|
8440 langp_T *lp;
|
|
|
8441 char_u badsound[MAXWLEN];
|
|
|
8442 int i;
|
|
|
8443 suggest_T *stp;
|
|
|
8444 suggest_T *sstp;
|
|
|
8445 int score;
|
|
|
8446
|
|
|
8447 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
8448 return;
|
|
|
8449
|
|
|
8450 /* Use the sound-folding of the first language that supports it. */
|
|
|
8451 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8452 lp->lp_slang != NULL; ++lp)
|
|
|
8453 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8454 {
|
|
|
8455 /* soundfold the bad word */
|
|
375
|
8456 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8457
|
|
|
8458 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8459 {
|
|
|
8460 stp = &SUG(su->su_ga, i);
|
|
|
8461
|
|
351
|
8462 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
8463 * sound-a-like score. */
|
|
|
8464 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
8465 if (score < SCORE_MAXMAX)
|
|
|
8466 {
|
|
|
8467 /* Add the suggestion. */
|
|
|
8468 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
8469 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
8470 if (sstp->st_word != NULL)
|
|
|
8471 {
|
|
|
8472 sstp->st_score = score;
|
|
|
8473 sstp->st_altscore = 0;
|
|
|
8474 sstp->st_orglen = stp->st_orglen;
|
|
|
8475 ++su->su_sga.ga_len;
|
|
|
8476 }
|
|
|
8477 }
|
|
|
8478 }
|
|
|
8479 break;
|
|
|
8480 }
|
|
|
8481 }
|
|
|
8482
|
|
|
8483 /*
|
|
|
8484 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
8485 * They are intwined.
|
|
|
8486 */
|
|
|
8487 static void
|
|
|
8488 score_combine(su)
|
|
|
8489 suginfo_T *su;
|
|
|
8490 {
|
|
|
8491 int i;
|
|
|
8492 int j;
|
|
|
8493 garray_T ga;
|
|
|
8494 garray_T *gap;
|
|
|
8495 langp_T *lp;
|
|
|
8496 suggest_T *stp;
|
|
|
8497 char_u *p;
|
|
|
8498 char_u badsound[MAXWLEN];
|
|
|
8499 int round;
|
|
|
8500
|
|
|
8501 /* Add the alternate score to su_ga. */
|
|
|
8502 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8503 lp->lp_slang != NULL; ++lp)
|
|
|
8504 {
|
|
|
8505 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8506 {
|
|
|
8507 /* soundfold the bad word */
|
|
375
|
8508 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8509
|
|
|
8510 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8511 {
|
|
|
8512 stp = &SUG(su->su_ga, i);
|
|
351
|
8513 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
8514 badsound);
|
|
344
|
8515 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8516 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
8517 else
|
|
|
8518 stp->st_score = (stp->st_score * 3
|
|
|
8519 + stp->st_altscore) / 4;
|
|
|
8520 stp->st_salscore = FALSE;
|
|
|
8521 }
|
|
|
8522 break;
|
|
|
8523 }
|
|
|
8524 }
|
|
|
8525
|
|
|
8526 /* Add the alternate score to su_sga. */
|
|
|
8527 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8528 {
|
|
|
8529 stp = &SUG(su->su_sga, i);
|
|
|
8530 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
8531 if (stp->st_score == SCORE_MAXMAX)
|
|
|
8532 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
8533 else
|
|
|
8534 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
8535 stp->st_salscore = TRUE;
|
|
|
8536 }
|
|
|
8537
|
|
|
8538 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
8539 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8540 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
8541
|
|
|
8542 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
8543 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
8544 return;
|
|
|
8545
|
|
|
8546 stp = &SUG(ga, 0);
|
|
|
8547 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
8548 {
|
|
|
8549 /* round 1: get a suggestion from su_ga
|
|
|
8550 * round 2: get a suggestion from su_sga */
|
|
|
8551 for (round = 1; round <= 2; ++round)
|
|
|
8552 {
|
|
|
8553 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
8554 if (i < gap->ga_len)
|
|
|
8555 {
|
|
|
8556 /* Don't add a word if it's already there. */
|
|
|
8557 p = SUG(*gap, i).st_word;
|
|
|
8558 for (j = 0; j < ga.ga_len; ++j)
|
|
|
8559 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
8560 break;
|
|
|
8561 if (j == ga.ga_len)
|
|
|
8562 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
8563 else
|
|
|
8564 vim_free(p);
|
|
|
8565 }
|
|
|
8566 }
|
|
|
8567 }
|
|
|
8568
|
|
|
8569 ga_clear(&su->su_ga);
|
|
|
8570 ga_clear(&su->su_sga);
|
|
|
8571
|
|
|
8572 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
8573 if (ga.ga_len > su->su_maxcount)
|
|
|
8574 {
|
|
|
8575 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
8576 vim_free(stp[i].st_word);
|
|
|
8577 ga.ga_len = su->su_maxcount;
|
|
|
8578 }
|
|
|
8579
|
|
|
8580 su->su_ga = ga;
|
|
|
8581 }
|
|
|
8582
|
|
|
8583 /*
|
|
351
|
8584 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
8585 * badword.
|
|
|
8586 */
|
|
|
8587 static int
|
|
|
8588 stp_sal_score(stp, su, slang, badsound)
|
|
|
8589 suggest_T *stp;
|
|
|
8590 suginfo_T *su;
|
|
|
8591 slang_T *slang;
|
|
|
8592 char_u *badsound; /* sound-folded badword */
|
|
|
8593 {
|
|
|
8594 char_u *p;
|
|
|
8595 char_u badsound2[MAXWLEN];
|
|
|
8596 char_u fword[MAXWLEN];
|
|
|
8597 char_u goodsound[MAXWLEN];
|
|
|
8598
|
|
|
8599 if (stp->st_orglen <= su->su_badlen)
|
|
|
8600 p = badsound;
|
|
|
8601 else
|
|
|
8602 {
|
|
|
8603 /* soundfold the bad word with more characters following */
|
|
|
8604 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
8605
|
|
|
8606 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
8607 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
8608 * removing the space. Don't do it when the good word also contains a
|
|
|
8609 * space. */
|
|
|
8610 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
8611 && *skiptowhite(stp->st_word) == NUL)
|
|
|
8612 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
8613 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
8614
|
|
375
|
8615 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
8616 p = badsound2;
|
|
|
8617 }
|
|
|
8618
|
|
375
|
8619 /* Sound-fold the word and compute the score for the difference. */
|
|
|
8620 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
8621
|
|
|
8622 return soundalike_score(goodsound, p);
|
|
|
8623 }
|
|
|
8624
|
|
|
8625 /*
|
|
323
|
8626 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
8627 */
|
|
|
8628 static void
|
|
346
|
8629 suggest_try_soundalike(su)
|
|
323
|
8630 suginfo_T *su;
|
|
|
8631 {
|
|
|
8632 char_u salword[MAXWLEN];
|
|
|
8633 char_u tword[MAXWLEN];
|
|
|
8634 char_u tsalword[MAXWLEN];
|
|
324
|
8635 idx_T arridx[MAXWLEN];
|
|
323
|
8636 int curi[MAXWLEN];
|
|
|
8637 langp_T *lp;
|
|
|
8638 char_u *byts;
|
|
324
|
8639 idx_T *idxs;
|
|
323
|
8640 int depth;
|
|
|
8641 int c;
|
|
324
|
8642 idx_T n;
|
|
323
|
8643 int round;
|
|
|
8644 int flags;
|
|
344
|
8645 int sound_score;
|
|
|
8646
|
|
|
8647 /* Do this for all languages that support sound folding. */
|
|
323
|
8648 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8649 lp->lp_slang != NULL; ++lp)
|
|
|
8650 {
|
|
|
8651 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8652 {
|
|
|
8653 /* soundfold the bad word */
|
|
375
|
8654 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
8655
|
|
|
8656 /*
|
|
|
8657 * Go through the whole tree, soundfold each word and compare.
|
|
|
8658 * round 1: use the case-folded tree.
|
|
|
8659 * round 2: use the keep-case tree.
|
|
|
8660 */
|
|
|
8661 for (round = 1; round <= 2; ++round)
|
|
|
8662 {
|
|
|
8663 if (round == 1)
|
|
|
8664 {
|
|
|
8665 byts = lp->lp_slang->sl_fbyts;
|
|
|
8666 idxs = lp->lp_slang->sl_fidxs;
|
|
|
8667 }
|
|
|
8668 else
|
|
|
8669 {
|
|
|
8670 byts = lp->lp_slang->sl_kbyts;
|
|
|
8671 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
8672 if (byts == NULL) /* no keep-case words */
|
|
|
8673 continue;
|
|
323
|
8674 }
|
|
|
8675
|
|
|
8676 depth = 0;
|
|
|
8677 arridx[0] = 0;
|
|
|
8678 curi[0] = 1;
|
|
|
8679 while (depth >= 0 && !got_int)
|
|
|
8680 {
|
|
|
8681 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
8682 {
|
|
323
|
8683 /* Done all bytes at this node, go up one level. */
|
|
|
8684 --depth;
|
|
351
|
8685 line_breakcheck();
|
|
|
8686 }
|
|
323
|
8687 else
|
|
|
8688 {
|
|
|
8689 /* Do one more byte at this node. */
|
|
|
8690 n = arridx[depth] + curi[depth];
|
|
|
8691 ++curi[depth];
|
|
|
8692 c = byts[n];
|
|
|
8693 if (c == 0)
|
|
|
8694 {
|
|
|
8695 /* End of word, deal with the word. */
|
|
324
|
8696 flags = (int)idxs[n];
|
|
323
|
8697 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
8698 {
|
|
|
8699 tword[depth] = NUL;
|
|
375
|
8700 /* Sound-fold. Only in keep-case tree need to
|
|
|
8701 * case-fold the word. */
|
|
|
8702 spell_soundfold(lp->lp_slang, tword,
|
|
|
8703 round == 1, tsalword);
|
|
323
|
8704
|
|
344
|
8705 /* Compute the edit distance between the
|
|
|
8706 * sound-a-like words. */
|
|
|
8707 sound_score = soundalike_score(salword,
|
|
|
8708 tsalword);
|
|
324
|
8709 if (sound_score < SCORE_MAXMAX)
|
|
323
|
8710 {
|
|
324
|
8711 char_u cword[MAXWLEN];
|
|
|
8712 char_u *p;
|
|
344
|
8713 int score;
|
|
324
|
8714
|
|
388
|
8715 flags |= su->su_badflags;
|
|
351
|
8716 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
8717 {
|
|
324
|
8718 /* Need to fix case according to
|
|
|
8719 * "flags". */
|
|
323
|
8720 make_case_word(tword, cword, flags);
|
|
324
|
8721 p = cword;
|
|
323
|
8722 }
|
|
|
8723 else
|
|
324
|
8724 p = tword;
|
|
|
8725
|
|
344
|
8726 if (sps_flags & SPS_DOUBLE)
|
|
|
8727 add_suggestion(su, &su->su_sga, p,
|
|
346
|
8728 su->su_badlen,
|
|
351
|
8729 sound_score, 0, FALSE);
|
|
344
|
8730 else
|
|
|
8731 {
|
|
|
8732 /* Compute the score. */
|
|
|
8733 score = spell_edit_score(
|
|
|
8734 su->su_badword, p);
|
|
|
8735 if (sps_flags & SPS_BEST)
|
|
|
8736 /* give a bonus for the good word
|
|
|
8737 * sounding the same as the bad
|
|
|
8738 * word */
|
|
|
8739 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8740 su->su_badlen,
|
|
344
|
8741 RESCORE(score, sound_score),
|
|
351
|
8742 sound_score, TRUE);
|
|
344
|
8743 else
|
|
|
8744 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8745 su->su_badlen,
|
|
351
|
8746 score + sound_score, 0, FALSE);
|
|
344
|
8747 }
|
|
323
|
8748 }
|
|
|
8749 }
|
|
|
8750
|
|
|
8751 /* Skip over other NUL bytes. */
|
|
|
8752 while (byts[n + 1] == 0)
|
|
|
8753 {
|
|
|
8754 ++n;
|
|
|
8755 ++curi[depth];
|
|
|
8756 }
|
|
|
8757 }
|
|
|
8758 else
|
|
|
8759 {
|
|
|
8760 /* Normal char, go one level deeper. */
|
|
|
8761 tword[depth++] = c;
|
|
|
8762 arridx[depth] = idxs[n];
|
|
|
8763 curi[depth] = 1;
|
|
|
8764 }
|
|
|
8765 }
|
|
|
8766 }
|
|
|
8767 }
|
|
|
8768 }
|
|
|
8769 }
|
|
|
8770 }
|
|
|
8771
|
|
|
8772 /*
|
|
324
|
8773 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
8774 */
|
|
|
8775 static void
|
|
|
8776 make_case_word(fword, cword, flags)
|
|
|
8777 char_u *fword;
|
|
|
8778 char_u *cword;
|
|
|
8779 int flags;
|
|
|
8780 {
|
|
|
8781 if (flags & WF_ALLCAP)
|
|
|
8782 /* Make it all upper-case */
|
|
|
8783 allcap_copy(fword, cword);
|
|
|
8784 else if (flags & WF_ONECAP)
|
|
|
8785 /* Make the first letter upper-case */
|
|
324
|
8786 onecap_copy(fword, cword, TRUE);
|
|
323
|
8787 else
|
|
|
8788 /* Use goodword as-is. */
|
|
|
8789 STRCPY(cword, fword);
|
|
|
8790 }
|
|
|
8791
|
|
330
|
8792 /*
|
|
|
8793 * Use map string "map" for languages "lp".
|
|
|
8794 */
|
|
|
8795 static void
|
|
|
8796 set_map_str(lp, map)
|
|
|
8797 slang_T *lp;
|
|
|
8798 char_u *map;
|
|
|
8799 {
|
|
|
8800 char_u *p;
|
|
|
8801 int headc = 0;
|
|
|
8802 int c;
|
|
|
8803 int i;
|
|
|
8804
|
|
|
8805 if (*map == NUL)
|
|
|
8806 {
|
|
|
8807 lp->sl_has_map = FALSE;
|
|
|
8808 return;
|
|
|
8809 }
|
|
|
8810 lp->sl_has_map = TRUE;
|
|
|
8811
|
|
|
8812 /* Init the array and hash table empty. */
|
|
|
8813 for (i = 0; i < 256; ++i)
|
|
|
8814 lp->sl_map_array[i] = 0;
|
|
|
8815 #ifdef FEAT_MBYTE
|
|
|
8816 hash_init(&lp->sl_map_hash);
|
|
|
8817 #endif
|
|
|
8818
|
|
|
8819 /*
|
|
|
8820 * The similar characters are stored separated with slashes:
|
|
|
8821 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
8822 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
8823 */
|
|
|
8824 for (p = map; *p != NUL; )
|
|
|
8825 {
|
|
|
8826 #ifdef FEAT_MBYTE
|
|
474
|
8827 c = mb_cptr2char_adv(&p);
|
|
330
|
8828 #else
|
|
|
8829 c = *p++;
|
|
|
8830 #endif
|
|
|
8831 if (c == '/')
|
|
|
8832 headc = 0;
|
|
|
8833 else
|
|
|
8834 {
|
|
|
8835 if (headc == 0)
|
|
|
8836 headc = c;
|
|
|
8837
|
|
|
8838 #ifdef FEAT_MBYTE
|
|
|
8839 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
8840 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
8841 * a NUL. */
|
|
|
8842 if (c >= 256)
|
|
|
8843 {
|
|
|
8844 int cl = mb_char2len(c);
|
|
|
8845 int headcl = mb_char2len(headc);
|
|
|
8846 char_u *b;
|
|
|
8847 hash_T hash;
|
|
|
8848 hashitem_T *hi;
|
|
|
8849
|
|
|
8850 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
8851 if (b == NULL)
|
|
|
8852 return;
|
|
|
8853 mb_char2bytes(c, b);
|
|
|
8854 b[cl] = NUL;
|
|
|
8855 mb_char2bytes(headc, b + cl + 1);
|
|
|
8856 b[cl + 1 + headcl] = NUL;
|
|
|
8857 hash = hash_hash(b);
|
|
|
8858 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
8859 if (HASHITEM_EMPTY(hi))
|
|
|
8860 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
8861 else
|
|
|
8862 {
|
|
|
8863 /* This should have been checked when generating the .spl
|
|
|
8864 * file. */
|
|
|
8865 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
8866 vim_free(b);
|
|
|
8867 }
|
|
|
8868 }
|
|
|
8869 else
|
|
|
8870 #endif
|
|
|
8871 lp->sl_map_array[c] = headc;
|
|
|
8872 }
|
|
|
8873 }
|
|
|
8874 }
|
|
|
8875
|
|
323
|
8876 /*
|
|
|
8877 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
8878 * lines in the .aff file.
|
|
|
8879 */
|
|
|
8880 static int
|
|
|
8881 similar_chars(slang, c1, c2)
|
|
|
8882 slang_T *slang;
|
|
|
8883 int c1;
|
|
|
8884 int c2;
|
|
|
8885 {
|
|
330
|
8886 int m1, m2;
|
|
|
8887 #ifdef FEAT_MBYTE
|
|
|
8888 char_u buf[MB_MAXBYTES];
|
|
|
8889 hashitem_T *hi;
|
|
|
8890
|
|
|
8891 if (c1 >= 256)
|
|
|
8892 {
|
|
|
8893 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
8894 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8895 if (HASHITEM_EMPTY(hi))
|
|
|
8896 m1 = 0;
|
|
|
8897 else
|
|
|
8898 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8899 }
|
|
|
8900 else
|
|
|
8901 #endif
|
|
|
8902 m1 = slang->sl_map_array[c1];
|
|
|
8903 if (m1 == 0)
|
|
323
|
8904 return FALSE;
|
|
330
|
8905
|
|
|
8906
|
|
|
8907 #ifdef FEAT_MBYTE
|
|
|
8908 if (c2 >= 256)
|
|
|
8909 {
|
|
|
8910 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
8911 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8912 if (HASHITEM_EMPTY(hi))
|
|
|
8913 m2 = 0;
|
|
|
8914 else
|
|
|
8915 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8916 }
|
|
|
8917 else
|
|
|
8918 #endif
|
|
|
8919 m2 = slang->sl_map_array[c2];
|
|
|
8920
|
|
|
8921 return m1 == m2;
|
|
323
|
8922 }
|
|
|
8923
|
|
|
8924 /*
|
|
|
8925 * Add a suggestion to the list of suggestions.
|
|
|
8926 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
8927 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
8928 * with spell_edit_score().
|
|
|
8929 */
|
|
|
8930 static void
|
|
351
|
8931 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
8932 suginfo_T *su;
|
|
344
|
8933 garray_T *gap;
|
|
323
|
8934 char_u *goodword;
|
|
346
|
8935 int badlen; /* length of bad word used */
|
|
324
|
8936 int score;
|
|
351
|
8937 int altscore;
|
|
344
|
8938 int had_bonus; /* value for st_had_bonus */
|
|
323
|
8939 {
|
|
|
8940 suggest_T *stp;
|
|
|
8941 int i;
|
|
346
|
8942 char_u *p = NULL;
|
|
|
8943 int c = 0;
|
|
323
|
8944
|
|
|
8945 /* Check that the word wasn't banned. */
|
|
|
8946 if (was_banned(su, goodword))
|
|
|
8947 return;
|
|
|
8948
|
|
346
|
8949 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
8950 * Remove the common part from "goodword". */
|
|
|
8951 i = badlen - su->su_badlen;
|
|
|
8952 if (i > 0)
|
|
|
8953 {
|
|
|
8954 /* This assumes there was no case folding or it didn't change the
|
|
|
8955 * length... */
|
|
|
8956 p = goodword + STRLEN(goodword) - i;
|
|
|
8957 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
8958 {
|
|
|
8959 badlen = su->su_badlen;
|
|
|
8960 c = *p;
|
|
|
8961 *p = NUL;
|
|
|
8962 }
|
|
|
8963 else
|
|
|
8964 p = NULL;
|
|
|
8965 }
|
|
372
|
8966 else if (i < 0)
|
|
|
8967 {
|
|
|
8968 /* When replacing part of the word check that we actually change
|
|
|
8969 * something. For "the the" a suggestion can be replacing the first
|
|
|
8970 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
8971 if (badlen == (int)STRLEN(goodword)
|
|
372
|
8972 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
8973 return;
|
|
|
8974 }
|
|
|
8975
|
|
346
|
8976
|
|
323
|
8977 if (score <= su->su_maxscore)
|
|
|
8978 {
|
|
366
|
8979 /* Check if the word is already there. Also check the length that is
|
|
|
8980 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
8981 * "thes" -> "these". */
|
|
344
|
8982 stp = &SUG(*gap, 0);
|
|
|
8983 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
8984 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
8985 && stp[i].st_orglen == badlen)
|
|
323
|
8986 {
|
|
|
8987 /* Found it. Remember the lowest score. */
|
|
|
8988 if (stp[i].st_score > score)
|
|
324
|
8989 {
|
|
323
|
8990 stp[i].st_score = score;
|
|
376
|
8991 stp[i].st_altscore = altscore;
|
|
324
|
8992 stp[i].st_had_bonus = had_bonus;
|
|
|
8993 }
|
|
323
|
8994 break;
|
|
|
8995 }
|
|
|
8996
|
|
344
|
8997 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
8998 {
|
|
|
8999 /* Add a suggestion. */
|
|
344
|
9000 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
9001 stp->st_word = vim_strsave(goodword);
|
|
|
9002 if (stp->st_word != NULL)
|
|
|
9003 {
|
|
|
9004 stp->st_score = score;
|
|
351
|
9005 stp->st_altscore = altscore;
|
|
324
|
9006 stp->st_had_bonus = had_bonus;
|
|
346
|
9007 stp->st_orglen = badlen;
|
|
344
|
9008 ++gap->ga_len;
|
|
323
|
9009
|
|
|
9010 /* If we have too many suggestions now, sort the list and keep
|
|
|
9011 * the best suggestions. */
|
|
344
|
9012 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
9013 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
9014 SUG_CLEAN_COUNT(su));
|
|
323
|
9015 }
|
|
|
9016 }
|
|
|
9017 }
|
|
346
|
9018
|
|
|
9019 if (p != NULL)
|
|
|
9020 *p = c; /* restore "goodword" */
|
|
323
|
9021 }
|
|
|
9022
|
|
|
9023 /*
|
|
|
9024 * Add a word to be banned.
|
|
|
9025 */
|
|
|
9026 static void
|
|
|
9027 add_banned(su, word)
|
|
|
9028 suginfo_T *su;
|
|
|
9029 char_u *word;
|
|
|
9030 {
|
|
|
9031 char_u *s = vim_strsave(word);
|
|
|
9032 hash_T hash;
|
|
|
9033 hashitem_T *hi;
|
|
|
9034
|
|
|
9035 if (s != NULL)
|
|
|
9036 {
|
|
|
9037 hash = hash_hash(s);
|
|
|
9038 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
9039 if (HASHITEM_EMPTY(hi))
|
|
|
9040 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
9041 else
|
|
|
9042 vim_free(s);
|
|
323
|
9043 }
|
|
|
9044 }
|
|
|
9045
|
|
|
9046 /*
|
|
|
9047 * Return TRUE if a word appears in the list of banned words.
|
|
|
9048 */
|
|
|
9049 static int
|
|
|
9050 was_banned(su, word)
|
|
|
9051 suginfo_T *su;
|
|
|
9052 char_u *word;
|
|
|
9053 {
|
|
324
|
9054 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
9055
|
|
|
9056 return !HASHITEM_EMPTY(hi);
|
|
323
|
9057 }
|
|
|
9058
|
|
|
9059 /*
|
|
|
9060 * Free the banned words in "su".
|
|
|
9061 */
|
|
|
9062 static void
|
|
|
9063 free_banned(su)
|
|
|
9064 suginfo_T *su;
|
|
|
9065 {
|
|
|
9066 int todo;
|
|
|
9067 hashitem_T *hi;
|
|
|
9068
|
|
|
9069 todo = su->su_banned.ht_used;
|
|
|
9070 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
9071 {
|
|
|
9072 if (!HASHITEM_EMPTY(hi))
|
|
|
9073 {
|
|
|
9074 vim_free(hi->hi_key);
|
|
|
9075 --todo;
|
|
|
9076 }
|
|
|
9077 }
|
|
|
9078 hash_clear(&su->su_banned);
|
|
|
9079 }
|
|
|
9080
|
|
324
|
9081 /*
|
|
|
9082 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
9083 * thus only done for the final results.
|
|
|
9084 */
|
|
|
9085 static void
|
|
|
9086 rescore_suggestions(su)
|
|
|
9087 suginfo_T *su;
|
|
|
9088 {
|
|
|
9089 langp_T *lp;
|
|
|
9090 suggest_T *stp;
|
|
|
9091 char_u sal_badword[MAXWLEN];
|
|
|
9092 int i;
|
|
|
9093
|
|
|
9094 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9095 lp->lp_slang != NULL; ++lp)
|
|
|
9096 {
|
|
|
9097 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9098 {
|
|
|
9099 /* soundfold the bad word */
|
|
375
|
9100 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
9101
|
|
|
9102 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9103 {
|
|
344
|
9104 stp = &SUG(su->su_ga, i);
|
|
324
|
9105 if (!stp->st_had_bonus)
|
|
|
9106 {
|
|
351
|
9107 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
9108 lp->lp_slang, sal_badword);
|
|
|
9109 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
9110 stp->st_altscore = SCORE_BIG;
|
|
|
9111 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
9112 }
|
|
|
9113 }
|
|
|
9114 break;
|
|
|
9115 }
|
|
|
9116 }
|
|
|
9117 }
|
|
|
9118
|
|
323
|
9119 static int
|
|
|
9120 #ifdef __BORLANDC__
|
|
|
9121 _RTLENTRYF
|
|
|
9122 #endif
|
|
|
9123 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
9124
|
|
|
9125 /*
|
|
|
9126 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
9127 */
|
|
|
9128 static int
|
|
|
9129 #ifdef __BORLANDC__
|
|
|
9130 _RTLENTRYF
|
|
|
9131 #endif
|
|
|
9132 sug_compare(s1, s2)
|
|
|
9133 const void *s1;
|
|
|
9134 const void *s2;
|
|
|
9135 {
|
|
|
9136 suggest_T *p1 = (suggest_T *)s1;
|
|
|
9137 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
9138 int n = p1->st_score - p2->st_score;
|
|
|
9139
|
|
|
9140 if (n == 0)
|
|
|
9141 return p1->st_altscore - p2->st_altscore;
|
|
|
9142 return n;
|
|
323
|
9143 }
|
|
|
9144
|
|
|
9145 /*
|
|
|
9146 * Cleanup the suggestions:
|
|
|
9147 * - Sort on score.
|
|
|
9148 * - Remove words that won't be displayed.
|
|
344
|
9149 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
9150 */
|
|
344
|
9151 static int
|
|
|
9152 cleanup_suggestions(gap, maxscore, keep)
|
|
|
9153 garray_T *gap;
|
|
|
9154 int maxscore;
|
|
324
|
9155 int keep; /* nr of suggestions to keep */
|
|
323
|
9156 {
|
|
344
|
9157 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
9158 int i;
|
|
|
9159
|
|
|
9160 /* Sort the list. */
|
|
344
|
9161 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
9162
|
|
|
9163 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
9164 if (gap->ga_len > keep)
|
|
323
|
9165 {
|
|
344
|
9166 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
9167 vim_free(stp[i].st_word);
|
|
344
|
9168 gap->ga_len = keep;
|
|
|
9169 return stp[keep - 1].st_score;
|
|
323
|
9170 }
|
|
344
|
9171 return maxscore;
|
|
323
|
9172 }
|
|
|
9173
|
|
372
|
9174 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
9175 /*
|
|
|
9176 * Soundfold a string, for soundfold().
|
|
|
9177 * Result is in allocated memory, NULL for an error.
|
|
|
9178 */
|
|
|
9179 char_u *
|
|
|
9180 eval_soundfold(word)
|
|
|
9181 char_u *word;
|
|
|
9182 {
|
|
|
9183 langp_T *lp;
|
|
|
9184 char_u sound[MAXWLEN];
|
|
|
9185
|
|
|
9186 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
9187 /* Use the sound-folding of the first language that supports it. */
|
|
|
9188 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9189 lp->lp_slang != NULL; ++lp)
|
|
|
9190 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9191 {
|
|
|
9192 /* soundfold the word */
|
|
375
|
9193 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
9194 return vim_strsave(sound);
|
|
|
9195 }
|
|
|
9196
|
|
|
9197 /* No language with sound folding, return word as-is. */
|
|
|
9198 return vim_strsave(word);
|
|
|
9199 }
|
|
|
9200 #endif
|
|
|
9201
|
|
323
|
9202 /*
|
|
|
9203 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
9204 */
|
|
|
9205 static void
|
|
375
|
9206 spell_soundfold(slang, inword, folded, res)
|
|
|
9207 slang_T *slang;
|
|
|
9208 char_u *inword;
|
|
|
9209 int folded; /* "inword" is already case-folded */
|
|
|
9210 char_u *res;
|
|
|
9211 {
|
|
|
9212 char_u fword[MAXWLEN];
|
|
|
9213 char_u *word;
|
|
|
9214
|
|
|
9215 if (slang->sl_sofo)
|
|
|
9216 /* SOFOFROM and SOFOTO used */
|
|
|
9217 spell_soundfold_sofo(slang, inword, res);
|
|
|
9218 else
|
|
|
9219 {
|
|
|
9220 /* SAL items used. Requires the word to be case-folded. */
|
|
|
9221 if (folded)
|
|
|
9222 word = inword;
|
|
|
9223 else
|
|
|
9224 {
|
|
|
9225 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
9226 word = fword;
|
|
|
9227 }
|
|
|
9228
|
|
|
9229 #ifdef FEAT_MBYTE
|
|
|
9230 if (has_mbyte)
|
|
|
9231 spell_soundfold_wsal(slang, word, res);
|
|
|
9232 else
|
|
|
9233 #endif
|
|
|
9234 spell_soundfold_sal(slang, word, res);
|
|
|
9235 }
|
|
|
9236 }
|
|
|
9237
|
|
|
9238 /*
|
|
|
9239 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
9240 * SOFOTO lines.
|
|
|
9241 */
|
|
|
9242 static void
|
|
|
9243 spell_soundfold_sofo(slang, inword, res)
|
|
|
9244 slang_T *slang;
|
|
|
9245 char_u *inword;
|
|
|
9246 char_u *res;
|
|
|
9247 {
|
|
|
9248 char_u *s;
|
|
|
9249 int ri = 0;
|
|
|
9250 int c;
|
|
|
9251
|
|
|
9252 #ifdef FEAT_MBYTE
|
|
|
9253 if (has_mbyte)
|
|
|
9254 {
|
|
|
9255 int prevc = 0;
|
|
|
9256 int *ip;
|
|
|
9257
|
|
|
9258 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
9259 * 255, sl_sal the rest. */
|
|
|
9260 for (s = inword; *s != NUL; )
|
|
|
9261 {
|
|
474
|
9262 c = mb_cptr2char_adv(&s);
|
|
375
|
9263 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9264 c = ' ';
|
|
|
9265 else if (c < 256)
|
|
|
9266 c = slang->sl_sal_first[c];
|
|
|
9267 else
|
|
|
9268 {
|
|
|
9269 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
9270 if (ip == NULL) /* empty list, can't match */
|
|
|
9271 c = NUL;
|
|
|
9272 else
|
|
|
9273 for (;;) /* find "c" in the list */
|
|
|
9274 {
|
|
|
9275 if (*ip == 0) /* not found */
|
|
|
9276 {
|
|
|
9277 c = NUL;
|
|
|
9278 break;
|
|
|
9279 }
|
|
|
9280 if (*ip == c) /* match! */
|
|
|
9281 {
|
|
|
9282 c = ip[1];
|
|
|
9283 break;
|
|
|
9284 }
|
|
|
9285 ip += 2;
|
|
|
9286 }
|
|
|
9287 }
|
|
|
9288
|
|
|
9289 if (c != NUL && c != prevc)
|
|
|
9290 {
|
|
|
9291 ri += mb_char2bytes(c, res + ri);
|
|
|
9292 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
9293 break;
|
|
|
9294 prevc = c;
|
|
|
9295 }
|
|
|
9296 }
|
|
|
9297 }
|
|
|
9298 else
|
|
|
9299 #endif
|
|
|
9300 {
|
|
|
9301 /* The sl_sal_first[] table contains the translation. */
|
|
|
9302 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
9303 {
|
|
|
9304 if (vim_iswhite(c))
|
|
|
9305 c = ' ';
|
|
|
9306 else
|
|
|
9307 c = slang->sl_sal_first[c];
|
|
|
9308 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
9309 res[ri++] = c;
|
|
|
9310 }
|
|
|
9311 }
|
|
|
9312
|
|
|
9313 res[ri] = NUL;
|
|
|
9314 }
|
|
|
9315
|
|
|
9316 static void
|
|
|
9317 spell_soundfold_sal(slang, inword, res)
|
|
323
|
9318 slang_T *slang;
|
|
|
9319 char_u *inword;
|
|
|
9320 char_u *res;
|
|
|
9321 {
|
|
344
|
9322 salitem_T *smp;
|
|
323
|
9323 char_u word[MAXWLEN];
|
|
375
|
9324 char_u *s = inword;
|
|
323
|
9325 char_u *t;
|
|
344
|
9326 char_u *pf;
|
|
323
|
9327 int i, j, z;
|
|
344
|
9328 int reslen;
|
|
323
|
9329 int n, k = 0;
|
|
|
9330 int z0;
|
|
|
9331 int k0;
|
|
|
9332 int n0;
|
|
|
9333 int c;
|
|
|
9334 int pri;
|
|
|
9335 int p0 = -333;
|
|
|
9336 int c0;
|
|
|
9337
|
|
324
|
9338 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
9339 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
9340 if (slang->sl_rem_accents)
|
|
|
9341 {
|
|
|
9342 t = word;
|
|
375
|
9343 while (*s != NUL)
|
|
323
|
9344 {
|
|
324
|
9345 if (vim_iswhite(*s))
|
|
344
|
9346 {
|
|
|
9347 *t++ = ' ';
|
|
|
9348 s = skipwhite(s);
|
|
|
9349 }
|
|
323
|
9350 else
|
|
|
9351 {
|
|
376
|
9352 if (spell_iswordp_nmw(s))
|
|
323
|
9353 *t++ = *s;
|
|
|
9354 ++s;
|
|
|
9355 }
|
|
|
9356 }
|
|
|
9357 *t = NUL;
|
|
|
9358 }
|
|
|
9359 else
|
|
375
|
9360 STRCPY(word, s);
|
|
323
|
9361
|
|
344
|
9362 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
9363
|
|
|
9364 /*
|
|
|
9365 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
9366 * Changed to keep spaces.
|
|
323
|
9367 */
|
|
344
|
9368 i = reslen = z = 0;
|
|
323
|
9369 while ((c = word[i]) != NUL)
|
|
|
9370 {
|
|
344
|
9371 /* Start with the first rule that has the character in the word. */
|
|
323
|
9372 n = slang->sl_sal_first[c];
|
|
|
9373 z0 = 0;
|
|
|
9374
|
|
|
9375 if (n >= 0)
|
|
|
9376 {
|
|
|
9377 /* check all rules for the same letter */
|
|
344
|
9378 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
9379 {
|
|
344
|
9380 /* Quickly skip entries that don't match the word. Most
|
|
|
9381 * entries are less then three chars, optimize for that. */
|
|
|
9382 k = smp[n].sm_leadlen;
|
|
|
9383 if (k > 1)
|
|
323
|
9384 {
|
|
344
|
9385 if (word[i + 1] != s[1])
|
|
|
9386 continue;
|
|
|
9387 if (k > 2)
|
|
|
9388 {
|
|
|
9389 for (j = 2; j < k; ++j)
|
|
|
9390 if (word[i + j] != s[j])
|
|
|
9391 break;
|
|
|
9392 if (j < k)
|
|
|
9393 continue;
|
|
|
9394 }
|
|
323
|
9395 }
|
|
|
9396
|
|
375
|
9397 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
9398 {
|
|
375
|
9399 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
9400 while (*pf != NUL && *pf != word[i + k])
|
|
|
9401 ++pf;
|
|
|
9402 if (*pf == NUL)
|
|
|
9403 continue;
|
|
|
9404 ++k;
|
|
323
|
9405 }
|
|
344
|
9406 s = smp[n].sm_rules;
|
|
|
9407 pri = 5; /* default priority */
|
|
323
|
9408
|
|
|
9409 p0 = *s;
|
|
|
9410 k0 = k;
|
|
|
9411 while (*s == '-' && k > 1)
|
|
|
9412 {
|
|
|
9413 k--;
|
|
|
9414 s++;
|
|
|
9415 }
|
|
|
9416 if (*s == '<')
|
|
|
9417 s++;
|
|
344
|
9418 if (VIM_ISDIGIT(*s))
|
|
323
|
9419 {
|
|
|
9420 /* determine priority */
|
|
|
9421 pri = *s - '0';
|
|
|
9422 s++;
|
|
|
9423 }
|
|
|
9424 if (*s == '^' && *(s + 1) == '^')
|
|
|
9425 s++;
|
|
|
9426
|
|
|
9427 if (*s == NUL
|
|
|
9428 || (*s == '^'
|
|
324
|
9429 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9430 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
9431 && (*(s + 1) != '$'
|
|
376
|
9432 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9433 || (*s == '$' && i > 0
|
|
376
|
9434 && spell_iswordp(word + i - 1, curbuf)
|
|
|
9435 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9436 {
|
|
|
9437 /* search for followup rules, if: */
|
|
|
9438 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9439 c0 = word[i + k - 1];
|
|
|
9440 n0 = slang->sl_sal_first[c0];
|
|
|
9441
|
|
|
9442 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
9443 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
9444 {
|
|
|
9445 /* test follow-up rule for "word[i + k]" */
|
|
344
|
9446 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
9447 {
|
|
344
|
9448 /* Quickly skip entries that don't match the word.
|
|
|
9449 * */
|
|
|
9450 k0 = smp[n0].sm_leadlen;
|
|
|
9451 if (k0 > 1)
|
|
323
|
9452 {
|
|
344
|
9453 if (word[i + k] != s[1])
|
|
|
9454 continue;
|
|
|
9455 if (k0 > 2)
|
|
|
9456 {
|
|
|
9457 pf = word + i + k + 1;
|
|
|
9458 for (j = 2; j < k0; ++j)
|
|
|
9459 if (*pf++ != s[j])
|
|
|
9460 break;
|
|
|
9461 if (j < k0)
|
|
|
9462 continue;
|
|
|
9463 }
|
|
323
|
9464 }
|
|
344
|
9465 k0 += k - 1;
|
|
|
9466
|
|
375
|
9467 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
9468 {
|
|
344
|
9469 /* Check for match with one of the chars in
|
|
375
|
9470 * "sm_oneof". */
|
|
344
|
9471 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9472 ++pf;
|
|
|
9473 if (*pf == NUL)
|
|
|
9474 continue;
|
|
|
9475 ++k0;
|
|
323
|
9476 }
|
|
344
|
9477
|
|
|
9478 p0 = 5;
|
|
|
9479 s = smp[n0].sm_rules;
|
|
323
|
9480 while (*s == '-')
|
|
|
9481 {
|
|
344
|
9482 /* "k0" gets NOT reduced because
|
|
|
9483 * "if (k0 == k)" */
|
|
323
|
9484 s++;
|
|
|
9485 }
|
|
|
9486 if (*s == '<')
|
|
|
9487 s++;
|
|
344
|
9488 if (VIM_ISDIGIT(*s))
|
|
323
|
9489 {
|
|
|
9490 p0 = *s - '0';
|
|
|
9491 s++;
|
|
|
9492 }
|
|
|
9493
|
|
|
9494 if (*s == NUL
|
|
|
9495 /* *s == '^' cuts */
|
|
|
9496 || (*s == '$'
|
|
376
|
9497 && !spell_iswordp(word + i + k0,
|
|
|
9498 curbuf)))
|
|
323
|
9499 {
|
|
|
9500 if (k0 == k)
|
|
|
9501 /* this is just a piece of the string */
|
|
|
9502 continue;
|
|
|
9503
|
|
|
9504 if (p0 < pri)
|
|
|
9505 /* priority too low */
|
|
|
9506 continue;
|
|
|
9507 /* rule fits; stop search */
|
|
|
9508 break;
|
|
|
9509 }
|
|
|
9510 }
|
|
|
9511
|
|
344
|
9512 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
9513 continue;
|
|
|
9514 }
|
|
|
9515
|
|
|
9516 /* replace string */
|
|
344
|
9517 s = smp[n].sm_to;
|
|
389
|
9518 if (s == NULL)
|
|
|
9519 s = (char_u *)"";
|
|
344
|
9520 pf = smp[n].sm_rules;
|
|
|
9521 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
9522 if (p0 == 1 && z == 0)
|
|
|
9523 {
|
|
|
9524 /* rule with '<' is used */
|
|
344
|
9525 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
9526 || res[reslen - 1] == *s))
|
|
|
9527 reslen--;
|
|
323
|
9528 z0 = 1;
|
|
|
9529 z = 1;
|
|
|
9530 k0 = 0;
|
|
372
|
9531 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
9532 {
|
|
|
9533 word[i + k0] = *s;
|
|
|
9534 k0++;
|
|
|
9535 s++;
|
|
|
9536 }
|
|
|
9537 if (k > k0)
|
|
|
9538 mch_memmove(word + i + k0, word + i + k,
|
|
|
9539 STRLEN(word + i + k) + 1);
|
|
|
9540
|
|
|
9541 /* new "actual letter" */
|
|
|
9542 c = word[i];
|
|
|
9543 }
|
|
|
9544 else
|
|
|
9545 {
|
|
|
9546 /* no '<' rule used */
|
|
|
9547 i += k - 1;
|
|
|
9548 z = 0;
|
|
344
|
9549 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
9550 {
|
|
344
|
9551 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
9552 res[reslen++] = *s;
|
|
323
|
9553 s++;
|
|
|
9554 }
|
|
|
9555 /* new "actual letter" */
|
|
|
9556 c = *s;
|
|
344
|
9557 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
9558 {
|
|
|
9559 if (c != NUL)
|
|
372
|
9560 res[reslen++] = c;
|
|
323
|
9561 mch_memmove(word, word + i + 1,
|
|
|
9562 STRLEN(word + i + 1) + 1);
|
|
|
9563 i = 0;
|
|
|
9564 z0 = 1;
|
|
|
9565 }
|
|
|
9566 }
|
|
|
9567 break;
|
|
|
9568 }
|
|
|
9569 }
|
|
|
9570 }
|
|
324
|
9571 else if (vim_iswhite(c))
|
|
|
9572 {
|
|
|
9573 c = ' ';
|
|
|
9574 k = 1;
|
|
|
9575 }
|
|
323
|
9576
|
|
|
9577 if (z0 == 0)
|
|
|
9578 {
|
|
344
|
9579 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9580 && (!slang->sl_collapse || reslen == 0
|
|
|
9581 || res[reslen - 1] != c))
|
|
323
|
9582 /* condense only double letters */
|
|
372
|
9583 res[reslen++] = c;
|
|
323
|
9584
|
|
|
9585 i++;
|
|
|
9586 z = 0;
|
|
|
9587 k = 0;
|
|
|
9588 }
|
|
|
9589 }
|
|
|
9590
|
|
344
|
9591 res[reslen] = NUL;
|
|
323
|
9592 }
|
|
|
9593
|
|
372
|
9594 #ifdef FEAT_MBYTE
|
|
|
9595 /*
|
|
|
9596 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
9597 * Multi-byte version of spell_soundfold().
|
|
|
9598 */
|
|
|
9599 static void
|
|
375
|
9600 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
9601 slang_T *slang;
|
|
|
9602 char_u *inword;
|
|
|
9603 char_u *res;
|
|
|
9604 {
|
|
375
|
9605 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
9606 int word[MAXWLEN];
|
|
|
9607 int wres[MAXWLEN];
|
|
|
9608 int l;
|
|
|
9609 char_u *s;
|
|
|
9610 int *ws;
|
|
|
9611 char_u *t;
|
|
|
9612 int *pf;
|
|
|
9613 int i, j, z;
|
|
|
9614 int reslen;
|
|
|
9615 int n, k = 0;
|
|
|
9616 int z0;
|
|
|
9617 int k0;
|
|
|
9618 int n0;
|
|
|
9619 int c;
|
|
|
9620 int pri;
|
|
|
9621 int p0 = -333;
|
|
|
9622 int c0;
|
|
|
9623 int did_white = FALSE;
|
|
|
9624
|
|
|
9625 /*
|
|
|
9626 * Convert the multi-byte string to a wide-character string.
|
|
|
9627 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
9628 * But keep white space.
|
|
|
9629 */
|
|
|
9630 n = 0;
|
|
|
9631 for (s = inword; *s != NUL; )
|
|
|
9632 {
|
|
|
9633 t = s;
|
|
474
|
9634 c = mb_cptr2char_adv(&s);
|
|
372
|
9635 if (slang->sl_rem_accents)
|
|
|
9636 {
|
|
|
9637 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9638 {
|
|
|
9639 if (did_white)
|
|
|
9640 continue;
|
|
|
9641 c = ' ';
|
|
|
9642 did_white = TRUE;
|
|
|
9643 }
|
|
|
9644 else
|
|
|
9645 {
|
|
|
9646 did_white = FALSE;
|
|
376
|
9647 if (!spell_iswordp_nmw(t))
|
|
372
|
9648 continue;
|
|
|
9649 }
|
|
|
9650 }
|
|
|
9651 word[n++] = c;
|
|
|
9652 }
|
|
|
9653 word[n] = NUL;
|
|
|
9654
|
|
|
9655 /*
|
|
|
9656 * This comes from Aspell phonet.cpp.
|
|
|
9657 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
9658 * Changed to keep spaces.
|
|
|
9659 */
|
|
|
9660 i = reslen = z = 0;
|
|
|
9661 while ((c = word[i]) != NUL)
|
|
|
9662 {
|
|
|
9663 /* Start with the first rule that has the character in the word. */
|
|
|
9664 n = slang->sl_sal_first[c & 0xff];
|
|
|
9665 z0 = 0;
|
|
|
9666
|
|
|
9667 if (n >= 0)
|
|
|
9668 {
|
|
375
|
9669 /* check all rules for the same index byte */
|
|
372
|
9670 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
9671 {
|
|
|
9672 /* Quickly skip entries that don't match the word. Most
|
|
|
9673 * entries are less then three chars, optimize for that. */
|
|
375
|
9674 if (c != ws[0])
|
|
|
9675 continue;
|
|
372
|
9676 k = smp[n].sm_leadlen;
|
|
|
9677 if (k > 1)
|
|
|
9678 {
|
|
|
9679 if (word[i + 1] != ws[1])
|
|
|
9680 continue;
|
|
|
9681 if (k > 2)
|
|
|
9682 {
|
|
|
9683 for (j = 2; j < k; ++j)
|
|
|
9684 if (word[i + j] != ws[j])
|
|
|
9685 break;
|
|
|
9686 if (j < k)
|
|
|
9687 continue;
|
|
|
9688 }
|
|
|
9689 }
|
|
|
9690
|
|
375
|
9691 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
9692 {
|
|
375
|
9693 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
9694 while (*pf != NUL && *pf != word[i + k])
|
|
|
9695 ++pf;
|
|
|
9696 if (*pf == NUL)
|
|
|
9697 continue;
|
|
|
9698 ++k;
|
|
|
9699 }
|
|
|
9700 s = smp[n].sm_rules;
|
|
|
9701 pri = 5; /* default priority */
|
|
|
9702
|
|
|
9703 p0 = *s;
|
|
|
9704 k0 = k;
|
|
|
9705 while (*s == '-' && k > 1)
|
|
|
9706 {
|
|
|
9707 k--;
|
|
|
9708 s++;
|
|
|
9709 }
|
|
|
9710 if (*s == '<')
|
|
|
9711 s++;
|
|
|
9712 if (VIM_ISDIGIT(*s))
|
|
|
9713 {
|
|
|
9714 /* determine priority */
|
|
|
9715 pri = *s - '0';
|
|
|
9716 s++;
|
|
|
9717 }
|
|
|
9718 if (*s == '^' && *(s + 1) == '^')
|
|
|
9719 s++;
|
|
|
9720
|
|
|
9721 if (*s == NUL
|
|
|
9722 || (*s == '^'
|
|
|
9723 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9724 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
9725 && (*(s + 1) != '$'
|
|
376
|
9726 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9727 || (*s == '$' && i > 0
|
|
376
|
9728 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
9729 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9730 {
|
|
|
9731 /* search for followup rules, if: */
|
|
|
9732 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9733 c0 = word[i + k - 1];
|
|
|
9734 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
9735
|
|
|
9736 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
9737 && p0 != '-' && word[i + k] != NUL)
|
|
|
9738 {
|
|
375
|
9739 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
9740 * all entries with the same index byte. */
|
|
372
|
9741 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
9742 == (c0 & 0xff); ++n0)
|
|
|
9743 {
|
|
|
9744 /* Quickly skip entries that don't match the word.
|
|
375
|
9745 */
|
|
|
9746 if (c0 != ws[0])
|
|
|
9747 continue;
|
|
372
|
9748 k0 = smp[n0].sm_leadlen;
|
|
|
9749 if (k0 > 1)
|
|
|
9750 {
|
|
|
9751 if (word[i + k] != ws[1])
|
|
|
9752 continue;
|
|
|
9753 if (k0 > 2)
|
|
|
9754 {
|
|
|
9755 pf = word + i + k + 1;
|
|
|
9756 for (j = 2; j < k0; ++j)
|
|
|
9757 if (*pf++ != ws[j])
|
|
|
9758 break;
|
|
|
9759 if (j < k0)
|
|
|
9760 continue;
|
|
|
9761 }
|
|
|
9762 }
|
|
|
9763 k0 += k - 1;
|
|
|
9764
|
|
375
|
9765 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
9766 {
|
|
|
9767 /* Check for match with one of the chars in
|
|
375
|
9768 * "sm_oneof". */
|
|
372
|
9769 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9770 ++pf;
|
|
|
9771 if (*pf == NUL)
|
|
|
9772 continue;
|
|
|
9773 ++k0;
|
|
|
9774 }
|
|
|
9775
|
|
|
9776 p0 = 5;
|
|
|
9777 s = smp[n0].sm_rules;
|
|
|
9778 while (*s == '-')
|
|
|
9779 {
|
|
|
9780 /* "k0" gets NOT reduced because
|
|
|
9781 * "if (k0 == k)" */
|
|
|
9782 s++;
|
|
|
9783 }
|
|
|
9784 if (*s == '<')
|
|
|
9785 s++;
|
|
|
9786 if (VIM_ISDIGIT(*s))
|
|
|
9787 {
|
|
|
9788 p0 = *s - '0';
|
|
|
9789 s++;
|
|
|
9790 }
|
|
|
9791
|
|
|
9792 if (*s == NUL
|
|
|
9793 /* *s == '^' cuts */
|
|
|
9794 || (*s == '$'
|
|
376
|
9795 && !spell_iswordp_w(word + i + k0,
|
|
|
9796 curbuf)))
|
|
372
|
9797 {
|
|
|
9798 if (k0 == k)
|
|
|
9799 /* this is just a piece of the string */
|
|
|
9800 continue;
|
|
|
9801
|
|
|
9802 if (p0 < pri)
|
|
|
9803 /* priority too low */
|
|
|
9804 continue;
|
|
|
9805 /* rule fits; stop search */
|
|
|
9806 break;
|
|
|
9807 }
|
|
|
9808 }
|
|
|
9809
|
|
|
9810 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
9811 == (c0 & 0xff))
|
|
|
9812 continue;
|
|
|
9813 }
|
|
|
9814
|
|
|
9815 /* replace string */
|
|
|
9816 ws = smp[n].sm_to_w;
|
|
|
9817 s = smp[n].sm_rules;
|
|
|
9818 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
9819 if (p0 == 1 && z == 0)
|
|
|
9820 {
|
|
|
9821 /* rule with '<' is used */
|
|
389
|
9822 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
9823 && (wres[reslen - 1] == c
|
|
372
|
9824 || wres[reslen - 1] == *ws))
|
|
|
9825 reslen--;
|
|
|
9826 z0 = 1;
|
|
|
9827 z = 1;
|
|
|
9828 k0 = 0;
|
|
389
|
9829 if (ws != NULL)
|
|
|
9830 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
9831 {
|
|
|
9832 word[i + k0] = *ws;
|
|
|
9833 k0++;
|
|
|
9834 ws++;
|
|
|
9835 }
|
|
372
|
9836 if (k > k0)
|
|
|
9837 mch_memmove(word + i + k0, word + i + k,
|
|
|
9838 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
9839
|
|
|
9840 /* new "actual letter" */
|
|
|
9841 c = word[i];
|
|
|
9842 }
|
|
|
9843 else
|
|
|
9844 {
|
|
|
9845 /* no '<' rule used */
|
|
|
9846 i += k - 1;
|
|
|
9847 z = 0;
|
|
389
|
9848 if (ws != NULL)
|
|
|
9849 while (*ws != NUL && ws[1] != NUL
|
|
|
9850 && reslen < MAXWLEN)
|
|
|
9851 {
|
|
|
9852 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
9853 wres[reslen++] = *ws;
|
|
|
9854 ws++;
|
|
|
9855 }
|
|
372
|
9856 /* new "actual letter" */
|
|
389
|
9857 if (ws == NULL)
|
|
|
9858 c = NUL;
|
|
|
9859 else
|
|
|
9860 c = *ws;
|
|
372
|
9861 if (strstr((char *)s, "^^") != NULL)
|
|
|
9862 {
|
|
|
9863 if (c != NUL)
|
|
|
9864 wres[reslen++] = c;
|
|
|
9865 mch_memmove(word, word + i + 1,
|
|
|
9866 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
9867 i = 0;
|
|
|
9868 z0 = 1;
|
|
|
9869 }
|
|
|
9870 }
|
|
|
9871 break;
|
|
|
9872 }
|
|
|
9873 }
|
|
|
9874 }
|
|
|
9875 else if (vim_iswhite(c))
|
|
|
9876 {
|
|
|
9877 c = ' ';
|
|
|
9878 k = 1;
|
|
|
9879 }
|
|
|
9880
|
|
|
9881 if (z0 == 0)
|
|
|
9882 {
|
|
|
9883 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9884 && (!slang->sl_collapse || reslen == 0
|
|
|
9885 || wres[reslen - 1] != c))
|
|
|
9886 /* condense only double letters */
|
|
|
9887 wres[reslen++] = c;
|
|
|
9888
|
|
|
9889 i++;
|
|
|
9890 z = 0;
|
|
|
9891 k = 0;
|
|
|
9892 }
|
|
|
9893 }
|
|
|
9894
|
|
|
9895 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
9896 l = 0;
|
|
|
9897 for (n = 0; n < reslen; ++n)
|
|
|
9898 {
|
|
|
9899 l += mb_char2bytes(wres[n], res + l);
|
|
|
9900 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
9901 break;
|
|
|
9902 }
|
|
|
9903 res[l] = NUL;
|
|
|
9904 }
|
|
|
9905 #endif
|
|
|
9906
|
|
324
|
9907 /*
|
|
344
|
9908 * Compute a score for two sound-a-like words.
|
|
|
9909 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
9910 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
9911 * avoiding checks that will not be possible.
|
|
|
9912 */
|
|
|
9913 static int
|
|
351
|
9914 soundalike_score(goodstart, badstart)
|
|
|
9915 char_u *goodstart; /* sound-folded good word */
|
|
|
9916 char_u *badstart; /* sound-folded bad word */
|
|
344
|
9917 {
|
|
351
|
9918 char_u *goodsound = goodstart;
|
|
|
9919 char_u *badsound = badstart;
|
|
|
9920 int goodlen;
|
|
|
9921 int badlen;
|
|
344
|
9922 int n;
|
|
|
9923 char_u *pl, *ps;
|
|
|
9924 char_u *pl2, *ps2;
|
|
351
|
9925 int score = 0;
|
|
|
9926
|
|
|
9927 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
9928 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
9929 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
9930 {
|
|
|
9931 score = SCORE_DEL / 2;
|
|
|
9932 if (*badsound == '*')
|
|
|
9933 ++badsound;
|
|
|
9934 else
|
|
|
9935 ++goodsound;
|
|
|
9936 }
|
|
|
9937
|
|
|
9938 goodlen = STRLEN(goodsound);
|
|
|
9939 badlen = STRLEN(badsound);
|
|
344
|
9940
|
|
|
9941 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
9942 * changes. */
|
|
|
9943 n = goodlen - badlen;
|
|
|
9944 if (n < -2 || n > 2)
|
|
|
9945 return SCORE_MAXMAX;
|
|
|
9946
|
|
|
9947 if (n > 0)
|
|
|
9948 {
|
|
351
|
9949 pl = goodsound; /* goodsound is longest */
|
|
344
|
9950 ps = badsound;
|
|
|
9951 }
|
|
|
9952 else
|
|
|
9953 {
|
|
351
|
9954 pl = badsound; /* badsound is longest */
|
|
344
|
9955 ps = goodsound;
|
|
|
9956 }
|
|
|
9957
|
|
|
9958 /* Skip over the identical part. */
|
|
|
9959 while (*pl == *ps && *pl != NUL)
|
|
|
9960 {
|
|
|
9961 ++pl;
|
|
|
9962 ++ps;
|
|
|
9963 }
|
|
|
9964
|
|
|
9965 switch (n)
|
|
|
9966 {
|
|
|
9967 case -2:
|
|
|
9968 case 2:
|
|
|
9969 /*
|
|
|
9970 * Must delete two characters from "pl".
|
|
|
9971 */
|
|
|
9972 ++pl; /* first delete */
|
|
|
9973 while (*pl == *ps)
|
|
|
9974 {
|
|
|
9975 ++pl;
|
|
|
9976 ++ps;
|
|
|
9977 }
|
|
|
9978 /* strings must be equal after second delete */
|
|
|
9979 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
9980 return score + SCORE_DEL * 2;
|
|
344
|
9981
|
|
|
9982 /* Failed to compare. */
|
|
|
9983 break;
|
|
|
9984
|
|
|
9985 case -1:
|
|
|
9986 case 1:
|
|
|
9987 /*
|
|
|
9988 * Minimal one delete from "pl" required.
|
|
|
9989 */
|
|
|
9990
|
|
|
9991 /* 1: delete */
|
|
|
9992 pl2 = pl + 1;
|
|
|
9993 ps2 = ps;
|
|
|
9994 while (*pl2 == *ps2)
|
|
|
9995 {
|
|
|
9996 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9997 return score + SCORE_DEL;
|
|
344
|
9998 ++pl2;
|
|
|
9999 ++ps2;
|
|
|
10000 }
|
|
|
10001
|
|
|
10002 /* 2: delete then swap, then rest must be equal */
|
|
|
10003 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10004 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10005 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
10006
|
|
|
10007 /* 3: delete then substitute, then the rest must be equal */
|
|
|
10008 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10009 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
10010
|
|
|
10011 /* 4: first swap then delete */
|
|
|
10012 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
10013 {
|
|
|
10014 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
10015 ps2 = ps + 2;
|
|
|
10016 while (*pl2 == *ps2)
|
|
|
10017 {
|
|
|
10018 ++pl2;
|
|
|
10019 ++ps2;
|
|
|
10020 }
|
|
|
10021 /* delete a char and then strings must be equal */
|
|
|
10022 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10023 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
10024 }
|
|
|
10025
|
|
|
10026 /* 5: first substitute then delete */
|
|
|
10027 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
10028 ps2 = ps + 1;
|
|
|
10029 while (*pl2 == *ps2)
|
|
|
10030 {
|
|
|
10031 ++pl2;
|
|
|
10032 ++ps2;
|
|
|
10033 }
|
|
|
10034 /* delete a char and then strings must be equal */
|
|
|
10035 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10036 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
10037
|
|
|
10038 /* Failed to compare. */
|
|
|
10039 break;
|
|
|
10040
|
|
|
10041 case 0:
|
|
|
10042 /*
|
|
|
10043 * Lenghts are equal, thus changes must result in same length: An
|
|
|
10044 * insert is only possible in combination with a delete.
|
|
|
10045 * 1: check if for identical strings
|
|
|
10046 */
|
|
|
10047 if (*pl == NUL)
|
|
351
|
10048 return score;
|
|
344
|
10049
|
|
|
10050 /* 2: swap */
|
|
|
10051 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
10052 {
|
|
|
10053 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
10054 ps2 = ps + 2;
|
|
|
10055 while (*pl2 == *ps2)
|
|
|
10056 {
|
|
|
10057 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10058 return score + SCORE_SWAP;
|
|
344
|
10059 ++pl2;
|
|
|
10060 ++ps2;
|
|
|
10061 }
|
|
|
10062 /* 3: swap and swap again */
|
|
|
10063 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10064 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10065 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
10066
|
|
|
10067 /* 4: swap and substitute */
|
|
|
10068 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10069 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
10070 }
|
|
|
10071
|
|
|
10072 /* 5: substitute */
|
|
|
10073 pl2 = pl + 1;
|
|
|
10074 ps2 = ps + 1;
|
|
|
10075 while (*pl2 == *ps2)
|
|
|
10076 {
|
|
|
10077 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10078 return score + SCORE_SUBST;
|
|
344
|
10079 ++pl2;
|
|
|
10080 ++ps2;
|
|
|
10081 }
|
|
|
10082
|
|
|
10083 /* 6: substitute and swap */
|
|
|
10084 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10085 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10086 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
10087
|
|
|
10088 /* 7: substitute and substitute */
|
|
|
10089 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10090 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
10091
|
|
|
10092 /* 8: insert then delete */
|
|
|
10093 pl2 = pl;
|
|
|
10094 ps2 = ps + 1;
|
|
|
10095 while (*pl2 == *ps2)
|
|
|
10096 {
|
|
|
10097 ++pl2;
|
|
|
10098 ++ps2;
|
|
|
10099 }
|
|
|
10100 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10101 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
10102
|
|
|
10103 /* 9: delete then insert */
|
|
|
10104 pl2 = pl + 1;
|
|
|
10105 ps2 = ps;
|
|
|
10106 while (*pl2 == *ps2)
|
|
|
10107 {
|
|
|
10108 ++pl2;
|
|
|
10109 ++ps2;
|
|
|
10110 }
|
|
|
10111 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
10112 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
10113
|
|
|
10114 /* Failed to compare. */
|
|
|
10115 break;
|
|
|
10116 }
|
|
|
10117
|
|
|
10118 return SCORE_MAXMAX;
|
|
|
10119 }
|
|
324
|
10120
|
|
323
|
10121 /*
|
|
|
10122 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
10123 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
10124 *
|
|
323
|
10125 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
10126 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
10127 * characters.
|
|
323
|
10128 */
|
|
|
10129 static int
|
|
|
10130 spell_edit_score(badword, goodword)
|
|
|
10131 char_u *badword;
|
|
|
10132 char_u *goodword;
|
|
|
10133 {
|
|
|
10134 int *cnt;
|
|
372
|
10135 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
10136 int j, i;
|
|
|
10137 int t;
|
|
|
10138 int bc, gc;
|
|
324
|
10139 int pbc, pgc;
|
|
|
10140 #ifdef FEAT_MBYTE
|
|
|
10141 char_u *p;
|
|
|
10142 int wbadword[MAXWLEN];
|
|
|
10143 int wgoodword[MAXWLEN];
|
|
|
10144
|
|
|
10145 if (has_mbyte)
|
|
|
10146 {
|
|
|
10147 /* Get the characters from the multi-byte strings and put them in an
|
|
|
10148 * int array for easy access. */
|
|
|
10149 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
10150 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
10151 wbadword[badlen++] = 0;
|
|
324
|
10152 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
10153 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
10154 wgoodword[goodlen++] = 0;
|
|
324
|
10155 }
|
|
|
10156 else
|
|
|
10157 #endif
|
|
|
10158 {
|
|
|
10159 badlen = STRLEN(badword) + 1;
|
|
|
10160 goodlen = STRLEN(goodword) + 1;
|
|
|
10161 }
|
|
323
|
10162
|
|
|
10163 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
10164 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
10165 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
10166 TRUE);
|
|
324
|
10167 if (cnt == NULL)
|
|
|
10168 return 0; /* out of memory */
|
|
323
|
10169
|
|
|
10170 CNT(0, 0) = 0;
|
|
|
10171 for (j = 1; j <= goodlen; ++j)
|
|
|
10172 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
10173
|
|
|
10174 for (i = 1; i <= badlen; ++i)
|
|
|
10175 {
|
|
|
10176 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
10177 for (j = 1; j <= goodlen; ++j)
|
|
|
10178 {
|
|
324
|
10179 #ifdef FEAT_MBYTE
|
|
|
10180 if (has_mbyte)
|
|
|
10181 {
|
|
|
10182 bc = wbadword[i - 1];
|
|
|
10183 gc = wgoodword[j - 1];
|
|
|
10184 }
|
|
|
10185 else
|
|
|
10186 #endif
|
|
|
10187 {
|
|
|
10188 bc = badword[i - 1];
|
|
|
10189 gc = goodword[j - 1];
|
|
|
10190 }
|
|
323
|
10191 if (bc == gc)
|
|
|
10192 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
10193 else
|
|
|
10194 {
|
|
|
10195 /* Use a better score when there is only a case difference. */
|
|
324
|
10196 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
10197 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
10198 else
|
|
|
10199 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
10200
|
|
324
|
10201 if (i > 1 && j > 1)
|
|
323
|
10202 {
|
|
324
|
10203 #ifdef FEAT_MBYTE
|
|
|
10204 if (has_mbyte)
|
|
|
10205 {
|
|
|
10206 pbc = wbadword[i - 2];
|
|
|
10207 pgc = wgoodword[j - 2];
|
|
|
10208 }
|
|
|
10209 else
|
|
|
10210 #endif
|
|
|
10211 {
|
|
|
10212 pbc = badword[i - 2];
|
|
|
10213 pgc = goodword[j - 2];
|
|
|
10214 }
|
|
|
10215 if (bc == pgc && pbc == gc)
|
|
|
10216 {
|
|
|
10217 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
10218 if (t < CNT(i, j))
|
|
|
10219 CNT(i, j) = t;
|
|
|
10220 }
|
|
323
|
10221 }
|
|
|
10222 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
10223 if (t < CNT(i, j))
|
|
|
10224 CNT(i, j) = t;
|
|
|
10225 t = SCORE_INS + CNT(i, j - 1);
|
|
|
10226 if (t < CNT(i, j))
|
|
|
10227 CNT(i, j) = t;
|
|
|
10228 }
|
|
|
10229 }
|
|
|
10230 }
|
|
344
|
10231
|
|
|
10232 i = CNT(badlen - 1, goodlen - 1);
|
|
|
10233 vim_free(cnt);
|
|
|
10234 return i;
|
|
323
|
10235 }
|
|
307
|
10236
|
|
351
|
10237 /*
|
|
|
10238 * ":spelldump"
|
|
|
10239 */
|
|
|
10240 /*ARGSUSED*/
|
|
|
10241 void
|
|
|
10242 ex_spelldump(eap)
|
|
|
10243 exarg_T *eap;
|
|
|
10244 {
|
|
|
10245 buf_T *buf = curbuf;
|
|
|
10246 langp_T *lp;
|
|
|
10247 slang_T *slang;
|
|
|
10248 idx_T arridx[MAXWLEN];
|
|
|
10249 int curi[MAXWLEN];
|
|
|
10250 char_u word[MAXWLEN];
|
|
|
10251 int c;
|
|
|
10252 char_u *byts;
|
|
|
10253 idx_T *idxs;
|
|
|
10254 linenr_T lnum = 0;
|
|
|
10255 int round;
|
|
|
10256 int depth;
|
|
|
10257 int n;
|
|
|
10258 int flags;
|
|
381
|
10259 char_u *region_names = NULL; /* region names being used */
|
|
|
10260 int do_region = TRUE; /* dump region names and numbers */
|
|
|
10261 char_u *p;
|
|
351
|
10262
|
|
|
10263 if (no_spell_checking())
|
|
|
10264 return;
|
|
|
10265
|
|
|
10266 /* Create a new empty buffer by splitting the window. */
|
|
|
10267 do_cmdline_cmd((char_u *)"new");
|
|
|
10268 if (!bufempty() || !buf_valid(buf))
|
|
|
10269 return;
|
|
|
10270
|
|
381
|
10271 /* Find out if we can support regions: All languages must support the same
|
|
|
10272 * regions or none at all. */
|
|
|
10273 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
10274 {
|
|
|
10275 p = lp->lp_slang->sl_regions;
|
|
|
10276 if (p[0] != 0)
|
|
|
10277 {
|
|
|
10278 if (region_names == NULL) /* first language with regions */
|
|
|
10279 region_names = p;
|
|
|
10280 else if (STRCMP(region_names, p) != 0)
|
|
|
10281 {
|
|
|
10282 do_region = FALSE; /* region names are different */
|
|
|
10283 break;
|
|
|
10284 }
|
|
|
10285 }
|
|
|
10286 }
|
|
|
10287
|
|
|
10288 if (do_region && region_names != NULL)
|
|
|
10289 {
|
|
|
10290 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
10291 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
10292 }
|
|
|
10293 else
|
|
|
10294 do_region = FALSE;
|
|
|
10295
|
|
|
10296 /*
|
|
|
10297 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
10298 */
|
|
351
|
10299 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
10300 {
|
|
|
10301 slang = lp->lp_slang;
|
|
|
10302
|
|
|
10303 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
10304 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
10305
|
|
|
10306 /* round 1: case-folded tree
|
|
|
10307 * round 2: keep-case tree */
|
|
|
10308 for (round = 1; round <= 2; ++round)
|
|
|
10309 {
|
|
|
10310 if (round == 1)
|
|
|
10311 {
|
|
|
10312 byts = slang->sl_fbyts;
|
|
|
10313 idxs = slang->sl_fidxs;
|
|
|
10314 }
|
|
|
10315 else
|
|
|
10316 {
|
|
|
10317 byts = slang->sl_kbyts;
|
|
|
10318 idxs = slang->sl_kidxs;
|
|
|
10319 }
|
|
|
10320 if (byts == NULL)
|
|
|
10321 continue; /* array is empty */
|
|
|
10322
|
|
|
10323 depth = 0;
|
|
|
10324 arridx[0] = 0;
|
|
|
10325 curi[0] = 1;
|
|
|
10326 while (depth >= 0 && !got_int)
|
|
|
10327 {
|
|
|
10328 if (curi[depth] > byts[arridx[depth]])
|
|
|
10329 {
|
|
|
10330 /* Done all bytes at this node, go up one level. */
|
|
|
10331 --depth;
|
|
|
10332 line_breakcheck();
|
|
|
10333 }
|
|
|
10334 else
|
|
|
10335 {
|
|
|
10336 /* Do one more byte at this node. */
|
|
|
10337 n = arridx[depth] + curi[depth];
|
|
|
10338 ++curi[depth];
|
|
|
10339 c = byts[n];
|
|
|
10340 if (c == 0)
|
|
|
10341 {
|
|
|
10342 /* End of word, deal with the word.
|
|
|
10343 * Don't use keep-case words in the fold-case tree,
|
|
|
10344 * they will appear in the keep-case tree.
|
|
|
10345 * Only use the word when the region matches. */
|
|
|
10346 flags = (int)idxs[n];
|
|
|
10347 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
10348 && (do_region
|
|
|
10349 || (flags & WF_REGION) == 0
|
|
390
|
10350 || (((unsigned)flags >> 16)
|
|
351
|
10351 & lp->lp_region) != 0))
|
|
|
10352 {
|
|
|
10353 word[depth] = NUL;
|
|
381
|
10354 if (!do_region)
|
|
|
10355 flags &= ~WF_REGION;
|
|
355
|
10356
|
|
|
10357 /* Dump the basic word if there is no prefix or
|
|
|
10358 * when it's the first one. */
|
|
390
|
10359 c = (unsigned)flags >> 24;
|
|
355
|
10360 if (c == 0 || curi[depth] == 2)
|
|
|
10361 dump_word(word, round, flags, lnum++);
|
|
351
|
10362
|
|
|
10363 /* Apply the prefix, if there is one. */
|
|
355
|
10364 if (c != 0)
|
|
351
|
10365 lnum = apply_prefixes(slang, word, round,
|
|
|
10366 flags, lnum);
|
|
|
10367 }
|
|
|
10368 }
|
|
|
10369 else
|
|
|
10370 {
|
|
|
10371 /* Normal char, go one level deeper. */
|
|
|
10372 word[depth++] = c;
|
|
|
10373 arridx[depth] = idxs[n];
|
|
|
10374 curi[depth] = 1;
|
|
|
10375 }
|
|
|
10376 }
|
|
|
10377 }
|
|
|
10378 }
|
|
|
10379 }
|
|
|
10380
|
|
|
10381 /* Delete the empty line that we started with. */
|
|
|
10382 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
10383 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
10384
|
|
|
10385 redraw_later(NOT_VALID);
|
|
|
10386 }
|
|
|
10387
|
|
|
10388 /*
|
|
|
10389 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
10390 */
|
|
|
10391 static void
|
|
|
10392 dump_word(word, round, flags, lnum)
|
|
|
10393 char_u *word;
|
|
|
10394 int round;
|
|
|
10395 int flags;
|
|
|
10396 linenr_T lnum;
|
|
|
10397 {
|
|
|
10398 int keepcap = FALSE;
|
|
|
10399 char_u *p;
|
|
|
10400 char_u cword[MAXWLEN];
|
|
381
|
10401 char_u badword[MAXWLEN + 10];
|
|
|
10402 int i;
|
|
351
|
10403
|
|
|
10404 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
10405 {
|
|
|
10406 /* Need to fix case according to "flags". */
|
|
|
10407 make_case_word(word, cword, flags);
|
|
|
10408 p = cword;
|
|
|
10409 }
|
|
|
10410 else
|
|
|
10411 {
|
|
|
10412 p = word;
|
|
389
|
10413 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
10414 || (flags & WF_FIXCAP) != 0))
|
|
351
|
10415 keepcap = TRUE;
|
|
|
10416 }
|
|
|
10417
|
|
381
|
10418 /* Add flags and regions after a slash. */
|
|
|
10419 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
10420 {
|
|
|
10421 STRCPY(badword, p);
|
|
|
10422 STRCAT(badword, "/");
|
|
351
|
10423 if (keepcap)
|
|
|
10424 STRCAT(badword, "=");
|
|
|
10425 if (flags & WF_BANNED)
|
|
|
10426 STRCAT(badword, "!");
|
|
|
10427 else if (flags & WF_RARE)
|
|
|
10428 STRCAT(badword, "?");
|
|
381
|
10429 if (flags & WF_REGION)
|
|
|
10430 for (i = 0; i < 7; ++i)
|
|
390
|
10431 if (flags & (0x10000 << i))
|
|
381
|
10432 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
10433 p = badword;
|
|
|
10434 }
|
|
|
10435
|
|
|
10436 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
10437 }
|
|
|
10438
|
|
|
10439 /*
|
|
372
|
10440 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
10441 * "word" and append a line to the buffer.
|
|
351
|
10442 * Return the updated line number.
|
|
|
10443 */
|
|
|
10444 static linenr_T
|
|
|
10445 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
10446 slang_T *slang;
|
|
|
10447 char_u *word; /* case-folded word */
|
|
|
10448 int round;
|
|
|
10449 int flags; /* flags with prefix ID */
|
|
|
10450 linenr_T startlnum;
|
|
|
10451 {
|
|
|
10452 idx_T arridx[MAXWLEN];
|
|
|
10453 int curi[MAXWLEN];
|
|
|
10454 char_u prefix[MAXWLEN];
|
|
455
|
10455 char_u word_up[MAXWLEN];
|
|
|
10456 int has_word_up = FALSE;
|
|
351
|
10457 int c;
|
|
|
10458 char_u *byts;
|
|
|
10459 idx_T *idxs;
|
|
|
10460 linenr_T lnum = startlnum;
|
|
|
10461 int depth;
|
|
|
10462 int n;
|
|
|
10463 int len;
|
|
|
10464 int i;
|
|
|
10465
|
|
455
|
10466 /* if the word starts with a lower-case letter make the word with an
|
|
|
10467 * upper-case letter in word_up[]. */
|
|
|
10468 c = PTR2CHAR(word);
|
|
|
10469 if (SPELL_TOUPPER(c) != c)
|
|
|
10470 {
|
|
|
10471 onecap_copy(word, word_up, TRUE);
|
|
|
10472 has_word_up = TRUE;
|
|
|
10473 }
|
|
|
10474
|
|
351
|
10475 byts = slang->sl_pbyts;
|
|
|
10476 idxs = slang->sl_pidxs;
|
|
|
10477 if (byts != NULL) /* array not is empty */
|
|
|
10478 {
|
|
|
10479 /*
|
|
|
10480 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
10481 * When at the end of a prefix check that it supports "flags".
|
|
351
|
10482 */
|
|
|
10483 depth = 0;
|
|
|
10484 arridx[0] = 0;
|
|
|
10485 curi[0] = 1;
|
|
|
10486 while (depth >= 0 && !got_int)
|
|
|
10487 {
|
|
390
|
10488 n = arridx[depth];
|
|
|
10489 len = byts[n];
|
|
|
10490 if (curi[depth] > len)
|
|
351
|
10491 {
|
|
|
10492 /* Done all bytes at this node, go up one level. */
|
|
|
10493 --depth;
|
|
|
10494 line_breakcheck();
|
|
|
10495 }
|
|
|
10496 else
|
|
|
10497 {
|
|
|
10498 /* Do one more byte at this node. */
|
|
390
|
10499 n += curi[depth];
|
|
351
|
10500 ++curi[depth];
|
|
|
10501 c = byts[n];
|
|
|
10502 if (c == 0)
|
|
|
10503 {
|
|
|
10504 /* End of prefix, find out how many IDs there are. */
|
|
|
10505 for (i = 1; i < len; ++i)
|
|
|
10506 if (byts[n + i] != 0)
|
|
|
10507 break;
|
|
|
10508 curi[depth] += i - 1;
|
|
|
10509
|
|
455
|
10510 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
10511 if (c != 0)
|
|
351
|
10512 {
|
|
376
|
10513 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
10514 dump_word(prefix, round,
|
|
455
|
10515 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
10516 : flags, lnum++);
|
|
351
|
10517 }
|
|
455
|
10518
|
|
|
10519 /* Check for prefix that matches the word when the
|
|
|
10520 * first letter is upper-case, but only if the prefix has
|
|
|
10521 * a condition. */
|
|
|
10522 if (has_word_up)
|
|
|
10523 {
|
|
|
10524 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
10525 TRUE);
|
|
|
10526 if (c != 0)
|
|
|
10527 {
|
|
|
10528 vim_strncpy(prefix + depth, word_up,
|
|
|
10529 MAXWLEN - depth - 1);
|
|
|
10530 dump_word(prefix, round,
|
|
|
10531 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
10532 : flags, lnum++);
|
|
|
10533 }
|
|
|
10534 }
|
|
351
|
10535 }
|
|
|
10536 else
|
|
|
10537 {
|
|
|
10538 /* Normal char, go one level deeper. */
|
|
|
10539 prefix[depth++] = c;
|
|
|
10540 arridx[depth] = idxs[n];
|
|
|
10541 curi[depth] = 1;
|
|
|
10542 }
|
|
|
10543 }
|
|
|
10544 }
|
|
|
10545 }
|
|
|
10546
|
|
|
10547 return lnum;
|
|
|
10548 }
|
|
|
10549
|
|
236
|
10550 #endif /* FEAT_SYN_HL */
|
