|
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 *
|
|
481
|
38 * There is one additional tree for when not all prefixes are applied when
|
|
339
|
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.
|
|
482
|
54 * Only use it for small word lists! */
|
|
469
|
55 #if 0
|
|
|
56 # define SPELL_PRINTTREE
|
|
482
|
57 #endif
|
|
|
58
|
|
|
59 /* SPELL_COMPRESS_CNT is after how many allocated blocks we compress the tree
|
|
|
60 * to limit the amount of memory used (esp. for Italian and Hungarian). The
|
|
|
61 * amount of memory used for nodes then is SPELL_COMPRESS_CNT times
|
|
|
62 * SBLOCKSIZE.
|
|
|
63 * Then compress again after allocating SPELL_COMPRESS_INC more blocks or
|
|
|
64 * adding SPELL_COMPRESS_ADDED words and running out of memory again. */
|
|
|
65 #define SPELL_COMPRESS_CNT 30000
|
|
|
66 #define SPELL_COMPRESS_INC 100
|
|
|
67 #define SPELL_COMPRESS_ADDED 500000
|
|
469
|
68
|
|
300
|
69 /*
|
|
324
|
70 * Use this to adjust the score after finding suggestions, based on the
|
|
|
71 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
72 * it for every possible suggestion.
|
|
|
73 * Disadvantage: When "the" is typed as "hte" it sounds different and goes
|
|
|
74 * down in the list.
|
|
344
|
75 * Used when 'spellsuggest' is set to "best".
|
|
|
76 */
|
|
|
77 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
78
|
|
|
79 /*
|
|
339
|
80 * Vim spell file format: <HEADER>
|
|
|
81 * <SUGGEST>
|
|
|
82 * <LWORDTREE>
|
|
|
83 * <KWORDTREE>
|
|
|
84 * <PREFIXTREE>
|
|
300
|
85 *
|
|
339
|
86 * <HEADER>: <fileID>
|
|
|
87 * <regioncnt> <regionname> ...
|
|
|
88 * <charflagslen> <charflags>
|
|
|
89 * <fcharslen> <fchars>
|
|
366
|
90 * <midwordlen> <midword>
|
|
481
|
91 * <compoundlen> <compoundtype> <compoundinfo>
|
|
339
|
92 * <prefcondcnt> <prefcond> ...
|
|
300
|
93 *
|
|
481
|
94 * <fileID> 10 bytes "VIMspell10"
|
|
300
|
95 * <regioncnt> 1 byte number of regions following (8 supported)
|
|
307
|
96 * <regionname> 2 bytes Region name: ca, au, etc. Lower case.
|
|
300
|
97 * First <regionname> is region 1.
|
|
|
98 *
|
|
|
99 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
100 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
101 * 0x01 word character CF_WORD
|
|
|
102 * 0x02 upper-case character CF_UPPER
|
|
300
|
103 * <fcharslen> 2 bytes Number of bytes in <fchars>.
|
|
|
104 * <fchars> N bytes Folded characters, first one is for character 128.
|
|
|
105 *
|
|
366
|
106 * <midwordlen> 2 bytes Number of bytes in <midword>.
|
|
|
107 * <midword> N bytes Characters that are word characters only when used
|
|
|
108 * in the middle of a word.
|
|
|
109 *
|
|
481
|
110 * <compoundlen> 2 bytes Number of bytes following for compound info (can
|
|
|
111 * be used to skip it when it's not understood).
|
|
|
112 *
|
|
|
113 * <compoundtype 1 byte 1: compound words using <comp1minlen> and
|
|
|
114 * <comp1flags>
|
|
|
115 *
|
|
|
116 * <comp1minlen> 1 byte minimal word length for compounding
|
|
|
117 *
|
|
|
118 * <comp1flags> N bytes flags used for compounding words
|
|
|
119 *
|
|
|
120 *
|
|
339
|
121 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
122 *
|
|
|
123 * <prefcond> : <condlen> <condstr>
|
|
|
124 *
|
|
|
125 * <condlen> 1 byte Length of <condstr>.
|
|
|
126 *
|
|
|
127 * <condstr> N bytes Condition for the prefix.
|
|
|
128 *
|
|
300
|
129 *
|
|
323
|
130 * <SUGGEST> : <repcount> <rep> ...
|
|
|
131 * <salflags> <salcount> <sal> ...
|
|
|
132 * <maplen> <mapstr>
|
|
|
133 *
|
|
|
134 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
|
135 *
|
|
|
136 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
|
137 *
|
|
|
138 * <repfromlen> 1 byte length of <repfrom>
|
|
|
139 *
|
|
|
140 * <repfrom> N bytes "from" part of replacement
|
|
|
141 *
|
|
|
142 * <reptolen> 1 byte length of <repto>
|
|
|
143 *
|
|
|
144 * <repto> N bytes "to" part of replacement
|
|
300
|
145 *
|
|
323
|
146 * <salflags> 1 byte flags for soundsalike conversion:
|
|
|
147 * SAL_F0LLOWUP
|
|
|
148 * SAL_COLLAPSE
|
|
|
149 * SAL_REM_ACCENTS
|
|
375
|
150 * SAL_SOFO: SOFOFROM and SOFOTO used instead of SAL
|
|
|
151 *
|
|
|
152 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
153 *
|
|
|
154 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
|
155 *
|
|
375
|
156 * <salfromlen> 1-2 bytes length of <salfrom> (2 bytes for SAL_SOFO)
|
|
323
|
157 *
|
|
|
158 * <salfrom> N bytes "from" part of soundsalike
|
|
|
159 *
|
|
375
|
160 * <saltolen> 1-2 bytes length of <salto> (2 bytes for SAL_SOFO)
|
|
323
|
161 *
|
|
|
162 * <salto> N bytes "to" part of soundsalike
|
|
|
163 *
|
|
|
164 * <maplen> 2 bytes length of <mapstr>, MSB first
|
|
|
165 *
|
|
|
166 * <mapstr> N bytes String with sequences of similar characters,
|
|
|
167 * separated by slashes.
|
|
300
|
168 *
|
|
|
169 *
|
|
|
170 * <LWORDTREE>: <wordtree>
|
|
|
171 *
|
|
339
|
172 * <KWORDTREE>: <wordtree>
|
|
|
173 *
|
|
|
174 * <PREFIXTREE>: <wordtree>
|
|
|
175 *
|
|
|
176 *
|
|
300
|
177 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
178 *
|
|
|
179 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
180 *
|
|
|
181 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
182 *
|
|
|
183 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
184 * follow in sorted order.
|
|
|
185 *
|
|
339
|
186 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
481
|
187 * | <flags> [<flags2>] [<region>] [<affixID>]
|
|
|
188 * | [<pflags>] <affixID> <prefcondnr> ]
|
|
300
|
189 *
|
|
|
190 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
191 * BY_NOFLAGS: End of word without flags and for all
|
|
|
192 * regions.
|
|
481
|
193 * For PREFIXTREE <affixID> and
|
|
366
|
194 * <prefcondnr> follow.
|
|
390
|
195 * BY_FLAGS: End of word, <flags> follow.
|
|
481
|
196 * For PREFIXTREE <pflags>, <affixID>
|
|
455
|
197 * and <prefcondnr> follow.
|
|
390
|
198 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
199 * follow. Not used in PREFIXTREE.
|
|
|
200 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
201 * and <xbyte> follow.
|
|
|
202 *
|
|
|
203 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
204 *
|
|
|
205 * <xbyte> 1 byte byte value of the sibling.
|
|
|
206 *
|
|
|
207 * <flags> 1 byte bitmask of:
|
|
|
208 * WF_ALLCAP word must have only capitals
|
|
|
209 * WF_ONECAP first char of word must be capital
|
|
389
|
210 * WF_KEEPCAP keep-case word
|
|
|
211 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
212 * WF_RARE rare word
|
|
389
|
213 * WF_BANNED bad word
|
|
300
|
214 * WF_REGION <region> follows
|
|
481
|
215 * WF_AFX <affixID> follows
|
|
300
|
216 *
|
|
390
|
217 * <flags2> 1 byte Only used when there are postponed prefixes.
|
|
|
218 * Bitmask of:
|
|
|
219 * WF_HAS_AFF >> 8 word includes affix
|
|
|
220 *
|
|
455
|
221 * <pflags> 1 byte bitmask of:
|
|
|
222 * WFP_RARE rare prefix
|
|
|
223 * WFP_NC non-combining prefix
|
|
|
224 * WFP_UP letter after prefix made upper case
|
|
|
225 *
|
|
300
|
226 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
227 * omitted it's valid in all regions.
|
|
|
228 * Lowest bit is for region 1.
|
|
|
229 *
|
|
481
|
230 * <affixID> 1 byte ID of affix that can be used with this word. In
|
|
339
|
231 * PREFIXTREE used for the required prefix ID.
|
|
|
232 *
|
|
|
233 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
234 * from HEADER.
|
|
300
|
235 *
|
|
|
236 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
237 */
|
|
|
238
|
|
223
|
239 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
240 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
241 #endif
|
|
|
242
|
|
|
243 #include "vim.h"
|
|
|
244
|
|
|
245 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
246
|
|
|
247 #ifdef HAVE_FCNTL_H
|
|
|
248 # include <fcntl.h>
|
|
|
249 #endif
|
|
|
250
|
|
323
|
251 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
252 Some places assume a word length fits in a
|
|
|
253 byte, thus it can't be above 255. */
|
|
226
|
254
|
|
324
|
255 /* Type used for indexes in the word tree need to be at least 3 bytes. If int
|
|
|
256 * is 8 bytes we could use something smaller, but what? */
|
|
|
257 #if SIZEOF_INT > 2
|
|
|
258 typedef int idx_T;
|
|
|
259 #else
|
|
|
260 typedef long idx_T;
|
|
|
261 #endif
|
|
|
262
|
|
|
263 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
264 * comes above it. */
|
|
300
|
265 #define WF_REGION 0x01 /* region byte follows */
|
|
|
266 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
267 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
268 #define WF_RARE 0x08 /* rare word */
|
|
307
|
269 #define WF_BANNED 0x10 /* bad word */
|
|
481
|
270 #define WF_AFX 0x20 /* affix ID follows */
|
|
389
|
271 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
272 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
273
|
|
390
|
274 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
275 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
|
276
|
|
389
|
277 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
278
|
|
455
|
279 /* flags for <pflags> */
|
|
|
280 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
281 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
282 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
283
|
|
481
|
284 /* Flags for postponed prefixes. Must be above affixID (one byte)
|
|
390
|
285 * and prefcondnr (two bytes). */
|
|
455
|
286 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
287 * postponed prefix */
|
|
|
288 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
289 * postponed prefix */
|
|
|
290 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
291 * postponed prefix */
|
|
390
|
292
|
|
|
293 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
294 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
295 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
296 * postponed prefix: no <pflags> */
|
|
|
297 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
298 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
299 * postponed prefix: <pflags> follows */
|
|
|
300 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
301 * follow; never used in prefix tree */
|
|
|
302 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
303
|
|
323
|
304 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
305 * and si_sal. Not for sl_sal!
|
|
323
|
306 * One replacement: from "ft_from" to "ft_to". */
|
|
|
307 typedef struct fromto_S
|
|
236
|
308 {
|
|
323
|
309 char_u *ft_from;
|
|
|
310 char_u *ft_to;
|
|
|
311 } fromto_T;
|
|
236
|
312
|
|
344
|
313 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
314 * The info is split for quick processing by spell_soundfold().
|
|
|
315 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
316 typedef struct salitem_S
|
|
|
317 {
|
|
|
318 char_u *sm_lead; /* leading letters */
|
|
|
319 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
320 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
321 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
322 char_u *sm_to; /* replacement. */
|
|
372
|
323 #ifdef FEAT_MBYTE
|
|
|
324 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
325 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
326 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
327 #endif
|
|
344
|
328 } salitem_T;
|
|
|
329
|
|
375
|
330 #ifdef FEAT_MBYTE
|
|
|
331 typedef int salfirst_T;
|
|
|
332 #else
|
|
|
333 typedef short salfirst_T;
|
|
|
334 #endif
|
|
|
335
|
|
236
|
336 /*
|
|
243
|
337 * Structure used to store words and other info for one language, loaded from
|
|
|
338 * a .spl file.
|
|
300
|
339 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
340 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
341 *
|
|
|
342 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
343 * the number of possible bytes, sorted on byte value:
|
|
|
344 * <len> <byte1> <byte2> ...
|
|
|
345 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
346 * byte in "byts".
|
|
|
347 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
481
|
348 * the flags, region mask and affixID for the word. There may be several
|
|
|
349 * zeros in sequence for alternative flag/region/affixID combinations.
|
|
236
|
350 */
|
|
|
351 typedef struct slang_S slang_T;
|
|
|
352 struct slang_S
|
|
|
353 {
|
|
|
354 slang_T *sl_next; /* next language */
|
|
|
355 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
356 char_u *sl_fname; /* name of .spl file */
|
|
323
|
357 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
358
|
|
300
|
359 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
360 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
361 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
362 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
363 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
364 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
365
|
|
236
|
366 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
367
|
|
376
|
368 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
369
|
|
481
|
370 int sl_compminlen; /* COMPOUNDMIN */
|
|
|
371 char_u *sl_compflags; /* COMPOUNDFLAGS (NULL when no compounding) */
|
|
|
372
|
|
339
|
373 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
374 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
375
|
|
323
|
376 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
377 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
378 there is none */
|
|
344
|
379 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
380 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
381 there is none */
|
|
375
|
382 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
383 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
384 * "sl_sal" is a list of wide char lists. */
|
|
323
|
385 int sl_followup; /* SAL followup */
|
|
|
386 int sl_collapse; /* SAL collapse_result */
|
|
|
387 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
388 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
389 #ifdef FEAT_MBYTE
|
|
|
390 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
391 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
392 #else
|
|
|
393 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
394 #endif
|
|
236
|
395 };
|
|
|
396
|
|
243
|
397 /* First language that is loaded, start of the linked list of loaded
|
|
|
398 * languages. */
|
|
236
|
399 static slang_T *first_lang = NULL;
|
|
|
400
|
|
323
|
401 /* Flags used in .spl file for soundsalike flags. */
|
|
|
402 #define SAL_F0LLOWUP 1
|
|
|
403 #define SAL_COLLAPSE 2
|
|
|
404 #define SAL_REM_ACCENTS 4
|
|
375
|
405 #define SAL_SOFO 8 /* SOFOFROM and SOFOTO instead of SAL */
|
|
323
|
406
|
|
236
|
407 /*
|
|
|
408 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
409 */
|
|
|
410 typedef struct langp_S
|
|
|
411 {
|
|
|
412 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
413 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
414 } langp_T;
|
|
|
415
|
|
|
416 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
417
|
|
307
|
418 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
419
|
|
|
420 /* Result values. Lower number is accepted over higher one. */
|
|
|
421 #define SP_BANNED -1
|
|
236
|
422 #define SP_OK 0
|
|
307
|
423 #define SP_RARE 1
|
|
|
424 #define SP_LOCAL 2
|
|
|
425 #define SP_BAD 3
|
|
236
|
426
|
|
481
|
427 #define VIMSPELLMAGIC "VIMspell10" /* string at start of Vim spell file */
|
|
236
|
428 #define VIMSPELLMAGICL 10
|
|
|
429
|
|
381
|
430 /* file used for "zG" and "zW" */
|
|
385
|
431 static char_u *int_wordlist = NULL;
|
|
381
|
432
|
|
236
|
433 /*
|
|
323
|
434 * Information used when looking for suggestions.
|
|
|
435 */
|
|
|
436 typedef struct suginfo_S
|
|
|
437 {
|
|
|
438 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
439 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
440 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
441 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
442 char_u *su_badptr; /* start of bad word in line */
|
|
|
443 int su_badlen; /* length of detected bad word in line */
|
|
346
|
444 int su_badflags; /* caps flags for bad word */
|
|
323
|
445 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
446 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
447 hashtab_T su_banned; /* table with banned words */
|
|
|
448 } suginfo_T;
|
|
|
449
|
|
|
450 /* One word suggestion. Used in "si_ga". */
|
|
|
451 typedef struct suggest_S
|
|
|
452 {
|
|
|
453 char_u *st_word; /* suggested word, allocated string */
|
|
|
454 int st_orglen; /* length of replaced text */
|
|
|
455 int st_score; /* lower is better */
|
|
344
|
456 int st_altscore; /* used when st_score compares equal */
|
|
|
457 int st_salscore; /* st_score is for soundalike */
|
|
324
|
458 int st_had_bonus; /* bonus already included in score */
|
|
323
|
459 } suggest_T;
|
|
|
460
|
|
344
|
461 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
462
|
|
324
|
463 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
464 * called the score may change, thus we need to keep more than what is
|
|
|
465 * displayed. */
|
|
351
|
466 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
467
|
|
|
468 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
469 * of suggestions that are not going to be displayed. */
|
|
344
|
470 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
471
|
|
|
472 /* score for various changes */
|
|
344
|
473 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
474 #define SCORE_ICASE 52 /* slightly different case */
|
|
|
475 #define SCORE_REGION 70 /* word is for different region */
|
|
|
476 #define SCORE_RARE 180 /* rare word */
|
|
|
477 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
478 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
479 #define SCORE_REP 87 /* REP replacement */
|
|
|
480 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
481 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
479
|
482 #define SCORE_SUBCOMP 33 /* substitute a composing character */
|
|
324
|
483 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
484 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
479
|
485 #define SCORE_DELCOMP 28 /* delete a composing character */
|
|
324
|
486 #define SCORE_INS 96 /* insert a character */
|
|
358
|
487 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
475
|
488 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
489 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
490
|
|
372
|
491 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
492 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
493 * 350 allows for about three changes. */
|
|
344
|
494
|
|
|
495 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
496 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
497
|
|
|
498 /*
|
|
236
|
499 * Structure to store info for word matching.
|
|
|
500 */
|
|
|
501 typedef struct matchinf_S
|
|
|
502 {
|
|
|
503 langp_T *mi_lp; /* info for language and region */
|
|
243
|
504
|
|
|
505 /* pointers to original text to be checked */
|
|
236
|
506 char_u *mi_word; /* start of word being checked */
|
|
339
|
507 char_u *mi_end; /* end of matching word so far */
|
|
243
|
508 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
509 char_u *mi_cend; /* char after what was used for
|
|
|
510 mi_capflags */
|
|
243
|
511
|
|
|
512 /* case-folded text */
|
|
|
513 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
514 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
515
|
|
339
|
516 /* for when checking word after a prefix */
|
|
|
517 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
481
|
518 affixID/condition */
|
|
339
|
519 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
520 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
521 #ifdef FEAT_MBYTE
|
|
|
522 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
523 case */
|
|
|
524 #else
|
|
|
525 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
526 #endif
|
|
339
|
527
|
|
481
|
528 /* for when checking a compound word */
|
|
|
529 int mi_compoff; /* start of following word offset */
|
|
|
530
|
|
243
|
531 /* others */
|
|
236
|
532 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
533 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
534 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
535 } matchinf_T;
|
|
|
536
|
|
307
|
537 /*
|
|
|
538 * The tables used for recognizing word characters according to spelling.
|
|
|
539 * These are only used for the first 256 characters of 'encoding'.
|
|
|
540 */
|
|
|
541 typedef struct spelltab_S
|
|
|
542 {
|
|
|
543 char_u st_isw[256]; /* flags: is word char */
|
|
|
544 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
545 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
546 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
547 } spelltab_T;
|
|
|
548
|
|
|
549 static spelltab_T spelltab;
|
|
|
550 static int did_set_spelltab;
|
|
|
551
|
|
324
|
552 #define CF_WORD 0x01
|
|
|
553 #define CF_UPPER 0x02
|
|
307
|
554
|
|
|
555 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
556 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
557 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
558 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
559 #ifdef FEAT_MBYTE
|
|
|
560 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
561 #endif
|
|
339
|
562 static void write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
563
|
|
|
564 /*
|
|
344
|
565 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
566 */
|
|
|
567 typedef enum
|
|
|
568 {
|
|
344
|
569 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
570 * ends); if badword ends there is a match, otherwise
|
|
|
571 * try splitting word. */
|
|
375
|
572 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
573 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
574 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
575 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
576 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
577 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
578 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
579 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
580 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
581 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
582 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
583 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
584 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
585 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
586 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
587 STATE_FINAL /* End of this node. */
|
|
|
588 } state_T;
|
|
|
589
|
|
|
590 /*
|
|
346
|
591 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
592 */
|
|
|
593 typedef struct trystate_S
|
|
|
594 {
|
|
330
|
595 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
596 int ts_score; /* score */
|
|
344
|
597 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
598 short ts_curi; /* index in list of child nodes */
|
|
|
599 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
600 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
601 char_u ts_twordlen; /* valid length of tword[] */
|
|
482
|
602 char_u ts_prefixdepth; /* stack depth for end of prefix or
|
|
|
603 * PFD_PREFIXTREE or PFD_NOPREFIX or
|
|
|
604 * PFD_COMPOUND */
|
|
330
|
605 #ifdef FEAT_MBYTE
|
|
|
606 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
607 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
608 char_u ts_isdiff; /* DIFF_ values */
|
|
|
609 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
610 #endif
|
|
323
|
611 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
330
|
612 char_u ts_save_splitoff; /* su_splitoff saved here */
|
|
346
|
613 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
614 } trystate_T;
|
|
|
615
|
|
330
|
616 /* values for ts_isdiff */
|
|
|
617 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
618 #define DIFF_YES 1 /* different byte found */
|
|
|
619 #define DIFF_INSERT 2 /* inserting character */
|
|
|
620
|
|
375
|
621 /* special values ts_prefixdepth */
|
|
482
|
622 #define PFD_COMPOUND 0xfd /* prefixed is a compound word */
|
|
|
623 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
624 #define PFD_NOPREFIX 0xff /* not using prefixes */
|
|
375
|
625
|
|
339
|
626 /* mode values for find_word */
|
|
481
|
627 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
628 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
629 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
630 #define FIND_COMPOUND 3 /* find case-folded compound word */
|
|
|
631 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */
|
|
339
|
632
|
|
236
|
633 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
634 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
635 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
636 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
482
|
637 static int can_compound __ARGS((slang_T *slang, int flags));
|
|
455
|
638 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
339
|
639 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
640 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
641 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
642 static int no_spell_checking __ARGS((void));
|
|
307
|
643 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
644 static char_u *spell_enc __ARGS((void));
|
|
385
|
645 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
646 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
647 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
648 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
381
|
649 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
650 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
651 #ifdef FEAT_MBYTE
|
|
|
652 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
653 #endif
|
|
339
|
654 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
655 static void clear_midword __ARGS((buf_T *buf));
|
|
|
656 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
657 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
658 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
659 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
660 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
389
|
661 static int set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
662 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
663 static void write_spell_chartab __ARGS((FILE *fd));
|
|
|
664 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
665 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
388
|
666 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
667 #ifdef FEAT_EVAL
|
|
|
668 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
669 #endif
|
|
|
670 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
671 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
672 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
673 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
674 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
675 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
676 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
677 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
678 #ifdef FEAT_MBYTE
|
|
|
679 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
680 #endif
|
|
323
|
681 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
682 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
683 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
684 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
685 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
686 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
687 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
688 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
689 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
690 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
691 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
692 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
693 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
694 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
695 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
696 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
697 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
698 #ifdef FEAT_MBYTE
|
|
375
|
699 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
700 #endif
|
|
344
|
701 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
702 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
703 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
704 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
705
|
|
324
|
706 /*
|
|
|
707 * Use our own character-case definitions, because the current locale may
|
|
|
708 * differ from what the .spl file uses.
|
|
|
709 * These must not be called with negative number!
|
|
|
710 */
|
|
|
711 #ifndef FEAT_MBYTE
|
|
|
712 /* Non-multi-byte implementation. */
|
|
|
713 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
714 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
715 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
716 #else
|
|
447
|
717 # if defined(HAVE_WCHAR_H)
|
|
|
718 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
719 # endif
|
|
324
|
720 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
721 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
722 * the "w" library function for characters above 255 if available. */
|
|
|
723 # ifdef HAVE_TOWLOWER
|
|
|
724 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
725 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
726 # else
|
|
|
727 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
728 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
729 # endif
|
|
|
730
|
|
|
731 # ifdef HAVE_TOWUPPER
|
|
|
732 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
733 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
734 # else
|
|
|
735 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
736 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
737 # endif
|
|
|
738
|
|
|
739 # ifdef HAVE_ISWUPPER
|
|
|
740 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
741 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
742 # else
|
|
|
743 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
744 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
745 # endif
|
|
|
746 #endif
|
|
|
747
|
|
307
|
748
|
|
|
749 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
750 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
482
|
751 static char *e_afftrailing = N_("Trailing text in %s line %d: %s");
|
|
469
|
752 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
753
|
|
|
754 /*
|
|
|
755 * Main spell-checking function.
|
|
300
|
756 * "ptr" points to a character that could be the start of a word.
|
|
236
|
757 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
758 * or when it's OK it remains unchanged.
|
|
|
759 * This must only be called when 'spelllang' is not empty.
|
|
323
|
760 *
|
|
385
|
761 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
762 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
763 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
764 * worry.
|
|
323
|
765 *
|
|
236
|
766 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
767 * caller can skip over the word.
|
|
|
768 */
|
|
|
769 int
|
|
385
|
770 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
771 win_T *wp; /* current window */
|
|
|
772 char_u *ptr;
|
|
|
773 int *attrp;
|
|
385
|
774 int *capcol; /* column to check for Capital */
|
|
236
|
775 {
|
|
|
776 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
777 be passed to functions quickly. */
|
|
344
|
778 int nrlen = 0; /* found a number first */
|
|
385
|
779 int c;
|
|
236
|
780
|
|
307
|
781 /* A word never starts at a space or a control character. Return quickly
|
|
|
782 * then, skipping over the character. */
|
|
|
783 if (*ptr <= ' ')
|
|
|
784 return 1;
|
|
236
|
785
|
|
344
|
786 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
787 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
788 * "3GPP". */
|
|
300
|
789 if (*ptr >= '0' && *ptr <= '9')
|
|
|
790 {
|
|
316
|
791 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
792 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
793 else
|
|
|
794 {
|
|
344
|
795 mi.mi_end = skipdigits(ptr);
|
|
|
796 nrlen = mi.mi_end - ptr;
|
|
300
|
797 }
|
|
376
|
798 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
799 return (int)(mi.mi_end - ptr);
|
|
346
|
800
|
|
|
801 /* Try including the digits in the word. */
|
|
|
802 mi.mi_fend = ptr + nrlen;
|
|
|
803 }
|
|
|
804 else
|
|
|
805 mi.mi_fend = ptr;
|
|
|
806
|
|
|
807 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
808 mi.mi_word = ptr;
|
|
376
|
809 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
810 {
|
|
|
811 do
|
|
|
812 {
|
|
307
|
813 mb_ptr_adv(mi.mi_fend);
|
|
376
|
814 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
815
|
|
|
816 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
817 {
|
|
|
818 /* Check word starting with capital letter. */
|
|
455
|
819 c = PTR2CHAR(ptr);
|
|
385
|
820 if (!SPELL_ISUPPER(c))
|
|
|
821 {
|
|
|
822 *attrp = highlight_attr[HLF_SPC];
|
|
|
823 return (int)(mi.mi_fend - ptr);
|
|
|
824 }
|
|
|
825 }
|
|
|
826 }
|
|
|
827 if (capcol != NULL)
|
|
|
828 *capcol = -1;
|
|
344
|
829
|
|
|
830 /* We always use the characters up to the next non-word character,
|
|
|
831 * also for bad words. */
|
|
|
832 mi.mi_end = mi.mi_fend;
|
|
|
833
|
|
|
834 /* Check caps type later. */
|
|
|
835 mi.mi_capflags = 0;
|
|
|
836 mi.mi_cend = NULL;
|
|
376
|
837 mi.mi_buf = wp->w_buffer;
|
|
344
|
838
|
|
|
839 /* Include one non-word character so that we can check for the
|
|
|
840 * word end. */
|
|
|
841 if (*mi.mi_fend != NUL)
|
|
|
842 mb_ptr_adv(mi.mi_fend);
|
|
|
843
|
|
|
844 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
845 MAXWLEN + 1);
|
|
|
846 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
847
|
|
|
848 /* The word is bad unless we recognize it. */
|
|
|
849 mi.mi_result = SP_BAD;
|
|
|
850
|
|
|
851 /*
|
|
|
852 * Loop over the languages specified in 'spelllang'.
|
|
|
853 * We check them all, because a matching word may be longer than an
|
|
|
854 * already found matching word.
|
|
|
855 */
|
|
|
856 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
857 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
858 {
|
|
|
859 /* Check for a matching word in case-folded words. */
|
|
|
860 find_word(&mi, FIND_FOLDWORD);
|
|
|
861
|
|
|
862 /* Check for a matching word in keep-case words. */
|
|
|
863 find_word(&mi, FIND_KEEPWORD);
|
|
|
864
|
|
|
865 /* Check for matching prefixes. */
|
|
|
866 find_prefix(&mi);
|
|
|
867 }
|
|
|
868
|
|
|
869 if (mi.mi_result != SP_OK)
|
|
|
870 {
|
|
346
|
871 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
872 * rare and local words, e.g., "3GPP". */
|
|
344
|
873 if (nrlen > 0)
|
|
346
|
874 {
|
|
|
875 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
876 return nrlen;
|
|
|
877 }
|
|
344
|
878
|
|
|
879 /* When we are at a non-word character there is no error, just
|
|
|
880 * skip over the character (try looking for a word after it). */
|
|
390
|
881 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
882 {
|
|
385
|
883 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
884 {
|
|
|
885 regmatch_T regmatch;
|
|
|
886
|
|
|
887 /* Check for end of sentence. */
|
|
|
888 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
889 regmatch.rm_ic = FALSE;
|
|
|
890 if (vim_regexec(®match, ptr, 0))
|
|
|
891 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
892 }
|
|
|
893
|
|
344
|
894 #ifdef FEAT_MBYTE
|
|
|
895 if (has_mbyte)
|
|
474
|
896 return (*mb_ptr2len)(ptr);
|
|
344
|
897 #endif
|
|
|
898 return 1;
|
|
300
|
899 }
|
|
243
|
900
|
|
344
|
901 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
902 *attrp = highlight_attr[HLF_SPB];
|
|
|
903 else if (mi.mi_result == SP_RARE)
|
|
|
904 *attrp = highlight_attr[HLF_SPR];
|
|
|
905 else
|
|
|
906 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
907 }
|
|
|
908
|
|
300
|
909 return (int)(mi.mi_end - ptr);
|
|
236
|
910 }
|
|
|
911
|
|
|
912 /*
|
|
300
|
913 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
914 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
915 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
916 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
917 * tree.
|
|
300
|
918 *
|
|
|
919 * For a match mip->mi_result is updated.
|
|
243
|
920 */
|
|
|
921 static void
|
|
339
|
922 find_word(mip, mode)
|
|
243
|
923 matchinf_T *mip;
|
|
339
|
924 int mode;
|
|
243
|
925 {
|
|
324
|
926 idx_T arridx = 0;
|
|
300
|
927 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
928 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
929 int endidxcnt = 0;
|
|
|
930 int len;
|
|
|
931 int wlen = 0;
|
|
|
932 int flen;
|
|
|
933 int c;
|
|
|
934 char_u *ptr;
|
|
324
|
935 idx_T lo, hi, m;
|
|
243
|
936 #ifdef FEAT_MBYTE
|
|
300
|
937 char_u *s;
|
|
339
|
938 char_u *p;
|
|
307
|
939 #endif
|
|
|
940 int res = SP_BAD;
|
|
300
|
941 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
942 unsigned flags;
|
|
|
943 char_u *byts;
|
|
324
|
944 idx_T *idxs;
|
|
481
|
945 int word_ends;
|
|
|
946
|
|
|
947 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND)
|
|
236
|
948 {
|
|
300
|
949 /* Check for word with matching case in keep-case tree. */
|
|
|
950 ptr = mip->mi_word;
|
|
|
951 flen = 9999; /* no case folding, always enough bytes */
|
|
|
952 byts = slang->sl_kbyts;
|
|
|
953 idxs = slang->sl_kidxs;
|
|
481
|
954
|
|
|
955 if (mode == FIND_KEEPCOMPOUND)
|
|
|
956 /* Skip over the previously found word(s). */
|
|
|
957 wlen += mip->mi_compoff;
|
|
236
|
958 }
|
|
|
959 else
|
|
|
960 {
|
|
300
|
961 /* Check for case-folded in case-folded tree. */
|
|
|
962 ptr = mip->mi_fword;
|
|
|
963 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
964 byts = slang->sl_fbyts;
|
|
|
965 idxs = slang->sl_fidxs;
|
|
339
|
966
|
|
|
967 if (mode == FIND_PREFIX)
|
|
|
968 {
|
|
|
969 /* Skip over the prefix. */
|
|
|
970 wlen = mip->mi_prefixlen;
|
|
|
971 flen -= mip->mi_prefixlen;
|
|
|
972 }
|
|
481
|
973 else if (mode == FIND_COMPOUND)
|
|
|
974 {
|
|
|
975 /* Skip over the previously found word(s). */
|
|
|
976 wlen = mip->mi_compoff;
|
|
|
977 flen -= mip->mi_compoff;
|
|
|
978 }
|
|
|
979
|
|
243
|
980 }
|
|
|
981
|
|
300
|
982 if (byts == NULL)
|
|
|
983 return; /* array is empty */
|
|
236
|
984
|
|
|
985 /*
|
|
307
|
986 * Repeat advancing in the tree until:
|
|
|
987 * - there is a byte that doesn't match,
|
|
|
988 * - we reach the end of the tree,
|
|
|
989 * - or we reach the end of the line.
|
|
236
|
990 */
|
|
300
|
991 for (;;)
|
|
236
|
992 {
|
|
346
|
993 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
994 flen = fold_more(mip);
|
|
300
|
995
|
|
|
996 len = byts[arridx++];
|
|
|
997
|
|
|
998 /* If the first possible byte is a zero the word could end here.
|
|
|
999 * Remember this index, we first check for the longest word. */
|
|
|
1000 if (byts[arridx] == 0)
|
|
|
1001 {
|
|
307
|
1002 if (endidxcnt == MAXWLEN)
|
|
|
1003 {
|
|
|
1004 /* Must be a corrupted spell file. */
|
|
|
1005 EMSG(_(e_format));
|
|
|
1006 return;
|
|
|
1007 }
|
|
300
|
1008 endlen[endidxcnt] = wlen;
|
|
|
1009 endidx[endidxcnt++] = arridx++;
|
|
|
1010 --len;
|
|
|
1011
|
|
|
1012 /* Skip over the zeros, there can be several flag/region
|
|
|
1013 * combinations. */
|
|
|
1014 while (len > 0 && byts[arridx] == 0)
|
|
|
1015 {
|
|
|
1016 ++arridx;
|
|
|
1017 --len;
|
|
|
1018 }
|
|
|
1019 if (len == 0)
|
|
|
1020 break; /* no children, word must end here */
|
|
|
1021 }
|
|
|
1022
|
|
|
1023 /* Stop looking at end of the line. */
|
|
|
1024 if (ptr[wlen] == NUL)
|
|
|
1025 break;
|
|
|
1026
|
|
|
1027 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1028 c = ptr[wlen];
|
|
346
|
1029 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1030 c = ' ';
|
|
300
|
1031 lo = arridx;
|
|
|
1032 hi = arridx + len - 1;
|
|
|
1033 while (lo < hi)
|
|
|
1034 {
|
|
|
1035 m = (lo + hi) / 2;
|
|
|
1036 if (byts[m] > c)
|
|
|
1037 hi = m - 1;
|
|
|
1038 else if (byts[m] < c)
|
|
|
1039 lo = m + 1;
|
|
|
1040 else
|
|
|
1041 {
|
|
|
1042 lo = hi = m;
|
|
|
1043 break;
|
|
236
|
1044 }
|
|
|
1045 }
|
|
300
|
1046
|
|
|
1047 /* Stop if there is no matching byte. */
|
|
|
1048 if (hi < lo || byts[lo] != c)
|
|
|
1049 break;
|
|
|
1050
|
|
|
1051 /* Continue at the child (if there is one). */
|
|
|
1052 arridx = idxs[lo];
|
|
|
1053 ++wlen;
|
|
|
1054 --flen;
|
|
346
|
1055
|
|
|
1056 /* One space in the good word may stand for several spaces in the
|
|
|
1057 * checked word. */
|
|
|
1058 if (c == ' ')
|
|
|
1059 {
|
|
|
1060 for (;;)
|
|
|
1061 {
|
|
|
1062 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1063 flen = fold_more(mip);
|
|
|
1064 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1065 break;
|
|
|
1066 ++wlen;
|
|
|
1067 --flen;
|
|
|
1068 }
|
|
|
1069 }
|
|
236
|
1070 }
|
|
|
1071
|
|
300
|
1072 /*
|
|
|
1073 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1074 * first.
|
|
|
1075 */
|
|
|
1076 while (endidxcnt > 0)
|
|
|
1077 {
|
|
|
1078 --endidxcnt;
|
|
|
1079 arridx = endidx[endidxcnt];
|
|
|
1080 wlen = endlen[endidxcnt];
|
|
236
|
1081
|
|
300
|
1082 #ifdef FEAT_MBYTE
|
|
|
1083 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1084 continue; /* not at first byte of character */
|
|
|
1085 #endif
|
|
376
|
1086 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
481
|
1087 {
|
|
|
1088 if (slang->sl_compflags == NULL)
|
|
|
1089 continue; /* next char is a word character */
|
|
|
1090 word_ends = FALSE;
|
|
|
1091 }
|
|
|
1092 else
|
|
|
1093 word_ends = TRUE;
|
|
300
|
1094
|
|
|
1095 #ifdef FEAT_MBYTE
|
|
339
|
1096 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1097 {
|
|
|
1098 /* Compute byte length in original word, length may change
|
|
339
|
1099 * when folding case. This can be slow, take a shortcut when the
|
|
|
1100 * case-folded word is equal to the keep-case word. */
|
|
300
|
1101 p = mip->mi_word;
|
|
339
|
1102 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1103 {
|
|
|
1104 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1105 mb_ptr_adv(p);
|
|
|
1106 wlen = p - mip->mi_word;
|
|
|
1107 }
|
|
300
|
1108 }
|
|
|
1109 #endif
|
|
236
|
1110
|
|
339
|
1111 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1112 * prefix ID.
|
|
|
1113 * Repeat this if there are more flags/region alternatives until there
|
|
|
1114 * is a match. */
|
|
|
1115 res = SP_BAD;
|
|
|
1116 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1117 --len, ++arridx)
|
|
300
|
1118 {
|
|
|
1119 flags = idxs[arridx];
|
|
324
|
1120
|
|
339
|
1121 /* For the fold-case tree check that the case of the checked word
|
|
|
1122 * matches with what the word in the tree requires.
|
|
|
1123 * For keep-case tree the case is always right. For prefixes we
|
|
|
1124 * don't bother to check. */
|
|
|
1125 if (mode == FIND_FOLDWORD)
|
|
300
|
1126 {
|
|
|
1127 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1128 {
|
|
323
|
1129 /* mi_capflags was set for a different word length, need
|
|
|
1130 * to do it again. */
|
|
300
|
1131 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1132 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1133 }
|
|
|
1134
|
|
346
|
1135 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1136 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1137 continue;
|
|
300
|
1138 }
|
|
236
|
1139
|
|
339
|
1140 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1141 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1142 * mip->mi_prefarridx that find_prefix() filled. */
|
|
481
|
1143 else if (mode == FIND_PREFIX)
|
|
|
1144 {
|
|
366
|
1145 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1146 flags,
|
|
455
|
1147 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1148 FALSE);
|
|
366
|
1149 if (c == 0)
|
|
339
|
1150 continue;
|
|
366
|
1151
|
|
|
1152 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1153 if (c & WF_RAREPFX)
|
|
|
1154 flags |= WF_RARE;
|
|
339
|
1155 }
|
|
|
1156
|
|
481
|
1157 if (mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND
|
|
|
1158 || !word_ends)
|
|
|
1159 {
|
|
|
1160 /* Makes you wonder why someone puts a compound flag on a word
|
|
|
1161 * that's too short... Myspell compatibility requires this
|
|
|
1162 * anyway. */
|
|
|
1163 if (wlen < slang->sl_compminlen)
|
|
|
1164 continue;
|
|
|
1165
|
|
|
1166 /* The word doesn't end or it comes after another: it must
|
|
|
1167 * have a compound flag. */
|
|
482
|
1168 if (!can_compound(slang, flags))
|
|
481
|
1169 continue;
|
|
|
1170 }
|
|
|
1171
|
|
|
1172 if (!word_ends)
|
|
|
1173 {
|
|
|
1174 /* Check that a valid word follows. If there is one, it will
|
|
|
1175 * set "mi_result", thus we are always finished here.
|
|
|
1176 * Recursive! */
|
|
|
1177
|
|
|
1178 /* Find following word in case-folded tree. */
|
|
|
1179 mip->mi_compoff = endlen[endidxcnt];
|
|
|
1180 #ifdef FEAT_MBYTE
|
|
|
1181 if (has_mbyte && mode == FIND_KEEPWORD)
|
|
|
1182 {
|
|
|
1183 /* Compute byte length in case-folded word from "wlen":
|
|
|
1184 * byte length in keep-case word. Length may change when
|
|
|
1185 * folding case. This can be slow, take a shortcut when
|
|
|
1186 * the case-folded word is equal to the keep-case word. */
|
|
|
1187 p = mip->mi_fword;
|
|
|
1188 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1189 {
|
|
|
1190 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1191 mb_ptr_adv(p);
|
|
|
1192 mip->mi_compoff = p - mip->mi_fword;
|
|
|
1193 }
|
|
|
1194 }
|
|
|
1195 #endif
|
|
|
1196 find_word(mip, FIND_COMPOUND);
|
|
|
1197 if (mip->mi_result == SP_OK)
|
|
|
1198 break;
|
|
|
1199
|
|
|
1200 /* Find following word in keep-case tree. */
|
|
|
1201 mip->mi_compoff = wlen;
|
|
|
1202 find_word(mip, FIND_KEEPCOMPOUND);
|
|
|
1203 if (mip->mi_result == SP_OK)
|
|
|
1204 break;
|
|
|
1205 continue;
|
|
|
1206 }
|
|
|
1207
|
|
339
|
1208 if (flags & WF_BANNED)
|
|
|
1209 res = SP_BANNED;
|
|
|
1210 else if (flags & WF_REGION)
|
|
|
1211 {
|
|
|
1212 /* Check region. */
|
|
390
|
1213 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1214 res = SP_OK;
|
|
339
|
1215 else
|
|
|
1216 res = SP_LOCAL;
|
|
300
|
1217 }
|
|
339
|
1218 else if (flags & WF_RARE)
|
|
|
1219 res = SP_RARE;
|
|
307
|
1220 else
|
|
339
|
1221 res = SP_OK;
|
|
|
1222
|
|
|
1223 /* Always use the longest match and the best result. */
|
|
|
1224 if (mip->mi_result > res)
|
|
|
1225 {
|
|
|
1226 mip->mi_result = res;
|
|
|
1227 mip->mi_end = mip->mi_word + wlen;
|
|
|
1228 }
|
|
351
|
1229 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1230 mip->mi_end = mip->mi_word + wlen;
|
|
|
1231
|
|
|
1232 if (res == SP_OK)
|
|
|
1233 break;
|
|
300
|
1234 }
|
|
|
1235
|
|
307
|
1236 if (res == SP_OK)
|
|
300
|
1237 break;
|
|
|
1238 }
|
|
236
|
1239 }
|
|
|
1240
|
|
323
|
1241 /*
|
|
482
|
1242 * Return TRUE if "flags" has a valid compound flag.
|
|
|
1243 * TODO: check flags in a more advanced way.
|
|
|
1244 */
|
|
|
1245 static int
|
|
|
1246 can_compound(slang, flags)
|
|
|
1247 slang_T *slang;
|
|
|
1248 int flags;
|
|
|
1249 {
|
|
|
1250 return slang->sl_compflags != NULL
|
|
|
1251 && *slang->sl_compflags == ((unsigned)flags >> 24);
|
|
|
1252 }
|
|
|
1253
|
|
|
1254 /*
|
|
390
|
1255 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1256 * ID in "flags" for the word "word".
|
|
366
|
1257 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1258 */
|
|
|
1259 static int
|
|
455
|
1260 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1261 int totprefcnt; /* nr of prefix IDs */
|
|
|
1262 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1263 int flags;
|
|
351
|
1264 char_u *word;
|
|
|
1265 slang_T *slang;
|
|
455
|
1266 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1267 {
|
|
|
1268 int prefcnt;
|
|
|
1269 int pidx;
|
|
|
1270 regprog_T *rp;
|
|
|
1271 regmatch_T regmatch;
|
|
390
|
1272 int prefid;
|
|
|
1273
|
|
|
1274 prefid = (unsigned)flags >> 24;
|
|
351
|
1275 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1276 {
|
|
|
1277 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1278
|
|
|
1279 /* Check the prefix ID. */
|
|
|
1280 if (prefid != (pidx & 0xff))
|
|
|
1281 continue;
|
|
|
1282
|
|
390
|
1283 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1284 * suffix. */
|
|
|
1285 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1286 continue;
|
|
|
1287
|
|
351
|
1288 /* Check the condition, if there is one. The condition index is
|
|
366
|
1289 * stored in the two bytes above the prefix ID byte. */
|
|
|
1290 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1291 if (rp != NULL)
|
|
|
1292 {
|
|
|
1293 regmatch.regprog = rp;
|
|
|
1294 regmatch.rm_ic = FALSE;
|
|
|
1295 if (!vim_regexec(®match, word, 0))
|
|
|
1296 continue;
|
|
|
1297 }
|
|
455
|
1298 else if (cond_req)
|
|
|
1299 continue;
|
|
|
1300
|
|
|
1301 /* It's a match! Return the WF_ flags. */
|
|
366
|
1302 return pidx;
|
|
|
1303 }
|
|
|
1304 return 0;
|
|
351
|
1305 }
|
|
|
1306
|
|
|
1307 /*
|
|
339
|
1308 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1309 * If it does, then check the following word.
|
|
|
1310 *
|
|
|
1311 * For a match mip->mi_result is updated.
|
|
|
1312 */
|
|
|
1313 static void
|
|
|
1314 find_prefix(mip)
|
|
|
1315 matchinf_T *mip;
|
|
|
1316 {
|
|
|
1317 idx_T arridx = 0;
|
|
|
1318 int len;
|
|
|
1319 int wlen = 0;
|
|
|
1320 int flen;
|
|
|
1321 int c;
|
|
|
1322 char_u *ptr;
|
|
|
1323 idx_T lo, hi, m;
|
|
|
1324 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1325 char_u *byts;
|
|
|
1326 idx_T *idxs;
|
|
|
1327
|
|
375
|
1328 byts = slang->sl_pbyts;
|
|
|
1329 if (byts == NULL)
|
|
|
1330 return; /* array is empty */
|
|
|
1331
|
|
339
|
1332 /* We use the case-folded word here, since prefixes are always
|
|
|
1333 * case-folded. */
|
|
|
1334 ptr = mip->mi_fword;
|
|
|
1335 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1336 idxs = slang->sl_pidxs;
|
|
|
1337
|
|
|
1338 /*
|
|
|
1339 * Repeat advancing in the tree until:
|
|
|
1340 * - there is a byte that doesn't match,
|
|
|
1341 * - we reach the end of the tree,
|
|
|
1342 * - or we reach the end of the line.
|
|
|
1343 */
|
|
|
1344 for (;;)
|
|
|
1345 {
|
|
|
1346 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1347 flen = fold_more(mip);
|
|
|
1348
|
|
|
1349 len = byts[arridx++];
|
|
|
1350
|
|
|
1351 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1352 * Check if the following word matches and supports the prefix. */
|
|
|
1353 if (byts[arridx] == 0)
|
|
|
1354 {
|
|
|
1355 /* There can be several prefixes with different conditions. We
|
|
|
1356 * try them all, since we don't know which one will give the
|
|
|
1357 * longest match. The word is the same each time, pass the list
|
|
|
1358 * of possible prefixes to find_word(). */
|
|
|
1359 mip->mi_prefarridx = arridx;
|
|
|
1360 mip->mi_prefcnt = len;
|
|
|
1361 while (len > 0 && byts[arridx] == 0)
|
|
|
1362 {
|
|
|
1363 ++arridx;
|
|
|
1364 --len;
|
|
|
1365 }
|
|
|
1366 mip->mi_prefcnt -= len;
|
|
|
1367
|
|
|
1368 /* Find the word that comes after the prefix. */
|
|
|
1369 mip->mi_prefixlen = wlen;
|
|
455
|
1370 #ifdef FEAT_MBYTE
|
|
|
1371 if (has_mbyte)
|
|
|
1372 {
|
|
|
1373 /* Case-folded length may differ from original length. */
|
|
|
1374 mip->mi_cprefixlen = nofold_len(mip->mi_fword, wlen,
|
|
|
1375 mip->mi_word);
|
|
|
1376 }
|
|
|
1377 else
|
|
|
1378 mip->mi_cprefixlen = wlen;
|
|
|
1379 #endif
|
|
339
|
1380 find_word(mip, FIND_PREFIX);
|
|
|
1381
|
|
|
1382
|
|
|
1383 if (len == 0)
|
|
|
1384 break; /* no children, word must end here */
|
|
|
1385 }
|
|
|
1386
|
|
|
1387 /* Stop looking at end of the line. */
|
|
|
1388 if (ptr[wlen] == NUL)
|
|
|
1389 break;
|
|
|
1390
|
|
|
1391 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1392 c = ptr[wlen];
|
|
|
1393 lo = arridx;
|
|
|
1394 hi = arridx + len - 1;
|
|
|
1395 while (lo < hi)
|
|
|
1396 {
|
|
|
1397 m = (lo + hi) / 2;
|
|
|
1398 if (byts[m] > c)
|
|
|
1399 hi = m - 1;
|
|
|
1400 else if (byts[m] < c)
|
|
|
1401 lo = m + 1;
|
|
|
1402 else
|
|
|
1403 {
|
|
|
1404 lo = hi = m;
|
|
|
1405 break;
|
|
|
1406 }
|
|
|
1407 }
|
|
|
1408
|
|
|
1409 /* Stop if there is no matching byte. */
|
|
|
1410 if (hi < lo || byts[lo] != c)
|
|
|
1411 break;
|
|
|
1412
|
|
|
1413 /* Continue at the child (if there is one). */
|
|
|
1414 arridx = idxs[lo];
|
|
|
1415 ++wlen;
|
|
|
1416 --flen;
|
|
|
1417 }
|
|
|
1418 }
|
|
|
1419
|
|
|
1420 /*
|
|
|
1421 * Need to fold at least one more character. Do until next non-word character
|
|
|
1422 * for efficiency.
|
|
|
1423 * Return the length of the folded chars in bytes.
|
|
|
1424 */
|
|
|
1425 static int
|
|
|
1426 fold_more(mip)
|
|
|
1427 matchinf_T *mip;
|
|
|
1428 {
|
|
|
1429 int flen;
|
|
|
1430 char_u *p;
|
|
|
1431
|
|
|
1432 p = mip->mi_fend;
|
|
|
1433 do
|
|
|
1434 {
|
|
|
1435 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1436 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1437
|
|
|
1438 /* Include the non-word character so that we can check for the
|
|
|
1439 * word end. */
|
|
|
1440 if (*mip->mi_fend != NUL)
|
|
|
1441 mb_ptr_adv(mip->mi_fend);
|
|
|
1442
|
|
|
1443 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1444 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1445 MAXWLEN - mip->mi_fwordlen);
|
|
|
1446 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1447 mip->mi_fwordlen += flen;
|
|
|
1448 return flen;
|
|
|
1449 }
|
|
|
1450
|
|
|
1451 /*
|
|
323
|
1452 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1453 * case.
|
|
|
1454 */
|
|
|
1455 static int
|
|
389
|
1456 spell_valid_case(wordflags, treeflags)
|
|
|
1457 int wordflags; /* flags for the checked word. */
|
|
323
|
1458 int treeflags; /* flags for the word in the spell tree */
|
|
|
1459 {
|
|
389
|
1460 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1461 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1462 && ((treeflags & WF_ONECAP) == 0
|
|
|
1463 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1464 }
|
|
|
1465
|
|
351
|
1466 /*
|
|
|
1467 * Return TRUE if spell checking is not enabled.
|
|
|
1468 */
|
|
|
1469 static int
|
|
|
1470 no_spell_checking()
|
|
|
1471 {
|
|
|
1472 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1473 {
|
|
|
1474 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1475 return TRUE;
|
|
|
1476 }
|
|
|
1477 return FALSE;
|
|
|
1478 }
|
|
300
|
1479
|
|
236
|
1480 /*
|
|
|
1481 * Move to next spell error.
|
|
323
|
1482 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
236
|
1483 * Return OK if found, FAIL otherwise.
|
|
|
1484 */
|
|
|
1485 int
|
|
323
|
1486 spell_move_to(dir, allwords, curline)
|
|
236
|
1487 int dir; /* FORWARD or BACKWARD */
|
|
|
1488 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1489 int curline;
|
|
236
|
1490 {
|
|
249
|
1491 linenr_T lnum;
|
|
|
1492 pos_T found_pos;
|
|
236
|
1493 char_u *line;
|
|
|
1494 char_u *p;
|
|
346
|
1495 char_u *endp;
|
|
|
1496 int attr;
|
|
236
|
1497 int len;
|
|
249
|
1498 int has_syntax = syntax_present(curbuf);
|
|
|
1499 int col;
|
|
|
1500 int can_spell;
|
|
346
|
1501 char_u *buf = NULL;
|
|
|
1502 int buflen = 0;
|
|
|
1503 int skip = 0;
|
|
385
|
1504 int capcol = -1;
|
|
236
|
1505
|
|
351
|
1506 if (no_spell_checking())
|
|
236
|
1507 return FAIL;
|
|
|
1508
|
|
249
|
1509 /*
|
|
|
1510 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1511 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1512 *
|
|
|
1513 * When searching backwards, we continue in the line to find the last
|
|
|
1514 * bad word (in the cursor line: before the cursor).
|
|
346
|
1515 *
|
|
|
1516 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1517 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1518 * though...
|
|
249
|
1519 */
|
|
|
1520 lnum = curwin->w_cursor.lnum;
|
|
|
1521 found_pos.lnum = 0;
|
|
236
|
1522
|
|
|
1523 while (!got_int)
|
|
|
1524 {
|
|
249
|
1525 line = ml_get(lnum);
|
|
346
|
1526
|
|
|
1527 len = STRLEN(line);
|
|
|
1528 if (buflen < len + MAXWLEN + 2)
|
|
|
1529 {
|
|
|
1530 vim_free(buf);
|
|
|
1531 buflen = len + MAXWLEN + 2;
|
|
|
1532 buf = alloc(buflen);
|
|
|
1533 if (buf == NULL)
|
|
|
1534 break;
|
|
|
1535 }
|
|
|
1536
|
|
385
|
1537 /* In first line check first word for Capital. */
|
|
|
1538 if (lnum == 1)
|
|
|
1539 capcol = 0;
|
|
|
1540
|
|
|
1541 /* For checking first word with a capital skip white space. */
|
|
|
1542 if (capcol == 0)
|
|
|
1543 capcol = skipwhite(line) - line;
|
|
|
1544
|
|
346
|
1545 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1546 * possible. */
|
|
|
1547 STRCPY(buf, line);
|
|
|
1548 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1549 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1550
|
|
|
1551 p = buf + skip;
|
|
|
1552 endp = buf + len;
|
|
|
1553 while (p < endp)
|
|
236
|
1554 {
|
|
300
|
1555 /* When searching backward don't search after the cursor. */
|
|
|
1556 if (dir == BACKWARD
|
|
|
1557 && lnum == curwin->w_cursor.lnum
|
|
346
|
1558 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1559 break;
|
|
249
|
1560
|
|
300
|
1561 /* start of word */
|
|
346
|
1562 attr = 0;
|
|
385
|
1563 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1564
|
|
300
|
1565 if (attr != 0)
|
|
|
1566 {
|
|
|
1567 /* We found a bad word. Check the attribute. */
|
|
|
1568 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1569 {
|
|
300
|
1570 /* When searching forward only accept a bad word after
|
|
|
1571 * the cursor. */
|
|
|
1572 if (dir == BACKWARD
|
|
|
1573 || lnum > curwin->w_cursor.lnum
|
|
|
1574 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1575 && (colnr_T)(curline ? p - buf + len
|
|
|
1576 : p - buf)
|
|
300
|
1577 > curwin->w_cursor.col))
|
|
236
|
1578 {
|
|
300
|
1579 if (has_syntax)
|
|
249
|
1580 {
|
|
346
|
1581 col = p - buf;
|
|
300
|
1582 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1583 FALSE, &can_spell);
|
|
|
1584 }
|
|
|
1585 else
|
|
|
1586 can_spell = TRUE;
|
|
249
|
1587
|
|
300
|
1588 if (can_spell)
|
|
|
1589 {
|
|
|
1590 found_pos.lnum = lnum;
|
|
346
|
1591 found_pos.col = p - buf;
|
|
249
|
1592 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1593 found_pos.coladd = 0;
|
|
249
|
1594 #endif
|
|
300
|
1595 if (dir == FORWARD)
|
|
|
1596 {
|
|
|
1597 /* No need to search further. */
|
|
|
1598 curwin->w_cursor = found_pos;
|
|
346
|
1599 vim_free(buf);
|
|
300
|
1600 return OK;
|
|
249
|
1601 }
|
|
|
1602 }
|
|
236
|
1603 }
|
|
|
1604 }
|
|
|
1605 }
|
|
|
1606
|
|
300
|
1607 /* advance to character after the word */
|
|
|
1608 p += len;
|
|
385
|
1609 capcol -= len;
|
|
236
|
1610 }
|
|
|
1611
|
|
323
|
1612 if (curline)
|
|
346
|
1613 break; /* only check cursor line */
|
|
323
|
1614
|
|
236
|
1615 /* Advance to next line. */
|
|
249
|
1616 if (dir == BACKWARD)
|
|
|
1617 {
|
|
|
1618 if (found_pos.lnum != 0)
|
|
|
1619 {
|
|
|
1620 /* Use the last match in the line. */
|
|
|
1621 curwin->w_cursor = found_pos;
|
|
346
|
1622 vim_free(buf);
|
|
249
|
1623 return OK;
|
|
|
1624 }
|
|
|
1625 if (lnum == 1)
|
|
346
|
1626 break;
|
|
249
|
1627 --lnum;
|
|
385
|
1628 capcol = -1;
|
|
249
|
1629 }
|
|
|
1630 else
|
|
|
1631 {
|
|
|
1632 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1633 break;
|
|
249
|
1634 ++lnum;
|
|
346
|
1635
|
|
|
1636 /* Skip the characters at the start of the next line that were
|
|
|
1637 * included in a match crossing line boundaries. */
|
|
|
1638 if (attr == 0)
|
|
|
1639 skip = p - endp;
|
|
|
1640 else
|
|
|
1641 skip = 0;
|
|
385
|
1642
|
|
|
1643 /* Capscol skips over the inserted space. */
|
|
|
1644 --capcol;
|
|
|
1645
|
|
|
1646 /* But after empty line check first word in next line */
|
|
|
1647 if (*skipwhite(line) == NUL)
|
|
|
1648 capcol = 0;
|
|
249
|
1649 }
|
|
236
|
1650
|
|
|
1651 line_breakcheck();
|
|
|
1652 }
|
|
|
1653
|
|
346
|
1654 vim_free(buf);
|
|
|
1655 return FAIL;
|
|
|
1656 }
|
|
|
1657
|
|
|
1658 /*
|
|
|
1659 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1660 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1661 */
|
|
|
1662 void
|
|
|
1663 spell_cat_line(buf, line, maxlen)
|
|
|
1664 char_u *buf;
|
|
|
1665 char_u *line;
|
|
|
1666 int maxlen;
|
|
|
1667 {
|
|
|
1668 char_u *p;
|
|
|
1669 int n;
|
|
|
1670
|
|
|
1671 p = skipwhite(line);
|
|
|
1672 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1673 p = skipwhite(p + 1);
|
|
|
1674
|
|
|
1675 if (*p != NUL)
|
|
|
1676 {
|
|
|
1677 *buf = ' ';
|
|
376
|
1678 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1679 n = p - line;
|
|
|
1680 if (n >= maxlen)
|
|
|
1681 n = maxlen - 1;
|
|
|
1682 vim_memset(buf + 1, ' ', n);
|
|
|
1683 }
|
|
236
|
1684 }
|
|
|
1685
|
|
|
1686 /*
|
|
307
|
1687 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1688 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1689 */
|
|
307
|
1690 static void
|
|
236
|
1691 spell_load_lang(lang)
|
|
|
1692 char_u *lang;
|
|
|
1693 {
|
|
310
|
1694 char_u fname_enc[85];
|
|
236
|
1695 int r;
|
|
307
|
1696 char_u langcp[MAXWLEN + 1];
|
|
|
1697
|
|
310
|
1698 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1699 * It's truncated when an error is detected. */
|
|
|
1700 STRCPY(langcp, lang);
|
|
|
1701
|
|
310
|
1702 /*
|
|
|
1703 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1704 */
|
|
|
1705 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1706 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1707 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1708
|
|
|
1709 if (r == FAIL && *langcp != NUL)
|
|
|
1710 {
|
|
|
1711 /* Try loading the ASCII version. */
|
|
310
|
1712 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1713 "spell/%s.ascii.spl", lang);
|
|
310
|
1714 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1715 }
|
|
|
1716
|
|
|
1717 if (r == FAIL)
|
|
|
1718 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
|
236
|
1719 fname_enc + 6);
|
|
310
|
1720 else if (*langcp != NUL)
|
|
|
1721 {
|
|
|
1722 /* Load all the additions. */
|
|
|
1723 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1724 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1725 }
|
|
|
1726 }
|
|
|
1727
|
|
|
1728 /*
|
|
|
1729 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1730 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1731 */
|
|
|
1732 static char_u *
|
|
|
1733 spell_enc()
|
|
|
1734 {
|
|
|
1735
|
|
|
1736 #ifdef FEAT_MBYTE
|
|
|
1737 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1738 return p_enc;
|
|
|
1739 #endif
|
|
|
1740 return (char_u *)"latin1";
|
|
236
|
1741 }
|
|
|
1742
|
|
|
1743 /*
|
|
385
|
1744 * Get the name of the .spl file for the internal wordlist into
|
|
|
1745 * "fname[MAXPATHL]".
|
|
|
1746 */
|
|
|
1747 static void
|
|
|
1748 int_wordlist_spl(fname)
|
|
|
1749 char_u *fname;
|
|
|
1750 {
|
|
|
1751 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
1752 int_wordlist, spell_enc());
|
|
|
1753 }
|
|
|
1754
|
|
|
1755 /*
|
|
236
|
1756 * Allocate a new slang_T.
|
|
|
1757 * Caller must fill "sl_next".
|
|
|
1758 */
|
|
|
1759 static slang_T *
|
|
|
1760 slang_alloc(lang)
|
|
|
1761 char_u *lang;
|
|
|
1762 {
|
|
|
1763 slang_T *lp;
|
|
|
1764
|
|
300
|
1765 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1766 if (lp != NULL)
|
|
|
1767 {
|
|
|
1768 lp->sl_name = vim_strsave(lang);
|
|
323
|
1769 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
236
|
1770 }
|
|
|
1771 return lp;
|
|
|
1772 }
|
|
|
1773
|
|
|
1774 /*
|
|
|
1775 * Free the contents of an slang_T and the structure itself.
|
|
|
1776 */
|
|
|
1777 static void
|
|
|
1778 slang_free(lp)
|
|
|
1779 slang_T *lp;
|
|
|
1780 {
|
|
|
1781 vim_free(lp->sl_name);
|
|
310
|
1782 vim_free(lp->sl_fname);
|
|
|
1783 slang_clear(lp);
|
|
|
1784 vim_free(lp);
|
|
|
1785 }
|
|
|
1786
|
|
|
1787 /*
|
|
|
1788 * Clear an slang_T so that the file can be reloaded.
|
|
|
1789 */
|
|
|
1790 static void
|
|
|
1791 slang_clear(lp)
|
|
|
1792 slang_T *lp;
|
|
|
1793 {
|
|
339
|
1794 garray_T *gap;
|
|
|
1795 fromto_T *ftp;
|
|
344
|
1796 salitem_T *smp;
|
|
339
|
1797 int i;
|
|
323
|
1798
|
|
300
|
1799 vim_free(lp->sl_fbyts);
|
|
310
|
1800 lp->sl_fbyts = NULL;
|
|
300
|
1801 vim_free(lp->sl_kbyts);
|
|
310
|
1802 lp->sl_kbyts = NULL;
|
|
339
|
1803 vim_free(lp->sl_pbyts);
|
|
|
1804 lp->sl_pbyts = NULL;
|
|
|
1805
|
|
300
|
1806 vim_free(lp->sl_fidxs);
|
|
310
|
1807 lp->sl_fidxs = NULL;
|
|
300
|
1808 vim_free(lp->sl_kidxs);
|
|
310
|
1809 lp->sl_kidxs = NULL;
|
|
339
|
1810 vim_free(lp->sl_pidxs);
|
|
|
1811 lp->sl_pidxs = NULL;
|
|
323
|
1812
|
|
344
|
1813 gap = &lp->sl_rep;
|
|
|
1814 while (gap->ga_len > 0)
|
|
323
|
1815 {
|
|
344
|
1816 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1817 vim_free(ftp->ft_from);
|
|
|
1818 vim_free(ftp->ft_to);
|
|
323
|
1819 }
|
|
344
|
1820 ga_clear(gap);
|
|
|
1821
|
|
|
1822 gap = &lp->sl_sal;
|
|
375
|
1823 if (lp->sl_sofo)
|
|
376
|
1824 {
|
|
|
1825 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1826 if (gap->ga_data != NULL)
|
|
|
1827 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1828 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1829 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1830 }
|
|
375
|
1831 else
|
|
|
1832 /* SAL items: free salitem_T items */
|
|
|
1833 while (gap->ga_len > 0)
|
|
|
1834 {
|
|
|
1835 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1836 vim_free(smp->sm_lead);
|
|
|
1837 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1838 vim_free(smp->sm_to);
|
|
|
1839 #ifdef FEAT_MBYTE
|
|
|
1840 vim_free(smp->sm_lead_w);
|
|
|
1841 vim_free(smp->sm_oneof_w);
|
|
|
1842 vim_free(smp->sm_to_w);
|
|
|
1843 #endif
|
|
|
1844 }
|
|
344
|
1845 ga_clear(gap);
|
|
323
|
1846
|
|
339
|
1847 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1848 vim_free(lp->sl_prefprog[i]);
|
|
376
|
1849 lp->sl_prefixcnt = 0;
|
|
339
|
1850 vim_free(lp->sl_prefprog);
|
|
376
|
1851 lp->sl_prefprog = NULL;
|
|
|
1852
|
|
|
1853 vim_free(lp->sl_midword);
|
|
|
1854 lp->sl_midword = NULL;
|
|
339
|
1855
|
|
481
|
1856 vim_free(lp->sl_compflags);
|
|
|
1857 lp->sl_compflags = NULL;
|
|
|
1858
|
|
330
|
1859 #ifdef FEAT_MBYTE
|
|
|
1860 {
|
|
|
1861 int todo = lp->sl_map_hash.ht_used;
|
|
|
1862 hashitem_T *hi;
|
|
|
1863
|
|
|
1864 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1865 if (!HASHITEM_EMPTY(hi))
|
|
|
1866 {
|
|
|
1867 --todo;
|
|
|
1868 vim_free(hi->hi_key);
|
|
|
1869 }
|
|
|
1870 }
|
|
|
1871 hash_clear(&lp->sl_map_hash);
|
|
|
1872 #endif
|
|
236
|
1873 }
|
|
|
1874
|
|
|
1875 /*
|
|
307
|
1876 * Load one spell file and store the info into a slang_T.
|
|
236
|
1877 * Invoked through do_in_runtimepath().
|
|
|
1878 */
|
|
|
1879 static void
|
|
310
|
1880 spell_load_cb(fname, cookie)
|
|
236
|
1881 char_u *fname;
|
|
307
|
1882 void *cookie; /* points to the language name */
|
|
236
|
1883 {
|
|
323
|
1884 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1885 }
|
|
|
1886
|
|
|
1887 /*
|
|
|
1888 * Load one spell file and store the info into a slang_T.
|
|
|
1889 *
|
|
|
1890 * This is invoked in two ways:
|
|
|
1891 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
1892 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
1893 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
1894 * points to the existing slang_T.
|
|
323
|
1895 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
1896 */
|
|
323
|
1897 static slang_T *
|
|
|
1898 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
1899 char_u *fname;
|
|
|
1900 char_u *lang;
|
|
|
1901 slang_T *old_lp;
|
|
323
|
1902 int silent; /* no error if file doesn't exist */
|
|
310
|
1903 {
|
|
236
|
1904 FILE *fd;
|
|
|
1905 char_u buf[MAXWLEN + 1];
|
|
|
1906 char_u *p;
|
|
339
|
1907 char_u *bp;
|
|
|
1908 idx_T *ip;
|
|
236
|
1909 int i;
|
|
339
|
1910 int n;
|
|
300
|
1911 int len;
|
|
236
|
1912 int round;
|
|
|
1913 char_u *save_sourcing_name = sourcing_name;
|
|
|
1914 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
255
|
1915 int cnt, ccnt;
|
|
|
1916 char_u *fol;
|
|
307
|
1917 slang_T *lp = NULL;
|
|
323
|
1918 garray_T *gap;
|
|
|
1919 fromto_T *ftp;
|
|
344
|
1920 salitem_T *smp;
|
|
323
|
1921 short *first;
|
|
324
|
1922 idx_T idx;
|
|
344
|
1923 int c = 0;
|
|
236
|
1924
|
|
310
|
1925 fd = mch_fopen((char *)fname, "r");
|
|
236
|
1926 if (fd == NULL)
|
|
|
1927 {
|
|
323
|
1928 if (!silent)
|
|
|
1929 EMSG2(_(e_notopen), fname);
|
|
|
1930 else if (p_verbose > 2)
|
|
|
1931 {
|
|
|
1932 verbose_enter();
|
|
|
1933 smsg((char_u *)e_notopen, fname);
|
|
|
1934 verbose_leave();
|
|
|
1935 }
|
|
255
|
1936 goto endFAIL;
|
|
236
|
1937 }
|
|
310
|
1938 if (p_verbose > 2)
|
|
|
1939 {
|
|
|
1940 verbose_enter();
|
|
|
1941 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
1942 verbose_leave();
|
|
|
1943 }
|
|
|
1944
|
|
|
1945 if (old_lp == NULL)
|
|
|
1946 {
|
|
|
1947 lp = slang_alloc(lang);
|
|
|
1948 if (lp == NULL)
|
|
|
1949 goto endFAIL;
|
|
|
1950
|
|
|
1951 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
1952 lp->sl_fname = vim_strsave(fname);
|
|
|
1953 if (lp->sl_fname == NULL)
|
|
|
1954 goto endFAIL;
|
|
|
1955
|
|
|
1956 /* Check for .add.spl. */
|
|
|
1957 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
1958 }
|
|
|
1959 else
|
|
|
1960 lp = old_lp;
|
|
307
|
1961
|
|
236
|
1962 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
1963 sourcing_name = fname;
|
|
|
1964 sourcing_lnum = 0;
|
|
|
1965
|
|
339
|
1966 /* <HEADER>: <fileID>
|
|
|
1967 * <regioncnt> <regionname> ...
|
|
|
1968 * <charflagslen> <charflags>
|
|
|
1969 * <fcharslen> <fchars>
|
|
366
|
1970 * <midwordlen> <midword>
|
|
481
|
1971 * <compoundlen> <compoundtype> <compoundinfo>
|
|
339
|
1972 * <prefcondcnt> <prefcond> ...
|
|
|
1973 */
|
|
236
|
1974 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
1975 buf[i] = getc(fd); /* <fileID> */
|
|
|
1976 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
1977 {
|
|
|
1978 EMSG(_("E757: Wrong file ID in spell file"));
|
|
255
|
1979 goto endFAIL;
|
|
236
|
1980 }
|
|
|
1981
|
|
|
1982 cnt = getc(fd); /* <regioncnt> */
|
|
255
|
1983 if (cnt < 0)
|
|
236
|
1984 {
|
|
|
1985 truncerr:
|
|
381
|
1986 EMSG(_(e_spell_trunc));
|
|
255
|
1987 goto endFAIL;
|
|
236
|
1988 }
|
|
|
1989 if (cnt > 8)
|
|
|
1990 {
|
|
|
1991 formerr:
|
|
307
|
1992 EMSG(_(e_format));
|
|
255
|
1993 goto endFAIL;
|
|
236
|
1994 }
|
|
|
1995 for (i = 0; i < cnt; ++i)
|
|
|
1996 {
|
|
|
1997 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
|
1998 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
|
1999 }
|
|
|
2000 lp->sl_regions[cnt * 2] = NUL;
|
|
|
2001
|
|
381
|
2002 /* <charflagslen> <charflags> */
|
|
|
2003 p = read_cnt_string(fd, 1, &cnt);
|
|
389
|
2004 if (cnt < 0)
|
|
381
|
2005 goto endFAIL;
|
|
|
2006
|
|
|
2007 /* <fcharslen> <fchars> */
|
|
389
|
2008 fol = read_cnt_string(fd, 2, &ccnt);
|
|
|
2009 if (ccnt < 0)
|
|
381
|
2010 {
|
|
300
|
2011 vim_free(p);
|
|
381
|
2012 goto endFAIL;
|
|
|
2013 }
|
|
|
2014
|
|
|
2015 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
2016 if (p != NULL && fol != NULL)
|
|
389
|
2017 i = set_spell_charflags(p, cnt, fol);
|
|
381
|
2018
|
|
|
2019 vim_free(p);
|
|
|
2020 vim_free(fol);
|
|
|
2021
|
|
|
2022 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2023 if ((p == NULL) != (fol == NULL))
|
|
|
2024 goto formerr;
|
|
255
|
2025
|
|
366
|
2026 /* <midwordlen> <midword> */
|
|
376
|
2027 lp->sl_midword = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2028 if (cnt < 0)
|
|
376
|
2029 goto endFAIL;
|
|
366
|
2030
|
|
481
|
2031 /* <compoundlen> <compoundtype> <compoundinfo> */
|
|
|
2032 cnt = (getc(fd) << 8) + getc(fd); /* <compoundlen> */
|
|
|
2033 if (cnt < 0)
|
|
|
2034 goto endFAIL;
|
|
|
2035 if (cnt > 0)
|
|
|
2036 {
|
|
|
2037 --cnt;
|
|
|
2038 c = getc(fd); /* <compoundtype> */
|
|
|
2039 if (c != 1)
|
|
|
2040 {
|
|
|
2041 /* Unknown kind of compound words, skip the info. */
|
|
|
2042 while (cnt-- > 0)
|
|
|
2043 getc(fd);
|
|
|
2044 }
|
|
|
2045 else if (cnt < 2)
|
|
|
2046 goto formerr;
|
|
|
2047 else
|
|
|
2048 {
|
|
|
2049 --cnt;
|
|
|
2050 c = getc(fd); /* <comp1minlen> */
|
|
|
2051 if (c < 1 || c > 50)
|
|
|
2052 c = 3;
|
|
|
2053 lp->sl_compminlen = c;
|
|
|
2054
|
|
|
2055 p = alloc(cnt + 1);
|
|
|
2056 if (p == NULL)
|
|
|
2057 goto endFAIL;
|
|
|
2058 lp->sl_compflags = p;
|
|
|
2059 while (cnt-- > 0)
|
|
|
2060 *p++ = getc(fd); /* <comp1flags> */
|
|
|
2061 *p = NUL;
|
|
|
2062 }
|
|
|
2063 }
|
|
|
2064
|
|
|
2065
|
|
339
|
2066 /* <prefcondcnt> <prefcond> ... */
|
|
|
2067 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
2068 if (cnt > 0)
|
|
|
2069 {
|
|
|
2070 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2071 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2072 if (lp->sl_prefprog == NULL)
|
|
|
2073 goto endFAIL;
|
|
|
2074 lp->sl_prefixcnt = cnt;
|
|
|
2075
|
|
|
2076 for (i = 0; i < cnt; ++i)
|
|
|
2077 {
|
|
|
2078 /* <prefcond> : <condlen> <condstr> */
|
|
|
2079 n = getc(fd); /* <condlen> */
|
|
481
|
2080 if (n < 0 || n >= MAXWLEN)
|
|
339
|
2081 goto formerr;
|
|
|
2082 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2083 * compile the regexp program used to check for the condition. */
|
|
|
2084 if (n > 0)
|
|
|
2085 {
|
|
344
|
2086 buf[0] = '^'; /* always match at one position only */
|
|
|
2087 p = buf + 1;
|
|
339
|
2088 while (n-- > 0)
|
|
|
2089 *p++ = getc(fd); /* <condstr> */
|
|
|
2090 *p = NUL;
|
|
|
2091 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
2092 }
|
|
|
2093 }
|
|
|
2094 }
|
|
|
2095
|
|
|
2096
|
|
323
|
2097 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
2098 * <salflags> <salcount> <sal> ...
|
|
|
2099 * <maplen> <mapstr> */
|
|
339
|
2100
|
|
344
|
2101 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
2102 if (cnt < 0)
|
|
|
2103 goto formerr;
|
|
|
2104
|
|
|
2105 gap = &lp->sl_rep;
|
|
|
2106 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2107 goto endFAIL;
|
|
|
2108
|
|
|
2109 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
2110 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2111 {
|
|
|
2112 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
381
|
2113 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
389
|
2114 if (i <= 0)
|
|
381
|
2115 goto endFAIL;
|
|
|
2116 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
389
|
2117 if (i <= 0)
|
|
381
|
2118 {
|
|
|
2119 vim_free(ftp->ft_from);
|
|
|
2120 goto endFAIL;
|
|
344
|
2121 }
|
|
|
2122 }
|
|
|
2123
|
|
|
2124 /* Fill the first-index table. */
|
|
|
2125 first = lp->sl_rep_first;
|
|
|
2126 for (i = 0; i < 256; ++i)
|
|
|
2127 first[i] = -1;
|
|
|
2128 for (i = 0; i < gap->ga_len; ++i)
|
|
323
|
2129 {
|
|
344
|
2130 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
2131 if (first[*ftp->ft_from] == -1)
|
|
|
2132 first[*ftp->ft_from] = i;
|
|
|
2133 }
|
|
|
2134
|
|
|
2135 i = getc(fd); /* <salflags> */
|
|
|
2136 if (i & SAL_F0LLOWUP)
|
|
|
2137 lp->sl_followup = TRUE;
|
|
|
2138 if (i & SAL_COLLAPSE)
|
|
|
2139 lp->sl_collapse = TRUE;
|
|
|
2140 if (i & SAL_REM_ACCENTS)
|
|
|
2141 lp->sl_rem_accents = TRUE;
|
|
375
|
2142 if (i & SAL_SOFO)
|
|
|
2143 lp->sl_sofo = TRUE;
|
|
389
|
2144 else
|
|
|
2145 lp->sl_sofo = FALSE;
|
|
344
|
2146
|
|
|
2147 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2148 if (cnt < 0)
|
|
|
2149 goto formerr;
|
|
|
2150
|
|
375
|
2151 if (lp->sl_sofo)
|
|
|
2152 {
|
|
|
2153 /*
|
|
|
2154 * SOFOFROM and SOFOTO items come in one <salfrom> and <salto>
|
|
|
2155 */
|
|
|
2156 if (cnt != 1)
|
|
|
2157 goto formerr;
|
|
|
2158
|
|
381
|
2159 /* <salfromlen> <salfrom> */
|
|
|
2160 bp = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2161 if (cnt < 0)
|
|
375
|
2162 goto endFAIL;
|
|
381
|
2163
|
|
|
2164 /* <saltolen> <salto> */
|
|
|
2165 fol = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2166 if (cnt < 0)
|
|
375
|
2167 {
|
|
|
2168 vim_free(bp);
|
|
344
|
2169 goto endFAIL;
|
|
375
|
2170 }
|
|
381
|
2171
|
|
|
2172 /* Store the info in lp->sl_sal and/or lp->sl_sal_first. */
|
|
389
|
2173 if (bp != NULL && fol != NULL)
|
|
|
2174 i = set_sofo(lp, bp, fol);
|
|
|
2175 else if (bp != NULL || fol != NULL)
|
|
|
2176 i = FAIL; /* only one of two strings is an error */
|
|
|
2177 else
|
|
|
2178 i = OK;
|
|
381
|
2179
|
|
375
|
2180 vim_free(bp);
|
|
|
2181 vim_free(fol);
|
|
381
|
2182 if (i == FAIL)
|
|
|
2183 goto formerr;
|
|
375
|
2184 }
|
|
|
2185 else
|
|
|
2186 {
|
|
|
2187 /*
|
|
|
2188 * SAL items
|
|
|
2189 */
|
|
|
2190 gap = &lp->sl_sal;
|
|
390
|
2191 ga_init2(gap, sizeof(salitem_T), 10);
|
|
375
|
2192 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2193 goto endFAIL;
|
|
|
2194
|
|
|
2195 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2196 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2197 {
|
|
|
2198 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2199 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2200 if (ccnt < 0)
|
|
|
2201 goto formerr;
|
|
|
2202 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2203 goto endFAIL;
|
|
|
2204 smp->sm_lead = p;
|
|
|
2205
|
|
|
2206 /* Read up to the first special char into sm_lead. */
|
|
|
2207 for (i = 0; i < ccnt; ++i)
|
|
|
2208 {
|
|
|
2209 c = getc(fd); /* <salfrom> */
|
|
|
2210 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2211 break;
|
|
|
2212 *p++ = c;
|
|
|
2213 }
|
|
|
2214 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2215 *p++ = NUL;
|
|
|
2216
|
|
|
2217 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2218 if (c == '(')
|
|
|
2219 {
|
|
|
2220 smp->sm_oneof = p;
|
|
|
2221 for (++i; i < ccnt; ++i)
|
|
|
2222 {
|
|
|
2223 c = getc(fd); /* <salfrom> */
|
|
|
2224 if (c == ')')
|
|
|
2225 break;
|
|
|
2226 *p++ = c;
|
|
|
2227 }
|
|
|
2228 *p++ = NUL;
|
|
|
2229 if (++i < ccnt)
|
|
|
2230 c = getc(fd);
|
|
|
2231 }
|
|
372
|
2232 else
|
|
375
|
2233 smp->sm_oneof = NULL;
|
|
|
2234
|
|
|
2235 /* Any following chars go in sm_rules. */
|
|
|
2236 smp->sm_rules = p;
|
|
|
2237 if (i < ccnt)
|
|
|
2238 /* store the char we got while checking for end of sm_lead */
|
|
|
2239 *p++ = c;
|
|
|
2240 for (++i; i < ccnt; ++i)
|
|
|
2241 *p++ = getc(fd); /* <salfrom> */
|
|
|
2242 *p++ = NUL;
|
|
|
2243
|
|
381
|
2244 /* <saltolen> <salto> */
|
|
|
2245 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
389
|
2246 if (ccnt < 0)
|
|
372
|
2247 {
|
|
|
2248 vim_free(smp->sm_lead);
|
|
375
|
2249 goto formerr;
|
|
|
2250 }
|
|
|
2251
|
|
372
|
2252 #ifdef FEAT_MBYTE
|
|
|
2253 if (has_mbyte)
|
|
|
2254 {
|
|
375
|
2255 /* convert the multi-byte strings to wide char strings */
|
|
|
2256 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2257 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2258 if (smp->sm_oneof == NULL)
|
|
|
2259 smp->sm_oneof_w = NULL;
|
|
|
2260 else
|
|
|
2261 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
389
|
2262 if (smp->sm_to == NULL)
|
|
|
2263 smp->sm_to_w = NULL;
|
|
|
2264 else
|
|
|
2265 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
375
|
2266 if (smp->sm_lead_w == NULL
|
|
|
2267 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
389
|
2268 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
372
|
2269 {
|
|
375
|
2270 vim_free(smp->sm_lead);
|
|
|
2271 vim_free(smp->sm_to);
|
|
|
2272 vim_free(smp->sm_lead_w);
|
|
|
2273 vim_free(smp->sm_oneof_w);
|
|
|
2274 vim_free(smp->sm_to_w);
|
|
|
2275 goto endFAIL;
|
|
372
|
2276 }
|
|
|
2277 }
|
|
|
2278 #endif
|
|
|
2279 }
|
|
375
|
2280
|
|
|
2281 /* Fill the first-index table. */
|
|
381
|
2282 set_sal_first(lp);
|
|
|
2283 }
|
|
|
2284
|
|
|
2285 /* <maplen> <mapstr> */
|
|
|
2286 p = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2287 if (cnt < 0)
|
|
323
|
2288 goto endFAIL;
|
|
389
|
2289 if (p != NULL)
|
|
|
2290 {
|
|
|
2291 set_map_str(lp, p);
|
|
|
2292 vim_free(p);
|
|
|
2293 }
|
|
236
|
2294
|
|
300
|
2295 /* round 1: <LWORDTREE>
|
|
339
|
2296 * round 2: <KWORDTREE>
|
|
|
2297 * round 3: <PREFIXTREE> */
|
|
|
2298 for (round = 1; round <= 3; ++round)
|
|
236
|
2299 {
|
|
300
|
2300 /* The tree size was computed when writing the file, so that we can
|
|
|
2301 * allocate it as one long block. <nodecount> */
|
|
|
2302 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2303 if (len < 0)
|
|
|
2304 goto truncerr;
|
|
|
2305 if (len > 0)
|
|
236
|
2306 {
|
|
300
|
2307 /* Allocate the byte array. */
|
|
339
|
2308 bp = lalloc((long_u)len, TRUE);
|
|
|
2309 if (bp == NULL)
|
|
300
|
2310 goto endFAIL;
|
|
|
2311 if (round == 1)
|
|
339
|
2312 lp->sl_fbyts = bp;
|
|
|
2313 else if (round == 2)
|
|
|
2314 lp->sl_kbyts = bp;
|
|
300
|
2315 else
|
|
339
|
2316 lp->sl_pbyts = bp;
|
|
236
|
2317
|
|
300
|
2318 /* Allocate the index array. */
|
|
339
|
2319 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2320 if (ip == NULL)
|
|
300
|
2321 goto endFAIL;
|
|
|
2322 if (round == 1)
|
|
339
|
2323 lp->sl_fidxs = ip;
|
|
|
2324 else if (round == 2)
|
|
|
2325 lp->sl_kidxs = ip;
|
|
300
|
2326 else
|
|
339
|
2327 lp->sl_pidxs = ip;
|
|
300
|
2328
|
|
|
2329 /* Read the tree and store it in the array. */
|
|
339
|
2330 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2331 if (idx == -1)
|
|
300
|
2332 goto truncerr;
|
|
324
|
2333 if (idx < 0)
|
|
236
|
2334 goto formerr;
|
|
|
2335 }
|
|
300
|
2336 }
|
|
243
|
2337
|
|
310
|
2338 /* For a new file link it in the list of spell files. */
|
|
|
2339 if (old_lp == NULL)
|
|
|
2340 {
|
|
|
2341 lp->sl_next = first_lang;
|
|
|
2342 first_lang = lp;
|
|
|
2343 }
|
|
307
|
2344
|
|
255
|
2345 goto endOK;
|
|
|
2346
|
|
|
2347 endFAIL:
|
|
310
|
2348 if (lang != NULL)
|
|
|
2349 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2350 *lang = NUL;
|
|
|
2351 if (lp != NULL && old_lp == NULL)
|
|
323
|
2352 {
|
|
307
|
2353 slang_free(lp);
|
|
323
|
2354 lp = NULL;
|
|
|
2355 }
|
|
255
|
2356
|
|
|
2357 endOK:
|
|
236
|
2358 if (fd != NULL)
|
|
|
2359 fclose(fd);
|
|
|
2360 sourcing_name = save_sourcing_name;
|
|
|
2361 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2362
|
|
|
2363 return lp;
|
|
236
|
2364 }
|
|
|
2365
|
|
376
|
2366 /*
|
|
|
2367 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2368 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2369 * Returns NULL when the count is zero.
|
|
389
|
2370 * Sets "*cntp" to -1 when there is an error, length of the result otherwise.
|
|
376
|
2371 */
|
|
|
2372 static char_u *
|
|
389
|
2373 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2374 FILE *fd;
|
|
|
2375 int cnt_bytes;
|
|
389
|
2376 int *cntp;
|
|
376
|
2377 {
|
|
|
2378 int cnt = 0;
|
|
|
2379 int i;
|
|
|
2380 char_u *str;
|
|
|
2381
|
|
|
2382 /* read the length bytes, MSB first */
|
|
|
2383 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2384 cnt = (cnt << 8) + getc(fd);
|
|
|
2385 if (cnt < 0)
|
|
|
2386 {
|
|
381
|
2387 EMSG(_(e_spell_trunc));
|
|
389
|
2388 *cntp = -1;
|
|
376
|
2389 return NULL;
|
|
|
2390 }
|
|
389
|
2391 *cntp = cnt;
|
|
|
2392 if (cnt == 0)
|
|
|
2393 return NULL; /* nothing to read, return NULL */
|
|
376
|
2394
|
|
|
2395 /* allocate memory */
|
|
|
2396 str = alloc((unsigned)cnt + 1);
|
|
|
2397 if (str == NULL)
|
|
|
2398 {
|
|
389
|
2399 *cntp = -1;
|
|
376
|
2400 return NULL;
|
|
|
2401 }
|
|
|
2402
|
|
|
2403 /* Read the string. Doesn't check for truncated file. */
|
|
|
2404 for (i = 0; i < cnt; ++i)
|
|
|
2405 str[i] = getc(fd);
|
|
|
2406 str[i] = NUL;
|
|
|
2407
|
|
|
2408 return str;
|
|
|
2409 }
|
|
|
2410
|
|
381
|
2411 /*
|
|
|
2412 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
|
2413 * Returns FAIL when there is something wrong.
|
|
|
2414 */
|
|
|
2415 static int
|
|
|
2416 set_sofo(lp, from, to)
|
|
|
2417 slang_T *lp;
|
|
|
2418 char_u *from;
|
|
|
2419 char_u *to;
|
|
|
2420 {
|
|
|
2421 int i;
|
|
|
2422
|
|
|
2423 #ifdef FEAT_MBYTE
|
|
|
2424 garray_T *gap;
|
|
|
2425 char_u *s;
|
|
|
2426 char_u *p;
|
|
|
2427 int c;
|
|
|
2428 int *inp;
|
|
|
2429
|
|
|
2430 if (has_mbyte)
|
|
|
2431 {
|
|
|
2432 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2433 * characters. The index is the low byte of the character.
|
|
|
2434 * The list contains from-to pairs with a terminating NUL.
|
|
|
2435 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2436 gap = &lp->sl_sal;
|
|
|
2437 ga_init2(gap, sizeof(int *), 1);
|
|
|
2438 if (ga_grow(gap, 256) == FAIL)
|
|
|
2439 return FAIL;
|
|
|
2440 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2441 gap->ga_len = 256;
|
|
|
2442
|
|
|
2443 /* First count the number of items for each list. Temporarily use
|
|
|
2444 * sl_sal_first[] for this. */
|
|
|
2445 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2446 {
|
|
474
|
2447 c = mb_cptr2char_adv(&p);
|
|
|
2448 mb_cptr_adv(s);
|
|
381
|
2449 if (c >= 256)
|
|
|
2450 ++lp->sl_sal_first[c & 0xff];
|
|
|
2451 }
|
|
|
2452 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
|
2453 return FAIL;
|
|
|
2454
|
|
|
2455 /* Allocate the lists. */
|
|
|
2456 for (i = 0; i < 256; ++i)
|
|
|
2457 if (lp->sl_sal_first[i] > 0)
|
|
|
2458 {
|
|
|
2459 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2460 if (p == NULL)
|
|
|
2461 return FAIL;
|
|
|
2462 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2463 *(int *)p = 0;
|
|
|
2464 }
|
|
|
2465
|
|
|
2466 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
2467 * list. */
|
|
|
2468 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2469 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2470 {
|
|
474
|
2471 c = mb_cptr2char_adv(&p);
|
|
|
2472 i = mb_cptr2char_adv(&s);
|
|
381
|
2473 if (c >= 256)
|
|
|
2474 {
|
|
|
2475 /* Append the from-to chars at the end of the list with
|
|
|
2476 * the low byte. */
|
|
|
2477 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2478 while (*inp != 0)
|
|
|
2479 ++inp;
|
|
|
2480 *inp++ = c; /* from char */
|
|
|
2481 *inp++ = i; /* to char */
|
|
|
2482 *inp++ = NUL; /* NUL at the end */
|
|
|
2483 }
|
|
|
2484 else
|
|
|
2485 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2486 lp->sl_sal_first[c] = i;
|
|
|
2487 }
|
|
|
2488 }
|
|
|
2489 else
|
|
|
2490 #endif
|
|
|
2491 {
|
|
|
2492 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2493 if (STRLEN(from) != STRLEN(to))
|
|
|
2494 return FAIL;
|
|
|
2495
|
|
|
2496 for (i = 0; to[i] != NUL; ++i)
|
|
|
2497 lp->sl_sal_first[from[i]] = to[i];
|
|
|
2498 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2499 }
|
|
|
2500
|
|
|
2501 return OK;
|
|
|
2502 }
|
|
|
2503
|
|
|
2504 /*
|
|
|
2505 * Fill the first-index table for "lp".
|
|
|
2506 */
|
|
|
2507 static void
|
|
|
2508 set_sal_first(lp)
|
|
|
2509 slang_T *lp;
|
|
|
2510 {
|
|
|
2511 salfirst_T *sfirst;
|
|
|
2512 int i;
|
|
|
2513 salitem_T *smp;
|
|
|
2514 int c;
|
|
|
2515 garray_T *gap = &lp->sl_sal;
|
|
|
2516
|
|
|
2517 sfirst = lp->sl_sal_first;
|
|
|
2518 for (i = 0; i < 256; ++i)
|
|
|
2519 sfirst[i] = -1;
|
|
|
2520 smp = (salitem_T *)gap->ga_data;
|
|
|
2521 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2522 {
|
|
|
2523 #ifdef FEAT_MBYTE
|
|
|
2524 if (has_mbyte)
|
|
|
2525 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2526 * the character, for other encodings it should differ for most
|
|
|
2527 * characters. */
|
|
|
2528 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2529 else
|
|
|
2530 #endif
|
|
|
2531 c = *smp[i].sm_lead;
|
|
|
2532 if (sfirst[c] == -1)
|
|
|
2533 {
|
|
|
2534 sfirst[c] = i;
|
|
|
2535 #ifdef FEAT_MBYTE
|
|
|
2536 if (has_mbyte)
|
|
|
2537 {
|
|
|
2538 int n;
|
|
|
2539
|
|
|
2540 /* Make sure all entries with this byte are following each
|
|
|
2541 * other. Move the ones that are in the wrong position. Do
|
|
|
2542 * keep the same ordering! */
|
|
|
2543 while (i + 1 < gap->ga_len
|
|
|
2544 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
2545 /* Skip over entry with same index byte. */
|
|
|
2546 ++i;
|
|
|
2547
|
|
|
2548 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
2549 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
2550 {
|
|
|
2551 salitem_T tsal;
|
|
|
2552
|
|
|
2553 /* Move entry with same index byte after the entries
|
|
|
2554 * we already found. */
|
|
|
2555 ++i;
|
|
|
2556 --n;
|
|
|
2557 tsal = smp[i + n];
|
|
|
2558 mch_memmove(smp + i + 1, smp + i,
|
|
|
2559 sizeof(salitem_T) * n);
|
|
|
2560 smp[i] = tsal;
|
|
|
2561 }
|
|
|
2562 }
|
|
|
2563 #endif
|
|
|
2564 }
|
|
|
2565 }
|
|
|
2566 }
|
|
376
|
2567
|
|
372
|
2568 #ifdef FEAT_MBYTE
|
|
|
2569 /*
|
|
|
2570 * Turn a multi-byte string into a wide character string.
|
|
|
2571 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
2572 */
|
|
|
2573 static int *
|
|
|
2574 mb_str2wide(s)
|
|
|
2575 char_u *s;
|
|
|
2576 {
|
|
|
2577 int *res;
|
|
|
2578 char_u *p;
|
|
|
2579 int i = 0;
|
|
|
2580
|
|
|
2581 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
2582 if (res != NULL)
|
|
|
2583 {
|
|
|
2584 for (p = s; *p != NUL; )
|
|
|
2585 res[i++] = mb_ptr2char_adv(&p);
|
|
|
2586 res[i] = NUL;
|
|
|
2587 }
|
|
|
2588 return res;
|
|
|
2589 }
|
|
|
2590 #endif
|
|
|
2591
|
|
236
|
2592 /*
|
|
300
|
2593 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
2594 * "byts" and index array "idxs". Recursively read the children.
|
|
|
2595 *
|
|
346
|
2596 * NOTE: The code here must match put_node().
|
|
300
|
2597 *
|
|
|
2598 * Returns the index follosing the siblings.
|
|
|
2599 * Returns -1 if the file is shorter than expected.
|
|
|
2600 * Returns -2 if there is a format error.
|
|
236
|
2601 */
|
|
324
|
2602 static idx_T
|
|
339
|
2603 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
2604 FILE *fd;
|
|
|
2605 char_u *byts;
|
|
324
|
2606 idx_T *idxs;
|
|
300
|
2607 int maxidx; /* size of arrays */
|
|
324
|
2608 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
2609 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
2610 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
2611 {
|
|
300
|
2612 int len;
|
|
|
2613 int i;
|
|
|
2614 int n;
|
|
324
|
2615 idx_T idx = startidx;
|
|
300
|
2616 int c;
|
|
366
|
2617 int c2;
|
|
300
|
2618 #define SHARED_MASK 0x8000000
|
|
236
|
2619
|
|
300
|
2620 len = getc(fd); /* <siblingcount> */
|
|
|
2621 if (len <= 0)
|
|
|
2622 return -1;
|
|
|
2623
|
|
|
2624 if (startidx + len >= maxidx)
|
|
|
2625 return -2;
|
|
|
2626 byts[idx++] = len;
|
|
|
2627
|
|
|
2628 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
2629 for (i = 1; i <= len; ++i)
|
|
236
|
2630 {
|
|
300
|
2631 c = getc(fd); /* <byte> */
|
|
|
2632 if (c < 0)
|
|
|
2633 return -1;
|
|
|
2634 if (c <= BY_SPECIAL)
|
|
|
2635 {
|
|
366
|
2636 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
2637 {
|
|
|
2638 /* No flags, all regions. */
|
|
|
2639 idxs[idx] = 0;
|
|
|
2640 c = 0;
|
|
|
2641 }
|
|
390
|
2642 else if (c != BY_INDEX)
|
|
300
|
2643 {
|
|
339
|
2644 if (prefixtree)
|
|
|
2645 {
|
|
455
|
2646 /* Read the optional pflags byte, the prefix ID and the
|
|
|
2647 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
2648 * byte, the condition index shifted up 8 bits, the flags
|
|
|
2649 * shifted up 24 bits. */
|
|
|
2650 if (c == BY_FLAGS)
|
|
|
2651 c = getc(fd) << 24; /* <pflags> */
|
|
|
2652 else
|
|
|
2653 c = 0;
|
|
|
2654
|
|
481
|
2655 c |= getc(fd); /* <affixID> */
|
|
455
|
2656
|
|
339
|
2657 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
2658 if (n >= maxprefcondnr)
|
|
|
2659 return -2;
|
|
455
|
2660 c |= (n << 8);
|
|
339
|
2661 }
|
|
390
|
2662 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
2663 {
|
|
|
2664 /* Read flags and optional region and prefix ID. In
|
|
390
|
2665 * idxs[] the flags go in the low two bytes, region above
|
|
|
2666 * that and prefix ID above the region. */
|
|
|
2667 c2 = c;
|
|
339
|
2668 c = getc(fd); /* <flags> */
|
|
390
|
2669 if (c2 == BY_FLAGS2)
|
|
|
2670 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
2671 if (c & WF_REGION)
|
|
390
|
2672 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
2673 if (c & WF_AFX)
|
|
|
2674 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
2675 }
|
|
|
2676
|
|
300
|
2677 idxs[idx] = c;
|
|
|
2678 c = 0;
|
|
|
2679 }
|
|
|
2680 else /* c == BY_INDEX */
|
|
|
2681 {
|
|
|
2682 /* <nodeidx> */
|
|
|
2683 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2684 if (n < 0 || n >= maxidx)
|
|
|
2685 return -2;
|
|
|
2686 idxs[idx] = n + SHARED_MASK;
|
|
|
2687 c = getc(fd); /* <xbyte> */
|
|
|
2688 }
|
|
|
2689 }
|
|
|
2690 byts[idx++] = c;
|
|
236
|
2691 }
|
|
|
2692
|
|
300
|
2693 /* Recursively read the children for non-shared siblings.
|
|
|
2694 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
2695 * remove SHARED_MASK) */
|
|
|
2696 for (i = 1; i <= len; ++i)
|
|
|
2697 if (byts[startidx + i] != 0)
|
|
|
2698 {
|
|
|
2699 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
2700 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
2701 else
|
|
|
2702 {
|
|
|
2703 idxs[startidx + i] = idx;
|
|
339
|
2704 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
2705 prefixtree, maxprefcondnr);
|
|
300
|
2706 if (idx < 0)
|
|
|
2707 break;
|
|
|
2708 }
|
|
|
2709 }
|
|
236
|
2710
|
|
300
|
2711 return idx;
|
|
236
|
2712 }
|
|
|
2713
|
|
|
2714 /*
|
|
|
2715 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
2716 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
2717 */
|
|
|
2718 char_u *
|
|
|
2719 did_set_spelllang(buf)
|
|
|
2720 buf_T *buf;
|
|
|
2721 {
|
|
|
2722 garray_T ga;
|
|
351
|
2723 char_u *splp;
|
|
236
|
2724 char_u *region;
|
|
416
|
2725 char_u region_cp[3];
|
|
355
|
2726 int filename;
|
|
236
|
2727 int region_mask;
|
|
|
2728 slang_T *lp;
|
|
|
2729 int c;
|
|
351
|
2730 char_u lang[MAXWLEN + 1];
|
|
323
|
2731 char_u spf_name[MAXPATHL];
|
|
351
|
2732 int len;
|
|
|
2733 char_u *p;
|
|
381
|
2734 int round;
|
|
385
|
2735 char_u *spf;
|
|
389
|
2736 char_u *use_region = NULL;
|
|
|
2737 int dont_use_region = FALSE;
|
|
236
|
2738
|
|
|
2739 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
2740 clear_midword(buf);
|
|
236
|
2741
|
|
351
|
2742 /* loop over comma separated language names. */
|
|
|
2743 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
2744 {
|
|
|
2745 /* Get one language name. */
|
|
|
2746 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
2747
|
|
240
|
2748 region = NULL;
|
|
351
|
2749 len = STRLEN(lang);
|
|
355
|
2750
|
|
|
2751 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
2752 * If there is a region name let "region" point to it and remove it
|
|
|
2753 * from the name. */
|
|
|
2754 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
2755 {
|
|
355
|
2756 filename = TRUE;
|
|
|
2757
|
|
416
|
2758 /* Locate a region and remove it from the file name. */
|
|
|
2759 p = vim_strchr(gettail(lang), '_');
|
|
|
2760 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
2761 && !ASCII_ISALPHA(p[3]))
|
|
|
2762 {
|
|
|
2763 vim_strncpy(region_cp, p + 1, 2);
|
|
|
2764 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
2765 len -= 3;
|
|
|
2766 region = region_cp;
|
|
|
2767 }
|
|
|
2768 else
|
|
|
2769 dont_use_region = TRUE;
|
|
|
2770
|
|
355
|
2771 /* Check if we loaded this language before. */
|
|
|
2772 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2773 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2774 break;
|
|
236
|
2775 }
|
|
355
|
2776 else
|
|
|
2777 {
|
|
|
2778 filename = FALSE;
|
|
|
2779 if (len > 3 && lang[len - 3] == '_')
|
|
|
2780 {
|
|
|
2781 region = lang + len - 2;
|
|
|
2782 len -= 3;
|
|
|
2783 lang[len] = NUL;
|
|
389
|
2784 }
|
|
|
2785 else
|
|
|
2786 dont_use_region = TRUE;
|
|
355
|
2787
|
|
|
2788 /* Check if we loaded this language before. */
|
|
|
2789 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2790 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
2791 break;
|
|
|
2792 }
|
|
236
|
2793
|
|
416
|
2794 if (region != NULL)
|
|
|
2795 {
|
|
|
2796 /* If the region differs from what was used before then don't
|
|
|
2797 * use it for 'spellfile'. */
|
|
|
2798 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
2799 dont_use_region = TRUE;
|
|
|
2800 use_region = region;
|
|
|
2801 }
|
|
|
2802
|
|
351
|
2803 /* If not found try loading the language now. */
|
|
236
|
2804 if (lp == NULL)
|
|
355
|
2805 {
|
|
|
2806 if (filename)
|
|
|
2807 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
2808 else
|
|
|
2809 spell_load_lang(lang);
|
|
|
2810 }
|
|
236
|
2811
|
|
307
|
2812 /*
|
|
351
|
2813 * Loop over the languages, there can be several files for "lang".
|
|
307
|
2814 */
|
|
|
2815 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
2816 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
2817 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
2818 {
|
|
316
|
2819 region_mask = REGION_ALL;
|
|
355
|
2820 if (!filename && region != NULL)
|
|
236
|
2821 {
|
|
307
|
2822 /* find region in sl_regions */
|
|
|
2823 c = find_region(lp->sl_regions, region);
|
|
|
2824 if (c == REGION_ALL)
|
|
|
2825 {
|
|
389
|
2826 if (lp->sl_add)
|
|
|
2827 {
|
|
|
2828 if (*lp->sl_regions != NUL)
|
|
|
2829 /* This addition file is for other regions. */
|
|
|
2830 region_mask = 0;
|
|
|
2831 }
|
|
|
2832 else
|
|
|
2833 /* This is probably an error. Give a warning and
|
|
|
2834 * accept the words anyway. */
|
|
351
|
2835 smsg((char_u *)
|
|
|
2836 _("Warning: region %s not supported"),
|
|
|
2837 region);
|
|
307
|
2838 }
|
|
|
2839 else
|
|
|
2840 region_mask = 1 << c;
|
|
236
|
2841 }
|
|
307
|
2842
|
|
389
|
2843 if (region_mask != 0)
|
|
307
|
2844 {
|
|
389
|
2845 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2846 {
|
|
|
2847 ga_clear(&ga);
|
|
|
2848 return e_outofmem;
|
|
|
2849 }
|
|
|
2850 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2851 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2852 ++ga.ga_len;
|
|
|
2853 use_midword(lp, buf);
|
|
307
|
2854 }
|
|
385
|
2855 }
|
|
|
2856 }
|
|
|
2857
|
|
|
2858 /* round 0: load int_wordlist, if possible.
|
|
|
2859 * round 1: load first name in 'spellfile'.
|
|
|
2860 * round 2: load second name in 'spellfile.
|
|
|
2861 * etc. */
|
|
|
2862 spf = curbuf->b_p_spf;
|
|
|
2863 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
2864 {
|
|
|
2865 if (round == 0)
|
|
|
2866 {
|
|
|
2867 /* Internal wordlist, if there is one. */
|
|
|
2868 if (int_wordlist == NULL)
|
|
381
|
2869 continue;
|
|
385
|
2870 int_wordlist_spl(spf_name);
|
|
381
|
2871 }
|
|
|
2872 else
|
|
|
2873 {
|
|
385
|
2874 /* One entry in 'spellfile'. */
|
|
|
2875 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
2876 STRCAT(spf_name, ".spl");
|
|
|
2877
|
|
|
2878 /* If it was already found above then skip it. */
|
|
|
2879 for (c = 0; c < ga.ga_len; ++c)
|
|
|
2880 if (fullpathcmp(spf_name,
|
|
|
2881 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
2882 FALSE) == FPC_SAME)
|
|
|
2883 break;
|
|
|
2884 if (c < ga.ga_len)
|
|
381
|
2885 continue;
|
|
|
2886 }
|
|
|
2887
|
|
351
|
2888 /* Check if it was loaded already. */
|
|
323
|
2889 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2890 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2891 break;
|
|
|
2892 if (lp == NULL)
|
|
|
2893 {
|
|
351
|
2894 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
2895 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
2896 * use an arbitrary name. */
|
|
|
2897 if (round == 0)
|
|
|
2898 STRCPY(lang, "internal wordlist");
|
|
|
2899 else
|
|
|
2900 {
|
|
|
2901 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
2902 p = vim_strchr(lang, '.');
|
|
|
2903 if (p != NULL)
|
|
|
2904 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
2905 }
|
|
351
|
2906 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
2907 }
|
|
|
2908 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
2909 {
|
|
389
|
2910 region_mask = REGION_ALL;
|
|
|
2911 if (use_region != NULL && !dont_use_region)
|
|
|
2912 {
|
|
|
2913 /* find region in sl_regions */
|
|
|
2914 c = find_region(lp->sl_regions, use_region);
|
|
|
2915 if (c != REGION_ALL)
|
|
|
2916 region_mask = 1 << c;
|
|
|
2917 else if (*lp->sl_regions != NUL)
|
|
|
2918 /* This spell file is for other regions. */
|
|
|
2919 region_mask = 0;
|
|
|
2920 }
|
|
|
2921
|
|
|
2922 if (region_mask != 0)
|
|
|
2923 {
|
|
|
2924 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2925 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2926 ++ga.ga_len;
|
|
|
2927 use_midword(lp, buf);
|
|
|
2928 }
|
|
323
|
2929 }
|
|
|
2930 }
|
|
|
2931
|
|
236
|
2932 /* Add a NULL entry to mark the end of the list. */
|
|
|
2933 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2934 {
|
|
|
2935 ga_clear(&ga);
|
|
|
2936 return e_outofmem;
|
|
|
2937 }
|
|
|
2938 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
2939 ++ga.ga_len;
|
|
|
2940
|
|
|
2941 /* Everything is fine, store the new b_langp value. */
|
|
|
2942 ga_clear(&buf->b_langp);
|
|
|
2943 buf->b_langp = ga;
|
|
|
2944
|
|
|
2945 return NULL;
|
|
|
2946 }
|
|
|
2947
|
|
|
2948 /*
|
|
376
|
2949 * Clear the midword characters for buffer "buf".
|
|
|
2950 */
|
|
|
2951 static void
|
|
|
2952 clear_midword(buf)
|
|
|
2953 buf_T *buf;
|
|
|
2954 {
|
|
|
2955 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
2956 #ifdef FEAT_MBYTE
|
|
|
2957 vim_free(buf->b_spell_ismw_mb);
|
|
|
2958 buf->b_spell_ismw_mb = NULL;
|
|
|
2959 #endif
|
|
|
2960 }
|
|
|
2961
|
|
|
2962 /*
|
|
|
2963 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
2964 * They add up to any currently used midword characters.
|
|
|
2965 */
|
|
|
2966 static void
|
|
|
2967 use_midword(lp, buf)
|
|
|
2968 slang_T *lp;
|
|
|
2969 buf_T *buf;
|
|
|
2970 {
|
|
|
2971 char_u *p;
|
|
|
2972
|
|
389
|
2973 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
2974 return;
|
|
|
2975
|
|
376
|
2976 for (p = lp->sl_midword; *p != NUL; )
|
|
|
2977 #ifdef FEAT_MBYTE
|
|
|
2978 if (has_mbyte)
|
|
|
2979 {
|
|
|
2980 int c, l, n;
|
|
|
2981 char_u *bp;
|
|
|
2982
|
|
|
2983 c = mb_ptr2char(p);
|
|
474
|
2984 l = (*mb_ptr2len)(p);
|
|
|
2985 if (c < 256 && l <= 2)
|
|
376
|
2986 buf->b_spell_ismw[c] = TRUE;
|
|
|
2987 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
2988 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
2989 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
2990 else
|
|
|
2991 {
|
|
|
2992 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
2993 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
2994 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
2995 if (bp != NULL)
|
|
|
2996 {
|
|
|
2997 vim_free(buf->b_spell_ismw_mb);
|
|
|
2998 buf->b_spell_ismw_mb = bp;
|
|
|
2999 vim_strncpy(bp + n, p, l);
|
|
|
3000 }
|
|
|
3001 }
|
|
|
3002 p += l;
|
|
|
3003 }
|
|
|
3004 else
|
|
|
3005 #endif
|
|
|
3006 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
3007 }
|
|
|
3008
|
|
|
3009 /*
|
|
236
|
3010 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
3011 * Each region is simply stored as the two characters of it's name.
|
|
381
|
3012 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
3013 */
|
|
|
3014 static int
|
|
|
3015 find_region(rp, region)
|
|
|
3016 char_u *rp;
|
|
|
3017 char_u *region;
|
|
|
3018 {
|
|
|
3019 int i;
|
|
|
3020
|
|
|
3021 for (i = 0; ; i += 2)
|
|
|
3022 {
|
|
|
3023 if (rp[i] == NUL)
|
|
|
3024 return REGION_ALL;
|
|
|
3025 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
3026 break;
|
|
|
3027 }
|
|
|
3028 return i / 2;
|
|
|
3029 }
|
|
|
3030
|
|
|
3031 /*
|
|
323
|
3032 * Return case type of word:
|
|
236
|
3033 * w word 0
|
|
300
|
3034 * Word WF_ONECAP
|
|
|
3035 * W WORD WF_ALLCAP
|
|
|
3036 * WoRd wOrd WF_KEEPCAP
|
|
236
|
3037 */
|
|
|
3038 static int
|
|
|
3039 captype(word, end)
|
|
|
3040 char_u *word;
|
|
323
|
3041 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
3042 {
|
|
|
3043 char_u *p;
|
|
|
3044 int c;
|
|
|
3045 int firstcap;
|
|
|
3046 int allcap;
|
|
|
3047 int past_second = FALSE; /* past second word char */
|
|
|
3048
|
|
|
3049 /* find first letter */
|
|
376
|
3050 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
3051 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
3052 return 0; /* only non-word characters, illegal word */
|
|
|
3053 #ifdef FEAT_MBYTE
|
|
310
|
3054 if (has_mbyte)
|
|
|
3055 c = mb_ptr2char_adv(&p);
|
|
|
3056 else
|
|
236
|
3057 #endif
|
|
310
|
3058 c = *p++;
|
|
324
|
3059 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
3060
|
|
|
3061 /*
|
|
|
3062 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
3063 * But a word with an upper char only at start is a ONECAP.
|
|
|
3064 */
|
|
323
|
3065 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
3066 if (spell_iswordp_nmw(p))
|
|
236
|
3067 {
|
|
455
|
3068 c = PTR2CHAR(p);
|
|
324
|
3069 if (!SPELL_ISUPPER(c))
|
|
236
|
3070 {
|
|
|
3071 /* UUl -> KEEPCAP */
|
|
|
3072 if (past_second && allcap)
|
|
300
|
3073 return WF_KEEPCAP;
|
|
236
|
3074 allcap = FALSE;
|
|
|
3075 }
|
|
|
3076 else if (!allcap)
|
|
|
3077 /* UlU -> KEEPCAP */
|
|
300
|
3078 return WF_KEEPCAP;
|
|
236
|
3079 past_second = TRUE;
|
|
|
3080 }
|
|
|
3081
|
|
|
3082 if (allcap)
|
|
300
|
3083 return WF_ALLCAP;
|
|
236
|
3084 if (firstcap)
|
|
300
|
3085 return WF_ONECAP;
|
|
236
|
3086 return 0;
|
|
|
3087 }
|
|
|
3088
|
|
474
|
3089 /*
|
|
|
3090 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
3091 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
3092 * Add ALLCAP for "WOrD".
|
|
|
3093 */
|
|
|
3094 static int
|
|
|
3095 badword_captype(word, end)
|
|
|
3096 char_u *word;
|
|
|
3097 char_u *end;
|
|
|
3098 {
|
|
|
3099 int flags = captype(word, end);
|
|
475
|
3100 int c;
|
|
474
|
3101 int l, u;
|
|
|
3102 int first;
|
|
|
3103 char_u *p;
|
|
|
3104
|
|
|
3105 if (flags & WF_KEEPCAP)
|
|
|
3106 {
|
|
|
3107 /* Count the number of UPPER and lower case letters. */
|
|
|
3108 l = u = 0;
|
|
|
3109 first = FALSE;
|
|
|
3110 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
3111 {
|
|
475
|
3112 c = PTR2CHAR(p);
|
|
|
3113 if (SPELL_ISUPPER(c))
|
|
474
|
3114 {
|
|
|
3115 ++u;
|
|
|
3116 if (p == word)
|
|
|
3117 first = TRUE;
|
|
|
3118 }
|
|
|
3119 else
|
|
|
3120 ++l;
|
|
|
3121 }
|
|
|
3122
|
|
|
3123 /* If there are more UPPER than lower case letters suggest an
|
|
|
3124 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
3125 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
3126 * require three upper case letters. */
|
|
|
3127 if (u > l && u > 2)
|
|
|
3128 flags |= WF_ALLCAP;
|
|
|
3129 else if (first)
|
|
|
3130 flags |= WF_ONECAP;
|
|
|
3131 }
|
|
|
3132 return flags;
|
|
|
3133 }
|
|
|
3134
|
|
355
|
3135 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
3136 /*
|
|
|
3137 * Free all languages.
|
|
|
3138 */
|
|
|
3139 void
|
|
|
3140 spell_free_all()
|
|
|
3141 {
|
|
|
3142 slang_T *lp;
|
|
|
3143 buf_T *buf;
|
|
385
|
3144 char_u fname[MAXPATHL];
|
|
355
|
3145
|
|
|
3146 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3147 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3148 ga_clear(&buf->b_langp);
|
|
|
3149
|
|
|
3150 while (first_lang != NULL)
|
|
|
3151 {
|
|
|
3152 lp = first_lang;
|
|
|
3153 first_lang = lp->sl_next;
|
|
|
3154 slang_free(lp);
|
|
|
3155 }
|
|
366
|
3156
|
|
385
|
3157 if (int_wordlist != NULL)
|
|
|
3158 {
|
|
|
3159 /* Delete the internal wordlist and its .spl file */
|
|
|
3160 mch_remove(int_wordlist);
|
|
|
3161 int_wordlist_spl(fname);
|
|
|
3162 mch_remove(fname);
|
|
|
3163 vim_free(int_wordlist);
|
|
|
3164 int_wordlist = NULL;
|
|
381
|
3165 }
|
|
|
3166
|
|
366
|
3167 init_spell_chartab();
|
|
355
|
3168 }
|
|
|
3169 # endif
|
|
|
3170
|
|
236
|
3171 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
3172 /*
|
|
|
3173 * Clear all spelling tables and reload them.
|
|
307
|
3174 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
3175 */
|
|
|
3176 void
|
|
|
3177 spell_reload()
|
|
|
3178 {
|
|
|
3179 buf_T *buf;
|
|
316
|
3180 win_T *wp;
|
|
236
|
3181
|
|
358
|
3182 /* Initialize the table for spell_iswordp(). */
|
|
236
|
3183 init_spell_chartab();
|
|
|
3184
|
|
|
3185 /* Unload all allocated memory. */
|
|
355
|
3186 spell_free_all();
|
|
236
|
3187
|
|
|
3188 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3189 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3190 {
|
|
316
|
3191 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
3192 * window for this buffer in which 'spell' is set. */
|
|
236
|
3193 if (*buf->b_p_spl != NUL)
|
|
316
|
3194 {
|
|
|
3195 FOR_ALL_WINDOWS(wp)
|
|
|
3196 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
3197 {
|
|
|
3198 (void)did_set_spelllang(buf);
|
|
|
3199 # ifdef FEAT_WINDOWS
|
|
|
3200 break;
|
|
|
3201 # endif
|
|
|
3202 }
|
|
|
3203 }
|
|
236
|
3204 }
|
|
|
3205 }
|
|
|
3206 # endif
|
|
|
3207
|
|
310
|
3208 /*
|
|
|
3209 * Reload the spell file "fname" if it's loaded.
|
|
|
3210 */
|
|
|
3211 static void
|
|
323
|
3212 spell_reload_one(fname, added_word)
|
|
310
|
3213 char_u *fname;
|
|
323
|
3214 int added_word; /* invoked through "zg" */
|
|
310
|
3215 {
|
|
|
3216 slang_T *lp;
|
|
323
|
3217 int didit = FALSE;
|
|
310
|
3218
|
|
|
3219 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3220 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3221 {
|
|
|
3222 slang_clear(lp);
|
|
323
|
3223 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3224 redraw_all_later(NOT_VALID);
|
|
323
|
3225 didit = TRUE;
|
|
310
|
3226 }
|
|
323
|
3227
|
|
|
3228 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3229 * 'spelllang' to get it loaded. */
|
|
|
3230 if (added_word && !didit)
|
|
|
3231 did_set_spelllang(curbuf);
|
|
310
|
3232 }
|
|
|
3233
|
|
|
3234
|
|
236
|
3235 /*
|
|
|
3236 * Functions for ":mkspell".
|
|
|
3237 */
|
|
|
3238
|
|
300
|
3239 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3240 and .dic file. */
|
|
|
3241 /*
|
|
|
3242 * Main structure to store the contents of a ".aff" file.
|
|
|
3243 */
|
|
|
3244 typedef struct afffile_S
|
|
|
3245 {
|
|
|
3246 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
481
|
3247 int af_slash; /* character used in word for slash */
|
|
310
|
3248 int af_rar; /* RAR ID for rare word */
|
|
|
3249 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
3250 int af_bad; /* BAD ID for banned word */
|
|
481
|
3251 char_u *af_compflags; /* COMPOUNDFLAG or COMPOUNDFLAGS */
|
|
|
3252 int af_compminlen; /* COMPOUNDMIN */
|
|
339
|
3253 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3254 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3255 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
3256 } afffile_T;
|
|
|
3257
|
|
|
3258 typedef struct affentry_S affentry_T;
|
|
|
3259 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3260 struct affentry_S
|
|
|
3261 {
|
|
|
3262 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3263 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3264 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3265 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3266 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
366
|
3267 int ae_rare; /* rare affix */
|
|
300
|
3268 };
|
|
|
3269
|
|
455
|
3270 #define AH_KEY_LEN 10
|
|
|
3271
|
|
300
|
3272 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3273 typedef struct affheader_S
|
|
|
3274 {
|
|
455
|
3275 /* key for hashtable == name of affix entry */
|
|
|
3276 #ifdef FEAT_MBYTE
|
|
|
3277 char_u ah_key[AH_KEY_LEN]; /* multi-byte char plus NUL */
|
|
|
3278 #else
|
|
|
3279 char_u ah_key[2]; /* one byte char plus NUL */
|
|
|
3280 #endif
|
|
339
|
3281 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
3282 int ah_combine; /* suffix may combine with prefix */
|
|
|
3283 affentry_T *ah_first; /* first affix entry */
|
|
|
3284 } affheader_T;
|
|
|
3285
|
|
|
3286 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
3287
|
|
|
3288 /*
|
|
|
3289 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
3290 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
3291 * once after ":mkspell" is done.
|
|
300
|
3292 */
|
|
|
3293 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
3294 typedef struct sblock_S sblock_T;
|
|
|
3295 struct sblock_S
|
|
|
3296 {
|
|
|
3297 sblock_T *sb_next; /* next block in list */
|
|
|
3298 int sb_used; /* nr of bytes already in use */
|
|
|
3299 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
3300 };
|
|
|
3301
|
|
|
3302 /*
|
|
300
|
3303 * A node in the tree.
|
|
236
|
3304 */
|
|
300
|
3305 typedef struct wordnode_S wordnode_T;
|
|
|
3306 struct wordnode_S
|
|
236
|
3307 {
|
|
346
|
3308 union /* shared to save space */
|
|
|
3309 {
|
|
469
|
3310 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
3311 int index; /* index in written nodes (valid after first
|
|
|
3312 round) */
|
|
|
3313 } wn_u1;
|
|
|
3314 union /* shared to save space */
|
|
|
3315 {
|
|
|
3316 wordnode_T *next; /* next node with same hash key */
|
|
|
3317 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
3318 } wn_u2;
|
|
300
|
3319 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
3320 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
3321 always sorted) */
|
|
474
|
3322 int wn_refs; /* Nr. of references to this node. Only
|
|
|
3323 relevant for first node in a list of
|
|
|
3324 siblings, in following siblings it is
|
|
|
3325 always one. */
|
|
300
|
3326 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
481
|
3327 char_u wn_affixID; /* when "wn_byte" is NUL: supported/required
|
|
469
|
3328 prefix ID or 0 */
|
|
|
3329 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
3330 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
3331 PREFIXTREE it's the prefcondnr */
|
|
469
|
3332 #ifdef SPELL_PRINTTREE
|
|
|
3333 int wn_nr; /* sequence nr for printing */
|
|
|
3334 #endif
|
|
236
|
3335 };
|
|
|
3336
|
|
390
|
3337 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
3338
|
|
300
|
3339 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
3340
|
|
300
|
3341 /*
|
|
|
3342 * Info used while reading the spell files.
|
|
|
3343 */
|
|
|
3344 typedef struct spellinfo_S
|
|
249
|
3345 {
|
|
300
|
3346 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
3347 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
3348
|
|
300
|
3349 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
3350 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
3351
|
|
339
|
3352 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
3353
|
|
300
|
3354 sblock_T *si_blocks; /* memory blocks used */
|
|
482
|
3355 long si_blocks_cnt; /* memory blocks allocated */
|
|
|
3356 long si_compress_cnt; /* words to add before lowering
|
|
|
3357 compression limit */
|
|
474
|
3358 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
3359 compression, linked by "wn_child" field. */
|
|
482
|
3360 long si_free_count; /* number of nodes in si_first_free */
|
|
474
|
3361 #ifdef SPELL_PRINTTREE
|
|
|
3362 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
3363 #endif
|
|
|
3364
|
|
|
3365
|
|
300
|
3366 int si_ascii; /* handling only ASCII words */
|
|
310
|
3367 int si_add; /* addition file */
|
|
351
|
3368 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
3369 int si_region; /* region mask */
|
|
|
3370 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
3371 int si_memtot; /* runtime memory used */
|
|
310
|
3372 int si_verbose; /* verbose messages */
|
|
474
|
3373 int si_msg_count; /* number of words added since last message */
|
|
316
|
3374 int si_region_count; /* number of regions supported (1 when there
|
|
|
3375 are no regions) */
|
|
|
3376 char_u si_region_name[16]; /* region names (if count > 1) */
|
|
323
|
3377
|
|
|
3378 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
3379 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
3380 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
3381 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
3382 int si_followup; /* soundsalike: ? */
|
|
|
3383 int si_collapse; /* soundsalike: ? */
|
|
|
3384 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
3385 garray_T si_map; /* MAP info concatenated */
|
|
366
|
3386 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
481
|
3387 int si_compminlen; /* minimal length for compounding */
|
|
|
3388 char_u *si_compflags; /* flags used for compounding */
|
|
339
|
3389 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
3390 * prefixes, each stored as a string */
|
|
|
3391 int si_newID; /* current value for ah_newID */
|
|
300
|
3392 } spellinfo_T;
|
|
249
|
3393
|
|
474
|
3394 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
339
|
3395 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
3396 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
3397 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
3398 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
3399 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
3400 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
|
3401 static char_u *get_pfxlist __ARGS((spellinfo_T *spin, afffile_T *affile, char_u *afflist));
|
|
481
|
3402 static char_u *get_compflags __ARGS((spellinfo_T *spin, char_u *afflist));
|
|
474
|
3403 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));
|
|
|
3404 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
3405 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
3406 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
3407 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
3408 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
|
3409 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist));
|
|
481
|
3410 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
3411 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
3412 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
3413 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
3414 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
3415 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
3416 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
474
|
3417 static void write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
3418 static void clear_node __ARGS((wordnode_T *node));
|
|
|
3419 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
3420 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3421 static void init_spellfile __ARGS((void));
|
|
236
|
3422
|
|
455
|
3423 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
3424 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
3425 * Use a negative number with the lower 8 bits zero. */
|
|
|
3426 #define PFX_FLAGS -256
|
|
390
|
3427
|
|
469
|
3428 #ifdef SPELL_PRINTTREE
|
|
|
3429 /*
|
|
|
3430 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
3431 * readable format, so that we can see what happens when adding a word and/or
|
|
|
3432 * compressing the tree.
|
|
|
3433 * Based on code from Olaf Seibert.
|
|
|
3434 */
|
|
|
3435 #define PRINTLINESIZE 1000
|
|
|
3436 #define PRINTWIDTH 6
|
|
|
3437
|
|
|
3438 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
3439 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
3440
|
|
|
3441 static char line1[PRINTLINESIZE];
|
|
|
3442 static char line2[PRINTLINESIZE];
|
|
|
3443 static char line3[PRINTLINESIZE];
|
|
|
3444
|
|
|
3445 static void
|
|
|
3446 spell_clear_flags(wordnode_T *node)
|
|
|
3447 {
|
|
|
3448 wordnode_T *np;
|
|
|
3449
|
|
|
3450 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
3451 {
|
|
|
3452 np->wn_u1.index = FALSE;
|
|
|
3453 spell_clear_flags(np->wn_child);
|
|
|
3454 }
|
|
|
3455 }
|
|
|
3456
|
|
|
3457 static void
|
|
|
3458 spell_print_node(wordnode_T *node, int depth)
|
|
|
3459 {
|
|
|
3460 if (node->wn_u1.index)
|
|
|
3461 {
|
|
|
3462 /* Done this node before, print the reference. */
|
|
|
3463 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
3464 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
3465 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3466 msg(line1);
|
|
|
3467 msg(line2);
|
|
|
3468 msg(line3);
|
|
|
3469 }
|
|
|
3470 else
|
|
|
3471 {
|
|
|
3472 node->wn_u1.index = TRUE;
|
|
|
3473
|
|
|
3474 if (node->wn_byte != NUL)
|
|
|
3475 {
|
|
|
3476 if (node->wn_child != NULL)
|
|
|
3477 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
3478 else
|
|
|
3479 /* Cannot happen? */
|
|
|
3480 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
3481 }
|
|
|
3482 else
|
|
|
3483 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
3484
|
|
|
3485 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
3486
|
|
|
3487 if (node->wn_sibling != NULL)
|
|
|
3488 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
3489 else
|
|
|
3490 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3491
|
|
|
3492 if (node->wn_byte == NUL)
|
|
|
3493 {
|
|
|
3494 msg(line1);
|
|
|
3495 msg(line2);
|
|
|
3496 msg(line3);
|
|
|
3497 }
|
|
|
3498
|
|
|
3499 /* do the children */
|
|
|
3500 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
3501 spell_print_node(node->wn_child, depth + 1);
|
|
|
3502
|
|
|
3503 /* do the siblings */
|
|
|
3504 if (node->wn_sibling != NULL)
|
|
|
3505 {
|
|
|
3506 /* get rid of all parent details except | */
|
|
|
3507 STRCPY(line1, line3);
|
|
|
3508 STRCPY(line2, line3);
|
|
|
3509 spell_print_node(node->wn_sibling, depth);
|
|
|
3510 }
|
|
|
3511 }
|
|
|
3512 }
|
|
|
3513
|
|
|
3514 static void
|
|
|
3515 spell_print_tree(wordnode_T *root)
|
|
|
3516 {
|
|
|
3517 if (root != NULL)
|
|
|
3518 {
|
|
|
3519 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
3520 * done. */
|
|
|
3521 spell_clear_flags(root);
|
|
|
3522
|
|
|
3523 /* Recursively print the tree. */
|
|
|
3524 spell_print_node(root, 0);
|
|
|
3525 }
|
|
|
3526 }
|
|
|
3527 #endif /* SPELL_PRINTTREE */
|
|
|
3528
|
|
236
|
3529 /*
|
|
323
|
3530 * Read the affix file "fname".
|
|
316
|
3531 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
3532 */
|
|
|
3533 static afffile_T *
|
|
474
|
3534 spell_read_aff(spin, fname)
|
|
|
3535 spellinfo_T *spin;
|
|
236
|
3536 char_u *fname;
|
|
|
3537 {
|
|
|
3538 FILE *fd;
|
|
|
3539 afffile_T *aff;
|
|
|
3540 char_u rline[MAXLINELEN];
|
|
|
3541 char_u *line;
|
|
|
3542 char_u *pc = NULL;
|
|
334
|
3543 #define MAXITEMCNT 7
|
|
|
3544 char_u *(items[MAXITEMCNT]);
|
|
236
|
3545 int itemcnt;
|
|
|
3546 char_u *p;
|
|
|
3547 int lnum = 0;
|
|
|
3548 affheader_T *cur_aff = NULL;
|
|
|
3549 int aff_todo = 0;
|
|
|
3550 hashtab_T *tp;
|
|
255
|
3551 char_u *low = NULL;
|
|
|
3552 char_u *fol = NULL;
|
|
|
3553 char_u *upp = NULL;
|
|
307
|
3554 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
3555 int do_rep;
|
|
|
3556 int do_sal;
|
|
|
3557 int do_map;
|
|
366
|
3558 int do_midword;
|
|
375
|
3559 int do_sofo;
|
|
323
|
3560 int found_map = FALSE;
|
|
324
|
3561 hashitem_T *hi;
|
|
455
|
3562 int l;
|
|
236
|
3563
|
|
300
|
3564 /*
|
|
|
3565 * Open the file.
|
|
|
3566 */
|
|
310
|
3567 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3568 if (fd == NULL)
|
|
|
3569 {
|
|
|
3570 EMSG2(_(e_notopen), fname);
|
|
|
3571 return NULL;
|
|
|
3572 }
|
|
|
3573
|
|
310
|
3574 if (spin->si_verbose || p_verbose > 2)
|
|
|
3575 {
|
|
|
3576 if (!spin->si_verbose)
|
|
|
3577 verbose_enter();
|
|
366
|
3578 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
3579 out_flush();
|
|
|
3580 if (!spin->si_verbose)
|
|
|
3581 verbose_leave();
|
|
|
3582 }
|
|
236
|
3583
|
|
323
|
3584 /* Only do REP lines when not done in another .aff file already. */
|
|
|
3585 do_rep = spin->si_rep.ga_len == 0;
|
|
|
3586
|
|
|
3587 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
3588 do_sal = spin->si_sal.ga_len == 0;
|
|
|
3589
|
|
|
3590 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
3591 do_map = spin->si_map.ga_len == 0;
|
|
|
3592
|
|
366
|
3593 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
3594 do_midword = spin->si_midword == NULL;
|
|
|
3595
|
|
375
|
3596 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
3597 do_sofo = spin->si_sofofr == NULL;
|
|
|
3598
|
|
300
|
3599 /*
|
|
|
3600 * Allocate and init the afffile_T structure.
|
|
|
3601 */
|
|
474
|
3602 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
3603 if (aff == NULL)
|
|
|
3604 return NULL;
|
|
|
3605 hash_init(&aff->af_pref);
|
|
|
3606 hash_init(&aff->af_suff);
|
|
|
3607
|
|
|
3608 /*
|
|
|
3609 * Read all the lines in the file one by one.
|
|
|
3610 */
|
|
255
|
3611 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
3612 {
|
|
255
|
3613 line_breakcheck();
|
|
236
|
3614 ++lnum;
|
|
|
3615
|
|
|
3616 /* Skip comment lines. */
|
|
|
3617 if (*rline == '#')
|
|
|
3618 continue;
|
|
|
3619
|
|
|
3620 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
3621 vim_free(pc);
|
|
310
|
3622 #ifdef FEAT_MBYTE
|
|
300
|
3623 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3624 {
|
|
300
|
3625 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
3626 if (pc == NULL)
|
|
|
3627 {
|
|
|
3628 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3629 fname, lnum, rline);
|
|
|
3630 continue;
|
|
|
3631 }
|
|
236
|
3632 line = pc;
|
|
|
3633 }
|
|
|
3634 else
|
|
310
|
3635 #endif
|
|
236
|
3636 {
|
|
|
3637 pc = NULL;
|
|
|
3638 line = rline;
|
|
|
3639 }
|
|
|
3640
|
|
|
3641 /* Split the line up in white separated items. Put a NUL after each
|
|
|
3642 * item. */
|
|
|
3643 itemcnt = 0;
|
|
|
3644 for (p = line; ; )
|
|
|
3645 {
|
|
|
3646 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
3647 ++p;
|
|
|
3648 if (*p == NUL)
|
|
|
3649 break;
|
|
334
|
3650 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
3651 break;
|
|
236
|
3652 items[itemcnt++] = p;
|
|
300
|
3653 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
3654 ++p;
|
|
|
3655 if (*p == NUL)
|
|
|
3656 break;
|
|
|
3657 *p++ = NUL;
|
|
|
3658 }
|
|
|
3659
|
|
|
3660 /* Handle non-empty lines. */
|
|
|
3661 if (itemcnt > 0)
|
|
|
3662 {
|
|
|
3663 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
3664 && aff->af_enc == NULL)
|
|
|
3665 {
|
|
310
|
3666 #ifdef FEAT_MBYTE
|
|
300
|
3667 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
3668 aff->af_enc = enc_canonize(items[1]);
|
|
|
3669 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
3670 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
3671 p_enc) == FAIL)
|
|
|
3672 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
3673 fname, aff->af_enc, p_enc);
|
|
375
|
3674 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3675 #else
|
|
|
3676 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3677 #endif
|
|
236
|
3678 }
|
|
366
|
3679 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
3680 {
|
|
|
3681 if (do_midword)
|
|
|
3682 spin->si_midword = vim_strsave(items[1]);
|
|
|
3683 }
|
|
302
|
3684 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
3685 {
|
|
323
|
3686 /* ignored, we always split */
|
|
302
|
3687 }
|
|
323
|
3688 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
3689 {
|
|
323
|
3690 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
3691 }
|
|
481
|
3692 else if (STRCMP(items[0], "SLASH") == 0 && itemcnt == 2
|
|
|
3693 && aff->af_slash == 0)
|
|
|
3694 {
|
|
|
3695 aff->af_slash = items[1][0];
|
|
|
3696 if (items[1][1] != NUL)
|
|
|
3697 smsg((char_u *)_("Character used for SLASH must be ASCII; in %s line %d: %s"),
|
|
|
3698 fname, lnum, items[1]);
|
|
|
3699 }
|
|
307
|
3700 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
3701 && aff->af_rar == 0)
|
|
|
3702 {
|
|
|
3703 aff->af_rar = items[1][0];
|
|
|
3704 if (items[1][1] != NUL)
|
|
|
3705 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3706 }
|
|
310
|
3707 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
3708 && aff->af_kep == 0)
|
|
307
|
3709 {
|
|
310
|
3710 aff->af_kep = items[1][0];
|
|
307
|
3711 if (items[1][1] != NUL)
|
|
|
3712 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3713 }
|
|
346
|
3714 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
3715 && aff->af_bad == 0)
|
|
|
3716 {
|
|
|
3717 aff->af_bad = items[1][0];
|
|
|
3718 if (items[1][1] != NUL)
|
|
|
3719 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3720 }
|
|
481
|
3721 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
|
3722 && aff->af_compflags == 0)
|
|
|
3723 {
|
|
|
3724 aff->af_compflags = getroom_save(spin, items[1]);
|
|
|
3725 if (items[1][1] != NUL)
|
|
|
3726 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3727 }
|
|
|
3728 else if (STRCMP(items[0], "COMPOUNDFLAGS") == 0 && itemcnt == 2
|
|
|
3729 && aff->af_compflags == 0)
|
|
|
3730 {
|
|
|
3731 aff->af_compflags = getroom_save(spin, items[1]);
|
|
|
3732 }
|
|
|
3733 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
|
3734 && aff->af_compminlen == 0)
|
|
|
3735 {
|
|
|
3736 aff->af_compminlen = atoi((char *)items[1]);
|
|
|
3737 if (aff->af_compminlen == 0)
|
|
|
3738 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
3739 fname, lnum, items[1]);
|
|
|
3740 }
|
|
339
|
3741 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
3742 {
|
|
|
3743 aff->af_pfxpostpone = TRUE;
|
|
|
3744 }
|
|
236
|
3745 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3746 || STRCMP(items[0], "SFX") == 0)
|
|
|
3747 && aff_todo == 0
|
|
334
|
3748 && itemcnt >= 4)
|
|
236
|
3749 {
|
|
334
|
3750 /* Myspell allows extra text after the item, but that might
|
|
|
3751 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
3752 if (itemcnt > 4 && *items[4] != '#')
|
|
|
3753 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
3754 fname, lnum, items[4]);
|
|
|
3755
|
|
236
|
3756 /* New affix letter. */
|
|
474
|
3757 cur_aff = (affheader_T *)getroom(spin,
|
|
447
|
3758 sizeof(affheader_T), TRUE);
|
|
236
|
3759 if (cur_aff == NULL)
|
|
|
3760 break;
|
|
455
|
3761 #ifdef FEAT_MBYTE
|
|
|
3762 if (has_mbyte)
|
|
|
3763 {
|
|
474
|
3764 l = (*mb_ptr2len)(items[1]);
|
|
455
|
3765 if (l >= AH_KEY_LEN)
|
|
|
3766 l = 1; /* too long, must be an overlong sequence */
|
|
|
3767 else
|
|
|
3768 mch_memmove(cur_aff->ah_key, items[1], l);
|
|
|
3769 }
|
|
|
3770 else
|
|
|
3771 #endif
|
|
|
3772 {
|
|
|
3773 *cur_aff->ah_key = *items[1];
|
|
|
3774 l = 1;
|
|
|
3775 }
|
|
|
3776 cur_aff->ah_key[l] = NUL;
|
|
|
3777 if (items[1][l] != NUL)
|
|
307
|
3778 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
3779 if (*items[2] == 'Y')
|
|
|
3780 cur_aff->ah_combine = TRUE;
|
|
300
|
3781 else if (*items[2] != 'N')
|
|
236
|
3782 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
3783 fname, lnum, items[2]);
|
|
339
|
3784
|
|
236
|
3785 if (*items[0] == 'P')
|
|
339
|
3786 {
|
|
236
|
3787 tp = &aff->af_pref;
|
|
339
|
3788 /* Use a new number in the .spl file later, to be able to
|
|
|
3789 * handle multiple .aff files. */
|
|
|
3790 if (aff->af_pfxpostpone)
|
|
344
|
3791 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
3792 }
|
|
236
|
3793 else
|
|
|
3794 tp = &aff->af_suff;
|
|
300
|
3795 aff_todo = atoi((char *)items[3]);
|
|
324
|
3796 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
3797 if (!HASHITEM_EMPTY(hi))
|
|
300
|
3798 {
|
|
236
|
3799 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
3800 fname, lnum, items[1]);
|
|
300
|
3801 aff_todo = 0;
|
|
|
3802 }
|
|
236
|
3803 else
|
|
|
3804 hash_add(tp, cur_aff->ah_key);
|
|
|
3805 }
|
|
|
3806 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3807 || STRCMP(items[0], "SFX") == 0)
|
|
|
3808 && aff_todo > 0
|
|
|
3809 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
3810 && itemcnt >= 5)
|
|
236
|
3811 {
|
|
|
3812 affentry_T *aff_entry;
|
|
366
|
3813 int rare = FALSE;
|
|
455
|
3814 int upper = FALSE;
|
|
366
|
3815 int lasti = 5;
|
|
|
3816
|
|
|
3817 /* Check for "rare" after the other info. */
|
|
|
3818 if (itemcnt > 5 && STRICMP(items[5], "rare") == 0)
|
|
|
3819 {
|
|
|
3820 rare = TRUE;
|
|
|
3821 lasti = 6;
|
|
|
3822 }
|
|
236
|
3823
|
|
334
|
3824 /* Myspell allows extra text after the item, but that might
|
|
|
3825 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
3826 if (itemcnt > lasti && *items[lasti] != '#')
|
|
482
|
3827 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]);
|
|
334
|
3828
|
|
236
|
3829 /* New item for an affix letter. */
|
|
|
3830 --aff_todo;
|
|
474
|
3831 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
3832 sizeof(affentry_T), TRUE);
|
|
236
|
3833 if (aff_entry == NULL)
|
|
|
3834 break;
|
|
366
|
3835 aff_entry->ae_rare = rare;
|
|
240
|
3836
|
|
236
|
3837 if (STRCMP(items[2], "0") != 0)
|
|
474
|
3838 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
3839 if (STRCMP(items[3], "0") != 0)
|
|
474
|
3840 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
3841
|
|
300
|
3842 /* Don't use an affix entry with non-ASCII characters when
|
|
|
3843 * "spin->si_ascii" is TRUE. */
|
|
|
3844 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
3845 || has_non_ascii(aff_entry->ae_add)))
|
|
|
3846 {
|
|
|
3847 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
3848 cur_aff->ah_first = aff_entry;
|
|
300
|
3849
|
|
|
3850 if (STRCMP(items[4], ".") != 0)
|
|
|
3851 {
|
|
|
3852 char_u buf[MAXLINELEN];
|
|
|
3853
|
|
474
|
3854 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
3855 if (*items[0] == 'P')
|
|
|
3856 sprintf((char *)buf, "^%s", items[4]);
|
|
|
3857 else
|
|
|
3858 sprintf((char *)buf, "%s$", items[4]);
|
|
|
3859 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
3860 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
3861 if (aff_entry->ae_prog == NULL)
|
|
|
3862 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
3863 fname, lnum, items[4]);
|
|
300
|
3864 }
|
|
339
|
3865
|
|
|
3866 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
3867 * for the condition. Use an existing one if possible. */
|
|
455
|
3868 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
3869 {
|
|
455
|
3870 /* When the chop string is one lower-case letter and
|
|
|
3871 * the add string ends in the upper-case letter we set
|
|
|
3872 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
3873 * letters from "ae_add". The condition must either
|
|
|
3874 * be empty or start with the same letter. */
|
|
|
3875 if (aff_entry->ae_chop != NULL
|
|
|
3876 && aff_entry->ae_add != NULL
|
|
|
3877 #ifdef FEAT_MBYTE
|
|
474
|
3878 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
3879 aff_entry->ae_chop)] == NUL
|
|
|
3880 #else
|
|
|
3881 && aff_entry->ae_chop[1] == NUL
|
|
|
3882 #endif
|
|
|
3883 )
|
|
339
|
3884 {
|
|
455
|
3885 int c, c_up;
|
|
|
3886
|
|
|
3887 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
3888 c_up = SPELL_TOUPPER(c);
|
|
|
3889 if (c_up != c
|
|
|
3890 && (aff_entry->ae_cond == NULL
|
|
|
3891 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
3892 {
|
|
|
3893 p = aff_entry->ae_add
|
|
|
3894 + STRLEN(aff_entry->ae_add);
|
|
|
3895 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
3896 if (PTR2CHAR(p) == c_up)
|
|
|
3897 {
|
|
|
3898 upper = TRUE;
|
|
|
3899 aff_entry->ae_chop = NULL;
|
|
|
3900 *p = NUL;
|
|
|
3901
|
|
|
3902 /* The condition is matched with the
|
|
|
3903 * actual word, thus must check for the
|
|
|
3904 * upper-case letter. */
|
|
|
3905 if (aff_entry->ae_cond != NULL)
|
|
|
3906 {
|
|
|
3907 char_u buf[MAXLINELEN];
|
|
|
3908 #ifdef FEAT_MBYTE
|
|
|
3909 if (has_mbyte)
|
|
|
3910 {
|
|
|
3911 onecap_copy(items[4], buf, TRUE);
|
|
|
3912 aff_entry->ae_cond = getroom_save(
|
|
474
|
3913 spin, buf);
|
|
455
|
3914 }
|
|
|
3915 else
|
|
|
3916 #endif
|
|
|
3917 *aff_entry->ae_cond = c_up;
|
|
|
3918 if (aff_entry->ae_cond != NULL)
|
|
|
3919 {
|
|
|
3920 sprintf((char *)buf, "^%s",
|
|
339
|
3921 aff_entry->ae_cond);
|
|
455
|
3922 vim_free(aff_entry->ae_prog);
|
|
|
3923 aff_entry->ae_prog = vim_regcomp(
|
|
|
3924 buf, RE_MAGIC + RE_STRING);
|
|
|
3925 }
|
|
|
3926 }
|
|
|
3927 }
|
|
|
3928 }
|
|
339
|
3929 }
|
|
|
3930
|
|
455
|
3931 if (aff_entry->ae_chop == NULL)
|
|
390
|
3932 {
|
|
455
|
3933 int idx;
|
|
|
3934 char_u **pp;
|
|
|
3935 int n;
|
|
|
3936
|
|
|
3937 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
3938 --idx)
|
|
|
3939 {
|
|
|
3940 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
3941 if (str_equal(p, aff_entry->ae_cond))
|
|
|
3942 break;
|
|
|
3943 }
|
|
|
3944 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
3945 {
|
|
|
3946 /* Not found, add a new condition. */
|
|
|
3947 idx = spin->si_prefcond.ga_len++;
|
|
|
3948 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
3949 + idx;
|
|
|
3950 if (aff_entry->ae_cond == NULL)
|
|
|
3951 *pp = NULL;
|
|
|
3952 else
|
|
474
|
3953 *pp = getroom_save(spin,
|
|
455
|
3954 aff_entry->ae_cond);
|
|
|
3955 }
|
|
|
3956
|
|
|
3957 /* Add the prefix to the prefix tree. */
|
|
|
3958 if (aff_entry->ae_add == NULL)
|
|
|
3959 p = (char_u *)"";
|
|
390
|
3960 else
|
|
455
|
3961 p = aff_entry->ae_add;
|
|
|
3962 /* PFX_FLAGS is a negative number, so that
|
|
|
3963 * tree_add_word() knows this is the prefix tree. */
|
|
|
3964 n = PFX_FLAGS;
|
|
|
3965 if (rare)
|
|
|
3966 n |= WFP_RARE;
|
|
|
3967 if (!cur_aff->ah_combine)
|
|
|
3968 n |= WFP_NC;
|
|
|
3969 if (upper)
|
|
|
3970 n |= WFP_UP;
|
|
474
|
3971 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
3972 idx, cur_aff->ah_newID);
|
|
390
|
3973 }
|
|
339
|
3974 }
|
|
240
|
3975 }
|
|
236
|
3976 }
|
|
255
|
3977 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
3978 {
|
|
|
3979 if (fol != NULL)
|
|
|
3980 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
3981 fname, lnum);
|
|
|
3982 else
|
|
|
3983 fol = vim_strsave(items[1]);
|
|
|
3984 }
|
|
|
3985 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
3986 {
|
|
|
3987 if (low != NULL)
|
|
|
3988 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
3989 fname, lnum);
|
|
|
3990 else
|
|
|
3991 low = vim_strsave(items[1]);
|
|
|
3992 }
|
|
|
3993 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
3994 {
|
|
|
3995 if (upp != NULL)
|
|
|
3996 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
3997 fname, lnum);
|
|
|
3998 else
|
|
|
3999 upp = vim_strsave(items[1]);
|
|
|
4000 }
|
|
236
|
4001 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
4002 {
|
|
236
|
4003 /* Ignore REP count */;
|
|
323
|
4004 if (!isdigit(*items[1]))
|
|
|
4005 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
4006 fname, lnum);
|
|
|
4007 }
|
|
482
|
4008 else if (STRCMP(items[0], "REP") == 0 && itemcnt >= 3)
|
|
236
|
4009 {
|
|
|
4010 /* REP item */
|
|
482
|
4011 /* Myspell ignores extra arguments, we require it starts with
|
|
|
4012 * # to detect mistakes. */
|
|
|
4013 if (itemcnt > 3 && items[3][0] != '#')
|
|
|
4014 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]);
|
|
323
|
4015 if (do_rep)
|
|
|
4016 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
4017 }
|
|
|
4018 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
4019 {
|
|
|
4020 /* MAP item or count */
|
|
|
4021 if (!found_map)
|
|
|
4022 {
|
|
|
4023 /* First line contains the count. */
|
|
|
4024 found_map = TRUE;
|
|
|
4025 if (!isdigit(*items[1]))
|
|
|
4026 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
4027 fname, lnum);
|
|
|
4028 }
|
|
|
4029 else if (do_map)
|
|
|
4030 {
|
|
346
|
4031 int c;
|
|
|
4032
|
|
|
4033 /* Check that every character appears only once. */
|
|
|
4034 for (p = items[1]; *p != NUL; )
|
|
|
4035 {
|
|
|
4036 #ifdef FEAT_MBYTE
|
|
|
4037 c = mb_ptr2char_adv(&p);
|
|
|
4038 #else
|
|
|
4039 c = *p++;
|
|
|
4040 #endif
|
|
|
4041 if ((spin->si_map.ga_len > 0
|
|
|
4042 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
4043 != NULL)
|
|
|
4044 || vim_strchr(p, c) != NULL)
|
|
|
4045 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
4046 fname, lnum);
|
|
|
4047 }
|
|
|
4048
|
|
323
|
4049 /* We simply concatenate all the MAP strings, separated by
|
|
|
4050 * slashes. */
|
|
|
4051 ga_concat(&spin->si_map, items[1]);
|
|
|
4052 ga_append(&spin->si_map, '/');
|
|
|
4053 }
|
|
|
4054 }
|
|
|
4055 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
4056 {
|
|
|
4057 if (do_sal)
|
|
|
4058 {
|
|
|
4059 /* SAL item (sounds-a-like)
|
|
|
4060 * Either one of the known keys or a from-to pair. */
|
|
|
4061 if (STRCMP(items[1], "followup") == 0)
|
|
|
4062 spin->si_followup = sal_to_bool(items[2]);
|
|
|
4063 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
4064 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
4065 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
4066 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
4067 else
|
|
|
4068 /* when "to" is "_" it means empty */
|
|
|
4069 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
4070 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
4071 : items[2]);
|
|
|
4072 }
|
|
236
|
4073 }
|
|
375
|
4074 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
4075 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
4076 {
|
|
|
4077 if (do_sofo)
|
|
|
4078 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
4079 }
|
|
|
4080 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
4081 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
4082 {
|
|
|
4083 if (do_sofo)
|
|
|
4084 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
4085 }
|
|
300
|
4086 else
|
|
481
|
4087 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
4088 fname, lnum, items[0]);
|
|
|
4089 }
|
|
|
4090 }
|
|
|
4091
|
|
375
|
4092 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
4093 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
4094 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
4095 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
4096 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
4097
|
|
255
|
4098 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
4099 {
|
|
351
|
4100 if (spin->si_clear_chartab)
|
|
|
4101 {
|
|
|
4102 /* Clear the char type tables, don't want to use any of the
|
|
|
4103 * currently used spell properties. */
|
|
|
4104 init_spell_chartab();
|
|
|
4105 spin->si_clear_chartab = FALSE;
|
|
|
4106 }
|
|
|
4107
|
|
316
|
4108 /*
|
|
|
4109 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
4110 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
4111 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
4112 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
4113 */
|
|
|
4114 if (!spin->si_ascii
|
|
|
4115 #ifdef FEAT_MBYTE
|
|
|
4116 && !enc_utf8
|
|
|
4117 #endif
|
|
|
4118 )
|
|
260
|
4119 {
|
|
|
4120 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
4121 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
4122 else
|
|
316
|
4123 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
4124 }
|
|
255
|
4125
|
|
|
4126 vim_free(fol);
|
|
|
4127 vim_free(low);
|
|
|
4128 vim_free(upp);
|
|
|
4129 }
|
|
|
4130
|
|
481
|
4131 /* Use compound specifications of the .aff file for the spell info. */
|
|
|
4132 if (aff->af_compminlen != 0)
|
|
|
4133 {
|
|
|
4134 if (spin->si_compminlen != 0
|
|
|
4135 && spin->si_compminlen != aff->af_compminlen)
|
|
|
4136 smsg((char_u *)_("COMPOUNDMIN value differs from what is used in another .aff file"));
|
|
|
4137 else
|
|
|
4138 spin->si_compminlen = aff->af_compminlen;
|
|
|
4139 }
|
|
|
4140
|
|
|
4141 if (aff->af_compflags != NULL)
|
|
|
4142 {
|
|
|
4143 if (spin->si_compflags != NULL
|
|
|
4144 && STRCMP(spin->si_compflags, aff->af_compflags) != 0)
|
|
|
4145 smsg((char_u *)_("COMPOUNDFLAG(S) value differs from what is used in another .aff file"));
|
|
|
4146 else
|
|
|
4147 spin->si_compflags = aff->af_compflags;
|
|
|
4148 }
|
|
|
4149
|
|
236
|
4150 vim_free(pc);
|
|
|
4151 fclose(fd);
|
|
|
4152 return aff;
|
|
|
4153 }
|
|
|
4154
|
|
|
4155 /*
|
|
339
|
4156 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
4157 * NULL as equal.
|
|
|
4158 */
|
|
|
4159 static int
|
|
|
4160 str_equal(s1, s2)
|
|
|
4161 char_u *s1;
|
|
|
4162 char_u *s2;
|
|
|
4163 {
|
|
|
4164 if (s1 == NULL || s2 == NULL)
|
|
|
4165 return s1 == s2;
|
|
|
4166 return STRCMP(s1, s2) == 0;
|
|
|
4167 }
|
|
|
4168
|
|
|
4169 /*
|
|
323
|
4170 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
4171 * They are stored case-folded.
|
|
|
4172 */
|
|
|
4173 static void
|
|
|
4174 add_fromto(spin, gap, from, to)
|
|
|
4175 spellinfo_T *spin;
|
|
|
4176 garray_T *gap;
|
|
|
4177 char_u *from;
|
|
|
4178 char_u *to;
|
|
|
4179 {
|
|
|
4180 fromto_T *ftp;
|
|
|
4181 char_u word[MAXWLEN];
|
|
|
4182
|
|
|
4183 if (ga_grow(gap, 1) == OK)
|
|
|
4184 {
|
|
|
4185 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
4186 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
4187 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
4188 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
4189 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
4190 ++gap->ga_len;
|
|
|
4191 }
|
|
|
4192 }
|
|
|
4193
|
|
|
4194 /*
|
|
|
4195 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
4196 */
|
|
|
4197 static int
|
|
|
4198 sal_to_bool(s)
|
|
|
4199 char_u *s;
|
|
|
4200 {
|
|
|
4201 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
4202 }
|
|
|
4203
|
|
|
4204 /*
|
|
240
|
4205 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
4206 * When "s" is NULL FALSE is returned.
|
|
|
4207 */
|
|
|
4208 static int
|
|
|
4209 has_non_ascii(s)
|
|
|
4210 char_u *s;
|
|
|
4211 {
|
|
|
4212 char_u *p;
|
|
|
4213
|
|
|
4214 if (s != NULL)
|
|
|
4215 for (p = s; *p != NUL; ++p)
|
|
|
4216 if (*p >= 128)
|
|
|
4217 return TRUE;
|
|
|
4218 return FALSE;
|
|
|
4219 }
|
|
|
4220
|
|
|
4221 /*
|
|
236
|
4222 * Free the structure filled by spell_read_aff().
|
|
|
4223 */
|
|
|
4224 static void
|
|
|
4225 spell_free_aff(aff)
|
|
|
4226 afffile_T *aff;
|
|
|
4227 {
|
|
|
4228 hashtab_T *ht;
|
|
|
4229 hashitem_T *hi;
|
|
|
4230 int todo;
|
|
|
4231 affheader_T *ah;
|
|
300
|
4232 affentry_T *ae;
|
|
236
|
4233
|
|
|
4234 vim_free(aff->af_enc);
|
|
|
4235
|
|
339
|
4236 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
4237 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
4238 {
|
|
|
4239 todo = ht->ht_used;
|
|
|
4240 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
4241 {
|
|
|
4242 if (!HASHITEM_EMPTY(hi))
|
|
|
4243 {
|
|
|
4244 --todo;
|
|
|
4245 ah = HI2AH(hi);
|
|
300
|
4246 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
4247 vim_free(ae->ae_prog);
|
|
236
|
4248 }
|
|
|
4249 }
|
|
|
4250 if (ht == &aff->af_suff)
|
|
|
4251 break;
|
|
|
4252 }
|
|
300
|
4253
|
|
236
|
4254 hash_clear(&aff->af_pref);
|
|
|
4255 hash_clear(&aff->af_suff);
|
|
|
4256 }
|
|
|
4257
|
|
|
4258 /*
|
|
300
|
4259 * Read dictionary file "fname".
|
|
236
|
4260 * Returns OK or FAIL;
|
|
|
4261 */
|
|
|
4262 static int
|
|
474
|
4263 spell_read_dic(spin, fname, affile)
|
|
|
4264 spellinfo_T *spin;
|
|
236
|
4265 char_u *fname;
|
|
300
|
4266 afffile_T *affile;
|
|
236
|
4267 {
|
|
300
|
4268 hashtab_T ht;
|
|
236
|
4269 char_u line[MAXLINELEN];
|
|
481
|
4270 char_u *p;
|
|
300
|
4271 char_u *afflist;
|
|
481
|
4272 char_u *store_afflist;
|
|
300
|
4273 char_u *dw;
|
|
236
|
4274 char_u *pc;
|
|
|
4275 char_u *w;
|
|
|
4276 int l;
|
|
|
4277 hash_T hash;
|
|
|
4278 hashitem_T *hi;
|
|
|
4279 FILE *fd;
|
|
|
4280 int lnum = 1;
|
|
300
|
4281 int non_ascii = 0;
|
|
|
4282 int retval = OK;
|
|
|
4283 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
4284 int flags;
|
|
481
|
4285 int duplicate = 0;
|
|
236
|
4286
|
|
300
|
4287 /*
|
|
|
4288 * Open the file.
|
|
|
4289 */
|
|
310
|
4290 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4291 if (fd == NULL)
|
|
|
4292 {
|
|
|
4293 EMSG2(_(e_notopen), fname);
|
|
|
4294 return FAIL;
|
|
|
4295 }
|
|
|
4296
|
|
300
|
4297 /* The hashtable is only used to detect duplicated words. */
|
|
|
4298 hash_init(&ht);
|
|
|
4299
|
|
310
|
4300 if (spin->si_verbose || p_verbose > 2)
|
|
|
4301 {
|
|
|
4302 if (!spin->si_verbose)
|
|
|
4303 verbose_enter();
|
|
366
|
4304 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
4305 out_flush();
|
|
|
4306 if (!spin->si_verbose)
|
|
|
4307 verbose_leave();
|
|
|
4308 }
|
|
236
|
4309
|
|
474
|
4310 /* start with a message for the first line */
|
|
|
4311 spin->si_msg_count = 999999;
|
|
|
4312
|
|
236
|
4313 /* Read and ignore the first line: word count. */
|
|
|
4314 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
4315 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
4316 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
4317
|
|
|
4318 /*
|
|
|
4319 * Read all the lines in the file one by one.
|
|
|
4320 * The words are converted to 'encoding' here, before being added to
|
|
|
4321 * the hashtable.
|
|
|
4322 */
|
|
255
|
4323 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
4324 {
|
|
255
|
4325 line_breakcheck();
|
|
236
|
4326 ++lnum;
|
|
455
|
4327 if (line[0] == '#' || line[0] == '/')
|
|
351
|
4328 continue; /* comment line */
|
|
236
|
4329
|
|
300
|
4330 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
4331 * the word is kept to allow e.g., "et al.". */
|
|
236
|
4332 l = STRLEN(line);
|
|
|
4333 while (l > 0 && line[l - 1] <= ' ')
|
|
|
4334 --l;
|
|
|
4335 if (l == 0)
|
|
|
4336 continue; /* empty line */
|
|
|
4337 line[l] = NUL;
|
|
|
4338
|
|
481
|
4339 /* Find the optional affix names. Replace the SLASH character by a
|
|
|
4340 * slash. */
|
|
|
4341 afflist = NULL;
|
|
|
4342 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
4343 {
|
|
|
4344 if (*p == affile->af_slash)
|
|
|
4345 *p = '/';
|
|
|
4346 else if (*p == '/')
|
|
|
4347 {
|
|
|
4348 *p = NUL;
|
|
|
4349 afflist = p + 1;
|
|
|
4350 break;
|
|
|
4351 }
|
|
|
4352 }
|
|
236
|
4353
|
|
300
|
4354 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4355 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4356 {
|
|
|
4357 ++non_ascii;
|
|
240
|
4358 continue;
|
|
300
|
4359 }
|
|
240
|
4360
|
|
310
|
4361 #ifdef FEAT_MBYTE
|
|
236
|
4362 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
4363 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4364 {
|
|
300
|
4365 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
4366 if (pc == NULL)
|
|
|
4367 {
|
|
|
4368 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4369 fname, lnum, line);
|
|
|
4370 continue;
|
|
|
4371 }
|
|
236
|
4372 w = pc;
|
|
|
4373 }
|
|
|
4374 else
|
|
310
|
4375 #endif
|
|
236
|
4376 {
|
|
|
4377 pc = NULL;
|
|
|
4378 w = line;
|
|
|
4379 }
|
|
|
4380
|
|
474
|
4381 /* This takes time, print a message every 10000 words. */
|
|
|
4382 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
4383 {
|
|
|
4384 spin->si_msg_count = 0;
|
|
339
|
4385 vim_snprintf((char *)message, sizeof(message),
|
|
|
4386 _("line %6d, word %6d - %s"),
|
|
|
4387 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
4388 msg_start();
|
|
|
4389 msg_puts_long_attr(message, 0);
|
|
|
4390 msg_clr_eos();
|
|
|
4391 msg_didout = FALSE;
|
|
|
4392 msg_col = 0;
|
|
|
4393 out_flush();
|
|
|
4394 }
|
|
|
4395
|
|
300
|
4396 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
4397 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
4398 if (dw == NULL)
|
|
300
|
4399 retval = FAIL;
|
|
|
4400 vim_free(pc);
|
|
|
4401 if (retval == FAIL)
|
|
236
|
4402 break;
|
|
|
4403
|
|
300
|
4404 hash = hash_hash(dw);
|
|
|
4405 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
4406 if (!HASHITEM_EMPTY(hi))
|
|
481
|
4407 {
|
|
|
4408 if (p_verbose > 0)
|
|
|
4409 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
4410 fname, lnum, dw);
|
|
481
|
4411 else if (duplicate == 0)
|
|
|
4412 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
4413 fname, lnum, dw);
|
|
|
4414 ++duplicate;
|
|
|
4415 }
|
|
236
|
4416 else
|
|
300
|
4417 hash_add_item(&ht, hi, dw, hash);
|
|
|
4418
|
|
307
|
4419 flags = 0;
|
|
481
|
4420 store_afflist = NULL;
|
|
307
|
4421 if (afflist != NULL)
|
|
|
4422 {
|
|
|
4423 /* Check for affix name that stands for keep-case word and stands
|
|
|
4424 * for rare word (if defined). */
|
|
310
|
4425 if (affile->af_kep != NUL
|
|
|
4426 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
389
|
4427 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
307
|
4428 if (affile->af_rar != NUL
|
|
|
4429 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
4430 flags |= WF_RARE;
|
|
346
|
4431 if (affile->af_bad != NUL
|
|
|
4432 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
4433 flags |= WF_BANNED;
|
|
339
|
4434
|
|
|
4435 if (affile->af_pfxpostpone)
|
|
|
4436 /* Need to store the list of prefix IDs with the word. */
|
|
481
|
4437 store_afflist = get_pfxlist(spin, affile, afflist);
|
|
482
|
4438
|
|
|
4439 if (spin->si_compflags)
|
|
|
4440 {
|
|
|
4441 /* Need to store the list of compound flags with the word. */
|
|
|
4442 p = get_compflags(spin, afflist);
|
|
|
4443 if (p != NULL)
|
|
|
4444 {
|
|
|
4445 if (store_afflist != NULL)
|
|
|
4446 {
|
|
|
4447 char_u *s;
|
|
|
4448
|
|
|
4449 /* Concatenate the prefix IDs with the compound flags.
|
|
|
4450 */
|
|
|
4451 s = getroom(spin, STRLEN(store_afflist)
|
|
|
4452 + STRLEN(p) + 1, FALSE);
|
|
|
4453 if (s != NULL)
|
|
|
4454 {
|
|
|
4455 STRCPY(s, store_afflist);
|
|
|
4456 STRCAT(s, p);
|
|
|
4457 store_afflist = s;
|
|
|
4458 }
|
|
|
4459 }
|
|
|
4460 else
|
|
|
4461 store_afflist = p;
|
|
|
4462 }
|
|
|
4463 }
|
|
307
|
4464 }
|
|
|
4465
|
|
300
|
4466 /* Add the word to the word tree(s). */
|
|
481
|
4467 if (store_word(spin, dw, flags, spin->si_region, store_afflist) == FAIL)
|
|
300
|
4468 retval = FAIL;
|
|
236
|
4469
|
|
300
|
4470 if (afflist != NULL)
|
|
|
4471 {
|
|
|
4472 /* Find all matching suffixes and add the resulting words.
|
|
|
4473 * Additionally do matching prefixes that combine. */
|
|
474
|
4474 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
4475 &affile->af_suff, &affile->af_pref,
|
|
481
|
4476 FALSE, flags, store_afflist) == FAIL)
|
|
300
|
4477 retval = FAIL;
|
|
|
4478
|
|
|
4479 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
4480 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
4481 &affile->af_pref, NULL,
|
|
481
|
4482 FALSE, flags, store_afflist) == FAIL)
|
|
300
|
4483 retval = FAIL;
|
|
|
4484 }
|
|
236
|
4485 }
|
|
|
4486
|
|
481
|
4487 if (duplicate > 0)
|
|
|
4488 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
4489 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
4490 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
4491 non_ascii, fname);
|
|
300
|
4492 hash_clear(&ht);
|
|
|
4493
|
|
236
|
4494 fclose(fd);
|
|
300
|
4495 return retval;
|
|
236
|
4496 }
|
|
|
4497
|
|
|
4498 /*
|
|
339
|
4499 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
4500 * Used for PFXPOSTPONE.
|
|
|
4501 * Returns a string allocated with getroom(). NULL when there are no prefixes
|
|
|
4502 * or when out of memory.
|
|
|
4503 */
|
|
|
4504 static char_u *
|
|
474
|
4505 get_pfxlist(spin, affile, afflist)
|
|
|
4506 spellinfo_T *spin;
|
|
339
|
4507 afffile_T *affile;
|
|
|
4508 char_u *afflist;
|
|
|
4509 {
|
|
|
4510 char_u *p;
|
|
|
4511 int cnt;
|
|
|
4512 int round;
|
|
|
4513 char_u *res = NULL;
|
|
|
4514 char_u key[2];
|
|
|
4515 hashitem_T *hi;
|
|
|
4516
|
|
|
4517 key[1] = NUL;
|
|
|
4518
|
|
|
4519 /* round 1: count the number of prefix IDs.
|
|
|
4520 * round 2: move prefix IDs to "res" */
|
|
|
4521 for (round = 1; round <= 2; ++round)
|
|
|
4522 {
|
|
|
4523 cnt = 0;
|
|
|
4524 for (p = afflist; *p != NUL; ++p)
|
|
|
4525 {
|
|
|
4526 key[0] = *p;
|
|
|
4527 hi = hash_find(&affile->af_pref, key);
|
|
|
4528 if (!HASHITEM_EMPTY(hi))
|
|
|
4529 {
|
|
|
4530 /* This is a prefix ID, use the new number. */
|
|
|
4531 if (round == 2)
|
|
|
4532 res[cnt] = HI2AH(hi)->ah_newID;
|
|
|
4533 ++cnt;
|
|
|
4534 }
|
|
|
4535 }
|
|
|
4536 if (round == 1 && cnt > 0)
|
|
474
|
4537 res = getroom(spin, cnt + 1, FALSE);
|
|
339
|
4538 if (res == NULL)
|
|
|
4539 break;
|
|
|
4540 }
|
|
|
4541
|
|
|
4542 if (res != NULL)
|
|
|
4543 res[cnt] = NUL;
|
|
|
4544 return res;
|
|
|
4545 }
|
|
|
4546
|
|
|
4547 /*
|
|
481
|
4548 * Get the list of affix IDs from the affix list "afflist" that are used for
|
|
|
4549 * compound words.
|
|
|
4550 * Returns a string allocated with getroom(). NULL when there are no relevant
|
|
|
4551 * affixes or when out of memory.
|
|
|
4552 */
|
|
|
4553 static char_u *
|
|
|
4554 get_compflags(spin, afflist)
|
|
|
4555 spellinfo_T *spin;
|
|
|
4556 char_u *afflist;
|
|
|
4557 {
|
|
|
4558 char_u *p;
|
|
|
4559 int cnt;
|
|
|
4560 int round;
|
|
|
4561 char_u *res = NULL;
|
|
|
4562
|
|
|
4563 /* round 1: count the number of affix IDs.
|
|
|
4564 * round 2: move affix IDs to "res" */
|
|
|
4565 for (round = 1; round <= 2; ++round)
|
|
|
4566 {
|
|
|
4567 cnt = 0;
|
|
|
4568 for (p = afflist; *p != NUL; ++p)
|
|
|
4569 {
|
|
|
4570 if (*p != ',' && *p != '-'
|
|
|
4571 && vim_strchr(spin->si_compflags, *p) != NULL)
|
|
|
4572 {
|
|
|
4573 /* This is a compount affix ID. */
|
|
|
4574 if (round == 2)
|
|
|
4575 res[cnt] = *p;
|
|
|
4576 ++cnt;
|
|
|
4577 }
|
|
|
4578 }
|
|
|
4579 if (round == 1 && cnt > 0)
|
|
|
4580 res = getroom(spin, cnt + 1, FALSE);
|
|
|
4581 if (res == NULL)
|
|
|
4582 break;
|
|
|
4583 }
|
|
|
4584
|
|
|
4585 if (res != NULL)
|
|
|
4586 res[cnt] = NUL;
|
|
|
4587 return res;
|
|
|
4588 }
|
|
|
4589
|
|
|
4590 /*
|
|
300
|
4591 * Apply affixes to a word and store the resulting words.
|
|
|
4592 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
4593 * prefixes or suffixes.
|
|
|
4594 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
4595 * the resulting words for combining affixes.
|
|
|
4596 *
|
|
|
4597 * Returns FAIL when out of memory.
|
|
236
|
4598 */
|
|
300
|
4599 static int
|
|
474
|
4600 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags, pfxlist)
|
|
|
4601 spellinfo_T *spin; /* spell info */
|
|
300
|
4602 char_u *word; /* basic word start */
|
|
|
4603 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
4604 afffile_T *affile;
|
|
300
|
4605 hashtab_T *ht;
|
|
|
4606 hashtab_T *xht;
|
|
|
4607 int comb; /* only use affixes that combine */
|
|
307
|
4608 int flags; /* flags for the word */
|
|
339
|
4609 char_u *pfxlist; /* list of prefix IDs */
|
|
236
|
4610 {
|
|
|
4611 int todo;
|
|
|
4612 hashitem_T *hi;
|
|
300
|
4613 affheader_T *ah;
|
|
|
4614 affentry_T *ae;
|
|
|
4615 regmatch_T regmatch;
|
|
|
4616 char_u newword[MAXWLEN];
|
|
|
4617 int retval = OK;
|
|
|
4618 int i;
|
|
|
4619 char_u *p;
|
|
366
|
4620 int use_flags;
|
|
390
|
4621 char_u *use_pfxlist;
|
|
455
|
4622 int c;
|
|
481
|
4623 int wordlen = STRLEN(word);
|
|
236
|
4624
|
|
300
|
4625 todo = ht->ht_used;
|
|
|
4626 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
4627 {
|
|
|
4628 if (!HASHITEM_EMPTY(hi))
|
|
|
4629 {
|
|
|
4630 --todo;
|
|
300
|
4631 ah = HI2AH(hi);
|
|
236
|
4632
|
|
300
|
4633 /* Check that the affix combines, if required, and that the word
|
|
|
4634 * supports this affix. */
|
|
455
|
4635 c = PTR2CHAR(ah->ah_key);
|
|
|
4636 if ((!comb || ah->ah_combine) && vim_strchr(afflist, c) != NULL)
|
|
236
|
4637 {
|
|
300
|
4638 /* Loop over all affix entries with this name. */
|
|
|
4639 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
4640 {
|
|
300
|
4641 /* Check the condition. It's not logical to match case
|
|
|
4642 * here, but it is required for compatibility with
|
|
339
|
4643 * Myspell.
|
|
481
|
4644 * Another requirement from Myspell is that the chop
|
|
|
4645 * string is shorter than the word itself.
|
|
339
|
4646 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
4647 * prefixes with a chop string. */
|
|
300
|
4648 regmatch.regprog = ae->ae_prog;
|
|
|
4649 regmatch.rm_ic = FALSE;
|
|
339
|
4650 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
4651 || ae->ae_chop != NULL)
|
|
481
|
4652 && (ae->ae_chop == NULL
|
|
|
4653 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
4654 && (ae->ae_prog == NULL
|
|
|
4655 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
4656 {
|
|
|
4657 /* Match. Remove the chop and add the affix. */
|
|
|
4658 if (xht == NULL)
|
|
240
|
4659 {
|
|
300
|
4660 /* prefix: chop/add at the start of the word */
|
|
|
4661 if (ae->ae_add == NULL)
|
|
|
4662 *newword = NUL;
|
|
|
4663 else
|
|
|
4664 STRCPY(newword, ae->ae_add);
|
|
|
4665 p = word;
|
|
|
4666 if (ae->ae_chop != NULL)
|
|
310
|
4667 {
|
|
300
|
4668 /* Skip chop string. */
|
|
310
|
4669 #ifdef FEAT_MBYTE
|
|
|
4670 if (has_mbyte)
|
|
324
|
4671 {
|
|
310
|
4672 i = mb_charlen(ae->ae_chop);
|
|
324
|
4673 for ( ; i > 0; --i)
|
|
|
4674 mb_ptr_adv(p);
|
|
|
4675 }
|
|
310
|
4676 else
|
|
|
4677 #endif
|
|
324
|
4678 p += STRLEN(ae->ae_chop);
|
|
310
|
4679 }
|
|
300
|
4680 STRCAT(newword, p);
|
|
|
4681 }
|
|
|
4682 else
|
|
|
4683 {
|
|
|
4684 /* suffix: chop/add at the end of the word */
|
|
|
4685 STRCPY(newword, word);
|
|
|
4686 if (ae->ae_chop != NULL)
|
|
|
4687 {
|
|
|
4688 /* Remove chop string. */
|
|
|
4689 p = newword + STRLEN(newword);
|
|
376
|
4690 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
4691 for ( ; i > 0; --i)
|
|
300
|
4692 mb_ptr_back(newword, p);
|
|
|
4693 *p = NUL;
|
|
|
4694 }
|
|
|
4695 if (ae->ae_add != NULL)
|
|
|
4696 STRCAT(newword, ae->ae_add);
|
|
240
|
4697 }
|
|
|
4698
|
|
366
|
4699 /* Obey the "rare" flag of the affix. */
|
|
|
4700 if (ae->ae_rare)
|
|
|
4701 use_flags = flags | WF_RARE;
|
|
|
4702 else
|
|
|
4703 use_flags = flags;
|
|
390
|
4704 use_pfxlist = pfxlist;
|
|
|
4705
|
|
|
4706 /* When there are postponed prefixes... */
|
|
393
|
4707 if (spin->si_prefroot != NULL
|
|
|
4708 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
4709 {
|
|
|
4710 /* ... add a flag to indicate an affix was used. */
|
|
|
4711 use_flags |= WF_HAS_AFF;
|
|
|
4712
|
|
|
4713 /* ... don't use a prefix list if combining
|
|
|
4714 * affixes is not allowed */
|
|
|
4715 if (!ah->ah_combine || comb)
|
|
|
4716 use_pfxlist = NULL;
|
|
|
4717 }
|
|
366
|
4718
|
|
300
|
4719 /* Store the modified word. */
|
|
474
|
4720 if (store_word(spin, newword, use_flags,
|
|
390
|
4721 spin->si_region, use_pfxlist) == FAIL)
|
|
300
|
4722 retval = FAIL;
|
|
236
|
4723
|
|
300
|
4724 /* When added a suffix and combining is allowed also
|
|
|
4725 * try adding prefixes additionally. */
|
|
|
4726 if (xht != NULL && ah->ah_combine)
|
|
474
|
4727 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
4728 xht, NULL, TRUE,
|
|
|
4729 use_flags, use_pfxlist) == FAIL)
|
|
300
|
4730 retval = FAIL;
|
|
236
|
4731 }
|
|
|
4732 }
|
|
|
4733 }
|
|
|
4734 }
|
|
|
4735 }
|
|
|
4736
|
|
|
4737 return retval;
|
|
|
4738 }
|
|
|
4739
|
|
|
4740 /*
|
|
300
|
4741 * Read a file with a list of words.
|
|
236
|
4742 */
|
|
|
4743 static int
|
|
474
|
4744 spell_read_wordfile(spin, fname)
|
|
|
4745 spellinfo_T *spin;
|
|
300
|
4746 char_u *fname;
|
|
236
|
4747 {
|
|
300
|
4748 FILE *fd;
|
|
|
4749 long lnum = 0;
|
|
|
4750 char_u rline[MAXLINELEN];
|
|
|
4751 char_u *line;
|
|
|
4752 char_u *pc = NULL;
|
|
381
|
4753 char_u *p;
|
|
300
|
4754 int l;
|
|
|
4755 int retval = OK;
|
|
|
4756 int did_word = FALSE;
|
|
|
4757 int non_ascii = 0;
|
|
307
|
4758 int flags;
|
|
316
|
4759 int regionmask;
|
|
236
|
4760
|
|
300
|
4761 /*
|
|
|
4762 * Open the file.
|
|
|
4763 */
|
|
310
|
4764 fd = mch_fopen((char *)fname, "r");
|
|
300
|
4765 if (fd == NULL)
|
|
236
|
4766 {
|
|
300
|
4767 EMSG2(_(e_notopen), fname);
|
|
|
4768 return FAIL;
|
|
236
|
4769 }
|
|
|
4770
|
|
310
|
4771 if (spin->si_verbose || p_verbose > 2)
|
|
|
4772 {
|
|
|
4773 if (!spin->si_verbose)
|
|
|
4774 verbose_enter();
|
|
366
|
4775 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
4776 out_flush();
|
|
|
4777 if (!spin->si_verbose)
|
|
|
4778 verbose_leave();
|
|
|
4779 }
|
|
300
|
4780
|
|
|
4781 /*
|
|
|
4782 * Read all the lines in the file one by one.
|
|
|
4783 */
|
|
|
4784 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
4785 {
|
|
|
4786 line_breakcheck();
|
|
|
4787 ++lnum;
|
|
|
4788
|
|
|
4789 /* Skip comment lines. */
|
|
|
4790 if (*rline == '#')
|
|
|
4791 continue;
|
|
|
4792
|
|
|
4793 /* Remove CR, LF and white space from the end. */
|
|
|
4794 l = STRLEN(rline);
|
|
|
4795 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
4796 --l;
|
|
|
4797 if (l == 0)
|
|
|
4798 continue; /* empty or blank line */
|
|
|
4799 rline[l] = NUL;
|
|
|
4800
|
|
|
4801 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
4802 vim_free(pc);
|
|
310
|
4803 #ifdef FEAT_MBYTE
|
|
300
|
4804 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
4805 {
|
|
|
4806 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
4807 if (pc == NULL)
|
|
|
4808 {
|
|
|
4809 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4810 fname, lnum, rline);
|
|
|
4811 continue;
|
|
|
4812 }
|
|
|
4813 line = pc;
|
|
|
4814 }
|
|
|
4815 else
|
|
310
|
4816 #endif
|
|
300
|
4817 {
|
|
|
4818 pc = NULL;
|
|
|
4819 line = rline;
|
|
|
4820 }
|
|
|
4821
|
|
307
|
4822 if (*line == '/')
|
|
300
|
4823 {
|
|
307
|
4824 ++line;
|
|
|
4825 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
4826 {
|
|
|
4827 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
4828 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
4829 fname, lnum, line - 1);
|
|
300
|
4830 else if (did_word)
|
|
316
|
4831 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
4832 fname, lnum, line - 1);
|
|
300
|
4833 else
|
|
|
4834 {
|
|
310
|
4835 #ifdef FEAT_MBYTE
|
|
|
4836 char_u *enc;
|
|
|
4837
|
|
300
|
4838 /* Setup for conversion to 'encoding'. */
|
|
316
|
4839 line += 10;
|
|
|
4840 enc = enc_canonize(line);
|
|
300
|
4841 if (enc != NULL && !spin->si_ascii
|
|
|
4842 && convert_setup(&spin->si_conv, enc,
|
|
|
4843 p_enc) == FAIL)
|
|
|
4844 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
4845 fname, line, p_enc);
|
|
300
|
4846 vim_free(enc);
|
|
375
|
4847 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4848 #else
|
|
|
4849 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4850 #endif
|
|
300
|
4851 }
|
|
307
|
4852 continue;
|
|
300
|
4853 }
|
|
307
|
4854
|
|
316
|
4855 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
4856 {
|
|
|
4857 if (spin->si_region_count > 1)
|
|
|
4858 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
4859 fname, lnum, line);
|
|
|
4860 else
|
|
|
4861 {
|
|
|
4862 line += 8;
|
|
|
4863 if (STRLEN(line) > 16)
|
|
|
4864 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
4865 fname, lnum, line);
|
|
|
4866 else
|
|
|
4867 {
|
|
|
4868 spin->si_region_count = STRLEN(line) / 2;
|
|
|
4869 STRCPY(spin->si_region_name, line);
|
|
389
|
4870
|
|
|
4871 /* Adjust the mask for a word valid in all regions. */
|
|
|
4872 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
4873 }
|
|
|
4874 }
|
|
|
4875 continue;
|
|
|
4876 }
|
|
|
4877
|
|
381
|
4878 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
4879 fname, lnum, line - 1);
|
|
|
4880 continue;
|
|
|
4881 }
|
|
|
4882
|
|
|
4883 flags = 0;
|
|
|
4884 regionmask = spin->si_region;
|
|
|
4885
|
|
|
4886 /* Check for flags and region after a slash. */
|
|
|
4887 p = vim_strchr(line, '/');
|
|
|
4888 if (p != NULL)
|
|
|
4889 {
|
|
|
4890 *p++ = NUL;
|
|
|
4891 while (*p != NUL)
|
|
|
4892 {
|
|
|
4893 if (*p == '=') /* keep-case word */
|
|
389
|
4894 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
4895 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
4896 flags |= WF_BANNED;
|
|
|
4897 else if (*p == '?') /* Rare word. */
|
|
|
4898 flags |= WF_RARE;
|
|
|
4899 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
4900 {
|
|
381
|
4901 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
4902 regionmask = 0;
|
|
|
4903 flags |= WF_REGION;
|
|
|
4904
|
|
|
4905 l = *p - '0';
|
|
316
|
4906 if (l > spin->si_region_count)
|
|
|
4907 {
|
|
|
4908 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
4909 fname, lnum, p);
|
|
316
|
4910 break;
|
|
|
4911 }
|
|
|
4912 regionmask |= 1 << (l - 1);
|
|
|
4913 }
|
|
381
|
4914 else
|
|
|
4915 {
|
|
|
4916 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
4917 fname, lnum, p);
|
|
|
4918 break;
|
|
|
4919 }
|
|
|
4920 ++p;
|
|
307
|
4921 }
|
|
300
|
4922 }
|
|
|
4923
|
|
|
4924 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4925 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4926 {
|
|
|
4927 ++non_ascii;
|
|
|
4928 continue;
|
|
|
4929 }
|
|
|
4930
|
|
|
4931 /* Normal word: store it. */
|
|
474
|
4932 if (store_word(spin, line, flags, regionmask, NULL) == FAIL)
|
|
300
|
4933 {
|
|
|
4934 retval = FAIL;
|
|
|
4935 break;
|
|
|
4936 }
|
|
|
4937 did_word = TRUE;
|
|
|
4938 }
|
|
|
4939
|
|
|
4940 vim_free(pc);
|
|
|
4941 fclose(fd);
|
|
|
4942
|
|
310
|
4943 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
4944 {
|
|
|
4945 if (p_verbose > 2)
|
|
|
4946 verbose_enter();
|
|
300
|
4947 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4948 non_ascii);
|
|
310
|
4949 if (p_verbose > 2)
|
|
|
4950 verbose_leave();
|
|
|
4951 }
|
|
300
|
4952 return retval;
|
|
236
|
4953 }
|
|
|
4954
|
|
|
4955 /*
|
|
300
|
4956 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
4957 * This avoids calling free() for every little struct we use (and keeping
|
|
|
4958 * track of them).
|
|
300
|
4959 * The memory is cleared to all zeros.
|
|
|
4960 * Returns NULL when out of memory.
|
|
|
4961 */
|
|
|
4962 static void *
|
|
474
|
4963 getroom(spin, len, align)
|
|
|
4964 spellinfo_T *spin;
|
|
447
|
4965 size_t len; /* length needed */
|
|
|
4966 int align; /* align for pointer */
|
|
300
|
4967 {
|
|
|
4968 char_u *p;
|
|
474
|
4969 sblock_T *bl = spin->si_blocks;
|
|
300
|
4970
|
|
447
|
4971 if (align && bl != NULL)
|
|
|
4972 /* Round size up for alignment. On some systems structures need to be
|
|
|
4973 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
4974 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
4975 & ~(sizeof(char *) - 1);
|
|
|
4976
|
|
300
|
4977 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
4978 {
|
|
|
4979 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
4980 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
4981 if (bl == NULL)
|
|
|
4982 return NULL;
|
|
474
|
4983 bl->sb_next = spin->si_blocks;
|
|
|
4984 spin->si_blocks = bl;
|
|
300
|
4985 bl->sb_used = 0;
|
|
482
|
4986 ++spin->si_blocks_cnt;
|
|
300
|
4987 }
|
|
|
4988
|
|
|
4989 p = bl->sb_data + bl->sb_used;
|
|
|
4990 bl->sb_used += len;
|
|
|
4991
|
|
|
4992 return p;
|
|
|
4993 }
|
|
|
4994
|
|
|
4995 /*
|
|
|
4996 * Make a copy of a string into memory allocated with getroom().
|
|
|
4997 */
|
|
|
4998 static char_u *
|
|
474
|
4999 getroom_save(spin, s)
|
|
|
5000 spellinfo_T *spin;
|
|
300
|
5001 char_u *s;
|
|
|
5002 {
|
|
|
5003 char_u *sc;
|
|
|
5004
|
|
474
|
5005 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
5006 if (sc != NULL)
|
|
|
5007 STRCPY(sc, s);
|
|
|
5008 return sc;
|
|
|
5009 }
|
|
|
5010
|
|
|
5011
|
|
|
5012 /*
|
|
|
5013 * Free the list of allocated sblock_T.
|
|
236
|
5014 */
|
|
|
5015 static void
|
|
300
|
5016 free_blocks(bl)
|
|
|
5017 sblock_T *bl;
|
|
236
|
5018 {
|
|
300
|
5019 sblock_T *next;
|
|
236
|
5020
|
|
300
|
5021 while (bl != NULL)
|
|
236
|
5022 {
|
|
300
|
5023 next = bl->sb_next;
|
|
|
5024 vim_free(bl);
|
|
|
5025 bl = next;
|
|
236
|
5026 }
|
|
|
5027 }
|
|
|
5028
|
|
|
5029 /*
|
|
300
|
5030 * Allocate the root of a word tree.
|
|
236
|
5031 */
|
|
300
|
5032 static wordnode_T *
|
|
474
|
5033 wordtree_alloc(spin)
|
|
|
5034 spellinfo_T *spin;
|
|
|
5035 {
|
|
|
5036 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
5037 }
|
|
|
5038
|
|
|
5039 /*
|
|
300
|
5040 * Store a word in the tree(s).
|
|
390
|
5041 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
5042 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
5043 * used to find suggestions.
|
|
300
|
5044 * For a keep-case word also store it in the keep-case tree.
|
|
482
|
5045 * When "pfxlist" is not NULL store the word for each postponed prefix ID and
|
|
|
5046 * compound flag.
|
|
236
|
5047 */
|
|
|
5048 static int
|
|
474
|
5049 store_word(spin, word, flags, region, pfxlist)
|
|
|
5050 spellinfo_T *spin;
|
|
300
|
5051 char_u *word;
|
|
307
|
5052 int flags; /* extra flags, WF_BANNED */
|
|
316
|
5053 int region; /* supported region(s) */
|
|
339
|
5054 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
236
|
5055 {
|
|
300
|
5056 int len = STRLEN(word);
|
|
|
5057 int ct = captype(word, word + len);
|
|
|
5058 char_u foldword[MAXWLEN];
|
|
339
|
5059 int res = OK;
|
|
|
5060 char_u *p;
|
|
236
|
5061
|
|
323
|
5062 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
5063 for (p = pfxlist; res == OK; ++p)
|
|
|
5064 {
|
|
474
|
5065 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
|
5066 region, p == NULL ? 0 : *p);
|
|
339
|
5067 if (p == NULL || *p == NUL)
|
|
|
5068 break;
|
|
|
5069 }
|
|
334
|
5070 ++spin->si_foldwcount;
|
|
307
|
5071
|
|
474
|
5072 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
5073 {
|
|
339
|
5074 for (p = pfxlist; res == OK; ++p)
|
|
|
5075 {
|
|
474
|
5076 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
|
5077 region, p == NULL ? 0 : *p);
|
|
339
|
5078 if (p == NULL || *p == NUL)
|
|
|
5079 break;
|
|
|
5080 }
|
|
334
|
5081 ++spin->si_keepwcount;
|
|
|
5082 }
|
|
300
|
5083 return res;
|
|
236
|
5084 }
|
|
|
5085
|
|
|
5086 /*
|
|
300
|
5087 * Add word "word" to a word tree at "root".
|
|
366
|
5088 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
5089 * "rare" and "region" is the condition nr.
|
|
255
|
5090 * Returns FAIL when out of memory.
|
|
236
|
5091 */
|
|
255
|
5092 static int
|
|
481
|
5093 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
5094 spellinfo_T *spin;
|
|
300
|
5095 char_u *word;
|
|
|
5096 wordnode_T *root;
|
|
|
5097 int flags;
|
|
|
5098 int region;
|
|
481
|
5099 int affixID;
|
|
236
|
5100 {
|
|
300
|
5101 wordnode_T *node = root;
|
|
|
5102 wordnode_T *np;
|
|
469
|
5103 wordnode_T *copyp, **copyprev;
|
|
300
|
5104 wordnode_T **prev = NULL;
|
|
|
5105 int i;
|
|
255
|
5106
|
|
300
|
5107 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
5108 for (i = 0; ; ++i)
|
|
255
|
5109 {
|
|
469
|
5110 /* When there is more than one reference to this node we need to make
|
|
|
5111 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
5112 * (we don't optimize for a partly shared list of siblings). */
|
|
|
5113 if (node != NULL && node->wn_refs > 1)
|
|
|
5114 {
|
|
|
5115 --node->wn_refs;
|
|
|
5116 copyprev = prev;
|
|
|
5117 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
5118 {
|
|
|
5119 /* Allocate a new node and copy the info. */
|
|
474
|
5120 np = get_wordnode(spin);
|
|
469
|
5121 if (np == NULL)
|
|
|
5122 return FAIL;
|
|
|
5123 np->wn_child = copyp->wn_child;
|
|
|
5124 if (np->wn_child != NULL)
|
|
|
5125 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
5126 np->wn_byte = copyp->wn_byte;
|
|
|
5127 if (np->wn_byte == NUL)
|
|
|
5128 {
|
|
|
5129 np->wn_flags = copyp->wn_flags;
|
|
|
5130 np->wn_region = copyp->wn_region;
|
|
481
|
5131 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
5132 }
|
|
|
5133
|
|
|
5134 /* Link the new node in the list, there will be one ref. */
|
|
|
5135 np->wn_refs = 1;
|
|
|
5136 *copyprev = np;
|
|
|
5137 copyprev = &np->wn_sibling;
|
|
|
5138
|
|
|
5139 /* Let "node" point to the head of the copied list. */
|
|
|
5140 if (copyp == node)
|
|
|
5141 node = np;
|
|
|
5142 }
|
|
|
5143 }
|
|
|
5144
|
|
300
|
5145 /* Look for the sibling that has the same character. They are sorted
|
|
|
5146 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
5147 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
5148 * done on flags and then on affixID. */
|
|
339
|
5149 while (node != NULL
|
|
|
5150 && (node->wn_byte < word[i]
|
|
|
5151 || (node->wn_byte == NUL
|
|
|
5152 && (flags < 0
|
|
481
|
5153 ? node->wn_affixID < affixID
|
|
390
|
5154 : node->wn_flags < (flags & WN_MASK)
|
|
|
5155 || (node->wn_flags == (flags & WN_MASK)
|
|
481
|
5156 && node->wn_affixID < affixID)))))
|
|
236
|
5157 {
|
|
300
|
5158 prev = &node->wn_sibling;
|
|
|
5159 node = *prev;
|
|
236
|
5160 }
|
|
339
|
5161 if (node == NULL
|
|
|
5162 || node->wn_byte != word[i]
|
|
|
5163 || (word[i] == NUL
|
|
|
5164 && (flags < 0
|
|
390
|
5165 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
5166 || node->wn_affixID != affixID)))
|
|
255
|
5167 {
|
|
300
|
5168 /* Allocate a new node. */
|
|
474
|
5169 np = get_wordnode(spin);
|
|
300
|
5170 if (np == NULL)
|
|
|
5171 return FAIL;
|
|
|
5172 np->wn_byte = word[i];
|
|
469
|
5173
|
|
|
5174 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
5175 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
5176 * make ref count of sibling one (matters when inserting in front
|
|
|
5177 * of the list of siblings). */
|
|
|
5178 if (node == NULL)
|
|
|
5179 np->wn_refs = 1;
|
|
|
5180 else
|
|
|
5181 {
|
|
|
5182 np->wn_refs = node->wn_refs;
|
|
|
5183 node->wn_refs = 1;
|
|
|
5184 }
|
|
300
|
5185 *prev = np;
|
|
|
5186 np->wn_sibling = node;
|
|
|
5187 node = np;
|
|
255
|
5188 }
|
|
300
|
5189
|
|
|
5190 if (word[i] == NUL)
|
|
|
5191 {
|
|
|
5192 node->wn_flags = flags;
|
|
|
5193 node->wn_region |= region;
|
|
481
|
5194 node->wn_affixID = affixID;
|
|
300
|
5195 break;
|
|
|
5196 }
|
|
|
5197 prev = &node->wn_child;
|
|
|
5198 node = *prev;
|
|
255
|
5199 }
|
|
469
|
5200 #ifdef SPELL_PRINTTREE
|
|
|
5201 smsg("Added \"%s\"", word);
|
|
|
5202 spell_print_tree(root->wn_sibling);
|
|
|
5203 #endif
|
|
|
5204
|
|
474
|
5205 /* count nr of words added since last message */
|
|
|
5206 ++spin->si_msg_count;
|
|
|
5207
|
|
482
|
5208 if (spin->si_compress_cnt > 1)
|
|
|
5209 {
|
|
|
5210 if (--spin->si_compress_cnt == 1)
|
|
|
5211 /* Did enough words to lower the block count limit. */
|
|
|
5212 spin->si_blocks_cnt += SPELL_COMPRESS_INC;
|
|
|
5213 }
|
|
|
5214
|
|
469
|
5215 /*
|
|
482
|
5216 * When we have allocated lots of memory we need to compress the word tree
|
|
|
5217 * to free up some room. But compression is slow, and we might actually
|
|
|
5218 * need that room, thus only compress in the following situations:
|
|
|
5219 * 1. When not compressed before (si_compress_cnt == 0): when using
|
|
|
5220 * SPELL_COMPRESS_CNT blocks.
|
|
|
5221 * 2. When compressed before and used SPELL_COMPRESS_INC blocks before
|
|
|
5222 * adding SPELL_COMPRESS_ADDED words (si_compress_cnt > 1).
|
|
|
5223 * 3. When compressed before, added SPELL_COMPRESS_ADDED words
|
|
|
5224 * (si_compress_cnt == 1) and the number of free nodes drops below the
|
|
|
5225 * maximum word length.
|
|
469
|
5226 */
|
|
482
|
5227 #ifndef SPELL_PRINTTREE
|
|
|
5228 if (spin->si_compress_cnt == 1
|
|
|
5229 ? spin->si_free_count < MAXWLEN
|
|
|
5230 : spin->si_blocks_cnt >= SPELL_COMPRESS_CNT)
|
|
|
5231 #endif
|
|
|
5232 {
|
|
|
5233 /* Decrement the block counter. The effect is that we compress again
|
|
|
5234 * when the freed up room has been used and another SPELL_COMPRESS_INC
|
|
|
5235 * blocks have been allocated. Unless SPELL_COMPRESS_ADDED words have
|
|
|
5236 * been added, then the limit is put back again. */
|
|
|
5237 spin->si_blocks_cnt -= SPELL_COMPRESS_INC;
|
|
|
5238 spin->si_compress_cnt = SPELL_COMPRESS_ADDED;
|
|
|
5239
|
|
474
|
5240 if (spin->si_verbose)
|
|
|
5241 {
|
|
|
5242 msg_start();
|
|
|
5243 msg_puts((char_u *)_(msg_compressing));
|
|
|
5244 msg_clr_eos();
|
|
|
5245 msg_didout = FALSE;
|
|
|
5246 msg_col = 0;
|
|
|
5247 out_flush();
|
|
|
5248 }
|
|
482
|
5249
|
|
|
5250 /* Compress both trees. Either they both have many nodes, which makes
|
|
|
5251 * compression useful, or one of them is small, which means
|
|
|
5252 * compression goes fast. */
|
|
|
5253 wordtree_compress(spin, spin->si_foldroot);
|
|
|
5254 wordtree_compress(spin, spin->si_keeproot);
|
|
469
|
5255 }
|
|
255
|
5256
|
|
|
5257 return OK;
|
|
236
|
5258 }
|
|
|
5259
|
|
469
|
5260 /*
|
|
|
5261 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
5262 * allocate a new one.
|
|
|
5263 */
|
|
|
5264 static wordnode_T *
|
|
474
|
5265 get_wordnode(spin)
|
|
|
5266 spellinfo_T *spin;
|
|
469
|
5267 {
|
|
|
5268 wordnode_T *n;
|
|
|
5269
|
|
474
|
5270 if (spin->si_first_free == NULL)
|
|
|
5271 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
5272 else
|
|
|
5273 {
|
|
474
|
5274 n = spin->si_first_free;
|
|
|
5275 spin->si_first_free = n->wn_child;
|
|
469
|
5276 vim_memset(n, 0, sizeof(wordnode_T));
|
|
482
|
5277 --spin->si_free_count;
|
|
469
|
5278 }
|
|
|
5279 #ifdef SPELL_PRINTTREE
|
|
474
|
5280 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
5281 #endif
|
|
|
5282 return n;
|
|
|
5283 }
|
|
|
5284
|
|
|
5285 /*
|
|
474
|
5286 * Decrement the reference count on a node (which is the head of a list of
|
|
|
5287 * siblings). If the reference count becomes zero free the node and its
|
|
|
5288 * siblings.
|
|
|
5289 */
|
|
|
5290 static void
|
|
|
5291 deref_wordnode(spin, node)
|
|
|
5292 spellinfo_T *spin;
|
|
|
5293 wordnode_T *node;
|
|
|
5294 {
|
|
|
5295 wordnode_T *np;
|
|
|
5296
|
|
|
5297 if (--node->wn_refs == 0)
|
|
|
5298 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5299 {
|
|
|
5300 if (np->wn_child != NULL)
|
|
|
5301 deref_wordnode(spin, np->wn_child);
|
|
|
5302 free_wordnode(spin, np);
|
|
|
5303 }
|
|
|
5304 }
|
|
|
5305
|
|
|
5306 /*
|
|
469
|
5307 * Free a wordnode_T for re-use later.
|
|
474
|
5308 * Only the "wn_child" field becomes invalid.
|
|
469
|
5309 */
|
|
|
5310 static void
|
|
474
|
5311 free_wordnode(spin, n)
|
|
|
5312 spellinfo_T *spin;
|
|
|
5313 wordnode_T *n;
|
|
|
5314 {
|
|
|
5315 n->wn_child = spin->si_first_free;
|
|
|
5316 spin->si_first_free = n;
|
|
482
|
5317 ++spin->si_free_count;
|
|
469
|
5318 }
|
|
|
5319
|
|
236
|
5320 /*
|
|
300
|
5321 * Compress a tree: find tails that are identical and can be shared.
|
|
|
5322 */
|
|
|
5323 static void
|
|
474
|
5324 wordtree_compress(spin, root)
|
|
|
5325 spellinfo_T *spin;
|
|
300
|
5326 wordnode_T *root;
|
|
|
5327 {
|
|
|
5328 hashtab_T ht;
|
|
|
5329 int n;
|
|
|
5330 int tot = 0;
|
|
469
|
5331 int perc;
|
|
300
|
5332
|
|
474
|
5333 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
5334 * start of the tree. */
|
|
|
5335 if (root->wn_sibling != NULL)
|
|
300
|
5336 {
|
|
|
5337 hash_init(&ht);
|
|
474
|
5338 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
5339
|
|
|
5340 #ifndef SPELL_PRINTTREE
|
|
310
|
5341 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
5342 #endif
|
|
310
|
5343 {
|
|
|
5344 if (!spin->si_verbose)
|
|
|
5345 verbose_enter();
|
|
469
|
5346 if (tot > 1000000)
|
|
|
5347 perc = (tot - n) / (tot / 100);
|
|
482
|
5348 else if (tot == 0)
|
|
|
5349 perc = 0;
|
|
469
|
5350 else
|
|
|
5351 perc = (tot - n) * 100 / tot;
|
|
310
|
5352 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
5353 n, tot, perc);
|
|
310
|
5354 if (p_verbose > 2)
|
|
|
5355 verbose_leave();
|
|
|
5356 }
|
|
469
|
5357 #ifdef SPELL_PRINTTREE
|
|
474
|
5358 spell_print_tree(root->wn_sibling);
|
|
469
|
5359 #endif
|
|
300
|
5360 hash_clear(&ht);
|
|
|
5361 }
|
|
|
5362 }
|
|
|
5363
|
|
|
5364 /*
|
|
|
5365 * Compress a node, its siblings and its children, depth first.
|
|
|
5366 * Returns the number of compressed nodes.
|
|
236
|
5367 */
|
|
255
|
5368 static int
|
|
474
|
5369 node_compress(spin, node, ht, tot)
|
|
|
5370 spellinfo_T *spin;
|
|
300
|
5371 wordnode_T *node;
|
|
|
5372 hashtab_T *ht;
|
|
|
5373 int *tot; /* total count of nodes before compressing,
|
|
|
5374 incremented while going through the tree */
|
|
236
|
5375 {
|
|
300
|
5376 wordnode_T *np;
|
|
|
5377 wordnode_T *tp;
|
|
|
5378 wordnode_T *child;
|
|
|
5379 hash_T hash;
|
|
236
|
5380 hashitem_T *hi;
|
|
300
|
5381 int len = 0;
|
|
|
5382 unsigned nr, n;
|
|
|
5383 int compressed = 0;
|
|
236
|
5384
|
|
300
|
5385 /*
|
|
|
5386 * Go through the list of siblings. Compress each child and then try
|
|
|
5387 * finding an identical child to replace it.
|
|
|
5388 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
5389 * but the whole list of siblings, of which the node is the first.
|
|
|
5390 */
|
|
482
|
5391 for (np = node; np != NULL && !got_int; np = np->wn_sibling)
|
|
236
|
5392 {
|
|
300
|
5393 ++len;
|
|
|
5394 if ((child = np->wn_child) != NULL)
|
|
|
5395 {
|
|
346
|
5396 /* Compress the child. This fills hashkey. */
|
|
474
|
5397 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
5398
|
|
|
5399 /* Try to find an identical child. */
|
|
346
|
5400 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
5401 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
5402 tp = NULL;
|
|
|
5403 if (!HASHITEM_EMPTY(hi))
|
|
|
5404 {
|
|
|
5405 /* There are children with an identical hash value. Now check
|
|
|
5406 * if there is one that is really identical. */
|
|
346
|
5407 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
5408 if (node_equal(child, tp))
|
|
|
5409 {
|
|
|
5410 /* Found one! Now use that child in place of the
|
|
469
|
5411 * current one. This means the current child and all
|
|
|
5412 * its siblings is unlinked from the tree. */
|
|
|
5413 ++tp->wn_refs;
|
|
474
|
5414 deref_wordnode(spin, child);
|
|
300
|
5415 np->wn_child = tp;
|
|
|
5416 ++compressed;
|
|
|
5417 break;
|
|
|
5418 }
|
|
|
5419 if (tp == NULL)
|
|
|
5420 {
|
|
|
5421 /* No other child with this hash value equals the child of
|
|
|
5422 * the node, add it to the linked list after the first
|
|
|
5423 * item. */
|
|
|
5424 tp = HI2WN(hi);
|
|
346
|
5425 child->wn_u2.next = tp->wn_u2.next;
|
|
|
5426 tp->wn_u2.next = child;
|
|
300
|
5427 }
|
|
|
5428 }
|
|
|
5429 else
|
|
|
5430 /* No other child has this hash value, add it to the
|
|
|
5431 * hashtable. */
|
|
346
|
5432 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
5433 }
|
|
236
|
5434 }
|
|
300
|
5435 *tot += len;
|
|
|
5436
|
|
|
5437 /*
|
|
|
5438 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
5439 * find a lookalike node. This must be done after compressing the sibling
|
|
|
5440 * list, otherwise the hash key would become invalid by the compression.
|
|
|
5441 */
|
|
346
|
5442 node->wn_u1.hashkey[0] = len;
|
|
300
|
5443 nr = 0;
|
|
|
5444 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5445 {
|
|
300
|
5446 if (np->wn_byte == NUL)
|
|
481
|
5447 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
5448 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
5449 else
|
|
|
5450 /* byte node: use the byte value and the child pointer */
|
|
|
5451 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
5452 nr = nr * 101 + n;
|
|
236
|
5453 }
|
|
300
|
5454
|
|
|
5455 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
5456 n = nr & 0xff;
|
|
346
|
5457 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
5458 n = (nr >> 8) & 0xff;
|
|
346
|
5459 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
5460 n = (nr >> 16) & 0xff;
|
|
346
|
5461 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
5462 n = (nr >> 24) & 0xff;
|
|
346
|
5463 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
5464 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
5465
|
|
482
|
5466 /* Check for CTRL-C pressed now and then. */
|
|
|
5467 fast_breakcheck();
|
|
|
5468
|
|
300
|
5469 return compressed;
|
|
|
5470 }
|
|
|
5471
|
|
|
5472 /*
|
|
|
5473 * Return TRUE when two nodes have identical siblings and children.
|
|
|
5474 */
|
|
|
5475 static int
|
|
|
5476 node_equal(n1, n2)
|
|
|
5477 wordnode_T *n1;
|
|
|
5478 wordnode_T *n2;
|
|
|
5479 {
|
|
|
5480 wordnode_T *p1;
|
|
|
5481 wordnode_T *p2;
|
|
|
5482
|
|
|
5483 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
5484 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
5485 if (p1->wn_byte != p2->wn_byte
|
|
|
5486 || (p1->wn_byte == NUL
|
|
|
5487 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
5488 || p1->wn_region != p2->wn_region
|
|
481
|
5489 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
5490 : (p1->wn_child != p2->wn_child)))
|
|
|
5491 break;
|
|
|
5492
|
|
|
5493 return p1 == NULL && p2 == NULL;
|
|
236
|
5494 }
|
|
|
5495
|
|
|
5496 /*
|
|
|
5497 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
5498 */
|
|
255
|
5499 void
|
|
236
|
5500 put_bytes(fd, nr, len)
|
|
|
5501 FILE *fd;
|
|
|
5502 long_u nr;
|
|
|
5503 int len;
|
|
|
5504 {
|
|
|
5505 int i;
|
|
|
5506
|
|
|
5507 for (i = len - 1; i >= 0; --i)
|
|
|
5508 putc((int)(nr >> (i * 8)), fd);
|
|
|
5509 }
|
|
|
5510
|
|
323
|
5511 static int
|
|
|
5512 #ifdef __BORLANDC__
|
|
|
5513 _RTLENTRYF
|
|
|
5514 #endif
|
|
|
5515 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
5516
|
|
|
5517 /*
|
|
|
5518 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
5519 */
|
|
|
5520 static int
|
|
|
5521 #ifdef __BORLANDC__
|
|
|
5522 _RTLENTRYF
|
|
|
5523 #endif
|
|
|
5524 rep_compare(s1, s2)
|
|
|
5525 const void *s1;
|
|
|
5526 const void *s2;
|
|
|
5527 {
|
|
|
5528 fromto_T *p1 = (fromto_T *)s1;
|
|
|
5529 fromto_T *p2 = (fromto_T *)s2;
|
|
|
5530
|
|
|
5531 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
5532 }
|
|
|
5533
|
|
236
|
5534 /*
|
|
|
5535 * Write the Vim spell file "fname".
|
|
|
5536 */
|
|
|
5537 static void
|
|
474
|
5538 write_vim_spell(spin, fname)
|
|
|
5539 spellinfo_T *spin;
|
|
236
|
5540 char_u *fname;
|
|
|
5541 {
|
|
300
|
5542 FILE *fd;
|
|
|
5543 int regionmask;
|
|
236
|
5544 int round;
|
|
300
|
5545 wordnode_T *tree;
|
|
|
5546 int nodecount;
|
|
323
|
5547 int i;
|
|
|
5548 int l;
|
|
|
5549 garray_T *gap;
|
|
|
5550 fromto_T *ftp;
|
|
|
5551 char_u *p;
|
|
|
5552 int rr;
|
|
236
|
5553
|
|
310
|
5554 fd = mch_fopen((char *)fname, "w");
|
|
300
|
5555 if (fd == NULL)
|
|
236
|
5556 {
|
|
|
5557 EMSG2(_(e_notopen), fname);
|
|
|
5558 return;
|
|
|
5559 }
|
|
|
5560
|
|
255
|
5561 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
339
|
5562 * <charflagslen> <charflags>
|
|
|
5563 * <fcharslen> <fchars>
|
|
366
|
5564 * <midwordlen> <midword>
|
|
481
|
5565 * <compoundlen> <compoundtype> <compoundinfo>
|
|
339
|
5566 * <prefcondcnt> <prefcond> ... */
|
|
300
|
5567
|
|
|
5568 /* <fileID> */
|
|
|
5569 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
5570 EMSG(_(e_write));
|
|
236
|
5571
|
|
|
5572 /* write the region names if there is more than one */
|
|
316
|
5573 if (spin->si_region_count > 1)
|
|
236
|
5574 {
|
|
316
|
5575 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
|
5576 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
|
5577 (size_t)1, fd);
|
|
|
5578 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
5579 }
|
|
|
5580 else
|
|
|
5581 {
|
|
300
|
5582 putc(0, fd);
|
|
|
5583 regionmask = 0;
|
|
236
|
5584 }
|
|
|
5585
|
|
323
|
5586 /*
|
|
|
5587 * Write the table with character flags and table for case folding.
|
|
260
|
5588 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
|
5589 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
5590 * 'encoding'.
|
|
|
5591 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
5592 * the table (avoids that it conflicts). File is shorter too.
|
|
|
5593 */
|
|
|
5594 if (spin->si_ascii || spin->si_add)
|
|
260
|
5595 {
|
|
300
|
5596 putc(0, fd);
|
|
|
5597 putc(0, fd);
|
|
|
5598 putc(0, fd);
|
|
260
|
5599 }
|
|
|
5600 else
|
|
300
|
5601 write_spell_chartab(fd);
|
|
255
|
5602
|
|
366
|
5603
|
|
|
5604 if (spin->si_midword == NULL)
|
|
|
5605 put_bytes(fd, 0L, 2); /* <midwordlen> */
|
|
|
5606 else
|
|
|
5607 {
|
|
|
5608 i = STRLEN(spin->si_midword);
|
|
|
5609 put_bytes(fd, (long_u)i, 2); /* <midwordlen> */
|
|
|
5610 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
5611 }
|
|
|
5612
|
|
|
5613
|
|
481
|
5614 /* Write the compound info. */
|
|
|
5615 if (spin->si_compflags == NULL)
|
|
|
5616 put_bytes(fd, 0L, 2); /* <compoundlen> */
|
|
|
5617 else
|
|
|
5618 {
|
|
|
5619 l = STRLEN(spin->si_compflags);
|
|
|
5620 put_bytes(fd, (long_u)(l + 2), 2); /* <compoundlen> */
|
|
|
5621 putc(1, fd); /* <compoundtype> */
|
|
|
5622 putc(spin->si_compminlen, fd); /* <comp1minlen> */
|
|
|
5623 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
5624 /* <comp1flags> */
|
|
|
5625 }
|
|
|
5626
|
|
|
5627
|
|
339
|
5628 /* Write the prefix conditions. */
|
|
|
5629 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
5630
|
|
366
|
5631 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
5632 * <salflags> <salcount> <sal> ...
|
|
|
5633 * <maplen> <mapstr> */
|
|
|
5634
|
|
323
|
5635 /* Sort the REP items. */
|
|
|
5636 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
5637 sizeof(fromto_T), rep_compare);
|
|
|
5638
|
|
375
|
5639 /* round 1: REP items
|
|
|
5640 * round 2: SAL items (unless SOFO is used) */
|
|
323
|
5641 for (round = 1; round <= 2; ++round)
|
|
|
5642 {
|
|
|
5643 if (round == 1)
|
|
|
5644 gap = &spin->si_rep;
|
|
|
5645 else
|
|
|
5646 {
|
|
|
5647 gap = &spin->si_sal;
|
|
|
5648
|
|
|
5649 i = 0;
|
|
|
5650 if (spin->si_followup)
|
|
|
5651 i |= SAL_F0LLOWUP;
|
|
|
5652 if (spin->si_collapse)
|
|
|
5653 i |= SAL_COLLAPSE;
|
|
|
5654 if (spin->si_rem_accents)
|
|
|
5655 i |= SAL_REM_ACCENTS;
|
|
375
|
5656 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5657 i |= SAL_SOFO;
|
|
323
|
5658 putc(i, fd); /* <salflags> */
|
|
375
|
5659 if (i & SAL_SOFO)
|
|
|
5660 break;
|
|
323
|
5661 }
|
|
|
5662
|
|
|
5663 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
5664 for (i = 0; i < gap->ga_len; ++i)
|
|
|
5665 {
|
|
|
5666 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
5667 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
5668 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
5669 for (rr = 1; rr <= 2; ++rr)
|
|
|
5670 {
|
|
|
5671 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
5672 l = STRLEN(p);
|
|
|
5673 putc(l, fd);
|
|
|
5674 fwrite(p, l, (size_t)1, fd);
|
|
|
5675 }
|
|
|
5676 }
|
|
|
5677 }
|
|
|
5678
|
|
375
|
5679 /* SOFOFROM and SOFOTO */
|
|
|
5680 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5681 {
|
|
|
5682 put_bytes(fd, 1L, 2); /* <salcount> */
|
|
|
5683
|
|
|
5684 l = STRLEN(spin->si_sofofr);
|
|
|
5685 put_bytes(fd, (long_u)l, 2); /* <salfromlen> */
|
|
|
5686 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <salfrom> */
|
|
|
5687
|
|
|
5688 l = STRLEN(spin->si_sofoto);
|
|
|
5689 put_bytes(fd, (long_u)l, 2); /* <saltolen> */
|
|
|
5690 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <salto> */
|
|
|
5691 }
|
|
|
5692
|
|
323
|
5693 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
|
5694 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
|
5695 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
|
5696 (size_t)1, fd);
|
|
302
|
5697
|
|
236
|
5698 /*
|
|
339
|
5699 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
5700 */
|
|
323
|
5701 spin->si_memtot = 0;
|
|
339
|
5702 for (round = 1; round <= 3; ++round)
|
|
236
|
5703 {
|
|
339
|
5704 if (round == 1)
|
|
469
|
5705 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
5706 else if (round == 2)
|
|
469
|
5707 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
5708 else
|
|
469
|
5709 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
5710
|
|
346
|
5711 /* Clear the index and wnode fields in the tree. */
|
|
|
5712 clear_node(tree);
|
|
|
5713
|
|
300
|
5714 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
5715 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
5716 * nodes. */
|
|
346
|
5717 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
5718
|
|
300
|
5719 /* number of nodes in 4 bytes */
|
|
|
5720 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
5721 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
5722
|
|
300
|
5723 /* Write the nodes. */
|
|
346
|
5724 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
5725 }
|
|
|
5726
|
|
300
|
5727 fclose(fd);
|
|
236
|
5728 }
|
|
|
5729
|
|
|
5730 /*
|
|
346
|
5731 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
5732 * children. This is needed because they are a union with other items to save
|
|
|
5733 * space.
|
|
|
5734 */
|
|
|
5735 static void
|
|
|
5736 clear_node(node)
|
|
|
5737 wordnode_T *node;
|
|
|
5738 {
|
|
|
5739 wordnode_T *np;
|
|
|
5740
|
|
|
5741 if (node != NULL)
|
|
|
5742 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5743 {
|
|
|
5744 np->wn_u1.index = 0;
|
|
|
5745 np->wn_u2.wnode = NULL;
|
|
|
5746
|
|
|
5747 if (np->wn_byte != NUL)
|
|
|
5748 clear_node(np->wn_child);
|
|
|
5749 }
|
|
|
5750 }
|
|
|
5751
|
|
|
5752
|
|
|
5753 /*
|
|
300
|
5754 * Dump a word tree at node "node".
|
|
|
5755 *
|
|
|
5756 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
5757 * byte recursively write the children.
|
|
|
5758 *
|
|
|
5759 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
5760 * are made about the indexes (so that we don't have to write them in the
|
|
|
5761 * file).
|
|
236
|
5762 *
|
|
300
|
5763 * Returns the number of nodes used.
|
|
236
|
5764 */
|
|
300
|
5765 static int
|
|
346
|
5766 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
5767 FILE *fd; /* NULL when only counting */
|
|
300
|
5768 wordnode_T *node;
|
|
|
5769 int index;
|
|
|
5770 int regionmask;
|
|
339
|
5771 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
5772 {
|
|
300
|
5773 int newindex = index;
|
|
|
5774 int siblingcount = 0;
|
|
|
5775 wordnode_T *np;
|
|
236
|
5776 int flags;
|
|
300
|
5777
|
|
|
5778 /* If "node" is zero the tree is empty. */
|
|
|
5779 if (node == NULL)
|
|
|
5780 return 0;
|
|
|
5781
|
|
|
5782 /* Store the index where this node is written. */
|
|
346
|
5783 node->wn_u1.index = index;
|
|
236
|
5784
|
|
300
|
5785 /* Count the number of siblings. */
|
|
|
5786 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5787 ++siblingcount;
|
|
236
|
5788
|
|
300
|
5789 /* Write the sibling count. */
|
|
|
5790 if (fd != NULL)
|
|
|
5791 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
5792
|
|
300
|
5793 /* Write each sibling byte and optionally extra info. */
|
|
|
5794 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5795 {
|
|
300
|
5796 if (np->wn_byte == 0)
|
|
|
5797 {
|
|
|
5798 if (fd != NULL)
|
|
|
5799 {
|
|
339
|
5800 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
5801 if (prefixtree)
|
|
300
|
5802 {
|
|
481
|
5803 /* In PREFIXTREE write the required affixID and the
|
|
390
|
5804 * associated condition nr (stored in wn_region). The
|
|
|
5805 * byte value is misused to store the "rare" and "not
|
|
|
5806 * combining" flags */
|
|
455
|
5807 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
5808 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5809 else
|
|
|
5810 {
|
|
390
|
5811 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
5812 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
5813 }
|
|
481
|
5814 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
5815 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
5816 }
|
|
|
5817 else
|
|
|
5818 {
|
|
339
|
5819 /* For word trees we write the flag/region items. */
|
|
|
5820 flags = np->wn_flags;
|
|
|
5821 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
5822 flags |= WF_REGION;
|
|
481
|
5823 if (np->wn_affixID != 0)
|
|
|
5824 flags |= WF_AFX;
|
|
339
|
5825 if (flags == 0)
|
|
|
5826 {
|
|
|
5827 /* word without flags or region */
|
|
|
5828 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5829 }
|
|
|
5830 else
|
|
|
5831 {
|
|
390
|
5832 if (np->wn_flags >= 0x100)
|
|
|
5833 {
|
|
|
5834 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
5835 putc(flags, fd); /* <flags> */
|
|
|
5836 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
5837 }
|
|
|
5838 else
|
|
|
5839 {
|
|
|
5840 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
5841 putc(flags, fd); /* <flags> */
|
|
|
5842 }
|
|
339
|
5843 if (flags & WF_REGION)
|
|
|
5844 putc(np->wn_region, fd); /* <region> */
|
|
481
|
5845 if (flags & WF_AFX)
|
|
|
5846 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
5847 }
|
|
300
|
5848 }
|
|
|
5849 }
|
|
|
5850 }
|
|
|
5851 else
|
|
|
5852 {
|
|
346
|
5853 if (np->wn_child->wn_u1.index != 0
|
|
|
5854 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
5855 {
|
|
|
5856 /* The child is written elsewhere, write the reference. */
|
|
|
5857 if (fd != NULL)
|
|
|
5858 {
|
|
|
5859 putc(BY_INDEX, fd); /* <byte> */
|
|
|
5860 /* <nodeidx> */
|
|
346
|
5861 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
5862 }
|
|
|
5863 }
|
|
346
|
5864 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
5865 /* We will write the child below and give it an index. */
|
|
346
|
5866 np->wn_child->wn_u2.wnode = node;
|
|
236
|
5867
|
|
300
|
5868 if (fd != NULL)
|
|
|
5869 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
5870 {
|
|
|
5871 EMSG(_(e_write));
|
|
|
5872 return 0;
|
|
|
5873 }
|
|
|
5874 }
|
|
236
|
5875 }
|
|
|
5876
|
|
300
|
5877 /* Space used in the array when reading: one for each sibling and one for
|
|
|
5878 * the count. */
|
|
|
5879 newindex += siblingcount + 1;
|
|
249
|
5880
|
|
300
|
5881 /* Recursively dump the children of each sibling. */
|
|
|
5882 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
5883 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
5884 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
5885 prefixtree);
|
|
249
|
5886
|
|
300
|
5887 return newindex;
|
|
236
|
5888 }
|
|
|
5889
|
|
|
5890
|
|
|
5891 /*
|
|
310
|
5892 * ":mkspell [-ascii] outfile infile ..."
|
|
|
5893 * ":mkspell [-ascii] addfile"
|
|
236
|
5894 */
|
|
|
5895 void
|
|
|
5896 ex_mkspell(eap)
|
|
|
5897 exarg_T *eap;
|
|
|
5898 {
|
|
|
5899 int fcount;
|
|
|
5900 char_u **fnames;
|
|
310
|
5901 char_u *arg = eap->arg;
|
|
|
5902 int ascii = FALSE;
|
|
|
5903
|
|
|
5904 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
5905 {
|
|
|
5906 ascii = TRUE;
|
|
|
5907 arg = skipwhite(arg + 6);
|
|
|
5908 }
|
|
|
5909
|
|
|
5910 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
5911 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
5912 {
|
|
323
|
5913 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
5914 FreeWild(fcount, fnames);
|
|
|
5915 }
|
|
|
5916 }
|
|
|
5917
|
|
|
5918 /*
|
|
|
5919 * Create a Vim spell file from one or more word lists.
|
|
|
5920 * "fnames[0]" is the output file name.
|
|
|
5921 * "fnames[fcount - 1]" is the last input file name.
|
|
|
5922 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
5923 * and ".spl" is appended to make the output file name.
|
|
|
5924 */
|
|
|
5925 static void
|
|
323
|
5926 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
5927 int fcount;
|
|
|
5928 char_u **fnames;
|
|
|
5929 int ascii; /* -ascii argument given */
|
|
|
5930 int overwrite; /* overwrite existing output file */
|
|
323
|
5931 int added_word; /* invoked through "zg" */
|
|
310
|
5932 {
|
|
236
|
5933 char_u fname[MAXPATHL];
|
|
|
5934 char_u wfname[MAXPATHL];
|
|
310
|
5935 char_u **innames;
|
|
|
5936 int incount;
|
|
236
|
5937 afffile_T *(afile[8]);
|
|
|
5938 int i;
|
|
|
5939 int len;
|
|
|
5940 struct stat st;
|
|
255
|
5941 int error = FALSE;
|
|
300
|
5942 spellinfo_T spin;
|
|
|
5943
|
|
|
5944 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
5945 spin.si_verbose = !added_word;
|
|
310
|
5946 spin.si_ascii = ascii;
|
|
323
|
5947 spin.si_followup = TRUE;
|
|
|
5948 spin.si_rem_accents = TRUE;
|
|
|
5949 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
5950 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
5951 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
5952 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
5953
|
|
|
5954 /* default: fnames[0] is output file, following are input files */
|
|
|
5955 innames = &fnames[1];
|
|
|
5956 incount = fcount - 1;
|
|
|
5957
|
|
|
5958 if (fcount >= 1)
|
|
240
|
5959 {
|
|
310
|
5960 len = STRLEN(fnames[0]);
|
|
|
5961 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
5962 {
|
|
|
5963 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
5964 * "path/en.latin1.add.spl". */
|
|
|
5965 innames = &fnames[0];
|
|
|
5966 incount = 1;
|
|
|
5967 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
5968 }
|
|
366
|
5969 else if (fcount == 1)
|
|
|
5970 {
|
|
|
5971 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
5972 innames = &fnames[0];
|
|
|
5973 incount = 1;
|
|
|
5974 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5975 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5976 }
|
|
310
|
5977 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
5978 {
|
|
|
5979 /* Name ends in ".spl", use as the file name. */
|
|
323
|
5980 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
5981 }
|
|
|
5982 else
|
|
|
5983 /* Name should be language, make the file name from it. */
|
|
|
5984 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5985 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5986
|
|
|
5987 /* Check for .ascii.spl. */
|
|
|
5988 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
5989 spin.si_ascii = TRUE;
|
|
|
5990
|
|
|
5991 /* Check for .add.spl. */
|
|
|
5992 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
5993 spin.si_add = TRUE;
|
|
240
|
5994 }
|
|
|
5995
|
|
310
|
5996 if (incount <= 0)
|
|
236
|
5997 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
5998 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
5999 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
6000 else if (incount > 8)
|
|
236
|
6001 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
6002 else
|
|
|
6003 {
|
|
|
6004 /* Check for overwriting before doing things that may take a lot of
|
|
|
6005 * time. */
|
|
310
|
6006 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
6007 {
|
|
|
6008 EMSG(_(e_exists));
|
|
310
|
6009 return;
|
|
236
|
6010 }
|
|
310
|
6011 if (mch_isdir(wfname))
|
|
236
|
6012 {
|
|
310
|
6013 EMSG2(_(e_isadir2), wfname);
|
|
|
6014 return;
|
|
236
|
6015 }
|
|
|
6016
|
|
|
6017 /*
|
|
|
6018 * Init the aff and dic pointers.
|
|
|
6019 * Get the region names if there are more than 2 arguments.
|
|
|
6020 */
|
|
310
|
6021 for (i = 0; i < incount; ++i)
|
|
236
|
6022 {
|
|
310
|
6023 afile[i] = NULL;
|
|
300
|
6024
|
|
316
|
6025 if (incount > 1)
|
|
236
|
6026 {
|
|
310
|
6027 len = STRLEN(innames[i]);
|
|
|
6028 if (STRLEN(gettail(innames[i])) < 5
|
|
|
6029 || innames[i][len - 3] != '_')
|
|
236
|
6030 {
|
|
310
|
6031 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
6032 return;
|
|
236
|
6033 }
|
|
316
|
6034 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
6035 spin.si_region_name[i * 2 + 1] =
|
|
|
6036 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
6037 }
|
|
|
6038 }
|
|
316
|
6039 spin.si_region_count = incount;
|
|
236
|
6040
|
|
474
|
6041 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
6042 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
6043 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
6044 if (spin.si_foldroot == NULL
|
|
|
6045 || spin.si_keeproot == NULL
|
|
|
6046 || spin.si_prefroot == NULL)
|
|
300
|
6047 {
|
|
469
|
6048 free_blocks(spin.si_blocks);
|
|
310
|
6049 return;
|
|
300
|
6050 }
|
|
|
6051
|
|
351
|
6052 /* When not producing a .add.spl file clear the character table when
|
|
|
6053 * we encounter one in the .aff file. This means we dump the current
|
|
|
6054 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
6055 * better than guessing the contents, the table will match a
|
|
|
6056 * previously loaded spell file. */
|
|
|
6057 if (!spin.si_add)
|
|
|
6058 spin.si_clear_chartab = TRUE;
|
|
|
6059
|
|
236
|
6060 /*
|
|
|
6061 * Read all the .aff and .dic files.
|
|
|
6062 * Text is converted to 'encoding'.
|
|
300
|
6063 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
6064 */
|
|
310
|
6065 for (i = 0; i < incount && !error; ++i)
|
|
236
|
6066 {
|
|
300
|
6067 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
6068 spin.si_region = 1 << i;
|
|
|
6069
|
|
|
6070 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
6071 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
6072 {
|
|
|
6073 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
6074 * "SET" line. */
|
|
474
|
6075 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
6076 if (afile[i] == NULL)
|
|
300
|
6077 error = TRUE;
|
|
|
6078 else
|
|
|
6079 {
|
|
|
6080 /* Read the .dic file and store the words in the trees. */
|
|
|
6081 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
6082 innames[i]);
|
|
474
|
6083 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
6084 error = TRUE;
|
|
|
6085 }
|
|
|
6086 }
|
|
|
6087 else
|
|
|
6088 {
|
|
|
6089 /* No .aff file, try reading the file as a word list. Store
|
|
|
6090 * the words in the trees. */
|
|
474
|
6091 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
6092 error = TRUE;
|
|
|
6093 }
|
|
236
|
6094
|
|
310
|
6095 #ifdef FEAT_MBYTE
|
|
236
|
6096 /* Free any conversion stuff. */
|
|
300
|
6097 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
6098 #endif
|
|
236
|
6099 }
|
|
|
6100
|
|
300
|
6101 if (!error)
|
|
236
|
6102 {
|
|
|
6103 /*
|
|
300
|
6104 * Combine tails in the tree.
|
|
236
|
6105 */
|
|
474
|
6106 if (spin.si_verbose || p_verbose > 2)
|
|
|
6107 {
|
|
|
6108 if (!spin.si_verbose)
|
|
310
|
6109 verbose_enter();
|
|
469
|
6110 MSG(_(msg_compressing));
|
|
310
|
6111 out_flush();
|
|
474
|
6112 if (!spin.si_verbose)
|
|
310
|
6113 verbose_leave();
|
|
|
6114 }
|
|
474
|
6115 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
6116 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
6117 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
6118 }
|
|
|
6119
|
|
300
|
6120 if (!error)
|
|
|
6121 {
|
|
|
6122 /*
|
|
|
6123 * Write the info in the spell file.
|
|
|
6124 */
|
|
474
|
6125 if (spin.si_verbose || p_verbose > 2)
|
|
|
6126 {
|
|
|
6127 if (!spin.si_verbose)
|
|
310
|
6128 verbose_enter();
|
|
366
|
6129 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
6130 out_flush();
|
|
474
|
6131 if (!spin.si_verbose)
|
|
310
|
6132 verbose_leave();
|
|
|
6133 }
|
|
|
6134
|
|
474
|
6135 write_vim_spell(&spin, wfname);
|
|
|
6136
|
|
|
6137 if (spin.si_verbose || p_verbose > 2)
|
|
|
6138 {
|
|
|
6139 if (!spin.si_verbose)
|
|
310
|
6140 verbose_enter();
|
|
|
6141 MSG(_("Done!"));
|
|
|
6142 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
6143 spin.si_memtot);
|
|
310
|
6144 out_flush();
|
|
474
|
6145 if (!spin.si_verbose)
|
|
310
|
6146 verbose_leave();
|
|
|
6147 }
|
|
|
6148
|
|
|
6149 /* If the file is loaded need to reload it. */
|
|
323
|
6150 spell_reload_one(wfname, added_word);
|
|
300
|
6151 }
|
|
|
6152
|
|
|
6153 /* Free the allocated memory. */
|
|
323
|
6154 ga_clear(&spin.si_rep);
|
|
|
6155 ga_clear(&spin.si_sal);
|
|
|
6156 ga_clear(&spin.si_map);
|
|
339
|
6157 ga_clear(&spin.si_prefcond);
|
|
366
|
6158 vim_free(spin.si_midword);
|
|
375
|
6159 vim_free(spin.si_sofofr);
|
|
|
6160 vim_free(spin.si_sofoto);
|
|
300
|
6161
|
|
|
6162 /* Free the .aff file structures. */
|
|
310
|
6163 for (i = 0; i < incount; ++i)
|
|
|
6164 if (afile[i] != NULL)
|
|
|
6165 spell_free_aff(afile[i]);
|
|
339
|
6166
|
|
|
6167 /* Free all the bits and pieces at once. */
|
|
|
6168 free_blocks(spin.si_blocks);
|
|
236
|
6169 }
|
|
310
|
6170 }
|
|
|
6171
|
|
|
6172
|
|
|
6173 /*
|
|
385
|
6174 * ":[count]spellgood {word}"
|
|
|
6175 * ":[count]spellwrong {word}"
|
|
310
|
6176 */
|
|
|
6177 void
|
|
|
6178 ex_spell(eap)
|
|
|
6179 exarg_T *eap;
|
|
|
6180 {
|
|
381
|
6181 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
6182 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
6183 }
|
|
|
6184
|
|
310
|
6185 /*
|
|
|
6186 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
6187 */
|
|
|
6188 void
|
|
385
|
6189 spell_add_word(word, len, bad, index)
|
|
310
|
6190 char_u *word;
|
|
|
6191 int len;
|
|
|
6192 int bad;
|
|
385
|
6193 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
6194 'spellfile' */
|
|
310
|
6195 {
|
|
|
6196 FILE *fd;
|
|
385
|
6197 buf_T *buf = NULL;
|
|
351
|
6198 int new_spf = FALSE;
|
|
|
6199 struct stat st;
|
|
381
|
6200 char_u *fname;
|
|
385
|
6201 char_u fnamebuf[MAXPATHL];
|
|
|
6202 char_u line[MAXWLEN * 2];
|
|
|
6203 long fpos, fpos_next = 0;
|
|
|
6204 int i;
|
|
|
6205 char_u *spf;
|
|
|
6206
|
|
|
6207 if (index == 0) /* use internal wordlist */
|
|
|
6208 {
|
|
|
6209 if (int_wordlist == NULL)
|
|
|
6210 {
|
|
|
6211 int_wordlist = vim_tempname('s');
|
|
|
6212 if (int_wordlist == NULL)
|
|
381
|
6213 return;
|
|
|
6214 }
|
|
385
|
6215 fname = int_wordlist;
|
|
381
|
6216 }
|
|
310
|
6217 else
|
|
|
6218 {
|
|
381
|
6219 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
6220 if (*curbuf->b_p_spf == NUL)
|
|
|
6221 {
|
|
|
6222 init_spellfile();
|
|
|
6223 new_spf = TRUE;
|
|
|
6224 }
|
|
|
6225
|
|
|
6226 if (*curbuf->b_p_spf == NUL)
|
|
|
6227 {
|
|
|
6228 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
6229 return;
|
|
|
6230 }
|
|
|
6231
|
|
385
|
6232 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
6233 {
|
|
|
6234 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
6235 if (i == index)
|
|
|
6236 break;
|
|
|
6237 if (*spf == NUL)
|
|
|
6238 {
|
|
|
6239 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
6240 return;
|
|
|
6241 }
|
|
|
6242 }
|
|
|
6243
|
|
310
|
6244 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
6245 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
6246 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
6247 buf = NULL;
|
|
|
6248 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
6249 {
|
|
310
|
6250 EMSG(_(e_bufloaded));
|
|
381
|
6251 return;
|
|
|
6252 }
|
|
|
6253
|
|
385
|
6254 fname = fnamebuf;
|
|
|
6255 }
|
|
|
6256
|
|
|
6257 if (bad)
|
|
|
6258 {
|
|
|
6259 /* When the word also appears as good word we need to remove that one,
|
|
|
6260 * since its flags sort before the one with WF_BANNED. */
|
|
|
6261 fd = mch_fopen((char *)fname, "r");
|
|
|
6262 if (fd != NULL)
|
|
|
6263 {
|
|
|
6264 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
6265 {
|
|
|
6266 fpos = fpos_next;
|
|
|
6267 fpos_next = ftell(fd);
|
|
|
6268 if (STRNCMP(word, line, len) == 0
|
|
|
6269 && (line[len] == '/' || line[len] < ' '))
|
|
|
6270 {
|
|
|
6271 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
6272 * the start of the line. Mixing reading and writing
|
|
|
6273 * doesn't work for all systems, close the file first. */
|
|
|
6274 fclose(fd);
|
|
|
6275 fd = mch_fopen((char *)fname, "r+");
|
|
|
6276 if (fd == NULL)
|
|
|
6277 break;
|
|
|
6278 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
6279 fputc('#', fd);
|
|
|
6280 fseek(fd, fpos_next, SEEK_SET);
|
|
|
6281 }
|
|
|
6282 }
|
|
|
6283 fclose(fd);
|
|
|
6284 }
|
|
381
|
6285 }
|
|
|
6286
|
|
|
6287 fd = mch_fopen((char *)fname, "a");
|
|
|
6288 if (fd == NULL && new_spf)
|
|
|
6289 {
|
|
|
6290 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
6291 * We may need to create the "spell" directory first. We already
|
|
|
6292 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
6293 STRCPY(NameBuff, fname);
|
|
|
6294 *gettail_sep(NameBuff) = NUL;
|
|
|
6295 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
6296 {
|
|
|
6297 /* The directory doesn't exist. Try creating it and opening the
|
|
|
6298 * file again. */
|
|
|
6299 vim_mkdir(NameBuff, 0755);
|
|
|
6300 fd = mch_fopen((char *)fname, "a");
|
|
|
6301 }
|
|
|
6302 }
|
|
|
6303
|
|
|
6304 if (fd == NULL)
|
|
|
6305 EMSG2(_(e_notopen), fname);
|
|
|
6306 else
|
|
|
6307 {
|
|
|
6308 if (bad)
|
|
|
6309 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
6310 else
|
|
381
|
6311 fprintf(fd, "%.*s\n", len, word);
|
|
|
6312 fclose(fd);
|
|
|
6313
|
|
|
6314 /* Update the .add.spl file. */
|
|
|
6315 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
6316
|
|
|
6317 /* If the .add file is edited somewhere, reload it. */
|
|
|
6318 if (buf != NULL)
|
|
|
6319 buf_reload(buf);
|
|
|
6320
|
|
|
6321 redraw_all_later(NOT_VALID);
|
|
310
|
6322 }
|
|
|
6323 }
|
|
|
6324
|
|
|
6325 /*
|
|
|
6326 * Initialize 'spellfile' for the current buffer.
|
|
|
6327 */
|
|
|
6328 static void
|
|
|
6329 init_spellfile()
|
|
|
6330 {
|
|
|
6331 char_u buf[MAXPATHL];
|
|
|
6332 int l;
|
|
|
6333 slang_T *sl;
|
|
|
6334 char_u *rtp;
|
|
351
|
6335 char_u *lend;
|
|
310
|
6336
|
|
|
6337 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
6338 {
|
|
351
|
6339 /* Find the end of the language name. Exclude the region. */
|
|
|
6340 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
6341 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
6342 ;
|
|
|
6343
|
|
|
6344 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
6345 * are allowed to write. */
|
|
310
|
6346 rtp = p_rtp;
|
|
|
6347 while (*rtp != NUL)
|
|
|
6348 {
|
|
|
6349 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
6350 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
6351 if (filewritable(buf) == 2)
|
|
|
6352 {
|
|
316
|
6353 /* Use the first language name from 'spelllang' and the
|
|
|
6354 * encoding used in the first loaded .spl file. */
|
|
310
|
6355 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
6356 l = STRLEN(buf);
|
|
|
6357 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
6358 "/spell/%.*s.%s.add",
|
|
351
|
6359 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
6360 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
6361 ? (char_u *)"ascii" : spell_enc());
|
|
|
6362 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
6363 break;
|
|
|
6364 }
|
|
|
6365 }
|
|
|
6366 }
|
|
|
6367 }
|
|
236
|
6368
|
|
300
|
6369
|
|
307
|
6370 /*
|
|
|
6371 * Init the chartab used for spelling for ASCII.
|
|
|
6372 * EBCDIC is not supported!
|
|
|
6373 */
|
|
|
6374 static void
|
|
|
6375 clear_spell_chartab(sp)
|
|
|
6376 spelltab_T *sp;
|
|
|
6377 {
|
|
324
|
6378 int i;
|
|
307
|
6379
|
|
|
6380 /* Init everything to FALSE. */
|
|
|
6381 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
6382 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
6383 for (i = 0; i < 256; ++i)
|
|
324
|
6384 {
|
|
307
|
6385 sp->st_fold[i] = i;
|
|
324
|
6386 sp->st_upper[i] = i;
|
|
|
6387 }
|
|
307
|
6388
|
|
|
6389 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
6390 * that is done separately. */
|
|
|
6391 for (i = '0'; i <= '9'; ++i)
|
|
|
6392 sp->st_isw[i] = TRUE;
|
|
|
6393 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
6394 {
|
|
|
6395 sp->st_isw[i] = TRUE;
|
|
|
6396 sp->st_isu[i] = TRUE;
|
|
|
6397 sp->st_fold[i] = i + 0x20;
|
|
|
6398 }
|
|
|
6399 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
6400 {
|
|
307
|
6401 sp->st_isw[i] = TRUE;
|
|
324
|
6402 sp->st_upper[i] = i - 0x20;
|
|
|
6403 }
|
|
307
|
6404 }
|
|
|
6405
|
|
|
6406 /*
|
|
|
6407 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
6408 * Called once while starting up and when 'encoding' changes.
|
|
|
6409 * The default is to use isalpha(), but the spell file should define the word
|
|
|
6410 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
6411 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
6412 */
|
|
|
6413 void
|
|
|
6414 init_spell_chartab()
|
|
|
6415 {
|
|
|
6416 int i;
|
|
|
6417
|
|
|
6418 did_set_spelltab = FALSE;
|
|
|
6419 clear_spell_chartab(&spelltab);
|
|
|
6420 #ifdef FEAT_MBYTE
|
|
|
6421 if (enc_dbcs)
|
|
|
6422 {
|
|
|
6423 /* DBCS: assume double-wide characters are word characters. */
|
|
|
6424 for (i = 128; i <= 255; ++i)
|
|
|
6425 if (MB_BYTE2LEN(i) == 2)
|
|
|
6426 spelltab.st_isw[i] = TRUE;
|
|
|
6427 }
|
|
324
|
6428 else if (enc_utf8)
|
|
|
6429 {
|
|
|
6430 for (i = 128; i < 256; ++i)
|
|
|
6431 {
|
|
|
6432 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
6433 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
6434 spelltab.st_fold[i] = utf_fold(i);
|
|
|
6435 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
6436 }
|
|
|
6437 }
|
|
307
|
6438 else
|
|
|
6439 #endif
|
|
|
6440 {
|
|
324
|
6441 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
6442 for (i = 128; i < 256; ++i)
|
|
|
6443 {
|
|
|
6444 if (MB_ISUPPER(i))
|
|
|
6445 {
|
|
324
|
6446 spelltab.st_isw[i] = TRUE;
|
|
307
|
6447 spelltab.st_isu[i] = TRUE;
|
|
|
6448 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
6449 }
|
|
324
|
6450 else if (MB_ISLOWER(i))
|
|
|
6451 {
|
|
|
6452 spelltab.st_isw[i] = TRUE;
|
|
|
6453 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
6454 }
|
|
307
|
6455 }
|
|
|
6456 }
|
|
|
6457 }
|
|
|
6458
|
|
|
6459 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
6460 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
6461
|
|
|
6462 /*
|
|
|
6463 * Set the spell character tables from strings in the affix file.
|
|
|
6464 */
|
|
|
6465 static int
|
|
|
6466 set_spell_chartab(fol, low, upp)
|
|
|
6467 char_u *fol;
|
|
|
6468 char_u *low;
|
|
|
6469 char_u *upp;
|
|
|
6470 {
|
|
|
6471 /* We build the new tables here first, so that we can compare with the
|
|
|
6472 * previous one. */
|
|
|
6473 spelltab_T new_st;
|
|
|
6474 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
6475 int f, l, u;
|
|
|
6476
|
|
|
6477 clear_spell_chartab(&new_st);
|
|
|
6478
|
|
|
6479 while (*pf != NUL)
|
|
|
6480 {
|
|
|
6481 if (*pl == NUL || *pu == NUL)
|
|
|
6482 {
|
|
|
6483 EMSG(_(e_affform));
|
|
|
6484 return FAIL;
|
|
|
6485 }
|
|
|
6486 #ifdef FEAT_MBYTE
|
|
|
6487 f = mb_ptr2char_adv(&pf);
|
|
|
6488 l = mb_ptr2char_adv(&pl);
|
|
|
6489 u = mb_ptr2char_adv(&pu);
|
|
|
6490 #else
|
|
|
6491 f = *pf++;
|
|
|
6492 l = *pl++;
|
|
|
6493 u = *pu++;
|
|
|
6494 #endif
|
|
|
6495 /* Every character that appears is a word character. */
|
|
|
6496 if (f < 256)
|
|
|
6497 new_st.st_isw[f] = TRUE;
|
|
|
6498 if (l < 256)
|
|
|
6499 new_st.st_isw[l] = TRUE;
|
|
|
6500 if (u < 256)
|
|
|
6501 new_st.st_isw[u] = TRUE;
|
|
|
6502
|
|
|
6503 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
6504 * case-folding */
|
|
|
6505 if (l < 256 && l != f)
|
|
|
6506 {
|
|
|
6507 if (f >= 256)
|
|
|
6508 {
|
|
|
6509 EMSG(_(e_affrange));
|
|
|
6510 return FAIL;
|
|
|
6511 }
|
|
|
6512 new_st.st_fold[l] = f;
|
|
|
6513 }
|
|
|
6514
|
|
|
6515 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
6516 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
6517 * "FOL" . */
|
|
307
|
6518 if (u < 256 && u != f)
|
|
|
6519 {
|
|
|
6520 if (f >= 256)
|
|
|
6521 {
|
|
|
6522 EMSG(_(e_affrange));
|
|
|
6523 return FAIL;
|
|
|
6524 }
|
|
|
6525 new_st.st_fold[u] = f;
|
|
|
6526 new_st.st_isu[u] = TRUE;
|
|
324
|
6527 new_st.st_upper[f] = u;
|
|
307
|
6528 }
|
|
|
6529 }
|
|
|
6530
|
|
|
6531 if (*pl != NUL || *pu != NUL)
|
|
|
6532 {
|
|
|
6533 EMSG(_(e_affform));
|
|
|
6534 return FAIL;
|
|
|
6535 }
|
|
|
6536
|
|
|
6537 return set_spell_finish(&new_st);
|
|
|
6538 }
|
|
|
6539
|
|
|
6540 /*
|
|
|
6541 * Set the spell character tables from strings in the .spl file.
|
|
|
6542 */
|
|
|
6543 static int
|
|
389
|
6544 set_spell_charflags(flags, cnt, fol)
|
|
307
|
6545 char_u *flags;
|
|
389
|
6546 int cnt; /* length of "flags" */
|
|
|
6547 char_u *fol;
|
|
307
|
6548 {
|
|
|
6549 /* We build the new tables here first, so that we can compare with the
|
|
|
6550 * previous one. */
|
|
|
6551 spelltab_T new_st;
|
|
|
6552 int i;
|
|
389
|
6553 char_u *p = fol;
|
|
324
|
6554 int c;
|
|
307
|
6555
|
|
|
6556 clear_spell_chartab(&new_st);
|
|
|
6557
|
|
389
|
6558 for (i = 0; i < 128; ++i)
|
|
|
6559 {
|
|
|
6560 if (i < cnt)
|
|
|
6561 {
|
|
|
6562 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
6563 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
6564 }
|
|
|
6565
|
|
|
6566 if (*p != NUL)
|
|
|
6567 {
|
|
307
|
6568 #ifdef FEAT_MBYTE
|
|
389
|
6569 c = mb_ptr2char_adv(&p);
|
|
307
|
6570 #else
|
|
389
|
6571 c = *p++;
|
|
307
|
6572 #endif
|
|
389
|
6573 new_st.st_fold[i + 128] = c;
|
|
|
6574 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
6575 new_st.st_upper[c] = i + 128;
|
|
|
6576 }
|
|
307
|
6577 }
|
|
|
6578
|
|
|
6579 return set_spell_finish(&new_st);
|
|
|
6580 }
|
|
|
6581
|
|
|
6582 static int
|
|
|
6583 set_spell_finish(new_st)
|
|
|
6584 spelltab_T *new_st;
|
|
|
6585 {
|
|
|
6586 int i;
|
|
|
6587
|
|
|
6588 if (did_set_spelltab)
|
|
|
6589 {
|
|
|
6590 /* check that it's the same table */
|
|
|
6591 for (i = 0; i < 256; ++i)
|
|
|
6592 {
|
|
|
6593 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
6594 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
6595 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
6596 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
6597 {
|
|
|
6598 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
6599 return FAIL;
|
|
|
6600 }
|
|
|
6601 }
|
|
|
6602 }
|
|
|
6603 else
|
|
|
6604 {
|
|
|
6605 /* copy the new spelltab into the one being used */
|
|
|
6606 spelltab = *new_st;
|
|
|
6607 did_set_spelltab = TRUE;
|
|
|
6608 }
|
|
|
6609
|
|
|
6610 return OK;
|
|
|
6611 }
|
|
|
6612
|
|
|
6613 /*
|
|
358
|
6614 * Return TRUE if "p" points to a word character.
|
|
366
|
6615 * As a special case we see "midword" characters as word character when it is
|
|
358
|
6616 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
6617 * Thus this only works properly when past the first character of the word.
|
|
358
|
6618 */
|
|
|
6619 static int
|
|
376
|
6620 spell_iswordp(p, buf)
|
|
358
|
6621 char_u *p;
|
|
376
|
6622 buf_T *buf; /* buffer used */
|
|
358
|
6623 {
|
|
366
|
6624 #ifdef FEAT_MBYTE
|
|
358
|
6625 char_u *s;
|
|
366
|
6626 int l;
|
|
|
6627 int c;
|
|
|
6628
|
|
|
6629 if (has_mbyte)
|
|
|
6630 {
|
|
|
6631 l = MB_BYTE2LEN(*p);
|
|
358
|
6632 s = p;
|
|
366
|
6633 if (l == 1)
|
|
|
6634 {
|
|
|
6635 /* be quick for ASCII */
|
|
376
|
6636 if (buf->b_spell_ismw[*p])
|
|
366
|
6637 {
|
|
|
6638 s = p + 1; /* skip a mid-word character */
|
|
|
6639 l = MB_BYTE2LEN(*s);
|
|
|
6640 }
|
|
|
6641 }
|
|
|
6642 else
|
|
|
6643 {
|
|
|
6644 c = mb_ptr2char(p);
|
|
376
|
6645 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
6646 : (buf->b_spell_ismw_mb != NULL
|
|
|
6647 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
6648 {
|
|
|
6649 s = p + l;
|
|
|
6650 l = MB_BYTE2LEN(*s);
|
|
|
6651 }
|
|
|
6652 }
|
|
|
6653
|
|
390
|
6654 c = mb_ptr2char(s);
|
|
|
6655 if (c > 255)
|
|
366
|
6656 return mb_get_class(s) >= 2;
|
|
390
|
6657 return spelltab.st_isw[c];
|
|
366
|
6658 }
|
|
358
|
6659 #endif
|
|
366
|
6660
|
|
376
|
6661 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
6662 }
|
|
|
6663
|
|
|
6664 /*
|
|
|
6665 * Return TRUE if "p" points to a word character.
|
|
|
6666 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
6667 */
|
|
|
6668 static int
|
|
|
6669 spell_iswordp_nmw(p)
|
|
|
6670 char_u *p;
|
|
|
6671 {
|
|
|
6672 #ifdef FEAT_MBYTE
|
|
390
|
6673 int c;
|
|
|
6674
|
|
|
6675 if (has_mbyte)
|
|
|
6676 {
|
|
|
6677 c = mb_ptr2char(p);
|
|
|
6678 if (c > 255)
|
|
|
6679 return mb_get_class(p) >= 2;
|
|
|
6680 return spelltab.st_isw[c];
|
|
|
6681 }
|
|
376
|
6682 #endif
|
|
|
6683 return spelltab.st_isw[*p];
|
|
358
|
6684 }
|
|
|
6685
|
|
372
|
6686 #ifdef FEAT_MBYTE
|
|
|
6687 /*
|
|
|
6688 * Return TRUE if "p" points to a word character.
|
|
|
6689 * Wide version of spell_iswordp().
|
|
|
6690 */
|
|
|
6691 static int
|
|
376
|
6692 spell_iswordp_w(p, buf)
|
|
372
|
6693 int *p;
|
|
376
|
6694 buf_T *buf;
|
|
372
|
6695 {
|
|
|
6696 int *s;
|
|
|
6697
|
|
376
|
6698 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
6699 : (buf->b_spell_ismw_mb != NULL
|
|
|
6700 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
6701 s = p + 1;
|
|
|
6702 else
|
|
|
6703 s = p;
|
|
|
6704
|
|
390
|
6705 if (*s > 255)
|
|
372
|
6706 {
|
|
|
6707 if (enc_utf8)
|
|
|
6708 return utf_class(*s) >= 2;
|
|
|
6709 if (enc_dbcs)
|
|
|
6710 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
6711 return 0;
|
|
|
6712 }
|
|
|
6713 return spelltab.st_isw[*s];
|
|
|
6714 }
|
|
|
6715 #endif
|
|
|
6716
|
|
358
|
6717 /*
|
|
339
|
6718 * Write the table with prefix conditions to the .spl file.
|
|
|
6719 */
|
|
|
6720 static void
|
|
|
6721 write_spell_prefcond(fd, gap)
|
|
|
6722 FILE *fd;
|
|
|
6723 garray_T *gap;
|
|
|
6724 {
|
|
|
6725 int i;
|
|
|
6726 char_u *p;
|
|
|
6727 int len;
|
|
|
6728
|
|
|
6729 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
6730
|
|
|
6731 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6732 {
|
|
|
6733 /* <prefcond> : <condlen> <condstr> */
|
|
|
6734 p = ((char_u **)gap->ga_data)[i];
|
|
|
6735 if (p == NULL)
|
|
|
6736 fputc(0, fd);
|
|
|
6737 else
|
|
|
6738 {
|
|
|
6739 len = STRLEN(p);
|
|
|
6740 fputc(len, fd);
|
|
|
6741 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
6742 }
|
|
|
6743 }
|
|
|
6744 }
|
|
|
6745
|
|
|
6746 /*
|
|
307
|
6747 * Write the current tables into the .spl file.
|
|
|
6748 * This makes sure the same characters are recognized as word characters when
|
|
|
6749 * generating an when using a spell file.
|
|
|
6750 */
|
|
|
6751 static void
|
|
|
6752 write_spell_chartab(fd)
|
|
|
6753 FILE *fd;
|
|
|
6754 {
|
|
|
6755 char_u charbuf[256 * 4];
|
|
|
6756 int len = 0;
|
|
|
6757 int flags;
|
|
|
6758 int i;
|
|
|
6759
|
|
|
6760 fputc(128, fd); /* <charflagslen> */
|
|
|
6761 for (i = 128; i < 256; ++i)
|
|
|
6762 {
|
|
|
6763 flags = 0;
|
|
|
6764 if (spelltab.st_isw[i])
|
|
324
|
6765 flags |= CF_WORD;
|
|
307
|
6766 if (spelltab.st_isu[i])
|
|
324
|
6767 flags |= CF_UPPER;
|
|
307
|
6768 fputc(flags, fd); /* <charflags> */
|
|
|
6769
|
|
310
|
6770 #ifdef FEAT_MBYTE
|
|
|
6771 if (has_mbyte)
|
|
|
6772 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
|
6773 else
|
|
|
6774 #endif
|
|
|
6775 charbuf[len++] = spelltab.st_fold[i];
|
|
307
|
6776 }
|
|
|
6777
|
|
|
6778 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
|
6779 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
|
6780 }
|
|
|
6781
|
|
|
6782 /*
|
|
324
|
6783 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
6784 * Uses the character definitions from the .spl file.
|
|
307
|
6785 * When using a multi-byte 'encoding' the length may change!
|
|
|
6786 * Returns FAIL when something wrong.
|
|
|
6787 */
|
|
|
6788 static int
|
|
324
|
6789 spell_casefold(str, len, buf, buflen)
|
|
|
6790 char_u *str;
|
|
307
|
6791 int len;
|
|
|
6792 char_u *buf;
|
|
|
6793 int buflen;
|
|
|
6794 {
|
|
|
6795 int i;
|
|
|
6796
|
|
|
6797 if (len >= buflen)
|
|
|
6798 {
|
|
|
6799 buf[0] = NUL;
|
|
|
6800 return FAIL; /* result will not fit */
|
|
|
6801 }
|
|
|
6802
|
|
|
6803 #ifdef FEAT_MBYTE
|
|
|
6804 if (has_mbyte)
|
|
|
6805 {
|
|
324
|
6806 int outi = 0;
|
|
|
6807 char_u *p;
|
|
307
|
6808 int c;
|
|
|
6809
|
|
|
6810 /* Fold one character at a time. */
|
|
324
|
6811 for (p = str; p < str + len; )
|
|
307
|
6812 {
|
|
|
6813 if (outi + MB_MAXBYTES > buflen)
|
|
|
6814 {
|
|
|
6815 buf[outi] = NUL;
|
|
|
6816 return FAIL;
|
|
|
6817 }
|
|
474
|
6818 c = mb_cptr2char_adv(&p);
|
|
324
|
6819 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
6820 }
|
|
|
6821 buf[outi] = NUL;
|
|
|
6822 }
|
|
|
6823 else
|
|
|
6824 #endif
|
|
|
6825 {
|
|
|
6826 /* Be quick for non-multibyte encodings. */
|
|
|
6827 for (i = 0; i < len; ++i)
|
|
324
|
6828 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
6829 buf[i] = NUL;
|
|
|
6830 }
|
|
|
6831
|
|
|
6832 return OK;
|
|
|
6833 }
|
|
|
6834
|
|
372
|
6835 #define SPS_BEST 1
|
|
|
6836 #define SPS_FAST 2
|
|
|
6837 #define SPS_DOUBLE 4
|
|
|
6838
|
|
|
6839 static int sps_flags = SPS_BEST;
|
|
|
6840
|
|
|
6841 /*
|
|
|
6842 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
|
6843 * Sets "sps_flags".
|
|
|
6844 */
|
|
|
6845 int
|
|
|
6846 spell_check_sps()
|
|
|
6847 {
|
|
|
6848 char_u *p;
|
|
|
6849 char_u buf[MAXPATHL];
|
|
|
6850 int f;
|
|
|
6851
|
|
|
6852 sps_flags = 0;
|
|
|
6853
|
|
|
6854 for (p = p_sps; *p != NUL; )
|
|
|
6855 {
|
|
|
6856 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6857
|
|
|
6858 f = 0;
|
|
|
6859 if (STRCMP(buf, "best") == 0)
|
|
|
6860 f = SPS_BEST;
|
|
|
6861 else if (STRCMP(buf, "fast") == 0)
|
|
|
6862 f = SPS_FAST;
|
|
|
6863 else if (STRCMP(buf, "double") == 0)
|
|
|
6864 f = SPS_DOUBLE;
|
|
|
6865 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
6866 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
6867 f = -1;
|
|
|
6868
|
|
|
6869 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
6870 {
|
|
|
6871 sps_flags = SPS_BEST;
|
|
|
6872 return FAIL;
|
|
|
6873 }
|
|
|
6874 if (f != 0)
|
|
|
6875 sps_flags = f;
|
|
|
6876 }
|
|
|
6877
|
|
|
6878 if (sps_flags == 0)
|
|
|
6879 sps_flags = SPS_BEST;
|
|
|
6880
|
|
|
6881 return OK;
|
|
|
6882 }
|
|
|
6883
|
|
|
6884 /* Remember what "z?" replaced. */
|
|
|
6885 static char_u *repl_from = NULL;
|
|
|
6886 static char_u *repl_to = NULL;
|
|
|
6887
|
|
323
|
6888 /*
|
|
|
6889 * "z?": Find badly spelled word under or after the cursor.
|
|
|
6890 * Give suggestions for the properly spelled word.
|
|
|
6891 */
|
|
|
6892 void
|
|
|
6893 spell_suggest()
|
|
|
6894 {
|
|
|
6895 char_u *line;
|
|
|
6896 pos_T prev_cursor = curwin->w_cursor;
|
|
|
6897 char_u wcopy[MAXWLEN + 2];
|
|
|
6898 char_u *p;
|
|
|
6899 int i;
|
|
|
6900 int c;
|
|
|
6901 suginfo_T sug;
|
|
|
6902 suggest_T *stp;
|
|
372
|
6903 int mouse_used;
|
|
388
|
6904 int need_cap;
|
|
323
|
6905
|
|
344
|
6906 /* Find the start of the badly spelled word. */
|
|
346
|
6907 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
6908 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
6909 {
|
|
|
6910 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
6911 return;
|
|
|
6912
|
|
|
6913 /* No bad word or it starts after the cursor: use the word under the
|
|
|
6914 * cursor. */
|
|
|
6915 curwin->w_cursor = prev_cursor;
|
|
|
6916 line = ml_get_curline();
|
|
|
6917 p = line + curwin->w_cursor.col;
|
|
|
6918 /* Backup to before start of word. */
|
|
390
|
6919 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
6920 mb_ptr_back(line, p);
|
|
|
6921 /* Forward to start of word. */
|
|
390
|
6922 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
6923 mb_ptr_adv(p);
|
|
|
6924
|
|
390
|
6925 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
6926 {
|
|
|
6927 beep_flush();
|
|
|
6928 return;
|
|
|
6929 }
|
|
|
6930 curwin->w_cursor.col = p - line;
|
|
323
|
6931 }
|
|
|
6932
|
|
344
|
6933 /* Get the word and its length. */
|
|
|
6934
|
|
388
|
6935 /* Figure out if the word should be capitalised. */
|
|
475
|
6936 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
6937
|
|
|
6938 line = ml_get_curline();
|
|
388
|
6939
|
|
344
|
6940 /* Get the list of suggestions */
|
|
388
|
6941 spell_find_suggest(line + curwin->w_cursor.col, &sug, (int)Rows - 2,
|
|
|
6942 TRUE, need_cap);
|
|
323
|
6943
|
|
|
6944 if (sug.su_ga.ga_len == 0)
|
|
|
6945 MSG(_("Sorry, no suggestions"));
|
|
|
6946 else
|
|
|
6947 {
|
|
372
|
6948 vim_free(repl_from);
|
|
|
6949 repl_from = NULL;
|
|
|
6950 vim_free(repl_to);
|
|
|
6951 repl_to = NULL;
|
|
|
6952
|
|
474
|
6953 #ifdef FEAT_RIGHTLEFT
|
|
|
6954 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
6955 cmdmsg_rl = curwin->w_p_rl;
|
|
|
6956 if (cmdmsg_rl)
|
|
|
6957 msg_col = Columns - 1;
|
|
|
6958 #endif
|
|
|
6959
|
|
323
|
6960 /* List the suggestions. */
|
|
|
6961 msg_start();
|
|
372
|
6962 lines_left = Rows; /* avoid more prompt */
|
|
323
|
6963 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
6964 sug.su_badlen, sug.su_badptr);
|
|
474
|
6965 #ifdef FEAT_RIGHTLEFT
|
|
|
6966 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
6967 {
|
|
|
6968 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
6969 * untranslated message rightleft. */
|
|
|
6970 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
6971 sug.su_badlen, sug.su_badptr);
|
|
|
6972 }
|
|
|
6973 #endif
|
|
323
|
6974 msg_puts(IObuff);
|
|
|
6975 msg_clr_eos();
|
|
|
6976 msg_putchar('\n');
|
|
346
|
6977
|
|
323
|
6978 msg_scroll = TRUE;
|
|
|
6979 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
6980 {
|
|
344
|
6981 stp = &SUG(sug.su_ga, i);
|
|
323
|
6982
|
|
|
6983 /* The suggested word may replace only part of the bad word, add
|
|
|
6984 * the not replaced part. */
|
|
|
6985 STRCPY(wcopy, stp->st_word);
|
|
|
6986 if (sug.su_badlen > stp->st_orglen)
|
|
|
6987 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
6988 sug.su_badptr + stp->st_orglen,
|
|
|
6989 sug.su_badlen - stp->st_orglen);
|
|
474
|
6990 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
6991 #ifdef FEAT_RIGHTLEFT
|
|
|
6992 if (cmdmsg_rl)
|
|
|
6993 rl_mirror(IObuff);
|
|
|
6994 #endif
|
|
|
6995 msg_puts(IObuff);
|
|
|
6996
|
|
|
6997 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
6998 msg_puts(IObuff);
|
|
|
6999
|
|
|
7000 /* The word may replace more than "su_badlen". */
|
|
|
7001 if (sug.su_badlen < stp->st_orglen)
|
|
|
7002 {
|
|
|
7003 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
7004 stp->st_orglen, sug.su_badptr);
|
|
|
7005 msg_puts(IObuff);
|
|
|
7006 }
|
|
|
7007
|
|
324
|
7008 if (p_verbose > 0)
|
|
344
|
7009 {
|
|
346
|
7010 /* Add the score. */
|
|
351
|
7011 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
7012 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
7013 stp->st_salscore ? "s " : "",
|
|
|
7014 stp->st_score, stp->st_altscore);
|
|
|
7015 else
|
|
474
|
7016 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
7017 stp->st_score);
|
|
474
|
7018 #ifdef FEAT_RIGHTLEFT
|
|
|
7019 if (cmdmsg_rl)
|
|
|
7020 /* Mirror the numbers, but keep the leading space. */
|
|
|
7021 rl_mirror(IObuff + 1);
|
|
|
7022 #endif
|
|
346
|
7023 msg_advance(30);
|
|
|
7024 msg_puts(IObuff);
|
|
344
|
7025 }
|
|
323
|
7026 msg_putchar('\n');
|
|
|
7027 }
|
|
|
7028
|
|
474
|
7029 #ifdef FEAT_RIGHTLEFT
|
|
|
7030 cmdmsg_rl = FALSE;
|
|
|
7031 msg_col = 0;
|
|
|
7032 #endif
|
|
323
|
7033 /* Ask for choice. */
|
|
372
|
7034 i = prompt_for_number(&mouse_used);
|
|
|
7035 if (mouse_used)
|
|
|
7036 i -= lines_left;
|
|
|
7037
|
|
344
|
7038 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
7039 {
|
|
372
|
7040 /* Save the from and to text for :spellrepall. */
|
|
|
7041 stp = &SUG(sug.su_ga, i - 1);
|
|
|
7042 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
7043 repl_to = vim_strsave(stp->st_word);
|
|
|
7044
|
|
323
|
7045 /* Replace the word. */
|
|
|
7046 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
7047 if (p != NULL)
|
|
|
7048 {
|
|
|
7049 c = sug.su_badptr - line;
|
|
|
7050 mch_memmove(p, line, c);
|
|
|
7051 STRCPY(p + c, stp->st_word);
|
|
|
7052 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
7053 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
7054 curwin->w_cursor.col = c;
|
|
|
7055 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
7056
|
|
|
7057 /* For redo we use a change-word command. */
|
|
|
7058 ResetRedobuff();
|
|
|
7059 AppendToRedobuff((char_u *)"ciw");
|
|
|
7060 AppendToRedobuff(stp->st_word);
|
|
|
7061 AppendCharToRedobuff(ESC);
|
|
323
|
7062 }
|
|
|
7063 }
|
|
|
7064 else
|
|
|
7065 curwin->w_cursor = prev_cursor;
|
|
|
7066 }
|
|
|
7067
|
|
344
|
7068 spell_find_cleanup(&sug);
|
|
|
7069 }
|
|
|
7070
|
|
|
7071 /*
|
|
475
|
7072 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
7073 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
7074 */
|
|
|
7075 static int
|
|
|
7076 check_need_cap(lnum, col)
|
|
|
7077 linenr_T lnum;
|
|
|
7078 colnr_T col;
|
|
|
7079 {
|
|
|
7080 int need_cap = FALSE;
|
|
|
7081 char_u *line;
|
|
|
7082 char_u *line_copy = NULL;
|
|
|
7083 char_u *p;
|
|
|
7084 colnr_T endcol;
|
|
|
7085 regmatch_T regmatch;
|
|
|
7086
|
|
|
7087 if (curbuf->b_cap_prog == NULL)
|
|
|
7088 return FALSE;
|
|
|
7089
|
|
|
7090 line = ml_get_curline();
|
|
|
7091 endcol = 0;
|
|
|
7092 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
7093 {
|
|
|
7094 /* At start of line, check if previous line is empty or sentence
|
|
|
7095 * ends there. */
|
|
|
7096 if (lnum == 1)
|
|
|
7097 need_cap = TRUE;
|
|
|
7098 else
|
|
|
7099 {
|
|
|
7100 line = ml_get(lnum - 1);
|
|
|
7101 if (*skipwhite(line) == NUL)
|
|
|
7102 need_cap = TRUE;
|
|
|
7103 else
|
|
|
7104 {
|
|
|
7105 /* Append a space in place of the line break. */
|
|
|
7106 line_copy = concat_str(line, (char_u *)" ");
|
|
|
7107 line = line_copy;
|
|
|
7108 endcol = STRLEN(line);
|
|
|
7109 }
|
|
|
7110 }
|
|
|
7111 }
|
|
|
7112 else
|
|
|
7113 endcol = col;
|
|
|
7114
|
|
|
7115 if (endcol > 0)
|
|
|
7116 {
|
|
|
7117 /* Check if sentence ends before the bad word. */
|
|
|
7118 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
7119 regmatch.rm_ic = FALSE;
|
|
|
7120 p = line + endcol;
|
|
|
7121 for (;;)
|
|
|
7122 {
|
|
|
7123 mb_ptr_back(line, p);
|
|
|
7124 if (p == line || spell_iswordp_nmw(p))
|
|
|
7125 break;
|
|
|
7126 if (vim_regexec(®match, p, 0)
|
|
|
7127 && regmatch.endp[0] == line + endcol)
|
|
|
7128 {
|
|
|
7129 need_cap = TRUE;
|
|
|
7130 break;
|
|
|
7131 }
|
|
|
7132 }
|
|
|
7133 }
|
|
|
7134
|
|
|
7135 vim_free(line_copy);
|
|
|
7136
|
|
|
7137 return need_cap;
|
|
|
7138 }
|
|
|
7139
|
|
|
7140
|
|
|
7141 /*
|
|
372
|
7142 * ":spellrepall"
|
|
|
7143 */
|
|
|
7144 /*ARGSUSED*/
|
|
|
7145 void
|
|
|
7146 ex_spellrepall(eap)
|
|
|
7147 exarg_T *eap;
|
|
|
7148 {
|
|
|
7149 pos_T pos = curwin->w_cursor;
|
|
|
7150 char_u *frompat;
|
|
|
7151 int addlen;
|
|
|
7152 char_u *line;
|
|
|
7153 char_u *p;
|
|
|
7154 int didone = FALSE;
|
|
|
7155 int save_ws = p_ws;
|
|
|
7156
|
|
|
7157 if (repl_from == NULL || repl_to == NULL)
|
|
|
7158 {
|
|
|
7159 EMSG(_("E752: No previous spell replacement"));
|
|
|
7160 return;
|
|
|
7161 }
|
|
|
7162 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
7163
|
|
|
7164 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
7165 if (frompat == NULL)
|
|
|
7166 return;
|
|
|
7167 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
7168 p_ws = FALSE;
|
|
|
7169
|
|
|
7170 curwin->w_cursor.lnum = 0;
|
|
|
7171 while (!got_int)
|
|
|
7172 {
|
|
|
7173 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
7174 || u_save_cursor() == FAIL)
|
|
|
7175 break;
|
|
|
7176
|
|
|
7177 /* Only replace when the right word isn't there yet. This happens
|
|
|
7178 * when changing "etc" to "etc.". */
|
|
|
7179 line = ml_get_curline();
|
|
|
7180 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
7181 repl_to, STRLEN(repl_to)) != 0)
|
|
|
7182 {
|
|
|
7183 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
7184 if (p == NULL)
|
|
|
7185 break;
|
|
|
7186 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
7187 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
7188 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
7189 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
7190 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
7191 didone = TRUE;
|
|
|
7192 }
|
|
|
7193 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
7194 }
|
|
|
7195
|
|
|
7196 p_ws = save_ws;
|
|
|
7197 curwin->w_cursor = pos;
|
|
|
7198 vim_free(frompat);
|
|
|
7199
|
|
|
7200 if (!didone)
|
|
|
7201 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
|
7202 }
|
|
|
7203
|
|
|
7204 /*
|
|
344
|
7205 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
7206 * a list of allocated strings.
|
|
|
7207 */
|
|
|
7208 void
|
|
475
|
7209 spell_suggest_list(gap, word, maxcount, need_cap)
|
|
344
|
7210 garray_T *gap;
|
|
|
7211 char_u *word;
|
|
|
7212 int maxcount; /* maximum nr of suggestions */
|
|
475
|
7213 int need_cap; /* 'spellcapcheck' matched */
|
|
344
|
7214 {
|
|
|
7215 suginfo_T sug;
|
|
|
7216 int i;
|
|
|
7217 suggest_T *stp;
|
|
|
7218 char_u *wcopy;
|
|
|
7219
|
|
475
|
7220 spell_find_suggest(word, &sug, maxcount, FALSE, need_cap);
|
|
344
|
7221
|
|
|
7222 /* Make room in "gap". */
|
|
|
7223 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
7224 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
7225 return;
|
|
|
7226
|
|
323
|
7227 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
7228 {
|
|
|
7229 stp = &SUG(sug.su_ga, i);
|
|
|
7230
|
|
|
7231 /* The suggested word may replace only part of "word", add the not
|
|
|
7232 * replaced part. */
|
|
|
7233 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
7234 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
7235 if (wcopy == NULL)
|
|
|
7236 break;
|
|
|
7237 STRCPY(wcopy, stp->st_word);
|
|
|
7238 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
7239 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
7240 }
|
|
|
7241
|
|
|
7242 spell_find_cleanup(&sug);
|
|
|
7243 }
|
|
|
7244
|
|
|
7245 /*
|
|
|
7246 * Find spell suggestions for the word at the start of "badptr".
|
|
|
7247 * Return the suggestions in "su->su_ga".
|
|
|
7248 * The maximum number of suggestions is "maxcount".
|
|
|
7249 * Note: does use info for the current window.
|
|
|
7250 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
7251 */
|
|
|
7252 static void
|
|
388
|
7253 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
7254 char_u *badptr;
|
|
|
7255 suginfo_T *su;
|
|
|
7256 int maxcount;
|
|
358
|
7257 int banbadword; /* don't include badword in suggestions */
|
|
388
|
7258 int need_cap; /* word should start with capital */
|
|
344
|
7259 {
|
|
385
|
7260 int attr = 0;
|
|
372
|
7261 char_u buf[MAXPATHL];
|
|
|
7262 char_u *p;
|
|
|
7263 int do_combine = FALSE;
|
|
|
7264 char_u *sps_copy;
|
|
|
7265 #ifdef FEAT_EVAL
|
|
|
7266 static int expr_busy = FALSE;
|
|
|
7267 #endif
|
|
385
|
7268 int c;
|
|
344
|
7269
|
|
|
7270 /*
|
|
|
7271 * Set the info in "*su".
|
|
|
7272 */
|
|
|
7273 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
7274 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
7275 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
7276 if (*badptr == NUL)
|
|
|
7277 return;
|
|
344
|
7278 hash_init(&su->su_banned);
|
|
|
7279
|
|
|
7280 su->su_badptr = badptr;
|
|
385
|
7281 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
7282 su->su_maxcount = maxcount;
|
|
372
|
7283 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
7284
|
|
|
7285 if (su->su_badlen >= MAXWLEN)
|
|
|
7286 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
7287 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
7288 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
7289 su->su_fbadword, MAXWLEN);
|
|
346
|
7290 /* get caps flags for bad word */
|
|
474
|
7291 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
7292 su->su_badptr + su->su_badlen);
|
|
388
|
7293 if (need_cap)
|
|
|
7294 su->su_badflags |= WF_ONECAP;
|
|
344
|
7295
|
|
385
|
7296 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
7297 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
7298 * for that. */
|
|
455
|
7299 c = PTR2CHAR(su->su_badptr);
|
|
385
|
7300 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
7301 {
|
|
|
7302 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
7303 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
7304 0, TRUE);
|
|
|
7305 }
|
|
|
7306
|
|
344
|
7307 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
7308 if (banbadword)
|
|
|
7309 add_banned(su, su->su_badword);
|
|
344
|
7310
|
|
372
|
7311 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
7312 sps_copy = vim_strsave(p_sps);
|
|
|
7313 if (sps_copy == NULL)
|
|
|
7314 return;
|
|
|
7315
|
|
|
7316 /* Loop over the items in 'spellsuggest'. */
|
|
|
7317 for (p = sps_copy; *p != NUL; )
|
|
|
7318 {
|
|
|
7319 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
7320
|
|
|
7321 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
7322 {
|
|
|
7323 #ifdef FEAT_EVAL
|
|
375
|
7324 /* Evaluate an expression. Skip this when called recursively,
|
|
|
7325 * when using spellsuggest() in the expression. */
|
|
372
|
7326 if (!expr_busy)
|
|
|
7327 {
|
|
|
7328 expr_busy = TRUE;
|
|
|
7329 spell_suggest_expr(su, buf + 5);
|
|
|
7330 expr_busy = FALSE;
|
|
|
7331 }
|
|
|
7332 #endif
|
|
|
7333 }
|
|
|
7334 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
7335 /* Use list of suggestions in a file. */
|
|
|
7336 spell_suggest_file(su, buf + 5);
|
|
|
7337 else
|
|
|
7338 {
|
|
|
7339 /* Use internal method. */
|
|
|
7340 spell_suggest_intern(su);
|
|
|
7341 if (sps_flags & SPS_DOUBLE)
|
|
|
7342 do_combine = TRUE;
|
|
|
7343 }
|
|
|
7344 }
|
|
|
7345
|
|
|
7346 vim_free(sps_copy);
|
|
|
7347
|
|
|
7348 if (do_combine)
|
|
|
7349 /* Combine the two list of suggestions. This must be done last,
|
|
|
7350 * because sorting changes the order again. */
|
|
|
7351 score_combine(su);
|
|
|
7352 }
|
|
|
7353
|
|
|
7354 #ifdef FEAT_EVAL
|
|
|
7355 /*
|
|
|
7356 * Find suggestions by evaluating expression "expr".
|
|
|
7357 */
|
|
|
7358 static void
|
|
|
7359 spell_suggest_expr(su, expr)
|
|
|
7360 suginfo_T *su;
|
|
|
7361 char_u *expr;
|
|
|
7362 {
|
|
|
7363 list_T *list;
|
|
|
7364 listitem_T *li;
|
|
|
7365 int score;
|
|
|
7366 char_u *p;
|
|
|
7367
|
|
|
7368 /* The work is split up in a few parts to avoid having to export
|
|
|
7369 * suginfo_T.
|
|
|
7370 * First evaluate the expression and get the resulting list. */
|
|
|
7371 list = eval_spell_expr(su->su_badword, expr);
|
|
|
7372 if (list != NULL)
|
|
|
7373 {
|
|
|
7374 /* Loop over the items in the list. */
|
|
|
7375 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
7376 if (li->li_tv.v_type == VAR_LIST)
|
|
|
7377 {
|
|
|
7378 /* Get the word and the score from the items. */
|
|
|
7379 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
7380 if (score >= 0)
|
|
|
7381 add_suggestion(su, &su->su_ga, p,
|
|
|
7382 su->su_badlen, score, 0, TRUE);
|
|
|
7383 }
|
|
|
7384 list_unref(list);
|
|
|
7385 }
|
|
|
7386
|
|
|
7387 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
7388 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7389 }
|
|
|
7390 #endif
|
|
|
7391
|
|
|
7392 /*
|
|
474
|
7393 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
7394 */
|
|
|
7395 static void
|
|
|
7396 spell_suggest_file(su, fname)
|
|
|
7397 suginfo_T *su;
|
|
|
7398 char_u *fname;
|
|
|
7399 {
|
|
|
7400 FILE *fd;
|
|
|
7401 char_u line[MAXWLEN * 2];
|
|
|
7402 char_u *p;
|
|
|
7403 int len;
|
|
|
7404 char_u cword[MAXWLEN];
|
|
|
7405
|
|
|
7406 /* Open the file. */
|
|
|
7407 fd = mch_fopen((char *)fname, "r");
|
|
|
7408 if (fd == NULL)
|
|
|
7409 {
|
|
|
7410 EMSG2(_(e_notopen), fname);
|
|
|
7411 return;
|
|
|
7412 }
|
|
|
7413
|
|
|
7414 /* Read it line by line. */
|
|
|
7415 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
7416 {
|
|
|
7417 line_breakcheck();
|
|
|
7418
|
|
|
7419 p = vim_strchr(line, '/');
|
|
|
7420 if (p == NULL)
|
|
|
7421 continue; /* No Tab found, just skip the line. */
|
|
|
7422 *p++ = NUL;
|
|
|
7423 if (STRICMP(su->su_badword, line) == 0)
|
|
|
7424 {
|
|
|
7425 /* Match! Isolate the good word, until CR or NL. */
|
|
|
7426 for (len = 0; p[len] >= ' '; ++len)
|
|
|
7427 ;
|
|
|
7428 p[len] = NUL;
|
|
|
7429
|
|
|
7430 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
7431 * of the bad word. */
|
|
|
7432 if (captype(p, NULL) == 0)
|
|
|
7433 {
|
|
|
7434 make_case_word(p, cword, su->su_badflags);
|
|
|
7435 p = cword;
|
|
|
7436 }
|
|
|
7437
|
|
|
7438 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
7439 SCORE_FILE, 0, TRUE);
|
|
|
7440 }
|
|
|
7441 }
|
|
|
7442
|
|
|
7443 fclose(fd);
|
|
|
7444
|
|
|
7445 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
7446 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7447 }
|
|
|
7448
|
|
|
7449 /*
|
|
|
7450 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
7451 */
|
|
|
7452 static void
|
|
|
7453 spell_suggest_intern(su)
|
|
|
7454 suginfo_T *su;
|
|
|
7455 {
|
|
344
|
7456 /*
|
|
346
|
7457 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
7458 *
|
|
|
7459 * Set a maximum score to limit the combination of operations that is
|
|
|
7460 * tried.
|
|
|
7461 */
|
|
346
|
7462 suggest_try_special(su);
|
|
|
7463
|
|
|
7464 /*
|
|
|
7465 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
7466 * from the .aff file and inserting a space (split the word).
|
|
|
7467 */
|
|
|
7468 suggest_try_change(su);
|
|
344
|
7469
|
|
|
7470 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
7471 if (sps_flags & SPS_DOUBLE)
|
|
|
7472 score_comp_sal(su);
|
|
|
7473
|
|
|
7474 /*
|
|
346
|
7475 * 3. Try finding sound-a-like words.
|
|
344
|
7476 *
|
|
|
7477 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
7478 * very slow and often doesn't find new suggestions.
|
|
|
7479 */
|
|
|
7480 if ((sps_flags & SPS_DOUBLE)
|
|
|
7481 || (!(sps_flags & SPS_FAST)
|
|
|
7482 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
7483 {
|
|
|
7484 /* Allow a higher score now. */
|
|
|
7485 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
7486 suggest_try_soundalike(su);
|
|
344
|
7487 }
|
|
|
7488
|
|
|
7489 /* When CTRL-C was hit while searching do show the results. */
|
|
|
7490 ui_breakcheck();
|
|
|
7491 if (got_int)
|
|
|
7492 {
|
|
|
7493 (void)vgetc();
|
|
|
7494 got_int = FALSE;
|
|
|
7495 }
|
|
|
7496
|
|
372
|
7497 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
7498 {
|
|
|
7499 if (sps_flags & SPS_BEST)
|
|
|
7500 /* Adjust the word score for how it sounds like. */
|
|
|
7501 rescore_suggestions(su);
|
|
|
7502
|
|
|
7503 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
7504 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
7505 }
|
|
|
7506 }
|
|
|
7507
|
|
|
7508 /*
|
|
|
7509 * Free the info put in "*su" by spell_find_suggest().
|
|
|
7510 */
|
|
|
7511 static void
|
|
|
7512 spell_find_cleanup(su)
|
|
|
7513 suginfo_T *su;
|
|
|
7514 {
|
|
|
7515 int i;
|
|
|
7516
|
|
|
7517 /* Free the suggestions. */
|
|
|
7518 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7519 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
7520 ga_clear(&su->su_ga);
|
|
|
7521 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
7522 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
7523 ga_clear(&su->su_sga);
|
|
323
|
7524
|
|
|
7525 /* Free the banned words. */
|
|
344
|
7526 free_banned(su);
|
|
323
|
7527 }
|
|
|
7528
|
|
|
7529 /*
|
|
324
|
7530 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
7531 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
7532 * The result is NUL terminated.
|
|
323
|
7533 */
|
|
|
7534 static void
|
|
324
|
7535 onecap_copy(word, wcopy, upper)
|
|
323
|
7536 char_u *word;
|
|
|
7537 char_u *wcopy;
|
|
|
7538 int upper; /* TRUE: first letter made upper case */
|
|
|
7539 {
|
|
|
7540 char_u *p;
|
|
|
7541 int c;
|
|
|
7542 int l;
|
|
|
7543
|
|
|
7544 p = word;
|
|
|
7545 #ifdef FEAT_MBYTE
|
|
|
7546 if (has_mbyte)
|
|
474
|
7547 c = mb_cptr2char_adv(&p);
|
|
323
|
7548 else
|
|
|
7549 #endif
|
|
|
7550 c = *p++;
|
|
|
7551 if (upper)
|
|
324
|
7552 c = SPELL_TOUPPER(c);
|
|
323
|
7553 else
|
|
324
|
7554 c = SPELL_TOFOLD(c);
|
|
323
|
7555 #ifdef FEAT_MBYTE
|
|
|
7556 if (has_mbyte)
|
|
|
7557 l = mb_char2bytes(c, wcopy);
|
|
|
7558 else
|
|
|
7559 #endif
|
|
|
7560 {
|
|
|
7561 l = 1;
|
|
|
7562 wcopy[0] = c;
|
|
|
7563 }
|
|
376
|
7564 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
7565 }
|
|
|
7566
|
|
|
7567 /*
|
|
324
|
7568 * Make a copy of "word" with all the letters upper cased into
|
|
|
7569 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
7570 */
|
|
|
7571 static void
|
|
|
7572 allcap_copy(word, wcopy)
|
|
|
7573 char_u *word;
|
|
|
7574 char_u *wcopy;
|
|
|
7575 {
|
|
|
7576 char_u *s;
|
|
|
7577 char_u *d;
|
|
|
7578 int c;
|
|
|
7579
|
|
|
7580 d = wcopy;
|
|
|
7581 for (s = word; *s != NUL; )
|
|
|
7582 {
|
|
|
7583 #ifdef FEAT_MBYTE
|
|
|
7584 if (has_mbyte)
|
|
474
|
7585 c = mb_cptr2char_adv(&s);
|
|
323
|
7586 else
|
|
|
7587 #endif
|
|
|
7588 c = *s++;
|
|
324
|
7589 c = SPELL_TOUPPER(c);
|
|
323
|
7590
|
|
|
7591 #ifdef FEAT_MBYTE
|
|
|
7592 if (has_mbyte)
|
|
|
7593 {
|
|
|
7594 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
7595 break;
|
|
|
7596 d += mb_char2bytes(c, d);
|
|
|
7597 }
|
|
|
7598 else
|
|
|
7599 #endif
|
|
|
7600 {
|
|
|
7601 if (d - wcopy >= MAXWLEN - 1)
|
|
|
7602 break;
|
|
|
7603 *d++ = c;
|
|
|
7604 }
|
|
|
7605 }
|
|
|
7606 *d = NUL;
|
|
|
7607 }
|
|
|
7608
|
|
|
7609 /*
|
|
346
|
7610 * Try finding suggestions by recognizing specific situations.
|
|
|
7611 */
|
|
|
7612 static void
|
|
|
7613 suggest_try_special(su)
|
|
|
7614 suginfo_T *su;
|
|
|
7615 {
|
|
|
7616 char_u *p;
|
|
376
|
7617 size_t len;
|
|
346
|
7618 int c;
|
|
|
7619 char_u word[MAXWLEN];
|
|
|
7620
|
|
|
7621 /*
|
|
|
7622 * Recognize a word that is repeated: "the the".
|
|
|
7623 */
|
|
|
7624 p = skiptowhite(su->su_fbadword);
|
|
|
7625 len = p - su->su_fbadword;
|
|
|
7626 p = skipwhite(p);
|
|
|
7627 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
7628 {
|
|
|
7629 /* Include badflags: if the badword is onecap or allcap
|
|
|
7630 * use that for the goodword too: "The the" -> "The". */
|
|
|
7631 c = su->su_fbadword[len];
|
|
|
7632 su->su_fbadword[len] = NUL;
|
|
|
7633 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
7634 su->su_fbadword[len] = c;
|
|
351
|
7635 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
7636 }
|
|
|
7637 }
|
|
|
7638
|
|
|
7639 /*
|
|
323
|
7640 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
7641 *
|
|
|
7642 * This uses a state machine. At each node in the tree we try various
|
|
|
7643 * operations. When trying if an operation work "depth" is increased and the
|
|
|
7644 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
7645 * character, replace one and delete another. The number of changes is
|
|
|
7646 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
7647 */
|
|
|
7648 static void
|
|
346
|
7649 suggest_try_change(su)
|
|
323
|
7650 suginfo_T *su;
|
|
|
7651 {
|
|
|
7652 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
7653 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
7654 trystate_T stack[MAXWLEN];
|
|
|
7655 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
|
7656 * to for word split) */
|
|
|
7657 char_u prewordlen = 0; /* length of word in "preword" */
|
|
|
7658 int splitoff = 0; /* index in tword after last split */
|
|
|
7659 trystate_T *sp;
|
|
|
7660 int newscore;
|
|
|
7661 langp_T *lp;
|
|
375
|
7662 char_u *byts, *fbyts, *pbyts;
|
|
|
7663 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
7664 int depth;
|
|
330
|
7665 int c, c2, c3;
|
|
390
|
7666 int n;
|
|
323
|
7667 int flags;
|
|
|
7668 garray_T *gap;
|
|
324
|
7669 idx_T arridx;
|
|
323
|
7670 int len;
|
|
|
7671 char_u *p;
|
|
|
7672 fromto_T *ftp;
|
|
330
|
7673 int fl = 0, tl;
|
|
346
|
7674 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
482
|
7675 slang_T *slang;
|
|
|
7676 int fword_ends;
|
|
323
|
7677
|
|
|
7678 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
7679 * to find matches (esp. REP items). Append some more text, changing
|
|
|
7680 * chars after the bad word may help. */
|
|
323
|
7681 STRCPY(fword, su->su_fbadword);
|
|
346
|
7682 n = STRLEN(fword);
|
|
|
7683 p = su->su_badptr + su->su_badlen;
|
|
|
7684 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
7685
|
|
|
7686 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7687 lp->lp_slang != NULL; ++lp)
|
|
|
7688 {
|
|
482
|
7689 slang = lp->lp_slang;
|
|
|
7690
|
|
323
|
7691 /*
|
|
|
7692 * Go through the whole case-fold tree, try changes at each node.
|
|
|
7693 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
7694 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
7695 * word).
|
|
|
7696 */
|
|
|
7697 depth = 0;
|
|
375
|
7698 sp = &stack[0];
|
|
|
7699 sp->ts_state = STATE_START;
|
|
|
7700 sp->ts_score = 0;
|
|
|
7701 sp->ts_curi = 1;
|
|
|
7702 sp->ts_fidx = 0;
|
|
|
7703 sp->ts_fidxtry = 0;
|
|
|
7704 sp->ts_twordlen = 0;
|
|
|
7705 sp->ts_arridx = 0;
|
|
330
|
7706 #ifdef FEAT_MBYTE
|
|
375
|
7707 sp->ts_tcharlen = 0;
|
|
330
|
7708 #endif
|
|
|
7709
|
|
|
7710 /*
|
|
375
|
7711 * When there are postponed prefixes we need to use these first. At
|
|
|
7712 * the end of the prefix we continue in the case-fold tree.
|
|
|
7713 */
|
|
482
|
7714 fbyts = slang->sl_fbyts;
|
|
|
7715 fidxs = slang->sl_fidxs;
|
|
|
7716 pbyts = slang->sl_pbyts;
|
|
|
7717 pidxs = slang->sl_pidxs;
|
|
375
|
7718 if (pbyts != NULL)
|
|
|
7719 {
|
|
|
7720 byts = pbyts;
|
|
|
7721 idxs = pidxs;
|
|
482
|
7722 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
375
|
7723 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
7724 }
|
|
|
7725 else
|
|
|
7726 {
|
|
|
7727 byts = fbyts;
|
|
|
7728 idxs = fidxs;
|
|
482
|
7729 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
375
|
7730 }
|
|
|
7731
|
|
|
7732 /*
|
|
330
|
7733 * Loop to find all suggestions. At each round we either:
|
|
|
7734 * - For the current state try one operation, advance "ts_curi",
|
|
|
7735 * increase "depth".
|
|
|
7736 * - When a state is done go to the next, set "ts_state".
|
|
|
7737 * - When all states are tried decrease "depth".
|
|
|
7738 */
|
|
323
|
7739 while (depth >= 0 && !got_int)
|
|
|
7740 {
|
|
|
7741 sp = &stack[depth];
|
|
|
7742 switch (sp->ts_state)
|
|
|
7743 {
|
|
|
7744 case STATE_START:
|
|
375
|
7745 case STATE_NOPREFIX:
|
|
323
|
7746 /*
|
|
|
7747 * Start of node: Deal with NUL bytes, which means
|
|
|
7748 * tword[] may end here.
|
|
|
7749 */
|
|
|
7750 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
7751 len = byts[arridx]; /* bytes in this node */
|
|
|
7752 arridx += sp->ts_curi; /* index of current byte */
|
|
|
7753
|
|
482
|
7754 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
7755 {
|
|
|
7756 /* Skip over the NUL bytes, we use them later. */
|
|
|
7757 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
7758 ;
|
|
|
7759 sp->ts_curi += n;
|
|
|
7760
|
|
390
|
7761 /* Always past NUL bytes now. */
|
|
|
7762 n = (int)sp->ts_state;
|
|
|
7763 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7764 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
7765
|
|
375
|
7766 /* At end of a prefix or at start of prefixtree: check for
|
|
|
7767 * following word. */
|
|
390
|
7768 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
7769 {
|
|
455
|
7770 /* Set su->su_badflags to the caps type at this
|
|
|
7771 * position. Use the caps type until here for the
|
|
|
7772 * prefix itself. */
|
|
|
7773 #ifdef FEAT_MBYTE
|
|
|
7774 if (has_mbyte)
|
|
|
7775 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
7776 else
|
|
|
7777 #endif
|
|
|
7778 n = sp->ts_fidx;
|
|
474
|
7779 flags = badword_captype(su->su_badptr,
|
|
|
7780 su->su_badptr + n);
|
|
|
7781 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
7782 su->su_badptr + su->su_badlen);
|
|
375
|
7783 ++depth;
|
|
|
7784 stack[depth] = stack[depth - 1];
|
|
|
7785 sp = &stack[depth];
|
|
|
7786 sp->ts_prefixdepth = depth - 1;
|
|
|
7787 byts = fbyts;
|
|
|
7788 idxs = fidxs;
|
|
|
7789 sp->ts_state = STATE_START;
|
|
|
7790 sp->ts_curi = 1; /* start just after length byte */
|
|
|
7791 sp->ts_arridx = 0;
|
|
|
7792
|
|
455
|
7793 /* Move the prefix to preword[] with the right case
|
|
|
7794 * and make find_keepcap_word() works. */
|
|
|
7795 splitoff = sp->ts_twordlen;
|
|
|
7796 tword[splitoff] = NUL;
|
|
|
7797 make_case_word(tword, preword, flags);
|
|
|
7798 prewordlen = STRLEN(preword);
|
|
375
|
7799 }
|
|
|
7800 break;
|
|
|
7801 }
|
|
|
7802
|
|
346
|
7803 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
7804 {
|
|
|
7805 /* Past bytes in node and/or past NUL bytes. */
|
|
|
7806 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7807 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7808 break;
|
|
|
7809 }
|
|
|
7810
|
|
|
7811 /*
|
|
|
7812 * End of word in tree.
|
|
|
7813 */
|
|
|
7814 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
7815
|
|
324
|
7816 flags = (int)idxs[arridx];
|
|
323
|
7817
|
|
482
|
7818 if (sp->ts_prefixdepth == PFD_COMPOUND)
|
|
|
7819 {
|
|
|
7820 /* There was a compound word before this word. If this
|
|
|
7821 * word does not support compounding then give up
|
|
|
7822 * (splitting is tried for the word without compound
|
|
|
7823 * flag). */
|
|
|
7824 if (sp->ts_twordlen - splitoff < slang->sl_compminlen
|
|
|
7825 || !can_compound(slang, flags))
|
|
|
7826 break;
|
|
|
7827 }
|
|
|
7828 else if (sp->ts_prefixdepth < MAXWLEN)
|
|
375
|
7829 {
|
|
|
7830 /* There was a prefix before the word. Check that the
|
|
|
7831 * prefix can be used with this word. */
|
|
|
7832 /* Count the length of the NULs in the prefix. If there
|
|
|
7833 * are none this must be the first try without a prefix.
|
|
|
7834 */
|
|
|
7835 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
7836 len = pbyts[n++];
|
|
|
7837 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
7838 ;
|
|
|
7839 if (c > 0)
|
|
|
7840 {
|
|
390
|
7841 c = valid_word_prefix(c, n, flags,
|
|
482
|
7842 tword + splitoff, slang, FALSE);
|
|
375
|
7843 if (c == 0)
|
|
|
7844 break;
|
|
|
7845
|
|
|
7846 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
7847 if (c & WF_RAREPFX)
|
|
|
7848 flags |= WF_RARE;
|
|
|
7849 }
|
|
|
7850 }
|
|
|
7851
|
|
323
|
7852 /*
|
|
|
7853 * Form the word with proper case in preword.
|
|
|
7854 * If there is a word from a previous split, append.
|
|
|
7855 */
|
|
|
7856 tword[sp->ts_twordlen] = NUL;
|
|
|
7857 if (flags & WF_KEEPCAP)
|
|
|
7858 /* Must find the word in the keep-case tree. */
|
|
482
|
7859 find_keepcap_word(slang, tword + splitoff,
|
|
323
|
7860 preword + prewordlen);
|
|
|
7861 else
|
|
346
|
7862 {
|
|
323
|
7863 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
7864 * use that for the goodword too. But if the badword is
|
|
|
7865 * allcap and it's only one char long use onecap. */
|
|
|
7866 c = su->su_badflags;
|
|
|
7867 if ((c & WF_ALLCAP)
|
|
|
7868 #ifdef FEAT_MBYTE
|
|
474
|
7869 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
7870 #else
|
|
|
7871 && su->su_badlen == 1
|
|
|
7872 #endif
|
|
|
7873 )
|
|
|
7874 c = WF_ONECAP;
|
|
323
|
7875 make_case_word(tword + splitoff,
|
|
346
|
7876 preword + prewordlen, flags | c);
|
|
|
7877 }
|
|
323
|
7878
|
|
|
7879 /* Don't use a banned word. It may appear again as a good
|
|
|
7880 * word, thus remember it. */
|
|
|
7881 if (flags & WF_BANNED)
|
|
|
7882 {
|
|
|
7883 add_banned(su, preword + prewordlen);
|
|
|
7884 break;
|
|
|
7885 }
|
|
372
|
7886 if (was_banned(su, preword + prewordlen)
|
|
|
7887 || was_banned(su, preword))
|
|
323
|
7888 break;
|
|
|
7889
|
|
|
7890 newscore = 0;
|
|
|
7891 if ((flags & WF_REGION)
|
|
390
|
7892 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
7893 newscore += SCORE_REGION;
|
|
|
7894 if (flags & WF_RARE)
|
|
|
7895 newscore += SCORE_RARE;
|
|
|
7896
|
|
346
|
7897 if (!spell_valid_case(su->su_badflags,
|
|
323
|
7898 captype(preword + prewordlen, NULL)))
|
|
|
7899 newscore += SCORE_ICASE;
|
|
|
7900
|
|
482
|
7901 fword_ends = (fword[sp->ts_fidx] == NUL
|
|
|
7902 || !spell_iswordp(fword + sp->ts_fidx, curbuf));
|
|
|
7903 if (fword_ends && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
7904 {
|
|
366
|
7905 /* The badword also ends: add suggestions. Give a penalty
|
|
|
7906 * when changing non-word char to word char, e.g., "thes,"
|
|
|
7907 * -> "these". */
|
|
|
7908 p = fword + sp->ts_fidx;
|
|
|
7909 #ifdef FEAT_MBYTE
|
|
|
7910 if (has_mbyte)
|
|
|
7911 mb_ptr_back(fword, p);
|
|
|
7912 else
|
|
|
7913 #endif
|
|
|
7914 --p;
|
|
376
|
7915 if (!spell_iswordp(p, curbuf))
|
|
366
|
7916 {
|
|
|
7917 p = preword + STRLEN(preword);
|
|
|
7918 #ifdef FEAT_MBYTE
|
|
|
7919 if (has_mbyte)
|
|
|
7920 mb_ptr_back(preword, p);
|
|
|
7921 else
|
|
|
7922 #endif
|
|
|
7923 --p;
|
|
376
|
7924 if (spell_iswordp(p, curbuf))
|
|
366
|
7925 newscore += SCORE_NONWORD;
|
|
|
7926 }
|
|
|
7927
|
|
344
|
7928 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
7929 sp->ts_fidx - repextra,
|
|
351
|
7930 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
7931 }
|
|
482
|
7932 else if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
|
|
330
|
7933 #ifdef FEAT_MBYTE
|
|
|
7934 /* Don't split halfway a character. */
|
|
|
7935 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
7936 #endif
|
|
|
7937 )
|
|
323
|
7938 {
|
|
482
|
7939 int try_compound;
|
|
|
7940
|
|
|
7941 /* Get here in two situations:
|
|
|
7942 * 1. The word in the tree ends but the badword continues:
|
|
|
7943 * If the word allows compounding try that. Otherwise
|
|
|
7944 * try a split by inserting a space. For both check
|
|
|
7945 * that a valid words starts at fword[sp->ts_fidx].
|
|
|
7946 * 2. The badword does end, but it was due to a change
|
|
|
7947 * (e.g., a swap). No need to split, but do check that
|
|
|
7948 * the following word is valid.
|
|
|
7949 */
|
|
|
7950 if (!fword_ends
|
|
|
7951 && spell_iswordp(fword + sp->ts_fidx, curbuf)
|
|
|
7952 && sp->ts_twordlen - splitoff
|
|
|
7953 >= slang->sl_compminlen
|
|
|
7954 && can_compound(slang, flags))
|
|
|
7955 try_compound = TRUE;
|
|
|
7956 else
|
|
|
7957 {
|
|
|
7958 try_compound = FALSE;
|
|
|
7959 if (!fword_ends)
|
|
|
7960 newscore += SCORE_SPLIT;
|
|
|
7961 }
|
|
|
7962
|
|
|
7963 if (try_deeper(su, stack, depth, newscore))
|
|
323
|
7964 {
|
|
|
7965 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
|
7966 sp->ts_save_prewordlen = prewordlen;
|
|
346
|
7967 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7968 sp->ts_save_splitoff = splitoff;
|
|
376
|
7969 sp->ts_state = STATE_SPLITUNDO;
|
|
|
7970
|
|
|
7971 ++depth;
|
|
|
7972 sp = &stack[depth];
|
|
323
|
7973
|
|
482
|
7974 /* Append a space to preword when splitting. */
|
|
|
7975 if (!try_compound && !fword_ends)
|
|
|
7976 STRCAT(preword, " ");
|
|
323
|
7977 prewordlen = STRLEN(preword);
|
|
|
7978 splitoff = sp->ts_twordlen;
|
|
376
|
7979
|
|
|
7980 /* If the badword has a non-word character at this
|
|
|
7981 * position skip it. That means replacing the
|
|
482
|
7982 * non-word character with a space. Always skip a
|
|
|
7983 * character when the word ends. */
|
|
|
7984 if ((!try_compound
|
|
|
7985 && !spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
7986 || fword_ends)
|
|
376
|
7987 {
|
|
482
|
7988 int l;
|
|
|
7989
|
|
376
|
7990 #ifdef FEAT_MBYTE
|
|
|
7991 if (has_mbyte)
|
|
482
|
7992 l = MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
376
|
7993 else
|
|
|
7994 #endif
|
|
482
|
7995 l = 1;
|
|
|
7996 if (fword_ends)
|
|
|
7997 {
|
|
|
7998 /* Copy the skipped character to preword. */
|
|
|
7999 mch_memmove(preword + prewordlen,
|
|
|
8000 fword + sp->ts_fidx, l);
|
|
|
8001 prewordlen += l;
|
|
|
8002 preword[prewordlen] = NUL;
|
|
|
8003 }
|
|
|
8004 else
|
|
|
8005 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
8006 sp->ts_fidx += l;
|
|
376
|
8007 }
|
|
455
|
8008
|
|
482
|
8009 /* set flag to check compound flag on following word */
|
|
|
8010 if (try_compound)
|
|
|
8011 sp->ts_prefixdepth = PFD_COMPOUND;
|
|
|
8012 else
|
|
|
8013 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
|
8014
|
|
455
|
8015 /* set su->su_badflags to the caps type at this
|
|
|
8016 * position */
|
|
324
|
8017 #ifdef FEAT_MBYTE
|
|
|
8018 if (has_mbyte)
|
|
455
|
8019 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
8020 else
|
|
|
8021 #endif
|
|
455
|
8022 n = sp->ts_fidx;
|
|
474
|
8023 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
8024 su->su_badptr + su->su_badlen);
|
|
323
|
8025
|
|
|
8026 /* Restart at top of the tree. */
|
|
376
|
8027 sp->ts_arridx = 0;
|
|
323
|
8028 }
|
|
|
8029 }
|
|
|
8030 break;
|
|
|
8031
|
|
|
8032 case STATE_SPLITUNDO:
|
|
482
|
8033 /* Undo the changes done for word split or compound word. */
|
|
346
|
8034 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
8035 splitoff = sp->ts_save_splitoff;
|
|
482
|
8036 prewordlen = sp->ts_save_prewordlen;
|
|
323
|
8037
|
|
|
8038 /* Continue looking for NUL bytes. */
|
|
|
8039 sp->ts_state = STATE_START;
|
|
|
8040 break;
|
|
|
8041
|
|
|
8042 case STATE_ENDNUL:
|
|
|
8043 /* Past the NUL bytes in the node. */
|
|
455
|
8044 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
8045 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
8046 {
|
|
|
8047 /* The badword ends, can't use the bytes in this node. */
|
|
|
8048 sp->ts_state = STATE_DEL;
|
|
|
8049 break;
|
|
|
8050 }
|
|
|
8051 sp->ts_state = STATE_PLAIN;
|
|
|
8052 /*FALLTHROUGH*/
|
|
|
8053
|
|
|
8054 case STATE_PLAIN:
|
|
|
8055 /*
|
|
|
8056 * Go over all possible bytes at this node, add each to
|
|
|
8057 * tword[] and use child node. "ts_curi" is the index.
|
|
|
8058 */
|
|
|
8059 arridx = sp->ts_arridx;
|
|
|
8060 if (sp->ts_curi > byts[arridx])
|
|
|
8061 {
|
|
|
8062 /* Done all bytes at this node, do next state. When still
|
|
|
8063 * at already changed bytes skip the other tricks. */
|
|
|
8064 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
8065 sp->ts_state = STATE_DEL;
|
|
|
8066 else
|
|
|
8067 sp->ts_state = STATE_FINAL;
|
|
|
8068 }
|
|
|
8069 else
|
|
|
8070 {
|
|
|
8071 arridx += sp->ts_curi++;
|
|
|
8072 c = byts[arridx];
|
|
|
8073
|
|
|
8074 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
8075 * the byte in the bad word adjust the score. But don't
|
|
|
8076 * even try when the byte was already changed. */
|
|
330
|
8077 if (c == fword[sp->ts_fidx]
|
|
|
8078 #ifdef FEAT_MBYTE
|
|
|
8079 || (sp->ts_tcharlen > 0
|
|
|
8080 && sp->ts_isdiff != DIFF_NONE)
|
|
|
8081 #endif
|
|
|
8082 )
|
|
323
|
8083 newscore = 0;
|
|
|
8084 else
|
|
|
8085 newscore = SCORE_SUBST;
|
|
|
8086 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
8087 && try_deeper(su, stack, depth, newscore))
|
|
|
8088 {
|
|
|
8089 ++depth;
|
|
330
|
8090 sp = &stack[depth];
|
|
|
8091 ++sp->ts_fidx;
|
|
|
8092 tword[sp->ts_twordlen++] = c;
|
|
|
8093 sp->ts_arridx = idxs[arridx];
|
|
|
8094 #ifdef FEAT_MBYTE
|
|
|
8095 if (newscore == SCORE_SUBST)
|
|
|
8096 sp->ts_isdiff = DIFF_YES;
|
|
|
8097 if (has_mbyte)
|
|
|
8098 {
|
|
|
8099 /* Multi-byte characters are a bit complicated to
|
|
|
8100 * handle: They differ when any of the bytes
|
|
|
8101 * differ and then their length may also differ. */
|
|
|
8102 if (sp->ts_tcharlen == 0)
|
|
|
8103 {
|
|
|
8104 /* First byte. */
|
|
|
8105 sp->ts_tcharidx = 0;
|
|
|
8106 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
8107 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
8108 sp->ts_isdiff = (newscore != 0)
|
|
|
8109 ? DIFF_YES : DIFF_NONE;
|
|
|
8110 }
|
|
|
8111 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
8112 /* When inserting trail bytes don't advance in
|
|
|
8113 * the bad word. */
|
|
|
8114 --sp->ts_fidx;
|
|
|
8115 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
8116 {
|
|
|
8117 /* Last byte of character. */
|
|
|
8118 if (sp->ts_isdiff == DIFF_YES)
|
|
|
8119 {
|
|
|
8120 /* Correct ts_fidx for the byte length of
|
|
|
8121 * the character (we didn't check that
|
|
|
8122 * before). */
|
|
|
8123 sp->ts_fidx = sp->ts_fcharstart
|
|
|
8124 + MB_BYTE2LEN(
|
|
|
8125 fword[sp->ts_fcharstart]);
|
|
|
8126
|
|
479
|
8127 /* For changing a composing character
|
|
|
8128 * adjust the score from SCORE_SUBST to
|
|
|
8129 * SCORE_SUBCOMP. */
|
|
|
8130 if (enc_utf8
|
|
|
8131 && utf_iscomposing(
|
|
|
8132 mb_ptr2char(tword
|
|
|
8133 + sp->ts_twordlen
|
|
|
8134 - sp->ts_tcharlen))
|
|
|
8135 && utf_iscomposing(
|
|
|
8136 mb_ptr2char(fword
|
|
|
8137 + sp->ts_fcharstart)))
|
|
|
8138 sp->ts_score -=
|
|
|
8139 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
8140
|
|
330
|
8141 /* For a similar character adjust score
|
|
|
8142 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
482
|
8143 else if (slang->sl_has_map
|
|
|
8144 && similar_chars(slang,
|
|
330
|
8145 mb_ptr2char(tword
|
|
|
8146 + sp->ts_twordlen
|
|
|
8147 - sp->ts_tcharlen),
|
|
|
8148 mb_ptr2char(fword
|
|
|
8149 + sp->ts_fcharstart)))
|
|
|
8150 sp->ts_score -=
|
|
|
8151 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
8152 }
|
|
358
|
8153 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
8154 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
8155 {
|
|
|
8156 p = tword + sp->ts_twordlen
|
|
|
8157 - sp->ts_tcharlen;
|
|
|
8158 c = mb_ptr2char(p);
|
|
475
|
8159 if (enc_utf8 && utf_iscomposing(c))
|
|
|
8160 {
|
|
|
8161 /* Inserting a composing char doesn't
|
|
|
8162 * count that much. */
|
|
358
|
8163 sp->ts_score -= SCORE_INS
|
|
475
|
8164 - SCORE_INSCOMP;
|
|
|
8165 }
|
|
|
8166 else
|
|
|
8167 {
|
|
|
8168 /* If the previous character was the
|
|
|
8169 * same, thus doubling a character,
|
|
|
8170 * give a bonus to the score. */
|
|
|
8171 mb_ptr_back(tword, p);
|
|
|
8172 if (c == mb_ptr2char(p))
|
|
|
8173 sp->ts_score -= SCORE_INS
|
|
358
|
8174 - SCORE_INSDUP;
|
|
475
|
8175 }
|
|
358
|
8176 }
|
|
330
|
8177
|
|
|
8178 /* Starting a new char, reset the length. */
|
|
|
8179 sp->ts_tcharlen = 0;
|
|
|
8180 }
|
|
|
8181 }
|
|
|
8182 else
|
|
|
8183 #endif
|
|
|
8184 {
|
|
|
8185 /* If we found a similar char adjust the score.
|
|
|
8186 * We do this after calling try_deeper() because
|
|
|
8187 * it's slow. */
|
|
|
8188 if (newscore != 0
|
|
482
|
8189 && slang->sl_has_map
|
|
|
8190 && similar_chars(slang,
|
|
330
|
8191 c, fword[sp->ts_fidx - 1]))
|
|
|
8192 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
8193 }
|
|
323
|
8194 }
|
|
|
8195 }
|
|
|
8196 break;
|
|
|
8197
|
|
|
8198 case STATE_DEL:
|
|
330
|
8199 #ifdef FEAT_MBYTE
|
|
|
8200 /* When past the first byte of a multi-byte char don't try
|
|
|
8201 * delete/insert/swap a character. */
|
|
|
8202 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
8203 {
|
|
|
8204 sp->ts_state = STATE_FINAL;
|
|
|
8205 break;
|
|
|
8206 }
|
|
|
8207 #endif
|
|
|
8208 /*
|
|
|
8209 * Try skipping one character in the bad word (delete it).
|
|
|
8210 */
|
|
323
|
8211 sp->ts_state = STATE_INS;
|
|
|
8212 sp->ts_curi = 1;
|
|
|
8213 if (fword[sp->ts_fidx] != NUL
|
|
|
8214 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
8215 {
|
|
|
8216 ++depth;
|
|
358
|
8217
|
|
|
8218 /* Advance over the character in fword[]. Give a bonus to
|
|
|
8219 * the score if the same character is following "nn" ->
|
|
|
8220 * "n". */
|
|
330
|
8221 #ifdef FEAT_MBYTE
|
|
|
8222 if (has_mbyte)
|
|
358
|
8223 {
|
|
|
8224 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
8225 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
479
|
8226 if (enc_utf8 && utf_iscomposing(c))
|
|
|
8227 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
8228 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
358
|
8229 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
8230 }
|
|
330
|
8231 else
|
|
|
8232 #endif
|
|
358
|
8233 {
|
|
330
|
8234 ++stack[depth].ts_fidx;
|
|
358
|
8235 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
8236 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
8237 }
|
|
323
|
8238 break;
|
|
|
8239 }
|
|
|
8240 /*FALLTHROUGH*/
|
|
|
8241
|
|
|
8242 case STATE_INS:
|
|
330
|
8243 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
8244 * node. */
|
|
|
8245 n = sp->ts_arridx;
|
|
|
8246 if (sp->ts_curi > byts[n])
|
|
|
8247 {
|
|
|
8248 /* Done all bytes at this node, do next state. */
|
|
|
8249 sp->ts_state = STATE_SWAP;
|
|
|
8250 }
|
|
|
8251 else
|
|
|
8252 {
|
|
330
|
8253 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
8254 n += sp->ts_curi++;
|
|
|
8255 c = byts[n];
|
|
|
8256 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
8257 {
|
|
|
8258 ++depth;
|
|
330
|
8259 sp = &stack[depth];
|
|
|
8260 tword[sp->ts_twordlen++] = c;
|
|
|
8261 sp->ts_arridx = idxs[n];
|
|
|
8262 #ifdef FEAT_MBYTE
|
|
|
8263 if (has_mbyte)
|
|
|
8264 {
|
|
|
8265 fl = MB_BYTE2LEN(c);
|
|
|
8266 if (fl > 1)
|
|
|
8267 {
|
|
|
8268 /* There are following bytes for the same
|
|
|
8269 * character. We must find all bytes before
|
|
|
8270 * trying delete/insert/swap/etc. */
|
|
|
8271 sp->ts_tcharlen = fl;
|
|
|
8272 sp->ts_tcharidx = 1;
|
|
|
8273 sp->ts_isdiff = DIFF_INSERT;
|
|
|
8274 }
|
|
|
8275 }
|
|
358
|
8276 else
|
|
|
8277 fl = 1;
|
|
|
8278 if (fl == 1)
|
|
330
|
8279 #endif
|
|
358
|
8280 {
|
|
|
8281 /* If the previous character was the same, thus
|
|
|
8282 * doubling a character, give a bonus to the
|
|
|
8283 * score. */
|
|
|
8284 if (sp->ts_twordlen >= 2
|
|
|
8285 && tword[sp->ts_twordlen - 2] == c)
|
|
|
8286 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
8287 }
|
|
323
|
8288 }
|
|
|
8289 }
|
|
|
8290 break;
|
|
|
8291
|
|
|
8292 case STATE_SWAP:
|
|
330
|
8293 /*
|
|
|
8294 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
8295 * We change "fword" here, it's changed back afterwards.
|
|
|
8296 */
|
|
|
8297 p = fword + sp->ts_fidx;
|
|
|
8298 c = *p;
|
|
|
8299 if (c == NUL)
|
|
|
8300 {
|
|
|
8301 /* End of word, can't swap or replace. */
|
|
|
8302 sp->ts_state = STATE_FINAL;
|
|
|
8303 break;
|
|
|
8304 }
|
|
|
8305 #ifdef FEAT_MBYTE
|
|
|
8306 if (has_mbyte)
|
|
|
8307 {
|
|
474
|
8308 n = mb_cptr2len(p);
|
|
330
|
8309 c = mb_ptr2char(p);
|
|
|
8310 c2 = mb_ptr2char(p + n);
|
|
|
8311 }
|
|
|
8312 else
|
|
|
8313 #endif
|
|
|
8314 c2 = p[1];
|
|
|
8315 if (c == c2)
|
|
323
|
8316 {
|
|
330
|
8317 /* Characters are identical, swap won't do anything. */
|
|
|
8318 sp->ts_state = STATE_SWAP3;
|
|
|
8319 break;
|
|
|
8320 }
|
|
|
8321 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
8322 {
|
|
|
8323 sp->ts_state = STATE_UNSWAP;
|
|
323
|
8324 ++depth;
|
|
330
|
8325 #ifdef FEAT_MBYTE
|
|
|
8326 if (has_mbyte)
|
|
|
8327 {
|
|
|
8328 fl = mb_char2len(c2);
|
|
|
8329 mch_memmove(p, p + n, fl);
|
|
|
8330 mb_char2bytes(c, p + fl);
|
|
|
8331 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
8332 }
|
|
|
8333 else
|
|
|
8334 #endif
|
|
|
8335 {
|
|
|
8336 p[0] = c2;
|
|
|
8337 p[1] = c;
|
|
|
8338 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
8339 }
|
|
323
|
8340 }
|
|
|
8341 else
|
|
|
8342 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
8343 sp->ts_state = STATE_REP_INI;
|
|
|
8344 break;
|
|
|
8345
|
|
330
|
8346 case STATE_UNSWAP:
|
|
|
8347 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
8348 p = fword + sp->ts_fidx;
|
|
|
8349 #ifdef FEAT_MBYTE
|
|
|
8350 if (has_mbyte)
|
|
|
8351 {
|
|
|
8352 n = MB_BYTE2LEN(*p);
|
|
|
8353 c = mb_ptr2char(p + n);
|
|
|
8354 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
8355 mb_char2bytes(c, p);
|
|
|
8356 }
|
|
|
8357 else
|
|
|
8358 #endif
|
|
|
8359 {
|
|
|
8360 c = *p;
|
|
|
8361 *p = p[1];
|
|
|
8362 p[1] = c;
|
|
|
8363 }
|
|
|
8364 /*FALLTHROUGH*/
|
|
|
8365
|
|
|
8366 case STATE_SWAP3:
|
|
323
|
8367 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
8368 * "fword" here, it's changed back afterwards. */
|
|
|
8369 p = fword + sp->ts_fidx;
|
|
|
8370 #ifdef FEAT_MBYTE
|
|
|
8371 if (has_mbyte)
|
|
|
8372 {
|
|
474
|
8373 n = mb_cptr2len(p);
|
|
330
|
8374 c = mb_ptr2char(p);
|
|
474
|
8375 fl = mb_cptr2len(p + n);
|
|
330
|
8376 c2 = mb_ptr2char(p + n);
|
|
|
8377 c3 = mb_ptr2char(p + n + fl);
|
|
|
8378 }
|
|
|
8379 else
|
|
|
8380 #endif
|
|
323
|
8381 {
|
|
330
|
8382 c = *p;
|
|
|
8383 c2 = p[1];
|
|
|
8384 c3 = p[2];
|
|
|
8385 }
|
|
|
8386
|
|
|
8387 /* When characters are identical: "121" then SWAP3 result is
|
|
|
8388 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
8389 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
8390 * swapping. Also skip when c3 is NUL. */
|
|
|
8391 if (c == c3 || c3 == NUL)
|
|
|
8392 {
|
|
|
8393 sp->ts_state = STATE_REP_INI;
|
|
|
8394 break;
|
|
|
8395 }
|
|
|
8396 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
8397 {
|
|
|
8398 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
8399 ++depth;
|
|
330
|
8400 #ifdef FEAT_MBYTE
|
|
|
8401 if (has_mbyte)
|
|
|
8402 {
|
|
|
8403 tl = mb_char2len(c3);
|
|
|
8404 mch_memmove(p, p + n + fl, tl);
|
|
|
8405 mb_char2bytes(c2, p + tl);
|
|
|
8406 mb_char2bytes(c, p + fl + tl);
|
|
|
8407 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
8408 }
|
|
|
8409 else
|
|
|
8410 #endif
|
|
|
8411 {
|
|
|
8412 p[0] = p[2];
|
|
|
8413 p[2] = c;
|
|
|
8414 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8415 }
|
|
323
|
8416 }
|
|
|
8417 else
|
|
|
8418 sp->ts_state = STATE_REP_INI;
|
|
|
8419 break;
|
|
|
8420
|
|
330
|
8421 case STATE_UNSWAP3:
|
|
|
8422 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
8423 p = fword + sp->ts_fidx;
|
|
|
8424 #ifdef FEAT_MBYTE
|
|
|
8425 if (has_mbyte)
|
|
|
8426 {
|
|
|
8427 n = MB_BYTE2LEN(*p);
|
|
|
8428 c2 = mb_ptr2char(p + n);
|
|
|
8429 fl = MB_BYTE2LEN(p[n]);
|
|
|
8430 c = mb_ptr2char(p + n + fl);
|
|
|
8431 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
8432 mch_memmove(p + fl + tl, p, n);
|
|
|
8433 mb_char2bytes(c, p);
|
|
|
8434 mb_char2bytes(c2, p + tl);
|
|
|
8435 }
|
|
|
8436 else
|
|
|
8437 #endif
|
|
|
8438 {
|
|
|
8439 c = *p;
|
|
|
8440 *p = p[2];
|
|
|
8441 p[2] = c;
|
|
|
8442 }
|
|
346
|
8443
|
|
330
|
8444 /* Rotate three characters left: "123" -> "231". We change
|
|
|
8445 * "fword" here, it's changed back afterwards. */
|
|
323
|
8446 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
8447 {
|
|
330
|
8448 sp->ts_state = STATE_UNROT3L;
|
|
323
|
8449 ++depth;
|
|
330
|
8450 p = fword + sp->ts_fidx;
|
|
|
8451 #ifdef FEAT_MBYTE
|
|
|
8452 if (has_mbyte)
|
|
|
8453 {
|
|
474
|
8454 n = mb_cptr2len(p);
|
|
330
|
8455 c = mb_ptr2char(p);
|
|
474
|
8456 fl = mb_cptr2len(p + n);
|
|
|
8457 fl += mb_cptr2len(p + n + fl);
|
|
330
|
8458 mch_memmove(p, p + n, fl);
|
|
|
8459 mb_char2bytes(c, p + fl);
|
|
|
8460 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
8461 }
|
|
|
8462 else
|
|
|
8463 #endif
|
|
|
8464 {
|
|
|
8465 c = *p;
|
|
|
8466 *p = p[1];
|
|
|
8467 p[1] = p[2];
|
|
|
8468 p[2] = c;
|
|
|
8469 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8470 }
|
|
323
|
8471 }
|
|
|
8472 else
|
|
|
8473 sp->ts_state = STATE_REP_INI;
|
|
|
8474 break;
|
|
|
8475
|
|
330
|
8476 case STATE_UNROT3L:
|
|
346
|
8477 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
8478 p = fword + sp->ts_fidx;
|
|
|
8479 #ifdef FEAT_MBYTE
|
|
|
8480 if (has_mbyte)
|
|
|
8481 {
|
|
|
8482 n = MB_BYTE2LEN(*p);
|
|
|
8483 n += MB_BYTE2LEN(p[n]);
|
|
|
8484 c = mb_ptr2char(p + n);
|
|
|
8485 tl = MB_BYTE2LEN(p[n]);
|
|
|
8486 mch_memmove(p + tl, p, n);
|
|
|
8487 mb_char2bytes(c, p);
|
|
|
8488 }
|
|
|
8489 else
|
|
|
8490 #endif
|
|
|
8491 {
|
|
|
8492 c = p[2];
|
|
|
8493 p[2] = p[1];
|
|
|
8494 p[1] = *p;
|
|
|
8495 *p = c;
|
|
|
8496 }
|
|
346
|
8497
|
|
323
|
8498 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
8499 * "fword" here, it's changed back afterwards. */
|
|
323
|
8500 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
8501 {
|
|
330
|
8502 sp->ts_state = STATE_UNROT3R;
|
|
323
|
8503 ++depth;
|
|
330
|
8504 p = fword + sp->ts_fidx;
|
|
|
8505 #ifdef FEAT_MBYTE
|
|
|
8506 if (has_mbyte)
|
|
|
8507 {
|
|
474
|
8508 n = mb_cptr2len(p);
|
|
|
8509 n += mb_cptr2len(p + n);
|
|
330
|
8510 c = mb_ptr2char(p + n);
|
|
474
|
8511 tl = mb_cptr2len(p + n);
|
|
330
|
8512 mch_memmove(p + tl, p, n);
|
|
|
8513 mb_char2bytes(c, p);
|
|
|
8514 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
8515 }
|
|
|
8516 else
|
|
|
8517 #endif
|
|
|
8518 {
|
|
|
8519 c = p[2];
|
|
|
8520 p[2] = p[1];
|
|
|
8521 p[1] = *p;
|
|
|
8522 *p = c;
|
|
|
8523 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8524 }
|
|
323
|
8525 }
|
|
|
8526 else
|
|
|
8527 sp->ts_state = STATE_REP_INI;
|
|
|
8528 break;
|
|
|
8529
|
|
330
|
8530 case STATE_UNROT3R:
|
|
346
|
8531 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
8532 p = fword + sp->ts_fidx;
|
|
|
8533 #ifdef FEAT_MBYTE
|
|
|
8534 if (has_mbyte)
|
|
|
8535 {
|
|
|
8536 c = mb_ptr2char(p);
|
|
|
8537 tl = MB_BYTE2LEN(*p);
|
|
|
8538 n = MB_BYTE2LEN(p[tl]);
|
|
|
8539 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
8540 mch_memmove(p, p + tl, n);
|
|
|
8541 mb_char2bytes(c, p + n);
|
|
|
8542 }
|
|
|
8543 else
|
|
|
8544 #endif
|
|
|
8545 {
|
|
|
8546 c = *p;
|
|
|
8547 *p = p[1];
|
|
|
8548 p[1] = p[2];
|
|
|
8549 p[2] = c;
|
|
|
8550 }
|
|
323
|
8551 /*FALLTHROUGH*/
|
|
|
8552
|
|
|
8553 case STATE_REP_INI:
|
|
|
8554 /* Check if matching with REP items from the .aff file would
|
|
|
8555 * work. Quickly skip if there are no REP items or the score
|
|
|
8556 * is going to be too high anyway. */
|
|
482
|
8557 gap = &slang->sl_rep;
|
|
323
|
8558 if (gap->ga_len == 0
|
|
|
8559 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
8560 {
|
|
|
8561 sp->ts_state = STATE_FINAL;
|
|
|
8562 break;
|
|
|
8563 }
|
|
|
8564
|
|
|
8565 /* Use the first byte to quickly find the first entry that
|
|
330
|
8566 * may match. If the index is -1 there is none. */
|
|
482
|
8567 sp->ts_curi = slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
323
|
8568 if (sp->ts_curi < 0)
|
|
|
8569 {
|
|
|
8570 sp->ts_state = STATE_FINAL;
|
|
|
8571 break;
|
|
|
8572 }
|
|
|
8573
|
|
|
8574 sp->ts_state = STATE_REP;
|
|
|
8575 /*FALLTHROUGH*/
|
|
|
8576
|
|
|
8577 case STATE_REP:
|
|
|
8578 /* Try matching with REP items from the .aff file. For each
|
|
330
|
8579 * match replace the characters and check if the resulting
|
|
|
8580 * word is valid. */
|
|
323
|
8581 p = fword + sp->ts_fidx;
|
|
|
8582
|
|
482
|
8583 gap = &slang->sl_rep;
|
|
323
|
8584 while (sp->ts_curi < gap->ga_len)
|
|
|
8585 {
|
|
|
8586 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
8587 if (*ftp->ft_from != *p)
|
|
|
8588 {
|
|
|
8589 /* past possible matching entries */
|
|
|
8590 sp->ts_curi = gap->ga_len;
|
|
|
8591 break;
|
|
|
8592 }
|
|
|
8593 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
8594 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
8595 {
|
|
|
8596 /* Need to undo this afterwards. */
|
|
|
8597 sp->ts_state = STATE_REP_UNDO;
|
|
|
8598
|
|
|
8599 /* Change the "from" to the "to" string. */
|
|
|
8600 ++depth;
|
|
|
8601 fl = STRLEN(ftp->ft_from);
|
|
|
8602 tl = STRLEN(ftp->ft_to);
|
|
|
8603 if (fl != tl)
|
|
346
|
8604 {
|
|
323
|
8605 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
8606 repextra += tl - fl;
|
|
|
8607 }
|
|
323
|
8608 mch_memmove(p, ftp->ft_to, tl);
|
|
|
8609 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
8610 #ifdef FEAT_MBYTE
|
|
|
8611 stack[depth].ts_tcharlen = 0;
|
|
|
8612 #endif
|
|
323
|
8613 break;
|
|
|
8614 }
|
|
|
8615 }
|
|
|
8616
|
|
|
8617 if (sp->ts_curi >= gap->ga_len)
|
|
|
8618 /* No (more) matches. */
|
|
|
8619 sp->ts_state = STATE_FINAL;
|
|
|
8620
|
|
|
8621 break;
|
|
|
8622
|
|
|
8623 case STATE_REP_UNDO:
|
|
|
8624 /* Undo a REP replacement and continue with the next one. */
|
|
482
|
8625 ftp = (fromto_T *)slang->sl_rep.ga_data + sp->ts_curi - 1;
|
|
323
|
8626 fl = STRLEN(ftp->ft_from);
|
|
|
8627 tl = STRLEN(ftp->ft_to);
|
|
|
8628 p = fword + sp->ts_fidx;
|
|
|
8629 if (fl != tl)
|
|
346
|
8630 {
|
|
323
|
8631 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
8632 repextra -= tl - fl;
|
|
|
8633 }
|
|
323
|
8634 mch_memmove(p, ftp->ft_from, fl);
|
|
|
8635 sp->ts_state = STATE_REP;
|
|
|
8636 break;
|
|
|
8637
|
|
|
8638 default:
|
|
|
8639 /* Did all possible states at this level, go up one level. */
|
|
|
8640 --depth;
|
|
344
|
8641
|
|
482
|
8642 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
8643 {
|
|
|
8644 /* Continue in or go back to the prefix tree. */
|
|
|
8645 byts = pbyts;
|
|
|
8646 idxs = pidxs;
|
|
|
8647 splitoff = 0;
|
|
|
8648 }
|
|
|
8649
|
|
344
|
8650 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
8651 line_breakcheck();
|
|
323
|
8652 }
|
|
|
8653 }
|
|
|
8654 }
|
|
|
8655 }
|
|
|
8656
|
|
|
8657 /*
|
|
|
8658 * Try going one level deeper in the tree.
|
|
|
8659 */
|
|
|
8660 static int
|
|
|
8661 try_deeper(su, stack, depth, score_add)
|
|
|
8662 suginfo_T *su;
|
|
|
8663 trystate_T *stack;
|
|
|
8664 int depth;
|
|
|
8665 int score_add;
|
|
|
8666 {
|
|
|
8667 int newscore;
|
|
|
8668
|
|
|
8669 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
8670 newscore = stack[depth].ts_score + score_add;
|
|
|
8671 if (newscore >= su->su_maxscore)
|
|
|
8672 return FALSE;
|
|
|
8673
|
|
330
|
8674 stack[depth + 1] = stack[depth];
|
|
323
|
8675 stack[depth + 1].ts_state = STATE_START;
|
|
|
8676 stack[depth + 1].ts_score = newscore;
|
|
|
8677 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
8678 return TRUE;
|
|
|
8679 }
|
|
|
8680
|
|
455
|
8681 #ifdef FEAT_MBYTE
|
|
|
8682 /*
|
|
|
8683 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
8684 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
8685 */
|
|
|
8686 static int
|
|
|
8687 nofold_len(fword, flen, word)
|
|
|
8688 char_u *fword;
|
|
|
8689 int flen;
|
|
|
8690 char_u *word;
|
|
|
8691 {
|
|
|
8692 char_u *p;
|
|
|
8693 int i = 0;
|
|
|
8694
|
|
|
8695 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
8696 ++i;
|
|
|
8697 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
8698 --i;
|
|
|
8699 return (int)(p - word);
|
|
|
8700 }
|
|
|
8701 #endif
|
|
|
8702
|
|
323
|
8703 /*
|
|
|
8704 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
8705 * words and put it in "kword".
|
|
|
8706 * Theoretically there could be several keep-case words that result in the
|
|
|
8707 * same case-folded word, but we only find one...
|
|
|
8708 */
|
|
|
8709 static void
|
|
|
8710 find_keepcap_word(slang, fword, kword)
|
|
|
8711 slang_T *slang;
|
|
|
8712 char_u *fword;
|
|
|
8713 char_u *kword;
|
|
|
8714 {
|
|
|
8715 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
8716 int depth;
|
|
324
|
8717 idx_T tryidx;
|
|
323
|
8718
|
|
|
8719 /* The following arrays are used at each depth in the tree. */
|
|
324
|
8720 idx_T arridx[MAXWLEN];
|
|
323
|
8721 int round[MAXWLEN];
|
|
|
8722 int fwordidx[MAXWLEN];
|
|
|
8723 int uwordidx[MAXWLEN];
|
|
|
8724 int kwordlen[MAXWLEN];
|
|
|
8725
|
|
|
8726 int flen, ulen;
|
|
|
8727 int l;
|
|
|
8728 int len;
|
|
|
8729 int c;
|
|
324
|
8730 idx_T lo, hi, m;
|
|
323
|
8731 char_u *p;
|
|
|
8732 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
8733 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
8734
|
|
|
8735 if (byts == NULL)
|
|
|
8736 {
|
|
|
8737 /* array is empty: "cannot happen" */
|
|
|
8738 *kword = NUL;
|
|
|
8739 return;
|
|
|
8740 }
|
|
|
8741
|
|
|
8742 /* Make an all-cap version of "fword". */
|
|
|
8743 allcap_copy(fword, uword);
|
|
|
8744
|
|
|
8745 /*
|
|
|
8746 * Each character needs to be tried both case-folded and upper-case.
|
|
|
8747 * All this gets very complicated if we keep in mind that changing case
|
|
|
8748 * may change the byte length of a multi-byte character...
|
|
|
8749 */
|
|
|
8750 depth = 0;
|
|
|
8751 arridx[0] = 0;
|
|
|
8752 round[0] = 0;
|
|
|
8753 fwordidx[0] = 0;
|
|
|
8754 uwordidx[0] = 0;
|
|
|
8755 kwordlen[0] = 0;
|
|
|
8756 while (depth >= 0)
|
|
|
8757 {
|
|
|
8758 if (fword[fwordidx[depth]] == NUL)
|
|
|
8759 {
|
|
|
8760 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
8761 * here we have found a match. */
|
|
|
8762 if (byts[arridx[depth] + 1] == 0)
|
|
|
8763 {
|
|
|
8764 kword[kwordlen[depth]] = NUL;
|
|
|
8765 return;
|
|
|
8766 }
|
|
|
8767
|
|
|
8768 /* kword is getting too long, continue one level up */
|
|
|
8769 --depth;
|
|
|
8770 }
|
|
|
8771 else if (++round[depth] > 2)
|
|
|
8772 {
|
|
|
8773 /* tried both fold-case and upper-case character, continue one
|
|
|
8774 * level up */
|
|
|
8775 --depth;
|
|
|
8776 }
|
|
|
8777 else
|
|
|
8778 {
|
|
|
8779 /*
|
|
|
8780 * round[depth] == 1: Try using the folded-case character.
|
|
|
8781 * round[depth] == 2: Try using the upper-case character.
|
|
|
8782 */
|
|
|
8783 #ifdef FEAT_MBYTE
|
|
|
8784 if (has_mbyte)
|
|
|
8785 {
|
|
474
|
8786 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
8787 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
8788 }
|
|
|
8789 else
|
|
|
8790 #endif
|
|
|
8791 ulen = flen = 1;
|
|
|
8792 if (round[depth] == 1)
|
|
|
8793 {
|
|
|
8794 p = fword + fwordidx[depth];
|
|
|
8795 l = flen;
|
|
|
8796 }
|
|
|
8797 else
|
|
|
8798 {
|
|
|
8799 p = uword + uwordidx[depth];
|
|
|
8800 l = ulen;
|
|
|
8801 }
|
|
|
8802
|
|
|
8803 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
8804 {
|
|
|
8805 /* Perform a binary search in the list of accepted bytes. */
|
|
|
8806 len = byts[tryidx++];
|
|
|
8807 c = *p++;
|
|
|
8808 lo = tryidx;
|
|
|
8809 hi = tryidx + len - 1;
|
|
|
8810 while (lo < hi)
|
|
|
8811 {
|
|
|
8812 m = (lo + hi) / 2;
|
|
|
8813 if (byts[m] > c)
|
|
|
8814 hi = m - 1;
|
|
|
8815 else if (byts[m] < c)
|
|
|
8816 lo = m + 1;
|
|
|
8817 else
|
|
|
8818 {
|
|
|
8819 lo = hi = m;
|
|
|
8820 break;
|
|
|
8821 }
|
|
|
8822 }
|
|
|
8823
|
|
|
8824 /* Stop if there is no matching byte. */
|
|
|
8825 if (hi < lo || byts[lo] != c)
|
|
|
8826 break;
|
|
|
8827
|
|
|
8828 /* Continue at the child (if there is one). */
|
|
|
8829 tryidx = idxs[lo];
|
|
|
8830 }
|
|
|
8831
|
|
|
8832 if (l == 0)
|
|
|
8833 {
|
|
|
8834 /*
|
|
|
8835 * Found the matching char. Copy it to "kword" and go a
|
|
|
8836 * level deeper.
|
|
|
8837 */
|
|
|
8838 if (round[depth] == 1)
|
|
|
8839 {
|
|
|
8840 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
8841 flen);
|
|
|
8842 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
8843 }
|
|
|
8844 else
|
|
|
8845 {
|
|
|
8846 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
8847 ulen);
|
|
|
8848 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
8849 }
|
|
|
8850 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
8851 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
8852
|
|
|
8853 ++depth;
|
|
|
8854 arridx[depth] = tryidx;
|
|
|
8855 round[depth] = 0;
|
|
|
8856 }
|
|
|
8857 }
|
|
|
8858 }
|
|
|
8859
|
|
|
8860 /* Didn't find it: "cannot happen". */
|
|
|
8861 *kword = NUL;
|
|
|
8862 }
|
|
|
8863
|
|
|
8864 /*
|
|
344
|
8865 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
8866 * su->su_sga.
|
|
|
8867 */
|
|
|
8868 static void
|
|
|
8869 score_comp_sal(su)
|
|
|
8870 suginfo_T *su;
|
|
|
8871 {
|
|
|
8872 langp_T *lp;
|
|
|
8873 char_u badsound[MAXWLEN];
|
|
|
8874 int i;
|
|
|
8875 suggest_T *stp;
|
|
|
8876 suggest_T *sstp;
|
|
|
8877 int score;
|
|
|
8878
|
|
|
8879 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
8880 return;
|
|
|
8881
|
|
|
8882 /* Use the sound-folding of the first language that supports it. */
|
|
|
8883 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8884 lp->lp_slang != NULL; ++lp)
|
|
|
8885 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8886 {
|
|
|
8887 /* soundfold the bad word */
|
|
375
|
8888 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8889
|
|
|
8890 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8891 {
|
|
|
8892 stp = &SUG(su->su_ga, i);
|
|
|
8893
|
|
351
|
8894 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
8895 * sound-a-like score. */
|
|
|
8896 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
8897 if (score < SCORE_MAXMAX)
|
|
|
8898 {
|
|
|
8899 /* Add the suggestion. */
|
|
|
8900 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
8901 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
8902 if (sstp->st_word != NULL)
|
|
|
8903 {
|
|
|
8904 sstp->st_score = score;
|
|
|
8905 sstp->st_altscore = 0;
|
|
|
8906 sstp->st_orglen = stp->st_orglen;
|
|
|
8907 ++su->su_sga.ga_len;
|
|
|
8908 }
|
|
|
8909 }
|
|
|
8910 }
|
|
|
8911 break;
|
|
|
8912 }
|
|
|
8913 }
|
|
|
8914
|
|
|
8915 /*
|
|
|
8916 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
8917 * They are intwined.
|
|
|
8918 */
|
|
|
8919 static void
|
|
|
8920 score_combine(su)
|
|
|
8921 suginfo_T *su;
|
|
|
8922 {
|
|
|
8923 int i;
|
|
|
8924 int j;
|
|
|
8925 garray_T ga;
|
|
|
8926 garray_T *gap;
|
|
|
8927 langp_T *lp;
|
|
|
8928 suggest_T *stp;
|
|
|
8929 char_u *p;
|
|
|
8930 char_u badsound[MAXWLEN];
|
|
|
8931 int round;
|
|
|
8932
|
|
|
8933 /* Add the alternate score to su_ga. */
|
|
|
8934 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8935 lp->lp_slang != NULL; ++lp)
|
|
|
8936 {
|
|
|
8937 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8938 {
|
|
|
8939 /* soundfold the bad word */
|
|
375
|
8940 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8941
|
|
|
8942 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8943 {
|
|
|
8944 stp = &SUG(su->su_ga, i);
|
|
351
|
8945 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
8946 badsound);
|
|
344
|
8947 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8948 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
8949 else
|
|
|
8950 stp->st_score = (stp->st_score * 3
|
|
|
8951 + stp->st_altscore) / 4;
|
|
|
8952 stp->st_salscore = FALSE;
|
|
|
8953 }
|
|
|
8954 break;
|
|
|
8955 }
|
|
|
8956 }
|
|
|
8957
|
|
|
8958 /* Add the alternate score to su_sga. */
|
|
|
8959 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8960 {
|
|
|
8961 stp = &SUG(su->su_sga, i);
|
|
|
8962 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
8963 if (stp->st_score == SCORE_MAXMAX)
|
|
|
8964 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
8965 else
|
|
|
8966 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
8967 stp->st_salscore = TRUE;
|
|
|
8968 }
|
|
|
8969
|
|
|
8970 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
8971 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8972 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
8973
|
|
|
8974 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
8975 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
8976 return;
|
|
|
8977
|
|
|
8978 stp = &SUG(ga, 0);
|
|
|
8979 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
8980 {
|
|
|
8981 /* round 1: get a suggestion from su_ga
|
|
|
8982 * round 2: get a suggestion from su_sga */
|
|
|
8983 for (round = 1; round <= 2; ++round)
|
|
|
8984 {
|
|
|
8985 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
8986 if (i < gap->ga_len)
|
|
|
8987 {
|
|
|
8988 /* Don't add a word if it's already there. */
|
|
|
8989 p = SUG(*gap, i).st_word;
|
|
|
8990 for (j = 0; j < ga.ga_len; ++j)
|
|
|
8991 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
8992 break;
|
|
|
8993 if (j == ga.ga_len)
|
|
|
8994 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
8995 else
|
|
|
8996 vim_free(p);
|
|
|
8997 }
|
|
|
8998 }
|
|
|
8999 }
|
|
|
9000
|
|
|
9001 ga_clear(&su->su_ga);
|
|
|
9002 ga_clear(&su->su_sga);
|
|
|
9003
|
|
|
9004 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
9005 if (ga.ga_len > su->su_maxcount)
|
|
|
9006 {
|
|
|
9007 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
9008 vim_free(stp[i].st_word);
|
|
|
9009 ga.ga_len = su->su_maxcount;
|
|
|
9010 }
|
|
|
9011
|
|
|
9012 su->su_ga = ga;
|
|
|
9013 }
|
|
|
9014
|
|
|
9015 /*
|
|
351
|
9016 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
9017 * badword.
|
|
|
9018 */
|
|
|
9019 static int
|
|
|
9020 stp_sal_score(stp, su, slang, badsound)
|
|
|
9021 suggest_T *stp;
|
|
|
9022 suginfo_T *su;
|
|
|
9023 slang_T *slang;
|
|
|
9024 char_u *badsound; /* sound-folded badword */
|
|
|
9025 {
|
|
|
9026 char_u *p;
|
|
|
9027 char_u badsound2[MAXWLEN];
|
|
|
9028 char_u fword[MAXWLEN];
|
|
|
9029 char_u goodsound[MAXWLEN];
|
|
|
9030
|
|
|
9031 if (stp->st_orglen <= su->su_badlen)
|
|
|
9032 p = badsound;
|
|
|
9033 else
|
|
|
9034 {
|
|
|
9035 /* soundfold the bad word with more characters following */
|
|
|
9036 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
9037
|
|
|
9038 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
9039 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
9040 * removing the space. Don't do it when the good word also contains a
|
|
|
9041 * space. */
|
|
|
9042 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
9043 && *skiptowhite(stp->st_word) == NUL)
|
|
|
9044 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
9045 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
9046
|
|
375
|
9047 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
9048 p = badsound2;
|
|
|
9049 }
|
|
|
9050
|
|
375
|
9051 /* Sound-fold the word and compute the score for the difference. */
|
|
|
9052 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
9053
|
|
|
9054 return soundalike_score(goodsound, p);
|
|
|
9055 }
|
|
|
9056
|
|
|
9057 /*
|
|
323
|
9058 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
9059 */
|
|
|
9060 static void
|
|
346
|
9061 suggest_try_soundalike(su)
|
|
323
|
9062 suginfo_T *su;
|
|
|
9063 {
|
|
|
9064 char_u salword[MAXWLEN];
|
|
|
9065 char_u tword[MAXWLEN];
|
|
|
9066 char_u tsalword[MAXWLEN];
|
|
324
|
9067 idx_T arridx[MAXWLEN];
|
|
323
|
9068 int curi[MAXWLEN];
|
|
|
9069 langp_T *lp;
|
|
|
9070 char_u *byts;
|
|
324
|
9071 idx_T *idxs;
|
|
323
|
9072 int depth;
|
|
|
9073 int c;
|
|
324
|
9074 idx_T n;
|
|
323
|
9075 int round;
|
|
|
9076 int flags;
|
|
344
|
9077 int sound_score;
|
|
|
9078
|
|
|
9079 /* Do this for all languages that support sound folding. */
|
|
323
|
9080 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9081 lp->lp_slang != NULL; ++lp)
|
|
|
9082 {
|
|
|
9083 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9084 {
|
|
|
9085 /* soundfold the bad word */
|
|
375
|
9086 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
9087
|
|
|
9088 /*
|
|
|
9089 * Go through the whole tree, soundfold each word and compare.
|
|
|
9090 * round 1: use the case-folded tree.
|
|
|
9091 * round 2: use the keep-case tree.
|
|
|
9092 */
|
|
|
9093 for (round = 1; round <= 2; ++round)
|
|
|
9094 {
|
|
|
9095 if (round == 1)
|
|
|
9096 {
|
|
|
9097 byts = lp->lp_slang->sl_fbyts;
|
|
|
9098 idxs = lp->lp_slang->sl_fidxs;
|
|
|
9099 }
|
|
|
9100 else
|
|
|
9101 {
|
|
|
9102 byts = lp->lp_slang->sl_kbyts;
|
|
|
9103 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
9104 if (byts == NULL) /* no keep-case words */
|
|
|
9105 continue;
|
|
323
|
9106 }
|
|
|
9107
|
|
|
9108 depth = 0;
|
|
|
9109 arridx[0] = 0;
|
|
|
9110 curi[0] = 1;
|
|
|
9111 while (depth >= 0 && !got_int)
|
|
|
9112 {
|
|
|
9113 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
9114 {
|
|
323
|
9115 /* Done all bytes at this node, go up one level. */
|
|
|
9116 --depth;
|
|
351
|
9117 line_breakcheck();
|
|
|
9118 }
|
|
323
|
9119 else
|
|
|
9120 {
|
|
|
9121 /* Do one more byte at this node. */
|
|
|
9122 n = arridx[depth] + curi[depth];
|
|
|
9123 ++curi[depth];
|
|
|
9124 c = byts[n];
|
|
|
9125 if (c == 0)
|
|
|
9126 {
|
|
|
9127 /* End of word, deal with the word. */
|
|
324
|
9128 flags = (int)idxs[n];
|
|
323
|
9129 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
9130 {
|
|
|
9131 tword[depth] = NUL;
|
|
375
|
9132 /* Sound-fold. Only in keep-case tree need to
|
|
|
9133 * case-fold the word. */
|
|
|
9134 spell_soundfold(lp->lp_slang, tword,
|
|
|
9135 round == 1, tsalword);
|
|
323
|
9136
|
|
344
|
9137 /* Compute the edit distance between the
|
|
|
9138 * sound-a-like words. */
|
|
|
9139 sound_score = soundalike_score(salword,
|
|
|
9140 tsalword);
|
|
324
|
9141 if (sound_score < SCORE_MAXMAX)
|
|
323
|
9142 {
|
|
324
|
9143 char_u cword[MAXWLEN];
|
|
|
9144 char_u *p;
|
|
344
|
9145 int score;
|
|
324
|
9146
|
|
388
|
9147 flags |= su->su_badflags;
|
|
351
|
9148 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
9149 {
|
|
324
|
9150 /* Need to fix case according to
|
|
|
9151 * "flags". */
|
|
323
|
9152 make_case_word(tword, cword, flags);
|
|
324
|
9153 p = cword;
|
|
323
|
9154 }
|
|
|
9155 else
|
|
324
|
9156 p = tword;
|
|
|
9157
|
|
344
|
9158 if (sps_flags & SPS_DOUBLE)
|
|
|
9159 add_suggestion(su, &su->su_sga, p,
|
|
346
|
9160 su->su_badlen,
|
|
351
|
9161 sound_score, 0, FALSE);
|
|
344
|
9162 else
|
|
|
9163 {
|
|
|
9164 /* Compute the score. */
|
|
|
9165 score = spell_edit_score(
|
|
|
9166 su->su_badword, p);
|
|
|
9167 if (sps_flags & SPS_BEST)
|
|
|
9168 /* give a bonus for the good word
|
|
|
9169 * sounding the same as the bad
|
|
|
9170 * word */
|
|
|
9171 add_suggestion(su, &su->su_ga, p,
|
|
346
|
9172 su->su_badlen,
|
|
344
|
9173 RESCORE(score, sound_score),
|
|
351
|
9174 sound_score, TRUE);
|
|
344
|
9175 else
|
|
|
9176 add_suggestion(su, &su->su_ga, p,
|
|
346
|
9177 su->su_badlen,
|
|
351
|
9178 score + sound_score, 0, FALSE);
|
|
344
|
9179 }
|
|
323
|
9180 }
|
|
|
9181 }
|
|
|
9182
|
|
|
9183 /* Skip over other NUL bytes. */
|
|
|
9184 while (byts[n + 1] == 0)
|
|
|
9185 {
|
|
|
9186 ++n;
|
|
|
9187 ++curi[depth];
|
|
|
9188 }
|
|
|
9189 }
|
|
|
9190 else
|
|
|
9191 {
|
|
|
9192 /* Normal char, go one level deeper. */
|
|
|
9193 tword[depth++] = c;
|
|
|
9194 arridx[depth] = idxs[n];
|
|
|
9195 curi[depth] = 1;
|
|
|
9196 }
|
|
|
9197 }
|
|
|
9198 }
|
|
|
9199 }
|
|
|
9200 }
|
|
|
9201 }
|
|
|
9202 }
|
|
|
9203
|
|
|
9204 /*
|
|
324
|
9205 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
9206 */
|
|
|
9207 static void
|
|
|
9208 make_case_word(fword, cword, flags)
|
|
|
9209 char_u *fword;
|
|
|
9210 char_u *cword;
|
|
|
9211 int flags;
|
|
|
9212 {
|
|
|
9213 if (flags & WF_ALLCAP)
|
|
|
9214 /* Make it all upper-case */
|
|
|
9215 allcap_copy(fword, cword);
|
|
|
9216 else if (flags & WF_ONECAP)
|
|
|
9217 /* Make the first letter upper-case */
|
|
324
|
9218 onecap_copy(fword, cword, TRUE);
|
|
323
|
9219 else
|
|
|
9220 /* Use goodword as-is. */
|
|
|
9221 STRCPY(cword, fword);
|
|
|
9222 }
|
|
|
9223
|
|
330
|
9224 /*
|
|
|
9225 * Use map string "map" for languages "lp".
|
|
|
9226 */
|
|
|
9227 static void
|
|
|
9228 set_map_str(lp, map)
|
|
|
9229 slang_T *lp;
|
|
|
9230 char_u *map;
|
|
|
9231 {
|
|
|
9232 char_u *p;
|
|
|
9233 int headc = 0;
|
|
|
9234 int c;
|
|
|
9235 int i;
|
|
|
9236
|
|
|
9237 if (*map == NUL)
|
|
|
9238 {
|
|
|
9239 lp->sl_has_map = FALSE;
|
|
|
9240 return;
|
|
|
9241 }
|
|
|
9242 lp->sl_has_map = TRUE;
|
|
|
9243
|
|
|
9244 /* Init the array and hash table empty. */
|
|
|
9245 for (i = 0; i < 256; ++i)
|
|
|
9246 lp->sl_map_array[i] = 0;
|
|
|
9247 #ifdef FEAT_MBYTE
|
|
|
9248 hash_init(&lp->sl_map_hash);
|
|
|
9249 #endif
|
|
|
9250
|
|
|
9251 /*
|
|
|
9252 * The similar characters are stored separated with slashes:
|
|
|
9253 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
9254 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
9255 */
|
|
|
9256 for (p = map; *p != NUL; )
|
|
|
9257 {
|
|
|
9258 #ifdef FEAT_MBYTE
|
|
474
|
9259 c = mb_cptr2char_adv(&p);
|
|
330
|
9260 #else
|
|
|
9261 c = *p++;
|
|
|
9262 #endif
|
|
|
9263 if (c == '/')
|
|
|
9264 headc = 0;
|
|
|
9265 else
|
|
|
9266 {
|
|
|
9267 if (headc == 0)
|
|
|
9268 headc = c;
|
|
|
9269
|
|
|
9270 #ifdef FEAT_MBYTE
|
|
|
9271 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
9272 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
9273 * a NUL. */
|
|
|
9274 if (c >= 256)
|
|
|
9275 {
|
|
|
9276 int cl = mb_char2len(c);
|
|
|
9277 int headcl = mb_char2len(headc);
|
|
|
9278 char_u *b;
|
|
|
9279 hash_T hash;
|
|
|
9280 hashitem_T *hi;
|
|
|
9281
|
|
|
9282 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
9283 if (b == NULL)
|
|
|
9284 return;
|
|
|
9285 mb_char2bytes(c, b);
|
|
|
9286 b[cl] = NUL;
|
|
|
9287 mb_char2bytes(headc, b + cl + 1);
|
|
|
9288 b[cl + 1 + headcl] = NUL;
|
|
|
9289 hash = hash_hash(b);
|
|
|
9290 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
9291 if (HASHITEM_EMPTY(hi))
|
|
|
9292 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
9293 else
|
|
|
9294 {
|
|
|
9295 /* This should have been checked when generating the .spl
|
|
|
9296 * file. */
|
|
|
9297 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
9298 vim_free(b);
|
|
|
9299 }
|
|
|
9300 }
|
|
|
9301 else
|
|
|
9302 #endif
|
|
|
9303 lp->sl_map_array[c] = headc;
|
|
|
9304 }
|
|
|
9305 }
|
|
|
9306 }
|
|
|
9307
|
|
323
|
9308 /*
|
|
|
9309 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
9310 * lines in the .aff file.
|
|
|
9311 */
|
|
|
9312 static int
|
|
|
9313 similar_chars(slang, c1, c2)
|
|
|
9314 slang_T *slang;
|
|
|
9315 int c1;
|
|
|
9316 int c2;
|
|
|
9317 {
|
|
330
|
9318 int m1, m2;
|
|
|
9319 #ifdef FEAT_MBYTE
|
|
|
9320 char_u buf[MB_MAXBYTES];
|
|
|
9321 hashitem_T *hi;
|
|
|
9322
|
|
|
9323 if (c1 >= 256)
|
|
|
9324 {
|
|
|
9325 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
9326 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
9327 if (HASHITEM_EMPTY(hi))
|
|
|
9328 m1 = 0;
|
|
|
9329 else
|
|
|
9330 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
9331 }
|
|
|
9332 else
|
|
|
9333 #endif
|
|
|
9334 m1 = slang->sl_map_array[c1];
|
|
|
9335 if (m1 == 0)
|
|
323
|
9336 return FALSE;
|
|
330
|
9337
|
|
|
9338
|
|
|
9339 #ifdef FEAT_MBYTE
|
|
|
9340 if (c2 >= 256)
|
|
|
9341 {
|
|
|
9342 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
9343 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
9344 if (HASHITEM_EMPTY(hi))
|
|
|
9345 m2 = 0;
|
|
|
9346 else
|
|
|
9347 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
9348 }
|
|
|
9349 else
|
|
|
9350 #endif
|
|
|
9351 m2 = slang->sl_map_array[c2];
|
|
|
9352
|
|
|
9353 return m1 == m2;
|
|
323
|
9354 }
|
|
|
9355
|
|
|
9356 /*
|
|
|
9357 * Add a suggestion to the list of suggestions.
|
|
|
9358 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
9359 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
9360 * with spell_edit_score().
|
|
|
9361 */
|
|
|
9362 static void
|
|
351
|
9363 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
9364 suginfo_T *su;
|
|
344
|
9365 garray_T *gap;
|
|
323
|
9366 char_u *goodword;
|
|
346
|
9367 int badlen; /* length of bad word used */
|
|
324
|
9368 int score;
|
|
351
|
9369 int altscore;
|
|
344
|
9370 int had_bonus; /* value for st_had_bonus */
|
|
323
|
9371 {
|
|
|
9372 suggest_T *stp;
|
|
|
9373 int i;
|
|
346
|
9374 char_u *p = NULL;
|
|
|
9375 int c = 0;
|
|
323
|
9376
|
|
|
9377 /* Check that the word wasn't banned. */
|
|
|
9378 if (was_banned(su, goodword))
|
|
|
9379 return;
|
|
|
9380
|
|
346
|
9381 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
9382 * Remove the common part from "goodword". */
|
|
|
9383 i = badlen - su->su_badlen;
|
|
|
9384 if (i > 0)
|
|
|
9385 {
|
|
|
9386 /* This assumes there was no case folding or it didn't change the
|
|
|
9387 * length... */
|
|
|
9388 p = goodword + STRLEN(goodword) - i;
|
|
|
9389 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
9390 {
|
|
|
9391 badlen = su->su_badlen;
|
|
|
9392 c = *p;
|
|
|
9393 *p = NUL;
|
|
|
9394 }
|
|
|
9395 else
|
|
|
9396 p = NULL;
|
|
|
9397 }
|
|
372
|
9398 else if (i < 0)
|
|
|
9399 {
|
|
|
9400 /* When replacing part of the word check that we actually change
|
|
|
9401 * something. For "the the" a suggestion can be replacing the first
|
|
|
9402 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
9403 if (badlen == (int)STRLEN(goodword)
|
|
372
|
9404 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
9405 return;
|
|
|
9406 }
|
|
|
9407
|
|
346
|
9408
|
|
323
|
9409 if (score <= su->su_maxscore)
|
|
|
9410 {
|
|
366
|
9411 /* Check if the word is already there. Also check the length that is
|
|
|
9412 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
9413 * "thes" -> "these". */
|
|
344
|
9414 stp = &SUG(*gap, 0);
|
|
|
9415 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
9416 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
9417 && stp[i].st_orglen == badlen)
|
|
323
|
9418 {
|
|
|
9419 /* Found it. Remember the lowest score. */
|
|
|
9420 if (stp[i].st_score > score)
|
|
324
|
9421 {
|
|
323
|
9422 stp[i].st_score = score;
|
|
376
|
9423 stp[i].st_altscore = altscore;
|
|
324
|
9424 stp[i].st_had_bonus = had_bonus;
|
|
|
9425 }
|
|
323
|
9426 break;
|
|
|
9427 }
|
|
|
9428
|
|
344
|
9429 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
9430 {
|
|
|
9431 /* Add a suggestion. */
|
|
344
|
9432 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
9433 stp->st_word = vim_strsave(goodword);
|
|
|
9434 if (stp->st_word != NULL)
|
|
|
9435 {
|
|
|
9436 stp->st_score = score;
|
|
351
|
9437 stp->st_altscore = altscore;
|
|
324
|
9438 stp->st_had_bonus = had_bonus;
|
|
346
|
9439 stp->st_orglen = badlen;
|
|
344
|
9440 ++gap->ga_len;
|
|
323
|
9441
|
|
|
9442 /* If we have too many suggestions now, sort the list and keep
|
|
|
9443 * the best suggestions. */
|
|
344
|
9444 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
9445 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
9446 SUG_CLEAN_COUNT(su));
|
|
323
|
9447 }
|
|
|
9448 }
|
|
|
9449 }
|
|
346
|
9450
|
|
|
9451 if (p != NULL)
|
|
|
9452 *p = c; /* restore "goodword" */
|
|
323
|
9453 }
|
|
|
9454
|
|
|
9455 /*
|
|
|
9456 * Add a word to be banned.
|
|
|
9457 */
|
|
|
9458 static void
|
|
|
9459 add_banned(su, word)
|
|
|
9460 suginfo_T *su;
|
|
|
9461 char_u *word;
|
|
|
9462 {
|
|
|
9463 char_u *s = vim_strsave(word);
|
|
|
9464 hash_T hash;
|
|
|
9465 hashitem_T *hi;
|
|
|
9466
|
|
|
9467 if (s != NULL)
|
|
|
9468 {
|
|
|
9469 hash = hash_hash(s);
|
|
|
9470 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
9471 if (HASHITEM_EMPTY(hi))
|
|
|
9472 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
9473 else
|
|
|
9474 vim_free(s);
|
|
323
|
9475 }
|
|
|
9476 }
|
|
|
9477
|
|
|
9478 /*
|
|
|
9479 * Return TRUE if a word appears in the list of banned words.
|
|
|
9480 */
|
|
|
9481 static int
|
|
|
9482 was_banned(su, word)
|
|
|
9483 suginfo_T *su;
|
|
|
9484 char_u *word;
|
|
|
9485 {
|
|
324
|
9486 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
9487
|
|
|
9488 return !HASHITEM_EMPTY(hi);
|
|
323
|
9489 }
|
|
|
9490
|
|
|
9491 /*
|
|
|
9492 * Free the banned words in "su".
|
|
|
9493 */
|
|
|
9494 static void
|
|
|
9495 free_banned(su)
|
|
|
9496 suginfo_T *su;
|
|
|
9497 {
|
|
|
9498 int todo;
|
|
|
9499 hashitem_T *hi;
|
|
|
9500
|
|
|
9501 todo = su->su_banned.ht_used;
|
|
|
9502 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
9503 {
|
|
|
9504 if (!HASHITEM_EMPTY(hi))
|
|
|
9505 {
|
|
|
9506 vim_free(hi->hi_key);
|
|
|
9507 --todo;
|
|
|
9508 }
|
|
|
9509 }
|
|
|
9510 hash_clear(&su->su_banned);
|
|
|
9511 }
|
|
|
9512
|
|
324
|
9513 /*
|
|
|
9514 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
9515 * thus only done for the final results.
|
|
|
9516 */
|
|
|
9517 static void
|
|
|
9518 rescore_suggestions(su)
|
|
|
9519 suginfo_T *su;
|
|
|
9520 {
|
|
|
9521 langp_T *lp;
|
|
|
9522 suggest_T *stp;
|
|
|
9523 char_u sal_badword[MAXWLEN];
|
|
|
9524 int i;
|
|
|
9525
|
|
|
9526 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9527 lp->lp_slang != NULL; ++lp)
|
|
|
9528 {
|
|
|
9529 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9530 {
|
|
|
9531 /* soundfold the bad word */
|
|
375
|
9532 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
9533
|
|
|
9534 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9535 {
|
|
344
|
9536 stp = &SUG(su->su_ga, i);
|
|
324
|
9537 if (!stp->st_had_bonus)
|
|
|
9538 {
|
|
351
|
9539 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
9540 lp->lp_slang, sal_badword);
|
|
|
9541 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
9542 stp->st_altscore = SCORE_BIG;
|
|
|
9543 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
9544 }
|
|
|
9545 }
|
|
|
9546 break;
|
|
|
9547 }
|
|
|
9548 }
|
|
|
9549 }
|
|
|
9550
|
|
323
|
9551 static int
|
|
|
9552 #ifdef __BORLANDC__
|
|
|
9553 _RTLENTRYF
|
|
|
9554 #endif
|
|
|
9555 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
9556
|
|
|
9557 /*
|
|
|
9558 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
9559 */
|
|
|
9560 static int
|
|
|
9561 #ifdef __BORLANDC__
|
|
|
9562 _RTLENTRYF
|
|
|
9563 #endif
|
|
|
9564 sug_compare(s1, s2)
|
|
|
9565 const void *s1;
|
|
|
9566 const void *s2;
|
|
|
9567 {
|
|
|
9568 suggest_T *p1 = (suggest_T *)s1;
|
|
|
9569 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
9570 int n = p1->st_score - p2->st_score;
|
|
|
9571
|
|
|
9572 if (n == 0)
|
|
|
9573 return p1->st_altscore - p2->st_altscore;
|
|
|
9574 return n;
|
|
323
|
9575 }
|
|
|
9576
|
|
|
9577 /*
|
|
|
9578 * Cleanup the suggestions:
|
|
|
9579 * - Sort on score.
|
|
|
9580 * - Remove words that won't be displayed.
|
|
344
|
9581 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
9582 */
|
|
344
|
9583 static int
|
|
|
9584 cleanup_suggestions(gap, maxscore, keep)
|
|
|
9585 garray_T *gap;
|
|
|
9586 int maxscore;
|
|
324
|
9587 int keep; /* nr of suggestions to keep */
|
|
323
|
9588 {
|
|
344
|
9589 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
9590 int i;
|
|
|
9591
|
|
|
9592 /* Sort the list. */
|
|
344
|
9593 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
9594
|
|
|
9595 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
9596 if (gap->ga_len > keep)
|
|
323
|
9597 {
|
|
344
|
9598 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
9599 vim_free(stp[i].st_word);
|
|
344
|
9600 gap->ga_len = keep;
|
|
|
9601 return stp[keep - 1].st_score;
|
|
323
|
9602 }
|
|
344
|
9603 return maxscore;
|
|
323
|
9604 }
|
|
|
9605
|
|
372
|
9606 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
9607 /*
|
|
|
9608 * Soundfold a string, for soundfold().
|
|
|
9609 * Result is in allocated memory, NULL for an error.
|
|
|
9610 */
|
|
|
9611 char_u *
|
|
|
9612 eval_soundfold(word)
|
|
|
9613 char_u *word;
|
|
|
9614 {
|
|
|
9615 langp_T *lp;
|
|
|
9616 char_u sound[MAXWLEN];
|
|
|
9617
|
|
|
9618 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
9619 /* Use the sound-folding of the first language that supports it. */
|
|
|
9620 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9621 lp->lp_slang != NULL; ++lp)
|
|
|
9622 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9623 {
|
|
|
9624 /* soundfold the word */
|
|
375
|
9625 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
9626 return vim_strsave(sound);
|
|
|
9627 }
|
|
|
9628
|
|
|
9629 /* No language with sound folding, return word as-is. */
|
|
|
9630 return vim_strsave(word);
|
|
|
9631 }
|
|
|
9632 #endif
|
|
|
9633
|
|
323
|
9634 /*
|
|
|
9635 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
9636 */
|
|
|
9637 static void
|
|
375
|
9638 spell_soundfold(slang, inword, folded, res)
|
|
|
9639 slang_T *slang;
|
|
|
9640 char_u *inword;
|
|
|
9641 int folded; /* "inword" is already case-folded */
|
|
|
9642 char_u *res;
|
|
|
9643 {
|
|
|
9644 char_u fword[MAXWLEN];
|
|
|
9645 char_u *word;
|
|
|
9646
|
|
|
9647 if (slang->sl_sofo)
|
|
|
9648 /* SOFOFROM and SOFOTO used */
|
|
|
9649 spell_soundfold_sofo(slang, inword, res);
|
|
|
9650 else
|
|
|
9651 {
|
|
|
9652 /* SAL items used. Requires the word to be case-folded. */
|
|
|
9653 if (folded)
|
|
|
9654 word = inword;
|
|
|
9655 else
|
|
|
9656 {
|
|
|
9657 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
9658 word = fword;
|
|
|
9659 }
|
|
|
9660
|
|
|
9661 #ifdef FEAT_MBYTE
|
|
|
9662 if (has_mbyte)
|
|
|
9663 spell_soundfold_wsal(slang, word, res);
|
|
|
9664 else
|
|
|
9665 #endif
|
|
|
9666 spell_soundfold_sal(slang, word, res);
|
|
|
9667 }
|
|
|
9668 }
|
|
|
9669
|
|
|
9670 /*
|
|
|
9671 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
9672 * SOFOTO lines.
|
|
|
9673 */
|
|
|
9674 static void
|
|
|
9675 spell_soundfold_sofo(slang, inword, res)
|
|
|
9676 slang_T *slang;
|
|
|
9677 char_u *inword;
|
|
|
9678 char_u *res;
|
|
|
9679 {
|
|
|
9680 char_u *s;
|
|
|
9681 int ri = 0;
|
|
|
9682 int c;
|
|
|
9683
|
|
|
9684 #ifdef FEAT_MBYTE
|
|
|
9685 if (has_mbyte)
|
|
|
9686 {
|
|
|
9687 int prevc = 0;
|
|
|
9688 int *ip;
|
|
|
9689
|
|
|
9690 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
9691 * 255, sl_sal the rest. */
|
|
|
9692 for (s = inword; *s != NUL; )
|
|
|
9693 {
|
|
474
|
9694 c = mb_cptr2char_adv(&s);
|
|
375
|
9695 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9696 c = ' ';
|
|
|
9697 else if (c < 256)
|
|
|
9698 c = slang->sl_sal_first[c];
|
|
|
9699 else
|
|
|
9700 {
|
|
|
9701 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
9702 if (ip == NULL) /* empty list, can't match */
|
|
|
9703 c = NUL;
|
|
|
9704 else
|
|
|
9705 for (;;) /* find "c" in the list */
|
|
|
9706 {
|
|
|
9707 if (*ip == 0) /* not found */
|
|
|
9708 {
|
|
|
9709 c = NUL;
|
|
|
9710 break;
|
|
|
9711 }
|
|
|
9712 if (*ip == c) /* match! */
|
|
|
9713 {
|
|
|
9714 c = ip[1];
|
|
|
9715 break;
|
|
|
9716 }
|
|
|
9717 ip += 2;
|
|
|
9718 }
|
|
|
9719 }
|
|
|
9720
|
|
|
9721 if (c != NUL && c != prevc)
|
|
|
9722 {
|
|
|
9723 ri += mb_char2bytes(c, res + ri);
|
|
|
9724 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
9725 break;
|
|
|
9726 prevc = c;
|
|
|
9727 }
|
|
|
9728 }
|
|
|
9729 }
|
|
|
9730 else
|
|
|
9731 #endif
|
|
|
9732 {
|
|
|
9733 /* The sl_sal_first[] table contains the translation. */
|
|
|
9734 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
9735 {
|
|
|
9736 if (vim_iswhite(c))
|
|
|
9737 c = ' ';
|
|
|
9738 else
|
|
|
9739 c = slang->sl_sal_first[c];
|
|
|
9740 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
9741 res[ri++] = c;
|
|
|
9742 }
|
|
|
9743 }
|
|
|
9744
|
|
|
9745 res[ri] = NUL;
|
|
|
9746 }
|
|
|
9747
|
|
|
9748 static void
|
|
|
9749 spell_soundfold_sal(slang, inword, res)
|
|
323
|
9750 slang_T *slang;
|
|
|
9751 char_u *inword;
|
|
|
9752 char_u *res;
|
|
|
9753 {
|
|
344
|
9754 salitem_T *smp;
|
|
323
|
9755 char_u word[MAXWLEN];
|
|
375
|
9756 char_u *s = inword;
|
|
323
|
9757 char_u *t;
|
|
344
|
9758 char_u *pf;
|
|
323
|
9759 int i, j, z;
|
|
344
|
9760 int reslen;
|
|
323
|
9761 int n, k = 0;
|
|
|
9762 int z0;
|
|
|
9763 int k0;
|
|
|
9764 int n0;
|
|
|
9765 int c;
|
|
|
9766 int pri;
|
|
|
9767 int p0 = -333;
|
|
|
9768 int c0;
|
|
|
9769
|
|
324
|
9770 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
9771 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
9772 if (slang->sl_rem_accents)
|
|
|
9773 {
|
|
|
9774 t = word;
|
|
375
|
9775 while (*s != NUL)
|
|
323
|
9776 {
|
|
324
|
9777 if (vim_iswhite(*s))
|
|
344
|
9778 {
|
|
|
9779 *t++ = ' ';
|
|
|
9780 s = skipwhite(s);
|
|
|
9781 }
|
|
323
|
9782 else
|
|
|
9783 {
|
|
376
|
9784 if (spell_iswordp_nmw(s))
|
|
323
|
9785 *t++ = *s;
|
|
|
9786 ++s;
|
|
|
9787 }
|
|
|
9788 }
|
|
|
9789 *t = NUL;
|
|
|
9790 }
|
|
|
9791 else
|
|
375
|
9792 STRCPY(word, s);
|
|
323
|
9793
|
|
344
|
9794 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
9795
|
|
|
9796 /*
|
|
|
9797 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
9798 * Changed to keep spaces.
|
|
323
|
9799 */
|
|
344
|
9800 i = reslen = z = 0;
|
|
323
|
9801 while ((c = word[i]) != NUL)
|
|
|
9802 {
|
|
344
|
9803 /* Start with the first rule that has the character in the word. */
|
|
323
|
9804 n = slang->sl_sal_first[c];
|
|
|
9805 z0 = 0;
|
|
|
9806
|
|
|
9807 if (n >= 0)
|
|
|
9808 {
|
|
|
9809 /* check all rules for the same letter */
|
|
344
|
9810 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
9811 {
|
|
344
|
9812 /* Quickly skip entries that don't match the word. Most
|
|
|
9813 * entries are less then three chars, optimize for that. */
|
|
|
9814 k = smp[n].sm_leadlen;
|
|
|
9815 if (k > 1)
|
|
323
|
9816 {
|
|
344
|
9817 if (word[i + 1] != s[1])
|
|
|
9818 continue;
|
|
|
9819 if (k > 2)
|
|
|
9820 {
|
|
|
9821 for (j = 2; j < k; ++j)
|
|
|
9822 if (word[i + j] != s[j])
|
|
|
9823 break;
|
|
|
9824 if (j < k)
|
|
|
9825 continue;
|
|
|
9826 }
|
|
323
|
9827 }
|
|
|
9828
|
|
375
|
9829 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
9830 {
|
|
375
|
9831 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
9832 while (*pf != NUL && *pf != word[i + k])
|
|
|
9833 ++pf;
|
|
|
9834 if (*pf == NUL)
|
|
|
9835 continue;
|
|
|
9836 ++k;
|
|
323
|
9837 }
|
|
344
|
9838 s = smp[n].sm_rules;
|
|
|
9839 pri = 5; /* default priority */
|
|
323
|
9840
|
|
|
9841 p0 = *s;
|
|
|
9842 k0 = k;
|
|
|
9843 while (*s == '-' && k > 1)
|
|
|
9844 {
|
|
|
9845 k--;
|
|
|
9846 s++;
|
|
|
9847 }
|
|
|
9848 if (*s == '<')
|
|
|
9849 s++;
|
|
344
|
9850 if (VIM_ISDIGIT(*s))
|
|
323
|
9851 {
|
|
|
9852 /* determine priority */
|
|
|
9853 pri = *s - '0';
|
|
|
9854 s++;
|
|
|
9855 }
|
|
|
9856 if (*s == '^' && *(s + 1) == '^')
|
|
|
9857 s++;
|
|
|
9858
|
|
|
9859 if (*s == NUL
|
|
|
9860 || (*s == '^'
|
|
324
|
9861 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9862 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
9863 && (*(s + 1) != '$'
|
|
376
|
9864 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9865 || (*s == '$' && i > 0
|
|
376
|
9866 && spell_iswordp(word + i - 1, curbuf)
|
|
|
9867 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9868 {
|
|
|
9869 /* search for followup rules, if: */
|
|
|
9870 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9871 c0 = word[i + k - 1];
|
|
|
9872 n0 = slang->sl_sal_first[c0];
|
|
|
9873
|
|
|
9874 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
9875 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
9876 {
|
|
|
9877 /* test follow-up rule for "word[i + k]" */
|
|
344
|
9878 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
9879 {
|
|
344
|
9880 /* Quickly skip entries that don't match the word.
|
|
|
9881 * */
|
|
|
9882 k0 = smp[n0].sm_leadlen;
|
|
|
9883 if (k0 > 1)
|
|
323
|
9884 {
|
|
344
|
9885 if (word[i + k] != s[1])
|
|
|
9886 continue;
|
|
|
9887 if (k0 > 2)
|
|
|
9888 {
|
|
|
9889 pf = word + i + k + 1;
|
|
|
9890 for (j = 2; j < k0; ++j)
|
|
|
9891 if (*pf++ != s[j])
|
|
|
9892 break;
|
|
|
9893 if (j < k0)
|
|
|
9894 continue;
|
|
|
9895 }
|
|
323
|
9896 }
|
|
344
|
9897 k0 += k - 1;
|
|
|
9898
|
|
375
|
9899 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
9900 {
|
|
344
|
9901 /* Check for match with one of the chars in
|
|
375
|
9902 * "sm_oneof". */
|
|
344
|
9903 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9904 ++pf;
|
|
|
9905 if (*pf == NUL)
|
|
|
9906 continue;
|
|
|
9907 ++k0;
|
|
323
|
9908 }
|
|
344
|
9909
|
|
|
9910 p0 = 5;
|
|
|
9911 s = smp[n0].sm_rules;
|
|
323
|
9912 while (*s == '-')
|
|
|
9913 {
|
|
344
|
9914 /* "k0" gets NOT reduced because
|
|
|
9915 * "if (k0 == k)" */
|
|
323
|
9916 s++;
|
|
|
9917 }
|
|
|
9918 if (*s == '<')
|
|
|
9919 s++;
|
|
344
|
9920 if (VIM_ISDIGIT(*s))
|
|
323
|
9921 {
|
|
|
9922 p0 = *s - '0';
|
|
|
9923 s++;
|
|
|
9924 }
|
|
|
9925
|
|
|
9926 if (*s == NUL
|
|
|
9927 /* *s == '^' cuts */
|
|
|
9928 || (*s == '$'
|
|
376
|
9929 && !spell_iswordp(word + i + k0,
|
|
|
9930 curbuf)))
|
|
323
|
9931 {
|
|
|
9932 if (k0 == k)
|
|
|
9933 /* this is just a piece of the string */
|
|
|
9934 continue;
|
|
|
9935
|
|
|
9936 if (p0 < pri)
|
|
|
9937 /* priority too low */
|
|
|
9938 continue;
|
|
|
9939 /* rule fits; stop search */
|
|
|
9940 break;
|
|
|
9941 }
|
|
|
9942 }
|
|
|
9943
|
|
344
|
9944 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
9945 continue;
|
|
|
9946 }
|
|
|
9947
|
|
|
9948 /* replace string */
|
|
344
|
9949 s = smp[n].sm_to;
|
|
389
|
9950 if (s == NULL)
|
|
|
9951 s = (char_u *)"";
|
|
344
|
9952 pf = smp[n].sm_rules;
|
|
|
9953 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
9954 if (p0 == 1 && z == 0)
|
|
|
9955 {
|
|
|
9956 /* rule with '<' is used */
|
|
344
|
9957 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
9958 || res[reslen - 1] == *s))
|
|
|
9959 reslen--;
|
|
323
|
9960 z0 = 1;
|
|
|
9961 z = 1;
|
|
|
9962 k0 = 0;
|
|
372
|
9963 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
9964 {
|
|
|
9965 word[i + k0] = *s;
|
|
|
9966 k0++;
|
|
|
9967 s++;
|
|
|
9968 }
|
|
|
9969 if (k > k0)
|
|
|
9970 mch_memmove(word + i + k0, word + i + k,
|
|
|
9971 STRLEN(word + i + k) + 1);
|
|
|
9972
|
|
|
9973 /* new "actual letter" */
|
|
|
9974 c = word[i];
|
|
|
9975 }
|
|
|
9976 else
|
|
|
9977 {
|
|
|
9978 /* no '<' rule used */
|
|
|
9979 i += k - 1;
|
|
|
9980 z = 0;
|
|
344
|
9981 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
9982 {
|
|
344
|
9983 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
9984 res[reslen++] = *s;
|
|
323
|
9985 s++;
|
|
|
9986 }
|
|
|
9987 /* new "actual letter" */
|
|
|
9988 c = *s;
|
|
344
|
9989 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
9990 {
|
|
|
9991 if (c != NUL)
|
|
372
|
9992 res[reslen++] = c;
|
|
323
|
9993 mch_memmove(word, word + i + 1,
|
|
|
9994 STRLEN(word + i + 1) + 1);
|
|
|
9995 i = 0;
|
|
|
9996 z0 = 1;
|
|
|
9997 }
|
|
|
9998 }
|
|
|
9999 break;
|
|
|
10000 }
|
|
|
10001 }
|
|
|
10002 }
|
|
324
|
10003 else if (vim_iswhite(c))
|
|
|
10004 {
|
|
|
10005 c = ' ';
|
|
|
10006 k = 1;
|
|
|
10007 }
|
|
323
|
10008
|
|
|
10009 if (z0 == 0)
|
|
|
10010 {
|
|
344
|
10011 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
10012 && (!slang->sl_collapse || reslen == 0
|
|
|
10013 || res[reslen - 1] != c))
|
|
323
|
10014 /* condense only double letters */
|
|
372
|
10015 res[reslen++] = c;
|
|
323
|
10016
|
|
|
10017 i++;
|
|
|
10018 z = 0;
|
|
|
10019 k = 0;
|
|
|
10020 }
|
|
|
10021 }
|
|
|
10022
|
|
344
|
10023 res[reslen] = NUL;
|
|
323
|
10024 }
|
|
|
10025
|
|
372
|
10026 #ifdef FEAT_MBYTE
|
|
|
10027 /*
|
|
|
10028 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
10029 * Multi-byte version of spell_soundfold().
|
|
|
10030 */
|
|
|
10031 static void
|
|
375
|
10032 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
10033 slang_T *slang;
|
|
|
10034 char_u *inword;
|
|
|
10035 char_u *res;
|
|
|
10036 {
|
|
375
|
10037 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
10038 int word[MAXWLEN];
|
|
|
10039 int wres[MAXWLEN];
|
|
|
10040 int l;
|
|
|
10041 char_u *s;
|
|
|
10042 int *ws;
|
|
|
10043 char_u *t;
|
|
|
10044 int *pf;
|
|
|
10045 int i, j, z;
|
|
|
10046 int reslen;
|
|
|
10047 int n, k = 0;
|
|
|
10048 int z0;
|
|
|
10049 int k0;
|
|
|
10050 int n0;
|
|
|
10051 int c;
|
|
|
10052 int pri;
|
|
|
10053 int p0 = -333;
|
|
|
10054 int c0;
|
|
|
10055 int did_white = FALSE;
|
|
|
10056
|
|
|
10057 /*
|
|
|
10058 * Convert the multi-byte string to a wide-character string.
|
|
|
10059 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
10060 * But keep white space.
|
|
|
10061 */
|
|
|
10062 n = 0;
|
|
|
10063 for (s = inword; *s != NUL; )
|
|
|
10064 {
|
|
|
10065 t = s;
|
|
474
|
10066 c = mb_cptr2char_adv(&s);
|
|
372
|
10067 if (slang->sl_rem_accents)
|
|
|
10068 {
|
|
|
10069 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
10070 {
|
|
|
10071 if (did_white)
|
|
|
10072 continue;
|
|
|
10073 c = ' ';
|
|
|
10074 did_white = TRUE;
|
|
|
10075 }
|
|
|
10076 else
|
|
|
10077 {
|
|
|
10078 did_white = FALSE;
|
|
376
|
10079 if (!spell_iswordp_nmw(t))
|
|
372
|
10080 continue;
|
|
|
10081 }
|
|
|
10082 }
|
|
|
10083 word[n++] = c;
|
|
|
10084 }
|
|
|
10085 word[n] = NUL;
|
|
|
10086
|
|
|
10087 /*
|
|
|
10088 * This comes from Aspell phonet.cpp.
|
|
|
10089 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
10090 * Changed to keep spaces.
|
|
|
10091 */
|
|
|
10092 i = reslen = z = 0;
|
|
|
10093 while ((c = word[i]) != NUL)
|
|
|
10094 {
|
|
|
10095 /* Start with the first rule that has the character in the word. */
|
|
|
10096 n = slang->sl_sal_first[c & 0xff];
|
|
|
10097 z0 = 0;
|
|
|
10098
|
|
|
10099 if (n >= 0)
|
|
|
10100 {
|
|
375
|
10101 /* check all rules for the same index byte */
|
|
372
|
10102 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
10103 {
|
|
|
10104 /* Quickly skip entries that don't match the word. Most
|
|
|
10105 * entries are less then three chars, optimize for that. */
|
|
375
|
10106 if (c != ws[0])
|
|
|
10107 continue;
|
|
372
|
10108 k = smp[n].sm_leadlen;
|
|
|
10109 if (k > 1)
|
|
|
10110 {
|
|
|
10111 if (word[i + 1] != ws[1])
|
|
|
10112 continue;
|
|
|
10113 if (k > 2)
|
|
|
10114 {
|
|
|
10115 for (j = 2; j < k; ++j)
|
|
|
10116 if (word[i + j] != ws[j])
|
|
|
10117 break;
|
|
|
10118 if (j < k)
|
|
|
10119 continue;
|
|
|
10120 }
|
|
|
10121 }
|
|
|
10122
|
|
375
|
10123 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
10124 {
|
|
375
|
10125 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
10126 while (*pf != NUL && *pf != word[i + k])
|
|
|
10127 ++pf;
|
|
|
10128 if (*pf == NUL)
|
|
|
10129 continue;
|
|
|
10130 ++k;
|
|
|
10131 }
|
|
|
10132 s = smp[n].sm_rules;
|
|
|
10133 pri = 5; /* default priority */
|
|
|
10134
|
|
|
10135 p0 = *s;
|
|
|
10136 k0 = k;
|
|
|
10137 while (*s == '-' && k > 1)
|
|
|
10138 {
|
|
|
10139 k--;
|
|
|
10140 s++;
|
|
|
10141 }
|
|
|
10142 if (*s == '<')
|
|
|
10143 s++;
|
|
|
10144 if (VIM_ISDIGIT(*s))
|
|
|
10145 {
|
|
|
10146 /* determine priority */
|
|
|
10147 pri = *s - '0';
|
|
|
10148 s++;
|
|
|
10149 }
|
|
|
10150 if (*s == '^' && *(s + 1) == '^')
|
|
|
10151 s++;
|
|
|
10152
|
|
|
10153 if (*s == NUL
|
|
|
10154 || (*s == '^'
|
|
|
10155 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
10156 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
10157 && (*(s + 1) != '$'
|
|
376
|
10158 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
10159 || (*s == '$' && i > 0
|
|
376
|
10160 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
10161 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
10162 {
|
|
|
10163 /* search for followup rules, if: */
|
|
|
10164 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
10165 c0 = word[i + k - 1];
|
|
|
10166 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
10167
|
|
|
10168 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
10169 && p0 != '-' && word[i + k] != NUL)
|
|
|
10170 {
|
|
375
|
10171 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
10172 * all entries with the same index byte. */
|
|
372
|
10173 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
10174 == (c0 & 0xff); ++n0)
|
|
|
10175 {
|
|
|
10176 /* Quickly skip entries that don't match the word.
|
|
375
|
10177 */
|
|
|
10178 if (c0 != ws[0])
|
|
|
10179 continue;
|
|
372
|
10180 k0 = smp[n0].sm_leadlen;
|
|
|
10181 if (k0 > 1)
|
|
|
10182 {
|
|
|
10183 if (word[i + k] != ws[1])
|
|
|
10184 continue;
|
|
|
10185 if (k0 > 2)
|
|
|
10186 {
|
|
|
10187 pf = word + i + k + 1;
|
|
|
10188 for (j = 2; j < k0; ++j)
|
|
|
10189 if (*pf++ != ws[j])
|
|
|
10190 break;
|
|
|
10191 if (j < k0)
|
|
|
10192 continue;
|
|
|
10193 }
|
|
|
10194 }
|
|
|
10195 k0 += k - 1;
|
|
|
10196
|
|
375
|
10197 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
10198 {
|
|
|
10199 /* Check for match with one of the chars in
|
|
375
|
10200 * "sm_oneof". */
|
|
372
|
10201 while (*pf != NUL && *pf != word[i + k0])
|
|
|
10202 ++pf;
|
|
|
10203 if (*pf == NUL)
|
|
|
10204 continue;
|
|
|
10205 ++k0;
|
|
|
10206 }
|
|
|
10207
|
|
|
10208 p0 = 5;
|
|
|
10209 s = smp[n0].sm_rules;
|
|
|
10210 while (*s == '-')
|
|
|
10211 {
|
|
|
10212 /* "k0" gets NOT reduced because
|
|
|
10213 * "if (k0 == k)" */
|
|
|
10214 s++;
|
|
|
10215 }
|
|
|
10216 if (*s == '<')
|
|
|
10217 s++;
|
|
|
10218 if (VIM_ISDIGIT(*s))
|
|
|
10219 {
|
|
|
10220 p0 = *s - '0';
|
|
|
10221 s++;
|
|
|
10222 }
|
|
|
10223
|
|
|
10224 if (*s == NUL
|
|
|
10225 /* *s == '^' cuts */
|
|
|
10226 || (*s == '$'
|
|
376
|
10227 && !spell_iswordp_w(word + i + k0,
|
|
|
10228 curbuf)))
|
|
372
|
10229 {
|
|
|
10230 if (k0 == k)
|
|
|
10231 /* this is just a piece of the string */
|
|
|
10232 continue;
|
|
|
10233
|
|
|
10234 if (p0 < pri)
|
|
|
10235 /* priority too low */
|
|
|
10236 continue;
|
|
|
10237 /* rule fits; stop search */
|
|
|
10238 break;
|
|
|
10239 }
|
|
|
10240 }
|
|
|
10241
|
|
|
10242 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
10243 == (c0 & 0xff))
|
|
|
10244 continue;
|
|
|
10245 }
|
|
|
10246
|
|
|
10247 /* replace string */
|
|
|
10248 ws = smp[n].sm_to_w;
|
|
|
10249 s = smp[n].sm_rules;
|
|
|
10250 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
10251 if (p0 == 1 && z == 0)
|
|
|
10252 {
|
|
|
10253 /* rule with '<' is used */
|
|
389
|
10254 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
10255 && (wres[reslen - 1] == c
|
|
372
|
10256 || wres[reslen - 1] == *ws))
|
|
|
10257 reslen--;
|
|
|
10258 z0 = 1;
|
|
|
10259 z = 1;
|
|
|
10260 k0 = 0;
|
|
389
|
10261 if (ws != NULL)
|
|
|
10262 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
10263 {
|
|
|
10264 word[i + k0] = *ws;
|
|
|
10265 k0++;
|
|
|
10266 ws++;
|
|
|
10267 }
|
|
372
|
10268 if (k > k0)
|
|
|
10269 mch_memmove(word + i + k0, word + i + k,
|
|
|
10270 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
10271
|
|
|
10272 /* new "actual letter" */
|
|
|
10273 c = word[i];
|
|
|
10274 }
|
|
|
10275 else
|
|
|
10276 {
|
|
|
10277 /* no '<' rule used */
|
|
|
10278 i += k - 1;
|
|
|
10279 z = 0;
|
|
389
|
10280 if (ws != NULL)
|
|
|
10281 while (*ws != NUL && ws[1] != NUL
|
|
|
10282 && reslen < MAXWLEN)
|
|
|
10283 {
|
|
|
10284 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
10285 wres[reslen++] = *ws;
|
|
|
10286 ws++;
|
|
|
10287 }
|
|
372
|
10288 /* new "actual letter" */
|
|
389
|
10289 if (ws == NULL)
|
|
|
10290 c = NUL;
|
|
|
10291 else
|
|
|
10292 c = *ws;
|
|
372
|
10293 if (strstr((char *)s, "^^") != NULL)
|
|
|
10294 {
|
|
|
10295 if (c != NUL)
|
|
|
10296 wres[reslen++] = c;
|
|
|
10297 mch_memmove(word, word + i + 1,
|
|
|
10298 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
10299 i = 0;
|
|
|
10300 z0 = 1;
|
|
|
10301 }
|
|
|
10302 }
|
|
|
10303 break;
|
|
|
10304 }
|
|
|
10305 }
|
|
|
10306 }
|
|
|
10307 else if (vim_iswhite(c))
|
|
|
10308 {
|
|
|
10309 c = ' ';
|
|
|
10310 k = 1;
|
|
|
10311 }
|
|
|
10312
|
|
|
10313 if (z0 == 0)
|
|
|
10314 {
|
|
|
10315 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
10316 && (!slang->sl_collapse || reslen == 0
|
|
|
10317 || wres[reslen - 1] != c))
|
|
|
10318 /* condense only double letters */
|
|
|
10319 wres[reslen++] = c;
|
|
|
10320
|
|
|
10321 i++;
|
|
|
10322 z = 0;
|
|
|
10323 k = 0;
|
|
|
10324 }
|
|
|
10325 }
|
|
|
10326
|
|
|
10327 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
10328 l = 0;
|
|
|
10329 for (n = 0; n < reslen; ++n)
|
|
|
10330 {
|
|
|
10331 l += mb_char2bytes(wres[n], res + l);
|
|
|
10332 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
10333 break;
|
|
|
10334 }
|
|
|
10335 res[l] = NUL;
|
|
|
10336 }
|
|
|
10337 #endif
|
|
|
10338
|
|
324
|
10339 /*
|
|
344
|
10340 * Compute a score for two sound-a-like words.
|
|
|
10341 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
10342 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
10343 * avoiding checks that will not be possible.
|
|
|
10344 */
|
|
|
10345 static int
|
|
351
|
10346 soundalike_score(goodstart, badstart)
|
|
|
10347 char_u *goodstart; /* sound-folded good word */
|
|
|
10348 char_u *badstart; /* sound-folded bad word */
|
|
344
|
10349 {
|
|
351
|
10350 char_u *goodsound = goodstart;
|
|
|
10351 char_u *badsound = badstart;
|
|
|
10352 int goodlen;
|
|
|
10353 int badlen;
|
|
344
|
10354 int n;
|
|
|
10355 char_u *pl, *ps;
|
|
|
10356 char_u *pl2, *ps2;
|
|
351
|
10357 int score = 0;
|
|
|
10358
|
|
|
10359 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
10360 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
10361 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
10362 {
|
|
|
10363 score = SCORE_DEL / 2;
|
|
|
10364 if (*badsound == '*')
|
|
|
10365 ++badsound;
|
|
|
10366 else
|
|
|
10367 ++goodsound;
|
|
|
10368 }
|
|
|
10369
|
|
|
10370 goodlen = STRLEN(goodsound);
|
|
|
10371 badlen = STRLEN(badsound);
|
|
344
|
10372
|
|
|
10373 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
10374 * changes. */
|
|
|
10375 n = goodlen - badlen;
|
|
|
10376 if (n < -2 || n > 2)
|
|
|
10377 return SCORE_MAXMAX;
|
|
|
10378
|
|
|
10379 if (n > 0)
|
|
|
10380 {
|
|
351
|
10381 pl = goodsound; /* goodsound is longest */
|
|
344
|
10382 ps = badsound;
|
|
|
10383 }
|
|
|
10384 else
|
|
|
10385 {
|
|
351
|
10386 pl = badsound; /* badsound is longest */
|
|
344
|
10387 ps = goodsound;
|
|
|
10388 }
|
|
|
10389
|
|
|
10390 /* Skip over the identical part. */
|
|
|
10391 while (*pl == *ps && *pl != NUL)
|
|
|
10392 {
|
|
|
10393 ++pl;
|
|
|
10394 ++ps;
|
|
|
10395 }
|
|
|
10396
|
|
|
10397 switch (n)
|
|
|
10398 {
|
|
|
10399 case -2:
|
|
|
10400 case 2:
|
|
|
10401 /*
|
|
|
10402 * Must delete two characters from "pl".
|
|
|
10403 */
|
|
|
10404 ++pl; /* first delete */
|
|
|
10405 while (*pl == *ps)
|
|
|
10406 {
|
|
|
10407 ++pl;
|
|
|
10408 ++ps;
|
|
|
10409 }
|
|
|
10410 /* strings must be equal after second delete */
|
|
|
10411 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
10412 return score + SCORE_DEL * 2;
|
|
344
|
10413
|
|
|
10414 /* Failed to compare. */
|
|
|
10415 break;
|
|
|
10416
|
|
|
10417 case -1:
|
|
|
10418 case 1:
|
|
|
10419 /*
|
|
|
10420 * Minimal one delete from "pl" required.
|
|
|
10421 */
|
|
|
10422
|
|
|
10423 /* 1: delete */
|
|
|
10424 pl2 = pl + 1;
|
|
|
10425 ps2 = ps;
|
|
|
10426 while (*pl2 == *ps2)
|
|
|
10427 {
|
|
|
10428 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10429 return score + SCORE_DEL;
|
|
344
|
10430 ++pl2;
|
|
|
10431 ++ps2;
|
|
|
10432 }
|
|
|
10433
|
|
|
10434 /* 2: delete then swap, then rest must be equal */
|
|
|
10435 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10436 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10437 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
10438
|
|
|
10439 /* 3: delete then substitute, then the rest must be equal */
|
|
|
10440 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10441 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
10442
|
|
|
10443 /* 4: first swap then delete */
|
|
|
10444 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
10445 {
|
|
|
10446 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
10447 ps2 = ps + 2;
|
|
|
10448 while (*pl2 == *ps2)
|
|
|
10449 {
|
|
|
10450 ++pl2;
|
|
|
10451 ++ps2;
|
|
|
10452 }
|
|
|
10453 /* delete a char and then strings must be equal */
|
|
|
10454 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10455 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
10456 }
|
|
|
10457
|
|
|
10458 /* 5: first substitute then delete */
|
|
|
10459 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
10460 ps2 = ps + 1;
|
|
|
10461 while (*pl2 == *ps2)
|
|
|
10462 {
|
|
|
10463 ++pl2;
|
|
|
10464 ++ps2;
|
|
|
10465 }
|
|
|
10466 /* delete a char and then strings must be equal */
|
|
|
10467 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10468 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
10469
|
|
|
10470 /* Failed to compare. */
|
|
|
10471 break;
|
|
|
10472
|
|
|
10473 case 0:
|
|
|
10474 /*
|
|
|
10475 * Lenghts are equal, thus changes must result in same length: An
|
|
|
10476 * insert is only possible in combination with a delete.
|
|
|
10477 * 1: check if for identical strings
|
|
|
10478 */
|
|
|
10479 if (*pl == NUL)
|
|
351
|
10480 return score;
|
|
344
|
10481
|
|
|
10482 /* 2: swap */
|
|
|
10483 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
10484 {
|
|
|
10485 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
10486 ps2 = ps + 2;
|
|
|
10487 while (*pl2 == *ps2)
|
|
|
10488 {
|
|
|
10489 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10490 return score + SCORE_SWAP;
|
|
344
|
10491 ++pl2;
|
|
|
10492 ++ps2;
|
|
|
10493 }
|
|
|
10494 /* 3: swap and swap again */
|
|
|
10495 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10496 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10497 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
10498
|
|
|
10499 /* 4: swap and substitute */
|
|
|
10500 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10501 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
10502 }
|
|
|
10503
|
|
|
10504 /* 5: substitute */
|
|
|
10505 pl2 = pl + 1;
|
|
|
10506 ps2 = ps + 1;
|
|
|
10507 while (*pl2 == *ps2)
|
|
|
10508 {
|
|
|
10509 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10510 return score + SCORE_SUBST;
|
|
344
|
10511 ++pl2;
|
|
|
10512 ++ps2;
|
|
|
10513 }
|
|
|
10514
|
|
|
10515 /* 6: substitute and swap */
|
|
|
10516 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10517 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10518 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
10519
|
|
|
10520 /* 7: substitute and substitute */
|
|
|
10521 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10522 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
10523
|
|
|
10524 /* 8: insert then delete */
|
|
|
10525 pl2 = pl;
|
|
|
10526 ps2 = ps + 1;
|
|
|
10527 while (*pl2 == *ps2)
|
|
|
10528 {
|
|
|
10529 ++pl2;
|
|
|
10530 ++ps2;
|
|
|
10531 }
|
|
|
10532 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10533 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
10534
|
|
|
10535 /* 9: delete then insert */
|
|
|
10536 pl2 = pl + 1;
|
|
|
10537 ps2 = ps;
|
|
|
10538 while (*pl2 == *ps2)
|
|
|
10539 {
|
|
|
10540 ++pl2;
|
|
|
10541 ++ps2;
|
|
|
10542 }
|
|
|
10543 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
10544 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
10545
|
|
|
10546 /* Failed to compare. */
|
|
|
10547 break;
|
|
|
10548 }
|
|
|
10549
|
|
|
10550 return SCORE_MAXMAX;
|
|
|
10551 }
|
|
324
|
10552
|
|
323
|
10553 /*
|
|
|
10554 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
10555 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
10556 *
|
|
323
|
10557 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
10558 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
10559 * characters.
|
|
323
|
10560 */
|
|
|
10561 static int
|
|
|
10562 spell_edit_score(badword, goodword)
|
|
|
10563 char_u *badword;
|
|
|
10564 char_u *goodword;
|
|
|
10565 {
|
|
|
10566 int *cnt;
|
|
372
|
10567 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
10568 int j, i;
|
|
|
10569 int t;
|
|
|
10570 int bc, gc;
|
|
324
|
10571 int pbc, pgc;
|
|
|
10572 #ifdef FEAT_MBYTE
|
|
|
10573 char_u *p;
|
|
|
10574 int wbadword[MAXWLEN];
|
|
|
10575 int wgoodword[MAXWLEN];
|
|
|
10576
|
|
|
10577 if (has_mbyte)
|
|
|
10578 {
|
|
|
10579 /* Get the characters from the multi-byte strings and put them in an
|
|
|
10580 * int array for easy access. */
|
|
|
10581 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
10582 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
10583 wbadword[badlen++] = 0;
|
|
324
|
10584 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
10585 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
10586 wgoodword[goodlen++] = 0;
|
|
324
|
10587 }
|
|
|
10588 else
|
|
|
10589 #endif
|
|
|
10590 {
|
|
|
10591 badlen = STRLEN(badword) + 1;
|
|
|
10592 goodlen = STRLEN(goodword) + 1;
|
|
|
10593 }
|
|
323
|
10594
|
|
|
10595 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
10596 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
10597 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
10598 TRUE);
|
|
324
|
10599 if (cnt == NULL)
|
|
|
10600 return 0; /* out of memory */
|
|
323
|
10601
|
|
|
10602 CNT(0, 0) = 0;
|
|
|
10603 for (j = 1; j <= goodlen; ++j)
|
|
|
10604 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
10605
|
|
|
10606 for (i = 1; i <= badlen; ++i)
|
|
|
10607 {
|
|
|
10608 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
10609 for (j = 1; j <= goodlen; ++j)
|
|
|
10610 {
|
|
324
|
10611 #ifdef FEAT_MBYTE
|
|
|
10612 if (has_mbyte)
|
|
|
10613 {
|
|
|
10614 bc = wbadword[i - 1];
|
|
|
10615 gc = wgoodword[j - 1];
|
|
|
10616 }
|
|
|
10617 else
|
|
|
10618 #endif
|
|
|
10619 {
|
|
|
10620 bc = badword[i - 1];
|
|
|
10621 gc = goodword[j - 1];
|
|
|
10622 }
|
|
323
|
10623 if (bc == gc)
|
|
|
10624 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
10625 else
|
|
|
10626 {
|
|
|
10627 /* Use a better score when there is only a case difference. */
|
|
324
|
10628 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
10629 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
10630 else
|
|
|
10631 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
10632
|
|
324
|
10633 if (i > 1 && j > 1)
|
|
323
|
10634 {
|
|
324
|
10635 #ifdef FEAT_MBYTE
|
|
|
10636 if (has_mbyte)
|
|
|
10637 {
|
|
|
10638 pbc = wbadword[i - 2];
|
|
|
10639 pgc = wgoodword[j - 2];
|
|
|
10640 }
|
|
|
10641 else
|
|
|
10642 #endif
|
|
|
10643 {
|
|
|
10644 pbc = badword[i - 2];
|
|
|
10645 pgc = goodword[j - 2];
|
|
|
10646 }
|
|
|
10647 if (bc == pgc && pbc == gc)
|
|
|
10648 {
|
|
|
10649 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
10650 if (t < CNT(i, j))
|
|
|
10651 CNT(i, j) = t;
|
|
|
10652 }
|
|
323
|
10653 }
|
|
|
10654 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
10655 if (t < CNT(i, j))
|
|
|
10656 CNT(i, j) = t;
|
|
|
10657 t = SCORE_INS + CNT(i, j - 1);
|
|
|
10658 if (t < CNT(i, j))
|
|
|
10659 CNT(i, j) = t;
|
|
|
10660 }
|
|
|
10661 }
|
|
|
10662 }
|
|
344
|
10663
|
|
|
10664 i = CNT(badlen - 1, goodlen - 1);
|
|
|
10665 vim_free(cnt);
|
|
|
10666 return i;
|
|
323
|
10667 }
|
|
307
|
10668
|
|
351
|
10669 /*
|
|
|
10670 * ":spelldump"
|
|
|
10671 */
|
|
|
10672 /*ARGSUSED*/
|
|
|
10673 void
|
|
|
10674 ex_spelldump(eap)
|
|
|
10675 exarg_T *eap;
|
|
|
10676 {
|
|
|
10677 buf_T *buf = curbuf;
|
|
|
10678 langp_T *lp;
|
|
|
10679 slang_T *slang;
|
|
|
10680 idx_T arridx[MAXWLEN];
|
|
|
10681 int curi[MAXWLEN];
|
|
|
10682 char_u word[MAXWLEN];
|
|
|
10683 int c;
|
|
|
10684 char_u *byts;
|
|
|
10685 idx_T *idxs;
|
|
|
10686 linenr_T lnum = 0;
|
|
|
10687 int round;
|
|
|
10688 int depth;
|
|
|
10689 int n;
|
|
|
10690 int flags;
|
|
381
|
10691 char_u *region_names = NULL; /* region names being used */
|
|
|
10692 int do_region = TRUE; /* dump region names and numbers */
|
|
|
10693 char_u *p;
|
|
351
|
10694
|
|
|
10695 if (no_spell_checking())
|
|
|
10696 return;
|
|
|
10697
|
|
|
10698 /* Create a new empty buffer by splitting the window. */
|
|
|
10699 do_cmdline_cmd((char_u *)"new");
|
|
|
10700 if (!bufempty() || !buf_valid(buf))
|
|
|
10701 return;
|
|
|
10702
|
|
381
|
10703 /* Find out if we can support regions: All languages must support the same
|
|
|
10704 * regions or none at all. */
|
|
|
10705 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
10706 {
|
|
|
10707 p = lp->lp_slang->sl_regions;
|
|
|
10708 if (p[0] != 0)
|
|
|
10709 {
|
|
|
10710 if (region_names == NULL) /* first language with regions */
|
|
|
10711 region_names = p;
|
|
|
10712 else if (STRCMP(region_names, p) != 0)
|
|
|
10713 {
|
|
|
10714 do_region = FALSE; /* region names are different */
|
|
|
10715 break;
|
|
|
10716 }
|
|
|
10717 }
|
|
|
10718 }
|
|
|
10719
|
|
|
10720 if (do_region && region_names != NULL)
|
|
|
10721 {
|
|
|
10722 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
10723 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
10724 }
|
|
|
10725 else
|
|
|
10726 do_region = FALSE;
|
|
|
10727
|
|
|
10728 /*
|
|
|
10729 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
10730 */
|
|
351
|
10731 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
10732 {
|
|
|
10733 slang = lp->lp_slang;
|
|
|
10734
|
|
|
10735 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
10736 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
10737
|
|
|
10738 /* round 1: case-folded tree
|
|
|
10739 * round 2: keep-case tree */
|
|
|
10740 for (round = 1; round <= 2; ++round)
|
|
|
10741 {
|
|
|
10742 if (round == 1)
|
|
|
10743 {
|
|
|
10744 byts = slang->sl_fbyts;
|
|
|
10745 idxs = slang->sl_fidxs;
|
|
|
10746 }
|
|
|
10747 else
|
|
|
10748 {
|
|
|
10749 byts = slang->sl_kbyts;
|
|
|
10750 idxs = slang->sl_kidxs;
|
|
|
10751 }
|
|
|
10752 if (byts == NULL)
|
|
|
10753 continue; /* array is empty */
|
|
|
10754
|
|
|
10755 depth = 0;
|
|
|
10756 arridx[0] = 0;
|
|
|
10757 curi[0] = 1;
|
|
|
10758 while (depth >= 0 && !got_int)
|
|
|
10759 {
|
|
|
10760 if (curi[depth] > byts[arridx[depth]])
|
|
|
10761 {
|
|
|
10762 /* Done all bytes at this node, go up one level. */
|
|
|
10763 --depth;
|
|
|
10764 line_breakcheck();
|
|
|
10765 }
|
|
|
10766 else
|
|
|
10767 {
|
|
|
10768 /* Do one more byte at this node. */
|
|
|
10769 n = arridx[depth] + curi[depth];
|
|
|
10770 ++curi[depth];
|
|
|
10771 c = byts[n];
|
|
|
10772 if (c == 0)
|
|
|
10773 {
|
|
|
10774 /* End of word, deal with the word.
|
|
|
10775 * Don't use keep-case words in the fold-case tree,
|
|
|
10776 * they will appear in the keep-case tree.
|
|
|
10777 * Only use the word when the region matches. */
|
|
|
10778 flags = (int)idxs[n];
|
|
|
10779 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
10780 && (do_region
|
|
|
10781 || (flags & WF_REGION) == 0
|
|
390
|
10782 || (((unsigned)flags >> 16)
|
|
351
|
10783 & lp->lp_region) != 0))
|
|
|
10784 {
|
|
|
10785 word[depth] = NUL;
|
|
381
|
10786 if (!do_region)
|
|
|
10787 flags &= ~WF_REGION;
|
|
355
|
10788
|
|
|
10789 /* Dump the basic word if there is no prefix or
|
|
|
10790 * when it's the first one. */
|
|
390
|
10791 c = (unsigned)flags >> 24;
|
|
355
|
10792 if (c == 0 || curi[depth] == 2)
|
|
|
10793 dump_word(word, round, flags, lnum++);
|
|
351
|
10794
|
|
|
10795 /* Apply the prefix, if there is one. */
|
|
355
|
10796 if (c != 0)
|
|
351
|
10797 lnum = apply_prefixes(slang, word, round,
|
|
|
10798 flags, lnum);
|
|
|
10799 }
|
|
|
10800 }
|
|
|
10801 else
|
|
|
10802 {
|
|
|
10803 /* Normal char, go one level deeper. */
|
|
|
10804 word[depth++] = c;
|
|
|
10805 arridx[depth] = idxs[n];
|
|
|
10806 curi[depth] = 1;
|
|
|
10807 }
|
|
|
10808 }
|
|
|
10809 }
|
|
|
10810 }
|
|
|
10811 }
|
|
|
10812
|
|
|
10813 /* Delete the empty line that we started with. */
|
|
|
10814 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
10815 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
10816
|
|
|
10817 redraw_later(NOT_VALID);
|
|
|
10818 }
|
|
|
10819
|
|
|
10820 /*
|
|
|
10821 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
10822 */
|
|
|
10823 static void
|
|
|
10824 dump_word(word, round, flags, lnum)
|
|
|
10825 char_u *word;
|
|
|
10826 int round;
|
|
|
10827 int flags;
|
|
|
10828 linenr_T lnum;
|
|
|
10829 {
|
|
|
10830 int keepcap = FALSE;
|
|
|
10831 char_u *p;
|
|
|
10832 char_u cword[MAXWLEN];
|
|
381
|
10833 char_u badword[MAXWLEN + 10];
|
|
|
10834 int i;
|
|
351
|
10835
|
|
|
10836 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
10837 {
|
|
|
10838 /* Need to fix case according to "flags". */
|
|
|
10839 make_case_word(word, cword, flags);
|
|
|
10840 p = cword;
|
|
|
10841 }
|
|
|
10842 else
|
|
|
10843 {
|
|
|
10844 p = word;
|
|
389
|
10845 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
10846 || (flags & WF_FIXCAP) != 0))
|
|
351
|
10847 keepcap = TRUE;
|
|
|
10848 }
|
|
|
10849
|
|
381
|
10850 /* Add flags and regions after a slash. */
|
|
|
10851 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
10852 {
|
|
|
10853 STRCPY(badword, p);
|
|
|
10854 STRCAT(badword, "/");
|
|
351
|
10855 if (keepcap)
|
|
|
10856 STRCAT(badword, "=");
|
|
|
10857 if (flags & WF_BANNED)
|
|
|
10858 STRCAT(badword, "!");
|
|
|
10859 else if (flags & WF_RARE)
|
|
|
10860 STRCAT(badword, "?");
|
|
381
|
10861 if (flags & WF_REGION)
|
|
|
10862 for (i = 0; i < 7; ++i)
|
|
390
|
10863 if (flags & (0x10000 << i))
|
|
381
|
10864 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
10865 p = badword;
|
|
|
10866 }
|
|
|
10867
|
|
|
10868 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
10869 }
|
|
|
10870
|
|
|
10871 /*
|
|
372
|
10872 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
10873 * "word" and append a line to the buffer.
|
|
351
|
10874 * Return the updated line number.
|
|
|
10875 */
|
|
|
10876 static linenr_T
|
|
|
10877 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
10878 slang_T *slang;
|
|
|
10879 char_u *word; /* case-folded word */
|
|
|
10880 int round;
|
|
|
10881 int flags; /* flags with prefix ID */
|
|
|
10882 linenr_T startlnum;
|
|
|
10883 {
|
|
|
10884 idx_T arridx[MAXWLEN];
|
|
|
10885 int curi[MAXWLEN];
|
|
|
10886 char_u prefix[MAXWLEN];
|
|
455
|
10887 char_u word_up[MAXWLEN];
|
|
|
10888 int has_word_up = FALSE;
|
|
351
|
10889 int c;
|
|
|
10890 char_u *byts;
|
|
|
10891 idx_T *idxs;
|
|
|
10892 linenr_T lnum = startlnum;
|
|
|
10893 int depth;
|
|
|
10894 int n;
|
|
|
10895 int len;
|
|
|
10896 int i;
|
|
|
10897
|
|
455
|
10898 /* if the word starts with a lower-case letter make the word with an
|
|
|
10899 * upper-case letter in word_up[]. */
|
|
|
10900 c = PTR2CHAR(word);
|
|
|
10901 if (SPELL_TOUPPER(c) != c)
|
|
|
10902 {
|
|
|
10903 onecap_copy(word, word_up, TRUE);
|
|
|
10904 has_word_up = TRUE;
|
|
|
10905 }
|
|
|
10906
|
|
351
|
10907 byts = slang->sl_pbyts;
|
|
|
10908 idxs = slang->sl_pidxs;
|
|
|
10909 if (byts != NULL) /* array not is empty */
|
|
|
10910 {
|
|
|
10911 /*
|
|
|
10912 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
10913 * When at the end of a prefix check that it supports "flags".
|
|
351
|
10914 */
|
|
|
10915 depth = 0;
|
|
|
10916 arridx[0] = 0;
|
|
|
10917 curi[0] = 1;
|
|
|
10918 while (depth >= 0 && !got_int)
|
|
|
10919 {
|
|
390
|
10920 n = arridx[depth];
|
|
|
10921 len = byts[n];
|
|
|
10922 if (curi[depth] > len)
|
|
351
|
10923 {
|
|
|
10924 /* Done all bytes at this node, go up one level. */
|
|
|
10925 --depth;
|
|
|
10926 line_breakcheck();
|
|
|
10927 }
|
|
|
10928 else
|
|
|
10929 {
|
|
|
10930 /* Do one more byte at this node. */
|
|
390
|
10931 n += curi[depth];
|
|
351
|
10932 ++curi[depth];
|
|
|
10933 c = byts[n];
|
|
|
10934 if (c == 0)
|
|
|
10935 {
|
|
|
10936 /* End of prefix, find out how many IDs there are. */
|
|
|
10937 for (i = 1; i < len; ++i)
|
|
|
10938 if (byts[n + i] != 0)
|
|
|
10939 break;
|
|
|
10940 curi[depth] += i - 1;
|
|
|
10941
|
|
455
|
10942 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
10943 if (c != 0)
|
|
351
|
10944 {
|
|
376
|
10945 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
10946 dump_word(prefix, round,
|
|
455
|
10947 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
10948 : flags, lnum++);
|
|
351
|
10949 }
|
|
455
|
10950
|
|
|
10951 /* Check for prefix that matches the word when the
|
|
|
10952 * first letter is upper-case, but only if the prefix has
|
|
|
10953 * a condition. */
|
|
|
10954 if (has_word_up)
|
|
|
10955 {
|
|
|
10956 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
10957 TRUE);
|
|
|
10958 if (c != 0)
|
|
|
10959 {
|
|
|
10960 vim_strncpy(prefix + depth, word_up,
|
|
|
10961 MAXWLEN - depth - 1);
|
|
|
10962 dump_word(prefix, round,
|
|
|
10963 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
10964 : flags, lnum++);
|
|
|
10965 }
|
|
|
10966 }
|
|
351
|
10967 }
|
|
|
10968 else
|
|
|
10969 {
|
|
|
10970 /* Normal char, go one level deeper. */
|
|
|
10971 prefix[depth++] = c;
|
|
|
10972 arridx[depth] = idxs[n];
|
|
|
10973 curi[depth] = 1;
|
|
|
10974 }
|
|
|
10975 }
|
|
|
10976 }
|
|
|
10977 }
|
|
|
10978
|
|
|
10979 return lnum;
|
|
|
10980 }
|
|
|
10981
|
|
475
|
10982 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
10983 static int spell_expand_need_cap;
|
|
|
10984
|
|
|
10985 /*
|
|
|
10986 * Find start of the word in front of the cursor. We don't check if it is
|
|
|
10987 * badly spelled, with completion we can only change the word in front of the
|
|
|
10988 * cursor.
|
|
|
10989 * Used for Insert mode completion CTRL-X ?.
|
|
|
10990 * Returns the column number of the word.
|
|
|
10991 */
|
|
|
10992 int
|
|
|
10993 spell_word_start(startcol)
|
|
|
10994 int startcol;
|
|
|
10995 {
|
|
|
10996 char_u *line;
|
|
|
10997 char_u *p;
|
|
|
10998 int col = 0;
|
|
|
10999
|
|
|
11000 if (no_spell_checking())
|
|
|
11001 return startcol;
|
|
|
11002
|
|
|
11003 /* Find a word character before "startcol". */
|
|
|
11004 line = ml_get_curline();
|
|
|
11005 for (p = line + startcol; p > line; )
|
|
|
11006 {
|
|
|
11007 mb_ptr_back(line, p);
|
|
|
11008 if (spell_iswordp_nmw(p))
|
|
|
11009 break;
|
|
|
11010 }
|
|
|
11011
|
|
|
11012 /* Go back to start of the word. */
|
|
|
11013 while (p > line)
|
|
|
11014 {
|
|
|
11015 col = p - line;
|
|
|
11016 mb_ptr_back(line, p);
|
|
|
11017 if (!spell_iswordp(p, curbuf))
|
|
|
11018 break;
|
|
|
11019 col = 0;
|
|
|
11020 }
|
|
|
11021
|
|
|
11022 /* Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
11023 * expand_spelling() is called. Therefore the ugly global variable. */
|
|
|
11024 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
11025
|
|
|
11026 return col;
|
|
|
11027 }
|
|
|
11028
|
|
|
11029 /*
|
|
|
11030 * Get list of spelling suggestions.
|
|
|
11031 * Used for Insert mode completion CTRL-X ?.
|
|
|
11032 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
11033 * allocated strings.
|
|
|
11034 */
|
|
|
11035 /*ARGSUSED*/
|
|
|
11036 int
|
|
|
11037 expand_spelling(lnum, col, pat, matchp)
|
|
|
11038 linenr_T lnum;
|
|
|
11039 int col;
|
|
|
11040 char_u *pat;
|
|
|
11041 char_u ***matchp;
|
|
|
11042 {
|
|
|
11043 garray_T ga;
|
|
|
11044
|
|
|
11045 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap);
|
|
|
11046 *matchp = ga.ga_data;
|
|
|
11047 return ga.ga_len;
|
|
|
11048 }
|
|
|
11049 #endif
|
|
|
11050
|
|
236
|
11051 #endif /* FEAT_SYN_HL */
|
