|
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.
|
|
474
|
54 * Only use it for small word lists!
|
|
|
55 * SPELL_COMPRESS_CNT is in how many words we compress the tree to limit the
|
|
|
56 * amount of memory used (esp. for Italian). */
|
|
469
|
57 #if 0
|
|
|
58 # define SPELL_PRINTTREE
|
|
|
59 # define SPELL_COMPRESS_CNT 1
|
|
|
60 #else
|
|
|
61 # define SPELL_COMPRESS_CNT 1000000
|
|
|
62 #endif
|
|
|
63
|
|
300
|
64 /*
|
|
324
|
65 * Use this to adjust the score after finding suggestions, based on the
|
|
|
66 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
67 * it for every possible suggestion.
|
|
|
68 * Disadvantage: When "the" is typed as "hte" it sounds different and goes
|
|
|
69 * down in the list.
|
|
344
|
70 * Used when 'spellsuggest' is set to "best".
|
|
|
71 */
|
|
|
72 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
73
|
|
|
74 /*
|
|
339
|
75 * Vim spell file format: <HEADER>
|
|
|
76 * <SUGGEST>
|
|
|
77 * <LWORDTREE>
|
|
|
78 * <KWORDTREE>
|
|
|
79 * <PREFIXTREE>
|
|
300
|
80 *
|
|
339
|
81 * <HEADER>: <fileID>
|
|
|
82 * <regioncnt> <regionname> ...
|
|
|
83 * <charflagslen> <charflags>
|
|
|
84 * <fcharslen> <fchars>
|
|
366
|
85 * <midwordlen> <midword>
|
|
481
|
86 * <compoundlen> <compoundtype> <compoundinfo>
|
|
339
|
87 * <prefcondcnt> <prefcond> ...
|
|
300
|
88 *
|
|
481
|
89 * <fileID> 10 bytes "VIMspell10"
|
|
300
|
90 * <regioncnt> 1 byte number of regions following (8 supported)
|
|
307
|
91 * <regionname> 2 bytes Region name: ca, au, etc. Lower case.
|
|
300
|
92 * First <regionname> is region 1.
|
|
|
93 *
|
|
|
94 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
95 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
96 * 0x01 word character CF_WORD
|
|
|
97 * 0x02 upper-case character CF_UPPER
|
|
300
|
98 * <fcharslen> 2 bytes Number of bytes in <fchars>.
|
|
|
99 * <fchars> N bytes Folded characters, first one is for character 128.
|
|
|
100 *
|
|
366
|
101 * <midwordlen> 2 bytes Number of bytes in <midword>.
|
|
|
102 * <midword> N bytes Characters that are word characters only when used
|
|
|
103 * in the middle of a word.
|
|
|
104 *
|
|
481
|
105 * <compoundlen> 2 bytes Number of bytes following for compound info (can
|
|
|
106 * be used to skip it when it's not understood).
|
|
|
107 *
|
|
|
108 * <compoundtype 1 byte 1: compound words using <comp1minlen> and
|
|
|
109 * <comp1flags>
|
|
|
110 *
|
|
|
111 * <comp1minlen> 1 byte minimal word length for compounding
|
|
|
112 *
|
|
|
113 * <comp1flags> N bytes flags used for compounding words
|
|
|
114 *
|
|
|
115 *
|
|
339
|
116 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
117 *
|
|
|
118 * <prefcond> : <condlen> <condstr>
|
|
|
119 *
|
|
|
120 * <condlen> 1 byte Length of <condstr>.
|
|
|
121 *
|
|
|
122 * <condstr> N bytes Condition for the prefix.
|
|
|
123 *
|
|
300
|
124 *
|
|
323
|
125 * <SUGGEST> : <repcount> <rep> ...
|
|
|
126 * <salflags> <salcount> <sal> ...
|
|
|
127 * <maplen> <mapstr>
|
|
|
128 *
|
|
|
129 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
|
130 *
|
|
|
131 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
|
132 *
|
|
|
133 * <repfromlen> 1 byte length of <repfrom>
|
|
|
134 *
|
|
|
135 * <repfrom> N bytes "from" part of replacement
|
|
|
136 *
|
|
|
137 * <reptolen> 1 byte length of <repto>
|
|
|
138 *
|
|
|
139 * <repto> N bytes "to" part of replacement
|
|
300
|
140 *
|
|
323
|
141 * <salflags> 1 byte flags for soundsalike conversion:
|
|
|
142 * SAL_F0LLOWUP
|
|
|
143 * SAL_COLLAPSE
|
|
|
144 * SAL_REM_ACCENTS
|
|
375
|
145 * SAL_SOFO: SOFOFROM and SOFOTO used instead of SAL
|
|
|
146 *
|
|
|
147 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
148 *
|
|
|
149 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
|
150 *
|
|
375
|
151 * <salfromlen> 1-2 bytes length of <salfrom> (2 bytes for SAL_SOFO)
|
|
323
|
152 *
|
|
|
153 * <salfrom> N bytes "from" part of soundsalike
|
|
|
154 *
|
|
375
|
155 * <saltolen> 1-2 bytes length of <salto> (2 bytes for SAL_SOFO)
|
|
323
|
156 *
|
|
|
157 * <salto> N bytes "to" part of soundsalike
|
|
|
158 *
|
|
|
159 * <maplen> 2 bytes length of <mapstr>, MSB first
|
|
|
160 *
|
|
|
161 * <mapstr> N bytes String with sequences of similar characters,
|
|
|
162 * separated by slashes.
|
|
300
|
163 *
|
|
|
164 *
|
|
|
165 * <LWORDTREE>: <wordtree>
|
|
|
166 *
|
|
339
|
167 * <KWORDTREE>: <wordtree>
|
|
|
168 *
|
|
|
169 * <PREFIXTREE>: <wordtree>
|
|
|
170 *
|
|
|
171 *
|
|
300
|
172 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
173 *
|
|
|
174 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
175 *
|
|
|
176 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
177 *
|
|
|
178 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
179 * follow in sorted order.
|
|
|
180 *
|
|
339
|
181 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
481
|
182 * | <flags> [<flags2>] [<region>] [<affixID>]
|
|
|
183 * | [<pflags>] <affixID> <prefcondnr> ]
|
|
300
|
184 *
|
|
|
185 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
186 * BY_NOFLAGS: End of word without flags and for all
|
|
|
187 * regions.
|
|
481
|
188 * For PREFIXTREE <affixID> and
|
|
366
|
189 * <prefcondnr> follow.
|
|
390
|
190 * BY_FLAGS: End of word, <flags> follow.
|
|
481
|
191 * For PREFIXTREE <pflags>, <affixID>
|
|
455
|
192 * and <prefcondnr> follow.
|
|
390
|
193 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
194 * follow. Not used in PREFIXTREE.
|
|
|
195 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
196 * and <xbyte> follow.
|
|
|
197 *
|
|
|
198 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
199 *
|
|
|
200 * <xbyte> 1 byte byte value of the sibling.
|
|
|
201 *
|
|
|
202 * <flags> 1 byte bitmask of:
|
|
|
203 * WF_ALLCAP word must have only capitals
|
|
|
204 * WF_ONECAP first char of word must be capital
|
|
389
|
205 * WF_KEEPCAP keep-case word
|
|
|
206 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
207 * WF_RARE rare word
|
|
389
|
208 * WF_BANNED bad word
|
|
300
|
209 * WF_REGION <region> follows
|
|
481
|
210 * WF_AFX <affixID> follows
|
|
300
|
211 *
|
|
390
|
212 * <flags2> 1 byte Only used when there are postponed prefixes.
|
|
|
213 * Bitmask of:
|
|
|
214 * WF_HAS_AFF >> 8 word includes affix
|
|
|
215 *
|
|
455
|
216 * <pflags> 1 byte bitmask of:
|
|
|
217 * WFP_RARE rare prefix
|
|
|
218 * WFP_NC non-combining prefix
|
|
|
219 * WFP_UP letter after prefix made upper case
|
|
|
220 *
|
|
300
|
221 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
222 * omitted it's valid in all regions.
|
|
|
223 * Lowest bit is for region 1.
|
|
|
224 *
|
|
481
|
225 * <affixID> 1 byte ID of affix that can be used with this word. In
|
|
339
|
226 * PREFIXTREE used for the required prefix ID.
|
|
|
227 *
|
|
|
228 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
229 * from HEADER.
|
|
300
|
230 *
|
|
|
231 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
232 */
|
|
|
233
|
|
223
|
234 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
235 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
236 #endif
|
|
|
237
|
|
|
238 #include "vim.h"
|
|
|
239
|
|
|
240 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
241
|
|
|
242 #ifdef HAVE_FCNTL_H
|
|
|
243 # include <fcntl.h>
|
|
|
244 #endif
|
|
|
245
|
|
323
|
246 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
247 Some places assume a word length fits in a
|
|
|
248 byte, thus it can't be above 255. */
|
|
226
|
249
|
|
324
|
250 /* Type used for indexes in the word tree need to be at least 3 bytes. If int
|
|
|
251 * is 8 bytes we could use something smaller, but what? */
|
|
|
252 #if SIZEOF_INT > 2
|
|
|
253 typedef int idx_T;
|
|
|
254 #else
|
|
|
255 typedef long idx_T;
|
|
|
256 #endif
|
|
|
257
|
|
|
258 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
259 * comes above it. */
|
|
300
|
260 #define WF_REGION 0x01 /* region byte follows */
|
|
|
261 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
262 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
263 #define WF_RARE 0x08 /* rare word */
|
|
307
|
264 #define WF_BANNED 0x10 /* bad word */
|
|
481
|
265 #define WF_AFX 0x20 /* affix ID follows */
|
|
389
|
266 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
267 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
268
|
|
390
|
269 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
270 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
|
271
|
|
389
|
272 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
273
|
|
455
|
274 /* flags for <pflags> */
|
|
|
275 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
276 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
277 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
278
|
|
481
|
279 /* Flags for postponed prefixes. Must be above affixID (one byte)
|
|
390
|
280 * and prefcondnr (two bytes). */
|
|
455
|
281 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
282 * postponed prefix */
|
|
|
283 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
284 * postponed prefix */
|
|
|
285 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
286 * postponed prefix */
|
|
390
|
287
|
|
|
288 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
289 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
290 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
291 * postponed prefix: no <pflags> */
|
|
|
292 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
293 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
294 * postponed prefix: <pflags> follows */
|
|
|
295 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
296 * follow; never used in prefix tree */
|
|
|
297 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
298
|
|
323
|
299 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
300 * and si_sal. Not for sl_sal!
|
|
323
|
301 * One replacement: from "ft_from" to "ft_to". */
|
|
|
302 typedef struct fromto_S
|
|
236
|
303 {
|
|
323
|
304 char_u *ft_from;
|
|
|
305 char_u *ft_to;
|
|
|
306 } fromto_T;
|
|
236
|
307
|
|
344
|
308 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
309 * The info is split for quick processing by spell_soundfold().
|
|
|
310 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
311 typedef struct salitem_S
|
|
|
312 {
|
|
|
313 char_u *sm_lead; /* leading letters */
|
|
|
314 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
315 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
316 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
317 char_u *sm_to; /* replacement. */
|
|
372
|
318 #ifdef FEAT_MBYTE
|
|
|
319 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
320 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
321 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
322 #endif
|
|
344
|
323 } salitem_T;
|
|
|
324
|
|
375
|
325 #ifdef FEAT_MBYTE
|
|
|
326 typedef int salfirst_T;
|
|
|
327 #else
|
|
|
328 typedef short salfirst_T;
|
|
|
329 #endif
|
|
|
330
|
|
236
|
331 /*
|
|
243
|
332 * Structure used to store words and other info for one language, loaded from
|
|
|
333 * a .spl file.
|
|
300
|
334 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
335 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
336 *
|
|
|
337 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
338 * the number of possible bytes, sorted on byte value:
|
|
|
339 * <len> <byte1> <byte2> ...
|
|
|
340 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
341 * byte in "byts".
|
|
|
342 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
481
|
343 * the flags, region mask and affixID for the word. There may be several
|
|
|
344 * zeros in sequence for alternative flag/region/affixID combinations.
|
|
236
|
345 */
|
|
|
346 typedef struct slang_S slang_T;
|
|
|
347 struct slang_S
|
|
|
348 {
|
|
|
349 slang_T *sl_next; /* next language */
|
|
|
350 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
351 char_u *sl_fname; /* name of .spl file */
|
|
323
|
352 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
353
|
|
300
|
354 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
355 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
356 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
357 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
358 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
359 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
360
|
|
236
|
361 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
362
|
|
376
|
363 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
364
|
|
481
|
365 int sl_compminlen; /* COMPOUNDMIN */
|
|
|
366 char_u *sl_compflags; /* COMPOUNDFLAGS (NULL when no compounding) */
|
|
|
367
|
|
339
|
368 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
369 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
370
|
|
323
|
371 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
372 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
373 there is none */
|
|
344
|
374 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
375 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
376 there is none */
|
|
375
|
377 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
378 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
379 * "sl_sal" is a list of wide char lists. */
|
|
323
|
380 int sl_followup; /* SAL followup */
|
|
|
381 int sl_collapse; /* SAL collapse_result */
|
|
|
382 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
383 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
384 #ifdef FEAT_MBYTE
|
|
|
385 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
386 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
387 #else
|
|
|
388 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
389 #endif
|
|
236
|
390 };
|
|
|
391
|
|
243
|
392 /* First language that is loaded, start of the linked list of loaded
|
|
|
393 * languages. */
|
|
236
|
394 static slang_T *first_lang = NULL;
|
|
|
395
|
|
323
|
396 /* Flags used in .spl file for soundsalike flags. */
|
|
|
397 #define SAL_F0LLOWUP 1
|
|
|
398 #define SAL_COLLAPSE 2
|
|
|
399 #define SAL_REM_ACCENTS 4
|
|
375
|
400 #define SAL_SOFO 8 /* SOFOFROM and SOFOTO instead of SAL */
|
|
323
|
401
|
|
236
|
402 /*
|
|
|
403 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
404 */
|
|
|
405 typedef struct langp_S
|
|
|
406 {
|
|
|
407 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
408 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
409 } langp_T;
|
|
|
410
|
|
|
411 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
412
|
|
307
|
413 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
414
|
|
|
415 /* Result values. Lower number is accepted over higher one. */
|
|
|
416 #define SP_BANNED -1
|
|
236
|
417 #define SP_OK 0
|
|
307
|
418 #define SP_RARE 1
|
|
|
419 #define SP_LOCAL 2
|
|
|
420 #define SP_BAD 3
|
|
236
|
421
|
|
481
|
422 #define VIMSPELLMAGIC "VIMspell10" /* string at start of Vim spell file */
|
|
236
|
423 #define VIMSPELLMAGICL 10
|
|
|
424
|
|
381
|
425 /* file used for "zG" and "zW" */
|
|
385
|
426 static char_u *int_wordlist = NULL;
|
|
381
|
427
|
|
236
|
428 /*
|
|
323
|
429 * Information used when looking for suggestions.
|
|
|
430 */
|
|
|
431 typedef struct suginfo_S
|
|
|
432 {
|
|
|
433 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
434 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
435 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
436 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
437 char_u *su_badptr; /* start of bad word in line */
|
|
|
438 int su_badlen; /* length of detected bad word in line */
|
|
346
|
439 int su_badflags; /* caps flags for bad word */
|
|
323
|
440 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
441 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
442 hashtab_T su_banned; /* table with banned words */
|
|
|
443 } suginfo_T;
|
|
|
444
|
|
|
445 /* One word suggestion. Used in "si_ga". */
|
|
|
446 typedef struct suggest_S
|
|
|
447 {
|
|
|
448 char_u *st_word; /* suggested word, allocated string */
|
|
|
449 int st_orglen; /* length of replaced text */
|
|
|
450 int st_score; /* lower is better */
|
|
344
|
451 int st_altscore; /* used when st_score compares equal */
|
|
|
452 int st_salscore; /* st_score is for soundalike */
|
|
324
|
453 int st_had_bonus; /* bonus already included in score */
|
|
323
|
454 } suggest_T;
|
|
|
455
|
|
344
|
456 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
457
|
|
324
|
458 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
459 * called the score may change, thus we need to keep more than what is
|
|
|
460 * displayed. */
|
|
351
|
461 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
462
|
|
|
463 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
464 * of suggestions that are not going to be displayed. */
|
|
344
|
465 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
466
|
|
|
467 /* score for various changes */
|
|
344
|
468 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
469 #define SCORE_ICASE 52 /* slightly different case */
|
|
|
470 #define SCORE_REGION 70 /* word is for different region */
|
|
|
471 #define SCORE_RARE 180 /* rare word */
|
|
|
472 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
473 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
474 #define SCORE_REP 87 /* REP replacement */
|
|
|
475 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
476 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
479
|
477 #define SCORE_SUBCOMP 33 /* substitute a composing character */
|
|
324
|
478 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
479 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
479
|
480 #define SCORE_DELCOMP 28 /* delete a composing character */
|
|
324
|
481 #define SCORE_INS 96 /* insert a character */
|
|
358
|
482 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
475
|
483 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
484 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
485
|
|
372
|
486 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
487 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
488 * 350 allows for about three changes. */
|
|
344
|
489
|
|
|
490 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
491 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
492
|
|
|
493 /*
|
|
236
|
494 * Structure to store info for word matching.
|
|
|
495 */
|
|
|
496 typedef struct matchinf_S
|
|
|
497 {
|
|
|
498 langp_T *mi_lp; /* info for language and region */
|
|
243
|
499
|
|
|
500 /* pointers to original text to be checked */
|
|
236
|
501 char_u *mi_word; /* start of word being checked */
|
|
339
|
502 char_u *mi_end; /* end of matching word so far */
|
|
243
|
503 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
504 char_u *mi_cend; /* char after what was used for
|
|
|
505 mi_capflags */
|
|
243
|
506
|
|
|
507 /* case-folded text */
|
|
|
508 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
509 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
510
|
|
339
|
511 /* for when checking word after a prefix */
|
|
|
512 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
481
|
513 affixID/condition */
|
|
339
|
514 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
515 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
516 #ifdef FEAT_MBYTE
|
|
|
517 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
518 case */
|
|
|
519 #else
|
|
|
520 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
521 #endif
|
|
339
|
522
|
|
481
|
523 /* for when checking a compound word */
|
|
|
524 int mi_compoff; /* start of following word offset */
|
|
|
525
|
|
243
|
526 /* others */
|
|
236
|
527 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
528 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
529 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
530 } matchinf_T;
|
|
|
531
|
|
307
|
532 /*
|
|
|
533 * The tables used for recognizing word characters according to spelling.
|
|
|
534 * These are only used for the first 256 characters of 'encoding'.
|
|
|
535 */
|
|
|
536 typedef struct spelltab_S
|
|
|
537 {
|
|
|
538 char_u st_isw[256]; /* flags: is word char */
|
|
|
539 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
540 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
541 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
542 } spelltab_T;
|
|
|
543
|
|
|
544 static spelltab_T spelltab;
|
|
|
545 static int did_set_spelltab;
|
|
|
546
|
|
324
|
547 #define CF_WORD 0x01
|
|
|
548 #define CF_UPPER 0x02
|
|
307
|
549
|
|
|
550 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
551 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
552 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
553 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
554 #ifdef FEAT_MBYTE
|
|
|
555 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
556 #endif
|
|
339
|
557 static void write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
558
|
|
|
559 /*
|
|
344
|
560 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
561 */
|
|
|
562 typedef enum
|
|
|
563 {
|
|
344
|
564 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
565 * ends); if badword ends there is a match, otherwise
|
|
|
566 * try splitting word. */
|
|
375
|
567 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
568 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
569 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
570 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
571 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
572 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
573 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
574 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
575 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
576 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
577 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
578 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
579 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
580 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
581 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
582 STATE_FINAL /* End of this node. */
|
|
|
583 } state_T;
|
|
|
584
|
|
|
585 /*
|
|
346
|
586 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
587 */
|
|
|
588 typedef struct trystate_S
|
|
|
589 {
|
|
330
|
590 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
591 int ts_score; /* score */
|
|
344
|
592 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
593 short ts_curi; /* index in list of child nodes */
|
|
|
594 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
595 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
596 char_u ts_twordlen; /* valid length of tword[] */
|
|
375
|
597 char_u ts_prefixdepth; /* stack depth for end of prefix or PREFIXTREE
|
|
|
598 * or NOPREFIX */
|
|
330
|
599 #ifdef FEAT_MBYTE
|
|
|
600 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
601 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
602 char_u ts_isdiff; /* DIFF_ values */
|
|
|
603 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
604 #endif
|
|
323
|
605 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
330
|
606 char_u ts_save_splitoff; /* su_splitoff saved here */
|
|
346
|
607 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
608 } trystate_T;
|
|
|
609
|
|
330
|
610 /* values for ts_isdiff */
|
|
|
611 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
612 #define DIFF_YES 1 /* different byte found */
|
|
|
613 #define DIFF_INSERT 2 /* inserting character */
|
|
|
614
|
|
375
|
615 /* special values ts_prefixdepth */
|
|
|
616 #define PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
617 #define NOPREFIX 0xff /* not using prefixes */
|
|
|
618
|
|
339
|
619 /* mode values for find_word */
|
|
481
|
620 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
621 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
622 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
623 #define FIND_COMPOUND 3 /* find case-folded compound word */
|
|
|
624 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */
|
|
339
|
625
|
|
236
|
626 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
627 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
628 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
629 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
455
|
630 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
339
|
631 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
632 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
633 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
634 static int no_spell_checking __ARGS((void));
|
|
307
|
635 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
636 static char_u *spell_enc __ARGS((void));
|
|
385
|
637 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
638 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
639 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
640 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
381
|
641 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
642 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
643 #ifdef FEAT_MBYTE
|
|
|
644 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
645 #endif
|
|
339
|
646 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
647 static void clear_midword __ARGS((buf_T *buf));
|
|
|
648 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
649 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
650 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
651 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
652 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
389
|
653 static int set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
654 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
655 static void write_spell_chartab __ARGS((FILE *fd));
|
|
|
656 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
657 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
388
|
658 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
659 #ifdef FEAT_EVAL
|
|
|
660 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
661 #endif
|
|
|
662 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
663 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
664 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
665 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
666 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
667 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
668 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
669 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
670 #ifdef FEAT_MBYTE
|
|
|
671 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
672 #endif
|
|
323
|
673 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
674 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
675 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
676 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
677 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
678 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
679 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
680 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
681 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
682 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
683 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
684 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
685 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
686 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
687 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
688 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
689 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
690 #ifdef FEAT_MBYTE
|
|
375
|
691 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
692 #endif
|
|
344
|
693 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
694 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
695 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
696 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
697
|
|
324
|
698 /*
|
|
|
699 * Use our own character-case definitions, because the current locale may
|
|
|
700 * differ from what the .spl file uses.
|
|
|
701 * These must not be called with negative number!
|
|
|
702 */
|
|
|
703 #ifndef FEAT_MBYTE
|
|
|
704 /* Non-multi-byte implementation. */
|
|
|
705 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
706 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
707 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
708 #else
|
|
447
|
709 # if defined(HAVE_WCHAR_H)
|
|
|
710 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
711 # endif
|
|
324
|
712 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
713 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
714 * the "w" library function for characters above 255 if available. */
|
|
|
715 # ifdef HAVE_TOWLOWER
|
|
|
716 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
717 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
718 # else
|
|
|
719 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
720 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
721 # endif
|
|
|
722
|
|
|
723 # ifdef HAVE_TOWUPPER
|
|
|
724 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
725 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
726 # else
|
|
|
727 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
728 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
729 # endif
|
|
|
730
|
|
|
731 # ifdef HAVE_ISWUPPER
|
|
|
732 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
733 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
734 # else
|
|
|
735 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
736 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
737 # endif
|
|
|
738 #endif
|
|
|
739
|
|
307
|
740
|
|
|
741 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
742 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
469
|
743 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
744
|
|
|
745 /*
|
|
|
746 * Main spell-checking function.
|
|
300
|
747 * "ptr" points to a character that could be the start of a word.
|
|
236
|
748 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
749 * or when it's OK it remains unchanged.
|
|
|
750 * This must only be called when 'spelllang' is not empty.
|
|
323
|
751 *
|
|
385
|
752 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
753 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
754 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
755 * worry.
|
|
323
|
756 *
|
|
236
|
757 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
758 * caller can skip over the word.
|
|
|
759 */
|
|
|
760 int
|
|
385
|
761 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
762 win_T *wp; /* current window */
|
|
|
763 char_u *ptr;
|
|
|
764 int *attrp;
|
|
385
|
765 int *capcol; /* column to check for Capital */
|
|
236
|
766 {
|
|
|
767 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
768 be passed to functions quickly. */
|
|
344
|
769 int nrlen = 0; /* found a number first */
|
|
385
|
770 int c;
|
|
236
|
771
|
|
307
|
772 /* A word never starts at a space or a control character. Return quickly
|
|
|
773 * then, skipping over the character. */
|
|
|
774 if (*ptr <= ' ')
|
|
|
775 return 1;
|
|
236
|
776
|
|
344
|
777 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
778 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
779 * "3GPP". */
|
|
300
|
780 if (*ptr >= '0' && *ptr <= '9')
|
|
|
781 {
|
|
316
|
782 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
783 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
784 else
|
|
|
785 {
|
|
344
|
786 mi.mi_end = skipdigits(ptr);
|
|
|
787 nrlen = mi.mi_end - ptr;
|
|
300
|
788 }
|
|
376
|
789 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
790 return (int)(mi.mi_end - ptr);
|
|
346
|
791
|
|
|
792 /* Try including the digits in the word. */
|
|
|
793 mi.mi_fend = ptr + nrlen;
|
|
|
794 }
|
|
|
795 else
|
|
|
796 mi.mi_fend = ptr;
|
|
|
797
|
|
|
798 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
799 mi.mi_word = ptr;
|
|
376
|
800 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
801 {
|
|
|
802 do
|
|
|
803 {
|
|
307
|
804 mb_ptr_adv(mi.mi_fend);
|
|
376
|
805 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
806
|
|
|
807 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
808 {
|
|
|
809 /* Check word starting with capital letter. */
|
|
455
|
810 c = PTR2CHAR(ptr);
|
|
385
|
811 if (!SPELL_ISUPPER(c))
|
|
|
812 {
|
|
|
813 *attrp = highlight_attr[HLF_SPC];
|
|
|
814 return (int)(mi.mi_fend - ptr);
|
|
|
815 }
|
|
|
816 }
|
|
|
817 }
|
|
|
818 if (capcol != NULL)
|
|
|
819 *capcol = -1;
|
|
344
|
820
|
|
|
821 /* We always use the characters up to the next non-word character,
|
|
|
822 * also for bad words. */
|
|
|
823 mi.mi_end = mi.mi_fend;
|
|
|
824
|
|
|
825 /* Check caps type later. */
|
|
|
826 mi.mi_capflags = 0;
|
|
|
827 mi.mi_cend = NULL;
|
|
376
|
828 mi.mi_buf = wp->w_buffer;
|
|
344
|
829
|
|
|
830 /* Include one non-word character so that we can check for the
|
|
|
831 * word end. */
|
|
|
832 if (*mi.mi_fend != NUL)
|
|
|
833 mb_ptr_adv(mi.mi_fend);
|
|
|
834
|
|
|
835 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
836 MAXWLEN + 1);
|
|
|
837 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
838
|
|
|
839 /* The word is bad unless we recognize it. */
|
|
|
840 mi.mi_result = SP_BAD;
|
|
|
841
|
|
|
842 /*
|
|
|
843 * Loop over the languages specified in 'spelllang'.
|
|
|
844 * We check them all, because a matching word may be longer than an
|
|
|
845 * already found matching word.
|
|
|
846 */
|
|
|
847 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
848 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
849 {
|
|
|
850 /* Check for a matching word in case-folded words. */
|
|
|
851 find_word(&mi, FIND_FOLDWORD);
|
|
|
852
|
|
|
853 /* Check for a matching word in keep-case words. */
|
|
|
854 find_word(&mi, FIND_KEEPWORD);
|
|
|
855
|
|
|
856 /* Check for matching prefixes. */
|
|
|
857 find_prefix(&mi);
|
|
|
858 }
|
|
|
859
|
|
|
860 if (mi.mi_result != SP_OK)
|
|
|
861 {
|
|
346
|
862 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
863 * rare and local words, e.g., "3GPP". */
|
|
344
|
864 if (nrlen > 0)
|
|
346
|
865 {
|
|
|
866 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
867 return nrlen;
|
|
|
868 }
|
|
344
|
869
|
|
|
870 /* When we are at a non-word character there is no error, just
|
|
|
871 * skip over the character (try looking for a word after it). */
|
|
390
|
872 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
873 {
|
|
385
|
874 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
875 {
|
|
|
876 regmatch_T regmatch;
|
|
|
877
|
|
|
878 /* Check for end of sentence. */
|
|
|
879 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
880 regmatch.rm_ic = FALSE;
|
|
|
881 if (vim_regexec(®match, ptr, 0))
|
|
|
882 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
883 }
|
|
|
884
|
|
344
|
885 #ifdef FEAT_MBYTE
|
|
|
886 if (has_mbyte)
|
|
474
|
887 return (*mb_ptr2len)(ptr);
|
|
344
|
888 #endif
|
|
|
889 return 1;
|
|
300
|
890 }
|
|
243
|
891
|
|
344
|
892 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
893 *attrp = highlight_attr[HLF_SPB];
|
|
|
894 else if (mi.mi_result == SP_RARE)
|
|
|
895 *attrp = highlight_attr[HLF_SPR];
|
|
|
896 else
|
|
|
897 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
898 }
|
|
|
899
|
|
300
|
900 return (int)(mi.mi_end - ptr);
|
|
236
|
901 }
|
|
|
902
|
|
|
903 /*
|
|
300
|
904 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
905 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
906 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
907 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
908 * tree.
|
|
300
|
909 *
|
|
|
910 * For a match mip->mi_result is updated.
|
|
243
|
911 */
|
|
|
912 static void
|
|
339
|
913 find_word(mip, mode)
|
|
243
|
914 matchinf_T *mip;
|
|
339
|
915 int mode;
|
|
243
|
916 {
|
|
324
|
917 idx_T arridx = 0;
|
|
300
|
918 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
919 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
920 int endidxcnt = 0;
|
|
|
921 int len;
|
|
|
922 int wlen = 0;
|
|
|
923 int flen;
|
|
|
924 int c;
|
|
|
925 char_u *ptr;
|
|
324
|
926 idx_T lo, hi, m;
|
|
243
|
927 #ifdef FEAT_MBYTE
|
|
300
|
928 char_u *s;
|
|
339
|
929 char_u *p;
|
|
307
|
930 #endif
|
|
|
931 int res = SP_BAD;
|
|
300
|
932 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
933 unsigned flags;
|
|
|
934 char_u *byts;
|
|
324
|
935 idx_T *idxs;
|
|
481
|
936 int word_ends;
|
|
|
937
|
|
|
938 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND)
|
|
236
|
939 {
|
|
300
|
940 /* Check for word with matching case in keep-case tree. */
|
|
|
941 ptr = mip->mi_word;
|
|
|
942 flen = 9999; /* no case folding, always enough bytes */
|
|
|
943 byts = slang->sl_kbyts;
|
|
|
944 idxs = slang->sl_kidxs;
|
|
481
|
945
|
|
|
946 if (mode == FIND_KEEPCOMPOUND)
|
|
|
947 /* Skip over the previously found word(s). */
|
|
|
948 wlen += mip->mi_compoff;
|
|
236
|
949 }
|
|
|
950 else
|
|
|
951 {
|
|
300
|
952 /* Check for case-folded in case-folded tree. */
|
|
|
953 ptr = mip->mi_fword;
|
|
|
954 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
955 byts = slang->sl_fbyts;
|
|
|
956 idxs = slang->sl_fidxs;
|
|
339
|
957
|
|
|
958 if (mode == FIND_PREFIX)
|
|
|
959 {
|
|
|
960 /* Skip over the prefix. */
|
|
|
961 wlen = mip->mi_prefixlen;
|
|
|
962 flen -= mip->mi_prefixlen;
|
|
|
963 }
|
|
481
|
964 else if (mode == FIND_COMPOUND)
|
|
|
965 {
|
|
|
966 /* Skip over the previously found word(s). */
|
|
|
967 wlen = mip->mi_compoff;
|
|
|
968 flen -= mip->mi_compoff;
|
|
|
969 }
|
|
|
970
|
|
243
|
971 }
|
|
|
972
|
|
300
|
973 if (byts == NULL)
|
|
|
974 return; /* array is empty */
|
|
236
|
975
|
|
|
976 /*
|
|
307
|
977 * Repeat advancing in the tree until:
|
|
|
978 * - there is a byte that doesn't match,
|
|
|
979 * - we reach the end of the tree,
|
|
|
980 * - or we reach the end of the line.
|
|
236
|
981 */
|
|
300
|
982 for (;;)
|
|
236
|
983 {
|
|
346
|
984 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
985 flen = fold_more(mip);
|
|
300
|
986
|
|
|
987 len = byts[arridx++];
|
|
|
988
|
|
|
989 /* If the first possible byte is a zero the word could end here.
|
|
|
990 * Remember this index, we first check for the longest word. */
|
|
|
991 if (byts[arridx] == 0)
|
|
|
992 {
|
|
307
|
993 if (endidxcnt == MAXWLEN)
|
|
|
994 {
|
|
|
995 /* Must be a corrupted spell file. */
|
|
|
996 EMSG(_(e_format));
|
|
|
997 return;
|
|
|
998 }
|
|
300
|
999 endlen[endidxcnt] = wlen;
|
|
|
1000 endidx[endidxcnt++] = arridx++;
|
|
|
1001 --len;
|
|
|
1002
|
|
|
1003 /* Skip over the zeros, there can be several flag/region
|
|
|
1004 * combinations. */
|
|
|
1005 while (len > 0 && byts[arridx] == 0)
|
|
|
1006 {
|
|
|
1007 ++arridx;
|
|
|
1008 --len;
|
|
|
1009 }
|
|
|
1010 if (len == 0)
|
|
|
1011 break; /* no children, word must end here */
|
|
|
1012 }
|
|
|
1013
|
|
|
1014 /* Stop looking at end of the line. */
|
|
|
1015 if (ptr[wlen] == NUL)
|
|
|
1016 break;
|
|
|
1017
|
|
|
1018 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1019 c = ptr[wlen];
|
|
346
|
1020 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1021 c = ' ';
|
|
300
|
1022 lo = arridx;
|
|
|
1023 hi = arridx + len - 1;
|
|
|
1024 while (lo < hi)
|
|
|
1025 {
|
|
|
1026 m = (lo + hi) / 2;
|
|
|
1027 if (byts[m] > c)
|
|
|
1028 hi = m - 1;
|
|
|
1029 else if (byts[m] < c)
|
|
|
1030 lo = m + 1;
|
|
|
1031 else
|
|
|
1032 {
|
|
|
1033 lo = hi = m;
|
|
|
1034 break;
|
|
236
|
1035 }
|
|
|
1036 }
|
|
300
|
1037
|
|
|
1038 /* Stop if there is no matching byte. */
|
|
|
1039 if (hi < lo || byts[lo] != c)
|
|
|
1040 break;
|
|
|
1041
|
|
|
1042 /* Continue at the child (if there is one). */
|
|
|
1043 arridx = idxs[lo];
|
|
|
1044 ++wlen;
|
|
|
1045 --flen;
|
|
346
|
1046
|
|
|
1047 /* One space in the good word may stand for several spaces in the
|
|
|
1048 * checked word. */
|
|
|
1049 if (c == ' ')
|
|
|
1050 {
|
|
|
1051 for (;;)
|
|
|
1052 {
|
|
|
1053 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1054 flen = fold_more(mip);
|
|
|
1055 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1056 break;
|
|
|
1057 ++wlen;
|
|
|
1058 --flen;
|
|
|
1059 }
|
|
|
1060 }
|
|
236
|
1061 }
|
|
|
1062
|
|
300
|
1063 /*
|
|
|
1064 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1065 * first.
|
|
|
1066 */
|
|
|
1067 while (endidxcnt > 0)
|
|
|
1068 {
|
|
|
1069 --endidxcnt;
|
|
|
1070 arridx = endidx[endidxcnt];
|
|
|
1071 wlen = endlen[endidxcnt];
|
|
236
|
1072
|
|
300
|
1073 #ifdef FEAT_MBYTE
|
|
|
1074 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1075 continue; /* not at first byte of character */
|
|
|
1076 #endif
|
|
376
|
1077 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
481
|
1078 {
|
|
|
1079 if (slang->sl_compflags == NULL)
|
|
|
1080 continue; /* next char is a word character */
|
|
|
1081 word_ends = FALSE;
|
|
|
1082 }
|
|
|
1083 else
|
|
|
1084 word_ends = TRUE;
|
|
300
|
1085
|
|
|
1086 #ifdef FEAT_MBYTE
|
|
339
|
1087 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1088 {
|
|
|
1089 /* Compute byte length in original word, length may change
|
|
339
|
1090 * when folding case. This can be slow, take a shortcut when the
|
|
|
1091 * case-folded word is equal to the keep-case word. */
|
|
300
|
1092 p = mip->mi_word;
|
|
339
|
1093 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1094 {
|
|
|
1095 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1096 mb_ptr_adv(p);
|
|
|
1097 wlen = p - mip->mi_word;
|
|
|
1098 }
|
|
300
|
1099 }
|
|
|
1100 #endif
|
|
236
|
1101
|
|
339
|
1102 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1103 * prefix ID.
|
|
|
1104 * Repeat this if there are more flags/region alternatives until there
|
|
|
1105 * is a match. */
|
|
|
1106 res = SP_BAD;
|
|
|
1107 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1108 --len, ++arridx)
|
|
300
|
1109 {
|
|
|
1110 flags = idxs[arridx];
|
|
324
|
1111
|
|
339
|
1112 /* For the fold-case tree check that the case of the checked word
|
|
|
1113 * matches with what the word in the tree requires.
|
|
|
1114 * For keep-case tree the case is always right. For prefixes we
|
|
|
1115 * don't bother to check. */
|
|
|
1116 if (mode == FIND_FOLDWORD)
|
|
300
|
1117 {
|
|
|
1118 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1119 {
|
|
323
|
1120 /* mi_capflags was set for a different word length, need
|
|
|
1121 * to do it again. */
|
|
300
|
1122 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1123 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1124 }
|
|
|
1125
|
|
346
|
1126 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1127 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1128 continue;
|
|
300
|
1129 }
|
|
236
|
1130
|
|
339
|
1131 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1132 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1133 * mip->mi_prefarridx that find_prefix() filled. */
|
|
481
|
1134 else if (mode == FIND_PREFIX)
|
|
|
1135 {
|
|
366
|
1136 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1137 flags,
|
|
455
|
1138 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1139 FALSE);
|
|
366
|
1140 if (c == 0)
|
|
339
|
1141 continue;
|
|
366
|
1142
|
|
|
1143 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1144 if (c & WF_RAREPFX)
|
|
|
1145 flags |= WF_RARE;
|
|
339
|
1146 }
|
|
|
1147
|
|
481
|
1148 if (mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND
|
|
|
1149 || !word_ends)
|
|
|
1150 {
|
|
|
1151 /* Makes you wonder why someone puts a compound flag on a word
|
|
|
1152 * that's too short... Myspell compatibility requires this
|
|
|
1153 * anyway. */
|
|
|
1154 if (wlen < slang->sl_compminlen)
|
|
|
1155 continue;
|
|
|
1156
|
|
|
1157 /* The word doesn't end or it comes after another: it must
|
|
|
1158 * have a compound flag. */
|
|
|
1159 /* TODO: check more flags */
|
|
|
1160 if (*slang->sl_compflags != ((unsigned)flags >> 24))
|
|
|
1161 continue;
|
|
|
1162 }
|
|
|
1163
|
|
|
1164 if (!word_ends)
|
|
|
1165 {
|
|
|
1166 /* Check that a valid word follows. If there is one, it will
|
|
|
1167 * set "mi_result", thus we are always finished here.
|
|
|
1168 * Recursive! */
|
|
|
1169
|
|
|
1170 /* Find following word in case-folded tree. */
|
|
|
1171 mip->mi_compoff = endlen[endidxcnt];
|
|
|
1172 #ifdef FEAT_MBYTE
|
|
|
1173 if (has_mbyte && mode == FIND_KEEPWORD)
|
|
|
1174 {
|
|
|
1175 /* Compute byte length in case-folded word from "wlen":
|
|
|
1176 * byte length in keep-case word. Length may change when
|
|
|
1177 * folding case. This can be slow, take a shortcut when
|
|
|
1178 * the case-folded word is equal to the keep-case word. */
|
|
|
1179 p = mip->mi_fword;
|
|
|
1180 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1181 {
|
|
|
1182 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1183 mb_ptr_adv(p);
|
|
|
1184 mip->mi_compoff = p - mip->mi_fword;
|
|
|
1185 }
|
|
|
1186 }
|
|
|
1187 #endif
|
|
|
1188 find_word(mip, FIND_COMPOUND);
|
|
|
1189 if (mip->mi_result == SP_OK)
|
|
|
1190 break;
|
|
|
1191
|
|
|
1192 /* Find following word in keep-case tree. */
|
|
|
1193 mip->mi_compoff = wlen;
|
|
|
1194 find_word(mip, FIND_KEEPCOMPOUND);
|
|
|
1195 if (mip->mi_result == SP_OK)
|
|
|
1196 break;
|
|
|
1197 continue;
|
|
|
1198 }
|
|
|
1199
|
|
339
|
1200 if (flags & WF_BANNED)
|
|
|
1201 res = SP_BANNED;
|
|
|
1202 else if (flags & WF_REGION)
|
|
|
1203 {
|
|
|
1204 /* Check region. */
|
|
390
|
1205 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1206 res = SP_OK;
|
|
339
|
1207 else
|
|
|
1208 res = SP_LOCAL;
|
|
300
|
1209 }
|
|
339
|
1210 else if (flags & WF_RARE)
|
|
|
1211 res = SP_RARE;
|
|
307
|
1212 else
|
|
339
|
1213 res = SP_OK;
|
|
|
1214
|
|
|
1215 /* Always use the longest match and the best result. */
|
|
|
1216 if (mip->mi_result > res)
|
|
|
1217 {
|
|
|
1218 mip->mi_result = res;
|
|
|
1219 mip->mi_end = mip->mi_word + wlen;
|
|
|
1220 }
|
|
351
|
1221 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1222 mip->mi_end = mip->mi_word + wlen;
|
|
|
1223
|
|
|
1224 if (res == SP_OK)
|
|
|
1225 break;
|
|
300
|
1226 }
|
|
|
1227
|
|
307
|
1228 if (res == SP_OK)
|
|
300
|
1229 break;
|
|
|
1230 }
|
|
236
|
1231 }
|
|
|
1232
|
|
323
|
1233 /*
|
|
390
|
1234 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1235 * ID in "flags" for the word "word".
|
|
366
|
1236 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1237 */
|
|
|
1238 static int
|
|
455
|
1239 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1240 int totprefcnt; /* nr of prefix IDs */
|
|
|
1241 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1242 int flags;
|
|
351
|
1243 char_u *word;
|
|
|
1244 slang_T *slang;
|
|
455
|
1245 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1246 {
|
|
|
1247 int prefcnt;
|
|
|
1248 int pidx;
|
|
|
1249 regprog_T *rp;
|
|
|
1250 regmatch_T regmatch;
|
|
390
|
1251 int prefid;
|
|
|
1252
|
|
|
1253 prefid = (unsigned)flags >> 24;
|
|
351
|
1254 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1255 {
|
|
|
1256 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1257
|
|
|
1258 /* Check the prefix ID. */
|
|
|
1259 if (prefid != (pidx & 0xff))
|
|
|
1260 continue;
|
|
|
1261
|
|
390
|
1262 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1263 * suffix. */
|
|
|
1264 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1265 continue;
|
|
|
1266
|
|
351
|
1267 /* Check the condition, if there is one. The condition index is
|
|
366
|
1268 * stored in the two bytes above the prefix ID byte. */
|
|
|
1269 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1270 if (rp != NULL)
|
|
|
1271 {
|
|
|
1272 regmatch.regprog = rp;
|
|
|
1273 regmatch.rm_ic = FALSE;
|
|
|
1274 if (!vim_regexec(®match, word, 0))
|
|
|
1275 continue;
|
|
|
1276 }
|
|
455
|
1277 else if (cond_req)
|
|
|
1278 continue;
|
|
|
1279
|
|
|
1280 /* It's a match! Return the WF_ flags. */
|
|
366
|
1281 return pidx;
|
|
|
1282 }
|
|
|
1283 return 0;
|
|
351
|
1284 }
|
|
|
1285
|
|
|
1286 /*
|
|
339
|
1287 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1288 * If it does, then check the following word.
|
|
|
1289 *
|
|
|
1290 * For a match mip->mi_result is updated.
|
|
|
1291 */
|
|
|
1292 static void
|
|
|
1293 find_prefix(mip)
|
|
|
1294 matchinf_T *mip;
|
|
|
1295 {
|
|
|
1296 idx_T arridx = 0;
|
|
|
1297 int len;
|
|
|
1298 int wlen = 0;
|
|
|
1299 int flen;
|
|
|
1300 int c;
|
|
|
1301 char_u *ptr;
|
|
|
1302 idx_T lo, hi, m;
|
|
|
1303 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1304 char_u *byts;
|
|
|
1305 idx_T *idxs;
|
|
|
1306
|
|
375
|
1307 byts = slang->sl_pbyts;
|
|
|
1308 if (byts == NULL)
|
|
|
1309 return; /* array is empty */
|
|
|
1310
|
|
339
|
1311 /* We use the case-folded word here, since prefixes are always
|
|
|
1312 * case-folded. */
|
|
|
1313 ptr = mip->mi_fword;
|
|
|
1314 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1315 idxs = slang->sl_pidxs;
|
|
|
1316
|
|
|
1317 /*
|
|
|
1318 * Repeat advancing in the tree until:
|
|
|
1319 * - there is a byte that doesn't match,
|
|
|
1320 * - we reach the end of the tree,
|
|
|
1321 * - or we reach the end of the line.
|
|
|
1322 */
|
|
|
1323 for (;;)
|
|
|
1324 {
|
|
|
1325 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1326 flen = fold_more(mip);
|
|
|
1327
|
|
|
1328 len = byts[arridx++];
|
|
|
1329
|
|
|
1330 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1331 * Check if the following word matches and supports the prefix. */
|
|
|
1332 if (byts[arridx] == 0)
|
|
|
1333 {
|
|
|
1334 /* There can be several prefixes with different conditions. We
|
|
|
1335 * try them all, since we don't know which one will give the
|
|
|
1336 * longest match. The word is the same each time, pass the list
|
|
|
1337 * of possible prefixes to find_word(). */
|
|
|
1338 mip->mi_prefarridx = arridx;
|
|
|
1339 mip->mi_prefcnt = len;
|
|
|
1340 while (len > 0 && byts[arridx] == 0)
|
|
|
1341 {
|
|
|
1342 ++arridx;
|
|
|
1343 --len;
|
|
|
1344 }
|
|
|
1345 mip->mi_prefcnt -= len;
|
|
|
1346
|
|
|
1347 /* Find the word that comes after the prefix. */
|
|
|
1348 mip->mi_prefixlen = wlen;
|
|
455
|
1349 #ifdef FEAT_MBYTE
|
|
|
1350 if (has_mbyte)
|
|
|
1351 {
|
|
|
1352 /* Case-folded length may differ from original length. */
|
|
|
1353 mip->mi_cprefixlen = nofold_len(mip->mi_fword, wlen,
|
|
|
1354 mip->mi_word);
|
|
|
1355 }
|
|
|
1356 else
|
|
|
1357 mip->mi_cprefixlen = wlen;
|
|
|
1358 #endif
|
|
339
|
1359 find_word(mip, FIND_PREFIX);
|
|
|
1360
|
|
|
1361
|
|
|
1362 if (len == 0)
|
|
|
1363 break; /* no children, word must end here */
|
|
|
1364 }
|
|
|
1365
|
|
|
1366 /* Stop looking at end of the line. */
|
|
|
1367 if (ptr[wlen] == NUL)
|
|
|
1368 break;
|
|
|
1369
|
|
|
1370 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1371 c = ptr[wlen];
|
|
|
1372 lo = arridx;
|
|
|
1373 hi = arridx + len - 1;
|
|
|
1374 while (lo < hi)
|
|
|
1375 {
|
|
|
1376 m = (lo + hi) / 2;
|
|
|
1377 if (byts[m] > c)
|
|
|
1378 hi = m - 1;
|
|
|
1379 else if (byts[m] < c)
|
|
|
1380 lo = m + 1;
|
|
|
1381 else
|
|
|
1382 {
|
|
|
1383 lo = hi = m;
|
|
|
1384 break;
|
|
|
1385 }
|
|
|
1386 }
|
|
|
1387
|
|
|
1388 /* Stop if there is no matching byte. */
|
|
|
1389 if (hi < lo || byts[lo] != c)
|
|
|
1390 break;
|
|
|
1391
|
|
|
1392 /* Continue at the child (if there is one). */
|
|
|
1393 arridx = idxs[lo];
|
|
|
1394 ++wlen;
|
|
|
1395 --flen;
|
|
|
1396 }
|
|
|
1397 }
|
|
|
1398
|
|
|
1399 /*
|
|
|
1400 * Need to fold at least one more character. Do until next non-word character
|
|
|
1401 * for efficiency.
|
|
|
1402 * Return the length of the folded chars in bytes.
|
|
|
1403 */
|
|
|
1404 static int
|
|
|
1405 fold_more(mip)
|
|
|
1406 matchinf_T *mip;
|
|
|
1407 {
|
|
|
1408 int flen;
|
|
|
1409 char_u *p;
|
|
|
1410
|
|
|
1411 p = mip->mi_fend;
|
|
|
1412 do
|
|
|
1413 {
|
|
|
1414 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1415 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1416
|
|
|
1417 /* Include the non-word character so that we can check for the
|
|
|
1418 * word end. */
|
|
|
1419 if (*mip->mi_fend != NUL)
|
|
|
1420 mb_ptr_adv(mip->mi_fend);
|
|
|
1421
|
|
|
1422 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1423 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1424 MAXWLEN - mip->mi_fwordlen);
|
|
|
1425 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1426 mip->mi_fwordlen += flen;
|
|
|
1427 return flen;
|
|
|
1428 }
|
|
|
1429
|
|
|
1430 /*
|
|
323
|
1431 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1432 * case.
|
|
|
1433 */
|
|
|
1434 static int
|
|
389
|
1435 spell_valid_case(wordflags, treeflags)
|
|
|
1436 int wordflags; /* flags for the checked word. */
|
|
323
|
1437 int treeflags; /* flags for the word in the spell tree */
|
|
|
1438 {
|
|
389
|
1439 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1440 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1441 && ((treeflags & WF_ONECAP) == 0
|
|
|
1442 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1443 }
|
|
|
1444
|
|
351
|
1445 /*
|
|
|
1446 * Return TRUE if spell checking is not enabled.
|
|
|
1447 */
|
|
|
1448 static int
|
|
|
1449 no_spell_checking()
|
|
|
1450 {
|
|
|
1451 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1452 {
|
|
|
1453 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1454 return TRUE;
|
|
|
1455 }
|
|
|
1456 return FALSE;
|
|
|
1457 }
|
|
300
|
1458
|
|
236
|
1459 /*
|
|
|
1460 * Move to next spell error.
|
|
323
|
1461 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
236
|
1462 * Return OK if found, FAIL otherwise.
|
|
|
1463 */
|
|
|
1464 int
|
|
323
|
1465 spell_move_to(dir, allwords, curline)
|
|
236
|
1466 int dir; /* FORWARD or BACKWARD */
|
|
|
1467 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1468 int curline;
|
|
236
|
1469 {
|
|
249
|
1470 linenr_T lnum;
|
|
|
1471 pos_T found_pos;
|
|
236
|
1472 char_u *line;
|
|
|
1473 char_u *p;
|
|
346
|
1474 char_u *endp;
|
|
|
1475 int attr;
|
|
236
|
1476 int len;
|
|
249
|
1477 int has_syntax = syntax_present(curbuf);
|
|
|
1478 int col;
|
|
|
1479 int can_spell;
|
|
346
|
1480 char_u *buf = NULL;
|
|
|
1481 int buflen = 0;
|
|
|
1482 int skip = 0;
|
|
385
|
1483 int capcol = -1;
|
|
236
|
1484
|
|
351
|
1485 if (no_spell_checking())
|
|
236
|
1486 return FAIL;
|
|
|
1487
|
|
249
|
1488 /*
|
|
|
1489 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1490 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1491 *
|
|
|
1492 * When searching backwards, we continue in the line to find the last
|
|
|
1493 * bad word (in the cursor line: before the cursor).
|
|
346
|
1494 *
|
|
|
1495 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1496 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1497 * though...
|
|
249
|
1498 */
|
|
|
1499 lnum = curwin->w_cursor.lnum;
|
|
|
1500 found_pos.lnum = 0;
|
|
236
|
1501
|
|
|
1502 while (!got_int)
|
|
|
1503 {
|
|
249
|
1504 line = ml_get(lnum);
|
|
346
|
1505
|
|
|
1506 len = STRLEN(line);
|
|
|
1507 if (buflen < len + MAXWLEN + 2)
|
|
|
1508 {
|
|
|
1509 vim_free(buf);
|
|
|
1510 buflen = len + MAXWLEN + 2;
|
|
|
1511 buf = alloc(buflen);
|
|
|
1512 if (buf == NULL)
|
|
|
1513 break;
|
|
|
1514 }
|
|
|
1515
|
|
385
|
1516 /* In first line check first word for Capital. */
|
|
|
1517 if (lnum == 1)
|
|
|
1518 capcol = 0;
|
|
|
1519
|
|
|
1520 /* For checking first word with a capital skip white space. */
|
|
|
1521 if (capcol == 0)
|
|
|
1522 capcol = skipwhite(line) - line;
|
|
|
1523
|
|
346
|
1524 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1525 * possible. */
|
|
|
1526 STRCPY(buf, line);
|
|
|
1527 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1528 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1529
|
|
|
1530 p = buf + skip;
|
|
|
1531 endp = buf + len;
|
|
|
1532 while (p < endp)
|
|
236
|
1533 {
|
|
300
|
1534 /* When searching backward don't search after the cursor. */
|
|
|
1535 if (dir == BACKWARD
|
|
|
1536 && lnum == curwin->w_cursor.lnum
|
|
346
|
1537 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1538 break;
|
|
249
|
1539
|
|
300
|
1540 /* start of word */
|
|
346
|
1541 attr = 0;
|
|
385
|
1542 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1543
|
|
300
|
1544 if (attr != 0)
|
|
|
1545 {
|
|
|
1546 /* We found a bad word. Check the attribute. */
|
|
|
1547 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1548 {
|
|
300
|
1549 /* When searching forward only accept a bad word after
|
|
|
1550 * the cursor. */
|
|
|
1551 if (dir == BACKWARD
|
|
|
1552 || lnum > curwin->w_cursor.lnum
|
|
|
1553 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1554 && (colnr_T)(curline ? p - buf + len
|
|
|
1555 : p - buf)
|
|
300
|
1556 > curwin->w_cursor.col))
|
|
236
|
1557 {
|
|
300
|
1558 if (has_syntax)
|
|
249
|
1559 {
|
|
346
|
1560 col = p - buf;
|
|
300
|
1561 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1562 FALSE, &can_spell);
|
|
|
1563 }
|
|
|
1564 else
|
|
|
1565 can_spell = TRUE;
|
|
249
|
1566
|
|
300
|
1567 if (can_spell)
|
|
|
1568 {
|
|
|
1569 found_pos.lnum = lnum;
|
|
346
|
1570 found_pos.col = p - buf;
|
|
249
|
1571 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1572 found_pos.coladd = 0;
|
|
249
|
1573 #endif
|
|
300
|
1574 if (dir == FORWARD)
|
|
|
1575 {
|
|
|
1576 /* No need to search further. */
|
|
|
1577 curwin->w_cursor = found_pos;
|
|
346
|
1578 vim_free(buf);
|
|
300
|
1579 return OK;
|
|
249
|
1580 }
|
|
|
1581 }
|
|
236
|
1582 }
|
|
|
1583 }
|
|
|
1584 }
|
|
|
1585
|
|
300
|
1586 /* advance to character after the word */
|
|
|
1587 p += len;
|
|
385
|
1588 capcol -= len;
|
|
236
|
1589 }
|
|
|
1590
|
|
323
|
1591 if (curline)
|
|
346
|
1592 break; /* only check cursor line */
|
|
323
|
1593
|
|
236
|
1594 /* Advance to next line. */
|
|
249
|
1595 if (dir == BACKWARD)
|
|
|
1596 {
|
|
|
1597 if (found_pos.lnum != 0)
|
|
|
1598 {
|
|
|
1599 /* Use the last match in the line. */
|
|
|
1600 curwin->w_cursor = found_pos;
|
|
346
|
1601 vim_free(buf);
|
|
249
|
1602 return OK;
|
|
|
1603 }
|
|
|
1604 if (lnum == 1)
|
|
346
|
1605 break;
|
|
249
|
1606 --lnum;
|
|
385
|
1607 capcol = -1;
|
|
249
|
1608 }
|
|
|
1609 else
|
|
|
1610 {
|
|
|
1611 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1612 break;
|
|
249
|
1613 ++lnum;
|
|
346
|
1614
|
|
|
1615 /* Skip the characters at the start of the next line that were
|
|
|
1616 * included in a match crossing line boundaries. */
|
|
|
1617 if (attr == 0)
|
|
|
1618 skip = p - endp;
|
|
|
1619 else
|
|
|
1620 skip = 0;
|
|
385
|
1621
|
|
|
1622 /* Capscol skips over the inserted space. */
|
|
|
1623 --capcol;
|
|
|
1624
|
|
|
1625 /* But after empty line check first word in next line */
|
|
|
1626 if (*skipwhite(line) == NUL)
|
|
|
1627 capcol = 0;
|
|
249
|
1628 }
|
|
236
|
1629
|
|
|
1630 line_breakcheck();
|
|
|
1631 }
|
|
|
1632
|
|
346
|
1633 vim_free(buf);
|
|
|
1634 return FAIL;
|
|
|
1635 }
|
|
|
1636
|
|
|
1637 /*
|
|
|
1638 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1639 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1640 */
|
|
|
1641 void
|
|
|
1642 spell_cat_line(buf, line, maxlen)
|
|
|
1643 char_u *buf;
|
|
|
1644 char_u *line;
|
|
|
1645 int maxlen;
|
|
|
1646 {
|
|
|
1647 char_u *p;
|
|
|
1648 int n;
|
|
|
1649
|
|
|
1650 p = skipwhite(line);
|
|
|
1651 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1652 p = skipwhite(p + 1);
|
|
|
1653
|
|
|
1654 if (*p != NUL)
|
|
|
1655 {
|
|
|
1656 *buf = ' ';
|
|
376
|
1657 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1658 n = p - line;
|
|
|
1659 if (n >= maxlen)
|
|
|
1660 n = maxlen - 1;
|
|
|
1661 vim_memset(buf + 1, ' ', n);
|
|
|
1662 }
|
|
236
|
1663 }
|
|
|
1664
|
|
|
1665 /*
|
|
307
|
1666 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1667 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1668 */
|
|
307
|
1669 static void
|
|
236
|
1670 spell_load_lang(lang)
|
|
|
1671 char_u *lang;
|
|
|
1672 {
|
|
310
|
1673 char_u fname_enc[85];
|
|
236
|
1674 int r;
|
|
307
|
1675 char_u langcp[MAXWLEN + 1];
|
|
|
1676
|
|
310
|
1677 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1678 * It's truncated when an error is detected. */
|
|
|
1679 STRCPY(langcp, lang);
|
|
|
1680
|
|
310
|
1681 /*
|
|
|
1682 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1683 */
|
|
|
1684 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1685 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1686 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1687
|
|
|
1688 if (r == FAIL && *langcp != NUL)
|
|
|
1689 {
|
|
|
1690 /* Try loading the ASCII version. */
|
|
310
|
1691 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1692 "spell/%s.ascii.spl", lang);
|
|
310
|
1693 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1694 }
|
|
|
1695
|
|
|
1696 if (r == FAIL)
|
|
|
1697 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
|
236
|
1698 fname_enc + 6);
|
|
310
|
1699 else if (*langcp != NUL)
|
|
|
1700 {
|
|
|
1701 /* Load all the additions. */
|
|
|
1702 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1703 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1704 }
|
|
|
1705 }
|
|
|
1706
|
|
|
1707 /*
|
|
|
1708 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1709 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1710 */
|
|
|
1711 static char_u *
|
|
|
1712 spell_enc()
|
|
|
1713 {
|
|
|
1714
|
|
|
1715 #ifdef FEAT_MBYTE
|
|
|
1716 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1717 return p_enc;
|
|
|
1718 #endif
|
|
|
1719 return (char_u *)"latin1";
|
|
236
|
1720 }
|
|
|
1721
|
|
|
1722 /*
|
|
385
|
1723 * Get the name of the .spl file for the internal wordlist into
|
|
|
1724 * "fname[MAXPATHL]".
|
|
|
1725 */
|
|
|
1726 static void
|
|
|
1727 int_wordlist_spl(fname)
|
|
|
1728 char_u *fname;
|
|
|
1729 {
|
|
|
1730 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
1731 int_wordlist, spell_enc());
|
|
|
1732 }
|
|
|
1733
|
|
|
1734 /*
|
|
236
|
1735 * Allocate a new slang_T.
|
|
|
1736 * Caller must fill "sl_next".
|
|
|
1737 */
|
|
|
1738 static slang_T *
|
|
|
1739 slang_alloc(lang)
|
|
|
1740 char_u *lang;
|
|
|
1741 {
|
|
|
1742 slang_T *lp;
|
|
|
1743
|
|
300
|
1744 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1745 if (lp != NULL)
|
|
|
1746 {
|
|
|
1747 lp->sl_name = vim_strsave(lang);
|
|
323
|
1748 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
236
|
1749 }
|
|
|
1750 return lp;
|
|
|
1751 }
|
|
|
1752
|
|
|
1753 /*
|
|
|
1754 * Free the contents of an slang_T and the structure itself.
|
|
|
1755 */
|
|
|
1756 static void
|
|
|
1757 slang_free(lp)
|
|
|
1758 slang_T *lp;
|
|
|
1759 {
|
|
|
1760 vim_free(lp->sl_name);
|
|
310
|
1761 vim_free(lp->sl_fname);
|
|
|
1762 slang_clear(lp);
|
|
|
1763 vim_free(lp);
|
|
|
1764 }
|
|
|
1765
|
|
|
1766 /*
|
|
|
1767 * Clear an slang_T so that the file can be reloaded.
|
|
|
1768 */
|
|
|
1769 static void
|
|
|
1770 slang_clear(lp)
|
|
|
1771 slang_T *lp;
|
|
|
1772 {
|
|
339
|
1773 garray_T *gap;
|
|
|
1774 fromto_T *ftp;
|
|
344
|
1775 salitem_T *smp;
|
|
339
|
1776 int i;
|
|
323
|
1777
|
|
300
|
1778 vim_free(lp->sl_fbyts);
|
|
310
|
1779 lp->sl_fbyts = NULL;
|
|
300
|
1780 vim_free(lp->sl_kbyts);
|
|
310
|
1781 lp->sl_kbyts = NULL;
|
|
339
|
1782 vim_free(lp->sl_pbyts);
|
|
|
1783 lp->sl_pbyts = NULL;
|
|
|
1784
|
|
300
|
1785 vim_free(lp->sl_fidxs);
|
|
310
|
1786 lp->sl_fidxs = NULL;
|
|
300
|
1787 vim_free(lp->sl_kidxs);
|
|
310
|
1788 lp->sl_kidxs = NULL;
|
|
339
|
1789 vim_free(lp->sl_pidxs);
|
|
|
1790 lp->sl_pidxs = NULL;
|
|
323
|
1791
|
|
344
|
1792 gap = &lp->sl_rep;
|
|
|
1793 while (gap->ga_len > 0)
|
|
323
|
1794 {
|
|
344
|
1795 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1796 vim_free(ftp->ft_from);
|
|
|
1797 vim_free(ftp->ft_to);
|
|
323
|
1798 }
|
|
344
|
1799 ga_clear(gap);
|
|
|
1800
|
|
|
1801 gap = &lp->sl_sal;
|
|
375
|
1802 if (lp->sl_sofo)
|
|
376
|
1803 {
|
|
|
1804 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1805 if (gap->ga_data != NULL)
|
|
|
1806 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1807 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1808 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1809 }
|
|
375
|
1810 else
|
|
|
1811 /* SAL items: free salitem_T items */
|
|
|
1812 while (gap->ga_len > 0)
|
|
|
1813 {
|
|
|
1814 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1815 vim_free(smp->sm_lead);
|
|
|
1816 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1817 vim_free(smp->sm_to);
|
|
|
1818 #ifdef FEAT_MBYTE
|
|
|
1819 vim_free(smp->sm_lead_w);
|
|
|
1820 vim_free(smp->sm_oneof_w);
|
|
|
1821 vim_free(smp->sm_to_w);
|
|
|
1822 #endif
|
|
|
1823 }
|
|
344
|
1824 ga_clear(gap);
|
|
323
|
1825
|
|
339
|
1826 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1827 vim_free(lp->sl_prefprog[i]);
|
|
376
|
1828 lp->sl_prefixcnt = 0;
|
|
339
|
1829 vim_free(lp->sl_prefprog);
|
|
376
|
1830 lp->sl_prefprog = NULL;
|
|
|
1831
|
|
|
1832 vim_free(lp->sl_midword);
|
|
|
1833 lp->sl_midword = NULL;
|
|
339
|
1834
|
|
481
|
1835 vim_free(lp->sl_compflags);
|
|
|
1836 lp->sl_compflags = NULL;
|
|
|
1837
|
|
330
|
1838 #ifdef FEAT_MBYTE
|
|
|
1839 {
|
|
|
1840 int todo = lp->sl_map_hash.ht_used;
|
|
|
1841 hashitem_T *hi;
|
|
|
1842
|
|
|
1843 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1844 if (!HASHITEM_EMPTY(hi))
|
|
|
1845 {
|
|
|
1846 --todo;
|
|
|
1847 vim_free(hi->hi_key);
|
|
|
1848 }
|
|
|
1849 }
|
|
|
1850 hash_clear(&lp->sl_map_hash);
|
|
|
1851 #endif
|
|
236
|
1852 }
|
|
|
1853
|
|
|
1854 /*
|
|
307
|
1855 * Load one spell file and store the info into a slang_T.
|
|
236
|
1856 * Invoked through do_in_runtimepath().
|
|
|
1857 */
|
|
|
1858 static void
|
|
310
|
1859 spell_load_cb(fname, cookie)
|
|
236
|
1860 char_u *fname;
|
|
307
|
1861 void *cookie; /* points to the language name */
|
|
236
|
1862 {
|
|
323
|
1863 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1864 }
|
|
|
1865
|
|
|
1866 /*
|
|
|
1867 * Load one spell file and store the info into a slang_T.
|
|
|
1868 *
|
|
|
1869 * This is invoked in two ways:
|
|
|
1870 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
1871 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
1872 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
1873 * points to the existing slang_T.
|
|
323
|
1874 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
1875 */
|
|
323
|
1876 static slang_T *
|
|
|
1877 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
1878 char_u *fname;
|
|
|
1879 char_u *lang;
|
|
|
1880 slang_T *old_lp;
|
|
323
|
1881 int silent; /* no error if file doesn't exist */
|
|
310
|
1882 {
|
|
236
|
1883 FILE *fd;
|
|
|
1884 char_u buf[MAXWLEN + 1];
|
|
|
1885 char_u *p;
|
|
339
|
1886 char_u *bp;
|
|
|
1887 idx_T *ip;
|
|
236
|
1888 int i;
|
|
339
|
1889 int n;
|
|
300
|
1890 int len;
|
|
236
|
1891 int round;
|
|
|
1892 char_u *save_sourcing_name = sourcing_name;
|
|
|
1893 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
255
|
1894 int cnt, ccnt;
|
|
|
1895 char_u *fol;
|
|
307
|
1896 slang_T *lp = NULL;
|
|
323
|
1897 garray_T *gap;
|
|
|
1898 fromto_T *ftp;
|
|
344
|
1899 salitem_T *smp;
|
|
323
|
1900 short *first;
|
|
324
|
1901 idx_T idx;
|
|
344
|
1902 int c = 0;
|
|
236
|
1903
|
|
310
|
1904 fd = mch_fopen((char *)fname, "r");
|
|
236
|
1905 if (fd == NULL)
|
|
|
1906 {
|
|
323
|
1907 if (!silent)
|
|
|
1908 EMSG2(_(e_notopen), fname);
|
|
|
1909 else if (p_verbose > 2)
|
|
|
1910 {
|
|
|
1911 verbose_enter();
|
|
|
1912 smsg((char_u *)e_notopen, fname);
|
|
|
1913 verbose_leave();
|
|
|
1914 }
|
|
255
|
1915 goto endFAIL;
|
|
236
|
1916 }
|
|
310
|
1917 if (p_verbose > 2)
|
|
|
1918 {
|
|
|
1919 verbose_enter();
|
|
|
1920 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
1921 verbose_leave();
|
|
|
1922 }
|
|
|
1923
|
|
|
1924 if (old_lp == NULL)
|
|
|
1925 {
|
|
|
1926 lp = slang_alloc(lang);
|
|
|
1927 if (lp == NULL)
|
|
|
1928 goto endFAIL;
|
|
|
1929
|
|
|
1930 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
1931 lp->sl_fname = vim_strsave(fname);
|
|
|
1932 if (lp->sl_fname == NULL)
|
|
|
1933 goto endFAIL;
|
|
|
1934
|
|
|
1935 /* Check for .add.spl. */
|
|
|
1936 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
1937 }
|
|
|
1938 else
|
|
|
1939 lp = old_lp;
|
|
307
|
1940
|
|
236
|
1941 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
1942 sourcing_name = fname;
|
|
|
1943 sourcing_lnum = 0;
|
|
|
1944
|
|
339
|
1945 /* <HEADER>: <fileID>
|
|
|
1946 * <regioncnt> <regionname> ...
|
|
|
1947 * <charflagslen> <charflags>
|
|
|
1948 * <fcharslen> <fchars>
|
|
366
|
1949 * <midwordlen> <midword>
|
|
481
|
1950 * <compoundlen> <compoundtype> <compoundinfo>
|
|
339
|
1951 * <prefcondcnt> <prefcond> ...
|
|
|
1952 */
|
|
236
|
1953 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
1954 buf[i] = getc(fd); /* <fileID> */
|
|
|
1955 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
1956 {
|
|
|
1957 EMSG(_("E757: Wrong file ID in spell file"));
|
|
255
|
1958 goto endFAIL;
|
|
236
|
1959 }
|
|
|
1960
|
|
|
1961 cnt = getc(fd); /* <regioncnt> */
|
|
255
|
1962 if (cnt < 0)
|
|
236
|
1963 {
|
|
|
1964 truncerr:
|
|
381
|
1965 EMSG(_(e_spell_trunc));
|
|
255
|
1966 goto endFAIL;
|
|
236
|
1967 }
|
|
|
1968 if (cnt > 8)
|
|
|
1969 {
|
|
|
1970 formerr:
|
|
307
|
1971 EMSG(_(e_format));
|
|
255
|
1972 goto endFAIL;
|
|
236
|
1973 }
|
|
|
1974 for (i = 0; i < cnt; ++i)
|
|
|
1975 {
|
|
|
1976 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
|
1977 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
|
1978 }
|
|
|
1979 lp->sl_regions[cnt * 2] = NUL;
|
|
|
1980
|
|
381
|
1981 /* <charflagslen> <charflags> */
|
|
|
1982 p = read_cnt_string(fd, 1, &cnt);
|
|
389
|
1983 if (cnt < 0)
|
|
381
|
1984 goto endFAIL;
|
|
|
1985
|
|
|
1986 /* <fcharslen> <fchars> */
|
|
389
|
1987 fol = read_cnt_string(fd, 2, &ccnt);
|
|
|
1988 if (ccnt < 0)
|
|
381
|
1989 {
|
|
300
|
1990 vim_free(p);
|
|
381
|
1991 goto endFAIL;
|
|
|
1992 }
|
|
|
1993
|
|
|
1994 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
1995 if (p != NULL && fol != NULL)
|
|
389
|
1996 i = set_spell_charflags(p, cnt, fol);
|
|
381
|
1997
|
|
|
1998 vim_free(p);
|
|
|
1999 vim_free(fol);
|
|
|
2000
|
|
|
2001 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2002 if ((p == NULL) != (fol == NULL))
|
|
|
2003 goto formerr;
|
|
255
|
2004
|
|
366
|
2005 /* <midwordlen> <midword> */
|
|
376
|
2006 lp->sl_midword = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2007 if (cnt < 0)
|
|
376
|
2008 goto endFAIL;
|
|
366
|
2009
|
|
481
|
2010 /* <compoundlen> <compoundtype> <compoundinfo> */
|
|
|
2011 cnt = (getc(fd) << 8) + getc(fd); /* <compoundlen> */
|
|
|
2012 if (cnt < 0)
|
|
|
2013 goto endFAIL;
|
|
|
2014 if (cnt > 0)
|
|
|
2015 {
|
|
|
2016 --cnt;
|
|
|
2017 c = getc(fd); /* <compoundtype> */
|
|
|
2018 if (c != 1)
|
|
|
2019 {
|
|
|
2020 /* Unknown kind of compound words, skip the info. */
|
|
|
2021 while (cnt-- > 0)
|
|
|
2022 getc(fd);
|
|
|
2023 }
|
|
|
2024 else if (cnt < 2)
|
|
|
2025 goto formerr;
|
|
|
2026 else
|
|
|
2027 {
|
|
|
2028 --cnt;
|
|
|
2029 c = getc(fd); /* <comp1minlen> */
|
|
|
2030 if (c < 1 || c > 50)
|
|
|
2031 c = 3;
|
|
|
2032 lp->sl_compminlen = c;
|
|
|
2033
|
|
|
2034 p = alloc(cnt + 1);
|
|
|
2035 if (p == NULL)
|
|
|
2036 goto endFAIL;
|
|
|
2037 lp->sl_compflags = p;
|
|
|
2038 while (cnt-- > 0)
|
|
|
2039 *p++ = getc(fd); /* <comp1flags> */
|
|
|
2040 *p = NUL;
|
|
|
2041 }
|
|
|
2042 }
|
|
|
2043
|
|
|
2044
|
|
339
|
2045 /* <prefcondcnt> <prefcond> ... */
|
|
|
2046 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
2047 if (cnt > 0)
|
|
|
2048 {
|
|
|
2049 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2050 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2051 if (lp->sl_prefprog == NULL)
|
|
|
2052 goto endFAIL;
|
|
|
2053 lp->sl_prefixcnt = cnt;
|
|
|
2054
|
|
|
2055 for (i = 0; i < cnt; ++i)
|
|
|
2056 {
|
|
|
2057 /* <prefcond> : <condlen> <condstr> */
|
|
|
2058 n = getc(fd); /* <condlen> */
|
|
481
|
2059 if (n < 0 || n >= MAXWLEN)
|
|
339
|
2060 goto formerr;
|
|
|
2061 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2062 * compile the regexp program used to check for the condition. */
|
|
|
2063 if (n > 0)
|
|
|
2064 {
|
|
344
|
2065 buf[0] = '^'; /* always match at one position only */
|
|
|
2066 p = buf + 1;
|
|
339
|
2067 while (n-- > 0)
|
|
|
2068 *p++ = getc(fd); /* <condstr> */
|
|
|
2069 *p = NUL;
|
|
|
2070 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
2071 }
|
|
|
2072 }
|
|
|
2073 }
|
|
|
2074
|
|
|
2075
|
|
323
|
2076 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
2077 * <salflags> <salcount> <sal> ...
|
|
|
2078 * <maplen> <mapstr> */
|
|
339
|
2079
|
|
344
|
2080 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
2081 if (cnt < 0)
|
|
|
2082 goto formerr;
|
|
|
2083
|
|
|
2084 gap = &lp->sl_rep;
|
|
|
2085 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2086 goto endFAIL;
|
|
|
2087
|
|
|
2088 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
2089 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2090 {
|
|
|
2091 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
381
|
2092 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
389
|
2093 if (i <= 0)
|
|
381
|
2094 goto endFAIL;
|
|
|
2095 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
389
|
2096 if (i <= 0)
|
|
381
|
2097 {
|
|
|
2098 vim_free(ftp->ft_from);
|
|
|
2099 goto endFAIL;
|
|
344
|
2100 }
|
|
|
2101 }
|
|
|
2102
|
|
|
2103 /* Fill the first-index table. */
|
|
|
2104 first = lp->sl_rep_first;
|
|
|
2105 for (i = 0; i < 256; ++i)
|
|
|
2106 first[i] = -1;
|
|
|
2107 for (i = 0; i < gap->ga_len; ++i)
|
|
323
|
2108 {
|
|
344
|
2109 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
2110 if (first[*ftp->ft_from] == -1)
|
|
|
2111 first[*ftp->ft_from] = i;
|
|
|
2112 }
|
|
|
2113
|
|
|
2114 i = getc(fd); /* <salflags> */
|
|
|
2115 if (i & SAL_F0LLOWUP)
|
|
|
2116 lp->sl_followup = TRUE;
|
|
|
2117 if (i & SAL_COLLAPSE)
|
|
|
2118 lp->sl_collapse = TRUE;
|
|
|
2119 if (i & SAL_REM_ACCENTS)
|
|
|
2120 lp->sl_rem_accents = TRUE;
|
|
375
|
2121 if (i & SAL_SOFO)
|
|
|
2122 lp->sl_sofo = TRUE;
|
|
389
|
2123 else
|
|
|
2124 lp->sl_sofo = FALSE;
|
|
344
|
2125
|
|
|
2126 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2127 if (cnt < 0)
|
|
|
2128 goto formerr;
|
|
|
2129
|
|
375
|
2130 if (lp->sl_sofo)
|
|
|
2131 {
|
|
|
2132 /*
|
|
|
2133 * SOFOFROM and SOFOTO items come in one <salfrom> and <salto>
|
|
|
2134 */
|
|
|
2135 if (cnt != 1)
|
|
|
2136 goto formerr;
|
|
|
2137
|
|
381
|
2138 /* <salfromlen> <salfrom> */
|
|
|
2139 bp = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2140 if (cnt < 0)
|
|
375
|
2141 goto endFAIL;
|
|
381
|
2142
|
|
|
2143 /* <saltolen> <salto> */
|
|
|
2144 fol = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2145 if (cnt < 0)
|
|
375
|
2146 {
|
|
|
2147 vim_free(bp);
|
|
344
|
2148 goto endFAIL;
|
|
375
|
2149 }
|
|
381
|
2150
|
|
|
2151 /* Store the info in lp->sl_sal and/or lp->sl_sal_first. */
|
|
389
|
2152 if (bp != NULL && fol != NULL)
|
|
|
2153 i = set_sofo(lp, bp, fol);
|
|
|
2154 else if (bp != NULL || fol != NULL)
|
|
|
2155 i = FAIL; /* only one of two strings is an error */
|
|
|
2156 else
|
|
|
2157 i = OK;
|
|
381
|
2158
|
|
375
|
2159 vim_free(bp);
|
|
|
2160 vim_free(fol);
|
|
381
|
2161 if (i == FAIL)
|
|
|
2162 goto formerr;
|
|
375
|
2163 }
|
|
|
2164 else
|
|
|
2165 {
|
|
|
2166 /*
|
|
|
2167 * SAL items
|
|
|
2168 */
|
|
|
2169 gap = &lp->sl_sal;
|
|
390
|
2170 ga_init2(gap, sizeof(salitem_T), 10);
|
|
375
|
2171 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2172 goto endFAIL;
|
|
|
2173
|
|
|
2174 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2175 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2176 {
|
|
|
2177 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2178 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2179 if (ccnt < 0)
|
|
|
2180 goto formerr;
|
|
|
2181 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2182 goto endFAIL;
|
|
|
2183 smp->sm_lead = p;
|
|
|
2184
|
|
|
2185 /* Read up to the first special char into sm_lead. */
|
|
|
2186 for (i = 0; i < ccnt; ++i)
|
|
|
2187 {
|
|
|
2188 c = getc(fd); /* <salfrom> */
|
|
|
2189 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2190 break;
|
|
|
2191 *p++ = c;
|
|
|
2192 }
|
|
|
2193 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2194 *p++ = NUL;
|
|
|
2195
|
|
|
2196 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2197 if (c == '(')
|
|
|
2198 {
|
|
|
2199 smp->sm_oneof = p;
|
|
|
2200 for (++i; i < ccnt; ++i)
|
|
|
2201 {
|
|
|
2202 c = getc(fd); /* <salfrom> */
|
|
|
2203 if (c == ')')
|
|
|
2204 break;
|
|
|
2205 *p++ = c;
|
|
|
2206 }
|
|
|
2207 *p++ = NUL;
|
|
|
2208 if (++i < ccnt)
|
|
|
2209 c = getc(fd);
|
|
|
2210 }
|
|
372
|
2211 else
|
|
375
|
2212 smp->sm_oneof = NULL;
|
|
|
2213
|
|
|
2214 /* Any following chars go in sm_rules. */
|
|
|
2215 smp->sm_rules = p;
|
|
|
2216 if (i < ccnt)
|
|
|
2217 /* store the char we got while checking for end of sm_lead */
|
|
|
2218 *p++ = c;
|
|
|
2219 for (++i; i < ccnt; ++i)
|
|
|
2220 *p++ = getc(fd); /* <salfrom> */
|
|
|
2221 *p++ = NUL;
|
|
|
2222
|
|
381
|
2223 /* <saltolen> <salto> */
|
|
|
2224 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
389
|
2225 if (ccnt < 0)
|
|
372
|
2226 {
|
|
|
2227 vim_free(smp->sm_lead);
|
|
375
|
2228 goto formerr;
|
|
|
2229 }
|
|
|
2230
|
|
372
|
2231 #ifdef FEAT_MBYTE
|
|
|
2232 if (has_mbyte)
|
|
|
2233 {
|
|
375
|
2234 /* convert the multi-byte strings to wide char strings */
|
|
|
2235 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2236 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2237 if (smp->sm_oneof == NULL)
|
|
|
2238 smp->sm_oneof_w = NULL;
|
|
|
2239 else
|
|
|
2240 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
389
|
2241 if (smp->sm_to == NULL)
|
|
|
2242 smp->sm_to_w = NULL;
|
|
|
2243 else
|
|
|
2244 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
375
|
2245 if (smp->sm_lead_w == NULL
|
|
|
2246 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
389
|
2247 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
372
|
2248 {
|
|
375
|
2249 vim_free(smp->sm_lead);
|
|
|
2250 vim_free(smp->sm_to);
|
|
|
2251 vim_free(smp->sm_lead_w);
|
|
|
2252 vim_free(smp->sm_oneof_w);
|
|
|
2253 vim_free(smp->sm_to_w);
|
|
|
2254 goto endFAIL;
|
|
372
|
2255 }
|
|
|
2256 }
|
|
|
2257 #endif
|
|
|
2258 }
|
|
375
|
2259
|
|
|
2260 /* Fill the first-index table. */
|
|
381
|
2261 set_sal_first(lp);
|
|
|
2262 }
|
|
|
2263
|
|
|
2264 /* <maplen> <mapstr> */
|
|
|
2265 p = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2266 if (cnt < 0)
|
|
323
|
2267 goto endFAIL;
|
|
389
|
2268 if (p != NULL)
|
|
|
2269 {
|
|
|
2270 set_map_str(lp, p);
|
|
|
2271 vim_free(p);
|
|
|
2272 }
|
|
236
|
2273
|
|
300
|
2274 /* round 1: <LWORDTREE>
|
|
339
|
2275 * round 2: <KWORDTREE>
|
|
|
2276 * round 3: <PREFIXTREE> */
|
|
|
2277 for (round = 1; round <= 3; ++round)
|
|
236
|
2278 {
|
|
300
|
2279 /* The tree size was computed when writing the file, so that we can
|
|
|
2280 * allocate it as one long block. <nodecount> */
|
|
|
2281 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2282 if (len < 0)
|
|
|
2283 goto truncerr;
|
|
|
2284 if (len > 0)
|
|
236
|
2285 {
|
|
300
|
2286 /* Allocate the byte array. */
|
|
339
|
2287 bp = lalloc((long_u)len, TRUE);
|
|
|
2288 if (bp == NULL)
|
|
300
|
2289 goto endFAIL;
|
|
|
2290 if (round == 1)
|
|
339
|
2291 lp->sl_fbyts = bp;
|
|
|
2292 else if (round == 2)
|
|
|
2293 lp->sl_kbyts = bp;
|
|
300
|
2294 else
|
|
339
|
2295 lp->sl_pbyts = bp;
|
|
236
|
2296
|
|
300
|
2297 /* Allocate the index array. */
|
|
339
|
2298 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2299 if (ip == NULL)
|
|
300
|
2300 goto endFAIL;
|
|
|
2301 if (round == 1)
|
|
339
|
2302 lp->sl_fidxs = ip;
|
|
|
2303 else if (round == 2)
|
|
|
2304 lp->sl_kidxs = ip;
|
|
300
|
2305 else
|
|
339
|
2306 lp->sl_pidxs = ip;
|
|
300
|
2307
|
|
|
2308 /* Read the tree and store it in the array. */
|
|
339
|
2309 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2310 if (idx == -1)
|
|
300
|
2311 goto truncerr;
|
|
324
|
2312 if (idx < 0)
|
|
236
|
2313 goto formerr;
|
|
|
2314 }
|
|
300
|
2315 }
|
|
243
|
2316
|
|
310
|
2317 /* For a new file link it in the list of spell files. */
|
|
|
2318 if (old_lp == NULL)
|
|
|
2319 {
|
|
|
2320 lp->sl_next = first_lang;
|
|
|
2321 first_lang = lp;
|
|
|
2322 }
|
|
307
|
2323
|
|
255
|
2324 goto endOK;
|
|
|
2325
|
|
|
2326 endFAIL:
|
|
310
|
2327 if (lang != NULL)
|
|
|
2328 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2329 *lang = NUL;
|
|
|
2330 if (lp != NULL && old_lp == NULL)
|
|
323
|
2331 {
|
|
307
|
2332 slang_free(lp);
|
|
323
|
2333 lp = NULL;
|
|
|
2334 }
|
|
255
|
2335
|
|
|
2336 endOK:
|
|
236
|
2337 if (fd != NULL)
|
|
|
2338 fclose(fd);
|
|
|
2339 sourcing_name = save_sourcing_name;
|
|
|
2340 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2341
|
|
|
2342 return lp;
|
|
236
|
2343 }
|
|
|
2344
|
|
376
|
2345 /*
|
|
|
2346 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2347 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2348 * Returns NULL when the count is zero.
|
|
389
|
2349 * Sets "*cntp" to -1 when there is an error, length of the result otherwise.
|
|
376
|
2350 */
|
|
|
2351 static char_u *
|
|
389
|
2352 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2353 FILE *fd;
|
|
|
2354 int cnt_bytes;
|
|
389
|
2355 int *cntp;
|
|
376
|
2356 {
|
|
|
2357 int cnt = 0;
|
|
|
2358 int i;
|
|
|
2359 char_u *str;
|
|
|
2360
|
|
|
2361 /* read the length bytes, MSB first */
|
|
|
2362 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2363 cnt = (cnt << 8) + getc(fd);
|
|
|
2364 if (cnt < 0)
|
|
|
2365 {
|
|
381
|
2366 EMSG(_(e_spell_trunc));
|
|
389
|
2367 *cntp = -1;
|
|
376
|
2368 return NULL;
|
|
|
2369 }
|
|
389
|
2370 *cntp = cnt;
|
|
|
2371 if (cnt == 0)
|
|
|
2372 return NULL; /* nothing to read, return NULL */
|
|
376
|
2373
|
|
|
2374 /* allocate memory */
|
|
|
2375 str = alloc((unsigned)cnt + 1);
|
|
|
2376 if (str == NULL)
|
|
|
2377 {
|
|
389
|
2378 *cntp = -1;
|
|
376
|
2379 return NULL;
|
|
|
2380 }
|
|
|
2381
|
|
|
2382 /* Read the string. Doesn't check for truncated file. */
|
|
|
2383 for (i = 0; i < cnt; ++i)
|
|
|
2384 str[i] = getc(fd);
|
|
|
2385 str[i] = NUL;
|
|
|
2386
|
|
|
2387 return str;
|
|
|
2388 }
|
|
|
2389
|
|
381
|
2390 /*
|
|
|
2391 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
|
2392 * Returns FAIL when there is something wrong.
|
|
|
2393 */
|
|
|
2394 static int
|
|
|
2395 set_sofo(lp, from, to)
|
|
|
2396 slang_T *lp;
|
|
|
2397 char_u *from;
|
|
|
2398 char_u *to;
|
|
|
2399 {
|
|
|
2400 int i;
|
|
|
2401
|
|
|
2402 #ifdef FEAT_MBYTE
|
|
|
2403 garray_T *gap;
|
|
|
2404 char_u *s;
|
|
|
2405 char_u *p;
|
|
|
2406 int c;
|
|
|
2407 int *inp;
|
|
|
2408
|
|
|
2409 if (has_mbyte)
|
|
|
2410 {
|
|
|
2411 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2412 * characters. The index is the low byte of the character.
|
|
|
2413 * The list contains from-to pairs with a terminating NUL.
|
|
|
2414 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2415 gap = &lp->sl_sal;
|
|
|
2416 ga_init2(gap, sizeof(int *), 1);
|
|
|
2417 if (ga_grow(gap, 256) == FAIL)
|
|
|
2418 return FAIL;
|
|
|
2419 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2420 gap->ga_len = 256;
|
|
|
2421
|
|
|
2422 /* First count the number of items for each list. Temporarily use
|
|
|
2423 * sl_sal_first[] for this. */
|
|
|
2424 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2425 {
|
|
474
|
2426 c = mb_cptr2char_adv(&p);
|
|
|
2427 mb_cptr_adv(s);
|
|
381
|
2428 if (c >= 256)
|
|
|
2429 ++lp->sl_sal_first[c & 0xff];
|
|
|
2430 }
|
|
|
2431 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
|
2432 return FAIL;
|
|
|
2433
|
|
|
2434 /* Allocate the lists. */
|
|
|
2435 for (i = 0; i < 256; ++i)
|
|
|
2436 if (lp->sl_sal_first[i] > 0)
|
|
|
2437 {
|
|
|
2438 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2439 if (p == NULL)
|
|
|
2440 return FAIL;
|
|
|
2441 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2442 *(int *)p = 0;
|
|
|
2443 }
|
|
|
2444
|
|
|
2445 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
2446 * list. */
|
|
|
2447 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2448 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2449 {
|
|
474
|
2450 c = mb_cptr2char_adv(&p);
|
|
|
2451 i = mb_cptr2char_adv(&s);
|
|
381
|
2452 if (c >= 256)
|
|
|
2453 {
|
|
|
2454 /* Append the from-to chars at the end of the list with
|
|
|
2455 * the low byte. */
|
|
|
2456 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2457 while (*inp != 0)
|
|
|
2458 ++inp;
|
|
|
2459 *inp++ = c; /* from char */
|
|
|
2460 *inp++ = i; /* to char */
|
|
|
2461 *inp++ = NUL; /* NUL at the end */
|
|
|
2462 }
|
|
|
2463 else
|
|
|
2464 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2465 lp->sl_sal_first[c] = i;
|
|
|
2466 }
|
|
|
2467 }
|
|
|
2468 else
|
|
|
2469 #endif
|
|
|
2470 {
|
|
|
2471 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2472 if (STRLEN(from) != STRLEN(to))
|
|
|
2473 return FAIL;
|
|
|
2474
|
|
|
2475 for (i = 0; to[i] != NUL; ++i)
|
|
|
2476 lp->sl_sal_first[from[i]] = to[i];
|
|
|
2477 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2478 }
|
|
|
2479
|
|
|
2480 return OK;
|
|
|
2481 }
|
|
|
2482
|
|
|
2483 /*
|
|
|
2484 * Fill the first-index table for "lp".
|
|
|
2485 */
|
|
|
2486 static void
|
|
|
2487 set_sal_first(lp)
|
|
|
2488 slang_T *lp;
|
|
|
2489 {
|
|
|
2490 salfirst_T *sfirst;
|
|
|
2491 int i;
|
|
|
2492 salitem_T *smp;
|
|
|
2493 int c;
|
|
|
2494 garray_T *gap = &lp->sl_sal;
|
|
|
2495
|
|
|
2496 sfirst = lp->sl_sal_first;
|
|
|
2497 for (i = 0; i < 256; ++i)
|
|
|
2498 sfirst[i] = -1;
|
|
|
2499 smp = (salitem_T *)gap->ga_data;
|
|
|
2500 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2501 {
|
|
|
2502 #ifdef FEAT_MBYTE
|
|
|
2503 if (has_mbyte)
|
|
|
2504 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2505 * the character, for other encodings it should differ for most
|
|
|
2506 * characters. */
|
|
|
2507 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2508 else
|
|
|
2509 #endif
|
|
|
2510 c = *smp[i].sm_lead;
|
|
|
2511 if (sfirst[c] == -1)
|
|
|
2512 {
|
|
|
2513 sfirst[c] = i;
|
|
|
2514 #ifdef FEAT_MBYTE
|
|
|
2515 if (has_mbyte)
|
|
|
2516 {
|
|
|
2517 int n;
|
|
|
2518
|
|
|
2519 /* Make sure all entries with this byte are following each
|
|
|
2520 * other. Move the ones that are in the wrong position. Do
|
|
|
2521 * keep the same ordering! */
|
|
|
2522 while (i + 1 < gap->ga_len
|
|
|
2523 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
2524 /* Skip over entry with same index byte. */
|
|
|
2525 ++i;
|
|
|
2526
|
|
|
2527 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
2528 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
2529 {
|
|
|
2530 salitem_T tsal;
|
|
|
2531
|
|
|
2532 /* Move entry with same index byte after the entries
|
|
|
2533 * we already found. */
|
|
|
2534 ++i;
|
|
|
2535 --n;
|
|
|
2536 tsal = smp[i + n];
|
|
|
2537 mch_memmove(smp + i + 1, smp + i,
|
|
|
2538 sizeof(salitem_T) * n);
|
|
|
2539 smp[i] = tsal;
|
|
|
2540 }
|
|
|
2541 }
|
|
|
2542 #endif
|
|
|
2543 }
|
|
|
2544 }
|
|
|
2545 }
|
|
376
|
2546
|
|
372
|
2547 #ifdef FEAT_MBYTE
|
|
|
2548 /*
|
|
|
2549 * Turn a multi-byte string into a wide character string.
|
|
|
2550 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
2551 */
|
|
|
2552 static int *
|
|
|
2553 mb_str2wide(s)
|
|
|
2554 char_u *s;
|
|
|
2555 {
|
|
|
2556 int *res;
|
|
|
2557 char_u *p;
|
|
|
2558 int i = 0;
|
|
|
2559
|
|
|
2560 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
2561 if (res != NULL)
|
|
|
2562 {
|
|
|
2563 for (p = s; *p != NUL; )
|
|
|
2564 res[i++] = mb_ptr2char_adv(&p);
|
|
|
2565 res[i] = NUL;
|
|
|
2566 }
|
|
|
2567 return res;
|
|
|
2568 }
|
|
|
2569 #endif
|
|
|
2570
|
|
236
|
2571 /*
|
|
300
|
2572 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
2573 * "byts" and index array "idxs". Recursively read the children.
|
|
|
2574 *
|
|
346
|
2575 * NOTE: The code here must match put_node().
|
|
300
|
2576 *
|
|
|
2577 * Returns the index follosing the siblings.
|
|
|
2578 * Returns -1 if the file is shorter than expected.
|
|
|
2579 * Returns -2 if there is a format error.
|
|
236
|
2580 */
|
|
324
|
2581 static idx_T
|
|
339
|
2582 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
2583 FILE *fd;
|
|
|
2584 char_u *byts;
|
|
324
|
2585 idx_T *idxs;
|
|
300
|
2586 int maxidx; /* size of arrays */
|
|
324
|
2587 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
2588 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
2589 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
2590 {
|
|
300
|
2591 int len;
|
|
|
2592 int i;
|
|
|
2593 int n;
|
|
324
|
2594 idx_T idx = startidx;
|
|
300
|
2595 int c;
|
|
366
|
2596 int c2;
|
|
300
|
2597 #define SHARED_MASK 0x8000000
|
|
236
|
2598
|
|
300
|
2599 len = getc(fd); /* <siblingcount> */
|
|
|
2600 if (len <= 0)
|
|
|
2601 return -1;
|
|
|
2602
|
|
|
2603 if (startidx + len >= maxidx)
|
|
|
2604 return -2;
|
|
|
2605 byts[idx++] = len;
|
|
|
2606
|
|
|
2607 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
2608 for (i = 1; i <= len; ++i)
|
|
236
|
2609 {
|
|
300
|
2610 c = getc(fd); /* <byte> */
|
|
|
2611 if (c < 0)
|
|
|
2612 return -1;
|
|
|
2613 if (c <= BY_SPECIAL)
|
|
|
2614 {
|
|
366
|
2615 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
2616 {
|
|
|
2617 /* No flags, all regions. */
|
|
|
2618 idxs[idx] = 0;
|
|
|
2619 c = 0;
|
|
|
2620 }
|
|
390
|
2621 else if (c != BY_INDEX)
|
|
300
|
2622 {
|
|
339
|
2623 if (prefixtree)
|
|
|
2624 {
|
|
455
|
2625 /* Read the optional pflags byte, the prefix ID and the
|
|
|
2626 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
2627 * byte, the condition index shifted up 8 bits, the flags
|
|
|
2628 * shifted up 24 bits. */
|
|
|
2629 if (c == BY_FLAGS)
|
|
|
2630 c = getc(fd) << 24; /* <pflags> */
|
|
|
2631 else
|
|
|
2632 c = 0;
|
|
|
2633
|
|
481
|
2634 c |= getc(fd); /* <affixID> */
|
|
455
|
2635
|
|
339
|
2636 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
2637 if (n >= maxprefcondnr)
|
|
|
2638 return -2;
|
|
455
|
2639 c |= (n << 8);
|
|
339
|
2640 }
|
|
390
|
2641 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
2642 {
|
|
|
2643 /* Read flags and optional region and prefix ID. In
|
|
390
|
2644 * idxs[] the flags go in the low two bytes, region above
|
|
|
2645 * that and prefix ID above the region. */
|
|
|
2646 c2 = c;
|
|
339
|
2647 c = getc(fd); /* <flags> */
|
|
390
|
2648 if (c2 == BY_FLAGS2)
|
|
|
2649 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
2650 if (c & WF_REGION)
|
|
390
|
2651 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
2652 if (c & WF_AFX)
|
|
|
2653 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
2654 }
|
|
|
2655
|
|
300
|
2656 idxs[idx] = c;
|
|
|
2657 c = 0;
|
|
|
2658 }
|
|
|
2659 else /* c == BY_INDEX */
|
|
|
2660 {
|
|
|
2661 /* <nodeidx> */
|
|
|
2662 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2663 if (n < 0 || n >= maxidx)
|
|
|
2664 return -2;
|
|
|
2665 idxs[idx] = n + SHARED_MASK;
|
|
|
2666 c = getc(fd); /* <xbyte> */
|
|
|
2667 }
|
|
|
2668 }
|
|
|
2669 byts[idx++] = c;
|
|
236
|
2670 }
|
|
|
2671
|
|
300
|
2672 /* Recursively read the children for non-shared siblings.
|
|
|
2673 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
2674 * remove SHARED_MASK) */
|
|
|
2675 for (i = 1; i <= len; ++i)
|
|
|
2676 if (byts[startidx + i] != 0)
|
|
|
2677 {
|
|
|
2678 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
2679 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
2680 else
|
|
|
2681 {
|
|
|
2682 idxs[startidx + i] = idx;
|
|
339
|
2683 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
2684 prefixtree, maxprefcondnr);
|
|
300
|
2685 if (idx < 0)
|
|
|
2686 break;
|
|
|
2687 }
|
|
|
2688 }
|
|
236
|
2689
|
|
300
|
2690 return idx;
|
|
236
|
2691 }
|
|
|
2692
|
|
|
2693 /*
|
|
|
2694 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
2695 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
2696 */
|
|
|
2697 char_u *
|
|
|
2698 did_set_spelllang(buf)
|
|
|
2699 buf_T *buf;
|
|
|
2700 {
|
|
|
2701 garray_T ga;
|
|
351
|
2702 char_u *splp;
|
|
236
|
2703 char_u *region;
|
|
416
|
2704 char_u region_cp[3];
|
|
355
|
2705 int filename;
|
|
236
|
2706 int region_mask;
|
|
|
2707 slang_T *lp;
|
|
|
2708 int c;
|
|
351
|
2709 char_u lang[MAXWLEN + 1];
|
|
323
|
2710 char_u spf_name[MAXPATHL];
|
|
351
|
2711 int len;
|
|
|
2712 char_u *p;
|
|
381
|
2713 int round;
|
|
385
|
2714 char_u *spf;
|
|
389
|
2715 char_u *use_region = NULL;
|
|
|
2716 int dont_use_region = FALSE;
|
|
236
|
2717
|
|
|
2718 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
2719 clear_midword(buf);
|
|
236
|
2720
|
|
351
|
2721 /* loop over comma separated language names. */
|
|
|
2722 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
2723 {
|
|
|
2724 /* Get one language name. */
|
|
|
2725 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
2726
|
|
240
|
2727 region = NULL;
|
|
351
|
2728 len = STRLEN(lang);
|
|
355
|
2729
|
|
|
2730 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
2731 * If there is a region name let "region" point to it and remove it
|
|
|
2732 * from the name. */
|
|
|
2733 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
2734 {
|
|
355
|
2735 filename = TRUE;
|
|
|
2736
|
|
416
|
2737 /* Locate a region and remove it from the file name. */
|
|
|
2738 p = vim_strchr(gettail(lang), '_');
|
|
|
2739 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
2740 && !ASCII_ISALPHA(p[3]))
|
|
|
2741 {
|
|
|
2742 vim_strncpy(region_cp, p + 1, 2);
|
|
|
2743 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
2744 len -= 3;
|
|
|
2745 region = region_cp;
|
|
|
2746 }
|
|
|
2747 else
|
|
|
2748 dont_use_region = TRUE;
|
|
|
2749
|
|
355
|
2750 /* Check if we loaded this language before. */
|
|
|
2751 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2752 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2753 break;
|
|
236
|
2754 }
|
|
355
|
2755 else
|
|
|
2756 {
|
|
|
2757 filename = FALSE;
|
|
|
2758 if (len > 3 && lang[len - 3] == '_')
|
|
|
2759 {
|
|
|
2760 region = lang + len - 2;
|
|
|
2761 len -= 3;
|
|
|
2762 lang[len] = NUL;
|
|
389
|
2763 }
|
|
|
2764 else
|
|
|
2765 dont_use_region = TRUE;
|
|
355
|
2766
|
|
|
2767 /* Check if we loaded this language before. */
|
|
|
2768 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2769 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
2770 break;
|
|
|
2771 }
|
|
236
|
2772
|
|
416
|
2773 if (region != NULL)
|
|
|
2774 {
|
|
|
2775 /* If the region differs from what was used before then don't
|
|
|
2776 * use it for 'spellfile'. */
|
|
|
2777 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
2778 dont_use_region = TRUE;
|
|
|
2779 use_region = region;
|
|
|
2780 }
|
|
|
2781
|
|
351
|
2782 /* If not found try loading the language now. */
|
|
236
|
2783 if (lp == NULL)
|
|
355
|
2784 {
|
|
|
2785 if (filename)
|
|
|
2786 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
2787 else
|
|
|
2788 spell_load_lang(lang);
|
|
|
2789 }
|
|
236
|
2790
|
|
307
|
2791 /*
|
|
351
|
2792 * Loop over the languages, there can be several files for "lang".
|
|
307
|
2793 */
|
|
|
2794 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
2795 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
2796 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
2797 {
|
|
316
|
2798 region_mask = REGION_ALL;
|
|
355
|
2799 if (!filename && region != NULL)
|
|
236
|
2800 {
|
|
307
|
2801 /* find region in sl_regions */
|
|
|
2802 c = find_region(lp->sl_regions, region);
|
|
|
2803 if (c == REGION_ALL)
|
|
|
2804 {
|
|
389
|
2805 if (lp->sl_add)
|
|
|
2806 {
|
|
|
2807 if (*lp->sl_regions != NUL)
|
|
|
2808 /* This addition file is for other regions. */
|
|
|
2809 region_mask = 0;
|
|
|
2810 }
|
|
|
2811 else
|
|
|
2812 /* This is probably an error. Give a warning and
|
|
|
2813 * accept the words anyway. */
|
|
351
|
2814 smsg((char_u *)
|
|
|
2815 _("Warning: region %s not supported"),
|
|
|
2816 region);
|
|
307
|
2817 }
|
|
|
2818 else
|
|
|
2819 region_mask = 1 << c;
|
|
236
|
2820 }
|
|
307
|
2821
|
|
389
|
2822 if (region_mask != 0)
|
|
307
|
2823 {
|
|
389
|
2824 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2825 {
|
|
|
2826 ga_clear(&ga);
|
|
|
2827 return e_outofmem;
|
|
|
2828 }
|
|
|
2829 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2830 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2831 ++ga.ga_len;
|
|
|
2832 use_midword(lp, buf);
|
|
307
|
2833 }
|
|
385
|
2834 }
|
|
|
2835 }
|
|
|
2836
|
|
|
2837 /* round 0: load int_wordlist, if possible.
|
|
|
2838 * round 1: load first name in 'spellfile'.
|
|
|
2839 * round 2: load second name in 'spellfile.
|
|
|
2840 * etc. */
|
|
|
2841 spf = curbuf->b_p_spf;
|
|
|
2842 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
2843 {
|
|
|
2844 if (round == 0)
|
|
|
2845 {
|
|
|
2846 /* Internal wordlist, if there is one. */
|
|
|
2847 if (int_wordlist == NULL)
|
|
381
|
2848 continue;
|
|
385
|
2849 int_wordlist_spl(spf_name);
|
|
381
|
2850 }
|
|
|
2851 else
|
|
|
2852 {
|
|
385
|
2853 /* One entry in 'spellfile'. */
|
|
|
2854 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
2855 STRCAT(spf_name, ".spl");
|
|
|
2856
|
|
|
2857 /* If it was already found above then skip it. */
|
|
|
2858 for (c = 0; c < ga.ga_len; ++c)
|
|
|
2859 if (fullpathcmp(spf_name,
|
|
|
2860 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
2861 FALSE) == FPC_SAME)
|
|
|
2862 break;
|
|
|
2863 if (c < ga.ga_len)
|
|
381
|
2864 continue;
|
|
|
2865 }
|
|
|
2866
|
|
351
|
2867 /* Check if it was loaded already. */
|
|
323
|
2868 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2869 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2870 break;
|
|
|
2871 if (lp == NULL)
|
|
|
2872 {
|
|
351
|
2873 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
2874 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
2875 * use an arbitrary name. */
|
|
|
2876 if (round == 0)
|
|
|
2877 STRCPY(lang, "internal wordlist");
|
|
|
2878 else
|
|
|
2879 {
|
|
|
2880 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
2881 p = vim_strchr(lang, '.');
|
|
|
2882 if (p != NULL)
|
|
|
2883 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
2884 }
|
|
351
|
2885 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
2886 }
|
|
|
2887 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
2888 {
|
|
389
|
2889 region_mask = REGION_ALL;
|
|
|
2890 if (use_region != NULL && !dont_use_region)
|
|
|
2891 {
|
|
|
2892 /* find region in sl_regions */
|
|
|
2893 c = find_region(lp->sl_regions, use_region);
|
|
|
2894 if (c != REGION_ALL)
|
|
|
2895 region_mask = 1 << c;
|
|
|
2896 else if (*lp->sl_regions != NUL)
|
|
|
2897 /* This spell file is for other regions. */
|
|
|
2898 region_mask = 0;
|
|
|
2899 }
|
|
|
2900
|
|
|
2901 if (region_mask != 0)
|
|
|
2902 {
|
|
|
2903 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2904 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2905 ++ga.ga_len;
|
|
|
2906 use_midword(lp, buf);
|
|
|
2907 }
|
|
323
|
2908 }
|
|
|
2909 }
|
|
|
2910
|
|
236
|
2911 /* Add a NULL entry to mark the end of the list. */
|
|
|
2912 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2913 {
|
|
|
2914 ga_clear(&ga);
|
|
|
2915 return e_outofmem;
|
|
|
2916 }
|
|
|
2917 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
2918 ++ga.ga_len;
|
|
|
2919
|
|
|
2920 /* Everything is fine, store the new b_langp value. */
|
|
|
2921 ga_clear(&buf->b_langp);
|
|
|
2922 buf->b_langp = ga;
|
|
|
2923
|
|
|
2924 return NULL;
|
|
|
2925 }
|
|
|
2926
|
|
|
2927 /*
|
|
376
|
2928 * Clear the midword characters for buffer "buf".
|
|
|
2929 */
|
|
|
2930 static void
|
|
|
2931 clear_midword(buf)
|
|
|
2932 buf_T *buf;
|
|
|
2933 {
|
|
|
2934 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
2935 #ifdef FEAT_MBYTE
|
|
|
2936 vim_free(buf->b_spell_ismw_mb);
|
|
|
2937 buf->b_spell_ismw_mb = NULL;
|
|
|
2938 #endif
|
|
|
2939 }
|
|
|
2940
|
|
|
2941 /*
|
|
|
2942 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
2943 * They add up to any currently used midword characters.
|
|
|
2944 */
|
|
|
2945 static void
|
|
|
2946 use_midword(lp, buf)
|
|
|
2947 slang_T *lp;
|
|
|
2948 buf_T *buf;
|
|
|
2949 {
|
|
|
2950 char_u *p;
|
|
|
2951
|
|
389
|
2952 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
2953 return;
|
|
|
2954
|
|
376
|
2955 for (p = lp->sl_midword; *p != NUL; )
|
|
|
2956 #ifdef FEAT_MBYTE
|
|
|
2957 if (has_mbyte)
|
|
|
2958 {
|
|
|
2959 int c, l, n;
|
|
|
2960 char_u *bp;
|
|
|
2961
|
|
|
2962 c = mb_ptr2char(p);
|
|
474
|
2963 l = (*mb_ptr2len)(p);
|
|
|
2964 if (c < 256 && l <= 2)
|
|
376
|
2965 buf->b_spell_ismw[c] = TRUE;
|
|
|
2966 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
2967 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
2968 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
2969 else
|
|
|
2970 {
|
|
|
2971 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
2972 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
2973 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
2974 if (bp != NULL)
|
|
|
2975 {
|
|
|
2976 vim_free(buf->b_spell_ismw_mb);
|
|
|
2977 buf->b_spell_ismw_mb = bp;
|
|
|
2978 vim_strncpy(bp + n, p, l);
|
|
|
2979 }
|
|
|
2980 }
|
|
|
2981 p += l;
|
|
|
2982 }
|
|
|
2983 else
|
|
|
2984 #endif
|
|
|
2985 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
2986 }
|
|
|
2987
|
|
|
2988 /*
|
|
236
|
2989 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
2990 * Each region is simply stored as the two characters of it's name.
|
|
381
|
2991 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
2992 */
|
|
|
2993 static int
|
|
|
2994 find_region(rp, region)
|
|
|
2995 char_u *rp;
|
|
|
2996 char_u *region;
|
|
|
2997 {
|
|
|
2998 int i;
|
|
|
2999
|
|
|
3000 for (i = 0; ; i += 2)
|
|
|
3001 {
|
|
|
3002 if (rp[i] == NUL)
|
|
|
3003 return REGION_ALL;
|
|
|
3004 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
3005 break;
|
|
|
3006 }
|
|
|
3007 return i / 2;
|
|
|
3008 }
|
|
|
3009
|
|
|
3010 /*
|
|
323
|
3011 * Return case type of word:
|
|
236
|
3012 * w word 0
|
|
300
|
3013 * Word WF_ONECAP
|
|
|
3014 * W WORD WF_ALLCAP
|
|
|
3015 * WoRd wOrd WF_KEEPCAP
|
|
236
|
3016 */
|
|
|
3017 static int
|
|
|
3018 captype(word, end)
|
|
|
3019 char_u *word;
|
|
323
|
3020 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
3021 {
|
|
|
3022 char_u *p;
|
|
|
3023 int c;
|
|
|
3024 int firstcap;
|
|
|
3025 int allcap;
|
|
|
3026 int past_second = FALSE; /* past second word char */
|
|
|
3027
|
|
|
3028 /* find first letter */
|
|
376
|
3029 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
3030 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
3031 return 0; /* only non-word characters, illegal word */
|
|
|
3032 #ifdef FEAT_MBYTE
|
|
310
|
3033 if (has_mbyte)
|
|
|
3034 c = mb_ptr2char_adv(&p);
|
|
|
3035 else
|
|
236
|
3036 #endif
|
|
310
|
3037 c = *p++;
|
|
324
|
3038 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
3039
|
|
|
3040 /*
|
|
|
3041 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
3042 * But a word with an upper char only at start is a ONECAP.
|
|
|
3043 */
|
|
323
|
3044 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
3045 if (spell_iswordp_nmw(p))
|
|
236
|
3046 {
|
|
455
|
3047 c = PTR2CHAR(p);
|
|
324
|
3048 if (!SPELL_ISUPPER(c))
|
|
236
|
3049 {
|
|
|
3050 /* UUl -> KEEPCAP */
|
|
|
3051 if (past_second && allcap)
|
|
300
|
3052 return WF_KEEPCAP;
|
|
236
|
3053 allcap = FALSE;
|
|
|
3054 }
|
|
|
3055 else if (!allcap)
|
|
|
3056 /* UlU -> KEEPCAP */
|
|
300
|
3057 return WF_KEEPCAP;
|
|
236
|
3058 past_second = TRUE;
|
|
|
3059 }
|
|
|
3060
|
|
|
3061 if (allcap)
|
|
300
|
3062 return WF_ALLCAP;
|
|
236
|
3063 if (firstcap)
|
|
300
|
3064 return WF_ONECAP;
|
|
236
|
3065 return 0;
|
|
|
3066 }
|
|
|
3067
|
|
474
|
3068 /*
|
|
|
3069 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
3070 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
3071 * Add ALLCAP for "WOrD".
|
|
|
3072 */
|
|
|
3073 static int
|
|
|
3074 badword_captype(word, end)
|
|
|
3075 char_u *word;
|
|
|
3076 char_u *end;
|
|
|
3077 {
|
|
|
3078 int flags = captype(word, end);
|
|
475
|
3079 int c;
|
|
474
|
3080 int l, u;
|
|
|
3081 int first;
|
|
|
3082 char_u *p;
|
|
|
3083
|
|
|
3084 if (flags & WF_KEEPCAP)
|
|
|
3085 {
|
|
|
3086 /* Count the number of UPPER and lower case letters. */
|
|
|
3087 l = u = 0;
|
|
|
3088 first = FALSE;
|
|
|
3089 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
3090 {
|
|
475
|
3091 c = PTR2CHAR(p);
|
|
|
3092 if (SPELL_ISUPPER(c))
|
|
474
|
3093 {
|
|
|
3094 ++u;
|
|
|
3095 if (p == word)
|
|
|
3096 first = TRUE;
|
|
|
3097 }
|
|
|
3098 else
|
|
|
3099 ++l;
|
|
|
3100 }
|
|
|
3101
|
|
|
3102 /* If there are more UPPER than lower case letters suggest an
|
|
|
3103 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
3104 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
3105 * require three upper case letters. */
|
|
|
3106 if (u > l && u > 2)
|
|
|
3107 flags |= WF_ALLCAP;
|
|
|
3108 else if (first)
|
|
|
3109 flags |= WF_ONECAP;
|
|
|
3110 }
|
|
|
3111 return flags;
|
|
|
3112 }
|
|
|
3113
|
|
355
|
3114 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
3115 /*
|
|
|
3116 * Free all languages.
|
|
|
3117 */
|
|
|
3118 void
|
|
|
3119 spell_free_all()
|
|
|
3120 {
|
|
|
3121 slang_T *lp;
|
|
|
3122 buf_T *buf;
|
|
385
|
3123 char_u fname[MAXPATHL];
|
|
355
|
3124
|
|
|
3125 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3126 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3127 ga_clear(&buf->b_langp);
|
|
|
3128
|
|
|
3129 while (first_lang != NULL)
|
|
|
3130 {
|
|
|
3131 lp = first_lang;
|
|
|
3132 first_lang = lp->sl_next;
|
|
|
3133 slang_free(lp);
|
|
|
3134 }
|
|
366
|
3135
|
|
385
|
3136 if (int_wordlist != NULL)
|
|
|
3137 {
|
|
|
3138 /* Delete the internal wordlist and its .spl file */
|
|
|
3139 mch_remove(int_wordlist);
|
|
|
3140 int_wordlist_spl(fname);
|
|
|
3141 mch_remove(fname);
|
|
|
3142 vim_free(int_wordlist);
|
|
|
3143 int_wordlist = NULL;
|
|
381
|
3144 }
|
|
|
3145
|
|
366
|
3146 init_spell_chartab();
|
|
355
|
3147 }
|
|
|
3148 # endif
|
|
|
3149
|
|
236
|
3150 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
3151 /*
|
|
|
3152 * Clear all spelling tables and reload them.
|
|
307
|
3153 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
3154 */
|
|
|
3155 void
|
|
|
3156 spell_reload()
|
|
|
3157 {
|
|
|
3158 buf_T *buf;
|
|
316
|
3159 win_T *wp;
|
|
236
|
3160
|
|
358
|
3161 /* Initialize the table for spell_iswordp(). */
|
|
236
|
3162 init_spell_chartab();
|
|
|
3163
|
|
|
3164 /* Unload all allocated memory. */
|
|
355
|
3165 spell_free_all();
|
|
236
|
3166
|
|
|
3167 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3168 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3169 {
|
|
316
|
3170 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
3171 * window for this buffer in which 'spell' is set. */
|
|
236
|
3172 if (*buf->b_p_spl != NUL)
|
|
316
|
3173 {
|
|
|
3174 FOR_ALL_WINDOWS(wp)
|
|
|
3175 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
3176 {
|
|
|
3177 (void)did_set_spelllang(buf);
|
|
|
3178 # ifdef FEAT_WINDOWS
|
|
|
3179 break;
|
|
|
3180 # endif
|
|
|
3181 }
|
|
|
3182 }
|
|
236
|
3183 }
|
|
|
3184 }
|
|
|
3185 # endif
|
|
|
3186
|
|
310
|
3187 /*
|
|
|
3188 * Reload the spell file "fname" if it's loaded.
|
|
|
3189 */
|
|
|
3190 static void
|
|
323
|
3191 spell_reload_one(fname, added_word)
|
|
310
|
3192 char_u *fname;
|
|
323
|
3193 int added_word; /* invoked through "zg" */
|
|
310
|
3194 {
|
|
|
3195 slang_T *lp;
|
|
323
|
3196 int didit = FALSE;
|
|
310
|
3197
|
|
|
3198 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3199 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3200 {
|
|
|
3201 slang_clear(lp);
|
|
323
|
3202 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3203 redraw_all_later(NOT_VALID);
|
|
323
|
3204 didit = TRUE;
|
|
310
|
3205 }
|
|
323
|
3206
|
|
|
3207 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3208 * 'spelllang' to get it loaded. */
|
|
|
3209 if (added_word && !didit)
|
|
|
3210 did_set_spelllang(curbuf);
|
|
310
|
3211 }
|
|
|
3212
|
|
|
3213
|
|
236
|
3214 /*
|
|
|
3215 * Functions for ":mkspell".
|
|
|
3216 */
|
|
|
3217
|
|
300
|
3218 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3219 and .dic file. */
|
|
|
3220 /*
|
|
|
3221 * Main structure to store the contents of a ".aff" file.
|
|
|
3222 */
|
|
|
3223 typedef struct afffile_S
|
|
|
3224 {
|
|
|
3225 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
481
|
3226 int af_slash; /* character used in word for slash */
|
|
310
|
3227 int af_rar; /* RAR ID for rare word */
|
|
|
3228 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
3229 int af_bad; /* BAD ID for banned word */
|
|
481
|
3230 char_u *af_compflags; /* COMPOUNDFLAG or COMPOUNDFLAGS */
|
|
|
3231 int af_compminlen; /* COMPOUNDMIN */
|
|
339
|
3232 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3233 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3234 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
3235 } afffile_T;
|
|
|
3236
|
|
|
3237 typedef struct affentry_S affentry_T;
|
|
|
3238 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3239 struct affentry_S
|
|
|
3240 {
|
|
|
3241 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3242 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3243 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3244 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3245 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
366
|
3246 int ae_rare; /* rare affix */
|
|
300
|
3247 };
|
|
|
3248
|
|
455
|
3249 #define AH_KEY_LEN 10
|
|
|
3250
|
|
300
|
3251 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3252 typedef struct affheader_S
|
|
|
3253 {
|
|
455
|
3254 /* key for hashtable == name of affix entry */
|
|
|
3255 #ifdef FEAT_MBYTE
|
|
|
3256 char_u ah_key[AH_KEY_LEN]; /* multi-byte char plus NUL */
|
|
|
3257 #else
|
|
|
3258 char_u ah_key[2]; /* one byte char plus NUL */
|
|
|
3259 #endif
|
|
339
|
3260 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
3261 int ah_combine; /* suffix may combine with prefix */
|
|
|
3262 affentry_T *ah_first; /* first affix entry */
|
|
|
3263 } affheader_T;
|
|
|
3264
|
|
|
3265 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
3266
|
|
|
3267 /*
|
|
|
3268 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
3269 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
3270 * once after ":mkspell" is done.
|
|
300
|
3271 */
|
|
|
3272 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
3273 typedef struct sblock_S sblock_T;
|
|
|
3274 struct sblock_S
|
|
|
3275 {
|
|
|
3276 sblock_T *sb_next; /* next block in list */
|
|
|
3277 int sb_used; /* nr of bytes already in use */
|
|
|
3278 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
3279 };
|
|
|
3280
|
|
|
3281 /*
|
|
300
|
3282 * A node in the tree.
|
|
236
|
3283 */
|
|
300
|
3284 typedef struct wordnode_S wordnode_T;
|
|
|
3285 struct wordnode_S
|
|
236
|
3286 {
|
|
346
|
3287 union /* shared to save space */
|
|
|
3288 {
|
|
469
|
3289 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
3290 int index; /* index in written nodes (valid after first
|
|
|
3291 round) */
|
|
|
3292 } wn_u1;
|
|
|
3293 union /* shared to save space */
|
|
|
3294 {
|
|
|
3295 wordnode_T *next; /* next node with same hash key */
|
|
|
3296 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
3297 } wn_u2;
|
|
300
|
3298 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
3299 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
3300 always sorted) */
|
|
474
|
3301 int wn_refs; /* Nr. of references to this node. Only
|
|
|
3302 relevant for first node in a list of
|
|
|
3303 siblings, in following siblings it is
|
|
|
3304 always one. */
|
|
300
|
3305 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
481
|
3306 char_u wn_affixID; /* when "wn_byte" is NUL: supported/required
|
|
469
|
3307 prefix ID or 0 */
|
|
|
3308 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
3309 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
3310 PREFIXTREE it's the prefcondnr */
|
|
469
|
3311 #ifdef SPELL_PRINTTREE
|
|
|
3312 int wn_nr; /* sequence nr for printing */
|
|
|
3313 #endif
|
|
236
|
3314 };
|
|
|
3315
|
|
390
|
3316 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
3317
|
|
300
|
3318 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
3319
|
|
300
|
3320 /*
|
|
|
3321 * Info used while reading the spell files.
|
|
|
3322 */
|
|
|
3323 typedef struct spellinfo_S
|
|
249
|
3324 {
|
|
300
|
3325 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
3326 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
3327 int si_fold_added; /* nr of words added since compressing */
|
|
|
3328
|
|
300
|
3329 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
3330 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
3331 int si_keep_added; /* nr of words added since compressing */
|
|
|
3332
|
|
339
|
3333 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
3334
|
|
300
|
3335 sblock_T *si_blocks; /* memory blocks used */
|
|
474
|
3336 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
3337 compression, linked by "wn_child" field. */
|
|
|
3338 #ifdef SPELL_PRINTTREE
|
|
|
3339 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
3340 #endif
|
|
|
3341
|
|
|
3342
|
|
300
|
3343 int si_ascii; /* handling only ASCII words */
|
|
310
|
3344 int si_add; /* addition file */
|
|
351
|
3345 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
3346 int si_region; /* region mask */
|
|
|
3347 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
3348 int si_memtot; /* runtime memory used */
|
|
310
|
3349 int si_verbose; /* verbose messages */
|
|
474
|
3350 int si_msg_count; /* number of words added since last message */
|
|
316
|
3351 int si_region_count; /* number of regions supported (1 when there
|
|
|
3352 are no regions) */
|
|
|
3353 char_u si_region_name[16]; /* region names (if count > 1) */
|
|
323
|
3354
|
|
|
3355 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
3356 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
3357 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
3358 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
3359 int si_followup; /* soundsalike: ? */
|
|
|
3360 int si_collapse; /* soundsalike: ? */
|
|
|
3361 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
3362 garray_T si_map; /* MAP info concatenated */
|
|
366
|
3363 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
481
|
3364 int si_compminlen; /* minimal length for compounding */
|
|
|
3365 char_u *si_compflags; /* flags used for compounding */
|
|
339
|
3366 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
3367 * prefixes, each stored as a string */
|
|
|
3368 int si_newID; /* current value for ah_newID */
|
|
300
|
3369 } spellinfo_T;
|
|
249
|
3370
|
|
474
|
3371 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
339
|
3372 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
3373 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
3374 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
3375 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
3376 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
3377 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
|
3378 static char_u *get_pfxlist __ARGS((spellinfo_T *spin, afffile_T *affile, char_u *afflist));
|
|
481
|
3379 static char_u *get_compflags __ARGS((spellinfo_T *spin, char_u *afflist));
|
|
474
|
3380 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));
|
|
|
3381 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
3382 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
3383 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
3384 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
3385 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
|
3386 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist));
|
|
481
|
3387 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
3388 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
3389 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
3390 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
3391 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
3392 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
3393 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
474
|
3394 static void write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
3395 static void clear_node __ARGS((wordnode_T *node));
|
|
|
3396 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
3397 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3398 static void init_spellfile __ARGS((void));
|
|
236
|
3399
|
|
455
|
3400 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
3401 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
3402 * Use a negative number with the lower 8 bits zero. */
|
|
|
3403 #define PFX_FLAGS -256
|
|
390
|
3404
|
|
469
|
3405 #ifdef SPELL_PRINTTREE
|
|
|
3406 /*
|
|
|
3407 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
3408 * readable format, so that we can see what happens when adding a word and/or
|
|
|
3409 * compressing the tree.
|
|
|
3410 * Based on code from Olaf Seibert.
|
|
|
3411 */
|
|
|
3412 #define PRINTLINESIZE 1000
|
|
|
3413 #define PRINTWIDTH 6
|
|
|
3414
|
|
|
3415 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
3416 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
3417
|
|
|
3418 static char line1[PRINTLINESIZE];
|
|
|
3419 static char line2[PRINTLINESIZE];
|
|
|
3420 static char line3[PRINTLINESIZE];
|
|
|
3421
|
|
|
3422 static void
|
|
|
3423 spell_clear_flags(wordnode_T *node)
|
|
|
3424 {
|
|
|
3425 wordnode_T *np;
|
|
|
3426
|
|
|
3427 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
3428 {
|
|
|
3429 np->wn_u1.index = FALSE;
|
|
|
3430 spell_clear_flags(np->wn_child);
|
|
|
3431 }
|
|
|
3432 }
|
|
|
3433
|
|
|
3434 static void
|
|
|
3435 spell_print_node(wordnode_T *node, int depth)
|
|
|
3436 {
|
|
|
3437 if (node->wn_u1.index)
|
|
|
3438 {
|
|
|
3439 /* Done this node before, print the reference. */
|
|
|
3440 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
3441 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
3442 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3443 msg(line1);
|
|
|
3444 msg(line2);
|
|
|
3445 msg(line3);
|
|
|
3446 }
|
|
|
3447 else
|
|
|
3448 {
|
|
|
3449 node->wn_u1.index = TRUE;
|
|
|
3450
|
|
|
3451 if (node->wn_byte != NUL)
|
|
|
3452 {
|
|
|
3453 if (node->wn_child != NULL)
|
|
|
3454 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
3455 else
|
|
|
3456 /* Cannot happen? */
|
|
|
3457 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
3458 }
|
|
|
3459 else
|
|
|
3460 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
3461
|
|
|
3462 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
3463
|
|
|
3464 if (node->wn_sibling != NULL)
|
|
|
3465 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
3466 else
|
|
|
3467 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3468
|
|
|
3469 if (node->wn_byte == NUL)
|
|
|
3470 {
|
|
|
3471 msg(line1);
|
|
|
3472 msg(line2);
|
|
|
3473 msg(line3);
|
|
|
3474 }
|
|
|
3475
|
|
|
3476 /* do the children */
|
|
|
3477 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
3478 spell_print_node(node->wn_child, depth + 1);
|
|
|
3479
|
|
|
3480 /* do the siblings */
|
|
|
3481 if (node->wn_sibling != NULL)
|
|
|
3482 {
|
|
|
3483 /* get rid of all parent details except | */
|
|
|
3484 STRCPY(line1, line3);
|
|
|
3485 STRCPY(line2, line3);
|
|
|
3486 spell_print_node(node->wn_sibling, depth);
|
|
|
3487 }
|
|
|
3488 }
|
|
|
3489 }
|
|
|
3490
|
|
|
3491 static void
|
|
|
3492 spell_print_tree(wordnode_T *root)
|
|
|
3493 {
|
|
|
3494 if (root != NULL)
|
|
|
3495 {
|
|
|
3496 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
3497 * done. */
|
|
|
3498 spell_clear_flags(root);
|
|
|
3499
|
|
|
3500 /* Recursively print the tree. */
|
|
|
3501 spell_print_node(root, 0);
|
|
|
3502 }
|
|
|
3503 }
|
|
|
3504 #endif /* SPELL_PRINTTREE */
|
|
|
3505
|
|
236
|
3506 /*
|
|
323
|
3507 * Read the affix file "fname".
|
|
316
|
3508 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
3509 */
|
|
|
3510 static afffile_T *
|
|
474
|
3511 spell_read_aff(spin, fname)
|
|
|
3512 spellinfo_T *spin;
|
|
236
|
3513 char_u *fname;
|
|
|
3514 {
|
|
|
3515 FILE *fd;
|
|
|
3516 afffile_T *aff;
|
|
|
3517 char_u rline[MAXLINELEN];
|
|
|
3518 char_u *line;
|
|
|
3519 char_u *pc = NULL;
|
|
334
|
3520 #define MAXITEMCNT 7
|
|
|
3521 char_u *(items[MAXITEMCNT]);
|
|
236
|
3522 int itemcnt;
|
|
|
3523 char_u *p;
|
|
|
3524 int lnum = 0;
|
|
|
3525 affheader_T *cur_aff = NULL;
|
|
|
3526 int aff_todo = 0;
|
|
|
3527 hashtab_T *tp;
|
|
255
|
3528 char_u *low = NULL;
|
|
|
3529 char_u *fol = NULL;
|
|
|
3530 char_u *upp = NULL;
|
|
307
|
3531 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
3532 int do_rep;
|
|
|
3533 int do_sal;
|
|
|
3534 int do_map;
|
|
366
|
3535 int do_midword;
|
|
375
|
3536 int do_sofo;
|
|
323
|
3537 int found_map = FALSE;
|
|
324
|
3538 hashitem_T *hi;
|
|
455
|
3539 int l;
|
|
236
|
3540
|
|
300
|
3541 /*
|
|
|
3542 * Open the file.
|
|
|
3543 */
|
|
310
|
3544 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3545 if (fd == NULL)
|
|
|
3546 {
|
|
|
3547 EMSG2(_(e_notopen), fname);
|
|
|
3548 return NULL;
|
|
|
3549 }
|
|
|
3550
|
|
310
|
3551 if (spin->si_verbose || p_verbose > 2)
|
|
|
3552 {
|
|
|
3553 if (!spin->si_verbose)
|
|
|
3554 verbose_enter();
|
|
366
|
3555 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
3556 out_flush();
|
|
|
3557 if (!spin->si_verbose)
|
|
|
3558 verbose_leave();
|
|
|
3559 }
|
|
236
|
3560
|
|
323
|
3561 /* Only do REP lines when not done in another .aff file already. */
|
|
|
3562 do_rep = spin->si_rep.ga_len == 0;
|
|
|
3563
|
|
|
3564 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
3565 do_sal = spin->si_sal.ga_len == 0;
|
|
|
3566
|
|
|
3567 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
3568 do_map = spin->si_map.ga_len == 0;
|
|
|
3569
|
|
366
|
3570 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
3571 do_midword = spin->si_midword == NULL;
|
|
|
3572
|
|
375
|
3573 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
3574 do_sofo = spin->si_sofofr == NULL;
|
|
|
3575
|
|
300
|
3576 /*
|
|
|
3577 * Allocate and init the afffile_T structure.
|
|
|
3578 */
|
|
474
|
3579 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
3580 if (aff == NULL)
|
|
|
3581 return NULL;
|
|
|
3582 hash_init(&aff->af_pref);
|
|
|
3583 hash_init(&aff->af_suff);
|
|
|
3584
|
|
|
3585 /*
|
|
|
3586 * Read all the lines in the file one by one.
|
|
|
3587 */
|
|
255
|
3588 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
3589 {
|
|
255
|
3590 line_breakcheck();
|
|
236
|
3591 ++lnum;
|
|
|
3592
|
|
|
3593 /* Skip comment lines. */
|
|
|
3594 if (*rline == '#')
|
|
|
3595 continue;
|
|
|
3596
|
|
|
3597 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
3598 vim_free(pc);
|
|
310
|
3599 #ifdef FEAT_MBYTE
|
|
300
|
3600 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3601 {
|
|
300
|
3602 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
3603 if (pc == NULL)
|
|
|
3604 {
|
|
|
3605 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3606 fname, lnum, rline);
|
|
|
3607 continue;
|
|
|
3608 }
|
|
236
|
3609 line = pc;
|
|
|
3610 }
|
|
|
3611 else
|
|
310
|
3612 #endif
|
|
236
|
3613 {
|
|
|
3614 pc = NULL;
|
|
|
3615 line = rline;
|
|
|
3616 }
|
|
|
3617
|
|
|
3618 /* Split the line up in white separated items. Put a NUL after each
|
|
|
3619 * item. */
|
|
|
3620 itemcnt = 0;
|
|
|
3621 for (p = line; ; )
|
|
|
3622 {
|
|
|
3623 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
3624 ++p;
|
|
|
3625 if (*p == NUL)
|
|
|
3626 break;
|
|
334
|
3627 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
3628 break;
|
|
236
|
3629 items[itemcnt++] = p;
|
|
300
|
3630 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
3631 ++p;
|
|
|
3632 if (*p == NUL)
|
|
|
3633 break;
|
|
|
3634 *p++ = NUL;
|
|
|
3635 }
|
|
|
3636
|
|
|
3637 /* Handle non-empty lines. */
|
|
|
3638 if (itemcnt > 0)
|
|
|
3639 {
|
|
|
3640 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
3641 && aff->af_enc == NULL)
|
|
|
3642 {
|
|
310
|
3643 #ifdef FEAT_MBYTE
|
|
300
|
3644 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
3645 aff->af_enc = enc_canonize(items[1]);
|
|
|
3646 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
3647 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
3648 p_enc) == FAIL)
|
|
|
3649 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
3650 fname, aff->af_enc, p_enc);
|
|
375
|
3651 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3652 #else
|
|
|
3653 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3654 #endif
|
|
236
|
3655 }
|
|
366
|
3656 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
3657 {
|
|
|
3658 if (do_midword)
|
|
|
3659 spin->si_midword = vim_strsave(items[1]);
|
|
|
3660 }
|
|
302
|
3661 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
3662 {
|
|
323
|
3663 /* ignored, we always split */
|
|
302
|
3664 }
|
|
323
|
3665 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
3666 {
|
|
323
|
3667 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
3668 }
|
|
481
|
3669 else if (STRCMP(items[0], "SLASH") == 0 && itemcnt == 2
|
|
|
3670 && aff->af_slash == 0)
|
|
|
3671 {
|
|
|
3672 aff->af_slash = items[1][0];
|
|
|
3673 if (items[1][1] != NUL)
|
|
|
3674 smsg((char_u *)_("Character used for SLASH must be ASCII; in %s line %d: %s"),
|
|
|
3675 fname, lnum, items[1]);
|
|
|
3676 }
|
|
307
|
3677 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
3678 && aff->af_rar == 0)
|
|
|
3679 {
|
|
|
3680 aff->af_rar = items[1][0];
|
|
|
3681 if (items[1][1] != NUL)
|
|
|
3682 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3683 }
|
|
310
|
3684 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
3685 && aff->af_kep == 0)
|
|
307
|
3686 {
|
|
310
|
3687 aff->af_kep = items[1][0];
|
|
307
|
3688 if (items[1][1] != NUL)
|
|
|
3689 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3690 }
|
|
346
|
3691 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
3692 && aff->af_bad == 0)
|
|
|
3693 {
|
|
|
3694 aff->af_bad = items[1][0];
|
|
|
3695 if (items[1][1] != NUL)
|
|
|
3696 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3697 }
|
|
481
|
3698 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
|
3699 && aff->af_compflags == 0)
|
|
|
3700 {
|
|
|
3701 aff->af_compflags = getroom_save(spin, items[1]);
|
|
|
3702 if (items[1][1] != NUL)
|
|
|
3703 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3704 }
|
|
|
3705 else if (STRCMP(items[0], "COMPOUNDFLAGS") == 0 && itemcnt == 2
|
|
|
3706 && aff->af_compflags == 0)
|
|
|
3707 {
|
|
|
3708 aff->af_compflags = getroom_save(spin, items[1]);
|
|
|
3709 }
|
|
|
3710 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
|
3711 && aff->af_compminlen == 0)
|
|
|
3712 {
|
|
|
3713 aff->af_compminlen = atoi((char *)items[1]);
|
|
|
3714 if (aff->af_compminlen == 0)
|
|
|
3715 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
3716 fname, lnum, items[1]);
|
|
|
3717 }
|
|
339
|
3718 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
3719 {
|
|
|
3720 aff->af_pfxpostpone = TRUE;
|
|
|
3721 }
|
|
236
|
3722 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3723 || STRCMP(items[0], "SFX") == 0)
|
|
|
3724 && aff_todo == 0
|
|
334
|
3725 && itemcnt >= 4)
|
|
236
|
3726 {
|
|
334
|
3727 /* Myspell allows extra text after the item, but that might
|
|
|
3728 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
3729 if (itemcnt > 4 && *items[4] != '#')
|
|
|
3730 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
3731 fname, lnum, items[4]);
|
|
|
3732
|
|
236
|
3733 /* New affix letter. */
|
|
474
|
3734 cur_aff = (affheader_T *)getroom(spin,
|
|
447
|
3735 sizeof(affheader_T), TRUE);
|
|
236
|
3736 if (cur_aff == NULL)
|
|
|
3737 break;
|
|
455
|
3738 #ifdef FEAT_MBYTE
|
|
|
3739 if (has_mbyte)
|
|
|
3740 {
|
|
474
|
3741 l = (*mb_ptr2len)(items[1]);
|
|
455
|
3742 if (l >= AH_KEY_LEN)
|
|
|
3743 l = 1; /* too long, must be an overlong sequence */
|
|
|
3744 else
|
|
|
3745 mch_memmove(cur_aff->ah_key, items[1], l);
|
|
|
3746 }
|
|
|
3747 else
|
|
|
3748 #endif
|
|
|
3749 {
|
|
|
3750 *cur_aff->ah_key = *items[1];
|
|
|
3751 l = 1;
|
|
|
3752 }
|
|
|
3753 cur_aff->ah_key[l] = NUL;
|
|
|
3754 if (items[1][l] != NUL)
|
|
307
|
3755 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
3756 if (*items[2] == 'Y')
|
|
|
3757 cur_aff->ah_combine = TRUE;
|
|
300
|
3758 else if (*items[2] != 'N')
|
|
236
|
3759 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
3760 fname, lnum, items[2]);
|
|
339
|
3761
|
|
236
|
3762 if (*items[0] == 'P')
|
|
339
|
3763 {
|
|
236
|
3764 tp = &aff->af_pref;
|
|
339
|
3765 /* Use a new number in the .spl file later, to be able to
|
|
|
3766 * handle multiple .aff files. */
|
|
|
3767 if (aff->af_pfxpostpone)
|
|
344
|
3768 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
3769 }
|
|
236
|
3770 else
|
|
|
3771 tp = &aff->af_suff;
|
|
300
|
3772 aff_todo = atoi((char *)items[3]);
|
|
324
|
3773 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
3774 if (!HASHITEM_EMPTY(hi))
|
|
300
|
3775 {
|
|
236
|
3776 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
3777 fname, lnum, items[1]);
|
|
300
|
3778 aff_todo = 0;
|
|
|
3779 }
|
|
236
|
3780 else
|
|
|
3781 hash_add(tp, cur_aff->ah_key);
|
|
|
3782 }
|
|
|
3783 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3784 || STRCMP(items[0], "SFX") == 0)
|
|
|
3785 && aff_todo > 0
|
|
|
3786 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
3787 && itemcnt >= 5)
|
|
236
|
3788 {
|
|
|
3789 affentry_T *aff_entry;
|
|
366
|
3790 int rare = FALSE;
|
|
455
|
3791 int upper = FALSE;
|
|
366
|
3792 int lasti = 5;
|
|
|
3793
|
|
|
3794 /* Check for "rare" after the other info. */
|
|
|
3795 if (itemcnt > 5 && STRICMP(items[5], "rare") == 0)
|
|
|
3796 {
|
|
|
3797 rare = TRUE;
|
|
|
3798 lasti = 6;
|
|
|
3799 }
|
|
236
|
3800
|
|
334
|
3801 /* Myspell allows extra text after the item, but that might
|
|
|
3802 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
3803 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
3804 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
366
|
3805 fname, lnum, items[lasti]);
|
|
334
|
3806
|
|
236
|
3807 /* New item for an affix letter. */
|
|
|
3808 --aff_todo;
|
|
474
|
3809 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
3810 sizeof(affentry_T), TRUE);
|
|
236
|
3811 if (aff_entry == NULL)
|
|
|
3812 break;
|
|
366
|
3813 aff_entry->ae_rare = rare;
|
|
240
|
3814
|
|
236
|
3815 if (STRCMP(items[2], "0") != 0)
|
|
474
|
3816 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
3817 if (STRCMP(items[3], "0") != 0)
|
|
474
|
3818 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
3819
|
|
300
|
3820 /* Don't use an affix entry with non-ASCII characters when
|
|
|
3821 * "spin->si_ascii" is TRUE. */
|
|
|
3822 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
3823 || has_non_ascii(aff_entry->ae_add)))
|
|
|
3824 {
|
|
|
3825 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
3826 cur_aff->ah_first = aff_entry;
|
|
300
|
3827
|
|
|
3828 if (STRCMP(items[4], ".") != 0)
|
|
|
3829 {
|
|
|
3830 char_u buf[MAXLINELEN];
|
|
|
3831
|
|
474
|
3832 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
3833 if (*items[0] == 'P')
|
|
|
3834 sprintf((char *)buf, "^%s", items[4]);
|
|
|
3835 else
|
|
|
3836 sprintf((char *)buf, "%s$", items[4]);
|
|
|
3837 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
3838 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
3839 if (aff_entry->ae_prog == NULL)
|
|
|
3840 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
3841 fname, lnum, items[4]);
|
|
300
|
3842 }
|
|
339
|
3843
|
|
|
3844 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
3845 * for the condition. Use an existing one if possible. */
|
|
455
|
3846 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
3847 {
|
|
455
|
3848 /* When the chop string is one lower-case letter and
|
|
|
3849 * the add string ends in the upper-case letter we set
|
|
|
3850 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
3851 * letters from "ae_add". The condition must either
|
|
|
3852 * be empty or start with the same letter. */
|
|
|
3853 if (aff_entry->ae_chop != NULL
|
|
|
3854 && aff_entry->ae_add != NULL
|
|
|
3855 #ifdef FEAT_MBYTE
|
|
474
|
3856 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
3857 aff_entry->ae_chop)] == NUL
|
|
|
3858 #else
|
|
|
3859 && aff_entry->ae_chop[1] == NUL
|
|
|
3860 #endif
|
|
|
3861 )
|
|
339
|
3862 {
|
|
455
|
3863 int c, c_up;
|
|
|
3864
|
|
|
3865 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
3866 c_up = SPELL_TOUPPER(c);
|
|
|
3867 if (c_up != c
|
|
|
3868 && (aff_entry->ae_cond == NULL
|
|
|
3869 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
3870 {
|
|
|
3871 p = aff_entry->ae_add
|
|
|
3872 + STRLEN(aff_entry->ae_add);
|
|
|
3873 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
3874 if (PTR2CHAR(p) == c_up)
|
|
|
3875 {
|
|
|
3876 upper = TRUE;
|
|
|
3877 aff_entry->ae_chop = NULL;
|
|
|
3878 *p = NUL;
|
|
|
3879
|
|
|
3880 /* The condition is matched with the
|
|
|
3881 * actual word, thus must check for the
|
|
|
3882 * upper-case letter. */
|
|
|
3883 if (aff_entry->ae_cond != NULL)
|
|
|
3884 {
|
|
|
3885 char_u buf[MAXLINELEN];
|
|
|
3886 #ifdef FEAT_MBYTE
|
|
|
3887 if (has_mbyte)
|
|
|
3888 {
|
|
|
3889 onecap_copy(items[4], buf, TRUE);
|
|
|
3890 aff_entry->ae_cond = getroom_save(
|
|
474
|
3891 spin, buf);
|
|
455
|
3892 }
|
|
|
3893 else
|
|
|
3894 #endif
|
|
|
3895 *aff_entry->ae_cond = c_up;
|
|
|
3896 if (aff_entry->ae_cond != NULL)
|
|
|
3897 {
|
|
|
3898 sprintf((char *)buf, "^%s",
|
|
339
|
3899 aff_entry->ae_cond);
|
|
455
|
3900 vim_free(aff_entry->ae_prog);
|
|
|
3901 aff_entry->ae_prog = vim_regcomp(
|
|
|
3902 buf, RE_MAGIC + RE_STRING);
|
|
|
3903 }
|
|
|
3904 }
|
|
|
3905 }
|
|
|
3906 }
|
|
339
|
3907 }
|
|
|
3908
|
|
455
|
3909 if (aff_entry->ae_chop == NULL)
|
|
390
|
3910 {
|
|
455
|
3911 int idx;
|
|
|
3912 char_u **pp;
|
|
|
3913 int n;
|
|
|
3914
|
|
|
3915 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
3916 --idx)
|
|
|
3917 {
|
|
|
3918 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
3919 if (str_equal(p, aff_entry->ae_cond))
|
|
|
3920 break;
|
|
|
3921 }
|
|
|
3922 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
3923 {
|
|
|
3924 /* Not found, add a new condition. */
|
|
|
3925 idx = spin->si_prefcond.ga_len++;
|
|
|
3926 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
3927 + idx;
|
|
|
3928 if (aff_entry->ae_cond == NULL)
|
|
|
3929 *pp = NULL;
|
|
|
3930 else
|
|
474
|
3931 *pp = getroom_save(spin,
|
|
455
|
3932 aff_entry->ae_cond);
|
|
|
3933 }
|
|
|
3934
|
|
|
3935 /* Add the prefix to the prefix tree. */
|
|
|
3936 if (aff_entry->ae_add == NULL)
|
|
|
3937 p = (char_u *)"";
|
|
390
|
3938 else
|
|
455
|
3939 p = aff_entry->ae_add;
|
|
|
3940 /* PFX_FLAGS is a negative number, so that
|
|
|
3941 * tree_add_word() knows this is the prefix tree. */
|
|
|
3942 n = PFX_FLAGS;
|
|
|
3943 if (rare)
|
|
|
3944 n |= WFP_RARE;
|
|
|
3945 if (!cur_aff->ah_combine)
|
|
|
3946 n |= WFP_NC;
|
|
|
3947 if (upper)
|
|
|
3948 n |= WFP_UP;
|
|
474
|
3949 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
3950 idx, cur_aff->ah_newID);
|
|
390
|
3951 }
|
|
339
|
3952 }
|
|
240
|
3953 }
|
|
236
|
3954 }
|
|
255
|
3955 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
3956 {
|
|
|
3957 if (fol != NULL)
|
|
|
3958 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
3959 fname, lnum);
|
|
|
3960 else
|
|
|
3961 fol = vim_strsave(items[1]);
|
|
|
3962 }
|
|
|
3963 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
3964 {
|
|
|
3965 if (low != NULL)
|
|
|
3966 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
3967 fname, lnum);
|
|
|
3968 else
|
|
|
3969 low = vim_strsave(items[1]);
|
|
|
3970 }
|
|
|
3971 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
3972 {
|
|
|
3973 if (upp != NULL)
|
|
|
3974 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
3975 fname, lnum);
|
|
|
3976 else
|
|
|
3977 upp = vim_strsave(items[1]);
|
|
|
3978 }
|
|
236
|
3979 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
3980 {
|
|
236
|
3981 /* Ignore REP count */;
|
|
323
|
3982 if (!isdigit(*items[1]))
|
|
|
3983 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
3984 fname, lnum);
|
|
|
3985 }
|
|
236
|
3986 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 3)
|
|
|
3987 {
|
|
|
3988 /* REP item */
|
|
323
|
3989 if (do_rep)
|
|
|
3990 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
3991 }
|
|
|
3992 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
3993 {
|
|
|
3994 /* MAP item or count */
|
|
|
3995 if (!found_map)
|
|
|
3996 {
|
|
|
3997 /* First line contains the count. */
|
|
|
3998 found_map = TRUE;
|
|
|
3999 if (!isdigit(*items[1]))
|
|
|
4000 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
4001 fname, lnum);
|
|
|
4002 }
|
|
|
4003 else if (do_map)
|
|
|
4004 {
|
|
346
|
4005 int c;
|
|
|
4006
|
|
|
4007 /* Check that every character appears only once. */
|
|
|
4008 for (p = items[1]; *p != NUL; )
|
|
|
4009 {
|
|
|
4010 #ifdef FEAT_MBYTE
|
|
|
4011 c = mb_ptr2char_adv(&p);
|
|
|
4012 #else
|
|
|
4013 c = *p++;
|
|
|
4014 #endif
|
|
|
4015 if ((spin->si_map.ga_len > 0
|
|
|
4016 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
4017 != NULL)
|
|
|
4018 || vim_strchr(p, c) != NULL)
|
|
|
4019 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
4020 fname, lnum);
|
|
|
4021 }
|
|
|
4022
|
|
323
|
4023 /* We simply concatenate all the MAP strings, separated by
|
|
|
4024 * slashes. */
|
|
|
4025 ga_concat(&spin->si_map, items[1]);
|
|
|
4026 ga_append(&spin->si_map, '/');
|
|
|
4027 }
|
|
|
4028 }
|
|
|
4029 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
4030 {
|
|
|
4031 if (do_sal)
|
|
|
4032 {
|
|
|
4033 /* SAL item (sounds-a-like)
|
|
|
4034 * Either one of the known keys or a from-to pair. */
|
|
|
4035 if (STRCMP(items[1], "followup") == 0)
|
|
|
4036 spin->si_followup = sal_to_bool(items[2]);
|
|
|
4037 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
4038 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
4039 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
4040 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
4041 else
|
|
|
4042 /* when "to" is "_" it means empty */
|
|
|
4043 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
4044 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
4045 : items[2]);
|
|
|
4046 }
|
|
236
|
4047 }
|
|
375
|
4048 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
4049 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
4050 {
|
|
|
4051 if (do_sofo)
|
|
|
4052 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
4053 }
|
|
|
4054 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
4055 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
4056 {
|
|
|
4057 if (do_sofo)
|
|
|
4058 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
4059 }
|
|
300
|
4060 else
|
|
481
|
4061 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
4062 fname, lnum, items[0]);
|
|
|
4063 }
|
|
|
4064 }
|
|
|
4065
|
|
375
|
4066 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
4067 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
4068 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
4069 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
4070 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
4071
|
|
255
|
4072 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
4073 {
|
|
351
|
4074 if (spin->si_clear_chartab)
|
|
|
4075 {
|
|
|
4076 /* Clear the char type tables, don't want to use any of the
|
|
|
4077 * currently used spell properties. */
|
|
|
4078 init_spell_chartab();
|
|
|
4079 spin->si_clear_chartab = FALSE;
|
|
|
4080 }
|
|
|
4081
|
|
316
|
4082 /*
|
|
|
4083 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
4084 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
4085 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
4086 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
4087 */
|
|
|
4088 if (!spin->si_ascii
|
|
|
4089 #ifdef FEAT_MBYTE
|
|
|
4090 && !enc_utf8
|
|
|
4091 #endif
|
|
|
4092 )
|
|
260
|
4093 {
|
|
|
4094 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
4095 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
4096 else
|
|
316
|
4097 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
4098 }
|
|
255
|
4099
|
|
|
4100 vim_free(fol);
|
|
|
4101 vim_free(low);
|
|
|
4102 vim_free(upp);
|
|
|
4103 }
|
|
|
4104
|
|
481
|
4105 /* Use compound specifications of the .aff file for the spell info. */
|
|
|
4106 if (aff->af_compminlen != 0)
|
|
|
4107 {
|
|
|
4108 if (spin->si_compminlen != 0
|
|
|
4109 && spin->si_compminlen != aff->af_compminlen)
|
|
|
4110 smsg((char_u *)_("COMPOUNDMIN value differs from what is used in another .aff file"));
|
|
|
4111 else
|
|
|
4112 spin->si_compminlen = aff->af_compminlen;
|
|
|
4113 }
|
|
|
4114
|
|
|
4115 if (aff->af_compflags != NULL)
|
|
|
4116 {
|
|
|
4117 if (spin->si_compflags != NULL
|
|
|
4118 && STRCMP(spin->si_compflags, aff->af_compflags) != 0)
|
|
|
4119 smsg((char_u *)_("COMPOUNDFLAG(S) value differs from what is used in another .aff file"));
|
|
|
4120 else
|
|
|
4121 spin->si_compflags = aff->af_compflags;
|
|
|
4122
|
|
|
4123 if (aff->af_pfxpostpone)
|
|
|
4124 smsg((char_u *)_("Cannot use both PFXPOSTPONE and COMPOUNDFLAG(S)"));
|
|
|
4125 }
|
|
|
4126
|
|
236
|
4127 vim_free(pc);
|
|
|
4128 fclose(fd);
|
|
|
4129 return aff;
|
|
|
4130 }
|
|
|
4131
|
|
|
4132 /*
|
|
339
|
4133 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
4134 * NULL as equal.
|
|
|
4135 */
|
|
|
4136 static int
|
|
|
4137 str_equal(s1, s2)
|
|
|
4138 char_u *s1;
|
|
|
4139 char_u *s2;
|
|
|
4140 {
|
|
|
4141 if (s1 == NULL || s2 == NULL)
|
|
|
4142 return s1 == s2;
|
|
|
4143 return STRCMP(s1, s2) == 0;
|
|
|
4144 }
|
|
|
4145
|
|
|
4146 /*
|
|
323
|
4147 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
4148 * They are stored case-folded.
|
|
|
4149 */
|
|
|
4150 static void
|
|
|
4151 add_fromto(spin, gap, from, to)
|
|
|
4152 spellinfo_T *spin;
|
|
|
4153 garray_T *gap;
|
|
|
4154 char_u *from;
|
|
|
4155 char_u *to;
|
|
|
4156 {
|
|
|
4157 fromto_T *ftp;
|
|
|
4158 char_u word[MAXWLEN];
|
|
|
4159
|
|
|
4160 if (ga_grow(gap, 1) == OK)
|
|
|
4161 {
|
|
|
4162 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
4163 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
4164 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
4165 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
4166 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
4167 ++gap->ga_len;
|
|
|
4168 }
|
|
|
4169 }
|
|
|
4170
|
|
|
4171 /*
|
|
|
4172 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
4173 */
|
|
|
4174 static int
|
|
|
4175 sal_to_bool(s)
|
|
|
4176 char_u *s;
|
|
|
4177 {
|
|
|
4178 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
4179 }
|
|
|
4180
|
|
|
4181 /*
|
|
240
|
4182 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
4183 * When "s" is NULL FALSE is returned.
|
|
|
4184 */
|
|
|
4185 static int
|
|
|
4186 has_non_ascii(s)
|
|
|
4187 char_u *s;
|
|
|
4188 {
|
|
|
4189 char_u *p;
|
|
|
4190
|
|
|
4191 if (s != NULL)
|
|
|
4192 for (p = s; *p != NUL; ++p)
|
|
|
4193 if (*p >= 128)
|
|
|
4194 return TRUE;
|
|
|
4195 return FALSE;
|
|
|
4196 }
|
|
|
4197
|
|
|
4198 /*
|
|
236
|
4199 * Free the structure filled by spell_read_aff().
|
|
|
4200 */
|
|
|
4201 static void
|
|
|
4202 spell_free_aff(aff)
|
|
|
4203 afffile_T *aff;
|
|
|
4204 {
|
|
|
4205 hashtab_T *ht;
|
|
|
4206 hashitem_T *hi;
|
|
|
4207 int todo;
|
|
|
4208 affheader_T *ah;
|
|
300
|
4209 affentry_T *ae;
|
|
236
|
4210
|
|
|
4211 vim_free(aff->af_enc);
|
|
|
4212
|
|
339
|
4213 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
4214 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
4215 {
|
|
|
4216 todo = ht->ht_used;
|
|
|
4217 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
4218 {
|
|
|
4219 if (!HASHITEM_EMPTY(hi))
|
|
|
4220 {
|
|
|
4221 --todo;
|
|
|
4222 ah = HI2AH(hi);
|
|
300
|
4223 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
4224 vim_free(ae->ae_prog);
|
|
236
|
4225 }
|
|
|
4226 }
|
|
|
4227 if (ht == &aff->af_suff)
|
|
|
4228 break;
|
|
|
4229 }
|
|
300
|
4230
|
|
236
|
4231 hash_clear(&aff->af_pref);
|
|
|
4232 hash_clear(&aff->af_suff);
|
|
|
4233 }
|
|
|
4234
|
|
|
4235 /*
|
|
300
|
4236 * Read dictionary file "fname".
|
|
236
|
4237 * Returns OK or FAIL;
|
|
|
4238 */
|
|
|
4239 static int
|
|
474
|
4240 spell_read_dic(spin, fname, affile)
|
|
|
4241 spellinfo_T *spin;
|
|
236
|
4242 char_u *fname;
|
|
300
|
4243 afffile_T *affile;
|
|
236
|
4244 {
|
|
300
|
4245 hashtab_T ht;
|
|
236
|
4246 char_u line[MAXLINELEN];
|
|
481
|
4247 char_u *p;
|
|
300
|
4248 char_u *afflist;
|
|
481
|
4249 char_u *store_afflist;
|
|
300
|
4250 char_u *dw;
|
|
236
|
4251 char_u *pc;
|
|
|
4252 char_u *w;
|
|
|
4253 int l;
|
|
|
4254 hash_T hash;
|
|
|
4255 hashitem_T *hi;
|
|
|
4256 FILE *fd;
|
|
|
4257 int lnum = 1;
|
|
300
|
4258 int non_ascii = 0;
|
|
|
4259 int retval = OK;
|
|
|
4260 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
4261 int flags;
|
|
481
|
4262 int duplicate = 0;
|
|
236
|
4263
|
|
300
|
4264 /*
|
|
|
4265 * Open the file.
|
|
|
4266 */
|
|
310
|
4267 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4268 if (fd == NULL)
|
|
|
4269 {
|
|
|
4270 EMSG2(_(e_notopen), fname);
|
|
|
4271 return FAIL;
|
|
|
4272 }
|
|
|
4273
|
|
300
|
4274 /* The hashtable is only used to detect duplicated words. */
|
|
|
4275 hash_init(&ht);
|
|
|
4276
|
|
310
|
4277 if (spin->si_verbose || p_verbose > 2)
|
|
|
4278 {
|
|
|
4279 if (!spin->si_verbose)
|
|
|
4280 verbose_enter();
|
|
366
|
4281 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
4282 out_flush();
|
|
|
4283 if (!spin->si_verbose)
|
|
|
4284 verbose_leave();
|
|
|
4285 }
|
|
236
|
4286
|
|
474
|
4287 /* start with a message for the first line */
|
|
|
4288 spin->si_msg_count = 999999;
|
|
|
4289
|
|
236
|
4290 /* Read and ignore the first line: word count. */
|
|
|
4291 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
4292 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
4293 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
4294
|
|
|
4295 /*
|
|
|
4296 * Read all the lines in the file one by one.
|
|
|
4297 * The words are converted to 'encoding' here, before being added to
|
|
|
4298 * the hashtable.
|
|
|
4299 */
|
|
255
|
4300 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
4301 {
|
|
255
|
4302 line_breakcheck();
|
|
236
|
4303 ++lnum;
|
|
455
|
4304 if (line[0] == '#' || line[0] == '/')
|
|
351
|
4305 continue; /* comment line */
|
|
236
|
4306
|
|
300
|
4307 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
4308 * the word is kept to allow e.g., "et al.". */
|
|
236
|
4309 l = STRLEN(line);
|
|
|
4310 while (l > 0 && line[l - 1] <= ' ')
|
|
|
4311 --l;
|
|
|
4312 if (l == 0)
|
|
|
4313 continue; /* empty line */
|
|
|
4314 line[l] = NUL;
|
|
|
4315
|
|
481
|
4316 /* Find the optional affix names. Replace the SLASH character by a
|
|
|
4317 * slash. */
|
|
|
4318 afflist = NULL;
|
|
|
4319 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
4320 {
|
|
|
4321 if (*p == affile->af_slash)
|
|
|
4322 *p = '/';
|
|
|
4323 else if (*p == '/')
|
|
|
4324 {
|
|
|
4325 *p = NUL;
|
|
|
4326 afflist = p + 1;
|
|
|
4327 break;
|
|
|
4328 }
|
|
|
4329 }
|
|
236
|
4330
|
|
300
|
4331 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4332 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4333 {
|
|
|
4334 ++non_ascii;
|
|
240
|
4335 continue;
|
|
300
|
4336 }
|
|
240
|
4337
|
|
310
|
4338 #ifdef FEAT_MBYTE
|
|
236
|
4339 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
4340 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4341 {
|
|
300
|
4342 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
4343 if (pc == NULL)
|
|
|
4344 {
|
|
|
4345 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4346 fname, lnum, line);
|
|
|
4347 continue;
|
|
|
4348 }
|
|
236
|
4349 w = pc;
|
|
|
4350 }
|
|
|
4351 else
|
|
310
|
4352 #endif
|
|
236
|
4353 {
|
|
|
4354 pc = NULL;
|
|
|
4355 w = line;
|
|
|
4356 }
|
|
|
4357
|
|
474
|
4358 /* This takes time, print a message every 10000 words. */
|
|
|
4359 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
4360 {
|
|
|
4361 spin->si_msg_count = 0;
|
|
339
|
4362 vim_snprintf((char *)message, sizeof(message),
|
|
|
4363 _("line %6d, word %6d - %s"),
|
|
|
4364 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
4365 msg_start();
|
|
|
4366 msg_puts_long_attr(message, 0);
|
|
|
4367 msg_clr_eos();
|
|
|
4368 msg_didout = FALSE;
|
|
|
4369 msg_col = 0;
|
|
|
4370 out_flush();
|
|
|
4371 }
|
|
|
4372
|
|
300
|
4373 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
4374 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
4375 if (dw == NULL)
|
|
300
|
4376 retval = FAIL;
|
|
|
4377 vim_free(pc);
|
|
|
4378 if (retval == FAIL)
|
|
236
|
4379 break;
|
|
|
4380
|
|
300
|
4381 hash = hash_hash(dw);
|
|
|
4382 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
4383 if (!HASHITEM_EMPTY(hi))
|
|
481
|
4384 {
|
|
|
4385 if (p_verbose > 0)
|
|
|
4386 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
4387 fname, lnum, dw);
|
|
481
|
4388 else if (duplicate == 0)
|
|
|
4389 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
4390 fname, lnum, dw);
|
|
|
4391 ++duplicate;
|
|
|
4392 }
|
|
236
|
4393 else
|
|
300
|
4394 hash_add_item(&ht, hi, dw, hash);
|
|
|
4395
|
|
307
|
4396 flags = 0;
|
|
481
|
4397 store_afflist = NULL;
|
|
307
|
4398 if (afflist != NULL)
|
|
|
4399 {
|
|
|
4400 /* Check for affix name that stands for keep-case word and stands
|
|
|
4401 * for rare word (if defined). */
|
|
310
|
4402 if (affile->af_kep != NUL
|
|
|
4403 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
389
|
4404 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
307
|
4405 if (affile->af_rar != NUL
|
|
|
4406 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
4407 flags |= WF_RARE;
|
|
346
|
4408 if (affile->af_bad != NUL
|
|
|
4409 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
4410 flags |= WF_BANNED;
|
|
339
|
4411
|
|
|
4412 if (affile->af_pfxpostpone)
|
|
|
4413 /* Need to store the list of prefix IDs with the word. */
|
|
481
|
4414 store_afflist = get_pfxlist(spin, affile, afflist);
|
|
|
4415 else if (spin->si_compflags)
|
|
|
4416 /* Need to store the list of affix IDs for compounding with
|
|
|
4417 * the word. */
|
|
|
4418 store_afflist = get_compflags(spin, afflist);
|
|
307
|
4419 }
|
|
|
4420
|
|
300
|
4421 /* Add the word to the word tree(s). */
|
|
481
|
4422 if (store_word(spin, dw, flags, spin->si_region, store_afflist) == FAIL)
|
|
300
|
4423 retval = FAIL;
|
|
236
|
4424
|
|
300
|
4425 if (afflist != NULL)
|
|
|
4426 {
|
|
|
4427 /* Find all matching suffixes and add the resulting words.
|
|
|
4428 * Additionally do matching prefixes that combine. */
|
|
474
|
4429 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
4430 &affile->af_suff, &affile->af_pref,
|
|
481
|
4431 FALSE, flags, store_afflist) == FAIL)
|
|
300
|
4432 retval = FAIL;
|
|
|
4433
|
|
|
4434 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
4435 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
4436 &affile->af_pref, NULL,
|
|
481
|
4437 FALSE, flags, store_afflist) == FAIL)
|
|
300
|
4438 retval = FAIL;
|
|
|
4439 }
|
|
236
|
4440 }
|
|
|
4441
|
|
481
|
4442 if (duplicate > 0)
|
|
|
4443 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
4444 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
4445 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
4446 non_ascii, fname);
|
|
300
|
4447 hash_clear(&ht);
|
|
|
4448
|
|
236
|
4449 fclose(fd);
|
|
300
|
4450 return retval;
|
|
236
|
4451 }
|
|
|
4452
|
|
|
4453 /*
|
|
339
|
4454 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
4455 * Used for PFXPOSTPONE.
|
|
|
4456 * Returns a string allocated with getroom(). NULL when there are no prefixes
|
|
|
4457 * or when out of memory.
|
|
|
4458 */
|
|
|
4459 static char_u *
|
|
474
|
4460 get_pfxlist(spin, affile, afflist)
|
|
|
4461 spellinfo_T *spin;
|
|
339
|
4462 afffile_T *affile;
|
|
|
4463 char_u *afflist;
|
|
|
4464 {
|
|
|
4465 char_u *p;
|
|
|
4466 int cnt;
|
|
|
4467 int round;
|
|
|
4468 char_u *res = NULL;
|
|
|
4469 char_u key[2];
|
|
|
4470 hashitem_T *hi;
|
|
|
4471
|
|
|
4472 key[1] = NUL;
|
|
|
4473
|
|
|
4474 /* round 1: count the number of prefix IDs.
|
|
|
4475 * round 2: move prefix IDs to "res" */
|
|
|
4476 for (round = 1; round <= 2; ++round)
|
|
|
4477 {
|
|
|
4478 cnt = 0;
|
|
|
4479 for (p = afflist; *p != NUL; ++p)
|
|
|
4480 {
|
|
|
4481 key[0] = *p;
|
|
|
4482 hi = hash_find(&affile->af_pref, key);
|
|
|
4483 if (!HASHITEM_EMPTY(hi))
|
|
|
4484 {
|
|
|
4485 /* This is a prefix ID, use the new number. */
|
|
|
4486 if (round == 2)
|
|
|
4487 res[cnt] = HI2AH(hi)->ah_newID;
|
|
|
4488 ++cnt;
|
|
|
4489 }
|
|
|
4490 }
|
|
|
4491 if (round == 1 && cnt > 0)
|
|
474
|
4492 res = getroom(spin, cnt + 1, FALSE);
|
|
339
|
4493 if (res == NULL)
|
|
|
4494 break;
|
|
|
4495 }
|
|
|
4496
|
|
|
4497 if (res != NULL)
|
|
|
4498 res[cnt] = NUL;
|
|
|
4499 return res;
|
|
|
4500 }
|
|
|
4501
|
|
|
4502 /*
|
|
481
|
4503 * Get the list of affix IDs from the affix list "afflist" that are used for
|
|
|
4504 * compound words.
|
|
|
4505 * Returns a string allocated with getroom(). NULL when there are no relevant
|
|
|
4506 * affixes or when out of memory.
|
|
|
4507 */
|
|
|
4508 static char_u *
|
|
|
4509 get_compflags(spin, afflist)
|
|
|
4510 spellinfo_T *spin;
|
|
|
4511 char_u *afflist;
|
|
|
4512 {
|
|
|
4513 char_u *p;
|
|
|
4514 int cnt;
|
|
|
4515 int round;
|
|
|
4516 char_u *res = NULL;
|
|
|
4517
|
|
|
4518 /* round 1: count the number of affix IDs.
|
|
|
4519 * round 2: move affix IDs to "res" */
|
|
|
4520 for (round = 1; round <= 2; ++round)
|
|
|
4521 {
|
|
|
4522 cnt = 0;
|
|
|
4523 for (p = afflist; *p != NUL; ++p)
|
|
|
4524 {
|
|
|
4525 if (*p != ',' && *p != '-'
|
|
|
4526 && vim_strchr(spin->si_compflags, *p) != NULL)
|
|
|
4527 {
|
|
|
4528 /* This is a compount affix ID. */
|
|
|
4529 if (round == 2)
|
|
|
4530 res[cnt] = *p;
|
|
|
4531 ++cnt;
|
|
|
4532 }
|
|
|
4533 }
|
|
|
4534 if (round == 1 && cnt > 0)
|
|
|
4535 res = getroom(spin, cnt + 1, FALSE);
|
|
|
4536 if (res == NULL)
|
|
|
4537 break;
|
|
|
4538 }
|
|
|
4539
|
|
|
4540 if (res != NULL)
|
|
|
4541 res[cnt] = NUL;
|
|
|
4542 return res;
|
|
|
4543 }
|
|
|
4544
|
|
|
4545 /*
|
|
300
|
4546 * Apply affixes to a word and store the resulting words.
|
|
|
4547 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
4548 * prefixes or suffixes.
|
|
|
4549 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
4550 * the resulting words for combining affixes.
|
|
|
4551 *
|
|
|
4552 * Returns FAIL when out of memory.
|
|
236
|
4553 */
|
|
300
|
4554 static int
|
|
474
|
4555 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags, pfxlist)
|
|
|
4556 spellinfo_T *spin; /* spell info */
|
|
300
|
4557 char_u *word; /* basic word start */
|
|
|
4558 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
4559 afffile_T *affile;
|
|
300
|
4560 hashtab_T *ht;
|
|
|
4561 hashtab_T *xht;
|
|
|
4562 int comb; /* only use affixes that combine */
|
|
307
|
4563 int flags; /* flags for the word */
|
|
339
|
4564 char_u *pfxlist; /* list of prefix IDs */
|
|
236
|
4565 {
|
|
|
4566 int todo;
|
|
|
4567 hashitem_T *hi;
|
|
300
|
4568 affheader_T *ah;
|
|
|
4569 affentry_T *ae;
|
|
|
4570 regmatch_T regmatch;
|
|
|
4571 char_u newword[MAXWLEN];
|
|
|
4572 int retval = OK;
|
|
|
4573 int i;
|
|
|
4574 char_u *p;
|
|
366
|
4575 int use_flags;
|
|
390
|
4576 char_u *use_pfxlist;
|
|
455
|
4577 int c;
|
|
481
|
4578 int wordlen = STRLEN(word);
|
|
236
|
4579
|
|
300
|
4580 todo = ht->ht_used;
|
|
|
4581 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
4582 {
|
|
|
4583 if (!HASHITEM_EMPTY(hi))
|
|
|
4584 {
|
|
|
4585 --todo;
|
|
300
|
4586 ah = HI2AH(hi);
|
|
236
|
4587
|
|
300
|
4588 /* Check that the affix combines, if required, and that the word
|
|
|
4589 * supports this affix. */
|
|
455
|
4590 c = PTR2CHAR(ah->ah_key);
|
|
|
4591 if ((!comb || ah->ah_combine) && vim_strchr(afflist, c) != NULL)
|
|
236
|
4592 {
|
|
300
|
4593 /* Loop over all affix entries with this name. */
|
|
|
4594 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
4595 {
|
|
300
|
4596 /* Check the condition. It's not logical to match case
|
|
|
4597 * here, but it is required for compatibility with
|
|
339
|
4598 * Myspell.
|
|
481
|
4599 * Another requirement from Myspell is that the chop
|
|
|
4600 * string is shorter than the word itself.
|
|
339
|
4601 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
4602 * prefixes with a chop string. */
|
|
300
|
4603 regmatch.regprog = ae->ae_prog;
|
|
|
4604 regmatch.rm_ic = FALSE;
|
|
339
|
4605 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
4606 || ae->ae_chop != NULL)
|
|
481
|
4607 && (ae->ae_chop == NULL
|
|
|
4608 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
4609 && (ae->ae_prog == NULL
|
|
|
4610 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
4611 {
|
|
|
4612 /* Match. Remove the chop and add the affix. */
|
|
|
4613 if (xht == NULL)
|
|
240
|
4614 {
|
|
300
|
4615 /* prefix: chop/add at the start of the word */
|
|
|
4616 if (ae->ae_add == NULL)
|
|
|
4617 *newword = NUL;
|
|
|
4618 else
|
|
|
4619 STRCPY(newword, ae->ae_add);
|
|
|
4620 p = word;
|
|
|
4621 if (ae->ae_chop != NULL)
|
|
310
|
4622 {
|
|
300
|
4623 /* Skip chop string. */
|
|
310
|
4624 #ifdef FEAT_MBYTE
|
|
|
4625 if (has_mbyte)
|
|
324
|
4626 {
|
|
310
|
4627 i = mb_charlen(ae->ae_chop);
|
|
324
|
4628 for ( ; i > 0; --i)
|
|
|
4629 mb_ptr_adv(p);
|
|
|
4630 }
|
|
310
|
4631 else
|
|
|
4632 #endif
|
|
324
|
4633 p += STRLEN(ae->ae_chop);
|
|
310
|
4634 }
|
|
300
|
4635 STRCAT(newword, p);
|
|
|
4636 }
|
|
|
4637 else
|
|
|
4638 {
|
|
|
4639 /* suffix: chop/add at the end of the word */
|
|
|
4640 STRCPY(newword, word);
|
|
|
4641 if (ae->ae_chop != NULL)
|
|
|
4642 {
|
|
|
4643 /* Remove chop string. */
|
|
|
4644 p = newword + STRLEN(newword);
|
|
376
|
4645 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
4646 for ( ; i > 0; --i)
|
|
300
|
4647 mb_ptr_back(newword, p);
|
|
|
4648 *p = NUL;
|
|
|
4649 }
|
|
|
4650 if (ae->ae_add != NULL)
|
|
|
4651 STRCAT(newword, ae->ae_add);
|
|
240
|
4652 }
|
|
|
4653
|
|
366
|
4654 /* Obey the "rare" flag of the affix. */
|
|
|
4655 if (ae->ae_rare)
|
|
|
4656 use_flags = flags | WF_RARE;
|
|
|
4657 else
|
|
|
4658 use_flags = flags;
|
|
390
|
4659 use_pfxlist = pfxlist;
|
|
|
4660
|
|
|
4661 /* When there are postponed prefixes... */
|
|
393
|
4662 if (spin->si_prefroot != NULL
|
|
|
4663 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
4664 {
|
|
|
4665 /* ... add a flag to indicate an affix was used. */
|
|
|
4666 use_flags |= WF_HAS_AFF;
|
|
|
4667
|
|
|
4668 /* ... don't use a prefix list if combining
|
|
|
4669 * affixes is not allowed */
|
|
|
4670 if (!ah->ah_combine || comb)
|
|
|
4671 use_pfxlist = NULL;
|
|
|
4672 }
|
|
366
|
4673
|
|
300
|
4674 /* Store the modified word. */
|
|
474
|
4675 if (store_word(spin, newword, use_flags,
|
|
390
|
4676 spin->si_region, use_pfxlist) == FAIL)
|
|
300
|
4677 retval = FAIL;
|
|
236
|
4678
|
|
300
|
4679 /* When added a suffix and combining is allowed also
|
|
|
4680 * try adding prefixes additionally. */
|
|
|
4681 if (xht != NULL && ah->ah_combine)
|
|
474
|
4682 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
4683 xht, NULL, TRUE,
|
|
|
4684 use_flags, use_pfxlist) == FAIL)
|
|
300
|
4685 retval = FAIL;
|
|
236
|
4686 }
|
|
|
4687 }
|
|
|
4688 }
|
|
|
4689 }
|
|
|
4690 }
|
|
|
4691
|
|
|
4692 return retval;
|
|
|
4693 }
|
|
|
4694
|
|
|
4695 /*
|
|
300
|
4696 * Read a file with a list of words.
|
|
236
|
4697 */
|
|
|
4698 static int
|
|
474
|
4699 spell_read_wordfile(spin, fname)
|
|
|
4700 spellinfo_T *spin;
|
|
300
|
4701 char_u *fname;
|
|
236
|
4702 {
|
|
300
|
4703 FILE *fd;
|
|
|
4704 long lnum = 0;
|
|
|
4705 char_u rline[MAXLINELEN];
|
|
|
4706 char_u *line;
|
|
|
4707 char_u *pc = NULL;
|
|
381
|
4708 char_u *p;
|
|
300
|
4709 int l;
|
|
|
4710 int retval = OK;
|
|
|
4711 int did_word = FALSE;
|
|
|
4712 int non_ascii = 0;
|
|
307
|
4713 int flags;
|
|
316
|
4714 int regionmask;
|
|
236
|
4715
|
|
300
|
4716 /*
|
|
|
4717 * Open the file.
|
|
|
4718 */
|
|
310
|
4719 fd = mch_fopen((char *)fname, "r");
|
|
300
|
4720 if (fd == NULL)
|
|
236
|
4721 {
|
|
300
|
4722 EMSG2(_(e_notopen), fname);
|
|
|
4723 return FAIL;
|
|
236
|
4724 }
|
|
|
4725
|
|
310
|
4726 if (spin->si_verbose || p_verbose > 2)
|
|
|
4727 {
|
|
|
4728 if (!spin->si_verbose)
|
|
|
4729 verbose_enter();
|
|
366
|
4730 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
4731 out_flush();
|
|
|
4732 if (!spin->si_verbose)
|
|
|
4733 verbose_leave();
|
|
|
4734 }
|
|
300
|
4735
|
|
|
4736 /*
|
|
|
4737 * Read all the lines in the file one by one.
|
|
|
4738 */
|
|
|
4739 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
4740 {
|
|
|
4741 line_breakcheck();
|
|
|
4742 ++lnum;
|
|
|
4743
|
|
|
4744 /* Skip comment lines. */
|
|
|
4745 if (*rline == '#')
|
|
|
4746 continue;
|
|
|
4747
|
|
|
4748 /* Remove CR, LF and white space from the end. */
|
|
|
4749 l = STRLEN(rline);
|
|
|
4750 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
4751 --l;
|
|
|
4752 if (l == 0)
|
|
|
4753 continue; /* empty or blank line */
|
|
|
4754 rline[l] = NUL;
|
|
|
4755
|
|
|
4756 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
4757 vim_free(pc);
|
|
310
|
4758 #ifdef FEAT_MBYTE
|
|
300
|
4759 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
4760 {
|
|
|
4761 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
4762 if (pc == NULL)
|
|
|
4763 {
|
|
|
4764 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4765 fname, lnum, rline);
|
|
|
4766 continue;
|
|
|
4767 }
|
|
|
4768 line = pc;
|
|
|
4769 }
|
|
|
4770 else
|
|
310
|
4771 #endif
|
|
300
|
4772 {
|
|
|
4773 pc = NULL;
|
|
|
4774 line = rline;
|
|
|
4775 }
|
|
|
4776
|
|
307
|
4777 if (*line == '/')
|
|
300
|
4778 {
|
|
307
|
4779 ++line;
|
|
|
4780 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
4781 {
|
|
|
4782 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
4783 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
4784 fname, lnum, line - 1);
|
|
300
|
4785 else if (did_word)
|
|
316
|
4786 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
4787 fname, lnum, line - 1);
|
|
300
|
4788 else
|
|
|
4789 {
|
|
310
|
4790 #ifdef FEAT_MBYTE
|
|
|
4791 char_u *enc;
|
|
|
4792
|
|
300
|
4793 /* Setup for conversion to 'encoding'. */
|
|
316
|
4794 line += 10;
|
|
|
4795 enc = enc_canonize(line);
|
|
300
|
4796 if (enc != NULL && !spin->si_ascii
|
|
|
4797 && convert_setup(&spin->si_conv, enc,
|
|
|
4798 p_enc) == FAIL)
|
|
|
4799 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
4800 fname, line, p_enc);
|
|
300
|
4801 vim_free(enc);
|
|
375
|
4802 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4803 #else
|
|
|
4804 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4805 #endif
|
|
300
|
4806 }
|
|
307
|
4807 continue;
|
|
300
|
4808 }
|
|
307
|
4809
|
|
316
|
4810 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
4811 {
|
|
|
4812 if (spin->si_region_count > 1)
|
|
|
4813 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
4814 fname, lnum, line);
|
|
|
4815 else
|
|
|
4816 {
|
|
|
4817 line += 8;
|
|
|
4818 if (STRLEN(line) > 16)
|
|
|
4819 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
4820 fname, lnum, line);
|
|
|
4821 else
|
|
|
4822 {
|
|
|
4823 spin->si_region_count = STRLEN(line) / 2;
|
|
|
4824 STRCPY(spin->si_region_name, line);
|
|
389
|
4825
|
|
|
4826 /* Adjust the mask for a word valid in all regions. */
|
|
|
4827 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
4828 }
|
|
|
4829 }
|
|
|
4830 continue;
|
|
|
4831 }
|
|
|
4832
|
|
381
|
4833 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
4834 fname, lnum, line - 1);
|
|
|
4835 continue;
|
|
|
4836 }
|
|
|
4837
|
|
|
4838 flags = 0;
|
|
|
4839 regionmask = spin->si_region;
|
|
|
4840
|
|
|
4841 /* Check for flags and region after a slash. */
|
|
|
4842 p = vim_strchr(line, '/');
|
|
|
4843 if (p != NULL)
|
|
|
4844 {
|
|
|
4845 *p++ = NUL;
|
|
|
4846 while (*p != NUL)
|
|
|
4847 {
|
|
|
4848 if (*p == '=') /* keep-case word */
|
|
389
|
4849 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
4850 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
4851 flags |= WF_BANNED;
|
|
|
4852 else if (*p == '?') /* Rare word. */
|
|
|
4853 flags |= WF_RARE;
|
|
|
4854 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
4855 {
|
|
381
|
4856 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
4857 regionmask = 0;
|
|
|
4858 flags |= WF_REGION;
|
|
|
4859
|
|
|
4860 l = *p - '0';
|
|
316
|
4861 if (l > spin->si_region_count)
|
|
|
4862 {
|
|
|
4863 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
4864 fname, lnum, p);
|
|
316
|
4865 break;
|
|
|
4866 }
|
|
|
4867 regionmask |= 1 << (l - 1);
|
|
|
4868 }
|
|
381
|
4869 else
|
|
|
4870 {
|
|
|
4871 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
4872 fname, lnum, p);
|
|
|
4873 break;
|
|
|
4874 }
|
|
|
4875 ++p;
|
|
307
|
4876 }
|
|
300
|
4877 }
|
|
|
4878
|
|
|
4879 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4880 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4881 {
|
|
|
4882 ++non_ascii;
|
|
|
4883 continue;
|
|
|
4884 }
|
|
|
4885
|
|
|
4886 /* Normal word: store it. */
|
|
474
|
4887 if (store_word(spin, line, flags, regionmask, NULL) == FAIL)
|
|
300
|
4888 {
|
|
|
4889 retval = FAIL;
|
|
|
4890 break;
|
|
|
4891 }
|
|
|
4892 did_word = TRUE;
|
|
|
4893 }
|
|
|
4894
|
|
|
4895 vim_free(pc);
|
|
|
4896 fclose(fd);
|
|
|
4897
|
|
310
|
4898 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
4899 {
|
|
|
4900 if (p_verbose > 2)
|
|
|
4901 verbose_enter();
|
|
300
|
4902 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4903 non_ascii);
|
|
310
|
4904 if (p_verbose > 2)
|
|
|
4905 verbose_leave();
|
|
|
4906 }
|
|
300
|
4907 return retval;
|
|
236
|
4908 }
|
|
|
4909
|
|
|
4910 /*
|
|
300
|
4911 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
4912 * This avoids calling free() for every little struct we use (and keeping
|
|
|
4913 * track of them).
|
|
300
|
4914 * The memory is cleared to all zeros.
|
|
|
4915 * Returns NULL when out of memory.
|
|
|
4916 */
|
|
|
4917 static void *
|
|
474
|
4918 getroom(spin, len, align)
|
|
|
4919 spellinfo_T *spin;
|
|
447
|
4920 size_t len; /* length needed */
|
|
|
4921 int align; /* align for pointer */
|
|
300
|
4922 {
|
|
|
4923 char_u *p;
|
|
474
|
4924 sblock_T *bl = spin->si_blocks;
|
|
300
|
4925
|
|
447
|
4926 if (align && bl != NULL)
|
|
|
4927 /* Round size up for alignment. On some systems structures need to be
|
|
|
4928 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
4929 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
4930 & ~(sizeof(char *) - 1);
|
|
|
4931
|
|
300
|
4932 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
4933 {
|
|
|
4934 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
4935 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
4936 if (bl == NULL)
|
|
|
4937 return NULL;
|
|
474
|
4938 bl->sb_next = spin->si_blocks;
|
|
|
4939 spin->si_blocks = bl;
|
|
300
|
4940 bl->sb_used = 0;
|
|
|
4941 }
|
|
|
4942
|
|
|
4943 p = bl->sb_data + bl->sb_used;
|
|
|
4944 bl->sb_used += len;
|
|
|
4945
|
|
|
4946 return p;
|
|
|
4947 }
|
|
|
4948
|
|
|
4949 /*
|
|
|
4950 * Make a copy of a string into memory allocated with getroom().
|
|
|
4951 */
|
|
|
4952 static char_u *
|
|
474
|
4953 getroom_save(spin, s)
|
|
|
4954 spellinfo_T *spin;
|
|
300
|
4955 char_u *s;
|
|
|
4956 {
|
|
|
4957 char_u *sc;
|
|
|
4958
|
|
474
|
4959 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
4960 if (sc != NULL)
|
|
|
4961 STRCPY(sc, s);
|
|
|
4962 return sc;
|
|
|
4963 }
|
|
|
4964
|
|
|
4965
|
|
|
4966 /*
|
|
|
4967 * Free the list of allocated sblock_T.
|
|
236
|
4968 */
|
|
|
4969 static void
|
|
300
|
4970 free_blocks(bl)
|
|
|
4971 sblock_T *bl;
|
|
236
|
4972 {
|
|
300
|
4973 sblock_T *next;
|
|
236
|
4974
|
|
300
|
4975 while (bl != NULL)
|
|
236
|
4976 {
|
|
300
|
4977 next = bl->sb_next;
|
|
|
4978 vim_free(bl);
|
|
|
4979 bl = next;
|
|
236
|
4980 }
|
|
|
4981 }
|
|
|
4982
|
|
|
4983 /*
|
|
300
|
4984 * Allocate the root of a word tree.
|
|
236
|
4985 */
|
|
300
|
4986 static wordnode_T *
|
|
474
|
4987 wordtree_alloc(spin)
|
|
|
4988 spellinfo_T *spin;
|
|
|
4989 {
|
|
|
4990 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
4991 }
|
|
|
4992
|
|
|
4993 /*
|
|
300
|
4994 * Store a word in the tree(s).
|
|
390
|
4995 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
4996 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
4997 * used to find suggestions.
|
|
300
|
4998 * For a keep-case word also store it in the keep-case tree.
|
|
390
|
4999 * When "pfxlist" is not NULL store the word for each postponed prefix ID.
|
|
236
|
5000 */
|
|
|
5001 static int
|
|
474
|
5002 store_word(spin, word, flags, region, pfxlist)
|
|
|
5003 spellinfo_T *spin;
|
|
300
|
5004 char_u *word;
|
|
307
|
5005 int flags; /* extra flags, WF_BANNED */
|
|
316
|
5006 int region; /* supported region(s) */
|
|
339
|
5007 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
236
|
5008 {
|
|
300
|
5009 int len = STRLEN(word);
|
|
|
5010 int ct = captype(word, word + len);
|
|
|
5011 char_u foldword[MAXWLEN];
|
|
339
|
5012 int res = OK;
|
|
|
5013 char_u *p;
|
|
236
|
5014
|
|
323
|
5015 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
5016 for (p = pfxlist; res == OK; ++p)
|
|
|
5017 {
|
|
474
|
5018 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
|
5019 region, p == NULL ? 0 : *p);
|
|
339
|
5020 if (p == NULL || *p == NUL)
|
|
|
5021 break;
|
|
|
5022 }
|
|
334
|
5023 ++spin->si_foldwcount;
|
|
307
|
5024
|
|
474
|
5025 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
5026 {
|
|
339
|
5027 for (p = pfxlist; res == OK; ++p)
|
|
|
5028 {
|
|
474
|
5029 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
|
5030 region, p == NULL ? 0 : *p);
|
|
339
|
5031 if (p == NULL || *p == NUL)
|
|
|
5032 break;
|
|
|
5033 }
|
|
334
|
5034 ++spin->si_keepwcount;
|
|
|
5035 }
|
|
300
|
5036 return res;
|
|
236
|
5037 }
|
|
|
5038
|
|
|
5039 /*
|
|
300
|
5040 * Add word "word" to a word tree at "root".
|
|
366
|
5041 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
5042 * "rare" and "region" is the condition nr.
|
|
255
|
5043 * Returns FAIL when out of memory.
|
|
236
|
5044 */
|
|
255
|
5045 static int
|
|
481
|
5046 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
5047 spellinfo_T *spin;
|
|
300
|
5048 char_u *word;
|
|
|
5049 wordnode_T *root;
|
|
|
5050 int flags;
|
|
|
5051 int region;
|
|
481
|
5052 int affixID;
|
|
236
|
5053 {
|
|
300
|
5054 wordnode_T *node = root;
|
|
|
5055 wordnode_T *np;
|
|
469
|
5056 wordnode_T *copyp, **copyprev;
|
|
300
|
5057 wordnode_T **prev = NULL;
|
|
|
5058 int i;
|
|
255
|
5059
|
|
300
|
5060 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
5061 for (i = 0; ; ++i)
|
|
255
|
5062 {
|
|
469
|
5063 /* When there is more than one reference to this node we need to make
|
|
|
5064 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
5065 * (we don't optimize for a partly shared list of siblings). */
|
|
|
5066 if (node != NULL && node->wn_refs > 1)
|
|
|
5067 {
|
|
|
5068 --node->wn_refs;
|
|
|
5069 copyprev = prev;
|
|
|
5070 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
5071 {
|
|
|
5072 /* Allocate a new node and copy the info. */
|
|
474
|
5073 np = get_wordnode(spin);
|
|
469
|
5074 if (np == NULL)
|
|
|
5075 return FAIL;
|
|
|
5076 np->wn_child = copyp->wn_child;
|
|
|
5077 if (np->wn_child != NULL)
|
|
|
5078 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
5079 np->wn_byte = copyp->wn_byte;
|
|
|
5080 if (np->wn_byte == NUL)
|
|
|
5081 {
|
|
|
5082 np->wn_flags = copyp->wn_flags;
|
|
|
5083 np->wn_region = copyp->wn_region;
|
|
481
|
5084 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
5085 }
|
|
|
5086
|
|
|
5087 /* Link the new node in the list, there will be one ref. */
|
|
|
5088 np->wn_refs = 1;
|
|
|
5089 *copyprev = np;
|
|
|
5090 copyprev = &np->wn_sibling;
|
|
|
5091
|
|
|
5092 /* Let "node" point to the head of the copied list. */
|
|
|
5093 if (copyp == node)
|
|
|
5094 node = np;
|
|
|
5095 }
|
|
|
5096 }
|
|
|
5097
|
|
300
|
5098 /* Look for the sibling that has the same character. They are sorted
|
|
|
5099 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
5100 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
5101 * done on flags and then on affixID. */
|
|
339
|
5102 while (node != NULL
|
|
|
5103 && (node->wn_byte < word[i]
|
|
|
5104 || (node->wn_byte == NUL
|
|
|
5105 && (flags < 0
|
|
481
|
5106 ? node->wn_affixID < affixID
|
|
390
|
5107 : node->wn_flags < (flags & WN_MASK)
|
|
|
5108 || (node->wn_flags == (flags & WN_MASK)
|
|
481
|
5109 && node->wn_affixID < affixID)))))
|
|
236
|
5110 {
|
|
300
|
5111 prev = &node->wn_sibling;
|
|
|
5112 node = *prev;
|
|
236
|
5113 }
|
|
339
|
5114 if (node == NULL
|
|
|
5115 || node->wn_byte != word[i]
|
|
|
5116 || (word[i] == NUL
|
|
|
5117 && (flags < 0
|
|
390
|
5118 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
5119 || node->wn_affixID != affixID)))
|
|
255
|
5120 {
|
|
300
|
5121 /* Allocate a new node. */
|
|
474
|
5122 np = get_wordnode(spin);
|
|
300
|
5123 if (np == NULL)
|
|
|
5124 return FAIL;
|
|
|
5125 np->wn_byte = word[i];
|
|
469
|
5126
|
|
|
5127 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
5128 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
5129 * make ref count of sibling one (matters when inserting in front
|
|
|
5130 * of the list of siblings). */
|
|
|
5131 if (node == NULL)
|
|
|
5132 np->wn_refs = 1;
|
|
|
5133 else
|
|
|
5134 {
|
|
|
5135 np->wn_refs = node->wn_refs;
|
|
|
5136 node->wn_refs = 1;
|
|
|
5137 }
|
|
300
|
5138 *prev = np;
|
|
|
5139 np->wn_sibling = node;
|
|
|
5140 node = np;
|
|
255
|
5141 }
|
|
300
|
5142
|
|
|
5143 if (word[i] == NUL)
|
|
|
5144 {
|
|
|
5145 node->wn_flags = flags;
|
|
|
5146 node->wn_region |= region;
|
|
481
|
5147 node->wn_affixID = affixID;
|
|
300
|
5148 break;
|
|
|
5149 }
|
|
|
5150 prev = &node->wn_child;
|
|
|
5151 node = *prev;
|
|
255
|
5152 }
|
|
469
|
5153 #ifdef SPELL_PRINTTREE
|
|
|
5154 smsg("Added \"%s\"", word);
|
|
|
5155 spell_print_tree(root->wn_sibling);
|
|
|
5156 #endif
|
|
|
5157
|
|
474
|
5158 /* count nr of words added since last message */
|
|
|
5159 ++spin->si_msg_count;
|
|
|
5160
|
|
469
|
5161 /*
|
|
|
5162 * Every so many words compress the tree, so that we don't use too much
|
|
|
5163 * memory.
|
|
|
5164 */
|
|
474
|
5165 i = FALSE;
|
|
|
5166 if (root == spin->si_foldroot)
|
|
|
5167 {
|
|
|
5168 if (++spin->si_fold_added >= SPELL_COMPRESS_CNT)
|
|
|
5169 {
|
|
|
5170 i = TRUE;
|
|
|
5171 spin->si_fold_added = 0;
|
|
|
5172 }
|
|
|
5173 }
|
|
|
5174 else if (root == spin->si_keeproot)
|
|
|
5175 {
|
|
|
5176 if (++spin->si_keep_added >= SPELL_COMPRESS_CNT)
|
|
|
5177 {
|
|
|
5178 i = TRUE;
|
|
|
5179 spin->si_keep_added = 0;
|
|
|
5180 }
|
|
|
5181 }
|
|
|
5182 if (i)
|
|
|
5183 {
|
|
|
5184 if (spin->si_verbose)
|
|
|
5185 {
|
|
|
5186 msg_start();
|
|
|
5187 msg_puts((char_u *)_(msg_compressing));
|
|
|
5188 msg_clr_eos();
|
|
|
5189 msg_didout = FALSE;
|
|
|
5190 msg_col = 0;
|
|
|
5191 out_flush();
|
|
|
5192 }
|
|
|
5193 wordtree_compress(spin, root);
|
|
469
|
5194 }
|
|
255
|
5195
|
|
|
5196 return OK;
|
|
236
|
5197 }
|
|
|
5198
|
|
469
|
5199 /*
|
|
|
5200 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
5201 * allocate a new one.
|
|
|
5202 */
|
|
|
5203 static wordnode_T *
|
|
474
|
5204 get_wordnode(spin)
|
|
|
5205 spellinfo_T *spin;
|
|
469
|
5206 {
|
|
|
5207 wordnode_T *n;
|
|
|
5208
|
|
474
|
5209 if (spin->si_first_free == NULL)
|
|
|
5210 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
5211 else
|
|
|
5212 {
|
|
474
|
5213 n = spin->si_first_free;
|
|
|
5214 spin->si_first_free = n->wn_child;
|
|
469
|
5215 vim_memset(n, 0, sizeof(wordnode_T));
|
|
|
5216 }
|
|
|
5217 #ifdef SPELL_PRINTTREE
|
|
474
|
5218 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
5219 #endif
|
|
|
5220 return n;
|
|
|
5221 }
|
|
|
5222
|
|
|
5223 /*
|
|
474
|
5224 * Decrement the reference count on a node (which is the head of a list of
|
|
|
5225 * siblings). If the reference count becomes zero free the node and its
|
|
|
5226 * siblings.
|
|
|
5227 */
|
|
|
5228 static void
|
|
|
5229 deref_wordnode(spin, node)
|
|
|
5230 spellinfo_T *spin;
|
|
|
5231 wordnode_T *node;
|
|
|
5232 {
|
|
|
5233 wordnode_T *np;
|
|
|
5234
|
|
|
5235 if (--node->wn_refs == 0)
|
|
|
5236 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5237 {
|
|
|
5238 if (np->wn_child != NULL)
|
|
|
5239 deref_wordnode(spin, np->wn_child);
|
|
|
5240 free_wordnode(spin, np);
|
|
|
5241 }
|
|
|
5242 }
|
|
|
5243
|
|
|
5244 /*
|
|
469
|
5245 * Free a wordnode_T for re-use later.
|
|
474
|
5246 * Only the "wn_child" field becomes invalid.
|
|
469
|
5247 */
|
|
|
5248 static void
|
|
474
|
5249 free_wordnode(spin, n)
|
|
|
5250 spellinfo_T *spin;
|
|
|
5251 wordnode_T *n;
|
|
|
5252 {
|
|
|
5253 n->wn_child = spin->si_first_free;
|
|
|
5254 spin->si_first_free = n;
|
|
469
|
5255 }
|
|
|
5256
|
|
236
|
5257 /*
|
|
300
|
5258 * Compress a tree: find tails that are identical and can be shared.
|
|
|
5259 */
|
|
|
5260 static void
|
|
474
|
5261 wordtree_compress(spin, root)
|
|
|
5262 spellinfo_T *spin;
|
|
300
|
5263 wordnode_T *root;
|
|
|
5264 {
|
|
|
5265 hashtab_T ht;
|
|
|
5266 int n;
|
|
|
5267 int tot = 0;
|
|
469
|
5268 int perc;
|
|
300
|
5269
|
|
474
|
5270 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
5271 * start of the tree. */
|
|
|
5272 if (root->wn_sibling != NULL)
|
|
300
|
5273 {
|
|
|
5274 hash_init(&ht);
|
|
474
|
5275 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
5276
|
|
|
5277 #ifndef SPELL_PRINTTREE
|
|
310
|
5278 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
5279 #endif
|
|
310
|
5280 {
|
|
|
5281 if (!spin->si_verbose)
|
|
|
5282 verbose_enter();
|
|
469
|
5283 if (tot > 1000000)
|
|
|
5284 perc = (tot - n) / (tot / 100);
|
|
|
5285 else
|
|
|
5286 perc = (tot - n) * 100 / tot;
|
|
310
|
5287 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
5288 n, tot, perc);
|
|
310
|
5289 if (p_verbose > 2)
|
|
|
5290 verbose_leave();
|
|
|
5291 }
|
|
469
|
5292 #ifdef SPELL_PRINTTREE
|
|
474
|
5293 spell_print_tree(root->wn_sibling);
|
|
469
|
5294 #endif
|
|
300
|
5295 hash_clear(&ht);
|
|
|
5296 }
|
|
|
5297 }
|
|
|
5298
|
|
|
5299 /*
|
|
|
5300 * Compress a node, its siblings and its children, depth first.
|
|
|
5301 * Returns the number of compressed nodes.
|
|
236
|
5302 */
|
|
255
|
5303 static int
|
|
474
|
5304 node_compress(spin, node, ht, tot)
|
|
|
5305 spellinfo_T *spin;
|
|
300
|
5306 wordnode_T *node;
|
|
|
5307 hashtab_T *ht;
|
|
|
5308 int *tot; /* total count of nodes before compressing,
|
|
|
5309 incremented while going through the tree */
|
|
236
|
5310 {
|
|
300
|
5311 wordnode_T *np;
|
|
|
5312 wordnode_T *tp;
|
|
|
5313 wordnode_T *child;
|
|
|
5314 hash_T hash;
|
|
236
|
5315 hashitem_T *hi;
|
|
300
|
5316 int len = 0;
|
|
|
5317 unsigned nr, n;
|
|
|
5318 int compressed = 0;
|
|
236
|
5319
|
|
300
|
5320 /*
|
|
|
5321 * Go through the list of siblings. Compress each child and then try
|
|
|
5322 * finding an identical child to replace it.
|
|
|
5323 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
5324 * but the whole list of siblings, of which the node is the first.
|
|
|
5325 */
|
|
|
5326 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5327 {
|
|
300
|
5328 ++len;
|
|
|
5329 if ((child = np->wn_child) != NULL)
|
|
|
5330 {
|
|
346
|
5331 /* Compress the child. This fills hashkey. */
|
|
474
|
5332 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
5333
|
|
|
5334 /* Try to find an identical child. */
|
|
346
|
5335 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
5336 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
5337 tp = NULL;
|
|
|
5338 if (!HASHITEM_EMPTY(hi))
|
|
|
5339 {
|
|
|
5340 /* There are children with an identical hash value. Now check
|
|
|
5341 * if there is one that is really identical. */
|
|
346
|
5342 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
5343 if (node_equal(child, tp))
|
|
|
5344 {
|
|
|
5345 /* Found one! Now use that child in place of the
|
|
469
|
5346 * current one. This means the current child and all
|
|
|
5347 * its siblings is unlinked from the tree. */
|
|
|
5348 ++tp->wn_refs;
|
|
474
|
5349 deref_wordnode(spin, child);
|
|
300
|
5350 np->wn_child = tp;
|
|
|
5351 ++compressed;
|
|
|
5352 break;
|
|
|
5353 }
|
|
|
5354 if (tp == NULL)
|
|
|
5355 {
|
|
|
5356 /* No other child with this hash value equals the child of
|
|
|
5357 * the node, add it to the linked list after the first
|
|
|
5358 * item. */
|
|
|
5359 tp = HI2WN(hi);
|
|
346
|
5360 child->wn_u2.next = tp->wn_u2.next;
|
|
|
5361 tp->wn_u2.next = child;
|
|
300
|
5362 }
|
|
|
5363 }
|
|
|
5364 else
|
|
|
5365 /* No other child has this hash value, add it to the
|
|
|
5366 * hashtable. */
|
|
346
|
5367 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
5368 }
|
|
236
|
5369 }
|
|
300
|
5370 *tot += len;
|
|
|
5371
|
|
|
5372 /*
|
|
|
5373 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
5374 * find a lookalike node. This must be done after compressing the sibling
|
|
|
5375 * list, otherwise the hash key would become invalid by the compression.
|
|
|
5376 */
|
|
346
|
5377 node->wn_u1.hashkey[0] = len;
|
|
300
|
5378 nr = 0;
|
|
|
5379 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5380 {
|
|
300
|
5381 if (np->wn_byte == NUL)
|
|
481
|
5382 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
5383 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
5384 else
|
|
|
5385 /* byte node: use the byte value and the child pointer */
|
|
|
5386 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
5387 nr = nr * 101 + n;
|
|
236
|
5388 }
|
|
300
|
5389
|
|
|
5390 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
5391 n = nr & 0xff;
|
|
346
|
5392 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
5393 n = (nr >> 8) & 0xff;
|
|
346
|
5394 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
5395 n = (nr >> 16) & 0xff;
|
|
346
|
5396 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
5397 n = (nr >> 24) & 0xff;
|
|
346
|
5398 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
5399 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
5400
|
|
|
5401 return compressed;
|
|
|
5402 }
|
|
|
5403
|
|
|
5404 /*
|
|
|
5405 * Return TRUE when two nodes have identical siblings and children.
|
|
|
5406 */
|
|
|
5407 static int
|
|
|
5408 node_equal(n1, n2)
|
|
|
5409 wordnode_T *n1;
|
|
|
5410 wordnode_T *n2;
|
|
|
5411 {
|
|
|
5412 wordnode_T *p1;
|
|
|
5413 wordnode_T *p2;
|
|
|
5414
|
|
|
5415 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
5416 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
5417 if (p1->wn_byte != p2->wn_byte
|
|
|
5418 || (p1->wn_byte == NUL
|
|
|
5419 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
5420 || p1->wn_region != p2->wn_region
|
|
481
|
5421 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
5422 : (p1->wn_child != p2->wn_child)))
|
|
|
5423 break;
|
|
|
5424
|
|
|
5425 return p1 == NULL && p2 == NULL;
|
|
236
|
5426 }
|
|
|
5427
|
|
|
5428 /*
|
|
|
5429 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
5430 */
|
|
255
|
5431 void
|
|
236
|
5432 put_bytes(fd, nr, len)
|
|
|
5433 FILE *fd;
|
|
|
5434 long_u nr;
|
|
|
5435 int len;
|
|
|
5436 {
|
|
|
5437 int i;
|
|
|
5438
|
|
|
5439 for (i = len - 1; i >= 0; --i)
|
|
|
5440 putc((int)(nr >> (i * 8)), fd);
|
|
|
5441 }
|
|
|
5442
|
|
323
|
5443 static int
|
|
|
5444 #ifdef __BORLANDC__
|
|
|
5445 _RTLENTRYF
|
|
|
5446 #endif
|
|
|
5447 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
5448
|
|
|
5449 /*
|
|
|
5450 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
5451 */
|
|
|
5452 static int
|
|
|
5453 #ifdef __BORLANDC__
|
|
|
5454 _RTLENTRYF
|
|
|
5455 #endif
|
|
|
5456 rep_compare(s1, s2)
|
|
|
5457 const void *s1;
|
|
|
5458 const void *s2;
|
|
|
5459 {
|
|
|
5460 fromto_T *p1 = (fromto_T *)s1;
|
|
|
5461 fromto_T *p2 = (fromto_T *)s2;
|
|
|
5462
|
|
|
5463 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
5464 }
|
|
|
5465
|
|
236
|
5466 /*
|
|
|
5467 * Write the Vim spell file "fname".
|
|
|
5468 */
|
|
|
5469 static void
|
|
474
|
5470 write_vim_spell(spin, fname)
|
|
|
5471 spellinfo_T *spin;
|
|
236
|
5472 char_u *fname;
|
|
|
5473 {
|
|
300
|
5474 FILE *fd;
|
|
|
5475 int regionmask;
|
|
236
|
5476 int round;
|
|
300
|
5477 wordnode_T *tree;
|
|
|
5478 int nodecount;
|
|
323
|
5479 int i;
|
|
|
5480 int l;
|
|
|
5481 garray_T *gap;
|
|
|
5482 fromto_T *ftp;
|
|
|
5483 char_u *p;
|
|
|
5484 int rr;
|
|
236
|
5485
|
|
310
|
5486 fd = mch_fopen((char *)fname, "w");
|
|
300
|
5487 if (fd == NULL)
|
|
236
|
5488 {
|
|
|
5489 EMSG2(_(e_notopen), fname);
|
|
|
5490 return;
|
|
|
5491 }
|
|
|
5492
|
|
255
|
5493 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
339
|
5494 * <charflagslen> <charflags>
|
|
|
5495 * <fcharslen> <fchars>
|
|
366
|
5496 * <midwordlen> <midword>
|
|
481
|
5497 * <compoundlen> <compoundtype> <compoundinfo>
|
|
339
|
5498 * <prefcondcnt> <prefcond> ... */
|
|
300
|
5499
|
|
|
5500 /* <fileID> */
|
|
|
5501 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
5502 EMSG(_(e_write));
|
|
236
|
5503
|
|
|
5504 /* write the region names if there is more than one */
|
|
316
|
5505 if (spin->si_region_count > 1)
|
|
236
|
5506 {
|
|
316
|
5507 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
|
5508 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
|
5509 (size_t)1, fd);
|
|
|
5510 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
5511 }
|
|
|
5512 else
|
|
|
5513 {
|
|
300
|
5514 putc(0, fd);
|
|
|
5515 regionmask = 0;
|
|
236
|
5516 }
|
|
|
5517
|
|
323
|
5518 /*
|
|
|
5519 * Write the table with character flags and table for case folding.
|
|
260
|
5520 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
|
5521 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
5522 * 'encoding'.
|
|
|
5523 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
5524 * the table (avoids that it conflicts). File is shorter too.
|
|
|
5525 */
|
|
|
5526 if (spin->si_ascii || spin->si_add)
|
|
260
|
5527 {
|
|
300
|
5528 putc(0, fd);
|
|
|
5529 putc(0, fd);
|
|
|
5530 putc(0, fd);
|
|
260
|
5531 }
|
|
|
5532 else
|
|
300
|
5533 write_spell_chartab(fd);
|
|
255
|
5534
|
|
366
|
5535
|
|
|
5536 if (spin->si_midword == NULL)
|
|
|
5537 put_bytes(fd, 0L, 2); /* <midwordlen> */
|
|
|
5538 else
|
|
|
5539 {
|
|
|
5540 i = STRLEN(spin->si_midword);
|
|
|
5541 put_bytes(fd, (long_u)i, 2); /* <midwordlen> */
|
|
|
5542 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
5543 }
|
|
|
5544
|
|
|
5545
|
|
481
|
5546 /* Write the compound info. */
|
|
|
5547 if (spin->si_compflags == NULL)
|
|
|
5548 put_bytes(fd, 0L, 2); /* <compoundlen> */
|
|
|
5549 else
|
|
|
5550 {
|
|
|
5551 l = STRLEN(spin->si_compflags);
|
|
|
5552 put_bytes(fd, (long_u)(l + 2), 2); /* <compoundlen> */
|
|
|
5553 putc(1, fd); /* <compoundtype> */
|
|
|
5554 putc(spin->si_compminlen, fd); /* <comp1minlen> */
|
|
|
5555 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
5556 /* <comp1flags> */
|
|
|
5557 }
|
|
|
5558
|
|
|
5559
|
|
339
|
5560 /* Write the prefix conditions. */
|
|
|
5561 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
5562
|
|
366
|
5563 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
5564 * <salflags> <salcount> <sal> ...
|
|
|
5565 * <maplen> <mapstr> */
|
|
|
5566
|
|
323
|
5567 /* Sort the REP items. */
|
|
|
5568 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
5569 sizeof(fromto_T), rep_compare);
|
|
|
5570
|
|
375
|
5571 /* round 1: REP items
|
|
|
5572 * round 2: SAL items (unless SOFO is used) */
|
|
323
|
5573 for (round = 1; round <= 2; ++round)
|
|
|
5574 {
|
|
|
5575 if (round == 1)
|
|
|
5576 gap = &spin->si_rep;
|
|
|
5577 else
|
|
|
5578 {
|
|
|
5579 gap = &spin->si_sal;
|
|
|
5580
|
|
|
5581 i = 0;
|
|
|
5582 if (spin->si_followup)
|
|
|
5583 i |= SAL_F0LLOWUP;
|
|
|
5584 if (spin->si_collapse)
|
|
|
5585 i |= SAL_COLLAPSE;
|
|
|
5586 if (spin->si_rem_accents)
|
|
|
5587 i |= SAL_REM_ACCENTS;
|
|
375
|
5588 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5589 i |= SAL_SOFO;
|
|
323
|
5590 putc(i, fd); /* <salflags> */
|
|
375
|
5591 if (i & SAL_SOFO)
|
|
|
5592 break;
|
|
323
|
5593 }
|
|
|
5594
|
|
|
5595 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
5596 for (i = 0; i < gap->ga_len; ++i)
|
|
|
5597 {
|
|
|
5598 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
5599 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
5600 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
5601 for (rr = 1; rr <= 2; ++rr)
|
|
|
5602 {
|
|
|
5603 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
5604 l = STRLEN(p);
|
|
|
5605 putc(l, fd);
|
|
|
5606 fwrite(p, l, (size_t)1, fd);
|
|
|
5607 }
|
|
|
5608 }
|
|
|
5609 }
|
|
|
5610
|
|
375
|
5611 /* SOFOFROM and SOFOTO */
|
|
|
5612 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5613 {
|
|
|
5614 put_bytes(fd, 1L, 2); /* <salcount> */
|
|
|
5615
|
|
|
5616 l = STRLEN(spin->si_sofofr);
|
|
|
5617 put_bytes(fd, (long_u)l, 2); /* <salfromlen> */
|
|
|
5618 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <salfrom> */
|
|
|
5619
|
|
|
5620 l = STRLEN(spin->si_sofoto);
|
|
|
5621 put_bytes(fd, (long_u)l, 2); /* <saltolen> */
|
|
|
5622 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <salto> */
|
|
|
5623 }
|
|
|
5624
|
|
323
|
5625 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
|
5626 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
|
5627 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
|
5628 (size_t)1, fd);
|
|
302
|
5629
|
|
236
|
5630 /*
|
|
339
|
5631 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
5632 */
|
|
323
|
5633 spin->si_memtot = 0;
|
|
339
|
5634 for (round = 1; round <= 3; ++round)
|
|
236
|
5635 {
|
|
339
|
5636 if (round == 1)
|
|
469
|
5637 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
5638 else if (round == 2)
|
|
469
|
5639 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
5640 else
|
|
469
|
5641 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
5642
|
|
346
|
5643 /* Clear the index and wnode fields in the tree. */
|
|
|
5644 clear_node(tree);
|
|
|
5645
|
|
300
|
5646 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
5647 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
5648 * nodes. */
|
|
346
|
5649 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
5650
|
|
300
|
5651 /* number of nodes in 4 bytes */
|
|
|
5652 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
5653 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
5654
|
|
300
|
5655 /* Write the nodes. */
|
|
346
|
5656 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
5657 }
|
|
|
5658
|
|
300
|
5659 fclose(fd);
|
|
236
|
5660 }
|
|
|
5661
|
|
|
5662 /*
|
|
346
|
5663 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
5664 * children. This is needed because they are a union with other items to save
|
|
|
5665 * space.
|
|
|
5666 */
|
|
|
5667 static void
|
|
|
5668 clear_node(node)
|
|
|
5669 wordnode_T *node;
|
|
|
5670 {
|
|
|
5671 wordnode_T *np;
|
|
|
5672
|
|
|
5673 if (node != NULL)
|
|
|
5674 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5675 {
|
|
|
5676 np->wn_u1.index = 0;
|
|
|
5677 np->wn_u2.wnode = NULL;
|
|
|
5678
|
|
|
5679 if (np->wn_byte != NUL)
|
|
|
5680 clear_node(np->wn_child);
|
|
|
5681 }
|
|
|
5682 }
|
|
|
5683
|
|
|
5684
|
|
|
5685 /*
|
|
300
|
5686 * Dump a word tree at node "node".
|
|
|
5687 *
|
|
|
5688 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
5689 * byte recursively write the children.
|
|
|
5690 *
|
|
|
5691 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
5692 * are made about the indexes (so that we don't have to write them in the
|
|
|
5693 * file).
|
|
236
|
5694 *
|
|
300
|
5695 * Returns the number of nodes used.
|
|
236
|
5696 */
|
|
300
|
5697 static int
|
|
346
|
5698 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
5699 FILE *fd; /* NULL when only counting */
|
|
300
|
5700 wordnode_T *node;
|
|
|
5701 int index;
|
|
|
5702 int regionmask;
|
|
339
|
5703 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
5704 {
|
|
300
|
5705 int newindex = index;
|
|
|
5706 int siblingcount = 0;
|
|
|
5707 wordnode_T *np;
|
|
236
|
5708 int flags;
|
|
300
|
5709
|
|
|
5710 /* If "node" is zero the tree is empty. */
|
|
|
5711 if (node == NULL)
|
|
|
5712 return 0;
|
|
|
5713
|
|
|
5714 /* Store the index where this node is written. */
|
|
346
|
5715 node->wn_u1.index = index;
|
|
236
|
5716
|
|
300
|
5717 /* Count the number of siblings. */
|
|
|
5718 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5719 ++siblingcount;
|
|
236
|
5720
|
|
300
|
5721 /* Write the sibling count. */
|
|
|
5722 if (fd != NULL)
|
|
|
5723 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
5724
|
|
300
|
5725 /* Write each sibling byte and optionally extra info. */
|
|
|
5726 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5727 {
|
|
300
|
5728 if (np->wn_byte == 0)
|
|
|
5729 {
|
|
|
5730 if (fd != NULL)
|
|
|
5731 {
|
|
339
|
5732 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
5733 if (prefixtree)
|
|
300
|
5734 {
|
|
481
|
5735 /* In PREFIXTREE write the required affixID and the
|
|
390
|
5736 * associated condition nr (stored in wn_region). The
|
|
|
5737 * byte value is misused to store the "rare" and "not
|
|
|
5738 * combining" flags */
|
|
455
|
5739 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
5740 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5741 else
|
|
|
5742 {
|
|
390
|
5743 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
5744 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
5745 }
|
|
481
|
5746 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
5747 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
5748 }
|
|
|
5749 else
|
|
|
5750 {
|
|
339
|
5751 /* For word trees we write the flag/region items. */
|
|
|
5752 flags = np->wn_flags;
|
|
|
5753 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
5754 flags |= WF_REGION;
|
|
481
|
5755 if (np->wn_affixID != 0)
|
|
|
5756 flags |= WF_AFX;
|
|
339
|
5757 if (flags == 0)
|
|
|
5758 {
|
|
|
5759 /* word without flags or region */
|
|
|
5760 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5761 }
|
|
|
5762 else
|
|
|
5763 {
|
|
390
|
5764 if (np->wn_flags >= 0x100)
|
|
|
5765 {
|
|
|
5766 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
5767 putc(flags, fd); /* <flags> */
|
|
|
5768 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
5769 }
|
|
|
5770 else
|
|
|
5771 {
|
|
|
5772 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
5773 putc(flags, fd); /* <flags> */
|
|
|
5774 }
|
|
339
|
5775 if (flags & WF_REGION)
|
|
|
5776 putc(np->wn_region, fd); /* <region> */
|
|
481
|
5777 if (flags & WF_AFX)
|
|
|
5778 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
5779 }
|
|
300
|
5780 }
|
|
|
5781 }
|
|
|
5782 }
|
|
|
5783 else
|
|
|
5784 {
|
|
346
|
5785 if (np->wn_child->wn_u1.index != 0
|
|
|
5786 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
5787 {
|
|
|
5788 /* The child is written elsewhere, write the reference. */
|
|
|
5789 if (fd != NULL)
|
|
|
5790 {
|
|
|
5791 putc(BY_INDEX, fd); /* <byte> */
|
|
|
5792 /* <nodeidx> */
|
|
346
|
5793 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
5794 }
|
|
|
5795 }
|
|
346
|
5796 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
5797 /* We will write the child below and give it an index. */
|
|
346
|
5798 np->wn_child->wn_u2.wnode = node;
|
|
236
|
5799
|
|
300
|
5800 if (fd != NULL)
|
|
|
5801 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
5802 {
|
|
|
5803 EMSG(_(e_write));
|
|
|
5804 return 0;
|
|
|
5805 }
|
|
|
5806 }
|
|
236
|
5807 }
|
|
|
5808
|
|
300
|
5809 /* Space used in the array when reading: one for each sibling and one for
|
|
|
5810 * the count. */
|
|
|
5811 newindex += siblingcount + 1;
|
|
249
|
5812
|
|
300
|
5813 /* Recursively dump the children of each sibling. */
|
|
|
5814 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
5815 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
5816 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
5817 prefixtree);
|
|
249
|
5818
|
|
300
|
5819 return newindex;
|
|
236
|
5820 }
|
|
|
5821
|
|
|
5822
|
|
|
5823 /*
|
|
310
|
5824 * ":mkspell [-ascii] outfile infile ..."
|
|
|
5825 * ":mkspell [-ascii] addfile"
|
|
236
|
5826 */
|
|
|
5827 void
|
|
|
5828 ex_mkspell(eap)
|
|
|
5829 exarg_T *eap;
|
|
|
5830 {
|
|
|
5831 int fcount;
|
|
|
5832 char_u **fnames;
|
|
310
|
5833 char_u *arg = eap->arg;
|
|
|
5834 int ascii = FALSE;
|
|
|
5835
|
|
|
5836 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
5837 {
|
|
|
5838 ascii = TRUE;
|
|
|
5839 arg = skipwhite(arg + 6);
|
|
|
5840 }
|
|
|
5841
|
|
|
5842 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
5843 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
5844 {
|
|
323
|
5845 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
5846 FreeWild(fcount, fnames);
|
|
|
5847 }
|
|
|
5848 }
|
|
|
5849
|
|
|
5850 /*
|
|
|
5851 * Create a Vim spell file from one or more word lists.
|
|
|
5852 * "fnames[0]" is the output file name.
|
|
|
5853 * "fnames[fcount - 1]" is the last input file name.
|
|
|
5854 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
5855 * and ".spl" is appended to make the output file name.
|
|
|
5856 */
|
|
|
5857 static void
|
|
323
|
5858 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
5859 int fcount;
|
|
|
5860 char_u **fnames;
|
|
|
5861 int ascii; /* -ascii argument given */
|
|
|
5862 int overwrite; /* overwrite existing output file */
|
|
323
|
5863 int added_word; /* invoked through "zg" */
|
|
310
|
5864 {
|
|
236
|
5865 char_u fname[MAXPATHL];
|
|
|
5866 char_u wfname[MAXPATHL];
|
|
310
|
5867 char_u **innames;
|
|
|
5868 int incount;
|
|
236
|
5869 afffile_T *(afile[8]);
|
|
|
5870 int i;
|
|
|
5871 int len;
|
|
|
5872 struct stat st;
|
|
255
|
5873 int error = FALSE;
|
|
300
|
5874 spellinfo_T spin;
|
|
|
5875
|
|
|
5876 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
5877 spin.si_verbose = !added_word;
|
|
310
|
5878 spin.si_ascii = ascii;
|
|
323
|
5879 spin.si_followup = TRUE;
|
|
|
5880 spin.si_rem_accents = TRUE;
|
|
|
5881 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
5882 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
5883 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
5884 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
5885
|
|
|
5886 /* default: fnames[0] is output file, following are input files */
|
|
|
5887 innames = &fnames[1];
|
|
|
5888 incount = fcount - 1;
|
|
|
5889
|
|
|
5890 if (fcount >= 1)
|
|
240
|
5891 {
|
|
310
|
5892 len = STRLEN(fnames[0]);
|
|
|
5893 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
5894 {
|
|
|
5895 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
5896 * "path/en.latin1.add.spl". */
|
|
|
5897 innames = &fnames[0];
|
|
|
5898 incount = 1;
|
|
|
5899 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
5900 }
|
|
366
|
5901 else if (fcount == 1)
|
|
|
5902 {
|
|
|
5903 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
5904 innames = &fnames[0];
|
|
|
5905 incount = 1;
|
|
|
5906 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5907 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5908 }
|
|
310
|
5909 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
5910 {
|
|
|
5911 /* Name ends in ".spl", use as the file name. */
|
|
323
|
5912 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
5913 }
|
|
|
5914 else
|
|
|
5915 /* Name should be language, make the file name from it. */
|
|
|
5916 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5917 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5918
|
|
|
5919 /* Check for .ascii.spl. */
|
|
|
5920 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
5921 spin.si_ascii = TRUE;
|
|
|
5922
|
|
|
5923 /* Check for .add.spl. */
|
|
|
5924 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
5925 spin.si_add = TRUE;
|
|
240
|
5926 }
|
|
|
5927
|
|
310
|
5928 if (incount <= 0)
|
|
236
|
5929 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
5930 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
5931 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
5932 else if (incount > 8)
|
|
236
|
5933 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
5934 else
|
|
|
5935 {
|
|
|
5936 /* Check for overwriting before doing things that may take a lot of
|
|
|
5937 * time. */
|
|
310
|
5938 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
5939 {
|
|
|
5940 EMSG(_(e_exists));
|
|
310
|
5941 return;
|
|
236
|
5942 }
|
|
310
|
5943 if (mch_isdir(wfname))
|
|
236
|
5944 {
|
|
310
|
5945 EMSG2(_(e_isadir2), wfname);
|
|
|
5946 return;
|
|
236
|
5947 }
|
|
|
5948
|
|
|
5949 /*
|
|
|
5950 * Init the aff and dic pointers.
|
|
|
5951 * Get the region names if there are more than 2 arguments.
|
|
|
5952 */
|
|
310
|
5953 for (i = 0; i < incount; ++i)
|
|
236
|
5954 {
|
|
310
|
5955 afile[i] = NULL;
|
|
300
|
5956
|
|
316
|
5957 if (incount > 1)
|
|
236
|
5958 {
|
|
310
|
5959 len = STRLEN(innames[i]);
|
|
|
5960 if (STRLEN(gettail(innames[i])) < 5
|
|
|
5961 || innames[i][len - 3] != '_')
|
|
236
|
5962 {
|
|
310
|
5963 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
5964 return;
|
|
236
|
5965 }
|
|
316
|
5966 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
5967 spin.si_region_name[i * 2 + 1] =
|
|
|
5968 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
5969 }
|
|
|
5970 }
|
|
316
|
5971 spin.si_region_count = incount;
|
|
236
|
5972
|
|
474
|
5973 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
5974 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
5975 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
5976 if (spin.si_foldroot == NULL
|
|
|
5977 || spin.si_keeproot == NULL
|
|
|
5978 || spin.si_prefroot == NULL)
|
|
300
|
5979 {
|
|
469
|
5980 free_blocks(spin.si_blocks);
|
|
310
|
5981 return;
|
|
300
|
5982 }
|
|
|
5983
|
|
351
|
5984 /* When not producing a .add.spl file clear the character table when
|
|
|
5985 * we encounter one in the .aff file. This means we dump the current
|
|
|
5986 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
5987 * better than guessing the contents, the table will match a
|
|
|
5988 * previously loaded spell file. */
|
|
|
5989 if (!spin.si_add)
|
|
|
5990 spin.si_clear_chartab = TRUE;
|
|
|
5991
|
|
236
|
5992 /*
|
|
|
5993 * Read all the .aff and .dic files.
|
|
|
5994 * Text is converted to 'encoding'.
|
|
300
|
5995 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
5996 */
|
|
310
|
5997 for (i = 0; i < incount && !error; ++i)
|
|
236
|
5998 {
|
|
300
|
5999 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
6000 spin.si_region = 1 << i;
|
|
|
6001
|
|
|
6002 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
6003 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
6004 {
|
|
|
6005 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
6006 * "SET" line. */
|
|
474
|
6007 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
6008 if (afile[i] == NULL)
|
|
300
|
6009 error = TRUE;
|
|
|
6010 else
|
|
|
6011 {
|
|
|
6012 /* Read the .dic file and store the words in the trees. */
|
|
|
6013 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
6014 innames[i]);
|
|
474
|
6015 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
6016 error = TRUE;
|
|
|
6017 }
|
|
|
6018 }
|
|
|
6019 else
|
|
|
6020 {
|
|
|
6021 /* No .aff file, try reading the file as a word list. Store
|
|
|
6022 * the words in the trees. */
|
|
474
|
6023 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
6024 error = TRUE;
|
|
|
6025 }
|
|
236
|
6026
|
|
310
|
6027 #ifdef FEAT_MBYTE
|
|
236
|
6028 /* Free any conversion stuff. */
|
|
300
|
6029 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
6030 #endif
|
|
236
|
6031 }
|
|
|
6032
|
|
300
|
6033 if (!error)
|
|
236
|
6034 {
|
|
|
6035 /*
|
|
300
|
6036 * Combine tails in the tree.
|
|
236
|
6037 */
|
|
474
|
6038 if (spin.si_verbose || p_verbose > 2)
|
|
|
6039 {
|
|
|
6040 if (!spin.si_verbose)
|
|
310
|
6041 verbose_enter();
|
|
469
|
6042 MSG(_(msg_compressing));
|
|
310
|
6043 out_flush();
|
|
474
|
6044 if (!spin.si_verbose)
|
|
310
|
6045 verbose_leave();
|
|
|
6046 }
|
|
474
|
6047 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
6048 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
6049 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
6050 }
|
|
|
6051
|
|
300
|
6052 if (!error)
|
|
|
6053 {
|
|
|
6054 /*
|
|
|
6055 * Write the info in the spell file.
|
|
|
6056 */
|
|
474
|
6057 if (spin.si_verbose || p_verbose > 2)
|
|
|
6058 {
|
|
|
6059 if (!spin.si_verbose)
|
|
310
|
6060 verbose_enter();
|
|
366
|
6061 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
6062 out_flush();
|
|
474
|
6063 if (!spin.si_verbose)
|
|
310
|
6064 verbose_leave();
|
|
|
6065 }
|
|
|
6066
|
|
474
|
6067 write_vim_spell(&spin, wfname);
|
|
|
6068
|
|
|
6069 if (spin.si_verbose || p_verbose > 2)
|
|
|
6070 {
|
|
|
6071 if (!spin.si_verbose)
|
|
310
|
6072 verbose_enter();
|
|
|
6073 MSG(_("Done!"));
|
|
|
6074 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
6075 spin.si_memtot);
|
|
310
|
6076 out_flush();
|
|
474
|
6077 if (!spin.si_verbose)
|
|
310
|
6078 verbose_leave();
|
|
|
6079 }
|
|
|
6080
|
|
|
6081 /* If the file is loaded need to reload it. */
|
|
323
|
6082 spell_reload_one(wfname, added_word);
|
|
300
|
6083 }
|
|
|
6084
|
|
|
6085 /* Free the allocated memory. */
|
|
323
|
6086 ga_clear(&spin.si_rep);
|
|
|
6087 ga_clear(&spin.si_sal);
|
|
|
6088 ga_clear(&spin.si_map);
|
|
339
|
6089 ga_clear(&spin.si_prefcond);
|
|
366
|
6090 vim_free(spin.si_midword);
|
|
375
|
6091 vim_free(spin.si_sofofr);
|
|
|
6092 vim_free(spin.si_sofoto);
|
|
300
|
6093
|
|
|
6094 /* Free the .aff file structures. */
|
|
310
|
6095 for (i = 0; i < incount; ++i)
|
|
|
6096 if (afile[i] != NULL)
|
|
|
6097 spell_free_aff(afile[i]);
|
|
339
|
6098
|
|
|
6099 /* Free all the bits and pieces at once. */
|
|
|
6100 free_blocks(spin.si_blocks);
|
|
236
|
6101 }
|
|
310
|
6102 }
|
|
|
6103
|
|
|
6104
|
|
|
6105 /*
|
|
385
|
6106 * ":[count]spellgood {word}"
|
|
|
6107 * ":[count]spellwrong {word}"
|
|
310
|
6108 */
|
|
|
6109 void
|
|
|
6110 ex_spell(eap)
|
|
|
6111 exarg_T *eap;
|
|
|
6112 {
|
|
381
|
6113 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
6114 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
6115 }
|
|
|
6116
|
|
310
|
6117 /*
|
|
|
6118 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
6119 */
|
|
|
6120 void
|
|
385
|
6121 spell_add_word(word, len, bad, index)
|
|
310
|
6122 char_u *word;
|
|
|
6123 int len;
|
|
|
6124 int bad;
|
|
385
|
6125 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
6126 'spellfile' */
|
|
310
|
6127 {
|
|
|
6128 FILE *fd;
|
|
385
|
6129 buf_T *buf = NULL;
|
|
351
|
6130 int new_spf = FALSE;
|
|
|
6131 struct stat st;
|
|
381
|
6132 char_u *fname;
|
|
385
|
6133 char_u fnamebuf[MAXPATHL];
|
|
|
6134 char_u line[MAXWLEN * 2];
|
|
|
6135 long fpos, fpos_next = 0;
|
|
|
6136 int i;
|
|
|
6137 char_u *spf;
|
|
|
6138
|
|
|
6139 if (index == 0) /* use internal wordlist */
|
|
|
6140 {
|
|
|
6141 if (int_wordlist == NULL)
|
|
|
6142 {
|
|
|
6143 int_wordlist = vim_tempname('s');
|
|
|
6144 if (int_wordlist == NULL)
|
|
381
|
6145 return;
|
|
|
6146 }
|
|
385
|
6147 fname = int_wordlist;
|
|
381
|
6148 }
|
|
310
|
6149 else
|
|
|
6150 {
|
|
381
|
6151 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
6152 if (*curbuf->b_p_spf == NUL)
|
|
|
6153 {
|
|
|
6154 init_spellfile();
|
|
|
6155 new_spf = TRUE;
|
|
|
6156 }
|
|
|
6157
|
|
|
6158 if (*curbuf->b_p_spf == NUL)
|
|
|
6159 {
|
|
|
6160 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
6161 return;
|
|
|
6162 }
|
|
|
6163
|
|
385
|
6164 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
6165 {
|
|
|
6166 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
6167 if (i == index)
|
|
|
6168 break;
|
|
|
6169 if (*spf == NUL)
|
|
|
6170 {
|
|
|
6171 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
6172 return;
|
|
|
6173 }
|
|
|
6174 }
|
|
|
6175
|
|
310
|
6176 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
6177 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
6178 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
6179 buf = NULL;
|
|
|
6180 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
6181 {
|
|
310
|
6182 EMSG(_(e_bufloaded));
|
|
381
|
6183 return;
|
|
|
6184 }
|
|
|
6185
|
|
385
|
6186 fname = fnamebuf;
|
|
|
6187 }
|
|
|
6188
|
|
|
6189 if (bad)
|
|
|
6190 {
|
|
|
6191 /* When the word also appears as good word we need to remove that one,
|
|
|
6192 * since its flags sort before the one with WF_BANNED. */
|
|
|
6193 fd = mch_fopen((char *)fname, "r");
|
|
|
6194 if (fd != NULL)
|
|
|
6195 {
|
|
|
6196 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
6197 {
|
|
|
6198 fpos = fpos_next;
|
|
|
6199 fpos_next = ftell(fd);
|
|
|
6200 if (STRNCMP(word, line, len) == 0
|
|
|
6201 && (line[len] == '/' || line[len] < ' '))
|
|
|
6202 {
|
|
|
6203 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
6204 * the start of the line. Mixing reading and writing
|
|
|
6205 * doesn't work for all systems, close the file first. */
|
|
|
6206 fclose(fd);
|
|
|
6207 fd = mch_fopen((char *)fname, "r+");
|
|
|
6208 if (fd == NULL)
|
|
|
6209 break;
|
|
|
6210 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
6211 fputc('#', fd);
|
|
|
6212 fseek(fd, fpos_next, SEEK_SET);
|
|
|
6213 }
|
|
|
6214 }
|
|
|
6215 fclose(fd);
|
|
|
6216 }
|
|
381
|
6217 }
|
|
|
6218
|
|
|
6219 fd = mch_fopen((char *)fname, "a");
|
|
|
6220 if (fd == NULL && new_spf)
|
|
|
6221 {
|
|
|
6222 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
6223 * We may need to create the "spell" directory first. We already
|
|
|
6224 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
6225 STRCPY(NameBuff, fname);
|
|
|
6226 *gettail_sep(NameBuff) = NUL;
|
|
|
6227 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
6228 {
|
|
|
6229 /* The directory doesn't exist. Try creating it and opening the
|
|
|
6230 * file again. */
|
|
|
6231 vim_mkdir(NameBuff, 0755);
|
|
|
6232 fd = mch_fopen((char *)fname, "a");
|
|
|
6233 }
|
|
|
6234 }
|
|
|
6235
|
|
|
6236 if (fd == NULL)
|
|
|
6237 EMSG2(_(e_notopen), fname);
|
|
|
6238 else
|
|
|
6239 {
|
|
|
6240 if (bad)
|
|
|
6241 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
6242 else
|
|
381
|
6243 fprintf(fd, "%.*s\n", len, word);
|
|
|
6244 fclose(fd);
|
|
|
6245
|
|
|
6246 /* Update the .add.spl file. */
|
|
|
6247 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
6248
|
|
|
6249 /* If the .add file is edited somewhere, reload it. */
|
|
|
6250 if (buf != NULL)
|
|
|
6251 buf_reload(buf);
|
|
|
6252
|
|
|
6253 redraw_all_later(NOT_VALID);
|
|
310
|
6254 }
|
|
|
6255 }
|
|
|
6256
|
|
|
6257 /*
|
|
|
6258 * Initialize 'spellfile' for the current buffer.
|
|
|
6259 */
|
|
|
6260 static void
|
|
|
6261 init_spellfile()
|
|
|
6262 {
|
|
|
6263 char_u buf[MAXPATHL];
|
|
|
6264 int l;
|
|
|
6265 slang_T *sl;
|
|
|
6266 char_u *rtp;
|
|
351
|
6267 char_u *lend;
|
|
310
|
6268
|
|
|
6269 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
6270 {
|
|
351
|
6271 /* Find the end of the language name. Exclude the region. */
|
|
|
6272 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
6273 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
6274 ;
|
|
|
6275
|
|
|
6276 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
6277 * are allowed to write. */
|
|
310
|
6278 rtp = p_rtp;
|
|
|
6279 while (*rtp != NUL)
|
|
|
6280 {
|
|
|
6281 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
6282 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
6283 if (filewritable(buf) == 2)
|
|
|
6284 {
|
|
316
|
6285 /* Use the first language name from 'spelllang' and the
|
|
|
6286 * encoding used in the first loaded .spl file. */
|
|
310
|
6287 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
6288 l = STRLEN(buf);
|
|
|
6289 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
6290 "/spell/%.*s.%s.add",
|
|
351
|
6291 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
6292 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
6293 ? (char_u *)"ascii" : spell_enc());
|
|
|
6294 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
6295 break;
|
|
|
6296 }
|
|
|
6297 }
|
|
|
6298 }
|
|
|
6299 }
|
|
236
|
6300
|
|
300
|
6301
|
|
307
|
6302 /*
|
|
|
6303 * Init the chartab used for spelling for ASCII.
|
|
|
6304 * EBCDIC is not supported!
|
|
|
6305 */
|
|
|
6306 static void
|
|
|
6307 clear_spell_chartab(sp)
|
|
|
6308 spelltab_T *sp;
|
|
|
6309 {
|
|
324
|
6310 int i;
|
|
307
|
6311
|
|
|
6312 /* Init everything to FALSE. */
|
|
|
6313 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
6314 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
6315 for (i = 0; i < 256; ++i)
|
|
324
|
6316 {
|
|
307
|
6317 sp->st_fold[i] = i;
|
|
324
|
6318 sp->st_upper[i] = i;
|
|
|
6319 }
|
|
307
|
6320
|
|
|
6321 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
6322 * that is done separately. */
|
|
|
6323 for (i = '0'; i <= '9'; ++i)
|
|
|
6324 sp->st_isw[i] = TRUE;
|
|
|
6325 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
6326 {
|
|
|
6327 sp->st_isw[i] = TRUE;
|
|
|
6328 sp->st_isu[i] = TRUE;
|
|
|
6329 sp->st_fold[i] = i + 0x20;
|
|
|
6330 }
|
|
|
6331 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
6332 {
|
|
307
|
6333 sp->st_isw[i] = TRUE;
|
|
324
|
6334 sp->st_upper[i] = i - 0x20;
|
|
|
6335 }
|
|
307
|
6336 }
|
|
|
6337
|
|
|
6338 /*
|
|
|
6339 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
6340 * Called once while starting up and when 'encoding' changes.
|
|
|
6341 * The default is to use isalpha(), but the spell file should define the word
|
|
|
6342 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
6343 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
6344 */
|
|
|
6345 void
|
|
|
6346 init_spell_chartab()
|
|
|
6347 {
|
|
|
6348 int i;
|
|
|
6349
|
|
|
6350 did_set_spelltab = FALSE;
|
|
|
6351 clear_spell_chartab(&spelltab);
|
|
|
6352 #ifdef FEAT_MBYTE
|
|
|
6353 if (enc_dbcs)
|
|
|
6354 {
|
|
|
6355 /* DBCS: assume double-wide characters are word characters. */
|
|
|
6356 for (i = 128; i <= 255; ++i)
|
|
|
6357 if (MB_BYTE2LEN(i) == 2)
|
|
|
6358 spelltab.st_isw[i] = TRUE;
|
|
|
6359 }
|
|
324
|
6360 else if (enc_utf8)
|
|
|
6361 {
|
|
|
6362 for (i = 128; i < 256; ++i)
|
|
|
6363 {
|
|
|
6364 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
6365 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
6366 spelltab.st_fold[i] = utf_fold(i);
|
|
|
6367 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
6368 }
|
|
|
6369 }
|
|
307
|
6370 else
|
|
|
6371 #endif
|
|
|
6372 {
|
|
324
|
6373 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
6374 for (i = 128; i < 256; ++i)
|
|
|
6375 {
|
|
|
6376 if (MB_ISUPPER(i))
|
|
|
6377 {
|
|
324
|
6378 spelltab.st_isw[i] = TRUE;
|
|
307
|
6379 spelltab.st_isu[i] = TRUE;
|
|
|
6380 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
6381 }
|
|
324
|
6382 else if (MB_ISLOWER(i))
|
|
|
6383 {
|
|
|
6384 spelltab.st_isw[i] = TRUE;
|
|
|
6385 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
6386 }
|
|
307
|
6387 }
|
|
|
6388 }
|
|
|
6389 }
|
|
|
6390
|
|
|
6391 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
6392 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
6393
|
|
|
6394 /*
|
|
|
6395 * Set the spell character tables from strings in the affix file.
|
|
|
6396 */
|
|
|
6397 static int
|
|
|
6398 set_spell_chartab(fol, low, upp)
|
|
|
6399 char_u *fol;
|
|
|
6400 char_u *low;
|
|
|
6401 char_u *upp;
|
|
|
6402 {
|
|
|
6403 /* We build the new tables here first, so that we can compare with the
|
|
|
6404 * previous one. */
|
|
|
6405 spelltab_T new_st;
|
|
|
6406 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
6407 int f, l, u;
|
|
|
6408
|
|
|
6409 clear_spell_chartab(&new_st);
|
|
|
6410
|
|
|
6411 while (*pf != NUL)
|
|
|
6412 {
|
|
|
6413 if (*pl == NUL || *pu == NUL)
|
|
|
6414 {
|
|
|
6415 EMSG(_(e_affform));
|
|
|
6416 return FAIL;
|
|
|
6417 }
|
|
|
6418 #ifdef FEAT_MBYTE
|
|
|
6419 f = mb_ptr2char_adv(&pf);
|
|
|
6420 l = mb_ptr2char_adv(&pl);
|
|
|
6421 u = mb_ptr2char_adv(&pu);
|
|
|
6422 #else
|
|
|
6423 f = *pf++;
|
|
|
6424 l = *pl++;
|
|
|
6425 u = *pu++;
|
|
|
6426 #endif
|
|
|
6427 /* Every character that appears is a word character. */
|
|
|
6428 if (f < 256)
|
|
|
6429 new_st.st_isw[f] = TRUE;
|
|
|
6430 if (l < 256)
|
|
|
6431 new_st.st_isw[l] = TRUE;
|
|
|
6432 if (u < 256)
|
|
|
6433 new_st.st_isw[u] = TRUE;
|
|
|
6434
|
|
|
6435 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
6436 * case-folding */
|
|
|
6437 if (l < 256 && l != f)
|
|
|
6438 {
|
|
|
6439 if (f >= 256)
|
|
|
6440 {
|
|
|
6441 EMSG(_(e_affrange));
|
|
|
6442 return FAIL;
|
|
|
6443 }
|
|
|
6444 new_st.st_fold[l] = f;
|
|
|
6445 }
|
|
|
6446
|
|
|
6447 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
6448 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
6449 * "FOL" . */
|
|
307
|
6450 if (u < 256 && u != f)
|
|
|
6451 {
|
|
|
6452 if (f >= 256)
|
|
|
6453 {
|
|
|
6454 EMSG(_(e_affrange));
|
|
|
6455 return FAIL;
|
|
|
6456 }
|
|
|
6457 new_st.st_fold[u] = f;
|
|
|
6458 new_st.st_isu[u] = TRUE;
|
|
324
|
6459 new_st.st_upper[f] = u;
|
|
307
|
6460 }
|
|
|
6461 }
|
|
|
6462
|
|
|
6463 if (*pl != NUL || *pu != NUL)
|
|
|
6464 {
|
|
|
6465 EMSG(_(e_affform));
|
|
|
6466 return FAIL;
|
|
|
6467 }
|
|
|
6468
|
|
|
6469 return set_spell_finish(&new_st);
|
|
|
6470 }
|
|
|
6471
|
|
|
6472 /*
|
|
|
6473 * Set the spell character tables from strings in the .spl file.
|
|
|
6474 */
|
|
|
6475 static int
|
|
389
|
6476 set_spell_charflags(flags, cnt, fol)
|
|
307
|
6477 char_u *flags;
|
|
389
|
6478 int cnt; /* length of "flags" */
|
|
|
6479 char_u *fol;
|
|
307
|
6480 {
|
|
|
6481 /* We build the new tables here first, so that we can compare with the
|
|
|
6482 * previous one. */
|
|
|
6483 spelltab_T new_st;
|
|
|
6484 int i;
|
|
389
|
6485 char_u *p = fol;
|
|
324
|
6486 int c;
|
|
307
|
6487
|
|
|
6488 clear_spell_chartab(&new_st);
|
|
|
6489
|
|
389
|
6490 for (i = 0; i < 128; ++i)
|
|
|
6491 {
|
|
|
6492 if (i < cnt)
|
|
|
6493 {
|
|
|
6494 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
6495 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
6496 }
|
|
|
6497
|
|
|
6498 if (*p != NUL)
|
|
|
6499 {
|
|
307
|
6500 #ifdef FEAT_MBYTE
|
|
389
|
6501 c = mb_ptr2char_adv(&p);
|
|
307
|
6502 #else
|
|
389
|
6503 c = *p++;
|
|
307
|
6504 #endif
|
|
389
|
6505 new_st.st_fold[i + 128] = c;
|
|
|
6506 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
6507 new_st.st_upper[c] = i + 128;
|
|
|
6508 }
|
|
307
|
6509 }
|
|
|
6510
|
|
|
6511 return set_spell_finish(&new_st);
|
|
|
6512 }
|
|
|
6513
|
|
|
6514 static int
|
|
|
6515 set_spell_finish(new_st)
|
|
|
6516 spelltab_T *new_st;
|
|
|
6517 {
|
|
|
6518 int i;
|
|
|
6519
|
|
|
6520 if (did_set_spelltab)
|
|
|
6521 {
|
|
|
6522 /* check that it's the same table */
|
|
|
6523 for (i = 0; i < 256; ++i)
|
|
|
6524 {
|
|
|
6525 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
6526 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
6527 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
6528 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
6529 {
|
|
|
6530 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
6531 return FAIL;
|
|
|
6532 }
|
|
|
6533 }
|
|
|
6534 }
|
|
|
6535 else
|
|
|
6536 {
|
|
|
6537 /* copy the new spelltab into the one being used */
|
|
|
6538 spelltab = *new_st;
|
|
|
6539 did_set_spelltab = TRUE;
|
|
|
6540 }
|
|
|
6541
|
|
|
6542 return OK;
|
|
|
6543 }
|
|
|
6544
|
|
|
6545 /*
|
|
358
|
6546 * Return TRUE if "p" points to a word character.
|
|
366
|
6547 * As a special case we see "midword" characters as word character when it is
|
|
358
|
6548 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
6549 * Thus this only works properly when past the first character of the word.
|
|
358
|
6550 */
|
|
|
6551 static int
|
|
376
|
6552 spell_iswordp(p, buf)
|
|
358
|
6553 char_u *p;
|
|
376
|
6554 buf_T *buf; /* buffer used */
|
|
358
|
6555 {
|
|
366
|
6556 #ifdef FEAT_MBYTE
|
|
358
|
6557 char_u *s;
|
|
366
|
6558 int l;
|
|
|
6559 int c;
|
|
|
6560
|
|
|
6561 if (has_mbyte)
|
|
|
6562 {
|
|
|
6563 l = MB_BYTE2LEN(*p);
|
|
358
|
6564 s = p;
|
|
366
|
6565 if (l == 1)
|
|
|
6566 {
|
|
|
6567 /* be quick for ASCII */
|
|
376
|
6568 if (buf->b_spell_ismw[*p])
|
|
366
|
6569 {
|
|
|
6570 s = p + 1; /* skip a mid-word character */
|
|
|
6571 l = MB_BYTE2LEN(*s);
|
|
|
6572 }
|
|
|
6573 }
|
|
|
6574 else
|
|
|
6575 {
|
|
|
6576 c = mb_ptr2char(p);
|
|
376
|
6577 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
6578 : (buf->b_spell_ismw_mb != NULL
|
|
|
6579 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
6580 {
|
|
|
6581 s = p + l;
|
|
|
6582 l = MB_BYTE2LEN(*s);
|
|
|
6583 }
|
|
|
6584 }
|
|
|
6585
|
|
390
|
6586 c = mb_ptr2char(s);
|
|
|
6587 if (c > 255)
|
|
366
|
6588 return mb_get_class(s) >= 2;
|
|
390
|
6589 return spelltab.st_isw[c];
|
|
366
|
6590 }
|
|
358
|
6591 #endif
|
|
366
|
6592
|
|
376
|
6593 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
6594 }
|
|
|
6595
|
|
|
6596 /*
|
|
|
6597 * Return TRUE if "p" points to a word character.
|
|
|
6598 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
6599 */
|
|
|
6600 static int
|
|
|
6601 spell_iswordp_nmw(p)
|
|
|
6602 char_u *p;
|
|
|
6603 {
|
|
|
6604 #ifdef FEAT_MBYTE
|
|
390
|
6605 int c;
|
|
|
6606
|
|
|
6607 if (has_mbyte)
|
|
|
6608 {
|
|
|
6609 c = mb_ptr2char(p);
|
|
|
6610 if (c > 255)
|
|
|
6611 return mb_get_class(p) >= 2;
|
|
|
6612 return spelltab.st_isw[c];
|
|
|
6613 }
|
|
376
|
6614 #endif
|
|
|
6615 return spelltab.st_isw[*p];
|
|
358
|
6616 }
|
|
|
6617
|
|
372
|
6618 #ifdef FEAT_MBYTE
|
|
|
6619 /*
|
|
|
6620 * Return TRUE if "p" points to a word character.
|
|
|
6621 * Wide version of spell_iswordp().
|
|
|
6622 */
|
|
|
6623 static int
|
|
376
|
6624 spell_iswordp_w(p, buf)
|
|
372
|
6625 int *p;
|
|
376
|
6626 buf_T *buf;
|
|
372
|
6627 {
|
|
|
6628 int *s;
|
|
|
6629
|
|
376
|
6630 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
6631 : (buf->b_spell_ismw_mb != NULL
|
|
|
6632 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
6633 s = p + 1;
|
|
|
6634 else
|
|
|
6635 s = p;
|
|
|
6636
|
|
390
|
6637 if (*s > 255)
|
|
372
|
6638 {
|
|
|
6639 if (enc_utf8)
|
|
|
6640 return utf_class(*s) >= 2;
|
|
|
6641 if (enc_dbcs)
|
|
|
6642 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
6643 return 0;
|
|
|
6644 }
|
|
|
6645 return spelltab.st_isw[*s];
|
|
|
6646 }
|
|
|
6647 #endif
|
|
|
6648
|
|
358
|
6649 /*
|
|
339
|
6650 * Write the table with prefix conditions to the .spl file.
|
|
|
6651 */
|
|
|
6652 static void
|
|
|
6653 write_spell_prefcond(fd, gap)
|
|
|
6654 FILE *fd;
|
|
|
6655 garray_T *gap;
|
|
|
6656 {
|
|
|
6657 int i;
|
|
|
6658 char_u *p;
|
|
|
6659 int len;
|
|
|
6660
|
|
|
6661 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
6662
|
|
|
6663 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6664 {
|
|
|
6665 /* <prefcond> : <condlen> <condstr> */
|
|
|
6666 p = ((char_u **)gap->ga_data)[i];
|
|
|
6667 if (p == NULL)
|
|
|
6668 fputc(0, fd);
|
|
|
6669 else
|
|
|
6670 {
|
|
|
6671 len = STRLEN(p);
|
|
|
6672 fputc(len, fd);
|
|
|
6673 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
6674 }
|
|
|
6675 }
|
|
|
6676 }
|
|
|
6677
|
|
|
6678 /*
|
|
307
|
6679 * Write the current tables into the .spl file.
|
|
|
6680 * This makes sure the same characters are recognized as word characters when
|
|
|
6681 * generating an when using a spell file.
|
|
|
6682 */
|
|
|
6683 static void
|
|
|
6684 write_spell_chartab(fd)
|
|
|
6685 FILE *fd;
|
|
|
6686 {
|
|
|
6687 char_u charbuf[256 * 4];
|
|
|
6688 int len = 0;
|
|
|
6689 int flags;
|
|
|
6690 int i;
|
|
|
6691
|
|
|
6692 fputc(128, fd); /* <charflagslen> */
|
|
|
6693 for (i = 128; i < 256; ++i)
|
|
|
6694 {
|
|
|
6695 flags = 0;
|
|
|
6696 if (spelltab.st_isw[i])
|
|
324
|
6697 flags |= CF_WORD;
|
|
307
|
6698 if (spelltab.st_isu[i])
|
|
324
|
6699 flags |= CF_UPPER;
|
|
307
|
6700 fputc(flags, fd); /* <charflags> */
|
|
|
6701
|
|
310
|
6702 #ifdef FEAT_MBYTE
|
|
|
6703 if (has_mbyte)
|
|
|
6704 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
|
6705 else
|
|
|
6706 #endif
|
|
|
6707 charbuf[len++] = spelltab.st_fold[i];
|
|
307
|
6708 }
|
|
|
6709
|
|
|
6710 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
|
6711 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
|
6712 }
|
|
|
6713
|
|
|
6714 /*
|
|
324
|
6715 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
6716 * Uses the character definitions from the .spl file.
|
|
307
|
6717 * When using a multi-byte 'encoding' the length may change!
|
|
|
6718 * Returns FAIL when something wrong.
|
|
|
6719 */
|
|
|
6720 static int
|
|
324
|
6721 spell_casefold(str, len, buf, buflen)
|
|
|
6722 char_u *str;
|
|
307
|
6723 int len;
|
|
|
6724 char_u *buf;
|
|
|
6725 int buflen;
|
|
|
6726 {
|
|
|
6727 int i;
|
|
|
6728
|
|
|
6729 if (len >= buflen)
|
|
|
6730 {
|
|
|
6731 buf[0] = NUL;
|
|
|
6732 return FAIL; /* result will not fit */
|
|
|
6733 }
|
|
|
6734
|
|
|
6735 #ifdef FEAT_MBYTE
|
|
|
6736 if (has_mbyte)
|
|
|
6737 {
|
|
324
|
6738 int outi = 0;
|
|
|
6739 char_u *p;
|
|
307
|
6740 int c;
|
|
|
6741
|
|
|
6742 /* Fold one character at a time. */
|
|
324
|
6743 for (p = str; p < str + len; )
|
|
307
|
6744 {
|
|
|
6745 if (outi + MB_MAXBYTES > buflen)
|
|
|
6746 {
|
|
|
6747 buf[outi] = NUL;
|
|
|
6748 return FAIL;
|
|
|
6749 }
|
|
474
|
6750 c = mb_cptr2char_adv(&p);
|
|
324
|
6751 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
6752 }
|
|
|
6753 buf[outi] = NUL;
|
|
|
6754 }
|
|
|
6755 else
|
|
|
6756 #endif
|
|
|
6757 {
|
|
|
6758 /* Be quick for non-multibyte encodings. */
|
|
|
6759 for (i = 0; i < len; ++i)
|
|
324
|
6760 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
6761 buf[i] = NUL;
|
|
|
6762 }
|
|
|
6763
|
|
|
6764 return OK;
|
|
|
6765 }
|
|
|
6766
|
|
372
|
6767 #define SPS_BEST 1
|
|
|
6768 #define SPS_FAST 2
|
|
|
6769 #define SPS_DOUBLE 4
|
|
|
6770
|
|
|
6771 static int sps_flags = SPS_BEST;
|
|
|
6772
|
|
|
6773 /*
|
|
|
6774 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
|
6775 * Sets "sps_flags".
|
|
|
6776 */
|
|
|
6777 int
|
|
|
6778 spell_check_sps()
|
|
|
6779 {
|
|
|
6780 char_u *p;
|
|
|
6781 char_u buf[MAXPATHL];
|
|
|
6782 int f;
|
|
|
6783
|
|
|
6784 sps_flags = 0;
|
|
|
6785
|
|
|
6786 for (p = p_sps; *p != NUL; )
|
|
|
6787 {
|
|
|
6788 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6789
|
|
|
6790 f = 0;
|
|
|
6791 if (STRCMP(buf, "best") == 0)
|
|
|
6792 f = SPS_BEST;
|
|
|
6793 else if (STRCMP(buf, "fast") == 0)
|
|
|
6794 f = SPS_FAST;
|
|
|
6795 else if (STRCMP(buf, "double") == 0)
|
|
|
6796 f = SPS_DOUBLE;
|
|
|
6797 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
6798 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
6799 f = -1;
|
|
|
6800
|
|
|
6801 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
6802 {
|
|
|
6803 sps_flags = SPS_BEST;
|
|
|
6804 return FAIL;
|
|
|
6805 }
|
|
|
6806 if (f != 0)
|
|
|
6807 sps_flags = f;
|
|
|
6808 }
|
|
|
6809
|
|
|
6810 if (sps_flags == 0)
|
|
|
6811 sps_flags = SPS_BEST;
|
|
|
6812
|
|
|
6813 return OK;
|
|
|
6814 }
|
|
|
6815
|
|
|
6816 /* Remember what "z?" replaced. */
|
|
|
6817 static char_u *repl_from = NULL;
|
|
|
6818 static char_u *repl_to = NULL;
|
|
|
6819
|
|
323
|
6820 /*
|
|
|
6821 * "z?": Find badly spelled word under or after the cursor.
|
|
|
6822 * Give suggestions for the properly spelled word.
|
|
|
6823 */
|
|
|
6824 void
|
|
|
6825 spell_suggest()
|
|
|
6826 {
|
|
|
6827 char_u *line;
|
|
|
6828 pos_T prev_cursor = curwin->w_cursor;
|
|
|
6829 char_u wcopy[MAXWLEN + 2];
|
|
|
6830 char_u *p;
|
|
|
6831 int i;
|
|
|
6832 int c;
|
|
|
6833 suginfo_T sug;
|
|
|
6834 suggest_T *stp;
|
|
372
|
6835 int mouse_used;
|
|
388
|
6836 int need_cap;
|
|
323
|
6837
|
|
344
|
6838 /* Find the start of the badly spelled word. */
|
|
346
|
6839 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
6840 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
6841 {
|
|
|
6842 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
6843 return;
|
|
|
6844
|
|
|
6845 /* No bad word or it starts after the cursor: use the word under the
|
|
|
6846 * cursor. */
|
|
|
6847 curwin->w_cursor = prev_cursor;
|
|
|
6848 line = ml_get_curline();
|
|
|
6849 p = line + curwin->w_cursor.col;
|
|
|
6850 /* Backup to before start of word. */
|
|
390
|
6851 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
6852 mb_ptr_back(line, p);
|
|
|
6853 /* Forward to start of word. */
|
|
390
|
6854 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
6855 mb_ptr_adv(p);
|
|
|
6856
|
|
390
|
6857 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
6858 {
|
|
|
6859 beep_flush();
|
|
|
6860 return;
|
|
|
6861 }
|
|
|
6862 curwin->w_cursor.col = p - line;
|
|
323
|
6863 }
|
|
|
6864
|
|
344
|
6865 /* Get the word and its length. */
|
|
|
6866
|
|
388
|
6867 /* Figure out if the word should be capitalised. */
|
|
475
|
6868 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
6869
|
|
|
6870 line = ml_get_curline();
|
|
388
|
6871
|
|
344
|
6872 /* Get the list of suggestions */
|
|
388
|
6873 spell_find_suggest(line + curwin->w_cursor.col, &sug, (int)Rows - 2,
|
|
|
6874 TRUE, need_cap);
|
|
323
|
6875
|
|
|
6876 if (sug.su_ga.ga_len == 0)
|
|
|
6877 MSG(_("Sorry, no suggestions"));
|
|
|
6878 else
|
|
|
6879 {
|
|
372
|
6880 vim_free(repl_from);
|
|
|
6881 repl_from = NULL;
|
|
|
6882 vim_free(repl_to);
|
|
|
6883 repl_to = NULL;
|
|
|
6884
|
|
474
|
6885 #ifdef FEAT_RIGHTLEFT
|
|
|
6886 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
6887 cmdmsg_rl = curwin->w_p_rl;
|
|
|
6888 if (cmdmsg_rl)
|
|
|
6889 msg_col = Columns - 1;
|
|
|
6890 #endif
|
|
|
6891
|
|
323
|
6892 /* List the suggestions. */
|
|
|
6893 msg_start();
|
|
372
|
6894 lines_left = Rows; /* avoid more prompt */
|
|
323
|
6895 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
6896 sug.su_badlen, sug.su_badptr);
|
|
474
|
6897 #ifdef FEAT_RIGHTLEFT
|
|
|
6898 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
6899 {
|
|
|
6900 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
6901 * untranslated message rightleft. */
|
|
|
6902 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
6903 sug.su_badlen, sug.su_badptr);
|
|
|
6904 }
|
|
|
6905 #endif
|
|
323
|
6906 msg_puts(IObuff);
|
|
|
6907 msg_clr_eos();
|
|
|
6908 msg_putchar('\n');
|
|
346
|
6909
|
|
323
|
6910 msg_scroll = TRUE;
|
|
|
6911 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
6912 {
|
|
344
|
6913 stp = &SUG(sug.su_ga, i);
|
|
323
|
6914
|
|
|
6915 /* The suggested word may replace only part of the bad word, add
|
|
|
6916 * the not replaced part. */
|
|
|
6917 STRCPY(wcopy, stp->st_word);
|
|
|
6918 if (sug.su_badlen > stp->st_orglen)
|
|
|
6919 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
6920 sug.su_badptr + stp->st_orglen,
|
|
|
6921 sug.su_badlen - stp->st_orglen);
|
|
474
|
6922 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
6923 #ifdef FEAT_RIGHTLEFT
|
|
|
6924 if (cmdmsg_rl)
|
|
|
6925 rl_mirror(IObuff);
|
|
|
6926 #endif
|
|
|
6927 msg_puts(IObuff);
|
|
|
6928
|
|
|
6929 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
6930 msg_puts(IObuff);
|
|
|
6931
|
|
|
6932 /* The word may replace more than "su_badlen". */
|
|
|
6933 if (sug.su_badlen < stp->st_orglen)
|
|
|
6934 {
|
|
|
6935 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
6936 stp->st_orglen, sug.su_badptr);
|
|
|
6937 msg_puts(IObuff);
|
|
|
6938 }
|
|
|
6939
|
|
324
|
6940 if (p_verbose > 0)
|
|
344
|
6941 {
|
|
346
|
6942 /* Add the score. */
|
|
351
|
6943 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
6944 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
6945 stp->st_salscore ? "s " : "",
|
|
|
6946 stp->st_score, stp->st_altscore);
|
|
|
6947 else
|
|
474
|
6948 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
6949 stp->st_score);
|
|
474
|
6950 #ifdef FEAT_RIGHTLEFT
|
|
|
6951 if (cmdmsg_rl)
|
|
|
6952 /* Mirror the numbers, but keep the leading space. */
|
|
|
6953 rl_mirror(IObuff + 1);
|
|
|
6954 #endif
|
|
346
|
6955 msg_advance(30);
|
|
|
6956 msg_puts(IObuff);
|
|
344
|
6957 }
|
|
323
|
6958 msg_putchar('\n');
|
|
|
6959 }
|
|
|
6960
|
|
474
|
6961 #ifdef FEAT_RIGHTLEFT
|
|
|
6962 cmdmsg_rl = FALSE;
|
|
|
6963 msg_col = 0;
|
|
|
6964 #endif
|
|
323
|
6965 /* Ask for choice. */
|
|
372
|
6966 i = prompt_for_number(&mouse_used);
|
|
|
6967 if (mouse_used)
|
|
|
6968 i -= lines_left;
|
|
|
6969
|
|
344
|
6970 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
6971 {
|
|
372
|
6972 /* Save the from and to text for :spellrepall. */
|
|
|
6973 stp = &SUG(sug.su_ga, i - 1);
|
|
|
6974 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
6975 repl_to = vim_strsave(stp->st_word);
|
|
|
6976
|
|
323
|
6977 /* Replace the word. */
|
|
|
6978 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
6979 if (p != NULL)
|
|
|
6980 {
|
|
|
6981 c = sug.su_badptr - line;
|
|
|
6982 mch_memmove(p, line, c);
|
|
|
6983 STRCPY(p + c, stp->st_word);
|
|
|
6984 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
6985 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6986 curwin->w_cursor.col = c;
|
|
|
6987 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
6988
|
|
|
6989 /* For redo we use a change-word command. */
|
|
|
6990 ResetRedobuff();
|
|
|
6991 AppendToRedobuff((char_u *)"ciw");
|
|
|
6992 AppendToRedobuff(stp->st_word);
|
|
|
6993 AppendCharToRedobuff(ESC);
|
|
323
|
6994 }
|
|
|
6995 }
|
|
|
6996 else
|
|
|
6997 curwin->w_cursor = prev_cursor;
|
|
|
6998 }
|
|
|
6999
|
|
344
|
7000 spell_find_cleanup(&sug);
|
|
|
7001 }
|
|
|
7002
|
|
|
7003 /*
|
|
475
|
7004 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
7005 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
7006 */
|
|
|
7007 static int
|
|
|
7008 check_need_cap(lnum, col)
|
|
|
7009 linenr_T lnum;
|
|
|
7010 colnr_T col;
|
|
|
7011 {
|
|
|
7012 int need_cap = FALSE;
|
|
|
7013 char_u *line;
|
|
|
7014 char_u *line_copy = NULL;
|
|
|
7015 char_u *p;
|
|
|
7016 colnr_T endcol;
|
|
|
7017 regmatch_T regmatch;
|
|
|
7018
|
|
|
7019 if (curbuf->b_cap_prog == NULL)
|
|
|
7020 return FALSE;
|
|
|
7021
|
|
|
7022 line = ml_get_curline();
|
|
|
7023 endcol = 0;
|
|
|
7024 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
7025 {
|
|
|
7026 /* At start of line, check if previous line is empty or sentence
|
|
|
7027 * ends there. */
|
|
|
7028 if (lnum == 1)
|
|
|
7029 need_cap = TRUE;
|
|
|
7030 else
|
|
|
7031 {
|
|
|
7032 line = ml_get(lnum - 1);
|
|
|
7033 if (*skipwhite(line) == NUL)
|
|
|
7034 need_cap = TRUE;
|
|
|
7035 else
|
|
|
7036 {
|
|
|
7037 /* Append a space in place of the line break. */
|
|
|
7038 line_copy = concat_str(line, (char_u *)" ");
|
|
|
7039 line = line_copy;
|
|
|
7040 endcol = STRLEN(line);
|
|
|
7041 }
|
|
|
7042 }
|
|
|
7043 }
|
|
|
7044 else
|
|
|
7045 endcol = col;
|
|
|
7046
|
|
|
7047 if (endcol > 0)
|
|
|
7048 {
|
|
|
7049 /* Check if sentence ends before the bad word. */
|
|
|
7050 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
7051 regmatch.rm_ic = FALSE;
|
|
|
7052 p = line + endcol;
|
|
|
7053 for (;;)
|
|
|
7054 {
|
|
|
7055 mb_ptr_back(line, p);
|
|
|
7056 if (p == line || spell_iswordp_nmw(p))
|
|
|
7057 break;
|
|
|
7058 if (vim_regexec(®match, p, 0)
|
|
|
7059 && regmatch.endp[0] == line + endcol)
|
|
|
7060 {
|
|
|
7061 need_cap = TRUE;
|
|
|
7062 break;
|
|
|
7063 }
|
|
|
7064 }
|
|
|
7065 }
|
|
|
7066
|
|
|
7067 vim_free(line_copy);
|
|
|
7068
|
|
|
7069 return need_cap;
|
|
|
7070 }
|
|
|
7071
|
|
|
7072
|
|
|
7073 /*
|
|
372
|
7074 * ":spellrepall"
|
|
|
7075 */
|
|
|
7076 /*ARGSUSED*/
|
|
|
7077 void
|
|
|
7078 ex_spellrepall(eap)
|
|
|
7079 exarg_T *eap;
|
|
|
7080 {
|
|
|
7081 pos_T pos = curwin->w_cursor;
|
|
|
7082 char_u *frompat;
|
|
|
7083 int addlen;
|
|
|
7084 char_u *line;
|
|
|
7085 char_u *p;
|
|
|
7086 int didone = FALSE;
|
|
|
7087 int save_ws = p_ws;
|
|
|
7088
|
|
|
7089 if (repl_from == NULL || repl_to == NULL)
|
|
|
7090 {
|
|
|
7091 EMSG(_("E752: No previous spell replacement"));
|
|
|
7092 return;
|
|
|
7093 }
|
|
|
7094 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
7095
|
|
|
7096 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
7097 if (frompat == NULL)
|
|
|
7098 return;
|
|
|
7099 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
7100 p_ws = FALSE;
|
|
|
7101
|
|
|
7102 curwin->w_cursor.lnum = 0;
|
|
|
7103 while (!got_int)
|
|
|
7104 {
|
|
|
7105 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
7106 || u_save_cursor() == FAIL)
|
|
|
7107 break;
|
|
|
7108
|
|
|
7109 /* Only replace when the right word isn't there yet. This happens
|
|
|
7110 * when changing "etc" to "etc.". */
|
|
|
7111 line = ml_get_curline();
|
|
|
7112 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
7113 repl_to, STRLEN(repl_to)) != 0)
|
|
|
7114 {
|
|
|
7115 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
7116 if (p == NULL)
|
|
|
7117 break;
|
|
|
7118 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
7119 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
7120 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
7121 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
7122 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
7123 didone = TRUE;
|
|
|
7124 }
|
|
|
7125 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
7126 }
|
|
|
7127
|
|
|
7128 p_ws = save_ws;
|
|
|
7129 curwin->w_cursor = pos;
|
|
|
7130 vim_free(frompat);
|
|
|
7131
|
|
|
7132 if (!didone)
|
|
|
7133 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
|
7134 }
|
|
|
7135
|
|
|
7136 /*
|
|
344
|
7137 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
7138 * a list of allocated strings.
|
|
|
7139 */
|
|
|
7140 void
|
|
475
|
7141 spell_suggest_list(gap, word, maxcount, need_cap)
|
|
344
|
7142 garray_T *gap;
|
|
|
7143 char_u *word;
|
|
|
7144 int maxcount; /* maximum nr of suggestions */
|
|
475
|
7145 int need_cap; /* 'spellcapcheck' matched */
|
|
344
|
7146 {
|
|
|
7147 suginfo_T sug;
|
|
|
7148 int i;
|
|
|
7149 suggest_T *stp;
|
|
|
7150 char_u *wcopy;
|
|
|
7151
|
|
475
|
7152 spell_find_suggest(word, &sug, maxcount, FALSE, need_cap);
|
|
344
|
7153
|
|
|
7154 /* Make room in "gap". */
|
|
|
7155 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
7156 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
7157 return;
|
|
|
7158
|
|
323
|
7159 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
7160 {
|
|
|
7161 stp = &SUG(sug.su_ga, i);
|
|
|
7162
|
|
|
7163 /* The suggested word may replace only part of "word", add the not
|
|
|
7164 * replaced part. */
|
|
|
7165 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
7166 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
7167 if (wcopy == NULL)
|
|
|
7168 break;
|
|
|
7169 STRCPY(wcopy, stp->st_word);
|
|
|
7170 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
7171 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
7172 }
|
|
|
7173
|
|
|
7174 spell_find_cleanup(&sug);
|
|
|
7175 }
|
|
|
7176
|
|
|
7177 /*
|
|
|
7178 * Find spell suggestions for the word at the start of "badptr".
|
|
|
7179 * Return the suggestions in "su->su_ga".
|
|
|
7180 * The maximum number of suggestions is "maxcount".
|
|
|
7181 * Note: does use info for the current window.
|
|
|
7182 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
7183 */
|
|
|
7184 static void
|
|
388
|
7185 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
7186 char_u *badptr;
|
|
|
7187 suginfo_T *su;
|
|
|
7188 int maxcount;
|
|
358
|
7189 int banbadword; /* don't include badword in suggestions */
|
|
388
|
7190 int need_cap; /* word should start with capital */
|
|
344
|
7191 {
|
|
385
|
7192 int attr = 0;
|
|
372
|
7193 char_u buf[MAXPATHL];
|
|
|
7194 char_u *p;
|
|
|
7195 int do_combine = FALSE;
|
|
|
7196 char_u *sps_copy;
|
|
|
7197 #ifdef FEAT_EVAL
|
|
|
7198 static int expr_busy = FALSE;
|
|
|
7199 #endif
|
|
385
|
7200 int c;
|
|
344
|
7201
|
|
|
7202 /*
|
|
|
7203 * Set the info in "*su".
|
|
|
7204 */
|
|
|
7205 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
7206 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
7207 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
7208 if (*badptr == NUL)
|
|
|
7209 return;
|
|
344
|
7210 hash_init(&su->su_banned);
|
|
|
7211
|
|
|
7212 su->su_badptr = badptr;
|
|
385
|
7213 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
7214 su->su_maxcount = maxcount;
|
|
372
|
7215 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
7216
|
|
|
7217 if (su->su_badlen >= MAXWLEN)
|
|
|
7218 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
7219 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
7220 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
7221 su->su_fbadword, MAXWLEN);
|
|
346
|
7222 /* get caps flags for bad word */
|
|
474
|
7223 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
7224 su->su_badptr + su->su_badlen);
|
|
388
|
7225 if (need_cap)
|
|
|
7226 su->su_badflags |= WF_ONECAP;
|
|
344
|
7227
|
|
385
|
7228 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
7229 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
7230 * for that. */
|
|
455
|
7231 c = PTR2CHAR(su->su_badptr);
|
|
385
|
7232 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
7233 {
|
|
|
7234 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
7235 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
7236 0, TRUE);
|
|
|
7237 }
|
|
|
7238
|
|
344
|
7239 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
7240 if (banbadword)
|
|
|
7241 add_banned(su, su->su_badword);
|
|
344
|
7242
|
|
372
|
7243 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
7244 sps_copy = vim_strsave(p_sps);
|
|
|
7245 if (sps_copy == NULL)
|
|
|
7246 return;
|
|
|
7247
|
|
|
7248 /* Loop over the items in 'spellsuggest'. */
|
|
|
7249 for (p = sps_copy; *p != NUL; )
|
|
|
7250 {
|
|
|
7251 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
7252
|
|
|
7253 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
7254 {
|
|
|
7255 #ifdef FEAT_EVAL
|
|
375
|
7256 /* Evaluate an expression. Skip this when called recursively,
|
|
|
7257 * when using spellsuggest() in the expression. */
|
|
372
|
7258 if (!expr_busy)
|
|
|
7259 {
|
|
|
7260 expr_busy = TRUE;
|
|
|
7261 spell_suggest_expr(su, buf + 5);
|
|
|
7262 expr_busy = FALSE;
|
|
|
7263 }
|
|
|
7264 #endif
|
|
|
7265 }
|
|
|
7266 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
7267 /* Use list of suggestions in a file. */
|
|
|
7268 spell_suggest_file(su, buf + 5);
|
|
|
7269 else
|
|
|
7270 {
|
|
|
7271 /* Use internal method. */
|
|
|
7272 spell_suggest_intern(su);
|
|
|
7273 if (sps_flags & SPS_DOUBLE)
|
|
|
7274 do_combine = TRUE;
|
|
|
7275 }
|
|
|
7276 }
|
|
|
7277
|
|
|
7278 vim_free(sps_copy);
|
|
|
7279
|
|
|
7280 if (do_combine)
|
|
|
7281 /* Combine the two list of suggestions. This must be done last,
|
|
|
7282 * because sorting changes the order again. */
|
|
|
7283 score_combine(su);
|
|
|
7284 }
|
|
|
7285
|
|
|
7286 #ifdef FEAT_EVAL
|
|
|
7287 /*
|
|
|
7288 * Find suggestions by evaluating expression "expr".
|
|
|
7289 */
|
|
|
7290 static void
|
|
|
7291 spell_suggest_expr(su, expr)
|
|
|
7292 suginfo_T *su;
|
|
|
7293 char_u *expr;
|
|
|
7294 {
|
|
|
7295 list_T *list;
|
|
|
7296 listitem_T *li;
|
|
|
7297 int score;
|
|
|
7298 char_u *p;
|
|
|
7299
|
|
|
7300 /* The work is split up in a few parts to avoid having to export
|
|
|
7301 * suginfo_T.
|
|
|
7302 * First evaluate the expression and get the resulting list. */
|
|
|
7303 list = eval_spell_expr(su->su_badword, expr);
|
|
|
7304 if (list != NULL)
|
|
|
7305 {
|
|
|
7306 /* Loop over the items in the list. */
|
|
|
7307 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
7308 if (li->li_tv.v_type == VAR_LIST)
|
|
|
7309 {
|
|
|
7310 /* Get the word and the score from the items. */
|
|
|
7311 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
7312 if (score >= 0)
|
|
|
7313 add_suggestion(su, &su->su_ga, p,
|
|
|
7314 su->su_badlen, score, 0, TRUE);
|
|
|
7315 }
|
|
|
7316 list_unref(list);
|
|
|
7317 }
|
|
|
7318
|
|
|
7319 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
7320 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7321 }
|
|
|
7322 #endif
|
|
|
7323
|
|
|
7324 /*
|
|
474
|
7325 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
7326 */
|
|
|
7327 static void
|
|
|
7328 spell_suggest_file(su, fname)
|
|
|
7329 suginfo_T *su;
|
|
|
7330 char_u *fname;
|
|
|
7331 {
|
|
|
7332 FILE *fd;
|
|
|
7333 char_u line[MAXWLEN * 2];
|
|
|
7334 char_u *p;
|
|
|
7335 int len;
|
|
|
7336 char_u cword[MAXWLEN];
|
|
|
7337
|
|
|
7338 /* Open the file. */
|
|
|
7339 fd = mch_fopen((char *)fname, "r");
|
|
|
7340 if (fd == NULL)
|
|
|
7341 {
|
|
|
7342 EMSG2(_(e_notopen), fname);
|
|
|
7343 return;
|
|
|
7344 }
|
|
|
7345
|
|
|
7346 /* Read it line by line. */
|
|
|
7347 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
7348 {
|
|
|
7349 line_breakcheck();
|
|
|
7350
|
|
|
7351 p = vim_strchr(line, '/');
|
|
|
7352 if (p == NULL)
|
|
|
7353 continue; /* No Tab found, just skip the line. */
|
|
|
7354 *p++ = NUL;
|
|
|
7355 if (STRICMP(su->su_badword, line) == 0)
|
|
|
7356 {
|
|
|
7357 /* Match! Isolate the good word, until CR or NL. */
|
|
|
7358 for (len = 0; p[len] >= ' '; ++len)
|
|
|
7359 ;
|
|
|
7360 p[len] = NUL;
|
|
|
7361
|
|
|
7362 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
7363 * of the bad word. */
|
|
|
7364 if (captype(p, NULL) == 0)
|
|
|
7365 {
|
|
|
7366 make_case_word(p, cword, su->su_badflags);
|
|
|
7367 p = cword;
|
|
|
7368 }
|
|
|
7369
|
|
|
7370 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
7371 SCORE_FILE, 0, TRUE);
|
|
|
7372 }
|
|
|
7373 }
|
|
|
7374
|
|
|
7375 fclose(fd);
|
|
|
7376
|
|
|
7377 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
7378 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
7379 }
|
|
|
7380
|
|
|
7381 /*
|
|
|
7382 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
7383 */
|
|
|
7384 static void
|
|
|
7385 spell_suggest_intern(su)
|
|
|
7386 suginfo_T *su;
|
|
|
7387 {
|
|
344
|
7388 /*
|
|
346
|
7389 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
7390 *
|
|
|
7391 * Set a maximum score to limit the combination of operations that is
|
|
|
7392 * tried.
|
|
|
7393 */
|
|
346
|
7394 suggest_try_special(su);
|
|
|
7395
|
|
|
7396 /*
|
|
|
7397 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
7398 * from the .aff file and inserting a space (split the word).
|
|
|
7399 */
|
|
|
7400 suggest_try_change(su);
|
|
344
|
7401
|
|
|
7402 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
7403 if (sps_flags & SPS_DOUBLE)
|
|
|
7404 score_comp_sal(su);
|
|
|
7405
|
|
|
7406 /*
|
|
346
|
7407 * 3. Try finding sound-a-like words.
|
|
344
|
7408 *
|
|
|
7409 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
7410 * very slow and often doesn't find new suggestions.
|
|
|
7411 */
|
|
|
7412 if ((sps_flags & SPS_DOUBLE)
|
|
|
7413 || (!(sps_flags & SPS_FAST)
|
|
|
7414 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
7415 {
|
|
|
7416 /* Allow a higher score now. */
|
|
|
7417 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
7418 suggest_try_soundalike(su);
|
|
344
|
7419 }
|
|
|
7420
|
|
|
7421 /* When CTRL-C was hit while searching do show the results. */
|
|
|
7422 ui_breakcheck();
|
|
|
7423 if (got_int)
|
|
|
7424 {
|
|
|
7425 (void)vgetc();
|
|
|
7426 got_int = FALSE;
|
|
|
7427 }
|
|
|
7428
|
|
372
|
7429 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
7430 {
|
|
|
7431 if (sps_flags & SPS_BEST)
|
|
|
7432 /* Adjust the word score for how it sounds like. */
|
|
|
7433 rescore_suggestions(su);
|
|
|
7434
|
|
|
7435 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
7436 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
7437 }
|
|
|
7438 }
|
|
|
7439
|
|
|
7440 /*
|
|
|
7441 * Free the info put in "*su" by spell_find_suggest().
|
|
|
7442 */
|
|
|
7443 static void
|
|
|
7444 spell_find_cleanup(su)
|
|
|
7445 suginfo_T *su;
|
|
|
7446 {
|
|
|
7447 int i;
|
|
|
7448
|
|
|
7449 /* Free the suggestions. */
|
|
|
7450 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7451 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
7452 ga_clear(&su->su_ga);
|
|
|
7453 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
7454 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
7455 ga_clear(&su->su_sga);
|
|
323
|
7456
|
|
|
7457 /* Free the banned words. */
|
|
344
|
7458 free_banned(su);
|
|
323
|
7459 }
|
|
|
7460
|
|
|
7461 /*
|
|
324
|
7462 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
7463 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
7464 * The result is NUL terminated.
|
|
323
|
7465 */
|
|
|
7466 static void
|
|
324
|
7467 onecap_copy(word, wcopy, upper)
|
|
323
|
7468 char_u *word;
|
|
|
7469 char_u *wcopy;
|
|
|
7470 int upper; /* TRUE: first letter made upper case */
|
|
|
7471 {
|
|
|
7472 char_u *p;
|
|
|
7473 int c;
|
|
|
7474 int l;
|
|
|
7475
|
|
|
7476 p = word;
|
|
|
7477 #ifdef FEAT_MBYTE
|
|
|
7478 if (has_mbyte)
|
|
474
|
7479 c = mb_cptr2char_adv(&p);
|
|
323
|
7480 else
|
|
|
7481 #endif
|
|
|
7482 c = *p++;
|
|
|
7483 if (upper)
|
|
324
|
7484 c = SPELL_TOUPPER(c);
|
|
323
|
7485 else
|
|
324
|
7486 c = SPELL_TOFOLD(c);
|
|
323
|
7487 #ifdef FEAT_MBYTE
|
|
|
7488 if (has_mbyte)
|
|
|
7489 l = mb_char2bytes(c, wcopy);
|
|
|
7490 else
|
|
|
7491 #endif
|
|
|
7492 {
|
|
|
7493 l = 1;
|
|
|
7494 wcopy[0] = c;
|
|
|
7495 }
|
|
376
|
7496 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
7497 }
|
|
|
7498
|
|
|
7499 /*
|
|
324
|
7500 * Make a copy of "word" with all the letters upper cased into
|
|
|
7501 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
7502 */
|
|
|
7503 static void
|
|
|
7504 allcap_copy(word, wcopy)
|
|
|
7505 char_u *word;
|
|
|
7506 char_u *wcopy;
|
|
|
7507 {
|
|
|
7508 char_u *s;
|
|
|
7509 char_u *d;
|
|
|
7510 int c;
|
|
|
7511
|
|
|
7512 d = wcopy;
|
|
|
7513 for (s = word; *s != NUL; )
|
|
|
7514 {
|
|
|
7515 #ifdef FEAT_MBYTE
|
|
|
7516 if (has_mbyte)
|
|
474
|
7517 c = mb_cptr2char_adv(&s);
|
|
323
|
7518 else
|
|
|
7519 #endif
|
|
|
7520 c = *s++;
|
|
324
|
7521 c = SPELL_TOUPPER(c);
|
|
323
|
7522
|
|
|
7523 #ifdef FEAT_MBYTE
|
|
|
7524 if (has_mbyte)
|
|
|
7525 {
|
|
|
7526 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
7527 break;
|
|
|
7528 d += mb_char2bytes(c, d);
|
|
|
7529 }
|
|
|
7530 else
|
|
|
7531 #endif
|
|
|
7532 {
|
|
|
7533 if (d - wcopy >= MAXWLEN - 1)
|
|
|
7534 break;
|
|
|
7535 *d++ = c;
|
|
|
7536 }
|
|
|
7537 }
|
|
|
7538 *d = NUL;
|
|
|
7539 }
|
|
|
7540
|
|
|
7541 /*
|
|
346
|
7542 * Try finding suggestions by recognizing specific situations.
|
|
|
7543 */
|
|
|
7544 static void
|
|
|
7545 suggest_try_special(su)
|
|
|
7546 suginfo_T *su;
|
|
|
7547 {
|
|
|
7548 char_u *p;
|
|
376
|
7549 size_t len;
|
|
346
|
7550 int c;
|
|
|
7551 char_u word[MAXWLEN];
|
|
|
7552
|
|
|
7553 /*
|
|
|
7554 * Recognize a word that is repeated: "the the".
|
|
|
7555 */
|
|
|
7556 p = skiptowhite(su->su_fbadword);
|
|
|
7557 len = p - su->su_fbadword;
|
|
|
7558 p = skipwhite(p);
|
|
|
7559 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
7560 {
|
|
|
7561 /* Include badflags: if the badword is onecap or allcap
|
|
|
7562 * use that for the goodword too: "The the" -> "The". */
|
|
|
7563 c = su->su_fbadword[len];
|
|
|
7564 su->su_fbadword[len] = NUL;
|
|
|
7565 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
7566 su->su_fbadword[len] = c;
|
|
351
|
7567 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
7568 }
|
|
|
7569 }
|
|
|
7570
|
|
|
7571 /*
|
|
323
|
7572 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
7573 *
|
|
|
7574 * This uses a state machine. At each node in the tree we try various
|
|
|
7575 * operations. When trying if an operation work "depth" is increased and the
|
|
|
7576 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
7577 * character, replace one and delete another. The number of changes is
|
|
|
7578 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
7579 */
|
|
|
7580 static void
|
|
346
|
7581 suggest_try_change(su)
|
|
323
|
7582 suginfo_T *su;
|
|
|
7583 {
|
|
|
7584 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
7585 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
7586 trystate_T stack[MAXWLEN];
|
|
|
7587 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
|
7588 * to for word split) */
|
|
|
7589 char_u prewordlen = 0; /* length of word in "preword" */
|
|
|
7590 int splitoff = 0; /* index in tword after last split */
|
|
|
7591 trystate_T *sp;
|
|
|
7592 int newscore;
|
|
|
7593 langp_T *lp;
|
|
375
|
7594 char_u *byts, *fbyts, *pbyts;
|
|
|
7595 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
7596 int depth;
|
|
330
|
7597 int c, c2, c3;
|
|
390
|
7598 int n;
|
|
323
|
7599 int flags;
|
|
|
7600 garray_T *gap;
|
|
324
|
7601 idx_T arridx;
|
|
323
|
7602 int len;
|
|
|
7603 char_u *p;
|
|
|
7604 fromto_T *ftp;
|
|
330
|
7605 int fl = 0, tl;
|
|
346
|
7606 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
323
|
7607
|
|
|
7608 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
7609 * to find matches (esp. REP items). Append some more text, changing
|
|
|
7610 * chars after the bad word may help. */
|
|
323
|
7611 STRCPY(fword, su->su_fbadword);
|
|
346
|
7612 n = STRLEN(fword);
|
|
|
7613 p = su->su_badptr + su->su_badlen;
|
|
|
7614 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
7615
|
|
|
7616 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7617 lp->lp_slang != NULL; ++lp)
|
|
|
7618 {
|
|
|
7619 /*
|
|
|
7620 * Go through the whole case-fold tree, try changes at each node.
|
|
|
7621 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
7622 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
7623 * word).
|
|
|
7624 */
|
|
|
7625 depth = 0;
|
|
375
|
7626 sp = &stack[0];
|
|
|
7627 sp->ts_state = STATE_START;
|
|
|
7628 sp->ts_score = 0;
|
|
|
7629 sp->ts_curi = 1;
|
|
|
7630 sp->ts_fidx = 0;
|
|
|
7631 sp->ts_fidxtry = 0;
|
|
|
7632 sp->ts_twordlen = 0;
|
|
|
7633 sp->ts_arridx = 0;
|
|
330
|
7634 #ifdef FEAT_MBYTE
|
|
375
|
7635 sp->ts_tcharlen = 0;
|
|
330
|
7636 #endif
|
|
|
7637
|
|
|
7638 /*
|
|
375
|
7639 * When there are postponed prefixes we need to use these first. At
|
|
|
7640 * the end of the prefix we continue in the case-fold tree.
|
|
|
7641 */
|
|
|
7642 fbyts = lp->lp_slang->sl_fbyts;
|
|
|
7643 fidxs = lp->lp_slang->sl_fidxs;
|
|
|
7644 pbyts = lp->lp_slang->sl_pbyts;
|
|
|
7645 pidxs = lp->lp_slang->sl_pidxs;
|
|
|
7646 if (pbyts != NULL)
|
|
|
7647 {
|
|
|
7648 byts = pbyts;
|
|
|
7649 idxs = pidxs;
|
|
|
7650 sp->ts_prefixdepth = PREFIXTREE;
|
|
|
7651 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
7652 }
|
|
|
7653 else
|
|
|
7654 {
|
|
|
7655 byts = fbyts;
|
|
|
7656 idxs = fidxs;
|
|
|
7657 sp->ts_prefixdepth = NOPREFIX;
|
|
|
7658 }
|
|
|
7659
|
|
|
7660 /*
|
|
330
|
7661 * Loop to find all suggestions. At each round we either:
|
|
|
7662 * - For the current state try one operation, advance "ts_curi",
|
|
|
7663 * increase "depth".
|
|
|
7664 * - When a state is done go to the next, set "ts_state".
|
|
|
7665 * - When all states are tried decrease "depth".
|
|
|
7666 */
|
|
323
|
7667 while (depth >= 0 && !got_int)
|
|
|
7668 {
|
|
|
7669 sp = &stack[depth];
|
|
|
7670 switch (sp->ts_state)
|
|
|
7671 {
|
|
|
7672 case STATE_START:
|
|
375
|
7673 case STATE_NOPREFIX:
|
|
323
|
7674 /*
|
|
|
7675 * Start of node: Deal with NUL bytes, which means
|
|
|
7676 * tword[] may end here.
|
|
|
7677 */
|
|
|
7678 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
7679 len = byts[arridx]; /* bytes in this node */
|
|
|
7680 arridx += sp->ts_curi; /* index of current byte */
|
|
|
7681
|
|
375
|
7682 if (sp->ts_prefixdepth == PREFIXTREE)
|
|
|
7683 {
|
|
|
7684 /* Skip over the NUL bytes, we use them later. */
|
|
|
7685 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
7686 ;
|
|
|
7687 sp->ts_curi += n;
|
|
|
7688
|
|
390
|
7689 /* Always past NUL bytes now. */
|
|
|
7690 n = (int)sp->ts_state;
|
|
|
7691 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7692 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
7693
|
|
375
|
7694 /* At end of a prefix or at start of prefixtree: check for
|
|
|
7695 * following word. */
|
|
390
|
7696 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
7697 {
|
|
455
|
7698 /* Set su->su_badflags to the caps type at this
|
|
|
7699 * position. Use the caps type until here for the
|
|
|
7700 * prefix itself. */
|
|
|
7701 #ifdef FEAT_MBYTE
|
|
|
7702 if (has_mbyte)
|
|
|
7703 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
7704 else
|
|
|
7705 #endif
|
|
|
7706 n = sp->ts_fidx;
|
|
474
|
7707 flags = badword_captype(su->su_badptr,
|
|
|
7708 su->su_badptr + n);
|
|
|
7709 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
7710 su->su_badptr + su->su_badlen);
|
|
375
|
7711 ++depth;
|
|
|
7712 stack[depth] = stack[depth - 1];
|
|
|
7713 sp = &stack[depth];
|
|
|
7714 sp->ts_prefixdepth = depth - 1;
|
|
|
7715 byts = fbyts;
|
|
|
7716 idxs = fidxs;
|
|
|
7717 sp->ts_state = STATE_START;
|
|
|
7718 sp->ts_curi = 1; /* start just after length byte */
|
|
|
7719 sp->ts_arridx = 0;
|
|
|
7720
|
|
455
|
7721 /* Move the prefix to preword[] with the right case
|
|
|
7722 * and make find_keepcap_word() works. */
|
|
|
7723 splitoff = sp->ts_twordlen;
|
|
|
7724 tword[splitoff] = NUL;
|
|
|
7725 make_case_word(tword, preword, flags);
|
|
|
7726 prewordlen = STRLEN(preword);
|
|
375
|
7727 }
|
|
|
7728 break;
|
|
|
7729 }
|
|
|
7730
|
|
346
|
7731 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
7732 {
|
|
|
7733 /* Past bytes in node and/or past NUL bytes. */
|
|
|
7734 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7735 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7736 break;
|
|
|
7737 }
|
|
|
7738
|
|
|
7739 /*
|
|
|
7740 * End of word in tree.
|
|
|
7741 */
|
|
|
7742 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
7743
|
|
324
|
7744 flags = (int)idxs[arridx];
|
|
323
|
7745
|
|
375
|
7746 if (sp->ts_prefixdepth < MAXWLEN)
|
|
|
7747 {
|
|
|
7748 /* There was a prefix before the word. Check that the
|
|
|
7749 * prefix can be used with this word. */
|
|
|
7750 /* Count the length of the NULs in the prefix. If there
|
|
|
7751 * are none this must be the first try without a prefix.
|
|
|
7752 */
|
|
|
7753 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
7754 len = pbyts[n++];
|
|
|
7755 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
7756 ;
|
|
|
7757 if (c > 0)
|
|
|
7758 {
|
|
390
|
7759 /* The prefix ID is stored three bytes above the
|
|
|
7760 * flags. */
|
|
|
7761 c = valid_word_prefix(c, n, flags,
|
|
455
|
7762 tword + splitoff, lp->lp_slang, FALSE);
|
|
375
|
7763 if (c == 0)
|
|
|
7764 break;
|
|
|
7765
|
|
|
7766 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
7767 if (c & WF_RAREPFX)
|
|
|
7768 flags |= WF_RARE;
|
|
|
7769 }
|
|
|
7770 }
|
|
|
7771
|
|
323
|
7772 /*
|
|
|
7773 * Form the word with proper case in preword.
|
|
|
7774 * If there is a word from a previous split, append.
|
|
|
7775 */
|
|
|
7776 tword[sp->ts_twordlen] = NUL;
|
|
|
7777 if (flags & WF_KEEPCAP)
|
|
|
7778 /* Must find the word in the keep-case tree. */
|
|
|
7779 find_keepcap_word(lp->lp_slang, tword + splitoff,
|
|
|
7780 preword + prewordlen);
|
|
|
7781 else
|
|
346
|
7782 {
|
|
323
|
7783 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
7784 * use that for the goodword too. But if the badword is
|
|
|
7785 * allcap and it's only one char long use onecap. */
|
|
|
7786 c = su->su_badflags;
|
|
|
7787 if ((c & WF_ALLCAP)
|
|
|
7788 #ifdef FEAT_MBYTE
|
|
474
|
7789 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
7790 #else
|
|
|
7791 && su->su_badlen == 1
|
|
|
7792 #endif
|
|
|
7793 )
|
|
|
7794 c = WF_ONECAP;
|
|
323
|
7795 make_case_word(tword + splitoff,
|
|
346
|
7796 preword + prewordlen, flags | c);
|
|
|
7797 }
|
|
323
|
7798
|
|
|
7799 /* Don't use a banned word. It may appear again as a good
|
|
|
7800 * word, thus remember it. */
|
|
|
7801 if (flags & WF_BANNED)
|
|
|
7802 {
|
|
|
7803 add_banned(su, preword + prewordlen);
|
|
|
7804 break;
|
|
|
7805 }
|
|
372
|
7806 if (was_banned(su, preword + prewordlen)
|
|
|
7807 || was_banned(su, preword))
|
|
323
|
7808 break;
|
|
|
7809
|
|
|
7810 newscore = 0;
|
|
|
7811 if ((flags & WF_REGION)
|
|
390
|
7812 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
7813 newscore += SCORE_REGION;
|
|
|
7814 if (flags & WF_RARE)
|
|
|
7815 newscore += SCORE_RARE;
|
|
|
7816
|
|
346
|
7817 if (!spell_valid_case(su->su_badflags,
|
|
323
|
7818 captype(preword + prewordlen, NULL)))
|
|
|
7819 newscore += SCORE_ICASE;
|
|
|
7820
|
|
346
|
7821 if ((fword[sp->ts_fidx] == NUL
|
|
376
|
7822 || !spell_iswordp(fword + sp->ts_fidx, curbuf))
|
|
346
|
7823 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
7824 {
|
|
366
|
7825 /* The badword also ends: add suggestions. Give a penalty
|
|
|
7826 * when changing non-word char to word char, e.g., "thes,"
|
|
|
7827 * -> "these". */
|
|
|
7828 p = fword + sp->ts_fidx;
|
|
|
7829 #ifdef FEAT_MBYTE
|
|
|
7830 if (has_mbyte)
|
|
|
7831 mb_ptr_back(fword, p);
|
|
|
7832 else
|
|
|
7833 #endif
|
|
|
7834 --p;
|
|
376
|
7835 if (!spell_iswordp(p, curbuf))
|
|
366
|
7836 {
|
|
|
7837 p = preword + STRLEN(preword);
|
|
|
7838 #ifdef FEAT_MBYTE
|
|
|
7839 if (has_mbyte)
|
|
|
7840 mb_ptr_back(preword, p);
|
|
|
7841 else
|
|
|
7842 #endif
|
|
|
7843 --p;
|
|
376
|
7844 if (spell_iswordp(p, curbuf))
|
|
366
|
7845 newscore += SCORE_NONWORD;
|
|
|
7846 }
|
|
|
7847
|
|
344
|
7848 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
7849 sp->ts_fidx - repextra,
|
|
351
|
7850 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
7851 }
|
|
330
|
7852 else if (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
7853 #ifdef FEAT_MBYTE
|
|
|
7854 /* Don't split halfway a character. */
|
|
|
7855 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
7856 #endif
|
|
|
7857 )
|
|
323
|
7858 {
|
|
|
7859 /* The word in the tree ends but the badword
|
|
|
7860 * continues: try inserting a space and check that a valid
|
|
|
7861 * words starts at fword[sp->ts_fidx]. */
|
|
|
7862 if (try_deeper(su, stack, depth, newscore + SCORE_SPLIT))
|
|
|
7863 {
|
|
|
7864 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
|
7865 sp->ts_save_prewordlen = prewordlen;
|
|
346
|
7866 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7867 sp->ts_save_splitoff = splitoff;
|
|
376
|
7868 sp->ts_state = STATE_SPLITUNDO;
|
|
|
7869
|
|
|
7870 ++depth;
|
|
|
7871 sp = &stack[depth];
|
|
323
|
7872
|
|
|
7873 /* Append a space to preword. */
|
|
|
7874 STRCAT(preword, " ");
|
|
|
7875 prewordlen = STRLEN(preword);
|
|
|
7876 splitoff = sp->ts_twordlen;
|
|
376
|
7877
|
|
|
7878 /* If the badword has a non-word character at this
|
|
|
7879 * position skip it. That means replacing the
|
|
|
7880 * non-word character with a space. */
|
|
|
7881 if (!spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
7882 {
|
|
|
7883 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
7884 #ifdef FEAT_MBYTE
|
|
|
7885 if (has_mbyte)
|
|
|
7886 sp->ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
|
7887 else
|
|
|
7888 #endif
|
|
|
7889 ++sp->ts_fidx;
|
|
|
7890 }
|
|
455
|
7891
|
|
|
7892 /* set su->su_badflags to the caps type at this
|
|
|
7893 * position */
|
|
|
7894
|
|
324
|
7895 #ifdef FEAT_MBYTE
|
|
|
7896 if (has_mbyte)
|
|
455
|
7897 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
7898 else
|
|
|
7899 #endif
|
|
455
|
7900 n = sp->ts_fidx;
|
|
474
|
7901 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
7902 su->su_badptr + su->su_badlen);
|
|
323
|
7903
|
|
|
7904 /* Restart at top of the tree. */
|
|
376
|
7905 sp->ts_arridx = 0;
|
|
323
|
7906 }
|
|
|
7907 }
|
|
|
7908 break;
|
|
|
7909
|
|
|
7910 case STATE_SPLITUNDO:
|
|
346
|
7911 /* Undo the changes done for word split. */
|
|
|
7912 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
7913 splitoff = sp->ts_save_splitoff;
|
|
|
7914 prewordlen = sp->ts_save_prewordlen;
|
|
|
7915
|
|
|
7916 /* Continue looking for NUL bytes. */
|
|
|
7917 sp->ts_state = STATE_START;
|
|
|
7918 break;
|
|
|
7919
|
|
|
7920 case STATE_ENDNUL:
|
|
|
7921 /* Past the NUL bytes in the node. */
|
|
455
|
7922 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
7923 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
7924 {
|
|
|
7925 /* The badword ends, can't use the bytes in this node. */
|
|
|
7926 sp->ts_state = STATE_DEL;
|
|
|
7927 break;
|
|
|
7928 }
|
|
|
7929 sp->ts_state = STATE_PLAIN;
|
|
|
7930 /*FALLTHROUGH*/
|
|
|
7931
|
|
|
7932 case STATE_PLAIN:
|
|
|
7933 /*
|
|
|
7934 * Go over all possible bytes at this node, add each to
|
|
|
7935 * tword[] and use child node. "ts_curi" is the index.
|
|
|
7936 */
|
|
|
7937 arridx = sp->ts_arridx;
|
|
|
7938 if (sp->ts_curi > byts[arridx])
|
|
|
7939 {
|
|
|
7940 /* Done all bytes at this node, do next state. When still
|
|
|
7941 * at already changed bytes skip the other tricks. */
|
|
|
7942 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7943 sp->ts_state = STATE_DEL;
|
|
|
7944 else
|
|
|
7945 sp->ts_state = STATE_FINAL;
|
|
|
7946 }
|
|
|
7947 else
|
|
|
7948 {
|
|
|
7949 arridx += sp->ts_curi++;
|
|
|
7950 c = byts[arridx];
|
|
|
7951
|
|
|
7952 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
7953 * the byte in the bad word adjust the score. But don't
|
|
|
7954 * even try when the byte was already changed. */
|
|
330
|
7955 if (c == fword[sp->ts_fidx]
|
|
|
7956 #ifdef FEAT_MBYTE
|
|
|
7957 || (sp->ts_tcharlen > 0
|
|
|
7958 && sp->ts_isdiff != DIFF_NONE)
|
|
|
7959 #endif
|
|
|
7960 )
|
|
323
|
7961 newscore = 0;
|
|
|
7962 else
|
|
|
7963 newscore = SCORE_SUBST;
|
|
|
7964 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7965 && try_deeper(su, stack, depth, newscore))
|
|
|
7966 {
|
|
|
7967 ++depth;
|
|
330
|
7968 sp = &stack[depth];
|
|
|
7969 ++sp->ts_fidx;
|
|
|
7970 tword[sp->ts_twordlen++] = c;
|
|
|
7971 sp->ts_arridx = idxs[arridx];
|
|
|
7972 #ifdef FEAT_MBYTE
|
|
|
7973 if (newscore == SCORE_SUBST)
|
|
|
7974 sp->ts_isdiff = DIFF_YES;
|
|
|
7975 if (has_mbyte)
|
|
|
7976 {
|
|
|
7977 /* Multi-byte characters are a bit complicated to
|
|
|
7978 * handle: They differ when any of the bytes
|
|
|
7979 * differ and then their length may also differ. */
|
|
|
7980 if (sp->ts_tcharlen == 0)
|
|
|
7981 {
|
|
|
7982 /* First byte. */
|
|
|
7983 sp->ts_tcharidx = 0;
|
|
|
7984 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
7985 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
7986 sp->ts_isdiff = (newscore != 0)
|
|
|
7987 ? DIFF_YES : DIFF_NONE;
|
|
|
7988 }
|
|
|
7989 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
7990 /* When inserting trail bytes don't advance in
|
|
|
7991 * the bad word. */
|
|
|
7992 --sp->ts_fidx;
|
|
|
7993 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
7994 {
|
|
|
7995 /* Last byte of character. */
|
|
|
7996 if (sp->ts_isdiff == DIFF_YES)
|
|
|
7997 {
|
|
|
7998 /* Correct ts_fidx for the byte length of
|
|
|
7999 * the character (we didn't check that
|
|
|
8000 * before). */
|
|
|
8001 sp->ts_fidx = sp->ts_fcharstart
|
|
|
8002 + MB_BYTE2LEN(
|
|
|
8003 fword[sp->ts_fcharstart]);
|
|
|
8004
|
|
479
|
8005 /* For changing a composing character
|
|
|
8006 * adjust the score from SCORE_SUBST to
|
|
|
8007 * SCORE_SUBCOMP. */
|
|
|
8008 if (enc_utf8
|
|
|
8009 && utf_iscomposing(
|
|
|
8010 mb_ptr2char(tword
|
|
|
8011 + sp->ts_twordlen
|
|
|
8012 - sp->ts_tcharlen))
|
|
|
8013 && utf_iscomposing(
|
|
|
8014 mb_ptr2char(fword
|
|
|
8015 + sp->ts_fcharstart)))
|
|
|
8016 sp->ts_score -=
|
|
|
8017 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
8018
|
|
330
|
8019 /* For a similar character adjust score
|
|
|
8020 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
479
|
8021 else if (lp->lp_slang->sl_has_map
|
|
330
|
8022 && similar_chars(lp->lp_slang,
|
|
|
8023 mb_ptr2char(tword
|
|
|
8024 + sp->ts_twordlen
|
|
|
8025 - sp->ts_tcharlen),
|
|
|
8026 mb_ptr2char(fword
|
|
|
8027 + sp->ts_fcharstart)))
|
|
|
8028 sp->ts_score -=
|
|
|
8029 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
8030 }
|
|
358
|
8031 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
8032 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
8033 {
|
|
|
8034 p = tword + sp->ts_twordlen
|
|
|
8035 - sp->ts_tcharlen;
|
|
|
8036 c = mb_ptr2char(p);
|
|
475
|
8037 if (enc_utf8 && utf_iscomposing(c))
|
|
|
8038 {
|
|
|
8039 /* Inserting a composing char doesn't
|
|
|
8040 * count that much. */
|
|
358
|
8041 sp->ts_score -= SCORE_INS
|
|
475
|
8042 - SCORE_INSCOMP;
|
|
|
8043 }
|
|
|
8044 else
|
|
|
8045 {
|
|
|
8046 /* If the previous character was the
|
|
|
8047 * same, thus doubling a character,
|
|
|
8048 * give a bonus to the score. */
|
|
|
8049 mb_ptr_back(tword, p);
|
|
|
8050 if (c == mb_ptr2char(p))
|
|
|
8051 sp->ts_score -= SCORE_INS
|
|
358
|
8052 - SCORE_INSDUP;
|
|
475
|
8053 }
|
|
358
|
8054 }
|
|
330
|
8055
|
|
|
8056 /* Starting a new char, reset the length. */
|
|
|
8057 sp->ts_tcharlen = 0;
|
|
|
8058 }
|
|
|
8059 }
|
|
|
8060 else
|
|
|
8061 #endif
|
|
|
8062 {
|
|
|
8063 /* If we found a similar char adjust the score.
|
|
|
8064 * We do this after calling try_deeper() because
|
|
|
8065 * it's slow. */
|
|
|
8066 if (newscore != 0
|
|
|
8067 && lp->lp_slang->sl_has_map
|
|
|
8068 && similar_chars(lp->lp_slang,
|
|
|
8069 c, fword[sp->ts_fidx - 1]))
|
|
|
8070 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
8071 }
|
|
323
|
8072 }
|
|
|
8073 }
|
|
|
8074 break;
|
|
|
8075
|
|
|
8076 case STATE_DEL:
|
|
330
|
8077 #ifdef FEAT_MBYTE
|
|
|
8078 /* When past the first byte of a multi-byte char don't try
|
|
|
8079 * delete/insert/swap a character. */
|
|
|
8080 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
8081 {
|
|
|
8082 sp->ts_state = STATE_FINAL;
|
|
|
8083 break;
|
|
|
8084 }
|
|
|
8085 #endif
|
|
|
8086 /*
|
|
|
8087 * Try skipping one character in the bad word (delete it).
|
|
|
8088 */
|
|
323
|
8089 sp->ts_state = STATE_INS;
|
|
|
8090 sp->ts_curi = 1;
|
|
|
8091 if (fword[sp->ts_fidx] != NUL
|
|
|
8092 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
8093 {
|
|
|
8094 ++depth;
|
|
358
|
8095
|
|
|
8096 /* Advance over the character in fword[]. Give a bonus to
|
|
|
8097 * the score if the same character is following "nn" ->
|
|
|
8098 * "n". */
|
|
330
|
8099 #ifdef FEAT_MBYTE
|
|
|
8100 if (has_mbyte)
|
|
358
|
8101 {
|
|
|
8102 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
8103 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
479
|
8104 if (enc_utf8 && utf_iscomposing(c))
|
|
|
8105 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
8106 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
358
|
8107 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
8108 }
|
|
330
|
8109 else
|
|
|
8110 #endif
|
|
358
|
8111 {
|
|
330
|
8112 ++stack[depth].ts_fidx;
|
|
358
|
8113 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
8114 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
8115 }
|
|
323
|
8116 break;
|
|
|
8117 }
|
|
|
8118 /*FALLTHROUGH*/
|
|
|
8119
|
|
|
8120 case STATE_INS:
|
|
330
|
8121 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
8122 * node. */
|
|
|
8123 n = sp->ts_arridx;
|
|
|
8124 if (sp->ts_curi > byts[n])
|
|
|
8125 {
|
|
|
8126 /* Done all bytes at this node, do next state. */
|
|
|
8127 sp->ts_state = STATE_SWAP;
|
|
|
8128 }
|
|
|
8129 else
|
|
|
8130 {
|
|
330
|
8131 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
8132 n += sp->ts_curi++;
|
|
|
8133 c = byts[n];
|
|
|
8134 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
8135 {
|
|
|
8136 ++depth;
|
|
330
|
8137 sp = &stack[depth];
|
|
|
8138 tword[sp->ts_twordlen++] = c;
|
|
|
8139 sp->ts_arridx = idxs[n];
|
|
|
8140 #ifdef FEAT_MBYTE
|
|
|
8141 if (has_mbyte)
|
|
|
8142 {
|
|
|
8143 fl = MB_BYTE2LEN(c);
|
|
|
8144 if (fl > 1)
|
|
|
8145 {
|
|
|
8146 /* There are following bytes for the same
|
|
|
8147 * character. We must find all bytes before
|
|
|
8148 * trying delete/insert/swap/etc. */
|
|
|
8149 sp->ts_tcharlen = fl;
|
|
|
8150 sp->ts_tcharidx = 1;
|
|
|
8151 sp->ts_isdiff = DIFF_INSERT;
|
|
|
8152 }
|
|
|
8153 }
|
|
358
|
8154 else
|
|
|
8155 fl = 1;
|
|
|
8156 if (fl == 1)
|
|
330
|
8157 #endif
|
|
358
|
8158 {
|
|
|
8159 /* If the previous character was the same, thus
|
|
|
8160 * doubling a character, give a bonus to the
|
|
|
8161 * score. */
|
|
|
8162 if (sp->ts_twordlen >= 2
|
|
|
8163 && tword[sp->ts_twordlen - 2] == c)
|
|
|
8164 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
8165 }
|
|
323
|
8166 }
|
|
|
8167 }
|
|
|
8168 break;
|
|
|
8169
|
|
|
8170 case STATE_SWAP:
|
|
330
|
8171 /*
|
|
|
8172 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
8173 * We change "fword" here, it's changed back afterwards.
|
|
|
8174 */
|
|
|
8175 p = fword + sp->ts_fidx;
|
|
|
8176 c = *p;
|
|
|
8177 if (c == NUL)
|
|
|
8178 {
|
|
|
8179 /* End of word, can't swap or replace. */
|
|
|
8180 sp->ts_state = STATE_FINAL;
|
|
|
8181 break;
|
|
|
8182 }
|
|
|
8183 #ifdef FEAT_MBYTE
|
|
|
8184 if (has_mbyte)
|
|
|
8185 {
|
|
474
|
8186 n = mb_cptr2len(p);
|
|
330
|
8187 c = mb_ptr2char(p);
|
|
|
8188 c2 = mb_ptr2char(p + n);
|
|
|
8189 }
|
|
|
8190 else
|
|
|
8191 #endif
|
|
|
8192 c2 = p[1];
|
|
|
8193 if (c == c2)
|
|
323
|
8194 {
|
|
330
|
8195 /* Characters are identical, swap won't do anything. */
|
|
|
8196 sp->ts_state = STATE_SWAP3;
|
|
|
8197 break;
|
|
|
8198 }
|
|
|
8199 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
8200 {
|
|
|
8201 sp->ts_state = STATE_UNSWAP;
|
|
323
|
8202 ++depth;
|
|
330
|
8203 #ifdef FEAT_MBYTE
|
|
|
8204 if (has_mbyte)
|
|
|
8205 {
|
|
|
8206 fl = mb_char2len(c2);
|
|
|
8207 mch_memmove(p, p + n, fl);
|
|
|
8208 mb_char2bytes(c, p + fl);
|
|
|
8209 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
8210 }
|
|
|
8211 else
|
|
|
8212 #endif
|
|
|
8213 {
|
|
|
8214 p[0] = c2;
|
|
|
8215 p[1] = c;
|
|
|
8216 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
8217 }
|
|
323
|
8218 }
|
|
|
8219 else
|
|
|
8220 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
8221 sp->ts_state = STATE_REP_INI;
|
|
|
8222 break;
|
|
|
8223
|
|
330
|
8224 case STATE_UNSWAP:
|
|
|
8225 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
8226 p = fword + sp->ts_fidx;
|
|
|
8227 #ifdef FEAT_MBYTE
|
|
|
8228 if (has_mbyte)
|
|
|
8229 {
|
|
|
8230 n = MB_BYTE2LEN(*p);
|
|
|
8231 c = mb_ptr2char(p + n);
|
|
|
8232 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
8233 mb_char2bytes(c, p);
|
|
|
8234 }
|
|
|
8235 else
|
|
|
8236 #endif
|
|
|
8237 {
|
|
|
8238 c = *p;
|
|
|
8239 *p = p[1];
|
|
|
8240 p[1] = c;
|
|
|
8241 }
|
|
|
8242 /*FALLTHROUGH*/
|
|
|
8243
|
|
|
8244 case STATE_SWAP3:
|
|
323
|
8245 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
8246 * "fword" here, it's changed back afterwards. */
|
|
|
8247 p = fword + sp->ts_fidx;
|
|
|
8248 #ifdef FEAT_MBYTE
|
|
|
8249 if (has_mbyte)
|
|
|
8250 {
|
|
474
|
8251 n = mb_cptr2len(p);
|
|
330
|
8252 c = mb_ptr2char(p);
|
|
474
|
8253 fl = mb_cptr2len(p + n);
|
|
330
|
8254 c2 = mb_ptr2char(p + n);
|
|
|
8255 c3 = mb_ptr2char(p + n + fl);
|
|
|
8256 }
|
|
|
8257 else
|
|
|
8258 #endif
|
|
323
|
8259 {
|
|
330
|
8260 c = *p;
|
|
|
8261 c2 = p[1];
|
|
|
8262 c3 = p[2];
|
|
|
8263 }
|
|
|
8264
|
|
|
8265 /* When characters are identical: "121" then SWAP3 result is
|
|
|
8266 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
8267 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
8268 * swapping. Also skip when c3 is NUL. */
|
|
|
8269 if (c == c3 || c3 == NUL)
|
|
|
8270 {
|
|
|
8271 sp->ts_state = STATE_REP_INI;
|
|
|
8272 break;
|
|
|
8273 }
|
|
|
8274 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
8275 {
|
|
|
8276 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
8277 ++depth;
|
|
330
|
8278 #ifdef FEAT_MBYTE
|
|
|
8279 if (has_mbyte)
|
|
|
8280 {
|
|
|
8281 tl = mb_char2len(c3);
|
|
|
8282 mch_memmove(p, p + n + fl, tl);
|
|
|
8283 mb_char2bytes(c2, p + tl);
|
|
|
8284 mb_char2bytes(c, p + fl + tl);
|
|
|
8285 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
8286 }
|
|
|
8287 else
|
|
|
8288 #endif
|
|
|
8289 {
|
|
|
8290 p[0] = p[2];
|
|
|
8291 p[2] = c;
|
|
|
8292 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8293 }
|
|
323
|
8294 }
|
|
|
8295 else
|
|
|
8296 sp->ts_state = STATE_REP_INI;
|
|
|
8297 break;
|
|
|
8298
|
|
330
|
8299 case STATE_UNSWAP3:
|
|
|
8300 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
8301 p = fword + sp->ts_fidx;
|
|
|
8302 #ifdef FEAT_MBYTE
|
|
|
8303 if (has_mbyte)
|
|
|
8304 {
|
|
|
8305 n = MB_BYTE2LEN(*p);
|
|
|
8306 c2 = mb_ptr2char(p + n);
|
|
|
8307 fl = MB_BYTE2LEN(p[n]);
|
|
|
8308 c = mb_ptr2char(p + n + fl);
|
|
|
8309 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
8310 mch_memmove(p + fl + tl, p, n);
|
|
|
8311 mb_char2bytes(c, p);
|
|
|
8312 mb_char2bytes(c2, p + tl);
|
|
|
8313 }
|
|
|
8314 else
|
|
|
8315 #endif
|
|
|
8316 {
|
|
|
8317 c = *p;
|
|
|
8318 *p = p[2];
|
|
|
8319 p[2] = c;
|
|
|
8320 }
|
|
346
|
8321
|
|
330
|
8322 /* Rotate three characters left: "123" -> "231". We change
|
|
|
8323 * "fword" here, it's changed back afterwards. */
|
|
323
|
8324 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
8325 {
|
|
330
|
8326 sp->ts_state = STATE_UNROT3L;
|
|
323
|
8327 ++depth;
|
|
330
|
8328 p = fword + sp->ts_fidx;
|
|
|
8329 #ifdef FEAT_MBYTE
|
|
|
8330 if (has_mbyte)
|
|
|
8331 {
|
|
474
|
8332 n = mb_cptr2len(p);
|
|
330
|
8333 c = mb_ptr2char(p);
|
|
474
|
8334 fl = mb_cptr2len(p + n);
|
|
|
8335 fl += mb_cptr2len(p + n + fl);
|
|
330
|
8336 mch_memmove(p, p + n, fl);
|
|
|
8337 mb_char2bytes(c, p + fl);
|
|
|
8338 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
8339 }
|
|
|
8340 else
|
|
|
8341 #endif
|
|
|
8342 {
|
|
|
8343 c = *p;
|
|
|
8344 *p = p[1];
|
|
|
8345 p[1] = p[2];
|
|
|
8346 p[2] = c;
|
|
|
8347 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8348 }
|
|
323
|
8349 }
|
|
|
8350 else
|
|
|
8351 sp->ts_state = STATE_REP_INI;
|
|
|
8352 break;
|
|
|
8353
|
|
330
|
8354 case STATE_UNROT3L:
|
|
346
|
8355 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
8356 p = fword + sp->ts_fidx;
|
|
|
8357 #ifdef FEAT_MBYTE
|
|
|
8358 if (has_mbyte)
|
|
|
8359 {
|
|
|
8360 n = MB_BYTE2LEN(*p);
|
|
|
8361 n += MB_BYTE2LEN(p[n]);
|
|
|
8362 c = mb_ptr2char(p + n);
|
|
|
8363 tl = MB_BYTE2LEN(p[n]);
|
|
|
8364 mch_memmove(p + tl, p, n);
|
|
|
8365 mb_char2bytes(c, p);
|
|
|
8366 }
|
|
|
8367 else
|
|
|
8368 #endif
|
|
|
8369 {
|
|
|
8370 c = p[2];
|
|
|
8371 p[2] = p[1];
|
|
|
8372 p[1] = *p;
|
|
|
8373 *p = c;
|
|
|
8374 }
|
|
346
|
8375
|
|
323
|
8376 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
8377 * "fword" here, it's changed back afterwards. */
|
|
323
|
8378 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
8379 {
|
|
330
|
8380 sp->ts_state = STATE_UNROT3R;
|
|
323
|
8381 ++depth;
|
|
330
|
8382 p = fword + sp->ts_fidx;
|
|
|
8383 #ifdef FEAT_MBYTE
|
|
|
8384 if (has_mbyte)
|
|
|
8385 {
|
|
474
|
8386 n = mb_cptr2len(p);
|
|
|
8387 n += mb_cptr2len(p + n);
|
|
330
|
8388 c = mb_ptr2char(p + n);
|
|
474
|
8389 tl = mb_cptr2len(p + n);
|
|
330
|
8390 mch_memmove(p + tl, p, n);
|
|
|
8391 mb_char2bytes(c, p);
|
|
|
8392 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
8393 }
|
|
|
8394 else
|
|
|
8395 #endif
|
|
|
8396 {
|
|
|
8397 c = p[2];
|
|
|
8398 p[2] = p[1];
|
|
|
8399 p[1] = *p;
|
|
|
8400 *p = c;
|
|
|
8401 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
8402 }
|
|
323
|
8403 }
|
|
|
8404 else
|
|
|
8405 sp->ts_state = STATE_REP_INI;
|
|
|
8406 break;
|
|
|
8407
|
|
330
|
8408 case STATE_UNROT3R:
|
|
346
|
8409 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
8410 p = fword + sp->ts_fidx;
|
|
|
8411 #ifdef FEAT_MBYTE
|
|
|
8412 if (has_mbyte)
|
|
|
8413 {
|
|
|
8414 c = mb_ptr2char(p);
|
|
|
8415 tl = MB_BYTE2LEN(*p);
|
|
|
8416 n = MB_BYTE2LEN(p[tl]);
|
|
|
8417 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
8418 mch_memmove(p, p + tl, n);
|
|
|
8419 mb_char2bytes(c, p + n);
|
|
|
8420 }
|
|
|
8421 else
|
|
|
8422 #endif
|
|
|
8423 {
|
|
|
8424 c = *p;
|
|
|
8425 *p = p[1];
|
|
|
8426 p[1] = p[2];
|
|
|
8427 p[2] = c;
|
|
|
8428 }
|
|
323
|
8429 /*FALLTHROUGH*/
|
|
|
8430
|
|
|
8431 case STATE_REP_INI:
|
|
|
8432 /* Check if matching with REP items from the .aff file would
|
|
|
8433 * work. Quickly skip if there are no REP items or the score
|
|
|
8434 * is going to be too high anyway. */
|
|
|
8435 gap = &lp->lp_slang->sl_rep;
|
|
|
8436 if (gap->ga_len == 0
|
|
|
8437 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
8438 {
|
|
|
8439 sp->ts_state = STATE_FINAL;
|
|
|
8440 break;
|
|
|
8441 }
|
|
|
8442
|
|
|
8443 /* Use the first byte to quickly find the first entry that
|
|
330
|
8444 * may match. If the index is -1 there is none. */
|
|
323
|
8445 sp->ts_curi = lp->lp_slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
8446 if (sp->ts_curi < 0)
|
|
|
8447 {
|
|
|
8448 sp->ts_state = STATE_FINAL;
|
|
|
8449 break;
|
|
|
8450 }
|
|
|
8451
|
|
|
8452 sp->ts_state = STATE_REP;
|
|
|
8453 /*FALLTHROUGH*/
|
|
|
8454
|
|
|
8455 case STATE_REP:
|
|
|
8456 /* Try matching with REP items from the .aff file. For each
|
|
330
|
8457 * match replace the characters and check if the resulting
|
|
|
8458 * word is valid. */
|
|
323
|
8459 p = fword + sp->ts_fidx;
|
|
|
8460
|
|
|
8461 gap = &lp->lp_slang->sl_rep;
|
|
|
8462 while (sp->ts_curi < gap->ga_len)
|
|
|
8463 {
|
|
|
8464 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
8465 if (*ftp->ft_from != *p)
|
|
|
8466 {
|
|
|
8467 /* past possible matching entries */
|
|
|
8468 sp->ts_curi = gap->ga_len;
|
|
|
8469 break;
|
|
|
8470 }
|
|
|
8471 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
8472 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
8473 {
|
|
|
8474 /* Need to undo this afterwards. */
|
|
|
8475 sp->ts_state = STATE_REP_UNDO;
|
|
|
8476
|
|
|
8477 /* Change the "from" to the "to" string. */
|
|
|
8478 ++depth;
|
|
|
8479 fl = STRLEN(ftp->ft_from);
|
|
|
8480 tl = STRLEN(ftp->ft_to);
|
|
|
8481 if (fl != tl)
|
|
346
|
8482 {
|
|
323
|
8483 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
8484 repextra += tl - fl;
|
|
|
8485 }
|
|
323
|
8486 mch_memmove(p, ftp->ft_to, tl);
|
|
|
8487 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
8488 #ifdef FEAT_MBYTE
|
|
|
8489 stack[depth].ts_tcharlen = 0;
|
|
|
8490 #endif
|
|
323
|
8491 break;
|
|
|
8492 }
|
|
|
8493 }
|
|
|
8494
|
|
|
8495 if (sp->ts_curi >= gap->ga_len)
|
|
|
8496 /* No (more) matches. */
|
|
|
8497 sp->ts_state = STATE_FINAL;
|
|
|
8498
|
|
|
8499 break;
|
|
|
8500
|
|
|
8501 case STATE_REP_UNDO:
|
|
|
8502 /* Undo a REP replacement and continue with the next one. */
|
|
|
8503 ftp = (fromto_T *)lp->lp_slang->sl_rep.ga_data
|
|
|
8504 + sp->ts_curi - 1;
|
|
|
8505 fl = STRLEN(ftp->ft_from);
|
|
|
8506 tl = STRLEN(ftp->ft_to);
|
|
|
8507 p = fword + sp->ts_fidx;
|
|
|
8508 if (fl != tl)
|
|
346
|
8509 {
|
|
323
|
8510 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
8511 repextra -= tl - fl;
|
|
|
8512 }
|
|
323
|
8513 mch_memmove(p, ftp->ft_from, fl);
|
|
|
8514 sp->ts_state = STATE_REP;
|
|
|
8515 break;
|
|
|
8516
|
|
|
8517 default:
|
|
|
8518 /* Did all possible states at this level, go up one level. */
|
|
|
8519 --depth;
|
|
344
|
8520
|
|
375
|
8521 if (depth >= 0 && stack[depth].ts_prefixdepth == PREFIXTREE)
|
|
|
8522 {
|
|
|
8523 /* Continue in or go back to the prefix tree. */
|
|
|
8524 byts = pbyts;
|
|
|
8525 idxs = pidxs;
|
|
|
8526 splitoff = 0;
|
|
|
8527 }
|
|
|
8528
|
|
344
|
8529 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
8530 line_breakcheck();
|
|
323
|
8531 }
|
|
|
8532 }
|
|
|
8533 }
|
|
|
8534 }
|
|
|
8535
|
|
|
8536 /*
|
|
|
8537 * Try going one level deeper in the tree.
|
|
|
8538 */
|
|
|
8539 static int
|
|
|
8540 try_deeper(su, stack, depth, score_add)
|
|
|
8541 suginfo_T *su;
|
|
|
8542 trystate_T *stack;
|
|
|
8543 int depth;
|
|
|
8544 int score_add;
|
|
|
8545 {
|
|
|
8546 int newscore;
|
|
|
8547
|
|
|
8548 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
8549 newscore = stack[depth].ts_score + score_add;
|
|
|
8550 if (newscore >= su->su_maxscore)
|
|
|
8551 return FALSE;
|
|
|
8552
|
|
330
|
8553 stack[depth + 1] = stack[depth];
|
|
323
|
8554 stack[depth + 1].ts_state = STATE_START;
|
|
|
8555 stack[depth + 1].ts_score = newscore;
|
|
|
8556 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
8557 return TRUE;
|
|
|
8558 }
|
|
|
8559
|
|
455
|
8560 #ifdef FEAT_MBYTE
|
|
|
8561 /*
|
|
|
8562 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
8563 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
8564 */
|
|
|
8565 static int
|
|
|
8566 nofold_len(fword, flen, word)
|
|
|
8567 char_u *fword;
|
|
|
8568 int flen;
|
|
|
8569 char_u *word;
|
|
|
8570 {
|
|
|
8571 char_u *p;
|
|
|
8572 int i = 0;
|
|
|
8573
|
|
|
8574 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
8575 ++i;
|
|
|
8576 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
8577 --i;
|
|
|
8578 return (int)(p - word);
|
|
|
8579 }
|
|
|
8580 #endif
|
|
|
8581
|
|
323
|
8582 /*
|
|
|
8583 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
8584 * words and put it in "kword".
|
|
|
8585 * Theoretically there could be several keep-case words that result in the
|
|
|
8586 * same case-folded word, but we only find one...
|
|
|
8587 */
|
|
|
8588 static void
|
|
|
8589 find_keepcap_word(slang, fword, kword)
|
|
|
8590 slang_T *slang;
|
|
|
8591 char_u *fword;
|
|
|
8592 char_u *kword;
|
|
|
8593 {
|
|
|
8594 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
8595 int depth;
|
|
324
|
8596 idx_T tryidx;
|
|
323
|
8597
|
|
|
8598 /* The following arrays are used at each depth in the tree. */
|
|
324
|
8599 idx_T arridx[MAXWLEN];
|
|
323
|
8600 int round[MAXWLEN];
|
|
|
8601 int fwordidx[MAXWLEN];
|
|
|
8602 int uwordidx[MAXWLEN];
|
|
|
8603 int kwordlen[MAXWLEN];
|
|
|
8604
|
|
|
8605 int flen, ulen;
|
|
|
8606 int l;
|
|
|
8607 int len;
|
|
|
8608 int c;
|
|
324
|
8609 idx_T lo, hi, m;
|
|
323
|
8610 char_u *p;
|
|
|
8611 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
8612 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
8613
|
|
|
8614 if (byts == NULL)
|
|
|
8615 {
|
|
|
8616 /* array is empty: "cannot happen" */
|
|
|
8617 *kword = NUL;
|
|
|
8618 return;
|
|
|
8619 }
|
|
|
8620
|
|
|
8621 /* Make an all-cap version of "fword". */
|
|
|
8622 allcap_copy(fword, uword);
|
|
|
8623
|
|
|
8624 /*
|
|
|
8625 * Each character needs to be tried both case-folded and upper-case.
|
|
|
8626 * All this gets very complicated if we keep in mind that changing case
|
|
|
8627 * may change the byte length of a multi-byte character...
|
|
|
8628 */
|
|
|
8629 depth = 0;
|
|
|
8630 arridx[0] = 0;
|
|
|
8631 round[0] = 0;
|
|
|
8632 fwordidx[0] = 0;
|
|
|
8633 uwordidx[0] = 0;
|
|
|
8634 kwordlen[0] = 0;
|
|
|
8635 while (depth >= 0)
|
|
|
8636 {
|
|
|
8637 if (fword[fwordidx[depth]] == NUL)
|
|
|
8638 {
|
|
|
8639 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
8640 * here we have found a match. */
|
|
|
8641 if (byts[arridx[depth] + 1] == 0)
|
|
|
8642 {
|
|
|
8643 kword[kwordlen[depth]] = NUL;
|
|
|
8644 return;
|
|
|
8645 }
|
|
|
8646
|
|
|
8647 /* kword is getting too long, continue one level up */
|
|
|
8648 --depth;
|
|
|
8649 }
|
|
|
8650 else if (++round[depth] > 2)
|
|
|
8651 {
|
|
|
8652 /* tried both fold-case and upper-case character, continue one
|
|
|
8653 * level up */
|
|
|
8654 --depth;
|
|
|
8655 }
|
|
|
8656 else
|
|
|
8657 {
|
|
|
8658 /*
|
|
|
8659 * round[depth] == 1: Try using the folded-case character.
|
|
|
8660 * round[depth] == 2: Try using the upper-case character.
|
|
|
8661 */
|
|
|
8662 #ifdef FEAT_MBYTE
|
|
|
8663 if (has_mbyte)
|
|
|
8664 {
|
|
474
|
8665 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
8666 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
8667 }
|
|
|
8668 else
|
|
|
8669 #endif
|
|
|
8670 ulen = flen = 1;
|
|
|
8671 if (round[depth] == 1)
|
|
|
8672 {
|
|
|
8673 p = fword + fwordidx[depth];
|
|
|
8674 l = flen;
|
|
|
8675 }
|
|
|
8676 else
|
|
|
8677 {
|
|
|
8678 p = uword + uwordidx[depth];
|
|
|
8679 l = ulen;
|
|
|
8680 }
|
|
|
8681
|
|
|
8682 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
8683 {
|
|
|
8684 /* Perform a binary search in the list of accepted bytes. */
|
|
|
8685 len = byts[tryidx++];
|
|
|
8686 c = *p++;
|
|
|
8687 lo = tryidx;
|
|
|
8688 hi = tryidx + len - 1;
|
|
|
8689 while (lo < hi)
|
|
|
8690 {
|
|
|
8691 m = (lo + hi) / 2;
|
|
|
8692 if (byts[m] > c)
|
|
|
8693 hi = m - 1;
|
|
|
8694 else if (byts[m] < c)
|
|
|
8695 lo = m + 1;
|
|
|
8696 else
|
|
|
8697 {
|
|
|
8698 lo = hi = m;
|
|
|
8699 break;
|
|
|
8700 }
|
|
|
8701 }
|
|
|
8702
|
|
|
8703 /* Stop if there is no matching byte. */
|
|
|
8704 if (hi < lo || byts[lo] != c)
|
|
|
8705 break;
|
|
|
8706
|
|
|
8707 /* Continue at the child (if there is one). */
|
|
|
8708 tryidx = idxs[lo];
|
|
|
8709 }
|
|
|
8710
|
|
|
8711 if (l == 0)
|
|
|
8712 {
|
|
|
8713 /*
|
|
|
8714 * Found the matching char. Copy it to "kword" and go a
|
|
|
8715 * level deeper.
|
|
|
8716 */
|
|
|
8717 if (round[depth] == 1)
|
|
|
8718 {
|
|
|
8719 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
8720 flen);
|
|
|
8721 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
8722 }
|
|
|
8723 else
|
|
|
8724 {
|
|
|
8725 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
8726 ulen);
|
|
|
8727 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
8728 }
|
|
|
8729 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
8730 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
8731
|
|
|
8732 ++depth;
|
|
|
8733 arridx[depth] = tryidx;
|
|
|
8734 round[depth] = 0;
|
|
|
8735 }
|
|
|
8736 }
|
|
|
8737 }
|
|
|
8738
|
|
|
8739 /* Didn't find it: "cannot happen". */
|
|
|
8740 *kword = NUL;
|
|
|
8741 }
|
|
|
8742
|
|
|
8743 /*
|
|
344
|
8744 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
8745 * su->su_sga.
|
|
|
8746 */
|
|
|
8747 static void
|
|
|
8748 score_comp_sal(su)
|
|
|
8749 suginfo_T *su;
|
|
|
8750 {
|
|
|
8751 langp_T *lp;
|
|
|
8752 char_u badsound[MAXWLEN];
|
|
|
8753 int i;
|
|
|
8754 suggest_T *stp;
|
|
|
8755 suggest_T *sstp;
|
|
|
8756 int score;
|
|
|
8757
|
|
|
8758 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
8759 return;
|
|
|
8760
|
|
|
8761 /* Use the sound-folding of the first language that supports it. */
|
|
|
8762 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8763 lp->lp_slang != NULL; ++lp)
|
|
|
8764 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8765 {
|
|
|
8766 /* soundfold the bad word */
|
|
375
|
8767 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8768
|
|
|
8769 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8770 {
|
|
|
8771 stp = &SUG(su->su_ga, i);
|
|
|
8772
|
|
351
|
8773 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
8774 * sound-a-like score. */
|
|
|
8775 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
8776 if (score < SCORE_MAXMAX)
|
|
|
8777 {
|
|
|
8778 /* Add the suggestion. */
|
|
|
8779 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
8780 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
8781 if (sstp->st_word != NULL)
|
|
|
8782 {
|
|
|
8783 sstp->st_score = score;
|
|
|
8784 sstp->st_altscore = 0;
|
|
|
8785 sstp->st_orglen = stp->st_orglen;
|
|
|
8786 ++su->su_sga.ga_len;
|
|
|
8787 }
|
|
|
8788 }
|
|
|
8789 }
|
|
|
8790 break;
|
|
|
8791 }
|
|
|
8792 }
|
|
|
8793
|
|
|
8794 /*
|
|
|
8795 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
8796 * They are intwined.
|
|
|
8797 */
|
|
|
8798 static void
|
|
|
8799 score_combine(su)
|
|
|
8800 suginfo_T *su;
|
|
|
8801 {
|
|
|
8802 int i;
|
|
|
8803 int j;
|
|
|
8804 garray_T ga;
|
|
|
8805 garray_T *gap;
|
|
|
8806 langp_T *lp;
|
|
|
8807 suggest_T *stp;
|
|
|
8808 char_u *p;
|
|
|
8809 char_u badsound[MAXWLEN];
|
|
|
8810 int round;
|
|
|
8811
|
|
|
8812 /* Add the alternate score to su_ga. */
|
|
|
8813 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8814 lp->lp_slang != NULL; ++lp)
|
|
|
8815 {
|
|
|
8816 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8817 {
|
|
|
8818 /* soundfold the bad word */
|
|
375
|
8819 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8820
|
|
|
8821 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8822 {
|
|
|
8823 stp = &SUG(su->su_ga, i);
|
|
351
|
8824 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
8825 badsound);
|
|
344
|
8826 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8827 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
8828 else
|
|
|
8829 stp->st_score = (stp->st_score * 3
|
|
|
8830 + stp->st_altscore) / 4;
|
|
|
8831 stp->st_salscore = FALSE;
|
|
|
8832 }
|
|
|
8833 break;
|
|
|
8834 }
|
|
|
8835 }
|
|
|
8836
|
|
|
8837 /* Add the alternate score to su_sga. */
|
|
|
8838 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8839 {
|
|
|
8840 stp = &SUG(su->su_sga, i);
|
|
|
8841 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
8842 if (stp->st_score == SCORE_MAXMAX)
|
|
|
8843 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
8844 else
|
|
|
8845 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
8846 stp->st_salscore = TRUE;
|
|
|
8847 }
|
|
|
8848
|
|
|
8849 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
8850 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8851 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
8852
|
|
|
8853 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
8854 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
8855 return;
|
|
|
8856
|
|
|
8857 stp = &SUG(ga, 0);
|
|
|
8858 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
8859 {
|
|
|
8860 /* round 1: get a suggestion from su_ga
|
|
|
8861 * round 2: get a suggestion from su_sga */
|
|
|
8862 for (round = 1; round <= 2; ++round)
|
|
|
8863 {
|
|
|
8864 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
8865 if (i < gap->ga_len)
|
|
|
8866 {
|
|
|
8867 /* Don't add a word if it's already there. */
|
|
|
8868 p = SUG(*gap, i).st_word;
|
|
|
8869 for (j = 0; j < ga.ga_len; ++j)
|
|
|
8870 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
8871 break;
|
|
|
8872 if (j == ga.ga_len)
|
|
|
8873 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
8874 else
|
|
|
8875 vim_free(p);
|
|
|
8876 }
|
|
|
8877 }
|
|
|
8878 }
|
|
|
8879
|
|
|
8880 ga_clear(&su->su_ga);
|
|
|
8881 ga_clear(&su->su_sga);
|
|
|
8882
|
|
|
8883 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
8884 if (ga.ga_len > su->su_maxcount)
|
|
|
8885 {
|
|
|
8886 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
8887 vim_free(stp[i].st_word);
|
|
|
8888 ga.ga_len = su->su_maxcount;
|
|
|
8889 }
|
|
|
8890
|
|
|
8891 su->su_ga = ga;
|
|
|
8892 }
|
|
|
8893
|
|
|
8894 /*
|
|
351
|
8895 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
8896 * badword.
|
|
|
8897 */
|
|
|
8898 static int
|
|
|
8899 stp_sal_score(stp, su, slang, badsound)
|
|
|
8900 suggest_T *stp;
|
|
|
8901 suginfo_T *su;
|
|
|
8902 slang_T *slang;
|
|
|
8903 char_u *badsound; /* sound-folded badword */
|
|
|
8904 {
|
|
|
8905 char_u *p;
|
|
|
8906 char_u badsound2[MAXWLEN];
|
|
|
8907 char_u fword[MAXWLEN];
|
|
|
8908 char_u goodsound[MAXWLEN];
|
|
|
8909
|
|
|
8910 if (stp->st_orglen <= su->su_badlen)
|
|
|
8911 p = badsound;
|
|
|
8912 else
|
|
|
8913 {
|
|
|
8914 /* soundfold the bad word with more characters following */
|
|
|
8915 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
8916
|
|
|
8917 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
8918 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
8919 * removing the space. Don't do it when the good word also contains a
|
|
|
8920 * space. */
|
|
|
8921 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
8922 && *skiptowhite(stp->st_word) == NUL)
|
|
|
8923 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
8924 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
8925
|
|
375
|
8926 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
8927 p = badsound2;
|
|
|
8928 }
|
|
|
8929
|
|
375
|
8930 /* Sound-fold the word and compute the score for the difference. */
|
|
|
8931 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
8932
|
|
|
8933 return soundalike_score(goodsound, p);
|
|
|
8934 }
|
|
|
8935
|
|
|
8936 /*
|
|
323
|
8937 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
8938 */
|
|
|
8939 static void
|
|
346
|
8940 suggest_try_soundalike(su)
|
|
323
|
8941 suginfo_T *su;
|
|
|
8942 {
|
|
|
8943 char_u salword[MAXWLEN];
|
|
|
8944 char_u tword[MAXWLEN];
|
|
|
8945 char_u tsalword[MAXWLEN];
|
|
324
|
8946 idx_T arridx[MAXWLEN];
|
|
323
|
8947 int curi[MAXWLEN];
|
|
|
8948 langp_T *lp;
|
|
|
8949 char_u *byts;
|
|
324
|
8950 idx_T *idxs;
|
|
323
|
8951 int depth;
|
|
|
8952 int c;
|
|
324
|
8953 idx_T n;
|
|
323
|
8954 int round;
|
|
|
8955 int flags;
|
|
344
|
8956 int sound_score;
|
|
|
8957
|
|
|
8958 /* Do this for all languages that support sound folding. */
|
|
323
|
8959 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8960 lp->lp_slang != NULL; ++lp)
|
|
|
8961 {
|
|
|
8962 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8963 {
|
|
|
8964 /* soundfold the bad word */
|
|
375
|
8965 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
8966
|
|
|
8967 /*
|
|
|
8968 * Go through the whole tree, soundfold each word and compare.
|
|
|
8969 * round 1: use the case-folded tree.
|
|
|
8970 * round 2: use the keep-case tree.
|
|
|
8971 */
|
|
|
8972 for (round = 1; round <= 2; ++round)
|
|
|
8973 {
|
|
|
8974 if (round == 1)
|
|
|
8975 {
|
|
|
8976 byts = lp->lp_slang->sl_fbyts;
|
|
|
8977 idxs = lp->lp_slang->sl_fidxs;
|
|
|
8978 }
|
|
|
8979 else
|
|
|
8980 {
|
|
|
8981 byts = lp->lp_slang->sl_kbyts;
|
|
|
8982 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
8983 if (byts == NULL) /* no keep-case words */
|
|
|
8984 continue;
|
|
323
|
8985 }
|
|
|
8986
|
|
|
8987 depth = 0;
|
|
|
8988 arridx[0] = 0;
|
|
|
8989 curi[0] = 1;
|
|
|
8990 while (depth >= 0 && !got_int)
|
|
|
8991 {
|
|
|
8992 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
8993 {
|
|
323
|
8994 /* Done all bytes at this node, go up one level. */
|
|
|
8995 --depth;
|
|
351
|
8996 line_breakcheck();
|
|
|
8997 }
|
|
323
|
8998 else
|
|
|
8999 {
|
|
|
9000 /* Do one more byte at this node. */
|
|
|
9001 n = arridx[depth] + curi[depth];
|
|
|
9002 ++curi[depth];
|
|
|
9003 c = byts[n];
|
|
|
9004 if (c == 0)
|
|
|
9005 {
|
|
|
9006 /* End of word, deal with the word. */
|
|
324
|
9007 flags = (int)idxs[n];
|
|
323
|
9008 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
9009 {
|
|
|
9010 tword[depth] = NUL;
|
|
375
|
9011 /* Sound-fold. Only in keep-case tree need to
|
|
|
9012 * case-fold the word. */
|
|
|
9013 spell_soundfold(lp->lp_slang, tword,
|
|
|
9014 round == 1, tsalword);
|
|
323
|
9015
|
|
344
|
9016 /* Compute the edit distance between the
|
|
|
9017 * sound-a-like words. */
|
|
|
9018 sound_score = soundalike_score(salword,
|
|
|
9019 tsalword);
|
|
324
|
9020 if (sound_score < SCORE_MAXMAX)
|
|
323
|
9021 {
|
|
324
|
9022 char_u cword[MAXWLEN];
|
|
|
9023 char_u *p;
|
|
344
|
9024 int score;
|
|
324
|
9025
|
|
388
|
9026 flags |= su->su_badflags;
|
|
351
|
9027 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
9028 {
|
|
324
|
9029 /* Need to fix case according to
|
|
|
9030 * "flags". */
|
|
323
|
9031 make_case_word(tword, cword, flags);
|
|
324
|
9032 p = cword;
|
|
323
|
9033 }
|
|
|
9034 else
|
|
324
|
9035 p = tword;
|
|
|
9036
|
|
344
|
9037 if (sps_flags & SPS_DOUBLE)
|
|
|
9038 add_suggestion(su, &su->su_sga, p,
|
|
346
|
9039 su->su_badlen,
|
|
351
|
9040 sound_score, 0, FALSE);
|
|
344
|
9041 else
|
|
|
9042 {
|
|
|
9043 /* Compute the score. */
|
|
|
9044 score = spell_edit_score(
|
|
|
9045 su->su_badword, p);
|
|
|
9046 if (sps_flags & SPS_BEST)
|
|
|
9047 /* give a bonus for the good word
|
|
|
9048 * sounding the same as the bad
|
|
|
9049 * word */
|
|
|
9050 add_suggestion(su, &su->su_ga, p,
|
|
346
|
9051 su->su_badlen,
|
|
344
|
9052 RESCORE(score, sound_score),
|
|
351
|
9053 sound_score, TRUE);
|
|
344
|
9054 else
|
|
|
9055 add_suggestion(su, &su->su_ga, p,
|
|
346
|
9056 su->su_badlen,
|
|
351
|
9057 score + sound_score, 0, FALSE);
|
|
344
|
9058 }
|
|
323
|
9059 }
|
|
|
9060 }
|
|
|
9061
|
|
|
9062 /* Skip over other NUL bytes. */
|
|
|
9063 while (byts[n + 1] == 0)
|
|
|
9064 {
|
|
|
9065 ++n;
|
|
|
9066 ++curi[depth];
|
|
|
9067 }
|
|
|
9068 }
|
|
|
9069 else
|
|
|
9070 {
|
|
|
9071 /* Normal char, go one level deeper. */
|
|
|
9072 tword[depth++] = c;
|
|
|
9073 arridx[depth] = idxs[n];
|
|
|
9074 curi[depth] = 1;
|
|
|
9075 }
|
|
|
9076 }
|
|
|
9077 }
|
|
|
9078 }
|
|
|
9079 }
|
|
|
9080 }
|
|
|
9081 }
|
|
|
9082
|
|
|
9083 /*
|
|
324
|
9084 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
9085 */
|
|
|
9086 static void
|
|
|
9087 make_case_word(fword, cword, flags)
|
|
|
9088 char_u *fword;
|
|
|
9089 char_u *cword;
|
|
|
9090 int flags;
|
|
|
9091 {
|
|
|
9092 if (flags & WF_ALLCAP)
|
|
|
9093 /* Make it all upper-case */
|
|
|
9094 allcap_copy(fword, cword);
|
|
|
9095 else if (flags & WF_ONECAP)
|
|
|
9096 /* Make the first letter upper-case */
|
|
324
|
9097 onecap_copy(fword, cword, TRUE);
|
|
323
|
9098 else
|
|
|
9099 /* Use goodword as-is. */
|
|
|
9100 STRCPY(cword, fword);
|
|
|
9101 }
|
|
|
9102
|
|
330
|
9103 /*
|
|
|
9104 * Use map string "map" for languages "lp".
|
|
|
9105 */
|
|
|
9106 static void
|
|
|
9107 set_map_str(lp, map)
|
|
|
9108 slang_T *lp;
|
|
|
9109 char_u *map;
|
|
|
9110 {
|
|
|
9111 char_u *p;
|
|
|
9112 int headc = 0;
|
|
|
9113 int c;
|
|
|
9114 int i;
|
|
|
9115
|
|
|
9116 if (*map == NUL)
|
|
|
9117 {
|
|
|
9118 lp->sl_has_map = FALSE;
|
|
|
9119 return;
|
|
|
9120 }
|
|
|
9121 lp->sl_has_map = TRUE;
|
|
|
9122
|
|
|
9123 /* Init the array and hash table empty. */
|
|
|
9124 for (i = 0; i < 256; ++i)
|
|
|
9125 lp->sl_map_array[i] = 0;
|
|
|
9126 #ifdef FEAT_MBYTE
|
|
|
9127 hash_init(&lp->sl_map_hash);
|
|
|
9128 #endif
|
|
|
9129
|
|
|
9130 /*
|
|
|
9131 * The similar characters are stored separated with slashes:
|
|
|
9132 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
9133 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
9134 */
|
|
|
9135 for (p = map; *p != NUL; )
|
|
|
9136 {
|
|
|
9137 #ifdef FEAT_MBYTE
|
|
474
|
9138 c = mb_cptr2char_adv(&p);
|
|
330
|
9139 #else
|
|
|
9140 c = *p++;
|
|
|
9141 #endif
|
|
|
9142 if (c == '/')
|
|
|
9143 headc = 0;
|
|
|
9144 else
|
|
|
9145 {
|
|
|
9146 if (headc == 0)
|
|
|
9147 headc = c;
|
|
|
9148
|
|
|
9149 #ifdef FEAT_MBYTE
|
|
|
9150 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
9151 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
9152 * a NUL. */
|
|
|
9153 if (c >= 256)
|
|
|
9154 {
|
|
|
9155 int cl = mb_char2len(c);
|
|
|
9156 int headcl = mb_char2len(headc);
|
|
|
9157 char_u *b;
|
|
|
9158 hash_T hash;
|
|
|
9159 hashitem_T *hi;
|
|
|
9160
|
|
|
9161 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
9162 if (b == NULL)
|
|
|
9163 return;
|
|
|
9164 mb_char2bytes(c, b);
|
|
|
9165 b[cl] = NUL;
|
|
|
9166 mb_char2bytes(headc, b + cl + 1);
|
|
|
9167 b[cl + 1 + headcl] = NUL;
|
|
|
9168 hash = hash_hash(b);
|
|
|
9169 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
9170 if (HASHITEM_EMPTY(hi))
|
|
|
9171 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
9172 else
|
|
|
9173 {
|
|
|
9174 /* This should have been checked when generating the .spl
|
|
|
9175 * file. */
|
|
|
9176 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
9177 vim_free(b);
|
|
|
9178 }
|
|
|
9179 }
|
|
|
9180 else
|
|
|
9181 #endif
|
|
|
9182 lp->sl_map_array[c] = headc;
|
|
|
9183 }
|
|
|
9184 }
|
|
|
9185 }
|
|
|
9186
|
|
323
|
9187 /*
|
|
|
9188 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
9189 * lines in the .aff file.
|
|
|
9190 */
|
|
|
9191 static int
|
|
|
9192 similar_chars(slang, c1, c2)
|
|
|
9193 slang_T *slang;
|
|
|
9194 int c1;
|
|
|
9195 int c2;
|
|
|
9196 {
|
|
330
|
9197 int m1, m2;
|
|
|
9198 #ifdef FEAT_MBYTE
|
|
|
9199 char_u buf[MB_MAXBYTES];
|
|
|
9200 hashitem_T *hi;
|
|
|
9201
|
|
|
9202 if (c1 >= 256)
|
|
|
9203 {
|
|
|
9204 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
9205 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
9206 if (HASHITEM_EMPTY(hi))
|
|
|
9207 m1 = 0;
|
|
|
9208 else
|
|
|
9209 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
9210 }
|
|
|
9211 else
|
|
|
9212 #endif
|
|
|
9213 m1 = slang->sl_map_array[c1];
|
|
|
9214 if (m1 == 0)
|
|
323
|
9215 return FALSE;
|
|
330
|
9216
|
|
|
9217
|
|
|
9218 #ifdef FEAT_MBYTE
|
|
|
9219 if (c2 >= 256)
|
|
|
9220 {
|
|
|
9221 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
9222 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
9223 if (HASHITEM_EMPTY(hi))
|
|
|
9224 m2 = 0;
|
|
|
9225 else
|
|
|
9226 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
9227 }
|
|
|
9228 else
|
|
|
9229 #endif
|
|
|
9230 m2 = slang->sl_map_array[c2];
|
|
|
9231
|
|
|
9232 return m1 == m2;
|
|
323
|
9233 }
|
|
|
9234
|
|
|
9235 /*
|
|
|
9236 * Add a suggestion to the list of suggestions.
|
|
|
9237 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
9238 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
9239 * with spell_edit_score().
|
|
|
9240 */
|
|
|
9241 static void
|
|
351
|
9242 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
9243 suginfo_T *su;
|
|
344
|
9244 garray_T *gap;
|
|
323
|
9245 char_u *goodword;
|
|
346
|
9246 int badlen; /* length of bad word used */
|
|
324
|
9247 int score;
|
|
351
|
9248 int altscore;
|
|
344
|
9249 int had_bonus; /* value for st_had_bonus */
|
|
323
|
9250 {
|
|
|
9251 suggest_T *stp;
|
|
|
9252 int i;
|
|
346
|
9253 char_u *p = NULL;
|
|
|
9254 int c = 0;
|
|
323
|
9255
|
|
|
9256 /* Check that the word wasn't banned. */
|
|
|
9257 if (was_banned(su, goodword))
|
|
|
9258 return;
|
|
|
9259
|
|
346
|
9260 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
9261 * Remove the common part from "goodword". */
|
|
|
9262 i = badlen - su->su_badlen;
|
|
|
9263 if (i > 0)
|
|
|
9264 {
|
|
|
9265 /* This assumes there was no case folding or it didn't change the
|
|
|
9266 * length... */
|
|
|
9267 p = goodword + STRLEN(goodword) - i;
|
|
|
9268 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
9269 {
|
|
|
9270 badlen = su->su_badlen;
|
|
|
9271 c = *p;
|
|
|
9272 *p = NUL;
|
|
|
9273 }
|
|
|
9274 else
|
|
|
9275 p = NULL;
|
|
|
9276 }
|
|
372
|
9277 else if (i < 0)
|
|
|
9278 {
|
|
|
9279 /* When replacing part of the word check that we actually change
|
|
|
9280 * something. For "the the" a suggestion can be replacing the first
|
|
|
9281 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
9282 if (badlen == (int)STRLEN(goodword)
|
|
372
|
9283 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
9284 return;
|
|
|
9285 }
|
|
|
9286
|
|
346
|
9287
|
|
323
|
9288 if (score <= su->su_maxscore)
|
|
|
9289 {
|
|
366
|
9290 /* Check if the word is already there. Also check the length that is
|
|
|
9291 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
9292 * "thes" -> "these". */
|
|
344
|
9293 stp = &SUG(*gap, 0);
|
|
|
9294 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
9295 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
9296 && stp[i].st_orglen == badlen)
|
|
323
|
9297 {
|
|
|
9298 /* Found it. Remember the lowest score. */
|
|
|
9299 if (stp[i].st_score > score)
|
|
324
|
9300 {
|
|
323
|
9301 stp[i].st_score = score;
|
|
376
|
9302 stp[i].st_altscore = altscore;
|
|
324
|
9303 stp[i].st_had_bonus = had_bonus;
|
|
|
9304 }
|
|
323
|
9305 break;
|
|
|
9306 }
|
|
|
9307
|
|
344
|
9308 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
9309 {
|
|
|
9310 /* Add a suggestion. */
|
|
344
|
9311 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
9312 stp->st_word = vim_strsave(goodword);
|
|
|
9313 if (stp->st_word != NULL)
|
|
|
9314 {
|
|
|
9315 stp->st_score = score;
|
|
351
|
9316 stp->st_altscore = altscore;
|
|
324
|
9317 stp->st_had_bonus = had_bonus;
|
|
346
|
9318 stp->st_orglen = badlen;
|
|
344
|
9319 ++gap->ga_len;
|
|
323
|
9320
|
|
|
9321 /* If we have too many suggestions now, sort the list and keep
|
|
|
9322 * the best suggestions. */
|
|
344
|
9323 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
9324 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
9325 SUG_CLEAN_COUNT(su));
|
|
323
|
9326 }
|
|
|
9327 }
|
|
|
9328 }
|
|
346
|
9329
|
|
|
9330 if (p != NULL)
|
|
|
9331 *p = c; /* restore "goodword" */
|
|
323
|
9332 }
|
|
|
9333
|
|
|
9334 /*
|
|
|
9335 * Add a word to be banned.
|
|
|
9336 */
|
|
|
9337 static void
|
|
|
9338 add_banned(su, word)
|
|
|
9339 suginfo_T *su;
|
|
|
9340 char_u *word;
|
|
|
9341 {
|
|
|
9342 char_u *s = vim_strsave(word);
|
|
|
9343 hash_T hash;
|
|
|
9344 hashitem_T *hi;
|
|
|
9345
|
|
|
9346 if (s != NULL)
|
|
|
9347 {
|
|
|
9348 hash = hash_hash(s);
|
|
|
9349 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
9350 if (HASHITEM_EMPTY(hi))
|
|
|
9351 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
9352 else
|
|
|
9353 vim_free(s);
|
|
323
|
9354 }
|
|
|
9355 }
|
|
|
9356
|
|
|
9357 /*
|
|
|
9358 * Return TRUE if a word appears in the list of banned words.
|
|
|
9359 */
|
|
|
9360 static int
|
|
|
9361 was_banned(su, word)
|
|
|
9362 suginfo_T *su;
|
|
|
9363 char_u *word;
|
|
|
9364 {
|
|
324
|
9365 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
9366
|
|
|
9367 return !HASHITEM_EMPTY(hi);
|
|
323
|
9368 }
|
|
|
9369
|
|
|
9370 /*
|
|
|
9371 * Free the banned words in "su".
|
|
|
9372 */
|
|
|
9373 static void
|
|
|
9374 free_banned(su)
|
|
|
9375 suginfo_T *su;
|
|
|
9376 {
|
|
|
9377 int todo;
|
|
|
9378 hashitem_T *hi;
|
|
|
9379
|
|
|
9380 todo = su->su_banned.ht_used;
|
|
|
9381 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
9382 {
|
|
|
9383 if (!HASHITEM_EMPTY(hi))
|
|
|
9384 {
|
|
|
9385 vim_free(hi->hi_key);
|
|
|
9386 --todo;
|
|
|
9387 }
|
|
|
9388 }
|
|
|
9389 hash_clear(&su->su_banned);
|
|
|
9390 }
|
|
|
9391
|
|
324
|
9392 /*
|
|
|
9393 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
9394 * thus only done for the final results.
|
|
|
9395 */
|
|
|
9396 static void
|
|
|
9397 rescore_suggestions(su)
|
|
|
9398 suginfo_T *su;
|
|
|
9399 {
|
|
|
9400 langp_T *lp;
|
|
|
9401 suggest_T *stp;
|
|
|
9402 char_u sal_badword[MAXWLEN];
|
|
|
9403 int i;
|
|
|
9404
|
|
|
9405 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9406 lp->lp_slang != NULL; ++lp)
|
|
|
9407 {
|
|
|
9408 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9409 {
|
|
|
9410 /* soundfold the bad word */
|
|
375
|
9411 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
9412
|
|
|
9413 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9414 {
|
|
344
|
9415 stp = &SUG(su->su_ga, i);
|
|
324
|
9416 if (!stp->st_had_bonus)
|
|
|
9417 {
|
|
351
|
9418 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
9419 lp->lp_slang, sal_badword);
|
|
|
9420 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
9421 stp->st_altscore = SCORE_BIG;
|
|
|
9422 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
9423 }
|
|
|
9424 }
|
|
|
9425 break;
|
|
|
9426 }
|
|
|
9427 }
|
|
|
9428 }
|
|
|
9429
|
|
323
|
9430 static int
|
|
|
9431 #ifdef __BORLANDC__
|
|
|
9432 _RTLENTRYF
|
|
|
9433 #endif
|
|
|
9434 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
9435
|
|
|
9436 /*
|
|
|
9437 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
9438 */
|
|
|
9439 static int
|
|
|
9440 #ifdef __BORLANDC__
|
|
|
9441 _RTLENTRYF
|
|
|
9442 #endif
|
|
|
9443 sug_compare(s1, s2)
|
|
|
9444 const void *s1;
|
|
|
9445 const void *s2;
|
|
|
9446 {
|
|
|
9447 suggest_T *p1 = (suggest_T *)s1;
|
|
|
9448 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
9449 int n = p1->st_score - p2->st_score;
|
|
|
9450
|
|
|
9451 if (n == 0)
|
|
|
9452 return p1->st_altscore - p2->st_altscore;
|
|
|
9453 return n;
|
|
323
|
9454 }
|
|
|
9455
|
|
|
9456 /*
|
|
|
9457 * Cleanup the suggestions:
|
|
|
9458 * - Sort on score.
|
|
|
9459 * - Remove words that won't be displayed.
|
|
344
|
9460 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
9461 */
|
|
344
|
9462 static int
|
|
|
9463 cleanup_suggestions(gap, maxscore, keep)
|
|
|
9464 garray_T *gap;
|
|
|
9465 int maxscore;
|
|
324
|
9466 int keep; /* nr of suggestions to keep */
|
|
323
|
9467 {
|
|
344
|
9468 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
9469 int i;
|
|
|
9470
|
|
|
9471 /* Sort the list. */
|
|
344
|
9472 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
9473
|
|
|
9474 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
9475 if (gap->ga_len > keep)
|
|
323
|
9476 {
|
|
344
|
9477 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
9478 vim_free(stp[i].st_word);
|
|
344
|
9479 gap->ga_len = keep;
|
|
|
9480 return stp[keep - 1].st_score;
|
|
323
|
9481 }
|
|
344
|
9482 return maxscore;
|
|
323
|
9483 }
|
|
|
9484
|
|
372
|
9485 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
9486 /*
|
|
|
9487 * Soundfold a string, for soundfold().
|
|
|
9488 * Result is in allocated memory, NULL for an error.
|
|
|
9489 */
|
|
|
9490 char_u *
|
|
|
9491 eval_soundfold(word)
|
|
|
9492 char_u *word;
|
|
|
9493 {
|
|
|
9494 langp_T *lp;
|
|
|
9495 char_u sound[MAXWLEN];
|
|
|
9496
|
|
|
9497 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
9498 /* Use the sound-folding of the first language that supports it. */
|
|
|
9499 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9500 lp->lp_slang != NULL; ++lp)
|
|
|
9501 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9502 {
|
|
|
9503 /* soundfold the word */
|
|
375
|
9504 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
9505 return vim_strsave(sound);
|
|
|
9506 }
|
|
|
9507
|
|
|
9508 /* No language with sound folding, return word as-is. */
|
|
|
9509 return vim_strsave(word);
|
|
|
9510 }
|
|
|
9511 #endif
|
|
|
9512
|
|
323
|
9513 /*
|
|
|
9514 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
9515 */
|
|
|
9516 static void
|
|
375
|
9517 spell_soundfold(slang, inword, folded, res)
|
|
|
9518 slang_T *slang;
|
|
|
9519 char_u *inword;
|
|
|
9520 int folded; /* "inword" is already case-folded */
|
|
|
9521 char_u *res;
|
|
|
9522 {
|
|
|
9523 char_u fword[MAXWLEN];
|
|
|
9524 char_u *word;
|
|
|
9525
|
|
|
9526 if (slang->sl_sofo)
|
|
|
9527 /* SOFOFROM and SOFOTO used */
|
|
|
9528 spell_soundfold_sofo(slang, inword, res);
|
|
|
9529 else
|
|
|
9530 {
|
|
|
9531 /* SAL items used. Requires the word to be case-folded. */
|
|
|
9532 if (folded)
|
|
|
9533 word = inword;
|
|
|
9534 else
|
|
|
9535 {
|
|
|
9536 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
9537 word = fword;
|
|
|
9538 }
|
|
|
9539
|
|
|
9540 #ifdef FEAT_MBYTE
|
|
|
9541 if (has_mbyte)
|
|
|
9542 spell_soundfold_wsal(slang, word, res);
|
|
|
9543 else
|
|
|
9544 #endif
|
|
|
9545 spell_soundfold_sal(slang, word, res);
|
|
|
9546 }
|
|
|
9547 }
|
|
|
9548
|
|
|
9549 /*
|
|
|
9550 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
9551 * SOFOTO lines.
|
|
|
9552 */
|
|
|
9553 static void
|
|
|
9554 spell_soundfold_sofo(slang, inword, res)
|
|
|
9555 slang_T *slang;
|
|
|
9556 char_u *inword;
|
|
|
9557 char_u *res;
|
|
|
9558 {
|
|
|
9559 char_u *s;
|
|
|
9560 int ri = 0;
|
|
|
9561 int c;
|
|
|
9562
|
|
|
9563 #ifdef FEAT_MBYTE
|
|
|
9564 if (has_mbyte)
|
|
|
9565 {
|
|
|
9566 int prevc = 0;
|
|
|
9567 int *ip;
|
|
|
9568
|
|
|
9569 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
9570 * 255, sl_sal the rest. */
|
|
|
9571 for (s = inword; *s != NUL; )
|
|
|
9572 {
|
|
474
|
9573 c = mb_cptr2char_adv(&s);
|
|
375
|
9574 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9575 c = ' ';
|
|
|
9576 else if (c < 256)
|
|
|
9577 c = slang->sl_sal_first[c];
|
|
|
9578 else
|
|
|
9579 {
|
|
|
9580 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
9581 if (ip == NULL) /* empty list, can't match */
|
|
|
9582 c = NUL;
|
|
|
9583 else
|
|
|
9584 for (;;) /* find "c" in the list */
|
|
|
9585 {
|
|
|
9586 if (*ip == 0) /* not found */
|
|
|
9587 {
|
|
|
9588 c = NUL;
|
|
|
9589 break;
|
|
|
9590 }
|
|
|
9591 if (*ip == c) /* match! */
|
|
|
9592 {
|
|
|
9593 c = ip[1];
|
|
|
9594 break;
|
|
|
9595 }
|
|
|
9596 ip += 2;
|
|
|
9597 }
|
|
|
9598 }
|
|
|
9599
|
|
|
9600 if (c != NUL && c != prevc)
|
|
|
9601 {
|
|
|
9602 ri += mb_char2bytes(c, res + ri);
|
|
|
9603 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
9604 break;
|
|
|
9605 prevc = c;
|
|
|
9606 }
|
|
|
9607 }
|
|
|
9608 }
|
|
|
9609 else
|
|
|
9610 #endif
|
|
|
9611 {
|
|
|
9612 /* The sl_sal_first[] table contains the translation. */
|
|
|
9613 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
9614 {
|
|
|
9615 if (vim_iswhite(c))
|
|
|
9616 c = ' ';
|
|
|
9617 else
|
|
|
9618 c = slang->sl_sal_first[c];
|
|
|
9619 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
9620 res[ri++] = c;
|
|
|
9621 }
|
|
|
9622 }
|
|
|
9623
|
|
|
9624 res[ri] = NUL;
|
|
|
9625 }
|
|
|
9626
|
|
|
9627 static void
|
|
|
9628 spell_soundfold_sal(slang, inword, res)
|
|
323
|
9629 slang_T *slang;
|
|
|
9630 char_u *inword;
|
|
|
9631 char_u *res;
|
|
|
9632 {
|
|
344
|
9633 salitem_T *smp;
|
|
323
|
9634 char_u word[MAXWLEN];
|
|
375
|
9635 char_u *s = inword;
|
|
323
|
9636 char_u *t;
|
|
344
|
9637 char_u *pf;
|
|
323
|
9638 int i, j, z;
|
|
344
|
9639 int reslen;
|
|
323
|
9640 int n, k = 0;
|
|
|
9641 int z0;
|
|
|
9642 int k0;
|
|
|
9643 int n0;
|
|
|
9644 int c;
|
|
|
9645 int pri;
|
|
|
9646 int p0 = -333;
|
|
|
9647 int c0;
|
|
|
9648
|
|
324
|
9649 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
9650 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
9651 if (slang->sl_rem_accents)
|
|
|
9652 {
|
|
|
9653 t = word;
|
|
375
|
9654 while (*s != NUL)
|
|
323
|
9655 {
|
|
324
|
9656 if (vim_iswhite(*s))
|
|
344
|
9657 {
|
|
|
9658 *t++ = ' ';
|
|
|
9659 s = skipwhite(s);
|
|
|
9660 }
|
|
323
|
9661 else
|
|
|
9662 {
|
|
376
|
9663 if (spell_iswordp_nmw(s))
|
|
323
|
9664 *t++ = *s;
|
|
|
9665 ++s;
|
|
|
9666 }
|
|
|
9667 }
|
|
|
9668 *t = NUL;
|
|
|
9669 }
|
|
|
9670 else
|
|
375
|
9671 STRCPY(word, s);
|
|
323
|
9672
|
|
344
|
9673 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
9674
|
|
|
9675 /*
|
|
|
9676 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
9677 * Changed to keep spaces.
|
|
323
|
9678 */
|
|
344
|
9679 i = reslen = z = 0;
|
|
323
|
9680 while ((c = word[i]) != NUL)
|
|
|
9681 {
|
|
344
|
9682 /* Start with the first rule that has the character in the word. */
|
|
323
|
9683 n = slang->sl_sal_first[c];
|
|
|
9684 z0 = 0;
|
|
|
9685
|
|
|
9686 if (n >= 0)
|
|
|
9687 {
|
|
|
9688 /* check all rules for the same letter */
|
|
344
|
9689 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
9690 {
|
|
344
|
9691 /* Quickly skip entries that don't match the word. Most
|
|
|
9692 * entries are less then three chars, optimize for that. */
|
|
|
9693 k = smp[n].sm_leadlen;
|
|
|
9694 if (k > 1)
|
|
323
|
9695 {
|
|
344
|
9696 if (word[i + 1] != s[1])
|
|
|
9697 continue;
|
|
|
9698 if (k > 2)
|
|
|
9699 {
|
|
|
9700 for (j = 2; j < k; ++j)
|
|
|
9701 if (word[i + j] != s[j])
|
|
|
9702 break;
|
|
|
9703 if (j < k)
|
|
|
9704 continue;
|
|
|
9705 }
|
|
323
|
9706 }
|
|
|
9707
|
|
375
|
9708 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
9709 {
|
|
375
|
9710 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
9711 while (*pf != NUL && *pf != word[i + k])
|
|
|
9712 ++pf;
|
|
|
9713 if (*pf == NUL)
|
|
|
9714 continue;
|
|
|
9715 ++k;
|
|
323
|
9716 }
|
|
344
|
9717 s = smp[n].sm_rules;
|
|
|
9718 pri = 5; /* default priority */
|
|
323
|
9719
|
|
|
9720 p0 = *s;
|
|
|
9721 k0 = k;
|
|
|
9722 while (*s == '-' && k > 1)
|
|
|
9723 {
|
|
|
9724 k--;
|
|
|
9725 s++;
|
|
|
9726 }
|
|
|
9727 if (*s == '<')
|
|
|
9728 s++;
|
|
344
|
9729 if (VIM_ISDIGIT(*s))
|
|
323
|
9730 {
|
|
|
9731 /* determine priority */
|
|
|
9732 pri = *s - '0';
|
|
|
9733 s++;
|
|
|
9734 }
|
|
|
9735 if (*s == '^' && *(s + 1) == '^')
|
|
|
9736 s++;
|
|
|
9737
|
|
|
9738 if (*s == NUL
|
|
|
9739 || (*s == '^'
|
|
324
|
9740 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9741 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
9742 && (*(s + 1) != '$'
|
|
376
|
9743 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9744 || (*s == '$' && i > 0
|
|
376
|
9745 && spell_iswordp(word + i - 1, curbuf)
|
|
|
9746 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9747 {
|
|
|
9748 /* search for followup rules, if: */
|
|
|
9749 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9750 c0 = word[i + k - 1];
|
|
|
9751 n0 = slang->sl_sal_first[c0];
|
|
|
9752
|
|
|
9753 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
9754 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
9755 {
|
|
|
9756 /* test follow-up rule for "word[i + k]" */
|
|
344
|
9757 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
9758 {
|
|
344
|
9759 /* Quickly skip entries that don't match the word.
|
|
|
9760 * */
|
|
|
9761 k0 = smp[n0].sm_leadlen;
|
|
|
9762 if (k0 > 1)
|
|
323
|
9763 {
|
|
344
|
9764 if (word[i + k] != s[1])
|
|
|
9765 continue;
|
|
|
9766 if (k0 > 2)
|
|
|
9767 {
|
|
|
9768 pf = word + i + k + 1;
|
|
|
9769 for (j = 2; j < k0; ++j)
|
|
|
9770 if (*pf++ != s[j])
|
|
|
9771 break;
|
|
|
9772 if (j < k0)
|
|
|
9773 continue;
|
|
|
9774 }
|
|
323
|
9775 }
|
|
344
|
9776 k0 += k - 1;
|
|
|
9777
|
|
375
|
9778 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
9779 {
|
|
344
|
9780 /* Check for match with one of the chars in
|
|
375
|
9781 * "sm_oneof". */
|
|
344
|
9782 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9783 ++pf;
|
|
|
9784 if (*pf == NUL)
|
|
|
9785 continue;
|
|
|
9786 ++k0;
|
|
323
|
9787 }
|
|
344
|
9788
|
|
|
9789 p0 = 5;
|
|
|
9790 s = smp[n0].sm_rules;
|
|
323
|
9791 while (*s == '-')
|
|
|
9792 {
|
|
344
|
9793 /* "k0" gets NOT reduced because
|
|
|
9794 * "if (k0 == k)" */
|
|
323
|
9795 s++;
|
|
|
9796 }
|
|
|
9797 if (*s == '<')
|
|
|
9798 s++;
|
|
344
|
9799 if (VIM_ISDIGIT(*s))
|
|
323
|
9800 {
|
|
|
9801 p0 = *s - '0';
|
|
|
9802 s++;
|
|
|
9803 }
|
|
|
9804
|
|
|
9805 if (*s == NUL
|
|
|
9806 /* *s == '^' cuts */
|
|
|
9807 || (*s == '$'
|
|
376
|
9808 && !spell_iswordp(word + i + k0,
|
|
|
9809 curbuf)))
|
|
323
|
9810 {
|
|
|
9811 if (k0 == k)
|
|
|
9812 /* this is just a piece of the string */
|
|
|
9813 continue;
|
|
|
9814
|
|
|
9815 if (p0 < pri)
|
|
|
9816 /* priority too low */
|
|
|
9817 continue;
|
|
|
9818 /* rule fits; stop search */
|
|
|
9819 break;
|
|
|
9820 }
|
|
|
9821 }
|
|
|
9822
|
|
344
|
9823 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
9824 continue;
|
|
|
9825 }
|
|
|
9826
|
|
|
9827 /* replace string */
|
|
344
|
9828 s = smp[n].sm_to;
|
|
389
|
9829 if (s == NULL)
|
|
|
9830 s = (char_u *)"";
|
|
344
|
9831 pf = smp[n].sm_rules;
|
|
|
9832 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
9833 if (p0 == 1 && z == 0)
|
|
|
9834 {
|
|
|
9835 /* rule with '<' is used */
|
|
344
|
9836 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
9837 || res[reslen - 1] == *s))
|
|
|
9838 reslen--;
|
|
323
|
9839 z0 = 1;
|
|
|
9840 z = 1;
|
|
|
9841 k0 = 0;
|
|
372
|
9842 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
9843 {
|
|
|
9844 word[i + k0] = *s;
|
|
|
9845 k0++;
|
|
|
9846 s++;
|
|
|
9847 }
|
|
|
9848 if (k > k0)
|
|
|
9849 mch_memmove(word + i + k0, word + i + k,
|
|
|
9850 STRLEN(word + i + k) + 1);
|
|
|
9851
|
|
|
9852 /* new "actual letter" */
|
|
|
9853 c = word[i];
|
|
|
9854 }
|
|
|
9855 else
|
|
|
9856 {
|
|
|
9857 /* no '<' rule used */
|
|
|
9858 i += k - 1;
|
|
|
9859 z = 0;
|
|
344
|
9860 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
9861 {
|
|
344
|
9862 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
9863 res[reslen++] = *s;
|
|
323
|
9864 s++;
|
|
|
9865 }
|
|
|
9866 /* new "actual letter" */
|
|
|
9867 c = *s;
|
|
344
|
9868 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
9869 {
|
|
|
9870 if (c != NUL)
|
|
372
|
9871 res[reslen++] = c;
|
|
323
|
9872 mch_memmove(word, word + i + 1,
|
|
|
9873 STRLEN(word + i + 1) + 1);
|
|
|
9874 i = 0;
|
|
|
9875 z0 = 1;
|
|
|
9876 }
|
|
|
9877 }
|
|
|
9878 break;
|
|
|
9879 }
|
|
|
9880 }
|
|
|
9881 }
|
|
324
|
9882 else if (vim_iswhite(c))
|
|
|
9883 {
|
|
|
9884 c = ' ';
|
|
|
9885 k = 1;
|
|
|
9886 }
|
|
323
|
9887
|
|
|
9888 if (z0 == 0)
|
|
|
9889 {
|
|
344
|
9890 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9891 && (!slang->sl_collapse || reslen == 0
|
|
|
9892 || res[reslen - 1] != c))
|
|
323
|
9893 /* condense only double letters */
|
|
372
|
9894 res[reslen++] = c;
|
|
323
|
9895
|
|
|
9896 i++;
|
|
|
9897 z = 0;
|
|
|
9898 k = 0;
|
|
|
9899 }
|
|
|
9900 }
|
|
|
9901
|
|
344
|
9902 res[reslen] = NUL;
|
|
323
|
9903 }
|
|
|
9904
|
|
372
|
9905 #ifdef FEAT_MBYTE
|
|
|
9906 /*
|
|
|
9907 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
9908 * Multi-byte version of spell_soundfold().
|
|
|
9909 */
|
|
|
9910 static void
|
|
375
|
9911 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
9912 slang_T *slang;
|
|
|
9913 char_u *inword;
|
|
|
9914 char_u *res;
|
|
|
9915 {
|
|
375
|
9916 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
9917 int word[MAXWLEN];
|
|
|
9918 int wres[MAXWLEN];
|
|
|
9919 int l;
|
|
|
9920 char_u *s;
|
|
|
9921 int *ws;
|
|
|
9922 char_u *t;
|
|
|
9923 int *pf;
|
|
|
9924 int i, j, z;
|
|
|
9925 int reslen;
|
|
|
9926 int n, k = 0;
|
|
|
9927 int z0;
|
|
|
9928 int k0;
|
|
|
9929 int n0;
|
|
|
9930 int c;
|
|
|
9931 int pri;
|
|
|
9932 int p0 = -333;
|
|
|
9933 int c0;
|
|
|
9934 int did_white = FALSE;
|
|
|
9935
|
|
|
9936 /*
|
|
|
9937 * Convert the multi-byte string to a wide-character string.
|
|
|
9938 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
9939 * But keep white space.
|
|
|
9940 */
|
|
|
9941 n = 0;
|
|
|
9942 for (s = inword; *s != NUL; )
|
|
|
9943 {
|
|
|
9944 t = s;
|
|
474
|
9945 c = mb_cptr2char_adv(&s);
|
|
372
|
9946 if (slang->sl_rem_accents)
|
|
|
9947 {
|
|
|
9948 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9949 {
|
|
|
9950 if (did_white)
|
|
|
9951 continue;
|
|
|
9952 c = ' ';
|
|
|
9953 did_white = TRUE;
|
|
|
9954 }
|
|
|
9955 else
|
|
|
9956 {
|
|
|
9957 did_white = FALSE;
|
|
376
|
9958 if (!spell_iswordp_nmw(t))
|
|
372
|
9959 continue;
|
|
|
9960 }
|
|
|
9961 }
|
|
|
9962 word[n++] = c;
|
|
|
9963 }
|
|
|
9964 word[n] = NUL;
|
|
|
9965
|
|
|
9966 /*
|
|
|
9967 * This comes from Aspell phonet.cpp.
|
|
|
9968 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
9969 * Changed to keep spaces.
|
|
|
9970 */
|
|
|
9971 i = reslen = z = 0;
|
|
|
9972 while ((c = word[i]) != NUL)
|
|
|
9973 {
|
|
|
9974 /* Start with the first rule that has the character in the word. */
|
|
|
9975 n = slang->sl_sal_first[c & 0xff];
|
|
|
9976 z0 = 0;
|
|
|
9977
|
|
|
9978 if (n >= 0)
|
|
|
9979 {
|
|
375
|
9980 /* check all rules for the same index byte */
|
|
372
|
9981 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
9982 {
|
|
|
9983 /* Quickly skip entries that don't match the word. Most
|
|
|
9984 * entries are less then three chars, optimize for that. */
|
|
375
|
9985 if (c != ws[0])
|
|
|
9986 continue;
|
|
372
|
9987 k = smp[n].sm_leadlen;
|
|
|
9988 if (k > 1)
|
|
|
9989 {
|
|
|
9990 if (word[i + 1] != ws[1])
|
|
|
9991 continue;
|
|
|
9992 if (k > 2)
|
|
|
9993 {
|
|
|
9994 for (j = 2; j < k; ++j)
|
|
|
9995 if (word[i + j] != ws[j])
|
|
|
9996 break;
|
|
|
9997 if (j < k)
|
|
|
9998 continue;
|
|
|
9999 }
|
|
|
10000 }
|
|
|
10001
|
|
375
|
10002 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
10003 {
|
|
375
|
10004 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
10005 while (*pf != NUL && *pf != word[i + k])
|
|
|
10006 ++pf;
|
|
|
10007 if (*pf == NUL)
|
|
|
10008 continue;
|
|
|
10009 ++k;
|
|
|
10010 }
|
|
|
10011 s = smp[n].sm_rules;
|
|
|
10012 pri = 5; /* default priority */
|
|
|
10013
|
|
|
10014 p0 = *s;
|
|
|
10015 k0 = k;
|
|
|
10016 while (*s == '-' && k > 1)
|
|
|
10017 {
|
|
|
10018 k--;
|
|
|
10019 s++;
|
|
|
10020 }
|
|
|
10021 if (*s == '<')
|
|
|
10022 s++;
|
|
|
10023 if (VIM_ISDIGIT(*s))
|
|
|
10024 {
|
|
|
10025 /* determine priority */
|
|
|
10026 pri = *s - '0';
|
|
|
10027 s++;
|
|
|
10028 }
|
|
|
10029 if (*s == '^' && *(s + 1) == '^')
|
|
|
10030 s++;
|
|
|
10031
|
|
|
10032 if (*s == NUL
|
|
|
10033 || (*s == '^'
|
|
|
10034 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
10035 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
10036 && (*(s + 1) != '$'
|
|
376
|
10037 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
10038 || (*s == '$' && i > 0
|
|
376
|
10039 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
10040 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
10041 {
|
|
|
10042 /* search for followup rules, if: */
|
|
|
10043 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
10044 c0 = word[i + k - 1];
|
|
|
10045 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
10046
|
|
|
10047 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
10048 && p0 != '-' && word[i + k] != NUL)
|
|
|
10049 {
|
|
375
|
10050 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
10051 * all entries with the same index byte. */
|
|
372
|
10052 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
10053 == (c0 & 0xff); ++n0)
|
|
|
10054 {
|
|
|
10055 /* Quickly skip entries that don't match the word.
|
|
375
|
10056 */
|
|
|
10057 if (c0 != ws[0])
|
|
|
10058 continue;
|
|
372
|
10059 k0 = smp[n0].sm_leadlen;
|
|
|
10060 if (k0 > 1)
|
|
|
10061 {
|
|
|
10062 if (word[i + k] != ws[1])
|
|
|
10063 continue;
|
|
|
10064 if (k0 > 2)
|
|
|
10065 {
|
|
|
10066 pf = word + i + k + 1;
|
|
|
10067 for (j = 2; j < k0; ++j)
|
|
|
10068 if (*pf++ != ws[j])
|
|
|
10069 break;
|
|
|
10070 if (j < k0)
|
|
|
10071 continue;
|
|
|
10072 }
|
|
|
10073 }
|
|
|
10074 k0 += k - 1;
|
|
|
10075
|
|
375
|
10076 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
10077 {
|
|
|
10078 /* Check for match with one of the chars in
|
|
375
|
10079 * "sm_oneof". */
|
|
372
|
10080 while (*pf != NUL && *pf != word[i + k0])
|
|
|
10081 ++pf;
|
|
|
10082 if (*pf == NUL)
|
|
|
10083 continue;
|
|
|
10084 ++k0;
|
|
|
10085 }
|
|
|
10086
|
|
|
10087 p0 = 5;
|
|
|
10088 s = smp[n0].sm_rules;
|
|
|
10089 while (*s == '-')
|
|
|
10090 {
|
|
|
10091 /* "k0" gets NOT reduced because
|
|
|
10092 * "if (k0 == k)" */
|
|
|
10093 s++;
|
|
|
10094 }
|
|
|
10095 if (*s == '<')
|
|
|
10096 s++;
|
|
|
10097 if (VIM_ISDIGIT(*s))
|
|
|
10098 {
|
|
|
10099 p0 = *s - '0';
|
|
|
10100 s++;
|
|
|
10101 }
|
|
|
10102
|
|
|
10103 if (*s == NUL
|
|
|
10104 /* *s == '^' cuts */
|
|
|
10105 || (*s == '$'
|
|
376
|
10106 && !spell_iswordp_w(word + i + k0,
|
|
|
10107 curbuf)))
|
|
372
|
10108 {
|
|
|
10109 if (k0 == k)
|
|
|
10110 /* this is just a piece of the string */
|
|
|
10111 continue;
|
|
|
10112
|
|
|
10113 if (p0 < pri)
|
|
|
10114 /* priority too low */
|
|
|
10115 continue;
|
|
|
10116 /* rule fits; stop search */
|
|
|
10117 break;
|
|
|
10118 }
|
|
|
10119 }
|
|
|
10120
|
|
|
10121 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
10122 == (c0 & 0xff))
|
|
|
10123 continue;
|
|
|
10124 }
|
|
|
10125
|
|
|
10126 /* replace string */
|
|
|
10127 ws = smp[n].sm_to_w;
|
|
|
10128 s = smp[n].sm_rules;
|
|
|
10129 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
10130 if (p0 == 1 && z == 0)
|
|
|
10131 {
|
|
|
10132 /* rule with '<' is used */
|
|
389
|
10133 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
10134 && (wres[reslen - 1] == c
|
|
372
|
10135 || wres[reslen - 1] == *ws))
|
|
|
10136 reslen--;
|
|
|
10137 z0 = 1;
|
|
|
10138 z = 1;
|
|
|
10139 k0 = 0;
|
|
389
|
10140 if (ws != NULL)
|
|
|
10141 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
10142 {
|
|
|
10143 word[i + k0] = *ws;
|
|
|
10144 k0++;
|
|
|
10145 ws++;
|
|
|
10146 }
|
|
372
|
10147 if (k > k0)
|
|
|
10148 mch_memmove(word + i + k0, word + i + k,
|
|
|
10149 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
10150
|
|
|
10151 /* new "actual letter" */
|
|
|
10152 c = word[i];
|
|
|
10153 }
|
|
|
10154 else
|
|
|
10155 {
|
|
|
10156 /* no '<' rule used */
|
|
|
10157 i += k - 1;
|
|
|
10158 z = 0;
|
|
389
|
10159 if (ws != NULL)
|
|
|
10160 while (*ws != NUL && ws[1] != NUL
|
|
|
10161 && reslen < MAXWLEN)
|
|
|
10162 {
|
|
|
10163 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
10164 wres[reslen++] = *ws;
|
|
|
10165 ws++;
|
|
|
10166 }
|
|
372
|
10167 /* new "actual letter" */
|
|
389
|
10168 if (ws == NULL)
|
|
|
10169 c = NUL;
|
|
|
10170 else
|
|
|
10171 c = *ws;
|
|
372
|
10172 if (strstr((char *)s, "^^") != NULL)
|
|
|
10173 {
|
|
|
10174 if (c != NUL)
|
|
|
10175 wres[reslen++] = c;
|
|
|
10176 mch_memmove(word, word + i + 1,
|
|
|
10177 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
10178 i = 0;
|
|
|
10179 z0 = 1;
|
|
|
10180 }
|
|
|
10181 }
|
|
|
10182 break;
|
|
|
10183 }
|
|
|
10184 }
|
|
|
10185 }
|
|
|
10186 else if (vim_iswhite(c))
|
|
|
10187 {
|
|
|
10188 c = ' ';
|
|
|
10189 k = 1;
|
|
|
10190 }
|
|
|
10191
|
|
|
10192 if (z0 == 0)
|
|
|
10193 {
|
|
|
10194 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
10195 && (!slang->sl_collapse || reslen == 0
|
|
|
10196 || wres[reslen - 1] != c))
|
|
|
10197 /* condense only double letters */
|
|
|
10198 wres[reslen++] = c;
|
|
|
10199
|
|
|
10200 i++;
|
|
|
10201 z = 0;
|
|
|
10202 k = 0;
|
|
|
10203 }
|
|
|
10204 }
|
|
|
10205
|
|
|
10206 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
10207 l = 0;
|
|
|
10208 for (n = 0; n < reslen; ++n)
|
|
|
10209 {
|
|
|
10210 l += mb_char2bytes(wres[n], res + l);
|
|
|
10211 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
10212 break;
|
|
|
10213 }
|
|
|
10214 res[l] = NUL;
|
|
|
10215 }
|
|
|
10216 #endif
|
|
|
10217
|
|
324
|
10218 /*
|
|
344
|
10219 * Compute a score for two sound-a-like words.
|
|
|
10220 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
10221 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
10222 * avoiding checks that will not be possible.
|
|
|
10223 */
|
|
|
10224 static int
|
|
351
|
10225 soundalike_score(goodstart, badstart)
|
|
|
10226 char_u *goodstart; /* sound-folded good word */
|
|
|
10227 char_u *badstart; /* sound-folded bad word */
|
|
344
|
10228 {
|
|
351
|
10229 char_u *goodsound = goodstart;
|
|
|
10230 char_u *badsound = badstart;
|
|
|
10231 int goodlen;
|
|
|
10232 int badlen;
|
|
344
|
10233 int n;
|
|
|
10234 char_u *pl, *ps;
|
|
|
10235 char_u *pl2, *ps2;
|
|
351
|
10236 int score = 0;
|
|
|
10237
|
|
|
10238 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
10239 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
10240 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
10241 {
|
|
|
10242 score = SCORE_DEL / 2;
|
|
|
10243 if (*badsound == '*')
|
|
|
10244 ++badsound;
|
|
|
10245 else
|
|
|
10246 ++goodsound;
|
|
|
10247 }
|
|
|
10248
|
|
|
10249 goodlen = STRLEN(goodsound);
|
|
|
10250 badlen = STRLEN(badsound);
|
|
344
|
10251
|
|
|
10252 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
10253 * changes. */
|
|
|
10254 n = goodlen - badlen;
|
|
|
10255 if (n < -2 || n > 2)
|
|
|
10256 return SCORE_MAXMAX;
|
|
|
10257
|
|
|
10258 if (n > 0)
|
|
|
10259 {
|
|
351
|
10260 pl = goodsound; /* goodsound is longest */
|
|
344
|
10261 ps = badsound;
|
|
|
10262 }
|
|
|
10263 else
|
|
|
10264 {
|
|
351
|
10265 pl = badsound; /* badsound is longest */
|
|
344
|
10266 ps = goodsound;
|
|
|
10267 }
|
|
|
10268
|
|
|
10269 /* Skip over the identical part. */
|
|
|
10270 while (*pl == *ps && *pl != NUL)
|
|
|
10271 {
|
|
|
10272 ++pl;
|
|
|
10273 ++ps;
|
|
|
10274 }
|
|
|
10275
|
|
|
10276 switch (n)
|
|
|
10277 {
|
|
|
10278 case -2:
|
|
|
10279 case 2:
|
|
|
10280 /*
|
|
|
10281 * Must delete two characters from "pl".
|
|
|
10282 */
|
|
|
10283 ++pl; /* first delete */
|
|
|
10284 while (*pl == *ps)
|
|
|
10285 {
|
|
|
10286 ++pl;
|
|
|
10287 ++ps;
|
|
|
10288 }
|
|
|
10289 /* strings must be equal after second delete */
|
|
|
10290 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
10291 return score + SCORE_DEL * 2;
|
|
344
|
10292
|
|
|
10293 /* Failed to compare. */
|
|
|
10294 break;
|
|
|
10295
|
|
|
10296 case -1:
|
|
|
10297 case 1:
|
|
|
10298 /*
|
|
|
10299 * Minimal one delete from "pl" required.
|
|
|
10300 */
|
|
|
10301
|
|
|
10302 /* 1: delete */
|
|
|
10303 pl2 = pl + 1;
|
|
|
10304 ps2 = ps;
|
|
|
10305 while (*pl2 == *ps2)
|
|
|
10306 {
|
|
|
10307 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10308 return score + SCORE_DEL;
|
|
344
|
10309 ++pl2;
|
|
|
10310 ++ps2;
|
|
|
10311 }
|
|
|
10312
|
|
|
10313 /* 2: delete then swap, then rest must be equal */
|
|
|
10314 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10315 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10316 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
10317
|
|
|
10318 /* 3: delete then substitute, then the rest must be equal */
|
|
|
10319 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10320 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
10321
|
|
|
10322 /* 4: first swap then delete */
|
|
|
10323 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
10324 {
|
|
|
10325 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
10326 ps2 = ps + 2;
|
|
|
10327 while (*pl2 == *ps2)
|
|
|
10328 {
|
|
|
10329 ++pl2;
|
|
|
10330 ++ps2;
|
|
|
10331 }
|
|
|
10332 /* delete a char and then strings must be equal */
|
|
|
10333 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10334 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
10335 }
|
|
|
10336
|
|
|
10337 /* 5: first substitute then delete */
|
|
|
10338 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
10339 ps2 = ps + 1;
|
|
|
10340 while (*pl2 == *ps2)
|
|
|
10341 {
|
|
|
10342 ++pl2;
|
|
|
10343 ++ps2;
|
|
|
10344 }
|
|
|
10345 /* delete a char and then strings must be equal */
|
|
|
10346 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10347 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
10348
|
|
|
10349 /* Failed to compare. */
|
|
|
10350 break;
|
|
|
10351
|
|
|
10352 case 0:
|
|
|
10353 /*
|
|
|
10354 * Lenghts are equal, thus changes must result in same length: An
|
|
|
10355 * insert is only possible in combination with a delete.
|
|
|
10356 * 1: check if for identical strings
|
|
|
10357 */
|
|
|
10358 if (*pl == NUL)
|
|
351
|
10359 return score;
|
|
344
|
10360
|
|
|
10361 /* 2: swap */
|
|
|
10362 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
10363 {
|
|
|
10364 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
10365 ps2 = ps + 2;
|
|
|
10366 while (*pl2 == *ps2)
|
|
|
10367 {
|
|
|
10368 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10369 return score + SCORE_SWAP;
|
|
344
|
10370 ++pl2;
|
|
|
10371 ++ps2;
|
|
|
10372 }
|
|
|
10373 /* 3: swap and swap again */
|
|
|
10374 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10375 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10376 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
10377
|
|
|
10378 /* 4: swap and substitute */
|
|
|
10379 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10380 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
10381 }
|
|
|
10382
|
|
|
10383 /* 5: substitute */
|
|
|
10384 pl2 = pl + 1;
|
|
|
10385 ps2 = ps + 1;
|
|
|
10386 while (*pl2 == *ps2)
|
|
|
10387 {
|
|
|
10388 if (*pl2 == NUL) /* reached the end */
|
|
351
|
10389 return score + SCORE_SUBST;
|
|
344
|
10390 ++pl2;
|
|
|
10391 ++ps2;
|
|
|
10392 }
|
|
|
10393
|
|
|
10394 /* 6: substitute and swap */
|
|
|
10395 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
10396 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
10397 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
10398
|
|
|
10399 /* 7: substitute and substitute */
|
|
|
10400 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
10401 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
10402
|
|
|
10403 /* 8: insert then delete */
|
|
|
10404 pl2 = pl;
|
|
|
10405 ps2 = ps + 1;
|
|
|
10406 while (*pl2 == *ps2)
|
|
|
10407 {
|
|
|
10408 ++pl2;
|
|
|
10409 ++ps2;
|
|
|
10410 }
|
|
|
10411 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
10412 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
10413
|
|
|
10414 /* 9: delete then insert */
|
|
|
10415 pl2 = pl + 1;
|
|
|
10416 ps2 = ps;
|
|
|
10417 while (*pl2 == *ps2)
|
|
|
10418 {
|
|
|
10419 ++pl2;
|
|
|
10420 ++ps2;
|
|
|
10421 }
|
|
|
10422 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
10423 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
10424
|
|
|
10425 /* Failed to compare. */
|
|
|
10426 break;
|
|
|
10427 }
|
|
|
10428
|
|
|
10429 return SCORE_MAXMAX;
|
|
|
10430 }
|
|
324
|
10431
|
|
323
|
10432 /*
|
|
|
10433 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
10434 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
10435 *
|
|
323
|
10436 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
10437 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
10438 * characters.
|
|
323
|
10439 */
|
|
|
10440 static int
|
|
|
10441 spell_edit_score(badword, goodword)
|
|
|
10442 char_u *badword;
|
|
|
10443 char_u *goodword;
|
|
|
10444 {
|
|
|
10445 int *cnt;
|
|
372
|
10446 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
10447 int j, i;
|
|
|
10448 int t;
|
|
|
10449 int bc, gc;
|
|
324
|
10450 int pbc, pgc;
|
|
|
10451 #ifdef FEAT_MBYTE
|
|
|
10452 char_u *p;
|
|
|
10453 int wbadword[MAXWLEN];
|
|
|
10454 int wgoodword[MAXWLEN];
|
|
|
10455
|
|
|
10456 if (has_mbyte)
|
|
|
10457 {
|
|
|
10458 /* Get the characters from the multi-byte strings and put them in an
|
|
|
10459 * int array for easy access. */
|
|
|
10460 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
10461 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
10462 wbadword[badlen++] = 0;
|
|
324
|
10463 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
10464 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
10465 wgoodword[goodlen++] = 0;
|
|
324
|
10466 }
|
|
|
10467 else
|
|
|
10468 #endif
|
|
|
10469 {
|
|
|
10470 badlen = STRLEN(badword) + 1;
|
|
|
10471 goodlen = STRLEN(goodword) + 1;
|
|
|
10472 }
|
|
323
|
10473
|
|
|
10474 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
10475 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
10476 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
10477 TRUE);
|
|
324
|
10478 if (cnt == NULL)
|
|
|
10479 return 0; /* out of memory */
|
|
323
|
10480
|
|
|
10481 CNT(0, 0) = 0;
|
|
|
10482 for (j = 1; j <= goodlen; ++j)
|
|
|
10483 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
10484
|
|
|
10485 for (i = 1; i <= badlen; ++i)
|
|
|
10486 {
|
|
|
10487 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
10488 for (j = 1; j <= goodlen; ++j)
|
|
|
10489 {
|
|
324
|
10490 #ifdef FEAT_MBYTE
|
|
|
10491 if (has_mbyte)
|
|
|
10492 {
|
|
|
10493 bc = wbadword[i - 1];
|
|
|
10494 gc = wgoodword[j - 1];
|
|
|
10495 }
|
|
|
10496 else
|
|
|
10497 #endif
|
|
|
10498 {
|
|
|
10499 bc = badword[i - 1];
|
|
|
10500 gc = goodword[j - 1];
|
|
|
10501 }
|
|
323
|
10502 if (bc == gc)
|
|
|
10503 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
10504 else
|
|
|
10505 {
|
|
|
10506 /* Use a better score when there is only a case difference. */
|
|
324
|
10507 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
10508 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
10509 else
|
|
|
10510 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
10511
|
|
324
|
10512 if (i > 1 && j > 1)
|
|
323
|
10513 {
|
|
324
|
10514 #ifdef FEAT_MBYTE
|
|
|
10515 if (has_mbyte)
|
|
|
10516 {
|
|
|
10517 pbc = wbadword[i - 2];
|
|
|
10518 pgc = wgoodword[j - 2];
|
|
|
10519 }
|
|
|
10520 else
|
|
|
10521 #endif
|
|
|
10522 {
|
|
|
10523 pbc = badword[i - 2];
|
|
|
10524 pgc = goodword[j - 2];
|
|
|
10525 }
|
|
|
10526 if (bc == pgc && pbc == gc)
|
|
|
10527 {
|
|
|
10528 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
10529 if (t < CNT(i, j))
|
|
|
10530 CNT(i, j) = t;
|
|
|
10531 }
|
|
323
|
10532 }
|
|
|
10533 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
10534 if (t < CNT(i, j))
|
|
|
10535 CNT(i, j) = t;
|
|
|
10536 t = SCORE_INS + CNT(i, j - 1);
|
|
|
10537 if (t < CNT(i, j))
|
|
|
10538 CNT(i, j) = t;
|
|
|
10539 }
|
|
|
10540 }
|
|
|
10541 }
|
|
344
|
10542
|
|
|
10543 i = CNT(badlen - 1, goodlen - 1);
|
|
|
10544 vim_free(cnt);
|
|
|
10545 return i;
|
|
323
|
10546 }
|
|
307
|
10547
|
|
351
|
10548 /*
|
|
|
10549 * ":spelldump"
|
|
|
10550 */
|
|
|
10551 /*ARGSUSED*/
|
|
|
10552 void
|
|
|
10553 ex_spelldump(eap)
|
|
|
10554 exarg_T *eap;
|
|
|
10555 {
|
|
|
10556 buf_T *buf = curbuf;
|
|
|
10557 langp_T *lp;
|
|
|
10558 slang_T *slang;
|
|
|
10559 idx_T arridx[MAXWLEN];
|
|
|
10560 int curi[MAXWLEN];
|
|
|
10561 char_u word[MAXWLEN];
|
|
|
10562 int c;
|
|
|
10563 char_u *byts;
|
|
|
10564 idx_T *idxs;
|
|
|
10565 linenr_T lnum = 0;
|
|
|
10566 int round;
|
|
|
10567 int depth;
|
|
|
10568 int n;
|
|
|
10569 int flags;
|
|
381
|
10570 char_u *region_names = NULL; /* region names being used */
|
|
|
10571 int do_region = TRUE; /* dump region names and numbers */
|
|
|
10572 char_u *p;
|
|
351
|
10573
|
|
|
10574 if (no_spell_checking())
|
|
|
10575 return;
|
|
|
10576
|
|
|
10577 /* Create a new empty buffer by splitting the window. */
|
|
|
10578 do_cmdline_cmd((char_u *)"new");
|
|
|
10579 if (!bufempty() || !buf_valid(buf))
|
|
|
10580 return;
|
|
|
10581
|
|
381
|
10582 /* Find out if we can support regions: All languages must support the same
|
|
|
10583 * regions or none at all. */
|
|
|
10584 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
10585 {
|
|
|
10586 p = lp->lp_slang->sl_regions;
|
|
|
10587 if (p[0] != 0)
|
|
|
10588 {
|
|
|
10589 if (region_names == NULL) /* first language with regions */
|
|
|
10590 region_names = p;
|
|
|
10591 else if (STRCMP(region_names, p) != 0)
|
|
|
10592 {
|
|
|
10593 do_region = FALSE; /* region names are different */
|
|
|
10594 break;
|
|
|
10595 }
|
|
|
10596 }
|
|
|
10597 }
|
|
|
10598
|
|
|
10599 if (do_region && region_names != NULL)
|
|
|
10600 {
|
|
|
10601 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
10602 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
10603 }
|
|
|
10604 else
|
|
|
10605 do_region = FALSE;
|
|
|
10606
|
|
|
10607 /*
|
|
|
10608 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
10609 */
|
|
351
|
10610 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
10611 {
|
|
|
10612 slang = lp->lp_slang;
|
|
|
10613
|
|
|
10614 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
10615 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
10616
|
|
|
10617 /* round 1: case-folded tree
|
|
|
10618 * round 2: keep-case tree */
|
|
|
10619 for (round = 1; round <= 2; ++round)
|
|
|
10620 {
|
|
|
10621 if (round == 1)
|
|
|
10622 {
|
|
|
10623 byts = slang->sl_fbyts;
|
|
|
10624 idxs = slang->sl_fidxs;
|
|
|
10625 }
|
|
|
10626 else
|
|
|
10627 {
|
|
|
10628 byts = slang->sl_kbyts;
|
|
|
10629 idxs = slang->sl_kidxs;
|
|
|
10630 }
|
|
|
10631 if (byts == NULL)
|
|
|
10632 continue; /* array is empty */
|
|
|
10633
|
|
|
10634 depth = 0;
|
|
|
10635 arridx[0] = 0;
|
|
|
10636 curi[0] = 1;
|
|
|
10637 while (depth >= 0 && !got_int)
|
|
|
10638 {
|
|
|
10639 if (curi[depth] > byts[arridx[depth]])
|
|
|
10640 {
|
|
|
10641 /* Done all bytes at this node, go up one level. */
|
|
|
10642 --depth;
|
|
|
10643 line_breakcheck();
|
|
|
10644 }
|
|
|
10645 else
|
|
|
10646 {
|
|
|
10647 /* Do one more byte at this node. */
|
|
|
10648 n = arridx[depth] + curi[depth];
|
|
|
10649 ++curi[depth];
|
|
|
10650 c = byts[n];
|
|
|
10651 if (c == 0)
|
|
|
10652 {
|
|
|
10653 /* End of word, deal with the word.
|
|
|
10654 * Don't use keep-case words in the fold-case tree,
|
|
|
10655 * they will appear in the keep-case tree.
|
|
|
10656 * Only use the word when the region matches. */
|
|
|
10657 flags = (int)idxs[n];
|
|
|
10658 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
10659 && (do_region
|
|
|
10660 || (flags & WF_REGION) == 0
|
|
390
|
10661 || (((unsigned)flags >> 16)
|
|
351
|
10662 & lp->lp_region) != 0))
|
|
|
10663 {
|
|
|
10664 word[depth] = NUL;
|
|
381
|
10665 if (!do_region)
|
|
|
10666 flags &= ~WF_REGION;
|
|
355
|
10667
|
|
|
10668 /* Dump the basic word if there is no prefix or
|
|
|
10669 * when it's the first one. */
|
|
390
|
10670 c = (unsigned)flags >> 24;
|
|
355
|
10671 if (c == 0 || curi[depth] == 2)
|
|
|
10672 dump_word(word, round, flags, lnum++);
|
|
351
|
10673
|
|
|
10674 /* Apply the prefix, if there is one. */
|
|
355
|
10675 if (c != 0)
|
|
351
|
10676 lnum = apply_prefixes(slang, word, round,
|
|
|
10677 flags, lnum);
|
|
|
10678 }
|
|
|
10679 }
|
|
|
10680 else
|
|
|
10681 {
|
|
|
10682 /* Normal char, go one level deeper. */
|
|
|
10683 word[depth++] = c;
|
|
|
10684 arridx[depth] = idxs[n];
|
|
|
10685 curi[depth] = 1;
|
|
|
10686 }
|
|
|
10687 }
|
|
|
10688 }
|
|
|
10689 }
|
|
|
10690 }
|
|
|
10691
|
|
|
10692 /* Delete the empty line that we started with. */
|
|
|
10693 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
10694 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
10695
|
|
|
10696 redraw_later(NOT_VALID);
|
|
|
10697 }
|
|
|
10698
|
|
|
10699 /*
|
|
|
10700 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
10701 */
|
|
|
10702 static void
|
|
|
10703 dump_word(word, round, flags, lnum)
|
|
|
10704 char_u *word;
|
|
|
10705 int round;
|
|
|
10706 int flags;
|
|
|
10707 linenr_T lnum;
|
|
|
10708 {
|
|
|
10709 int keepcap = FALSE;
|
|
|
10710 char_u *p;
|
|
|
10711 char_u cword[MAXWLEN];
|
|
381
|
10712 char_u badword[MAXWLEN + 10];
|
|
|
10713 int i;
|
|
351
|
10714
|
|
|
10715 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
10716 {
|
|
|
10717 /* Need to fix case according to "flags". */
|
|
|
10718 make_case_word(word, cword, flags);
|
|
|
10719 p = cword;
|
|
|
10720 }
|
|
|
10721 else
|
|
|
10722 {
|
|
|
10723 p = word;
|
|
389
|
10724 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
10725 || (flags & WF_FIXCAP) != 0))
|
|
351
|
10726 keepcap = TRUE;
|
|
|
10727 }
|
|
|
10728
|
|
381
|
10729 /* Add flags and regions after a slash. */
|
|
|
10730 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
10731 {
|
|
|
10732 STRCPY(badword, p);
|
|
|
10733 STRCAT(badword, "/");
|
|
351
|
10734 if (keepcap)
|
|
|
10735 STRCAT(badword, "=");
|
|
|
10736 if (flags & WF_BANNED)
|
|
|
10737 STRCAT(badword, "!");
|
|
|
10738 else if (flags & WF_RARE)
|
|
|
10739 STRCAT(badword, "?");
|
|
381
|
10740 if (flags & WF_REGION)
|
|
|
10741 for (i = 0; i < 7; ++i)
|
|
390
|
10742 if (flags & (0x10000 << i))
|
|
381
|
10743 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
10744 p = badword;
|
|
|
10745 }
|
|
|
10746
|
|
|
10747 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
10748 }
|
|
|
10749
|
|
|
10750 /*
|
|
372
|
10751 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
10752 * "word" and append a line to the buffer.
|
|
351
|
10753 * Return the updated line number.
|
|
|
10754 */
|
|
|
10755 static linenr_T
|
|
|
10756 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
10757 slang_T *slang;
|
|
|
10758 char_u *word; /* case-folded word */
|
|
|
10759 int round;
|
|
|
10760 int flags; /* flags with prefix ID */
|
|
|
10761 linenr_T startlnum;
|
|
|
10762 {
|
|
|
10763 idx_T arridx[MAXWLEN];
|
|
|
10764 int curi[MAXWLEN];
|
|
|
10765 char_u prefix[MAXWLEN];
|
|
455
|
10766 char_u word_up[MAXWLEN];
|
|
|
10767 int has_word_up = FALSE;
|
|
351
|
10768 int c;
|
|
|
10769 char_u *byts;
|
|
|
10770 idx_T *idxs;
|
|
|
10771 linenr_T lnum = startlnum;
|
|
|
10772 int depth;
|
|
|
10773 int n;
|
|
|
10774 int len;
|
|
|
10775 int i;
|
|
|
10776
|
|
455
|
10777 /* if the word starts with a lower-case letter make the word with an
|
|
|
10778 * upper-case letter in word_up[]. */
|
|
|
10779 c = PTR2CHAR(word);
|
|
|
10780 if (SPELL_TOUPPER(c) != c)
|
|
|
10781 {
|
|
|
10782 onecap_copy(word, word_up, TRUE);
|
|
|
10783 has_word_up = TRUE;
|
|
|
10784 }
|
|
|
10785
|
|
351
|
10786 byts = slang->sl_pbyts;
|
|
|
10787 idxs = slang->sl_pidxs;
|
|
|
10788 if (byts != NULL) /* array not is empty */
|
|
|
10789 {
|
|
|
10790 /*
|
|
|
10791 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
10792 * When at the end of a prefix check that it supports "flags".
|
|
351
|
10793 */
|
|
|
10794 depth = 0;
|
|
|
10795 arridx[0] = 0;
|
|
|
10796 curi[0] = 1;
|
|
|
10797 while (depth >= 0 && !got_int)
|
|
|
10798 {
|
|
390
|
10799 n = arridx[depth];
|
|
|
10800 len = byts[n];
|
|
|
10801 if (curi[depth] > len)
|
|
351
|
10802 {
|
|
|
10803 /* Done all bytes at this node, go up one level. */
|
|
|
10804 --depth;
|
|
|
10805 line_breakcheck();
|
|
|
10806 }
|
|
|
10807 else
|
|
|
10808 {
|
|
|
10809 /* Do one more byte at this node. */
|
|
390
|
10810 n += curi[depth];
|
|
351
|
10811 ++curi[depth];
|
|
|
10812 c = byts[n];
|
|
|
10813 if (c == 0)
|
|
|
10814 {
|
|
|
10815 /* End of prefix, find out how many IDs there are. */
|
|
|
10816 for (i = 1; i < len; ++i)
|
|
|
10817 if (byts[n + i] != 0)
|
|
|
10818 break;
|
|
|
10819 curi[depth] += i - 1;
|
|
|
10820
|
|
455
|
10821 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
10822 if (c != 0)
|
|
351
|
10823 {
|
|
376
|
10824 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
10825 dump_word(prefix, round,
|
|
455
|
10826 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
10827 : flags, lnum++);
|
|
351
|
10828 }
|
|
455
|
10829
|
|
|
10830 /* Check for prefix that matches the word when the
|
|
|
10831 * first letter is upper-case, but only if the prefix has
|
|
|
10832 * a condition. */
|
|
|
10833 if (has_word_up)
|
|
|
10834 {
|
|
|
10835 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
10836 TRUE);
|
|
|
10837 if (c != 0)
|
|
|
10838 {
|
|
|
10839 vim_strncpy(prefix + depth, word_up,
|
|
|
10840 MAXWLEN - depth - 1);
|
|
|
10841 dump_word(prefix, round,
|
|
|
10842 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
10843 : flags, lnum++);
|
|
|
10844 }
|
|
|
10845 }
|
|
351
|
10846 }
|
|
|
10847 else
|
|
|
10848 {
|
|
|
10849 /* Normal char, go one level deeper. */
|
|
|
10850 prefix[depth++] = c;
|
|
|
10851 arridx[depth] = idxs[n];
|
|
|
10852 curi[depth] = 1;
|
|
|
10853 }
|
|
|
10854 }
|
|
|
10855 }
|
|
|
10856 }
|
|
|
10857
|
|
|
10858 return lnum;
|
|
|
10859 }
|
|
|
10860
|
|
475
|
10861 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
10862 static int spell_expand_need_cap;
|
|
|
10863
|
|
|
10864 /*
|
|
|
10865 * Find start of the word in front of the cursor. We don't check if it is
|
|
|
10866 * badly spelled, with completion we can only change the word in front of the
|
|
|
10867 * cursor.
|
|
|
10868 * Used for Insert mode completion CTRL-X ?.
|
|
|
10869 * Returns the column number of the word.
|
|
|
10870 */
|
|
|
10871 int
|
|
|
10872 spell_word_start(startcol)
|
|
|
10873 int startcol;
|
|
|
10874 {
|
|
|
10875 char_u *line;
|
|
|
10876 char_u *p;
|
|
|
10877 int col = 0;
|
|
|
10878
|
|
|
10879 if (no_spell_checking())
|
|
|
10880 return startcol;
|
|
|
10881
|
|
|
10882 /* Find a word character before "startcol". */
|
|
|
10883 line = ml_get_curline();
|
|
|
10884 for (p = line + startcol; p > line; )
|
|
|
10885 {
|
|
|
10886 mb_ptr_back(line, p);
|
|
|
10887 if (spell_iswordp_nmw(p))
|
|
|
10888 break;
|
|
|
10889 }
|
|
|
10890
|
|
|
10891 /* Go back to start of the word. */
|
|
|
10892 while (p > line)
|
|
|
10893 {
|
|
|
10894 col = p - line;
|
|
|
10895 mb_ptr_back(line, p);
|
|
|
10896 if (!spell_iswordp(p, curbuf))
|
|
|
10897 break;
|
|
|
10898 col = 0;
|
|
|
10899 }
|
|
|
10900
|
|
|
10901 /* Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
10902 * expand_spelling() is called. Therefore the ugly global variable. */
|
|
|
10903 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
10904
|
|
|
10905 return col;
|
|
|
10906 }
|
|
|
10907
|
|
|
10908 /*
|
|
|
10909 * Get list of spelling suggestions.
|
|
|
10910 * Used for Insert mode completion CTRL-X ?.
|
|
|
10911 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
10912 * allocated strings.
|
|
|
10913 */
|
|
|
10914 /*ARGSUSED*/
|
|
|
10915 int
|
|
|
10916 expand_spelling(lnum, col, pat, matchp)
|
|
|
10917 linenr_T lnum;
|
|
|
10918 int col;
|
|
|
10919 char_u *pat;
|
|
|
10920 char_u ***matchp;
|
|
|
10921 {
|
|
|
10922 garray_T ga;
|
|
|
10923
|
|
|
10924 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap);
|
|
|
10925 *matchp = ga.ga_data;
|
|
|
10926 return ga.ga_len;
|
|
|
10927 }
|
|
|
10928 #endif
|
|
|
10929
|
|
236
|
10930 #endif /* FEAT_SYN_HL */
|
