|
223
|
1 /* vi:set ts=8 sts=4 sw=4:
|
|
|
2 *
|
|
|
3 * VIM - Vi IMproved by Bram Moolenaar
|
|
|
4 *
|
|
|
5 * Do ":help uganda" in Vim to read copying and usage conditions.
|
|
|
6 * Do ":help credits" in Vim to see a list of people who contributed.
|
|
|
7 * See README.txt for an overview of the Vim source code.
|
|
|
8 */
|
|
|
9
|
|
|
10 /*
|
|
|
11 * spell.c: code for spell checking
|
|
226
|
12 *
|
|
300
|
13 * The spell checking mechanism uses a tree (aka trie). Each node in the tree
|
|
|
14 * has a list of bytes that can appear (siblings). For each byte there is a
|
|
|
15 * pointer to the node with the byte that follows in the word (child).
|
|
324
|
16 *
|
|
|
17 * A NUL byte is used where the word may end. The bytes are sorted, so that
|
|
|
18 * binary searching can be used and the NUL bytes are at the start. The
|
|
|
19 * number of possible bytes is stored before the list of bytes.
|
|
|
20 *
|
|
|
21 * The tree uses two arrays: "byts" stores the characters, "idxs" stores
|
|
|
22 * either the next index or flags. The tree starts at index 0. For example,
|
|
|
23 * to lookup "vi" this sequence is followed:
|
|
|
24 * i = 0
|
|
|
25 * len = byts[i]
|
|
|
26 * n = where "v" appears in byts[i + 1] to byts[i + len]
|
|
|
27 * i = idxs[n]
|
|
|
28 * len = byts[i]
|
|
|
29 * n = where "i" appears in byts[i + 1] to byts[i + len]
|
|
|
30 * i = idxs[n]
|
|
|
31 * len = byts[i]
|
|
|
32 * find that byts[i + 1] is 0, idxs[i + 1] has flags for "vi".
|
|
300
|
33 *
|
|
339
|
34 * There are two word trees: one with case-folded words and one with words in
|
|
300
|
35 * original case. The second one is only used for keep-case words and is
|
|
|
36 * usually small.
|
|
|
37 *
|
|
339
|
38 * There is one additional tree for when prefixes are not applied when
|
|
|
39 * generating the .spl file. This tree stores all the possible prefixes, as
|
|
|
40 * if they were words. At each word (prefix) end the prefix nr is stored, the
|
|
|
41 * following word must support this prefix nr. And the condition nr is
|
|
|
42 * stored, used to lookup the condition that the word must match with.
|
|
|
43 *
|
|
300
|
44 * Thanks to Olaf Seibert for providing an example implementation of this tree
|
|
|
45 * and the compression mechanism.
|
|
243
|
46 *
|
|
|
47 * Matching involves checking the caps type: Onecap ALLCAP KeepCap.
|
|
|
48 *
|
|
236
|
49 * Why doesn't Vim use aspell/ispell/myspell/etc.?
|
|
|
50 * See ":help develop-spell".
|
|
|
51 */
|
|
|
52
|
|
469
|
53 /* Use SPELL_PRINTTREE for debugging: dump the word tree after adding a word.
|
|
|
54 * Only use for small word lists!
|
|
|
55 * SPELL_COMPRESS_CNT is in how many words we compress the tree. */
|
|
|
56 #if 0
|
|
|
57 # define SPELL_PRINTTREE
|
|
|
58 # define SPELL_COMPRESS_CNT 1
|
|
|
59 #else
|
|
|
60 # define SPELL_COMPRESS_CNT 1000000
|
|
|
61 #endif
|
|
|
62
|
|
300
|
63 /*
|
|
324
|
64 * Use this to adjust the score after finding suggestions, based on the
|
|
|
65 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
66 * it for every possible suggestion.
|
|
|
67 * Disadvantage: When "the" is typed as "hte" it sounds different and goes
|
|
|
68 * down in the list.
|
|
344
|
69 * Used when 'spellsuggest' is set to "best".
|
|
|
70 */
|
|
|
71 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
72
|
|
|
73 /*
|
|
|
74 * The double scoring mechanism is based on the principle that there are two
|
|
|
75 * kinds of spelling mistakes:
|
|
|
76 * 1. You know how to spell the word, but mistype something. This results in
|
|
|
77 * a small editing distance (character swapped/omitted/inserted) and
|
|
|
78 * possibly a word that sounds completely different.
|
|
|
79 * 2. You don't know how to spell the word and type something that sounds
|
|
|
80 * right. The edit distance can be big but the word is similar after
|
|
|
81 * sound-folding.
|
|
|
82 * Since scores for these two mistakes will be very different we use a list
|
|
|
83 * for each.
|
|
|
84 * The sound-folding is slow, only do double scoring when 'spellsuggest' is
|
|
|
85 * "double".
|
|
323
|
86 */
|
|
|
87
|
|
|
88 /*
|
|
339
|
89 * Vim spell file format: <HEADER>
|
|
|
90 * <SUGGEST>
|
|
|
91 * <LWORDTREE>
|
|
|
92 * <KWORDTREE>
|
|
|
93 * <PREFIXTREE>
|
|
300
|
94 *
|
|
339
|
95 * <HEADER>: <fileID>
|
|
|
96 * <regioncnt> <regionname> ...
|
|
|
97 * <charflagslen> <charflags>
|
|
|
98 * <fcharslen> <fchars>
|
|
366
|
99 * <midwordlen> <midword>
|
|
339
|
100 * <prefcondcnt> <prefcond> ...
|
|
300
|
101 *
|
|
455
|
102 * <fileID> 10 bytes "VIMspell09"
|
|
300
|
103 * <regioncnt> 1 byte number of regions following (8 supported)
|
|
307
|
104 * <regionname> 2 bytes Region name: ca, au, etc. Lower case.
|
|
300
|
105 * First <regionname> is region 1.
|
|
|
106 *
|
|
|
107 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
108 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
109 * 0x01 word character CF_WORD
|
|
|
110 * 0x02 upper-case character CF_UPPER
|
|
300
|
111 * <fcharslen> 2 bytes Number of bytes in <fchars>.
|
|
|
112 * <fchars> N bytes Folded characters, first one is for character 128.
|
|
|
113 *
|
|
366
|
114 * <midwordlen> 2 bytes Number of bytes in <midword>.
|
|
|
115 * <midword> N bytes Characters that are word characters only when used
|
|
|
116 * in the middle of a word.
|
|
|
117 *
|
|
339
|
118 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
119 *
|
|
|
120 * <prefcond> : <condlen> <condstr>
|
|
|
121 *
|
|
|
122 * <condlen> 1 byte Length of <condstr>.
|
|
|
123 *
|
|
|
124 * <condstr> N bytes Condition for the prefix.
|
|
|
125 *
|
|
300
|
126 *
|
|
323
|
127 * <SUGGEST> : <repcount> <rep> ...
|
|
|
128 * <salflags> <salcount> <sal> ...
|
|
|
129 * <maplen> <mapstr>
|
|
|
130 *
|
|
|
131 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
|
132 *
|
|
|
133 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
|
134 *
|
|
|
135 * <repfromlen> 1 byte length of <repfrom>
|
|
|
136 *
|
|
|
137 * <repfrom> N bytes "from" part of replacement
|
|
|
138 *
|
|
|
139 * <reptolen> 1 byte length of <repto>
|
|
|
140 *
|
|
|
141 * <repto> N bytes "to" part of replacement
|
|
300
|
142 *
|
|
323
|
143 * <salflags> 1 byte flags for soundsalike conversion:
|
|
|
144 * SAL_F0LLOWUP
|
|
|
145 * SAL_COLLAPSE
|
|
|
146 * SAL_REM_ACCENTS
|
|
375
|
147 * SAL_SOFO: SOFOFROM and SOFOTO used instead of SAL
|
|
|
148 *
|
|
|
149 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
150 *
|
|
|
151 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
|
152 *
|
|
375
|
153 * <salfromlen> 1-2 bytes length of <salfrom> (2 bytes for SAL_SOFO)
|
|
323
|
154 *
|
|
|
155 * <salfrom> N bytes "from" part of soundsalike
|
|
|
156 *
|
|
375
|
157 * <saltolen> 1-2 bytes length of <salto> (2 bytes for SAL_SOFO)
|
|
323
|
158 *
|
|
|
159 * <salto> N bytes "to" part of soundsalike
|
|
|
160 *
|
|
|
161 * <maplen> 2 bytes length of <mapstr>, MSB first
|
|
|
162 *
|
|
|
163 * <mapstr> N bytes String with sequences of similar characters,
|
|
|
164 * separated by slashes.
|
|
300
|
165 *
|
|
|
166 *
|
|
|
167 * <LWORDTREE>: <wordtree>
|
|
|
168 *
|
|
339
|
169 * <KWORDTREE>: <wordtree>
|
|
|
170 *
|
|
|
171 * <PREFIXTREE>: <wordtree>
|
|
|
172 *
|
|
|
173 *
|
|
300
|
174 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
175 *
|
|
|
176 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
177 *
|
|
|
178 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
179 *
|
|
|
180 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
181 * follow in sorted order.
|
|
|
182 *
|
|
339
|
183 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
455
|
184 * | <flags> [<flags2>] [<region>] [<prefixID>]
|
|
|
185 * | [<pflags>] <prefixID> <prefcondnr> ]
|
|
300
|
186 *
|
|
|
187 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
188 * BY_NOFLAGS: End of word without flags and for all
|
|
|
189 * regions.
|
|
366
|
190 * For PREFIXTREE <prefixID> and
|
|
|
191 * <prefcondnr> follow.
|
|
390
|
192 * BY_FLAGS: End of word, <flags> follow.
|
|
455
|
193 * For PREFIXTREE <pflags>, <prefixID>
|
|
|
194 * and <prefcondnr> follow.
|
|
390
|
195 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
196 * follow. Not used in PREFIXTREE.
|
|
|
197 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
198 * and <xbyte> follow.
|
|
|
199 *
|
|
|
200 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
201 *
|
|
|
202 * <xbyte> 1 byte byte value of the sibling.
|
|
|
203 *
|
|
|
204 * <flags> 1 byte bitmask of:
|
|
|
205 * WF_ALLCAP word must have only capitals
|
|
|
206 * WF_ONECAP first char of word must be capital
|
|
389
|
207 * WF_KEEPCAP keep-case word
|
|
|
208 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
209 * WF_RARE rare word
|
|
389
|
210 * WF_BANNED bad word
|
|
300
|
211 * WF_REGION <region> follows
|
|
339
|
212 * WF_PFX <prefixID> follows
|
|
300
|
213 *
|
|
390
|
214 * <flags2> 1 byte Only used when there are postponed prefixes.
|
|
|
215 * Bitmask of:
|
|
|
216 * WF_HAS_AFF >> 8 word includes affix
|
|
|
217 *
|
|
455
|
218 * <pflags> 1 byte bitmask of:
|
|
|
219 * WFP_RARE rare prefix
|
|
|
220 * WFP_NC non-combining prefix
|
|
|
221 * WFP_UP letter after prefix made upper case
|
|
|
222 *
|
|
300
|
223 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
224 * omitted it's valid in all regions.
|
|
|
225 * Lowest bit is for region 1.
|
|
|
226 *
|
|
339
|
227 * <prefixID> 1 byte ID of prefix that can be used with this word. For
|
|
|
228 * PREFIXTREE used for the required prefix ID.
|
|
|
229 *
|
|
|
230 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
231 * from HEADER.
|
|
300
|
232 *
|
|
|
233 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
234 */
|
|
|
235
|
|
223
|
236 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
237 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
238 #endif
|
|
|
239
|
|
|
240 #include "vim.h"
|
|
|
241
|
|
|
242 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
243
|
|
|
244 #ifdef HAVE_FCNTL_H
|
|
|
245 # include <fcntl.h>
|
|
|
246 #endif
|
|
|
247
|
|
323
|
248 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
249 Some places assume a word length fits in a
|
|
|
250 byte, thus it can't be above 255. */
|
|
226
|
251
|
|
324
|
252 /* Type used for indexes in the word tree need to be at least 3 bytes. If int
|
|
|
253 * is 8 bytes we could use something smaller, but what? */
|
|
|
254 #if SIZEOF_INT > 2
|
|
|
255 typedef int idx_T;
|
|
|
256 #else
|
|
|
257 typedef long idx_T;
|
|
|
258 #endif
|
|
|
259
|
|
|
260 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
261 * comes above it. */
|
|
300
|
262 #define WF_REGION 0x01 /* region byte follows */
|
|
|
263 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
264 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
265 #define WF_RARE 0x08 /* rare word */
|
|
307
|
266 #define WF_BANNED 0x10 /* bad word */
|
|
390
|
267 #define WF_PFX 0x20 /* prefix ID follows */
|
|
389
|
268 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
269 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
270
|
|
390
|
271 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
272 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
|
273
|
|
389
|
274 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
275
|
|
455
|
276 /* flags for <pflags> */
|
|
|
277 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
278 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
279 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
280
|
|
390
|
281 /* flags for postponed prefixes. Must be above prefixID (one byte)
|
|
|
282 * and prefcondnr (two bytes). */
|
|
455
|
283 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
284 * postponed prefix */
|
|
|
285 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
286 * postponed prefix */
|
|
|
287 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
288 * postponed prefix */
|
|
390
|
289
|
|
|
290 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
291 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
292 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
293 * postponed prefix: no <pflags> */
|
|
|
294 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
295 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
296 * postponed prefix: <pflags> follows */
|
|
|
297 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
298 * follow; never used in prefix tree */
|
|
|
299 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
300
|
|
323
|
301 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
302 * and si_sal. Not for sl_sal!
|
|
323
|
303 * One replacement: from "ft_from" to "ft_to". */
|
|
|
304 typedef struct fromto_S
|
|
236
|
305 {
|
|
323
|
306 char_u *ft_from;
|
|
|
307 char_u *ft_to;
|
|
|
308 } fromto_T;
|
|
236
|
309
|
|
344
|
310 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
311 * The info is split for quick processing by spell_soundfold().
|
|
|
312 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
313 typedef struct salitem_S
|
|
|
314 {
|
|
|
315 char_u *sm_lead; /* leading letters */
|
|
|
316 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
317 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
318 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
319 char_u *sm_to; /* replacement. */
|
|
372
|
320 #ifdef FEAT_MBYTE
|
|
|
321 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
322 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
323 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
324 #endif
|
|
344
|
325 } salitem_T;
|
|
|
326
|
|
375
|
327 #ifdef FEAT_MBYTE
|
|
|
328 typedef int salfirst_T;
|
|
|
329 #else
|
|
|
330 typedef short salfirst_T;
|
|
|
331 #endif
|
|
|
332
|
|
236
|
333 /*
|
|
243
|
334 * Structure used to store words and other info for one language, loaded from
|
|
|
335 * a .spl file.
|
|
300
|
336 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
337 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
338 *
|
|
|
339 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
340 * the number of possible bytes, sorted on byte value:
|
|
|
341 * <len> <byte1> <byte2> ...
|
|
|
342 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
343 * byte in "byts".
|
|
|
344 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
390
|
345 * the flags, region mask and prefixID for the word. There may be several
|
|
|
346 * zeros in sequence for alternative flag/region combinations.
|
|
236
|
347 */
|
|
|
348 typedef struct slang_S slang_T;
|
|
|
349 struct slang_S
|
|
|
350 {
|
|
|
351 slang_T *sl_next; /* next language */
|
|
|
352 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
353 char_u *sl_fname; /* name of .spl file */
|
|
323
|
354 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
355
|
|
300
|
356 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
357 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
358 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
359 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
360 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
361 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
362
|
|
236
|
363 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
364
|
|
376
|
365 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
366
|
|
339
|
367 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
368 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
369
|
|
323
|
370 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
371 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
372 there is none */
|
|
344
|
373 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
374 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
375 there is none */
|
|
375
|
376 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
377 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
378 * "sl_sal" is a list of wide char lists. */
|
|
323
|
379 int sl_followup; /* SAL followup */
|
|
|
380 int sl_collapse; /* SAL collapse_result */
|
|
|
381 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
382 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
383 #ifdef FEAT_MBYTE
|
|
|
384 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
385 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
386 #else
|
|
|
387 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
388 #endif
|
|
236
|
389 };
|
|
|
390
|
|
243
|
391 /* First language that is loaded, start of the linked list of loaded
|
|
|
392 * languages. */
|
|
236
|
393 static slang_T *first_lang = NULL;
|
|
|
394
|
|
323
|
395 /* Flags used in .spl file for soundsalike flags. */
|
|
|
396 #define SAL_F0LLOWUP 1
|
|
|
397 #define SAL_COLLAPSE 2
|
|
|
398 #define SAL_REM_ACCENTS 4
|
|
375
|
399 #define SAL_SOFO 8 /* SOFOFROM and SOFOTO instead of SAL */
|
|
323
|
400
|
|
236
|
401 /*
|
|
|
402 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
403 */
|
|
|
404 typedef struct langp_S
|
|
|
405 {
|
|
|
406 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
407 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
408 } langp_T;
|
|
|
409
|
|
|
410 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
411
|
|
307
|
412 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
413
|
|
|
414 /* Result values. Lower number is accepted over higher one. */
|
|
|
415 #define SP_BANNED -1
|
|
236
|
416 #define SP_OK 0
|
|
307
|
417 #define SP_RARE 1
|
|
|
418 #define SP_LOCAL 2
|
|
|
419 #define SP_BAD 3
|
|
236
|
420
|
|
455
|
421 #define VIMSPELLMAGIC "VIMspell09" /* string at start of Vim spell file */
|
|
236
|
422 #define VIMSPELLMAGICL 10
|
|
|
423
|
|
381
|
424 /* file used for "zG" and "zW" */
|
|
385
|
425 static char_u *int_wordlist = NULL;
|
|
381
|
426
|
|
236
|
427 /*
|
|
323
|
428 * Information used when looking for suggestions.
|
|
|
429 */
|
|
|
430 typedef struct suginfo_S
|
|
|
431 {
|
|
|
432 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
433 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
434 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
435 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
436 char_u *su_badptr; /* start of bad word in line */
|
|
|
437 int su_badlen; /* length of detected bad word in line */
|
|
346
|
438 int su_badflags; /* caps flags for bad word */
|
|
323
|
439 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
440 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
441 hashtab_T su_banned; /* table with banned words */
|
|
|
442 } suginfo_T;
|
|
|
443
|
|
|
444 /* One word suggestion. Used in "si_ga". */
|
|
|
445 typedef struct suggest_S
|
|
|
446 {
|
|
|
447 char_u *st_word; /* suggested word, allocated string */
|
|
|
448 int st_orglen; /* length of replaced text */
|
|
|
449 int st_score; /* lower is better */
|
|
344
|
450 int st_altscore; /* used when st_score compares equal */
|
|
|
451 int st_salscore; /* st_score is for soundalike */
|
|
324
|
452 int st_had_bonus; /* bonus already included in score */
|
|
323
|
453 } suggest_T;
|
|
|
454
|
|
344
|
455 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
456
|
|
324
|
457 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
458 * called the score may change, thus we need to keep more than what is
|
|
|
459 * displayed. */
|
|
351
|
460 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
461
|
|
|
462 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
463 * of suggestions that are not going to be displayed. */
|
|
344
|
464 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
465
|
|
|
466 /* score for various changes */
|
|
344
|
467 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
468 #define SCORE_ICASE 52 /* slightly different case */
|
|
|
469 #define SCORE_REGION 70 /* word is for different region */
|
|
|
470 #define SCORE_RARE 180 /* rare word */
|
|
|
471 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
472 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
473 #define SCORE_REP 87 /* REP replacement */
|
|
|
474 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
475 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
324
|
476 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
477 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
324
|
478 #define SCORE_INS 96 /* insert a character */
|
|
358
|
479 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
366
|
480 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
481
|
|
372
|
482 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
483 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
484 * 350 allows for about three changes. */
|
|
344
|
485
|
|
|
486 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
487 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
488
|
|
|
489 /*
|
|
236
|
490 * Structure to store info for word matching.
|
|
|
491 */
|
|
|
492 typedef struct matchinf_S
|
|
|
493 {
|
|
|
494 langp_T *mi_lp; /* info for language and region */
|
|
243
|
495
|
|
|
496 /* pointers to original text to be checked */
|
|
236
|
497 char_u *mi_word; /* start of word being checked */
|
|
339
|
498 char_u *mi_end; /* end of matching word so far */
|
|
243
|
499 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
500 char_u *mi_cend; /* char after what was used for
|
|
|
501 mi_capflags */
|
|
243
|
502
|
|
|
503 /* case-folded text */
|
|
|
504 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
505 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
506
|
|
339
|
507 /* for when checking word after a prefix */
|
|
|
508 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
|
509 prefixID/condition */
|
|
|
510 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
511 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
512 #ifdef FEAT_MBYTE
|
|
|
513 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
514 case */
|
|
|
515 #else
|
|
|
516 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
517 #endif
|
|
339
|
518
|
|
243
|
519 /* others */
|
|
236
|
520 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
521 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
522 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
523 } matchinf_T;
|
|
|
524
|
|
307
|
525 /*
|
|
|
526 * The tables used for recognizing word characters according to spelling.
|
|
|
527 * These are only used for the first 256 characters of 'encoding'.
|
|
|
528 */
|
|
|
529 typedef struct spelltab_S
|
|
|
530 {
|
|
|
531 char_u st_isw[256]; /* flags: is word char */
|
|
|
532 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
533 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
534 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
535 } spelltab_T;
|
|
|
536
|
|
|
537 static spelltab_T spelltab;
|
|
|
538 static int did_set_spelltab;
|
|
|
539
|
|
324
|
540 #define CF_WORD 0x01
|
|
|
541 #define CF_UPPER 0x02
|
|
307
|
542
|
|
|
543 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
544 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
545 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
546 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
547 #ifdef FEAT_MBYTE
|
|
|
548 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
549 #endif
|
|
339
|
550 static void write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
551
|
|
|
552 /*
|
|
344
|
553 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
554 */
|
|
|
555 typedef enum
|
|
|
556 {
|
|
344
|
557 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
558 * ends); if badword ends there is a match, otherwise
|
|
|
559 * try splitting word. */
|
|
375
|
560 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
561 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
562 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
563 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
564 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
565 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
566 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
567 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
568 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
569 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
570 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
571 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
572 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
573 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
574 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
575 STATE_FINAL /* End of this node. */
|
|
|
576 } state_T;
|
|
|
577
|
|
|
578 /*
|
|
346
|
579 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
580 */
|
|
|
581 typedef struct trystate_S
|
|
|
582 {
|
|
330
|
583 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
584 int ts_score; /* score */
|
|
344
|
585 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
586 short ts_curi; /* index in list of child nodes */
|
|
|
587 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
588 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
589 char_u ts_twordlen; /* valid length of tword[] */
|
|
375
|
590 char_u ts_prefixdepth; /* stack depth for end of prefix or PREFIXTREE
|
|
|
591 * or NOPREFIX */
|
|
330
|
592 #ifdef FEAT_MBYTE
|
|
|
593 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
594 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
595 char_u ts_isdiff; /* DIFF_ values */
|
|
|
596 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
597 #endif
|
|
323
|
598 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
330
|
599 char_u ts_save_splitoff; /* su_splitoff saved here */
|
|
346
|
600 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
601 } trystate_T;
|
|
|
602
|
|
330
|
603 /* values for ts_isdiff */
|
|
|
604 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
605 #define DIFF_YES 1 /* different byte found */
|
|
|
606 #define DIFF_INSERT 2 /* inserting character */
|
|
|
607
|
|
375
|
608 /* special values ts_prefixdepth */
|
|
|
609 #define PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
610 #define NOPREFIX 0xff /* not using prefixes */
|
|
|
611
|
|
339
|
612 /* mode values for find_word */
|
|
|
613 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
614 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
615 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
616
|
|
236
|
617 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
618 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
619 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
620 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
455
|
621 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
339
|
622 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
623 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
624 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
625 static int no_spell_checking __ARGS((void));
|
|
307
|
626 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
627 static char_u *spell_enc __ARGS((void));
|
|
385
|
628 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
629 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
630 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
631 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
381
|
632 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
633 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
634 #ifdef FEAT_MBYTE
|
|
|
635 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
636 #endif
|
|
339
|
637 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
638 static void clear_midword __ARGS((buf_T *buf));
|
|
|
639 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
640 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
641 static int captype __ARGS((char_u *word, char_u *end));
|
|
323
|
642 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
389
|
643 static int set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
644 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
645 static void write_spell_chartab __ARGS((FILE *fd));
|
|
|
646 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
388
|
647 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
648 #ifdef FEAT_EVAL
|
|
|
649 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
650 #endif
|
|
|
651 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
652 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
653 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
654 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
655 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
656 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
657 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
658 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
659 #ifdef FEAT_MBYTE
|
|
|
660 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
661 #endif
|
|
323
|
662 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
663 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
664 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
665 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
666 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
667 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
668 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
669 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
670 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
671 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
672 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
673 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
674 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
675 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
676 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
677 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
678 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
679 #ifdef FEAT_MBYTE
|
|
375
|
680 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
681 #endif
|
|
344
|
682 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
683 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
684 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
685 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
686
|
|
324
|
687 /*
|
|
|
688 * Use our own character-case definitions, because the current locale may
|
|
|
689 * differ from what the .spl file uses.
|
|
|
690 * These must not be called with negative number!
|
|
|
691 */
|
|
|
692 #ifndef FEAT_MBYTE
|
|
|
693 /* Non-multi-byte implementation. */
|
|
|
694 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
695 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
696 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
697 #else
|
|
447
|
698 # if defined(HAVE_WCHAR_H)
|
|
|
699 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
700 # endif
|
|
324
|
701 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
702 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
703 * the "w" library function for characters above 255 if available. */
|
|
|
704 # ifdef HAVE_TOWLOWER
|
|
|
705 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
706 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
707 # else
|
|
|
708 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
709 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
710 # endif
|
|
|
711
|
|
|
712 # ifdef HAVE_TOWUPPER
|
|
|
713 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
714 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
715 # else
|
|
|
716 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
717 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
718 # endif
|
|
|
719
|
|
|
720 # ifdef HAVE_ISWUPPER
|
|
|
721 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
722 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
723 # else
|
|
|
724 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
725 : (c) < 256 ? spelltab.st_isu[c] : (c))
|
|
|
726 # endif
|
|
|
727 #endif
|
|
|
728
|
|
307
|
729
|
|
|
730 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
731 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
469
|
732 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
733
|
|
|
734 /*
|
|
|
735 * Main spell-checking function.
|
|
300
|
736 * "ptr" points to a character that could be the start of a word.
|
|
236
|
737 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
738 * or when it's OK it remains unchanged.
|
|
|
739 * This must only be called when 'spelllang' is not empty.
|
|
323
|
740 *
|
|
385
|
741 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
742 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
743 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
744 * worry.
|
|
323
|
745 *
|
|
236
|
746 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
747 * caller can skip over the word.
|
|
|
748 */
|
|
|
749 int
|
|
385
|
750 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
751 win_T *wp; /* current window */
|
|
|
752 char_u *ptr;
|
|
|
753 int *attrp;
|
|
385
|
754 int *capcol; /* column to check for Capital */
|
|
236
|
755 {
|
|
|
756 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
757 be passed to functions quickly. */
|
|
344
|
758 int nrlen = 0; /* found a number first */
|
|
385
|
759 int c;
|
|
236
|
760
|
|
307
|
761 /* A word never starts at a space or a control character. Return quickly
|
|
|
762 * then, skipping over the character. */
|
|
|
763 if (*ptr <= ' ')
|
|
|
764 return 1;
|
|
236
|
765
|
|
344
|
766 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
767 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
768 * "3GPP". */
|
|
300
|
769 if (*ptr >= '0' && *ptr <= '9')
|
|
|
770 {
|
|
316
|
771 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
772 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
773 else
|
|
|
774 {
|
|
344
|
775 mi.mi_end = skipdigits(ptr);
|
|
|
776 nrlen = mi.mi_end - ptr;
|
|
300
|
777 }
|
|
376
|
778 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
779 return (int)(mi.mi_end - ptr);
|
|
346
|
780
|
|
|
781 /* Try including the digits in the word. */
|
|
|
782 mi.mi_fend = ptr + nrlen;
|
|
|
783 }
|
|
|
784 else
|
|
|
785 mi.mi_fend = ptr;
|
|
|
786
|
|
|
787 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
788 mi.mi_word = ptr;
|
|
376
|
789 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
790 {
|
|
|
791 do
|
|
|
792 {
|
|
307
|
793 mb_ptr_adv(mi.mi_fend);
|
|
376
|
794 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
795
|
|
|
796 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
797 {
|
|
|
798 /* Check word starting with capital letter. */
|
|
455
|
799 c = PTR2CHAR(ptr);
|
|
385
|
800 if (!SPELL_ISUPPER(c))
|
|
|
801 {
|
|
|
802 *attrp = highlight_attr[HLF_SPC];
|
|
|
803 return (int)(mi.mi_fend - ptr);
|
|
|
804 }
|
|
|
805 }
|
|
|
806 }
|
|
|
807 if (capcol != NULL)
|
|
|
808 *capcol = -1;
|
|
344
|
809
|
|
|
810 /* We always use the characters up to the next non-word character,
|
|
|
811 * also for bad words. */
|
|
|
812 mi.mi_end = mi.mi_fend;
|
|
|
813
|
|
|
814 /* Check caps type later. */
|
|
|
815 mi.mi_capflags = 0;
|
|
|
816 mi.mi_cend = NULL;
|
|
376
|
817 mi.mi_buf = wp->w_buffer;
|
|
344
|
818
|
|
|
819 /* Include one non-word character so that we can check for the
|
|
|
820 * word end. */
|
|
|
821 if (*mi.mi_fend != NUL)
|
|
|
822 mb_ptr_adv(mi.mi_fend);
|
|
|
823
|
|
|
824 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
825 MAXWLEN + 1);
|
|
|
826 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
827
|
|
|
828 /* The word is bad unless we recognize it. */
|
|
|
829 mi.mi_result = SP_BAD;
|
|
|
830
|
|
|
831 /*
|
|
|
832 * Loop over the languages specified in 'spelllang'.
|
|
|
833 * We check them all, because a matching word may be longer than an
|
|
|
834 * already found matching word.
|
|
|
835 */
|
|
|
836 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
837 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
838 {
|
|
|
839 /* Check for a matching word in case-folded words. */
|
|
|
840 find_word(&mi, FIND_FOLDWORD);
|
|
|
841
|
|
|
842 /* Check for a matching word in keep-case words. */
|
|
|
843 find_word(&mi, FIND_KEEPWORD);
|
|
|
844
|
|
|
845 /* Check for matching prefixes. */
|
|
|
846 find_prefix(&mi);
|
|
|
847 }
|
|
|
848
|
|
|
849 if (mi.mi_result != SP_OK)
|
|
|
850 {
|
|
346
|
851 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
852 * rare and local words, e.g., "3GPP". */
|
|
344
|
853 if (nrlen > 0)
|
|
346
|
854 {
|
|
|
855 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
856 return nrlen;
|
|
|
857 }
|
|
344
|
858
|
|
|
859 /* When we are at a non-word character there is no error, just
|
|
|
860 * skip over the character (try looking for a word after it). */
|
|
390
|
861 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
862 {
|
|
385
|
863 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
864 {
|
|
|
865 regmatch_T regmatch;
|
|
|
866
|
|
|
867 /* Check for end of sentence. */
|
|
|
868 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
869 regmatch.rm_ic = FALSE;
|
|
|
870 if (vim_regexec(®match, ptr, 0))
|
|
|
871 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
872 }
|
|
|
873
|
|
344
|
874 #ifdef FEAT_MBYTE
|
|
|
875 if (has_mbyte)
|
|
|
876 return mb_ptr2len_check(ptr);
|
|
|
877 #endif
|
|
|
878 return 1;
|
|
300
|
879 }
|
|
243
|
880
|
|
344
|
881 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
882 *attrp = highlight_attr[HLF_SPB];
|
|
|
883 else if (mi.mi_result == SP_RARE)
|
|
|
884 *attrp = highlight_attr[HLF_SPR];
|
|
|
885 else
|
|
|
886 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
887 }
|
|
|
888
|
|
300
|
889 return (int)(mi.mi_end - ptr);
|
|
236
|
890 }
|
|
|
891
|
|
|
892 /*
|
|
300
|
893 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
894 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
895 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
896 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
897 * tree.
|
|
300
|
898 *
|
|
|
899 * For a match mip->mi_result is updated.
|
|
243
|
900 */
|
|
|
901 static void
|
|
339
|
902 find_word(mip, mode)
|
|
243
|
903 matchinf_T *mip;
|
|
339
|
904 int mode;
|
|
243
|
905 {
|
|
324
|
906 idx_T arridx = 0;
|
|
300
|
907 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
908 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
909 int endidxcnt = 0;
|
|
|
910 int len;
|
|
|
911 int wlen = 0;
|
|
|
912 int flen;
|
|
|
913 int c;
|
|
|
914 char_u *ptr;
|
|
324
|
915 idx_T lo, hi, m;
|
|
243
|
916 #ifdef FEAT_MBYTE
|
|
300
|
917 char_u *s;
|
|
339
|
918 char_u *p;
|
|
307
|
919 #endif
|
|
|
920 int res = SP_BAD;
|
|
300
|
921 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
922 unsigned flags;
|
|
|
923 char_u *byts;
|
|
324
|
924 idx_T *idxs;
|
|
339
|
925
|
|
|
926 if (mode == FIND_KEEPWORD)
|
|
236
|
927 {
|
|
300
|
928 /* Check for word with matching case in keep-case tree. */
|
|
|
929 ptr = mip->mi_word;
|
|
|
930 flen = 9999; /* no case folding, always enough bytes */
|
|
|
931 byts = slang->sl_kbyts;
|
|
|
932 idxs = slang->sl_kidxs;
|
|
236
|
933 }
|
|
|
934 else
|
|
|
935 {
|
|
300
|
936 /* Check for case-folded in case-folded tree. */
|
|
|
937 ptr = mip->mi_fword;
|
|
|
938 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
939 byts = slang->sl_fbyts;
|
|
|
940 idxs = slang->sl_fidxs;
|
|
339
|
941
|
|
|
942 if (mode == FIND_PREFIX)
|
|
|
943 {
|
|
|
944 /* Skip over the prefix. */
|
|
|
945 wlen = mip->mi_prefixlen;
|
|
|
946 flen -= mip->mi_prefixlen;
|
|
|
947 }
|
|
243
|
948 }
|
|
|
949
|
|
300
|
950 if (byts == NULL)
|
|
|
951 return; /* array is empty */
|
|
236
|
952
|
|
|
953 /*
|
|
307
|
954 * Repeat advancing in the tree until:
|
|
|
955 * - there is a byte that doesn't match,
|
|
|
956 * - we reach the end of the tree,
|
|
|
957 * - or we reach the end of the line.
|
|
236
|
958 */
|
|
300
|
959 for (;;)
|
|
236
|
960 {
|
|
346
|
961 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
962 flen = fold_more(mip);
|
|
300
|
963
|
|
|
964 len = byts[arridx++];
|
|
|
965
|
|
|
966 /* If the first possible byte is a zero the word could end here.
|
|
|
967 * Remember this index, we first check for the longest word. */
|
|
|
968 if (byts[arridx] == 0)
|
|
|
969 {
|
|
307
|
970 if (endidxcnt == MAXWLEN)
|
|
|
971 {
|
|
|
972 /* Must be a corrupted spell file. */
|
|
|
973 EMSG(_(e_format));
|
|
|
974 return;
|
|
|
975 }
|
|
300
|
976 endlen[endidxcnt] = wlen;
|
|
|
977 endidx[endidxcnt++] = arridx++;
|
|
|
978 --len;
|
|
|
979
|
|
|
980 /* Skip over the zeros, there can be several flag/region
|
|
|
981 * combinations. */
|
|
|
982 while (len > 0 && byts[arridx] == 0)
|
|
|
983 {
|
|
|
984 ++arridx;
|
|
|
985 --len;
|
|
|
986 }
|
|
|
987 if (len == 0)
|
|
|
988 break; /* no children, word must end here */
|
|
|
989 }
|
|
|
990
|
|
|
991 /* Stop looking at end of the line. */
|
|
|
992 if (ptr[wlen] == NUL)
|
|
|
993 break;
|
|
|
994
|
|
|
995 /* Perform a binary search in the list of accepted bytes. */
|
|
|
996 c = ptr[wlen];
|
|
346
|
997 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
998 c = ' ';
|
|
300
|
999 lo = arridx;
|
|
|
1000 hi = arridx + len - 1;
|
|
|
1001 while (lo < hi)
|
|
|
1002 {
|
|
|
1003 m = (lo + hi) / 2;
|
|
|
1004 if (byts[m] > c)
|
|
|
1005 hi = m - 1;
|
|
|
1006 else if (byts[m] < c)
|
|
|
1007 lo = m + 1;
|
|
|
1008 else
|
|
|
1009 {
|
|
|
1010 lo = hi = m;
|
|
|
1011 break;
|
|
236
|
1012 }
|
|
|
1013 }
|
|
300
|
1014
|
|
|
1015 /* Stop if there is no matching byte. */
|
|
|
1016 if (hi < lo || byts[lo] != c)
|
|
|
1017 break;
|
|
|
1018
|
|
|
1019 /* Continue at the child (if there is one). */
|
|
|
1020 arridx = idxs[lo];
|
|
|
1021 ++wlen;
|
|
|
1022 --flen;
|
|
346
|
1023
|
|
|
1024 /* One space in the good word may stand for several spaces in the
|
|
|
1025 * checked word. */
|
|
|
1026 if (c == ' ')
|
|
|
1027 {
|
|
|
1028 for (;;)
|
|
|
1029 {
|
|
|
1030 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1031 flen = fold_more(mip);
|
|
|
1032 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1033 break;
|
|
|
1034 ++wlen;
|
|
|
1035 --flen;
|
|
|
1036 }
|
|
|
1037 }
|
|
236
|
1038 }
|
|
|
1039
|
|
300
|
1040 /*
|
|
|
1041 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1042 * first.
|
|
|
1043 */
|
|
|
1044 while (endidxcnt > 0)
|
|
|
1045 {
|
|
|
1046 --endidxcnt;
|
|
|
1047 arridx = endidx[endidxcnt];
|
|
|
1048 wlen = endlen[endidxcnt];
|
|
236
|
1049
|
|
300
|
1050 #ifdef FEAT_MBYTE
|
|
|
1051 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1052 continue; /* not at first byte of character */
|
|
|
1053 #endif
|
|
376
|
1054 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
300
|
1055 continue; /* next char is a word character */
|
|
|
1056
|
|
|
1057 #ifdef FEAT_MBYTE
|
|
339
|
1058 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1059 {
|
|
|
1060 /* Compute byte length in original word, length may change
|
|
339
|
1061 * when folding case. This can be slow, take a shortcut when the
|
|
|
1062 * case-folded word is equal to the keep-case word. */
|
|
300
|
1063 p = mip->mi_word;
|
|
339
|
1064 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1065 {
|
|
|
1066 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1067 mb_ptr_adv(p);
|
|
|
1068 wlen = p - mip->mi_word;
|
|
|
1069 }
|
|
300
|
1070 }
|
|
|
1071 #endif
|
|
236
|
1072
|
|
339
|
1073 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1074 * prefix ID.
|
|
|
1075 * Repeat this if there are more flags/region alternatives until there
|
|
|
1076 * is a match. */
|
|
|
1077 res = SP_BAD;
|
|
|
1078 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1079 --len, ++arridx)
|
|
300
|
1080 {
|
|
|
1081 flags = idxs[arridx];
|
|
324
|
1082
|
|
339
|
1083 /* For the fold-case tree check that the case of the checked word
|
|
|
1084 * matches with what the word in the tree requires.
|
|
|
1085 * For keep-case tree the case is always right. For prefixes we
|
|
|
1086 * don't bother to check. */
|
|
|
1087 if (mode == FIND_FOLDWORD)
|
|
300
|
1088 {
|
|
|
1089 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1090 {
|
|
323
|
1091 /* mi_capflags was set for a different word length, need
|
|
|
1092 * to do it again. */
|
|
300
|
1093 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1094 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1095 }
|
|
|
1096
|
|
346
|
1097 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1098 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1099 continue;
|
|
300
|
1100 }
|
|
236
|
1101
|
|
339
|
1102 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1103 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1104 * mip->mi_prefarridx that find_prefix() filled. */
|
|
339
|
1105 if (mode == FIND_PREFIX)
|
|
300
|
1106 {
|
|
339
|
1107 /* The prefix ID is stored two bytes above the flags. */
|
|
366
|
1108 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1109 flags,
|
|
455
|
1110 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1111 FALSE);
|
|
366
|
1112 if (c == 0)
|
|
339
|
1113 continue;
|
|
366
|
1114
|
|
|
1115 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1116 if (c & WF_RAREPFX)
|
|
|
1117 flags |= WF_RARE;
|
|
339
|
1118 }
|
|
|
1119
|
|
|
1120 if (flags & WF_BANNED)
|
|
|
1121 res = SP_BANNED;
|
|
|
1122 else if (flags & WF_REGION)
|
|
|
1123 {
|
|
|
1124 /* Check region. */
|
|
390
|
1125 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1126 res = SP_OK;
|
|
339
|
1127 else
|
|
|
1128 res = SP_LOCAL;
|
|
300
|
1129 }
|
|
339
|
1130 else if (flags & WF_RARE)
|
|
|
1131 res = SP_RARE;
|
|
307
|
1132 else
|
|
339
|
1133 res = SP_OK;
|
|
|
1134
|
|
|
1135 /* Always use the longest match and the best result. */
|
|
|
1136 if (mip->mi_result > res)
|
|
|
1137 {
|
|
|
1138 mip->mi_result = res;
|
|
|
1139 mip->mi_end = mip->mi_word + wlen;
|
|
|
1140 }
|
|
351
|
1141 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1142 mip->mi_end = mip->mi_word + wlen;
|
|
|
1143
|
|
|
1144 if (res == SP_OK)
|
|
|
1145 break;
|
|
300
|
1146 }
|
|
|
1147
|
|
307
|
1148 if (res == SP_OK)
|
|
300
|
1149 break;
|
|
|
1150 }
|
|
236
|
1151 }
|
|
|
1152
|
|
323
|
1153 /*
|
|
390
|
1154 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1155 * ID in "flags" for the word "word".
|
|
366
|
1156 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1157 */
|
|
|
1158 static int
|
|
455
|
1159 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1160 int totprefcnt; /* nr of prefix IDs */
|
|
|
1161 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1162 int flags;
|
|
351
|
1163 char_u *word;
|
|
|
1164 slang_T *slang;
|
|
455
|
1165 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1166 {
|
|
|
1167 int prefcnt;
|
|
|
1168 int pidx;
|
|
|
1169 regprog_T *rp;
|
|
|
1170 regmatch_T regmatch;
|
|
390
|
1171 int prefid;
|
|
|
1172
|
|
|
1173 prefid = (unsigned)flags >> 24;
|
|
351
|
1174 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1175 {
|
|
|
1176 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1177
|
|
|
1178 /* Check the prefix ID. */
|
|
|
1179 if (prefid != (pidx & 0xff))
|
|
|
1180 continue;
|
|
|
1181
|
|
390
|
1182 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1183 * suffix. */
|
|
|
1184 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1185 continue;
|
|
|
1186
|
|
351
|
1187 /* Check the condition, if there is one. The condition index is
|
|
366
|
1188 * stored in the two bytes above the prefix ID byte. */
|
|
|
1189 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1190 if (rp != NULL)
|
|
|
1191 {
|
|
|
1192 regmatch.regprog = rp;
|
|
|
1193 regmatch.rm_ic = FALSE;
|
|
|
1194 if (!vim_regexec(®match, word, 0))
|
|
|
1195 continue;
|
|
|
1196 }
|
|
455
|
1197 else if (cond_req)
|
|
|
1198 continue;
|
|
|
1199
|
|
|
1200 /* It's a match! Return the WF_ flags. */
|
|
366
|
1201 return pidx;
|
|
|
1202 }
|
|
|
1203 return 0;
|
|
351
|
1204 }
|
|
|
1205
|
|
|
1206 /*
|
|
339
|
1207 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1208 * If it does, then check the following word.
|
|
|
1209 *
|
|
|
1210 * For a match mip->mi_result is updated.
|
|
|
1211 */
|
|
|
1212 static void
|
|
|
1213 find_prefix(mip)
|
|
|
1214 matchinf_T *mip;
|
|
|
1215 {
|
|
|
1216 idx_T arridx = 0;
|
|
|
1217 int len;
|
|
|
1218 int wlen = 0;
|
|
|
1219 int flen;
|
|
|
1220 int c;
|
|
|
1221 char_u *ptr;
|
|
|
1222 idx_T lo, hi, m;
|
|
|
1223 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1224 char_u *byts;
|
|
|
1225 idx_T *idxs;
|
|
|
1226
|
|
375
|
1227 byts = slang->sl_pbyts;
|
|
|
1228 if (byts == NULL)
|
|
|
1229 return; /* array is empty */
|
|
|
1230
|
|
339
|
1231 /* We use the case-folded word here, since prefixes are always
|
|
|
1232 * case-folded. */
|
|
|
1233 ptr = mip->mi_fword;
|
|
|
1234 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1235 idxs = slang->sl_pidxs;
|
|
|
1236
|
|
|
1237 /*
|
|
|
1238 * Repeat advancing in the tree until:
|
|
|
1239 * - there is a byte that doesn't match,
|
|
|
1240 * - we reach the end of the tree,
|
|
|
1241 * - or we reach the end of the line.
|
|
|
1242 */
|
|
|
1243 for (;;)
|
|
|
1244 {
|
|
|
1245 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1246 flen = fold_more(mip);
|
|
|
1247
|
|
|
1248 len = byts[arridx++];
|
|
|
1249
|
|
|
1250 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1251 * Check if the following word matches and supports the prefix. */
|
|
|
1252 if (byts[arridx] == 0)
|
|
|
1253 {
|
|
|
1254 /* There can be several prefixes with different conditions. We
|
|
|
1255 * try them all, since we don't know which one will give the
|
|
|
1256 * longest match. The word is the same each time, pass the list
|
|
|
1257 * of possible prefixes to find_word(). */
|
|
|
1258 mip->mi_prefarridx = arridx;
|
|
|
1259 mip->mi_prefcnt = len;
|
|
|
1260 while (len > 0 && byts[arridx] == 0)
|
|
|
1261 {
|
|
|
1262 ++arridx;
|
|
|
1263 --len;
|
|
|
1264 }
|
|
|
1265 mip->mi_prefcnt -= len;
|
|
|
1266
|
|
|
1267 /* Find the word that comes after the prefix. */
|
|
|
1268 mip->mi_prefixlen = wlen;
|
|
455
|
1269 #ifdef FEAT_MBYTE
|
|
|
1270 if (has_mbyte)
|
|
|
1271 {
|
|
|
1272 /* Case-folded length may differ from original length. */
|
|
|
1273 mip->mi_cprefixlen = nofold_len(mip->mi_fword, wlen,
|
|
|
1274 mip->mi_word);
|
|
|
1275 }
|
|
|
1276 else
|
|
|
1277 mip->mi_cprefixlen = wlen;
|
|
|
1278 #endif
|
|
339
|
1279 find_word(mip, FIND_PREFIX);
|
|
|
1280
|
|
|
1281
|
|
|
1282 if (len == 0)
|
|
|
1283 break; /* no children, word must end here */
|
|
|
1284 }
|
|
|
1285
|
|
|
1286 /* Stop looking at end of the line. */
|
|
|
1287 if (ptr[wlen] == NUL)
|
|
|
1288 break;
|
|
|
1289
|
|
|
1290 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1291 c = ptr[wlen];
|
|
|
1292 lo = arridx;
|
|
|
1293 hi = arridx + len - 1;
|
|
|
1294 while (lo < hi)
|
|
|
1295 {
|
|
|
1296 m = (lo + hi) / 2;
|
|
|
1297 if (byts[m] > c)
|
|
|
1298 hi = m - 1;
|
|
|
1299 else if (byts[m] < c)
|
|
|
1300 lo = m + 1;
|
|
|
1301 else
|
|
|
1302 {
|
|
|
1303 lo = hi = m;
|
|
|
1304 break;
|
|
|
1305 }
|
|
|
1306 }
|
|
|
1307
|
|
|
1308 /* Stop if there is no matching byte. */
|
|
|
1309 if (hi < lo || byts[lo] != c)
|
|
|
1310 break;
|
|
|
1311
|
|
|
1312 /* Continue at the child (if there is one). */
|
|
|
1313 arridx = idxs[lo];
|
|
|
1314 ++wlen;
|
|
|
1315 --flen;
|
|
|
1316 }
|
|
|
1317 }
|
|
|
1318
|
|
|
1319 /*
|
|
|
1320 * Need to fold at least one more character. Do until next non-word character
|
|
|
1321 * for efficiency.
|
|
|
1322 * Return the length of the folded chars in bytes.
|
|
|
1323 */
|
|
|
1324 static int
|
|
|
1325 fold_more(mip)
|
|
|
1326 matchinf_T *mip;
|
|
|
1327 {
|
|
|
1328 int flen;
|
|
|
1329 char_u *p;
|
|
|
1330
|
|
|
1331 p = mip->mi_fend;
|
|
|
1332 do
|
|
|
1333 {
|
|
|
1334 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1335 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1336
|
|
|
1337 /* Include the non-word character so that we can check for the
|
|
|
1338 * word end. */
|
|
|
1339 if (*mip->mi_fend != NUL)
|
|
|
1340 mb_ptr_adv(mip->mi_fend);
|
|
|
1341
|
|
|
1342 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1343 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1344 MAXWLEN - mip->mi_fwordlen);
|
|
|
1345 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1346 mip->mi_fwordlen += flen;
|
|
|
1347 return flen;
|
|
|
1348 }
|
|
|
1349
|
|
|
1350 /*
|
|
323
|
1351 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1352 * case.
|
|
|
1353 */
|
|
|
1354 static int
|
|
389
|
1355 spell_valid_case(wordflags, treeflags)
|
|
|
1356 int wordflags; /* flags for the checked word. */
|
|
323
|
1357 int treeflags; /* flags for the word in the spell tree */
|
|
|
1358 {
|
|
389
|
1359 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1360 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
389
|
1361 && ((treeflags & WF_ONECAP) == 0 || wordflags == WF_ONECAP)));
|
|
323
|
1362 }
|
|
|
1363
|
|
351
|
1364 /*
|
|
|
1365 * Return TRUE if spell checking is not enabled.
|
|
|
1366 */
|
|
|
1367 static int
|
|
|
1368 no_spell_checking()
|
|
|
1369 {
|
|
|
1370 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1371 {
|
|
|
1372 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1373 return TRUE;
|
|
|
1374 }
|
|
|
1375 return FALSE;
|
|
|
1376 }
|
|
300
|
1377
|
|
236
|
1378 /*
|
|
|
1379 * Move to next spell error.
|
|
323
|
1380 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
236
|
1381 * Return OK if found, FAIL otherwise.
|
|
|
1382 */
|
|
|
1383 int
|
|
323
|
1384 spell_move_to(dir, allwords, curline)
|
|
236
|
1385 int dir; /* FORWARD or BACKWARD */
|
|
|
1386 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1387 int curline;
|
|
236
|
1388 {
|
|
249
|
1389 linenr_T lnum;
|
|
|
1390 pos_T found_pos;
|
|
236
|
1391 char_u *line;
|
|
|
1392 char_u *p;
|
|
346
|
1393 char_u *endp;
|
|
|
1394 int attr;
|
|
236
|
1395 int len;
|
|
249
|
1396 int has_syntax = syntax_present(curbuf);
|
|
|
1397 int col;
|
|
|
1398 int can_spell;
|
|
346
|
1399 char_u *buf = NULL;
|
|
|
1400 int buflen = 0;
|
|
|
1401 int skip = 0;
|
|
385
|
1402 int capcol = -1;
|
|
236
|
1403
|
|
351
|
1404 if (no_spell_checking())
|
|
236
|
1405 return FAIL;
|
|
|
1406
|
|
249
|
1407 /*
|
|
|
1408 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1409 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1410 *
|
|
|
1411 * When searching backwards, we continue in the line to find the last
|
|
|
1412 * bad word (in the cursor line: before the cursor).
|
|
346
|
1413 *
|
|
|
1414 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1415 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1416 * though...
|
|
249
|
1417 */
|
|
|
1418 lnum = curwin->w_cursor.lnum;
|
|
|
1419 found_pos.lnum = 0;
|
|
236
|
1420
|
|
|
1421 while (!got_int)
|
|
|
1422 {
|
|
249
|
1423 line = ml_get(lnum);
|
|
346
|
1424
|
|
|
1425 len = STRLEN(line);
|
|
|
1426 if (buflen < len + MAXWLEN + 2)
|
|
|
1427 {
|
|
|
1428 vim_free(buf);
|
|
|
1429 buflen = len + MAXWLEN + 2;
|
|
|
1430 buf = alloc(buflen);
|
|
|
1431 if (buf == NULL)
|
|
|
1432 break;
|
|
|
1433 }
|
|
|
1434
|
|
385
|
1435 /* In first line check first word for Capital. */
|
|
|
1436 if (lnum == 1)
|
|
|
1437 capcol = 0;
|
|
|
1438
|
|
|
1439 /* For checking first word with a capital skip white space. */
|
|
|
1440 if (capcol == 0)
|
|
|
1441 capcol = skipwhite(line) - line;
|
|
|
1442
|
|
346
|
1443 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1444 * possible. */
|
|
|
1445 STRCPY(buf, line);
|
|
|
1446 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1447 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1448
|
|
|
1449 p = buf + skip;
|
|
|
1450 endp = buf + len;
|
|
|
1451 while (p < endp)
|
|
236
|
1452 {
|
|
300
|
1453 /* When searching backward don't search after the cursor. */
|
|
|
1454 if (dir == BACKWARD
|
|
|
1455 && lnum == curwin->w_cursor.lnum
|
|
346
|
1456 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1457 break;
|
|
249
|
1458
|
|
300
|
1459 /* start of word */
|
|
346
|
1460 attr = 0;
|
|
385
|
1461 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1462
|
|
300
|
1463 if (attr != 0)
|
|
|
1464 {
|
|
|
1465 /* We found a bad word. Check the attribute. */
|
|
|
1466 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1467 {
|
|
300
|
1468 /* When searching forward only accept a bad word after
|
|
|
1469 * the cursor. */
|
|
|
1470 if (dir == BACKWARD
|
|
|
1471 || lnum > curwin->w_cursor.lnum
|
|
|
1472 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1473 && (colnr_T)(curline ? p - buf + len
|
|
|
1474 : p - buf)
|
|
300
|
1475 > curwin->w_cursor.col))
|
|
236
|
1476 {
|
|
300
|
1477 if (has_syntax)
|
|
249
|
1478 {
|
|
346
|
1479 col = p - buf;
|
|
300
|
1480 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1481 FALSE, &can_spell);
|
|
|
1482 }
|
|
|
1483 else
|
|
|
1484 can_spell = TRUE;
|
|
249
|
1485
|
|
300
|
1486 if (can_spell)
|
|
|
1487 {
|
|
|
1488 found_pos.lnum = lnum;
|
|
346
|
1489 found_pos.col = p - buf;
|
|
249
|
1490 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1491 found_pos.coladd = 0;
|
|
249
|
1492 #endif
|
|
300
|
1493 if (dir == FORWARD)
|
|
|
1494 {
|
|
|
1495 /* No need to search further. */
|
|
|
1496 curwin->w_cursor = found_pos;
|
|
346
|
1497 vim_free(buf);
|
|
300
|
1498 return OK;
|
|
249
|
1499 }
|
|
|
1500 }
|
|
236
|
1501 }
|
|
|
1502 }
|
|
|
1503 }
|
|
|
1504
|
|
300
|
1505 /* advance to character after the word */
|
|
|
1506 p += len;
|
|
385
|
1507 capcol -= len;
|
|
236
|
1508 }
|
|
|
1509
|
|
323
|
1510 if (curline)
|
|
346
|
1511 break; /* only check cursor line */
|
|
323
|
1512
|
|
236
|
1513 /* Advance to next line. */
|
|
249
|
1514 if (dir == BACKWARD)
|
|
|
1515 {
|
|
|
1516 if (found_pos.lnum != 0)
|
|
|
1517 {
|
|
|
1518 /* Use the last match in the line. */
|
|
|
1519 curwin->w_cursor = found_pos;
|
|
346
|
1520 vim_free(buf);
|
|
249
|
1521 return OK;
|
|
|
1522 }
|
|
|
1523 if (lnum == 1)
|
|
346
|
1524 break;
|
|
249
|
1525 --lnum;
|
|
385
|
1526 capcol = -1;
|
|
249
|
1527 }
|
|
|
1528 else
|
|
|
1529 {
|
|
|
1530 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1531 break;
|
|
249
|
1532 ++lnum;
|
|
346
|
1533
|
|
|
1534 /* Skip the characters at the start of the next line that were
|
|
|
1535 * included in a match crossing line boundaries. */
|
|
|
1536 if (attr == 0)
|
|
|
1537 skip = p - endp;
|
|
|
1538 else
|
|
|
1539 skip = 0;
|
|
385
|
1540
|
|
|
1541 /* Capscol skips over the inserted space. */
|
|
|
1542 --capcol;
|
|
|
1543
|
|
|
1544 /* But after empty line check first word in next line */
|
|
|
1545 if (*skipwhite(line) == NUL)
|
|
|
1546 capcol = 0;
|
|
249
|
1547 }
|
|
236
|
1548
|
|
|
1549 line_breakcheck();
|
|
|
1550 }
|
|
|
1551
|
|
346
|
1552 vim_free(buf);
|
|
|
1553 return FAIL;
|
|
|
1554 }
|
|
|
1555
|
|
|
1556 /*
|
|
|
1557 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1558 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1559 */
|
|
|
1560 void
|
|
|
1561 spell_cat_line(buf, line, maxlen)
|
|
|
1562 char_u *buf;
|
|
|
1563 char_u *line;
|
|
|
1564 int maxlen;
|
|
|
1565 {
|
|
|
1566 char_u *p;
|
|
|
1567 int n;
|
|
|
1568
|
|
|
1569 p = skipwhite(line);
|
|
|
1570 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1571 p = skipwhite(p + 1);
|
|
|
1572
|
|
|
1573 if (*p != NUL)
|
|
|
1574 {
|
|
|
1575 *buf = ' ';
|
|
376
|
1576 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1577 n = p - line;
|
|
|
1578 if (n >= maxlen)
|
|
|
1579 n = maxlen - 1;
|
|
|
1580 vim_memset(buf + 1, ' ', n);
|
|
|
1581 }
|
|
236
|
1582 }
|
|
|
1583
|
|
|
1584 /*
|
|
307
|
1585 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1586 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1587 */
|
|
307
|
1588 static void
|
|
236
|
1589 spell_load_lang(lang)
|
|
|
1590 char_u *lang;
|
|
|
1591 {
|
|
310
|
1592 char_u fname_enc[85];
|
|
236
|
1593 int r;
|
|
307
|
1594 char_u langcp[MAXWLEN + 1];
|
|
|
1595
|
|
310
|
1596 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1597 * It's truncated when an error is detected. */
|
|
|
1598 STRCPY(langcp, lang);
|
|
|
1599
|
|
310
|
1600 /*
|
|
|
1601 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1602 */
|
|
|
1603 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1604 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1605 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1606
|
|
|
1607 if (r == FAIL && *langcp != NUL)
|
|
|
1608 {
|
|
|
1609 /* Try loading the ASCII version. */
|
|
310
|
1610 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1611 "spell/%s.ascii.spl", lang);
|
|
310
|
1612 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1613 }
|
|
|
1614
|
|
|
1615 if (r == FAIL)
|
|
|
1616 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
|
236
|
1617 fname_enc + 6);
|
|
310
|
1618 else if (*langcp != NUL)
|
|
|
1619 {
|
|
|
1620 /* Load all the additions. */
|
|
|
1621 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1622 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1623 }
|
|
|
1624 }
|
|
|
1625
|
|
|
1626 /*
|
|
|
1627 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1628 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1629 */
|
|
|
1630 static char_u *
|
|
|
1631 spell_enc()
|
|
|
1632 {
|
|
|
1633
|
|
|
1634 #ifdef FEAT_MBYTE
|
|
|
1635 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1636 return p_enc;
|
|
|
1637 #endif
|
|
|
1638 return (char_u *)"latin1";
|
|
236
|
1639 }
|
|
|
1640
|
|
|
1641 /*
|
|
385
|
1642 * Get the name of the .spl file for the internal wordlist into
|
|
|
1643 * "fname[MAXPATHL]".
|
|
|
1644 */
|
|
|
1645 static void
|
|
|
1646 int_wordlist_spl(fname)
|
|
|
1647 char_u *fname;
|
|
|
1648 {
|
|
|
1649 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
1650 int_wordlist, spell_enc());
|
|
|
1651 }
|
|
|
1652
|
|
|
1653 /*
|
|
236
|
1654 * Allocate a new slang_T.
|
|
|
1655 * Caller must fill "sl_next".
|
|
|
1656 */
|
|
|
1657 static slang_T *
|
|
|
1658 slang_alloc(lang)
|
|
|
1659 char_u *lang;
|
|
|
1660 {
|
|
|
1661 slang_T *lp;
|
|
|
1662
|
|
300
|
1663 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1664 if (lp != NULL)
|
|
|
1665 {
|
|
|
1666 lp->sl_name = vim_strsave(lang);
|
|
323
|
1667 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
236
|
1668 }
|
|
|
1669 return lp;
|
|
|
1670 }
|
|
|
1671
|
|
|
1672 /*
|
|
|
1673 * Free the contents of an slang_T and the structure itself.
|
|
|
1674 */
|
|
|
1675 static void
|
|
|
1676 slang_free(lp)
|
|
|
1677 slang_T *lp;
|
|
|
1678 {
|
|
|
1679 vim_free(lp->sl_name);
|
|
310
|
1680 vim_free(lp->sl_fname);
|
|
|
1681 slang_clear(lp);
|
|
|
1682 vim_free(lp);
|
|
|
1683 }
|
|
|
1684
|
|
|
1685 /*
|
|
|
1686 * Clear an slang_T so that the file can be reloaded.
|
|
|
1687 */
|
|
|
1688 static void
|
|
|
1689 slang_clear(lp)
|
|
|
1690 slang_T *lp;
|
|
|
1691 {
|
|
339
|
1692 garray_T *gap;
|
|
|
1693 fromto_T *ftp;
|
|
344
|
1694 salitem_T *smp;
|
|
339
|
1695 int i;
|
|
323
|
1696
|
|
300
|
1697 vim_free(lp->sl_fbyts);
|
|
310
|
1698 lp->sl_fbyts = NULL;
|
|
300
|
1699 vim_free(lp->sl_kbyts);
|
|
310
|
1700 lp->sl_kbyts = NULL;
|
|
339
|
1701 vim_free(lp->sl_pbyts);
|
|
|
1702 lp->sl_pbyts = NULL;
|
|
|
1703
|
|
300
|
1704 vim_free(lp->sl_fidxs);
|
|
310
|
1705 lp->sl_fidxs = NULL;
|
|
300
|
1706 vim_free(lp->sl_kidxs);
|
|
310
|
1707 lp->sl_kidxs = NULL;
|
|
339
|
1708 vim_free(lp->sl_pidxs);
|
|
|
1709 lp->sl_pidxs = NULL;
|
|
323
|
1710
|
|
344
|
1711 gap = &lp->sl_rep;
|
|
|
1712 while (gap->ga_len > 0)
|
|
323
|
1713 {
|
|
344
|
1714 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1715 vim_free(ftp->ft_from);
|
|
|
1716 vim_free(ftp->ft_to);
|
|
323
|
1717 }
|
|
344
|
1718 ga_clear(gap);
|
|
|
1719
|
|
|
1720 gap = &lp->sl_sal;
|
|
375
|
1721 if (lp->sl_sofo)
|
|
376
|
1722 {
|
|
|
1723 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1724 if (gap->ga_data != NULL)
|
|
|
1725 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1726 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1727 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1728 }
|
|
375
|
1729 else
|
|
|
1730 /* SAL items: free salitem_T items */
|
|
|
1731 while (gap->ga_len > 0)
|
|
|
1732 {
|
|
|
1733 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1734 vim_free(smp->sm_lead);
|
|
|
1735 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1736 vim_free(smp->sm_to);
|
|
|
1737 #ifdef FEAT_MBYTE
|
|
|
1738 vim_free(smp->sm_lead_w);
|
|
|
1739 vim_free(smp->sm_oneof_w);
|
|
|
1740 vim_free(smp->sm_to_w);
|
|
|
1741 #endif
|
|
|
1742 }
|
|
344
|
1743 ga_clear(gap);
|
|
323
|
1744
|
|
339
|
1745 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1746 vim_free(lp->sl_prefprog[i]);
|
|
376
|
1747 lp->sl_prefixcnt = 0;
|
|
339
|
1748 vim_free(lp->sl_prefprog);
|
|
376
|
1749 lp->sl_prefprog = NULL;
|
|
|
1750
|
|
|
1751 vim_free(lp->sl_midword);
|
|
|
1752 lp->sl_midword = NULL;
|
|
339
|
1753
|
|
330
|
1754 #ifdef FEAT_MBYTE
|
|
|
1755 {
|
|
|
1756 int todo = lp->sl_map_hash.ht_used;
|
|
|
1757 hashitem_T *hi;
|
|
|
1758
|
|
|
1759 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1760 if (!HASHITEM_EMPTY(hi))
|
|
|
1761 {
|
|
|
1762 --todo;
|
|
|
1763 vim_free(hi->hi_key);
|
|
|
1764 }
|
|
|
1765 }
|
|
|
1766 hash_clear(&lp->sl_map_hash);
|
|
|
1767 #endif
|
|
236
|
1768 }
|
|
|
1769
|
|
|
1770 /*
|
|
307
|
1771 * Load one spell file and store the info into a slang_T.
|
|
236
|
1772 * Invoked through do_in_runtimepath().
|
|
|
1773 */
|
|
|
1774 static void
|
|
310
|
1775 spell_load_cb(fname, cookie)
|
|
236
|
1776 char_u *fname;
|
|
307
|
1777 void *cookie; /* points to the language name */
|
|
236
|
1778 {
|
|
323
|
1779 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1780 }
|
|
|
1781
|
|
|
1782 /*
|
|
|
1783 * Load one spell file and store the info into a slang_T.
|
|
|
1784 *
|
|
|
1785 * This is invoked in two ways:
|
|
|
1786 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
1787 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
1788 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
1789 * points to the existing slang_T.
|
|
323
|
1790 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
1791 */
|
|
323
|
1792 static slang_T *
|
|
|
1793 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
1794 char_u *fname;
|
|
|
1795 char_u *lang;
|
|
|
1796 slang_T *old_lp;
|
|
323
|
1797 int silent; /* no error if file doesn't exist */
|
|
310
|
1798 {
|
|
236
|
1799 FILE *fd;
|
|
|
1800 char_u buf[MAXWLEN + 1];
|
|
|
1801 char_u *p;
|
|
339
|
1802 char_u *bp;
|
|
|
1803 idx_T *ip;
|
|
236
|
1804 int i;
|
|
339
|
1805 int n;
|
|
300
|
1806 int len;
|
|
236
|
1807 int round;
|
|
|
1808 char_u *save_sourcing_name = sourcing_name;
|
|
|
1809 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
255
|
1810 int cnt, ccnt;
|
|
|
1811 char_u *fol;
|
|
307
|
1812 slang_T *lp = NULL;
|
|
323
|
1813 garray_T *gap;
|
|
|
1814 fromto_T *ftp;
|
|
344
|
1815 salitem_T *smp;
|
|
323
|
1816 short *first;
|
|
324
|
1817 idx_T idx;
|
|
344
|
1818 int c = 0;
|
|
236
|
1819
|
|
310
|
1820 fd = mch_fopen((char *)fname, "r");
|
|
236
|
1821 if (fd == NULL)
|
|
|
1822 {
|
|
323
|
1823 if (!silent)
|
|
|
1824 EMSG2(_(e_notopen), fname);
|
|
|
1825 else if (p_verbose > 2)
|
|
|
1826 {
|
|
|
1827 verbose_enter();
|
|
|
1828 smsg((char_u *)e_notopen, fname);
|
|
|
1829 verbose_leave();
|
|
|
1830 }
|
|
255
|
1831 goto endFAIL;
|
|
236
|
1832 }
|
|
310
|
1833 if (p_verbose > 2)
|
|
|
1834 {
|
|
|
1835 verbose_enter();
|
|
|
1836 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
1837 verbose_leave();
|
|
|
1838 }
|
|
|
1839
|
|
|
1840 if (old_lp == NULL)
|
|
|
1841 {
|
|
|
1842 lp = slang_alloc(lang);
|
|
|
1843 if (lp == NULL)
|
|
|
1844 goto endFAIL;
|
|
|
1845
|
|
|
1846 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
1847 lp->sl_fname = vim_strsave(fname);
|
|
|
1848 if (lp->sl_fname == NULL)
|
|
|
1849 goto endFAIL;
|
|
|
1850
|
|
|
1851 /* Check for .add.spl. */
|
|
|
1852 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
1853 }
|
|
|
1854 else
|
|
|
1855 lp = old_lp;
|
|
307
|
1856
|
|
236
|
1857 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
1858 sourcing_name = fname;
|
|
|
1859 sourcing_lnum = 0;
|
|
|
1860
|
|
339
|
1861 /* <HEADER>: <fileID>
|
|
|
1862 * <regioncnt> <regionname> ...
|
|
|
1863 * <charflagslen> <charflags>
|
|
|
1864 * <fcharslen> <fchars>
|
|
366
|
1865 * <midwordlen> <midword>
|
|
339
|
1866 * <prefcondcnt> <prefcond> ...
|
|
|
1867 */
|
|
236
|
1868 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
1869 buf[i] = getc(fd); /* <fileID> */
|
|
|
1870 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
1871 {
|
|
|
1872 EMSG(_("E757: Wrong file ID in spell file"));
|
|
255
|
1873 goto endFAIL;
|
|
236
|
1874 }
|
|
|
1875
|
|
|
1876 cnt = getc(fd); /* <regioncnt> */
|
|
255
|
1877 if (cnt < 0)
|
|
236
|
1878 {
|
|
|
1879 truncerr:
|
|
381
|
1880 EMSG(_(e_spell_trunc));
|
|
255
|
1881 goto endFAIL;
|
|
236
|
1882 }
|
|
|
1883 if (cnt > 8)
|
|
|
1884 {
|
|
|
1885 formerr:
|
|
307
|
1886 EMSG(_(e_format));
|
|
255
|
1887 goto endFAIL;
|
|
236
|
1888 }
|
|
|
1889 for (i = 0; i < cnt; ++i)
|
|
|
1890 {
|
|
|
1891 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
|
1892 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
|
1893 }
|
|
|
1894 lp->sl_regions[cnt * 2] = NUL;
|
|
|
1895
|
|
381
|
1896 /* <charflagslen> <charflags> */
|
|
|
1897 p = read_cnt_string(fd, 1, &cnt);
|
|
389
|
1898 if (cnt < 0)
|
|
381
|
1899 goto endFAIL;
|
|
|
1900
|
|
|
1901 /* <fcharslen> <fchars> */
|
|
389
|
1902 fol = read_cnt_string(fd, 2, &ccnt);
|
|
|
1903 if (ccnt < 0)
|
|
381
|
1904 {
|
|
300
|
1905 vim_free(p);
|
|
381
|
1906 goto endFAIL;
|
|
|
1907 }
|
|
|
1908
|
|
|
1909 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
1910 if (p != NULL && fol != NULL)
|
|
389
|
1911 i = set_spell_charflags(p, cnt, fol);
|
|
381
|
1912
|
|
|
1913 vim_free(p);
|
|
|
1914 vim_free(fol);
|
|
|
1915
|
|
|
1916 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
1917 if ((p == NULL) != (fol == NULL))
|
|
|
1918 goto formerr;
|
|
255
|
1919
|
|
366
|
1920 /* <midwordlen> <midword> */
|
|
376
|
1921 lp->sl_midword = read_cnt_string(fd, 2, &cnt);
|
|
389
|
1922 if (cnt < 0)
|
|
376
|
1923 goto endFAIL;
|
|
366
|
1924
|
|
339
|
1925 /* <prefcondcnt> <prefcond> ... */
|
|
|
1926 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
1927 if (cnt > 0)
|
|
|
1928 {
|
|
|
1929 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
1930 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
1931 if (lp->sl_prefprog == NULL)
|
|
|
1932 goto endFAIL;
|
|
|
1933 lp->sl_prefixcnt = cnt;
|
|
|
1934
|
|
|
1935 for (i = 0; i < cnt; ++i)
|
|
|
1936 {
|
|
|
1937 /* <prefcond> : <condlen> <condstr> */
|
|
|
1938 n = getc(fd); /* <condlen> */
|
|
|
1939 if (n < 0)
|
|
|
1940 goto formerr;
|
|
|
1941 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
1942 * compile the regexp program used to check for the condition. */
|
|
|
1943 if (n > 0)
|
|
|
1944 {
|
|
344
|
1945 buf[0] = '^'; /* always match at one position only */
|
|
|
1946 p = buf + 1;
|
|
339
|
1947 while (n-- > 0)
|
|
|
1948 *p++ = getc(fd); /* <condstr> */
|
|
|
1949 *p = NUL;
|
|
|
1950 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
1951 }
|
|
|
1952 }
|
|
|
1953 }
|
|
|
1954
|
|
|
1955
|
|
323
|
1956 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
1957 * <salflags> <salcount> <sal> ...
|
|
|
1958 * <maplen> <mapstr> */
|
|
339
|
1959
|
|
344
|
1960 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
1961 if (cnt < 0)
|
|
|
1962 goto formerr;
|
|
|
1963
|
|
|
1964 gap = &lp->sl_rep;
|
|
|
1965 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1966 goto endFAIL;
|
|
|
1967
|
|
|
1968 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
1969 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
1970 {
|
|
|
1971 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
381
|
1972 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
389
|
1973 if (i <= 0)
|
|
381
|
1974 goto endFAIL;
|
|
|
1975 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
389
|
1976 if (i <= 0)
|
|
381
|
1977 {
|
|
|
1978 vim_free(ftp->ft_from);
|
|
|
1979 goto endFAIL;
|
|
344
|
1980 }
|
|
|
1981 }
|
|
|
1982
|
|
|
1983 /* Fill the first-index table. */
|
|
|
1984 first = lp->sl_rep_first;
|
|
|
1985 for (i = 0; i < 256; ++i)
|
|
|
1986 first[i] = -1;
|
|
|
1987 for (i = 0; i < gap->ga_len; ++i)
|
|
323
|
1988 {
|
|
344
|
1989 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
1990 if (first[*ftp->ft_from] == -1)
|
|
|
1991 first[*ftp->ft_from] = i;
|
|
|
1992 }
|
|
|
1993
|
|
|
1994 i = getc(fd); /* <salflags> */
|
|
|
1995 if (i & SAL_F0LLOWUP)
|
|
|
1996 lp->sl_followup = TRUE;
|
|
|
1997 if (i & SAL_COLLAPSE)
|
|
|
1998 lp->sl_collapse = TRUE;
|
|
|
1999 if (i & SAL_REM_ACCENTS)
|
|
|
2000 lp->sl_rem_accents = TRUE;
|
|
375
|
2001 if (i & SAL_SOFO)
|
|
|
2002 lp->sl_sofo = TRUE;
|
|
389
|
2003 else
|
|
|
2004 lp->sl_sofo = FALSE;
|
|
344
|
2005
|
|
|
2006 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2007 if (cnt < 0)
|
|
|
2008 goto formerr;
|
|
|
2009
|
|
375
|
2010 if (lp->sl_sofo)
|
|
|
2011 {
|
|
|
2012 /*
|
|
|
2013 * SOFOFROM and SOFOTO items come in one <salfrom> and <salto>
|
|
|
2014 */
|
|
|
2015 if (cnt != 1)
|
|
|
2016 goto formerr;
|
|
|
2017
|
|
381
|
2018 /* <salfromlen> <salfrom> */
|
|
|
2019 bp = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2020 if (cnt < 0)
|
|
375
|
2021 goto endFAIL;
|
|
381
|
2022
|
|
|
2023 /* <saltolen> <salto> */
|
|
|
2024 fol = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2025 if (cnt < 0)
|
|
375
|
2026 {
|
|
|
2027 vim_free(bp);
|
|
344
|
2028 goto endFAIL;
|
|
375
|
2029 }
|
|
381
|
2030
|
|
|
2031 /* Store the info in lp->sl_sal and/or lp->sl_sal_first. */
|
|
389
|
2032 if (bp != NULL && fol != NULL)
|
|
|
2033 i = set_sofo(lp, bp, fol);
|
|
|
2034 else if (bp != NULL || fol != NULL)
|
|
|
2035 i = FAIL; /* only one of two strings is an error */
|
|
|
2036 else
|
|
|
2037 i = OK;
|
|
381
|
2038
|
|
375
|
2039 vim_free(bp);
|
|
|
2040 vim_free(fol);
|
|
381
|
2041 if (i == FAIL)
|
|
|
2042 goto formerr;
|
|
375
|
2043 }
|
|
|
2044 else
|
|
|
2045 {
|
|
|
2046 /*
|
|
|
2047 * SAL items
|
|
|
2048 */
|
|
|
2049 gap = &lp->sl_sal;
|
|
390
|
2050 ga_init2(gap, sizeof(salitem_T), 10);
|
|
375
|
2051 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2052 goto endFAIL;
|
|
|
2053
|
|
|
2054 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2055 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2056 {
|
|
|
2057 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2058 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2059 if (ccnt < 0)
|
|
|
2060 goto formerr;
|
|
|
2061 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2062 goto endFAIL;
|
|
|
2063 smp->sm_lead = p;
|
|
|
2064
|
|
|
2065 /* Read up to the first special char into sm_lead. */
|
|
|
2066 for (i = 0; i < ccnt; ++i)
|
|
|
2067 {
|
|
|
2068 c = getc(fd); /* <salfrom> */
|
|
|
2069 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2070 break;
|
|
|
2071 *p++ = c;
|
|
|
2072 }
|
|
|
2073 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2074 *p++ = NUL;
|
|
|
2075
|
|
|
2076 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2077 if (c == '(')
|
|
|
2078 {
|
|
|
2079 smp->sm_oneof = p;
|
|
|
2080 for (++i; i < ccnt; ++i)
|
|
|
2081 {
|
|
|
2082 c = getc(fd); /* <salfrom> */
|
|
|
2083 if (c == ')')
|
|
|
2084 break;
|
|
|
2085 *p++ = c;
|
|
|
2086 }
|
|
|
2087 *p++ = NUL;
|
|
|
2088 if (++i < ccnt)
|
|
|
2089 c = getc(fd);
|
|
|
2090 }
|
|
372
|
2091 else
|
|
375
|
2092 smp->sm_oneof = NULL;
|
|
|
2093
|
|
|
2094 /* Any following chars go in sm_rules. */
|
|
|
2095 smp->sm_rules = p;
|
|
|
2096 if (i < ccnt)
|
|
|
2097 /* store the char we got while checking for end of sm_lead */
|
|
|
2098 *p++ = c;
|
|
|
2099 for (++i; i < ccnt; ++i)
|
|
|
2100 *p++ = getc(fd); /* <salfrom> */
|
|
|
2101 *p++ = NUL;
|
|
|
2102
|
|
381
|
2103 /* <saltolen> <salto> */
|
|
|
2104 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
389
|
2105 if (ccnt < 0)
|
|
372
|
2106 {
|
|
|
2107 vim_free(smp->sm_lead);
|
|
375
|
2108 goto formerr;
|
|
|
2109 }
|
|
|
2110
|
|
372
|
2111 #ifdef FEAT_MBYTE
|
|
|
2112 if (has_mbyte)
|
|
|
2113 {
|
|
375
|
2114 /* convert the multi-byte strings to wide char strings */
|
|
|
2115 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2116 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2117 if (smp->sm_oneof == NULL)
|
|
|
2118 smp->sm_oneof_w = NULL;
|
|
|
2119 else
|
|
|
2120 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
389
|
2121 if (smp->sm_to == NULL)
|
|
|
2122 smp->sm_to_w = NULL;
|
|
|
2123 else
|
|
|
2124 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
375
|
2125 if (smp->sm_lead_w == NULL
|
|
|
2126 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
389
|
2127 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
372
|
2128 {
|
|
375
|
2129 vim_free(smp->sm_lead);
|
|
|
2130 vim_free(smp->sm_to);
|
|
|
2131 vim_free(smp->sm_lead_w);
|
|
|
2132 vim_free(smp->sm_oneof_w);
|
|
|
2133 vim_free(smp->sm_to_w);
|
|
|
2134 goto endFAIL;
|
|
372
|
2135 }
|
|
|
2136 }
|
|
|
2137 #endif
|
|
|
2138 }
|
|
375
|
2139
|
|
|
2140 /* Fill the first-index table. */
|
|
381
|
2141 set_sal_first(lp);
|
|
|
2142 }
|
|
|
2143
|
|
|
2144 /* <maplen> <mapstr> */
|
|
|
2145 p = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2146 if (cnt < 0)
|
|
323
|
2147 goto endFAIL;
|
|
389
|
2148 if (p != NULL)
|
|
|
2149 {
|
|
|
2150 set_map_str(lp, p);
|
|
|
2151 vim_free(p);
|
|
|
2152 }
|
|
236
|
2153
|
|
300
|
2154 /* round 1: <LWORDTREE>
|
|
339
|
2155 * round 2: <KWORDTREE>
|
|
|
2156 * round 3: <PREFIXTREE> */
|
|
|
2157 for (round = 1; round <= 3; ++round)
|
|
236
|
2158 {
|
|
300
|
2159 /* The tree size was computed when writing the file, so that we can
|
|
|
2160 * allocate it as one long block. <nodecount> */
|
|
|
2161 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2162 if (len < 0)
|
|
|
2163 goto truncerr;
|
|
|
2164 if (len > 0)
|
|
236
|
2165 {
|
|
300
|
2166 /* Allocate the byte array. */
|
|
339
|
2167 bp = lalloc((long_u)len, TRUE);
|
|
|
2168 if (bp == NULL)
|
|
300
|
2169 goto endFAIL;
|
|
|
2170 if (round == 1)
|
|
339
|
2171 lp->sl_fbyts = bp;
|
|
|
2172 else if (round == 2)
|
|
|
2173 lp->sl_kbyts = bp;
|
|
300
|
2174 else
|
|
339
|
2175 lp->sl_pbyts = bp;
|
|
236
|
2176
|
|
300
|
2177 /* Allocate the index array. */
|
|
339
|
2178 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2179 if (ip == NULL)
|
|
300
|
2180 goto endFAIL;
|
|
|
2181 if (round == 1)
|
|
339
|
2182 lp->sl_fidxs = ip;
|
|
|
2183 else if (round == 2)
|
|
|
2184 lp->sl_kidxs = ip;
|
|
300
|
2185 else
|
|
339
|
2186 lp->sl_pidxs = ip;
|
|
300
|
2187
|
|
|
2188 /* Read the tree and store it in the array. */
|
|
339
|
2189 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2190 if (idx == -1)
|
|
300
|
2191 goto truncerr;
|
|
324
|
2192 if (idx < 0)
|
|
236
|
2193 goto formerr;
|
|
|
2194 }
|
|
300
|
2195 }
|
|
243
|
2196
|
|
310
|
2197 /* For a new file link it in the list of spell files. */
|
|
|
2198 if (old_lp == NULL)
|
|
|
2199 {
|
|
|
2200 lp->sl_next = first_lang;
|
|
|
2201 first_lang = lp;
|
|
|
2202 }
|
|
307
|
2203
|
|
255
|
2204 goto endOK;
|
|
|
2205
|
|
|
2206 endFAIL:
|
|
310
|
2207 if (lang != NULL)
|
|
|
2208 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2209 *lang = NUL;
|
|
|
2210 if (lp != NULL && old_lp == NULL)
|
|
323
|
2211 {
|
|
307
|
2212 slang_free(lp);
|
|
323
|
2213 lp = NULL;
|
|
|
2214 }
|
|
255
|
2215
|
|
|
2216 endOK:
|
|
236
|
2217 if (fd != NULL)
|
|
|
2218 fclose(fd);
|
|
|
2219 sourcing_name = save_sourcing_name;
|
|
|
2220 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2221
|
|
|
2222 return lp;
|
|
236
|
2223 }
|
|
|
2224
|
|
376
|
2225 /*
|
|
|
2226 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2227 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2228 * Returns NULL when the count is zero.
|
|
389
|
2229 * Sets "*cntp" to -1 when there is an error, length of the result otherwise.
|
|
376
|
2230 */
|
|
|
2231 static char_u *
|
|
389
|
2232 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2233 FILE *fd;
|
|
|
2234 int cnt_bytes;
|
|
389
|
2235 int *cntp;
|
|
376
|
2236 {
|
|
|
2237 int cnt = 0;
|
|
|
2238 int i;
|
|
|
2239 char_u *str;
|
|
|
2240
|
|
|
2241 /* read the length bytes, MSB first */
|
|
|
2242 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2243 cnt = (cnt << 8) + getc(fd);
|
|
|
2244 if (cnt < 0)
|
|
|
2245 {
|
|
381
|
2246 EMSG(_(e_spell_trunc));
|
|
389
|
2247 *cntp = -1;
|
|
376
|
2248 return NULL;
|
|
|
2249 }
|
|
389
|
2250 *cntp = cnt;
|
|
|
2251 if (cnt == 0)
|
|
|
2252 return NULL; /* nothing to read, return NULL */
|
|
376
|
2253
|
|
|
2254 /* allocate memory */
|
|
|
2255 str = alloc((unsigned)cnt + 1);
|
|
|
2256 if (str == NULL)
|
|
|
2257 {
|
|
389
|
2258 *cntp = -1;
|
|
376
|
2259 return NULL;
|
|
|
2260 }
|
|
|
2261
|
|
|
2262 /* Read the string. Doesn't check for truncated file. */
|
|
|
2263 for (i = 0; i < cnt; ++i)
|
|
|
2264 str[i] = getc(fd);
|
|
|
2265 str[i] = NUL;
|
|
|
2266
|
|
|
2267 return str;
|
|
|
2268 }
|
|
|
2269
|
|
381
|
2270 /*
|
|
|
2271 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
|
2272 * Returns FAIL when there is something wrong.
|
|
|
2273 */
|
|
|
2274 static int
|
|
|
2275 set_sofo(lp, from, to)
|
|
|
2276 slang_T *lp;
|
|
|
2277 char_u *from;
|
|
|
2278 char_u *to;
|
|
|
2279 {
|
|
|
2280 int i;
|
|
|
2281
|
|
|
2282 #ifdef FEAT_MBYTE
|
|
|
2283 garray_T *gap;
|
|
|
2284 char_u *s;
|
|
|
2285 char_u *p;
|
|
|
2286 int c;
|
|
|
2287 int *inp;
|
|
|
2288
|
|
|
2289 if (has_mbyte)
|
|
|
2290 {
|
|
|
2291 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2292 * characters. The index is the low byte of the character.
|
|
|
2293 * The list contains from-to pairs with a terminating NUL.
|
|
|
2294 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2295 gap = &lp->sl_sal;
|
|
|
2296 ga_init2(gap, sizeof(int *), 1);
|
|
|
2297 if (ga_grow(gap, 256) == FAIL)
|
|
|
2298 return FAIL;
|
|
|
2299 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2300 gap->ga_len = 256;
|
|
|
2301
|
|
|
2302 /* First count the number of items for each list. Temporarily use
|
|
|
2303 * sl_sal_first[] for this. */
|
|
|
2304 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2305 {
|
|
|
2306 c = mb_ptr2char_adv(&p);
|
|
|
2307 mb_ptr_adv(s);
|
|
|
2308 if (c >= 256)
|
|
|
2309 ++lp->sl_sal_first[c & 0xff];
|
|
|
2310 }
|
|
|
2311 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
|
2312 return FAIL;
|
|
|
2313
|
|
|
2314 /* Allocate the lists. */
|
|
|
2315 for (i = 0; i < 256; ++i)
|
|
|
2316 if (lp->sl_sal_first[i] > 0)
|
|
|
2317 {
|
|
|
2318 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2319 if (p == NULL)
|
|
|
2320 return FAIL;
|
|
|
2321 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2322 *(int *)p = 0;
|
|
|
2323 }
|
|
|
2324
|
|
|
2325 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
2326 * list. */
|
|
|
2327 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2328 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2329 {
|
|
|
2330 c = mb_ptr2char_adv(&p);
|
|
|
2331 i = mb_ptr2char_adv(&s);
|
|
|
2332 if (c >= 256)
|
|
|
2333 {
|
|
|
2334 /* Append the from-to chars at the end of the list with
|
|
|
2335 * the low byte. */
|
|
|
2336 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2337 while (*inp != 0)
|
|
|
2338 ++inp;
|
|
|
2339 *inp++ = c; /* from char */
|
|
|
2340 *inp++ = i; /* to char */
|
|
|
2341 *inp++ = NUL; /* NUL at the end */
|
|
|
2342 }
|
|
|
2343 else
|
|
|
2344 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2345 lp->sl_sal_first[c] = i;
|
|
|
2346 }
|
|
|
2347 }
|
|
|
2348 else
|
|
|
2349 #endif
|
|
|
2350 {
|
|
|
2351 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2352 if (STRLEN(from) != STRLEN(to))
|
|
|
2353 return FAIL;
|
|
|
2354
|
|
|
2355 for (i = 0; to[i] != NUL; ++i)
|
|
|
2356 lp->sl_sal_first[from[i]] = to[i];
|
|
|
2357 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2358 }
|
|
|
2359
|
|
|
2360 return OK;
|
|
|
2361 }
|
|
|
2362
|
|
|
2363 /*
|
|
|
2364 * Fill the first-index table for "lp".
|
|
|
2365 */
|
|
|
2366 static void
|
|
|
2367 set_sal_first(lp)
|
|
|
2368 slang_T *lp;
|
|
|
2369 {
|
|
|
2370 salfirst_T *sfirst;
|
|
|
2371 int i;
|
|
|
2372 salitem_T *smp;
|
|
|
2373 int c;
|
|
|
2374 garray_T *gap = &lp->sl_sal;
|
|
|
2375
|
|
|
2376 sfirst = lp->sl_sal_first;
|
|
|
2377 for (i = 0; i < 256; ++i)
|
|
|
2378 sfirst[i] = -1;
|
|
|
2379 smp = (salitem_T *)gap->ga_data;
|
|
|
2380 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2381 {
|
|
|
2382 #ifdef FEAT_MBYTE
|
|
|
2383 if (has_mbyte)
|
|
|
2384 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2385 * the character, for other encodings it should differ for most
|
|
|
2386 * characters. */
|
|
|
2387 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2388 else
|
|
|
2389 #endif
|
|
|
2390 c = *smp[i].sm_lead;
|
|
|
2391 if (sfirst[c] == -1)
|
|
|
2392 {
|
|
|
2393 sfirst[c] = i;
|
|
|
2394 #ifdef FEAT_MBYTE
|
|
|
2395 if (has_mbyte)
|
|
|
2396 {
|
|
|
2397 int n;
|
|
|
2398
|
|
|
2399 /* Make sure all entries with this byte are following each
|
|
|
2400 * other. Move the ones that are in the wrong position. Do
|
|
|
2401 * keep the same ordering! */
|
|
|
2402 while (i + 1 < gap->ga_len
|
|
|
2403 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
2404 /* Skip over entry with same index byte. */
|
|
|
2405 ++i;
|
|
|
2406
|
|
|
2407 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
2408 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
2409 {
|
|
|
2410 salitem_T tsal;
|
|
|
2411
|
|
|
2412 /* Move entry with same index byte after the entries
|
|
|
2413 * we already found. */
|
|
|
2414 ++i;
|
|
|
2415 --n;
|
|
|
2416 tsal = smp[i + n];
|
|
|
2417 mch_memmove(smp + i + 1, smp + i,
|
|
|
2418 sizeof(salitem_T) * n);
|
|
|
2419 smp[i] = tsal;
|
|
|
2420 }
|
|
|
2421 }
|
|
|
2422 #endif
|
|
|
2423 }
|
|
|
2424 }
|
|
|
2425 }
|
|
376
|
2426
|
|
372
|
2427 #ifdef FEAT_MBYTE
|
|
|
2428 /*
|
|
|
2429 * Turn a multi-byte string into a wide character string.
|
|
|
2430 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
2431 */
|
|
|
2432 static int *
|
|
|
2433 mb_str2wide(s)
|
|
|
2434 char_u *s;
|
|
|
2435 {
|
|
|
2436 int *res;
|
|
|
2437 char_u *p;
|
|
|
2438 int i = 0;
|
|
|
2439
|
|
|
2440 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
2441 if (res != NULL)
|
|
|
2442 {
|
|
|
2443 for (p = s; *p != NUL; )
|
|
|
2444 res[i++] = mb_ptr2char_adv(&p);
|
|
|
2445 res[i] = NUL;
|
|
|
2446 }
|
|
|
2447 return res;
|
|
|
2448 }
|
|
|
2449 #endif
|
|
|
2450
|
|
236
|
2451 /*
|
|
300
|
2452 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
2453 * "byts" and index array "idxs". Recursively read the children.
|
|
|
2454 *
|
|
346
|
2455 * NOTE: The code here must match put_node().
|
|
300
|
2456 *
|
|
|
2457 * Returns the index follosing the siblings.
|
|
|
2458 * Returns -1 if the file is shorter than expected.
|
|
|
2459 * Returns -2 if there is a format error.
|
|
236
|
2460 */
|
|
324
|
2461 static idx_T
|
|
339
|
2462 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
2463 FILE *fd;
|
|
|
2464 char_u *byts;
|
|
324
|
2465 idx_T *idxs;
|
|
300
|
2466 int maxidx; /* size of arrays */
|
|
324
|
2467 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
2468 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
2469 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
2470 {
|
|
300
|
2471 int len;
|
|
|
2472 int i;
|
|
|
2473 int n;
|
|
324
|
2474 idx_T idx = startidx;
|
|
300
|
2475 int c;
|
|
366
|
2476 int c2;
|
|
300
|
2477 #define SHARED_MASK 0x8000000
|
|
236
|
2478
|
|
300
|
2479 len = getc(fd); /* <siblingcount> */
|
|
|
2480 if (len <= 0)
|
|
|
2481 return -1;
|
|
|
2482
|
|
|
2483 if (startidx + len >= maxidx)
|
|
|
2484 return -2;
|
|
|
2485 byts[idx++] = len;
|
|
|
2486
|
|
|
2487 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
2488 for (i = 1; i <= len; ++i)
|
|
236
|
2489 {
|
|
300
|
2490 c = getc(fd); /* <byte> */
|
|
|
2491 if (c < 0)
|
|
|
2492 return -1;
|
|
|
2493 if (c <= BY_SPECIAL)
|
|
|
2494 {
|
|
366
|
2495 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
2496 {
|
|
|
2497 /* No flags, all regions. */
|
|
|
2498 idxs[idx] = 0;
|
|
|
2499 c = 0;
|
|
|
2500 }
|
|
390
|
2501 else if (c != BY_INDEX)
|
|
300
|
2502 {
|
|
339
|
2503 if (prefixtree)
|
|
|
2504 {
|
|
455
|
2505 /* Read the optional pflags byte, the prefix ID and the
|
|
|
2506 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
2507 * byte, the condition index shifted up 8 bits, the flags
|
|
|
2508 * shifted up 24 bits. */
|
|
|
2509 if (c == BY_FLAGS)
|
|
|
2510 c = getc(fd) << 24; /* <pflags> */
|
|
|
2511 else
|
|
|
2512 c = 0;
|
|
|
2513
|
|
|
2514 c |= getc(fd); /* <prefixID> */
|
|
|
2515
|
|
339
|
2516 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
2517 if (n >= maxprefcondnr)
|
|
|
2518 return -2;
|
|
455
|
2519 c |= (n << 8);
|
|
339
|
2520 }
|
|
390
|
2521 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
2522 {
|
|
|
2523 /* Read flags and optional region and prefix ID. In
|
|
390
|
2524 * idxs[] the flags go in the low two bytes, region above
|
|
|
2525 * that and prefix ID above the region. */
|
|
|
2526 c2 = c;
|
|
339
|
2527 c = getc(fd); /* <flags> */
|
|
390
|
2528 if (c2 == BY_FLAGS2)
|
|
|
2529 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
2530 if (c & WF_REGION)
|
|
390
|
2531 c = (getc(fd) << 16) + c; /* <region> */
|
|
339
|
2532 if (c & WF_PFX)
|
|
390
|
2533 c = (getc(fd) << 24) + c; /* <prefixID> */
|
|
339
|
2534 }
|
|
|
2535
|
|
300
|
2536 idxs[idx] = c;
|
|
|
2537 c = 0;
|
|
|
2538 }
|
|
|
2539 else /* c == BY_INDEX */
|
|
|
2540 {
|
|
|
2541 /* <nodeidx> */
|
|
|
2542 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2543 if (n < 0 || n >= maxidx)
|
|
|
2544 return -2;
|
|
|
2545 idxs[idx] = n + SHARED_MASK;
|
|
|
2546 c = getc(fd); /* <xbyte> */
|
|
|
2547 }
|
|
|
2548 }
|
|
|
2549 byts[idx++] = c;
|
|
236
|
2550 }
|
|
|
2551
|
|
300
|
2552 /* Recursively read the children for non-shared siblings.
|
|
|
2553 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
2554 * remove SHARED_MASK) */
|
|
|
2555 for (i = 1; i <= len; ++i)
|
|
|
2556 if (byts[startidx + i] != 0)
|
|
|
2557 {
|
|
|
2558 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
2559 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
2560 else
|
|
|
2561 {
|
|
|
2562 idxs[startidx + i] = idx;
|
|
339
|
2563 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
2564 prefixtree, maxprefcondnr);
|
|
300
|
2565 if (idx < 0)
|
|
|
2566 break;
|
|
|
2567 }
|
|
|
2568 }
|
|
236
|
2569
|
|
300
|
2570 return idx;
|
|
236
|
2571 }
|
|
|
2572
|
|
|
2573 /*
|
|
|
2574 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
2575 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
2576 */
|
|
|
2577 char_u *
|
|
|
2578 did_set_spelllang(buf)
|
|
|
2579 buf_T *buf;
|
|
|
2580 {
|
|
|
2581 garray_T ga;
|
|
351
|
2582 char_u *splp;
|
|
236
|
2583 char_u *region;
|
|
416
|
2584 char_u region_cp[3];
|
|
355
|
2585 int filename;
|
|
236
|
2586 int region_mask;
|
|
|
2587 slang_T *lp;
|
|
|
2588 int c;
|
|
351
|
2589 char_u lang[MAXWLEN + 1];
|
|
323
|
2590 char_u spf_name[MAXPATHL];
|
|
351
|
2591 int len;
|
|
|
2592 char_u *p;
|
|
381
|
2593 int round;
|
|
385
|
2594 char_u *spf;
|
|
389
|
2595 char_u *use_region = NULL;
|
|
|
2596 int dont_use_region = FALSE;
|
|
236
|
2597
|
|
|
2598 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
2599 clear_midword(buf);
|
|
236
|
2600
|
|
351
|
2601 /* loop over comma separated language names. */
|
|
|
2602 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
2603 {
|
|
|
2604 /* Get one language name. */
|
|
|
2605 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
2606
|
|
240
|
2607 region = NULL;
|
|
351
|
2608 len = STRLEN(lang);
|
|
355
|
2609
|
|
|
2610 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
2611 * If there is a region name let "region" point to it and remove it
|
|
|
2612 * from the name. */
|
|
|
2613 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
2614 {
|
|
355
|
2615 filename = TRUE;
|
|
|
2616
|
|
416
|
2617 /* Locate a region and remove it from the file name. */
|
|
|
2618 p = vim_strchr(gettail(lang), '_');
|
|
|
2619 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
2620 && !ASCII_ISALPHA(p[3]))
|
|
|
2621 {
|
|
|
2622 vim_strncpy(region_cp, p + 1, 2);
|
|
|
2623 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
2624 len -= 3;
|
|
|
2625 region = region_cp;
|
|
|
2626 }
|
|
|
2627 else
|
|
|
2628 dont_use_region = TRUE;
|
|
|
2629
|
|
355
|
2630 /* Check if we loaded this language before. */
|
|
|
2631 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2632 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2633 break;
|
|
236
|
2634 }
|
|
355
|
2635 else
|
|
|
2636 {
|
|
|
2637 filename = FALSE;
|
|
|
2638 if (len > 3 && lang[len - 3] == '_')
|
|
|
2639 {
|
|
|
2640 region = lang + len - 2;
|
|
|
2641 len -= 3;
|
|
|
2642 lang[len] = NUL;
|
|
389
|
2643 }
|
|
|
2644 else
|
|
|
2645 dont_use_region = TRUE;
|
|
355
|
2646
|
|
|
2647 /* Check if we loaded this language before. */
|
|
|
2648 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2649 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
2650 break;
|
|
|
2651 }
|
|
236
|
2652
|
|
416
|
2653 if (region != NULL)
|
|
|
2654 {
|
|
|
2655 /* If the region differs from what was used before then don't
|
|
|
2656 * use it for 'spellfile'. */
|
|
|
2657 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
2658 dont_use_region = TRUE;
|
|
|
2659 use_region = region;
|
|
|
2660 }
|
|
|
2661
|
|
351
|
2662 /* If not found try loading the language now. */
|
|
236
|
2663 if (lp == NULL)
|
|
355
|
2664 {
|
|
|
2665 if (filename)
|
|
|
2666 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
2667 else
|
|
|
2668 spell_load_lang(lang);
|
|
|
2669 }
|
|
236
|
2670
|
|
307
|
2671 /*
|
|
351
|
2672 * Loop over the languages, there can be several files for "lang".
|
|
307
|
2673 */
|
|
|
2674 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
2675 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
2676 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
2677 {
|
|
316
|
2678 region_mask = REGION_ALL;
|
|
355
|
2679 if (!filename && region != NULL)
|
|
236
|
2680 {
|
|
307
|
2681 /* find region in sl_regions */
|
|
|
2682 c = find_region(lp->sl_regions, region);
|
|
|
2683 if (c == REGION_ALL)
|
|
|
2684 {
|
|
389
|
2685 if (lp->sl_add)
|
|
|
2686 {
|
|
|
2687 if (*lp->sl_regions != NUL)
|
|
|
2688 /* This addition file is for other regions. */
|
|
|
2689 region_mask = 0;
|
|
|
2690 }
|
|
|
2691 else
|
|
|
2692 /* This is probably an error. Give a warning and
|
|
|
2693 * accept the words anyway. */
|
|
351
|
2694 smsg((char_u *)
|
|
|
2695 _("Warning: region %s not supported"),
|
|
|
2696 region);
|
|
307
|
2697 }
|
|
|
2698 else
|
|
|
2699 region_mask = 1 << c;
|
|
236
|
2700 }
|
|
307
|
2701
|
|
389
|
2702 if (region_mask != 0)
|
|
307
|
2703 {
|
|
389
|
2704 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2705 {
|
|
|
2706 ga_clear(&ga);
|
|
|
2707 return e_outofmem;
|
|
|
2708 }
|
|
|
2709 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2710 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2711 ++ga.ga_len;
|
|
|
2712 use_midword(lp, buf);
|
|
307
|
2713 }
|
|
385
|
2714 }
|
|
|
2715 }
|
|
|
2716
|
|
|
2717 /* round 0: load int_wordlist, if possible.
|
|
|
2718 * round 1: load first name in 'spellfile'.
|
|
|
2719 * round 2: load second name in 'spellfile.
|
|
|
2720 * etc. */
|
|
|
2721 spf = curbuf->b_p_spf;
|
|
|
2722 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
2723 {
|
|
|
2724 if (round == 0)
|
|
|
2725 {
|
|
|
2726 /* Internal wordlist, if there is one. */
|
|
|
2727 if (int_wordlist == NULL)
|
|
381
|
2728 continue;
|
|
385
|
2729 int_wordlist_spl(spf_name);
|
|
381
|
2730 }
|
|
|
2731 else
|
|
|
2732 {
|
|
385
|
2733 /* One entry in 'spellfile'. */
|
|
|
2734 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
2735 STRCAT(spf_name, ".spl");
|
|
|
2736
|
|
|
2737 /* If it was already found above then skip it. */
|
|
|
2738 for (c = 0; c < ga.ga_len; ++c)
|
|
|
2739 if (fullpathcmp(spf_name,
|
|
|
2740 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
2741 FALSE) == FPC_SAME)
|
|
|
2742 break;
|
|
|
2743 if (c < ga.ga_len)
|
|
381
|
2744 continue;
|
|
|
2745 }
|
|
|
2746
|
|
351
|
2747 /* Check if it was loaded already. */
|
|
323
|
2748 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2749 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2750 break;
|
|
|
2751 if (lp == NULL)
|
|
|
2752 {
|
|
351
|
2753 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
2754 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
2755 * use an arbitrary name. */
|
|
|
2756 if (round == 0)
|
|
|
2757 STRCPY(lang, "internal wordlist");
|
|
|
2758 else
|
|
|
2759 {
|
|
|
2760 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
2761 p = vim_strchr(lang, '.');
|
|
|
2762 if (p != NULL)
|
|
|
2763 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
2764 }
|
|
351
|
2765 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
2766 }
|
|
|
2767 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
2768 {
|
|
389
|
2769 region_mask = REGION_ALL;
|
|
|
2770 if (use_region != NULL && !dont_use_region)
|
|
|
2771 {
|
|
|
2772 /* find region in sl_regions */
|
|
|
2773 c = find_region(lp->sl_regions, use_region);
|
|
|
2774 if (c != REGION_ALL)
|
|
|
2775 region_mask = 1 << c;
|
|
|
2776 else if (*lp->sl_regions != NUL)
|
|
|
2777 /* This spell file is for other regions. */
|
|
|
2778 region_mask = 0;
|
|
|
2779 }
|
|
|
2780
|
|
|
2781 if (region_mask != 0)
|
|
|
2782 {
|
|
|
2783 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2784 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2785 ++ga.ga_len;
|
|
|
2786 use_midword(lp, buf);
|
|
|
2787 }
|
|
323
|
2788 }
|
|
|
2789 }
|
|
|
2790
|
|
236
|
2791 /* Add a NULL entry to mark the end of the list. */
|
|
|
2792 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2793 {
|
|
|
2794 ga_clear(&ga);
|
|
|
2795 return e_outofmem;
|
|
|
2796 }
|
|
|
2797 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
2798 ++ga.ga_len;
|
|
|
2799
|
|
|
2800 /* Everything is fine, store the new b_langp value. */
|
|
|
2801 ga_clear(&buf->b_langp);
|
|
|
2802 buf->b_langp = ga;
|
|
|
2803
|
|
|
2804 return NULL;
|
|
|
2805 }
|
|
|
2806
|
|
|
2807 /*
|
|
376
|
2808 * Clear the midword characters for buffer "buf".
|
|
|
2809 */
|
|
|
2810 static void
|
|
|
2811 clear_midword(buf)
|
|
|
2812 buf_T *buf;
|
|
|
2813 {
|
|
|
2814 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
2815 #ifdef FEAT_MBYTE
|
|
|
2816 vim_free(buf->b_spell_ismw_mb);
|
|
|
2817 buf->b_spell_ismw_mb = NULL;
|
|
|
2818 #endif
|
|
|
2819 }
|
|
|
2820
|
|
|
2821 /*
|
|
|
2822 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
2823 * They add up to any currently used midword characters.
|
|
|
2824 */
|
|
|
2825 static void
|
|
|
2826 use_midword(lp, buf)
|
|
|
2827 slang_T *lp;
|
|
|
2828 buf_T *buf;
|
|
|
2829 {
|
|
|
2830 char_u *p;
|
|
|
2831
|
|
389
|
2832 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
2833 return;
|
|
|
2834
|
|
376
|
2835 for (p = lp->sl_midword; *p != NUL; )
|
|
|
2836 #ifdef FEAT_MBYTE
|
|
|
2837 if (has_mbyte)
|
|
|
2838 {
|
|
|
2839 int c, l, n;
|
|
|
2840 char_u *bp;
|
|
|
2841
|
|
|
2842 c = mb_ptr2char(p);
|
|
|
2843 l = mb_ptr2len_check(p);
|
|
|
2844 if (c < 256)
|
|
|
2845 buf->b_spell_ismw[c] = TRUE;
|
|
|
2846 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
2847 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
2848 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
2849 else
|
|
|
2850 {
|
|
|
2851 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
2852 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
2853 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
2854 if (bp != NULL)
|
|
|
2855 {
|
|
|
2856 vim_free(buf->b_spell_ismw_mb);
|
|
|
2857 buf->b_spell_ismw_mb = bp;
|
|
|
2858 vim_strncpy(bp + n, p, l);
|
|
|
2859 }
|
|
|
2860 }
|
|
|
2861 p += l;
|
|
|
2862 }
|
|
|
2863 else
|
|
|
2864 #endif
|
|
|
2865 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
2866 }
|
|
|
2867
|
|
|
2868 /*
|
|
236
|
2869 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
2870 * Each region is simply stored as the two characters of it's name.
|
|
381
|
2871 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
2872 */
|
|
|
2873 static int
|
|
|
2874 find_region(rp, region)
|
|
|
2875 char_u *rp;
|
|
|
2876 char_u *region;
|
|
|
2877 {
|
|
|
2878 int i;
|
|
|
2879
|
|
|
2880 for (i = 0; ; i += 2)
|
|
|
2881 {
|
|
|
2882 if (rp[i] == NUL)
|
|
|
2883 return REGION_ALL;
|
|
|
2884 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
2885 break;
|
|
|
2886 }
|
|
|
2887 return i / 2;
|
|
|
2888 }
|
|
|
2889
|
|
|
2890 /*
|
|
323
|
2891 * Return case type of word:
|
|
236
|
2892 * w word 0
|
|
300
|
2893 * Word WF_ONECAP
|
|
|
2894 * W WORD WF_ALLCAP
|
|
|
2895 * WoRd wOrd WF_KEEPCAP
|
|
236
|
2896 */
|
|
|
2897 static int
|
|
|
2898 captype(word, end)
|
|
|
2899 char_u *word;
|
|
323
|
2900 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
2901 {
|
|
|
2902 char_u *p;
|
|
|
2903 int c;
|
|
|
2904 int firstcap;
|
|
|
2905 int allcap;
|
|
|
2906 int past_second = FALSE; /* past second word char */
|
|
|
2907
|
|
|
2908 /* find first letter */
|
|
376
|
2909 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
2910 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
2911 return 0; /* only non-word characters, illegal word */
|
|
|
2912 #ifdef FEAT_MBYTE
|
|
310
|
2913 if (has_mbyte)
|
|
|
2914 c = mb_ptr2char_adv(&p);
|
|
|
2915 else
|
|
236
|
2916 #endif
|
|
310
|
2917 c = *p++;
|
|
324
|
2918 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
2919
|
|
|
2920 /*
|
|
|
2921 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
2922 * But a word with an upper char only at start is a ONECAP.
|
|
|
2923 */
|
|
323
|
2924 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
2925 if (spell_iswordp_nmw(p))
|
|
236
|
2926 {
|
|
455
|
2927 c = PTR2CHAR(p);
|
|
324
|
2928 if (!SPELL_ISUPPER(c))
|
|
236
|
2929 {
|
|
|
2930 /* UUl -> KEEPCAP */
|
|
|
2931 if (past_second && allcap)
|
|
300
|
2932 return WF_KEEPCAP;
|
|
236
|
2933 allcap = FALSE;
|
|
|
2934 }
|
|
|
2935 else if (!allcap)
|
|
|
2936 /* UlU -> KEEPCAP */
|
|
300
|
2937 return WF_KEEPCAP;
|
|
236
|
2938 past_second = TRUE;
|
|
|
2939 }
|
|
|
2940
|
|
|
2941 if (allcap)
|
|
300
|
2942 return WF_ALLCAP;
|
|
236
|
2943 if (firstcap)
|
|
300
|
2944 return WF_ONECAP;
|
|
236
|
2945 return 0;
|
|
|
2946 }
|
|
|
2947
|
|
355
|
2948 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
2949 /*
|
|
|
2950 * Free all languages.
|
|
|
2951 */
|
|
|
2952 void
|
|
|
2953 spell_free_all()
|
|
|
2954 {
|
|
|
2955 slang_T *lp;
|
|
|
2956 buf_T *buf;
|
|
385
|
2957 char_u fname[MAXPATHL];
|
|
355
|
2958
|
|
|
2959 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2960 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2961 ga_clear(&buf->b_langp);
|
|
|
2962
|
|
|
2963 while (first_lang != NULL)
|
|
|
2964 {
|
|
|
2965 lp = first_lang;
|
|
|
2966 first_lang = lp->sl_next;
|
|
|
2967 slang_free(lp);
|
|
|
2968 }
|
|
366
|
2969
|
|
385
|
2970 if (int_wordlist != NULL)
|
|
|
2971 {
|
|
|
2972 /* Delete the internal wordlist and its .spl file */
|
|
|
2973 mch_remove(int_wordlist);
|
|
|
2974 int_wordlist_spl(fname);
|
|
|
2975 mch_remove(fname);
|
|
|
2976 vim_free(int_wordlist);
|
|
|
2977 int_wordlist = NULL;
|
|
381
|
2978 }
|
|
|
2979
|
|
366
|
2980 init_spell_chartab();
|
|
355
|
2981 }
|
|
|
2982 # endif
|
|
|
2983
|
|
236
|
2984 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
2985 /*
|
|
|
2986 * Clear all spelling tables and reload them.
|
|
307
|
2987 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
2988 */
|
|
|
2989 void
|
|
|
2990 spell_reload()
|
|
|
2991 {
|
|
|
2992 buf_T *buf;
|
|
316
|
2993 win_T *wp;
|
|
236
|
2994
|
|
358
|
2995 /* Initialize the table for spell_iswordp(). */
|
|
236
|
2996 init_spell_chartab();
|
|
|
2997
|
|
|
2998 /* Unload all allocated memory. */
|
|
355
|
2999 spell_free_all();
|
|
236
|
3000
|
|
|
3001 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3002 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3003 {
|
|
316
|
3004 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
3005 * window for this buffer in which 'spell' is set. */
|
|
236
|
3006 if (*buf->b_p_spl != NUL)
|
|
316
|
3007 {
|
|
|
3008 FOR_ALL_WINDOWS(wp)
|
|
|
3009 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
3010 {
|
|
|
3011 (void)did_set_spelllang(buf);
|
|
|
3012 # ifdef FEAT_WINDOWS
|
|
|
3013 break;
|
|
|
3014 # endif
|
|
|
3015 }
|
|
|
3016 }
|
|
236
|
3017 }
|
|
|
3018 }
|
|
|
3019 # endif
|
|
|
3020
|
|
310
|
3021 /*
|
|
|
3022 * Reload the spell file "fname" if it's loaded.
|
|
|
3023 */
|
|
|
3024 static void
|
|
323
|
3025 spell_reload_one(fname, added_word)
|
|
310
|
3026 char_u *fname;
|
|
323
|
3027 int added_word; /* invoked through "zg" */
|
|
310
|
3028 {
|
|
|
3029 slang_T *lp;
|
|
323
|
3030 int didit = FALSE;
|
|
310
|
3031
|
|
|
3032 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3033 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3034 {
|
|
|
3035 slang_clear(lp);
|
|
323
|
3036 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3037 redraw_all_later(NOT_VALID);
|
|
323
|
3038 didit = TRUE;
|
|
310
|
3039 }
|
|
323
|
3040
|
|
|
3041 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3042 * 'spelllang' to get it loaded. */
|
|
|
3043 if (added_word && !didit)
|
|
|
3044 did_set_spelllang(curbuf);
|
|
310
|
3045 }
|
|
|
3046
|
|
|
3047
|
|
236
|
3048 /*
|
|
|
3049 * Functions for ":mkspell".
|
|
|
3050 */
|
|
|
3051
|
|
300
|
3052 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3053 and .dic file. */
|
|
|
3054 /*
|
|
|
3055 * Main structure to store the contents of a ".aff" file.
|
|
|
3056 */
|
|
|
3057 typedef struct afffile_S
|
|
|
3058 {
|
|
|
3059 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
310
|
3060 int af_rar; /* RAR ID for rare word */
|
|
|
3061 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
3062 int af_bad; /* BAD ID for banned word */
|
|
339
|
3063 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3064 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3065 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
3066 } afffile_T;
|
|
|
3067
|
|
|
3068 typedef struct affentry_S affentry_T;
|
|
|
3069 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3070 struct affentry_S
|
|
|
3071 {
|
|
|
3072 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3073 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3074 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3075 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3076 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
366
|
3077 int ae_rare; /* rare affix */
|
|
300
|
3078 };
|
|
|
3079
|
|
455
|
3080 #define AH_KEY_LEN 10
|
|
|
3081
|
|
300
|
3082 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3083 typedef struct affheader_S
|
|
|
3084 {
|
|
455
|
3085 /* key for hashtable == name of affix entry */
|
|
|
3086 #ifdef FEAT_MBYTE
|
|
|
3087 char_u ah_key[AH_KEY_LEN]; /* multi-byte char plus NUL */
|
|
|
3088 #else
|
|
|
3089 char_u ah_key[2]; /* one byte char plus NUL */
|
|
|
3090 #endif
|
|
339
|
3091 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
3092 int ah_combine; /* suffix may combine with prefix */
|
|
|
3093 affentry_T *ah_first; /* first affix entry */
|
|
|
3094 } affheader_T;
|
|
|
3095
|
|
|
3096 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
3097
|
|
|
3098 /*
|
|
|
3099 * Structure that is used to store the items in the word tree. This avoids
|
|
|
3100 * the need to keep track of each allocated thing, it's freed all at once
|
|
|
3101 * after ":mkspell" is done.
|
|
|
3102 */
|
|
|
3103 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
3104 typedef struct sblock_S sblock_T;
|
|
|
3105 struct sblock_S
|
|
|
3106 {
|
|
|
3107 sblock_T *sb_next; /* next block in list */
|
|
|
3108 int sb_used; /* nr of bytes already in use */
|
|
|
3109 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
3110 };
|
|
|
3111
|
|
|
3112 /*
|
|
300
|
3113 * A node in the tree.
|
|
236
|
3114 */
|
|
300
|
3115 typedef struct wordnode_S wordnode_T;
|
|
|
3116 struct wordnode_S
|
|
236
|
3117 {
|
|
346
|
3118 union /* shared to save space */
|
|
|
3119 {
|
|
469
|
3120 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
3121 int index; /* index in written nodes (valid after first
|
|
|
3122 round) */
|
|
|
3123 } wn_u1;
|
|
|
3124 union /* shared to save space */
|
|
|
3125 {
|
|
|
3126 wordnode_T *next; /* next node with same hash key */
|
|
|
3127 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
3128 } wn_u2;
|
|
300
|
3129 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
3130 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
3131 always sorted) */
|
|
469
|
3132 int wn_refs; /* nr of references to this node */
|
|
300
|
3133 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
469
|
3134 char_u wn_prefixID; /* when "wn_byte" is NUL: supported/required
|
|
|
3135 prefix ID or 0 */
|
|
|
3136 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
3137 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
3138 PREFIXTREE it's the prefcondnr */
|
|
469
|
3139 #ifdef SPELL_PRINTTREE
|
|
|
3140 int wn_nr; /* sequence nr for printing */
|
|
|
3141 #endif
|
|
236
|
3142 };
|
|
|
3143
|
|
390
|
3144 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
3145
|
|
300
|
3146 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
3147
|
|
300
|
3148 /*
|
|
|
3149 * Info used while reading the spell files.
|
|
|
3150 */
|
|
|
3151 typedef struct spellinfo_S
|
|
249
|
3152 {
|
|
300
|
3153 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
3154 long si_foldwcount; /* nr of words in si_foldroot */
|
|
300
|
3155 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
3156 long si_keepwcount; /* nr of words in si_keeproot */
|
|
339
|
3157 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
300
|
3158 sblock_T *si_blocks; /* memory blocks used */
|
|
|
3159 int si_ascii; /* handling only ASCII words */
|
|
310
|
3160 int si_add; /* addition file */
|
|
351
|
3161 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
3162 int si_region; /* region mask */
|
|
|
3163 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
3164 int si_memtot; /* runtime memory used */
|
|
310
|
3165 int si_verbose; /* verbose messages */
|
|
316
|
3166 int si_region_count; /* number of regions supported (1 when there
|
|
|
3167 are no regions) */
|
|
|
3168 char_u si_region_name[16]; /* region names (if count > 1) */
|
|
323
|
3169
|
|
|
3170 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
3171 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
3172 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
3173 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
3174 int si_followup; /* soundsalike: ? */
|
|
|
3175 int si_collapse; /* soundsalike: ? */
|
|
|
3176 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
3177 garray_T si_map; /* MAP info concatenated */
|
|
366
|
3178 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
339
|
3179 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
3180 * prefixes, each stored as a string */
|
|
|
3181 int si_newID; /* current value for ah_newID */
|
|
300
|
3182 } spellinfo_T;
|
|
249
|
3183
|
|
300
|
3184 static afffile_T *spell_read_aff __ARGS((char_u *fname, spellinfo_T *spin));
|
|
339
|
3185 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
3186 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
3187 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
3188 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
3189 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
|
3190 static int spell_read_dic __ARGS((char_u *fname, spellinfo_T *spin, afffile_T *affile));
|
|
339
|
3191 static char_u *get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, sblock_T **blp));
|
|
|
3192 static int store_aff_word __ARGS((char_u *word, spellinfo_T *spin, char_u *afflist, afffile_T *affile, hashtab_T *ht, hashtab_T *xht, int comb, int flags, char_u *pfxlist));
|
|
300
|
3193 static int spell_read_wordfile __ARGS((char_u *fname, spellinfo_T *spin));
|
|
447
|
3194 static void *getroom __ARGS((sblock_T **blp, size_t len, int align));
|
|
300
|
3195 static char_u *getroom_save __ARGS((sblock_T **blp, char_u *s));
|
|
|
3196 static void free_blocks __ARGS((sblock_T *bl));
|
|
|
3197 static wordnode_T *wordtree_alloc __ARGS((sblock_T **blp));
|
|
339
|
3198 static int store_word __ARGS((char_u *word, spellinfo_T *spin, int flags, int region, char_u *pfxlist));
|
|
469
|
3199 static int tree_add_word __ARGS((char_u *word, wordnode_T *tree, int flags, int region, int prefixID, spellinfo_T *spin));
|
|
|
3200 static wordnode_T *get_wordnode __ARGS((sblock_T **blp));
|
|
|
3201 static void free_wordnode __ARGS((wordnode_T *n));
|
|
310
|
3202 static void wordtree_compress __ARGS((wordnode_T *root, spellinfo_T *spin));
|
|
300
|
3203 static int node_compress __ARGS((wordnode_T *node, hashtab_T *ht, int *tot));
|
|
|
3204 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
316
|
3205 static void write_vim_spell __ARGS((char_u *fname, spellinfo_T *spin));
|
|
346
|
3206 static void clear_node __ARGS((wordnode_T *node));
|
|
|
3207 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
3208 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3209 static void init_spellfile __ARGS((void));
|
|
236
|
3210
|
|
455
|
3211 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
3212 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
3213 * Use a negative number with the lower 8 bits zero. */
|
|
|
3214 #define PFX_FLAGS -256
|
|
390
|
3215
|
|
469
|
3216 static int words_added = 0; /* number of words added to tree */
|
|
|
3217
|
|
|
3218 #ifdef SPELL_PRINTTREE
|
|
|
3219 /*
|
|
|
3220 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
3221 * readable format, so that we can see what happens when adding a word and/or
|
|
|
3222 * compressing the tree.
|
|
|
3223 * Based on code from Olaf Seibert.
|
|
|
3224 */
|
|
|
3225 #define PRINTLINESIZE 1000
|
|
|
3226 #define PRINTWIDTH 6
|
|
|
3227
|
|
|
3228 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
3229 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
3230
|
|
|
3231 static char line1[PRINTLINESIZE];
|
|
|
3232 static char line2[PRINTLINESIZE];
|
|
|
3233 static char line3[PRINTLINESIZE];
|
|
|
3234
|
|
|
3235 static void
|
|
|
3236 spell_clear_flags(wordnode_T *node)
|
|
|
3237 {
|
|
|
3238 wordnode_T *np;
|
|
|
3239
|
|
|
3240 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
3241 {
|
|
|
3242 np->wn_u1.index = FALSE;
|
|
|
3243 spell_clear_flags(np->wn_child);
|
|
|
3244 }
|
|
|
3245 }
|
|
|
3246
|
|
|
3247 static void
|
|
|
3248 spell_print_node(wordnode_T *node, int depth)
|
|
|
3249 {
|
|
|
3250 if (node->wn_u1.index)
|
|
|
3251 {
|
|
|
3252 /* Done this node before, print the reference. */
|
|
|
3253 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
3254 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
3255 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3256 msg(line1);
|
|
|
3257 msg(line2);
|
|
|
3258 msg(line3);
|
|
|
3259 }
|
|
|
3260 else
|
|
|
3261 {
|
|
|
3262 node->wn_u1.index = TRUE;
|
|
|
3263
|
|
|
3264 if (node->wn_byte != NUL)
|
|
|
3265 {
|
|
|
3266 if (node->wn_child != NULL)
|
|
|
3267 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
3268 else
|
|
|
3269 /* Cannot happen? */
|
|
|
3270 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
3271 }
|
|
|
3272 else
|
|
|
3273 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
3274
|
|
|
3275 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
3276
|
|
|
3277 if (node->wn_sibling != NULL)
|
|
|
3278 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
3279 else
|
|
|
3280 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3281
|
|
|
3282 if (node->wn_byte == NUL)
|
|
|
3283 {
|
|
|
3284 msg(line1);
|
|
|
3285 msg(line2);
|
|
|
3286 msg(line3);
|
|
|
3287 }
|
|
|
3288
|
|
|
3289 /* do the children */
|
|
|
3290 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
3291 spell_print_node(node->wn_child, depth + 1);
|
|
|
3292
|
|
|
3293 /* do the siblings */
|
|
|
3294 if (node->wn_sibling != NULL)
|
|
|
3295 {
|
|
|
3296 /* get rid of all parent details except | */
|
|
|
3297 STRCPY(line1, line3);
|
|
|
3298 STRCPY(line2, line3);
|
|
|
3299 spell_print_node(node->wn_sibling, depth);
|
|
|
3300 }
|
|
|
3301 }
|
|
|
3302 }
|
|
|
3303
|
|
|
3304 static void
|
|
|
3305 spell_print_tree(wordnode_T *root)
|
|
|
3306 {
|
|
|
3307 if (root != NULL)
|
|
|
3308 {
|
|
|
3309 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
3310 * done. */
|
|
|
3311 spell_clear_flags(root);
|
|
|
3312
|
|
|
3313 /* Recursively print the tree. */
|
|
|
3314 spell_print_node(root, 0);
|
|
|
3315 }
|
|
|
3316 }
|
|
|
3317 #endif /* SPELL_PRINTTREE */
|
|
|
3318
|
|
236
|
3319 /*
|
|
323
|
3320 * Read the affix file "fname".
|
|
316
|
3321 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
3322 */
|
|
|
3323 static afffile_T *
|
|
300
|
3324 spell_read_aff(fname, spin)
|
|
236
|
3325 char_u *fname;
|
|
300
|
3326 spellinfo_T *spin;
|
|
236
|
3327 {
|
|
|
3328 FILE *fd;
|
|
|
3329 afffile_T *aff;
|
|
|
3330 char_u rline[MAXLINELEN];
|
|
|
3331 char_u *line;
|
|
|
3332 char_u *pc = NULL;
|
|
334
|
3333 #define MAXITEMCNT 7
|
|
|
3334 char_u *(items[MAXITEMCNT]);
|
|
236
|
3335 int itemcnt;
|
|
|
3336 char_u *p;
|
|
|
3337 int lnum = 0;
|
|
|
3338 affheader_T *cur_aff = NULL;
|
|
|
3339 int aff_todo = 0;
|
|
|
3340 hashtab_T *tp;
|
|
255
|
3341 char_u *low = NULL;
|
|
|
3342 char_u *fol = NULL;
|
|
|
3343 char_u *upp = NULL;
|
|
307
|
3344 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
3345 int do_rep;
|
|
|
3346 int do_sal;
|
|
|
3347 int do_map;
|
|
366
|
3348 int do_midword;
|
|
375
|
3349 int do_sofo;
|
|
323
|
3350 int found_map = FALSE;
|
|
324
|
3351 hashitem_T *hi;
|
|
455
|
3352 int l;
|
|
236
|
3353
|
|
300
|
3354 /*
|
|
|
3355 * Open the file.
|
|
|
3356 */
|
|
310
|
3357 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3358 if (fd == NULL)
|
|
|
3359 {
|
|
|
3360 EMSG2(_(e_notopen), fname);
|
|
|
3361 return NULL;
|
|
|
3362 }
|
|
|
3363
|
|
310
|
3364 if (spin->si_verbose || p_verbose > 2)
|
|
|
3365 {
|
|
|
3366 if (!spin->si_verbose)
|
|
|
3367 verbose_enter();
|
|
366
|
3368 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
3369 out_flush();
|
|
|
3370 if (!spin->si_verbose)
|
|
|
3371 verbose_leave();
|
|
|
3372 }
|
|
236
|
3373
|
|
323
|
3374 /* Only do REP lines when not done in another .aff file already. */
|
|
|
3375 do_rep = spin->si_rep.ga_len == 0;
|
|
|
3376
|
|
|
3377 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
3378 do_sal = spin->si_sal.ga_len == 0;
|
|
|
3379
|
|
|
3380 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
3381 do_map = spin->si_map.ga_len == 0;
|
|
|
3382
|
|
366
|
3383 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
3384 do_midword = spin->si_midword == NULL;
|
|
|
3385
|
|
375
|
3386 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
3387 do_sofo = spin->si_sofofr == NULL;
|
|
|
3388
|
|
300
|
3389 /*
|
|
|
3390 * Allocate and init the afffile_T structure.
|
|
|
3391 */
|
|
447
|
3392 aff = (afffile_T *)getroom(&spin->si_blocks, sizeof(afffile_T), TRUE);
|
|
236
|
3393 if (aff == NULL)
|
|
|
3394 return NULL;
|
|
|
3395 hash_init(&aff->af_pref);
|
|
|
3396 hash_init(&aff->af_suff);
|
|
|
3397
|
|
|
3398 /*
|
|
|
3399 * Read all the lines in the file one by one.
|
|
|
3400 */
|
|
255
|
3401 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
3402 {
|
|
255
|
3403 line_breakcheck();
|
|
236
|
3404 ++lnum;
|
|
|
3405
|
|
|
3406 /* Skip comment lines. */
|
|
|
3407 if (*rline == '#')
|
|
|
3408 continue;
|
|
|
3409
|
|
|
3410 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
3411 vim_free(pc);
|
|
310
|
3412 #ifdef FEAT_MBYTE
|
|
300
|
3413 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3414 {
|
|
300
|
3415 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
3416 if (pc == NULL)
|
|
|
3417 {
|
|
|
3418 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3419 fname, lnum, rline);
|
|
|
3420 continue;
|
|
|
3421 }
|
|
236
|
3422 line = pc;
|
|
|
3423 }
|
|
|
3424 else
|
|
310
|
3425 #endif
|
|
236
|
3426 {
|
|
|
3427 pc = NULL;
|
|
|
3428 line = rline;
|
|
|
3429 }
|
|
|
3430
|
|
|
3431 /* Split the line up in white separated items. Put a NUL after each
|
|
|
3432 * item. */
|
|
|
3433 itemcnt = 0;
|
|
|
3434 for (p = line; ; )
|
|
|
3435 {
|
|
|
3436 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
3437 ++p;
|
|
|
3438 if (*p == NUL)
|
|
|
3439 break;
|
|
334
|
3440 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
3441 break;
|
|
236
|
3442 items[itemcnt++] = p;
|
|
300
|
3443 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
3444 ++p;
|
|
|
3445 if (*p == NUL)
|
|
|
3446 break;
|
|
|
3447 *p++ = NUL;
|
|
|
3448 }
|
|
|
3449
|
|
|
3450 /* Handle non-empty lines. */
|
|
|
3451 if (itemcnt > 0)
|
|
|
3452 {
|
|
|
3453 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
3454 && aff->af_enc == NULL)
|
|
|
3455 {
|
|
310
|
3456 #ifdef FEAT_MBYTE
|
|
300
|
3457 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
3458 aff->af_enc = enc_canonize(items[1]);
|
|
|
3459 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
3460 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
3461 p_enc) == FAIL)
|
|
|
3462 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
3463 fname, aff->af_enc, p_enc);
|
|
375
|
3464 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3465 #else
|
|
|
3466 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3467 #endif
|
|
236
|
3468 }
|
|
366
|
3469 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
3470 {
|
|
|
3471 if (do_midword)
|
|
|
3472 spin->si_midword = vim_strsave(items[1]);
|
|
|
3473 }
|
|
302
|
3474 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
3475 {
|
|
323
|
3476 /* ignored, we always split */
|
|
302
|
3477 }
|
|
323
|
3478 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
3479 {
|
|
323
|
3480 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
3481 }
|
|
307
|
3482 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
3483 && aff->af_rar == 0)
|
|
|
3484 {
|
|
|
3485 aff->af_rar = items[1][0];
|
|
|
3486 if (items[1][1] != NUL)
|
|
|
3487 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3488 }
|
|
310
|
3489 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
3490 && aff->af_kep == 0)
|
|
307
|
3491 {
|
|
310
|
3492 aff->af_kep = items[1][0];
|
|
307
|
3493 if (items[1][1] != NUL)
|
|
|
3494 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3495 }
|
|
346
|
3496 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
3497 && aff->af_bad == 0)
|
|
|
3498 {
|
|
|
3499 aff->af_bad = items[1][0];
|
|
|
3500 if (items[1][1] != NUL)
|
|
|
3501 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3502 }
|
|
339
|
3503 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
3504 {
|
|
|
3505 aff->af_pfxpostpone = TRUE;
|
|
|
3506 }
|
|
236
|
3507 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3508 || STRCMP(items[0], "SFX") == 0)
|
|
|
3509 && aff_todo == 0
|
|
334
|
3510 && itemcnt >= 4)
|
|
236
|
3511 {
|
|
334
|
3512 /* Myspell allows extra text after the item, but that might
|
|
|
3513 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
3514 if (itemcnt > 4 && *items[4] != '#')
|
|
|
3515 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
3516 fname, lnum, items[4]);
|
|
|
3517
|
|
236
|
3518 /* New affix letter. */
|
|
300
|
3519 cur_aff = (affheader_T *)getroom(&spin->si_blocks,
|
|
447
|
3520 sizeof(affheader_T), TRUE);
|
|
236
|
3521 if (cur_aff == NULL)
|
|
|
3522 break;
|
|
455
|
3523 #ifdef FEAT_MBYTE
|
|
|
3524 if (has_mbyte)
|
|
|
3525 {
|
|
|
3526 l = mb_ptr2len_check(items[1]);
|
|
|
3527 if (l >= AH_KEY_LEN)
|
|
|
3528 l = 1; /* too long, must be an overlong sequence */
|
|
|
3529 else
|
|
|
3530 mch_memmove(cur_aff->ah_key, items[1], l);
|
|
|
3531 }
|
|
|
3532 else
|
|
|
3533 #endif
|
|
|
3534 {
|
|
|
3535 *cur_aff->ah_key = *items[1];
|
|
|
3536 l = 1;
|
|
|
3537 }
|
|
|
3538 cur_aff->ah_key[l] = NUL;
|
|
|
3539 if (items[1][l] != NUL)
|
|
307
|
3540 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
3541 if (*items[2] == 'Y')
|
|
|
3542 cur_aff->ah_combine = TRUE;
|
|
300
|
3543 else if (*items[2] != 'N')
|
|
236
|
3544 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
3545 fname, lnum, items[2]);
|
|
339
|
3546
|
|
236
|
3547 if (*items[0] == 'P')
|
|
339
|
3548 {
|
|
236
|
3549 tp = &aff->af_pref;
|
|
339
|
3550 /* Use a new number in the .spl file later, to be able to
|
|
|
3551 * handle multiple .aff files. */
|
|
|
3552 if (aff->af_pfxpostpone)
|
|
344
|
3553 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
3554 }
|
|
236
|
3555 else
|
|
|
3556 tp = &aff->af_suff;
|
|
300
|
3557 aff_todo = atoi((char *)items[3]);
|
|
324
|
3558 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
3559 if (!HASHITEM_EMPTY(hi))
|
|
300
|
3560 {
|
|
236
|
3561 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
3562 fname, lnum, items[1]);
|
|
300
|
3563 aff_todo = 0;
|
|
|
3564 }
|
|
236
|
3565 else
|
|
|
3566 hash_add(tp, cur_aff->ah_key);
|
|
|
3567 }
|
|
|
3568 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3569 || STRCMP(items[0], "SFX") == 0)
|
|
|
3570 && aff_todo > 0
|
|
|
3571 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
3572 && itemcnt >= 5)
|
|
236
|
3573 {
|
|
|
3574 affentry_T *aff_entry;
|
|
366
|
3575 int rare = FALSE;
|
|
455
|
3576 int upper = FALSE;
|
|
366
|
3577 int lasti = 5;
|
|
|
3578
|
|
|
3579 /* Check for "rare" after the other info. */
|
|
|
3580 if (itemcnt > 5 && STRICMP(items[5], "rare") == 0)
|
|
|
3581 {
|
|
|
3582 rare = TRUE;
|
|
|
3583 lasti = 6;
|
|
|
3584 }
|
|
236
|
3585
|
|
334
|
3586 /* Myspell allows extra text after the item, but that might
|
|
|
3587 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
3588 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
3589 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
366
|
3590 fname, lnum, items[lasti]);
|
|
334
|
3591
|
|
236
|
3592 /* New item for an affix letter. */
|
|
|
3593 --aff_todo;
|
|
300
|
3594 aff_entry = (affentry_T *)getroom(&spin->si_blocks,
|
|
447
|
3595 sizeof(affentry_T), TRUE);
|
|
236
|
3596 if (aff_entry == NULL)
|
|
|
3597 break;
|
|
366
|
3598 aff_entry->ae_rare = rare;
|
|
240
|
3599
|
|
236
|
3600 if (STRCMP(items[2], "0") != 0)
|
|
300
|
3601 aff_entry->ae_chop = getroom_save(&spin->si_blocks,
|
|
|
3602 items[2]);
|
|
236
|
3603 if (STRCMP(items[3], "0") != 0)
|
|
300
|
3604 aff_entry->ae_add = getroom_save(&spin->si_blocks,
|
|
|
3605 items[3]);
|
|
236
|
3606
|
|
300
|
3607 /* Don't use an affix entry with non-ASCII characters when
|
|
|
3608 * "spin->si_ascii" is TRUE. */
|
|
|
3609 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
3610 || has_non_ascii(aff_entry->ae_add)))
|
|
|
3611 {
|
|
|
3612 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
3613 cur_aff->ah_first = aff_entry;
|
|
300
|
3614
|
|
|
3615 if (STRCMP(items[4], ".") != 0)
|
|
|
3616 {
|
|
|
3617 char_u buf[MAXLINELEN];
|
|
|
3618
|
|
|
3619 aff_entry->ae_cond = getroom_save(&spin->si_blocks,
|
|
|
3620 items[4]);
|
|
|
3621 if (*items[0] == 'P')
|
|
|
3622 sprintf((char *)buf, "^%s", items[4]);
|
|
|
3623 else
|
|
|
3624 sprintf((char *)buf, "%s$", items[4]);
|
|
|
3625 aff_entry->ae_prog = vim_regcomp(buf,
|
|
|
3626 RE_MAGIC + RE_STRING);
|
|
|
3627 }
|
|
339
|
3628
|
|
|
3629 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
3630 * for the condition. Use an existing one if possible. */
|
|
455
|
3631 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
3632 {
|
|
455
|
3633 /* When the chop string is one lower-case letter and
|
|
|
3634 * the add string ends in the upper-case letter we set
|
|
|
3635 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
3636 * letters from "ae_add". The condition must either
|
|
|
3637 * be empty or start with the same letter. */
|
|
|
3638 if (aff_entry->ae_chop != NULL
|
|
|
3639 && aff_entry->ae_add != NULL
|
|
|
3640 #ifdef FEAT_MBYTE
|
|
|
3641 && aff_entry->ae_chop[mb_ptr2len_check(
|
|
|
3642 aff_entry->ae_chop)] == NUL
|
|
|
3643 #else
|
|
|
3644 && aff_entry->ae_chop[1] == NUL
|
|
|
3645 #endif
|
|
|
3646 )
|
|
339
|
3647 {
|
|
455
|
3648 int c, c_up;
|
|
|
3649
|
|
|
3650 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
3651 c_up = SPELL_TOUPPER(c);
|
|
|
3652 if (c_up != c
|
|
|
3653 && (aff_entry->ae_cond == NULL
|
|
|
3654 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
3655 {
|
|
|
3656 p = aff_entry->ae_add
|
|
|
3657 + STRLEN(aff_entry->ae_add);
|
|
|
3658 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
3659 if (PTR2CHAR(p) == c_up)
|
|
|
3660 {
|
|
|
3661 upper = TRUE;
|
|
|
3662 aff_entry->ae_chop = NULL;
|
|
|
3663 *p = NUL;
|
|
|
3664
|
|
|
3665 /* The condition is matched with the
|
|
|
3666 * actual word, thus must check for the
|
|
|
3667 * upper-case letter. */
|
|
|
3668 if (aff_entry->ae_cond != NULL)
|
|
|
3669 {
|
|
|
3670 char_u buf[MAXLINELEN];
|
|
|
3671 #ifdef FEAT_MBYTE
|
|
|
3672 if (has_mbyte)
|
|
|
3673 {
|
|
|
3674 onecap_copy(items[4], buf, TRUE);
|
|
|
3675 aff_entry->ae_cond = getroom_save(
|
|
|
3676 &spin->si_blocks, buf);
|
|
|
3677 }
|
|
|
3678 else
|
|
|
3679 #endif
|
|
|
3680 *aff_entry->ae_cond = c_up;
|
|
|
3681 if (aff_entry->ae_cond != NULL)
|
|
|
3682 {
|
|
|
3683 sprintf((char *)buf, "^%s",
|
|
339
|
3684 aff_entry->ae_cond);
|
|
455
|
3685 vim_free(aff_entry->ae_prog);
|
|
|
3686 aff_entry->ae_prog = vim_regcomp(
|
|
|
3687 buf, RE_MAGIC + RE_STRING);
|
|
|
3688 }
|
|
|
3689 }
|
|
|
3690 }
|
|
|
3691 }
|
|
339
|
3692 }
|
|
|
3693
|
|
455
|
3694 if (aff_entry->ae_chop == NULL)
|
|
390
|
3695 {
|
|
455
|
3696 int idx;
|
|
|
3697 char_u **pp;
|
|
|
3698 int n;
|
|
|
3699
|
|
|
3700 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
3701 --idx)
|
|
|
3702 {
|
|
|
3703 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
3704 if (str_equal(p, aff_entry->ae_cond))
|
|
|
3705 break;
|
|
|
3706 }
|
|
|
3707 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
3708 {
|
|
|
3709 /* Not found, add a new condition. */
|
|
|
3710 idx = spin->si_prefcond.ga_len++;
|
|
|
3711 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
3712 + idx;
|
|
|
3713 if (aff_entry->ae_cond == NULL)
|
|
|
3714 *pp = NULL;
|
|
|
3715 else
|
|
|
3716 *pp = getroom_save(&spin->si_blocks,
|
|
|
3717 aff_entry->ae_cond);
|
|
|
3718 }
|
|
|
3719
|
|
|
3720 /* Add the prefix to the prefix tree. */
|
|
|
3721 if (aff_entry->ae_add == NULL)
|
|
|
3722 p = (char_u *)"";
|
|
390
|
3723 else
|
|
455
|
3724 p = aff_entry->ae_add;
|
|
|
3725 /* PFX_FLAGS is a negative number, so that
|
|
|
3726 * tree_add_word() knows this is the prefix tree. */
|
|
|
3727 n = PFX_FLAGS;
|
|
|
3728 if (rare)
|
|
|
3729 n |= WFP_RARE;
|
|
|
3730 if (!cur_aff->ah_combine)
|
|
|
3731 n |= WFP_NC;
|
|
|
3732 if (upper)
|
|
|
3733 n |= WFP_UP;
|
|
|
3734 tree_add_word(p, spin->si_prefroot, n,
|
|
469
|
3735 idx, cur_aff->ah_newID, spin);
|
|
390
|
3736 }
|
|
339
|
3737 }
|
|
240
|
3738 }
|
|
236
|
3739 }
|
|
255
|
3740 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
3741 {
|
|
|
3742 if (fol != NULL)
|
|
|
3743 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
3744 fname, lnum);
|
|
|
3745 else
|
|
|
3746 fol = vim_strsave(items[1]);
|
|
|
3747 }
|
|
|
3748 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
3749 {
|
|
|
3750 if (low != NULL)
|
|
|
3751 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
3752 fname, lnum);
|
|
|
3753 else
|
|
|
3754 low = vim_strsave(items[1]);
|
|
|
3755 }
|
|
|
3756 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
3757 {
|
|
|
3758 if (upp != NULL)
|
|
|
3759 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
3760 fname, lnum);
|
|
|
3761 else
|
|
|
3762 upp = vim_strsave(items[1]);
|
|
|
3763 }
|
|
236
|
3764 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
3765 {
|
|
236
|
3766 /* Ignore REP count */;
|
|
323
|
3767 if (!isdigit(*items[1]))
|
|
|
3768 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
3769 fname, lnum);
|
|
|
3770 }
|
|
236
|
3771 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 3)
|
|
|
3772 {
|
|
|
3773 /* REP item */
|
|
323
|
3774 if (do_rep)
|
|
|
3775 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
3776 }
|
|
|
3777 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
3778 {
|
|
|
3779 /* MAP item or count */
|
|
|
3780 if (!found_map)
|
|
|
3781 {
|
|
|
3782 /* First line contains the count. */
|
|
|
3783 found_map = TRUE;
|
|
|
3784 if (!isdigit(*items[1]))
|
|
|
3785 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
3786 fname, lnum);
|
|
|
3787 }
|
|
|
3788 else if (do_map)
|
|
|
3789 {
|
|
346
|
3790 int c;
|
|
|
3791
|
|
|
3792 /* Check that every character appears only once. */
|
|
|
3793 for (p = items[1]; *p != NUL; )
|
|
|
3794 {
|
|
|
3795 #ifdef FEAT_MBYTE
|
|
|
3796 c = mb_ptr2char_adv(&p);
|
|
|
3797 #else
|
|
|
3798 c = *p++;
|
|
|
3799 #endif
|
|
|
3800 if ((spin->si_map.ga_len > 0
|
|
|
3801 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
3802 != NULL)
|
|
|
3803 || vim_strchr(p, c) != NULL)
|
|
|
3804 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
3805 fname, lnum);
|
|
|
3806 }
|
|
|
3807
|
|
323
|
3808 /* We simply concatenate all the MAP strings, separated by
|
|
|
3809 * slashes. */
|
|
|
3810 ga_concat(&spin->si_map, items[1]);
|
|
|
3811 ga_append(&spin->si_map, '/');
|
|
|
3812 }
|
|
|
3813 }
|
|
|
3814 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
3815 {
|
|
|
3816 if (do_sal)
|
|
|
3817 {
|
|
|
3818 /* SAL item (sounds-a-like)
|
|
|
3819 * Either one of the known keys or a from-to pair. */
|
|
|
3820 if (STRCMP(items[1], "followup") == 0)
|
|
|
3821 spin->si_followup = sal_to_bool(items[2]);
|
|
|
3822 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
3823 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
3824 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
3825 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
3826 else
|
|
|
3827 /* when "to" is "_" it means empty */
|
|
|
3828 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
3829 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
3830 : items[2]);
|
|
|
3831 }
|
|
236
|
3832 }
|
|
375
|
3833 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
3834 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
3835 {
|
|
|
3836 if (do_sofo)
|
|
|
3837 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
3838 }
|
|
|
3839 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
3840 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
3841 {
|
|
|
3842 if (do_sofo)
|
|
|
3843 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
3844 }
|
|
300
|
3845 else
|
|
236
|
3846 smsg((char_u *)_("Unrecognized item in %s line %d: %s"),
|
|
|
3847 fname, lnum, items[0]);
|
|
|
3848 }
|
|
|
3849 }
|
|
|
3850
|
|
375
|
3851 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
3852 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
3853 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
3854 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
3855 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
3856
|
|
255
|
3857 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
3858 {
|
|
351
|
3859 if (spin->si_clear_chartab)
|
|
|
3860 {
|
|
|
3861 /* Clear the char type tables, don't want to use any of the
|
|
|
3862 * currently used spell properties. */
|
|
|
3863 init_spell_chartab();
|
|
|
3864 spin->si_clear_chartab = FALSE;
|
|
|
3865 }
|
|
|
3866
|
|
316
|
3867 /*
|
|
|
3868 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
3869 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
3870 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
3871 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
3872 */
|
|
|
3873 if (!spin->si_ascii
|
|
|
3874 #ifdef FEAT_MBYTE
|
|
|
3875 && !enc_utf8
|
|
|
3876 #endif
|
|
|
3877 )
|
|
260
|
3878 {
|
|
|
3879 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
3880 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
3881 else
|
|
316
|
3882 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
3883 }
|
|
255
|
3884
|
|
|
3885 vim_free(fol);
|
|
|
3886 vim_free(low);
|
|
|
3887 vim_free(upp);
|
|
|
3888 }
|
|
|
3889
|
|
236
|
3890 vim_free(pc);
|
|
|
3891 fclose(fd);
|
|
|
3892 return aff;
|
|
|
3893 }
|
|
|
3894
|
|
|
3895 /*
|
|
339
|
3896 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
3897 * NULL as equal.
|
|
|
3898 */
|
|
|
3899 static int
|
|
|
3900 str_equal(s1, s2)
|
|
|
3901 char_u *s1;
|
|
|
3902 char_u *s2;
|
|
|
3903 {
|
|
|
3904 if (s1 == NULL || s2 == NULL)
|
|
|
3905 return s1 == s2;
|
|
|
3906 return STRCMP(s1, s2) == 0;
|
|
|
3907 }
|
|
|
3908
|
|
|
3909 /*
|
|
323
|
3910 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
3911 * They are stored case-folded.
|
|
|
3912 */
|
|
|
3913 static void
|
|
|
3914 add_fromto(spin, gap, from, to)
|
|
|
3915 spellinfo_T *spin;
|
|
|
3916 garray_T *gap;
|
|
|
3917 char_u *from;
|
|
|
3918 char_u *to;
|
|
|
3919 {
|
|
|
3920 fromto_T *ftp;
|
|
|
3921 char_u word[MAXWLEN];
|
|
|
3922
|
|
|
3923 if (ga_grow(gap, 1) == OK)
|
|
|
3924 {
|
|
|
3925 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
3926 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
|
3927 ftp->ft_from = getroom_save(&spin->si_blocks, word);
|
|
|
3928 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
|
3929 ftp->ft_to = getroom_save(&spin->si_blocks, word);
|
|
|
3930 ++gap->ga_len;
|
|
|
3931 }
|
|
|
3932 }
|
|
|
3933
|
|
|
3934 /*
|
|
|
3935 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
3936 */
|
|
|
3937 static int
|
|
|
3938 sal_to_bool(s)
|
|
|
3939 char_u *s;
|
|
|
3940 {
|
|
|
3941 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
3942 }
|
|
|
3943
|
|
|
3944 /*
|
|
240
|
3945 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
3946 * When "s" is NULL FALSE is returned.
|
|
|
3947 */
|
|
|
3948 static int
|
|
|
3949 has_non_ascii(s)
|
|
|
3950 char_u *s;
|
|
|
3951 {
|
|
|
3952 char_u *p;
|
|
|
3953
|
|
|
3954 if (s != NULL)
|
|
|
3955 for (p = s; *p != NUL; ++p)
|
|
|
3956 if (*p >= 128)
|
|
|
3957 return TRUE;
|
|
|
3958 return FALSE;
|
|
|
3959 }
|
|
|
3960
|
|
|
3961 /*
|
|
236
|
3962 * Free the structure filled by spell_read_aff().
|
|
|
3963 */
|
|
|
3964 static void
|
|
|
3965 spell_free_aff(aff)
|
|
|
3966 afffile_T *aff;
|
|
|
3967 {
|
|
|
3968 hashtab_T *ht;
|
|
|
3969 hashitem_T *hi;
|
|
|
3970 int todo;
|
|
|
3971 affheader_T *ah;
|
|
300
|
3972 affentry_T *ae;
|
|
236
|
3973
|
|
|
3974 vim_free(aff->af_enc);
|
|
|
3975
|
|
339
|
3976 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
3977 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
3978 {
|
|
|
3979 todo = ht->ht_used;
|
|
|
3980 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
3981 {
|
|
|
3982 if (!HASHITEM_EMPTY(hi))
|
|
|
3983 {
|
|
|
3984 --todo;
|
|
|
3985 ah = HI2AH(hi);
|
|
300
|
3986 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
3987 vim_free(ae->ae_prog);
|
|
236
|
3988 }
|
|
|
3989 }
|
|
|
3990 if (ht == &aff->af_suff)
|
|
|
3991 break;
|
|
|
3992 }
|
|
300
|
3993
|
|
236
|
3994 hash_clear(&aff->af_pref);
|
|
|
3995 hash_clear(&aff->af_suff);
|
|
|
3996 }
|
|
|
3997
|
|
|
3998 /*
|
|
300
|
3999 * Read dictionary file "fname".
|
|
236
|
4000 * Returns OK or FAIL;
|
|
|
4001 */
|
|
|
4002 static int
|
|
300
|
4003 spell_read_dic(fname, spin, affile)
|
|
236
|
4004 char_u *fname;
|
|
300
|
4005 spellinfo_T *spin;
|
|
|
4006 afffile_T *affile;
|
|
236
|
4007 {
|
|
300
|
4008 hashtab_T ht;
|
|
236
|
4009 char_u line[MAXLINELEN];
|
|
300
|
4010 char_u *afflist;
|
|
339
|
4011 char_u *pfxlist;
|
|
300
|
4012 char_u *dw;
|
|
236
|
4013 char_u *pc;
|
|
|
4014 char_u *w;
|
|
|
4015 int l;
|
|
|
4016 hash_T hash;
|
|
|
4017 hashitem_T *hi;
|
|
|
4018 FILE *fd;
|
|
|
4019 int lnum = 1;
|
|
300
|
4020 int non_ascii = 0;
|
|
|
4021 int retval = OK;
|
|
|
4022 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
4023 int flags;
|
|
236
|
4024
|
|
300
|
4025 /*
|
|
|
4026 * Open the file.
|
|
|
4027 */
|
|
310
|
4028 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4029 if (fd == NULL)
|
|
|
4030 {
|
|
|
4031 EMSG2(_(e_notopen), fname);
|
|
|
4032 return FAIL;
|
|
|
4033 }
|
|
|
4034
|
|
300
|
4035 /* The hashtable is only used to detect duplicated words. */
|
|
|
4036 hash_init(&ht);
|
|
|
4037
|
|
334
|
4038 spin->si_foldwcount = 0;
|
|
|
4039 spin->si_keepwcount = 0;
|
|
|
4040
|
|
310
|
4041 if (spin->si_verbose || p_verbose > 2)
|
|
|
4042 {
|
|
|
4043 if (!spin->si_verbose)
|
|
|
4044 verbose_enter();
|
|
366
|
4045 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
4046 out_flush();
|
|
|
4047 if (!spin->si_verbose)
|
|
|
4048 verbose_leave();
|
|
|
4049 }
|
|
236
|
4050
|
|
|
4051 /* Read and ignore the first line: word count. */
|
|
|
4052 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
4053 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
4054 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
4055
|
|
|
4056 /*
|
|
|
4057 * Read all the lines in the file one by one.
|
|
|
4058 * The words are converted to 'encoding' here, before being added to
|
|
|
4059 * the hashtable.
|
|
|
4060 */
|
|
255
|
4061 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
4062 {
|
|
255
|
4063 line_breakcheck();
|
|
236
|
4064 ++lnum;
|
|
455
|
4065 if (line[0] == '#' || line[0] == '/')
|
|
351
|
4066 continue; /* comment line */
|
|
236
|
4067
|
|
300
|
4068 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
4069 * the word is kept to allow e.g., "et al.". */
|
|
236
|
4070 l = STRLEN(line);
|
|
|
4071 while (l > 0 && line[l - 1] <= ' ')
|
|
|
4072 --l;
|
|
|
4073 if (l == 0)
|
|
|
4074 continue; /* empty line */
|
|
|
4075 line[l] = NUL;
|
|
|
4076
|
|
|
4077 /* Find the optional affix names. */
|
|
300
|
4078 afflist = vim_strchr(line, '/');
|
|
|
4079 if (afflist != NULL)
|
|
|
4080 *afflist++ = NUL;
|
|
236
|
4081
|
|
300
|
4082 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4083 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4084 {
|
|
|
4085 ++non_ascii;
|
|
240
|
4086 continue;
|
|
300
|
4087 }
|
|
240
|
4088
|
|
310
|
4089 #ifdef FEAT_MBYTE
|
|
236
|
4090 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
4091 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4092 {
|
|
300
|
4093 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
4094 if (pc == NULL)
|
|
|
4095 {
|
|
|
4096 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4097 fname, lnum, line);
|
|
|
4098 continue;
|
|
|
4099 }
|
|
236
|
4100 w = pc;
|
|
|
4101 }
|
|
|
4102 else
|
|
310
|
4103 #endif
|
|
236
|
4104 {
|
|
|
4105 pc = NULL;
|
|
|
4106 w = line;
|
|
|
4107 }
|
|
|
4108
|
|
339
|
4109 /* This takes time, print a message now and then. */
|
|
|
4110 if (spin->si_verbose && (lnum & 0x3ff) == 0)
|
|
|
4111 {
|
|
|
4112 vim_snprintf((char *)message, sizeof(message),
|
|
|
4113 _("line %6d, word %6d - %s"),
|
|
|
4114 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
4115 msg_start();
|
|
|
4116 msg_puts_long_attr(message, 0);
|
|
|
4117 msg_clr_eos();
|
|
|
4118 msg_didout = FALSE;
|
|
|
4119 msg_col = 0;
|
|
|
4120 out_flush();
|
|
|
4121 }
|
|
|
4122
|
|
300
|
4123 /* Store the word in the hashtable to be able to find duplicates. */
|
|
|
4124 dw = (char_u *)getroom_save(&spin->si_blocks, w);
|
|
236
|
4125 if (dw == NULL)
|
|
300
|
4126 retval = FAIL;
|
|
|
4127 vim_free(pc);
|
|
|
4128 if (retval == FAIL)
|
|
236
|
4129 break;
|
|
|
4130
|
|
300
|
4131 hash = hash_hash(dw);
|
|
|
4132 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
4133 if (!HASHITEM_EMPTY(hi))
|
|
|
4134 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
4135 fname, lnum, dw);
|
|
236
|
4136 else
|
|
300
|
4137 hash_add_item(&ht, hi, dw, hash);
|
|
|
4138
|
|
307
|
4139 flags = 0;
|
|
339
|
4140 pfxlist = NULL;
|
|
307
|
4141 if (afflist != NULL)
|
|
|
4142 {
|
|
|
4143 /* Check for affix name that stands for keep-case word and stands
|
|
|
4144 * for rare word (if defined). */
|
|
310
|
4145 if (affile->af_kep != NUL
|
|
|
4146 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
389
|
4147 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
307
|
4148 if (affile->af_rar != NUL
|
|
|
4149 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
4150 flags |= WF_RARE;
|
|
346
|
4151 if (affile->af_bad != NUL
|
|
|
4152 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
4153 flags |= WF_BANNED;
|
|
339
|
4154
|
|
|
4155 if (affile->af_pfxpostpone)
|
|
|
4156 /* Need to store the list of prefix IDs with the word. */
|
|
|
4157 pfxlist = get_pfxlist(affile, afflist, &spin->si_blocks);
|
|
307
|
4158 }
|
|
|
4159
|
|
300
|
4160 /* Add the word to the word tree(s). */
|
|
339
|
4161 if (store_word(dw, spin, flags, spin->si_region, pfxlist) == FAIL)
|
|
300
|
4162 retval = FAIL;
|
|
236
|
4163
|
|
300
|
4164 if (afflist != NULL)
|
|
|
4165 {
|
|
|
4166 /* Find all matching suffixes and add the resulting words.
|
|
|
4167 * Additionally do matching prefixes that combine. */
|
|
339
|
4168 if (store_aff_word(dw, spin, afflist, affile,
|
|
307
|
4169 &affile->af_suff, &affile->af_pref,
|
|
339
|
4170 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
4171 retval = FAIL;
|
|
|
4172
|
|
|
4173 /* Find all matching prefixes and add the resulting words. */
|
|
339
|
4174 if (store_aff_word(dw, spin, afflist, affile,
|
|
|
4175 &affile->af_pref, NULL,
|
|
|
4176 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
4177 retval = FAIL;
|
|
|
4178 }
|
|
236
|
4179 }
|
|
|
4180
|
|
300
|
4181 if (spin->si_ascii && non_ascii > 0)
|
|
|
4182 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4183 non_ascii);
|
|
|
4184 hash_clear(&ht);
|
|
|
4185
|
|
236
|
4186 fclose(fd);
|
|
300
|
4187 return retval;
|
|
236
|
4188 }
|
|
|
4189
|
|
|
4190 /*
|
|
339
|
4191 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
4192 * Used for PFXPOSTPONE.
|
|
|
4193 * Returns a string allocated with getroom(). NULL when there are no prefixes
|
|
|
4194 * or when out of memory.
|
|
|
4195 */
|
|
|
4196 static char_u *
|
|
|
4197 get_pfxlist(affile, afflist, blp)
|
|
|
4198 afffile_T *affile;
|
|
|
4199 char_u *afflist;
|
|
|
4200 sblock_T **blp;
|
|
|
4201 {
|
|
|
4202 char_u *p;
|
|
|
4203 int cnt;
|
|
|
4204 int round;
|
|
|
4205 char_u *res = NULL;
|
|
|
4206 char_u key[2];
|
|
|
4207 hashitem_T *hi;
|
|
|
4208
|
|
|
4209 key[1] = NUL;
|
|
|
4210
|
|
|
4211 /* round 1: count the number of prefix IDs.
|
|
|
4212 * round 2: move prefix IDs to "res" */
|
|
|
4213 for (round = 1; round <= 2; ++round)
|
|
|
4214 {
|
|
|
4215 cnt = 0;
|
|
|
4216 for (p = afflist; *p != NUL; ++p)
|
|
|
4217 {
|
|
|
4218 key[0] = *p;
|
|
|
4219 hi = hash_find(&affile->af_pref, key);
|
|
|
4220 if (!HASHITEM_EMPTY(hi))
|
|
|
4221 {
|
|
|
4222 /* This is a prefix ID, use the new number. */
|
|
|
4223 if (round == 2)
|
|
|
4224 res[cnt] = HI2AH(hi)->ah_newID;
|
|
|
4225 ++cnt;
|
|
|
4226 }
|
|
|
4227 }
|
|
|
4228 if (round == 1 && cnt > 0)
|
|
447
|
4229 res = getroom(blp, cnt + 1, FALSE);
|
|
339
|
4230 if (res == NULL)
|
|
|
4231 break;
|
|
|
4232 }
|
|
|
4233
|
|
|
4234 if (res != NULL)
|
|
|
4235 res[cnt] = NUL;
|
|
|
4236 return res;
|
|
|
4237 }
|
|
|
4238
|
|
|
4239 /*
|
|
300
|
4240 * Apply affixes to a word and store the resulting words.
|
|
|
4241 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
4242 * prefixes or suffixes.
|
|
|
4243 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
4244 * the resulting words for combining affixes.
|
|
|
4245 *
|
|
|
4246 * Returns FAIL when out of memory.
|
|
236
|
4247 */
|
|
300
|
4248 static int
|
|
339
|
4249 store_aff_word(word, spin, afflist, affile, ht, xht, comb, flags, pfxlist)
|
|
300
|
4250 char_u *word; /* basic word start */
|
|
|
4251 spellinfo_T *spin; /* spell info */
|
|
|
4252 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
4253 afffile_T *affile;
|
|
300
|
4254 hashtab_T *ht;
|
|
|
4255 hashtab_T *xht;
|
|
|
4256 int comb; /* only use affixes that combine */
|
|
307
|
4257 int flags; /* flags for the word */
|
|
339
|
4258 char_u *pfxlist; /* list of prefix IDs */
|
|
236
|
4259 {
|
|
|
4260 int todo;
|
|
|
4261 hashitem_T *hi;
|
|
300
|
4262 affheader_T *ah;
|
|
|
4263 affentry_T *ae;
|
|
|
4264 regmatch_T regmatch;
|
|
|
4265 char_u newword[MAXWLEN];
|
|
|
4266 int retval = OK;
|
|
|
4267 int i;
|
|
|
4268 char_u *p;
|
|
366
|
4269 int use_flags;
|
|
390
|
4270 char_u *use_pfxlist;
|
|
455
|
4271 int c;
|
|
236
|
4272
|
|
300
|
4273 todo = ht->ht_used;
|
|
|
4274 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
4275 {
|
|
|
4276 if (!HASHITEM_EMPTY(hi))
|
|
|
4277 {
|
|
|
4278 --todo;
|
|
300
|
4279 ah = HI2AH(hi);
|
|
236
|
4280
|
|
300
|
4281 /* Check that the affix combines, if required, and that the word
|
|
|
4282 * supports this affix. */
|
|
455
|
4283 c = PTR2CHAR(ah->ah_key);
|
|
|
4284 if ((!comb || ah->ah_combine) && vim_strchr(afflist, c) != NULL)
|
|
236
|
4285 {
|
|
300
|
4286 /* Loop over all affix entries with this name. */
|
|
|
4287 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
4288 {
|
|
300
|
4289 /* Check the condition. It's not logical to match case
|
|
|
4290 * here, but it is required for compatibility with
|
|
339
|
4291 * Myspell.
|
|
|
4292 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
4293 * prefixes with a chop string. */
|
|
300
|
4294 regmatch.regprog = ae->ae_prog;
|
|
|
4295 regmatch.rm_ic = FALSE;
|
|
339
|
4296 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
4297 || ae->ae_chop != NULL)
|
|
|
4298 && (ae->ae_prog == NULL
|
|
|
4299 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
4300 {
|
|
|
4301 /* Match. Remove the chop and add the affix. */
|
|
|
4302 if (xht == NULL)
|
|
240
|
4303 {
|
|
300
|
4304 /* prefix: chop/add at the start of the word */
|
|
|
4305 if (ae->ae_add == NULL)
|
|
|
4306 *newword = NUL;
|
|
|
4307 else
|
|
|
4308 STRCPY(newword, ae->ae_add);
|
|
|
4309 p = word;
|
|
|
4310 if (ae->ae_chop != NULL)
|
|
310
|
4311 {
|
|
300
|
4312 /* Skip chop string. */
|
|
310
|
4313 #ifdef FEAT_MBYTE
|
|
|
4314 if (has_mbyte)
|
|
324
|
4315 {
|
|
310
|
4316 i = mb_charlen(ae->ae_chop);
|
|
324
|
4317 for ( ; i > 0; --i)
|
|
|
4318 mb_ptr_adv(p);
|
|
|
4319 }
|
|
310
|
4320 else
|
|
|
4321 #endif
|
|
324
|
4322 p += STRLEN(ae->ae_chop);
|
|
310
|
4323 }
|
|
300
|
4324 STRCAT(newword, p);
|
|
|
4325 }
|
|
|
4326 else
|
|
|
4327 {
|
|
|
4328 /* suffix: chop/add at the end of the word */
|
|
|
4329 STRCPY(newword, word);
|
|
|
4330 if (ae->ae_chop != NULL)
|
|
|
4331 {
|
|
|
4332 /* Remove chop string. */
|
|
|
4333 p = newword + STRLEN(newword);
|
|
376
|
4334 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
4335 for ( ; i > 0; --i)
|
|
300
|
4336 mb_ptr_back(newword, p);
|
|
|
4337 *p = NUL;
|
|
|
4338 }
|
|
|
4339 if (ae->ae_add != NULL)
|
|
|
4340 STRCAT(newword, ae->ae_add);
|
|
240
|
4341 }
|
|
|
4342
|
|
366
|
4343 /* Obey the "rare" flag of the affix. */
|
|
|
4344 if (ae->ae_rare)
|
|
|
4345 use_flags = flags | WF_RARE;
|
|
|
4346 else
|
|
|
4347 use_flags = flags;
|
|
390
|
4348 use_pfxlist = pfxlist;
|
|
|
4349
|
|
|
4350 /* When there are postponed prefixes... */
|
|
393
|
4351 if (spin->si_prefroot != NULL
|
|
|
4352 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
4353 {
|
|
|
4354 /* ... add a flag to indicate an affix was used. */
|
|
|
4355 use_flags |= WF_HAS_AFF;
|
|
|
4356
|
|
|
4357 /* ... don't use a prefix list if combining
|
|
|
4358 * affixes is not allowed */
|
|
|
4359 if (!ah->ah_combine || comb)
|
|
|
4360 use_pfxlist = NULL;
|
|
|
4361 }
|
|
366
|
4362
|
|
300
|
4363 /* Store the modified word. */
|
|
366
|
4364 if (store_word(newword, spin, use_flags,
|
|
390
|
4365 spin->si_region, use_pfxlist) == FAIL)
|
|
300
|
4366 retval = FAIL;
|
|
236
|
4367
|
|
300
|
4368 /* When added a suffix and combining is allowed also
|
|
|
4369 * try adding prefixes additionally. */
|
|
|
4370 if (xht != NULL && ah->ah_combine)
|
|
339
|
4371 if (store_aff_word(newword, spin, afflist, affile,
|
|
390
|
4372 xht, NULL, TRUE,
|
|
|
4373 use_flags, use_pfxlist) == FAIL)
|
|
300
|
4374 retval = FAIL;
|
|
236
|
4375 }
|
|
|
4376 }
|
|
|
4377 }
|
|
|
4378 }
|
|
|
4379 }
|
|
|
4380
|
|
|
4381 return retval;
|
|
|
4382 }
|
|
|
4383
|
|
|
4384 /*
|
|
300
|
4385 * Read a file with a list of words.
|
|
236
|
4386 */
|
|
|
4387 static int
|
|
300
|
4388 spell_read_wordfile(fname, spin)
|
|
|
4389 char_u *fname;
|
|
|
4390 spellinfo_T *spin;
|
|
236
|
4391 {
|
|
300
|
4392 FILE *fd;
|
|
|
4393 long lnum = 0;
|
|
|
4394 char_u rline[MAXLINELEN];
|
|
|
4395 char_u *line;
|
|
|
4396 char_u *pc = NULL;
|
|
381
|
4397 char_u *p;
|
|
300
|
4398 int l;
|
|
|
4399 int retval = OK;
|
|
|
4400 int did_word = FALSE;
|
|
|
4401 int non_ascii = 0;
|
|
307
|
4402 int flags;
|
|
316
|
4403 int regionmask;
|
|
236
|
4404
|
|
300
|
4405 /*
|
|
|
4406 * Open the file.
|
|
|
4407 */
|
|
310
|
4408 fd = mch_fopen((char *)fname, "r");
|
|
300
|
4409 if (fd == NULL)
|
|
236
|
4410 {
|
|
300
|
4411 EMSG2(_(e_notopen), fname);
|
|
|
4412 return FAIL;
|
|
236
|
4413 }
|
|
|
4414
|
|
310
|
4415 if (spin->si_verbose || p_verbose > 2)
|
|
|
4416 {
|
|
|
4417 if (!spin->si_verbose)
|
|
|
4418 verbose_enter();
|
|
366
|
4419 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
4420 out_flush();
|
|
|
4421 if (!spin->si_verbose)
|
|
|
4422 verbose_leave();
|
|
|
4423 }
|
|
300
|
4424
|
|
|
4425 /*
|
|
|
4426 * Read all the lines in the file one by one.
|
|
|
4427 */
|
|
|
4428 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
4429 {
|
|
|
4430 line_breakcheck();
|
|
|
4431 ++lnum;
|
|
|
4432
|
|
|
4433 /* Skip comment lines. */
|
|
|
4434 if (*rline == '#')
|
|
|
4435 continue;
|
|
|
4436
|
|
|
4437 /* Remove CR, LF and white space from the end. */
|
|
|
4438 l = STRLEN(rline);
|
|
|
4439 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
4440 --l;
|
|
|
4441 if (l == 0)
|
|
|
4442 continue; /* empty or blank line */
|
|
|
4443 rline[l] = NUL;
|
|
|
4444
|
|
|
4445 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
4446 vim_free(pc);
|
|
310
|
4447 #ifdef FEAT_MBYTE
|
|
300
|
4448 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
4449 {
|
|
|
4450 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
4451 if (pc == NULL)
|
|
|
4452 {
|
|
|
4453 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4454 fname, lnum, rline);
|
|
|
4455 continue;
|
|
|
4456 }
|
|
|
4457 line = pc;
|
|
|
4458 }
|
|
|
4459 else
|
|
310
|
4460 #endif
|
|
300
|
4461 {
|
|
|
4462 pc = NULL;
|
|
|
4463 line = rline;
|
|
|
4464 }
|
|
|
4465
|
|
307
|
4466 if (*line == '/')
|
|
300
|
4467 {
|
|
307
|
4468 ++line;
|
|
|
4469 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
4470 {
|
|
|
4471 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
4472 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
4473 fname, lnum, line - 1);
|
|
300
|
4474 else if (did_word)
|
|
316
|
4475 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
4476 fname, lnum, line - 1);
|
|
300
|
4477 else
|
|
|
4478 {
|
|
310
|
4479 #ifdef FEAT_MBYTE
|
|
|
4480 char_u *enc;
|
|
|
4481
|
|
300
|
4482 /* Setup for conversion to 'encoding'. */
|
|
316
|
4483 line += 10;
|
|
|
4484 enc = enc_canonize(line);
|
|
300
|
4485 if (enc != NULL && !spin->si_ascii
|
|
|
4486 && convert_setup(&spin->si_conv, enc,
|
|
|
4487 p_enc) == FAIL)
|
|
|
4488 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
4489 fname, line, p_enc);
|
|
300
|
4490 vim_free(enc);
|
|
375
|
4491 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4492 #else
|
|
|
4493 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4494 #endif
|
|
300
|
4495 }
|
|
307
|
4496 continue;
|
|
300
|
4497 }
|
|
307
|
4498
|
|
316
|
4499 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
4500 {
|
|
|
4501 if (spin->si_region_count > 1)
|
|
|
4502 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
4503 fname, lnum, line);
|
|
|
4504 else
|
|
|
4505 {
|
|
|
4506 line += 8;
|
|
|
4507 if (STRLEN(line) > 16)
|
|
|
4508 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
4509 fname, lnum, line);
|
|
|
4510 else
|
|
|
4511 {
|
|
|
4512 spin->si_region_count = STRLEN(line) / 2;
|
|
|
4513 STRCPY(spin->si_region_name, line);
|
|
389
|
4514
|
|
|
4515 /* Adjust the mask for a word valid in all regions. */
|
|
|
4516 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
4517 }
|
|
|
4518 }
|
|
|
4519 continue;
|
|
|
4520 }
|
|
|
4521
|
|
381
|
4522 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
4523 fname, lnum, line - 1);
|
|
|
4524 continue;
|
|
|
4525 }
|
|
|
4526
|
|
|
4527 flags = 0;
|
|
|
4528 regionmask = spin->si_region;
|
|
|
4529
|
|
|
4530 /* Check for flags and region after a slash. */
|
|
|
4531 p = vim_strchr(line, '/');
|
|
|
4532 if (p != NULL)
|
|
|
4533 {
|
|
|
4534 *p++ = NUL;
|
|
|
4535 while (*p != NUL)
|
|
|
4536 {
|
|
|
4537 if (*p == '=') /* keep-case word */
|
|
389
|
4538 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
4539 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
4540 flags |= WF_BANNED;
|
|
|
4541 else if (*p == '?') /* Rare word. */
|
|
|
4542 flags |= WF_RARE;
|
|
|
4543 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
4544 {
|
|
381
|
4545 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
4546 regionmask = 0;
|
|
|
4547 flags |= WF_REGION;
|
|
|
4548
|
|
|
4549 l = *p - '0';
|
|
316
|
4550 if (l > spin->si_region_count)
|
|
|
4551 {
|
|
|
4552 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
4553 fname, lnum, p);
|
|
316
|
4554 break;
|
|
|
4555 }
|
|
|
4556 regionmask |= 1 << (l - 1);
|
|
|
4557 }
|
|
381
|
4558 else
|
|
|
4559 {
|
|
|
4560 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
4561 fname, lnum, p);
|
|
|
4562 break;
|
|
|
4563 }
|
|
|
4564 ++p;
|
|
307
|
4565 }
|
|
300
|
4566 }
|
|
|
4567
|
|
|
4568 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4569 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4570 {
|
|
|
4571 ++non_ascii;
|
|
|
4572 continue;
|
|
|
4573 }
|
|
|
4574
|
|
|
4575 /* Normal word: store it. */
|
|
339
|
4576 if (store_word(line, spin, flags, regionmask, NULL) == FAIL)
|
|
300
|
4577 {
|
|
|
4578 retval = FAIL;
|
|
|
4579 break;
|
|
|
4580 }
|
|
|
4581 did_word = TRUE;
|
|
|
4582 }
|
|
|
4583
|
|
|
4584 vim_free(pc);
|
|
|
4585 fclose(fd);
|
|
|
4586
|
|
310
|
4587 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
4588 {
|
|
|
4589 if (p_verbose > 2)
|
|
|
4590 verbose_enter();
|
|
300
|
4591 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4592 non_ascii);
|
|
310
|
4593 if (p_verbose > 2)
|
|
|
4594 verbose_leave();
|
|
|
4595 }
|
|
300
|
4596 return retval;
|
|
236
|
4597 }
|
|
|
4598
|
|
|
4599 /*
|
|
300
|
4600 * Get part of an sblock_T, "len" bytes long.
|
|
|
4601 * This avoids calling free() for every little struct we use.
|
|
|
4602 * The memory is cleared to all zeros.
|
|
|
4603 * Returns NULL when out of memory.
|
|
|
4604 */
|
|
|
4605 static void *
|
|
447
|
4606 getroom(blp, len, align)
|
|
300
|
4607 sblock_T **blp;
|
|
447
|
4608 size_t len; /* length needed */
|
|
|
4609 int align; /* align for pointer */
|
|
300
|
4610 {
|
|
|
4611 char_u *p;
|
|
|
4612 sblock_T *bl = *blp;
|
|
|
4613
|
|
447
|
4614 if (align && bl != NULL)
|
|
|
4615 /* Round size up for alignment. On some systems structures need to be
|
|
|
4616 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
4617 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
4618 & ~(sizeof(char *) - 1);
|
|
|
4619
|
|
300
|
4620 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
4621 {
|
|
|
4622 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
4623 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
4624 if (bl == NULL)
|
|
|
4625 return NULL;
|
|
|
4626 bl->sb_next = *blp;
|
|
|
4627 *blp = bl;
|
|
|
4628 bl->sb_used = 0;
|
|
|
4629 }
|
|
|
4630
|
|
|
4631 p = bl->sb_data + bl->sb_used;
|
|
|
4632 bl->sb_used += len;
|
|
|
4633
|
|
|
4634 return p;
|
|
|
4635 }
|
|
|
4636
|
|
|
4637 /*
|
|
|
4638 * Make a copy of a string into memory allocated with getroom().
|
|
|
4639 */
|
|
|
4640 static char_u *
|
|
|
4641 getroom_save(blp, s)
|
|
|
4642 sblock_T **blp;
|
|
|
4643 char_u *s;
|
|
|
4644 {
|
|
|
4645 char_u *sc;
|
|
|
4646
|
|
447
|
4647 sc = (char_u *)getroom(blp, STRLEN(s) + 1, FALSE);
|
|
300
|
4648 if (sc != NULL)
|
|
|
4649 STRCPY(sc, s);
|
|
|
4650 return sc;
|
|
|
4651 }
|
|
|
4652
|
|
|
4653
|
|
|
4654 /*
|
|
|
4655 * Free the list of allocated sblock_T.
|
|
236
|
4656 */
|
|
|
4657 static void
|
|
300
|
4658 free_blocks(bl)
|
|
|
4659 sblock_T *bl;
|
|
236
|
4660 {
|
|
300
|
4661 sblock_T *next;
|
|
236
|
4662
|
|
300
|
4663 while (bl != NULL)
|
|
236
|
4664 {
|
|
300
|
4665 next = bl->sb_next;
|
|
|
4666 vim_free(bl);
|
|
|
4667 bl = next;
|
|
236
|
4668 }
|
|
|
4669 }
|
|
|
4670
|
|
|
4671 /*
|
|
300
|
4672 * Allocate the root of a word tree.
|
|
236
|
4673 */
|
|
300
|
4674 static wordnode_T *
|
|
|
4675 wordtree_alloc(blp)
|
|
|
4676 sblock_T **blp;
|
|
236
|
4677 {
|
|
447
|
4678 return (wordnode_T *)getroom(blp, sizeof(wordnode_T), TRUE);
|
|
236
|
4679 }
|
|
|
4680
|
|
|
4681 /*
|
|
300
|
4682 * Store a word in the tree(s).
|
|
390
|
4683 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
4684 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
4685 * used to find suggestions.
|
|
300
|
4686 * For a keep-case word also store it in the keep-case tree.
|
|
390
|
4687 * When "pfxlist" is not NULL store the word for each postponed prefix ID.
|
|
236
|
4688 */
|
|
|
4689 static int
|
|
339
|
4690 store_word(word, spin, flags, region, pfxlist)
|
|
300
|
4691 char_u *word;
|
|
|
4692 spellinfo_T *spin;
|
|
307
|
4693 int flags; /* extra flags, WF_BANNED */
|
|
316
|
4694 int region; /* supported region(s) */
|
|
339
|
4695 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
236
|
4696 {
|
|
300
|
4697 int len = STRLEN(word);
|
|
|
4698 int ct = captype(word, word + len);
|
|
|
4699 char_u foldword[MAXWLEN];
|
|
339
|
4700 int res = OK;
|
|
|
4701 char_u *p;
|
|
236
|
4702
|
|
323
|
4703 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
4704 for (p = pfxlist; res == OK; ++p)
|
|
|
4705 {
|
|
|
4706 res = tree_add_word(foldword, spin->si_foldroot, ct | flags,
|
|
469
|
4707 region, p == NULL ? 0 : *p, spin);
|
|
339
|
4708 if (p == NULL || *p == NUL)
|
|
|
4709 break;
|
|
|
4710 }
|
|
334
|
4711 ++spin->si_foldwcount;
|
|
307
|
4712
|
|
|
4713 if (res == OK && (ct == WF_KEEPCAP || flags & WF_KEEPCAP))
|
|
334
|
4714 {
|
|
339
|
4715 for (p = pfxlist; res == OK; ++p)
|
|
|
4716 {
|
|
|
4717 res = tree_add_word(word, spin->si_keeproot, flags,
|
|
469
|
4718 region, p == NULL ? 0 : *p, spin);
|
|
339
|
4719 if (p == NULL || *p == NUL)
|
|
|
4720 break;
|
|
|
4721 }
|
|
334
|
4722 ++spin->si_keepwcount;
|
|
|
4723 }
|
|
300
|
4724 return res;
|
|
236
|
4725 }
|
|
|
4726
|
|
|
4727 /*
|
|
300
|
4728 * Add word "word" to a word tree at "root".
|
|
366
|
4729 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
4730 * "rare" and "region" is the condition nr.
|
|
255
|
4731 * Returns FAIL when out of memory.
|
|
236
|
4732 */
|
|
255
|
4733 static int
|
|
469
|
4734 tree_add_word(word, root, flags, region, prefixID, spin)
|
|
300
|
4735 char_u *word;
|
|
|
4736 wordnode_T *root;
|
|
|
4737 int flags;
|
|
|
4738 int region;
|
|
339
|
4739 int prefixID;
|
|
469
|
4740 spellinfo_T *spin;
|
|
236
|
4741 {
|
|
300
|
4742 wordnode_T *node = root;
|
|
|
4743 wordnode_T *np;
|
|
469
|
4744 wordnode_T *copyp, **copyprev;
|
|
300
|
4745 wordnode_T **prev = NULL;
|
|
|
4746 int i;
|
|
469
|
4747 sblock_T **blp = &spin->si_blocks;
|
|
255
|
4748
|
|
300
|
4749 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
4750 for (i = 0; ; ++i)
|
|
255
|
4751 {
|
|
469
|
4752 /* When there is more than one reference to this node we need to make
|
|
|
4753 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
4754 * (we don't optimize for a partly shared list of siblings). */
|
|
|
4755 if (node != NULL && node->wn_refs > 1)
|
|
|
4756 {
|
|
|
4757 --node->wn_refs;
|
|
|
4758 copyprev = prev;
|
|
|
4759 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
4760 {
|
|
|
4761 /* Allocate a new node and copy the info. */
|
|
|
4762 np = get_wordnode(blp);
|
|
|
4763 if (np == NULL)
|
|
|
4764 return FAIL;
|
|
|
4765 np->wn_child = copyp->wn_child;
|
|
|
4766 if (np->wn_child != NULL)
|
|
|
4767 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
4768 np->wn_byte = copyp->wn_byte;
|
|
|
4769 if (np->wn_byte == NUL)
|
|
|
4770 {
|
|
|
4771 np->wn_flags = copyp->wn_flags;
|
|
|
4772 np->wn_region = copyp->wn_region;
|
|
|
4773 np->wn_prefixID = copyp->wn_prefixID;
|
|
|
4774 }
|
|
|
4775
|
|
|
4776 /* Link the new node in the list, there will be one ref. */
|
|
|
4777 np->wn_refs = 1;
|
|
|
4778 *copyprev = np;
|
|
|
4779 copyprev = &np->wn_sibling;
|
|
|
4780
|
|
|
4781 /* Let "node" point to the head of the copied list. */
|
|
|
4782 if (copyp == node)
|
|
|
4783 node = np;
|
|
|
4784 }
|
|
|
4785 }
|
|
|
4786
|
|
300
|
4787 /* Look for the sibling that has the same character. They are sorted
|
|
|
4788 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
4789 * higher byte value. For zero bytes (end of word) the sorting is
|
|
469
|
4790 * done on flags and then on prefixID. */
|
|
339
|
4791 while (node != NULL
|
|
|
4792 && (node->wn_byte < word[i]
|
|
|
4793 || (node->wn_byte == NUL
|
|
|
4794 && (flags < 0
|
|
|
4795 ? node->wn_prefixID < prefixID
|
|
390
|
4796 : node->wn_flags < (flags & WN_MASK)
|
|
|
4797 || (node->wn_flags == (flags & WN_MASK)
|
|
339
|
4798 && node->wn_prefixID < prefixID)))))
|
|
236
|
4799 {
|
|
300
|
4800 prev = &node->wn_sibling;
|
|
|
4801 node = *prev;
|
|
236
|
4802 }
|
|
339
|
4803 if (node == NULL
|
|
|
4804 || node->wn_byte != word[i]
|
|
|
4805 || (word[i] == NUL
|
|
|
4806 && (flags < 0
|
|
390
|
4807 || node->wn_flags != (flags & WN_MASK)
|
|
339
|
4808 || node->wn_prefixID != prefixID)))
|
|
255
|
4809 {
|
|
300
|
4810 /* Allocate a new node. */
|
|
469
|
4811 np = get_wordnode(blp);
|
|
300
|
4812 if (np == NULL)
|
|
|
4813 return FAIL;
|
|
|
4814 np->wn_byte = word[i];
|
|
469
|
4815
|
|
|
4816 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
4817 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
4818 * make ref count of sibling one (matters when inserting in front
|
|
|
4819 * of the list of siblings). */
|
|
|
4820 if (node == NULL)
|
|
|
4821 np->wn_refs = 1;
|
|
|
4822 else
|
|
|
4823 {
|
|
|
4824 np->wn_refs = node->wn_refs;
|
|
|
4825 node->wn_refs = 1;
|
|
|
4826 }
|
|
300
|
4827 *prev = np;
|
|
|
4828 np->wn_sibling = node;
|
|
|
4829 node = np;
|
|
255
|
4830 }
|
|
300
|
4831
|
|
|
4832 if (word[i] == NUL)
|
|
|
4833 {
|
|
|
4834 node->wn_flags = flags;
|
|
|
4835 node->wn_region |= region;
|
|
339
|
4836 node->wn_prefixID = prefixID;
|
|
300
|
4837 break;
|
|
|
4838 }
|
|
|
4839 prev = &node->wn_child;
|
|
|
4840 node = *prev;
|
|
255
|
4841 }
|
|
469
|
4842 #ifdef SPELL_PRINTTREE
|
|
|
4843 smsg("Added \"%s\"", word);
|
|
|
4844 spell_print_tree(root->wn_sibling);
|
|
|
4845 #endif
|
|
|
4846
|
|
|
4847 /*
|
|
|
4848 * Every so many words compress the tree, so that we don't use too much
|
|
|
4849 * memory.
|
|
|
4850 */
|
|
|
4851 if (++words_added >= SPELL_COMPRESS_CNT)
|
|
|
4852 {
|
|
|
4853 words_added = 0;
|
|
|
4854
|
|
|
4855 msg_start();
|
|
|
4856 msg_puts((char_u *)_(msg_compressing));
|
|
|
4857 msg_clr_eos();
|
|
|
4858 msg_didout = FALSE;
|
|
|
4859 msg_col = 0;
|
|
|
4860 out_flush();
|
|
|
4861 wordtree_compress(root->wn_sibling, spin);
|
|
|
4862 }
|
|
255
|
4863
|
|
|
4864 return OK;
|
|
236
|
4865 }
|
|
|
4866
|
|
469
|
4867 /* We keep a list of nodes that have been freed during compression. They are
|
|
|
4868 * re-used when adding a new node. The "wn_child" fields links them. */
|
|
|
4869 static wordnode_T *first_free_node = NULL;
|
|
|
4870 #ifdef SPELL_PRINTTREE
|
|
|
4871 static int wordnode_nr = 0;
|
|
|
4872 #endif
|
|
|
4873
|
|
|
4874 /*
|
|
|
4875 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
4876 * allocate a new one.
|
|
|
4877 */
|
|
|
4878 static wordnode_T *
|
|
|
4879 get_wordnode(blp)
|
|
|
4880 sblock_T **blp;
|
|
|
4881 {
|
|
|
4882 wordnode_T *n;
|
|
|
4883
|
|
|
4884 if (first_free_node == NULL)
|
|
|
4885 n = (wordnode_T *)getroom(blp, sizeof(wordnode_T), TRUE);
|
|
|
4886 else
|
|
|
4887 {
|
|
|
4888 n = first_free_node;
|
|
|
4889 first_free_node = n->wn_child;
|
|
|
4890 vim_memset(n, 0, sizeof(wordnode_T));
|
|
|
4891 }
|
|
|
4892 #ifdef SPELL_PRINTTREE
|
|
|
4893 n->wn_nr = ++wordnode_nr;
|
|
|
4894 #endif
|
|
|
4895 return n;
|
|
|
4896 }
|
|
|
4897
|
|
|
4898 /*
|
|
|
4899 * Free a wordnode_T for re-use later.
|
|
|
4900 */
|
|
|
4901 static void
|
|
|
4902 free_wordnode(n)
|
|
|
4903 wordnode_T *n;
|
|
|
4904 {
|
|
|
4905 n->wn_child = first_free_node;
|
|
|
4906 first_free_node = n;
|
|
|
4907 }
|
|
|
4908
|
|
236
|
4909 /*
|
|
300
|
4910 * Compress a tree: find tails that are identical and can be shared.
|
|
|
4911 */
|
|
|
4912 static void
|
|
310
|
4913 wordtree_compress(root, spin)
|
|
300
|
4914 wordnode_T *root;
|
|
310
|
4915 spellinfo_T *spin;
|
|
300
|
4916 {
|
|
|
4917 hashtab_T ht;
|
|
|
4918 int n;
|
|
|
4919 int tot = 0;
|
|
469
|
4920 int perc;
|
|
300
|
4921
|
|
|
4922 if (root != NULL)
|
|
|
4923 {
|
|
|
4924 hash_init(&ht);
|
|
|
4925 n = node_compress(root, &ht, &tot);
|
|
469
|
4926
|
|
|
4927 #ifndef SPELL_PRINTTREE
|
|
310
|
4928 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
4929 #endif
|
|
310
|
4930 {
|
|
|
4931 if (!spin->si_verbose)
|
|
|
4932 verbose_enter();
|
|
469
|
4933 if (tot > 1000000)
|
|
|
4934 perc = (tot - n) / (tot / 100);
|
|
|
4935 else
|
|
|
4936 perc = (tot - n) * 100 / tot;
|
|
310
|
4937 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
4938 n, tot, perc);
|
|
310
|
4939 if (p_verbose > 2)
|
|
|
4940 verbose_leave();
|
|
|
4941 }
|
|
469
|
4942 #ifdef SPELL_PRINTTREE
|
|
|
4943 spell_print_tree(root);
|
|
|
4944 #endif
|
|
300
|
4945 hash_clear(&ht);
|
|
|
4946 }
|
|
|
4947 }
|
|
|
4948
|
|
|
4949 /*
|
|
|
4950 * Compress a node, its siblings and its children, depth first.
|
|
|
4951 * Returns the number of compressed nodes.
|
|
236
|
4952 */
|
|
255
|
4953 static int
|
|
300
|
4954 node_compress(node, ht, tot)
|
|
|
4955 wordnode_T *node;
|
|
|
4956 hashtab_T *ht;
|
|
|
4957 int *tot; /* total count of nodes before compressing,
|
|
|
4958 incremented while going through the tree */
|
|
236
|
4959 {
|
|
300
|
4960 wordnode_T *np;
|
|
|
4961 wordnode_T *tp;
|
|
|
4962 wordnode_T *child;
|
|
|
4963 hash_T hash;
|
|
236
|
4964 hashitem_T *hi;
|
|
300
|
4965 int len = 0;
|
|
|
4966 unsigned nr, n;
|
|
|
4967 int compressed = 0;
|
|
236
|
4968
|
|
300
|
4969 /*
|
|
|
4970 * Go through the list of siblings. Compress each child and then try
|
|
|
4971 * finding an identical child to replace it.
|
|
|
4972 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
4973 * but the whole list of siblings, of which the node is the first.
|
|
|
4974 */
|
|
|
4975 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4976 {
|
|
300
|
4977 ++len;
|
|
|
4978 if ((child = np->wn_child) != NULL)
|
|
|
4979 {
|
|
346
|
4980 /* Compress the child. This fills hashkey. */
|
|
300
|
4981 compressed += node_compress(child, ht, tot);
|
|
|
4982
|
|
|
4983 /* Try to find an identical child. */
|
|
346
|
4984 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
4985 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
4986 tp = NULL;
|
|
|
4987 if (!HASHITEM_EMPTY(hi))
|
|
|
4988 {
|
|
|
4989 /* There are children with an identical hash value. Now check
|
|
|
4990 * if there is one that is really identical. */
|
|
346
|
4991 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
4992 if (node_equal(child, tp))
|
|
|
4993 {
|
|
|
4994 /* Found one! Now use that child in place of the
|
|
469
|
4995 * current one. This means the current child and all
|
|
|
4996 * its siblings is unlinked from the tree. */
|
|
|
4997 ++tp->wn_refs;
|
|
|
4998 --child->wn_refs;
|
|
|
4999 if (child->wn_refs == 0)
|
|
|
5000 for (; child != NULL; child = child->wn_sibling)
|
|
|
5001 {
|
|
|
5002 if (child->wn_child != NULL)
|
|
|
5003 --child->wn_child->wn_refs;
|
|
|
5004 free_wordnode(child);
|
|
|
5005 }
|
|
300
|
5006 np->wn_child = tp;
|
|
|
5007 ++compressed;
|
|
|
5008 break;
|
|
|
5009 }
|
|
|
5010 if (tp == NULL)
|
|
|
5011 {
|
|
|
5012 /* No other child with this hash value equals the child of
|
|
|
5013 * the node, add it to the linked list after the first
|
|
|
5014 * item. */
|
|
|
5015 tp = HI2WN(hi);
|
|
346
|
5016 child->wn_u2.next = tp->wn_u2.next;
|
|
|
5017 tp->wn_u2.next = child;
|
|
300
|
5018 }
|
|
|
5019 }
|
|
|
5020 else
|
|
|
5021 /* No other child has this hash value, add it to the
|
|
|
5022 * hashtable. */
|
|
346
|
5023 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
5024 }
|
|
236
|
5025 }
|
|
300
|
5026 *tot += len;
|
|
|
5027
|
|
|
5028 /*
|
|
|
5029 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
5030 * find a lookalike node. This must be done after compressing the sibling
|
|
|
5031 * list, otherwise the hash key would become invalid by the compression.
|
|
|
5032 */
|
|
346
|
5033 node->wn_u1.hashkey[0] = len;
|
|
300
|
5034 nr = 0;
|
|
|
5035 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5036 {
|
|
300
|
5037 if (np->wn_byte == NUL)
|
|
339
|
5038 /* end node: use wn_flags, wn_region and wn_prefixID */
|
|
|
5039 n = np->wn_flags + (np->wn_region << 8) + (np->wn_prefixID << 16);
|
|
300
|
5040 else
|
|
|
5041 /* byte node: use the byte value and the child pointer */
|
|
|
5042 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
5043 nr = nr * 101 + n;
|
|
236
|
5044 }
|
|
300
|
5045
|
|
|
5046 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
5047 n = nr & 0xff;
|
|
346
|
5048 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
5049 n = (nr >> 8) & 0xff;
|
|
346
|
5050 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
5051 n = (nr >> 16) & 0xff;
|
|
346
|
5052 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
5053 n = (nr >> 24) & 0xff;
|
|
346
|
5054 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
5055 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
5056
|
|
|
5057 return compressed;
|
|
|
5058 }
|
|
|
5059
|
|
|
5060 /*
|
|
|
5061 * Return TRUE when two nodes have identical siblings and children.
|
|
|
5062 */
|
|
|
5063 static int
|
|
|
5064 node_equal(n1, n2)
|
|
|
5065 wordnode_T *n1;
|
|
|
5066 wordnode_T *n2;
|
|
|
5067 {
|
|
|
5068 wordnode_T *p1;
|
|
|
5069 wordnode_T *p2;
|
|
|
5070
|
|
|
5071 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
5072 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
5073 if (p1->wn_byte != p2->wn_byte
|
|
|
5074 || (p1->wn_byte == NUL
|
|
|
5075 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
5076 || p1->wn_region != p2->wn_region
|
|
|
5077 || p1->wn_prefixID != p2->wn_prefixID)
|
|
300
|
5078 : (p1->wn_child != p2->wn_child)))
|
|
|
5079 break;
|
|
|
5080
|
|
|
5081 return p1 == NULL && p2 == NULL;
|
|
236
|
5082 }
|
|
|
5083
|
|
|
5084 /*
|
|
|
5085 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
5086 */
|
|
255
|
5087 void
|
|
236
|
5088 put_bytes(fd, nr, len)
|
|
|
5089 FILE *fd;
|
|
|
5090 long_u nr;
|
|
|
5091 int len;
|
|
|
5092 {
|
|
|
5093 int i;
|
|
|
5094
|
|
|
5095 for (i = len - 1; i >= 0; --i)
|
|
|
5096 putc((int)(nr >> (i * 8)), fd);
|
|
|
5097 }
|
|
|
5098
|
|
323
|
5099 static int
|
|
|
5100 #ifdef __BORLANDC__
|
|
|
5101 _RTLENTRYF
|
|
|
5102 #endif
|
|
|
5103 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
5104
|
|
|
5105 /*
|
|
|
5106 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
5107 */
|
|
|
5108 static int
|
|
|
5109 #ifdef __BORLANDC__
|
|
|
5110 _RTLENTRYF
|
|
|
5111 #endif
|
|
|
5112 rep_compare(s1, s2)
|
|
|
5113 const void *s1;
|
|
|
5114 const void *s2;
|
|
|
5115 {
|
|
|
5116 fromto_T *p1 = (fromto_T *)s1;
|
|
|
5117 fromto_T *p2 = (fromto_T *)s2;
|
|
|
5118
|
|
|
5119 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
5120 }
|
|
|
5121
|
|
236
|
5122 /*
|
|
|
5123 * Write the Vim spell file "fname".
|
|
|
5124 */
|
|
|
5125 static void
|
|
316
|
5126 write_vim_spell(fname, spin)
|
|
236
|
5127 char_u *fname;
|
|
300
|
5128 spellinfo_T *spin;
|
|
236
|
5129 {
|
|
300
|
5130 FILE *fd;
|
|
|
5131 int regionmask;
|
|
236
|
5132 int round;
|
|
300
|
5133 wordnode_T *tree;
|
|
|
5134 int nodecount;
|
|
323
|
5135 int i;
|
|
|
5136 int l;
|
|
|
5137 garray_T *gap;
|
|
|
5138 fromto_T *ftp;
|
|
|
5139 char_u *p;
|
|
|
5140 int rr;
|
|
236
|
5141
|
|
310
|
5142 fd = mch_fopen((char *)fname, "w");
|
|
300
|
5143 if (fd == NULL)
|
|
236
|
5144 {
|
|
|
5145 EMSG2(_(e_notopen), fname);
|
|
|
5146 return;
|
|
|
5147 }
|
|
|
5148
|
|
255
|
5149 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
339
|
5150 * <charflagslen> <charflags>
|
|
|
5151 * <fcharslen> <fchars>
|
|
366
|
5152 * <midwordlen> <midword>
|
|
339
|
5153 * <prefcondcnt> <prefcond> ... */
|
|
300
|
5154
|
|
|
5155 /* <fileID> */
|
|
|
5156 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
5157 EMSG(_(e_write));
|
|
236
|
5158
|
|
|
5159 /* write the region names if there is more than one */
|
|
316
|
5160 if (spin->si_region_count > 1)
|
|
236
|
5161 {
|
|
316
|
5162 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
|
5163 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
|
5164 (size_t)1, fd);
|
|
|
5165 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
5166 }
|
|
|
5167 else
|
|
|
5168 {
|
|
300
|
5169 putc(0, fd);
|
|
|
5170 regionmask = 0;
|
|
236
|
5171 }
|
|
|
5172
|
|
323
|
5173 /*
|
|
|
5174 * Write the table with character flags and table for case folding.
|
|
260
|
5175 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
|
5176 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
5177 * 'encoding'.
|
|
|
5178 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
5179 * the table (avoids that it conflicts). File is shorter too.
|
|
|
5180 */
|
|
|
5181 if (spin->si_ascii || spin->si_add)
|
|
260
|
5182 {
|
|
300
|
5183 putc(0, fd);
|
|
|
5184 putc(0, fd);
|
|
|
5185 putc(0, fd);
|
|
260
|
5186 }
|
|
|
5187 else
|
|
300
|
5188 write_spell_chartab(fd);
|
|
255
|
5189
|
|
366
|
5190
|
|
|
5191 if (spin->si_midword == NULL)
|
|
|
5192 put_bytes(fd, 0L, 2); /* <midwordlen> */
|
|
|
5193 else
|
|
|
5194 {
|
|
|
5195 i = STRLEN(spin->si_midword);
|
|
|
5196 put_bytes(fd, (long_u)i, 2); /* <midwordlen> */
|
|
|
5197 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
5198 }
|
|
|
5199
|
|
|
5200
|
|
339
|
5201 /* Write the prefix conditions. */
|
|
|
5202 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
5203
|
|
366
|
5204 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
5205 * <salflags> <salcount> <sal> ...
|
|
|
5206 * <maplen> <mapstr> */
|
|
|
5207
|
|
323
|
5208 /* Sort the REP items. */
|
|
|
5209 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
5210 sizeof(fromto_T), rep_compare);
|
|
|
5211
|
|
375
|
5212 /* round 1: REP items
|
|
|
5213 * round 2: SAL items (unless SOFO is used) */
|
|
323
|
5214 for (round = 1; round <= 2; ++round)
|
|
|
5215 {
|
|
|
5216 if (round == 1)
|
|
|
5217 gap = &spin->si_rep;
|
|
|
5218 else
|
|
|
5219 {
|
|
|
5220 gap = &spin->si_sal;
|
|
|
5221
|
|
|
5222 i = 0;
|
|
|
5223 if (spin->si_followup)
|
|
|
5224 i |= SAL_F0LLOWUP;
|
|
|
5225 if (spin->si_collapse)
|
|
|
5226 i |= SAL_COLLAPSE;
|
|
|
5227 if (spin->si_rem_accents)
|
|
|
5228 i |= SAL_REM_ACCENTS;
|
|
375
|
5229 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5230 i |= SAL_SOFO;
|
|
323
|
5231 putc(i, fd); /* <salflags> */
|
|
375
|
5232 if (i & SAL_SOFO)
|
|
|
5233 break;
|
|
323
|
5234 }
|
|
|
5235
|
|
|
5236 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
5237 for (i = 0; i < gap->ga_len; ++i)
|
|
|
5238 {
|
|
|
5239 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
5240 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
5241 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
5242 for (rr = 1; rr <= 2; ++rr)
|
|
|
5243 {
|
|
|
5244 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
5245 l = STRLEN(p);
|
|
|
5246 putc(l, fd);
|
|
|
5247 fwrite(p, l, (size_t)1, fd);
|
|
|
5248 }
|
|
|
5249 }
|
|
|
5250 }
|
|
|
5251
|
|
375
|
5252 /* SOFOFROM and SOFOTO */
|
|
|
5253 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5254 {
|
|
|
5255 put_bytes(fd, 1L, 2); /* <salcount> */
|
|
|
5256
|
|
|
5257 l = STRLEN(spin->si_sofofr);
|
|
|
5258 put_bytes(fd, (long_u)l, 2); /* <salfromlen> */
|
|
|
5259 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <salfrom> */
|
|
|
5260
|
|
|
5261 l = STRLEN(spin->si_sofoto);
|
|
|
5262 put_bytes(fd, (long_u)l, 2); /* <saltolen> */
|
|
|
5263 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <salto> */
|
|
|
5264 }
|
|
|
5265
|
|
323
|
5266 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
|
5267 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
|
5268 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
|
5269 (size_t)1, fd);
|
|
302
|
5270
|
|
236
|
5271 /*
|
|
339
|
5272 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
5273 */
|
|
323
|
5274 spin->si_memtot = 0;
|
|
339
|
5275 for (round = 1; round <= 3; ++round)
|
|
236
|
5276 {
|
|
339
|
5277 if (round == 1)
|
|
469
|
5278 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
5279 else if (round == 2)
|
|
469
|
5280 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
5281 else
|
|
469
|
5282 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
5283
|
|
346
|
5284 /* Clear the index and wnode fields in the tree. */
|
|
|
5285 clear_node(tree);
|
|
|
5286
|
|
300
|
5287 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
5288 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
5289 * nodes. */
|
|
346
|
5290 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
5291
|
|
300
|
5292 /* number of nodes in 4 bytes */
|
|
|
5293 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
5294 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
5295
|
|
300
|
5296 /* Write the nodes. */
|
|
346
|
5297 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
5298 }
|
|
|
5299
|
|
300
|
5300 fclose(fd);
|
|
236
|
5301 }
|
|
|
5302
|
|
|
5303 /*
|
|
346
|
5304 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
5305 * children. This is needed because they are a union with other items to save
|
|
|
5306 * space.
|
|
|
5307 */
|
|
|
5308 static void
|
|
|
5309 clear_node(node)
|
|
|
5310 wordnode_T *node;
|
|
|
5311 {
|
|
|
5312 wordnode_T *np;
|
|
|
5313
|
|
|
5314 if (node != NULL)
|
|
|
5315 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5316 {
|
|
|
5317 np->wn_u1.index = 0;
|
|
|
5318 np->wn_u2.wnode = NULL;
|
|
|
5319
|
|
|
5320 if (np->wn_byte != NUL)
|
|
|
5321 clear_node(np->wn_child);
|
|
|
5322 }
|
|
|
5323 }
|
|
|
5324
|
|
|
5325
|
|
|
5326 /*
|
|
300
|
5327 * Dump a word tree at node "node".
|
|
|
5328 *
|
|
|
5329 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
5330 * byte recursively write the children.
|
|
|
5331 *
|
|
|
5332 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
5333 * are made about the indexes (so that we don't have to write them in the
|
|
|
5334 * file).
|
|
236
|
5335 *
|
|
300
|
5336 * Returns the number of nodes used.
|
|
236
|
5337 */
|
|
300
|
5338 static int
|
|
346
|
5339 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
5340 FILE *fd; /* NULL when only counting */
|
|
300
|
5341 wordnode_T *node;
|
|
|
5342 int index;
|
|
|
5343 int regionmask;
|
|
339
|
5344 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
5345 {
|
|
300
|
5346 int newindex = index;
|
|
|
5347 int siblingcount = 0;
|
|
|
5348 wordnode_T *np;
|
|
236
|
5349 int flags;
|
|
300
|
5350
|
|
|
5351 /* If "node" is zero the tree is empty. */
|
|
|
5352 if (node == NULL)
|
|
|
5353 return 0;
|
|
|
5354
|
|
|
5355 /* Store the index where this node is written. */
|
|
346
|
5356 node->wn_u1.index = index;
|
|
236
|
5357
|
|
300
|
5358 /* Count the number of siblings. */
|
|
|
5359 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5360 ++siblingcount;
|
|
236
|
5361
|
|
300
|
5362 /* Write the sibling count. */
|
|
|
5363 if (fd != NULL)
|
|
|
5364 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
5365
|
|
300
|
5366 /* Write each sibling byte and optionally extra info. */
|
|
|
5367 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5368 {
|
|
300
|
5369 if (np->wn_byte == 0)
|
|
|
5370 {
|
|
|
5371 if (fd != NULL)
|
|
|
5372 {
|
|
339
|
5373 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
5374 if (prefixtree)
|
|
300
|
5375 {
|
|
339
|
5376 /* In PREFIXTREE write the required prefixID and the
|
|
390
|
5377 * associated condition nr (stored in wn_region). The
|
|
|
5378 * byte value is misused to store the "rare" and "not
|
|
|
5379 * combining" flags */
|
|
455
|
5380 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
5381 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5382 else
|
|
|
5383 {
|
|
390
|
5384 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
5385 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
5386 }
|
|
339
|
5387 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
5388 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
5389 }
|
|
|
5390 else
|
|
|
5391 {
|
|
339
|
5392 /* For word trees we write the flag/region items. */
|
|
|
5393 flags = np->wn_flags;
|
|
|
5394 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
5395 flags |= WF_REGION;
|
|
|
5396 if (np->wn_prefixID != 0)
|
|
|
5397 flags |= WF_PFX;
|
|
|
5398 if (flags == 0)
|
|
|
5399 {
|
|
|
5400 /* word without flags or region */
|
|
|
5401 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5402 }
|
|
|
5403 else
|
|
|
5404 {
|
|
390
|
5405 if (np->wn_flags >= 0x100)
|
|
|
5406 {
|
|
|
5407 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
5408 putc(flags, fd); /* <flags> */
|
|
|
5409 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
5410 }
|
|
|
5411 else
|
|
|
5412 {
|
|
|
5413 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
5414 putc(flags, fd); /* <flags> */
|
|
|
5415 }
|
|
339
|
5416 if (flags & WF_REGION)
|
|
|
5417 putc(np->wn_region, fd); /* <region> */
|
|
|
5418 if (flags & WF_PFX)
|
|
|
5419 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
5420 }
|
|
300
|
5421 }
|
|
|
5422 }
|
|
|
5423 }
|
|
|
5424 else
|
|
|
5425 {
|
|
346
|
5426 if (np->wn_child->wn_u1.index != 0
|
|
|
5427 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
5428 {
|
|
|
5429 /* The child is written elsewhere, write the reference. */
|
|
|
5430 if (fd != NULL)
|
|
|
5431 {
|
|
|
5432 putc(BY_INDEX, fd); /* <byte> */
|
|
|
5433 /* <nodeidx> */
|
|
346
|
5434 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
5435 }
|
|
|
5436 }
|
|
346
|
5437 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
5438 /* We will write the child below and give it an index. */
|
|
346
|
5439 np->wn_child->wn_u2.wnode = node;
|
|
236
|
5440
|
|
300
|
5441 if (fd != NULL)
|
|
|
5442 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
5443 {
|
|
|
5444 EMSG(_(e_write));
|
|
|
5445 return 0;
|
|
|
5446 }
|
|
|
5447 }
|
|
236
|
5448 }
|
|
|
5449
|
|
300
|
5450 /* Space used in the array when reading: one for each sibling and one for
|
|
|
5451 * the count. */
|
|
|
5452 newindex += siblingcount + 1;
|
|
249
|
5453
|
|
300
|
5454 /* Recursively dump the children of each sibling. */
|
|
|
5455 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
5456 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
5457 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
5458 prefixtree);
|
|
249
|
5459
|
|
300
|
5460 return newindex;
|
|
236
|
5461 }
|
|
|
5462
|
|
|
5463
|
|
|
5464 /*
|
|
310
|
5465 * ":mkspell [-ascii] outfile infile ..."
|
|
|
5466 * ":mkspell [-ascii] addfile"
|
|
236
|
5467 */
|
|
|
5468 void
|
|
|
5469 ex_mkspell(eap)
|
|
|
5470 exarg_T *eap;
|
|
|
5471 {
|
|
|
5472 int fcount;
|
|
|
5473 char_u **fnames;
|
|
310
|
5474 char_u *arg = eap->arg;
|
|
|
5475 int ascii = FALSE;
|
|
|
5476
|
|
|
5477 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
5478 {
|
|
|
5479 ascii = TRUE;
|
|
|
5480 arg = skipwhite(arg + 6);
|
|
|
5481 }
|
|
|
5482
|
|
|
5483 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
5484 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
5485 {
|
|
323
|
5486 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
5487 FreeWild(fcount, fnames);
|
|
|
5488 }
|
|
|
5489 }
|
|
|
5490
|
|
|
5491 /*
|
|
|
5492 * Create a Vim spell file from one or more word lists.
|
|
|
5493 * "fnames[0]" is the output file name.
|
|
|
5494 * "fnames[fcount - 1]" is the last input file name.
|
|
|
5495 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
5496 * and ".spl" is appended to make the output file name.
|
|
|
5497 */
|
|
|
5498 static void
|
|
323
|
5499 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
5500 int fcount;
|
|
|
5501 char_u **fnames;
|
|
|
5502 int ascii; /* -ascii argument given */
|
|
|
5503 int overwrite; /* overwrite existing output file */
|
|
323
|
5504 int added_word; /* invoked through "zg" */
|
|
310
|
5505 {
|
|
236
|
5506 char_u fname[MAXPATHL];
|
|
|
5507 char_u wfname[MAXPATHL];
|
|
310
|
5508 char_u **innames;
|
|
|
5509 int incount;
|
|
236
|
5510 afffile_T *(afile[8]);
|
|
|
5511 int i;
|
|
|
5512 int len;
|
|
|
5513 struct stat st;
|
|
255
|
5514 int error = FALSE;
|
|
300
|
5515 spellinfo_T spin;
|
|
|
5516
|
|
|
5517 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
5518 spin.si_verbose = !added_word;
|
|
310
|
5519 spin.si_ascii = ascii;
|
|
323
|
5520 spin.si_followup = TRUE;
|
|
|
5521 spin.si_rem_accents = TRUE;
|
|
|
5522 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
5523 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
5524 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
5525 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
469
|
5526 first_free_node = NULL;
|
|
|
5527 #ifdef SPELL_PRINTTREE
|
|
|
5528 wordnode_nr = 0;
|
|
|
5529 #endif
|
|
|
5530 words_added = 0;
|
|
310
|
5531
|
|
|
5532 /* default: fnames[0] is output file, following are input files */
|
|
|
5533 innames = &fnames[1];
|
|
|
5534 incount = fcount - 1;
|
|
|
5535
|
|
|
5536 if (fcount >= 1)
|
|
240
|
5537 {
|
|
310
|
5538 len = STRLEN(fnames[0]);
|
|
|
5539 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
5540 {
|
|
|
5541 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
5542 * "path/en.latin1.add.spl". */
|
|
|
5543 innames = &fnames[0];
|
|
|
5544 incount = 1;
|
|
|
5545 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
5546 }
|
|
366
|
5547 else if (fcount == 1)
|
|
|
5548 {
|
|
|
5549 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
5550 innames = &fnames[0];
|
|
|
5551 incount = 1;
|
|
|
5552 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5553 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5554 }
|
|
310
|
5555 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
5556 {
|
|
|
5557 /* Name ends in ".spl", use as the file name. */
|
|
323
|
5558 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
5559 }
|
|
|
5560 else
|
|
|
5561 /* Name should be language, make the file name from it. */
|
|
|
5562 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5563 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5564
|
|
|
5565 /* Check for .ascii.spl. */
|
|
|
5566 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
5567 spin.si_ascii = TRUE;
|
|
|
5568
|
|
|
5569 /* Check for .add.spl. */
|
|
|
5570 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
5571 spin.si_add = TRUE;
|
|
240
|
5572 }
|
|
|
5573
|
|
310
|
5574 if (incount <= 0)
|
|
236
|
5575 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
5576 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
5577 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
5578 else if (incount > 8)
|
|
236
|
5579 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
5580 else
|
|
|
5581 {
|
|
|
5582 /* Check for overwriting before doing things that may take a lot of
|
|
|
5583 * time. */
|
|
310
|
5584 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
5585 {
|
|
|
5586 EMSG(_(e_exists));
|
|
310
|
5587 return;
|
|
236
|
5588 }
|
|
310
|
5589 if (mch_isdir(wfname))
|
|
236
|
5590 {
|
|
310
|
5591 EMSG2(_(e_isadir2), wfname);
|
|
|
5592 return;
|
|
236
|
5593 }
|
|
|
5594
|
|
|
5595 /*
|
|
|
5596 * Init the aff and dic pointers.
|
|
|
5597 * Get the region names if there are more than 2 arguments.
|
|
|
5598 */
|
|
310
|
5599 for (i = 0; i < incount; ++i)
|
|
236
|
5600 {
|
|
310
|
5601 afile[i] = NULL;
|
|
300
|
5602
|
|
316
|
5603 if (incount > 1)
|
|
236
|
5604 {
|
|
310
|
5605 len = STRLEN(innames[i]);
|
|
|
5606 if (STRLEN(gettail(innames[i])) < 5
|
|
|
5607 || innames[i][len - 3] != '_')
|
|
236
|
5608 {
|
|
310
|
5609 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
5610 return;
|
|
236
|
5611 }
|
|
316
|
5612 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
5613 spin.si_region_name[i * 2 + 1] =
|
|
|
5614 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
5615 }
|
|
|
5616 }
|
|
316
|
5617 spin.si_region_count = incount;
|
|
236
|
5618
|
|
300
|
5619 spin.si_foldroot = wordtree_alloc(&spin.si_blocks);
|
|
|
5620 spin.si_keeproot = wordtree_alloc(&spin.si_blocks);
|
|
339
|
5621 spin.si_prefroot = wordtree_alloc(&spin.si_blocks);
|
|
|
5622 if (spin.si_foldroot == NULL
|
|
|
5623 || spin.si_keeproot == NULL
|
|
|
5624 || spin.si_prefroot == NULL)
|
|
300
|
5625 {
|
|
469
|
5626 free_blocks(spin.si_blocks);
|
|
310
|
5627 return;
|
|
300
|
5628 }
|
|
|
5629
|
|
351
|
5630 /* When not producing a .add.spl file clear the character table when
|
|
|
5631 * we encounter one in the .aff file. This means we dump the current
|
|
|
5632 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
5633 * better than guessing the contents, the table will match a
|
|
|
5634 * previously loaded spell file. */
|
|
|
5635 if (!spin.si_add)
|
|
|
5636 spin.si_clear_chartab = TRUE;
|
|
|
5637
|
|
236
|
5638 /*
|
|
|
5639 * Read all the .aff and .dic files.
|
|
|
5640 * Text is converted to 'encoding'.
|
|
300
|
5641 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
5642 */
|
|
310
|
5643 for (i = 0; i < incount && !error; ++i)
|
|
236
|
5644 {
|
|
300
|
5645 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
5646 spin.si_region = 1 << i;
|
|
|
5647
|
|
|
5648 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
5649 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
5650 {
|
|
|
5651 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
5652 * "SET" line. */
|
|
310
|
5653 afile[i] = spell_read_aff(fname, &spin);
|
|
|
5654 if (afile[i] == NULL)
|
|
300
|
5655 error = TRUE;
|
|
|
5656 else
|
|
|
5657 {
|
|
|
5658 /* Read the .dic file and store the words in the trees. */
|
|
|
5659 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
5660 innames[i]);
|
|
|
5661 if (spell_read_dic(fname, &spin, afile[i]) == FAIL)
|
|
300
|
5662 error = TRUE;
|
|
|
5663 }
|
|
|
5664 }
|
|
|
5665 else
|
|
|
5666 {
|
|
|
5667 /* No .aff file, try reading the file as a word list. Store
|
|
|
5668 * the words in the trees. */
|
|
310
|
5669 if (spell_read_wordfile(innames[i], &spin) == FAIL)
|
|
300
|
5670 error = TRUE;
|
|
|
5671 }
|
|
236
|
5672
|
|
310
|
5673 #ifdef FEAT_MBYTE
|
|
236
|
5674 /* Free any conversion stuff. */
|
|
300
|
5675 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
5676 #endif
|
|
236
|
5677 }
|
|
|
5678
|
|
300
|
5679 if (!error)
|
|
236
|
5680 {
|
|
|
5681 /*
|
|
300
|
5682 * Combine tails in the tree.
|
|
236
|
5683 */
|
|
323
|
5684 if (!added_word || p_verbose > 2)
|
|
310
|
5685 {
|
|
323
|
5686 if (added_word)
|
|
310
|
5687 verbose_enter();
|
|
469
|
5688 MSG(_(msg_compressing));
|
|
310
|
5689 out_flush();
|
|
323
|
5690 if (added_word)
|
|
310
|
5691 verbose_leave();
|
|
|
5692 }
|
|
469
|
5693 wordtree_compress(spin.si_foldroot->wn_sibling, &spin);
|
|
|
5694 wordtree_compress(spin.si_keeproot->wn_sibling, &spin);
|
|
|
5695 wordtree_compress(spin.si_prefroot->wn_sibling, &spin);
|
|
236
|
5696 }
|
|
|
5697
|
|
300
|
5698 if (!error)
|
|
|
5699 {
|
|
|
5700 /*
|
|
|
5701 * Write the info in the spell file.
|
|
|
5702 */
|
|
323
|
5703 if (!added_word || p_verbose > 2)
|
|
310
|
5704 {
|
|
323
|
5705 if (added_word)
|
|
310
|
5706 verbose_enter();
|
|
366
|
5707 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
5708 out_flush();
|
|
323
|
5709 if (added_word)
|
|
310
|
5710 verbose_leave();
|
|
|
5711 }
|
|
|
5712
|
|
316
|
5713 write_vim_spell(wfname, &spin);
|
|
310
|
5714
|
|
323
|
5715 if (!added_word || p_verbose > 2)
|
|
310
|
5716 {
|
|
323
|
5717 if (added_word)
|
|
310
|
5718 verbose_enter();
|
|
|
5719 MSG(_("Done!"));
|
|
|
5720 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
5721 spin.si_memtot);
|
|
310
|
5722 out_flush();
|
|
323
|
5723 if (added_word)
|
|
310
|
5724 verbose_leave();
|
|
|
5725 }
|
|
|
5726
|
|
|
5727 /* If the file is loaded need to reload it. */
|
|
323
|
5728 spell_reload_one(wfname, added_word);
|
|
300
|
5729 }
|
|
|
5730
|
|
|
5731 /* Free the allocated memory. */
|
|
323
|
5732 ga_clear(&spin.si_rep);
|
|
|
5733 ga_clear(&spin.si_sal);
|
|
|
5734 ga_clear(&spin.si_map);
|
|
339
|
5735 ga_clear(&spin.si_prefcond);
|
|
366
|
5736 vim_free(spin.si_midword);
|
|
375
|
5737 vim_free(spin.si_sofofr);
|
|
|
5738 vim_free(spin.si_sofoto);
|
|
300
|
5739
|
|
|
5740 /* Free the .aff file structures. */
|
|
310
|
5741 for (i = 0; i < incount; ++i)
|
|
|
5742 if (afile[i] != NULL)
|
|
|
5743 spell_free_aff(afile[i]);
|
|
339
|
5744
|
|
|
5745 /* Free all the bits and pieces at once. */
|
|
|
5746 free_blocks(spin.si_blocks);
|
|
236
|
5747 }
|
|
310
|
5748 }
|
|
|
5749
|
|
|
5750
|
|
|
5751 /*
|
|
385
|
5752 * ":[count]spellgood {word}"
|
|
|
5753 * ":[count]spellwrong {word}"
|
|
310
|
5754 */
|
|
|
5755 void
|
|
|
5756 ex_spell(eap)
|
|
|
5757 exarg_T *eap;
|
|
|
5758 {
|
|
381
|
5759 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
5760 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
5761 }
|
|
|
5762
|
|
310
|
5763 /*
|
|
|
5764 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
5765 */
|
|
|
5766 void
|
|
385
|
5767 spell_add_word(word, len, bad, index)
|
|
310
|
5768 char_u *word;
|
|
|
5769 int len;
|
|
|
5770 int bad;
|
|
385
|
5771 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
5772 'spellfile' */
|
|
310
|
5773 {
|
|
|
5774 FILE *fd;
|
|
385
|
5775 buf_T *buf = NULL;
|
|
351
|
5776 int new_spf = FALSE;
|
|
|
5777 struct stat st;
|
|
381
|
5778 char_u *fname;
|
|
385
|
5779 char_u fnamebuf[MAXPATHL];
|
|
|
5780 char_u line[MAXWLEN * 2];
|
|
|
5781 long fpos, fpos_next = 0;
|
|
|
5782 int i;
|
|
|
5783 char_u *spf;
|
|
|
5784
|
|
|
5785 if (index == 0) /* use internal wordlist */
|
|
|
5786 {
|
|
|
5787 if (int_wordlist == NULL)
|
|
|
5788 {
|
|
|
5789 int_wordlist = vim_tempname('s');
|
|
|
5790 if (int_wordlist == NULL)
|
|
381
|
5791 return;
|
|
|
5792 }
|
|
385
|
5793 fname = int_wordlist;
|
|
381
|
5794 }
|
|
310
|
5795 else
|
|
|
5796 {
|
|
381
|
5797 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
5798 if (*curbuf->b_p_spf == NUL)
|
|
|
5799 {
|
|
|
5800 init_spellfile();
|
|
|
5801 new_spf = TRUE;
|
|
|
5802 }
|
|
|
5803
|
|
|
5804 if (*curbuf->b_p_spf == NUL)
|
|
|
5805 {
|
|
|
5806 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
5807 return;
|
|
|
5808 }
|
|
|
5809
|
|
385
|
5810 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
5811 {
|
|
|
5812 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
5813 if (i == index)
|
|
|
5814 break;
|
|
|
5815 if (*spf == NUL)
|
|
|
5816 {
|
|
|
5817 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
5818 return;
|
|
|
5819 }
|
|
|
5820 }
|
|
|
5821
|
|
310
|
5822 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
5823 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
5824 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
5825 buf = NULL;
|
|
|
5826 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
5827 {
|
|
310
|
5828 EMSG(_(e_bufloaded));
|
|
381
|
5829 return;
|
|
|
5830 }
|
|
|
5831
|
|
385
|
5832 fname = fnamebuf;
|
|
|
5833 }
|
|
|
5834
|
|
|
5835 if (bad)
|
|
|
5836 {
|
|
|
5837 /* When the word also appears as good word we need to remove that one,
|
|
|
5838 * since its flags sort before the one with WF_BANNED. */
|
|
|
5839 fd = mch_fopen((char *)fname, "r");
|
|
|
5840 if (fd != NULL)
|
|
|
5841 {
|
|
|
5842 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
5843 {
|
|
|
5844 fpos = fpos_next;
|
|
|
5845 fpos_next = ftell(fd);
|
|
|
5846 if (STRNCMP(word, line, len) == 0
|
|
|
5847 && (line[len] == '/' || line[len] < ' '))
|
|
|
5848 {
|
|
|
5849 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
5850 * the start of the line. Mixing reading and writing
|
|
|
5851 * doesn't work for all systems, close the file first. */
|
|
|
5852 fclose(fd);
|
|
|
5853 fd = mch_fopen((char *)fname, "r+");
|
|
|
5854 if (fd == NULL)
|
|
|
5855 break;
|
|
|
5856 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
5857 fputc('#', fd);
|
|
|
5858 fseek(fd, fpos_next, SEEK_SET);
|
|
|
5859 }
|
|
|
5860 }
|
|
|
5861 fclose(fd);
|
|
|
5862 }
|
|
381
|
5863 }
|
|
|
5864
|
|
|
5865 fd = mch_fopen((char *)fname, "a");
|
|
|
5866 if (fd == NULL && new_spf)
|
|
|
5867 {
|
|
|
5868 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
5869 * We may need to create the "spell" directory first. We already
|
|
|
5870 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
5871 STRCPY(NameBuff, fname);
|
|
|
5872 *gettail_sep(NameBuff) = NUL;
|
|
|
5873 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
5874 {
|
|
|
5875 /* The directory doesn't exist. Try creating it and opening the
|
|
|
5876 * file again. */
|
|
|
5877 vim_mkdir(NameBuff, 0755);
|
|
|
5878 fd = mch_fopen((char *)fname, "a");
|
|
|
5879 }
|
|
|
5880 }
|
|
|
5881
|
|
|
5882 if (fd == NULL)
|
|
|
5883 EMSG2(_(e_notopen), fname);
|
|
|
5884 else
|
|
|
5885 {
|
|
|
5886 if (bad)
|
|
|
5887 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
5888 else
|
|
381
|
5889 fprintf(fd, "%.*s\n", len, word);
|
|
|
5890 fclose(fd);
|
|
|
5891
|
|
|
5892 /* Update the .add.spl file. */
|
|
|
5893 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
5894
|
|
|
5895 /* If the .add file is edited somewhere, reload it. */
|
|
|
5896 if (buf != NULL)
|
|
|
5897 buf_reload(buf);
|
|
|
5898
|
|
|
5899 redraw_all_later(NOT_VALID);
|
|
310
|
5900 }
|
|
|
5901 }
|
|
|
5902
|
|
|
5903 /*
|
|
|
5904 * Initialize 'spellfile' for the current buffer.
|
|
|
5905 */
|
|
|
5906 static void
|
|
|
5907 init_spellfile()
|
|
|
5908 {
|
|
|
5909 char_u buf[MAXPATHL];
|
|
|
5910 int l;
|
|
|
5911 slang_T *sl;
|
|
|
5912 char_u *rtp;
|
|
351
|
5913 char_u *lend;
|
|
310
|
5914
|
|
|
5915 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
5916 {
|
|
351
|
5917 /* Find the end of the language name. Exclude the region. */
|
|
|
5918 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
5919 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
5920 ;
|
|
|
5921
|
|
|
5922 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
5923 * are allowed to write. */
|
|
310
|
5924 rtp = p_rtp;
|
|
|
5925 while (*rtp != NUL)
|
|
|
5926 {
|
|
|
5927 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
5928 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
5929 if (filewritable(buf) == 2)
|
|
|
5930 {
|
|
316
|
5931 /* Use the first language name from 'spelllang' and the
|
|
|
5932 * encoding used in the first loaded .spl file. */
|
|
310
|
5933 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
5934 l = STRLEN(buf);
|
|
|
5935 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
5936 "/spell/%.*s.%s.add",
|
|
351
|
5937 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
5938 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
5939 ? (char_u *)"ascii" : spell_enc());
|
|
|
5940 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
5941 break;
|
|
|
5942 }
|
|
|
5943 }
|
|
|
5944 }
|
|
|
5945 }
|
|
236
|
5946
|
|
300
|
5947
|
|
307
|
5948 /*
|
|
|
5949 * Init the chartab used for spelling for ASCII.
|
|
|
5950 * EBCDIC is not supported!
|
|
|
5951 */
|
|
|
5952 static void
|
|
|
5953 clear_spell_chartab(sp)
|
|
|
5954 spelltab_T *sp;
|
|
|
5955 {
|
|
324
|
5956 int i;
|
|
307
|
5957
|
|
|
5958 /* Init everything to FALSE. */
|
|
|
5959 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
5960 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
5961 for (i = 0; i < 256; ++i)
|
|
324
|
5962 {
|
|
307
|
5963 sp->st_fold[i] = i;
|
|
324
|
5964 sp->st_upper[i] = i;
|
|
|
5965 }
|
|
307
|
5966
|
|
|
5967 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
5968 * that is done separately. */
|
|
|
5969 for (i = '0'; i <= '9'; ++i)
|
|
|
5970 sp->st_isw[i] = TRUE;
|
|
|
5971 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
5972 {
|
|
|
5973 sp->st_isw[i] = TRUE;
|
|
|
5974 sp->st_isu[i] = TRUE;
|
|
|
5975 sp->st_fold[i] = i + 0x20;
|
|
|
5976 }
|
|
|
5977 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
5978 {
|
|
307
|
5979 sp->st_isw[i] = TRUE;
|
|
324
|
5980 sp->st_upper[i] = i - 0x20;
|
|
|
5981 }
|
|
307
|
5982 }
|
|
|
5983
|
|
|
5984 /*
|
|
|
5985 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
5986 * Called once while starting up and when 'encoding' changes.
|
|
|
5987 * The default is to use isalpha(), but the spell file should define the word
|
|
|
5988 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
5989 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
5990 */
|
|
|
5991 void
|
|
|
5992 init_spell_chartab()
|
|
|
5993 {
|
|
|
5994 int i;
|
|
|
5995
|
|
|
5996 did_set_spelltab = FALSE;
|
|
|
5997 clear_spell_chartab(&spelltab);
|
|
|
5998 #ifdef FEAT_MBYTE
|
|
|
5999 if (enc_dbcs)
|
|
|
6000 {
|
|
|
6001 /* DBCS: assume double-wide characters are word characters. */
|
|
|
6002 for (i = 128; i <= 255; ++i)
|
|
|
6003 if (MB_BYTE2LEN(i) == 2)
|
|
|
6004 spelltab.st_isw[i] = TRUE;
|
|
|
6005 }
|
|
324
|
6006 else if (enc_utf8)
|
|
|
6007 {
|
|
|
6008 for (i = 128; i < 256; ++i)
|
|
|
6009 {
|
|
|
6010 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
6011 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
6012 spelltab.st_fold[i] = utf_fold(i);
|
|
|
6013 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
6014 }
|
|
|
6015 }
|
|
307
|
6016 else
|
|
|
6017 #endif
|
|
|
6018 {
|
|
324
|
6019 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
6020 for (i = 128; i < 256; ++i)
|
|
|
6021 {
|
|
|
6022 if (MB_ISUPPER(i))
|
|
|
6023 {
|
|
324
|
6024 spelltab.st_isw[i] = TRUE;
|
|
307
|
6025 spelltab.st_isu[i] = TRUE;
|
|
|
6026 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
6027 }
|
|
324
|
6028 else if (MB_ISLOWER(i))
|
|
|
6029 {
|
|
|
6030 spelltab.st_isw[i] = TRUE;
|
|
|
6031 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
6032 }
|
|
307
|
6033 }
|
|
|
6034 }
|
|
|
6035 }
|
|
|
6036
|
|
|
6037 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
6038 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
6039
|
|
|
6040 /*
|
|
|
6041 * Set the spell character tables from strings in the affix file.
|
|
|
6042 */
|
|
|
6043 static int
|
|
|
6044 set_spell_chartab(fol, low, upp)
|
|
|
6045 char_u *fol;
|
|
|
6046 char_u *low;
|
|
|
6047 char_u *upp;
|
|
|
6048 {
|
|
|
6049 /* We build the new tables here first, so that we can compare with the
|
|
|
6050 * previous one. */
|
|
|
6051 spelltab_T new_st;
|
|
|
6052 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
6053 int f, l, u;
|
|
|
6054
|
|
|
6055 clear_spell_chartab(&new_st);
|
|
|
6056
|
|
|
6057 while (*pf != NUL)
|
|
|
6058 {
|
|
|
6059 if (*pl == NUL || *pu == NUL)
|
|
|
6060 {
|
|
|
6061 EMSG(_(e_affform));
|
|
|
6062 return FAIL;
|
|
|
6063 }
|
|
|
6064 #ifdef FEAT_MBYTE
|
|
|
6065 f = mb_ptr2char_adv(&pf);
|
|
|
6066 l = mb_ptr2char_adv(&pl);
|
|
|
6067 u = mb_ptr2char_adv(&pu);
|
|
|
6068 #else
|
|
|
6069 f = *pf++;
|
|
|
6070 l = *pl++;
|
|
|
6071 u = *pu++;
|
|
|
6072 #endif
|
|
|
6073 /* Every character that appears is a word character. */
|
|
|
6074 if (f < 256)
|
|
|
6075 new_st.st_isw[f] = TRUE;
|
|
|
6076 if (l < 256)
|
|
|
6077 new_st.st_isw[l] = TRUE;
|
|
|
6078 if (u < 256)
|
|
|
6079 new_st.st_isw[u] = TRUE;
|
|
|
6080
|
|
|
6081 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
6082 * case-folding */
|
|
|
6083 if (l < 256 && l != f)
|
|
|
6084 {
|
|
|
6085 if (f >= 256)
|
|
|
6086 {
|
|
|
6087 EMSG(_(e_affrange));
|
|
|
6088 return FAIL;
|
|
|
6089 }
|
|
|
6090 new_st.st_fold[l] = f;
|
|
|
6091 }
|
|
|
6092
|
|
|
6093 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
6094 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
6095 * "FOL" . */
|
|
307
|
6096 if (u < 256 && u != f)
|
|
|
6097 {
|
|
|
6098 if (f >= 256)
|
|
|
6099 {
|
|
|
6100 EMSG(_(e_affrange));
|
|
|
6101 return FAIL;
|
|
|
6102 }
|
|
|
6103 new_st.st_fold[u] = f;
|
|
|
6104 new_st.st_isu[u] = TRUE;
|
|
324
|
6105 new_st.st_upper[f] = u;
|
|
307
|
6106 }
|
|
|
6107 }
|
|
|
6108
|
|
|
6109 if (*pl != NUL || *pu != NUL)
|
|
|
6110 {
|
|
|
6111 EMSG(_(e_affform));
|
|
|
6112 return FAIL;
|
|
|
6113 }
|
|
|
6114
|
|
|
6115 return set_spell_finish(&new_st);
|
|
|
6116 }
|
|
|
6117
|
|
|
6118 /*
|
|
|
6119 * Set the spell character tables from strings in the .spl file.
|
|
|
6120 */
|
|
|
6121 static int
|
|
389
|
6122 set_spell_charflags(flags, cnt, fol)
|
|
307
|
6123 char_u *flags;
|
|
389
|
6124 int cnt; /* length of "flags" */
|
|
|
6125 char_u *fol;
|
|
307
|
6126 {
|
|
|
6127 /* We build the new tables here first, so that we can compare with the
|
|
|
6128 * previous one. */
|
|
|
6129 spelltab_T new_st;
|
|
|
6130 int i;
|
|
389
|
6131 char_u *p = fol;
|
|
324
|
6132 int c;
|
|
307
|
6133
|
|
|
6134 clear_spell_chartab(&new_st);
|
|
|
6135
|
|
389
|
6136 for (i = 0; i < 128; ++i)
|
|
|
6137 {
|
|
|
6138 if (i < cnt)
|
|
|
6139 {
|
|
|
6140 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
6141 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
6142 }
|
|
|
6143
|
|
|
6144 if (*p != NUL)
|
|
|
6145 {
|
|
307
|
6146 #ifdef FEAT_MBYTE
|
|
389
|
6147 c = mb_ptr2char_adv(&p);
|
|
307
|
6148 #else
|
|
389
|
6149 c = *p++;
|
|
307
|
6150 #endif
|
|
389
|
6151 new_st.st_fold[i + 128] = c;
|
|
|
6152 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
6153 new_st.st_upper[c] = i + 128;
|
|
|
6154 }
|
|
307
|
6155 }
|
|
|
6156
|
|
|
6157 return set_spell_finish(&new_st);
|
|
|
6158 }
|
|
|
6159
|
|
|
6160 static int
|
|
|
6161 set_spell_finish(new_st)
|
|
|
6162 spelltab_T *new_st;
|
|
|
6163 {
|
|
|
6164 int i;
|
|
|
6165
|
|
|
6166 if (did_set_spelltab)
|
|
|
6167 {
|
|
|
6168 /* check that it's the same table */
|
|
|
6169 for (i = 0; i < 256; ++i)
|
|
|
6170 {
|
|
|
6171 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
6172 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
6173 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
6174 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
6175 {
|
|
|
6176 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
6177 return FAIL;
|
|
|
6178 }
|
|
|
6179 }
|
|
|
6180 }
|
|
|
6181 else
|
|
|
6182 {
|
|
|
6183 /* copy the new spelltab into the one being used */
|
|
|
6184 spelltab = *new_st;
|
|
|
6185 did_set_spelltab = TRUE;
|
|
|
6186 }
|
|
|
6187
|
|
|
6188 return OK;
|
|
|
6189 }
|
|
|
6190
|
|
|
6191 /*
|
|
358
|
6192 * Return TRUE if "p" points to a word character.
|
|
366
|
6193 * As a special case we see "midword" characters as word character when it is
|
|
358
|
6194 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
6195 * Thus this only works properly when past the first character of the word.
|
|
358
|
6196 */
|
|
|
6197 static int
|
|
376
|
6198 spell_iswordp(p, buf)
|
|
358
|
6199 char_u *p;
|
|
376
|
6200 buf_T *buf; /* buffer used */
|
|
358
|
6201 {
|
|
366
|
6202 #ifdef FEAT_MBYTE
|
|
358
|
6203 char_u *s;
|
|
366
|
6204 int l;
|
|
|
6205 int c;
|
|
|
6206
|
|
|
6207 if (has_mbyte)
|
|
|
6208 {
|
|
|
6209 l = MB_BYTE2LEN(*p);
|
|
358
|
6210 s = p;
|
|
366
|
6211 if (l == 1)
|
|
|
6212 {
|
|
|
6213 /* be quick for ASCII */
|
|
376
|
6214 if (buf->b_spell_ismw[*p])
|
|
366
|
6215 {
|
|
|
6216 s = p + 1; /* skip a mid-word character */
|
|
|
6217 l = MB_BYTE2LEN(*s);
|
|
|
6218 }
|
|
|
6219 }
|
|
|
6220 else
|
|
|
6221 {
|
|
|
6222 c = mb_ptr2char(p);
|
|
376
|
6223 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
6224 : (buf->b_spell_ismw_mb != NULL
|
|
|
6225 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
6226 {
|
|
|
6227 s = p + l;
|
|
|
6228 l = MB_BYTE2LEN(*s);
|
|
|
6229 }
|
|
|
6230 }
|
|
|
6231
|
|
390
|
6232 c = mb_ptr2char(s);
|
|
|
6233 if (c > 255)
|
|
366
|
6234 return mb_get_class(s) >= 2;
|
|
390
|
6235 return spelltab.st_isw[c];
|
|
366
|
6236 }
|
|
358
|
6237 #endif
|
|
366
|
6238
|
|
376
|
6239 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
6240 }
|
|
|
6241
|
|
|
6242 /*
|
|
|
6243 * Return TRUE if "p" points to a word character.
|
|
|
6244 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
6245 */
|
|
|
6246 static int
|
|
|
6247 spell_iswordp_nmw(p)
|
|
|
6248 char_u *p;
|
|
|
6249 {
|
|
|
6250 #ifdef FEAT_MBYTE
|
|
390
|
6251 int c;
|
|
|
6252
|
|
|
6253 if (has_mbyte)
|
|
|
6254 {
|
|
|
6255 c = mb_ptr2char(p);
|
|
|
6256 if (c > 255)
|
|
|
6257 return mb_get_class(p) >= 2;
|
|
|
6258 return spelltab.st_isw[c];
|
|
|
6259 }
|
|
376
|
6260 #endif
|
|
|
6261 return spelltab.st_isw[*p];
|
|
358
|
6262 }
|
|
|
6263
|
|
372
|
6264 #ifdef FEAT_MBYTE
|
|
|
6265 /*
|
|
|
6266 * Return TRUE if "p" points to a word character.
|
|
|
6267 * Wide version of spell_iswordp().
|
|
|
6268 */
|
|
|
6269 static int
|
|
376
|
6270 spell_iswordp_w(p, buf)
|
|
372
|
6271 int *p;
|
|
376
|
6272 buf_T *buf;
|
|
372
|
6273 {
|
|
|
6274 int *s;
|
|
|
6275
|
|
376
|
6276 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
6277 : (buf->b_spell_ismw_mb != NULL
|
|
|
6278 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
6279 s = p + 1;
|
|
|
6280 else
|
|
|
6281 s = p;
|
|
|
6282
|
|
390
|
6283 if (*s > 255)
|
|
372
|
6284 {
|
|
|
6285 if (enc_utf8)
|
|
|
6286 return utf_class(*s) >= 2;
|
|
|
6287 if (enc_dbcs)
|
|
|
6288 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
6289 return 0;
|
|
|
6290 }
|
|
|
6291 return spelltab.st_isw[*s];
|
|
|
6292 }
|
|
|
6293 #endif
|
|
|
6294
|
|
358
|
6295 /*
|
|
339
|
6296 * Write the table with prefix conditions to the .spl file.
|
|
|
6297 */
|
|
|
6298 static void
|
|
|
6299 write_spell_prefcond(fd, gap)
|
|
|
6300 FILE *fd;
|
|
|
6301 garray_T *gap;
|
|
|
6302 {
|
|
|
6303 int i;
|
|
|
6304 char_u *p;
|
|
|
6305 int len;
|
|
|
6306
|
|
|
6307 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
6308
|
|
|
6309 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6310 {
|
|
|
6311 /* <prefcond> : <condlen> <condstr> */
|
|
|
6312 p = ((char_u **)gap->ga_data)[i];
|
|
|
6313 if (p == NULL)
|
|
|
6314 fputc(0, fd);
|
|
|
6315 else
|
|
|
6316 {
|
|
|
6317 len = STRLEN(p);
|
|
|
6318 fputc(len, fd);
|
|
|
6319 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
6320 }
|
|
|
6321 }
|
|
|
6322 }
|
|
|
6323
|
|
|
6324 /*
|
|
307
|
6325 * Write the current tables into the .spl file.
|
|
|
6326 * This makes sure the same characters are recognized as word characters when
|
|
|
6327 * generating an when using a spell file.
|
|
|
6328 */
|
|
|
6329 static void
|
|
|
6330 write_spell_chartab(fd)
|
|
|
6331 FILE *fd;
|
|
|
6332 {
|
|
|
6333 char_u charbuf[256 * 4];
|
|
|
6334 int len = 0;
|
|
|
6335 int flags;
|
|
|
6336 int i;
|
|
|
6337
|
|
|
6338 fputc(128, fd); /* <charflagslen> */
|
|
|
6339 for (i = 128; i < 256; ++i)
|
|
|
6340 {
|
|
|
6341 flags = 0;
|
|
|
6342 if (spelltab.st_isw[i])
|
|
324
|
6343 flags |= CF_WORD;
|
|
307
|
6344 if (spelltab.st_isu[i])
|
|
324
|
6345 flags |= CF_UPPER;
|
|
307
|
6346 fputc(flags, fd); /* <charflags> */
|
|
|
6347
|
|
310
|
6348 #ifdef FEAT_MBYTE
|
|
|
6349 if (has_mbyte)
|
|
|
6350 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
|
6351 else
|
|
|
6352 #endif
|
|
|
6353 charbuf[len++] = spelltab.st_fold[i];
|
|
307
|
6354 }
|
|
|
6355
|
|
|
6356 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
|
6357 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
|
6358 }
|
|
|
6359
|
|
|
6360 /*
|
|
324
|
6361 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
6362 * Uses the character definitions from the .spl file.
|
|
307
|
6363 * When using a multi-byte 'encoding' the length may change!
|
|
|
6364 * Returns FAIL when something wrong.
|
|
|
6365 */
|
|
|
6366 static int
|
|
324
|
6367 spell_casefold(str, len, buf, buflen)
|
|
|
6368 char_u *str;
|
|
307
|
6369 int len;
|
|
|
6370 char_u *buf;
|
|
|
6371 int buflen;
|
|
|
6372 {
|
|
|
6373 int i;
|
|
|
6374
|
|
|
6375 if (len >= buflen)
|
|
|
6376 {
|
|
|
6377 buf[0] = NUL;
|
|
|
6378 return FAIL; /* result will not fit */
|
|
|
6379 }
|
|
|
6380
|
|
|
6381 #ifdef FEAT_MBYTE
|
|
|
6382 if (has_mbyte)
|
|
|
6383 {
|
|
324
|
6384 int outi = 0;
|
|
|
6385 char_u *p;
|
|
307
|
6386 int c;
|
|
|
6387
|
|
|
6388 /* Fold one character at a time. */
|
|
324
|
6389 for (p = str; p < str + len; )
|
|
307
|
6390 {
|
|
|
6391 if (outi + MB_MAXBYTES > buflen)
|
|
|
6392 {
|
|
|
6393 buf[outi] = NUL;
|
|
|
6394 return FAIL;
|
|
|
6395 }
|
|
324
|
6396 c = mb_ptr2char_adv(&p);
|
|
|
6397 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
6398 }
|
|
|
6399 buf[outi] = NUL;
|
|
|
6400 }
|
|
|
6401 else
|
|
|
6402 #endif
|
|
|
6403 {
|
|
|
6404 /* Be quick for non-multibyte encodings. */
|
|
|
6405 for (i = 0; i < len; ++i)
|
|
324
|
6406 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
6407 buf[i] = NUL;
|
|
|
6408 }
|
|
|
6409
|
|
|
6410 return OK;
|
|
|
6411 }
|
|
|
6412
|
|
372
|
6413 #define SPS_BEST 1
|
|
|
6414 #define SPS_FAST 2
|
|
|
6415 #define SPS_DOUBLE 4
|
|
|
6416
|
|
|
6417 static int sps_flags = SPS_BEST;
|
|
|
6418
|
|
|
6419 /*
|
|
|
6420 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
|
6421 * Sets "sps_flags".
|
|
|
6422 */
|
|
|
6423 int
|
|
|
6424 spell_check_sps()
|
|
|
6425 {
|
|
|
6426 char_u *p;
|
|
|
6427 char_u buf[MAXPATHL];
|
|
|
6428 int f;
|
|
|
6429
|
|
|
6430 sps_flags = 0;
|
|
|
6431
|
|
|
6432 for (p = p_sps; *p != NUL; )
|
|
|
6433 {
|
|
|
6434 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6435
|
|
|
6436 f = 0;
|
|
|
6437 if (STRCMP(buf, "best") == 0)
|
|
|
6438 f = SPS_BEST;
|
|
|
6439 else if (STRCMP(buf, "fast") == 0)
|
|
|
6440 f = SPS_FAST;
|
|
|
6441 else if (STRCMP(buf, "double") == 0)
|
|
|
6442 f = SPS_DOUBLE;
|
|
|
6443 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
6444 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
6445 f = -1;
|
|
|
6446
|
|
|
6447 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
6448 {
|
|
|
6449 sps_flags = SPS_BEST;
|
|
|
6450 return FAIL;
|
|
|
6451 }
|
|
|
6452 if (f != 0)
|
|
|
6453 sps_flags = f;
|
|
|
6454 }
|
|
|
6455
|
|
|
6456 if (sps_flags == 0)
|
|
|
6457 sps_flags = SPS_BEST;
|
|
|
6458
|
|
|
6459 return OK;
|
|
|
6460 }
|
|
|
6461
|
|
|
6462 /* Remember what "z?" replaced. */
|
|
|
6463 static char_u *repl_from = NULL;
|
|
|
6464 static char_u *repl_to = NULL;
|
|
|
6465
|
|
323
|
6466 /*
|
|
|
6467 * "z?": Find badly spelled word under or after the cursor.
|
|
|
6468 * Give suggestions for the properly spelled word.
|
|
|
6469 */
|
|
|
6470 void
|
|
|
6471 spell_suggest()
|
|
|
6472 {
|
|
|
6473 char_u *line;
|
|
|
6474 pos_T prev_cursor = curwin->w_cursor;
|
|
|
6475 char_u wcopy[MAXWLEN + 2];
|
|
|
6476 char_u *p;
|
|
|
6477 int i;
|
|
|
6478 int c;
|
|
|
6479 suginfo_T sug;
|
|
|
6480 suggest_T *stp;
|
|
372
|
6481 int mouse_used;
|
|
388
|
6482 int need_cap;
|
|
|
6483 regmatch_T regmatch;
|
|
|
6484 int endcol;
|
|
|
6485 char_u *line_copy = NULL;
|
|
323
|
6486
|
|
344
|
6487 /* Find the start of the badly spelled word. */
|
|
346
|
6488 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
6489 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
6490 {
|
|
|
6491 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
6492 return;
|
|
|
6493
|
|
|
6494 /* No bad word or it starts after the cursor: use the word under the
|
|
|
6495 * cursor. */
|
|
|
6496 curwin->w_cursor = prev_cursor;
|
|
|
6497 line = ml_get_curline();
|
|
|
6498 p = line + curwin->w_cursor.col;
|
|
|
6499 /* Backup to before start of word. */
|
|
390
|
6500 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
6501 mb_ptr_back(line, p);
|
|
|
6502 /* Forward to start of word. */
|
|
390
|
6503 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
6504 mb_ptr_adv(p);
|
|
|
6505
|
|
390
|
6506 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
6507 {
|
|
|
6508 beep_flush();
|
|
|
6509 return;
|
|
|
6510 }
|
|
|
6511 curwin->w_cursor.col = p - line;
|
|
323
|
6512 }
|
|
|
6513
|
|
344
|
6514 /* Get the word and its length. */
|
|
323
|
6515 line = ml_get_curline();
|
|
344
|
6516
|
|
388
|
6517 /* Figure out if the word should be capitalised. */
|
|
|
6518 need_cap = FALSE;
|
|
|
6519 if (curbuf->b_cap_prog != NULL)
|
|
|
6520 {
|
|
|
6521 endcol = 0;
|
|
398
|
6522 if ((int)(skipwhite(line) - line) == (int)curwin->w_cursor.col)
|
|
388
|
6523 {
|
|
|
6524 /* At start of line, check if previous line is empty or sentence
|
|
|
6525 * ends there. */
|
|
|
6526 if (curwin->w_cursor.lnum == 1)
|
|
|
6527 need_cap = TRUE;
|
|
|
6528 else
|
|
|
6529 {
|
|
|
6530 line = ml_get(curwin->w_cursor.lnum - 1);
|
|
|
6531 if (*skipwhite(line) == NUL)
|
|
|
6532 need_cap = TRUE;
|
|
|
6533 else
|
|
|
6534 {
|
|
|
6535 /* Append a space in place of the line break. */
|
|
|
6536 line_copy = concat_str(line, (char_u *)" ");
|
|
|
6537 line = line_copy;
|
|
|
6538 endcol = STRLEN(line);
|
|
|
6539 }
|
|
|
6540 }
|
|
|
6541 }
|
|
|
6542 else
|
|
|
6543 endcol = curwin->w_cursor.col;
|
|
|
6544
|
|
|
6545 if (endcol > 0)
|
|
|
6546 {
|
|
|
6547 /* Check if sentence ends before the bad word. */
|
|
|
6548 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
6549 regmatch.rm_ic = FALSE;
|
|
|
6550 p = line + endcol;
|
|
|
6551 for (;;)
|
|
|
6552 {
|
|
|
6553 mb_ptr_back(line, p);
|
|
390
|
6554 if (p == line || spell_iswordp_nmw(p))
|
|
388
|
6555 break;
|
|
|
6556 if (vim_regexec(®match, p, 0)
|
|
|
6557 && regmatch.endp[0] == line + endcol)
|
|
|
6558 {
|
|
|
6559 need_cap = TRUE;
|
|
|
6560 break;
|
|
|
6561 }
|
|
|
6562 }
|
|
|
6563 }
|
|
|
6564
|
|
|
6565 /* get the line again, we may have been using the previous one */
|
|
|
6566 line = ml_get_curline();
|
|
|
6567 vim_free(line_copy);
|
|
|
6568 }
|
|
|
6569
|
|
344
|
6570 /* Get the list of suggestions */
|
|
388
|
6571 spell_find_suggest(line + curwin->w_cursor.col, &sug, (int)Rows - 2,
|
|
|
6572 TRUE, need_cap);
|
|
323
|
6573
|
|
|
6574 if (sug.su_ga.ga_len == 0)
|
|
|
6575 MSG(_("Sorry, no suggestions"));
|
|
|
6576 else
|
|
|
6577 {
|
|
372
|
6578 vim_free(repl_from);
|
|
|
6579 repl_from = NULL;
|
|
|
6580 vim_free(repl_to);
|
|
|
6581 repl_to = NULL;
|
|
|
6582
|
|
323
|
6583 /* List the suggestions. */
|
|
|
6584 msg_start();
|
|
372
|
6585 lines_left = Rows; /* avoid more prompt */
|
|
323
|
6586 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
6587 sug.su_badlen, sug.su_badptr);
|
|
|
6588 msg_puts(IObuff);
|
|
|
6589 msg_clr_eos();
|
|
|
6590 msg_putchar('\n');
|
|
346
|
6591
|
|
323
|
6592 msg_scroll = TRUE;
|
|
|
6593 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
6594 {
|
|
344
|
6595 stp = &SUG(sug.su_ga, i);
|
|
323
|
6596
|
|
|
6597 /* The suggested word may replace only part of the bad word, add
|
|
|
6598 * the not replaced part. */
|
|
|
6599 STRCPY(wcopy, stp->st_word);
|
|
|
6600 if (sug.su_badlen > stp->st_orglen)
|
|
|
6601 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
6602 sug.su_badptr + stp->st_orglen,
|
|
|
6603 sug.su_badlen - stp->st_orglen);
|
|
346
|
6604 vim_snprintf((char *)IObuff, IOSIZE, _("%2d \"%s\""), i + 1, wcopy);
|
|
|
6605 msg_puts(IObuff);
|
|
|
6606
|
|
|
6607 /* The word may replace more than "su_badlen". */
|
|
|
6608 if (sug.su_badlen < stp->st_orglen)
|
|
|
6609 {
|
|
|
6610 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
6611 stp->st_orglen, sug.su_badptr);
|
|
|
6612 msg_puts(IObuff);
|
|
|
6613 }
|
|
|
6614
|
|
324
|
6615 if (p_verbose > 0)
|
|
344
|
6616 {
|
|
346
|
6617 /* Add the score. */
|
|
351
|
6618 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
346
|
6619 vim_snprintf((char *)IObuff, IOSIZE, _(" (%s%d - %d)"),
|
|
344
|
6620 stp->st_salscore ? "s " : "",
|
|
|
6621 stp->st_score, stp->st_altscore);
|
|
|
6622 else
|
|
346
|
6623 vim_snprintf((char *)IObuff, IOSIZE, _(" (%d)"),
|
|
|
6624 stp->st_score);
|
|
|
6625 msg_advance(30);
|
|
|
6626 msg_puts(IObuff);
|
|
344
|
6627 }
|
|
323
|
6628 msg_putchar('\n');
|
|
|
6629 }
|
|
|
6630
|
|
|
6631 /* Ask for choice. */
|
|
372
|
6632 i = prompt_for_number(&mouse_used);
|
|
|
6633 if (mouse_used)
|
|
|
6634 i -= lines_left;
|
|
|
6635
|
|
344
|
6636 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
6637 {
|
|
372
|
6638 /* Save the from and to text for :spellrepall. */
|
|
|
6639 stp = &SUG(sug.su_ga, i - 1);
|
|
|
6640 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
6641 repl_to = vim_strsave(stp->st_word);
|
|
|
6642
|
|
323
|
6643 /* Replace the word. */
|
|
|
6644 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
6645 if (p != NULL)
|
|
|
6646 {
|
|
|
6647 c = sug.su_badptr - line;
|
|
|
6648 mch_memmove(p, line, c);
|
|
|
6649 STRCPY(p + c, stp->st_word);
|
|
|
6650 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
6651 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6652 curwin->w_cursor.col = c;
|
|
|
6653 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
6654
|
|
|
6655 /* For redo we use a change-word command. */
|
|
|
6656 ResetRedobuff();
|
|
|
6657 AppendToRedobuff((char_u *)"ciw");
|
|
|
6658 AppendToRedobuff(stp->st_word);
|
|
|
6659 AppendCharToRedobuff(ESC);
|
|
323
|
6660 }
|
|
|
6661 }
|
|
|
6662 else
|
|
|
6663 curwin->w_cursor = prev_cursor;
|
|
|
6664 }
|
|
|
6665
|
|
344
|
6666 spell_find_cleanup(&sug);
|
|
|
6667 }
|
|
|
6668
|
|
|
6669 /*
|
|
372
|
6670 * ":spellrepall"
|
|
|
6671 */
|
|
|
6672 /*ARGSUSED*/
|
|
|
6673 void
|
|
|
6674 ex_spellrepall(eap)
|
|
|
6675 exarg_T *eap;
|
|
|
6676 {
|
|
|
6677 pos_T pos = curwin->w_cursor;
|
|
|
6678 char_u *frompat;
|
|
|
6679 int addlen;
|
|
|
6680 char_u *line;
|
|
|
6681 char_u *p;
|
|
|
6682 int didone = FALSE;
|
|
|
6683 int save_ws = p_ws;
|
|
|
6684
|
|
|
6685 if (repl_from == NULL || repl_to == NULL)
|
|
|
6686 {
|
|
|
6687 EMSG(_("E752: No previous spell replacement"));
|
|
|
6688 return;
|
|
|
6689 }
|
|
|
6690 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
6691
|
|
|
6692 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
6693 if (frompat == NULL)
|
|
|
6694 return;
|
|
|
6695 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
6696 p_ws = FALSE;
|
|
|
6697
|
|
|
6698 curwin->w_cursor.lnum = 0;
|
|
|
6699 while (!got_int)
|
|
|
6700 {
|
|
|
6701 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
6702 || u_save_cursor() == FAIL)
|
|
|
6703 break;
|
|
|
6704
|
|
|
6705 /* Only replace when the right word isn't there yet. This happens
|
|
|
6706 * when changing "etc" to "etc.". */
|
|
|
6707 line = ml_get_curline();
|
|
|
6708 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
6709 repl_to, STRLEN(repl_to)) != 0)
|
|
|
6710 {
|
|
|
6711 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
6712 if (p == NULL)
|
|
|
6713 break;
|
|
|
6714 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
6715 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
6716 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
6717 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6718 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
6719 didone = TRUE;
|
|
|
6720 }
|
|
|
6721 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
6722 }
|
|
|
6723
|
|
|
6724 p_ws = save_ws;
|
|
|
6725 curwin->w_cursor = pos;
|
|
|
6726 vim_free(frompat);
|
|
|
6727
|
|
|
6728 if (!didone)
|
|
|
6729 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
|
6730 }
|
|
|
6731
|
|
|
6732 /*
|
|
344
|
6733 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
6734 * a list of allocated strings.
|
|
|
6735 */
|
|
|
6736 void
|
|
|
6737 spell_suggest_list(gap, word, maxcount)
|
|
|
6738 garray_T *gap;
|
|
|
6739 char_u *word;
|
|
|
6740 int maxcount; /* maximum nr of suggestions */
|
|
|
6741 {
|
|
|
6742 suginfo_T sug;
|
|
|
6743 int i;
|
|
|
6744 suggest_T *stp;
|
|
|
6745 char_u *wcopy;
|
|
|
6746
|
|
388
|
6747 spell_find_suggest(word, &sug, maxcount, FALSE, FALSE);
|
|
344
|
6748
|
|
|
6749 /* Make room in "gap". */
|
|
|
6750 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
6751 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
6752 return;
|
|
|
6753
|
|
323
|
6754 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
6755 {
|
|
|
6756 stp = &SUG(sug.su_ga, i);
|
|
|
6757
|
|
|
6758 /* The suggested word may replace only part of "word", add the not
|
|
|
6759 * replaced part. */
|
|
|
6760 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
6761 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
6762 if (wcopy == NULL)
|
|
|
6763 break;
|
|
|
6764 STRCPY(wcopy, stp->st_word);
|
|
|
6765 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
6766 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
6767 }
|
|
|
6768
|
|
|
6769 spell_find_cleanup(&sug);
|
|
|
6770 }
|
|
|
6771
|
|
|
6772 /*
|
|
|
6773 * Find spell suggestions for the word at the start of "badptr".
|
|
|
6774 * Return the suggestions in "su->su_ga".
|
|
|
6775 * The maximum number of suggestions is "maxcount".
|
|
|
6776 * Note: does use info for the current window.
|
|
|
6777 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
6778 */
|
|
|
6779 static void
|
|
388
|
6780 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
6781 char_u *badptr;
|
|
|
6782 suginfo_T *su;
|
|
|
6783 int maxcount;
|
|
358
|
6784 int banbadword; /* don't include badword in suggestions */
|
|
388
|
6785 int need_cap; /* word should start with capital */
|
|
344
|
6786 {
|
|
385
|
6787 int attr = 0;
|
|
372
|
6788 char_u buf[MAXPATHL];
|
|
|
6789 char_u *p;
|
|
|
6790 int do_combine = FALSE;
|
|
|
6791 char_u *sps_copy;
|
|
|
6792 #ifdef FEAT_EVAL
|
|
|
6793 static int expr_busy = FALSE;
|
|
|
6794 #endif
|
|
385
|
6795 int c;
|
|
344
|
6796
|
|
|
6797 /*
|
|
|
6798 * Set the info in "*su".
|
|
|
6799 */
|
|
|
6800 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
6801 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
6802 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
6803 if (*badptr == NUL)
|
|
|
6804 return;
|
|
344
|
6805 hash_init(&su->su_banned);
|
|
|
6806
|
|
|
6807 su->su_badptr = badptr;
|
|
385
|
6808 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
6809 su->su_maxcount = maxcount;
|
|
372
|
6810 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
6811
|
|
|
6812 if (su->su_badlen >= MAXWLEN)
|
|
|
6813 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
6814 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
6815 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
6816 su->su_fbadword, MAXWLEN);
|
|
346
|
6817 /* get caps flags for bad word */
|
|
|
6818 su->su_badflags = captype(su->su_badptr, su->su_badptr + su->su_badlen);
|
|
388
|
6819 if (need_cap)
|
|
|
6820 su->su_badflags |= WF_ONECAP;
|
|
344
|
6821
|
|
385
|
6822 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
6823 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
6824 * for that. */
|
|
455
|
6825 c = PTR2CHAR(su->su_badptr);
|
|
385
|
6826 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
6827 {
|
|
|
6828 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
6829 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
6830 0, TRUE);
|
|
|
6831 }
|
|
|
6832
|
|
344
|
6833 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
6834 if (banbadword)
|
|
|
6835 add_banned(su, su->su_badword);
|
|
344
|
6836
|
|
372
|
6837 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
6838 sps_copy = vim_strsave(p_sps);
|
|
|
6839 if (sps_copy == NULL)
|
|
|
6840 return;
|
|
|
6841
|
|
|
6842 /* Loop over the items in 'spellsuggest'. */
|
|
|
6843 for (p = sps_copy; *p != NUL; )
|
|
|
6844 {
|
|
|
6845 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6846
|
|
|
6847 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
6848 {
|
|
|
6849 #ifdef FEAT_EVAL
|
|
375
|
6850 /* Evaluate an expression. Skip this when called recursively,
|
|
|
6851 * when using spellsuggest() in the expression. */
|
|
372
|
6852 if (!expr_busy)
|
|
|
6853 {
|
|
|
6854 expr_busy = TRUE;
|
|
|
6855 spell_suggest_expr(su, buf + 5);
|
|
|
6856 expr_busy = FALSE;
|
|
|
6857 }
|
|
|
6858 #endif
|
|
|
6859 }
|
|
|
6860 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
6861 /* Use list of suggestions in a file. */
|
|
|
6862 spell_suggest_file(su, buf + 5);
|
|
|
6863 else
|
|
|
6864 {
|
|
|
6865 /* Use internal method. */
|
|
|
6866 spell_suggest_intern(su);
|
|
|
6867 if (sps_flags & SPS_DOUBLE)
|
|
|
6868 do_combine = TRUE;
|
|
|
6869 }
|
|
|
6870 }
|
|
|
6871
|
|
|
6872 vim_free(sps_copy);
|
|
|
6873
|
|
|
6874 if (do_combine)
|
|
|
6875 /* Combine the two list of suggestions. This must be done last,
|
|
|
6876 * because sorting changes the order again. */
|
|
|
6877 score_combine(su);
|
|
|
6878 }
|
|
|
6879
|
|
|
6880 #ifdef FEAT_EVAL
|
|
|
6881 /*
|
|
|
6882 * Find suggestions by evaluating expression "expr".
|
|
|
6883 */
|
|
|
6884 static void
|
|
|
6885 spell_suggest_expr(su, expr)
|
|
|
6886 suginfo_T *su;
|
|
|
6887 char_u *expr;
|
|
|
6888 {
|
|
|
6889 list_T *list;
|
|
|
6890 listitem_T *li;
|
|
|
6891 int score;
|
|
|
6892 char_u *p;
|
|
|
6893
|
|
|
6894 /* The work is split up in a few parts to avoid having to export
|
|
|
6895 * suginfo_T.
|
|
|
6896 * First evaluate the expression and get the resulting list. */
|
|
|
6897 list = eval_spell_expr(su->su_badword, expr);
|
|
|
6898 if (list != NULL)
|
|
|
6899 {
|
|
|
6900 /* Loop over the items in the list. */
|
|
|
6901 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
6902 if (li->li_tv.v_type == VAR_LIST)
|
|
|
6903 {
|
|
|
6904 /* Get the word and the score from the items. */
|
|
|
6905 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
6906 if (score >= 0)
|
|
|
6907 add_suggestion(su, &su->su_ga, p,
|
|
|
6908 su->su_badlen, score, 0, TRUE);
|
|
|
6909 }
|
|
|
6910 list_unref(list);
|
|
|
6911 }
|
|
|
6912
|
|
|
6913 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6914 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6915 }
|
|
|
6916 #endif
|
|
|
6917
|
|
|
6918 /*
|
|
|
6919 * Find suggestions a file "fname".
|
|
|
6920 */
|
|
|
6921 static void
|
|
|
6922 spell_suggest_file(su, fname)
|
|
|
6923 suginfo_T *su;
|
|
|
6924 char_u *fname;
|
|
|
6925 {
|
|
|
6926 FILE *fd;
|
|
|
6927 char_u line[MAXWLEN * 2];
|
|
|
6928 char_u *p;
|
|
|
6929 int len;
|
|
|
6930 char_u cword[MAXWLEN];
|
|
|
6931
|
|
|
6932 /* Open the file. */
|
|
|
6933 fd = mch_fopen((char *)fname, "r");
|
|
|
6934 if (fd == NULL)
|
|
|
6935 {
|
|
|
6936 EMSG2(_(e_notopen), fname);
|
|
|
6937 return;
|
|
|
6938 }
|
|
|
6939
|
|
|
6940 /* Read it line by line. */
|
|
|
6941 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
6942 {
|
|
|
6943 line_breakcheck();
|
|
|
6944
|
|
|
6945 p = vim_strchr(line, '/');
|
|
|
6946 if (p == NULL)
|
|
|
6947 continue; /* No Tab found, just skip the line. */
|
|
|
6948 *p++ = NUL;
|
|
|
6949 if (STRICMP(su->su_badword, line) == 0)
|
|
|
6950 {
|
|
|
6951 /* Match! Isolate the good word, until CR or NL. */
|
|
|
6952 for (len = 0; p[len] >= ' '; ++len)
|
|
|
6953 ;
|
|
|
6954 p[len] = NUL;
|
|
|
6955
|
|
|
6956 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
6957 * of the bad word. */
|
|
|
6958 if (captype(p, NULL) == 0)
|
|
|
6959 {
|
|
|
6960 make_case_word(p, cword, su->su_badflags);
|
|
|
6961 p = cword;
|
|
|
6962 }
|
|
|
6963
|
|
|
6964 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
6965 SCORE_FILE, 0, TRUE);
|
|
|
6966 }
|
|
|
6967 }
|
|
|
6968
|
|
|
6969 fclose(fd);
|
|
|
6970
|
|
|
6971 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6972 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6973 }
|
|
|
6974
|
|
|
6975 /*
|
|
|
6976 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
6977 */
|
|
|
6978 static void
|
|
|
6979 spell_suggest_intern(su)
|
|
|
6980 suginfo_T *su;
|
|
|
6981 {
|
|
344
|
6982 /*
|
|
346
|
6983 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
6984 *
|
|
|
6985 * Set a maximum score to limit the combination of operations that is
|
|
|
6986 * tried.
|
|
|
6987 */
|
|
346
|
6988 suggest_try_special(su);
|
|
|
6989
|
|
|
6990 /*
|
|
|
6991 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
6992 * from the .aff file and inserting a space (split the word).
|
|
|
6993 */
|
|
|
6994 suggest_try_change(su);
|
|
344
|
6995
|
|
|
6996 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
6997 if (sps_flags & SPS_DOUBLE)
|
|
|
6998 score_comp_sal(su);
|
|
|
6999
|
|
|
7000 /*
|
|
346
|
7001 * 3. Try finding sound-a-like words.
|
|
344
|
7002 *
|
|
|
7003 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
7004 * very slow and often doesn't find new suggestions.
|
|
|
7005 */
|
|
|
7006 if ((sps_flags & SPS_DOUBLE)
|
|
|
7007 || (!(sps_flags & SPS_FAST)
|
|
|
7008 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
7009 {
|
|
|
7010 /* Allow a higher score now. */
|
|
|
7011 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
7012 suggest_try_soundalike(su);
|
|
344
|
7013 }
|
|
|
7014
|
|
|
7015 /* When CTRL-C was hit while searching do show the results. */
|
|
|
7016 ui_breakcheck();
|
|
|
7017 if (got_int)
|
|
|
7018 {
|
|
|
7019 (void)vgetc();
|
|
|
7020 got_int = FALSE;
|
|
|
7021 }
|
|
|
7022
|
|
372
|
7023 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
7024 {
|
|
|
7025 if (sps_flags & SPS_BEST)
|
|
|
7026 /* Adjust the word score for how it sounds like. */
|
|
|
7027 rescore_suggestions(su);
|
|
|
7028
|
|
|
7029 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
7030 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
7031 }
|
|
|
7032 }
|
|
|
7033
|
|
|
7034 /*
|
|
|
7035 * Free the info put in "*su" by spell_find_suggest().
|
|
|
7036 */
|
|
|
7037 static void
|
|
|
7038 spell_find_cleanup(su)
|
|
|
7039 suginfo_T *su;
|
|
|
7040 {
|
|
|
7041 int i;
|
|
|
7042
|
|
|
7043 /* Free the suggestions. */
|
|
|
7044 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
7045 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
7046 ga_clear(&su->su_ga);
|
|
|
7047 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
7048 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
7049 ga_clear(&su->su_sga);
|
|
323
|
7050
|
|
|
7051 /* Free the banned words. */
|
|
344
|
7052 free_banned(su);
|
|
323
|
7053 }
|
|
|
7054
|
|
|
7055 /*
|
|
324
|
7056 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
7057 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
7058 * The result is NUL terminated.
|
|
323
|
7059 */
|
|
|
7060 static void
|
|
324
|
7061 onecap_copy(word, wcopy, upper)
|
|
323
|
7062 char_u *word;
|
|
|
7063 char_u *wcopy;
|
|
|
7064 int upper; /* TRUE: first letter made upper case */
|
|
|
7065 {
|
|
|
7066 char_u *p;
|
|
|
7067 int c;
|
|
|
7068 int l;
|
|
|
7069
|
|
|
7070 p = word;
|
|
|
7071 #ifdef FEAT_MBYTE
|
|
|
7072 if (has_mbyte)
|
|
|
7073 c = mb_ptr2char_adv(&p);
|
|
|
7074 else
|
|
|
7075 #endif
|
|
|
7076 c = *p++;
|
|
|
7077 if (upper)
|
|
324
|
7078 c = SPELL_TOUPPER(c);
|
|
323
|
7079 else
|
|
324
|
7080 c = SPELL_TOFOLD(c);
|
|
323
|
7081 #ifdef FEAT_MBYTE
|
|
|
7082 if (has_mbyte)
|
|
|
7083 l = mb_char2bytes(c, wcopy);
|
|
|
7084 else
|
|
|
7085 #endif
|
|
|
7086 {
|
|
|
7087 l = 1;
|
|
|
7088 wcopy[0] = c;
|
|
|
7089 }
|
|
376
|
7090 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
7091 }
|
|
|
7092
|
|
|
7093 /*
|
|
324
|
7094 * Make a copy of "word" with all the letters upper cased into
|
|
|
7095 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
7096 */
|
|
|
7097 static void
|
|
|
7098 allcap_copy(word, wcopy)
|
|
|
7099 char_u *word;
|
|
|
7100 char_u *wcopy;
|
|
|
7101 {
|
|
|
7102 char_u *s;
|
|
|
7103 char_u *d;
|
|
|
7104 int c;
|
|
|
7105
|
|
|
7106 d = wcopy;
|
|
|
7107 for (s = word; *s != NUL; )
|
|
|
7108 {
|
|
|
7109 #ifdef FEAT_MBYTE
|
|
|
7110 if (has_mbyte)
|
|
|
7111 c = mb_ptr2char_adv(&s);
|
|
|
7112 else
|
|
|
7113 #endif
|
|
|
7114 c = *s++;
|
|
324
|
7115 c = SPELL_TOUPPER(c);
|
|
323
|
7116
|
|
|
7117 #ifdef FEAT_MBYTE
|
|
|
7118 if (has_mbyte)
|
|
|
7119 {
|
|
|
7120 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
7121 break;
|
|
|
7122 d += mb_char2bytes(c, d);
|
|
|
7123 }
|
|
|
7124 else
|
|
|
7125 #endif
|
|
|
7126 {
|
|
|
7127 if (d - wcopy >= MAXWLEN - 1)
|
|
|
7128 break;
|
|
|
7129 *d++ = c;
|
|
|
7130 }
|
|
|
7131 }
|
|
|
7132 *d = NUL;
|
|
|
7133 }
|
|
|
7134
|
|
|
7135 /*
|
|
346
|
7136 * Try finding suggestions by recognizing specific situations.
|
|
|
7137 */
|
|
|
7138 static void
|
|
|
7139 suggest_try_special(su)
|
|
|
7140 suginfo_T *su;
|
|
|
7141 {
|
|
|
7142 char_u *p;
|
|
376
|
7143 size_t len;
|
|
346
|
7144 int c;
|
|
|
7145 char_u word[MAXWLEN];
|
|
|
7146
|
|
|
7147 /*
|
|
|
7148 * Recognize a word that is repeated: "the the".
|
|
|
7149 */
|
|
|
7150 p = skiptowhite(su->su_fbadword);
|
|
|
7151 len = p - su->su_fbadword;
|
|
|
7152 p = skipwhite(p);
|
|
|
7153 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
7154 {
|
|
|
7155 /* Include badflags: if the badword is onecap or allcap
|
|
|
7156 * use that for the goodword too: "The the" -> "The". */
|
|
|
7157 c = su->su_fbadword[len];
|
|
|
7158 su->su_fbadword[len] = NUL;
|
|
|
7159 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
7160 su->su_fbadword[len] = c;
|
|
351
|
7161 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
7162 }
|
|
|
7163 }
|
|
|
7164
|
|
|
7165 /*
|
|
323
|
7166 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
7167 *
|
|
|
7168 * This uses a state machine. At each node in the tree we try various
|
|
|
7169 * operations. When trying if an operation work "depth" is increased and the
|
|
|
7170 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
7171 * character, replace one and delete another. The number of changes is
|
|
|
7172 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
7173 */
|
|
|
7174 static void
|
|
346
|
7175 suggest_try_change(su)
|
|
323
|
7176 suginfo_T *su;
|
|
|
7177 {
|
|
|
7178 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
7179 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
7180 trystate_T stack[MAXWLEN];
|
|
|
7181 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
|
7182 * to for word split) */
|
|
|
7183 char_u prewordlen = 0; /* length of word in "preword" */
|
|
|
7184 int splitoff = 0; /* index in tword after last split */
|
|
|
7185 trystate_T *sp;
|
|
|
7186 int newscore;
|
|
|
7187 langp_T *lp;
|
|
375
|
7188 char_u *byts, *fbyts, *pbyts;
|
|
|
7189 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
7190 int depth;
|
|
330
|
7191 int c, c2, c3;
|
|
390
|
7192 int n;
|
|
323
|
7193 int flags;
|
|
|
7194 garray_T *gap;
|
|
324
|
7195 idx_T arridx;
|
|
323
|
7196 int len;
|
|
|
7197 char_u *p;
|
|
|
7198 fromto_T *ftp;
|
|
330
|
7199 int fl = 0, tl;
|
|
346
|
7200 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
323
|
7201
|
|
|
7202 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
7203 * to find matches (esp. REP items). Append some more text, changing
|
|
|
7204 * chars after the bad word may help. */
|
|
323
|
7205 STRCPY(fword, su->su_fbadword);
|
|
346
|
7206 n = STRLEN(fword);
|
|
|
7207 p = su->su_badptr + su->su_badlen;
|
|
|
7208 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
7209
|
|
|
7210 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
7211 lp->lp_slang != NULL; ++lp)
|
|
|
7212 {
|
|
|
7213 /*
|
|
|
7214 * Go through the whole case-fold tree, try changes at each node.
|
|
|
7215 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
7216 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
7217 * word).
|
|
|
7218 */
|
|
|
7219 depth = 0;
|
|
375
|
7220 sp = &stack[0];
|
|
|
7221 sp->ts_state = STATE_START;
|
|
|
7222 sp->ts_score = 0;
|
|
|
7223 sp->ts_curi = 1;
|
|
|
7224 sp->ts_fidx = 0;
|
|
|
7225 sp->ts_fidxtry = 0;
|
|
|
7226 sp->ts_twordlen = 0;
|
|
|
7227 sp->ts_arridx = 0;
|
|
330
|
7228 #ifdef FEAT_MBYTE
|
|
375
|
7229 sp->ts_tcharlen = 0;
|
|
330
|
7230 #endif
|
|
|
7231
|
|
|
7232 /*
|
|
375
|
7233 * When there are postponed prefixes we need to use these first. At
|
|
|
7234 * the end of the prefix we continue in the case-fold tree.
|
|
|
7235 */
|
|
|
7236 fbyts = lp->lp_slang->sl_fbyts;
|
|
|
7237 fidxs = lp->lp_slang->sl_fidxs;
|
|
|
7238 pbyts = lp->lp_slang->sl_pbyts;
|
|
|
7239 pidxs = lp->lp_slang->sl_pidxs;
|
|
|
7240 if (pbyts != NULL)
|
|
|
7241 {
|
|
|
7242 byts = pbyts;
|
|
|
7243 idxs = pidxs;
|
|
|
7244 sp->ts_prefixdepth = PREFIXTREE;
|
|
|
7245 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
7246 }
|
|
|
7247 else
|
|
|
7248 {
|
|
|
7249 byts = fbyts;
|
|
|
7250 idxs = fidxs;
|
|
|
7251 sp->ts_prefixdepth = NOPREFIX;
|
|
|
7252 }
|
|
|
7253
|
|
|
7254 /*
|
|
330
|
7255 * Loop to find all suggestions. At each round we either:
|
|
|
7256 * - For the current state try one operation, advance "ts_curi",
|
|
|
7257 * increase "depth".
|
|
|
7258 * - When a state is done go to the next, set "ts_state".
|
|
|
7259 * - When all states are tried decrease "depth".
|
|
|
7260 */
|
|
323
|
7261 while (depth >= 0 && !got_int)
|
|
|
7262 {
|
|
|
7263 sp = &stack[depth];
|
|
|
7264 switch (sp->ts_state)
|
|
|
7265 {
|
|
|
7266 case STATE_START:
|
|
375
|
7267 case STATE_NOPREFIX:
|
|
323
|
7268 /*
|
|
|
7269 * Start of node: Deal with NUL bytes, which means
|
|
|
7270 * tword[] may end here.
|
|
|
7271 */
|
|
|
7272 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
7273 len = byts[arridx]; /* bytes in this node */
|
|
|
7274 arridx += sp->ts_curi; /* index of current byte */
|
|
|
7275
|
|
375
|
7276 if (sp->ts_prefixdepth == PREFIXTREE)
|
|
|
7277 {
|
|
|
7278 /* Skip over the NUL bytes, we use them later. */
|
|
|
7279 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
7280 ;
|
|
|
7281 sp->ts_curi += n;
|
|
|
7282
|
|
390
|
7283 /* Always past NUL bytes now. */
|
|
|
7284 n = (int)sp->ts_state;
|
|
|
7285 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7286 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
7287
|
|
375
|
7288 /* At end of a prefix or at start of prefixtree: check for
|
|
|
7289 * following word. */
|
|
390
|
7290 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
7291 {
|
|
455
|
7292 /* Set su->su_badflags to the caps type at this
|
|
|
7293 * position. Use the caps type until here for the
|
|
|
7294 * prefix itself. */
|
|
|
7295 #ifdef FEAT_MBYTE
|
|
|
7296 if (has_mbyte)
|
|
|
7297 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
7298 else
|
|
|
7299 #endif
|
|
|
7300 n = sp->ts_fidx;
|
|
|
7301 flags = captype(su->su_badptr, su->su_badptr + n);
|
|
|
7302 su->su_badflags = captype(su->su_badptr + n,
|
|
|
7303 su->su_badptr + su->su_badlen);
|
|
|
7304
|
|
375
|
7305 ++depth;
|
|
|
7306 stack[depth] = stack[depth - 1];
|
|
|
7307 sp = &stack[depth];
|
|
|
7308 sp->ts_prefixdepth = depth - 1;
|
|
|
7309 byts = fbyts;
|
|
|
7310 idxs = fidxs;
|
|
|
7311 sp->ts_state = STATE_START;
|
|
|
7312 sp->ts_curi = 1; /* start just after length byte */
|
|
|
7313 sp->ts_arridx = 0;
|
|
|
7314
|
|
455
|
7315 /* Move the prefix to preword[] with the right case
|
|
|
7316 * and make find_keepcap_word() works. */
|
|
|
7317 splitoff = sp->ts_twordlen;
|
|
|
7318 tword[splitoff] = NUL;
|
|
|
7319 make_case_word(tword, preword, flags);
|
|
|
7320 prewordlen = STRLEN(preword);
|
|
375
|
7321 }
|
|
|
7322 break;
|
|
|
7323 }
|
|
|
7324
|
|
346
|
7325 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
7326 {
|
|
|
7327 /* Past bytes in node and/or past NUL bytes. */
|
|
|
7328 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7329 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7330 break;
|
|
|
7331 }
|
|
|
7332
|
|
|
7333 /*
|
|
|
7334 * End of word in tree.
|
|
|
7335 */
|
|
|
7336 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
7337
|
|
324
|
7338 flags = (int)idxs[arridx];
|
|
323
|
7339
|
|
375
|
7340 if (sp->ts_prefixdepth < MAXWLEN)
|
|
|
7341 {
|
|
|
7342 /* There was a prefix before the word. Check that the
|
|
|
7343 * prefix can be used with this word. */
|
|
|
7344 /* Count the length of the NULs in the prefix. If there
|
|
|
7345 * are none this must be the first try without a prefix.
|
|
|
7346 */
|
|
|
7347 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
7348 len = pbyts[n++];
|
|
|
7349 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
7350 ;
|
|
|
7351 if (c > 0)
|
|
|
7352 {
|
|
390
|
7353 /* The prefix ID is stored three bytes above the
|
|
|
7354 * flags. */
|
|
|
7355 c = valid_word_prefix(c, n, flags,
|
|
455
|
7356 tword + splitoff, lp->lp_slang, FALSE);
|
|
375
|
7357 if (c == 0)
|
|
|
7358 break;
|
|
|
7359
|
|
|
7360 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
7361 if (c & WF_RAREPFX)
|
|
|
7362 flags |= WF_RARE;
|
|
|
7363 }
|
|
|
7364 }
|
|
|
7365
|
|
323
|
7366 /*
|
|
|
7367 * Form the word with proper case in preword.
|
|
|
7368 * If there is a word from a previous split, append.
|
|
|
7369 */
|
|
|
7370 tword[sp->ts_twordlen] = NUL;
|
|
|
7371 if (flags & WF_KEEPCAP)
|
|
|
7372 /* Must find the word in the keep-case tree. */
|
|
|
7373 find_keepcap_word(lp->lp_slang, tword + splitoff,
|
|
|
7374 preword + prewordlen);
|
|
|
7375 else
|
|
346
|
7376 {
|
|
323
|
7377 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
7378 * use that for the goodword too. But if the badword is
|
|
|
7379 * allcap and it's only one char long use onecap. */
|
|
|
7380 c = su->su_badflags;
|
|
|
7381 if ((c & WF_ALLCAP)
|
|
|
7382 #ifdef FEAT_MBYTE
|
|
|
7383 && su->su_badlen == mb_ptr2len_check(su->su_badptr)
|
|
|
7384 #else
|
|
|
7385 && su->su_badlen == 1
|
|
|
7386 #endif
|
|
|
7387 )
|
|
|
7388 c = WF_ONECAP;
|
|
323
|
7389 make_case_word(tword + splitoff,
|
|
346
|
7390 preword + prewordlen, flags | c);
|
|
|
7391 }
|
|
323
|
7392
|
|
|
7393 /* Don't use a banned word. It may appear again as a good
|
|
|
7394 * word, thus remember it. */
|
|
|
7395 if (flags & WF_BANNED)
|
|
|
7396 {
|
|
|
7397 add_banned(su, preword + prewordlen);
|
|
|
7398 break;
|
|
|
7399 }
|
|
372
|
7400 if (was_banned(su, preword + prewordlen)
|
|
|
7401 || was_banned(su, preword))
|
|
323
|
7402 break;
|
|
|
7403
|
|
|
7404 newscore = 0;
|
|
|
7405 if ((flags & WF_REGION)
|
|
390
|
7406 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
7407 newscore += SCORE_REGION;
|
|
|
7408 if (flags & WF_RARE)
|
|
|
7409 newscore += SCORE_RARE;
|
|
|
7410
|
|
346
|
7411 if (!spell_valid_case(su->su_badflags,
|
|
323
|
7412 captype(preword + prewordlen, NULL)))
|
|
|
7413 newscore += SCORE_ICASE;
|
|
|
7414
|
|
346
|
7415 if ((fword[sp->ts_fidx] == NUL
|
|
376
|
7416 || !spell_iswordp(fword + sp->ts_fidx, curbuf))
|
|
346
|
7417 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
7418 {
|
|
366
|
7419 /* The badword also ends: add suggestions. Give a penalty
|
|
|
7420 * when changing non-word char to word char, e.g., "thes,"
|
|
|
7421 * -> "these". */
|
|
|
7422 p = fword + sp->ts_fidx;
|
|
|
7423 #ifdef FEAT_MBYTE
|
|
|
7424 if (has_mbyte)
|
|
|
7425 mb_ptr_back(fword, p);
|
|
|
7426 else
|
|
|
7427 #endif
|
|
|
7428 --p;
|
|
376
|
7429 if (!spell_iswordp(p, curbuf))
|
|
366
|
7430 {
|
|
|
7431 p = preword + STRLEN(preword);
|
|
|
7432 #ifdef FEAT_MBYTE
|
|
|
7433 if (has_mbyte)
|
|
|
7434 mb_ptr_back(preword, p);
|
|
|
7435 else
|
|
|
7436 #endif
|
|
|
7437 --p;
|
|
376
|
7438 if (spell_iswordp(p, curbuf))
|
|
366
|
7439 newscore += SCORE_NONWORD;
|
|
|
7440 }
|
|
|
7441
|
|
344
|
7442 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
7443 sp->ts_fidx - repextra,
|
|
351
|
7444 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
7445 }
|
|
330
|
7446 else if (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
7447 #ifdef FEAT_MBYTE
|
|
|
7448 /* Don't split halfway a character. */
|
|
|
7449 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
7450 #endif
|
|
|
7451 )
|
|
323
|
7452 {
|
|
|
7453 /* The word in the tree ends but the badword
|
|
|
7454 * continues: try inserting a space and check that a valid
|
|
|
7455 * words starts at fword[sp->ts_fidx]. */
|
|
|
7456 if (try_deeper(su, stack, depth, newscore + SCORE_SPLIT))
|
|
|
7457 {
|
|
|
7458 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
|
7459 sp->ts_save_prewordlen = prewordlen;
|
|
346
|
7460 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7461 sp->ts_save_splitoff = splitoff;
|
|
376
|
7462 sp->ts_state = STATE_SPLITUNDO;
|
|
|
7463
|
|
|
7464 ++depth;
|
|
|
7465 sp = &stack[depth];
|
|
323
|
7466
|
|
|
7467 /* Append a space to preword. */
|
|
|
7468 STRCAT(preword, " ");
|
|
|
7469 prewordlen = STRLEN(preword);
|
|
|
7470 splitoff = sp->ts_twordlen;
|
|
376
|
7471
|
|
|
7472 /* If the badword has a non-word character at this
|
|
|
7473 * position skip it. That means replacing the
|
|
|
7474 * non-word character with a space. */
|
|
|
7475 if (!spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
7476 {
|
|
|
7477 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
7478 #ifdef FEAT_MBYTE
|
|
|
7479 if (has_mbyte)
|
|
|
7480 sp->ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
|
7481 else
|
|
|
7482 #endif
|
|
|
7483 ++sp->ts_fidx;
|
|
|
7484 }
|
|
455
|
7485
|
|
|
7486 /* set su->su_badflags to the caps type at this
|
|
|
7487 * position */
|
|
|
7488
|
|
324
|
7489 #ifdef FEAT_MBYTE
|
|
|
7490 if (has_mbyte)
|
|
455
|
7491 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
7492 else
|
|
|
7493 #endif
|
|
455
|
7494 n = sp->ts_fidx;
|
|
|
7495 su->su_badflags = captype(su->su_badptr + n,
|
|
|
7496 su->su_badptr + su->su_badlen);
|
|
323
|
7497
|
|
|
7498 /* Restart at top of the tree. */
|
|
376
|
7499 sp->ts_arridx = 0;
|
|
323
|
7500 }
|
|
|
7501 }
|
|
|
7502 break;
|
|
|
7503
|
|
|
7504 case STATE_SPLITUNDO:
|
|
346
|
7505 /* Undo the changes done for word split. */
|
|
|
7506 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
7507 splitoff = sp->ts_save_splitoff;
|
|
|
7508 prewordlen = sp->ts_save_prewordlen;
|
|
|
7509
|
|
|
7510 /* Continue looking for NUL bytes. */
|
|
|
7511 sp->ts_state = STATE_START;
|
|
|
7512 break;
|
|
|
7513
|
|
|
7514 case STATE_ENDNUL:
|
|
|
7515 /* Past the NUL bytes in the node. */
|
|
455
|
7516 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
7517 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
7518 {
|
|
|
7519 /* The badword ends, can't use the bytes in this node. */
|
|
|
7520 sp->ts_state = STATE_DEL;
|
|
|
7521 break;
|
|
|
7522 }
|
|
|
7523 sp->ts_state = STATE_PLAIN;
|
|
|
7524 /*FALLTHROUGH*/
|
|
|
7525
|
|
|
7526 case STATE_PLAIN:
|
|
|
7527 /*
|
|
|
7528 * Go over all possible bytes at this node, add each to
|
|
|
7529 * tword[] and use child node. "ts_curi" is the index.
|
|
|
7530 */
|
|
|
7531 arridx = sp->ts_arridx;
|
|
|
7532 if (sp->ts_curi > byts[arridx])
|
|
|
7533 {
|
|
|
7534 /* Done all bytes at this node, do next state. When still
|
|
|
7535 * at already changed bytes skip the other tricks. */
|
|
|
7536 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7537 sp->ts_state = STATE_DEL;
|
|
|
7538 else
|
|
|
7539 sp->ts_state = STATE_FINAL;
|
|
|
7540 }
|
|
|
7541 else
|
|
|
7542 {
|
|
|
7543 arridx += sp->ts_curi++;
|
|
|
7544 c = byts[arridx];
|
|
|
7545
|
|
|
7546 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
7547 * the byte in the bad word adjust the score. But don't
|
|
|
7548 * even try when the byte was already changed. */
|
|
330
|
7549 if (c == fword[sp->ts_fidx]
|
|
|
7550 #ifdef FEAT_MBYTE
|
|
|
7551 || (sp->ts_tcharlen > 0
|
|
|
7552 && sp->ts_isdiff != DIFF_NONE)
|
|
|
7553 #endif
|
|
|
7554 )
|
|
323
|
7555 newscore = 0;
|
|
|
7556 else
|
|
|
7557 newscore = SCORE_SUBST;
|
|
|
7558 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7559 && try_deeper(su, stack, depth, newscore))
|
|
|
7560 {
|
|
|
7561 ++depth;
|
|
330
|
7562 sp = &stack[depth];
|
|
|
7563 ++sp->ts_fidx;
|
|
|
7564 tword[sp->ts_twordlen++] = c;
|
|
|
7565 sp->ts_arridx = idxs[arridx];
|
|
|
7566 #ifdef FEAT_MBYTE
|
|
|
7567 if (newscore == SCORE_SUBST)
|
|
|
7568 sp->ts_isdiff = DIFF_YES;
|
|
|
7569 if (has_mbyte)
|
|
|
7570 {
|
|
|
7571 /* Multi-byte characters are a bit complicated to
|
|
|
7572 * handle: They differ when any of the bytes
|
|
|
7573 * differ and then their length may also differ. */
|
|
|
7574 if (sp->ts_tcharlen == 0)
|
|
|
7575 {
|
|
|
7576 /* First byte. */
|
|
|
7577 sp->ts_tcharidx = 0;
|
|
|
7578 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
7579 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
7580 sp->ts_isdiff = (newscore != 0)
|
|
|
7581 ? DIFF_YES : DIFF_NONE;
|
|
|
7582 }
|
|
|
7583 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
7584 /* When inserting trail bytes don't advance in
|
|
|
7585 * the bad word. */
|
|
|
7586 --sp->ts_fidx;
|
|
|
7587 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
7588 {
|
|
|
7589 /* Last byte of character. */
|
|
|
7590 if (sp->ts_isdiff == DIFF_YES)
|
|
|
7591 {
|
|
|
7592 /* Correct ts_fidx for the byte length of
|
|
|
7593 * the character (we didn't check that
|
|
|
7594 * before). */
|
|
|
7595 sp->ts_fidx = sp->ts_fcharstart
|
|
|
7596 + MB_BYTE2LEN(
|
|
|
7597 fword[sp->ts_fcharstart]);
|
|
|
7598
|
|
|
7599 /* For a similar character adjust score
|
|
|
7600 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
|
7601 if (lp->lp_slang->sl_has_map
|
|
|
7602 && similar_chars(lp->lp_slang,
|
|
|
7603 mb_ptr2char(tword
|
|
|
7604 + sp->ts_twordlen
|
|
|
7605 - sp->ts_tcharlen),
|
|
|
7606 mb_ptr2char(fword
|
|
|
7607 + sp->ts_fcharstart)))
|
|
|
7608 sp->ts_score -=
|
|
|
7609 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7610 }
|
|
358
|
7611 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
7612 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
7613 {
|
|
|
7614 /* If the previous character was the same,
|
|
|
7615 * thus doubling a character, give a bonus
|
|
|
7616 * to the score. */
|
|
|
7617 p = tword + sp->ts_twordlen
|
|
|
7618 - sp->ts_tcharlen;
|
|
|
7619 c = mb_ptr2char(p);
|
|
|
7620 mb_ptr_back(tword, p);
|
|
|
7621 if (c == mb_ptr2char(p))
|
|
|
7622 sp->ts_score -= SCORE_INS
|
|
|
7623 - SCORE_INSDUP;
|
|
|
7624 }
|
|
330
|
7625
|
|
|
7626 /* Starting a new char, reset the length. */
|
|
|
7627 sp->ts_tcharlen = 0;
|
|
|
7628 }
|
|
|
7629 }
|
|
|
7630 else
|
|
|
7631 #endif
|
|
|
7632 {
|
|
|
7633 /* If we found a similar char adjust the score.
|
|
|
7634 * We do this after calling try_deeper() because
|
|
|
7635 * it's slow. */
|
|
|
7636 if (newscore != 0
|
|
|
7637 && lp->lp_slang->sl_has_map
|
|
|
7638 && similar_chars(lp->lp_slang,
|
|
|
7639 c, fword[sp->ts_fidx - 1]))
|
|
|
7640 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7641 }
|
|
323
|
7642 }
|
|
|
7643 }
|
|
|
7644 break;
|
|
|
7645
|
|
|
7646 case STATE_DEL:
|
|
330
|
7647 #ifdef FEAT_MBYTE
|
|
|
7648 /* When past the first byte of a multi-byte char don't try
|
|
|
7649 * delete/insert/swap a character. */
|
|
|
7650 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
7651 {
|
|
|
7652 sp->ts_state = STATE_FINAL;
|
|
|
7653 break;
|
|
|
7654 }
|
|
|
7655 #endif
|
|
|
7656 /*
|
|
|
7657 * Try skipping one character in the bad word (delete it).
|
|
|
7658 */
|
|
323
|
7659 sp->ts_state = STATE_INS;
|
|
|
7660 sp->ts_curi = 1;
|
|
|
7661 if (fword[sp->ts_fidx] != NUL
|
|
|
7662 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
7663 {
|
|
|
7664 ++depth;
|
|
358
|
7665
|
|
|
7666 /* Advance over the character in fword[]. Give a bonus to
|
|
|
7667 * the score if the same character is following "nn" ->
|
|
|
7668 * "n". */
|
|
330
|
7669 #ifdef FEAT_MBYTE
|
|
|
7670 if (has_mbyte)
|
|
358
|
7671 {
|
|
|
7672 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
7673 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
358
|
7674 if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
|
7675 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7676 }
|
|
330
|
7677 else
|
|
|
7678 #endif
|
|
358
|
7679 {
|
|
330
|
7680 ++stack[depth].ts_fidx;
|
|
358
|
7681 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
7682 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7683 }
|
|
323
|
7684 break;
|
|
|
7685 }
|
|
|
7686 /*FALLTHROUGH*/
|
|
|
7687
|
|
|
7688 case STATE_INS:
|
|
330
|
7689 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
7690 * node. */
|
|
|
7691 n = sp->ts_arridx;
|
|
|
7692 if (sp->ts_curi > byts[n])
|
|
|
7693 {
|
|
|
7694 /* Done all bytes at this node, do next state. */
|
|
|
7695 sp->ts_state = STATE_SWAP;
|
|
|
7696 }
|
|
|
7697 else
|
|
|
7698 {
|
|
330
|
7699 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
7700 n += sp->ts_curi++;
|
|
|
7701 c = byts[n];
|
|
|
7702 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
7703 {
|
|
|
7704 ++depth;
|
|
330
|
7705 sp = &stack[depth];
|
|
|
7706 tword[sp->ts_twordlen++] = c;
|
|
|
7707 sp->ts_arridx = idxs[n];
|
|
|
7708 #ifdef FEAT_MBYTE
|
|
|
7709 if (has_mbyte)
|
|
|
7710 {
|
|
|
7711 fl = MB_BYTE2LEN(c);
|
|
|
7712 if (fl > 1)
|
|
|
7713 {
|
|
|
7714 /* There are following bytes for the same
|
|
|
7715 * character. We must find all bytes before
|
|
|
7716 * trying delete/insert/swap/etc. */
|
|
|
7717 sp->ts_tcharlen = fl;
|
|
|
7718 sp->ts_tcharidx = 1;
|
|
|
7719 sp->ts_isdiff = DIFF_INSERT;
|
|
|
7720 }
|
|
|
7721 }
|
|
358
|
7722 else
|
|
|
7723 fl = 1;
|
|
|
7724 if (fl == 1)
|
|
330
|
7725 #endif
|
|
358
|
7726 {
|
|
|
7727 /* If the previous character was the same, thus
|
|
|
7728 * doubling a character, give a bonus to the
|
|
|
7729 * score. */
|
|
|
7730 if (sp->ts_twordlen >= 2
|
|
|
7731 && tword[sp->ts_twordlen - 2] == c)
|
|
|
7732 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
7733 }
|
|
323
|
7734 }
|
|
|
7735 }
|
|
|
7736 break;
|
|
|
7737
|
|
|
7738 case STATE_SWAP:
|
|
330
|
7739 /*
|
|
|
7740 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
7741 * We change "fword" here, it's changed back afterwards.
|
|
|
7742 */
|
|
|
7743 p = fword + sp->ts_fidx;
|
|
|
7744 c = *p;
|
|
|
7745 if (c == NUL)
|
|
|
7746 {
|
|
|
7747 /* End of word, can't swap or replace. */
|
|
|
7748 sp->ts_state = STATE_FINAL;
|
|
|
7749 break;
|
|
|
7750 }
|
|
|
7751 #ifdef FEAT_MBYTE
|
|
|
7752 if (has_mbyte)
|
|
|
7753 {
|
|
|
7754 n = mb_ptr2len_check(p);
|
|
|
7755 c = mb_ptr2char(p);
|
|
|
7756 c2 = mb_ptr2char(p + n);
|
|
|
7757 }
|
|
|
7758 else
|
|
|
7759 #endif
|
|
|
7760 c2 = p[1];
|
|
|
7761 if (c == c2)
|
|
323
|
7762 {
|
|
330
|
7763 /* Characters are identical, swap won't do anything. */
|
|
|
7764 sp->ts_state = STATE_SWAP3;
|
|
|
7765 break;
|
|
|
7766 }
|
|
|
7767 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
7768 {
|
|
|
7769 sp->ts_state = STATE_UNSWAP;
|
|
323
|
7770 ++depth;
|
|
330
|
7771 #ifdef FEAT_MBYTE
|
|
|
7772 if (has_mbyte)
|
|
|
7773 {
|
|
|
7774 fl = mb_char2len(c2);
|
|
|
7775 mch_memmove(p, p + n, fl);
|
|
|
7776 mb_char2bytes(c, p + fl);
|
|
|
7777 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7778 }
|
|
|
7779 else
|
|
|
7780 #endif
|
|
|
7781 {
|
|
|
7782 p[0] = c2;
|
|
|
7783 p[1] = c;
|
|
|
7784 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
7785 }
|
|
323
|
7786 }
|
|
|
7787 else
|
|
|
7788 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
7789 sp->ts_state = STATE_REP_INI;
|
|
|
7790 break;
|
|
|
7791
|
|
330
|
7792 case STATE_UNSWAP:
|
|
|
7793 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
7794 p = fword + sp->ts_fidx;
|
|
|
7795 #ifdef FEAT_MBYTE
|
|
|
7796 if (has_mbyte)
|
|
|
7797 {
|
|
|
7798 n = MB_BYTE2LEN(*p);
|
|
|
7799 c = mb_ptr2char(p + n);
|
|
|
7800 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
7801 mb_char2bytes(c, p);
|
|
|
7802 }
|
|
|
7803 else
|
|
|
7804 #endif
|
|
|
7805 {
|
|
|
7806 c = *p;
|
|
|
7807 *p = p[1];
|
|
|
7808 p[1] = c;
|
|
|
7809 }
|
|
|
7810 /*FALLTHROUGH*/
|
|
|
7811
|
|
|
7812 case STATE_SWAP3:
|
|
323
|
7813 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
7814 * "fword" here, it's changed back afterwards. */
|
|
|
7815 p = fword + sp->ts_fidx;
|
|
|
7816 #ifdef FEAT_MBYTE
|
|
|
7817 if (has_mbyte)
|
|
|
7818 {
|
|
|
7819 n = mb_ptr2len_check(p);
|
|
|
7820 c = mb_ptr2char(p);
|
|
|
7821 fl = mb_ptr2len_check(p + n);
|
|
|
7822 c2 = mb_ptr2char(p + n);
|
|
|
7823 c3 = mb_ptr2char(p + n + fl);
|
|
|
7824 }
|
|
|
7825 else
|
|
|
7826 #endif
|
|
323
|
7827 {
|
|
330
|
7828 c = *p;
|
|
|
7829 c2 = p[1];
|
|
|
7830 c3 = p[2];
|
|
|
7831 }
|
|
|
7832
|
|
|
7833 /* When characters are identical: "121" then SWAP3 result is
|
|
|
7834 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
7835 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
7836 * swapping. Also skip when c3 is NUL. */
|
|
|
7837 if (c == c3 || c3 == NUL)
|
|
|
7838 {
|
|
|
7839 sp->ts_state = STATE_REP_INI;
|
|
|
7840 break;
|
|
|
7841 }
|
|
|
7842 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7843 {
|
|
|
7844 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
7845 ++depth;
|
|
330
|
7846 #ifdef FEAT_MBYTE
|
|
|
7847 if (has_mbyte)
|
|
|
7848 {
|
|
|
7849 tl = mb_char2len(c3);
|
|
|
7850 mch_memmove(p, p + n + fl, tl);
|
|
|
7851 mb_char2bytes(c2, p + tl);
|
|
|
7852 mb_char2bytes(c, p + fl + tl);
|
|
|
7853 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
7854 }
|
|
|
7855 else
|
|
|
7856 #endif
|
|
|
7857 {
|
|
|
7858 p[0] = p[2];
|
|
|
7859 p[2] = c;
|
|
|
7860 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7861 }
|
|
323
|
7862 }
|
|
|
7863 else
|
|
|
7864 sp->ts_state = STATE_REP_INI;
|
|
|
7865 break;
|
|
|
7866
|
|
330
|
7867 case STATE_UNSWAP3:
|
|
|
7868 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
7869 p = fword + sp->ts_fidx;
|
|
|
7870 #ifdef FEAT_MBYTE
|
|
|
7871 if (has_mbyte)
|
|
|
7872 {
|
|
|
7873 n = MB_BYTE2LEN(*p);
|
|
|
7874 c2 = mb_ptr2char(p + n);
|
|
|
7875 fl = MB_BYTE2LEN(p[n]);
|
|
|
7876 c = mb_ptr2char(p + n + fl);
|
|
|
7877 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
7878 mch_memmove(p + fl + tl, p, n);
|
|
|
7879 mb_char2bytes(c, p);
|
|
|
7880 mb_char2bytes(c2, p + tl);
|
|
|
7881 }
|
|
|
7882 else
|
|
|
7883 #endif
|
|
|
7884 {
|
|
|
7885 c = *p;
|
|
|
7886 *p = p[2];
|
|
|
7887 p[2] = c;
|
|
|
7888 }
|
|
346
|
7889
|
|
330
|
7890 /* Rotate three characters left: "123" -> "231". We change
|
|
|
7891 * "fword" here, it's changed back afterwards. */
|
|
323
|
7892 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7893 {
|
|
330
|
7894 sp->ts_state = STATE_UNROT3L;
|
|
323
|
7895 ++depth;
|
|
330
|
7896 p = fword + sp->ts_fidx;
|
|
|
7897 #ifdef FEAT_MBYTE
|
|
|
7898 if (has_mbyte)
|
|
|
7899 {
|
|
|
7900 n = mb_ptr2len_check(p);
|
|
|
7901 c = mb_ptr2char(p);
|
|
|
7902 fl = mb_ptr2len_check(p + n);
|
|
|
7903 fl += mb_ptr2len_check(p + n + fl);
|
|
|
7904 mch_memmove(p, p + n, fl);
|
|
|
7905 mb_char2bytes(c, p + fl);
|
|
|
7906 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7907 }
|
|
|
7908 else
|
|
|
7909 #endif
|
|
|
7910 {
|
|
|
7911 c = *p;
|
|
|
7912 *p = p[1];
|
|
|
7913 p[1] = p[2];
|
|
|
7914 p[2] = c;
|
|
|
7915 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7916 }
|
|
323
|
7917 }
|
|
|
7918 else
|
|
|
7919 sp->ts_state = STATE_REP_INI;
|
|
|
7920 break;
|
|
|
7921
|
|
330
|
7922 case STATE_UNROT3L:
|
|
346
|
7923 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
7924 p = fword + sp->ts_fidx;
|
|
|
7925 #ifdef FEAT_MBYTE
|
|
|
7926 if (has_mbyte)
|
|
|
7927 {
|
|
|
7928 n = MB_BYTE2LEN(*p);
|
|
|
7929 n += MB_BYTE2LEN(p[n]);
|
|
|
7930 c = mb_ptr2char(p + n);
|
|
|
7931 tl = MB_BYTE2LEN(p[n]);
|
|
|
7932 mch_memmove(p + tl, p, n);
|
|
|
7933 mb_char2bytes(c, p);
|
|
|
7934 }
|
|
|
7935 else
|
|
|
7936 #endif
|
|
|
7937 {
|
|
|
7938 c = p[2];
|
|
|
7939 p[2] = p[1];
|
|
|
7940 p[1] = *p;
|
|
|
7941 *p = c;
|
|
|
7942 }
|
|
346
|
7943
|
|
323
|
7944 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
7945 * "fword" here, it's changed back afterwards. */
|
|
323
|
7946 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7947 {
|
|
330
|
7948 sp->ts_state = STATE_UNROT3R;
|
|
323
|
7949 ++depth;
|
|
330
|
7950 p = fword + sp->ts_fidx;
|
|
|
7951 #ifdef FEAT_MBYTE
|
|
|
7952 if (has_mbyte)
|
|
|
7953 {
|
|
|
7954 n = mb_ptr2len_check(p);
|
|
|
7955 n += mb_ptr2len_check(p + n);
|
|
|
7956 c = mb_ptr2char(p + n);
|
|
|
7957 tl = mb_ptr2len_check(p + n);
|
|
|
7958 mch_memmove(p + tl, p, n);
|
|
|
7959 mb_char2bytes(c, p);
|
|
|
7960 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
7961 }
|
|
|
7962 else
|
|
|
7963 #endif
|
|
|
7964 {
|
|
|
7965 c = p[2];
|
|
|
7966 p[2] = p[1];
|
|
|
7967 p[1] = *p;
|
|
|
7968 *p = c;
|
|
|
7969 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7970 }
|
|
323
|
7971 }
|
|
|
7972 else
|
|
|
7973 sp->ts_state = STATE_REP_INI;
|
|
|
7974 break;
|
|
|
7975
|
|
330
|
7976 case STATE_UNROT3R:
|
|
346
|
7977 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
7978 p = fword + sp->ts_fidx;
|
|
|
7979 #ifdef FEAT_MBYTE
|
|
|
7980 if (has_mbyte)
|
|
|
7981 {
|
|
|
7982 c = mb_ptr2char(p);
|
|
|
7983 tl = MB_BYTE2LEN(*p);
|
|
|
7984 n = MB_BYTE2LEN(p[tl]);
|
|
|
7985 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
7986 mch_memmove(p, p + tl, n);
|
|
|
7987 mb_char2bytes(c, p + n);
|
|
|
7988 }
|
|
|
7989 else
|
|
|
7990 #endif
|
|
|
7991 {
|
|
|
7992 c = *p;
|
|
|
7993 *p = p[1];
|
|
|
7994 p[1] = p[2];
|
|
|
7995 p[2] = c;
|
|
|
7996 }
|
|
323
|
7997 /*FALLTHROUGH*/
|
|
|
7998
|
|
|
7999 case STATE_REP_INI:
|
|
|
8000 /* Check if matching with REP items from the .aff file would
|
|
|
8001 * work. Quickly skip if there are no REP items or the score
|
|
|
8002 * is going to be too high anyway. */
|
|
|
8003 gap = &lp->lp_slang->sl_rep;
|
|
|
8004 if (gap->ga_len == 0
|
|
|
8005 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
8006 {
|
|
|
8007 sp->ts_state = STATE_FINAL;
|
|
|
8008 break;
|
|
|
8009 }
|
|
|
8010
|
|
|
8011 /* Use the first byte to quickly find the first entry that
|
|
330
|
8012 * may match. If the index is -1 there is none. */
|
|
323
|
8013 sp->ts_curi = lp->lp_slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
8014 if (sp->ts_curi < 0)
|
|
|
8015 {
|
|
|
8016 sp->ts_state = STATE_FINAL;
|
|
|
8017 break;
|
|
|
8018 }
|
|
|
8019
|
|
|
8020 sp->ts_state = STATE_REP;
|
|
|
8021 /*FALLTHROUGH*/
|
|
|
8022
|
|
|
8023 case STATE_REP:
|
|
|
8024 /* Try matching with REP items from the .aff file. For each
|
|
330
|
8025 * match replace the characters and check if the resulting
|
|
|
8026 * word is valid. */
|
|
323
|
8027 p = fword + sp->ts_fidx;
|
|
|
8028
|
|
|
8029 gap = &lp->lp_slang->sl_rep;
|
|
|
8030 while (sp->ts_curi < gap->ga_len)
|
|
|
8031 {
|
|
|
8032 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
8033 if (*ftp->ft_from != *p)
|
|
|
8034 {
|
|
|
8035 /* past possible matching entries */
|
|
|
8036 sp->ts_curi = gap->ga_len;
|
|
|
8037 break;
|
|
|
8038 }
|
|
|
8039 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
8040 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
8041 {
|
|
|
8042 /* Need to undo this afterwards. */
|
|
|
8043 sp->ts_state = STATE_REP_UNDO;
|
|
|
8044
|
|
|
8045 /* Change the "from" to the "to" string. */
|
|
|
8046 ++depth;
|
|
|
8047 fl = STRLEN(ftp->ft_from);
|
|
|
8048 tl = STRLEN(ftp->ft_to);
|
|
|
8049 if (fl != tl)
|
|
346
|
8050 {
|
|
323
|
8051 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
8052 repextra += tl - fl;
|
|
|
8053 }
|
|
323
|
8054 mch_memmove(p, ftp->ft_to, tl);
|
|
|
8055 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
8056 #ifdef FEAT_MBYTE
|
|
|
8057 stack[depth].ts_tcharlen = 0;
|
|
|
8058 #endif
|
|
323
|
8059 break;
|
|
|
8060 }
|
|
|
8061 }
|
|
|
8062
|
|
|
8063 if (sp->ts_curi >= gap->ga_len)
|
|
|
8064 /* No (more) matches. */
|
|
|
8065 sp->ts_state = STATE_FINAL;
|
|
|
8066
|
|
|
8067 break;
|
|
|
8068
|
|
|
8069 case STATE_REP_UNDO:
|
|
|
8070 /* Undo a REP replacement and continue with the next one. */
|
|
|
8071 ftp = (fromto_T *)lp->lp_slang->sl_rep.ga_data
|
|
|
8072 + sp->ts_curi - 1;
|
|
|
8073 fl = STRLEN(ftp->ft_from);
|
|
|
8074 tl = STRLEN(ftp->ft_to);
|
|
|
8075 p = fword + sp->ts_fidx;
|
|
|
8076 if (fl != tl)
|
|
346
|
8077 {
|
|
323
|
8078 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
8079 repextra -= tl - fl;
|
|
|
8080 }
|
|
323
|
8081 mch_memmove(p, ftp->ft_from, fl);
|
|
|
8082 sp->ts_state = STATE_REP;
|
|
|
8083 break;
|
|
|
8084
|
|
|
8085 default:
|
|
|
8086 /* Did all possible states at this level, go up one level. */
|
|
|
8087 --depth;
|
|
344
|
8088
|
|
375
|
8089 if (depth >= 0 && stack[depth].ts_prefixdepth == PREFIXTREE)
|
|
|
8090 {
|
|
|
8091 /* Continue in or go back to the prefix tree. */
|
|
|
8092 byts = pbyts;
|
|
|
8093 idxs = pidxs;
|
|
|
8094 splitoff = 0;
|
|
|
8095 }
|
|
|
8096
|
|
344
|
8097 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
8098 line_breakcheck();
|
|
323
|
8099 }
|
|
|
8100 }
|
|
|
8101 }
|
|
|
8102 }
|
|
|
8103
|
|
|
8104 /*
|
|
|
8105 * Try going one level deeper in the tree.
|
|
|
8106 */
|
|
|
8107 static int
|
|
|
8108 try_deeper(su, stack, depth, score_add)
|
|
|
8109 suginfo_T *su;
|
|
|
8110 trystate_T *stack;
|
|
|
8111 int depth;
|
|
|
8112 int score_add;
|
|
|
8113 {
|
|
|
8114 int newscore;
|
|
|
8115
|
|
|
8116 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
8117 newscore = stack[depth].ts_score + score_add;
|
|
|
8118 if (newscore >= su->su_maxscore)
|
|
|
8119 return FALSE;
|
|
|
8120
|
|
330
|
8121 stack[depth + 1] = stack[depth];
|
|
323
|
8122 stack[depth + 1].ts_state = STATE_START;
|
|
|
8123 stack[depth + 1].ts_score = newscore;
|
|
|
8124 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
8125 return TRUE;
|
|
|
8126 }
|
|
|
8127
|
|
455
|
8128 #ifdef FEAT_MBYTE
|
|
|
8129 /*
|
|
|
8130 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
8131 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
8132 */
|
|
|
8133 static int
|
|
|
8134 nofold_len(fword, flen, word)
|
|
|
8135 char_u *fword;
|
|
|
8136 int flen;
|
|
|
8137 char_u *word;
|
|
|
8138 {
|
|
|
8139 char_u *p;
|
|
|
8140 int i = 0;
|
|
|
8141
|
|
|
8142 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
8143 ++i;
|
|
|
8144 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
8145 --i;
|
|
|
8146 return (int)(p - word);
|
|
|
8147 }
|
|
|
8148 #endif
|
|
|
8149
|
|
323
|
8150 /*
|
|
|
8151 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
8152 * words and put it in "kword".
|
|
|
8153 * Theoretically there could be several keep-case words that result in the
|
|
|
8154 * same case-folded word, but we only find one...
|
|
|
8155 */
|
|
|
8156 static void
|
|
|
8157 find_keepcap_word(slang, fword, kword)
|
|
|
8158 slang_T *slang;
|
|
|
8159 char_u *fword;
|
|
|
8160 char_u *kword;
|
|
|
8161 {
|
|
|
8162 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
8163 int depth;
|
|
324
|
8164 idx_T tryidx;
|
|
323
|
8165
|
|
|
8166 /* The following arrays are used at each depth in the tree. */
|
|
324
|
8167 idx_T arridx[MAXWLEN];
|
|
323
|
8168 int round[MAXWLEN];
|
|
|
8169 int fwordidx[MAXWLEN];
|
|
|
8170 int uwordidx[MAXWLEN];
|
|
|
8171 int kwordlen[MAXWLEN];
|
|
|
8172
|
|
|
8173 int flen, ulen;
|
|
|
8174 int l;
|
|
|
8175 int len;
|
|
|
8176 int c;
|
|
324
|
8177 idx_T lo, hi, m;
|
|
323
|
8178 char_u *p;
|
|
|
8179 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
8180 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
8181
|
|
|
8182 if (byts == NULL)
|
|
|
8183 {
|
|
|
8184 /* array is empty: "cannot happen" */
|
|
|
8185 *kword = NUL;
|
|
|
8186 return;
|
|
|
8187 }
|
|
|
8188
|
|
|
8189 /* Make an all-cap version of "fword". */
|
|
|
8190 allcap_copy(fword, uword);
|
|
|
8191
|
|
|
8192 /*
|
|
|
8193 * Each character needs to be tried both case-folded and upper-case.
|
|
|
8194 * All this gets very complicated if we keep in mind that changing case
|
|
|
8195 * may change the byte length of a multi-byte character...
|
|
|
8196 */
|
|
|
8197 depth = 0;
|
|
|
8198 arridx[0] = 0;
|
|
|
8199 round[0] = 0;
|
|
|
8200 fwordidx[0] = 0;
|
|
|
8201 uwordidx[0] = 0;
|
|
|
8202 kwordlen[0] = 0;
|
|
|
8203 while (depth >= 0)
|
|
|
8204 {
|
|
|
8205 if (fword[fwordidx[depth]] == NUL)
|
|
|
8206 {
|
|
|
8207 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
8208 * here we have found a match. */
|
|
|
8209 if (byts[arridx[depth] + 1] == 0)
|
|
|
8210 {
|
|
|
8211 kword[kwordlen[depth]] = NUL;
|
|
|
8212 return;
|
|
|
8213 }
|
|
|
8214
|
|
|
8215 /* kword is getting too long, continue one level up */
|
|
|
8216 --depth;
|
|
|
8217 }
|
|
|
8218 else if (++round[depth] > 2)
|
|
|
8219 {
|
|
|
8220 /* tried both fold-case and upper-case character, continue one
|
|
|
8221 * level up */
|
|
|
8222 --depth;
|
|
|
8223 }
|
|
|
8224 else
|
|
|
8225 {
|
|
|
8226 /*
|
|
|
8227 * round[depth] == 1: Try using the folded-case character.
|
|
|
8228 * round[depth] == 2: Try using the upper-case character.
|
|
|
8229 */
|
|
|
8230 #ifdef FEAT_MBYTE
|
|
|
8231 if (has_mbyte)
|
|
|
8232 {
|
|
|
8233 flen = mb_ptr2len_check(fword + fwordidx[depth]);
|
|
|
8234 ulen = mb_ptr2len_check(uword + uwordidx[depth]);
|
|
|
8235 }
|
|
|
8236 else
|
|
|
8237 #endif
|
|
|
8238 ulen = flen = 1;
|
|
|
8239 if (round[depth] == 1)
|
|
|
8240 {
|
|
|
8241 p = fword + fwordidx[depth];
|
|
|
8242 l = flen;
|
|
|
8243 }
|
|
|
8244 else
|
|
|
8245 {
|
|
|
8246 p = uword + uwordidx[depth];
|
|
|
8247 l = ulen;
|
|
|
8248 }
|
|
|
8249
|
|
|
8250 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
8251 {
|
|
|
8252 /* Perform a binary search in the list of accepted bytes. */
|
|
|
8253 len = byts[tryidx++];
|
|
|
8254 c = *p++;
|
|
|
8255 lo = tryidx;
|
|
|
8256 hi = tryidx + len - 1;
|
|
|
8257 while (lo < hi)
|
|
|
8258 {
|
|
|
8259 m = (lo + hi) / 2;
|
|
|
8260 if (byts[m] > c)
|
|
|
8261 hi = m - 1;
|
|
|
8262 else if (byts[m] < c)
|
|
|
8263 lo = m + 1;
|
|
|
8264 else
|
|
|
8265 {
|
|
|
8266 lo = hi = m;
|
|
|
8267 break;
|
|
|
8268 }
|
|
|
8269 }
|
|
|
8270
|
|
|
8271 /* Stop if there is no matching byte. */
|
|
|
8272 if (hi < lo || byts[lo] != c)
|
|
|
8273 break;
|
|
|
8274
|
|
|
8275 /* Continue at the child (if there is one). */
|
|
|
8276 tryidx = idxs[lo];
|
|
|
8277 }
|
|
|
8278
|
|
|
8279 if (l == 0)
|
|
|
8280 {
|
|
|
8281 /*
|
|
|
8282 * Found the matching char. Copy it to "kword" and go a
|
|
|
8283 * level deeper.
|
|
|
8284 */
|
|
|
8285 if (round[depth] == 1)
|
|
|
8286 {
|
|
|
8287 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
8288 flen);
|
|
|
8289 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
8290 }
|
|
|
8291 else
|
|
|
8292 {
|
|
|
8293 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
8294 ulen);
|
|
|
8295 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
8296 }
|
|
|
8297 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
8298 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
8299
|
|
|
8300 ++depth;
|
|
|
8301 arridx[depth] = tryidx;
|
|
|
8302 round[depth] = 0;
|
|
|
8303 }
|
|
|
8304 }
|
|
|
8305 }
|
|
|
8306
|
|
|
8307 /* Didn't find it: "cannot happen". */
|
|
|
8308 *kword = NUL;
|
|
|
8309 }
|
|
|
8310
|
|
|
8311 /*
|
|
344
|
8312 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
8313 * su->su_sga.
|
|
|
8314 */
|
|
|
8315 static void
|
|
|
8316 score_comp_sal(su)
|
|
|
8317 suginfo_T *su;
|
|
|
8318 {
|
|
|
8319 langp_T *lp;
|
|
|
8320 char_u badsound[MAXWLEN];
|
|
|
8321 int i;
|
|
|
8322 suggest_T *stp;
|
|
|
8323 suggest_T *sstp;
|
|
|
8324 int score;
|
|
|
8325
|
|
|
8326 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
8327 return;
|
|
|
8328
|
|
|
8329 /* Use the sound-folding of the first language that supports it. */
|
|
|
8330 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8331 lp->lp_slang != NULL; ++lp)
|
|
|
8332 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8333 {
|
|
|
8334 /* soundfold the bad word */
|
|
375
|
8335 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8336
|
|
|
8337 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8338 {
|
|
|
8339 stp = &SUG(su->su_ga, i);
|
|
|
8340
|
|
351
|
8341 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
8342 * sound-a-like score. */
|
|
|
8343 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
8344 if (score < SCORE_MAXMAX)
|
|
|
8345 {
|
|
|
8346 /* Add the suggestion. */
|
|
|
8347 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
8348 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
8349 if (sstp->st_word != NULL)
|
|
|
8350 {
|
|
|
8351 sstp->st_score = score;
|
|
|
8352 sstp->st_altscore = 0;
|
|
|
8353 sstp->st_orglen = stp->st_orglen;
|
|
|
8354 ++su->su_sga.ga_len;
|
|
|
8355 }
|
|
|
8356 }
|
|
|
8357 }
|
|
|
8358 break;
|
|
|
8359 }
|
|
|
8360 }
|
|
|
8361
|
|
|
8362 /*
|
|
|
8363 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
8364 * They are intwined.
|
|
|
8365 */
|
|
|
8366 static void
|
|
|
8367 score_combine(su)
|
|
|
8368 suginfo_T *su;
|
|
|
8369 {
|
|
|
8370 int i;
|
|
|
8371 int j;
|
|
|
8372 garray_T ga;
|
|
|
8373 garray_T *gap;
|
|
|
8374 langp_T *lp;
|
|
|
8375 suggest_T *stp;
|
|
|
8376 char_u *p;
|
|
|
8377 char_u badsound[MAXWLEN];
|
|
|
8378 int round;
|
|
|
8379
|
|
|
8380 /* Add the alternate score to su_ga. */
|
|
|
8381 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8382 lp->lp_slang != NULL; ++lp)
|
|
|
8383 {
|
|
|
8384 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8385 {
|
|
|
8386 /* soundfold the bad word */
|
|
375
|
8387 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8388
|
|
|
8389 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8390 {
|
|
|
8391 stp = &SUG(su->su_ga, i);
|
|
351
|
8392 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
8393 badsound);
|
|
344
|
8394 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8395 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
8396 else
|
|
|
8397 stp->st_score = (stp->st_score * 3
|
|
|
8398 + stp->st_altscore) / 4;
|
|
|
8399 stp->st_salscore = FALSE;
|
|
|
8400 }
|
|
|
8401 break;
|
|
|
8402 }
|
|
|
8403 }
|
|
|
8404
|
|
|
8405 /* Add the alternate score to su_sga. */
|
|
|
8406 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8407 {
|
|
|
8408 stp = &SUG(su->su_sga, i);
|
|
|
8409 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
8410 if (stp->st_score == SCORE_MAXMAX)
|
|
|
8411 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
8412 else
|
|
|
8413 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
8414 stp->st_salscore = TRUE;
|
|
|
8415 }
|
|
|
8416
|
|
|
8417 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
8418 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8419 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
8420
|
|
|
8421 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
8422 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
8423 return;
|
|
|
8424
|
|
|
8425 stp = &SUG(ga, 0);
|
|
|
8426 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
8427 {
|
|
|
8428 /* round 1: get a suggestion from su_ga
|
|
|
8429 * round 2: get a suggestion from su_sga */
|
|
|
8430 for (round = 1; round <= 2; ++round)
|
|
|
8431 {
|
|
|
8432 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
8433 if (i < gap->ga_len)
|
|
|
8434 {
|
|
|
8435 /* Don't add a word if it's already there. */
|
|
|
8436 p = SUG(*gap, i).st_word;
|
|
|
8437 for (j = 0; j < ga.ga_len; ++j)
|
|
|
8438 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
8439 break;
|
|
|
8440 if (j == ga.ga_len)
|
|
|
8441 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
8442 else
|
|
|
8443 vim_free(p);
|
|
|
8444 }
|
|
|
8445 }
|
|
|
8446 }
|
|
|
8447
|
|
|
8448 ga_clear(&su->su_ga);
|
|
|
8449 ga_clear(&su->su_sga);
|
|
|
8450
|
|
|
8451 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
8452 if (ga.ga_len > su->su_maxcount)
|
|
|
8453 {
|
|
|
8454 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
8455 vim_free(stp[i].st_word);
|
|
|
8456 ga.ga_len = su->su_maxcount;
|
|
|
8457 }
|
|
|
8458
|
|
|
8459 su->su_ga = ga;
|
|
|
8460 }
|
|
|
8461
|
|
|
8462 /*
|
|
351
|
8463 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
8464 * badword.
|
|
|
8465 */
|
|
|
8466 static int
|
|
|
8467 stp_sal_score(stp, su, slang, badsound)
|
|
|
8468 suggest_T *stp;
|
|
|
8469 suginfo_T *su;
|
|
|
8470 slang_T *slang;
|
|
|
8471 char_u *badsound; /* sound-folded badword */
|
|
|
8472 {
|
|
|
8473 char_u *p;
|
|
|
8474 char_u badsound2[MAXWLEN];
|
|
|
8475 char_u fword[MAXWLEN];
|
|
|
8476 char_u goodsound[MAXWLEN];
|
|
|
8477
|
|
|
8478 if (stp->st_orglen <= su->su_badlen)
|
|
|
8479 p = badsound;
|
|
|
8480 else
|
|
|
8481 {
|
|
|
8482 /* soundfold the bad word with more characters following */
|
|
|
8483 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
8484
|
|
|
8485 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
8486 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
8487 * removing the space. Don't do it when the good word also contains a
|
|
|
8488 * space. */
|
|
|
8489 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
8490 && *skiptowhite(stp->st_word) == NUL)
|
|
|
8491 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
8492 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
8493
|
|
375
|
8494 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
8495 p = badsound2;
|
|
|
8496 }
|
|
|
8497
|
|
375
|
8498 /* Sound-fold the word and compute the score for the difference. */
|
|
|
8499 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
8500
|
|
|
8501 return soundalike_score(goodsound, p);
|
|
|
8502 }
|
|
|
8503
|
|
|
8504 /*
|
|
323
|
8505 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
8506 */
|
|
|
8507 static void
|
|
346
|
8508 suggest_try_soundalike(su)
|
|
323
|
8509 suginfo_T *su;
|
|
|
8510 {
|
|
|
8511 char_u salword[MAXWLEN];
|
|
|
8512 char_u tword[MAXWLEN];
|
|
|
8513 char_u tsalword[MAXWLEN];
|
|
324
|
8514 idx_T arridx[MAXWLEN];
|
|
323
|
8515 int curi[MAXWLEN];
|
|
|
8516 langp_T *lp;
|
|
|
8517 char_u *byts;
|
|
324
|
8518 idx_T *idxs;
|
|
323
|
8519 int depth;
|
|
|
8520 int c;
|
|
324
|
8521 idx_T n;
|
|
323
|
8522 int round;
|
|
|
8523 int flags;
|
|
344
|
8524 int sound_score;
|
|
|
8525
|
|
|
8526 /* Do this for all languages that support sound folding. */
|
|
323
|
8527 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8528 lp->lp_slang != NULL; ++lp)
|
|
|
8529 {
|
|
|
8530 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8531 {
|
|
|
8532 /* soundfold the bad word */
|
|
375
|
8533 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
8534
|
|
|
8535 /*
|
|
|
8536 * Go through the whole tree, soundfold each word and compare.
|
|
|
8537 * round 1: use the case-folded tree.
|
|
|
8538 * round 2: use the keep-case tree.
|
|
|
8539 */
|
|
|
8540 for (round = 1; round <= 2; ++round)
|
|
|
8541 {
|
|
|
8542 if (round == 1)
|
|
|
8543 {
|
|
|
8544 byts = lp->lp_slang->sl_fbyts;
|
|
|
8545 idxs = lp->lp_slang->sl_fidxs;
|
|
|
8546 }
|
|
|
8547 else
|
|
|
8548 {
|
|
|
8549 byts = lp->lp_slang->sl_kbyts;
|
|
|
8550 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
8551 if (byts == NULL) /* no keep-case words */
|
|
|
8552 continue;
|
|
323
|
8553 }
|
|
|
8554
|
|
|
8555 depth = 0;
|
|
|
8556 arridx[0] = 0;
|
|
|
8557 curi[0] = 1;
|
|
|
8558 while (depth >= 0 && !got_int)
|
|
|
8559 {
|
|
|
8560 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
8561 {
|
|
323
|
8562 /* Done all bytes at this node, go up one level. */
|
|
|
8563 --depth;
|
|
351
|
8564 line_breakcheck();
|
|
|
8565 }
|
|
323
|
8566 else
|
|
|
8567 {
|
|
|
8568 /* Do one more byte at this node. */
|
|
|
8569 n = arridx[depth] + curi[depth];
|
|
|
8570 ++curi[depth];
|
|
|
8571 c = byts[n];
|
|
|
8572 if (c == 0)
|
|
|
8573 {
|
|
|
8574 /* End of word, deal with the word. */
|
|
324
|
8575 flags = (int)idxs[n];
|
|
323
|
8576 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
8577 {
|
|
|
8578 tword[depth] = NUL;
|
|
375
|
8579 /* Sound-fold. Only in keep-case tree need to
|
|
|
8580 * case-fold the word. */
|
|
|
8581 spell_soundfold(lp->lp_slang, tword,
|
|
|
8582 round == 1, tsalword);
|
|
323
|
8583
|
|
344
|
8584 /* Compute the edit distance between the
|
|
|
8585 * sound-a-like words. */
|
|
|
8586 sound_score = soundalike_score(salword,
|
|
|
8587 tsalword);
|
|
324
|
8588 if (sound_score < SCORE_MAXMAX)
|
|
323
|
8589 {
|
|
324
|
8590 char_u cword[MAXWLEN];
|
|
|
8591 char_u *p;
|
|
344
|
8592 int score;
|
|
324
|
8593
|
|
388
|
8594 flags |= su->su_badflags;
|
|
351
|
8595 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
8596 {
|
|
324
|
8597 /* Need to fix case according to
|
|
|
8598 * "flags". */
|
|
323
|
8599 make_case_word(tword, cword, flags);
|
|
324
|
8600 p = cword;
|
|
323
|
8601 }
|
|
|
8602 else
|
|
324
|
8603 p = tword;
|
|
|
8604
|
|
344
|
8605 if (sps_flags & SPS_DOUBLE)
|
|
|
8606 add_suggestion(su, &su->su_sga, p,
|
|
346
|
8607 su->su_badlen,
|
|
351
|
8608 sound_score, 0, FALSE);
|
|
344
|
8609 else
|
|
|
8610 {
|
|
|
8611 /* Compute the score. */
|
|
|
8612 score = spell_edit_score(
|
|
|
8613 su->su_badword, p);
|
|
|
8614 if (sps_flags & SPS_BEST)
|
|
|
8615 /* give a bonus for the good word
|
|
|
8616 * sounding the same as the bad
|
|
|
8617 * word */
|
|
|
8618 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8619 su->su_badlen,
|
|
344
|
8620 RESCORE(score, sound_score),
|
|
351
|
8621 sound_score, TRUE);
|
|
344
|
8622 else
|
|
|
8623 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8624 su->su_badlen,
|
|
351
|
8625 score + sound_score, 0, FALSE);
|
|
344
|
8626 }
|
|
323
|
8627 }
|
|
|
8628 }
|
|
|
8629
|
|
|
8630 /* Skip over other NUL bytes. */
|
|
|
8631 while (byts[n + 1] == 0)
|
|
|
8632 {
|
|
|
8633 ++n;
|
|
|
8634 ++curi[depth];
|
|
|
8635 }
|
|
|
8636 }
|
|
|
8637 else
|
|
|
8638 {
|
|
|
8639 /* Normal char, go one level deeper. */
|
|
|
8640 tword[depth++] = c;
|
|
|
8641 arridx[depth] = idxs[n];
|
|
|
8642 curi[depth] = 1;
|
|
|
8643 }
|
|
|
8644 }
|
|
|
8645 }
|
|
|
8646 }
|
|
|
8647 }
|
|
|
8648 }
|
|
|
8649 }
|
|
|
8650
|
|
|
8651 /*
|
|
324
|
8652 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
8653 */
|
|
|
8654 static void
|
|
|
8655 make_case_word(fword, cword, flags)
|
|
|
8656 char_u *fword;
|
|
|
8657 char_u *cword;
|
|
|
8658 int flags;
|
|
|
8659 {
|
|
|
8660 if (flags & WF_ALLCAP)
|
|
|
8661 /* Make it all upper-case */
|
|
|
8662 allcap_copy(fword, cword);
|
|
|
8663 else if (flags & WF_ONECAP)
|
|
|
8664 /* Make the first letter upper-case */
|
|
324
|
8665 onecap_copy(fword, cword, TRUE);
|
|
323
|
8666 else
|
|
|
8667 /* Use goodword as-is. */
|
|
|
8668 STRCPY(cword, fword);
|
|
|
8669 }
|
|
|
8670
|
|
330
|
8671 /*
|
|
|
8672 * Use map string "map" for languages "lp".
|
|
|
8673 */
|
|
|
8674 static void
|
|
|
8675 set_map_str(lp, map)
|
|
|
8676 slang_T *lp;
|
|
|
8677 char_u *map;
|
|
|
8678 {
|
|
|
8679 char_u *p;
|
|
|
8680 int headc = 0;
|
|
|
8681 int c;
|
|
|
8682 int i;
|
|
|
8683
|
|
|
8684 if (*map == NUL)
|
|
|
8685 {
|
|
|
8686 lp->sl_has_map = FALSE;
|
|
|
8687 return;
|
|
|
8688 }
|
|
|
8689 lp->sl_has_map = TRUE;
|
|
|
8690
|
|
|
8691 /* Init the array and hash table empty. */
|
|
|
8692 for (i = 0; i < 256; ++i)
|
|
|
8693 lp->sl_map_array[i] = 0;
|
|
|
8694 #ifdef FEAT_MBYTE
|
|
|
8695 hash_init(&lp->sl_map_hash);
|
|
|
8696 #endif
|
|
|
8697
|
|
|
8698 /*
|
|
|
8699 * The similar characters are stored separated with slashes:
|
|
|
8700 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
8701 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
8702 */
|
|
|
8703 for (p = map; *p != NUL; )
|
|
|
8704 {
|
|
|
8705 #ifdef FEAT_MBYTE
|
|
|
8706 c = mb_ptr2char_adv(&p);
|
|
|
8707 #else
|
|
|
8708 c = *p++;
|
|
|
8709 #endif
|
|
|
8710 if (c == '/')
|
|
|
8711 headc = 0;
|
|
|
8712 else
|
|
|
8713 {
|
|
|
8714 if (headc == 0)
|
|
|
8715 headc = c;
|
|
|
8716
|
|
|
8717 #ifdef FEAT_MBYTE
|
|
|
8718 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
8719 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
8720 * a NUL. */
|
|
|
8721 if (c >= 256)
|
|
|
8722 {
|
|
|
8723 int cl = mb_char2len(c);
|
|
|
8724 int headcl = mb_char2len(headc);
|
|
|
8725 char_u *b;
|
|
|
8726 hash_T hash;
|
|
|
8727 hashitem_T *hi;
|
|
|
8728
|
|
|
8729 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
8730 if (b == NULL)
|
|
|
8731 return;
|
|
|
8732 mb_char2bytes(c, b);
|
|
|
8733 b[cl] = NUL;
|
|
|
8734 mb_char2bytes(headc, b + cl + 1);
|
|
|
8735 b[cl + 1 + headcl] = NUL;
|
|
|
8736 hash = hash_hash(b);
|
|
|
8737 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
8738 if (HASHITEM_EMPTY(hi))
|
|
|
8739 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
8740 else
|
|
|
8741 {
|
|
|
8742 /* This should have been checked when generating the .spl
|
|
|
8743 * file. */
|
|
|
8744 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
8745 vim_free(b);
|
|
|
8746 }
|
|
|
8747 }
|
|
|
8748 else
|
|
|
8749 #endif
|
|
|
8750 lp->sl_map_array[c] = headc;
|
|
|
8751 }
|
|
|
8752 }
|
|
|
8753 }
|
|
|
8754
|
|
323
|
8755 /*
|
|
|
8756 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
8757 * lines in the .aff file.
|
|
|
8758 */
|
|
|
8759 static int
|
|
|
8760 similar_chars(slang, c1, c2)
|
|
|
8761 slang_T *slang;
|
|
|
8762 int c1;
|
|
|
8763 int c2;
|
|
|
8764 {
|
|
330
|
8765 int m1, m2;
|
|
|
8766 #ifdef FEAT_MBYTE
|
|
|
8767 char_u buf[MB_MAXBYTES];
|
|
|
8768 hashitem_T *hi;
|
|
|
8769
|
|
|
8770 if (c1 >= 256)
|
|
|
8771 {
|
|
|
8772 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
8773 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8774 if (HASHITEM_EMPTY(hi))
|
|
|
8775 m1 = 0;
|
|
|
8776 else
|
|
|
8777 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8778 }
|
|
|
8779 else
|
|
|
8780 #endif
|
|
|
8781 m1 = slang->sl_map_array[c1];
|
|
|
8782 if (m1 == 0)
|
|
323
|
8783 return FALSE;
|
|
330
|
8784
|
|
|
8785
|
|
|
8786 #ifdef FEAT_MBYTE
|
|
|
8787 if (c2 >= 256)
|
|
|
8788 {
|
|
|
8789 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
8790 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8791 if (HASHITEM_EMPTY(hi))
|
|
|
8792 m2 = 0;
|
|
|
8793 else
|
|
|
8794 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8795 }
|
|
|
8796 else
|
|
|
8797 #endif
|
|
|
8798 m2 = slang->sl_map_array[c2];
|
|
|
8799
|
|
|
8800 return m1 == m2;
|
|
323
|
8801 }
|
|
|
8802
|
|
|
8803 /*
|
|
|
8804 * Add a suggestion to the list of suggestions.
|
|
|
8805 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
8806 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
8807 * with spell_edit_score().
|
|
|
8808 */
|
|
|
8809 static void
|
|
351
|
8810 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
8811 suginfo_T *su;
|
|
344
|
8812 garray_T *gap;
|
|
323
|
8813 char_u *goodword;
|
|
346
|
8814 int badlen; /* length of bad word used */
|
|
324
|
8815 int score;
|
|
351
|
8816 int altscore;
|
|
344
|
8817 int had_bonus; /* value for st_had_bonus */
|
|
323
|
8818 {
|
|
|
8819 suggest_T *stp;
|
|
|
8820 int i;
|
|
346
|
8821 char_u *p = NULL;
|
|
|
8822 int c = 0;
|
|
323
|
8823
|
|
|
8824 /* Check that the word wasn't banned. */
|
|
|
8825 if (was_banned(su, goodword))
|
|
|
8826 return;
|
|
|
8827
|
|
346
|
8828 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
8829 * Remove the common part from "goodword". */
|
|
|
8830 i = badlen - su->su_badlen;
|
|
|
8831 if (i > 0)
|
|
|
8832 {
|
|
|
8833 /* This assumes there was no case folding or it didn't change the
|
|
|
8834 * length... */
|
|
|
8835 p = goodword + STRLEN(goodword) - i;
|
|
|
8836 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
8837 {
|
|
|
8838 badlen = su->su_badlen;
|
|
|
8839 c = *p;
|
|
|
8840 *p = NUL;
|
|
|
8841 }
|
|
|
8842 else
|
|
|
8843 p = NULL;
|
|
|
8844 }
|
|
372
|
8845 else if (i < 0)
|
|
|
8846 {
|
|
|
8847 /* When replacing part of the word check that we actually change
|
|
|
8848 * something. For "the the" a suggestion can be replacing the first
|
|
|
8849 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
8850 if (badlen == (int)STRLEN(goodword)
|
|
372
|
8851 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
8852 return;
|
|
|
8853 }
|
|
|
8854
|
|
346
|
8855
|
|
323
|
8856 if (score <= su->su_maxscore)
|
|
|
8857 {
|
|
366
|
8858 /* Check if the word is already there. Also check the length that is
|
|
|
8859 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
8860 * "thes" -> "these". */
|
|
344
|
8861 stp = &SUG(*gap, 0);
|
|
|
8862 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
8863 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
8864 && stp[i].st_orglen == badlen)
|
|
323
|
8865 {
|
|
|
8866 /* Found it. Remember the lowest score. */
|
|
|
8867 if (stp[i].st_score > score)
|
|
324
|
8868 {
|
|
323
|
8869 stp[i].st_score = score;
|
|
376
|
8870 stp[i].st_altscore = altscore;
|
|
324
|
8871 stp[i].st_had_bonus = had_bonus;
|
|
|
8872 }
|
|
323
|
8873 break;
|
|
|
8874 }
|
|
|
8875
|
|
344
|
8876 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
8877 {
|
|
|
8878 /* Add a suggestion. */
|
|
344
|
8879 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
8880 stp->st_word = vim_strsave(goodword);
|
|
|
8881 if (stp->st_word != NULL)
|
|
|
8882 {
|
|
|
8883 stp->st_score = score;
|
|
351
|
8884 stp->st_altscore = altscore;
|
|
324
|
8885 stp->st_had_bonus = had_bonus;
|
|
346
|
8886 stp->st_orglen = badlen;
|
|
344
|
8887 ++gap->ga_len;
|
|
323
|
8888
|
|
|
8889 /* If we have too many suggestions now, sort the list and keep
|
|
|
8890 * the best suggestions. */
|
|
344
|
8891 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
8892 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
8893 SUG_CLEAN_COUNT(su));
|
|
323
|
8894 }
|
|
|
8895 }
|
|
|
8896 }
|
|
346
|
8897
|
|
|
8898 if (p != NULL)
|
|
|
8899 *p = c; /* restore "goodword" */
|
|
323
|
8900 }
|
|
|
8901
|
|
|
8902 /*
|
|
|
8903 * Add a word to be banned.
|
|
|
8904 */
|
|
|
8905 static void
|
|
|
8906 add_banned(su, word)
|
|
|
8907 suginfo_T *su;
|
|
|
8908 char_u *word;
|
|
|
8909 {
|
|
|
8910 char_u *s = vim_strsave(word);
|
|
|
8911 hash_T hash;
|
|
|
8912 hashitem_T *hi;
|
|
|
8913
|
|
|
8914 if (s != NULL)
|
|
|
8915 {
|
|
|
8916 hash = hash_hash(s);
|
|
|
8917 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
8918 if (HASHITEM_EMPTY(hi))
|
|
|
8919 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
8920 else
|
|
|
8921 vim_free(s);
|
|
323
|
8922 }
|
|
|
8923 }
|
|
|
8924
|
|
|
8925 /*
|
|
|
8926 * Return TRUE if a word appears in the list of banned words.
|
|
|
8927 */
|
|
|
8928 static int
|
|
|
8929 was_banned(su, word)
|
|
|
8930 suginfo_T *su;
|
|
|
8931 char_u *word;
|
|
|
8932 {
|
|
324
|
8933 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
8934
|
|
|
8935 return !HASHITEM_EMPTY(hi);
|
|
323
|
8936 }
|
|
|
8937
|
|
|
8938 /*
|
|
|
8939 * Free the banned words in "su".
|
|
|
8940 */
|
|
|
8941 static void
|
|
|
8942 free_banned(su)
|
|
|
8943 suginfo_T *su;
|
|
|
8944 {
|
|
|
8945 int todo;
|
|
|
8946 hashitem_T *hi;
|
|
|
8947
|
|
|
8948 todo = su->su_banned.ht_used;
|
|
|
8949 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
8950 {
|
|
|
8951 if (!HASHITEM_EMPTY(hi))
|
|
|
8952 {
|
|
|
8953 vim_free(hi->hi_key);
|
|
|
8954 --todo;
|
|
|
8955 }
|
|
|
8956 }
|
|
|
8957 hash_clear(&su->su_banned);
|
|
|
8958 }
|
|
|
8959
|
|
324
|
8960 /*
|
|
|
8961 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
8962 * thus only done for the final results.
|
|
|
8963 */
|
|
|
8964 static void
|
|
|
8965 rescore_suggestions(su)
|
|
|
8966 suginfo_T *su;
|
|
|
8967 {
|
|
|
8968 langp_T *lp;
|
|
|
8969 suggest_T *stp;
|
|
|
8970 char_u sal_badword[MAXWLEN];
|
|
|
8971 int i;
|
|
|
8972
|
|
|
8973 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8974 lp->lp_slang != NULL; ++lp)
|
|
|
8975 {
|
|
|
8976 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8977 {
|
|
|
8978 /* soundfold the bad word */
|
|
375
|
8979 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
8980
|
|
|
8981 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8982 {
|
|
344
|
8983 stp = &SUG(su->su_ga, i);
|
|
324
|
8984 if (!stp->st_had_bonus)
|
|
|
8985 {
|
|
351
|
8986 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
8987 lp->lp_slang, sal_badword);
|
|
|
8988 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8989 stp->st_altscore = SCORE_BIG;
|
|
|
8990 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
8991 }
|
|
|
8992 }
|
|
|
8993 break;
|
|
|
8994 }
|
|
|
8995 }
|
|
|
8996 }
|
|
|
8997
|
|
323
|
8998 static int
|
|
|
8999 #ifdef __BORLANDC__
|
|
|
9000 _RTLENTRYF
|
|
|
9001 #endif
|
|
|
9002 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
9003
|
|
|
9004 /*
|
|
|
9005 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
9006 */
|
|
|
9007 static int
|
|
|
9008 #ifdef __BORLANDC__
|
|
|
9009 _RTLENTRYF
|
|
|
9010 #endif
|
|
|
9011 sug_compare(s1, s2)
|
|
|
9012 const void *s1;
|
|
|
9013 const void *s2;
|
|
|
9014 {
|
|
|
9015 suggest_T *p1 = (suggest_T *)s1;
|
|
|
9016 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
9017 int n = p1->st_score - p2->st_score;
|
|
|
9018
|
|
|
9019 if (n == 0)
|
|
|
9020 return p1->st_altscore - p2->st_altscore;
|
|
|
9021 return n;
|
|
323
|
9022 }
|
|
|
9023
|
|
|
9024 /*
|
|
|
9025 * Cleanup the suggestions:
|
|
|
9026 * - Sort on score.
|
|
|
9027 * - Remove words that won't be displayed.
|
|
344
|
9028 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
9029 */
|
|
344
|
9030 static int
|
|
|
9031 cleanup_suggestions(gap, maxscore, keep)
|
|
|
9032 garray_T *gap;
|
|
|
9033 int maxscore;
|
|
324
|
9034 int keep; /* nr of suggestions to keep */
|
|
323
|
9035 {
|
|
344
|
9036 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
9037 int i;
|
|
|
9038
|
|
|
9039 /* Sort the list. */
|
|
344
|
9040 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
9041
|
|
|
9042 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
9043 if (gap->ga_len > keep)
|
|
323
|
9044 {
|
|
344
|
9045 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
9046 vim_free(stp[i].st_word);
|
|
344
|
9047 gap->ga_len = keep;
|
|
|
9048 return stp[keep - 1].st_score;
|
|
323
|
9049 }
|
|
344
|
9050 return maxscore;
|
|
323
|
9051 }
|
|
|
9052
|
|
372
|
9053 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
9054 /*
|
|
|
9055 * Soundfold a string, for soundfold().
|
|
|
9056 * Result is in allocated memory, NULL for an error.
|
|
|
9057 */
|
|
|
9058 char_u *
|
|
|
9059 eval_soundfold(word)
|
|
|
9060 char_u *word;
|
|
|
9061 {
|
|
|
9062 langp_T *lp;
|
|
|
9063 char_u sound[MAXWLEN];
|
|
|
9064
|
|
|
9065 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
9066 /* Use the sound-folding of the first language that supports it. */
|
|
|
9067 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9068 lp->lp_slang != NULL; ++lp)
|
|
|
9069 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9070 {
|
|
|
9071 /* soundfold the word */
|
|
375
|
9072 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
9073 return vim_strsave(sound);
|
|
|
9074 }
|
|
|
9075
|
|
|
9076 /* No language with sound folding, return word as-is. */
|
|
|
9077 return vim_strsave(word);
|
|
|
9078 }
|
|
|
9079 #endif
|
|
|
9080
|
|
323
|
9081 /*
|
|
|
9082 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
9083 */
|
|
|
9084 static void
|
|
375
|
9085 spell_soundfold(slang, inword, folded, res)
|
|
|
9086 slang_T *slang;
|
|
|
9087 char_u *inword;
|
|
|
9088 int folded; /* "inword" is already case-folded */
|
|
|
9089 char_u *res;
|
|
|
9090 {
|
|
|
9091 char_u fword[MAXWLEN];
|
|
|
9092 char_u *word;
|
|
|
9093
|
|
|
9094 if (slang->sl_sofo)
|
|
|
9095 /* SOFOFROM and SOFOTO used */
|
|
|
9096 spell_soundfold_sofo(slang, inword, res);
|
|
|
9097 else
|
|
|
9098 {
|
|
|
9099 /* SAL items used. Requires the word to be case-folded. */
|
|
|
9100 if (folded)
|
|
|
9101 word = inword;
|
|
|
9102 else
|
|
|
9103 {
|
|
|
9104 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
9105 word = fword;
|
|
|
9106 }
|
|
|
9107
|
|
|
9108 #ifdef FEAT_MBYTE
|
|
|
9109 if (has_mbyte)
|
|
|
9110 spell_soundfold_wsal(slang, word, res);
|
|
|
9111 else
|
|
|
9112 #endif
|
|
|
9113 spell_soundfold_sal(slang, word, res);
|
|
|
9114 }
|
|
|
9115 }
|
|
|
9116
|
|
|
9117 /*
|
|
|
9118 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
9119 * SOFOTO lines.
|
|
|
9120 */
|
|
|
9121 static void
|
|
|
9122 spell_soundfold_sofo(slang, inword, res)
|
|
|
9123 slang_T *slang;
|
|
|
9124 char_u *inword;
|
|
|
9125 char_u *res;
|
|
|
9126 {
|
|
|
9127 char_u *s;
|
|
|
9128 int ri = 0;
|
|
|
9129 int c;
|
|
|
9130
|
|
|
9131 #ifdef FEAT_MBYTE
|
|
|
9132 if (has_mbyte)
|
|
|
9133 {
|
|
|
9134 int prevc = 0;
|
|
|
9135 int *ip;
|
|
|
9136
|
|
|
9137 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
9138 * 255, sl_sal the rest. */
|
|
|
9139 for (s = inword; *s != NUL; )
|
|
|
9140 {
|
|
|
9141 c = mb_ptr2char_adv(&s);
|
|
|
9142 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9143 c = ' ';
|
|
|
9144 else if (c < 256)
|
|
|
9145 c = slang->sl_sal_first[c];
|
|
|
9146 else
|
|
|
9147 {
|
|
|
9148 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
9149 if (ip == NULL) /* empty list, can't match */
|
|
|
9150 c = NUL;
|
|
|
9151 else
|
|
|
9152 for (;;) /* find "c" in the list */
|
|
|
9153 {
|
|
|
9154 if (*ip == 0) /* not found */
|
|
|
9155 {
|
|
|
9156 c = NUL;
|
|
|
9157 break;
|
|
|
9158 }
|
|
|
9159 if (*ip == c) /* match! */
|
|
|
9160 {
|
|
|
9161 c = ip[1];
|
|
|
9162 break;
|
|
|
9163 }
|
|
|
9164 ip += 2;
|
|
|
9165 }
|
|
|
9166 }
|
|
|
9167
|
|
|
9168 if (c != NUL && c != prevc)
|
|
|
9169 {
|
|
|
9170 ri += mb_char2bytes(c, res + ri);
|
|
|
9171 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
9172 break;
|
|
|
9173 prevc = c;
|
|
|
9174 }
|
|
|
9175 }
|
|
|
9176 }
|
|
|
9177 else
|
|
|
9178 #endif
|
|
|
9179 {
|
|
|
9180 /* The sl_sal_first[] table contains the translation. */
|
|
|
9181 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
9182 {
|
|
|
9183 if (vim_iswhite(c))
|
|
|
9184 c = ' ';
|
|
|
9185 else
|
|
|
9186 c = slang->sl_sal_first[c];
|
|
|
9187 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
9188 res[ri++] = c;
|
|
|
9189 }
|
|
|
9190 }
|
|
|
9191
|
|
|
9192 res[ri] = NUL;
|
|
|
9193 }
|
|
|
9194
|
|
|
9195 static void
|
|
|
9196 spell_soundfold_sal(slang, inword, res)
|
|
323
|
9197 slang_T *slang;
|
|
|
9198 char_u *inword;
|
|
|
9199 char_u *res;
|
|
|
9200 {
|
|
344
|
9201 salitem_T *smp;
|
|
323
|
9202 char_u word[MAXWLEN];
|
|
375
|
9203 char_u *s = inword;
|
|
323
|
9204 char_u *t;
|
|
344
|
9205 char_u *pf;
|
|
323
|
9206 int i, j, z;
|
|
344
|
9207 int reslen;
|
|
323
|
9208 int n, k = 0;
|
|
|
9209 int z0;
|
|
|
9210 int k0;
|
|
|
9211 int n0;
|
|
|
9212 int c;
|
|
|
9213 int pri;
|
|
|
9214 int p0 = -333;
|
|
|
9215 int c0;
|
|
|
9216
|
|
324
|
9217 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
9218 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
9219 if (slang->sl_rem_accents)
|
|
|
9220 {
|
|
|
9221 t = word;
|
|
375
|
9222 while (*s != NUL)
|
|
323
|
9223 {
|
|
324
|
9224 if (vim_iswhite(*s))
|
|
344
|
9225 {
|
|
|
9226 *t++ = ' ';
|
|
|
9227 s = skipwhite(s);
|
|
|
9228 }
|
|
323
|
9229 else
|
|
|
9230 {
|
|
376
|
9231 if (spell_iswordp_nmw(s))
|
|
323
|
9232 *t++ = *s;
|
|
|
9233 ++s;
|
|
|
9234 }
|
|
|
9235 }
|
|
|
9236 *t = NUL;
|
|
|
9237 }
|
|
|
9238 else
|
|
375
|
9239 STRCPY(word, s);
|
|
323
|
9240
|
|
344
|
9241 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
9242
|
|
|
9243 /*
|
|
|
9244 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
9245 * Changed to keep spaces.
|
|
323
|
9246 */
|
|
344
|
9247 i = reslen = z = 0;
|
|
323
|
9248 while ((c = word[i]) != NUL)
|
|
|
9249 {
|
|
344
|
9250 /* Start with the first rule that has the character in the word. */
|
|
323
|
9251 n = slang->sl_sal_first[c];
|
|
|
9252 z0 = 0;
|
|
|
9253
|
|
|
9254 if (n >= 0)
|
|
|
9255 {
|
|
|
9256 /* check all rules for the same letter */
|
|
344
|
9257 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
9258 {
|
|
344
|
9259 /* Quickly skip entries that don't match the word. Most
|
|
|
9260 * entries are less then three chars, optimize for that. */
|
|
|
9261 k = smp[n].sm_leadlen;
|
|
|
9262 if (k > 1)
|
|
323
|
9263 {
|
|
344
|
9264 if (word[i + 1] != s[1])
|
|
|
9265 continue;
|
|
|
9266 if (k > 2)
|
|
|
9267 {
|
|
|
9268 for (j = 2; j < k; ++j)
|
|
|
9269 if (word[i + j] != s[j])
|
|
|
9270 break;
|
|
|
9271 if (j < k)
|
|
|
9272 continue;
|
|
|
9273 }
|
|
323
|
9274 }
|
|
|
9275
|
|
375
|
9276 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
9277 {
|
|
375
|
9278 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
9279 while (*pf != NUL && *pf != word[i + k])
|
|
|
9280 ++pf;
|
|
|
9281 if (*pf == NUL)
|
|
|
9282 continue;
|
|
|
9283 ++k;
|
|
323
|
9284 }
|
|
344
|
9285 s = smp[n].sm_rules;
|
|
|
9286 pri = 5; /* default priority */
|
|
323
|
9287
|
|
|
9288 p0 = *s;
|
|
|
9289 k0 = k;
|
|
|
9290 while (*s == '-' && k > 1)
|
|
|
9291 {
|
|
|
9292 k--;
|
|
|
9293 s++;
|
|
|
9294 }
|
|
|
9295 if (*s == '<')
|
|
|
9296 s++;
|
|
344
|
9297 if (VIM_ISDIGIT(*s))
|
|
323
|
9298 {
|
|
|
9299 /* determine priority */
|
|
|
9300 pri = *s - '0';
|
|
|
9301 s++;
|
|
|
9302 }
|
|
|
9303 if (*s == '^' && *(s + 1) == '^')
|
|
|
9304 s++;
|
|
|
9305
|
|
|
9306 if (*s == NUL
|
|
|
9307 || (*s == '^'
|
|
324
|
9308 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9309 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
9310 && (*(s + 1) != '$'
|
|
376
|
9311 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9312 || (*s == '$' && i > 0
|
|
376
|
9313 && spell_iswordp(word + i - 1, curbuf)
|
|
|
9314 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9315 {
|
|
|
9316 /* search for followup rules, if: */
|
|
|
9317 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9318 c0 = word[i + k - 1];
|
|
|
9319 n0 = slang->sl_sal_first[c0];
|
|
|
9320
|
|
|
9321 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
9322 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
9323 {
|
|
|
9324 /* test follow-up rule for "word[i + k]" */
|
|
344
|
9325 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
9326 {
|
|
344
|
9327 /* Quickly skip entries that don't match the word.
|
|
|
9328 * */
|
|
|
9329 k0 = smp[n0].sm_leadlen;
|
|
|
9330 if (k0 > 1)
|
|
323
|
9331 {
|
|
344
|
9332 if (word[i + k] != s[1])
|
|
|
9333 continue;
|
|
|
9334 if (k0 > 2)
|
|
|
9335 {
|
|
|
9336 pf = word + i + k + 1;
|
|
|
9337 for (j = 2; j < k0; ++j)
|
|
|
9338 if (*pf++ != s[j])
|
|
|
9339 break;
|
|
|
9340 if (j < k0)
|
|
|
9341 continue;
|
|
|
9342 }
|
|
323
|
9343 }
|
|
344
|
9344 k0 += k - 1;
|
|
|
9345
|
|
375
|
9346 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
9347 {
|
|
344
|
9348 /* Check for match with one of the chars in
|
|
375
|
9349 * "sm_oneof". */
|
|
344
|
9350 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9351 ++pf;
|
|
|
9352 if (*pf == NUL)
|
|
|
9353 continue;
|
|
|
9354 ++k0;
|
|
323
|
9355 }
|
|
344
|
9356
|
|
|
9357 p0 = 5;
|
|
|
9358 s = smp[n0].sm_rules;
|
|
323
|
9359 while (*s == '-')
|
|
|
9360 {
|
|
344
|
9361 /* "k0" gets NOT reduced because
|
|
|
9362 * "if (k0 == k)" */
|
|
323
|
9363 s++;
|
|
|
9364 }
|
|
|
9365 if (*s == '<')
|
|
|
9366 s++;
|
|
344
|
9367 if (VIM_ISDIGIT(*s))
|
|
323
|
9368 {
|
|
|
9369 p0 = *s - '0';
|
|
|
9370 s++;
|
|
|
9371 }
|
|
|
9372
|
|
|
9373 if (*s == NUL
|
|
|
9374 /* *s == '^' cuts */
|
|
|
9375 || (*s == '$'
|
|
376
|
9376 && !spell_iswordp(word + i + k0,
|
|
|
9377 curbuf)))
|
|
323
|
9378 {
|
|
|
9379 if (k0 == k)
|
|
|
9380 /* this is just a piece of the string */
|
|
|
9381 continue;
|
|
|
9382
|
|
|
9383 if (p0 < pri)
|
|
|
9384 /* priority too low */
|
|
|
9385 continue;
|
|
|
9386 /* rule fits; stop search */
|
|
|
9387 break;
|
|
|
9388 }
|
|
|
9389 }
|
|
|
9390
|
|
344
|
9391 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
9392 continue;
|
|
|
9393 }
|
|
|
9394
|
|
|
9395 /* replace string */
|
|
344
|
9396 s = smp[n].sm_to;
|
|
389
|
9397 if (s == NULL)
|
|
|
9398 s = (char_u *)"";
|
|
344
|
9399 pf = smp[n].sm_rules;
|
|
|
9400 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
9401 if (p0 == 1 && z == 0)
|
|
|
9402 {
|
|
|
9403 /* rule with '<' is used */
|
|
344
|
9404 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
9405 || res[reslen - 1] == *s))
|
|
|
9406 reslen--;
|
|
323
|
9407 z0 = 1;
|
|
|
9408 z = 1;
|
|
|
9409 k0 = 0;
|
|
372
|
9410 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
9411 {
|
|
|
9412 word[i + k0] = *s;
|
|
|
9413 k0++;
|
|
|
9414 s++;
|
|
|
9415 }
|
|
|
9416 if (k > k0)
|
|
|
9417 mch_memmove(word + i + k0, word + i + k,
|
|
|
9418 STRLEN(word + i + k) + 1);
|
|
|
9419
|
|
|
9420 /* new "actual letter" */
|
|
|
9421 c = word[i];
|
|
|
9422 }
|
|
|
9423 else
|
|
|
9424 {
|
|
|
9425 /* no '<' rule used */
|
|
|
9426 i += k - 1;
|
|
|
9427 z = 0;
|
|
344
|
9428 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
9429 {
|
|
344
|
9430 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
9431 res[reslen++] = *s;
|
|
323
|
9432 s++;
|
|
|
9433 }
|
|
|
9434 /* new "actual letter" */
|
|
|
9435 c = *s;
|
|
344
|
9436 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
9437 {
|
|
|
9438 if (c != NUL)
|
|
372
|
9439 res[reslen++] = c;
|
|
323
|
9440 mch_memmove(word, word + i + 1,
|
|
|
9441 STRLEN(word + i + 1) + 1);
|
|
|
9442 i = 0;
|
|
|
9443 z0 = 1;
|
|
|
9444 }
|
|
|
9445 }
|
|
|
9446 break;
|
|
|
9447 }
|
|
|
9448 }
|
|
|
9449 }
|
|
324
|
9450 else if (vim_iswhite(c))
|
|
|
9451 {
|
|
|
9452 c = ' ';
|
|
|
9453 k = 1;
|
|
|
9454 }
|
|
323
|
9455
|
|
|
9456 if (z0 == 0)
|
|
|
9457 {
|
|
344
|
9458 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9459 && (!slang->sl_collapse || reslen == 0
|
|
|
9460 || res[reslen - 1] != c))
|
|
323
|
9461 /* condense only double letters */
|
|
372
|
9462 res[reslen++] = c;
|
|
323
|
9463
|
|
|
9464 i++;
|
|
|
9465 z = 0;
|
|
|
9466 k = 0;
|
|
|
9467 }
|
|
|
9468 }
|
|
|
9469
|
|
344
|
9470 res[reslen] = NUL;
|
|
323
|
9471 }
|
|
|
9472
|
|
372
|
9473 #ifdef FEAT_MBYTE
|
|
|
9474 /*
|
|
|
9475 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
9476 * Multi-byte version of spell_soundfold().
|
|
|
9477 */
|
|
|
9478 static void
|
|
375
|
9479 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
9480 slang_T *slang;
|
|
|
9481 char_u *inword;
|
|
|
9482 char_u *res;
|
|
|
9483 {
|
|
375
|
9484 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
9485 int word[MAXWLEN];
|
|
|
9486 int wres[MAXWLEN];
|
|
|
9487 int l;
|
|
|
9488 char_u *s;
|
|
|
9489 int *ws;
|
|
|
9490 char_u *t;
|
|
|
9491 int *pf;
|
|
|
9492 int i, j, z;
|
|
|
9493 int reslen;
|
|
|
9494 int n, k = 0;
|
|
|
9495 int z0;
|
|
|
9496 int k0;
|
|
|
9497 int n0;
|
|
|
9498 int c;
|
|
|
9499 int pri;
|
|
|
9500 int p0 = -333;
|
|
|
9501 int c0;
|
|
|
9502 int did_white = FALSE;
|
|
|
9503
|
|
|
9504 /*
|
|
|
9505 * Convert the multi-byte string to a wide-character string.
|
|
|
9506 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
9507 * But keep white space.
|
|
|
9508 */
|
|
|
9509 n = 0;
|
|
|
9510 for (s = inword; *s != NUL; )
|
|
|
9511 {
|
|
|
9512 t = s;
|
|
|
9513 c = mb_ptr2char_adv(&s);
|
|
|
9514 if (slang->sl_rem_accents)
|
|
|
9515 {
|
|
|
9516 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9517 {
|
|
|
9518 if (did_white)
|
|
|
9519 continue;
|
|
|
9520 c = ' ';
|
|
|
9521 did_white = TRUE;
|
|
|
9522 }
|
|
|
9523 else
|
|
|
9524 {
|
|
|
9525 did_white = FALSE;
|
|
376
|
9526 if (!spell_iswordp_nmw(t))
|
|
372
|
9527 continue;
|
|
|
9528 }
|
|
|
9529 }
|
|
|
9530 word[n++] = c;
|
|
|
9531 }
|
|
|
9532 word[n] = NUL;
|
|
|
9533
|
|
|
9534 /*
|
|
|
9535 * This comes from Aspell phonet.cpp.
|
|
|
9536 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
9537 * Changed to keep spaces.
|
|
|
9538 */
|
|
|
9539 i = reslen = z = 0;
|
|
|
9540 while ((c = word[i]) != NUL)
|
|
|
9541 {
|
|
|
9542 /* Start with the first rule that has the character in the word. */
|
|
|
9543 n = slang->sl_sal_first[c & 0xff];
|
|
|
9544 z0 = 0;
|
|
|
9545
|
|
|
9546 if (n >= 0)
|
|
|
9547 {
|
|
375
|
9548 /* check all rules for the same index byte */
|
|
372
|
9549 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
9550 {
|
|
|
9551 /* Quickly skip entries that don't match the word. Most
|
|
|
9552 * entries are less then three chars, optimize for that. */
|
|
375
|
9553 if (c != ws[0])
|
|
|
9554 continue;
|
|
372
|
9555 k = smp[n].sm_leadlen;
|
|
|
9556 if (k > 1)
|
|
|
9557 {
|
|
|
9558 if (word[i + 1] != ws[1])
|
|
|
9559 continue;
|
|
|
9560 if (k > 2)
|
|
|
9561 {
|
|
|
9562 for (j = 2; j < k; ++j)
|
|
|
9563 if (word[i + j] != ws[j])
|
|
|
9564 break;
|
|
|
9565 if (j < k)
|
|
|
9566 continue;
|
|
|
9567 }
|
|
|
9568 }
|
|
|
9569
|
|
375
|
9570 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
9571 {
|
|
375
|
9572 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
9573 while (*pf != NUL && *pf != word[i + k])
|
|
|
9574 ++pf;
|
|
|
9575 if (*pf == NUL)
|
|
|
9576 continue;
|
|
|
9577 ++k;
|
|
|
9578 }
|
|
|
9579 s = smp[n].sm_rules;
|
|
|
9580 pri = 5; /* default priority */
|
|
|
9581
|
|
|
9582 p0 = *s;
|
|
|
9583 k0 = k;
|
|
|
9584 while (*s == '-' && k > 1)
|
|
|
9585 {
|
|
|
9586 k--;
|
|
|
9587 s++;
|
|
|
9588 }
|
|
|
9589 if (*s == '<')
|
|
|
9590 s++;
|
|
|
9591 if (VIM_ISDIGIT(*s))
|
|
|
9592 {
|
|
|
9593 /* determine priority */
|
|
|
9594 pri = *s - '0';
|
|
|
9595 s++;
|
|
|
9596 }
|
|
|
9597 if (*s == '^' && *(s + 1) == '^')
|
|
|
9598 s++;
|
|
|
9599
|
|
|
9600 if (*s == NUL
|
|
|
9601 || (*s == '^'
|
|
|
9602 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9603 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
9604 && (*(s + 1) != '$'
|
|
376
|
9605 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9606 || (*s == '$' && i > 0
|
|
376
|
9607 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
9608 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9609 {
|
|
|
9610 /* search for followup rules, if: */
|
|
|
9611 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9612 c0 = word[i + k - 1];
|
|
|
9613 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
9614
|
|
|
9615 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
9616 && p0 != '-' && word[i + k] != NUL)
|
|
|
9617 {
|
|
375
|
9618 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
9619 * all entries with the same index byte. */
|
|
372
|
9620 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
9621 == (c0 & 0xff); ++n0)
|
|
|
9622 {
|
|
|
9623 /* Quickly skip entries that don't match the word.
|
|
375
|
9624 */
|
|
|
9625 if (c0 != ws[0])
|
|
|
9626 continue;
|
|
372
|
9627 k0 = smp[n0].sm_leadlen;
|
|
|
9628 if (k0 > 1)
|
|
|
9629 {
|
|
|
9630 if (word[i + k] != ws[1])
|
|
|
9631 continue;
|
|
|
9632 if (k0 > 2)
|
|
|
9633 {
|
|
|
9634 pf = word + i + k + 1;
|
|
|
9635 for (j = 2; j < k0; ++j)
|
|
|
9636 if (*pf++ != ws[j])
|
|
|
9637 break;
|
|
|
9638 if (j < k0)
|
|
|
9639 continue;
|
|
|
9640 }
|
|
|
9641 }
|
|
|
9642 k0 += k - 1;
|
|
|
9643
|
|
375
|
9644 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
9645 {
|
|
|
9646 /* Check for match with one of the chars in
|
|
375
|
9647 * "sm_oneof". */
|
|
372
|
9648 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9649 ++pf;
|
|
|
9650 if (*pf == NUL)
|
|
|
9651 continue;
|
|
|
9652 ++k0;
|
|
|
9653 }
|
|
|
9654
|
|
|
9655 p0 = 5;
|
|
|
9656 s = smp[n0].sm_rules;
|
|
|
9657 while (*s == '-')
|
|
|
9658 {
|
|
|
9659 /* "k0" gets NOT reduced because
|
|
|
9660 * "if (k0 == k)" */
|
|
|
9661 s++;
|
|
|
9662 }
|
|
|
9663 if (*s == '<')
|
|
|
9664 s++;
|
|
|
9665 if (VIM_ISDIGIT(*s))
|
|
|
9666 {
|
|
|
9667 p0 = *s - '0';
|
|
|
9668 s++;
|
|
|
9669 }
|
|
|
9670
|
|
|
9671 if (*s == NUL
|
|
|
9672 /* *s == '^' cuts */
|
|
|
9673 || (*s == '$'
|
|
376
|
9674 && !spell_iswordp_w(word + i + k0,
|
|
|
9675 curbuf)))
|
|
372
|
9676 {
|
|
|
9677 if (k0 == k)
|
|
|
9678 /* this is just a piece of the string */
|
|
|
9679 continue;
|
|
|
9680
|
|
|
9681 if (p0 < pri)
|
|
|
9682 /* priority too low */
|
|
|
9683 continue;
|
|
|
9684 /* rule fits; stop search */
|
|
|
9685 break;
|
|
|
9686 }
|
|
|
9687 }
|
|
|
9688
|
|
|
9689 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
9690 == (c0 & 0xff))
|
|
|
9691 continue;
|
|
|
9692 }
|
|
|
9693
|
|
|
9694 /* replace string */
|
|
|
9695 ws = smp[n].sm_to_w;
|
|
|
9696 s = smp[n].sm_rules;
|
|
|
9697 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
9698 if (p0 == 1 && z == 0)
|
|
|
9699 {
|
|
|
9700 /* rule with '<' is used */
|
|
389
|
9701 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
9702 && (wres[reslen - 1] == c
|
|
372
|
9703 || wres[reslen - 1] == *ws))
|
|
|
9704 reslen--;
|
|
|
9705 z0 = 1;
|
|
|
9706 z = 1;
|
|
|
9707 k0 = 0;
|
|
389
|
9708 if (ws != NULL)
|
|
|
9709 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
9710 {
|
|
|
9711 word[i + k0] = *ws;
|
|
|
9712 k0++;
|
|
|
9713 ws++;
|
|
|
9714 }
|
|
372
|
9715 if (k > k0)
|
|
|
9716 mch_memmove(word + i + k0, word + i + k,
|
|
|
9717 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
9718
|
|
|
9719 /* new "actual letter" */
|
|
|
9720 c = word[i];
|
|
|
9721 }
|
|
|
9722 else
|
|
|
9723 {
|
|
|
9724 /* no '<' rule used */
|
|
|
9725 i += k - 1;
|
|
|
9726 z = 0;
|
|
389
|
9727 if (ws != NULL)
|
|
|
9728 while (*ws != NUL && ws[1] != NUL
|
|
|
9729 && reslen < MAXWLEN)
|
|
|
9730 {
|
|
|
9731 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
9732 wres[reslen++] = *ws;
|
|
|
9733 ws++;
|
|
|
9734 }
|
|
372
|
9735 /* new "actual letter" */
|
|
389
|
9736 if (ws == NULL)
|
|
|
9737 c = NUL;
|
|
|
9738 else
|
|
|
9739 c = *ws;
|
|
372
|
9740 if (strstr((char *)s, "^^") != NULL)
|
|
|
9741 {
|
|
|
9742 if (c != NUL)
|
|
|
9743 wres[reslen++] = c;
|
|
|
9744 mch_memmove(word, word + i + 1,
|
|
|
9745 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
9746 i = 0;
|
|
|
9747 z0 = 1;
|
|
|
9748 }
|
|
|
9749 }
|
|
|
9750 break;
|
|
|
9751 }
|
|
|
9752 }
|
|
|
9753 }
|
|
|
9754 else if (vim_iswhite(c))
|
|
|
9755 {
|
|
|
9756 c = ' ';
|
|
|
9757 k = 1;
|
|
|
9758 }
|
|
|
9759
|
|
|
9760 if (z0 == 0)
|
|
|
9761 {
|
|
|
9762 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9763 && (!slang->sl_collapse || reslen == 0
|
|
|
9764 || wres[reslen - 1] != c))
|
|
|
9765 /* condense only double letters */
|
|
|
9766 wres[reslen++] = c;
|
|
|
9767
|
|
|
9768 i++;
|
|
|
9769 z = 0;
|
|
|
9770 k = 0;
|
|
|
9771 }
|
|
|
9772 }
|
|
|
9773
|
|
|
9774 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
9775 l = 0;
|
|
|
9776 for (n = 0; n < reslen; ++n)
|
|
|
9777 {
|
|
|
9778 l += mb_char2bytes(wres[n], res + l);
|
|
|
9779 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
9780 break;
|
|
|
9781 }
|
|
|
9782 res[l] = NUL;
|
|
|
9783 }
|
|
|
9784 #endif
|
|
|
9785
|
|
324
|
9786 /*
|
|
344
|
9787 * Compute a score for two sound-a-like words.
|
|
|
9788 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
9789 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
9790 * avoiding checks that will not be possible.
|
|
|
9791 */
|
|
|
9792 static int
|
|
351
|
9793 soundalike_score(goodstart, badstart)
|
|
|
9794 char_u *goodstart; /* sound-folded good word */
|
|
|
9795 char_u *badstart; /* sound-folded bad word */
|
|
344
|
9796 {
|
|
351
|
9797 char_u *goodsound = goodstart;
|
|
|
9798 char_u *badsound = badstart;
|
|
|
9799 int goodlen;
|
|
|
9800 int badlen;
|
|
344
|
9801 int n;
|
|
|
9802 char_u *pl, *ps;
|
|
|
9803 char_u *pl2, *ps2;
|
|
351
|
9804 int score = 0;
|
|
|
9805
|
|
|
9806 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
9807 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
9808 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
9809 {
|
|
|
9810 score = SCORE_DEL / 2;
|
|
|
9811 if (*badsound == '*')
|
|
|
9812 ++badsound;
|
|
|
9813 else
|
|
|
9814 ++goodsound;
|
|
|
9815 }
|
|
|
9816
|
|
|
9817 goodlen = STRLEN(goodsound);
|
|
|
9818 badlen = STRLEN(badsound);
|
|
344
|
9819
|
|
|
9820 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
9821 * changes. */
|
|
|
9822 n = goodlen - badlen;
|
|
|
9823 if (n < -2 || n > 2)
|
|
|
9824 return SCORE_MAXMAX;
|
|
|
9825
|
|
|
9826 if (n > 0)
|
|
|
9827 {
|
|
351
|
9828 pl = goodsound; /* goodsound is longest */
|
|
344
|
9829 ps = badsound;
|
|
|
9830 }
|
|
|
9831 else
|
|
|
9832 {
|
|
351
|
9833 pl = badsound; /* badsound is longest */
|
|
344
|
9834 ps = goodsound;
|
|
|
9835 }
|
|
|
9836
|
|
|
9837 /* Skip over the identical part. */
|
|
|
9838 while (*pl == *ps && *pl != NUL)
|
|
|
9839 {
|
|
|
9840 ++pl;
|
|
|
9841 ++ps;
|
|
|
9842 }
|
|
|
9843
|
|
|
9844 switch (n)
|
|
|
9845 {
|
|
|
9846 case -2:
|
|
|
9847 case 2:
|
|
|
9848 /*
|
|
|
9849 * Must delete two characters from "pl".
|
|
|
9850 */
|
|
|
9851 ++pl; /* first delete */
|
|
|
9852 while (*pl == *ps)
|
|
|
9853 {
|
|
|
9854 ++pl;
|
|
|
9855 ++ps;
|
|
|
9856 }
|
|
|
9857 /* strings must be equal after second delete */
|
|
|
9858 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
9859 return score + SCORE_DEL * 2;
|
|
344
|
9860
|
|
|
9861 /* Failed to compare. */
|
|
|
9862 break;
|
|
|
9863
|
|
|
9864 case -1:
|
|
|
9865 case 1:
|
|
|
9866 /*
|
|
|
9867 * Minimal one delete from "pl" required.
|
|
|
9868 */
|
|
|
9869
|
|
|
9870 /* 1: delete */
|
|
|
9871 pl2 = pl + 1;
|
|
|
9872 ps2 = ps;
|
|
|
9873 while (*pl2 == *ps2)
|
|
|
9874 {
|
|
|
9875 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9876 return score + SCORE_DEL;
|
|
344
|
9877 ++pl2;
|
|
|
9878 ++ps2;
|
|
|
9879 }
|
|
|
9880
|
|
|
9881 /* 2: delete then swap, then rest must be equal */
|
|
|
9882 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9883 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9884 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
9885
|
|
|
9886 /* 3: delete then substitute, then the rest must be equal */
|
|
|
9887 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9888 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
9889
|
|
|
9890 /* 4: first swap then delete */
|
|
|
9891 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9892 {
|
|
|
9893 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9894 ps2 = ps + 2;
|
|
|
9895 while (*pl2 == *ps2)
|
|
|
9896 {
|
|
|
9897 ++pl2;
|
|
|
9898 ++ps2;
|
|
|
9899 }
|
|
|
9900 /* delete a char and then strings must be equal */
|
|
|
9901 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9902 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
9903 }
|
|
|
9904
|
|
|
9905 /* 5: first substitute then delete */
|
|
|
9906 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
9907 ps2 = ps + 1;
|
|
|
9908 while (*pl2 == *ps2)
|
|
|
9909 {
|
|
|
9910 ++pl2;
|
|
|
9911 ++ps2;
|
|
|
9912 }
|
|
|
9913 /* delete a char and then strings must be equal */
|
|
|
9914 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9915 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
9916
|
|
|
9917 /* Failed to compare. */
|
|
|
9918 break;
|
|
|
9919
|
|
|
9920 case 0:
|
|
|
9921 /*
|
|
|
9922 * Lenghts are equal, thus changes must result in same length: An
|
|
|
9923 * insert is only possible in combination with a delete.
|
|
|
9924 * 1: check if for identical strings
|
|
|
9925 */
|
|
|
9926 if (*pl == NUL)
|
|
351
|
9927 return score;
|
|
344
|
9928
|
|
|
9929 /* 2: swap */
|
|
|
9930 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9931 {
|
|
|
9932 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9933 ps2 = ps + 2;
|
|
|
9934 while (*pl2 == *ps2)
|
|
|
9935 {
|
|
|
9936 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9937 return score + SCORE_SWAP;
|
|
344
|
9938 ++pl2;
|
|
|
9939 ++ps2;
|
|
|
9940 }
|
|
|
9941 /* 3: swap and swap again */
|
|
|
9942 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9943 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9944 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
9945
|
|
|
9946 /* 4: swap and substitute */
|
|
|
9947 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9948 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
9949 }
|
|
|
9950
|
|
|
9951 /* 5: substitute */
|
|
|
9952 pl2 = pl + 1;
|
|
|
9953 ps2 = ps + 1;
|
|
|
9954 while (*pl2 == *ps2)
|
|
|
9955 {
|
|
|
9956 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9957 return score + SCORE_SUBST;
|
|
344
|
9958 ++pl2;
|
|
|
9959 ++ps2;
|
|
|
9960 }
|
|
|
9961
|
|
|
9962 /* 6: substitute and swap */
|
|
|
9963 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9964 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9965 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
9966
|
|
|
9967 /* 7: substitute and substitute */
|
|
|
9968 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9969 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
9970
|
|
|
9971 /* 8: insert then delete */
|
|
|
9972 pl2 = pl;
|
|
|
9973 ps2 = ps + 1;
|
|
|
9974 while (*pl2 == *ps2)
|
|
|
9975 {
|
|
|
9976 ++pl2;
|
|
|
9977 ++ps2;
|
|
|
9978 }
|
|
|
9979 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9980 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9981
|
|
|
9982 /* 9: delete then insert */
|
|
|
9983 pl2 = pl + 1;
|
|
|
9984 ps2 = ps;
|
|
|
9985 while (*pl2 == *ps2)
|
|
|
9986 {
|
|
|
9987 ++pl2;
|
|
|
9988 ++ps2;
|
|
|
9989 }
|
|
|
9990 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
9991 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9992
|
|
|
9993 /* Failed to compare. */
|
|
|
9994 break;
|
|
|
9995 }
|
|
|
9996
|
|
|
9997 return SCORE_MAXMAX;
|
|
|
9998 }
|
|
324
|
9999
|
|
323
|
10000 /*
|
|
|
10001 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
10002 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
10003 *
|
|
323
|
10004 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
10005 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
10006 * characters.
|
|
323
|
10007 */
|
|
|
10008 static int
|
|
|
10009 spell_edit_score(badword, goodword)
|
|
|
10010 char_u *badword;
|
|
|
10011 char_u *goodword;
|
|
|
10012 {
|
|
|
10013 int *cnt;
|
|
372
|
10014 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
10015 int j, i;
|
|
|
10016 int t;
|
|
|
10017 int bc, gc;
|
|
324
|
10018 int pbc, pgc;
|
|
|
10019 #ifdef FEAT_MBYTE
|
|
|
10020 char_u *p;
|
|
|
10021 int wbadword[MAXWLEN];
|
|
|
10022 int wgoodword[MAXWLEN];
|
|
|
10023
|
|
|
10024 if (has_mbyte)
|
|
|
10025 {
|
|
|
10026 /* Get the characters from the multi-byte strings and put them in an
|
|
|
10027 * int array for easy access. */
|
|
|
10028 for (p = badword, badlen = 0; *p != NUL; )
|
|
|
10029 wbadword[badlen++] = mb_ptr2char_adv(&p);
|
|
398
|
10030 wbadword[badlen++] = 0;
|
|
324
|
10031 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
|
10032 wgoodword[goodlen++] = mb_ptr2char_adv(&p);
|
|
398
|
10033 wgoodword[goodlen++] = 0;
|
|
324
|
10034 }
|
|
|
10035 else
|
|
|
10036 #endif
|
|
|
10037 {
|
|
|
10038 badlen = STRLEN(badword) + 1;
|
|
|
10039 goodlen = STRLEN(goodword) + 1;
|
|
|
10040 }
|
|
323
|
10041
|
|
|
10042 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
10043 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
10044 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
10045 TRUE);
|
|
324
|
10046 if (cnt == NULL)
|
|
|
10047 return 0; /* out of memory */
|
|
323
|
10048
|
|
|
10049 CNT(0, 0) = 0;
|
|
|
10050 for (j = 1; j <= goodlen; ++j)
|
|
|
10051 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
10052
|
|
|
10053 for (i = 1; i <= badlen; ++i)
|
|
|
10054 {
|
|
|
10055 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
10056 for (j = 1; j <= goodlen; ++j)
|
|
|
10057 {
|
|
324
|
10058 #ifdef FEAT_MBYTE
|
|
|
10059 if (has_mbyte)
|
|
|
10060 {
|
|
|
10061 bc = wbadword[i - 1];
|
|
|
10062 gc = wgoodword[j - 1];
|
|
|
10063 }
|
|
|
10064 else
|
|
|
10065 #endif
|
|
|
10066 {
|
|
|
10067 bc = badword[i - 1];
|
|
|
10068 gc = goodword[j - 1];
|
|
|
10069 }
|
|
323
|
10070 if (bc == gc)
|
|
|
10071 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
10072 else
|
|
|
10073 {
|
|
|
10074 /* Use a better score when there is only a case difference. */
|
|
324
|
10075 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
10076 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
10077 else
|
|
|
10078 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
10079
|
|
324
|
10080 if (i > 1 && j > 1)
|
|
323
|
10081 {
|
|
324
|
10082 #ifdef FEAT_MBYTE
|
|
|
10083 if (has_mbyte)
|
|
|
10084 {
|
|
|
10085 pbc = wbadword[i - 2];
|
|
|
10086 pgc = wgoodword[j - 2];
|
|
|
10087 }
|
|
|
10088 else
|
|
|
10089 #endif
|
|
|
10090 {
|
|
|
10091 pbc = badword[i - 2];
|
|
|
10092 pgc = goodword[j - 2];
|
|
|
10093 }
|
|
|
10094 if (bc == pgc && pbc == gc)
|
|
|
10095 {
|
|
|
10096 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
10097 if (t < CNT(i, j))
|
|
|
10098 CNT(i, j) = t;
|
|
|
10099 }
|
|
323
|
10100 }
|
|
|
10101 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
10102 if (t < CNT(i, j))
|
|
|
10103 CNT(i, j) = t;
|
|
|
10104 t = SCORE_INS + CNT(i, j - 1);
|
|
|
10105 if (t < CNT(i, j))
|
|
|
10106 CNT(i, j) = t;
|
|
|
10107 }
|
|
|
10108 }
|
|
|
10109 }
|
|
344
|
10110
|
|
|
10111 i = CNT(badlen - 1, goodlen - 1);
|
|
|
10112 vim_free(cnt);
|
|
|
10113 return i;
|
|
323
|
10114 }
|
|
307
|
10115
|
|
351
|
10116 /*
|
|
|
10117 * ":spelldump"
|
|
|
10118 */
|
|
|
10119 /*ARGSUSED*/
|
|
|
10120 void
|
|
|
10121 ex_spelldump(eap)
|
|
|
10122 exarg_T *eap;
|
|
|
10123 {
|
|
|
10124 buf_T *buf = curbuf;
|
|
|
10125 langp_T *lp;
|
|
|
10126 slang_T *slang;
|
|
|
10127 idx_T arridx[MAXWLEN];
|
|
|
10128 int curi[MAXWLEN];
|
|
|
10129 char_u word[MAXWLEN];
|
|
|
10130 int c;
|
|
|
10131 char_u *byts;
|
|
|
10132 idx_T *idxs;
|
|
|
10133 linenr_T lnum = 0;
|
|
|
10134 int round;
|
|
|
10135 int depth;
|
|
|
10136 int n;
|
|
|
10137 int flags;
|
|
381
|
10138 char_u *region_names = NULL; /* region names being used */
|
|
|
10139 int do_region = TRUE; /* dump region names and numbers */
|
|
|
10140 char_u *p;
|
|
351
|
10141
|
|
|
10142 if (no_spell_checking())
|
|
|
10143 return;
|
|
|
10144
|
|
|
10145 /* Create a new empty buffer by splitting the window. */
|
|
|
10146 do_cmdline_cmd((char_u *)"new");
|
|
|
10147 if (!bufempty() || !buf_valid(buf))
|
|
|
10148 return;
|
|
|
10149
|
|
381
|
10150 /* Find out if we can support regions: All languages must support the same
|
|
|
10151 * regions or none at all. */
|
|
|
10152 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
10153 {
|
|
|
10154 p = lp->lp_slang->sl_regions;
|
|
|
10155 if (p[0] != 0)
|
|
|
10156 {
|
|
|
10157 if (region_names == NULL) /* first language with regions */
|
|
|
10158 region_names = p;
|
|
|
10159 else if (STRCMP(region_names, p) != 0)
|
|
|
10160 {
|
|
|
10161 do_region = FALSE; /* region names are different */
|
|
|
10162 break;
|
|
|
10163 }
|
|
|
10164 }
|
|
|
10165 }
|
|
|
10166
|
|
|
10167 if (do_region && region_names != NULL)
|
|
|
10168 {
|
|
|
10169 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
10170 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
10171 }
|
|
|
10172 else
|
|
|
10173 do_region = FALSE;
|
|
|
10174
|
|
|
10175 /*
|
|
|
10176 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
10177 */
|
|
351
|
10178 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
10179 {
|
|
|
10180 slang = lp->lp_slang;
|
|
|
10181
|
|
|
10182 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
10183 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
10184
|
|
|
10185 /* round 1: case-folded tree
|
|
|
10186 * round 2: keep-case tree */
|
|
|
10187 for (round = 1; round <= 2; ++round)
|
|
|
10188 {
|
|
|
10189 if (round == 1)
|
|
|
10190 {
|
|
|
10191 byts = slang->sl_fbyts;
|
|
|
10192 idxs = slang->sl_fidxs;
|
|
|
10193 }
|
|
|
10194 else
|
|
|
10195 {
|
|
|
10196 byts = slang->sl_kbyts;
|
|
|
10197 idxs = slang->sl_kidxs;
|
|
|
10198 }
|
|
|
10199 if (byts == NULL)
|
|
|
10200 continue; /* array is empty */
|
|
|
10201
|
|
|
10202 depth = 0;
|
|
|
10203 arridx[0] = 0;
|
|
|
10204 curi[0] = 1;
|
|
|
10205 while (depth >= 0 && !got_int)
|
|
|
10206 {
|
|
|
10207 if (curi[depth] > byts[arridx[depth]])
|
|
|
10208 {
|
|
|
10209 /* Done all bytes at this node, go up one level. */
|
|
|
10210 --depth;
|
|
|
10211 line_breakcheck();
|
|
|
10212 }
|
|
|
10213 else
|
|
|
10214 {
|
|
|
10215 /* Do one more byte at this node. */
|
|
|
10216 n = arridx[depth] + curi[depth];
|
|
|
10217 ++curi[depth];
|
|
|
10218 c = byts[n];
|
|
|
10219 if (c == 0)
|
|
|
10220 {
|
|
|
10221 /* End of word, deal with the word.
|
|
|
10222 * Don't use keep-case words in the fold-case tree,
|
|
|
10223 * they will appear in the keep-case tree.
|
|
|
10224 * Only use the word when the region matches. */
|
|
|
10225 flags = (int)idxs[n];
|
|
|
10226 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
10227 && (do_region
|
|
|
10228 || (flags & WF_REGION) == 0
|
|
390
|
10229 || (((unsigned)flags >> 16)
|
|
351
|
10230 & lp->lp_region) != 0))
|
|
|
10231 {
|
|
|
10232 word[depth] = NUL;
|
|
381
|
10233 if (!do_region)
|
|
|
10234 flags &= ~WF_REGION;
|
|
355
|
10235
|
|
|
10236 /* Dump the basic word if there is no prefix or
|
|
|
10237 * when it's the first one. */
|
|
390
|
10238 c = (unsigned)flags >> 24;
|
|
355
|
10239 if (c == 0 || curi[depth] == 2)
|
|
|
10240 dump_word(word, round, flags, lnum++);
|
|
351
|
10241
|
|
|
10242 /* Apply the prefix, if there is one. */
|
|
355
|
10243 if (c != 0)
|
|
351
|
10244 lnum = apply_prefixes(slang, word, round,
|
|
|
10245 flags, lnum);
|
|
|
10246 }
|
|
|
10247 }
|
|
|
10248 else
|
|
|
10249 {
|
|
|
10250 /* Normal char, go one level deeper. */
|
|
|
10251 word[depth++] = c;
|
|
|
10252 arridx[depth] = idxs[n];
|
|
|
10253 curi[depth] = 1;
|
|
|
10254 }
|
|
|
10255 }
|
|
|
10256 }
|
|
|
10257 }
|
|
|
10258 }
|
|
|
10259
|
|
|
10260 /* Delete the empty line that we started with. */
|
|
|
10261 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
10262 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
10263
|
|
|
10264 redraw_later(NOT_VALID);
|
|
|
10265 }
|
|
|
10266
|
|
|
10267 /*
|
|
|
10268 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
10269 */
|
|
|
10270 static void
|
|
|
10271 dump_word(word, round, flags, lnum)
|
|
|
10272 char_u *word;
|
|
|
10273 int round;
|
|
|
10274 int flags;
|
|
|
10275 linenr_T lnum;
|
|
|
10276 {
|
|
|
10277 int keepcap = FALSE;
|
|
|
10278 char_u *p;
|
|
|
10279 char_u cword[MAXWLEN];
|
|
381
|
10280 char_u badword[MAXWLEN + 10];
|
|
|
10281 int i;
|
|
351
|
10282
|
|
|
10283 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
10284 {
|
|
|
10285 /* Need to fix case according to "flags". */
|
|
|
10286 make_case_word(word, cword, flags);
|
|
|
10287 p = cword;
|
|
|
10288 }
|
|
|
10289 else
|
|
|
10290 {
|
|
|
10291 p = word;
|
|
389
|
10292 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
10293 || (flags & WF_FIXCAP) != 0))
|
|
351
|
10294 keepcap = TRUE;
|
|
|
10295 }
|
|
|
10296
|
|
381
|
10297 /* Add flags and regions after a slash. */
|
|
|
10298 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
10299 {
|
|
|
10300 STRCPY(badword, p);
|
|
|
10301 STRCAT(badword, "/");
|
|
351
|
10302 if (keepcap)
|
|
|
10303 STRCAT(badword, "=");
|
|
|
10304 if (flags & WF_BANNED)
|
|
|
10305 STRCAT(badword, "!");
|
|
|
10306 else if (flags & WF_RARE)
|
|
|
10307 STRCAT(badword, "?");
|
|
381
|
10308 if (flags & WF_REGION)
|
|
|
10309 for (i = 0; i < 7; ++i)
|
|
390
|
10310 if (flags & (0x10000 << i))
|
|
381
|
10311 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
10312 p = badword;
|
|
|
10313 }
|
|
|
10314
|
|
|
10315 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
10316 }
|
|
|
10317
|
|
|
10318 /*
|
|
372
|
10319 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
10320 * "word" and append a line to the buffer.
|
|
351
|
10321 * Return the updated line number.
|
|
|
10322 */
|
|
|
10323 static linenr_T
|
|
|
10324 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
10325 slang_T *slang;
|
|
|
10326 char_u *word; /* case-folded word */
|
|
|
10327 int round;
|
|
|
10328 int flags; /* flags with prefix ID */
|
|
|
10329 linenr_T startlnum;
|
|
|
10330 {
|
|
|
10331 idx_T arridx[MAXWLEN];
|
|
|
10332 int curi[MAXWLEN];
|
|
|
10333 char_u prefix[MAXWLEN];
|
|
455
|
10334 char_u word_up[MAXWLEN];
|
|
|
10335 int has_word_up = FALSE;
|
|
351
|
10336 int c;
|
|
|
10337 char_u *byts;
|
|
|
10338 idx_T *idxs;
|
|
|
10339 linenr_T lnum = startlnum;
|
|
|
10340 int depth;
|
|
|
10341 int n;
|
|
|
10342 int len;
|
|
|
10343 int i;
|
|
|
10344
|
|
455
|
10345 /* if the word starts with a lower-case letter make the word with an
|
|
|
10346 * upper-case letter in word_up[]. */
|
|
|
10347 c = PTR2CHAR(word);
|
|
|
10348 if (SPELL_TOUPPER(c) != c)
|
|
|
10349 {
|
|
|
10350 onecap_copy(word, word_up, TRUE);
|
|
|
10351 has_word_up = TRUE;
|
|
|
10352 }
|
|
|
10353
|
|
351
|
10354 byts = slang->sl_pbyts;
|
|
|
10355 idxs = slang->sl_pidxs;
|
|
|
10356 if (byts != NULL) /* array not is empty */
|
|
|
10357 {
|
|
|
10358 /*
|
|
|
10359 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
10360 * When at the end of a prefix check that it supports "flags".
|
|
351
|
10361 */
|
|
|
10362 depth = 0;
|
|
|
10363 arridx[0] = 0;
|
|
|
10364 curi[0] = 1;
|
|
|
10365 while (depth >= 0 && !got_int)
|
|
|
10366 {
|
|
390
|
10367 n = arridx[depth];
|
|
|
10368 len = byts[n];
|
|
|
10369 if (curi[depth] > len)
|
|
351
|
10370 {
|
|
|
10371 /* Done all bytes at this node, go up one level. */
|
|
|
10372 --depth;
|
|
|
10373 line_breakcheck();
|
|
|
10374 }
|
|
|
10375 else
|
|
|
10376 {
|
|
|
10377 /* Do one more byte at this node. */
|
|
390
|
10378 n += curi[depth];
|
|
351
|
10379 ++curi[depth];
|
|
|
10380 c = byts[n];
|
|
|
10381 if (c == 0)
|
|
|
10382 {
|
|
|
10383 /* End of prefix, find out how many IDs there are. */
|
|
|
10384 for (i = 1; i < len; ++i)
|
|
|
10385 if (byts[n + i] != 0)
|
|
|
10386 break;
|
|
|
10387 curi[depth] += i - 1;
|
|
|
10388
|
|
455
|
10389 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
10390 if (c != 0)
|
|
351
|
10391 {
|
|
376
|
10392 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
10393 dump_word(prefix, round,
|
|
455
|
10394 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
10395 : flags, lnum++);
|
|
351
|
10396 }
|
|
455
|
10397
|
|
|
10398 /* Check for prefix that matches the word when the
|
|
|
10399 * first letter is upper-case, but only if the prefix has
|
|
|
10400 * a condition. */
|
|
|
10401 if (has_word_up)
|
|
|
10402 {
|
|
|
10403 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
10404 TRUE);
|
|
|
10405 if (c != 0)
|
|
|
10406 {
|
|
|
10407 vim_strncpy(prefix + depth, word_up,
|
|
|
10408 MAXWLEN - depth - 1);
|
|
|
10409 dump_word(prefix, round,
|
|
|
10410 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
10411 : flags, lnum++);
|
|
|
10412 }
|
|
|
10413 }
|
|
351
|
10414 }
|
|
|
10415 else
|
|
|
10416 {
|
|
|
10417 /* Normal char, go one level deeper. */
|
|
|
10418 prefix[depth++] = c;
|
|
|
10419 arridx[depth] = idxs[n];
|
|
|
10420 curi[depth] = 1;
|
|
|
10421 }
|
|
|
10422 }
|
|
|
10423 }
|
|
|
10424 }
|
|
|
10425
|
|
|
10426 return lnum;
|
|
|
10427 }
|
|
|
10428
|
|
236
|
10429 #endif /* FEAT_SYN_HL */
|
