|
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
|
|
300
|
53 /*
|
|
324
|
54 * Use this to adjust the score after finding suggestions, based on the
|
|
|
55 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
56 * it for every possible suggestion.
|
|
|
57 * Disadvantage: When "the" is typed as "hte" it sounds different and goes
|
|
|
58 * down in the list.
|
|
344
|
59 * Used when 'spellsuggest' is set to "best".
|
|
|
60 */
|
|
|
61 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
62
|
|
|
63 /*
|
|
|
64 * The double scoring mechanism is based on the principle that there are two
|
|
|
65 * kinds of spelling mistakes:
|
|
|
66 * 1. You know how to spell the word, but mistype something. This results in
|
|
|
67 * a small editing distance (character swapped/omitted/inserted) and
|
|
|
68 * possibly a word that sounds completely different.
|
|
|
69 * 2. You don't know how to spell the word and type something that sounds
|
|
|
70 * right. The edit distance can be big but the word is similar after
|
|
|
71 * sound-folding.
|
|
|
72 * Since scores for these two mistakes will be very different we use a list
|
|
|
73 * for each.
|
|
|
74 * The sound-folding is slow, only do double scoring when 'spellsuggest' is
|
|
|
75 * "double".
|
|
323
|
76 */
|
|
|
77
|
|
|
78 /*
|
|
339
|
79 * Vim spell file format: <HEADER>
|
|
|
80 * <SUGGEST>
|
|
|
81 * <LWORDTREE>
|
|
|
82 * <KWORDTREE>
|
|
|
83 * <PREFIXTREE>
|
|
300
|
84 *
|
|
339
|
85 * <HEADER>: <fileID>
|
|
|
86 * <regioncnt> <regionname> ...
|
|
|
87 * <charflagslen> <charflags>
|
|
|
88 * <fcharslen> <fchars>
|
|
366
|
89 * <midwordlen> <midword>
|
|
339
|
90 * <prefcondcnt> <prefcond> ...
|
|
300
|
91 *
|
|
455
|
92 * <fileID> 10 bytes "VIMspell09"
|
|
300
|
93 * <regioncnt> 1 byte number of regions following (8 supported)
|
|
307
|
94 * <regionname> 2 bytes Region name: ca, au, etc. Lower case.
|
|
300
|
95 * First <regionname> is region 1.
|
|
|
96 *
|
|
|
97 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
98 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
99 * 0x01 word character CF_WORD
|
|
|
100 * 0x02 upper-case character CF_UPPER
|
|
300
|
101 * <fcharslen> 2 bytes Number of bytes in <fchars>.
|
|
|
102 * <fchars> N bytes Folded characters, first one is for character 128.
|
|
|
103 *
|
|
366
|
104 * <midwordlen> 2 bytes Number of bytes in <midword>.
|
|
|
105 * <midword> N bytes Characters that are word characters only when used
|
|
|
106 * in the middle of a word.
|
|
|
107 *
|
|
339
|
108 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
109 *
|
|
|
110 * <prefcond> : <condlen> <condstr>
|
|
|
111 *
|
|
|
112 * <condlen> 1 byte Length of <condstr>.
|
|
|
113 *
|
|
|
114 * <condstr> N bytes Condition for the prefix.
|
|
|
115 *
|
|
300
|
116 *
|
|
323
|
117 * <SUGGEST> : <repcount> <rep> ...
|
|
|
118 * <salflags> <salcount> <sal> ...
|
|
|
119 * <maplen> <mapstr>
|
|
|
120 *
|
|
|
121 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
|
122 *
|
|
|
123 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
|
124 *
|
|
|
125 * <repfromlen> 1 byte length of <repfrom>
|
|
|
126 *
|
|
|
127 * <repfrom> N bytes "from" part of replacement
|
|
|
128 *
|
|
|
129 * <reptolen> 1 byte length of <repto>
|
|
|
130 *
|
|
|
131 * <repto> N bytes "to" part of replacement
|
|
300
|
132 *
|
|
323
|
133 * <salflags> 1 byte flags for soundsalike conversion:
|
|
|
134 * SAL_F0LLOWUP
|
|
|
135 * SAL_COLLAPSE
|
|
|
136 * SAL_REM_ACCENTS
|
|
375
|
137 * SAL_SOFO: SOFOFROM and SOFOTO used instead of SAL
|
|
|
138 *
|
|
|
139 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
140 *
|
|
|
141 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
|
142 *
|
|
375
|
143 * <salfromlen> 1-2 bytes length of <salfrom> (2 bytes for SAL_SOFO)
|
|
323
|
144 *
|
|
|
145 * <salfrom> N bytes "from" part of soundsalike
|
|
|
146 *
|
|
375
|
147 * <saltolen> 1-2 bytes length of <salto> (2 bytes for SAL_SOFO)
|
|
323
|
148 *
|
|
|
149 * <salto> N bytes "to" part of soundsalike
|
|
|
150 *
|
|
|
151 * <maplen> 2 bytes length of <mapstr>, MSB first
|
|
|
152 *
|
|
|
153 * <mapstr> N bytes String with sequences of similar characters,
|
|
|
154 * separated by slashes.
|
|
300
|
155 *
|
|
|
156 *
|
|
|
157 * <LWORDTREE>: <wordtree>
|
|
|
158 *
|
|
339
|
159 * <KWORDTREE>: <wordtree>
|
|
|
160 *
|
|
|
161 * <PREFIXTREE>: <wordtree>
|
|
|
162 *
|
|
|
163 *
|
|
300
|
164 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
165 *
|
|
|
166 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
167 *
|
|
|
168 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
169 *
|
|
|
170 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
171 * follow in sorted order.
|
|
|
172 *
|
|
339
|
173 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
455
|
174 * | <flags> [<flags2>] [<region>] [<prefixID>]
|
|
|
175 * | [<pflags>] <prefixID> <prefcondnr> ]
|
|
300
|
176 *
|
|
|
177 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
178 * BY_NOFLAGS: End of word without flags and for all
|
|
|
179 * regions.
|
|
366
|
180 * For PREFIXTREE <prefixID> and
|
|
|
181 * <prefcondnr> follow.
|
|
390
|
182 * BY_FLAGS: End of word, <flags> follow.
|
|
455
|
183 * For PREFIXTREE <pflags>, <prefixID>
|
|
|
184 * and <prefcondnr> follow.
|
|
390
|
185 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
186 * follow. Not used in PREFIXTREE.
|
|
|
187 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
188 * and <xbyte> follow.
|
|
|
189 *
|
|
|
190 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
191 *
|
|
|
192 * <xbyte> 1 byte byte value of the sibling.
|
|
|
193 *
|
|
|
194 * <flags> 1 byte bitmask of:
|
|
|
195 * WF_ALLCAP word must have only capitals
|
|
|
196 * WF_ONECAP first char of word must be capital
|
|
389
|
197 * WF_KEEPCAP keep-case word
|
|
|
198 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
199 * WF_RARE rare word
|
|
389
|
200 * WF_BANNED bad word
|
|
300
|
201 * WF_REGION <region> follows
|
|
339
|
202 * WF_PFX <prefixID> follows
|
|
300
|
203 *
|
|
390
|
204 * <flags2> 1 byte Only used when there are postponed prefixes.
|
|
|
205 * Bitmask of:
|
|
|
206 * WF_HAS_AFF >> 8 word includes affix
|
|
|
207 *
|
|
455
|
208 * <pflags> 1 byte bitmask of:
|
|
|
209 * WFP_RARE rare prefix
|
|
|
210 * WFP_NC non-combining prefix
|
|
|
211 * WFP_UP letter after prefix made upper case
|
|
|
212 *
|
|
300
|
213 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
214 * omitted it's valid in all regions.
|
|
|
215 * Lowest bit is for region 1.
|
|
|
216 *
|
|
339
|
217 * <prefixID> 1 byte ID of prefix that can be used with this word. For
|
|
|
218 * PREFIXTREE used for the required prefix ID.
|
|
|
219 *
|
|
|
220 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
221 * from HEADER.
|
|
300
|
222 *
|
|
|
223 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
224 */
|
|
|
225
|
|
223
|
226 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
227 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
228 #endif
|
|
|
229
|
|
|
230 #include "vim.h"
|
|
|
231
|
|
|
232 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
233
|
|
|
234 #ifdef HAVE_FCNTL_H
|
|
|
235 # include <fcntl.h>
|
|
|
236 #endif
|
|
|
237
|
|
323
|
238 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
239 Some places assume a word length fits in a
|
|
|
240 byte, thus it can't be above 255. */
|
|
226
|
241
|
|
324
|
242 /* Type used for indexes in the word tree need to be at least 3 bytes. If int
|
|
|
243 * is 8 bytes we could use something smaller, but what? */
|
|
|
244 #if SIZEOF_INT > 2
|
|
|
245 typedef int idx_T;
|
|
|
246 #else
|
|
|
247 typedef long idx_T;
|
|
|
248 #endif
|
|
|
249
|
|
|
250 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
251 * comes above it. */
|
|
300
|
252 #define WF_REGION 0x01 /* region byte follows */
|
|
|
253 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
254 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
255 #define WF_RARE 0x08 /* rare word */
|
|
307
|
256 #define WF_BANNED 0x10 /* bad word */
|
|
390
|
257 #define WF_PFX 0x20 /* prefix ID follows */
|
|
389
|
258 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
259 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
260
|
|
390
|
261 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
262 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
|
263
|
|
389
|
264 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
265
|
|
455
|
266 /* flags for <pflags> */
|
|
|
267 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
268 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
269 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
270
|
|
390
|
271 /* flags for postponed prefixes. Must be above prefixID (one byte)
|
|
|
272 * and prefcondnr (two bytes). */
|
|
455
|
273 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
274 * postponed prefix */
|
|
|
275 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
276 * postponed prefix */
|
|
|
277 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
278 * postponed prefix */
|
|
390
|
279
|
|
|
280 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
281 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
282 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
283 * postponed prefix: no <pflags> */
|
|
|
284 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
285 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
286 * postponed prefix: <pflags> follows */
|
|
|
287 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
288 * follow; never used in prefix tree */
|
|
|
289 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
290
|
|
323
|
291 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
292 * and si_sal. Not for sl_sal!
|
|
323
|
293 * One replacement: from "ft_from" to "ft_to". */
|
|
|
294 typedef struct fromto_S
|
|
236
|
295 {
|
|
323
|
296 char_u *ft_from;
|
|
|
297 char_u *ft_to;
|
|
|
298 } fromto_T;
|
|
236
|
299
|
|
344
|
300 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
301 * The info is split for quick processing by spell_soundfold().
|
|
|
302 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
303 typedef struct salitem_S
|
|
|
304 {
|
|
|
305 char_u *sm_lead; /* leading letters */
|
|
|
306 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
307 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
308 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
309 char_u *sm_to; /* replacement. */
|
|
372
|
310 #ifdef FEAT_MBYTE
|
|
|
311 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
312 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
313 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
314 #endif
|
|
344
|
315 } salitem_T;
|
|
|
316
|
|
375
|
317 #ifdef FEAT_MBYTE
|
|
|
318 typedef int salfirst_T;
|
|
|
319 #else
|
|
|
320 typedef short salfirst_T;
|
|
|
321 #endif
|
|
|
322
|
|
236
|
323 /*
|
|
243
|
324 * Structure used to store words and other info for one language, loaded from
|
|
|
325 * a .spl file.
|
|
300
|
326 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
327 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
328 *
|
|
|
329 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
330 * the number of possible bytes, sorted on byte value:
|
|
|
331 * <len> <byte1> <byte2> ...
|
|
|
332 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
333 * byte in "byts".
|
|
|
334 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
390
|
335 * the flags, region mask and prefixID for the word. There may be several
|
|
|
336 * zeros in sequence for alternative flag/region combinations.
|
|
236
|
337 */
|
|
|
338 typedef struct slang_S slang_T;
|
|
|
339 struct slang_S
|
|
|
340 {
|
|
|
341 slang_T *sl_next; /* next language */
|
|
|
342 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
343 char_u *sl_fname; /* name of .spl file */
|
|
323
|
344 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
345
|
|
300
|
346 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
347 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
348 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
349 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
350 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
351 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
352
|
|
236
|
353 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
354
|
|
376
|
355 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
356
|
|
339
|
357 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
358 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
359
|
|
323
|
360 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
361 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
362 there is none */
|
|
344
|
363 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
364 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
365 there is none */
|
|
375
|
366 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
367 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
368 * "sl_sal" is a list of wide char lists. */
|
|
323
|
369 int sl_followup; /* SAL followup */
|
|
|
370 int sl_collapse; /* SAL collapse_result */
|
|
|
371 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
372 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
373 #ifdef FEAT_MBYTE
|
|
|
374 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
375 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
376 #else
|
|
|
377 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
378 #endif
|
|
236
|
379 };
|
|
|
380
|
|
243
|
381 /* First language that is loaded, start of the linked list of loaded
|
|
|
382 * languages. */
|
|
236
|
383 static slang_T *first_lang = NULL;
|
|
|
384
|
|
323
|
385 /* Flags used in .spl file for soundsalike flags. */
|
|
|
386 #define SAL_F0LLOWUP 1
|
|
|
387 #define SAL_COLLAPSE 2
|
|
|
388 #define SAL_REM_ACCENTS 4
|
|
375
|
389 #define SAL_SOFO 8 /* SOFOFROM and SOFOTO instead of SAL */
|
|
323
|
390
|
|
236
|
391 /*
|
|
|
392 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
393 */
|
|
|
394 typedef struct langp_S
|
|
|
395 {
|
|
|
396 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
397 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
398 } langp_T;
|
|
|
399
|
|
|
400 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
401
|
|
307
|
402 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
403
|
|
|
404 /* Result values. Lower number is accepted over higher one. */
|
|
|
405 #define SP_BANNED -1
|
|
236
|
406 #define SP_OK 0
|
|
307
|
407 #define SP_RARE 1
|
|
|
408 #define SP_LOCAL 2
|
|
|
409 #define SP_BAD 3
|
|
236
|
410
|
|
455
|
411 #define VIMSPELLMAGIC "VIMspell09" /* string at start of Vim spell file */
|
|
236
|
412 #define VIMSPELLMAGICL 10
|
|
|
413
|
|
381
|
414 /* file used for "zG" and "zW" */
|
|
385
|
415 static char_u *int_wordlist = NULL;
|
|
381
|
416
|
|
236
|
417 /*
|
|
323
|
418 * Information used when looking for suggestions.
|
|
|
419 */
|
|
|
420 typedef struct suginfo_S
|
|
|
421 {
|
|
|
422 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
423 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
424 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
425 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
426 char_u *su_badptr; /* start of bad word in line */
|
|
|
427 int su_badlen; /* length of detected bad word in line */
|
|
346
|
428 int su_badflags; /* caps flags for bad word */
|
|
323
|
429 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
430 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
431 hashtab_T su_banned; /* table with banned words */
|
|
|
432 } suginfo_T;
|
|
|
433
|
|
|
434 /* One word suggestion. Used in "si_ga". */
|
|
|
435 typedef struct suggest_S
|
|
|
436 {
|
|
|
437 char_u *st_word; /* suggested word, allocated string */
|
|
|
438 int st_orglen; /* length of replaced text */
|
|
|
439 int st_score; /* lower is better */
|
|
344
|
440 int st_altscore; /* used when st_score compares equal */
|
|
|
441 int st_salscore; /* st_score is for soundalike */
|
|
324
|
442 int st_had_bonus; /* bonus already included in score */
|
|
323
|
443 } suggest_T;
|
|
|
444
|
|
344
|
445 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
446
|
|
324
|
447 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
448 * called the score may change, thus we need to keep more than what is
|
|
|
449 * displayed. */
|
|
351
|
450 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
451
|
|
|
452 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
453 * of suggestions that are not going to be displayed. */
|
|
344
|
454 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
455
|
|
|
456 /* score for various changes */
|
|
344
|
457 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
458 #define SCORE_ICASE 52 /* slightly different case */
|
|
|
459 #define SCORE_REGION 70 /* word is for different region */
|
|
|
460 #define SCORE_RARE 180 /* rare word */
|
|
|
461 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
462 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
463 #define SCORE_REP 87 /* REP replacement */
|
|
|
464 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
465 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
324
|
466 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
467 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
324
|
468 #define SCORE_INS 96 /* insert a character */
|
|
358
|
469 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
366
|
470 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
471
|
|
372
|
472 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
473 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
474 * 350 allows for about three changes. */
|
|
344
|
475
|
|
|
476 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
477 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
478
|
|
|
479 /*
|
|
236
|
480 * Structure to store info for word matching.
|
|
|
481 */
|
|
|
482 typedef struct matchinf_S
|
|
|
483 {
|
|
|
484 langp_T *mi_lp; /* info for language and region */
|
|
243
|
485
|
|
|
486 /* pointers to original text to be checked */
|
|
236
|
487 char_u *mi_word; /* start of word being checked */
|
|
339
|
488 char_u *mi_end; /* end of matching word so far */
|
|
243
|
489 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
490 char_u *mi_cend; /* char after what was used for
|
|
|
491 mi_capflags */
|
|
243
|
492
|
|
|
493 /* case-folded text */
|
|
|
494 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
495 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
496
|
|
339
|
497 /* for when checking word after a prefix */
|
|
|
498 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
|
499 prefixID/condition */
|
|
|
500 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
501 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
502 #ifdef FEAT_MBYTE
|
|
|
503 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
504 case */
|
|
|
505 #else
|
|
|
506 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
507 #endif
|
|
339
|
508
|
|
243
|
509 /* others */
|
|
236
|
510 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
511 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
512 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
513 } matchinf_T;
|
|
|
514
|
|
307
|
515 /*
|
|
|
516 * The tables used for recognizing word characters according to spelling.
|
|
|
517 * These are only used for the first 256 characters of 'encoding'.
|
|
|
518 */
|
|
|
519 typedef struct spelltab_S
|
|
|
520 {
|
|
|
521 char_u st_isw[256]; /* flags: is word char */
|
|
|
522 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
523 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
524 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
525 } spelltab_T;
|
|
|
526
|
|
|
527 static spelltab_T spelltab;
|
|
|
528 static int did_set_spelltab;
|
|
|
529
|
|
324
|
530 #define CF_WORD 0x01
|
|
|
531 #define CF_UPPER 0x02
|
|
307
|
532
|
|
|
533 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
534 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
535 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
536 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
537 #ifdef FEAT_MBYTE
|
|
|
538 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
539 #endif
|
|
339
|
540 static void write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
541
|
|
|
542 /*
|
|
344
|
543 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
544 */
|
|
|
545 typedef enum
|
|
|
546 {
|
|
344
|
547 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
548 * ends); if badword ends there is a match, otherwise
|
|
|
549 * try splitting word. */
|
|
375
|
550 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
551 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
552 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
553 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
554 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
555 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
556 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
557 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
558 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
559 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
560 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
561 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
562 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
563 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
564 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
565 STATE_FINAL /* End of this node. */
|
|
|
566 } state_T;
|
|
|
567
|
|
|
568 /*
|
|
346
|
569 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
570 */
|
|
|
571 typedef struct trystate_S
|
|
|
572 {
|
|
330
|
573 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
574 int ts_score; /* score */
|
|
344
|
575 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
576 short ts_curi; /* index in list of child nodes */
|
|
|
577 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
578 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
579 char_u ts_twordlen; /* valid length of tword[] */
|
|
375
|
580 char_u ts_prefixdepth; /* stack depth for end of prefix or PREFIXTREE
|
|
|
581 * or NOPREFIX */
|
|
330
|
582 #ifdef FEAT_MBYTE
|
|
|
583 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
584 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
585 char_u ts_isdiff; /* DIFF_ values */
|
|
|
586 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
587 #endif
|
|
323
|
588 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
330
|
589 char_u ts_save_splitoff; /* su_splitoff saved here */
|
|
346
|
590 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
591 } trystate_T;
|
|
|
592
|
|
330
|
593 /* values for ts_isdiff */
|
|
|
594 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
595 #define DIFF_YES 1 /* different byte found */
|
|
|
596 #define DIFF_INSERT 2 /* inserting character */
|
|
|
597
|
|
375
|
598 /* special values ts_prefixdepth */
|
|
|
599 #define PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
600 #define NOPREFIX 0xff /* not using prefixes */
|
|
|
601
|
|
339
|
602 /* mode values for find_word */
|
|
|
603 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
604 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
605 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
606
|
|
236
|
607 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
608 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
609 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
610 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
455
|
611 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
339
|
612 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
613 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
614 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
615 static int no_spell_checking __ARGS((void));
|
|
307
|
616 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
617 static char_u *spell_enc __ARGS((void));
|
|
385
|
618 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
619 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
620 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
621 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
381
|
622 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
623 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
624 #ifdef FEAT_MBYTE
|
|
|
625 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
626 #endif
|
|
339
|
627 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
628 static void clear_midword __ARGS((buf_T *buf));
|
|
|
629 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
630 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
631 static int captype __ARGS((char_u *word, char_u *end));
|
|
323
|
632 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
389
|
633 static int set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
634 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
635 static void write_spell_chartab __ARGS((FILE *fd));
|
|
|
636 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
388
|
637 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
638 #ifdef FEAT_EVAL
|
|
|
639 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
640 #endif
|
|
|
641 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
642 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
643 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
644 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
645 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
646 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
647 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
648 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
649 #ifdef FEAT_MBYTE
|
|
|
650 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
651 #endif
|
|
323
|
652 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
653 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
654 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
655 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
656 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
657 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
658 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
659 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
660 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
661 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
662 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
663 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
664 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
665 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
666 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
667 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
668 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
669 #ifdef FEAT_MBYTE
|
|
375
|
670 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
671 #endif
|
|
344
|
672 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
673 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
674 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
675 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
676
|
|
324
|
677 /*
|
|
|
678 * Use our own character-case definitions, because the current locale may
|
|
|
679 * differ from what the .spl file uses.
|
|
|
680 * These must not be called with negative number!
|
|
|
681 */
|
|
|
682 #ifndef FEAT_MBYTE
|
|
|
683 /* Non-multi-byte implementation. */
|
|
|
684 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
685 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
686 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
687 #else
|
|
447
|
688 # if defined(HAVE_WCHAR_H)
|
|
|
689 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
690 # endif
|
|
324
|
691 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
692 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
693 * the "w" library function for characters above 255 if available. */
|
|
|
694 # ifdef HAVE_TOWLOWER
|
|
|
695 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
696 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
697 # else
|
|
|
698 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
699 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
700 # endif
|
|
|
701
|
|
|
702 # ifdef HAVE_TOWUPPER
|
|
|
703 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
704 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
705 # else
|
|
|
706 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
707 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
708 # endif
|
|
|
709
|
|
|
710 # ifdef HAVE_ISWUPPER
|
|
|
711 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
712 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
713 # else
|
|
|
714 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
715 : (c) < 256 ? spelltab.st_isu[c] : (c))
|
|
|
716 # endif
|
|
|
717 #endif
|
|
|
718
|
|
307
|
719
|
|
|
720 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
721 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
236
|
722
|
|
|
723 /*
|
|
|
724 * Main spell-checking function.
|
|
300
|
725 * "ptr" points to a character that could be the start of a word.
|
|
236
|
726 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
727 * or when it's OK it remains unchanged.
|
|
|
728 * This must only be called when 'spelllang' is not empty.
|
|
323
|
729 *
|
|
385
|
730 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
731 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
732 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
733 * worry.
|
|
323
|
734 *
|
|
236
|
735 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
736 * caller can skip over the word.
|
|
|
737 */
|
|
|
738 int
|
|
385
|
739 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
740 win_T *wp; /* current window */
|
|
|
741 char_u *ptr;
|
|
|
742 int *attrp;
|
|
385
|
743 int *capcol; /* column to check for Capital */
|
|
236
|
744 {
|
|
|
745 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
746 be passed to functions quickly. */
|
|
344
|
747 int nrlen = 0; /* found a number first */
|
|
385
|
748 int c;
|
|
236
|
749
|
|
307
|
750 /* A word never starts at a space or a control character. Return quickly
|
|
|
751 * then, skipping over the character. */
|
|
|
752 if (*ptr <= ' ')
|
|
|
753 return 1;
|
|
236
|
754
|
|
344
|
755 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
756 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
757 * "3GPP". */
|
|
300
|
758 if (*ptr >= '0' && *ptr <= '9')
|
|
|
759 {
|
|
316
|
760 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
761 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
762 else
|
|
|
763 {
|
|
344
|
764 mi.mi_end = skipdigits(ptr);
|
|
|
765 nrlen = mi.mi_end - ptr;
|
|
300
|
766 }
|
|
376
|
767 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
768 return (int)(mi.mi_end - ptr);
|
|
346
|
769
|
|
|
770 /* Try including the digits in the word. */
|
|
|
771 mi.mi_fend = ptr + nrlen;
|
|
|
772 }
|
|
|
773 else
|
|
|
774 mi.mi_fend = ptr;
|
|
|
775
|
|
|
776 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
777 mi.mi_word = ptr;
|
|
376
|
778 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
779 {
|
|
|
780 do
|
|
|
781 {
|
|
307
|
782 mb_ptr_adv(mi.mi_fend);
|
|
376
|
783 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
784
|
|
|
785 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
786 {
|
|
|
787 /* Check word starting with capital letter. */
|
|
455
|
788 c = PTR2CHAR(ptr);
|
|
385
|
789 if (!SPELL_ISUPPER(c))
|
|
|
790 {
|
|
|
791 *attrp = highlight_attr[HLF_SPC];
|
|
|
792 return (int)(mi.mi_fend - ptr);
|
|
|
793 }
|
|
|
794 }
|
|
|
795 }
|
|
|
796 if (capcol != NULL)
|
|
|
797 *capcol = -1;
|
|
344
|
798
|
|
|
799 /* We always use the characters up to the next non-word character,
|
|
|
800 * also for bad words. */
|
|
|
801 mi.mi_end = mi.mi_fend;
|
|
|
802
|
|
|
803 /* Check caps type later. */
|
|
|
804 mi.mi_capflags = 0;
|
|
|
805 mi.mi_cend = NULL;
|
|
376
|
806 mi.mi_buf = wp->w_buffer;
|
|
344
|
807
|
|
|
808 /* Include one non-word character so that we can check for the
|
|
|
809 * word end. */
|
|
|
810 if (*mi.mi_fend != NUL)
|
|
|
811 mb_ptr_adv(mi.mi_fend);
|
|
|
812
|
|
|
813 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
814 MAXWLEN + 1);
|
|
|
815 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
816
|
|
|
817 /* The word is bad unless we recognize it. */
|
|
|
818 mi.mi_result = SP_BAD;
|
|
|
819
|
|
|
820 /*
|
|
|
821 * Loop over the languages specified in 'spelllang'.
|
|
|
822 * We check them all, because a matching word may be longer than an
|
|
|
823 * already found matching word.
|
|
|
824 */
|
|
|
825 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
826 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
827 {
|
|
|
828 /* Check for a matching word in case-folded words. */
|
|
|
829 find_word(&mi, FIND_FOLDWORD);
|
|
|
830
|
|
|
831 /* Check for a matching word in keep-case words. */
|
|
|
832 find_word(&mi, FIND_KEEPWORD);
|
|
|
833
|
|
|
834 /* Check for matching prefixes. */
|
|
|
835 find_prefix(&mi);
|
|
|
836 }
|
|
|
837
|
|
|
838 if (mi.mi_result != SP_OK)
|
|
|
839 {
|
|
346
|
840 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
841 * rare and local words, e.g., "3GPP". */
|
|
344
|
842 if (nrlen > 0)
|
|
346
|
843 {
|
|
|
844 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
845 return nrlen;
|
|
|
846 }
|
|
344
|
847
|
|
|
848 /* When we are at a non-word character there is no error, just
|
|
|
849 * skip over the character (try looking for a word after it). */
|
|
390
|
850 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
851 {
|
|
385
|
852 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
853 {
|
|
|
854 regmatch_T regmatch;
|
|
|
855
|
|
|
856 /* Check for end of sentence. */
|
|
|
857 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
858 regmatch.rm_ic = FALSE;
|
|
|
859 if (vim_regexec(®match, ptr, 0))
|
|
|
860 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
861 }
|
|
|
862
|
|
344
|
863 #ifdef FEAT_MBYTE
|
|
|
864 if (has_mbyte)
|
|
|
865 return mb_ptr2len_check(ptr);
|
|
|
866 #endif
|
|
|
867 return 1;
|
|
300
|
868 }
|
|
243
|
869
|
|
344
|
870 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
871 *attrp = highlight_attr[HLF_SPB];
|
|
|
872 else if (mi.mi_result == SP_RARE)
|
|
|
873 *attrp = highlight_attr[HLF_SPR];
|
|
|
874 else
|
|
|
875 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
876 }
|
|
|
877
|
|
300
|
878 return (int)(mi.mi_end - ptr);
|
|
236
|
879 }
|
|
|
880
|
|
|
881 /*
|
|
300
|
882 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
883 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
884 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
885 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
886 * tree.
|
|
300
|
887 *
|
|
|
888 * For a match mip->mi_result is updated.
|
|
243
|
889 */
|
|
|
890 static void
|
|
339
|
891 find_word(mip, mode)
|
|
243
|
892 matchinf_T *mip;
|
|
339
|
893 int mode;
|
|
243
|
894 {
|
|
324
|
895 idx_T arridx = 0;
|
|
300
|
896 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
897 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
898 int endidxcnt = 0;
|
|
|
899 int len;
|
|
|
900 int wlen = 0;
|
|
|
901 int flen;
|
|
|
902 int c;
|
|
|
903 char_u *ptr;
|
|
324
|
904 idx_T lo, hi, m;
|
|
243
|
905 #ifdef FEAT_MBYTE
|
|
300
|
906 char_u *s;
|
|
339
|
907 char_u *p;
|
|
307
|
908 #endif
|
|
|
909 int res = SP_BAD;
|
|
300
|
910 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
911 unsigned flags;
|
|
|
912 char_u *byts;
|
|
324
|
913 idx_T *idxs;
|
|
339
|
914
|
|
|
915 if (mode == FIND_KEEPWORD)
|
|
236
|
916 {
|
|
300
|
917 /* Check for word with matching case in keep-case tree. */
|
|
|
918 ptr = mip->mi_word;
|
|
|
919 flen = 9999; /* no case folding, always enough bytes */
|
|
|
920 byts = slang->sl_kbyts;
|
|
|
921 idxs = slang->sl_kidxs;
|
|
236
|
922 }
|
|
|
923 else
|
|
|
924 {
|
|
300
|
925 /* Check for case-folded in case-folded tree. */
|
|
|
926 ptr = mip->mi_fword;
|
|
|
927 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
928 byts = slang->sl_fbyts;
|
|
|
929 idxs = slang->sl_fidxs;
|
|
339
|
930
|
|
|
931 if (mode == FIND_PREFIX)
|
|
|
932 {
|
|
|
933 /* Skip over the prefix. */
|
|
|
934 wlen = mip->mi_prefixlen;
|
|
|
935 flen -= mip->mi_prefixlen;
|
|
|
936 }
|
|
243
|
937 }
|
|
|
938
|
|
300
|
939 if (byts == NULL)
|
|
|
940 return; /* array is empty */
|
|
236
|
941
|
|
|
942 /*
|
|
307
|
943 * Repeat advancing in the tree until:
|
|
|
944 * - there is a byte that doesn't match,
|
|
|
945 * - we reach the end of the tree,
|
|
|
946 * - or we reach the end of the line.
|
|
236
|
947 */
|
|
300
|
948 for (;;)
|
|
236
|
949 {
|
|
346
|
950 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
951 flen = fold_more(mip);
|
|
300
|
952
|
|
|
953 len = byts[arridx++];
|
|
|
954
|
|
|
955 /* If the first possible byte is a zero the word could end here.
|
|
|
956 * Remember this index, we first check for the longest word. */
|
|
|
957 if (byts[arridx] == 0)
|
|
|
958 {
|
|
307
|
959 if (endidxcnt == MAXWLEN)
|
|
|
960 {
|
|
|
961 /* Must be a corrupted spell file. */
|
|
|
962 EMSG(_(e_format));
|
|
|
963 return;
|
|
|
964 }
|
|
300
|
965 endlen[endidxcnt] = wlen;
|
|
|
966 endidx[endidxcnt++] = arridx++;
|
|
|
967 --len;
|
|
|
968
|
|
|
969 /* Skip over the zeros, there can be several flag/region
|
|
|
970 * combinations. */
|
|
|
971 while (len > 0 && byts[arridx] == 0)
|
|
|
972 {
|
|
|
973 ++arridx;
|
|
|
974 --len;
|
|
|
975 }
|
|
|
976 if (len == 0)
|
|
|
977 break; /* no children, word must end here */
|
|
|
978 }
|
|
|
979
|
|
|
980 /* Stop looking at end of the line. */
|
|
|
981 if (ptr[wlen] == NUL)
|
|
|
982 break;
|
|
|
983
|
|
|
984 /* Perform a binary search in the list of accepted bytes. */
|
|
|
985 c = ptr[wlen];
|
|
346
|
986 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
987 c = ' ';
|
|
300
|
988 lo = arridx;
|
|
|
989 hi = arridx + len - 1;
|
|
|
990 while (lo < hi)
|
|
|
991 {
|
|
|
992 m = (lo + hi) / 2;
|
|
|
993 if (byts[m] > c)
|
|
|
994 hi = m - 1;
|
|
|
995 else if (byts[m] < c)
|
|
|
996 lo = m + 1;
|
|
|
997 else
|
|
|
998 {
|
|
|
999 lo = hi = m;
|
|
|
1000 break;
|
|
236
|
1001 }
|
|
|
1002 }
|
|
300
|
1003
|
|
|
1004 /* Stop if there is no matching byte. */
|
|
|
1005 if (hi < lo || byts[lo] != c)
|
|
|
1006 break;
|
|
|
1007
|
|
|
1008 /* Continue at the child (if there is one). */
|
|
|
1009 arridx = idxs[lo];
|
|
|
1010 ++wlen;
|
|
|
1011 --flen;
|
|
346
|
1012
|
|
|
1013 /* One space in the good word may stand for several spaces in the
|
|
|
1014 * checked word. */
|
|
|
1015 if (c == ' ')
|
|
|
1016 {
|
|
|
1017 for (;;)
|
|
|
1018 {
|
|
|
1019 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1020 flen = fold_more(mip);
|
|
|
1021 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1022 break;
|
|
|
1023 ++wlen;
|
|
|
1024 --flen;
|
|
|
1025 }
|
|
|
1026 }
|
|
236
|
1027 }
|
|
|
1028
|
|
300
|
1029 /*
|
|
|
1030 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1031 * first.
|
|
|
1032 */
|
|
|
1033 while (endidxcnt > 0)
|
|
|
1034 {
|
|
|
1035 --endidxcnt;
|
|
|
1036 arridx = endidx[endidxcnt];
|
|
|
1037 wlen = endlen[endidxcnt];
|
|
236
|
1038
|
|
300
|
1039 #ifdef FEAT_MBYTE
|
|
|
1040 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1041 continue; /* not at first byte of character */
|
|
|
1042 #endif
|
|
376
|
1043 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
300
|
1044 continue; /* next char is a word character */
|
|
|
1045
|
|
|
1046 #ifdef FEAT_MBYTE
|
|
339
|
1047 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1048 {
|
|
|
1049 /* Compute byte length in original word, length may change
|
|
339
|
1050 * when folding case. This can be slow, take a shortcut when the
|
|
|
1051 * case-folded word is equal to the keep-case word. */
|
|
300
|
1052 p = mip->mi_word;
|
|
339
|
1053 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1054 {
|
|
|
1055 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1056 mb_ptr_adv(p);
|
|
|
1057 wlen = p - mip->mi_word;
|
|
|
1058 }
|
|
300
|
1059 }
|
|
|
1060 #endif
|
|
236
|
1061
|
|
339
|
1062 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1063 * prefix ID.
|
|
|
1064 * Repeat this if there are more flags/region alternatives until there
|
|
|
1065 * is a match. */
|
|
|
1066 res = SP_BAD;
|
|
|
1067 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1068 --len, ++arridx)
|
|
300
|
1069 {
|
|
|
1070 flags = idxs[arridx];
|
|
324
|
1071
|
|
339
|
1072 /* For the fold-case tree check that the case of the checked word
|
|
|
1073 * matches with what the word in the tree requires.
|
|
|
1074 * For keep-case tree the case is always right. For prefixes we
|
|
|
1075 * don't bother to check. */
|
|
|
1076 if (mode == FIND_FOLDWORD)
|
|
300
|
1077 {
|
|
|
1078 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1079 {
|
|
323
|
1080 /* mi_capflags was set for a different word length, need
|
|
|
1081 * to do it again. */
|
|
300
|
1082 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1083 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1084 }
|
|
|
1085
|
|
346
|
1086 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1087 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1088 continue;
|
|
300
|
1089 }
|
|
236
|
1090
|
|
339
|
1091 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1092 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1093 * mip->mi_prefarridx that find_prefix() filled. */
|
|
339
|
1094 if (mode == FIND_PREFIX)
|
|
300
|
1095 {
|
|
339
|
1096 /* The prefix ID is stored two bytes above the flags. */
|
|
366
|
1097 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1098 flags,
|
|
455
|
1099 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1100 FALSE);
|
|
366
|
1101 if (c == 0)
|
|
339
|
1102 continue;
|
|
366
|
1103
|
|
|
1104 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1105 if (c & WF_RAREPFX)
|
|
|
1106 flags |= WF_RARE;
|
|
339
|
1107 }
|
|
|
1108
|
|
|
1109 if (flags & WF_BANNED)
|
|
|
1110 res = SP_BANNED;
|
|
|
1111 else if (flags & WF_REGION)
|
|
|
1112 {
|
|
|
1113 /* Check region. */
|
|
390
|
1114 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1115 res = SP_OK;
|
|
339
|
1116 else
|
|
|
1117 res = SP_LOCAL;
|
|
300
|
1118 }
|
|
339
|
1119 else if (flags & WF_RARE)
|
|
|
1120 res = SP_RARE;
|
|
307
|
1121 else
|
|
339
|
1122 res = SP_OK;
|
|
|
1123
|
|
|
1124 /* Always use the longest match and the best result. */
|
|
|
1125 if (mip->mi_result > res)
|
|
|
1126 {
|
|
|
1127 mip->mi_result = res;
|
|
|
1128 mip->mi_end = mip->mi_word + wlen;
|
|
|
1129 }
|
|
351
|
1130 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1131 mip->mi_end = mip->mi_word + wlen;
|
|
|
1132
|
|
|
1133 if (res == SP_OK)
|
|
|
1134 break;
|
|
300
|
1135 }
|
|
|
1136
|
|
307
|
1137 if (res == SP_OK)
|
|
300
|
1138 break;
|
|
|
1139 }
|
|
236
|
1140 }
|
|
|
1141
|
|
323
|
1142 /*
|
|
390
|
1143 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1144 * ID in "flags" for the word "word".
|
|
366
|
1145 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1146 */
|
|
|
1147 static int
|
|
455
|
1148 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1149 int totprefcnt; /* nr of prefix IDs */
|
|
|
1150 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1151 int flags;
|
|
351
|
1152 char_u *word;
|
|
|
1153 slang_T *slang;
|
|
455
|
1154 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1155 {
|
|
|
1156 int prefcnt;
|
|
|
1157 int pidx;
|
|
|
1158 regprog_T *rp;
|
|
|
1159 regmatch_T regmatch;
|
|
390
|
1160 int prefid;
|
|
|
1161
|
|
|
1162 prefid = (unsigned)flags >> 24;
|
|
351
|
1163 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1164 {
|
|
|
1165 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1166
|
|
|
1167 /* Check the prefix ID. */
|
|
|
1168 if (prefid != (pidx & 0xff))
|
|
|
1169 continue;
|
|
|
1170
|
|
390
|
1171 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1172 * suffix. */
|
|
|
1173 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1174 continue;
|
|
|
1175
|
|
351
|
1176 /* Check the condition, if there is one. The condition index is
|
|
366
|
1177 * stored in the two bytes above the prefix ID byte. */
|
|
|
1178 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1179 if (rp != NULL)
|
|
|
1180 {
|
|
|
1181 regmatch.regprog = rp;
|
|
|
1182 regmatch.rm_ic = FALSE;
|
|
|
1183 if (!vim_regexec(®match, word, 0))
|
|
|
1184 continue;
|
|
|
1185 }
|
|
455
|
1186 else if (cond_req)
|
|
|
1187 continue;
|
|
|
1188
|
|
|
1189 /* It's a match! Return the WF_ flags. */
|
|
366
|
1190 return pidx;
|
|
|
1191 }
|
|
|
1192 return 0;
|
|
351
|
1193 }
|
|
|
1194
|
|
|
1195 /*
|
|
339
|
1196 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1197 * If it does, then check the following word.
|
|
|
1198 *
|
|
|
1199 * For a match mip->mi_result is updated.
|
|
|
1200 */
|
|
|
1201 static void
|
|
|
1202 find_prefix(mip)
|
|
|
1203 matchinf_T *mip;
|
|
|
1204 {
|
|
|
1205 idx_T arridx = 0;
|
|
|
1206 int len;
|
|
|
1207 int wlen = 0;
|
|
|
1208 int flen;
|
|
|
1209 int c;
|
|
|
1210 char_u *ptr;
|
|
|
1211 idx_T lo, hi, m;
|
|
|
1212 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1213 char_u *byts;
|
|
|
1214 idx_T *idxs;
|
|
|
1215
|
|
375
|
1216 byts = slang->sl_pbyts;
|
|
|
1217 if (byts == NULL)
|
|
|
1218 return; /* array is empty */
|
|
|
1219
|
|
339
|
1220 /* We use the case-folded word here, since prefixes are always
|
|
|
1221 * case-folded. */
|
|
|
1222 ptr = mip->mi_fword;
|
|
|
1223 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1224 idxs = slang->sl_pidxs;
|
|
|
1225
|
|
|
1226 /*
|
|
|
1227 * Repeat advancing in the tree until:
|
|
|
1228 * - there is a byte that doesn't match,
|
|
|
1229 * - we reach the end of the tree,
|
|
|
1230 * - or we reach the end of the line.
|
|
|
1231 */
|
|
|
1232 for (;;)
|
|
|
1233 {
|
|
|
1234 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1235 flen = fold_more(mip);
|
|
|
1236
|
|
|
1237 len = byts[arridx++];
|
|
|
1238
|
|
|
1239 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1240 * Check if the following word matches and supports the prefix. */
|
|
|
1241 if (byts[arridx] == 0)
|
|
|
1242 {
|
|
|
1243 /* There can be several prefixes with different conditions. We
|
|
|
1244 * try them all, since we don't know which one will give the
|
|
|
1245 * longest match. The word is the same each time, pass the list
|
|
|
1246 * of possible prefixes to find_word(). */
|
|
|
1247 mip->mi_prefarridx = arridx;
|
|
|
1248 mip->mi_prefcnt = len;
|
|
|
1249 while (len > 0 && byts[arridx] == 0)
|
|
|
1250 {
|
|
|
1251 ++arridx;
|
|
|
1252 --len;
|
|
|
1253 }
|
|
|
1254 mip->mi_prefcnt -= len;
|
|
|
1255
|
|
|
1256 /* Find the word that comes after the prefix. */
|
|
|
1257 mip->mi_prefixlen = wlen;
|
|
455
|
1258 #ifdef FEAT_MBYTE
|
|
|
1259 if (has_mbyte)
|
|
|
1260 {
|
|
|
1261 /* Case-folded length may differ from original length. */
|
|
|
1262 mip->mi_cprefixlen = nofold_len(mip->mi_fword, wlen,
|
|
|
1263 mip->mi_word);
|
|
|
1264 }
|
|
|
1265 else
|
|
|
1266 mip->mi_cprefixlen = wlen;
|
|
|
1267 #endif
|
|
339
|
1268 find_word(mip, FIND_PREFIX);
|
|
|
1269
|
|
|
1270
|
|
|
1271 if (len == 0)
|
|
|
1272 break; /* no children, word must end here */
|
|
|
1273 }
|
|
|
1274
|
|
|
1275 /* Stop looking at end of the line. */
|
|
|
1276 if (ptr[wlen] == NUL)
|
|
|
1277 break;
|
|
|
1278
|
|
|
1279 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1280 c = ptr[wlen];
|
|
|
1281 lo = arridx;
|
|
|
1282 hi = arridx + len - 1;
|
|
|
1283 while (lo < hi)
|
|
|
1284 {
|
|
|
1285 m = (lo + hi) / 2;
|
|
|
1286 if (byts[m] > c)
|
|
|
1287 hi = m - 1;
|
|
|
1288 else if (byts[m] < c)
|
|
|
1289 lo = m + 1;
|
|
|
1290 else
|
|
|
1291 {
|
|
|
1292 lo = hi = m;
|
|
|
1293 break;
|
|
|
1294 }
|
|
|
1295 }
|
|
|
1296
|
|
|
1297 /* Stop if there is no matching byte. */
|
|
|
1298 if (hi < lo || byts[lo] != c)
|
|
|
1299 break;
|
|
|
1300
|
|
|
1301 /* Continue at the child (if there is one). */
|
|
|
1302 arridx = idxs[lo];
|
|
|
1303 ++wlen;
|
|
|
1304 --flen;
|
|
|
1305 }
|
|
|
1306 }
|
|
|
1307
|
|
|
1308 /*
|
|
|
1309 * Need to fold at least one more character. Do until next non-word character
|
|
|
1310 * for efficiency.
|
|
|
1311 * Return the length of the folded chars in bytes.
|
|
|
1312 */
|
|
|
1313 static int
|
|
|
1314 fold_more(mip)
|
|
|
1315 matchinf_T *mip;
|
|
|
1316 {
|
|
|
1317 int flen;
|
|
|
1318 char_u *p;
|
|
|
1319
|
|
|
1320 p = mip->mi_fend;
|
|
|
1321 do
|
|
|
1322 {
|
|
|
1323 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1324 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1325
|
|
|
1326 /* Include the non-word character so that we can check for the
|
|
|
1327 * word end. */
|
|
|
1328 if (*mip->mi_fend != NUL)
|
|
|
1329 mb_ptr_adv(mip->mi_fend);
|
|
|
1330
|
|
|
1331 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1332 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1333 MAXWLEN - mip->mi_fwordlen);
|
|
|
1334 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1335 mip->mi_fwordlen += flen;
|
|
|
1336 return flen;
|
|
|
1337 }
|
|
|
1338
|
|
|
1339 /*
|
|
323
|
1340 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1341 * case.
|
|
|
1342 */
|
|
|
1343 static int
|
|
389
|
1344 spell_valid_case(wordflags, treeflags)
|
|
|
1345 int wordflags; /* flags for the checked word. */
|
|
323
|
1346 int treeflags; /* flags for the word in the spell tree */
|
|
|
1347 {
|
|
389
|
1348 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1349 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
389
|
1350 && ((treeflags & WF_ONECAP) == 0 || wordflags == WF_ONECAP)));
|
|
323
|
1351 }
|
|
|
1352
|
|
351
|
1353 /*
|
|
|
1354 * Return TRUE if spell checking is not enabled.
|
|
|
1355 */
|
|
|
1356 static int
|
|
|
1357 no_spell_checking()
|
|
|
1358 {
|
|
|
1359 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1360 {
|
|
|
1361 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1362 return TRUE;
|
|
|
1363 }
|
|
|
1364 return FALSE;
|
|
|
1365 }
|
|
300
|
1366
|
|
236
|
1367 /*
|
|
|
1368 * Move to next spell error.
|
|
323
|
1369 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
236
|
1370 * Return OK if found, FAIL otherwise.
|
|
|
1371 */
|
|
|
1372 int
|
|
323
|
1373 spell_move_to(dir, allwords, curline)
|
|
236
|
1374 int dir; /* FORWARD or BACKWARD */
|
|
|
1375 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1376 int curline;
|
|
236
|
1377 {
|
|
249
|
1378 linenr_T lnum;
|
|
|
1379 pos_T found_pos;
|
|
236
|
1380 char_u *line;
|
|
|
1381 char_u *p;
|
|
346
|
1382 char_u *endp;
|
|
|
1383 int attr;
|
|
236
|
1384 int len;
|
|
249
|
1385 int has_syntax = syntax_present(curbuf);
|
|
|
1386 int col;
|
|
|
1387 int can_spell;
|
|
346
|
1388 char_u *buf = NULL;
|
|
|
1389 int buflen = 0;
|
|
|
1390 int skip = 0;
|
|
385
|
1391 int capcol = -1;
|
|
236
|
1392
|
|
351
|
1393 if (no_spell_checking())
|
|
236
|
1394 return FAIL;
|
|
|
1395
|
|
249
|
1396 /*
|
|
|
1397 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1398 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1399 *
|
|
|
1400 * When searching backwards, we continue in the line to find the last
|
|
|
1401 * bad word (in the cursor line: before the cursor).
|
|
346
|
1402 *
|
|
|
1403 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1404 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1405 * though...
|
|
249
|
1406 */
|
|
|
1407 lnum = curwin->w_cursor.lnum;
|
|
|
1408 found_pos.lnum = 0;
|
|
236
|
1409
|
|
|
1410 while (!got_int)
|
|
|
1411 {
|
|
249
|
1412 line = ml_get(lnum);
|
|
346
|
1413
|
|
|
1414 len = STRLEN(line);
|
|
|
1415 if (buflen < len + MAXWLEN + 2)
|
|
|
1416 {
|
|
|
1417 vim_free(buf);
|
|
|
1418 buflen = len + MAXWLEN + 2;
|
|
|
1419 buf = alloc(buflen);
|
|
|
1420 if (buf == NULL)
|
|
|
1421 break;
|
|
|
1422 }
|
|
|
1423
|
|
385
|
1424 /* In first line check first word for Capital. */
|
|
|
1425 if (lnum == 1)
|
|
|
1426 capcol = 0;
|
|
|
1427
|
|
|
1428 /* For checking first word with a capital skip white space. */
|
|
|
1429 if (capcol == 0)
|
|
|
1430 capcol = skipwhite(line) - line;
|
|
|
1431
|
|
346
|
1432 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1433 * possible. */
|
|
|
1434 STRCPY(buf, line);
|
|
|
1435 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1436 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1437
|
|
|
1438 p = buf + skip;
|
|
|
1439 endp = buf + len;
|
|
|
1440 while (p < endp)
|
|
236
|
1441 {
|
|
300
|
1442 /* When searching backward don't search after the cursor. */
|
|
|
1443 if (dir == BACKWARD
|
|
|
1444 && lnum == curwin->w_cursor.lnum
|
|
346
|
1445 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1446 break;
|
|
249
|
1447
|
|
300
|
1448 /* start of word */
|
|
346
|
1449 attr = 0;
|
|
385
|
1450 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1451
|
|
300
|
1452 if (attr != 0)
|
|
|
1453 {
|
|
|
1454 /* We found a bad word. Check the attribute. */
|
|
|
1455 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1456 {
|
|
300
|
1457 /* When searching forward only accept a bad word after
|
|
|
1458 * the cursor. */
|
|
|
1459 if (dir == BACKWARD
|
|
|
1460 || lnum > curwin->w_cursor.lnum
|
|
|
1461 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1462 && (colnr_T)(curline ? p - buf + len
|
|
|
1463 : p - buf)
|
|
300
|
1464 > curwin->w_cursor.col))
|
|
236
|
1465 {
|
|
300
|
1466 if (has_syntax)
|
|
249
|
1467 {
|
|
346
|
1468 col = p - buf;
|
|
300
|
1469 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1470 FALSE, &can_spell);
|
|
|
1471 }
|
|
|
1472 else
|
|
|
1473 can_spell = TRUE;
|
|
249
|
1474
|
|
300
|
1475 if (can_spell)
|
|
|
1476 {
|
|
|
1477 found_pos.lnum = lnum;
|
|
346
|
1478 found_pos.col = p - buf;
|
|
249
|
1479 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1480 found_pos.coladd = 0;
|
|
249
|
1481 #endif
|
|
300
|
1482 if (dir == FORWARD)
|
|
|
1483 {
|
|
|
1484 /* No need to search further. */
|
|
|
1485 curwin->w_cursor = found_pos;
|
|
346
|
1486 vim_free(buf);
|
|
300
|
1487 return OK;
|
|
249
|
1488 }
|
|
|
1489 }
|
|
236
|
1490 }
|
|
|
1491 }
|
|
|
1492 }
|
|
|
1493
|
|
300
|
1494 /* advance to character after the word */
|
|
|
1495 p += len;
|
|
385
|
1496 capcol -= len;
|
|
236
|
1497 }
|
|
|
1498
|
|
323
|
1499 if (curline)
|
|
346
|
1500 break; /* only check cursor line */
|
|
323
|
1501
|
|
236
|
1502 /* Advance to next line. */
|
|
249
|
1503 if (dir == BACKWARD)
|
|
|
1504 {
|
|
|
1505 if (found_pos.lnum != 0)
|
|
|
1506 {
|
|
|
1507 /* Use the last match in the line. */
|
|
|
1508 curwin->w_cursor = found_pos;
|
|
346
|
1509 vim_free(buf);
|
|
249
|
1510 return OK;
|
|
|
1511 }
|
|
|
1512 if (lnum == 1)
|
|
346
|
1513 break;
|
|
249
|
1514 --lnum;
|
|
385
|
1515 capcol = -1;
|
|
249
|
1516 }
|
|
|
1517 else
|
|
|
1518 {
|
|
|
1519 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1520 break;
|
|
249
|
1521 ++lnum;
|
|
346
|
1522
|
|
|
1523 /* Skip the characters at the start of the next line that were
|
|
|
1524 * included in a match crossing line boundaries. */
|
|
|
1525 if (attr == 0)
|
|
|
1526 skip = p - endp;
|
|
|
1527 else
|
|
|
1528 skip = 0;
|
|
385
|
1529
|
|
|
1530 /* Capscol skips over the inserted space. */
|
|
|
1531 --capcol;
|
|
|
1532
|
|
|
1533 /* But after empty line check first word in next line */
|
|
|
1534 if (*skipwhite(line) == NUL)
|
|
|
1535 capcol = 0;
|
|
249
|
1536 }
|
|
236
|
1537
|
|
|
1538 line_breakcheck();
|
|
|
1539 }
|
|
|
1540
|
|
346
|
1541 vim_free(buf);
|
|
|
1542 return FAIL;
|
|
|
1543 }
|
|
|
1544
|
|
|
1545 /*
|
|
|
1546 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1547 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1548 */
|
|
|
1549 void
|
|
|
1550 spell_cat_line(buf, line, maxlen)
|
|
|
1551 char_u *buf;
|
|
|
1552 char_u *line;
|
|
|
1553 int maxlen;
|
|
|
1554 {
|
|
|
1555 char_u *p;
|
|
|
1556 int n;
|
|
|
1557
|
|
|
1558 p = skipwhite(line);
|
|
|
1559 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1560 p = skipwhite(p + 1);
|
|
|
1561
|
|
|
1562 if (*p != NUL)
|
|
|
1563 {
|
|
|
1564 *buf = ' ';
|
|
376
|
1565 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1566 n = p - line;
|
|
|
1567 if (n >= maxlen)
|
|
|
1568 n = maxlen - 1;
|
|
|
1569 vim_memset(buf + 1, ' ', n);
|
|
|
1570 }
|
|
236
|
1571 }
|
|
|
1572
|
|
|
1573 /*
|
|
307
|
1574 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1575 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1576 */
|
|
307
|
1577 static void
|
|
236
|
1578 spell_load_lang(lang)
|
|
|
1579 char_u *lang;
|
|
|
1580 {
|
|
310
|
1581 char_u fname_enc[85];
|
|
236
|
1582 int r;
|
|
307
|
1583 char_u langcp[MAXWLEN + 1];
|
|
|
1584
|
|
310
|
1585 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1586 * It's truncated when an error is detected. */
|
|
|
1587 STRCPY(langcp, lang);
|
|
|
1588
|
|
310
|
1589 /*
|
|
|
1590 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1591 */
|
|
|
1592 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1593 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1594 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1595
|
|
|
1596 if (r == FAIL && *langcp != NUL)
|
|
|
1597 {
|
|
|
1598 /* Try loading the ASCII version. */
|
|
310
|
1599 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1600 "spell/%s.ascii.spl", lang);
|
|
310
|
1601 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1602 }
|
|
|
1603
|
|
|
1604 if (r == FAIL)
|
|
|
1605 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
|
236
|
1606 fname_enc + 6);
|
|
310
|
1607 else if (*langcp != NUL)
|
|
|
1608 {
|
|
|
1609 /* Load all the additions. */
|
|
|
1610 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1611 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1612 }
|
|
|
1613 }
|
|
|
1614
|
|
|
1615 /*
|
|
|
1616 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1617 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1618 */
|
|
|
1619 static char_u *
|
|
|
1620 spell_enc()
|
|
|
1621 {
|
|
|
1622
|
|
|
1623 #ifdef FEAT_MBYTE
|
|
|
1624 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1625 return p_enc;
|
|
|
1626 #endif
|
|
|
1627 return (char_u *)"latin1";
|
|
236
|
1628 }
|
|
|
1629
|
|
|
1630 /*
|
|
385
|
1631 * Get the name of the .spl file for the internal wordlist into
|
|
|
1632 * "fname[MAXPATHL]".
|
|
|
1633 */
|
|
|
1634 static void
|
|
|
1635 int_wordlist_spl(fname)
|
|
|
1636 char_u *fname;
|
|
|
1637 {
|
|
|
1638 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
1639 int_wordlist, spell_enc());
|
|
|
1640 }
|
|
|
1641
|
|
|
1642 /*
|
|
236
|
1643 * Allocate a new slang_T.
|
|
|
1644 * Caller must fill "sl_next".
|
|
|
1645 */
|
|
|
1646 static slang_T *
|
|
|
1647 slang_alloc(lang)
|
|
|
1648 char_u *lang;
|
|
|
1649 {
|
|
|
1650 slang_T *lp;
|
|
|
1651
|
|
300
|
1652 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1653 if (lp != NULL)
|
|
|
1654 {
|
|
|
1655 lp->sl_name = vim_strsave(lang);
|
|
323
|
1656 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
236
|
1657 }
|
|
|
1658 return lp;
|
|
|
1659 }
|
|
|
1660
|
|
|
1661 /*
|
|
|
1662 * Free the contents of an slang_T and the structure itself.
|
|
|
1663 */
|
|
|
1664 static void
|
|
|
1665 slang_free(lp)
|
|
|
1666 slang_T *lp;
|
|
|
1667 {
|
|
|
1668 vim_free(lp->sl_name);
|
|
310
|
1669 vim_free(lp->sl_fname);
|
|
|
1670 slang_clear(lp);
|
|
|
1671 vim_free(lp);
|
|
|
1672 }
|
|
|
1673
|
|
|
1674 /*
|
|
|
1675 * Clear an slang_T so that the file can be reloaded.
|
|
|
1676 */
|
|
|
1677 static void
|
|
|
1678 slang_clear(lp)
|
|
|
1679 slang_T *lp;
|
|
|
1680 {
|
|
339
|
1681 garray_T *gap;
|
|
|
1682 fromto_T *ftp;
|
|
344
|
1683 salitem_T *smp;
|
|
339
|
1684 int i;
|
|
323
|
1685
|
|
300
|
1686 vim_free(lp->sl_fbyts);
|
|
310
|
1687 lp->sl_fbyts = NULL;
|
|
300
|
1688 vim_free(lp->sl_kbyts);
|
|
310
|
1689 lp->sl_kbyts = NULL;
|
|
339
|
1690 vim_free(lp->sl_pbyts);
|
|
|
1691 lp->sl_pbyts = NULL;
|
|
|
1692
|
|
300
|
1693 vim_free(lp->sl_fidxs);
|
|
310
|
1694 lp->sl_fidxs = NULL;
|
|
300
|
1695 vim_free(lp->sl_kidxs);
|
|
310
|
1696 lp->sl_kidxs = NULL;
|
|
339
|
1697 vim_free(lp->sl_pidxs);
|
|
|
1698 lp->sl_pidxs = NULL;
|
|
323
|
1699
|
|
344
|
1700 gap = &lp->sl_rep;
|
|
|
1701 while (gap->ga_len > 0)
|
|
323
|
1702 {
|
|
344
|
1703 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1704 vim_free(ftp->ft_from);
|
|
|
1705 vim_free(ftp->ft_to);
|
|
323
|
1706 }
|
|
344
|
1707 ga_clear(gap);
|
|
|
1708
|
|
|
1709 gap = &lp->sl_sal;
|
|
375
|
1710 if (lp->sl_sofo)
|
|
376
|
1711 {
|
|
|
1712 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1713 if (gap->ga_data != NULL)
|
|
|
1714 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1715 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1716 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1717 }
|
|
375
|
1718 else
|
|
|
1719 /* SAL items: free salitem_T items */
|
|
|
1720 while (gap->ga_len > 0)
|
|
|
1721 {
|
|
|
1722 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1723 vim_free(smp->sm_lead);
|
|
|
1724 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1725 vim_free(smp->sm_to);
|
|
|
1726 #ifdef FEAT_MBYTE
|
|
|
1727 vim_free(smp->sm_lead_w);
|
|
|
1728 vim_free(smp->sm_oneof_w);
|
|
|
1729 vim_free(smp->sm_to_w);
|
|
|
1730 #endif
|
|
|
1731 }
|
|
344
|
1732 ga_clear(gap);
|
|
323
|
1733
|
|
339
|
1734 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1735 vim_free(lp->sl_prefprog[i]);
|
|
376
|
1736 lp->sl_prefixcnt = 0;
|
|
339
|
1737 vim_free(lp->sl_prefprog);
|
|
376
|
1738 lp->sl_prefprog = NULL;
|
|
|
1739
|
|
|
1740 vim_free(lp->sl_midword);
|
|
|
1741 lp->sl_midword = NULL;
|
|
339
|
1742
|
|
330
|
1743 #ifdef FEAT_MBYTE
|
|
|
1744 {
|
|
|
1745 int todo = lp->sl_map_hash.ht_used;
|
|
|
1746 hashitem_T *hi;
|
|
|
1747
|
|
|
1748 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1749 if (!HASHITEM_EMPTY(hi))
|
|
|
1750 {
|
|
|
1751 --todo;
|
|
|
1752 vim_free(hi->hi_key);
|
|
|
1753 }
|
|
|
1754 }
|
|
|
1755 hash_clear(&lp->sl_map_hash);
|
|
|
1756 #endif
|
|
236
|
1757 }
|
|
|
1758
|
|
|
1759 /*
|
|
307
|
1760 * Load one spell file and store the info into a slang_T.
|
|
236
|
1761 * Invoked through do_in_runtimepath().
|
|
|
1762 */
|
|
|
1763 static void
|
|
310
|
1764 spell_load_cb(fname, cookie)
|
|
236
|
1765 char_u *fname;
|
|
307
|
1766 void *cookie; /* points to the language name */
|
|
236
|
1767 {
|
|
323
|
1768 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1769 }
|
|
|
1770
|
|
|
1771 /*
|
|
|
1772 * Load one spell file and store the info into a slang_T.
|
|
|
1773 *
|
|
|
1774 * This is invoked in two ways:
|
|
|
1775 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
1776 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
1777 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
1778 * points to the existing slang_T.
|
|
323
|
1779 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
1780 */
|
|
323
|
1781 static slang_T *
|
|
|
1782 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
1783 char_u *fname;
|
|
|
1784 char_u *lang;
|
|
|
1785 slang_T *old_lp;
|
|
323
|
1786 int silent; /* no error if file doesn't exist */
|
|
310
|
1787 {
|
|
236
|
1788 FILE *fd;
|
|
|
1789 char_u buf[MAXWLEN + 1];
|
|
|
1790 char_u *p;
|
|
339
|
1791 char_u *bp;
|
|
|
1792 idx_T *ip;
|
|
236
|
1793 int i;
|
|
339
|
1794 int n;
|
|
300
|
1795 int len;
|
|
236
|
1796 int round;
|
|
|
1797 char_u *save_sourcing_name = sourcing_name;
|
|
|
1798 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
255
|
1799 int cnt, ccnt;
|
|
|
1800 char_u *fol;
|
|
307
|
1801 slang_T *lp = NULL;
|
|
323
|
1802 garray_T *gap;
|
|
|
1803 fromto_T *ftp;
|
|
344
|
1804 salitem_T *smp;
|
|
323
|
1805 short *first;
|
|
324
|
1806 idx_T idx;
|
|
344
|
1807 int c = 0;
|
|
236
|
1808
|
|
310
|
1809 fd = mch_fopen((char *)fname, "r");
|
|
236
|
1810 if (fd == NULL)
|
|
|
1811 {
|
|
323
|
1812 if (!silent)
|
|
|
1813 EMSG2(_(e_notopen), fname);
|
|
|
1814 else if (p_verbose > 2)
|
|
|
1815 {
|
|
|
1816 verbose_enter();
|
|
|
1817 smsg((char_u *)e_notopen, fname);
|
|
|
1818 verbose_leave();
|
|
|
1819 }
|
|
255
|
1820 goto endFAIL;
|
|
236
|
1821 }
|
|
310
|
1822 if (p_verbose > 2)
|
|
|
1823 {
|
|
|
1824 verbose_enter();
|
|
|
1825 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
1826 verbose_leave();
|
|
|
1827 }
|
|
|
1828
|
|
|
1829 if (old_lp == NULL)
|
|
|
1830 {
|
|
|
1831 lp = slang_alloc(lang);
|
|
|
1832 if (lp == NULL)
|
|
|
1833 goto endFAIL;
|
|
|
1834
|
|
|
1835 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
1836 lp->sl_fname = vim_strsave(fname);
|
|
|
1837 if (lp->sl_fname == NULL)
|
|
|
1838 goto endFAIL;
|
|
|
1839
|
|
|
1840 /* Check for .add.spl. */
|
|
|
1841 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
1842 }
|
|
|
1843 else
|
|
|
1844 lp = old_lp;
|
|
307
|
1845
|
|
236
|
1846 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
1847 sourcing_name = fname;
|
|
|
1848 sourcing_lnum = 0;
|
|
|
1849
|
|
339
|
1850 /* <HEADER>: <fileID>
|
|
|
1851 * <regioncnt> <regionname> ...
|
|
|
1852 * <charflagslen> <charflags>
|
|
|
1853 * <fcharslen> <fchars>
|
|
366
|
1854 * <midwordlen> <midword>
|
|
339
|
1855 * <prefcondcnt> <prefcond> ...
|
|
|
1856 */
|
|
236
|
1857 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
1858 buf[i] = getc(fd); /* <fileID> */
|
|
|
1859 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
1860 {
|
|
|
1861 EMSG(_("E757: Wrong file ID in spell file"));
|
|
255
|
1862 goto endFAIL;
|
|
236
|
1863 }
|
|
|
1864
|
|
|
1865 cnt = getc(fd); /* <regioncnt> */
|
|
255
|
1866 if (cnt < 0)
|
|
236
|
1867 {
|
|
|
1868 truncerr:
|
|
381
|
1869 EMSG(_(e_spell_trunc));
|
|
255
|
1870 goto endFAIL;
|
|
236
|
1871 }
|
|
|
1872 if (cnt > 8)
|
|
|
1873 {
|
|
|
1874 formerr:
|
|
307
|
1875 EMSG(_(e_format));
|
|
255
|
1876 goto endFAIL;
|
|
236
|
1877 }
|
|
|
1878 for (i = 0; i < cnt; ++i)
|
|
|
1879 {
|
|
|
1880 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
|
1881 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
|
1882 }
|
|
|
1883 lp->sl_regions[cnt * 2] = NUL;
|
|
|
1884
|
|
381
|
1885 /* <charflagslen> <charflags> */
|
|
|
1886 p = read_cnt_string(fd, 1, &cnt);
|
|
389
|
1887 if (cnt < 0)
|
|
381
|
1888 goto endFAIL;
|
|
|
1889
|
|
|
1890 /* <fcharslen> <fchars> */
|
|
389
|
1891 fol = read_cnt_string(fd, 2, &ccnt);
|
|
|
1892 if (ccnt < 0)
|
|
381
|
1893 {
|
|
300
|
1894 vim_free(p);
|
|
381
|
1895 goto endFAIL;
|
|
|
1896 }
|
|
|
1897
|
|
|
1898 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
1899 if (p != NULL && fol != NULL)
|
|
389
|
1900 i = set_spell_charflags(p, cnt, fol);
|
|
381
|
1901
|
|
|
1902 vim_free(p);
|
|
|
1903 vim_free(fol);
|
|
|
1904
|
|
|
1905 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
1906 if ((p == NULL) != (fol == NULL))
|
|
|
1907 goto formerr;
|
|
255
|
1908
|
|
366
|
1909 /* <midwordlen> <midword> */
|
|
376
|
1910 lp->sl_midword = read_cnt_string(fd, 2, &cnt);
|
|
389
|
1911 if (cnt < 0)
|
|
376
|
1912 goto endFAIL;
|
|
366
|
1913
|
|
339
|
1914 /* <prefcondcnt> <prefcond> ... */
|
|
|
1915 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
1916 if (cnt > 0)
|
|
|
1917 {
|
|
|
1918 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
1919 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
1920 if (lp->sl_prefprog == NULL)
|
|
|
1921 goto endFAIL;
|
|
|
1922 lp->sl_prefixcnt = cnt;
|
|
|
1923
|
|
|
1924 for (i = 0; i < cnt; ++i)
|
|
|
1925 {
|
|
|
1926 /* <prefcond> : <condlen> <condstr> */
|
|
|
1927 n = getc(fd); /* <condlen> */
|
|
|
1928 if (n < 0)
|
|
|
1929 goto formerr;
|
|
|
1930 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
1931 * compile the regexp program used to check for the condition. */
|
|
|
1932 if (n > 0)
|
|
|
1933 {
|
|
344
|
1934 buf[0] = '^'; /* always match at one position only */
|
|
|
1935 p = buf + 1;
|
|
339
|
1936 while (n-- > 0)
|
|
|
1937 *p++ = getc(fd); /* <condstr> */
|
|
|
1938 *p = NUL;
|
|
|
1939 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
1940 }
|
|
|
1941 }
|
|
|
1942 }
|
|
|
1943
|
|
|
1944
|
|
323
|
1945 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
1946 * <salflags> <salcount> <sal> ...
|
|
|
1947 * <maplen> <mapstr> */
|
|
339
|
1948
|
|
344
|
1949 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
1950 if (cnt < 0)
|
|
|
1951 goto formerr;
|
|
|
1952
|
|
|
1953 gap = &lp->sl_rep;
|
|
|
1954 if (ga_grow(gap, cnt) == FAIL)
|
|
|
1955 goto endFAIL;
|
|
|
1956
|
|
|
1957 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
1958 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
1959 {
|
|
|
1960 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
381
|
1961 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
389
|
1962 if (i <= 0)
|
|
381
|
1963 goto endFAIL;
|
|
|
1964 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
389
|
1965 if (i <= 0)
|
|
381
|
1966 {
|
|
|
1967 vim_free(ftp->ft_from);
|
|
|
1968 goto endFAIL;
|
|
344
|
1969 }
|
|
|
1970 }
|
|
|
1971
|
|
|
1972 /* Fill the first-index table. */
|
|
|
1973 first = lp->sl_rep_first;
|
|
|
1974 for (i = 0; i < 256; ++i)
|
|
|
1975 first[i] = -1;
|
|
|
1976 for (i = 0; i < gap->ga_len; ++i)
|
|
323
|
1977 {
|
|
344
|
1978 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
1979 if (first[*ftp->ft_from] == -1)
|
|
|
1980 first[*ftp->ft_from] = i;
|
|
|
1981 }
|
|
|
1982
|
|
|
1983 i = getc(fd); /* <salflags> */
|
|
|
1984 if (i & SAL_F0LLOWUP)
|
|
|
1985 lp->sl_followup = TRUE;
|
|
|
1986 if (i & SAL_COLLAPSE)
|
|
|
1987 lp->sl_collapse = TRUE;
|
|
|
1988 if (i & SAL_REM_ACCENTS)
|
|
|
1989 lp->sl_rem_accents = TRUE;
|
|
375
|
1990 if (i & SAL_SOFO)
|
|
|
1991 lp->sl_sofo = TRUE;
|
|
389
|
1992 else
|
|
|
1993 lp->sl_sofo = FALSE;
|
|
344
|
1994
|
|
|
1995 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
1996 if (cnt < 0)
|
|
|
1997 goto formerr;
|
|
|
1998
|
|
375
|
1999 if (lp->sl_sofo)
|
|
|
2000 {
|
|
|
2001 /*
|
|
|
2002 * SOFOFROM and SOFOTO items come in one <salfrom> and <salto>
|
|
|
2003 */
|
|
|
2004 if (cnt != 1)
|
|
|
2005 goto formerr;
|
|
|
2006
|
|
381
|
2007 /* <salfromlen> <salfrom> */
|
|
|
2008 bp = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2009 if (cnt < 0)
|
|
375
|
2010 goto endFAIL;
|
|
381
|
2011
|
|
|
2012 /* <saltolen> <salto> */
|
|
|
2013 fol = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2014 if (cnt < 0)
|
|
375
|
2015 {
|
|
|
2016 vim_free(bp);
|
|
344
|
2017 goto endFAIL;
|
|
375
|
2018 }
|
|
381
|
2019
|
|
|
2020 /* Store the info in lp->sl_sal and/or lp->sl_sal_first. */
|
|
389
|
2021 if (bp != NULL && fol != NULL)
|
|
|
2022 i = set_sofo(lp, bp, fol);
|
|
|
2023 else if (bp != NULL || fol != NULL)
|
|
|
2024 i = FAIL; /* only one of two strings is an error */
|
|
|
2025 else
|
|
|
2026 i = OK;
|
|
381
|
2027
|
|
375
|
2028 vim_free(bp);
|
|
|
2029 vim_free(fol);
|
|
381
|
2030 if (i == FAIL)
|
|
|
2031 goto formerr;
|
|
375
|
2032 }
|
|
|
2033 else
|
|
|
2034 {
|
|
|
2035 /*
|
|
|
2036 * SAL items
|
|
|
2037 */
|
|
|
2038 gap = &lp->sl_sal;
|
|
390
|
2039 ga_init2(gap, sizeof(salitem_T), 10);
|
|
375
|
2040 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2041 goto endFAIL;
|
|
|
2042
|
|
|
2043 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2044 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2045 {
|
|
|
2046 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2047 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2048 if (ccnt < 0)
|
|
|
2049 goto formerr;
|
|
|
2050 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2051 goto endFAIL;
|
|
|
2052 smp->sm_lead = p;
|
|
|
2053
|
|
|
2054 /* Read up to the first special char into sm_lead. */
|
|
|
2055 for (i = 0; i < ccnt; ++i)
|
|
|
2056 {
|
|
|
2057 c = getc(fd); /* <salfrom> */
|
|
|
2058 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2059 break;
|
|
|
2060 *p++ = c;
|
|
|
2061 }
|
|
|
2062 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2063 *p++ = NUL;
|
|
|
2064
|
|
|
2065 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2066 if (c == '(')
|
|
|
2067 {
|
|
|
2068 smp->sm_oneof = p;
|
|
|
2069 for (++i; i < ccnt; ++i)
|
|
|
2070 {
|
|
|
2071 c = getc(fd); /* <salfrom> */
|
|
|
2072 if (c == ')')
|
|
|
2073 break;
|
|
|
2074 *p++ = c;
|
|
|
2075 }
|
|
|
2076 *p++ = NUL;
|
|
|
2077 if (++i < ccnt)
|
|
|
2078 c = getc(fd);
|
|
|
2079 }
|
|
372
|
2080 else
|
|
375
|
2081 smp->sm_oneof = NULL;
|
|
|
2082
|
|
|
2083 /* Any following chars go in sm_rules. */
|
|
|
2084 smp->sm_rules = p;
|
|
|
2085 if (i < ccnt)
|
|
|
2086 /* store the char we got while checking for end of sm_lead */
|
|
|
2087 *p++ = c;
|
|
|
2088 for (++i; i < ccnt; ++i)
|
|
|
2089 *p++ = getc(fd); /* <salfrom> */
|
|
|
2090 *p++ = NUL;
|
|
|
2091
|
|
381
|
2092 /* <saltolen> <salto> */
|
|
|
2093 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
389
|
2094 if (ccnt < 0)
|
|
372
|
2095 {
|
|
|
2096 vim_free(smp->sm_lead);
|
|
375
|
2097 goto formerr;
|
|
|
2098 }
|
|
|
2099
|
|
372
|
2100 #ifdef FEAT_MBYTE
|
|
|
2101 if (has_mbyte)
|
|
|
2102 {
|
|
375
|
2103 /* convert the multi-byte strings to wide char strings */
|
|
|
2104 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2105 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2106 if (smp->sm_oneof == NULL)
|
|
|
2107 smp->sm_oneof_w = NULL;
|
|
|
2108 else
|
|
|
2109 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
389
|
2110 if (smp->sm_to == NULL)
|
|
|
2111 smp->sm_to_w = NULL;
|
|
|
2112 else
|
|
|
2113 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
375
|
2114 if (smp->sm_lead_w == NULL
|
|
|
2115 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
389
|
2116 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
372
|
2117 {
|
|
375
|
2118 vim_free(smp->sm_lead);
|
|
|
2119 vim_free(smp->sm_to);
|
|
|
2120 vim_free(smp->sm_lead_w);
|
|
|
2121 vim_free(smp->sm_oneof_w);
|
|
|
2122 vim_free(smp->sm_to_w);
|
|
|
2123 goto endFAIL;
|
|
372
|
2124 }
|
|
|
2125 }
|
|
|
2126 #endif
|
|
|
2127 }
|
|
375
|
2128
|
|
|
2129 /* Fill the first-index table. */
|
|
381
|
2130 set_sal_first(lp);
|
|
|
2131 }
|
|
|
2132
|
|
|
2133 /* <maplen> <mapstr> */
|
|
|
2134 p = read_cnt_string(fd, 2, &cnt);
|
|
389
|
2135 if (cnt < 0)
|
|
323
|
2136 goto endFAIL;
|
|
389
|
2137 if (p != NULL)
|
|
|
2138 {
|
|
|
2139 set_map_str(lp, p);
|
|
|
2140 vim_free(p);
|
|
|
2141 }
|
|
236
|
2142
|
|
300
|
2143 /* round 1: <LWORDTREE>
|
|
339
|
2144 * round 2: <KWORDTREE>
|
|
|
2145 * round 3: <PREFIXTREE> */
|
|
|
2146 for (round = 1; round <= 3; ++round)
|
|
236
|
2147 {
|
|
300
|
2148 /* The tree size was computed when writing the file, so that we can
|
|
|
2149 * allocate it as one long block. <nodecount> */
|
|
|
2150 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2151 if (len < 0)
|
|
|
2152 goto truncerr;
|
|
|
2153 if (len > 0)
|
|
236
|
2154 {
|
|
300
|
2155 /* Allocate the byte array. */
|
|
339
|
2156 bp = lalloc((long_u)len, TRUE);
|
|
|
2157 if (bp == NULL)
|
|
300
|
2158 goto endFAIL;
|
|
|
2159 if (round == 1)
|
|
339
|
2160 lp->sl_fbyts = bp;
|
|
|
2161 else if (round == 2)
|
|
|
2162 lp->sl_kbyts = bp;
|
|
300
|
2163 else
|
|
339
|
2164 lp->sl_pbyts = bp;
|
|
236
|
2165
|
|
300
|
2166 /* Allocate the index array. */
|
|
339
|
2167 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2168 if (ip == NULL)
|
|
300
|
2169 goto endFAIL;
|
|
|
2170 if (round == 1)
|
|
339
|
2171 lp->sl_fidxs = ip;
|
|
|
2172 else if (round == 2)
|
|
|
2173 lp->sl_kidxs = ip;
|
|
300
|
2174 else
|
|
339
|
2175 lp->sl_pidxs = ip;
|
|
300
|
2176
|
|
|
2177 /* Read the tree and store it in the array. */
|
|
339
|
2178 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2179 if (idx == -1)
|
|
300
|
2180 goto truncerr;
|
|
324
|
2181 if (idx < 0)
|
|
236
|
2182 goto formerr;
|
|
|
2183 }
|
|
300
|
2184 }
|
|
243
|
2185
|
|
310
|
2186 /* For a new file link it in the list of spell files. */
|
|
|
2187 if (old_lp == NULL)
|
|
|
2188 {
|
|
|
2189 lp->sl_next = first_lang;
|
|
|
2190 first_lang = lp;
|
|
|
2191 }
|
|
307
|
2192
|
|
255
|
2193 goto endOK;
|
|
|
2194
|
|
|
2195 endFAIL:
|
|
310
|
2196 if (lang != NULL)
|
|
|
2197 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2198 *lang = NUL;
|
|
|
2199 if (lp != NULL && old_lp == NULL)
|
|
323
|
2200 {
|
|
307
|
2201 slang_free(lp);
|
|
323
|
2202 lp = NULL;
|
|
|
2203 }
|
|
255
|
2204
|
|
|
2205 endOK:
|
|
236
|
2206 if (fd != NULL)
|
|
|
2207 fclose(fd);
|
|
|
2208 sourcing_name = save_sourcing_name;
|
|
|
2209 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2210
|
|
|
2211 return lp;
|
|
236
|
2212 }
|
|
|
2213
|
|
376
|
2214 /*
|
|
|
2215 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2216 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2217 * Returns NULL when the count is zero.
|
|
389
|
2218 * Sets "*cntp" to -1 when there is an error, length of the result otherwise.
|
|
376
|
2219 */
|
|
|
2220 static char_u *
|
|
389
|
2221 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2222 FILE *fd;
|
|
|
2223 int cnt_bytes;
|
|
389
|
2224 int *cntp;
|
|
376
|
2225 {
|
|
|
2226 int cnt = 0;
|
|
|
2227 int i;
|
|
|
2228 char_u *str;
|
|
|
2229
|
|
|
2230 /* read the length bytes, MSB first */
|
|
|
2231 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2232 cnt = (cnt << 8) + getc(fd);
|
|
|
2233 if (cnt < 0)
|
|
|
2234 {
|
|
381
|
2235 EMSG(_(e_spell_trunc));
|
|
389
|
2236 *cntp = -1;
|
|
376
|
2237 return NULL;
|
|
|
2238 }
|
|
389
|
2239 *cntp = cnt;
|
|
|
2240 if (cnt == 0)
|
|
|
2241 return NULL; /* nothing to read, return NULL */
|
|
376
|
2242
|
|
|
2243 /* allocate memory */
|
|
|
2244 str = alloc((unsigned)cnt + 1);
|
|
|
2245 if (str == NULL)
|
|
|
2246 {
|
|
389
|
2247 *cntp = -1;
|
|
376
|
2248 return NULL;
|
|
|
2249 }
|
|
|
2250
|
|
|
2251 /* Read the string. Doesn't check for truncated file. */
|
|
|
2252 for (i = 0; i < cnt; ++i)
|
|
|
2253 str[i] = getc(fd);
|
|
|
2254 str[i] = NUL;
|
|
|
2255
|
|
|
2256 return str;
|
|
|
2257 }
|
|
|
2258
|
|
381
|
2259 /*
|
|
|
2260 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
|
2261 * Returns FAIL when there is something wrong.
|
|
|
2262 */
|
|
|
2263 static int
|
|
|
2264 set_sofo(lp, from, to)
|
|
|
2265 slang_T *lp;
|
|
|
2266 char_u *from;
|
|
|
2267 char_u *to;
|
|
|
2268 {
|
|
|
2269 int i;
|
|
|
2270
|
|
|
2271 #ifdef FEAT_MBYTE
|
|
|
2272 garray_T *gap;
|
|
|
2273 char_u *s;
|
|
|
2274 char_u *p;
|
|
|
2275 int c;
|
|
|
2276 int *inp;
|
|
|
2277
|
|
|
2278 if (has_mbyte)
|
|
|
2279 {
|
|
|
2280 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2281 * characters. The index is the low byte of the character.
|
|
|
2282 * The list contains from-to pairs with a terminating NUL.
|
|
|
2283 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2284 gap = &lp->sl_sal;
|
|
|
2285 ga_init2(gap, sizeof(int *), 1);
|
|
|
2286 if (ga_grow(gap, 256) == FAIL)
|
|
|
2287 return FAIL;
|
|
|
2288 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2289 gap->ga_len = 256;
|
|
|
2290
|
|
|
2291 /* First count the number of items for each list. Temporarily use
|
|
|
2292 * sl_sal_first[] for this. */
|
|
|
2293 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2294 {
|
|
|
2295 c = mb_ptr2char_adv(&p);
|
|
|
2296 mb_ptr_adv(s);
|
|
|
2297 if (c >= 256)
|
|
|
2298 ++lp->sl_sal_first[c & 0xff];
|
|
|
2299 }
|
|
|
2300 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
|
2301 return FAIL;
|
|
|
2302
|
|
|
2303 /* Allocate the lists. */
|
|
|
2304 for (i = 0; i < 256; ++i)
|
|
|
2305 if (lp->sl_sal_first[i] > 0)
|
|
|
2306 {
|
|
|
2307 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2308 if (p == NULL)
|
|
|
2309 return FAIL;
|
|
|
2310 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2311 *(int *)p = 0;
|
|
|
2312 }
|
|
|
2313
|
|
|
2314 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
2315 * list. */
|
|
|
2316 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2317 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2318 {
|
|
|
2319 c = mb_ptr2char_adv(&p);
|
|
|
2320 i = mb_ptr2char_adv(&s);
|
|
|
2321 if (c >= 256)
|
|
|
2322 {
|
|
|
2323 /* Append the from-to chars at the end of the list with
|
|
|
2324 * the low byte. */
|
|
|
2325 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2326 while (*inp != 0)
|
|
|
2327 ++inp;
|
|
|
2328 *inp++ = c; /* from char */
|
|
|
2329 *inp++ = i; /* to char */
|
|
|
2330 *inp++ = NUL; /* NUL at the end */
|
|
|
2331 }
|
|
|
2332 else
|
|
|
2333 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2334 lp->sl_sal_first[c] = i;
|
|
|
2335 }
|
|
|
2336 }
|
|
|
2337 else
|
|
|
2338 #endif
|
|
|
2339 {
|
|
|
2340 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2341 if (STRLEN(from) != STRLEN(to))
|
|
|
2342 return FAIL;
|
|
|
2343
|
|
|
2344 for (i = 0; to[i] != NUL; ++i)
|
|
|
2345 lp->sl_sal_first[from[i]] = to[i];
|
|
|
2346 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2347 }
|
|
|
2348
|
|
|
2349 return OK;
|
|
|
2350 }
|
|
|
2351
|
|
|
2352 /*
|
|
|
2353 * Fill the first-index table for "lp".
|
|
|
2354 */
|
|
|
2355 static void
|
|
|
2356 set_sal_first(lp)
|
|
|
2357 slang_T *lp;
|
|
|
2358 {
|
|
|
2359 salfirst_T *sfirst;
|
|
|
2360 int i;
|
|
|
2361 salitem_T *smp;
|
|
|
2362 int c;
|
|
|
2363 garray_T *gap = &lp->sl_sal;
|
|
|
2364
|
|
|
2365 sfirst = lp->sl_sal_first;
|
|
|
2366 for (i = 0; i < 256; ++i)
|
|
|
2367 sfirst[i] = -1;
|
|
|
2368 smp = (salitem_T *)gap->ga_data;
|
|
|
2369 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2370 {
|
|
|
2371 #ifdef FEAT_MBYTE
|
|
|
2372 if (has_mbyte)
|
|
|
2373 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2374 * the character, for other encodings it should differ for most
|
|
|
2375 * characters. */
|
|
|
2376 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2377 else
|
|
|
2378 #endif
|
|
|
2379 c = *smp[i].sm_lead;
|
|
|
2380 if (sfirst[c] == -1)
|
|
|
2381 {
|
|
|
2382 sfirst[c] = i;
|
|
|
2383 #ifdef FEAT_MBYTE
|
|
|
2384 if (has_mbyte)
|
|
|
2385 {
|
|
|
2386 int n;
|
|
|
2387
|
|
|
2388 /* Make sure all entries with this byte are following each
|
|
|
2389 * other. Move the ones that are in the wrong position. Do
|
|
|
2390 * keep the same ordering! */
|
|
|
2391 while (i + 1 < gap->ga_len
|
|
|
2392 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
2393 /* Skip over entry with same index byte. */
|
|
|
2394 ++i;
|
|
|
2395
|
|
|
2396 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
2397 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
2398 {
|
|
|
2399 salitem_T tsal;
|
|
|
2400
|
|
|
2401 /* Move entry with same index byte after the entries
|
|
|
2402 * we already found. */
|
|
|
2403 ++i;
|
|
|
2404 --n;
|
|
|
2405 tsal = smp[i + n];
|
|
|
2406 mch_memmove(smp + i + 1, smp + i,
|
|
|
2407 sizeof(salitem_T) * n);
|
|
|
2408 smp[i] = tsal;
|
|
|
2409 }
|
|
|
2410 }
|
|
|
2411 #endif
|
|
|
2412 }
|
|
|
2413 }
|
|
|
2414 }
|
|
376
|
2415
|
|
372
|
2416 #ifdef FEAT_MBYTE
|
|
|
2417 /*
|
|
|
2418 * Turn a multi-byte string into a wide character string.
|
|
|
2419 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
2420 */
|
|
|
2421 static int *
|
|
|
2422 mb_str2wide(s)
|
|
|
2423 char_u *s;
|
|
|
2424 {
|
|
|
2425 int *res;
|
|
|
2426 char_u *p;
|
|
|
2427 int i = 0;
|
|
|
2428
|
|
|
2429 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
2430 if (res != NULL)
|
|
|
2431 {
|
|
|
2432 for (p = s; *p != NUL; )
|
|
|
2433 res[i++] = mb_ptr2char_adv(&p);
|
|
|
2434 res[i] = NUL;
|
|
|
2435 }
|
|
|
2436 return res;
|
|
|
2437 }
|
|
|
2438 #endif
|
|
|
2439
|
|
236
|
2440 /*
|
|
300
|
2441 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
2442 * "byts" and index array "idxs". Recursively read the children.
|
|
|
2443 *
|
|
346
|
2444 * NOTE: The code here must match put_node().
|
|
300
|
2445 *
|
|
|
2446 * Returns the index follosing the siblings.
|
|
|
2447 * Returns -1 if the file is shorter than expected.
|
|
|
2448 * Returns -2 if there is a format error.
|
|
236
|
2449 */
|
|
324
|
2450 static idx_T
|
|
339
|
2451 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
2452 FILE *fd;
|
|
|
2453 char_u *byts;
|
|
324
|
2454 idx_T *idxs;
|
|
300
|
2455 int maxidx; /* size of arrays */
|
|
324
|
2456 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
2457 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
2458 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
2459 {
|
|
300
|
2460 int len;
|
|
|
2461 int i;
|
|
|
2462 int n;
|
|
324
|
2463 idx_T idx = startidx;
|
|
300
|
2464 int c;
|
|
366
|
2465 int c2;
|
|
300
|
2466 #define SHARED_MASK 0x8000000
|
|
236
|
2467
|
|
300
|
2468 len = getc(fd); /* <siblingcount> */
|
|
|
2469 if (len <= 0)
|
|
|
2470 return -1;
|
|
|
2471
|
|
|
2472 if (startidx + len >= maxidx)
|
|
|
2473 return -2;
|
|
|
2474 byts[idx++] = len;
|
|
|
2475
|
|
|
2476 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
2477 for (i = 1; i <= len; ++i)
|
|
236
|
2478 {
|
|
300
|
2479 c = getc(fd); /* <byte> */
|
|
|
2480 if (c < 0)
|
|
|
2481 return -1;
|
|
|
2482 if (c <= BY_SPECIAL)
|
|
|
2483 {
|
|
366
|
2484 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
2485 {
|
|
|
2486 /* No flags, all regions. */
|
|
|
2487 idxs[idx] = 0;
|
|
|
2488 c = 0;
|
|
|
2489 }
|
|
390
|
2490 else if (c != BY_INDEX)
|
|
300
|
2491 {
|
|
339
|
2492 if (prefixtree)
|
|
|
2493 {
|
|
455
|
2494 /* Read the optional pflags byte, the prefix ID and the
|
|
|
2495 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
2496 * byte, the condition index shifted up 8 bits, the flags
|
|
|
2497 * shifted up 24 bits. */
|
|
|
2498 if (c == BY_FLAGS)
|
|
|
2499 c = getc(fd) << 24; /* <pflags> */
|
|
|
2500 else
|
|
|
2501 c = 0;
|
|
|
2502
|
|
|
2503 c |= getc(fd); /* <prefixID> */
|
|
|
2504
|
|
339
|
2505 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
2506 if (n >= maxprefcondnr)
|
|
|
2507 return -2;
|
|
455
|
2508 c |= (n << 8);
|
|
339
|
2509 }
|
|
390
|
2510 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
2511 {
|
|
|
2512 /* Read flags and optional region and prefix ID. In
|
|
390
|
2513 * idxs[] the flags go in the low two bytes, region above
|
|
|
2514 * that and prefix ID above the region. */
|
|
|
2515 c2 = c;
|
|
339
|
2516 c = getc(fd); /* <flags> */
|
|
390
|
2517 if (c2 == BY_FLAGS2)
|
|
|
2518 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
2519 if (c & WF_REGION)
|
|
390
|
2520 c = (getc(fd) << 16) + c; /* <region> */
|
|
339
|
2521 if (c & WF_PFX)
|
|
390
|
2522 c = (getc(fd) << 24) + c; /* <prefixID> */
|
|
339
|
2523 }
|
|
|
2524
|
|
300
|
2525 idxs[idx] = c;
|
|
|
2526 c = 0;
|
|
|
2527 }
|
|
|
2528 else /* c == BY_INDEX */
|
|
|
2529 {
|
|
|
2530 /* <nodeidx> */
|
|
|
2531 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2532 if (n < 0 || n >= maxidx)
|
|
|
2533 return -2;
|
|
|
2534 idxs[idx] = n + SHARED_MASK;
|
|
|
2535 c = getc(fd); /* <xbyte> */
|
|
|
2536 }
|
|
|
2537 }
|
|
|
2538 byts[idx++] = c;
|
|
236
|
2539 }
|
|
|
2540
|
|
300
|
2541 /* Recursively read the children for non-shared siblings.
|
|
|
2542 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
2543 * remove SHARED_MASK) */
|
|
|
2544 for (i = 1; i <= len; ++i)
|
|
|
2545 if (byts[startidx + i] != 0)
|
|
|
2546 {
|
|
|
2547 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
2548 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
2549 else
|
|
|
2550 {
|
|
|
2551 idxs[startidx + i] = idx;
|
|
339
|
2552 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
2553 prefixtree, maxprefcondnr);
|
|
300
|
2554 if (idx < 0)
|
|
|
2555 break;
|
|
|
2556 }
|
|
|
2557 }
|
|
236
|
2558
|
|
300
|
2559 return idx;
|
|
236
|
2560 }
|
|
|
2561
|
|
|
2562 /*
|
|
|
2563 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
2564 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
2565 */
|
|
|
2566 char_u *
|
|
|
2567 did_set_spelllang(buf)
|
|
|
2568 buf_T *buf;
|
|
|
2569 {
|
|
|
2570 garray_T ga;
|
|
351
|
2571 char_u *splp;
|
|
236
|
2572 char_u *region;
|
|
416
|
2573 char_u region_cp[3];
|
|
355
|
2574 int filename;
|
|
236
|
2575 int region_mask;
|
|
|
2576 slang_T *lp;
|
|
|
2577 int c;
|
|
351
|
2578 char_u lang[MAXWLEN + 1];
|
|
323
|
2579 char_u spf_name[MAXPATHL];
|
|
351
|
2580 int len;
|
|
|
2581 char_u *p;
|
|
381
|
2582 int round;
|
|
385
|
2583 char_u *spf;
|
|
389
|
2584 char_u *use_region = NULL;
|
|
|
2585 int dont_use_region = FALSE;
|
|
236
|
2586
|
|
|
2587 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
2588 clear_midword(buf);
|
|
236
|
2589
|
|
351
|
2590 /* loop over comma separated language names. */
|
|
|
2591 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
2592 {
|
|
|
2593 /* Get one language name. */
|
|
|
2594 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
2595
|
|
240
|
2596 region = NULL;
|
|
351
|
2597 len = STRLEN(lang);
|
|
355
|
2598
|
|
|
2599 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
2600 * If there is a region name let "region" point to it and remove it
|
|
|
2601 * from the name. */
|
|
|
2602 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
2603 {
|
|
355
|
2604 filename = TRUE;
|
|
|
2605
|
|
416
|
2606 /* Locate a region and remove it from the file name. */
|
|
|
2607 p = vim_strchr(gettail(lang), '_');
|
|
|
2608 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
2609 && !ASCII_ISALPHA(p[3]))
|
|
|
2610 {
|
|
|
2611 vim_strncpy(region_cp, p + 1, 2);
|
|
|
2612 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
2613 len -= 3;
|
|
|
2614 region = region_cp;
|
|
|
2615 }
|
|
|
2616 else
|
|
|
2617 dont_use_region = TRUE;
|
|
|
2618
|
|
355
|
2619 /* Check if we loaded this language before. */
|
|
|
2620 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2621 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2622 break;
|
|
236
|
2623 }
|
|
355
|
2624 else
|
|
|
2625 {
|
|
|
2626 filename = FALSE;
|
|
|
2627 if (len > 3 && lang[len - 3] == '_')
|
|
|
2628 {
|
|
|
2629 region = lang + len - 2;
|
|
|
2630 len -= 3;
|
|
|
2631 lang[len] = NUL;
|
|
389
|
2632 }
|
|
|
2633 else
|
|
|
2634 dont_use_region = TRUE;
|
|
355
|
2635
|
|
|
2636 /* Check if we loaded this language before. */
|
|
|
2637 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2638 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
2639 break;
|
|
|
2640 }
|
|
236
|
2641
|
|
416
|
2642 if (region != NULL)
|
|
|
2643 {
|
|
|
2644 /* If the region differs from what was used before then don't
|
|
|
2645 * use it for 'spellfile'. */
|
|
|
2646 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
2647 dont_use_region = TRUE;
|
|
|
2648 use_region = region;
|
|
|
2649 }
|
|
|
2650
|
|
351
|
2651 /* If not found try loading the language now. */
|
|
236
|
2652 if (lp == NULL)
|
|
355
|
2653 {
|
|
|
2654 if (filename)
|
|
|
2655 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
2656 else
|
|
|
2657 spell_load_lang(lang);
|
|
|
2658 }
|
|
236
|
2659
|
|
307
|
2660 /*
|
|
351
|
2661 * Loop over the languages, there can be several files for "lang".
|
|
307
|
2662 */
|
|
|
2663 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
2664 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
2665 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
2666 {
|
|
316
|
2667 region_mask = REGION_ALL;
|
|
355
|
2668 if (!filename && region != NULL)
|
|
236
|
2669 {
|
|
307
|
2670 /* find region in sl_regions */
|
|
|
2671 c = find_region(lp->sl_regions, region);
|
|
|
2672 if (c == REGION_ALL)
|
|
|
2673 {
|
|
389
|
2674 if (lp->sl_add)
|
|
|
2675 {
|
|
|
2676 if (*lp->sl_regions != NUL)
|
|
|
2677 /* This addition file is for other regions. */
|
|
|
2678 region_mask = 0;
|
|
|
2679 }
|
|
|
2680 else
|
|
|
2681 /* This is probably an error. Give a warning and
|
|
|
2682 * accept the words anyway. */
|
|
351
|
2683 smsg((char_u *)
|
|
|
2684 _("Warning: region %s not supported"),
|
|
|
2685 region);
|
|
307
|
2686 }
|
|
|
2687 else
|
|
|
2688 region_mask = 1 << c;
|
|
236
|
2689 }
|
|
307
|
2690
|
|
389
|
2691 if (region_mask != 0)
|
|
307
|
2692 {
|
|
389
|
2693 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2694 {
|
|
|
2695 ga_clear(&ga);
|
|
|
2696 return e_outofmem;
|
|
|
2697 }
|
|
|
2698 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2699 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2700 ++ga.ga_len;
|
|
|
2701 use_midword(lp, buf);
|
|
307
|
2702 }
|
|
385
|
2703 }
|
|
|
2704 }
|
|
|
2705
|
|
|
2706 /* round 0: load int_wordlist, if possible.
|
|
|
2707 * round 1: load first name in 'spellfile'.
|
|
|
2708 * round 2: load second name in 'spellfile.
|
|
|
2709 * etc. */
|
|
|
2710 spf = curbuf->b_p_spf;
|
|
|
2711 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
2712 {
|
|
|
2713 if (round == 0)
|
|
|
2714 {
|
|
|
2715 /* Internal wordlist, if there is one. */
|
|
|
2716 if (int_wordlist == NULL)
|
|
381
|
2717 continue;
|
|
385
|
2718 int_wordlist_spl(spf_name);
|
|
381
|
2719 }
|
|
|
2720 else
|
|
|
2721 {
|
|
385
|
2722 /* One entry in 'spellfile'. */
|
|
|
2723 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
2724 STRCAT(spf_name, ".spl");
|
|
|
2725
|
|
|
2726 /* If it was already found above then skip it. */
|
|
|
2727 for (c = 0; c < ga.ga_len; ++c)
|
|
|
2728 if (fullpathcmp(spf_name,
|
|
|
2729 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
2730 FALSE) == FPC_SAME)
|
|
|
2731 break;
|
|
|
2732 if (c < ga.ga_len)
|
|
381
|
2733 continue;
|
|
|
2734 }
|
|
|
2735
|
|
351
|
2736 /* Check if it was loaded already. */
|
|
323
|
2737 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
2738 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
2739 break;
|
|
|
2740 if (lp == NULL)
|
|
|
2741 {
|
|
351
|
2742 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
2743 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
2744 * use an arbitrary name. */
|
|
|
2745 if (round == 0)
|
|
|
2746 STRCPY(lang, "internal wordlist");
|
|
|
2747 else
|
|
|
2748 {
|
|
|
2749 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
2750 p = vim_strchr(lang, '.');
|
|
|
2751 if (p != NULL)
|
|
|
2752 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
2753 }
|
|
351
|
2754 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
2755 }
|
|
|
2756 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
2757 {
|
|
389
|
2758 region_mask = REGION_ALL;
|
|
|
2759 if (use_region != NULL && !dont_use_region)
|
|
|
2760 {
|
|
|
2761 /* find region in sl_regions */
|
|
|
2762 c = find_region(lp->sl_regions, use_region);
|
|
|
2763 if (c != REGION_ALL)
|
|
|
2764 region_mask = 1 << c;
|
|
|
2765 else if (*lp->sl_regions != NUL)
|
|
|
2766 /* This spell file is for other regions. */
|
|
|
2767 region_mask = 0;
|
|
|
2768 }
|
|
|
2769
|
|
|
2770 if (region_mask != 0)
|
|
|
2771 {
|
|
|
2772 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
2773 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
2774 ++ga.ga_len;
|
|
|
2775 use_midword(lp, buf);
|
|
|
2776 }
|
|
323
|
2777 }
|
|
|
2778 }
|
|
|
2779
|
|
236
|
2780 /* Add a NULL entry to mark the end of the list. */
|
|
|
2781 if (ga_grow(&ga, 1) == FAIL)
|
|
|
2782 {
|
|
|
2783 ga_clear(&ga);
|
|
|
2784 return e_outofmem;
|
|
|
2785 }
|
|
|
2786 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
2787 ++ga.ga_len;
|
|
|
2788
|
|
|
2789 /* Everything is fine, store the new b_langp value. */
|
|
|
2790 ga_clear(&buf->b_langp);
|
|
|
2791 buf->b_langp = ga;
|
|
|
2792
|
|
|
2793 return NULL;
|
|
|
2794 }
|
|
|
2795
|
|
|
2796 /*
|
|
376
|
2797 * Clear the midword characters for buffer "buf".
|
|
|
2798 */
|
|
|
2799 static void
|
|
|
2800 clear_midword(buf)
|
|
|
2801 buf_T *buf;
|
|
|
2802 {
|
|
|
2803 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
2804 #ifdef FEAT_MBYTE
|
|
|
2805 vim_free(buf->b_spell_ismw_mb);
|
|
|
2806 buf->b_spell_ismw_mb = NULL;
|
|
|
2807 #endif
|
|
|
2808 }
|
|
|
2809
|
|
|
2810 /*
|
|
|
2811 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
2812 * They add up to any currently used midword characters.
|
|
|
2813 */
|
|
|
2814 static void
|
|
|
2815 use_midword(lp, buf)
|
|
|
2816 slang_T *lp;
|
|
|
2817 buf_T *buf;
|
|
|
2818 {
|
|
|
2819 char_u *p;
|
|
|
2820
|
|
389
|
2821 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
2822 return;
|
|
|
2823
|
|
376
|
2824 for (p = lp->sl_midword; *p != NUL; )
|
|
|
2825 #ifdef FEAT_MBYTE
|
|
|
2826 if (has_mbyte)
|
|
|
2827 {
|
|
|
2828 int c, l, n;
|
|
|
2829 char_u *bp;
|
|
|
2830
|
|
|
2831 c = mb_ptr2char(p);
|
|
|
2832 l = mb_ptr2len_check(p);
|
|
|
2833 if (c < 256)
|
|
|
2834 buf->b_spell_ismw[c] = TRUE;
|
|
|
2835 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
2836 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
2837 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
2838 else
|
|
|
2839 {
|
|
|
2840 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
2841 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
2842 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
2843 if (bp != NULL)
|
|
|
2844 {
|
|
|
2845 vim_free(buf->b_spell_ismw_mb);
|
|
|
2846 buf->b_spell_ismw_mb = bp;
|
|
|
2847 vim_strncpy(bp + n, p, l);
|
|
|
2848 }
|
|
|
2849 }
|
|
|
2850 p += l;
|
|
|
2851 }
|
|
|
2852 else
|
|
|
2853 #endif
|
|
|
2854 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
2855 }
|
|
|
2856
|
|
|
2857 /*
|
|
236
|
2858 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
2859 * Each region is simply stored as the two characters of it's name.
|
|
381
|
2860 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
2861 */
|
|
|
2862 static int
|
|
|
2863 find_region(rp, region)
|
|
|
2864 char_u *rp;
|
|
|
2865 char_u *region;
|
|
|
2866 {
|
|
|
2867 int i;
|
|
|
2868
|
|
|
2869 for (i = 0; ; i += 2)
|
|
|
2870 {
|
|
|
2871 if (rp[i] == NUL)
|
|
|
2872 return REGION_ALL;
|
|
|
2873 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
2874 break;
|
|
|
2875 }
|
|
|
2876 return i / 2;
|
|
|
2877 }
|
|
|
2878
|
|
|
2879 /*
|
|
323
|
2880 * Return case type of word:
|
|
236
|
2881 * w word 0
|
|
300
|
2882 * Word WF_ONECAP
|
|
|
2883 * W WORD WF_ALLCAP
|
|
|
2884 * WoRd wOrd WF_KEEPCAP
|
|
236
|
2885 */
|
|
|
2886 static int
|
|
|
2887 captype(word, end)
|
|
|
2888 char_u *word;
|
|
323
|
2889 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
2890 {
|
|
|
2891 char_u *p;
|
|
|
2892 int c;
|
|
|
2893 int firstcap;
|
|
|
2894 int allcap;
|
|
|
2895 int past_second = FALSE; /* past second word char */
|
|
|
2896
|
|
|
2897 /* find first letter */
|
|
376
|
2898 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
2899 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
2900 return 0; /* only non-word characters, illegal word */
|
|
|
2901 #ifdef FEAT_MBYTE
|
|
310
|
2902 if (has_mbyte)
|
|
|
2903 c = mb_ptr2char_adv(&p);
|
|
|
2904 else
|
|
236
|
2905 #endif
|
|
310
|
2906 c = *p++;
|
|
324
|
2907 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
2908
|
|
|
2909 /*
|
|
|
2910 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
2911 * But a word with an upper char only at start is a ONECAP.
|
|
|
2912 */
|
|
323
|
2913 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
2914 if (spell_iswordp_nmw(p))
|
|
236
|
2915 {
|
|
455
|
2916 c = PTR2CHAR(p);
|
|
324
|
2917 if (!SPELL_ISUPPER(c))
|
|
236
|
2918 {
|
|
|
2919 /* UUl -> KEEPCAP */
|
|
|
2920 if (past_second && allcap)
|
|
300
|
2921 return WF_KEEPCAP;
|
|
236
|
2922 allcap = FALSE;
|
|
|
2923 }
|
|
|
2924 else if (!allcap)
|
|
|
2925 /* UlU -> KEEPCAP */
|
|
300
|
2926 return WF_KEEPCAP;
|
|
236
|
2927 past_second = TRUE;
|
|
|
2928 }
|
|
|
2929
|
|
|
2930 if (allcap)
|
|
300
|
2931 return WF_ALLCAP;
|
|
236
|
2932 if (firstcap)
|
|
300
|
2933 return WF_ONECAP;
|
|
236
|
2934 return 0;
|
|
|
2935 }
|
|
|
2936
|
|
355
|
2937 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
2938 /*
|
|
|
2939 * Free all languages.
|
|
|
2940 */
|
|
|
2941 void
|
|
|
2942 spell_free_all()
|
|
|
2943 {
|
|
|
2944 slang_T *lp;
|
|
|
2945 buf_T *buf;
|
|
385
|
2946 char_u fname[MAXPATHL];
|
|
355
|
2947
|
|
|
2948 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2949 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2950 ga_clear(&buf->b_langp);
|
|
|
2951
|
|
|
2952 while (first_lang != NULL)
|
|
|
2953 {
|
|
|
2954 lp = first_lang;
|
|
|
2955 first_lang = lp->sl_next;
|
|
|
2956 slang_free(lp);
|
|
|
2957 }
|
|
366
|
2958
|
|
385
|
2959 if (int_wordlist != NULL)
|
|
|
2960 {
|
|
|
2961 /* Delete the internal wordlist and its .spl file */
|
|
|
2962 mch_remove(int_wordlist);
|
|
|
2963 int_wordlist_spl(fname);
|
|
|
2964 mch_remove(fname);
|
|
|
2965 vim_free(int_wordlist);
|
|
|
2966 int_wordlist = NULL;
|
|
381
|
2967 }
|
|
|
2968
|
|
366
|
2969 init_spell_chartab();
|
|
355
|
2970 }
|
|
|
2971 # endif
|
|
|
2972
|
|
236
|
2973 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
2974 /*
|
|
|
2975 * Clear all spelling tables and reload them.
|
|
307
|
2976 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
2977 */
|
|
|
2978 void
|
|
|
2979 spell_reload()
|
|
|
2980 {
|
|
|
2981 buf_T *buf;
|
|
316
|
2982 win_T *wp;
|
|
236
|
2983
|
|
358
|
2984 /* Initialize the table for spell_iswordp(). */
|
|
236
|
2985 init_spell_chartab();
|
|
|
2986
|
|
|
2987 /* Unload all allocated memory. */
|
|
355
|
2988 spell_free_all();
|
|
236
|
2989
|
|
|
2990 /* Go through all buffers and handle 'spelllang'. */
|
|
|
2991 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
2992 {
|
|
316
|
2993 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
2994 * window for this buffer in which 'spell' is set. */
|
|
236
|
2995 if (*buf->b_p_spl != NUL)
|
|
316
|
2996 {
|
|
|
2997 FOR_ALL_WINDOWS(wp)
|
|
|
2998 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
2999 {
|
|
|
3000 (void)did_set_spelllang(buf);
|
|
|
3001 # ifdef FEAT_WINDOWS
|
|
|
3002 break;
|
|
|
3003 # endif
|
|
|
3004 }
|
|
|
3005 }
|
|
236
|
3006 }
|
|
|
3007 }
|
|
|
3008 # endif
|
|
|
3009
|
|
310
|
3010 /*
|
|
|
3011 * Reload the spell file "fname" if it's loaded.
|
|
|
3012 */
|
|
|
3013 static void
|
|
323
|
3014 spell_reload_one(fname, added_word)
|
|
310
|
3015 char_u *fname;
|
|
323
|
3016 int added_word; /* invoked through "zg" */
|
|
310
|
3017 {
|
|
|
3018 slang_T *lp;
|
|
323
|
3019 int didit = FALSE;
|
|
310
|
3020
|
|
|
3021 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3022 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3023 {
|
|
|
3024 slang_clear(lp);
|
|
323
|
3025 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3026 redraw_all_later(NOT_VALID);
|
|
323
|
3027 didit = TRUE;
|
|
310
|
3028 }
|
|
323
|
3029
|
|
|
3030 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3031 * 'spelllang' to get it loaded. */
|
|
|
3032 if (added_word && !didit)
|
|
|
3033 did_set_spelllang(curbuf);
|
|
310
|
3034 }
|
|
|
3035
|
|
|
3036
|
|
236
|
3037 /*
|
|
|
3038 * Functions for ":mkspell".
|
|
|
3039 */
|
|
|
3040
|
|
300
|
3041 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3042 and .dic file. */
|
|
|
3043 /*
|
|
|
3044 * Main structure to store the contents of a ".aff" file.
|
|
|
3045 */
|
|
|
3046 typedef struct afffile_S
|
|
|
3047 {
|
|
|
3048 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
310
|
3049 int af_rar; /* RAR ID for rare word */
|
|
|
3050 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
3051 int af_bad; /* BAD ID for banned word */
|
|
339
|
3052 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3053 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3054 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
3055 } afffile_T;
|
|
|
3056
|
|
|
3057 typedef struct affentry_S affentry_T;
|
|
|
3058 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3059 struct affentry_S
|
|
|
3060 {
|
|
|
3061 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3062 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3063 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3064 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3065 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
366
|
3066 int ae_rare; /* rare affix */
|
|
300
|
3067 };
|
|
|
3068
|
|
455
|
3069 #define AH_KEY_LEN 10
|
|
|
3070
|
|
300
|
3071 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3072 typedef struct affheader_S
|
|
|
3073 {
|
|
455
|
3074 /* key for hashtable == name of affix entry */
|
|
|
3075 #ifdef FEAT_MBYTE
|
|
|
3076 char_u ah_key[AH_KEY_LEN]; /* multi-byte char plus NUL */
|
|
|
3077 #else
|
|
|
3078 char_u ah_key[2]; /* one byte char plus NUL */
|
|
|
3079 #endif
|
|
339
|
3080 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
3081 int ah_combine; /* suffix may combine with prefix */
|
|
|
3082 affentry_T *ah_first; /* first affix entry */
|
|
|
3083 } affheader_T;
|
|
|
3084
|
|
|
3085 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
3086
|
|
|
3087 /*
|
|
|
3088 * Structure that is used to store the items in the word tree. This avoids
|
|
|
3089 * the need to keep track of each allocated thing, it's freed all at once
|
|
|
3090 * after ":mkspell" is done.
|
|
|
3091 */
|
|
|
3092 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
3093 typedef struct sblock_S sblock_T;
|
|
|
3094 struct sblock_S
|
|
|
3095 {
|
|
|
3096 sblock_T *sb_next; /* next block in list */
|
|
|
3097 int sb_used; /* nr of bytes already in use */
|
|
|
3098 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
3099 };
|
|
|
3100
|
|
|
3101 /*
|
|
300
|
3102 * A node in the tree.
|
|
236
|
3103 */
|
|
300
|
3104 typedef struct wordnode_S wordnode_T;
|
|
|
3105 struct wordnode_S
|
|
236
|
3106 {
|
|
346
|
3107 union /* shared to save space */
|
|
|
3108 {
|
|
|
3109 char_u hashkey[6]; /* room for the hash key */
|
|
|
3110 int index; /* index in written nodes (valid after first
|
|
|
3111 round) */
|
|
|
3112 } wn_u1;
|
|
|
3113 union /* shared to save space */
|
|
|
3114 {
|
|
|
3115 wordnode_T *next; /* next node with same hash key */
|
|
|
3116 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
3117 } wn_u2;
|
|
300
|
3118 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
3119 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
3120 always sorted) */
|
|
|
3121 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
390
|
3122 short_u wn_flags; /* when wn_byte is NUL: WF_ flags */
|
|
339
|
3123 short wn_region; /* when wn_byte is NUL: region mask; for
|
|
|
3124 PREFIXTREE it's the prefcondnr */
|
|
|
3125 char_u wn_prefixID; /* supported/required prefix ID or 0 */
|
|
236
|
3126 };
|
|
|
3127
|
|
390
|
3128 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
3129
|
|
300
|
3130 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
3131
|
|
300
|
3132 /*
|
|
|
3133 * Info used while reading the spell files.
|
|
|
3134 */
|
|
|
3135 typedef struct spellinfo_S
|
|
249
|
3136 {
|
|
300
|
3137 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
3138 long si_foldwcount; /* nr of words in si_foldroot */
|
|
300
|
3139 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
3140 long si_keepwcount; /* nr of words in si_keeproot */
|
|
339
|
3141 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
300
|
3142 sblock_T *si_blocks; /* memory blocks used */
|
|
|
3143 int si_ascii; /* handling only ASCII words */
|
|
310
|
3144 int si_add; /* addition file */
|
|
351
|
3145 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
3146 int si_region; /* region mask */
|
|
|
3147 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
3148 int si_memtot; /* runtime memory used */
|
|
310
|
3149 int si_verbose; /* verbose messages */
|
|
316
|
3150 int si_region_count; /* number of regions supported (1 when there
|
|
|
3151 are no regions) */
|
|
|
3152 char_u si_region_name[16]; /* region names (if count > 1) */
|
|
323
|
3153
|
|
|
3154 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
3155 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
3156 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
3157 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
3158 int si_followup; /* soundsalike: ? */
|
|
|
3159 int si_collapse; /* soundsalike: ? */
|
|
|
3160 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
3161 garray_T si_map; /* MAP info concatenated */
|
|
366
|
3162 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
339
|
3163 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
3164 * prefixes, each stored as a string */
|
|
|
3165 int si_newID; /* current value for ah_newID */
|
|
300
|
3166 } spellinfo_T;
|
|
249
|
3167
|
|
300
|
3168 static afffile_T *spell_read_aff __ARGS((char_u *fname, spellinfo_T *spin));
|
|
339
|
3169 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
3170 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
3171 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
3172 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
3173 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
|
3174 static int spell_read_dic __ARGS((char_u *fname, spellinfo_T *spin, afffile_T *affile));
|
|
339
|
3175 static char_u *get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, sblock_T **blp));
|
|
|
3176 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
|
3177 static int spell_read_wordfile __ARGS((char_u *fname, spellinfo_T *spin));
|
|
447
|
3178 static void *getroom __ARGS((sblock_T **blp, size_t len, int align));
|
|
300
|
3179 static char_u *getroom_save __ARGS((sblock_T **blp, char_u *s));
|
|
|
3180 static void free_blocks __ARGS((sblock_T *bl));
|
|
|
3181 static wordnode_T *wordtree_alloc __ARGS((sblock_T **blp));
|
|
339
|
3182 static int store_word __ARGS((char_u *word, spellinfo_T *spin, int flags, int region, char_u *pfxlist));
|
|
|
3183 static int tree_add_word __ARGS((char_u *word, wordnode_T *tree, int flags, int region, int prefixID, sblock_T **blp));
|
|
310
|
3184 static void wordtree_compress __ARGS((wordnode_T *root, spellinfo_T *spin));
|
|
300
|
3185 static int node_compress __ARGS((wordnode_T *node, hashtab_T *ht, int *tot));
|
|
|
3186 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
316
|
3187 static void write_vim_spell __ARGS((char_u *fname, spellinfo_T *spin));
|
|
346
|
3188 static void clear_node __ARGS((wordnode_T *node));
|
|
|
3189 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
3190 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3191 static void init_spellfile __ARGS((void));
|
|
236
|
3192
|
|
455
|
3193 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
3194 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
3195 * Use a negative number with the lower 8 bits zero. */
|
|
|
3196 #define PFX_FLAGS -256
|
|
390
|
3197
|
|
236
|
3198 /*
|
|
323
|
3199 * Read the affix file "fname".
|
|
316
|
3200 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
3201 */
|
|
|
3202 static afffile_T *
|
|
300
|
3203 spell_read_aff(fname, spin)
|
|
236
|
3204 char_u *fname;
|
|
300
|
3205 spellinfo_T *spin;
|
|
236
|
3206 {
|
|
|
3207 FILE *fd;
|
|
|
3208 afffile_T *aff;
|
|
|
3209 char_u rline[MAXLINELEN];
|
|
|
3210 char_u *line;
|
|
|
3211 char_u *pc = NULL;
|
|
334
|
3212 #define MAXITEMCNT 7
|
|
|
3213 char_u *(items[MAXITEMCNT]);
|
|
236
|
3214 int itemcnt;
|
|
|
3215 char_u *p;
|
|
|
3216 int lnum = 0;
|
|
|
3217 affheader_T *cur_aff = NULL;
|
|
|
3218 int aff_todo = 0;
|
|
|
3219 hashtab_T *tp;
|
|
255
|
3220 char_u *low = NULL;
|
|
|
3221 char_u *fol = NULL;
|
|
|
3222 char_u *upp = NULL;
|
|
307
|
3223 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
3224 int do_rep;
|
|
|
3225 int do_sal;
|
|
|
3226 int do_map;
|
|
366
|
3227 int do_midword;
|
|
375
|
3228 int do_sofo;
|
|
323
|
3229 int found_map = FALSE;
|
|
324
|
3230 hashitem_T *hi;
|
|
455
|
3231 int l;
|
|
236
|
3232
|
|
300
|
3233 /*
|
|
|
3234 * Open the file.
|
|
|
3235 */
|
|
310
|
3236 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3237 if (fd == NULL)
|
|
|
3238 {
|
|
|
3239 EMSG2(_(e_notopen), fname);
|
|
|
3240 return NULL;
|
|
|
3241 }
|
|
|
3242
|
|
310
|
3243 if (spin->si_verbose || p_verbose > 2)
|
|
|
3244 {
|
|
|
3245 if (!spin->si_verbose)
|
|
|
3246 verbose_enter();
|
|
366
|
3247 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
3248 out_flush();
|
|
|
3249 if (!spin->si_verbose)
|
|
|
3250 verbose_leave();
|
|
|
3251 }
|
|
236
|
3252
|
|
323
|
3253 /* Only do REP lines when not done in another .aff file already. */
|
|
|
3254 do_rep = spin->si_rep.ga_len == 0;
|
|
|
3255
|
|
|
3256 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
3257 do_sal = spin->si_sal.ga_len == 0;
|
|
|
3258
|
|
|
3259 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
3260 do_map = spin->si_map.ga_len == 0;
|
|
|
3261
|
|
366
|
3262 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
3263 do_midword = spin->si_midword == NULL;
|
|
|
3264
|
|
375
|
3265 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
3266 do_sofo = spin->si_sofofr == NULL;
|
|
|
3267
|
|
300
|
3268 /*
|
|
|
3269 * Allocate and init the afffile_T structure.
|
|
|
3270 */
|
|
447
|
3271 aff = (afffile_T *)getroom(&spin->si_blocks, sizeof(afffile_T), TRUE);
|
|
236
|
3272 if (aff == NULL)
|
|
|
3273 return NULL;
|
|
|
3274 hash_init(&aff->af_pref);
|
|
|
3275 hash_init(&aff->af_suff);
|
|
|
3276
|
|
|
3277 /*
|
|
|
3278 * Read all the lines in the file one by one.
|
|
|
3279 */
|
|
255
|
3280 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
3281 {
|
|
255
|
3282 line_breakcheck();
|
|
236
|
3283 ++lnum;
|
|
|
3284
|
|
|
3285 /* Skip comment lines. */
|
|
|
3286 if (*rline == '#')
|
|
|
3287 continue;
|
|
|
3288
|
|
|
3289 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
3290 vim_free(pc);
|
|
310
|
3291 #ifdef FEAT_MBYTE
|
|
300
|
3292 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3293 {
|
|
300
|
3294 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
3295 if (pc == NULL)
|
|
|
3296 {
|
|
|
3297 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3298 fname, lnum, rline);
|
|
|
3299 continue;
|
|
|
3300 }
|
|
236
|
3301 line = pc;
|
|
|
3302 }
|
|
|
3303 else
|
|
310
|
3304 #endif
|
|
236
|
3305 {
|
|
|
3306 pc = NULL;
|
|
|
3307 line = rline;
|
|
|
3308 }
|
|
|
3309
|
|
|
3310 /* Split the line up in white separated items. Put a NUL after each
|
|
|
3311 * item. */
|
|
|
3312 itemcnt = 0;
|
|
|
3313 for (p = line; ; )
|
|
|
3314 {
|
|
|
3315 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
3316 ++p;
|
|
|
3317 if (*p == NUL)
|
|
|
3318 break;
|
|
334
|
3319 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
3320 break;
|
|
236
|
3321 items[itemcnt++] = p;
|
|
300
|
3322 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
3323 ++p;
|
|
|
3324 if (*p == NUL)
|
|
|
3325 break;
|
|
|
3326 *p++ = NUL;
|
|
|
3327 }
|
|
|
3328
|
|
|
3329 /* Handle non-empty lines. */
|
|
|
3330 if (itemcnt > 0)
|
|
|
3331 {
|
|
|
3332 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
3333 && aff->af_enc == NULL)
|
|
|
3334 {
|
|
310
|
3335 #ifdef FEAT_MBYTE
|
|
300
|
3336 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
3337 aff->af_enc = enc_canonize(items[1]);
|
|
|
3338 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
3339 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
3340 p_enc) == FAIL)
|
|
|
3341 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
3342 fname, aff->af_enc, p_enc);
|
|
375
|
3343 spin->si_conv.vc_fail = TRUE;
|
|
310
|
3344 #else
|
|
|
3345 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
3346 #endif
|
|
236
|
3347 }
|
|
366
|
3348 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
3349 {
|
|
|
3350 if (do_midword)
|
|
|
3351 spin->si_midword = vim_strsave(items[1]);
|
|
|
3352 }
|
|
302
|
3353 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
3354 {
|
|
323
|
3355 /* ignored, we always split */
|
|
302
|
3356 }
|
|
323
|
3357 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
3358 {
|
|
323
|
3359 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
3360 }
|
|
307
|
3361 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
3362 && aff->af_rar == 0)
|
|
|
3363 {
|
|
|
3364 aff->af_rar = items[1][0];
|
|
|
3365 if (items[1][1] != NUL)
|
|
|
3366 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3367 }
|
|
310
|
3368 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
3369 && aff->af_kep == 0)
|
|
307
|
3370 {
|
|
310
|
3371 aff->af_kep = items[1][0];
|
|
307
|
3372 if (items[1][1] != NUL)
|
|
|
3373 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3374 }
|
|
346
|
3375 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
3376 && aff->af_bad == 0)
|
|
|
3377 {
|
|
|
3378 aff->af_bad = items[1][0];
|
|
|
3379 if (items[1][1] != NUL)
|
|
|
3380 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
3381 }
|
|
339
|
3382 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
3383 {
|
|
|
3384 aff->af_pfxpostpone = TRUE;
|
|
|
3385 }
|
|
236
|
3386 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3387 || STRCMP(items[0], "SFX") == 0)
|
|
|
3388 && aff_todo == 0
|
|
334
|
3389 && itemcnt >= 4)
|
|
236
|
3390 {
|
|
334
|
3391 /* Myspell allows extra text after the item, but that might
|
|
|
3392 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
3393 if (itemcnt > 4 && *items[4] != '#')
|
|
|
3394 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
3395 fname, lnum, items[4]);
|
|
|
3396
|
|
236
|
3397 /* New affix letter. */
|
|
300
|
3398 cur_aff = (affheader_T *)getroom(&spin->si_blocks,
|
|
447
|
3399 sizeof(affheader_T), TRUE);
|
|
236
|
3400 if (cur_aff == NULL)
|
|
|
3401 break;
|
|
455
|
3402 #ifdef FEAT_MBYTE
|
|
|
3403 if (has_mbyte)
|
|
|
3404 {
|
|
|
3405 l = mb_ptr2len_check(items[1]);
|
|
|
3406 if (l >= AH_KEY_LEN)
|
|
|
3407 l = 1; /* too long, must be an overlong sequence */
|
|
|
3408 else
|
|
|
3409 mch_memmove(cur_aff->ah_key, items[1], l);
|
|
|
3410 }
|
|
|
3411 else
|
|
|
3412 #endif
|
|
|
3413 {
|
|
|
3414 *cur_aff->ah_key = *items[1];
|
|
|
3415 l = 1;
|
|
|
3416 }
|
|
|
3417 cur_aff->ah_key[l] = NUL;
|
|
|
3418 if (items[1][l] != NUL)
|
|
307
|
3419 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
3420 if (*items[2] == 'Y')
|
|
|
3421 cur_aff->ah_combine = TRUE;
|
|
300
|
3422 else if (*items[2] != 'N')
|
|
236
|
3423 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
3424 fname, lnum, items[2]);
|
|
339
|
3425
|
|
236
|
3426 if (*items[0] == 'P')
|
|
339
|
3427 {
|
|
236
|
3428 tp = &aff->af_pref;
|
|
339
|
3429 /* Use a new number in the .spl file later, to be able to
|
|
|
3430 * handle multiple .aff files. */
|
|
|
3431 if (aff->af_pfxpostpone)
|
|
344
|
3432 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
3433 }
|
|
236
|
3434 else
|
|
|
3435 tp = &aff->af_suff;
|
|
300
|
3436 aff_todo = atoi((char *)items[3]);
|
|
324
|
3437 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
3438 if (!HASHITEM_EMPTY(hi))
|
|
300
|
3439 {
|
|
236
|
3440 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
3441 fname, lnum, items[1]);
|
|
300
|
3442 aff_todo = 0;
|
|
|
3443 }
|
|
236
|
3444 else
|
|
|
3445 hash_add(tp, cur_aff->ah_key);
|
|
|
3446 }
|
|
|
3447 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
3448 || STRCMP(items[0], "SFX") == 0)
|
|
|
3449 && aff_todo > 0
|
|
|
3450 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
3451 && itemcnt >= 5)
|
|
236
|
3452 {
|
|
|
3453 affentry_T *aff_entry;
|
|
366
|
3454 int rare = FALSE;
|
|
455
|
3455 int upper = FALSE;
|
|
366
|
3456 int lasti = 5;
|
|
|
3457
|
|
|
3458 /* Check for "rare" after the other info. */
|
|
|
3459 if (itemcnt > 5 && STRICMP(items[5], "rare") == 0)
|
|
|
3460 {
|
|
|
3461 rare = TRUE;
|
|
|
3462 lasti = 6;
|
|
|
3463 }
|
|
236
|
3464
|
|
334
|
3465 /* Myspell allows extra text after the item, but that might
|
|
|
3466 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
3467 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
3468 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
366
|
3469 fname, lnum, items[lasti]);
|
|
334
|
3470
|
|
236
|
3471 /* New item for an affix letter. */
|
|
|
3472 --aff_todo;
|
|
300
|
3473 aff_entry = (affentry_T *)getroom(&spin->si_blocks,
|
|
447
|
3474 sizeof(affentry_T), TRUE);
|
|
236
|
3475 if (aff_entry == NULL)
|
|
|
3476 break;
|
|
366
|
3477 aff_entry->ae_rare = rare;
|
|
240
|
3478
|
|
236
|
3479 if (STRCMP(items[2], "0") != 0)
|
|
300
|
3480 aff_entry->ae_chop = getroom_save(&spin->si_blocks,
|
|
|
3481 items[2]);
|
|
236
|
3482 if (STRCMP(items[3], "0") != 0)
|
|
300
|
3483 aff_entry->ae_add = getroom_save(&spin->si_blocks,
|
|
|
3484 items[3]);
|
|
236
|
3485
|
|
300
|
3486 /* Don't use an affix entry with non-ASCII characters when
|
|
|
3487 * "spin->si_ascii" is TRUE. */
|
|
|
3488 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
3489 || has_non_ascii(aff_entry->ae_add)))
|
|
|
3490 {
|
|
|
3491 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
3492 cur_aff->ah_first = aff_entry;
|
|
300
|
3493
|
|
|
3494 if (STRCMP(items[4], ".") != 0)
|
|
|
3495 {
|
|
|
3496 char_u buf[MAXLINELEN];
|
|
|
3497
|
|
|
3498 aff_entry->ae_cond = getroom_save(&spin->si_blocks,
|
|
|
3499 items[4]);
|
|
|
3500 if (*items[0] == 'P')
|
|
|
3501 sprintf((char *)buf, "^%s", items[4]);
|
|
|
3502 else
|
|
|
3503 sprintf((char *)buf, "%s$", items[4]);
|
|
|
3504 aff_entry->ae_prog = vim_regcomp(buf,
|
|
|
3505 RE_MAGIC + RE_STRING);
|
|
|
3506 }
|
|
339
|
3507
|
|
|
3508 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
3509 * for the condition. Use an existing one if possible. */
|
|
455
|
3510 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
3511 {
|
|
455
|
3512 /* When the chop string is one lower-case letter and
|
|
|
3513 * the add string ends in the upper-case letter we set
|
|
|
3514 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
3515 * letters from "ae_add". The condition must either
|
|
|
3516 * be empty or start with the same letter. */
|
|
|
3517 if (aff_entry->ae_chop != NULL
|
|
|
3518 && aff_entry->ae_add != NULL
|
|
|
3519 #ifdef FEAT_MBYTE
|
|
|
3520 && aff_entry->ae_chop[mb_ptr2len_check(
|
|
|
3521 aff_entry->ae_chop)] == NUL
|
|
|
3522 #else
|
|
|
3523 && aff_entry->ae_chop[1] == NUL
|
|
|
3524 #endif
|
|
|
3525 )
|
|
339
|
3526 {
|
|
455
|
3527 int c, c_up;
|
|
|
3528
|
|
|
3529 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
3530 c_up = SPELL_TOUPPER(c);
|
|
|
3531 if (c_up != c
|
|
|
3532 && (aff_entry->ae_cond == NULL
|
|
|
3533 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
3534 {
|
|
|
3535 p = aff_entry->ae_add
|
|
|
3536 + STRLEN(aff_entry->ae_add);
|
|
|
3537 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
3538 if (PTR2CHAR(p) == c_up)
|
|
|
3539 {
|
|
|
3540 upper = TRUE;
|
|
|
3541 aff_entry->ae_chop = NULL;
|
|
|
3542 *p = NUL;
|
|
|
3543
|
|
|
3544 /* The condition is matched with the
|
|
|
3545 * actual word, thus must check for the
|
|
|
3546 * upper-case letter. */
|
|
|
3547 if (aff_entry->ae_cond != NULL)
|
|
|
3548 {
|
|
|
3549 char_u buf[MAXLINELEN];
|
|
|
3550 #ifdef FEAT_MBYTE
|
|
|
3551 if (has_mbyte)
|
|
|
3552 {
|
|
|
3553 onecap_copy(items[4], buf, TRUE);
|
|
|
3554 aff_entry->ae_cond = getroom_save(
|
|
|
3555 &spin->si_blocks, buf);
|
|
|
3556 }
|
|
|
3557 else
|
|
|
3558 #endif
|
|
|
3559 *aff_entry->ae_cond = c_up;
|
|
|
3560 if (aff_entry->ae_cond != NULL)
|
|
|
3561 {
|
|
|
3562 sprintf((char *)buf, "^%s",
|
|
339
|
3563 aff_entry->ae_cond);
|
|
455
|
3564 vim_free(aff_entry->ae_prog);
|
|
|
3565 aff_entry->ae_prog = vim_regcomp(
|
|
|
3566 buf, RE_MAGIC + RE_STRING);
|
|
|
3567 }
|
|
|
3568 }
|
|
|
3569 }
|
|
|
3570 }
|
|
339
|
3571 }
|
|
|
3572
|
|
455
|
3573 if (aff_entry->ae_chop == NULL)
|
|
390
|
3574 {
|
|
455
|
3575 int idx;
|
|
|
3576 char_u **pp;
|
|
|
3577 int n;
|
|
|
3578
|
|
|
3579 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
3580 --idx)
|
|
|
3581 {
|
|
|
3582 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
3583 if (str_equal(p, aff_entry->ae_cond))
|
|
|
3584 break;
|
|
|
3585 }
|
|
|
3586 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
3587 {
|
|
|
3588 /* Not found, add a new condition. */
|
|
|
3589 idx = spin->si_prefcond.ga_len++;
|
|
|
3590 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
3591 + idx;
|
|
|
3592 if (aff_entry->ae_cond == NULL)
|
|
|
3593 *pp = NULL;
|
|
|
3594 else
|
|
|
3595 *pp = getroom_save(&spin->si_blocks,
|
|
|
3596 aff_entry->ae_cond);
|
|
|
3597 }
|
|
|
3598
|
|
|
3599 /* Add the prefix to the prefix tree. */
|
|
|
3600 if (aff_entry->ae_add == NULL)
|
|
|
3601 p = (char_u *)"";
|
|
390
|
3602 else
|
|
455
|
3603 p = aff_entry->ae_add;
|
|
|
3604 /* PFX_FLAGS is a negative number, so that
|
|
|
3605 * tree_add_word() knows this is the prefix tree. */
|
|
|
3606 n = PFX_FLAGS;
|
|
|
3607 if (rare)
|
|
|
3608 n |= WFP_RARE;
|
|
|
3609 if (!cur_aff->ah_combine)
|
|
|
3610 n |= WFP_NC;
|
|
|
3611 if (upper)
|
|
|
3612 n |= WFP_UP;
|
|
|
3613 tree_add_word(p, spin->si_prefroot, n,
|
|
|
3614 idx, cur_aff->ah_newID, &spin->si_blocks);
|
|
390
|
3615 }
|
|
339
|
3616 }
|
|
240
|
3617 }
|
|
236
|
3618 }
|
|
255
|
3619 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
3620 {
|
|
|
3621 if (fol != NULL)
|
|
|
3622 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
3623 fname, lnum);
|
|
|
3624 else
|
|
|
3625 fol = vim_strsave(items[1]);
|
|
|
3626 }
|
|
|
3627 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
3628 {
|
|
|
3629 if (low != NULL)
|
|
|
3630 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
3631 fname, lnum);
|
|
|
3632 else
|
|
|
3633 low = vim_strsave(items[1]);
|
|
|
3634 }
|
|
|
3635 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
3636 {
|
|
|
3637 if (upp != NULL)
|
|
|
3638 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
3639 fname, lnum);
|
|
|
3640 else
|
|
|
3641 upp = vim_strsave(items[1]);
|
|
|
3642 }
|
|
236
|
3643 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
3644 {
|
|
236
|
3645 /* Ignore REP count */;
|
|
323
|
3646 if (!isdigit(*items[1]))
|
|
|
3647 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
3648 fname, lnum);
|
|
|
3649 }
|
|
236
|
3650 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 3)
|
|
|
3651 {
|
|
|
3652 /* REP item */
|
|
323
|
3653 if (do_rep)
|
|
|
3654 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
3655 }
|
|
|
3656 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
3657 {
|
|
|
3658 /* MAP item or count */
|
|
|
3659 if (!found_map)
|
|
|
3660 {
|
|
|
3661 /* First line contains the count. */
|
|
|
3662 found_map = TRUE;
|
|
|
3663 if (!isdigit(*items[1]))
|
|
|
3664 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
3665 fname, lnum);
|
|
|
3666 }
|
|
|
3667 else if (do_map)
|
|
|
3668 {
|
|
346
|
3669 int c;
|
|
|
3670
|
|
|
3671 /* Check that every character appears only once. */
|
|
|
3672 for (p = items[1]; *p != NUL; )
|
|
|
3673 {
|
|
|
3674 #ifdef FEAT_MBYTE
|
|
|
3675 c = mb_ptr2char_adv(&p);
|
|
|
3676 #else
|
|
|
3677 c = *p++;
|
|
|
3678 #endif
|
|
|
3679 if ((spin->si_map.ga_len > 0
|
|
|
3680 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
3681 != NULL)
|
|
|
3682 || vim_strchr(p, c) != NULL)
|
|
|
3683 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
3684 fname, lnum);
|
|
|
3685 }
|
|
|
3686
|
|
323
|
3687 /* We simply concatenate all the MAP strings, separated by
|
|
|
3688 * slashes. */
|
|
|
3689 ga_concat(&spin->si_map, items[1]);
|
|
|
3690 ga_append(&spin->si_map, '/');
|
|
|
3691 }
|
|
|
3692 }
|
|
|
3693 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
3694 {
|
|
|
3695 if (do_sal)
|
|
|
3696 {
|
|
|
3697 /* SAL item (sounds-a-like)
|
|
|
3698 * Either one of the known keys or a from-to pair. */
|
|
|
3699 if (STRCMP(items[1], "followup") == 0)
|
|
|
3700 spin->si_followup = sal_to_bool(items[2]);
|
|
|
3701 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
3702 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
3703 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
3704 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
3705 else
|
|
|
3706 /* when "to" is "_" it means empty */
|
|
|
3707 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
3708 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
3709 : items[2]);
|
|
|
3710 }
|
|
236
|
3711 }
|
|
375
|
3712 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
3713 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
3714 {
|
|
|
3715 if (do_sofo)
|
|
|
3716 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
3717 }
|
|
|
3718 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
3719 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
3720 {
|
|
|
3721 if (do_sofo)
|
|
|
3722 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
3723 }
|
|
300
|
3724 else
|
|
236
|
3725 smsg((char_u *)_("Unrecognized item in %s line %d: %s"),
|
|
|
3726 fname, lnum, items[0]);
|
|
|
3727 }
|
|
|
3728 }
|
|
|
3729
|
|
375
|
3730 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
3731 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
3732 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
3733 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
3734 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
3735
|
|
255
|
3736 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
3737 {
|
|
351
|
3738 if (spin->si_clear_chartab)
|
|
|
3739 {
|
|
|
3740 /* Clear the char type tables, don't want to use any of the
|
|
|
3741 * currently used spell properties. */
|
|
|
3742 init_spell_chartab();
|
|
|
3743 spin->si_clear_chartab = FALSE;
|
|
|
3744 }
|
|
|
3745
|
|
316
|
3746 /*
|
|
|
3747 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
3748 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
3749 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
3750 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
3751 */
|
|
|
3752 if (!spin->si_ascii
|
|
|
3753 #ifdef FEAT_MBYTE
|
|
|
3754 && !enc_utf8
|
|
|
3755 #endif
|
|
|
3756 )
|
|
260
|
3757 {
|
|
|
3758 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
3759 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
3760 else
|
|
316
|
3761 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
3762 }
|
|
255
|
3763
|
|
|
3764 vim_free(fol);
|
|
|
3765 vim_free(low);
|
|
|
3766 vim_free(upp);
|
|
|
3767 }
|
|
|
3768
|
|
236
|
3769 vim_free(pc);
|
|
|
3770 fclose(fd);
|
|
|
3771 return aff;
|
|
|
3772 }
|
|
|
3773
|
|
|
3774 /*
|
|
339
|
3775 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
3776 * NULL as equal.
|
|
|
3777 */
|
|
|
3778 static int
|
|
|
3779 str_equal(s1, s2)
|
|
|
3780 char_u *s1;
|
|
|
3781 char_u *s2;
|
|
|
3782 {
|
|
|
3783 if (s1 == NULL || s2 == NULL)
|
|
|
3784 return s1 == s2;
|
|
|
3785 return STRCMP(s1, s2) == 0;
|
|
|
3786 }
|
|
|
3787
|
|
|
3788 /*
|
|
323
|
3789 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
3790 * They are stored case-folded.
|
|
|
3791 */
|
|
|
3792 static void
|
|
|
3793 add_fromto(spin, gap, from, to)
|
|
|
3794 spellinfo_T *spin;
|
|
|
3795 garray_T *gap;
|
|
|
3796 char_u *from;
|
|
|
3797 char_u *to;
|
|
|
3798 {
|
|
|
3799 fromto_T *ftp;
|
|
|
3800 char_u word[MAXWLEN];
|
|
|
3801
|
|
|
3802 if (ga_grow(gap, 1) == OK)
|
|
|
3803 {
|
|
|
3804 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
3805 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
|
3806 ftp->ft_from = getroom_save(&spin->si_blocks, word);
|
|
|
3807 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
|
3808 ftp->ft_to = getroom_save(&spin->si_blocks, word);
|
|
|
3809 ++gap->ga_len;
|
|
|
3810 }
|
|
|
3811 }
|
|
|
3812
|
|
|
3813 /*
|
|
|
3814 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
3815 */
|
|
|
3816 static int
|
|
|
3817 sal_to_bool(s)
|
|
|
3818 char_u *s;
|
|
|
3819 {
|
|
|
3820 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
3821 }
|
|
|
3822
|
|
|
3823 /*
|
|
240
|
3824 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
3825 * When "s" is NULL FALSE is returned.
|
|
|
3826 */
|
|
|
3827 static int
|
|
|
3828 has_non_ascii(s)
|
|
|
3829 char_u *s;
|
|
|
3830 {
|
|
|
3831 char_u *p;
|
|
|
3832
|
|
|
3833 if (s != NULL)
|
|
|
3834 for (p = s; *p != NUL; ++p)
|
|
|
3835 if (*p >= 128)
|
|
|
3836 return TRUE;
|
|
|
3837 return FALSE;
|
|
|
3838 }
|
|
|
3839
|
|
|
3840 /*
|
|
236
|
3841 * Free the structure filled by spell_read_aff().
|
|
|
3842 */
|
|
|
3843 static void
|
|
|
3844 spell_free_aff(aff)
|
|
|
3845 afffile_T *aff;
|
|
|
3846 {
|
|
|
3847 hashtab_T *ht;
|
|
|
3848 hashitem_T *hi;
|
|
|
3849 int todo;
|
|
|
3850 affheader_T *ah;
|
|
300
|
3851 affentry_T *ae;
|
|
236
|
3852
|
|
|
3853 vim_free(aff->af_enc);
|
|
|
3854
|
|
339
|
3855 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
3856 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
3857 {
|
|
|
3858 todo = ht->ht_used;
|
|
|
3859 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
3860 {
|
|
|
3861 if (!HASHITEM_EMPTY(hi))
|
|
|
3862 {
|
|
|
3863 --todo;
|
|
|
3864 ah = HI2AH(hi);
|
|
300
|
3865 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
3866 vim_free(ae->ae_prog);
|
|
236
|
3867 }
|
|
|
3868 }
|
|
|
3869 if (ht == &aff->af_suff)
|
|
|
3870 break;
|
|
|
3871 }
|
|
300
|
3872
|
|
236
|
3873 hash_clear(&aff->af_pref);
|
|
|
3874 hash_clear(&aff->af_suff);
|
|
|
3875 }
|
|
|
3876
|
|
|
3877 /*
|
|
300
|
3878 * Read dictionary file "fname".
|
|
236
|
3879 * Returns OK or FAIL;
|
|
|
3880 */
|
|
|
3881 static int
|
|
300
|
3882 spell_read_dic(fname, spin, affile)
|
|
236
|
3883 char_u *fname;
|
|
300
|
3884 spellinfo_T *spin;
|
|
|
3885 afffile_T *affile;
|
|
236
|
3886 {
|
|
300
|
3887 hashtab_T ht;
|
|
236
|
3888 char_u line[MAXLINELEN];
|
|
300
|
3889 char_u *afflist;
|
|
339
|
3890 char_u *pfxlist;
|
|
300
|
3891 char_u *dw;
|
|
236
|
3892 char_u *pc;
|
|
|
3893 char_u *w;
|
|
|
3894 int l;
|
|
|
3895 hash_T hash;
|
|
|
3896 hashitem_T *hi;
|
|
|
3897 FILE *fd;
|
|
|
3898 int lnum = 1;
|
|
300
|
3899 int non_ascii = 0;
|
|
|
3900 int retval = OK;
|
|
|
3901 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
3902 int flags;
|
|
236
|
3903
|
|
300
|
3904 /*
|
|
|
3905 * Open the file.
|
|
|
3906 */
|
|
310
|
3907 fd = mch_fopen((char *)fname, "r");
|
|
236
|
3908 if (fd == NULL)
|
|
|
3909 {
|
|
|
3910 EMSG2(_(e_notopen), fname);
|
|
|
3911 return FAIL;
|
|
|
3912 }
|
|
|
3913
|
|
300
|
3914 /* The hashtable is only used to detect duplicated words. */
|
|
|
3915 hash_init(&ht);
|
|
|
3916
|
|
334
|
3917 spin->si_foldwcount = 0;
|
|
|
3918 spin->si_keepwcount = 0;
|
|
|
3919
|
|
310
|
3920 if (spin->si_verbose || p_verbose > 2)
|
|
|
3921 {
|
|
|
3922 if (!spin->si_verbose)
|
|
|
3923 verbose_enter();
|
|
366
|
3924 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
3925 out_flush();
|
|
|
3926 if (!spin->si_verbose)
|
|
|
3927 verbose_leave();
|
|
|
3928 }
|
|
236
|
3929
|
|
|
3930 /* Read and ignore the first line: word count. */
|
|
|
3931 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
3932 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
3933 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
3934
|
|
|
3935 /*
|
|
|
3936 * Read all the lines in the file one by one.
|
|
|
3937 * The words are converted to 'encoding' here, before being added to
|
|
|
3938 * the hashtable.
|
|
|
3939 */
|
|
255
|
3940 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
3941 {
|
|
255
|
3942 line_breakcheck();
|
|
236
|
3943 ++lnum;
|
|
455
|
3944 if (line[0] == '#' || line[0] == '/')
|
|
351
|
3945 continue; /* comment line */
|
|
236
|
3946
|
|
300
|
3947 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
3948 * the word is kept to allow e.g., "et al.". */
|
|
236
|
3949 l = STRLEN(line);
|
|
|
3950 while (l > 0 && line[l - 1] <= ' ')
|
|
|
3951 --l;
|
|
|
3952 if (l == 0)
|
|
|
3953 continue; /* empty line */
|
|
|
3954 line[l] = NUL;
|
|
|
3955
|
|
|
3956 /* Find the optional affix names. */
|
|
300
|
3957 afflist = vim_strchr(line, '/');
|
|
|
3958 if (afflist != NULL)
|
|
|
3959 *afflist++ = NUL;
|
|
236
|
3960
|
|
300
|
3961 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
3962 if (spin->si_ascii && has_non_ascii(line))
|
|
|
3963 {
|
|
|
3964 ++non_ascii;
|
|
240
|
3965 continue;
|
|
300
|
3966 }
|
|
240
|
3967
|
|
310
|
3968 #ifdef FEAT_MBYTE
|
|
236
|
3969 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
3970 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
3971 {
|
|
300
|
3972 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
3973 if (pc == NULL)
|
|
|
3974 {
|
|
|
3975 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
3976 fname, lnum, line);
|
|
|
3977 continue;
|
|
|
3978 }
|
|
236
|
3979 w = pc;
|
|
|
3980 }
|
|
|
3981 else
|
|
310
|
3982 #endif
|
|
236
|
3983 {
|
|
|
3984 pc = NULL;
|
|
|
3985 w = line;
|
|
|
3986 }
|
|
|
3987
|
|
339
|
3988 /* This takes time, print a message now and then. */
|
|
|
3989 if (spin->si_verbose && (lnum & 0x3ff) == 0)
|
|
|
3990 {
|
|
|
3991 vim_snprintf((char *)message, sizeof(message),
|
|
|
3992 _("line %6d, word %6d - %s"),
|
|
|
3993 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
3994 msg_start();
|
|
|
3995 msg_puts_long_attr(message, 0);
|
|
|
3996 msg_clr_eos();
|
|
|
3997 msg_didout = FALSE;
|
|
|
3998 msg_col = 0;
|
|
|
3999 out_flush();
|
|
|
4000 }
|
|
|
4001
|
|
300
|
4002 /* Store the word in the hashtable to be able to find duplicates. */
|
|
|
4003 dw = (char_u *)getroom_save(&spin->si_blocks, w);
|
|
236
|
4004 if (dw == NULL)
|
|
300
|
4005 retval = FAIL;
|
|
|
4006 vim_free(pc);
|
|
|
4007 if (retval == FAIL)
|
|
236
|
4008 break;
|
|
|
4009
|
|
300
|
4010 hash = hash_hash(dw);
|
|
|
4011 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
4012 if (!HASHITEM_EMPTY(hi))
|
|
|
4013 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
4014 fname, lnum, dw);
|
|
236
|
4015 else
|
|
300
|
4016 hash_add_item(&ht, hi, dw, hash);
|
|
|
4017
|
|
307
|
4018 flags = 0;
|
|
339
|
4019 pfxlist = NULL;
|
|
307
|
4020 if (afflist != NULL)
|
|
|
4021 {
|
|
|
4022 /* Check for affix name that stands for keep-case word and stands
|
|
|
4023 * for rare word (if defined). */
|
|
310
|
4024 if (affile->af_kep != NUL
|
|
|
4025 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
389
|
4026 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
307
|
4027 if (affile->af_rar != NUL
|
|
|
4028 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
4029 flags |= WF_RARE;
|
|
346
|
4030 if (affile->af_bad != NUL
|
|
|
4031 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
4032 flags |= WF_BANNED;
|
|
339
|
4033
|
|
|
4034 if (affile->af_pfxpostpone)
|
|
|
4035 /* Need to store the list of prefix IDs with the word. */
|
|
|
4036 pfxlist = get_pfxlist(affile, afflist, &spin->si_blocks);
|
|
307
|
4037 }
|
|
|
4038
|
|
300
|
4039 /* Add the word to the word tree(s). */
|
|
339
|
4040 if (store_word(dw, spin, flags, spin->si_region, pfxlist) == FAIL)
|
|
300
|
4041 retval = FAIL;
|
|
236
|
4042
|
|
300
|
4043 if (afflist != NULL)
|
|
|
4044 {
|
|
|
4045 /* Find all matching suffixes and add the resulting words.
|
|
|
4046 * Additionally do matching prefixes that combine. */
|
|
339
|
4047 if (store_aff_word(dw, spin, afflist, affile,
|
|
307
|
4048 &affile->af_suff, &affile->af_pref,
|
|
339
|
4049 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
4050 retval = FAIL;
|
|
|
4051
|
|
|
4052 /* Find all matching prefixes and add the resulting words. */
|
|
339
|
4053 if (store_aff_word(dw, spin, afflist, affile,
|
|
|
4054 &affile->af_pref, NULL,
|
|
|
4055 FALSE, flags, pfxlist) == FAIL)
|
|
300
|
4056 retval = FAIL;
|
|
|
4057 }
|
|
236
|
4058 }
|
|
|
4059
|
|
300
|
4060 if (spin->si_ascii && non_ascii > 0)
|
|
|
4061 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4062 non_ascii);
|
|
|
4063 hash_clear(&ht);
|
|
|
4064
|
|
236
|
4065 fclose(fd);
|
|
300
|
4066 return retval;
|
|
236
|
4067 }
|
|
|
4068
|
|
|
4069 /*
|
|
339
|
4070 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
4071 * Used for PFXPOSTPONE.
|
|
|
4072 * Returns a string allocated with getroom(). NULL when there are no prefixes
|
|
|
4073 * or when out of memory.
|
|
|
4074 */
|
|
|
4075 static char_u *
|
|
|
4076 get_pfxlist(affile, afflist, blp)
|
|
|
4077 afffile_T *affile;
|
|
|
4078 char_u *afflist;
|
|
|
4079 sblock_T **blp;
|
|
|
4080 {
|
|
|
4081 char_u *p;
|
|
|
4082 int cnt;
|
|
|
4083 int round;
|
|
|
4084 char_u *res = NULL;
|
|
|
4085 char_u key[2];
|
|
|
4086 hashitem_T *hi;
|
|
|
4087
|
|
|
4088 key[1] = NUL;
|
|
|
4089
|
|
|
4090 /* round 1: count the number of prefix IDs.
|
|
|
4091 * round 2: move prefix IDs to "res" */
|
|
|
4092 for (round = 1; round <= 2; ++round)
|
|
|
4093 {
|
|
|
4094 cnt = 0;
|
|
|
4095 for (p = afflist; *p != NUL; ++p)
|
|
|
4096 {
|
|
|
4097 key[0] = *p;
|
|
|
4098 hi = hash_find(&affile->af_pref, key);
|
|
|
4099 if (!HASHITEM_EMPTY(hi))
|
|
|
4100 {
|
|
|
4101 /* This is a prefix ID, use the new number. */
|
|
|
4102 if (round == 2)
|
|
|
4103 res[cnt] = HI2AH(hi)->ah_newID;
|
|
|
4104 ++cnt;
|
|
|
4105 }
|
|
|
4106 }
|
|
|
4107 if (round == 1 && cnt > 0)
|
|
447
|
4108 res = getroom(blp, cnt + 1, FALSE);
|
|
339
|
4109 if (res == NULL)
|
|
|
4110 break;
|
|
|
4111 }
|
|
|
4112
|
|
|
4113 if (res != NULL)
|
|
|
4114 res[cnt] = NUL;
|
|
|
4115 return res;
|
|
|
4116 }
|
|
|
4117
|
|
|
4118 /*
|
|
300
|
4119 * Apply affixes to a word and store the resulting words.
|
|
|
4120 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
4121 * prefixes or suffixes.
|
|
|
4122 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
4123 * the resulting words for combining affixes.
|
|
|
4124 *
|
|
|
4125 * Returns FAIL when out of memory.
|
|
236
|
4126 */
|
|
300
|
4127 static int
|
|
339
|
4128 store_aff_word(word, spin, afflist, affile, ht, xht, comb, flags, pfxlist)
|
|
300
|
4129 char_u *word; /* basic word start */
|
|
|
4130 spellinfo_T *spin; /* spell info */
|
|
|
4131 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
4132 afffile_T *affile;
|
|
300
|
4133 hashtab_T *ht;
|
|
|
4134 hashtab_T *xht;
|
|
|
4135 int comb; /* only use affixes that combine */
|
|
307
|
4136 int flags; /* flags for the word */
|
|
339
|
4137 char_u *pfxlist; /* list of prefix IDs */
|
|
236
|
4138 {
|
|
|
4139 int todo;
|
|
|
4140 hashitem_T *hi;
|
|
300
|
4141 affheader_T *ah;
|
|
|
4142 affentry_T *ae;
|
|
|
4143 regmatch_T regmatch;
|
|
|
4144 char_u newword[MAXWLEN];
|
|
|
4145 int retval = OK;
|
|
|
4146 int i;
|
|
|
4147 char_u *p;
|
|
366
|
4148 int use_flags;
|
|
390
|
4149 char_u *use_pfxlist;
|
|
455
|
4150 int c;
|
|
236
|
4151
|
|
300
|
4152 todo = ht->ht_used;
|
|
|
4153 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
4154 {
|
|
|
4155 if (!HASHITEM_EMPTY(hi))
|
|
|
4156 {
|
|
|
4157 --todo;
|
|
300
|
4158 ah = HI2AH(hi);
|
|
236
|
4159
|
|
300
|
4160 /* Check that the affix combines, if required, and that the word
|
|
|
4161 * supports this affix. */
|
|
455
|
4162 c = PTR2CHAR(ah->ah_key);
|
|
|
4163 if ((!comb || ah->ah_combine) && vim_strchr(afflist, c) != NULL)
|
|
236
|
4164 {
|
|
300
|
4165 /* Loop over all affix entries with this name. */
|
|
|
4166 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
4167 {
|
|
300
|
4168 /* Check the condition. It's not logical to match case
|
|
|
4169 * here, but it is required for compatibility with
|
|
339
|
4170 * Myspell.
|
|
|
4171 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
4172 * prefixes with a chop string. */
|
|
300
|
4173 regmatch.regprog = ae->ae_prog;
|
|
|
4174 regmatch.rm_ic = FALSE;
|
|
339
|
4175 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
4176 || ae->ae_chop != NULL)
|
|
|
4177 && (ae->ae_prog == NULL
|
|
|
4178 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
4179 {
|
|
|
4180 /* Match. Remove the chop and add the affix. */
|
|
|
4181 if (xht == NULL)
|
|
240
|
4182 {
|
|
300
|
4183 /* prefix: chop/add at the start of the word */
|
|
|
4184 if (ae->ae_add == NULL)
|
|
|
4185 *newword = NUL;
|
|
|
4186 else
|
|
|
4187 STRCPY(newword, ae->ae_add);
|
|
|
4188 p = word;
|
|
|
4189 if (ae->ae_chop != NULL)
|
|
310
|
4190 {
|
|
300
|
4191 /* Skip chop string. */
|
|
310
|
4192 #ifdef FEAT_MBYTE
|
|
|
4193 if (has_mbyte)
|
|
324
|
4194 {
|
|
310
|
4195 i = mb_charlen(ae->ae_chop);
|
|
324
|
4196 for ( ; i > 0; --i)
|
|
|
4197 mb_ptr_adv(p);
|
|
|
4198 }
|
|
310
|
4199 else
|
|
|
4200 #endif
|
|
324
|
4201 p += STRLEN(ae->ae_chop);
|
|
310
|
4202 }
|
|
300
|
4203 STRCAT(newword, p);
|
|
|
4204 }
|
|
|
4205 else
|
|
|
4206 {
|
|
|
4207 /* suffix: chop/add at the end of the word */
|
|
|
4208 STRCPY(newword, word);
|
|
|
4209 if (ae->ae_chop != NULL)
|
|
|
4210 {
|
|
|
4211 /* Remove chop string. */
|
|
|
4212 p = newword + STRLEN(newword);
|
|
376
|
4213 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
4214 for ( ; i > 0; --i)
|
|
300
|
4215 mb_ptr_back(newword, p);
|
|
|
4216 *p = NUL;
|
|
|
4217 }
|
|
|
4218 if (ae->ae_add != NULL)
|
|
|
4219 STRCAT(newword, ae->ae_add);
|
|
240
|
4220 }
|
|
|
4221
|
|
366
|
4222 /* Obey the "rare" flag of the affix. */
|
|
|
4223 if (ae->ae_rare)
|
|
|
4224 use_flags = flags | WF_RARE;
|
|
|
4225 else
|
|
|
4226 use_flags = flags;
|
|
390
|
4227 use_pfxlist = pfxlist;
|
|
|
4228
|
|
|
4229 /* When there are postponed prefixes... */
|
|
393
|
4230 if (spin->si_prefroot != NULL
|
|
|
4231 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
4232 {
|
|
|
4233 /* ... add a flag to indicate an affix was used. */
|
|
|
4234 use_flags |= WF_HAS_AFF;
|
|
|
4235
|
|
|
4236 /* ... don't use a prefix list if combining
|
|
|
4237 * affixes is not allowed */
|
|
|
4238 if (!ah->ah_combine || comb)
|
|
|
4239 use_pfxlist = NULL;
|
|
|
4240 }
|
|
366
|
4241
|
|
300
|
4242 /* Store the modified word. */
|
|
366
|
4243 if (store_word(newword, spin, use_flags,
|
|
390
|
4244 spin->si_region, use_pfxlist) == FAIL)
|
|
300
|
4245 retval = FAIL;
|
|
236
|
4246
|
|
300
|
4247 /* When added a suffix and combining is allowed also
|
|
|
4248 * try adding prefixes additionally. */
|
|
|
4249 if (xht != NULL && ah->ah_combine)
|
|
339
|
4250 if (store_aff_word(newword, spin, afflist, affile,
|
|
390
|
4251 xht, NULL, TRUE,
|
|
|
4252 use_flags, use_pfxlist) == FAIL)
|
|
300
|
4253 retval = FAIL;
|
|
236
|
4254 }
|
|
|
4255 }
|
|
|
4256 }
|
|
|
4257 }
|
|
|
4258 }
|
|
|
4259
|
|
|
4260 return retval;
|
|
|
4261 }
|
|
|
4262
|
|
|
4263 /*
|
|
300
|
4264 * Read a file with a list of words.
|
|
236
|
4265 */
|
|
|
4266 static int
|
|
300
|
4267 spell_read_wordfile(fname, spin)
|
|
|
4268 char_u *fname;
|
|
|
4269 spellinfo_T *spin;
|
|
236
|
4270 {
|
|
300
|
4271 FILE *fd;
|
|
|
4272 long lnum = 0;
|
|
|
4273 char_u rline[MAXLINELEN];
|
|
|
4274 char_u *line;
|
|
|
4275 char_u *pc = NULL;
|
|
381
|
4276 char_u *p;
|
|
300
|
4277 int l;
|
|
|
4278 int retval = OK;
|
|
|
4279 int did_word = FALSE;
|
|
|
4280 int non_ascii = 0;
|
|
307
|
4281 int flags;
|
|
316
|
4282 int regionmask;
|
|
236
|
4283
|
|
300
|
4284 /*
|
|
|
4285 * Open the file.
|
|
|
4286 */
|
|
310
|
4287 fd = mch_fopen((char *)fname, "r");
|
|
300
|
4288 if (fd == NULL)
|
|
236
|
4289 {
|
|
300
|
4290 EMSG2(_(e_notopen), fname);
|
|
|
4291 return FAIL;
|
|
236
|
4292 }
|
|
|
4293
|
|
310
|
4294 if (spin->si_verbose || p_verbose > 2)
|
|
|
4295 {
|
|
|
4296 if (!spin->si_verbose)
|
|
|
4297 verbose_enter();
|
|
366
|
4298 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
4299 out_flush();
|
|
|
4300 if (!spin->si_verbose)
|
|
|
4301 verbose_leave();
|
|
|
4302 }
|
|
300
|
4303
|
|
|
4304 /*
|
|
|
4305 * Read all the lines in the file one by one.
|
|
|
4306 */
|
|
|
4307 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
4308 {
|
|
|
4309 line_breakcheck();
|
|
|
4310 ++lnum;
|
|
|
4311
|
|
|
4312 /* Skip comment lines. */
|
|
|
4313 if (*rline == '#')
|
|
|
4314 continue;
|
|
|
4315
|
|
|
4316 /* Remove CR, LF and white space from the end. */
|
|
|
4317 l = STRLEN(rline);
|
|
|
4318 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
4319 --l;
|
|
|
4320 if (l == 0)
|
|
|
4321 continue; /* empty or blank line */
|
|
|
4322 rline[l] = NUL;
|
|
|
4323
|
|
|
4324 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
4325 vim_free(pc);
|
|
310
|
4326 #ifdef FEAT_MBYTE
|
|
300
|
4327 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
4328 {
|
|
|
4329 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
4330 if (pc == NULL)
|
|
|
4331 {
|
|
|
4332 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4333 fname, lnum, rline);
|
|
|
4334 continue;
|
|
|
4335 }
|
|
|
4336 line = pc;
|
|
|
4337 }
|
|
|
4338 else
|
|
310
|
4339 #endif
|
|
300
|
4340 {
|
|
|
4341 pc = NULL;
|
|
|
4342 line = rline;
|
|
|
4343 }
|
|
|
4344
|
|
307
|
4345 if (*line == '/')
|
|
300
|
4346 {
|
|
307
|
4347 ++line;
|
|
|
4348 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
4349 {
|
|
|
4350 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
4351 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
4352 fname, lnum, line - 1);
|
|
300
|
4353 else if (did_word)
|
|
316
|
4354 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
4355 fname, lnum, line - 1);
|
|
300
|
4356 else
|
|
|
4357 {
|
|
310
|
4358 #ifdef FEAT_MBYTE
|
|
|
4359 char_u *enc;
|
|
|
4360
|
|
300
|
4361 /* Setup for conversion to 'encoding'. */
|
|
316
|
4362 line += 10;
|
|
|
4363 enc = enc_canonize(line);
|
|
300
|
4364 if (enc != NULL && !spin->si_ascii
|
|
|
4365 && convert_setup(&spin->si_conv, enc,
|
|
|
4366 p_enc) == FAIL)
|
|
|
4367 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
4368 fname, line, p_enc);
|
|
300
|
4369 vim_free(enc);
|
|
375
|
4370 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4371 #else
|
|
|
4372 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4373 #endif
|
|
300
|
4374 }
|
|
307
|
4375 continue;
|
|
300
|
4376 }
|
|
307
|
4377
|
|
316
|
4378 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
4379 {
|
|
|
4380 if (spin->si_region_count > 1)
|
|
|
4381 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
4382 fname, lnum, line);
|
|
|
4383 else
|
|
|
4384 {
|
|
|
4385 line += 8;
|
|
|
4386 if (STRLEN(line) > 16)
|
|
|
4387 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
4388 fname, lnum, line);
|
|
|
4389 else
|
|
|
4390 {
|
|
|
4391 spin->si_region_count = STRLEN(line) / 2;
|
|
|
4392 STRCPY(spin->si_region_name, line);
|
|
389
|
4393
|
|
|
4394 /* Adjust the mask for a word valid in all regions. */
|
|
|
4395 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
4396 }
|
|
|
4397 }
|
|
|
4398 continue;
|
|
|
4399 }
|
|
|
4400
|
|
381
|
4401 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
4402 fname, lnum, line - 1);
|
|
|
4403 continue;
|
|
|
4404 }
|
|
|
4405
|
|
|
4406 flags = 0;
|
|
|
4407 regionmask = spin->si_region;
|
|
|
4408
|
|
|
4409 /* Check for flags and region after a slash. */
|
|
|
4410 p = vim_strchr(line, '/');
|
|
|
4411 if (p != NULL)
|
|
|
4412 {
|
|
|
4413 *p++ = NUL;
|
|
|
4414 while (*p != NUL)
|
|
|
4415 {
|
|
|
4416 if (*p == '=') /* keep-case word */
|
|
389
|
4417 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
4418 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
4419 flags |= WF_BANNED;
|
|
|
4420 else if (*p == '?') /* Rare word. */
|
|
|
4421 flags |= WF_RARE;
|
|
|
4422 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
4423 {
|
|
381
|
4424 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
4425 regionmask = 0;
|
|
|
4426 flags |= WF_REGION;
|
|
|
4427
|
|
|
4428 l = *p - '0';
|
|
316
|
4429 if (l > spin->si_region_count)
|
|
|
4430 {
|
|
|
4431 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
4432 fname, lnum, p);
|
|
316
|
4433 break;
|
|
|
4434 }
|
|
|
4435 regionmask |= 1 << (l - 1);
|
|
|
4436 }
|
|
381
|
4437 else
|
|
|
4438 {
|
|
|
4439 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
4440 fname, lnum, p);
|
|
|
4441 break;
|
|
|
4442 }
|
|
|
4443 ++p;
|
|
307
|
4444 }
|
|
300
|
4445 }
|
|
|
4446
|
|
|
4447 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4448 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4449 {
|
|
|
4450 ++non_ascii;
|
|
|
4451 continue;
|
|
|
4452 }
|
|
|
4453
|
|
|
4454 /* Normal word: store it. */
|
|
339
|
4455 if (store_word(line, spin, flags, regionmask, NULL) == FAIL)
|
|
300
|
4456 {
|
|
|
4457 retval = FAIL;
|
|
|
4458 break;
|
|
|
4459 }
|
|
|
4460 did_word = TRUE;
|
|
|
4461 }
|
|
|
4462
|
|
|
4463 vim_free(pc);
|
|
|
4464 fclose(fd);
|
|
|
4465
|
|
310
|
4466 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
4467 {
|
|
|
4468 if (p_verbose > 2)
|
|
|
4469 verbose_enter();
|
|
300
|
4470 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
4471 non_ascii);
|
|
310
|
4472 if (p_verbose > 2)
|
|
|
4473 verbose_leave();
|
|
|
4474 }
|
|
300
|
4475 return retval;
|
|
236
|
4476 }
|
|
|
4477
|
|
|
4478 /*
|
|
300
|
4479 * Get part of an sblock_T, "len" bytes long.
|
|
|
4480 * This avoids calling free() for every little struct we use.
|
|
|
4481 * The memory is cleared to all zeros.
|
|
|
4482 * Returns NULL when out of memory.
|
|
|
4483 */
|
|
|
4484 static void *
|
|
447
|
4485 getroom(blp, len, align)
|
|
300
|
4486 sblock_T **blp;
|
|
447
|
4487 size_t len; /* length needed */
|
|
|
4488 int align; /* align for pointer */
|
|
300
|
4489 {
|
|
|
4490 char_u *p;
|
|
|
4491 sblock_T *bl = *blp;
|
|
|
4492
|
|
447
|
4493 if (align && bl != NULL)
|
|
|
4494 /* Round size up for alignment. On some systems structures need to be
|
|
|
4495 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
4496 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
4497 & ~(sizeof(char *) - 1);
|
|
|
4498
|
|
300
|
4499 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
4500 {
|
|
|
4501 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
4502 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
4503 if (bl == NULL)
|
|
|
4504 return NULL;
|
|
|
4505 bl->sb_next = *blp;
|
|
|
4506 *blp = bl;
|
|
|
4507 bl->sb_used = 0;
|
|
|
4508 }
|
|
|
4509
|
|
|
4510 p = bl->sb_data + bl->sb_used;
|
|
|
4511 bl->sb_used += len;
|
|
|
4512
|
|
|
4513 return p;
|
|
|
4514 }
|
|
|
4515
|
|
|
4516 /*
|
|
|
4517 * Make a copy of a string into memory allocated with getroom().
|
|
|
4518 */
|
|
|
4519 static char_u *
|
|
|
4520 getroom_save(blp, s)
|
|
|
4521 sblock_T **blp;
|
|
|
4522 char_u *s;
|
|
|
4523 {
|
|
|
4524 char_u *sc;
|
|
|
4525
|
|
447
|
4526 sc = (char_u *)getroom(blp, STRLEN(s) + 1, FALSE);
|
|
300
|
4527 if (sc != NULL)
|
|
|
4528 STRCPY(sc, s);
|
|
|
4529 return sc;
|
|
|
4530 }
|
|
|
4531
|
|
|
4532
|
|
|
4533 /*
|
|
|
4534 * Free the list of allocated sblock_T.
|
|
236
|
4535 */
|
|
|
4536 static void
|
|
300
|
4537 free_blocks(bl)
|
|
|
4538 sblock_T *bl;
|
|
236
|
4539 {
|
|
300
|
4540 sblock_T *next;
|
|
236
|
4541
|
|
300
|
4542 while (bl != NULL)
|
|
236
|
4543 {
|
|
300
|
4544 next = bl->sb_next;
|
|
|
4545 vim_free(bl);
|
|
|
4546 bl = next;
|
|
236
|
4547 }
|
|
|
4548 }
|
|
|
4549
|
|
|
4550 /*
|
|
300
|
4551 * Allocate the root of a word tree.
|
|
236
|
4552 */
|
|
300
|
4553 static wordnode_T *
|
|
|
4554 wordtree_alloc(blp)
|
|
|
4555 sblock_T **blp;
|
|
236
|
4556 {
|
|
447
|
4557 return (wordnode_T *)getroom(blp, sizeof(wordnode_T), TRUE);
|
|
236
|
4558 }
|
|
|
4559
|
|
|
4560 /*
|
|
300
|
4561 * Store a word in the tree(s).
|
|
390
|
4562 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
4563 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
4564 * used to find suggestions.
|
|
300
|
4565 * For a keep-case word also store it in the keep-case tree.
|
|
390
|
4566 * When "pfxlist" is not NULL store the word for each postponed prefix ID.
|
|
236
|
4567 */
|
|
|
4568 static int
|
|
339
|
4569 store_word(word, spin, flags, region, pfxlist)
|
|
300
|
4570 char_u *word;
|
|
|
4571 spellinfo_T *spin;
|
|
307
|
4572 int flags; /* extra flags, WF_BANNED */
|
|
316
|
4573 int region; /* supported region(s) */
|
|
339
|
4574 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
236
|
4575 {
|
|
300
|
4576 int len = STRLEN(word);
|
|
|
4577 int ct = captype(word, word + len);
|
|
|
4578 char_u foldword[MAXWLEN];
|
|
339
|
4579 int res = OK;
|
|
|
4580 char_u *p;
|
|
236
|
4581
|
|
323
|
4582 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
4583 for (p = pfxlist; res == OK; ++p)
|
|
|
4584 {
|
|
|
4585 res = tree_add_word(foldword, spin->si_foldroot, ct | flags,
|
|
|
4586 region, p == NULL ? 0 : *p, &spin->si_blocks);
|
|
|
4587 if (p == NULL || *p == NUL)
|
|
|
4588 break;
|
|
|
4589 }
|
|
334
|
4590 ++spin->si_foldwcount;
|
|
307
|
4591
|
|
|
4592 if (res == OK && (ct == WF_KEEPCAP || flags & WF_KEEPCAP))
|
|
334
|
4593 {
|
|
339
|
4594 for (p = pfxlist; res == OK; ++p)
|
|
|
4595 {
|
|
|
4596 res = tree_add_word(word, spin->si_keeproot, flags,
|
|
|
4597 region, p == NULL ? 0 : *p, &spin->si_blocks);
|
|
|
4598 if (p == NULL || *p == NUL)
|
|
|
4599 break;
|
|
|
4600 }
|
|
334
|
4601 ++spin->si_keepwcount;
|
|
|
4602 }
|
|
300
|
4603 return res;
|
|
236
|
4604 }
|
|
|
4605
|
|
|
4606 /*
|
|
300
|
4607 * Add word "word" to a word tree at "root".
|
|
366
|
4608 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
4609 * "rare" and "region" is the condition nr.
|
|
255
|
4610 * Returns FAIL when out of memory.
|
|
236
|
4611 */
|
|
255
|
4612 static int
|
|
339
|
4613 tree_add_word(word, root, flags, region, prefixID, blp)
|
|
300
|
4614 char_u *word;
|
|
|
4615 wordnode_T *root;
|
|
|
4616 int flags;
|
|
|
4617 int region;
|
|
339
|
4618 int prefixID;
|
|
300
|
4619 sblock_T **blp;
|
|
236
|
4620 {
|
|
300
|
4621 wordnode_T *node = root;
|
|
|
4622 wordnode_T *np;
|
|
|
4623 wordnode_T **prev = NULL;
|
|
|
4624 int i;
|
|
255
|
4625
|
|
300
|
4626 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
4627 for (i = 0; ; ++i)
|
|
255
|
4628 {
|
|
300
|
4629 /* Look for the sibling that has the same character. They are sorted
|
|
|
4630 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
4631 * higher byte value. For zero bytes (end of word) the sorting is
|
|
|
4632 * done on flags and then on prefixID
|
|
300
|
4633 */
|
|
339
|
4634 while (node != NULL
|
|
|
4635 && (node->wn_byte < word[i]
|
|
|
4636 || (node->wn_byte == NUL
|
|
|
4637 && (flags < 0
|
|
|
4638 ? node->wn_prefixID < prefixID
|
|
390
|
4639 : node->wn_flags < (flags & WN_MASK)
|
|
|
4640 || (node->wn_flags == (flags & WN_MASK)
|
|
339
|
4641 && node->wn_prefixID < prefixID)))))
|
|
236
|
4642 {
|
|
300
|
4643 prev = &node->wn_sibling;
|
|
|
4644 node = *prev;
|
|
236
|
4645 }
|
|
339
|
4646 if (node == NULL
|
|
|
4647 || node->wn_byte != word[i]
|
|
|
4648 || (word[i] == NUL
|
|
|
4649 && (flags < 0
|
|
390
|
4650 || node->wn_flags != (flags & WN_MASK)
|
|
339
|
4651 || node->wn_prefixID != prefixID)))
|
|
255
|
4652 {
|
|
300
|
4653 /* Allocate a new node. */
|
|
447
|
4654 np = (wordnode_T *)getroom(blp, sizeof(wordnode_T), TRUE);
|
|
300
|
4655 if (np == NULL)
|
|
|
4656 return FAIL;
|
|
|
4657 np->wn_byte = word[i];
|
|
|
4658 *prev = np;
|
|
|
4659 np->wn_sibling = node;
|
|
|
4660 node = np;
|
|
255
|
4661 }
|
|
300
|
4662
|
|
|
4663 if (word[i] == NUL)
|
|
|
4664 {
|
|
|
4665 node->wn_flags = flags;
|
|
|
4666 node->wn_region |= region;
|
|
339
|
4667 node->wn_prefixID = prefixID;
|
|
300
|
4668 break;
|
|
|
4669 }
|
|
|
4670 prev = &node->wn_child;
|
|
|
4671 node = *prev;
|
|
255
|
4672 }
|
|
|
4673
|
|
|
4674 return OK;
|
|
236
|
4675 }
|
|
|
4676
|
|
|
4677 /*
|
|
300
|
4678 * Compress a tree: find tails that are identical and can be shared.
|
|
|
4679 */
|
|
|
4680 static void
|
|
310
|
4681 wordtree_compress(root, spin)
|
|
300
|
4682 wordnode_T *root;
|
|
310
|
4683 spellinfo_T *spin;
|
|
300
|
4684 {
|
|
|
4685 hashtab_T ht;
|
|
|
4686 int n;
|
|
|
4687 int tot = 0;
|
|
|
4688
|
|
|
4689 if (root != NULL)
|
|
|
4690 {
|
|
|
4691 hash_init(&ht);
|
|
|
4692 n = node_compress(root, &ht, &tot);
|
|
310
|
4693 if (spin->si_verbose || p_verbose > 2)
|
|
|
4694 {
|
|
|
4695 if (!spin->si_verbose)
|
|
|
4696 verbose_enter();
|
|
|
4697 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
300
|
4698 n, tot, (tot - n) * 100 / tot);
|
|
310
|
4699 if (p_verbose > 2)
|
|
|
4700 verbose_leave();
|
|
|
4701 }
|
|
300
|
4702 hash_clear(&ht);
|
|
|
4703 }
|
|
|
4704 }
|
|
|
4705
|
|
|
4706 /*
|
|
|
4707 * Compress a node, its siblings and its children, depth first.
|
|
|
4708 * Returns the number of compressed nodes.
|
|
236
|
4709 */
|
|
255
|
4710 static int
|
|
300
|
4711 node_compress(node, ht, tot)
|
|
|
4712 wordnode_T *node;
|
|
|
4713 hashtab_T *ht;
|
|
|
4714 int *tot; /* total count of nodes before compressing,
|
|
|
4715 incremented while going through the tree */
|
|
236
|
4716 {
|
|
300
|
4717 wordnode_T *np;
|
|
|
4718 wordnode_T *tp;
|
|
|
4719 wordnode_T *child;
|
|
|
4720 hash_T hash;
|
|
236
|
4721 hashitem_T *hi;
|
|
300
|
4722 int len = 0;
|
|
|
4723 unsigned nr, n;
|
|
|
4724 int compressed = 0;
|
|
236
|
4725
|
|
300
|
4726 /*
|
|
|
4727 * Go through the list of siblings. Compress each child and then try
|
|
|
4728 * finding an identical child to replace it.
|
|
|
4729 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
4730 * but the whole list of siblings, of which the node is the first.
|
|
|
4731 */
|
|
|
4732 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4733 {
|
|
300
|
4734 ++len;
|
|
|
4735 if ((child = np->wn_child) != NULL)
|
|
|
4736 {
|
|
346
|
4737 /* Compress the child. This fills hashkey. */
|
|
300
|
4738 compressed += node_compress(child, ht, tot);
|
|
|
4739
|
|
|
4740 /* Try to find an identical child. */
|
|
346
|
4741 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
4742 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
4743 tp = NULL;
|
|
|
4744 if (!HASHITEM_EMPTY(hi))
|
|
|
4745 {
|
|
|
4746 /* There are children with an identical hash value. Now check
|
|
|
4747 * if there is one that is really identical. */
|
|
346
|
4748 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
4749 if (node_equal(child, tp))
|
|
|
4750 {
|
|
|
4751 /* Found one! Now use that child in place of the
|
|
|
4752 * current one. This means the current child is
|
|
|
4753 * dropped from the tree. */
|
|
|
4754 np->wn_child = tp;
|
|
|
4755 ++compressed;
|
|
|
4756 break;
|
|
|
4757 }
|
|
|
4758 if (tp == NULL)
|
|
|
4759 {
|
|
|
4760 /* No other child with this hash value equals the child of
|
|
|
4761 * the node, add it to the linked list after the first
|
|
|
4762 * item. */
|
|
|
4763 tp = HI2WN(hi);
|
|
346
|
4764 child->wn_u2.next = tp->wn_u2.next;
|
|
|
4765 tp->wn_u2.next = child;
|
|
300
|
4766 }
|
|
|
4767 }
|
|
|
4768 else
|
|
|
4769 /* No other child has this hash value, add it to the
|
|
|
4770 * hashtable. */
|
|
346
|
4771 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
4772 }
|
|
236
|
4773 }
|
|
300
|
4774 *tot += len;
|
|
|
4775
|
|
|
4776 /*
|
|
|
4777 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
4778 * find a lookalike node. This must be done after compressing the sibling
|
|
|
4779 * list, otherwise the hash key would become invalid by the compression.
|
|
|
4780 */
|
|
346
|
4781 node->wn_u1.hashkey[0] = len;
|
|
300
|
4782 nr = 0;
|
|
|
4783 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
4784 {
|
|
300
|
4785 if (np->wn_byte == NUL)
|
|
339
|
4786 /* end node: use wn_flags, wn_region and wn_prefixID */
|
|
|
4787 n = np->wn_flags + (np->wn_region << 8) + (np->wn_prefixID << 16);
|
|
300
|
4788 else
|
|
|
4789 /* byte node: use the byte value and the child pointer */
|
|
|
4790 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
4791 nr = nr * 101 + n;
|
|
236
|
4792 }
|
|
300
|
4793
|
|
|
4794 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
4795 n = nr & 0xff;
|
|
346
|
4796 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
4797 n = (nr >> 8) & 0xff;
|
|
346
|
4798 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
4799 n = (nr >> 16) & 0xff;
|
|
346
|
4800 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
4801 n = (nr >> 24) & 0xff;
|
|
346
|
4802 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
4803 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
4804
|
|
|
4805 return compressed;
|
|
|
4806 }
|
|
|
4807
|
|
|
4808 /*
|
|
|
4809 * Return TRUE when two nodes have identical siblings and children.
|
|
|
4810 */
|
|
|
4811 static int
|
|
|
4812 node_equal(n1, n2)
|
|
|
4813 wordnode_T *n1;
|
|
|
4814 wordnode_T *n2;
|
|
|
4815 {
|
|
|
4816 wordnode_T *p1;
|
|
|
4817 wordnode_T *p2;
|
|
|
4818
|
|
|
4819 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
4820 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
4821 if (p1->wn_byte != p2->wn_byte
|
|
|
4822 || (p1->wn_byte == NUL
|
|
|
4823 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
4824 || p1->wn_region != p2->wn_region
|
|
|
4825 || p1->wn_prefixID != p2->wn_prefixID)
|
|
300
|
4826 : (p1->wn_child != p2->wn_child)))
|
|
|
4827 break;
|
|
|
4828
|
|
|
4829 return p1 == NULL && p2 == NULL;
|
|
236
|
4830 }
|
|
|
4831
|
|
|
4832 /*
|
|
|
4833 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
4834 */
|
|
255
|
4835 void
|
|
236
|
4836 put_bytes(fd, nr, len)
|
|
|
4837 FILE *fd;
|
|
|
4838 long_u nr;
|
|
|
4839 int len;
|
|
|
4840 {
|
|
|
4841 int i;
|
|
|
4842
|
|
|
4843 for (i = len - 1; i >= 0; --i)
|
|
|
4844 putc((int)(nr >> (i * 8)), fd);
|
|
|
4845 }
|
|
|
4846
|
|
323
|
4847 static int
|
|
|
4848 #ifdef __BORLANDC__
|
|
|
4849 _RTLENTRYF
|
|
|
4850 #endif
|
|
|
4851 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
4852
|
|
|
4853 /*
|
|
|
4854 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
4855 */
|
|
|
4856 static int
|
|
|
4857 #ifdef __BORLANDC__
|
|
|
4858 _RTLENTRYF
|
|
|
4859 #endif
|
|
|
4860 rep_compare(s1, s2)
|
|
|
4861 const void *s1;
|
|
|
4862 const void *s2;
|
|
|
4863 {
|
|
|
4864 fromto_T *p1 = (fromto_T *)s1;
|
|
|
4865 fromto_T *p2 = (fromto_T *)s2;
|
|
|
4866
|
|
|
4867 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
4868 }
|
|
|
4869
|
|
236
|
4870 /*
|
|
|
4871 * Write the Vim spell file "fname".
|
|
|
4872 */
|
|
|
4873 static void
|
|
316
|
4874 write_vim_spell(fname, spin)
|
|
236
|
4875 char_u *fname;
|
|
300
|
4876 spellinfo_T *spin;
|
|
236
|
4877 {
|
|
300
|
4878 FILE *fd;
|
|
|
4879 int regionmask;
|
|
236
|
4880 int round;
|
|
300
|
4881 wordnode_T *tree;
|
|
|
4882 int nodecount;
|
|
323
|
4883 int i;
|
|
|
4884 int l;
|
|
|
4885 garray_T *gap;
|
|
|
4886 fromto_T *ftp;
|
|
|
4887 char_u *p;
|
|
|
4888 int rr;
|
|
236
|
4889
|
|
310
|
4890 fd = mch_fopen((char *)fname, "w");
|
|
300
|
4891 if (fd == NULL)
|
|
236
|
4892 {
|
|
|
4893 EMSG2(_(e_notopen), fname);
|
|
|
4894 return;
|
|
|
4895 }
|
|
|
4896
|
|
255
|
4897 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
339
|
4898 * <charflagslen> <charflags>
|
|
|
4899 * <fcharslen> <fchars>
|
|
366
|
4900 * <midwordlen> <midword>
|
|
339
|
4901 * <prefcondcnt> <prefcond> ... */
|
|
300
|
4902
|
|
|
4903 /* <fileID> */
|
|
|
4904 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
4905 EMSG(_(e_write));
|
|
236
|
4906
|
|
|
4907 /* write the region names if there is more than one */
|
|
316
|
4908 if (spin->si_region_count > 1)
|
|
236
|
4909 {
|
|
316
|
4910 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
|
4911 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
|
4912 (size_t)1, fd);
|
|
|
4913 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
4914 }
|
|
|
4915 else
|
|
|
4916 {
|
|
300
|
4917 putc(0, fd);
|
|
|
4918 regionmask = 0;
|
|
236
|
4919 }
|
|
|
4920
|
|
323
|
4921 /*
|
|
|
4922 * Write the table with character flags and table for case folding.
|
|
260
|
4923 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
|
4924 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
4925 * 'encoding'.
|
|
|
4926 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
4927 * the table (avoids that it conflicts). File is shorter too.
|
|
|
4928 */
|
|
|
4929 if (spin->si_ascii || spin->si_add)
|
|
260
|
4930 {
|
|
300
|
4931 putc(0, fd);
|
|
|
4932 putc(0, fd);
|
|
|
4933 putc(0, fd);
|
|
260
|
4934 }
|
|
|
4935 else
|
|
300
|
4936 write_spell_chartab(fd);
|
|
255
|
4937
|
|
366
|
4938
|
|
|
4939 if (spin->si_midword == NULL)
|
|
|
4940 put_bytes(fd, 0L, 2); /* <midwordlen> */
|
|
|
4941 else
|
|
|
4942 {
|
|
|
4943 i = STRLEN(spin->si_midword);
|
|
|
4944 put_bytes(fd, (long_u)i, 2); /* <midwordlen> */
|
|
|
4945 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
4946 }
|
|
|
4947
|
|
|
4948
|
|
339
|
4949 /* Write the prefix conditions. */
|
|
|
4950 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
4951
|
|
366
|
4952 /* <SUGGEST> : <repcount> <rep> ...
|
|
|
4953 * <salflags> <salcount> <sal> ...
|
|
|
4954 * <maplen> <mapstr> */
|
|
|
4955
|
|
323
|
4956 /* Sort the REP items. */
|
|
|
4957 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
4958 sizeof(fromto_T), rep_compare);
|
|
|
4959
|
|
375
|
4960 /* round 1: REP items
|
|
|
4961 * round 2: SAL items (unless SOFO is used) */
|
|
323
|
4962 for (round = 1; round <= 2; ++round)
|
|
|
4963 {
|
|
|
4964 if (round == 1)
|
|
|
4965 gap = &spin->si_rep;
|
|
|
4966 else
|
|
|
4967 {
|
|
|
4968 gap = &spin->si_sal;
|
|
|
4969
|
|
|
4970 i = 0;
|
|
|
4971 if (spin->si_followup)
|
|
|
4972 i |= SAL_F0LLOWUP;
|
|
|
4973 if (spin->si_collapse)
|
|
|
4974 i |= SAL_COLLAPSE;
|
|
|
4975 if (spin->si_rem_accents)
|
|
|
4976 i |= SAL_REM_ACCENTS;
|
|
375
|
4977 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
4978 i |= SAL_SOFO;
|
|
323
|
4979 putc(i, fd); /* <salflags> */
|
|
375
|
4980 if (i & SAL_SOFO)
|
|
|
4981 break;
|
|
323
|
4982 }
|
|
|
4983
|
|
|
4984 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
4985 for (i = 0; i < gap->ga_len; ++i)
|
|
|
4986 {
|
|
|
4987 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
4988 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
4989 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
4990 for (rr = 1; rr <= 2; ++rr)
|
|
|
4991 {
|
|
|
4992 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
4993 l = STRLEN(p);
|
|
|
4994 putc(l, fd);
|
|
|
4995 fwrite(p, l, (size_t)1, fd);
|
|
|
4996 }
|
|
|
4997 }
|
|
|
4998 }
|
|
|
4999
|
|
375
|
5000 /* SOFOFROM and SOFOTO */
|
|
|
5001 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
5002 {
|
|
|
5003 put_bytes(fd, 1L, 2); /* <salcount> */
|
|
|
5004
|
|
|
5005 l = STRLEN(spin->si_sofofr);
|
|
|
5006 put_bytes(fd, (long_u)l, 2); /* <salfromlen> */
|
|
|
5007 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <salfrom> */
|
|
|
5008
|
|
|
5009 l = STRLEN(spin->si_sofoto);
|
|
|
5010 put_bytes(fd, (long_u)l, 2); /* <saltolen> */
|
|
|
5011 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <salto> */
|
|
|
5012 }
|
|
|
5013
|
|
323
|
5014 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
|
5015 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
|
5016 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
|
5017 (size_t)1, fd);
|
|
302
|
5018
|
|
236
|
5019 /*
|
|
339
|
5020 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
5021 */
|
|
323
|
5022 spin->si_memtot = 0;
|
|
339
|
5023 for (round = 1; round <= 3; ++round)
|
|
236
|
5024 {
|
|
339
|
5025 if (round == 1)
|
|
|
5026 tree = spin->si_foldroot;
|
|
|
5027 else if (round == 2)
|
|
|
5028 tree = spin->si_keeproot;
|
|
|
5029 else
|
|
|
5030 tree = spin->si_prefroot;
|
|
236
|
5031
|
|
346
|
5032 /* Clear the index and wnode fields in the tree. */
|
|
|
5033 clear_node(tree);
|
|
|
5034
|
|
300
|
5035 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
5036 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
5037 * nodes. */
|
|
346
|
5038 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
5039
|
|
300
|
5040 /* number of nodes in 4 bytes */
|
|
|
5041 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
5042 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
5043
|
|
300
|
5044 /* Write the nodes. */
|
|
346
|
5045 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
5046 }
|
|
|
5047
|
|
300
|
5048 fclose(fd);
|
|
236
|
5049 }
|
|
|
5050
|
|
|
5051 /*
|
|
346
|
5052 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
5053 * children. This is needed because they are a union with other items to save
|
|
|
5054 * space.
|
|
|
5055 */
|
|
|
5056 static void
|
|
|
5057 clear_node(node)
|
|
|
5058 wordnode_T *node;
|
|
|
5059 {
|
|
|
5060 wordnode_T *np;
|
|
|
5061
|
|
|
5062 if (node != NULL)
|
|
|
5063 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5064 {
|
|
|
5065 np->wn_u1.index = 0;
|
|
|
5066 np->wn_u2.wnode = NULL;
|
|
|
5067
|
|
|
5068 if (np->wn_byte != NUL)
|
|
|
5069 clear_node(np->wn_child);
|
|
|
5070 }
|
|
|
5071 }
|
|
|
5072
|
|
|
5073
|
|
|
5074 /*
|
|
300
|
5075 * Dump a word tree at node "node".
|
|
|
5076 *
|
|
|
5077 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
5078 * byte recursively write the children.
|
|
|
5079 *
|
|
|
5080 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
5081 * are made about the indexes (so that we don't have to write them in the
|
|
|
5082 * file).
|
|
236
|
5083 *
|
|
300
|
5084 * Returns the number of nodes used.
|
|
236
|
5085 */
|
|
300
|
5086 static int
|
|
346
|
5087 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
5088 FILE *fd; /* NULL when only counting */
|
|
300
|
5089 wordnode_T *node;
|
|
|
5090 int index;
|
|
|
5091 int regionmask;
|
|
339
|
5092 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
5093 {
|
|
300
|
5094 int newindex = index;
|
|
|
5095 int siblingcount = 0;
|
|
|
5096 wordnode_T *np;
|
|
236
|
5097 int flags;
|
|
300
|
5098
|
|
|
5099 /* If "node" is zero the tree is empty. */
|
|
|
5100 if (node == NULL)
|
|
|
5101 return 0;
|
|
|
5102
|
|
|
5103 /* Store the index where this node is written. */
|
|
346
|
5104 node->wn_u1.index = index;
|
|
236
|
5105
|
|
300
|
5106 /* Count the number of siblings. */
|
|
|
5107 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5108 ++siblingcount;
|
|
236
|
5109
|
|
300
|
5110 /* Write the sibling count. */
|
|
|
5111 if (fd != NULL)
|
|
|
5112 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
5113
|
|
300
|
5114 /* Write each sibling byte and optionally extra info. */
|
|
|
5115 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
5116 {
|
|
300
|
5117 if (np->wn_byte == 0)
|
|
|
5118 {
|
|
|
5119 if (fd != NULL)
|
|
|
5120 {
|
|
339
|
5121 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
5122 if (prefixtree)
|
|
300
|
5123 {
|
|
339
|
5124 /* In PREFIXTREE write the required prefixID and the
|
|
390
|
5125 * associated condition nr (stored in wn_region). The
|
|
|
5126 * byte value is misused to store the "rare" and "not
|
|
|
5127 * combining" flags */
|
|
455
|
5128 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
5129 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5130 else
|
|
|
5131 {
|
|
390
|
5132 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
5133 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
5134 }
|
|
339
|
5135 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
5136 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
5137 }
|
|
|
5138 else
|
|
|
5139 {
|
|
339
|
5140 /* For word trees we write the flag/region items. */
|
|
|
5141 flags = np->wn_flags;
|
|
|
5142 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
5143 flags |= WF_REGION;
|
|
|
5144 if (np->wn_prefixID != 0)
|
|
|
5145 flags |= WF_PFX;
|
|
|
5146 if (flags == 0)
|
|
|
5147 {
|
|
|
5148 /* word without flags or region */
|
|
|
5149 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
5150 }
|
|
|
5151 else
|
|
|
5152 {
|
|
390
|
5153 if (np->wn_flags >= 0x100)
|
|
|
5154 {
|
|
|
5155 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
5156 putc(flags, fd); /* <flags> */
|
|
|
5157 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
5158 }
|
|
|
5159 else
|
|
|
5160 {
|
|
|
5161 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
5162 putc(flags, fd); /* <flags> */
|
|
|
5163 }
|
|
339
|
5164 if (flags & WF_REGION)
|
|
|
5165 putc(np->wn_region, fd); /* <region> */
|
|
|
5166 if (flags & WF_PFX)
|
|
|
5167 putc(np->wn_prefixID, fd); /* <prefixID> */
|
|
|
5168 }
|
|
300
|
5169 }
|
|
|
5170 }
|
|
|
5171 }
|
|
|
5172 else
|
|
|
5173 {
|
|
346
|
5174 if (np->wn_child->wn_u1.index != 0
|
|
|
5175 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
5176 {
|
|
|
5177 /* The child is written elsewhere, write the reference. */
|
|
|
5178 if (fd != NULL)
|
|
|
5179 {
|
|
|
5180 putc(BY_INDEX, fd); /* <byte> */
|
|
|
5181 /* <nodeidx> */
|
|
346
|
5182 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
5183 }
|
|
|
5184 }
|
|
346
|
5185 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
5186 /* We will write the child below and give it an index. */
|
|
346
|
5187 np->wn_child->wn_u2.wnode = node;
|
|
236
|
5188
|
|
300
|
5189 if (fd != NULL)
|
|
|
5190 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
5191 {
|
|
|
5192 EMSG(_(e_write));
|
|
|
5193 return 0;
|
|
|
5194 }
|
|
|
5195 }
|
|
236
|
5196 }
|
|
|
5197
|
|
300
|
5198 /* Space used in the array when reading: one for each sibling and one for
|
|
|
5199 * the count. */
|
|
|
5200 newindex += siblingcount + 1;
|
|
249
|
5201
|
|
300
|
5202 /* Recursively dump the children of each sibling. */
|
|
|
5203 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
5204 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
5205 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
5206 prefixtree);
|
|
249
|
5207
|
|
300
|
5208 return newindex;
|
|
236
|
5209 }
|
|
|
5210
|
|
|
5211
|
|
|
5212 /*
|
|
310
|
5213 * ":mkspell [-ascii] outfile infile ..."
|
|
|
5214 * ":mkspell [-ascii] addfile"
|
|
236
|
5215 */
|
|
|
5216 void
|
|
|
5217 ex_mkspell(eap)
|
|
|
5218 exarg_T *eap;
|
|
|
5219 {
|
|
|
5220 int fcount;
|
|
|
5221 char_u **fnames;
|
|
310
|
5222 char_u *arg = eap->arg;
|
|
|
5223 int ascii = FALSE;
|
|
|
5224
|
|
|
5225 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
5226 {
|
|
|
5227 ascii = TRUE;
|
|
|
5228 arg = skipwhite(arg + 6);
|
|
|
5229 }
|
|
|
5230
|
|
|
5231 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
5232 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
5233 {
|
|
323
|
5234 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
5235 FreeWild(fcount, fnames);
|
|
|
5236 }
|
|
|
5237 }
|
|
|
5238
|
|
|
5239 /*
|
|
|
5240 * Create a Vim spell file from one or more word lists.
|
|
|
5241 * "fnames[0]" is the output file name.
|
|
|
5242 * "fnames[fcount - 1]" is the last input file name.
|
|
|
5243 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
5244 * and ".spl" is appended to make the output file name.
|
|
|
5245 */
|
|
|
5246 static void
|
|
323
|
5247 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
5248 int fcount;
|
|
|
5249 char_u **fnames;
|
|
|
5250 int ascii; /* -ascii argument given */
|
|
|
5251 int overwrite; /* overwrite existing output file */
|
|
323
|
5252 int added_word; /* invoked through "zg" */
|
|
310
|
5253 {
|
|
236
|
5254 char_u fname[MAXPATHL];
|
|
|
5255 char_u wfname[MAXPATHL];
|
|
310
|
5256 char_u **innames;
|
|
|
5257 int incount;
|
|
236
|
5258 afffile_T *(afile[8]);
|
|
|
5259 int i;
|
|
|
5260 int len;
|
|
|
5261 struct stat st;
|
|
255
|
5262 int error = FALSE;
|
|
300
|
5263 spellinfo_T spin;
|
|
|
5264
|
|
|
5265 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
5266 spin.si_verbose = !added_word;
|
|
310
|
5267 spin.si_ascii = ascii;
|
|
323
|
5268 spin.si_followup = TRUE;
|
|
|
5269 spin.si_rem_accents = TRUE;
|
|
|
5270 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
5271 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
5272 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
5273 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
5274
|
|
|
5275 /* default: fnames[0] is output file, following are input files */
|
|
|
5276 innames = &fnames[1];
|
|
|
5277 incount = fcount - 1;
|
|
|
5278
|
|
|
5279 if (fcount >= 1)
|
|
240
|
5280 {
|
|
310
|
5281 len = STRLEN(fnames[0]);
|
|
|
5282 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
5283 {
|
|
|
5284 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
5285 * "path/en.latin1.add.spl". */
|
|
|
5286 innames = &fnames[0];
|
|
|
5287 incount = 1;
|
|
|
5288 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
5289 }
|
|
366
|
5290 else if (fcount == 1)
|
|
|
5291 {
|
|
|
5292 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
5293 innames = &fnames[0];
|
|
|
5294 incount = 1;
|
|
|
5295 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5296 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5297 }
|
|
310
|
5298 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
5299 {
|
|
|
5300 /* Name ends in ".spl", use as the file name. */
|
|
323
|
5301 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
5302 }
|
|
|
5303 else
|
|
|
5304 /* Name should be language, make the file name from it. */
|
|
|
5305 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
5306 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
5307
|
|
|
5308 /* Check for .ascii.spl. */
|
|
|
5309 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
5310 spin.si_ascii = TRUE;
|
|
|
5311
|
|
|
5312 /* Check for .add.spl. */
|
|
|
5313 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
5314 spin.si_add = TRUE;
|
|
240
|
5315 }
|
|
|
5316
|
|
310
|
5317 if (incount <= 0)
|
|
236
|
5318 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
5319 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
5320 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
5321 else if (incount > 8)
|
|
236
|
5322 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
5323 else
|
|
|
5324 {
|
|
|
5325 /* Check for overwriting before doing things that may take a lot of
|
|
|
5326 * time. */
|
|
310
|
5327 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
5328 {
|
|
|
5329 EMSG(_(e_exists));
|
|
310
|
5330 return;
|
|
236
|
5331 }
|
|
310
|
5332 if (mch_isdir(wfname))
|
|
236
|
5333 {
|
|
310
|
5334 EMSG2(_(e_isadir2), wfname);
|
|
|
5335 return;
|
|
236
|
5336 }
|
|
|
5337
|
|
|
5338 /*
|
|
|
5339 * Init the aff and dic pointers.
|
|
|
5340 * Get the region names if there are more than 2 arguments.
|
|
|
5341 */
|
|
310
|
5342 for (i = 0; i < incount; ++i)
|
|
236
|
5343 {
|
|
310
|
5344 afile[i] = NULL;
|
|
300
|
5345
|
|
316
|
5346 if (incount > 1)
|
|
236
|
5347 {
|
|
310
|
5348 len = STRLEN(innames[i]);
|
|
|
5349 if (STRLEN(gettail(innames[i])) < 5
|
|
|
5350 || innames[i][len - 3] != '_')
|
|
236
|
5351 {
|
|
310
|
5352 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
5353 return;
|
|
236
|
5354 }
|
|
316
|
5355 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
5356 spin.si_region_name[i * 2 + 1] =
|
|
|
5357 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
5358 }
|
|
|
5359 }
|
|
316
|
5360 spin.si_region_count = incount;
|
|
236
|
5361
|
|
300
|
5362 spin.si_foldroot = wordtree_alloc(&spin.si_blocks);
|
|
|
5363 spin.si_keeproot = wordtree_alloc(&spin.si_blocks);
|
|
339
|
5364 spin.si_prefroot = wordtree_alloc(&spin.si_blocks);
|
|
|
5365 if (spin.si_foldroot == NULL
|
|
|
5366 || spin.si_keeproot == NULL
|
|
|
5367 || spin.si_prefroot == NULL)
|
|
300
|
5368 {
|
|
|
5369 error = TRUE;
|
|
310
|
5370 return;
|
|
300
|
5371 }
|
|
|
5372
|
|
351
|
5373 /* When not producing a .add.spl file clear the character table when
|
|
|
5374 * we encounter one in the .aff file. This means we dump the current
|
|
|
5375 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
5376 * better than guessing the contents, the table will match a
|
|
|
5377 * previously loaded spell file. */
|
|
|
5378 if (!spin.si_add)
|
|
|
5379 spin.si_clear_chartab = TRUE;
|
|
|
5380
|
|
236
|
5381 /*
|
|
|
5382 * Read all the .aff and .dic files.
|
|
|
5383 * Text is converted to 'encoding'.
|
|
300
|
5384 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
5385 */
|
|
310
|
5386 for (i = 0; i < incount && !error; ++i)
|
|
236
|
5387 {
|
|
300
|
5388 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
5389 spin.si_region = 1 << i;
|
|
|
5390
|
|
|
5391 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
5392 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
5393 {
|
|
|
5394 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
5395 * "SET" line. */
|
|
310
|
5396 afile[i] = spell_read_aff(fname, &spin);
|
|
|
5397 if (afile[i] == NULL)
|
|
300
|
5398 error = TRUE;
|
|
|
5399 else
|
|
|
5400 {
|
|
|
5401 /* Read the .dic file and store the words in the trees. */
|
|
|
5402 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
5403 innames[i]);
|
|
|
5404 if (spell_read_dic(fname, &spin, afile[i]) == FAIL)
|
|
300
|
5405 error = TRUE;
|
|
|
5406 }
|
|
|
5407 }
|
|
|
5408 else
|
|
|
5409 {
|
|
|
5410 /* No .aff file, try reading the file as a word list. Store
|
|
|
5411 * the words in the trees. */
|
|
310
|
5412 if (spell_read_wordfile(innames[i], &spin) == FAIL)
|
|
300
|
5413 error = TRUE;
|
|
|
5414 }
|
|
236
|
5415
|
|
310
|
5416 #ifdef FEAT_MBYTE
|
|
236
|
5417 /* Free any conversion stuff. */
|
|
300
|
5418 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
5419 #endif
|
|
236
|
5420 }
|
|
|
5421
|
|
300
|
5422 if (!error)
|
|
236
|
5423 {
|
|
|
5424 /*
|
|
300
|
5425 * Remove the dummy NUL from the start of the tree root.
|
|
236
|
5426 */
|
|
300
|
5427 spin.si_foldroot = spin.si_foldroot->wn_sibling;
|
|
|
5428 spin.si_keeproot = spin.si_keeproot->wn_sibling;
|
|
339
|
5429 spin.si_prefroot = spin.si_prefroot->wn_sibling;
|
|
236
|
5430
|
|
|
5431 /*
|
|
300
|
5432 * Combine tails in the tree.
|
|
236
|
5433 */
|
|
323
|
5434 if (!added_word || p_verbose > 2)
|
|
310
|
5435 {
|
|
323
|
5436 if (added_word)
|
|
310
|
5437 verbose_enter();
|
|
|
5438 MSG(_("Compressing word tree..."));
|
|
|
5439 out_flush();
|
|
323
|
5440 if (added_word)
|
|
310
|
5441 verbose_leave();
|
|
|
5442 }
|
|
|
5443 wordtree_compress(spin.si_foldroot, &spin);
|
|
|
5444 wordtree_compress(spin.si_keeproot, &spin);
|
|
339
|
5445 wordtree_compress(spin.si_prefroot, &spin);
|
|
236
|
5446 }
|
|
|
5447
|
|
300
|
5448 if (!error)
|
|
|
5449 {
|
|
|
5450 /*
|
|
|
5451 * Write the info in the spell file.
|
|
|
5452 */
|
|
323
|
5453 if (!added_word || p_verbose > 2)
|
|
310
|
5454 {
|
|
323
|
5455 if (added_word)
|
|
310
|
5456 verbose_enter();
|
|
366
|
5457 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
5458 out_flush();
|
|
323
|
5459 if (added_word)
|
|
310
|
5460 verbose_leave();
|
|
|
5461 }
|
|
|
5462
|
|
316
|
5463 write_vim_spell(wfname, &spin);
|
|
310
|
5464
|
|
323
|
5465 if (!added_word || p_verbose > 2)
|
|
310
|
5466 {
|
|
323
|
5467 if (added_word)
|
|
310
|
5468 verbose_enter();
|
|
|
5469 MSG(_("Done!"));
|
|
|
5470 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
5471 spin.si_memtot);
|
|
310
|
5472 out_flush();
|
|
323
|
5473 if (added_word)
|
|
310
|
5474 verbose_leave();
|
|
|
5475 }
|
|
|
5476
|
|
|
5477 /* If the file is loaded need to reload it. */
|
|
323
|
5478 spell_reload_one(wfname, added_word);
|
|
300
|
5479 }
|
|
|
5480
|
|
|
5481 /* Free the allocated memory. */
|
|
323
|
5482 ga_clear(&spin.si_rep);
|
|
|
5483 ga_clear(&spin.si_sal);
|
|
|
5484 ga_clear(&spin.si_map);
|
|
339
|
5485 ga_clear(&spin.si_prefcond);
|
|
366
|
5486 vim_free(spin.si_midword);
|
|
375
|
5487 vim_free(spin.si_sofofr);
|
|
|
5488 vim_free(spin.si_sofoto);
|
|
300
|
5489
|
|
|
5490 /* Free the .aff file structures. */
|
|
310
|
5491 for (i = 0; i < incount; ++i)
|
|
|
5492 if (afile[i] != NULL)
|
|
|
5493 spell_free_aff(afile[i]);
|
|
339
|
5494
|
|
|
5495 /* Free all the bits and pieces at once. */
|
|
|
5496 free_blocks(spin.si_blocks);
|
|
236
|
5497 }
|
|
310
|
5498 }
|
|
|
5499
|
|
|
5500
|
|
|
5501 /*
|
|
385
|
5502 * ":[count]spellgood {word}"
|
|
|
5503 * ":[count]spellwrong {word}"
|
|
310
|
5504 */
|
|
|
5505 void
|
|
|
5506 ex_spell(eap)
|
|
|
5507 exarg_T *eap;
|
|
|
5508 {
|
|
381
|
5509 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
5510 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
5511 }
|
|
|
5512
|
|
310
|
5513 /*
|
|
|
5514 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
5515 */
|
|
|
5516 void
|
|
385
|
5517 spell_add_word(word, len, bad, index)
|
|
310
|
5518 char_u *word;
|
|
|
5519 int len;
|
|
|
5520 int bad;
|
|
385
|
5521 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
5522 'spellfile' */
|
|
310
|
5523 {
|
|
|
5524 FILE *fd;
|
|
385
|
5525 buf_T *buf = NULL;
|
|
351
|
5526 int new_spf = FALSE;
|
|
|
5527 struct stat st;
|
|
381
|
5528 char_u *fname;
|
|
385
|
5529 char_u fnamebuf[MAXPATHL];
|
|
|
5530 char_u line[MAXWLEN * 2];
|
|
|
5531 long fpos, fpos_next = 0;
|
|
|
5532 int i;
|
|
|
5533 char_u *spf;
|
|
|
5534
|
|
|
5535 if (index == 0) /* use internal wordlist */
|
|
|
5536 {
|
|
|
5537 if (int_wordlist == NULL)
|
|
|
5538 {
|
|
|
5539 int_wordlist = vim_tempname('s');
|
|
|
5540 if (int_wordlist == NULL)
|
|
381
|
5541 return;
|
|
|
5542 }
|
|
385
|
5543 fname = int_wordlist;
|
|
381
|
5544 }
|
|
310
|
5545 else
|
|
|
5546 {
|
|
381
|
5547 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
5548 if (*curbuf->b_p_spf == NUL)
|
|
|
5549 {
|
|
|
5550 init_spellfile();
|
|
|
5551 new_spf = TRUE;
|
|
|
5552 }
|
|
|
5553
|
|
|
5554 if (*curbuf->b_p_spf == NUL)
|
|
|
5555 {
|
|
|
5556 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
5557 return;
|
|
|
5558 }
|
|
|
5559
|
|
385
|
5560 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
5561 {
|
|
|
5562 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
5563 if (i == index)
|
|
|
5564 break;
|
|
|
5565 if (*spf == NUL)
|
|
|
5566 {
|
|
|
5567 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
5568 return;
|
|
|
5569 }
|
|
|
5570 }
|
|
|
5571
|
|
310
|
5572 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
5573 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
5574 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
5575 buf = NULL;
|
|
|
5576 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
5577 {
|
|
310
|
5578 EMSG(_(e_bufloaded));
|
|
381
|
5579 return;
|
|
|
5580 }
|
|
|
5581
|
|
385
|
5582 fname = fnamebuf;
|
|
|
5583 }
|
|
|
5584
|
|
|
5585 if (bad)
|
|
|
5586 {
|
|
|
5587 /* When the word also appears as good word we need to remove that one,
|
|
|
5588 * since its flags sort before the one with WF_BANNED. */
|
|
|
5589 fd = mch_fopen((char *)fname, "r");
|
|
|
5590 if (fd != NULL)
|
|
|
5591 {
|
|
|
5592 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
5593 {
|
|
|
5594 fpos = fpos_next;
|
|
|
5595 fpos_next = ftell(fd);
|
|
|
5596 if (STRNCMP(word, line, len) == 0
|
|
|
5597 && (line[len] == '/' || line[len] < ' '))
|
|
|
5598 {
|
|
|
5599 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
5600 * the start of the line. Mixing reading and writing
|
|
|
5601 * doesn't work for all systems, close the file first. */
|
|
|
5602 fclose(fd);
|
|
|
5603 fd = mch_fopen((char *)fname, "r+");
|
|
|
5604 if (fd == NULL)
|
|
|
5605 break;
|
|
|
5606 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
5607 fputc('#', fd);
|
|
|
5608 fseek(fd, fpos_next, SEEK_SET);
|
|
|
5609 }
|
|
|
5610 }
|
|
|
5611 fclose(fd);
|
|
|
5612 }
|
|
381
|
5613 }
|
|
|
5614
|
|
|
5615 fd = mch_fopen((char *)fname, "a");
|
|
|
5616 if (fd == NULL && new_spf)
|
|
|
5617 {
|
|
|
5618 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
5619 * We may need to create the "spell" directory first. We already
|
|
|
5620 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
5621 STRCPY(NameBuff, fname);
|
|
|
5622 *gettail_sep(NameBuff) = NUL;
|
|
|
5623 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
5624 {
|
|
|
5625 /* The directory doesn't exist. Try creating it and opening the
|
|
|
5626 * file again. */
|
|
|
5627 vim_mkdir(NameBuff, 0755);
|
|
|
5628 fd = mch_fopen((char *)fname, "a");
|
|
|
5629 }
|
|
|
5630 }
|
|
|
5631
|
|
|
5632 if (fd == NULL)
|
|
|
5633 EMSG2(_(e_notopen), fname);
|
|
|
5634 else
|
|
|
5635 {
|
|
|
5636 if (bad)
|
|
|
5637 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
5638 else
|
|
381
|
5639 fprintf(fd, "%.*s\n", len, word);
|
|
|
5640 fclose(fd);
|
|
|
5641
|
|
|
5642 /* Update the .add.spl file. */
|
|
|
5643 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
5644
|
|
|
5645 /* If the .add file is edited somewhere, reload it. */
|
|
|
5646 if (buf != NULL)
|
|
|
5647 buf_reload(buf);
|
|
|
5648
|
|
|
5649 redraw_all_later(NOT_VALID);
|
|
310
|
5650 }
|
|
|
5651 }
|
|
|
5652
|
|
|
5653 /*
|
|
|
5654 * Initialize 'spellfile' for the current buffer.
|
|
|
5655 */
|
|
|
5656 static void
|
|
|
5657 init_spellfile()
|
|
|
5658 {
|
|
|
5659 char_u buf[MAXPATHL];
|
|
|
5660 int l;
|
|
|
5661 slang_T *sl;
|
|
|
5662 char_u *rtp;
|
|
351
|
5663 char_u *lend;
|
|
310
|
5664
|
|
|
5665 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
5666 {
|
|
351
|
5667 /* Find the end of the language name. Exclude the region. */
|
|
|
5668 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
5669 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
5670 ;
|
|
|
5671
|
|
|
5672 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
5673 * are allowed to write. */
|
|
310
|
5674 rtp = p_rtp;
|
|
|
5675 while (*rtp != NUL)
|
|
|
5676 {
|
|
|
5677 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
5678 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
5679 if (filewritable(buf) == 2)
|
|
|
5680 {
|
|
316
|
5681 /* Use the first language name from 'spelllang' and the
|
|
|
5682 * encoding used in the first loaded .spl file. */
|
|
310
|
5683 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
5684 l = STRLEN(buf);
|
|
|
5685 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
5686 "/spell/%.*s.%s.add",
|
|
351
|
5687 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
5688 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
5689 ? (char_u *)"ascii" : spell_enc());
|
|
|
5690 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
5691 break;
|
|
|
5692 }
|
|
|
5693 }
|
|
|
5694 }
|
|
|
5695 }
|
|
236
|
5696
|
|
300
|
5697
|
|
307
|
5698 /*
|
|
|
5699 * Init the chartab used for spelling for ASCII.
|
|
|
5700 * EBCDIC is not supported!
|
|
|
5701 */
|
|
|
5702 static void
|
|
|
5703 clear_spell_chartab(sp)
|
|
|
5704 spelltab_T *sp;
|
|
|
5705 {
|
|
324
|
5706 int i;
|
|
307
|
5707
|
|
|
5708 /* Init everything to FALSE. */
|
|
|
5709 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
5710 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
5711 for (i = 0; i < 256; ++i)
|
|
324
|
5712 {
|
|
307
|
5713 sp->st_fold[i] = i;
|
|
324
|
5714 sp->st_upper[i] = i;
|
|
|
5715 }
|
|
307
|
5716
|
|
|
5717 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
5718 * that is done separately. */
|
|
|
5719 for (i = '0'; i <= '9'; ++i)
|
|
|
5720 sp->st_isw[i] = TRUE;
|
|
|
5721 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
5722 {
|
|
|
5723 sp->st_isw[i] = TRUE;
|
|
|
5724 sp->st_isu[i] = TRUE;
|
|
|
5725 sp->st_fold[i] = i + 0x20;
|
|
|
5726 }
|
|
|
5727 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
5728 {
|
|
307
|
5729 sp->st_isw[i] = TRUE;
|
|
324
|
5730 sp->st_upper[i] = i - 0x20;
|
|
|
5731 }
|
|
307
|
5732 }
|
|
|
5733
|
|
|
5734 /*
|
|
|
5735 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
5736 * Called once while starting up and when 'encoding' changes.
|
|
|
5737 * The default is to use isalpha(), but the spell file should define the word
|
|
|
5738 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
5739 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
5740 */
|
|
|
5741 void
|
|
|
5742 init_spell_chartab()
|
|
|
5743 {
|
|
|
5744 int i;
|
|
|
5745
|
|
|
5746 did_set_spelltab = FALSE;
|
|
|
5747 clear_spell_chartab(&spelltab);
|
|
|
5748 #ifdef FEAT_MBYTE
|
|
|
5749 if (enc_dbcs)
|
|
|
5750 {
|
|
|
5751 /* DBCS: assume double-wide characters are word characters. */
|
|
|
5752 for (i = 128; i <= 255; ++i)
|
|
|
5753 if (MB_BYTE2LEN(i) == 2)
|
|
|
5754 spelltab.st_isw[i] = TRUE;
|
|
|
5755 }
|
|
324
|
5756 else if (enc_utf8)
|
|
|
5757 {
|
|
|
5758 for (i = 128; i < 256; ++i)
|
|
|
5759 {
|
|
|
5760 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
5761 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
5762 spelltab.st_fold[i] = utf_fold(i);
|
|
|
5763 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
5764 }
|
|
|
5765 }
|
|
307
|
5766 else
|
|
|
5767 #endif
|
|
|
5768 {
|
|
324
|
5769 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
5770 for (i = 128; i < 256; ++i)
|
|
|
5771 {
|
|
|
5772 if (MB_ISUPPER(i))
|
|
|
5773 {
|
|
324
|
5774 spelltab.st_isw[i] = TRUE;
|
|
307
|
5775 spelltab.st_isu[i] = TRUE;
|
|
|
5776 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
5777 }
|
|
324
|
5778 else if (MB_ISLOWER(i))
|
|
|
5779 {
|
|
|
5780 spelltab.st_isw[i] = TRUE;
|
|
|
5781 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
5782 }
|
|
307
|
5783 }
|
|
|
5784 }
|
|
|
5785 }
|
|
|
5786
|
|
|
5787 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
5788 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
5789
|
|
|
5790 /*
|
|
|
5791 * Set the spell character tables from strings in the affix file.
|
|
|
5792 */
|
|
|
5793 static int
|
|
|
5794 set_spell_chartab(fol, low, upp)
|
|
|
5795 char_u *fol;
|
|
|
5796 char_u *low;
|
|
|
5797 char_u *upp;
|
|
|
5798 {
|
|
|
5799 /* We build the new tables here first, so that we can compare with the
|
|
|
5800 * previous one. */
|
|
|
5801 spelltab_T new_st;
|
|
|
5802 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
5803 int f, l, u;
|
|
|
5804
|
|
|
5805 clear_spell_chartab(&new_st);
|
|
|
5806
|
|
|
5807 while (*pf != NUL)
|
|
|
5808 {
|
|
|
5809 if (*pl == NUL || *pu == NUL)
|
|
|
5810 {
|
|
|
5811 EMSG(_(e_affform));
|
|
|
5812 return FAIL;
|
|
|
5813 }
|
|
|
5814 #ifdef FEAT_MBYTE
|
|
|
5815 f = mb_ptr2char_adv(&pf);
|
|
|
5816 l = mb_ptr2char_adv(&pl);
|
|
|
5817 u = mb_ptr2char_adv(&pu);
|
|
|
5818 #else
|
|
|
5819 f = *pf++;
|
|
|
5820 l = *pl++;
|
|
|
5821 u = *pu++;
|
|
|
5822 #endif
|
|
|
5823 /* Every character that appears is a word character. */
|
|
|
5824 if (f < 256)
|
|
|
5825 new_st.st_isw[f] = TRUE;
|
|
|
5826 if (l < 256)
|
|
|
5827 new_st.st_isw[l] = TRUE;
|
|
|
5828 if (u < 256)
|
|
|
5829 new_st.st_isw[u] = TRUE;
|
|
|
5830
|
|
|
5831 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
5832 * case-folding */
|
|
|
5833 if (l < 256 && l != f)
|
|
|
5834 {
|
|
|
5835 if (f >= 256)
|
|
|
5836 {
|
|
|
5837 EMSG(_(e_affrange));
|
|
|
5838 return FAIL;
|
|
|
5839 }
|
|
|
5840 new_st.st_fold[l] = f;
|
|
|
5841 }
|
|
|
5842
|
|
|
5843 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
5844 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
5845 * "FOL" . */
|
|
307
|
5846 if (u < 256 && u != f)
|
|
|
5847 {
|
|
|
5848 if (f >= 256)
|
|
|
5849 {
|
|
|
5850 EMSG(_(e_affrange));
|
|
|
5851 return FAIL;
|
|
|
5852 }
|
|
|
5853 new_st.st_fold[u] = f;
|
|
|
5854 new_st.st_isu[u] = TRUE;
|
|
324
|
5855 new_st.st_upper[f] = u;
|
|
307
|
5856 }
|
|
|
5857 }
|
|
|
5858
|
|
|
5859 if (*pl != NUL || *pu != NUL)
|
|
|
5860 {
|
|
|
5861 EMSG(_(e_affform));
|
|
|
5862 return FAIL;
|
|
|
5863 }
|
|
|
5864
|
|
|
5865 return set_spell_finish(&new_st);
|
|
|
5866 }
|
|
|
5867
|
|
|
5868 /*
|
|
|
5869 * Set the spell character tables from strings in the .spl file.
|
|
|
5870 */
|
|
|
5871 static int
|
|
389
|
5872 set_spell_charflags(flags, cnt, fol)
|
|
307
|
5873 char_u *flags;
|
|
389
|
5874 int cnt; /* length of "flags" */
|
|
|
5875 char_u *fol;
|
|
307
|
5876 {
|
|
|
5877 /* We build the new tables here first, so that we can compare with the
|
|
|
5878 * previous one. */
|
|
|
5879 spelltab_T new_st;
|
|
|
5880 int i;
|
|
389
|
5881 char_u *p = fol;
|
|
324
|
5882 int c;
|
|
307
|
5883
|
|
|
5884 clear_spell_chartab(&new_st);
|
|
|
5885
|
|
389
|
5886 for (i = 0; i < 128; ++i)
|
|
|
5887 {
|
|
|
5888 if (i < cnt)
|
|
|
5889 {
|
|
|
5890 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
5891 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
5892 }
|
|
|
5893
|
|
|
5894 if (*p != NUL)
|
|
|
5895 {
|
|
307
|
5896 #ifdef FEAT_MBYTE
|
|
389
|
5897 c = mb_ptr2char_adv(&p);
|
|
307
|
5898 #else
|
|
389
|
5899 c = *p++;
|
|
307
|
5900 #endif
|
|
389
|
5901 new_st.st_fold[i + 128] = c;
|
|
|
5902 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
5903 new_st.st_upper[c] = i + 128;
|
|
|
5904 }
|
|
307
|
5905 }
|
|
|
5906
|
|
|
5907 return set_spell_finish(&new_st);
|
|
|
5908 }
|
|
|
5909
|
|
|
5910 static int
|
|
|
5911 set_spell_finish(new_st)
|
|
|
5912 spelltab_T *new_st;
|
|
|
5913 {
|
|
|
5914 int i;
|
|
|
5915
|
|
|
5916 if (did_set_spelltab)
|
|
|
5917 {
|
|
|
5918 /* check that it's the same table */
|
|
|
5919 for (i = 0; i < 256; ++i)
|
|
|
5920 {
|
|
|
5921 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
5922 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
5923 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
5924 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
5925 {
|
|
|
5926 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
5927 return FAIL;
|
|
|
5928 }
|
|
|
5929 }
|
|
|
5930 }
|
|
|
5931 else
|
|
|
5932 {
|
|
|
5933 /* copy the new spelltab into the one being used */
|
|
|
5934 spelltab = *new_st;
|
|
|
5935 did_set_spelltab = TRUE;
|
|
|
5936 }
|
|
|
5937
|
|
|
5938 return OK;
|
|
|
5939 }
|
|
|
5940
|
|
|
5941 /*
|
|
358
|
5942 * Return TRUE if "p" points to a word character.
|
|
366
|
5943 * As a special case we see "midword" characters as word character when it is
|
|
358
|
5944 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
5945 * Thus this only works properly when past the first character of the word.
|
|
358
|
5946 */
|
|
|
5947 static int
|
|
376
|
5948 spell_iswordp(p, buf)
|
|
358
|
5949 char_u *p;
|
|
376
|
5950 buf_T *buf; /* buffer used */
|
|
358
|
5951 {
|
|
366
|
5952 #ifdef FEAT_MBYTE
|
|
358
|
5953 char_u *s;
|
|
366
|
5954 int l;
|
|
|
5955 int c;
|
|
|
5956
|
|
|
5957 if (has_mbyte)
|
|
|
5958 {
|
|
|
5959 l = MB_BYTE2LEN(*p);
|
|
358
|
5960 s = p;
|
|
366
|
5961 if (l == 1)
|
|
|
5962 {
|
|
|
5963 /* be quick for ASCII */
|
|
376
|
5964 if (buf->b_spell_ismw[*p])
|
|
366
|
5965 {
|
|
|
5966 s = p + 1; /* skip a mid-word character */
|
|
|
5967 l = MB_BYTE2LEN(*s);
|
|
|
5968 }
|
|
|
5969 }
|
|
|
5970 else
|
|
|
5971 {
|
|
|
5972 c = mb_ptr2char(p);
|
|
376
|
5973 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
5974 : (buf->b_spell_ismw_mb != NULL
|
|
|
5975 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
5976 {
|
|
|
5977 s = p + l;
|
|
|
5978 l = MB_BYTE2LEN(*s);
|
|
|
5979 }
|
|
|
5980 }
|
|
|
5981
|
|
390
|
5982 c = mb_ptr2char(s);
|
|
|
5983 if (c > 255)
|
|
366
|
5984 return mb_get_class(s) >= 2;
|
|
390
|
5985 return spelltab.st_isw[c];
|
|
366
|
5986 }
|
|
358
|
5987 #endif
|
|
366
|
5988
|
|
376
|
5989 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
5990 }
|
|
|
5991
|
|
|
5992 /*
|
|
|
5993 * Return TRUE if "p" points to a word character.
|
|
|
5994 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
5995 */
|
|
|
5996 static int
|
|
|
5997 spell_iswordp_nmw(p)
|
|
|
5998 char_u *p;
|
|
|
5999 {
|
|
|
6000 #ifdef FEAT_MBYTE
|
|
390
|
6001 int c;
|
|
|
6002
|
|
|
6003 if (has_mbyte)
|
|
|
6004 {
|
|
|
6005 c = mb_ptr2char(p);
|
|
|
6006 if (c > 255)
|
|
|
6007 return mb_get_class(p) >= 2;
|
|
|
6008 return spelltab.st_isw[c];
|
|
|
6009 }
|
|
376
|
6010 #endif
|
|
|
6011 return spelltab.st_isw[*p];
|
|
358
|
6012 }
|
|
|
6013
|
|
372
|
6014 #ifdef FEAT_MBYTE
|
|
|
6015 /*
|
|
|
6016 * Return TRUE if "p" points to a word character.
|
|
|
6017 * Wide version of spell_iswordp().
|
|
|
6018 */
|
|
|
6019 static int
|
|
376
|
6020 spell_iswordp_w(p, buf)
|
|
372
|
6021 int *p;
|
|
376
|
6022 buf_T *buf;
|
|
372
|
6023 {
|
|
|
6024 int *s;
|
|
|
6025
|
|
376
|
6026 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
6027 : (buf->b_spell_ismw_mb != NULL
|
|
|
6028 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
6029 s = p + 1;
|
|
|
6030 else
|
|
|
6031 s = p;
|
|
|
6032
|
|
390
|
6033 if (*s > 255)
|
|
372
|
6034 {
|
|
|
6035 if (enc_utf8)
|
|
|
6036 return utf_class(*s) >= 2;
|
|
|
6037 if (enc_dbcs)
|
|
|
6038 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
6039 return 0;
|
|
|
6040 }
|
|
|
6041 return spelltab.st_isw[*s];
|
|
|
6042 }
|
|
|
6043 #endif
|
|
|
6044
|
|
358
|
6045 /*
|
|
339
|
6046 * Write the table with prefix conditions to the .spl file.
|
|
|
6047 */
|
|
|
6048 static void
|
|
|
6049 write_spell_prefcond(fd, gap)
|
|
|
6050 FILE *fd;
|
|
|
6051 garray_T *gap;
|
|
|
6052 {
|
|
|
6053 int i;
|
|
|
6054 char_u *p;
|
|
|
6055 int len;
|
|
|
6056
|
|
|
6057 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
6058
|
|
|
6059 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6060 {
|
|
|
6061 /* <prefcond> : <condlen> <condstr> */
|
|
|
6062 p = ((char_u **)gap->ga_data)[i];
|
|
|
6063 if (p == NULL)
|
|
|
6064 fputc(0, fd);
|
|
|
6065 else
|
|
|
6066 {
|
|
|
6067 len = STRLEN(p);
|
|
|
6068 fputc(len, fd);
|
|
|
6069 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
6070 }
|
|
|
6071 }
|
|
|
6072 }
|
|
|
6073
|
|
|
6074 /*
|
|
307
|
6075 * Write the current tables into the .spl file.
|
|
|
6076 * This makes sure the same characters are recognized as word characters when
|
|
|
6077 * generating an when using a spell file.
|
|
|
6078 */
|
|
|
6079 static void
|
|
|
6080 write_spell_chartab(fd)
|
|
|
6081 FILE *fd;
|
|
|
6082 {
|
|
|
6083 char_u charbuf[256 * 4];
|
|
|
6084 int len = 0;
|
|
|
6085 int flags;
|
|
|
6086 int i;
|
|
|
6087
|
|
|
6088 fputc(128, fd); /* <charflagslen> */
|
|
|
6089 for (i = 128; i < 256; ++i)
|
|
|
6090 {
|
|
|
6091 flags = 0;
|
|
|
6092 if (spelltab.st_isw[i])
|
|
324
|
6093 flags |= CF_WORD;
|
|
307
|
6094 if (spelltab.st_isu[i])
|
|
324
|
6095 flags |= CF_UPPER;
|
|
307
|
6096 fputc(flags, fd); /* <charflags> */
|
|
|
6097
|
|
310
|
6098 #ifdef FEAT_MBYTE
|
|
|
6099 if (has_mbyte)
|
|
|
6100 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
|
6101 else
|
|
|
6102 #endif
|
|
|
6103 charbuf[len++] = spelltab.st_fold[i];
|
|
307
|
6104 }
|
|
|
6105
|
|
|
6106 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
|
6107 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
|
6108 }
|
|
|
6109
|
|
|
6110 /*
|
|
324
|
6111 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
6112 * Uses the character definitions from the .spl file.
|
|
307
|
6113 * When using a multi-byte 'encoding' the length may change!
|
|
|
6114 * Returns FAIL when something wrong.
|
|
|
6115 */
|
|
|
6116 static int
|
|
324
|
6117 spell_casefold(str, len, buf, buflen)
|
|
|
6118 char_u *str;
|
|
307
|
6119 int len;
|
|
|
6120 char_u *buf;
|
|
|
6121 int buflen;
|
|
|
6122 {
|
|
|
6123 int i;
|
|
|
6124
|
|
|
6125 if (len >= buflen)
|
|
|
6126 {
|
|
|
6127 buf[0] = NUL;
|
|
|
6128 return FAIL; /* result will not fit */
|
|
|
6129 }
|
|
|
6130
|
|
|
6131 #ifdef FEAT_MBYTE
|
|
|
6132 if (has_mbyte)
|
|
|
6133 {
|
|
324
|
6134 int outi = 0;
|
|
|
6135 char_u *p;
|
|
307
|
6136 int c;
|
|
|
6137
|
|
|
6138 /* Fold one character at a time. */
|
|
324
|
6139 for (p = str; p < str + len; )
|
|
307
|
6140 {
|
|
|
6141 if (outi + MB_MAXBYTES > buflen)
|
|
|
6142 {
|
|
|
6143 buf[outi] = NUL;
|
|
|
6144 return FAIL;
|
|
|
6145 }
|
|
324
|
6146 c = mb_ptr2char_adv(&p);
|
|
|
6147 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
6148 }
|
|
|
6149 buf[outi] = NUL;
|
|
|
6150 }
|
|
|
6151 else
|
|
|
6152 #endif
|
|
|
6153 {
|
|
|
6154 /* Be quick for non-multibyte encodings. */
|
|
|
6155 for (i = 0; i < len; ++i)
|
|
324
|
6156 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
6157 buf[i] = NUL;
|
|
|
6158 }
|
|
|
6159
|
|
|
6160 return OK;
|
|
|
6161 }
|
|
|
6162
|
|
372
|
6163 #define SPS_BEST 1
|
|
|
6164 #define SPS_FAST 2
|
|
|
6165 #define SPS_DOUBLE 4
|
|
|
6166
|
|
|
6167 static int sps_flags = SPS_BEST;
|
|
|
6168
|
|
|
6169 /*
|
|
|
6170 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
|
6171 * Sets "sps_flags".
|
|
|
6172 */
|
|
|
6173 int
|
|
|
6174 spell_check_sps()
|
|
|
6175 {
|
|
|
6176 char_u *p;
|
|
|
6177 char_u buf[MAXPATHL];
|
|
|
6178 int f;
|
|
|
6179
|
|
|
6180 sps_flags = 0;
|
|
|
6181
|
|
|
6182 for (p = p_sps; *p != NUL; )
|
|
|
6183 {
|
|
|
6184 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6185
|
|
|
6186 f = 0;
|
|
|
6187 if (STRCMP(buf, "best") == 0)
|
|
|
6188 f = SPS_BEST;
|
|
|
6189 else if (STRCMP(buf, "fast") == 0)
|
|
|
6190 f = SPS_FAST;
|
|
|
6191 else if (STRCMP(buf, "double") == 0)
|
|
|
6192 f = SPS_DOUBLE;
|
|
|
6193 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
6194 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
6195 f = -1;
|
|
|
6196
|
|
|
6197 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
6198 {
|
|
|
6199 sps_flags = SPS_BEST;
|
|
|
6200 return FAIL;
|
|
|
6201 }
|
|
|
6202 if (f != 0)
|
|
|
6203 sps_flags = f;
|
|
|
6204 }
|
|
|
6205
|
|
|
6206 if (sps_flags == 0)
|
|
|
6207 sps_flags = SPS_BEST;
|
|
|
6208
|
|
|
6209 return OK;
|
|
|
6210 }
|
|
|
6211
|
|
|
6212 /* Remember what "z?" replaced. */
|
|
|
6213 static char_u *repl_from = NULL;
|
|
|
6214 static char_u *repl_to = NULL;
|
|
|
6215
|
|
323
|
6216 /*
|
|
|
6217 * "z?": Find badly spelled word under or after the cursor.
|
|
|
6218 * Give suggestions for the properly spelled word.
|
|
|
6219 */
|
|
|
6220 void
|
|
|
6221 spell_suggest()
|
|
|
6222 {
|
|
|
6223 char_u *line;
|
|
|
6224 pos_T prev_cursor = curwin->w_cursor;
|
|
|
6225 char_u wcopy[MAXWLEN + 2];
|
|
|
6226 char_u *p;
|
|
|
6227 int i;
|
|
|
6228 int c;
|
|
|
6229 suginfo_T sug;
|
|
|
6230 suggest_T *stp;
|
|
372
|
6231 int mouse_used;
|
|
388
|
6232 int need_cap;
|
|
|
6233 regmatch_T regmatch;
|
|
|
6234 int endcol;
|
|
|
6235 char_u *line_copy = NULL;
|
|
323
|
6236
|
|
344
|
6237 /* Find the start of the badly spelled word. */
|
|
346
|
6238 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
6239 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
6240 {
|
|
|
6241 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
6242 return;
|
|
|
6243
|
|
|
6244 /* No bad word or it starts after the cursor: use the word under the
|
|
|
6245 * cursor. */
|
|
|
6246 curwin->w_cursor = prev_cursor;
|
|
|
6247 line = ml_get_curline();
|
|
|
6248 p = line + curwin->w_cursor.col;
|
|
|
6249 /* Backup to before start of word. */
|
|
390
|
6250 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
6251 mb_ptr_back(line, p);
|
|
|
6252 /* Forward to start of word. */
|
|
390
|
6253 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
6254 mb_ptr_adv(p);
|
|
|
6255
|
|
390
|
6256 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
6257 {
|
|
|
6258 beep_flush();
|
|
|
6259 return;
|
|
|
6260 }
|
|
|
6261 curwin->w_cursor.col = p - line;
|
|
323
|
6262 }
|
|
|
6263
|
|
344
|
6264 /* Get the word and its length. */
|
|
323
|
6265 line = ml_get_curline();
|
|
344
|
6266
|
|
388
|
6267 /* Figure out if the word should be capitalised. */
|
|
|
6268 need_cap = FALSE;
|
|
|
6269 if (curbuf->b_cap_prog != NULL)
|
|
|
6270 {
|
|
|
6271 endcol = 0;
|
|
398
|
6272 if ((int)(skipwhite(line) - line) == (int)curwin->w_cursor.col)
|
|
388
|
6273 {
|
|
|
6274 /* At start of line, check if previous line is empty or sentence
|
|
|
6275 * ends there. */
|
|
|
6276 if (curwin->w_cursor.lnum == 1)
|
|
|
6277 need_cap = TRUE;
|
|
|
6278 else
|
|
|
6279 {
|
|
|
6280 line = ml_get(curwin->w_cursor.lnum - 1);
|
|
|
6281 if (*skipwhite(line) == NUL)
|
|
|
6282 need_cap = TRUE;
|
|
|
6283 else
|
|
|
6284 {
|
|
|
6285 /* Append a space in place of the line break. */
|
|
|
6286 line_copy = concat_str(line, (char_u *)" ");
|
|
|
6287 line = line_copy;
|
|
|
6288 endcol = STRLEN(line);
|
|
|
6289 }
|
|
|
6290 }
|
|
|
6291 }
|
|
|
6292 else
|
|
|
6293 endcol = curwin->w_cursor.col;
|
|
|
6294
|
|
|
6295 if (endcol > 0)
|
|
|
6296 {
|
|
|
6297 /* Check if sentence ends before the bad word. */
|
|
|
6298 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
6299 regmatch.rm_ic = FALSE;
|
|
|
6300 p = line + endcol;
|
|
|
6301 for (;;)
|
|
|
6302 {
|
|
|
6303 mb_ptr_back(line, p);
|
|
390
|
6304 if (p == line || spell_iswordp_nmw(p))
|
|
388
|
6305 break;
|
|
|
6306 if (vim_regexec(®match, p, 0)
|
|
|
6307 && regmatch.endp[0] == line + endcol)
|
|
|
6308 {
|
|
|
6309 need_cap = TRUE;
|
|
|
6310 break;
|
|
|
6311 }
|
|
|
6312 }
|
|
|
6313 }
|
|
|
6314
|
|
|
6315 /* get the line again, we may have been using the previous one */
|
|
|
6316 line = ml_get_curline();
|
|
|
6317 vim_free(line_copy);
|
|
|
6318 }
|
|
|
6319
|
|
344
|
6320 /* Get the list of suggestions */
|
|
388
|
6321 spell_find_suggest(line + curwin->w_cursor.col, &sug, (int)Rows - 2,
|
|
|
6322 TRUE, need_cap);
|
|
323
|
6323
|
|
|
6324 if (sug.su_ga.ga_len == 0)
|
|
|
6325 MSG(_("Sorry, no suggestions"));
|
|
|
6326 else
|
|
|
6327 {
|
|
372
|
6328 vim_free(repl_from);
|
|
|
6329 repl_from = NULL;
|
|
|
6330 vim_free(repl_to);
|
|
|
6331 repl_to = NULL;
|
|
|
6332
|
|
323
|
6333 /* List the suggestions. */
|
|
|
6334 msg_start();
|
|
372
|
6335 lines_left = Rows; /* avoid more prompt */
|
|
323
|
6336 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
6337 sug.su_badlen, sug.su_badptr);
|
|
|
6338 msg_puts(IObuff);
|
|
|
6339 msg_clr_eos();
|
|
|
6340 msg_putchar('\n');
|
|
346
|
6341
|
|
323
|
6342 msg_scroll = TRUE;
|
|
|
6343 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
6344 {
|
|
344
|
6345 stp = &SUG(sug.su_ga, i);
|
|
323
|
6346
|
|
|
6347 /* The suggested word may replace only part of the bad word, add
|
|
|
6348 * the not replaced part. */
|
|
|
6349 STRCPY(wcopy, stp->st_word);
|
|
|
6350 if (sug.su_badlen > stp->st_orglen)
|
|
|
6351 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
6352 sug.su_badptr + stp->st_orglen,
|
|
|
6353 sug.su_badlen - stp->st_orglen);
|
|
346
|
6354 vim_snprintf((char *)IObuff, IOSIZE, _("%2d \"%s\""), i + 1, wcopy);
|
|
|
6355 msg_puts(IObuff);
|
|
|
6356
|
|
|
6357 /* The word may replace more than "su_badlen". */
|
|
|
6358 if (sug.su_badlen < stp->st_orglen)
|
|
|
6359 {
|
|
|
6360 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
6361 stp->st_orglen, sug.su_badptr);
|
|
|
6362 msg_puts(IObuff);
|
|
|
6363 }
|
|
|
6364
|
|
324
|
6365 if (p_verbose > 0)
|
|
344
|
6366 {
|
|
346
|
6367 /* Add the score. */
|
|
351
|
6368 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
346
|
6369 vim_snprintf((char *)IObuff, IOSIZE, _(" (%s%d - %d)"),
|
|
344
|
6370 stp->st_salscore ? "s " : "",
|
|
|
6371 stp->st_score, stp->st_altscore);
|
|
|
6372 else
|
|
346
|
6373 vim_snprintf((char *)IObuff, IOSIZE, _(" (%d)"),
|
|
|
6374 stp->st_score);
|
|
|
6375 msg_advance(30);
|
|
|
6376 msg_puts(IObuff);
|
|
344
|
6377 }
|
|
323
|
6378 msg_putchar('\n');
|
|
|
6379 }
|
|
|
6380
|
|
|
6381 /* Ask for choice. */
|
|
372
|
6382 i = prompt_for_number(&mouse_used);
|
|
|
6383 if (mouse_used)
|
|
|
6384 i -= lines_left;
|
|
|
6385
|
|
344
|
6386 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
6387 {
|
|
372
|
6388 /* Save the from and to text for :spellrepall. */
|
|
|
6389 stp = &SUG(sug.su_ga, i - 1);
|
|
|
6390 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
6391 repl_to = vim_strsave(stp->st_word);
|
|
|
6392
|
|
323
|
6393 /* Replace the word. */
|
|
|
6394 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
6395 if (p != NULL)
|
|
|
6396 {
|
|
|
6397 c = sug.su_badptr - line;
|
|
|
6398 mch_memmove(p, line, c);
|
|
|
6399 STRCPY(p + c, stp->st_word);
|
|
|
6400 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
6401 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6402 curwin->w_cursor.col = c;
|
|
|
6403 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
6404
|
|
|
6405 /* For redo we use a change-word command. */
|
|
|
6406 ResetRedobuff();
|
|
|
6407 AppendToRedobuff((char_u *)"ciw");
|
|
|
6408 AppendToRedobuff(stp->st_word);
|
|
|
6409 AppendCharToRedobuff(ESC);
|
|
323
|
6410 }
|
|
|
6411 }
|
|
|
6412 else
|
|
|
6413 curwin->w_cursor = prev_cursor;
|
|
|
6414 }
|
|
|
6415
|
|
344
|
6416 spell_find_cleanup(&sug);
|
|
|
6417 }
|
|
|
6418
|
|
|
6419 /*
|
|
372
|
6420 * ":spellrepall"
|
|
|
6421 */
|
|
|
6422 /*ARGSUSED*/
|
|
|
6423 void
|
|
|
6424 ex_spellrepall(eap)
|
|
|
6425 exarg_T *eap;
|
|
|
6426 {
|
|
|
6427 pos_T pos = curwin->w_cursor;
|
|
|
6428 char_u *frompat;
|
|
|
6429 int addlen;
|
|
|
6430 char_u *line;
|
|
|
6431 char_u *p;
|
|
|
6432 int didone = FALSE;
|
|
|
6433 int save_ws = p_ws;
|
|
|
6434
|
|
|
6435 if (repl_from == NULL || repl_to == NULL)
|
|
|
6436 {
|
|
|
6437 EMSG(_("E752: No previous spell replacement"));
|
|
|
6438 return;
|
|
|
6439 }
|
|
|
6440 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
6441
|
|
|
6442 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
6443 if (frompat == NULL)
|
|
|
6444 return;
|
|
|
6445 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
6446 p_ws = FALSE;
|
|
|
6447
|
|
|
6448 curwin->w_cursor.lnum = 0;
|
|
|
6449 while (!got_int)
|
|
|
6450 {
|
|
|
6451 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
6452 || u_save_cursor() == FAIL)
|
|
|
6453 break;
|
|
|
6454
|
|
|
6455 /* Only replace when the right word isn't there yet. This happens
|
|
|
6456 * when changing "etc" to "etc.". */
|
|
|
6457 line = ml_get_curline();
|
|
|
6458 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
6459 repl_to, STRLEN(repl_to)) != 0)
|
|
|
6460 {
|
|
|
6461 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
6462 if (p == NULL)
|
|
|
6463 break;
|
|
|
6464 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
6465 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
6466 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
6467 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
6468 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
6469 didone = TRUE;
|
|
|
6470 }
|
|
|
6471 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
6472 }
|
|
|
6473
|
|
|
6474 p_ws = save_ws;
|
|
|
6475 curwin->w_cursor = pos;
|
|
|
6476 vim_free(frompat);
|
|
|
6477
|
|
|
6478 if (!didone)
|
|
|
6479 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
|
6480 }
|
|
|
6481
|
|
|
6482 /*
|
|
344
|
6483 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
6484 * a list of allocated strings.
|
|
|
6485 */
|
|
|
6486 void
|
|
|
6487 spell_suggest_list(gap, word, maxcount)
|
|
|
6488 garray_T *gap;
|
|
|
6489 char_u *word;
|
|
|
6490 int maxcount; /* maximum nr of suggestions */
|
|
|
6491 {
|
|
|
6492 suginfo_T sug;
|
|
|
6493 int i;
|
|
|
6494 suggest_T *stp;
|
|
|
6495 char_u *wcopy;
|
|
|
6496
|
|
388
|
6497 spell_find_suggest(word, &sug, maxcount, FALSE, FALSE);
|
|
344
|
6498
|
|
|
6499 /* Make room in "gap". */
|
|
|
6500 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
6501 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
6502 return;
|
|
|
6503
|
|
323
|
6504 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
6505 {
|
|
|
6506 stp = &SUG(sug.su_ga, i);
|
|
|
6507
|
|
|
6508 /* The suggested word may replace only part of "word", add the not
|
|
|
6509 * replaced part. */
|
|
|
6510 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
6511 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
6512 if (wcopy == NULL)
|
|
|
6513 break;
|
|
|
6514 STRCPY(wcopy, stp->st_word);
|
|
|
6515 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
6516 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
6517 }
|
|
|
6518
|
|
|
6519 spell_find_cleanup(&sug);
|
|
|
6520 }
|
|
|
6521
|
|
|
6522 /*
|
|
|
6523 * Find spell suggestions for the word at the start of "badptr".
|
|
|
6524 * Return the suggestions in "su->su_ga".
|
|
|
6525 * The maximum number of suggestions is "maxcount".
|
|
|
6526 * Note: does use info for the current window.
|
|
|
6527 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
6528 */
|
|
|
6529 static void
|
|
388
|
6530 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
6531 char_u *badptr;
|
|
|
6532 suginfo_T *su;
|
|
|
6533 int maxcount;
|
|
358
|
6534 int banbadword; /* don't include badword in suggestions */
|
|
388
|
6535 int need_cap; /* word should start with capital */
|
|
344
|
6536 {
|
|
385
|
6537 int attr = 0;
|
|
372
|
6538 char_u buf[MAXPATHL];
|
|
|
6539 char_u *p;
|
|
|
6540 int do_combine = FALSE;
|
|
|
6541 char_u *sps_copy;
|
|
|
6542 #ifdef FEAT_EVAL
|
|
|
6543 static int expr_busy = FALSE;
|
|
|
6544 #endif
|
|
385
|
6545 int c;
|
|
344
|
6546
|
|
|
6547 /*
|
|
|
6548 * Set the info in "*su".
|
|
|
6549 */
|
|
|
6550 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
6551 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
6552 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
6553 if (*badptr == NUL)
|
|
|
6554 return;
|
|
344
|
6555 hash_init(&su->su_banned);
|
|
|
6556
|
|
|
6557 su->su_badptr = badptr;
|
|
385
|
6558 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
6559 su->su_maxcount = maxcount;
|
|
372
|
6560 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
6561
|
|
|
6562 if (su->su_badlen >= MAXWLEN)
|
|
|
6563 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
6564 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
6565 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
6566 su->su_fbadword, MAXWLEN);
|
|
346
|
6567 /* get caps flags for bad word */
|
|
|
6568 su->su_badflags = captype(su->su_badptr, su->su_badptr + su->su_badlen);
|
|
388
|
6569 if (need_cap)
|
|
|
6570 su->su_badflags |= WF_ONECAP;
|
|
344
|
6571
|
|
385
|
6572 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
6573 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
6574 * for that. */
|
|
455
|
6575 c = PTR2CHAR(su->su_badptr);
|
|
385
|
6576 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
6577 {
|
|
|
6578 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
6579 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
6580 0, TRUE);
|
|
|
6581 }
|
|
|
6582
|
|
344
|
6583 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
6584 if (banbadword)
|
|
|
6585 add_banned(su, su->su_badword);
|
|
344
|
6586
|
|
372
|
6587 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
6588 sps_copy = vim_strsave(p_sps);
|
|
|
6589 if (sps_copy == NULL)
|
|
|
6590 return;
|
|
|
6591
|
|
|
6592 /* Loop over the items in 'spellsuggest'. */
|
|
|
6593 for (p = sps_copy; *p != NUL; )
|
|
|
6594 {
|
|
|
6595 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
6596
|
|
|
6597 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
6598 {
|
|
|
6599 #ifdef FEAT_EVAL
|
|
375
|
6600 /* Evaluate an expression. Skip this when called recursively,
|
|
|
6601 * when using spellsuggest() in the expression. */
|
|
372
|
6602 if (!expr_busy)
|
|
|
6603 {
|
|
|
6604 expr_busy = TRUE;
|
|
|
6605 spell_suggest_expr(su, buf + 5);
|
|
|
6606 expr_busy = FALSE;
|
|
|
6607 }
|
|
|
6608 #endif
|
|
|
6609 }
|
|
|
6610 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
6611 /* Use list of suggestions in a file. */
|
|
|
6612 spell_suggest_file(su, buf + 5);
|
|
|
6613 else
|
|
|
6614 {
|
|
|
6615 /* Use internal method. */
|
|
|
6616 spell_suggest_intern(su);
|
|
|
6617 if (sps_flags & SPS_DOUBLE)
|
|
|
6618 do_combine = TRUE;
|
|
|
6619 }
|
|
|
6620 }
|
|
|
6621
|
|
|
6622 vim_free(sps_copy);
|
|
|
6623
|
|
|
6624 if (do_combine)
|
|
|
6625 /* Combine the two list of suggestions. This must be done last,
|
|
|
6626 * because sorting changes the order again. */
|
|
|
6627 score_combine(su);
|
|
|
6628 }
|
|
|
6629
|
|
|
6630 #ifdef FEAT_EVAL
|
|
|
6631 /*
|
|
|
6632 * Find suggestions by evaluating expression "expr".
|
|
|
6633 */
|
|
|
6634 static void
|
|
|
6635 spell_suggest_expr(su, expr)
|
|
|
6636 suginfo_T *su;
|
|
|
6637 char_u *expr;
|
|
|
6638 {
|
|
|
6639 list_T *list;
|
|
|
6640 listitem_T *li;
|
|
|
6641 int score;
|
|
|
6642 char_u *p;
|
|
|
6643
|
|
|
6644 /* The work is split up in a few parts to avoid having to export
|
|
|
6645 * suginfo_T.
|
|
|
6646 * First evaluate the expression and get the resulting list. */
|
|
|
6647 list = eval_spell_expr(su->su_badword, expr);
|
|
|
6648 if (list != NULL)
|
|
|
6649 {
|
|
|
6650 /* Loop over the items in the list. */
|
|
|
6651 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
6652 if (li->li_tv.v_type == VAR_LIST)
|
|
|
6653 {
|
|
|
6654 /* Get the word and the score from the items. */
|
|
|
6655 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
6656 if (score >= 0)
|
|
|
6657 add_suggestion(su, &su->su_ga, p,
|
|
|
6658 su->su_badlen, score, 0, TRUE);
|
|
|
6659 }
|
|
|
6660 list_unref(list);
|
|
|
6661 }
|
|
|
6662
|
|
|
6663 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6664 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6665 }
|
|
|
6666 #endif
|
|
|
6667
|
|
|
6668 /*
|
|
|
6669 * Find suggestions a file "fname".
|
|
|
6670 */
|
|
|
6671 static void
|
|
|
6672 spell_suggest_file(su, fname)
|
|
|
6673 suginfo_T *su;
|
|
|
6674 char_u *fname;
|
|
|
6675 {
|
|
|
6676 FILE *fd;
|
|
|
6677 char_u line[MAXWLEN * 2];
|
|
|
6678 char_u *p;
|
|
|
6679 int len;
|
|
|
6680 char_u cword[MAXWLEN];
|
|
|
6681
|
|
|
6682 /* Open the file. */
|
|
|
6683 fd = mch_fopen((char *)fname, "r");
|
|
|
6684 if (fd == NULL)
|
|
|
6685 {
|
|
|
6686 EMSG2(_(e_notopen), fname);
|
|
|
6687 return;
|
|
|
6688 }
|
|
|
6689
|
|
|
6690 /* Read it line by line. */
|
|
|
6691 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
6692 {
|
|
|
6693 line_breakcheck();
|
|
|
6694
|
|
|
6695 p = vim_strchr(line, '/');
|
|
|
6696 if (p == NULL)
|
|
|
6697 continue; /* No Tab found, just skip the line. */
|
|
|
6698 *p++ = NUL;
|
|
|
6699 if (STRICMP(su->su_badword, line) == 0)
|
|
|
6700 {
|
|
|
6701 /* Match! Isolate the good word, until CR or NL. */
|
|
|
6702 for (len = 0; p[len] >= ' '; ++len)
|
|
|
6703 ;
|
|
|
6704 p[len] = NUL;
|
|
|
6705
|
|
|
6706 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
6707 * of the bad word. */
|
|
|
6708 if (captype(p, NULL) == 0)
|
|
|
6709 {
|
|
|
6710 make_case_word(p, cword, su->su_badflags);
|
|
|
6711 p = cword;
|
|
|
6712 }
|
|
|
6713
|
|
|
6714 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
6715 SCORE_FILE, 0, TRUE);
|
|
|
6716 }
|
|
|
6717 }
|
|
|
6718
|
|
|
6719 fclose(fd);
|
|
|
6720
|
|
|
6721 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
6722 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
6723 }
|
|
|
6724
|
|
|
6725 /*
|
|
|
6726 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
6727 */
|
|
|
6728 static void
|
|
|
6729 spell_suggest_intern(su)
|
|
|
6730 suginfo_T *su;
|
|
|
6731 {
|
|
344
|
6732 /*
|
|
346
|
6733 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
6734 *
|
|
|
6735 * Set a maximum score to limit the combination of operations that is
|
|
|
6736 * tried.
|
|
|
6737 */
|
|
346
|
6738 suggest_try_special(su);
|
|
|
6739
|
|
|
6740 /*
|
|
|
6741 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
6742 * from the .aff file and inserting a space (split the word).
|
|
|
6743 */
|
|
|
6744 suggest_try_change(su);
|
|
344
|
6745
|
|
|
6746 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
6747 if (sps_flags & SPS_DOUBLE)
|
|
|
6748 score_comp_sal(su);
|
|
|
6749
|
|
|
6750 /*
|
|
346
|
6751 * 3. Try finding sound-a-like words.
|
|
344
|
6752 *
|
|
|
6753 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
6754 * very slow and often doesn't find new suggestions.
|
|
|
6755 */
|
|
|
6756 if ((sps_flags & SPS_DOUBLE)
|
|
|
6757 || (!(sps_flags & SPS_FAST)
|
|
|
6758 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
6759 {
|
|
|
6760 /* Allow a higher score now. */
|
|
|
6761 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
6762 suggest_try_soundalike(su);
|
|
344
|
6763 }
|
|
|
6764
|
|
|
6765 /* When CTRL-C was hit while searching do show the results. */
|
|
|
6766 ui_breakcheck();
|
|
|
6767 if (got_int)
|
|
|
6768 {
|
|
|
6769 (void)vgetc();
|
|
|
6770 got_int = FALSE;
|
|
|
6771 }
|
|
|
6772
|
|
372
|
6773 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
6774 {
|
|
|
6775 if (sps_flags & SPS_BEST)
|
|
|
6776 /* Adjust the word score for how it sounds like. */
|
|
|
6777 rescore_suggestions(su);
|
|
|
6778
|
|
|
6779 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
6780 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
6781 }
|
|
|
6782 }
|
|
|
6783
|
|
|
6784 /*
|
|
|
6785 * Free the info put in "*su" by spell_find_suggest().
|
|
|
6786 */
|
|
|
6787 static void
|
|
|
6788 spell_find_cleanup(su)
|
|
|
6789 suginfo_T *su;
|
|
|
6790 {
|
|
|
6791 int i;
|
|
|
6792
|
|
|
6793 /* Free the suggestions. */
|
|
|
6794 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
6795 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
6796 ga_clear(&su->su_ga);
|
|
|
6797 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
6798 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
6799 ga_clear(&su->su_sga);
|
|
323
|
6800
|
|
|
6801 /* Free the banned words. */
|
|
344
|
6802 free_banned(su);
|
|
323
|
6803 }
|
|
|
6804
|
|
|
6805 /*
|
|
324
|
6806 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
6807 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
6808 * The result is NUL terminated.
|
|
323
|
6809 */
|
|
|
6810 static void
|
|
324
|
6811 onecap_copy(word, wcopy, upper)
|
|
323
|
6812 char_u *word;
|
|
|
6813 char_u *wcopy;
|
|
|
6814 int upper; /* TRUE: first letter made upper case */
|
|
|
6815 {
|
|
|
6816 char_u *p;
|
|
|
6817 int c;
|
|
|
6818 int l;
|
|
|
6819
|
|
|
6820 p = word;
|
|
|
6821 #ifdef FEAT_MBYTE
|
|
|
6822 if (has_mbyte)
|
|
|
6823 c = mb_ptr2char_adv(&p);
|
|
|
6824 else
|
|
|
6825 #endif
|
|
|
6826 c = *p++;
|
|
|
6827 if (upper)
|
|
324
|
6828 c = SPELL_TOUPPER(c);
|
|
323
|
6829 else
|
|
324
|
6830 c = SPELL_TOFOLD(c);
|
|
323
|
6831 #ifdef FEAT_MBYTE
|
|
|
6832 if (has_mbyte)
|
|
|
6833 l = mb_char2bytes(c, wcopy);
|
|
|
6834 else
|
|
|
6835 #endif
|
|
|
6836 {
|
|
|
6837 l = 1;
|
|
|
6838 wcopy[0] = c;
|
|
|
6839 }
|
|
376
|
6840 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
6841 }
|
|
|
6842
|
|
|
6843 /*
|
|
324
|
6844 * Make a copy of "word" with all the letters upper cased into
|
|
|
6845 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
6846 */
|
|
|
6847 static void
|
|
|
6848 allcap_copy(word, wcopy)
|
|
|
6849 char_u *word;
|
|
|
6850 char_u *wcopy;
|
|
|
6851 {
|
|
|
6852 char_u *s;
|
|
|
6853 char_u *d;
|
|
|
6854 int c;
|
|
|
6855
|
|
|
6856 d = wcopy;
|
|
|
6857 for (s = word; *s != NUL; )
|
|
|
6858 {
|
|
|
6859 #ifdef FEAT_MBYTE
|
|
|
6860 if (has_mbyte)
|
|
|
6861 c = mb_ptr2char_adv(&s);
|
|
|
6862 else
|
|
|
6863 #endif
|
|
|
6864 c = *s++;
|
|
324
|
6865 c = SPELL_TOUPPER(c);
|
|
323
|
6866
|
|
|
6867 #ifdef FEAT_MBYTE
|
|
|
6868 if (has_mbyte)
|
|
|
6869 {
|
|
|
6870 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
6871 break;
|
|
|
6872 d += mb_char2bytes(c, d);
|
|
|
6873 }
|
|
|
6874 else
|
|
|
6875 #endif
|
|
|
6876 {
|
|
|
6877 if (d - wcopy >= MAXWLEN - 1)
|
|
|
6878 break;
|
|
|
6879 *d++ = c;
|
|
|
6880 }
|
|
|
6881 }
|
|
|
6882 *d = NUL;
|
|
|
6883 }
|
|
|
6884
|
|
|
6885 /*
|
|
346
|
6886 * Try finding suggestions by recognizing specific situations.
|
|
|
6887 */
|
|
|
6888 static void
|
|
|
6889 suggest_try_special(su)
|
|
|
6890 suginfo_T *su;
|
|
|
6891 {
|
|
|
6892 char_u *p;
|
|
376
|
6893 size_t len;
|
|
346
|
6894 int c;
|
|
|
6895 char_u word[MAXWLEN];
|
|
|
6896
|
|
|
6897 /*
|
|
|
6898 * Recognize a word that is repeated: "the the".
|
|
|
6899 */
|
|
|
6900 p = skiptowhite(su->su_fbadword);
|
|
|
6901 len = p - su->su_fbadword;
|
|
|
6902 p = skipwhite(p);
|
|
|
6903 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
6904 {
|
|
|
6905 /* Include badflags: if the badword is onecap or allcap
|
|
|
6906 * use that for the goodword too: "The the" -> "The". */
|
|
|
6907 c = su->su_fbadword[len];
|
|
|
6908 su->su_fbadword[len] = NUL;
|
|
|
6909 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
6910 su->su_fbadword[len] = c;
|
|
351
|
6911 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
6912 }
|
|
|
6913 }
|
|
|
6914
|
|
|
6915 /*
|
|
323
|
6916 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
6917 *
|
|
|
6918 * This uses a state machine. At each node in the tree we try various
|
|
|
6919 * operations. When trying if an operation work "depth" is increased and the
|
|
|
6920 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
6921 * character, replace one and delete another. The number of changes is
|
|
|
6922 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
6923 */
|
|
|
6924 static void
|
|
346
|
6925 suggest_try_change(su)
|
|
323
|
6926 suginfo_T *su;
|
|
|
6927 {
|
|
|
6928 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
6929 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
6930 trystate_T stack[MAXWLEN];
|
|
|
6931 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
|
6932 * to for word split) */
|
|
|
6933 char_u prewordlen = 0; /* length of word in "preword" */
|
|
|
6934 int splitoff = 0; /* index in tword after last split */
|
|
|
6935 trystate_T *sp;
|
|
|
6936 int newscore;
|
|
|
6937 langp_T *lp;
|
|
375
|
6938 char_u *byts, *fbyts, *pbyts;
|
|
|
6939 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
6940 int depth;
|
|
330
|
6941 int c, c2, c3;
|
|
390
|
6942 int n;
|
|
323
|
6943 int flags;
|
|
|
6944 garray_T *gap;
|
|
324
|
6945 idx_T arridx;
|
|
323
|
6946 int len;
|
|
|
6947 char_u *p;
|
|
|
6948 fromto_T *ftp;
|
|
330
|
6949 int fl = 0, tl;
|
|
346
|
6950 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
323
|
6951
|
|
|
6952 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
6953 * to find matches (esp. REP items). Append some more text, changing
|
|
|
6954 * chars after the bad word may help. */
|
|
323
|
6955 STRCPY(fword, su->su_fbadword);
|
|
346
|
6956 n = STRLEN(fword);
|
|
|
6957 p = su->su_badptr + su->su_badlen;
|
|
|
6958 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
6959
|
|
|
6960 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
6961 lp->lp_slang != NULL; ++lp)
|
|
|
6962 {
|
|
|
6963 /*
|
|
|
6964 * Go through the whole case-fold tree, try changes at each node.
|
|
|
6965 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
6966 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
6967 * word).
|
|
|
6968 */
|
|
|
6969 depth = 0;
|
|
375
|
6970 sp = &stack[0];
|
|
|
6971 sp->ts_state = STATE_START;
|
|
|
6972 sp->ts_score = 0;
|
|
|
6973 sp->ts_curi = 1;
|
|
|
6974 sp->ts_fidx = 0;
|
|
|
6975 sp->ts_fidxtry = 0;
|
|
|
6976 sp->ts_twordlen = 0;
|
|
|
6977 sp->ts_arridx = 0;
|
|
330
|
6978 #ifdef FEAT_MBYTE
|
|
375
|
6979 sp->ts_tcharlen = 0;
|
|
330
|
6980 #endif
|
|
|
6981
|
|
|
6982 /*
|
|
375
|
6983 * When there are postponed prefixes we need to use these first. At
|
|
|
6984 * the end of the prefix we continue in the case-fold tree.
|
|
|
6985 */
|
|
|
6986 fbyts = lp->lp_slang->sl_fbyts;
|
|
|
6987 fidxs = lp->lp_slang->sl_fidxs;
|
|
|
6988 pbyts = lp->lp_slang->sl_pbyts;
|
|
|
6989 pidxs = lp->lp_slang->sl_pidxs;
|
|
|
6990 if (pbyts != NULL)
|
|
|
6991 {
|
|
|
6992 byts = pbyts;
|
|
|
6993 idxs = pidxs;
|
|
|
6994 sp->ts_prefixdepth = PREFIXTREE;
|
|
|
6995 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
6996 }
|
|
|
6997 else
|
|
|
6998 {
|
|
|
6999 byts = fbyts;
|
|
|
7000 idxs = fidxs;
|
|
|
7001 sp->ts_prefixdepth = NOPREFIX;
|
|
|
7002 }
|
|
|
7003
|
|
|
7004 /*
|
|
330
|
7005 * Loop to find all suggestions. At each round we either:
|
|
|
7006 * - For the current state try one operation, advance "ts_curi",
|
|
|
7007 * increase "depth".
|
|
|
7008 * - When a state is done go to the next, set "ts_state".
|
|
|
7009 * - When all states are tried decrease "depth".
|
|
|
7010 */
|
|
323
|
7011 while (depth >= 0 && !got_int)
|
|
|
7012 {
|
|
|
7013 sp = &stack[depth];
|
|
|
7014 switch (sp->ts_state)
|
|
|
7015 {
|
|
|
7016 case STATE_START:
|
|
375
|
7017 case STATE_NOPREFIX:
|
|
323
|
7018 /*
|
|
|
7019 * Start of node: Deal with NUL bytes, which means
|
|
|
7020 * tword[] may end here.
|
|
|
7021 */
|
|
|
7022 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
7023 len = byts[arridx]; /* bytes in this node */
|
|
|
7024 arridx += sp->ts_curi; /* index of current byte */
|
|
|
7025
|
|
375
|
7026 if (sp->ts_prefixdepth == PREFIXTREE)
|
|
|
7027 {
|
|
|
7028 /* Skip over the NUL bytes, we use them later. */
|
|
|
7029 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
7030 ;
|
|
|
7031 sp->ts_curi += n;
|
|
|
7032
|
|
390
|
7033 /* Always past NUL bytes now. */
|
|
|
7034 n = (int)sp->ts_state;
|
|
|
7035 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7036 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
7037
|
|
375
|
7038 /* At end of a prefix or at start of prefixtree: check for
|
|
|
7039 * following word. */
|
|
390
|
7040 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
7041 {
|
|
455
|
7042 /* Set su->su_badflags to the caps type at this
|
|
|
7043 * position. Use the caps type until here for the
|
|
|
7044 * prefix itself. */
|
|
|
7045 #ifdef FEAT_MBYTE
|
|
|
7046 if (has_mbyte)
|
|
|
7047 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
7048 else
|
|
|
7049 #endif
|
|
|
7050 n = sp->ts_fidx;
|
|
|
7051 flags = captype(su->su_badptr, su->su_badptr + n);
|
|
|
7052 su->su_badflags = captype(su->su_badptr + n,
|
|
|
7053 su->su_badptr + su->su_badlen);
|
|
|
7054
|
|
375
|
7055 ++depth;
|
|
|
7056 stack[depth] = stack[depth - 1];
|
|
|
7057 sp = &stack[depth];
|
|
|
7058 sp->ts_prefixdepth = depth - 1;
|
|
|
7059 byts = fbyts;
|
|
|
7060 idxs = fidxs;
|
|
|
7061 sp->ts_state = STATE_START;
|
|
|
7062 sp->ts_curi = 1; /* start just after length byte */
|
|
|
7063 sp->ts_arridx = 0;
|
|
|
7064
|
|
455
|
7065 /* Move the prefix to preword[] with the right case
|
|
|
7066 * and make find_keepcap_word() works. */
|
|
|
7067 splitoff = sp->ts_twordlen;
|
|
|
7068 tword[splitoff] = NUL;
|
|
|
7069 make_case_word(tword, preword, flags);
|
|
|
7070 prewordlen = STRLEN(preword);
|
|
375
|
7071 }
|
|
|
7072 break;
|
|
|
7073 }
|
|
|
7074
|
|
346
|
7075 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
7076 {
|
|
|
7077 /* Past bytes in node and/or past NUL bytes. */
|
|
|
7078 sp->ts_state = STATE_ENDNUL;
|
|
455
|
7079 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7080 break;
|
|
|
7081 }
|
|
|
7082
|
|
|
7083 /*
|
|
|
7084 * End of word in tree.
|
|
|
7085 */
|
|
|
7086 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
7087
|
|
324
|
7088 flags = (int)idxs[arridx];
|
|
323
|
7089
|
|
375
|
7090 if (sp->ts_prefixdepth < MAXWLEN)
|
|
|
7091 {
|
|
|
7092 /* There was a prefix before the word. Check that the
|
|
|
7093 * prefix can be used with this word. */
|
|
|
7094 /* Count the length of the NULs in the prefix. If there
|
|
|
7095 * are none this must be the first try without a prefix.
|
|
|
7096 */
|
|
|
7097 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
7098 len = pbyts[n++];
|
|
|
7099 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
7100 ;
|
|
|
7101 if (c > 0)
|
|
|
7102 {
|
|
390
|
7103 /* The prefix ID is stored three bytes above the
|
|
|
7104 * flags. */
|
|
|
7105 c = valid_word_prefix(c, n, flags,
|
|
455
|
7106 tword + splitoff, lp->lp_slang, FALSE);
|
|
375
|
7107 if (c == 0)
|
|
|
7108 break;
|
|
|
7109
|
|
|
7110 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
7111 if (c & WF_RAREPFX)
|
|
|
7112 flags |= WF_RARE;
|
|
|
7113 }
|
|
|
7114 }
|
|
|
7115
|
|
323
|
7116 /*
|
|
|
7117 * Form the word with proper case in preword.
|
|
|
7118 * If there is a word from a previous split, append.
|
|
|
7119 */
|
|
|
7120 tword[sp->ts_twordlen] = NUL;
|
|
|
7121 if (flags & WF_KEEPCAP)
|
|
|
7122 /* Must find the word in the keep-case tree. */
|
|
|
7123 find_keepcap_word(lp->lp_slang, tword + splitoff,
|
|
|
7124 preword + prewordlen);
|
|
|
7125 else
|
|
346
|
7126 {
|
|
323
|
7127 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
7128 * use that for the goodword too. But if the badword is
|
|
|
7129 * allcap and it's only one char long use onecap. */
|
|
|
7130 c = su->su_badflags;
|
|
|
7131 if ((c & WF_ALLCAP)
|
|
|
7132 #ifdef FEAT_MBYTE
|
|
|
7133 && su->su_badlen == mb_ptr2len_check(su->su_badptr)
|
|
|
7134 #else
|
|
|
7135 && su->su_badlen == 1
|
|
|
7136 #endif
|
|
|
7137 )
|
|
|
7138 c = WF_ONECAP;
|
|
323
|
7139 make_case_word(tword + splitoff,
|
|
346
|
7140 preword + prewordlen, flags | c);
|
|
|
7141 }
|
|
323
|
7142
|
|
|
7143 /* Don't use a banned word. It may appear again as a good
|
|
|
7144 * word, thus remember it. */
|
|
|
7145 if (flags & WF_BANNED)
|
|
|
7146 {
|
|
|
7147 add_banned(su, preword + prewordlen);
|
|
|
7148 break;
|
|
|
7149 }
|
|
372
|
7150 if (was_banned(su, preword + prewordlen)
|
|
|
7151 || was_banned(su, preword))
|
|
323
|
7152 break;
|
|
|
7153
|
|
|
7154 newscore = 0;
|
|
|
7155 if ((flags & WF_REGION)
|
|
390
|
7156 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
7157 newscore += SCORE_REGION;
|
|
|
7158 if (flags & WF_RARE)
|
|
|
7159 newscore += SCORE_RARE;
|
|
|
7160
|
|
346
|
7161 if (!spell_valid_case(su->su_badflags,
|
|
323
|
7162 captype(preword + prewordlen, NULL)))
|
|
|
7163 newscore += SCORE_ICASE;
|
|
|
7164
|
|
346
|
7165 if ((fword[sp->ts_fidx] == NUL
|
|
376
|
7166 || !spell_iswordp(fword + sp->ts_fidx, curbuf))
|
|
346
|
7167 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
7168 {
|
|
366
|
7169 /* The badword also ends: add suggestions. Give a penalty
|
|
|
7170 * when changing non-word char to word char, e.g., "thes,"
|
|
|
7171 * -> "these". */
|
|
|
7172 p = fword + sp->ts_fidx;
|
|
|
7173 #ifdef FEAT_MBYTE
|
|
|
7174 if (has_mbyte)
|
|
|
7175 mb_ptr_back(fword, p);
|
|
|
7176 else
|
|
|
7177 #endif
|
|
|
7178 --p;
|
|
376
|
7179 if (!spell_iswordp(p, curbuf))
|
|
366
|
7180 {
|
|
|
7181 p = preword + STRLEN(preword);
|
|
|
7182 #ifdef FEAT_MBYTE
|
|
|
7183 if (has_mbyte)
|
|
|
7184 mb_ptr_back(preword, p);
|
|
|
7185 else
|
|
|
7186 #endif
|
|
|
7187 --p;
|
|
376
|
7188 if (spell_iswordp(p, curbuf))
|
|
366
|
7189 newscore += SCORE_NONWORD;
|
|
|
7190 }
|
|
|
7191
|
|
344
|
7192 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
7193 sp->ts_fidx - repextra,
|
|
351
|
7194 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
7195 }
|
|
330
|
7196 else if (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
7197 #ifdef FEAT_MBYTE
|
|
|
7198 /* Don't split halfway a character. */
|
|
|
7199 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
7200 #endif
|
|
|
7201 )
|
|
323
|
7202 {
|
|
|
7203 /* The word in the tree ends but the badword
|
|
|
7204 * continues: try inserting a space and check that a valid
|
|
|
7205 * words starts at fword[sp->ts_fidx]. */
|
|
|
7206 if (try_deeper(su, stack, depth, newscore + SCORE_SPLIT))
|
|
|
7207 {
|
|
|
7208 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
|
7209 sp->ts_save_prewordlen = prewordlen;
|
|
346
|
7210 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
7211 sp->ts_save_splitoff = splitoff;
|
|
376
|
7212 sp->ts_state = STATE_SPLITUNDO;
|
|
|
7213
|
|
|
7214 ++depth;
|
|
|
7215 sp = &stack[depth];
|
|
323
|
7216
|
|
|
7217 /* Append a space to preword. */
|
|
|
7218 STRCAT(preword, " ");
|
|
|
7219 prewordlen = STRLEN(preword);
|
|
|
7220 splitoff = sp->ts_twordlen;
|
|
376
|
7221
|
|
|
7222 /* If the badword has a non-word character at this
|
|
|
7223 * position skip it. That means replacing the
|
|
|
7224 * non-word character with a space. */
|
|
|
7225 if (!spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
7226 {
|
|
|
7227 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
7228 #ifdef FEAT_MBYTE
|
|
|
7229 if (has_mbyte)
|
|
|
7230 sp->ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
|
7231 else
|
|
|
7232 #endif
|
|
|
7233 ++sp->ts_fidx;
|
|
|
7234 }
|
|
455
|
7235
|
|
|
7236 /* set su->su_badflags to the caps type at this
|
|
|
7237 * position */
|
|
|
7238
|
|
324
|
7239 #ifdef FEAT_MBYTE
|
|
|
7240 if (has_mbyte)
|
|
455
|
7241 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
7242 else
|
|
|
7243 #endif
|
|
455
|
7244 n = sp->ts_fidx;
|
|
|
7245 su->su_badflags = captype(su->su_badptr + n,
|
|
|
7246 su->su_badptr + su->su_badlen);
|
|
323
|
7247
|
|
|
7248 /* Restart at top of the tree. */
|
|
376
|
7249 sp->ts_arridx = 0;
|
|
323
|
7250 }
|
|
|
7251 }
|
|
|
7252 break;
|
|
|
7253
|
|
|
7254 case STATE_SPLITUNDO:
|
|
346
|
7255 /* Undo the changes done for word split. */
|
|
|
7256 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
7257 splitoff = sp->ts_save_splitoff;
|
|
|
7258 prewordlen = sp->ts_save_prewordlen;
|
|
|
7259
|
|
|
7260 /* Continue looking for NUL bytes. */
|
|
|
7261 sp->ts_state = STATE_START;
|
|
|
7262 break;
|
|
|
7263
|
|
|
7264 case STATE_ENDNUL:
|
|
|
7265 /* Past the NUL bytes in the node. */
|
|
455
|
7266 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
7267 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
7268 {
|
|
|
7269 /* The badword ends, can't use the bytes in this node. */
|
|
|
7270 sp->ts_state = STATE_DEL;
|
|
|
7271 break;
|
|
|
7272 }
|
|
|
7273 sp->ts_state = STATE_PLAIN;
|
|
|
7274 /*FALLTHROUGH*/
|
|
|
7275
|
|
|
7276 case STATE_PLAIN:
|
|
|
7277 /*
|
|
|
7278 * Go over all possible bytes at this node, add each to
|
|
|
7279 * tword[] and use child node. "ts_curi" is the index.
|
|
|
7280 */
|
|
|
7281 arridx = sp->ts_arridx;
|
|
|
7282 if (sp->ts_curi > byts[arridx])
|
|
|
7283 {
|
|
|
7284 /* Done all bytes at this node, do next state. When still
|
|
|
7285 * at already changed bytes skip the other tricks. */
|
|
|
7286 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7287 sp->ts_state = STATE_DEL;
|
|
|
7288 else
|
|
|
7289 sp->ts_state = STATE_FINAL;
|
|
|
7290 }
|
|
|
7291 else
|
|
|
7292 {
|
|
|
7293 arridx += sp->ts_curi++;
|
|
|
7294 c = byts[arridx];
|
|
|
7295
|
|
|
7296 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
7297 * the byte in the bad word adjust the score. But don't
|
|
|
7298 * even try when the byte was already changed. */
|
|
330
|
7299 if (c == fword[sp->ts_fidx]
|
|
|
7300 #ifdef FEAT_MBYTE
|
|
|
7301 || (sp->ts_tcharlen > 0
|
|
|
7302 && sp->ts_isdiff != DIFF_NONE)
|
|
|
7303 #endif
|
|
|
7304 )
|
|
323
|
7305 newscore = 0;
|
|
|
7306 else
|
|
|
7307 newscore = SCORE_SUBST;
|
|
|
7308 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
7309 && try_deeper(su, stack, depth, newscore))
|
|
|
7310 {
|
|
|
7311 ++depth;
|
|
330
|
7312 sp = &stack[depth];
|
|
|
7313 ++sp->ts_fidx;
|
|
|
7314 tword[sp->ts_twordlen++] = c;
|
|
|
7315 sp->ts_arridx = idxs[arridx];
|
|
|
7316 #ifdef FEAT_MBYTE
|
|
|
7317 if (newscore == SCORE_SUBST)
|
|
|
7318 sp->ts_isdiff = DIFF_YES;
|
|
|
7319 if (has_mbyte)
|
|
|
7320 {
|
|
|
7321 /* Multi-byte characters are a bit complicated to
|
|
|
7322 * handle: They differ when any of the bytes
|
|
|
7323 * differ and then their length may also differ. */
|
|
|
7324 if (sp->ts_tcharlen == 0)
|
|
|
7325 {
|
|
|
7326 /* First byte. */
|
|
|
7327 sp->ts_tcharidx = 0;
|
|
|
7328 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
7329 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
7330 sp->ts_isdiff = (newscore != 0)
|
|
|
7331 ? DIFF_YES : DIFF_NONE;
|
|
|
7332 }
|
|
|
7333 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
7334 /* When inserting trail bytes don't advance in
|
|
|
7335 * the bad word. */
|
|
|
7336 --sp->ts_fidx;
|
|
|
7337 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
7338 {
|
|
|
7339 /* Last byte of character. */
|
|
|
7340 if (sp->ts_isdiff == DIFF_YES)
|
|
|
7341 {
|
|
|
7342 /* Correct ts_fidx for the byte length of
|
|
|
7343 * the character (we didn't check that
|
|
|
7344 * before). */
|
|
|
7345 sp->ts_fidx = sp->ts_fcharstart
|
|
|
7346 + MB_BYTE2LEN(
|
|
|
7347 fword[sp->ts_fcharstart]);
|
|
|
7348
|
|
|
7349 /* For a similar character adjust score
|
|
|
7350 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
|
7351 if (lp->lp_slang->sl_has_map
|
|
|
7352 && similar_chars(lp->lp_slang,
|
|
|
7353 mb_ptr2char(tword
|
|
|
7354 + sp->ts_twordlen
|
|
|
7355 - sp->ts_tcharlen),
|
|
|
7356 mb_ptr2char(fword
|
|
|
7357 + sp->ts_fcharstart)))
|
|
|
7358 sp->ts_score -=
|
|
|
7359 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7360 }
|
|
358
|
7361 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
7362 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
7363 {
|
|
|
7364 /* If the previous character was the same,
|
|
|
7365 * thus doubling a character, give a bonus
|
|
|
7366 * to the score. */
|
|
|
7367 p = tword + sp->ts_twordlen
|
|
|
7368 - sp->ts_tcharlen;
|
|
|
7369 c = mb_ptr2char(p);
|
|
|
7370 mb_ptr_back(tword, p);
|
|
|
7371 if (c == mb_ptr2char(p))
|
|
|
7372 sp->ts_score -= SCORE_INS
|
|
|
7373 - SCORE_INSDUP;
|
|
|
7374 }
|
|
330
|
7375
|
|
|
7376 /* Starting a new char, reset the length. */
|
|
|
7377 sp->ts_tcharlen = 0;
|
|
|
7378 }
|
|
|
7379 }
|
|
|
7380 else
|
|
|
7381 #endif
|
|
|
7382 {
|
|
|
7383 /* If we found a similar char adjust the score.
|
|
|
7384 * We do this after calling try_deeper() because
|
|
|
7385 * it's slow. */
|
|
|
7386 if (newscore != 0
|
|
|
7387 && lp->lp_slang->sl_has_map
|
|
|
7388 && similar_chars(lp->lp_slang,
|
|
|
7389 c, fword[sp->ts_fidx - 1]))
|
|
|
7390 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
7391 }
|
|
323
|
7392 }
|
|
|
7393 }
|
|
|
7394 break;
|
|
|
7395
|
|
|
7396 case STATE_DEL:
|
|
330
|
7397 #ifdef FEAT_MBYTE
|
|
|
7398 /* When past the first byte of a multi-byte char don't try
|
|
|
7399 * delete/insert/swap a character. */
|
|
|
7400 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
7401 {
|
|
|
7402 sp->ts_state = STATE_FINAL;
|
|
|
7403 break;
|
|
|
7404 }
|
|
|
7405 #endif
|
|
|
7406 /*
|
|
|
7407 * Try skipping one character in the bad word (delete it).
|
|
|
7408 */
|
|
323
|
7409 sp->ts_state = STATE_INS;
|
|
|
7410 sp->ts_curi = 1;
|
|
|
7411 if (fword[sp->ts_fidx] != NUL
|
|
|
7412 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
7413 {
|
|
|
7414 ++depth;
|
|
358
|
7415
|
|
|
7416 /* Advance over the character in fword[]. Give a bonus to
|
|
|
7417 * the score if the same character is following "nn" ->
|
|
|
7418 * "n". */
|
|
330
|
7419 #ifdef FEAT_MBYTE
|
|
|
7420 if (has_mbyte)
|
|
358
|
7421 {
|
|
|
7422 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
7423 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
358
|
7424 if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
|
7425 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7426 }
|
|
330
|
7427 else
|
|
|
7428 #endif
|
|
358
|
7429 {
|
|
330
|
7430 ++stack[depth].ts_fidx;
|
|
358
|
7431 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
7432 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
7433 }
|
|
323
|
7434 break;
|
|
|
7435 }
|
|
|
7436 /*FALLTHROUGH*/
|
|
|
7437
|
|
|
7438 case STATE_INS:
|
|
330
|
7439 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
7440 * node. */
|
|
|
7441 n = sp->ts_arridx;
|
|
|
7442 if (sp->ts_curi > byts[n])
|
|
|
7443 {
|
|
|
7444 /* Done all bytes at this node, do next state. */
|
|
|
7445 sp->ts_state = STATE_SWAP;
|
|
|
7446 }
|
|
|
7447 else
|
|
|
7448 {
|
|
330
|
7449 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
7450 n += sp->ts_curi++;
|
|
|
7451 c = byts[n];
|
|
|
7452 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
7453 {
|
|
|
7454 ++depth;
|
|
330
|
7455 sp = &stack[depth];
|
|
|
7456 tword[sp->ts_twordlen++] = c;
|
|
|
7457 sp->ts_arridx = idxs[n];
|
|
|
7458 #ifdef FEAT_MBYTE
|
|
|
7459 if (has_mbyte)
|
|
|
7460 {
|
|
|
7461 fl = MB_BYTE2LEN(c);
|
|
|
7462 if (fl > 1)
|
|
|
7463 {
|
|
|
7464 /* There are following bytes for the same
|
|
|
7465 * character. We must find all bytes before
|
|
|
7466 * trying delete/insert/swap/etc. */
|
|
|
7467 sp->ts_tcharlen = fl;
|
|
|
7468 sp->ts_tcharidx = 1;
|
|
|
7469 sp->ts_isdiff = DIFF_INSERT;
|
|
|
7470 }
|
|
|
7471 }
|
|
358
|
7472 else
|
|
|
7473 fl = 1;
|
|
|
7474 if (fl == 1)
|
|
330
|
7475 #endif
|
|
358
|
7476 {
|
|
|
7477 /* If the previous character was the same, thus
|
|
|
7478 * doubling a character, give a bonus to the
|
|
|
7479 * score. */
|
|
|
7480 if (sp->ts_twordlen >= 2
|
|
|
7481 && tword[sp->ts_twordlen - 2] == c)
|
|
|
7482 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
7483 }
|
|
323
|
7484 }
|
|
|
7485 }
|
|
|
7486 break;
|
|
|
7487
|
|
|
7488 case STATE_SWAP:
|
|
330
|
7489 /*
|
|
|
7490 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
7491 * We change "fword" here, it's changed back afterwards.
|
|
|
7492 */
|
|
|
7493 p = fword + sp->ts_fidx;
|
|
|
7494 c = *p;
|
|
|
7495 if (c == NUL)
|
|
|
7496 {
|
|
|
7497 /* End of word, can't swap or replace. */
|
|
|
7498 sp->ts_state = STATE_FINAL;
|
|
|
7499 break;
|
|
|
7500 }
|
|
|
7501 #ifdef FEAT_MBYTE
|
|
|
7502 if (has_mbyte)
|
|
|
7503 {
|
|
|
7504 n = mb_ptr2len_check(p);
|
|
|
7505 c = mb_ptr2char(p);
|
|
|
7506 c2 = mb_ptr2char(p + n);
|
|
|
7507 }
|
|
|
7508 else
|
|
|
7509 #endif
|
|
|
7510 c2 = p[1];
|
|
|
7511 if (c == c2)
|
|
323
|
7512 {
|
|
330
|
7513 /* Characters are identical, swap won't do anything. */
|
|
|
7514 sp->ts_state = STATE_SWAP3;
|
|
|
7515 break;
|
|
|
7516 }
|
|
|
7517 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
7518 {
|
|
|
7519 sp->ts_state = STATE_UNSWAP;
|
|
323
|
7520 ++depth;
|
|
330
|
7521 #ifdef FEAT_MBYTE
|
|
|
7522 if (has_mbyte)
|
|
|
7523 {
|
|
|
7524 fl = mb_char2len(c2);
|
|
|
7525 mch_memmove(p, p + n, fl);
|
|
|
7526 mb_char2bytes(c, p + fl);
|
|
|
7527 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7528 }
|
|
|
7529 else
|
|
|
7530 #endif
|
|
|
7531 {
|
|
|
7532 p[0] = c2;
|
|
|
7533 p[1] = c;
|
|
|
7534 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
7535 }
|
|
323
|
7536 }
|
|
|
7537 else
|
|
|
7538 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
7539 sp->ts_state = STATE_REP_INI;
|
|
|
7540 break;
|
|
|
7541
|
|
330
|
7542 case STATE_UNSWAP:
|
|
|
7543 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
7544 p = fword + sp->ts_fidx;
|
|
|
7545 #ifdef FEAT_MBYTE
|
|
|
7546 if (has_mbyte)
|
|
|
7547 {
|
|
|
7548 n = MB_BYTE2LEN(*p);
|
|
|
7549 c = mb_ptr2char(p + n);
|
|
|
7550 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
7551 mb_char2bytes(c, p);
|
|
|
7552 }
|
|
|
7553 else
|
|
|
7554 #endif
|
|
|
7555 {
|
|
|
7556 c = *p;
|
|
|
7557 *p = p[1];
|
|
|
7558 p[1] = c;
|
|
|
7559 }
|
|
|
7560 /*FALLTHROUGH*/
|
|
|
7561
|
|
|
7562 case STATE_SWAP3:
|
|
323
|
7563 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
7564 * "fword" here, it's changed back afterwards. */
|
|
|
7565 p = fword + sp->ts_fidx;
|
|
|
7566 #ifdef FEAT_MBYTE
|
|
|
7567 if (has_mbyte)
|
|
|
7568 {
|
|
|
7569 n = mb_ptr2len_check(p);
|
|
|
7570 c = mb_ptr2char(p);
|
|
|
7571 fl = mb_ptr2len_check(p + n);
|
|
|
7572 c2 = mb_ptr2char(p + n);
|
|
|
7573 c3 = mb_ptr2char(p + n + fl);
|
|
|
7574 }
|
|
|
7575 else
|
|
|
7576 #endif
|
|
323
|
7577 {
|
|
330
|
7578 c = *p;
|
|
|
7579 c2 = p[1];
|
|
|
7580 c3 = p[2];
|
|
|
7581 }
|
|
|
7582
|
|
|
7583 /* When characters are identical: "121" then SWAP3 result is
|
|
|
7584 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
7585 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
7586 * swapping. Also skip when c3 is NUL. */
|
|
|
7587 if (c == c3 || c3 == NUL)
|
|
|
7588 {
|
|
|
7589 sp->ts_state = STATE_REP_INI;
|
|
|
7590 break;
|
|
|
7591 }
|
|
|
7592 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7593 {
|
|
|
7594 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
7595 ++depth;
|
|
330
|
7596 #ifdef FEAT_MBYTE
|
|
|
7597 if (has_mbyte)
|
|
|
7598 {
|
|
|
7599 tl = mb_char2len(c3);
|
|
|
7600 mch_memmove(p, p + n + fl, tl);
|
|
|
7601 mb_char2bytes(c2, p + tl);
|
|
|
7602 mb_char2bytes(c, p + fl + tl);
|
|
|
7603 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
7604 }
|
|
|
7605 else
|
|
|
7606 #endif
|
|
|
7607 {
|
|
|
7608 p[0] = p[2];
|
|
|
7609 p[2] = c;
|
|
|
7610 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7611 }
|
|
323
|
7612 }
|
|
|
7613 else
|
|
|
7614 sp->ts_state = STATE_REP_INI;
|
|
|
7615 break;
|
|
|
7616
|
|
330
|
7617 case STATE_UNSWAP3:
|
|
|
7618 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
7619 p = fword + sp->ts_fidx;
|
|
|
7620 #ifdef FEAT_MBYTE
|
|
|
7621 if (has_mbyte)
|
|
|
7622 {
|
|
|
7623 n = MB_BYTE2LEN(*p);
|
|
|
7624 c2 = mb_ptr2char(p + n);
|
|
|
7625 fl = MB_BYTE2LEN(p[n]);
|
|
|
7626 c = mb_ptr2char(p + n + fl);
|
|
|
7627 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
7628 mch_memmove(p + fl + tl, p, n);
|
|
|
7629 mb_char2bytes(c, p);
|
|
|
7630 mb_char2bytes(c2, p + tl);
|
|
|
7631 }
|
|
|
7632 else
|
|
|
7633 #endif
|
|
|
7634 {
|
|
|
7635 c = *p;
|
|
|
7636 *p = p[2];
|
|
|
7637 p[2] = c;
|
|
|
7638 }
|
|
346
|
7639
|
|
330
|
7640 /* Rotate three characters left: "123" -> "231". We change
|
|
|
7641 * "fword" here, it's changed back afterwards. */
|
|
323
|
7642 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7643 {
|
|
330
|
7644 sp->ts_state = STATE_UNROT3L;
|
|
323
|
7645 ++depth;
|
|
330
|
7646 p = fword + sp->ts_fidx;
|
|
|
7647 #ifdef FEAT_MBYTE
|
|
|
7648 if (has_mbyte)
|
|
|
7649 {
|
|
|
7650 n = mb_ptr2len_check(p);
|
|
|
7651 c = mb_ptr2char(p);
|
|
|
7652 fl = mb_ptr2len_check(p + n);
|
|
|
7653 fl += mb_ptr2len_check(p + n + fl);
|
|
|
7654 mch_memmove(p, p + n, fl);
|
|
|
7655 mb_char2bytes(c, p + fl);
|
|
|
7656 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
7657 }
|
|
|
7658 else
|
|
|
7659 #endif
|
|
|
7660 {
|
|
|
7661 c = *p;
|
|
|
7662 *p = p[1];
|
|
|
7663 p[1] = p[2];
|
|
|
7664 p[2] = c;
|
|
|
7665 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7666 }
|
|
323
|
7667 }
|
|
|
7668 else
|
|
|
7669 sp->ts_state = STATE_REP_INI;
|
|
|
7670 break;
|
|
|
7671
|
|
330
|
7672 case STATE_UNROT3L:
|
|
346
|
7673 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
7674 p = fword + sp->ts_fidx;
|
|
|
7675 #ifdef FEAT_MBYTE
|
|
|
7676 if (has_mbyte)
|
|
|
7677 {
|
|
|
7678 n = MB_BYTE2LEN(*p);
|
|
|
7679 n += MB_BYTE2LEN(p[n]);
|
|
|
7680 c = mb_ptr2char(p + n);
|
|
|
7681 tl = MB_BYTE2LEN(p[n]);
|
|
|
7682 mch_memmove(p + tl, p, n);
|
|
|
7683 mb_char2bytes(c, p);
|
|
|
7684 }
|
|
|
7685 else
|
|
|
7686 #endif
|
|
|
7687 {
|
|
|
7688 c = p[2];
|
|
|
7689 p[2] = p[1];
|
|
|
7690 p[1] = *p;
|
|
|
7691 *p = c;
|
|
|
7692 }
|
|
346
|
7693
|
|
323
|
7694 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
7695 * "fword" here, it's changed back afterwards. */
|
|
323
|
7696 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
7697 {
|
|
330
|
7698 sp->ts_state = STATE_UNROT3R;
|
|
323
|
7699 ++depth;
|
|
330
|
7700 p = fword + sp->ts_fidx;
|
|
|
7701 #ifdef FEAT_MBYTE
|
|
|
7702 if (has_mbyte)
|
|
|
7703 {
|
|
|
7704 n = mb_ptr2len_check(p);
|
|
|
7705 n += mb_ptr2len_check(p + n);
|
|
|
7706 c = mb_ptr2char(p + n);
|
|
|
7707 tl = mb_ptr2len_check(p + n);
|
|
|
7708 mch_memmove(p + tl, p, n);
|
|
|
7709 mb_char2bytes(c, p);
|
|
|
7710 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
7711 }
|
|
|
7712 else
|
|
|
7713 #endif
|
|
|
7714 {
|
|
|
7715 c = p[2];
|
|
|
7716 p[2] = p[1];
|
|
|
7717 p[1] = *p;
|
|
|
7718 *p = c;
|
|
|
7719 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
7720 }
|
|
323
|
7721 }
|
|
|
7722 else
|
|
|
7723 sp->ts_state = STATE_REP_INI;
|
|
|
7724 break;
|
|
|
7725
|
|
330
|
7726 case STATE_UNROT3R:
|
|
346
|
7727 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
7728 p = fword + sp->ts_fidx;
|
|
|
7729 #ifdef FEAT_MBYTE
|
|
|
7730 if (has_mbyte)
|
|
|
7731 {
|
|
|
7732 c = mb_ptr2char(p);
|
|
|
7733 tl = MB_BYTE2LEN(*p);
|
|
|
7734 n = MB_BYTE2LEN(p[tl]);
|
|
|
7735 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
7736 mch_memmove(p, p + tl, n);
|
|
|
7737 mb_char2bytes(c, p + n);
|
|
|
7738 }
|
|
|
7739 else
|
|
|
7740 #endif
|
|
|
7741 {
|
|
|
7742 c = *p;
|
|
|
7743 *p = p[1];
|
|
|
7744 p[1] = p[2];
|
|
|
7745 p[2] = c;
|
|
|
7746 }
|
|
323
|
7747 /*FALLTHROUGH*/
|
|
|
7748
|
|
|
7749 case STATE_REP_INI:
|
|
|
7750 /* Check if matching with REP items from the .aff file would
|
|
|
7751 * work. Quickly skip if there are no REP items or the score
|
|
|
7752 * is going to be too high anyway. */
|
|
|
7753 gap = &lp->lp_slang->sl_rep;
|
|
|
7754 if (gap->ga_len == 0
|
|
|
7755 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
7756 {
|
|
|
7757 sp->ts_state = STATE_FINAL;
|
|
|
7758 break;
|
|
|
7759 }
|
|
|
7760
|
|
|
7761 /* Use the first byte to quickly find the first entry that
|
|
330
|
7762 * may match. If the index is -1 there is none. */
|
|
323
|
7763 sp->ts_curi = lp->lp_slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
7764 if (sp->ts_curi < 0)
|
|
|
7765 {
|
|
|
7766 sp->ts_state = STATE_FINAL;
|
|
|
7767 break;
|
|
|
7768 }
|
|
|
7769
|
|
|
7770 sp->ts_state = STATE_REP;
|
|
|
7771 /*FALLTHROUGH*/
|
|
|
7772
|
|
|
7773 case STATE_REP:
|
|
|
7774 /* Try matching with REP items from the .aff file. For each
|
|
330
|
7775 * match replace the characters and check if the resulting
|
|
|
7776 * word is valid. */
|
|
323
|
7777 p = fword + sp->ts_fidx;
|
|
|
7778
|
|
|
7779 gap = &lp->lp_slang->sl_rep;
|
|
|
7780 while (sp->ts_curi < gap->ga_len)
|
|
|
7781 {
|
|
|
7782 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
7783 if (*ftp->ft_from != *p)
|
|
|
7784 {
|
|
|
7785 /* past possible matching entries */
|
|
|
7786 sp->ts_curi = gap->ga_len;
|
|
|
7787 break;
|
|
|
7788 }
|
|
|
7789 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
7790 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
7791 {
|
|
|
7792 /* Need to undo this afterwards. */
|
|
|
7793 sp->ts_state = STATE_REP_UNDO;
|
|
|
7794
|
|
|
7795 /* Change the "from" to the "to" string. */
|
|
|
7796 ++depth;
|
|
|
7797 fl = STRLEN(ftp->ft_from);
|
|
|
7798 tl = STRLEN(ftp->ft_to);
|
|
|
7799 if (fl != tl)
|
|
346
|
7800 {
|
|
323
|
7801 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
7802 repextra += tl - fl;
|
|
|
7803 }
|
|
323
|
7804 mch_memmove(p, ftp->ft_to, tl);
|
|
|
7805 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
7806 #ifdef FEAT_MBYTE
|
|
|
7807 stack[depth].ts_tcharlen = 0;
|
|
|
7808 #endif
|
|
323
|
7809 break;
|
|
|
7810 }
|
|
|
7811 }
|
|
|
7812
|
|
|
7813 if (sp->ts_curi >= gap->ga_len)
|
|
|
7814 /* No (more) matches. */
|
|
|
7815 sp->ts_state = STATE_FINAL;
|
|
|
7816
|
|
|
7817 break;
|
|
|
7818
|
|
|
7819 case STATE_REP_UNDO:
|
|
|
7820 /* Undo a REP replacement and continue with the next one. */
|
|
|
7821 ftp = (fromto_T *)lp->lp_slang->sl_rep.ga_data
|
|
|
7822 + sp->ts_curi - 1;
|
|
|
7823 fl = STRLEN(ftp->ft_from);
|
|
|
7824 tl = STRLEN(ftp->ft_to);
|
|
|
7825 p = fword + sp->ts_fidx;
|
|
|
7826 if (fl != tl)
|
|
346
|
7827 {
|
|
323
|
7828 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
7829 repextra -= tl - fl;
|
|
|
7830 }
|
|
323
|
7831 mch_memmove(p, ftp->ft_from, fl);
|
|
|
7832 sp->ts_state = STATE_REP;
|
|
|
7833 break;
|
|
|
7834
|
|
|
7835 default:
|
|
|
7836 /* Did all possible states at this level, go up one level. */
|
|
|
7837 --depth;
|
|
344
|
7838
|
|
375
|
7839 if (depth >= 0 && stack[depth].ts_prefixdepth == PREFIXTREE)
|
|
|
7840 {
|
|
|
7841 /* Continue in or go back to the prefix tree. */
|
|
|
7842 byts = pbyts;
|
|
|
7843 idxs = pidxs;
|
|
|
7844 splitoff = 0;
|
|
|
7845 }
|
|
|
7846
|
|
344
|
7847 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
7848 line_breakcheck();
|
|
323
|
7849 }
|
|
|
7850 }
|
|
|
7851 }
|
|
|
7852 }
|
|
|
7853
|
|
|
7854 /*
|
|
|
7855 * Try going one level deeper in the tree.
|
|
|
7856 */
|
|
|
7857 static int
|
|
|
7858 try_deeper(su, stack, depth, score_add)
|
|
|
7859 suginfo_T *su;
|
|
|
7860 trystate_T *stack;
|
|
|
7861 int depth;
|
|
|
7862 int score_add;
|
|
|
7863 {
|
|
|
7864 int newscore;
|
|
|
7865
|
|
|
7866 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
7867 newscore = stack[depth].ts_score + score_add;
|
|
|
7868 if (newscore >= su->su_maxscore)
|
|
|
7869 return FALSE;
|
|
|
7870
|
|
330
|
7871 stack[depth + 1] = stack[depth];
|
|
323
|
7872 stack[depth + 1].ts_state = STATE_START;
|
|
|
7873 stack[depth + 1].ts_score = newscore;
|
|
|
7874 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
7875 return TRUE;
|
|
|
7876 }
|
|
|
7877
|
|
455
|
7878 #ifdef FEAT_MBYTE
|
|
|
7879 /*
|
|
|
7880 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
7881 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
7882 */
|
|
|
7883 static int
|
|
|
7884 nofold_len(fword, flen, word)
|
|
|
7885 char_u *fword;
|
|
|
7886 int flen;
|
|
|
7887 char_u *word;
|
|
|
7888 {
|
|
|
7889 char_u *p;
|
|
|
7890 int i = 0;
|
|
|
7891
|
|
|
7892 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
7893 ++i;
|
|
|
7894 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
7895 --i;
|
|
|
7896 return (int)(p - word);
|
|
|
7897 }
|
|
|
7898 #endif
|
|
|
7899
|
|
323
|
7900 /*
|
|
|
7901 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
7902 * words and put it in "kword".
|
|
|
7903 * Theoretically there could be several keep-case words that result in the
|
|
|
7904 * same case-folded word, but we only find one...
|
|
|
7905 */
|
|
|
7906 static void
|
|
|
7907 find_keepcap_word(slang, fword, kword)
|
|
|
7908 slang_T *slang;
|
|
|
7909 char_u *fword;
|
|
|
7910 char_u *kword;
|
|
|
7911 {
|
|
|
7912 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
7913 int depth;
|
|
324
|
7914 idx_T tryidx;
|
|
323
|
7915
|
|
|
7916 /* The following arrays are used at each depth in the tree. */
|
|
324
|
7917 idx_T arridx[MAXWLEN];
|
|
323
|
7918 int round[MAXWLEN];
|
|
|
7919 int fwordidx[MAXWLEN];
|
|
|
7920 int uwordidx[MAXWLEN];
|
|
|
7921 int kwordlen[MAXWLEN];
|
|
|
7922
|
|
|
7923 int flen, ulen;
|
|
|
7924 int l;
|
|
|
7925 int len;
|
|
|
7926 int c;
|
|
324
|
7927 idx_T lo, hi, m;
|
|
323
|
7928 char_u *p;
|
|
|
7929 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
7930 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
7931
|
|
|
7932 if (byts == NULL)
|
|
|
7933 {
|
|
|
7934 /* array is empty: "cannot happen" */
|
|
|
7935 *kword = NUL;
|
|
|
7936 return;
|
|
|
7937 }
|
|
|
7938
|
|
|
7939 /* Make an all-cap version of "fword". */
|
|
|
7940 allcap_copy(fword, uword);
|
|
|
7941
|
|
|
7942 /*
|
|
|
7943 * Each character needs to be tried both case-folded and upper-case.
|
|
|
7944 * All this gets very complicated if we keep in mind that changing case
|
|
|
7945 * may change the byte length of a multi-byte character...
|
|
|
7946 */
|
|
|
7947 depth = 0;
|
|
|
7948 arridx[0] = 0;
|
|
|
7949 round[0] = 0;
|
|
|
7950 fwordidx[0] = 0;
|
|
|
7951 uwordidx[0] = 0;
|
|
|
7952 kwordlen[0] = 0;
|
|
|
7953 while (depth >= 0)
|
|
|
7954 {
|
|
|
7955 if (fword[fwordidx[depth]] == NUL)
|
|
|
7956 {
|
|
|
7957 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
7958 * here we have found a match. */
|
|
|
7959 if (byts[arridx[depth] + 1] == 0)
|
|
|
7960 {
|
|
|
7961 kword[kwordlen[depth]] = NUL;
|
|
|
7962 return;
|
|
|
7963 }
|
|
|
7964
|
|
|
7965 /* kword is getting too long, continue one level up */
|
|
|
7966 --depth;
|
|
|
7967 }
|
|
|
7968 else if (++round[depth] > 2)
|
|
|
7969 {
|
|
|
7970 /* tried both fold-case and upper-case character, continue one
|
|
|
7971 * level up */
|
|
|
7972 --depth;
|
|
|
7973 }
|
|
|
7974 else
|
|
|
7975 {
|
|
|
7976 /*
|
|
|
7977 * round[depth] == 1: Try using the folded-case character.
|
|
|
7978 * round[depth] == 2: Try using the upper-case character.
|
|
|
7979 */
|
|
|
7980 #ifdef FEAT_MBYTE
|
|
|
7981 if (has_mbyte)
|
|
|
7982 {
|
|
|
7983 flen = mb_ptr2len_check(fword + fwordidx[depth]);
|
|
|
7984 ulen = mb_ptr2len_check(uword + uwordidx[depth]);
|
|
|
7985 }
|
|
|
7986 else
|
|
|
7987 #endif
|
|
|
7988 ulen = flen = 1;
|
|
|
7989 if (round[depth] == 1)
|
|
|
7990 {
|
|
|
7991 p = fword + fwordidx[depth];
|
|
|
7992 l = flen;
|
|
|
7993 }
|
|
|
7994 else
|
|
|
7995 {
|
|
|
7996 p = uword + uwordidx[depth];
|
|
|
7997 l = ulen;
|
|
|
7998 }
|
|
|
7999
|
|
|
8000 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
8001 {
|
|
|
8002 /* Perform a binary search in the list of accepted bytes. */
|
|
|
8003 len = byts[tryidx++];
|
|
|
8004 c = *p++;
|
|
|
8005 lo = tryidx;
|
|
|
8006 hi = tryidx + len - 1;
|
|
|
8007 while (lo < hi)
|
|
|
8008 {
|
|
|
8009 m = (lo + hi) / 2;
|
|
|
8010 if (byts[m] > c)
|
|
|
8011 hi = m - 1;
|
|
|
8012 else if (byts[m] < c)
|
|
|
8013 lo = m + 1;
|
|
|
8014 else
|
|
|
8015 {
|
|
|
8016 lo = hi = m;
|
|
|
8017 break;
|
|
|
8018 }
|
|
|
8019 }
|
|
|
8020
|
|
|
8021 /* Stop if there is no matching byte. */
|
|
|
8022 if (hi < lo || byts[lo] != c)
|
|
|
8023 break;
|
|
|
8024
|
|
|
8025 /* Continue at the child (if there is one). */
|
|
|
8026 tryidx = idxs[lo];
|
|
|
8027 }
|
|
|
8028
|
|
|
8029 if (l == 0)
|
|
|
8030 {
|
|
|
8031 /*
|
|
|
8032 * Found the matching char. Copy it to "kword" and go a
|
|
|
8033 * level deeper.
|
|
|
8034 */
|
|
|
8035 if (round[depth] == 1)
|
|
|
8036 {
|
|
|
8037 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
8038 flen);
|
|
|
8039 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
8040 }
|
|
|
8041 else
|
|
|
8042 {
|
|
|
8043 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
8044 ulen);
|
|
|
8045 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
8046 }
|
|
|
8047 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
8048 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
8049
|
|
|
8050 ++depth;
|
|
|
8051 arridx[depth] = tryidx;
|
|
|
8052 round[depth] = 0;
|
|
|
8053 }
|
|
|
8054 }
|
|
|
8055 }
|
|
|
8056
|
|
|
8057 /* Didn't find it: "cannot happen". */
|
|
|
8058 *kword = NUL;
|
|
|
8059 }
|
|
|
8060
|
|
|
8061 /*
|
|
344
|
8062 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
8063 * su->su_sga.
|
|
|
8064 */
|
|
|
8065 static void
|
|
|
8066 score_comp_sal(su)
|
|
|
8067 suginfo_T *su;
|
|
|
8068 {
|
|
|
8069 langp_T *lp;
|
|
|
8070 char_u badsound[MAXWLEN];
|
|
|
8071 int i;
|
|
|
8072 suggest_T *stp;
|
|
|
8073 suggest_T *sstp;
|
|
|
8074 int score;
|
|
|
8075
|
|
|
8076 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
8077 return;
|
|
|
8078
|
|
|
8079 /* Use the sound-folding of the first language that supports it. */
|
|
|
8080 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8081 lp->lp_slang != NULL; ++lp)
|
|
|
8082 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8083 {
|
|
|
8084 /* soundfold the bad word */
|
|
375
|
8085 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8086
|
|
|
8087 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8088 {
|
|
|
8089 stp = &SUG(su->su_ga, i);
|
|
|
8090
|
|
351
|
8091 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
8092 * sound-a-like score. */
|
|
|
8093 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
8094 if (score < SCORE_MAXMAX)
|
|
|
8095 {
|
|
|
8096 /* Add the suggestion. */
|
|
|
8097 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
8098 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
8099 if (sstp->st_word != NULL)
|
|
|
8100 {
|
|
|
8101 sstp->st_score = score;
|
|
|
8102 sstp->st_altscore = 0;
|
|
|
8103 sstp->st_orglen = stp->st_orglen;
|
|
|
8104 ++su->su_sga.ga_len;
|
|
|
8105 }
|
|
|
8106 }
|
|
|
8107 }
|
|
|
8108 break;
|
|
|
8109 }
|
|
|
8110 }
|
|
|
8111
|
|
|
8112 /*
|
|
|
8113 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
8114 * They are intwined.
|
|
|
8115 */
|
|
|
8116 static void
|
|
|
8117 score_combine(su)
|
|
|
8118 suginfo_T *su;
|
|
|
8119 {
|
|
|
8120 int i;
|
|
|
8121 int j;
|
|
|
8122 garray_T ga;
|
|
|
8123 garray_T *gap;
|
|
|
8124 langp_T *lp;
|
|
|
8125 suggest_T *stp;
|
|
|
8126 char_u *p;
|
|
|
8127 char_u badsound[MAXWLEN];
|
|
|
8128 int round;
|
|
|
8129
|
|
|
8130 /* Add the alternate score to su_ga. */
|
|
|
8131 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8132 lp->lp_slang != NULL; ++lp)
|
|
|
8133 {
|
|
|
8134 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8135 {
|
|
|
8136 /* soundfold the bad word */
|
|
375
|
8137 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
8138
|
|
|
8139 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8140 {
|
|
|
8141 stp = &SUG(su->su_ga, i);
|
|
351
|
8142 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
8143 badsound);
|
|
344
|
8144 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8145 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
8146 else
|
|
|
8147 stp->st_score = (stp->st_score * 3
|
|
|
8148 + stp->st_altscore) / 4;
|
|
|
8149 stp->st_salscore = FALSE;
|
|
|
8150 }
|
|
|
8151 break;
|
|
|
8152 }
|
|
|
8153 }
|
|
|
8154
|
|
|
8155 /* Add the alternate score to su_sga. */
|
|
|
8156 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8157 {
|
|
|
8158 stp = &SUG(su->su_sga, i);
|
|
|
8159 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
8160 if (stp->st_score == SCORE_MAXMAX)
|
|
|
8161 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
8162 else
|
|
|
8163 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
8164 stp->st_salscore = TRUE;
|
|
|
8165 }
|
|
|
8166
|
|
|
8167 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
8168 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8169 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
8170
|
|
|
8171 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
8172 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
8173 return;
|
|
|
8174
|
|
|
8175 stp = &SUG(ga, 0);
|
|
|
8176 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
8177 {
|
|
|
8178 /* round 1: get a suggestion from su_ga
|
|
|
8179 * round 2: get a suggestion from su_sga */
|
|
|
8180 for (round = 1; round <= 2; ++round)
|
|
|
8181 {
|
|
|
8182 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
8183 if (i < gap->ga_len)
|
|
|
8184 {
|
|
|
8185 /* Don't add a word if it's already there. */
|
|
|
8186 p = SUG(*gap, i).st_word;
|
|
|
8187 for (j = 0; j < ga.ga_len; ++j)
|
|
|
8188 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
8189 break;
|
|
|
8190 if (j == ga.ga_len)
|
|
|
8191 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
8192 else
|
|
|
8193 vim_free(p);
|
|
|
8194 }
|
|
|
8195 }
|
|
|
8196 }
|
|
|
8197
|
|
|
8198 ga_clear(&su->su_ga);
|
|
|
8199 ga_clear(&su->su_sga);
|
|
|
8200
|
|
|
8201 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
8202 if (ga.ga_len > su->su_maxcount)
|
|
|
8203 {
|
|
|
8204 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
8205 vim_free(stp[i].st_word);
|
|
|
8206 ga.ga_len = su->su_maxcount;
|
|
|
8207 }
|
|
|
8208
|
|
|
8209 su->su_ga = ga;
|
|
|
8210 }
|
|
|
8211
|
|
|
8212 /*
|
|
351
|
8213 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
8214 * badword.
|
|
|
8215 */
|
|
|
8216 static int
|
|
|
8217 stp_sal_score(stp, su, slang, badsound)
|
|
|
8218 suggest_T *stp;
|
|
|
8219 suginfo_T *su;
|
|
|
8220 slang_T *slang;
|
|
|
8221 char_u *badsound; /* sound-folded badword */
|
|
|
8222 {
|
|
|
8223 char_u *p;
|
|
|
8224 char_u badsound2[MAXWLEN];
|
|
|
8225 char_u fword[MAXWLEN];
|
|
|
8226 char_u goodsound[MAXWLEN];
|
|
|
8227
|
|
|
8228 if (stp->st_orglen <= su->su_badlen)
|
|
|
8229 p = badsound;
|
|
|
8230 else
|
|
|
8231 {
|
|
|
8232 /* soundfold the bad word with more characters following */
|
|
|
8233 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
8234
|
|
|
8235 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
8236 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
8237 * removing the space. Don't do it when the good word also contains a
|
|
|
8238 * space. */
|
|
|
8239 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
8240 && *skiptowhite(stp->st_word) == NUL)
|
|
|
8241 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
8242 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
8243
|
|
375
|
8244 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
8245 p = badsound2;
|
|
|
8246 }
|
|
|
8247
|
|
375
|
8248 /* Sound-fold the word and compute the score for the difference. */
|
|
|
8249 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
8250
|
|
|
8251 return soundalike_score(goodsound, p);
|
|
|
8252 }
|
|
|
8253
|
|
|
8254 /*
|
|
323
|
8255 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
8256 */
|
|
|
8257 static void
|
|
346
|
8258 suggest_try_soundalike(su)
|
|
323
|
8259 suginfo_T *su;
|
|
|
8260 {
|
|
|
8261 char_u salword[MAXWLEN];
|
|
|
8262 char_u tword[MAXWLEN];
|
|
|
8263 char_u tsalword[MAXWLEN];
|
|
324
|
8264 idx_T arridx[MAXWLEN];
|
|
323
|
8265 int curi[MAXWLEN];
|
|
|
8266 langp_T *lp;
|
|
|
8267 char_u *byts;
|
|
324
|
8268 idx_T *idxs;
|
|
323
|
8269 int depth;
|
|
|
8270 int c;
|
|
324
|
8271 idx_T n;
|
|
323
|
8272 int round;
|
|
|
8273 int flags;
|
|
344
|
8274 int sound_score;
|
|
|
8275
|
|
|
8276 /* Do this for all languages that support sound folding. */
|
|
323
|
8277 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8278 lp->lp_slang != NULL; ++lp)
|
|
|
8279 {
|
|
|
8280 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8281 {
|
|
|
8282 /* soundfold the bad word */
|
|
375
|
8283 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
8284
|
|
|
8285 /*
|
|
|
8286 * Go through the whole tree, soundfold each word and compare.
|
|
|
8287 * round 1: use the case-folded tree.
|
|
|
8288 * round 2: use the keep-case tree.
|
|
|
8289 */
|
|
|
8290 for (round = 1; round <= 2; ++round)
|
|
|
8291 {
|
|
|
8292 if (round == 1)
|
|
|
8293 {
|
|
|
8294 byts = lp->lp_slang->sl_fbyts;
|
|
|
8295 idxs = lp->lp_slang->sl_fidxs;
|
|
|
8296 }
|
|
|
8297 else
|
|
|
8298 {
|
|
|
8299 byts = lp->lp_slang->sl_kbyts;
|
|
|
8300 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
8301 if (byts == NULL) /* no keep-case words */
|
|
|
8302 continue;
|
|
323
|
8303 }
|
|
|
8304
|
|
|
8305 depth = 0;
|
|
|
8306 arridx[0] = 0;
|
|
|
8307 curi[0] = 1;
|
|
|
8308 while (depth >= 0 && !got_int)
|
|
|
8309 {
|
|
|
8310 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
8311 {
|
|
323
|
8312 /* Done all bytes at this node, go up one level. */
|
|
|
8313 --depth;
|
|
351
|
8314 line_breakcheck();
|
|
|
8315 }
|
|
323
|
8316 else
|
|
|
8317 {
|
|
|
8318 /* Do one more byte at this node. */
|
|
|
8319 n = arridx[depth] + curi[depth];
|
|
|
8320 ++curi[depth];
|
|
|
8321 c = byts[n];
|
|
|
8322 if (c == 0)
|
|
|
8323 {
|
|
|
8324 /* End of word, deal with the word. */
|
|
324
|
8325 flags = (int)idxs[n];
|
|
323
|
8326 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
8327 {
|
|
|
8328 tword[depth] = NUL;
|
|
375
|
8329 /* Sound-fold. Only in keep-case tree need to
|
|
|
8330 * case-fold the word. */
|
|
|
8331 spell_soundfold(lp->lp_slang, tword,
|
|
|
8332 round == 1, tsalword);
|
|
323
|
8333
|
|
344
|
8334 /* Compute the edit distance between the
|
|
|
8335 * sound-a-like words. */
|
|
|
8336 sound_score = soundalike_score(salword,
|
|
|
8337 tsalword);
|
|
324
|
8338 if (sound_score < SCORE_MAXMAX)
|
|
323
|
8339 {
|
|
324
|
8340 char_u cword[MAXWLEN];
|
|
|
8341 char_u *p;
|
|
344
|
8342 int score;
|
|
324
|
8343
|
|
388
|
8344 flags |= su->su_badflags;
|
|
351
|
8345 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
8346 {
|
|
324
|
8347 /* Need to fix case according to
|
|
|
8348 * "flags". */
|
|
323
|
8349 make_case_word(tword, cword, flags);
|
|
324
|
8350 p = cword;
|
|
323
|
8351 }
|
|
|
8352 else
|
|
324
|
8353 p = tword;
|
|
|
8354
|
|
344
|
8355 if (sps_flags & SPS_DOUBLE)
|
|
|
8356 add_suggestion(su, &su->su_sga, p,
|
|
346
|
8357 su->su_badlen,
|
|
351
|
8358 sound_score, 0, FALSE);
|
|
344
|
8359 else
|
|
|
8360 {
|
|
|
8361 /* Compute the score. */
|
|
|
8362 score = spell_edit_score(
|
|
|
8363 su->su_badword, p);
|
|
|
8364 if (sps_flags & SPS_BEST)
|
|
|
8365 /* give a bonus for the good word
|
|
|
8366 * sounding the same as the bad
|
|
|
8367 * word */
|
|
|
8368 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8369 su->su_badlen,
|
|
344
|
8370 RESCORE(score, sound_score),
|
|
351
|
8371 sound_score, TRUE);
|
|
344
|
8372 else
|
|
|
8373 add_suggestion(su, &su->su_ga, p,
|
|
346
|
8374 su->su_badlen,
|
|
351
|
8375 score + sound_score, 0, FALSE);
|
|
344
|
8376 }
|
|
323
|
8377 }
|
|
|
8378 }
|
|
|
8379
|
|
|
8380 /* Skip over other NUL bytes. */
|
|
|
8381 while (byts[n + 1] == 0)
|
|
|
8382 {
|
|
|
8383 ++n;
|
|
|
8384 ++curi[depth];
|
|
|
8385 }
|
|
|
8386 }
|
|
|
8387 else
|
|
|
8388 {
|
|
|
8389 /* Normal char, go one level deeper. */
|
|
|
8390 tword[depth++] = c;
|
|
|
8391 arridx[depth] = idxs[n];
|
|
|
8392 curi[depth] = 1;
|
|
|
8393 }
|
|
|
8394 }
|
|
|
8395 }
|
|
|
8396 }
|
|
|
8397 }
|
|
|
8398 }
|
|
|
8399 }
|
|
|
8400
|
|
|
8401 /*
|
|
324
|
8402 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
8403 */
|
|
|
8404 static void
|
|
|
8405 make_case_word(fword, cword, flags)
|
|
|
8406 char_u *fword;
|
|
|
8407 char_u *cword;
|
|
|
8408 int flags;
|
|
|
8409 {
|
|
|
8410 if (flags & WF_ALLCAP)
|
|
|
8411 /* Make it all upper-case */
|
|
|
8412 allcap_copy(fword, cword);
|
|
|
8413 else if (flags & WF_ONECAP)
|
|
|
8414 /* Make the first letter upper-case */
|
|
324
|
8415 onecap_copy(fword, cword, TRUE);
|
|
323
|
8416 else
|
|
|
8417 /* Use goodword as-is. */
|
|
|
8418 STRCPY(cword, fword);
|
|
|
8419 }
|
|
|
8420
|
|
330
|
8421 /*
|
|
|
8422 * Use map string "map" for languages "lp".
|
|
|
8423 */
|
|
|
8424 static void
|
|
|
8425 set_map_str(lp, map)
|
|
|
8426 slang_T *lp;
|
|
|
8427 char_u *map;
|
|
|
8428 {
|
|
|
8429 char_u *p;
|
|
|
8430 int headc = 0;
|
|
|
8431 int c;
|
|
|
8432 int i;
|
|
|
8433
|
|
|
8434 if (*map == NUL)
|
|
|
8435 {
|
|
|
8436 lp->sl_has_map = FALSE;
|
|
|
8437 return;
|
|
|
8438 }
|
|
|
8439 lp->sl_has_map = TRUE;
|
|
|
8440
|
|
|
8441 /* Init the array and hash table empty. */
|
|
|
8442 for (i = 0; i < 256; ++i)
|
|
|
8443 lp->sl_map_array[i] = 0;
|
|
|
8444 #ifdef FEAT_MBYTE
|
|
|
8445 hash_init(&lp->sl_map_hash);
|
|
|
8446 #endif
|
|
|
8447
|
|
|
8448 /*
|
|
|
8449 * The similar characters are stored separated with slashes:
|
|
|
8450 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
8451 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
8452 */
|
|
|
8453 for (p = map; *p != NUL; )
|
|
|
8454 {
|
|
|
8455 #ifdef FEAT_MBYTE
|
|
|
8456 c = mb_ptr2char_adv(&p);
|
|
|
8457 #else
|
|
|
8458 c = *p++;
|
|
|
8459 #endif
|
|
|
8460 if (c == '/')
|
|
|
8461 headc = 0;
|
|
|
8462 else
|
|
|
8463 {
|
|
|
8464 if (headc == 0)
|
|
|
8465 headc = c;
|
|
|
8466
|
|
|
8467 #ifdef FEAT_MBYTE
|
|
|
8468 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
8469 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
8470 * a NUL. */
|
|
|
8471 if (c >= 256)
|
|
|
8472 {
|
|
|
8473 int cl = mb_char2len(c);
|
|
|
8474 int headcl = mb_char2len(headc);
|
|
|
8475 char_u *b;
|
|
|
8476 hash_T hash;
|
|
|
8477 hashitem_T *hi;
|
|
|
8478
|
|
|
8479 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
8480 if (b == NULL)
|
|
|
8481 return;
|
|
|
8482 mb_char2bytes(c, b);
|
|
|
8483 b[cl] = NUL;
|
|
|
8484 mb_char2bytes(headc, b + cl + 1);
|
|
|
8485 b[cl + 1 + headcl] = NUL;
|
|
|
8486 hash = hash_hash(b);
|
|
|
8487 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
8488 if (HASHITEM_EMPTY(hi))
|
|
|
8489 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
8490 else
|
|
|
8491 {
|
|
|
8492 /* This should have been checked when generating the .spl
|
|
|
8493 * file. */
|
|
|
8494 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
8495 vim_free(b);
|
|
|
8496 }
|
|
|
8497 }
|
|
|
8498 else
|
|
|
8499 #endif
|
|
|
8500 lp->sl_map_array[c] = headc;
|
|
|
8501 }
|
|
|
8502 }
|
|
|
8503 }
|
|
|
8504
|
|
323
|
8505 /*
|
|
|
8506 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
8507 * lines in the .aff file.
|
|
|
8508 */
|
|
|
8509 static int
|
|
|
8510 similar_chars(slang, c1, c2)
|
|
|
8511 slang_T *slang;
|
|
|
8512 int c1;
|
|
|
8513 int c2;
|
|
|
8514 {
|
|
330
|
8515 int m1, m2;
|
|
|
8516 #ifdef FEAT_MBYTE
|
|
|
8517 char_u buf[MB_MAXBYTES];
|
|
|
8518 hashitem_T *hi;
|
|
|
8519
|
|
|
8520 if (c1 >= 256)
|
|
|
8521 {
|
|
|
8522 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
8523 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8524 if (HASHITEM_EMPTY(hi))
|
|
|
8525 m1 = 0;
|
|
|
8526 else
|
|
|
8527 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8528 }
|
|
|
8529 else
|
|
|
8530 #endif
|
|
|
8531 m1 = slang->sl_map_array[c1];
|
|
|
8532 if (m1 == 0)
|
|
323
|
8533 return FALSE;
|
|
330
|
8534
|
|
|
8535
|
|
|
8536 #ifdef FEAT_MBYTE
|
|
|
8537 if (c2 >= 256)
|
|
|
8538 {
|
|
|
8539 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
8540 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
8541 if (HASHITEM_EMPTY(hi))
|
|
|
8542 m2 = 0;
|
|
|
8543 else
|
|
|
8544 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
8545 }
|
|
|
8546 else
|
|
|
8547 #endif
|
|
|
8548 m2 = slang->sl_map_array[c2];
|
|
|
8549
|
|
|
8550 return m1 == m2;
|
|
323
|
8551 }
|
|
|
8552
|
|
|
8553 /*
|
|
|
8554 * Add a suggestion to the list of suggestions.
|
|
|
8555 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
8556 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
8557 * with spell_edit_score().
|
|
|
8558 */
|
|
|
8559 static void
|
|
351
|
8560 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
8561 suginfo_T *su;
|
|
344
|
8562 garray_T *gap;
|
|
323
|
8563 char_u *goodword;
|
|
346
|
8564 int badlen; /* length of bad word used */
|
|
324
|
8565 int score;
|
|
351
|
8566 int altscore;
|
|
344
|
8567 int had_bonus; /* value for st_had_bonus */
|
|
323
|
8568 {
|
|
|
8569 suggest_T *stp;
|
|
|
8570 int i;
|
|
346
|
8571 char_u *p = NULL;
|
|
|
8572 int c = 0;
|
|
323
|
8573
|
|
|
8574 /* Check that the word wasn't banned. */
|
|
|
8575 if (was_banned(su, goodword))
|
|
|
8576 return;
|
|
|
8577
|
|
346
|
8578 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
8579 * Remove the common part from "goodword". */
|
|
|
8580 i = badlen - su->su_badlen;
|
|
|
8581 if (i > 0)
|
|
|
8582 {
|
|
|
8583 /* This assumes there was no case folding or it didn't change the
|
|
|
8584 * length... */
|
|
|
8585 p = goodword + STRLEN(goodword) - i;
|
|
|
8586 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
8587 {
|
|
|
8588 badlen = su->su_badlen;
|
|
|
8589 c = *p;
|
|
|
8590 *p = NUL;
|
|
|
8591 }
|
|
|
8592 else
|
|
|
8593 p = NULL;
|
|
|
8594 }
|
|
372
|
8595 else if (i < 0)
|
|
|
8596 {
|
|
|
8597 /* When replacing part of the word check that we actually change
|
|
|
8598 * something. For "the the" a suggestion can be replacing the first
|
|
|
8599 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
8600 if (badlen == (int)STRLEN(goodword)
|
|
372
|
8601 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
8602 return;
|
|
|
8603 }
|
|
|
8604
|
|
346
|
8605
|
|
323
|
8606 if (score <= su->su_maxscore)
|
|
|
8607 {
|
|
366
|
8608 /* Check if the word is already there. Also check the length that is
|
|
|
8609 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
8610 * "thes" -> "these". */
|
|
344
|
8611 stp = &SUG(*gap, 0);
|
|
|
8612 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
8613 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
8614 && stp[i].st_orglen == badlen)
|
|
323
|
8615 {
|
|
|
8616 /* Found it. Remember the lowest score. */
|
|
|
8617 if (stp[i].st_score > score)
|
|
324
|
8618 {
|
|
323
|
8619 stp[i].st_score = score;
|
|
376
|
8620 stp[i].st_altscore = altscore;
|
|
324
|
8621 stp[i].st_had_bonus = had_bonus;
|
|
|
8622 }
|
|
323
|
8623 break;
|
|
|
8624 }
|
|
|
8625
|
|
344
|
8626 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
8627 {
|
|
|
8628 /* Add a suggestion. */
|
|
344
|
8629 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
8630 stp->st_word = vim_strsave(goodword);
|
|
|
8631 if (stp->st_word != NULL)
|
|
|
8632 {
|
|
|
8633 stp->st_score = score;
|
|
351
|
8634 stp->st_altscore = altscore;
|
|
324
|
8635 stp->st_had_bonus = had_bonus;
|
|
346
|
8636 stp->st_orglen = badlen;
|
|
344
|
8637 ++gap->ga_len;
|
|
323
|
8638
|
|
|
8639 /* If we have too many suggestions now, sort the list and keep
|
|
|
8640 * the best suggestions. */
|
|
344
|
8641 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
8642 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
8643 SUG_CLEAN_COUNT(su));
|
|
323
|
8644 }
|
|
|
8645 }
|
|
|
8646 }
|
|
346
|
8647
|
|
|
8648 if (p != NULL)
|
|
|
8649 *p = c; /* restore "goodword" */
|
|
323
|
8650 }
|
|
|
8651
|
|
|
8652 /*
|
|
|
8653 * Add a word to be banned.
|
|
|
8654 */
|
|
|
8655 static void
|
|
|
8656 add_banned(su, word)
|
|
|
8657 suginfo_T *su;
|
|
|
8658 char_u *word;
|
|
|
8659 {
|
|
|
8660 char_u *s = vim_strsave(word);
|
|
|
8661 hash_T hash;
|
|
|
8662 hashitem_T *hi;
|
|
|
8663
|
|
|
8664 if (s != NULL)
|
|
|
8665 {
|
|
|
8666 hash = hash_hash(s);
|
|
|
8667 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
8668 if (HASHITEM_EMPTY(hi))
|
|
|
8669 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
8670 else
|
|
|
8671 vim_free(s);
|
|
323
|
8672 }
|
|
|
8673 }
|
|
|
8674
|
|
|
8675 /*
|
|
|
8676 * Return TRUE if a word appears in the list of banned words.
|
|
|
8677 */
|
|
|
8678 static int
|
|
|
8679 was_banned(su, word)
|
|
|
8680 suginfo_T *su;
|
|
|
8681 char_u *word;
|
|
|
8682 {
|
|
324
|
8683 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
8684
|
|
|
8685 return !HASHITEM_EMPTY(hi);
|
|
323
|
8686 }
|
|
|
8687
|
|
|
8688 /*
|
|
|
8689 * Free the banned words in "su".
|
|
|
8690 */
|
|
|
8691 static void
|
|
|
8692 free_banned(su)
|
|
|
8693 suginfo_T *su;
|
|
|
8694 {
|
|
|
8695 int todo;
|
|
|
8696 hashitem_T *hi;
|
|
|
8697
|
|
|
8698 todo = su->su_banned.ht_used;
|
|
|
8699 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
8700 {
|
|
|
8701 if (!HASHITEM_EMPTY(hi))
|
|
|
8702 {
|
|
|
8703 vim_free(hi->hi_key);
|
|
|
8704 --todo;
|
|
|
8705 }
|
|
|
8706 }
|
|
|
8707 hash_clear(&su->su_banned);
|
|
|
8708 }
|
|
|
8709
|
|
324
|
8710 /*
|
|
|
8711 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
8712 * thus only done for the final results.
|
|
|
8713 */
|
|
|
8714 static void
|
|
|
8715 rescore_suggestions(su)
|
|
|
8716 suginfo_T *su;
|
|
|
8717 {
|
|
|
8718 langp_T *lp;
|
|
|
8719 suggest_T *stp;
|
|
|
8720 char_u sal_badword[MAXWLEN];
|
|
|
8721 int i;
|
|
|
8722
|
|
|
8723 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8724 lp->lp_slang != NULL; ++lp)
|
|
|
8725 {
|
|
|
8726 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8727 {
|
|
|
8728 /* soundfold the bad word */
|
|
375
|
8729 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
8730
|
|
|
8731 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8732 {
|
|
344
|
8733 stp = &SUG(su->su_ga, i);
|
|
324
|
8734 if (!stp->st_had_bonus)
|
|
|
8735 {
|
|
351
|
8736 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
8737 lp->lp_slang, sal_badword);
|
|
|
8738 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
8739 stp->st_altscore = SCORE_BIG;
|
|
|
8740 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
8741 }
|
|
|
8742 }
|
|
|
8743 break;
|
|
|
8744 }
|
|
|
8745 }
|
|
|
8746 }
|
|
|
8747
|
|
323
|
8748 static int
|
|
|
8749 #ifdef __BORLANDC__
|
|
|
8750 _RTLENTRYF
|
|
|
8751 #endif
|
|
|
8752 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
8753
|
|
|
8754 /*
|
|
|
8755 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
8756 */
|
|
|
8757 static int
|
|
|
8758 #ifdef __BORLANDC__
|
|
|
8759 _RTLENTRYF
|
|
|
8760 #endif
|
|
|
8761 sug_compare(s1, s2)
|
|
|
8762 const void *s1;
|
|
|
8763 const void *s2;
|
|
|
8764 {
|
|
|
8765 suggest_T *p1 = (suggest_T *)s1;
|
|
|
8766 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
8767 int n = p1->st_score - p2->st_score;
|
|
|
8768
|
|
|
8769 if (n == 0)
|
|
|
8770 return p1->st_altscore - p2->st_altscore;
|
|
|
8771 return n;
|
|
323
|
8772 }
|
|
|
8773
|
|
|
8774 /*
|
|
|
8775 * Cleanup the suggestions:
|
|
|
8776 * - Sort on score.
|
|
|
8777 * - Remove words that won't be displayed.
|
|
344
|
8778 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
8779 */
|
|
344
|
8780 static int
|
|
|
8781 cleanup_suggestions(gap, maxscore, keep)
|
|
|
8782 garray_T *gap;
|
|
|
8783 int maxscore;
|
|
324
|
8784 int keep; /* nr of suggestions to keep */
|
|
323
|
8785 {
|
|
344
|
8786 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
8787 int i;
|
|
|
8788
|
|
|
8789 /* Sort the list. */
|
|
344
|
8790 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
8791
|
|
|
8792 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
8793 if (gap->ga_len > keep)
|
|
323
|
8794 {
|
|
344
|
8795 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
8796 vim_free(stp[i].st_word);
|
|
344
|
8797 gap->ga_len = keep;
|
|
|
8798 return stp[keep - 1].st_score;
|
|
323
|
8799 }
|
|
344
|
8800 return maxscore;
|
|
323
|
8801 }
|
|
|
8802
|
|
372
|
8803 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
8804 /*
|
|
|
8805 * Soundfold a string, for soundfold().
|
|
|
8806 * Result is in allocated memory, NULL for an error.
|
|
|
8807 */
|
|
|
8808 char_u *
|
|
|
8809 eval_soundfold(word)
|
|
|
8810 char_u *word;
|
|
|
8811 {
|
|
|
8812 langp_T *lp;
|
|
|
8813 char_u sound[MAXWLEN];
|
|
|
8814
|
|
|
8815 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
8816 /* Use the sound-folding of the first language that supports it. */
|
|
|
8817 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8818 lp->lp_slang != NULL; ++lp)
|
|
|
8819 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
8820 {
|
|
|
8821 /* soundfold the word */
|
|
375
|
8822 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
8823 return vim_strsave(sound);
|
|
|
8824 }
|
|
|
8825
|
|
|
8826 /* No language with sound folding, return word as-is. */
|
|
|
8827 return vim_strsave(word);
|
|
|
8828 }
|
|
|
8829 #endif
|
|
|
8830
|
|
323
|
8831 /*
|
|
|
8832 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
8833 */
|
|
|
8834 static void
|
|
375
|
8835 spell_soundfold(slang, inword, folded, res)
|
|
|
8836 slang_T *slang;
|
|
|
8837 char_u *inword;
|
|
|
8838 int folded; /* "inword" is already case-folded */
|
|
|
8839 char_u *res;
|
|
|
8840 {
|
|
|
8841 char_u fword[MAXWLEN];
|
|
|
8842 char_u *word;
|
|
|
8843
|
|
|
8844 if (slang->sl_sofo)
|
|
|
8845 /* SOFOFROM and SOFOTO used */
|
|
|
8846 spell_soundfold_sofo(slang, inword, res);
|
|
|
8847 else
|
|
|
8848 {
|
|
|
8849 /* SAL items used. Requires the word to be case-folded. */
|
|
|
8850 if (folded)
|
|
|
8851 word = inword;
|
|
|
8852 else
|
|
|
8853 {
|
|
|
8854 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
8855 word = fword;
|
|
|
8856 }
|
|
|
8857
|
|
|
8858 #ifdef FEAT_MBYTE
|
|
|
8859 if (has_mbyte)
|
|
|
8860 spell_soundfold_wsal(slang, word, res);
|
|
|
8861 else
|
|
|
8862 #endif
|
|
|
8863 spell_soundfold_sal(slang, word, res);
|
|
|
8864 }
|
|
|
8865 }
|
|
|
8866
|
|
|
8867 /*
|
|
|
8868 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
8869 * SOFOTO lines.
|
|
|
8870 */
|
|
|
8871 static void
|
|
|
8872 spell_soundfold_sofo(slang, inword, res)
|
|
|
8873 slang_T *slang;
|
|
|
8874 char_u *inword;
|
|
|
8875 char_u *res;
|
|
|
8876 {
|
|
|
8877 char_u *s;
|
|
|
8878 int ri = 0;
|
|
|
8879 int c;
|
|
|
8880
|
|
|
8881 #ifdef FEAT_MBYTE
|
|
|
8882 if (has_mbyte)
|
|
|
8883 {
|
|
|
8884 int prevc = 0;
|
|
|
8885 int *ip;
|
|
|
8886
|
|
|
8887 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
8888 * 255, sl_sal the rest. */
|
|
|
8889 for (s = inword; *s != NUL; )
|
|
|
8890 {
|
|
|
8891 c = mb_ptr2char_adv(&s);
|
|
|
8892 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
8893 c = ' ';
|
|
|
8894 else if (c < 256)
|
|
|
8895 c = slang->sl_sal_first[c];
|
|
|
8896 else
|
|
|
8897 {
|
|
|
8898 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
8899 if (ip == NULL) /* empty list, can't match */
|
|
|
8900 c = NUL;
|
|
|
8901 else
|
|
|
8902 for (;;) /* find "c" in the list */
|
|
|
8903 {
|
|
|
8904 if (*ip == 0) /* not found */
|
|
|
8905 {
|
|
|
8906 c = NUL;
|
|
|
8907 break;
|
|
|
8908 }
|
|
|
8909 if (*ip == c) /* match! */
|
|
|
8910 {
|
|
|
8911 c = ip[1];
|
|
|
8912 break;
|
|
|
8913 }
|
|
|
8914 ip += 2;
|
|
|
8915 }
|
|
|
8916 }
|
|
|
8917
|
|
|
8918 if (c != NUL && c != prevc)
|
|
|
8919 {
|
|
|
8920 ri += mb_char2bytes(c, res + ri);
|
|
|
8921 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
8922 break;
|
|
|
8923 prevc = c;
|
|
|
8924 }
|
|
|
8925 }
|
|
|
8926 }
|
|
|
8927 else
|
|
|
8928 #endif
|
|
|
8929 {
|
|
|
8930 /* The sl_sal_first[] table contains the translation. */
|
|
|
8931 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
8932 {
|
|
|
8933 if (vim_iswhite(c))
|
|
|
8934 c = ' ';
|
|
|
8935 else
|
|
|
8936 c = slang->sl_sal_first[c];
|
|
|
8937 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
8938 res[ri++] = c;
|
|
|
8939 }
|
|
|
8940 }
|
|
|
8941
|
|
|
8942 res[ri] = NUL;
|
|
|
8943 }
|
|
|
8944
|
|
|
8945 static void
|
|
|
8946 spell_soundfold_sal(slang, inword, res)
|
|
323
|
8947 slang_T *slang;
|
|
|
8948 char_u *inword;
|
|
|
8949 char_u *res;
|
|
|
8950 {
|
|
344
|
8951 salitem_T *smp;
|
|
323
|
8952 char_u word[MAXWLEN];
|
|
375
|
8953 char_u *s = inword;
|
|
323
|
8954 char_u *t;
|
|
344
|
8955 char_u *pf;
|
|
323
|
8956 int i, j, z;
|
|
344
|
8957 int reslen;
|
|
323
|
8958 int n, k = 0;
|
|
|
8959 int z0;
|
|
|
8960 int k0;
|
|
|
8961 int n0;
|
|
|
8962 int c;
|
|
|
8963 int pri;
|
|
|
8964 int p0 = -333;
|
|
|
8965 int c0;
|
|
|
8966
|
|
324
|
8967 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
8968 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
8969 if (slang->sl_rem_accents)
|
|
|
8970 {
|
|
|
8971 t = word;
|
|
375
|
8972 while (*s != NUL)
|
|
323
|
8973 {
|
|
324
|
8974 if (vim_iswhite(*s))
|
|
344
|
8975 {
|
|
|
8976 *t++ = ' ';
|
|
|
8977 s = skipwhite(s);
|
|
|
8978 }
|
|
323
|
8979 else
|
|
|
8980 {
|
|
376
|
8981 if (spell_iswordp_nmw(s))
|
|
323
|
8982 *t++ = *s;
|
|
|
8983 ++s;
|
|
|
8984 }
|
|
|
8985 }
|
|
|
8986 *t = NUL;
|
|
|
8987 }
|
|
|
8988 else
|
|
375
|
8989 STRCPY(word, s);
|
|
323
|
8990
|
|
344
|
8991 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
8992
|
|
|
8993 /*
|
|
|
8994 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
8995 * Changed to keep spaces.
|
|
323
|
8996 */
|
|
344
|
8997 i = reslen = z = 0;
|
|
323
|
8998 while ((c = word[i]) != NUL)
|
|
|
8999 {
|
|
344
|
9000 /* Start with the first rule that has the character in the word. */
|
|
323
|
9001 n = slang->sl_sal_first[c];
|
|
|
9002 z0 = 0;
|
|
|
9003
|
|
|
9004 if (n >= 0)
|
|
|
9005 {
|
|
|
9006 /* check all rules for the same letter */
|
|
344
|
9007 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
9008 {
|
|
344
|
9009 /* Quickly skip entries that don't match the word. Most
|
|
|
9010 * entries are less then three chars, optimize for that. */
|
|
|
9011 k = smp[n].sm_leadlen;
|
|
|
9012 if (k > 1)
|
|
323
|
9013 {
|
|
344
|
9014 if (word[i + 1] != s[1])
|
|
|
9015 continue;
|
|
|
9016 if (k > 2)
|
|
|
9017 {
|
|
|
9018 for (j = 2; j < k; ++j)
|
|
|
9019 if (word[i + j] != s[j])
|
|
|
9020 break;
|
|
|
9021 if (j < k)
|
|
|
9022 continue;
|
|
|
9023 }
|
|
323
|
9024 }
|
|
|
9025
|
|
375
|
9026 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
9027 {
|
|
375
|
9028 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
9029 while (*pf != NUL && *pf != word[i + k])
|
|
|
9030 ++pf;
|
|
|
9031 if (*pf == NUL)
|
|
|
9032 continue;
|
|
|
9033 ++k;
|
|
323
|
9034 }
|
|
344
|
9035 s = smp[n].sm_rules;
|
|
|
9036 pri = 5; /* default priority */
|
|
323
|
9037
|
|
|
9038 p0 = *s;
|
|
|
9039 k0 = k;
|
|
|
9040 while (*s == '-' && k > 1)
|
|
|
9041 {
|
|
|
9042 k--;
|
|
|
9043 s++;
|
|
|
9044 }
|
|
|
9045 if (*s == '<')
|
|
|
9046 s++;
|
|
344
|
9047 if (VIM_ISDIGIT(*s))
|
|
323
|
9048 {
|
|
|
9049 /* determine priority */
|
|
|
9050 pri = *s - '0';
|
|
|
9051 s++;
|
|
|
9052 }
|
|
|
9053 if (*s == '^' && *(s + 1) == '^')
|
|
|
9054 s++;
|
|
|
9055
|
|
|
9056 if (*s == NUL
|
|
|
9057 || (*s == '^'
|
|
324
|
9058 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9059 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
9060 && (*(s + 1) != '$'
|
|
376
|
9061 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9062 || (*s == '$' && i > 0
|
|
376
|
9063 && spell_iswordp(word + i - 1, curbuf)
|
|
|
9064 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
9065 {
|
|
|
9066 /* search for followup rules, if: */
|
|
|
9067 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9068 c0 = word[i + k - 1];
|
|
|
9069 n0 = slang->sl_sal_first[c0];
|
|
|
9070
|
|
|
9071 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
9072 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
9073 {
|
|
|
9074 /* test follow-up rule for "word[i + k]" */
|
|
344
|
9075 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
9076 {
|
|
344
|
9077 /* Quickly skip entries that don't match the word.
|
|
|
9078 * */
|
|
|
9079 k0 = smp[n0].sm_leadlen;
|
|
|
9080 if (k0 > 1)
|
|
323
|
9081 {
|
|
344
|
9082 if (word[i + k] != s[1])
|
|
|
9083 continue;
|
|
|
9084 if (k0 > 2)
|
|
|
9085 {
|
|
|
9086 pf = word + i + k + 1;
|
|
|
9087 for (j = 2; j < k0; ++j)
|
|
|
9088 if (*pf++ != s[j])
|
|
|
9089 break;
|
|
|
9090 if (j < k0)
|
|
|
9091 continue;
|
|
|
9092 }
|
|
323
|
9093 }
|
|
344
|
9094 k0 += k - 1;
|
|
|
9095
|
|
375
|
9096 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
9097 {
|
|
344
|
9098 /* Check for match with one of the chars in
|
|
375
|
9099 * "sm_oneof". */
|
|
344
|
9100 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9101 ++pf;
|
|
|
9102 if (*pf == NUL)
|
|
|
9103 continue;
|
|
|
9104 ++k0;
|
|
323
|
9105 }
|
|
344
|
9106
|
|
|
9107 p0 = 5;
|
|
|
9108 s = smp[n0].sm_rules;
|
|
323
|
9109 while (*s == '-')
|
|
|
9110 {
|
|
344
|
9111 /* "k0" gets NOT reduced because
|
|
|
9112 * "if (k0 == k)" */
|
|
323
|
9113 s++;
|
|
|
9114 }
|
|
|
9115 if (*s == '<')
|
|
|
9116 s++;
|
|
344
|
9117 if (VIM_ISDIGIT(*s))
|
|
323
|
9118 {
|
|
|
9119 p0 = *s - '0';
|
|
|
9120 s++;
|
|
|
9121 }
|
|
|
9122
|
|
|
9123 if (*s == NUL
|
|
|
9124 /* *s == '^' cuts */
|
|
|
9125 || (*s == '$'
|
|
376
|
9126 && !spell_iswordp(word + i + k0,
|
|
|
9127 curbuf)))
|
|
323
|
9128 {
|
|
|
9129 if (k0 == k)
|
|
|
9130 /* this is just a piece of the string */
|
|
|
9131 continue;
|
|
|
9132
|
|
|
9133 if (p0 < pri)
|
|
|
9134 /* priority too low */
|
|
|
9135 continue;
|
|
|
9136 /* rule fits; stop search */
|
|
|
9137 break;
|
|
|
9138 }
|
|
|
9139 }
|
|
|
9140
|
|
344
|
9141 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
9142 continue;
|
|
|
9143 }
|
|
|
9144
|
|
|
9145 /* replace string */
|
|
344
|
9146 s = smp[n].sm_to;
|
|
389
|
9147 if (s == NULL)
|
|
|
9148 s = (char_u *)"";
|
|
344
|
9149 pf = smp[n].sm_rules;
|
|
|
9150 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
9151 if (p0 == 1 && z == 0)
|
|
|
9152 {
|
|
|
9153 /* rule with '<' is used */
|
|
344
|
9154 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
9155 || res[reslen - 1] == *s))
|
|
|
9156 reslen--;
|
|
323
|
9157 z0 = 1;
|
|
|
9158 z = 1;
|
|
|
9159 k0 = 0;
|
|
372
|
9160 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
9161 {
|
|
|
9162 word[i + k0] = *s;
|
|
|
9163 k0++;
|
|
|
9164 s++;
|
|
|
9165 }
|
|
|
9166 if (k > k0)
|
|
|
9167 mch_memmove(word + i + k0, word + i + k,
|
|
|
9168 STRLEN(word + i + k) + 1);
|
|
|
9169
|
|
|
9170 /* new "actual letter" */
|
|
|
9171 c = word[i];
|
|
|
9172 }
|
|
|
9173 else
|
|
|
9174 {
|
|
|
9175 /* no '<' rule used */
|
|
|
9176 i += k - 1;
|
|
|
9177 z = 0;
|
|
344
|
9178 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
9179 {
|
|
344
|
9180 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
9181 res[reslen++] = *s;
|
|
323
|
9182 s++;
|
|
|
9183 }
|
|
|
9184 /* new "actual letter" */
|
|
|
9185 c = *s;
|
|
344
|
9186 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
9187 {
|
|
|
9188 if (c != NUL)
|
|
372
|
9189 res[reslen++] = c;
|
|
323
|
9190 mch_memmove(word, word + i + 1,
|
|
|
9191 STRLEN(word + i + 1) + 1);
|
|
|
9192 i = 0;
|
|
|
9193 z0 = 1;
|
|
|
9194 }
|
|
|
9195 }
|
|
|
9196 break;
|
|
|
9197 }
|
|
|
9198 }
|
|
|
9199 }
|
|
324
|
9200 else if (vim_iswhite(c))
|
|
|
9201 {
|
|
|
9202 c = ' ';
|
|
|
9203 k = 1;
|
|
|
9204 }
|
|
323
|
9205
|
|
|
9206 if (z0 == 0)
|
|
|
9207 {
|
|
344
|
9208 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9209 && (!slang->sl_collapse || reslen == 0
|
|
|
9210 || res[reslen - 1] != c))
|
|
323
|
9211 /* condense only double letters */
|
|
372
|
9212 res[reslen++] = c;
|
|
323
|
9213
|
|
|
9214 i++;
|
|
|
9215 z = 0;
|
|
|
9216 k = 0;
|
|
|
9217 }
|
|
|
9218 }
|
|
|
9219
|
|
344
|
9220 res[reslen] = NUL;
|
|
323
|
9221 }
|
|
|
9222
|
|
372
|
9223 #ifdef FEAT_MBYTE
|
|
|
9224 /*
|
|
|
9225 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
9226 * Multi-byte version of spell_soundfold().
|
|
|
9227 */
|
|
|
9228 static void
|
|
375
|
9229 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
9230 slang_T *slang;
|
|
|
9231 char_u *inword;
|
|
|
9232 char_u *res;
|
|
|
9233 {
|
|
375
|
9234 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
9235 int word[MAXWLEN];
|
|
|
9236 int wres[MAXWLEN];
|
|
|
9237 int l;
|
|
|
9238 char_u *s;
|
|
|
9239 int *ws;
|
|
|
9240 char_u *t;
|
|
|
9241 int *pf;
|
|
|
9242 int i, j, z;
|
|
|
9243 int reslen;
|
|
|
9244 int n, k = 0;
|
|
|
9245 int z0;
|
|
|
9246 int k0;
|
|
|
9247 int n0;
|
|
|
9248 int c;
|
|
|
9249 int pri;
|
|
|
9250 int p0 = -333;
|
|
|
9251 int c0;
|
|
|
9252 int did_white = FALSE;
|
|
|
9253
|
|
|
9254 /*
|
|
|
9255 * Convert the multi-byte string to a wide-character string.
|
|
|
9256 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
9257 * But keep white space.
|
|
|
9258 */
|
|
|
9259 n = 0;
|
|
|
9260 for (s = inword; *s != NUL; )
|
|
|
9261 {
|
|
|
9262 t = s;
|
|
|
9263 c = mb_ptr2char_adv(&s);
|
|
|
9264 if (slang->sl_rem_accents)
|
|
|
9265 {
|
|
|
9266 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
9267 {
|
|
|
9268 if (did_white)
|
|
|
9269 continue;
|
|
|
9270 c = ' ';
|
|
|
9271 did_white = TRUE;
|
|
|
9272 }
|
|
|
9273 else
|
|
|
9274 {
|
|
|
9275 did_white = FALSE;
|
|
376
|
9276 if (!spell_iswordp_nmw(t))
|
|
372
|
9277 continue;
|
|
|
9278 }
|
|
|
9279 }
|
|
|
9280 word[n++] = c;
|
|
|
9281 }
|
|
|
9282 word[n] = NUL;
|
|
|
9283
|
|
|
9284 /*
|
|
|
9285 * This comes from Aspell phonet.cpp.
|
|
|
9286 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
9287 * Changed to keep spaces.
|
|
|
9288 */
|
|
|
9289 i = reslen = z = 0;
|
|
|
9290 while ((c = word[i]) != NUL)
|
|
|
9291 {
|
|
|
9292 /* Start with the first rule that has the character in the word. */
|
|
|
9293 n = slang->sl_sal_first[c & 0xff];
|
|
|
9294 z0 = 0;
|
|
|
9295
|
|
|
9296 if (n >= 0)
|
|
|
9297 {
|
|
375
|
9298 /* check all rules for the same index byte */
|
|
372
|
9299 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
9300 {
|
|
|
9301 /* Quickly skip entries that don't match the word. Most
|
|
|
9302 * entries are less then three chars, optimize for that. */
|
|
375
|
9303 if (c != ws[0])
|
|
|
9304 continue;
|
|
372
|
9305 k = smp[n].sm_leadlen;
|
|
|
9306 if (k > 1)
|
|
|
9307 {
|
|
|
9308 if (word[i + 1] != ws[1])
|
|
|
9309 continue;
|
|
|
9310 if (k > 2)
|
|
|
9311 {
|
|
|
9312 for (j = 2; j < k; ++j)
|
|
|
9313 if (word[i + j] != ws[j])
|
|
|
9314 break;
|
|
|
9315 if (j < k)
|
|
|
9316 continue;
|
|
|
9317 }
|
|
|
9318 }
|
|
|
9319
|
|
375
|
9320 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
9321 {
|
|
375
|
9322 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
9323 while (*pf != NUL && *pf != word[i + k])
|
|
|
9324 ++pf;
|
|
|
9325 if (*pf == NUL)
|
|
|
9326 continue;
|
|
|
9327 ++k;
|
|
|
9328 }
|
|
|
9329 s = smp[n].sm_rules;
|
|
|
9330 pri = 5; /* default priority */
|
|
|
9331
|
|
|
9332 p0 = *s;
|
|
|
9333 k0 = k;
|
|
|
9334 while (*s == '-' && k > 1)
|
|
|
9335 {
|
|
|
9336 k--;
|
|
|
9337 s++;
|
|
|
9338 }
|
|
|
9339 if (*s == '<')
|
|
|
9340 s++;
|
|
|
9341 if (VIM_ISDIGIT(*s))
|
|
|
9342 {
|
|
|
9343 /* determine priority */
|
|
|
9344 pri = *s - '0';
|
|
|
9345 s++;
|
|
|
9346 }
|
|
|
9347 if (*s == '^' && *(s + 1) == '^')
|
|
|
9348 s++;
|
|
|
9349
|
|
|
9350 if (*s == NUL
|
|
|
9351 || (*s == '^'
|
|
|
9352 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
9353 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
9354 && (*(s + 1) != '$'
|
|
376
|
9355 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9356 || (*s == '$' && i > 0
|
|
376
|
9357 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
9358 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
9359 {
|
|
|
9360 /* search for followup rules, if: */
|
|
|
9361 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
9362 c0 = word[i + k - 1];
|
|
|
9363 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
9364
|
|
|
9365 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
9366 && p0 != '-' && word[i + k] != NUL)
|
|
|
9367 {
|
|
375
|
9368 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
9369 * all entries with the same index byte. */
|
|
372
|
9370 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
9371 == (c0 & 0xff); ++n0)
|
|
|
9372 {
|
|
|
9373 /* Quickly skip entries that don't match the word.
|
|
375
|
9374 */
|
|
|
9375 if (c0 != ws[0])
|
|
|
9376 continue;
|
|
372
|
9377 k0 = smp[n0].sm_leadlen;
|
|
|
9378 if (k0 > 1)
|
|
|
9379 {
|
|
|
9380 if (word[i + k] != ws[1])
|
|
|
9381 continue;
|
|
|
9382 if (k0 > 2)
|
|
|
9383 {
|
|
|
9384 pf = word + i + k + 1;
|
|
|
9385 for (j = 2; j < k0; ++j)
|
|
|
9386 if (*pf++ != ws[j])
|
|
|
9387 break;
|
|
|
9388 if (j < k0)
|
|
|
9389 continue;
|
|
|
9390 }
|
|
|
9391 }
|
|
|
9392 k0 += k - 1;
|
|
|
9393
|
|
375
|
9394 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
9395 {
|
|
|
9396 /* Check for match with one of the chars in
|
|
375
|
9397 * "sm_oneof". */
|
|
372
|
9398 while (*pf != NUL && *pf != word[i + k0])
|
|
|
9399 ++pf;
|
|
|
9400 if (*pf == NUL)
|
|
|
9401 continue;
|
|
|
9402 ++k0;
|
|
|
9403 }
|
|
|
9404
|
|
|
9405 p0 = 5;
|
|
|
9406 s = smp[n0].sm_rules;
|
|
|
9407 while (*s == '-')
|
|
|
9408 {
|
|
|
9409 /* "k0" gets NOT reduced because
|
|
|
9410 * "if (k0 == k)" */
|
|
|
9411 s++;
|
|
|
9412 }
|
|
|
9413 if (*s == '<')
|
|
|
9414 s++;
|
|
|
9415 if (VIM_ISDIGIT(*s))
|
|
|
9416 {
|
|
|
9417 p0 = *s - '0';
|
|
|
9418 s++;
|
|
|
9419 }
|
|
|
9420
|
|
|
9421 if (*s == NUL
|
|
|
9422 /* *s == '^' cuts */
|
|
|
9423 || (*s == '$'
|
|
376
|
9424 && !spell_iswordp_w(word + i + k0,
|
|
|
9425 curbuf)))
|
|
372
|
9426 {
|
|
|
9427 if (k0 == k)
|
|
|
9428 /* this is just a piece of the string */
|
|
|
9429 continue;
|
|
|
9430
|
|
|
9431 if (p0 < pri)
|
|
|
9432 /* priority too low */
|
|
|
9433 continue;
|
|
|
9434 /* rule fits; stop search */
|
|
|
9435 break;
|
|
|
9436 }
|
|
|
9437 }
|
|
|
9438
|
|
|
9439 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
9440 == (c0 & 0xff))
|
|
|
9441 continue;
|
|
|
9442 }
|
|
|
9443
|
|
|
9444 /* replace string */
|
|
|
9445 ws = smp[n].sm_to_w;
|
|
|
9446 s = smp[n].sm_rules;
|
|
|
9447 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
9448 if (p0 == 1 && z == 0)
|
|
|
9449 {
|
|
|
9450 /* rule with '<' is used */
|
|
389
|
9451 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
9452 && (wres[reslen - 1] == c
|
|
372
|
9453 || wres[reslen - 1] == *ws))
|
|
|
9454 reslen--;
|
|
|
9455 z0 = 1;
|
|
|
9456 z = 1;
|
|
|
9457 k0 = 0;
|
|
389
|
9458 if (ws != NULL)
|
|
|
9459 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
9460 {
|
|
|
9461 word[i + k0] = *ws;
|
|
|
9462 k0++;
|
|
|
9463 ws++;
|
|
|
9464 }
|
|
372
|
9465 if (k > k0)
|
|
|
9466 mch_memmove(word + i + k0, word + i + k,
|
|
|
9467 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
9468
|
|
|
9469 /* new "actual letter" */
|
|
|
9470 c = word[i];
|
|
|
9471 }
|
|
|
9472 else
|
|
|
9473 {
|
|
|
9474 /* no '<' rule used */
|
|
|
9475 i += k - 1;
|
|
|
9476 z = 0;
|
|
389
|
9477 if (ws != NULL)
|
|
|
9478 while (*ws != NUL && ws[1] != NUL
|
|
|
9479 && reslen < MAXWLEN)
|
|
|
9480 {
|
|
|
9481 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
9482 wres[reslen++] = *ws;
|
|
|
9483 ws++;
|
|
|
9484 }
|
|
372
|
9485 /* new "actual letter" */
|
|
389
|
9486 if (ws == NULL)
|
|
|
9487 c = NUL;
|
|
|
9488 else
|
|
|
9489 c = *ws;
|
|
372
|
9490 if (strstr((char *)s, "^^") != NULL)
|
|
|
9491 {
|
|
|
9492 if (c != NUL)
|
|
|
9493 wres[reslen++] = c;
|
|
|
9494 mch_memmove(word, word + i + 1,
|
|
|
9495 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
9496 i = 0;
|
|
|
9497 z0 = 1;
|
|
|
9498 }
|
|
|
9499 }
|
|
|
9500 break;
|
|
|
9501 }
|
|
|
9502 }
|
|
|
9503 }
|
|
|
9504 else if (vim_iswhite(c))
|
|
|
9505 {
|
|
|
9506 c = ' ';
|
|
|
9507 k = 1;
|
|
|
9508 }
|
|
|
9509
|
|
|
9510 if (z0 == 0)
|
|
|
9511 {
|
|
|
9512 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
9513 && (!slang->sl_collapse || reslen == 0
|
|
|
9514 || wres[reslen - 1] != c))
|
|
|
9515 /* condense only double letters */
|
|
|
9516 wres[reslen++] = c;
|
|
|
9517
|
|
|
9518 i++;
|
|
|
9519 z = 0;
|
|
|
9520 k = 0;
|
|
|
9521 }
|
|
|
9522 }
|
|
|
9523
|
|
|
9524 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
9525 l = 0;
|
|
|
9526 for (n = 0; n < reslen; ++n)
|
|
|
9527 {
|
|
|
9528 l += mb_char2bytes(wres[n], res + l);
|
|
|
9529 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
9530 break;
|
|
|
9531 }
|
|
|
9532 res[l] = NUL;
|
|
|
9533 }
|
|
|
9534 #endif
|
|
|
9535
|
|
324
|
9536 /*
|
|
344
|
9537 * Compute a score for two sound-a-like words.
|
|
|
9538 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
9539 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
9540 * avoiding checks that will not be possible.
|
|
|
9541 */
|
|
|
9542 static int
|
|
351
|
9543 soundalike_score(goodstart, badstart)
|
|
|
9544 char_u *goodstart; /* sound-folded good word */
|
|
|
9545 char_u *badstart; /* sound-folded bad word */
|
|
344
|
9546 {
|
|
351
|
9547 char_u *goodsound = goodstart;
|
|
|
9548 char_u *badsound = badstart;
|
|
|
9549 int goodlen;
|
|
|
9550 int badlen;
|
|
344
|
9551 int n;
|
|
|
9552 char_u *pl, *ps;
|
|
|
9553 char_u *pl2, *ps2;
|
|
351
|
9554 int score = 0;
|
|
|
9555
|
|
|
9556 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
9557 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
9558 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
9559 {
|
|
|
9560 score = SCORE_DEL / 2;
|
|
|
9561 if (*badsound == '*')
|
|
|
9562 ++badsound;
|
|
|
9563 else
|
|
|
9564 ++goodsound;
|
|
|
9565 }
|
|
|
9566
|
|
|
9567 goodlen = STRLEN(goodsound);
|
|
|
9568 badlen = STRLEN(badsound);
|
|
344
|
9569
|
|
|
9570 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
9571 * changes. */
|
|
|
9572 n = goodlen - badlen;
|
|
|
9573 if (n < -2 || n > 2)
|
|
|
9574 return SCORE_MAXMAX;
|
|
|
9575
|
|
|
9576 if (n > 0)
|
|
|
9577 {
|
|
351
|
9578 pl = goodsound; /* goodsound is longest */
|
|
344
|
9579 ps = badsound;
|
|
|
9580 }
|
|
|
9581 else
|
|
|
9582 {
|
|
351
|
9583 pl = badsound; /* badsound is longest */
|
|
344
|
9584 ps = goodsound;
|
|
|
9585 }
|
|
|
9586
|
|
|
9587 /* Skip over the identical part. */
|
|
|
9588 while (*pl == *ps && *pl != NUL)
|
|
|
9589 {
|
|
|
9590 ++pl;
|
|
|
9591 ++ps;
|
|
|
9592 }
|
|
|
9593
|
|
|
9594 switch (n)
|
|
|
9595 {
|
|
|
9596 case -2:
|
|
|
9597 case 2:
|
|
|
9598 /*
|
|
|
9599 * Must delete two characters from "pl".
|
|
|
9600 */
|
|
|
9601 ++pl; /* first delete */
|
|
|
9602 while (*pl == *ps)
|
|
|
9603 {
|
|
|
9604 ++pl;
|
|
|
9605 ++ps;
|
|
|
9606 }
|
|
|
9607 /* strings must be equal after second delete */
|
|
|
9608 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
9609 return score + SCORE_DEL * 2;
|
|
344
|
9610
|
|
|
9611 /* Failed to compare. */
|
|
|
9612 break;
|
|
|
9613
|
|
|
9614 case -1:
|
|
|
9615 case 1:
|
|
|
9616 /*
|
|
|
9617 * Minimal one delete from "pl" required.
|
|
|
9618 */
|
|
|
9619
|
|
|
9620 /* 1: delete */
|
|
|
9621 pl2 = pl + 1;
|
|
|
9622 ps2 = ps;
|
|
|
9623 while (*pl2 == *ps2)
|
|
|
9624 {
|
|
|
9625 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9626 return score + SCORE_DEL;
|
|
344
|
9627 ++pl2;
|
|
|
9628 ++ps2;
|
|
|
9629 }
|
|
|
9630
|
|
|
9631 /* 2: delete then swap, then rest must be equal */
|
|
|
9632 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9633 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9634 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
9635
|
|
|
9636 /* 3: delete then substitute, then the rest must be equal */
|
|
|
9637 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9638 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
9639
|
|
|
9640 /* 4: first swap then delete */
|
|
|
9641 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9642 {
|
|
|
9643 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9644 ps2 = ps + 2;
|
|
|
9645 while (*pl2 == *ps2)
|
|
|
9646 {
|
|
|
9647 ++pl2;
|
|
|
9648 ++ps2;
|
|
|
9649 }
|
|
|
9650 /* delete a char and then strings must be equal */
|
|
|
9651 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9652 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
9653 }
|
|
|
9654
|
|
|
9655 /* 5: first substitute then delete */
|
|
|
9656 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
9657 ps2 = ps + 1;
|
|
|
9658 while (*pl2 == *ps2)
|
|
|
9659 {
|
|
|
9660 ++pl2;
|
|
|
9661 ++ps2;
|
|
|
9662 }
|
|
|
9663 /* delete a char and then strings must be equal */
|
|
|
9664 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9665 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
9666
|
|
|
9667 /* Failed to compare. */
|
|
|
9668 break;
|
|
|
9669
|
|
|
9670 case 0:
|
|
|
9671 /*
|
|
|
9672 * Lenghts are equal, thus changes must result in same length: An
|
|
|
9673 * insert is only possible in combination with a delete.
|
|
|
9674 * 1: check if for identical strings
|
|
|
9675 */
|
|
|
9676 if (*pl == NUL)
|
|
351
|
9677 return score;
|
|
344
|
9678
|
|
|
9679 /* 2: swap */
|
|
|
9680 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
9681 {
|
|
|
9682 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
9683 ps2 = ps + 2;
|
|
|
9684 while (*pl2 == *ps2)
|
|
|
9685 {
|
|
|
9686 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9687 return score + SCORE_SWAP;
|
|
344
|
9688 ++pl2;
|
|
|
9689 ++ps2;
|
|
|
9690 }
|
|
|
9691 /* 3: swap and swap again */
|
|
|
9692 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9693 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9694 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
9695
|
|
|
9696 /* 4: swap and substitute */
|
|
|
9697 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9698 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
9699 }
|
|
|
9700
|
|
|
9701 /* 5: substitute */
|
|
|
9702 pl2 = pl + 1;
|
|
|
9703 ps2 = ps + 1;
|
|
|
9704 while (*pl2 == *ps2)
|
|
|
9705 {
|
|
|
9706 if (*pl2 == NUL) /* reached the end */
|
|
351
|
9707 return score + SCORE_SUBST;
|
|
344
|
9708 ++pl2;
|
|
|
9709 ++ps2;
|
|
|
9710 }
|
|
|
9711
|
|
|
9712 /* 6: substitute and swap */
|
|
|
9713 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
9714 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
9715 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
9716
|
|
|
9717 /* 7: substitute and substitute */
|
|
|
9718 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
9719 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
9720
|
|
|
9721 /* 8: insert then delete */
|
|
|
9722 pl2 = pl;
|
|
|
9723 ps2 = ps + 1;
|
|
|
9724 while (*pl2 == *ps2)
|
|
|
9725 {
|
|
|
9726 ++pl2;
|
|
|
9727 ++ps2;
|
|
|
9728 }
|
|
|
9729 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
9730 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9731
|
|
|
9732 /* 9: delete then insert */
|
|
|
9733 pl2 = pl + 1;
|
|
|
9734 ps2 = ps;
|
|
|
9735 while (*pl2 == *ps2)
|
|
|
9736 {
|
|
|
9737 ++pl2;
|
|
|
9738 ++ps2;
|
|
|
9739 }
|
|
|
9740 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
9741 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
9742
|
|
|
9743 /* Failed to compare. */
|
|
|
9744 break;
|
|
|
9745 }
|
|
|
9746
|
|
|
9747 return SCORE_MAXMAX;
|
|
|
9748 }
|
|
324
|
9749
|
|
323
|
9750 /*
|
|
|
9751 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
9752 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
9753 *
|
|
323
|
9754 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
9755 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
9756 * characters.
|
|
323
|
9757 */
|
|
|
9758 static int
|
|
|
9759 spell_edit_score(badword, goodword)
|
|
|
9760 char_u *badword;
|
|
|
9761 char_u *goodword;
|
|
|
9762 {
|
|
|
9763 int *cnt;
|
|
372
|
9764 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
9765 int j, i;
|
|
|
9766 int t;
|
|
|
9767 int bc, gc;
|
|
324
|
9768 int pbc, pgc;
|
|
|
9769 #ifdef FEAT_MBYTE
|
|
|
9770 char_u *p;
|
|
|
9771 int wbadword[MAXWLEN];
|
|
|
9772 int wgoodword[MAXWLEN];
|
|
|
9773
|
|
|
9774 if (has_mbyte)
|
|
|
9775 {
|
|
|
9776 /* Get the characters from the multi-byte strings and put them in an
|
|
|
9777 * int array for easy access. */
|
|
|
9778 for (p = badword, badlen = 0; *p != NUL; )
|
|
|
9779 wbadword[badlen++] = mb_ptr2char_adv(&p);
|
|
398
|
9780 wbadword[badlen++] = 0;
|
|
324
|
9781 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
|
9782 wgoodword[goodlen++] = mb_ptr2char_adv(&p);
|
|
398
|
9783 wgoodword[goodlen++] = 0;
|
|
324
|
9784 }
|
|
|
9785 else
|
|
|
9786 #endif
|
|
|
9787 {
|
|
|
9788 badlen = STRLEN(badword) + 1;
|
|
|
9789 goodlen = STRLEN(goodword) + 1;
|
|
|
9790 }
|
|
323
|
9791
|
|
|
9792 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
9793 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
9794 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
9795 TRUE);
|
|
324
|
9796 if (cnt == NULL)
|
|
|
9797 return 0; /* out of memory */
|
|
323
|
9798
|
|
|
9799 CNT(0, 0) = 0;
|
|
|
9800 for (j = 1; j <= goodlen; ++j)
|
|
|
9801 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
9802
|
|
|
9803 for (i = 1; i <= badlen; ++i)
|
|
|
9804 {
|
|
|
9805 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
9806 for (j = 1; j <= goodlen; ++j)
|
|
|
9807 {
|
|
324
|
9808 #ifdef FEAT_MBYTE
|
|
|
9809 if (has_mbyte)
|
|
|
9810 {
|
|
|
9811 bc = wbadword[i - 1];
|
|
|
9812 gc = wgoodword[j - 1];
|
|
|
9813 }
|
|
|
9814 else
|
|
|
9815 #endif
|
|
|
9816 {
|
|
|
9817 bc = badword[i - 1];
|
|
|
9818 gc = goodword[j - 1];
|
|
|
9819 }
|
|
323
|
9820 if (bc == gc)
|
|
|
9821 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
9822 else
|
|
|
9823 {
|
|
|
9824 /* Use a better score when there is only a case difference. */
|
|
324
|
9825 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
9826 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
9827 else
|
|
|
9828 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
9829
|
|
324
|
9830 if (i > 1 && j > 1)
|
|
323
|
9831 {
|
|
324
|
9832 #ifdef FEAT_MBYTE
|
|
|
9833 if (has_mbyte)
|
|
|
9834 {
|
|
|
9835 pbc = wbadword[i - 2];
|
|
|
9836 pgc = wgoodword[j - 2];
|
|
|
9837 }
|
|
|
9838 else
|
|
|
9839 #endif
|
|
|
9840 {
|
|
|
9841 pbc = badword[i - 2];
|
|
|
9842 pgc = goodword[j - 2];
|
|
|
9843 }
|
|
|
9844 if (bc == pgc && pbc == gc)
|
|
|
9845 {
|
|
|
9846 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
9847 if (t < CNT(i, j))
|
|
|
9848 CNT(i, j) = t;
|
|
|
9849 }
|
|
323
|
9850 }
|
|
|
9851 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
9852 if (t < CNT(i, j))
|
|
|
9853 CNT(i, j) = t;
|
|
|
9854 t = SCORE_INS + CNT(i, j - 1);
|
|
|
9855 if (t < CNT(i, j))
|
|
|
9856 CNT(i, j) = t;
|
|
|
9857 }
|
|
|
9858 }
|
|
|
9859 }
|
|
344
|
9860
|
|
|
9861 i = CNT(badlen - 1, goodlen - 1);
|
|
|
9862 vim_free(cnt);
|
|
|
9863 return i;
|
|
323
|
9864 }
|
|
307
|
9865
|
|
351
|
9866 /*
|
|
|
9867 * ":spelldump"
|
|
|
9868 */
|
|
|
9869 /*ARGSUSED*/
|
|
|
9870 void
|
|
|
9871 ex_spelldump(eap)
|
|
|
9872 exarg_T *eap;
|
|
|
9873 {
|
|
|
9874 buf_T *buf = curbuf;
|
|
|
9875 langp_T *lp;
|
|
|
9876 slang_T *slang;
|
|
|
9877 idx_T arridx[MAXWLEN];
|
|
|
9878 int curi[MAXWLEN];
|
|
|
9879 char_u word[MAXWLEN];
|
|
|
9880 int c;
|
|
|
9881 char_u *byts;
|
|
|
9882 idx_T *idxs;
|
|
|
9883 linenr_T lnum = 0;
|
|
|
9884 int round;
|
|
|
9885 int depth;
|
|
|
9886 int n;
|
|
|
9887 int flags;
|
|
381
|
9888 char_u *region_names = NULL; /* region names being used */
|
|
|
9889 int do_region = TRUE; /* dump region names and numbers */
|
|
|
9890 char_u *p;
|
|
351
|
9891
|
|
|
9892 if (no_spell_checking())
|
|
|
9893 return;
|
|
|
9894
|
|
|
9895 /* Create a new empty buffer by splitting the window. */
|
|
|
9896 do_cmdline_cmd((char_u *)"new");
|
|
|
9897 if (!bufempty() || !buf_valid(buf))
|
|
|
9898 return;
|
|
|
9899
|
|
381
|
9900 /* Find out if we can support regions: All languages must support the same
|
|
|
9901 * regions or none at all. */
|
|
|
9902 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
9903 {
|
|
|
9904 p = lp->lp_slang->sl_regions;
|
|
|
9905 if (p[0] != 0)
|
|
|
9906 {
|
|
|
9907 if (region_names == NULL) /* first language with regions */
|
|
|
9908 region_names = p;
|
|
|
9909 else if (STRCMP(region_names, p) != 0)
|
|
|
9910 {
|
|
|
9911 do_region = FALSE; /* region names are different */
|
|
|
9912 break;
|
|
|
9913 }
|
|
|
9914 }
|
|
|
9915 }
|
|
|
9916
|
|
|
9917 if (do_region && region_names != NULL)
|
|
|
9918 {
|
|
|
9919 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
9920 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
9921 }
|
|
|
9922 else
|
|
|
9923 do_region = FALSE;
|
|
|
9924
|
|
|
9925 /*
|
|
|
9926 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
9927 */
|
|
351
|
9928 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
9929 {
|
|
|
9930 slang = lp->lp_slang;
|
|
|
9931
|
|
|
9932 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
9933 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
9934
|
|
|
9935 /* round 1: case-folded tree
|
|
|
9936 * round 2: keep-case tree */
|
|
|
9937 for (round = 1; round <= 2; ++round)
|
|
|
9938 {
|
|
|
9939 if (round == 1)
|
|
|
9940 {
|
|
|
9941 byts = slang->sl_fbyts;
|
|
|
9942 idxs = slang->sl_fidxs;
|
|
|
9943 }
|
|
|
9944 else
|
|
|
9945 {
|
|
|
9946 byts = slang->sl_kbyts;
|
|
|
9947 idxs = slang->sl_kidxs;
|
|
|
9948 }
|
|
|
9949 if (byts == NULL)
|
|
|
9950 continue; /* array is empty */
|
|
|
9951
|
|
|
9952 depth = 0;
|
|
|
9953 arridx[0] = 0;
|
|
|
9954 curi[0] = 1;
|
|
|
9955 while (depth >= 0 && !got_int)
|
|
|
9956 {
|
|
|
9957 if (curi[depth] > byts[arridx[depth]])
|
|
|
9958 {
|
|
|
9959 /* Done all bytes at this node, go up one level. */
|
|
|
9960 --depth;
|
|
|
9961 line_breakcheck();
|
|
|
9962 }
|
|
|
9963 else
|
|
|
9964 {
|
|
|
9965 /* Do one more byte at this node. */
|
|
|
9966 n = arridx[depth] + curi[depth];
|
|
|
9967 ++curi[depth];
|
|
|
9968 c = byts[n];
|
|
|
9969 if (c == 0)
|
|
|
9970 {
|
|
|
9971 /* End of word, deal with the word.
|
|
|
9972 * Don't use keep-case words in the fold-case tree,
|
|
|
9973 * they will appear in the keep-case tree.
|
|
|
9974 * Only use the word when the region matches. */
|
|
|
9975 flags = (int)idxs[n];
|
|
|
9976 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
9977 && (do_region
|
|
|
9978 || (flags & WF_REGION) == 0
|
|
390
|
9979 || (((unsigned)flags >> 16)
|
|
351
|
9980 & lp->lp_region) != 0))
|
|
|
9981 {
|
|
|
9982 word[depth] = NUL;
|
|
381
|
9983 if (!do_region)
|
|
|
9984 flags &= ~WF_REGION;
|
|
355
|
9985
|
|
|
9986 /* Dump the basic word if there is no prefix or
|
|
|
9987 * when it's the first one. */
|
|
390
|
9988 c = (unsigned)flags >> 24;
|
|
355
|
9989 if (c == 0 || curi[depth] == 2)
|
|
|
9990 dump_word(word, round, flags, lnum++);
|
|
351
|
9991
|
|
|
9992 /* Apply the prefix, if there is one. */
|
|
355
|
9993 if (c != 0)
|
|
351
|
9994 lnum = apply_prefixes(slang, word, round,
|
|
|
9995 flags, lnum);
|
|
|
9996 }
|
|
|
9997 }
|
|
|
9998 else
|
|
|
9999 {
|
|
|
10000 /* Normal char, go one level deeper. */
|
|
|
10001 word[depth++] = c;
|
|
|
10002 arridx[depth] = idxs[n];
|
|
|
10003 curi[depth] = 1;
|
|
|
10004 }
|
|
|
10005 }
|
|
|
10006 }
|
|
|
10007 }
|
|
|
10008 }
|
|
|
10009
|
|
|
10010 /* Delete the empty line that we started with. */
|
|
|
10011 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
10012 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
10013
|
|
|
10014 redraw_later(NOT_VALID);
|
|
|
10015 }
|
|
|
10016
|
|
|
10017 /*
|
|
|
10018 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
10019 */
|
|
|
10020 static void
|
|
|
10021 dump_word(word, round, flags, lnum)
|
|
|
10022 char_u *word;
|
|
|
10023 int round;
|
|
|
10024 int flags;
|
|
|
10025 linenr_T lnum;
|
|
|
10026 {
|
|
|
10027 int keepcap = FALSE;
|
|
|
10028 char_u *p;
|
|
|
10029 char_u cword[MAXWLEN];
|
|
381
|
10030 char_u badword[MAXWLEN + 10];
|
|
|
10031 int i;
|
|
351
|
10032
|
|
|
10033 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
10034 {
|
|
|
10035 /* Need to fix case according to "flags". */
|
|
|
10036 make_case_word(word, cword, flags);
|
|
|
10037 p = cword;
|
|
|
10038 }
|
|
|
10039 else
|
|
|
10040 {
|
|
|
10041 p = word;
|
|
389
|
10042 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
10043 || (flags & WF_FIXCAP) != 0))
|
|
351
|
10044 keepcap = TRUE;
|
|
|
10045 }
|
|
|
10046
|
|
381
|
10047 /* Add flags and regions after a slash. */
|
|
|
10048 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
10049 {
|
|
|
10050 STRCPY(badword, p);
|
|
|
10051 STRCAT(badword, "/");
|
|
351
|
10052 if (keepcap)
|
|
|
10053 STRCAT(badword, "=");
|
|
|
10054 if (flags & WF_BANNED)
|
|
|
10055 STRCAT(badword, "!");
|
|
|
10056 else if (flags & WF_RARE)
|
|
|
10057 STRCAT(badword, "?");
|
|
381
|
10058 if (flags & WF_REGION)
|
|
|
10059 for (i = 0; i < 7; ++i)
|
|
390
|
10060 if (flags & (0x10000 << i))
|
|
381
|
10061 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
10062 p = badword;
|
|
|
10063 }
|
|
|
10064
|
|
|
10065 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
10066 }
|
|
|
10067
|
|
|
10068 /*
|
|
372
|
10069 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
10070 * "word" and append a line to the buffer.
|
|
351
|
10071 * Return the updated line number.
|
|
|
10072 */
|
|
|
10073 static linenr_T
|
|
|
10074 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
10075 slang_T *slang;
|
|
|
10076 char_u *word; /* case-folded word */
|
|
|
10077 int round;
|
|
|
10078 int flags; /* flags with prefix ID */
|
|
|
10079 linenr_T startlnum;
|
|
|
10080 {
|
|
|
10081 idx_T arridx[MAXWLEN];
|
|
|
10082 int curi[MAXWLEN];
|
|
|
10083 char_u prefix[MAXWLEN];
|
|
455
|
10084 char_u word_up[MAXWLEN];
|
|
|
10085 int has_word_up = FALSE;
|
|
351
|
10086 int c;
|
|
|
10087 char_u *byts;
|
|
|
10088 idx_T *idxs;
|
|
|
10089 linenr_T lnum = startlnum;
|
|
|
10090 int depth;
|
|
|
10091 int n;
|
|
|
10092 int len;
|
|
|
10093 int i;
|
|
|
10094
|
|
455
|
10095 /* if the word starts with a lower-case letter make the word with an
|
|
|
10096 * upper-case letter in word_up[]. */
|
|
|
10097 c = PTR2CHAR(word);
|
|
|
10098 if (SPELL_TOUPPER(c) != c)
|
|
|
10099 {
|
|
|
10100 onecap_copy(word, word_up, TRUE);
|
|
|
10101 has_word_up = TRUE;
|
|
|
10102 }
|
|
|
10103
|
|
351
|
10104 byts = slang->sl_pbyts;
|
|
|
10105 idxs = slang->sl_pidxs;
|
|
|
10106 if (byts != NULL) /* array not is empty */
|
|
|
10107 {
|
|
|
10108 /*
|
|
|
10109 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
10110 * When at the end of a prefix check that it supports "flags".
|
|
351
|
10111 */
|
|
|
10112 depth = 0;
|
|
|
10113 arridx[0] = 0;
|
|
|
10114 curi[0] = 1;
|
|
|
10115 while (depth >= 0 && !got_int)
|
|
|
10116 {
|
|
390
|
10117 n = arridx[depth];
|
|
|
10118 len = byts[n];
|
|
|
10119 if (curi[depth] > len)
|
|
351
|
10120 {
|
|
|
10121 /* Done all bytes at this node, go up one level. */
|
|
|
10122 --depth;
|
|
|
10123 line_breakcheck();
|
|
|
10124 }
|
|
|
10125 else
|
|
|
10126 {
|
|
|
10127 /* Do one more byte at this node. */
|
|
390
|
10128 n += curi[depth];
|
|
351
|
10129 ++curi[depth];
|
|
|
10130 c = byts[n];
|
|
|
10131 if (c == 0)
|
|
|
10132 {
|
|
|
10133 /* End of prefix, find out how many IDs there are. */
|
|
|
10134 for (i = 1; i < len; ++i)
|
|
|
10135 if (byts[n + i] != 0)
|
|
|
10136 break;
|
|
|
10137 curi[depth] += i - 1;
|
|
|
10138
|
|
455
|
10139 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
10140 if (c != 0)
|
|
351
|
10141 {
|
|
376
|
10142 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
10143 dump_word(prefix, round,
|
|
455
|
10144 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
10145 : flags, lnum++);
|
|
351
|
10146 }
|
|
455
|
10147
|
|
|
10148 /* Check for prefix that matches the word when the
|
|
|
10149 * first letter is upper-case, but only if the prefix has
|
|
|
10150 * a condition. */
|
|
|
10151 if (has_word_up)
|
|
|
10152 {
|
|
|
10153 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
10154 TRUE);
|
|
|
10155 if (c != 0)
|
|
|
10156 {
|
|
|
10157 vim_strncpy(prefix + depth, word_up,
|
|
|
10158 MAXWLEN - depth - 1);
|
|
|
10159 dump_word(prefix, round,
|
|
|
10160 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
10161 : flags, lnum++);
|
|
|
10162 }
|
|
|
10163 }
|
|
351
|
10164 }
|
|
|
10165 else
|
|
|
10166 {
|
|
|
10167 /* Normal char, go one level deeper. */
|
|
|
10168 prefix[depth++] = c;
|
|
|
10169 arridx[depth] = idxs[n];
|
|
|
10170 curi[depth] = 1;
|
|
|
10171 }
|
|
|
10172 }
|
|
|
10173 }
|
|
|
10174 }
|
|
|
10175
|
|
|
10176 return lnum;
|
|
|
10177 }
|
|
|
10178
|
|
236
|
10179 #endif /* FEAT_SYN_HL */
|
