|
223
|
1 /* vi:set ts=8 sts=4 sw=4:
|
|
|
2 *
|
|
|
3 * VIM - Vi IMproved by Bram Moolenaar
|
|
|
4 *
|
|
|
5 * Do ":help uganda" in Vim to read copying and usage conditions.
|
|
|
6 * Do ":help credits" in Vim to see a list of people who contributed.
|
|
|
7 * See README.txt for an overview of the Vim source code.
|
|
|
8 */
|
|
|
9
|
|
|
10 /*
|
|
|
11 * spell.c: code for spell checking
|
|
226
|
12 *
|
|
300
|
13 * The spell checking mechanism uses a tree (aka trie). Each node in the tree
|
|
|
14 * has a list of bytes that can appear (siblings). For each byte there is a
|
|
|
15 * pointer to the node with the byte that follows in the word (child).
|
|
324
|
16 *
|
|
|
17 * A NUL byte is used where the word may end. The bytes are sorted, so that
|
|
|
18 * binary searching can be used and the NUL bytes are at the start. The
|
|
|
19 * number of possible bytes is stored before the list of bytes.
|
|
|
20 *
|
|
|
21 * The tree uses two arrays: "byts" stores the characters, "idxs" stores
|
|
|
22 * either the next index or flags. The tree starts at index 0. For example,
|
|
|
23 * to lookup "vi" this sequence is followed:
|
|
|
24 * i = 0
|
|
|
25 * len = byts[i]
|
|
|
26 * n = where "v" appears in byts[i + 1] to byts[i + len]
|
|
|
27 * i = idxs[n]
|
|
|
28 * len = byts[i]
|
|
|
29 * n = where "i" appears in byts[i + 1] to byts[i + len]
|
|
|
30 * i = idxs[n]
|
|
|
31 * len = byts[i]
|
|
|
32 * find that byts[i + 1] is 0, idxs[i + 1] has flags for "vi".
|
|
300
|
33 *
|
|
339
|
34 * There are two word trees: one with case-folded words and one with words in
|
|
300
|
35 * original case. The second one is only used for keep-case words and is
|
|
|
36 * usually small.
|
|
|
37 *
|
|
481
|
38 * There is one additional tree for when not all prefixes are applied when
|
|
339
|
39 * generating the .spl file. This tree stores all the possible prefixes, as
|
|
|
40 * if they were words. At each word (prefix) end the prefix nr is stored, the
|
|
|
41 * following word must support this prefix nr. And the condition nr is
|
|
|
42 * stored, used to lookup the condition that the word must match with.
|
|
|
43 *
|
|
300
|
44 * Thanks to Olaf Seibert for providing an example implementation of this tree
|
|
|
45 * and the compression mechanism.
|
|
625
|
46 * LZ trie ideas:
|
|
|
47 * http://www.irb.hr/hr/home/ristov/papers/RistovLZtrieRevision1.pdf
|
|
|
48 * More papers: http://www-igm.univ-mlv.fr/~laporte/publi_en.html
|
|
243
|
49 *
|
|
|
50 * Matching involves checking the caps type: Onecap ALLCAP KeepCap.
|
|
|
51 *
|
|
236
|
52 * Why doesn't Vim use aspell/ispell/myspell/etc.?
|
|
|
53 * See ":help develop-spell".
|
|
|
54 */
|
|
|
55
|
|
469
|
56 /* Use SPELL_PRINTTREE for debugging: dump the word tree after adding a word.
|
|
482
|
57 * Only use it for small word lists! */
|
|
469
|
58 #if 0
|
|
|
59 # define SPELL_PRINTTREE
|
|
482
|
60 #endif
|
|
|
61
|
|
633
|
62 /* Use DEBUG_TRIEWALK to print the changes made in suggest_trie_walk() for a
|
|
|
63 * specific word. */
|
|
625
|
64 #if 0
|
|
|
65 # define DEBUG_TRIEWALK
|
|
|
66 #endif
|
|
|
67
|
|
300
|
68 /*
|
|
324
|
69 * Use this to adjust the score after finding suggestions, based on the
|
|
|
70 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
71 * it for every possible suggestion.
|
|
625
|
72 * Disadvantage: When "the" is typed as "hte" it sounds quite different ("@"
|
|
|
73 * vs "ht") and goes down in the list.
|
|
344
|
74 * Used when 'spellsuggest' is set to "best".
|
|
|
75 */
|
|
|
76 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
77
|
|
|
78 /*
|
|
625
|
79 * Do the opposite: based on a maximum end score and a known sound score,
|
|
|
80 * compute the the maximum word score that can be used.
|
|
|
81 */
|
|
|
82 #define MAXSCORE(word_score, sound_score) ((4 * word_score - sound_score) / 3)
|
|
|
83
|
|
|
84 /*
|
|
339
|
85 * Vim spell file format: <HEADER>
|
|
483
|
86 * <SECTIONS>
|
|
339
|
87 * <LWORDTREE>
|
|
|
88 * <KWORDTREE>
|
|
|
89 * <PREFIXTREE>
|
|
300
|
90 *
|
|
483
|
91 * <HEADER>: <fileID> <versionnr>
|
|
|
92 *
|
|
|
93 * <fileID> 8 bytes "VIMspell"
|
|
|
94 * <versionnr> 1 byte VIMSPELLVERSION
|
|
|
95 *
|
|
|
96 *
|
|
|
97 * Sections make it possible to add information to the .spl file without
|
|
|
98 * making it incompatible with previous versions. There are two kinds of
|
|
|
99 * sections:
|
|
|
100 * 1. Not essential for correct spell checking. E.g. for making suggestions.
|
|
|
101 * These are skipped when not supported.
|
|
|
102 * 2. Optional information, but essential for spell checking when present.
|
|
|
103 * E.g. conditions for affixes. When this section is present but not
|
|
|
104 * supported an error message is given.
|
|
300
|
105 *
|
|
483
|
106 * <SECTIONS>: <section> ... <sectionend>
|
|
|
107 *
|
|
|
108 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
109 *
|
|
|
110 * <sectionID> 1 byte number from 0 to 254 identifying the section
|
|
|
111 *
|
|
|
112 * <sectionflags> 1 byte SNF_REQUIRED: this section is required for correct
|
|
|
113 * spell checking
|
|
|
114 *
|
|
|
115 * <sectionlen> 4 bytes length of section contents, MSB first
|
|
|
116 *
|
|
|
117 * <sectionend> 1 byte SN_END
|
|
|
118 *
|
|
|
119 *
|
|
|
120 * sectionID == SN_REGION: <regionname> ...
|
|
|
121 * <regionname> 2 bytes Up to 8 region names: ca, au, etc. Lower case.
|
|
300
|
122 * First <regionname> is region 1.
|
|
|
123 *
|
|
483
|
124 * sectionID == SN_CHARFLAGS: <charflagslen> <charflags>
|
|
|
125 * <folcharslen> <folchars>
|
|
300
|
126 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
127 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
128 * 0x01 word character CF_WORD
|
|
|
129 * 0x02 upper-case character CF_UPPER
|
|
483
|
130 * <folcharslen> 2 bytes Number of bytes in <folchars>.
|
|
|
131 * <folchars> N bytes Folded characters, first one is for character 128.
|
|
300
|
132 *
|
|
483
|
133 * sectionID == SN_MIDWORD: <midword>
|
|
|
134 * <midword> N bytes Characters that are word characters only when used
|
|
366
|
135 * in the middle of a word.
|
|
|
136 *
|
|
483
|
137 * sectionID == SN_PREFCOND: <prefcondcnt> <prefcond> ...
|
|
339
|
138 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
139 * <prefcond> : <condlen> <condstr>
|
|
|
140 * <condlen> 1 byte Length of <condstr>.
|
|
|
141 * <condstr> N bytes Condition for the prefix.
|
|
|
142 *
|
|
483
|
143 * sectionID == SN_REP: <repcount> <rep> ...
|
|
|
144 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
323
|
145 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
483
|
146 * <repfromlen> 1 byte length of <repfrom>
|
|
|
147 * <repfrom> N bytes "from" part of replacement
|
|
|
148 * <reptolen> 1 byte length of <repto>
|
|
|
149 * <repto> N bytes "to" part of replacement
|
|
323
|
150 *
|
|
625
|
151 * sectionID == SN_REPSAL: <repcount> <rep> ...
|
|
|
152 * just like SN_REP but for soundfolded words
|
|
|
153 *
|
|
483
|
154 * sectionID == SN_SAL: <salflags> <salcount> <sal> ...
|
|
|
155 * <salflags> 1 byte flags for soundsalike conversion:
|
|
323
|
156 * SAL_F0LLOWUP
|
|
|
157 * SAL_COLLAPSE
|
|
|
158 * SAL_REM_ACCENTS
|
|
483
|
159 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
160 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
483
|
161 * <salfromlen> 1 byte length of <salfrom>
|
|
|
162 * <salfrom> N bytes "from" part of soundsalike
|
|
|
163 * <saltolen> 1 byte length of <salto>
|
|
|
164 * <salto> N bytes "to" part of soundsalike
|
|
323
|
165 *
|
|
483
|
166 * sectionID == SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
167 * <sofofromlen> 2 bytes length of <sofofrom>
|
|
|
168 * <sofofrom> N bytes "from" part of soundfold
|
|
|
169 * <sofotolen> 2 bytes length of <sofoto>
|
|
|
170 * <sofoto> N bytes "to" part of soundfold
|
|
323
|
171 *
|
|
625
|
172 * sectionID == SN_SUGFILE: <timestamp>
|
|
|
173 * <timestamp> 8 bytes time in seconds that must match with .sug file
|
|
|
174 *
|
|
|
175 * sectionID == SN_WORDS: <word> ...
|
|
|
176 * <word> N bytes NUL terminated common word
|
|
|
177 *
|
|
483
|
178 * sectionID == SN_MAP: <mapstr>
|
|
|
179 * <mapstr> N bytes String with sequences of similar characters,
|
|
323
|
180 * separated by slashes.
|
|
300
|
181 *
|
|
483
|
182 * sectionID == SN_COMPOUND: <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
183 * <compmax> 1 byte Maximum nr of words in compound word.
|
|
|
184 * <compminlen> 1 byte Minimal word length for compounding.
|
|
|
185 * <compsylmax> 1 byte Maximum nr of syllables in compound word.
|
|
|
186 * <compflags> N bytes Flags from COMPOUNDFLAGS items, separated by
|
|
|
187 * slashes.
|
|
|
188 *
|
|
492
|
189 * sectionID == SN_NOBREAK: (empty, its presence is enough)
|
|
|
190 *
|
|
483
|
191 * sectionID == SN_SYLLABLE: <syllable>
|
|
|
192 * <syllable> N bytes String from SYLLABLE item.
|
|
300
|
193 *
|
|
|
194 * <LWORDTREE>: <wordtree>
|
|
|
195 *
|
|
339
|
196 * <KWORDTREE>: <wordtree>
|
|
|
197 *
|
|
|
198 * <PREFIXTREE>: <wordtree>
|
|
|
199 *
|
|
|
200 *
|
|
300
|
201 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
202 *
|
|
|
203 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
204 *
|
|
|
205 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
206 *
|
|
|
207 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
208 * follow in sorted order.
|
|
|
209 *
|
|
339
|
210 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
481
|
211 * | <flags> [<flags2>] [<region>] [<affixID>]
|
|
|
212 * | [<pflags>] <affixID> <prefcondnr> ]
|
|
300
|
213 *
|
|
|
214 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
215 * BY_NOFLAGS: End of word without flags and for all
|
|
|
216 * regions.
|
|
481
|
217 * For PREFIXTREE <affixID> and
|
|
366
|
218 * <prefcondnr> follow.
|
|
390
|
219 * BY_FLAGS: End of word, <flags> follow.
|
|
481
|
220 * For PREFIXTREE <pflags>, <affixID>
|
|
455
|
221 * and <prefcondnr> follow.
|
|
390
|
222 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
223 * follow. Not used in PREFIXTREE.
|
|
|
224 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
225 * and <xbyte> follow.
|
|
|
226 *
|
|
|
227 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
228 *
|
|
|
229 * <xbyte> 1 byte byte value of the sibling.
|
|
|
230 *
|
|
|
231 * <flags> 1 byte bitmask of:
|
|
|
232 * WF_ALLCAP word must have only capitals
|
|
|
233 * WF_ONECAP first char of word must be capital
|
|
389
|
234 * WF_KEEPCAP keep-case word
|
|
|
235 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
236 * WF_RARE rare word
|
|
389
|
237 * WF_BANNED bad word
|
|
300
|
238 * WF_REGION <region> follows
|
|
481
|
239 * WF_AFX <affixID> follows
|
|
300
|
240 *
|
|
500
|
241 * <flags2> 1 byte Bitmask of:
|
|
390
|
242 * WF_HAS_AFF >> 8 word includes affix
|
|
500
|
243 * WF_NEEDCOMP >> 8 word only valid in compound
|
|
390
|
244 *
|
|
455
|
245 * <pflags> 1 byte bitmask of:
|
|
|
246 * WFP_RARE rare prefix
|
|
|
247 * WFP_NC non-combining prefix
|
|
|
248 * WFP_UP letter after prefix made upper case
|
|
|
249 *
|
|
300
|
250 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
251 * omitted it's valid in all regions.
|
|
|
252 * Lowest bit is for region 1.
|
|
|
253 *
|
|
481
|
254 * <affixID> 1 byte ID of affix that can be used with this word. In
|
|
339
|
255 * PREFIXTREE used for the required prefix ID.
|
|
|
256 *
|
|
|
257 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
258 * from HEADER.
|
|
300
|
259 *
|
|
|
260 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
261 */
|
|
|
262
|
|
625
|
263 /*
|
|
|
264 * Vim .sug file format: <SUGHEADER>
|
|
|
265 * <SUGWORDTREE>
|
|
|
266 * <SUGTABLE>
|
|
|
267 *
|
|
|
268 * <SUGHEADER>: <fileID> <versionnr> <timestamp>
|
|
|
269 *
|
|
|
270 * <fileID> 6 bytes "VIMsug"
|
|
|
271 * <versionnr> 1 byte VIMSUGVERSION
|
|
|
272 * <timestamp> 8 bytes timestamp that must match with .spl file
|
|
|
273 *
|
|
|
274 *
|
|
|
275 * <SUGWORDTREE>: <wordtree> (see above, no flags or region used)
|
|
|
276 *
|
|
|
277 *
|
|
|
278 * <SUGTABLE>: <sugwcount> <sugline> ...
|
|
|
279 *
|
|
|
280 * <sugwcount> 4 bytes number of <sugline> following
|
|
|
281 *
|
|
|
282 * <sugline>: <sugnr> ... NUL
|
|
|
283 *
|
|
|
284 * <sugnr>: X bytes word number that results in this soundfolded word,
|
|
|
285 * stored as an offset to the previous number in as
|
|
|
286 * few bytes as possible, see offset2bytes())
|
|
|
287 */
|
|
|
288
|
|
223
|
289 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
290 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
291 #endif
|
|
|
292
|
|
|
293 #include "vim.h"
|
|
|
294
|
|
|
295 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
296
|
|
|
297 #ifdef HAVE_FCNTL_H
|
|
|
298 # include <fcntl.h>
|
|
|
299 #endif
|
|
|
300
|
|
625
|
301 #ifndef UNIX /* it's in os_unix.h for Unix */
|
|
|
302 # include <time.h> /* for time_t */
|
|
|
303 #endif
|
|
|
304
|
|
323
|
305 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
306 Some places assume a word length fits in a
|
|
|
307 byte, thus it can't be above 255. */
|
|
226
|
308
|
|
490
|
309 /* Type used for indexes in the word tree need to be at least 4 bytes. If int
|
|
324
|
310 * is 8 bytes we could use something smaller, but what? */
|
|
490
|
311 #if SIZEOF_INT > 3
|
|
324
|
312 typedef int idx_T;
|
|
|
313 #else
|
|
|
314 typedef long idx_T;
|
|
|
315 #endif
|
|
|
316
|
|
|
317 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
318 * comes above it. */
|
|
300
|
319 #define WF_REGION 0x01 /* region byte follows */
|
|
|
320 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
321 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
322 #define WF_RARE 0x08 /* rare word */
|
|
307
|
323 #define WF_BANNED 0x10 /* bad word */
|
|
481
|
324 #define WF_AFX 0x20 /* affix ID follows */
|
|
389
|
325 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
326 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
327
|
|
390
|
328 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
329 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
500
|
330 #define WF_NEEDCOMP 0x0200 /* word only valid in compound */
|
|
390
|
331
|
|
633
|
332 /* only used for su_badflags */
|
|
|
333 #define WF_MIXCAP 0x20 /* mix of upper and lower case: macaRONI */
|
|
|
334
|
|
389
|
335 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
336
|
|
455
|
337 /* flags for <pflags> */
|
|
|
338 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
339 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
340 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
341
|
|
481
|
342 /* Flags for postponed prefixes. Must be above affixID (one byte)
|
|
390
|
343 * and prefcondnr (two bytes). */
|
|
455
|
344 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
345 * postponed prefix */
|
|
|
346 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
347 * postponed prefix */
|
|
|
348 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
349 * postponed prefix */
|
|
390
|
350
|
|
|
351 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
352 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
353 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
354 * postponed prefix: no <pflags> */
|
|
|
355 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
356 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
357 * postponed prefix: <pflags> follows */
|
|
|
358 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
359 * follow; never used in prefix tree */
|
|
|
360 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
361
|
|
625
|
362 /* Info from "REP", "REPSAL" and "SAL" entries in ".aff" file used in si_rep,
|
|
|
363 * si_repsal, sl_rep, and si_sal. Not for sl_sal!
|
|
323
|
364 * One replacement: from "ft_from" to "ft_to". */
|
|
|
365 typedef struct fromto_S
|
|
236
|
366 {
|
|
323
|
367 char_u *ft_from;
|
|
|
368 char_u *ft_to;
|
|
|
369 } fromto_T;
|
|
236
|
370
|
|
344
|
371 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
372 * The info is split for quick processing by spell_soundfold().
|
|
|
373 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
374 typedef struct salitem_S
|
|
|
375 {
|
|
|
376 char_u *sm_lead; /* leading letters */
|
|
|
377 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
378 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
379 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
380 char_u *sm_to; /* replacement. */
|
|
372
|
381 #ifdef FEAT_MBYTE
|
|
|
382 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
383 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
384 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
385 #endif
|
|
344
|
386 } salitem_T;
|
|
|
387
|
|
375
|
388 #ifdef FEAT_MBYTE
|
|
|
389 typedef int salfirst_T;
|
|
|
390 #else
|
|
|
391 typedef short salfirst_T;
|
|
|
392 #endif
|
|
|
393
|
|
483
|
394 /* Values for SP_*ERROR are negative, positive values are used by
|
|
|
395 * read_cnt_string(). */
|
|
|
396 #define SP_TRUNCERROR -1 /* spell file truncated error */
|
|
|
397 #define SP_FORMERROR -2 /* format error in spell file */
|
|
495
|
398 #define SP_OTHERERROR -3 /* other error while reading spell file */
|
|
483
|
399
|
|
236
|
400 /*
|
|
243
|
401 * Structure used to store words and other info for one language, loaded from
|
|
|
402 * a .spl file.
|
|
300
|
403 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
404 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
405 *
|
|
|
406 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
407 * the number of possible bytes, sorted on byte value:
|
|
|
408 * <len> <byte1> <byte2> ...
|
|
|
409 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
410 * byte in "byts".
|
|
|
411 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
481
|
412 * the flags, region mask and affixID for the word. There may be several
|
|
|
413 * zeros in sequence for alternative flag/region/affixID combinations.
|
|
236
|
414 */
|
|
|
415 typedef struct slang_S slang_T;
|
|
|
416 struct slang_S
|
|
|
417 {
|
|
|
418 slang_T *sl_next; /* next language */
|
|
|
419 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
420 char_u *sl_fname; /* name of .spl file */
|
|
323
|
421 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
422
|
|
300
|
423 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
424 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
425 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
426 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
427 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
428 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
429
|
|
236
|
430 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
431
|
|
376
|
432 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
433
|
|
625
|
434 hashtab_T sl_wordcount; /* hashtable with word count, wordcount_T */
|
|
|
435
|
|
483
|
436 int sl_compmax; /* COMPOUNDMAX (default: MAXWLEN) */
|
|
501
|
437 int sl_compminlen; /* COMPOUNDMIN (default: 0) */
|
|
483
|
438 int sl_compsylmax; /* COMPOUNDSYLMAX (default: MAXWLEN) */
|
|
|
439 regprog_T *sl_compprog; /* COMPOUNDFLAGS turned into a regexp progrm
|
|
|
440 * (NULL when no compounding) */
|
|
|
441 char_u *sl_compstartflags; /* flags for first compound word */
|
|
485
|
442 char_u *sl_compallflags; /* all flags for compound words */
|
|
492
|
443 char_u sl_nobreak; /* When TRUE: no spaces between words */
|
|
483
|
444 char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */
|
|
|
445 garray_T sl_syl_items; /* syllable items */
|
|
481
|
446
|
|
339
|
447 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
448 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
449
|
|
323
|
450 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
451 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
452 there is none */
|
|
344
|
453 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
454 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
455 there is none */
|
|
625
|
456 int sl_followup; /* SAL followup */
|
|
|
457 int sl_collapse; /* SAL collapse_result */
|
|
|
458 int sl_rem_accents; /* SAL remove_accents */
|
|
375
|
459 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
460 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
461 * "sl_sal" is a list of wide char lists. */
|
|
625
|
462 garray_T sl_repsal; /* list of fromto_T entries from REPSAL lines */
|
|
|
463 short sl_repsal_first[256]; /* sl_rep_first for REPSAL lines */
|
|
|
464
|
|
|
465 /* Info from the .sug file. Loaded on demand. */
|
|
|
466 time_t sl_sugtime; /* timestamp for .sug file */
|
|
|
467 char_u *sl_sbyts; /* soundfolded word bytes */
|
|
|
468 idx_T *sl_sidxs; /* soundfolded word indexes */
|
|
|
469 buf_T *sl_sugbuf; /* buffer with word number table */
|
|
|
470 int sl_sugloaded; /* TRUE when .sug file was loaded or failed to
|
|
|
471 load */
|
|
|
472
|
|
330
|
473 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
474 #ifdef FEAT_MBYTE
|
|
|
475 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
476 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
477 #else
|
|
|
478 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
479 #endif
|
|
625
|
480 hashtab_T sl_sounddone; /* table with soundfolded words that have
|
|
|
481 handled, see add_sound_suggest() */
|
|
236
|
482 };
|
|
|
483
|
|
243
|
484 /* First language that is loaded, start of the linked list of loaded
|
|
|
485 * languages. */
|
|
236
|
486 static slang_T *first_lang = NULL;
|
|
|
487
|
|
323
|
488 /* Flags used in .spl file for soundsalike flags. */
|
|
|
489 #define SAL_F0LLOWUP 1
|
|
|
490 #define SAL_COLLAPSE 2
|
|
|
491 #define SAL_REM_ACCENTS 4
|
|
|
492
|
|
236
|
493 /*
|
|
|
494 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
495 */
|
|
|
496 typedef struct langp_S
|
|
|
497 {
|
|
503
|
498 slang_T *lp_slang; /* info for this language */
|
|
|
499 slang_T *lp_sallang; /* language used for sound folding or NULL */
|
|
|
500 slang_T *lp_replang; /* language used for REP items or NULL */
|
|
236
|
501 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
502 } langp_T;
|
|
|
503
|
|
|
504 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
505
|
|
307
|
506 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
507
|
|
483
|
508 #define VIMSPELLMAGIC "VIMspell" /* string at start of Vim spell file */
|
|
|
509 #define VIMSPELLMAGICL 8
|
|
|
510 #define VIMSPELLVERSION 50
|
|
|
511
|
|
625
|
512 #define VIMSUGMAGIC "VIMsug" /* string at start of Vim .sug file */
|
|
|
513 #define VIMSUGMAGICL 6
|
|
|
514 #define VIMSUGVERSION 1
|
|
|
515
|
|
483
|
516 /* Section IDs. Only renumber them when VIMSPELLVERSION changes! */
|
|
|
517 #define SN_REGION 0 /* <regionname> section */
|
|
|
518 #define SN_CHARFLAGS 1 /* charflags section */
|
|
|
519 #define SN_MIDWORD 2 /* <midword> section */
|
|
|
520 #define SN_PREFCOND 3 /* <prefcond> section */
|
|
|
521 #define SN_REP 4 /* REP items section */
|
|
|
522 #define SN_SAL 5 /* SAL items section */
|
|
|
523 #define SN_SOFO 6 /* soundfolding section */
|
|
|
524 #define SN_MAP 7 /* MAP items section */
|
|
|
525 #define SN_COMPOUND 8 /* compound words section */
|
|
|
526 #define SN_SYLLABLE 9 /* syllable section */
|
|
492
|
527 #define SN_NOBREAK 10 /* NOBREAK section */
|
|
625
|
528 #define SN_SUGFILE 11 /* timestamp for .sug file */
|
|
|
529 #define SN_REPSAL 12 /* REPSAL items section */
|
|
|
530 #define SN_WORDS 13 /* common words */
|
|
483
|
531 #define SN_END 255 /* end of sections */
|
|
|
532
|
|
|
533 #define SNF_REQUIRED 1 /* <sectionflags>: required section */
|
|
|
534
|
|
307
|
535 /* Result values. Lower number is accepted over higher one. */
|
|
|
536 #define SP_BANNED -1
|
|
236
|
537 #define SP_OK 0
|
|
307
|
538 #define SP_RARE 1
|
|
|
539 #define SP_LOCAL 2
|
|
|
540 #define SP_BAD 3
|
|
236
|
541
|
|
381
|
542 /* file used for "zG" and "zW" */
|
|
385
|
543 static char_u *int_wordlist = NULL;
|
|
381
|
544
|
|
625
|
545 typedef struct wordcount_S
|
|
|
546 {
|
|
|
547 short_u wc_count; /* nr of times word was seen */
|
|
|
548 char_u wc_word[1]; /* word, actually longer */
|
|
|
549 } wordcount_T;
|
|
|
550
|
|
|
551 static wordcount_T dumwc;
|
|
|
552 #define WC_KEY_OFF (dumwc.wc_word - (char_u *)&dumwc)
|
|
|
553 #define HI2WC(hi) ((wordcount_T *)((hi)->hi_key - WC_KEY_OFF))
|
|
|
554 #define MAXWORDCOUNT 0xffff
|
|
|
555
|
|
236
|
556 /*
|
|
323
|
557 * Information used when looking for suggestions.
|
|
|
558 */
|
|
|
559 typedef struct suginfo_S
|
|
|
560 {
|
|
|
561 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
562 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
563 int su_maxscore; /* maximum score for adding to su_ga */
|
|
625
|
564 int su_sfmaxscore; /* idem, for when doing soundfold words */
|
|
344
|
565 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
566 char_u *su_badptr; /* start of bad word in line */
|
|
|
567 int su_badlen; /* length of detected bad word in line */
|
|
346
|
568 int su_badflags; /* caps flags for bad word */
|
|
323
|
569 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
570 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
534
|
571 char_u su_sal_badword[MAXWLEN]; /* su_badword soundfolded */
|
|
323
|
572 hashtab_T su_banned; /* table with banned words */
|
|
503
|
573 slang_T *su_sallang; /* default language for sound folding */
|
|
323
|
574 } suginfo_T;
|
|
|
575
|
|
|
576 /* One word suggestion. Used in "si_ga". */
|
|
|
577 typedef struct suggest_S
|
|
|
578 {
|
|
|
579 char_u *st_word; /* suggested word, allocated string */
|
|
625
|
580 int st_wordlen; /* STRLEN(st_word) */
|
|
323
|
581 int st_orglen; /* length of replaced text */
|
|
|
582 int st_score; /* lower is better */
|
|
344
|
583 int st_altscore; /* used when st_score compares equal */
|
|
|
584 int st_salscore; /* st_score is for soundalike */
|
|
324
|
585 int st_had_bonus; /* bonus already included in score */
|
|
503
|
586 slang_T *st_slang; /* language used for sound folding */
|
|
323
|
587 } suggest_T;
|
|
|
588
|
|
344
|
589 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
590
|
|
625
|
591 /* TRUE if a word appears in the list of banned words. */
|
|
|
592 #define WAS_BANNED(su, word) (!HASHITEM_EMPTY(hash_find(&su->su_banned, word)))
|
|
|
593
|
|
|
594 /* Number of suggestions kept when cleaning up. we need to keep more than
|
|
|
595 * what is displayed, because when rescore_suggestions() is called the score
|
|
|
596 * may change and wrong suggestions may be removed later. */
|
|
|
597 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 130 ? 150 : (su)->su_maxcount + 20)
|
|
324
|
598
|
|
|
599 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
600 * of suggestions that are not going to be displayed. */
|
|
625
|
601 #define SUG_MAX_COUNT(su) (SUG_CLEAN_COUNT(su) + 50)
|
|
323
|
602
|
|
|
603 /* score for various changes */
|
|
344
|
604 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
605 #define SCORE_ICASE 52 /* slightly different case */
|
|
483
|
606 #define SCORE_REGION 200 /* word is for different region */
|
|
323
|
607 #define SCORE_RARE 180 /* rare word */
|
|
625
|
608 #define SCORE_SWAP 75 /* swap two characters */
|
|
323
|
609 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
532
|
610 #define SCORE_REP 65 /* REP replacement */
|
|
323
|
611 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
612 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
479
|
613 #define SCORE_SUBCOMP 33 /* substitute a composing character */
|
|
324
|
614 #define SCORE_DEL 94 /* delete a character */
|
|
532
|
615 #define SCORE_DELDUP 66 /* delete a duplicated character */
|
|
479
|
616 #define SCORE_DELCOMP 28 /* delete a composing character */
|
|
324
|
617 #define SCORE_INS 96 /* insert a character */
|
|
532
|
618 #define SCORE_INSDUP 67 /* insert a duplicate character */
|
|
475
|
619 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
620 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
621
|
|
372
|
622 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
623 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
624 * 350 allows for about three changes. */
|
|
344
|
625
|
|
625
|
626 #define SCORE_COMMON1 30 /* subtracted for words seen before */
|
|
|
627 #define SCORE_COMMON2 40 /* subtracted for words often seen */
|
|
|
628 #define SCORE_COMMON3 50 /* subtracted for words very often seen */
|
|
|
629 #define SCORE_THRES2 10 /* word count threshold for COMMON2 */
|
|
|
630 #define SCORE_THRES3 100 /* word count threshold for COMMON3 */
|
|
|
631
|
|
|
632 /* When trying changed soundfold words it becomes slow when trying more than
|
|
|
633 * two changes. With less then two changes it's slightly faster but we miss a
|
|
|
634 * few good suggestions. In rare cases we need to try three of four changes.
|
|
|
635 */
|
|
|
636 #define SCORE_SFMAX1 200 /* maximum score for first try */
|
|
|
637 #define SCORE_SFMAX2 300 /* maximum score for second try */
|
|
|
638 #define SCORE_SFMAX3 400 /* maximum score for third try */
|
|
|
639
|
|
344
|
640 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
625
|
641 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
642 #define SCORE_LIMITMAX 350 /* for spell_edit_score_limit() */
|
|
|
643
|
|
|
644 /* for spell_edit_score_limit() we need to know the minimum value of
|
|
|
645 * SCORE_ICASE, SCORE_SWAP, SCORE_DEL, SCORE_SIMILAR and SCORE_INS */
|
|
|
646 #define SCORE_EDIT_MIN SCORE_SIMILAR
|
|
323
|
647
|
|
|
648 /*
|
|
236
|
649 * Structure to store info for word matching.
|
|
|
650 */
|
|
|
651 typedef struct matchinf_S
|
|
|
652 {
|
|
|
653 langp_T *mi_lp; /* info for language and region */
|
|
243
|
654
|
|
|
655 /* pointers to original text to be checked */
|
|
236
|
656 char_u *mi_word; /* start of word being checked */
|
|
339
|
657 char_u *mi_end; /* end of matching word so far */
|
|
243
|
658 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
659 char_u *mi_cend; /* char after what was used for
|
|
|
660 mi_capflags */
|
|
243
|
661
|
|
|
662 /* case-folded text */
|
|
|
663 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
664 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
665
|
|
339
|
666 /* for when checking word after a prefix */
|
|
|
667 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
481
|
668 affixID/condition */
|
|
339
|
669 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
670 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
671 #ifdef FEAT_MBYTE
|
|
|
672 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
673 case */
|
|
|
674 #else
|
|
|
675 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
676 #endif
|
|
339
|
677
|
|
481
|
678 /* for when checking a compound word */
|
|
|
679 int mi_compoff; /* start of following word offset */
|
|
483
|
680 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */
|
|
|
681 int mi_complen; /* nr of compound words used */
|
|
481
|
682
|
|
243
|
683 /* others */
|
|
236
|
684 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
685 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
686 buf_T *mi_buf; /* buffer being checked */
|
|
492
|
687
|
|
|
688 /* for NOBREAK */
|
|
|
689 int mi_result2; /* "mi_resul" without following word */
|
|
|
690 char_u *mi_end2; /* "mi_end" without following word */
|
|
236
|
691 } matchinf_T;
|
|
|
692
|
|
307
|
693 /*
|
|
|
694 * The tables used for recognizing word characters according to spelling.
|
|
|
695 * These are only used for the first 256 characters of 'encoding'.
|
|
|
696 */
|
|
|
697 typedef struct spelltab_S
|
|
|
698 {
|
|
|
699 char_u st_isw[256]; /* flags: is word char */
|
|
|
700 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
701 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
702 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
703 } spelltab_T;
|
|
|
704
|
|
|
705 static spelltab_T spelltab;
|
|
|
706 static int did_set_spelltab;
|
|
|
707
|
|
324
|
708 #define CF_WORD 0x01
|
|
|
709 #define CF_UPPER 0x02
|
|
307
|
710
|
|
|
711 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
712 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
713 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
714 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
715 #ifdef FEAT_MBYTE
|
|
|
716 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
717 #endif
|
|
483
|
718 static int write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
719
|
|
|
720 /*
|
|
344
|
721 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
722 */
|
|
|
723 typedef enum
|
|
|
724 {
|
|
344
|
725 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
726 * ends); if badword ends there is a match, otherwise
|
|
|
727 * try splitting word. */
|
|
375
|
728 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
729 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
730 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
731 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
732 STATE_DEL, /* Delete a byte from the bad word. */
|
|
625
|
733 STATE_INS_PREP, /* Prepare for inserting bytes. */
|
|
330
|
734 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
735 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
736 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
737 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
738 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
739 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
740 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
741 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
742 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
743 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
744 STATE_FINAL /* End of this node. */
|
|
|
745 } state_T;
|
|
|
746
|
|
|
747 /*
|
|
346
|
748 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
749 */
|
|
|
750 typedef struct trystate_S
|
|
|
751 {
|
|
330
|
752 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
753 int ts_score; /* score */
|
|
344
|
754 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
755 short ts_curi; /* index in list of child nodes */
|
|
|
756 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
757 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
758 char_u ts_twordlen; /* valid length of tword[] */
|
|
482
|
759 char_u ts_prefixdepth; /* stack depth for end of prefix or
|
|
485
|
760 * PFD_PREFIXTREE or PFD_NOPREFIX */
|
|
|
761 char_u ts_flags; /* TSF_ flags */
|
|
330
|
762 #ifdef FEAT_MBYTE
|
|
|
763 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
764 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
765 char_u ts_isdiff; /* DIFF_ values */
|
|
|
766 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
767 #endif
|
|
483
|
768 char_u ts_prewordlen; /* length of word in "preword[]" */
|
|
|
769 char_u ts_splitoff; /* index in "tword" after last split */
|
|
492
|
770 char_u ts_splitfidx; /* "ts_fidx" at word split */
|
|
483
|
771 char_u ts_complen; /* nr of compound words used */
|
|
485
|
772 char_u ts_compsplit; /* index for "compflags" where word was spit */
|
|
346
|
773 char_u ts_save_badflags; /* su_badflags saved here */
|
|
625
|
774 char_u ts_delidx; /* index in fword for char that was deleted,
|
|
|
775 valid when "ts_flags" has TSF_DIDDEL */
|
|
323
|
776 } trystate_T;
|
|
|
777
|
|
330
|
778 /* values for ts_isdiff */
|
|
|
779 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
780 #define DIFF_YES 1 /* different byte found */
|
|
|
781 #define DIFF_INSERT 2 /* inserting character */
|
|
|
782
|
|
485
|
783 /* values for ts_flags */
|
|
|
784 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */
|
|
|
785 #define TSF_DIDSPLIT 2 /* tried split at this point */
|
|
625
|
786 #define TSF_DIDDEL 4 /* did a delete, "ts_delidx" has index */
|
|
485
|
787
|
|
375
|
788 /* special values ts_prefixdepth */
|
|
485
|
789 #define PFD_NOPREFIX 0xff /* not using prefixes */
|
|
482
|
790 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
625
|
791 #define PFD_NOTSPECIAL 0xfd /* highest value that's not special */
|
|
375
|
792
|
|
339
|
793 /* mode values for find_word */
|
|
481
|
794 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
795 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
796 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
797 #define FIND_COMPOUND 3 /* find case-folded compound word */
|
|
|
798 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */
|
|
339
|
799
|
|
236
|
800 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
801 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
802 static void slang_clear __ARGS((slang_T *lp));
|
|
625
|
803 static void slang_clear_sug __ARGS((slang_T *lp));
|
|
339
|
804 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
483
|
805 static int can_compound __ARGS((slang_T *slang, char_u *word, char_u *flags));
|
|
455
|
806 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
485
|
807 static void find_prefix __ARGS((matchinf_T *mip, int mode));
|
|
339
|
808 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
809 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
498
|
810 static int no_spell_checking __ARGS((win_T *wp));
|
|
307
|
811 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
812 static char_u *spell_enc __ARGS((void));
|
|
385
|
813 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
814 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
815 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
816 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
483
|
817 static char_u *read_string __ARGS((FILE *fd, int cnt));
|
|
|
818 static int read_region_section __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
819 static int read_charflags_section __ARGS((FILE *fd));
|
|
|
820 static int read_prefcond_section __ARGS((FILE *fd, slang_T *lp));
|
|
625
|
821 static int read_rep_section __ARGS((FILE *fd, garray_T *gap, short *first));
|
|
483
|
822 static int read_sal_section __ARGS((FILE *fd, slang_T *slang));
|
|
625
|
823 static int read_words_section __ARGS((FILE *fd, slang_T *lp, int len));
|
|
|
824 static void count_common_word __ARGS((slang_T *lp, char_u *word, int len, int count));
|
|
|
825 static int score_wordcount_adj __ARGS((slang_T *slang, int score, char_u *word, int split));
|
|
483
|
826 static int read_sofo_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
827 static int read_compound __ARGS((FILE *fd, slang_T *slang, int len));
|
|
495
|
828 static int byte_in_str __ARGS((char_u *str, int byte));
|
|
483
|
829 static int init_syl_tab __ARGS((slang_T *slang));
|
|
|
830 static int count_syllables __ARGS((slang_T *slang, char_u *word));
|
|
381
|
831 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
832 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
833 #ifdef FEAT_MBYTE
|
|
|
834 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
835 #endif
|
|
625
|
836 static int spell_read_tree __ARGS((FILE *fd, char_u **bytsp, idx_T **idxsp, int prefixtree, int prefixcnt));
|
|
|
837 static idx_T read_tree_node __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
838 static void clear_midword __ARGS((buf_T *buf));
|
|
|
839 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
840 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
841 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
842 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
843 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
483
|
844 static void set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
845 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
846 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
847 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
638
|
848 static void spell_find_suggest __ARGS((char_u *badptr, int badlen, suginfo_T *su, int maxcount, int banbadword, int need_cap, int interactive));
|
|
372
|
849 #ifdef FEAT_EVAL
|
|
|
850 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
851 #endif
|
|
|
852 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
625
|
853 static void spell_suggest_intern __ARGS((suginfo_T *su, int interactive));
|
|
|
854 static void suggest_load_files __ARGS((void));
|
|
|
855 static void tree_count_words __ARGS((char_u *byts, idx_T *idxs));
|
|
344
|
856 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
857 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
858 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
859 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
860 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
625
|
861 static void suggest_trie_walk __ARGS((suginfo_T *su, langp_T *lp, char_u *fword, int soundfold));
|
|
|
862 static void go_deeper __ARGS((trystate_T *stack, int depth, int score_add));
|
|
455
|
863 #ifdef FEAT_MBYTE
|
|
|
864 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
865 #endif
|
|
323
|
866 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
867 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
868 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
869 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
625
|
870 static void suggest_try_soundalike_prep __ARGS((void));
|
|
346
|
871 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
625
|
872 static void suggest_try_soundalike_finish __ARGS((void));
|
|
|
873 static void add_sound_suggest __ARGS((suginfo_T *su, char_u *goodword, int score, langp_T *lp));
|
|
|
874 static int soundfold_find __ARGS((slang_T *slang, char_u *word));
|
|
323
|
875 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
876 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
877 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
625
|
878 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus, slang_T *slang, int maxsf));
|
|
|
879 static void check_suggestions __ARGS((suginfo_T *su, garray_T *gap));
|
|
323
|
880 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
324
|
881 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
534
|
882 static void rescore_one __ARGS((suginfo_T *su, suggest_T *stp));
|
|
344
|
883 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
884 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
885 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
886 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
887 #ifdef FEAT_MBYTE
|
|
375
|
888 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
889 #endif
|
|
344
|
890 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
625
|
891 static int spell_edit_score __ARGS((slang_T *slang, char_u *badword, char_u *goodword));
|
|
|
892 static int spell_edit_score_limit __ARGS((slang_T *slang, char_u *badword, char_u *goodword, int limit));
|
|
|
893 #ifdef FEAT_MBYTE
|
|
|
894 static int spell_edit_score_limit_w __ARGS((slang_T *slang, char_u *badword, char_u *goodword, int limit));
|
|
|
895 #endif
|
|
|
896 static void dump_word __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T lnum));
|
|
500
|
897 static linenr_T dump_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
625
|
898 static buf_T *open_spellbuf __ARGS((void));
|
|
|
899 static void close_spellbuf __ARGS((buf_T *buf));
|
|
323
|
900
|
|
324
|
901 /*
|
|
|
902 * Use our own character-case definitions, because the current locale may
|
|
|
903 * differ from what the .spl file uses.
|
|
|
904 * These must not be called with negative number!
|
|
|
905 */
|
|
|
906 #ifndef FEAT_MBYTE
|
|
|
907 /* Non-multi-byte implementation. */
|
|
|
908 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
909 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
910 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
911 #else
|
|
447
|
912 # if defined(HAVE_WCHAR_H)
|
|
|
913 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
914 # endif
|
|
324
|
915 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
916 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
917 * the "w" library function for characters above 255 if available. */
|
|
|
918 # ifdef HAVE_TOWLOWER
|
|
|
919 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
920 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
921 # else
|
|
|
922 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
923 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
924 # endif
|
|
|
925
|
|
|
926 # ifdef HAVE_TOWUPPER
|
|
|
927 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
928 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
929 # else
|
|
|
930 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
931 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
932 # endif
|
|
|
933
|
|
|
934 # ifdef HAVE_ISWUPPER
|
|
|
935 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
936 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
937 # else
|
|
|
938 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
939 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
940 # endif
|
|
|
941 #endif
|
|
|
942
|
|
307
|
943
|
|
|
944 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
945 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
482
|
946 static char *e_afftrailing = N_("Trailing text in %s line %d: %s");
|
|
495
|
947 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
|
948 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
949 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
469
|
950 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
951
|
|
626
|
952 /* Remember what "z?" replaced. */
|
|
|
953 static char_u *repl_from = NULL;
|
|
|
954 static char_u *repl_to = NULL;
|
|
|
955
|
|
236
|
956 /*
|
|
|
957 * Main spell-checking function.
|
|
300
|
958 * "ptr" points to a character that could be the start of a word.
|
|
534
|
959 * "*attrp" is set to the highlight index for a badly spelled word. For a
|
|
|
960 * non-word or when it's OK it remains unchanged.
|
|
236
|
961 * This must only be called when 'spelllang' is not empty.
|
|
323
|
962 *
|
|
385
|
963 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
964 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
965 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
966 * worry.
|
|
323
|
967 *
|
|
236
|
968 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
969 * caller can skip over the word.
|
|
|
970 */
|
|
|
971 int
|
|
625
|
972 spell_check(wp, ptr, attrp, capcol, docount)
|
|
236
|
973 win_T *wp; /* current window */
|
|
|
974 char_u *ptr;
|
|
534
|
975 hlf_T *attrp;
|
|
385
|
976 int *capcol; /* column to check for Capital */
|
|
625
|
977 int docount; /* count good words */
|
|
236
|
978 {
|
|
|
979 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
980 be passed to functions quickly. */
|
|
344
|
981 int nrlen = 0; /* found a number first */
|
|
385
|
982 int c;
|
|
483
|
983 int wrongcaplen = 0;
|
|
500
|
984 int lpi;
|
|
625
|
985 int count_word = docount;
|
|
236
|
986
|
|
307
|
987 /* A word never starts at a space or a control character. Return quickly
|
|
|
988 * then, skipping over the character. */
|
|
|
989 if (*ptr <= ' ')
|
|
|
990 return 1;
|
|
483
|
991 vim_memset(&mi, 0, sizeof(matchinf_T));
|
|
236
|
992
|
|
344
|
993 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
584
|
994 * 0X99FF. But always do check spelling to find "3GPP" and "11
|
|
|
995 * julifeest". */
|
|
300
|
996 if (*ptr >= '0' && *ptr <= '9')
|
|
|
997 {
|
|
316
|
998 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
999 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
1000 else
|
|
344
|
1001 mi.mi_end = skipdigits(ptr);
|
|
584
|
1002 nrlen = mi.mi_end - ptr;
|
|
|
1003 }
|
|
346
|
1004
|
|
|
1005 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
1006 mi.mi_word = ptr;
|
|
584
|
1007 mi.mi_fend = ptr;
|
|
376
|
1008 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
1009 {
|
|
|
1010 do
|
|
|
1011 {
|
|
307
|
1012 mb_ptr_adv(mi.mi_fend);
|
|
376
|
1013 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
1014
|
|
|
1015 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
1016 {
|
|
|
1017 /* Check word starting with capital letter. */
|
|
455
|
1018 c = PTR2CHAR(ptr);
|
|
385
|
1019 if (!SPELL_ISUPPER(c))
|
|
483
|
1020 wrongcaplen = (int)(mi.mi_fend - ptr);
|
|
385
|
1021 }
|
|
|
1022 }
|
|
|
1023 if (capcol != NULL)
|
|
|
1024 *capcol = -1;
|
|
344
|
1025
|
|
|
1026 /* We always use the characters up to the next non-word character,
|
|
|
1027 * also for bad words. */
|
|
|
1028 mi.mi_end = mi.mi_fend;
|
|
|
1029
|
|
|
1030 /* Check caps type later. */
|
|
376
|
1031 mi.mi_buf = wp->w_buffer;
|
|
344
|
1032
|
|
483
|
1033 /* case-fold the word with one non-word character, so that we can check
|
|
|
1034 * for the word end. */
|
|
344
|
1035 if (*mi.mi_fend != NUL)
|
|
|
1036 mb_ptr_adv(mi.mi_fend);
|
|
|
1037
|
|
|
1038 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
1039 MAXWLEN + 1);
|
|
|
1040 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
1041
|
|
|
1042 /* The word is bad unless we recognize it. */
|
|
|
1043 mi.mi_result = SP_BAD;
|
|
492
|
1044 mi.mi_result2 = SP_BAD;
|
|
344
|
1045
|
|
|
1046 /*
|
|
|
1047 * Loop over the languages specified in 'spelllang'.
|
|
625
|
1048 * We check them all, because a word may be matched longer in another
|
|
|
1049 * language.
|
|
344
|
1050 */
|
|
500
|
1051 for (lpi = 0; lpi < wp->w_buffer->b_langp.ga_len; ++lpi)
|
|
|
1052 {
|
|
|
1053 mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, lpi);
|
|
|
1054
|
|
|
1055 /* If reloading fails the language is still in the list but everything
|
|
|
1056 * has been cleared. */
|
|
|
1057 if (mi.mi_lp->lp_slang->sl_fidxs == NULL)
|
|
|
1058 continue;
|
|
|
1059
|
|
344
|
1060 /* Check for a matching word in case-folded words. */
|
|
|
1061 find_word(&mi, FIND_FOLDWORD);
|
|
|
1062
|
|
|
1063 /* Check for a matching word in keep-case words. */
|
|
|
1064 find_word(&mi, FIND_KEEPWORD);
|
|
|
1065
|
|
|
1066 /* Check for matching prefixes. */
|
|
485
|
1067 find_prefix(&mi, FIND_FOLDWORD);
|
|
492
|
1068
|
|
|
1069 /* For a NOBREAK language, may want to use a word without a following
|
|
|
1070 * word as a backup. */
|
|
|
1071 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD
|
|
|
1072 && mi.mi_result2 != SP_BAD)
|
|
|
1073 {
|
|
|
1074 mi.mi_result = mi.mi_result2;
|
|
|
1075 mi.mi_end = mi.mi_end2;
|
|
|
1076 }
|
|
625
|
1077
|
|
|
1078 /* Count the word in the first language where it's found to be OK. */
|
|
|
1079 if (count_word && mi.mi_result == SP_OK)
|
|
|
1080 {
|
|
|
1081 count_common_word(mi.mi_lp->lp_slang, ptr,
|
|
|
1082 (int)(mi.mi_end - ptr), 1);
|
|
|
1083 count_word = FALSE;
|
|
|
1084 }
|
|
344
|
1085 }
|
|
|
1086
|
|
|
1087 if (mi.mi_result != SP_OK)
|
|
|
1088 {
|
|
346
|
1089 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
1090 * rare and local words, e.g., "3GPP". */
|
|
344
|
1091 if (nrlen > 0)
|
|
346
|
1092 {
|
|
|
1093 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
1094 return nrlen;
|
|
|
1095 }
|
|
344
|
1096
|
|
|
1097 /* When we are at a non-word character there is no error, just
|
|
|
1098 * skip over the character (try looking for a word after it). */
|
|
390
|
1099 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
1100 {
|
|
385
|
1101 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
1102 {
|
|
|
1103 regmatch_T regmatch;
|
|
|
1104
|
|
|
1105 /* Check for end of sentence. */
|
|
|
1106 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
1107 regmatch.rm_ic = FALSE;
|
|
|
1108 if (vim_regexec(®match, ptr, 0))
|
|
|
1109 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
1110 }
|
|
|
1111
|
|
344
|
1112 #ifdef FEAT_MBYTE
|
|
|
1113 if (has_mbyte)
|
|
474
|
1114 return (*mb_ptr2len)(ptr);
|
|
344
|
1115 #endif
|
|
|
1116 return 1;
|
|
300
|
1117 }
|
|
483
|
1118 else if (mi.mi_end == ptr)
|
|
|
1119 /* Always include at least one character. Required for when there
|
|
|
1120 * is a mixup in "midword". */
|
|
|
1121 mb_ptr_adv(mi.mi_end);
|
|
492
|
1122 else if (mi.mi_result == SP_BAD
|
|
|
1123 && LANGP_ENTRY(wp->w_buffer->b_langp, 0)->lp_slang->sl_nobreak)
|
|
|
1124 {
|
|
|
1125 char_u *p, *fp;
|
|
|
1126 int save_result = mi.mi_result;
|
|
|
1127
|
|
|
1128 /* First language in 'spelllang' is NOBREAK. Find first position
|
|
|
1129 * at which any word would be valid. */
|
|
|
1130 mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
500
|
1131 if (mi.mi_lp->lp_slang->sl_fidxs != NULL)
|
|
|
1132 {
|
|
|
1133 p = mi.mi_word;
|
|
|
1134 fp = mi.mi_fword;
|
|
|
1135 for (;;)
|
|
|
1136 {
|
|
|
1137 mb_ptr_adv(p);
|
|
|
1138 mb_ptr_adv(fp);
|
|
|
1139 if (p >= mi.mi_end)
|
|
|
1140 break;
|
|
|
1141 mi.mi_compoff = fp - mi.mi_fword;
|
|
|
1142 find_word(&mi, FIND_COMPOUND);
|
|
|
1143 if (mi.mi_result != SP_BAD)
|
|
|
1144 {
|
|
|
1145 mi.mi_end = p;
|
|
|
1146 break;
|
|
|
1147 }
|
|
|
1148 }
|
|
|
1149 mi.mi_result = save_result;
|
|
|
1150 }
|
|
492
|
1151 }
|
|
243
|
1152
|
|
344
|
1153 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
534
|
1154 *attrp = HLF_SPB;
|
|
344
|
1155 else if (mi.mi_result == SP_RARE)
|
|
534
|
1156 *attrp = HLF_SPR;
|
|
344
|
1157 else
|
|
534
|
1158 *attrp = HLF_SPL;
|
|
243
|
1159 }
|
|
|
1160
|
|
483
|
1161 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE))
|
|
|
1162 {
|
|
|
1163 /* Report SpellCap only when the word isn't badly spelled. */
|
|
534
|
1164 *attrp = HLF_SPC;
|
|
483
|
1165 return wrongcaplen;
|
|
|
1166 }
|
|
|
1167
|
|
300
|
1168 return (int)(mi.mi_end - ptr);
|
|
236
|
1169 }
|
|
|
1170
|
|
|
1171 /*
|
|
300
|
1172 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
1173 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
1174 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
1175 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
1176 * tree.
|
|
300
|
1177 *
|
|
|
1178 * For a match mip->mi_result is updated.
|
|
243
|
1179 */
|
|
|
1180 static void
|
|
339
|
1181 find_word(mip, mode)
|
|
243
|
1182 matchinf_T *mip;
|
|
339
|
1183 int mode;
|
|
243
|
1184 {
|
|
324
|
1185 idx_T arridx = 0;
|
|
300
|
1186 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
1187 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
1188 int endidxcnt = 0;
|
|
|
1189 int len;
|
|
|
1190 int wlen = 0;
|
|
|
1191 int flen;
|
|
|
1192 int c;
|
|
|
1193 char_u *ptr;
|
|
324
|
1194 idx_T lo, hi, m;
|
|
243
|
1195 #ifdef FEAT_MBYTE
|
|
300
|
1196 char_u *s;
|
|
490
|
1197 #endif
|
|
339
|
1198 char_u *p;
|
|
307
|
1199 int res = SP_BAD;
|
|
300
|
1200 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1201 unsigned flags;
|
|
|
1202 char_u *byts;
|
|
324
|
1203 idx_T *idxs;
|
|
481
|
1204 int word_ends;
|
|
485
|
1205 int prefix_found;
|
|
492
|
1206 int nobreak_result;
|
|
481
|
1207
|
|
|
1208 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND)
|
|
236
|
1209 {
|
|
300
|
1210 /* Check for word with matching case in keep-case tree. */
|
|
|
1211 ptr = mip->mi_word;
|
|
|
1212 flen = 9999; /* no case folding, always enough bytes */
|
|
|
1213 byts = slang->sl_kbyts;
|
|
|
1214 idxs = slang->sl_kidxs;
|
|
481
|
1215
|
|
|
1216 if (mode == FIND_KEEPCOMPOUND)
|
|
|
1217 /* Skip over the previously found word(s). */
|
|
|
1218 wlen += mip->mi_compoff;
|
|
236
|
1219 }
|
|
|
1220 else
|
|
|
1221 {
|
|
300
|
1222 /* Check for case-folded in case-folded tree. */
|
|
|
1223 ptr = mip->mi_fword;
|
|
|
1224 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1225 byts = slang->sl_fbyts;
|
|
|
1226 idxs = slang->sl_fidxs;
|
|
339
|
1227
|
|
|
1228 if (mode == FIND_PREFIX)
|
|
|
1229 {
|
|
|
1230 /* Skip over the prefix. */
|
|
|
1231 wlen = mip->mi_prefixlen;
|
|
|
1232 flen -= mip->mi_prefixlen;
|
|
|
1233 }
|
|
481
|
1234 else if (mode == FIND_COMPOUND)
|
|
|
1235 {
|
|
|
1236 /* Skip over the previously found word(s). */
|
|
|
1237 wlen = mip->mi_compoff;
|
|
|
1238 flen -= mip->mi_compoff;
|
|
|
1239 }
|
|
|
1240
|
|
243
|
1241 }
|
|
|
1242
|
|
300
|
1243 if (byts == NULL)
|
|
|
1244 return; /* array is empty */
|
|
236
|
1245
|
|
|
1246 /*
|
|
307
|
1247 * Repeat advancing in the tree until:
|
|
|
1248 * - there is a byte that doesn't match,
|
|
|
1249 * - we reach the end of the tree,
|
|
|
1250 * - or we reach the end of the line.
|
|
236
|
1251 */
|
|
300
|
1252 for (;;)
|
|
236
|
1253 {
|
|
346
|
1254 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
1255 flen = fold_more(mip);
|
|
300
|
1256
|
|
|
1257 len = byts[arridx++];
|
|
|
1258
|
|
|
1259 /* If the first possible byte is a zero the word could end here.
|
|
|
1260 * Remember this index, we first check for the longest word. */
|
|
|
1261 if (byts[arridx] == 0)
|
|
|
1262 {
|
|
307
|
1263 if (endidxcnt == MAXWLEN)
|
|
|
1264 {
|
|
|
1265 /* Must be a corrupted spell file. */
|
|
|
1266 EMSG(_(e_format));
|
|
|
1267 return;
|
|
|
1268 }
|
|
300
|
1269 endlen[endidxcnt] = wlen;
|
|
|
1270 endidx[endidxcnt++] = arridx++;
|
|
|
1271 --len;
|
|
|
1272
|
|
|
1273 /* Skip over the zeros, there can be several flag/region
|
|
|
1274 * combinations. */
|
|
|
1275 while (len > 0 && byts[arridx] == 0)
|
|
|
1276 {
|
|
|
1277 ++arridx;
|
|
|
1278 --len;
|
|
|
1279 }
|
|
|
1280 if (len == 0)
|
|
|
1281 break; /* no children, word must end here */
|
|
|
1282 }
|
|
|
1283
|
|
|
1284 /* Stop looking at end of the line. */
|
|
|
1285 if (ptr[wlen] == NUL)
|
|
|
1286 break;
|
|
|
1287
|
|
|
1288 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1289 c = ptr[wlen];
|
|
346
|
1290 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1291 c = ' ';
|
|
300
|
1292 lo = arridx;
|
|
|
1293 hi = arridx + len - 1;
|
|
|
1294 while (lo < hi)
|
|
|
1295 {
|
|
|
1296 m = (lo + hi) / 2;
|
|
|
1297 if (byts[m] > c)
|
|
|
1298 hi = m - 1;
|
|
|
1299 else if (byts[m] < c)
|
|
|
1300 lo = m + 1;
|
|
|
1301 else
|
|
|
1302 {
|
|
|
1303 lo = hi = m;
|
|
|
1304 break;
|
|
236
|
1305 }
|
|
|
1306 }
|
|
300
|
1307
|
|
|
1308 /* Stop if there is no matching byte. */
|
|
|
1309 if (hi < lo || byts[lo] != c)
|
|
|
1310 break;
|
|
|
1311
|
|
|
1312 /* Continue at the child (if there is one). */
|
|
|
1313 arridx = idxs[lo];
|
|
|
1314 ++wlen;
|
|
|
1315 --flen;
|
|
346
|
1316
|
|
|
1317 /* One space in the good word may stand for several spaces in the
|
|
|
1318 * checked word. */
|
|
|
1319 if (c == ' ')
|
|
|
1320 {
|
|
|
1321 for (;;)
|
|
|
1322 {
|
|
|
1323 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1324 flen = fold_more(mip);
|
|
|
1325 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1326 break;
|
|
|
1327 ++wlen;
|
|
|
1328 --flen;
|
|
|
1329 }
|
|
|
1330 }
|
|
236
|
1331 }
|
|
|
1332
|
|
300
|
1333 /*
|
|
|
1334 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1335 * first.
|
|
|
1336 */
|
|
|
1337 while (endidxcnt > 0)
|
|
|
1338 {
|
|
|
1339 --endidxcnt;
|
|
|
1340 arridx = endidx[endidxcnt];
|
|
|
1341 wlen = endlen[endidxcnt];
|
|
236
|
1342
|
|
300
|
1343 #ifdef FEAT_MBYTE
|
|
|
1344 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1345 continue; /* not at first byte of character */
|
|
|
1346 #endif
|
|
376
|
1347 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
481
|
1348 {
|
|
492
|
1349 if (slang->sl_compprog == NULL && !slang->sl_nobreak)
|
|
481
|
1350 continue; /* next char is a word character */
|
|
|
1351 word_ends = FALSE;
|
|
|
1352 }
|
|
|
1353 else
|
|
|
1354 word_ends = TRUE;
|
|
485
|
1355 /* The prefix flag is before compound flags. Once a valid prefix flag
|
|
|
1356 * has been found we try compound flags. */
|
|
|
1357 prefix_found = FALSE;
|
|
300
|
1358
|
|
|
1359 #ifdef FEAT_MBYTE
|
|
339
|
1360 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1361 {
|
|
|
1362 /* Compute byte length in original word, length may change
|
|
339
|
1363 * when folding case. This can be slow, take a shortcut when the
|
|
|
1364 * case-folded word is equal to the keep-case word. */
|
|
300
|
1365 p = mip->mi_word;
|
|
339
|
1366 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1367 {
|
|
|
1368 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1369 mb_ptr_adv(p);
|
|
|
1370 wlen = p - mip->mi_word;
|
|
|
1371 }
|
|
300
|
1372 }
|
|
|
1373 #endif
|
|
236
|
1374
|
|
339
|
1375 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1376 * prefix ID.
|
|
|
1377 * Repeat this if there are more flags/region alternatives until there
|
|
|
1378 * is a match. */
|
|
|
1379 res = SP_BAD;
|
|
|
1380 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1381 --len, ++arridx)
|
|
300
|
1382 {
|
|
|
1383 flags = idxs[arridx];
|
|
324
|
1384
|
|
339
|
1385 /* For the fold-case tree check that the case of the checked word
|
|
|
1386 * matches with what the word in the tree requires.
|
|
|
1387 * For keep-case tree the case is always right. For prefixes we
|
|
|
1388 * don't bother to check. */
|
|
|
1389 if (mode == FIND_FOLDWORD)
|
|
300
|
1390 {
|
|
|
1391 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1392 {
|
|
323
|
1393 /* mi_capflags was set for a different word length, need
|
|
|
1394 * to do it again. */
|
|
300
|
1395 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1396 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1397 }
|
|
|
1398
|
|
346
|
1399 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1400 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1401 continue;
|
|
300
|
1402 }
|
|
236
|
1403
|
|
339
|
1404 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1405 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1406 * mip->mi_prefarridx that find_prefix() filled. */
|
|
485
|
1407 else if (mode == FIND_PREFIX && !prefix_found)
|
|
481
|
1408 {
|
|
366
|
1409 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1410 flags,
|
|
455
|
1411 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1412 FALSE);
|
|
366
|
1413 if (c == 0)
|
|
339
|
1414 continue;
|
|
366
|
1415
|
|
|
1416 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1417 if (c & WF_RAREPFX)
|
|
|
1418 flags |= WF_RARE;
|
|
485
|
1419 prefix_found = TRUE;
|
|
339
|
1420 }
|
|
|
1421
|
|
492
|
1422 if (slang->sl_nobreak)
|
|
|
1423 {
|
|
|
1424 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND)
|
|
|
1425 && (flags & WF_BANNED) == 0)
|
|
|
1426 {
|
|
|
1427 /* NOBREAK: found a valid following word. That's all we
|
|
|
1428 * need to know, so return. */
|
|
|
1429 mip->mi_result = SP_OK;
|
|
|
1430 break;
|
|
|
1431 }
|
|
|
1432 }
|
|
|
1433
|
|
|
1434 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND
|
|
|
1435 || !word_ends))
|
|
481
|
1436 {
|
|
483
|
1437 /* If there is no flag or the word is shorter than
|
|
|
1438 * COMPOUNDMIN reject it quickly.
|
|
|
1439 * Makes you wonder why someone puts a compound flag on a word
|
|
481
|
1440 * that's too short... Myspell compatibility requires this
|
|
|
1441 * anyway. */
|
|
490
|
1442 if (((unsigned)flags >> 24) == 0
|
|
|
1443 || wlen - mip->mi_compoff < slang->sl_compminlen)
|
|
483
|
1444 continue;
|
|
500
|
1445 #ifdef FEAT_MBYTE
|
|
|
1446 /* For multi-byte chars check character length against
|
|
|
1447 * COMPOUNDMIN. */
|
|
|
1448 if (has_mbyte
|
|
501
|
1449 && slang->sl_compminlen > 0
|
|
500
|
1450 && mb_charlen_len(mip->mi_word + mip->mi_compoff,
|
|
|
1451 wlen - mip->mi_compoff) < slang->sl_compminlen)
|
|
|
1452 continue;
|
|
|
1453 #endif
|
|
483
|
1454
|
|
490
|
1455 /* Limit the number of compound words to COMPOUNDMAX if no
|
|
|
1456 * maximum for syllables is specified. */
|
|
|
1457 if (!word_ends && mip->mi_complen + 2 > slang->sl_compmax
|
|
|
1458 && slang->sl_compsylmax == MAXWLEN)
|
|
483
|
1459 continue;
|
|
|
1460
|
|
485
|
1461 /* Quickly check if compounding is possible with this flag. */
|
|
495
|
1462 if (!byte_in_str(mip->mi_complen == 0
|
|
485
|
1463 ? slang->sl_compstartflags
|
|
|
1464 : slang->sl_compallflags,
|
|
495
|
1465 ((unsigned)flags >> 24)))
|
|
481
|
1466 continue;
|
|
|
1467
|
|
490
|
1468 if (mode == FIND_COMPOUND)
|
|
|
1469 {
|
|
|
1470 int capflags;
|
|
|
1471
|
|
|
1472 /* Need to check the caps type of the appended compound
|
|
|
1473 * word. */
|
|
|
1474 #ifdef FEAT_MBYTE
|
|
|
1475 if (has_mbyte && STRNCMP(ptr, mip->mi_word,
|
|
|
1476 mip->mi_compoff) != 0)
|
|
|
1477 {
|
|
|
1478 /* case folding may have changed the length */
|
|
|
1479 p = mip->mi_word;
|
|
|
1480 for (s = ptr; s < ptr + mip->mi_compoff; mb_ptr_adv(s))
|
|
|
1481 mb_ptr_adv(p);
|
|
|
1482 }
|
|
|
1483 else
|
|
|
1484 #endif
|
|
|
1485 p = mip->mi_word + mip->mi_compoff;
|
|
|
1486 capflags = captype(p, mip->mi_word + wlen);
|
|
|
1487 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP
|
|
|
1488 && (flags & WF_FIXCAP) != 0))
|
|
|
1489 continue;
|
|
|
1490
|
|
|
1491 if (capflags != WF_ALLCAP)
|
|
|
1492 {
|
|
|
1493 /* When the character before the word is a word
|
|
|
1494 * character we do not accept a Onecap word. We do
|
|
|
1495 * accept a no-caps word, even when the dictionary
|
|
|
1496 * word specifies ONECAP. */
|
|
|
1497 mb_ptr_back(mip->mi_word, p);
|
|
|
1498 if (spell_iswordp_nmw(p)
|
|
|
1499 ? capflags == WF_ONECAP
|
|
|
1500 : (flags & WF_ONECAP) != 0
|
|
|
1501 && capflags != WF_ONECAP)
|
|
|
1502 continue;
|
|
|
1503 }
|
|
|
1504 }
|
|
|
1505
|
|
483
|
1506 /* If the word ends the sequence of compound flags of the
|
|
|
1507 * words must match with one of the COMPOUNDFLAGS items and
|
|
|
1508 * the number of syllables must not be too large. */
|
|
|
1509 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24);
|
|
|
1510 mip->mi_compflags[mip->mi_complen + 1] = NUL;
|
|
|
1511 if (word_ends)
|
|
|
1512 {
|
|
|
1513 char_u fword[MAXWLEN];
|
|
|
1514
|
|
|
1515 if (slang->sl_compsylmax < MAXWLEN)
|
|
|
1516 {
|
|
|
1517 /* "fword" is only needed for checking syllables. */
|
|
|
1518 if (ptr == mip->mi_word)
|
|
|
1519 (void)spell_casefold(ptr, wlen, fword, MAXWLEN);
|
|
|
1520 else
|
|
|
1521 vim_strncpy(fword, ptr, endlen[endidxcnt]);
|
|
|
1522 }
|
|
|
1523 if (!can_compound(slang, fword, mip->mi_compflags))
|
|
|
1524 continue;
|
|
|
1525 }
|
|
481
|
1526 }
|
|
|
1527
|
|
500
|
1528 /* Check NEEDCOMPOUND: can't use word without compounding. */
|
|
|
1529 else if (flags & WF_NEEDCOMP)
|
|
|
1530 continue;
|
|
|
1531
|
|
492
|
1532 nobreak_result = SP_OK;
|
|
|
1533
|
|
481
|
1534 if (!word_ends)
|
|
|
1535 {
|
|
492
|
1536 int save_result = mip->mi_result;
|
|
|
1537 char_u *save_end = mip->mi_end;
|
|
501
|
1538 langp_T *save_lp = mip->mi_lp;
|
|
|
1539 int lpi;
|
|
492
|
1540
|
|
|
1541 /* Check that a valid word follows. If there is one and we
|
|
|
1542 * are compounding, it will set "mi_result", thus we are
|
|
|
1543 * always finished here. For NOBREAK we only check that a
|
|
|
1544 * valid word follows.
|
|
481
|
1545 * Recursive! */
|
|
492
|
1546 if (slang->sl_nobreak)
|
|
|
1547 mip->mi_result = SP_BAD;
|
|
481
|
1548
|
|
|
1549 /* Find following word in case-folded tree. */
|
|
|
1550 mip->mi_compoff = endlen[endidxcnt];
|
|
|
1551 #ifdef FEAT_MBYTE
|
|
|
1552 if (has_mbyte && mode == FIND_KEEPWORD)
|
|
|
1553 {
|
|
|
1554 /* Compute byte length in case-folded word from "wlen":
|
|
|
1555 * byte length in keep-case word. Length may change when
|
|
|
1556 * folding case. This can be slow, take a shortcut when
|
|
|
1557 * the case-folded word is equal to the keep-case word. */
|
|
|
1558 p = mip->mi_fword;
|
|
|
1559 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1560 {
|
|
|
1561 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1562 mb_ptr_adv(p);
|
|
|
1563 mip->mi_compoff = p - mip->mi_fword;
|
|
|
1564 }
|
|
|
1565 }
|
|
|
1566 #endif
|
|
485
|
1567 c = mip->mi_compoff;
|
|
483
|
1568 ++mip->mi_complen;
|
|
501
|
1569
|
|
|
1570 /* For NOBREAK we need to try all NOBREAK languages, at least
|
|
|
1571 * to find the ".add" file(s). */
|
|
|
1572 for (lpi = 0; lpi < mip->mi_buf->b_langp.ga_len; ++lpi)
|
|
|
1573 {
|
|
|
1574 if (slang->sl_nobreak)
|
|
|
1575 {
|
|
|
1576 mip->mi_lp = LANGP_ENTRY(mip->mi_buf->b_langp, lpi);
|
|
|
1577 if (mip->mi_lp->lp_slang->sl_fidxs == NULL
|
|
|
1578 || !mip->mi_lp->lp_slang->sl_nobreak)
|
|
|
1579 continue;
|
|
|
1580 }
|
|
|
1581
|
|
|
1582 find_word(mip, FIND_COMPOUND);
|
|
|
1583
|
|
|
1584 /* When NOBREAK any word that matches is OK. Otherwise we
|
|
|
1585 * need to find the longest match, thus try with keep-case
|
|
|
1586 * and prefix too. */
|
|
492
|
1587 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1588 {
|
|
501
|
1589 /* Find following word in keep-case tree. */
|
|
|
1590 mip->mi_compoff = wlen;
|
|
|
1591 find_word(mip, FIND_KEEPCOMPOUND);
|
|
|
1592
|
|
|
1593 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1594 {
|
|
|
1595 /* Check for following word with prefix. */
|
|
|
1596 mip->mi_compoff = c;
|
|
|
1597 find_prefix(mip, FIND_COMPOUND);
|
|
|
1598 }
|
|
492
|
1599 }
|
|
501
|
1600
|
|
|
1601 if (!slang->sl_nobreak)
|
|
|
1602 break;
|
|
492
|
1603 }
|
|
483
|
1604 --mip->mi_complen;
|
|
501
|
1605 mip->mi_lp = save_lp;
|
|
485
|
1606
|
|
492
|
1607 if (slang->sl_nobreak)
|
|
|
1608 {
|
|
|
1609 nobreak_result = mip->mi_result;
|
|
|
1610 mip->mi_result = save_result;
|
|
|
1611 mip->mi_end = save_end;
|
|
|
1612 }
|
|
|
1613 else
|
|
|
1614 {
|
|
|
1615 if (mip->mi_result == SP_OK)
|
|
|
1616 break;
|
|
|
1617 continue;
|
|
|
1618 }
|
|
481
|
1619 }
|
|
|
1620
|
|
339
|
1621 if (flags & WF_BANNED)
|
|
|
1622 res = SP_BANNED;
|
|
|
1623 else if (flags & WF_REGION)
|
|
|
1624 {
|
|
|
1625 /* Check region. */
|
|
390
|
1626 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1627 res = SP_OK;
|
|
339
|
1628 else
|
|
|
1629 res = SP_LOCAL;
|
|
300
|
1630 }
|
|
339
|
1631 else if (flags & WF_RARE)
|
|
|
1632 res = SP_RARE;
|
|
307
|
1633 else
|
|
339
|
1634 res = SP_OK;
|
|
|
1635
|
|
492
|
1636 /* Always use the longest match and the best result. For NOBREAK
|
|
|
1637 * we separately keep the longest match without a following good
|
|
|
1638 * word as a fall-back. */
|
|
|
1639 if (nobreak_result == SP_BAD)
|
|
|
1640 {
|
|
|
1641 if (mip->mi_result2 > res)
|
|
|
1642 {
|
|
|
1643 mip->mi_result2 = res;
|
|
|
1644 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1645 }
|
|
|
1646 else if (mip->mi_result2 == res
|
|
|
1647 && mip->mi_end2 < mip->mi_word + wlen)
|
|
|
1648 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1649 }
|
|
|
1650 else if (mip->mi_result > res)
|
|
339
|
1651 {
|
|
|
1652 mip->mi_result = res;
|
|
|
1653 mip->mi_end = mip->mi_word + wlen;
|
|
|
1654 }
|
|
351
|
1655 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1656 mip->mi_end = mip->mi_word + wlen;
|
|
|
1657
|
|
492
|
1658 if (mip->mi_result == SP_OK)
|
|
339
|
1659 break;
|
|
300
|
1660 }
|
|
|
1661
|
|
492
|
1662 if (mip->mi_result == SP_OK)
|
|
300
|
1663 break;
|
|
|
1664 }
|
|
236
|
1665 }
|
|
|
1666
|
|
323
|
1667 /*
|
|
626
|
1668 * Return TRUE if "flags" is a valid sequence of compound flags and "word"
|
|
|
1669 * does not have too many syllables.
|
|
482
|
1670 */
|
|
|
1671 static int
|
|
483
|
1672 can_compound(slang, word, flags)
|
|
482
|
1673 slang_T *slang;
|
|
483
|
1674 char_u *word;
|
|
|
1675 char_u *flags;
|
|
|
1676 {
|
|
|
1677 regmatch_T regmatch;
|
|
495
|
1678 #ifdef FEAT_MBYTE
|
|
|
1679 char_u uflags[MAXWLEN * 2];
|
|
|
1680 int i;
|
|
|
1681 #endif
|
|
|
1682 char_u *p;
|
|
483
|
1683
|
|
|
1684 if (slang->sl_compprog == NULL)
|
|
|
1685 return FALSE;
|
|
495
|
1686 #ifdef FEAT_MBYTE
|
|
|
1687 if (enc_utf8)
|
|
|
1688 {
|
|
|
1689 /* Need to convert the single byte flags to utf8 characters. */
|
|
|
1690 p = uflags;
|
|
|
1691 for (i = 0; flags[i] != NUL; ++i)
|
|
|
1692 p += mb_char2bytes(flags[i], p);
|
|
|
1693 *p = NUL;
|
|
|
1694 p = uflags;
|
|
|
1695 }
|
|
|
1696 else
|
|
|
1697 #endif
|
|
|
1698 p = flags;
|
|
483
|
1699 regmatch.regprog = slang->sl_compprog;
|
|
|
1700 regmatch.rm_ic = FALSE;
|
|
495
|
1701 if (!vim_regexec(®match, p, 0))
|
|
483
|
1702 return FALSE;
|
|
|
1703
|
|
490
|
1704 /* Count the number of syllables. This may be slow, do it last. If there
|
|
|
1705 * are too many syllables AND the number of compound words is above
|
|
|
1706 * COMPOUNDMAX then compounding is not allowed. */
|
|
483
|
1707 if (slang->sl_compsylmax < MAXWLEN
|
|
|
1708 && count_syllables(slang, word) > slang->sl_compsylmax)
|
|
495
|
1709 return (int)STRLEN(flags) < slang->sl_compmax;
|
|
483
|
1710 return TRUE;
|
|
482
|
1711 }
|
|
|
1712
|
|
|
1713 /*
|
|
390
|
1714 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1715 * ID in "flags" for the word "word".
|
|
366
|
1716 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1717 */
|
|
|
1718 static int
|
|
455
|
1719 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1720 int totprefcnt; /* nr of prefix IDs */
|
|
|
1721 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1722 int flags;
|
|
351
|
1723 char_u *word;
|
|
|
1724 slang_T *slang;
|
|
455
|
1725 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1726 {
|
|
|
1727 int prefcnt;
|
|
|
1728 int pidx;
|
|
|
1729 regprog_T *rp;
|
|
|
1730 regmatch_T regmatch;
|
|
390
|
1731 int prefid;
|
|
|
1732
|
|
|
1733 prefid = (unsigned)flags >> 24;
|
|
351
|
1734 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1735 {
|
|
|
1736 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1737
|
|
|
1738 /* Check the prefix ID. */
|
|
|
1739 if (prefid != (pidx & 0xff))
|
|
|
1740 continue;
|
|
|
1741
|
|
390
|
1742 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1743 * suffix. */
|
|
|
1744 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1745 continue;
|
|
|
1746
|
|
351
|
1747 /* Check the condition, if there is one. The condition index is
|
|
366
|
1748 * stored in the two bytes above the prefix ID byte. */
|
|
|
1749 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1750 if (rp != NULL)
|
|
|
1751 {
|
|
|
1752 regmatch.regprog = rp;
|
|
|
1753 regmatch.rm_ic = FALSE;
|
|
|
1754 if (!vim_regexec(®match, word, 0))
|
|
|
1755 continue;
|
|
|
1756 }
|
|
455
|
1757 else if (cond_req)
|
|
|
1758 continue;
|
|
|
1759
|
|
|
1760 /* It's a match! Return the WF_ flags. */
|
|
366
|
1761 return pidx;
|
|
|
1762 }
|
|
|
1763 return 0;
|
|
351
|
1764 }
|
|
|
1765
|
|
|
1766 /*
|
|
339
|
1767 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1768 * If it does, then check the following word.
|
|
|
1769 *
|
|
485
|
1770 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a
|
|
|
1771 * prefix in a compound word.
|
|
|
1772 *
|
|
339
|
1773 * For a match mip->mi_result is updated.
|
|
|
1774 */
|
|
|
1775 static void
|
|
485
|
1776 find_prefix(mip, mode)
|
|
339
|
1777 matchinf_T *mip;
|
|
485
|
1778 int mode;
|
|
339
|
1779 {
|
|
|
1780 idx_T arridx = 0;
|
|
|
1781 int len;
|
|
|
1782 int wlen = 0;
|
|
|
1783 int flen;
|
|
|
1784 int c;
|
|
|
1785 char_u *ptr;
|
|
|
1786 idx_T lo, hi, m;
|
|
|
1787 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1788 char_u *byts;
|
|
|
1789 idx_T *idxs;
|
|
|
1790
|
|
375
|
1791 byts = slang->sl_pbyts;
|
|
|
1792 if (byts == NULL)
|
|
|
1793 return; /* array is empty */
|
|
|
1794
|
|
339
|
1795 /* We use the case-folded word here, since prefixes are always
|
|
|
1796 * case-folded. */
|
|
|
1797 ptr = mip->mi_fword;
|
|
|
1798 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
485
|
1799 if (mode == FIND_COMPOUND)
|
|
|
1800 {
|
|
|
1801 /* Skip over the previously found word(s). */
|
|
|
1802 ptr += mip->mi_compoff;
|
|
|
1803 flen -= mip->mi_compoff;
|
|
|
1804 }
|
|
339
|
1805 idxs = slang->sl_pidxs;
|
|
|
1806
|
|
|
1807 /*
|
|
|
1808 * Repeat advancing in the tree until:
|
|
|
1809 * - there is a byte that doesn't match,
|
|
|
1810 * - we reach the end of the tree,
|
|
|
1811 * - or we reach the end of the line.
|
|
|
1812 */
|
|
|
1813 for (;;)
|
|
|
1814 {
|
|
|
1815 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1816 flen = fold_more(mip);
|
|
|
1817
|
|
|
1818 len = byts[arridx++];
|
|
|
1819
|
|
|
1820 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1821 * Check if the following word matches and supports the prefix. */
|
|
|
1822 if (byts[arridx] == 0)
|
|
|
1823 {
|
|
|
1824 /* There can be several prefixes with different conditions. We
|
|
|
1825 * try them all, since we don't know which one will give the
|
|
|
1826 * longest match. The word is the same each time, pass the list
|
|
|
1827 * of possible prefixes to find_word(). */
|
|
|
1828 mip->mi_prefarridx = arridx;
|
|
|
1829 mip->mi_prefcnt = len;
|
|
|
1830 while (len > 0 && byts[arridx] == 0)
|
|
|
1831 {
|
|
|
1832 ++arridx;
|
|
|
1833 --len;
|
|
|
1834 }
|
|
|
1835 mip->mi_prefcnt -= len;
|
|
|
1836
|
|
|
1837 /* Find the word that comes after the prefix. */
|
|
|
1838 mip->mi_prefixlen = wlen;
|
|
485
|
1839 if (mode == FIND_COMPOUND)
|
|
|
1840 /* Skip over the previously found word(s). */
|
|
|
1841 mip->mi_prefixlen += mip->mi_compoff;
|
|
|
1842
|
|
455
|
1843 #ifdef FEAT_MBYTE
|
|
|
1844 if (has_mbyte)
|
|
|
1845 {
|
|
|
1846 /* Case-folded length may differ from original length. */
|
|
485
|
1847 mip->mi_cprefixlen = nofold_len(mip->mi_fword,
|
|
|
1848 mip->mi_prefixlen, mip->mi_word);
|
|
455
|
1849 }
|
|
|
1850 else
|
|
485
|
1851 mip->mi_cprefixlen = mip->mi_prefixlen;
|
|
455
|
1852 #endif
|
|
339
|
1853 find_word(mip, FIND_PREFIX);
|
|
|
1854
|
|
|
1855
|
|
|
1856 if (len == 0)
|
|
|
1857 break; /* no children, word must end here */
|
|
|
1858 }
|
|
|
1859
|
|
|
1860 /* Stop looking at end of the line. */
|
|
|
1861 if (ptr[wlen] == NUL)
|
|
|
1862 break;
|
|
|
1863
|
|
|
1864 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1865 c = ptr[wlen];
|
|
|
1866 lo = arridx;
|
|
|
1867 hi = arridx + len - 1;
|
|
|
1868 while (lo < hi)
|
|
|
1869 {
|
|
|
1870 m = (lo + hi) / 2;
|
|
|
1871 if (byts[m] > c)
|
|
|
1872 hi = m - 1;
|
|
|
1873 else if (byts[m] < c)
|
|
|
1874 lo = m + 1;
|
|
|
1875 else
|
|
|
1876 {
|
|
|
1877 lo = hi = m;
|
|
|
1878 break;
|
|
|
1879 }
|
|
|
1880 }
|
|
|
1881
|
|
|
1882 /* Stop if there is no matching byte. */
|
|
|
1883 if (hi < lo || byts[lo] != c)
|
|
|
1884 break;
|
|
|
1885
|
|
|
1886 /* Continue at the child (if there is one). */
|
|
|
1887 arridx = idxs[lo];
|
|
|
1888 ++wlen;
|
|
|
1889 --flen;
|
|
|
1890 }
|
|
|
1891 }
|
|
|
1892
|
|
|
1893 /*
|
|
|
1894 * Need to fold at least one more character. Do until next non-word character
|
|
626
|
1895 * for efficiency. Include the non-word character too.
|
|
339
|
1896 * Return the length of the folded chars in bytes.
|
|
|
1897 */
|
|
|
1898 static int
|
|
|
1899 fold_more(mip)
|
|
|
1900 matchinf_T *mip;
|
|
|
1901 {
|
|
|
1902 int flen;
|
|
|
1903 char_u *p;
|
|
|
1904
|
|
|
1905 p = mip->mi_fend;
|
|
|
1906 do
|
|
|
1907 {
|
|
|
1908 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1909 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1910
|
|
626
|
1911 /* Include the non-word character so that we can check for the word end. */
|
|
339
|
1912 if (*mip->mi_fend != NUL)
|
|
|
1913 mb_ptr_adv(mip->mi_fend);
|
|
|
1914
|
|
|
1915 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1916 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1917 MAXWLEN - mip->mi_fwordlen);
|
|
|
1918 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1919 mip->mi_fwordlen += flen;
|
|
|
1920 return flen;
|
|
|
1921 }
|
|
|
1922
|
|
|
1923 /*
|
|
323
|
1924 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1925 * case.
|
|
|
1926 */
|
|
|
1927 static int
|
|
389
|
1928 spell_valid_case(wordflags, treeflags)
|
|
|
1929 int wordflags; /* flags for the checked word. */
|
|
323
|
1930 int treeflags; /* flags for the word in the spell tree */
|
|
|
1931 {
|
|
389
|
1932 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1933 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1934 && ((treeflags & WF_ONECAP) == 0
|
|
|
1935 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1936 }
|
|
|
1937
|
|
351
|
1938 /*
|
|
|
1939 * Return TRUE if spell checking is not enabled.
|
|
|
1940 */
|
|
|
1941 static int
|
|
498
|
1942 no_spell_checking(wp)
|
|
|
1943 win_T *wp;
|
|
|
1944 {
|
|
|
1945 if (!wp->w_p_spell || *wp->w_buffer->b_p_spl == NUL)
|
|
351
|
1946 {
|
|
|
1947 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1948 return TRUE;
|
|
|
1949 }
|
|
|
1950 return FALSE;
|
|
|
1951 }
|
|
300
|
1952
|
|
236
|
1953 /*
|
|
|
1954 * Move to next spell error.
|
|
500
|
1955 * "curline" is FALSE for "[s", "]s", "[S" and "]S".
|
|
|
1956 * "curline" is TRUE to find word under/after cursor in the same line.
|
|
483
|
1957 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move
|
|
|
1958 * to after badly spelled word before the cursor.
|
|
495
|
1959 * Return 0 if not found, length of the badly spelled word otherwise.
|
|
236
|
1960 */
|
|
|
1961 int
|
|
498
|
1962 spell_move_to(wp, dir, allwords, curline, attrp)
|
|
|
1963 win_T *wp;
|
|
236
|
1964 int dir; /* FORWARD or BACKWARD */
|
|
500
|
1965 int allwords; /* TRUE for "[s"/"]s", FALSE for "[S"/"]S" */
|
|
323
|
1966 int curline;
|
|
534
|
1967 hlf_T *attrp; /* return: attributes of bad word or NULL
|
|
|
1968 (only when "dir" is FORWARD) */
|
|
236
|
1969 {
|
|
249
|
1970 linenr_T lnum;
|
|
|
1971 pos_T found_pos;
|
|
495
|
1972 int found_len = 0;
|
|
236
|
1973 char_u *line;
|
|
|
1974 char_u *p;
|
|
346
|
1975 char_u *endp;
|
|
534
|
1976 hlf_T attr;
|
|
236
|
1977 int len;
|
|
498
|
1978 int has_syntax = syntax_present(wp->w_buffer);
|
|
249
|
1979 int col;
|
|
|
1980 int can_spell;
|
|
346
|
1981 char_u *buf = NULL;
|
|
|
1982 int buflen = 0;
|
|
|
1983 int skip = 0;
|
|
385
|
1984 int capcol = -1;
|
|
500
|
1985 int found_one = FALSE;
|
|
|
1986 int wrapped = FALSE;
|
|
236
|
1987
|
|
498
|
1988 if (no_spell_checking(wp))
|
|
495
|
1989 return 0;
|
|
236
|
1990
|
|
249
|
1991 /*
|
|
|
1992 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1993 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1994 *
|
|
|
1995 * When searching backwards, we continue in the line to find the last
|
|
|
1996 * bad word (in the cursor line: before the cursor).
|
|
346
|
1997 *
|
|
|
1998 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1999 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
2000 * though...
|
|
249
|
2001 */
|
|
498
|
2002 lnum = wp->w_cursor.lnum;
|
|
249
|
2003 found_pos.lnum = 0;
|
|
236
|
2004
|
|
|
2005 while (!got_int)
|
|
|
2006 {
|
|
498
|
2007 line = ml_get_buf(wp->w_buffer, lnum, FALSE);
|
|
346
|
2008
|
|
|
2009 len = STRLEN(line);
|
|
|
2010 if (buflen < len + MAXWLEN + 2)
|
|
|
2011 {
|
|
|
2012 vim_free(buf);
|
|
|
2013 buflen = len + MAXWLEN + 2;
|
|
|
2014 buf = alloc(buflen);
|
|
|
2015 if (buf == NULL)
|
|
|
2016 break;
|
|
|
2017 }
|
|
|
2018
|
|
385
|
2019 /* In first line check first word for Capital. */
|
|
|
2020 if (lnum == 1)
|
|
|
2021 capcol = 0;
|
|
|
2022
|
|
|
2023 /* For checking first word with a capital skip white space. */
|
|
|
2024 if (capcol == 0)
|
|
|
2025 capcol = skipwhite(line) - line;
|
|
|
2026
|
|
346
|
2027 /* Copy the line into "buf" and append the start of the next line if
|
|
|
2028 * possible. */
|
|
|
2029 STRCPY(buf, line);
|
|
498
|
2030 if (lnum < wp->w_buffer->b_ml.ml_line_count)
|
|
346
|
2031 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
2032
|
|
|
2033 p = buf + skip;
|
|
|
2034 endp = buf + len;
|
|
|
2035 while (p < endp)
|
|
236
|
2036 {
|
|
500
|
2037 /* When searching backward don't search after the cursor. Unless
|
|
|
2038 * we wrapped around the end of the buffer. */
|
|
300
|
2039 if (dir == BACKWARD
|
|
498
|
2040 && lnum == wp->w_cursor.lnum
|
|
500
|
2041 && !wrapped
|
|
498
|
2042 && (colnr_T)(p - buf) >= wp->w_cursor.col)
|
|
300
|
2043 break;
|
|
249
|
2044
|
|
300
|
2045 /* start of word */
|
|
534
|
2046 attr = HLF_COUNT;
|
|
625
|
2047 len = spell_check(wp, p, &attr, &capcol, FALSE);
|
|
249
|
2048
|
|
534
|
2049 if (attr != HLF_COUNT)
|
|
300
|
2050 {
|
|
|
2051 /* We found a bad word. Check the attribute. */
|
|
534
|
2052 if (allwords || attr == HLF_SPB)
|
|
236
|
2053 {
|
|
500
|
2054 found_one = TRUE;
|
|
|
2055
|
|
300
|
2056 /* When searching forward only accept a bad word after
|
|
|
2057 * the cursor. */
|
|
|
2058 if (dir == BACKWARD
|
|
500
|
2059 || lnum != wp->w_cursor.lnum
|
|
498
|
2060 || (lnum == wp->w_cursor.lnum
|
|
500
|
2061 && (wrapped
|
|
|
2062 || (colnr_T)(curline ? p - buf + len
|
|
346
|
2063 : p - buf)
|
|
500
|
2064 > wp->w_cursor.col)))
|
|
236
|
2065 {
|
|
300
|
2066 if (has_syntax)
|
|
249
|
2067 {
|
|
346
|
2068 col = p - buf;
|
|
498
|
2069 (void)syn_get_id(wp, lnum, (colnr_T)col,
|
|
300
|
2070 FALSE, &can_spell);
|
|
|
2071 }
|
|
|
2072 else
|
|
|
2073 can_spell = TRUE;
|
|
249
|
2074
|
|
300
|
2075 if (can_spell)
|
|
|
2076 {
|
|
|
2077 found_pos.lnum = lnum;
|
|
346
|
2078 found_pos.col = p - buf;
|
|
249
|
2079 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
2080 found_pos.coladd = 0;
|
|
249
|
2081 #endif
|
|
300
|
2082 if (dir == FORWARD)
|
|
|
2083 {
|
|
|
2084 /* No need to search further. */
|
|
498
|
2085 wp->w_cursor = found_pos;
|
|
346
|
2086 vim_free(buf);
|
|
498
|
2087 if (attrp != NULL)
|
|
|
2088 *attrp = attr;
|
|
495
|
2089 return len;
|
|
249
|
2090 }
|
|
483
|
2091 else if (curline)
|
|
|
2092 /* Insert mode completion: put cursor after
|
|
|
2093 * the bad word. */
|
|
|
2094 found_pos.col += len;
|
|
495
|
2095 found_len = len;
|
|
249
|
2096 }
|
|
236
|
2097 }
|
|
|
2098 }
|
|
|
2099 }
|
|
|
2100
|
|
300
|
2101 /* advance to character after the word */
|
|
|
2102 p += len;
|
|
385
|
2103 capcol -= len;
|
|
236
|
2104 }
|
|
|
2105
|
|
483
|
2106 if (dir == BACKWARD && found_pos.lnum != 0)
|
|
|
2107 {
|
|
500
|
2108 /* Use the last match in the line (before the cursor). */
|
|
498
|
2109 wp->w_cursor = found_pos;
|
|
483
|
2110 vim_free(buf);
|
|
495
|
2111 return found_len;
|
|
483
|
2112 }
|
|
|
2113
|
|
323
|
2114 if (curline)
|
|
346
|
2115 break; /* only check cursor line */
|
|
323
|
2116
|
|
236
|
2117 /* Advance to next line. */
|
|
249
|
2118 if (dir == BACKWARD)
|
|
|
2119 {
|
|
500
|
2120 /* If we are back at the starting line and searched it again there
|
|
|
2121 * is no match, give up. */
|
|
|
2122 if (lnum == wp->w_cursor.lnum && wrapped)
|
|
346
|
2123 break;
|
|
500
|
2124
|
|
|
2125 if (lnum > 1)
|
|
|
2126 --lnum;
|
|
|
2127 else if (!p_ws)
|
|
|
2128 break; /* at first line and 'nowrapscan' */
|
|
|
2129 else
|
|
|
2130 {
|
|
|
2131 /* Wrap around to the end of the buffer. May search the
|
|
|
2132 * starting line again and accept the last match. */
|
|
|
2133 lnum = wp->w_buffer->b_ml.ml_line_count;
|
|
|
2134 wrapped = TRUE;
|
|
503
|
2135 if (!shortmess(SHM_SEARCH))
|
|
|
2136 give_warning((char_u *)_(top_bot_msg), TRUE);
|
|
500
|
2137 }
|
|
385
|
2138 capcol = -1;
|
|
249
|
2139 }
|
|
|
2140 else
|
|
|
2141 {
|
|
500
|
2142 if (lnum < wp->w_buffer->b_ml.ml_line_count)
|
|
|
2143 ++lnum;
|
|
|
2144 else if (!p_ws)
|
|
|
2145 break; /* at first line and 'nowrapscan' */
|
|
|
2146 else
|
|
|
2147 {
|
|
|
2148 /* Wrap around to the start of the buffer. May search the
|
|
|
2149 * starting line again and accept the first match. */
|
|
|
2150 lnum = 1;
|
|
|
2151 wrapped = TRUE;
|
|
503
|
2152 if (!shortmess(SHM_SEARCH))
|
|
|
2153 give_warning((char_u *)_(bot_top_msg), TRUE);
|
|
500
|
2154 }
|
|
|
2155
|
|
|
2156 /* If we are back at the starting line and there is no match then
|
|
|
2157 * give up. */
|
|
|
2158 if (lnum == wp->w_cursor.lnum && !found_one)
|
|
346
|
2159 break;
|
|
|
2160
|
|
|
2161 /* Skip the characters at the start of the next line that were
|
|
|
2162 * included in a match crossing line boundaries. */
|
|
534
|
2163 if (attr == HLF_COUNT)
|
|
346
|
2164 skip = p - endp;
|
|
|
2165 else
|
|
|
2166 skip = 0;
|
|
385
|
2167
|
|
|
2168 /* Capscol skips over the inserted space. */
|
|
|
2169 --capcol;
|
|
|
2170
|
|
|
2171 /* But after empty line check first word in next line */
|
|
|
2172 if (*skipwhite(line) == NUL)
|
|
|
2173 capcol = 0;
|
|
249
|
2174 }
|
|
236
|
2175
|
|
|
2176 line_breakcheck();
|
|
|
2177 }
|
|
|
2178
|
|
346
|
2179 vim_free(buf);
|
|
495
|
2180 return 0;
|
|
346
|
2181 }
|
|
|
2182
|
|
|
2183 /*
|
|
|
2184 * For spell checking: concatenate the start of the following line "line" into
|
|
|
2185 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
2186 */
|
|
|
2187 void
|
|
|
2188 spell_cat_line(buf, line, maxlen)
|
|
|
2189 char_u *buf;
|
|
|
2190 char_u *line;
|
|
|
2191 int maxlen;
|
|
|
2192 {
|
|
|
2193 char_u *p;
|
|
|
2194 int n;
|
|
|
2195
|
|
|
2196 p = skipwhite(line);
|
|
|
2197 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
2198 p = skipwhite(p + 1);
|
|
|
2199
|
|
|
2200 if (*p != NUL)
|
|
|
2201 {
|
|
|
2202 *buf = ' ';
|
|
376
|
2203 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
2204 n = p - line;
|
|
|
2205 if (n >= maxlen)
|
|
|
2206 n = maxlen - 1;
|
|
|
2207 vim_memset(buf + 1, ' ', n);
|
|
|
2208 }
|
|
236
|
2209 }
|
|
|
2210
|
|
626
|
2211 /*
|
|
|
2212 * Structure used for the cookie argument of do_in_runtimepath().
|
|
|
2213 */
|
|
501
|
2214 typedef struct spelload_S
|
|
|
2215 {
|
|
|
2216 char_u sl_lang[MAXWLEN + 1]; /* language name */
|
|
|
2217 slang_T *sl_slang; /* resulting slang_T struct */
|
|
|
2218 int sl_nobreak; /* NOBREAK language found */
|
|
|
2219 } spelload_T;
|
|
|
2220
|
|
236
|
2221 /*
|
|
307
|
2222 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
2223 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
2224 */
|
|
307
|
2225 static void
|
|
236
|
2226 spell_load_lang(lang)
|
|
|
2227 char_u *lang;
|
|
|
2228 {
|
|
310
|
2229 char_u fname_enc[85];
|
|
236
|
2230 int r;
|
|
501
|
2231 spelload_T sl;
|
|
307
|
2232
|
|
310
|
2233 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
2234 * It's truncated when an error is detected. */
|
|
501
|
2235 STRCPY(sl.sl_lang, lang);
|
|
|
2236 sl.sl_slang = NULL;
|
|
|
2237 sl.sl_nobreak = FALSE;
|
|
307
|
2238
|
|
310
|
2239 /*
|
|
|
2240 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
2241 */
|
|
|
2242 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
2243 "spell/%s.%s.spl", lang, spell_enc());
|
|
501
|
2244 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl);
|
|
|
2245
|
|
|
2246 if (r == FAIL && *sl.sl_lang != NUL)
|
|
307
|
2247 {
|
|
|
2248 /* Try loading the ASCII version. */
|
|
310
|
2249 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
2250 "spell/%s.ascii.spl", lang);
|
|
501
|
2251 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl);
|
|
307
|
2252 }
|
|
|
2253
|
|
|
2254 if (r == FAIL)
|
|
483
|
2255 smsg((char_u *)_("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""),
|
|
|
2256 lang, spell_enc(), lang);
|
|
501
|
2257 else if (sl.sl_slang != NULL)
|
|
|
2258 {
|
|
626
|
2259 /* At least one file was loaded, now load ALL the additions. */
|
|
310
|
2260 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
501
|
2261 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &sl);
|
|
310
|
2262 }
|
|
|
2263 }
|
|
|
2264
|
|
|
2265 /*
|
|
|
2266 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
2267 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
2268 */
|
|
|
2269 static char_u *
|
|
|
2270 spell_enc()
|
|
|
2271 {
|
|
|
2272
|
|
|
2273 #ifdef FEAT_MBYTE
|
|
|
2274 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
2275 return p_enc;
|
|
|
2276 #endif
|
|
|
2277 return (char_u *)"latin1";
|
|
236
|
2278 }
|
|
|
2279
|
|
|
2280 /*
|
|
385
|
2281 * Get the name of the .spl file for the internal wordlist into
|
|
|
2282 * "fname[MAXPATHL]".
|
|
|
2283 */
|
|
|
2284 static void
|
|
|
2285 int_wordlist_spl(fname)
|
|
|
2286 char_u *fname;
|
|
|
2287 {
|
|
|
2288 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
2289 int_wordlist, spell_enc());
|
|
|
2290 }
|
|
|
2291
|
|
|
2292 /*
|
|
625
|
2293 * Allocate a new slang_T for language "lang". "lang" can be NULL.
|
|
236
|
2294 * Caller must fill "sl_next".
|
|
|
2295 */
|
|
|
2296 static slang_T *
|
|
|
2297 slang_alloc(lang)
|
|
|
2298 char_u *lang;
|
|
|
2299 {
|
|
|
2300 slang_T *lp;
|
|
|
2301
|
|
300
|
2302 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
2303 if (lp != NULL)
|
|
|
2304 {
|
|
625
|
2305 if (lang != NULL)
|
|
|
2306 lp->sl_name = vim_strsave(lang);
|
|
323
|
2307 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
625
|
2308 ga_init2(&lp->sl_repsal, sizeof(fromto_T), 10);
|
|
483
|
2309 lp->sl_compmax = MAXWLEN;
|
|
|
2310 lp->sl_compsylmax = MAXWLEN;
|
|
625
|
2311 hash_init(&lp->sl_wordcount);
|
|
|
2312 }
|
|
|
2313
|
|
236
|
2314 return lp;
|
|
|
2315 }
|
|
|
2316
|
|
|
2317 /*
|
|
|
2318 * Free the contents of an slang_T and the structure itself.
|
|
|
2319 */
|
|
|
2320 static void
|
|
|
2321 slang_free(lp)
|
|
|
2322 slang_T *lp;
|
|
|
2323 {
|
|
|
2324 vim_free(lp->sl_name);
|
|
310
|
2325 vim_free(lp->sl_fname);
|
|
|
2326 slang_clear(lp);
|
|
|
2327 vim_free(lp);
|
|
|
2328 }
|
|
|
2329
|
|
|
2330 /*
|
|
|
2331 * Clear an slang_T so that the file can be reloaded.
|
|
|
2332 */
|
|
|
2333 static void
|
|
|
2334 slang_clear(lp)
|
|
|
2335 slang_T *lp;
|
|
|
2336 {
|
|
339
|
2337 garray_T *gap;
|
|
|
2338 fromto_T *ftp;
|
|
344
|
2339 salitem_T *smp;
|
|
339
|
2340 int i;
|
|
625
|
2341 int round;
|
|
323
|
2342
|
|
300
|
2343 vim_free(lp->sl_fbyts);
|
|
310
|
2344 lp->sl_fbyts = NULL;
|
|
300
|
2345 vim_free(lp->sl_kbyts);
|
|
310
|
2346 lp->sl_kbyts = NULL;
|
|
339
|
2347 vim_free(lp->sl_pbyts);
|
|
|
2348 lp->sl_pbyts = NULL;
|
|
|
2349
|
|
300
|
2350 vim_free(lp->sl_fidxs);
|
|
310
|
2351 lp->sl_fidxs = NULL;
|
|
300
|
2352 vim_free(lp->sl_kidxs);
|
|
310
|
2353 lp->sl_kidxs = NULL;
|
|
339
|
2354 vim_free(lp->sl_pidxs);
|
|
|
2355 lp->sl_pidxs = NULL;
|
|
323
|
2356
|
|
625
|
2357 for (round = 1; round <= 2; ++round)
|
|
|
2358 {
|
|
|
2359 gap = round == 1 ? &lp->sl_rep : &lp->sl_repsal;
|
|
|
2360 while (gap->ga_len > 0)
|
|
|
2361 {
|
|
|
2362 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2363 vim_free(ftp->ft_from);
|
|
|
2364 vim_free(ftp->ft_to);
|
|
|
2365 }
|
|
|
2366 ga_clear(gap);
|
|
|
2367 }
|
|
344
|
2368
|
|
|
2369 gap = &lp->sl_sal;
|
|
375
|
2370 if (lp->sl_sofo)
|
|
376
|
2371 {
|
|
|
2372 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
2373 if (gap->ga_data != NULL)
|
|
|
2374 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
2375 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2376 vim_free(((int **)gap->ga_data)[i]);
|
|
|
2377 }
|
|
375
|
2378 else
|
|
|
2379 /* SAL items: free salitem_T items */
|
|
|
2380 while (gap->ga_len > 0)
|
|
|
2381 {
|
|
|
2382 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2383 vim_free(smp->sm_lead);
|
|
|
2384 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
2385 vim_free(smp->sm_to);
|
|
|
2386 #ifdef FEAT_MBYTE
|
|
|
2387 vim_free(smp->sm_lead_w);
|
|
|
2388 vim_free(smp->sm_oneof_w);
|
|
|
2389 vim_free(smp->sm_to_w);
|
|
|
2390 #endif
|
|
|
2391 }
|
|
344
|
2392 ga_clear(gap);
|
|
323
|
2393
|
|
339
|
2394 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
2395 vim_free(lp->sl_prefprog[i]);
|
|
376
|
2396 lp->sl_prefixcnt = 0;
|
|
339
|
2397 vim_free(lp->sl_prefprog);
|
|
376
|
2398 lp->sl_prefprog = NULL;
|
|
|
2399
|
|
|
2400 vim_free(lp->sl_midword);
|
|
|
2401 lp->sl_midword = NULL;
|
|
339
|
2402
|
|
483
|
2403 vim_free(lp->sl_compprog);
|
|
|
2404 vim_free(lp->sl_compstartflags);
|
|
485
|
2405 vim_free(lp->sl_compallflags);
|
|
483
|
2406 lp->sl_compprog = NULL;
|
|
|
2407 lp->sl_compstartflags = NULL;
|
|
485
|
2408 lp->sl_compallflags = NULL;
|
|
483
|
2409
|
|
|
2410 vim_free(lp->sl_syllable);
|
|
|
2411 lp->sl_syllable = NULL;
|
|
|
2412 ga_clear(&lp->sl_syl_items);
|
|
481
|
2413
|
|
625
|
2414 hash_clear_all(&lp->sl_wordcount, WC_KEY_OFF);
|
|
|
2415 hash_init(&lp->sl_wordcount);
|
|
|
2416
|
|
|
2417 #ifdef FEAT_MBYTE
|
|
|
2418 hash_clear_all(&lp->sl_map_hash, 0);
|
|
|
2419 #endif
|
|
|
2420
|
|
|
2421 /* Clear info from .sug file. */
|
|
|
2422 slang_clear_sug(lp);
|
|
483
|
2423
|
|
|
2424 lp->sl_compmax = MAXWLEN;
|
|
501
|
2425 lp->sl_compminlen = 0;
|
|
483
|
2426 lp->sl_compsylmax = MAXWLEN;
|
|
|
2427 lp->sl_regions[0] = NUL;
|
|
236
|
2428 }
|
|
|
2429
|
|
|
2430 /*
|
|
625
|
2431 * Clear the info from the .sug file in "lp".
|
|
|
2432 */
|
|
|
2433 static void
|
|
|
2434 slang_clear_sug(lp)
|
|
|
2435 slang_T *lp;
|
|
|
2436 {
|
|
|
2437 vim_free(lp->sl_sbyts);
|
|
|
2438 lp->sl_sbyts = NULL;
|
|
|
2439 vim_free(lp->sl_sidxs);
|
|
|
2440 lp->sl_sidxs = NULL;
|
|
|
2441 close_spellbuf(lp->sl_sugbuf);
|
|
|
2442 lp->sl_sugbuf = NULL;
|
|
|
2443 lp->sl_sugloaded = FALSE;
|
|
|
2444 lp->sl_sugtime = 0;
|
|
|
2445 }
|
|
|
2446
|
|
|
2447 /*
|
|
307
|
2448 * Load one spell file and store the info into a slang_T.
|
|
236
|
2449 * Invoked through do_in_runtimepath().
|
|
|
2450 */
|
|
|
2451 static void
|
|
310
|
2452 spell_load_cb(fname, cookie)
|
|
236
|
2453 char_u *fname;
|
|
501
|
2454 void *cookie;
|
|
|
2455 {
|
|
|
2456 spelload_T *slp = (spelload_T *)cookie;
|
|
|
2457 slang_T *slang;
|
|
|
2458
|
|
|
2459 slang = spell_load_file(fname, slp->sl_lang, NULL, FALSE);
|
|
|
2460 if (slang != NULL)
|
|
|
2461 {
|
|
|
2462 /* When a previously loaded file has NOBREAK also use it for the
|
|
|
2463 * ".add" files. */
|
|
|
2464 if (slp->sl_nobreak && slang->sl_add)
|
|
|
2465 slang->sl_nobreak = TRUE;
|
|
|
2466 else if (slang->sl_nobreak)
|
|
|
2467 slp->sl_nobreak = TRUE;
|
|
|
2468
|
|
|
2469 slp->sl_slang = slang;
|
|
|
2470 }
|
|
310
|
2471 }
|
|
|
2472
|
|
|
2473 /*
|
|
|
2474 * Load one spell file and store the info into a slang_T.
|
|
|
2475 *
|
|
625
|
2476 * This is invoked in three ways:
|
|
310
|
2477 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
2478 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
2479 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
2480 * points to the existing slang_T.
|
|
625
|
2481 * - Just after writing a .spl file; it's read back to produce the .sug file.
|
|
626
|
2482 * "old_lp" is NULL and "lang" is NULL. Will allocate an slang_T.
|
|
|
2483 *
|
|
323
|
2484 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
2485 */
|
|
323
|
2486 static slang_T *
|
|
|
2487 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
2488 char_u *fname;
|
|
|
2489 char_u *lang;
|
|
|
2490 slang_T *old_lp;
|
|
323
|
2491 int silent; /* no error if file doesn't exist */
|
|
310
|
2492 {
|
|
236
|
2493 FILE *fd;
|
|
483
|
2494 char_u buf[VIMSPELLMAGICL];
|
|
236
|
2495 char_u *p;
|
|
|
2496 int i;
|
|
339
|
2497 int n;
|
|
300
|
2498 int len;
|
|
236
|
2499 char_u *save_sourcing_name = sourcing_name;
|
|
|
2500 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
307
|
2501 slang_T *lp = NULL;
|
|
344
|
2502 int c = 0;
|
|
483
|
2503 int res;
|
|
236
|
2504
|
|
310
|
2505 fd = mch_fopen((char *)fname, "r");
|
|
236
|
2506 if (fd == NULL)
|
|
|
2507 {
|
|
323
|
2508 if (!silent)
|
|
|
2509 EMSG2(_(e_notopen), fname);
|
|
|
2510 else if (p_verbose > 2)
|
|
|
2511 {
|
|
|
2512 verbose_enter();
|
|
|
2513 smsg((char_u *)e_notopen, fname);
|
|
|
2514 verbose_leave();
|
|
|
2515 }
|
|
255
|
2516 goto endFAIL;
|
|
236
|
2517 }
|
|
310
|
2518 if (p_verbose > 2)
|
|
|
2519 {
|
|
|
2520 verbose_enter();
|
|
|
2521 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
2522 verbose_leave();
|
|
|
2523 }
|
|
|
2524
|
|
|
2525 if (old_lp == NULL)
|
|
|
2526 {
|
|
|
2527 lp = slang_alloc(lang);
|
|
|
2528 if (lp == NULL)
|
|
|
2529 goto endFAIL;
|
|
|
2530
|
|
|
2531 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
2532 lp->sl_fname = vim_strsave(fname);
|
|
|
2533 if (lp->sl_fname == NULL)
|
|
|
2534 goto endFAIL;
|
|
|
2535
|
|
|
2536 /* Check for .add.spl. */
|
|
|
2537 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
2538 }
|
|
|
2539 else
|
|
|
2540 lp = old_lp;
|
|
307
|
2541
|
|
236
|
2542 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
2543 sourcing_name = fname;
|
|
|
2544 sourcing_lnum = 0;
|
|
|
2545
|
|
625
|
2546 /*
|
|
|
2547 * <HEADER>: <fileID>
|
|
339
|
2548 */
|
|
236
|
2549 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
2550 buf[i] = getc(fd); /* <fileID> */
|
|
|
2551 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
2552 {
|
|
483
|
2553 EMSG(_("E757: This does not look like a spell file"));
|
|
381
|
2554 goto endFAIL;
|
|
483
|
2555 }
|
|
|
2556 c = getc(fd); /* <versionnr> */
|
|
|
2557 if (c < VIMSPELLVERSION)
|
|
|
2558 {
|
|
|
2559 EMSG(_("E771: Old spell file, needs to be updated"));
|
|
381
|
2560 goto endFAIL;
|
|
|
2561 }
|
|
483
|
2562 else if (c > VIMSPELLVERSION)
|
|
|
2563 {
|
|
|
2564 EMSG(_("E772: Spell file is for newer version of Vim"));
|
|
481
|
2565 goto endFAIL;
|
|
483
|
2566 }
|
|
|
2567
|
|
|
2568
|
|
|
2569 /*
|
|
|
2570 * <SECTIONS>: <section> ... <sectionend>
|
|
|
2571 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
2572 */
|
|
|
2573 for (;;)
|
|
|
2574 {
|
|
|
2575 n = getc(fd); /* <sectionID> or <sectionend> */
|
|
|
2576 if (n == SN_END)
|
|
|
2577 break;
|
|
|
2578 c = getc(fd); /* <sectionflags> */
|
|
|
2579 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2580 /* <sectionlen> */
|
|
|
2581 if (len < 0)
|
|
|
2582 goto truncerr;
|
|
|
2583
|
|
|
2584 res = 0;
|
|
|
2585 switch (n)
|
|
|
2586 {
|
|
|
2587 case SN_REGION:
|
|
|
2588 res = read_region_section(fd, lp, len);
|
|
|
2589 break;
|
|
|
2590
|
|
|
2591 case SN_CHARFLAGS:
|
|
|
2592 res = read_charflags_section(fd);
|
|
|
2593 break;
|
|
|
2594
|
|
|
2595 case SN_MIDWORD:
|
|
|
2596 lp->sl_midword = read_string(fd, len); /* <midword> */
|
|
|
2597 if (lp->sl_midword == NULL)
|
|
|
2598 goto endFAIL;
|
|
|
2599 break;
|
|
|
2600
|
|
|
2601 case SN_PREFCOND:
|
|
|
2602 res = read_prefcond_section(fd, lp);
|
|
|
2603 break;
|
|
|
2604
|
|
|
2605 case SN_REP:
|
|
625
|
2606 res = read_rep_section(fd, &lp->sl_rep, lp->sl_rep_first);
|
|
|
2607 break;
|
|
|
2608
|
|
|
2609 case SN_REPSAL:
|
|
|
2610 res = read_rep_section(fd, &lp->sl_repsal, lp->sl_repsal_first);
|
|
483
|
2611 break;
|
|
|
2612
|
|
|
2613 case SN_SAL:
|
|
|
2614 res = read_sal_section(fd, lp);
|
|
|
2615 break;
|
|
|
2616
|
|
|
2617 case SN_SOFO:
|
|
|
2618 res = read_sofo_section(fd, lp);
|
|
|
2619 break;
|
|
|
2620
|
|
|
2621 case SN_MAP:
|
|
|
2622 p = read_string(fd, len); /* <mapstr> */
|
|
|
2623 if (p == NULL)
|
|
|
2624 goto endFAIL;
|
|
|
2625 set_map_str(lp, p);
|
|
|
2626 vim_free(p);
|
|
|
2627 break;
|
|
|
2628
|
|
625
|
2629 case SN_WORDS:
|
|
|
2630 res = read_words_section(fd, lp, len);
|
|
|
2631 break;
|
|
|
2632
|
|
|
2633 case SN_SUGFILE:
|
|
|
2634 for (i = 7; i >= 0; --i) /* <timestamp> */
|
|
|
2635 lp->sl_sugtime += getc(fd) << (i * 8);
|
|
|
2636 break;
|
|
|
2637
|
|
483
|
2638 case SN_COMPOUND:
|
|
|
2639 res = read_compound(fd, lp, len);
|
|
|
2640 break;
|
|
|
2641
|
|
492
|
2642 case SN_NOBREAK:
|
|
|
2643 lp->sl_nobreak = TRUE;
|
|
|
2644 break;
|
|
|
2645
|
|
483
|
2646 case SN_SYLLABLE:
|
|
|
2647 lp->sl_syllable = read_string(fd, len); /* <syllable> */
|
|
|
2648 if (lp->sl_syllable == NULL)
|
|
|
2649 goto endFAIL;
|
|
|
2650 if (init_syl_tab(lp) == FAIL)
|
|
|
2651 goto endFAIL;
|
|
|
2652 break;
|
|
|
2653
|
|
|
2654 default:
|
|
|
2655 /* Unsupported section. When it's required give an error
|
|
|
2656 * message. When it's not required skip the contents. */
|
|
|
2657 if (c & SNF_REQUIRED)
|
|
|
2658 {
|
|
|
2659 EMSG(_("E770: Unsupported section in spell file"));
|
|
|
2660 goto endFAIL;
|
|
|
2661 }
|
|
|
2662 while (--len >= 0)
|
|
|
2663 if (getc(fd) < 0)
|
|
|
2664 goto truncerr;
|
|
|
2665 break;
|
|
|
2666 }
|
|
625
|
2667 someerror:
|
|
483
|
2668 if (res == SP_FORMERROR)
|
|
|
2669 {
|
|
|
2670 EMSG(_(e_format));
|
|
381
|
2671 goto endFAIL;
|
|
344
|
2672 }
|
|
483
|
2673 if (res == SP_TRUNCERROR)
|
|
|
2674 {
|
|
|
2675 truncerr:
|
|
|
2676 EMSG(_(e_spell_trunc));
|
|
375
|
2677 goto endFAIL;
|
|
483
|
2678 }
|
|
495
|
2679 if (res == SP_OTHERERROR)
|
|
483
|
2680 goto endFAIL;
|
|
389
|
2681 }
|
|
236
|
2682
|
|
625
|
2683 /* <LWORDTREE> */
|
|
|
2684 res = spell_read_tree(fd, &lp->sl_fbyts, &lp->sl_fidxs, FALSE, 0);
|
|
|
2685 if (res != 0)
|
|
|
2686 goto someerror;
|
|
|
2687
|
|
|
2688 /* <KWORDTREE> */
|
|
|
2689 res = spell_read_tree(fd, &lp->sl_kbyts, &lp->sl_kidxs, FALSE, 0);
|
|
|
2690 if (res != 0)
|
|
|
2691 goto someerror;
|
|
|
2692
|
|
|
2693 /* <PREFIXTREE> */
|
|
|
2694 res = spell_read_tree(fd, &lp->sl_pbyts, &lp->sl_pidxs, TRUE,
|
|
|
2695 lp->sl_prefixcnt);
|
|
|
2696 if (res != 0)
|
|
|
2697 goto someerror;
|
|
243
|
2698
|
|
310
|
2699 /* For a new file link it in the list of spell files. */
|
|
626
|
2700 if (old_lp == NULL && lang != NULL)
|
|
310
|
2701 {
|
|
|
2702 lp->sl_next = first_lang;
|
|
|
2703 first_lang = lp;
|
|
|
2704 }
|
|
307
|
2705
|
|
255
|
2706 goto endOK;
|
|
|
2707
|
|
|
2708 endFAIL:
|
|
310
|
2709 if (lang != NULL)
|
|
|
2710 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2711 *lang = NUL;
|
|
|
2712 if (lp != NULL && old_lp == NULL)
|
|
307
|
2713 slang_free(lp);
|
|
500
|
2714 lp = NULL;
|
|
255
|
2715
|
|
|
2716 endOK:
|
|
236
|
2717 if (fd != NULL)
|
|
|
2718 fclose(fd);
|
|
|
2719 sourcing_name = save_sourcing_name;
|
|
|
2720 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2721
|
|
|
2722 return lp;
|
|
236
|
2723 }
|
|
|
2724
|
|
376
|
2725 /*
|
|
|
2726 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2727 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2728 * Returns NULL when the count is zero.
|
|
483
|
2729 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result
|
|
|
2730 * otherwise.
|
|
376
|
2731 */
|
|
|
2732 static char_u *
|
|
389
|
2733 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2734 FILE *fd;
|
|
|
2735 int cnt_bytes;
|
|
389
|
2736 int *cntp;
|
|
376
|
2737 {
|
|
|
2738 int cnt = 0;
|
|
|
2739 int i;
|
|
|
2740 char_u *str;
|
|
|
2741
|
|
|
2742 /* read the length bytes, MSB first */
|
|
|
2743 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2744 cnt = (cnt << 8) + getc(fd);
|
|
|
2745 if (cnt < 0)
|
|
|
2746 {
|
|
483
|
2747 *cntp = SP_TRUNCERROR;
|
|
376
|
2748 return NULL;
|
|
|
2749 }
|
|
389
|
2750 *cntp = cnt;
|
|
|
2751 if (cnt == 0)
|
|
|
2752 return NULL; /* nothing to read, return NULL */
|
|
376
|
2753
|
|
483
|
2754 str = read_string(fd, cnt);
|
|
|
2755 if (str == NULL)
|
|
495
|
2756 *cntp = SP_OTHERERROR;
|
|
483
|
2757 return str;
|
|
|
2758 }
|
|
|
2759
|
|
|
2760 /*
|
|
|
2761 * Read a string of length "cnt" from "fd" into allocated memory.
|
|
|
2762 * Returns NULL when out of memory.
|
|
|
2763 */
|
|
|
2764 static char_u *
|
|
|
2765 read_string(fd, cnt)
|
|
|
2766 FILE *fd;
|
|
|
2767 int cnt;
|
|
|
2768 {
|
|
|
2769 char_u *str;
|
|
|
2770 int i;
|
|
|
2771
|
|
376
|
2772 /* allocate memory */
|
|
|
2773 str = alloc((unsigned)cnt + 1);
|
|
483
|
2774 if (str != NULL)
|
|
|
2775 {
|
|
|
2776 /* Read the string. Doesn't check for truncated file. */
|
|
|
2777 for (i = 0; i < cnt; ++i)
|
|
|
2778 str[i] = getc(fd);
|
|
|
2779 str[i] = NUL;
|
|
|
2780 }
|
|
|
2781 return str;
|
|
|
2782 }
|
|
|
2783
|
|
|
2784 /*
|
|
|
2785 * Read SN_REGION: <regionname> ...
|
|
|
2786 * Return SP_*ERROR flags.
|
|
|
2787 */
|
|
|
2788 static int
|
|
|
2789 read_region_section(fd, lp, len)
|
|
|
2790 FILE *fd;
|
|
|
2791 slang_T *lp;
|
|
|
2792 int len;
|
|
|
2793 {
|
|
|
2794 int i;
|
|
|
2795
|
|
|
2796 if (len > 16)
|
|
|
2797 return SP_FORMERROR;
|
|
|
2798 for (i = 0; i < len; ++i)
|
|
|
2799 lp->sl_regions[i] = getc(fd); /* <regionname> */
|
|
|
2800 lp->sl_regions[len] = NUL;
|
|
|
2801 return 0;
|
|
|
2802 }
|
|
|
2803
|
|
|
2804 /*
|
|
|
2805 * Read SN_CHARFLAGS section: <charflagslen> <charflags>
|
|
|
2806 * <folcharslen> <folchars>
|
|
|
2807 * Return SP_*ERROR flags.
|
|
|
2808 */
|
|
|
2809 static int
|
|
|
2810 read_charflags_section(fd)
|
|
|
2811 FILE *fd;
|
|
|
2812 {
|
|
|
2813 char_u *flags;
|
|
|
2814 char_u *fol;
|
|
|
2815 int flagslen, follen;
|
|
|
2816
|
|
|
2817 /* <charflagslen> <charflags> */
|
|
|
2818 flags = read_cnt_string(fd, 1, &flagslen);
|
|
|
2819 if (flagslen < 0)
|
|
|
2820 return flagslen;
|
|
|
2821
|
|
|
2822 /* <folcharslen> <folchars> */
|
|
|
2823 fol = read_cnt_string(fd, 2, &follen);
|
|
|
2824 if (follen < 0)
|
|
|
2825 {
|
|
|
2826 vim_free(flags);
|
|
|
2827 return follen;
|
|
|
2828 }
|
|
|
2829
|
|
|
2830 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
2831 if (flags != NULL && fol != NULL)
|
|
|
2832 set_spell_charflags(flags, flagslen, fol);
|
|
|
2833
|
|
|
2834 vim_free(flags);
|
|
|
2835 vim_free(fol);
|
|
|
2836
|
|
|
2837 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2838 if ((flags == NULL) != (fol == NULL))
|
|
|
2839 return SP_FORMERROR;
|
|
|
2840 return 0;
|
|
|
2841 }
|
|
|
2842
|
|
|
2843 /*
|
|
|
2844 * Read SN_PREFCOND section.
|
|
|
2845 * Return SP_*ERROR flags.
|
|
|
2846 */
|
|
|
2847 static int
|
|
|
2848 read_prefcond_section(fd, lp)
|
|
|
2849 FILE *fd;
|
|
|
2850 slang_T *lp;
|
|
|
2851 {
|
|
|
2852 int cnt;
|
|
|
2853 int i;
|
|
|
2854 int n;
|
|
|
2855 char_u *p;
|
|
|
2856 char_u buf[MAXWLEN + 1];
|
|
|
2857
|
|
|
2858 /* <prefcondcnt> <prefcond> ... */
|
|
|
2859 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
2860 if (cnt <= 0)
|
|
|
2861 return SP_FORMERROR;
|
|
|
2862
|
|
|
2863 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2864 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2865 if (lp->sl_prefprog == NULL)
|
|
495
|
2866 return SP_OTHERERROR;
|
|
483
|
2867 lp->sl_prefixcnt = cnt;
|
|
|
2868
|
|
376
|
2869 for (i = 0; i < cnt; ++i)
|
|
483
|
2870 {
|
|
|
2871 /* <prefcond> : <condlen> <condstr> */
|
|
|
2872 n = getc(fd); /* <condlen> */
|
|
|
2873 if (n < 0 || n >= MAXWLEN)
|
|
|
2874 return SP_FORMERROR;
|
|
|
2875
|
|
|
2876 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2877 * compile the regexp program used to check for the condition. */
|
|
|
2878 if (n > 0)
|
|
|
2879 {
|
|
|
2880 buf[0] = '^'; /* always match at one position only */
|
|
|
2881 p = buf + 1;
|
|
|
2882 while (n-- > 0)
|
|
|
2883 *p++ = getc(fd); /* <condstr> */
|
|
|
2884 *p = NUL;
|
|
|
2885 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
2886 }
|
|
|
2887 }
|
|
|
2888 return 0;
|
|
|
2889 }
|
|
|
2890
|
|
|
2891 /*
|
|
625
|
2892 * Read REP or REPSAL items section from "fd": <repcount> <rep> ...
|
|
483
|
2893 * Return SP_*ERROR flags.
|
|
|
2894 */
|
|
|
2895 static int
|
|
625
|
2896 read_rep_section(fd, gap, first)
|
|
483
|
2897 FILE *fd;
|
|
625
|
2898 garray_T *gap;
|
|
|
2899 short *first;
|
|
483
|
2900 {
|
|
|
2901 int cnt;
|
|
|
2902 fromto_T *ftp;
|
|
|
2903 int i;
|
|
|
2904
|
|
|
2905 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
2906 if (cnt < 0)
|
|
|
2907 return SP_TRUNCERROR;
|
|
|
2908
|
|
|
2909 if (ga_grow(gap, cnt) == FAIL)
|
|
495
|
2910 return SP_OTHERERROR;
|
|
483
|
2911
|
|
|
2912 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
2913 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2914 {
|
|
|
2915 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
|
2916 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
2917 if (i < 0)
|
|
|
2918 return i;
|
|
|
2919 if (i == 0)
|
|
|
2920 return SP_FORMERROR;
|
|
|
2921 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
2922 if (i <= 0)
|
|
|
2923 {
|
|
|
2924 vim_free(ftp->ft_from);
|
|
|
2925 if (i < 0)
|
|
|
2926 return i;
|
|
|
2927 return SP_FORMERROR;
|
|
|
2928 }
|
|
|
2929 }
|
|
|
2930
|
|
|
2931 /* Fill the first-index table. */
|
|
|
2932 for (i = 0; i < 256; ++i)
|
|
|
2933 first[i] = -1;
|
|
|
2934 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2935 {
|
|
|
2936 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
2937 if (first[*ftp->ft_from] == -1)
|
|
|
2938 first[*ftp->ft_from] = i;
|
|
|
2939 }
|
|
|
2940 return 0;
|
|
|
2941 }
|
|
|
2942
|
|
|
2943 /*
|
|
|
2944 * Read SN_SAL section: <salflags> <salcount> <sal> ...
|
|
|
2945 * Return SP_*ERROR flags.
|
|
|
2946 */
|
|
|
2947 static int
|
|
|
2948 read_sal_section(fd, slang)
|
|
|
2949 FILE *fd;
|
|
|
2950 slang_T *slang;
|
|
|
2951 {
|
|
|
2952 int i;
|
|
|
2953 int cnt;
|
|
|
2954 garray_T *gap;
|
|
|
2955 salitem_T *smp;
|
|
|
2956 int ccnt;
|
|
|
2957 char_u *p;
|
|
485
|
2958 int c = NUL;
|
|
483
|
2959
|
|
|
2960 slang->sl_sofo = FALSE;
|
|
|
2961
|
|
|
2962 i = getc(fd); /* <salflags> */
|
|
|
2963 if (i & SAL_F0LLOWUP)
|
|
|
2964 slang->sl_followup = TRUE;
|
|
|
2965 if (i & SAL_COLLAPSE)
|
|
|
2966 slang->sl_collapse = TRUE;
|
|
|
2967 if (i & SAL_REM_ACCENTS)
|
|
|
2968 slang->sl_rem_accents = TRUE;
|
|
|
2969
|
|
|
2970 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2971 if (cnt < 0)
|
|
|
2972 return SP_TRUNCERROR;
|
|
|
2973
|
|
|
2974 gap = &slang->sl_sal;
|
|
|
2975 ga_init2(gap, sizeof(salitem_T), 10);
|
|
548
|
2976 if (ga_grow(gap, cnt + 1) == FAIL)
|
|
495
|
2977 return SP_OTHERERROR;
|
|
483
|
2978
|
|
|
2979 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2980 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2981 {
|
|
|
2982 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2983 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2984 if (ccnt < 0)
|
|
|
2985 return SP_TRUNCERROR;
|
|
|
2986 if ((p = alloc(ccnt + 2)) == NULL)
|
|
495
|
2987 return SP_OTHERERROR;
|
|
483
|
2988 smp->sm_lead = p;
|
|
|
2989
|
|
|
2990 /* Read up to the first special char into sm_lead. */
|
|
|
2991 for (i = 0; i < ccnt; ++i)
|
|
|
2992 {
|
|
|
2993 c = getc(fd); /* <salfrom> */
|
|
|
2994 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2995 break;
|
|
|
2996 *p++ = c;
|
|
|
2997 }
|
|
|
2998 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2999 *p++ = NUL;
|
|
|
3000
|
|
|
3001 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
3002 if (c == '(')
|
|
|
3003 {
|
|
|
3004 smp->sm_oneof = p;
|
|
|
3005 for (++i; i < ccnt; ++i)
|
|
|
3006 {
|
|
|
3007 c = getc(fd); /* <salfrom> */
|
|
|
3008 if (c == ')')
|
|
|
3009 break;
|
|
|
3010 *p++ = c;
|
|
|
3011 }
|
|
|
3012 *p++ = NUL;
|
|
|
3013 if (++i < ccnt)
|
|
|
3014 c = getc(fd);
|
|
|
3015 }
|
|
|
3016 else
|
|
|
3017 smp->sm_oneof = NULL;
|
|
|
3018
|
|
|
3019 /* Any following chars go in sm_rules. */
|
|
|
3020 smp->sm_rules = p;
|
|
|
3021 if (i < ccnt)
|
|
|
3022 /* store the char we got while checking for end of sm_lead */
|
|
|
3023 *p++ = c;
|
|
|
3024 for (++i; i < ccnt; ++i)
|
|
|
3025 *p++ = getc(fd); /* <salfrom> */
|
|
|
3026 *p++ = NUL;
|
|
|
3027
|
|
|
3028 /* <saltolen> <salto> */
|
|
|
3029 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
3030 if (ccnt < 0)
|
|
|
3031 {
|
|
|
3032 vim_free(smp->sm_lead);
|
|
|
3033 return ccnt;
|
|
|
3034 }
|
|
|
3035
|
|
|
3036 #ifdef FEAT_MBYTE
|
|
|
3037 if (has_mbyte)
|
|
|
3038 {
|
|
|
3039 /* convert the multi-byte strings to wide char strings */
|
|
|
3040 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
3041 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
3042 if (smp->sm_oneof == NULL)
|
|
|
3043 smp->sm_oneof_w = NULL;
|
|
|
3044 else
|
|
|
3045 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
3046 if (smp->sm_to == NULL)
|
|
|
3047 smp->sm_to_w = NULL;
|
|
|
3048 else
|
|
|
3049 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
3050 if (smp->sm_lead_w == NULL
|
|
|
3051 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
3052 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
|
3053 {
|
|
|
3054 vim_free(smp->sm_lead);
|
|
|
3055 vim_free(smp->sm_to);
|
|
|
3056 vim_free(smp->sm_lead_w);
|
|
|
3057 vim_free(smp->sm_oneof_w);
|
|
|
3058 vim_free(smp->sm_to_w);
|
|
495
|
3059 return SP_OTHERERROR;
|
|
483
|
3060 }
|
|
|
3061 }
|
|
|
3062 #endif
|
|
|
3063 }
|
|
|
3064
|
|
548
|
3065 if (gap->ga_len > 0)
|
|
|
3066 {
|
|
|
3067 /* Add one extra entry to mark the end with an empty sm_lead. Avoids
|
|
|
3068 * that we need to check the index every time. */
|
|
|
3069 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
3070 if ((p = alloc(1)) == NULL)
|
|
|
3071 return SP_OTHERERROR;
|
|
|
3072 p[0] = NUL;
|
|
|
3073 smp->sm_lead = p;
|
|
|
3074 smp->sm_leadlen = 0;
|
|
|
3075 smp->sm_oneof = NULL;
|
|
|
3076 smp->sm_rules = p;
|
|
|
3077 smp->sm_to = NULL;
|
|
|
3078 #ifdef FEAT_MBYTE
|
|
|
3079 if (has_mbyte)
|
|
|
3080 {
|
|
|
3081 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
3082 smp->sm_leadlen = 0;
|
|
|
3083 smp->sm_oneof_w = NULL;
|
|
|
3084 smp->sm_to_w = NULL;
|
|
|
3085 }
|
|
|
3086 #endif
|
|
|
3087 ++gap->ga_len;
|
|
|
3088 }
|
|
|
3089
|
|
483
|
3090 /* Fill the first-index table. */
|
|
|
3091 set_sal_first(slang);
|
|
|
3092
|
|
|
3093 return 0;
|
|
|
3094 }
|
|
|
3095
|
|
|
3096 /*
|
|
625
|
3097 * Read SN_WORDS: <word> ...
|
|
|
3098 * Return SP_*ERROR flags.
|
|
|
3099 */
|
|
|
3100 static int
|
|
|
3101 read_words_section(fd, lp, len)
|
|
|
3102 FILE *fd;
|
|
|
3103 slang_T *lp;
|
|
|
3104 int len;
|
|
|
3105 {
|
|
|
3106 int done = 0;
|
|
|
3107 int i;
|
|
|
3108 char_u word[MAXWLEN];
|
|
|
3109
|
|
|
3110 while (done < len)
|
|
|
3111 {
|
|
|
3112 /* Read one word at a time. */
|
|
|
3113 for (i = 0; ; ++i)
|
|
|
3114 {
|
|
|
3115 word[i] = getc(fd);
|
|
|
3116 if (word[i] == NUL)
|
|
|
3117 break;
|
|
|
3118 if (i == MAXWLEN - 1)
|
|
|
3119 return SP_FORMERROR;
|
|
|
3120 }
|
|
|
3121
|
|
|
3122 /* Init the count to 10. */
|
|
|
3123 count_common_word(lp, word, -1, 10);
|
|
|
3124 done += i + 1;
|
|
|
3125 }
|
|
|
3126 return 0;
|
|
|
3127 }
|
|
|
3128
|
|
|
3129 /*
|
|
|
3130 * Add a word to the hashtable of common words.
|
|
|
3131 * If it's already there then the counter is increased.
|
|
|
3132 */
|
|
|
3133 static void
|
|
|
3134 count_common_word(lp, word, len, count)
|
|
|
3135 slang_T *lp;
|
|
|
3136 char_u *word;
|
|
|
3137 int len; /* word length, -1 for upto NUL */
|
|
|
3138 int count; /* 1 to count once, 10 to init */
|
|
|
3139 {
|
|
|
3140 hash_T hash;
|
|
|
3141 hashitem_T *hi;
|
|
|
3142 wordcount_T *wc;
|
|
|
3143 char_u buf[MAXWLEN];
|
|
|
3144 char_u *p;
|
|
|
3145
|
|
|
3146 if (len == -1)
|
|
|
3147 p = word;
|
|
|
3148 else
|
|
|
3149 {
|
|
|
3150 vim_strncpy(buf, word, len);
|
|
|
3151 p = buf;
|
|
|
3152 }
|
|
|
3153
|
|
|
3154 hash = hash_hash(p);
|
|
|
3155 hi = hash_lookup(&lp->sl_wordcount, p, hash);
|
|
|
3156 if (HASHITEM_EMPTY(hi))
|
|
|
3157 {
|
|
|
3158 wc = (wordcount_T *)alloc(sizeof(wordcount_T) + STRLEN(p));
|
|
|
3159 if (wc == NULL)
|
|
|
3160 return;
|
|
|
3161 STRCPY(wc->wc_word, p);
|
|
|
3162 wc->wc_count = count;
|
|
|
3163 hash_add_item(&lp->sl_wordcount, hi, wc->wc_word, hash);
|
|
|
3164 }
|
|
|
3165 else
|
|
|
3166 {
|
|
|
3167 wc = HI2WC(hi);
|
|
|
3168 if ((wc->wc_count += count) < (unsigned)count) /* check for overflow */
|
|
|
3169 wc->wc_count = MAXWORDCOUNT;
|
|
|
3170 }
|
|
|
3171 }
|
|
|
3172
|
|
|
3173 /*
|
|
|
3174 * Adjust the score of common words.
|
|
|
3175 */
|
|
|
3176 static int
|
|
|
3177 score_wordcount_adj(slang, score, word, split)
|
|
|
3178 slang_T *slang;
|
|
|
3179 int score;
|
|
|
3180 char_u *word;
|
|
|
3181 int split; /* word was split, less bonus */
|
|
|
3182 {
|
|
|
3183 hashitem_T *hi;
|
|
|
3184 wordcount_T *wc;
|
|
|
3185 int bonus;
|
|
|
3186 int newscore;
|
|
|
3187
|
|
|
3188 hi = hash_find(&slang->sl_wordcount, word);
|
|
|
3189 if (!HASHITEM_EMPTY(hi))
|
|
|
3190 {
|
|
|
3191 wc = HI2WC(hi);
|
|
|
3192 if (wc->wc_count < SCORE_THRES2)
|
|
|
3193 bonus = SCORE_COMMON1;
|
|
|
3194 else if (wc->wc_count < SCORE_THRES3)
|
|
|
3195 bonus = SCORE_COMMON2;
|
|
|
3196 else
|
|
|
3197 bonus = SCORE_COMMON3;
|
|
|
3198 if (split)
|
|
|
3199 newscore = score - bonus / 2;
|
|
|
3200 else
|
|
|
3201 newscore = score - bonus;
|
|
|
3202 if (newscore < 0)
|
|
|
3203 return 0;
|
|
|
3204 return newscore;
|
|
|
3205 }
|
|
|
3206 return score;
|
|
|
3207 }
|
|
|
3208
|
|
|
3209 /*
|
|
483
|
3210 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
3211 * Return SP_*ERROR flags.
|
|
|
3212 */
|
|
|
3213 static int
|
|
|
3214 read_sofo_section(fd, slang)
|
|
|
3215 FILE *fd;
|
|
|
3216 slang_T *slang;
|
|
|
3217 {
|
|
|
3218 int cnt;
|
|
|
3219 char_u *from, *to;
|
|
|
3220 int res;
|
|
|
3221
|
|
|
3222 slang->sl_sofo = TRUE;
|
|
|
3223
|
|
|
3224 /* <sofofromlen> <sofofrom> */
|
|
|
3225 from = read_cnt_string(fd, 2, &cnt);
|
|
|
3226 if (cnt < 0)
|
|
|
3227 return cnt;
|
|
|
3228
|
|
|
3229 /* <sofotolen> <sofoto> */
|
|
|
3230 to = read_cnt_string(fd, 2, &cnt);
|
|
|
3231 if (cnt < 0)
|
|
|
3232 {
|
|
|
3233 vim_free(from);
|
|
|
3234 return cnt;
|
|
|
3235 }
|
|
|
3236
|
|
|
3237 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */
|
|
|
3238 if (from != NULL && to != NULL)
|
|
|
3239 res = set_sofo(slang, from, to);
|
|
|
3240 else if (from != NULL || to != NULL)
|
|
|
3241 res = SP_FORMERROR; /* only one of two strings is an error */
|
|
|
3242 else
|
|
|
3243 res = 0;
|
|
|
3244
|
|
|
3245 vim_free(from);
|
|
|
3246 vim_free(to);
|
|
|
3247 return res;
|
|
|
3248 }
|
|
|
3249
|
|
|
3250 /*
|
|
|
3251 * Read the compound section from the .spl file:
|
|
|
3252 * <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
3253 * Returns SP_*ERROR flags.
|
|
|
3254 */
|
|
|
3255 static int
|
|
|
3256 read_compound(fd, slang, len)
|
|
|
3257 FILE *fd;
|
|
|
3258 slang_T *slang;
|
|
|
3259 int len;
|
|
|
3260 {
|
|
|
3261 int todo = len;
|
|
|
3262 int c;
|
|
|
3263 int atstart;
|
|
|
3264 char_u *pat;
|
|
|
3265 char_u *pp;
|
|
|
3266 char_u *cp;
|
|
485
|
3267 char_u *ap;
|
|
483
|
3268
|
|
|
3269 if (todo < 2)
|
|
|
3270 return SP_FORMERROR; /* need at least two bytes */
|
|
|
3271
|
|
|
3272 --todo;
|
|
|
3273 c = getc(fd); /* <compmax> */
|
|
|
3274 if (c < 2)
|
|
|
3275 c = MAXWLEN;
|
|
|
3276 slang->sl_compmax = c;
|
|
|
3277
|
|
|
3278 --todo;
|
|
|
3279 c = getc(fd); /* <compminlen> */
|
|
|
3280 if (c < 1)
|
|
501
|
3281 c = 0;
|
|
483
|
3282 slang->sl_compminlen = c;
|
|
|
3283
|
|
|
3284 --todo;
|
|
|
3285 c = getc(fd); /* <compsylmax> */
|
|
|
3286 if (c < 1)
|
|
|
3287 c = MAXWLEN;
|
|
|
3288 slang->sl_compsylmax = c;
|
|
|
3289
|
|
|
3290 /* Turn the COMPOUNDFLAGS items into a regexp pattern:
|
|
|
3291 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$".
|
|
495
|
3292 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes.
|
|
|
3293 * Conversion to utf-8 may double the size. */
|
|
|
3294 c = todo * 2 + 7;
|
|
|
3295 #ifdef FEAT_MBYTE
|
|
|
3296 if (enc_utf8)
|
|
|
3297 c += todo * 2;
|
|
|
3298 #endif
|
|
|
3299 pat = alloc((unsigned)c);
|
|
483
|
3300 if (pat == NULL)
|
|
495
|
3301 return SP_OTHERERROR;
|
|
483
|
3302
|
|
485
|
3303 /* We also need a list of all flags that can appear at the start and one
|
|
|
3304 * for all flags. */
|
|
483
|
3305 cp = alloc(todo + 1);
|
|
|
3306 if (cp == NULL)
|
|
|
3307 {
|
|
|
3308 vim_free(pat);
|
|
495
|
3309 return SP_OTHERERROR;
|
|
483
|
3310 }
|
|
|
3311 slang->sl_compstartflags = cp;
|
|
|
3312 *cp = NUL;
|
|
|
3313
|
|
485
|
3314 ap = alloc(todo + 1);
|
|
|
3315 if (ap == NULL)
|
|
|
3316 {
|
|
|
3317 vim_free(pat);
|
|
495
|
3318 return SP_OTHERERROR;
|
|
485
|
3319 }
|
|
|
3320 slang->sl_compallflags = ap;
|
|
|
3321 *ap = NUL;
|
|
|
3322
|
|
483
|
3323 pp = pat;
|
|
|
3324 *pp++ = '^';
|
|
|
3325 *pp++ = '\\';
|
|
|
3326 *pp++ = '(';
|
|
|
3327
|
|
|
3328 atstart = 1;
|
|
|
3329 while (todo-- > 0)
|
|
|
3330 {
|
|
|
3331 c = getc(fd); /* <compflags> */
|
|
485
|
3332
|
|
|
3333 /* Add all flags to "sl_compallflags". */
|
|
|
3334 if (vim_strchr((char_u *)"+*[]/", c) == NULL
|
|
495
|
3335 && !byte_in_str(slang->sl_compallflags, c))
|
|
485
|
3336 {
|
|
|
3337 *ap++ = c;
|
|
|
3338 *ap = NUL;
|
|
|
3339 }
|
|
|
3340
|
|
483
|
3341 if (atstart != 0)
|
|
|
3342 {
|
|
|
3343 /* At start of item: copy flags to "sl_compstartflags". For a
|
|
|
3344 * [abc] item set "atstart" to 2 and copy up to the ']'. */
|
|
|
3345 if (c == '[')
|
|
|
3346 atstart = 2;
|
|
|
3347 else if (c == ']')
|
|
|
3348 atstart = 0;
|
|
|
3349 else
|
|
|
3350 {
|
|
495
|
3351 if (!byte_in_str(slang->sl_compstartflags, c))
|
|
483
|
3352 {
|
|
|
3353 *cp++ = c;
|
|
|
3354 *cp = NUL;
|
|
|
3355 }
|
|
|
3356 if (atstart == 1)
|
|
|
3357 atstart = 0;
|
|
|
3358 }
|
|
|
3359 }
|
|
|
3360 if (c == '/') /* slash separates two items */
|
|
|
3361 {
|
|
|
3362 *pp++ = '\\';
|
|
|
3363 *pp++ = '|';
|
|
|
3364 atstart = 1;
|
|
|
3365 }
|
|
|
3366 else /* normal char, "[abc]" and '*' are copied as-is */
|
|
|
3367 {
|
|
500
|
3368 if (c == '+' || c == '~')
|
|
483
|
3369 *pp++ = '\\'; /* "a+" becomes "a\+" */
|
|
495
|
3370 #ifdef FEAT_MBYTE
|
|
|
3371 if (enc_utf8)
|
|
|
3372 pp += mb_char2bytes(c, pp);
|
|
|
3373 else
|
|
|
3374 #endif
|
|
|
3375 *pp++ = c;
|
|
483
|
3376 }
|
|
|
3377 }
|
|
|
3378
|
|
|
3379 *pp++ = '\\';
|
|
|
3380 *pp++ = ')';
|
|
|
3381 *pp++ = '$';
|
|
|
3382 *pp = NUL;
|
|
|
3383
|
|
|
3384 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
3385 vim_free(pat);
|
|
|
3386 if (slang->sl_compprog == NULL)
|
|
|
3387 return SP_FORMERROR;
|
|
|
3388
|
|
|
3389 return 0;
|
|
|
3390 }
|
|
|
3391
|
|
495
|
3392 /*
|
|
498
|
3393 * Return TRUE if byte "n" appears in "str".
|
|
495
|
3394 * Like strchr() but independent of locale.
|
|
|
3395 */
|
|
|
3396 static int
|
|
498
|
3397 byte_in_str(str, n)
|
|
495
|
3398 char_u *str;
|
|
498
|
3399 int n;
|
|
495
|
3400 {
|
|
|
3401 char_u *p;
|
|
|
3402
|
|
|
3403 for (p = str; *p != NUL; ++p)
|
|
498
|
3404 if (*p == n)
|
|
495
|
3405 return TRUE;
|
|
|
3406 return FALSE;
|
|
|
3407 }
|
|
|
3408
|
|
483
|
3409 #define SY_MAXLEN 30
|
|
|
3410 typedef struct syl_item_S
|
|
|
3411 {
|
|
|
3412 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */
|
|
|
3413 int sy_len;
|
|
|
3414 } syl_item_T;
|
|
|
3415
|
|
|
3416 /*
|
|
|
3417 * Truncate "slang->sl_syllable" at the first slash and put the following items
|
|
|
3418 * in "slang->sl_syl_items".
|
|
|
3419 */
|
|
|
3420 static int
|
|
|
3421 init_syl_tab(slang)
|
|
|
3422 slang_T *slang;
|
|
|
3423 {
|
|
|
3424 char_u *p;
|
|
|
3425 char_u *s;
|
|
|
3426 int l;
|
|
|
3427 syl_item_T *syl;
|
|
|
3428
|
|
|
3429 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4);
|
|
|
3430 p = vim_strchr(slang->sl_syllable, '/');
|
|
|
3431 while (p != NULL)
|
|
|
3432 {
|
|
|
3433 *p++ = NUL;
|
|
495
|
3434 if (*p == NUL) /* trailing slash */
|
|
483
|
3435 break;
|
|
|
3436 s = p;
|
|
|
3437 p = vim_strchr(p, '/');
|
|
|
3438 if (p == NULL)
|
|
|
3439 l = STRLEN(s);
|
|
|
3440 else
|
|
|
3441 l = p - s;
|
|
|
3442 if (l >= SY_MAXLEN)
|
|
|
3443 return SP_FORMERROR;
|
|
|
3444 if (ga_grow(&slang->sl_syl_items, 1) == FAIL)
|
|
495
|
3445 return SP_OTHERERROR;
|
|
483
|
3446 syl = ((syl_item_T *)slang->sl_syl_items.ga_data)
|
|
|
3447 + slang->sl_syl_items.ga_len++;
|
|
|
3448 vim_strncpy(syl->sy_chars, s, l);
|
|
|
3449 syl->sy_len = l;
|
|
|
3450 }
|
|
|
3451 return OK;
|
|
|
3452 }
|
|
|
3453
|
|
|
3454 /*
|
|
|
3455 * Count the number of syllables in "word".
|
|
|
3456 * When "word" contains spaces the syllables after the last space are counted.
|
|
|
3457 * Returns zero if syllables are not defines.
|
|
|
3458 */
|
|
|
3459 static int
|
|
|
3460 count_syllables(slang, word)
|
|
|
3461 slang_T *slang;
|
|
|
3462 char_u *word;
|
|
|
3463 {
|
|
|
3464 int cnt = 0;
|
|
|
3465 int skip = FALSE;
|
|
|
3466 char_u *p;
|
|
|
3467 int len;
|
|
|
3468 int i;
|
|
|
3469 syl_item_T *syl;
|
|
|
3470 int c;
|
|
|
3471
|
|
|
3472 if (slang->sl_syllable == NULL)
|
|
|
3473 return 0;
|
|
|
3474
|
|
|
3475 for (p = word; *p != NUL; p += len)
|
|
|
3476 {
|
|
|
3477 /* When running into a space reset counter. */
|
|
|
3478 if (*p == ' ')
|
|
|
3479 {
|
|
|
3480 len = 1;
|
|
|
3481 cnt = 0;
|
|
|
3482 continue;
|
|
|
3483 }
|
|
|
3484
|
|
|
3485 /* Find longest match of syllable items. */
|
|
|
3486 len = 0;
|
|
|
3487 for (i = 0; i < slang->sl_syl_items.ga_len; ++i)
|
|
|
3488 {
|
|
|
3489 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i;
|
|
|
3490 if (syl->sy_len > len
|
|
|
3491 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0)
|
|
|
3492 len = syl->sy_len;
|
|
|
3493 }
|
|
|
3494 if (len != 0) /* found a match, count syllable */
|
|
|
3495 {
|
|
|
3496 ++cnt;
|
|
|
3497 skip = FALSE;
|
|
|
3498 }
|
|
|
3499 else
|
|
|
3500 {
|
|
|
3501 /* No recognized syllable item, at least a syllable char then? */
|
|
|
3502 #ifdef FEAT_MBYTE
|
|
|
3503 c = mb_ptr2char(p);
|
|
|
3504 len = (*mb_ptr2len)(p);
|
|
|
3505 #else
|
|
|
3506 c = *p;
|
|
|
3507 len = 1;
|
|
|
3508 #endif
|
|
|
3509 if (vim_strchr(slang->sl_syllable, c) == NULL)
|
|
|
3510 skip = FALSE; /* No, search for next syllable */
|
|
|
3511 else if (!skip)
|
|
|
3512 {
|
|
|
3513 ++cnt; /* Yes, count it */
|
|
|
3514 skip = TRUE; /* don't count following syllable chars */
|
|
|
3515 }
|
|
|
3516 }
|
|
|
3517 }
|
|
|
3518 return cnt;
|
|
376
|
3519 }
|
|
|
3520
|
|
381
|
3521 /*
|
|
|
3522 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
483
|
3523 * Returns SP_*ERROR flags when there is something wrong.
|
|
381
|
3524 */
|
|
|
3525 static int
|
|
|
3526 set_sofo(lp, from, to)
|
|
|
3527 slang_T *lp;
|
|
|
3528 char_u *from;
|
|
|
3529 char_u *to;
|
|
|
3530 {
|
|
|
3531 int i;
|
|
|
3532
|
|
|
3533 #ifdef FEAT_MBYTE
|
|
|
3534 garray_T *gap;
|
|
|
3535 char_u *s;
|
|
|
3536 char_u *p;
|
|
|
3537 int c;
|
|
|
3538 int *inp;
|
|
|
3539
|
|
|
3540 if (has_mbyte)
|
|
|
3541 {
|
|
|
3542 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
3543 * characters. The index is the low byte of the character.
|
|
|
3544 * The list contains from-to pairs with a terminating NUL.
|
|
|
3545 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
3546 gap = &lp->sl_sal;
|
|
|
3547 ga_init2(gap, sizeof(int *), 1);
|
|
|
3548 if (ga_grow(gap, 256) == FAIL)
|
|
495
|
3549 return SP_OTHERERROR;
|
|
381
|
3550 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
3551 gap->ga_len = 256;
|
|
|
3552
|
|
|
3553 /* First count the number of items for each list. Temporarily use
|
|
|
3554 * sl_sal_first[] for this. */
|
|
|
3555 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3556 {
|
|
474
|
3557 c = mb_cptr2char_adv(&p);
|
|
|
3558 mb_cptr_adv(s);
|
|
381
|
3559 if (c >= 256)
|
|
|
3560 ++lp->sl_sal_first[c & 0xff];
|
|
|
3561 }
|
|
|
3562 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
483
|
3563 return SP_FORMERROR;
|
|
381
|
3564
|
|
|
3565 /* Allocate the lists. */
|
|
|
3566 for (i = 0; i < 256; ++i)
|
|
|
3567 if (lp->sl_sal_first[i] > 0)
|
|
|
3568 {
|
|
|
3569 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
3570 if (p == NULL)
|
|
495
|
3571 return SP_OTHERERROR;
|
|
381
|
3572 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
3573 *(int *)p = 0;
|
|
|
3574 }
|
|
|
3575
|
|
|
3576 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
3577 * list. */
|
|
|
3578 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
3579 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3580 {
|
|
474
|
3581 c = mb_cptr2char_adv(&p);
|
|
|
3582 i = mb_cptr2char_adv(&s);
|
|
381
|
3583 if (c >= 256)
|
|
|
3584 {
|
|
|
3585 /* Append the from-to chars at the end of the list with
|
|
|
3586 * the low byte. */
|
|
|
3587 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
3588 while (*inp != 0)
|
|
|
3589 ++inp;
|
|
|
3590 *inp++ = c; /* from char */
|
|
|
3591 *inp++ = i; /* to char */
|
|
|
3592 *inp++ = NUL; /* NUL at the end */
|
|
|
3593 }
|
|
|
3594 else
|
|
|
3595 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
3596 lp->sl_sal_first[c] = i;
|
|
|
3597 }
|
|
|
3598 }
|
|
|
3599 else
|
|
|
3600 #endif
|
|
|
3601 {
|
|
|
3602 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
3603 if (STRLEN(from) != STRLEN(to))
|
|
483
|
3604 return SP_FORMERROR;
|
|
381
|
3605
|
|
|
3606 for (i = 0; to[i] != NUL; ++i)
|
|
|
3607 lp->sl_sal_first[from[i]] = to[i];
|
|
|
3608 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
3609 }
|
|
|
3610
|
|
483
|
3611 return 0;
|
|
381
|
3612 }
|
|
|
3613
|
|
|
3614 /*
|
|
|
3615 * Fill the first-index table for "lp".
|
|
|
3616 */
|
|
|
3617 static void
|
|
|
3618 set_sal_first(lp)
|
|
|
3619 slang_T *lp;
|
|
|
3620 {
|
|
|
3621 salfirst_T *sfirst;
|
|
|
3622 int i;
|
|
|
3623 salitem_T *smp;
|
|
|
3624 int c;
|
|
|
3625 garray_T *gap = &lp->sl_sal;
|
|
|
3626
|
|
|
3627 sfirst = lp->sl_sal_first;
|
|
|
3628 for (i = 0; i < 256; ++i)
|
|
|
3629 sfirst[i] = -1;
|
|
|
3630 smp = (salitem_T *)gap->ga_data;
|
|
|
3631 for (i = 0; i < gap->ga_len; ++i)
|
|
|
3632 {
|
|
|
3633 #ifdef FEAT_MBYTE
|
|
|
3634 if (has_mbyte)
|
|
|
3635 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
3636 * the character, for other encodings it should differ for most
|
|
|
3637 * characters. */
|
|
|
3638 c = *smp[i].sm_lead_w & 0xff;
|
|
|
3639 else
|
|
|
3640 #endif
|
|
|
3641 c = *smp[i].sm_lead;
|
|
|
3642 if (sfirst[c] == -1)
|
|
|
3643 {
|
|
|
3644 sfirst[c] = i;
|
|
|
3645 #ifdef FEAT_MBYTE
|
|
|
3646 if (has_mbyte)
|
|
|
3647 {
|
|
|
3648 int n;
|
|
|
3649
|
|
|
3650 /* Make sure all entries with this byte are following each
|
|
|
3651 * other. Move the ones that are in the wrong position. Do
|
|
|
3652 * keep the same ordering! */
|
|
|
3653 while (i + 1 < gap->ga_len
|
|
|
3654 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
3655 /* Skip over entry with same index byte. */
|
|
|
3656 ++i;
|
|
|
3657
|
|
|
3658 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
3659 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
3660 {
|
|
|
3661 salitem_T tsal;
|
|
|
3662
|
|
|
3663 /* Move entry with same index byte after the entries
|
|
|
3664 * we already found. */
|
|
|
3665 ++i;
|
|
|
3666 --n;
|
|
|
3667 tsal = smp[i + n];
|
|
|
3668 mch_memmove(smp + i + 1, smp + i,
|
|
|
3669 sizeof(salitem_T) * n);
|
|
|
3670 smp[i] = tsal;
|
|
|
3671 }
|
|
|
3672 }
|
|
|
3673 #endif
|
|
|
3674 }
|
|
|
3675 }
|
|
|
3676 }
|
|
376
|
3677
|
|
372
|
3678 #ifdef FEAT_MBYTE
|
|
|
3679 /*
|
|
|
3680 * Turn a multi-byte string into a wide character string.
|
|
|
3681 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
3682 */
|
|
|
3683 static int *
|
|
|
3684 mb_str2wide(s)
|
|
|
3685 char_u *s;
|
|
|
3686 {
|
|
|
3687 int *res;
|
|
|
3688 char_u *p;
|
|
|
3689 int i = 0;
|
|
|
3690
|
|
|
3691 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
3692 if (res != NULL)
|
|
|
3693 {
|
|
|
3694 for (p = s; *p != NUL; )
|
|
|
3695 res[i++] = mb_ptr2char_adv(&p);
|
|
|
3696 res[i] = NUL;
|
|
|
3697 }
|
|
|
3698 return res;
|
|
|
3699 }
|
|
|
3700 #endif
|
|
|
3701
|
|
236
|
3702 /*
|
|
625
|
3703 * Read a tree from the .spl or .sug file.
|
|
|
3704 * Allocates the memory and stores pointers in "bytsp" and "idxsp".
|
|
|
3705 * This is skipped when the tree has zero length.
|
|
|
3706 * Returns zero when OK, SP_ value for an error.
|
|
|
3707 */
|
|
|
3708 static int
|
|
|
3709 spell_read_tree(fd, bytsp, idxsp, prefixtree, prefixcnt)
|
|
|
3710 FILE *fd;
|
|
|
3711 char_u **bytsp;
|
|
|
3712 idx_T **idxsp;
|
|
|
3713 int prefixtree; /* TRUE for the prefix tree */
|
|
|
3714 int prefixcnt; /* when "prefixtree" is TRUE: prefix count */
|
|
|
3715 {
|
|
|
3716 int len;
|
|
|
3717 int idx;
|
|
|
3718 char_u *bp;
|
|
|
3719 idx_T *ip;
|
|
|
3720
|
|
|
3721 /* The tree size was computed when writing the file, so that we can
|
|
|
3722 * allocate it as one long block. <nodecount> */
|
|
|
3723 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
3724 if (len < 0)
|
|
|
3725 return SP_TRUNCERROR;
|
|
|
3726 if (len > 0)
|
|
|
3727 {
|
|
|
3728 /* Allocate the byte array. */
|
|
|
3729 bp = lalloc((long_u)len, TRUE);
|
|
|
3730 if (bp == NULL)
|
|
|
3731 return SP_OTHERERROR;
|
|
|
3732 *bytsp = bp;
|
|
|
3733
|
|
|
3734 /* Allocate the index array. */
|
|
|
3735 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
3736 if (ip == NULL)
|
|
|
3737 return SP_OTHERERROR;
|
|
|
3738 *idxsp = ip;
|
|
|
3739
|
|
|
3740 /* Recursively read the tree and store it in the array. */
|
|
|
3741 idx = read_tree_node(fd, bp, ip, len, 0, prefixtree, prefixcnt);
|
|
|
3742 if (idx < 0)
|
|
|
3743 return idx;
|
|
|
3744 }
|
|
|
3745 return 0;
|
|
|
3746 }
|
|
|
3747
|
|
|
3748 /*
|
|
300
|
3749 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
3750 * "byts" and index array "idxs". Recursively read the children.
|
|
|
3751 *
|
|
625
|
3752 * NOTE: The code here must match put_node()!
|
|
|
3753 *
|
|
|
3754 * Returns the index (>= 0) following the siblings.
|
|
|
3755 * Returns SP_TRUNCERROR if the file is shorter than expected.
|
|
|
3756 * Returns SP_FORMERROR if there is a format error.
|
|
236
|
3757 */
|
|
324
|
3758 static idx_T
|
|
625
|
3759 read_tree_node(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
3760 FILE *fd;
|
|
|
3761 char_u *byts;
|
|
324
|
3762 idx_T *idxs;
|
|
300
|
3763 int maxidx; /* size of arrays */
|
|
324
|
3764 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
3765 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
3766 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
3767 {
|
|
300
|
3768 int len;
|
|
|
3769 int i;
|
|
|
3770 int n;
|
|
324
|
3771 idx_T idx = startidx;
|
|
300
|
3772 int c;
|
|
366
|
3773 int c2;
|
|
300
|
3774 #define SHARED_MASK 0x8000000
|
|
236
|
3775
|
|
300
|
3776 len = getc(fd); /* <siblingcount> */
|
|
|
3777 if (len <= 0)
|
|
625
|
3778 return SP_TRUNCERROR;
|
|
300
|
3779
|
|
|
3780 if (startidx + len >= maxidx)
|
|
625
|
3781 return SP_FORMERROR;
|
|
300
|
3782 byts[idx++] = len;
|
|
|
3783
|
|
|
3784 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
3785 for (i = 1; i <= len; ++i)
|
|
236
|
3786 {
|
|
300
|
3787 c = getc(fd); /* <byte> */
|
|
|
3788 if (c < 0)
|
|
625
|
3789 return SP_TRUNCERROR;
|
|
300
|
3790 if (c <= BY_SPECIAL)
|
|
|
3791 {
|
|
366
|
3792 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
3793 {
|
|
|
3794 /* No flags, all regions. */
|
|
|
3795 idxs[idx] = 0;
|
|
|
3796 c = 0;
|
|
|
3797 }
|
|
390
|
3798 else if (c != BY_INDEX)
|
|
300
|
3799 {
|
|
339
|
3800 if (prefixtree)
|
|
|
3801 {
|
|
455
|
3802 /* Read the optional pflags byte, the prefix ID and the
|
|
|
3803 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
3804 * byte, the condition index shifted up 8 bits, the flags
|
|
|
3805 * shifted up 24 bits. */
|
|
|
3806 if (c == BY_FLAGS)
|
|
|
3807 c = getc(fd) << 24; /* <pflags> */
|
|
|
3808 else
|
|
|
3809 c = 0;
|
|
|
3810
|
|
481
|
3811 c |= getc(fd); /* <affixID> */
|
|
455
|
3812
|
|
339
|
3813 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
3814 if (n >= maxprefcondnr)
|
|
625
|
3815 return SP_FORMERROR;
|
|
455
|
3816 c |= (n << 8);
|
|
339
|
3817 }
|
|
390
|
3818 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
3819 {
|
|
|
3820 /* Read flags and optional region and prefix ID. In
|
|
390
|
3821 * idxs[] the flags go in the low two bytes, region above
|
|
|
3822 * that and prefix ID above the region. */
|
|
|
3823 c2 = c;
|
|
339
|
3824 c = getc(fd); /* <flags> */
|
|
390
|
3825 if (c2 == BY_FLAGS2)
|
|
|
3826 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
3827 if (c & WF_REGION)
|
|
390
|
3828 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
3829 if (c & WF_AFX)
|
|
|
3830 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
3831 }
|
|
|
3832
|
|
300
|
3833 idxs[idx] = c;
|
|
|
3834 c = 0;
|
|
|
3835 }
|
|
|
3836 else /* c == BY_INDEX */
|
|
|
3837 {
|
|
|
3838 /* <nodeidx> */
|
|
|
3839 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
3840 if (n < 0 || n >= maxidx)
|
|
625
|
3841 return SP_FORMERROR;
|
|
300
|
3842 idxs[idx] = n + SHARED_MASK;
|
|
|
3843 c = getc(fd); /* <xbyte> */
|
|
|
3844 }
|
|
|
3845 }
|
|
|
3846 byts[idx++] = c;
|
|
236
|
3847 }
|
|
|
3848
|
|
300
|
3849 /* Recursively read the children for non-shared siblings.
|
|
|
3850 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
3851 * remove SHARED_MASK) */
|
|
|
3852 for (i = 1; i <= len; ++i)
|
|
|
3853 if (byts[startidx + i] != 0)
|
|
|
3854 {
|
|
|
3855 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
3856 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
3857 else
|
|
|
3858 {
|
|
|
3859 idxs[startidx + i] = idx;
|
|
625
|
3860 idx = read_tree_node(fd, byts, idxs, maxidx, idx,
|
|
339
|
3861 prefixtree, maxprefcondnr);
|
|
300
|
3862 if (idx < 0)
|
|
|
3863 break;
|
|
|
3864 }
|
|
|
3865 }
|
|
236
|
3866
|
|
300
|
3867 return idx;
|
|
236
|
3868 }
|
|
|
3869
|
|
|
3870 /*
|
|
|
3871 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
3872 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
3873 */
|
|
|
3874 char_u *
|
|
|
3875 did_set_spelllang(buf)
|
|
|
3876 buf_T *buf;
|
|
|
3877 {
|
|
|
3878 garray_T ga;
|
|
351
|
3879 char_u *splp;
|
|
236
|
3880 char_u *region;
|
|
416
|
3881 char_u region_cp[3];
|
|
355
|
3882 int filename;
|
|
236
|
3883 int region_mask;
|
|
503
|
3884 slang_T *slang;
|
|
236
|
3885 int c;
|
|
351
|
3886 char_u lang[MAXWLEN + 1];
|
|
323
|
3887 char_u spf_name[MAXPATHL];
|
|
351
|
3888 int len;
|
|
|
3889 char_u *p;
|
|
381
|
3890 int round;
|
|
385
|
3891 char_u *spf;
|
|
389
|
3892 char_u *use_region = NULL;
|
|
|
3893 int dont_use_region = FALSE;
|
|
501
|
3894 int nobreak = FALSE;
|
|
503
|
3895 int i, j;
|
|
|
3896 langp_T *lp, *lp2;
|
|
236
|
3897
|
|
|
3898 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
3899 clear_midword(buf);
|
|
236
|
3900
|
|
351
|
3901 /* loop over comma separated language names. */
|
|
|
3902 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
3903 {
|
|
|
3904 /* Get one language name. */
|
|
|
3905 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
3906
|
|
240
|
3907 region = NULL;
|
|
351
|
3908 len = STRLEN(lang);
|
|
355
|
3909
|
|
|
3910 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
3911 * If there is a region name let "region" point to it and remove it
|
|
|
3912 * from the name. */
|
|
|
3913 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
3914 {
|
|
355
|
3915 filename = TRUE;
|
|
|
3916
|
|
416
|
3917 /* Locate a region and remove it from the file name. */
|
|
|
3918 p = vim_strchr(gettail(lang), '_');
|
|
|
3919 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
3920 && !ASCII_ISALPHA(p[3]))
|
|
|
3921 {
|
|
|
3922 vim_strncpy(region_cp, p + 1, 2);
|
|
|
3923 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
3924 len -= 3;
|
|
|
3925 region = region_cp;
|
|
|
3926 }
|
|
|
3927 else
|
|
|
3928 dont_use_region = TRUE;
|
|
|
3929
|
|
355
|
3930 /* Check if we loaded this language before. */
|
|
503
|
3931 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3932 if (fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME)
|
|
355
|
3933 break;
|
|
236
|
3934 }
|
|
355
|
3935 else
|
|
|
3936 {
|
|
|
3937 filename = FALSE;
|
|
|
3938 if (len > 3 && lang[len - 3] == '_')
|
|
|
3939 {
|
|
|
3940 region = lang + len - 2;
|
|
|
3941 len -= 3;
|
|
|
3942 lang[len] = NUL;
|
|
389
|
3943 }
|
|
|
3944 else
|
|
|
3945 dont_use_region = TRUE;
|
|
355
|
3946
|
|
|
3947 /* Check if we loaded this language before. */
|
|
503
|
3948 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3949 if (STRICMP(lang, slang->sl_name) == 0)
|
|
355
|
3950 break;
|
|
|
3951 }
|
|
236
|
3952
|
|
416
|
3953 if (region != NULL)
|
|
|
3954 {
|
|
|
3955 /* If the region differs from what was used before then don't
|
|
|
3956 * use it for 'spellfile'. */
|
|
|
3957 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
3958 dont_use_region = TRUE;
|
|
|
3959 use_region = region;
|
|
|
3960 }
|
|
|
3961
|
|
351
|
3962 /* If not found try loading the language now. */
|
|
503
|
3963 if (slang == NULL)
|
|
355
|
3964 {
|
|
|
3965 if (filename)
|
|
|
3966 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
3967 else
|
|
|
3968 spell_load_lang(lang);
|
|
|
3969 }
|
|
236
|
3970
|
|
307
|
3971 /*
|
|
351
|
3972 * Loop over the languages, there can be several files for "lang".
|
|
307
|
3973 */
|
|
503
|
3974 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3975 if (filename ? fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME
|
|
|
3976 : STRICMP(lang, slang->sl_name) == 0)
|
|
236
|
3977 {
|
|
316
|
3978 region_mask = REGION_ALL;
|
|
355
|
3979 if (!filename && region != NULL)
|
|
236
|
3980 {
|
|
307
|
3981 /* find region in sl_regions */
|
|
503
|
3982 c = find_region(slang->sl_regions, region);
|
|
307
|
3983 if (c == REGION_ALL)
|
|
|
3984 {
|
|
503
|
3985 if (slang->sl_add)
|
|
389
|
3986 {
|
|
503
|
3987 if (*slang->sl_regions != NUL)
|
|
389
|
3988 /* This addition file is for other regions. */
|
|
|
3989 region_mask = 0;
|
|
|
3990 }
|
|
|
3991 else
|
|
|
3992 /* This is probably an error. Give a warning and
|
|
|
3993 * accept the words anyway. */
|
|
351
|
3994 smsg((char_u *)
|
|
|
3995 _("Warning: region %s not supported"),
|
|
|
3996 region);
|
|
307
|
3997 }
|
|
|
3998 else
|
|
|
3999 region_mask = 1 << c;
|
|
236
|
4000 }
|
|
307
|
4001
|
|
389
|
4002 if (region_mask != 0)
|
|
307
|
4003 {
|
|
389
|
4004 if (ga_grow(&ga, 1) == FAIL)
|
|
|
4005 {
|
|
|
4006 ga_clear(&ga);
|
|
|
4007 return e_outofmem;
|
|
|
4008 }
|
|
503
|
4009 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang;
|
|
389
|
4010 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
4011 ++ga.ga_len;
|
|
503
|
4012 use_midword(slang, buf);
|
|
|
4013 if (slang->sl_nobreak)
|
|
501
|
4014 nobreak = TRUE;
|
|
307
|
4015 }
|
|
385
|
4016 }
|
|
|
4017 }
|
|
|
4018
|
|
|
4019 /* round 0: load int_wordlist, if possible.
|
|
|
4020 * round 1: load first name in 'spellfile'.
|
|
|
4021 * round 2: load second name in 'spellfile.
|
|
|
4022 * etc. */
|
|
|
4023 spf = curbuf->b_p_spf;
|
|
|
4024 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
4025 {
|
|
|
4026 if (round == 0)
|
|
|
4027 {
|
|
|
4028 /* Internal wordlist, if there is one. */
|
|
|
4029 if (int_wordlist == NULL)
|
|
381
|
4030 continue;
|
|
385
|
4031 int_wordlist_spl(spf_name);
|
|
381
|
4032 }
|
|
|
4033 else
|
|
|
4034 {
|
|
385
|
4035 /* One entry in 'spellfile'. */
|
|
|
4036 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
4037 STRCAT(spf_name, ".spl");
|
|
|
4038
|
|
|
4039 /* If it was already found above then skip it. */
|
|
|
4040 for (c = 0; c < ga.ga_len; ++c)
|
|
500
|
4041 {
|
|
|
4042 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname;
|
|
|
4043 if (p != NULL && fullpathcmp(spf_name, p, FALSE) == FPC_SAME)
|
|
385
|
4044 break;
|
|
500
|
4045 }
|
|
385
|
4046 if (c < ga.ga_len)
|
|
381
|
4047 continue;
|
|
|
4048 }
|
|
|
4049
|
|
351
|
4050 /* Check if it was loaded already. */
|
|
503
|
4051 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
4052 if (fullpathcmp(spf_name, slang->sl_fname, FALSE) == FPC_SAME)
|
|
323
|
4053 break;
|
|
503
|
4054 if (slang == NULL)
|
|
323
|
4055 {
|
|
351
|
4056 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
4057 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
4058 * use an arbitrary name. */
|
|
|
4059 if (round == 0)
|
|
|
4060 STRCPY(lang, "internal wordlist");
|
|
|
4061 else
|
|
|
4062 {
|
|
|
4063 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
4064 p = vim_strchr(lang, '.');
|
|
|
4065 if (p != NULL)
|
|
|
4066 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
4067 }
|
|
503
|
4068 slang = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
501
|
4069
|
|
|
4070 /* If one of the languages has NOBREAK we assume the addition
|
|
|
4071 * files also have this. */
|
|
503
|
4072 if (slang != NULL && nobreak)
|
|
|
4073 slang->sl_nobreak = TRUE;
|
|
|
4074 }
|
|
|
4075 if (slang != NULL && ga_grow(&ga, 1) == OK)
|
|
323
|
4076 {
|
|
389
|
4077 region_mask = REGION_ALL;
|
|
|
4078 if (use_region != NULL && !dont_use_region)
|
|
|
4079 {
|
|
|
4080 /* find region in sl_regions */
|
|
503
|
4081 c = find_region(slang->sl_regions, use_region);
|
|
389
|
4082 if (c != REGION_ALL)
|
|
|
4083 region_mask = 1 << c;
|
|
503
|
4084 else if (*slang->sl_regions != NUL)
|
|
389
|
4085 /* This spell file is for other regions. */
|
|
|
4086 region_mask = 0;
|
|
|
4087 }
|
|
|
4088
|
|
|
4089 if (region_mask != 0)
|
|
|
4090 {
|
|
503
|
4091 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang;
|
|
|
4092 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL;
|
|
|
4093 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL;
|
|
389
|
4094 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
4095 ++ga.ga_len;
|
|
503
|
4096 use_midword(slang, buf);
|
|
389
|
4097 }
|
|
323
|
4098 }
|
|
|
4099 }
|
|
|
4100
|
|
236
|
4101 /* Everything is fine, store the new b_langp value. */
|
|
|
4102 ga_clear(&buf->b_langp);
|
|
|
4103 buf->b_langp = ga;
|
|
|
4104
|
|
503
|
4105 /* For each language figure out what language to use for sound folding and
|
|
|
4106 * REP items. If the language doesn't support it itself use another one
|
|
|
4107 * with the same name. E.g. for "en-math" use "en". */
|
|
|
4108 for (i = 0; i < ga.ga_len; ++i)
|
|
|
4109 {
|
|
|
4110 lp = LANGP_ENTRY(ga, i);
|
|
|
4111
|
|
|
4112 /* sound folding */
|
|
|
4113 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
4114 /* language does sound folding itself */
|
|
|
4115 lp->lp_sallang = lp->lp_slang;
|
|
|
4116 else
|
|
|
4117 /* find first similar language that does sound folding */
|
|
|
4118 for (j = 0; j < ga.ga_len; ++j)
|
|
|
4119 {
|
|
|
4120 lp2 = LANGP_ENTRY(ga, j);
|
|
|
4121 if (lp2->lp_slang->sl_sal.ga_len > 0
|
|
|
4122 && STRNCMP(lp->lp_slang->sl_name,
|
|
|
4123 lp2->lp_slang->sl_name, 2) == 0)
|
|
|
4124 {
|
|
|
4125 lp->lp_sallang = lp2->lp_slang;
|
|
|
4126 break;
|
|
|
4127 }
|
|
|
4128 }
|
|
|
4129
|
|
|
4130 /* REP items */
|
|
|
4131 if (lp->lp_slang->sl_rep.ga_len > 0)
|
|
|
4132 /* language has REP items itself */
|
|
|
4133 lp->lp_replang = lp->lp_slang;
|
|
|
4134 else
|
|
625
|
4135 /* find first similar language that has REP items */
|
|
503
|
4136 for (j = 0; j < ga.ga_len; ++j)
|
|
|
4137 {
|
|
|
4138 lp2 = LANGP_ENTRY(ga, j);
|
|
|
4139 if (lp2->lp_slang->sl_rep.ga_len > 0
|
|
|
4140 && STRNCMP(lp->lp_slang->sl_name,
|
|
|
4141 lp2->lp_slang->sl_name, 2) == 0)
|
|
|
4142 {
|
|
|
4143 lp->lp_replang = lp2->lp_slang;
|
|
|
4144 break;
|
|
|
4145 }
|
|
|
4146 }
|
|
|
4147 }
|
|
|
4148
|
|
236
|
4149 return NULL;
|
|
|
4150 }
|
|
|
4151
|
|
|
4152 /*
|
|
376
|
4153 * Clear the midword characters for buffer "buf".
|
|
|
4154 */
|
|
|
4155 static void
|
|
|
4156 clear_midword(buf)
|
|
|
4157 buf_T *buf;
|
|
|
4158 {
|
|
|
4159 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
4160 #ifdef FEAT_MBYTE
|
|
|
4161 vim_free(buf->b_spell_ismw_mb);
|
|
|
4162 buf->b_spell_ismw_mb = NULL;
|
|
|
4163 #endif
|
|
|
4164 }
|
|
|
4165
|
|
|
4166 /*
|
|
|
4167 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
4168 * They add up to any currently used midword characters.
|
|
|
4169 */
|
|
|
4170 static void
|
|
|
4171 use_midword(lp, buf)
|
|
|
4172 slang_T *lp;
|
|
|
4173 buf_T *buf;
|
|
|
4174 {
|
|
|
4175 char_u *p;
|
|
|
4176
|
|
389
|
4177 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
4178 return;
|
|
|
4179
|
|
376
|
4180 for (p = lp->sl_midword; *p != NUL; )
|
|
|
4181 #ifdef FEAT_MBYTE
|
|
|
4182 if (has_mbyte)
|
|
|
4183 {
|
|
|
4184 int c, l, n;
|
|
|
4185 char_u *bp;
|
|
|
4186
|
|
|
4187 c = mb_ptr2char(p);
|
|
474
|
4188 l = (*mb_ptr2len)(p);
|
|
|
4189 if (c < 256 && l <= 2)
|
|
376
|
4190 buf->b_spell_ismw[c] = TRUE;
|
|
|
4191 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
4192 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
4193 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
4194 else
|
|
|
4195 {
|
|
|
4196 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
4197 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
4198 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
4199 if (bp != NULL)
|
|
|
4200 {
|
|
|
4201 vim_free(buf->b_spell_ismw_mb);
|
|
|
4202 buf->b_spell_ismw_mb = bp;
|
|
|
4203 vim_strncpy(bp + n, p, l);
|
|
|
4204 }
|
|
|
4205 }
|
|
|
4206 p += l;
|
|
|
4207 }
|
|
|
4208 else
|
|
|
4209 #endif
|
|
|
4210 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
4211 }
|
|
|
4212
|
|
|
4213 /*
|
|
236
|
4214 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
4215 * Each region is simply stored as the two characters of it's name.
|
|
381
|
4216 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
4217 */
|
|
|
4218 static int
|
|
|
4219 find_region(rp, region)
|
|
|
4220 char_u *rp;
|
|
|
4221 char_u *region;
|
|
|
4222 {
|
|
|
4223 int i;
|
|
|
4224
|
|
|
4225 for (i = 0; ; i += 2)
|
|
|
4226 {
|
|
|
4227 if (rp[i] == NUL)
|
|
|
4228 return REGION_ALL;
|
|
|
4229 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
4230 break;
|
|
|
4231 }
|
|
|
4232 return i / 2;
|
|
|
4233 }
|
|
|
4234
|
|
|
4235 /*
|
|
323
|
4236 * Return case type of word:
|
|
236
|
4237 * w word 0
|
|
300
|
4238 * Word WF_ONECAP
|
|
|
4239 * W WORD WF_ALLCAP
|
|
|
4240 * WoRd wOrd WF_KEEPCAP
|
|
236
|
4241 */
|
|
|
4242 static int
|
|
|
4243 captype(word, end)
|
|
|
4244 char_u *word;
|
|
323
|
4245 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
4246 {
|
|
|
4247 char_u *p;
|
|
|
4248 int c;
|
|
|
4249 int firstcap;
|
|
|
4250 int allcap;
|
|
|
4251 int past_second = FALSE; /* past second word char */
|
|
|
4252
|
|
|
4253 /* find first letter */
|
|
376
|
4254 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
4255 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
4256 return 0; /* only non-word characters, illegal word */
|
|
|
4257 #ifdef FEAT_MBYTE
|
|
310
|
4258 if (has_mbyte)
|
|
|
4259 c = mb_ptr2char_adv(&p);
|
|
|
4260 else
|
|
236
|
4261 #endif
|
|
310
|
4262 c = *p++;
|
|
324
|
4263 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
4264
|
|
|
4265 /*
|
|
|
4266 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
4267 * But a word with an upper char only at start is a ONECAP.
|
|
|
4268 */
|
|
323
|
4269 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
4270 if (spell_iswordp_nmw(p))
|
|
236
|
4271 {
|
|
455
|
4272 c = PTR2CHAR(p);
|
|
324
|
4273 if (!SPELL_ISUPPER(c))
|
|
236
|
4274 {
|
|
|
4275 /* UUl -> KEEPCAP */
|
|
|
4276 if (past_second && allcap)
|
|
300
|
4277 return WF_KEEPCAP;
|
|
236
|
4278 allcap = FALSE;
|
|
|
4279 }
|
|
|
4280 else if (!allcap)
|
|
|
4281 /* UlU -> KEEPCAP */
|
|
300
|
4282 return WF_KEEPCAP;
|
|
236
|
4283 past_second = TRUE;
|
|
|
4284 }
|
|
|
4285
|
|
|
4286 if (allcap)
|
|
300
|
4287 return WF_ALLCAP;
|
|
236
|
4288 if (firstcap)
|
|
300
|
4289 return WF_ONECAP;
|
|
236
|
4290 return 0;
|
|
|
4291 }
|
|
|
4292
|
|
474
|
4293 /*
|
|
|
4294 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
4295 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
4296 * Add ALLCAP for "WOrD".
|
|
|
4297 */
|
|
|
4298 static int
|
|
|
4299 badword_captype(word, end)
|
|
|
4300 char_u *word;
|
|
|
4301 char_u *end;
|
|
|
4302 {
|
|
|
4303 int flags = captype(word, end);
|
|
475
|
4304 int c;
|
|
474
|
4305 int l, u;
|
|
|
4306 int first;
|
|
|
4307 char_u *p;
|
|
|
4308
|
|
|
4309 if (flags & WF_KEEPCAP)
|
|
|
4310 {
|
|
|
4311 /* Count the number of UPPER and lower case letters. */
|
|
|
4312 l = u = 0;
|
|
|
4313 first = FALSE;
|
|
|
4314 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
4315 {
|
|
475
|
4316 c = PTR2CHAR(p);
|
|
|
4317 if (SPELL_ISUPPER(c))
|
|
474
|
4318 {
|
|
|
4319 ++u;
|
|
|
4320 if (p == word)
|
|
|
4321 first = TRUE;
|
|
|
4322 }
|
|
|
4323 else
|
|
|
4324 ++l;
|
|
|
4325 }
|
|
|
4326
|
|
|
4327 /* If there are more UPPER than lower case letters suggest an
|
|
|
4328 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
4329 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
4330 * require three upper case letters. */
|
|
|
4331 if (u > l && u > 2)
|
|
|
4332 flags |= WF_ALLCAP;
|
|
|
4333 else if (first)
|
|
|
4334 flags |= WF_ONECAP;
|
|
633
|
4335
|
|
|
4336 if (u >= 2 && l >= 2) /* maCARONI maCAroni */
|
|
|
4337 flags |= WF_MIXCAP;
|
|
474
|
4338 }
|
|
|
4339 return flags;
|
|
|
4340 }
|
|
|
4341
|
|
355
|
4342 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
4343 /*
|
|
|
4344 * Free all languages.
|
|
|
4345 */
|
|
|
4346 void
|
|
|
4347 spell_free_all()
|
|
|
4348 {
|
|
503
|
4349 slang_T *slang;
|
|
355
|
4350 buf_T *buf;
|
|
385
|
4351 char_u fname[MAXPATHL];
|
|
355
|
4352
|
|
|
4353 /* Go through all buffers and handle 'spelllang'. */
|
|
|
4354 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
4355 ga_clear(&buf->b_langp);
|
|
|
4356
|
|
|
4357 while (first_lang != NULL)
|
|
|
4358 {
|
|
503
|
4359 slang = first_lang;
|
|
|
4360 first_lang = slang->sl_next;
|
|
|
4361 slang_free(slang);
|
|
355
|
4362 }
|
|
366
|
4363
|
|
385
|
4364 if (int_wordlist != NULL)
|
|
|
4365 {
|
|
|
4366 /* Delete the internal wordlist and its .spl file */
|
|
|
4367 mch_remove(int_wordlist);
|
|
|
4368 int_wordlist_spl(fname);
|
|
|
4369 mch_remove(fname);
|
|
|
4370 vim_free(int_wordlist);
|
|
|
4371 int_wordlist = NULL;
|
|
381
|
4372 }
|
|
|
4373
|
|
366
|
4374 init_spell_chartab();
|
|
626
|
4375
|
|
|
4376 vim_free(repl_to);
|
|
|
4377 repl_to = NULL;
|
|
|
4378 vim_free(repl_from);
|
|
|
4379 repl_from = NULL;
|
|
355
|
4380 }
|
|
|
4381 # endif
|
|
|
4382
|
|
236
|
4383 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
4384 /*
|
|
|
4385 * Clear all spelling tables and reload them.
|
|
307
|
4386 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
4387 */
|
|
|
4388 void
|
|
|
4389 spell_reload()
|
|
|
4390 {
|
|
|
4391 buf_T *buf;
|
|
316
|
4392 win_T *wp;
|
|
236
|
4393
|
|
358
|
4394 /* Initialize the table for spell_iswordp(). */
|
|
236
|
4395 init_spell_chartab();
|
|
|
4396
|
|
|
4397 /* Unload all allocated memory. */
|
|
355
|
4398 spell_free_all();
|
|
236
|
4399
|
|
|
4400 /* Go through all buffers and handle 'spelllang'. */
|
|
|
4401 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
4402 {
|
|
316
|
4403 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
4404 * window for this buffer in which 'spell' is set. */
|
|
236
|
4405 if (*buf->b_p_spl != NUL)
|
|
316
|
4406 {
|
|
|
4407 FOR_ALL_WINDOWS(wp)
|
|
|
4408 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
4409 {
|
|
|
4410 (void)did_set_spelllang(buf);
|
|
|
4411 # ifdef FEAT_WINDOWS
|
|
|
4412 break;
|
|
|
4413 # endif
|
|
|
4414 }
|
|
|
4415 }
|
|
236
|
4416 }
|
|
|
4417 }
|
|
|
4418 # endif
|
|
|
4419
|
|
310
|
4420 /*
|
|
|
4421 * Reload the spell file "fname" if it's loaded.
|
|
|
4422 */
|
|
|
4423 static void
|
|
323
|
4424 spell_reload_one(fname, added_word)
|
|
310
|
4425 char_u *fname;
|
|
323
|
4426 int added_word; /* invoked through "zg" */
|
|
310
|
4427 {
|
|
503
|
4428 slang_T *slang;
|
|
323
|
4429 int didit = FALSE;
|
|
310
|
4430
|
|
503
|
4431 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
4432 {
|
|
|
4433 if (fullpathcmp(fname, slang->sl_fname, FALSE) == FPC_SAME)
|
|
|
4434 {
|
|
|
4435 slang_clear(slang);
|
|
|
4436 if (spell_load_file(fname, NULL, slang, FALSE) == NULL)
|
|
500
|
4437 /* reloading failed, clear the language */
|
|
503
|
4438 slang_clear(slang);
|
|
310
|
4439 redraw_all_later(NOT_VALID);
|
|
323
|
4440 didit = TRUE;
|
|
310
|
4441 }
|
|
500
|
4442 }
|
|
323
|
4443
|
|
|
4444 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
626
|
4445 * 'spelllang' to load it now. */
|
|
323
|
4446 if (added_word && !didit)
|
|
|
4447 did_set_spelllang(curbuf);
|
|
310
|
4448 }
|
|
|
4449
|
|
|
4450
|
|
236
|
4451 /*
|
|
|
4452 * Functions for ":mkspell".
|
|
|
4453 */
|
|
|
4454
|
|
300
|
4455 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
4456 and .dic file. */
|
|
|
4457 /*
|
|
|
4458 * Main structure to store the contents of a ".aff" file.
|
|
|
4459 */
|
|
|
4460 typedef struct afffile_S
|
|
|
4461 {
|
|
|
4462 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
498
|
4463 int af_flagtype; /* AFT_CHAR, AFT_LONG, AFT_NUM or AFT_CAPLONG */
|
|
621
|
4464 unsigned af_rare; /* RARE ID for rare word */
|
|
|
4465 unsigned af_keepcase; /* KEEPCASE ID for keep-case word */
|
|
495
|
4466 unsigned af_bad; /* BAD ID for banned word */
|
|
|
4467 unsigned af_needaffix; /* NEEDAFFIX ID */
|
|
500
|
4468 unsigned af_needcomp; /* NEEDCOMPOUND ID */
|
|
339
|
4469 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
4470 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
4471 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
495
|
4472 hashtab_T af_comp; /* hashtable for compound flags, compitem_T */
|
|
236
|
4473 } afffile_T;
|
|
|
4474
|
|
495
|
4475 #define AFT_CHAR 0 /* flags are one character */
|
|
498
|
4476 #define AFT_LONG 1 /* flags are two characters */
|
|
|
4477 #define AFT_CAPLONG 2 /* flags are one or two characters */
|
|
|
4478 #define AFT_NUM 3 /* flags are numbers, comma separated */
|
|
495
|
4479
|
|
236
|
4480 typedef struct affentry_S affentry_T;
|
|
|
4481 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
4482 struct affentry_S
|
|
|
4483 {
|
|
|
4484 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
4485 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
4486 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
4487 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
4488 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
483
|
4489 char_u ae_rare; /* rare affix */
|
|
|
4490 char_u ae_nocomp; /* word with affix not compoundable */
|
|
300
|
4491 };
|
|
|
4492
|
|
495
|
4493 #ifdef FEAT_MBYTE
|
|
|
4494 # define AH_KEY_LEN 17 /* 2 x 8 bytes + NUL */
|
|
|
4495 #else
|
|
498
|
4496 # define AH_KEY_LEN 7 /* 6 digits + NUL */
|
|
495
|
4497 #endif
|
|
455
|
4498
|
|
300
|
4499 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
4500 typedef struct affheader_S
|
|
|
4501 {
|
|
495
|
4502 char_u ah_key[AH_KEY_LEN]; /* key for hashtab == name of affix */
|
|
|
4503 unsigned ah_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4504 int ah_newID; /* prefix ID after renumbering; 0 if not used */
|
|
300
|
4505 int ah_combine; /* suffix may combine with prefix */
|
|
498
|
4506 int ah_follows; /* another affix block should be following */
|
|
300
|
4507 affentry_T *ah_first; /* first affix entry */
|
|
|
4508 } affheader_T;
|
|
|
4509
|
|
|
4510 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
4511
|
|
495
|
4512 /* Flag used in compound items. */
|
|
|
4513 typedef struct compitem_S
|
|
|
4514 {
|
|
|
4515 char_u ci_key[AH_KEY_LEN]; /* key for hashtab == name of compound */
|
|
|
4516 unsigned ci_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4517 int ci_newID; /* affix ID after renumbering. */
|
|
|
4518 } compitem_T;
|
|
|
4519
|
|
|
4520 #define HI2CI(hi) ((compitem_T *)(hi)->hi_key)
|
|
|
4521
|
|
300
|
4522 /*
|
|
|
4523 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
4524 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
4525 * once after ":mkspell" is done.
|
|
300
|
4526 */
|
|
|
4527 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
4528 typedef struct sblock_S sblock_T;
|
|
|
4529 struct sblock_S
|
|
|
4530 {
|
|
|
4531 sblock_T *sb_next; /* next block in list */
|
|
|
4532 int sb_used; /* nr of bytes already in use */
|
|
|
4533 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
4534 };
|
|
|
4535
|
|
|
4536 /*
|
|
300
|
4537 * A node in the tree.
|
|
236
|
4538 */
|
|
300
|
4539 typedef struct wordnode_S wordnode_T;
|
|
|
4540 struct wordnode_S
|
|
236
|
4541 {
|
|
346
|
4542 union /* shared to save space */
|
|
|
4543 {
|
|
469
|
4544 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
4545 int index; /* index in written nodes (valid after first
|
|
|
4546 round) */
|
|
|
4547 } wn_u1;
|
|
|
4548 union /* shared to save space */
|
|
|
4549 {
|
|
|
4550 wordnode_T *next; /* next node with same hash key */
|
|
|
4551 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
4552 } wn_u2;
|
|
300
|
4553 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
4554 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
4555 always sorted) */
|
|
474
|
4556 int wn_refs; /* Nr. of references to this node. Only
|
|
|
4557 relevant for first node in a list of
|
|
|
4558 siblings, in following siblings it is
|
|
|
4559 always one. */
|
|
300
|
4560 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
625
|
4561
|
|
|
4562 /* Info for when "wn_byte" is NUL.
|
|
|
4563 * In PREFIXTREE "wn_region" is used for the prefcondnr.
|
|
|
4564 * In the soundfolded word tree "wn_flags" has the MSW of the wordnr and
|
|
|
4565 * "wn_region" the LSW of the wordnr. */
|
|
|
4566 char_u wn_affixID; /* supported/required prefix ID or 0 */
|
|
|
4567 short_u wn_flags; /* WF_ flags */
|
|
|
4568 short wn_region; /* region mask */
|
|
|
4569
|
|
469
|
4570 #ifdef SPELL_PRINTTREE
|
|
|
4571 int wn_nr; /* sequence nr for printing */
|
|
|
4572 #endif
|
|
236
|
4573 };
|
|
|
4574
|
|
390
|
4575 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
4576
|
|
300
|
4577 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
4578
|
|
300
|
4579 /*
|
|
|
4580 * Info used while reading the spell files.
|
|
|
4581 */
|
|
|
4582 typedef struct spellinfo_S
|
|
249
|
4583 {
|
|
300
|
4584 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
4585 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
4586
|
|
300
|
4587 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
4588 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
4589
|
|
339
|
4590 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
4591
|
|
625
|
4592 long si_sugtree; /* creating the soundfolding trie */
|
|
|
4593
|
|
300
|
4594 sblock_T *si_blocks; /* memory blocks used */
|
|
482
|
4595 long si_blocks_cnt; /* memory blocks allocated */
|
|
|
4596 long si_compress_cnt; /* words to add before lowering
|
|
|
4597 compression limit */
|
|
474
|
4598 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
4599 compression, linked by "wn_child" field. */
|
|
482
|
4600 long si_free_count; /* number of nodes in si_first_free */
|
|
474
|
4601 #ifdef SPELL_PRINTTREE
|
|
|
4602 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
4603 #endif
|
|
625
|
4604 buf_T *si_spellbuf; /* buffer used to store soundfold word table */
|
|
474
|
4605
|
|
300
|
4606 int si_ascii; /* handling only ASCII words */
|
|
310
|
4607 int si_add; /* addition file */
|
|
351
|
4608 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
4609 int si_region; /* region mask */
|
|
|
4610 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
4611 int si_memtot; /* runtime memory used */
|
|
310
|
4612 int si_verbose; /* verbose messages */
|
|
474
|
4613 int si_msg_count; /* number of words added since last message */
|
|
316
|
4614 int si_region_count; /* number of regions supported (1 when there
|
|
|
4615 are no regions) */
|
|
483
|
4616 char_u si_region_name[16]; /* region names; used only if
|
|
|
4617 * si_region_count > 1) */
|
|
323
|
4618
|
|
|
4619 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
625
|
4620 garray_T si_repsal; /* list of fromto_T entries from REPSAL lines */
|
|
323
|
4621 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
4622 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
4623 char_u *si_sofoto; /* SOFOTO text */
|
|
625
|
4624 int si_nosugfile; /* NOSUGFILE item found */
|
|
323
|
4625 int si_followup; /* soundsalike: ? */
|
|
|
4626 int si_collapse; /* soundsalike: ? */
|
|
625
|
4627 hashtab_T si_commonwords; /* hashtable for common words */
|
|
|
4628 time_t si_sugtime; /* timestamp for .sug file */
|
|
323
|
4629 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
4630 garray_T si_map; /* MAP info concatenated */
|
|
495
|
4631 char_u *si_midword; /* MIDWORD chars or NULL */
|
|
483
|
4632 int si_compmax; /* max nr of words for compounding */
|
|
481
|
4633 int si_compminlen; /* minimal length for compounding */
|
|
483
|
4634 int si_compsylmax; /* max nr of syllables for compounding */
|
|
481
|
4635 char_u *si_compflags; /* flags used for compounding */
|
|
492
|
4636 char_u si_nobreak; /* NOBREAK */
|
|
483
|
4637 char_u *si_syllable; /* syllable string */
|
|
339
|
4638 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
4639 * prefixes, each stored as a string */
|
|
495
|
4640 int si_newprefID; /* current value for ah_newID */
|
|
500
|
4641 int si_newcompID; /* current value for compound ID */
|
|
300
|
4642 } spellinfo_T;
|
|
249
|
4643
|
|
474
|
4644 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
495
|
4645 static unsigned affitem2flag __ARGS((int flagtype, char_u *item, char_u *fname, int lnum));
|
|
|
4646 static unsigned get_affitem __ARGS((int flagtype, char_u **pp));
|
|
|
4647 static void process_compflags __ARGS((spellinfo_T *spin, afffile_T *aff, char_u *compflags));
|
|
500
|
4648 static void check_renumber __ARGS((spellinfo_T *spin));
|
|
495
|
4649 static int flag_in_afflist __ARGS((int flagtype, char_u *afflist, unsigned flag));
|
|
|
4650 static void aff_check_number __ARGS((int spinval, int affval, char *name));
|
|
|
4651 static void aff_check_string __ARGS((char_u *spinval, char_u *affval, char *name));
|
|
339
|
4652 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
4653 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
4654 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
4655 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
4656 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
4657 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
483
|
4658 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
495
|
4659 static void get_compflags __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
483
|
4660 static int store_aff_word __ARGS((spellinfo_T *spin, char_u *word, char_u *afflist, afffile_T *affile, hashtab_T *ht, hashtab_T *xht, int comb, int flags, char_u *pfxlist, int pfxlen));
|
|
474
|
4661 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
4662 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
4663 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
4664 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
4665 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
483
|
4666 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix));
|
|
481
|
4667 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
4668 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
625
|
4669 static int deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
474
|
4670 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
4671 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
4672 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
4673 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
625
|
4674 static void put_sugtime __ARGS((spellinfo_T *spin, FILE *fd));
|
|
500
|
4675 static int write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
4676 static void clear_node __ARGS((wordnode_T *node));
|
|
|
4677 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
625
|
4678 static void spell_make_sugfile __ARGS((spellinfo_T *spin, char_u *wfname));
|
|
|
4679 static int sug_filltree __ARGS((spellinfo_T *spin, slang_T *slang));
|
|
|
4680 static int sug_maketable __ARGS((spellinfo_T *spin));
|
|
|
4681 static int sug_filltable __ARGS((spellinfo_T *spin, wordnode_T *node, int startwordnr, garray_T *gap));
|
|
|
4682 static int offset2bytes __ARGS((int nr, char_u *buf));
|
|
|
4683 static int bytes2offset __ARGS((char_u **pp));
|
|
|
4684 static void sug_write __ARGS((spellinfo_T *spin, char_u *fname));
|
|
323
|
4685 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
625
|
4686 static void spell_message __ARGS((spellinfo_T *spin, char_u *str));
|
|
310
|
4687 static void init_spellfile __ARGS((void));
|
|
236
|
4688
|
|
455
|
4689 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
4690 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
4691 * Use a negative number with the lower 8 bits zero. */
|
|
|
4692 #define PFX_FLAGS -256
|
|
390
|
4693
|
|
483
|
4694 /*
|
|
|
4695 * Tunable parameters for when the tree is compressed. See 'mkspellmem'.
|
|
|
4696 */
|
|
|
4697 static long compress_start = 30000; /* memory / SBLOCKSIZE */
|
|
|
4698 static long compress_inc = 100; /* memory / SBLOCKSIZE */
|
|
|
4699 static long compress_added = 500000; /* word count */
|
|
|
4700
|
|
469
|
4701 #ifdef SPELL_PRINTTREE
|
|
|
4702 /*
|
|
|
4703 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
4704 * readable format, so that we can see what happens when adding a word and/or
|
|
|
4705 * compressing the tree.
|
|
|
4706 * Based on code from Olaf Seibert.
|
|
|
4707 */
|
|
|
4708 #define PRINTLINESIZE 1000
|
|
|
4709 #define PRINTWIDTH 6
|
|
|
4710
|
|
|
4711 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
4712 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
4713
|
|
|
4714 static char line1[PRINTLINESIZE];
|
|
|
4715 static char line2[PRINTLINESIZE];
|
|
|
4716 static char line3[PRINTLINESIZE];
|
|
|
4717
|
|
|
4718 static void
|
|
|
4719 spell_clear_flags(wordnode_T *node)
|
|
|
4720 {
|
|
|
4721 wordnode_T *np;
|
|
|
4722
|
|
|
4723 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4724 {
|
|
|
4725 np->wn_u1.index = FALSE;
|
|
|
4726 spell_clear_flags(np->wn_child);
|
|
|
4727 }
|
|
|
4728 }
|
|
|
4729
|
|
|
4730 static void
|
|
|
4731 spell_print_node(wordnode_T *node, int depth)
|
|
|
4732 {
|
|
|
4733 if (node->wn_u1.index)
|
|
|
4734 {
|
|
|
4735 /* Done this node before, print the reference. */
|
|
|
4736 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
4737 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
4738 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4739 msg(line1);
|
|
|
4740 msg(line2);
|
|
|
4741 msg(line3);
|
|
|
4742 }
|
|
|
4743 else
|
|
|
4744 {
|
|
|
4745 node->wn_u1.index = TRUE;
|
|
|
4746
|
|
|
4747 if (node->wn_byte != NUL)
|
|
|
4748 {
|
|
|
4749 if (node->wn_child != NULL)
|
|
|
4750 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
4751 else
|
|
|
4752 /* Cannot happen? */
|
|
|
4753 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
4754 }
|
|
|
4755 else
|
|
|
4756 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
4757
|
|
|
4758 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
4759
|
|
|
4760 if (node->wn_sibling != NULL)
|
|
|
4761 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
4762 else
|
|
|
4763 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4764
|
|
|
4765 if (node->wn_byte == NUL)
|
|
|
4766 {
|
|
|
4767 msg(line1);
|
|
|
4768 msg(line2);
|
|
|
4769 msg(line3);
|
|
|
4770 }
|
|
|
4771
|
|
|
4772 /* do the children */
|
|
|
4773 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
4774 spell_print_node(node->wn_child, depth + 1);
|
|
|
4775
|
|
|
4776 /* do the siblings */
|
|
|
4777 if (node->wn_sibling != NULL)
|
|
|
4778 {
|
|
|
4779 /* get rid of all parent details except | */
|
|
|
4780 STRCPY(line1, line3);
|
|
|
4781 STRCPY(line2, line3);
|
|
|
4782 spell_print_node(node->wn_sibling, depth);
|
|
|
4783 }
|
|
|
4784 }
|
|
|
4785 }
|
|
|
4786
|
|
|
4787 static void
|
|
|
4788 spell_print_tree(wordnode_T *root)
|
|
|
4789 {
|
|
|
4790 if (root != NULL)
|
|
|
4791 {
|
|
|
4792 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
4793 * done. */
|
|
|
4794 spell_clear_flags(root);
|
|
|
4795
|
|
|
4796 /* Recursively print the tree. */
|
|
|
4797 spell_print_node(root, 0);
|
|
|
4798 }
|
|
|
4799 }
|
|
|
4800 #endif /* SPELL_PRINTTREE */
|
|
|
4801
|
|
236
|
4802 /*
|
|
323
|
4803 * Read the affix file "fname".
|
|
316
|
4804 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
4805 */
|
|
|
4806 static afffile_T *
|
|
474
|
4807 spell_read_aff(spin, fname)
|
|
|
4808 spellinfo_T *spin;
|
|
236
|
4809 char_u *fname;
|
|
|
4810 {
|
|
|
4811 FILE *fd;
|
|
|
4812 afffile_T *aff;
|
|
|
4813 char_u rline[MAXLINELEN];
|
|
|
4814 char_u *line;
|
|
|
4815 char_u *pc = NULL;
|
|
625
|
4816 #define MAXITEMCNT 30
|
|
334
|
4817 char_u *(items[MAXITEMCNT]);
|
|
236
|
4818 int itemcnt;
|
|
|
4819 char_u *p;
|
|
|
4820 int lnum = 0;
|
|
|
4821 affheader_T *cur_aff = NULL;
|
|
495
|
4822 int did_postpone_prefix = FALSE;
|
|
236
|
4823 int aff_todo = 0;
|
|
|
4824 hashtab_T *tp;
|
|
255
|
4825 char_u *low = NULL;
|
|
|
4826 char_u *fol = NULL;
|
|
|
4827 char_u *upp = NULL;
|
|
323
|
4828 int do_rep;
|
|
625
|
4829 int do_repsal;
|
|
323
|
4830 int do_sal;
|
|
|
4831 int do_map;
|
|
|
4832 int found_map = FALSE;
|
|
324
|
4833 hashitem_T *hi;
|
|
455
|
4834 int l;
|
|
495
|
4835 int compminlen = 0; /* COMPOUNDMIN value */
|
|
|
4836 int compsylmax = 0; /* COMPOUNDSYLMAX value */
|
|
|
4837 int compmax = 0; /* COMPOUNDMAX value */
|
|
|
4838 char_u *compflags = NULL; /* COMPOUNDFLAG and COMPOUNDFLAGS
|
|
|
4839 concatenated */
|
|
|
4840 char_u *midword = NULL; /* MIDWORD value */
|
|
|
4841 char_u *syllable = NULL; /* SYLLABLE value */
|
|
|
4842 char_u *sofofrom = NULL; /* SOFOFROM value */
|
|
|
4843 char_u *sofoto = NULL; /* SOFOTO value */
|
|
236
|
4844
|
|
300
|
4845 /*
|
|
|
4846 * Open the file.
|
|
|
4847 */
|
|
310
|
4848 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4849 if (fd == NULL)
|
|
|
4850 {
|
|
|
4851 EMSG2(_(e_notopen), fname);
|
|
|
4852 return NULL;
|
|
|
4853 }
|
|
|
4854
|
|
625
|
4855 vim_snprintf((char *)IObuff, IOSIZE, _("Reading affix file %s ..."), fname);
|
|
|
4856 spell_message(spin, IObuff);
|
|
236
|
4857
|
|
323
|
4858 /* Only do REP lines when not done in another .aff file already. */
|
|
|
4859 do_rep = spin->si_rep.ga_len == 0;
|
|
|
4860
|
|
625
|
4861 /* Only do REPSAL lines when not done in another .aff file already. */
|
|
|
4862 do_repsal = spin->si_repsal.ga_len == 0;
|
|
|
4863
|
|
323
|
4864 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
4865 do_sal = spin->si_sal.ga_len == 0;
|
|
|
4866
|
|
|
4867 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
4868 do_map = spin->si_map.ga_len == 0;
|
|
|
4869
|
|
300
|
4870 /*
|
|
|
4871 * Allocate and init the afffile_T structure.
|
|
|
4872 */
|
|
474
|
4873 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
4874 if (aff == NULL)
|
|
|
4875 return NULL;
|
|
|
4876 hash_init(&aff->af_pref);
|
|
|
4877 hash_init(&aff->af_suff);
|
|
495
|
4878 hash_init(&aff->af_comp);
|
|
236
|
4879
|
|
|
4880 /*
|
|
|
4881 * Read all the lines in the file one by one.
|
|
|
4882 */
|
|
255
|
4883 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
4884 {
|
|
255
|
4885 line_breakcheck();
|
|
236
|
4886 ++lnum;
|
|
|
4887
|
|
|
4888 /* Skip comment lines. */
|
|
|
4889 if (*rline == '#')
|
|
|
4890 continue;
|
|
|
4891
|
|
|
4892 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
4893 vim_free(pc);
|
|
310
|
4894 #ifdef FEAT_MBYTE
|
|
300
|
4895 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4896 {
|
|
300
|
4897 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
4898 if (pc == NULL)
|
|
|
4899 {
|
|
|
4900 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4901 fname, lnum, rline);
|
|
|
4902 continue;
|
|
|
4903 }
|
|
236
|
4904 line = pc;
|
|
|
4905 }
|
|
|
4906 else
|
|
310
|
4907 #endif
|
|
236
|
4908 {
|
|
|
4909 pc = NULL;
|
|
|
4910 line = rline;
|
|
|
4911 }
|
|
|
4912
|
|
|
4913 /* Split the line up in white separated items. Put a NUL after each
|
|
|
4914 * item. */
|
|
|
4915 itemcnt = 0;
|
|
|
4916 for (p = line; ; )
|
|
|
4917 {
|
|
|
4918 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
4919 ++p;
|
|
|
4920 if (*p == NUL)
|
|
|
4921 break;
|
|
334
|
4922 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
4923 break;
|
|
236
|
4924 items[itemcnt++] = p;
|
|
300
|
4925 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
4926 ++p;
|
|
|
4927 if (*p == NUL)
|
|
|
4928 break;
|
|
|
4929 *p++ = NUL;
|
|
|
4930 }
|
|
|
4931
|
|
|
4932 /* Handle non-empty lines. */
|
|
|
4933 if (itemcnt > 0)
|
|
|
4934 {
|
|
|
4935 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
4936 && aff->af_enc == NULL)
|
|
|
4937 {
|
|
310
|
4938 #ifdef FEAT_MBYTE
|
|
300
|
4939 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
4940 aff->af_enc = enc_canonize(items[1]);
|
|
|
4941 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
4942 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
4943 p_enc) == FAIL)
|
|
|
4944 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
4945 fname, aff->af_enc, p_enc);
|
|
375
|
4946 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4947 #else
|
|
|
4948 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4949 #endif
|
|
236
|
4950 }
|
|
495
|
4951 else if (STRCMP(items[0], "FLAG") == 0 && itemcnt == 2
|
|
|
4952 && aff->af_flagtype == AFT_CHAR)
|
|
|
4953 {
|
|
|
4954 if (STRCMP(items[1], "long") == 0)
|
|
498
|
4955 aff->af_flagtype = AFT_LONG;
|
|
495
|
4956 else if (STRCMP(items[1], "num") == 0)
|
|
498
|
4957 aff->af_flagtype = AFT_NUM;
|
|
|
4958 else if (STRCMP(items[1], "caplong") == 0)
|
|
|
4959 aff->af_flagtype = AFT_CAPLONG;
|
|
495
|
4960 else
|
|
|
4961 smsg((char_u *)_("Invalid value for FLAG in %s line %d: %s"),
|
|
|
4962 fname, lnum, items[1]);
|
|
621
|
4963 if (aff->af_rare != 0
|
|
|
4964 || aff->af_keepcase != 0
|
|
|
4965 || aff->af_bad != 0
|
|
500
|
4966 || aff->af_needaffix != 0
|
|
|
4967 || aff->af_needcomp != 0
|
|
|
4968 || compflags != NULL
|
|
495
|
4969 || aff->af_suff.ht_used > 0
|
|
|
4970 || aff->af_pref.ht_used > 0)
|
|
|
4971 smsg((char_u *)_("FLAG after using flags in %s line %d: %s"),
|
|
|
4972 fname, lnum, items[1]);
|
|
|
4973 }
|
|
|
4974 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2
|
|
|
4975 && midword == NULL)
|
|
|
4976 {
|
|
|
4977 midword = getroom_save(spin, items[1]);
|
|
366
|
4978 }
|
|
302
|
4979 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
4980 {
|
|
323
|
4981 /* ignored, we always split */
|
|
302
|
4982 }
|
|
323
|
4983 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
4984 {
|
|
323
|
4985 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
4986 }
|
|
621
|
4987 /* TODO: remove "RAR" later */
|
|
|
4988 else if ((STRCMP(items[0], "RAR") == 0
|
|
|
4989 || STRCMP(items[0], "RARE") == 0) && itemcnt == 2
|
|
|
4990 && aff->af_rare == 0)
|
|
|
4991 {
|
|
|
4992 aff->af_rare = affitem2flag(aff->af_flagtype, items[1],
|
|
495
|
4993 fname, lnum);
|
|
307
|
4994 }
|
|
621
|
4995 /* TODO: remove "KEP" later */
|
|
|
4996 else if ((STRCMP(items[0], "KEP") == 0
|
|
|
4997 || STRCMP(items[0], "KEEPCASE") == 0) && itemcnt == 2
|
|
|
4998 && aff->af_keepcase == 0)
|
|
|
4999 {
|
|
|
5000 aff->af_keepcase = affitem2flag(aff->af_flagtype, items[1],
|
|
495
|
5001 fname, lnum);
|
|
307
|
5002 }
|
|
346
|
5003 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
5004 && aff->af_bad == 0)
|
|
|
5005 {
|
|
495
|
5006 aff->af_bad = affitem2flag(aff->af_flagtype, items[1],
|
|
|
5007 fname, lnum);
|
|
346
|
5008 }
|
|
483
|
5009 else if (STRCMP(items[0], "NEEDAFFIX") == 0 && itemcnt == 2
|
|
|
5010 && aff->af_needaffix == 0)
|
|
|
5011 {
|
|
495
|
5012 aff->af_needaffix = affitem2flag(aff->af_flagtype, items[1],
|
|
|
5013 fname, lnum);
|
|
483
|
5014 }
|
|
500
|
5015 else if (STRCMP(items[0], "NEEDCOMPOUND") == 0 && itemcnt == 2
|
|
|
5016 && aff->af_needcomp == 0)
|
|
|
5017 {
|
|
|
5018 aff->af_needcomp = affitem2flag(aff->af_flagtype, items[1],
|
|
|
5019 fname, lnum);
|
|
|
5020 }
|
|
481
|
5021 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
495
|
5022 && compflags == NULL)
|
|
|
5023 {
|
|
|
5024 /* Turn flag "c" into COMPOUNDFLAGS compatible string "c+",
|
|
|
5025 * "Na" into "Na+", "1234" into "1234+". */
|
|
|
5026 p = getroom(spin, STRLEN(items[1]) + 2, FALSE);
|
|
483
|
5027 if (p != NULL)
|
|
|
5028 {
|
|
495
|
5029 STRCPY(p, items[1]);
|
|
|
5030 STRCAT(p, "+");
|
|
|
5031 compflags = p;
|
|
|
5032 }
|
|
481
|
5033 }
|
|
483
|
5034 else if (STRCMP(items[0], "COMPOUNDFLAGS") == 0 && itemcnt == 2)
|
|
|
5035 {
|
|
|
5036 /* Concatenate this string to previously defined ones, using a
|
|
|
5037 * slash to separate them. */
|
|
|
5038 l = STRLEN(items[1]) + 1;
|
|
495
|
5039 if (compflags != NULL)
|
|
|
5040 l += STRLEN(compflags) + 1;
|
|
483
|
5041 p = getroom(spin, l, FALSE);
|
|
|
5042 if (p != NULL)
|
|
|
5043 {
|
|
495
|
5044 if (compflags != NULL)
|
|
483
|
5045 {
|
|
495
|
5046 STRCPY(p, compflags);
|
|
483
|
5047 STRCAT(p, "/");
|
|
|
5048 }
|
|
|
5049 STRCAT(p, items[1]);
|
|
495
|
5050 compflags = p;
|
|
483
|
5051 }
|
|
|
5052 }
|
|
|
5053 else if (STRCMP(items[0], "COMPOUNDMAX") == 0 && itemcnt == 2
|
|
495
|
5054 && compmax == 0)
|
|
|
5055 {
|
|
|
5056 compmax = atoi((char *)items[1]);
|
|
|
5057 if (compmax == 0)
|
|
483
|
5058 smsg((char_u *)_("Wrong COMPOUNDMAX value in %s line %d: %s"),
|
|
|
5059 fname, lnum, items[1]);
|
|
481
|
5060 }
|
|
|
5061 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
495
|
5062 && compminlen == 0)
|
|
|
5063 {
|
|
|
5064 compminlen = atoi((char *)items[1]);
|
|
|
5065 if (compminlen == 0)
|
|
481
|
5066 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
5067 fname, lnum, items[1]);
|
|
|
5068 }
|
|
483
|
5069 else if (STRCMP(items[0], "COMPOUNDSYLMAX") == 0 && itemcnt == 2
|
|
495
|
5070 && compsylmax == 0)
|
|
|
5071 {
|
|
|
5072 compsylmax = atoi((char *)items[1]);
|
|
|
5073 if (compsylmax == 0)
|
|
483
|
5074 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"),
|
|
|
5075 fname, lnum, items[1]);
|
|
|
5076 }
|
|
|
5077 else if (STRCMP(items[0], "SYLLABLE") == 0 && itemcnt == 2
|
|
495
|
5078 && syllable == NULL)
|
|
|
5079 {
|
|
|
5080 syllable = getroom_save(spin, items[1]);
|
|
483
|
5081 }
|
|
492
|
5082 else if (STRCMP(items[0], "NOBREAK") == 0 && itemcnt == 1)
|
|
|
5083 {
|
|
|
5084 spin->si_nobreak = TRUE;
|
|
|
5085 }
|
|
625
|
5086 else if (STRCMP(items[0], "NOSUGFILE") == 0 && itemcnt == 1)
|
|
|
5087 {
|
|
|
5088 spin->si_nosugfile = TRUE;
|
|
|
5089 }
|
|
339
|
5090 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
5091 {
|
|
|
5092 aff->af_pfxpostpone = TRUE;
|
|
|
5093 }
|
|
236
|
5094 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
5095 || STRCMP(items[0], "SFX") == 0)
|
|
|
5096 && aff_todo == 0
|
|
334
|
5097 && itemcnt >= 4)
|
|
236
|
5098 {
|
|
498
|
5099 int lasti = 4;
|
|
|
5100 char_u key[AH_KEY_LEN];
|
|
|
5101
|
|
|
5102 if (*items[0] == 'P')
|
|
|
5103 tp = &aff->af_pref;
|
|
|
5104 else
|
|
|
5105 tp = &aff->af_suff;
|
|
|
5106
|
|
|
5107 /* Myspell allows the same affix name to be used multiple
|
|
|
5108 * times. The affix files that do this have an undocumented
|
|
|
5109 * "S" flag on all but the last block, thus we check for that
|
|
|
5110 * and store it in ah_follows. */
|
|
|
5111 vim_strncpy(key, items[1], AH_KEY_LEN - 1);
|
|
|
5112 hi = hash_find(tp, key);
|
|
|
5113 if (!HASHITEM_EMPTY(hi))
|
|
|
5114 {
|
|
|
5115 cur_aff = HI2AH(hi);
|
|
|
5116 if (cur_aff->ah_combine != (*items[2] == 'Y'))
|
|
|
5117 smsg((char_u *)_("Different combining flag in continued affix block in %s line %d: %s"),
|
|
|
5118 fname, lnum, items[1]);
|
|
|
5119 if (!cur_aff->ah_follows)
|
|
|
5120 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
5121 fname, lnum, items[1]);
|
|
|
5122 }
|
|
|
5123 else
|
|
|
5124 {
|
|
|
5125 /* New affix letter. */
|
|
|
5126 cur_aff = (affheader_T *)getroom(spin,
|
|
|
5127 sizeof(affheader_T), TRUE);
|
|
|
5128 if (cur_aff == NULL)
|
|
|
5129 break;
|
|
|
5130 cur_aff->ah_flag = affitem2flag(aff->af_flagtype, items[1],
|
|
|
5131 fname, lnum);
|
|
|
5132 if (cur_aff->ah_flag == 0 || STRLEN(items[1]) >= AH_KEY_LEN)
|
|
|
5133 break;
|
|
|
5134 if (cur_aff->ah_flag == aff->af_bad
|
|
621
|
5135 || cur_aff->ah_flag == aff->af_rare
|
|
|
5136 || cur_aff->ah_flag == aff->af_keepcase
|
|
500
|
5137 || cur_aff->ah_flag == aff->af_needaffix
|
|
|
5138 || cur_aff->ah_flag == aff->af_needcomp)
|
|
621
|
5139 smsg((char_u *)_("Affix also used for BAD/RARE/KEEPCASE/NEEDAFFIX/NEEDCOMPOUND in %s line %d: %s"),
|
|
498
|
5140 fname, lnum, items[1]);
|
|
|
5141 STRCPY(cur_aff->ah_key, items[1]);
|
|
|
5142 hash_add(tp, cur_aff->ah_key);
|
|
|
5143
|
|
|
5144 cur_aff->ah_combine = (*items[2] == 'Y');
|
|
|
5145 }
|
|
|
5146
|
|
|
5147 /* Check for the "S" flag, which apparently means that another
|
|
|
5148 * block with the same affix name is following. */
|
|
|
5149 if (itemcnt > lasti && STRCMP(items[lasti], "S") == 0)
|
|
|
5150 {
|
|
|
5151 ++lasti;
|
|
|
5152 cur_aff->ah_follows = TRUE;
|
|
|
5153 }
|
|
|
5154 else
|
|
|
5155 cur_aff->ah_follows = FALSE;
|
|
|
5156
|
|
334
|
5157 /* Myspell allows extra text after the item, but that might
|
|
|
5158 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
498
|
5159 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
5160 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
5161 fname, lnum, items[4]);
|
|
|
5162
|
|
498
|
5163 if (STRCMP(items[2], "Y") != 0 && STRCMP(items[2], "N") != 0)
|
|
236
|
5164 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
5165 fname, lnum, items[2]);
|
|
339
|
5166
|
|
498
|
5167 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
|
5168 {
|
|
|
5169 if (cur_aff->ah_newID == 0)
|
|
495
|
5170 {
|
|
|
5171 /* Use a new number in the .spl file later, to be able
|
|
|
5172 * to handle multiple .aff files. */
|
|
500
|
5173 check_renumber(spin);
|
|
495
|
5174 cur_aff->ah_newID = ++spin->si_newprefID;
|
|
|
5175
|
|
|
5176 /* We only really use ah_newID if the prefix is
|
|
|
5177 * postponed. We know that only after handling all
|
|
|
5178 * the items. */
|
|
|
5179 did_postpone_prefix = FALSE;
|
|
|
5180 }
|
|
498
|
5181 else
|
|
|
5182 /* Did use the ID in a previous block. */
|
|
|
5183 did_postpone_prefix = TRUE;
|
|
|
5184 }
|
|
|
5185
|
|
300
|
5186 aff_todo = atoi((char *)items[3]);
|
|
236
|
5187 }
|
|
|
5188 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
5189 || STRCMP(items[0], "SFX") == 0)
|
|
|
5190 && aff_todo > 0
|
|
|
5191 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
5192 && itemcnt >= 5)
|
|
236
|
5193 {
|
|
|
5194 affentry_T *aff_entry;
|
|
366
|
5195 int rare = FALSE;
|
|
483
|
5196 int nocomp = FALSE;
|
|
455
|
5197 int upper = FALSE;
|
|
366
|
5198 int lasti = 5;
|
|
|
5199
|
|
483
|
5200 /* Check for "rare" and "nocomp" after the other info. */
|
|
|
5201 while (itemcnt > lasti)
|
|
|
5202 {
|
|
|
5203 if (!rare && STRICMP(items[lasti], "rare") == 0)
|
|
|
5204 {
|
|
|
5205 rare = TRUE;
|
|
|
5206 ++lasti;
|
|
|
5207 }
|
|
|
5208 else if (!nocomp && STRICMP(items[lasti], "nocomp") == 0)
|
|
|
5209 {
|
|
|
5210 nocomp = TRUE;
|
|
|
5211 ++lasti;
|
|
|
5212 }
|
|
|
5213 else
|
|
|
5214 break;
|
|
366
|
5215 }
|
|
236
|
5216
|
|
334
|
5217 /* Myspell allows extra text after the item, but that might
|
|
|
5218 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
5219 if (itemcnt > lasti && *items[lasti] != '#')
|
|
482
|
5220 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]);
|
|
334
|
5221
|
|
236
|
5222 /* New item for an affix letter. */
|
|
|
5223 --aff_todo;
|
|
474
|
5224 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
5225 sizeof(affentry_T), TRUE);
|
|
236
|
5226 if (aff_entry == NULL)
|
|
|
5227 break;
|
|
366
|
5228 aff_entry->ae_rare = rare;
|
|
483
|
5229 aff_entry->ae_nocomp = nocomp;
|
|
240
|
5230
|
|
236
|
5231 if (STRCMP(items[2], "0") != 0)
|
|
474
|
5232 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
5233 if (STRCMP(items[3], "0") != 0)
|
|
474
|
5234 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
5235
|
|
300
|
5236 /* Don't use an affix entry with non-ASCII characters when
|
|
|
5237 * "spin->si_ascii" is TRUE. */
|
|
|
5238 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
5239 || has_non_ascii(aff_entry->ae_add)))
|
|
|
5240 {
|
|
|
5241 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
5242 cur_aff->ah_first = aff_entry;
|
|
300
|
5243
|
|
|
5244 if (STRCMP(items[4], ".") != 0)
|
|
|
5245 {
|
|
|
5246 char_u buf[MAXLINELEN];
|
|
|
5247
|
|
474
|
5248 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
5249 if (*items[0] == 'P')
|
|
|
5250 sprintf((char *)buf, "^%s", items[4]);
|
|
|
5251 else
|
|
|
5252 sprintf((char *)buf, "%s$", items[4]);
|
|
|
5253 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
5254 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
5255 if (aff_entry->ae_prog == NULL)
|
|
|
5256 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
5257 fname, lnum, items[4]);
|
|
300
|
5258 }
|
|
339
|
5259
|
|
|
5260 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
5261 * for the condition. Use an existing one if possible. */
|
|
455
|
5262 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
5263 {
|
|
455
|
5264 /* When the chop string is one lower-case letter and
|
|
|
5265 * the add string ends in the upper-case letter we set
|
|
|
5266 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
5267 * letters from "ae_add". The condition must either
|
|
|
5268 * be empty or start with the same letter. */
|
|
|
5269 if (aff_entry->ae_chop != NULL
|
|
|
5270 && aff_entry->ae_add != NULL
|
|
|
5271 #ifdef FEAT_MBYTE
|
|
474
|
5272 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
5273 aff_entry->ae_chop)] == NUL
|
|
|
5274 #else
|
|
|
5275 && aff_entry->ae_chop[1] == NUL
|
|
|
5276 #endif
|
|
|
5277 )
|
|
339
|
5278 {
|
|
455
|
5279 int c, c_up;
|
|
|
5280
|
|
|
5281 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
5282 c_up = SPELL_TOUPPER(c);
|
|
|
5283 if (c_up != c
|
|
|
5284 && (aff_entry->ae_cond == NULL
|
|
|
5285 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
5286 {
|
|
|
5287 p = aff_entry->ae_add
|
|
|
5288 + STRLEN(aff_entry->ae_add);
|
|
|
5289 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
5290 if (PTR2CHAR(p) == c_up)
|
|
|
5291 {
|
|
|
5292 upper = TRUE;
|
|
|
5293 aff_entry->ae_chop = NULL;
|
|
|
5294 *p = NUL;
|
|
|
5295
|
|
|
5296 /* The condition is matched with the
|
|
|
5297 * actual word, thus must check for the
|
|
|
5298 * upper-case letter. */
|
|
|
5299 if (aff_entry->ae_cond != NULL)
|
|
|
5300 {
|
|
|
5301 char_u buf[MAXLINELEN];
|
|
|
5302 #ifdef FEAT_MBYTE
|
|
|
5303 if (has_mbyte)
|
|
|
5304 {
|
|
|
5305 onecap_copy(items[4], buf, TRUE);
|
|
|
5306 aff_entry->ae_cond = getroom_save(
|
|
474
|
5307 spin, buf);
|
|
455
|
5308 }
|
|
|
5309 else
|
|
|
5310 #endif
|
|
|
5311 *aff_entry->ae_cond = c_up;
|
|
|
5312 if (aff_entry->ae_cond != NULL)
|
|
|
5313 {
|
|
|
5314 sprintf((char *)buf, "^%s",
|
|
339
|
5315 aff_entry->ae_cond);
|
|
455
|
5316 vim_free(aff_entry->ae_prog);
|
|
|
5317 aff_entry->ae_prog = vim_regcomp(
|
|
|
5318 buf, RE_MAGIC + RE_STRING);
|
|
|
5319 }
|
|
|
5320 }
|
|
|
5321 }
|
|
|
5322 }
|
|
339
|
5323 }
|
|
|
5324
|
|
455
|
5325 if (aff_entry->ae_chop == NULL)
|
|
390
|
5326 {
|
|
455
|
5327 int idx;
|
|
|
5328 char_u **pp;
|
|
|
5329 int n;
|
|
|
5330
|
|
495
|
5331 /* Find a previously used condition. */
|
|
455
|
5332 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
5333 --idx)
|
|
|
5334 {
|
|
|
5335 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
5336 if (str_equal(p, aff_entry->ae_cond))
|
|
|
5337 break;
|
|
|
5338 }
|
|
|
5339 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
5340 {
|
|
|
5341 /* Not found, add a new condition. */
|
|
|
5342 idx = spin->si_prefcond.ga_len++;
|
|
|
5343 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
5344 + idx;
|
|
|
5345 if (aff_entry->ae_cond == NULL)
|
|
|
5346 *pp = NULL;
|
|
|
5347 else
|
|
474
|
5348 *pp = getroom_save(spin,
|
|
455
|
5349 aff_entry->ae_cond);
|
|
|
5350 }
|
|
|
5351
|
|
|
5352 /* Add the prefix to the prefix tree. */
|
|
|
5353 if (aff_entry->ae_add == NULL)
|
|
|
5354 p = (char_u *)"";
|
|
390
|
5355 else
|
|
455
|
5356 p = aff_entry->ae_add;
|
|
|
5357 /* PFX_FLAGS is a negative number, so that
|
|
|
5358 * tree_add_word() knows this is the prefix tree. */
|
|
|
5359 n = PFX_FLAGS;
|
|
|
5360 if (rare)
|
|
|
5361 n |= WFP_RARE;
|
|
|
5362 if (!cur_aff->ah_combine)
|
|
|
5363 n |= WFP_NC;
|
|
|
5364 if (upper)
|
|
|
5365 n |= WFP_UP;
|
|
474
|
5366 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
5367 idx, cur_aff->ah_newID);
|
|
495
|
5368 did_postpone_prefix = TRUE;
|
|
|
5369 }
|
|
|
5370
|
|
|
5371 /* Didn't actually use ah_newID, backup si_newprefID. */
|
|
|
5372 if (aff_todo == 0 && !did_postpone_prefix)
|
|
|
5373 {
|
|
|
5374 --spin->si_newprefID;
|
|
|
5375 cur_aff->ah_newID = 0;
|
|
390
|
5376 }
|
|
339
|
5377 }
|
|
240
|
5378 }
|
|
236
|
5379 }
|
|
495
|
5380 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2
|
|
|
5381 && fol == NULL)
|
|
|
5382 {
|
|
|
5383 fol = vim_strsave(items[1]);
|
|
|
5384 }
|
|
|
5385 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2
|
|
|
5386 && low == NULL)
|
|
|
5387 {
|
|
|
5388 low = vim_strsave(items[1]);
|
|
|
5389 }
|
|
|
5390 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2
|
|
|
5391 && upp == NULL)
|
|
|
5392 {
|
|
|
5393 upp = vim_strsave(items[1]);
|
|
255
|
5394 }
|
|
625
|
5395 else if ((STRCMP(items[0], "REP") == 0
|
|
|
5396 || STRCMP(items[0], "REPSAL") == 0)
|
|
|
5397 && itemcnt == 2)
|
|
|
5398 {
|
|
|
5399 /* Ignore REP/REPSAL count */;
|
|
323
|
5400 if (!isdigit(*items[1]))
|
|
625
|
5401 smsg((char_u *)_("Expected REP(SAL) count in %s line %d"),
|
|
323
|
5402 fname, lnum);
|
|
|
5403 }
|
|
625
|
5404 else if ((STRCMP(items[0], "REP") == 0
|
|
|
5405 || STRCMP(items[0], "REPSAL") == 0)
|
|
|
5406 && itemcnt >= 3)
|
|
|
5407 {
|
|
|
5408 /* REP/REPSAL item */
|
|
482
|
5409 /* Myspell ignores extra arguments, we require it starts with
|
|
|
5410 * # to detect mistakes. */
|
|
|
5411 if (itemcnt > 3 && items[3][0] != '#')
|
|
|
5412 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]);
|
|
625
|
5413 if (items[0][3] == 'S' ? do_repsal : do_rep)
|
|
532
|
5414 {
|
|
|
5415 /* Replace underscore with space (can't include a space
|
|
|
5416 * directly). */
|
|
|
5417 for (p = items[1]; *p != NUL; mb_ptr_adv(p))
|
|
|
5418 if (*p == '_')
|
|
|
5419 *p = ' ';
|
|
|
5420 for (p = items[2]; *p != NUL; mb_ptr_adv(p))
|
|
|
5421 if (*p == '_')
|
|
|
5422 *p = ' ';
|
|
625
|
5423 add_fromto(spin, items[0][3] == 'S'
|
|
|
5424 ? &spin->si_repsal
|
|
|
5425 : &spin->si_rep, items[1], items[2]);
|
|
532
|
5426 }
|
|
323
|
5427 }
|
|
|
5428 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
5429 {
|
|
|
5430 /* MAP item or count */
|
|
|
5431 if (!found_map)
|
|
|
5432 {
|
|
|
5433 /* First line contains the count. */
|
|
|
5434 found_map = TRUE;
|
|
|
5435 if (!isdigit(*items[1]))
|
|
|
5436 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
5437 fname, lnum);
|
|
|
5438 }
|
|
|
5439 else if (do_map)
|
|
|
5440 {
|
|
346
|
5441 int c;
|
|
|
5442
|
|
|
5443 /* Check that every character appears only once. */
|
|
|
5444 for (p = items[1]; *p != NUL; )
|
|
|
5445 {
|
|
|
5446 #ifdef FEAT_MBYTE
|
|
|
5447 c = mb_ptr2char_adv(&p);
|
|
|
5448 #else
|
|
|
5449 c = *p++;
|
|
|
5450 #endif
|
|
|
5451 if ((spin->si_map.ga_len > 0
|
|
|
5452 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
5453 != NULL)
|
|
|
5454 || vim_strchr(p, c) != NULL)
|
|
|
5455 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
5456 fname, lnum);
|
|
|
5457 }
|
|
|
5458
|
|
323
|
5459 /* We simply concatenate all the MAP strings, separated by
|
|
|
5460 * slashes. */
|
|
|
5461 ga_concat(&spin->si_map, items[1]);
|
|
|
5462 ga_append(&spin->si_map, '/');
|
|
|
5463 }
|
|
|
5464 }
|
|
534
|
5465 /* Accept "SAL from to" and "SAL from to # comment". */
|
|
|
5466 else if (STRCMP(items[0], "SAL") == 0
|
|
|
5467 && (itemcnt == 3 || (itemcnt > 3 && items[3][0] == '#')))
|
|
323
|
5468 {
|
|
|
5469 if (do_sal)
|
|
|
5470 {
|
|
|
5471 /* SAL item (sounds-a-like)
|
|
|
5472 * Either one of the known keys or a from-to pair. */
|
|
|
5473 if (STRCMP(items[1], "followup") == 0)
|
|
|
5474 spin->si_followup = sal_to_bool(items[2]);
|
|
|
5475 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
5476 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
5477 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
5478 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
5479 else
|
|
|
5480 /* when "to" is "_" it means empty */
|
|
|
5481 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
5482 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
5483 : items[2]);
|
|
|
5484 }
|
|
236
|
5485 }
|
|
375
|
5486 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
495
|
5487 && sofofrom == NULL)
|
|
|
5488 {
|
|
|
5489 sofofrom = getroom_save(spin, items[1]);
|
|
375
|
5490 }
|
|
|
5491 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
495
|
5492 && sofoto == NULL)
|
|
|
5493 {
|
|
|
5494 sofoto = getroom_save(spin, items[1]);
|
|
375
|
5495 }
|
|
625
|
5496 else if (STRCMP(items[0], "COMMON") == 0)
|
|
|
5497 {
|
|
|
5498 int i;
|
|
|
5499
|
|
|
5500 for (i = 1; i < itemcnt; ++i)
|
|
|
5501 {
|
|
|
5502 if (HASHITEM_EMPTY(hash_find(&spin->si_commonwords,
|
|
|
5503 items[i])))
|
|
|
5504 {
|
|
|
5505 p = vim_strsave(items[i]);
|
|
|
5506 if (p == NULL)
|
|
|
5507 break;
|
|
|
5508 hash_add(&spin->si_commonwords, p);
|
|
|
5509 }
|
|
|
5510 }
|
|
|
5511 }
|
|
300
|
5512 else
|
|
481
|
5513 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
5514 fname, lnum, items[0]);
|
|
|
5515 }
|
|
|
5516 }
|
|
|
5517
|
|
255
|
5518 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
5519 {
|
|
351
|
5520 if (spin->si_clear_chartab)
|
|
|
5521 {
|
|
|
5522 /* Clear the char type tables, don't want to use any of the
|
|
|
5523 * currently used spell properties. */
|
|
|
5524 init_spell_chartab();
|
|
|
5525 spin->si_clear_chartab = FALSE;
|
|
|
5526 }
|
|
|
5527
|
|
316
|
5528 /*
|
|
|
5529 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
5530 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
5531 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
5532 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
5533 */
|
|
|
5534 if (!spin->si_ascii
|
|
|
5535 #ifdef FEAT_MBYTE
|
|
|
5536 && !enc_utf8
|
|
|
5537 #endif
|
|
|
5538 )
|
|
260
|
5539 {
|
|
|
5540 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
5541 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
5542 else
|
|
316
|
5543 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
5544 }
|
|
255
|
5545
|
|
|
5546 vim_free(fol);
|
|
|
5547 vim_free(low);
|
|
|
5548 vim_free(upp);
|
|
|
5549 }
|
|
|
5550
|
|
481
|
5551 /* Use compound specifications of the .aff file for the spell info. */
|
|
495
|
5552 if (compmax != 0)
|
|
|
5553 {
|
|
|
5554 aff_check_number(spin->si_compmax, compmax, "COMPOUNDMAX");
|
|
|
5555 spin->si_compmax = compmax;
|
|
|
5556 }
|
|
|
5557
|
|
|
5558 if (compminlen != 0)
|
|
|
5559 {
|
|
|
5560 aff_check_number(spin->si_compminlen, compminlen, "COMPOUNDMIN");
|
|
|
5561 spin->si_compminlen = compminlen;
|
|
|
5562 }
|
|
|
5563
|
|
|
5564 if (compsylmax != 0)
|
|
|
5565 {
|
|
|
5566 if (syllable == NULL)
|
|
|
5567 smsg((char_u *)_("COMPOUNDSYLMAX used without SYLLABLE"));
|
|
|
5568 aff_check_number(spin->si_compsylmax, compsylmax, "COMPOUNDSYLMAX");
|
|
|
5569 spin->si_compsylmax = compsylmax;
|
|
|
5570 }
|
|
|
5571
|
|
|
5572 if (compflags != NULL)
|
|
|
5573 process_compflags(spin, aff, compflags);
|
|
|
5574
|
|
|
5575 /* Check that we didn't use too many renumbered flags. */
|
|
500
|
5576 if (spin->si_newcompID < spin->si_newprefID)
|
|
|
5577 {
|
|
|
5578 if (spin->si_newcompID == 127 || spin->si_newcompID == 255)
|
|
495
|
5579 MSG(_("Too many postponed prefixes"));
|
|
500
|
5580 else if (spin->si_newprefID == 0 || spin->si_newprefID == 127)
|
|
495
|
5581 MSG(_("Too many compound flags"));
|
|
481
|
5582 else
|
|
495
|
5583 MSG(_("Too many posponed prefixes and/or compound flags"));
|
|
|
5584 }
|
|
|
5585
|
|
|
5586 if (syllable != NULL)
|
|
|
5587 {
|
|
|
5588 aff_check_string(spin->si_syllable, syllable, "SYLLABLE");
|
|
|
5589 spin->si_syllable = syllable;
|
|
|
5590 }
|
|
|
5591
|
|
|
5592 if (sofofrom != NULL || sofoto != NULL)
|
|
|
5593 {
|
|
|
5594 if (sofofrom == NULL || sofoto == NULL)
|
|
|
5595 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
5596 sofofrom == NULL ? "FROM" : "TO", fname);
|
|
|
5597 else if (spin->si_sal.ga_len > 0)
|
|
|
5598 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
483
|
5599 else
|
|
495
|
5600 {
|
|
|
5601 aff_check_string(spin->si_sofofr, sofofrom, "SOFOFROM");
|
|
|
5602 aff_check_string(spin->si_sofoto, sofoto, "SOFOTO");
|
|
|
5603 spin->si_sofofr = sofofrom;
|
|
|
5604 spin->si_sofoto = sofoto;
|
|
|
5605 }
|
|
|
5606 }
|
|
|
5607
|
|
|
5608 if (midword != NULL)
|
|
|
5609 {
|
|
|
5610 aff_check_string(spin->si_midword, midword, "MIDWORD");
|
|
|
5611 spin->si_midword = midword;
|
|
483
|
5612 }
|
|
|
5613
|
|
236
|
5614 vim_free(pc);
|
|
|
5615 fclose(fd);
|
|
|
5616 return aff;
|
|
|
5617 }
|
|
|
5618
|
|
|
5619 /*
|
|
495
|
5620 * Turn an affix flag name into a number, according to the FLAG type.
|
|
|
5621 * returns zero for failure.
|
|
|
5622 */
|
|
|
5623 static unsigned
|
|
|
5624 affitem2flag(flagtype, item, fname, lnum)
|
|
|
5625 int flagtype;
|
|
|
5626 char_u *item;
|
|
|
5627 char_u *fname;
|
|
|
5628 int lnum;
|
|
|
5629 {
|
|
|
5630 unsigned res;
|
|
|
5631 char_u *p = item;
|
|
|
5632
|
|
|
5633 res = get_affitem(flagtype, &p);
|
|
|
5634 if (res == 0)
|
|
|
5635 {
|
|
498
|
5636 if (flagtype == AFT_NUM)
|
|
495
|
5637 smsg((char_u *)_("Flag is not a number in %s line %d: %s"),
|
|
|
5638 fname, lnum, item);
|
|
|
5639 else
|
|
|
5640 smsg((char_u *)_("Illegal flag in %s line %d: %s"),
|
|
|
5641 fname, lnum, item);
|
|
|
5642 }
|
|
|
5643 if (*p != NUL)
|
|
|
5644 {
|
|
|
5645 smsg((char_u *)_(e_affname), fname, lnum, item);
|
|
|
5646 return 0;
|
|
|
5647 }
|
|
|
5648
|
|
|
5649 return res;
|
|
|
5650 }
|
|
|
5651
|
|
|
5652 /*
|
|
|
5653 * Get one affix name from "*pp" and advance the pointer.
|
|
|
5654 * Returns zero for an error, still advances the pointer then.
|
|
|
5655 */
|
|
|
5656 static unsigned
|
|
|
5657 get_affitem(flagtype, pp)
|
|
|
5658 int flagtype;
|
|
|
5659 char_u **pp;
|
|
|
5660 {
|
|
|
5661 int res;
|
|
|
5662
|
|
498
|
5663 if (flagtype == AFT_NUM)
|
|
495
|
5664 {
|
|
|
5665 if (!VIM_ISDIGIT(**pp))
|
|
|
5666 {
|
|
498
|
5667 ++*pp; /* always advance, avoid getting stuck */
|
|
495
|
5668 return 0;
|
|
|
5669 }
|
|
|
5670 res = getdigits(pp);
|
|
|
5671 }
|
|
|
5672 else
|
|
|
5673 {
|
|
|
5674 #ifdef FEAT_MBYTE
|
|
|
5675 res = mb_ptr2char_adv(pp);
|
|
|
5676 #else
|
|
|
5677 res = *(*pp)++;
|
|
|
5678 #endif
|
|
498
|
5679 if (flagtype == AFT_LONG || (flagtype == AFT_CAPLONG
|
|
495
|
5680 && res >= 'A' && res <= 'Z'))
|
|
|
5681 {
|
|
|
5682 if (**pp == NUL)
|
|
|
5683 return 0;
|
|
|
5684 #ifdef FEAT_MBYTE
|
|
|
5685 res = mb_ptr2char_adv(pp) + (res << 16);
|
|
|
5686 #else
|
|
|
5687 res = *(*pp)++ + (res << 16);
|
|
|
5688 #endif
|
|
|
5689 }
|
|
|
5690 }
|
|
|
5691 return res;
|
|
|
5692 }
|
|
|
5693
|
|
|
5694 /*
|
|
|
5695 * Process the "compflags" string used in an affix file and append it to
|
|
|
5696 * spin->si_compflags.
|
|
|
5697 * The processing involves changing the affix names to ID numbers, so that
|
|
|
5698 * they fit in one byte.
|
|
|
5699 */
|
|
|
5700 static void
|
|
|
5701 process_compflags(spin, aff, compflags)
|
|
|
5702 spellinfo_T *spin;
|
|
|
5703 afffile_T *aff;
|
|
|
5704 char_u *compflags;
|
|
|
5705 {
|
|
|
5706 char_u *p;
|
|
|
5707 char_u *prevp;
|
|
|
5708 unsigned flag;
|
|
|
5709 compitem_T *ci;
|
|
|
5710 int id;
|
|
|
5711 int len;
|
|
|
5712 char_u *tp;
|
|
|
5713 char_u key[AH_KEY_LEN];
|
|
|
5714 hashitem_T *hi;
|
|
|
5715
|
|
|
5716 /* Make room for the old and the new compflags, concatenated with a / in
|
|
|
5717 * between. Processing it makes it shorter, but we don't know by how
|
|
|
5718 * much, thus allocate the maximum. */
|
|
|
5719 len = STRLEN(compflags) + 1;
|
|
|
5720 if (spin->si_compflags != NULL)
|
|
|
5721 len += STRLEN(spin->si_compflags) + 1;
|
|
|
5722 p = getroom(spin, len, FALSE);
|
|
|
5723 if (p == NULL)
|
|
|
5724 return;
|
|
|
5725 if (spin->si_compflags != NULL)
|
|
|
5726 {
|
|
|
5727 STRCPY(p, spin->si_compflags);
|
|
|
5728 STRCAT(p, "/");
|
|
|
5729 }
|
|
|
5730 spin->si_compflags = p;
|
|
|
5731 tp = p + STRLEN(p);
|
|
|
5732
|
|
|
5733 for (p = compflags; *p != NUL; )
|
|
|
5734 {
|
|
|
5735 if (vim_strchr((char_u *)"/*+[]", *p) != NULL)
|
|
|
5736 /* Copy non-flag characters directly. */
|
|
|
5737 *tp++ = *p++;
|
|
|
5738 else
|
|
|
5739 {
|
|
|
5740 /* First get the flag number, also checks validity. */
|
|
|
5741 prevp = p;
|
|
|
5742 flag = get_affitem(aff->af_flagtype, &p);
|
|
|
5743 if (flag != 0)
|
|
|
5744 {
|
|
|
5745 /* Find the flag in the hashtable. If it was used before, use
|
|
|
5746 * the existing ID. Otherwise add a new entry. */
|
|
|
5747 vim_strncpy(key, prevp, p - prevp);
|
|
|
5748 hi = hash_find(&aff->af_comp, key);
|
|
|
5749 if (!HASHITEM_EMPTY(hi))
|
|
|
5750 id = HI2CI(hi)->ci_newID;
|
|
|
5751 else
|
|
|
5752 {
|
|
|
5753 ci = (compitem_T *)getroom(spin, sizeof(compitem_T), TRUE);
|
|
|
5754 if (ci == NULL)
|
|
|
5755 break;
|
|
|
5756 STRCPY(ci->ci_key, key);
|
|
|
5757 ci->ci_flag = flag;
|
|
|
5758 /* Avoid using a flag ID that has a special meaning in a
|
|
|
5759 * regexp (also inside []). */
|
|
|
5760 do
|
|
|
5761 {
|
|
500
|
5762 check_renumber(spin);
|
|
|
5763 id = spin->si_newcompID--;
|
|
|
5764 } while (vim_strchr((char_u *)"/+*[]\\-^", id) != NULL);
|
|
495
|
5765 ci->ci_newID = id;
|
|
|
5766 hash_add(&aff->af_comp, ci->ci_key);
|
|
|
5767 }
|
|
|
5768 *tp++ = id;
|
|
|
5769 }
|
|
498
|
5770 if (aff->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5771 ++p;
|
|
|
5772 }
|
|
|
5773 }
|
|
|
5774
|
|
|
5775 *tp = NUL;
|
|
|
5776 }
|
|
|
5777
|
|
|
5778 /*
|
|
500
|
5779 * Check that the new IDs for postponed affixes and compounding don't overrun
|
|
|
5780 * each other. We have almost 255 available, but start at 0-127 to avoid
|
|
|
5781 * using two bytes for utf-8. When the 0-127 range is used up go to 128-255.
|
|
|
5782 * When that is used up an error message is given.
|
|
|
5783 */
|
|
|
5784 static void
|
|
|
5785 check_renumber(spin)
|
|
|
5786 spellinfo_T *spin;
|
|
|
5787 {
|
|
|
5788 if (spin->si_newprefID == spin->si_newcompID && spin->si_newcompID < 128)
|
|
|
5789 {
|
|
|
5790 spin->si_newprefID = 127;
|
|
|
5791 spin->si_newcompID = 255;
|
|
|
5792 }
|
|
|
5793 }
|
|
|
5794
|
|
|
5795 /*
|
|
495
|
5796 * Return TRUE if flag "flag" appears in affix list "afflist".
|
|
|
5797 */
|
|
|
5798 static int
|
|
|
5799 flag_in_afflist(flagtype, afflist, flag)
|
|
|
5800 int flagtype;
|
|
|
5801 char_u *afflist;
|
|
|
5802 unsigned flag;
|
|
|
5803 {
|
|
|
5804 char_u *p;
|
|
|
5805 unsigned n;
|
|
|
5806
|
|
|
5807 switch (flagtype)
|
|
|
5808 {
|
|
|
5809 case AFT_CHAR:
|
|
|
5810 return vim_strchr(afflist, flag) != NULL;
|
|
|
5811
|
|
498
|
5812 case AFT_CAPLONG:
|
|
|
5813 case AFT_LONG:
|
|
495
|
5814 for (p = afflist; *p != NUL; )
|
|
|
5815 {
|
|
|
5816 #ifdef FEAT_MBYTE
|
|
|
5817 n = mb_ptr2char_adv(&p);
|
|
|
5818 #else
|
|
|
5819 n = *p++;
|
|
|
5820 #endif
|
|
498
|
5821 if ((flagtype == AFT_LONG || (n >= 'A' && n <= 'Z'))
|
|
495
|
5822 && *p != NUL)
|
|
|
5823 #ifdef FEAT_MBYTE
|
|
|
5824 n = mb_ptr2char_adv(&p) + (n << 16);
|
|
|
5825 #else
|
|
|
5826 n = *p++ + (n << 16);
|
|
|
5827 #endif
|
|
|
5828 if (n == flag)
|
|
|
5829 return TRUE;
|
|
|
5830 }
|
|
|
5831 break;
|
|
|
5832
|
|
498
|
5833 case AFT_NUM:
|
|
495
|
5834 for (p = afflist; *p != NUL; )
|
|
|
5835 {
|
|
|
5836 n = getdigits(&p);
|
|
|
5837 if (n == flag)
|
|
|
5838 return TRUE;
|
|
|
5839 if (*p != NUL) /* skip over comma */
|
|
|
5840 ++p;
|
|
|
5841 }
|
|
|
5842 break;
|
|
|
5843 }
|
|
|
5844 return FALSE;
|
|
|
5845 }
|
|
|
5846
|
|
|
5847 /*
|
|
|
5848 * Give a warning when "spinval" and "affval" numbers are set and not the same.
|
|
|
5849 */
|
|
|
5850 static void
|
|
|
5851 aff_check_number(spinval, affval, name)
|
|
|
5852 int spinval;
|
|
|
5853 int affval;
|
|
|
5854 char *name;
|
|
|
5855 {
|
|
|
5856 if (spinval != 0 && spinval != affval)
|
|
|
5857 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
5858 }
|
|
|
5859
|
|
|
5860 /*
|
|
|
5861 * Give a warning when "spinval" and "affval" strings are set and not the same.
|
|
|
5862 */
|
|
|
5863 static void
|
|
|
5864 aff_check_string(spinval, affval, name)
|
|
|
5865 char_u *spinval;
|
|
|
5866 char_u *affval;
|
|
|
5867 char *name;
|
|
|
5868 {
|
|
|
5869 if (spinval != NULL && STRCMP(spinval, affval) != 0)
|
|
|
5870 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
5871 }
|
|
|
5872
|
|
|
5873 /*
|
|
339
|
5874 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
5875 * NULL as equal.
|
|
|
5876 */
|
|
|
5877 static int
|
|
|
5878 str_equal(s1, s2)
|
|
|
5879 char_u *s1;
|
|
|
5880 char_u *s2;
|
|
|
5881 {
|
|
|
5882 if (s1 == NULL || s2 == NULL)
|
|
|
5883 return s1 == s2;
|
|
|
5884 return STRCMP(s1, s2) == 0;
|
|
|
5885 }
|
|
|
5886
|
|
|
5887 /*
|
|
323
|
5888 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
5889 * They are stored case-folded.
|
|
|
5890 */
|
|
|
5891 static void
|
|
|
5892 add_fromto(spin, gap, from, to)
|
|
|
5893 spellinfo_T *spin;
|
|
|
5894 garray_T *gap;
|
|
|
5895 char_u *from;
|
|
|
5896 char_u *to;
|
|
|
5897 {
|
|
|
5898 fromto_T *ftp;
|
|
|
5899 char_u word[MAXWLEN];
|
|
|
5900
|
|
|
5901 if (ga_grow(gap, 1) == OK)
|
|
|
5902 {
|
|
|
5903 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
5904 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
5905 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
5906 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
5907 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
5908 ++gap->ga_len;
|
|
|
5909 }
|
|
|
5910 }
|
|
|
5911
|
|
|
5912 /*
|
|
|
5913 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
5914 */
|
|
|
5915 static int
|
|
|
5916 sal_to_bool(s)
|
|
|
5917 char_u *s;
|
|
|
5918 {
|
|
|
5919 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
5920 }
|
|
|
5921
|
|
|
5922 /*
|
|
240
|
5923 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
5924 * When "s" is NULL FALSE is returned.
|
|
|
5925 */
|
|
|
5926 static int
|
|
|
5927 has_non_ascii(s)
|
|
|
5928 char_u *s;
|
|
|
5929 {
|
|
|
5930 char_u *p;
|
|
|
5931
|
|
|
5932 if (s != NULL)
|
|
|
5933 for (p = s; *p != NUL; ++p)
|
|
|
5934 if (*p >= 128)
|
|
|
5935 return TRUE;
|
|
|
5936 return FALSE;
|
|
|
5937 }
|
|
|
5938
|
|
|
5939 /*
|
|
236
|
5940 * Free the structure filled by spell_read_aff().
|
|
|
5941 */
|
|
|
5942 static void
|
|
|
5943 spell_free_aff(aff)
|
|
|
5944 afffile_T *aff;
|
|
|
5945 {
|
|
|
5946 hashtab_T *ht;
|
|
|
5947 hashitem_T *hi;
|
|
|
5948 int todo;
|
|
|
5949 affheader_T *ah;
|
|
300
|
5950 affentry_T *ae;
|
|
236
|
5951
|
|
|
5952 vim_free(aff->af_enc);
|
|
|
5953
|
|
339
|
5954 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
5955 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
5956 {
|
|
|
5957 todo = ht->ht_used;
|
|
|
5958 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
5959 {
|
|
|
5960 if (!HASHITEM_EMPTY(hi))
|
|
|
5961 {
|
|
|
5962 --todo;
|
|
|
5963 ah = HI2AH(hi);
|
|
300
|
5964 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
5965 vim_free(ae->ae_prog);
|
|
236
|
5966 }
|
|
|
5967 }
|
|
|
5968 if (ht == &aff->af_suff)
|
|
|
5969 break;
|
|
|
5970 }
|
|
300
|
5971
|
|
236
|
5972 hash_clear(&aff->af_pref);
|
|
|
5973 hash_clear(&aff->af_suff);
|
|
495
|
5974 hash_clear(&aff->af_comp);
|
|
236
|
5975 }
|
|
|
5976
|
|
|
5977 /*
|
|
300
|
5978 * Read dictionary file "fname".
|
|
236
|
5979 * Returns OK or FAIL;
|
|
|
5980 */
|
|
|
5981 static int
|
|
474
|
5982 spell_read_dic(spin, fname, affile)
|
|
|
5983 spellinfo_T *spin;
|
|
236
|
5984 char_u *fname;
|
|
300
|
5985 afffile_T *affile;
|
|
236
|
5986 {
|
|
300
|
5987 hashtab_T ht;
|
|
236
|
5988 char_u line[MAXLINELEN];
|
|
481
|
5989 char_u *p;
|
|
300
|
5990 char_u *afflist;
|
|
483
|
5991 char_u store_afflist[MAXWLEN];
|
|
|
5992 int pfxlen;
|
|
|
5993 int need_affix;
|
|
300
|
5994 char_u *dw;
|
|
236
|
5995 char_u *pc;
|
|
|
5996 char_u *w;
|
|
|
5997 int l;
|
|
|
5998 hash_T hash;
|
|
|
5999 hashitem_T *hi;
|
|
|
6000 FILE *fd;
|
|
|
6001 int lnum = 1;
|
|
300
|
6002 int non_ascii = 0;
|
|
|
6003 int retval = OK;
|
|
|
6004 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
6005 int flags;
|
|
481
|
6006 int duplicate = 0;
|
|
236
|
6007
|
|
300
|
6008 /*
|
|
|
6009 * Open the file.
|
|
|
6010 */
|
|
310
|
6011 fd = mch_fopen((char *)fname, "r");
|
|
236
|
6012 if (fd == NULL)
|
|
|
6013 {
|
|
|
6014 EMSG2(_(e_notopen), fname);
|
|
|
6015 return FAIL;
|
|
|
6016 }
|
|
|
6017
|
|
300
|
6018 /* The hashtable is only used to detect duplicated words. */
|
|
|
6019 hash_init(&ht);
|
|
|
6020
|
|
625
|
6021 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
6022 _("Reading dictionary file %s ..."), fname);
|
|
|
6023 spell_message(spin, IObuff);
|
|
236
|
6024
|
|
474
|
6025 /* start with a message for the first line */
|
|
|
6026 spin->si_msg_count = 999999;
|
|
|
6027
|
|
236
|
6028 /* Read and ignore the first line: word count. */
|
|
|
6029 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
6030 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
6031 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
6032
|
|
|
6033 /*
|
|
|
6034 * Read all the lines in the file one by one.
|
|
|
6035 * The words are converted to 'encoding' here, before being added to
|
|
|
6036 * the hashtable.
|
|
|
6037 */
|
|
255
|
6038 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
6039 {
|
|
255
|
6040 line_breakcheck();
|
|
236
|
6041 ++lnum;
|
|
455
|
6042 if (line[0] == '#' || line[0] == '/')
|
|
351
|
6043 continue; /* comment line */
|
|
236
|
6044
|
|
300
|
6045 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
6046 * the word is kept to allow e.g., "et al.". */
|
|
236
|
6047 l = STRLEN(line);
|
|
|
6048 while (l > 0 && line[l - 1] <= ' ')
|
|
|
6049 --l;
|
|
|
6050 if (l == 0)
|
|
|
6051 continue; /* empty line */
|
|
|
6052 line[l] = NUL;
|
|
|
6053
|
|
638
|
6054 /* Truncate the word at the "/", set "afflist" to what follows.
|
|
|
6055 * Replace "\/" by "/" and "\\" by "\". */
|
|
481
|
6056 afflist = NULL;
|
|
|
6057 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
6058 {
|
|
638
|
6059 if (*p == '\\' && (p[1] == '\\' || p[1] == '/'))
|
|
|
6060 mch_memmove(p, p + 1, STRLEN(p));
|
|
481
|
6061 else if (*p == '/')
|
|
|
6062 {
|
|
|
6063 *p = NUL;
|
|
|
6064 afflist = p + 1;
|
|
|
6065 break;
|
|
|
6066 }
|
|
|
6067 }
|
|
236
|
6068
|
|
300
|
6069 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
6070 if (spin->si_ascii && has_non_ascii(line))
|
|
|
6071 {
|
|
|
6072 ++non_ascii;
|
|
240
|
6073 continue;
|
|
300
|
6074 }
|
|
240
|
6075
|
|
310
|
6076 #ifdef FEAT_MBYTE
|
|
236
|
6077 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
6078 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
6079 {
|
|
300
|
6080 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
6081 if (pc == NULL)
|
|
|
6082 {
|
|
|
6083 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
6084 fname, lnum, line);
|
|
|
6085 continue;
|
|
|
6086 }
|
|
236
|
6087 w = pc;
|
|
|
6088 }
|
|
|
6089 else
|
|
310
|
6090 #endif
|
|
236
|
6091 {
|
|
|
6092 pc = NULL;
|
|
|
6093 w = line;
|
|
|
6094 }
|
|
|
6095
|
|
474
|
6096 /* This takes time, print a message every 10000 words. */
|
|
|
6097 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
6098 {
|
|
|
6099 spin->si_msg_count = 0;
|
|
339
|
6100 vim_snprintf((char *)message, sizeof(message),
|
|
|
6101 _("line %6d, word %6d - %s"),
|
|
|
6102 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
6103 msg_start();
|
|
|
6104 msg_puts_long_attr(message, 0);
|
|
|
6105 msg_clr_eos();
|
|
|
6106 msg_didout = FALSE;
|
|
|
6107 msg_col = 0;
|
|
|
6108 out_flush();
|
|
|
6109 }
|
|
|
6110
|
|
300
|
6111 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
6112 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
6113 if (dw == NULL)
|
|
300
|
6114 retval = FAIL;
|
|
|
6115 vim_free(pc);
|
|
|
6116 if (retval == FAIL)
|
|
236
|
6117 break;
|
|
|
6118
|
|
300
|
6119 hash = hash_hash(dw);
|
|
|
6120 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
6121 if (!HASHITEM_EMPTY(hi))
|
|
481
|
6122 {
|
|
|
6123 if (p_verbose > 0)
|
|
|
6124 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
6125 fname, lnum, dw);
|
|
481
|
6126 else if (duplicate == 0)
|
|
|
6127 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
6128 fname, lnum, dw);
|
|
|
6129 ++duplicate;
|
|
|
6130 }
|
|
236
|
6131 else
|
|
300
|
6132 hash_add_item(&ht, hi, dw, hash);
|
|
|
6133
|
|
307
|
6134 flags = 0;
|
|
483
|
6135 store_afflist[0] = NUL;
|
|
|
6136 pfxlen = 0;
|
|
|
6137 need_affix = FALSE;
|
|
307
|
6138 if (afflist != NULL)
|
|
|
6139 {
|
|
|
6140 /* Check for affix name that stands for keep-case word and stands
|
|
|
6141 * for rare word (if defined). */
|
|
621
|
6142 if (affile->af_keepcase != 0 && flag_in_afflist(
|
|
|
6143 affile->af_flagtype, afflist, affile->af_keepcase))
|
|
389
|
6144 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
621
|
6145 if (affile->af_rare != 0 && flag_in_afflist(
|
|
|
6146 affile->af_flagtype, afflist, affile->af_rare))
|
|
307
|
6147 flags |= WF_RARE;
|
|
495
|
6148 if (affile->af_bad != 0 && flag_in_afflist(
|
|
|
6149 affile->af_flagtype, afflist, affile->af_bad))
|
|
346
|
6150 flags |= WF_BANNED;
|
|
495
|
6151 if (affile->af_needaffix != 0 && flag_in_afflist(
|
|
|
6152 affile->af_flagtype, afflist, affile->af_needaffix))
|
|
483
|
6153 need_affix = TRUE;
|
|
500
|
6154 if (affile->af_needcomp != 0 && flag_in_afflist(
|
|
|
6155 affile->af_flagtype, afflist, affile->af_needcomp))
|
|
|
6156 flags |= WF_NEEDCOMP;
|
|
339
|
6157
|
|
|
6158 if (affile->af_pfxpostpone)
|
|
|
6159 /* Need to store the list of prefix IDs with the word. */
|
|
483
|
6160 pfxlen = get_pfxlist(affile, afflist, store_afflist);
|
|
|
6161
|
|
|
6162 if (spin->si_compflags != NULL)
|
|
|
6163 /* Need to store the list of compound flags with the word.
|
|
|
6164 * Concatenate them to the list of prefix IDs. */
|
|
495
|
6165 get_compflags(affile, afflist, store_afflist + pfxlen);
|
|
307
|
6166 }
|
|
|
6167
|
|
300
|
6168 /* Add the word to the word tree(s). */
|
|
483
|
6169 if (store_word(spin, dw, flags, spin->si_region,
|
|
|
6170 store_afflist, need_affix) == FAIL)
|
|
300
|
6171 retval = FAIL;
|
|
236
|
6172
|
|
300
|
6173 if (afflist != NULL)
|
|
|
6174 {
|
|
|
6175 /* Find all matching suffixes and add the resulting words.
|
|
|
6176 * Additionally do matching prefixes that combine. */
|
|
474
|
6177 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
6178 &affile->af_suff, &affile->af_pref,
|
|
483
|
6179 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
6180 retval = FAIL;
|
|
|
6181
|
|
|
6182 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
6183 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
6184 &affile->af_pref, NULL,
|
|
483
|
6185 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
6186 retval = FAIL;
|
|
|
6187 }
|
|
236
|
6188 }
|
|
|
6189
|
|
481
|
6190 if (duplicate > 0)
|
|
|
6191 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
6192 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
6193 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
6194 non_ascii, fname);
|
|
300
|
6195 hash_clear(&ht);
|
|
|
6196
|
|
236
|
6197 fclose(fd);
|
|
300
|
6198 return retval;
|
|
236
|
6199 }
|
|
|
6200
|
|
|
6201 /*
|
|
339
|
6202 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
6203 * Used for PFXPOSTPONE.
|
|
483
|
6204 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL
|
|
|
6205 * and return the number of affixes.
|
|
|
6206 */
|
|
|
6207 static int
|
|
|
6208 get_pfxlist(affile, afflist, store_afflist)
|
|
339
|
6209 afffile_T *affile;
|
|
|
6210 char_u *afflist;
|
|
483
|
6211 char_u *store_afflist;
|
|
339
|
6212 {
|
|
|
6213 char_u *p;
|
|
495
|
6214 char_u *prevp;
|
|
483
|
6215 int cnt = 0;
|
|
495
|
6216 int id;
|
|
|
6217 char_u key[AH_KEY_LEN];
|
|
339
|
6218 hashitem_T *hi;
|
|
|
6219
|
|
495
|
6220 for (p = afflist; *p != NUL; )
|
|
|
6221 {
|
|
|
6222 prevp = p;
|
|
|
6223 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
6224 {
|
|
|
6225 /* A flag is a postponed prefix flag if it appears in "af_pref"
|
|
|
6226 * and it's ID is not zero. */
|
|
|
6227 vim_strncpy(key, prevp, p - prevp);
|
|
|
6228 hi = hash_find(&affile->af_pref, key);
|
|
|
6229 if (!HASHITEM_EMPTY(hi))
|
|
|
6230 {
|
|
|
6231 id = HI2AH(hi)->ah_newID;
|
|
|
6232 if (id != 0)
|
|
|
6233 store_afflist[cnt++] = id;
|
|
|
6234 }
|
|
|
6235 }
|
|
498
|
6236 if (affile->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
6237 ++p;
|
|
483
|
6238 }
|
|
|
6239
|
|
|
6240 store_afflist[cnt] = NUL;
|
|
|
6241 return cnt;
|
|
339
|
6242 }
|
|
|
6243
|
|
|
6244 /*
|
|
495
|
6245 * Get the list of compound IDs from the affix list "afflist" that are used
|
|
|
6246 * for compound words.
|
|
483
|
6247 * Puts the flags in "store_afflist[]".
|
|
|
6248 */
|
|
|
6249 static void
|
|
495
|
6250 get_compflags(affile, afflist, store_afflist)
|
|
|
6251 afffile_T *affile;
|
|
481
|
6252 char_u *afflist;
|
|
483
|
6253 char_u *store_afflist;
|
|
481
|
6254 {
|
|
|
6255 char_u *p;
|
|
495
|
6256 char_u *prevp;
|
|
483
|
6257 int cnt = 0;
|
|
495
|
6258 char_u key[AH_KEY_LEN];
|
|
|
6259 hashitem_T *hi;
|
|
|
6260
|
|
|
6261 for (p = afflist; *p != NUL; )
|
|
|
6262 {
|
|
|
6263 prevp = p;
|
|
|
6264 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
6265 {
|
|
|
6266 /* A flag is a compound flag if it appears in "af_comp". */
|
|
|
6267 vim_strncpy(key, prevp, p - prevp);
|
|
|
6268 hi = hash_find(&affile->af_comp, key);
|
|
|
6269 if (!HASHITEM_EMPTY(hi))
|
|
|
6270 store_afflist[cnt++] = HI2CI(hi)->ci_newID;
|
|
|
6271 }
|
|
498
|
6272 if (affile->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
6273 ++p;
|
|
|
6274 }
|
|
483
|
6275
|
|
|
6276 store_afflist[cnt] = NUL;
|
|
481
|
6277 }
|
|
|
6278
|
|
|
6279 /*
|
|
300
|
6280 * Apply affixes to a word and store the resulting words.
|
|
|
6281 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
6282 * prefixes or suffixes.
|
|
|
6283 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
6284 * the resulting words for combining affixes.
|
|
|
6285 *
|
|
|
6286 * Returns FAIL when out of memory.
|
|
236
|
6287 */
|
|
300
|
6288 static int
|
|
483
|
6289 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags,
|
|
|
6290 pfxlist, pfxlen)
|
|
474
|
6291 spellinfo_T *spin; /* spell info */
|
|
300
|
6292 char_u *word; /* basic word start */
|
|
|
6293 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
6294 afffile_T *affile;
|
|
300
|
6295 hashtab_T *ht;
|
|
|
6296 hashtab_T *xht;
|
|
|
6297 int comb; /* only use affixes that combine */
|
|
307
|
6298 int flags; /* flags for the word */
|
|
339
|
6299 char_u *pfxlist; /* list of prefix IDs */
|
|
483
|
6300 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest
|
|
|
6301 * is compound flags */
|
|
236
|
6302 {
|
|
|
6303 int todo;
|
|
|
6304 hashitem_T *hi;
|
|
300
|
6305 affheader_T *ah;
|
|
|
6306 affentry_T *ae;
|
|
|
6307 regmatch_T regmatch;
|
|
|
6308 char_u newword[MAXWLEN];
|
|
|
6309 int retval = OK;
|
|
|
6310 int i;
|
|
|
6311 char_u *p;
|
|
366
|
6312 int use_flags;
|
|
390
|
6313 char_u *use_pfxlist;
|
|
483
|
6314 char_u pfx_pfxlist[MAXWLEN];
|
|
|
6315 size_t wordlen = STRLEN(word);
|
|
236
|
6316
|
|
300
|
6317 todo = ht->ht_used;
|
|
|
6318 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
6319 {
|
|
|
6320 if (!HASHITEM_EMPTY(hi))
|
|
|
6321 {
|
|
|
6322 --todo;
|
|
300
|
6323 ah = HI2AH(hi);
|
|
236
|
6324
|
|
300
|
6325 /* Check that the affix combines, if required, and that the word
|
|
|
6326 * supports this affix. */
|
|
495
|
6327 if ((!comb || ah->ah_combine) && flag_in_afflist(
|
|
|
6328 affile->af_flagtype, afflist, ah->ah_flag))
|
|
236
|
6329 {
|
|
300
|
6330 /* Loop over all affix entries with this name. */
|
|
|
6331 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
6332 {
|
|
300
|
6333 /* Check the condition. It's not logical to match case
|
|
|
6334 * here, but it is required for compatibility with
|
|
339
|
6335 * Myspell.
|
|
481
|
6336 * Another requirement from Myspell is that the chop
|
|
|
6337 * string is shorter than the word itself.
|
|
339
|
6338 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
6339 * prefixes with a chop string. */
|
|
300
|
6340 regmatch.regprog = ae->ae_prog;
|
|
|
6341 regmatch.rm_ic = FALSE;
|
|
339
|
6342 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
6343 || ae->ae_chop != NULL)
|
|
481
|
6344 && (ae->ae_chop == NULL
|
|
|
6345 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
6346 && (ae->ae_prog == NULL
|
|
|
6347 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
6348 {
|
|
|
6349 /* Match. Remove the chop and add the affix. */
|
|
|
6350 if (xht == NULL)
|
|
240
|
6351 {
|
|
300
|
6352 /* prefix: chop/add at the start of the word */
|
|
|
6353 if (ae->ae_add == NULL)
|
|
|
6354 *newword = NUL;
|
|
|
6355 else
|
|
|
6356 STRCPY(newword, ae->ae_add);
|
|
|
6357 p = word;
|
|
|
6358 if (ae->ae_chop != NULL)
|
|
310
|
6359 {
|
|
300
|
6360 /* Skip chop string. */
|
|
310
|
6361 #ifdef FEAT_MBYTE
|
|
|
6362 if (has_mbyte)
|
|
324
|
6363 {
|
|
310
|
6364 i = mb_charlen(ae->ae_chop);
|
|
324
|
6365 for ( ; i > 0; --i)
|
|
|
6366 mb_ptr_adv(p);
|
|
|
6367 }
|
|
310
|
6368 else
|
|
|
6369 #endif
|
|
324
|
6370 p += STRLEN(ae->ae_chop);
|
|
310
|
6371 }
|
|
300
|
6372 STRCAT(newword, p);
|
|
|
6373 }
|
|
|
6374 else
|
|
|
6375 {
|
|
|
6376 /* suffix: chop/add at the end of the word */
|
|
|
6377 STRCPY(newword, word);
|
|
|
6378 if (ae->ae_chop != NULL)
|
|
|
6379 {
|
|
|
6380 /* Remove chop string. */
|
|
|
6381 p = newword + STRLEN(newword);
|
|
376
|
6382 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
6383 for ( ; i > 0; --i)
|
|
300
|
6384 mb_ptr_back(newword, p);
|
|
|
6385 *p = NUL;
|
|
|
6386 }
|
|
|
6387 if (ae->ae_add != NULL)
|
|
|
6388 STRCAT(newword, ae->ae_add);
|
|
240
|
6389 }
|
|
|
6390
|
|
366
|
6391 /* Obey the "rare" flag of the affix. */
|
|
|
6392 if (ae->ae_rare)
|
|
|
6393 use_flags = flags | WF_RARE;
|
|
|
6394 else
|
|
|
6395 use_flags = flags;
|
|
483
|
6396
|
|
|
6397 /* Obey the "nocomp" flag of the affix: don't use the
|
|
|
6398 * compound flags. */
|
|
390
|
6399 use_pfxlist = pfxlist;
|
|
483
|
6400 if (ae->ae_nocomp && pfxlist != NULL)
|
|
|
6401 {
|
|
|
6402 vim_strncpy(pfx_pfxlist, pfxlist, pfxlen);
|
|
|
6403 use_pfxlist = pfx_pfxlist;
|
|
|
6404 }
|
|
390
|
6405
|
|
|
6406 /* When there are postponed prefixes... */
|
|
393
|
6407 if (spin->si_prefroot != NULL
|
|
|
6408 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
6409 {
|
|
|
6410 /* ... add a flag to indicate an affix was used. */
|
|
|
6411 use_flags |= WF_HAS_AFF;
|
|
|
6412
|
|
|
6413 /* ... don't use a prefix list if combining
|
|
483
|
6414 * affixes is not allowed. But do use the
|
|
|
6415 * compound flags after them. */
|
|
|
6416 if ((!ah->ah_combine || comb) && pfxlist != NULL)
|
|
|
6417 use_pfxlist += pfxlen;
|
|
390
|
6418 }
|
|
366
|
6419
|
|
300
|
6420 /* Store the modified word. */
|
|
474
|
6421 if (store_word(spin, newword, use_flags,
|
|
483
|
6422 spin->si_region, use_pfxlist, FALSE) == FAIL)
|
|
300
|
6423 retval = FAIL;
|
|
236
|
6424
|
|
300
|
6425 /* When added a suffix and combining is allowed also
|
|
|
6426 * try adding prefixes additionally. */
|
|
|
6427 if (xht != NULL && ah->ah_combine)
|
|
474
|
6428 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
6429 xht, NULL, TRUE,
|
|
483
|
6430 use_flags, use_pfxlist, pfxlen) == FAIL)
|
|
300
|
6431 retval = FAIL;
|
|
236
|
6432 }
|
|
|
6433 }
|
|
|
6434 }
|
|
|
6435 }
|
|
|
6436 }
|
|
|
6437
|
|
|
6438 return retval;
|
|
|
6439 }
|
|
|
6440
|
|
|
6441 /*
|
|
300
|
6442 * Read a file with a list of words.
|
|
236
|
6443 */
|
|
|
6444 static int
|
|
474
|
6445 spell_read_wordfile(spin, fname)
|
|
|
6446 spellinfo_T *spin;
|
|
300
|
6447 char_u *fname;
|
|
236
|
6448 {
|
|
300
|
6449 FILE *fd;
|
|
|
6450 long lnum = 0;
|
|
|
6451 char_u rline[MAXLINELEN];
|
|
|
6452 char_u *line;
|
|
|
6453 char_u *pc = NULL;
|
|
381
|
6454 char_u *p;
|
|
300
|
6455 int l;
|
|
|
6456 int retval = OK;
|
|
|
6457 int did_word = FALSE;
|
|
|
6458 int non_ascii = 0;
|
|
307
|
6459 int flags;
|
|
316
|
6460 int regionmask;
|
|
236
|
6461
|
|
300
|
6462 /*
|
|
|
6463 * Open the file.
|
|
|
6464 */
|
|
310
|
6465 fd = mch_fopen((char *)fname, "r");
|
|
300
|
6466 if (fd == NULL)
|
|
236
|
6467 {
|
|
300
|
6468 EMSG2(_(e_notopen), fname);
|
|
|
6469 return FAIL;
|
|
236
|
6470 }
|
|
|
6471
|
|
625
|
6472 vim_snprintf((char *)IObuff, IOSIZE, _("Reading word file %s ..."), fname);
|
|
|
6473 spell_message(spin, IObuff);
|
|
300
|
6474
|
|
|
6475 /*
|
|
|
6476 * Read all the lines in the file one by one.
|
|
|
6477 */
|
|
|
6478 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
6479 {
|
|
|
6480 line_breakcheck();
|
|
|
6481 ++lnum;
|
|
|
6482
|
|
|
6483 /* Skip comment lines. */
|
|
|
6484 if (*rline == '#')
|
|
|
6485 continue;
|
|
|
6486
|
|
|
6487 /* Remove CR, LF and white space from the end. */
|
|
|
6488 l = STRLEN(rline);
|
|
|
6489 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
6490 --l;
|
|
|
6491 if (l == 0)
|
|
|
6492 continue; /* empty or blank line */
|
|
|
6493 rline[l] = NUL;
|
|
|
6494
|
|
|
6495 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
6496 vim_free(pc);
|
|
310
|
6497 #ifdef FEAT_MBYTE
|
|
300
|
6498 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
6499 {
|
|
|
6500 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
6501 if (pc == NULL)
|
|
|
6502 {
|
|
|
6503 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
6504 fname, lnum, rline);
|
|
|
6505 continue;
|
|
|
6506 }
|
|
|
6507 line = pc;
|
|
|
6508 }
|
|
|
6509 else
|
|
310
|
6510 #endif
|
|
300
|
6511 {
|
|
|
6512 pc = NULL;
|
|
|
6513 line = rline;
|
|
|
6514 }
|
|
|
6515
|
|
307
|
6516 if (*line == '/')
|
|
300
|
6517 {
|
|
307
|
6518 ++line;
|
|
|
6519 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
6520 {
|
|
|
6521 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
6522 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
6523 fname, lnum, line - 1);
|
|
300
|
6524 else if (did_word)
|
|
316
|
6525 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
6526 fname, lnum, line - 1);
|
|
300
|
6527 else
|
|
|
6528 {
|
|
310
|
6529 #ifdef FEAT_MBYTE
|
|
|
6530 char_u *enc;
|
|
|
6531
|
|
300
|
6532 /* Setup for conversion to 'encoding'. */
|
|
316
|
6533 line += 10;
|
|
|
6534 enc = enc_canonize(line);
|
|
300
|
6535 if (enc != NULL && !spin->si_ascii
|
|
|
6536 && convert_setup(&spin->si_conv, enc,
|
|
|
6537 p_enc) == FAIL)
|
|
|
6538 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
6539 fname, line, p_enc);
|
|
300
|
6540 vim_free(enc);
|
|
375
|
6541 spin->si_conv.vc_fail = TRUE;
|
|
310
|
6542 #else
|
|
|
6543 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
6544 #endif
|
|
300
|
6545 }
|
|
307
|
6546 continue;
|
|
300
|
6547 }
|
|
307
|
6548
|
|
316
|
6549 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
6550 {
|
|
|
6551 if (spin->si_region_count > 1)
|
|
|
6552 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
6553 fname, lnum, line);
|
|
|
6554 else
|
|
|
6555 {
|
|
|
6556 line += 8;
|
|
|
6557 if (STRLEN(line) > 16)
|
|
|
6558 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
6559 fname, lnum, line);
|
|
|
6560 else
|
|
|
6561 {
|
|
|
6562 spin->si_region_count = STRLEN(line) / 2;
|
|
|
6563 STRCPY(spin->si_region_name, line);
|
|
389
|
6564
|
|
|
6565 /* Adjust the mask for a word valid in all regions. */
|
|
|
6566 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
6567 }
|
|
|
6568 }
|
|
|
6569 continue;
|
|
|
6570 }
|
|
|
6571
|
|
381
|
6572 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
6573 fname, lnum, line - 1);
|
|
|
6574 continue;
|
|
|
6575 }
|
|
|
6576
|
|
|
6577 flags = 0;
|
|
|
6578 regionmask = spin->si_region;
|
|
|
6579
|
|
|
6580 /* Check for flags and region after a slash. */
|
|
|
6581 p = vim_strchr(line, '/');
|
|
|
6582 if (p != NULL)
|
|
|
6583 {
|
|
|
6584 *p++ = NUL;
|
|
|
6585 while (*p != NUL)
|
|
|
6586 {
|
|
|
6587 if (*p == '=') /* keep-case word */
|
|
389
|
6588 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
6589 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
6590 flags |= WF_BANNED;
|
|
|
6591 else if (*p == '?') /* Rare word. */
|
|
|
6592 flags |= WF_RARE;
|
|
|
6593 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
6594 {
|
|
381
|
6595 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
6596 regionmask = 0;
|
|
|
6597 flags |= WF_REGION;
|
|
|
6598
|
|
|
6599 l = *p - '0';
|
|
316
|
6600 if (l > spin->si_region_count)
|
|
|
6601 {
|
|
|
6602 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
6603 fname, lnum, p);
|
|
316
|
6604 break;
|
|
|
6605 }
|
|
|
6606 regionmask |= 1 << (l - 1);
|
|
|
6607 }
|
|
381
|
6608 else
|
|
|
6609 {
|
|
|
6610 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
6611 fname, lnum, p);
|
|
|
6612 break;
|
|
|
6613 }
|
|
|
6614 ++p;
|
|
307
|
6615 }
|
|
300
|
6616 }
|
|
|
6617
|
|
|
6618 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
6619 if (spin->si_ascii && has_non_ascii(line))
|
|
|
6620 {
|
|
|
6621 ++non_ascii;
|
|
|
6622 continue;
|
|
|
6623 }
|
|
|
6624
|
|
|
6625 /* Normal word: store it. */
|
|
483
|
6626 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL)
|
|
300
|
6627 {
|
|
|
6628 retval = FAIL;
|
|
|
6629 break;
|
|
|
6630 }
|
|
|
6631 did_word = TRUE;
|
|
|
6632 }
|
|
|
6633
|
|
|
6634 vim_free(pc);
|
|
|
6635 fclose(fd);
|
|
|
6636
|
|
625
|
6637 if (spin->si_ascii && non_ascii > 0)
|
|
|
6638 {
|
|
|
6639 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
6640 _("Ignored %d words with non-ASCII characters"), non_ascii);
|
|
|
6641 spell_message(spin, IObuff);
|
|
|
6642 }
|
|
|
6643
|
|
300
|
6644 return retval;
|
|
236
|
6645 }
|
|
|
6646
|
|
|
6647 /*
|
|
300
|
6648 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
6649 * This avoids calling free() for every little struct we use (and keeping
|
|
|
6650 * track of them).
|
|
300
|
6651 * The memory is cleared to all zeros.
|
|
|
6652 * Returns NULL when out of memory.
|
|
|
6653 */
|
|
|
6654 static void *
|
|
474
|
6655 getroom(spin, len, align)
|
|
|
6656 spellinfo_T *spin;
|
|
447
|
6657 size_t len; /* length needed */
|
|
|
6658 int align; /* align for pointer */
|
|
300
|
6659 {
|
|
|
6660 char_u *p;
|
|
474
|
6661 sblock_T *bl = spin->si_blocks;
|
|
300
|
6662
|
|
447
|
6663 if (align && bl != NULL)
|
|
|
6664 /* Round size up for alignment. On some systems structures need to be
|
|
|
6665 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
6666 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
6667 & ~(sizeof(char *) - 1);
|
|
|
6668
|
|
300
|
6669 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
6670 {
|
|
|
6671 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
6672 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
6673 if (bl == NULL)
|
|
|
6674 return NULL;
|
|
474
|
6675 bl->sb_next = spin->si_blocks;
|
|
|
6676 spin->si_blocks = bl;
|
|
300
|
6677 bl->sb_used = 0;
|
|
482
|
6678 ++spin->si_blocks_cnt;
|
|
300
|
6679 }
|
|
|
6680
|
|
|
6681 p = bl->sb_data + bl->sb_used;
|
|
|
6682 bl->sb_used += len;
|
|
|
6683
|
|
|
6684 return p;
|
|
|
6685 }
|
|
|
6686
|
|
|
6687 /*
|
|
|
6688 * Make a copy of a string into memory allocated with getroom().
|
|
|
6689 */
|
|
|
6690 static char_u *
|
|
474
|
6691 getroom_save(spin, s)
|
|
|
6692 spellinfo_T *spin;
|
|
300
|
6693 char_u *s;
|
|
|
6694 {
|
|
|
6695 char_u *sc;
|
|
|
6696
|
|
474
|
6697 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
6698 if (sc != NULL)
|
|
|
6699 STRCPY(sc, s);
|
|
|
6700 return sc;
|
|
|
6701 }
|
|
|
6702
|
|
|
6703
|
|
|
6704 /*
|
|
|
6705 * Free the list of allocated sblock_T.
|
|
236
|
6706 */
|
|
|
6707 static void
|
|
300
|
6708 free_blocks(bl)
|
|
|
6709 sblock_T *bl;
|
|
236
|
6710 {
|
|
300
|
6711 sblock_T *next;
|
|
236
|
6712
|
|
300
|
6713 while (bl != NULL)
|
|
236
|
6714 {
|
|
300
|
6715 next = bl->sb_next;
|
|
|
6716 vim_free(bl);
|
|
|
6717 bl = next;
|
|
236
|
6718 }
|
|
|
6719 }
|
|
|
6720
|
|
|
6721 /*
|
|
300
|
6722 * Allocate the root of a word tree.
|
|
236
|
6723 */
|
|
300
|
6724 static wordnode_T *
|
|
474
|
6725 wordtree_alloc(spin)
|
|
|
6726 spellinfo_T *spin;
|
|
|
6727 {
|
|
|
6728 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
6729 }
|
|
|
6730
|
|
|
6731 /*
|
|
300
|
6732 * Store a word in the tree(s).
|
|
390
|
6733 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
6734 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
6735 * used to find suggestions.
|
|
300
|
6736 * For a keep-case word also store it in the keep-case tree.
|
|
482
|
6737 * When "pfxlist" is not NULL store the word for each postponed prefix ID and
|
|
|
6738 * compound flag.
|
|
236
|
6739 */
|
|
|
6740 static int
|
|
483
|
6741 store_word(spin, word, flags, region, pfxlist, need_affix)
|
|
474
|
6742 spellinfo_T *spin;
|
|
300
|
6743 char_u *word;
|
|
307
|
6744 int flags; /* extra flags, WF_BANNED */
|
|
316
|
6745 int region; /* supported region(s) */
|
|
339
|
6746 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
483
|
6747 int need_affix; /* only store word with affix ID */
|
|
236
|
6748 {
|
|
300
|
6749 int len = STRLEN(word);
|
|
|
6750 int ct = captype(word, word + len);
|
|
|
6751 char_u foldword[MAXWLEN];
|
|
339
|
6752 int res = OK;
|
|
|
6753 char_u *p;
|
|
236
|
6754
|
|
323
|
6755 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
6756 for (p = pfxlist; res == OK; ++p)
|
|
|
6757 {
|
|
483
|
6758 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6759 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
474
|
6760 region, p == NULL ? 0 : *p);
|
|
339
|
6761 if (p == NULL || *p == NUL)
|
|
|
6762 break;
|
|
|
6763 }
|
|
334
|
6764 ++spin->si_foldwcount;
|
|
307
|
6765
|
|
474
|
6766 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
6767 {
|
|
339
|
6768 for (p = pfxlist; res == OK; ++p)
|
|
|
6769 {
|
|
483
|
6770 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6771 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
474
|
6772 region, p == NULL ? 0 : *p);
|
|
339
|
6773 if (p == NULL || *p == NUL)
|
|
|
6774 break;
|
|
|
6775 }
|
|
334
|
6776 ++spin->si_keepwcount;
|
|
|
6777 }
|
|
300
|
6778 return res;
|
|
236
|
6779 }
|
|
|
6780
|
|
|
6781 /*
|
|
300
|
6782 * Add word "word" to a word tree at "root".
|
|
625
|
6783 * When "flags" < 0 we are adding to the prefix tree where "flags" is used for
|
|
366
|
6784 * "rare" and "region" is the condition nr.
|
|
255
|
6785 * Returns FAIL when out of memory.
|
|
236
|
6786 */
|
|
255
|
6787 static int
|
|
481
|
6788 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
6789 spellinfo_T *spin;
|
|
300
|
6790 char_u *word;
|
|
|
6791 wordnode_T *root;
|
|
|
6792 int flags;
|
|
|
6793 int region;
|
|
481
|
6794 int affixID;
|
|
236
|
6795 {
|
|
300
|
6796 wordnode_T *node = root;
|
|
|
6797 wordnode_T *np;
|
|
469
|
6798 wordnode_T *copyp, **copyprev;
|
|
300
|
6799 wordnode_T **prev = NULL;
|
|
|
6800 int i;
|
|
255
|
6801
|
|
300
|
6802 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
6803 for (i = 0; ; ++i)
|
|
255
|
6804 {
|
|
469
|
6805 /* When there is more than one reference to this node we need to make
|
|
|
6806 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
6807 * (we don't optimize for a partly shared list of siblings). */
|
|
|
6808 if (node != NULL && node->wn_refs > 1)
|
|
|
6809 {
|
|
|
6810 --node->wn_refs;
|
|
|
6811 copyprev = prev;
|
|
|
6812 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
6813 {
|
|
|
6814 /* Allocate a new node and copy the info. */
|
|
474
|
6815 np = get_wordnode(spin);
|
|
469
|
6816 if (np == NULL)
|
|
|
6817 return FAIL;
|
|
|
6818 np->wn_child = copyp->wn_child;
|
|
|
6819 if (np->wn_child != NULL)
|
|
|
6820 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
6821 np->wn_byte = copyp->wn_byte;
|
|
|
6822 if (np->wn_byte == NUL)
|
|
|
6823 {
|
|
|
6824 np->wn_flags = copyp->wn_flags;
|
|
|
6825 np->wn_region = copyp->wn_region;
|
|
481
|
6826 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
6827 }
|
|
|
6828
|
|
|
6829 /* Link the new node in the list, there will be one ref. */
|
|
|
6830 np->wn_refs = 1;
|
|
|
6831 *copyprev = np;
|
|
|
6832 copyprev = &np->wn_sibling;
|
|
|
6833
|
|
|
6834 /* Let "node" point to the head of the copied list. */
|
|
|
6835 if (copyp == node)
|
|
|
6836 node = np;
|
|
|
6837 }
|
|
|
6838 }
|
|
|
6839
|
|
300
|
6840 /* Look for the sibling that has the same character. They are sorted
|
|
|
6841 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
6842 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
6843 * done on flags and then on affixID. */
|
|
339
|
6844 while (node != NULL
|
|
|
6845 && (node->wn_byte < word[i]
|
|
|
6846 || (node->wn_byte == NUL
|
|
|
6847 && (flags < 0
|
|
625
|
6848 ? node->wn_affixID < (unsigned)affixID
|
|
|
6849 : (node->wn_flags < (unsigned)(flags & WN_MASK)
|
|
390
|
6850 || (node->wn_flags == (flags & WN_MASK)
|
|
625
|
6851 && (spin->si_sugtree
|
|
|
6852 ? (node->wn_region & 0xffff) < region
|
|
|
6853 : node->wn_affixID
|
|
|
6854 < (unsigned)affixID)))))))
|
|
236
|
6855 {
|
|
300
|
6856 prev = &node->wn_sibling;
|
|
|
6857 node = *prev;
|
|
236
|
6858 }
|
|
339
|
6859 if (node == NULL
|
|
|
6860 || node->wn_byte != word[i]
|
|
|
6861 || (word[i] == NUL
|
|
|
6862 && (flags < 0
|
|
625
|
6863 || spin->si_sugtree
|
|
390
|
6864 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
6865 || node->wn_affixID != affixID)))
|
|
255
|
6866 {
|
|
300
|
6867 /* Allocate a new node. */
|
|
474
|
6868 np = get_wordnode(spin);
|
|
300
|
6869 if (np == NULL)
|
|
|
6870 return FAIL;
|
|
|
6871 np->wn_byte = word[i];
|
|
469
|
6872
|
|
|
6873 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
6874 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
6875 * make ref count of sibling one (matters when inserting in front
|
|
|
6876 * of the list of siblings). */
|
|
|
6877 if (node == NULL)
|
|
|
6878 np->wn_refs = 1;
|
|
|
6879 else
|
|
|
6880 {
|
|
|
6881 np->wn_refs = node->wn_refs;
|
|
|
6882 node->wn_refs = 1;
|
|
|
6883 }
|
|
300
|
6884 *prev = np;
|
|
|
6885 np->wn_sibling = node;
|
|
|
6886 node = np;
|
|
255
|
6887 }
|
|
300
|
6888
|
|
|
6889 if (word[i] == NUL)
|
|
|
6890 {
|
|
|
6891 node->wn_flags = flags;
|
|
|
6892 node->wn_region |= region;
|
|
481
|
6893 node->wn_affixID = affixID;
|
|
300
|
6894 break;
|
|
|
6895 }
|
|
|
6896 prev = &node->wn_child;
|
|
|
6897 node = *prev;
|
|
255
|
6898 }
|
|
469
|
6899 #ifdef SPELL_PRINTTREE
|
|
|
6900 smsg("Added \"%s\"", word);
|
|
|
6901 spell_print_tree(root->wn_sibling);
|
|
|
6902 #endif
|
|
|
6903
|
|
474
|
6904 /* count nr of words added since last message */
|
|
|
6905 ++spin->si_msg_count;
|
|
|
6906
|
|
482
|
6907 if (spin->si_compress_cnt > 1)
|
|
|
6908 {
|
|
|
6909 if (--spin->si_compress_cnt == 1)
|
|
|
6910 /* Did enough words to lower the block count limit. */
|
|
483
|
6911 spin->si_blocks_cnt += compress_inc;
|
|
482
|
6912 }
|
|
|
6913
|
|
469
|
6914 /*
|
|
482
|
6915 * When we have allocated lots of memory we need to compress the word tree
|
|
|
6916 * to free up some room. But compression is slow, and we might actually
|
|
|
6917 * need that room, thus only compress in the following situations:
|
|
|
6918 * 1. When not compressed before (si_compress_cnt == 0): when using
|
|
483
|
6919 * "compress_start" blocks.
|
|
|
6920 * 2. When compressed before and used "compress_inc" blocks before
|
|
|
6921 * adding "compress_added" words (si_compress_cnt > 1).
|
|
|
6922 * 3. When compressed before, added "compress_added" words
|
|
482
|
6923 * (si_compress_cnt == 1) and the number of free nodes drops below the
|
|
|
6924 * maximum word length.
|
|
469
|
6925 */
|
|
482
|
6926 #ifndef SPELL_PRINTTREE
|
|
|
6927 if (spin->si_compress_cnt == 1
|
|
|
6928 ? spin->si_free_count < MAXWLEN
|
|
483
|
6929 : spin->si_blocks_cnt >= compress_start)
|
|
482
|
6930 #endif
|
|
|
6931 {
|
|
|
6932 /* Decrement the block counter. The effect is that we compress again
|
|
483
|
6933 * when the freed up room has been used and another "compress_inc"
|
|
|
6934 * blocks have been allocated. Unless "compress_added" words have
|
|
482
|
6935 * been added, then the limit is put back again. */
|
|
483
|
6936 spin->si_blocks_cnt -= compress_inc;
|
|
|
6937 spin->si_compress_cnt = compress_added;
|
|
482
|
6938
|
|
474
|
6939 if (spin->si_verbose)
|
|
|
6940 {
|
|
|
6941 msg_start();
|
|
|
6942 msg_puts((char_u *)_(msg_compressing));
|
|
|
6943 msg_clr_eos();
|
|
|
6944 msg_didout = FALSE;
|
|
|
6945 msg_col = 0;
|
|
|
6946 out_flush();
|
|
|
6947 }
|
|
482
|
6948
|
|
|
6949 /* Compress both trees. Either they both have many nodes, which makes
|
|
|
6950 * compression useful, or one of them is small, which means
|
|
625
|
6951 * compression goes fast. But when filling the souldfold word tree
|
|
|
6952 * there is no keep-case tree. */
|
|
482
|
6953 wordtree_compress(spin, spin->si_foldroot);
|
|
625
|
6954 if (affixID >= 0)
|
|
|
6955 wordtree_compress(spin, spin->si_keeproot);
|
|
469
|
6956 }
|
|
255
|
6957
|
|
|
6958 return OK;
|
|
236
|
6959 }
|
|
|
6960
|
|
469
|
6961 /*
|
|
483
|
6962 * Check the 'mkspellmem' option. Return FAIL if it's wrong.
|
|
|
6963 * Sets "sps_flags".
|
|
|
6964 */
|
|
|
6965 int
|
|
|
6966 spell_check_msm()
|
|
|
6967 {
|
|
|
6968 char_u *p = p_msm;
|
|
|
6969 long start = 0;
|
|
|
6970 long inc = 0;
|
|
|
6971 long added = 0;
|
|
|
6972
|
|
|
6973 if (!VIM_ISDIGIT(*p))
|
|
|
6974 return FAIL;
|
|
|
6975 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/
|
|
|
6976 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102);
|
|
|
6977 if (*p != ',')
|
|
|
6978 return FAIL;
|
|
|
6979 ++p;
|
|
|
6980 if (!VIM_ISDIGIT(*p))
|
|
|
6981 return FAIL;
|
|
|
6982 inc = (getdigits(&p) * 102) / (SBLOCKSIZE / 10);
|
|
|
6983 if (*p != ',')
|
|
|
6984 return FAIL;
|
|
|
6985 ++p;
|
|
|
6986 if (!VIM_ISDIGIT(*p))
|
|
|
6987 return FAIL;
|
|
|
6988 added = getdigits(&p) * 1024;
|
|
|
6989 if (*p != NUL)
|
|
|
6990 return FAIL;
|
|
|
6991
|
|
|
6992 if (start == 0 || inc == 0 || added == 0 || inc > start)
|
|
|
6993 return FAIL;
|
|
|
6994
|
|
|
6995 compress_start = start;
|
|
|
6996 compress_inc = inc;
|
|
|
6997 compress_added = added;
|
|
|
6998 return OK;
|
|
|
6999 }
|
|
|
7000
|
|
|
7001
|
|
|
7002 /*
|
|
469
|
7003 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
7004 * allocate a new one.
|
|
|
7005 */
|
|
|
7006 static wordnode_T *
|
|
474
|
7007 get_wordnode(spin)
|
|
|
7008 spellinfo_T *spin;
|
|
469
|
7009 {
|
|
|
7010 wordnode_T *n;
|
|
|
7011
|
|
474
|
7012 if (spin->si_first_free == NULL)
|
|
|
7013 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
7014 else
|
|
|
7015 {
|
|
474
|
7016 n = spin->si_first_free;
|
|
|
7017 spin->si_first_free = n->wn_child;
|
|
469
|
7018 vim_memset(n, 0, sizeof(wordnode_T));
|
|
482
|
7019 --spin->si_free_count;
|
|
469
|
7020 }
|
|
|
7021 #ifdef SPELL_PRINTTREE
|
|
474
|
7022 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
7023 #endif
|
|
|
7024 return n;
|
|
|
7025 }
|
|
|
7026
|
|
|
7027 /*
|
|
474
|
7028 * Decrement the reference count on a node (which is the head of a list of
|
|
|
7029 * siblings). If the reference count becomes zero free the node and its
|
|
|
7030 * siblings.
|
|
625
|
7031 * Returns the number of nodes actually freed.
|
|
|
7032 */
|
|
|
7033 static int
|
|
474
|
7034 deref_wordnode(spin, node)
|
|
|
7035 spellinfo_T *spin;
|
|
|
7036 wordnode_T *node;
|
|
|
7037 {
|
|
625
|
7038 wordnode_T *np;
|
|
|
7039 int cnt = 0;
|
|
474
|
7040
|
|
|
7041 if (--node->wn_refs == 0)
|
|
625
|
7042 {
|
|
474
|
7043 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7044 {
|
|
|
7045 if (np->wn_child != NULL)
|
|
625
|
7046 cnt += deref_wordnode(spin, np->wn_child);
|
|
474
|
7047 free_wordnode(spin, np);
|
|
625
|
7048 ++cnt;
|
|
|
7049 }
|
|
|
7050 ++cnt; /* length field */
|
|
|
7051 }
|
|
|
7052 return cnt;
|
|
474
|
7053 }
|
|
|
7054
|
|
|
7055 /*
|
|
469
|
7056 * Free a wordnode_T for re-use later.
|
|
474
|
7057 * Only the "wn_child" field becomes invalid.
|
|
469
|
7058 */
|
|
|
7059 static void
|
|
474
|
7060 free_wordnode(spin, n)
|
|
|
7061 spellinfo_T *spin;
|
|
|
7062 wordnode_T *n;
|
|
|
7063 {
|
|
|
7064 n->wn_child = spin->si_first_free;
|
|
|
7065 spin->si_first_free = n;
|
|
482
|
7066 ++spin->si_free_count;
|
|
469
|
7067 }
|
|
|
7068
|
|
236
|
7069 /*
|
|
300
|
7070 * Compress a tree: find tails that are identical and can be shared.
|
|
|
7071 */
|
|
|
7072 static void
|
|
474
|
7073 wordtree_compress(spin, root)
|
|
|
7074 spellinfo_T *spin;
|
|
300
|
7075 wordnode_T *root;
|
|
|
7076 {
|
|
|
7077 hashtab_T ht;
|
|
|
7078 int n;
|
|
|
7079 int tot = 0;
|
|
469
|
7080 int perc;
|
|
300
|
7081
|
|
474
|
7082 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
7083 * start of the tree. */
|
|
|
7084 if (root->wn_sibling != NULL)
|
|
300
|
7085 {
|
|
|
7086 hash_init(&ht);
|
|
474
|
7087 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
7088
|
|
|
7089 #ifndef SPELL_PRINTTREE
|
|
310
|
7090 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
7091 #endif
|
|
310
|
7092 {
|
|
469
|
7093 if (tot > 1000000)
|
|
|
7094 perc = (tot - n) / (tot / 100);
|
|
482
|
7095 else if (tot == 0)
|
|
|
7096 perc = 0;
|
|
469
|
7097 else
|
|
|
7098 perc = (tot - n) * 100 / tot;
|
|
625
|
7099 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
7100 _("Compressed %d of %d nodes; %d (%d%%) remaining"),
|
|
|
7101 n, tot, tot - n, perc);
|
|
|
7102 spell_message(spin, IObuff);
|
|
310
|
7103 }
|
|
469
|
7104 #ifdef SPELL_PRINTTREE
|
|
474
|
7105 spell_print_tree(root->wn_sibling);
|
|
469
|
7106 #endif
|
|
300
|
7107 hash_clear(&ht);
|
|
|
7108 }
|
|
|
7109 }
|
|
|
7110
|
|
|
7111 /*
|
|
|
7112 * Compress a node, its siblings and its children, depth first.
|
|
|
7113 * Returns the number of compressed nodes.
|
|
236
|
7114 */
|
|
255
|
7115 static int
|
|
474
|
7116 node_compress(spin, node, ht, tot)
|
|
|
7117 spellinfo_T *spin;
|
|
300
|
7118 wordnode_T *node;
|
|
|
7119 hashtab_T *ht;
|
|
|
7120 int *tot; /* total count of nodes before compressing,
|
|
|
7121 incremented while going through the tree */
|
|
236
|
7122 {
|
|
300
|
7123 wordnode_T *np;
|
|
|
7124 wordnode_T *tp;
|
|
|
7125 wordnode_T *child;
|
|
|
7126 hash_T hash;
|
|
236
|
7127 hashitem_T *hi;
|
|
300
|
7128 int len = 0;
|
|
|
7129 unsigned nr, n;
|
|
|
7130 int compressed = 0;
|
|
236
|
7131
|
|
300
|
7132 /*
|
|
|
7133 * Go through the list of siblings. Compress each child and then try
|
|
|
7134 * finding an identical child to replace it.
|
|
|
7135 * Note that with "child" we mean not just the node that is pointed to,
|
|
625
|
7136 * but the whole list of siblings of which the child node is the first.
|
|
300
|
7137 */
|
|
482
|
7138 for (np = node; np != NULL && !got_int; np = np->wn_sibling)
|
|
236
|
7139 {
|
|
300
|
7140 ++len;
|
|
|
7141 if ((child = np->wn_child) != NULL)
|
|
|
7142 {
|
|
625
|
7143 /* Compress the child first. This fills hashkey. */
|
|
474
|
7144 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
7145
|
|
|
7146 /* Try to find an identical child. */
|
|
346
|
7147 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
7148 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
7149 if (!HASHITEM_EMPTY(hi))
|
|
|
7150 {
|
|
625
|
7151 /* There are children we encountered before with a hash value
|
|
|
7152 * identical to the current child. Now check if there is one
|
|
|
7153 * that is really identical. */
|
|
346
|
7154 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
7155 if (node_equal(child, tp))
|
|
|
7156 {
|
|
|
7157 /* Found one! Now use that child in place of the
|
|
469
|
7158 * current one. This means the current child and all
|
|
|
7159 * its siblings is unlinked from the tree. */
|
|
|
7160 ++tp->wn_refs;
|
|
625
|
7161 compressed += deref_wordnode(spin, child);
|
|
300
|
7162 np->wn_child = tp;
|
|
|
7163 break;
|
|
|
7164 }
|
|
|
7165 if (tp == NULL)
|
|
|
7166 {
|
|
|
7167 /* No other child with this hash value equals the child of
|
|
|
7168 * the node, add it to the linked list after the first
|
|
|
7169 * item. */
|
|
|
7170 tp = HI2WN(hi);
|
|
346
|
7171 child->wn_u2.next = tp->wn_u2.next;
|
|
|
7172 tp->wn_u2.next = child;
|
|
300
|
7173 }
|
|
|
7174 }
|
|
|
7175 else
|
|
|
7176 /* No other child has this hash value, add it to the
|
|
|
7177 * hashtable. */
|
|
346
|
7178 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
7179 }
|
|
236
|
7180 }
|
|
625
|
7181 *tot += len + 1; /* add one for the node that stores the length */
|
|
300
|
7182
|
|
|
7183 /*
|
|
|
7184 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
7185 * find a lookalike node. This must be done after compressing the sibling
|
|
|
7186 * list, otherwise the hash key would become invalid by the compression.
|
|
|
7187 */
|
|
346
|
7188 node->wn_u1.hashkey[0] = len;
|
|
300
|
7189 nr = 0;
|
|
|
7190 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
7191 {
|
|
300
|
7192 if (np->wn_byte == NUL)
|
|
481
|
7193 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
7194 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
7195 else
|
|
|
7196 /* byte node: use the byte value and the child pointer */
|
|
|
7197 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
7198 nr = nr * 101 + n;
|
|
236
|
7199 }
|
|
300
|
7200
|
|
|
7201 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
7202 n = nr & 0xff;
|
|
346
|
7203 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
7204 n = (nr >> 8) & 0xff;
|
|
346
|
7205 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
7206 n = (nr >> 16) & 0xff;
|
|
346
|
7207 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
7208 n = (nr >> 24) & 0xff;
|
|
346
|
7209 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
7210 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
7211
|
|
482
|
7212 /* Check for CTRL-C pressed now and then. */
|
|
|
7213 fast_breakcheck();
|
|
|
7214
|
|
300
|
7215 return compressed;
|
|
|
7216 }
|
|
|
7217
|
|
|
7218 /*
|
|
|
7219 * Return TRUE when two nodes have identical siblings and children.
|
|
|
7220 */
|
|
|
7221 static int
|
|
|
7222 node_equal(n1, n2)
|
|
|
7223 wordnode_T *n1;
|
|
|
7224 wordnode_T *n2;
|
|
|
7225 {
|
|
|
7226 wordnode_T *p1;
|
|
|
7227 wordnode_T *p2;
|
|
|
7228
|
|
|
7229 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
7230 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
7231 if (p1->wn_byte != p2->wn_byte
|
|
|
7232 || (p1->wn_byte == NUL
|
|
|
7233 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
7234 || p1->wn_region != p2->wn_region
|
|
481
|
7235 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
7236 : (p1->wn_child != p2->wn_child)))
|
|
|
7237 break;
|
|
|
7238
|
|
|
7239 return p1 == NULL && p2 == NULL;
|
|
236
|
7240 }
|
|
|
7241
|
|
|
7242 /*
|
|
|
7243 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
7244 */
|
|
255
|
7245 void
|
|
236
|
7246 put_bytes(fd, nr, len)
|
|
|
7247 FILE *fd;
|
|
|
7248 long_u nr;
|
|
|
7249 int len;
|
|
|
7250 {
|
|
|
7251 int i;
|
|
|
7252
|
|
|
7253 for (i = len - 1; i >= 0; --i)
|
|
|
7254 putc((int)(nr >> (i * 8)), fd);
|
|
|
7255 }
|
|
|
7256
|
|
625
|
7257 /*
|
|
|
7258 * Write spin->si_sugtime to file "fd".
|
|
|
7259 */
|
|
|
7260 static void
|
|
|
7261 put_sugtime(spin, fd)
|
|
|
7262 spellinfo_T *spin;
|
|
|
7263 FILE *fd;
|
|
|
7264 {
|
|
|
7265 int c;
|
|
|
7266 int i;
|
|
|
7267
|
|
|
7268 /* time_t can be up to 8 bytes in size, more than long_u, thus we
|
|
|
7269 * can't use put_bytes() here. */
|
|
|
7270 for (i = 7; i >= 0; --i)
|
|
|
7271 if (i + 1 > sizeof(time_t))
|
|
|
7272 /* ">>" doesn't work well when shifting more bits than avail */
|
|
|
7273 putc(0, fd);
|
|
|
7274 else
|
|
|
7275 {
|
|
|
7276 c = (unsigned)spin->si_sugtime >> (i * 8);
|
|
|
7277 putc(c, fd);
|
|
|
7278 }
|
|
|
7279 }
|
|
|
7280
|
|
323
|
7281 static int
|
|
|
7282 #ifdef __BORLANDC__
|
|
|
7283 _RTLENTRYF
|
|
|
7284 #endif
|
|
|
7285 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
7286
|
|
|
7287 /*
|
|
|
7288 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
7289 */
|
|
|
7290 static int
|
|
|
7291 #ifdef __BORLANDC__
|
|
|
7292 _RTLENTRYF
|
|
|
7293 #endif
|
|
|
7294 rep_compare(s1, s2)
|
|
|
7295 const void *s1;
|
|
|
7296 const void *s2;
|
|
|
7297 {
|
|
|
7298 fromto_T *p1 = (fromto_T *)s1;
|
|
|
7299 fromto_T *p2 = (fromto_T *)s2;
|
|
|
7300
|
|
|
7301 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
7302 }
|
|
|
7303
|
|
236
|
7304 /*
|
|
483
|
7305 * Write the Vim .spl file "fname".
|
|
500
|
7306 * Return FAIL or OK;
|
|
|
7307 */
|
|
|
7308 static int
|
|
474
|
7309 write_vim_spell(spin, fname)
|
|
|
7310 spellinfo_T *spin;
|
|
236
|
7311 char_u *fname;
|
|
|
7312 {
|
|
300
|
7313 FILE *fd;
|
|
|
7314 int regionmask;
|
|
236
|
7315 int round;
|
|
300
|
7316 wordnode_T *tree;
|
|
|
7317 int nodecount;
|
|
323
|
7318 int i;
|
|
|
7319 int l;
|
|
|
7320 garray_T *gap;
|
|
|
7321 fromto_T *ftp;
|
|
|
7322 char_u *p;
|
|
|
7323 int rr;
|
|
500
|
7324 int retval = OK;
|
|
236
|
7325
|
|
310
|
7326 fd = mch_fopen((char *)fname, "w");
|
|
300
|
7327 if (fd == NULL)
|
|
236
|
7328 {
|
|
|
7329 EMSG2(_(e_notopen), fname);
|
|
500
|
7330 return FAIL;
|
|
236
|
7331 }
|
|
|
7332
|
|
483
|
7333 /* <HEADER>: <fileID> <versionnr> */
|
|
300
|
7334 /* <fileID> */
|
|
|
7335 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
500
|
7336 {
|
|
300
|
7337 EMSG(_(e_write));
|
|
500
|
7338 retval = FAIL;
|
|
|
7339 }
|
|
483
|
7340 putc(VIMSPELLVERSION, fd); /* <versionnr> */
|
|
|
7341
|
|
|
7342 /*
|
|
|
7343 * <SECTIONS>: <section> ... <sectionend>
|
|
|
7344 */
|
|
|
7345
|
|
|
7346 /* SN_REGION: <regionname> ...
|
|
|
7347 * Write the region names only if there is more than one. */
|
|
316
|
7348 if (spin->si_region_count > 1)
|
|
236
|
7349 {
|
|
483
|
7350 putc(SN_REGION, fd); /* <sectionID> */
|
|
|
7351 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7352 l = spin->si_region_count * 2;
|
|
|
7353 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7354 fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd);
|
|
|
7355 /* <regionname> ... */
|
|
316
|
7356 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
7357 }
|
|
|
7358 else
|
|
300
|
7359 regionmask = 0;
|
|
483
|
7360
|
|
|
7361 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars>
|
|
|
7362 *
|
|
|
7363 * The table with character flags and the table for case folding.
|
|
|
7364 * This makes sure the same characters are recognized as word characters
|
|
|
7365 * when generating an when using a spell file.
|
|
260
|
7366 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
7367 * 'encoding'.
|
|
|
7368 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
7369 * the table (avoids that it conflicts). File is shorter too.
|
|
|
7370 */
|
|
483
|
7371 if (!spin->si_ascii && !spin->si_add)
|
|
|
7372 {
|
|
|
7373 char_u folchars[128 * 8];
|
|
|
7374 int flags;
|
|
|
7375
|
|
485
|
7376 putc(SN_CHARFLAGS, fd); /* <sectionID> */
|
|
483
|
7377 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7378
|
|
|
7379 /* Form the <folchars> string first, we need to know its length. */
|
|
|
7380 l = 0;
|
|
|
7381 for (i = 128; i < 256; ++i)
|
|
|
7382 {
|
|
|
7383 #ifdef FEAT_MBYTE
|
|
|
7384 if (has_mbyte)
|
|
|
7385 l += mb_char2bytes(spelltab.st_fold[i], folchars + l);
|
|
|
7386 else
|
|
|
7387 #endif
|
|
|
7388 folchars[l++] = spelltab.st_fold[i];
|
|
|
7389 }
|
|
|
7390 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */
|
|
|
7391
|
|
|
7392 fputc(128, fd); /* <charflagslen> */
|
|
|
7393 for (i = 128; i < 256; ++i)
|
|
|
7394 {
|
|
|
7395 flags = 0;
|
|
|
7396 if (spelltab.st_isw[i])
|
|
|
7397 flags |= CF_WORD;
|
|
|
7398 if (spelltab.st_isu[i])
|
|
|
7399 flags |= CF_UPPER;
|
|
|
7400 fputc(flags, fd); /* <charflags> */
|
|
|
7401 }
|
|
|
7402
|
|
|
7403 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */
|
|
|
7404 fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */
|
|
|
7405 }
|
|
|
7406
|
|
|
7407 /* SN_MIDWORD: <midword> */
|
|
|
7408 if (spin->si_midword != NULL)
|
|
|
7409 {
|
|
|
7410 putc(SN_MIDWORD, fd); /* <sectionID> */
|
|
|
7411 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7412
|
|
366
|
7413 i = STRLEN(spin->si_midword);
|
|
483
|
7414 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */
|
|
366
|
7415 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
7416 }
|
|
|
7417
|
|
483
|
7418 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */
|
|
|
7419 if (spin->si_prefcond.ga_len > 0)
|
|
|
7420 {
|
|
|
7421 putc(SN_PREFCOND, fd); /* <sectionID> */
|
|
|
7422 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7423
|
|
|
7424 l = write_spell_prefcond(NULL, &spin->si_prefcond);
|
|
|
7425 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7426
|
|
|
7427 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
7428 }
|
|
|
7429
|
|
|
7430 /* SN_REP: <repcount> <rep> ...
|
|
625
|
7431 * SN_SAL: <salflags> <salcount> <sal> ...
|
|
|
7432 * SN_REPSAL: <repcount> <rep> ... */
|
|
323
|
7433
|
|
483
|
7434 /* round 1: SN_REP section
|
|
625
|
7435 * round 2: SN_SAL section (unless SN_SOFO is used)
|
|
|
7436 * round 3: SN_REPSAL section */
|
|
|
7437 for (round = 1; round <= 3; ++round)
|
|
323
|
7438 {
|
|
|
7439 if (round == 1)
|
|
|
7440 gap = &spin->si_rep;
|
|
625
|
7441 else if (round == 2)
|
|
|
7442 {
|
|
|
7443 /* Don't write SN_SAL when using a SN_SOFO section */
|
|
|
7444 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
7445 continue;
|
|
|
7446 gap = &spin->si_sal;
|
|
483
|
7447 }
|
|
323
|
7448 else
|
|
625
|
7449 gap = &spin->si_repsal;
|
|
|
7450
|
|
|
7451 /* Don't write the section if there are no items. */
|
|
|
7452 if (gap->ga_len == 0)
|
|
|
7453 continue;
|
|
|
7454
|
|
|
7455 /* Sort the REP/REPSAL items. */
|
|
|
7456 if (round != 2)
|
|
|
7457 qsort(gap->ga_data, (size_t)gap->ga_len,
|
|
|
7458 sizeof(fromto_T), rep_compare);
|
|
|
7459
|
|
|
7460 i = round == 1 ? SN_REP : (round == 2 ? SN_SAL : SN_REPSAL);
|
|
|
7461 putc(i, fd); /* <sectionID> */
|
|
483
|
7462
|
|
|
7463 /* This is for making suggestions, section is not required. */
|
|
|
7464 putc(0, fd); /* <sectionflags> */
|
|
|
7465
|
|
|
7466 /* Compute the length of what follows. */
|
|
|
7467 l = 2; /* count <repcount> or <salcount> */
|
|
|
7468 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7469 {
|
|
|
7470 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
7471 l += 1 + STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */
|
|
|
7472 l += 1 + STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */
|
|
|
7473 }
|
|
|
7474 if (round == 2)
|
|
|
7475 ++l; /* count <salflags> */
|
|
|
7476 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7477
|
|
|
7478 if (round == 2)
|
|
|
7479 {
|
|
323
|
7480 i = 0;
|
|
|
7481 if (spin->si_followup)
|
|
|
7482 i |= SAL_F0LLOWUP;
|
|
|
7483 if (spin->si_collapse)
|
|
|
7484 i |= SAL_COLLAPSE;
|
|
|
7485 if (spin->si_rem_accents)
|
|
|
7486 i |= SAL_REM_ACCENTS;
|
|
|
7487 putc(i, fd); /* <salflags> */
|
|
|
7488 }
|
|
|
7489
|
|
|
7490 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
7491 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7492 {
|
|
|
7493 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
7494 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
7495 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
7496 for (rr = 1; rr <= 2; ++rr)
|
|
|
7497 {
|
|
|
7498 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
7499 l = STRLEN(p);
|
|
|
7500 putc(l, fd);
|
|
|
7501 fwrite(p, l, (size_t)1, fd);
|
|
|
7502 }
|
|
|
7503 }
|
|
483
|
7504
|
|
|
7505 }
|
|
|
7506
|
|
|
7507 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
7508 * This is for making suggestions, section is not required. */
|
|
375
|
7509 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
7510 {
|
|
483
|
7511 putc(SN_SOFO, fd); /* <sectionID> */
|
|
|
7512 putc(0, fd); /* <sectionflags> */
|
|
375
|
7513
|
|
|
7514 l = STRLEN(spin->si_sofofr);
|
|
483
|
7515 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4);
|
|
|
7516 /* <sectionlen> */
|
|
|
7517
|
|
|
7518 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */
|
|
|
7519 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */
|
|
375
|
7520
|
|
|
7521 l = STRLEN(spin->si_sofoto);
|
|
483
|
7522 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */
|
|
|
7523 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */
|
|
|
7524 }
|
|
|
7525
|
|
625
|
7526 /* SN_WORDS: <word> ...
|
|
|
7527 * This is for making suggestions, section is not required. */
|
|
|
7528 if (spin->si_commonwords.ht_used > 0)
|
|
|
7529 {
|
|
|
7530 putc(SN_WORDS, fd); /* <sectionID> */
|
|
|
7531 putc(0, fd); /* <sectionflags> */
|
|
|
7532
|
|
|
7533 /* round 1: count the bytes
|
|
|
7534 * round 2: write the bytes */
|
|
|
7535 for (round = 1; round <= 2; ++round)
|
|
|
7536 {
|
|
|
7537 int todo;
|
|
|
7538 int len = 0;
|
|
|
7539 hashitem_T *hi;
|
|
|
7540
|
|
|
7541 todo = spin->si_commonwords.ht_used;
|
|
|
7542 for (hi = spin->si_commonwords.ht_array; todo > 0; ++hi)
|
|
|
7543 if (!HASHITEM_EMPTY(hi))
|
|
|
7544 {
|
|
|
7545 l = STRLEN(hi->hi_key) + 1;
|
|
|
7546 len += l;
|
|
|
7547 if (round == 2) /* <word> */
|
|
|
7548 fwrite(hi->hi_key, (size_t)l, (size_t)1, fd);
|
|
|
7549 --todo;
|
|
|
7550 }
|
|
|
7551 if (round == 1)
|
|
|
7552 put_bytes(fd, (long_u)len, 4); /* <sectionlen> */
|
|
|
7553 }
|
|
|
7554 }
|
|
|
7555
|
|
483
|
7556 /* SN_MAP: <mapstr>
|
|
|
7557 * This is for making suggestions, section is not required. */
|
|
|
7558 if (spin->si_map.ga_len > 0)
|
|
|
7559 {
|
|
|
7560 putc(SN_MAP, fd); /* <sectionID> */
|
|
|
7561 putc(0, fd); /* <sectionflags> */
|
|
|
7562 l = spin->si_map.ga_len;
|
|
|
7563 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7564 fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd);
|
|
|
7565 /* <mapstr> */
|
|
|
7566 }
|
|
|
7567
|
|
625
|
7568 /* SN_SUGFILE: <timestamp>
|
|
|
7569 * This is used to notify that a .sug file may be available and at the
|
|
|
7570 * same time allows for checking that a .sug file that is found matches
|
|
|
7571 * with this .spl file. That's because the word numbers must be exactly
|
|
|
7572 * right. */
|
|
|
7573 if (!spin->si_nosugfile
|
|
|
7574 && (spin->si_sal.ga_len > 0
|
|
|
7575 || (spin->si_sofofr != NULL && spin->si_sofoto != NULL)))
|
|
|
7576 {
|
|
|
7577 putc(SN_SUGFILE, fd); /* <sectionID> */
|
|
|
7578 putc(0, fd); /* <sectionflags> */
|
|
|
7579 put_bytes(fd, (long_u)8, 4); /* <sectionlen> */
|
|
|
7580
|
|
|
7581 /* Set si_sugtime and write it to the file. */
|
|
|
7582 spin->si_sugtime = time(NULL);
|
|
|
7583 put_sugtime(spin, fd); /* <timestamp> */
|
|
|
7584 }
|
|
|
7585
|
|
483
|
7586 /* SN_COMPOUND: compound info.
|
|
|
7587 * We don't mark it required, when not supported all compound words will
|
|
|
7588 * be bad words. */
|
|
|
7589 if (spin->si_compflags != NULL)
|
|
|
7590 {
|
|
|
7591 putc(SN_COMPOUND, fd); /* <sectionID> */
|
|
|
7592 putc(0, fd); /* <sectionflags> */
|
|
|
7593
|
|
|
7594 l = STRLEN(spin->si_compflags);
|
|
|
7595 put_bytes(fd, (long_u)(l + 3), 4); /* <sectionlen> */
|
|
|
7596 putc(spin->si_compmax, fd); /* <compmax> */
|
|
|
7597 putc(spin->si_compminlen, fd); /* <compminlen> */
|
|
|
7598 putc(spin->si_compsylmax, fd); /* <compsylmax> */
|
|
|
7599 /* <compflags> */
|
|
|
7600 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
7601 }
|
|
|
7602
|
|
492
|
7603 /* SN_NOBREAK: NOBREAK flag */
|
|
|
7604 if (spin->si_nobreak)
|
|
|
7605 {
|
|
|
7606 putc(SN_NOBREAK, fd); /* <sectionID> */
|
|
|
7607 putc(0, fd); /* <sectionflags> */
|
|
|
7608
|
|
|
7609 /* It's empty, the precense of the section flags the feature. */
|
|
|
7610 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */
|
|
|
7611 }
|
|
|
7612
|
|
483
|
7613 /* SN_SYLLABLE: syllable info.
|
|
|
7614 * We don't mark it required, when not supported syllables will not be
|
|
|
7615 * counted. */
|
|
|
7616 if (spin->si_syllable != NULL)
|
|
|
7617 {
|
|
|
7618 putc(SN_SYLLABLE, fd); /* <sectionID> */
|
|
|
7619 putc(0, fd); /* <sectionflags> */
|
|
|
7620
|
|
|
7621 l = STRLEN(spin->si_syllable);
|
|
|
7622 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7623 fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); /* <syllable> */
|
|
|
7624 }
|
|
|
7625
|
|
|
7626 /* end of <SECTIONS> */
|
|
|
7627 putc(SN_END, fd); /* <sectionend> */
|
|
|
7628
|
|
302
|
7629
|
|
236
|
7630 /*
|
|
339
|
7631 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
7632 */
|
|
323
|
7633 spin->si_memtot = 0;
|
|
339
|
7634 for (round = 1; round <= 3; ++round)
|
|
236
|
7635 {
|
|
339
|
7636 if (round == 1)
|
|
469
|
7637 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
7638 else if (round == 2)
|
|
469
|
7639 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
7640 else
|
|
469
|
7641 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
7642
|
|
346
|
7643 /* Clear the index and wnode fields in the tree. */
|
|
|
7644 clear_node(tree);
|
|
|
7645
|
|
300
|
7646 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
7647 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
7648 * nodes. */
|
|
346
|
7649 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
7650
|
|
300
|
7651 /* number of nodes in 4 bytes */
|
|
|
7652 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
7653 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
7654
|
|
300
|
7655 /* Write the nodes. */
|
|
346
|
7656 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
7657 }
|
|
|
7658
|
|
500
|
7659 /* Write another byte to check for errors. */
|
|
|
7660 if (putc(0, fd) == EOF)
|
|
|
7661 retval = FAIL;
|
|
|
7662
|
|
|
7663 if (fclose(fd) == EOF)
|
|
|
7664 retval = FAIL;
|
|
|
7665
|
|
|
7666 return retval;
|
|
236
|
7667 }
|
|
|
7668
|
|
|
7669 /*
|
|
346
|
7670 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
7671 * children. This is needed because they are a union with other items to save
|
|
|
7672 * space.
|
|
|
7673 */
|
|
|
7674 static void
|
|
|
7675 clear_node(node)
|
|
|
7676 wordnode_T *node;
|
|
|
7677 {
|
|
|
7678 wordnode_T *np;
|
|
|
7679
|
|
|
7680 if (node != NULL)
|
|
|
7681 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7682 {
|
|
|
7683 np->wn_u1.index = 0;
|
|
|
7684 np->wn_u2.wnode = NULL;
|
|
|
7685
|
|
|
7686 if (np->wn_byte != NUL)
|
|
|
7687 clear_node(np->wn_child);
|
|
|
7688 }
|
|
|
7689 }
|
|
|
7690
|
|
|
7691
|
|
|
7692 /*
|
|
300
|
7693 * Dump a word tree at node "node".
|
|
|
7694 *
|
|
|
7695 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
7696 * byte recursively write the children.
|
|
|
7697 *
|
|
625
|
7698 * NOTE: The code here must match the code in read_tree_node(), since
|
|
|
7699 * assumptions are made about the indexes (so that we don't have to write them
|
|
|
7700 * in the file).
|
|
236
|
7701 *
|
|
300
|
7702 * Returns the number of nodes used.
|
|
236
|
7703 */
|
|
300
|
7704 static int
|
|
346
|
7705 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
7706 FILE *fd; /* NULL when only counting */
|
|
300
|
7707 wordnode_T *node;
|
|
|
7708 int index;
|
|
|
7709 int regionmask;
|
|
339
|
7710 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
7711 {
|
|
300
|
7712 int newindex = index;
|
|
|
7713 int siblingcount = 0;
|
|
|
7714 wordnode_T *np;
|
|
236
|
7715 int flags;
|
|
300
|
7716
|
|
|
7717 /* If "node" is zero the tree is empty. */
|
|
|
7718 if (node == NULL)
|
|
|
7719 return 0;
|
|
|
7720
|
|
|
7721 /* Store the index where this node is written. */
|
|
346
|
7722 node->wn_u1.index = index;
|
|
236
|
7723
|
|
300
|
7724 /* Count the number of siblings. */
|
|
|
7725 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7726 ++siblingcount;
|
|
236
|
7727
|
|
300
|
7728 /* Write the sibling count. */
|
|
|
7729 if (fd != NULL)
|
|
|
7730 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
7731
|
|
300
|
7732 /* Write each sibling byte and optionally extra info. */
|
|
|
7733 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
7734 {
|
|
300
|
7735 if (np->wn_byte == 0)
|
|
|
7736 {
|
|
|
7737 if (fd != NULL)
|
|
|
7738 {
|
|
339
|
7739 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
7740 if (prefixtree)
|
|
300
|
7741 {
|
|
481
|
7742 /* In PREFIXTREE write the required affixID and the
|
|
390
|
7743 * associated condition nr (stored in wn_region). The
|
|
|
7744 * byte value is misused to store the "rare" and "not
|
|
|
7745 * combining" flags */
|
|
455
|
7746 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
7747 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7748 else
|
|
|
7749 {
|
|
390
|
7750 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
7751 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
7752 }
|
|
481
|
7753 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7754 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
7755 }
|
|
|
7756 else
|
|
|
7757 {
|
|
339
|
7758 /* For word trees we write the flag/region items. */
|
|
|
7759 flags = np->wn_flags;
|
|
|
7760 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
7761 flags |= WF_REGION;
|
|
481
|
7762 if (np->wn_affixID != 0)
|
|
|
7763 flags |= WF_AFX;
|
|
339
|
7764 if (flags == 0)
|
|
|
7765 {
|
|
|
7766 /* word without flags or region */
|
|
|
7767 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7768 }
|
|
|
7769 else
|
|
|
7770 {
|
|
390
|
7771 if (np->wn_flags >= 0x100)
|
|
|
7772 {
|
|
|
7773 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
7774 putc(flags, fd); /* <flags> */
|
|
|
7775 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
7776 }
|
|
|
7777 else
|
|
|
7778 {
|
|
|
7779 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
7780 putc(flags, fd); /* <flags> */
|
|
|
7781 }
|
|
339
|
7782 if (flags & WF_REGION)
|
|
|
7783 putc(np->wn_region, fd); /* <region> */
|
|
481
|
7784 if (flags & WF_AFX)
|
|
|
7785 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7786 }
|
|
300
|
7787 }
|
|
|
7788 }
|
|
|
7789 }
|
|
|
7790 else
|
|
|
7791 {
|
|
346
|
7792 if (np->wn_child->wn_u1.index != 0
|
|
|
7793 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
7794 {
|
|
|
7795 /* The child is written elsewhere, write the reference. */
|
|
|
7796 if (fd != NULL)
|
|
|
7797 {
|
|
|
7798 putc(BY_INDEX, fd); /* <byte> */
|
|
|
7799 /* <nodeidx> */
|
|
346
|
7800 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
7801 }
|
|
|
7802 }
|
|
346
|
7803 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
7804 /* We will write the child below and give it an index. */
|
|
346
|
7805 np->wn_child->wn_u2.wnode = node;
|
|
236
|
7806
|
|
300
|
7807 if (fd != NULL)
|
|
|
7808 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
7809 {
|
|
|
7810 EMSG(_(e_write));
|
|
|
7811 return 0;
|
|
|
7812 }
|
|
|
7813 }
|
|
236
|
7814 }
|
|
|
7815
|
|
300
|
7816 /* Space used in the array when reading: one for each sibling and one for
|
|
|
7817 * the count. */
|
|
|
7818 newindex += siblingcount + 1;
|
|
249
|
7819
|
|
300
|
7820 /* Recursively dump the children of each sibling. */
|
|
|
7821 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
7822 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
7823 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
7824 prefixtree);
|
|
249
|
7825
|
|
300
|
7826 return newindex;
|
|
236
|
7827 }
|
|
|
7828
|
|
|
7829
|
|
|
7830 /*
|
|
310
|
7831 * ":mkspell [-ascii] outfile infile ..."
|
|
|
7832 * ":mkspell [-ascii] addfile"
|
|
236
|
7833 */
|
|
|
7834 void
|
|
|
7835 ex_mkspell(eap)
|
|
|
7836 exarg_T *eap;
|
|
|
7837 {
|
|
|
7838 int fcount;
|
|
|
7839 char_u **fnames;
|
|
310
|
7840 char_u *arg = eap->arg;
|
|
|
7841 int ascii = FALSE;
|
|
|
7842
|
|
|
7843 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
7844 {
|
|
|
7845 ascii = TRUE;
|
|
|
7846 arg = skipwhite(arg + 6);
|
|
|
7847 }
|
|
|
7848
|
|
|
7849 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
7850 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
7851 {
|
|
323
|
7852 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
7853 FreeWild(fcount, fnames);
|
|
|
7854 }
|
|
|
7855 }
|
|
|
7856
|
|
|
7857 /*
|
|
625
|
7858 * Create the .sug file.
|
|
|
7859 * Uses the soundfold info in "spin".
|
|
|
7860 * Writes the file with the name "wfname", with ".spl" changed to ".sug".
|
|
|
7861 */
|
|
|
7862 static void
|
|
|
7863 spell_make_sugfile(spin, wfname)
|
|
|
7864 spellinfo_T *spin;
|
|
|
7865 char_u *wfname;
|
|
|
7866 {
|
|
|
7867 char_u fname[MAXPATHL];
|
|
|
7868 int len;
|
|
|
7869 slang_T *slang;
|
|
|
7870 int free_slang = FALSE;
|
|
|
7871
|
|
|
7872 /*
|
|
|
7873 * Read back the .spl file that was written. This fills the required
|
|
|
7874 * info for soundfolding. This also uses less memory than the
|
|
|
7875 * pointer-linked version of the trie. And it avoids having two versions
|
|
|
7876 * of the code for the soundfolding stuff.
|
|
|
7877 * It might have been done already by spell_reload_one().
|
|
|
7878 */
|
|
|
7879 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
7880 if (fullpathcmp(wfname, slang->sl_fname, FALSE) == FPC_SAME)
|
|
|
7881 break;
|
|
|
7882 if (slang == NULL)
|
|
|
7883 {
|
|
|
7884 spell_message(spin, (char_u *)_("Reading back spell file..."));
|
|
|
7885 slang = spell_load_file(wfname, NULL, NULL, FALSE);
|
|
|
7886 if (slang == NULL)
|
|
|
7887 return;
|
|
|
7888 free_slang = TRUE;
|
|
|
7889 }
|
|
|
7890
|
|
|
7891 /*
|
|
|
7892 * Clear the info in "spin" that is used.
|
|
|
7893 */
|
|
|
7894 spin->si_blocks = NULL;
|
|
|
7895 spin->si_blocks_cnt = 0;
|
|
|
7896 spin->si_compress_cnt = 0; /* will stay at 0 all the time*/
|
|
|
7897 spin->si_free_count = 0;
|
|
|
7898 spin->si_first_free = NULL;
|
|
|
7899 spin->si_foldwcount = 0;
|
|
|
7900
|
|
|
7901 /*
|
|
|
7902 * Go through the trie of good words, soundfold each word and add it to
|
|
|
7903 * the soundfold trie.
|
|
|
7904 */
|
|
|
7905 spell_message(spin, (char_u *)_("Performing soundfolding..."));
|
|
|
7906 if (sug_filltree(spin, slang) == FAIL)
|
|
|
7907 goto theend;
|
|
|
7908
|
|
|
7909 /*
|
|
|
7910 * Create the table which links each soundfold word with a list of the
|
|
|
7911 * good words it may come from. Creates buffer "spin->si_spellbuf".
|
|
|
7912 * This also removes the wordnr from the NUL byte entries to make
|
|
|
7913 * compression possible.
|
|
|
7914 */
|
|
|
7915 if (sug_maketable(spin) == FAIL)
|
|
|
7916 goto theend;
|
|
|
7917
|
|
|
7918 smsg((char_u *)_("Number of words after soundfolding: %ld"),
|
|
|
7919 (long)spin->si_spellbuf->b_ml.ml_line_count);
|
|
|
7920
|
|
|
7921 /*
|
|
|
7922 * Compress the soundfold trie.
|
|
|
7923 */
|
|
|
7924 spell_message(spin, (char_u *)_(msg_compressing));
|
|
|
7925 wordtree_compress(spin, spin->si_foldroot);
|
|
|
7926
|
|
|
7927 /*
|
|
|
7928 * Write the .sug file.
|
|
|
7929 * Make the file name by changing ".spl" to ".sug".
|
|
|
7930 */
|
|
|
7931 STRCPY(fname, wfname);
|
|
|
7932 len = STRLEN(fname);
|
|
|
7933 fname[len - 2] = 'u';
|
|
|
7934 fname[len - 1] = 'g';
|
|
|
7935 sug_write(spin, fname);
|
|
|
7936
|
|
|
7937 theend:
|
|
|
7938 if (free_slang)
|
|
|
7939 slang_free(slang);
|
|
|
7940 free_blocks(spin->si_blocks);
|
|
|
7941 close_spellbuf(spin->si_spellbuf);
|
|
|
7942 }
|
|
|
7943
|
|
|
7944 /*
|
|
|
7945 * Build the soundfold trie for language "slang".
|
|
|
7946 */
|
|
|
7947 static int
|
|
|
7948 sug_filltree(spin, slang)
|
|
|
7949 spellinfo_T *spin;
|
|
|
7950 slang_T *slang;
|
|
|
7951 {
|
|
|
7952 char_u *byts;
|
|
|
7953 idx_T *idxs;
|
|
|
7954 int depth;
|
|
|
7955 idx_T arridx[MAXWLEN];
|
|
|
7956 int curi[MAXWLEN];
|
|
|
7957 char_u tword[MAXWLEN];
|
|
|
7958 char_u tsalword[MAXWLEN];
|
|
|
7959 int c;
|
|
|
7960 idx_T n;
|
|
|
7961 unsigned words_done = 0;
|
|
|
7962 int wordcount[MAXWLEN];
|
|
|
7963
|
|
|
7964 /* We use si_foldroot for the souldfolded trie. */
|
|
|
7965 spin->si_foldroot = wordtree_alloc(spin);
|
|
|
7966 if (spin->si_foldroot == NULL)
|
|
|
7967 return FAIL;
|
|
|
7968
|
|
|
7969 /* let tree_add_word() know we're adding to the soundfolded tree */
|
|
|
7970 spin->si_sugtree = TRUE;
|
|
|
7971
|
|
|
7972 /*
|
|
|
7973 * Go through the whole case-folded tree, soundfold each word and put it
|
|
|
7974 * in the trie.
|
|
|
7975 */
|
|
|
7976 byts = slang->sl_fbyts;
|
|
|
7977 idxs = slang->sl_fidxs;
|
|
|
7978
|
|
|
7979 arridx[0] = 0;
|
|
|
7980 curi[0] = 1;
|
|
|
7981 wordcount[0] = 0;
|
|
|
7982
|
|
|
7983 depth = 0;
|
|
|
7984 while (depth >= 0 && !got_int)
|
|
|
7985 {
|
|
|
7986 if (curi[depth] > byts[arridx[depth]])
|
|
|
7987 {
|
|
|
7988 /* Done all bytes at this node, go up one level. */
|
|
|
7989 idxs[arridx[depth]] = wordcount[depth];
|
|
|
7990 if (depth > 0)
|
|
|
7991 wordcount[depth - 1] += wordcount[depth];
|
|
|
7992
|
|
|
7993 --depth;
|
|
|
7994 line_breakcheck();
|
|
|
7995 }
|
|
|
7996 else
|
|
|
7997 {
|
|
|
7998
|
|
|
7999 /* Do one more byte at this node. */
|
|
|
8000 n = arridx[depth] + curi[depth];
|
|
|
8001 ++curi[depth];
|
|
|
8002
|
|
|
8003 c = byts[n];
|
|
|
8004 if (c == 0)
|
|
|
8005 {
|
|
|
8006 /* Sound-fold the word. */
|
|
|
8007 tword[depth] = NUL;
|
|
|
8008 spell_soundfold(slang, tword, TRUE, tsalword);
|
|
|
8009
|
|
|
8010 /* We use the "flags" field for the MSB of the wordnr,
|
|
|
8011 * "region" for the LSB of the wordnr. */
|
|
|
8012 if (tree_add_word(spin, tsalword, spin->si_foldroot,
|
|
|
8013 words_done >> 16, words_done & 0xffff,
|
|
|
8014 0) == FAIL)
|
|
|
8015 return FAIL;
|
|
|
8016
|
|
|
8017 ++words_done;
|
|
|
8018 ++wordcount[depth];
|
|
|
8019
|
|
|
8020 /* Reset the block count each time to avoid compression
|
|
|
8021 * kicking in. */
|
|
|
8022 spin->si_blocks_cnt = 0;
|
|
|
8023
|
|
|
8024 /* Skip over any other NUL bytes (same word with different
|
|
|
8025 * flags). */
|
|
|
8026 while (byts[n + 1] == 0)
|
|
|
8027 {
|
|
|
8028 ++n;
|
|
|
8029 ++curi[depth];
|
|
|
8030 }
|
|
|
8031 }
|
|
|
8032 else
|
|
|
8033 {
|
|
|
8034 /* Normal char, go one level deeper. */
|
|
|
8035 tword[depth++] = c;
|
|
|
8036 arridx[depth] = idxs[n];
|
|
|
8037 curi[depth] = 1;
|
|
|
8038 wordcount[depth] = 0;
|
|
|
8039 }
|
|
|
8040 }
|
|
|
8041 }
|
|
|
8042
|
|
|
8043 smsg((char_u *)_("Total number of words: %d"), words_done);
|
|
|
8044
|
|
|
8045 return OK;
|
|
|
8046 }
|
|
|
8047
|
|
|
8048 /*
|
|
|
8049 * Make the table that links each word in the soundfold trie to the words it
|
|
|
8050 * can be produced from.
|
|
|
8051 * This is not unlike lines in a file, thus use a memfile to be able to access
|
|
|
8052 * the table efficiently.
|
|
|
8053 * Returns FAIL when out of memory.
|
|
|
8054 */
|
|
|
8055 static int
|
|
|
8056 sug_maketable(spin)
|
|
|
8057 spellinfo_T *spin;
|
|
|
8058 {
|
|
|
8059 garray_T ga;
|
|
|
8060 int res = OK;
|
|
|
8061
|
|
|
8062 /* Allocate a buffer, open a memline for it and create the swap file
|
|
|
8063 * (uses a temp file, not a .swp file). */
|
|
|
8064 spin->si_spellbuf = open_spellbuf();
|
|
|
8065 if (spin->si_spellbuf == NULL)
|
|
|
8066 return FAIL;
|
|
|
8067
|
|
|
8068 /* Use a buffer to store the line info, avoids allocating many small
|
|
|
8069 * pieces of memory. */
|
|
|
8070 ga_init2(&ga, 1, 100);
|
|
|
8071
|
|
|
8072 /* recursively go through the tree */
|
|
|
8073 if (sug_filltable(spin, spin->si_foldroot->wn_sibling, 0, &ga) == -1)
|
|
|
8074 res = FAIL;
|
|
|
8075
|
|
|
8076 ga_clear(&ga);
|
|
|
8077 return res;
|
|
|
8078 }
|
|
|
8079
|
|
|
8080 /*
|
|
|
8081 * Fill the table for one node and its children.
|
|
|
8082 * Returns the wordnr at the start of the node.
|
|
|
8083 * Returns -1 when out of memory.
|
|
|
8084 */
|
|
|
8085 static int
|
|
|
8086 sug_filltable(spin, node, startwordnr, gap)
|
|
|
8087 spellinfo_T *spin;
|
|
|
8088 wordnode_T *node;
|
|
|
8089 int startwordnr;
|
|
|
8090 garray_T *gap; /* place to store line of numbers */
|
|
|
8091 {
|
|
|
8092 wordnode_T *p, *np;
|
|
|
8093 int wordnr = startwordnr;
|
|
|
8094 int nr;
|
|
|
8095 int prev_nr;
|
|
|
8096
|
|
|
8097 for (p = node; p != NULL; p = p->wn_sibling)
|
|
|
8098 {
|
|
|
8099 if (p->wn_byte == NUL)
|
|
|
8100 {
|
|
|
8101 gap->ga_len = 0;
|
|
|
8102 prev_nr = 0;
|
|
|
8103 for (np = p; np != NULL && np->wn_byte == NUL; np = np->wn_sibling)
|
|
|
8104 {
|
|
|
8105 if (ga_grow(gap, 10) == FAIL)
|
|
|
8106 return -1;
|
|
|
8107
|
|
|
8108 nr = (np->wn_flags << 16) + (np->wn_region & 0xffff);
|
|
|
8109 /* Compute the offset from the previous nr and store the
|
|
|
8110 * offset in a way that it takes a minimum number of bytes.
|
|
|
8111 * It's a bit like utf-8, but without the need to mark
|
|
|
8112 * following bytes. */
|
|
|
8113 nr -= prev_nr;
|
|
|
8114 prev_nr += nr;
|
|
|
8115 gap->ga_len += offset2bytes(nr,
|
|
|
8116 (char_u *)gap->ga_data + gap->ga_len);
|
|
|
8117 }
|
|
|
8118
|
|
|
8119 /* add the NUL byte */
|
|
|
8120 ((char_u *)gap->ga_data)[gap->ga_len++] = NUL;
|
|
|
8121
|
|
|
8122 if (ml_append_buf(spin->si_spellbuf, (linenr_T)wordnr,
|
|
|
8123 gap->ga_data, gap->ga_len, TRUE) == FAIL)
|
|
|
8124 return -1;
|
|
|
8125 ++wordnr;
|
|
|
8126
|
|
|
8127 /* Remove extra NUL entries, we no longer need them. We don't
|
|
|
8128 * bother freeing the nodes, the won't be reused anyway. */
|
|
|
8129 while (p->wn_sibling != NULL && p->wn_sibling->wn_byte == NUL)
|
|
|
8130 p->wn_sibling = p->wn_sibling->wn_sibling;
|
|
|
8131
|
|
|
8132 /* Clear the flags on the remaining NUL node, so that compression
|
|
|
8133 * works a lot better. */
|
|
|
8134 p->wn_flags = 0;
|
|
|
8135 p->wn_region = 0;
|
|
|
8136 }
|
|
|
8137 else
|
|
|
8138 {
|
|
|
8139 wordnr = sug_filltable(spin, p->wn_child, wordnr, gap);
|
|
|
8140 if (wordnr == -1)
|
|
|
8141 return -1;
|
|
|
8142 }
|
|
|
8143 }
|
|
|
8144 return wordnr;
|
|
|
8145 }
|
|
|
8146
|
|
|
8147 /*
|
|
|
8148 * Convert an offset into a minimal number of bytes.
|
|
|
8149 * Similar to utf_char2byters, but use 8 bits in followup bytes and avoid NUL
|
|
|
8150 * bytes.
|
|
|
8151 */
|
|
|
8152 static int
|
|
|
8153 offset2bytes(nr, buf)
|
|
|
8154 int nr;
|
|
|
8155 char_u *buf;
|
|
|
8156 {
|
|
|
8157 int rem;
|
|
|
8158 int b1, b2, b3, b4;
|
|
|
8159
|
|
|
8160 /* Split the number in parts of base 255. We need to avoid NUL bytes. */
|
|
|
8161 b1 = nr % 255 + 1;
|
|
|
8162 rem = nr / 255;
|
|
|
8163 b2 = rem % 255 + 1;
|
|
|
8164 rem = rem / 255;
|
|
|
8165 b3 = rem % 255 + 1;
|
|
|
8166 b4 = rem / 255 + 1;
|
|
|
8167
|
|
|
8168 if (b4 > 1 || b3 > 0x1f) /* 4 bytes */
|
|
|
8169 {
|
|
|
8170 buf[0] = 0xe0 + b4;
|
|
|
8171 buf[1] = b3;
|
|
|
8172 buf[2] = b2;
|
|
|
8173 buf[3] = b1;
|
|
|
8174 return 4;
|
|
|
8175 }
|
|
|
8176 if (b3 > 1 || b2 > 0x3f ) /* 3 bytes */
|
|
|
8177 {
|
|
|
8178 buf[0] = 0xc0 + b3;
|
|
|
8179 buf[1] = b2;
|
|
|
8180 buf[2] = b1;
|
|
|
8181 return 3;
|
|
|
8182 }
|
|
|
8183 if (b2 > 1 || b1 > 0x7f ) /* 2 bytes */
|
|
|
8184 {
|
|
|
8185 buf[0] = 0x80 + b2;
|
|
|
8186 buf[1] = b1;
|
|
|
8187 return 2;
|
|
|
8188 }
|
|
|
8189 /* 1 byte */
|
|
|
8190 buf[0] = b1;
|
|
|
8191 return 1;
|
|
|
8192 }
|
|
|
8193
|
|
|
8194 /*
|
|
|
8195 * Opposite of offset2bytes().
|
|
|
8196 * "pp" points to the bytes and is advanced over it.
|
|
|
8197 * Returns the offset.
|
|
|
8198 */
|
|
|
8199 static int
|
|
|
8200 bytes2offset(pp)
|
|
|
8201 char_u **pp;
|
|
|
8202 {
|
|
|
8203 char_u *p = *pp;
|
|
|
8204 int nr;
|
|
|
8205 int c;
|
|
|
8206
|
|
|
8207 c = *p++;
|
|
|
8208 if ((c & 0x80) == 0x00) /* 1 byte */
|
|
|
8209 {
|
|
|
8210 nr = c - 1;
|
|
|
8211 }
|
|
|
8212 else if ((c & 0xc0) == 0x80) /* 2 bytes */
|
|
|
8213 {
|
|
|
8214 nr = (c & 0x3f) - 1;
|
|
|
8215 nr = nr * 255 + (*p++ - 1);
|
|
|
8216 }
|
|
|
8217 else if ((c & 0xe0) == 0xc0) /* 3 bytes */
|
|
|
8218 {
|
|
|
8219 nr = (c & 0x1f) - 1;
|
|
|
8220 nr = nr * 255 + (*p++ - 1);
|
|
|
8221 nr = nr * 255 + (*p++ - 1);
|
|
|
8222 }
|
|
|
8223 else /* 4 bytes */
|
|
|
8224 {
|
|
|
8225 nr = (c & 0x0f) - 1;
|
|
|
8226 nr = nr * 255 + (*p++ - 1);
|
|
|
8227 nr = nr * 255 + (*p++ - 1);
|
|
|
8228 nr = nr * 255 + (*p++ - 1);
|
|
|
8229 }
|
|
|
8230
|
|
|
8231 *pp = p;
|
|
|
8232 return nr;
|
|
|
8233 }
|
|
|
8234
|
|
|
8235 /*
|
|
|
8236 * Write the .sug file in "fname".
|
|
|
8237 */
|
|
|
8238 static void
|
|
|
8239 sug_write(spin, fname)
|
|
|
8240 spellinfo_T *spin;
|
|
|
8241 char_u *fname;
|
|
|
8242 {
|
|
|
8243 FILE *fd;
|
|
|
8244 wordnode_T *tree;
|
|
|
8245 int nodecount;
|
|
|
8246 int wcount;
|
|
|
8247 char_u *line;
|
|
|
8248 linenr_T lnum;
|
|
|
8249 int len;
|
|
|
8250
|
|
|
8251 /* Create the file. Note that an existing file is silently overwritten! */
|
|
|
8252 fd = mch_fopen((char *)fname, "w");
|
|
|
8253 if (fd == NULL)
|
|
|
8254 {
|
|
|
8255 EMSG2(_(e_notopen), fname);
|
|
|
8256 return;
|
|
|
8257 }
|
|
|
8258
|
|
|
8259 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
8260 _("Writing suggestion file %s ..."), fname);
|
|
|
8261 spell_message(spin, IObuff);
|
|
|
8262
|
|
|
8263 /*
|
|
|
8264 * <SUGHEADER>: <fileID> <versionnr> <timestamp>
|
|
|
8265 */
|
|
|
8266 if (fwrite(VIMSUGMAGIC, VIMSUGMAGICL, (size_t)1, fd) != 1) /* <fileID> */
|
|
|
8267 {
|
|
|
8268 EMSG(_(e_write));
|
|
|
8269 goto theend;
|
|
|
8270 }
|
|
|
8271 putc(VIMSUGVERSION, fd); /* <versionnr> */
|
|
|
8272
|
|
|
8273 /* Write si_sugtime to the file. */
|
|
|
8274 put_sugtime(spin, fd); /* <timestamp> */
|
|
|
8275
|
|
|
8276 /*
|
|
|
8277 * <SUGWORDTREE>
|
|
|
8278 */
|
|
|
8279 spin->si_memtot = 0;
|
|
|
8280 tree = spin->si_foldroot->wn_sibling;
|
|
|
8281
|
|
|
8282 /* Clear the index and wnode fields in the tree. */
|
|
|
8283 clear_node(tree);
|
|
|
8284
|
|
|
8285 /* Count the number of nodes. Needed to be able to allocate the
|
|
|
8286 * memory when reading the nodes. Also fills in index for shared
|
|
|
8287 * nodes. */
|
|
|
8288 nodecount = put_node(NULL, tree, 0, 0, FALSE);
|
|
|
8289
|
|
|
8290 /* number of nodes in 4 bytes */
|
|
|
8291 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
|
8292 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
|
8293
|
|
|
8294 /* Write the nodes. */
|
|
|
8295 (void)put_node(fd, tree, 0, 0, FALSE);
|
|
|
8296
|
|
|
8297 /*
|
|
|
8298 * <SUGTABLE>: <sugwcount> <sugline> ...
|
|
|
8299 */
|
|
|
8300 wcount = spin->si_spellbuf->b_ml.ml_line_count;
|
|
|
8301 put_bytes(fd, (long_u)wcount, 4); /* <sugwcount> */
|
|
|
8302
|
|
|
8303 for (lnum = 1; lnum <= (linenr_T)wcount; ++lnum)
|
|
|
8304 {
|
|
|
8305 /* <sugline>: <sugnr> ... NUL */
|
|
|
8306 line = ml_get_buf(spin->si_spellbuf, lnum, FALSE);
|
|
|
8307 len = STRLEN(line) + 1;
|
|
|
8308 if (fwrite(line, (size_t)len, (size_t)1, fd) == 0)
|
|
|
8309 {
|
|
|
8310 EMSG(_(e_write));
|
|
|
8311 goto theend;
|
|
|
8312 }
|
|
|
8313 spin->si_memtot += len;
|
|
|
8314 }
|
|
|
8315
|
|
|
8316 /* Write another byte to check for errors. */
|
|
|
8317 if (putc(0, fd) == EOF)
|
|
|
8318 EMSG(_(e_write));
|
|
|
8319
|
|
|
8320 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
8321 _("Estimated runtime memory use: %d bytes"), spin->si_memtot);
|
|
|
8322 spell_message(spin, IObuff);
|
|
|
8323
|
|
|
8324 theend:
|
|
|
8325 /* close the file */
|
|
|
8326 fclose(fd);
|
|
|
8327 }
|
|
|
8328
|
|
|
8329 /*
|
|
|
8330 * Open a spell buffer. This is a nameless buffer that is not in the buffer
|
|
|
8331 * list and only contains text lines. Can use a swapfile to reduce memory
|
|
|
8332 * use.
|
|
|
8333 * Most other fields are invalid! Esp. watch out for string options being
|
|
|
8334 * NULL and there is no undo info.
|
|
|
8335 * Returns NULL when out of memory.
|
|
|
8336 */
|
|
|
8337 static buf_T *
|
|
|
8338 open_spellbuf()
|
|
|
8339 {
|
|
|
8340 buf_T *buf;
|
|
|
8341
|
|
|
8342 buf = (buf_T *)alloc_clear(sizeof(buf_T));
|
|
|
8343 if (buf != NULL)
|
|
|
8344 {
|
|
|
8345 buf->b_spell = TRUE;
|
|
|
8346 buf->b_p_swf = TRUE; /* may create a swap file */
|
|
|
8347 ml_open(buf);
|
|
|
8348 ml_open_file(buf); /* create swap file now */
|
|
|
8349 }
|
|
|
8350 return buf;
|
|
|
8351 }
|
|
|
8352
|
|
|
8353 /*
|
|
|
8354 * Close the buffer used for spell info.
|
|
|
8355 */
|
|
|
8356 static void
|
|
|
8357 close_spellbuf(buf)
|
|
|
8358 buf_T *buf;
|
|
|
8359 {
|
|
|
8360 if (buf != NULL)
|
|
|
8361 {
|
|
|
8362 ml_close(buf, TRUE);
|
|
|
8363 vim_free(buf);
|
|
|
8364 }
|
|
|
8365 }
|
|
|
8366
|
|
|
8367
|
|
|
8368 /*
|
|
310
|
8369 * Create a Vim spell file from one or more word lists.
|
|
|
8370 * "fnames[0]" is the output file name.
|
|
|
8371 * "fnames[fcount - 1]" is the last input file name.
|
|
|
8372 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
8373 * and ".spl" is appended to make the output file name.
|
|
|
8374 */
|
|
|
8375 static void
|
|
323
|
8376 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
8377 int fcount;
|
|
|
8378 char_u **fnames;
|
|
|
8379 int ascii; /* -ascii argument given */
|
|
|
8380 int overwrite; /* overwrite existing output file */
|
|
323
|
8381 int added_word; /* invoked through "zg" */
|
|
310
|
8382 {
|
|
236
|
8383 char_u fname[MAXPATHL];
|
|
|
8384 char_u wfname[MAXPATHL];
|
|
310
|
8385 char_u **innames;
|
|
|
8386 int incount;
|
|
236
|
8387 afffile_T *(afile[8]);
|
|
|
8388 int i;
|
|
|
8389 int len;
|
|
|
8390 struct stat st;
|
|
255
|
8391 int error = FALSE;
|
|
300
|
8392 spellinfo_T spin;
|
|
|
8393
|
|
|
8394 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
8395 spin.si_verbose = !added_word;
|
|
310
|
8396 spin.si_ascii = ascii;
|
|
323
|
8397 spin.si_followup = TRUE;
|
|
|
8398 spin.si_rem_accents = TRUE;
|
|
|
8399 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
625
|
8400 ga_init2(&spin.si_repsal, (int)sizeof(fromto_T), 20);
|
|
323
|
8401 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
8402 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
8403 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
625
|
8404 hash_init(&spin.si_commonwords);
|
|
500
|
8405 spin.si_newcompID = 127; /* start compound ID at first maximum */
|
|
310
|
8406
|
|
|
8407 /* default: fnames[0] is output file, following are input files */
|
|
|
8408 innames = &fnames[1];
|
|
|
8409 incount = fcount - 1;
|
|
|
8410
|
|
|
8411 if (fcount >= 1)
|
|
240
|
8412 {
|
|
310
|
8413 len = STRLEN(fnames[0]);
|
|
|
8414 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
8415 {
|
|
|
8416 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
8417 * "path/en.latin1.add.spl". */
|
|
|
8418 innames = &fnames[0];
|
|
|
8419 incount = 1;
|
|
|
8420 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
8421 }
|
|
366
|
8422 else if (fcount == 1)
|
|
|
8423 {
|
|
|
8424 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
8425 innames = &fnames[0];
|
|
|
8426 incount = 1;
|
|
|
8427 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
8428 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
8429 }
|
|
310
|
8430 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
8431 {
|
|
|
8432 /* Name ends in ".spl", use as the file name. */
|
|
323
|
8433 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
8434 }
|
|
|
8435 else
|
|
|
8436 /* Name should be language, make the file name from it. */
|
|
|
8437 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
8438 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
8439
|
|
|
8440 /* Check for .ascii.spl. */
|
|
|
8441 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
8442 spin.si_ascii = TRUE;
|
|
|
8443
|
|
|
8444 /* Check for .add.spl. */
|
|
|
8445 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
8446 spin.si_add = TRUE;
|
|
240
|
8447 }
|
|
|
8448
|
|
310
|
8449 if (incount <= 0)
|
|
236
|
8450 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
8451 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
8452 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
8453 else if (incount > 8)
|
|
236
|
8454 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
8455 else
|
|
|
8456 {
|
|
|
8457 /* Check for overwriting before doing things that may take a lot of
|
|
|
8458 * time. */
|
|
310
|
8459 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
8460 {
|
|
|
8461 EMSG(_(e_exists));
|
|
310
|
8462 return;
|
|
236
|
8463 }
|
|
310
|
8464 if (mch_isdir(wfname))
|
|
236
|
8465 {
|
|
310
|
8466 EMSG2(_(e_isadir2), wfname);
|
|
|
8467 return;
|
|
236
|
8468 }
|
|
|
8469
|
|
|
8470 /*
|
|
|
8471 * Init the aff and dic pointers.
|
|
|
8472 * Get the region names if there are more than 2 arguments.
|
|
|
8473 */
|
|
310
|
8474 for (i = 0; i < incount; ++i)
|
|
236
|
8475 {
|
|
310
|
8476 afile[i] = NULL;
|
|
300
|
8477
|
|
316
|
8478 if (incount > 1)
|
|
236
|
8479 {
|
|
310
|
8480 len = STRLEN(innames[i]);
|
|
|
8481 if (STRLEN(gettail(innames[i])) < 5
|
|
|
8482 || innames[i][len - 3] != '_')
|
|
236
|
8483 {
|
|
310
|
8484 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
8485 return;
|
|
236
|
8486 }
|
|
316
|
8487 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
8488 spin.si_region_name[i * 2 + 1] =
|
|
|
8489 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
8490 }
|
|
|
8491 }
|
|
316
|
8492 spin.si_region_count = incount;
|
|
236
|
8493
|
|
474
|
8494 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
8495 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
8496 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
8497 if (spin.si_foldroot == NULL
|
|
|
8498 || spin.si_keeproot == NULL
|
|
|
8499 || spin.si_prefroot == NULL)
|
|
300
|
8500 {
|
|
469
|
8501 free_blocks(spin.si_blocks);
|
|
310
|
8502 return;
|
|
300
|
8503 }
|
|
|
8504
|
|
351
|
8505 /* When not producing a .add.spl file clear the character table when
|
|
|
8506 * we encounter one in the .aff file. This means we dump the current
|
|
|
8507 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
8508 * better than guessing the contents, the table will match a
|
|
|
8509 * previously loaded spell file. */
|
|
|
8510 if (!spin.si_add)
|
|
|
8511 spin.si_clear_chartab = TRUE;
|
|
|
8512
|
|
236
|
8513 /*
|
|
|
8514 * Read all the .aff and .dic files.
|
|
|
8515 * Text is converted to 'encoding'.
|
|
300
|
8516 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
8517 */
|
|
310
|
8518 for (i = 0; i < incount && !error; ++i)
|
|
236
|
8519 {
|
|
300
|
8520 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
8521 spin.si_region = 1 << i;
|
|
|
8522
|
|
|
8523 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
8524 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
8525 {
|
|
|
8526 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
8527 * "SET" line. */
|
|
474
|
8528 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
8529 if (afile[i] == NULL)
|
|
300
|
8530 error = TRUE;
|
|
|
8531 else
|
|
|
8532 {
|
|
|
8533 /* Read the .dic file and store the words in the trees. */
|
|
|
8534 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
8535 innames[i]);
|
|
474
|
8536 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
8537 error = TRUE;
|
|
|
8538 }
|
|
|
8539 }
|
|
|
8540 else
|
|
|
8541 {
|
|
|
8542 /* No .aff file, try reading the file as a word list. Store
|
|
|
8543 * the words in the trees. */
|
|
474
|
8544 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
8545 error = TRUE;
|
|
|
8546 }
|
|
236
|
8547
|
|
310
|
8548 #ifdef FEAT_MBYTE
|
|
236
|
8549 /* Free any conversion stuff. */
|
|
300
|
8550 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
8551 #endif
|
|
236
|
8552 }
|
|
|
8553
|
|
492
|
8554 if (spin.si_compflags != NULL && spin.si_nobreak)
|
|
|
8555 MSG(_("Warning: both compounding and NOBREAK specified"));
|
|
|
8556
|
|
625
|
8557 if (!error && !got_int)
|
|
236
|
8558 {
|
|
|
8559 /*
|
|
300
|
8560 * Combine tails in the tree.
|
|
236
|
8561 */
|
|
625
|
8562 spell_message(&spin, (char_u *)_(msg_compressing));
|
|
474
|
8563 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
8564 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
8565 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
8566 }
|
|
|
8567
|
|
625
|
8568 if (!error && !got_int)
|
|
300
|
8569 {
|
|
|
8570 /*
|
|
|
8571 * Write the info in the spell file.
|
|
|
8572 */
|
|
625
|
8573 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
8574 _("Writing spell file %s ..."), wfname);
|
|
|
8575 spell_message(&spin, IObuff);
|
|
310
|
8576
|
|
500
|
8577 error = write_vim_spell(&spin, wfname) == FAIL;
|
|
474
|
8578
|
|
625
|
8579 spell_message(&spin, (char_u *)_("Done!"));
|
|
|
8580 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
8581 _("Estimated runtime memory use: %d bytes"), spin.si_memtot);
|
|
|
8582 spell_message(&spin, IObuff);
|
|
|
8583
|
|
|
8584 /*
|
|
|
8585 * If the file is loaded need to reload it.
|
|
|
8586 */
|
|
500
|
8587 if (!error)
|
|
|
8588 spell_reload_one(wfname, added_word);
|
|
300
|
8589 }
|
|
|
8590
|
|
|
8591 /* Free the allocated memory. */
|
|
323
|
8592 ga_clear(&spin.si_rep);
|
|
625
|
8593 ga_clear(&spin.si_repsal);
|
|
323
|
8594 ga_clear(&spin.si_sal);
|
|
|
8595 ga_clear(&spin.si_map);
|
|
339
|
8596 ga_clear(&spin.si_prefcond);
|
|
625
|
8597 hash_clear_all(&spin.si_commonwords, 0);
|
|
300
|
8598
|
|
|
8599 /* Free the .aff file structures. */
|
|
310
|
8600 for (i = 0; i < incount; ++i)
|
|
|
8601 if (afile[i] != NULL)
|
|
|
8602 spell_free_aff(afile[i]);
|
|
339
|
8603
|
|
|
8604 /* Free all the bits and pieces at once. */
|
|
|
8605 free_blocks(spin.si_blocks);
|
|
625
|
8606
|
|
|
8607 /*
|
|
|
8608 * If there is soundfolding info and no NOSUGFILE item create the
|
|
|
8609 * .sug file with the soundfolded word trie.
|
|
|
8610 */
|
|
|
8611 if (spin.si_sugtime != 0 && !error && !got_int)
|
|
|
8612 spell_make_sugfile(&spin, wfname);
|
|
|
8613
|
|
|
8614 }
|
|
|
8615 }
|
|
|
8616
|
|
|
8617 /*
|
|
|
8618 * Display a message for spell file processing when 'verbose' is set or using
|
|
|
8619 * ":mkspell". "str" can be IObuff.
|
|
|
8620 */
|
|
|
8621 static void
|
|
|
8622 spell_message(spin, str)
|
|
|
8623 spellinfo_T *spin;
|
|
|
8624 char_u *str;
|
|
|
8625 {
|
|
|
8626 if (spin->si_verbose || p_verbose > 2)
|
|
|
8627 {
|
|
|
8628 if (!spin->si_verbose)
|
|
|
8629 verbose_enter();
|
|
|
8630 MSG(str);
|
|
|
8631 out_flush();
|
|
|
8632 if (!spin->si_verbose)
|
|
|
8633 verbose_leave();
|
|
|
8634 }
|
|
|
8635 }
|
|
310
|
8636
|
|
|
8637 /*
|
|
385
|
8638 * ":[count]spellgood {word}"
|
|
|
8639 * ":[count]spellwrong {word}"
|
|
310
|
8640 */
|
|
|
8641 void
|
|
|
8642 ex_spell(eap)
|
|
|
8643 exarg_T *eap;
|
|
|
8644 {
|
|
381
|
8645 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
8646 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
8647 }
|
|
|
8648
|
|
310
|
8649 /*
|
|
|
8650 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
8651 */
|
|
|
8652 void
|
|
385
|
8653 spell_add_word(word, len, bad, index)
|
|
310
|
8654 char_u *word;
|
|
|
8655 int len;
|
|
|
8656 int bad;
|
|
385
|
8657 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
8658 'spellfile' */
|
|
310
|
8659 {
|
|
|
8660 FILE *fd;
|
|
385
|
8661 buf_T *buf = NULL;
|
|
351
|
8662 int new_spf = FALSE;
|
|
381
|
8663 char_u *fname;
|
|
385
|
8664 char_u fnamebuf[MAXPATHL];
|
|
|
8665 char_u line[MAXWLEN * 2];
|
|
|
8666 long fpos, fpos_next = 0;
|
|
|
8667 int i;
|
|
|
8668 char_u *spf;
|
|
|
8669
|
|
|
8670 if (index == 0) /* use internal wordlist */
|
|
|
8671 {
|
|
|
8672 if (int_wordlist == NULL)
|
|
|
8673 {
|
|
|
8674 int_wordlist = vim_tempname('s');
|
|
|
8675 if (int_wordlist == NULL)
|
|
381
|
8676 return;
|
|
|
8677 }
|
|
385
|
8678 fname = int_wordlist;
|
|
381
|
8679 }
|
|
310
|
8680 else
|
|
|
8681 {
|
|
381
|
8682 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
8683 if (*curbuf->b_p_spf == NUL)
|
|
|
8684 {
|
|
|
8685 init_spellfile();
|
|
|
8686 new_spf = TRUE;
|
|
|
8687 }
|
|
|
8688
|
|
|
8689 if (*curbuf->b_p_spf == NUL)
|
|
|
8690 {
|
|
523
|
8691 EMSG2(_(e_notset), "spellfile");
|
|
381
|
8692 return;
|
|
|
8693 }
|
|
|
8694
|
|
385
|
8695 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
8696 {
|
|
|
8697 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
8698 if (i == index)
|
|
|
8699 break;
|
|
|
8700 if (*spf == NUL)
|
|
|
8701 {
|
|
502
|
8702 EMSGN(_("E765: 'spellfile' does not have %ld entries"), index);
|
|
385
|
8703 return;
|
|
|
8704 }
|
|
|
8705 }
|
|
|
8706
|
|
310
|
8707 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
8708 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
8709 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
8710 buf = NULL;
|
|
|
8711 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
8712 {
|
|
310
|
8713 EMSG(_(e_bufloaded));
|
|
381
|
8714 return;
|
|
|
8715 }
|
|
|
8716
|
|
385
|
8717 fname = fnamebuf;
|
|
|
8718 }
|
|
|
8719
|
|
|
8720 if (bad)
|
|
|
8721 {
|
|
|
8722 /* When the word also appears as good word we need to remove that one,
|
|
|
8723 * since its flags sort before the one with WF_BANNED. */
|
|
|
8724 fd = mch_fopen((char *)fname, "r");
|
|
|
8725 if (fd != NULL)
|
|
|
8726 {
|
|
|
8727 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
8728 {
|
|
|
8729 fpos = fpos_next;
|
|
|
8730 fpos_next = ftell(fd);
|
|
|
8731 if (STRNCMP(word, line, len) == 0
|
|
|
8732 && (line[len] == '/' || line[len] < ' '))
|
|
|
8733 {
|
|
|
8734 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
8735 * the start of the line. Mixing reading and writing
|
|
|
8736 * doesn't work for all systems, close the file first. */
|
|
|
8737 fclose(fd);
|
|
|
8738 fd = mch_fopen((char *)fname, "r+");
|
|
|
8739 if (fd == NULL)
|
|
|
8740 break;
|
|
|
8741 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
8742 fputc('#', fd);
|
|
|
8743 fseek(fd, fpos_next, SEEK_SET);
|
|
|
8744 }
|
|
|
8745 }
|
|
|
8746 fclose(fd);
|
|
|
8747 }
|
|
381
|
8748 }
|
|
|
8749
|
|
|
8750 fd = mch_fopen((char *)fname, "a");
|
|
|
8751 if (fd == NULL && new_spf)
|
|
|
8752 {
|
|
|
8753 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
8754 * We may need to create the "spell" directory first. We already
|
|
|
8755 * checked the runtime directory is writable in init_spellfile(). */
|
|
592
|
8756 if (!dir_of_file_exists(fname))
|
|
381
|
8757 {
|
|
|
8758 /* The directory doesn't exist. Try creating it and opening the
|
|
|
8759 * file again. */
|
|
|
8760 vim_mkdir(NameBuff, 0755);
|
|
|
8761 fd = mch_fopen((char *)fname, "a");
|
|
|
8762 }
|
|
|
8763 }
|
|
|
8764
|
|
|
8765 if (fd == NULL)
|
|
|
8766 EMSG2(_(e_notopen), fname);
|
|
|
8767 else
|
|
|
8768 {
|
|
|
8769 if (bad)
|
|
|
8770 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
8771 else
|
|
381
|
8772 fprintf(fd, "%.*s\n", len, word);
|
|
|
8773 fclose(fd);
|
|
|
8774
|
|
|
8775 /* Update the .add.spl file. */
|
|
|
8776 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
8777
|
|
|
8778 /* If the .add file is edited somewhere, reload it. */
|
|
|
8779 if (buf != NULL)
|
|
627
|
8780 buf_reload(buf, buf->b_orig_mode);
|
|
381
|
8781
|
|
|
8782 redraw_all_later(NOT_VALID);
|
|
310
|
8783 }
|
|
|
8784 }
|
|
|
8785
|
|
|
8786 /*
|
|
|
8787 * Initialize 'spellfile' for the current buffer.
|
|
|
8788 */
|
|
|
8789 static void
|
|
|
8790 init_spellfile()
|
|
|
8791 {
|
|
|
8792 char_u buf[MAXPATHL];
|
|
|
8793 int l;
|
|
500
|
8794 char_u *fname;
|
|
310
|
8795 char_u *rtp;
|
|
351
|
8796 char_u *lend;
|
|
501
|
8797 int aspath = FALSE;
|
|
|
8798 char_u *lstart = curbuf->b_p_spl;
|
|
310
|
8799
|
|
|
8800 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
8801 {
|
|
501
|
8802 /* Find the end of the language name. Exclude the region. If there
|
|
|
8803 * is a path separator remember the start of the tail. */
|
|
351
|
8804 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
8805 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
501
|
8806 if (vim_ispathsep(*lend))
|
|
|
8807 {
|
|
|
8808 aspath = TRUE;
|
|
|
8809 lstart = lend + 1;
|
|
|
8810 }
|
|
351
|
8811
|
|
|
8812 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
8813 * are allowed to write. */
|
|
310
|
8814 rtp = p_rtp;
|
|
|
8815 while (*rtp != NUL)
|
|
|
8816 {
|
|
501
|
8817 if (aspath)
|
|
|
8818 /* Use directory of an entry with path, e.g., for
|
|
|
8819 * "/dir/lg.utf-8.spl" use "/dir". */
|
|
|
8820 vim_strncpy(buf, curbuf->b_p_spl, lstart - curbuf->b_p_spl - 1);
|
|
|
8821 else
|
|
|
8822 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
8823 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
310
|
8824 if (filewritable(buf) == 2)
|
|
|
8825 {
|
|
316
|
8826 /* Use the first language name from 'spelllang' and the
|
|
|
8827 * encoding used in the first loaded .spl file. */
|
|
501
|
8828 if (aspath)
|
|
|
8829 vim_strncpy(buf, curbuf->b_p_spl, lend - curbuf->b_p_spl);
|
|
|
8830 else
|
|
|
8831 {
|
|
|
8832 l = STRLEN(buf);
|
|
|
8833 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
|
8834 "/spell/%.*s", (int)(lend - lstart), lstart);
|
|
|
8835 }
|
|
310
|
8836 l = STRLEN(buf);
|
|
501
|
8837 fname = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang->sl_fname;
|
|
|
8838 vim_snprintf((char *)buf + l, MAXPATHL - l, ".%s.add",
|
|
|
8839 fname != NULL
|
|
|
8840 && strstr((char *)gettail(fname), ".ascii.") != NULL
|
|
|
8841 ? (char_u *)"ascii" : spell_enc());
|
|
310
|
8842 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
8843 break;
|
|
|
8844 }
|
|
501
|
8845 aspath = FALSE;
|
|
310
|
8846 }
|
|
|
8847 }
|
|
|
8848 }
|
|
236
|
8849
|
|
300
|
8850
|
|
307
|
8851 /*
|
|
|
8852 * Init the chartab used for spelling for ASCII.
|
|
|
8853 * EBCDIC is not supported!
|
|
|
8854 */
|
|
|
8855 static void
|
|
|
8856 clear_spell_chartab(sp)
|
|
|
8857 spelltab_T *sp;
|
|
|
8858 {
|
|
324
|
8859 int i;
|
|
307
|
8860
|
|
|
8861 /* Init everything to FALSE. */
|
|
|
8862 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
8863 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
8864 for (i = 0; i < 256; ++i)
|
|
324
|
8865 {
|
|
307
|
8866 sp->st_fold[i] = i;
|
|
324
|
8867 sp->st_upper[i] = i;
|
|
|
8868 }
|
|
307
|
8869
|
|
|
8870 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
8871 * that is done separately. */
|
|
|
8872 for (i = '0'; i <= '9'; ++i)
|
|
|
8873 sp->st_isw[i] = TRUE;
|
|
|
8874 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
8875 {
|
|
|
8876 sp->st_isw[i] = TRUE;
|
|
|
8877 sp->st_isu[i] = TRUE;
|
|
|
8878 sp->st_fold[i] = i + 0x20;
|
|
|
8879 }
|
|
|
8880 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
8881 {
|
|
307
|
8882 sp->st_isw[i] = TRUE;
|
|
324
|
8883 sp->st_upper[i] = i - 0x20;
|
|
|
8884 }
|
|
307
|
8885 }
|
|
|
8886
|
|
|
8887 /*
|
|
|
8888 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
8889 * Called once while starting up and when 'encoding' changes.
|
|
|
8890 * The default is to use isalpha(), but the spell file should define the word
|
|
|
8891 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
8892 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
8893 */
|
|
|
8894 void
|
|
|
8895 init_spell_chartab()
|
|
|
8896 {
|
|
|
8897 int i;
|
|
|
8898
|
|
|
8899 did_set_spelltab = FALSE;
|
|
|
8900 clear_spell_chartab(&spelltab);
|
|
|
8901 #ifdef FEAT_MBYTE
|
|
|
8902 if (enc_dbcs)
|
|
|
8903 {
|
|
|
8904 /* DBCS: assume double-wide characters are word characters. */
|
|
|
8905 for (i = 128; i <= 255; ++i)
|
|
|
8906 if (MB_BYTE2LEN(i) == 2)
|
|
|
8907 spelltab.st_isw[i] = TRUE;
|
|
|
8908 }
|
|
324
|
8909 else if (enc_utf8)
|
|
|
8910 {
|
|
|
8911 for (i = 128; i < 256; ++i)
|
|
|
8912 {
|
|
|
8913 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
8914 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
8915 spelltab.st_fold[i] = utf_fold(i);
|
|
|
8916 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
8917 }
|
|
|
8918 }
|
|
307
|
8919 else
|
|
|
8920 #endif
|
|
|
8921 {
|
|
324
|
8922 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
8923 for (i = 128; i < 256; ++i)
|
|
|
8924 {
|
|
|
8925 if (MB_ISUPPER(i))
|
|
|
8926 {
|
|
324
|
8927 spelltab.st_isw[i] = TRUE;
|
|
307
|
8928 spelltab.st_isu[i] = TRUE;
|
|
|
8929 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
8930 }
|
|
324
|
8931 else if (MB_ISLOWER(i))
|
|
|
8932 {
|
|
|
8933 spelltab.st_isw[i] = TRUE;
|
|
|
8934 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
8935 }
|
|
307
|
8936 }
|
|
|
8937 }
|
|
|
8938 }
|
|
|
8939
|
|
|
8940 /*
|
|
|
8941 * Set the spell character tables from strings in the affix file.
|
|
|
8942 */
|
|
|
8943 static int
|
|
|
8944 set_spell_chartab(fol, low, upp)
|
|
|
8945 char_u *fol;
|
|
|
8946 char_u *low;
|
|
|
8947 char_u *upp;
|
|
|
8948 {
|
|
|
8949 /* We build the new tables here first, so that we can compare with the
|
|
|
8950 * previous one. */
|
|
|
8951 spelltab_T new_st;
|
|
|
8952 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
8953 int f, l, u;
|
|
|
8954
|
|
|
8955 clear_spell_chartab(&new_st);
|
|
|
8956
|
|
|
8957 while (*pf != NUL)
|
|
|
8958 {
|
|
|
8959 if (*pl == NUL || *pu == NUL)
|
|
|
8960 {
|
|
|
8961 EMSG(_(e_affform));
|
|
|
8962 return FAIL;
|
|
|
8963 }
|
|
|
8964 #ifdef FEAT_MBYTE
|
|
|
8965 f = mb_ptr2char_adv(&pf);
|
|
|
8966 l = mb_ptr2char_adv(&pl);
|
|
|
8967 u = mb_ptr2char_adv(&pu);
|
|
|
8968 #else
|
|
|
8969 f = *pf++;
|
|
|
8970 l = *pl++;
|
|
|
8971 u = *pu++;
|
|
|
8972 #endif
|
|
|
8973 /* Every character that appears is a word character. */
|
|
|
8974 if (f < 256)
|
|
|
8975 new_st.st_isw[f] = TRUE;
|
|
|
8976 if (l < 256)
|
|
|
8977 new_st.st_isw[l] = TRUE;
|
|
|
8978 if (u < 256)
|
|
|
8979 new_st.st_isw[u] = TRUE;
|
|
|
8980
|
|
|
8981 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
8982 * case-folding */
|
|
|
8983 if (l < 256 && l != f)
|
|
|
8984 {
|
|
|
8985 if (f >= 256)
|
|
|
8986 {
|
|
|
8987 EMSG(_(e_affrange));
|
|
|
8988 return FAIL;
|
|
|
8989 }
|
|
|
8990 new_st.st_fold[l] = f;
|
|
|
8991 }
|
|
|
8992
|
|
|
8993 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
8994 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
8995 * "FOL" . */
|
|
307
|
8996 if (u < 256 && u != f)
|
|
|
8997 {
|
|
|
8998 if (f >= 256)
|
|
|
8999 {
|
|
|
9000 EMSG(_(e_affrange));
|
|
|
9001 return FAIL;
|
|
|
9002 }
|
|
|
9003 new_st.st_fold[u] = f;
|
|
|
9004 new_st.st_isu[u] = TRUE;
|
|
324
|
9005 new_st.st_upper[f] = u;
|
|
307
|
9006 }
|
|
|
9007 }
|
|
|
9008
|
|
|
9009 if (*pl != NUL || *pu != NUL)
|
|
|
9010 {
|
|
|
9011 EMSG(_(e_affform));
|
|
|
9012 return FAIL;
|
|
|
9013 }
|
|
|
9014
|
|
|
9015 return set_spell_finish(&new_st);
|
|
|
9016 }
|
|
|
9017
|
|
|
9018 /*
|
|
|
9019 * Set the spell character tables from strings in the .spl file.
|
|
|
9020 */
|
|
483
|
9021 static void
|
|
389
|
9022 set_spell_charflags(flags, cnt, fol)
|
|
307
|
9023 char_u *flags;
|
|
389
|
9024 int cnt; /* length of "flags" */
|
|
|
9025 char_u *fol;
|
|
307
|
9026 {
|
|
|
9027 /* We build the new tables here first, so that we can compare with the
|
|
|
9028 * previous one. */
|
|
|
9029 spelltab_T new_st;
|
|
|
9030 int i;
|
|
389
|
9031 char_u *p = fol;
|
|
324
|
9032 int c;
|
|
307
|
9033
|
|
|
9034 clear_spell_chartab(&new_st);
|
|
|
9035
|
|
389
|
9036 for (i = 0; i < 128; ++i)
|
|
|
9037 {
|
|
|
9038 if (i < cnt)
|
|
|
9039 {
|
|
|
9040 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
9041 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
9042 }
|
|
|
9043
|
|
|
9044 if (*p != NUL)
|
|
|
9045 {
|
|
307
|
9046 #ifdef FEAT_MBYTE
|
|
389
|
9047 c = mb_ptr2char_adv(&p);
|
|
307
|
9048 #else
|
|
389
|
9049 c = *p++;
|
|
307
|
9050 #endif
|
|
389
|
9051 new_st.st_fold[i + 128] = c;
|
|
|
9052 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
9053 new_st.st_upper[c] = i + 128;
|
|
|
9054 }
|
|
307
|
9055 }
|
|
|
9056
|
|
483
|
9057 (void)set_spell_finish(&new_st);
|
|
307
|
9058 }
|
|
|
9059
|
|
|
9060 static int
|
|
|
9061 set_spell_finish(new_st)
|
|
|
9062 spelltab_T *new_st;
|
|
|
9063 {
|
|
|
9064 int i;
|
|
|
9065
|
|
|
9066 if (did_set_spelltab)
|
|
|
9067 {
|
|
|
9068 /* check that it's the same table */
|
|
|
9069 for (i = 0; i < 256; ++i)
|
|
|
9070 {
|
|
|
9071 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
9072 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
9073 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
9074 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
9075 {
|
|
|
9076 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
9077 return FAIL;
|
|
|
9078 }
|
|
|
9079 }
|
|
|
9080 }
|
|
|
9081 else
|
|
|
9082 {
|
|
|
9083 /* copy the new spelltab into the one being used */
|
|
|
9084 spelltab = *new_st;
|
|
|
9085 did_set_spelltab = TRUE;
|
|
|
9086 }
|
|
|
9087
|
|
|
9088 return OK;
|
|
|
9089 }
|
|
|
9090
|
|
|
9091 /*
|
|
358
|
9092 * Return TRUE if "p" points to a word character.
|
|
366
|
9093 * As a special case we see "midword" characters as word character when it is
|
|
358
|
9094 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
9095 * Thus this only works properly when past the first character of the word.
|
|
358
|
9096 */
|
|
|
9097 static int
|
|
376
|
9098 spell_iswordp(p, buf)
|
|
358
|
9099 char_u *p;
|
|
376
|
9100 buf_T *buf; /* buffer used */
|
|
358
|
9101 {
|
|
366
|
9102 #ifdef FEAT_MBYTE
|
|
358
|
9103 char_u *s;
|
|
366
|
9104 int l;
|
|
|
9105 int c;
|
|
|
9106
|
|
|
9107 if (has_mbyte)
|
|
|
9108 {
|
|
|
9109 l = MB_BYTE2LEN(*p);
|
|
358
|
9110 s = p;
|
|
366
|
9111 if (l == 1)
|
|
|
9112 {
|
|
|
9113 /* be quick for ASCII */
|
|
376
|
9114 if (buf->b_spell_ismw[*p])
|
|
366
|
9115 {
|
|
|
9116 s = p + 1; /* skip a mid-word character */
|
|
|
9117 l = MB_BYTE2LEN(*s);
|
|
|
9118 }
|
|
|
9119 }
|
|
|
9120 else
|
|
|
9121 {
|
|
|
9122 c = mb_ptr2char(p);
|
|
376
|
9123 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
9124 : (buf->b_spell_ismw_mb != NULL
|
|
|
9125 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
9126 {
|
|
|
9127 s = p + l;
|
|
|
9128 l = MB_BYTE2LEN(*s);
|
|
|
9129 }
|
|
|
9130 }
|
|
|
9131
|
|
390
|
9132 c = mb_ptr2char(s);
|
|
|
9133 if (c > 255)
|
|
366
|
9134 return mb_get_class(s) >= 2;
|
|
390
|
9135 return spelltab.st_isw[c];
|
|
366
|
9136 }
|
|
358
|
9137 #endif
|
|
366
|
9138
|
|
376
|
9139 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
9140 }
|
|
|
9141
|
|
|
9142 /*
|
|
|
9143 * Return TRUE if "p" points to a word character.
|
|
|
9144 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
9145 */
|
|
|
9146 static int
|
|
|
9147 spell_iswordp_nmw(p)
|
|
|
9148 char_u *p;
|
|
|
9149 {
|
|
|
9150 #ifdef FEAT_MBYTE
|
|
390
|
9151 int c;
|
|
|
9152
|
|
|
9153 if (has_mbyte)
|
|
|
9154 {
|
|
|
9155 c = mb_ptr2char(p);
|
|
|
9156 if (c > 255)
|
|
|
9157 return mb_get_class(p) >= 2;
|
|
|
9158 return spelltab.st_isw[c];
|
|
|
9159 }
|
|
376
|
9160 #endif
|
|
|
9161 return spelltab.st_isw[*p];
|
|
358
|
9162 }
|
|
|
9163
|
|
372
|
9164 #ifdef FEAT_MBYTE
|
|
|
9165 /*
|
|
|
9166 * Return TRUE if "p" points to a word character.
|
|
|
9167 * Wide version of spell_iswordp().
|
|
|
9168 */
|
|
|
9169 static int
|
|
376
|
9170 spell_iswordp_w(p, buf)
|
|
372
|
9171 int *p;
|
|
376
|
9172 buf_T *buf;
|
|
372
|
9173 {
|
|
|
9174 int *s;
|
|
|
9175
|
|
376
|
9176 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
9177 : (buf->b_spell_ismw_mb != NULL
|
|
|
9178 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
9179 s = p + 1;
|
|
|
9180 else
|
|
|
9181 s = p;
|
|
|
9182
|
|
390
|
9183 if (*s > 255)
|
|
372
|
9184 {
|
|
|
9185 if (enc_utf8)
|
|
|
9186 return utf_class(*s) >= 2;
|
|
|
9187 if (enc_dbcs)
|
|
|
9188 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
9189 return 0;
|
|
|
9190 }
|
|
|
9191 return spelltab.st_isw[*s];
|
|
|
9192 }
|
|
|
9193 #endif
|
|
|
9194
|
|
358
|
9195 /*
|
|
339
|
9196 * Write the table with prefix conditions to the .spl file.
|
|
483
|
9197 * When "fd" is NULL only count the length of what is written.
|
|
|
9198 */
|
|
|
9199 static int
|
|
339
|
9200 write_spell_prefcond(fd, gap)
|
|
|
9201 FILE *fd;
|
|
|
9202 garray_T *gap;
|
|
|
9203 {
|
|
|
9204 int i;
|
|
|
9205 char_u *p;
|
|
|
9206 int len;
|
|
483
|
9207 int totlen;
|
|
|
9208
|
|
|
9209 if (fd != NULL)
|
|
|
9210 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
9211
|
|
|
9212 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */
|
|
339
|
9213
|
|
|
9214 for (i = 0; i < gap->ga_len; ++i)
|
|
|
9215 {
|
|
|
9216 /* <prefcond> : <condlen> <condstr> */
|
|
|
9217 p = ((char_u **)gap->ga_data)[i];
|
|
483
|
9218 if (p != NULL)
|
|
339
|
9219 {
|
|
|
9220 len = STRLEN(p);
|
|
483
|
9221 if (fd != NULL)
|
|
|
9222 {
|
|
|
9223 fputc(len, fd);
|
|
|
9224 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
9225 }
|
|
|
9226 totlen += len;
|
|
|
9227 }
|
|
|
9228 else if (fd != NULL)
|
|
|
9229 fputc(0, fd);
|
|
|
9230 }
|
|
|
9231
|
|
|
9232 return totlen;
|
|
307
|
9233 }
|
|
|
9234
|
|
|
9235 /*
|
|
324
|
9236 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
9237 * Uses the character definitions from the .spl file.
|
|
307
|
9238 * When using a multi-byte 'encoding' the length may change!
|
|
|
9239 * Returns FAIL when something wrong.
|
|
|
9240 */
|
|
|
9241 static int
|
|
324
|
9242 spell_casefold(str, len, buf, buflen)
|
|
|
9243 char_u *str;
|
|
307
|
9244 int len;
|
|
|
9245 char_u *buf;
|
|
|
9246 int buflen;
|
|
|
9247 {
|
|
|
9248 int i;
|
|
|
9249
|
|
|
9250 if (len >= buflen)
|
|
|
9251 {
|
|
|
9252 buf[0] = NUL;
|
|
|
9253 return FAIL; /* result will not fit */
|
|
|
9254 }
|
|
|
9255
|
|
|
9256 #ifdef FEAT_MBYTE
|
|
|
9257 if (has_mbyte)
|
|
|
9258 {
|
|
324
|
9259 int outi = 0;
|
|
|
9260 char_u *p;
|
|
307
|
9261 int c;
|
|
|
9262
|
|
|
9263 /* Fold one character at a time. */
|
|
324
|
9264 for (p = str; p < str + len; )
|
|
307
|
9265 {
|
|
|
9266 if (outi + MB_MAXBYTES > buflen)
|
|
|
9267 {
|
|
|
9268 buf[outi] = NUL;
|
|
|
9269 return FAIL;
|
|
|
9270 }
|
|
474
|
9271 c = mb_cptr2char_adv(&p);
|
|
324
|
9272 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
9273 }
|
|
|
9274 buf[outi] = NUL;
|
|
|
9275 }
|
|
|
9276 else
|
|
|
9277 #endif
|
|
|
9278 {
|
|
|
9279 /* Be quick for non-multibyte encodings. */
|
|
|
9280 for (i = 0; i < len; ++i)
|
|
324
|
9281 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
9282 buf[i] = NUL;
|
|
|
9283 }
|
|
|
9284
|
|
|
9285 return OK;
|
|
|
9286 }
|
|
|
9287
|
|
625
|
9288 /* values for sps_flags */
|
|
372
|
9289 #define SPS_BEST 1
|
|
|
9290 #define SPS_FAST 2
|
|
|
9291 #define SPS_DOUBLE 4
|
|
|
9292
|
|
625
|
9293 static int sps_flags = SPS_BEST; /* flags from 'spellsuggest' */
|
|
|
9294 static int sps_limit = 9999; /* max nr of suggestions given */
|
|
372
|
9295
|
|
|
9296 /*
|
|
|
9297 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
483
|
9298 * Sets "sps_flags" and "sps_limit".
|
|
372
|
9299 */
|
|
|
9300 int
|
|
|
9301 spell_check_sps()
|
|
|
9302 {
|
|
|
9303 char_u *p;
|
|
483
|
9304 char_u *s;
|
|
372
|
9305 char_u buf[MAXPATHL];
|
|
|
9306 int f;
|
|
|
9307
|
|
|
9308 sps_flags = 0;
|
|
483
|
9309 sps_limit = 9999;
|
|
372
|
9310
|
|
|
9311 for (p = p_sps; *p != NUL; )
|
|
|
9312 {
|
|
|
9313 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
9314
|
|
|
9315 f = 0;
|
|
483
|
9316 if (VIM_ISDIGIT(*buf))
|
|
|
9317 {
|
|
|
9318 s = buf;
|
|
|
9319 sps_limit = getdigits(&s);
|
|
|
9320 if (*s != NUL && !VIM_ISDIGIT(*s))
|
|
|
9321 f = -1;
|
|
|
9322 }
|
|
|
9323 else if (STRCMP(buf, "best") == 0)
|
|
372
|
9324 f = SPS_BEST;
|
|
|
9325 else if (STRCMP(buf, "fast") == 0)
|
|
|
9326 f = SPS_FAST;
|
|
|
9327 else if (STRCMP(buf, "double") == 0)
|
|
|
9328 f = SPS_DOUBLE;
|
|
|
9329 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
9330 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
9331 f = -1;
|
|
|
9332
|
|
|
9333 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
9334 {
|
|
|
9335 sps_flags = SPS_BEST;
|
|
483
|
9336 sps_limit = 9999;
|
|
372
|
9337 return FAIL;
|
|
|
9338 }
|
|
|
9339 if (f != 0)
|
|
|
9340 sps_flags = f;
|
|
|
9341 }
|
|
|
9342
|
|
|
9343 if (sps_flags == 0)
|
|
|
9344 sps_flags = SPS_BEST;
|
|
|
9345
|
|
|
9346 return OK;
|
|
|
9347 }
|
|
|
9348
|
|
323
|
9349 /*
|
|
|
9350 * "z?": Find badly spelled word under or after the cursor.
|
|
|
9351 * Give suggestions for the properly spelled word.
|
|
638
|
9352 * In Visual mode use the highlighted word as the bad word.
|
|
485
|
9353 * When "count" is non-zero use that suggestion.
|
|
323
|
9354 */
|
|
|
9355 void
|
|
485
|
9356 spell_suggest(count)
|
|
|
9357 int count;
|
|
323
|
9358 {
|
|
|
9359 char_u *line;
|
|
|
9360 pos_T prev_cursor = curwin->w_cursor;
|
|
|
9361 char_u wcopy[MAXWLEN + 2];
|
|
|
9362 char_u *p;
|
|
|
9363 int i;
|
|
|
9364 int c;
|
|
|
9365 suginfo_T sug;
|
|
|
9366 suggest_T *stp;
|
|
372
|
9367 int mouse_used;
|
|
388
|
9368 int need_cap;
|
|
483
|
9369 int limit;
|
|
485
|
9370 int selected = count;
|
|
638
|
9371 int badlen = 0;
|
|
|
9372
|
|
|
9373 if (no_spell_checking(curwin))
|
|
|
9374 return;
|
|
|
9375
|
|
|
9376 #ifdef FEAT_VISUAL
|
|
|
9377 if (VIsual_active)
|
|
|
9378 {
|
|
|
9379 /* Use the Visually selected text as the bad word. But reject
|
|
|
9380 * a multi-line selection. */
|
|
|
9381 if (curwin->w_cursor.lnum != VIsual.lnum)
|
|
|
9382 {
|
|
|
9383 vim_beep();
|
|
|
9384 return;
|
|
|
9385 }
|
|
|
9386 badlen = (int)curwin->w_cursor.col - (int)VIsual.col;
|
|
|
9387 if (badlen < 0)
|
|
|
9388 badlen = -badlen;
|
|
|
9389 else
|
|
|
9390 curwin->w_cursor.col = VIsual.col;
|
|
|
9391 ++badlen;
|
|
|
9392 end_visual_mode();
|
|
|
9393 }
|
|
|
9394 else
|
|
|
9395 #endif
|
|
|
9396 /* Find the start of the badly spelled word. */
|
|
|
9397 if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0
|
|
346
|
9398 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
9399 {
|
|
|
9400 /* No bad word or it starts after the cursor: use the word under the
|
|
|
9401 * cursor. */
|
|
|
9402 curwin->w_cursor = prev_cursor;
|
|
|
9403 line = ml_get_curline();
|
|
|
9404 p = line + curwin->w_cursor.col;
|
|
|
9405 /* Backup to before start of word. */
|
|
390
|
9406 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
9407 mb_ptr_back(line, p);
|
|
|
9408 /* Forward to start of word. */
|
|
390
|
9409 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
9410 mb_ptr_adv(p);
|
|
|
9411
|
|
390
|
9412 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
9413 {
|
|
|
9414 beep_flush();
|
|
|
9415 return;
|
|
|
9416 }
|
|
|
9417 curwin->w_cursor.col = p - line;
|
|
323
|
9418 }
|
|
|
9419
|
|
344
|
9420 /* Get the word and its length. */
|
|
|
9421
|
|
388
|
9422 /* Figure out if the word should be capitalised. */
|
|
475
|
9423 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
9424
|
|
|
9425 line = ml_get_curline();
|
|
388
|
9426
|
|
483
|
9427 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in
|
|
|
9428 * 'spellsuggest', whatever is smaller. */
|
|
|
9429 if (sps_limit > (int)Rows - 2)
|
|
|
9430 limit = (int)Rows - 2;
|
|
|
9431 else
|
|
|
9432 limit = sps_limit;
|
|
638
|
9433 spell_find_suggest(line + curwin->w_cursor.col, badlen, &sug, limit,
|
|
625
|
9434 TRUE, need_cap, TRUE);
|
|
323
|
9435
|
|
|
9436 if (sug.su_ga.ga_len == 0)
|
|
|
9437 MSG(_("Sorry, no suggestions"));
|
|
485
|
9438 else if (count > 0)
|
|
|
9439 {
|
|
|
9440 if (count > sug.su_ga.ga_len)
|
|
|
9441 smsg((char_u *)_("Sorry, only %ld suggestions"),
|
|
|
9442 (long)sug.su_ga.ga_len);
|
|
|
9443 }
|
|
323
|
9444 else
|
|
|
9445 {
|
|
372
|
9446 vim_free(repl_from);
|
|
|
9447 repl_from = NULL;
|
|
|
9448 vim_free(repl_to);
|
|
|
9449 repl_to = NULL;
|
|
|
9450
|
|
474
|
9451 #ifdef FEAT_RIGHTLEFT
|
|
|
9452 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
9453 cmdmsg_rl = curwin->w_p_rl;
|
|
|
9454 if (cmdmsg_rl)
|
|
|
9455 msg_col = Columns - 1;
|
|
|
9456 #endif
|
|
|
9457
|
|
323
|
9458 /* List the suggestions. */
|
|
|
9459 msg_start();
|
|
372
|
9460 lines_left = Rows; /* avoid more prompt */
|
|
323
|
9461 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
9462 sug.su_badlen, sug.su_badptr);
|
|
474
|
9463 #ifdef FEAT_RIGHTLEFT
|
|
|
9464 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
9465 {
|
|
|
9466 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
9467 * untranslated message rightleft. */
|
|
|
9468 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
9469 sug.su_badlen, sug.su_badptr);
|
|
|
9470 }
|
|
|
9471 #endif
|
|
323
|
9472 msg_puts(IObuff);
|
|
|
9473 msg_clr_eos();
|
|
|
9474 msg_putchar('\n');
|
|
346
|
9475
|
|
323
|
9476 msg_scroll = TRUE;
|
|
|
9477 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
9478 {
|
|
344
|
9479 stp = &SUG(sug.su_ga, i);
|
|
323
|
9480
|
|
|
9481 /* The suggested word may replace only part of the bad word, add
|
|
|
9482 * the not replaced part. */
|
|
|
9483 STRCPY(wcopy, stp->st_word);
|
|
|
9484 if (sug.su_badlen > stp->st_orglen)
|
|
625
|
9485 vim_strncpy(wcopy + stp->st_wordlen,
|
|
323
|
9486 sug.su_badptr + stp->st_orglen,
|
|
|
9487 sug.su_badlen - stp->st_orglen);
|
|
474
|
9488 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
9489 #ifdef FEAT_RIGHTLEFT
|
|
|
9490 if (cmdmsg_rl)
|
|
|
9491 rl_mirror(IObuff);
|
|
|
9492 #endif
|
|
|
9493 msg_puts(IObuff);
|
|
|
9494
|
|
|
9495 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
9496 msg_puts(IObuff);
|
|
|
9497
|
|
|
9498 /* The word may replace more than "su_badlen". */
|
|
|
9499 if (sug.su_badlen < stp->st_orglen)
|
|
|
9500 {
|
|
|
9501 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
9502 stp->st_orglen, sug.su_badptr);
|
|
|
9503 msg_puts(IObuff);
|
|
|
9504 }
|
|
|
9505
|
|
324
|
9506 if (p_verbose > 0)
|
|
344
|
9507 {
|
|
346
|
9508 /* Add the score. */
|
|
351
|
9509 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
9510 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
9511 stp->st_salscore ? "s " : "",
|
|
|
9512 stp->st_score, stp->st_altscore);
|
|
|
9513 else
|
|
474
|
9514 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
9515 stp->st_score);
|
|
474
|
9516 #ifdef FEAT_RIGHTLEFT
|
|
|
9517 if (cmdmsg_rl)
|
|
|
9518 /* Mirror the numbers, but keep the leading space. */
|
|
|
9519 rl_mirror(IObuff + 1);
|
|
|
9520 #endif
|
|
346
|
9521 msg_advance(30);
|
|
|
9522 msg_puts(IObuff);
|
|
344
|
9523 }
|
|
323
|
9524 msg_putchar('\n');
|
|
|
9525 }
|
|
|
9526
|
|
474
|
9527 #ifdef FEAT_RIGHTLEFT
|
|
|
9528 cmdmsg_rl = FALSE;
|
|
|
9529 msg_col = 0;
|
|
|
9530 #endif
|
|
323
|
9531 /* Ask for choice. */
|
|
485
|
9532 selected = prompt_for_number(&mouse_used);
|
|
372
|
9533 if (mouse_used)
|
|
485
|
9534 selected -= lines_left;
|
|
|
9535 }
|
|
|
9536
|
|
|
9537 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
|
9538 {
|
|
|
9539 /* Save the from and to text for :spellrepall. */
|
|
|
9540 stp = &SUG(sug.su_ga, selected - 1);
|
|
548
|
9541 if (sug.su_badlen > stp->st_orglen)
|
|
|
9542 {
|
|
|
9543 /* Replacing less than "su_badlen", append the remainder to
|
|
|
9544 * repl_to. */
|
|
|
9545 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen);
|
|
|
9546 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word,
|
|
|
9547 sug.su_badlen - stp->st_orglen,
|
|
|
9548 sug.su_badptr + stp->st_orglen);
|
|
|
9549 repl_to = vim_strsave(IObuff);
|
|
|
9550 }
|
|
|
9551 else
|
|
|
9552 {
|
|
|
9553 /* Replacing su_badlen or more, use the whole word. */
|
|
|
9554 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
9555 repl_to = vim_strsave(stp->st_word);
|
|
|
9556 }
|
|
485
|
9557
|
|
|
9558 /* Replace the word. */
|
|
625
|
9559 p = alloc(STRLEN(line) - stp->st_orglen + stp->st_wordlen + 1);
|
|
485
|
9560 if (p != NULL)
|
|
|
9561 {
|
|
|
9562 c = sug.su_badptr - line;
|
|
|
9563 mch_memmove(p, line, c);
|
|
|
9564 STRCPY(p + c, stp->st_word);
|
|
|
9565 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
9566 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
9567 curwin->w_cursor.col = c;
|
|
|
9568 changed_bytes(curwin->w_cursor.lnum, c);
|
|
|
9569
|
|
|
9570 /* For redo we use a change-word command. */
|
|
|
9571 ResetRedobuff();
|
|
|
9572 AppendToRedobuff((char_u *)"ciw");
|
|
620
|
9573 AppendToRedobuffLit(p + c,
|
|
625
|
9574 stp->st_wordlen + sug.su_badlen - stp->st_orglen);
|
|
485
|
9575 AppendCharToRedobuff(ESC);
|
|
|
9576 }
|
|
|
9577 }
|
|
|
9578 else
|
|
|
9579 curwin->w_cursor = prev_cursor;
|
|
323
|
9580
|
|
344
|
9581 spell_find_cleanup(&sug);
|
|
|
9582 }
|
|
|
9583
|
|
|
9584 /*
|
|
475
|
9585 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
9586 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
9587 */
|
|
|
9588 static int
|
|
|
9589 check_need_cap(lnum, col)
|
|
|
9590 linenr_T lnum;
|
|
|
9591 colnr_T col;
|
|
|
9592 {
|
|
|
9593 int need_cap = FALSE;
|
|
|
9594 char_u *line;
|
|
|
9595 char_u *line_copy = NULL;
|
|
|
9596 char_u *p;
|
|
|
9597 colnr_T endcol;
|
|
|
9598 regmatch_T regmatch;
|
|
|
9599
|
|
|
9600 if (curbuf->b_cap_prog == NULL)
|
|
|
9601 return FALSE;
|
|
|
9602
|
|
|
9603 line = ml_get_curline();
|
|
|
9604 endcol = 0;
|
|
|
9605 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
9606 {
|
|
|
9607 /* At start of line, check if previous line is empty or sentence
|
|
|
9608 * ends there. */
|
|
|
9609 if (lnum == 1)
|
|
|
9610 need_cap = TRUE;
|
|
|
9611 else
|
|
|
9612 {
|
|
|
9613 line = ml_get(lnum - 1);
|
|
|
9614 if (*skipwhite(line) == NUL)
|
|
|
9615 need_cap = TRUE;
|
|
|
9616 else
|
|
|
9617 {
|
|
|
9618 /* Append a space in place of the line break. */
|
|
|
9619 line_copy = concat_str(line, (char_u *)" ");
|
|
|
9620 line = line_copy;
|
|
|
9621 endcol = STRLEN(line);
|
|
|
9622 }
|
|
|
9623 }
|
|
|
9624 }
|
|
|
9625 else
|
|
|
9626 endcol = col;
|
|
|
9627
|
|
|
9628 if (endcol > 0)
|
|
|
9629 {
|
|
|
9630 /* Check if sentence ends before the bad word. */
|
|
|
9631 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
9632 regmatch.rm_ic = FALSE;
|
|
|
9633 p = line + endcol;
|
|
|
9634 for (;;)
|
|
|
9635 {
|
|
|
9636 mb_ptr_back(line, p);
|
|
|
9637 if (p == line || spell_iswordp_nmw(p))
|
|
|
9638 break;
|
|
|
9639 if (vim_regexec(®match, p, 0)
|
|
|
9640 && regmatch.endp[0] == line + endcol)
|
|
|
9641 {
|
|
|
9642 need_cap = TRUE;
|
|
|
9643 break;
|
|
|
9644 }
|
|
|
9645 }
|
|
|
9646 }
|
|
|
9647
|
|
|
9648 vim_free(line_copy);
|
|
|
9649
|
|
|
9650 return need_cap;
|
|
|
9651 }
|
|
|
9652
|
|
|
9653
|
|
|
9654 /*
|
|
372
|
9655 * ":spellrepall"
|
|
|
9656 */
|
|
|
9657 /*ARGSUSED*/
|
|
|
9658 void
|
|
|
9659 ex_spellrepall(eap)
|
|
|
9660 exarg_T *eap;
|
|
|
9661 {
|
|
|
9662 pos_T pos = curwin->w_cursor;
|
|
|
9663 char_u *frompat;
|
|
|
9664 int addlen;
|
|
|
9665 char_u *line;
|
|
|
9666 char_u *p;
|
|
|
9667 int save_ws = p_ws;
|
|
483
|
9668 linenr_T prev_lnum = 0;
|
|
372
|
9669
|
|
|
9670 if (repl_from == NULL || repl_to == NULL)
|
|
|
9671 {
|
|
|
9672 EMSG(_("E752: No previous spell replacement"));
|
|
|
9673 return;
|
|
|
9674 }
|
|
|
9675 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
9676
|
|
|
9677 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
9678 if (frompat == NULL)
|
|
|
9679 return;
|
|
|
9680 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
9681 p_ws = FALSE;
|
|
|
9682
|
|
483
|
9683 sub_nsubs = 0;
|
|
|
9684 sub_nlines = 0;
|
|
372
|
9685 curwin->w_cursor.lnum = 0;
|
|
|
9686 while (!got_int)
|
|
|
9687 {
|
|
|
9688 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
9689 || u_save_cursor() == FAIL)
|
|
|
9690 break;
|
|
|
9691
|
|
|
9692 /* Only replace when the right word isn't there yet. This happens
|
|
|
9693 * when changing "etc" to "etc.". */
|
|
|
9694 line = ml_get_curline();
|
|
|
9695 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
9696 repl_to, STRLEN(repl_to)) != 0)
|
|
|
9697 {
|
|
|
9698 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
9699 if (p == NULL)
|
|
|
9700 break;
|
|
|
9701 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
9702 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
9703 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
9704 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
9705 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
483
|
9706
|
|
|
9707 if (curwin->w_cursor.lnum != prev_lnum)
|
|
|
9708 {
|
|
|
9709 ++sub_nlines;
|
|
|
9710 prev_lnum = curwin->w_cursor.lnum;
|
|
|
9711 }
|
|
|
9712 ++sub_nsubs;
|
|
372
|
9713 }
|
|
|
9714 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
9715 }
|
|
|
9716
|
|
|
9717 p_ws = save_ws;
|
|
|
9718 curwin->w_cursor = pos;
|
|
|
9719 vim_free(frompat);
|
|
|
9720
|
|
483
|
9721 if (sub_nsubs == 0)
|
|
372
|
9722 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
483
|
9723 else
|
|
|
9724 do_sub_msg(FALSE);
|
|
372
|
9725 }
|
|
|
9726
|
|
|
9727 /*
|
|
344
|
9728 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
9729 * a list of allocated strings.
|
|
|
9730 */
|
|
|
9731 void
|
|
625
|
9732 spell_suggest_list(gap, word, maxcount, need_cap, interactive)
|
|
344
|
9733 garray_T *gap;
|
|
|
9734 char_u *word;
|
|
|
9735 int maxcount; /* maximum nr of suggestions */
|
|
475
|
9736 int need_cap; /* 'spellcapcheck' matched */
|
|
625
|
9737 int interactive;
|
|
344
|
9738 {
|
|
|
9739 suginfo_T sug;
|
|
|
9740 int i;
|
|
|
9741 suggest_T *stp;
|
|
|
9742 char_u *wcopy;
|
|
|
9743
|
|
638
|
9744 spell_find_suggest(word, 0, &sug, maxcount, FALSE, need_cap, interactive);
|
|
344
|
9745
|
|
|
9746 /* Make room in "gap". */
|
|
|
9747 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
626
|
9748 if (ga_grow(gap, sug.su_ga.ga_len) == OK)
|
|
|
9749 {
|
|
|
9750 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
9751 {
|
|
|
9752 stp = &SUG(sug.su_ga, i);
|
|
|
9753
|
|
|
9754 /* The suggested word may replace only part of "word", add the not
|
|
|
9755 * replaced part. */
|
|
|
9756 wcopy = alloc(stp->st_wordlen
|
|
344
|
9757 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
626
|
9758 if (wcopy == NULL)
|
|
|
9759 break;
|
|
|
9760 STRCPY(wcopy, stp->st_word);
|
|
|
9761 STRCPY(wcopy + stp->st_wordlen, sug.su_badptr + stp->st_orglen);
|
|
|
9762 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
9763 }
|
|
344
|
9764 }
|
|
|
9765
|
|
|
9766 spell_find_cleanup(&sug);
|
|
|
9767 }
|
|
|
9768
|
|
|
9769 /*
|
|
|
9770 * Find spell suggestions for the word at the start of "badptr".
|
|
|
9771 * Return the suggestions in "su->su_ga".
|
|
|
9772 * The maximum number of suggestions is "maxcount".
|
|
|
9773 * Note: does use info for the current window.
|
|
|
9774 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
9775 */
|
|
|
9776 static void
|
|
638
|
9777 spell_find_suggest(badptr, badlen, su, maxcount, banbadword, need_cap, interactive)
|
|
344
|
9778 char_u *badptr;
|
|
638
|
9779 int badlen; /* length of bad word or 0 if unknown */
|
|
344
|
9780 suginfo_T *su;
|
|
|
9781 int maxcount;
|
|
358
|
9782 int banbadword; /* don't include badword in suggestions */
|
|
388
|
9783 int need_cap; /* word should start with capital */
|
|
625
|
9784 int interactive;
|
|
344
|
9785 {
|
|
534
|
9786 hlf_T attr = HLF_COUNT;
|
|
372
|
9787 char_u buf[MAXPATHL];
|
|
|
9788 char_u *p;
|
|
|
9789 int do_combine = FALSE;
|
|
|
9790 char_u *sps_copy;
|
|
|
9791 #ifdef FEAT_EVAL
|
|
|
9792 static int expr_busy = FALSE;
|
|
|
9793 #endif
|
|
385
|
9794 int c;
|
|
503
|
9795 int i;
|
|
|
9796 langp_T *lp;
|
|
344
|
9797
|
|
|
9798 /*
|
|
|
9799 * Set the info in "*su".
|
|
|
9800 */
|
|
|
9801 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
9802 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
9803 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
9804 if (*badptr == NUL)
|
|
|
9805 return;
|
|
344
|
9806 hash_init(&su->su_banned);
|
|
|
9807
|
|
|
9808 su->su_badptr = badptr;
|
|
638
|
9809 if (badlen != 0)
|
|
|
9810 su->su_badlen = badlen;
|
|
|
9811 else
|
|
|
9812 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL, FALSE);
|
|
344
|
9813 su->su_maxcount = maxcount;
|
|
372
|
9814 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
9815
|
|
|
9816 if (su->su_badlen >= MAXWLEN)
|
|
|
9817 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
9818 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
9819 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
9820 su->su_fbadword, MAXWLEN);
|
|
346
|
9821 /* get caps flags for bad word */
|
|
474
|
9822 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
9823 su->su_badptr + su->su_badlen);
|
|
388
|
9824 if (need_cap)
|
|
|
9825 su->su_badflags |= WF_ONECAP;
|
|
344
|
9826
|
|
503
|
9827 /* Find the default language for sound folding. We simply use the first
|
|
|
9828 * one in 'spelllang' that supports sound folding. That's good for when
|
|
|
9829 * using multiple files for one language, it's not that bad when mixing
|
|
|
9830 * languages (e.g., "pl,en"). */
|
|
|
9831 for (i = 0; i < curbuf->b_langp.ga_len; ++i)
|
|
|
9832 {
|
|
|
9833 lp = LANGP_ENTRY(curbuf->b_langp, i);
|
|
|
9834 if (lp->lp_sallang != NULL)
|
|
|
9835 {
|
|
|
9836 su->su_sallang = lp->lp_sallang;
|
|
|
9837 break;
|
|
|
9838 }
|
|
|
9839 }
|
|
|
9840
|
|
534
|
9841 /* Soundfold the bad word with the default sound folding, so that we don't
|
|
|
9842 * have to do this many times. */
|
|
|
9843 if (su->su_sallang != NULL)
|
|
|
9844 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE,
|
|
|
9845 su->su_sal_badword);
|
|
|
9846
|
|
385
|
9847 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
9848 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
9849 * for that. */
|
|
455
|
9850 c = PTR2CHAR(su->su_badptr);
|
|
534
|
9851 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT)
|
|
385
|
9852 {
|
|
|
9853 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
9854 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
625
|
9855 0, TRUE, su->su_sallang, FALSE);
|
|
385
|
9856 }
|
|
|
9857
|
|
344
|
9858 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
9859 if (banbadword)
|
|
|
9860 add_banned(su, su->su_badword);
|
|
344
|
9861
|
|
372
|
9862 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
9863 sps_copy = vim_strsave(p_sps);
|
|
|
9864 if (sps_copy == NULL)
|
|
|
9865 return;
|
|
|
9866
|
|
|
9867 /* Loop over the items in 'spellsuggest'. */
|
|
|
9868 for (p = sps_copy; *p != NUL; )
|
|
|
9869 {
|
|
|
9870 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
9871
|
|
|
9872 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
9873 {
|
|
|
9874 #ifdef FEAT_EVAL
|
|
375
|
9875 /* Evaluate an expression. Skip this when called recursively,
|
|
|
9876 * when using spellsuggest() in the expression. */
|
|
372
|
9877 if (!expr_busy)
|
|
|
9878 {
|
|
|
9879 expr_busy = TRUE;
|
|
|
9880 spell_suggest_expr(su, buf + 5);
|
|
|
9881 expr_busy = FALSE;
|
|
|
9882 }
|
|
|
9883 #endif
|
|
|
9884 }
|
|
|
9885 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
9886 /* Use list of suggestions in a file. */
|
|
|
9887 spell_suggest_file(su, buf + 5);
|
|
|
9888 else
|
|
|
9889 {
|
|
|
9890 /* Use internal method. */
|
|
625
|
9891 spell_suggest_intern(su, interactive);
|
|
372
|
9892 if (sps_flags & SPS_DOUBLE)
|
|
|
9893 do_combine = TRUE;
|
|
|
9894 }
|
|
|
9895 }
|
|
|
9896
|
|
|
9897 vim_free(sps_copy);
|
|
|
9898
|
|
|
9899 if (do_combine)
|
|
|
9900 /* Combine the two list of suggestions. This must be done last,
|
|
|
9901 * because sorting changes the order again. */
|
|
|
9902 score_combine(su);
|
|
|
9903 }
|
|
|
9904
|
|
|
9905 #ifdef FEAT_EVAL
|
|
|
9906 /*
|
|
|
9907 * Find suggestions by evaluating expression "expr".
|
|
|
9908 */
|
|
|
9909 static void
|
|
|
9910 spell_suggest_expr(su, expr)
|
|
|
9911 suginfo_T *su;
|
|
|
9912 char_u *expr;
|
|
|
9913 {
|
|
|
9914 list_T *list;
|
|
|
9915 listitem_T *li;
|
|
|
9916 int score;
|
|
|
9917 char_u *p;
|
|
|
9918
|
|
|
9919 /* The work is split up in a few parts to avoid having to export
|
|
|
9920 * suginfo_T.
|
|
|
9921 * First evaluate the expression and get the resulting list. */
|
|
|
9922 list = eval_spell_expr(su->su_badword, expr);
|
|
|
9923 if (list != NULL)
|
|
|
9924 {
|
|
|
9925 /* Loop over the items in the list. */
|
|
|
9926 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
9927 if (li->li_tv.v_type == VAR_LIST)
|
|
|
9928 {
|
|
|
9929 /* Get the word and the score from the items. */
|
|
|
9930 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
625
|
9931 if (score >= 0 && score <= su->su_maxscore)
|
|
|
9932 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
9933 score, 0, TRUE, su->su_sallang, FALSE);
|
|
372
|
9934 }
|
|
|
9935 list_unref(list);
|
|
|
9936 }
|
|
|
9937
|
|
625
|
9938 /* Remove bogus suggestions, sort and truncate at "maxcount". */
|
|
|
9939 check_suggestions(su, &su->su_ga);
|
|
372
|
9940 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
9941 }
|
|
|
9942 #endif
|
|
|
9943
|
|
|
9944 /*
|
|
474
|
9945 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
9946 */
|
|
|
9947 static void
|
|
|
9948 spell_suggest_file(su, fname)
|
|
|
9949 suginfo_T *su;
|
|
|
9950 char_u *fname;
|
|
|
9951 {
|
|
|
9952 FILE *fd;
|
|
|
9953 char_u line[MAXWLEN * 2];
|
|
|
9954 char_u *p;
|
|
|
9955 int len;
|
|
|
9956 char_u cword[MAXWLEN];
|
|
|
9957
|
|
|
9958 /* Open the file. */
|
|
|
9959 fd = mch_fopen((char *)fname, "r");
|
|
|
9960 if (fd == NULL)
|
|
|
9961 {
|
|
|
9962 EMSG2(_(e_notopen), fname);
|
|
|
9963 return;
|
|
|
9964 }
|
|
|
9965
|
|
|
9966 /* Read it line by line. */
|
|
|
9967 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
9968 {
|
|
|
9969 line_breakcheck();
|
|
|
9970
|
|
|
9971 p = vim_strchr(line, '/');
|
|
|
9972 if (p == NULL)
|
|
|
9973 continue; /* No Tab found, just skip the line. */
|
|
|
9974 *p++ = NUL;
|
|
|
9975 if (STRICMP(su->su_badword, line) == 0)
|
|
|
9976 {
|
|
|
9977 /* Match! Isolate the good word, until CR or NL. */
|
|
|
9978 for (len = 0; p[len] >= ' '; ++len)
|
|
|
9979 ;
|
|
|
9980 p[len] = NUL;
|
|
|
9981
|
|
|
9982 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
9983 * of the bad word. */
|
|
|
9984 if (captype(p, NULL) == 0)
|
|
|
9985 {
|
|
|
9986 make_case_word(p, cword, su->su_badflags);
|
|
|
9987 p = cword;
|
|
|
9988 }
|
|
|
9989
|
|
|
9990 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
625
|
9991 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE);
|
|
372
|
9992 }
|
|
|
9993 }
|
|
|
9994
|
|
|
9995 fclose(fd);
|
|
|
9996
|
|
625
|
9997 /* Remove bogus suggestions, sort and truncate at "maxcount". */
|
|
|
9998 check_suggestions(su, &su->su_ga);
|
|
372
|
9999 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
10000 }
|
|
|
10001
|
|
|
10002 /*
|
|
|
10003 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
10004 */
|
|
|
10005 static void
|
|
625
|
10006 spell_suggest_intern(su, interactive)
|
|
372
|
10007 suginfo_T *su;
|
|
625
|
10008 int interactive;
|
|
|
10009 {
|
|
|
10010 /*
|
|
|
10011 * Load the .sug file(s) that are available and not done yet.
|
|
|
10012 */
|
|
|
10013 suggest_load_files();
|
|
|
10014
|
|
344
|
10015 /*
|
|
346
|
10016 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
10017 *
|
|
|
10018 * Set a maximum score to limit the combination of operations that is
|
|
|
10019 * tried.
|
|
|
10020 */
|
|
346
|
10021 suggest_try_special(su);
|
|
|
10022
|
|
|
10023 /*
|
|
|
10024 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
10025 * from the .aff file and inserting a space (split the word).
|
|
|
10026 */
|
|
|
10027 suggest_try_change(su);
|
|
344
|
10028
|
|
|
10029 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
10030 if (sps_flags & SPS_DOUBLE)
|
|
|
10031 score_comp_sal(su);
|
|
|
10032
|
|
|
10033 /*
|
|
346
|
10034 * 3. Try finding sound-a-like words.
|
|
344
|
10035 */
|
|
625
|
10036 if ((sps_flags & SPS_FAST) == 0)
|
|
|
10037 {
|
|
|
10038 if (sps_flags & SPS_BEST)
|
|
|
10039 /* Adjust the word score for the suggestions found so far for how
|
|
|
10040 * they sounds like. */
|
|
|
10041 rescore_suggestions(su);
|
|
|
10042
|
|
|
10043 /*
|
|
|
10044 * While going throught the soundfold tree "su_maxscore" is the score
|
|
|
10045 * for the soundfold word, limits the changes that are being tried,
|
|
|
10046 * and "su_sfmaxscore" the rescored score, which is set by
|
|
|
10047 * cleanup_suggestions().
|
|
|
10048 * First find words with a small edit distance, because this is much
|
|
|
10049 * faster and often already finds the top-N suggestions. If we didn't
|
|
|
10050 * find many suggestions try again with a higher edit distance.
|
|
|
10051 * "sl_sounddone" is used to avoid doing the same word twice.
|
|
|
10052 */
|
|
|
10053 suggest_try_soundalike_prep();
|
|
|
10054 su->su_maxscore = SCORE_SFMAX1;
|
|
|
10055 su->su_sfmaxscore = SCORE_MAXINIT * 3;
|
|
346
|
10056 suggest_try_soundalike(su);
|
|
625
|
10057 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su))
|
|
|
10058 {
|
|
|
10059 /* We didn't find enough matches, try again, allowing more
|
|
|
10060 * changes to the soundfold word. */
|
|
|
10061 su->su_maxscore = SCORE_SFMAX2;
|
|
|
10062 suggest_try_soundalike(su);
|
|
|
10063 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su))
|
|
|
10064 {
|
|
|
10065 /* Still didn't find enough matches, try again, allowing even
|
|
|
10066 * more changes to the soundfold word. */
|
|
|
10067 su->su_maxscore = SCORE_SFMAX3;
|
|
|
10068 suggest_try_soundalike(su);
|
|
|
10069 }
|
|
|
10070 }
|
|
|
10071 su->su_maxscore = su->su_sfmaxscore;
|
|
|
10072 suggest_try_soundalike_finish();
|
|
|
10073 }
|
|
|
10074
|
|
|
10075 /* When CTRL-C was hit while searching do show the results. Only clear
|
|
|
10076 * got_int when using a command, not for spellsuggest(). */
|
|
344
|
10077 ui_breakcheck();
|
|
625
|
10078 if (interactive && got_int)
|
|
344
|
10079 {
|
|
|
10080 (void)vgetc();
|
|
|
10081 got_int = FALSE;
|
|
|
10082 }
|
|
|
10083
|
|
372
|
10084 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
10085 {
|
|
|
10086 if (sps_flags & SPS_BEST)
|
|
|
10087 /* Adjust the word score for how it sounds like. */
|
|
|
10088 rescore_suggestions(su);
|
|
|
10089
|
|
625
|
10090 /* Remove bogus suggestions, sort and truncate at "maxcount". */
|
|
|
10091 check_suggestions(su, &su->su_ga);
|
|
372
|
10092 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
10093 }
|
|
|
10094 }
|
|
|
10095
|
|
|
10096 /*
|
|
625
|
10097 * Load the .sug files for languages that have one and weren't loaded yet.
|
|
|
10098 */
|
|
|
10099 static void
|
|
|
10100 suggest_load_files()
|
|
|
10101 {
|
|
|
10102 langp_T *lp;
|
|
|
10103 int lpi;
|
|
|
10104 slang_T *slang;
|
|
|
10105 char_u *dotp;
|
|
|
10106 FILE *fd;
|
|
|
10107 char_u buf[MAXWLEN];
|
|
|
10108 int i;
|
|
|
10109 time_t timestamp;
|
|
|
10110 int wcount;
|
|
|
10111 int wordnr;
|
|
|
10112 garray_T ga;
|
|
|
10113 int c;
|
|
|
10114
|
|
|
10115 /* Do this for all languages that support sound folding. */
|
|
|
10116 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
10117 {
|
|
|
10118 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
10119 slang = lp->lp_slang;
|
|
|
10120 if (slang->sl_sugtime != 0 && !slang->sl_sugloaded)
|
|
|
10121 {
|
|
|
10122 /* Change ".spl" to ".sug" and open the file. When the file isn't
|
|
|
10123 * found silently skip it. Do set "sl_sugloaded" so that we
|
|
|
10124 * don't try again and again. */
|
|
|
10125 slang->sl_sugloaded = TRUE;
|
|
|
10126
|
|
|
10127 dotp = vim_strrchr(slang->sl_fname, '.');
|
|
|
10128 if (dotp == NULL || fnamecmp(dotp, ".spl") != 0)
|
|
|
10129 continue;
|
|
|
10130 STRCPY(dotp, ".sug");
|
|
|
10131 fd = fopen((char *)slang->sl_fname, "r");
|
|
|
10132 if (fd == NULL)
|
|
|
10133 goto nextone;
|
|
|
10134
|
|
|
10135 /*
|
|
|
10136 * <SUGHEADER>: <fileID> <versionnr> <timestamp>
|
|
|
10137 */
|
|
|
10138 for (i = 0; i < VIMSUGMAGICL; ++i)
|
|
|
10139 buf[i] = getc(fd); /* <fileID> */
|
|
|
10140 if (STRNCMP(buf, VIMSUGMAGIC, VIMSUGMAGICL) != 0)
|
|
|
10141 {
|
|
626
|
10142 EMSG2(_("E778: This does not look like a .sug file: %s"),
|
|
625
|
10143 slang->sl_fname);
|
|
|
10144 goto nextone;
|
|
|
10145 }
|
|
|
10146 c = getc(fd); /* <versionnr> */
|
|
|
10147 if (c < VIMSUGVERSION)
|
|
|
10148 {
|
|
626
|
10149 EMSG2(_("E779: Old .sug file, needs to be updated: %s"),
|
|
625
|
10150 slang->sl_fname);
|
|
|
10151 goto nextone;
|
|
|
10152 }
|
|
|
10153 else if (c > VIMSUGVERSION)
|
|
|
10154 {
|
|
626
|
10155 EMSG2(_("E780: .sug file is for newer version of Vim: %s"),
|
|
625
|
10156 slang->sl_fname);
|
|
|
10157 goto nextone;
|
|
|
10158 }
|
|
|
10159
|
|
|
10160 /* Check the timestamp, it must be exactly the same as the one in
|
|
|
10161 * the .spl file. Otherwise the word numbers won't match. */
|
|
|
10162 timestamp = 0;
|
|
|
10163 for (i = 7; i >= 0; --i) /* <timestamp> */
|
|
|
10164 timestamp += getc(fd) << (i * 8);
|
|
|
10165 if (timestamp != slang->sl_sugtime)
|
|
|
10166 {
|
|
626
|
10167 EMSG2(_("E781: .sug file doesn't match .spl file: %s"),
|
|
625
|
10168 slang->sl_fname);
|
|
|
10169 goto nextone;
|
|
|
10170 }
|
|
|
10171
|
|
|
10172 /*
|
|
|
10173 * <SUGWORDTREE>: <wordtree>
|
|
|
10174 * Read the trie with the soundfolded words.
|
|
|
10175 */
|
|
|
10176 if (spell_read_tree(fd, &slang->sl_sbyts, &slang->sl_sidxs,
|
|
|
10177 FALSE, 0) != 0)
|
|
|
10178 {
|
|
|
10179 someerror:
|
|
626
|
10180 EMSG2(_("E782: error while reading .sug file: %s"),
|
|
625
|
10181 slang->sl_fname);
|
|
|
10182 slang_clear_sug(slang);
|
|
|
10183 goto nextone;
|
|
|
10184 }
|
|
|
10185
|
|
|
10186 /*
|
|
|
10187 * <SUGTABLE>: <sugwcount> <sugline> ...
|
|
|
10188 *
|
|
|
10189 * Read the table with word numbers. We use a file buffer for
|
|
|
10190 * this, because it's so much like a file with lines. Makes it
|
|
|
10191 * possible to swap the info and save on memory use.
|
|
|
10192 */
|
|
|
10193 slang->sl_sugbuf = open_spellbuf();
|
|
|
10194 if (slang->sl_sugbuf == NULL)
|
|
|
10195 goto someerror;
|
|
|
10196 /* <sugwcount> */
|
|
|
10197 wcount = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8)
|
|
|
10198 + getc(fd);
|
|
|
10199 if (wcount < 0)
|
|
|
10200 goto someerror;
|
|
|
10201
|
|
|
10202 /* Read all the wordnr lists into the buffer, one NUL terminated
|
|
|
10203 * list per line. */
|
|
|
10204 ga_init2(&ga, 1, 100);
|
|
|
10205 for (wordnr = 0; wordnr < wcount; ++wordnr)
|
|
|
10206 {
|
|
|
10207 ga.ga_len = 0;
|
|
|
10208 for (;;)
|
|
|
10209 {
|
|
|
10210 c = getc(fd); /* <sugline> */
|
|
|
10211 if (c < 0 || ga_grow(&ga, 1) == FAIL)
|
|
|
10212 goto someerror;
|
|
|
10213 ((char_u *)ga.ga_data)[ga.ga_len++] = c;
|
|
|
10214 if (c == NUL)
|
|
|
10215 break;
|
|
|
10216 }
|
|
|
10217 if (ml_append_buf(slang->sl_sugbuf, (linenr_T)wordnr,
|
|
|
10218 ga.ga_data, ga.ga_len, TRUE) == FAIL)
|
|
|
10219 goto someerror;
|
|
|
10220 }
|
|
|
10221 ga_clear(&ga);
|
|
|
10222
|
|
|
10223 /*
|
|
|
10224 * Need to put word counts in the word tries, so that we can find
|
|
|
10225 * a word by its number.
|
|
|
10226 */
|
|
|
10227 tree_count_words(slang->sl_fbyts, slang->sl_fidxs);
|
|
|
10228 tree_count_words(slang->sl_sbyts, slang->sl_sidxs);
|
|
|
10229
|
|
|
10230 nextone:
|
|
|
10231 if (fd != NULL)
|
|
|
10232 fclose(fd);
|
|
|
10233 STRCPY(dotp, ".spl");
|
|
|
10234 }
|
|
|
10235 }
|
|
|
10236 }
|
|
|
10237
|
|
|
10238
|
|
|
10239 /*
|
|
|
10240 * Fill in the wordcount fields for a trie.
|
|
|
10241 * Returns the total number of words.
|
|
|
10242 */
|
|
|
10243 static void
|
|
|
10244 tree_count_words(byts, idxs)
|
|
|
10245 char_u *byts;
|
|
|
10246 idx_T *idxs;
|
|
|
10247 {
|
|
|
10248 int depth;
|
|
|
10249 idx_T arridx[MAXWLEN];
|
|
|
10250 int curi[MAXWLEN];
|
|
|
10251 int c;
|
|
|
10252 idx_T n;
|
|
|
10253 int wordcount[MAXWLEN];
|
|
|
10254
|
|
|
10255 arridx[0] = 0;
|
|
|
10256 curi[0] = 1;
|
|
|
10257 wordcount[0] = 0;
|
|
|
10258 depth = 0;
|
|
|
10259 while (depth >= 0 && !got_int)
|
|
|
10260 {
|
|
|
10261 if (curi[depth] > byts[arridx[depth]])
|
|
|
10262 {
|
|
|
10263 /* Done all bytes at this node, go up one level. */
|
|
|
10264 idxs[arridx[depth]] = wordcount[depth];
|
|
|
10265 if (depth > 0)
|
|
|
10266 wordcount[depth - 1] += wordcount[depth];
|
|
|
10267
|
|
|
10268 --depth;
|
|
|
10269 fast_breakcheck();
|
|
|
10270 }
|
|
|
10271 else
|
|
|
10272 {
|
|
|
10273 /* Do one more byte at this node. */
|
|
|
10274 n = arridx[depth] + curi[depth];
|
|
|
10275 ++curi[depth];
|
|
|
10276
|
|
|
10277 c = byts[n];
|
|
|
10278 if (c == 0)
|
|
|
10279 {
|
|
|
10280 /* End of word, count it. */
|
|
|
10281 ++wordcount[depth];
|
|
|
10282
|
|
|
10283 /* Skip over any other NUL bytes (same word with different
|
|
|
10284 * flags). */
|
|
|
10285 while (byts[n + 1] == 0)
|
|
|
10286 {
|
|
|
10287 ++n;
|
|
|
10288 ++curi[depth];
|
|
|
10289 }
|
|
|
10290 }
|
|
|
10291 else
|
|
|
10292 {
|
|
|
10293 /* Normal char, go one level deeper to count the words. */
|
|
|
10294 ++depth;
|
|
|
10295 arridx[depth] = idxs[n];
|
|
|
10296 curi[depth] = 1;
|
|
|
10297 wordcount[depth] = 0;
|
|
|
10298 }
|
|
|
10299 }
|
|
|
10300 }
|
|
|
10301 }
|
|
|
10302
|
|
|
10303 /*
|
|
344
|
10304 * Free the info put in "*su" by spell_find_suggest().
|
|
|
10305 */
|
|
|
10306 static void
|
|
|
10307 spell_find_cleanup(su)
|
|
|
10308 suginfo_T *su;
|
|
|
10309 {
|
|
|
10310 int i;
|
|
|
10311
|
|
|
10312 /* Free the suggestions. */
|
|
|
10313 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10314 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
10315 ga_clear(&su->su_ga);
|
|
|
10316 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
10317 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
10318 ga_clear(&su->su_sga);
|
|
323
|
10319
|
|
|
10320 /* Free the banned words. */
|
|
625
|
10321 hash_clear_all(&su->su_banned, 0);
|
|
323
|
10322 }
|
|
|
10323
|
|
|
10324 /*
|
|
324
|
10325 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
10326 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
10327 * The result is NUL terminated.
|
|
323
|
10328 */
|
|
|
10329 static void
|
|
324
|
10330 onecap_copy(word, wcopy, upper)
|
|
323
|
10331 char_u *word;
|
|
|
10332 char_u *wcopy;
|
|
|
10333 int upper; /* TRUE: first letter made upper case */
|
|
|
10334 {
|
|
|
10335 char_u *p;
|
|
|
10336 int c;
|
|
|
10337 int l;
|
|
|
10338
|
|
|
10339 p = word;
|
|
|
10340 #ifdef FEAT_MBYTE
|
|
|
10341 if (has_mbyte)
|
|
474
|
10342 c = mb_cptr2char_adv(&p);
|
|
323
|
10343 else
|
|
|
10344 #endif
|
|
|
10345 c = *p++;
|
|
|
10346 if (upper)
|
|
324
|
10347 c = SPELL_TOUPPER(c);
|
|
323
|
10348 else
|
|
324
|
10349 c = SPELL_TOFOLD(c);
|
|
323
|
10350 #ifdef FEAT_MBYTE
|
|
|
10351 if (has_mbyte)
|
|
|
10352 l = mb_char2bytes(c, wcopy);
|
|
|
10353 else
|
|
|
10354 #endif
|
|
|
10355 {
|
|
|
10356 l = 1;
|
|
|
10357 wcopy[0] = c;
|
|
|
10358 }
|
|
376
|
10359 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
10360 }
|
|
|
10361
|
|
|
10362 /*
|
|
324
|
10363 * Make a copy of "word" with all the letters upper cased into
|
|
|
10364 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
10365 */
|
|
|
10366 static void
|
|
|
10367 allcap_copy(word, wcopy)
|
|
|
10368 char_u *word;
|
|
|
10369 char_u *wcopy;
|
|
|
10370 {
|
|
|
10371 char_u *s;
|
|
|
10372 char_u *d;
|
|
|
10373 int c;
|
|
|
10374
|
|
|
10375 d = wcopy;
|
|
|
10376 for (s = word; *s != NUL; )
|
|
|
10377 {
|
|
|
10378 #ifdef FEAT_MBYTE
|
|
|
10379 if (has_mbyte)
|
|
474
|
10380 c = mb_cptr2char_adv(&s);
|
|
323
|
10381 else
|
|
|
10382 #endif
|
|
|
10383 c = *s++;
|
|
492
|
10384
|
|
|
10385 #ifdef FEAT_MBYTE
|
|
|
10386 /* We only change ß to SS when we are certain latin1 is used. It
|
|
|
10387 * would cause weird errors in other 8-bit encodings. */
|
|
|
10388 if (enc_latin1like && c == 0xdf)
|
|
|
10389 {
|
|
|
10390 c = 'S';
|
|
|
10391 if (d - wcopy >= MAXWLEN - 1)
|
|
|
10392 break;
|
|
|
10393 *d++ = c;
|
|
|
10394 }
|
|
|
10395 else
|
|
|
10396 #endif
|
|
|
10397 c = SPELL_TOUPPER(c);
|
|
323
|
10398
|
|
|
10399 #ifdef FEAT_MBYTE
|
|
|
10400 if (has_mbyte)
|
|
|
10401 {
|
|
|
10402 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
10403 break;
|
|
|
10404 d += mb_char2bytes(c, d);
|
|
|
10405 }
|
|
|
10406 else
|
|
|
10407 #endif
|
|
|
10408 {
|
|
|
10409 if (d - wcopy >= MAXWLEN - 1)
|
|
|
10410 break;
|
|
|
10411 *d++ = c;
|
|
|
10412 }
|
|
|
10413 }
|
|
|
10414 *d = NUL;
|
|
|
10415 }
|
|
|
10416
|
|
|
10417 /*
|
|
346
|
10418 * Try finding suggestions by recognizing specific situations.
|
|
|
10419 */
|
|
|
10420 static void
|
|
|
10421 suggest_try_special(su)
|
|
|
10422 suginfo_T *su;
|
|
|
10423 {
|
|
|
10424 char_u *p;
|
|
376
|
10425 size_t len;
|
|
346
|
10426 int c;
|
|
|
10427 char_u word[MAXWLEN];
|
|
|
10428
|
|
|
10429 /*
|
|
|
10430 * Recognize a word that is repeated: "the the".
|
|
|
10431 */
|
|
|
10432 p = skiptowhite(su->su_fbadword);
|
|
|
10433 len = p - su->su_fbadword;
|
|
|
10434 p = skipwhite(p);
|
|
|
10435 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
10436 {
|
|
|
10437 /* Include badflags: if the badword is onecap or allcap
|
|
|
10438 * use that for the goodword too: "The the" -> "The". */
|
|
|
10439 c = su->su_fbadword[len];
|
|
|
10440 su->su_fbadword[len] = NUL;
|
|
|
10441 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
10442 su->su_fbadword[len] = c;
|
|
534
|
10443
|
|
|
10444 /* Give a soundalike score of 0, compute the score as if deleting one
|
|
|
10445 * character. */
|
|
|
10446 add_suggestion(su, &su->su_ga, word, su->su_badlen,
|
|
625
|
10447 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE);
|
|
346
|
10448 }
|
|
|
10449 }
|
|
|
10450
|
|
|
10451 /*
|
|
323
|
10452 * Try finding suggestions by adding/removing/swapping letters.
|
|
|
10453 */
|
|
|
10454 static void
|
|
346
|
10455 suggest_try_change(su)
|
|
323
|
10456 suginfo_T *su;
|
|
|
10457 {
|
|
|
10458 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
625
|
10459 int n;
|
|
|
10460 char_u *p;
|
|
|
10461 int lpi;
|
|
323
|
10462 langp_T *lp;
|
|
|
10463
|
|
|
10464 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
10465 * to find matches (esp. REP items). Append some more text, changing
|
|
|
10466 * chars after the bad word may help. */
|
|
323
|
10467 STRCPY(fword, su->su_fbadword);
|
|
346
|
10468 n = STRLEN(fword);
|
|
|
10469 p = su->su_badptr + su->su_badlen;
|
|
|
10470 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
10471
|
|
503
|
10472 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
10473 {
|
|
|
10474 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
482
|
10475
|
|
500
|
10476 /* If reloading a spell file fails it's still in the list but
|
|
|
10477 * everything has been cleared. */
|
|
625
|
10478 if (lp->lp_slang->sl_fbyts == NULL)
|
|
500
|
10479 continue;
|
|
|
10480
|
|
625
|
10481 /* Try it for this language. Will add possible suggestions. */
|
|
|
10482 suggest_trie_walk(su, lp, fword, FALSE);
|
|
|
10483 }
|
|
|
10484 }
|
|
|
10485
|
|
|
10486 /* Check the maximum score, if we go over it we won't try this change. */
|
|
|
10487 #define TRY_DEEPER(su, stack, depth, add) \
|
|
|
10488 (stack[depth].ts_score + (add) < su->su_maxscore)
|
|
|
10489
|
|
|
10490 /*
|
|
|
10491 * Try finding suggestions by adding/removing/swapping letters.
|
|
|
10492 *
|
|
|
10493 * This uses a state machine. At each node in the tree we try various
|
|
|
10494 * operations. When trying if an operation works "depth" is increased and the
|
|
|
10495 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
10496 * character, replace one and delete another. The number of changes is
|
|
|
10497 * limited by su->su_maxscore.
|
|
|
10498 *
|
|
|
10499 * After implementing this I noticed an article by Kemal Oflazer that
|
|
|
10500 * describes something similar: "Error-tolerant Finite State Recognition with
|
|
|
10501 * Applications to Morphological Analysis and Spelling Correction" (1996).
|
|
|
10502 * The implementation in the article is simplified and requires a stack of
|
|
|
10503 * unknown depth. The implementation here only needs a stack depth equal to
|
|
|
10504 * the length of the word.
|
|
|
10505 *
|
|
|
10506 * This is also used for the sound-folded word, "soundfold" is TRUE then.
|
|
|
10507 * The mechanism is the same, but we find a match with a sound-folded word
|
|
|
10508 * that comes from one or more original words. Each of these words may be
|
|
|
10509 * added, this is done by add_sound_suggest().
|
|
|
10510 * Don't use:
|
|
|
10511 * the prefix tree or the keep-case tree
|
|
|
10512 * "su->su_badlen"
|
|
|
10513 * anything to do with upper and lower case
|
|
|
10514 * anything to do with word or non-word characters ("spell_iswordp()")
|
|
|
10515 * banned words
|
|
|
10516 * word flags (rare, region, compounding)
|
|
|
10517 * word splitting for now
|
|
|
10518 * "similar_chars()"
|
|
|
10519 * use "slang->sl_repsal" instead of "lp->lp_replang->sl_rep"
|
|
|
10520 */
|
|
|
10521 static void
|
|
|
10522 suggest_trie_walk(su, lp, fword, soundfold)
|
|
|
10523 suginfo_T *su;
|
|
|
10524 langp_T *lp;
|
|
|
10525 char_u *fword;
|
|
|
10526 int soundfold;
|
|
|
10527 {
|
|
|
10528 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
10529 trystate_T stack[MAXWLEN];
|
|
|
10530 char_u preword[MAXWLEN * 3]; /* word found with proper case;
|
|
|
10531 * concatanation of prefix compound
|
|
|
10532 * words and split word. NUL terminated
|
|
|
10533 * when going deeper but not when coming
|
|
|
10534 * back. */
|
|
|
10535 char_u compflags[MAXWLEN]; /* compound flags, one for each word */
|
|
|
10536 trystate_T *sp;
|
|
|
10537 int newscore;
|
|
|
10538 int score;
|
|
|
10539 char_u *byts, *fbyts, *pbyts;
|
|
|
10540 idx_T *idxs, *fidxs, *pidxs;
|
|
|
10541 int depth;
|
|
|
10542 int c, c2, c3;
|
|
|
10543 int n = 0;
|
|
|
10544 int flags;
|
|
|
10545 garray_T *gap;
|
|
|
10546 idx_T arridx;
|
|
|
10547 int len;
|
|
|
10548 char_u *p;
|
|
|
10549 fromto_T *ftp;
|
|
|
10550 int fl = 0, tl;
|
|
|
10551 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
|
10552 slang_T *slang = lp->lp_slang;
|
|
|
10553 int fword_ends;
|
|
|
10554 int goodword_ends;
|
|
|
10555 #ifdef DEBUG_TRIEWALK
|
|
|
10556 /* Stores the name of the change made at each level. */
|
|
|
10557 char_u changename[MAXWLEN][80];
|
|
|
10558 #endif
|
|
|
10559 int breakcheckcount = 1000;
|
|
|
10560 int compound_ok;
|
|
|
10561
|
|
|
10562 /*
|
|
|
10563 * Go through the whole case-fold tree, try changes at each node.
|
|
|
10564 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
10565 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
10566 * word).
|
|
|
10567 */
|
|
|
10568 depth = 0;
|
|
|
10569 sp = &stack[0];
|
|
|
10570 vim_memset(sp, 0, sizeof(trystate_T));
|
|
|
10571 sp->ts_curi = 1;
|
|
|
10572
|
|
|
10573 if (soundfold)
|
|
|
10574 {
|
|
|
10575 /* Going through the soundfold tree. */
|
|
|
10576 byts = fbyts = slang->sl_sbyts;
|
|
|
10577 idxs = fidxs = slang->sl_sidxs;
|
|
|
10578 pbyts = NULL;
|
|
|
10579 pidxs = NULL;
|
|
|
10580 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
|
10581 sp->ts_state = STATE_START;
|
|
|
10582 }
|
|
|
10583 else
|
|
|
10584 {
|
|
330
|
10585 /*
|
|
375
|
10586 * When there are postponed prefixes we need to use these first. At
|
|
|
10587 * the end of the prefix we continue in the case-fold tree.
|
|
|
10588 */
|
|
482
|
10589 fbyts = slang->sl_fbyts;
|
|
|
10590 fidxs = slang->sl_fidxs;
|
|
|
10591 pbyts = slang->sl_pbyts;
|
|
|
10592 pidxs = slang->sl_pidxs;
|
|
375
|
10593 if (pbyts != NULL)
|
|
|
10594 {
|
|
|
10595 byts = pbyts;
|
|
|
10596 idxs = pidxs;
|
|
482
|
10597 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
375
|
10598 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
10599 }
|
|
|
10600 else
|
|
|
10601 {
|
|
|
10602 byts = fbyts;
|
|
|
10603 idxs = fidxs;
|
|
482
|
10604 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
485
|
10605 sp->ts_state = STATE_START;
|
|
375
|
10606 }
|
|
625
|
10607 }
|
|
|
10608
|
|
|
10609 /*
|
|
|
10610 * Loop to find all suggestions. At each round we either:
|
|
|
10611 * - For the current state try one operation, advance "ts_curi",
|
|
|
10612 * increase "depth".
|
|
|
10613 * - When a state is done go to the next, set "ts_state".
|
|
|
10614 * - When all states are tried decrease "depth".
|
|
|
10615 */
|
|
|
10616 while (depth >= 0 && !got_int)
|
|
|
10617 {
|
|
|
10618 sp = &stack[depth];
|
|
|
10619 switch (sp->ts_state)
|
|
|
10620 {
|
|
|
10621 case STATE_START:
|
|
|
10622 case STATE_NOPREFIX:
|
|
|
10623 /*
|
|
|
10624 * Start of node: Deal with NUL bytes, which means
|
|
|
10625 * tword[] may end here.
|
|
|
10626 */
|
|
|
10627 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
10628 len = byts[arridx]; /* bytes in this node */
|
|
|
10629 arridx += sp->ts_curi; /* index of current byte */
|
|
|
10630
|
|
|
10631 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
|
|
|
10632 {
|
|
|
10633 /* Skip over the NUL bytes, we use them later. */
|
|
|
10634 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
10635 ;
|
|
|
10636 sp->ts_curi += n;
|
|
|
10637
|
|
|
10638 /* Always past NUL bytes now. */
|
|
|
10639 n = (int)sp->ts_state;
|
|
|
10640 sp->ts_state = STATE_ENDNUL;
|
|
|
10641 sp->ts_save_badflags = su->su_badflags;
|
|
|
10642
|
|
|
10643 /* At end of a prefix or at start of prefixtree: check for
|
|
|
10644 * following word. */
|
|
|
10645 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
|
10646 {
|
|
|
10647 /* Set su->su_badflags to the caps type at this position.
|
|
|
10648 * Use the caps type until here for the prefix itself. */
|
|
|
10649 #ifdef FEAT_MBYTE
|
|
|
10650 if (has_mbyte)
|
|
|
10651 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
10652 else
|
|
|
10653 #endif
|
|
|
10654 n = sp->ts_fidx;
|
|
|
10655 flags = badword_captype(su->su_badptr, su->su_badptr + n);
|
|
|
10656 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
10657 su->su_badptr + su->su_badlen);
|
|
625
|
10658 #ifdef DEBUG_TRIEWALK
|
|
|
10659 sprintf(changename[depth], "prefix");
|
|
|
10660 #endif
|
|
|
10661 go_deeper(stack, depth, 0);
|
|
|
10662 ++depth;
|
|
|
10663 sp = &stack[depth];
|
|
|
10664 sp->ts_prefixdepth = depth - 1;
|
|
|
10665 byts = fbyts;
|
|
|
10666 idxs = fidxs;
|
|
|
10667 sp->ts_arridx = 0;
|
|
|
10668
|
|
|
10669 /* Move the prefix to preword[] with the right case
|
|
|
10670 * and make find_keepcap_word() works. */
|
|
|
10671 tword[sp->ts_twordlen] = NUL;
|
|
|
10672 make_case_word(tword + sp->ts_splitoff,
|
|
|
10673 preword + sp->ts_prewordlen, flags);
|
|
|
10674 sp->ts_prewordlen = STRLEN(preword);
|
|
|
10675 sp->ts_splitoff = sp->ts_twordlen;
|
|
|
10676 }
|
|
|
10677 break;
|
|
|
10678 }
|
|
|
10679
|
|
|
10680 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
|
10681 {
|
|
|
10682 /* Past bytes in node and/or past NUL bytes. */
|
|
|
10683 sp->ts_state = STATE_ENDNUL;
|
|
|
10684 sp->ts_save_badflags = su->su_badflags;
|
|
|
10685 break;
|
|
|
10686 }
|
|
|
10687
|
|
|
10688 /*
|
|
|
10689 * End of word in tree.
|
|
|
10690 */
|
|
|
10691 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
10692
|
|
|
10693 flags = (int)idxs[arridx];
|
|
|
10694 fword_ends = (fword[sp->ts_fidx] == NUL
|
|
|
10695 || (soundfold
|
|
|
10696 ? vim_iswhite(fword[sp->ts_fidx])
|
|
|
10697 : !spell_iswordp(fword + sp->ts_fidx, curbuf)));
|
|
|
10698 tword[sp->ts_twordlen] = NUL;
|
|
|
10699
|
|
|
10700 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL
|
|
485
|
10701 && (sp->ts_flags & TSF_PREFIXOK) == 0)
|
|
625
|
10702 {
|
|
|
10703 /* There was a prefix before the word. Check that the prefix
|
|
|
10704 * can be used with this word. */
|
|
|
10705 /* Count the length of the NULs in the prefix. If there are
|
|
|
10706 * none this must be the first try without a prefix. */
|
|
|
10707 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
10708 len = pbyts[n++];
|
|
|
10709 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
10710 ;
|
|
|
10711 if (c > 0)
|
|
|
10712 {
|
|
|
10713 c = valid_word_prefix(c, n, flags,
|
|
483
|
10714 tword + sp->ts_splitoff, slang, FALSE);
|
|
625
|
10715 if (c == 0)
|
|
|
10716 break;
|
|
|
10717
|
|
|
10718 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
10719 if (c & WF_RAREPFX)
|
|
|
10720 flags |= WF_RARE;
|
|
|
10721
|
|
|
10722 /* Tricky: when checking for both prefix and compounding
|
|
|
10723 * we run into the prefix flag first.
|
|
|
10724 * Remember that it's OK, so that we accept the prefix
|
|
|
10725 * when arriving at a compound flag. */
|
|
|
10726 sp->ts_flags |= TSF_PREFIXOK;
|
|
|
10727 }
|
|
|
10728 }
|
|
|
10729
|
|
|
10730 /* Check NEEDCOMPOUND: can't use word without compounding. Do try
|
|
|
10731 * appending another compound word below. */
|
|
|
10732 if (sp->ts_complen == sp->ts_compsplit && fword_ends
|
|
500
|
10733 && (flags & WF_NEEDCOMP))
|
|
625
|
10734 goodword_ends = FALSE;
|
|
|
10735 else
|
|
|
10736 goodword_ends = TRUE;
|
|
|
10737
|
|
|
10738 p = NULL;
|
|
|
10739 compound_ok = TRUE;
|
|
|
10740 if (sp->ts_complen > sp->ts_compsplit)
|
|
|
10741 {
|
|
|
10742 if (slang->sl_nobreak)
|
|
|
10743 {
|
|
|
10744 /* There was a word before this word. When there was no
|
|
|
10745 * change in this word (it was correct) add the first word
|
|
|
10746 * as a suggestion. If this word was corrected too, we
|
|
|
10747 * need to check if a correct word follows. */
|
|
|
10748 if (sp->ts_fidx - sp->ts_splitfidx
|
|
|
10749 == sp->ts_twordlen - sp->ts_splitoff
|
|
|
10750 && STRNCMP(fword + sp->ts_splitfidx,
|
|
|
10751 tword + sp->ts_splitoff,
|
|
|
10752 sp->ts_fidx - sp->ts_splitfidx) == 0)
|
|
492
|
10753 {
|
|
625
|
10754 preword[sp->ts_prewordlen] = NUL;
|
|
|
10755 newscore = score_wordcount_adj(slang, sp->ts_score,
|
|
|
10756 preword + sp->ts_prewordlen,
|
|
|
10757 sp->ts_prewordlen > 0);
|
|
|
10758 /* Add the suggestion if the score isn't too bad. */
|
|
|
10759 if (newscore <= su->su_maxscore)
|
|
492
|
10760 add_suggestion(su, &su->su_ga, preword,
|
|
503
|
10761 sp->ts_splitfidx - repextra,
|
|
625
|
10762 newscore, 0, FALSE,
|
|
|
10763 lp->lp_sallang, FALSE);
|
|
|
10764 break;
|
|
492
|
10765 }
|
|
490
|
10766 }
|
|
|
10767 else
|
|
625
|
10768 {
|
|
|
10769 /* There was a compound word before this word. If this
|
|
|
10770 * word does not support compounding then give up
|
|
|
10771 * (splitting is tried for the word without compound
|
|
|
10772 * flag). */
|
|
|
10773 if (((unsigned)flags >> 24) == 0
|
|
|
10774 || sp->ts_twordlen - sp->ts_splitoff
|
|
|
10775 < slang->sl_compminlen)
|
|
|
10776 break;
|
|
|
10777 #ifdef FEAT_MBYTE
|
|
|
10778 /* For multi-byte chars check character length against
|
|
|
10779 * COMPOUNDMIN. */
|
|
|
10780 if (has_mbyte
|
|
|
10781 && slang->sl_compminlen > 0
|
|
|
10782 && mb_charlen(tword + sp->ts_splitoff)
|
|
|
10783 < slang->sl_compminlen)
|
|
|
10784 break;
|
|
|
10785 #endif
|
|
|
10786
|
|
|
10787 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
10788 compflags[sp->ts_complen + 1] = NUL;
|
|
|
10789 vim_strncpy(preword + sp->ts_prewordlen,
|
|
|
10790 tword + sp->ts_splitoff,
|
|
|
10791 sp->ts_twordlen - sp->ts_splitoff);
|
|
|
10792 p = preword;
|
|
|
10793 while (*skiptowhite(p) != NUL)
|
|
|
10794 p = skipwhite(skiptowhite(p));
|
|
|
10795 if (fword_ends && !can_compound(slang, p,
|
|
|
10796 compflags + sp->ts_compsplit))
|
|
|
10797 /* Compound is not allowed. But it may still be
|
|
|
10798 * possible if we add another (short) word. */
|
|
|
10799 compound_ok = FALSE;
|
|
|
10800
|
|
|
10801 /* Get pointer to last char of previous word. */
|
|
|
10802 p = preword + sp->ts_prewordlen;
|
|
|
10803 mb_ptr_back(preword, p);
|
|
|
10804 }
|
|
|
10805 }
|
|
|
10806
|
|
|
10807 /*
|
|
|
10808 * Form the word with proper case in preword.
|
|
|
10809 * If there is a word from a previous split, append.
|
|
|
10810 * For the soundfold tree don't change the case, simply append.
|
|
|
10811 */
|
|
|
10812 if (soundfold)
|
|
|
10813 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff);
|
|
|
10814 else if (flags & WF_KEEPCAP)
|
|
|
10815 /* Must find the word in the keep-case tree. */
|
|
|
10816 find_keepcap_word(slang, tword + sp->ts_splitoff,
|
|
483
|
10817 preword + sp->ts_prewordlen);
|
|
625
|
10818 else
|
|
|
10819 {
|
|
|
10820 /* Include badflags: If the badword is onecap or allcap
|
|
|
10821 * use that for the goodword too. But if the badword is
|
|
|
10822 * allcap and it's only one char long use onecap. */
|
|
|
10823 c = su->su_badflags;
|
|
|
10824 if ((c & WF_ALLCAP)
|
|
|
10825 #ifdef FEAT_MBYTE
|
|
|
10826 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
10827 #else
|
|
625
|
10828 && su->su_badlen == 1
|
|
|
10829 #endif
|
|
|
10830 )
|
|
|
10831 c = WF_ONECAP;
|
|
|
10832 c |= flags;
|
|
|
10833
|
|
|
10834 /* When appending a compound word after a word character don't
|
|
|
10835 * use Onecap. */
|
|
|
10836 if (p != NULL && spell_iswordp_nmw(p))
|
|
|
10837 c &= ~WF_ONECAP;
|
|
|
10838 make_case_word(tword + sp->ts_splitoff,
|
|
490
|
10839 preword + sp->ts_prewordlen, c);
|
|
625
|
10840 }
|
|
|
10841
|
|
|
10842 if (!soundfold)
|
|
|
10843 {
|
|
323
|
10844 /* Don't use a banned word. It may appear again as a good
|
|
|
10845 * word, thus remember it. */
|
|
|
10846 if (flags & WF_BANNED)
|
|
|
10847 {
|
|
483
|
10848 add_banned(su, preword + sp->ts_prewordlen);
|
|
323
|
10849 break;
|
|
|
10850 }
|
|
534
|
10851 if ((sp->ts_complen == sp->ts_compsplit
|
|
625
|
10852 && WAS_BANNED(su, preword + sp->ts_prewordlen))
|
|
|
10853 || WAS_BANNED(su, preword))
|
|
534
|
10854 {
|
|
|
10855 if (slang->sl_compprog == NULL)
|
|
|
10856 break;
|
|
|
10857 /* the word so far was banned but we may try compounding */
|
|
|
10858 goodword_ends = FALSE;
|
|
|
10859 }
|
|
625
|
10860 }
|
|
|
10861
|
|
|
10862 newscore = 0;
|
|
|
10863 if (!soundfold) /* soundfold words don't have flags */
|
|
|
10864 {
|
|
323
|
10865 if ((flags & WF_REGION)
|
|
390
|
10866 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
10867 newscore += SCORE_REGION;
|
|
|
10868 if (flags & WF_RARE)
|
|
|
10869 newscore += SCORE_RARE;
|
|
|
10870
|
|
346
|
10871 if (!spell_valid_case(su->su_badflags,
|
|
483
|
10872 captype(preword + sp->ts_prewordlen, NULL)))
|
|
323
|
10873 newscore += SCORE_ICASE;
|
|
625
|
10874 }
|
|
|
10875
|
|
|
10876 /* TODO: how about splitting in the soundfold tree? */
|
|
|
10877 if (fword_ends
|
|
|
10878 && goodword_ends
|
|
|
10879 && sp->ts_fidx >= sp->ts_fidxtry
|
|
|
10880 && compound_ok)
|
|
|
10881 {
|
|
|
10882 /* The badword also ends: add suggestions. */
|
|
|
10883 #ifdef DEBUG_TRIEWALK
|
|
|
10884 if (soundfold && STRCMP(preword, "smwrd") == 0)
|
|
|
10885 {
|
|
|
10886 int j;
|
|
|
10887
|
|
|
10888 /* print the stack of changes that brought us here */
|
|
|
10889 smsg("------ %s -------", fword);
|
|
|
10890 for (j = 0; j < depth; ++j)
|
|
|
10891 smsg("%s", changename[j]);
|
|
|
10892 }
|
|
|
10893 #endif
|
|
|
10894 if (soundfold)
|
|
|
10895 {
|
|
|
10896 /* For soundfolded words we need to find the original
|
|
|
10897 * words, the edit distrance and then add them. */
|
|
|
10898 add_sound_suggest(su, preword, sp->ts_score, lp);
|
|
|
10899 }
|
|
|
10900 else
|
|
|
10901 {
|
|
|
10902 /* Give a penalty when changing non-word char to word
|
|
|
10903 * char, e.g., "thes," -> "these". */
|
|
366
|
10904 p = fword + sp->ts_fidx;
|
|
625
|
10905 mb_ptr_back(fword, p);
|
|
376
|
10906 if (!spell_iswordp(p, curbuf))
|
|
366
|
10907 {
|
|
|
10908 p = preword + STRLEN(preword);
|
|
625
|
10909 mb_ptr_back(preword, p);
|
|
376
|
10910 if (spell_iswordp(p, curbuf))
|
|
366
|
10911 newscore += SCORE_NONWORD;
|
|
|
10912 }
|
|
|
10913
|
|
625
|
10914 /* Give a bonus to words seen before. */
|
|
|
10915 score = score_wordcount_adj(slang,
|
|
|
10916 sp->ts_score + newscore,
|
|
|
10917 preword + sp->ts_prewordlen,
|
|
|
10918 sp->ts_prewordlen > 0);
|
|
|
10919
|
|
|
10920 /* Add the suggestion if the score isn't too bad. */
|
|
|
10921 if (score <= su->su_maxscore)
|
|
633
|
10922 {
|
|
625
|
10923 add_suggestion(su, &su->su_ga, preword,
|
|
|
10924 sp->ts_fidx - repextra,
|
|
|
10925 score, 0, FALSE, lp->lp_sallang, FALSE);
|
|
633
|
10926
|
|
|
10927 if (su->su_badflags & WF_MIXCAP)
|
|
|
10928 {
|
|
|
10929 /* We really don't know if the word should be
|
|
|
10930 * upper or lower case, add both. */
|
|
|
10931 c = captype(preword, NULL);
|
|
|
10932 if (c == 0 || c == WF_ALLCAP)
|
|
|
10933 {
|
|
|
10934 make_case_word(tword + sp->ts_splitoff,
|
|
|
10935 preword + sp->ts_prewordlen,
|
|
|
10936 c == 0 ? WF_ALLCAP : 0);
|
|
|
10937
|
|
|
10938 add_suggestion(su, &su->su_ga, preword,
|
|
|
10939 sp->ts_fidx - repextra,
|
|
|
10940 score + SCORE_ICASE, 0, FALSE,
|
|
|
10941 lp->lp_sallang, FALSE);
|
|
|
10942 }
|
|
|
10943 }
|
|
|
10944 }
|
|
625
|
10945 }
|
|
|
10946 }
|
|
|
10947
|
|
|
10948 /*
|
|
|
10949 * Try word split and/or compounding.
|
|
|
10950 */
|
|
|
10951 if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
|
|
|
10952 #ifdef FEAT_MBYTE
|
|
|
10953 /* Don't split halfway a character. */
|
|
|
10954 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
10955 #endif
|
|
|
10956 )
|
|
|
10957 {
|
|
|
10958 int try_compound;
|
|
|
10959 int try_split;
|
|
|
10960
|
|
|
10961 /* If past the end of the bad word don't try a split.
|
|
|
10962 * Otherwise try changing the next word. E.g., find
|
|
|
10963 * suggestions for "the the" where the second "the" is
|
|
|
10964 * different. It's done like a split.
|
|
|
10965 * TODO: word split for soundfold words */
|
|
|
10966 try_split = (sp->ts_fidx - repextra < su->su_badlen)
|
|
|
10967 && !soundfold;
|
|
|
10968
|
|
|
10969 /* Get here in several situations:
|
|
|
10970 * 1. The word in the tree ends:
|
|
|
10971 * If the word allows compounding try that. Otherwise try
|
|
|
10972 * a split by inserting a space. For both check that a
|
|
|
10973 * valid words starts at fword[sp->ts_fidx].
|
|
|
10974 * For NOBREAK do like compounding to be able to check if
|
|
|
10975 * the next word is valid.
|
|
|
10976 * 2. The badword does end, but it was due to a change (e.g.,
|
|
|
10977 * a swap). No need to split, but do check that the
|
|
|
10978 * following word is valid.
|
|
|
10979 * 3. The badword and the word in the tree end. It may still
|
|
|
10980 * be possible to compound another (short) word.
|
|
|
10981 */
|
|
|
10982 try_compound = FALSE;
|
|
|
10983 if (!soundfold
|
|
|
10984 && slang->sl_compprog != NULL
|
|
|
10985 && ((unsigned)flags >> 24) != 0
|
|
|
10986 && sp->ts_twordlen - sp->ts_splitoff
|
|
|
10987 >= slang->sl_compminlen
|
|
|
10988 #ifdef FEAT_MBYTE
|
|
|
10989 && (!has_mbyte
|
|
|
10990 || slang->sl_compminlen == 0
|
|
|
10991 || mb_charlen(tword + sp->ts_splitoff)
|
|
|
10992 >= slang->sl_compminlen)
|
|
|
10993 #endif
|
|
|
10994 && (slang->sl_compsylmax < MAXWLEN
|
|
|
10995 || sp->ts_complen + 1 - sp->ts_compsplit
|
|
|
10996 < slang->sl_compmax)
|
|
|
10997 && (byte_in_str(sp->ts_complen == sp->ts_compsplit
|
|
|
10998 ? slang->sl_compstartflags
|
|
|
10999 : slang->sl_compallflags,
|
|
|
11000 ((unsigned)flags >> 24))))
|
|
|
11001 {
|
|
|
11002 try_compound = TRUE;
|
|
|
11003 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
11004 compflags[sp->ts_complen + 1] = NUL;
|
|
|
11005 }
|
|
|
11006
|
|
|
11007 /* For NOBREAK we never try splitting, it won't make any word
|
|
|
11008 * valid. */
|
|
|
11009 if (slang->sl_nobreak)
|
|
|
11010 try_compound = TRUE;
|
|
|
11011
|
|
|
11012 /* If we could add a compound word, and it's also possible to
|
|
|
11013 * split at this point, do the split first and set
|
|
|
11014 * TSF_DIDSPLIT to avoid doing it again. */
|
|
|
11015 else if (!fword_ends
|
|
|
11016 && try_compound
|
|
|
11017 && (sp->ts_flags & TSF_DIDSPLIT) == 0)
|
|
|
11018 {
|
|
485
|
11019 try_compound = FALSE;
|
|
625
|
11020 sp->ts_flags |= TSF_DIDSPLIT;
|
|
|
11021 --sp->ts_curi; /* do the same NUL again */
|
|
|
11022 compflags[sp->ts_complen] = NUL;
|
|
|
11023 }
|
|
|
11024 else
|
|
|
11025 sp->ts_flags &= ~TSF_DIDSPLIT;
|
|
|
11026
|
|
|
11027 if (try_split || try_compound)
|
|
|
11028 {
|
|
534
|
11029 if (!try_compound && (!fword_ends || !goodword_ends))
|
|
485
|
11030 {
|
|
|
11031 /* If we're going to split need to check that the
|
|
501
|
11032 * words so far are valid for compounding. If there
|
|
|
11033 * is only one word it must not have the NEEDCOMPOUND
|
|
|
11034 * flag. */
|
|
|
11035 if (sp->ts_complen == sp->ts_compsplit
|
|
|
11036 && (flags & WF_NEEDCOMP))
|
|
|
11037 break;
|
|
490
|
11038 p = preword;
|
|
|
11039 while (*skiptowhite(p) != NUL)
|
|
|
11040 p = skipwhite(skiptowhite(p));
|
|
485
|
11041 if (sp->ts_complen > sp->ts_compsplit
|
|
490
|
11042 && !can_compound(slang, p,
|
|
485
|
11043 compflags + sp->ts_compsplit))
|
|
|
11044 break;
|
|
|
11045 newscore += SCORE_SPLIT;
|
|
625
|
11046
|
|
|
11047 /* Give a bonus to words seen before. */
|
|
|
11048 newscore = score_wordcount_adj(slang, newscore,
|
|
|
11049 preword + sp->ts_prewordlen, TRUE);
|
|
482
|
11050 }
|
|
|
11051
|
|
625
|
11052 if (TRY_DEEPER(su, stack, depth, newscore))
|
|
323
|
11053 {
|
|
625
|
11054 go_deeper(stack, depth, newscore);
|
|
|
11055 #ifdef DEBUG_TRIEWALK
|
|
|
11056 if (!try_compound && !fword_ends)
|
|
|
11057 sprintf(changename[depth], "%.*s-%s: split",
|
|
|
11058 sp->ts_twordlen, tword, fword + sp->ts_fidx);
|
|
|
11059 else
|
|
|
11060 sprintf(changename[depth], "%.*s-%s: compound",
|
|
|
11061 sp->ts_twordlen, tword, fword + sp->ts_fidx);
|
|
|
11062 #endif
|
|
323
|
11063 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
346
|
11064 sp->ts_save_badflags = su->su_badflags;
|
|
376
|
11065 sp->ts_state = STATE_SPLITUNDO;
|
|
|
11066
|
|
|
11067 ++depth;
|
|
|
11068 sp = &stack[depth];
|
|
323
|
11069
|
|
482
|
11070 /* Append a space to preword when splitting. */
|
|
|
11071 if (!try_compound && !fword_ends)
|
|
|
11072 STRCAT(preword, " ");
|
|
483
|
11073 sp->ts_prewordlen = STRLEN(preword);
|
|
|
11074 sp->ts_splitoff = sp->ts_twordlen;
|
|
492
|
11075 sp->ts_splitfidx = sp->ts_fidx;
|
|
376
|
11076
|
|
|
11077 /* If the badword has a non-word character at this
|
|
|
11078 * position skip it. That means replacing the
|
|
482
|
11079 * non-word character with a space. Always skip a
|
|
534
|
11080 * character when the word ends. But only when the
|
|
|
11081 * good word can end. */
|
|
625
|
11082 if (((!try_compound && !spell_iswordp_nmw(fword
|
|
|
11083 + sp->ts_fidx))
|
|
|
11084 || fword_ends)
|
|
|
11085 && fword[sp->ts_fidx] != NUL
|
|
|
11086 && goodword_ends)
|
|
376
|
11087 {
|
|
482
|
11088 int l;
|
|
|
11089
|
|
376
|
11090 #ifdef FEAT_MBYTE
|
|
|
11091 if (has_mbyte)
|
|
482
|
11092 l = MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
376
|
11093 else
|
|
|
11094 #endif
|
|
482
|
11095 l = 1;
|
|
|
11096 if (fword_ends)
|
|
|
11097 {
|
|
|
11098 /* Copy the skipped character to preword. */
|
|
483
|
11099 mch_memmove(preword + sp->ts_prewordlen,
|
|
482
|
11100 fword + sp->ts_fidx, l);
|
|
483
|
11101 sp->ts_prewordlen += l;
|
|
|
11102 preword[sp->ts_prewordlen] = NUL;
|
|
482
|
11103 }
|
|
|
11104 else
|
|
|
11105 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
11106 sp->ts_fidx += l;
|
|
376
|
11107 }
|
|
455
|
11108
|
|
485
|
11109 /* When compounding include compound flag in
|
|
|
11110 * compflags[] (already set above). When splitting we
|
|
|
11111 * may start compounding over again. */
|
|
482
|
11112 if (try_compound)
|
|
483
|
11113 ++sp->ts_complen;
|
|
482
|
11114 else
|
|
485
|
11115 sp->ts_compsplit = sp->ts_complen;
|
|
|
11116 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
482
|
11117
|
|
455
|
11118 /* set su->su_badflags to the caps type at this
|
|
|
11119 * position */
|
|
324
|
11120 #ifdef FEAT_MBYTE
|
|
|
11121 if (has_mbyte)
|
|
455
|
11122 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
11123 else
|
|
|
11124 #endif
|
|
455
|
11125 n = sp->ts_fidx;
|
|
474
|
11126 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
11127 su->su_badptr + su->su_badlen);
|
|
323
|
11128
|
|
|
11129 /* Restart at top of the tree. */
|
|
376
|
11130 sp->ts_arridx = 0;
|
|
485
|
11131
|
|
|
11132 /* If there are postponed prefixes, try these too. */
|
|
|
11133 if (pbyts != NULL)
|
|
|
11134 {
|
|
|
11135 byts = pbyts;
|
|
|
11136 idxs = pidxs;
|
|
|
11137 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
|
11138 sp->ts_state = STATE_NOPREFIX;
|
|
|
11139 }
|
|
323
|
11140 }
|
|
|
11141 }
|
|
625
|
11142 }
|
|
|
11143 break;
|
|
|
11144
|
|
|
11145 case STATE_SPLITUNDO:
|
|
|
11146 /* Undo the changes done for word split or compound word. */
|
|
|
11147 su->su_badflags = sp->ts_save_badflags;
|
|
|
11148
|
|
|
11149 /* Continue looking for NUL bytes. */
|
|
|
11150 sp->ts_state = STATE_START;
|
|
|
11151
|
|
|
11152 /* In case we went into the prefix tree. */
|
|
|
11153 byts = fbyts;
|
|
|
11154 idxs = fidxs;
|
|
|
11155 break;
|
|
|
11156
|
|
|
11157 case STATE_ENDNUL:
|
|
|
11158 /* Past the NUL bytes in the node. */
|
|
|
11159 su->su_badflags = sp->ts_save_badflags;
|
|
|
11160 if (fword[sp->ts_fidx] == NUL
|
|
|
11161 #ifdef FEAT_MBYTE
|
|
|
11162 && sp->ts_tcharlen == 0
|
|
|
11163 #endif
|
|
|
11164 )
|
|
|
11165 {
|
|
|
11166 /* The badword ends, can't use STATE_PLAIN. */
|
|
|
11167 sp->ts_state = STATE_DEL;
|
|
323
|
11168 break;
|
|
625
|
11169 }
|
|
|
11170 sp->ts_state = STATE_PLAIN;
|
|
|
11171 /*FALLTHROUGH*/
|
|
|
11172
|
|
|
11173 case STATE_PLAIN:
|
|
|
11174 /*
|
|
|
11175 * Go over all possible bytes at this node, add each to tword[]
|
|
|
11176 * and use child node. "ts_curi" is the index.
|
|
|
11177 */
|
|
|
11178 arridx = sp->ts_arridx;
|
|
|
11179 if (sp->ts_curi > byts[arridx])
|
|
|
11180 {
|
|
|
11181 /* Done all bytes at this node, do next state. When still at
|
|
|
11182 * already changed bytes skip the other tricks. */
|
|
|
11183 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
11184 sp->ts_state = STATE_DEL;
|
|
|
11185 else
|
|
625
|
11186 sp->ts_state = STATE_FINAL;
|
|
|
11187 }
|
|
|
11188 else
|
|
|
11189 {
|
|
|
11190 arridx += sp->ts_curi++;
|
|
|
11191 c = byts[arridx];
|
|
|
11192
|
|
|
11193 /* Normal byte, go one level deeper. If it's not equal to the
|
|
|
11194 * byte in the bad word adjust the score. But don't even try
|
|
|
11195 * when the byte was already changed. And don't try when we
|
|
|
11196 * just deleted this byte, accepting it is always cheaper then
|
|
|
11197 * delete + substitute. */
|
|
|
11198 if (c == fword[sp->ts_fidx]
|
|
|
11199 #ifdef FEAT_MBYTE
|
|
|
11200 || (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE)
|
|
|
11201 #endif
|
|
|
11202 )
|
|
|
11203 newscore = 0;
|
|
|
11204 else
|
|
|
11205 newscore = SCORE_SUBST;
|
|
|
11206 if ((newscore == 0
|
|
|
11207 || (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
11208 && ((sp->ts_flags & TSF_DIDDEL) == 0
|
|
|
11209 || c != fword[sp->ts_delidx])))
|
|
|
11210 && TRY_DEEPER(su, stack, depth, newscore))
|
|
|
11211 {
|
|
|
11212 go_deeper(stack, depth, newscore);
|
|
|
11213 #ifdef DEBUG_TRIEWALK
|
|
|
11214 if (newscore > 0)
|
|
|
11215 sprintf(changename[depth], "%.*s-%s: subst %c to %c",
|
|
|
11216 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11217 fword[sp->ts_fidx], c);
|
|
323
|
11218 else
|
|
625
|
11219 sprintf(changename[depth], "%.*s-%s: accept %c",
|
|
|
11220 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11221 fword[sp->ts_fidx]);
|
|
|
11222 #endif
|
|
323
|
11223 ++depth;
|
|
625
|
11224 sp = &stack[depth];
|
|
|
11225 ++sp->ts_fidx;
|
|
|
11226 tword[sp->ts_twordlen++] = c;
|
|
|
11227 sp->ts_arridx = idxs[arridx];
|
|
|
11228 #ifdef FEAT_MBYTE
|
|
|
11229 if (newscore == SCORE_SUBST)
|
|
|
11230 sp->ts_isdiff = DIFF_YES;
|
|
330
|
11231 if (has_mbyte)
|
|
358
|
11232 {
|
|
625
|
11233 /* Multi-byte characters are a bit complicated to
|
|
|
11234 * handle: They differ when any of the bytes differ
|
|
|
11235 * and then their length may also differ. */
|
|
|
11236 if (sp->ts_tcharlen == 0)
|
|
|
11237 {
|
|
|
11238 /* First byte. */
|
|
|
11239 sp->ts_tcharidx = 0;
|
|
|
11240 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
11241 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
11242 sp->ts_isdiff = (newscore != 0)
|
|
|
11243 ? DIFF_YES : DIFF_NONE;
|
|
|
11244 }
|
|
|
11245 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
11246 /* When inserting trail bytes don't advance in the
|
|
|
11247 * bad word. */
|
|
|
11248 --sp->ts_fidx;
|
|
|
11249 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
11250 {
|
|
|
11251 /* Last byte of character. */
|
|
|
11252 if (sp->ts_isdiff == DIFF_YES)
|
|
|
11253 {
|
|
|
11254 /* Correct ts_fidx for the byte length of the
|
|
|
11255 * character (we didn't check that before). */
|
|
|
11256 sp->ts_fidx = sp->ts_fcharstart
|
|
|
11257 + MB_BYTE2LEN(
|
|
|
11258 fword[sp->ts_fcharstart]);
|
|
|
11259
|
|
|
11260 /* For changing a composing character adjust
|
|
|
11261 * the score from SCORE_SUBST to
|
|
|
11262 * SCORE_SUBCOMP. */
|
|
|
11263 if (enc_utf8
|
|
|
11264 && utf_iscomposing(
|
|
|
11265 mb_ptr2char(tword
|
|
|
11266 + sp->ts_twordlen
|
|
|
11267 - sp->ts_tcharlen))
|
|
|
11268 && utf_iscomposing(
|
|
|
11269 mb_ptr2char(fword
|
|
|
11270 + sp->ts_fcharstart)))
|
|
|
11271 sp->ts_score -=
|
|
|
11272 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
11273
|
|
|
11274 /* For a similar character adjust score from
|
|
|
11275 * SCORE_SUBST to SCORE_SIMILAR. */
|
|
|
11276 else if (!soundfold
|
|
|
11277 && slang->sl_has_map
|
|
|
11278 && similar_chars(slang,
|
|
|
11279 mb_ptr2char(tword
|
|
|
11280 + sp->ts_twordlen
|
|
|
11281 - sp->ts_tcharlen),
|
|
|
11282 mb_ptr2char(fword
|
|
|
11283 + sp->ts_fcharstart)))
|
|
|
11284 sp->ts_score -=
|
|
|
11285 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
11286 }
|
|
|
11287 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
11288 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
11289 {
|
|
|
11290 p = tword + sp->ts_twordlen - sp->ts_tcharlen;
|
|
|
11291 c = mb_ptr2char(p);
|
|
|
11292 if (enc_utf8 && utf_iscomposing(c))
|
|
|
11293 {
|
|
|
11294 /* Inserting a composing char doesn't
|
|
|
11295 * count that much. */
|
|
|
11296 sp->ts_score -= SCORE_INS - SCORE_INSCOMP;
|
|
|
11297 }
|
|
|
11298 else
|
|
|
11299 {
|
|
|
11300 /* If the previous character was the same,
|
|
|
11301 * thus doubling a character, give a bonus
|
|
|
11302 * to the score. Also for the soundfold
|
|
|
11303 * tree (might seem illogical but does
|
|
|
11304 * give better scores). */
|
|
|
11305 mb_ptr_back(tword, p);
|
|
|
11306 if (c == mb_ptr2char(p))
|
|
|
11307 sp->ts_score -= SCORE_INS
|
|
|
11308 - SCORE_INSDUP;
|
|
|
11309 }
|
|
|
11310 }
|
|
|
11311
|
|
|
11312 /* Starting a new char, reset the length. */
|
|
|
11313 sp->ts_tcharlen = 0;
|
|
|
11314 }
|
|
358
|
11315 }
|
|
330
|
11316 else
|
|
|
11317 #endif
|
|
358
|
11318 {
|
|
625
|
11319 /* If we found a similar char adjust the score.
|
|
|
11320 * We do this after calling go_deeper() because
|
|
|
11321 * it's slow. */
|
|
|
11322 if (newscore != 0
|
|
|
11323 && !soundfold
|
|
|
11324 && slang->sl_has_map
|
|
|
11325 && similar_chars(slang,
|
|
|
11326 c, fword[sp->ts_fidx - 1]))
|
|
|
11327 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
358
|
11328 }
|
|
625
|
11329 }
|
|
|
11330 }
|
|
|
11331 break;
|
|
|
11332
|
|
|
11333 case STATE_DEL:
|
|
|
11334 #ifdef FEAT_MBYTE
|
|
|
11335 /* When past the first byte of a multi-byte char don't try
|
|
|
11336 * delete/insert/swap a character. */
|
|
|
11337 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
11338 {
|
|
|
11339 sp->ts_state = STATE_FINAL;
|
|
|
11340 break;
|
|
|
11341 }
|
|
|
11342 #endif
|
|
|
11343 /*
|
|
|
11344 * Try skipping one character in the bad word (delete it).
|
|
|
11345 */
|
|
|
11346 sp->ts_state = STATE_INS_PREP;
|
|
|
11347 sp->ts_curi = 1;
|
|
|
11348 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*')
|
|
|
11349 /* Deleting a vowel at the start of a word counts less, see
|
|
|
11350 * soundalike_score(). */
|
|
|
11351 newscore = 2 * SCORE_DEL / 3;
|
|
|
11352 else
|
|
|
11353 newscore = SCORE_DEL;
|
|
|
11354 if (fword[sp->ts_fidx] != NUL
|
|
|
11355 && TRY_DEEPER(su, stack, depth, newscore))
|
|
|
11356 {
|
|
|
11357 go_deeper(stack, depth, newscore);
|
|
|
11358 #ifdef DEBUG_TRIEWALK
|
|
|
11359 sprintf(changename[depth], "%.*s-%s: delete %c",
|
|
|
11360 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11361 fword[sp->ts_fidx]);
|
|
|
11362 #endif
|
|
|
11363 ++depth;
|
|
|
11364
|
|
|
11365 /* Remember what character we deleted, so that we can avoid
|
|
|
11366 * inserting it again. */
|
|
|
11367 stack[depth].ts_flags |= TSF_DIDDEL;
|
|
|
11368 stack[depth].ts_delidx = sp->ts_fidx;
|
|
|
11369
|
|
|
11370 /* Advance over the character in fword[]. Give a bonus to the
|
|
|
11371 * score if the same character is following "nn" -> "n". It's
|
|
|
11372 * a bit illogical for soundfold tree but it does give better
|
|
|
11373 * results. */
|
|
|
11374 #ifdef FEAT_MBYTE
|
|
|
11375 if (has_mbyte)
|
|
|
11376 {
|
|
|
11377 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
|
11378 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
|
11379 if (enc_utf8 && utf_iscomposing(c))
|
|
|
11380 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
11381 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
|
11382 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
323
|
11383 }
|
|
|
11384 else
|
|
625
|
11385 #endif
|
|
|
11386 {
|
|
|
11387 ++stack[depth].ts_fidx;
|
|
|
11388 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
11389 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
323
|
11390 }
|
|
|
11391 break;
|
|
625
|
11392 }
|
|
|
11393 /*FALLTHROUGH*/
|
|
|
11394
|
|
|
11395 case STATE_INS_PREP:
|
|
|
11396 if (sp->ts_flags & TSF_DIDDEL)
|
|
|
11397 {
|
|
|
11398 /* If we just deleted a byte then inserting won't make sense,
|
|
|
11399 * a substitute is always cheaper. */
|
|
|
11400 sp->ts_state = STATE_SWAP;
|
|
|
11401 break;
|
|
|
11402 }
|
|
|
11403
|
|
|
11404 /* skip over NUL bytes */
|
|
|
11405 n = sp->ts_arridx;
|
|
|
11406 for (;;)
|
|
|
11407 {
|
|
|
11408 if (sp->ts_curi > byts[n])
|
|
|
11409 {
|
|
|
11410 /* Only NUL bytes at this node, go to next state. */
|
|
|
11411 sp->ts_state = STATE_SWAP;
|
|
|
11412 break;
|
|
|
11413 }
|
|
|
11414 if (byts[n + sp->ts_curi] != NUL)
|
|
|
11415 {
|
|
|
11416 /* Found a byte to insert. */
|
|
|
11417 sp->ts_state = STATE_INS;
|
|
330
|
11418 break;
|
|
|
11419 }
|
|
625
|
11420 ++sp->ts_curi;
|
|
|
11421 }
|
|
|
11422 break;
|
|
|
11423
|
|
|
11424 /*FALLTHROUGH*/
|
|
|
11425
|
|
|
11426 case STATE_INS:
|
|
|
11427 /* Insert one byte. Repeat this for each possible byte at this
|
|
|
11428 * node. */
|
|
|
11429 n = sp->ts_arridx;
|
|
|
11430 if (sp->ts_curi > byts[n])
|
|
|
11431 {
|
|
|
11432 /* Done all bytes at this node, go to next state. */
|
|
|
11433 sp->ts_state = STATE_SWAP;
|
|
|
11434 break;
|
|
|
11435 }
|
|
|
11436
|
|
|
11437 /* Do one more byte at this node, but:
|
|
|
11438 * - Skip NUL bytes.
|
|
|
11439 * - Skip the byte if it's equal to the byte in the word,
|
|
|
11440 * accepting that byte is always better.
|
|
|
11441 */
|
|
|
11442 n += sp->ts_curi++;
|
|
|
11443 c = byts[n];
|
|
|
11444 if (soundfold && sp->ts_twordlen == 0 && c == '*')
|
|
|
11445 /* Inserting a vowel at the start of a word counts less,
|
|
|
11446 * see soundalike_score(). */
|
|
|
11447 newscore = 2 * SCORE_INS / 3;
|
|
|
11448 else
|
|
|
11449 newscore = SCORE_INS;
|
|
|
11450 if (c != fword[sp->ts_fidx]
|
|
|
11451 && TRY_DEEPER(su, stack, depth, newscore))
|
|
|
11452 {
|
|
|
11453 go_deeper(stack, depth, newscore);
|
|
|
11454 #ifdef DEBUG_TRIEWALK
|
|
|
11455 sprintf(changename[depth], "%.*s-%s: insert %c",
|
|
|
11456 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11457 c);
|
|
|
11458 #endif
|
|
|
11459 ++depth;
|
|
|
11460 sp = &stack[depth];
|
|
|
11461 tword[sp->ts_twordlen++] = c;
|
|
|
11462 sp->ts_arridx = idxs[n];
|
|
330
|
11463 #ifdef FEAT_MBYTE
|
|
|
11464 if (has_mbyte)
|
|
|
11465 {
|
|
625
|
11466 fl = MB_BYTE2LEN(c);
|
|
|
11467 if (fl > 1)
|
|
330
|
11468 {
|
|
625
|
11469 /* There are following bytes for the same character.
|
|
|
11470 * We must find all bytes before trying
|
|
|
11471 * delete/insert/swap/etc. */
|
|
|
11472 sp->ts_tcharlen = fl;
|
|
|
11473 sp->ts_tcharidx = 1;
|
|
|
11474 sp->ts_isdiff = DIFF_INSERT;
|
|
330
|
11475 }
|
|
323
|
11476 }
|
|
|
11477 else
|
|
625
|
11478 fl = 1;
|
|
|
11479 if (fl == 1)
|
|
|
11480 #endif
|
|
|
11481 {
|
|
|
11482 /* If the previous character was the same, thus doubling a
|
|
|
11483 * character, give a bonus to the score. Also for
|
|
|
11484 * soundfold words (illogical but does give a better
|
|
|
11485 * score). */
|
|
|
11486 if (sp->ts_twordlen >= 2
|
|
|
11487 && tword[sp->ts_twordlen - 2] == c)
|
|
|
11488 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
11489 }
|
|
|
11490 }
|
|
|
11491 break;
|
|
|
11492
|
|
|
11493 case STATE_SWAP:
|
|
|
11494 /*
|
|
|
11495 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
11496 * We change "fword" here, it's changed back afterwards at
|
|
|
11497 * STATE_UNSWAP.
|
|
|
11498 */
|
|
|
11499 p = fword + sp->ts_fidx;
|
|
|
11500 c = *p;
|
|
|
11501 if (c == NUL)
|
|
|
11502 {
|
|
|
11503 /* End of word, can't swap or replace. */
|
|
|
11504 sp->ts_state = STATE_FINAL;
|
|
|
11505 break;
|
|
|
11506 }
|
|
|
11507
|
|
|
11508 /* Don't swap if the first character is not a word character.
|
|
|
11509 * SWAP3 etc. also don't make sense then. */
|
|
|
11510 if (!soundfold && !spell_iswordp(p, curbuf))
|
|
|
11511 {
|
|
|
11512 sp->ts_state = STATE_REP_INI;
|
|
323
|
11513 break;
|
|
625
|
11514 }
|
|
|
11515
|
|
|
11516 #ifdef FEAT_MBYTE
|
|
|
11517 if (has_mbyte)
|
|
|
11518 {
|
|
|
11519 n = mb_cptr2len(p);
|
|
|
11520 c = mb_ptr2char(p);
|
|
|
11521 if (!soundfold && !spell_iswordp(p + n, curbuf))
|
|
|
11522 c2 = c; /* don't swap non-word char */
|
|
|
11523 else
|
|
|
11524 c2 = mb_ptr2char(p + n);
|
|
|
11525 }
|
|
|
11526 else
|
|
|
11527 #endif
|
|
|
11528 {
|
|
|
11529 if (!soundfold && !spell_iswordp(p + 1, curbuf))
|
|
|
11530 c2 = c; /* don't swap non-word char */
|
|
|
11531 else
|
|
|
11532 c2 = p[1];
|
|
|
11533 }
|
|
|
11534
|
|
|
11535 /* When characters are identical, swap won't do anything.
|
|
|
11536 * Also get here if the second char is not a word character. */
|
|
|
11537 if (c == c2)
|
|
|
11538 {
|
|
|
11539 sp->ts_state = STATE_SWAP3;
|
|
|
11540 break;
|
|
|
11541 }
|
|
|
11542 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP))
|
|
|
11543 {
|
|
|
11544 go_deeper(stack, depth, SCORE_SWAP);
|
|
|
11545 #ifdef DEBUG_TRIEWALK
|
|
|
11546 sprintf(changename[depth], "%.*s-%s: swap %c and %c",
|
|
|
11547 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11548 c, c2);
|
|
|
11549 #endif
|
|
|
11550 sp->ts_state = STATE_UNSWAP;
|
|
|
11551 ++depth;
|
|
330
|
11552 #ifdef FEAT_MBYTE
|
|
|
11553 if (has_mbyte)
|
|
|
11554 {
|
|
625
|
11555 fl = mb_char2len(c2);
|
|
|
11556 mch_memmove(p, p + n, fl);
|
|
|
11557 mb_char2bytes(c, p + fl);
|
|
|
11558 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
330
|
11559 }
|
|
|
11560 else
|
|
|
11561 #endif
|
|
|
11562 {
|
|
625
|
11563 p[0] = c2;
|
|
330
|
11564 p[1] = c;
|
|
625
|
11565 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
11566 }
|
|
|
11567 }
|
|
|
11568 else
|
|
|
11569 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
11570 sp->ts_state = STATE_REP_INI;
|
|
|
11571 break;
|
|
|
11572
|
|
|
11573 case STATE_UNSWAP:
|
|
|
11574 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
11575 p = fword + sp->ts_fidx;
|
|
|
11576 #ifdef FEAT_MBYTE
|
|
|
11577 if (has_mbyte)
|
|
|
11578 {
|
|
|
11579 n = MB_BYTE2LEN(*p);
|
|
|
11580 c = mb_ptr2char(p + n);
|
|
|
11581 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
11582 mb_char2bytes(c, p);
|
|
|
11583 }
|
|
|
11584 else
|
|
|
11585 #endif
|
|
|
11586 {
|
|
|
11587 c = *p;
|
|
|
11588 *p = p[1];
|
|
|
11589 p[1] = c;
|
|
|
11590 }
|
|
|
11591 /*FALLTHROUGH*/
|
|
|
11592
|
|
|
11593 case STATE_SWAP3:
|
|
|
11594 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
|
11595 * "fword" here, it's changed back afterwards at STATE_UNSWAP3. */
|
|
|
11596 p = fword + sp->ts_fidx;
|
|
|
11597 #ifdef FEAT_MBYTE
|
|
|
11598 if (has_mbyte)
|
|
|
11599 {
|
|
|
11600 n = mb_cptr2len(p);
|
|
|
11601 c = mb_ptr2char(p);
|
|
|
11602 fl = mb_cptr2len(p + n);
|
|
|
11603 c2 = mb_ptr2char(p + n);
|
|
|
11604 if (!soundfold && !spell_iswordp(p + n + fl, curbuf))
|
|
|
11605 c3 = c; /* don't swap non-word char */
|
|
|
11606 else
|
|
|
11607 c3 = mb_ptr2char(p + n + fl);
|
|
|
11608 }
|
|
|
11609 else
|
|
|
11610 #endif
|
|
|
11611 {
|
|
|
11612 c = *p;
|
|
|
11613 c2 = p[1];
|
|
|
11614 if (!soundfold && !spell_iswordp(p + 2, curbuf))
|
|
|
11615 c3 = c; /* don't swap non-word char */
|
|
|
11616 else
|
|
|
11617 c3 = p[2];
|
|
|
11618 }
|
|
|
11619
|
|
|
11620 /* When characters are identical: "121" then SWAP3 result is
|
|
|
11621 * identical, ROT3L result is same as SWAP: "211", ROT3L result is
|
|
|
11622 * same as SWAP on next char: "112". Thus skip all swapping.
|
|
|
11623 * Also skip when c3 is NUL.
|
|
|
11624 * Also get here when the third character is not a word character.
|
|
|
11625 * Second character may any char: "a.b" -> "b.a" */
|
|
|
11626 if (c == c3 || c3 == NUL)
|
|
|
11627 {
|
|
|
11628 sp->ts_state = STATE_REP_INI;
|
|
|
11629 break;
|
|
|
11630 }
|
|
|
11631 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
|
|
|
11632 {
|
|
|
11633 go_deeper(stack, depth, SCORE_SWAP3);
|
|
|
11634 #ifdef DEBUG_TRIEWALK
|
|
|
11635 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c",
|
|
|
11636 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11637 c, c3);
|
|
|
11638 #endif
|
|
|
11639 sp->ts_state = STATE_UNSWAP3;
|
|
|
11640 ++depth;
|
|
|
11641 #ifdef FEAT_MBYTE
|
|
|
11642 if (has_mbyte)
|
|
|
11643 {
|
|
|
11644 tl = mb_char2len(c3);
|
|
|
11645 mch_memmove(p, p + n + fl, tl);
|
|
|
11646 mb_char2bytes(c2, p + tl);
|
|
|
11647 mb_char2bytes(c, p + fl + tl);
|
|
|
11648 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
11649 }
|
|
|
11650 else
|
|
|
11651 #endif
|
|
|
11652 {
|
|
|
11653 p[0] = p[2];
|
|
|
11654 p[2] = c;
|
|
|
11655 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
11656 }
|
|
|
11657 }
|
|
|
11658 else
|
|
|
11659 sp->ts_state = STATE_REP_INI;
|
|
|
11660 break;
|
|
|
11661
|
|
|
11662 case STATE_UNSWAP3:
|
|
|
11663 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
11664 p = fword + sp->ts_fidx;
|
|
|
11665 #ifdef FEAT_MBYTE
|
|
|
11666 if (has_mbyte)
|
|
|
11667 {
|
|
|
11668 n = MB_BYTE2LEN(*p);
|
|
|
11669 c2 = mb_ptr2char(p + n);
|
|
|
11670 fl = MB_BYTE2LEN(p[n]);
|
|
|
11671 c = mb_ptr2char(p + n + fl);
|
|
|
11672 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
11673 mch_memmove(p + fl + tl, p, n);
|
|
|
11674 mb_char2bytes(c, p);
|
|
|
11675 mb_char2bytes(c2, p + tl);
|
|
|
11676 p = p + tl;
|
|
|
11677 }
|
|
|
11678 else
|
|
|
11679 #endif
|
|
|
11680 {
|
|
|
11681 c = *p;
|
|
|
11682 *p = p[2];
|
|
|
11683 p[2] = c;
|
|
|
11684 ++p;
|
|
|
11685 }
|
|
|
11686
|
|
|
11687 if (!soundfold && !spell_iswordp(p, curbuf))
|
|
|
11688 {
|
|
|
11689 /* Middle char is not a word char, skip the rotate. First and
|
|
|
11690 * third char were already checked at swap and swap3. */
|
|
|
11691 sp->ts_state = STATE_REP_INI;
|
|
|
11692 break;
|
|
|
11693 }
|
|
|
11694
|
|
|
11695 /* Rotate three characters left: "123" -> "231". We change
|
|
|
11696 * "fword" here, it's changed back afterwards at STATE_UNROT3L. */
|
|
|
11697 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
|
|
|
11698 {
|
|
|
11699 go_deeper(stack, depth, SCORE_SWAP3);
|
|
|
11700 #ifdef DEBUG_TRIEWALK
|
|
|
11701 p = fword + sp->ts_fidx;
|
|
|
11702 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c",
|
|
|
11703 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11704 p[0], p[1], p[2]);
|
|
|
11705 #endif
|
|
|
11706 sp->ts_state = STATE_UNROT3L;
|
|
|
11707 ++depth;
|
|
330
|
11708 p = fword + sp->ts_fidx;
|
|
|
11709 #ifdef FEAT_MBYTE
|
|
|
11710 if (has_mbyte)
|
|
|
11711 {
|
|
474
|
11712 n = mb_cptr2len(p);
|
|
330
|
11713 c = mb_ptr2char(p);
|
|
474
|
11714 fl = mb_cptr2len(p + n);
|
|
625
|
11715 fl += mb_cptr2len(p + n + fl);
|
|
|
11716 mch_memmove(p, p + n, fl);
|
|
|
11717 mb_char2bytes(c, p + fl);
|
|
|
11718 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
330
|
11719 }
|
|
|
11720 else
|
|
|
11721 #endif
|
|
323
|
11722 {
|
|
330
|
11723 c = *p;
|
|
625
|
11724 *p = p[1];
|
|
|
11725 p[1] = p[2];
|
|
|
11726 p[2] = c;
|
|
|
11727 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
11728 }
|
|
|
11729 }
|
|
|
11730 else
|
|
|
11731 sp->ts_state = STATE_REP_INI;
|
|
|
11732 break;
|
|
|
11733
|
|
|
11734 case STATE_UNROT3L:
|
|
|
11735 /* Undo ROT3L: "231" -> "123" */
|
|
|
11736 p = fword + sp->ts_fidx;
|
|
|
11737 #ifdef FEAT_MBYTE
|
|
|
11738 if (has_mbyte)
|
|
|
11739 {
|
|
|
11740 n = MB_BYTE2LEN(*p);
|
|
|
11741 n += MB_BYTE2LEN(p[n]);
|
|
|
11742 c = mb_ptr2char(p + n);
|
|
|
11743 tl = MB_BYTE2LEN(p[n]);
|
|
|
11744 mch_memmove(p + tl, p, n);
|
|
|
11745 mb_char2bytes(c, p);
|
|
|
11746 }
|
|
|
11747 else
|
|
|
11748 #endif
|
|
|
11749 {
|
|
|
11750 c = p[2];
|
|
|
11751 p[2] = p[1];
|
|
|
11752 p[1] = *p;
|
|
|
11753 *p = c;
|
|
|
11754 }
|
|
|
11755
|
|
|
11756 /* Rotate three bytes right: "123" -> "312". We change "fword"
|
|
|
11757 * here, it's changed back afterwards at STATE_UNROT3R. */
|
|
|
11758 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
|
|
|
11759 {
|
|
|
11760 go_deeper(stack, depth, SCORE_SWAP3);
|
|
|
11761 #ifdef DEBUG_TRIEWALK
|
|
|
11762 p = fword + sp->ts_fidx;
|
|
|
11763 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c",
|
|
|
11764 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11765 p[0], p[1], p[2]);
|
|
|
11766 #endif
|
|
|
11767 sp->ts_state = STATE_UNROT3R;
|
|
|
11768 ++depth;
|
|
330
|
11769 p = fword + sp->ts_fidx;
|
|
|
11770 #ifdef FEAT_MBYTE
|
|
|
11771 if (has_mbyte)
|
|
|
11772 {
|
|
625
|
11773 n = mb_cptr2len(p);
|
|
|
11774 n += mb_cptr2len(p + n);
|
|
330
|
11775 c = mb_ptr2char(p + n);
|
|
625
|
11776 tl = mb_cptr2len(p + n);
|
|
330
|
11777 mch_memmove(p + tl, p, n);
|
|
|
11778 mb_char2bytes(c, p);
|
|
625
|
11779 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
330
|
11780 }
|
|
|
11781 else
|
|
|
11782 #endif
|
|
|
11783 {
|
|
|
11784 c = p[2];
|
|
|
11785 p[2] = p[1];
|
|
|
11786 p[1] = *p;
|
|
|
11787 *p = c;
|
|
625
|
11788 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
11789 }
|
|
|
11790 }
|
|
|
11791 else
|
|
|
11792 sp->ts_state = STATE_REP_INI;
|
|
|
11793 break;
|
|
|
11794
|
|
|
11795 case STATE_UNROT3R:
|
|
|
11796 /* Undo ROT3R: "312" -> "123" */
|
|
|
11797 p = fword + sp->ts_fidx;
|
|
|
11798 #ifdef FEAT_MBYTE
|
|
|
11799 if (has_mbyte)
|
|
|
11800 {
|
|
|
11801 c = mb_ptr2char(p);
|
|
|
11802 tl = MB_BYTE2LEN(*p);
|
|
|
11803 n = MB_BYTE2LEN(p[tl]);
|
|
|
11804 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
11805 mch_memmove(p, p + tl, n);
|
|
|
11806 mb_char2bytes(c, p + n);
|
|
|
11807 }
|
|
|
11808 else
|
|
|
11809 #endif
|
|
|
11810 {
|
|
|
11811 c = *p;
|
|
|
11812 *p = p[1];
|
|
|
11813 p[1] = p[2];
|
|
|
11814 p[2] = c;
|
|
|
11815 }
|
|
|
11816 /*FALLTHROUGH*/
|
|
|
11817
|
|
|
11818 case STATE_REP_INI:
|
|
|
11819 /* Check if matching with REP items from the .aff file would work.
|
|
|
11820 * Quickly skip if:
|
|
|
11821 * - there are no REP items and we are not in the soundfold trie
|
|
|
11822 * - the score is going to be too high anyway
|
|
|
11823 * - already applied a REP item or swapped here */
|
|
|
11824 if ((lp->lp_replang == NULL && !soundfold)
|
|
|
11825 || sp->ts_score + SCORE_REP >= su->su_maxscore
|
|
|
11826 || sp->ts_fidx < sp->ts_fidxtry)
|
|
|
11827 {
|
|
|
11828 sp->ts_state = STATE_FINAL;
|
|
323
|
11829 break;
|
|
625
|
11830 }
|
|
|
11831
|
|
|
11832 /* Use the first byte to quickly find the first entry that may
|
|
|
11833 * match. If the index is -1 there is none. */
|
|
|
11834 if (soundfold)
|
|
|
11835 sp->ts_curi = slang->sl_repsal_first[fword[sp->ts_fidx]];
|
|
|
11836 else
|
|
|
11837 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
11838
|
|
|
11839 if (sp->ts_curi < 0)
|
|
|
11840 {
|
|
|
11841 sp->ts_state = STATE_FINAL;
|
|
|
11842 break;
|
|
|
11843 }
|
|
|
11844
|
|
|
11845 sp->ts_state = STATE_REP;
|
|
|
11846 /*FALLTHROUGH*/
|
|
|
11847
|
|
|
11848 case STATE_REP:
|
|
|
11849 /* Try matching with REP items from the .aff file. For each match
|
|
|
11850 * replace the characters and check if the resulting word is
|
|
|
11851 * valid. */
|
|
|
11852 p = fword + sp->ts_fidx;
|
|
|
11853
|
|
|
11854 if (soundfold)
|
|
|
11855 gap = &slang->sl_repsal;
|
|
|
11856 else
|
|
503
|
11857 gap = &lp->lp_replang->sl_rep;
|
|
625
|
11858 while (sp->ts_curi < gap->ga_len)
|
|
|
11859 {
|
|
|
11860 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
11861 if (*ftp->ft_from != *p)
|
|
|
11862 {
|
|
|
11863 /* past possible matching entries */
|
|
|
11864 sp->ts_curi = gap->ga_len;
|
|
323
|
11865 break;
|
|
|
11866 }
|
|
625
|
11867 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
11868 && TRY_DEEPER(su, stack, depth, SCORE_REP))
|
|
|
11869 {
|
|
|
11870 go_deeper(stack, depth, SCORE_REP);
|
|
|
11871 #ifdef DEBUG_TRIEWALK
|
|
|
11872 sprintf(changename[depth], "%.*s-%s: replace %s with %s",
|
|
|
11873 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11874 ftp->ft_from, ftp->ft_to);
|
|
|
11875 #endif
|
|
|
11876 /* Need to undo this afterwards. */
|
|
|
11877 sp->ts_state = STATE_REP_UNDO;
|
|
|
11878
|
|
|
11879 /* Change the "from" to the "to" string. */
|
|
|
11880 ++depth;
|
|
|
11881 fl = STRLEN(ftp->ft_from);
|
|
|
11882 tl = STRLEN(ftp->ft_to);
|
|
|
11883 if (fl != tl)
|
|
323
|
11884 {
|
|
625
|
11885 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
|
11886 repextra += tl - fl;
|
|
323
|
11887 }
|
|
625
|
11888 mch_memmove(p, ftp->ft_to, tl);
|
|
|
11889 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
|
11890 #ifdef FEAT_MBYTE
|
|
|
11891 stack[depth].ts_tcharlen = 0;
|
|
|
11892 #endif
|
|
|
11893 break;
|
|
|
11894 }
|
|
|
11895 }
|
|
|
11896
|
|
|
11897 if (sp->ts_curi >= gap->ga_len && sp->ts_state == STATE_REP)
|
|
|
11898 /* No (more) matches. */
|
|
|
11899 sp->ts_state = STATE_FINAL;
|
|
|
11900
|
|
|
11901 break;
|
|
|
11902
|
|
|
11903 case STATE_REP_UNDO:
|
|
|
11904 /* Undo a REP replacement and continue with the next one. */
|
|
|
11905 if (soundfold)
|
|
|
11906 gap = &slang->sl_repsal;
|
|
|
11907 else
|
|
|
11908 gap = &lp->lp_replang->sl_rep;
|
|
|
11909 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1;
|
|
|
11910 fl = STRLEN(ftp->ft_from);
|
|
|
11911 tl = STRLEN(ftp->ft_to);
|
|
|
11912 p = fword + sp->ts_fidx;
|
|
|
11913 if (fl != tl)
|
|
|
11914 {
|
|
|
11915 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
|
11916 repextra -= tl - fl;
|
|
|
11917 }
|
|
|
11918 mch_memmove(p, ftp->ft_from, fl);
|
|
|
11919 sp->ts_state = STATE_REP;
|
|
|
11920 break;
|
|
|
11921
|
|
|
11922 default:
|
|
|
11923 /* Did all possible states at this level, go up one level. */
|
|
|
11924 --depth;
|
|
|
11925
|
|
|
11926 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
|
|
|
11927 {
|
|
|
11928 /* Continue in or go back to the prefix tree. */
|
|
|
11929 byts = pbyts;
|
|
|
11930 idxs = pidxs;
|
|
|
11931 }
|
|
|
11932
|
|
|
11933 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
11934 if (--breakcheckcount == 0)
|
|
|
11935 {
|
|
|
11936 ui_breakcheck();
|
|
|
11937 breakcheckcount = 1000;
|
|
|
11938 }
|
|
|
11939 }
|
|
|
11940 }
|
|
|
11941 }
|
|
|
11942
|
|
|
11943
|
|
|
11944 /*
|
|
|
11945 * Go one level deeper in the tree.
|
|
|
11946 */
|
|
|
11947 static void
|
|
|
11948 go_deeper(stack, depth, score_add)
|
|
323
|
11949 trystate_T *stack;
|
|
|
11950 int depth;
|
|
|
11951 int score_add;
|
|
|
11952 {
|
|
330
|
11953 stack[depth + 1] = stack[depth];
|
|
323
|
11954 stack[depth + 1].ts_state = STATE_START;
|
|
625
|
11955 stack[depth + 1].ts_score = stack[depth].ts_score + score_add;
|
|
323
|
11956 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
485
|
11957 stack[depth + 1].ts_flags = 0;
|
|
323
|
11958 }
|
|
|
11959
|
|
455
|
11960 #ifdef FEAT_MBYTE
|
|
|
11961 /*
|
|
|
11962 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
11963 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
11964 */
|
|
|
11965 static int
|
|
|
11966 nofold_len(fword, flen, word)
|
|
|
11967 char_u *fword;
|
|
|
11968 int flen;
|
|
|
11969 char_u *word;
|
|
|
11970 {
|
|
|
11971 char_u *p;
|
|
|
11972 int i = 0;
|
|
|
11973
|
|
|
11974 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
11975 ++i;
|
|
|
11976 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
11977 --i;
|
|
|
11978 return (int)(p - word);
|
|
|
11979 }
|
|
|
11980 #endif
|
|
|
11981
|
|
323
|
11982 /*
|
|
|
11983 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
11984 * words and put it in "kword".
|
|
|
11985 * Theoretically there could be several keep-case words that result in the
|
|
|
11986 * same case-folded word, but we only find one...
|
|
|
11987 */
|
|
|
11988 static void
|
|
|
11989 find_keepcap_word(slang, fword, kword)
|
|
|
11990 slang_T *slang;
|
|
|
11991 char_u *fword;
|
|
|
11992 char_u *kword;
|
|
|
11993 {
|
|
|
11994 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
11995 int depth;
|
|
324
|
11996 idx_T tryidx;
|
|
323
|
11997
|
|
|
11998 /* The following arrays are used at each depth in the tree. */
|
|
324
|
11999 idx_T arridx[MAXWLEN];
|
|
323
|
12000 int round[MAXWLEN];
|
|
|
12001 int fwordidx[MAXWLEN];
|
|
|
12002 int uwordidx[MAXWLEN];
|
|
|
12003 int kwordlen[MAXWLEN];
|
|
|
12004
|
|
|
12005 int flen, ulen;
|
|
|
12006 int l;
|
|
|
12007 int len;
|
|
|
12008 int c;
|
|
324
|
12009 idx_T lo, hi, m;
|
|
323
|
12010 char_u *p;
|
|
|
12011 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
12012 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
12013
|
|
|
12014 if (byts == NULL)
|
|
|
12015 {
|
|
|
12016 /* array is empty: "cannot happen" */
|
|
|
12017 *kword = NUL;
|
|
|
12018 return;
|
|
|
12019 }
|
|
|
12020
|
|
|
12021 /* Make an all-cap version of "fword". */
|
|
|
12022 allcap_copy(fword, uword);
|
|
|
12023
|
|
|
12024 /*
|
|
|
12025 * Each character needs to be tried both case-folded and upper-case.
|
|
|
12026 * All this gets very complicated if we keep in mind that changing case
|
|
|
12027 * may change the byte length of a multi-byte character...
|
|
|
12028 */
|
|
|
12029 depth = 0;
|
|
|
12030 arridx[0] = 0;
|
|
|
12031 round[0] = 0;
|
|
|
12032 fwordidx[0] = 0;
|
|
|
12033 uwordidx[0] = 0;
|
|
|
12034 kwordlen[0] = 0;
|
|
|
12035 while (depth >= 0)
|
|
|
12036 {
|
|
|
12037 if (fword[fwordidx[depth]] == NUL)
|
|
|
12038 {
|
|
|
12039 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
12040 * here we have found a match. */
|
|
|
12041 if (byts[arridx[depth] + 1] == 0)
|
|
|
12042 {
|
|
|
12043 kword[kwordlen[depth]] = NUL;
|
|
|
12044 return;
|
|
|
12045 }
|
|
|
12046
|
|
|
12047 /* kword is getting too long, continue one level up */
|
|
|
12048 --depth;
|
|
|
12049 }
|
|
|
12050 else if (++round[depth] > 2)
|
|
|
12051 {
|
|
|
12052 /* tried both fold-case and upper-case character, continue one
|
|
|
12053 * level up */
|
|
|
12054 --depth;
|
|
|
12055 }
|
|
|
12056 else
|
|
|
12057 {
|
|
|
12058 /*
|
|
|
12059 * round[depth] == 1: Try using the folded-case character.
|
|
|
12060 * round[depth] == 2: Try using the upper-case character.
|
|
|
12061 */
|
|
|
12062 #ifdef FEAT_MBYTE
|
|
|
12063 if (has_mbyte)
|
|
|
12064 {
|
|
474
|
12065 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
12066 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
12067 }
|
|
|
12068 else
|
|
|
12069 #endif
|
|
|
12070 ulen = flen = 1;
|
|
|
12071 if (round[depth] == 1)
|
|
|
12072 {
|
|
|
12073 p = fword + fwordidx[depth];
|
|
|
12074 l = flen;
|
|
|
12075 }
|
|
|
12076 else
|
|
|
12077 {
|
|
|
12078 p = uword + uwordidx[depth];
|
|
|
12079 l = ulen;
|
|
|
12080 }
|
|
|
12081
|
|
|
12082 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
12083 {
|
|
|
12084 /* Perform a binary search in the list of accepted bytes. */
|
|
|
12085 len = byts[tryidx++];
|
|
|
12086 c = *p++;
|
|
|
12087 lo = tryidx;
|
|
|
12088 hi = tryidx + len - 1;
|
|
|
12089 while (lo < hi)
|
|
|
12090 {
|
|
|
12091 m = (lo + hi) / 2;
|
|
|
12092 if (byts[m] > c)
|
|
|
12093 hi = m - 1;
|
|
|
12094 else if (byts[m] < c)
|
|
|
12095 lo = m + 1;
|
|
|
12096 else
|
|
|
12097 {
|
|
|
12098 lo = hi = m;
|
|
|
12099 break;
|
|
|
12100 }
|
|
|
12101 }
|
|
|
12102
|
|
|
12103 /* Stop if there is no matching byte. */
|
|
|
12104 if (hi < lo || byts[lo] != c)
|
|
|
12105 break;
|
|
|
12106
|
|
|
12107 /* Continue at the child (if there is one). */
|
|
|
12108 tryidx = idxs[lo];
|
|
|
12109 }
|
|
|
12110
|
|
|
12111 if (l == 0)
|
|
|
12112 {
|
|
|
12113 /*
|
|
|
12114 * Found the matching char. Copy it to "kword" and go a
|
|
|
12115 * level deeper.
|
|
|
12116 */
|
|
|
12117 if (round[depth] == 1)
|
|
|
12118 {
|
|
|
12119 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
12120 flen);
|
|
|
12121 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
12122 }
|
|
|
12123 else
|
|
|
12124 {
|
|
|
12125 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
12126 ulen);
|
|
|
12127 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
12128 }
|
|
|
12129 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
12130 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
12131
|
|
|
12132 ++depth;
|
|
|
12133 arridx[depth] = tryidx;
|
|
|
12134 round[depth] = 0;
|
|
|
12135 }
|
|
|
12136 }
|
|
|
12137 }
|
|
|
12138
|
|
|
12139 /* Didn't find it: "cannot happen". */
|
|
|
12140 *kword = NUL;
|
|
|
12141 }
|
|
|
12142
|
|
|
12143 /*
|
|
344
|
12144 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
12145 * su->su_sga.
|
|
|
12146 */
|
|
|
12147 static void
|
|
|
12148 score_comp_sal(su)
|
|
|
12149 suginfo_T *su;
|
|
|
12150 {
|
|
|
12151 langp_T *lp;
|
|
|
12152 char_u badsound[MAXWLEN];
|
|
|
12153 int i;
|
|
|
12154 suggest_T *stp;
|
|
|
12155 suggest_T *sstp;
|
|
|
12156 int score;
|
|
500
|
12157 int lpi;
|
|
344
|
12158
|
|
|
12159 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
12160 return;
|
|
|
12161
|
|
|
12162 /* Use the sound-folding of the first language that supports it. */
|
|
503
|
12163 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
12164 {
|
|
|
12165 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
344
|
12166 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
12167 {
|
|
|
12168 /* soundfold the bad word */
|
|
375
|
12169 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
12170
|
|
|
12171 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
12172 {
|
|
|
12173 stp = &SUG(su->su_ga, i);
|
|
|
12174
|
|
351
|
12175 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
12176 * sound-a-like score. */
|
|
|
12177 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
12178 if (score < SCORE_MAXMAX)
|
|
|
12179 {
|
|
|
12180 /* Add the suggestion. */
|
|
|
12181 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
12182 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
12183 if (sstp->st_word != NULL)
|
|
|
12184 {
|
|
625
|
12185 sstp->st_wordlen = stp->st_wordlen;
|
|
344
|
12186 sstp->st_score = score;
|
|
|
12187 sstp->st_altscore = 0;
|
|
|
12188 sstp->st_orglen = stp->st_orglen;
|
|
|
12189 ++su->su_sga.ga_len;
|
|
|
12190 }
|
|
|
12191 }
|
|
|
12192 }
|
|
|
12193 break;
|
|
|
12194 }
|
|
500
|
12195 }
|
|
344
|
12196 }
|
|
|
12197
|
|
|
12198 /*
|
|
|
12199 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
12200 * They are intwined.
|
|
|
12201 */
|
|
|
12202 static void
|
|
|
12203 score_combine(su)
|
|
|
12204 suginfo_T *su;
|
|
|
12205 {
|
|
|
12206 int i;
|
|
|
12207 int j;
|
|
|
12208 garray_T ga;
|
|
|
12209 garray_T *gap;
|
|
|
12210 langp_T *lp;
|
|
|
12211 suggest_T *stp;
|
|
|
12212 char_u *p;
|
|
|
12213 char_u badsound[MAXWLEN];
|
|
|
12214 int round;
|
|
500
|
12215 int lpi;
|
|
625
|
12216 slang_T *slang = NULL;
|
|
344
|
12217
|
|
|
12218 /* Add the alternate score to su_ga. */
|
|
503
|
12219 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
12220 {
|
|
|
12221 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
344
|
12222 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
12223 {
|
|
|
12224 /* soundfold the bad word */
|
|
625
|
12225 slang = lp->lp_slang;
|
|
|
12226 spell_soundfold(slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
12227
|
|
|
12228 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
12229 {
|
|
|
12230 stp = &SUG(su->su_ga, i);
|
|
625
|
12231 stp->st_altscore = stp_sal_score(stp, su, slang, badsound);
|
|
344
|
12232 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
12233 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
12234 else
|
|
|
12235 stp->st_score = (stp->st_score * 3
|
|
|
12236 + stp->st_altscore) / 4;
|
|
|
12237 stp->st_salscore = FALSE;
|
|
|
12238 }
|
|
|
12239 break;
|
|
|
12240 }
|
|
|
12241 }
|
|
|
12242
|
|
625
|
12243 if (slang == NULL) /* just in case */
|
|
|
12244 return;
|
|
|
12245
|
|
344
|
12246 /* Add the alternate score to su_sga. */
|
|
|
12247 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
12248 {
|
|
|
12249 stp = &SUG(su->su_sga, i);
|
|
625
|
12250 stp->st_altscore = spell_edit_score(slang,
|
|
|
12251 su->su_badword, stp->st_word);
|
|
344
|
12252 if (stp->st_score == SCORE_MAXMAX)
|
|
|
12253 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
12254 else
|
|
|
12255 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
12256 stp->st_salscore = TRUE;
|
|
|
12257 }
|
|
|
12258
|
|
625
|
12259 /* Remove bad suggestions, sort the suggestions and truncate at "maxcount"
|
|
|
12260 * for both lists. */
|
|
|
12261 check_suggestions(su, &su->su_ga);
|
|
344
|
12262 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
625
|
12263 check_suggestions(su, &su->su_sga);
|
|
344
|
12264 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
12265
|
|
|
12266 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
12267 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
12268 return;
|
|
|
12269
|
|
|
12270 stp = &SUG(ga, 0);
|
|
|
12271 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
12272 {
|
|
|
12273 /* round 1: get a suggestion from su_ga
|
|
|
12274 * round 2: get a suggestion from su_sga */
|
|
|
12275 for (round = 1; round <= 2; ++round)
|
|
|
12276 {
|
|
|
12277 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
12278 if (i < gap->ga_len)
|
|
|
12279 {
|
|
|
12280 /* Don't add a word if it's already there. */
|
|
|
12281 p = SUG(*gap, i).st_word;
|
|
|
12282 for (j = 0; j < ga.ga_len; ++j)
|
|
|
12283 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
12284 break;
|
|
|
12285 if (j == ga.ga_len)
|
|
|
12286 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
12287 else
|
|
|
12288 vim_free(p);
|
|
|
12289 }
|
|
|
12290 }
|
|
|
12291 }
|
|
|
12292
|
|
|
12293 ga_clear(&su->su_ga);
|
|
|
12294 ga_clear(&su->su_sga);
|
|
|
12295
|
|
|
12296 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
12297 if (ga.ga_len > su->su_maxcount)
|
|
|
12298 {
|
|
|
12299 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
12300 vim_free(stp[i].st_word);
|
|
|
12301 ga.ga_len = su->su_maxcount;
|
|
|
12302 }
|
|
|
12303
|
|
|
12304 su->su_ga = ga;
|
|
|
12305 }
|
|
|
12306
|
|
|
12307 /*
|
|
351
|
12308 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
12309 * badword.
|
|
|
12310 */
|
|
|
12311 static int
|
|
|
12312 stp_sal_score(stp, su, slang, badsound)
|
|
|
12313 suggest_T *stp;
|
|
|
12314 suginfo_T *su;
|
|
|
12315 slang_T *slang;
|
|
|
12316 char_u *badsound; /* sound-folded badword */
|
|
|
12317 {
|
|
|
12318 char_u *p;
|
|
534
|
12319 char_u *pbad;
|
|
|
12320 char_u *pgood;
|
|
351
|
12321 char_u badsound2[MAXWLEN];
|
|
|
12322 char_u fword[MAXWLEN];
|
|
|
12323 char_u goodsound[MAXWLEN];
|
|
534
|
12324 char_u goodword[MAXWLEN];
|
|
|
12325 int lendiff;
|
|
|
12326
|
|
|
12327 lendiff = (int)(su->su_badlen - stp->st_orglen);
|
|
|
12328 if (lendiff >= 0)
|
|
|
12329 pbad = badsound;
|
|
351
|
12330 else
|
|
|
12331 {
|
|
|
12332 /* soundfold the bad word with more characters following */
|
|
|
12333 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
12334
|
|
|
12335 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
12336 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
12337 * removing the space. Don't do it when the good word also contains a
|
|
|
12338 * space. */
|
|
|
12339 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
12340 && *skiptowhite(stp->st_word) == NUL)
|
|
|
12341 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
12342 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
12343
|
|
375
|
12344 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
534
|
12345 pbad = badsound2;
|
|
|
12346 }
|
|
|
12347
|
|
|
12348 if (lendiff > 0)
|
|
|
12349 {
|
|
|
12350 /* Add part of the bad word to the good word, so that we soundfold
|
|
|
12351 * what replaces the bad word. */
|
|
|
12352 STRCPY(goodword, stp->st_word);
|
|
625
|
12353 vim_strncpy(goodword + stp->st_wordlen,
|
|
|
12354 su->su_badptr + su->su_badlen - lendiff, lendiff);
|
|
534
|
12355 pgood = goodword;
|
|
|
12356 }
|
|
|
12357 else
|
|
|
12358 pgood = stp->st_word;
|
|
351
|
12359
|
|
375
|
12360 /* Sound-fold the word and compute the score for the difference. */
|
|
534
|
12361 spell_soundfold(slang, pgood, FALSE, goodsound);
|
|
|
12362
|
|
|
12363 return soundalike_score(goodsound, pbad);
|
|
351
|
12364 }
|
|
|
12365
|
|
625
|
12366 /* structure used to store soundfolded words that add_sound_suggest() has
|
|
|
12367 * handled already. */
|
|
|
12368 typedef struct
|
|
|
12369 {
|
|
|
12370 short sft_score; /* lowest score used */
|
|
|
12371 char_u sft_word[1]; /* soundfolded word, actually longer */
|
|
|
12372 } sftword_T;
|
|
|
12373
|
|
|
12374 static sftword_T dumsft;
|
|
|
12375 #define HIKEY2SFT(p) ((sftword_T *)(p - (dumsft.sft_word - (char_u *)&dumsft)))
|
|
|
12376 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key)
|
|
|
12377
|
|
|
12378 /*
|
|
|
12379 * Prepare for calling suggest_try_soundalike().
|
|
|
12380 */
|
|
|
12381 static void
|
|
|
12382 suggest_try_soundalike_prep()
|
|
|
12383 {
|
|
|
12384 langp_T *lp;
|
|
|
12385 int lpi;
|
|
|
12386 slang_T *slang;
|
|
|
12387
|
|
|
12388 /* Do this for all languages that support sound folding and for which a
|
|
|
12389 * .sug file has been loaded. */
|
|
|
12390 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
12391 {
|
|
|
12392 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
12393 slang = lp->lp_slang;
|
|
|
12394 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL)
|
|
|
12395 /* prepare the hashtable used by add_sound_suggest() */
|
|
|
12396 hash_init(&slang->sl_sounddone);
|
|
|
12397 }
|
|
|
12398 }
|
|
|
12399
|
|
351
|
12400 /*
|
|
323
|
12401 * Find suggestions by comparing the word in a sound-a-like form.
|
|
503
|
12402 * Note: This doesn't support postponed prefixes.
|
|
323
|
12403 */
|
|
|
12404 static void
|
|
346
|
12405 suggest_try_soundalike(su)
|
|
323
|
12406 suginfo_T *su;
|
|
|
12407 {
|
|
|
12408 char_u salword[MAXWLEN];
|
|
|
12409 langp_T *lp;
|
|
500
|
12410 int lpi;
|
|
503
|
12411 slang_T *slang;
|
|
344
|
12412
|
|
625
|
12413 /* Do this for all languages that support sound folding and for which a
|
|
|
12414 * .sug file has been loaded. */
|
|
|
12415 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
12416 {
|
|
|
12417 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
12418 slang = lp->lp_slang;
|
|
|
12419 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL)
|
|
|
12420 {
|
|
|
12421 /* soundfold the bad word */
|
|
|
12422 spell_soundfold(slang, su->su_fbadword, TRUE, salword);
|
|
|
12423
|
|
|
12424 /* try all kinds of inserts/deletes/swaps/etc. */
|
|
|
12425 /* TODO: also soundfold the next words, so that we can try joining
|
|
|
12426 * and splitting */
|
|
|
12427 suggest_trie_walk(su, lp, salword, TRUE);
|
|
|
12428 }
|
|
|
12429 }
|
|
|
12430 }
|
|
|
12431
|
|
|
12432 /*
|
|
|
12433 * Finish up after calling suggest_try_soundalike().
|
|
|
12434 */
|
|
|
12435 static void
|
|
|
12436 suggest_try_soundalike_finish()
|
|
|
12437 {
|
|
|
12438 langp_T *lp;
|
|
|
12439 int lpi;
|
|
|
12440 slang_T *slang;
|
|
|
12441 int todo;
|
|
|
12442 hashitem_T *hi;
|
|
|
12443
|
|
|
12444 /* Do this for all languages that support sound folding and for which a
|
|
|
12445 * .sug file has been loaded. */
|
|
503
|
12446 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
12447 {
|
|
|
12448 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
12449 slang = lp->lp_slang;
|
|
625
|
12450 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL)
|
|
|
12451 {
|
|
|
12452 /* Free the info about handled words. */
|
|
|
12453 todo = slang->sl_sounddone.ht_used;
|
|
|
12454 for (hi = slang->sl_sounddone.ht_array; todo > 0; ++hi)
|
|
|
12455 if (!HASHITEM_EMPTY(hi))
|
|
|
12456 {
|
|
|
12457 vim_free(HI2SFT(hi));
|
|
|
12458 --todo;
|
|
|
12459 }
|
|
|
12460 hash_clear(&slang->sl_sounddone);
|
|
|
12461 }
|
|
|
12462 }
|
|
|
12463 }
|
|
|
12464
|
|
|
12465 /*
|
|
|
12466 * A match with a soundfolded word is found. Add the good word(s) that
|
|
|
12467 * produce this soundfolded word.
|
|
|
12468 */
|
|
|
12469 static void
|
|
|
12470 add_sound_suggest(su, goodword, score, lp)
|
|
|
12471 suginfo_T *su;
|
|
|
12472 char_u *goodword;
|
|
|
12473 int score; /* soundfold score */
|
|
|
12474 langp_T *lp;
|
|
|
12475 {
|
|
|
12476 slang_T *slang = lp->lp_slang; /* language for sound folding */
|
|
|
12477 int sfwordnr;
|
|
|
12478 char_u *nrline;
|
|
|
12479 int orgnr;
|
|
|
12480 char_u theword[MAXWLEN];
|
|
|
12481 int i;
|
|
|
12482 int wlen;
|
|
|
12483 char_u *byts;
|
|
|
12484 idx_T *idxs;
|
|
|
12485 int n;
|
|
|
12486 int wordcount;
|
|
|
12487 int wc;
|
|
|
12488 int goodscore;
|
|
|
12489 hash_T hash;
|
|
|
12490 hashitem_T *hi;
|
|
|
12491 sftword_T *sft;
|
|
|
12492 int bc, gc;
|
|
|
12493 int limit;
|
|
|
12494
|
|
|
12495 /*
|
|
|
12496 * It's very well possible that the same soundfold word is found several
|
|
|
12497 * times with different scores. Since the following is quite slow only do
|
|
|
12498 * the words that have a better score than before. Use a hashtable to
|
|
|
12499 * remember the words that have been done.
|
|
|
12500 */
|
|
|
12501 hash = hash_hash(goodword);
|
|
|
12502 hi = hash_lookup(&slang->sl_sounddone, goodword, hash);
|
|
|
12503 if (HASHITEM_EMPTY(hi))
|
|
|
12504 {
|
|
|
12505 sft = (sftword_T *)alloc(sizeof(sftword_T) + STRLEN(goodword));
|
|
|
12506 if (sft != NULL)
|
|
|
12507 {
|
|
|
12508 sft->sft_score = score;
|
|
|
12509 STRCPY(sft->sft_word, goodword);
|
|
|
12510 hash_add_item(&slang->sl_sounddone, hi, sft->sft_word, hash);
|
|
|
12511 }
|
|
|
12512 }
|
|
|
12513 else
|
|
|
12514 {
|
|
|
12515 sft = HI2SFT(hi);
|
|
|
12516 if (score >= sft->sft_score)
|
|
|
12517 return;
|
|
|
12518 sft->sft_score = score;
|
|
|
12519 }
|
|
|
12520
|
|
|
12521 /*
|
|
|
12522 * Find the word nr in the soundfold tree.
|
|
|
12523 */
|
|
|
12524 sfwordnr = soundfold_find(slang, goodword);
|
|
|
12525 if (sfwordnr < 0)
|
|
|
12526 {
|
|
|
12527 EMSG2(_(e_intern2), "add_sound_suggest()");
|
|
|
12528 return;
|
|
|
12529 }
|
|
|
12530
|
|
|
12531 /*
|
|
|
12532 * go over the list of good words that produce this soundfold word
|
|
|
12533 */
|
|
|
12534 nrline = ml_get_buf(slang->sl_sugbuf, (linenr_T)(sfwordnr + 1), FALSE);
|
|
|
12535 orgnr = 0;
|
|
|
12536 while (*nrline != NUL)
|
|
|
12537 {
|
|
|
12538 /* The wordnr was stored in a minimal nr of bytes as an offset to the
|
|
|
12539 * previous wordnr. */
|
|
|
12540 orgnr += bytes2offset(&nrline);
|
|
|
12541
|
|
|
12542 byts = slang->sl_fbyts;
|
|
|
12543 idxs = slang->sl_fidxs;
|
|
|
12544
|
|
|
12545 /* Lookup the word "orgnr" one of the two tries. */
|
|
|
12546 n = 0;
|
|
|
12547 wlen = 0;
|
|
|
12548 wordcount = 0;
|
|
|
12549 for (;;)
|
|
|
12550 {
|
|
|
12551 i = 1;
|
|
|
12552 if (wordcount == orgnr && byts[n + 1] == NUL)
|
|
|
12553 break; /* found end of word */
|
|
|
12554
|
|
|
12555 if (byts[n + 1] == NUL)
|
|
|
12556 ++wordcount;
|
|
|
12557
|
|
|
12558 /* skip over the NUL bytes */
|
|
|
12559 for ( ; byts[n + i] == NUL; ++i)
|
|
|
12560 if (i > byts[n]) /* safety check */
|
|
|
12561 {
|
|
|
12562 STRCPY(theword + wlen, "BAD");
|
|
|
12563 goto badword;
|
|
|
12564 }
|
|
|
12565
|
|
|
12566 /* One of the siblings must have the word. */
|
|
|
12567 for ( ; i < byts[n]; ++i)
|
|
|
12568 {
|
|
|
12569 wc = idxs[idxs[n + i]]; /* nr of words under this byte */
|
|
|
12570 if (wordcount + wc > orgnr)
|
|
|
12571 break;
|
|
|
12572 wordcount += wc;
|
|
|
12573 }
|
|
|
12574
|
|
|
12575 theword[wlen++] = byts[n + i];
|
|
|
12576 n = idxs[n + i];
|
|
|
12577 }
|
|
|
12578 badword:
|
|
|
12579 theword[wlen] = NUL;
|
|
|
12580
|
|
|
12581 /* Go over the possible flags and regions. */
|
|
|
12582 for (; i <= byts[n] && byts[n + i] == NUL; ++i)
|
|
|
12583 {
|
|
|
12584 char_u cword[MAXWLEN];
|
|
|
12585 char_u *p;
|
|
|
12586 int flags = (int)idxs[n + i];
|
|
|
12587
|
|
|
12588 if (flags & WF_KEEPCAP)
|
|
|
12589 {
|
|
|
12590 /* Must find the word in the keep-case tree. */
|
|
|
12591 find_keepcap_word(slang, theword, cword);
|
|
|
12592 p = cword;
|
|
|
12593 }
|
|
|
12594 else
|
|
|
12595 {
|
|
|
12596 flags |= su->su_badflags;
|
|
|
12597 if ((flags & WF_CAPMASK) != 0)
|
|
|
12598 {
|
|
|
12599 /* Need to fix case according to "flags". */
|
|
|
12600 make_case_word(theword, cword, flags);
|
|
|
12601 p = cword;
|
|
323
|
12602 }
|
|
|
12603 else
|
|
625
|
12604 p = theword;
|
|
|
12605 }
|
|
|
12606
|
|
|
12607 /* Add the suggestion. */
|
|
|
12608 if (sps_flags & SPS_DOUBLE)
|
|
|
12609 {
|
|
|
12610 /* Add the suggestion if the score isn't too bad. */
|
|
|
12611 if (score <= su->su_maxscore)
|
|
|
12612 add_suggestion(su, &su->su_sga, p, su->su_badlen,
|
|
|
12613 score, 0, FALSE, slang, FALSE);
|
|
|
12614 }
|
|
|
12615 else
|
|
|
12616 {
|
|
|
12617 /* Add a penalty for words in another region. */
|
|
|
12618 if ((flags & WF_REGION)
|
|
|
12619 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
|
12620 goodscore = SCORE_REGION;
|
|
|
12621 else
|
|
|
12622 goodscore = 0;
|
|
|
12623
|
|
|
12624 /* Add a small penalty for changing the first letter from
|
|
|
12625 * lower to upper case. Helps for "tath" -> "Kath", which is
|
|
|
12626 * less common thatn "tath" -> "path". Don't do it when the
|
|
|
12627 * letter is the same, that has already been counted. */
|
|
|
12628 gc = PTR2CHAR(p);
|
|
|
12629 if (SPELL_ISUPPER(gc))
|
|
|
12630 {
|
|
|
12631 bc = PTR2CHAR(su->su_badword);
|
|
|
12632 if (!SPELL_ISUPPER(bc)
|
|
|
12633 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc))
|
|
|
12634 goodscore += SCORE_ICASE / 2;
|
|
|
12635 }
|
|
|
12636
|
|
|
12637 /* Compute the score for the good word. This only does letter
|
|
|
12638 * insert/delete/swap/replace. REP items are not considered,
|
|
|
12639 * which may make the score a bit higher.
|
|
|
12640 * Use a limit for the score to make it work faster. Use
|
|
|
12641 * MAXSCORE(), because RESCORE() will change the score.
|
|
|
12642 * If the limit is very high then the iterative method is
|
|
|
12643 * inefficient, using an array is quicker. */
|
|
|
12644 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score);
|
|
|
12645 if (limit > SCORE_LIMITMAX)
|
|
|
12646 goodscore += spell_edit_score(slang, su->su_badword, p);
|
|
|
12647 else
|
|
|
12648 goodscore += spell_edit_score_limit(slang, su->su_badword,
|
|
|
12649 p, limit);
|
|
|
12650
|
|
|
12651 /* When going over the limit don't bother to do the rest. */
|
|
|
12652 if (goodscore < SCORE_MAXMAX)
|
|
|
12653 {
|
|
|
12654 /* Give a bonus to words seen before. */
|
|
|
12655 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE);
|
|
|
12656
|
|
|
12657 /* Add the suggestion if the score isn't too bad. */
|
|
|
12658 goodscore = RESCORE(goodscore, score);
|
|
|
12659 if (goodscore <= su->su_sfmaxscore)
|
|
|
12660 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
12661 goodscore, score, TRUE, slang, TRUE);
|
|
|
12662 }
|
|
|
12663 }
|
|
|
12664 }
|
|
|
12665 /* smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr); */
|
|
|
12666 }
|
|
|
12667 }
|
|
|
12668
|
|
|
12669 /*
|
|
|
12670 * Find word "word" in fold-case tree for "slang" and return the word number.
|
|
|
12671 */
|
|
|
12672 static int
|
|
|
12673 soundfold_find(slang, word)
|
|
|
12674 slang_T *slang;
|
|
|
12675 char_u *word;
|
|
|
12676 {
|
|
|
12677 idx_T arridx = 0;
|
|
|
12678 int len;
|
|
|
12679 int wlen = 0;
|
|
|
12680 int c;
|
|
|
12681 char_u *ptr = word;
|
|
|
12682 char_u *byts;
|
|
|
12683 idx_T *idxs;
|
|
|
12684 int wordnr = 0;
|
|
|
12685
|
|
|
12686 byts = slang->sl_sbyts;
|
|
|
12687 idxs = slang->sl_sidxs;
|
|
|
12688
|
|
|
12689 for (;;)
|
|
|
12690 {
|
|
|
12691 /* First byte is the number of possible bytes. */
|
|
|
12692 len = byts[arridx++];
|
|
|
12693
|
|
|
12694 /* If the first possible byte is a zero the word could end here.
|
|
|
12695 * If the word ends we found the word. If not skip the NUL bytes. */
|
|
|
12696 c = ptr[wlen];
|
|
|
12697 if (byts[arridx] == NUL)
|
|
|
12698 {
|
|
|
12699 if (c == NUL)
|
|
|
12700 break;
|
|
|
12701
|
|
|
12702 /* Skip over the zeros, there can be several. */
|
|
|
12703 while (len > 0 && byts[arridx] == NUL)
|
|
|
12704 {
|
|
|
12705 ++arridx;
|
|
|
12706 --len;
|
|
|
12707 }
|
|
|
12708 if (len == 0)
|
|
|
12709 return -1; /* no children, word should have ended here */
|
|
|
12710 ++wordnr;
|
|
|
12711 }
|
|
|
12712
|
|
|
12713 /* If the word ends we didn't find it. */
|
|
|
12714 if (c == NUL)
|
|
|
12715 return -1;
|
|
|
12716
|
|
|
12717 /* Perform a binary search in the list of accepted bytes. */
|
|
|
12718 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
12719 c = ' ';
|
|
|
12720 while (byts[arridx] < c)
|
|
|
12721 {
|
|
|
12722 /* The word count is in the first idxs[] entry of the child. */
|
|
|
12723 wordnr += idxs[idxs[arridx]];
|
|
|
12724 ++arridx;
|
|
|
12725 if (--len == 0) /* end of the bytes, didn't find it */
|
|
|
12726 return -1;
|
|
|
12727 }
|
|
|
12728 if (byts[arridx] != c) /* didn't find the byte */
|
|
|
12729 return -1;
|
|
|
12730
|
|
|
12731 /* Continue at the child (if there is one). */
|
|
|
12732 arridx = idxs[arridx];
|
|
|
12733 ++wlen;
|
|
|
12734
|
|
|
12735 /* One space in the good word may stand for several spaces in the
|
|
|
12736 * checked word. */
|
|
|
12737 if (c == ' ')
|
|
|
12738 while (ptr[wlen] == ' ' || ptr[wlen] == TAB)
|
|
|
12739 ++wlen;
|
|
|
12740 }
|
|
|
12741
|
|
|
12742 return wordnr;
|
|
323
|
12743 }
|
|
|
12744
|
|
|
12745 /*
|
|
324
|
12746 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
12747 */
|
|
|
12748 static void
|
|
|
12749 make_case_word(fword, cword, flags)
|
|
|
12750 char_u *fword;
|
|
|
12751 char_u *cword;
|
|
|
12752 int flags;
|
|
|
12753 {
|
|
|
12754 if (flags & WF_ALLCAP)
|
|
|
12755 /* Make it all upper-case */
|
|
|
12756 allcap_copy(fword, cword);
|
|
|
12757 else if (flags & WF_ONECAP)
|
|
|
12758 /* Make the first letter upper-case */
|
|
324
|
12759 onecap_copy(fword, cword, TRUE);
|
|
323
|
12760 else
|
|
|
12761 /* Use goodword as-is. */
|
|
|
12762 STRCPY(cword, fword);
|
|
|
12763 }
|
|
|
12764
|
|
330
|
12765 /*
|
|
|
12766 * Use map string "map" for languages "lp".
|
|
|
12767 */
|
|
|
12768 static void
|
|
|
12769 set_map_str(lp, map)
|
|
|
12770 slang_T *lp;
|
|
|
12771 char_u *map;
|
|
|
12772 {
|
|
|
12773 char_u *p;
|
|
|
12774 int headc = 0;
|
|
|
12775 int c;
|
|
|
12776 int i;
|
|
|
12777
|
|
|
12778 if (*map == NUL)
|
|
|
12779 {
|
|
|
12780 lp->sl_has_map = FALSE;
|
|
|
12781 return;
|
|
|
12782 }
|
|
|
12783 lp->sl_has_map = TRUE;
|
|
|
12784
|
|
625
|
12785 /* Init the array and hash tables empty. */
|
|
330
|
12786 for (i = 0; i < 256; ++i)
|
|
|
12787 lp->sl_map_array[i] = 0;
|
|
|
12788 #ifdef FEAT_MBYTE
|
|
|
12789 hash_init(&lp->sl_map_hash);
|
|
|
12790 #endif
|
|
|
12791
|
|
|
12792 /*
|
|
|
12793 * The similar characters are stored separated with slashes:
|
|
|
12794 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
12795 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
12796 */
|
|
|
12797 for (p = map; *p != NUL; )
|
|
|
12798 {
|
|
|
12799 #ifdef FEAT_MBYTE
|
|
474
|
12800 c = mb_cptr2char_adv(&p);
|
|
330
|
12801 #else
|
|
|
12802 c = *p++;
|
|
|
12803 #endif
|
|
|
12804 if (c == '/')
|
|
|
12805 headc = 0;
|
|
|
12806 else
|
|
|
12807 {
|
|
|
12808 if (headc == 0)
|
|
|
12809 headc = c;
|
|
|
12810
|
|
|
12811 #ifdef FEAT_MBYTE
|
|
|
12812 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
12813 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
12814 * a NUL. */
|
|
|
12815 if (c >= 256)
|
|
|
12816 {
|
|
|
12817 int cl = mb_char2len(c);
|
|
|
12818 int headcl = mb_char2len(headc);
|
|
|
12819 char_u *b;
|
|
|
12820 hash_T hash;
|
|
|
12821 hashitem_T *hi;
|
|
|
12822
|
|
|
12823 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
12824 if (b == NULL)
|
|
|
12825 return;
|
|
|
12826 mb_char2bytes(c, b);
|
|
|
12827 b[cl] = NUL;
|
|
|
12828 mb_char2bytes(headc, b + cl + 1);
|
|
|
12829 b[cl + 1 + headcl] = NUL;
|
|
|
12830 hash = hash_hash(b);
|
|
|
12831 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
12832 if (HASHITEM_EMPTY(hi))
|
|
|
12833 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
12834 else
|
|
|
12835 {
|
|
|
12836 /* This should have been checked when generating the .spl
|
|
|
12837 * file. */
|
|
626
|
12838 EMSG(_("E783: duplicate char in MAP entry"));
|
|
330
|
12839 vim_free(b);
|
|
|
12840 }
|
|
|
12841 }
|
|
|
12842 else
|
|
|
12843 #endif
|
|
|
12844 lp->sl_map_array[c] = headc;
|
|
|
12845 }
|
|
|
12846 }
|
|
|
12847 }
|
|
|
12848
|
|
323
|
12849 /*
|
|
|
12850 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
12851 * lines in the .aff file.
|
|
|
12852 */
|
|
|
12853 static int
|
|
|
12854 similar_chars(slang, c1, c2)
|
|
|
12855 slang_T *slang;
|
|
|
12856 int c1;
|
|
|
12857 int c2;
|
|
|
12858 {
|
|
330
|
12859 int m1, m2;
|
|
|
12860 #ifdef FEAT_MBYTE
|
|
|
12861 char_u buf[MB_MAXBYTES];
|
|
|
12862 hashitem_T *hi;
|
|
|
12863
|
|
|
12864 if (c1 >= 256)
|
|
|
12865 {
|
|
|
12866 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
12867 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
12868 if (HASHITEM_EMPTY(hi))
|
|
|
12869 m1 = 0;
|
|
|
12870 else
|
|
|
12871 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
12872 }
|
|
|
12873 else
|
|
|
12874 #endif
|
|
|
12875 m1 = slang->sl_map_array[c1];
|
|
|
12876 if (m1 == 0)
|
|
323
|
12877 return FALSE;
|
|
330
|
12878
|
|
|
12879
|
|
|
12880 #ifdef FEAT_MBYTE
|
|
|
12881 if (c2 >= 256)
|
|
|
12882 {
|
|
|
12883 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
12884 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
12885 if (HASHITEM_EMPTY(hi))
|
|
|
12886 m2 = 0;
|
|
|
12887 else
|
|
|
12888 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
12889 }
|
|
|
12890 else
|
|
|
12891 #endif
|
|
|
12892 m2 = slang->sl_map_array[c2];
|
|
|
12893
|
|
|
12894 return m1 == m2;
|
|
323
|
12895 }
|
|
|
12896
|
|
|
12897 /*
|
|
|
12898 * Add a suggestion to the list of suggestions.
|
|
625
|
12899 * For a suggestion that is already in the list the lowest score is remembered.
|
|
|
12900 */
|
|
|
12901 static void
|
|
|
12902 add_suggestion(su, gap, goodword, badlenarg, score, altscore, had_bonus,
|
|
|
12903 slang, maxsf)
|
|
323
|
12904 suginfo_T *su;
|
|
625
|
12905 garray_T *gap; /* either su_ga or su_sga */
|
|
323
|
12906 char_u *goodword;
|
|
534
|
12907 int badlenarg; /* len of bad word replaced with "goodword" */
|
|
324
|
12908 int score;
|
|
351
|
12909 int altscore;
|
|
344
|
12910 int had_bonus; /* value for st_had_bonus */
|
|
503
|
12911 slang_T *slang; /* language for sound folding */
|
|
625
|
12912 int maxsf; /* su_maxscore applies to soundfold score,
|
|
|
12913 su_sfmaxscore to the total score. */
|
|
|
12914 {
|
|
|
12915 int goodlen; /* len of goodword changed */
|
|
|
12916 int badlen; /* len of bad word changed */
|
|
323
|
12917 suggest_T *stp;
|
|
534
|
12918 suggest_T new_sug;
|
|
323
|
12919 int i;
|
|
534
|
12920 char_u *pgood, *pbad;
|
|
532
|
12921
|
|
534
|
12922 /* Minimize "badlen" for consistency. Avoids that changing "the the" to
|
|
|
12923 * "thee the" is added next to changing the first "the" the "thee". */
|
|
|
12924 pgood = goodword + STRLEN(goodword);
|
|
625
|
12925 pbad = su->su_badptr + badlenarg;
|
|
|
12926 for (;;)
|
|
|
12927 {
|
|
|
12928 goodlen = pgood - goodword;
|
|
|
12929 badlen = pbad - su->su_badptr;
|
|
|
12930 if (goodlen <= 0 || badlen <= 0)
|
|
|
12931 break;
|
|
534
|
12932 mb_ptr_back(goodword, pgood);
|
|
|
12933 mb_ptr_back(su->su_badptr, pbad);
|
|
|
12934 #ifdef FEAT_MBYTE
|
|
|
12935 if (has_mbyte)
|
|
|
12936 {
|
|
|
12937 if (mb_ptr2char(pgood) != mb_ptr2char(pbad))
|
|
|
12938 break;
|
|
346
|
12939 }
|
|
|
12940 else
|
|
534
|
12941 #endif
|
|
|
12942 if (*pgood != *pbad)
|
|
|
12943 break;
|
|
625
|
12944 }
|
|
|
12945
|
|
534
|
12946 if (badlen == 0 && goodlen == 0)
|
|
|
12947 /* goodword doesn't change anything; may happen for "the the" changing
|
|
|
12948 * the first "the" to itself. */
|
|
|
12949 return;
|
|
346
|
12950
|
|
625
|
12951 /* Check if the word is already there. Also check the length that is
|
|
|
12952 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
12953 * "thes" -> "these". */
|
|
|
12954 stp = &SUG(*gap, 0);
|
|
|
12955 for (i = gap->ga_len; --i >= 0; ++stp)
|
|
|
12956 if (stp->st_wordlen == goodlen
|
|
|
12957 && stp->st_orglen == badlen
|
|
|
12958 && STRNCMP(stp->st_word, goodword, goodlen) == 0)
|
|
|
12959 {
|
|
|
12960 /*
|
|
|
12961 * Found it. Remember the word with the lowest score.
|
|
|
12962 */
|
|
|
12963 if (stp->st_slang == NULL)
|
|
503
|
12964 stp->st_slang = slang;
|
|
625
|
12965
|
|
|
12966 new_sug.st_score = score;
|
|
|
12967 new_sug.st_altscore = altscore;
|
|
|
12968 new_sug.st_had_bonus = had_bonus;
|
|
|
12969
|
|
|
12970 if (stp->st_had_bonus != had_bonus)
|
|
|
12971 {
|
|
|
12972 /* Only one of the two had the soundalike score computed.
|
|
|
12973 * Need to do that for the other one now, otherwise the
|
|
|
12974 * scores can't be compared. This happens because
|
|
|
12975 * suggest_try_change() doesn't compute the soundalike
|
|
|
12976 * word to keep it fast, while some special methods set
|
|
|
12977 * the soundalike score to zero. */
|
|
|
12978 if (had_bonus)
|
|
|
12979 rescore_one(su, stp);
|
|
|
12980 else
|
|
|
12981 {
|
|
|
12982 new_sug.st_word = stp->st_word;
|
|
|
12983 new_sug.st_wordlen = stp->st_wordlen;
|
|
|
12984 new_sug.st_slang = stp->st_slang;
|
|
|
12985 new_sug.st_orglen = badlen;
|
|
|
12986 rescore_one(su, &new_sug);
|
|
|
12987 }
|
|
|
12988 }
|
|
|
12989
|
|
|
12990 if (stp->st_score > new_sug.st_score)
|
|
|
12991 {
|
|
|
12992 stp->st_score = new_sug.st_score;
|
|
|
12993 stp->st_altscore = new_sug.st_altscore;
|
|
|
12994 stp->st_had_bonus = new_sug.st_had_bonus;
|
|
|
12995 }
|
|
|
12996 break;
|
|
|
12997 }
|
|
|
12998
|
|
|
12999 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
|
13000 {
|
|
|
13001 /* Add a suggestion. */
|
|
|
13002 stp = &SUG(*gap, gap->ga_len);
|
|
|
13003 stp->st_word = vim_strnsave(goodword, goodlen);
|
|
|
13004 if (stp->st_word != NULL)
|
|
|
13005 {
|
|
|
13006 stp->st_wordlen = goodlen;
|
|
|
13007 stp->st_score = score;
|
|
|
13008 stp->st_altscore = altscore;
|
|
|
13009 stp->st_had_bonus = had_bonus;
|
|
|
13010 stp->st_orglen = badlen;
|
|
|
13011 stp->st_slang = slang;
|
|
|
13012 ++gap->ga_len;
|
|
|
13013
|
|
|
13014 /* If we have too many suggestions now, sort the list and keep
|
|
|
13015 * the best suggestions. */
|
|
|
13016 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
13017 {
|
|
|
13018 if (maxsf)
|
|
|
13019 su->su_sfmaxscore = cleanup_suggestions(gap,
|
|
|
13020 su->su_sfmaxscore, SUG_CLEAN_COUNT(su));
|
|
|
13021 else
|
|
|
13022 {
|
|
|
13023 i = su->su_maxscore;
|
|
|
13024 su->su_maxscore = cleanup_suggestions(gap,
|
|
|
13025 su->su_maxscore, SUG_CLEAN_COUNT(su));
|
|
|
13026 }
|
|
|
13027 }
|
|
|
13028 }
|
|
|
13029 }
|
|
|
13030 }
|
|
|
13031
|
|
|
13032 /*
|
|
|
13033 * Suggestions may in fact be flagged as errors. Esp. for banned words and
|
|
|
13034 * for split words, such as "the the". Remove these from the list here.
|
|
|
13035 */
|
|
|
13036 static void
|
|
|
13037 check_suggestions(su, gap)
|
|
|
13038 suginfo_T *su;
|
|
|
13039 garray_T *gap; /* either su_ga or su_sga */
|
|
|
13040 {
|
|
|
13041 suggest_T *stp;
|
|
|
13042 int i;
|
|
|
13043 char_u longword[MAXWLEN + 1];
|
|
|
13044 int len;
|
|
|
13045 hlf_T attr;
|
|
|
13046
|
|
|
13047 stp = &SUG(*gap, 0);
|
|
|
13048 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
|
13049 {
|
|
|
13050 /* Need to append what follows to check for "the the". */
|
|
|
13051 STRCPY(longword, stp[i].st_word);
|
|
|
13052 len = stp[i].st_wordlen;
|
|
|
13053 vim_strncpy(longword + len, su->su_badptr + stp[i].st_orglen,
|
|
|
13054 MAXWLEN - len);
|
|
|
13055 attr = HLF_COUNT;
|
|
|
13056 (void)spell_check(curwin, longword, &attr, NULL, FALSE);
|
|
|
13057 if (attr != HLF_COUNT)
|
|
|
13058 {
|
|
|
13059 /* Remove this entry. */
|
|
|
13060 vim_free(stp[i].st_word);
|
|
|
13061 --gap->ga_len;
|
|
|
13062 if (i < gap->ga_len)
|
|
|
13063 mch_memmove(stp + i, stp + i + 1,
|
|
|
13064 sizeof(suggest_T) * (gap->ga_len - i));
|
|
|
13065 }
|
|
|
13066 }
|
|
|
13067 }
|
|
|
13068
|
|
323
|
13069
|
|
|
13070 /*
|
|
|
13071 * Add a word to be banned.
|
|
|
13072 */
|
|
|
13073 static void
|
|
|
13074 add_banned(su, word)
|
|
|
13075 suginfo_T *su;
|
|
|
13076 char_u *word;
|
|
|
13077 {
|
|
626
|
13078 char_u *s;
|
|
323
|
13079 hash_T hash;
|
|
|
13080 hashitem_T *hi;
|
|
|
13081
|
|
625
|
13082 hash = hash_hash(word);
|
|
|
13083 hi = hash_lookup(&su->su_banned, word, hash);
|
|
|
13084 if (HASHITEM_EMPTY(hi))
|
|
|
13085 {
|
|
|
13086 s = vim_strsave(word);
|
|
|
13087 if (s != NULL)
|
|
323
|
13088 hash_add_item(&su->su_banned, hi, s, hash);
|
|
625
|
13089 }
|
|
323
|
13090 }
|
|
|
13091
|
|
324
|
13092 /*
|
|
534
|
13093 * Recompute the score for all suggestions if sound-folding is possible. This
|
|
|
13094 * is slow, thus only done for the final results.
|
|
324
|
13095 */
|
|
|
13096 static void
|
|
|
13097 rescore_suggestions(su)
|
|
|
13098 suginfo_T *su;
|
|
|
13099 {
|
|
|
13100 int i;
|
|
534
|
13101
|
|
|
13102 if (su->su_sallang != NULL)
|
|
503
|
13103 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
534
|
13104 rescore_one(su, &SUG(su->su_ga, i));
|
|
|
13105 }
|
|
|
13106
|
|
|
13107 /*
|
|
|
13108 * Recompute the score for one suggestion if sound-folding is possible.
|
|
|
13109 */
|
|
|
13110 static void
|
|
|
13111 rescore_one(su, stp)
|
|
535
|
13112 suginfo_T *su;
|
|
|
13113 suggest_T *stp;
|
|
534
|
13114 {
|
|
|
13115 slang_T *slang = stp->st_slang;
|
|
|
13116 char_u sal_badword[MAXWLEN];
|
|
535
|
13117 char_u *p;
|
|
534
|
13118
|
|
|
13119 /* Only rescore suggestions that have no sal score yet and do have a
|
|
|
13120 * language. */
|
|
|
13121 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus)
|
|
|
13122 {
|
|
|
13123 if (slang == su->su_sallang)
|
|
535
|
13124 p = su->su_sal_badword;
|
|
534
|
13125 else
|
|
|
13126 {
|
|
|
13127 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword);
|
|
535
|
13128 p = sal_badword;
|
|
|
13129 }
|
|
|
13130
|
|
|
13131 stp->st_altscore = stp_sal_score(stp, su, slang, p);
|
|
534
|
13132 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
13133 stp->st_altscore = SCORE_BIG;
|
|
|
13134 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
|
13135 stp->st_had_bonus = TRUE;
|
|
324
|
13136 }
|
|
|
13137 }
|
|
|
13138
|
|
323
|
13139 static int
|
|
|
13140 #ifdef __BORLANDC__
|
|
|
13141 _RTLENTRYF
|
|
|
13142 #endif
|
|
|
13143 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
13144
|
|
|
13145 /*
|
|
|
13146 * Function given to qsort() to sort the suggestions on st_score.
|
|
529
|
13147 * First on "st_score", then "st_altscore" then alphabetically.
|
|
323
|
13148 */
|
|
|
13149 static int
|
|
|
13150 #ifdef __BORLANDC__
|
|
|
13151 _RTLENTRYF
|
|
|
13152 #endif
|
|
|
13153 sug_compare(s1, s2)
|
|
|
13154 const void *s1;
|
|
|
13155 const void *s2;
|
|
|
13156 {
|
|
|
13157 suggest_T *p1 = (suggest_T *)s1;
|
|
|
13158 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
13159 int n = p1->st_score - p2->st_score;
|
|
|
13160
|
|
|
13161 if (n == 0)
|
|
529
|
13162 {
|
|
|
13163 n = p1->st_altscore - p2->st_altscore;
|
|
|
13164 if (n == 0)
|
|
|
13165 n = STRICMP(p1->st_word, p2->st_word);
|
|
|
13166 }
|
|
344
|
13167 return n;
|
|
323
|
13168 }
|
|
|
13169
|
|
|
13170 /*
|
|
|
13171 * Cleanup the suggestions:
|
|
|
13172 * - Sort on score.
|
|
|
13173 * - Remove words that won't be displayed.
|
|
344
|
13174 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
13175 */
|
|
344
|
13176 static int
|
|
|
13177 cleanup_suggestions(gap, maxscore, keep)
|
|
|
13178 garray_T *gap;
|
|
|
13179 int maxscore;
|
|
324
|
13180 int keep; /* nr of suggestions to keep */
|
|
323
|
13181 {
|
|
344
|
13182 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
13183 int i;
|
|
|
13184
|
|
|
13185 /* Sort the list. */
|
|
344
|
13186 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
13187
|
|
|
13188 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
13189 if (gap->ga_len > keep)
|
|
323
|
13190 {
|
|
344
|
13191 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
13192 vim_free(stp[i].st_word);
|
|
344
|
13193 gap->ga_len = keep;
|
|
|
13194 return stp[keep - 1].st_score;
|
|
323
|
13195 }
|
|
344
|
13196 return maxscore;
|
|
323
|
13197 }
|
|
|
13198
|
|
372
|
13199 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
13200 /*
|
|
|
13201 * Soundfold a string, for soundfold().
|
|
|
13202 * Result is in allocated memory, NULL for an error.
|
|
|
13203 */
|
|
|
13204 char_u *
|
|
|
13205 eval_soundfold(word)
|
|
|
13206 char_u *word;
|
|
|
13207 {
|
|
|
13208 langp_T *lp;
|
|
|
13209 char_u sound[MAXWLEN];
|
|
500
|
13210 int lpi;
|
|
372
|
13211
|
|
|
13212 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
13213 /* Use the sound-folding of the first language that supports it. */
|
|
503
|
13214 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
13215 {
|
|
|
13216 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
372
|
13217 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
13218 {
|
|
|
13219 /* soundfold the word */
|
|
375
|
13220 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
13221 return vim_strsave(sound);
|
|
|
13222 }
|
|
500
|
13223 }
|
|
372
|
13224
|
|
|
13225 /* No language with sound folding, return word as-is. */
|
|
|
13226 return vim_strsave(word);
|
|
|
13227 }
|
|
|
13228 #endif
|
|
|
13229
|
|
323
|
13230 /*
|
|
|
13231 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
485
|
13232 *
|
|
|
13233 * There are many ways to turn a word into a sound-a-like representation. The
|
|
|
13234 * oldest is Soundex (1918!). A nice overview can be found in "Approximate
|
|
|
13235 * swedish name matching - survey and test of different algorithms" by Klas
|
|
|
13236 * Erikson.
|
|
|
13237 *
|
|
|
13238 * We support two methods:
|
|
|
13239 * 1. SOFOFROM/SOFOTO do a simple character mapping.
|
|
|
13240 * 2. SAL items define a more advanced sound-folding (and much slower).
|
|
323
|
13241 */
|
|
|
13242 static void
|
|
375
|
13243 spell_soundfold(slang, inword, folded, res)
|
|
|
13244 slang_T *slang;
|
|
|
13245 char_u *inword;
|
|
|
13246 int folded; /* "inword" is already case-folded */
|
|
|
13247 char_u *res;
|
|
|
13248 {
|
|
|
13249 char_u fword[MAXWLEN];
|
|
|
13250 char_u *word;
|
|
|
13251
|
|
|
13252 if (slang->sl_sofo)
|
|
|
13253 /* SOFOFROM and SOFOTO used */
|
|
|
13254 spell_soundfold_sofo(slang, inword, res);
|
|
|
13255 else
|
|
|
13256 {
|
|
|
13257 /* SAL items used. Requires the word to be case-folded. */
|
|
|
13258 if (folded)
|
|
|
13259 word = inword;
|
|
|
13260 else
|
|
|
13261 {
|
|
|
13262 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
13263 word = fword;
|
|
|
13264 }
|
|
|
13265
|
|
|
13266 #ifdef FEAT_MBYTE
|
|
|
13267 if (has_mbyte)
|
|
|
13268 spell_soundfold_wsal(slang, word, res);
|
|
|
13269 else
|
|
|
13270 #endif
|
|
|
13271 spell_soundfold_sal(slang, word, res);
|
|
|
13272 }
|
|
|
13273 }
|
|
|
13274
|
|
|
13275 /*
|
|
|
13276 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
13277 * SOFOTO lines.
|
|
|
13278 */
|
|
|
13279 static void
|
|
|
13280 spell_soundfold_sofo(slang, inword, res)
|
|
|
13281 slang_T *slang;
|
|
|
13282 char_u *inword;
|
|
|
13283 char_u *res;
|
|
|
13284 {
|
|
|
13285 char_u *s;
|
|
|
13286 int ri = 0;
|
|
|
13287 int c;
|
|
|
13288
|
|
|
13289 #ifdef FEAT_MBYTE
|
|
|
13290 if (has_mbyte)
|
|
|
13291 {
|
|
|
13292 int prevc = 0;
|
|
|
13293 int *ip;
|
|
|
13294
|
|
|
13295 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
13296 * 255, sl_sal the rest. */
|
|
|
13297 for (s = inword; *s != NUL; )
|
|
|
13298 {
|
|
474
|
13299 c = mb_cptr2char_adv(&s);
|
|
375
|
13300 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
13301 c = ' ';
|
|
|
13302 else if (c < 256)
|
|
|
13303 c = slang->sl_sal_first[c];
|
|
|
13304 else
|
|
|
13305 {
|
|
|
13306 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
13307 if (ip == NULL) /* empty list, can't match */
|
|
|
13308 c = NUL;
|
|
|
13309 else
|
|
|
13310 for (;;) /* find "c" in the list */
|
|
|
13311 {
|
|
|
13312 if (*ip == 0) /* not found */
|
|
|
13313 {
|
|
|
13314 c = NUL;
|
|
|
13315 break;
|
|
|
13316 }
|
|
|
13317 if (*ip == c) /* match! */
|
|
|
13318 {
|
|
|
13319 c = ip[1];
|
|
|
13320 break;
|
|
|
13321 }
|
|
|
13322 ip += 2;
|
|
|
13323 }
|
|
|
13324 }
|
|
|
13325
|
|
|
13326 if (c != NUL && c != prevc)
|
|
|
13327 {
|
|
|
13328 ri += mb_char2bytes(c, res + ri);
|
|
|
13329 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
13330 break;
|
|
|
13331 prevc = c;
|
|
|
13332 }
|
|
|
13333 }
|
|
|
13334 }
|
|
|
13335 else
|
|
|
13336 #endif
|
|
|
13337 {
|
|
|
13338 /* The sl_sal_first[] table contains the translation. */
|
|
|
13339 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
13340 {
|
|
|
13341 if (vim_iswhite(c))
|
|
|
13342 c = ' ';
|
|
|
13343 else
|
|
|
13344 c = slang->sl_sal_first[c];
|
|
|
13345 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
13346 res[ri++] = c;
|
|
|
13347 }
|
|
|
13348 }
|
|
|
13349
|
|
|
13350 res[ri] = NUL;
|
|
|
13351 }
|
|
|
13352
|
|
|
13353 static void
|
|
|
13354 spell_soundfold_sal(slang, inword, res)
|
|
323
|
13355 slang_T *slang;
|
|
|
13356 char_u *inword;
|
|
|
13357 char_u *res;
|
|
|
13358 {
|
|
344
|
13359 salitem_T *smp;
|
|
323
|
13360 char_u word[MAXWLEN];
|
|
375
|
13361 char_u *s = inword;
|
|
323
|
13362 char_u *t;
|
|
344
|
13363 char_u *pf;
|
|
323
|
13364 int i, j, z;
|
|
344
|
13365 int reslen;
|
|
323
|
13366 int n, k = 0;
|
|
|
13367 int z0;
|
|
|
13368 int k0;
|
|
|
13369 int n0;
|
|
|
13370 int c;
|
|
|
13371 int pri;
|
|
|
13372 int p0 = -333;
|
|
|
13373 int c0;
|
|
|
13374
|
|
324
|
13375 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
13376 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
13377 if (slang->sl_rem_accents)
|
|
|
13378 {
|
|
|
13379 t = word;
|
|
375
|
13380 while (*s != NUL)
|
|
323
|
13381 {
|
|
324
|
13382 if (vim_iswhite(*s))
|
|
344
|
13383 {
|
|
|
13384 *t++ = ' ';
|
|
|
13385 s = skipwhite(s);
|
|
|
13386 }
|
|
323
|
13387 else
|
|
|
13388 {
|
|
376
|
13389 if (spell_iswordp_nmw(s))
|
|
323
|
13390 *t++ = *s;
|
|
|
13391 ++s;
|
|
|
13392 }
|
|
|
13393 }
|
|
|
13394 *t = NUL;
|
|
|
13395 }
|
|
|
13396 else
|
|
375
|
13397 STRCPY(word, s);
|
|
323
|
13398
|
|
344
|
13399 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
13400
|
|
|
13401 /*
|
|
|
13402 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
13403 * Changed to keep spaces.
|
|
323
|
13404 */
|
|
344
|
13405 i = reslen = z = 0;
|
|
323
|
13406 while ((c = word[i]) != NUL)
|
|
|
13407 {
|
|
344
|
13408 /* Start with the first rule that has the character in the word. */
|
|
323
|
13409 n = slang->sl_sal_first[c];
|
|
|
13410 z0 = 0;
|
|
|
13411
|
|
|
13412 if (n >= 0)
|
|
|
13413 {
|
|
|
13414 /* check all rules for the same letter */
|
|
344
|
13415 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
13416 {
|
|
344
|
13417 /* Quickly skip entries that don't match the word. Most
|
|
|
13418 * entries are less then three chars, optimize for that. */
|
|
|
13419 k = smp[n].sm_leadlen;
|
|
|
13420 if (k > 1)
|
|
323
|
13421 {
|
|
344
|
13422 if (word[i + 1] != s[1])
|
|
|
13423 continue;
|
|
|
13424 if (k > 2)
|
|
|
13425 {
|
|
|
13426 for (j = 2; j < k; ++j)
|
|
|
13427 if (word[i + j] != s[j])
|
|
|
13428 break;
|
|
|
13429 if (j < k)
|
|
|
13430 continue;
|
|
|
13431 }
|
|
323
|
13432 }
|
|
|
13433
|
|
375
|
13434 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
13435 {
|
|
375
|
13436 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
13437 while (*pf != NUL && *pf != word[i + k])
|
|
|
13438 ++pf;
|
|
|
13439 if (*pf == NUL)
|
|
|
13440 continue;
|
|
|
13441 ++k;
|
|
323
|
13442 }
|
|
344
|
13443 s = smp[n].sm_rules;
|
|
|
13444 pri = 5; /* default priority */
|
|
323
|
13445
|
|
|
13446 p0 = *s;
|
|
|
13447 k0 = k;
|
|
|
13448 while (*s == '-' && k > 1)
|
|
|
13449 {
|
|
|
13450 k--;
|
|
|
13451 s++;
|
|
|
13452 }
|
|
|
13453 if (*s == '<')
|
|
|
13454 s++;
|
|
344
|
13455 if (VIM_ISDIGIT(*s))
|
|
323
|
13456 {
|
|
|
13457 /* determine priority */
|
|
|
13458 pri = *s - '0';
|
|
|
13459 s++;
|
|
|
13460 }
|
|
|
13461 if (*s == '^' && *(s + 1) == '^')
|
|
|
13462 s++;
|
|
|
13463
|
|
|
13464 if (*s == NUL
|
|
|
13465 || (*s == '^'
|
|
324
|
13466 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
13467 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
13468 && (*(s + 1) != '$'
|
|
376
|
13469 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
13470 || (*s == '$' && i > 0
|
|
376
|
13471 && spell_iswordp(word + i - 1, curbuf)
|
|
|
13472 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
13473 {
|
|
|
13474 /* search for followup rules, if: */
|
|
|
13475 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
13476 c0 = word[i + k - 1];
|
|
|
13477 n0 = slang->sl_sal_first[c0];
|
|
|
13478
|
|
|
13479 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
13480 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
13481 {
|
|
|
13482 /* test follow-up rule for "word[i + k]" */
|
|
344
|
13483 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
13484 {
|
|
344
|
13485 /* Quickly skip entries that don't match the word.
|
|
|
13486 * */
|
|
|
13487 k0 = smp[n0].sm_leadlen;
|
|
|
13488 if (k0 > 1)
|
|
323
|
13489 {
|
|
344
|
13490 if (word[i + k] != s[1])
|
|
|
13491 continue;
|
|
|
13492 if (k0 > 2)
|
|
|
13493 {
|
|
|
13494 pf = word + i + k + 1;
|
|
|
13495 for (j = 2; j < k0; ++j)
|
|
|
13496 if (*pf++ != s[j])
|
|
|
13497 break;
|
|
|
13498 if (j < k0)
|
|
|
13499 continue;
|
|
|
13500 }
|
|
323
|
13501 }
|
|
344
|
13502 k0 += k - 1;
|
|
|
13503
|
|
375
|
13504 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
13505 {
|
|
344
|
13506 /* Check for match with one of the chars in
|
|
375
|
13507 * "sm_oneof". */
|
|
344
|
13508 while (*pf != NUL && *pf != word[i + k0])
|
|
|
13509 ++pf;
|
|
|
13510 if (*pf == NUL)
|
|
|
13511 continue;
|
|
|
13512 ++k0;
|
|
323
|
13513 }
|
|
344
|
13514
|
|
|
13515 p0 = 5;
|
|
|
13516 s = smp[n0].sm_rules;
|
|
323
|
13517 while (*s == '-')
|
|
|
13518 {
|
|
344
|
13519 /* "k0" gets NOT reduced because
|
|
|
13520 * "if (k0 == k)" */
|
|
323
|
13521 s++;
|
|
|
13522 }
|
|
|
13523 if (*s == '<')
|
|
|
13524 s++;
|
|
344
|
13525 if (VIM_ISDIGIT(*s))
|
|
323
|
13526 {
|
|
|
13527 p0 = *s - '0';
|
|
|
13528 s++;
|
|
|
13529 }
|
|
|
13530
|
|
|
13531 if (*s == NUL
|
|
|
13532 /* *s == '^' cuts */
|
|
|
13533 || (*s == '$'
|
|
376
|
13534 && !spell_iswordp(word + i + k0,
|
|
|
13535 curbuf)))
|
|
323
|
13536 {
|
|
|
13537 if (k0 == k)
|
|
|
13538 /* this is just a piece of the string */
|
|
|
13539 continue;
|
|
|
13540
|
|
|
13541 if (p0 < pri)
|
|
|
13542 /* priority too low */
|
|
|
13543 continue;
|
|
|
13544 /* rule fits; stop search */
|
|
|
13545 break;
|
|
|
13546 }
|
|
|
13547 }
|
|
|
13548
|
|
344
|
13549 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
13550 continue;
|
|
|
13551 }
|
|
|
13552
|
|
|
13553 /* replace string */
|
|
344
|
13554 s = smp[n].sm_to;
|
|
389
|
13555 if (s == NULL)
|
|
|
13556 s = (char_u *)"";
|
|
344
|
13557 pf = smp[n].sm_rules;
|
|
|
13558 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
13559 if (p0 == 1 && z == 0)
|
|
|
13560 {
|
|
|
13561 /* rule with '<' is used */
|
|
344
|
13562 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
13563 || res[reslen - 1] == *s))
|
|
|
13564 reslen--;
|
|
323
|
13565 z0 = 1;
|
|
|
13566 z = 1;
|
|
|
13567 k0 = 0;
|
|
372
|
13568 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
13569 {
|
|
|
13570 word[i + k0] = *s;
|
|
|
13571 k0++;
|
|
|
13572 s++;
|
|
|
13573 }
|
|
|
13574 if (k > k0)
|
|
|
13575 mch_memmove(word + i + k0, word + i + k,
|
|
|
13576 STRLEN(word + i + k) + 1);
|
|
|
13577
|
|
|
13578 /* new "actual letter" */
|
|
|
13579 c = word[i];
|
|
|
13580 }
|
|
|
13581 else
|
|
|
13582 {
|
|
|
13583 /* no '<' rule used */
|
|
|
13584 i += k - 1;
|
|
|
13585 z = 0;
|
|
344
|
13586 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
13587 {
|
|
344
|
13588 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
13589 res[reslen++] = *s;
|
|
323
|
13590 s++;
|
|
|
13591 }
|
|
|
13592 /* new "actual letter" */
|
|
|
13593 c = *s;
|
|
344
|
13594 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
13595 {
|
|
|
13596 if (c != NUL)
|
|
372
|
13597 res[reslen++] = c;
|
|
323
|
13598 mch_memmove(word, word + i + 1,
|
|
|
13599 STRLEN(word + i + 1) + 1);
|
|
|
13600 i = 0;
|
|
|
13601 z0 = 1;
|
|
|
13602 }
|
|
|
13603 }
|
|
|
13604 break;
|
|
|
13605 }
|
|
|
13606 }
|
|
|
13607 }
|
|
324
|
13608 else if (vim_iswhite(c))
|
|
|
13609 {
|
|
|
13610 c = ' ';
|
|
|
13611 k = 1;
|
|
|
13612 }
|
|
323
|
13613
|
|
|
13614 if (z0 == 0)
|
|
|
13615 {
|
|
344
|
13616 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
13617 && (!slang->sl_collapse || reslen == 0
|
|
|
13618 || res[reslen - 1] != c))
|
|
323
|
13619 /* condense only double letters */
|
|
372
|
13620 res[reslen++] = c;
|
|
323
|
13621
|
|
|
13622 i++;
|
|
|
13623 z = 0;
|
|
|
13624 k = 0;
|
|
|
13625 }
|
|
|
13626 }
|
|
|
13627
|
|
344
|
13628 res[reslen] = NUL;
|
|
323
|
13629 }
|
|
|
13630
|
|
372
|
13631 #ifdef FEAT_MBYTE
|
|
|
13632 /*
|
|
|
13633 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
13634 * Multi-byte version of spell_soundfold().
|
|
|
13635 */
|
|
|
13636 static void
|
|
375
|
13637 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
13638 slang_T *slang;
|
|
|
13639 char_u *inword;
|
|
|
13640 char_u *res;
|
|
|
13641 {
|
|
375
|
13642 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
13643 int word[MAXWLEN];
|
|
|
13644 int wres[MAXWLEN];
|
|
|
13645 int l;
|
|
|
13646 char_u *s;
|
|
|
13647 int *ws;
|
|
|
13648 char_u *t;
|
|
|
13649 int *pf;
|
|
|
13650 int i, j, z;
|
|
|
13651 int reslen;
|
|
|
13652 int n, k = 0;
|
|
|
13653 int z0;
|
|
|
13654 int k0;
|
|
|
13655 int n0;
|
|
|
13656 int c;
|
|
|
13657 int pri;
|
|
|
13658 int p0 = -333;
|
|
|
13659 int c0;
|
|
|
13660 int did_white = FALSE;
|
|
|
13661
|
|
|
13662 /*
|
|
|
13663 * Convert the multi-byte string to a wide-character string.
|
|
|
13664 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
13665 * But keep white space.
|
|
|
13666 */
|
|
|
13667 n = 0;
|
|
|
13668 for (s = inword; *s != NUL; )
|
|
|
13669 {
|
|
|
13670 t = s;
|
|
474
|
13671 c = mb_cptr2char_adv(&s);
|
|
372
|
13672 if (slang->sl_rem_accents)
|
|
|
13673 {
|
|
|
13674 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
13675 {
|
|
|
13676 if (did_white)
|
|
|
13677 continue;
|
|
|
13678 c = ' ';
|
|
|
13679 did_white = TRUE;
|
|
|
13680 }
|
|
|
13681 else
|
|
|
13682 {
|
|
|
13683 did_white = FALSE;
|
|
376
|
13684 if (!spell_iswordp_nmw(t))
|
|
372
|
13685 continue;
|
|
|
13686 }
|
|
|
13687 }
|
|
|
13688 word[n++] = c;
|
|
|
13689 }
|
|
|
13690 word[n] = NUL;
|
|
|
13691
|
|
|
13692 /*
|
|
|
13693 * This comes from Aspell phonet.cpp.
|
|
|
13694 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
13695 * Changed to keep spaces.
|
|
|
13696 */
|
|
|
13697 i = reslen = z = 0;
|
|
|
13698 while ((c = word[i]) != NUL)
|
|
|
13699 {
|
|
|
13700 /* Start with the first rule that has the character in the word. */
|
|
|
13701 n = slang->sl_sal_first[c & 0xff];
|
|
|
13702 z0 = 0;
|
|
|
13703
|
|
|
13704 if (n >= 0)
|
|
|
13705 {
|
|
375
|
13706 /* check all rules for the same index byte */
|
|
372
|
13707 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
13708 {
|
|
|
13709 /* Quickly skip entries that don't match the word. Most
|
|
|
13710 * entries are less then three chars, optimize for that. */
|
|
375
|
13711 if (c != ws[0])
|
|
|
13712 continue;
|
|
372
|
13713 k = smp[n].sm_leadlen;
|
|
|
13714 if (k > 1)
|
|
|
13715 {
|
|
|
13716 if (word[i + 1] != ws[1])
|
|
|
13717 continue;
|
|
|
13718 if (k > 2)
|
|
|
13719 {
|
|
|
13720 for (j = 2; j < k; ++j)
|
|
|
13721 if (word[i + j] != ws[j])
|
|
|
13722 break;
|
|
|
13723 if (j < k)
|
|
|
13724 continue;
|
|
|
13725 }
|
|
|
13726 }
|
|
|
13727
|
|
375
|
13728 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
13729 {
|
|
375
|
13730 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
13731 while (*pf != NUL && *pf != word[i + k])
|
|
|
13732 ++pf;
|
|
|
13733 if (*pf == NUL)
|
|
|
13734 continue;
|
|
|
13735 ++k;
|
|
|
13736 }
|
|
|
13737 s = smp[n].sm_rules;
|
|
|
13738 pri = 5; /* default priority */
|
|
|
13739
|
|
|
13740 p0 = *s;
|
|
|
13741 k0 = k;
|
|
|
13742 while (*s == '-' && k > 1)
|
|
|
13743 {
|
|
|
13744 k--;
|
|
|
13745 s++;
|
|
|
13746 }
|
|
|
13747 if (*s == '<')
|
|
|
13748 s++;
|
|
|
13749 if (VIM_ISDIGIT(*s))
|
|
|
13750 {
|
|
|
13751 /* determine priority */
|
|
|
13752 pri = *s - '0';
|
|
|
13753 s++;
|
|
|
13754 }
|
|
|
13755 if (*s == '^' && *(s + 1) == '^')
|
|
|
13756 s++;
|
|
|
13757
|
|
|
13758 if (*s == NUL
|
|
|
13759 || (*s == '^'
|
|
|
13760 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
13761 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
13762 && (*(s + 1) != '$'
|
|
376
|
13763 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
13764 || (*s == '$' && i > 0
|
|
376
|
13765 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
13766 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
13767 {
|
|
|
13768 /* search for followup rules, if: */
|
|
|
13769 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
13770 c0 = word[i + k - 1];
|
|
|
13771 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
13772
|
|
|
13773 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
13774 && p0 != '-' && word[i + k] != NUL)
|
|
|
13775 {
|
|
375
|
13776 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
13777 * all entries with the same index byte. */
|
|
372
|
13778 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
13779 == (c0 & 0xff); ++n0)
|
|
|
13780 {
|
|
|
13781 /* Quickly skip entries that don't match the word.
|
|
375
|
13782 */
|
|
|
13783 if (c0 != ws[0])
|
|
|
13784 continue;
|
|
372
|
13785 k0 = smp[n0].sm_leadlen;
|
|
|
13786 if (k0 > 1)
|
|
|
13787 {
|
|
|
13788 if (word[i + k] != ws[1])
|
|
|
13789 continue;
|
|
|
13790 if (k0 > 2)
|
|
|
13791 {
|
|
|
13792 pf = word + i + k + 1;
|
|
|
13793 for (j = 2; j < k0; ++j)
|
|
|
13794 if (*pf++ != ws[j])
|
|
|
13795 break;
|
|
|
13796 if (j < k0)
|
|
|
13797 continue;
|
|
|
13798 }
|
|
|
13799 }
|
|
|
13800 k0 += k - 1;
|
|
|
13801
|
|
375
|
13802 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
13803 {
|
|
|
13804 /* Check for match with one of the chars in
|
|
375
|
13805 * "sm_oneof". */
|
|
372
|
13806 while (*pf != NUL && *pf != word[i + k0])
|
|
|
13807 ++pf;
|
|
|
13808 if (*pf == NUL)
|
|
|
13809 continue;
|
|
|
13810 ++k0;
|
|
|
13811 }
|
|
|
13812
|
|
|
13813 p0 = 5;
|
|
|
13814 s = smp[n0].sm_rules;
|
|
|
13815 while (*s == '-')
|
|
|
13816 {
|
|
|
13817 /* "k0" gets NOT reduced because
|
|
|
13818 * "if (k0 == k)" */
|
|
|
13819 s++;
|
|
|
13820 }
|
|
|
13821 if (*s == '<')
|
|
|
13822 s++;
|
|
|
13823 if (VIM_ISDIGIT(*s))
|
|
|
13824 {
|
|
|
13825 p0 = *s - '0';
|
|
|
13826 s++;
|
|
|
13827 }
|
|
|
13828
|
|
|
13829 if (*s == NUL
|
|
|
13830 /* *s == '^' cuts */
|
|
|
13831 || (*s == '$'
|
|
376
|
13832 && !spell_iswordp_w(word + i + k0,
|
|
|
13833 curbuf)))
|
|
372
|
13834 {
|
|
|
13835 if (k0 == k)
|
|
|
13836 /* this is just a piece of the string */
|
|
|
13837 continue;
|
|
|
13838
|
|
|
13839 if (p0 < pri)
|
|
|
13840 /* priority too low */
|
|
|
13841 continue;
|
|
|
13842 /* rule fits; stop search */
|
|
|
13843 break;
|
|
|
13844 }
|
|
|
13845 }
|
|
|
13846
|
|
|
13847 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
13848 == (c0 & 0xff))
|
|
|
13849 continue;
|
|
|
13850 }
|
|
|
13851
|
|
|
13852 /* replace string */
|
|
|
13853 ws = smp[n].sm_to_w;
|
|
|
13854 s = smp[n].sm_rules;
|
|
|
13855 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
13856 if (p0 == 1 && z == 0)
|
|
|
13857 {
|
|
|
13858 /* rule with '<' is used */
|
|
389
|
13859 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
13860 && (wres[reslen - 1] == c
|
|
372
|
13861 || wres[reslen - 1] == *ws))
|
|
|
13862 reslen--;
|
|
|
13863 z0 = 1;
|
|
|
13864 z = 1;
|
|
|
13865 k0 = 0;
|
|
389
|
13866 if (ws != NULL)
|
|
|
13867 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
13868 {
|
|
|
13869 word[i + k0] = *ws;
|
|
|
13870 k0++;
|
|
|
13871 ws++;
|
|
|
13872 }
|
|
372
|
13873 if (k > k0)
|
|
|
13874 mch_memmove(word + i + k0, word + i + k,
|
|
|
13875 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
13876
|
|
|
13877 /* new "actual letter" */
|
|
|
13878 c = word[i];
|
|
|
13879 }
|
|
|
13880 else
|
|
|
13881 {
|
|
|
13882 /* no '<' rule used */
|
|
|
13883 i += k - 1;
|
|
|
13884 z = 0;
|
|
389
|
13885 if (ws != NULL)
|
|
|
13886 while (*ws != NUL && ws[1] != NUL
|
|
|
13887 && reslen < MAXWLEN)
|
|
|
13888 {
|
|
|
13889 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
13890 wres[reslen++] = *ws;
|
|
|
13891 ws++;
|
|
|
13892 }
|
|
372
|
13893 /* new "actual letter" */
|
|
389
|
13894 if (ws == NULL)
|
|
|
13895 c = NUL;
|
|
|
13896 else
|
|
|
13897 c = *ws;
|
|
372
|
13898 if (strstr((char *)s, "^^") != NULL)
|
|
|
13899 {
|
|
|
13900 if (c != NUL)
|
|
|
13901 wres[reslen++] = c;
|
|
|
13902 mch_memmove(word, word + i + 1,
|
|
|
13903 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
13904 i = 0;
|
|
|
13905 z0 = 1;
|
|
|
13906 }
|
|
|
13907 }
|
|
|
13908 break;
|
|
|
13909 }
|
|
|
13910 }
|
|
|
13911 }
|
|
|
13912 else if (vim_iswhite(c))
|
|
|
13913 {
|
|
|
13914 c = ' ';
|
|
|
13915 k = 1;
|
|
|
13916 }
|
|
|
13917
|
|
|
13918 if (z0 == 0)
|
|
|
13919 {
|
|
|
13920 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
13921 && (!slang->sl_collapse || reslen == 0
|
|
|
13922 || wres[reslen - 1] != c))
|
|
|
13923 /* condense only double letters */
|
|
|
13924 wres[reslen++] = c;
|
|
|
13925
|
|
|
13926 i++;
|
|
|
13927 z = 0;
|
|
|
13928 k = 0;
|
|
|
13929 }
|
|
|
13930 }
|
|
|
13931
|
|
|
13932 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
13933 l = 0;
|
|
|
13934 for (n = 0; n < reslen; ++n)
|
|
|
13935 {
|
|
|
13936 l += mb_char2bytes(wres[n], res + l);
|
|
|
13937 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
13938 break;
|
|
|
13939 }
|
|
|
13940 res[l] = NUL;
|
|
|
13941 }
|
|
|
13942 #endif
|
|
|
13943
|
|
324
|
13944 /*
|
|
344
|
13945 * Compute a score for two sound-a-like words.
|
|
|
13946 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
13947 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
13948 * avoiding checks that will not be possible.
|
|
|
13949 */
|
|
|
13950 static int
|
|
351
|
13951 soundalike_score(goodstart, badstart)
|
|
|
13952 char_u *goodstart; /* sound-folded good word */
|
|
|
13953 char_u *badstart; /* sound-folded bad word */
|
|
344
|
13954 {
|
|
351
|
13955 char_u *goodsound = goodstart;
|
|
|
13956 char_u *badsound = badstart;
|
|
|
13957 int goodlen;
|
|
|
13958 int badlen;
|
|
344
|
13959 int n;
|
|
|
13960 char_u *pl, *ps;
|
|
|
13961 char_u *pl2, *ps2;
|
|
351
|
13962 int score = 0;
|
|
|
13963
|
|
|
13964 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
13965 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
13966 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
13967 {
|
|
625
|
13968 if (badsound[1] == goodsound[1]
|
|
|
13969 || (badsound[1] != NUL
|
|
|
13970 && goodsound[1] != NUL
|
|
|
13971 && badsound[2] == goodsound[2]))
|
|
|
13972 {
|
|
|
13973 /* handle like a substitute */
|
|
|
13974 }
|
|
351
|
13975 else
|
|
625
|
13976 {
|
|
|
13977 score = 2 * SCORE_DEL / 3;
|
|
|
13978 if (*badsound == '*')
|
|
|
13979 ++badsound;
|
|
|
13980 else
|
|
|
13981 ++goodsound;
|
|
|
13982 }
|
|
351
|
13983 }
|
|
|
13984
|
|
|
13985 goodlen = STRLEN(goodsound);
|
|
|
13986 badlen = STRLEN(badsound);
|
|
344
|
13987
|
|
|
13988 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
13989 * changes. */
|
|
|
13990 n = goodlen - badlen;
|
|
|
13991 if (n < -2 || n > 2)
|
|
|
13992 return SCORE_MAXMAX;
|
|
|
13993
|
|
|
13994 if (n > 0)
|
|
|
13995 {
|
|
351
|
13996 pl = goodsound; /* goodsound is longest */
|
|
344
|
13997 ps = badsound;
|
|
|
13998 }
|
|
|
13999 else
|
|
|
14000 {
|
|
351
|
14001 pl = badsound; /* badsound is longest */
|
|
344
|
14002 ps = goodsound;
|
|
|
14003 }
|
|
|
14004
|
|
|
14005 /* Skip over the identical part. */
|
|
|
14006 while (*pl == *ps && *pl != NUL)
|
|
|
14007 {
|
|
|
14008 ++pl;
|
|
|
14009 ++ps;
|
|
|
14010 }
|
|
|
14011
|
|
|
14012 switch (n)
|
|
|
14013 {
|
|
|
14014 case -2:
|
|
|
14015 case 2:
|
|
|
14016 /*
|
|
|
14017 * Must delete two characters from "pl".
|
|
|
14018 */
|
|
|
14019 ++pl; /* first delete */
|
|
|
14020 while (*pl == *ps)
|
|
|
14021 {
|
|
|
14022 ++pl;
|
|
|
14023 ++ps;
|
|
|
14024 }
|
|
|
14025 /* strings must be equal after second delete */
|
|
|
14026 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
14027 return score + SCORE_DEL * 2;
|
|
344
|
14028
|
|
|
14029 /* Failed to compare. */
|
|
|
14030 break;
|
|
|
14031
|
|
|
14032 case -1:
|
|
|
14033 case 1:
|
|
|
14034 /*
|
|
|
14035 * Minimal one delete from "pl" required.
|
|
|
14036 */
|
|
|
14037
|
|
|
14038 /* 1: delete */
|
|
|
14039 pl2 = pl + 1;
|
|
|
14040 ps2 = ps;
|
|
|
14041 while (*pl2 == *ps2)
|
|
|
14042 {
|
|
|
14043 if (*pl2 == NUL) /* reached the end */
|
|
351
|
14044 return score + SCORE_DEL;
|
|
344
|
14045 ++pl2;
|
|
|
14046 ++ps2;
|
|
|
14047 }
|
|
|
14048
|
|
|
14049 /* 2: delete then swap, then rest must be equal */
|
|
|
14050 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
14051 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
14052 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
14053
|
|
|
14054 /* 3: delete then substitute, then the rest must be equal */
|
|
|
14055 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
14056 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
14057
|
|
|
14058 /* 4: first swap then delete */
|
|
|
14059 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
14060 {
|
|
|
14061 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
14062 ps2 = ps + 2;
|
|
|
14063 while (*pl2 == *ps2)
|
|
|
14064 {
|
|
|
14065 ++pl2;
|
|
|
14066 ++ps2;
|
|
|
14067 }
|
|
|
14068 /* delete a char and then strings must be equal */
|
|
|
14069 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
14070 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
14071 }
|
|
|
14072
|
|
|
14073 /* 5: first substitute then delete */
|
|
|
14074 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
14075 ps2 = ps + 1;
|
|
|
14076 while (*pl2 == *ps2)
|
|
|
14077 {
|
|
|
14078 ++pl2;
|
|
|
14079 ++ps2;
|
|
|
14080 }
|
|
|
14081 /* delete a char and then strings must be equal */
|
|
|
14082 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
14083 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
14084
|
|
|
14085 /* Failed to compare. */
|
|
|
14086 break;
|
|
|
14087
|
|
|
14088 case 0:
|
|
|
14089 /*
|
|
|
14090 * Lenghts are equal, thus changes must result in same length: An
|
|
|
14091 * insert is only possible in combination with a delete.
|
|
|
14092 * 1: check if for identical strings
|
|
|
14093 */
|
|
|
14094 if (*pl == NUL)
|
|
351
|
14095 return score;
|
|
344
|
14096
|
|
|
14097 /* 2: swap */
|
|
|
14098 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
14099 {
|
|
|
14100 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
14101 ps2 = ps + 2;
|
|
|
14102 while (*pl2 == *ps2)
|
|
|
14103 {
|
|
|
14104 if (*pl2 == NUL) /* reached the end */
|
|
351
|
14105 return score + SCORE_SWAP;
|
|
344
|
14106 ++pl2;
|
|
|
14107 ++ps2;
|
|
|
14108 }
|
|
|
14109 /* 3: swap and swap again */
|
|
|
14110 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
14111 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
14112 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
14113
|
|
|
14114 /* 4: swap and substitute */
|
|
|
14115 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
14116 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
14117 }
|
|
|
14118
|
|
|
14119 /* 5: substitute */
|
|
|
14120 pl2 = pl + 1;
|
|
|
14121 ps2 = ps + 1;
|
|
|
14122 while (*pl2 == *ps2)
|
|
|
14123 {
|
|
|
14124 if (*pl2 == NUL) /* reached the end */
|
|
351
|
14125 return score + SCORE_SUBST;
|
|
344
|
14126 ++pl2;
|
|
|
14127 ++ps2;
|
|
|
14128 }
|
|
|
14129
|
|
|
14130 /* 6: substitute and swap */
|
|
|
14131 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
14132 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
14133 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
14134
|
|
|
14135 /* 7: substitute and substitute */
|
|
|
14136 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
14137 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
14138
|
|
|
14139 /* 8: insert then delete */
|
|
|
14140 pl2 = pl;
|
|
|
14141 ps2 = ps + 1;
|
|
|
14142 while (*pl2 == *ps2)
|
|
|
14143 {
|
|
|
14144 ++pl2;
|
|
|
14145 ++ps2;
|
|
|
14146 }
|
|
|
14147 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
14148 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
14149
|
|
|
14150 /* 9: delete then insert */
|
|
|
14151 pl2 = pl + 1;
|
|
|
14152 ps2 = ps;
|
|
|
14153 while (*pl2 == *ps2)
|
|
|
14154 {
|
|
|
14155 ++pl2;
|
|
|
14156 ++ps2;
|
|
|
14157 }
|
|
|
14158 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
14159 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
14160
|
|
|
14161 /* Failed to compare. */
|
|
|
14162 break;
|
|
|
14163 }
|
|
|
14164
|
|
|
14165 return SCORE_MAXMAX;
|
|
|
14166 }
|
|
324
|
14167
|
|
323
|
14168 /*
|
|
|
14169 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
14170 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
14171 *
|
|
485
|
14172 * The algorithm is described by Du and Chang, 1992.
|
|
|
14173 * The implementation of the algorithm comes from Aspell editdist.cpp,
|
|
|
14174 * edit_distance(). It has been converted from C++ to C and modified to
|
|
|
14175 * support multi-byte characters.
|
|
323
|
14176 */
|
|
|
14177 static int
|
|
625
|
14178 spell_edit_score(slang, badword, goodword)
|
|
|
14179 slang_T *slang;
|
|
323
|
14180 char_u *badword;
|
|
|
14181 char_u *goodword;
|
|
|
14182 {
|
|
|
14183 int *cnt;
|
|
372
|
14184 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
14185 int j, i;
|
|
|
14186 int t;
|
|
|
14187 int bc, gc;
|
|
324
|
14188 int pbc, pgc;
|
|
|
14189 #ifdef FEAT_MBYTE
|
|
|
14190 char_u *p;
|
|
|
14191 int wbadword[MAXWLEN];
|
|
|
14192 int wgoodword[MAXWLEN];
|
|
|
14193
|
|
|
14194 if (has_mbyte)
|
|
|
14195 {
|
|
|
14196 /* Get the characters from the multi-byte strings and put them in an
|
|
|
14197 * int array for easy access. */
|
|
|
14198 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
14199 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
14200 wbadword[badlen++] = 0;
|
|
324
|
14201 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
14202 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
14203 wgoodword[goodlen++] = 0;
|
|
324
|
14204 }
|
|
|
14205 else
|
|
|
14206 #endif
|
|
|
14207 {
|
|
|
14208 badlen = STRLEN(badword) + 1;
|
|
|
14209 goodlen = STRLEN(goodword) + 1;
|
|
|
14210 }
|
|
323
|
14211
|
|
|
14212 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
14213 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
14214 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
14215 TRUE);
|
|
324
|
14216 if (cnt == NULL)
|
|
|
14217 return 0; /* out of memory */
|
|
323
|
14218
|
|
|
14219 CNT(0, 0) = 0;
|
|
|
14220 for (j = 1; j <= goodlen; ++j)
|
|
625
|
14221 CNT(0, j) = CNT(0, j - 1) + SCORE_INS;
|
|
323
|
14222
|
|
|
14223 for (i = 1; i <= badlen; ++i)
|
|
|
14224 {
|
|
625
|
14225 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL;
|
|
323
|
14226 for (j = 1; j <= goodlen; ++j)
|
|
|
14227 {
|
|
324
|
14228 #ifdef FEAT_MBYTE
|
|
|
14229 if (has_mbyte)
|
|
|
14230 {
|
|
|
14231 bc = wbadword[i - 1];
|
|
|
14232 gc = wgoodword[j - 1];
|
|
|
14233 }
|
|
|
14234 else
|
|
|
14235 #endif
|
|
|
14236 {
|
|
|
14237 bc = badword[i - 1];
|
|
|
14238 gc = goodword[j - 1];
|
|
|
14239 }
|
|
323
|
14240 if (bc == gc)
|
|
|
14241 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
14242 else
|
|
|
14243 {
|
|
|
14244 /* Use a better score when there is only a case difference. */
|
|
324
|
14245 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
14246 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
14247 else
|
|
625
|
14248 {
|
|
|
14249 /* For a similar character use SCORE_SIMILAR. */
|
|
|
14250 if (slang != NULL
|
|
|
14251 && slang->sl_has_map
|
|
|
14252 && similar_chars(slang, gc, bc))
|
|
|
14253 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1);
|
|
|
14254 else
|
|
|
14255 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
14256 }
|
|
323
|
14257
|
|
324
|
14258 if (i > 1 && j > 1)
|
|
323
|
14259 {
|
|
324
|
14260 #ifdef FEAT_MBYTE
|
|
|
14261 if (has_mbyte)
|
|
|
14262 {
|
|
|
14263 pbc = wbadword[i - 2];
|
|
|
14264 pgc = wgoodword[j - 2];
|
|
|
14265 }
|
|
|
14266 else
|
|
|
14267 #endif
|
|
|
14268 {
|
|
|
14269 pbc = badword[i - 2];
|
|
|
14270 pgc = goodword[j - 2];
|
|
|
14271 }
|
|
|
14272 if (bc == pgc && pbc == gc)
|
|
|
14273 {
|
|
|
14274 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
14275 if (t < CNT(i, j))
|
|
|
14276 CNT(i, j) = t;
|
|
|
14277 }
|
|
323
|
14278 }
|
|
|
14279 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
14280 if (t < CNT(i, j))
|
|
|
14281 CNT(i, j) = t;
|
|
|
14282 t = SCORE_INS + CNT(i, j - 1);
|
|
|
14283 if (t < CNT(i, j))
|
|
|
14284 CNT(i, j) = t;
|
|
|
14285 }
|
|
|
14286 }
|
|
|
14287 }
|
|
344
|
14288
|
|
|
14289 i = CNT(badlen - 1, goodlen - 1);
|
|
|
14290 vim_free(cnt);
|
|
|
14291 return i;
|
|
323
|
14292 }
|
|
307
|
14293
|
|
625
|
14294 typedef struct
|
|
|
14295 {
|
|
|
14296 int badi;
|
|
|
14297 int goodi;
|
|
|
14298 int score;
|
|
|
14299 } limitscore_T;
|
|
|
14300
|
|
|
14301 /*
|
|
|
14302 * Like spell_edit_score(), but with a limit on the score to make it faster.
|
|
|
14303 * May return SCORE_MAXMAX when the score is higher than "limit".
|
|
|
14304 *
|
|
|
14305 * This uses a stack for the edits still to be tried.
|
|
|
14306 * The idea comes from Aspell leditdist.cpp. Rewritten in C and added support
|
|
|
14307 * for multi-byte characters.
|
|
|
14308 */
|
|
|
14309 static int
|
|
|
14310 spell_edit_score_limit(slang, badword, goodword, limit)
|
|
|
14311 slang_T *slang;
|
|
|
14312 char_u *badword;
|
|
|
14313 char_u *goodword;
|
|
|
14314 int limit;
|
|
|
14315 {
|
|
|
14316 limitscore_T stack[10]; /* allow for over 3 * 2 edits */
|
|
|
14317 int stackidx;
|
|
|
14318 int bi, gi;
|
|
|
14319 int bi2, gi2;
|
|
|
14320 int bc, gc;
|
|
|
14321 int score;
|
|
|
14322 int score_off;
|
|
|
14323 int minscore;
|
|
|
14324 int round;
|
|
|
14325
|
|
|
14326 #ifdef FEAT_MBYTE
|
|
|
14327 /* Multi-byte characters require a bit more work, use a different function
|
|
|
14328 * to avoid testing "has_mbyte" quite often. */
|
|
|
14329 if (has_mbyte)
|
|
|
14330 return spell_edit_score_limit_w(slang, badword, goodword, limit);
|
|
|
14331 #endif
|
|
|
14332
|
|
|
14333 /*
|
|
|
14334 * The idea is to go from start to end over the words. So long as
|
|
|
14335 * characters are equal just continue, this always gives the lowest score.
|
|
|
14336 * When there is a difference try several alternatives. Each alternative
|
|
|
14337 * increases "score" for the edit distance. Some of the alternatives are
|
|
|
14338 * pushed unto a stack and tried later, some are tried right away. At the
|
|
|
14339 * end of the word the score for one alternative is known. The lowest
|
|
|
14340 * possible score is stored in "minscore".
|
|
|
14341 */
|
|
|
14342 stackidx = 0;
|
|
|
14343 bi = 0;
|
|
|
14344 gi = 0;
|
|
|
14345 score = 0;
|
|
|
14346 minscore = limit + 1;
|
|
|
14347
|
|
|
14348 for (;;)
|
|
|
14349 {
|
|
|
14350 /* Skip over an equal part, score remains the same. */
|
|
|
14351 for (;;)
|
|
|
14352 {
|
|
|
14353 bc = badword[bi];
|
|
|
14354 gc = goodword[gi];
|
|
|
14355 if (bc != gc) /* stop at a char that's different */
|
|
|
14356 break;
|
|
|
14357 if (bc == NUL) /* both words end */
|
|
|
14358 {
|
|
|
14359 if (score < minscore)
|
|
|
14360 minscore = score;
|
|
|
14361 goto pop; /* do next alternative */
|
|
|
14362 }
|
|
|
14363 ++bi;
|
|
|
14364 ++gi;
|
|
|
14365 }
|
|
|
14366
|
|
|
14367 if (gc == NUL) /* goodword ends, delete badword chars */
|
|
|
14368 {
|
|
|
14369 do
|
|
|
14370 {
|
|
|
14371 if ((score += SCORE_DEL) >= minscore)
|
|
|
14372 goto pop; /* do next alternative */
|
|
|
14373 } while (badword[++bi] != NUL);
|
|
|
14374 minscore = score;
|
|
|
14375 }
|
|
|
14376 else if (bc == NUL) /* badword ends, insert badword chars */
|
|
|
14377 {
|
|
|
14378 do
|
|
|
14379 {
|
|
|
14380 if ((score += SCORE_INS) >= minscore)
|
|
|
14381 goto pop; /* do next alternative */
|
|
|
14382 } while (goodword[++gi] != NUL);
|
|
|
14383 minscore = score;
|
|
|
14384 }
|
|
|
14385 else /* both words continue */
|
|
|
14386 {
|
|
|
14387 /* If not close to the limit, perform a change. Only try changes
|
|
|
14388 * that may lead to a lower score than "minscore".
|
|
|
14389 * round 0: try deleting a char from badword
|
|
|
14390 * round 1: try inserting a char in badword */
|
|
|
14391 for (round = 0; round <= 1; ++round)
|
|
|
14392 {
|
|
|
14393 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS);
|
|
|
14394 if (score_off < minscore)
|
|
|
14395 {
|
|
|
14396 if (score_off + SCORE_EDIT_MIN >= minscore)
|
|
|
14397 {
|
|
|
14398 /* Near the limit, rest of the words must match. We
|
|
|
14399 * can check that right now, no need to push an item
|
|
|
14400 * onto the stack. */
|
|
|
14401 bi2 = bi + 1 - round;
|
|
|
14402 gi2 = gi + round;
|
|
|
14403 while (goodword[gi2] == badword[bi2])
|
|
|
14404 {
|
|
|
14405 if (goodword[gi2] == NUL)
|
|
|
14406 {
|
|
|
14407 minscore = score_off;
|
|
|
14408 break;
|
|
|
14409 }
|
|
|
14410 ++bi2;
|
|
|
14411 ++gi2;
|
|
|
14412 }
|
|
|
14413 }
|
|
|
14414 else
|
|
|
14415 {
|
|
|
14416 /* try deleting/inserting a character later */
|
|
|
14417 stack[stackidx].badi = bi + 1 - round;
|
|
|
14418 stack[stackidx].goodi = gi + round;
|
|
|
14419 stack[stackidx].score = score_off;
|
|
|
14420 ++stackidx;
|
|
|
14421 }
|
|
|
14422 }
|
|
|
14423 }
|
|
|
14424
|
|
|
14425 if (score + SCORE_SWAP < minscore)
|
|
|
14426 {
|
|
|
14427 /* If swapping two characters makes a match then the
|
|
|
14428 * substitution is more expensive, thus there is no need to
|
|
|
14429 * try both. */
|
|
|
14430 if (gc == badword[bi + 1] && bc == goodword[gi + 1])
|
|
|
14431 {
|
|
|
14432 /* Swap two characters, that is: skip them. */
|
|
|
14433 gi += 2;
|
|
|
14434 bi += 2;
|
|
|
14435 score += SCORE_SWAP;
|
|
|
14436 continue;
|
|
|
14437 }
|
|
|
14438 }
|
|
|
14439
|
|
|
14440 /* Substitute one character for another which is the same
|
|
|
14441 * thing as deleting a character from both goodword and badword.
|
|
|
14442 * Use a better score when there is only a case difference. */
|
|
|
14443 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
|
14444 score += SCORE_ICASE;
|
|
|
14445 else
|
|
|
14446 {
|
|
|
14447 /* For a similar character use SCORE_SIMILAR. */
|
|
|
14448 if (slang != NULL
|
|
|
14449 && slang->sl_has_map
|
|
|
14450 && similar_chars(slang, gc, bc))
|
|
|
14451 score += SCORE_SIMILAR;
|
|
|
14452 else
|
|
|
14453 score += SCORE_SUBST;
|
|
|
14454 }
|
|
|
14455
|
|
|
14456 if (score < minscore)
|
|
|
14457 {
|
|
|
14458 /* Do the substitution. */
|
|
|
14459 ++gi;
|
|
|
14460 ++bi;
|
|
|
14461 continue;
|
|
|
14462 }
|
|
|
14463 }
|
|
|
14464 pop:
|
|
|
14465 /*
|
|
|
14466 * Get here to try the next alternative, pop it from the stack.
|
|
|
14467 */
|
|
|
14468 if (stackidx == 0) /* stack is empty, finished */
|
|
|
14469 break;
|
|
|
14470
|
|
|
14471 /* pop an item from the stack */
|
|
|
14472 --stackidx;
|
|
|
14473 gi = stack[stackidx].goodi;
|
|
|
14474 bi = stack[stackidx].badi;
|
|
|
14475 score = stack[stackidx].score;
|
|
|
14476 }
|
|
|
14477
|
|
|
14478 /* When the score goes over "limit" it may actually be much higher.
|
|
|
14479 * Return a very large number to avoid going below the limit when giving a
|
|
|
14480 * bonus. */
|
|
|
14481 if (minscore > limit)
|
|
|
14482 return SCORE_MAXMAX;
|
|
|
14483 return minscore;
|
|
|
14484 }
|
|
|
14485
|
|
|
14486 #ifdef FEAT_MBYTE
|
|
|
14487 /*
|
|
|
14488 * Multi-byte version of spell_edit_score_limit().
|
|
|
14489 * Keep it in sync with the above!
|
|
|
14490 */
|
|
|
14491 static int
|
|
|
14492 spell_edit_score_limit_w(slang, badword, goodword, limit)
|
|
|
14493 slang_T *slang;
|
|
|
14494 char_u *badword;
|
|
|
14495 char_u *goodword;
|
|
|
14496 int limit;
|
|
|
14497 {
|
|
|
14498 limitscore_T stack[10]; /* allow for over 3 * 2 edits */
|
|
|
14499 int stackidx;
|
|
|
14500 int bi, gi;
|
|
|
14501 int bi2, gi2;
|
|
|
14502 int bc, gc;
|
|
|
14503 int score;
|
|
|
14504 int score_off;
|
|
|
14505 int minscore;
|
|
|
14506 int round;
|
|
|
14507 char_u *p;
|
|
|
14508 int wbadword[MAXWLEN];
|
|
|
14509 int wgoodword[MAXWLEN];
|
|
|
14510
|
|
|
14511 /* Get the characters from the multi-byte strings and put them in an
|
|
|
14512 * int array for easy access. */
|
|
|
14513 bi = 0;
|
|
|
14514 for (p = badword; *p != NUL; )
|
|
|
14515 wbadword[bi++] = mb_cptr2char_adv(&p);
|
|
|
14516 wbadword[bi++] = 0;
|
|
|
14517 gi = 0;
|
|
|
14518 for (p = goodword; *p != NUL; )
|
|
|
14519 wgoodword[gi++] = mb_cptr2char_adv(&p);
|
|
|
14520 wgoodword[gi++] = 0;
|
|
|
14521
|
|
|
14522 /*
|
|
|
14523 * The idea is to go from start to end over the words. So long as
|
|
|
14524 * characters are equal just continue, this always gives the lowest score.
|
|
|
14525 * When there is a difference try several alternatives. Each alternative
|
|
|
14526 * increases "score" for the edit distance. Some of the alternatives are
|
|
|
14527 * pushed unto a stack and tried later, some are tried right away. At the
|
|
|
14528 * end of the word the score for one alternative is known. The lowest
|
|
|
14529 * possible score is stored in "minscore".
|
|
|
14530 */
|
|
|
14531 stackidx = 0;
|
|
|
14532 bi = 0;
|
|
|
14533 gi = 0;
|
|
|
14534 score = 0;
|
|
|
14535 minscore = limit + 1;
|
|
|
14536
|
|
|
14537 for (;;)
|
|
|
14538 {
|
|
|
14539 /* Skip over an equal part, score remains the same. */
|
|
|
14540 for (;;)
|
|
|
14541 {
|
|
|
14542 bc = wbadword[bi];
|
|
|
14543 gc = wgoodword[gi];
|
|
|
14544
|
|
|
14545 if (bc != gc) /* stop at a char that's different */
|
|
|
14546 break;
|
|
|
14547 if (bc == NUL) /* both words end */
|
|
|
14548 {
|
|
|
14549 if (score < minscore)
|
|
|
14550 minscore = score;
|
|
|
14551 goto pop; /* do next alternative */
|
|
|
14552 }
|
|
|
14553 ++bi;
|
|
|
14554 ++gi;
|
|
|
14555 }
|
|
|
14556
|
|
|
14557 if (gc == NUL) /* goodword ends, delete badword chars */
|
|
|
14558 {
|
|
|
14559 do
|
|
|
14560 {
|
|
|
14561 if ((score += SCORE_DEL) >= minscore)
|
|
|
14562 goto pop; /* do next alternative */
|
|
|
14563 } while (wbadword[++bi] != NUL);
|
|
|
14564 minscore = score;
|
|
|
14565 }
|
|
|
14566 else if (bc == NUL) /* badword ends, insert badword chars */
|
|
|
14567 {
|
|
|
14568 do
|
|
|
14569 {
|
|
|
14570 if ((score += SCORE_INS) >= minscore)
|
|
|
14571 goto pop; /* do next alternative */
|
|
|
14572 } while (wgoodword[++gi] != NUL);
|
|
|
14573 minscore = score;
|
|
|
14574 }
|
|
|
14575 else /* both words continue */
|
|
|
14576 {
|
|
|
14577 /* If not close to the limit, perform a change. Only try changes
|
|
|
14578 * that may lead to a lower score than "minscore".
|
|
|
14579 * round 0: try deleting a char from badword
|
|
|
14580 * round 1: try inserting a char in badword */
|
|
|
14581 for (round = 0; round <= 1; ++round)
|
|
|
14582 {
|
|
|
14583 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS);
|
|
|
14584 if (score_off < minscore)
|
|
|
14585 {
|
|
|
14586 if (score_off + SCORE_EDIT_MIN >= minscore)
|
|
|
14587 {
|
|
|
14588 /* Near the limit, rest of the words must match. We
|
|
|
14589 * can check that right now, no need to push an item
|
|
|
14590 * onto the stack. */
|
|
|
14591 bi2 = bi + 1 - round;
|
|
|
14592 gi2 = gi + round;
|
|
|
14593 while (wgoodword[gi2] == wbadword[bi2])
|
|
|
14594 {
|
|
|
14595 if (wgoodword[gi2] == NUL)
|
|
|
14596 {
|
|
|
14597 minscore = score_off;
|
|
|
14598 break;
|
|
|
14599 }
|
|
|
14600 ++bi2;
|
|
|
14601 ++gi2;
|
|
|
14602 }
|
|
|
14603 }
|
|
|
14604 else
|
|
|
14605 {
|
|
|
14606 /* try deleting a character from badword later */
|
|
|
14607 stack[stackidx].badi = bi + 1 - round;
|
|
|
14608 stack[stackidx].goodi = gi + round;
|
|
|
14609 stack[stackidx].score = score_off;
|
|
|
14610 ++stackidx;
|
|
|
14611 }
|
|
|
14612 }
|
|
|
14613 }
|
|
|
14614
|
|
|
14615 if (score + SCORE_SWAP < minscore)
|
|
|
14616 {
|
|
|
14617 /* If swapping two characters makes a match then the
|
|
|
14618 * substitution is more expensive, thus there is no need to
|
|
|
14619 * try both. */
|
|
|
14620 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1])
|
|
|
14621 {
|
|
|
14622 /* Swap two characters, that is: skip them. */
|
|
|
14623 gi += 2;
|
|
|
14624 bi += 2;
|
|
|
14625 score += SCORE_SWAP;
|
|
|
14626 continue;
|
|
|
14627 }
|
|
|
14628 }
|
|
|
14629
|
|
|
14630 /* Substitute one character for another which is the same
|
|
|
14631 * thing as deleting a character from both goodword and badword.
|
|
|
14632 * Use a better score when there is only a case difference. */
|
|
|
14633 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
|
14634 score += SCORE_ICASE;
|
|
|
14635 else
|
|
|
14636 {
|
|
|
14637 /* For a similar character use SCORE_SIMILAR. */
|
|
|
14638 if (slang != NULL
|
|
|
14639 && slang->sl_has_map
|
|
|
14640 && similar_chars(slang, gc, bc))
|
|
|
14641 score += SCORE_SIMILAR;
|
|
|
14642 else
|
|
|
14643 score += SCORE_SUBST;
|
|
|
14644 }
|
|
|
14645
|
|
|
14646 if (score < minscore)
|
|
|
14647 {
|
|
|
14648 /* Do the substitution. */
|
|
|
14649 ++gi;
|
|
|
14650 ++bi;
|
|
|
14651 continue;
|
|
|
14652 }
|
|
|
14653 }
|
|
|
14654 pop:
|
|
|
14655 /*
|
|
|
14656 * Get here to try the next alternative, pop it from the stack.
|
|
|
14657 */
|
|
|
14658 if (stackidx == 0) /* stack is empty, finished */
|
|
|
14659 break;
|
|
|
14660
|
|
|
14661 /* pop an item from the stack */
|
|
|
14662 --stackidx;
|
|
|
14663 gi = stack[stackidx].goodi;
|
|
|
14664 bi = stack[stackidx].badi;
|
|
|
14665 score = stack[stackidx].score;
|
|
|
14666 }
|
|
|
14667
|
|
|
14668 /* When the score goes over "limit" it may actually be much higher.
|
|
|
14669 * Return a very large number to avoid going below the limit when giving a
|
|
|
14670 * bonus. */
|
|
|
14671 if (minscore > limit)
|
|
|
14672 return SCORE_MAXMAX;
|
|
|
14673 return minscore;
|
|
|
14674 }
|
|
|
14675 #endif
|
|
|
14676
|
|
|
14677 #define DUMPFLAG_KEEPCASE 1 /* round 2: keep-case tree */
|
|
|
14678 #define DUMPFLAG_COUNT 2 /* include word count */
|
|
|
14679
|
|
351
|
14680 /*
|
|
|
14681 * ":spelldump"
|
|
|
14682 */
|
|
|
14683 /*ARGSUSED*/
|
|
|
14684 void
|
|
|
14685 ex_spelldump(eap)
|
|
|
14686 exarg_T *eap;
|
|
|
14687 {
|
|
|
14688 buf_T *buf = curbuf;
|
|
|
14689 langp_T *lp;
|
|
|
14690 slang_T *slang;
|
|
|
14691 idx_T arridx[MAXWLEN];
|
|
|
14692 int curi[MAXWLEN];
|
|
|
14693 char_u word[MAXWLEN];
|
|
|
14694 int c;
|
|
|
14695 char_u *byts;
|
|
|
14696 idx_T *idxs;
|
|
|
14697 linenr_T lnum = 0;
|
|
|
14698 int round;
|
|
|
14699 int depth;
|
|
|
14700 int n;
|
|
|
14701 int flags;
|
|
381
|
14702 char_u *region_names = NULL; /* region names being used */
|
|
|
14703 int do_region = TRUE; /* dump region names and numbers */
|
|
|
14704 char_u *p;
|
|
500
|
14705 int lpi;
|
|
625
|
14706 int dumpflags;
|
|
351
|
14707
|
|
498
|
14708 if (no_spell_checking(curwin))
|
|
351
|
14709 return;
|
|
|
14710
|
|
|
14711 /* Create a new empty buffer by splitting the window. */
|
|
|
14712 do_cmdline_cmd((char_u *)"new");
|
|
|
14713 if (!bufempty() || !buf_valid(buf))
|
|
|
14714 return;
|
|
|
14715
|
|
381
|
14716 /* Find out if we can support regions: All languages must support the same
|
|
|
14717 * regions or none at all. */
|
|
500
|
14718 for (lpi = 0; lpi < buf->b_langp.ga_len; ++lpi)
|
|
|
14719 {
|
|
|
14720 lp = LANGP_ENTRY(buf->b_langp, lpi);
|
|
381
|
14721 p = lp->lp_slang->sl_regions;
|
|
|
14722 if (p[0] != 0)
|
|
|
14723 {
|
|
|
14724 if (region_names == NULL) /* first language with regions */
|
|
|
14725 region_names = p;
|
|
|
14726 else if (STRCMP(region_names, p) != 0)
|
|
|
14727 {
|
|
|
14728 do_region = FALSE; /* region names are different */
|
|
|
14729 break;
|
|
|
14730 }
|
|
|
14731 }
|
|
|
14732 }
|
|
|
14733
|
|
|
14734 if (do_region && region_names != NULL)
|
|
|
14735 {
|
|
|
14736 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
14737 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
14738 }
|
|
|
14739 else
|
|
|
14740 do_region = FALSE;
|
|
|
14741
|
|
|
14742 /*
|
|
|
14743 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
14744 */
|
|
500
|
14745 for (lpi = 0; lpi < buf->b_langp.ga_len; ++lpi)
|
|
|
14746 {
|
|
|
14747 lp = LANGP_ENTRY(buf->b_langp, lpi);
|
|
351
|
14748 slang = lp->lp_slang;
|
|
500
|
14749 if (slang->sl_fbyts == NULL) /* reloading failed */
|
|
|
14750 continue;
|
|
351
|
14751
|
|
|
14752 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
14753 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
14754
|
|
|
14755 /* round 1: case-folded tree
|
|
|
14756 * round 2: keep-case tree */
|
|
|
14757 for (round = 1; round <= 2; ++round)
|
|
|
14758 {
|
|
|
14759 if (round == 1)
|
|
|
14760 {
|
|
625
|
14761 dumpflags = 0;
|
|
351
|
14762 byts = slang->sl_fbyts;
|
|
|
14763 idxs = slang->sl_fidxs;
|
|
|
14764 }
|
|
|
14765 else
|
|
|
14766 {
|
|
625
|
14767 dumpflags = DUMPFLAG_KEEPCASE;
|
|
351
|
14768 byts = slang->sl_kbyts;
|
|
|
14769 idxs = slang->sl_kidxs;
|
|
|
14770 }
|
|
|
14771 if (byts == NULL)
|
|
|
14772 continue; /* array is empty */
|
|
|
14773
|
|
625
|
14774 if (eap->forceit)
|
|
|
14775 dumpflags |= DUMPFLAG_COUNT;
|
|
|
14776
|
|
351
|
14777 depth = 0;
|
|
|
14778 arridx[0] = 0;
|
|
|
14779 curi[0] = 1;
|
|
|
14780 while (depth >= 0 && !got_int)
|
|
|
14781 {
|
|
|
14782 if (curi[depth] > byts[arridx[depth]])
|
|
|
14783 {
|
|
|
14784 /* Done all bytes at this node, go up one level. */
|
|
|
14785 --depth;
|
|
|
14786 line_breakcheck();
|
|
|
14787 }
|
|
|
14788 else
|
|
|
14789 {
|
|
|
14790 /* Do one more byte at this node. */
|
|
|
14791 n = arridx[depth] + curi[depth];
|
|
|
14792 ++curi[depth];
|
|
|
14793 c = byts[n];
|
|
|
14794 if (c == 0)
|
|
|
14795 {
|
|
|
14796 /* End of word, deal with the word.
|
|
|
14797 * Don't use keep-case words in the fold-case tree,
|
|
|
14798 * they will appear in the keep-case tree.
|
|
|
14799 * Only use the word when the region matches. */
|
|
|
14800 flags = (int)idxs[n];
|
|
|
14801 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
500
|
14802 && (flags & WF_NEEDCOMP) == 0
|
|
381
|
14803 && (do_region
|
|
|
14804 || (flags & WF_REGION) == 0
|
|
390
|
14805 || (((unsigned)flags >> 16)
|
|
351
|
14806 & lp->lp_region) != 0))
|
|
|
14807 {
|
|
|
14808 word[depth] = NUL;
|
|
381
|
14809 if (!do_region)
|
|
|
14810 flags &= ~WF_REGION;
|
|
355
|
14811
|
|
|
14812 /* Dump the basic word if there is no prefix or
|
|
|
14813 * when it's the first one. */
|
|
390
|
14814 c = (unsigned)flags >> 24;
|
|
355
|
14815 if (c == 0 || curi[depth] == 2)
|
|
625
|
14816 dump_word(slang, word, dumpflags,
|
|
|
14817 flags, lnum++);
|
|
351
|
14818
|
|
|
14819 /* Apply the prefix, if there is one. */
|
|
355
|
14820 if (c != 0)
|
|
625
|
14821 lnum = dump_prefixes(slang, word, dumpflags,
|
|
351
|
14822 flags, lnum);
|
|
|
14823 }
|
|
|
14824 }
|
|
|
14825 else
|
|
|
14826 {
|
|
|
14827 /* Normal char, go one level deeper. */
|
|
|
14828 word[depth++] = c;
|
|
|
14829 arridx[depth] = idxs[n];
|
|
|
14830 curi[depth] = 1;
|
|
|
14831 }
|
|
|
14832 }
|
|
|
14833 }
|
|
|
14834 }
|
|
|
14835 }
|
|
|
14836
|
|
|
14837 /* Delete the empty line that we started with. */
|
|
|
14838 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
14839 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
14840
|
|
|
14841 redraw_later(NOT_VALID);
|
|
|
14842 }
|
|
|
14843
|
|
|
14844 /*
|
|
|
14845 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
14846 */
|
|
|
14847 static void
|
|
625
|
14848 dump_word(slang, word, dumpflags, flags, lnum)
|
|
|
14849 slang_T *slang;
|
|
351
|
14850 char_u *word;
|
|
625
|
14851 int dumpflags;
|
|
351
|
14852 int flags;
|
|
|
14853 linenr_T lnum;
|
|
|
14854 {
|
|
|
14855 int keepcap = FALSE;
|
|
|
14856 char_u *p;
|
|
625
|
14857 char_u *tw;
|
|
351
|
14858 char_u cword[MAXWLEN];
|
|
381
|
14859 char_u badword[MAXWLEN + 10];
|
|
|
14860 int i;
|
|
351
|
14861
|
|
625
|
14862 if ((dumpflags & DUMPFLAG_KEEPCASE) == 0 && (flags & WF_CAPMASK) != 0)
|
|
351
|
14863 {
|
|
|
14864 /* Need to fix case according to "flags". */
|
|
|
14865 make_case_word(word, cword, flags);
|
|
|
14866 p = cword;
|
|
|
14867 }
|
|
|
14868 else
|
|
|
14869 {
|
|
|
14870 p = word;
|
|
625
|
14871 if ((dumpflags & DUMPFLAG_KEEPCASE)
|
|
|
14872 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
389
|
14873 || (flags & WF_FIXCAP) != 0))
|
|
351
|
14874 keepcap = TRUE;
|
|
|
14875 }
|
|
625
|
14876 tw = p;
|
|
351
|
14877
|
|
381
|
14878 /* Add flags and regions after a slash. */
|
|
|
14879 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
14880 {
|
|
|
14881 STRCPY(badword, p);
|
|
|
14882 STRCAT(badword, "/");
|
|
351
|
14883 if (keepcap)
|
|
|
14884 STRCAT(badword, "=");
|
|
|
14885 if (flags & WF_BANNED)
|
|
|
14886 STRCAT(badword, "!");
|
|
|
14887 else if (flags & WF_RARE)
|
|
|
14888 STRCAT(badword, "?");
|
|
381
|
14889 if (flags & WF_REGION)
|
|
|
14890 for (i = 0; i < 7; ++i)
|
|
390
|
14891 if (flags & (0x10000 << i))
|
|
381
|
14892 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
14893 p = badword;
|
|
|
14894 }
|
|
|
14895
|
|
625
|
14896 if (dumpflags & DUMPFLAG_COUNT)
|
|
|
14897 {
|
|
|
14898 hashitem_T *hi;
|
|
|
14899
|
|
|
14900 /* Include the word count for ":spelldump!". */
|
|
|
14901 hi = hash_find(&slang->sl_wordcount, tw);
|
|
|
14902 if (!HASHITEM_EMPTY(hi))
|
|
|
14903 {
|
|
|
14904 vim_snprintf((char *)IObuff, IOSIZE, "%s\t%d",
|
|
|
14905 tw, HI2WC(hi)->wc_count);
|
|
|
14906 p = IObuff;
|
|
|
14907 }
|
|
|
14908 }
|
|
|
14909
|
|
351
|
14910 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
14911 }
|
|
|
14912
|
|
|
14913 /*
|
|
372
|
14914 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
14915 * "word" and append a line to the buffer.
|
|
351
|
14916 * Return the updated line number.
|
|
|
14917 */
|
|
|
14918 static linenr_T
|
|
625
|
14919 dump_prefixes(slang, word, dumpflags, flags, startlnum)
|
|
351
|
14920 slang_T *slang;
|
|
|
14921 char_u *word; /* case-folded word */
|
|
625
|
14922 int dumpflags;
|
|
351
|
14923 int flags; /* flags with prefix ID */
|
|
|
14924 linenr_T startlnum;
|
|
|
14925 {
|
|
|
14926 idx_T arridx[MAXWLEN];
|
|
|
14927 int curi[MAXWLEN];
|
|
|
14928 char_u prefix[MAXWLEN];
|
|
455
|
14929 char_u word_up[MAXWLEN];
|
|
|
14930 int has_word_up = FALSE;
|
|
351
|
14931 int c;
|
|
|
14932 char_u *byts;
|
|
|
14933 idx_T *idxs;
|
|
|
14934 linenr_T lnum = startlnum;
|
|
|
14935 int depth;
|
|
|
14936 int n;
|
|
|
14937 int len;
|
|
|
14938 int i;
|
|
|
14939
|
|
626
|
14940 /* If the word starts with a lower-case letter make the word with an
|
|
455
|
14941 * upper-case letter in word_up[]. */
|
|
|
14942 c = PTR2CHAR(word);
|
|
|
14943 if (SPELL_TOUPPER(c) != c)
|
|
|
14944 {
|
|
|
14945 onecap_copy(word, word_up, TRUE);
|
|
|
14946 has_word_up = TRUE;
|
|
|
14947 }
|
|
|
14948
|
|
351
|
14949 byts = slang->sl_pbyts;
|
|
|
14950 idxs = slang->sl_pidxs;
|
|
|
14951 if (byts != NULL) /* array not is empty */
|
|
|
14952 {
|
|
|
14953 /*
|
|
|
14954 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
14955 * When at the end of a prefix check that it supports "flags".
|
|
351
|
14956 */
|
|
|
14957 depth = 0;
|
|
|
14958 arridx[0] = 0;
|
|
|
14959 curi[0] = 1;
|
|
|
14960 while (depth >= 0 && !got_int)
|
|
|
14961 {
|
|
390
|
14962 n = arridx[depth];
|
|
|
14963 len = byts[n];
|
|
|
14964 if (curi[depth] > len)
|
|
351
|
14965 {
|
|
|
14966 /* Done all bytes at this node, go up one level. */
|
|
|
14967 --depth;
|
|
|
14968 line_breakcheck();
|
|
|
14969 }
|
|
|
14970 else
|
|
|
14971 {
|
|
|
14972 /* Do one more byte at this node. */
|
|
390
|
14973 n += curi[depth];
|
|
351
|
14974 ++curi[depth];
|
|
|
14975 c = byts[n];
|
|
|
14976 if (c == 0)
|
|
|
14977 {
|
|
|
14978 /* End of prefix, find out how many IDs there are. */
|
|
|
14979 for (i = 1; i < len; ++i)
|
|
|
14980 if (byts[n + i] != 0)
|
|
|
14981 break;
|
|
|
14982 curi[depth] += i - 1;
|
|
|
14983
|
|
455
|
14984 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
14985 if (c != 0)
|
|
351
|
14986 {
|
|
376
|
14987 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
625
|
14988 dump_word(slang, prefix, dumpflags,
|
|
455
|
14989 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
14990 : flags, lnum++);
|
|
351
|
14991 }
|
|
455
|
14992
|
|
|
14993 /* Check for prefix that matches the word when the
|
|
|
14994 * first letter is upper-case, but only if the prefix has
|
|
|
14995 * a condition. */
|
|
|
14996 if (has_word_up)
|
|
|
14997 {
|
|
|
14998 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
14999 TRUE);
|
|
|
15000 if (c != 0)
|
|
|
15001 {
|
|
|
15002 vim_strncpy(prefix + depth, word_up,
|
|
|
15003 MAXWLEN - depth - 1);
|
|
625
|
15004 dump_word(slang, prefix, dumpflags,
|
|
455
|
15005 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
15006 : flags, lnum++);
|
|
|
15007 }
|
|
|
15008 }
|
|
351
|
15009 }
|
|
|
15010 else
|
|
|
15011 {
|
|
|
15012 /* Normal char, go one level deeper. */
|
|
|
15013 prefix[depth++] = c;
|
|
|
15014 arridx[depth] = idxs[n];
|
|
|
15015 curi[depth] = 1;
|
|
|
15016 }
|
|
|
15017 }
|
|
|
15018 }
|
|
|
15019 }
|
|
|
15020
|
|
|
15021 return lnum;
|
|
|
15022 }
|
|
|
15023
|
|
498
|
15024 /*
|
|
626
|
15025 * Move "p" to the end of word "start".
|
|
|
15026 * Uses the spell-checking word characters.
|
|
498
|
15027 */
|
|
|
15028 char_u *
|
|
|
15029 spell_to_word_end(start, buf)
|
|
|
15030 char_u *start;
|
|
|
15031 buf_T *buf;
|
|
|
15032 {
|
|
|
15033 char_u *p = start;
|
|
|
15034
|
|
|
15035 while (*p != NUL && spell_iswordp(p, buf))
|
|
|
15036 mb_ptr_adv(p);
|
|
|
15037 return p;
|
|
|
15038 }
|
|
|
15039
|
|
475
|
15040 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
15041 /*
|
|
626
|
15042 * For Insert mode completion CTRL-X s:
|
|
|
15043 * Find start of the word in front of column "startcol".
|
|
|
15044 * We don't check if it is badly spelled, with completion we can only change
|
|
|
15045 * the word in front of the cursor.
|
|
475
|
15046 * Returns the column number of the word.
|
|
|
15047 */
|
|
|
15048 int
|
|
|
15049 spell_word_start(startcol)
|
|
|
15050 int startcol;
|
|
|
15051 {
|
|
|
15052 char_u *line;
|
|
|
15053 char_u *p;
|
|
|
15054 int col = 0;
|
|
|
15055
|
|
498
|
15056 if (no_spell_checking(curwin))
|
|
475
|
15057 return startcol;
|
|
|
15058
|
|
|
15059 /* Find a word character before "startcol". */
|
|
|
15060 line = ml_get_curline();
|
|
|
15061 for (p = line + startcol; p > line; )
|
|
|
15062 {
|
|
|
15063 mb_ptr_back(line, p);
|
|
|
15064 if (spell_iswordp_nmw(p))
|
|
|
15065 break;
|
|
|
15066 }
|
|
|
15067
|
|
|
15068 /* Go back to start of the word. */
|
|
|
15069 while (p > line)
|
|
|
15070 {
|
|
|
15071 col = p - line;
|
|
|
15072 mb_ptr_back(line, p);
|
|
|
15073 if (!spell_iswordp(p, curbuf))
|
|
|
15074 break;
|
|
|
15075 col = 0;
|
|
|
15076 }
|
|
|
15077
|
|
535
|
15078 return col;
|
|
|
15079 }
|
|
|
15080
|
|
|
15081 /*
|
|
|
15082 * Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
15083 * expand_spelling() is called. Therefore the ugly global variable.
|
|
|
15084 */
|
|
|
15085 static int spell_expand_need_cap;
|
|
|
15086
|
|
|
15087 void
|
|
|
15088 spell_expand_check_cap(col)
|
|
|
15089 colnr_T col;
|
|
|
15090 {
|
|
475
|
15091 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
15092 }
|
|
|
15093
|
|
|
15094 /*
|
|
|
15095 * Get list of spelling suggestions.
|
|
|
15096 * Used for Insert mode completion CTRL-X ?.
|
|
|
15097 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
15098 * allocated strings.
|
|
|
15099 */
|
|
|
15100 /*ARGSUSED*/
|
|
|
15101 int
|
|
|
15102 expand_spelling(lnum, col, pat, matchp)
|
|
|
15103 linenr_T lnum;
|
|
|
15104 int col;
|
|
|
15105 char_u *pat;
|
|
|
15106 char_u ***matchp;
|
|
|
15107 {
|
|
|
15108 garray_T ga;
|
|
|
15109
|
|
625
|
15110 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap, TRUE);
|
|
475
|
15111 *matchp = ga.ga_data;
|
|
|
15112 return ga.ga_len;
|
|
|
15113 }
|
|
|
15114 #endif
|
|
|
15115
|
|
236
|
15116 #endif /* FEAT_SYN_HL */
|
