|
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 *
|
|
714
|
120 * sectionID == SN_INFO: <infotext>
|
|
|
121 * <infotext> N bytes free format text with spell file info (version,
|
|
|
122 * website, etc)
|
|
|
123 *
|
|
483
|
124 * sectionID == SN_REGION: <regionname> ...
|
|
|
125 * <regionname> 2 bytes Up to 8 region names: ca, au, etc. Lower case.
|
|
300
|
126 * First <regionname> is region 1.
|
|
|
127 *
|
|
483
|
128 * sectionID == SN_CHARFLAGS: <charflagslen> <charflags>
|
|
|
129 * <folcharslen> <folchars>
|
|
300
|
130 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
131 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
132 * 0x01 word character CF_WORD
|
|
|
133 * 0x02 upper-case character CF_UPPER
|
|
483
|
134 * <folcharslen> 2 bytes Number of bytes in <folchars>.
|
|
|
135 * <folchars> N bytes Folded characters, first one is for character 128.
|
|
300
|
136 *
|
|
483
|
137 * sectionID == SN_MIDWORD: <midword>
|
|
|
138 * <midword> N bytes Characters that are word characters only when used
|
|
366
|
139 * in the middle of a word.
|
|
|
140 *
|
|
483
|
141 * sectionID == SN_PREFCOND: <prefcondcnt> <prefcond> ...
|
|
339
|
142 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
143 * <prefcond> : <condlen> <condstr>
|
|
|
144 * <condlen> 1 byte Length of <condstr>.
|
|
|
145 * <condstr> N bytes Condition for the prefix.
|
|
|
146 *
|
|
483
|
147 * sectionID == SN_REP: <repcount> <rep> ...
|
|
|
148 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
323
|
149 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
483
|
150 * <repfromlen> 1 byte length of <repfrom>
|
|
|
151 * <repfrom> N bytes "from" part of replacement
|
|
|
152 * <reptolen> 1 byte length of <repto>
|
|
|
153 * <repto> N bytes "to" part of replacement
|
|
323
|
154 *
|
|
625
|
155 * sectionID == SN_REPSAL: <repcount> <rep> ...
|
|
|
156 * just like SN_REP but for soundfolded words
|
|
|
157 *
|
|
483
|
158 * sectionID == SN_SAL: <salflags> <salcount> <sal> ...
|
|
|
159 * <salflags> 1 byte flags for soundsalike conversion:
|
|
323
|
160 * SAL_F0LLOWUP
|
|
|
161 * SAL_COLLAPSE
|
|
|
162 * SAL_REM_ACCENTS
|
|
483
|
163 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
164 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
483
|
165 * <salfromlen> 1 byte length of <salfrom>
|
|
|
166 * <salfrom> N bytes "from" part of soundsalike
|
|
|
167 * <saltolen> 1 byte length of <salto>
|
|
|
168 * <salto> N bytes "to" part of soundsalike
|
|
323
|
169 *
|
|
483
|
170 * sectionID == SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
171 * <sofofromlen> 2 bytes length of <sofofrom>
|
|
|
172 * <sofofrom> N bytes "from" part of soundfold
|
|
|
173 * <sofotolen> 2 bytes length of <sofoto>
|
|
|
174 * <sofoto> N bytes "to" part of soundfold
|
|
323
|
175 *
|
|
625
|
176 * sectionID == SN_SUGFILE: <timestamp>
|
|
|
177 * <timestamp> 8 bytes time in seconds that must match with .sug file
|
|
|
178 *
|
|
698
|
179 * sectionID == SN_NOSPLITSUGS: nothing
|
|
|
180 *
|
|
625
|
181 * sectionID == SN_WORDS: <word> ...
|
|
|
182 * <word> N bytes NUL terminated common word
|
|
|
183 *
|
|
483
|
184 * sectionID == SN_MAP: <mapstr>
|
|
|
185 * <mapstr> N bytes String with sequences of similar characters,
|
|
323
|
186 * separated by slashes.
|
|
300
|
187 *
|
|
483
|
188 * sectionID == SN_COMPOUND: <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
189 * <compmax> 1 byte Maximum nr of words in compound word.
|
|
|
190 * <compminlen> 1 byte Minimal word length for compounding.
|
|
|
191 * <compsylmax> 1 byte Maximum nr of syllables in compound word.
|
|
714
|
192 * <compflags> N bytes Flags from COMPOUNDRULE items, separated by
|
|
483
|
193 * slashes.
|
|
|
194 *
|
|
492
|
195 * sectionID == SN_NOBREAK: (empty, its presence is enough)
|
|
|
196 *
|
|
483
|
197 * sectionID == SN_SYLLABLE: <syllable>
|
|
|
198 * <syllable> N bytes String from SYLLABLE item.
|
|
300
|
199 *
|
|
|
200 * <LWORDTREE>: <wordtree>
|
|
|
201 *
|
|
339
|
202 * <KWORDTREE>: <wordtree>
|
|
|
203 *
|
|
|
204 * <PREFIXTREE>: <wordtree>
|
|
|
205 *
|
|
|
206 *
|
|
300
|
207 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
208 *
|
|
|
209 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
210 *
|
|
|
211 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
212 *
|
|
|
213 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
214 * follow in sorted order.
|
|
|
215 *
|
|
339
|
216 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
481
|
217 * | <flags> [<flags2>] [<region>] [<affixID>]
|
|
|
218 * | [<pflags>] <affixID> <prefcondnr> ]
|
|
300
|
219 *
|
|
|
220 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
221 * BY_NOFLAGS: End of word without flags and for all
|
|
|
222 * regions.
|
|
481
|
223 * For PREFIXTREE <affixID> and
|
|
366
|
224 * <prefcondnr> follow.
|
|
390
|
225 * BY_FLAGS: End of word, <flags> follow.
|
|
481
|
226 * For PREFIXTREE <pflags>, <affixID>
|
|
455
|
227 * and <prefcondnr> follow.
|
|
390
|
228 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
229 * follow. Not used in PREFIXTREE.
|
|
|
230 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
231 * and <xbyte> follow.
|
|
|
232 *
|
|
|
233 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
234 *
|
|
|
235 * <xbyte> 1 byte byte value of the sibling.
|
|
|
236 *
|
|
|
237 * <flags> 1 byte bitmask of:
|
|
|
238 * WF_ALLCAP word must have only capitals
|
|
|
239 * WF_ONECAP first char of word must be capital
|
|
389
|
240 * WF_KEEPCAP keep-case word
|
|
|
241 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
242 * WF_RARE rare word
|
|
389
|
243 * WF_BANNED bad word
|
|
300
|
244 * WF_REGION <region> follows
|
|
481
|
245 * WF_AFX <affixID> follows
|
|
300
|
246 *
|
|
500
|
247 * <flags2> 1 byte Bitmask of:
|
|
390
|
248 * WF_HAS_AFF >> 8 word includes affix
|
|
500
|
249 * WF_NEEDCOMP >> 8 word only valid in compound
|
|
698
|
250 * WF_NOSUGGEST >> 8 word not used for suggestions
|
|
390
|
251 *
|
|
455
|
252 * <pflags> 1 byte bitmask of:
|
|
|
253 * WFP_RARE rare prefix
|
|
|
254 * WFP_NC non-combining prefix
|
|
|
255 * WFP_UP letter after prefix made upper case
|
|
|
256 *
|
|
300
|
257 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
258 * omitted it's valid in all regions.
|
|
|
259 * Lowest bit is for region 1.
|
|
|
260 *
|
|
481
|
261 * <affixID> 1 byte ID of affix that can be used with this word. In
|
|
339
|
262 * PREFIXTREE used for the required prefix ID.
|
|
|
263 *
|
|
|
264 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
265 * from HEADER.
|
|
300
|
266 *
|
|
|
267 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
268 */
|
|
|
269
|
|
625
|
270 /*
|
|
|
271 * Vim .sug file format: <SUGHEADER>
|
|
|
272 * <SUGWORDTREE>
|
|
|
273 * <SUGTABLE>
|
|
|
274 *
|
|
|
275 * <SUGHEADER>: <fileID> <versionnr> <timestamp>
|
|
|
276 *
|
|
|
277 * <fileID> 6 bytes "VIMsug"
|
|
|
278 * <versionnr> 1 byte VIMSUGVERSION
|
|
|
279 * <timestamp> 8 bytes timestamp that must match with .spl file
|
|
|
280 *
|
|
|
281 *
|
|
|
282 * <SUGWORDTREE>: <wordtree> (see above, no flags or region used)
|
|
|
283 *
|
|
|
284 *
|
|
|
285 * <SUGTABLE>: <sugwcount> <sugline> ...
|
|
|
286 *
|
|
|
287 * <sugwcount> 4 bytes number of <sugline> following
|
|
|
288 *
|
|
|
289 * <sugline>: <sugnr> ... NUL
|
|
|
290 *
|
|
|
291 * <sugnr>: X bytes word number that results in this soundfolded word,
|
|
|
292 * stored as an offset to the previous number in as
|
|
|
293 * few bytes as possible, see offset2bytes())
|
|
|
294 */
|
|
|
295
|
|
223
|
296 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
714
|
297 # include "vimio.h" /* for lseek(), must be before vim.h */
|
|
223
|
298 #endif
|
|
|
299
|
|
|
300 #include "vim.h"
|
|
|
301
|
|
|
302 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
303
|
|
|
304 #ifdef HAVE_FCNTL_H
|
|
|
305 # include <fcntl.h>
|
|
|
306 #endif
|
|
|
307
|
|
625
|
308 #ifndef UNIX /* it's in os_unix.h for Unix */
|
|
|
309 # include <time.h> /* for time_t */
|
|
|
310 #endif
|
|
|
311
|
|
323
|
312 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
313 Some places assume a word length fits in a
|
|
|
314 byte, thus it can't be above 255. */
|
|
226
|
315
|
|
490
|
316 /* Type used for indexes in the word tree need to be at least 4 bytes. If int
|
|
324
|
317 * is 8 bytes we could use something smaller, but what? */
|
|
490
|
318 #if SIZEOF_INT > 3
|
|
324
|
319 typedef int idx_T;
|
|
|
320 #else
|
|
|
321 typedef long idx_T;
|
|
|
322 #endif
|
|
|
323
|
|
|
324 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
325 * comes above it. */
|
|
300
|
326 #define WF_REGION 0x01 /* region byte follows */
|
|
|
327 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
328 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
329 #define WF_RARE 0x08 /* rare word */
|
|
307
|
330 #define WF_BANNED 0x10 /* bad word */
|
|
481
|
331 #define WF_AFX 0x20 /* affix ID follows */
|
|
389
|
332 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
333 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
334
|
|
390
|
335 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
336 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
500
|
337 #define WF_NEEDCOMP 0x0200 /* word only valid in compound */
|
|
698
|
338 #define WF_NOSUGGEST 0x0400 /* word not to be suggested */
|
|
390
|
339
|
|
633
|
340 /* only used for su_badflags */
|
|
|
341 #define WF_MIXCAP 0x20 /* mix of upper and lower case: macaRONI */
|
|
|
342
|
|
389
|
343 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
344
|
|
455
|
345 /* flags for <pflags> */
|
|
|
346 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
347 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
348 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
349
|
|
481
|
350 /* Flags for postponed prefixes. Must be above affixID (one byte)
|
|
390
|
351 * and prefcondnr (two bytes). */
|
|
455
|
352 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
353 * postponed prefix */
|
|
|
354 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
355 * postponed prefix */
|
|
|
356 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
357 * postponed prefix */
|
|
390
|
358
|
|
|
359 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
360 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
361 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
362 * postponed prefix: no <pflags> */
|
|
|
363 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
364 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
365 * postponed prefix: <pflags> follows */
|
|
|
366 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
367 * follow; never used in prefix tree */
|
|
|
368 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
369
|
|
625
|
370 /* Info from "REP", "REPSAL" and "SAL" entries in ".aff" file used in si_rep,
|
|
|
371 * si_repsal, sl_rep, and si_sal. Not for sl_sal!
|
|
323
|
372 * One replacement: from "ft_from" to "ft_to". */
|
|
|
373 typedef struct fromto_S
|
|
236
|
374 {
|
|
323
|
375 char_u *ft_from;
|
|
|
376 char_u *ft_to;
|
|
|
377 } fromto_T;
|
|
236
|
378
|
|
344
|
379 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
380 * The info is split for quick processing by spell_soundfold().
|
|
|
381 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
382 typedef struct salitem_S
|
|
|
383 {
|
|
|
384 char_u *sm_lead; /* leading letters */
|
|
|
385 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
386 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
387 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
388 char_u *sm_to; /* replacement. */
|
|
372
|
389 #ifdef FEAT_MBYTE
|
|
|
390 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
391 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
392 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
393 #endif
|
|
344
|
394 } salitem_T;
|
|
|
395
|
|
375
|
396 #ifdef FEAT_MBYTE
|
|
|
397 typedef int salfirst_T;
|
|
|
398 #else
|
|
|
399 typedef short salfirst_T;
|
|
|
400 #endif
|
|
|
401
|
|
483
|
402 /* Values for SP_*ERROR are negative, positive values are used by
|
|
|
403 * read_cnt_string(). */
|
|
|
404 #define SP_TRUNCERROR -1 /* spell file truncated error */
|
|
|
405 #define SP_FORMERROR -2 /* format error in spell file */
|
|
495
|
406 #define SP_OTHERERROR -3 /* other error while reading spell file */
|
|
483
|
407
|
|
236
|
408 /*
|
|
243
|
409 * Structure used to store words and other info for one language, loaded from
|
|
|
410 * a .spl file.
|
|
300
|
411 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
412 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
413 *
|
|
|
414 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
415 * the number of possible bytes, sorted on byte value:
|
|
|
416 * <len> <byte1> <byte2> ...
|
|
|
417 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
418 * byte in "byts".
|
|
|
419 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
481
|
420 * the flags, region mask and affixID for the word. There may be several
|
|
|
421 * zeros in sequence for alternative flag/region/affixID combinations.
|
|
236
|
422 */
|
|
|
423 typedef struct slang_S slang_T;
|
|
|
424 struct slang_S
|
|
|
425 {
|
|
|
426 slang_T *sl_next; /* next language */
|
|
|
427 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
428 char_u *sl_fname; /* name of .spl file */
|
|
323
|
429 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
430
|
|
300
|
431 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
432 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
433 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
434 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
435 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
436 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
437
|
|
714
|
438 char_u *sl_info; /* infotext string or NULL */
|
|
|
439
|
|
236
|
440 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
441
|
|
376
|
442 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
443
|
|
625
|
444 hashtab_T sl_wordcount; /* hashtable with word count, wordcount_T */
|
|
|
445
|
|
483
|
446 int sl_compmax; /* COMPOUNDMAX (default: MAXWLEN) */
|
|
501
|
447 int sl_compminlen; /* COMPOUNDMIN (default: 0) */
|
|
483
|
448 int sl_compsylmax; /* COMPOUNDSYLMAX (default: MAXWLEN) */
|
|
714
|
449 regprog_T *sl_compprog; /* COMPOUNDRULE turned into a regexp progrm
|
|
483
|
450 * (NULL when no compounding) */
|
|
|
451 char_u *sl_compstartflags; /* flags for first compound word */
|
|
485
|
452 char_u *sl_compallflags; /* all flags for compound words */
|
|
492
|
453 char_u sl_nobreak; /* When TRUE: no spaces between words */
|
|
483
|
454 char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */
|
|
|
455 garray_T sl_syl_items; /* syllable items */
|
|
481
|
456
|
|
339
|
457 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
458 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
459
|
|
323
|
460 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
461 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
462 there is none */
|
|
344
|
463 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
464 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
465 there is none */
|
|
625
|
466 int sl_followup; /* SAL followup */
|
|
|
467 int sl_collapse; /* SAL collapse_result */
|
|
|
468 int sl_rem_accents; /* SAL remove_accents */
|
|
375
|
469 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
470 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
471 * "sl_sal" is a list of wide char lists. */
|
|
625
|
472 garray_T sl_repsal; /* list of fromto_T entries from REPSAL lines */
|
|
|
473 short sl_repsal_first[256]; /* sl_rep_first for REPSAL lines */
|
|
698
|
474 int sl_nosplitsugs; /* don't suggest splitting a word */
|
|
625
|
475
|
|
|
476 /* Info from the .sug file. Loaded on demand. */
|
|
|
477 time_t sl_sugtime; /* timestamp for .sug file */
|
|
|
478 char_u *sl_sbyts; /* soundfolded word bytes */
|
|
|
479 idx_T *sl_sidxs; /* soundfolded word indexes */
|
|
|
480 buf_T *sl_sugbuf; /* buffer with word number table */
|
|
|
481 int sl_sugloaded; /* TRUE when .sug file was loaded or failed to
|
|
|
482 load */
|
|
|
483
|
|
330
|
484 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
485 #ifdef FEAT_MBYTE
|
|
|
486 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
487 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
488 #else
|
|
|
489 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
490 #endif
|
|
625
|
491 hashtab_T sl_sounddone; /* table with soundfolded words that have
|
|
|
492 handled, see add_sound_suggest() */
|
|
236
|
493 };
|
|
|
494
|
|
243
|
495 /* First language that is loaded, start of the linked list of loaded
|
|
|
496 * languages. */
|
|
236
|
497 static slang_T *first_lang = NULL;
|
|
|
498
|
|
323
|
499 /* Flags used in .spl file for soundsalike flags. */
|
|
|
500 #define SAL_F0LLOWUP 1
|
|
|
501 #define SAL_COLLAPSE 2
|
|
|
502 #define SAL_REM_ACCENTS 4
|
|
|
503
|
|
236
|
504 /*
|
|
|
505 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
506 */
|
|
|
507 typedef struct langp_S
|
|
|
508 {
|
|
503
|
509 slang_T *lp_slang; /* info for this language */
|
|
|
510 slang_T *lp_sallang; /* language used for sound folding or NULL */
|
|
|
511 slang_T *lp_replang; /* language used for REP items or NULL */
|
|
236
|
512 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
513 } langp_T;
|
|
|
514
|
|
|
515 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
516
|
|
307
|
517 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
518
|
|
483
|
519 #define VIMSPELLMAGIC "VIMspell" /* string at start of Vim spell file */
|
|
|
520 #define VIMSPELLMAGICL 8
|
|
|
521 #define VIMSPELLVERSION 50
|
|
|
522
|
|
625
|
523 #define VIMSUGMAGIC "VIMsug" /* string at start of Vim .sug file */
|
|
|
524 #define VIMSUGMAGICL 6
|
|
|
525 #define VIMSUGVERSION 1
|
|
|
526
|
|
483
|
527 /* Section IDs. Only renumber them when VIMSPELLVERSION changes! */
|
|
|
528 #define SN_REGION 0 /* <regionname> section */
|
|
|
529 #define SN_CHARFLAGS 1 /* charflags section */
|
|
|
530 #define SN_MIDWORD 2 /* <midword> section */
|
|
|
531 #define SN_PREFCOND 3 /* <prefcond> section */
|
|
|
532 #define SN_REP 4 /* REP items section */
|
|
|
533 #define SN_SAL 5 /* SAL items section */
|
|
|
534 #define SN_SOFO 6 /* soundfolding section */
|
|
|
535 #define SN_MAP 7 /* MAP items section */
|
|
|
536 #define SN_COMPOUND 8 /* compound words section */
|
|
|
537 #define SN_SYLLABLE 9 /* syllable section */
|
|
492
|
538 #define SN_NOBREAK 10 /* NOBREAK section */
|
|
625
|
539 #define SN_SUGFILE 11 /* timestamp for .sug file */
|
|
|
540 #define SN_REPSAL 12 /* REPSAL items section */
|
|
|
541 #define SN_WORDS 13 /* common words */
|
|
698
|
542 #define SN_NOSPLITSUGS 14 /* don't split word for suggestions */
|
|
714
|
543 #define SN_INFO 15 /* info section */
|
|
483
|
544 #define SN_END 255 /* end of sections */
|
|
|
545
|
|
|
546 #define SNF_REQUIRED 1 /* <sectionflags>: required section */
|
|
|
547
|
|
307
|
548 /* Result values. Lower number is accepted over higher one. */
|
|
|
549 #define SP_BANNED -1
|
|
236
|
550 #define SP_OK 0
|
|
307
|
551 #define SP_RARE 1
|
|
|
552 #define SP_LOCAL 2
|
|
|
553 #define SP_BAD 3
|
|
236
|
554
|
|
381
|
555 /* file used for "zG" and "zW" */
|
|
385
|
556 static char_u *int_wordlist = NULL;
|
|
381
|
557
|
|
625
|
558 typedef struct wordcount_S
|
|
|
559 {
|
|
|
560 short_u wc_count; /* nr of times word was seen */
|
|
|
561 char_u wc_word[1]; /* word, actually longer */
|
|
|
562 } wordcount_T;
|
|
|
563
|
|
|
564 static wordcount_T dumwc;
|
|
|
565 #define WC_KEY_OFF (dumwc.wc_word - (char_u *)&dumwc)
|
|
|
566 #define HI2WC(hi) ((wordcount_T *)((hi)->hi_key - WC_KEY_OFF))
|
|
|
567 #define MAXWORDCOUNT 0xffff
|
|
|
568
|
|
236
|
569 /*
|
|
323
|
570 * Information used when looking for suggestions.
|
|
|
571 */
|
|
|
572 typedef struct suginfo_S
|
|
|
573 {
|
|
|
574 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
575 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
576 int su_maxscore; /* maximum score for adding to su_ga */
|
|
625
|
577 int su_sfmaxscore; /* idem, for when doing soundfold words */
|
|
344
|
578 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
579 char_u *su_badptr; /* start of bad word in line */
|
|
|
580 int su_badlen; /* length of detected bad word in line */
|
|
346
|
581 int su_badflags; /* caps flags for bad word */
|
|
323
|
582 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
583 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
534
|
584 char_u su_sal_badword[MAXWLEN]; /* su_badword soundfolded */
|
|
323
|
585 hashtab_T su_banned; /* table with banned words */
|
|
503
|
586 slang_T *su_sallang; /* default language for sound folding */
|
|
323
|
587 } suginfo_T;
|
|
|
588
|
|
|
589 /* One word suggestion. Used in "si_ga". */
|
|
|
590 typedef struct suggest_S
|
|
|
591 {
|
|
|
592 char_u *st_word; /* suggested word, allocated string */
|
|
625
|
593 int st_wordlen; /* STRLEN(st_word) */
|
|
323
|
594 int st_orglen; /* length of replaced text */
|
|
|
595 int st_score; /* lower is better */
|
|
344
|
596 int st_altscore; /* used when st_score compares equal */
|
|
|
597 int st_salscore; /* st_score is for soundalike */
|
|
324
|
598 int st_had_bonus; /* bonus already included in score */
|
|
503
|
599 slang_T *st_slang; /* language used for sound folding */
|
|
323
|
600 } suggest_T;
|
|
|
601
|
|
344
|
602 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
603
|
|
625
|
604 /* TRUE if a word appears in the list of banned words. */
|
|
|
605 #define WAS_BANNED(su, word) (!HASHITEM_EMPTY(hash_find(&su->su_banned, word)))
|
|
|
606
|
|
|
607 /* Number of suggestions kept when cleaning up. we need to keep more than
|
|
|
608 * what is displayed, because when rescore_suggestions() is called the score
|
|
|
609 * may change and wrong suggestions may be removed later. */
|
|
|
610 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 130 ? 150 : (su)->su_maxcount + 20)
|
|
324
|
611
|
|
|
612 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
613 * of suggestions that are not going to be displayed. */
|
|
625
|
614 #define SUG_MAX_COUNT(su) (SUG_CLEAN_COUNT(su) + 50)
|
|
323
|
615
|
|
|
616 /* score for various changes */
|
|
344
|
617 #define SCORE_SPLIT 149 /* split bad word */
|
|
698
|
618 #define SCORE_SPLIT_NO 249 /* split bad word with NOSPLITSUGS */
|
|
323
|
619 #define SCORE_ICASE 52 /* slightly different case */
|
|
483
|
620 #define SCORE_REGION 200 /* word is for different region */
|
|
323
|
621 #define SCORE_RARE 180 /* rare word */
|
|
625
|
622 #define SCORE_SWAP 75 /* swap two characters */
|
|
323
|
623 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
532
|
624 #define SCORE_REP 65 /* REP replacement */
|
|
323
|
625 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
626 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
479
|
627 #define SCORE_SUBCOMP 33 /* substitute a composing character */
|
|
324
|
628 #define SCORE_DEL 94 /* delete a character */
|
|
532
|
629 #define SCORE_DELDUP 66 /* delete a duplicated character */
|
|
479
|
630 #define SCORE_DELCOMP 28 /* delete a composing character */
|
|
324
|
631 #define SCORE_INS 96 /* insert a character */
|
|
532
|
632 #define SCORE_INSDUP 67 /* insert a duplicate character */
|
|
475
|
633 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
634 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
635
|
|
372
|
636 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
637 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
638 * 350 allows for about three changes. */
|
|
344
|
639
|
|
625
|
640 #define SCORE_COMMON1 30 /* subtracted for words seen before */
|
|
|
641 #define SCORE_COMMON2 40 /* subtracted for words often seen */
|
|
|
642 #define SCORE_COMMON3 50 /* subtracted for words very often seen */
|
|
|
643 #define SCORE_THRES2 10 /* word count threshold for COMMON2 */
|
|
|
644 #define SCORE_THRES3 100 /* word count threshold for COMMON3 */
|
|
|
645
|
|
|
646 /* When trying changed soundfold words it becomes slow when trying more than
|
|
|
647 * two changes. With less then two changes it's slightly faster but we miss a
|
|
|
648 * few good suggestions. In rare cases we need to try three of four changes.
|
|
|
649 */
|
|
|
650 #define SCORE_SFMAX1 200 /* maximum score for first try */
|
|
|
651 #define SCORE_SFMAX2 300 /* maximum score for second try */
|
|
|
652 #define SCORE_SFMAX3 400 /* maximum score for third try */
|
|
|
653
|
|
344
|
654 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
625
|
655 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
656 #define SCORE_LIMITMAX 350 /* for spell_edit_score_limit() */
|
|
|
657
|
|
|
658 /* for spell_edit_score_limit() we need to know the minimum value of
|
|
|
659 * SCORE_ICASE, SCORE_SWAP, SCORE_DEL, SCORE_SIMILAR and SCORE_INS */
|
|
|
660 #define SCORE_EDIT_MIN SCORE_SIMILAR
|
|
323
|
661
|
|
|
662 /*
|
|
236
|
663 * Structure to store info for word matching.
|
|
|
664 */
|
|
|
665 typedef struct matchinf_S
|
|
|
666 {
|
|
|
667 langp_T *mi_lp; /* info for language and region */
|
|
243
|
668
|
|
|
669 /* pointers to original text to be checked */
|
|
236
|
670 char_u *mi_word; /* start of word being checked */
|
|
339
|
671 char_u *mi_end; /* end of matching word so far */
|
|
243
|
672 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
673 char_u *mi_cend; /* char after what was used for
|
|
|
674 mi_capflags */
|
|
243
|
675
|
|
|
676 /* case-folded text */
|
|
|
677 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
678 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
679
|
|
339
|
680 /* for when checking word after a prefix */
|
|
|
681 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
481
|
682 affixID/condition */
|
|
339
|
683 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
684 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
685 #ifdef FEAT_MBYTE
|
|
|
686 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
687 case */
|
|
|
688 #else
|
|
|
689 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
690 #endif
|
|
339
|
691
|
|
481
|
692 /* for when checking a compound word */
|
|
|
693 int mi_compoff; /* start of following word offset */
|
|
483
|
694 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */
|
|
|
695 int mi_complen; /* nr of compound words used */
|
|
481
|
696
|
|
243
|
697 /* others */
|
|
236
|
698 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
699 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
700 buf_T *mi_buf; /* buffer being checked */
|
|
492
|
701
|
|
|
702 /* for NOBREAK */
|
|
|
703 int mi_result2; /* "mi_resul" without following word */
|
|
|
704 char_u *mi_end2; /* "mi_end" without following word */
|
|
236
|
705 } matchinf_T;
|
|
|
706
|
|
307
|
707 /*
|
|
|
708 * The tables used for recognizing word characters according to spelling.
|
|
|
709 * These are only used for the first 256 characters of 'encoding'.
|
|
|
710 */
|
|
|
711 typedef struct spelltab_S
|
|
|
712 {
|
|
|
713 char_u st_isw[256]; /* flags: is word char */
|
|
|
714 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
715 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
716 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
717 } spelltab_T;
|
|
|
718
|
|
|
719 static spelltab_T spelltab;
|
|
|
720 static int did_set_spelltab;
|
|
|
721
|
|
324
|
722 #define CF_WORD 0x01
|
|
|
723 #define CF_UPPER 0x02
|
|
307
|
724
|
|
|
725 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
726 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
727 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
728 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
729 #ifdef FEAT_MBYTE
|
|
|
730 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
731 #endif
|
|
483
|
732 static int write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
733
|
|
|
734 /*
|
|
344
|
735 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
736 */
|
|
|
737 typedef enum
|
|
|
738 {
|
|
344
|
739 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
740 * ends); if badword ends there is a match, otherwise
|
|
|
741 * try splitting word. */
|
|
375
|
742 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
743 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
744 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
745 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
746 STATE_DEL, /* Delete a byte from the bad word. */
|
|
625
|
747 STATE_INS_PREP, /* Prepare for inserting bytes. */
|
|
330
|
748 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
749 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
750 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
751 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
752 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
753 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
754 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
755 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
756 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
757 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
758 STATE_FINAL /* End of this node. */
|
|
|
759 } state_T;
|
|
|
760
|
|
|
761 /*
|
|
346
|
762 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
763 */
|
|
|
764 typedef struct trystate_S
|
|
|
765 {
|
|
330
|
766 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
767 int ts_score; /* score */
|
|
344
|
768 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
769 short ts_curi; /* index in list of child nodes */
|
|
|
770 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
771 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
772 char_u ts_twordlen; /* valid length of tword[] */
|
|
482
|
773 char_u ts_prefixdepth; /* stack depth for end of prefix or
|
|
485
|
774 * PFD_PREFIXTREE or PFD_NOPREFIX */
|
|
|
775 char_u ts_flags; /* TSF_ flags */
|
|
330
|
776 #ifdef FEAT_MBYTE
|
|
|
777 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
778 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
779 char_u ts_isdiff; /* DIFF_ values */
|
|
|
780 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
781 #endif
|
|
483
|
782 char_u ts_prewordlen; /* length of word in "preword[]" */
|
|
|
783 char_u ts_splitoff; /* index in "tword" after last split */
|
|
492
|
784 char_u ts_splitfidx; /* "ts_fidx" at word split */
|
|
483
|
785 char_u ts_complen; /* nr of compound words used */
|
|
485
|
786 char_u ts_compsplit; /* index for "compflags" where word was spit */
|
|
346
|
787 char_u ts_save_badflags; /* su_badflags saved here */
|
|
625
|
788 char_u ts_delidx; /* index in fword for char that was deleted,
|
|
|
789 valid when "ts_flags" has TSF_DIDDEL */
|
|
323
|
790 } trystate_T;
|
|
|
791
|
|
330
|
792 /* values for ts_isdiff */
|
|
|
793 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
794 #define DIFF_YES 1 /* different byte found */
|
|
|
795 #define DIFF_INSERT 2 /* inserting character */
|
|
|
796
|
|
485
|
797 /* values for ts_flags */
|
|
|
798 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */
|
|
|
799 #define TSF_DIDSPLIT 2 /* tried split at this point */
|
|
625
|
800 #define TSF_DIDDEL 4 /* did a delete, "ts_delidx" has index */
|
|
485
|
801
|
|
375
|
802 /* special values ts_prefixdepth */
|
|
485
|
803 #define PFD_NOPREFIX 0xff /* not using prefixes */
|
|
482
|
804 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
625
|
805 #define PFD_NOTSPECIAL 0xfd /* highest value that's not special */
|
|
375
|
806
|
|
339
|
807 /* mode values for find_word */
|
|
481
|
808 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
809 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
810 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
811 #define FIND_COMPOUND 3 /* find case-folded compound word */
|
|
|
812 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */
|
|
339
|
813
|
|
236
|
814 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
815 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
816 static void slang_clear __ARGS((slang_T *lp));
|
|
625
|
817 static void slang_clear_sug __ARGS((slang_T *lp));
|
|
339
|
818 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
483
|
819 static int can_compound __ARGS((slang_T *slang, char_u *word, char_u *flags));
|
|
455
|
820 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
485
|
821 static void find_prefix __ARGS((matchinf_T *mip, int mode));
|
|
339
|
822 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
823 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
498
|
824 static int no_spell_checking __ARGS((win_T *wp));
|
|
307
|
825 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
826 static char_u *spell_enc __ARGS((void));
|
|
385
|
827 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
828 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
829 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
695
|
830 static int get2c __ARGS((FILE *fd));
|
|
|
831 static int get3c __ARGS((FILE *fd));
|
|
|
832 static int get4c __ARGS((FILE *fd));
|
|
|
833 static time_t get8c __ARGS((FILE *fd));
|
|
389
|
834 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
483
|
835 static char_u *read_string __ARGS((FILE *fd, int cnt));
|
|
|
836 static int read_region_section __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
837 static int read_charflags_section __ARGS((FILE *fd));
|
|
|
838 static int read_prefcond_section __ARGS((FILE *fd, slang_T *lp));
|
|
625
|
839 static int read_rep_section __ARGS((FILE *fd, garray_T *gap, short *first));
|
|
483
|
840 static int read_sal_section __ARGS((FILE *fd, slang_T *slang));
|
|
625
|
841 static int read_words_section __ARGS((FILE *fd, slang_T *lp, int len));
|
|
|
842 static void count_common_word __ARGS((slang_T *lp, char_u *word, int len, int count));
|
|
|
843 static int score_wordcount_adj __ARGS((slang_T *slang, int score, char_u *word, int split));
|
|
483
|
844 static int read_sofo_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
845 static int read_compound __ARGS((FILE *fd, slang_T *slang, int len));
|
|
495
|
846 static int byte_in_str __ARGS((char_u *str, int byte));
|
|
483
|
847 static int init_syl_tab __ARGS((slang_T *slang));
|
|
|
848 static int count_syllables __ARGS((slang_T *slang, char_u *word));
|
|
381
|
849 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
850 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
851 #ifdef FEAT_MBYTE
|
|
|
852 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
853 #endif
|
|
625
|
854 static int spell_read_tree __ARGS((FILE *fd, char_u **bytsp, idx_T **idxsp, int prefixtree, int prefixcnt));
|
|
|
855 static idx_T read_tree_node __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
856 static void clear_midword __ARGS((buf_T *buf));
|
|
|
857 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
858 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
859 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
860 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
861 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
483
|
862 static void set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
863 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
864 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
865 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
638
|
866 static void spell_find_suggest __ARGS((char_u *badptr, int badlen, suginfo_T *su, int maxcount, int banbadword, int need_cap, int interactive));
|
|
372
|
867 #ifdef FEAT_EVAL
|
|
|
868 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
869 #endif
|
|
|
870 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
625
|
871 static void spell_suggest_intern __ARGS((suginfo_T *su, int interactive));
|
|
|
872 static void suggest_load_files __ARGS((void));
|
|
|
873 static void tree_count_words __ARGS((char_u *byts, idx_T *idxs));
|
|
344
|
874 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
875 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
876 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
877 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
878 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
625
|
879 static void suggest_trie_walk __ARGS((suginfo_T *su, langp_T *lp, char_u *fword, int soundfold));
|
|
|
880 static void go_deeper __ARGS((trystate_T *stack, int depth, int score_add));
|
|
455
|
881 #ifdef FEAT_MBYTE
|
|
|
882 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
883 #endif
|
|
323
|
884 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
885 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
886 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
887 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
625
|
888 static void suggest_try_soundalike_prep __ARGS((void));
|
|
346
|
889 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
625
|
890 static void suggest_try_soundalike_finish __ARGS((void));
|
|
|
891 static void add_sound_suggest __ARGS((suginfo_T *su, char_u *goodword, int score, langp_T *lp));
|
|
|
892 static int soundfold_find __ARGS((slang_T *slang, char_u *word));
|
|
323
|
893 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
894 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
895 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
625
|
896 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));
|
|
|
897 static void check_suggestions __ARGS((suginfo_T *su, garray_T *gap));
|
|
323
|
898 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
324
|
899 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
534
|
900 static void rescore_one __ARGS((suginfo_T *su, suggest_T *stp));
|
|
344
|
901 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
902 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
903 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
904 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
905 #ifdef FEAT_MBYTE
|
|
375
|
906 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
907 #endif
|
|
344
|
908 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
625
|
909 static int spell_edit_score __ARGS((slang_T *slang, char_u *badword, char_u *goodword));
|
|
|
910 static int spell_edit_score_limit __ARGS((slang_T *slang, char_u *badword, char_u *goodword, int limit));
|
|
|
911 #ifdef FEAT_MBYTE
|
|
|
912 static int spell_edit_score_limit_w __ARGS((slang_T *slang, char_u *badword, char_u *goodword, int limit));
|
|
|
913 #endif
|
|
701
|
914 static void dump_word __ARGS((slang_T *slang, char_u *word, char_u *pat, int *dir, int round, int flags, linenr_T lnum));
|
|
|
915 static linenr_T dump_prefixes __ARGS((slang_T *slang, char_u *word, char_u *pat, int *dir, int round, int flags, linenr_T startlnum));
|
|
625
|
916 static buf_T *open_spellbuf __ARGS((void));
|
|
|
917 static void close_spellbuf __ARGS((buf_T *buf));
|
|
323
|
918
|
|
324
|
919 /*
|
|
|
920 * Use our own character-case definitions, because the current locale may
|
|
|
921 * differ from what the .spl file uses.
|
|
|
922 * These must not be called with negative number!
|
|
|
923 */
|
|
|
924 #ifndef FEAT_MBYTE
|
|
|
925 /* Non-multi-byte implementation. */
|
|
|
926 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
927 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
928 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
929 #else
|
|
447
|
930 # if defined(HAVE_WCHAR_H)
|
|
|
931 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
932 # endif
|
|
324
|
933 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
934 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
935 * the "w" library function for characters above 255 if available. */
|
|
|
936 # ifdef HAVE_TOWLOWER
|
|
|
937 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
938 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
939 # else
|
|
|
940 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
941 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
942 # endif
|
|
|
943
|
|
|
944 # ifdef HAVE_TOWUPPER
|
|
|
945 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
946 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
947 # else
|
|
|
948 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
949 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
950 # endif
|
|
|
951
|
|
|
952 # ifdef HAVE_ISWUPPER
|
|
|
953 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
954 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
955 # else
|
|
|
956 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
957 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
958 # endif
|
|
|
959 #endif
|
|
|
960
|
|
307
|
961
|
|
|
962 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
963 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
482
|
964 static char *e_afftrailing = N_("Trailing text in %s line %d: %s");
|
|
495
|
965 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
|
966 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
967 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
469
|
968 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
969
|
|
626
|
970 /* Remember what "z?" replaced. */
|
|
|
971 static char_u *repl_from = NULL;
|
|
|
972 static char_u *repl_to = NULL;
|
|
|
973
|
|
236
|
974 /*
|
|
|
975 * Main spell-checking function.
|
|
300
|
976 * "ptr" points to a character that could be the start of a word.
|
|
534
|
977 * "*attrp" is set to the highlight index for a badly spelled word. For a
|
|
|
978 * non-word or when it's OK it remains unchanged.
|
|
236
|
979 * This must only be called when 'spelllang' is not empty.
|
|
323
|
980 *
|
|
385
|
981 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
982 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
983 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
984 * worry.
|
|
323
|
985 *
|
|
236
|
986 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
987 * caller can skip over the word.
|
|
|
988 */
|
|
|
989 int
|
|
625
|
990 spell_check(wp, ptr, attrp, capcol, docount)
|
|
236
|
991 win_T *wp; /* current window */
|
|
|
992 char_u *ptr;
|
|
534
|
993 hlf_T *attrp;
|
|
385
|
994 int *capcol; /* column to check for Capital */
|
|
625
|
995 int docount; /* count good words */
|
|
236
|
996 {
|
|
|
997 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
998 be passed to functions quickly. */
|
|
344
|
999 int nrlen = 0; /* found a number first */
|
|
385
|
1000 int c;
|
|
483
|
1001 int wrongcaplen = 0;
|
|
500
|
1002 int lpi;
|
|
625
|
1003 int count_word = docount;
|
|
236
|
1004
|
|
307
|
1005 /* A word never starts at a space or a control character. Return quickly
|
|
|
1006 * then, skipping over the character. */
|
|
|
1007 if (*ptr <= ' ')
|
|
|
1008 return 1;
|
|
690
|
1009
|
|
|
1010 /* Return here when loading language files failed. */
|
|
|
1011 if (wp->w_buffer->b_langp.ga_len == 0)
|
|
|
1012 return 1;
|
|
|
1013
|
|
483
|
1014 vim_memset(&mi, 0, sizeof(matchinf_T));
|
|
236
|
1015
|
|
344
|
1016 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
584
|
1017 * 0X99FF. But always do check spelling to find "3GPP" and "11
|
|
|
1018 * julifeest". */
|
|
300
|
1019 if (*ptr >= '0' && *ptr <= '9')
|
|
|
1020 {
|
|
316
|
1021 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
1022 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
1023 else
|
|
344
|
1024 mi.mi_end = skipdigits(ptr);
|
|
584
|
1025 nrlen = mi.mi_end - ptr;
|
|
|
1026 }
|
|
346
|
1027
|
|
|
1028 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
1029 mi.mi_word = ptr;
|
|
584
|
1030 mi.mi_fend = ptr;
|
|
376
|
1031 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
1032 {
|
|
|
1033 do
|
|
|
1034 {
|
|
307
|
1035 mb_ptr_adv(mi.mi_fend);
|
|
376
|
1036 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
1037
|
|
|
1038 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
1039 {
|
|
|
1040 /* Check word starting with capital letter. */
|
|
455
|
1041 c = PTR2CHAR(ptr);
|
|
385
|
1042 if (!SPELL_ISUPPER(c))
|
|
483
|
1043 wrongcaplen = (int)(mi.mi_fend - ptr);
|
|
385
|
1044 }
|
|
|
1045 }
|
|
|
1046 if (capcol != NULL)
|
|
|
1047 *capcol = -1;
|
|
344
|
1048
|
|
|
1049 /* We always use the characters up to the next non-word character,
|
|
|
1050 * also for bad words. */
|
|
|
1051 mi.mi_end = mi.mi_fend;
|
|
|
1052
|
|
|
1053 /* Check caps type later. */
|
|
376
|
1054 mi.mi_buf = wp->w_buffer;
|
|
344
|
1055
|
|
483
|
1056 /* case-fold the word with one non-word character, so that we can check
|
|
|
1057 * for the word end. */
|
|
344
|
1058 if (*mi.mi_fend != NUL)
|
|
|
1059 mb_ptr_adv(mi.mi_fend);
|
|
|
1060
|
|
|
1061 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
1062 MAXWLEN + 1);
|
|
|
1063 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
1064
|
|
|
1065 /* The word is bad unless we recognize it. */
|
|
|
1066 mi.mi_result = SP_BAD;
|
|
492
|
1067 mi.mi_result2 = SP_BAD;
|
|
344
|
1068
|
|
|
1069 /*
|
|
|
1070 * Loop over the languages specified in 'spelllang'.
|
|
625
|
1071 * We check them all, because a word may be matched longer in another
|
|
|
1072 * language.
|
|
344
|
1073 */
|
|
500
|
1074 for (lpi = 0; lpi < wp->w_buffer->b_langp.ga_len; ++lpi)
|
|
|
1075 {
|
|
|
1076 mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, lpi);
|
|
|
1077
|
|
|
1078 /* If reloading fails the language is still in the list but everything
|
|
|
1079 * has been cleared. */
|
|
|
1080 if (mi.mi_lp->lp_slang->sl_fidxs == NULL)
|
|
|
1081 continue;
|
|
|
1082
|
|
344
|
1083 /* Check for a matching word in case-folded words. */
|
|
|
1084 find_word(&mi, FIND_FOLDWORD);
|
|
|
1085
|
|
|
1086 /* Check for a matching word in keep-case words. */
|
|
|
1087 find_word(&mi, FIND_KEEPWORD);
|
|
|
1088
|
|
|
1089 /* Check for matching prefixes. */
|
|
485
|
1090 find_prefix(&mi, FIND_FOLDWORD);
|
|
492
|
1091
|
|
|
1092 /* For a NOBREAK language, may want to use a word without a following
|
|
|
1093 * word as a backup. */
|
|
|
1094 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD
|
|
|
1095 && mi.mi_result2 != SP_BAD)
|
|
|
1096 {
|
|
|
1097 mi.mi_result = mi.mi_result2;
|
|
|
1098 mi.mi_end = mi.mi_end2;
|
|
|
1099 }
|
|
625
|
1100
|
|
|
1101 /* Count the word in the first language where it's found to be OK. */
|
|
|
1102 if (count_word && mi.mi_result == SP_OK)
|
|
|
1103 {
|
|
|
1104 count_common_word(mi.mi_lp->lp_slang, ptr,
|
|
|
1105 (int)(mi.mi_end - ptr), 1);
|
|
|
1106 count_word = FALSE;
|
|
|
1107 }
|
|
344
|
1108 }
|
|
|
1109
|
|
|
1110 if (mi.mi_result != SP_OK)
|
|
|
1111 {
|
|
346
|
1112 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
1113 * rare and local words, e.g., "3GPP". */
|
|
344
|
1114 if (nrlen > 0)
|
|
346
|
1115 {
|
|
|
1116 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
1117 return nrlen;
|
|
|
1118 }
|
|
344
|
1119
|
|
|
1120 /* When we are at a non-word character there is no error, just
|
|
|
1121 * skip over the character (try looking for a word after it). */
|
|
390
|
1122 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
1123 {
|
|
385
|
1124 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
1125 {
|
|
|
1126 regmatch_T regmatch;
|
|
|
1127
|
|
|
1128 /* Check for end of sentence. */
|
|
|
1129 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
1130 regmatch.rm_ic = FALSE;
|
|
|
1131 if (vim_regexec(®match, ptr, 0))
|
|
|
1132 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
1133 }
|
|
|
1134
|
|
344
|
1135 #ifdef FEAT_MBYTE
|
|
|
1136 if (has_mbyte)
|
|
474
|
1137 return (*mb_ptr2len)(ptr);
|
|
344
|
1138 #endif
|
|
|
1139 return 1;
|
|
300
|
1140 }
|
|
483
|
1141 else if (mi.mi_end == ptr)
|
|
|
1142 /* Always include at least one character. Required for when there
|
|
|
1143 * is a mixup in "midword". */
|
|
|
1144 mb_ptr_adv(mi.mi_end);
|
|
492
|
1145 else if (mi.mi_result == SP_BAD
|
|
|
1146 && LANGP_ENTRY(wp->w_buffer->b_langp, 0)->lp_slang->sl_nobreak)
|
|
|
1147 {
|
|
|
1148 char_u *p, *fp;
|
|
|
1149 int save_result = mi.mi_result;
|
|
|
1150
|
|
|
1151 /* First language in 'spelllang' is NOBREAK. Find first position
|
|
|
1152 * at which any word would be valid. */
|
|
|
1153 mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
500
|
1154 if (mi.mi_lp->lp_slang->sl_fidxs != NULL)
|
|
|
1155 {
|
|
|
1156 p = mi.mi_word;
|
|
|
1157 fp = mi.mi_fword;
|
|
|
1158 for (;;)
|
|
|
1159 {
|
|
|
1160 mb_ptr_adv(p);
|
|
|
1161 mb_ptr_adv(fp);
|
|
|
1162 if (p >= mi.mi_end)
|
|
|
1163 break;
|
|
|
1164 mi.mi_compoff = fp - mi.mi_fword;
|
|
|
1165 find_word(&mi, FIND_COMPOUND);
|
|
|
1166 if (mi.mi_result != SP_BAD)
|
|
|
1167 {
|
|
|
1168 mi.mi_end = p;
|
|
|
1169 break;
|
|
|
1170 }
|
|
|
1171 }
|
|
|
1172 mi.mi_result = save_result;
|
|
|
1173 }
|
|
492
|
1174 }
|
|
243
|
1175
|
|
344
|
1176 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
534
|
1177 *attrp = HLF_SPB;
|
|
344
|
1178 else if (mi.mi_result == SP_RARE)
|
|
534
|
1179 *attrp = HLF_SPR;
|
|
344
|
1180 else
|
|
534
|
1181 *attrp = HLF_SPL;
|
|
243
|
1182 }
|
|
|
1183
|
|
483
|
1184 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE))
|
|
|
1185 {
|
|
|
1186 /* Report SpellCap only when the word isn't badly spelled. */
|
|
534
|
1187 *attrp = HLF_SPC;
|
|
483
|
1188 return wrongcaplen;
|
|
|
1189 }
|
|
|
1190
|
|
300
|
1191 return (int)(mi.mi_end - ptr);
|
|
236
|
1192 }
|
|
|
1193
|
|
|
1194 /*
|
|
300
|
1195 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
1196 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
1197 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
1198 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
1199 * tree.
|
|
300
|
1200 *
|
|
|
1201 * For a match mip->mi_result is updated.
|
|
243
|
1202 */
|
|
|
1203 static void
|
|
339
|
1204 find_word(mip, mode)
|
|
243
|
1205 matchinf_T *mip;
|
|
339
|
1206 int mode;
|
|
243
|
1207 {
|
|
324
|
1208 idx_T arridx = 0;
|
|
300
|
1209 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
1210 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
1211 int endidxcnt = 0;
|
|
|
1212 int len;
|
|
|
1213 int wlen = 0;
|
|
|
1214 int flen;
|
|
|
1215 int c;
|
|
|
1216 char_u *ptr;
|
|
324
|
1217 idx_T lo, hi, m;
|
|
243
|
1218 #ifdef FEAT_MBYTE
|
|
300
|
1219 char_u *s;
|
|
490
|
1220 #endif
|
|
339
|
1221 char_u *p;
|
|
307
|
1222 int res = SP_BAD;
|
|
300
|
1223 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1224 unsigned flags;
|
|
|
1225 char_u *byts;
|
|
324
|
1226 idx_T *idxs;
|
|
481
|
1227 int word_ends;
|
|
485
|
1228 int prefix_found;
|
|
492
|
1229 int nobreak_result;
|
|
481
|
1230
|
|
|
1231 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND)
|
|
236
|
1232 {
|
|
300
|
1233 /* Check for word with matching case in keep-case tree. */
|
|
|
1234 ptr = mip->mi_word;
|
|
|
1235 flen = 9999; /* no case folding, always enough bytes */
|
|
|
1236 byts = slang->sl_kbyts;
|
|
|
1237 idxs = slang->sl_kidxs;
|
|
481
|
1238
|
|
|
1239 if (mode == FIND_KEEPCOMPOUND)
|
|
|
1240 /* Skip over the previously found word(s). */
|
|
|
1241 wlen += mip->mi_compoff;
|
|
236
|
1242 }
|
|
|
1243 else
|
|
|
1244 {
|
|
300
|
1245 /* Check for case-folded in case-folded tree. */
|
|
|
1246 ptr = mip->mi_fword;
|
|
|
1247 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1248 byts = slang->sl_fbyts;
|
|
|
1249 idxs = slang->sl_fidxs;
|
|
339
|
1250
|
|
|
1251 if (mode == FIND_PREFIX)
|
|
|
1252 {
|
|
|
1253 /* Skip over the prefix. */
|
|
|
1254 wlen = mip->mi_prefixlen;
|
|
|
1255 flen -= mip->mi_prefixlen;
|
|
|
1256 }
|
|
481
|
1257 else if (mode == FIND_COMPOUND)
|
|
|
1258 {
|
|
|
1259 /* Skip over the previously found word(s). */
|
|
|
1260 wlen = mip->mi_compoff;
|
|
|
1261 flen -= mip->mi_compoff;
|
|
|
1262 }
|
|
|
1263
|
|
243
|
1264 }
|
|
|
1265
|
|
300
|
1266 if (byts == NULL)
|
|
|
1267 return; /* array is empty */
|
|
236
|
1268
|
|
|
1269 /*
|
|
307
|
1270 * Repeat advancing in the tree until:
|
|
|
1271 * - there is a byte that doesn't match,
|
|
|
1272 * - we reach the end of the tree,
|
|
|
1273 * - or we reach the end of the line.
|
|
236
|
1274 */
|
|
300
|
1275 for (;;)
|
|
236
|
1276 {
|
|
346
|
1277 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
1278 flen = fold_more(mip);
|
|
300
|
1279
|
|
|
1280 len = byts[arridx++];
|
|
|
1281
|
|
|
1282 /* If the first possible byte is a zero the word could end here.
|
|
|
1283 * Remember this index, we first check for the longest word. */
|
|
|
1284 if (byts[arridx] == 0)
|
|
|
1285 {
|
|
307
|
1286 if (endidxcnt == MAXWLEN)
|
|
|
1287 {
|
|
|
1288 /* Must be a corrupted spell file. */
|
|
|
1289 EMSG(_(e_format));
|
|
|
1290 return;
|
|
|
1291 }
|
|
300
|
1292 endlen[endidxcnt] = wlen;
|
|
|
1293 endidx[endidxcnt++] = arridx++;
|
|
|
1294 --len;
|
|
|
1295
|
|
|
1296 /* Skip over the zeros, there can be several flag/region
|
|
|
1297 * combinations. */
|
|
|
1298 while (len > 0 && byts[arridx] == 0)
|
|
|
1299 {
|
|
|
1300 ++arridx;
|
|
|
1301 --len;
|
|
|
1302 }
|
|
|
1303 if (len == 0)
|
|
|
1304 break; /* no children, word must end here */
|
|
|
1305 }
|
|
|
1306
|
|
|
1307 /* Stop looking at end of the line. */
|
|
|
1308 if (ptr[wlen] == NUL)
|
|
|
1309 break;
|
|
|
1310
|
|
|
1311 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1312 c = ptr[wlen];
|
|
346
|
1313 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1314 c = ' ';
|
|
300
|
1315 lo = arridx;
|
|
|
1316 hi = arridx + len - 1;
|
|
|
1317 while (lo < hi)
|
|
|
1318 {
|
|
|
1319 m = (lo + hi) / 2;
|
|
|
1320 if (byts[m] > c)
|
|
|
1321 hi = m - 1;
|
|
|
1322 else if (byts[m] < c)
|
|
|
1323 lo = m + 1;
|
|
|
1324 else
|
|
|
1325 {
|
|
|
1326 lo = hi = m;
|
|
|
1327 break;
|
|
236
|
1328 }
|
|
|
1329 }
|
|
300
|
1330
|
|
|
1331 /* Stop if there is no matching byte. */
|
|
|
1332 if (hi < lo || byts[lo] != c)
|
|
|
1333 break;
|
|
|
1334
|
|
|
1335 /* Continue at the child (if there is one). */
|
|
|
1336 arridx = idxs[lo];
|
|
|
1337 ++wlen;
|
|
|
1338 --flen;
|
|
346
|
1339
|
|
|
1340 /* One space in the good word may stand for several spaces in the
|
|
|
1341 * checked word. */
|
|
|
1342 if (c == ' ')
|
|
|
1343 {
|
|
|
1344 for (;;)
|
|
|
1345 {
|
|
|
1346 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1347 flen = fold_more(mip);
|
|
|
1348 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1349 break;
|
|
|
1350 ++wlen;
|
|
|
1351 --flen;
|
|
|
1352 }
|
|
|
1353 }
|
|
236
|
1354 }
|
|
|
1355
|
|
300
|
1356 /*
|
|
|
1357 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1358 * first.
|
|
|
1359 */
|
|
|
1360 while (endidxcnt > 0)
|
|
|
1361 {
|
|
|
1362 --endidxcnt;
|
|
|
1363 arridx = endidx[endidxcnt];
|
|
|
1364 wlen = endlen[endidxcnt];
|
|
236
|
1365
|
|
300
|
1366 #ifdef FEAT_MBYTE
|
|
|
1367 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1368 continue; /* not at first byte of character */
|
|
|
1369 #endif
|
|
376
|
1370 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
481
|
1371 {
|
|
492
|
1372 if (slang->sl_compprog == NULL && !slang->sl_nobreak)
|
|
481
|
1373 continue; /* next char is a word character */
|
|
|
1374 word_ends = FALSE;
|
|
|
1375 }
|
|
|
1376 else
|
|
|
1377 word_ends = TRUE;
|
|
485
|
1378 /* The prefix flag is before compound flags. Once a valid prefix flag
|
|
|
1379 * has been found we try compound flags. */
|
|
|
1380 prefix_found = FALSE;
|
|
300
|
1381
|
|
|
1382 #ifdef FEAT_MBYTE
|
|
339
|
1383 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1384 {
|
|
|
1385 /* Compute byte length in original word, length may change
|
|
339
|
1386 * when folding case. This can be slow, take a shortcut when the
|
|
|
1387 * case-folded word is equal to the keep-case word. */
|
|
300
|
1388 p = mip->mi_word;
|
|
339
|
1389 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1390 {
|
|
|
1391 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1392 mb_ptr_adv(p);
|
|
|
1393 wlen = p - mip->mi_word;
|
|
|
1394 }
|
|
300
|
1395 }
|
|
|
1396 #endif
|
|
236
|
1397
|
|
339
|
1398 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1399 * prefix ID.
|
|
|
1400 * Repeat this if there are more flags/region alternatives until there
|
|
|
1401 * is a match. */
|
|
|
1402 res = SP_BAD;
|
|
|
1403 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1404 --len, ++arridx)
|
|
300
|
1405 {
|
|
|
1406 flags = idxs[arridx];
|
|
324
|
1407
|
|
339
|
1408 /* For the fold-case tree check that the case of the checked word
|
|
|
1409 * matches with what the word in the tree requires.
|
|
|
1410 * For keep-case tree the case is always right. For prefixes we
|
|
|
1411 * don't bother to check. */
|
|
|
1412 if (mode == FIND_FOLDWORD)
|
|
300
|
1413 {
|
|
|
1414 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1415 {
|
|
323
|
1416 /* mi_capflags was set for a different word length, need
|
|
|
1417 * to do it again. */
|
|
300
|
1418 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1419 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1420 }
|
|
|
1421
|
|
346
|
1422 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1423 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1424 continue;
|
|
300
|
1425 }
|
|
236
|
1426
|
|
339
|
1427 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1428 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1429 * mip->mi_prefarridx that find_prefix() filled. */
|
|
485
|
1430 else if (mode == FIND_PREFIX && !prefix_found)
|
|
481
|
1431 {
|
|
366
|
1432 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1433 flags,
|
|
455
|
1434 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1435 FALSE);
|
|
366
|
1436 if (c == 0)
|
|
339
|
1437 continue;
|
|
366
|
1438
|
|
|
1439 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1440 if (c & WF_RAREPFX)
|
|
|
1441 flags |= WF_RARE;
|
|
485
|
1442 prefix_found = TRUE;
|
|
339
|
1443 }
|
|
|
1444
|
|
492
|
1445 if (slang->sl_nobreak)
|
|
|
1446 {
|
|
|
1447 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND)
|
|
|
1448 && (flags & WF_BANNED) == 0)
|
|
|
1449 {
|
|
|
1450 /* NOBREAK: found a valid following word. That's all we
|
|
|
1451 * need to know, so return. */
|
|
|
1452 mip->mi_result = SP_OK;
|
|
|
1453 break;
|
|
|
1454 }
|
|
|
1455 }
|
|
|
1456
|
|
|
1457 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND
|
|
|
1458 || !word_ends))
|
|
481
|
1459 {
|
|
483
|
1460 /* If there is no flag or the word is shorter than
|
|
|
1461 * COMPOUNDMIN reject it quickly.
|
|
|
1462 * Makes you wonder why someone puts a compound flag on a word
|
|
481
|
1463 * that's too short... Myspell compatibility requires this
|
|
|
1464 * anyway. */
|
|
490
|
1465 if (((unsigned)flags >> 24) == 0
|
|
|
1466 || wlen - mip->mi_compoff < slang->sl_compminlen)
|
|
483
|
1467 continue;
|
|
500
|
1468 #ifdef FEAT_MBYTE
|
|
|
1469 /* For multi-byte chars check character length against
|
|
|
1470 * COMPOUNDMIN. */
|
|
|
1471 if (has_mbyte
|
|
501
|
1472 && slang->sl_compminlen > 0
|
|
500
|
1473 && mb_charlen_len(mip->mi_word + mip->mi_compoff,
|
|
|
1474 wlen - mip->mi_compoff) < slang->sl_compminlen)
|
|
|
1475 continue;
|
|
|
1476 #endif
|
|
483
|
1477
|
|
490
|
1478 /* Limit the number of compound words to COMPOUNDMAX if no
|
|
|
1479 * maximum for syllables is specified. */
|
|
|
1480 if (!word_ends && mip->mi_complen + 2 > slang->sl_compmax
|
|
|
1481 && slang->sl_compsylmax == MAXWLEN)
|
|
483
|
1482 continue;
|
|
|
1483
|
|
485
|
1484 /* Quickly check if compounding is possible with this flag. */
|
|
495
|
1485 if (!byte_in_str(mip->mi_complen == 0
|
|
485
|
1486 ? slang->sl_compstartflags
|
|
|
1487 : slang->sl_compallflags,
|
|
495
|
1488 ((unsigned)flags >> 24)))
|
|
481
|
1489 continue;
|
|
|
1490
|
|
490
|
1491 if (mode == FIND_COMPOUND)
|
|
|
1492 {
|
|
|
1493 int capflags;
|
|
|
1494
|
|
|
1495 /* Need to check the caps type of the appended compound
|
|
|
1496 * word. */
|
|
|
1497 #ifdef FEAT_MBYTE
|
|
|
1498 if (has_mbyte && STRNCMP(ptr, mip->mi_word,
|
|
|
1499 mip->mi_compoff) != 0)
|
|
|
1500 {
|
|
|
1501 /* case folding may have changed the length */
|
|
|
1502 p = mip->mi_word;
|
|
|
1503 for (s = ptr; s < ptr + mip->mi_compoff; mb_ptr_adv(s))
|
|
|
1504 mb_ptr_adv(p);
|
|
|
1505 }
|
|
|
1506 else
|
|
|
1507 #endif
|
|
|
1508 p = mip->mi_word + mip->mi_compoff;
|
|
|
1509 capflags = captype(p, mip->mi_word + wlen);
|
|
|
1510 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP
|
|
|
1511 && (flags & WF_FIXCAP) != 0))
|
|
|
1512 continue;
|
|
|
1513
|
|
|
1514 if (capflags != WF_ALLCAP)
|
|
|
1515 {
|
|
|
1516 /* When the character before the word is a word
|
|
|
1517 * character we do not accept a Onecap word. We do
|
|
|
1518 * accept a no-caps word, even when the dictionary
|
|
|
1519 * word specifies ONECAP. */
|
|
|
1520 mb_ptr_back(mip->mi_word, p);
|
|
|
1521 if (spell_iswordp_nmw(p)
|
|
|
1522 ? capflags == WF_ONECAP
|
|
|
1523 : (flags & WF_ONECAP) != 0
|
|
|
1524 && capflags != WF_ONECAP)
|
|
|
1525 continue;
|
|
|
1526 }
|
|
|
1527 }
|
|
|
1528
|
|
483
|
1529 /* If the word ends the sequence of compound flags of the
|
|
714
|
1530 * words must match with one of the COMPOUNDRULE items and
|
|
483
|
1531 * the number of syllables must not be too large. */
|
|
|
1532 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24);
|
|
|
1533 mip->mi_compflags[mip->mi_complen + 1] = NUL;
|
|
|
1534 if (word_ends)
|
|
|
1535 {
|
|
|
1536 char_u fword[MAXWLEN];
|
|
|
1537
|
|
|
1538 if (slang->sl_compsylmax < MAXWLEN)
|
|
|
1539 {
|
|
|
1540 /* "fword" is only needed for checking syllables. */
|
|
|
1541 if (ptr == mip->mi_word)
|
|
|
1542 (void)spell_casefold(ptr, wlen, fword, MAXWLEN);
|
|
|
1543 else
|
|
|
1544 vim_strncpy(fword, ptr, endlen[endidxcnt]);
|
|
|
1545 }
|
|
|
1546 if (!can_compound(slang, fword, mip->mi_compflags))
|
|
|
1547 continue;
|
|
|
1548 }
|
|
481
|
1549 }
|
|
|
1550
|
|
500
|
1551 /* Check NEEDCOMPOUND: can't use word without compounding. */
|
|
|
1552 else if (flags & WF_NEEDCOMP)
|
|
|
1553 continue;
|
|
|
1554
|
|
492
|
1555 nobreak_result = SP_OK;
|
|
|
1556
|
|
481
|
1557 if (!word_ends)
|
|
|
1558 {
|
|
492
|
1559 int save_result = mip->mi_result;
|
|
|
1560 char_u *save_end = mip->mi_end;
|
|
501
|
1561 langp_T *save_lp = mip->mi_lp;
|
|
|
1562 int lpi;
|
|
492
|
1563
|
|
|
1564 /* Check that a valid word follows. If there is one and we
|
|
|
1565 * are compounding, it will set "mi_result", thus we are
|
|
|
1566 * always finished here. For NOBREAK we only check that a
|
|
|
1567 * valid word follows.
|
|
481
|
1568 * Recursive! */
|
|
492
|
1569 if (slang->sl_nobreak)
|
|
|
1570 mip->mi_result = SP_BAD;
|
|
481
|
1571
|
|
|
1572 /* Find following word in case-folded tree. */
|
|
|
1573 mip->mi_compoff = endlen[endidxcnt];
|
|
|
1574 #ifdef FEAT_MBYTE
|
|
|
1575 if (has_mbyte && mode == FIND_KEEPWORD)
|
|
|
1576 {
|
|
|
1577 /* Compute byte length in case-folded word from "wlen":
|
|
|
1578 * byte length in keep-case word. Length may change when
|
|
|
1579 * folding case. This can be slow, take a shortcut when
|
|
|
1580 * the case-folded word is equal to the keep-case word. */
|
|
|
1581 p = mip->mi_fword;
|
|
|
1582 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1583 {
|
|
|
1584 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1585 mb_ptr_adv(p);
|
|
|
1586 mip->mi_compoff = p - mip->mi_fword;
|
|
|
1587 }
|
|
|
1588 }
|
|
|
1589 #endif
|
|
485
|
1590 c = mip->mi_compoff;
|
|
483
|
1591 ++mip->mi_complen;
|
|
501
|
1592
|
|
|
1593 /* For NOBREAK we need to try all NOBREAK languages, at least
|
|
|
1594 * to find the ".add" file(s). */
|
|
|
1595 for (lpi = 0; lpi < mip->mi_buf->b_langp.ga_len; ++lpi)
|
|
|
1596 {
|
|
|
1597 if (slang->sl_nobreak)
|
|
|
1598 {
|
|
|
1599 mip->mi_lp = LANGP_ENTRY(mip->mi_buf->b_langp, lpi);
|
|
|
1600 if (mip->mi_lp->lp_slang->sl_fidxs == NULL
|
|
|
1601 || !mip->mi_lp->lp_slang->sl_nobreak)
|
|
|
1602 continue;
|
|
|
1603 }
|
|
|
1604
|
|
|
1605 find_word(mip, FIND_COMPOUND);
|
|
|
1606
|
|
|
1607 /* When NOBREAK any word that matches is OK. Otherwise we
|
|
|
1608 * need to find the longest match, thus try with keep-case
|
|
|
1609 * and prefix too. */
|
|
492
|
1610 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1611 {
|
|
501
|
1612 /* Find following word in keep-case tree. */
|
|
|
1613 mip->mi_compoff = wlen;
|
|
|
1614 find_word(mip, FIND_KEEPCOMPOUND);
|
|
|
1615
|
|
|
1616 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1617 {
|
|
|
1618 /* Check for following word with prefix. */
|
|
|
1619 mip->mi_compoff = c;
|
|
|
1620 find_prefix(mip, FIND_COMPOUND);
|
|
|
1621 }
|
|
492
|
1622 }
|
|
501
|
1623
|
|
|
1624 if (!slang->sl_nobreak)
|
|
|
1625 break;
|
|
492
|
1626 }
|
|
483
|
1627 --mip->mi_complen;
|
|
501
|
1628 mip->mi_lp = save_lp;
|
|
485
|
1629
|
|
492
|
1630 if (slang->sl_nobreak)
|
|
|
1631 {
|
|
|
1632 nobreak_result = mip->mi_result;
|
|
|
1633 mip->mi_result = save_result;
|
|
|
1634 mip->mi_end = save_end;
|
|
|
1635 }
|
|
|
1636 else
|
|
|
1637 {
|
|
|
1638 if (mip->mi_result == SP_OK)
|
|
|
1639 break;
|
|
|
1640 continue;
|
|
|
1641 }
|
|
481
|
1642 }
|
|
|
1643
|
|
339
|
1644 if (flags & WF_BANNED)
|
|
|
1645 res = SP_BANNED;
|
|
|
1646 else if (flags & WF_REGION)
|
|
|
1647 {
|
|
|
1648 /* Check region. */
|
|
390
|
1649 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1650 res = SP_OK;
|
|
339
|
1651 else
|
|
|
1652 res = SP_LOCAL;
|
|
300
|
1653 }
|
|
339
|
1654 else if (flags & WF_RARE)
|
|
|
1655 res = SP_RARE;
|
|
307
|
1656 else
|
|
339
|
1657 res = SP_OK;
|
|
|
1658
|
|
492
|
1659 /* Always use the longest match and the best result. For NOBREAK
|
|
|
1660 * we separately keep the longest match without a following good
|
|
|
1661 * word as a fall-back. */
|
|
|
1662 if (nobreak_result == SP_BAD)
|
|
|
1663 {
|
|
|
1664 if (mip->mi_result2 > res)
|
|
|
1665 {
|
|
|
1666 mip->mi_result2 = res;
|
|
|
1667 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1668 }
|
|
|
1669 else if (mip->mi_result2 == res
|
|
|
1670 && mip->mi_end2 < mip->mi_word + wlen)
|
|
|
1671 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1672 }
|
|
|
1673 else if (mip->mi_result > res)
|
|
339
|
1674 {
|
|
|
1675 mip->mi_result = res;
|
|
|
1676 mip->mi_end = mip->mi_word + wlen;
|
|
|
1677 }
|
|
351
|
1678 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1679 mip->mi_end = mip->mi_word + wlen;
|
|
|
1680
|
|
492
|
1681 if (mip->mi_result == SP_OK)
|
|
339
|
1682 break;
|
|
300
|
1683 }
|
|
|
1684
|
|
492
|
1685 if (mip->mi_result == SP_OK)
|
|
300
|
1686 break;
|
|
|
1687 }
|
|
236
|
1688 }
|
|
|
1689
|
|
323
|
1690 /*
|
|
626
|
1691 * Return TRUE if "flags" is a valid sequence of compound flags and "word"
|
|
|
1692 * does not have too many syllables.
|
|
482
|
1693 */
|
|
|
1694 static int
|
|
483
|
1695 can_compound(slang, word, flags)
|
|
482
|
1696 slang_T *slang;
|
|
483
|
1697 char_u *word;
|
|
|
1698 char_u *flags;
|
|
|
1699 {
|
|
|
1700 regmatch_T regmatch;
|
|
495
|
1701 #ifdef FEAT_MBYTE
|
|
|
1702 char_u uflags[MAXWLEN * 2];
|
|
|
1703 int i;
|
|
|
1704 #endif
|
|
|
1705 char_u *p;
|
|
483
|
1706
|
|
|
1707 if (slang->sl_compprog == NULL)
|
|
|
1708 return FALSE;
|
|
495
|
1709 #ifdef FEAT_MBYTE
|
|
|
1710 if (enc_utf8)
|
|
|
1711 {
|
|
|
1712 /* Need to convert the single byte flags to utf8 characters. */
|
|
|
1713 p = uflags;
|
|
|
1714 for (i = 0; flags[i] != NUL; ++i)
|
|
|
1715 p += mb_char2bytes(flags[i], p);
|
|
|
1716 *p = NUL;
|
|
|
1717 p = uflags;
|
|
|
1718 }
|
|
|
1719 else
|
|
|
1720 #endif
|
|
|
1721 p = flags;
|
|
483
|
1722 regmatch.regprog = slang->sl_compprog;
|
|
|
1723 regmatch.rm_ic = FALSE;
|
|
495
|
1724 if (!vim_regexec(®match, p, 0))
|
|
483
|
1725 return FALSE;
|
|
|
1726
|
|
490
|
1727 /* Count the number of syllables. This may be slow, do it last. If there
|
|
|
1728 * are too many syllables AND the number of compound words is above
|
|
|
1729 * COMPOUNDMAX then compounding is not allowed. */
|
|
483
|
1730 if (slang->sl_compsylmax < MAXWLEN
|
|
|
1731 && count_syllables(slang, word) > slang->sl_compsylmax)
|
|
495
|
1732 return (int)STRLEN(flags) < slang->sl_compmax;
|
|
483
|
1733 return TRUE;
|
|
482
|
1734 }
|
|
|
1735
|
|
|
1736 /*
|
|
390
|
1737 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1738 * ID in "flags" for the word "word".
|
|
366
|
1739 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1740 */
|
|
|
1741 static int
|
|
455
|
1742 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1743 int totprefcnt; /* nr of prefix IDs */
|
|
|
1744 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1745 int flags;
|
|
351
|
1746 char_u *word;
|
|
|
1747 slang_T *slang;
|
|
455
|
1748 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1749 {
|
|
|
1750 int prefcnt;
|
|
|
1751 int pidx;
|
|
|
1752 regprog_T *rp;
|
|
|
1753 regmatch_T regmatch;
|
|
390
|
1754 int prefid;
|
|
|
1755
|
|
|
1756 prefid = (unsigned)flags >> 24;
|
|
351
|
1757 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1758 {
|
|
|
1759 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1760
|
|
|
1761 /* Check the prefix ID. */
|
|
|
1762 if (prefid != (pidx & 0xff))
|
|
|
1763 continue;
|
|
|
1764
|
|
390
|
1765 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1766 * suffix. */
|
|
|
1767 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1768 continue;
|
|
|
1769
|
|
351
|
1770 /* Check the condition, if there is one. The condition index is
|
|
366
|
1771 * stored in the two bytes above the prefix ID byte. */
|
|
|
1772 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1773 if (rp != NULL)
|
|
|
1774 {
|
|
|
1775 regmatch.regprog = rp;
|
|
|
1776 regmatch.rm_ic = FALSE;
|
|
|
1777 if (!vim_regexec(®match, word, 0))
|
|
|
1778 continue;
|
|
|
1779 }
|
|
455
|
1780 else if (cond_req)
|
|
|
1781 continue;
|
|
|
1782
|
|
|
1783 /* It's a match! Return the WF_ flags. */
|
|
366
|
1784 return pidx;
|
|
|
1785 }
|
|
|
1786 return 0;
|
|
351
|
1787 }
|
|
|
1788
|
|
|
1789 /*
|
|
339
|
1790 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1791 * If it does, then check the following word.
|
|
|
1792 *
|
|
485
|
1793 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a
|
|
|
1794 * prefix in a compound word.
|
|
|
1795 *
|
|
339
|
1796 * For a match mip->mi_result is updated.
|
|
|
1797 */
|
|
|
1798 static void
|
|
485
|
1799 find_prefix(mip, mode)
|
|
339
|
1800 matchinf_T *mip;
|
|
485
|
1801 int mode;
|
|
339
|
1802 {
|
|
|
1803 idx_T arridx = 0;
|
|
|
1804 int len;
|
|
|
1805 int wlen = 0;
|
|
|
1806 int flen;
|
|
|
1807 int c;
|
|
|
1808 char_u *ptr;
|
|
|
1809 idx_T lo, hi, m;
|
|
|
1810 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1811 char_u *byts;
|
|
|
1812 idx_T *idxs;
|
|
|
1813
|
|
375
|
1814 byts = slang->sl_pbyts;
|
|
|
1815 if (byts == NULL)
|
|
|
1816 return; /* array is empty */
|
|
|
1817
|
|
339
|
1818 /* We use the case-folded word here, since prefixes are always
|
|
|
1819 * case-folded. */
|
|
|
1820 ptr = mip->mi_fword;
|
|
|
1821 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
485
|
1822 if (mode == FIND_COMPOUND)
|
|
|
1823 {
|
|
|
1824 /* Skip over the previously found word(s). */
|
|
|
1825 ptr += mip->mi_compoff;
|
|
|
1826 flen -= mip->mi_compoff;
|
|
|
1827 }
|
|
339
|
1828 idxs = slang->sl_pidxs;
|
|
|
1829
|
|
|
1830 /*
|
|
|
1831 * Repeat advancing in the tree until:
|
|
|
1832 * - there is a byte that doesn't match,
|
|
|
1833 * - we reach the end of the tree,
|
|
|
1834 * - or we reach the end of the line.
|
|
|
1835 */
|
|
|
1836 for (;;)
|
|
|
1837 {
|
|
|
1838 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1839 flen = fold_more(mip);
|
|
|
1840
|
|
|
1841 len = byts[arridx++];
|
|
|
1842
|
|
|
1843 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1844 * Check if the following word matches and supports the prefix. */
|
|
|
1845 if (byts[arridx] == 0)
|
|
|
1846 {
|
|
|
1847 /* There can be several prefixes with different conditions. We
|
|
|
1848 * try them all, since we don't know which one will give the
|
|
|
1849 * longest match. The word is the same each time, pass the list
|
|
|
1850 * of possible prefixes to find_word(). */
|
|
|
1851 mip->mi_prefarridx = arridx;
|
|
|
1852 mip->mi_prefcnt = len;
|
|
|
1853 while (len > 0 && byts[arridx] == 0)
|
|
|
1854 {
|
|
|
1855 ++arridx;
|
|
|
1856 --len;
|
|
|
1857 }
|
|
|
1858 mip->mi_prefcnt -= len;
|
|
|
1859
|
|
|
1860 /* Find the word that comes after the prefix. */
|
|
|
1861 mip->mi_prefixlen = wlen;
|
|
485
|
1862 if (mode == FIND_COMPOUND)
|
|
|
1863 /* Skip over the previously found word(s). */
|
|
|
1864 mip->mi_prefixlen += mip->mi_compoff;
|
|
|
1865
|
|
455
|
1866 #ifdef FEAT_MBYTE
|
|
|
1867 if (has_mbyte)
|
|
|
1868 {
|
|
|
1869 /* Case-folded length may differ from original length. */
|
|
485
|
1870 mip->mi_cprefixlen = nofold_len(mip->mi_fword,
|
|
|
1871 mip->mi_prefixlen, mip->mi_word);
|
|
455
|
1872 }
|
|
|
1873 else
|
|
485
|
1874 mip->mi_cprefixlen = mip->mi_prefixlen;
|
|
455
|
1875 #endif
|
|
339
|
1876 find_word(mip, FIND_PREFIX);
|
|
|
1877
|
|
|
1878
|
|
|
1879 if (len == 0)
|
|
|
1880 break; /* no children, word must end here */
|
|
|
1881 }
|
|
|
1882
|
|
|
1883 /* Stop looking at end of the line. */
|
|
|
1884 if (ptr[wlen] == NUL)
|
|
|
1885 break;
|
|
|
1886
|
|
|
1887 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1888 c = ptr[wlen];
|
|
|
1889 lo = arridx;
|
|
|
1890 hi = arridx + len - 1;
|
|
|
1891 while (lo < hi)
|
|
|
1892 {
|
|
|
1893 m = (lo + hi) / 2;
|
|
|
1894 if (byts[m] > c)
|
|
|
1895 hi = m - 1;
|
|
|
1896 else if (byts[m] < c)
|
|
|
1897 lo = m + 1;
|
|
|
1898 else
|
|
|
1899 {
|
|
|
1900 lo = hi = m;
|
|
|
1901 break;
|
|
|
1902 }
|
|
|
1903 }
|
|
|
1904
|
|
|
1905 /* Stop if there is no matching byte. */
|
|
|
1906 if (hi < lo || byts[lo] != c)
|
|
|
1907 break;
|
|
|
1908
|
|
|
1909 /* Continue at the child (if there is one). */
|
|
|
1910 arridx = idxs[lo];
|
|
|
1911 ++wlen;
|
|
|
1912 --flen;
|
|
|
1913 }
|
|
|
1914 }
|
|
|
1915
|
|
|
1916 /*
|
|
|
1917 * Need to fold at least one more character. Do until next non-word character
|
|
626
|
1918 * for efficiency. Include the non-word character too.
|
|
339
|
1919 * Return the length of the folded chars in bytes.
|
|
|
1920 */
|
|
|
1921 static int
|
|
|
1922 fold_more(mip)
|
|
|
1923 matchinf_T *mip;
|
|
|
1924 {
|
|
|
1925 int flen;
|
|
|
1926 char_u *p;
|
|
|
1927
|
|
|
1928 p = mip->mi_fend;
|
|
|
1929 do
|
|
|
1930 {
|
|
|
1931 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1932 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1933
|
|
626
|
1934 /* Include the non-word character so that we can check for the word end. */
|
|
339
|
1935 if (*mip->mi_fend != NUL)
|
|
|
1936 mb_ptr_adv(mip->mi_fend);
|
|
|
1937
|
|
|
1938 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1939 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1940 MAXWLEN - mip->mi_fwordlen);
|
|
|
1941 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1942 mip->mi_fwordlen += flen;
|
|
|
1943 return flen;
|
|
|
1944 }
|
|
|
1945
|
|
|
1946 /*
|
|
323
|
1947 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1948 * case.
|
|
|
1949 */
|
|
|
1950 static int
|
|
389
|
1951 spell_valid_case(wordflags, treeflags)
|
|
|
1952 int wordflags; /* flags for the checked word. */
|
|
323
|
1953 int treeflags; /* flags for the word in the spell tree */
|
|
|
1954 {
|
|
389
|
1955 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1956 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1957 && ((treeflags & WF_ONECAP) == 0
|
|
|
1958 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1959 }
|
|
|
1960
|
|
351
|
1961 /*
|
|
|
1962 * Return TRUE if spell checking is not enabled.
|
|
|
1963 */
|
|
|
1964 static int
|
|
498
|
1965 no_spell_checking(wp)
|
|
|
1966 win_T *wp;
|
|
|
1967 {
|
|
690
|
1968 if (!wp->w_p_spell || *wp->w_buffer->b_p_spl == NUL
|
|
|
1969 || wp->w_buffer->b_langp.ga_len == 0)
|
|
351
|
1970 {
|
|
|
1971 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1972 return TRUE;
|
|
|
1973 }
|
|
|
1974 return FALSE;
|
|
|
1975 }
|
|
300
|
1976
|
|
236
|
1977 /*
|
|
|
1978 * Move to next spell error.
|
|
500
|
1979 * "curline" is FALSE for "[s", "]s", "[S" and "]S".
|
|
|
1980 * "curline" is TRUE to find word under/after cursor in the same line.
|
|
483
|
1981 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move
|
|
|
1982 * to after badly spelled word before the cursor.
|
|
495
|
1983 * Return 0 if not found, length of the badly spelled word otherwise.
|
|
236
|
1984 */
|
|
|
1985 int
|
|
498
|
1986 spell_move_to(wp, dir, allwords, curline, attrp)
|
|
|
1987 win_T *wp;
|
|
236
|
1988 int dir; /* FORWARD or BACKWARD */
|
|
500
|
1989 int allwords; /* TRUE for "[s"/"]s", FALSE for "[S"/"]S" */
|
|
323
|
1990 int curline;
|
|
534
|
1991 hlf_T *attrp; /* return: attributes of bad word or NULL
|
|
|
1992 (only when "dir" is FORWARD) */
|
|
236
|
1993 {
|
|
249
|
1994 linenr_T lnum;
|
|
|
1995 pos_T found_pos;
|
|
495
|
1996 int found_len = 0;
|
|
236
|
1997 char_u *line;
|
|
|
1998 char_u *p;
|
|
346
|
1999 char_u *endp;
|
|
534
|
2000 hlf_T attr;
|
|
236
|
2001 int len;
|
|
498
|
2002 int has_syntax = syntax_present(wp->w_buffer);
|
|
249
|
2003 int col;
|
|
|
2004 int can_spell;
|
|
346
|
2005 char_u *buf = NULL;
|
|
|
2006 int buflen = 0;
|
|
|
2007 int skip = 0;
|
|
385
|
2008 int capcol = -1;
|
|
500
|
2009 int found_one = FALSE;
|
|
|
2010 int wrapped = FALSE;
|
|
236
|
2011
|
|
498
|
2012 if (no_spell_checking(wp))
|
|
495
|
2013 return 0;
|
|
236
|
2014
|
|
249
|
2015 /*
|
|
|
2016 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
2017 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
2018 *
|
|
|
2019 * When searching backwards, we continue in the line to find the last
|
|
|
2020 * bad word (in the cursor line: before the cursor).
|
|
346
|
2021 *
|
|
|
2022 * We concatenate the start of the next line, so that wrapped words work
|
|
|
2023 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
2024 * though...
|
|
249
|
2025 */
|
|
498
|
2026 lnum = wp->w_cursor.lnum;
|
|
698
|
2027 clearpos(&found_pos);
|
|
236
|
2028
|
|
|
2029 while (!got_int)
|
|
|
2030 {
|
|
498
|
2031 line = ml_get_buf(wp->w_buffer, lnum, FALSE);
|
|
346
|
2032
|
|
|
2033 len = STRLEN(line);
|
|
|
2034 if (buflen < len + MAXWLEN + 2)
|
|
|
2035 {
|
|
|
2036 vim_free(buf);
|
|
|
2037 buflen = len + MAXWLEN + 2;
|
|
|
2038 buf = alloc(buflen);
|
|
|
2039 if (buf == NULL)
|
|
|
2040 break;
|
|
|
2041 }
|
|
|
2042
|
|
385
|
2043 /* In first line check first word for Capital. */
|
|
|
2044 if (lnum == 1)
|
|
|
2045 capcol = 0;
|
|
|
2046
|
|
|
2047 /* For checking first word with a capital skip white space. */
|
|
|
2048 if (capcol == 0)
|
|
|
2049 capcol = skipwhite(line) - line;
|
|
|
2050
|
|
346
|
2051 /* Copy the line into "buf" and append the start of the next line if
|
|
|
2052 * possible. */
|
|
|
2053 STRCPY(buf, line);
|
|
498
|
2054 if (lnum < wp->w_buffer->b_ml.ml_line_count)
|
|
346
|
2055 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
2056
|
|
|
2057 p = buf + skip;
|
|
|
2058 endp = buf + len;
|
|
|
2059 while (p < endp)
|
|
236
|
2060 {
|
|
500
|
2061 /* When searching backward don't search after the cursor. Unless
|
|
|
2062 * we wrapped around the end of the buffer. */
|
|
300
|
2063 if (dir == BACKWARD
|
|
498
|
2064 && lnum == wp->w_cursor.lnum
|
|
500
|
2065 && !wrapped
|
|
498
|
2066 && (colnr_T)(p - buf) >= wp->w_cursor.col)
|
|
300
|
2067 break;
|
|
249
|
2068
|
|
300
|
2069 /* start of word */
|
|
534
|
2070 attr = HLF_COUNT;
|
|
625
|
2071 len = spell_check(wp, p, &attr, &capcol, FALSE);
|
|
249
|
2072
|
|
534
|
2073 if (attr != HLF_COUNT)
|
|
300
|
2074 {
|
|
|
2075 /* We found a bad word. Check the attribute. */
|
|
534
|
2076 if (allwords || attr == HLF_SPB)
|
|
236
|
2077 {
|
|
500
|
2078 found_one = TRUE;
|
|
|
2079
|
|
300
|
2080 /* When searching forward only accept a bad word after
|
|
|
2081 * the cursor. */
|
|
|
2082 if (dir == BACKWARD
|
|
500
|
2083 || lnum != wp->w_cursor.lnum
|
|
498
|
2084 || (lnum == wp->w_cursor.lnum
|
|
500
|
2085 && (wrapped
|
|
|
2086 || (colnr_T)(curline ? p - buf + len
|
|
346
|
2087 : p - buf)
|
|
500
|
2088 > wp->w_cursor.col)))
|
|
236
|
2089 {
|
|
300
|
2090 if (has_syntax)
|
|
249
|
2091 {
|
|
346
|
2092 col = p - buf;
|
|
498
|
2093 (void)syn_get_id(wp, lnum, (colnr_T)col,
|
|
300
|
2094 FALSE, &can_spell);
|
|
|
2095 }
|
|
|
2096 else
|
|
|
2097 can_spell = TRUE;
|
|
249
|
2098
|
|
300
|
2099 if (can_spell)
|
|
|
2100 {
|
|
|
2101 found_pos.lnum = lnum;
|
|
346
|
2102 found_pos.col = p - buf;
|
|
249
|
2103 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
2104 found_pos.coladd = 0;
|
|
249
|
2105 #endif
|
|
300
|
2106 if (dir == FORWARD)
|
|
|
2107 {
|
|
|
2108 /* No need to search further. */
|
|
498
|
2109 wp->w_cursor = found_pos;
|
|
346
|
2110 vim_free(buf);
|
|
498
|
2111 if (attrp != NULL)
|
|
|
2112 *attrp = attr;
|
|
495
|
2113 return len;
|
|
249
|
2114 }
|
|
483
|
2115 else if (curline)
|
|
|
2116 /* Insert mode completion: put cursor after
|
|
|
2117 * the bad word. */
|
|
|
2118 found_pos.col += len;
|
|
495
|
2119 found_len = len;
|
|
249
|
2120 }
|
|
236
|
2121 }
|
|
|
2122 }
|
|
|
2123 }
|
|
|
2124
|
|
300
|
2125 /* advance to character after the word */
|
|
|
2126 p += len;
|
|
385
|
2127 capcol -= len;
|
|
236
|
2128 }
|
|
|
2129
|
|
483
|
2130 if (dir == BACKWARD && found_pos.lnum != 0)
|
|
|
2131 {
|
|
500
|
2132 /* Use the last match in the line (before the cursor). */
|
|
498
|
2133 wp->w_cursor = found_pos;
|
|
483
|
2134 vim_free(buf);
|
|
495
|
2135 return found_len;
|
|
483
|
2136 }
|
|
|
2137
|
|
323
|
2138 if (curline)
|
|
346
|
2139 break; /* only check cursor line */
|
|
323
|
2140
|
|
236
|
2141 /* Advance to next line. */
|
|
249
|
2142 if (dir == BACKWARD)
|
|
|
2143 {
|
|
500
|
2144 /* If we are back at the starting line and searched it again there
|
|
|
2145 * is no match, give up. */
|
|
|
2146 if (lnum == wp->w_cursor.lnum && wrapped)
|
|
346
|
2147 break;
|
|
500
|
2148
|
|
|
2149 if (lnum > 1)
|
|
|
2150 --lnum;
|
|
|
2151 else if (!p_ws)
|
|
|
2152 break; /* at first line and 'nowrapscan' */
|
|
|
2153 else
|
|
|
2154 {
|
|
|
2155 /* Wrap around to the end of the buffer. May search the
|
|
|
2156 * starting line again and accept the last match. */
|
|
|
2157 lnum = wp->w_buffer->b_ml.ml_line_count;
|
|
|
2158 wrapped = TRUE;
|
|
503
|
2159 if (!shortmess(SHM_SEARCH))
|
|
|
2160 give_warning((char_u *)_(top_bot_msg), TRUE);
|
|
500
|
2161 }
|
|
385
|
2162 capcol = -1;
|
|
249
|
2163 }
|
|
|
2164 else
|
|
|
2165 {
|
|
500
|
2166 if (lnum < wp->w_buffer->b_ml.ml_line_count)
|
|
|
2167 ++lnum;
|
|
|
2168 else if (!p_ws)
|
|
|
2169 break; /* at first line and 'nowrapscan' */
|
|
|
2170 else
|
|
|
2171 {
|
|
|
2172 /* Wrap around to the start of the buffer. May search the
|
|
|
2173 * starting line again and accept the first match. */
|
|
|
2174 lnum = 1;
|
|
|
2175 wrapped = TRUE;
|
|
503
|
2176 if (!shortmess(SHM_SEARCH))
|
|
|
2177 give_warning((char_u *)_(bot_top_msg), TRUE);
|
|
500
|
2178 }
|
|
|
2179
|
|
|
2180 /* If we are back at the starting line and there is no match then
|
|
|
2181 * give up. */
|
|
|
2182 if (lnum == wp->w_cursor.lnum && !found_one)
|
|
346
|
2183 break;
|
|
|
2184
|
|
|
2185 /* Skip the characters at the start of the next line that were
|
|
|
2186 * included in a match crossing line boundaries. */
|
|
534
|
2187 if (attr == HLF_COUNT)
|
|
346
|
2188 skip = p - endp;
|
|
|
2189 else
|
|
|
2190 skip = 0;
|
|
385
|
2191
|
|
|
2192 /* Capscol skips over the inserted space. */
|
|
|
2193 --capcol;
|
|
|
2194
|
|
|
2195 /* But after empty line check first word in next line */
|
|
|
2196 if (*skipwhite(line) == NUL)
|
|
|
2197 capcol = 0;
|
|
249
|
2198 }
|
|
236
|
2199
|
|
|
2200 line_breakcheck();
|
|
|
2201 }
|
|
|
2202
|
|
346
|
2203 vim_free(buf);
|
|
495
|
2204 return 0;
|
|
346
|
2205 }
|
|
|
2206
|
|
|
2207 /*
|
|
|
2208 * For spell checking: concatenate the start of the following line "line" into
|
|
|
2209 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
2210 */
|
|
|
2211 void
|
|
|
2212 spell_cat_line(buf, line, maxlen)
|
|
|
2213 char_u *buf;
|
|
|
2214 char_u *line;
|
|
|
2215 int maxlen;
|
|
|
2216 {
|
|
|
2217 char_u *p;
|
|
|
2218 int n;
|
|
|
2219
|
|
|
2220 p = skipwhite(line);
|
|
|
2221 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
2222 p = skipwhite(p + 1);
|
|
|
2223
|
|
|
2224 if (*p != NUL)
|
|
|
2225 {
|
|
|
2226 *buf = ' ';
|
|
376
|
2227 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
2228 n = p - line;
|
|
|
2229 if (n >= maxlen)
|
|
|
2230 n = maxlen - 1;
|
|
|
2231 vim_memset(buf + 1, ' ', n);
|
|
|
2232 }
|
|
236
|
2233 }
|
|
|
2234
|
|
626
|
2235 /*
|
|
|
2236 * Structure used for the cookie argument of do_in_runtimepath().
|
|
|
2237 */
|
|
501
|
2238 typedef struct spelload_S
|
|
|
2239 {
|
|
|
2240 char_u sl_lang[MAXWLEN + 1]; /* language name */
|
|
|
2241 slang_T *sl_slang; /* resulting slang_T struct */
|
|
|
2242 int sl_nobreak; /* NOBREAK language found */
|
|
|
2243 } spelload_T;
|
|
|
2244
|
|
236
|
2245 /*
|
|
307
|
2246 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
2247 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
2248 */
|
|
307
|
2249 static void
|
|
236
|
2250 spell_load_lang(lang)
|
|
|
2251 char_u *lang;
|
|
|
2252 {
|
|
310
|
2253 char_u fname_enc[85];
|
|
236
|
2254 int r;
|
|
501
|
2255 spelload_T sl;
|
|
649
|
2256 #ifdef FEAT_AUTOCMD
|
|
|
2257 int round;
|
|
|
2258 #endif
|
|
307
|
2259
|
|
310
|
2260 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
2261 * It's truncated when an error is detected. */
|
|
501
|
2262 STRCPY(sl.sl_lang, lang);
|
|
|
2263 sl.sl_slang = NULL;
|
|
|
2264 sl.sl_nobreak = FALSE;
|
|
307
|
2265
|
|
649
|
2266 #ifdef FEAT_AUTOCMD
|
|
|
2267 /* We may retry when no spell file is found for the language, an
|
|
|
2268 * autocommand may load it then. */
|
|
|
2269 for (round = 1; round <= 2; ++round)
|
|
|
2270 #endif
|
|
|
2271 {
|
|
|
2272 /*
|
|
|
2273 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
2274 */
|
|
|
2275 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
310
|
2276 "spell/%s.%s.spl", lang, spell_enc());
|
|
649
|
2277 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl);
|
|
|
2278
|
|
|
2279 if (r == FAIL && *sl.sl_lang != NUL)
|
|
|
2280 {
|
|
|
2281 /* Try loading the ASCII version. */
|
|
|
2282 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
2283 "spell/%s.ascii.spl", lang);
|
|
649
|
2284 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl);
|
|
|
2285
|
|
|
2286 #ifdef FEAT_AUTOCMD
|
|
|
2287 if (r == FAIL && *sl.sl_lang != NUL && round == 1
|
|
|
2288 && apply_autocmds(EVENT_SPELLFILEMISSING, lang,
|
|
|
2289 curbuf->b_fname, FALSE, curbuf))
|
|
|
2290 continue;
|
|
|
2291 break;
|
|
|
2292 #endif
|
|
|
2293 }
|
|
714
|
2294 #ifdef FEAT_AUTOCMD
|
|
|
2295 break;
|
|
|
2296 #endif
|
|
307
|
2297 }
|
|
|
2298
|
|
|
2299 if (r == FAIL)
|
|
649
|
2300 {
|
|
483
|
2301 smsg((char_u *)_("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""),
|
|
|
2302 lang, spell_enc(), lang);
|
|
649
|
2303 }
|
|
501
|
2304 else if (sl.sl_slang != NULL)
|
|
|
2305 {
|
|
626
|
2306 /* At least one file was loaded, now load ALL the additions. */
|
|
310
|
2307 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
501
|
2308 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &sl);
|
|
310
|
2309 }
|
|
|
2310 }
|
|
|
2311
|
|
|
2312 /*
|
|
|
2313 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
2314 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
2315 */
|
|
|
2316 static char_u *
|
|
|
2317 spell_enc()
|
|
|
2318 {
|
|
|
2319
|
|
|
2320 #ifdef FEAT_MBYTE
|
|
|
2321 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
2322 return p_enc;
|
|
|
2323 #endif
|
|
|
2324 return (char_u *)"latin1";
|
|
236
|
2325 }
|
|
|
2326
|
|
|
2327 /*
|
|
385
|
2328 * Get the name of the .spl file for the internal wordlist into
|
|
|
2329 * "fname[MAXPATHL]".
|
|
|
2330 */
|
|
|
2331 static void
|
|
|
2332 int_wordlist_spl(fname)
|
|
|
2333 char_u *fname;
|
|
|
2334 {
|
|
|
2335 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
2336 int_wordlist, spell_enc());
|
|
|
2337 }
|
|
|
2338
|
|
|
2339 /*
|
|
625
|
2340 * Allocate a new slang_T for language "lang". "lang" can be NULL.
|
|
236
|
2341 * Caller must fill "sl_next".
|
|
|
2342 */
|
|
|
2343 static slang_T *
|
|
|
2344 slang_alloc(lang)
|
|
|
2345 char_u *lang;
|
|
|
2346 {
|
|
|
2347 slang_T *lp;
|
|
|
2348
|
|
300
|
2349 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
2350 if (lp != NULL)
|
|
|
2351 {
|
|
625
|
2352 if (lang != NULL)
|
|
|
2353 lp->sl_name = vim_strsave(lang);
|
|
323
|
2354 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
625
|
2355 ga_init2(&lp->sl_repsal, sizeof(fromto_T), 10);
|
|
483
|
2356 lp->sl_compmax = MAXWLEN;
|
|
|
2357 lp->sl_compsylmax = MAXWLEN;
|
|
625
|
2358 hash_init(&lp->sl_wordcount);
|
|
|
2359 }
|
|
|
2360
|
|
236
|
2361 return lp;
|
|
|
2362 }
|
|
|
2363
|
|
|
2364 /*
|
|
|
2365 * Free the contents of an slang_T and the structure itself.
|
|
|
2366 */
|
|
|
2367 static void
|
|
|
2368 slang_free(lp)
|
|
|
2369 slang_T *lp;
|
|
|
2370 {
|
|
|
2371 vim_free(lp->sl_name);
|
|
310
|
2372 vim_free(lp->sl_fname);
|
|
|
2373 slang_clear(lp);
|
|
|
2374 vim_free(lp);
|
|
|
2375 }
|
|
|
2376
|
|
|
2377 /*
|
|
|
2378 * Clear an slang_T so that the file can be reloaded.
|
|
|
2379 */
|
|
|
2380 static void
|
|
|
2381 slang_clear(lp)
|
|
|
2382 slang_T *lp;
|
|
|
2383 {
|
|
339
|
2384 garray_T *gap;
|
|
|
2385 fromto_T *ftp;
|
|
344
|
2386 salitem_T *smp;
|
|
339
|
2387 int i;
|
|
625
|
2388 int round;
|
|
323
|
2389
|
|
300
|
2390 vim_free(lp->sl_fbyts);
|
|
310
|
2391 lp->sl_fbyts = NULL;
|
|
300
|
2392 vim_free(lp->sl_kbyts);
|
|
310
|
2393 lp->sl_kbyts = NULL;
|
|
339
|
2394 vim_free(lp->sl_pbyts);
|
|
|
2395 lp->sl_pbyts = NULL;
|
|
|
2396
|
|
300
|
2397 vim_free(lp->sl_fidxs);
|
|
310
|
2398 lp->sl_fidxs = NULL;
|
|
300
|
2399 vim_free(lp->sl_kidxs);
|
|
310
|
2400 lp->sl_kidxs = NULL;
|
|
339
|
2401 vim_free(lp->sl_pidxs);
|
|
|
2402 lp->sl_pidxs = NULL;
|
|
323
|
2403
|
|
625
|
2404 for (round = 1; round <= 2; ++round)
|
|
|
2405 {
|
|
|
2406 gap = round == 1 ? &lp->sl_rep : &lp->sl_repsal;
|
|
|
2407 while (gap->ga_len > 0)
|
|
|
2408 {
|
|
|
2409 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2410 vim_free(ftp->ft_from);
|
|
|
2411 vim_free(ftp->ft_to);
|
|
|
2412 }
|
|
|
2413 ga_clear(gap);
|
|
|
2414 }
|
|
344
|
2415
|
|
|
2416 gap = &lp->sl_sal;
|
|
375
|
2417 if (lp->sl_sofo)
|
|
376
|
2418 {
|
|
|
2419 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
2420 if (gap->ga_data != NULL)
|
|
|
2421 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
2422 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2423 vim_free(((int **)gap->ga_data)[i]);
|
|
|
2424 }
|
|
375
|
2425 else
|
|
|
2426 /* SAL items: free salitem_T items */
|
|
|
2427 while (gap->ga_len > 0)
|
|
|
2428 {
|
|
|
2429 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2430 vim_free(smp->sm_lead);
|
|
|
2431 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
2432 vim_free(smp->sm_to);
|
|
|
2433 #ifdef FEAT_MBYTE
|
|
|
2434 vim_free(smp->sm_lead_w);
|
|
|
2435 vim_free(smp->sm_oneof_w);
|
|
|
2436 vim_free(smp->sm_to_w);
|
|
|
2437 #endif
|
|
|
2438 }
|
|
344
|
2439 ga_clear(gap);
|
|
323
|
2440
|
|
339
|
2441 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
2442 vim_free(lp->sl_prefprog[i]);
|
|
376
|
2443 lp->sl_prefixcnt = 0;
|
|
339
|
2444 vim_free(lp->sl_prefprog);
|
|
376
|
2445 lp->sl_prefprog = NULL;
|
|
|
2446
|
|
714
|
2447 vim_free(lp->sl_info);
|
|
|
2448 lp->sl_info = NULL;
|
|
|
2449
|
|
376
|
2450 vim_free(lp->sl_midword);
|
|
|
2451 lp->sl_midword = NULL;
|
|
339
|
2452
|
|
483
|
2453 vim_free(lp->sl_compprog);
|
|
|
2454 vim_free(lp->sl_compstartflags);
|
|
485
|
2455 vim_free(lp->sl_compallflags);
|
|
483
|
2456 lp->sl_compprog = NULL;
|
|
|
2457 lp->sl_compstartflags = NULL;
|
|
485
|
2458 lp->sl_compallflags = NULL;
|
|
483
|
2459
|
|
|
2460 vim_free(lp->sl_syllable);
|
|
|
2461 lp->sl_syllable = NULL;
|
|
|
2462 ga_clear(&lp->sl_syl_items);
|
|
481
|
2463
|
|
625
|
2464 hash_clear_all(&lp->sl_wordcount, WC_KEY_OFF);
|
|
|
2465 hash_init(&lp->sl_wordcount);
|
|
|
2466
|
|
|
2467 #ifdef FEAT_MBYTE
|
|
|
2468 hash_clear_all(&lp->sl_map_hash, 0);
|
|
|
2469 #endif
|
|
|
2470
|
|
|
2471 /* Clear info from .sug file. */
|
|
|
2472 slang_clear_sug(lp);
|
|
483
|
2473
|
|
|
2474 lp->sl_compmax = MAXWLEN;
|
|
501
|
2475 lp->sl_compminlen = 0;
|
|
483
|
2476 lp->sl_compsylmax = MAXWLEN;
|
|
|
2477 lp->sl_regions[0] = NUL;
|
|
236
|
2478 }
|
|
|
2479
|
|
|
2480 /*
|
|
625
|
2481 * Clear the info from the .sug file in "lp".
|
|
|
2482 */
|
|
|
2483 static void
|
|
|
2484 slang_clear_sug(lp)
|
|
|
2485 slang_T *lp;
|
|
|
2486 {
|
|
|
2487 vim_free(lp->sl_sbyts);
|
|
|
2488 lp->sl_sbyts = NULL;
|
|
|
2489 vim_free(lp->sl_sidxs);
|
|
|
2490 lp->sl_sidxs = NULL;
|
|
|
2491 close_spellbuf(lp->sl_sugbuf);
|
|
|
2492 lp->sl_sugbuf = NULL;
|
|
|
2493 lp->sl_sugloaded = FALSE;
|
|
|
2494 lp->sl_sugtime = 0;
|
|
|
2495 }
|
|
|
2496
|
|
|
2497 /*
|
|
307
|
2498 * Load one spell file and store the info into a slang_T.
|
|
236
|
2499 * Invoked through do_in_runtimepath().
|
|
|
2500 */
|
|
|
2501 static void
|
|
310
|
2502 spell_load_cb(fname, cookie)
|
|
236
|
2503 char_u *fname;
|
|
501
|
2504 void *cookie;
|
|
|
2505 {
|
|
|
2506 spelload_T *slp = (spelload_T *)cookie;
|
|
|
2507 slang_T *slang;
|
|
|
2508
|
|
|
2509 slang = spell_load_file(fname, slp->sl_lang, NULL, FALSE);
|
|
|
2510 if (slang != NULL)
|
|
|
2511 {
|
|
|
2512 /* When a previously loaded file has NOBREAK also use it for the
|
|
|
2513 * ".add" files. */
|
|
|
2514 if (slp->sl_nobreak && slang->sl_add)
|
|
|
2515 slang->sl_nobreak = TRUE;
|
|
|
2516 else if (slang->sl_nobreak)
|
|
|
2517 slp->sl_nobreak = TRUE;
|
|
|
2518
|
|
|
2519 slp->sl_slang = slang;
|
|
|
2520 }
|
|
310
|
2521 }
|
|
|
2522
|
|
|
2523 /*
|
|
|
2524 * Load one spell file and store the info into a slang_T.
|
|
|
2525 *
|
|
625
|
2526 * This is invoked in three ways:
|
|
310
|
2527 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
2528 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
2529 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
2530 * points to the existing slang_T.
|
|
625
|
2531 * - Just after writing a .spl file; it's read back to produce the .sug file.
|
|
626
|
2532 * "old_lp" is NULL and "lang" is NULL. Will allocate an slang_T.
|
|
|
2533 *
|
|
323
|
2534 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
2535 */
|
|
323
|
2536 static slang_T *
|
|
|
2537 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
2538 char_u *fname;
|
|
|
2539 char_u *lang;
|
|
|
2540 slang_T *old_lp;
|
|
323
|
2541 int silent; /* no error if file doesn't exist */
|
|
310
|
2542 {
|
|
236
|
2543 FILE *fd;
|
|
483
|
2544 char_u buf[VIMSPELLMAGICL];
|
|
236
|
2545 char_u *p;
|
|
|
2546 int i;
|
|
339
|
2547 int n;
|
|
300
|
2548 int len;
|
|
236
|
2549 char_u *save_sourcing_name = sourcing_name;
|
|
|
2550 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
307
|
2551 slang_T *lp = NULL;
|
|
344
|
2552 int c = 0;
|
|
483
|
2553 int res;
|
|
236
|
2554
|
|
310
|
2555 fd = mch_fopen((char *)fname, "r");
|
|
236
|
2556 if (fd == NULL)
|
|
|
2557 {
|
|
323
|
2558 if (!silent)
|
|
|
2559 EMSG2(_(e_notopen), fname);
|
|
|
2560 else if (p_verbose > 2)
|
|
|
2561 {
|
|
|
2562 verbose_enter();
|
|
|
2563 smsg((char_u *)e_notopen, fname);
|
|
|
2564 verbose_leave();
|
|
|
2565 }
|
|
255
|
2566 goto endFAIL;
|
|
236
|
2567 }
|
|
310
|
2568 if (p_verbose > 2)
|
|
|
2569 {
|
|
|
2570 verbose_enter();
|
|
|
2571 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
2572 verbose_leave();
|
|
|
2573 }
|
|
|
2574
|
|
|
2575 if (old_lp == NULL)
|
|
|
2576 {
|
|
|
2577 lp = slang_alloc(lang);
|
|
|
2578 if (lp == NULL)
|
|
|
2579 goto endFAIL;
|
|
|
2580
|
|
|
2581 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
2582 lp->sl_fname = vim_strsave(fname);
|
|
|
2583 if (lp->sl_fname == NULL)
|
|
|
2584 goto endFAIL;
|
|
|
2585
|
|
|
2586 /* Check for .add.spl. */
|
|
|
2587 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
2588 }
|
|
|
2589 else
|
|
|
2590 lp = old_lp;
|
|
307
|
2591
|
|
236
|
2592 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
2593 sourcing_name = fname;
|
|
|
2594 sourcing_lnum = 0;
|
|
|
2595
|
|
625
|
2596 /*
|
|
|
2597 * <HEADER>: <fileID>
|
|
339
|
2598 */
|
|
236
|
2599 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
2600 buf[i] = getc(fd); /* <fileID> */
|
|
|
2601 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
2602 {
|
|
483
|
2603 EMSG(_("E757: This does not look like a spell file"));
|
|
381
|
2604 goto endFAIL;
|
|
483
|
2605 }
|
|
|
2606 c = getc(fd); /* <versionnr> */
|
|
|
2607 if (c < VIMSPELLVERSION)
|
|
|
2608 {
|
|
|
2609 EMSG(_("E771: Old spell file, needs to be updated"));
|
|
381
|
2610 goto endFAIL;
|
|
|
2611 }
|
|
483
|
2612 else if (c > VIMSPELLVERSION)
|
|
|
2613 {
|
|
|
2614 EMSG(_("E772: Spell file is for newer version of Vim"));
|
|
481
|
2615 goto endFAIL;
|
|
483
|
2616 }
|
|
|
2617
|
|
|
2618
|
|
|
2619 /*
|
|
|
2620 * <SECTIONS>: <section> ... <sectionend>
|
|
|
2621 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
2622 */
|
|
|
2623 for (;;)
|
|
|
2624 {
|
|
|
2625 n = getc(fd); /* <sectionID> or <sectionend> */
|
|
|
2626 if (n == SN_END)
|
|
|
2627 break;
|
|
|
2628 c = getc(fd); /* <sectionflags> */
|
|
695
|
2629 len = get4c(fd); /* <sectionlen> */
|
|
483
|
2630 if (len < 0)
|
|
|
2631 goto truncerr;
|
|
|
2632
|
|
|
2633 res = 0;
|
|
|
2634 switch (n)
|
|
|
2635 {
|
|
714
|
2636 case SN_INFO:
|
|
|
2637 lp->sl_info = read_string(fd, len); /* <infotext> */
|
|
|
2638 if (lp->sl_info == NULL)
|
|
|
2639 goto endFAIL;
|
|
|
2640 break;
|
|
|
2641
|
|
483
|
2642 case SN_REGION:
|
|
|
2643 res = read_region_section(fd, lp, len);
|
|
|
2644 break;
|
|
|
2645
|
|
|
2646 case SN_CHARFLAGS:
|
|
|
2647 res = read_charflags_section(fd);
|
|
|
2648 break;
|
|
|
2649
|
|
|
2650 case SN_MIDWORD:
|
|
|
2651 lp->sl_midword = read_string(fd, len); /* <midword> */
|
|
|
2652 if (lp->sl_midword == NULL)
|
|
|
2653 goto endFAIL;
|
|
|
2654 break;
|
|
|
2655
|
|
|
2656 case SN_PREFCOND:
|
|
|
2657 res = read_prefcond_section(fd, lp);
|
|
|
2658 break;
|
|
|
2659
|
|
|
2660 case SN_REP:
|
|
625
|
2661 res = read_rep_section(fd, &lp->sl_rep, lp->sl_rep_first);
|
|
|
2662 break;
|
|
|
2663
|
|
|
2664 case SN_REPSAL:
|
|
|
2665 res = read_rep_section(fd, &lp->sl_repsal, lp->sl_repsal_first);
|
|
483
|
2666 break;
|
|
|
2667
|
|
|
2668 case SN_SAL:
|
|
|
2669 res = read_sal_section(fd, lp);
|
|
|
2670 break;
|
|
|
2671
|
|
|
2672 case SN_SOFO:
|
|
|
2673 res = read_sofo_section(fd, lp);
|
|
|
2674 break;
|
|
|
2675
|
|
|
2676 case SN_MAP:
|
|
|
2677 p = read_string(fd, len); /* <mapstr> */
|
|
|
2678 if (p == NULL)
|
|
|
2679 goto endFAIL;
|
|
|
2680 set_map_str(lp, p);
|
|
|
2681 vim_free(p);
|
|
|
2682 break;
|
|
|
2683
|
|
625
|
2684 case SN_WORDS:
|
|
|
2685 res = read_words_section(fd, lp, len);
|
|
|
2686 break;
|
|
|
2687
|
|
|
2688 case SN_SUGFILE:
|
|
695
|
2689 lp->sl_sugtime = get8c(fd); /* <timestamp> */
|
|
625
|
2690 break;
|
|
|
2691
|
|
698
|
2692 case SN_NOSPLITSUGS:
|
|
|
2693 lp->sl_nosplitsugs = TRUE; /* <timestamp> */
|
|
|
2694 break;
|
|
|
2695
|
|
483
|
2696 case SN_COMPOUND:
|
|
|
2697 res = read_compound(fd, lp, len);
|
|
|
2698 break;
|
|
|
2699
|
|
492
|
2700 case SN_NOBREAK:
|
|
|
2701 lp->sl_nobreak = TRUE;
|
|
|
2702 break;
|
|
|
2703
|
|
483
|
2704 case SN_SYLLABLE:
|
|
|
2705 lp->sl_syllable = read_string(fd, len); /* <syllable> */
|
|
|
2706 if (lp->sl_syllable == NULL)
|
|
|
2707 goto endFAIL;
|
|
|
2708 if (init_syl_tab(lp) == FAIL)
|
|
|
2709 goto endFAIL;
|
|
|
2710 break;
|
|
|
2711
|
|
|
2712 default:
|
|
|
2713 /* Unsupported section. When it's required give an error
|
|
|
2714 * message. When it's not required skip the contents. */
|
|
|
2715 if (c & SNF_REQUIRED)
|
|
|
2716 {
|
|
|
2717 EMSG(_("E770: Unsupported section in spell file"));
|
|
|
2718 goto endFAIL;
|
|
|
2719 }
|
|
|
2720 while (--len >= 0)
|
|
|
2721 if (getc(fd) < 0)
|
|
|
2722 goto truncerr;
|
|
|
2723 break;
|
|
|
2724 }
|
|
625
|
2725 someerror:
|
|
483
|
2726 if (res == SP_FORMERROR)
|
|
|
2727 {
|
|
|
2728 EMSG(_(e_format));
|
|
381
|
2729 goto endFAIL;
|
|
344
|
2730 }
|
|
483
|
2731 if (res == SP_TRUNCERROR)
|
|
|
2732 {
|
|
|
2733 truncerr:
|
|
|
2734 EMSG(_(e_spell_trunc));
|
|
375
|
2735 goto endFAIL;
|
|
483
|
2736 }
|
|
495
|
2737 if (res == SP_OTHERERROR)
|
|
483
|
2738 goto endFAIL;
|
|
389
|
2739 }
|
|
236
|
2740
|
|
625
|
2741 /* <LWORDTREE> */
|
|
|
2742 res = spell_read_tree(fd, &lp->sl_fbyts, &lp->sl_fidxs, FALSE, 0);
|
|
|
2743 if (res != 0)
|
|
|
2744 goto someerror;
|
|
|
2745
|
|
|
2746 /* <KWORDTREE> */
|
|
|
2747 res = spell_read_tree(fd, &lp->sl_kbyts, &lp->sl_kidxs, FALSE, 0);
|
|
|
2748 if (res != 0)
|
|
|
2749 goto someerror;
|
|
|
2750
|
|
|
2751 /* <PREFIXTREE> */
|
|
|
2752 res = spell_read_tree(fd, &lp->sl_pbyts, &lp->sl_pidxs, TRUE,
|
|
|
2753 lp->sl_prefixcnt);
|
|
|
2754 if (res != 0)
|
|
|
2755 goto someerror;
|
|
243
|
2756
|
|
310
|
2757 /* For a new file link it in the list of spell files. */
|
|
626
|
2758 if (old_lp == NULL && lang != NULL)
|
|
310
|
2759 {
|
|
|
2760 lp->sl_next = first_lang;
|
|
|
2761 first_lang = lp;
|
|
|
2762 }
|
|
307
|
2763
|
|
255
|
2764 goto endOK;
|
|
|
2765
|
|
|
2766 endFAIL:
|
|
310
|
2767 if (lang != NULL)
|
|
|
2768 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2769 *lang = NUL;
|
|
|
2770 if (lp != NULL && old_lp == NULL)
|
|
307
|
2771 slang_free(lp);
|
|
500
|
2772 lp = NULL;
|
|
255
|
2773
|
|
|
2774 endOK:
|
|
236
|
2775 if (fd != NULL)
|
|
|
2776 fclose(fd);
|
|
|
2777 sourcing_name = save_sourcing_name;
|
|
|
2778 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2779
|
|
|
2780 return lp;
|
|
236
|
2781 }
|
|
|
2782
|
|
376
|
2783 /*
|
|
695
|
2784 * Read 2 bytes from "fd" and turn them into an int, MSB first.
|
|
|
2785 */
|
|
|
2786 static int
|
|
|
2787 get2c(fd)
|
|
|
2788 FILE *fd;
|
|
|
2789 {
|
|
|
2790 long n;
|
|
|
2791
|
|
|
2792 n = getc(fd);
|
|
|
2793 n = (n << 8) + getc(fd);
|
|
|
2794 return n;
|
|
|
2795 }
|
|
|
2796
|
|
|
2797 /*
|
|
|
2798 * Read 3 bytes from "fd" and turn them into an int, MSB first.
|
|
|
2799 */
|
|
|
2800 static int
|
|
|
2801 get3c(fd)
|
|
|
2802 FILE *fd;
|
|
|
2803 {
|
|
|
2804 long n;
|
|
|
2805
|
|
|
2806 n = getc(fd);
|
|
|
2807 n = (n << 8) + getc(fd);
|
|
|
2808 n = (n << 8) + getc(fd);
|
|
|
2809 return n;
|
|
|
2810 }
|
|
|
2811
|
|
|
2812 /*
|
|
|
2813 * Read 4 bytes from "fd" and turn them into an int, MSB first.
|
|
|
2814 */
|
|
|
2815 static int
|
|
|
2816 get4c(fd)
|
|
|
2817 FILE *fd;
|
|
|
2818 {
|
|
|
2819 long n;
|
|
|
2820
|
|
|
2821 n = getc(fd);
|
|
|
2822 n = (n << 8) + getc(fd);
|
|
|
2823 n = (n << 8) + getc(fd);
|
|
|
2824 n = (n << 8) + getc(fd);
|
|
|
2825 return n;
|
|
|
2826 }
|
|
|
2827
|
|
|
2828 /*
|
|
|
2829 * Read 8 bytes from "fd" and turn them into a time_t, MSB first.
|
|
|
2830 */
|
|
|
2831 static time_t
|
|
|
2832 get8c(fd)
|
|
|
2833 FILE *fd;
|
|
|
2834 {
|
|
|
2835 time_t n = 0;
|
|
|
2836 int i;
|
|
|
2837
|
|
|
2838 for (i = 0; i < 8; ++i)
|
|
|
2839 n = (n << 8) + getc(fd);
|
|
|
2840 return n;
|
|
|
2841 }
|
|
|
2842
|
|
|
2843 /*
|
|
376
|
2844 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2845 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2846 * Returns NULL when the count is zero.
|
|
483
|
2847 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result
|
|
|
2848 * otherwise.
|
|
376
|
2849 */
|
|
|
2850 static char_u *
|
|
389
|
2851 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2852 FILE *fd;
|
|
|
2853 int cnt_bytes;
|
|
389
|
2854 int *cntp;
|
|
376
|
2855 {
|
|
|
2856 int cnt = 0;
|
|
|
2857 int i;
|
|
|
2858 char_u *str;
|
|
|
2859
|
|
|
2860 /* read the length bytes, MSB first */
|
|
|
2861 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2862 cnt = (cnt << 8) + getc(fd);
|
|
|
2863 if (cnt < 0)
|
|
|
2864 {
|
|
483
|
2865 *cntp = SP_TRUNCERROR;
|
|
376
|
2866 return NULL;
|
|
|
2867 }
|
|
389
|
2868 *cntp = cnt;
|
|
|
2869 if (cnt == 0)
|
|
|
2870 return NULL; /* nothing to read, return NULL */
|
|
376
|
2871
|
|
483
|
2872 str = read_string(fd, cnt);
|
|
|
2873 if (str == NULL)
|
|
495
|
2874 *cntp = SP_OTHERERROR;
|
|
483
|
2875 return str;
|
|
|
2876 }
|
|
|
2877
|
|
|
2878 /*
|
|
|
2879 * Read a string of length "cnt" from "fd" into allocated memory.
|
|
|
2880 * Returns NULL when out of memory.
|
|
|
2881 */
|
|
|
2882 static char_u *
|
|
|
2883 read_string(fd, cnt)
|
|
|
2884 FILE *fd;
|
|
|
2885 int cnt;
|
|
|
2886 {
|
|
|
2887 char_u *str;
|
|
|
2888 int i;
|
|
|
2889
|
|
376
|
2890 /* allocate memory */
|
|
|
2891 str = alloc((unsigned)cnt + 1);
|
|
483
|
2892 if (str != NULL)
|
|
|
2893 {
|
|
|
2894 /* Read the string. Doesn't check for truncated file. */
|
|
|
2895 for (i = 0; i < cnt; ++i)
|
|
|
2896 str[i] = getc(fd);
|
|
|
2897 str[i] = NUL;
|
|
|
2898 }
|
|
|
2899 return str;
|
|
|
2900 }
|
|
|
2901
|
|
|
2902 /*
|
|
|
2903 * Read SN_REGION: <regionname> ...
|
|
|
2904 * Return SP_*ERROR flags.
|
|
|
2905 */
|
|
|
2906 static int
|
|
|
2907 read_region_section(fd, lp, len)
|
|
|
2908 FILE *fd;
|
|
|
2909 slang_T *lp;
|
|
|
2910 int len;
|
|
|
2911 {
|
|
|
2912 int i;
|
|
|
2913
|
|
|
2914 if (len > 16)
|
|
|
2915 return SP_FORMERROR;
|
|
|
2916 for (i = 0; i < len; ++i)
|
|
|
2917 lp->sl_regions[i] = getc(fd); /* <regionname> */
|
|
|
2918 lp->sl_regions[len] = NUL;
|
|
|
2919 return 0;
|
|
|
2920 }
|
|
|
2921
|
|
|
2922 /*
|
|
|
2923 * Read SN_CHARFLAGS section: <charflagslen> <charflags>
|
|
|
2924 * <folcharslen> <folchars>
|
|
|
2925 * Return SP_*ERROR flags.
|
|
|
2926 */
|
|
|
2927 static int
|
|
|
2928 read_charflags_section(fd)
|
|
|
2929 FILE *fd;
|
|
|
2930 {
|
|
|
2931 char_u *flags;
|
|
|
2932 char_u *fol;
|
|
|
2933 int flagslen, follen;
|
|
|
2934
|
|
|
2935 /* <charflagslen> <charflags> */
|
|
|
2936 flags = read_cnt_string(fd, 1, &flagslen);
|
|
|
2937 if (flagslen < 0)
|
|
|
2938 return flagslen;
|
|
|
2939
|
|
|
2940 /* <folcharslen> <folchars> */
|
|
|
2941 fol = read_cnt_string(fd, 2, &follen);
|
|
|
2942 if (follen < 0)
|
|
|
2943 {
|
|
|
2944 vim_free(flags);
|
|
|
2945 return follen;
|
|
|
2946 }
|
|
|
2947
|
|
|
2948 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
2949 if (flags != NULL && fol != NULL)
|
|
|
2950 set_spell_charflags(flags, flagslen, fol);
|
|
|
2951
|
|
|
2952 vim_free(flags);
|
|
|
2953 vim_free(fol);
|
|
|
2954
|
|
|
2955 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2956 if ((flags == NULL) != (fol == NULL))
|
|
|
2957 return SP_FORMERROR;
|
|
|
2958 return 0;
|
|
|
2959 }
|
|
|
2960
|
|
|
2961 /*
|
|
|
2962 * Read SN_PREFCOND section.
|
|
|
2963 * Return SP_*ERROR flags.
|
|
|
2964 */
|
|
|
2965 static int
|
|
|
2966 read_prefcond_section(fd, lp)
|
|
|
2967 FILE *fd;
|
|
|
2968 slang_T *lp;
|
|
|
2969 {
|
|
|
2970 int cnt;
|
|
|
2971 int i;
|
|
|
2972 int n;
|
|
|
2973 char_u *p;
|
|
|
2974 char_u buf[MAXWLEN + 1];
|
|
|
2975
|
|
|
2976 /* <prefcondcnt> <prefcond> ... */
|
|
695
|
2977 cnt = get2c(fd); /* <prefcondcnt> */
|
|
483
|
2978 if (cnt <= 0)
|
|
|
2979 return SP_FORMERROR;
|
|
|
2980
|
|
|
2981 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2982 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2983 if (lp->sl_prefprog == NULL)
|
|
495
|
2984 return SP_OTHERERROR;
|
|
483
|
2985 lp->sl_prefixcnt = cnt;
|
|
|
2986
|
|
376
|
2987 for (i = 0; i < cnt; ++i)
|
|
483
|
2988 {
|
|
|
2989 /* <prefcond> : <condlen> <condstr> */
|
|
|
2990 n = getc(fd); /* <condlen> */
|
|
|
2991 if (n < 0 || n >= MAXWLEN)
|
|
|
2992 return SP_FORMERROR;
|
|
|
2993
|
|
|
2994 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2995 * compile the regexp program used to check for the condition. */
|
|
|
2996 if (n > 0)
|
|
|
2997 {
|
|
|
2998 buf[0] = '^'; /* always match at one position only */
|
|
|
2999 p = buf + 1;
|
|
|
3000 while (n-- > 0)
|
|
|
3001 *p++ = getc(fd); /* <condstr> */
|
|
|
3002 *p = NUL;
|
|
|
3003 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
3004 }
|
|
|
3005 }
|
|
|
3006 return 0;
|
|
|
3007 }
|
|
|
3008
|
|
|
3009 /*
|
|
625
|
3010 * Read REP or REPSAL items section from "fd": <repcount> <rep> ...
|
|
483
|
3011 * Return SP_*ERROR flags.
|
|
|
3012 */
|
|
|
3013 static int
|
|
625
|
3014 read_rep_section(fd, gap, first)
|
|
483
|
3015 FILE *fd;
|
|
625
|
3016 garray_T *gap;
|
|
|
3017 short *first;
|
|
483
|
3018 {
|
|
|
3019 int cnt;
|
|
|
3020 fromto_T *ftp;
|
|
|
3021 int i;
|
|
|
3022
|
|
695
|
3023 cnt = get2c(fd); /* <repcount> */
|
|
483
|
3024 if (cnt < 0)
|
|
|
3025 return SP_TRUNCERROR;
|
|
|
3026
|
|
|
3027 if (ga_grow(gap, cnt) == FAIL)
|
|
495
|
3028 return SP_OTHERERROR;
|
|
483
|
3029
|
|
|
3030 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
3031 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
3032 {
|
|
|
3033 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
|
3034 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
3035 if (i < 0)
|
|
|
3036 return i;
|
|
|
3037 if (i == 0)
|
|
|
3038 return SP_FORMERROR;
|
|
|
3039 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
3040 if (i <= 0)
|
|
|
3041 {
|
|
|
3042 vim_free(ftp->ft_from);
|
|
|
3043 if (i < 0)
|
|
|
3044 return i;
|
|
|
3045 return SP_FORMERROR;
|
|
|
3046 }
|
|
|
3047 }
|
|
|
3048
|
|
|
3049 /* Fill the first-index table. */
|
|
|
3050 for (i = 0; i < 256; ++i)
|
|
|
3051 first[i] = -1;
|
|
|
3052 for (i = 0; i < gap->ga_len; ++i)
|
|
|
3053 {
|
|
|
3054 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
3055 if (first[*ftp->ft_from] == -1)
|
|
|
3056 first[*ftp->ft_from] = i;
|
|
|
3057 }
|
|
|
3058 return 0;
|
|
|
3059 }
|
|
|
3060
|
|
|
3061 /*
|
|
|
3062 * Read SN_SAL section: <salflags> <salcount> <sal> ...
|
|
|
3063 * Return SP_*ERROR flags.
|
|
|
3064 */
|
|
|
3065 static int
|
|
|
3066 read_sal_section(fd, slang)
|
|
|
3067 FILE *fd;
|
|
|
3068 slang_T *slang;
|
|
|
3069 {
|
|
|
3070 int i;
|
|
|
3071 int cnt;
|
|
|
3072 garray_T *gap;
|
|
|
3073 salitem_T *smp;
|
|
|
3074 int ccnt;
|
|
|
3075 char_u *p;
|
|
485
|
3076 int c = NUL;
|
|
483
|
3077
|
|
|
3078 slang->sl_sofo = FALSE;
|
|
|
3079
|
|
|
3080 i = getc(fd); /* <salflags> */
|
|
|
3081 if (i & SAL_F0LLOWUP)
|
|
|
3082 slang->sl_followup = TRUE;
|
|
|
3083 if (i & SAL_COLLAPSE)
|
|
|
3084 slang->sl_collapse = TRUE;
|
|
|
3085 if (i & SAL_REM_ACCENTS)
|
|
|
3086 slang->sl_rem_accents = TRUE;
|
|
|
3087
|
|
695
|
3088 cnt = get2c(fd); /* <salcount> */
|
|
483
|
3089 if (cnt < 0)
|
|
|
3090 return SP_TRUNCERROR;
|
|
|
3091
|
|
|
3092 gap = &slang->sl_sal;
|
|
|
3093 ga_init2(gap, sizeof(salitem_T), 10);
|
|
548
|
3094 if (ga_grow(gap, cnt + 1) == FAIL)
|
|
495
|
3095 return SP_OTHERERROR;
|
|
483
|
3096
|
|
|
3097 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
3098 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
3099 {
|
|
|
3100 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
3101 ccnt = getc(fd); /* <salfromlen> */
|
|
|
3102 if (ccnt < 0)
|
|
|
3103 return SP_TRUNCERROR;
|
|
|
3104 if ((p = alloc(ccnt + 2)) == NULL)
|
|
495
|
3105 return SP_OTHERERROR;
|
|
483
|
3106 smp->sm_lead = p;
|
|
|
3107
|
|
|
3108 /* Read up to the first special char into sm_lead. */
|
|
|
3109 for (i = 0; i < ccnt; ++i)
|
|
|
3110 {
|
|
|
3111 c = getc(fd); /* <salfrom> */
|
|
|
3112 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
3113 break;
|
|
|
3114 *p++ = c;
|
|
|
3115 }
|
|
|
3116 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
3117 *p++ = NUL;
|
|
|
3118
|
|
|
3119 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
3120 if (c == '(')
|
|
|
3121 {
|
|
|
3122 smp->sm_oneof = p;
|
|
|
3123 for (++i; i < ccnt; ++i)
|
|
|
3124 {
|
|
|
3125 c = getc(fd); /* <salfrom> */
|
|
|
3126 if (c == ')')
|
|
|
3127 break;
|
|
|
3128 *p++ = c;
|
|
|
3129 }
|
|
|
3130 *p++ = NUL;
|
|
|
3131 if (++i < ccnt)
|
|
|
3132 c = getc(fd);
|
|
|
3133 }
|
|
|
3134 else
|
|
|
3135 smp->sm_oneof = NULL;
|
|
|
3136
|
|
|
3137 /* Any following chars go in sm_rules. */
|
|
|
3138 smp->sm_rules = p;
|
|
|
3139 if (i < ccnt)
|
|
|
3140 /* store the char we got while checking for end of sm_lead */
|
|
|
3141 *p++ = c;
|
|
|
3142 for (++i; i < ccnt; ++i)
|
|
|
3143 *p++ = getc(fd); /* <salfrom> */
|
|
|
3144 *p++ = NUL;
|
|
|
3145
|
|
|
3146 /* <saltolen> <salto> */
|
|
|
3147 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
3148 if (ccnt < 0)
|
|
|
3149 {
|
|
|
3150 vim_free(smp->sm_lead);
|
|
|
3151 return ccnt;
|
|
|
3152 }
|
|
|
3153
|
|
|
3154 #ifdef FEAT_MBYTE
|
|
|
3155 if (has_mbyte)
|
|
|
3156 {
|
|
|
3157 /* convert the multi-byte strings to wide char strings */
|
|
|
3158 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
3159 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
3160 if (smp->sm_oneof == NULL)
|
|
|
3161 smp->sm_oneof_w = NULL;
|
|
|
3162 else
|
|
|
3163 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
3164 if (smp->sm_to == NULL)
|
|
|
3165 smp->sm_to_w = NULL;
|
|
|
3166 else
|
|
|
3167 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
3168 if (smp->sm_lead_w == NULL
|
|
|
3169 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
3170 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
|
3171 {
|
|
|
3172 vim_free(smp->sm_lead);
|
|
|
3173 vim_free(smp->sm_to);
|
|
|
3174 vim_free(smp->sm_lead_w);
|
|
|
3175 vim_free(smp->sm_oneof_w);
|
|
|
3176 vim_free(smp->sm_to_w);
|
|
495
|
3177 return SP_OTHERERROR;
|
|
483
|
3178 }
|
|
|
3179 }
|
|
|
3180 #endif
|
|
|
3181 }
|
|
|
3182
|
|
548
|
3183 if (gap->ga_len > 0)
|
|
|
3184 {
|
|
|
3185 /* Add one extra entry to mark the end with an empty sm_lead. Avoids
|
|
|
3186 * that we need to check the index every time. */
|
|
|
3187 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
3188 if ((p = alloc(1)) == NULL)
|
|
|
3189 return SP_OTHERERROR;
|
|
|
3190 p[0] = NUL;
|
|
|
3191 smp->sm_lead = p;
|
|
|
3192 smp->sm_leadlen = 0;
|
|
|
3193 smp->sm_oneof = NULL;
|
|
|
3194 smp->sm_rules = p;
|
|
|
3195 smp->sm_to = NULL;
|
|
|
3196 #ifdef FEAT_MBYTE
|
|
|
3197 if (has_mbyte)
|
|
|
3198 {
|
|
|
3199 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
3200 smp->sm_leadlen = 0;
|
|
|
3201 smp->sm_oneof_w = NULL;
|
|
|
3202 smp->sm_to_w = NULL;
|
|
|
3203 }
|
|
|
3204 #endif
|
|
|
3205 ++gap->ga_len;
|
|
|
3206 }
|
|
|
3207
|
|
483
|
3208 /* Fill the first-index table. */
|
|
|
3209 set_sal_first(slang);
|
|
|
3210
|
|
|
3211 return 0;
|
|
|
3212 }
|
|
|
3213
|
|
|
3214 /*
|
|
625
|
3215 * Read SN_WORDS: <word> ...
|
|
|
3216 * Return SP_*ERROR flags.
|
|
|
3217 */
|
|
|
3218 static int
|
|
|
3219 read_words_section(fd, lp, len)
|
|
|
3220 FILE *fd;
|
|
|
3221 slang_T *lp;
|
|
|
3222 int len;
|
|
|
3223 {
|
|
|
3224 int done = 0;
|
|
|
3225 int i;
|
|
|
3226 char_u word[MAXWLEN];
|
|
|
3227
|
|
|
3228 while (done < len)
|
|
|
3229 {
|
|
|
3230 /* Read one word at a time. */
|
|
|
3231 for (i = 0; ; ++i)
|
|
|
3232 {
|
|
|
3233 word[i] = getc(fd);
|
|
|
3234 if (word[i] == NUL)
|
|
|
3235 break;
|
|
|
3236 if (i == MAXWLEN - 1)
|
|
|
3237 return SP_FORMERROR;
|
|
|
3238 }
|
|
|
3239
|
|
|
3240 /* Init the count to 10. */
|
|
|
3241 count_common_word(lp, word, -1, 10);
|
|
|
3242 done += i + 1;
|
|
|
3243 }
|
|
|
3244 return 0;
|
|
|
3245 }
|
|
|
3246
|
|
|
3247 /*
|
|
|
3248 * Add a word to the hashtable of common words.
|
|
|
3249 * If it's already there then the counter is increased.
|
|
|
3250 */
|
|
|
3251 static void
|
|
|
3252 count_common_word(lp, word, len, count)
|
|
|
3253 slang_T *lp;
|
|
|
3254 char_u *word;
|
|
|
3255 int len; /* word length, -1 for upto NUL */
|
|
|
3256 int count; /* 1 to count once, 10 to init */
|
|
|
3257 {
|
|
|
3258 hash_T hash;
|
|
|
3259 hashitem_T *hi;
|
|
|
3260 wordcount_T *wc;
|
|
|
3261 char_u buf[MAXWLEN];
|
|
|
3262 char_u *p;
|
|
|
3263
|
|
|
3264 if (len == -1)
|
|
|
3265 p = word;
|
|
|
3266 else
|
|
|
3267 {
|
|
|
3268 vim_strncpy(buf, word, len);
|
|
|
3269 p = buf;
|
|
|
3270 }
|
|
|
3271
|
|
|
3272 hash = hash_hash(p);
|
|
|
3273 hi = hash_lookup(&lp->sl_wordcount, p, hash);
|
|
|
3274 if (HASHITEM_EMPTY(hi))
|
|
|
3275 {
|
|
|
3276 wc = (wordcount_T *)alloc(sizeof(wordcount_T) + STRLEN(p));
|
|
|
3277 if (wc == NULL)
|
|
|
3278 return;
|
|
|
3279 STRCPY(wc->wc_word, p);
|
|
|
3280 wc->wc_count = count;
|
|
|
3281 hash_add_item(&lp->sl_wordcount, hi, wc->wc_word, hash);
|
|
|
3282 }
|
|
|
3283 else
|
|
|
3284 {
|
|
|
3285 wc = HI2WC(hi);
|
|
|
3286 if ((wc->wc_count += count) < (unsigned)count) /* check for overflow */
|
|
|
3287 wc->wc_count = MAXWORDCOUNT;
|
|
|
3288 }
|
|
|
3289 }
|
|
|
3290
|
|
|
3291 /*
|
|
|
3292 * Adjust the score of common words.
|
|
|
3293 */
|
|
|
3294 static int
|
|
|
3295 score_wordcount_adj(slang, score, word, split)
|
|
|
3296 slang_T *slang;
|
|
|
3297 int score;
|
|
|
3298 char_u *word;
|
|
|
3299 int split; /* word was split, less bonus */
|
|
|
3300 {
|
|
|
3301 hashitem_T *hi;
|
|
|
3302 wordcount_T *wc;
|
|
|
3303 int bonus;
|
|
|
3304 int newscore;
|
|
|
3305
|
|
|
3306 hi = hash_find(&slang->sl_wordcount, word);
|
|
|
3307 if (!HASHITEM_EMPTY(hi))
|
|
|
3308 {
|
|
|
3309 wc = HI2WC(hi);
|
|
|
3310 if (wc->wc_count < SCORE_THRES2)
|
|
|
3311 bonus = SCORE_COMMON1;
|
|
|
3312 else if (wc->wc_count < SCORE_THRES3)
|
|
|
3313 bonus = SCORE_COMMON2;
|
|
|
3314 else
|
|
|
3315 bonus = SCORE_COMMON3;
|
|
|
3316 if (split)
|
|
|
3317 newscore = score - bonus / 2;
|
|
|
3318 else
|
|
|
3319 newscore = score - bonus;
|
|
|
3320 if (newscore < 0)
|
|
|
3321 return 0;
|
|
|
3322 return newscore;
|
|
|
3323 }
|
|
|
3324 return score;
|
|
|
3325 }
|
|
|
3326
|
|
|
3327 /*
|
|
483
|
3328 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
3329 * Return SP_*ERROR flags.
|
|
|
3330 */
|
|
|
3331 static int
|
|
|
3332 read_sofo_section(fd, slang)
|
|
|
3333 FILE *fd;
|
|
|
3334 slang_T *slang;
|
|
|
3335 {
|
|
|
3336 int cnt;
|
|
|
3337 char_u *from, *to;
|
|
|
3338 int res;
|
|
|
3339
|
|
|
3340 slang->sl_sofo = TRUE;
|
|
|
3341
|
|
|
3342 /* <sofofromlen> <sofofrom> */
|
|
|
3343 from = read_cnt_string(fd, 2, &cnt);
|
|
|
3344 if (cnt < 0)
|
|
|
3345 return cnt;
|
|
|
3346
|
|
|
3347 /* <sofotolen> <sofoto> */
|
|
|
3348 to = read_cnt_string(fd, 2, &cnt);
|
|
|
3349 if (cnt < 0)
|
|
|
3350 {
|
|
|
3351 vim_free(from);
|
|
|
3352 return cnt;
|
|
|
3353 }
|
|
|
3354
|
|
|
3355 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */
|
|
|
3356 if (from != NULL && to != NULL)
|
|
|
3357 res = set_sofo(slang, from, to);
|
|
|
3358 else if (from != NULL || to != NULL)
|
|
|
3359 res = SP_FORMERROR; /* only one of two strings is an error */
|
|
|
3360 else
|
|
|
3361 res = 0;
|
|
|
3362
|
|
|
3363 vim_free(from);
|
|
|
3364 vim_free(to);
|
|
|
3365 return res;
|
|
|
3366 }
|
|
|
3367
|
|
|
3368 /*
|
|
|
3369 * Read the compound section from the .spl file:
|
|
|
3370 * <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
3371 * Returns SP_*ERROR flags.
|
|
|
3372 */
|
|
|
3373 static int
|
|
|
3374 read_compound(fd, slang, len)
|
|
|
3375 FILE *fd;
|
|
|
3376 slang_T *slang;
|
|
|
3377 int len;
|
|
|
3378 {
|
|
|
3379 int todo = len;
|
|
|
3380 int c;
|
|
|
3381 int atstart;
|
|
|
3382 char_u *pat;
|
|
|
3383 char_u *pp;
|
|
|
3384 char_u *cp;
|
|
485
|
3385 char_u *ap;
|
|
483
|
3386
|
|
|
3387 if (todo < 2)
|
|
|
3388 return SP_FORMERROR; /* need at least two bytes */
|
|
|
3389
|
|
|
3390 --todo;
|
|
|
3391 c = getc(fd); /* <compmax> */
|
|
|
3392 if (c < 2)
|
|
|
3393 c = MAXWLEN;
|
|
|
3394 slang->sl_compmax = c;
|
|
|
3395
|
|
|
3396 --todo;
|
|
|
3397 c = getc(fd); /* <compminlen> */
|
|
|
3398 if (c < 1)
|
|
501
|
3399 c = 0;
|
|
483
|
3400 slang->sl_compminlen = c;
|
|
|
3401
|
|
|
3402 --todo;
|
|
|
3403 c = getc(fd); /* <compsylmax> */
|
|
|
3404 if (c < 1)
|
|
|
3405 c = MAXWLEN;
|
|
|
3406 slang->sl_compsylmax = c;
|
|
|
3407
|
|
714
|
3408 /* Turn the COMPOUNDRULE items into a regexp pattern:
|
|
483
|
3409 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$".
|
|
495
|
3410 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes.
|
|
|
3411 * Conversion to utf-8 may double the size. */
|
|
|
3412 c = todo * 2 + 7;
|
|
|
3413 #ifdef FEAT_MBYTE
|
|
|
3414 if (enc_utf8)
|
|
|
3415 c += todo * 2;
|
|
|
3416 #endif
|
|
|
3417 pat = alloc((unsigned)c);
|
|
483
|
3418 if (pat == NULL)
|
|
495
|
3419 return SP_OTHERERROR;
|
|
483
|
3420
|
|
485
|
3421 /* We also need a list of all flags that can appear at the start and one
|
|
|
3422 * for all flags. */
|
|
483
|
3423 cp = alloc(todo + 1);
|
|
|
3424 if (cp == NULL)
|
|
|
3425 {
|
|
|
3426 vim_free(pat);
|
|
495
|
3427 return SP_OTHERERROR;
|
|
483
|
3428 }
|
|
|
3429 slang->sl_compstartflags = cp;
|
|
|
3430 *cp = NUL;
|
|
|
3431
|
|
485
|
3432 ap = alloc(todo + 1);
|
|
|
3433 if (ap == NULL)
|
|
|
3434 {
|
|
|
3435 vim_free(pat);
|
|
495
|
3436 return SP_OTHERERROR;
|
|
485
|
3437 }
|
|
|
3438 slang->sl_compallflags = ap;
|
|
|
3439 *ap = NUL;
|
|
|
3440
|
|
483
|
3441 pp = pat;
|
|
|
3442 *pp++ = '^';
|
|
|
3443 *pp++ = '\\';
|
|
|
3444 *pp++ = '(';
|
|
|
3445
|
|
|
3446 atstart = 1;
|
|
|
3447 while (todo-- > 0)
|
|
|
3448 {
|
|
|
3449 c = getc(fd); /* <compflags> */
|
|
485
|
3450
|
|
|
3451 /* Add all flags to "sl_compallflags". */
|
|
|
3452 if (vim_strchr((char_u *)"+*[]/", c) == NULL
|
|
495
|
3453 && !byte_in_str(slang->sl_compallflags, c))
|
|
485
|
3454 {
|
|
|
3455 *ap++ = c;
|
|
|
3456 *ap = NUL;
|
|
|
3457 }
|
|
|
3458
|
|
483
|
3459 if (atstart != 0)
|
|
|
3460 {
|
|
|
3461 /* At start of item: copy flags to "sl_compstartflags". For a
|
|
|
3462 * [abc] item set "atstart" to 2 and copy up to the ']'. */
|
|
|
3463 if (c == '[')
|
|
|
3464 atstart = 2;
|
|
|
3465 else if (c == ']')
|
|
|
3466 atstart = 0;
|
|
|
3467 else
|
|
|
3468 {
|
|
495
|
3469 if (!byte_in_str(slang->sl_compstartflags, c))
|
|
483
|
3470 {
|
|
|
3471 *cp++ = c;
|
|
|
3472 *cp = NUL;
|
|
|
3473 }
|
|
|
3474 if (atstart == 1)
|
|
|
3475 atstart = 0;
|
|
|
3476 }
|
|
|
3477 }
|
|
|
3478 if (c == '/') /* slash separates two items */
|
|
|
3479 {
|
|
|
3480 *pp++ = '\\';
|
|
|
3481 *pp++ = '|';
|
|
|
3482 atstart = 1;
|
|
|
3483 }
|
|
|
3484 else /* normal char, "[abc]" and '*' are copied as-is */
|
|
|
3485 {
|
|
500
|
3486 if (c == '+' || c == '~')
|
|
483
|
3487 *pp++ = '\\'; /* "a+" becomes "a\+" */
|
|
495
|
3488 #ifdef FEAT_MBYTE
|
|
|
3489 if (enc_utf8)
|
|
|
3490 pp += mb_char2bytes(c, pp);
|
|
|
3491 else
|
|
|
3492 #endif
|
|
|
3493 *pp++ = c;
|
|
483
|
3494 }
|
|
|
3495 }
|
|
|
3496
|
|
|
3497 *pp++ = '\\';
|
|
|
3498 *pp++ = ')';
|
|
|
3499 *pp++ = '$';
|
|
|
3500 *pp = NUL;
|
|
|
3501
|
|
|
3502 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
3503 vim_free(pat);
|
|
|
3504 if (slang->sl_compprog == NULL)
|
|
|
3505 return SP_FORMERROR;
|
|
|
3506
|
|
|
3507 return 0;
|
|
|
3508 }
|
|
|
3509
|
|
495
|
3510 /*
|
|
498
|
3511 * Return TRUE if byte "n" appears in "str".
|
|
495
|
3512 * Like strchr() but independent of locale.
|
|
|
3513 */
|
|
|
3514 static int
|
|
498
|
3515 byte_in_str(str, n)
|
|
495
|
3516 char_u *str;
|
|
498
|
3517 int n;
|
|
495
|
3518 {
|
|
|
3519 char_u *p;
|
|
|
3520
|
|
|
3521 for (p = str; *p != NUL; ++p)
|
|
498
|
3522 if (*p == n)
|
|
495
|
3523 return TRUE;
|
|
|
3524 return FALSE;
|
|
|
3525 }
|
|
|
3526
|
|
483
|
3527 #define SY_MAXLEN 30
|
|
|
3528 typedef struct syl_item_S
|
|
|
3529 {
|
|
|
3530 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */
|
|
|
3531 int sy_len;
|
|
|
3532 } syl_item_T;
|
|
|
3533
|
|
|
3534 /*
|
|
|
3535 * Truncate "slang->sl_syllable" at the first slash and put the following items
|
|
|
3536 * in "slang->sl_syl_items".
|
|
|
3537 */
|
|
|
3538 static int
|
|
|
3539 init_syl_tab(slang)
|
|
|
3540 slang_T *slang;
|
|
|
3541 {
|
|
|
3542 char_u *p;
|
|
|
3543 char_u *s;
|
|
|
3544 int l;
|
|
|
3545 syl_item_T *syl;
|
|
|
3546
|
|
|
3547 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4);
|
|
|
3548 p = vim_strchr(slang->sl_syllable, '/');
|
|
|
3549 while (p != NULL)
|
|
|
3550 {
|
|
|
3551 *p++ = NUL;
|
|
495
|
3552 if (*p == NUL) /* trailing slash */
|
|
483
|
3553 break;
|
|
|
3554 s = p;
|
|
|
3555 p = vim_strchr(p, '/');
|
|
|
3556 if (p == NULL)
|
|
|
3557 l = STRLEN(s);
|
|
|
3558 else
|
|
|
3559 l = p - s;
|
|
|
3560 if (l >= SY_MAXLEN)
|
|
|
3561 return SP_FORMERROR;
|
|
|
3562 if (ga_grow(&slang->sl_syl_items, 1) == FAIL)
|
|
495
|
3563 return SP_OTHERERROR;
|
|
483
|
3564 syl = ((syl_item_T *)slang->sl_syl_items.ga_data)
|
|
|
3565 + slang->sl_syl_items.ga_len++;
|
|
|
3566 vim_strncpy(syl->sy_chars, s, l);
|
|
|
3567 syl->sy_len = l;
|
|
|
3568 }
|
|
|
3569 return OK;
|
|
|
3570 }
|
|
|
3571
|
|
|
3572 /*
|
|
|
3573 * Count the number of syllables in "word".
|
|
|
3574 * When "word" contains spaces the syllables after the last space are counted.
|
|
|
3575 * Returns zero if syllables are not defines.
|
|
|
3576 */
|
|
|
3577 static int
|
|
|
3578 count_syllables(slang, word)
|
|
|
3579 slang_T *slang;
|
|
|
3580 char_u *word;
|
|
|
3581 {
|
|
|
3582 int cnt = 0;
|
|
|
3583 int skip = FALSE;
|
|
|
3584 char_u *p;
|
|
|
3585 int len;
|
|
|
3586 int i;
|
|
|
3587 syl_item_T *syl;
|
|
|
3588 int c;
|
|
|
3589
|
|
|
3590 if (slang->sl_syllable == NULL)
|
|
|
3591 return 0;
|
|
|
3592
|
|
|
3593 for (p = word; *p != NUL; p += len)
|
|
|
3594 {
|
|
|
3595 /* When running into a space reset counter. */
|
|
|
3596 if (*p == ' ')
|
|
|
3597 {
|
|
|
3598 len = 1;
|
|
|
3599 cnt = 0;
|
|
|
3600 continue;
|
|
|
3601 }
|
|
|
3602
|
|
|
3603 /* Find longest match of syllable items. */
|
|
|
3604 len = 0;
|
|
|
3605 for (i = 0; i < slang->sl_syl_items.ga_len; ++i)
|
|
|
3606 {
|
|
|
3607 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i;
|
|
|
3608 if (syl->sy_len > len
|
|
|
3609 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0)
|
|
|
3610 len = syl->sy_len;
|
|
|
3611 }
|
|
|
3612 if (len != 0) /* found a match, count syllable */
|
|
|
3613 {
|
|
|
3614 ++cnt;
|
|
|
3615 skip = FALSE;
|
|
|
3616 }
|
|
|
3617 else
|
|
|
3618 {
|
|
|
3619 /* No recognized syllable item, at least a syllable char then? */
|
|
|
3620 #ifdef FEAT_MBYTE
|
|
|
3621 c = mb_ptr2char(p);
|
|
|
3622 len = (*mb_ptr2len)(p);
|
|
|
3623 #else
|
|
|
3624 c = *p;
|
|
|
3625 len = 1;
|
|
|
3626 #endif
|
|
|
3627 if (vim_strchr(slang->sl_syllable, c) == NULL)
|
|
|
3628 skip = FALSE; /* No, search for next syllable */
|
|
|
3629 else if (!skip)
|
|
|
3630 {
|
|
|
3631 ++cnt; /* Yes, count it */
|
|
|
3632 skip = TRUE; /* don't count following syllable chars */
|
|
|
3633 }
|
|
|
3634 }
|
|
|
3635 }
|
|
|
3636 return cnt;
|
|
376
|
3637 }
|
|
|
3638
|
|
381
|
3639 /*
|
|
|
3640 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
483
|
3641 * Returns SP_*ERROR flags when there is something wrong.
|
|
381
|
3642 */
|
|
|
3643 static int
|
|
|
3644 set_sofo(lp, from, to)
|
|
|
3645 slang_T *lp;
|
|
|
3646 char_u *from;
|
|
|
3647 char_u *to;
|
|
|
3648 {
|
|
|
3649 int i;
|
|
|
3650
|
|
|
3651 #ifdef FEAT_MBYTE
|
|
|
3652 garray_T *gap;
|
|
|
3653 char_u *s;
|
|
|
3654 char_u *p;
|
|
|
3655 int c;
|
|
|
3656 int *inp;
|
|
|
3657
|
|
|
3658 if (has_mbyte)
|
|
|
3659 {
|
|
|
3660 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
3661 * characters. The index is the low byte of the character.
|
|
|
3662 * The list contains from-to pairs with a terminating NUL.
|
|
|
3663 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
3664 gap = &lp->sl_sal;
|
|
|
3665 ga_init2(gap, sizeof(int *), 1);
|
|
|
3666 if (ga_grow(gap, 256) == FAIL)
|
|
495
|
3667 return SP_OTHERERROR;
|
|
381
|
3668 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
3669 gap->ga_len = 256;
|
|
|
3670
|
|
|
3671 /* First count the number of items for each list. Temporarily use
|
|
|
3672 * sl_sal_first[] for this. */
|
|
|
3673 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3674 {
|
|
474
|
3675 c = mb_cptr2char_adv(&p);
|
|
|
3676 mb_cptr_adv(s);
|
|
381
|
3677 if (c >= 256)
|
|
|
3678 ++lp->sl_sal_first[c & 0xff];
|
|
|
3679 }
|
|
|
3680 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
483
|
3681 return SP_FORMERROR;
|
|
381
|
3682
|
|
|
3683 /* Allocate the lists. */
|
|
|
3684 for (i = 0; i < 256; ++i)
|
|
|
3685 if (lp->sl_sal_first[i] > 0)
|
|
|
3686 {
|
|
|
3687 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
3688 if (p == NULL)
|
|
495
|
3689 return SP_OTHERERROR;
|
|
381
|
3690 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
3691 *(int *)p = 0;
|
|
|
3692 }
|
|
|
3693
|
|
|
3694 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
3695 * list. */
|
|
|
3696 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
3697 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3698 {
|
|
474
|
3699 c = mb_cptr2char_adv(&p);
|
|
|
3700 i = mb_cptr2char_adv(&s);
|
|
381
|
3701 if (c >= 256)
|
|
|
3702 {
|
|
|
3703 /* Append the from-to chars at the end of the list with
|
|
|
3704 * the low byte. */
|
|
|
3705 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
3706 while (*inp != 0)
|
|
|
3707 ++inp;
|
|
|
3708 *inp++ = c; /* from char */
|
|
|
3709 *inp++ = i; /* to char */
|
|
|
3710 *inp++ = NUL; /* NUL at the end */
|
|
|
3711 }
|
|
|
3712 else
|
|
|
3713 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
3714 lp->sl_sal_first[c] = i;
|
|
|
3715 }
|
|
|
3716 }
|
|
|
3717 else
|
|
|
3718 #endif
|
|
|
3719 {
|
|
|
3720 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
3721 if (STRLEN(from) != STRLEN(to))
|
|
483
|
3722 return SP_FORMERROR;
|
|
381
|
3723
|
|
|
3724 for (i = 0; to[i] != NUL; ++i)
|
|
|
3725 lp->sl_sal_first[from[i]] = to[i];
|
|
|
3726 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
3727 }
|
|
|
3728
|
|
483
|
3729 return 0;
|
|
381
|
3730 }
|
|
|
3731
|
|
|
3732 /*
|
|
|
3733 * Fill the first-index table for "lp".
|
|
|
3734 */
|
|
|
3735 static void
|
|
|
3736 set_sal_first(lp)
|
|
|
3737 slang_T *lp;
|
|
|
3738 {
|
|
|
3739 salfirst_T *sfirst;
|
|
|
3740 int i;
|
|
|
3741 salitem_T *smp;
|
|
|
3742 int c;
|
|
|
3743 garray_T *gap = &lp->sl_sal;
|
|
|
3744
|
|
|
3745 sfirst = lp->sl_sal_first;
|
|
|
3746 for (i = 0; i < 256; ++i)
|
|
|
3747 sfirst[i] = -1;
|
|
|
3748 smp = (salitem_T *)gap->ga_data;
|
|
|
3749 for (i = 0; i < gap->ga_len; ++i)
|
|
|
3750 {
|
|
|
3751 #ifdef FEAT_MBYTE
|
|
|
3752 if (has_mbyte)
|
|
|
3753 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
3754 * the character, for other encodings it should differ for most
|
|
|
3755 * characters. */
|
|
|
3756 c = *smp[i].sm_lead_w & 0xff;
|
|
|
3757 else
|
|
|
3758 #endif
|
|
|
3759 c = *smp[i].sm_lead;
|
|
|
3760 if (sfirst[c] == -1)
|
|
|
3761 {
|
|
|
3762 sfirst[c] = i;
|
|
|
3763 #ifdef FEAT_MBYTE
|
|
|
3764 if (has_mbyte)
|
|
|
3765 {
|
|
|
3766 int n;
|
|
|
3767
|
|
|
3768 /* Make sure all entries with this byte are following each
|
|
|
3769 * other. Move the ones that are in the wrong position. Do
|
|
|
3770 * keep the same ordering! */
|
|
|
3771 while (i + 1 < gap->ga_len
|
|
|
3772 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
3773 /* Skip over entry with same index byte. */
|
|
|
3774 ++i;
|
|
|
3775
|
|
|
3776 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
3777 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
3778 {
|
|
|
3779 salitem_T tsal;
|
|
|
3780
|
|
|
3781 /* Move entry with same index byte after the entries
|
|
|
3782 * we already found. */
|
|
|
3783 ++i;
|
|
|
3784 --n;
|
|
|
3785 tsal = smp[i + n];
|
|
|
3786 mch_memmove(smp + i + 1, smp + i,
|
|
|
3787 sizeof(salitem_T) * n);
|
|
|
3788 smp[i] = tsal;
|
|
|
3789 }
|
|
|
3790 }
|
|
|
3791 #endif
|
|
|
3792 }
|
|
|
3793 }
|
|
|
3794 }
|
|
376
|
3795
|
|
372
|
3796 #ifdef FEAT_MBYTE
|
|
|
3797 /*
|
|
|
3798 * Turn a multi-byte string into a wide character string.
|
|
|
3799 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
3800 */
|
|
|
3801 static int *
|
|
|
3802 mb_str2wide(s)
|
|
|
3803 char_u *s;
|
|
|
3804 {
|
|
|
3805 int *res;
|
|
|
3806 char_u *p;
|
|
|
3807 int i = 0;
|
|
|
3808
|
|
|
3809 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
3810 if (res != NULL)
|
|
|
3811 {
|
|
|
3812 for (p = s; *p != NUL; )
|
|
|
3813 res[i++] = mb_ptr2char_adv(&p);
|
|
|
3814 res[i] = NUL;
|
|
|
3815 }
|
|
|
3816 return res;
|
|
|
3817 }
|
|
|
3818 #endif
|
|
|
3819
|
|
236
|
3820 /*
|
|
625
|
3821 * Read a tree from the .spl or .sug file.
|
|
|
3822 * Allocates the memory and stores pointers in "bytsp" and "idxsp".
|
|
|
3823 * This is skipped when the tree has zero length.
|
|
|
3824 * Returns zero when OK, SP_ value for an error.
|
|
|
3825 */
|
|
|
3826 static int
|
|
|
3827 spell_read_tree(fd, bytsp, idxsp, prefixtree, prefixcnt)
|
|
|
3828 FILE *fd;
|
|
|
3829 char_u **bytsp;
|
|
|
3830 idx_T **idxsp;
|
|
|
3831 int prefixtree; /* TRUE for the prefix tree */
|
|
|
3832 int prefixcnt; /* when "prefixtree" is TRUE: prefix count */
|
|
|
3833 {
|
|
|
3834 int len;
|
|
|
3835 int idx;
|
|
|
3836 char_u *bp;
|
|
|
3837 idx_T *ip;
|
|
|
3838
|
|
|
3839 /* The tree size was computed when writing the file, so that we can
|
|
|
3840 * allocate it as one long block. <nodecount> */
|
|
695
|
3841 len = get4c(fd);
|
|
625
|
3842 if (len < 0)
|
|
|
3843 return SP_TRUNCERROR;
|
|
|
3844 if (len > 0)
|
|
|
3845 {
|
|
|
3846 /* Allocate the byte array. */
|
|
|
3847 bp = lalloc((long_u)len, TRUE);
|
|
|
3848 if (bp == NULL)
|
|
|
3849 return SP_OTHERERROR;
|
|
|
3850 *bytsp = bp;
|
|
|
3851
|
|
|
3852 /* Allocate the index array. */
|
|
|
3853 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
3854 if (ip == NULL)
|
|
|
3855 return SP_OTHERERROR;
|
|
|
3856 *idxsp = ip;
|
|
|
3857
|
|
|
3858 /* Recursively read the tree and store it in the array. */
|
|
|
3859 idx = read_tree_node(fd, bp, ip, len, 0, prefixtree, prefixcnt);
|
|
|
3860 if (idx < 0)
|
|
|
3861 return idx;
|
|
|
3862 }
|
|
|
3863 return 0;
|
|
|
3864 }
|
|
|
3865
|
|
|
3866 /*
|
|
300
|
3867 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
3868 * "byts" and index array "idxs". Recursively read the children.
|
|
|
3869 *
|
|
625
|
3870 * NOTE: The code here must match put_node()!
|
|
|
3871 *
|
|
|
3872 * Returns the index (>= 0) following the siblings.
|
|
|
3873 * Returns SP_TRUNCERROR if the file is shorter than expected.
|
|
|
3874 * Returns SP_FORMERROR if there is a format error.
|
|
236
|
3875 */
|
|
324
|
3876 static idx_T
|
|
625
|
3877 read_tree_node(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
3878 FILE *fd;
|
|
|
3879 char_u *byts;
|
|
324
|
3880 idx_T *idxs;
|
|
300
|
3881 int maxidx; /* size of arrays */
|
|
324
|
3882 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
3883 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
3884 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
3885 {
|
|
300
|
3886 int len;
|
|
|
3887 int i;
|
|
|
3888 int n;
|
|
324
|
3889 idx_T idx = startidx;
|
|
300
|
3890 int c;
|
|
366
|
3891 int c2;
|
|
300
|
3892 #define SHARED_MASK 0x8000000
|
|
236
|
3893
|
|
300
|
3894 len = getc(fd); /* <siblingcount> */
|
|
|
3895 if (len <= 0)
|
|
625
|
3896 return SP_TRUNCERROR;
|
|
300
|
3897
|
|
|
3898 if (startidx + len >= maxidx)
|
|
625
|
3899 return SP_FORMERROR;
|
|
300
|
3900 byts[idx++] = len;
|
|
|
3901
|
|
|
3902 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
3903 for (i = 1; i <= len; ++i)
|
|
236
|
3904 {
|
|
300
|
3905 c = getc(fd); /* <byte> */
|
|
|
3906 if (c < 0)
|
|
625
|
3907 return SP_TRUNCERROR;
|
|
300
|
3908 if (c <= BY_SPECIAL)
|
|
|
3909 {
|
|
366
|
3910 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
3911 {
|
|
|
3912 /* No flags, all regions. */
|
|
|
3913 idxs[idx] = 0;
|
|
|
3914 c = 0;
|
|
|
3915 }
|
|
390
|
3916 else if (c != BY_INDEX)
|
|
300
|
3917 {
|
|
339
|
3918 if (prefixtree)
|
|
|
3919 {
|
|
455
|
3920 /* Read the optional pflags byte, the prefix ID and the
|
|
|
3921 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
3922 * byte, the condition index shifted up 8 bits, the flags
|
|
|
3923 * shifted up 24 bits. */
|
|
|
3924 if (c == BY_FLAGS)
|
|
|
3925 c = getc(fd) << 24; /* <pflags> */
|
|
|
3926 else
|
|
|
3927 c = 0;
|
|
|
3928
|
|
481
|
3929 c |= getc(fd); /* <affixID> */
|
|
455
|
3930
|
|
695
|
3931 n = get2c(fd); /* <prefcondnr> */
|
|
339
|
3932 if (n >= maxprefcondnr)
|
|
625
|
3933 return SP_FORMERROR;
|
|
455
|
3934 c |= (n << 8);
|
|
339
|
3935 }
|
|
390
|
3936 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
3937 {
|
|
|
3938 /* Read flags and optional region and prefix ID. In
|
|
390
|
3939 * idxs[] the flags go in the low two bytes, region above
|
|
|
3940 * that and prefix ID above the region. */
|
|
|
3941 c2 = c;
|
|
339
|
3942 c = getc(fd); /* <flags> */
|
|
390
|
3943 if (c2 == BY_FLAGS2)
|
|
|
3944 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
3945 if (c & WF_REGION)
|
|
390
|
3946 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
3947 if (c & WF_AFX)
|
|
|
3948 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
3949 }
|
|
|
3950
|
|
300
|
3951 idxs[idx] = c;
|
|
|
3952 c = 0;
|
|
|
3953 }
|
|
|
3954 else /* c == BY_INDEX */
|
|
|
3955 {
|
|
|
3956 /* <nodeidx> */
|
|
695
|
3957 n = get3c(fd);
|
|
300
|
3958 if (n < 0 || n >= maxidx)
|
|
625
|
3959 return SP_FORMERROR;
|
|
300
|
3960 idxs[idx] = n + SHARED_MASK;
|
|
|
3961 c = getc(fd); /* <xbyte> */
|
|
|
3962 }
|
|
|
3963 }
|
|
|
3964 byts[idx++] = c;
|
|
236
|
3965 }
|
|
|
3966
|
|
300
|
3967 /* Recursively read the children for non-shared siblings.
|
|
|
3968 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
3969 * remove SHARED_MASK) */
|
|
|
3970 for (i = 1; i <= len; ++i)
|
|
|
3971 if (byts[startidx + i] != 0)
|
|
|
3972 {
|
|
|
3973 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
3974 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
3975 else
|
|
|
3976 {
|
|
|
3977 idxs[startidx + i] = idx;
|
|
625
|
3978 idx = read_tree_node(fd, byts, idxs, maxidx, idx,
|
|
339
|
3979 prefixtree, maxprefcondnr);
|
|
300
|
3980 if (idx < 0)
|
|
|
3981 break;
|
|
|
3982 }
|
|
|
3983 }
|
|
236
|
3984
|
|
300
|
3985 return idx;
|
|
236
|
3986 }
|
|
|
3987
|
|
|
3988 /*
|
|
|
3989 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
3990 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
3991 */
|
|
|
3992 char_u *
|
|
|
3993 did_set_spelllang(buf)
|
|
|
3994 buf_T *buf;
|
|
|
3995 {
|
|
|
3996 garray_T ga;
|
|
351
|
3997 char_u *splp;
|
|
236
|
3998 char_u *region;
|
|
416
|
3999 char_u region_cp[3];
|
|
355
|
4000 int filename;
|
|
236
|
4001 int region_mask;
|
|
503
|
4002 slang_T *slang;
|
|
236
|
4003 int c;
|
|
351
|
4004 char_u lang[MAXWLEN + 1];
|
|
323
|
4005 char_u spf_name[MAXPATHL];
|
|
351
|
4006 int len;
|
|
|
4007 char_u *p;
|
|
381
|
4008 int round;
|
|
385
|
4009 char_u *spf;
|
|
389
|
4010 char_u *use_region = NULL;
|
|
|
4011 int dont_use_region = FALSE;
|
|
501
|
4012 int nobreak = FALSE;
|
|
503
|
4013 int i, j;
|
|
|
4014 langp_T *lp, *lp2;
|
|
236
|
4015
|
|
|
4016 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
4017 clear_midword(buf);
|
|
236
|
4018
|
|
351
|
4019 /* loop over comma separated language names. */
|
|
|
4020 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
4021 {
|
|
|
4022 /* Get one language name. */
|
|
|
4023 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
4024
|
|
240
|
4025 region = NULL;
|
|
351
|
4026 len = STRLEN(lang);
|
|
355
|
4027
|
|
|
4028 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
4029 * If there is a region name let "region" point to it and remove it
|
|
|
4030 * from the name. */
|
|
|
4031 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
4032 {
|
|
355
|
4033 filename = TRUE;
|
|
|
4034
|
|
416
|
4035 /* Locate a region and remove it from the file name. */
|
|
|
4036 p = vim_strchr(gettail(lang), '_');
|
|
|
4037 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
4038 && !ASCII_ISALPHA(p[3]))
|
|
|
4039 {
|
|
|
4040 vim_strncpy(region_cp, p + 1, 2);
|
|
|
4041 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
4042 len -= 3;
|
|
|
4043 region = region_cp;
|
|
|
4044 }
|
|
|
4045 else
|
|
|
4046 dont_use_region = TRUE;
|
|
|
4047
|
|
355
|
4048 /* Check if we loaded this language before. */
|
|
503
|
4049 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
4050 if (fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME)
|
|
355
|
4051 break;
|
|
236
|
4052 }
|
|
355
|
4053 else
|
|
|
4054 {
|
|
|
4055 filename = FALSE;
|
|
|
4056 if (len > 3 && lang[len - 3] == '_')
|
|
|
4057 {
|
|
|
4058 region = lang + len - 2;
|
|
|
4059 len -= 3;
|
|
|
4060 lang[len] = NUL;
|
|
389
|
4061 }
|
|
|
4062 else
|
|
|
4063 dont_use_region = TRUE;
|
|
355
|
4064
|
|
|
4065 /* Check if we loaded this language before. */
|
|
503
|
4066 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
4067 if (STRICMP(lang, slang->sl_name) == 0)
|
|
355
|
4068 break;
|
|
|
4069 }
|
|
236
|
4070
|
|
416
|
4071 if (region != NULL)
|
|
|
4072 {
|
|
|
4073 /* If the region differs from what was used before then don't
|
|
|
4074 * use it for 'spellfile'. */
|
|
|
4075 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
4076 dont_use_region = TRUE;
|
|
|
4077 use_region = region;
|
|
|
4078 }
|
|
|
4079
|
|
351
|
4080 /* If not found try loading the language now. */
|
|
503
|
4081 if (slang == NULL)
|
|
355
|
4082 {
|
|
|
4083 if (filename)
|
|
|
4084 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
4085 else
|
|
|
4086 spell_load_lang(lang);
|
|
|
4087 }
|
|
236
|
4088
|
|
307
|
4089 /*
|
|
351
|
4090 * Loop over the languages, there can be several files for "lang".
|
|
307
|
4091 */
|
|
503
|
4092 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
4093 if (filename ? fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME
|
|
|
4094 : STRICMP(lang, slang->sl_name) == 0)
|
|
236
|
4095 {
|
|
316
|
4096 region_mask = REGION_ALL;
|
|
355
|
4097 if (!filename && region != NULL)
|
|
236
|
4098 {
|
|
307
|
4099 /* find region in sl_regions */
|
|
503
|
4100 c = find_region(slang->sl_regions, region);
|
|
307
|
4101 if (c == REGION_ALL)
|
|
|
4102 {
|
|
503
|
4103 if (slang->sl_add)
|
|
389
|
4104 {
|
|
503
|
4105 if (*slang->sl_regions != NUL)
|
|
389
|
4106 /* This addition file is for other regions. */
|
|
|
4107 region_mask = 0;
|
|
|
4108 }
|
|
|
4109 else
|
|
|
4110 /* This is probably an error. Give a warning and
|
|
|
4111 * accept the words anyway. */
|
|
351
|
4112 smsg((char_u *)
|
|
|
4113 _("Warning: region %s not supported"),
|
|
|
4114 region);
|
|
307
|
4115 }
|
|
|
4116 else
|
|
|
4117 region_mask = 1 << c;
|
|
236
|
4118 }
|
|
307
|
4119
|
|
389
|
4120 if (region_mask != 0)
|
|
307
|
4121 {
|
|
389
|
4122 if (ga_grow(&ga, 1) == FAIL)
|
|
|
4123 {
|
|
|
4124 ga_clear(&ga);
|
|
|
4125 return e_outofmem;
|
|
|
4126 }
|
|
503
|
4127 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang;
|
|
389
|
4128 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
4129 ++ga.ga_len;
|
|
503
|
4130 use_midword(slang, buf);
|
|
|
4131 if (slang->sl_nobreak)
|
|
501
|
4132 nobreak = TRUE;
|
|
307
|
4133 }
|
|
385
|
4134 }
|
|
|
4135 }
|
|
|
4136
|
|
|
4137 /* round 0: load int_wordlist, if possible.
|
|
|
4138 * round 1: load first name in 'spellfile'.
|
|
|
4139 * round 2: load second name in 'spellfile.
|
|
|
4140 * etc. */
|
|
|
4141 spf = curbuf->b_p_spf;
|
|
|
4142 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
4143 {
|
|
|
4144 if (round == 0)
|
|
|
4145 {
|
|
|
4146 /* Internal wordlist, if there is one. */
|
|
|
4147 if (int_wordlist == NULL)
|
|
381
|
4148 continue;
|
|
385
|
4149 int_wordlist_spl(spf_name);
|
|
381
|
4150 }
|
|
|
4151 else
|
|
|
4152 {
|
|
385
|
4153 /* One entry in 'spellfile'. */
|
|
|
4154 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
4155 STRCAT(spf_name, ".spl");
|
|
|
4156
|
|
|
4157 /* If it was already found above then skip it. */
|
|
|
4158 for (c = 0; c < ga.ga_len; ++c)
|
|
500
|
4159 {
|
|
|
4160 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname;
|
|
|
4161 if (p != NULL && fullpathcmp(spf_name, p, FALSE) == FPC_SAME)
|
|
385
|
4162 break;
|
|
500
|
4163 }
|
|
385
|
4164 if (c < ga.ga_len)
|
|
381
|
4165 continue;
|
|
|
4166 }
|
|
|
4167
|
|
351
|
4168 /* Check if it was loaded already. */
|
|
503
|
4169 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
4170 if (fullpathcmp(spf_name, slang->sl_fname, FALSE) == FPC_SAME)
|
|
323
|
4171 break;
|
|
503
|
4172 if (slang == NULL)
|
|
323
|
4173 {
|
|
351
|
4174 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
4175 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
4176 * use an arbitrary name. */
|
|
|
4177 if (round == 0)
|
|
|
4178 STRCPY(lang, "internal wordlist");
|
|
|
4179 else
|
|
|
4180 {
|
|
|
4181 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
4182 p = vim_strchr(lang, '.');
|
|
|
4183 if (p != NULL)
|
|
|
4184 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
4185 }
|
|
503
|
4186 slang = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
501
|
4187
|
|
|
4188 /* If one of the languages has NOBREAK we assume the addition
|
|
|
4189 * files also have this. */
|
|
503
|
4190 if (slang != NULL && nobreak)
|
|
|
4191 slang->sl_nobreak = TRUE;
|
|
|
4192 }
|
|
|
4193 if (slang != NULL && ga_grow(&ga, 1) == OK)
|
|
323
|
4194 {
|
|
389
|
4195 region_mask = REGION_ALL;
|
|
|
4196 if (use_region != NULL && !dont_use_region)
|
|
|
4197 {
|
|
|
4198 /* find region in sl_regions */
|
|
503
|
4199 c = find_region(slang->sl_regions, use_region);
|
|
389
|
4200 if (c != REGION_ALL)
|
|
|
4201 region_mask = 1 << c;
|
|
503
|
4202 else if (*slang->sl_regions != NUL)
|
|
389
|
4203 /* This spell file is for other regions. */
|
|
|
4204 region_mask = 0;
|
|
|
4205 }
|
|
|
4206
|
|
|
4207 if (region_mask != 0)
|
|
|
4208 {
|
|
503
|
4209 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang;
|
|
|
4210 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL;
|
|
|
4211 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL;
|
|
389
|
4212 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
4213 ++ga.ga_len;
|
|
503
|
4214 use_midword(slang, buf);
|
|
389
|
4215 }
|
|
323
|
4216 }
|
|
|
4217 }
|
|
|
4218
|
|
236
|
4219 /* Everything is fine, store the new b_langp value. */
|
|
|
4220 ga_clear(&buf->b_langp);
|
|
|
4221 buf->b_langp = ga;
|
|
|
4222
|
|
503
|
4223 /* For each language figure out what language to use for sound folding and
|
|
|
4224 * REP items. If the language doesn't support it itself use another one
|
|
|
4225 * with the same name. E.g. for "en-math" use "en". */
|
|
|
4226 for (i = 0; i < ga.ga_len; ++i)
|
|
|
4227 {
|
|
|
4228 lp = LANGP_ENTRY(ga, i);
|
|
|
4229
|
|
|
4230 /* sound folding */
|
|
|
4231 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
4232 /* language does sound folding itself */
|
|
|
4233 lp->lp_sallang = lp->lp_slang;
|
|
|
4234 else
|
|
|
4235 /* find first similar language that does sound folding */
|
|
|
4236 for (j = 0; j < ga.ga_len; ++j)
|
|
|
4237 {
|
|
|
4238 lp2 = LANGP_ENTRY(ga, j);
|
|
|
4239 if (lp2->lp_slang->sl_sal.ga_len > 0
|
|
|
4240 && STRNCMP(lp->lp_slang->sl_name,
|
|
|
4241 lp2->lp_slang->sl_name, 2) == 0)
|
|
|
4242 {
|
|
|
4243 lp->lp_sallang = lp2->lp_slang;
|
|
|
4244 break;
|
|
|
4245 }
|
|
|
4246 }
|
|
|
4247
|
|
|
4248 /* REP items */
|
|
|
4249 if (lp->lp_slang->sl_rep.ga_len > 0)
|
|
|
4250 /* language has REP items itself */
|
|
|
4251 lp->lp_replang = lp->lp_slang;
|
|
|
4252 else
|
|
625
|
4253 /* find first similar language that has REP items */
|
|
503
|
4254 for (j = 0; j < ga.ga_len; ++j)
|
|
|
4255 {
|
|
|
4256 lp2 = LANGP_ENTRY(ga, j);
|
|
|
4257 if (lp2->lp_slang->sl_rep.ga_len > 0
|
|
|
4258 && STRNCMP(lp->lp_slang->sl_name,
|
|
|
4259 lp2->lp_slang->sl_name, 2) == 0)
|
|
|
4260 {
|
|
|
4261 lp->lp_replang = lp2->lp_slang;
|
|
|
4262 break;
|
|
|
4263 }
|
|
|
4264 }
|
|
|
4265 }
|
|
|
4266
|
|
236
|
4267 return NULL;
|
|
|
4268 }
|
|
|
4269
|
|
|
4270 /*
|
|
376
|
4271 * Clear the midword characters for buffer "buf".
|
|
|
4272 */
|
|
|
4273 static void
|
|
|
4274 clear_midword(buf)
|
|
|
4275 buf_T *buf;
|
|
|
4276 {
|
|
|
4277 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
4278 #ifdef FEAT_MBYTE
|
|
|
4279 vim_free(buf->b_spell_ismw_mb);
|
|
|
4280 buf->b_spell_ismw_mb = NULL;
|
|
|
4281 #endif
|
|
|
4282 }
|
|
|
4283
|
|
|
4284 /*
|
|
|
4285 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
4286 * They add up to any currently used midword characters.
|
|
|
4287 */
|
|
|
4288 static void
|
|
|
4289 use_midword(lp, buf)
|
|
|
4290 slang_T *lp;
|
|
|
4291 buf_T *buf;
|
|
|
4292 {
|
|
|
4293 char_u *p;
|
|
|
4294
|
|
389
|
4295 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
4296 return;
|
|
|
4297
|
|
376
|
4298 for (p = lp->sl_midword; *p != NUL; )
|
|
|
4299 #ifdef FEAT_MBYTE
|
|
|
4300 if (has_mbyte)
|
|
|
4301 {
|
|
|
4302 int c, l, n;
|
|
|
4303 char_u *bp;
|
|
|
4304
|
|
|
4305 c = mb_ptr2char(p);
|
|
474
|
4306 l = (*mb_ptr2len)(p);
|
|
|
4307 if (c < 256 && l <= 2)
|
|
376
|
4308 buf->b_spell_ismw[c] = TRUE;
|
|
|
4309 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
4310 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
4311 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
4312 else
|
|
|
4313 {
|
|
|
4314 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
4315 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
4316 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
4317 if (bp != NULL)
|
|
|
4318 {
|
|
|
4319 vim_free(buf->b_spell_ismw_mb);
|
|
|
4320 buf->b_spell_ismw_mb = bp;
|
|
|
4321 vim_strncpy(bp + n, p, l);
|
|
|
4322 }
|
|
|
4323 }
|
|
|
4324 p += l;
|
|
|
4325 }
|
|
|
4326 else
|
|
|
4327 #endif
|
|
|
4328 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
4329 }
|
|
|
4330
|
|
|
4331 /*
|
|
236
|
4332 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
4333 * Each region is simply stored as the two characters of it's name.
|
|
381
|
4334 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
4335 */
|
|
|
4336 static int
|
|
|
4337 find_region(rp, region)
|
|
|
4338 char_u *rp;
|
|
|
4339 char_u *region;
|
|
|
4340 {
|
|
|
4341 int i;
|
|
|
4342
|
|
|
4343 for (i = 0; ; i += 2)
|
|
|
4344 {
|
|
|
4345 if (rp[i] == NUL)
|
|
|
4346 return REGION_ALL;
|
|
|
4347 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
4348 break;
|
|
|
4349 }
|
|
|
4350 return i / 2;
|
|
|
4351 }
|
|
|
4352
|
|
|
4353 /*
|
|
323
|
4354 * Return case type of word:
|
|
236
|
4355 * w word 0
|
|
300
|
4356 * Word WF_ONECAP
|
|
|
4357 * W WORD WF_ALLCAP
|
|
|
4358 * WoRd wOrd WF_KEEPCAP
|
|
236
|
4359 */
|
|
|
4360 static int
|
|
|
4361 captype(word, end)
|
|
|
4362 char_u *word;
|
|
323
|
4363 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
4364 {
|
|
|
4365 char_u *p;
|
|
|
4366 int c;
|
|
|
4367 int firstcap;
|
|
|
4368 int allcap;
|
|
|
4369 int past_second = FALSE; /* past second word char */
|
|
|
4370
|
|
|
4371 /* find first letter */
|
|
376
|
4372 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
4373 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
4374 return 0; /* only non-word characters, illegal word */
|
|
|
4375 #ifdef FEAT_MBYTE
|
|
310
|
4376 if (has_mbyte)
|
|
|
4377 c = mb_ptr2char_adv(&p);
|
|
|
4378 else
|
|
236
|
4379 #endif
|
|
310
|
4380 c = *p++;
|
|
324
|
4381 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
4382
|
|
|
4383 /*
|
|
|
4384 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
4385 * But a word with an upper char only at start is a ONECAP.
|
|
|
4386 */
|
|
323
|
4387 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
4388 if (spell_iswordp_nmw(p))
|
|
236
|
4389 {
|
|
455
|
4390 c = PTR2CHAR(p);
|
|
324
|
4391 if (!SPELL_ISUPPER(c))
|
|
236
|
4392 {
|
|
|
4393 /* UUl -> KEEPCAP */
|
|
|
4394 if (past_second && allcap)
|
|
300
|
4395 return WF_KEEPCAP;
|
|
236
|
4396 allcap = FALSE;
|
|
|
4397 }
|
|
|
4398 else if (!allcap)
|
|
|
4399 /* UlU -> KEEPCAP */
|
|
300
|
4400 return WF_KEEPCAP;
|
|
236
|
4401 past_second = TRUE;
|
|
|
4402 }
|
|
|
4403
|
|
|
4404 if (allcap)
|
|
300
|
4405 return WF_ALLCAP;
|
|
236
|
4406 if (firstcap)
|
|
300
|
4407 return WF_ONECAP;
|
|
236
|
4408 return 0;
|
|
|
4409 }
|
|
|
4410
|
|
474
|
4411 /*
|
|
|
4412 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
4413 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
4414 * Add ALLCAP for "WOrD".
|
|
|
4415 */
|
|
|
4416 static int
|
|
|
4417 badword_captype(word, end)
|
|
|
4418 char_u *word;
|
|
|
4419 char_u *end;
|
|
|
4420 {
|
|
|
4421 int flags = captype(word, end);
|
|
475
|
4422 int c;
|
|
474
|
4423 int l, u;
|
|
|
4424 int first;
|
|
|
4425 char_u *p;
|
|
|
4426
|
|
|
4427 if (flags & WF_KEEPCAP)
|
|
|
4428 {
|
|
|
4429 /* Count the number of UPPER and lower case letters. */
|
|
|
4430 l = u = 0;
|
|
|
4431 first = FALSE;
|
|
|
4432 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
4433 {
|
|
475
|
4434 c = PTR2CHAR(p);
|
|
|
4435 if (SPELL_ISUPPER(c))
|
|
474
|
4436 {
|
|
|
4437 ++u;
|
|
|
4438 if (p == word)
|
|
|
4439 first = TRUE;
|
|
|
4440 }
|
|
|
4441 else
|
|
|
4442 ++l;
|
|
|
4443 }
|
|
|
4444
|
|
|
4445 /* If there are more UPPER than lower case letters suggest an
|
|
|
4446 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
4447 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
4448 * require three upper case letters. */
|
|
|
4449 if (u > l && u > 2)
|
|
|
4450 flags |= WF_ALLCAP;
|
|
|
4451 else if (first)
|
|
|
4452 flags |= WF_ONECAP;
|
|
633
|
4453
|
|
|
4454 if (u >= 2 && l >= 2) /* maCARONI maCAroni */
|
|
|
4455 flags |= WF_MIXCAP;
|
|
474
|
4456 }
|
|
|
4457 return flags;
|
|
|
4458 }
|
|
|
4459
|
|
355
|
4460 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
4461 /*
|
|
|
4462 * Free all languages.
|
|
|
4463 */
|
|
|
4464 void
|
|
|
4465 spell_free_all()
|
|
|
4466 {
|
|
503
|
4467 slang_T *slang;
|
|
355
|
4468 buf_T *buf;
|
|
385
|
4469 char_u fname[MAXPATHL];
|
|
355
|
4470
|
|
|
4471 /* Go through all buffers and handle 'spelllang'. */
|
|
|
4472 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
4473 ga_clear(&buf->b_langp);
|
|
|
4474
|
|
|
4475 while (first_lang != NULL)
|
|
|
4476 {
|
|
503
|
4477 slang = first_lang;
|
|
|
4478 first_lang = slang->sl_next;
|
|
|
4479 slang_free(slang);
|
|
355
|
4480 }
|
|
366
|
4481
|
|
385
|
4482 if (int_wordlist != NULL)
|
|
|
4483 {
|
|
|
4484 /* Delete the internal wordlist and its .spl file */
|
|
|
4485 mch_remove(int_wordlist);
|
|
|
4486 int_wordlist_spl(fname);
|
|
|
4487 mch_remove(fname);
|
|
|
4488 vim_free(int_wordlist);
|
|
|
4489 int_wordlist = NULL;
|
|
381
|
4490 }
|
|
|
4491
|
|
366
|
4492 init_spell_chartab();
|
|
626
|
4493
|
|
|
4494 vim_free(repl_to);
|
|
|
4495 repl_to = NULL;
|
|
|
4496 vim_free(repl_from);
|
|
|
4497 repl_from = NULL;
|
|
355
|
4498 }
|
|
|
4499 # endif
|
|
|
4500
|
|
236
|
4501 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
4502 /*
|
|
|
4503 * Clear all spelling tables and reload them.
|
|
307
|
4504 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
4505 */
|
|
|
4506 void
|
|
|
4507 spell_reload()
|
|
|
4508 {
|
|
|
4509 buf_T *buf;
|
|
316
|
4510 win_T *wp;
|
|
236
|
4511
|
|
358
|
4512 /* Initialize the table for spell_iswordp(). */
|
|
236
|
4513 init_spell_chartab();
|
|
|
4514
|
|
|
4515 /* Unload all allocated memory. */
|
|
355
|
4516 spell_free_all();
|
|
236
|
4517
|
|
|
4518 /* Go through all buffers and handle 'spelllang'. */
|
|
|
4519 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
4520 {
|
|
316
|
4521 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
4522 * window for this buffer in which 'spell' is set. */
|
|
236
|
4523 if (*buf->b_p_spl != NUL)
|
|
316
|
4524 {
|
|
|
4525 FOR_ALL_WINDOWS(wp)
|
|
|
4526 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
4527 {
|
|
|
4528 (void)did_set_spelllang(buf);
|
|
|
4529 # ifdef FEAT_WINDOWS
|
|
|
4530 break;
|
|
|
4531 # endif
|
|
|
4532 }
|
|
|
4533 }
|
|
236
|
4534 }
|
|
|
4535 }
|
|
|
4536 # endif
|
|
|
4537
|
|
310
|
4538 /*
|
|
|
4539 * Reload the spell file "fname" if it's loaded.
|
|
|
4540 */
|
|
|
4541 static void
|
|
323
|
4542 spell_reload_one(fname, added_word)
|
|
310
|
4543 char_u *fname;
|
|
323
|
4544 int added_word; /* invoked through "zg" */
|
|
310
|
4545 {
|
|
503
|
4546 slang_T *slang;
|
|
323
|
4547 int didit = FALSE;
|
|
310
|
4548
|
|
503
|
4549 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
4550 {
|
|
|
4551 if (fullpathcmp(fname, slang->sl_fname, FALSE) == FPC_SAME)
|
|
|
4552 {
|
|
|
4553 slang_clear(slang);
|
|
|
4554 if (spell_load_file(fname, NULL, slang, FALSE) == NULL)
|
|
500
|
4555 /* reloading failed, clear the language */
|
|
503
|
4556 slang_clear(slang);
|
|
310
|
4557 redraw_all_later(NOT_VALID);
|
|
323
|
4558 didit = TRUE;
|
|
310
|
4559 }
|
|
500
|
4560 }
|
|
323
|
4561
|
|
|
4562 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
626
|
4563 * 'spelllang' to load it now. */
|
|
323
|
4564 if (added_word && !didit)
|
|
|
4565 did_set_spelllang(curbuf);
|
|
310
|
4566 }
|
|
|
4567
|
|
|
4568
|
|
236
|
4569 /*
|
|
|
4570 * Functions for ":mkspell".
|
|
|
4571 */
|
|
|
4572
|
|
300
|
4573 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
4574 and .dic file. */
|
|
|
4575 /*
|
|
|
4576 * Main structure to store the contents of a ".aff" file.
|
|
|
4577 */
|
|
|
4578 typedef struct afffile_S
|
|
|
4579 {
|
|
|
4580 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
498
|
4581 int af_flagtype; /* AFT_CHAR, AFT_LONG, AFT_NUM or AFT_CAPLONG */
|
|
621
|
4582 unsigned af_rare; /* RARE ID for rare word */
|
|
|
4583 unsigned af_keepcase; /* KEEPCASE ID for keep-case word */
|
|
495
|
4584 unsigned af_bad; /* BAD ID for banned word */
|
|
|
4585 unsigned af_needaffix; /* NEEDAFFIX ID */
|
|
500
|
4586 unsigned af_needcomp; /* NEEDCOMPOUND ID */
|
|
698
|
4587 unsigned af_nosuggest; /* NOSUGGEST ID */
|
|
339
|
4588 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
4589 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
4590 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
495
|
4591 hashtab_T af_comp; /* hashtable for compound flags, compitem_T */
|
|
236
|
4592 } afffile_T;
|
|
|
4593
|
|
495
|
4594 #define AFT_CHAR 0 /* flags are one character */
|
|
498
|
4595 #define AFT_LONG 1 /* flags are two characters */
|
|
|
4596 #define AFT_CAPLONG 2 /* flags are one or two characters */
|
|
|
4597 #define AFT_NUM 3 /* flags are numbers, comma separated */
|
|
495
|
4598
|
|
236
|
4599 typedef struct affentry_S affentry_T;
|
|
|
4600 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
4601 struct affentry_S
|
|
|
4602 {
|
|
|
4603 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
4604 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
4605 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
4606 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
4607 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
483
|
4608 char_u ae_rare; /* rare affix */
|
|
|
4609 char_u ae_nocomp; /* word with affix not compoundable */
|
|
300
|
4610 };
|
|
|
4611
|
|
495
|
4612 #ifdef FEAT_MBYTE
|
|
|
4613 # define AH_KEY_LEN 17 /* 2 x 8 bytes + NUL */
|
|
|
4614 #else
|
|
498
|
4615 # define AH_KEY_LEN 7 /* 6 digits + NUL */
|
|
495
|
4616 #endif
|
|
455
|
4617
|
|
300
|
4618 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
4619 typedef struct affheader_S
|
|
|
4620 {
|
|
495
|
4621 char_u ah_key[AH_KEY_LEN]; /* key for hashtab == name of affix */
|
|
|
4622 unsigned ah_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4623 int ah_newID; /* prefix ID after renumbering; 0 if not used */
|
|
300
|
4624 int ah_combine; /* suffix may combine with prefix */
|
|
498
|
4625 int ah_follows; /* another affix block should be following */
|
|
300
|
4626 affentry_T *ah_first; /* first affix entry */
|
|
|
4627 } affheader_T;
|
|
|
4628
|
|
|
4629 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
4630
|
|
495
|
4631 /* Flag used in compound items. */
|
|
|
4632 typedef struct compitem_S
|
|
|
4633 {
|
|
|
4634 char_u ci_key[AH_KEY_LEN]; /* key for hashtab == name of compound */
|
|
|
4635 unsigned ci_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4636 int ci_newID; /* affix ID after renumbering. */
|
|
|
4637 } compitem_T;
|
|
|
4638
|
|
|
4639 #define HI2CI(hi) ((compitem_T *)(hi)->hi_key)
|
|
|
4640
|
|
300
|
4641 /*
|
|
|
4642 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
4643 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
4644 * once after ":mkspell" is done.
|
|
300
|
4645 */
|
|
|
4646 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
4647 typedef struct sblock_S sblock_T;
|
|
|
4648 struct sblock_S
|
|
|
4649 {
|
|
|
4650 sblock_T *sb_next; /* next block in list */
|
|
|
4651 int sb_used; /* nr of bytes already in use */
|
|
|
4652 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
4653 };
|
|
|
4654
|
|
|
4655 /*
|
|
300
|
4656 * A node in the tree.
|
|
236
|
4657 */
|
|
300
|
4658 typedef struct wordnode_S wordnode_T;
|
|
|
4659 struct wordnode_S
|
|
236
|
4660 {
|
|
346
|
4661 union /* shared to save space */
|
|
|
4662 {
|
|
469
|
4663 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
4664 int index; /* index in written nodes (valid after first
|
|
|
4665 round) */
|
|
|
4666 } wn_u1;
|
|
|
4667 union /* shared to save space */
|
|
|
4668 {
|
|
|
4669 wordnode_T *next; /* next node with same hash key */
|
|
|
4670 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
4671 } wn_u2;
|
|
300
|
4672 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
4673 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
4674 always sorted) */
|
|
474
|
4675 int wn_refs; /* Nr. of references to this node. Only
|
|
|
4676 relevant for first node in a list of
|
|
|
4677 siblings, in following siblings it is
|
|
|
4678 always one. */
|
|
300
|
4679 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
625
|
4680
|
|
|
4681 /* Info for when "wn_byte" is NUL.
|
|
|
4682 * In PREFIXTREE "wn_region" is used for the prefcondnr.
|
|
|
4683 * In the soundfolded word tree "wn_flags" has the MSW of the wordnr and
|
|
|
4684 * "wn_region" the LSW of the wordnr. */
|
|
|
4685 char_u wn_affixID; /* supported/required prefix ID or 0 */
|
|
|
4686 short_u wn_flags; /* WF_ flags */
|
|
|
4687 short wn_region; /* region mask */
|
|
|
4688
|
|
469
|
4689 #ifdef SPELL_PRINTTREE
|
|
|
4690 int wn_nr; /* sequence nr for printing */
|
|
|
4691 #endif
|
|
236
|
4692 };
|
|
|
4693
|
|
390
|
4694 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
4695
|
|
300
|
4696 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
4697
|
|
300
|
4698 /*
|
|
|
4699 * Info used while reading the spell files.
|
|
|
4700 */
|
|
|
4701 typedef struct spellinfo_S
|
|
249
|
4702 {
|
|
300
|
4703 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
4704 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
4705
|
|
300
|
4706 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
4707 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
4708
|
|
339
|
4709 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
4710
|
|
625
|
4711 long si_sugtree; /* creating the soundfolding trie */
|
|
|
4712
|
|
300
|
4713 sblock_T *si_blocks; /* memory blocks used */
|
|
482
|
4714 long si_blocks_cnt; /* memory blocks allocated */
|
|
|
4715 long si_compress_cnt; /* words to add before lowering
|
|
|
4716 compression limit */
|
|
474
|
4717 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
4718 compression, linked by "wn_child" field. */
|
|
482
|
4719 long si_free_count; /* number of nodes in si_first_free */
|
|
474
|
4720 #ifdef SPELL_PRINTTREE
|
|
|
4721 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
4722 #endif
|
|
625
|
4723 buf_T *si_spellbuf; /* buffer used to store soundfold word table */
|
|
474
|
4724
|
|
300
|
4725 int si_ascii; /* handling only ASCII words */
|
|
310
|
4726 int si_add; /* addition file */
|
|
351
|
4727 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
4728 int si_region; /* region mask */
|
|
|
4729 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
4730 int si_memtot; /* runtime memory used */
|
|
310
|
4731 int si_verbose; /* verbose messages */
|
|
474
|
4732 int si_msg_count; /* number of words added since last message */
|
|
714
|
4733 char_u *si_info; /* info text chars or NULL */
|
|
316
|
4734 int si_region_count; /* number of regions supported (1 when there
|
|
|
4735 are no regions) */
|
|
483
|
4736 char_u si_region_name[16]; /* region names; used only if
|
|
|
4737 * si_region_count > 1) */
|
|
323
|
4738
|
|
|
4739 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
625
|
4740 garray_T si_repsal; /* list of fromto_T entries from REPSAL lines */
|
|
323
|
4741 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
4742 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
4743 char_u *si_sofoto; /* SOFOTO text */
|
|
625
|
4744 int si_nosugfile; /* NOSUGFILE item found */
|
|
698
|
4745 int si_nosplitsugs; /* NOSPLITSUGS item found */
|
|
323
|
4746 int si_followup; /* soundsalike: ? */
|
|
|
4747 int si_collapse; /* soundsalike: ? */
|
|
625
|
4748 hashtab_T si_commonwords; /* hashtable for common words */
|
|
|
4749 time_t si_sugtime; /* timestamp for .sug file */
|
|
323
|
4750 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
4751 garray_T si_map; /* MAP info concatenated */
|
|
495
|
4752 char_u *si_midword; /* MIDWORD chars or NULL */
|
|
483
|
4753 int si_compmax; /* max nr of words for compounding */
|
|
481
|
4754 int si_compminlen; /* minimal length for compounding */
|
|
483
|
4755 int si_compsylmax; /* max nr of syllables for compounding */
|
|
481
|
4756 char_u *si_compflags; /* flags used for compounding */
|
|
492
|
4757 char_u si_nobreak; /* NOBREAK */
|
|
483
|
4758 char_u *si_syllable; /* syllable string */
|
|
339
|
4759 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
4760 * prefixes, each stored as a string */
|
|
495
|
4761 int si_newprefID; /* current value for ah_newID */
|
|
500
|
4762 int si_newcompID; /* current value for compound ID */
|
|
300
|
4763 } spellinfo_T;
|
|
249
|
4764
|
|
474
|
4765 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
714
|
4766 static int spell_info_item __ARGS((char_u *s));
|
|
495
|
4767 static unsigned affitem2flag __ARGS((int flagtype, char_u *item, char_u *fname, int lnum));
|
|
|
4768 static unsigned get_affitem __ARGS((int flagtype, char_u **pp));
|
|
|
4769 static void process_compflags __ARGS((spellinfo_T *spin, afffile_T *aff, char_u *compflags));
|
|
500
|
4770 static void check_renumber __ARGS((spellinfo_T *spin));
|
|
495
|
4771 static int flag_in_afflist __ARGS((int flagtype, char_u *afflist, unsigned flag));
|
|
|
4772 static void aff_check_number __ARGS((int spinval, int affval, char *name));
|
|
|
4773 static void aff_check_string __ARGS((char_u *spinval, char_u *affval, char *name));
|
|
339
|
4774 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
4775 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
4776 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
4777 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
4778 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
4779 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
483
|
4780 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
495
|
4781 static void get_compflags __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
483
|
4782 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
|
4783 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
4784 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
4785 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
4786 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
4787 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
483
|
4788 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix));
|
|
481
|
4789 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
4790 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
625
|
4791 static int deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
474
|
4792 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
4793 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
4794 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
4795 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
625
|
4796 static void put_sugtime __ARGS((spellinfo_T *spin, FILE *fd));
|
|
500
|
4797 static int write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
4798 static void clear_node __ARGS((wordnode_T *node));
|
|
|
4799 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
625
|
4800 static void spell_make_sugfile __ARGS((spellinfo_T *spin, char_u *wfname));
|
|
|
4801 static int sug_filltree __ARGS((spellinfo_T *spin, slang_T *slang));
|
|
|
4802 static int sug_maketable __ARGS((spellinfo_T *spin));
|
|
|
4803 static int sug_filltable __ARGS((spellinfo_T *spin, wordnode_T *node, int startwordnr, garray_T *gap));
|
|
|
4804 static int offset2bytes __ARGS((int nr, char_u *buf));
|
|
|
4805 static int bytes2offset __ARGS((char_u **pp));
|
|
|
4806 static void sug_write __ARGS((spellinfo_T *spin, char_u *fname));
|
|
323
|
4807 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
625
|
4808 static void spell_message __ARGS((spellinfo_T *spin, char_u *str));
|
|
310
|
4809 static void init_spellfile __ARGS((void));
|
|
236
|
4810
|
|
455
|
4811 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
4812 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
4813 * Use a negative number with the lower 8 bits zero. */
|
|
|
4814 #define PFX_FLAGS -256
|
|
390
|
4815
|
|
483
|
4816 /*
|
|
|
4817 * Tunable parameters for when the tree is compressed. See 'mkspellmem'.
|
|
|
4818 */
|
|
|
4819 static long compress_start = 30000; /* memory / SBLOCKSIZE */
|
|
|
4820 static long compress_inc = 100; /* memory / SBLOCKSIZE */
|
|
|
4821 static long compress_added = 500000; /* word count */
|
|
|
4822
|
|
469
|
4823 #ifdef SPELL_PRINTTREE
|
|
|
4824 /*
|
|
|
4825 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
4826 * readable format, so that we can see what happens when adding a word and/or
|
|
|
4827 * compressing the tree.
|
|
|
4828 * Based on code from Olaf Seibert.
|
|
|
4829 */
|
|
|
4830 #define PRINTLINESIZE 1000
|
|
|
4831 #define PRINTWIDTH 6
|
|
|
4832
|
|
|
4833 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
4834 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
4835
|
|
|
4836 static char line1[PRINTLINESIZE];
|
|
|
4837 static char line2[PRINTLINESIZE];
|
|
|
4838 static char line3[PRINTLINESIZE];
|
|
|
4839
|
|
|
4840 static void
|
|
|
4841 spell_clear_flags(wordnode_T *node)
|
|
|
4842 {
|
|
|
4843 wordnode_T *np;
|
|
|
4844
|
|
|
4845 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4846 {
|
|
|
4847 np->wn_u1.index = FALSE;
|
|
|
4848 spell_clear_flags(np->wn_child);
|
|
|
4849 }
|
|
|
4850 }
|
|
|
4851
|
|
|
4852 static void
|
|
|
4853 spell_print_node(wordnode_T *node, int depth)
|
|
|
4854 {
|
|
|
4855 if (node->wn_u1.index)
|
|
|
4856 {
|
|
|
4857 /* Done this node before, print the reference. */
|
|
|
4858 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
4859 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
4860 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4861 msg(line1);
|
|
|
4862 msg(line2);
|
|
|
4863 msg(line3);
|
|
|
4864 }
|
|
|
4865 else
|
|
|
4866 {
|
|
|
4867 node->wn_u1.index = TRUE;
|
|
|
4868
|
|
|
4869 if (node->wn_byte != NUL)
|
|
|
4870 {
|
|
|
4871 if (node->wn_child != NULL)
|
|
|
4872 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
4873 else
|
|
|
4874 /* Cannot happen? */
|
|
|
4875 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
4876 }
|
|
|
4877 else
|
|
|
4878 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
4879
|
|
|
4880 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
4881
|
|
|
4882 if (node->wn_sibling != NULL)
|
|
|
4883 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
4884 else
|
|
|
4885 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4886
|
|
|
4887 if (node->wn_byte == NUL)
|
|
|
4888 {
|
|
|
4889 msg(line1);
|
|
|
4890 msg(line2);
|
|
|
4891 msg(line3);
|
|
|
4892 }
|
|
|
4893
|
|
|
4894 /* do the children */
|
|
|
4895 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
4896 spell_print_node(node->wn_child, depth + 1);
|
|
|
4897
|
|
|
4898 /* do the siblings */
|
|
|
4899 if (node->wn_sibling != NULL)
|
|
|
4900 {
|
|
|
4901 /* get rid of all parent details except | */
|
|
|
4902 STRCPY(line1, line3);
|
|
|
4903 STRCPY(line2, line3);
|
|
|
4904 spell_print_node(node->wn_sibling, depth);
|
|
|
4905 }
|
|
|
4906 }
|
|
|
4907 }
|
|
|
4908
|
|
|
4909 static void
|
|
|
4910 spell_print_tree(wordnode_T *root)
|
|
|
4911 {
|
|
|
4912 if (root != NULL)
|
|
|
4913 {
|
|
|
4914 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
4915 * done. */
|
|
|
4916 spell_clear_flags(root);
|
|
|
4917
|
|
|
4918 /* Recursively print the tree. */
|
|
|
4919 spell_print_node(root, 0);
|
|
|
4920 }
|
|
|
4921 }
|
|
|
4922 #endif /* SPELL_PRINTTREE */
|
|
|
4923
|
|
236
|
4924 /*
|
|
323
|
4925 * Read the affix file "fname".
|
|
316
|
4926 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
4927 */
|
|
|
4928 static afffile_T *
|
|
474
|
4929 spell_read_aff(spin, fname)
|
|
|
4930 spellinfo_T *spin;
|
|
236
|
4931 char_u *fname;
|
|
|
4932 {
|
|
|
4933 FILE *fd;
|
|
|
4934 afffile_T *aff;
|
|
|
4935 char_u rline[MAXLINELEN];
|
|
|
4936 char_u *line;
|
|
|
4937 char_u *pc = NULL;
|
|
625
|
4938 #define MAXITEMCNT 30
|
|
334
|
4939 char_u *(items[MAXITEMCNT]);
|
|
236
|
4940 int itemcnt;
|
|
|
4941 char_u *p;
|
|
|
4942 int lnum = 0;
|
|
|
4943 affheader_T *cur_aff = NULL;
|
|
495
|
4944 int did_postpone_prefix = FALSE;
|
|
236
|
4945 int aff_todo = 0;
|
|
|
4946 hashtab_T *tp;
|
|
255
|
4947 char_u *low = NULL;
|
|
|
4948 char_u *fol = NULL;
|
|
|
4949 char_u *upp = NULL;
|
|
323
|
4950 int do_rep;
|
|
625
|
4951 int do_repsal;
|
|
323
|
4952 int do_sal;
|
|
|
4953 int do_map;
|
|
|
4954 int found_map = FALSE;
|
|
324
|
4955 hashitem_T *hi;
|
|
455
|
4956 int l;
|
|
495
|
4957 int compminlen = 0; /* COMPOUNDMIN value */
|
|
|
4958 int compsylmax = 0; /* COMPOUNDSYLMAX value */
|
|
|
4959 int compmax = 0; /* COMPOUNDMAX value */
|
|
714
|
4960 char_u *compflags = NULL; /* COMPOUNDFLAG and COMPOUNDRULE
|
|
495
|
4961 concatenated */
|
|
|
4962 char_u *midword = NULL; /* MIDWORD value */
|
|
|
4963 char_u *syllable = NULL; /* SYLLABLE value */
|
|
|
4964 char_u *sofofrom = NULL; /* SOFOFROM value */
|
|
|
4965 char_u *sofoto = NULL; /* SOFOTO value */
|
|
236
|
4966
|
|
300
|
4967 /*
|
|
|
4968 * Open the file.
|
|
|
4969 */
|
|
310
|
4970 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4971 if (fd == NULL)
|
|
|
4972 {
|
|
|
4973 EMSG2(_(e_notopen), fname);
|
|
|
4974 return NULL;
|
|
|
4975 }
|
|
|
4976
|
|
625
|
4977 vim_snprintf((char *)IObuff, IOSIZE, _("Reading affix file %s ..."), fname);
|
|
|
4978 spell_message(spin, IObuff);
|
|
236
|
4979
|
|
323
|
4980 /* Only do REP lines when not done in another .aff file already. */
|
|
|
4981 do_rep = spin->si_rep.ga_len == 0;
|
|
|
4982
|
|
625
|
4983 /* Only do REPSAL lines when not done in another .aff file already. */
|
|
|
4984 do_repsal = spin->si_repsal.ga_len == 0;
|
|
|
4985
|
|
323
|
4986 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
4987 do_sal = spin->si_sal.ga_len == 0;
|
|
|
4988
|
|
|
4989 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
4990 do_map = spin->si_map.ga_len == 0;
|
|
|
4991
|
|
300
|
4992 /*
|
|
|
4993 * Allocate and init the afffile_T structure.
|
|
|
4994 */
|
|
474
|
4995 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
4996 if (aff == NULL)
|
|
|
4997 return NULL;
|
|
|
4998 hash_init(&aff->af_pref);
|
|
|
4999 hash_init(&aff->af_suff);
|
|
495
|
5000 hash_init(&aff->af_comp);
|
|
236
|
5001
|
|
|
5002 /*
|
|
|
5003 * Read all the lines in the file one by one.
|
|
|
5004 */
|
|
255
|
5005 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
5006 {
|
|
255
|
5007 line_breakcheck();
|
|
236
|
5008 ++lnum;
|
|
|
5009
|
|
|
5010 /* Skip comment lines. */
|
|
|
5011 if (*rline == '#')
|
|
|
5012 continue;
|
|
|
5013
|
|
|
5014 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
5015 vim_free(pc);
|
|
310
|
5016 #ifdef FEAT_MBYTE
|
|
300
|
5017 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
5018 {
|
|
300
|
5019 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
5020 if (pc == NULL)
|
|
|
5021 {
|
|
|
5022 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5023 fname, lnum, rline);
|
|
|
5024 continue;
|
|
|
5025 }
|
|
236
|
5026 line = pc;
|
|
|
5027 }
|
|
|
5028 else
|
|
310
|
5029 #endif
|
|
236
|
5030 {
|
|
|
5031 pc = NULL;
|
|
|
5032 line = rline;
|
|
|
5033 }
|
|
|
5034
|
|
|
5035 /* Split the line up in white separated items. Put a NUL after each
|
|
|
5036 * item. */
|
|
|
5037 itemcnt = 0;
|
|
|
5038 for (p = line; ; )
|
|
|
5039 {
|
|
|
5040 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
5041 ++p;
|
|
|
5042 if (*p == NUL)
|
|
|
5043 break;
|
|
334
|
5044 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
5045 break;
|
|
236
|
5046 items[itemcnt++] = p;
|
|
714
|
5047 /* A few items have arbitrary text argument, don't split them. */
|
|
|
5048 if (itemcnt == 2 && spell_info_item(items[0]))
|
|
|
5049 while (*p >= ' ' || *p == TAB) /* skip until CR/NL */
|
|
|
5050 ++p;
|
|
|
5051 else
|
|
|
5052 while (*p > ' ') /* skip until white space or CR/NL */
|
|
|
5053 ++p;
|
|
236
|
5054 if (*p == NUL)
|
|
|
5055 break;
|
|
|
5056 *p++ = NUL;
|
|
|
5057 }
|
|
|
5058
|
|
|
5059 /* Handle non-empty lines. */
|
|
|
5060 if (itemcnt > 0)
|
|
|
5061 {
|
|
|
5062 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
5063 && aff->af_enc == NULL)
|
|
|
5064 {
|
|
310
|
5065 #ifdef FEAT_MBYTE
|
|
300
|
5066 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
5067 aff->af_enc = enc_canonize(items[1]);
|
|
|
5068 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
5069 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
5070 p_enc) == FAIL)
|
|
|
5071 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
5072 fname, aff->af_enc, p_enc);
|
|
375
|
5073 spin->si_conv.vc_fail = TRUE;
|
|
310
|
5074 #else
|
|
|
5075 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
5076 #endif
|
|
236
|
5077 }
|
|
495
|
5078 else if (STRCMP(items[0], "FLAG") == 0 && itemcnt == 2
|
|
|
5079 && aff->af_flagtype == AFT_CHAR)
|
|
|
5080 {
|
|
|
5081 if (STRCMP(items[1], "long") == 0)
|
|
498
|
5082 aff->af_flagtype = AFT_LONG;
|
|
495
|
5083 else if (STRCMP(items[1], "num") == 0)
|
|
498
|
5084 aff->af_flagtype = AFT_NUM;
|
|
|
5085 else if (STRCMP(items[1], "caplong") == 0)
|
|
|
5086 aff->af_flagtype = AFT_CAPLONG;
|
|
495
|
5087 else
|
|
|
5088 smsg((char_u *)_("Invalid value for FLAG in %s line %d: %s"),
|
|
|
5089 fname, lnum, items[1]);
|
|
621
|
5090 if (aff->af_rare != 0
|
|
|
5091 || aff->af_keepcase != 0
|
|
|
5092 || aff->af_bad != 0
|
|
500
|
5093 || aff->af_needaffix != 0
|
|
|
5094 || aff->af_needcomp != 0
|
|
698
|
5095 || aff->af_nosuggest != 0
|
|
500
|
5096 || compflags != NULL
|
|
495
|
5097 || aff->af_suff.ht_used > 0
|
|
|
5098 || aff->af_pref.ht_used > 0)
|
|
|
5099 smsg((char_u *)_("FLAG after using flags in %s line %d: %s"),
|
|
|
5100 fname, lnum, items[1]);
|
|
|
5101 }
|
|
714
|
5102 else if (spell_info_item(items[0]))
|
|
|
5103 {
|
|
|
5104 p = (char_u *)getroom(spin,
|
|
|
5105 (spin->si_info == NULL ? 0 : STRLEN(spin->si_info))
|
|
|
5106 + STRLEN(items[0])
|
|
|
5107 + STRLEN(items[1]) + 3, FALSE);
|
|
|
5108 if (p != NULL)
|
|
|
5109 {
|
|
|
5110 if (spin->si_info != NULL)
|
|
|
5111 {
|
|
|
5112 STRCPY(p, spin->si_info);
|
|
|
5113 STRCAT(p, "\n");
|
|
|
5114 }
|
|
|
5115 STRCAT(p, items[0]);
|
|
|
5116 STRCAT(p, " ");
|
|
|
5117 STRCAT(p, items[1]);
|
|
|
5118 spin->si_info = p;
|
|
|
5119 }
|
|
|
5120 }
|
|
495
|
5121 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2
|
|
|
5122 && midword == NULL)
|
|
|
5123 {
|
|
|
5124 midword = getroom_save(spin, items[1]);
|
|
366
|
5125 }
|
|
323
|
5126 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
5127 {
|
|
323
|
5128 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
5129 }
|
|
621
|
5130 /* TODO: remove "RAR" later */
|
|
|
5131 else if ((STRCMP(items[0], "RAR") == 0
|
|
|
5132 || STRCMP(items[0], "RARE") == 0) && itemcnt == 2
|
|
|
5133 && aff->af_rare == 0)
|
|
|
5134 {
|
|
|
5135 aff->af_rare = affitem2flag(aff->af_flagtype, items[1],
|
|
495
|
5136 fname, lnum);
|
|
307
|
5137 }
|
|
621
|
5138 /* TODO: remove "KEP" later */
|
|
|
5139 else if ((STRCMP(items[0], "KEP") == 0
|
|
|
5140 || STRCMP(items[0], "KEEPCASE") == 0) && itemcnt == 2
|
|
|
5141 && aff->af_keepcase == 0)
|
|
|
5142 {
|
|
|
5143 aff->af_keepcase = affitem2flag(aff->af_flagtype, items[1],
|
|
495
|
5144 fname, lnum);
|
|
307
|
5145 }
|
|
346
|
5146 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
5147 && aff->af_bad == 0)
|
|
|
5148 {
|
|
495
|
5149 aff->af_bad = affitem2flag(aff->af_flagtype, items[1],
|
|
|
5150 fname, lnum);
|
|
346
|
5151 }
|
|
483
|
5152 else if (STRCMP(items[0], "NEEDAFFIX") == 0 && itemcnt == 2
|
|
|
5153 && aff->af_needaffix == 0)
|
|
|
5154 {
|
|
495
|
5155 aff->af_needaffix = affitem2flag(aff->af_flagtype, items[1],
|
|
|
5156 fname, lnum);
|
|
483
|
5157 }
|
|
698
|
5158 else if (STRCMP(items[0], "NOSUGGEST") == 0 && itemcnt == 2
|
|
|
5159 && aff->af_nosuggest == 0)
|
|
|
5160 {
|
|
|
5161 aff->af_nosuggest = affitem2flag(aff->af_flagtype, items[1],
|
|
|
5162 fname, lnum);
|
|
|
5163 }
|
|
500
|
5164 else if (STRCMP(items[0], "NEEDCOMPOUND") == 0 && itemcnt == 2
|
|
|
5165 && aff->af_needcomp == 0)
|
|
|
5166 {
|
|
|
5167 aff->af_needcomp = affitem2flag(aff->af_flagtype, items[1],
|
|
|
5168 fname, lnum);
|
|
|
5169 }
|
|
481
|
5170 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
495
|
5171 && compflags == NULL)
|
|
|
5172 {
|
|
714
|
5173 /* Turn flag "c" into COMPOUNDRULE compatible string "c+",
|
|
495
|
5174 * "Na" into "Na+", "1234" into "1234+". */
|
|
|
5175 p = getroom(spin, STRLEN(items[1]) + 2, FALSE);
|
|
483
|
5176 if (p != NULL)
|
|
|
5177 {
|
|
495
|
5178 STRCPY(p, items[1]);
|
|
|
5179 STRCAT(p, "+");
|
|
|
5180 compflags = p;
|
|
|
5181 }
|
|
481
|
5182 }
|
|
714
|
5183 else if (STRCMP(items[0], "COMPOUNDRULE") == 0 && itemcnt == 2)
|
|
483
|
5184 {
|
|
|
5185 /* Concatenate this string to previously defined ones, using a
|
|
|
5186 * slash to separate them. */
|
|
|
5187 l = STRLEN(items[1]) + 1;
|
|
495
|
5188 if (compflags != NULL)
|
|
|
5189 l += STRLEN(compflags) + 1;
|
|
483
|
5190 p = getroom(spin, l, FALSE);
|
|
|
5191 if (p != NULL)
|
|
|
5192 {
|
|
495
|
5193 if (compflags != NULL)
|
|
483
|
5194 {
|
|
495
|
5195 STRCPY(p, compflags);
|
|
483
|
5196 STRCAT(p, "/");
|
|
|
5197 }
|
|
|
5198 STRCAT(p, items[1]);
|
|
495
|
5199 compflags = p;
|
|
483
|
5200 }
|
|
|
5201 }
|
|
|
5202 else if (STRCMP(items[0], "COMPOUNDMAX") == 0 && itemcnt == 2
|
|
495
|
5203 && compmax == 0)
|
|
|
5204 {
|
|
|
5205 compmax = atoi((char *)items[1]);
|
|
|
5206 if (compmax == 0)
|
|
483
|
5207 smsg((char_u *)_("Wrong COMPOUNDMAX value in %s line %d: %s"),
|
|
|
5208 fname, lnum, items[1]);
|
|
481
|
5209 }
|
|
|
5210 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
495
|
5211 && compminlen == 0)
|
|
|
5212 {
|
|
|
5213 compminlen = atoi((char *)items[1]);
|
|
|
5214 if (compminlen == 0)
|
|
481
|
5215 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
5216 fname, lnum, items[1]);
|
|
|
5217 }
|
|
483
|
5218 else if (STRCMP(items[0], "COMPOUNDSYLMAX") == 0 && itemcnt == 2
|
|
495
|
5219 && compsylmax == 0)
|
|
|
5220 {
|
|
|
5221 compsylmax = atoi((char *)items[1]);
|
|
|
5222 if (compsylmax == 0)
|
|
483
|
5223 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"),
|
|
|
5224 fname, lnum, items[1]);
|
|
|
5225 }
|
|
|
5226 else if (STRCMP(items[0], "SYLLABLE") == 0 && itemcnt == 2
|
|
495
|
5227 && syllable == NULL)
|
|
|
5228 {
|
|
|
5229 syllable = getroom_save(spin, items[1]);
|
|
483
|
5230 }
|
|
492
|
5231 else if (STRCMP(items[0], "NOBREAK") == 0 && itemcnt == 1)
|
|
|
5232 {
|
|
|
5233 spin->si_nobreak = TRUE;
|
|
|
5234 }
|
|
698
|
5235 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
5236 {
|
|
|
5237 spin->si_nosplitsugs = TRUE;
|
|
|
5238 }
|
|
625
|
5239 else if (STRCMP(items[0], "NOSUGFILE") == 0 && itemcnt == 1)
|
|
|
5240 {
|
|
|
5241 spin->si_nosugfile = TRUE;
|
|
|
5242 }
|
|
339
|
5243 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
5244 {
|
|
|
5245 aff->af_pfxpostpone = TRUE;
|
|
|
5246 }
|
|
236
|
5247 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
5248 || STRCMP(items[0], "SFX") == 0)
|
|
|
5249 && aff_todo == 0
|
|
334
|
5250 && itemcnt >= 4)
|
|
236
|
5251 {
|
|
498
|
5252 int lasti = 4;
|
|
|
5253 char_u key[AH_KEY_LEN];
|
|
|
5254
|
|
|
5255 if (*items[0] == 'P')
|
|
|
5256 tp = &aff->af_pref;
|
|
|
5257 else
|
|
|
5258 tp = &aff->af_suff;
|
|
|
5259
|
|
|
5260 /* Myspell allows the same affix name to be used multiple
|
|
|
5261 * times. The affix files that do this have an undocumented
|
|
|
5262 * "S" flag on all but the last block, thus we check for that
|
|
|
5263 * and store it in ah_follows. */
|
|
|
5264 vim_strncpy(key, items[1], AH_KEY_LEN - 1);
|
|
|
5265 hi = hash_find(tp, key);
|
|
|
5266 if (!HASHITEM_EMPTY(hi))
|
|
|
5267 {
|
|
|
5268 cur_aff = HI2AH(hi);
|
|
|
5269 if (cur_aff->ah_combine != (*items[2] == 'Y'))
|
|
|
5270 smsg((char_u *)_("Different combining flag in continued affix block in %s line %d: %s"),
|
|
|
5271 fname, lnum, items[1]);
|
|
|
5272 if (!cur_aff->ah_follows)
|
|
|
5273 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
5274 fname, lnum, items[1]);
|
|
|
5275 }
|
|
|
5276 else
|
|
|
5277 {
|
|
|
5278 /* New affix letter. */
|
|
|
5279 cur_aff = (affheader_T *)getroom(spin,
|
|
|
5280 sizeof(affheader_T), TRUE);
|
|
|
5281 if (cur_aff == NULL)
|
|
|
5282 break;
|
|
|
5283 cur_aff->ah_flag = affitem2flag(aff->af_flagtype, items[1],
|
|
|
5284 fname, lnum);
|
|
|
5285 if (cur_aff->ah_flag == 0 || STRLEN(items[1]) >= AH_KEY_LEN)
|
|
|
5286 break;
|
|
|
5287 if (cur_aff->ah_flag == aff->af_bad
|
|
621
|
5288 || cur_aff->ah_flag == aff->af_rare
|
|
|
5289 || cur_aff->ah_flag == aff->af_keepcase
|
|
500
|
5290 || cur_aff->ah_flag == aff->af_needaffix
|
|
698
|
5291 || cur_aff->ah_flag == aff->af_nosuggest
|
|
500
|
5292 || cur_aff->ah_flag == aff->af_needcomp)
|
|
698
|
5293 smsg((char_u *)_("Affix also used for BAD/RARE/KEEPCASE/NEEDAFFIX/NEEDCOMPOUND/NOSUGGEST in %s line %d: %s"),
|
|
498
|
5294 fname, lnum, items[1]);
|
|
|
5295 STRCPY(cur_aff->ah_key, items[1]);
|
|
|
5296 hash_add(tp, cur_aff->ah_key);
|
|
|
5297
|
|
|
5298 cur_aff->ah_combine = (*items[2] == 'Y');
|
|
|
5299 }
|
|
|
5300
|
|
|
5301 /* Check for the "S" flag, which apparently means that another
|
|
|
5302 * block with the same affix name is following. */
|
|
|
5303 if (itemcnt > lasti && STRCMP(items[lasti], "S") == 0)
|
|
|
5304 {
|
|
|
5305 ++lasti;
|
|
|
5306 cur_aff->ah_follows = TRUE;
|
|
|
5307 }
|
|
|
5308 else
|
|
|
5309 cur_aff->ah_follows = FALSE;
|
|
|
5310
|
|
334
|
5311 /* Myspell allows extra text after the item, but that might
|
|
|
5312 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
498
|
5313 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
5314 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
5315 fname, lnum, items[4]);
|
|
|
5316
|
|
498
|
5317 if (STRCMP(items[2], "Y") != 0 && STRCMP(items[2], "N") != 0)
|
|
236
|
5318 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
5319 fname, lnum, items[2]);
|
|
339
|
5320
|
|
498
|
5321 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
|
5322 {
|
|
|
5323 if (cur_aff->ah_newID == 0)
|
|
495
|
5324 {
|
|
|
5325 /* Use a new number in the .spl file later, to be able
|
|
|
5326 * to handle multiple .aff files. */
|
|
500
|
5327 check_renumber(spin);
|
|
495
|
5328 cur_aff->ah_newID = ++spin->si_newprefID;
|
|
|
5329
|
|
|
5330 /* We only really use ah_newID if the prefix is
|
|
|
5331 * postponed. We know that only after handling all
|
|
|
5332 * the items. */
|
|
|
5333 did_postpone_prefix = FALSE;
|
|
|
5334 }
|
|
498
|
5335 else
|
|
|
5336 /* Did use the ID in a previous block. */
|
|
|
5337 did_postpone_prefix = TRUE;
|
|
|
5338 }
|
|
|
5339
|
|
300
|
5340 aff_todo = atoi((char *)items[3]);
|
|
236
|
5341 }
|
|
|
5342 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
5343 || STRCMP(items[0], "SFX") == 0)
|
|
|
5344 && aff_todo > 0
|
|
|
5345 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
5346 && itemcnt >= 5)
|
|
236
|
5347 {
|
|
|
5348 affentry_T *aff_entry;
|
|
366
|
5349 int rare = FALSE;
|
|
483
|
5350 int nocomp = FALSE;
|
|
455
|
5351 int upper = FALSE;
|
|
366
|
5352 int lasti = 5;
|
|
|
5353
|
|
483
|
5354 /* Check for "rare" and "nocomp" after the other info. */
|
|
|
5355 while (itemcnt > lasti)
|
|
|
5356 {
|
|
|
5357 if (!rare && STRICMP(items[lasti], "rare") == 0)
|
|
|
5358 {
|
|
|
5359 rare = TRUE;
|
|
|
5360 ++lasti;
|
|
|
5361 }
|
|
|
5362 else if (!nocomp && STRICMP(items[lasti], "nocomp") == 0)
|
|
|
5363 {
|
|
|
5364 nocomp = TRUE;
|
|
|
5365 ++lasti;
|
|
|
5366 }
|
|
|
5367 else
|
|
|
5368 break;
|
|
366
|
5369 }
|
|
236
|
5370
|
|
334
|
5371 /* Myspell allows extra text after the item, but that might
|
|
|
5372 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
5373 if (itemcnt > lasti && *items[lasti] != '#')
|
|
482
|
5374 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]);
|
|
334
|
5375
|
|
236
|
5376 /* New item for an affix letter. */
|
|
|
5377 --aff_todo;
|
|
474
|
5378 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
5379 sizeof(affentry_T), TRUE);
|
|
236
|
5380 if (aff_entry == NULL)
|
|
|
5381 break;
|
|
366
|
5382 aff_entry->ae_rare = rare;
|
|
483
|
5383 aff_entry->ae_nocomp = nocomp;
|
|
240
|
5384
|
|
236
|
5385 if (STRCMP(items[2], "0") != 0)
|
|
474
|
5386 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
5387 if (STRCMP(items[3], "0") != 0)
|
|
474
|
5388 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
5389
|
|
300
|
5390 /* Don't use an affix entry with non-ASCII characters when
|
|
|
5391 * "spin->si_ascii" is TRUE. */
|
|
|
5392 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
5393 || has_non_ascii(aff_entry->ae_add)))
|
|
|
5394 {
|
|
|
5395 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
5396 cur_aff->ah_first = aff_entry;
|
|
300
|
5397
|
|
|
5398 if (STRCMP(items[4], ".") != 0)
|
|
|
5399 {
|
|
|
5400 char_u buf[MAXLINELEN];
|
|
|
5401
|
|
474
|
5402 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
5403 if (*items[0] == 'P')
|
|
|
5404 sprintf((char *)buf, "^%s", items[4]);
|
|
|
5405 else
|
|
|
5406 sprintf((char *)buf, "%s$", items[4]);
|
|
|
5407 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
5408 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
5409 if (aff_entry->ae_prog == NULL)
|
|
|
5410 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
5411 fname, lnum, items[4]);
|
|
300
|
5412 }
|
|
339
|
5413
|
|
|
5414 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
5415 * for the condition. Use an existing one if possible. */
|
|
455
|
5416 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
5417 {
|
|
455
|
5418 /* When the chop string is one lower-case letter and
|
|
|
5419 * the add string ends in the upper-case letter we set
|
|
|
5420 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
5421 * letters from "ae_add". The condition must either
|
|
|
5422 * be empty or start with the same letter. */
|
|
|
5423 if (aff_entry->ae_chop != NULL
|
|
|
5424 && aff_entry->ae_add != NULL
|
|
|
5425 #ifdef FEAT_MBYTE
|
|
474
|
5426 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
5427 aff_entry->ae_chop)] == NUL
|
|
|
5428 #else
|
|
|
5429 && aff_entry->ae_chop[1] == NUL
|
|
|
5430 #endif
|
|
|
5431 )
|
|
339
|
5432 {
|
|
455
|
5433 int c, c_up;
|
|
|
5434
|
|
|
5435 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
5436 c_up = SPELL_TOUPPER(c);
|
|
|
5437 if (c_up != c
|
|
|
5438 && (aff_entry->ae_cond == NULL
|
|
|
5439 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
5440 {
|
|
|
5441 p = aff_entry->ae_add
|
|
|
5442 + STRLEN(aff_entry->ae_add);
|
|
|
5443 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
5444 if (PTR2CHAR(p) == c_up)
|
|
|
5445 {
|
|
|
5446 upper = TRUE;
|
|
|
5447 aff_entry->ae_chop = NULL;
|
|
|
5448 *p = NUL;
|
|
|
5449
|
|
|
5450 /* The condition is matched with the
|
|
|
5451 * actual word, thus must check for the
|
|
|
5452 * upper-case letter. */
|
|
|
5453 if (aff_entry->ae_cond != NULL)
|
|
|
5454 {
|
|
|
5455 char_u buf[MAXLINELEN];
|
|
|
5456 #ifdef FEAT_MBYTE
|
|
|
5457 if (has_mbyte)
|
|
|
5458 {
|
|
|
5459 onecap_copy(items[4], buf, TRUE);
|
|
|
5460 aff_entry->ae_cond = getroom_save(
|
|
474
|
5461 spin, buf);
|
|
455
|
5462 }
|
|
|
5463 else
|
|
|
5464 #endif
|
|
|
5465 *aff_entry->ae_cond = c_up;
|
|
|
5466 if (aff_entry->ae_cond != NULL)
|
|
|
5467 {
|
|
|
5468 sprintf((char *)buf, "^%s",
|
|
339
|
5469 aff_entry->ae_cond);
|
|
455
|
5470 vim_free(aff_entry->ae_prog);
|
|
|
5471 aff_entry->ae_prog = vim_regcomp(
|
|
|
5472 buf, RE_MAGIC + RE_STRING);
|
|
|
5473 }
|
|
|
5474 }
|
|
|
5475 }
|
|
|
5476 }
|
|
339
|
5477 }
|
|
|
5478
|
|
455
|
5479 if (aff_entry->ae_chop == NULL)
|
|
390
|
5480 {
|
|
455
|
5481 int idx;
|
|
|
5482 char_u **pp;
|
|
|
5483 int n;
|
|
|
5484
|
|
495
|
5485 /* Find a previously used condition. */
|
|
455
|
5486 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
5487 --idx)
|
|
|
5488 {
|
|
|
5489 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
5490 if (str_equal(p, aff_entry->ae_cond))
|
|
|
5491 break;
|
|
|
5492 }
|
|
|
5493 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
5494 {
|
|
|
5495 /* Not found, add a new condition. */
|
|
|
5496 idx = spin->si_prefcond.ga_len++;
|
|
|
5497 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
5498 + idx;
|
|
|
5499 if (aff_entry->ae_cond == NULL)
|
|
|
5500 *pp = NULL;
|
|
|
5501 else
|
|
474
|
5502 *pp = getroom_save(spin,
|
|
455
|
5503 aff_entry->ae_cond);
|
|
|
5504 }
|
|
|
5505
|
|
|
5506 /* Add the prefix to the prefix tree. */
|
|
|
5507 if (aff_entry->ae_add == NULL)
|
|
|
5508 p = (char_u *)"";
|
|
390
|
5509 else
|
|
455
|
5510 p = aff_entry->ae_add;
|
|
|
5511 /* PFX_FLAGS is a negative number, so that
|
|
|
5512 * tree_add_word() knows this is the prefix tree. */
|
|
|
5513 n = PFX_FLAGS;
|
|
|
5514 if (rare)
|
|
|
5515 n |= WFP_RARE;
|
|
|
5516 if (!cur_aff->ah_combine)
|
|
|
5517 n |= WFP_NC;
|
|
|
5518 if (upper)
|
|
|
5519 n |= WFP_UP;
|
|
474
|
5520 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
5521 idx, cur_aff->ah_newID);
|
|
495
|
5522 did_postpone_prefix = TRUE;
|
|
|
5523 }
|
|
|
5524
|
|
|
5525 /* Didn't actually use ah_newID, backup si_newprefID. */
|
|
|
5526 if (aff_todo == 0 && !did_postpone_prefix)
|
|
|
5527 {
|
|
|
5528 --spin->si_newprefID;
|
|
|
5529 cur_aff->ah_newID = 0;
|
|
390
|
5530 }
|
|
339
|
5531 }
|
|
240
|
5532 }
|
|
236
|
5533 }
|
|
495
|
5534 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2
|
|
|
5535 && fol == NULL)
|
|
|
5536 {
|
|
|
5537 fol = vim_strsave(items[1]);
|
|
|
5538 }
|
|
|
5539 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2
|
|
|
5540 && low == NULL)
|
|
|
5541 {
|
|
|
5542 low = vim_strsave(items[1]);
|
|
|
5543 }
|
|
|
5544 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2
|
|
|
5545 && upp == NULL)
|
|
|
5546 {
|
|
|
5547 upp = vim_strsave(items[1]);
|
|
255
|
5548 }
|
|
625
|
5549 else if ((STRCMP(items[0], "REP") == 0
|
|
|
5550 || STRCMP(items[0], "REPSAL") == 0)
|
|
|
5551 && itemcnt == 2)
|
|
|
5552 {
|
|
|
5553 /* Ignore REP/REPSAL count */;
|
|
323
|
5554 if (!isdigit(*items[1]))
|
|
625
|
5555 smsg((char_u *)_("Expected REP(SAL) count in %s line %d"),
|
|
323
|
5556 fname, lnum);
|
|
|
5557 }
|
|
625
|
5558 else if ((STRCMP(items[0], "REP") == 0
|
|
|
5559 || STRCMP(items[0], "REPSAL") == 0)
|
|
|
5560 && itemcnt >= 3)
|
|
|
5561 {
|
|
|
5562 /* REP/REPSAL item */
|
|
482
|
5563 /* Myspell ignores extra arguments, we require it starts with
|
|
|
5564 * # to detect mistakes. */
|
|
|
5565 if (itemcnt > 3 && items[3][0] != '#')
|
|
|
5566 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]);
|
|
625
|
5567 if (items[0][3] == 'S' ? do_repsal : do_rep)
|
|
532
|
5568 {
|
|
|
5569 /* Replace underscore with space (can't include a space
|
|
|
5570 * directly). */
|
|
|
5571 for (p = items[1]; *p != NUL; mb_ptr_adv(p))
|
|
|
5572 if (*p == '_')
|
|
|
5573 *p = ' ';
|
|
|
5574 for (p = items[2]; *p != NUL; mb_ptr_adv(p))
|
|
|
5575 if (*p == '_')
|
|
|
5576 *p = ' ';
|
|
625
|
5577 add_fromto(spin, items[0][3] == 'S'
|
|
|
5578 ? &spin->si_repsal
|
|
|
5579 : &spin->si_rep, items[1], items[2]);
|
|
532
|
5580 }
|
|
323
|
5581 }
|
|
|
5582 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
5583 {
|
|
|
5584 /* MAP item or count */
|
|
|
5585 if (!found_map)
|
|
|
5586 {
|
|
|
5587 /* First line contains the count. */
|
|
|
5588 found_map = TRUE;
|
|
|
5589 if (!isdigit(*items[1]))
|
|
|
5590 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
5591 fname, lnum);
|
|
|
5592 }
|
|
|
5593 else if (do_map)
|
|
|
5594 {
|
|
346
|
5595 int c;
|
|
|
5596
|
|
|
5597 /* Check that every character appears only once. */
|
|
|
5598 for (p = items[1]; *p != NUL; )
|
|
|
5599 {
|
|
|
5600 #ifdef FEAT_MBYTE
|
|
|
5601 c = mb_ptr2char_adv(&p);
|
|
|
5602 #else
|
|
|
5603 c = *p++;
|
|
|
5604 #endif
|
|
|
5605 if ((spin->si_map.ga_len > 0
|
|
|
5606 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
5607 != NULL)
|
|
|
5608 || vim_strchr(p, c) != NULL)
|
|
|
5609 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
5610 fname, lnum);
|
|
|
5611 }
|
|
|
5612
|
|
323
|
5613 /* We simply concatenate all the MAP strings, separated by
|
|
|
5614 * slashes. */
|
|
|
5615 ga_concat(&spin->si_map, items[1]);
|
|
|
5616 ga_append(&spin->si_map, '/');
|
|
|
5617 }
|
|
|
5618 }
|
|
534
|
5619 /* Accept "SAL from to" and "SAL from to # comment". */
|
|
|
5620 else if (STRCMP(items[0], "SAL") == 0
|
|
|
5621 && (itemcnt == 3 || (itemcnt > 3 && items[3][0] == '#')))
|
|
323
|
5622 {
|
|
|
5623 if (do_sal)
|
|
|
5624 {
|
|
|
5625 /* SAL item (sounds-a-like)
|
|
|
5626 * Either one of the known keys or a from-to pair. */
|
|
|
5627 if (STRCMP(items[1], "followup") == 0)
|
|
|
5628 spin->si_followup = sal_to_bool(items[2]);
|
|
|
5629 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
5630 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
5631 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
5632 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
5633 else
|
|
|
5634 /* when "to" is "_" it means empty */
|
|
|
5635 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
5636 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
5637 : items[2]);
|
|
|
5638 }
|
|
236
|
5639 }
|
|
375
|
5640 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
495
|
5641 && sofofrom == NULL)
|
|
|
5642 {
|
|
|
5643 sofofrom = getroom_save(spin, items[1]);
|
|
375
|
5644 }
|
|
|
5645 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
495
|
5646 && sofoto == NULL)
|
|
|
5647 {
|
|
|
5648 sofoto = getroom_save(spin, items[1]);
|
|
375
|
5649 }
|
|
625
|
5650 else if (STRCMP(items[0], "COMMON") == 0)
|
|
|
5651 {
|
|
|
5652 int i;
|
|
|
5653
|
|
|
5654 for (i = 1; i < itemcnt; ++i)
|
|
|
5655 {
|
|
|
5656 if (HASHITEM_EMPTY(hash_find(&spin->si_commonwords,
|
|
|
5657 items[i])))
|
|
|
5658 {
|
|
|
5659 p = vim_strsave(items[i]);
|
|
|
5660 if (p == NULL)
|
|
|
5661 break;
|
|
|
5662 hash_add(&spin->si_commonwords, p);
|
|
|
5663 }
|
|
|
5664 }
|
|
|
5665 }
|
|
300
|
5666 else
|
|
481
|
5667 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
5668 fname, lnum, items[0]);
|
|
|
5669 }
|
|
|
5670 }
|
|
|
5671
|
|
255
|
5672 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
5673 {
|
|
351
|
5674 if (spin->si_clear_chartab)
|
|
|
5675 {
|
|
|
5676 /* Clear the char type tables, don't want to use any of the
|
|
|
5677 * currently used spell properties. */
|
|
|
5678 init_spell_chartab();
|
|
|
5679 spin->si_clear_chartab = FALSE;
|
|
|
5680 }
|
|
|
5681
|
|
316
|
5682 /*
|
|
|
5683 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
5684 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
5685 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
5686 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
5687 */
|
|
|
5688 if (!spin->si_ascii
|
|
|
5689 #ifdef FEAT_MBYTE
|
|
|
5690 && !enc_utf8
|
|
|
5691 #endif
|
|
|
5692 )
|
|
260
|
5693 {
|
|
|
5694 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
5695 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
5696 else
|
|
316
|
5697 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
5698 }
|
|
255
|
5699
|
|
|
5700 vim_free(fol);
|
|
|
5701 vim_free(low);
|
|
|
5702 vim_free(upp);
|
|
|
5703 }
|
|
|
5704
|
|
481
|
5705 /* Use compound specifications of the .aff file for the spell info. */
|
|
495
|
5706 if (compmax != 0)
|
|
|
5707 {
|
|
|
5708 aff_check_number(spin->si_compmax, compmax, "COMPOUNDMAX");
|
|
|
5709 spin->si_compmax = compmax;
|
|
|
5710 }
|
|
|
5711
|
|
|
5712 if (compminlen != 0)
|
|
|
5713 {
|
|
|
5714 aff_check_number(spin->si_compminlen, compminlen, "COMPOUNDMIN");
|
|
|
5715 spin->si_compminlen = compminlen;
|
|
|
5716 }
|
|
|
5717
|
|
|
5718 if (compsylmax != 0)
|
|
|
5719 {
|
|
|
5720 if (syllable == NULL)
|
|
|
5721 smsg((char_u *)_("COMPOUNDSYLMAX used without SYLLABLE"));
|
|
|
5722 aff_check_number(spin->si_compsylmax, compsylmax, "COMPOUNDSYLMAX");
|
|
|
5723 spin->si_compsylmax = compsylmax;
|
|
|
5724 }
|
|
|
5725
|
|
|
5726 if (compflags != NULL)
|
|
|
5727 process_compflags(spin, aff, compflags);
|
|
|
5728
|
|
|
5729 /* Check that we didn't use too many renumbered flags. */
|
|
500
|
5730 if (spin->si_newcompID < spin->si_newprefID)
|
|
|
5731 {
|
|
|
5732 if (spin->si_newcompID == 127 || spin->si_newcompID == 255)
|
|
495
|
5733 MSG(_("Too many postponed prefixes"));
|
|
500
|
5734 else if (spin->si_newprefID == 0 || spin->si_newprefID == 127)
|
|
495
|
5735 MSG(_("Too many compound flags"));
|
|
481
|
5736 else
|
|
495
|
5737 MSG(_("Too many posponed prefixes and/or compound flags"));
|
|
|
5738 }
|
|
|
5739
|
|
|
5740 if (syllable != NULL)
|
|
|
5741 {
|
|
|
5742 aff_check_string(spin->si_syllable, syllable, "SYLLABLE");
|
|
|
5743 spin->si_syllable = syllable;
|
|
|
5744 }
|
|
|
5745
|
|
|
5746 if (sofofrom != NULL || sofoto != NULL)
|
|
|
5747 {
|
|
|
5748 if (sofofrom == NULL || sofoto == NULL)
|
|
|
5749 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
5750 sofofrom == NULL ? "FROM" : "TO", fname);
|
|
|
5751 else if (spin->si_sal.ga_len > 0)
|
|
|
5752 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
483
|
5753 else
|
|
495
|
5754 {
|
|
|
5755 aff_check_string(spin->si_sofofr, sofofrom, "SOFOFROM");
|
|
|
5756 aff_check_string(spin->si_sofoto, sofoto, "SOFOTO");
|
|
|
5757 spin->si_sofofr = sofofrom;
|
|
|
5758 spin->si_sofoto = sofoto;
|
|
|
5759 }
|
|
|
5760 }
|
|
|
5761
|
|
|
5762 if (midword != NULL)
|
|
|
5763 {
|
|
|
5764 aff_check_string(spin->si_midword, midword, "MIDWORD");
|
|
|
5765 spin->si_midword = midword;
|
|
483
|
5766 }
|
|
|
5767
|
|
236
|
5768 vim_free(pc);
|
|
|
5769 fclose(fd);
|
|
|
5770 return aff;
|
|
|
5771 }
|
|
|
5772
|
|
|
5773 /*
|
|
714
|
5774 * Return TRUE if "s" is the name of an info item in the affix file.
|
|
|
5775 */
|
|
|
5776 static int
|
|
|
5777 spell_info_item(s)
|
|
|
5778 char_u *s;
|
|
|
5779 {
|
|
|
5780 return STRCMP(s, "NAME") == 0
|
|
|
5781 || STRCMP(s, "HOME") == 0
|
|
|
5782 || STRCMP(s, "VERSION") == 0
|
|
|
5783 || STRCMP(s, "AUTHOR") == 0
|
|
|
5784 || STRCMP(s, "EMAIL") == 0
|
|
|
5785 || STRCMP(s, "COPYRIGHT") == 0;
|
|
|
5786 }
|
|
|
5787
|
|
|
5788 /*
|
|
495
|
5789 * Turn an affix flag name into a number, according to the FLAG type.
|
|
|
5790 * returns zero for failure.
|
|
|
5791 */
|
|
|
5792 static unsigned
|
|
|
5793 affitem2flag(flagtype, item, fname, lnum)
|
|
|
5794 int flagtype;
|
|
|
5795 char_u *item;
|
|
|
5796 char_u *fname;
|
|
|
5797 int lnum;
|
|
|
5798 {
|
|
|
5799 unsigned res;
|
|
|
5800 char_u *p = item;
|
|
|
5801
|
|
|
5802 res = get_affitem(flagtype, &p);
|
|
|
5803 if (res == 0)
|
|
|
5804 {
|
|
498
|
5805 if (flagtype == AFT_NUM)
|
|
495
|
5806 smsg((char_u *)_("Flag is not a number in %s line %d: %s"),
|
|
|
5807 fname, lnum, item);
|
|
|
5808 else
|
|
|
5809 smsg((char_u *)_("Illegal flag in %s line %d: %s"),
|
|
|
5810 fname, lnum, item);
|
|
|
5811 }
|
|
|
5812 if (*p != NUL)
|
|
|
5813 {
|
|
|
5814 smsg((char_u *)_(e_affname), fname, lnum, item);
|
|
|
5815 return 0;
|
|
|
5816 }
|
|
|
5817
|
|
|
5818 return res;
|
|
|
5819 }
|
|
|
5820
|
|
|
5821 /*
|
|
|
5822 * Get one affix name from "*pp" and advance the pointer.
|
|
|
5823 * Returns zero for an error, still advances the pointer then.
|
|
|
5824 */
|
|
|
5825 static unsigned
|
|
|
5826 get_affitem(flagtype, pp)
|
|
|
5827 int flagtype;
|
|
|
5828 char_u **pp;
|
|
|
5829 {
|
|
|
5830 int res;
|
|
|
5831
|
|
498
|
5832 if (flagtype == AFT_NUM)
|
|
495
|
5833 {
|
|
|
5834 if (!VIM_ISDIGIT(**pp))
|
|
|
5835 {
|
|
498
|
5836 ++*pp; /* always advance, avoid getting stuck */
|
|
495
|
5837 return 0;
|
|
|
5838 }
|
|
|
5839 res = getdigits(pp);
|
|
|
5840 }
|
|
|
5841 else
|
|
|
5842 {
|
|
|
5843 #ifdef FEAT_MBYTE
|
|
|
5844 res = mb_ptr2char_adv(pp);
|
|
|
5845 #else
|
|
|
5846 res = *(*pp)++;
|
|
|
5847 #endif
|
|
498
|
5848 if (flagtype == AFT_LONG || (flagtype == AFT_CAPLONG
|
|
495
|
5849 && res >= 'A' && res <= 'Z'))
|
|
|
5850 {
|
|
|
5851 if (**pp == NUL)
|
|
|
5852 return 0;
|
|
|
5853 #ifdef FEAT_MBYTE
|
|
|
5854 res = mb_ptr2char_adv(pp) + (res << 16);
|
|
|
5855 #else
|
|
|
5856 res = *(*pp)++ + (res << 16);
|
|
|
5857 #endif
|
|
|
5858 }
|
|
|
5859 }
|
|
|
5860 return res;
|
|
|
5861 }
|
|
|
5862
|
|
|
5863 /*
|
|
|
5864 * Process the "compflags" string used in an affix file and append it to
|
|
|
5865 * spin->si_compflags.
|
|
|
5866 * The processing involves changing the affix names to ID numbers, so that
|
|
|
5867 * they fit in one byte.
|
|
|
5868 */
|
|
|
5869 static void
|
|
|
5870 process_compflags(spin, aff, compflags)
|
|
|
5871 spellinfo_T *spin;
|
|
|
5872 afffile_T *aff;
|
|
|
5873 char_u *compflags;
|
|
|
5874 {
|
|
|
5875 char_u *p;
|
|
|
5876 char_u *prevp;
|
|
|
5877 unsigned flag;
|
|
|
5878 compitem_T *ci;
|
|
|
5879 int id;
|
|
|
5880 int len;
|
|
|
5881 char_u *tp;
|
|
|
5882 char_u key[AH_KEY_LEN];
|
|
|
5883 hashitem_T *hi;
|
|
|
5884
|
|
|
5885 /* Make room for the old and the new compflags, concatenated with a / in
|
|
|
5886 * between. Processing it makes it shorter, but we don't know by how
|
|
|
5887 * much, thus allocate the maximum. */
|
|
|
5888 len = STRLEN(compflags) + 1;
|
|
|
5889 if (spin->si_compflags != NULL)
|
|
|
5890 len += STRLEN(spin->si_compflags) + 1;
|
|
|
5891 p = getroom(spin, len, FALSE);
|
|
|
5892 if (p == NULL)
|
|
|
5893 return;
|
|
|
5894 if (spin->si_compflags != NULL)
|
|
|
5895 {
|
|
|
5896 STRCPY(p, spin->si_compflags);
|
|
|
5897 STRCAT(p, "/");
|
|
|
5898 }
|
|
|
5899 spin->si_compflags = p;
|
|
|
5900 tp = p + STRLEN(p);
|
|
|
5901
|
|
|
5902 for (p = compflags; *p != NUL; )
|
|
|
5903 {
|
|
|
5904 if (vim_strchr((char_u *)"/*+[]", *p) != NULL)
|
|
|
5905 /* Copy non-flag characters directly. */
|
|
|
5906 *tp++ = *p++;
|
|
|
5907 else
|
|
|
5908 {
|
|
|
5909 /* First get the flag number, also checks validity. */
|
|
|
5910 prevp = p;
|
|
|
5911 flag = get_affitem(aff->af_flagtype, &p);
|
|
|
5912 if (flag != 0)
|
|
|
5913 {
|
|
|
5914 /* Find the flag in the hashtable. If it was used before, use
|
|
|
5915 * the existing ID. Otherwise add a new entry. */
|
|
|
5916 vim_strncpy(key, prevp, p - prevp);
|
|
|
5917 hi = hash_find(&aff->af_comp, key);
|
|
|
5918 if (!HASHITEM_EMPTY(hi))
|
|
|
5919 id = HI2CI(hi)->ci_newID;
|
|
|
5920 else
|
|
|
5921 {
|
|
|
5922 ci = (compitem_T *)getroom(spin, sizeof(compitem_T), TRUE);
|
|
|
5923 if (ci == NULL)
|
|
|
5924 break;
|
|
|
5925 STRCPY(ci->ci_key, key);
|
|
|
5926 ci->ci_flag = flag;
|
|
|
5927 /* Avoid using a flag ID that has a special meaning in a
|
|
|
5928 * regexp (also inside []). */
|
|
|
5929 do
|
|
|
5930 {
|
|
500
|
5931 check_renumber(spin);
|
|
|
5932 id = spin->si_newcompID--;
|
|
|
5933 } while (vim_strchr((char_u *)"/+*[]\\-^", id) != NULL);
|
|
495
|
5934 ci->ci_newID = id;
|
|
|
5935 hash_add(&aff->af_comp, ci->ci_key);
|
|
|
5936 }
|
|
|
5937 *tp++ = id;
|
|
|
5938 }
|
|
498
|
5939 if (aff->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5940 ++p;
|
|
|
5941 }
|
|
|
5942 }
|
|
|
5943
|
|
|
5944 *tp = NUL;
|
|
|
5945 }
|
|
|
5946
|
|
|
5947 /*
|
|
500
|
5948 * Check that the new IDs for postponed affixes and compounding don't overrun
|
|
|
5949 * each other. We have almost 255 available, but start at 0-127 to avoid
|
|
|
5950 * using two bytes for utf-8. When the 0-127 range is used up go to 128-255.
|
|
|
5951 * When that is used up an error message is given.
|
|
|
5952 */
|
|
|
5953 static void
|
|
|
5954 check_renumber(spin)
|
|
|
5955 spellinfo_T *spin;
|
|
|
5956 {
|
|
|
5957 if (spin->si_newprefID == spin->si_newcompID && spin->si_newcompID < 128)
|
|
|
5958 {
|
|
|
5959 spin->si_newprefID = 127;
|
|
|
5960 spin->si_newcompID = 255;
|
|
|
5961 }
|
|
|
5962 }
|
|
|
5963
|
|
|
5964 /*
|
|
495
|
5965 * Return TRUE if flag "flag" appears in affix list "afflist".
|
|
|
5966 */
|
|
|
5967 static int
|
|
|
5968 flag_in_afflist(flagtype, afflist, flag)
|
|
|
5969 int flagtype;
|
|
|
5970 char_u *afflist;
|
|
|
5971 unsigned flag;
|
|
|
5972 {
|
|
|
5973 char_u *p;
|
|
|
5974 unsigned n;
|
|
|
5975
|
|
|
5976 switch (flagtype)
|
|
|
5977 {
|
|
|
5978 case AFT_CHAR:
|
|
|
5979 return vim_strchr(afflist, flag) != NULL;
|
|
|
5980
|
|
498
|
5981 case AFT_CAPLONG:
|
|
|
5982 case AFT_LONG:
|
|
495
|
5983 for (p = afflist; *p != NUL; )
|
|
|
5984 {
|
|
|
5985 #ifdef FEAT_MBYTE
|
|
|
5986 n = mb_ptr2char_adv(&p);
|
|
|
5987 #else
|
|
|
5988 n = *p++;
|
|
|
5989 #endif
|
|
498
|
5990 if ((flagtype == AFT_LONG || (n >= 'A' && n <= 'Z'))
|
|
495
|
5991 && *p != NUL)
|
|
|
5992 #ifdef FEAT_MBYTE
|
|
|
5993 n = mb_ptr2char_adv(&p) + (n << 16);
|
|
|
5994 #else
|
|
|
5995 n = *p++ + (n << 16);
|
|
|
5996 #endif
|
|
|
5997 if (n == flag)
|
|
|
5998 return TRUE;
|
|
|
5999 }
|
|
|
6000 break;
|
|
|
6001
|
|
498
|
6002 case AFT_NUM:
|
|
495
|
6003 for (p = afflist; *p != NUL; )
|
|
|
6004 {
|
|
|
6005 n = getdigits(&p);
|
|
|
6006 if (n == flag)
|
|
|
6007 return TRUE;
|
|
|
6008 if (*p != NUL) /* skip over comma */
|
|
|
6009 ++p;
|
|
|
6010 }
|
|
|
6011 break;
|
|
|
6012 }
|
|
|
6013 return FALSE;
|
|
|
6014 }
|
|
|
6015
|
|
|
6016 /*
|
|
|
6017 * Give a warning when "spinval" and "affval" numbers are set and not the same.
|
|
|
6018 */
|
|
|
6019 static void
|
|
|
6020 aff_check_number(spinval, affval, name)
|
|
|
6021 int spinval;
|
|
|
6022 int affval;
|
|
|
6023 char *name;
|
|
|
6024 {
|
|
|
6025 if (spinval != 0 && spinval != affval)
|
|
|
6026 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
6027 }
|
|
|
6028
|
|
|
6029 /*
|
|
|
6030 * Give a warning when "spinval" and "affval" strings are set and not the same.
|
|
|
6031 */
|
|
|
6032 static void
|
|
|
6033 aff_check_string(spinval, affval, name)
|
|
|
6034 char_u *spinval;
|
|
|
6035 char_u *affval;
|
|
|
6036 char *name;
|
|
|
6037 {
|
|
|
6038 if (spinval != NULL && STRCMP(spinval, affval) != 0)
|
|
|
6039 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
6040 }
|
|
|
6041
|
|
|
6042 /*
|
|
339
|
6043 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
6044 * NULL as equal.
|
|
|
6045 */
|
|
|
6046 static int
|
|
|
6047 str_equal(s1, s2)
|
|
|
6048 char_u *s1;
|
|
|
6049 char_u *s2;
|
|
|
6050 {
|
|
|
6051 if (s1 == NULL || s2 == NULL)
|
|
|
6052 return s1 == s2;
|
|
|
6053 return STRCMP(s1, s2) == 0;
|
|
|
6054 }
|
|
|
6055
|
|
|
6056 /*
|
|
323
|
6057 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
6058 * They are stored case-folded.
|
|
|
6059 */
|
|
|
6060 static void
|
|
|
6061 add_fromto(spin, gap, from, to)
|
|
|
6062 spellinfo_T *spin;
|
|
|
6063 garray_T *gap;
|
|
|
6064 char_u *from;
|
|
|
6065 char_u *to;
|
|
|
6066 {
|
|
|
6067 fromto_T *ftp;
|
|
|
6068 char_u word[MAXWLEN];
|
|
|
6069
|
|
|
6070 if (ga_grow(gap, 1) == OK)
|
|
|
6071 {
|
|
|
6072 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
6073 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
6074 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
6075 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
6076 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
6077 ++gap->ga_len;
|
|
|
6078 }
|
|
|
6079 }
|
|
|
6080
|
|
|
6081 /*
|
|
|
6082 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
6083 */
|
|
|
6084 static int
|
|
|
6085 sal_to_bool(s)
|
|
|
6086 char_u *s;
|
|
|
6087 {
|
|
|
6088 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
6089 }
|
|
|
6090
|
|
|
6091 /*
|
|
240
|
6092 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
6093 * When "s" is NULL FALSE is returned.
|
|
|
6094 */
|
|
|
6095 static int
|
|
|
6096 has_non_ascii(s)
|
|
|
6097 char_u *s;
|
|
|
6098 {
|
|
|
6099 char_u *p;
|
|
|
6100
|
|
|
6101 if (s != NULL)
|
|
|
6102 for (p = s; *p != NUL; ++p)
|
|
|
6103 if (*p >= 128)
|
|
|
6104 return TRUE;
|
|
|
6105 return FALSE;
|
|
|
6106 }
|
|
|
6107
|
|
|
6108 /*
|
|
236
|
6109 * Free the structure filled by spell_read_aff().
|
|
|
6110 */
|
|
|
6111 static void
|
|
|
6112 spell_free_aff(aff)
|
|
|
6113 afffile_T *aff;
|
|
|
6114 {
|
|
|
6115 hashtab_T *ht;
|
|
|
6116 hashitem_T *hi;
|
|
|
6117 int todo;
|
|
|
6118 affheader_T *ah;
|
|
300
|
6119 affentry_T *ae;
|
|
236
|
6120
|
|
|
6121 vim_free(aff->af_enc);
|
|
|
6122
|
|
339
|
6123 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
6124 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
6125 {
|
|
|
6126 todo = ht->ht_used;
|
|
|
6127 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
6128 {
|
|
|
6129 if (!HASHITEM_EMPTY(hi))
|
|
|
6130 {
|
|
|
6131 --todo;
|
|
|
6132 ah = HI2AH(hi);
|
|
300
|
6133 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
6134 vim_free(ae->ae_prog);
|
|
236
|
6135 }
|
|
|
6136 }
|
|
|
6137 if (ht == &aff->af_suff)
|
|
|
6138 break;
|
|
|
6139 }
|
|
300
|
6140
|
|
236
|
6141 hash_clear(&aff->af_pref);
|
|
|
6142 hash_clear(&aff->af_suff);
|
|
495
|
6143 hash_clear(&aff->af_comp);
|
|
236
|
6144 }
|
|
|
6145
|
|
|
6146 /*
|
|
300
|
6147 * Read dictionary file "fname".
|
|
236
|
6148 * Returns OK or FAIL;
|
|
|
6149 */
|
|
|
6150 static int
|
|
474
|
6151 spell_read_dic(spin, fname, affile)
|
|
|
6152 spellinfo_T *spin;
|
|
236
|
6153 char_u *fname;
|
|
300
|
6154 afffile_T *affile;
|
|
236
|
6155 {
|
|
300
|
6156 hashtab_T ht;
|
|
236
|
6157 char_u line[MAXLINELEN];
|
|
481
|
6158 char_u *p;
|
|
300
|
6159 char_u *afflist;
|
|
483
|
6160 char_u store_afflist[MAXWLEN];
|
|
|
6161 int pfxlen;
|
|
|
6162 int need_affix;
|
|
300
|
6163 char_u *dw;
|
|
236
|
6164 char_u *pc;
|
|
|
6165 char_u *w;
|
|
|
6166 int l;
|
|
|
6167 hash_T hash;
|
|
|
6168 hashitem_T *hi;
|
|
|
6169 FILE *fd;
|
|
|
6170 int lnum = 1;
|
|
300
|
6171 int non_ascii = 0;
|
|
|
6172 int retval = OK;
|
|
|
6173 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
6174 int flags;
|
|
481
|
6175 int duplicate = 0;
|
|
236
|
6176
|
|
300
|
6177 /*
|
|
|
6178 * Open the file.
|
|
|
6179 */
|
|
310
|
6180 fd = mch_fopen((char *)fname, "r");
|
|
236
|
6181 if (fd == NULL)
|
|
|
6182 {
|
|
|
6183 EMSG2(_(e_notopen), fname);
|
|
|
6184 return FAIL;
|
|
|
6185 }
|
|
|
6186
|
|
300
|
6187 /* The hashtable is only used to detect duplicated words. */
|
|
|
6188 hash_init(&ht);
|
|
|
6189
|
|
625
|
6190 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
6191 _("Reading dictionary file %s ..."), fname);
|
|
|
6192 spell_message(spin, IObuff);
|
|
236
|
6193
|
|
474
|
6194 /* start with a message for the first line */
|
|
|
6195 spin->si_msg_count = 999999;
|
|
|
6196
|
|
236
|
6197 /* Read and ignore the first line: word count. */
|
|
|
6198 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
6199 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
6200 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
6201
|
|
|
6202 /*
|
|
|
6203 * Read all the lines in the file one by one.
|
|
|
6204 * The words are converted to 'encoding' here, before being added to
|
|
|
6205 * the hashtable.
|
|
|
6206 */
|
|
255
|
6207 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
6208 {
|
|
255
|
6209 line_breakcheck();
|
|
236
|
6210 ++lnum;
|
|
455
|
6211 if (line[0] == '#' || line[0] == '/')
|
|
351
|
6212 continue; /* comment line */
|
|
236
|
6213
|
|
300
|
6214 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
6215 * the word is kept to allow e.g., "et al.". */
|
|
236
|
6216 l = STRLEN(line);
|
|
|
6217 while (l > 0 && line[l - 1] <= ' ')
|
|
|
6218 --l;
|
|
|
6219 if (l == 0)
|
|
|
6220 continue; /* empty line */
|
|
|
6221 line[l] = NUL;
|
|
|
6222
|
|
638
|
6223 /* Truncate the word at the "/", set "afflist" to what follows.
|
|
|
6224 * Replace "\/" by "/" and "\\" by "\". */
|
|
481
|
6225 afflist = NULL;
|
|
|
6226 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
6227 {
|
|
638
|
6228 if (*p == '\\' && (p[1] == '\\' || p[1] == '/'))
|
|
|
6229 mch_memmove(p, p + 1, STRLEN(p));
|
|
481
|
6230 else if (*p == '/')
|
|
|
6231 {
|
|
|
6232 *p = NUL;
|
|
|
6233 afflist = p + 1;
|
|
|
6234 break;
|
|
|
6235 }
|
|
|
6236 }
|
|
236
|
6237
|
|
300
|
6238 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
6239 if (spin->si_ascii && has_non_ascii(line))
|
|
|
6240 {
|
|
|
6241 ++non_ascii;
|
|
240
|
6242 continue;
|
|
300
|
6243 }
|
|
240
|
6244
|
|
310
|
6245 #ifdef FEAT_MBYTE
|
|
236
|
6246 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
6247 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
6248 {
|
|
300
|
6249 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
6250 if (pc == NULL)
|
|
|
6251 {
|
|
|
6252 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
6253 fname, lnum, line);
|
|
|
6254 continue;
|
|
|
6255 }
|
|
236
|
6256 w = pc;
|
|
|
6257 }
|
|
|
6258 else
|
|
310
|
6259 #endif
|
|
236
|
6260 {
|
|
|
6261 pc = NULL;
|
|
|
6262 w = line;
|
|
|
6263 }
|
|
|
6264
|
|
474
|
6265 /* This takes time, print a message every 10000 words. */
|
|
|
6266 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
6267 {
|
|
|
6268 spin->si_msg_count = 0;
|
|
339
|
6269 vim_snprintf((char *)message, sizeof(message),
|
|
|
6270 _("line %6d, word %6d - %s"),
|
|
|
6271 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
6272 msg_start();
|
|
|
6273 msg_puts_long_attr(message, 0);
|
|
|
6274 msg_clr_eos();
|
|
|
6275 msg_didout = FALSE;
|
|
|
6276 msg_col = 0;
|
|
|
6277 out_flush();
|
|
|
6278 }
|
|
|
6279
|
|
300
|
6280 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
6281 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
6282 if (dw == NULL)
|
|
300
|
6283 retval = FAIL;
|
|
|
6284 vim_free(pc);
|
|
|
6285 if (retval == FAIL)
|
|
236
|
6286 break;
|
|
|
6287
|
|
300
|
6288 hash = hash_hash(dw);
|
|
|
6289 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
6290 if (!HASHITEM_EMPTY(hi))
|
|
481
|
6291 {
|
|
|
6292 if (p_verbose > 0)
|
|
|
6293 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
6294 fname, lnum, dw);
|
|
481
|
6295 else if (duplicate == 0)
|
|
|
6296 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
6297 fname, lnum, dw);
|
|
|
6298 ++duplicate;
|
|
|
6299 }
|
|
236
|
6300 else
|
|
300
|
6301 hash_add_item(&ht, hi, dw, hash);
|
|
|
6302
|
|
307
|
6303 flags = 0;
|
|
483
|
6304 store_afflist[0] = NUL;
|
|
|
6305 pfxlen = 0;
|
|
|
6306 need_affix = FALSE;
|
|
307
|
6307 if (afflist != NULL)
|
|
|
6308 {
|
|
|
6309 /* Check for affix name that stands for keep-case word and stands
|
|
|
6310 * for rare word (if defined). */
|
|
621
|
6311 if (affile->af_keepcase != 0 && flag_in_afflist(
|
|
|
6312 affile->af_flagtype, afflist, affile->af_keepcase))
|
|
389
|
6313 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
621
|
6314 if (affile->af_rare != 0 && flag_in_afflist(
|
|
|
6315 affile->af_flagtype, afflist, affile->af_rare))
|
|
307
|
6316 flags |= WF_RARE;
|
|
495
|
6317 if (affile->af_bad != 0 && flag_in_afflist(
|
|
|
6318 affile->af_flagtype, afflist, affile->af_bad))
|
|
346
|
6319 flags |= WF_BANNED;
|
|
495
|
6320 if (affile->af_needaffix != 0 && flag_in_afflist(
|
|
|
6321 affile->af_flagtype, afflist, affile->af_needaffix))
|
|
483
|
6322 need_affix = TRUE;
|
|
500
|
6323 if (affile->af_needcomp != 0 && flag_in_afflist(
|
|
|
6324 affile->af_flagtype, afflist, affile->af_needcomp))
|
|
|
6325 flags |= WF_NEEDCOMP;
|
|
698
|
6326 if (affile->af_nosuggest != 0 && flag_in_afflist(
|
|
|
6327 affile->af_flagtype, afflist, affile->af_nosuggest))
|
|
|
6328 flags |= WF_NOSUGGEST;
|
|
339
|
6329
|
|
|
6330 if (affile->af_pfxpostpone)
|
|
|
6331 /* Need to store the list of prefix IDs with the word. */
|
|
483
|
6332 pfxlen = get_pfxlist(affile, afflist, store_afflist);
|
|
|
6333
|
|
|
6334 if (spin->si_compflags != NULL)
|
|
|
6335 /* Need to store the list of compound flags with the word.
|
|
|
6336 * Concatenate them to the list of prefix IDs. */
|
|
495
|
6337 get_compflags(affile, afflist, store_afflist + pfxlen);
|
|
307
|
6338 }
|
|
|
6339
|
|
300
|
6340 /* Add the word to the word tree(s). */
|
|
483
|
6341 if (store_word(spin, dw, flags, spin->si_region,
|
|
|
6342 store_afflist, need_affix) == FAIL)
|
|
300
|
6343 retval = FAIL;
|
|
236
|
6344
|
|
300
|
6345 if (afflist != NULL)
|
|
|
6346 {
|
|
|
6347 /* Find all matching suffixes and add the resulting words.
|
|
|
6348 * Additionally do matching prefixes that combine. */
|
|
474
|
6349 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
6350 &affile->af_suff, &affile->af_pref,
|
|
483
|
6351 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
6352 retval = FAIL;
|
|
|
6353
|
|
|
6354 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
6355 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
6356 &affile->af_pref, NULL,
|
|
483
|
6357 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
6358 retval = FAIL;
|
|
|
6359 }
|
|
236
|
6360 }
|
|
|
6361
|
|
481
|
6362 if (duplicate > 0)
|
|
|
6363 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
6364 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
6365 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
6366 non_ascii, fname);
|
|
300
|
6367 hash_clear(&ht);
|
|
|
6368
|
|
236
|
6369 fclose(fd);
|
|
300
|
6370 return retval;
|
|
236
|
6371 }
|
|
|
6372
|
|
|
6373 /*
|
|
339
|
6374 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
6375 * Used for PFXPOSTPONE.
|
|
483
|
6376 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL
|
|
|
6377 * and return the number of affixes.
|
|
|
6378 */
|
|
|
6379 static int
|
|
|
6380 get_pfxlist(affile, afflist, store_afflist)
|
|
339
|
6381 afffile_T *affile;
|
|
|
6382 char_u *afflist;
|
|
483
|
6383 char_u *store_afflist;
|
|
339
|
6384 {
|
|
|
6385 char_u *p;
|
|
495
|
6386 char_u *prevp;
|
|
483
|
6387 int cnt = 0;
|
|
495
|
6388 int id;
|
|
|
6389 char_u key[AH_KEY_LEN];
|
|
339
|
6390 hashitem_T *hi;
|
|
|
6391
|
|
495
|
6392 for (p = afflist; *p != NUL; )
|
|
|
6393 {
|
|
|
6394 prevp = p;
|
|
|
6395 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
6396 {
|
|
|
6397 /* A flag is a postponed prefix flag if it appears in "af_pref"
|
|
|
6398 * and it's ID is not zero. */
|
|
|
6399 vim_strncpy(key, prevp, p - prevp);
|
|
|
6400 hi = hash_find(&affile->af_pref, key);
|
|
|
6401 if (!HASHITEM_EMPTY(hi))
|
|
|
6402 {
|
|
|
6403 id = HI2AH(hi)->ah_newID;
|
|
|
6404 if (id != 0)
|
|
|
6405 store_afflist[cnt++] = id;
|
|
|
6406 }
|
|
|
6407 }
|
|
498
|
6408 if (affile->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
6409 ++p;
|
|
483
|
6410 }
|
|
|
6411
|
|
|
6412 store_afflist[cnt] = NUL;
|
|
|
6413 return cnt;
|
|
339
|
6414 }
|
|
|
6415
|
|
|
6416 /*
|
|
495
|
6417 * Get the list of compound IDs from the affix list "afflist" that are used
|
|
|
6418 * for compound words.
|
|
483
|
6419 * Puts the flags in "store_afflist[]".
|
|
|
6420 */
|
|
|
6421 static void
|
|
495
|
6422 get_compflags(affile, afflist, store_afflist)
|
|
|
6423 afffile_T *affile;
|
|
481
|
6424 char_u *afflist;
|
|
483
|
6425 char_u *store_afflist;
|
|
481
|
6426 {
|
|
|
6427 char_u *p;
|
|
495
|
6428 char_u *prevp;
|
|
483
|
6429 int cnt = 0;
|
|
495
|
6430 char_u key[AH_KEY_LEN];
|
|
|
6431 hashitem_T *hi;
|
|
|
6432
|
|
|
6433 for (p = afflist; *p != NUL; )
|
|
|
6434 {
|
|
|
6435 prevp = p;
|
|
|
6436 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
6437 {
|
|
|
6438 /* A flag is a compound flag if it appears in "af_comp". */
|
|
|
6439 vim_strncpy(key, prevp, p - prevp);
|
|
|
6440 hi = hash_find(&affile->af_comp, key);
|
|
|
6441 if (!HASHITEM_EMPTY(hi))
|
|
|
6442 store_afflist[cnt++] = HI2CI(hi)->ci_newID;
|
|
|
6443 }
|
|
498
|
6444 if (affile->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
6445 ++p;
|
|
|
6446 }
|
|
483
|
6447
|
|
|
6448 store_afflist[cnt] = NUL;
|
|
481
|
6449 }
|
|
|
6450
|
|
|
6451 /*
|
|
300
|
6452 * Apply affixes to a word and store the resulting words.
|
|
|
6453 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
6454 * prefixes or suffixes.
|
|
|
6455 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
6456 * the resulting words for combining affixes.
|
|
|
6457 *
|
|
|
6458 * Returns FAIL when out of memory.
|
|
236
|
6459 */
|
|
300
|
6460 static int
|
|
483
|
6461 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags,
|
|
|
6462 pfxlist, pfxlen)
|
|
474
|
6463 spellinfo_T *spin; /* spell info */
|
|
300
|
6464 char_u *word; /* basic word start */
|
|
|
6465 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
6466 afffile_T *affile;
|
|
300
|
6467 hashtab_T *ht;
|
|
|
6468 hashtab_T *xht;
|
|
|
6469 int comb; /* only use affixes that combine */
|
|
307
|
6470 int flags; /* flags for the word */
|
|
339
|
6471 char_u *pfxlist; /* list of prefix IDs */
|
|
483
|
6472 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest
|
|
|
6473 * is compound flags */
|
|
236
|
6474 {
|
|
|
6475 int todo;
|
|
|
6476 hashitem_T *hi;
|
|
300
|
6477 affheader_T *ah;
|
|
|
6478 affentry_T *ae;
|
|
|
6479 regmatch_T regmatch;
|
|
|
6480 char_u newword[MAXWLEN];
|
|
|
6481 int retval = OK;
|
|
|
6482 int i;
|
|
|
6483 char_u *p;
|
|
366
|
6484 int use_flags;
|
|
390
|
6485 char_u *use_pfxlist;
|
|
483
|
6486 char_u pfx_pfxlist[MAXWLEN];
|
|
|
6487 size_t wordlen = STRLEN(word);
|
|
236
|
6488
|
|
300
|
6489 todo = ht->ht_used;
|
|
|
6490 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
6491 {
|
|
|
6492 if (!HASHITEM_EMPTY(hi))
|
|
|
6493 {
|
|
|
6494 --todo;
|
|
300
|
6495 ah = HI2AH(hi);
|
|
236
|
6496
|
|
300
|
6497 /* Check that the affix combines, if required, and that the word
|
|
|
6498 * supports this affix. */
|
|
495
|
6499 if ((!comb || ah->ah_combine) && flag_in_afflist(
|
|
|
6500 affile->af_flagtype, afflist, ah->ah_flag))
|
|
236
|
6501 {
|
|
300
|
6502 /* Loop over all affix entries with this name. */
|
|
|
6503 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
6504 {
|
|
300
|
6505 /* Check the condition. It's not logical to match case
|
|
|
6506 * here, but it is required for compatibility with
|
|
339
|
6507 * Myspell.
|
|
481
|
6508 * Another requirement from Myspell is that the chop
|
|
|
6509 * string is shorter than the word itself.
|
|
339
|
6510 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
6511 * prefixes with a chop string. */
|
|
300
|
6512 regmatch.regprog = ae->ae_prog;
|
|
|
6513 regmatch.rm_ic = FALSE;
|
|
339
|
6514 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
6515 || ae->ae_chop != NULL)
|
|
481
|
6516 && (ae->ae_chop == NULL
|
|
|
6517 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
6518 && (ae->ae_prog == NULL
|
|
|
6519 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
6520 {
|
|
|
6521 /* Match. Remove the chop and add the affix. */
|
|
|
6522 if (xht == NULL)
|
|
240
|
6523 {
|
|
300
|
6524 /* prefix: chop/add at the start of the word */
|
|
|
6525 if (ae->ae_add == NULL)
|
|
|
6526 *newword = NUL;
|
|
|
6527 else
|
|
|
6528 STRCPY(newword, ae->ae_add);
|
|
|
6529 p = word;
|
|
|
6530 if (ae->ae_chop != NULL)
|
|
310
|
6531 {
|
|
300
|
6532 /* Skip chop string. */
|
|
310
|
6533 #ifdef FEAT_MBYTE
|
|
|
6534 if (has_mbyte)
|
|
324
|
6535 {
|
|
310
|
6536 i = mb_charlen(ae->ae_chop);
|
|
324
|
6537 for ( ; i > 0; --i)
|
|
|
6538 mb_ptr_adv(p);
|
|
|
6539 }
|
|
310
|
6540 else
|
|
|
6541 #endif
|
|
324
|
6542 p += STRLEN(ae->ae_chop);
|
|
310
|
6543 }
|
|
300
|
6544 STRCAT(newword, p);
|
|
|
6545 }
|
|
|
6546 else
|
|
|
6547 {
|
|
|
6548 /* suffix: chop/add at the end of the word */
|
|
|
6549 STRCPY(newword, word);
|
|
|
6550 if (ae->ae_chop != NULL)
|
|
|
6551 {
|
|
|
6552 /* Remove chop string. */
|
|
|
6553 p = newword + STRLEN(newword);
|
|
376
|
6554 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
6555 for ( ; i > 0; --i)
|
|
300
|
6556 mb_ptr_back(newword, p);
|
|
|
6557 *p = NUL;
|
|
|
6558 }
|
|
|
6559 if (ae->ae_add != NULL)
|
|
|
6560 STRCAT(newword, ae->ae_add);
|
|
240
|
6561 }
|
|
|
6562
|
|
366
|
6563 /* Obey the "rare" flag of the affix. */
|
|
|
6564 if (ae->ae_rare)
|
|
|
6565 use_flags = flags | WF_RARE;
|
|
|
6566 else
|
|
|
6567 use_flags = flags;
|
|
483
|
6568
|
|
|
6569 /* Obey the "nocomp" flag of the affix: don't use the
|
|
|
6570 * compound flags. */
|
|
390
|
6571 use_pfxlist = pfxlist;
|
|
483
|
6572 if (ae->ae_nocomp && pfxlist != NULL)
|
|
|
6573 {
|
|
|
6574 vim_strncpy(pfx_pfxlist, pfxlist, pfxlen);
|
|
|
6575 use_pfxlist = pfx_pfxlist;
|
|
|
6576 }
|
|
390
|
6577
|
|
|
6578 /* When there are postponed prefixes... */
|
|
393
|
6579 if (spin->si_prefroot != NULL
|
|
|
6580 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
6581 {
|
|
|
6582 /* ... add a flag to indicate an affix was used. */
|
|
|
6583 use_flags |= WF_HAS_AFF;
|
|
|
6584
|
|
|
6585 /* ... don't use a prefix list if combining
|
|
483
|
6586 * affixes is not allowed. But do use the
|
|
|
6587 * compound flags after them. */
|
|
|
6588 if ((!ah->ah_combine || comb) && pfxlist != NULL)
|
|
|
6589 use_pfxlist += pfxlen;
|
|
390
|
6590 }
|
|
366
|
6591
|
|
300
|
6592 /* Store the modified word. */
|
|
474
|
6593 if (store_word(spin, newword, use_flags,
|
|
483
|
6594 spin->si_region, use_pfxlist, FALSE) == FAIL)
|
|
300
|
6595 retval = FAIL;
|
|
236
|
6596
|
|
300
|
6597 /* When added a suffix and combining is allowed also
|
|
|
6598 * try adding prefixes additionally. */
|
|
|
6599 if (xht != NULL && ah->ah_combine)
|
|
474
|
6600 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
6601 xht, NULL, TRUE,
|
|
483
|
6602 use_flags, use_pfxlist, pfxlen) == FAIL)
|
|
300
|
6603 retval = FAIL;
|
|
236
|
6604 }
|
|
|
6605 }
|
|
|
6606 }
|
|
|
6607 }
|
|
|
6608 }
|
|
|
6609
|
|
|
6610 return retval;
|
|
|
6611 }
|
|
|
6612
|
|
|
6613 /*
|
|
300
|
6614 * Read a file with a list of words.
|
|
236
|
6615 */
|
|
|
6616 static int
|
|
474
|
6617 spell_read_wordfile(spin, fname)
|
|
|
6618 spellinfo_T *spin;
|
|
300
|
6619 char_u *fname;
|
|
236
|
6620 {
|
|
300
|
6621 FILE *fd;
|
|
|
6622 long lnum = 0;
|
|
|
6623 char_u rline[MAXLINELEN];
|
|
|
6624 char_u *line;
|
|
|
6625 char_u *pc = NULL;
|
|
381
|
6626 char_u *p;
|
|
300
|
6627 int l;
|
|
|
6628 int retval = OK;
|
|
|
6629 int did_word = FALSE;
|
|
|
6630 int non_ascii = 0;
|
|
307
|
6631 int flags;
|
|
316
|
6632 int regionmask;
|
|
236
|
6633
|
|
300
|
6634 /*
|
|
|
6635 * Open the file.
|
|
|
6636 */
|
|
310
|
6637 fd = mch_fopen((char *)fname, "r");
|
|
300
|
6638 if (fd == NULL)
|
|
236
|
6639 {
|
|
300
|
6640 EMSG2(_(e_notopen), fname);
|
|
|
6641 return FAIL;
|
|
236
|
6642 }
|
|
|
6643
|
|
625
|
6644 vim_snprintf((char *)IObuff, IOSIZE, _("Reading word file %s ..."), fname);
|
|
|
6645 spell_message(spin, IObuff);
|
|
300
|
6646
|
|
|
6647 /*
|
|
|
6648 * Read all the lines in the file one by one.
|
|
|
6649 */
|
|
|
6650 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
6651 {
|
|
|
6652 line_breakcheck();
|
|
|
6653 ++lnum;
|
|
|
6654
|
|
|
6655 /* Skip comment lines. */
|
|
|
6656 if (*rline == '#')
|
|
|
6657 continue;
|
|
|
6658
|
|
|
6659 /* Remove CR, LF and white space from the end. */
|
|
|
6660 l = STRLEN(rline);
|
|
|
6661 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
6662 --l;
|
|
|
6663 if (l == 0)
|
|
|
6664 continue; /* empty or blank line */
|
|
|
6665 rline[l] = NUL;
|
|
|
6666
|
|
|
6667 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
6668 vim_free(pc);
|
|
310
|
6669 #ifdef FEAT_MBYTE
|
|
300
|
6670 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
6671 {
|
|
|
6672 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
6673 if (pc == NULL)
|
|
|
6674 {
|
|
|
6675 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
6676 fname, lnum, rline);
|
|
|
6677 continue;
|
|
|
6678 }
|
|
|
6679 line = pc;
|
|
|
6680 }
|
|
|
6681 else
|
|
310
|
6682 #endif
|
|
300
|
6683 {
|
|
|
6684 pc = NULL;
|
|
|
6685 line = rline;
|
|
|
6686 }
|
|
|
6687
|
|
307
|
6688 if (*line == '/')
|
|
300
|
6689 {
|
|
307
|
6690 ++line;
|
|
|
6691 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
6692 {
|
|
|
6693 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
6694 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
6695 fname, lnum, line - 1);
|
|
300
|
6696 else if (did_word)
|
|
316
|
6697 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
6698 fname, lnum, line - 1);
|
|
300
|
6699 else
|
|
|
6700 {
|
|
310
|
6701 #ifdef FEAT_MBYTE
|
|
|
6702 char_u *enc;
|
|
|
6703
|
|
300
|
6704 /* Setup for conversion to 'encoding'. */
|
|
316
|
6705 line += 10;
|
|
|
6706 enc = enc_canonize(line);
|
|
300
|
6707 if (enc != NULL && !spin->si_ascii
|
|
|
6708 && convert_setup(&spin->si_conv, enc,
|
|
|
6709 p_enc) == FAIL)
|
|
|
6710 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
6711 fname, line, p_enc);
|
|
300
|
6712 vim_free(enc);
|
|
375
|
6713 spin->si_conv.vc_fail = TRUE;
|
|
310
|
6714 #else
|
|
|
6715 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
6716 #endif
|
|
300
|
6717 }
|
|
307
|
6718 continue;
|
|
300
|
6719 }
|
|
307
|
6720
|
|
316
|
6721 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
6722 {
|
|
|
6723 if (spin->si_region_count > 1)
|
|
|
6724 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
6725 fname, lnum, line);
|
|
|
6726 else
|
|
|
6727 {
|
|
|
6728 line += 8;
|
|
|
6729 if (STRLEN(line) > 16)
|
|
|
6730 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
6731 fname, lnum, line);
|
|
|
6732 else
|
|
|
6733 {
|
|
|
6734 spin->si_region_count = STRLEN(line) / 2;
|
|
|
6735 STRCPY(spin->si_region_name, line);
|
|
389
|
6736
|
|
|
6737 /* Adjust the mask for a word valid in all regions. */
|
|
|
6738 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
6739 }
|
|
|
6740 }
|
|
|
6741 continue;
|
|
|
6742 }
|
|
|
6743
|
|
381
|
6744 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
6745 fname, lnum, line - 1);
|
|
|
6746 continue;
|
|
|
6747 }
|
|
|
6748
|
|
|
6749 flags = 0;
|
|
|
6750 regionmask = spin->si_region;
|
|
|
6751
|
|
|
6752 /* Check for flags and region after a slash. */
|
|
|
6753 p = vim_strchr(line, '/');
|
|
|
6754 if (p != NULL)
|
|
|
6755 {
|
|
|
6756 *p++ = NUL;
|
|
|
6757 while (*p != NUL)
|
|
|
6758 {
|
|
|
6759 if (*p == '=') /* keep-case word */
|
|
389
|
6760 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
6761 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
6762 flags |= WF_BANNED;
|
|
|
6763 else if (*p == '?') /* Rare word. */
|
|
|
6764 flags |= WF_RARE;
|
|
|
6765 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
6766 {
|
|
381
|
6767 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
6768 regionmask = 0;
|
|
|
6769 flags |= WF_REGION;
|
|
|
6770
|
|
|
6771 l = *p - '0';
|
|
316
|
6772 if (l > spin->si_region_count)
|
|
|
6773 {
|
|
|
6774 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
6775 fname, lnum, p);
|
|
316
|
6776 break;
|
|
|
6777 }
|
|
|
6778 regionmask |= 1 << (l - 1);
|
|
|
6779 }
|
|
381
|
6780 else
|
|
|
6781 {
|
|
|
6782 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
6783 fname, lnum, p);
|
|
|
6784 break;
|
|
|
6785 }
|
|
|
6786 ++p;
|
|
307
|
6787 }
|
|
300
|
6788 }
|
|
|
6789
|
|
|
6790 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
6791 if (spin->si_ascii && has_non_ascii(line))
|
|
|
6792 {
|
|
|
6793 ++non_ascii;
|
|
|
6794 continue;
|
|
|
6795 }
|
|
|
6796
|
|
|
6797 /* Normal word: store it. */
|
|
483
|
6798 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL)
|
|
300
|
6799 {
|
|
|
6800 retval = FAIL;
|
|
|
6801 break;
|
|
|
6802 }
|
|
|
6803 did_word = TRUE;
|
|
|
6804 }
|
|
|
6805
|
|
|
6806 vim_free(pc);
|
|
|
6807 fclose(fd);
|
|
|
6808
|
|
625
|
6809 if (spin->si_ascii && non_ascii > 0)
|
|
|
6810 {
|
|
|
6811 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
6812 _("Ignored %d words with non-ASCII characters"), non_ascii);
|
|
|
6813 spell_message(spin, IObuff);
|
|
|
6814 }
|
|
|
6815
|
|
300
|
6816 return retval;
|
|
236
|
6817 }
|
|
|
6818
|
|
|
6819 /*
|
|
300
|
6820 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
6821 * This avoids calling free() for every little struct we use (and keeping
|
|
|
6822 * track of them).
|
|
300
|
6823 * The memory is cleared to all zeros.
|
|
|
6824 * Returns NULL when out of memory.
|
|
|
6825 */
|
|
|
6826 static void *
|
|
474
|
6827 getroom(spin, len, align)
|
|
|
6828 spellinfo_T *spin;
|
|
447
|
6829 size_t len; /* length needed */
|
|
|
6830 int align; /* align for pointer */
|
|
300
|
6831 {
|
|
|
6832 char_u *p;
|
|
474
|
6833 sblock_T *bl = spin->si_blocks;
|
|
300
|
6834
|
|
447
|
6835 if (align && bl != NULL)
|
|
|
6836 /* Round size up for alignment. On some systems structures need to be
|
|
|
6837 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
6838 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
6839 & ~(sizeof(char *) - 1);
|
|
|
6840
|
|
300
|
6841 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
6842 {
|
|
|
6843 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
6844 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
6845 if (bl == NULL)
|
|
|
6846 return NULL;
|
|
474
|
6847 bl->sb_next = spin->si_blocks;
|
|
|
6848 spin->si_blocks = bl;
|
|
300
|
6849 bl->sb_used = 0;
|
|
482
|
6850 ++spin->si_blocks_cnt;
|
|
300
|
6851 }
|
|
|
6852
|
|
|
6853 p = bl->sb_data + bl->sb_used;
|
|
|
6854 bl->sb_used += len;
|
|
|
6855
|
|
|
6856 return p;
|
|
|
6857 }
|
|
|
6858
|
|
|
6859 /*
|
|
|
6860 * Make a copy of a string into memory allocated with getroom().
|
|
|
6861 */
|
|
|
6862 static char_u *
|
|
474
|
6863 getroom_save(spin, s)
|
|
|
6864 spellinfo_T *spin;
|
|
300
|
6865 char_u *s;
|
|
|
6866 {
|
|
|
6867 char_u *sc;
|
|
|
6868
|
|
474
|
6869 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
6870 if (sc != NULL)
|
|
|
6871 STRCPY(sc, s);
|
|
|
6872 return sc;
|
|
|
6873 }
|
|
|
6874
|
|
|
6875
|
|
|
6876 /*
|
|
|
6877 * Free the list of allocated sblock_T.
|
|
236
|
6878 */
|
|
|
6879 static void
|
|
300
|
6880 free_blocks(bl)
|
|
|
6881 sblock_T *bl;
|
|
236
|
6882 {
|
|
300
|
6883 sblock_T *next;
|
|
236
|
6884
|
|
300
|
6885 while (bl != NULL)
|
|
236
|
6886 {
|
|
300
|
6887 next = bl->sb_next;
|
|
|
6888 vim_free(bl);
|
|
|
6889 bl = next;
|
|
236
|
6890 }
|
|
|
6891 }
|
|
|
6892
|
|
|
6893 /*
|
|
300
|
6894 * Allocate the root of a word tree.
|
|
236
|
6895 */
|
|
300
|
6896 static wordnode_T *
|
|
474
|
6897 wordtree_alloc(spin)
|
|
|
6898 spellinfo_T *spin;
|
|
|
6899 {
|
|
|
6900 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
6901 }
|
|
|
6902
|
|
|
6903 /*
|
|
300
|
6904 * Store a word in the tree(s).
|
|
390
|
6905 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
6906 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
6907 * used to find suggestions.
|
|
300
|
6908 * For a keep-case word also store it in the keep-case tree.
|
|
482
|
6909 * When "pfxlist" is not NULL store the word for each postponed prefix ID and
|
|
|
6910 * compound flag.
|
|
236
|
6911 */
|
|
|
6912 static int
|
|
483
|
6913 store_word(spin, word, flags, region, pfxlist, need_affix)
|
|
474
|
6914 spellinfo_T *spin;
|
|
300
|
6915 char_u *word;
|
|
307
|
6916 int flags; /* extra flags, WF_BANNED */
|
|
316
|
6917 int region; /* supported region(s) */
|
|
339
|
6918 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
483
|
6919 int need_affix; /* only store word with affix ID */
|
|
236
|
6920 {
|
|
300
|
6921 int len = STRLEN(word);
|
|
|
6922 int ct = captype(word, word + len);
|
|
|
6923 char_u foldword[MAXWLEN];
|
|
339
|
6924 int res = OK;
|
|
|
6925 char_u *p;
|
|
236
|
6926
|
|
323
|
6927 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
6928 for (p = pfxlist; res == OK; ++p)
|
|
|
6929 {
|
|
483
|
6930 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6931 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
474
|
6932 region, p == NULL ? 0 : *p);
|
|
339
|
6933 if (p == NULL || *p == NUL)
|
|
|
6934 break;
|
|
|
6935 }
|
|
334
|
6936 ++spin->si_foldwcount;
|
|
307
|
6937
|
|
474
|
6938 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
6939 {
|
|
339
|
6940 for (p = pfxlist; res == OK; ++p)
|
|
|
6941 {
|
|
483
|
6942 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6943 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
474
|
6944 region, p == NULL ? 0 : *p);
|
|
339
|
6945 if (p == NULL || *p == NUL)
|
|
|
6946 break;
|
|
|
6947 }
|
|
334
|
6948 ++spin->si_keepwcount;
|
|
|
6949 }
|
|
300
|
6950 return res;
|
|
236
|
6951 }
|
|
|
6952
|
|
|
6953 /*
|
|
300
|
6954 * Add word "word" to a word tree at "root".
|
|
625
|
6955 * When "flags" < 0 we are adding to the prefix tree where "flags" is used for
|
|
366
|
6956 * "rare" and "region" is the condition nr.
|
|
255
|
6957 * Returns FAIL when out of memory.
|
|
236
|
6958 */
|
|
255
|
6959 static int
|
|
481
|
6960 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
6961 spellinfo_T *spin;
|
|
300
|
6962 char_u *word;
|
|
|
6963 wordnode_T *root;
|
|
|
6964 int flags;
|
|
|
6965 int region;
|
|
481
|
6966 int affixID;
|
|
236
|
6967 {
|
|
300
|
6968 wordnode_T *node = root;
|
|
|
6969 wordnode_T *np;
|
|
469
|
6970 wordnode_T *copyp, **copyprev;
|
|
300
|
6971 wordnode_T **prev = NULL;
|
|
|
6972 int i;
|
|
255
|
6973
|
|
300
|
6974 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
6975 for (i = 0; ; ++i)
|
|
255
|
6976 {
|
|
469
|
6977 /* When there is more than one reference to this node we need to make
|
|
|
6978 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
6979 * (we don't optimize for a partly shared list of siblings). */
|
|
|
6980 if (node != NULL && node->wn_refs > 1)
|
|
|
6981 {
|
|
|
6982 --node->wn_refs;
|
|
|
6983 copyprev = prev;
|
|
|
6984 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
6985 {
|
|
|
6986 /* Allocate a new node and copy the info. */
|
|
474
|
6987 np = get_wordnode(spin);
|
|
469
|
6988 if (np == NULL)
|
|
|
6989 return FAIL;
|
|
|
6990 np->wn_child = copyp->wn_child;
|
|
|
6991 if (np->wn_child != NULL)
|
|
|
6992 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
6993 np->wn_byte = copyp->wn_byte;
|
|
|
6994 if (np->wn_byte == NUL)
|
|
|
6995 {
|
|
|
6996 np->wn_flags = copyp->wn_flags;
|
|
|
6997 np->wn_region = copyp->wn_region;
|
|
481
|
6998 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
6999 }
|
|
|
7000
|
|
|
7001 /* Link the new node in the list, there will be one ref. */
|
|
|
7002 np->wn_refs = 1;
|
|
|
7003 *copyprev = np;
|
|
|
7004 copyprev = &np->wn_sibling;
|
|
|
7005
|
|
|
7006 /* Let "node" point to the head of the copied list. */
|
|
|
7007 if (copyp == node)
|
|
|
7008 node = np;
|
|
|
7009 }
|
|
|
7010 }
|
|
|
7011
|
|
300
|
7012 /* Look for the sibling that has the same character. They are sorted
|
|
|
7013 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
7014 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
7015 * done on flags and then on affixID. */
|
|
339
|
7016 while (node != NULL
|
|
|
7017 && (node->wn_byte < word[i]
|
|
|
7018 || (node->wn_byte == NUL
|
|
|
7019 && (flags < 0
|
|
625
|
7020 ? node->wn_affixID < (unsigned)affixID
|
|
|
7021 : (node->wn_flags < (unsigned)(flags & WN_MASK)
|
|
390
|
7022 || (node->wn_flags == (flags & WN_MASK)
|
|
625
|
7023 && (spin->si_sugtree
|
|
|
7024 ? (node->wn_region & 0xffff) < region
|
|
|
7025 : node->wn_affixID
|
|
|
7026 < (unsigned)affixID)))))))
|
|
236
|
7027 {
|
|
300
|
7028 prev = &node->wn_sibling;
|
|
|
7029 node = *prev;
|
|
236
|
7030 }
|
|
339
|
7031 if (node == NULL
|
|
|
7032 || node->wn_byte != word[i]
|
|
|
7033 || (word[i] == NUL
|
|
|
7034 && (flags < 0
|
|
625
|
7035 || spin->si_sugtree
|
|
390
|
7036 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
7037 || node->wn_affixID != affixID)))
|
|
255
|
7038 {
|
|
300
|
7039 /* Allocate a new node. */
|
|
474
|
7040 np = get_wordnode(spin);
|
|
300
|
7041 if (np == NULL)
|
|
|
7042 return FAIL;
|
|
|
7043 np->wn_byte = word[i];
|
|
469
|
7044
|
|
|
7045 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
7046 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
7047 * make ref count of sibling one (matters when inserting in front
|
|
|
7048 * of the list of siblings). */
|
|
|
7049 if (node == NULL)
|
|
|
7050 np->wn_refs = 1;
|
|
|
7051 else
|
|
|
7052 {
|
|
|
7053 np->wn_refs = node->wn_refs;
|
|
|
7054 node->wn_refs = 1;
|
|
|
7055 }
|
|
300
|
7056 *prev = np;
|
|
|
7057 np->wn_sibling = node;
|
|
|
7058 node = np;
|
|
255
|
7059 }
|
|
300
|
7060
|
|
|
7061 if (word[i] == NUL)
|
|
|
7062 {
|
|
|
7063 node->wn_flags = flags;
|
|
|
7064 node->wn_region |= region;
|
|
481
|
7065 node->wn_affixID = affixID;
|
|
300
|
7066 break;
|
|
|
7067 }
|
|
|
7068 prev = &node->wn_child;
|
|
|
7069 node = *prev;
|
|
255
|
7070 }
|
|
469
|
7071 #ifdef SPELL_PRINTTREE
|
|
|
7072 smsg("Added \"%s\"", word);
|
|
|
7073 spell_print_tree(root->wn_sibling);
|
|
|
7074 #endif
|
|
|
7075
|
|
474
|
7076 /* count nr of words added since last message */
|
|
|
7077 ++spin->si_msg_count;
|
|
|
7078
|
|
482
|
7079 if (spin->si_compress_cnt > 1)
|
|
|
7080 {
|
|
|
7081 if (--spin->si_compress_cnt == 1)
|
|
|
7082 /* Did enough words to lower the block count limit. */
|
|
483
|
7083 spin->si_blocks_cnt += compress_inc;
|
|
482
|
7084 }
|
|
|
7085
|
|
469
|
7086 /*
|
|
482
|
7087 * When we have allocated lots of memory we need to compress the word tree
|
|
|
7088 * to free up some room. But compression is slow, and we might actually
|
|
|
7089 * need that room, thus only compress in the following situations:
|
|
|
7090 * 1. When not compressed before (si_compress_cnt == 0): when using
|
|
483
|
7091 * "compress_start" blocks.
|
|
|
7092 * 2. When compressed before and used "compress_inc" blocks before
|
|
|
7093 * adding "compress_added" words (si_compress_cnt > 1).
|
|
|
7094 * 3. When compressed before, added "compress_added" words
|
|
482
|
7095 * (si_compress_cnt == 1) and the number of free nodes drops below the
|
|
|
7096 * maximum word length.
|
|
469
|
7097 */
|
|
482
|
7098 #ifndef SPELL_PRINTTREE
|
|
|
7099 if (spin->si_compress_cnt == 1
|
|
|
7100 ? spin->si_free_count < MAXWLEN
|
|
483
|
7101 : spin->si_blocks_cnt >= compress_start)
|
|
482
|
7102 #endif
|
|
|
7103 {
|
|
|
7104 /* Decrement the block counter. The effect is that we compress again
|
|
483
|
7105 * when the freed up room has been used and another "compress_inc"
|
|
|
7106 * blocks have been allocated. Unless "compress_added" words have
|
|
482
|
7107 * been added, then the limit is put back again. */
|
|
483
|
7108 spin->si_blocks_cnt -= compress_inc;
|
|
|
7109 spin->si_compress_cnt = compress_added;
|
|
482
|
7110
|
|
474
|
7111 if (spin->si_verbose)
|
|
|
7112 {
|
|
|
7113 msg_start();
|
|
|
7114 msg_puts((char_u *)_(msg_compressing));
|
|
|
7115 msg_clr_eos();
|
|
|
7116 msg_didout = FALSE;
|
|
|
7117 msg_col = 0;
|
|
|
7118 out_flush();
|
|
|
7119 }
|
|
482
|
7120
|
|
|
7121 /* Compress both trees. Either they both have many nodes, which makes
|
|
|
7122 * compression useful, or one of them is small, which means
|
|
625
|
7123 * compression goes fast. But when filling the souldfold word tree
|
|
|
7124 * there is no keep-case tree. */
|
|
482
|
7125 wordtree_compress(spin, spin->si_foldroot);
|
|
625
|
7126 if (affixID >= 0)
|
|
|
7127 wordtree_compress(spin, spin->si_keeproot);
|
|
469
|
7128 }
|
|
255
|
7129
|
|
|
7130 return OK;
|
|
236
|
7131 }
|
|
|
7132
|
|
469
|
7133 /*
|
|
483
|
7134 * Check the 'mkspellmem' option. Return FAIL if it's wrong.
|
|
|
7135 * Sets "sps_flags".
|
|
|
7136 */
|
|
|
7137 int
|
|
|
7138 spell_check_msm()
|
|
|
7139 {
|
|
|
7140 char_u *p = p_msm;
|
|
|
7141 long start = 0;
|
|
|
7142 long inc = 0;
|
|
|
7143 long added = 0;
|
|
|
7144
|
|
|
7145 if (!VIM_ISDIGIT(*p))
|
|
|
7146 return FAIL;
|
|
|
7147 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/
|
|
|
7148 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102);
|
|
|
7149 if (*p != ',')
|
|
|
7150 return FAIL;
|
|
|
7151 ++p;
|
|
|
7152 if (!VIM_ISDIGIT(*p))
|
|
|
7153 return FAIL;
|
|
|
7154 inc = (getdigits(&p) * 102) / (SBLOCKSIZE / 10);
|
|
|
7155 if (*p != ',')
|
|
|
7156 return FAIL;
|
|
|
7157 ++p;
|
|
|
7158 if (!VIM_ISDIGIT(*p))
|
|
|
7159 return FAIL;
|
|
|
7160 added = getdigits(&p) * 1024;
|
|
|
7161 if (*p != NUL)
|
|
|
7162 return FAIL;
|
|
|
7163
|
|
|
7164 if (start == 0 || inc == 0 || added == 0 || inc > start)
|
|
|
7165 return FAIL;
|
|
|
7166
|
|
|
7167 compress_start = start;
|
|
|
7168 compress_inc = inc;
|
|
|
7169 compress_added = added;
|
|
|
7170 return OK;
|
|
|
7171 }
|
|
|
7172
|
|
|
7173
|
|
|
7174 /*
|
|
469
|
7175 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
7176 * allocate a new one.
|
|
|
7177 */
|
|
|
7178 static wordnode_T *
|
|
474
|
7179 get_wordnode(spin)
|
|
|
7180 spellinfo_T *spin;
|
|
469
|
7181 {
|
|
|
7182 wordnode_T *n;
|
|
|
7183
|
|
474
|
7184 if (spin->si_first_free == NULL)
|
|
|
7185 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
7186 else
|
|
|
7187 {
|
|
474
|
7188 n = spin->si_first_free;
|
|
|
7189 spin->si_first_free = n->wn_child;
|
|
469
|
7190 vim_memset(n, 0, sizeof(wordnode_T));
|
|
482
|
7191 --spin->si_free_count;
|
|
469
|
7192 }
|
|
|
7193 #ifdef SPELL_PRINTTREE
|
|
474
|
7194 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
7195 #endif
|
|
|
7196 return n;
|
|
|
7197 }
|
|
|
7198
|
|
|
7199 /*
|
|
474
|
7200 * Decrement the reference count on a node (which is the head of a list of
|
|
|
7201 * siblings). If the reference count becomes zero free the node and its
|
|
|
7202 * siblings.
|
|
625
|
7203 * Returns the number of nodes actually freed.
|
|
|
7204 */
|
|
|
7205 static int
|
|
474
|
7206 deref_wordnode(spin, node)
|
|
|
7207 spellinfo_T *spin;
|
|
|
7208 wordnode_T *node;
|
|
|
7209 {
|
|
625
|
7210 wordnode_T *np;
|
|
|
7211 int cnt = 0;
|
|
474
|
7212
|
|
|
7213 if (--node->wn_refs == 0)
|
|
625
|
7214 {
|
|
474
|
7215 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7216 {
|
|
|
7217 if (np->wn_child != NULL)
|
|
625
|
7218 cnt += deref_wordnode(spin, np->wn_child);
|
|
474
|
7219 free_wordnode(spin, np);
|
|
625
|
7220 ++cnt;
|
|
|
7221 }
|
|
|
7222 ++cnt; /* length field */
|
|
|
7223 }
|
|
|
7224 return cnt;
|
|
474
|
7225 }
|
|
|
7226
|
|
|
7227 /*
|
|
469
|
7228 * Free a wordnode_T for re-use later.
|
|
474
|
7229 * Only the "wn_child" field becomes invalid.
|
|
469
|
7230 */
|
|
|
7231 static void
|
|
474
|
7232 free_wordnode(spin, n)
|
|
|
7233 spellinfo_T *spin;
|
|
|
7234 wordnode_T *n;
|
|
|
7235 {
|
|
|
7236 n->wn_child = spin->si_first_free;
|
|
|
7237 spin->si_first_free = n;
|
|
482
|
7238 ++spin->si_free_count;
|
|
469
|
7239 }
|
|
|
7240
|
|
236
|
7241 /*
|
|
300
|
7242 * Compress a tree: find tails that are identical and can be shared.
|
|
|
7243 */
|
|
|
7244 static void
|
|
474
|
7245 wordtree_compress(spin, root)
|
|
|
7246 spellinfo_T *spin;
|
|
300
|
7247 wordnode_T *root;
|
|
|
7248 {
|
|
|
7249 hashtab_T ht;
|
|
|
7250 int n;
|
|
|
7251 int tot = 0;
|
|
469
|
7252 int perc;
|
|
300
|
7253
|
|
474
|
7254 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
7255 * start of the tree. */
|
|
|
7256 if (root->wn_sibling != NULL)
|
|
300
|
7257 {
|
|
|
7258 hash_init(&ht);
|
|
474
|
7259 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
7260
|
|
|
7261 #ifndef SPELL_PRINTTREE
|
|
310
|
7262 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
7263 #endif
|
|
310
|
7264 {
|
|
469
|
7265 if (tot > 1000000)
|
|
|
7266 perc = (tot - n) / (tot / 100);
|
|
482
|
7267 else if (tot == 0)
|
|
|
7268 perc = 0;
|
|
469
|
7269 else
|
|
|
7270 perc = (tot - n) * 100 / tot;
|
|
625
|
7271 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
7272 _("Compressed %d of %d nodes; %d (%d%%) remaining"),
|
|
|
7273 n, tot, tot - n, perc);
|
|
|
7274 spell_message(spin, IObuff);
|
|
310
|
7275 }
|
|
469
|
7276 #ifdef SPELL_PRINTTREE
|
|
474
|
7277 spell_print_tree(root->wn_sibling);
|
|
469
|
7278 #endif
|
|
300
|
7279 hash_clear(&ht);
|
|
|
7280 }
|
|
|
7281 }
|
|
|
7282
|
|
|
7283 /*
|
|
|
7284 * Compress a node, its siblings and its children, depth first.
|
|
|
7285 * Returns the number of compressed nodes.
|
|
236
|
7286 */
|
|
255
|
7287 static int
|
|
474
|
7288 node_compress(spin, node, ht, tot)
|
|
|
7289 spellinfo_T *spin;
|
|
300
|
7290 wordnode_T *node;
|
|
|
7291 hashtab_T *ht;
|
|
|
7292 int *tot; /* total count of nodes before compressing,
|
|
|
7293 incremented while going through the tree */
|
|
236
|
7294 {
|
|
300
|
7295 wordnode_T *np;
|
|
|
7296 wordnode_T *tp;
|
|
|
7297 wordnode_T *child;
|
|
|
7298 hash_T hash;
|
|
236
|
7299 hashitem_T *hi;
|
|
300
|
7300 int len = 0;
|
|
|
7301 unsigned nr, n;
|
|
|
7302 int compressed = 0;
|
|
236
|
7303
|
|
300
|
7304 /*
|
|
|
7305 * Go through the list of siblings. Compress each child and then try
|
|
|
7306 * finding an identical child to replace it.
|
|
|
7307 * Note that with "child" we mean not just the node that is pointed to,
|
|
625
|
7308 * but the whole list of siblings of which the child node is the first.
|
|
300
|
7309 */
|
|
482
|
7310 for (np = node; np != NULL && !got_int; np = np->wn_sibling)
|
|
236
|
7311 {
|
|
300
|
7312 ++len;
|
|
|
7313 if ((child = np->wn_child) != NULL)
|
|
|
7314 {
|
|
625
|
7315 /* Compress the child first. This fills hashkey. */
|
|
474
|
7316 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
7317
|
|
|
7318 /* Try to find an identical child. */
|
|
346
|
7319 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
7320 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
7321 if (!HASHITEM_EMPTY(hi))
|
|
|
7322 {
|
|
625
|
7323 /* There are children we encountered before with a hash value
|
|
|
7324 * identical to the current child. Now check if there is one
|
|
|
7325 * that is really identical. */
|
|
346
|
7326 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
7327 if (node_equal(child, tp))
|
|
|
7328 {
|
|
|
7329 /* Found one! Now use that child in place of the
|
|
469
|
7330 * current one. This means the current child and all
|
|
|
7331 * its siblings is unlinked from the tree. */
|
|
|
7332 ++tp->wn_refs;
|
|
625
|
7333 compressed += deref_wordnode(spin, child);
|
|
300
|
7334 np->wn_child = tp;
|
|
|
7335 break;
|
|
|
7336 }
|
|
|
7337 if (tp == NULL)
|
|
|
7338 {
|
|
|
7339 /* No other child with this hash value equals the child of
|
|
|
7340 * the node, add it to the linked list after the first
|
|
|
7341 * item. */
|
|
|
7342 tp = HI2WN(hi);
|
|
346
|
7343 child->wn_u2.next = tp->wn_u2.next;
|
|
|
7344 tp->wn_u2.next = child;
|
|
300
|
7345 }
|
|
|
7346 }
|
|
|
7347 else
|
|
|
7348 /* No other child has this hash value, add it to the
|
|
|
7349 * hashtable. */
|
|
346
|
7350 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
7351 }
|
|
236
|
7352 }
|
|
625
|
7353 *tot += len + 1; /* add one for the node that stores the length */
|
|
300
|
7354
|
|
|
7355 /*
|
|
|
7356 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
7357 * find a lookalike node. This must be done after compressing the sibling
|
|
|
7358 * list, otherwise the hash key would become invalid by the compression.
|
|
|
7359 */
|
|
346
|
7360 node->wn_u1.hashkey[0] = len;
|
|
300
|
7361 nr = 0;
|
|
|
7362 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
7363 {
|
|
300
|
7364 if (np->wn_byte == NUL)
|
|
481
|
7365 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
7366 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
7367 else
|
|
|
7368 /* byte node: use the byte value and the child pointer */
|
|
|
7369 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
7370 nr = nr * 101 + n;
|
|
236
|
7371 }
|
|
300
|
7372
|
|
|
7373 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
7374 n = nr & 0xff;
|
|
346
|
7375 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
7376 n = (nr >> 8) & 0xff;
|
|
346
|
7377 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
7378 n = (nr >> 16) & 0xff;
|
|
346
|
7379 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
7380 n = (nr >> 24) & 0xff;
|
|
346
|
7381 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
7382 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
7383
|
|
482
|
7384 /* Check for CTRL-C pressed now and then. */
|
|
|
7385 fast_breakcheck();
|
|
|
7386
|
|
300
|
7387 return compressed;
|
|
|
7388 }
|
|
|
7389
|
|
|
7390 /*
|
|
|
7391 * Return TRUE when two nodes have identical siblings and children.
|
|
|
7392 */
|
|
|
7393 static int
|
|
|
7394 node_equal(n1, n2)
|
|
|
7395 wordnode_T *n1;
|
|
|
7396 wordnode_T *n2;
|
|
|
7397 {
|
|
|
7398 wordnode_T *p1;
|
|
|
7399 wordnode_T *p2;
|
|
|
7400
|
|
|
7401 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
7402 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
7403 if (p1->wn_byte != p2->wn_byte
|
|
|
7404 || (p1->wn_byte == NUL
|
|
|
7405 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
7406 || p1->wn_region != p2->wn_region
|
|
481
|
7407 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
7408 : (p1->wn_child != p2->wn_child)))
|
|
|
7409 break;
|
|
|
7410
|
|
|
7411 return p1 == NULL && p2 == NULL;
|
|
236
|
7412 }
|
|
|
7413
|
|
|
7414 /*
|
|
|
7415 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
7416 */
|
|
255
|
7417 void
|
|
236
|
7418 put_bytes(fd, nr, len)
|
|
|
7419 FILE *fd;
|
|
|
7420 long_u nr;
|
|
|
7421 int len;
|
|
|
7422 {
|
|
|
7423 int i;
|
|
|
7424
|
|
|
7425 for (i = len - 1; i >= 0; --i)
|
|
|
7426 putc((int)(nr >> (i * 8)), fd);
|
|
|
7427 }
|
|
|
7428
|
|
714
|
7429 #ifdef _MSC_VER
|
|
|
7430 # if (_MSC_VER <= 1200)
|
|
|
7431 /* This line is required for VC6 without the service pack. Also see the
|
|
|
7432 * matching #pragma below. */
|
|
|
7433 /* # pragma optimize("", off) */
|
|
|
7434 # endif
|
|
|
7435 #endif
|
|
|
7436
|
|
625
|
7437 /*
|
|
|
7438 * Write spin->si_sugtime to file "fd".
|
|
|
7439 */
|
|
|
7440 static void
|
|
|
7441 put_sugtime(spin, fd)
|
|
|
7442 spellinfo_T *spin;
|
|
|
7443 FILE *fd;
|
|
|
7444 {
|
|
|
7445 int c;
|
|
|
7446 int i;
|
|
|
7447
|
|
|
7448 /* time_t can be up to 8 bytes in size, more than long_u, thus we
|
|
|
7449 * can't use put_bytes() here. */
|
|
|
7450 for (i = 7; i >= 0; --i)
|
|
|
7451 if (i + 1 > sizeof(time_t))
|
|
|
7452 /* ">>" doesn't work well when shifting more bits than avail */
|
|
|
7453 putc(0, fd);
|
|
|
7454 else
|
|
|
7455 {
|
|
|
7456 c = (unsigned)spin->si_sugtime >> (i * 8);
|
|
|
7457 putc(c, fd);
|
|
|
7458 }
|
|
|
7459 }
|
|
|
7460
|
|
714
|
7461 #ifdef _MSC_VER
|
|
|
7462 # if (_MSC_VER <= 1200)
|
|
|
7463 /* # pragma optimize("", on) */
|
|
|
7464 # endif
|
|
|
7465 #endif
|
|
|
7466
|
|
323
|
7467 static int
|
|
|
7468 #ifdef __BORLANDC__
|
|
|
7469 _RTLENTRYF
|
|
|
7470 #endif
|
|
|
7471 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
7472
|
|
|
7473 /*
|
|
|
7474 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
7475 */
|
|
|
7476 static int
|
|
|
7477 #ifdef __BORLANDC__
|
|
|
7478 _RTLENTRYF
|
|
|
7479 #endif
|
|
|
7480 rep_compare(s1, s2)
|
|
|
7481 const void *s1;
|
|
|
7482 const void *s2;
|
|
|
7483 {
|
|
|
7484 fromto_T *p1 = (fromto_T *)s1;
|
|
|
7485 fromto_T *p2 = (fromto_T *)s2;
|
|
|
7486
|
|
|
7487 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
7488 }
|
|
|
7489
|
|
236
|
7490 /*
|
|
483
|
7491 * Write the Vim .spl file "fname".
|
|
500
|
7492 * Return FAIL or OK;
|
|
|
7493 */
|
|
|
7494 static int
|
|
474
|
7495 write_vim_spell(spin, fname)
|
|
|
7496 spellinfo_T *spin;
|
|
236
|
7497 char_u *fname;
|
|
|
7498 {
|
|
300
|
7499 FILE *fd;
|
|
|
7500 int regionmask;
|
|
236
|
7501 int round;
|
|
300
|
7502 wordnode_T *tree;
|
|
|
7503 int nodecount;
|
|
323
|
7504 int i;
|
|
|
7505 int l;
|
|
|
7506 garray_T *gap;
|
|
|
7507 fromto_T *ftp;
|
|
|
7508 char_u *p;
|
|
|
7509 int rr;
|
|
500
|
7510 int retval = OK;
|
|
236
|
7511
|
|
310
|
7512 fd = mch_fopen((char *)fname, "w");
|
|
300
|
7513 if (fd == NULL)
|
|
236
|
7514 {
|
|
|
7515 EMSG2(_(e_notopen), fname);
|
|
500
|
7516 return FAIL;
|
|
236
|
7517 }
|
|
|
7518
|
|
483
|
7519 /* <HEADER>: <fileID> <versionnr> */
|
|
300
|
7520 /* <fileID> */
|
|
|
7521 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
500
|
7522 {
|
|
300
|
7523 EMSG(_(e_write));
|
|
500
|
7524 retval = FAIL;
|
|
|
7525 }
|
|
483
|
7526 putc(VIMSPELLVERSION, fd); /* <versionnr> */
|
|
|
7527
|
|
|
7528 /*
|
|
|
7529 * <SECTIONS>: <section> ... <sectionend>
|
|
|
7530 */
|
|
|
7531
|
|
714
|
7532 /* SN_INFO: <infotext> */
|
|
|
7533 if (spin->si_info != NULL)
|
|
|
7534 {
|
|
|
7535 putc(SN_INFO, fd); /* <sectionID> */
|
|
|
7536 putc(0, fd); /* <sectionflags> */
|
|
|
7537
|
|
|
7538 i = STRLEN(spin->si_info);
|
|
|
7539 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */
|
|
|
7540 fwrite(spin->si_info, (size_t)i, (size_t)1, fd); /* <infotext> */
|
|
|
7541 }
|
|
|
7542
|
|
483
|
7543 /* SN_REGION: <regionname> ...
|
|
|
7544 * Write the region names only if there is more than one. */
|
|
316
|
7545 if (spin->si_region_count > 1)
|
|
236
|
7546 {
|
|
483
|
7547 putc(SN_REGION, fd); /* <sectionID> */
|
|
|
7548 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7549 l = spin->si_region_count * 2;
|
|
|
7550 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7551 fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd);
|
|
|
7552 /* <regionname> ... */
|
|
316
|
7553 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
7554 }
|
|
|
7555 else
|
|
300
|
7556 regionmask = 0;
|
|
483
|
7557
|
|
|
7558 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars>
|
|
|
7559 *
|
|
|
7560 * The table with character flags and the table for case folding.
|
|
|
7561 * This makes sure the same characters are recognized as word characters
|
|
|
7562 * when generating an when using a spell file.
|
|
260
|
7563 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
7564 * 'encoding'.
|
|
|
7565 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
7566 * the table (avoids that it conflicts). File is shorter too.
|
|
|
7567 */
|
|
483
|
7568 if (!spin->si_ascii && !spin->si_add)
|
|
|
7569 {
|
|
|
7570 char_u folchars[128 * 8];
|
|
|
7571 int flags;
|
|
|
7572
|
|
485
|
7573 putc(SN_CHARFLAGS, fd); /* <sectionID> */
|
|
483
|
7574 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7575
|
|
|
7576 /* Form the <folchars> string first, we need to know its length. */
|
|
|
7577 l = 0;
|
|
|
7578 for (i = 128; i < 256; ++i)
|
|
|
7579 {
|
|
|
7580 #ifdef FEAT_MBYTE
|
|
|
7581 if (has_mbyte)
|
|
|
7582 l += mb_char2bytes(spelltab.st_fold[i], folchars + l);
|
|
|
7583 else
|
|
|
7584 #endif
|
|
|
7585 folchars[l++] = spelltab.st_fold[i];
|
|
|
7586 }
|
|
|
7587 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */
|
|
|
7588
|
|
|
7589 fputc(128, fd); /* <charflagslen> */
|
|
|
7590 for (i = 128; i < 256; ++i)
|
|
|
7591 {
|
|
|
7592 flags = 0;
|
|
|
7593 if (spelltab.st_isw[i])
|
|
|
7594 flags |= CF_WORD;
|
|
|
7595 if (spelltab.st_isu[i])
|
|
|
7596 flags |= CF_UPPER;
|
|
|
7597 fputc(flags, fd); /* <charflags> */
|
|
|
7598 }
|
|
|
7599
|
|
|
7600 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */
|
|
|
7601 fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */
|
|
|
7602 }
|
|
|
7603
|
|
|
7604 /* SN_MIDWORD: <midword> */
|
|
|
7605 if (spin->si_midword != NULL)
|
|
|
7606 {
|
|
|
7607 putc(SN_MIDWORD, fd); /* <sectionID> */
|
|
|
7608 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7609
|
|
366
|
7610 i = STRLEN(spin->si_midword);
|
|
483
|
7611 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */
|
|
366
|
7612 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
7613 }
|
|
|
7614
|
|
483
|
7615 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */
|
|
|
7616 if (spin->si_prefcond.ga_len > 0)
|
|
|
7617 {
|
|
|
7618 putc(SN_PREFCOND, fd); /* <sectionID> */
|
|
|
7619 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7620
|
|
|
7621 l = write_spell_prefcond(NULL, &spin->si_prefcond);
|
|
|
7622 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7623
|
|
|
7624 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
7625 }
|
|
|
7626
|
|
|
7627 /* SN_REP: <repcount> <rep> ...
|
|
625
|
7628 * SN_SAL: <salflags> <salcount> <sal> ...
|
|
|
7629 * SN_REPSAL: <repcount> <rep> ... */
|
|
323
|
7630
|
|
483
|
7631 /* round 1: SN_REP section
|
|
625
|
7632 * round 2: SN_SAL section (unless SN_SOFO is used)
|
|
|
7633 * round 3: SN_REPSAL section */
|
|
|
7634 for (round = 1; round <= 3; ++round)
|
|
323
|
7635 {
|
|
|
7636 if (round == 1)
|
|
|
7637 gap = &spin->si_rep;
|
|
625
|
7638 else if (round == 2)
|
|
|
7639 {
|
|
|
7640 /* Don't write SN_SAL when using a SN_SOFO section */
|
|
|
7641 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
7642 continue;
|
|
|
7643 gap = &spin->si_sal;
|
|
483
|
7644 }
|
|
323
|
7645 else
|
|
625
|
7646 gap = &spin->si_repsal;
|
|
|
7647
|
|
|
7648 /* Don't write the section if there are no items. */
|
|
|
7649 if (gap->ga_len == 0)
|
|
|
7650 continue;
|
|
|
7651
|
|
|
7652 /* Sort the REP/REPSAL items. */
|
|
|
7653 if (round != 2)
|
|
|
7654 qsort(gap->ga_data, (size_t)gap->ga_len,
|
|
|
7655 sizeof(fromto_T), rep_compare);
|
|
|
7656
|
|
|
7657 i = round == 1 ? SN_REP : (round == 2 ? SN_SAL : SN_REPSAL);
|
|
|
7658 putc(i, fd); /* <sectionID> */
|
|
483
|
7659
|
|
|
7660 /* This is for making suggestions, section is not required. */
|
|
|
7661 putc(0, fd); /* <sectionflags> */
|
|
|
7662
|
|
|
7663 /* Compute the length of what follows. */
|
|
|
7664 l = 2; /* count <repcount> or <salcount> */
|
|
|
7665 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7666 {
|
|
|
7667 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
7668 l += 1 + STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */
|
|
|
7669 l += 1 + STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */
|
|
|
7670 }
|
|
|
7671 if (round == 2)
|
|
|
7672 ++l; /* count <salflags> */
|
|
|
7673 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7674
|
|
|
7675 if (round == 2)
|
|
|
7676 {
|
|
323
|
7677 i = 0;
|
|
|
7678 if (spin->si_followup)
|
|
|
7679 i |= SAL_F0LLOWUP;
|
|
|
7680 if (spin->si_collapse)
|
|
|
7681 i |= SAL_COLLAPSE;
|
|
|
7682 if (spin->si_rem_accents)
|
|
|
7683 i |= SAL_REM_ACCENTS;
|
|
|
7684 putc(i, fd); /* <salflags> */
|
|
|
7685 }
|
|
|
7686
|
|
|
7687 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
7688 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7689 {
|
|
|
7690 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
7691 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
7692 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
7693 for (rr = 1; rr <= 2; ++rr)
|
|
|
7694 {
|
|
|
7695 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
7696 l = STRLEN(p);
|
|
|
7697 putc(l, fd);
|
|
|
7698 fwrite(p, l, (size_t)1, fd);
|
|
|
7699 }
|
|
|
7700 }
|
|
483
|
7701
|
|
|
7702 }
|
|
|
7703
|
|
|
7704 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
7705 * This is for making suggestions, section is not required. */
|
|
375
|
7706 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
7707 {
|
|
483
|
7708 putc(SN_SOFO, fd); /* <sectionID> */
|
|
|
7709 putc(0, fd); /* <sectionflags> */
|
|
375
|
7710
|
|
|
7711 l = STRLEN(spin->si_sofofr);
|
|
483
|
7712 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4);
|
|
|
7713 /* <sectionlen> */
|
|
|
7714
|
|
|
7715 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */
|
|
|
7716 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */
|
|
375
|
7717
|
|
|
7718 l = STRLEN(spin->si_sofoto);
|
|
483
|
7719 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */
|
|
|
7720 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */
|
|
|
7721 }
|
|
|
7722
|
|
625
|
7723 /* SN_WORDS: <word> ...
|
|
|
7724 * This is for making suggestions, section is not required. */
|
|
|
7725 if (spin->si_commonwords.ht_used > 0)
|
|
|
7726 {
|
|
|
7727 putc(SN_WORDS, fd); /* <sectionID> */
|
|
|
7728 putc(0, fd); /* <sectionflags> */
|
|
|
7729
|
|
|
7730 /* round 1: count the bytes
|
|
|
7731 * round 2: write the bytes */
|
|
|
7732 for (round = 1; round <= 2; ++round)
|
|
|
7733 {
|
|
|
7734 int todo;
|
|
|
7735 int len = 0;
|
|
|
7736 hashitem_T *hi;
|
|
|
7737
|
|
|
7738 todo = spin->si_commonwords.ht_used;
|
|
|
7739 for (hi = spin->si_commonwords.ht_array; todo > 0; ++hi)
|
|
|
7740 if (!HASHITEM_EMPTY(hi))
|
|
|
7741 {
|
|
|
7742 l = STRLEN(hi->hi_key) + 1;
|
|
|
7743 len += l;
|
|
|
7744 if (round == 2) /* <word> */
|
|
|
7745 fwrite(hi->hi_key, (size_t)l, (size_t)1, fd);
|
|
|
7746 --todo;
|
|
|
7747 }
|
|
|
7748 if (round == 1)
|
|
|
7749 put_bytes(fd, (long_u)len, 4); /* <sectionlen> */
|
|
|
7750 }
|
|
|
7751 }
|
|
|
7752
|
|
483
|
7753 /* SN_MAP: <mapstr>
|
|
|
7754 * This is for making suggestions, section is not required. */
|
|
|
7755 if (spin->si_map.ga_len > 0)
|
|
|
7756 {
|
|
|
7757 putc(SN_MAP, fd); /* <sectionID> */
|
|
|
7758 putc(0, fd); /* <sectionflags> */
|
|
|
7759 l = spin->si_map.ga_len;
|
|
|
7760 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7761 fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd);
|
|
|
7762 /* <mapstr> */
|
|
|
7763 }
|
|
|
7764
|
|
625
|
7765 /* SN_SUGFILE: <timestamp>
|
|
|
7766 * This is used to notify that a .sug file may be available and at the
|
|
|
7767 * same time allows for checking that a .sug file that is found matches
|
|
|
7768 * with this .spl file. That's because the word numbers must be exactly
|
|
|
7769 * right. */
|
|
|
7770 if (!spin->si_nosugfile
|
|
|
7771 && (spin->si_sal.ga_len > 0
|
|
|
7772 || (spin->si_sofofr != NULL && spin->si_sofoto != NULL)))
|
|
|
7773 {
|
|
|
7774 putc(SN_SUGFILE, fd); /* <sectionID> */
|
|
|
7775 putc(0, fd); /* <sectionflags> */
|
|
|
7776 put_bytes(fd, (long_u)8, 4); /* <sectionlen> */
|
|
|
7777
|
|
|
7778 /* Set si_sugtime and write it to the file. */
|
|
|
7779 spin->si_sugtime = time(NULL);
|
|
|
7780 put_sugtime(spin, fd); /* <timestamp> */
|
|
|
7781 }
|
|
|
7782
|
|
698
|
7783 /* SN_NOSPLITSUGS: nothing
|
|
|
7784 * This is used to notify that no suggestions with word splits are to be
|
|
|
7785 * made. */
|
|
|
7786 if (spin->si_nosplitsugs)
|
|
|
7787 {
|
|
|
7788 putc(SN_NOSPLITSUGS, fd); /* <sectionID> */
|
|
|
7789 putc(0, fd); /* <sectionflags> */
|
|
|
7790 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */
|
|
|
7791 }
|
|
|
7792
|
|
483
|
7793 /* SN_COMPOUND: compound info.
|
|
|
7794 * We don't mark it required, when not supported all compound words will
|
|
|
7795 * be bad words. */
|
|
|
7796 if (spin->si_compflags != NULL)
|
|
|
7797 {
|
|
|
7798 putc(SN_COMPOUND, fd); /* <sectionID> */
|
|
|
7799 putc(0, fd); /* <sectionflags> */
|
|
|
7800
|
|
|
7801 l = STRLEN(spin->si_compflags);
|
|
|
7802 put_bytes(fd, (long_u)(l + 3), 4); /* <sectionlen> */
|
|
|
7803 putc(spin->si_compmax, fd); /* <compmax> */
|
|
|
7804 putc(spin->si_compminlen, fd); /* <compminlen> */
|
|
|
7805 putc(spin->si_compsylmax, fd); /* <compsylmax> */
|
|
|
7806 /* <compflags> */
|
|
|
7807 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
7808 }
|
|
|
7809
|
|
492
|
7810 /* SN_NOBREAK: NOBREAK flag */
|
|
|
7811 if (spin->si_nobreak)
|
|
|
7812 {
|
|
|
7813 putc(SN_NOBREAK, fd); /* <sectionID> */
|
|
|
7814 putc(0, fd); /* <sectionflags> */
|
|
|
7815
|
|
|
7816 /* It's empty, the precense of the section flags the feature. */
|
|
|
7817 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */
|
|
|
7818 }
|
|
|
7819
|
|
483
|
7820 /* SN_SYLLABLE: syllable info.
|
|
|
7821 * We don't mark it required, when not supported syllables will not be
|
|
|
7822 * counted. */
|
|
|
7823 if (spin->si_syllable != NULL)
|
|
|
7824 {
|
|
|
7825 putc(SN_SYLLABLE, fd); /* <sectionID> */
|
|
|
7826 putc(0, fd); /* <sectionflags> */
|
|
|
7827
|
|
|
7828 l = STRLEN(spin->si_syllable);
|
|
|
7829 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7830 fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); /* <syllable> */
|
|
|
7831 }
|
|
|
7832
|
|
|
7833 /* end of <SECTIONS> */
|
|
|
7834 putc(SN_END, fd); /* <sectionend> */
|
|
|
7835
|
|
302
|
7836
|
|
236
|
7837 /*
|
|
339
|
7838 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
7839 */
|
|
323
|
7840 spin->si_memtot = 0;
|
|
339
|
7841 for (round = 1; round <= 3; ++round)
|
|
236
|
7842 {
|
|
339
|
7843 if (round == 1)
|
|
469
|
7844 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
7845 else if (round == 2)
|
|
469
|
7846 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
7847 else
|
|
469
|
7848 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
7849
|
|
346
|
7850 /* Clear the index and wnode fields in the tree. */
|
|
|
7851 clear_node(tree);
|
|
|
7852
|
|
300
|
7853 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
7854 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
7855 * nodes. */
|
|
346
|
7856 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
7857
|
|
300
|
7858 /* number of nodes in 4 bytes */
|
|
|
7859 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
7860 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
7861
|
|
300
|
7862 /* Write the nodes. */
|
|
346
|
7863 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
7864 }
|
|
|
7865
|
|
500
|
7866 /* Write another byte to check for errors. */
|
|
|
7867 if (putc(0, fd) == EOF)
|
|
|
7868 retval = FAIL;
|
|
|
7869
|
|
|
7870 if (fclose(fd) == EOF)
|
|
|
7871 retval = FAIL;
|
|
|
7872
|
|
|
7873 return retval;
|
|
236
|
7874 }
|
|
|
7875
|
|
|
7876 /*
|
|
346
|
7877 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
7878 * children. This is needed because they are a union with other items to save
|
|
|
7879 * space.
|
|
|
7880 */
|
|
|
7881 static void
|
|
|
7882 clear_node(node)
|
|
|
7883 wordnode_T *node;
|
|
|
7884 {
|
|
|
7885 wordnode_T *np;
|
|
|
7886
|
|
|
7887 if (node != NULL)
|
|
|
7888 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7889 {
|
|
|
7890 np->wn_u1.index = 0;
|
|
|
7891 np->wn_u2.wnode = NULL;
|
|
|
7892
|
|
|
7893 if (np->wn_byte != NUL)
|
|
|
7894 clear_node(np->wn_child);
|
|
|
7895 }
|
|
|
7896 }
|
|
|
7897
|
|
|
7898
|
|
|
7899 /*
|
|
300
|
7900 * Dump a word tree at node "node".
|
|
|
7901 *
|
|
|
7902 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
7903 * byte recursively write the children.
|
|
|
7904 *
|
|
625
|
7905 * NOTE: The code here must match the code in read_tree_node(), since
|
|
|
7906 * assumptions are made about the indexes (so that we don't have to write them
|
|
|
7907 * in the file).
|
|
236
|
7908 *
|
|
300
|
7909 * Returns the number of nodes used.
|
|
236
|
7910 */
|
|
300
|
7911 static int
|
|
346
|
7912 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
7913 FILE *fd; /* NULL when only counting */
|
|
300
|
7914 wordnode_T *node;
|
|
|
7915 int index;
|
|
|
7916 int regionmask;
|
|
339
|
7917 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
7918 {
|
|
300
|
7919 int newindex = index;
|
|
|
7920 int siblingcount = 0;
|
|
|
7921 wordnode_T *np;
|
|
236
|
7922 int flags;
|
|
300
|
7923
|
|
|
7924 /* If "node" is zero the tree is empty. */
|
|
|
7925 if (node == NULL)
|
|
|
7926 return 0;
|
|
|
7927
|
|
|
7928 /* Store the index where this node is written. */
|
|
346
|
7929 node->wn_u1.index = index;
|
|
236
|
7930
|
|
300
|
7931 /* Count the number of siblings. */
|
|
|
7932 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7933 ++siblingcount;
|
|
236
|
7934
|
|
300
|
7935 /* Write the sibling count. */
|
|
|
7936 if (fd != NULL)
|
|
|
7937 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
7938
|
|
300
|
7939 /* Write each sibling byte and optionally extra info. */
|
|
|
7940 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
7941 {
|
|
300
|
7942 if (np->wn_byte == 0)
|
|
|
7943 {
|
|
|
7944 if (fd != NULL)
|
|
|
7945 {
|
|
339
|
7946 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
7947 if (prefixtree)
|
|
300
|
7948 {
|
|
481
|
7949 /* In PREFIXTREE write the required affixID and the
|
|
390
|
7950 * associated condition nr (stored in wn_region). The
|
|
|
7951 * byte value is misused to store the "rare" and "not
|
|
|
7952 * combining" flags */
|
|
455
|
7953 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
7954 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7955 else
|
|
|
7956 {
|
|
390
|
7957 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
7958 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
7959 }
|
|
481
|
7960 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7961 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
7962 }
|
|
|
7963 else
|
|
|
7964 {
|
|
339
|
7965 /* For word trees we write the flag/region items. */
|
|
|
7966 flags = np->wn_flags;
|
|
|
7967 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
7968 flags |= WF_REGION;
|
|
481
|
7969 if (np->wn_affixID != 0)
|
|
|
7970 flags |= WF_AFX;
|
|
339
|
7971 if (flags == 0)
|
|
|
7972 {
|
|
|
7973 /* word without flags or region */
|
|
|
7974 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7975 }
|
|
|
7976 else
|
|
|
7977 {
|
|
390
|
7978 if (np->wn_flags >= 0x100)
|
|
|
7979 {
|
|
|
7980 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
7981 putc(flags, fd); /* <flags> */
|
|
|
7982 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
7983 }
|
|
|
7984 else
|
|
|
7985 {
|
|
|
7986 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
7987 putc(flags, fd); /* <flags> */
|
|
|
7988 }
|
|
339
|
7989 if (flags & WF_REGION)
|
|
|
7990 putc(np->wn_region, fd); /* <region> */
|
|
481
|
7991 if (flags & WF_AFX)
|
|
|
7992 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7993 }
|
|
300
|
7994 }
|
|
|
7995 }
|
|
|
7996 }
|
|
|
7997 else
|
|
|
7998 {
|
|
346
|
7999 if (np->wn_child->wn_u1.index != 0
|
|
|
8000 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
8001 {
|
|
|
8002 /* The child is written elsewhere, write the reference. */
|
|
|
8003 if (fd != NULL)
|
|
|
8004 {
|
|
|
8005 putc(BY_INDEX, fd); /* <byte> */
|
|
|
8006 /* <nodeidx> */
|
|
346
|
8007 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
8008 }
|
|
|
8009 }
|
|
346
|
8010 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
8011 /* We will write the child below and give it an index. */
|
|
346
|
8012 np->wn_child->wn_u2.wnode = node;
|
|
236
|
8013
|
|
300
|
8014 if (fd != NULL)
|
|
|
8015 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
8016 {
|
|
|
8017 EMSG(_(e_write));
|
|
|
8018 return 0;
|
|
|
8019 }
|
|
|
8020 }
|
|
236
|
8021 }
|
|
|
8022
|
|
300
|
8023 /* Space used in the array when reading: one for each sibling and one for
|
|
|
8024 * the count. */
|
|
|
8025 newindex += siblingcount + 1;
|
|
249
|
8026
|
|
300
|
8027 /* Recursively dump the children of each sibling. */
|
|
|
8028 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
8029 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
8030 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
8031 prefixtree);
|
|
249
|
8032
|
|
300
|
8033 return newindex;
|
|
236
|
8034 }
|
|
|
8035
|
|
|
8036
|
|
|
8037 /*
|
|
310
|
8038 * ":mkspell [-ascii] outfile infile ..."
|
|
|
8039 * ":mkspell [-ascii] addfile"
|
|
236
|
8040 */
|
|
|
8041 void
|
|
|
8042 ex_mkspell(eap)
|
|
|
8043 exarg_T *eap;
|
|
|
8044 {
|
|
|
8045 int fcount;
|
|
|
8046 char_u **fnames;
|
|
310
|
8047 char_u *arg = eap->arg;
|
|
|
8048 int ascii = FALSE;
|
|
|
8049
|
|
|
8050 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
8051 {
|
|
|
8052 ascii = TRUE;
|
|
|
8053 arg = skipwhite(arg + 6);
|
|
|
8054 }
|
|
|
8055
|
|
|
8056 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
8057 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
8058 {
|
|
323
|
8059 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
8060 FreeWild(fcount, fnames);
|
|
|
8061 }
|
|
|
8062 }
|
|
|
8063
|
|
|
8064 /*
|
|
625
|
8065 * Create the .sug file.
|
|
|
8066 * Uses the soundfold info in "spin".
|
|
|
8067 * Writes the file with the name "wfname", with ".spl" changed to ".sug".
|
|
|
8068 */
|
|
|
8069 static void
|
|
|
8070 spell_make_sugfile(spin, wfname)
|
|
|
8071 spellinfo_T *spin;
|
|
|
8072 char_u *wfname;
|
|
|
8073 {
|
|
|
8074 char_u fname[MAXPATHL];
|
|
|
8075 int len;
|
|
|
8076 slang_T *slang;
|
|
|
8077 int free_slang = FALSE;
|
|
|
8078
|
|
|
8079 /*
|
|
|
8080 * Read back the .spl file that was written. This fills the required
|
|
|
8081 * info for soundfolding. This also uses less memory than the
|
|
|
8082 * pointer-linked version of the trie. And it avoids having two versions
|
|
|
8083 * of the code for the soundfolding stuff.
|
|
|
8084 * It might have been done already by spell_reload_one().
|
|
|
8085 */
|
|
|
8086 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
8087 if (fullpathcmp(wfname, slang->sl_fname, FALSE) == FPC_SAME)
|
|
|
8088 break;
|
|
|
8089 if (slang == NULL)
|
|
|
8090 {
|
|
|
8091 spell_message(spin, (char_u *)_("Reading back spell file..."));
|
|
|
8092 slang = spell_load_file(wfname, NULL, NULL, FALSE);
|
|
|
8093 if (slang == NULL)
|
|
|
8094 return;
|
|
|
8095 free_slang = TRUE;
|
|
|
8096 }
|
|
|
8097
|
|
|
8098 /*
|
|
|
8099 * Clear the info in "spin" that is used.
|
|
|
8100 */
|
|
|
8101 spin->si_blocks = NULL;
|
|
|
8102 spin->si_blocks_cnt = 0;
|
|
|
8103 spin->si_compress_cnt = 0; /* will stay at 0 all the time*/
|
|
|
8104 spin->si_free_count = 0;
|
|
|
8105 spin->si_first_free = NULL;
|
|
|
8106 spin->si_foldwcount = 0;
|
|
|
8107
|
|
|
8108 /*
|
|
|
8109 * Go through the trie of good words, soundfold each word and add it to
|
|
|
8110 * the soundfold trie.
|
|
|
8111 */
|
|
|
8112 spell_message(spin, (char_u *)_("Performing soundfolding..."));
|
|
|
8113 if (sug_filltree(spin, slang) == FAIL)
|
|
|
8114 goto theend;
|
|
|
8115
|
|
|
8116 /*
|
|
|
8117 * Create the table which links each soundfold word with a list of the
|
|
|
8118 * good words it may come from. Creates buffer "spin->si_spellbuf".
|
|
|
8119 * This also removes the wordnr from the NUL byte entries to make
|
|
|
8120 * compression possible.
|
|
|
8121 */
|
|
|
8122 if (sug_maketable(spin) == FAIL)
|
|
|
8123 goto theend;
|
|
|
8124
|
|
|
8125 smsg((char_u *)_("Number of words after soundfolding: %ld"),
|
|
|
8126 (long)spin->si_spellbuf->b_ml.ml_line_count);
|
|
|
8127
|
|
|
8128 /*
|
|
|
8129 * Compress the soundfold trie.
|
|
|
8130 */
|
|
|
8131 spell_message(spin, (char_u *)_(msg_compressing));
|
|
|
8132 wordtree_compress(spin, spin->si_foldroot);
|
|
|
8133
|
|
|
8134 /*
|
|
|
8135 * Write the .sug file.
|
|
|
8136 * Make the file name by changing ".spl" to ".sug".
|
|
|
8137 */
|
|
|
8138 STRCPY(fname, wfname);
|
|
|
8139 len = STRLEN(fname);
|
|
|
8140 fname[len - 2] = 'u';
|
|
|
8141 fname[len - 1] = 'g';
|
|
|
8142 sug_write(spin, fname);
|
|
|
8143
|
|
|
8144 theend:
|
|
|
8145 if (free_slang)
|
|
|
8146 slang_free(slang);
|
|
|
8147 free_blocks(spin->si_blocks);
|
|
|
8148 close_spellbuf(spin->si_spellbuf);
|
|
|
8149 }
|
|
|
8150
|
|
|
8151 /*
|
|
|
8152 * Build the soundfold trie for language "slang".
|
|
|
8153 */
|
|
|
8154 static int
|
|
|
8155 sug_filltree(spin, slang)
|
|
|
8156 spellinfo_T *spin;
|
|
|
8157 slang_T *slang;
|
|
|
8158 {
|
|
|
8159 char_u *byts;
|
|
|
8160 idx_T *idxs;
|
|
|
8161 int depth;
|
|
|
8162 idx_T arridx[MAXWLEN];
|
|
|
8163 int curi[MAXWLEN];
|
|
|
8164 char_u tword[MAXWLEN];
|
|
|
8165 char_u tsalword[MAXWLEN];
|
|
|
8166 int c;
|
|
|
8167 idx_T n;
|
|
|
8168 unsigned words_done = 0;
|
|
|
8169 int wordcount[MAXWLEN];
|
|
|
8170
|
|
|
8171 /* We use si_foldroot for the souldfolded trie. */
|
|
|
8172 spin->si_foldroot = wordtree_alloc(spin);
|
|
|
8173 if (spin->si_foldroot == NULL)
|
|
|
8174 return FAIL;
|
|
|
8175
|
|
|
8176 /* let tree_add_word() know we're adding to the soundfolded tree */
|
|
|
8177 spin->si_sugtree = TRUE;
|
|
|
8178
|
|
|
8179 /*
|
|
|
8180 * Go through the whole case-folded tree, soundfold each word and put it
|
|
|
8181 * in the trie.
|
|
|
8182 */
|
|
|
8183 byts = slang->sl_fbyts;
|
|
|
8184 idxs = slang->sl_fidxs;
|
|
|
8185
|
|
|
8186 arridx[0] = 0;
|
|
|
8187 curi[0] = 1;
|
|
|
8188 wordcount[0] = 0;
|
|
|
8189
|
|
|
8190 depth = 0;
|
|
|
8191 while (depth >= 0 && !got_int)
|
|
|
8192 {
|
|
|
8193 if (curi[depth] > byts[arridx[depth]])
|
|
|
8194 {
|
|
|
8195 /* Done all bytes at this node, go up one level. */
|
|
|
8196 idxs[arridx[depth]] = wordcount[depth];
|
|
|
8197 if (depth > 0)
|
|
|
8198 wordcount[depth - 1] += wordcount[depth];
|
|
|
8199
|
|
|
8200 --depth;
|
|
|
8201 line_breakcheck();
|
|
|
8202 }
|
|
|
8203 else
|
|
|
8204 {
|
|
|
8205
|
|
|
8206 /* Do one more byte at this node. */
|
|
|
8207 n = arridx[depth] + curi[depth];
|
|
|
8208 ++curi[depth];
|
|
|
8209
|
|
|
8210 c = byts[n];
|
|
|
8211 if (c == 0)
|
|
|
8212 {
|
|
|
8213 /* Sound-fold the word. */
|
|
|
8214 tword[depth] = NUL;
|
|
|
8215 spell_soundfold(slang, tword, TRUE, tsalword);
|
|
|
8216
|
|
|
8217 /* We use the "flags" field for the MSB of the wordnr,
|
|
|
8218 * "region" for the LSB of the wordnr. */
|
|
|
8219 if (tree_add_word(spin, tsalword, spin->si_foldroot,
|
|
|
8220 words_done >> 16, words_done & 0xffff,
|
|
|
8221 0) == FAIL)
|
|
|
8222 return FAIL;
|
|
|
8223
|
|
|
8224 ++words_done;
|
|
|
8225 ++wordcount[depth];
|
|
|
8226
|
|
|
8227 /* Reset the block count each time to avoid compression
|
|
|
8228 * kicking in. */
|
|
|
8229 spin->si_blocks_cnt = 0;
|
|
|
8230
|
|
|
8231 /* Skip over any other NUL bytes (same word with different
|
|
|
8232 * flags). */
|
|
|
8233 while (byts[n + 1] == 0)
|
|
|
8234 {
|
|
|
8235 ++n;
|
|
|
8236 ++curi[depth];
|
|
|
8237 }
|
|
|
8238 }
|
|
|
8239 else
|
|
|
8240 {
|
|
|
8241 /* Normal char, go one level deeper. */
|
|
|
8242 tword[depth++] = c;
|
|
|
8243 arridx[depth] = idxs[n];
|
|
|
8244 curi[depth] = 1;
|
|
|
8245 wordcount[depth] = 0;
|
|
|
8246 }
|
|
|
8247 }
|
|
|
8248 }
|
|
|
8249
|
|
|
8250 smsg((char_u *)_("Total number of words: %d"), words_done);
|
|
|
8251
|
|
|
8252 return OK;
|
|
|
8253 }
|
|
|
8254
|
|
|
8255 /*
|
|
|
8256 * Make the table that links each word in the soundfold trie to the words it
|
|
|
8257 * can be produced from.
|
|
|
8258 * This is not unlike lines in a file, thus use a memfile to be able to access
|
|
|
8259 * the table efficiently.
|
|
|
8260 * Returns FAIL when out of memory.
|
|
|
8261 */
|
|
|
8262 static int
|
|
|
8263 sug_maketable(spin)
|
|
|
8264 spellinfo_T *spin;
|
|
|
8265 {
|
|
|
8266 garray_T ga;
|
|
|
8267 int res = OK;
|
|
|
8268
|
|
|
8269 /* Allocate a buffer, open a memline for it and create the swap file
|
|
|
8270 * (uses a temp file, not a .swp file). */
|
|
|
8271 spin->si_spellbuf = open_spellbuf();
|
|
|
8272 if (spin->si_spellbuf == NULL)
|
|
|
8273 return FAIL;
|
|
|
8274
|
|
|
8275 /* Use a buffer to store the line info, avoids allocating many small
|
|
|
8276 * pieces of memory. */
|
|
|
8277 ga_init2(&ga, 1, 100);
|
|
|
8278
|
|
|
8279 /* recursively go through the tree */
|
|
|
8280 if (sug_filltable(spin, spin->si_foldroot->wn_sibling, 0, &ga) == -1)
|
|
|
8281 res = FAIL;
|
|
|
8282
|
|
|
8283 ga_clear(&ga);
|
|
|
8284 return res;
|
|
|
8285 }
|
|
|
8286
|
|
|
8287 /*
|
|
|
8288 * Fill the table for one node and its children.
|
|
|
8289 * Returns the wordnr at the start of the node.
|
|
|
8290 * Returns -1 when out of memory.
|
|
|
8291 */
|
|
|
8292 static int
|
|
|
8293 sug_filltable(spin, node, startwordnr, gap)
|
|
|
8294 spellinfo_T *spin;
|
|
|
8295 wordnode_T *node;
|
|
|
8296 int startwordnr;
|
|
|
8297 garray_T *gap; /* place to store line of numbers */
|
|
|
8298 {
|
|
|
8299 wordnode_T *p, *np;
|
|
|
8300 int wordnr = startwordnr;
|
|
|
8301 int nr;
|
|
|
8302 int prev_nr;
|
|
|
8303
|
|
|
8304 for (p = node; p != NULL; p = p->wn_sibling)
|
|
|
8305 {
|
|
|
8306 if (p->wn_byte == NUL)
|
|
|
8307 {
|
|
|
8308 gap->ga_len = 0;
|
|
|
8309 prev_nr = 0;
|
|
|
8310 for (np = p; np != NULL && np->wn_byte == NUL; np = np->wn_sibling)
|
|
|
8311 {
|
|
|
8312 if (ga_grow(gap, 10) == FAIL)
|
|
|
8313 return -1;
|
|
|
8314
|
|
|
8315 nr = (np->wn_flags << 16) + (np->wn_region & 0xffff);
|
|
|
8316 /* Compute the offset from the previous nr and store the
|
|
|
8317 * offset in a way that it takes a minimum number of bytes.
|
|
|
8318 * It's a bit like utf-8, but without the need to mark
|
|
|
8319 * following bytes. */
|
|
|
8320 nr -= prev_nr;
|
|
|
8321 prev_nr += nr;
|
|
|
8322 gap->ga_len += offset2bytes(nr,
|
|
|
8323 (char_u *)gap->ga_data + gap->ga_len);
|
|
|
8324 }
|
|
|
8325
|
|
|
8326 /* add the NUL byte */
|
|
|
8327 ((char_u *)gap->ga_data)[gap->ga_len++] = NUL;
|
|
|
8328
|
|
|
8329 if (ml_append_buf(spin->si_spellbuf, (linenr_T)wordnr,
|
|
|
8330 gap->ga_data, gap->ga_len, TRUE) == FAIL)
|
|
|
8331 return -1;
|
|
|
8332 ++wordnr;
|
|
|
8333
|
|
|
8334 /* Remove extra NUL entries, we no longer need them. We don't
|
|
|
8335 * bother freeing the nodes, the won't be reused anyway. */
|
|
|
8336 while (p->wn_sibling != NULL && p->wn_sibling->wn_byte == NUL)
|
|
|
8337 p->wn_sibling = p->wn_sibling->wn_sibling;
|
|
|
8338
|
|
|
8339 /* Clear the flags on the remaining NUL node, so that compression
|
|
|
8340 * works a lot better. */
|
|
|
8341 p->wn_flags = 0;
|
|
|
8342 p->wn_region = 0;
|
|
|
8343 }
|
|
|
8344 else
|
|
|
8345 {
|
|
|
8346 wordnr = sug_filltable(spin, p->wn_child, wordnr, gap);
|
|
|
8347 if (wordnr == -1)
|
|
|
8348 return -1;
|
|
|
8349 }
|
|
|
8350 }
|
|
|
8351 return wordnr;
|
|
|
8352 }
|
|
|
8353
|
|
|
8354 /*
|
|
|
8355 * Convert an offset into a minimal number of bytes.
|
|
|
8356 * Similar to utf_char2byters, but use 8 bits in followup bytes and avoid NUL
|
|
|
8357 * bytes.
|
|
|
8358 */
|
|
|
8359 static int
|
|
|
8360 offset2bytes(nr, buf)
|
|
|
8361 int nr;
|
|
|
8362 char_u *buf;
|
|
|
8363 {
|
|
|
8364 int rem;
|
|
|
8365 int b1, b2, b3, b4;
|
|
|
8366
|
|
|
8367 /* Split the number in parts of base 255. We need to avoid NUL bytes. */
|
|
|
8368 b1 = nr % 255 + 1;
|
|
|
8369 rem = nr / 255;
|
|
|
8370 b2 = rem % 255 + 1;
|
|
|
8371 rem = rem / 255;
|
|
|
8372 b3 = rem % 255 + 1;
|
|
|
8373 b4 = rem / 255 + 1;
|
|
|
8374
|
|
|
8375 if (b4 > 1 || b3 > 0x1f) /* 4 bytes */
|
|
|
8376 {
|
|
|
8377 buf[0] = 0xe0 + b4;
|
|
|
8378 buf[1] = b3;
|
|
|
8379 buf[2] = b2;
|
|
|
8380 buf[3] = b1;
|
|
|
8381 return 4;
|
|
|
8382 }
|
|
|
8383 if (b3 > 1 || b2 > 0x3f ) /* 3 bytes */
|
|
|
8384 {
|
|
|
8385 buf[0] = 0xc0 + b3;
|
|
|
8386 buf[1] = b2;
|
|
|
8387 buf[2] = b1;
|
|
|
8388 return 3;
|
|
|
8389 }
|
|
|
8390 if (b2 > 1 || b1 > 0x7f ) /* 2 bytes */
|
|
|
8391 {
|
|
|
8392 buf[0] = 0x80 + b2;
|
|
|
8393 buf[1] = b1;
|
|
|
8394 return 2;
|
|
|
8395 }
|
|
|
8396 /* 1 byte */
|
|
|
8397 buf[0] = b1;
|
|
|
8398 return 1;
|
|
|
8399 }
|
|
|
8400
|
|
|
8401 /*
|
|
|
8402 * Opposite of offset2bytes().
|
|
|
8403 * "pp" points to the bytes and is advanced over it.
|
|
|
8404 * Returns the offset.
|
|
|
8405 */
|
|
|
8406 static int
|
|
|
8407 bytes2offset(pp)
|
|
|
8408 char_u **pp;
|
|
|
8409 {
|
|
|
8410 char_u *p = *pp;
|
|
|
8411 int nr;
|
|
|
8412 int c;
|
|
|
8413
|
|
|
8414 c = *p++;
|
|
|
8415 if ((c & 0x80) == 0x00) /* 1 byte */
|
|
|
8416 {
|
|
|
8417 nr = c - 1;
|
|
|
8418 }
|
|
|
8419 else if ((c & 0xc0) == 0x80) /* 2 bytes */
|
|
|
8420 {
|
|
|
8421 nr = (c & 0x3f) - 1;
|
|
|
8422 nr = nr * 255 + (*p++ - 1);
|
|
|
8423 }
|
|
|
8424 else if ((c & 0xe0) == 0xc0) /* 3 bytes */
|
|
|
8425 {
|
|
|
8426 nr = (c & 0x1f) - 1;
|
|
|
8427 nr = nr * 255 + (*p++ - 1);
|
|
|
8428 nr = nr * 255 + (*p++ - 1);
|
|
|
8429 }
|
|
|
8430 else /* 4 bytes */
|
|
|
8431 {
|
|
|
8432 nr = (c & 0x0f) - 1;
|
|
|
8433 nr = nr * 255 + (*p++ - 1);
|
|
|
8434 nr = nr * 255 + (*p++ - 1);
|
|
|
8435 nr = nr * 255 + (*p++ - 1);
|
|
|
8436 }
|
|
|
8437
|
|
|
8438 *pp = p;
|
|
|
8439 return nr;
|
|
|
8440 }
|
|
|
8441
|
|
|
8442 /*
|
|
|
8443 * Write the .sug file in "fname".
|
|
|
8444 */
|
|
|
8445 static void
|
|
|
8446 sug_write(spin, fname)
|
|
|
8447 spellinfo_T *spin;
|
|
|
8448 char_u *fname;
|
|
|
8449 {
|
|
|
8450 FILE *fd;
|
|
|
8451 wordnode_T *tree;
|
|
|
8452 int nodecount;
|
|
|
8453 int wcount;
|
|
|
8454 char_u *line;
|
|
|
8455 linenr_T lnum;
|
|
|
8456 int len;
|
|
|
8457
|
|
|
8458 /* Create the file. Note that an existing file is silently overwritten! */
|
|
|
8459 fd = mch_fopen((char *)fname, "w");
|
|
|
8460 if (fd == NULL)
|
|
|
8461 {
|
|
|
8462 EMSG2(_(e_notopen), fname);
|
|
|
8463 return;
|
|
|
8464 }
|
|
|
8465
|
|
|
8466 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
8467 _("Writing suggestion file %s ..."), fname);
|
|
|
8468 spell_message(spin, IObuff);
|
|
|
8469
|
|
|
8470 /*
|
|
|
8471 * <SUGHEADER>: <fileID> <versionnr> <timestamp>
|
|
|
8472 */
|
|
|
8473 if (fwrite(VIMSUGMAGIC, VIMSUGMAGICL, (size_t)1, fd) != 1) /* <fileID> */
|
|
|
8474 {
|
|
|
8475 EMSG(_(e_write));
|
|
|
8476 goto theend;
|
|
|
8477 }
|
|
|
8478 putc(VIMSUGVERSION, fd); /* <versionnr> */
|
|
|
8479
|
|
|
8480 /* Write si_sugtime to the file. */
|
|
|
8481 put_sugtime(spin, fd); /* <timestamp> */
|
|
|
8482
|
|
|
8483 /*
|
|
|
8484 * <SUGWORDTREE>
|
|
|
8485 */
|
|
|
8486 spin->si_memtot = 0;
|
|
|
8487 tree = spin->si_foldroot->wn_sibling;
|
|
|
8488
|
|
|
8489 /* Clear the index and wnode fields in the tree. */
|
|
|
8490 clear_node(tree);
|
|
|
8491
|
|
|
8492 /* Count the number of nodes. Needed to be able to allocate the
|
|
|
8493 * memory when reading the nodes. Also fills in index for shared
|
|
|
8494 * nodes. */
|
|
|
8495 nodecount = put_node(NULL, tree, 0, 0, FALSE);
|
|
|
8496
|
|
|
8497 /* number of nodes in 4 bytes */
|
|
|
8498 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
|
8499 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
|
8500
|
|
|
8501 /* Write the nodes. */
|
|
|
8502 (void)put_node(fd, tree, 0, 0, FALSE);
|
|
|
8503
|
|
|
8504 /*
|
|
|
8505 * <SUGTABLE>: <sugwcount> <sugline> ...
|
|
|
8506 */
|
|
|
8507 wcount = spin->si_spellbuf->b_ml.ml_line_count;
|
|
|
8508 put_bytes(fd, (long_u)wcount, 4); /* <sugwcount> */
|
|
|
8509
|
|
|
8510 for (lnum = 1; lnum <= (linenr_T)wcount; ++lnum)
|
|
|
8511 {
|
|
|
8512 /* <sugline>: <sugnr> ... NUL */
|
|
|
8513 line = ml_get_buf(spin->si_spellbuf, lnum, FALSE);
|
|
|
8514 len = STRLEN(line) + 1;
|
|
|
8515 if (fwrite(line, (size_t)len, (size_t)1, fd) == 0)
|
|
|
8516 {
|
|
|
8517 EMSG(_(e_write));
|
|
|
8518 goto theend;
|
|
|
8519 }
|
|
|
8520 spin->si_memtot += len;
|
|
|
8521 }
|
|
|
8522
|
|
|
8523 /* Write another byte to check for errors. */
|
|
|
8524 if (putc(0, fd) == EOF)
|
|
|
8525 EMSG(_(e_write));
|
|
|
8526
|
|
|
8527 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
8528 _("Estimated runtime memory use: %d bytes"), spin->si_memtot);
|
|
|
8529 spell_message(spin, IObuff);
|
|
|
8530
|
|
|
8531 theend:
|
|
|
8532 /* close the file */
|
|
|
8533 fclose(fd);
|
|
|
8534 }
|
|
|
8535
|
|
|
8536 /*
|
|
|
8537 * Open a spell buffer. This is a nameless buffer that is not in the buffer
|
|
|
8538 * list and only contains text lines. Can use a swapfile to reduce memory
|
|
|
8539 * use.
|
|
|
8540 * Most other fields are invalid! Esp. watch out for string options being
|
|
|
8541 * NULL and there is no undo info.
|
|
|
8542 * Returns NULL when out of memory.
|
|
|
8543 */
|
|
|
8544 static buf_T *
|
|
|
8545 open_spellbuf()
|
|
|
8546 {
|
|
|
8547 buf_T *buf;
|
|
|
8548
|
|
|
8549 buf = (buf_T *)alloc_clear(sizeof(buf_T));
|
|
|
8550 if (buf != NULL)
|
|
|
8551 {
|
|
|
8552 buf->b_spell = TRUE;
|
|
|
8553 buf->b_p_swf = TRUE; /* may create a swap file */
|
|
|
8554 ml_open(buf);
|
|
|
8555 ml_open_file(buf); /* create swap file now */
|
|
|
8556 }
|
|
|
8557 return buf;
|
|
|
8558 }
|
|
|
8559
|
|
|
8560 /*
|
|
|
8561 * Close the buffer used for spell info.
|
|
|
8562 */
|
|
|
8563 static void
|
|
|
8564 close_spellbuf(buf)
|
|
|
8565 buf_T *buf;
|
|
|
8566 {
|
|
|
8567 if (buf != NULL)
|
|
|
8568 {
|
|
|
8569 ml_close(buf, TRUE);
|
|
|
8570 vim_free(buf);
|
|
|
8571 }
|
|
|
8572 }
|
|
|
8573
|
|
|
8574
|
|
|
8575 /*
|
|
310
|
8576 * Create a Vim spell file from one or more word lists.
|
|
|
8577 * "fnames[0]" is the output file name.
|
|
|
8578 * "fnames[fcount - 1]" is the last input file name.
|
|
|
8579 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
8580 * and ".spl" is appended to make the output file name.
|
|
|
8581 */
|
|
|
8582 static void
|
|
323
|
8583 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
8584 int fcount;
|
|
|
8585 char_u **fnames;
|
|
|
8586 int ascii; /* -ascii argument given */
|
|
|
8587 int overwrite; /* overwrite existing output file */
|
|
323
|
8588 int added_word; /* invoked through "zg" */
|
|
310
|
8589 {
|
|
236
|
8590 char_u fname[MAXPATHL];
|
|
|
8591 char_u wfname[MAXPATHL];
|
|
310
|
8592 char_u **innames;
|
|
|
8593 int incount;
|
|
236
|
8594 afffile_T *(afile[8]);
|
|
|
8595 int i;
|
|
|
8596 int len;
|
|
|
8597 struct stat st;
|
|
255
|
8598 int error = FALSE;
|
|
300
|
8599 spellinfo_T spin;
|
|
|
8600
|
|
|
8601 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
8602 spin.si_verbose = !added_word;
|
|
310
|
8603 spin.si_ascii = ascii;
|
|
323
|
8604 spin.si_followup = TRUE;
|
|
|
8605 spin.si_rem_accents = TRUE;
|
|
|
8606 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
625
|
8607 ga_init2(&spin.si_repsal, (int)sizeof(fromto_T), 20);
|
|
323
|
8608 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
8609 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
8610 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
625
|
8611 hash_init(&spin.si_commonwords);
|
|
500
|
8612 spin.si_newcompID = 127; /* start compound ID at first maximum */
|
|
310
|
8613
|
|
|
8614 /* default: fnames[0] is output file, following are input files */
|
|
|
8615 innames = &fnames[1];
|
|
|
8616 incount = fcount - 1;
|
|
|
8617
|
|
|
8618 if (fcount >= 1)
|
|
240
|
8619 {
|
|
310
|
8620 len = STRLEN(fnames[0]);
|
|
|
8621 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
8622 {
|
|
|
8623 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
8624 * "path/en.latin1.add.spl". */
|
|
|
8625 innames = &fnames[0];
|
|
|
8626 incount = 1;
|
|
|
8627 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
8628 }
|
|
366
|
8629 else if (fcount == 1)
|
|
|
8630 {
|
|
|
8631 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
8632 innames = &fnames[0];
|
|
|
8633 incount = 1;
|
|
|
8634 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
8635 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
8636 }
|
|
310
|
8637 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
8638 {
|
|
|
8639 /* Name ends in ".spl", use as the file name. */
|
|
323
|
8640 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
8641 }
|
|
|
8642 else
|
|
|
8643 /* Name should be language, make the file name from it. */
|
|
|
8644 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
8645 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
8646
|
|
|
8647 /* Check for .ascii.spl. */
|
|
|
8648 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
8649 spin.si_ascii = TRUE;
|
|
|
8650
|
|
|
8651 /* Check for .add.spl. */
|
|
|
8652 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
8653 spin.si_add = TRUE;
|
|
240
|
8654 }
|
|
|
8655
|
|
310
|
8656 if (incount <= 0)
|
|
236
|
8657 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
8658 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
8659 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
8660 else if (incount > 8)
|
|
236
|
8661 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
8662 else
|
|
|
8663 {
|
|
|
8664 /* Check for overwriting before doing things that may take a lot of
|
|
|
8665 * time. */
|
|
310
|
8666 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
8667 {
|
|
|
8668 EMSG(_(e_exists));
|
|
310
|
8669 return;
|
|
236
|
8670 }
|
|
310
|
8671 if (mch_isdir(wfname))
|
|
236
|
8672 {
|
|
310
|
8673 EMSG2(_(e_isadir2), wfname);
|
|
|
8674 return;
|
|
236
|
8675 }
|
|
|
8676
|
|
|
8677 /*
|
|
|
8678 * Init the aff and dic pointers.
|
|
|
8679 * Get the region names if there are more than 2 arguments.
|
|
|
8680 */
|
|
310
|
8681 for (i = 0; i < incount; ++i)
|
|
236
|
8682 {
|
|
310
|
8683 afile[i] = NULL;
|
|
300
|
8684
|
|
316
|
8685 if (incount > 1)
|
|
236
|
8686 {
|
|
310
|
8687 len = STRLEN(innames[i]);
|
|
|
8688 if (STRLEN(gettail(innames[i])) < 5
|
|
|
8689 || innames[i][len - 3] != '_')
|
|
236
|
8690 {
|
|
310
|
8691 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
8692 return;
|
|
236
|
8693 }
|
|
316
|
8694 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
8695 spin.si_region_name[i * 2 + 1] =
|
|
|
8696 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
8697 }
|
|
|
8698 }
|
|
316
|
8699 spin.si_region_count = incount;
|
|
236
|
8700
|
|
474
|
8701 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
8702 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
8703 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
8704 if (spin.si_foldroot == NULL
|
|
|
8705 || spin.si_keeproot == NULL
|
|
|
8706 || spin.si_prefroot == NULL)
|
|
300
|
8707 {
|
|
469
|
8708 free_blocks(spin.si_blocks);
|
|
310
|
8709 return;
|
|
300
|
8710 }
|
|
|
8711
|
|
351
|
8712 /* When not producing a .add.spl file clear the character table when
|
|
|
8713 * we encounter one in the .aff file. This means we dump the current
|
|
|
8714 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
8715 * better than guessing the contents, the table will match a
|
|
|
8716 * previously loaded spell file. */
|
|
|
8717 if (!spin.si_add)
|
|
|
8718 spin.si_clear_chartab = TRUE;
|
|
|
8719
|
|
236
|
8720 /*
|
|
|
8721 * Read all the .aff and .dic files.
|
|
|
8722 * Text is converted to 'encoding'.
|
|
300
|
8723 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
8724 */
|
|
310
|
8725 for (i = 0; i < incount && !error; ++i)
|
|
236
|
8726 {
|
|
300
|
8727 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
8728 spin.si_region = 1 << i;
|
|
|
8729
|
|
|
8730 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
8731 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
8732 {
|
|
|
8733 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
8734 * "SET" line. */
|
|
474
|
8735 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
8736 if (afile[i] == NULL)
|
|
300
|
8737 error = TRUE;
|
|
|
8738 else
|
|
|
8739 {
|
|
|
8740 /* Read the .dic file and store the words in the trees. */
|
|
|
8741 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
8742 innames[i]);
|
|
474
|
8743 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
8744 error = TRUE;
|
|
|
8745 }
|
|
|
8746 }
|
|
|
8747 else
|
|
|
8748 {
|
|
|
8749 /* No .aff file, try reading the file as a word list. Store
|
|
|
8750 * the words in the trees. */
|
|
474
|
8751 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
8752 error = TRUE;
|
|
|
8753 }
|
|
236
|
8754
|
|
310
|
8755 #ifdef FEAT_MBYTE
|
|
236
|
8756 /* Free any conversion stuff. */
|
|
300
|
8757 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
8758 #endif
|
|
236
|
8759 }
|
|
|
8760
|
|
492
|
8761 if (spin.si_compflags != NULL && spin.si_nobreak)
|
|
|
8762 MSG(_("Warning: both compounding and NOBREAK specified"));
|
|
|
8763
|
|
625
|
8764 if (!error && !got_int)
|
|
236
|
8765 {
|
|
|
8766 /*
|
|
300
|
8767 * Combine tails in the tree.
|
|
236
|
8768 */
|
|
625
|
8769 spell_message(&spin, (char_u *)_(msg_compressing));
|
|
474
|
8770 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
8771 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
8772 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
8773 }
|
|
|
8774
|
|
625
|
8775 if (!error && !got_int)
|
|
300
|
8776 {
|
|
|
8777 /*
|
|
|
8778 * Write the info in the spell file.
|
|
|
8779 */
|
|
625
|
8780 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
8781 _("Writing spell file %s ..."), wfname);
|
|
|
8782 spell_message(&spin, IObuff);
|
|
310
|
8783
|
|
500
|
8784 error = write_vim_spell(&spin, wfname) == FAIL;
|
|
474
|
8785
|
|
625
|
8786 spell_message(&spin, (char_u *)_("Done!"));
|
|
|
8787 vim_snprintf((char *)IObuff, IOSIZE,
|
|
|
8788 _("Estimated runtime memory use: %d bytes"), spin.si_memtot);
|
|
|
8789 spell_message(&spin, IObuff);
|
|
|
8790
|
|
|
8791 /*
|
|
|
8792 * If the file is loaded need to reload it.
|
|
|
8793 */
|
|
500
|
8794 if (!error)
|
|
|
8795 spell_reload_one(wfname, added_word);
|
|
300
|
8796 }
|
|
|
8797
|
|
|
8798 /* Free the allocated memory. */
|
|
323
|
8799 ga_clear(&spin.si_rep);
|
|
625
|
8800 ga_clear(&spin.si_repsal);
|
|
323
|
8801 ga_clear(&spin.si_sal);
|
|
|
8802 ga_clear(&spin.si_map);
|
|
339
|
8803 ga_clear(&spin.si_prefcond);
|
|
625
|
8804 hash_clear_all(&spin.si_commonwords, 0);
|
|
300
|
8805
|
|
|
8806 /* Free the .aff file structures. */
|
|
310
|
8807 for (i = 0; i < incount; ++i)
|
|
|
8808 if (afile[i] != NULL)
|
|
|
8809 spell_free_aff(afile[i]);
|
|
339
|
8810
|
|
|
8811 /* Free all the bits and pieces at once. */
|
|
|
8812 free_blocks(spin.si_blocks);
|
|
625
|
8813
|
|
|
8814 /*
|
|
|
8815 * If there is soundfolding info and no NOSUGFILE item create the
|
|
|
8816 * .sug file with the soundfolded word trie.
|
|
|
8817 */
|
|
|
8818 if (spin.si_sugtime != 0 && !error && !got_int)
|
|
|
8819 spell_make_sugfile(&spin, wfname);
|
|
|
8820
|
|
|
8821 }
|
|
|
8822 }
|
|
|
8823
|
|
|
8824 /*
|
|
|
8825 * Display a message for spell file processing when 'verbose' is set or using
|
|
|
8826 * ":mkspell". "str" can be IObuff.
|
|
|
8827 */
|
|
|
8828 static void
|
|
|
8829 spell_message(spin, str)
|
|
|
8830 spellinfo_T *spin;
|
|
|
8831 char_u *str;
|
|
|
8832 {
|
|
|
8833 if (spin->si_verbose || p_verbose > 2)
|
|
|
8834 {
|
|
|
8835 if (!spin->si_verbose)
|
|
|
8836 verbose_enter();
|
|
|
8837 MSG(str);
|
|
|
8838 out_flush();
|
|
|
8839 if (!spin->si_verbose)
|
|
|
8840 verbose_leave();
|
|
|
8841 }
|
|
|
8842 }
|
|
310
|
8843
|
|
|
8844 /*
|
|
385
|
8845 * ":[count]spellgood {word}"
|
|
|
8846 * ":[count]spellwrong {word}"
|
|
710
|
8847 * ":[count]spellundo {word}"
|
|
310
|
8848 */
|
|
|
8849 void
|
|
|
8850 ex_spell(eap)
|
|
|
8851 exarg_T *eap;
|
|
|
8852 {
|
|
381
|
8853 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
710
|
8854 eap->forceit ? 0 : (int)eap->line2,
|
|
|
8855 eap->cmdidx == CMD_spellundo);
|
|
236
|
8856 }
|
|
|
8857
|
|
310
|
8858 /*
|
|
|
8859 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
8860 */
|
|
|
8861 void
|
|
710
|
8862 spell_add_word(word, len, bad, index, undo)
|
|
310
|
8863 char_u *word;
|
|
|
8864 int len;
|
|
|
8865 int bad;
|
|
385
|
8866 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
8867 'spellfile' */
|
|
710
|
8868 int undo; /* TRUE for "zug", "zuG", "zuw" and "zuW" */
|
|
310
|
8869 {
|
|
|
8870 FILE *fd;
|
|
385
|
8871 buf_T *buf = NULL;
|
|
351
|
8872 int new_spf = FALSE;
|
|
381
|
8873 char_u *fname;
|
|
385
|
8874 char_u fnamebuf[MAXPATHL];
|
|
|
8875 char_u line[MAXWLEN * 2];
|
|
|
8876 long fpos, fpos_next = 0;
|
|
|
8877 int i;
|
|
|
8878 char_u *spf;
|
|
|
8879
|
|
|
8880 if (index == 0) /* use internal wordlist */
|
|
|
8881 {
|
|
|
8882 if (int_wordlist == NULL)
|
|
|
8883 {
|
|
|
8884 int_wordlist = vim_tempname('s');
|
|
|
8885 if (int_wordlist == NULL)
|
|
381
|
8886 return;
|
|
|
8887 }
|
|
385
|
8888 fname = int_wordlist;
|
|
381
|
8889 }
|
|
310
|
8890 else
|
|
|
8891 {
|
|
381
|
8892 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
8893 if (*curbuf->b_p_spf == NUL)
|
|
|
8894 {
|
|
|
8895 init_spellfile();
|
|
|
8896 new_spf = TRUE;
|
|
|
8897 }
|
|
|
8898
|
|
|
8899 if (*curbuf->b_p_spf == NUL)
|
|
|
8900 {
|
|
523
|
8901 EMSG2(_(e_notset), "spellfile");
|
|
381
|
8902 return;
|
|
|
8903 }
|
|
|
8904
|
|
385
|
8905 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
8906 {
|
|
|
8907 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
8908 if (i == index)
|
|
|
8909 break;
|
|
|
8910 if (*spf == NUL)
|
|
|
8911 {
|
|
502
|
8912 EMSGN(_("E765: 'spellfile' does not have %ld entries"), index);
|
|
385
|
8913 return;
|
|
|
8914 }
|
|
|
8915 }
|
|
|
8916
|
|
310
|
8917 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
8918 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
8919 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
8920 buf = NULL;
|
|
|
8921 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
8922 {
|
|
310
|
8923 EMSG(_(e_bufloaded));
|
|
381
|
8924 return;
|
|
|
8925 }
|
|
|
8926
|
|
385
|
8927 fname = fnamebuf;
|
|
|
8928 }
|
|
|
8929
|
|
710
|
8930 if (bad || undo)
|
|
|
8931 {
|
|
|
8932 /* When the word appears as good word we need to remove that one,
|
|
385
|
8933 * since its flags sort before the one with WF_BANNED. */
|
|
|
8934 fd = mch_fopen((char *)fname, "r");
|
|
|
8935 if (fd != NULL)
|
|
|
8936 {
|
|
|
8937 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
8938 {
|
|
|
8939 fpos = fpos_next;
|
|
|
8940 fpos_next = ftell(fd);
|
|
|
8941 if (STRNCMP(word, line, len) == 0
|
|
|
8942 && (line[len] == '/' || line[len] < ' '))
|
|
|
8943 {
|
|
|
8944 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
8945 * the start of the line. Mixing reading and writing
|
|
|
8946 * doesn't work for all systems, close the file first. */
|
|
|
8947 fclose(fd);
|
|
|
8948 fd = mch_fopen((char *)fname, "r+");
|
|
|
8949 if (fd == NULL)
|
|
|
8950 break;
|
|
|
8951 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
710
|
8952 {
|
|
385
|
8953 fputc('#', fd);
|
|
710
|
8954 if (undo)
|
|
|
8955 smsg((char_u *)_("Word removed from %s"), NameBuff);
|
|
|
8956 }
|
|
385
|
8957 fseek(fd, fpos_next, SEEK_SET);
|
|
|
8958 }
|
|
|
8959 }
|
|
|
8960 fclose(fd);
|
|
|
8961 }
|
|
381
|
8962 }
|
|
|
8963
|
|
710
|
8964 if (!undo)
|
|
|
8965 {
|
|
|
8966 fd = mch_fopen((char *)fname, "a");
|
|
|
8967 if (fd == NULL && new_spf)
|
|
|
8968 {
|
|
|
8969 /* We just initialized the 'spellfile' option and can't open the
|
|
|
8970 * file. We may need to create the "spell" directory first. We
|
|
|
8971 * already checked the runtime directory is writable in
|
|
|
8972 * init_spellfile(). */
|
|
|
8973 if (!dir_of_file_exists(fname))
|
|
|
8974 {
|
|
|
8975 /* The directory doesn't exist. Try creating it and opening
|
|
|
8976 * the file again. */
|
|
|
8977 vim_mkdir(NameBuff, 0755);
|
|
|
8978 fd = mch_fopen((char *)fname, "a");
|
|
|
8979 }
|
|
|
8980 }
|
|
|
8981
|
|
|
8982 if (fd == NULL)
|
|
|
8983 EMSG2(_(e_notopen), fname);
|
|
310
|
8984 else
|
|
710
|
8985 {
|
|
|
8986 if (bad)
|
|
|
8987 fprintf(fd, "%.*s/!\n", len, word);
|
|
|
8988 else
|
|
|
8989 fprintf(fd, "%.*s\n", len, word);
|
|
|
8990 fclose(fd);
|
|
|
8991
|
|
|
8992 home_replace(NULL, fname, NameBuff, MAXPATHL, TRUE);
|
|
|
8993 smsg((char_u *)_("Word added to %s"), NameBuff);
|
|
|
8994 }
|
|
|
8995 }
|
|
|
8996
|
|
|
8997 if (fd != NULL)
|
|
|
8998 {
|
|
381
|
8999 /* Update the .add.spl file. */
|
|
|
9000 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
9001
|
|
|
9002 /* If the .add file is edited somewhere, reload it. */
|
|
|
9003 if (buf != NULL)
|
|
627
|
9004 buf_reload(buf, buf->b_orig_mode);
|
|
381
|
9005
|
|
|
9006 redraw_all_later(NOT_VALID);
|
|
310
|
9007 }
|
|
|
9008 }
|
|
|
9009
|
|
|
9010 /*
|
|
|
9011 * Initialize 'spellfile' for the current buffer.
|
|
|
9012 */
|
|
|
9013 static void
|
|
|
9014 init_spellfile()
|
|
|
9015 {
|
|
|
9016 char_u buf[MAXPATHL];
|
|
|
9017 int l;
|
|
500
|
9018 char_u *fname;
|
|
310
|
9019 char_u *rtp;
|
|
351
|
9020 char_u *lend;
|
|
501
|
9021 int aspath = FALSE;
|
|
|
9022 char_u *lstart = curbuf->b_p_spl;
|
|
310
|
9023
|
|
|
9024 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
9025 {
|
|
501
|
9026 /* Find the end of the language name. Exclude the region. If there
|
|
|
9027 * is a path separator remember the start of the tail. */
|
|
351
|
9028 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
9029 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
501
|
9030 if (vim_ispathsep(*lend))
|
|
|
9031 {
|
|
|
9032 aspath = TRUE;
|
|
|
9033 lstart = lend + 1;
|
|
|
9034 }
|
|
351
|
9035
|
|
|
9036 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
9037 * are allowed to write. */
|
|
310
|
9038 rtp = p_rtp;
|
|
|
9039 while (*rtp != NUL)
|
|
|
9040 {
|
|
501
|
9041 if (aspath)
|
|
|
9042 /* Use directory of an entry with path, e.g., for
|
|
|
9043 * "/dir/lg.utf-8.spl" use "/dir". */
|
|
|
9044 vim_strncpy(buf, curbuf->b_p_spl, lstart - curbuf->b_p_spl - 1);
|
|
|
9045 else
|
|
|
9046 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
9047 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
310
|
9048 if (filewritable(buf) == 2)
|
|
|
9049 {
|
|
316
|
9050 /* Use the first language name from 'spelllang' and the
|
|
|
9051 * encoding used in the first loaded .spl file. */
|
|
501
|
9052 if (aspath)
|
|
|
9053 vim_strncpy(buf, curbuf->b_p_spl, lend - curbuf->b_p_spl);
|
|
|
9054 else
|
|
|
9055 {
|
|
|
9056 l = STRLEN(buf);
|
|
|
9057 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
|
9058 "/spell/%.*s", (int)(lend - lstart), lstart);
|
|
|
9059 }
|
|
310
|
9060 l = STRLEN(buf);
|
|
501
|
9061 fname = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang->sl_fname;
|
|
|
9062 vim_snprintf((char *)buf + l, MAXPATHL - l, ".%s.add",
|
|
|
9063 fname != NULL
|
|
|
9064 && strstr((char *)gettail(fname), ".ascii.") != NULL
|
|
|
9065 ? (char_u *)"ascii" : spell_enc());
|
|
310
|
9066 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
9067 break;
|
|
|
9068 }
|
|
501
|
9069 aspath = FALSE;
|
|
310
|
9070 }
|
|
|
9071 }
|
|
|
9072 }
|
|
236
|
9073
|
|
300
|
9074
|
|
307
|
9075 /*
|
|
|
9076 * Init the chartab used for spelling for ASCII.
|
|
|
9077 * EBCDIC is not supported!
|
|
|
9078 */
|
|
|
9079 static void
|
|
|
9080 clear_spell_chartab(sp)
|
|
|
9081 spelltab_T *sp;
|
|
|
9082 {
|
|
324
|
9083 int i;
|
|
307
|
9084
|
|
|
9085 /* Init everything to FALSE. */
|
|
|
9086 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
9087 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
9088 for (i = 0; i < 256; ++i)
|
|
324
|
9089 {
|
|
307
|
9090 sp->st_fold[i] = i;
|
|
324
|
9091 sp->st_upper[i] = i;
|
|
|
9092 }
|
|
307
|
9093
|
|
|
9094 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
9095 * that is done separately. */
|
|
|
9096 for (i = '0'; i <= '9'; ++i)
|
|
|
9097 sp->st_isw[i] = TRUE;
|
|
|
9098 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
9099 {
|
|
|
9100 sp->st_isw[i] = TRUE;
|
|
|
9101 sp->st_isu[i] = TRUE;
|
|
|
9102 sp->st_fold[i] = i + 0x20;
|
|
|
9103 }
|
|
|
9104 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
9105 {
|
|
307
|
9106 sp->st_isw[i] = TRUE;
|
|
324
|
9107 sp->st_upper[i] = i - 0x20;
|
|
|
9108 }
|
|
307
|
9109 }
|
|
|
9110
|
|
|
9111 /*
|
|
|
9112 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
9113 * Called once while starting up and when 'encoding' changes.
|
|
|
9114 * The default is to use isalpha(), but the spell file should define the word
|
|
|
9115 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
9116 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
9117 */
|
|
|
9118 void
|
|
|
9119 init_spell_chartab()
|
|
|
9120 {
|
|
|
9121 int i;
|
|
|
9122
|
|
|
9123 did_set_spelltab = FALSE;
|
|
|
9124 clear_spell_chartab(&spelltab);
|
|
|
9125 #ifdef FEAT_MBYTE
|
|
|
9126 if (enc_dbcs)
|
|
|
9127 {
|
|
|
9128 /* DBCS: assume double-wide characters are word characters. */
|
|
|
9129 for (i = 128; i <= 255; ++i)
|
|
|
9130 if (MB_BYTE2LEN(i) == 2)
|
|
|
9131 spelltab.st_isw[i] = TRUE;
|
|
|
9132 }
|
|
324
|
9133 else if (enc_utf8)
|
|
|
9134 {
|
|
|
9135 for (i = 128; i < 256; ++i)
|
|
|
9136 {
|
|
|
9137 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
9138 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
9139 spelltab.st_fold[i] = utf_fold(i);
|
|
|
9140 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
9141 }
|
|
|
9142 }
|
|
307
|
9143 else
|
|
|
9144 #endif
|
|
|
9145 {
|
|
324
|
9146 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
9147 for (i = 128; i < 256; ++i)
|
|
|
9148 {
|
|
|
9149 if (MB_ISUPPER(i))
|
|
|
9150 {
|
|
324
|
9151 spelltab.st_isw[i] = TRUE;
|
|
307
|
9152 spelltab.st_isu[i] = TRUE;
|
|
|
9153 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
9154 }
|
|
324
|
9155 else if (MB_ISLOWER(i))
|
|
|
9156 {
|
|
|
9157 spelltab.st_isw[i] = TRUE;
|
|
|
9158 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
9159 }
|
|
307
|
9160 }
|
|
|
9161 }
|
|
|
9162 }
|
|
|
9163
|
|
|
9164 /*
|
|
|
9165 * Set the spell character tables from strings in the affix file.
|
|
|
9166 */
|
|
|
9167 static int
|
|
|
9168 set_spell_chartab(fol, low, upp)
|
|
|
9169 char_u *fol;
|
|
|
9170 char_u *low;
|
|
|
9171 char_u *upp;
|
|
|
9172 {
|
|
|
9173 /* We build the new tables here first, so that we can compare with the
|
|
|
9174 * previous one. */
|
|
|
9175 spelltab_T new_st;
|
|
|
9176 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
9177 int f, l, u;
|
|
|
9178
|
|
|
9179 clear_spell_chartab(&new_st);
|
|
|
9180
|
|
|
9181 while (*pf != NUL)
|
|
|
9182 {
|
|
|
9183 if (*pl == NUL || *pu == NUL)
|
|
|
9184 {
|
|
|
9185 EMSG(_(e_affform));
|
|
|
9186 return FAIL;
|
|
|
9187 }
|
|
|
9188 #ifdef FEAT_MBYTE
|
|
|
9189 f = mb_ptr2char_adv(&pf);
|
|
|
9190 l = mb_ptr2char_adv(&pl);
|
|
|
9191 u = mb_ptr2char_adv(&pu);
|
|
|
9192 #else
|
|
|
9193 f = *pf++;
|
|
|
9194 l = *pl++;
|
|
|
9195 u = *pu++;
|
|
|
9196 #endif
|
|
|
9197 /* Every character that appears is a word character. */
|
|
|
9198 if (f < 256)
|
|
|
9199 new_st.st_isw[f] = TRUE;
|
|
|
9200 if (l < 256)
|
|
|
9201 new_st.st_isw[l] = TRUE;
|
|
|
9202 if (u < 256)
|
|
|
9203 new_st.st_isw[u] = TRUE;
|
|
|
9204
|
|
|
9205 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
9206 * case-folding */
|
|
|
9207 if (l < 256 && l != f)
|
|
|
9208 {
|
|
|
9209 if (f >= 256)
|
|
|
9210 {
|
|
|
9211 EMSG(_(e_affrange));
|
|
|
9212 return FAIL;
|
|
|
9213 }
|
|
|
9214 new_st.st_fold[l] = f;
|
|
|
9215 }
|
|
|
9216
|
|
|
9217 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
9218 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
9219 * "FOL" . */
|
|
307
|
9220 if (u < 256 && u != f)
|
|
|
9221 {
|
|
|
9222 if (f >= 256)
|
|
|
9223 {
|
|
|
9224 EMSG(_(e_affrange));
|
|
|
9225 return FAIL;
|
|
|
9226 }
|
|
|
9227 new_st.st_fold[u] = f;
|
|
|
9228 new_st.st_isu[u] = TRUE;
|
|
324
|
9229 new_st.st_upper[f] = u;
|
|
307
|
9230 }
|
|
|
9231 }
|
|
|
9232
|
|
|
9233 if (*pl != NUL || *pu != NUL)
|
|
|
9234 {
|
|
|
9235 EMSG(_(e_affform));
|
|
|
9236 return FAIL;
|
|
|
9237 }
|
|
|
9238
|
|
|
9239 return set_spell_finish(&new_st);
|
|
|
9240 }
|
|
|
9241
|
|
|
9242 /*
|
|
|
9243 * Set the spell character tables from strings in the .spl file.
|
|
|
9244 */
|
|
483
|
9245 static void
|
|
389
|
9246 set_spell_charflags(flags, cnt, fol)
|
|
307
|
9247 char_u *flags;
|
|
389
|
9248 int cnt; /* length of "flags" */
|
|
|
9249 char_u *fol;
|
|
307
|
9250 {
|
|
|
9251 /* We build the new tables here first, so that we can compare with the
|
|
|
9252 * previous one. */
|
|
|
9253 spelltab_T new_st;
|
|
|
9254 int i;
|
|
389
|
9255 char_u *p = fol;
|
|
324
|
9256 int c;
|
|
307
|
9257
|
|
|
9258 clear_spell_chartab(&new_st);
|
|
|
9259
|
|
389
|
9260 for (i = 0; i < 128; ++i)
|
|
|
9261 {
|
|
|
9262 if (i < cnt)
|
|
|
9263 {
|
|
|
9264 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
9265 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
9266 }
|
|
|
9267
|
|
|
9268 if (*p != NUL)
|
|
|
9269 {
|
|
307
|
9270 #ifdef FEAT_MBYTE
|
|
389
|
9271 c = mb_ptr2char_adv(&p);
|
|
307
|
9272 #else
|
|
389
|
9273 c = *p++;
|
|
307
|
9274 #endif
|
|
389
|
9275 new_st.st_fold[i + 128] = c;
|
|
|
9276 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
9277 new_st.st_upper[c] = i + 128;
|
|
|
9278 }
|
|
307
|
9279 }
|
|
|
9280
|
|
483
|
9281 (void)set_spell_finish(&new_st);
|
|
307
|
9282 }
|
|
|
9283
|
|
|
9284 static int
|
|
|
9285 set_spell_finish(new_st)
|
|
|
9286 spelltab_T *new_st;
|
|
|
9287 {
|
|
|
9288 int i;
|
|
|
9289
|
|
|
9290 if (did_set_spelltab)
|
|
|
9291 {
|
|
|
9292 /* check that it's the same table */
|
|
|
9293 for (i = 0; i < 256; ++i)
|
|
|
9294 {
|
|
|
9295 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
9296 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
9297 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
9298 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
9299 {
|
|
|
9300 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
9301 return FAIL;
|
|
|
9302 }
|
|
|
9303 }
|
|
|
9304 }
|
|
|
9305 else
|
|
|
9306 {
|
|
|
9307 /* copy the new spelltab into the one being used */
|
|
|
9308 spelltab = *new_st;
|
|
|
9309 did_set_spelltab = TRUE;
|
|
|
9310 }
|
|
|
9311
|
|
|
9312 return OK;
|
|
|
9313 }
|
|
|
9314
|
|
|
9315 /*
|
|
358
|
9316 * Return TRUE if "p" points to a word character.
|
|
366
|
9317 * As a special case we see "midword" characters as word character when it is
|
|
358
|
9318 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
9319 * Thus this only works properly when past the first character of the word.
|
|
358
|
9320 */
|
|
|
9321 static int
|
|
376
|
9322 spell_iswordp(p, buf)
|
|
358
|
9323 char_u *p;
|
|
376
|
9324 buf_T *buf; /* buffer used */
|
|
358
|
9325 {
|
|
366
|
9326 #ifdef FEAT_MBYTE
|
|
358
|
9327 char_u *s;
|
|
366
|
9328 int l;
|
|
|
9329 int c;
|
|
|
9330
|
|
|
9331 if (has_mbyte)
|
|
|
9332 {
|
|
|
9333 l = MB_BYTE2LEN(*p);
|
|
358
|
9334 s = p;
|
|
366
|
9335 if (l == 1)
|
|
|
9336 {
|
|
|
9337 /* be quick for ASCII */
|
|
376
|
9338 if (buf->b_spell_ismw[*p])
|
|
366
|
9339 {
|
|
|
9340 s = p + 1; /* skip a mid-word character */
|
|
|
9341 l = MB_BYTE2LEN(*s);
|
|
|
9342 }
|
|
|
9343 }
|
|
|
9344 else
|
|
|
9345 {
|
|
|
9346 c = mb_ptr2char(p);
|
|
376
|
9347 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
9348 : (buf->b_spell_ismw_mb != NULL
|
|
|
9349 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
9350 {
|
|
|
9351 s = p + l;
|
|
|
9352 l = MB_BYTE2LEN(*s);
|
|
|
9353 }
|
|
|
9354 }
|
|
|
9355
|
|
390
|
9356 c = mb_ptr2char(s);
|
|
|
9357 if (c > 255)
|
|
366
|
9358 return mb_get_class(s) >= 2;
|
|
390
|
9359 return spelltab.st_isw[c];
|
|
366
|
9360 }
|
|
358
|
9361 #endif
|
|
366
|
9362
|
|
376
|
9363 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
9364 }
|
|
|
9365
|
|
|
9366 /*
|
|
|
9367 * Return TRUE if "p" points to a word character.
|
|
|
9368 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
9369 */
|
|
|
9370 static int
|
|
|
9371 spell_iswordp_nmw(p)
|
|
|
9372 char_u *p;
|
|
|
9373 {
|
|
|
9374 #ifdef FEAT_MBYTE
|
|
390
|
9375 int c;
|
|
|
9376
|
|
|
9377 if (has_mbyte)
|
|
|
9378 {
|
|
|
9379 c = mb_ptr2char(p);
|
|
|
9380 if (c > 255)
|
|
|
9381 return mb_get_class(p) >= 2;
|
|
|
9382 return spelltab.st_isw[c];
|
|
|
9383 }
|
|
376
|
9384 #endif
|
|
|
9385 return spelltab.st_isw[*p];
|
|
358
|
9386 }
|
|
|
9387
|
|
372
|
9388 #ifdef FEAT_MBYTE
|
|
|
9389 /*
|
|
|
9390 * Return TRUE if "p" points to a word character.
|
|
|
9391 * Wide version of spell_iswordp().
|
|
|
9392 */
|
|
|
9393 static int
|
|
376
|
9394 spell_iswordp_w(p, buf)
|
|
372
|
9395 int *p;
|
|
376
|
9396 buf_T *buf;
|
|
372
|
9397 {
|
|
|
9398 int *s;
|
|
|
9399
|
|
376
|
9400 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
9401 : (buf->b_spell_ismw_mb != NULL
|
|
|
9402 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
9403 s = p + 1;
|
|
|
9404 else
|
|
|
9405 s = p;
|
|
|
9406
|
|
390
|
9407 if (*s > 255)
|
|
372
|
9408 {
|
|
|
9409 if (enc_utf8)
|
|
|
9410 return utf_class(*s) >= 2;
|
|
|
9411 if (enc_dbcs)
|
|
|
9412 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
9413 return 0;
|
|
|
9414 }
|
|
|
9415 return spelltab.st_isw[*s];
|
|
|
9416 }
|
|
|
9417 #endif
|
|
|
9418
|
|
358
|
9419 /*
|
|
339
|
9420 * Write the table with prefix conditions to the .spl file.
|
|
483
|
9421 * When "fd" is NULL only count the length of what is written.
|
|
|
9422 */
|
|
|
9423 static int
|
|
339
|
9424 write_spell_prefcond(fd, gap)
|
|
|
9425 FILE *fd;
|
|
|
9426 garray_T *gap;
|
|
|
9427 {
|
|
|
9428 int i;
|
|
|
9429 char_u *p;
|
|
|
9430 int len;
|
|
483
|
9431 int totlen;
|
|
|
9432
|
|
|
9433 if (fd != NULL)
|
|
|
9434 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
9435
|
|
|
9436 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */
|
|
339
|
9437
|
|
|
9438 for (i = 0; i < gap->ga_len; ++i)
|
|
|
9439 {
|
|
|
9440 /* <prefcond> : <condlen> <condstr> */
|
|
|
9441 p = ((char_u **)gap->ga_data)[i];
|
|
483
|
9442 if (p != NULL)
|
|
339
|
9443 {
|
|
|
9444 len = STRLEN(p);
|
|
483
|
9445 if (fd != NULL)
|
|
|
9446 {
|
|
|
9447 fputc(len, fd);
|
|
|
9448 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
9449 }
|
|
|
9450 totlen += len;
|
|
|
9451 }
|
|
|
9452 else if (fd != NULL)
|
|
|
9453 fputc(0, fd);
|
|
|
9454 }
|
|
|
9455
|
|
|
9456 return totlen;
|
|
307
|
9457 }
|
|
|
9458
|
|
|
9459 /*
|
|
324
|
9460 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
9461 * Uses the character definitions from the .spl file.
|
|
307
|
9462 * When using a multi-byte 'encoding' the length may change!
|
|
|
9463 * Returns FAIL when something wrong.
|
|
|
9464 */
|
|
|
9465 static int
|
|
324
|
9466 spell_casefold(str, len, buf, buflen)
|
|
|
9467 char_u *str;
|
|
307
|
9468 int len;
|
|
|
9469 char_u *buf;
|
|
|
9470 int buflen;
|
|
|
9471 {
|
|
|
9472 int i;
|
|
|
9473
|
|
|
9474 if (len >= buflen)
|
|
|
9475 {
|
|
|
9476 buf[0] = NUL;
|
|
|
9477 return FAIL; /* result will not fit */
|
|
|
9478 }
|
|
|
9479
|
|
|
9480 #ifdef FEAT_MBYTE
|
|
|
9481 if (has_mbyte)
|
|
|
9482 {
|
|
324
|
9483 int outi = 0;
|
|
|
9484 char_u *p;
|
|
307
|
9485 int c;
|
|
|
9486
|
|
|
9487 /* Fold one character at a time. */
|
|
324
|
9488 for (p = str; p < str + len; )
|
|
307
|
9489 {
|
|
|
9490 if (outi + MB_MAXBYTES > buflen)
|
|
|
9491 {
|
|
|
9492 buf[outi] = NUL;
|
|
|
9493 return FAIL;
|
|
|
9494 }
|
|
474
|
9495 c = mb_cptr2char_adv(&p);
|
|
324
|
9496 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
9497 }
|
|
|
9498 buf[outi] = NUL;
|
|
|
9499 }
|
|
|
9500 else
|
|
|
9501 #endif
|
|
|
9502 {
|
|
|
9503 /* Be quick for non-multibyte encodings. */
|
|
|
9504 for (i = 0; i < len; ++i)
|
|
324
|
9505 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
9506 buf[i] = NUL;
|
|
|
9507 }
|
|
|
9508
|
|
|
9509 return OK;
|
|
|
9510 }
|
|
|
9511
|
|
625
|
9512 /* values for sps_flags */
|
|
372
|
9513 #define SPS_BEST 1
|
|
|
9514 #define SPS_FAST 2
|
|
|
9515 #define SPS_DOUBLE 4
|
|
|
9516
|
|
625
|
9517 static int sps_flags = SPS_BEST; /* flags from 'spellsuggest' */
|
|
|
9518 static int sps_limit = 9999; /* max nr of suggestions given */
|
|
372
|
9519
|
|
|
9520 /*
|
|
|
9521 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
483
|
9522 * Sets "sps_flags" and "sps_limit".
|
|
372
|
9523 */
|
|
|
9524 int
|
|
|
9525 spell_check_sps()
|
|
|
9526 {
|
|
|
9527 char_u *p;
|
|
483
|
9528 char_u *s;
|
|
372
|
9529 char_u buf[MAXPATHL];
|
|
|
9530 int f;
|
|
|
9531
|
|
|
9532 sps_flags = 0;
|
|
483
|
9533 sps_limit = 9999;
|
|
372
|
9534
|
|
|
9535 for (p = p_sps; *p != NUL; )
|
|
|
9536 {
|
|
|
9537 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
9538
|
|
|
9539 f = 0;
|
|
483
|
9540 if (VIM_ISDIGIT(*buf))
|
|
|
9541 {
|
|
|
9542 s = buf;
|
|
|
9543 sps_limit = getdigits(&s);
|
|
|
9544 if (*s != NUL && !VIM_ISDIGIT(*s))
|
|
|
9545 f = -1;
|
|
|
9546 }
|
|
|
9547 else if (STRCMP(buf, "best") == 0)
|
|
372
|
9548 f = SPS_BEST;
|
|
|
9549 else if (STRCMP(buf, "fast") == 0)
|
|
|
9550 f = SPS_FAST;
|
|
|
9551 else if (STRCMP(buf, "double") == 0)
|
|
|
9552 f = SPS_DOUBLE;
|
|
|
9553 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
9554 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
9555 f = -1;
|
|
|
9556
|
|
|
9557 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
9558 {
|
|
|
9559 sps_flags = SPS_BEST;
|
|
483
|
9560 sps_limit = 9999;
|
|
372
|
9561 return FAIL;
|
|
|
9562 }
|
|
|
9563 if (f != 0)
|
|
|
9564 sps_flags = f;
|
|
|
9565 }
|
|
|
9566
|
|
|
9567 if (sps_flags == 0)
|
|
|
9568 sps_flags = SPS_BEST;
|
|
|
9569
|
|
|
9570 return OK;
|
|
|
9571 }
|
|
|
9572
|
|
323
|
9573 /*
|
|
|
9574 * "z?": Find badly spelled word under or after the cursor.
|
|
|
9575 * Give suggestions for the properly spelled word.
|
|
638
|
9576 * In Visual mode use the highlighted word as the bad word.
|
|
485
|
9577 * When "count" is non-zero use that suggestion.
|
|
323
|
9578 */
|
|
|
9579 void
|
|
485
|
9580 spell_suggest(count)
|
|
|
9581 int count;
|
|
323
|
9582 {
|
|
|
9583 char_u *line;
|
|
|
9584 pos_T prev_cursor = curwin->w_cursor;
|
|
|
9585 char_u wcopy[MAXWLEN + 2];
|
|
|
9586 char_u *p;
|
|
|
9587 int i;
|
|
|
9588 int c;
|
|
|
9589 suginfo_T sug;
|
|
|
9590 suggest_T *stp;
|
|
372
|
9591 int mouse_used;
|
|
388
|
9592 int need_cap;
|
|
483
|
9593 int limit;
|
|
485
|
9594 int selected = count;
|
|
638
|
9595 int badlen = 0;
|
|
|
9596
|
|
|
9597 if (no_spell_checking(curwin))
|
|
|
9598 return;
|
|
|
9599
|
|
|
9600 #ifdef FEAT_VISUAL
|
|
|
9601 if (VIsual_active)
|
|
|
9602 {
|
|
|
9603 /* Use the Visually selected text as the bad word. But reject
|
|
|
9604 * a multi-line selection. */
|
|
|
9605 if (curwin->w_cursor.lnum != VIsual.lnum)
|
|
|
9606 {
|
|
|
9607 vim_beep();
|
|
|
9608 return;
|
|
|
9609 }
|
|
|
9610 badlen = (int)curwin->w_cursor.col - (int)VIsual.col;
|
|
|
9611 if (badlen < 0)
|
|
|
9612 badlen = -badlen;
|
|
|
9613 else
|
|
|
9614 curwin->w_cursor.col = VIsual.col;
|
|
|
9615 ++badlen;
|
|
|
9616 end_visual_mode();
|
|
|
9617 }
|
|
|
9618 else
|
|
|
9619 #endif
|
|
|
9620 /* Find the start of the badly spelled word. */
|
|
|
9621 if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0
|
|
346
|
9622 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
9623 {
|
|
|
9624 /* No bad word or it starts after the cursor: use the word under the
|
|
|
9625 * cursor. */
|
|
|
9626 curwin->w_cursor = prev_cursor;
|
|
|
9627 line = ml_get_curline();
|
|
|
9628 p = line + curwin->w_cursor.col;
|
|
|
9629 /* Backup to before start of word. */
|
|
390
|
9630 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
9631 mb_ptr_back(line, p);
|
|
|
9632 /* Forward to start of word. */
|
|
390
|
9633 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
9634 mb_ptr_adv(p);
|
|
|
9635
|
|
390
|
9636 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
9637 {
|
|
|
9638 beep_flush();
|
|
|
9639 return;
|
|
|
9640 }
|
|
|
9641 curwin->w_cursor.col = p - line;
|
|
323
|
9642 }
|
|
|
9643
|
|
344
|
9644 /* Get the word and its length. */
|
|
|
9645
|
|
388
|
9646 /* Figure out if the word should be capitalised. */
|
|
475
|
9647 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
9648
|
|
|
9649 line = ml_get_curline();
|
|
388
|
9650
|
|
483
|
9651 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in
|
|
|
9652 * 'spellsuggest', whatever is smaller. */
|
|
|
9653 if (sps_limit > (int)Rows - 2)
|
|
|
9654 limit = (int)Rows - 2;
|
|
|
9655 else
|
|
|
9656 limit = sps_limit;
|
|
638
|
9657 spell_find_suggest(line + curwin->w_cursor.col, badlen, &sug, limit,
|
|
625
|
9658 TRUE, need_cap, TRUE);
|
|
323
|
9659
|
|
|
9660 if (sug.su_ga.ga_len == 0)
|
|
|
9661 MSG(_("Sorry, no suggestions"));
|
|
485
|
9662 else if (count > 0)
|
|
|
9663 {
|
|
|
9664 if (count > sug.su_ga.ga_len)
|
|
|
9665 smsg((char_u *)_("Sorry, only %ld suggestions"),
|
|
|
9666 (long)sug.su_ga.ga_len);
|
|
|
9667 }
|
|
323
|
9668 else
|
|
|
9669 {
|
|
372
|
9670 vim_free(repl_from);
|
|
|
9671 repl_from = NULL;
|
|
|
9672 vim_free(repl_to);
|
|
|
9673 repl_to = NULL;
|
|
|
9674
|
|
474
|
9675 #ifdef FEAT_RIGHTLEFT
|
|
|
9676 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
9677 cmdmsg_rl = curwin->w_p_rl;
|
|
|
9678 if (cmdmsg_rl)
|
|
|
9679 msg_col = Columns - 1;
|
|
|
9680 #endif
|
|
|
9681
|
|
323
|
9682 /* List the suggestions. */
|
|
|
9683 msg_start();
|
|
372
|
9684 lines_left = Rows; /* avoid more prompt */
|
|
323
|
9685 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
9686 sug.su_badlen, sug.su_badptr);
|
|
474
|
9687 #ifdef FEAT_RIGHTLEFT
|
|
|
9688 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
9689 {
|
|
|
9690 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
9691 * untranslated message rightleft. */
|
|
|
9692 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
9693 sug.su_badlen, sug.su_badptr);
|
|
|
9694 }
|
|
|
9695 #endif
|
|
323
|
9696 msg_puts(IObuff);
|
|
|
9697 msg_clr_eos();
|
|
|
9698 msg_putchar('\n');
|
|
346
|
9699
|
|
323
|
9700 msg_scroll = TRUE;
|
|
|
9701 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
9702 {
|
|
344
|
9703 stp = &SUG(sug.su_ga, i);
|
|
323
|
9704
|
|
|
9705 /* The suggested word may replace only part of the bad word, add
|
|
|
9706 * the not replaced part. */
|
|
|
9707 STRCPY(wcopy, stp->st_word);
|
|
|
9708 if (sug.su_badlen > stp->st_orglen)
|
|
625
|
9709 vim_strncpy(wcopy + stp->st_wordlen,
|
|
323
|
9710 sug.su_badptr + stp->st_orglen,
|
|
|
9711 sug.su_badlen - stp->st_orglen);
|
|
474
|
9712 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
9713 #ifdef FEAT_RIGHTLEFT
|
|
|
9714 if (cmdmsg_rl)
|
|
|
9715 rl_mirror(IObuff);
|
|
|
9716 #endif
|
|
|
9717 msg_puts(IObuff);
|
|
|
9718
|
|
|
9719 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
9720 msg_puts(IObuff);
|
|
|
9721
|
|
|
9722 /* The word may replace more than "su_badlen". */
|
|
|
9723 if (sug.su_badlen < stp->st_orglen)
|
|
|
9724 {
|
|
|
9725 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
9726 stp->st_orglen, sug.su_badptr);
|
|
|
9727 msg_puts(IObuff);
|
|
|
9728 }
|
|
|
9729
|
|
324
|
9730 if (p_verbose > 0)
|
|
344
|
9731 {
|
|
346
|
9732 /* Add the score. */
|
|
351
|
9733 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
9734 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
9735 stp->st_salscore ? "s " : "",
|
|
|
9736 stp->st_score, stp->st_altscore);
|
|
|
9737 else
|
|
474
|
9738 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
9739 stp->st_score);
|
|
474
|
9740 #ifdef FEAT_RIGHTLEFT
|
|
|
9741 if (cmdmsg_rl)
|
|
|
9742 /* Mirror the numbers, but keep the leading space. */
|
|
|
9743 rl_mirror(IObuff + 1);
|
|
|
9744 #endif
|
|
346
|
9745 msg_advance(30);
|
|
|
9746 msg_puts(IObuff);
|
|
344
|
9747 }
|
|
323
|
9748 msg_putchar('\n');
|
|
|
9749 }
|
|
|
9750
|
|
474
|
9751 #ifdef FEAT_RIGHTLEFT
|
|
|
9752 cmdmsg_rl = FALSE;
|
|
|
9753 msg_col = 0;
|
|
|
9754 #endif
|
|
323
|
9755 /* Ask for choice. */
|
|
485
|
9756 selected = prompt_for_number(&mouse_used);
|
|
372
|
9757 if (mouse_used)
|
|
485
|
9758 selected -= lines_left;
|
|
|
9759 }
|
|
|
9760
|
|
|
9761 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
|
9762 {
|
|
|
9763 /* Save the from and to text for :spellrepall. */
|
|
|
9764 stp = &SUG(sug.su_ga, selected - 1);
|
|
548
|
9765 if (sug.su_badlen > stp->st_orglen)
|
|
|
9766 {
|
|
|
9767 /* Replacing less than "su_badlen", append the remainder to
|
|
|
9768 * repl_to. */
|
|
|
9769 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen);
|
|
|
9770 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word,
|
|
|
9771 sug.su_badlen - stp->st_orglen,
|
|
|
9772 sug.su_badptr + stp->st_orglen);
|
|
|
9773 repl_to = vim_strsave(IObuff);
|
|
|
9774 }
|
|
|
9775 else
|
|
|
9776 {
|
|
|
9777 /* Replacing su_badlen or more, use the whole word. */
|
|
|
9778 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
9779 repl_to = vim_strsave(stp->st_word);
|
|
|
9780 }
|
|
485
|
9781
|
|
|
9782 /* Replace the word. */
|
|
625
|
9783 p = alloc(STRLEN(line) - stp->st_orglen + stp->st_wordlen + 1);
|
|
485
|
9784 if (p != NULL)
|
|
|
9785 {
|
|
|
9786 c = sug.su_badptr - line;
|
|
|
9787 mch_memmove(p, line, c);
|
|
|
9788 STRCPY(p + c, stp->st_word);
|
|
|
9789 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
9790 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
9791 curwin->w_cursor.col = c;
|
|
|
9792 changed_bytes(curwin->w_cursor.lnum, c);
|
|
|
9793
|
|
|
9794 /* For redo we use a change-word command. */
|
|
|
9795 ResetRedobuff();
|
|
|
9796 AppendToRedobuff((char_u *)"ciw");
|
|
620
|
9797 AppendToRedobuffLit(p + c,
|
|
625
|
9798 stp->st_wordlen + sug.su_badlen - stp->st_orglen);
|
|
485
|
9799 AppendCharToRedobuff(ESC);
|
|
|
9800 }
|
|
|
9801 }
|
|
|
9802 else
|
|
|
9803 curwin->w_cursor = prev_cursor;
|
|
323
|
9804
|
|
344
|
9805 spell_find_cleanup(&sug);
|
|
|
9806 }
|
|
|
9807
|
|
|
9808 /*
|
|
475
|
9809 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
9810 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
9811 */
|
|
|
9812 static int
|
|
|
9813 check_need_cap(lnum, col)
|
|
|
9814 linenr_T lnum;
|
|
|
9815 colnr_T col;
|
|
|
9816 {
|
|
|
9817 int need_cap = FALSE;
|
|
|
9818 char_u *line;
|
|
|
9819 char_u *line_copy = NULL;
|
|
|
9820 char_u *p;
|
|
|
9821 colnr_T endcol;
|
|
|
9822 regmatch_T regmatch;
|
|
|
9823
|
|
|
9824 if (curbuf->b_cap_prog == NULL)
|
|
|
9825 return FALSE;
|
|
|
9826
|
|
|
9827 line = ml_get_curline();
|
|
|
9828 endcol = 0;
|
|
|
9829 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
9830 {
|
|
|
9831 /* At start of line, check if previous line is empty or sentence
|
|
|
9832 * ends there. */
|
|
|
9833 if (lnum == 1)
|
|
|
9834 need_cap = TRUE;
|
|
|
9835 else
|
|
|
9836 {
|
|
|
9837 line = ml_get(lnum - 1);
|
|
|
9838 if (*skipwhite(line) == NUL)
|
|
|
9839 need_cap = TRUE;
|
|
|
9840 else
|
|
|
9841 {
|
|
|
9842 /* Append a space in place of the line break. */
|
|
|
9843 line_copy = concat_str(line, (char_u *)" ");
|
|
|
9844 line = line_copy;
|
|
|
9845 endcol = STRLEN(line);
|
|
|
9846 }
|
|
|
9847 }
|
|
|
9848 }
|
|
|
9849 else
|
|
|
9850 endcol = col;
|
|
|
9851
|
|
|
9852 if (endcol > 0)
|
|
|
9853 {
|
|
|
9854 /* Check if sentence ends before the bad word. */
|
|
|
9855 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
9856 regmatch.rm_ic = FALSE;
|
|
|
9857 p = line + endcol;
|
|
|
9858 for (;;)
|
|
|
9859 {
|
|
|
9860 mb_ptr_back(line, p);
|
|
|
9861 if (p == line || spell_iswordp_nmw(p))
|
|
|
9862 break;
|
|
|
9863 if (vim_regexec(®match, p, 0)
|
|
|
9864 && regmatch.endp[0] == line + endcol)
|
|
|
9865 {
|
|
|
9866 need_cap = TRUE;
|
|
|
9867 break;
|
|
|
9868 }
|
|
|
9869 }
|
|
|
9870 }
|
|
|
9871
|
|
|
9872 vim_free(line_copy);
|
|
|
9873
|
|
|
9874 return need_cap;
|
|
|
9875 }
|
|
|
9876
|
|
|
9877
|
|
|
9878 /*
|
|
372
|
9879 * ":spellrepall"
|
|
|
9880 */
|
|
|
9881 /*ARGSUSED*/
|
|
|
9882 void
|
|
|
9883 ex_spellrepall(eap)
|
|
|
9884 exarg_T *eap;
|
|
|
9885 {
|
|
|
9886 pos_T pos = curwin->w_cursor;
|
|
|
9887 char_u *frompat;
|
|
|
9888 int addlen;
|
|
|
9889 char_u *line;
|
|
|
9890 char_u *p;
|
|
|
9891 int save_ws = p_ws;
|
|
483
|
9892 linenr_T prev_lnum = 0;
|
|
372
|
9893
|
|
|
9894 if (repl_from == NULL || repl_to == NULL)
|
|
|
9895 {
|
|
|
9896 EMSG(_("E752: No previous spell replacement"));
|
|
|
9897 return;
|
|
|
9898 }
|
|
|
9899 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
9900
|
|
|
9901 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
9902 if (frompat == NULL)
|
|
|
9903 return;
|
|
|
9904 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
9905 p_ws = FALSE;
|
|
|
9906
|
|
483
|
9907 sub_nsubs = 0;
|
|
|
9908 sub_nlines = 0;
|
|
372
|
9909 curwin->w_cursor.lnum = 0;
|
|
|
9910 while (!got_int)
|
|
|
9911 {
|
|
|
9912 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
9913 || u_save_cursor() == FAIL)
|
|
|
9914 break;
|
|
|
9915
|
|
|
9916 /* Only replace when the right word isn't there yet. This happens
|
|
|
9917 * when changing "etc" to "etc.". */
|
|
|
9918 line = ml_get_curline();
|
|
|
9919 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
9920 repl_to, STRLEN(repl_to)) != 0)
|
|
|
9921 {
|
|
|
9922 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
9923 if (p == NULL)
|
|
|
9924 break;
|
|
|
9925 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
9926 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
9927 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
9928 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
9929 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
483
|
9930
|
|
|
9931 if (curwin->w_cursor.lnum != prev_lnum)
|
|
|
9932 {
|
|
|
9933 ++sub_nlines;
|
|
|
9934 prev_lnum = curwin->w_cursor.lnum;
|
|
|
9935 }
|
|
|
9936 ++sub_nsubs;
|
|
372
|
9937 }
|
|
|
9938 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
9939 }
|
|
|
9940
|
|
|
9941 p_ws = save_ws;
|
|
|
9942 curwin->w_cursor = pos;
|
|
|
9943 vim_free(frompat);
|
|
|
9944
|
|
483
|
9945 if (sub_nsubs == 0)
|
|
372
|
9946 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
483
|
9947 else
|
|
|
9948 do_sub_msg(FALSE);
|
|
372
|
9949 }
|
|
|
9950
|
|
|
9951 /*
|
|
344
|
9952 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
9953 * a list of allocated strings.
|
|
|
9954 */
|
|
|
9955 void
|
|
625
|
9956 spell_suggest_list(gap, word, maxcount, need_cap, interactive)
|
|
344
|
9957 garray_T *gap;
|
|
|
9958 char_u *word;
|
|
|
9959 int maxcount; /* maximum nr of suggestions */
|
|
475
|
9960 int need_cap; /* 'spellcapcheck' matched */
|
|
625
|
9961 int interactive;
|
|
344
|
9962 {
|
|
|
9963 suginfo_T sug;
|
|
|
9964 int i;
|
|
|
9965 suggest_T *stp;
|
|
|
9966 char_u *wcopy;
|
|
|
9967
|
|
638
|
9968 spell_find_suggest(word, 0, &sug, maxcount, FALSE, need_cap, interactive);
|
|
344
|
9969
|
|
|
9970 /* Make room in "gap". */
|
|
|
9971 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
626
|
9972 if (ga_grow(gap, sug.su_ga.ga_len) == OK)
|
|
|
9973 {
|
|
|
9974 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
9975 {
|
|
|
9976 stp = &SUG(sug.su_ga, i);
|
|
|
9977
|
|
|
9978 /* The suggested word may replace only part of "word", add the not
|
|
|
9979 * replaced part. */
|
|
|
9980 wcopy = alloc(stp->st_wordlen
|
|
344
|
9981 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
626
|
9982 if (wcopy == NULL)
|
|
|
9983 break;
|
|
|
9984 STRCPY(wcopy, stp->st_word);
|
|
|
9985 STRCPY(wcopy + stp->st_wordlen, sug.su_badptr + stp->st_orglen);
|
|
|
9986 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
9987 }
|
|
344
|
9988 }
|
|
|
9989
|
|
|
9990 spell_find_cleanup(&sug);
|
|
|
9991 }
|
|
|
9992
|
|
|
9993 /*
|
|
|
9994 * Find spell suggestions for the word at the start of "badptr".
|
|
|
9995 * Return the suggestions in "su->su_ga".
|
|
|
9996 * The maximum number of suggestions is "maxcount".
|
|
|
9997 * Note: does use info for the current window.
|
|
|
9998 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
9999 */
|
|
|
10000 static void
|
|
638
|
10001 spell_find_suggest(badptr, badlen, su, maxcount, banbadword, need_cap, interactive)
|
|
344
|
10002 char_u *badptr;
|
|
638
|
10003 int badlen; /* length of bad word or 0 if unknown */
|
|
344
|
10004 suginfo_T *su;
|
|
|
10005 int maxcount;
|
|
358
|
10006 int banbadword; /* don't include badword in suggestions */
|
|
388
|
10007 int need_cap; /* word should start with capital */
|
|
625
|
10008 int interactive;
|
|
344
|
10009 {
|
|
534
|
10010 hlf_T attr = HLF_COUNT;
|
|
372
|
10011 char_u buf[MAXPATHL];
|
|
|
10012 char_u *p;
|
|
|
10013 int do_combine = FALSE;
|
|
|
10014 char_u *sps_copy;
|
|
|
10015 #ifdef FEAT_EVAL
|
|
|
10016 static int expr_busy = FALSE;
|
|
|
10017 #endif
|
|
385
|
10018 int c;
|
|
503
|
10019 int i;
|
|
|
10020 langp_T *lp;
|
|
344
|
10021
|
|
|
10022 /*
|
|
|
10023 * Set the info in "*su".
|
|
|
10024 */
|
|
|
10025 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
10026 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
10027 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
10028 if (*badptr == NUL)
|
|
|
10029 return;
|
|
344
|
10030 hash_init(&su->su_banned);
|
|
|
10031
|
|
|
10032 su->su_badptr = badptr;
|
|
638
|
10033 if (badlen != 0)
|
|
|
10034 su->su_badlen = badlen;
|
|
|
10035 else
|
|
|
10036 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL, FALSE);
|
|
344
|
10037 su->su_maxcount = maxcount;
|
|
372
|
10038 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
10039
|
|
|
10040 if (su->su_badlen >= MAXWLEN)
|
|
|
10041 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
10042 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
10043 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
10044 su->su_fbadword, MAXWLEN);
|
|
346
|
10045 /* get caps flags for bad word */
|
|
474
|
10046 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
10047 su->su_badptr + su->su_badlen);
|
|
388
|
10048 if (need_cap)
|
|
|
10049 su->su_badflags |= WF_ONECAP;
|
|
344
|
10050
|
|
503
|
10051 /* Find the default language for sound folding. We simply use the first
|
|
|
10052 * one in 'spelllang' that supports sound folding. That's good for when
|
|
|
10053 * using multiple files for one language, it's not that bad when mixing
|
|
|
10054 * languages (e.g., "pl,en"). */
|
|
|
10055 for (i = 0; i < curbuf->b_langp.ga_len; ++i)
|
|
|
10056 {
|
|
|
10057 lp = LANGP_ENTRY(curbuf->b_langp, i);
|
|
|
10058 if (lp->lp_sallang != NULL)
|
|
|
10059 {
|
|
|
10060 su->su_sallang = lp->lp_sallang;
|
|
|
10061 break;
|
|
|
10062 }
|
|
|
10063 }
|
|
|
10064
|
|
534
|
10065 /* Soundfold the bad word with the default sound folding, so that we don't
|
|
|
10066 * have to do this many times. */
|
|
|
10067 if (su->su_sallang != NULL)
|
|
|
10068 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE,
|
|
|
10069 su->su_sal_badword);
|
|
|
10070
|
|
385
|
10071 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
10072 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
10073 * for that. */
|
|
455
|
10074 c = PTR2CHAR(su->su_badptr);
|
|
534
|
10075 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT)
|
|
385
|
10076 {
|
|
|
10077 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
10078 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
625
|
10079 0, TRUE, su->su_sallang, FALSE);
|
|
385
|
10080 }
|
|
|
10081
|
|
344
|
10082 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
10083 if (banbadword)
|
|
|
10084 add_banned(su, su->su_badword);
|
|
344
|
10085
|
|
372
|
10086 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
10087 sps_copy = vim_strsave(p_sps);
|
|
|
10088 if (sps_copy == NULL)
|
|
|
10089 return;
|
|
|
10090
|
|
|
10091 /* Loop over the items in 'spellsuggest'. */
|
|
|
10092 for (p = sps_copy; *p != NUL; )
|
|
|
10093 {
|
|
|
10094 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
10095
|
|
|
10096 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
10097 {
|
|
|
10098 #ifdef FEAT_EVAL
|
|
375
|
10099 /* Evaluate an expression. Skip this when called recursively,
|
|
|
10100 * when using spellsuggest() in the expression. */
|
|
372
|
10101 if (!expr_busy)
|
|
|
10102 {
|
|
|
10103 expr_busy = TRUE;
|
|
|
10104 spell_suggest_expr(su, buf + 5);
|
|
|
10105 expr_busy = FALSE;
|
|
|
10106 }
|
|
|
10107 #endif
|
|
|
10108 }
|
|
|
10109 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
10110 /* Use list of suggestions in a file. */
|
|
|
10111 spell_suggest_file(su, buf + 5);
|
|
|
10112 else
|
|
|
10113 {
|
|
|
10114 /* Use internal method. */
|
|
625
|
10115 spell_suggest_intern(su, interactive);
|
|
372
|
10116 if (sps_flags & SPS_DOUBLE)
|
|
|
10117 do_combine = TRUE;
|
|
|
10118 }
|
|
|
10119 }
|
|
|
10120
|
|
|
10121 vim_free(sps_copy);
|
|
|
10122
|
|
|
10123 if (do_combine)
|
|
|
10124 /* Combine the two list of suggestions. This must be done last,
|
|
|
10125 * because sorting changes the order again. */
|
|
|
10126 score_combine(su);
|
|
|
10127 }
|
|
|
10128
|
|
|
10129 #ifdef FEAT_EVAL
|
|
|
10130 /*
|
|
|
10131 * Find suggestions by evaluating expression "expr".
|
|
|
10132 */
|
|
|
10133 static void
|
|
|
10134 spell_suggest_expr(su, expr)
|
|
|
10135 suginfo_T *su;
|
|
|
10136 char_u *expr;
|
|
|
10137 {
|
|
|
10138 list_T *list;
|
|
|
10139 listitem_T *li;
|
|
|
10140 int score;
|
|
|
10141 char_u *p;
|
|
|
10142
|
|
|
10143 /* The work is split up in a few parts to avoid having to export
|
|
|
10144 * suginfo_T.
|
|
|
10145 * First evaluate the expression and get the resulting list. */
|
|
|
10146 list = eval_spell_expr(su->su_badword, expr);
|
|
|
10147 if (list != NULL)
|
|
|
10148 {
|
|
|
10149 /* Loop over the items in the list. */
|
|
|
10150 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
10151 if (li->li_tv.v_type == VAR_LIST)
|
|
|
10152 {
|
|
|
10153 /* Get the word and the score from the items. */
|
|
|
10154 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
625
|
10155 if (score >= 0 && score <= su->su_maxscore)
|
|
|
10156 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
10157 score, 0, TRUE, su->su_sallang, FALSE);
|
|
372
|
10158 }
|
|
|
10159 list_unref(list);
|
|
|
10160 }
|
|
|
10161
|
|
625
|
10162 /* Remove bogus suggestions, sort and truncate at "maxcount". */
|
|
|
10163 check_suggestions(su, &su->su_ga);
|
|
372
|
10164 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
10165 }
|
|
|
10166 #endif
|
|
|
10167
|
|
|
10168 /*
|
|
474
|
10169 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
10170 */
|
|
|
10171 static void
|
|
|
10172 spell_suggest_file(su, fname)
|
|
|
10173 suginfo_T *su;
|
|
|
10174 char_u *fname;
|
|
|
10175 {
|
|
|
10176 FILE *fd;
|
|
|
10177 char_u line[MAXWLEN * 2];
|
|
|
10178 char_u *p;
|
|
|
10179 int len;
|
|
|
10180 char_u cword[MAXWLEN];
|
|
|
10181
|
|
|
10182 /* Open the file. */
|
|
|
10183 fd = mch_fopen((char *)fname, "r");
|
|
|
10184 if (fd == NULL)
|
|
|
10185 {
|
|
|
10186 EMSG2(_(e_notopen), fname);
|
|
|
10187 return;
|
|
|
10188 }
|
|
|
10189
|
|
|
10190 /* Read it line by line. */
|
|
|
10191 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
10192 {
|
|
|
10193 line_breakcheck();
|
|
|
10194
|
|
|
10195 p = vim_strchr(line, '/');
|
|
|
10196 if (p == NULL)
|
|
|
10197 continue; /* No Tab found, just skip the line. */
|
|
|
10198 *p++ = NUL;
|
|
|
10199 if (STRICMP(su->su_badword, line) == 0)
|
|
|
10200 {
|
|
|
10201 /* Match! Isolate the good word, until CR or NL. */
|
|
|
10202 for (len = 0; p[len] >= ' '; ++len)
|
|
|
10203 ;
|
|
|
10204 p[len] = NUL;
|
|
|
10205
|
|
|
10206 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
10207 * of the bad word. */
|
|
|
10208 if (captype(p, NULL) == 0)
|
|
|
10209 {
|
|
|
10210 make_case_word(p, cword, su->su_badflags);
|
|
|
10211 p = cword;
|
|
|
10212 }
|
|
|
10213
|
|
|
10214 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
625
|
10215 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE);
|
|
372
|
10216 }
|
|
|
10217 }
|
|
|
10218
|
|
|
10219 fclose(fd);
|
|
|
10220
|
|
625
|
10221 /* Remove bogus suggestions, sort and truncate at "maxcount". */
|
|
|
10222 check_suggestions(su, &su->su_ga);
|
|
372
|
10223 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
10224 }
|
|
|
10225
|
|
|
10226 /*
|
|
|
10227 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
10228 */
|
|
|
10229 static void
|
|
625
|
10230 spell_suggest_intern(su, interactive)
|
|
372
|
10231 suginfo_T *su;
|
|
625
|
10232 int interactive;
|
|
|
10233 {
|
|
|
10234 /*
|
|
|
10235 * Load the .sug file(s) that are available and not done yet.
|
|
|
10236 */
|
|
|
10237 suggest_load_files();
|
|
|
10238
|
|
344
|
10239 /*
|
|
346
|
10240 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
10241 *
|
|
|
10242 * Set a maximum score to limit the combination of operations that is
|
|
|
10243 * tried.
|
|
|
10244 */
|
|
346
|
10245 suggest_try_special(su);
|
|
|
10246
|
|
|
10247 /*
|
|
|
10248 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
10249 * from the .aff file and inserting a space (split the word).
|
|
|
10250 */
|
|
|
10251 suggest_try_change(su);
|
|
344
|
10252
|
|
|
10253 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
10254 if (sps_flags & SPS_DOUBLE)
|
|
|
10255 score_comp_sal(su);
|
|
|
10256
|
|
|
10257 /*
|
|
346
|
10258 * 3. Try finding sound-a-like words.
|
|
344
|
10259 */
|
|
625
|
10260 if ((sps_flags & SPS_FAST) == 0)
|
|
|
10261 {
|
|
|
10262 if (sps_flags & SPS_BEST)
|
|
|
10263 /* Adjust the word score for the suggestions found so far for how
|
|
|
10264 * they sounds like. */
|
|
|
10265 rescore_suggestions(su);
|
|
|
10266
|
|
|
10267 /*
|
|
|
10268 * While going throught the soundfold tree "su_maxscore" is the score
|
|
|
10269 * for the soundfold word, limits the changes that are being tried,
|
|
|
10270 * and "su_sfmaxscore" the rescored score, which is set by
|
|
|
10271 * cleanup_suggestions().
|
|
|
10272 * First find words with a small edit distance, because this is much
|
|
|
10273 * faster and often already finds the top-N suggestions. If we didn't
|
|
|
10274 * find many suggestions try again with a higher edit distance.
|
|
|
10275 * "sl_sounddone" is used to avoid doing the same word twice.
|
|
|
10276 */
|
|
|
10277 suggest_try_soundalike_prep();
|
|
|
10278 su->su_maxscore = SCORE_SFMAX1;
|
|
|
10279 su->su_sfmaxscore = SCORE_MAXINIT * 3;
|
|
346
|
10280 suggest_try_soundalike(su);
|
|
625
|
10281 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su))
|
|
|
10282 {
|
|
|
10283 /* We didn't find enough matches, try again, allowing more
|
|
|
10284 * changes to the soundfold word. */
|
|
|
10285 su->su_maxscore = SCORE_SFMAX2;
|
|
|
10286 suggest_try_soundalike(su);
|
|
|
10287 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su))
|
|
|
10288 {
|
|
|
10289 /* Still didn't find enough matches, try again, allowing even
|
|
|
10290 * more changes to the soundfold word. */
|
|
|
10291 su->su_maxscore = SCORE_SFMAX3;
|
|
|
10292 suggest_try_soundalike(su);
|
|
|
10293 }
|
|
|
10294 }
|
|
|
10295 su->su_maxscore = su->su_sfmaxscore;
|
|
|
10296 suggest_try_soundalike_finish();
|
|
|
10297 }
|
|
|
10298
|
|
|
10299 /* When CTRL-C was hit while searching do show the results. Only clear
|
|
|
10300 * got_int when using a command, not for spellsuggest(). */
|
|
344
|
10301 ui_breakcheck();
|
|
625
|
10302 if (interactive && got_int)
|
|
344
|
10303 {
|
|
|
10304 (void)vgetc();
|
|
|
10305 got_int = FALSE;
|
|
|
10306 }
|
|
|
10307
|
|
372
|
10308 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
10309 {
|
|
|
10310 if (sps_flags & SPS_BEST)
|
|
|
10311 /* Adjust the word score for how it sounds like. */
|
|
|
10312 rescore_suggestions(su);
|
|
|
10313
|
|
625
|
10314 /* Remove bogus suggestions, sort and truncate at "maxcount". */
|
|
|
10315 check_suggestions(su, &su->su_ga);
|
|
372
|
10316 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
10317 }
|
|
|
10318 }
|
|
|
10319
|
|
|
10320 /*
|
|
625
|
10321 * Load the .sug files for languages that have one and weren't loaded yet.
|
|
|
10322 */
|
|
|
10323 static void
|
|
|
10324 suggest_load_files()
|
|
|
10325 {
|
|
|
10326 langp_T *lp;
|
|
|
10327 int lpi;
|
|
|
10328 slang_T *slang;
|
|
|
10329 char_u *dotp;
|
|
|
10330 FILE *fd;
|
|
|
10331 char_u buf[MAXWLEN];
|
|
|
10332 int i;
|
|
|
10333 time_t timestamp;
|
|
|
10334 int wcount;
|
|
|
10335 int wordnr;
|
|
|
10336 garray_T ga;
|
|
|
10337 int c;
|
|
|
10338
|
|
|
10339 /* Do this for all languages that support sound folding. */
|
|
|
10340 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
10341 {
|
|
|
10342 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
10343 slang = lp->lp_slang;
|
|
|
10344 if (slang->sl_sugtime != 0 && !slang->sl_sugloaded)
|
|
|
10345 {
|
|
|
10346 /* Change ".spl" to ".sug" and open the file. When the file isn't
|
|
|
10347 * found silently skip it. Do set "sl_sugloaded" so that we
|
|
|
10348 * don't try again and again. */
|
|
|
10349 slang->sl_sugloaded = TRUE;
|
|
|
10350
|
|
|
10351 dotp = vim_strrchr(slang->sl_fname, '.');
|
|
|
10352 if (dotp == NULL || fnamecmp(dotp, ".spl") != 0)
|
|
|
10353 continue;
|
|
|
10354 STRCPY(dotp, ".sug");
|
|
|
10355 fd = fopen((char *)slang->sl_fname, "r");
|
|
|
10356 if (fd == NULL)
|
|
|
10357 goto nextone;
|
|
|
10358
|
|
|
10359 /*
|
|
|
10360 * <SUGHEADER>: <fileID> <versionnr> <timestamp>
|
|
|
10361 */
|
|
|
10362 for (i = 0; i < VIMSUGMAGICL; ++i)
|
|
|
10363 buf[i] = getc(fd); /* <fileID> */
|
|
|
10364 if (STRNCMP(buf, VIMSUGMAGIC, VIMSUGMAGICL) != 0)
|
|
|
10365 {
|
|
626
|
10366 EMSG2(_("E778: This does not look like a .sug file: %s"),
|
|
625
|
10367 slang->sl_fname);
|
|
|
10368 goto nextone;
|
|
|
10369 }
|
|
|
10370 c = getc(fd); /* <versionnr> */
|
|
|
10371 if (c < VIMSUGVERSION)
|
|
|
10372 {
|
|
626
|
10373 EMSG2(_("E779: Old .sug file, needs to be updated: %s"),
|
|
625
|
10374 slang->sl_fname);
|
|
|
10375 goto nextone;
|
|
|
10376 }
|
|
|
10377 else if (c > VIMSUGVERSION)
|
|
|
10378 {
|
|
626
|
10379 EMSG2(_("E780: .sug file is for newer version of Vim: %s"),
|
|
625
|
10380 slang->sl_fname);
|
|
|
10381 goto nextone;
|
|
|
10382 }
|
|
|
10383
|
|
|
10384 /* Check the timestamp, it must be exactly the same as the one in
|
|
|
10385 * the .spl file. Otherwise the word numbers won't match. */
|
|
695
|
10386 timestamp = get8c(fd); /* <timestamp> */
|
|
625
|
10387 if (timestamp != slang->sl_sugtime)
|
|
|
10388 {
|
|
626
|
10389 EMSG2(_("E781: .sug file doesn't match .spl file: %s"),
|
|
625
|
10390 slang->sl_fname);
|
|
|
10391 goto nextone;
|
|
|
10392 }
|
|
|
10393
|
|
|
10394 /*
|
|
|
10395 * <SUGWORDTREE>: <wordtree>
|
|
|
10396 * Read the trie with the soundfolded words.
|
|
|
10397 */
|
|
|
10398 if (spell_read_tree(fd, &slang->sl_sbyts, &slang->sl_sidxs,
|
|
|
10399 FALSE, 0) != 0)
|
|
|
10400 {
|
|
|
10401 someerror:
|
|
626
|
10402 EMSG2(_("E782: error while reading .sug file: %s"),
|
|
625
|
10403 slang->sl_fname);
|
|
|
10404 slang_clear_sug(slang);
|
|
|
10405 goto nextone;
|
|
|
10406 }
|
|
|
10407
|
|
|
10408 /*
|
|
|
10409 * <SUGTABLE>: <sugwcount> <sugline> ...
|
|
|
10410 *
|
|
|
10411 * Read the table with word numbers. We use a file buffer for
|
|
|
10412 * this, because it's so much like a file with lines. Makes it
|
|
|
10413 * possible to swap the info and save on memory use.
|
|
|
10414 */
|
|
|
10415 slang->sl_sugbuf = open_spellbuf();
|
|
|
10416 if (slang->sl_sugbuf == NULL)
|
|
|
10417 goto someerror;
|
|
|
10418 /* <sugwcount> */
|
|
695
|
10419 wcount = get4c(fd);
|
|
625
|
10420 if (wcount < 0)
|
|
|
10421 goto someerror;
|
|
|
10422
|
|
|
10423 /* Read all the wordnr lists into the buffer, one NUL terminated
|
|
|
10424 * list per line. */
|
|
|
10425 ga_init2(&ga, 1, 100);
|
|
|
10426 for (wordnr = 0; wordnr < wcount; ++wordnr)
|
|
|
10427 {
|
|
|
10428 ga.ga_len = 0;
|
|
|
10429 for (;;)
|
|
|
10430 {
|
|
|
10431 c = getc(fd); /* <sugline> */
|
|
|
10432 if (c < 0 || ga_grow(&ga, 1) == FAIL)
|
|
|
10433 goto someerror;
|
|
|
10434 ((char_u *)ga.ga_data)[ga.ga_len++] = c;
|
|
|
10435 if (c == NUL)
|
|
|
10436 break;
|
|
|
10437 }
|
|
|
10438 if (ml_append_buf(slang->sl_sugbuf, (linenr_T)wordnr,
|
|
|
10439 ga.ga_data, ga.ga_len, TRUE) == FAIL)
|
|
|
10440 goto someerror;
|
|
|
10441 }
|
|
|
10442 ga_clear(&ga);
|
|
|
10443
|
|
|
10444 /*
|
|
|
10445 * Need to put word counts in the word tries, so that we can find
|
|
|
10446 * a word by its number.
|
|
|
10447 */
|
|
|
10448 tree_count_words(slang->sl_fbyts, slang->sl_fidxs);
|
|
|
10449 tree_count_words(slang->sl_sbyts, slang->sl_sidxs);
|
|
|
10450
|
|
|
10451 nextone:
|
|
|
10452 if (fd != NULL)
|
|
|
10453 fclose(fd);
|
|
|
10454 STRCPY(dotp, ".spl");
|
|
|
10455 }
|
|
|
10456 }
|
|
|
10457 }
|
|
|
10458
|
|
|
10459
|
|
|
10460 /*
|
|
|
10461 * Fill in the wordcount fields for a trie.
|
|
|
10462 * Returns the total number of words.
|
|
|
10463 */
|
|
|
10464 static void
|
|
|
10465 tree_count_words(byts, idxs)
|
|
|
10466 char_u *byts;
|
|
|
10467 idx_T *idxs;
|
|
|
10468 {
|
|
|
10469 int depth;
|
|
|
10470 idx_T arridx[MAXWLEN];
|
|
|
10471 int curi[MAXWLEN];
|
|
|
10472 int c;
|
|
|
10473 idx_T n;
|
|
|
10474 int wordcount[MAXWLEN];
|
|
|
10475
|
|
|
10476 arridx[0] = 0;
|
|
|
10477 curi[0] = 1;
|
|
|
10478 wordcount[0] = 0;
|
|
|
10479 depth = 0;
|
|
|
10480 while (depth >= 0 && !got_int)
|
|
|
10481 {
|
|
|
10482 if (curi[depth] > byts[arridx[depth]])
|
|
|
10483 {
|
|
|
10484 /* Done all bytes at this node, go up one level. */
|
|
|
10485 idxs[arridx[depth]] = wordcount[depth];
|
|
|
10486 if (depth > 0)
|
|
|
10487 wordcount[depth - 1] += wordcount[depth];
|
|
|
10488
|
|
|
10489 --depth;
|
|
|
10490 fast_breakcheck();
|
|
|
10491 }
|
|
|
10492 else
|
|
|
10493 {
|
|
|
10494 /* Do one more byte at this node. */
|
|
|
10495 n = arridx[depth] + curi[depth];
|
|
|
10496 ++curi[depth];
|
|
|
10497
|
|
|
10498 c = byts[n];
|
|
|
10499 if (c == 0)
|
|
|
10500 {
|
|
|
10501 /* End of word, count it. */
|
|
|
10502 ++wordcount[depth];
|
|
|
10503
|
|
|
10504 /* Skip over any other NUL bytes (same word with different
|
|
|
10505 * flags). */
|
|
|
10506 while (byts[n + 1] == 0)
|
|
|
10507 {
|
|
|
10508 ++n;
|
|
|
10509 ++curi[depth];
|
|
|
10510 }
|
|
|
10511 }
|
|
|
10512 else
|
|
|
10513 {
|
|
|
10514 /* Normal char, go one level deeper to count the words. */
|
|
|
10515 ++depth;
|
|
|
10516 arridx[depth] = idxs[n];
|
|
|
10517 curi[depth] = 1;
|
|
|
10518 wordcount[depth] = 0;
|
|
|
10519 }
|
|
|
10520 }
|
|
|
10521 }
|
|
|
10522 }
|
|
|
10523
|
|
|
10524 /*
|
|
344
|
10525 * Free the info put in "*su" by spell_find_suggest().
|
|
|
10526 */
|
|
|
10527 static void
|
|
|
10528 spell_find_cleanup(su)
|
|
|
10529 suginfo_T *su;
|
|
|
10530 {
|
|
|
10531 int i;
|
|
|
10532
|
|
|
10533 /* Free the suggestions. */
|
|
|
10534 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10535 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
10536 ga_clear(&su->su_ga);
|
|
|
10537 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
10538 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
10539 ga_clear(&su->su_sga);
|
|
323
|
10540
|
|
|
10541 /* Free the banned words. */
|
|
625
|
10542 hash_clear_all(&su->su_banned, 0);
|
|
323
|
10543 }
|
|
|
10544
|
|
|
10545 /*
|
|
324
|
10546 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
10547 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
10548 * The result is NUL terminated.
|
|
323
|
10549 */
|
|
|
10550 static void
|
|
324
|
10551 onecap_copy(word, wcopy, upper)
|
|
323
|
10552 char_u *word;
|
|
|
10553 char_u *wcopy;
|
|
|
10554 int upper; /* TRUE: first letter made upper case */
|
|
|
10555 {
|
|
|
10556 char_u *p;
|
|
|
10557 int c;
|
|
|
10558 int l;
|
|
|
10559
|
|
|
10560 p = word;
|
|
|
10561 #ifdef FEAT_MBYTE
|
|
|
10562 if (has_mbyte)
|
|
474
|
10563 c = mb_cptr2char_adv(&p);
|
|
323
|
10564 else
|
|
|
10565 #endif
|
|
|
10566 c = *p++;
|
|
|
10567 if (upper)
|
|
324
|
10568 c = SPELL_TOUPPER(c);
|
|
323
|
10569 else
|
|
324
|
10570 c = SPELL_TOFOLD(c);
|
|
323
|
10571 #ifdef FEAT_MBYTE
|
|
|
10572 if (has_mbyte)
|
|
|
10573 l = mb_char2bytes(c, wcopy);
|
|
|
10574 else
|
|
|
10575 #endif
|
|
|
10576 {
|
|
|
10577 l = 1;
|
|
|
10578 wcopy[0] = c;
|
|
|
10579 }
|
|
376
|
10580 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
10581 }
|
|
|
10582
|
|
|
10583 /*
|
|
324
|
10584 * Make a copy of "word" with all the letters upper cased into
|
|
|
10585 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
10586 */
|
|
|
10587 static void
|
|
|
10588 allcap_copy(word, wcopy)
|
|
|
10589 char_u *word;
|
|
|
10590 char_u *wcopy;
|
|
|
10591 {
|
|
|
10592 char_u *s;
|
|
|
10593 char_u *d;
|
|
|
10594 int c;
|
|
|
10595
|
|
|
10596 d = wcopy;
|
|
|
10597 for (s = word; *s != NUL; )
|
|
|
10598 {
|
|
|
10599 #ifdef FEAT_MBYTE
|
|
|
10600 if (has_mbyte)
|
|
474
|
10601 c = mb_cptr2char_adv(&s);
|
|
323
|
10602 else
|
|
|
10603 #endif
|
|
|
10604 c = *s++;
|
|
492
|
10605
|
|
|
10606 #ifdef FEAT_MBYTE
|
|
|
10607 /* We only change ß to SS when we are certain latin1 is used. It
|
|
|
10608 * would cause weird errors in other 8-bit encodings. */
|
|
|
10609 if (enc_latin1like && c == 0xdf)
|
|
|
10610 {
|
|
|
10611 c = 'S';
|
|
|
10612 if (d - wcopy >= MAXWLEN - 1)
|
|
|
10613 break;
|
|
|
10614 *d++ = c;
|
|
|
10615 }
|
|
|
10616 else
|
|
|
10617 #endif
|
|
|
10618 c = SPELL_TOUPPER(c);
|
|
323
|
10619
|
|
|
10620 #ifdef FEAT_MBYTE
|
|
|
10621 if (has_mbyte)
|
|
|
10622 {
|
|
|
10623 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
10624 break;
|
|
|
10625 d += mb_char2bytes(c, d);
|
|
|
10626 }
|
|
|
10627 else
|
|
|
10628 #endif
|
|
|
10629 {
|
|
|
10630 if (d - wcopy >= MAXWLEN - 1)
|
|
|
10631 break;
|
|
|
10632 *d++ = c;
|
|
|
10633 }
|
|
|
10634 }
|
|
|
10635 *d = NUL;
|
|
|
10636 }
|
|
|
10637
|
|
|
10638 /*
|
|
346
|
10639 * Try finding suggestions by recognizing specific situations.
|
|
|
10640 */
|
|
|
10641 static void
|
|
|
10642 suggest_try_special(su)
|
|
|
10643 suginfo_T *su;
|
|
|
10644 {
|
|
|
10645 char_u *p;
|
|
376
|
10646 size_t len;
|
|
346
|
10647 int c;
|
|
|
10648 char_u word[MAXWLEN];
|
|
|
10649
|
|
|
10650 /*
|
|
|
10651 * Recognize a word that is repeated: "the the".
|
|
|
10652 */
|
|
|
10653 p = skiptowhite(su->su_fbadword);
|
|
|
10654 len = p - su->su_fbadword;
|
|
|
10655 p = skipwhite(p);
|
|
|
10656 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
10657 {
|
|
|
10658 /* Include badflags: if the badword is onecap or allcap
|
|
|
10659 * use that for the goodword too: "The the" -> "The". */
|
|
|
10660 c = su->su_fbadword[len];
|
|
|
10661 su->su_fbadword[len] = NUL;
|
|
|
10662 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
10663 su->su_fbadword[len] = c;
|
|
534
|
10664
|
|
|
10665 /* Give a soundalike score of 0, compute the score as if deleting one
|
|
|
10666 * character. */
|
|
|
10667 add_suggestion(su, &su->su_ga, word, su->su_badlen,
|
|
625
|
10668 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE);
|
|
346
|
10669 }
|
|
|
10670 }
|
|
|
10671
|
|
|
10672 /*
|
|
323
|
10673 * Try finding suggestions by adding/removing/swapping letters.
|
|
|
10674 */
|
|
|
10675 static void
|
|
346
|
10676 suggest_try_change(su)
|
|
323
|
10677 suginfo_T *su;
|
|
|
10678 {
|
|
|
10679 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
625
|
10680 int n;
|
|
|
10681 char_u *p;
|
|
|
10682 int lpi;
|
|
323
|
10683 langp_T *lp;
|
|
|
10684
|
|
|
10685 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
10686 * to find matches (esp. REP items). Append some more text, changing
|
|
|
10687 * chars after the bad word may help. */
|
|
323
|
10688 STRCPY(fword, su->su_fbadword);
|
|
346
|
10689 n = STRLEN(fword);
|
|
|
10690 p = su->su_badptr + su->su_badlen;
|
|
|
10691 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
10692
|
|
503
|
10693 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
10694 {
|
|
|
10695 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
482
|
10696
|
|
500
|
10697 /* If reloading a spell file fails it's still in the list but
|
|
|
10698 * everything has been cleared. */
|
|
625
|
10699 if (lp->lp_slang->sl_fbyts == NULL)
|
|
500
|
10700 continue;
|
|
|
10701
|
|
625
|
10702 /* Try it for this language. Will add possible suggestions. */
|
|
|
10703 suggest_trie_walk(su, lp, fword, FALSE);
|
|
|
10704 }
|
|
|
10705 }
|
|
|
10706
|
|
|
10707 /* Check the maximum score, if we go over it we won't try this change. */
|
|
|
10708 #define TRY_DEEPER(su, stack, depth, add) \
|
|
|
10709 (stack[depth].ts_score + (add) < su->su_maxscore)
|
|
|
10710
|
|
|
10711 /*
|
|
|
10712 * Try finding suggestions by adding/removing/swapping letters.
|
|
|
10713 *
|
|
|
10714 * This uses a state machine. At each node in the tree we try various
|
|
|
10715 * operations. When trying if an operation works "depth" is increased and the
|
|
|
10716 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
10717 * character, replace one and delete another. The number of changes is
|
|
|
10718 * limited by su->su_maxscore.
|
|
|
10719 *
|
|
|
10720 * After implementing this I noticed an article by Kemal Oflazer that
|
|
|
10721 * describes something similar: "Error-tolerant Finite State Recognition with
|
|
|
10722 * Applications to Morphological Analysis and Spelling Correction" (1996).
|
|
|
10723 * The implementation in the article is simplified and requires a stack of
|
|
|
10724 * unknown depth. The implementation here only needs a stack depth equal to
|
|
|
10725 * the length of the word.
|
|
|
10726 *
|
|
|
10727 * This is also used for the sound-folded word, "soundfold" is TRUE then.
|
|
|
10728 * The mechanism is the same, but we find a match with a sound-folded word
|
|
|
10729 * that comes from one or more original words. Each of these words may be
|
|
|
10730 * added, this is done by add_sound_suggest().
|
|
|
10731 * Don't use:
|
|
|
10732 * the prefix tree or the keep-case tree
|
|
|
10733 * "su->su_badlen"
|
|
|
10734 * anything to do with upper and lower case
|
|
|
10735 * anything to do with word or non-word characters ("spell_iswordp()")
|
|
|
10736 * banned words
|
|
|
10737 * word flags (rare, region, compounding)
|
|
|
10738 * word splitting for now
|
|
|
10739 * "similar_chars()"
|
|
|
10740 * use "slang->sl_repsal" instead of "lp->lp_replang->sl_rep"
|
|
|
10741 */
|
|
|
10742 static void
|
|
|
10743 suggest_trie_walk(su, lp, fword, soundfold)
|
|
|
10744 suginfo_T *su;
|
|
|
10745 langp_T *lp;
|
|
|
10746 char_u *fword;
|
|
|
10747 int soundfold;
|
|
|
10748 {
|
|
|
10749 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
10750 trystate_T stack[MAXWLEN];
|
|
|
10751 char_u preword[MAXWLEN * 3]; /* word found with proper case;
|
|
|
10752 * concatanation of prefix compound
|
|
|
10753 * words and split word. NUL terminated
|
|
|
10754 * when going deeper but not when coming
|
|
|
10755 * back. */
|
|
|
10756 char_u compflags[MAXWLEN]; /* compound flags, one for each word */
|
|
|
10757 trystate_T *sp;
|
|
|
10758 int newscore;
|
|
|
10759 int score;
|
|
|
10760 char_u *byts, *fbyts, *pbyts;
|
|
|
10761 idx_T *idxs, *fidxs, *pidxs;
|
|
|
10762 int depth;
|
|
|
10763 int c, c2, c3;
|
|
|
10764 int n = 0;
|
|
|
10765 int flags;
|
|
|
10766 garray_T *gap;
|
|
|
10767 idx_T arridx;
|
|
|
10768 int len;
|
|
|
10769 char_u *p;
|
|
|
10770 fromto_T *ftp;
|
|
|
10771 int fl = 0, tl;
|
|
|
10772 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
|
10773 slang_T *slang = lp->lp_slang;
|
|
|
10774 int fword_ends;
|
|
|
10775 int goodword_ends;
|
|
|
10776 #ifdef DEBUG_TRIEWALK
|
|
|
10777 /* Stores the name of the change made at each level. */
|
|
|
10778 char_u changename[MAXWLEN][80];
|
|
|
10779 #endif
|
|
|
10780 int breakcheckcount = 1000;
|
|
|
10781 int compound_ok;
|
|
|
10782
|
|
|
10783 /*
|
|
|
10784 * Go through the whole case-fold tree, try changes at each node.
|
|
|
10785 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
10786 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
10787 * word).
|
|
|
10788 */
|
|
|
10789 depth = 0;
|
|
|
10790 sp = &stack[0];
|
|
|
10791 vim_memset(sp, 0, sizeof(trystate_T));
|
|
|
10792 sp->ts_curi = 1;
|
|
|
10793
|
|
|
10794 if (soundfold)
|
|
|
10795 {
|
|
|
10796 /* Going through the soundfold tree. */
|
|
|
10797 byts = fbyts = slang->sl_sbyts;
|
|
|
10798 idxs = fidxs = slang->sl_sidxs;
|
|
|
10799 pbyts = NULL;
|
|
|
10800 pidxs = NULL;
|
|
|
10801 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
|
10802 sp->ts_state = STATE_START;
|
|
|
10803 }
|
|
|
10804 else
|
|
|
10805 {
|
|
330
|
10806 /*
|
|
375
|
10807 * When there are postponed prefixes we need to use these first. At
|
|
|
10808 * the end of the prefix we continue in the case-fold tree.
|
|
|
10809 */
|
|
482
|
10810 fbyts = slang->sl_fbyts;
|
|
|
10811 fidxs = slang->sl_fidxs;
|
|
|
10812 pbyts = slang->sl_pbyts;
|
|
|
10813 pidxs = slang->sl_pidxs;
|
|
375
|
10814 if (pbyts != NULL)
|
|
|
10815 {
|
|
|
10816 byts = pbyts;
|
|
|
10817 idxs = pidxs;
|
|
482
|
10818 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
375
|
10819 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
10820 }
|
|
|
10821 else
|
|
|
10822 {
|
|
|
10823 byts = fbyts;
|
|
|
10824 idxs = fidxs;
|
|
482
|
10825 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
485
|
10826 sp->ts_state = STATE_START;
|
|
375
|
10827 }
|
|
625
|
10828 }
|
|
|
10829
|
|
|
10830 /*
|
|
|
10831 * Loop to find all suggestions. At each round we either:
|
|
|
10832 * - For the current state try one operation, advance "ts_curi",
|
|
|
10833 * increase "depth".
|
|
|
10834 * - When a state is done go to the next, set "ts_state".
|
|
|
10835 * - When all states are tried decrease "depth".
|
|
|
10836 */
|
|
|
10837 while (depth >= 0 && !got_int)
|
|
|
10838 {
|
|
|
10839 sp = &stack[depth];
|
|
|
10840 switch (sp->ts_state)
|
|
|
10841 {
|
|
|
10842 case STATE_START:
|
|
|
10843 case STATE_NOPREFIX:
|
|
|
10844 /*
|
|
|
10845 * Start of node: Deal with NUL bytes, which means
|
|
|
10846 * tword[] may end here.
|
|
|
10847 */
|
|
|
10848 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
10849 len = byts[arridx]; /* bytes in this node */
|
|
|
10850 arridx += sp->ts_curi; /* index of current byte */
|
|
|
10851
|
|
|
10852 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
|
|
|
10853 {
|
|
|
10854 /* Skip over the NUL bytes, we use them later. */
|
|
|
10855 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
10856 ;
|
|
|
10857 sp->ts_curi += n;
|
|
|
10858
|
|
|
10859 /* Always past NUL bytes now. */
|
|
|
10860 n = (int)sp->ts_state;
|
|
|
10861 sp->ts_state = STATE_ENDNUL;
|
|
|
10862 sp->ts_save_badflags = su->su_badflags;
|
|
|
10863
|
|
|
10864 /* At end of a prefix or at start of prefixtree: check for
|
|
|
10865 * following word. */
|
|
|
10866 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
|
10867 {
|
|
|
10868 /* Set su->su_badflags to the caps type at this position.
|
|
|
10869 * Use the caps type until here for the prefix itself. */
|
|
|
10870 #ifdef FEAT_MBYTE
|
|
|
10871 if (has_mbyte)
|
|
|
10872 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
10873 else
|
|
|
10874 #endif
|
|
|
10875 n = sp->ts_fidx;
|
|
|
10876 flags = badword_captype(su->su_badptr, su->su_badptr + n);
|
|
|
10877 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
10878 su->su_badptr + su->su_badlen);
|
|
625
|
10879 #ifdef DEBUG_TRIEWALK
|
|
|
10880 sprintf(changename[depth], "prefix");
|
|
|
10881 #endif
|
|
|
10882 go_deeper(stack, depth, 0);
|
|
|
10883 ++depth;
|
|
|
10884 sp = &stack[depth];
|
|
|
10885 sp->ts_prefixdepth = depth - 1;
|
|
|
10886 byts = fbyts;
|
|
|
10887 idxs = fidxs;
|
|
|
10888 sp->ts_arridx = 0;
|
|
|
10889
|
|
|
10890 /* Move the prefix to preword[] with the right case
|
|
|
10891 * and make find_keepcap_word() works. */
|
|
|
10892 tword[sp->ts_twordlen] = NUL;
|
|
|
10893 make_case_word(tword + sp->ts_splitoff,
|
|
|
10894 preword + sp->ts_prewordlen, flags);
|
|
|
10895 sp->ts_prewordlen = STRLEN(preword);
|
|
|
10896 sp->ts_splitoff = sp->ts_twordlen;
|
|
|
10897 }
|
|
|
10898 break;
|
|
|
10899 }
|
|
|
10900
|
|
|
10901 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
|
10902 {
|
|
|
10903 /* Past bytes in node and/or past NUL bytes. */
|
|
|
10904 sp->ts_state = STATE_ENDNUL;
|
|
|
10905 sp->ts_save_badflags = su->su_badflags;
|
|
|
10906 break;
|
|
|
10907 }
|
|
|
10908
|
|
|
10909 /*
|
|
|
10910 * End of word in tree.
|
|
|
10911 */
|
|
|
10912 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
10913
|
|
|
10914 flags = (int)idxs[arridx];
|
|
698
|
10915
|
|
|
10916 /* Skip words with the NOSUGGEST flag. */
|
|
|
10917 if (flags & WF_NOSUGGEST)
|
|
|
10918 break;
|
|
|
10919
|
|
625
|
10920 fword_ends = (fword[sp->ts_fidx] == NUL
|
|
|
10921 || (soundfold
|
|
|
10922 ? vim_iswhite(fword[sp->ts_fidx])
|
|
|
10923 : !spell_iswordp(fword + sp->ts_fidx, curbuf)));
|
|
|
10924 tword[sp->ts_twordlen] = NUL;
|
|
|
10925
|
|
|
10926 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL
|
|
485
|
10927 && (sp->ts_flags & TSF_PREFIXOK) == 0)
|
|
625
|
10928 {
|
|
|
10929 /* There was a prefix before the word. Check that the prefix
|
|
|
10930 * can be used with this word. */
|
|
|
10931 /* Count the length of the NULs in the prefix. If there are
|
|
|
10932 * none this must be the first try without a prefix. */
|
|
|
10933 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
10934 len = pbyts[n++];
|
|
|
10935 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
10936 ;
|
|
|
10937 if (c > 0)
|
|
|
10938 {
|
|
|
10939 c = valid_word_prefix(c, n, flags,
|
|
483
|
10940 tword + sp->ts_splitoff, slang, FALSE);
|
|
625
|
10941 if (c == 0)
|
|
|
10942 break;
|
|
|
10943
|
|
|
10944 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
10945 if (c & WF_RAREPFX)
|
|
|
10946 flags |= WF_RARE;
|
|
|
10947
|
|
|
10948 /* Tricky: when checking for both prefix and compounding
|
|
|
10949 * we run into the prefix flag first.
|
|
|
10950 * Remember that it's OK, so that we accept the prefix
|
|
|
10951 * when arriving at a compound flag. */
|
|
|
10952 sp->ts_flags |= TSF_PREFIXOK;
|
|
|
10953 }
|
|
|
10954 }
|
|
|
10955
|
|
|
10956 /* Check NEEDCOMPOUND: can't use word without compounding. Do try
|
|
|
10957 * appending another compound word below. */
|
|
|
10958 if (sp->ts_complen == sp->ts_compsplit && fword_ends
|
|
500
|
10959 && (flags & WF_NEEDCOMP))
|
|
625
|
10960 goodword_ends = FALSE;
|
|
|
10961 else
|
|
|
10962 goodword_ends = TRUE;
|
|
|
10963
|
|
|
10964 p = NULL;
|
|
|
10965 compound_ok = TRUE;
|
|
|
10966 if (sp->ts_complen > sp->ts_compsplit)
|
|
|
10967 {
|
|
|
10968 if (slang->sl_nobreak)
|
|
|
10969 {
|
|
|
10970 /* There was a word before this word. When there was no
|
|
|
10971 * change in this word (it was correct) add the first word
|
|
|
10972 * as a suggestion. If this word was corrected too, we
|
|
|
10973 * need to check if a correct word follows. */
|
|
|
10974 if (sp->ts_fidx - sp->ts_splitfidx
|
|
|
10975 == sp->ts_twordlen - sp->ts_splitoff
|
|
|
10976 && STRNCMP(fword + sp->ts_splitfidx,
|
|
|
10977 tword + sp->ts_splitoff,
|
|
|
10978 sp->ts_fidx - sp->ts_splitfidx) == 0)
|
|
492
|
10979 {
|
|
625
|
10980 preword[sp->ts_prewordlen] = NUL;
|
|
|
10981 newscore = score_wordcount_adj(slang, sp->ts_score,
|
|
|
10982 preword + sp->ts_prewordlen,
|
|
|
10983 sp->ts_prewordlen > 0);
|
|
|
10984 /* Add the suggestion if the score isn't too bad. */
|
|
|
10985 if (newscore <= su->su_maxscore)
|
|
492
|
10986 add_suggestion(su, &su->su_ga, preword,
|
|
503
|
10987 sp->ts_splitfidx - repextra,
|
|
625
|
10988 newscore, 0, FALSE,
|
|
|
10989 lp->lp_sallang, FALSE);
|
|
|
10990 break;
|
|
492
|
10991 }
|
|
490
|
10992 }
|
|
|
10993 else
|
|
625
|
10994 {
|
|
|
10995 /* There was a compound word before this word. If this
|
|
|
10996 * word does not support compounding then give up
|
|
|
10997 * (splitting is tried for the word without compound
|
|
|
10998 * flag). */
|
|
|
10999 if (((unsigned)flags >> 24) == 0
|
|
|
11000 || sp->ts_twordlen - sp->ts_splitoff
|
|
|
11001 < slang->sl_compminlen)
|
|
|
11002 break;
|
|
|
11003 #ifdef FEAT_MBYTE
|
|
|
11004 /* For multi-byte chars check character length against
|
|
|
11005 * COMPOUNDMIN. */
|
|
|
11006 if (has_mbyte
|
|
|
11007 && slang->sl_compminlen > 0
|
|
|
11008 && mb_charlen(tword + sp->ts_splitoff)
|
|
|
11009 < slang->sl_compminlen)
|
|
|
11010 break;
|
|
|
11011 #endif
|
|
|
11012
|
|
|
11013 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
11014 compflags[sp->ts_complen + 1] = NUL;
|
|
|
11015 vim_strncpy(preword + sp->ts_prewordlen,
|
|
|
11016 tword + sp->ts_splitoff,
|
|
|
11017 sp->ts_twordlen - sp->ts_splitoff);
|
|
|
11018 p = preword;
|
|
|
11019 while (*skiptowhite(p) != NUL)
|
|
|
11020 p = skipwhite(skiptowhite(p));
|
|
|
11021 if (fword_ends && !can_compound(slang, p,
|
|
|
11022 compflags + sp->ts_compsplit))
|
|
|
11023 /* Compound is not allowed. But it may still be
|
|
|
11024 * possible if we add another (short) word. */
|
|
|
11025 compound_ok = FALSE;
|
|
|
11026
|
|
|
11027 /* Get pointer to last char of previous word. */
|
|
|
11028 p = preword + sp->ts_prewordlen;
|
|
|
11029 mb_ptr_back(preword, p);
|
|
|
11030 }
|
|
|
11031 }
|
|
|
11032
|
|
|
11033 /*
|
|
|
11034 * Form the word with proper case in preword.
|
|
|
11035 * If there is a word from a previous split, append.
|
|
|
11036 * For the soundfold tree don't change the case, simply append.
|
|
|
11037 */
|
|
|
11038 if (soundfold)
|
|
|
11039 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff);
|
|
|
11040 else if (flags & WF_KEEPCAP)
|
|
|
11041 /* Must find the word in the keep-case tree. */
|
|
|
11042 find_keepcap_word(slang, tword + sp->ts_splitoff,
|
|
483
|
11043 preword + sp->ts_prewordlen);
|
|
625
|
11044 else
|
|
|
11045 {
|
|
|
11046 /* Include badflags: If the badword is onecap or allcap
|
|
|
11047 * use that for the goodword too. But if the badword is
|
|
|
11048 * allcap and it's only one char long use onecap. */
|
|
|
11049 c = su->su_badflags;
|
|
|
11050 if ((c & WF_ALLCAP)
|
|
|
11051 #ifdef FEAT_MBYTE
|
|
|
11052 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
11053 #else
|
|
625
|
11054 && su->su_badlen == 1
|
|
|
11055 #endif
|
|
|
11056 )
|
|
|
11057 c = WF_ONECAP;
|
|
|
11058 c |= flags;
|
|
|
11059
|
|
|
11060 /* When appending a compound word after a word character don't
|
|
|
11061 * use Onecap. */
|
|
|
11062 if (p != NULL && spell_iswordp_nmw(p))
|
|
|
11063 c &= ~WF_ONECAP;
|
|
|
11064 make_case_word(tword + sp->ts_splitoff,
|
|
490
|
11065 preword + sp->ts_prewordlen, c);
|
|
625
|
11066 }
|
|
|
11067
|
|
|
11068 if (!soundfold)
|
|
|
11069 {
|
|
323
|
11070 /* Don't use a banned word. It may appear again as a good
|
|
|
11071 * word, thus remember it. */
|
|
|
11072 if (flags & WF_BANNED)
|
|
|
11073 {
|
|
483
|
11074 add_banned(su, preword + sp->ts_prewordlen);
|
|
323
|
11075 break;
|
|
|
11076 }
|
|
534
|
11077 if ((sp->ts_complen == sp->ts_compsplit
|
|
625
|
11078 && WAS_BANNED(su, preword + sp->ts_prewordlen))
|
|
|
11079 || WAS_BANNED(su, preword))
|
|
534
|
11080 {
|
|
|
11081 if (slang->sl_compprog == NULL)
|
|
|
11082 break;
|
|
|
11083 /* the word so far was banned but we may try compounding */
|
|
|
11084 goodword_ends = FALSE;
|
|
|
11085 }
|
|
625
|
11086 }
|
|
|
11087
|
|
|
11088 newscore = 0;
|
|
|
11089 if (!soundfold) /* soundfold words don't have flags */
|
|
|
11090 {
|
|
323
|
11091 if ((flags & WF_REGION)
|
|
390
|
11092 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
11093 newscore += SCORE_REGION;
|
|
|
11094 if (flags & WF_RARE)
|
|
|
11095 newscore += SCORE_RARE;
|
|
|
11096
|
|
346
|
11097 if (!spell_valid_case(su->su_badflags,
|
|
483
|
11098 captype(preword + sp->ts_prewordlen, NULL)))
|
|
323
|
11099 newscore += SCORE_ICASE;
|
|
625
|
11100 }
|
|
|
11101
|
|
|
11102 /* TODO: how about splitting in the soundfold tree? */
|
|
|
11103 if (fword_ends
|
|
|
11104 && goodword_ends
|
|
|
11105 && sp->ts_fidx >= sp->ts_fidxtry
|
|
|
11106 && compound_ok)
|
|
|
11107 {
|
|
|
11108 /* The badword also ends: add suggestions. */
|
|
|
11109 #ifdef DEBUG_TRIEWALK
|
|
|
11110 if (soundfold && STRCMP(preword, "smwrd") == 0)
|
|
|
11111 {
|
|
|
11112 int j;
|
|
|
11113
|
|
|
11114 /* print the stack of changes that brought us here */
|
|
|
11115 smsg("------ %s -------", fword);
|
|
|
11116 for (j = 0; j < depth; ++j)
|
|
|
11117 smsg("%s", changename[j]);
|
|
|
11118 }
|
|
|
11119 #endif
|
|
|
11120 if (soundfold)
|
|
|
11121 {
|
|
|
11122 /* For soundfolded words we need to find the original
|
|
|
11123 * words, the edit distrance and then add them. */
|
|
|
11124 add_sound_suggest(su, preword, sp->ts_score, lp);
|
|
|
11125 }
|
|
|
11126 else
|
|
|
11127 {
|
|
|
11128 /* Give a penalty when changing non-word char to word
|
|
|
11129 * char, e.g., "thes," -> "these". */
|
|
366
|
11130 p = fword + sp->ts_fidx;
|
|
625
|
11131 mb_ptr_back(fword, p);
|
|
376
|
11132 if (!spell_iswordp(p, curbuf))
|
|
366
|
11133 {
|
|
|
11134 p = preword + STRLEN(preword);
|
|
625
|
11135 mb_ptr_back(preword, p);
|
|
376
|
11136 if (spell_iswordp(p, curbuf))
|
|
366
|
11137 newscore += SCORE_NONWORD;
|
|
|
11138 }
|
|
|
11139
|
|
625
|
11140 /* Give a bonus to words seen before. */
|
|
|
11141 score = score_wordcount_adj(slang,
|
|
|
11142 sp->ts_score + newscore,
|
|
|
11143 preword + sp->ts_prewordlen,
|
|
|
11144 sp->ts_prewordlen > 0);
|
|
|
11145
|
|
|
11146 /* Add the suggestion if the score isn't too bad. */
|
|
|
11147 if (score <= su->su_maxscore)
|
|
633
|
11148 {
|
|
625
|
11149 add_suggestion(su, &su->su_ga, preword,
|
|
|
11150 sp->ts_fidx - repextra,
|
|
|
11151 score, 0, FALSE, lp->lp_sallang, FALSE);
|
|
633
|
11152
|
|
|
11153 if (su->su_badflags & WF_MIXCAP)
|
|
|
11154 {
|
|
|
11155 /* We really don't know if the word should be
|
|
|
11156 * upper or lower case, add both. */
|
|
|
11157 c = captype(preword, NULL);
|
|
|
11158 if (c == 0 || c == WF_ALLCAP)
|
|
|
11159 {
|
|
|
11160 make_case_word(tword + sp->ts_splitoff,
|
|
|
11161 preword + sp->ts_prewordlen,
|
|
|
11162 c == 0 ? WF_ALLCAP : 0);
|
|
|
11163
|
|
|
11164 add_suggestion(su, &su->su_ga, preword,
|
|
|
11165 sp->ts_fidx - repextra,
|
|
|
11166 score + SCORE_ICASE, 0, FALSE,
|
|
|
11167 lp->lp_sallang, FALSE);
|
|
|
11168 }
|
|
|
11169 }
|
|
|
11170 }
|
|
625
|
11171 }
|
|
|
11172 }
|
|
|
11173
|
|
|
11174 /*
|
|
|
11175 * Try word split and/or compounding.
|
|
|
11176 */
|
|
|
11177 if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
|
|
|
11178 #ifdef FEAT_MBYTE
|
|
|
11179 /* Don't split halfway a character. */
|
|
|
11180 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
11181 #endif
|
|
|
11182 )
|
|
|
11183 {
|
|
|
11184 int try_compound;
|
|
|
11185 int try_split;
|
|
|
11186
|
|
|
11187 /* If past the end of the bad word don't try a split.
|
|
|
11188 * Otherwise try changing the next word. E.g., find
|
|
|
11189 * suggestions for "the the" where the second "the" is
|
|
|
11190 * different. It's done like a split.
|
|
|
11191 * TODO: word split for soundfold words */
|
|
|
11192 try_split = (sp->ts_fidx - repextra < su->su_badlen)
|
|
|
11193 && !soundfold;
|
|
|
11194
|
|
|
11195 /* Get here in several situations:
|
|
|
11196 * 1. The word in the tree ends:
|
|
|
11197 * If the word allows compounding try that. Otherwise try
|
|
|
11198 * a split by inserting a space. For both check that a
|
|
|
11199 * valid words starts at fword[sp->ts_fidx].
|
|
|
11200 * For NOBREAK do like compounding to be able to check if
|
|
|
11201 * the next word is valid.
|
|
|
11202 * 2. The badword does end, but it was due to a change (e.g.,
|
|
|
11203 * a swap). No need to split, but do check that the
|
|
|
11204 * following word is valid.
|
|
|
11205 * 3. The badword and the word in the tree end. It may still
|
|
|
11206 * be possible to compound another (short) word.
|
|
|
11207 */
|
|
|
11208 try_compound = FALSE;
|
|
|
11209 if (!soundfold
|
|
|
11210 && slang->sl_compprog != NULL
|
|
|
11211 && ((unsigned)flags >> 24) != 0
|
|
|
11212 && sp->ts_twordlen - sp->ts_splitoff
|
|
|
11213 >= slang->sl_compminlen
|
|
|
11214 #ifdef FEAT_MBYTE
|
|
|
11215 && (!has_mbyte
|
|
|
11216 || slang->sl_compminlen == 0
|
|
|
11217 || mb_charlen(tword + sp->ts_splitoff)
|
|
|
11218 >= slang->sl_compminlen)
|
|
|
11219 #endif
|
|
|
11220 && (slang->sl_compsylmax < MAXWLEN
|
|
|
11221 || sp->ts_complen + 1 - sp->ts_compsplit
|
|
|
11222 < slang->sl_compmax)
|
|
|
11223 && (byte_in_str(sp->ts_complen == sp->ts_compsplit
|
|
|
11224 ? slang->sl_compstartflags
|
|
|
11225 : slang->sl_compallflags,
|
|
|
11226 ((unsigned)flags >> 24))))
|
|
|
11227 {
|
|
|
11228 try_compound = TRUE;
|
|
|
11229 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
11230 compflags[sp->ts_complen + 1] = NUL;
|
|
|
11231 }
|
|
|
11232
|
|
|
11233 /* For NOBREAK we never try splitting, it won't make any word
|
|
|
11234 * valid. */
|
|
|
11235 if (slang->sl_nobreak)
|
|
|
11236 try_compound = TRUE;
|
|
|
11237
|
|
|
11238 /* If we could add a compound word, and it's also possible to
|
|
|
11239 * split at this point, do the split first and set
|
|
|
11240 * TSF_DIDSPLIT to avoid doing it again. */
|
|
|
11241 else if (!fword_ends
|
|
|
11242 && try_compound
|
|
|
11243 && (sp->ts_flags & TSF_DIDSPLIT) == 0)
|
|
|
11244 {
|
|
485
|
11245 try_compound = FALSE;
|
|
625
|
11246 sp->ts_flags |= TSF_DIDSPLIT;
|
|
|
11247 --sp->ts_curi; /* do the same NUL again */
|
|
|
11248 compflags[sp->ts_complen] = NUL;
|
|
|
11249 }
|
|
|
11250 else
|
|
|
11251 sp->ts_flags &= ~TSF_DIDSPLIT;
|
|
|
11252
|
|
|
11253 if (try_split || try_compound)
|
|
|
11254 {
|
|
534
|
11255 if (!try_compound && (!fword_ends || !goodword_ends))
|
|
485
|
11256 {
|
|
|
11257 /* If we're going to split need to check that the
|
|
501
|
11258 * words so far are valid for compounding. If there
|
|
|
11259 * is only one word it must not have the NEEDCOMPOUND
|
|
|
11260 * flag. */
|
|
|
11261 if (sp->ts_complen == sp->ts_compsplit
|
|
|
11262 && (flags & WF_NEEDCOMP))
|
|
|
11263 break;
|
|
490
|
11264 p = preword;
|
|
|
11265 while (*skiptowhite(p) != NUL)
|
|
|
11266 p = skipwhite(skiptowhite(p));
|
|
485
|
11267 if (sp->ts_complen > sp->ts_compsplit
|
|
490
|
11268 && !can_compound(slang, p,
|
|
485
|
11269 compflags + sp->ts_compsplit))
|
|
|
11270 break;
|
|
698
|
11271
|
|
|
11272 if (slang->sl_nosplitsugs)
|
|
|
11273 newscore += SCORE_SPLIT_NO;
|
|
|
11274 else
|
|
|
11275 newscore += SCORE_SPLIT;
|
|
625
|
11276
|
|
|
11277 /* Give a bonus to words seen before. */
|
|
|
11278 newscore = score_wordcount_adj(slang, newscore,
|
|
|
11279 preword + sp->ts_prewordlen, TRUE);
|
|
482
|
11280 }
|
|
|
11281
|
|
625
|
11282 if (TRY_DEEPER(su, stack, depth, newscore))
|
|
323
|
11283 {
|
|
625
|
11284 go_deeper(stack, depth, newscore);
|
|
|
11285 #ifdef DEBUG_TRIEWALK
|
|
|
11286 if (!try_compound && !fword_ends)
|
|
|
11287 sprintf(changename[depth], "%.*s-%s: split",
|
|
|
11288 sp->ts_twordlen, tword, fword + sp->ts_fidx);
|
|
|
11289 else
|
|
|
11290 sprintf(changename[depth], "%.*s-%s: compound",
|
|
|
11291 sp->ts_twordlen, tword, fword + sp->ts_fidx);
|
|
|
11292 #endif
|
|
323
|
11293 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
346
|
11294 sp->ts_save_badflags = su->su_badflags;
|
|
376
|
11295 sp->ts_state = STATE_SPLITUNDO;
|
|
|
11296
|
|
|
11297 ++depth;
|
|
|
11298 sp = &stack[depth];
|
|
323
|
11299
|
|
482
|
11300 /* Append a space to preword when splitting. */
|
|
|
11301 if (!try_compound && !fword_ends)
|
|
|
11302 STRCAT(preword, " ");
|
|
483
|
11303 sp->ts_prewordlen = STRLEN(preword);
|
|
|
11304 sp->ts_splitoff = sp->ts_twordlen;
|
|
492
|
11305 sp->ts_splitfidx = sp->ts_fidx;
|
|
376
|
11306
|
|
|
11307 /* If the badword has a non-word character at this
|
|
|
11308 * position skip it. That means replacing the
|
|
482
|
11309 * non-word character with a space. Always skip a
|
|
534
|
11310 * character when the word ends. But only when the
|
|
|
11311 * good word can end. */
|
|
625
|
11312 if (((!try_compound && !spell_iswordp_nmw(fword
|
|
|
11313 + sp->ts_fidx))
|
|
|
11314 || fword_ends)
|
|
|
11315 && fword[sp->ts_fidx] != NUL
|
|
|
11316 && goodword_ends)
|
|
376
|
11317 {
|
|
482
|
11318 int l;
|
|
|
11319
|
|
376
|
11320 #ifdef FEAT_MBYTE
|
|
|
11321 if (has_mbyte)
|
|
482
|
11322 l = MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
376
|
11323 else
|
|
|
11324 #endif
|
|
482
|
11325 l = 1;
|
|
|
11326 if (fword_ends)
|
|
|
11327 {
|
|
|
11328 /* Copy the skipped character to preword. */
|
|
483
|
11329 mch_memmove(preword + sp->ts_prewordlen,
|
|
482
|
11330 fword + sp->ts_fidx, l);
|
|
483
|
11331 sp->ts_prewordlen += l;
|
|
|
11332 preword[sp->ts_prewordlen] = NUL;
|
|
482
|
11333 }
|
|
|
11334 else
|
|
|
11335 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
11336 sp->ts_fidx += l;
|
|
376
|
11337 }
|
|
455
|
11338
|
|
485
|
11339 /* When compounding include compound flag in
|
|
|
11340 * compflags[] (already set above). When splitting we
|
|
|
11341 * may start compounding over again. */
|
|
482
|
11342 if (try_compound)
|
|
483
|
11343 ++sp->ts_complen;
|
|
482
|
11344 else
|
|
485
|
11345 sp->ts_compsplit = sp->ts_complen;
|
|
|
11346 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
482
|
11347
|
|
455
|
11348 /* set su->su_badflags to the caps type at this
|
|
|
11349 * position */
|
|
324
|
11350 #ifdef FEAT_MBYTE
|
|
|
11351 if (has_mbyte)
|
|
455
|
11352 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
11353 else
|
|
|
11354 #endif
|
|
455
|
11355 n = sp->ts_fidx;
|
|
474
|
11356 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
11357 su->su_badptr + su->su_badlen);
|
|
323
|
11358
|
|
|
11359 /* Restart at top of the tree. */
|
|
376
|
11360 sp->ts_arridx = 0;
|
|
485
|
11361
|
|
|
11362 /* If there are postponed prefixes, try these too. */
|
|
|
11363 if (pbyts != NULL)
|
|
|
11364 {
|
|
|
11365 byts = pbyts;
|
|
|
11366 idxs = pidxs;
|
|
|
11367 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
|
11368 sp->ts_state = STATE_NOPREFIX;
|
|
|
11369 }
|
|
323
|
11370 }
|
|
|
11371 }
|
|
625
|
11372 }
|
|
|
11373 break;
|
|
|
11374
|
|
|
11375 case STATE_SPLITUNDO:
|
|
|
11376 /* Undo the changes done for word split or compound word. */
|
|
|
11377 su->su_badflags = sp->ts_save_badflags;
|
|
|
11378
|
|
|
11379 /* Continue looking for NUL bytes. */
|
|
|
11380 sp->ts_state = STATE_START;
|
|
|
11381
|
|
|
11382 /* In case we went into the prefix tree. */
|
|
|
11383 byts = fbyts;
|
|
|
11384 idxs = fidxs;
|
|
|
11385 break;
|
|
|
11386
|
|
|
11387 case STATE_ENDNUL:
|
|
|
11388 /* Past the NUL bytes in the node. */
|
|
|
11389 su->su_badflags = sp->ts_save_badflags;
|
|
|
11390 if (fword[sp->ts_fidx] == NUL
|
|
|
11391 #ifdef FEAT_MBYTE
|
|
|
11392 && sp->ts_tcharlen == 0
|
|
|
11393 #endif
|
|
|
11394 )
|
|
|
11395 {
|
|
|
11396 /* The badword ends, can't use STATE_PLAIN. */
|
|
|
11397 sp->ts_state = STATE_DEL;
|
|
323
|
11398 break;
|
|
625
|
11399 }
|
|
|
11400 sp->ts_state = STATE_PLAIN;
|
|
|
11401 /*FALLTHROUGH*/
|
|
|
11402
|
|
|
11403 case STATE_PLAIN:
|
|
|
11404 /*
|
|
|
11405 * Go over all possible bytes at this node, add each to tword[]
|
|
|
11406 * and use child node. "ts_curi" is the index.
|
|
|
11407 */
|
|
|
11408 arridx = sp->ts_arridx;
|
|
|
11409 if (sp->ts_curi > byts[arridx])
|
|
|
11410 {
|
|
|
11411 /* Done all bytes at this node, do next state. When still at
|
|
|
11412 * already changed bytes skip the other tricks. */
|
|
|
11413 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
11414 sp->ts_state = STATE_DEL;
|
|
|
11415 else
|
|
625
|
11416 sp->ts_state = STATE_FINAL;
|
|
|
11417 }
|
|
|
11418 else
|
|
|
11419 {
|
|
|
11420 arridx += sp->ts_curi++;
|
|
|
11421 c = byts[arridx];
|
|
|
11422
|
|
|
11423 /* Normal byte, go one level deeper. If it's not equal to the
|
|
|
11424 * byte in the bad word adjust the score. But don't even try
|
|
|
11425 * when the byte was already changed. And don't try when we
|
|
|
11426 * just deleted this byte, accepting it is always cheaper then
|
|
|
11427 * delete + substitute. */
|
|
|
11428 if (c == fword[sp->ts_fidx]
|
|
|
11429 #ifdef FEAT_MBYTE
|
|
|
11430 || (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE)
|
|
|
11431 #endif
|
|
|
11432 )
|
|
|
11433 newscore = 0;
|
|
|
11434 else
|
|
|
11435 newscore = SCORE_SUBST;
|
|
|
11436 if ((newscore == 0
|
|
|
11437 || (sp->ts_fidx >= sp->ts_fidxtry
|
|
|
11438 && ((sp->ts_flags & TSF_DIDDEL) == 0
|
|
|
11439 || c != fword[sp->ts_delidx])))
|
|
|
11440 && TRY_DEEPER(su, stack, depth, newscore))
|
|
|
11441 {
|
|
|
11442 go_deeper(stack, depth, newscore);
|
|
|
11443 #ifdef DEBUG_TRIEWALK
|
|
|
11444 if (newscore > 0)
|
|
|
11445 sprintf(changename[depth], "%.*s-%s: subst %c to %c",
|
|
|
11446 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11447 fword[sp->ts_fidx], c);
|
|
323
|
11448 else
|
|
625
|
11449 sprintf(changename[depth], "%.*s-%s: accept %c",
|
|
|
11450 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11451 fword[sp->ts_fidx]);
|
|
|
11452 #endif
|
|
323
|
11453 ++depth;
|
|
625
|
11454 sp = &stack[depth];
|
|
|
11455 ++sp->ts_fidx;
|
|
|
11456 tword[sp->ts_twordlen++] = c;
|
|
|
11457 sp->ts_arridx = idxs[arridx];
|
|
|
11458 #ifdef FEAT_MBYTE
|
|
|
11459 if (newscore == SCORE_SUBST)
|
|
|
11460 sp->ts_isdiff = DIFF_YES;
|
|
330
|
11461 if (has_mbyte)
|
|
358
|
11462 {
|
|
625
|
11463 /* Multi-byte characters are a bit complicated to
|
|
|
11464 * handle: They differ when any of the bytes differ
|
|
|
11465 * and then their length may also differ. */
|
|
|
11466 if (sp->ts_tcharlen == 0)
|
|
|
11467 {
|
|
|
11468 /* First byte. */
|
|
|
11469 sp->ts_tcharidx = 0;
|
|
|
11470 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
11471 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
11472 sp->ts_isdiff = (newscore != 0)
|
|
|
11473 ? DIFF_YES : DIFF_NONE;
|
|
|
11474 }
|
|
|
11475 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
11476 /* When inserting trail bytes don't advance in the
|
|
|
11477 * bad word. */
|
|
|
11478 --sp->ts_fidx;
|
|
|
11479 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
11480 {
|
|
|
11481 /* Last byte of character. */
|
|
|
11482 if (sp->ts_isdiff == DIFF_YES)
|
|
|
11483 {
|
|
|
11484 /* Correct ts_fidx for the byte length of the
|
|
|
11485 * character (we didn't check that before). */
|
|
|
11486 sp->ts_fidx = sp->ts_fcharstart
|
|
|
11487 + MB_BYTE2LEN(
|
|
|
11488 fword[sp->ts_fcharstart]);
|
|
|
11489
|
|
|
11490 /* For changing a composing character adjust
|
|
|
11491 * the score from SCORE_SUBST to
|
|
|
11492 * SCORE_SUBCOMP. */
|
|
|
11493 if (enc_utf8
|
|
|
11494 && utf_iscomposing(
|
|
|
11495 mb_ptr2char(tword
|
|
|
11496 + sp->ts_twordlen
|
|
|
11497 - sp->ts_tcharlen))
|
|
|
11498 && utf_iscomposing(
|
|
|
11499 mb_ptr2char(fword
|
|
|
11500 + sp->ts_fcharstart)))
|
|
|
11501 sp->ts_score -=
|
|
|
11502 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
11503
|
|
|
11504 /* For a similar character adjust score from
|
|
|
11505 * SCORE_SUBST to SCORE_SIMILAR. */
|
|
|
11506 else if (!soundfold
|
|
|
11507 && slang->sl_has_map
|
|
|
11508 && similar_chars(slang,
|
|
|
11509 mb_ptr2char(tword
|
|
|
11510 + sp->ts_twordlen
|
|
|
11511 - sp->ts_tcharlen),
|
|
|
11512 mb_ptr2char(fword
|
|
|
11513 + sp->ts_fcharstart)))
|
|
|
11514 sp->ts_score -=
|
|
|
11515 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
11516 }
|
|
|
11517 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
11518 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
11519 {
|
|
|
11520 p = tword + sp->ts_twordlen - sp->ts_tcharlen;
|
|
|
11521 c = mb_ptr2char(p);
|
|
|
11522 if (enc_utf8 && utf_iscomposing(c))
|
|
|
11523 {
|
|
|
11524 /* Inserting a composing char doesn't
|
|
|
11525 * count that much. */
|
|
|
11526 sp->ts_score -= SCORE_INS - SCORE_INSCOMP;
|
|
|
11527 }
|
|
|
11528 else
|
|
|
11529 {
|
|
|
11530 /* If the previous character was the same,
|
|
|
11531 * thus doubling a character, give a bonus
|
|
|
11532 * to the score. Also for the soundfold
|
|
|
11533 * tree (might seem illogical but does
|
|
|
11534 * give better scores). */
|
|
|
11535 mb_ptr_back(tword, p);
|
|
|
11536 if (c == mb_ptr2char(p))
|
|
|
11537 sp->ts_score -= SCORE_INS
|
|
|
11538 - SCORE_INSDUP;
|
|
|
11539 }
|
|
|
11540 }
|
|
|
11541
|
|
|
11542 /* Starting a new char, reset the length. */
|
|
|
11543 sp->ts_tcharlen = 0;
|
|
|
11544 }
|
|
358
|
11545 }
|
|
330
|
11546 else
|
|
|
11547 #endif
|
|
358
|
11548 {
|
|
625
|
11549 /* If we found a similar char adjust the score.
|
|
|
11550 * We do this after calling go_deeper() because
|
|
|
11551 * it's slow. */
|
|
|
11552 if (newscore != 0
|
|
|
11553 && !soundfold
|
|
|
11554 && slang->sl_has_map
|
|
|
11555 && similar_chars(slang,
|
|
|
11556 c, fword[sp->ts_fidx - 1]))
|
|
|
11557 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
358
|
11558 }
|
|
625
|
11559 }
|
|
|
11560 }
|
|
|
11561 break;
|
|
|
11562
|
|
|
11563 case STATE_DEL:
|
|
|
11564 #ifdef FEAT_MBYTE
|
|
|
11565 /* When past the first byte of a multi-byte char don't try
|
|
|
11566 * delete/insert/swap a character. */
|
|
|
11567 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
11568 {
|
|
|
11569 sp->ts_state = STATE_FINAL;
|
|
|
11570 break;
|
|
|
11571 }
|
|
|
11572 #endif
|
|
|
11573 /*
|
|
|
11574 * Try skipping one character in the bad word (delete it).
|
|
|
11575 */
|
|
|
11576 sp->ts_state = STATE_INS_PREP;
|
|
|
11577 sp->ts_curi = 1;
|
|
|
11578 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*')
|
|
|
11579 /* Deleting a vowel at the start of a word counts less, see
|
|
|
11580 * soundalike_score(). */
|
|
|
11581 newscore = 2 * SCORE_DEL / 3;
|
|
|
11582 else
|
|
|
11583 newscore = SCORE_DEL;
|
|
|
11584 if (fword[sp->ts_fidx] != NUL
|
|
|
11585 && TRY_DEEPER(su, stack, depth, newscore))
|
|
|
11586 {
|
|
|
11587 go_deeper(stack, depth, newscore);
|
|
|
11588 #ifdef DEBUG_TRIEWALK
|
|
|
11589 sprintf(changename[depth], "%.*s-%s: delete %c",
|
|
|
11590 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11591 fword[sp->ts_fidx]);
|
|
|
11592 #endif
|
|
|
11593 ++depth;
|
|
|
11594
|
|
|
11595 /* Remember what character we deleted, so that we can avoid
|
|
|
11596 * inserting it again. */
|
|
|
11597 stack[depth].ts_flags |= TSF_DIDDEL;
|
|
|
11598 stack[depth].ts_delidx = sp->ts_fidx;
|
|
|
11599
|
|
|
11600 /* Advance over the character in fword[]. Give a bonus to the
|
|
|
11601 * score if the same character is following "nn" -> "n". It's
|
|
|
11602 * a bit illogical for soundfold tree but it does give better
|
|
|
11603 * results. */
|
|
|
11604 #ifdef FEAT_MBYTE
|
|
|
11605 if (has_mbyte)
|
|
|
11606 {
|
|
|
11607 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
|
11608 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
|
11609 if (enc_utf8 && utf_iscomposing(c))
|
|
|
11610 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
11611 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
|
11612 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
323
|
11613 }
|
|
|
11614 else
|
|
625
|
11615 #endif
|
|
|
11616 {
|
|
|
11617 ++stack[depth].ts_fidx;
|
|
|
11618 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
11619 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
323
|
11620 }
|
|
|
11621 break;
|
|
625
|
11622 }
|
|
|
11623 /*FALLTHROUGH*/
|
|
|
11624
|
|
|
11625 case STATE_INS_PREP:
|
|
|
11626 if (sp->ts_flags & TSF_DIDDEL)
|
|
|
11627 {
|
|
|
11628 /* If we just deleted a byte then inserting won't make sense,
|
|
|
11629 * a substitute is always cheaper. */
|
|
|
11630 sp->ts_state = STATE_SWAP;
|
|
|
11631 break;
|
|
|
11632 }
|
|
|
11633
|
|
|
11634 /* skip over NUL bytes */
|
|
|
11635 n = sp->ts_arridx;
|
|
|
11636 for (;;)
|
|
|
11637 {
|
|
|
11638 if (sp->ts_curi > byts[n])
|
|
|
11639 {
|
|
|
11640 /* Only NUL bytes at this node, go to next state. */
|
|
|
11641 sp->ts_state = STATE_SWAP;
|
|
|
11642 break;
|
|
|
11643 }
|
|
|
11644 if (byts[n + sp->ts_curi] != NUL)
|
|
|
11645 {
|
|
|
11646 /* Found a byte to insert. */
|
|
|
11647 sp->ts_state = STATE_INS;
|
|
330
|
11648 break;
|
|
|
11649 }
|
|
625
|
11650 ++sp->ts_curi;
|
|
|
11651 }
|
|
|
11652 break;
|
|
|
11653
|
|
|
11654 /*FALLTHROUGH*/
|
|
|
11655
|
|
|
11656 case STATE_INS:
|
|
|
11657 /* Insert one byte. Repeat this for each possible byte at this
|
|
|
11658 * node. */
|
|
|
11659 n = sp->ts_arridx;
|
|
|
11660 if (sp->ts_curi > byts[n])
|
|
|
11661 {
|
|
|
11662 /* Done all bytes at this node, go to next state. */
|
|
|
11663 sp->ts_state = STATE_SWAP;
|
|
|
11664 break;
|
|
|
11665 }
|
|
|
11666
|
|
|
11667 /* Do one more byte at this node, but:
|
|
|
11668 * - Skip NUL bytes.
|
|
|
11669 * - Skip the byte if it's equal to the byte in the word,
|
|
|
11670 * accepting that byte is always better.
|
|
|
11671 */
|
|
|
11672 n += sp->ts_curi++;
|
|
|
11673 c = byts[n];
|
|
|
11674 if (soundfold && sp->ts_twordlen == 0 && c == '*')
|
|
|
11675 /* Inserting a vowel at the start of a word counts less,
|
|
|
11676 * see soundalike_score(). */
|
|
|
11677 newscore = 2 * SCORE_INS / 3;
|
|
|
11678 else
|
|
|
11679 newscore = SCORE_INS;
|
|
|
11680 if (c != fword[sp->ts_fidx]
|
|
|
11681 && TRY_DEEPER(su, stack, depth, newscore))
|
|
|
11682 {
|
|
|
11683 go_deeper(stack, depth, newscore);
|
|
|
11684 #ifdef DEBUG_TRIEWALK
|
|
|
11685 sprintf(changename[depth], "%.*s-%s: insert %c",
|
|
|
11686 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11687 c);
|
|
|
11688 #endif
|
|
|
11689 ++depth;
|
|
|
11690 sp = &stack[depth];
|
|
|
11691 tword[sp->ts_twordlen++] = c;
|
|
|
11692 sp->ts_arridx = idxs[n];
|
|
330
|
11693 #ifdef FEAT_MBYTE
|
|
|
11694 if (has_mbyte)
|
|
|
11695 {
|
|
625
|
11696 fl = MB_BYTE2LEN(c);
|
|
|
11697 if (fl > 1)
|
|
330
|
11698 {
|
|
625
|
11699 /* There are following bytes for the same character.
|
|
|
11700 * We must find all bytes before trying
|
|
|
11701 * delete/insert/swap/etc. */
|
|
|
11702 sp->ts_tcharlen = fl;
|
|
|
11703 sp->ts_tcharidx = 1;
|
|
|
11704 sp->ts_isdiff = DIFF_INSERT;
|
|
330
|
11705 }
|
|
323
|
11706 }
|
|
|
11707 else
|
|
625
|
11708 fl = 1;
|
|
|
11709 if (fl == 1)
|
|
|
11710 #endif
|
|
|
11711 {
|
|
|
11712 /* If the previous character was the same, thus doubling a
|
|
|
11713 * character, give a bonus to the score. Also for
|
|
|
11714 * soundfold words (illogical but does give a better
|
|
|
11715 * score). */
|
|
|
11716 if (sp->ts_twordlen >= 2
|
|
|
11717 && tword[sp->ts_twordlen - 2] == c)
|
|
|
11718 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
11719 }
|
|
|
11720 }
|
|
|
11721 break;
|
|
|
11722
|
|
|
11723 case STATE_SWAP:
|
|
|
11724 /*
|
|
|
11725 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
11726 * We change "fword" here, it's changed back afterwards at
|
|
|
11727 * STATE_UNSWAP.
|
|
|
11728 */
|
|
|
11729 p = fword + sp->ts_fidx;
|
|
|
11730 c = *p;
|
|
|
11731 if (c == NUL)
|
|
|
11732 {
|
|
|
11733 /* End of word, can't swap or replace. */
|
|
|
11734 sp->ts_state = STATE_FINAL;
|
|
|
11735 break;
|
|
|
11736 }
|
|
|
11737
|
|
|
11738 /* Don't swap if the first character is not a word character.
|
|
|
11739 * SWAP3 etc. also don't make sense then. */
|
|
|
11740 if (!soundfold && !spell_iswordp(p, curbuf))
|
|
|
11741 {
|
|
|
11742 sp->ts_state = STATE_REP_INI;
|
|
323
|
11743 break;
|
|
625
|
11744 }
|
|
|
11745
|
|
|
11746 #ifdef FEAT_MBYTE
|
|
|
11747 if (has_mbyte)
|
|
|
11748 {
|
|
|
11749 n = mb_cptr2len(p);
|
|
|
11750 c = mb_ptr2char(p);
|
|
|
11751 if (!soundfold && !spell_iswordp(p + n, curbuf))
|
|
|
11752 c2 = c; /* don't swap non-word char */
|
|
|
11753 else
|
|
|
11754 c2 = mb_ptr2char(p + n);
|
|
|
11755 }
|
|
|
11756 else
|
|
|
11757 #endif
|
|
|
11758 {
|
|
|
11759 if (!soundfold && !spell_iswordp(p + 1, curbuf))
|
|
|
11760 c2 = c; /* don't swap non-word char */
|
|
|
11761 else
|
|
|
11762 c2 = p[1];
|
|
|
11763 }
|
|
|
11764
|
|
|
11765 /* When characters are identical, swap won't do anything.
|
|
|
11766 * Also get here if the second char is not a word character. */
|
|
|
11767 if (c == c2)
|
|
|
11768 {
|
|
|
11769 sp->ts_state = STATE_SWAP3;
|
|
|
11770 break;
|
|
|
11771 }
|
|
|
11772 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP))
|
|
|
11773 {
|
|
|
11774 go_deeper(stack, depth, SCORE_SWAP);
|
|
|
11775 #ifdef DEBUG_TRIEWALK
|
|
|
11776 sprintf(changename[depth], "%.*s-%s: swap %c and %c",
|
|
|
11777 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11778 c, c2);
|
|
|
11779 #endif
|
|
|
11780 sp->ts_state = STATE_UNSWAP;
|
|
|
11781 ++depth;
|
|
330
|
11782 #ifdef FEAT_MBYTE
|
|
|
11783 if (has_mbyte)
|
|
|
11784 {
|
|
625
|
11785 fl = mb_char2len(c2);
|
|
|
11786 mch_memmove(p, p + n, fl);
|
|
|
11787 mb_char2bytes(c, p + fl);
|
|
|
11788 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
330
|
11789 }
|
|
|
11790 else
|
|
|
11791 #endif
|
|
|
11792 {
|
|
625
|
11793 p[0] = c2;
|
|
330
|
11794 p[1] = c;
|
|
625
|
11795 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
11796 }
|
|
|
11797 }
|
|
|
11798 else
|
|
|
11799 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
11800 sp->ts_state = STATE_REP_INI;
|
|
|
11801 break;
|
|
|
11802
|
|
|
11803 case STATE_UNSWAP:
|
|
|
11804 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
11805 p = fword + sp->ts_fidx;
|
|
|
11806 #ifdef FEAT_MBYTE
|
|
|
11807 if (has_mbyte)
|
|
|
11808 {
|
|
|
11809 n = MB_BYTE2LEN(*p);
|
|
|
11810 c = mb_ptr2char(p + n);
|
|
|
11811 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
11812 mb_char2bytes(c, p);
|
|
|
11813 }
|
|
|
11814 else
|
|
|
11815 #endif
|
|
|
11816 {
|
|
|
11817 c = *p;
|
|
|
11818 *p = p[1];
|
|
|
11819 p[1] = c;
|
|
|
11820 }
|
|
|
11821 /*FALLTHROUGH*/
|
|
|
11822
|
|
|
11823 case STATE_SWAP3:
|
|
|
11824 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
|
11825 * "fword" here, it's changed back afterwards at STATE_UNSWAP3. */
|
|
|
11826 p = fword + sp->ts_fidx;
|
|
|
11827 #ifdef FEAT_MBYTE
|
|
|
11828 if (has_mbyte)
|
|
|
11829 {
|
|
|
11830 n = mb_cptr2len(p);
|
|
|
11831 c = mb_ptr2char(p);
|
|
|
11832 fl = mb_cptr2len(p + n);
|
|
|
11833 c2 = mb_ptr2char(p + n);
|
|
|
11834 if (!soundfold && !spell_iswordp(p + n + fl, curbuf))
|
|
|
11835 c3 = c; /* don't swap non-word char */
|
|
|
11836 else
|
|
|
11837 c3 = mb_ptr2char(p + n + fl);
|
|
|
11838 }
|
|
|
11839 else
|
|
|
11840 #endif
|
|
|
11841 {
|
|
|
11842 c = *p;
|
|
|
11843 c2 = p[1];
|
|
|
11844 if (!soundfold && !spell_iswordp(p + 2, curbuf))
|
|
|
11845 c3 = c; /* don't swap non-word char */
|
|
|
11846 else
|
|
|
11847 c3 = p[2];
|
|
|
11848 }
|
|
|
11849
|
|
|
11850 /* When characters are identical: "121" then SWAP3 result is
|
|
|
11851 * identical, ROT3L result is same as SWAP: "211", ROT3L result is
|
|
|
11852 * same as SWAP on next char: "112". Thus skip all swapping.
|
|
|
11853 * Also skip when c3 is NUL.
|
|
|
11854 * Also get here when the third character is not a word character.
|
|
|
11855 * Second character may any char: "a.b" -> "b.a" */
|
|
|
11856 if (c == c3 || c3 == NUL)
|
|
|
11857 {
|
|
|
11858 sp->ts_state = STATE_REP_INI;
|
|
|
11859 break;
|
|
|
11860 }
|
|
|
11861 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
|
|
|
11862 {
|
|
|
11863 go_deeper(stack, depth, SCORE_SWAP3);
|
|
|
11864 #ifdef DEBUG_TRIEWALK
|
|
|
11865 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c",
|
|
|
11866 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11867 c, c3);
|
|
|
11868 #endif
|
|
|
11869 sp->ts_state = STATE_UNSWAP3;
|
|
|
11870 ++depth;
|
|
|
11871 #ifdef FEAT_MBYTE
|
|
|
11872 if (has_mbyte)
|
|
|
11873 {
|
|
|
11874 tl = mb_char2len(c3);
|
|
|
11875 mch_memmove(p, p + n + fl, tl);
|
|
|
11876 mb_char2bytes(c2, p + tl);
|
|
|
11877 mb_char2bytes(c, p + fl + tl);
|
|
|
11878 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
11879 }
|
|
|
11880 else
|
|
|
11881 #endif
|
|
|
11882 {
|
|
|
11883 p[0] = p[2];
|
|
|
11884 p[2] = c;
|
|
|
11885 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
11886 }
|
|
|
11887 }
|
|
|
11888 else
|
|
|
11889 sp->ts_state = STATE_REP_INI;
|
|
|
11890 break;
|
|
|
11891
|
|
|
11892 case STATE_UNSWAP3:
|
|
|
11893 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
11894 p = fword + sp->ts_fidx;
|
|
|
11895 #ifdef FEAT_MBYTE
|
|
|
11896 if (has_mbyte)
|
|
|
11897 {
|
|
|
11898 n = MB_BYTE2LEN(*p);
|
|
|
11899 c2 = mb_ptr2char(p + n);
|
|
|
11900 fl = MB_BYTE2LEN(p[n]);
|
|
|
11901 c = mb_ptr2char(p + n + fl);
|
|
|
11902 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
11903 mch_memmove(p + fl + tl, p, n);
|
|
|
11904 mb_char2bytes(c, p);
|
|
|
11905 mb_char2bytes(c2, p + tl);
|
|
|
11906 p = p + tl;
|
|
|
11907 }
|
|
|
11908 else
|
|
|
11909 #endif
|
|
|
11910 {
|
|
|
11911 c = *p;
|
|
|
11912 *p = p[2];
|
|
|
11913 p[2] = c;
|
|
|
11914 ++p;
|
|
|
11915 }
|
|
|
11916
|
|
|
11917 if (!soundfold && !spell_iswordp(p, curbuf))
|
|
|
11918 {
|
|
|
11919 /* Middle char is not a word char, skip the rotate. First and
|
|
|
11920 * third char were already checked at swap and swap3. */
|
|
|
11921 sp->ts_state = STATE_REP_INI;
|
|
|
11922 break;
|
|
|
11923 }
|
|
|
11924
|
|
|
11925 /* Rotate three characters left: "123" -> "231". We change
|
|
|
11926 * "fword" here, it's changed back afterwards at STATE_UNROT3L. */
|
|
|
11927 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
|
|
|
11928 {
|
|
|
11929 go_deeper(stack, depth, SCORE_SWAP3);
|
|
|
11930 #ifdef DEBUG_TRIEWALK
|
|
|
11931 p = fword + sp->ts_fidx;
|
|
|
11932 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c",
|
|
|
11933 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11934 p[0], p[1], p[2]);
|
|
|
11935 #endif
|
|
|
11936 sp->ts_state = STATE_UNROT3L;
|
|
|
11937 ++depth;
|
|
330
|
11938 p = fword + sp->ts_fidx;
|
|
|
11939 #ifdef FEAT_MBYTE
|
|
|
11940 if (has_mbyte)
|
|
|
11941 {
|
|
474
|
11942 n = mb_cptr2len(p);
|
|
330
|
11943 c = mb_ptr2char(p);
|
|
474
|
11944 fl = mb_cptr2len(p + n);
|
|
625
|
11945 fl += mb_cptr2len(p + n + fl);
|
|
|
11946 mch_memmove(p, p + n, fl);
|
|
|
11947 mb_char2bytes(c, p + fl);
|
|
|
11948 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
330
|
11949 }
|
|
|
11950 else
|
|
|
11951 #endif
|
|
323
|
11952 {
|
|
330
|
11953 c = *p;
|
|
625
|
11954 *p = p[1];
|
|
|
11955 p[1] = p[2];
|
|
|
11956 p[2] = c;
|
|
|
11957 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
11958 }
|
|
|
11959 }
|
|
|
11960 else
|
|
|
11961 sp->ts_state = STATE_REP_INI;
|
|
|
11962 break;
|
|
|
11963
|
|
|
11964 case STATE_UNROT3L:
|
|
|
11965 /* Undo ROT3L: "231" -> "123" */
|
|
|
11966 p = fword + sp->ts_fidx;
|
|
|
11967 #ifdef FEAT_MBYTE
|
|
|
11968 if (has_mbyte)
|
|
|
11969 {
|
|
|
11970 n = MB_BYTE2LEN(*p);
|
|
|
11971 n += MB_BYTE2LEN(p[n]);
|
|
|
11972 c = mb_ptr2char(p + n);
|
|
|
11973 tl = MB_BYTE2LEN(p[n]);
|
|
|
11974 mch_memmove(p + tl, p, n);
|
|
|
11975 mb_char2bytes(c, p);
|
|
|
11976 }
|
|
|
11977 else
|
|
|
11978 #endif
|
|
|
11979 {
|
|
|
11980 c = p[2];
|
|
|
11981 p[2] = p[1];
|
|
|
11982 p[1] = *p;
|
|
|
11983 *p = c;
|
|
|
11984 }
|
|
|
11985
|
|
|
11986 /* Rotate three bytes right: "123" -> "312". We change "fword"
|
|
|
11987 * here, it's changed back afterwards at STATE_UNROT3R. */
|
|
|
11988 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
|
|
|
11989 {
|
|
|
11990 go_deeper(stack, depth, SCORE_SWAP3);
|
|
|
11991 #ifdef DEBUG_TRIEWALK
|
|
|
11992 p = fword + sp->ts_fidx;
|
|
|
11993 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c",
|
|
|
11994 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
11995 p[0], p[1], p[2]);
|
|
|
11996 #endif
|
|
|
11997 sp->ts_state = STATE_UNROT3R;
|
|
|
11998 ++depth;
|
|
330
|
11999 p = fword + sp->ts_fidx;
|
|
|
12000 #ifdef FEAT_MBYTE
|
|
|
12001 if (has_mbyte)
|
|
|
12002 {
|
|
625
|
12003 n = mb_cptr2len(p);
|
|
|
12004 n += mb_cptr2len(p + n);
|
|
330
|
12005 c = mb_ptr2char(p + n);
|
|
625
|
12006 tl = mb_cptr2len(p + n);
|
|
330
|
12007 mch_memmove(p + tl, p, n);
|
|
|
12008 mb_char2bytes(c, p);
|
|
625
|
12009 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
330
|
12010 }
|
|
|
12011 else
|
|
|
12012 #endif
|
|
|
12013 {
|
|
|
12014 c = p[2];
|
|
|
12015 p[2] = p[1];
|
|
|
12016 p[1] = *p;
|
|
|
12017 *p = c;
|
|
625
|
12018 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
12019 }
|
|
|
12020 }
|
|
|
12021 else
|
|
|
12022 sp->ts_state = STATE_REP_INI;
|
|
|
12023 break;
|
|
|
12024
|
|
|
12025 case STATE_UNROT3R:
|
|
|
12026 /* Undo ROT3R: "312" -> "123" */
|
|
|
12027 p = fword + sp->ts_fidx;
|
|
|
12028 #ifdef FEAT_MBYTE
|
|
|
12029 if (has_mbyte)
|
|
|
12030 {
|
|
|
12031 c = mb_ptr2char(p);
|
|
|
12032 tl = MB_BYTE2LEN(*p);
|
|
|
12033 n = MB_BYTE2LEN(p[tl]);
|
|
|
12034 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
12035 mch_memmove(p, p + tl, n);
|
|
|
12036 mb_char2bytes(c, p + n);
|
|
|
12037 }
|
|
|
12038 else
|
|
|
12039 #endif
|
|
|
12040 {
|
|
|
12041 c = *p;
|
|
|
12042 *p = p[1];
|
|
|
12043 p[1] = p[2];
|
|
|
12044 p[2] = c;
|
|
|
12045 }
|
|
|
12046 /*FALLTHROUGH*/
|
|
|
12047
|
|
|
12048 case STATE_REP_INI:
|
|
|
12049 /* Check if matching with REP items from the .aff file would work.
|
|
|
12050 * Quickly skip if:
|
|
|
12051 * - there are no REP items and we are not in the soundfold trie
|
|
|
12052 * - the score is going to be too high anyway
|
|
|
12053 * - already applied a REP item or swapped here */
|
|
|
12054 if ((lp->lp_replang == NULL && !soundfold)
|
|
|
12055 || sp->ts_score + SCORE_REP >= su->su_maxscore
|
|
|
12056 || sp->ts_fidx < sp->ts_fidxtry)
|
|
|
12057 {
|
|
|
12058 sp->ts_state = STATE_FINAL;
|
|
323
|
12059 break;
|
|
625
|
12060 }
|
|
|
12061
|
|
|
12062 /* Use the first byte to quickly find the first entry that may
|
|
|
12063 * match. If the index is -1 there is none. */
|
|
|
12064 if (soundfold)
|
|
|
12065 sp->ts_curi = slang->sl_repsal_first[fword[sp->ts_fidx]];
|
|
|
12066 else
|
|
|
12067 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]];
|
|
|
12068
|
|
|
12069 if (sp->ts_curi < 0)
|
|
|
12070 {
|
|
|
12071 sp->ts_state = STATE_FINAL;
|
|
|
12072 break;
|
|
|
12073 }
|
|
|
12074
|
|
|
12075 sp->ts_state = STATE_REP;
|
|
|
12076 /*FALLTHROUGH*/
|
|
|
12077
|
|
|
12078 case STATE_REP:
|
|
|
12079 /* Try matching with REP items from the .aff file. For each match
|
|
|
12080 * replace the characters and check if the resulting word is
|
|
|
12081 * valid. */
|
|
|
12082 p = fword + sp->ts_fidx;
|
|
|
12083
|
|
|
12084 if (soundfold)
|
|
|
12085 gap = &slang->sl_repsal;
|
|
|
12086 else
|
|
503
|
12087 gap = &lp->lp_replang->sl_rep;
|
|
625
|
12088 while (sp->ts_curi < gap->ga_len)
|
|
|
12089 {
|
|
|
12090 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
12091 if (*ftp->ft_from != *p)
|
|
|
12092 {
|
|
|
12093 /* past possible matching entries */
|
|
|
12094 sp->ts_curi = gap->ga_len;
|
|
323
|
12095 break;
|
|
|
12096 }
|
|
625
|
12097 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
12098 && TRY_DEEPER(su, stack, depth, SCORE_REP))
|
|
|
12099 {
|
|
|
12100 go_deeper(stack, depth, SCORE_REP);
|
|
|
12101 #ifdef DEBUG_TRIEWALK
|
|
|
12102 sprintf(changename[depth], "%.*s-%s: replace %s with %s",
|
|
|
12103 sp->ts_twordlen, tword, fword + sp->ts_fidx,
|
|
|
12104 ftp->ft_from, ftp->ft_to);
|
|
|
12105 #endif
|
|
|
12106 /* Need to undo this afterwards. */
|
|
|
12107 sp->ts_state = STATE_REP_UNDO;
|
|
|
12108
|
|
|
12109 /* Change the "from" to the "to" string. */
|
|
|
12110 ++depth;
|
|
|
12111 fl = STRLEN(ftp->ft_from);
|
|
|
12112 tl = STRLEN(ftp->ft_to);
|
|
|
12113 if (fl != tl)
|
|
323
|
12114 {
|
|
625
|
12115 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
|
12116 repextra += tl - fl;
|
|
323
|
12117 }
|
|
625
|
12118 mch_memmove(p, ftp->ft_to, tl);
|
|
|
12119 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
|
12120 #ifdef FEAT_MBYTE
|
|
|
12121 stack[depth].ts_tcharlen = 0;
|
|
|
12122 #endif
|
|
|
12123 break;
|
|
|
12124 }
|
|
|
12125 }
|
|
|
12126
|
|
|
12127 if (sp->ts_curi >= gap->ga_len && sp->ts_state == STATE_REP)
|
|
|
12128 /* No (more) matches. */
|
|
|
12129 sp->ts_state = STATE_FINAL;
|
|
|
12130
|
|
|
12131 break;
|
|
|
12132
|
|
|
12133 case STATE_REP_UNDO:
|
|
|
12134 /* Undo a REP replacement and continue with the next one. */
|
|
|
12135 if (soundfold)
|
|
|
12136 gap = &slang->sl_repsal;
|
|
|
12137 else
|
|
|
12138 gap = &lp->lp_replang->sl_rep;
|
|
|
12139 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1;
|
|
|
12140 fl = STRLEN(ftp->ft_from);
|
|
|
12141 tl = STRLEN(ftp->ft_to);
|
|
|
12142 p = fword + sp->ts_fidx;
|
|
|
12143 if (fl != tl)
|
|
|
12144 {
|
|
|
12145 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
|
12146 repextra -= tl - fl;
|
|
|
12147 }
|
|
|
12148 mch_memmove(p, ftp->ft_from, fl);
|
|
|
12149 sp->ts_state = STATE_REP;
|
|
|
12150 break;
|
|
|
12151
|
|
|
12152 default:
|
|
|
12153 /* Did all possible states at this level, go up one level. */
|
|
|
12154 --depth;
|
|
|
12155
|
|
|
12156 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
|
|
|
12157 {
|
|
|
12158 /* Continue in or go back to the prefix tree. */
|
|
|
12159 byts = pbyts;
|
|
|
12160 idxs = pidxs;
|
|
|
12161 }
|
|
|
12162
|
|
|
12163 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
12164 if (--breakcheckcount == 0)
|
|
|
12165 {
|
|
|
12166 ui_breakcheck();
|
|
|
12167 breakcheckcount = 1000;
|
|
|
12168 }
|
|
|
12169 }
|
|
|
12170 }
|
|
|
12171 }
|
|
|
12172
|
|
|
12173
|
|
|
12174 /*
|
|
|
12175 * Go one level deeper in the tree.
|
|
|
12176 */
|
|
|
12177 static void
|
|
|
12178 go_deeper(stack, depth, score_add)
|
|
323
|
12179 trystate_T *stack;
|
|
|
12180 int depth;
|
|
|
12181 int score_add;
|
|
|
12182 {
|
|
330
|
12183 stack[depth + 1] = stack[depth];
|
|
323
|
12184 stack[depth + 1].ts_state = STATE_START;
|
|
625
|
12185 stack[depth + 1].ts_score = stack[depth].ts_score + score_add;
|
|
323
|
12186 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
485
|
12187 stack[depth + 1].ts_flags = 0;
|
|
323
|
12188 }
|
|
|
12189
|
|
455
|
12190 #ifdef FEAT_MBYTE
|
|
|
12191 /*
|
|
|
12192 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
12193 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
12194 */
|
|
|
12195 static int
|
|
|
12196 nofold_len(fword, flen, word)
|
|
|
12197 char_u *fword;
|
|
|
12198 int flen;
|
|
|
12199 char_u *word;
|
|
|
12200 {
|
|
|
12201 char_u *p;
|
|
|
12202 int i = 0;
|
|
|
12203
|
|
|
12204 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
12205 ++i;
|
|
|
12206 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
12207 --i;
|
|
|
12208 return (int)(p - word);
|
|
|
12209 }
|
|
|
12210 #endif
|
|
|
12211
|
|
323
|
12212 /*
|
|
|
12213 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
12214 * words and put it in "kword".
|
|
|
12215 * Theoretically there could be several keep-case words that result in the
|
|
|
12216 * same case-folded word, but we only find one...
|
|
|
12217 */
|
|
|
12218 static void
|
|
|
12219 find_keepcap_word(slang, fword, kword)
|
|
|
12220 slang_T *slang;
|
|
|
12221 char_u *fword;
|
|
|
12222 char_u *kword;
|
|
|
12223 {
|
|
|
12224 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
12225 int depth;
|
|
324
|
12226 idx_T tryidx;
|
|
323
|
12227
|
|
|
12228 /* The following arrays are used at each depth in the tree. */
|
|
324
|
12229 idx_T arridx[MAXWLEN];
|
|
323
|
12230 int round[MAXWLEN];
|
|
|
12231 int fwordidx[MAXWLEN];
|
|
|
12232 int uwordidx[MAXWLEN];
|
|
|
12233 int kwordlen[MAXWLEN];
|
|
|
12234
|
|
|
12235 int flen, ulen;
|
|
|
12236 int l;
|
|
|
12237 int len;
|
|
|
12238 int c;
|
|
324
|
12239 idx_T lo, hi, m;
|
|
323
|
12240 char_u *p;
|
|
|
12241 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
12242 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
12243
|
|
|
12244 if (byts == NULL)
|
|
|
12245 {
|
|
|
12246 /* array is empty: "cannot happen" */
|
|
|
12247 *kword = NUL;
|
|
|
12248 return;
|
|
|
12249 }
|
|
|
12250
|
|
|
12251 /* Make an all-cap version of "fword". */
|
|
|
12252 allcap_copy(fword, uword);
|
|
|
12253
|
|
|
12254 /*
|
|
|
12255 * Each character needs to be tried both case-folded and upper-case.
|
|
|
12256 * All this gets very complicated if we keep in mind that changing case
|
|
|
12257 * may change the byte length of a multi-byte character...
|
|
|
12258 */
|
|
|
12259 depth = 0;
|
|
|
12260 arridx[0] = 0;
|
|
|
12261 round[0] = 0;
|
|
|
12262 fwordidx[0] = 0;
|
|
|
12263 uwordidx[0] = 0;
|
|
|
12264 kwordlen[0] = 0;
|
|
|
12265 while (depth >= 0)
|
|
|
12266 {
|
|
|
12267 if (fword[fwordidx[depth]] == NUL)
|
|
|
12268 {
|
|
|
12269 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
12270 * here we have found a match. */
|
|
|
12271 if (byts[arridx[depth] + 1] == 0)
|
|
|
12272 {
|
|
|
12273 kword[kwordlen[depth]] = NUL;
|
|
|
12274 return;
|
|
|
12275 }
|
|
|
12276
|
|
|
12277 /* kword is getting too long, continue one level up */
|
|
|
12278 --depth;
|
|
|
12279 }
|
|
|
12280 else if (++round[depth] > 2)
|
|
|
12281 {
|
|
|
12282 /* tried both fold-case and upper-case character, continue one
|
|
|
12283 * level up */
|
|
|
12284 --depth;
|
|
|
12285 }
|
|
|
12286 else
|
|
|
12287 {
|
|
|
12288 /*
|
|
|
12289 * round[depth] == 1: Try using the folded-case character.
|
|
|
12290 * round[depth] == 2: Try using the upper-case character.
|
|
|
12291 */
|
|
|
12292 #ifdef FEAT_MBYTE
|
|
|
12293 if (has_mbyte)
|
|
|
12294 {
|
|
474
|
12295 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
12296 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
12297 }
|
|
|
12298 else
|
|
|
12299 #endif
|
|
|
12300 ulen = flen = 1;
|
|
|
12301 if (round[depth] == 1)
|
|
|
12302 {
|
|
|
12303 p = fword + fwordidx[depth];
|
|
|
12304 l = flen;
|
|
|
12305 }
|
|
|
12306 else
|
|
|
12307 {
|
|
|
12308 p = uword + uwordidx[depth];
|
|
|
12309 l = ulen;
|
|
|
12310 }
|
|
|
12311
|
|
|
12312 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
12313 {
|
|
|
12314 /* Perform a binary search in the list of accepted bytes. */
|
|
|
12315 len = byts[tryidx++];
|
|
|
12316 c = *p++;
|
|
|
12317 lo = tryidx;
|
|
|
12318 hi = tryidx + len - 1;
|
|
|
12319 while (lo < hi)
|
|
|
12320 {
|
|
|
12321 m = (lo + hi) / 2;
|
|
|
12322 if (byts[m] > c)
|
|
|
12323 hi = m - 1;
|
|
|
12324 else if (byts[m] < c)
|
|
|
12325 lo = m + 1;
|
|
|
12326 else
|
|
|
12327 {
|
|
|
12328 lo = hi = m;
|
|
|
12329 break;
|
|
|
12330 }
|
|
|
12331 }
|
|
|
12332
|
|
|
12333 /* Stop if there is no matching byte. */
|
|
|
12334 if (hi < lo || byts[lo] != c)
|
|
|
12335 break;
|
|
|
12336
|
|
|
12337 /* Continue at the child (if there is one). */
|
|
|
12338 tryidx = idxs[lo];
|
|
|
12339 }
|
|
|
12340
|
|
|
12341 if (l == 0)
|
|
|
12342 {
|
|
|
12343 /*
|
|
|
12344 * Found the matching char. Copy it to "kword" and go a
|
|
|
12345 * level deeper.
|
|
|
12346 */
|
|
|
12347 if (round[depth] == 1)
|
|
|
12348 {
|
|
|
12349 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
12350 flen);
|
|
|
12351 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
12352 }
|
|
|
12353 else
|
|
|
12354 {
|
|
|
12355 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
12356 ulen);
|
|
|
12357 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
12358 }
|
|
|
12359 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
12360 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
12361
|
|
|
12362 ++depth;
|
|
|
12363 arridx[depth] = tryidx;
|
|
|
12364 round[depth] = 0;
|
|
|
12365 }
|
|
|
12366 }
|
|
|
12367 }
|
|
|
12368
|
|
|
12369 /* Didn't find it: "cannot happen". */
|
|
|
12370 *kword = NUL;
|
|
|
12371 }
|
|
|
12372
|
|
|
12373 /*
|
|
344
|
12374 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
12375 * su->su_sga.
|
|
|
12376 */
|
|
|
12377 static void
|
|
|
12378 score_comp_sal(su)
|
|
|
12379 suginfo_T *su;
|
|
|
12380 {
|
|
|
12381 langp_T *lp;
|
|
|
12382 char_u badsound[MAXWLEN];
|
|
|
12383 int i;
|
|
|
12384 suggest_T *stp;
|
|
|
12385 suggest_T *sstp;
|
|
|
12386 int score;
|
|
500
|
12387 int lpi;
|
|
344
|
12388
|
|
|
12389 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
12390 return;
|
|
|
12391
|
|
|
12392 /* Use the sound-folding of the first language that supports it. */
|
|
503
|
12393 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
12394 {
|
|
|
12395 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
344
|
12396 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
12397 {
|
|
|
12398 /* soundfold the bad word */
|
|
375
|
12399 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
12400
|
|
|
12401 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
12402 {
|
|
|
12403 stp = &SUG(su->su_ga, i);
|
|
|
12404
|
|
351
|
12405 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
12406 * sound-a-like score. */
|
|
|
12407 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
12408 if (score < SCORE_MAXMAX)
|
|
|
12409 {
|
|
|
12410 /* Add the suggestion. */
|
|
|
12411 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
12412 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
12413 if (sstp->st_word != NULL)
|
|
|
12414 {
|
|
625
|
12415 sstp->st_wordlen = stp->st_wordlen;
|
|
344
|
12416 sstp->st_score = score;
|
|
|
12417 sstp->st_altscore = 0;
|
|
|
12418 sstp->st_orglen = stp->st_orglen;
|
|
|
12419 ++su->su_sga.ga_len;
|
|
|
12420 }
|
|
|
12421 }
|
|
|
12422 }
|
|
|
12423 break;
|
|
|
12424 }
|
|
500
|
12425 }
|
|
344
|
12426 }
|
|
|
12427
|
|
|
12428 /*
|
|
|
12429 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
12430 * They are intwined.
|
|
|
12431 */
|
|
|
12432 static void
|
|
|
12433 score_combine(su)
|
|
|
12434 suginfo_T *su;
|
|
|
12435 {
|
|
|
12436 int i;
|
|
|
12437 int j;
|
|
|
12438 garray_T ga;
|
|
|
12439 garray_T *gap;
|
|
|
12440 langp_T *lp;
|
|
|
12441 suggest_T *stp;
|
|
|
12442 char_u *p;
|
|
|
12443 char_u badsound[MAXWLEN];
|
|
|
12444 int round;
|
|
500
|
12445 int lpi;
|
|
625
|
12446 slang_T *slang = NULL;
|
|
344
|
12447
|
|
|
12448 /* Add the alternate score to su_ga. */
|
|
503
|
12449 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
12450 {
|
|
|
12451 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
344
|
12452 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
12453 {
|
|
|
12454 /* soundfold the bad word */
|
|
625
|
12455 slang = lp->lp_slang;
|
|
|
12456 spell_soundfold(slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
12457
|
|
|
12458 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
12459 {
|
|
|
12460 stp = &SUG(su->su_ga, i);
|
|
625
|
12461 stp->st_altscore = stp_sal_score(stp, su, slang, badsound);
|
|
344
|
12462 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
12463 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
12464 else
|
|
|
12465 stp->st_score = (stp->st_score * 3
|
|
|
12466 + stp->st_altscore) / 4;
|
|
|
12467 stp->st_salscore = FALSE;
|
|
|
12468 }
|
|
|
12469 break;
|
|
|
12470 }
|
|
|
12471 }
|
|
|
12472
|
|
625
|
12473 if (slang == NULL) /* just in case */
|
|
|
12474 return;
|
|
|
12475
|
|
344
|
12476 /* Add the alternate score to su_sga. */
|
|
|
12477 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
12478 {
|
|
|
12479 stp = &SUG(su->su_sga, i);
|
|
625
|
12480 stp->st_altscore = spell_edit_score(slang,
|
|
|
12481 su->su_badword, stp->st_word);
|
|
344
|
12482 if (stp->st_score == SCORE_MAXMAX)
|
|
|
12483 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
12484 else
|
|
|
12485 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
12486 stp->st_salscore = TRUE;
|
|
|
12487 }
|
|
|
12488
|
|
625
|
12489 /* Remove bad suggestions, sort the suggestions and truncate at "maxcount"
|
|
|
12490 * for both lists. */
|
|
|
12491 check_suggestions(su, &su->su_ga);
|
|
344
|
12492 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
625
|
12493 check_suggestions(su, &su->su_sga);
|
|
344
|
12494 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
12495
|
|
|
12496 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
12497 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
12498 return;
|
|
|
12499
|
|
|
12500 stp = &SUG(ga, 0);
|
|
|
12501 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
12502 {
|
|
|
12503 /* round 1: get a suggestion from su_ga
|
|
|
12504 * round 2: get a suggestion from su_sga */
|
|
|
12505 for (round = 1; round <= 2; ++round)
|
|
|
12506 {
|
|
|
12507 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
12508 if (i < gap->ga_len)
|
|
|
12509 {
|
|
|
12510 /* Don't add a word if it's already there. */
|
|
|
12511 p = SUG(*gap, i).st_word;
|
|
|
12512 for (j = 0; j < ga.ga_len; ++j)
|
|
|
12513 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
12514 break;
|
|
|
12515 if (j == ga.ga_len)
|
|
|
12516 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
12517 else
|
|
|
12518 vim_free(p);
|
|
|
12519 }
|
|
|
12520 }
|
|
|
12521 }
|
|
|
12522
|
|
|
12523 ga_clear(&su->su_ga);
|
|
|
12524 ga_clear(&su->su_sga);
|
|
|
12525
|
|
|
12526 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
12527 if (ga.ga_len > su->su_maxcount)
|
|
|
12528 {
|
|
|
12529 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
12530 vim_free(stp[i].st_word);
|
|
|
12531 ga.ga_len = su->su_maxcount;
|
|
|
12532 }
|
|
|
12533
|
|
|
12534 su->su_ga = ga;
|
|
|
12535 }
|
|
|
12536
|
|
|
12537 /*
|
|
351
|
12538 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
12539 * badword.
|
|
|
12540 */
|
|
|
12541 static int
|
|
|
12542 stp_sal_score(stp, su, slang, badsound)
|
|
|
12543 suggest_T *stp;
|
|
|
12544 suginfo_T *su;
|
|
|
12545 slang_T *slang;
|
|
|
12546 char_u *badsound; /* sound-folded badword */
|
|
|
12547 {
|
|
|
12548 char_u *p;
|
|
534
|
12549 char_u *pbad;
|
|
|
12550 char_u *pgood;
|
|
351
|
12551 char_u badsound2[MAXWLEN];
|
|
|
12552 char_u fword[MAXWLEN];
|
|
|
12553 char_u goodsound[MAXWLEN];
|
|
534
|
12554 char_u goodword[MAXWLEN];
|
|
|
12555 int lendiff;
|
|
|
12556
|
|
|
12557 lendiff = (int)(su->su_badlen - stp->st_orglen);
|
|
|
12558 if (lendiff >= 0)
|
|
|
12559 pbad = badsound;
|
|
351
|
12560 else
|
|
|
12561 {
|
|
|
12562 /* soundfold the bad word with more characters following */
|
|
|
12563 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
12564
|
|
|
12565 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
12566 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
12567 * removing the space. Don't do it when the good word also contains a
|
|
|
12568 * space. */
|
|
|
12569 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
12570 && *skiptowhite(stp->st_word) == NUL)
|
|
|
12571 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
12572 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
12573
|
|
375
|
12574 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
534
|
12575 pbad = badsound2;
|
|
|
12576 }
|
|
|
12577
|
|
|
12578 if (lendiff > 0)
|
|
|
12579 {
|
|
|
12580 /* Add part of the bad word to the good word, so that we soundfold
|
|
|
12581 * what replaces the bad word. */
|
|
|
12582 STRCPY(goodword, stp->st_word);
|
|
625
|
12583 vim_strncpy(goodword + stp->st_wordlen,
|
|
|
12584 su->su_badptr + su->su_badlen - lendiff, lendiff);
|
|
534
|
12585 pgood = goodword;
|
|
|
12586 }
|
|
|
12587 else
|
|
|
12588 pgood = stp->st_word;
|
|
351
|
12589
|
|
375
|
12590 /* Sound-fold the word and compute the score for the difference. */
|
|
534
|
12591 spell_soundfold(slang, pgood, FALSE, goodsound);
|
|
|
12592
|
|
|
12593 return soundalike_score(goodsound, pbad);
|
|
351
|
12594 }
|
|
|
12595
|
|
625
|
12596 /* structure used to store soundfolded words that add_sound_suggest() has
|
|
|
12597 * handled already. */
|
|
|
12598 typedef struct
|
|
|
12599 {
|
|
|
12600 short sft_score; /* lowest score used */
|
|
|
12601 char_u sft_word[1]; /* soundfolded word, actually longer */
|
|
|
12602 } sftword_T;
|
|
|
12603
|
|
|
12604 static sftword_T dumsft;
|
|
|
12605 #define HIKEY2SFT(p) ((sftword_T *)(p - (dumsft.sft_word - (char_u *)&dumsft)))
|
|
|
12606 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key)
|
|
|
12607
|
|
|
12608 /*
|
|
|
12609 * Prepare for calling suggest_try_soundalike().
|
|
|
12610 */
|
|
|
12611 static void
|
|
|
12612 suggest_try_soundalike_prep()
|
|
|
12613 {
|
|
|
12614 langp_T *lp;
|
|
|
12615 int lpi;
|
|
|
12616 slang_T *slang;
|
|
|
12617
|
|
|
12618 /* Do this for all languages that support sound folding and for which a
|
|
|
12619 * .sug file has been loaded. */
|
|
|
12620 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
12621 {
|
|
|
12622 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
12623 slang = lp->lp_slang;
|
|
|
12624 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL)
|
|
|
12625 /* prepare the hashtable used by add_sound_suggest() */
|
|
|
12626 hash_init(&slang->sl_sounddone);
|
|
|
12627 }
|
|
|
12628 }
|
|
|
12629
|
|
351
|
12630 /*
|
|
323
|
12631 * Find suggestions by comparing the word in a sound-a-like form.
|
|
503
|
12632 * Note: This doesn't support postponed prefixes.
|
|
323
|
12633 */
|
|
|
12634 static void
|
|
346
|
12635 suggest_try_soundalike(su)
|
|
323
|
12636 suginfo_T *su;
|
|
|
12637 {
|
|
|
12638 char_u salword[MAXWLEN];
|
|
|
12639 langp_T *lp;
|
|
500
|
12640 int lpi;
|
|
503
|
12641 slang_T *slang;
|
|
344
|
12642
|
|
625
|
12643 /* Do this for all languages that support sound folding and for which a
|
|
|
12644 * .sug file has been loaded. */
|
|
|
12645 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
12646 {
|
|
|
12647 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
12648 slang = lp->lp_slang;
|
|
|
12649 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL)
|
|
|
12650 {
|
|
|
12651 /* soundfold the bad word */
|
|
|
12652 spell_soundfold(slang, su->su_fbadword, TRUE, salword);
|
|
|
12653
|
|
|
12654 /* try all kinds of inserts/deletes/swaps/etc. */
|
|
|
12655 /* TODO: also soundfold the next words, so that we can try joining
|
|
|
12656 * and splitting */
|
|
|
12657 suggest_trie_walk(su, lp, salword, TRUE);
|
|
|
12658 }
|
|
|
12659 }
|
|
|
12660 }
|
|
|
12661
|
|
|
12662 /*
|
|
|
12663 * Finish up after calling suggest_try_soundalike().
|
|
|
12664 */
|
|
|
12665 static void
|
|
|
12666 suggest_try_soundalike_finish()
|
|
|
12667 {
|
|
|
12668 langp_T *lp;
|
|
|
12669 int lpi;
|
|
|
12670 slang_T *slang;
|
|
|
12671 int todo;
|
|
|
12672 hashitem_T *hi;
|
|
|
12673
|
|
|
12674 /* Do this for all languages that support sound folding and for which a
|
|
|
12675 * .sug file has been loaded. */
|
|
503
|
12676 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
12677 {
|
|
|
12678 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
12679 slang = lp->lp_slang;
|
|
625
|
12680 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL)
|
|
|
12681 {
|
|
|
12682 /* Free the info about handled words. */
|
|
|
12683 todo = slang->sl_sounddone.ht_used;
|
|
|
12684 for (hi = slang->sl_sounddone.ht_array; todo > 0; ++hi)
|
|
|
12685 if (!HASHITEM_EMPTY(hi))
|
|
|
12686 {
|
|
|
12687 vim_free(HI2SFT(hi));
|
|
|
12688 --todo;
|
|
|
12689 }
|
|
|
12690 hash_clear(&slang->sl_sounddone);
|
|
|
12691 }
|
|
|
12692 }
|
|
|
12693 }
|
|
|
12694
|
|
|
12695 /*
|
|
|
12696 * A match with a soundfolded word is found. Add the good word(s) that
|
|
|
12697 * produce this soundfolded word.
|
|
|
12698 */
|
|
|
12699 static void
|
|
|
12700 add_sound_suggest(su, goodword, score, lp)
|
|
|
12701 suginfo_T *su;
|
|
|
12702 char_u *goodword;
|
|
|
12703 int score; /* soundfold score */
|
|
|
12704 langp_T *lp;
|
|
|
12705 {
|
|
|
12706 slang_T *slang = lp->lp_slang; /* language for sound folding */
|
|
|
12707 int sfwordnr;
|
|
|
12708 char_u *nrline;
|
|
|
12709 int orgnr;
|
|
|
12710 char_u theword[MAXWLEN];
|
|
|
12711 int i;
|
|
|
12712 int wlen;
|
|
|
12713 char_u *byts;
|
|
|
12714 idx_T *idxs;
|
|
|
12715 int n;
|
|
|
12716 int wordcount;
|
|
|
12717 int wc;
|
|
|
12718 int goodscore;
|
|
|
12719 hash_T hash;
|
|
|
12720 hashitem_T *hi;
|
|
|
12721 sftword_T *sft;
|
|
|
12722 int bc, gc;
|
|
|
12723 int limit;
|
|
|
12724
|
|
|
12725 /*
|
|
|
12726 * It's very well possible that the same soundfold word is found several
|
|
|
12727 * times with different scores. Since the following is quite slow only do
|
|
|
12728 * the words that have a better score than before. Use a hashtable to
|
|
|
12729 * remember the words that have been done.
|
|
|
12730 */
|
|
|
12731 hash = hash_hash(goodword);
|
|
|
12732 hi = hash_lookup(&slang->sl_sounddone, goodword, hash);
|
|
|
12733 if (HASHITEM_EMPTY(hi))
|
|
|
12734 {
|
|
|
12735 sft = (sftword_T *)alloc(sizeof(sftword_T) + STRLEN(goodword));
|
|
|
12736 if (sft != NULL)
|
|
|
12737 {
|
|
|
12738 sft->sft_score = score;
|
|
|
12739 STRCPY(sft->sft_word, goodword);
|
|
|
12740 hash_add_item(&slang->sl_sounddone, hi, sft->sft_word, hash);
|
|
|
12741 }
|
|
|
12742 }
|
|
|
12743 else
|
|
|
12744 {
|
|
|
12745 sft = HI2SFT(hi);
|
|
|
12746 if (score >= sft->sft_score)
|
|
|
12747 return;
|
|
|
12748 sft->sft_score = score;
|
|
|
12749 }
|
|
|
12750
|
|
|
12751 /*
|
|
|
12752 * Find the word nr in the soundfold tree.
|
|
|
12753 */
|
|
|
12754 sfwordnr = soundfold_find(slang, goodword);
|
|
|
12755 if (sfwordnr < 0)
|
|
|
12756 {
|
|
|
12757 EMSG2(_(e_intern2), "add_sound_suggest()");
|
|
|
12758 return;
|
|
|
12759 }
|
|
|
12760
|
|
|
12761 /*
|
|
|
12762 * go over the list of good words that produce this soundfold word
|
|
|
12763 */
|
|
|
12764 nrline = ml_get_buf(slang->sl_sugbuf, (linenr_T)(sfwordnr + 1), FALSE);
|
|
|
12765 orgnr = 0;
|
|
|
12766 while (*nrline != NUL)
|
|
|
12767 {
|
|
|
12768 /* The wordnr was stored in a minimal nr of bytes as an offset to the
|
|
|
12769 * previous wordnr. */
|
|
|
12770 orgnr += bytes2offset(&nrline);
|
|
|
12771
|
|
|
12772 byts = slang->sl_fbyts;
|
|
|
12773 idxs = slang->sl_fidxs;
|
|
|
12774
|
|
|
12775 /* Lookup the word "orgnr" one of the two tries. */
|
|
|
12776 n = 0;
|
|
|
12777 wlen = 0;
|
|
|
12778 wordcount = 0;
|
|
|
12779 for (;;)
|
|
|
12780 {
|
|
|
12781 i = 1;
|
|
|
12782 if (wordcount == orgnr && byts[n + 1] == NUL)
|
|
|
12783 break; /* found end of word */
|
|
|
12784
|
|
|
12785 if (byts[n + 1] == NUL)
|
|
|
12786 ++wordcount;
|
|
|
12787
|
|
|
12788 /* skip over the NUL bytes */
|
|
|
12789 for ( ; byts[n + i] == NUL; ++i)
|
|
|
12790 if (i > byts[n]) /* safety check */
|
|
|
12791 {
|
|
|
12792 STRCPY(theword + wlen, "BAD");
|
|
|
12793 goto badword;
|
|
|
12794 }
|
|
|
12795
|
|
|
12796 /* One of the siblings must have the word. */
|
|
|
12797 for ( ; i < byts[n]; ++i)
|
|
|
12798 {
|
|
|
12799 wc = idxs[idxs[n + i]]; /* nr of words under this byte */
|
|
|
12800 if (wordcount + wc > orgnr)
|
|
|
12801 break;
|
|
|
12802 wordcount += wc;
|
|
|
12803 }
|
|
|
12804
|
|
|
12805 theword[wlen++] = byts[n + i];
|
|
|
12806 n = idxs[n + i];
|
|
|
12807 }
|
|
|
12808 badword:
|
|
|
12809 theword[wlen] = NUL;
|
|
|
12810
|
|
|
12811 /* Go over the possible flags and regions. */
|
|
|
12812 for (; i <= byts[n] && byts[n + i] == NUL; ++i)
|
|
|
12813 {
|
|
|
12814 char_u cword[MAXWLEN];
|
|
|
12815 char_u *p;
|
|
|
12816 int flags = (int)idxs[n + i];
|
|
|
12817
|
|
698
|
12818 /* Skip words with the NOSUGGEST flag */
|
|
|
12819 if (flags & WF_NOSUGGEST)
|
|
|
12820 continue;
|
|
|
12821
|
|
625
|
12822 if (flags & WF_KEEPCAP)
|
|
|
12823 {
|
|
|
12824 /* Must find the word in the keep-case tree. */
|
|
|
12825 find_keepcap_word(slang, theword, cword);
|
|
|
12826 p = cword;
|
|
|
12827 }
|
|
|
12828 else
|
|
|
12829 {
|
|
|
12830 flags |= su->su_badflags;
|
|
|
12831 if ((flags & WF_CAPMASK) != 0)
|
|
|
12832 {
|
|
|
12833 /* Need to fix case according to "flags". */
|
|
|
12834 make_case_word(theword, cword, flags);
|
|
|
12835 p = cword;
|
|
323
|
12836 }
|
|
|
12837 else
|
|
625
|
12838 p = theword;
|
|
|
12839 }
|
|
|
12840
|
|
|
12841 /* Add the suggestion. */
|
|
|
12842 if (sps_flags & SPS_DOUBLE)
|
|
|
12843 {
|
|
|
12844 /* Add the suggestion if the score isn't too bad. */
|
|
|
12845 if (score <= su->su_maxscore)
|
|
|
12846 add_suggestion(su, &su->su_sga, p, su->su_badlen,
|
|
|
12847 score, 0, FALSE, slang, FALSE);
|
|
|
12848 }
|
|
|
12849 else
|
|
|
12850 {
|
|
|
12851 /* Add a penalty for words in another region. */
|
|
|
12852 if ((flags & WF_REGION)
|
|
|
12853 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
|
12854 goodscore = SCORE_REGION;
|
|
|
12855 else
|
|
|
12856 goodscore = 0;
|
|
|
12857
|
|
|
12858 /* Add a small penalty for changing the first letter from
|
|
|
12859 * lower to upper case. Helps for "tath" -> "Kath", which is
|
|
|
12860 * less common thatn "tath" -> "path". Don't do it when the
|
|
|
12861 * letter is the same, that has already been counted. */
|
|
|
12862 gc = PTR2CHAR(p);
|
|
|
12863 if (SPELL_ISUPPER(gc))
|
|
|
12864 {
|
|
|
12865 bc = PTR2CHAR(su->su_badword);
|
|
|
12866 if (!SPELL_ISUPPER(bc)
|
|
|
12867 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc))
|
|
|
12868 goodscore += SCORE_ICASE / 2;
|
|
|
12869 }
|
|
|
12870
|
|
|
12871 /* Compute the score for the good word. This only does letter
|
|
|
12872 * insert/delete/swap/replace. REP items are not considered,
|
|
|
12873 * which may make the score a bit higher.
|
|
|
12874 * Use a limit for the score to make it work faster. Use
|
|
|
12875 * MAXSCORE(), because RESCORE() will change the score.
|
|
|
12876 * If the limit is very high then the iterative method is
|
|
|
12877 * inefficient, using an array is quicker. */
|
|
|
12878 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score);
|
|
|
12879 if (limit > SCORE_LIMITMAX)
|
|
|
12880 goodscore += spell_edit_score(slang, su->su_badword, p);
|
|
|
12881 else
|
|
|
12882 goodscore += spell_edit_score_limit(slang, su->su_badword,
|
|
|
12883 p, limit);
|
|
|
12884
|
|
|
12885 /* When going over the limit don't bother to do the rest. */
|
|
|
12886 if (goodscore < SCORE_MAXMAX)
|
|
|
12887 {
|
|
|
12888 /* Give a bonus to words seen before. */
|
|
|
12889 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE);
|
|
|
12890
|
|
|
12891 /* Add the suggestion if the score isn't too bad. */
|
|
|
12892 goodscore = RESCORE(goodscore, score);
|
|
|
12893 if (goodscore <= su->su_sfmaxscore)
|
|
|
12894 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
12895 goodscore, score, TRUE, slang, TRUE);
|
|
|
12896 }
|
|
|
12897 }
|
|
|
12898 }
|
|
|
12899 /* smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr); */
|
|
|
12900 }
|
|
|
12901 }
|
|
|
12902
|
|
|
12903 /*
|
|
|
12904 * Find word "word" in fold-case tree for "slang" and return the word number.
|
|
|
12905 */
|
|
|
12906 static int
|
|
|
12907 soundfold_find(slang, word)
|
|
|
12908 slang_T *slang;
|
|
|
12909 char_u *word;
|
|
|
12910 {
|
|
|
12911 idx_T arridx = 0;
|
|
|
12912 int len;
|
|
|
12913 int wlen = 0;
|
|
|
12914 int c;
|
|
|
12915 char_u *ptr = word;
|
|
|
12916 char_u *byts;
|
|
|
12917 idx_T *idxs;
|
|
|
12918 int wordnr = 0;
|
|
|
12919
|
|
|
12920 byts = slang->sl_sbyts;
|
|
|
12921 idxs = slang->sl_sidxs;
|
|
|
12922
|
|
|
12923 for (;;)
|
|
|
12924 {
|
|
|
12925 /* First byte is the number of possible bytes. */
|
|
|
12926 len = byts[arridx++];
|
|
|
12927
|
|
|
12928 /* If the first possible byte is a zero the word could end here.
|
|
|
12929 * If the word ends we found the word. If not skip the NUL bytes. */
|
|
|
12930 c = ptr[wlen];
|
|
|
12931 if (byts[arridx] == NUL)
|
|
|
12932 {
|
|
|
12933 if (c == NUL)
|
|
|
12934 break;
|
|
|
12935
|
|
|
12936 /* Skip over the zeros, there can be several. */
|
|
|
12937 while (len > 0 && byts[arridx] == NUL)
|
|
|
12938 {
|
|
|
12939 ++arridx;
|
|
|
12940 --len;
|
|
|
12941 }
|
|
|
12942 if (len == 0)
|
|
|
12943 return -1; /* no children, word should have ended here */
|
|
|
12944 ++wordnr;
|
|
|
12945 }
|
|
|
12946
|
|
|
12947 /* If the word ends we didn't find it. */
|
|
|
12948 if (c == NUL)
|
|
|
12949 return -1;
|
|
|
12950
|
|
|
12951 /* Perform a binary search in the list of accepted bytes. */
|
|
|
12952 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
12953 c = ' ';
|
|
|
12954 while (byts[arridx] < c)
|
|
|
12955 {
|
|
|
12956 /* The word count is in the first idxs[] entry of the child. */
|
|
|
12957 wordnr += idxs[idxs[arridx]];
|
|
|
12958 ++arridx;
|
|
|
12959 if (--len == 0) /* end of the bytes, didn't find it */
|
|
|
12960 return -1;
|
|
|
12961 }
|
|
|
12962 if (byts[arridx] != c) /* didn't find the byte */
|
|
|
12963 return -1;
|
|
|
12964
|
|
|
12965 /* Continue at the child (if there is one). */
|
|
|
12966 arridx = idxs[arridx];
|
|
|
12967 ++wlen;
|
|
|
12968
|
|
|
12969 /* One space in the good word may stand for several spaces in the
|
|
|
12970 * checked word. */
|
|
|
12971 if (c == ' ')
|
|
|
12972 while (ptr[wlen] == ' ' || ptr[wlen] == TAB)
|
|
|
12973 ++wlen;
|
|
|
12974 }
|
|
|
12975
|
|
|
12976 return wordnr;
|
|
323
|
12977 }
|
|
|
12978
|
|
|
12979 /*
|
|
324
|
12980 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
12981 */
|
|
|
12982 static void
|
|
|
12983 make_case_word(fword, cword, flags)
|
|
|
12984 char_u *fword;
|
|
|
12985 char_u *cword;
|
|
|
12986 int flags;
|
|
|
12987 {
|
|
|
12988 if (flags & WF_ALLCAP)
|
|
|
12989 /* Make it all upper-case */
|
|
|
12990 allcap_copy(fword, cword);
|
|
|
12991 else if (flags & WF_ONECAP)
|
|
|
12992 /* Make the first letter upper-case */
|
|
324
|
12993 onecap_copy(fword, cword, TRUE);
|
|
323
|
12994 else
|
|
|
12995 /* Use goodword as-is. */
|
|
|
12996 STRCPY(cword, fword);
|
|
|
12997 }
|
|
|
12998
|
|
330
|
12999 /*
|
|
|
13000 * Use map string "map" for languages "lp".
|
|
|
13001 */
|
|
|
13002 static void
|
|
|
13003 set_map_str(lp, map)
|
|
|
13004 slang_T *lp;
|
|
|
13005 char_u *map;
|
|
|
13006 {
|
|
|
13007 char_u *p;
|
|
|
13008 int headc = 0;
|
|
|
13009 int c;
|
|
|
13010 int i;
|
|
|
13011
|
|
|
13012 if (*map == NUL)
|
|
|
13013 {
|
|
|
13014 lp->sl_has_map = FALSE;
|
|
|
13015 return;
|
|
|
13016 }
|
|
|
13017 lp->sl_has_map = TRUE;
|
|
|
13018
|
|
625
|
13019 /* Init the array and hash tables empty. */
|
|
330
|
13020 for (i = 0; i < 256; ++i)
|
|
|
13021 lp->sl_map_array[i] = 0;
|
|
|
13022 #ifdef FEAT_MBYTE
|
|
|
13023 hash_init(&lp->sl_map_hash);
|
|
|
13024 #endif
|
|
|
13025
|
|
|
13026 /*
|
|
|
13027 * The similar characters are stored separated with slashes:
|
|
|
13028 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
13029 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
13030 */
|
|
|
13031 for (p = map; *p != NUL; )
|
|
|
13032 {
|
|
|
13033 #ifdef FEAT_MBYTE
|
|
474
|
13034 c = mb_cptr2char_adv(&p);
|
|
330
|
13035 #else
|
|
|
13036 c = *p++;
|
|
|
13037 #endif
|
|
|
13038 if (c == '/')
|
|
|
13039 headc = 0;
|
|
|
13040 else
|
|
|
13041 {
|
|
|
13042 if (headc == 0)
|
|
|
13043 headc = c;
|
|
|
13044
|
|
|
13045 #ifdef FEAT_MBYTE
|
|
|
13046 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
13047 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
13048 * a NUL. */
|
|
|
13049 if (c >= 256)
|
|
|
13050 {
|
|
|
13051 int cl = mb_char2len(c);
|
|
|
13052 int headcl = mb_char2len(headc);
|
|
|
13053 char_u *b;
|
|
|
13054 hash_T hash;
|
|
|
13055 hashitem_T *hi;
|
|
|
13056
|
|
|
13057 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
13058 if (b == NULL)
|
|
|
13059 return;
|
|
|
13060 mb_char2bytes(c, b);
|
|
|
13061 b[cl] = NUL;
|
|
|
13062 mb_char2bytes(headc, b + cl + 1);
|
|
|
13063 b[cl + 1 + headcl] = NUL;
|
|
|
13064 hash = hash_hash(b);
|
|
|
13065 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
13066 if (HASHITEM_EMPTY(hi))
|
|
|
13067 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
13068 else
|
|
|
13069 {
|
|
|
13070 /* This should have been checked when generating the .spl
|
|
|
13071 * file. */
|
|
626
|
13072 EMSG(_("E783: duplicate char in MAP entry"));
|
|
330
|
13073 vim_free(b);
|
|
|
13074 }
|
|
|
13075 }
|
|
|
13076 else
|
|
|
13077 #endif
|
|
|
13078 lp->sl_map_array[c] = headc;
|
|
|
13079 }
|
|
|
13080 }
|
|
|
13081 }
|
|
|
13082
|
|
323
|
13083 /*
|
|
|
13084 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
13085 * lines in the .aff file.
|
|
|
13086 */
|
|
|
13087 static int
|
|
|
13088 similar_chars(slang, c1, c2)
|
|
|
13089 slang_T *slang;
|
|
|
13090 int c1;
|
|
|
13091 int c2;
|
|
|
13092 {
|
|
330
|
13093 int m1, m2;
|
|
|
13094 #ifdef FEAT_MBYTE
|
|
|
13095 char_u buf[MB_MAXBYTES];
|
|
|
13096 hashitem_T *hi;
|
|
|
13097
|
|
|
13098 if (c1 >= 256)
|
|
|
13099 {
|
|
|
13100 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
13101 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
13102 if (HASHITEM_EMPTY(hi))
|
|
|
13103 m1 = 0;
|
|
|
13104 else
|
|
|
13105 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
13106 }
|
|
|
13107 else
|
|
|
13108 #endif
|
|
|
13109 m1 = slang->sl_map_array[c1];
|
|
|
13110 if (m1 == 0)
|
|
323
|
13111 return FALSE;
|
|
330
|
13112
|
|
|
13113
|
|
|
13114 #ifdef FEAT_MBYTE
|
|
|
13115 if (c2 >= 256)
|
|
|
13116 {
|
|
|
13117 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
13118 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
13119 if (HASHITEM_EMPTY(hi))
|
|
|
13120 m2 = 0;
|
|
|
13121 else
|
|
|
13122 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
13123 }
|
|
|
13124 else
|
|
|
13125 #endif
|
|
|
13126 m2 = slang->sl_map_array[c2];
|
|
|
13127
|
|
|
13128 return m1 == m2;
|
|
323
|
13129 }
|
|
|
13130
|
|
|
13131 /*
|
|
|
13132 * Add a suggestion to the list of suggestions.
|
|
625
|
13133 * For a suggestion that is already in the list the lowest score is remembered.
|
|
|
13134 */
|
|
|
13135 static void
|
|
|
13136 add_suggestion(su, gap, goodword, badlenarg, score, altscore, had_bonus,
|
|
|
13137 slang, maxsf)
|
|
323
|
13138 suginfo_T *su;
|
|
625
|
13139 garray_T *gap; /* either su_ga or su_sga */
|
|
323
|
13140 char_u *goodword;
|
|
534
|
13141 int badlenarg; /* len of bad word replaced with "goodword" */
|
|
324
|
13142 int score;
|
|
351
|
13143 int altscore;
|
|
344
|
13144 int had_bonus; /* value for st_had_bonus */
|
|
503
|
13145 slang_T *slang; /* language for sound folding */
|
|
625
|
13146 int maxsf; /* su_maxscore applies to soundfold score,
|
|
|
13147 su_sfmaxscore to the total score. */
|
|
|
13148 {
|
|
|
13149 int goodlen; /* len of goodword changed */
|
|
|
13150 int badlen; /* len of bad word changed */
|
|
323
|
13151 suggest_T *stp;
|
|
534
|
13152 suggest_T new_sug;
|
|
323
|
13153 int i;
|
|
534
|
13154 char_u *pgood, *pbad;
|
|
532
|
13155
|
|
534
|
13156 /* Minimize "badlen" for consistency. Avoids that changing "the the" to
|
|
|
13157 * "thee the" is added next to changing the first "the" the "thee". */
|
|
|
13158 pgood = goodword + STRLEN(goodword);
|
|
625
|
13159 pbad = su->su_badptr + badlenarg;
|
|
|
13160 for (;;)
|
|
|
13161 {
|
|
|
13162 goodlen = pgood - goodword;
|
|
|
13163 badlen = pbad - su->su_badptr;
|
|
|
13164 if (goodlen <= 0 || badlen <= 0)
|
|
|
13165 break;
|
|
534
|
13166 mb_ptr_back(goodword, pgood);
|
|
|
13167 mb_ptr_back(su->su_badptr, pbad);
|
|
|
13168 #ifdef FEAT_MBYTE
|
|
|
13169 if (has_mbyte)
|
|
|
13170 {
|
|
|
13171 if (mb_ptr2char(pgood) != mb_ptr2char(pbad))
|
|
|
13172 break;
|
|
346
|
13173 }
|
|
|
13174 else
|
|
534
|
13175 #endif
|
|
|
13176 if (*pgood != *pbad)
|
|
|
13177 break;
|
|
625
|
13178 }
|
|
|
13179
|
|
534
|
13180 if (badlen == 0 && goodlen == 0)
|
|
|
13181 /* goodword doesn't change anything; may happen for "the the" changing
|
|
|
13182 * the first "the" to itself. */
|
|
|
13183 return;
|
|
346
|
13184
|
|
625
|
13185 /* Check if the word is already there. Also check the length that is
|
|
|
13186 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
13187 * "thes" -> "these". */
|
|
|
13188 stp = &SUG(*gap, 0);
|
|
|
13189 for (i = gap->ga_len; --i >= 0; ++stp)
|
|
|
13190 if (stp->st_wordlen == goodlen
|
|
|
13191 && stp->st_orglen == badlen
|
|
|
13192 && STRNCMP(stp->st_word, goodword, goodlen) == 0)
|
|
|
13193 {
|
|
|
13194 /*
|
|
|
13195 * Found it. Remember the word with the lowest score.
|
|
|
13196 */
|
|
|
13197 if (stp->st_slang == NULL)
|
|
503
|
13198 stp->st_slang = slang;
|
|
625
|
13199
|
|
|
13200 new_sug.st_score = score;
|
|
|
13201 new_sug.st_altscore = altscore;
|
|
|
13202 new_sug.st_had_bonus = had_bonus;
|
|
|
13203
|
|
|
13204 if (stp->st_had_bonus != had_bonus)
|
|
|
13205 {
|
|
|
13206 /* Only one of the two had the soundalike score computed.
|
|
|
13207 * Need to do that for the other one now, otherwise the
|
|
|
13208 * scores can't be compared. This happens because
|
|
|
13209 * suggest_try_change() doesn't compute the soundalike
|
|
|
13210 * word to keep it fast, while some special methods set
|
|
|
13211 * the soundalike score to zero. */
|
|
|
13212 if (had_bonus)
|
|
|
13213 rescore_one(su, stp);
|
|
|
13214 else
|
|
|
13215 {
|
|
|
13216 new_sug.st_word = stp->st_word;
|
|
|
13217 new_sug.st_wordlen = stp->st_wordlen;
|
|
|
13218 new_sug.st_slang = stp->st_slang;
|
|
|
13219 new_sug.st_orglen = badlen;
|
|
|
13220 rescore_one(su, &new_sug);
|
|
|
13221 }
|
|
|
13222 }
|
|
|
13223
|
|
|
13224 if (stp->st_score > new_sug.st_score)
|
|
|
13225 {
|
|
|
13226 stp->st_score = new_sug.st_score;
|
|
|
13227 stp->st_altscore = new_sug.st_altscore;
|
|
|
13228 stp->st_had_bonus = new_sug.st_had_bonus;
|
|
|
13229 }
|
|
|
13230 break;
|
|
|
13231 }
|
|
|
13232
|
|
|
13233 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
|
13234 {
|
|
|
13235 /* Add a suggestion. */
|
|
|
13236 stp = &SUG(*gap, gap->ga_len);
|
|
|
13237 stp->st_word = vim_strnsave(goodword, goodlen);
|
|
|
13238 if (stp->st_word != NULL)
|
|
|
13239 {
|
|
|
13240 stp->st_wordlen = goodlen;
|
|
|
13241 stp->st_score = score;
|
|
|
13242 stp->st_altscore = altscore;
|
|
|
13243 stp->st_had_bonus = had_bonus;
|
|
|
13244 stp->st_orglen = badlen;
|
|
|
13245 stp->st_slang = slang;
|
|
|
13246 ++gap->ga_len;
|
|
|
13247
|
|
|
13248 /* If we have too many suggestions now, sort the list and keep
|
|
|
13249 * the best suggestions. */
|
|
|
13250 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
13251 {
|
|
|
13252 if (maxsf)
|
|
|
13253 su->su_sfmaxscore = cleanup_suggestions(gap,
|
|
|
13254 su->su_sfmaxscore, SUG_CLEAN_COUNT(su));
|
|
|
13255 else
|
|
|
13256 {
|
|
|
13257 i = su->su_maxscore;
|
|
|
13258 su->su_maxscore = cleanup_suggestions(gap,
|
|
|
13259 su->su_maxscore, SUG_CLEAN_COUNT(su));
|
|
|
13260 }
|
|
|
13261 }
|
|
|
13262 }
|
|
|
13263 }
|
|
|
13264 }
|
|
|
13265
|
|
|
13266 /*
|
|
|
13267 * Suggestions may in fact be flagged as errors. Esp. for banned words and
|
|
|
13268 * for split words, such as "the the". Remove these from the list here.
|
|
|
13269 */
|
|
|
13270 static void
|
|
|
13271 check_suggestions(su, gap)
|
|
|
13272 suginfo_T *su;
|
|
|
13273 garray_T *gap; /* either su_ga or su_sga */
|
|
|
13274 {
|
|
|
13275 suggest_T *stp;
|
|
|
13276 int i;
|
|
|
13277 char_u longword[MAXWLEN + 1];
|
|
|
13278 int len;
|
|
|
13279 hlf_T attr;
|
|
|
13280
|
|
|
13281 stp = &SUG(*gap, 0);
|
|
|
13282 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
|
13283 {
|
|
|
13284 /* Need to append what follows to check for "the the". */
|
|
|
13285 STRCPY(longword, stp[i].st_word);
|
|
|
13286 len = stp[i].st_wordlen;
|
|
|
13287 vim_strncpy(longword + len, su->su_badptr + stp[i].st_orglen,
|
|
|
13288 MAXWLEN - len);
|
|
|
13289 attr = HLF_COUNT;
|
|
|
13290 (void)spell_check(curwin, longword, &attr, NULL, FALSE);
|
|
|
13291 if (attr != HLF_COUNT)
|
|
|
13292 {
|
|
|
13293 /* Remove this entry. */
|
|
|
13294 vim_free(stp[i].st_word);
|
|
|
13295 --gap->ga_len;
|
|
|
13296 if (i < gap->ga_len)
|
|
|
13297 mch_memmove(stp + i, stp + i + 1,
|
|
|
13298 sizeof(suggest_T) * (gap->ga_len - i));
|
|
|
13299 }
|
|
|
13300 }
|
|
|
13301 }
|
|
|
13302
|
|
323
|
13303
|
|
|
13304 /*
|
|
|
13305 * Add a word to be banned.
|
|
|
13306 */
|
|
|
13307 static void
|
|
|
13308 add_banned(su, word)
|
|
|
13309 suginfo_T *su;
|
|
|
13310 char_u *word;
|
|
|
13311 {
|
|
626
|
13312 char_u *s;
|
|
323
|
13313 hash_T hash;
|
|
|
13314 hashitem_T *hi;
|
|
|
13315
|
|
625
|
13316 hash = hash_hash(word);
|
|
|
13317 hi = hash_lookup(&su->su_banned, word, hash);
|
|
|
13318 if (HASHITEM_EMPTY(hi))
|
|
|
13319 {
|
|
|
13320 s = vim_strsave(word);
|
|
|
13321 if (s != NULL)
|
|
323
|
13322 hash_add_item(&su->su_banned, hi, s, hash);
|
|
625
|
13323 }
|
|
323
|
13324 }
|
|
|
13325
|
|
324
|
13326 /*
|
|
534
|
13327 * Recompute the score for all suggestions if sound-folding is possible. This
|
|
|
13328 * is slow, thus only done for the final results.
|
|
324
|
13329 */
|
|
|
13330 static void
|
|
|
13331 rescore_suggestions(su)
|
|
|
13332 suginfo_T *su;
|
|
|
13333 {
|
|
|
13334 int i;
|
|
534
|
13335
|
|
|
13336 if (su->su_sallang != NULL)
|
|
503
|
13337 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
534
|
13338 rescore_one(su, &SUG(su->su_ga, i));
|
|
|
13339 }
|
|
|
13340
|
|
|
13341 /*
|
|
|
13342 * Recompute the score for one suggestion if sound-folding is possible.
|
|
|
13343 */
|
|
|
13344 static void
|
|
|
13345 rescore_one(su, stp)
|
|
535
|
13346 suginfo_T *su;
|
|
|
13347 suggest_T *stp;
|
|
534
|
13348 {
|
|
|
13349 slang_T *slang = stp->st_slang;
|
|
|
13350 char_u sal_badword[MAXWLEN];
|
|
535
|
13351 char_u *p;
|
|
534
|
13352
|
|
|
13353 /* Only rescore suggestions that have no sal score yet and do have a
|
|
|
13354 * language. */
|
|
|
13355 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus)
|
|
|
13356 {
|
|
|
13357 if (slang == su->su_sallang)
|
|
535
|
13358 p = su->su_sal_badword;
|
|
534
|
13359 else
|
|
|
13360 {
|
|
|
13361 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword);
|
|
535
|
13362 p = sal_badword;
|
|
|
13363 }
|
|
|
13364
|
|
|
13365 stp->st_altscore = stp_sal_score(stp, su, slang, p);
|
|
534
|
13366 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
13367 stp->st_altscore = SCORE_BIG;
|
|
|
13368 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
|
13369 stp->st_had_bonus = TRUE;
|
|
324
|
13370 }
|
|
|
13371 }
|
|
|
13372
|
|
323
|
13373 static int
|
|
|
13374 #ifdef __BORLANDC__
|
|
|
13375 _RTLENTRYF
|
|
|
13376 #endif
|
|
|
13377 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
13378
|
|
|
13379 /*
|
|
|
13380 * Function given to qsort() to sort the suggestions on st_score.
|
|
529
|
13381 * First on "st_score", then "st_altscore" then alphabetically.
|
|
323
|
13382 */
|
|
|
13383 static int
|
|
|
13384 #ifdef __BORLANDC__
|
|
|
13385 _RTLENTRYF
|
|
|
13386 #endif
|
|
|
13387 sug_compare(s1, s2)
|
|
|
13388 const void *s1;
|
|
|
13389 const void *s2;
|
|
|
13390 {
|
|
|
13391 suggest_T *p1 = (suggest_T *)s1;
|
|
|
13392 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
13393 int n = p1->st_score - p2->st_score;
|
|
|
13394
|
|
|
13395 if (n == 0)
|
|
529
|
13396 {
|
|
|
13397 n = p1->st_altscore - p2->st_altscore;
|
|
|
13398 if (n == 0)
|
|
|
13399 n = STRICMP(p1->st_word, p2->st_word);
|
|
|
13400 }
|
|
344
|
13401 return n;
|
|
323
|
13402 }
|
|
|
13403
|
|
|
13404 /*
|
|
|
13405 * Cleanup the suggestions:
|
|
|
13406 * - Sort on score.
|
|
|
13407 * - Remove words that won't be displayed.
|
|
344
|
13408 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
13409 */
|
|
344
|
13410 static int
|
|
|
13411 cleanup_suggestions(gap, maxscore, keep)
|
|
|
13412 garray_T *gap;
|
|
|
13413 int maxscore;
|
|
324
|
13414 int keep; /* nr of suggestions to keep */
|
|
323
|
13415 {
|
|
344
|
13416 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
13417 int i;
|
|
|
13418
|
|
|
13419 /* Sort the list. */
|
|
344
|
13420 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
13421
|
|
|
13422 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
13423 if (gap->ga_len > keep)
|
|
323
|
13424 {
|
|
344
|
13425 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
13426 vim_free(stp[i].st_word);
|
|
344
|
13427 gap->ga_len = keep;
|
|
|
13428 return stp[keep - 1].st_score;
|
|
323
|
13429 }
|
|
344
|
13430 return maxscore;
|
|
323
|
13431 }
|
|
|
13432
|
|
372
|
13433 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
13434 /*
|
|
|
13435 * Soundfold a string, for soundfold().
|
|
|
13436 * Result is in allocated memory, NULL for an error.
|
|
|
13437 */
|
|
|
13438 char_u *
|
|
|
13439 eval_soundfold(word)
|
|
|
13440 char_u *word;
|
|
|
13441 {
|
|
|
13442 langp_T *lp;
|
|
|
13443 char_u sound[MAXWLEN];
|
|
500
|
13444 int lpi;
|
|
372
|
13445
|
|
|
13446 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
13447 /* Use the sound-folding of the first language that supports it. */
|
|
503
|
13448 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
13449 {
|
|
|
13450 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
372
|
13451 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
13452 {
|
|
|
13453 /* soundfold the word */
|
|
375
|
13454 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
13455 return vim_strsave(sound);
|
|
|
13456 }
|
|
500
|
13457 }
|
|
372
|
13458
|
|
|
13459 /* No language with sound folding, return word as-is. */
|
|
|
13460 return vim_strsave(word);
|
|
|
13461 }
|
|
|
13462 #endif
|
|
|
13463
|
|
323
|
13464 /*
|
|
|
13465 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
485
|
13466 *
|
|
|
13467 * There are many ways to turn a word into a sound-a-like representation. The
|
|
|
13468 * oldest is Soundex (1918!). A nice overview can be found in "Approximate
|
|
|
13469 * swedish name matching - survey and test of different algorithms" by Klas
|
|
|
13470 * Erikson.
|
|
|
13471 *
|
|
|
13472 * We support two methods:
|
|
|
13473 * 1. SOFOFROM/SOFOTO do a simple character mapping.
|
|
|
13474 * 2. SAL items define a more advanced sound-folding (and much slower).
|
|
323
|
13475 */
|
|
|
13476 static void
|
|
375
|
13477 spell_soundfold(slang, inword, folded, res)
|
|
|
13478 slang_T *slang;
|
|
|
13479 char_u *inword;
|
|
|
13480 int folded; /* "inword" is already case-folded */
|
|
|
13481 char_u *res;
|
|
|
13482 {
|
|
|
13483 char_u fword[MAXWLEN];
|
|
|
13484 char_u *word;
|
|
|
13485
|
|
|
13486 if (slang->sl_sofo)
|
|
|
13487 /* SOFOFROM and SOFOTO used */
|
|
|
13488 spell_soundfold_sofo(slang, inword, res);
|
|
|
13489 else
|
|
|
13490 {
|
|
|
13491 /* SAL items used. Requires the word to be case-folded. */
|
|
|
13492 if (folded)
|
|
|
13493 word = inword;
|
|
|
13494 else
|
|
|
13495 {
|
|
|
13496 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
13497 word = fword;
|
|
|
13498 }
|
|
|
13499
|
|
|
13500 #ifdef FEAT_MBYTE
|
|
|
13501 if (has_mbyte)
|
|
|
13502 spell_soundfold_wsal(slang, word, res);
|
|
|
13503 else
|
|
|
13504 #endif
|
|
|
13505 spell_soundfold_sal(slang, word, res);
|
|
|
13506 }
|
|
|
13507 }
|
|
|
13508
|
|
|
13509 /*
|
|
|
13510 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
13511 * SOFOTO lines.
|
|
|
13512 */
|
|
|
13513 static void
|
|
|
13514 spell_soundfold_sofo(slang, inword, res)
|
|
|
13515 slang_T *slang;
|
|
|
13516 char_u *inword;
|
|
|
13517 char_u *res;
|
|
|
13518 {
|
|
|
13519 char_u *s;
|
|
|
13520 int ri = 0;
|
|
|
13521 int c;
|
|
|
13522
|
|
|
13523 #ifdef FEAT_MBYTE
|
|
|
13524 if (has_mbyte)
|
|
|
13525 {
|
|
|
13526 int prevc = 0;
|
|
|
13527 int *ip;
|
|
|
13528
|
|
|
13529 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
13530 * 255, sl_sal the rest. */
|
|
|
13531 for (s = inword; *s != NUL; )
|
|
|
13532 {
|
|
474
|
13533 c = mb_cptr2char_adv(&s);
|
|
375
|
13534 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
13535 c = ' ';
|
|
|
13536 else if (c < 256)
|
|
|
13537 c = slang->sl_sal_first[c];
|
|
|
13538 else
|
|
|
13539 {
|
|
|
13540 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
13541 if (ip == NULL) /* empty list, can't match */
|
|
|
13542 c = NUL;
|
|
|
13543 else
|
|
|
13544 for (;;) /* find "c" in the list */
|
|
|
13545 {
|
|
|
13546 if (*ip == 0) /* not found */
|
|
|
13547 {
|
|
|
13548 c = NUL;
|
|
|
13549 break;
|
|
|
13550 }
|
|
|
13551 if (*ip == c) /* match! */
|
|
|
13552 {
|
|
|
13553 c = ip[1];
|
|
|
13554 break;
|
|
|
13555 }
|
|
|
13556 ip += 2;
|
|
|
13557 }
|
|
|
13558 }
|
|
|
13559
|
|
|
13560 if (c != NUL && c != prevc)
|
|
|
13561 {
|
|
|
13562 ri += mb_char2bytes(c, res + ri);
|
|
|
13563 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
13564 break;
|
|
|
13565 prevc = c;
|
|
|
13566 }
|
|
|
13567 }
|
|
|
13568 }
|
|
|
13569 else
|
|
|
13570 #endif
|
|
|
13571 {
|
|
|
13572 /* The sl_sal_first[] table contains the translation. */
|
|
|
13573 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
13574 {
|
|
|
13575 if (vim_iswhite(c))
|
|
|
13576 c = ' ';
|
|
|
13577 else
|
|
|
13578 c = slang->sl_sal_first[c];
|
|
|
13579 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
13580 res[ri++] = c;
|
|
|
13581 }
|
|
|
13582 }
|
|
|
13583
|
|
|
13584 res[ri] = NUL;
|
|
|
13585 }
|
|
|
13586
|
|
|
13587 static void
|
|
|
13588 spell_soundfold_sal(slang, inword, res)
|
|
323
|
13589 slang_T *slang;
|
|
|
13590 char_u *inword;
|
|
|
13591 char_u *res;
|
|
|
13592 {
|
|
344
|
13593 salitem_T *smp;
|
|
323
|
13594 char_u word[MAXWLEN];
|
|
375
|
13595 char_u *s = inword;
|
|
323
|
13596 char_u *t;
|
|
344
|
13597 char_u *pf;
|
|
323
|
13598 int i, j, z;
|
|
344
|
13599 int reslen;
|
|
323
|
13600 int n, k = 0;
|
|
|
13601 int z0;
|
|
|
13602 int k0;
|
|
|
13603 int n0;
|
|
|
13604 int c;
|
|
|
13605 int pri;
|
|
|
13606 int p0 = -333;
|
|
|
13607 int c0;
|
|
|
13608
|
|
324
|
13609 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
13610 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
13611 if (slang->sl_rem_accents)
|
|
|
13612 {
|
|
|
13613 t = word;
|
|
375
|
13614 while (*s != NUL)
|
|
323
|
13615 {
|
|
324
|
13616 if (vim_iswhite(*s))
|
|
344
|
13617 {
|
|
|
13618 *t++ = ' ';
|
|
|
13619 s = skipwhite(s);
|
|
|
13620 }
|
|
323
|
13621 else
|
|
|
13622 {
|
|
376
|
13623 if (spell_iswordp_nmw(s))
|
|
323
|
13624 *t++ = *s;
|
|
|
13625 ++s;
|
|
|
13626 }
|
|
|
13627 }
|
|
|
13628 *t = NUL;
|
|
|
13629 }
|
|
|
13630 else
|
|
375
|
13631 STRCPY(word, s);
|
|
323
|
13632
|
|
344
|
13633 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
13634
|
|
|
13635 /*
|
|
|
13636 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
13637 * Changed to keep spaces.
|
|
323
|
13638 */
|
|
344
|
13639 i = reslen = z = 0;
|
|
323
|
13640 while ((c = word[i]) != NUL)
|
|
|
13641 {
|
|
344
|
13642 /* Start with the first rule that has the character in the word. */
|
|
323
|
13643 n = slang->sl_sal_first[c];
|
|
|
13644 z0 = 0;
|
|
|
13645
|
|
|
13646 if (n >= 0)
|
|
|
13647 {
|
|
|
13648 /* check all rules for the same letter */
|
|
344
|
13649 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
13650 {
|
|
344
|
13651 /* Quickly skip entries that don't match the word. Most
|
|
|
13652 * entries are less then three chars, optimize for that. */
|
|
|
13653 k = smp[n].sm_leadlen;
|
|
|
13654 if (k > 1)
|
|
323
|
13655 {
|
|
344
|
13656 if (word[i + 1] != s[1])
|
|
|
13657 continue;
|
|
|
13658 if (k > 2)
|
|
|
13659 {
|
|
|
13660 for (j = 2; j < k; ++j)
|
|
|
13661 if (word[i + j] != s[j])
|
|
|
13662 break;
|
|
|
13663 if (j < k)
|
|
|
13664 continue;
|
|
|
13665 }
|
|
323
|
13666 }
|
|
|
13667
|
|
375
|
13668 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
13669 {
|
|
375
|
13670 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
13671 while (*pf != NUL && *pf != word[i + k])
|
|
|
13672 ++pf;
|
|
|
13673 if (*pf == NUL)
|
|
|
13674 continue;
|
|
|
13675 ++k;
|
|
323
|
13676 }
|
|
344
|
13677 s = smp[n].sm_rules;
|
|
|
13678 pri = 5; /* default priority */
|
|
323
|
13679
|
|
|
13680 p0 = *s;
|
|
|
13681 k0 = k;
|
|
|
13682 while (*s == '-' && k > 1)
|
|
|
13683 {
|
|
|
13684 k--;
|
|
|
13685 s++;
|
|
|
13686 }
|
|
|
13687 if (*s == '<')
|
|
|
13688 s++;
|
|
344
|
13689 if (VIM_ISDIGIT(*s))
|
|
323
|
13690 {
|
|
|
13691 /* determine priority */
|
|
|
13692 pri = *s - '0';
|
|
|
13693 s++;
|
|
|
13694 }
|
|
|
13695 if (*s == '^' && *(s + 1) == '^')
|
|
|
13696 s++;
|
|
|
13697
|
|
|
13698 if (*s == NUL
|
|
|
13699 || (*s == '^'
|
|
324
|
13700 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
13701 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
13702 && (*(s + 1) != '$'
|
|
376
|
13703 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
13704 || (*s == '$' && i > 0
|
|
376
|
13705 && spell_iswordp(word + i - 1, curbuf)
|
|
|
13706 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
13707 {
|
|
|
13708 /* search for followup rules, if: */
|
|
|
13709 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
13710 c0 = word[i + k - 1];
|
|
|
13711 n0 = slang->sl_sal_first[c0];
|
|
|
13712
|
|
|
13713 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
13714 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
13715 {
|
|
|
13716 /* test follow-up rule for "word[i + k]" */
|
|
344
|
13717 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
13718 {
|
|
344
|
13719 /* Quickly skip entries that don't match the word.
|
|
|
13720 * */
|
|
|
13721 k0 = smp[n0].sm_leadlen;
|
|
|
13722 if (k0 > 1)
|
|
323
|
13723 {
|
|
344
|
13724 if (word[i + k] != s[1])
|
|
|
13725 continue;
|
|
|
13726 if (k0 > 2)
|
|
|
13727 {
|
|
|
13728 pf = word + i + k + 1;
|
|
|
13729 for (j = 2; j < k0; ++j)
|
|
|
13730 if (*pf++ != s[j])
|
|
|
13731 break;
|
|
|
13732 if (j < k0)
|
|
|
13733 continue;
|
|
|
13734 }
|
|
323
|
13735 }
|
|
344
|
13736 k0 += k - 1;
|
|
|
13737
|
|
375
|
13738 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
13739 {
|
|
344
|
13740 /* Check for match with one of the chars in
|
|
375
|
13741 * "sm_oneof". */
|
|
344
|
13742 while (*pf != NUL && *pf != word[i + k0])
|
|
|
13743 ++pf;
|
|
|
13744 if (*pf == NUL)
|
|
|
13745 continue;
|
|
|
13746 ++k0;
|
|
323
|
13747 }
|
|
344
|
13748
|
|
|
13749 p0 = 5;
|
|
|
13750 s = smp[n0].sm_rules;
|
|
323
|
13751 while (*s == '-')
|
|
|
13752 {
|
|
344
|
13753 /* "k0" gets NOT reduced because
|
|
|
13754 * "if (k0 == k)" */
|
|
323
|
13755 s++;
|
|
|
13756 }
|
|
|
13757 if (*s == '<')
|
|
|
13758 s++;
|
|
344
|
13759 if (VIM_ISDIGIT(*s))
|
|
323
|
13760 {
|
|
|
13761 p0 = *s - '0';
|
|
|
13762 s++;
|
|
|
13763 }
|
|
|
13764
|
|
|
13765 if (*s == NUL
|
|
|
13766 /* *s == '^' cuts */
|
|
|
13767 || (*s == '$'
|
|
376
|
13768 && !spell_iswordp(word + i + k0,
|
|
|
13769 curbuf)))
|
|
323
|
13770 {
|
|
|
13771 if (k0 == k)
|
|
|
13772 /* this is just a piece of the string */
|
|
|
13773 continue;
|
|
|
13774
|
|
|
13775 if (p0 < pri)
|
|
|
13776 /* priority too low */
|
|
|
13777 continue;
|
|
|
13778 /* rule fits; stop search */
|
|
|
13779 break;
|
|
|
13780 }
|
|
|
13781 }
|
|
|
13782
|
|
344
|
13783 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
13784 continue;
|
|
|
13785 }
|
|
|
13786
|
|
|
13787 /* replace string */
|
|
344
|
13788 s = smp[n].sm_to;
|
|
389
|
13789 if (s == NULL)
|
|
|
13790 s = (char_u *)"";
|
|
344
|
13791 pf = smp[n].sm_rules;
|
|
|
13792 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
13793 if (p0 == 1 && z == 0)
|
|
|
13794 {
|
|
|
13795 /* rule with '<' is used */
|
|
344
|
13796 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
13797 || res[reslen - 1] == *s))
|
|
|
13798 reslen--;
|
|
323
|
13799 z0 = 1;
|
|
|
13800 z = 1;
|
|
|
13801 k0 = 0;
|
|
372
|
13802 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
13803 {
|
|
|
13804 word[i + k0] = *s;
|
|
|
13805 k0++;
|
|
|
13806 s++;
|
|
|
13807 }
|
|
|
13808 if (k > k0)
|
|
|
13809 mch_memmove(word + i + k0, word + i + k,
|
|
|
13810 STRLEN(word + i + k) + 1);
|
|
|
13811
|
|
|
13812 /* new "actual letter" */
|
|
|
13813 c = word[i];
|
|
|
13814 }
|
|
|
13815 else
|
|
|
13816 {
|
|
|
13817 /* no '<' rule used */
|
|
|
13818 i += k - 1;
|
|
|
13819 z = 0;
|
|
344
|
13820 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
13821 {
|
|
344
|
13822 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
13823 res[reslen++] = *s;
|
|
323
|
13824 s++;
|
|
|
13825 }
|
|
|
13826 /* new "actual letter" */
|
|
|
13827 c = *s;
|
|
344
|
13828 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
13829 {
|
|
|
13830 if (c != NUL)
|
|
372
|
13831 res[reslen++] = c;
|
|
323
|
13832 mch_memmove(word, word + i + 1,
|
|
|
13833 STRLEN(word + i + 1) + 1);
|
|
|
13834 i = 0;
|
|
|
13835 z0 = 1;
|
|
|
13836 }
|
|
|
13837 }
|
|
|
13838 break;
|
|
|
13839 }
|
|
|
13840 }
|
|
|
13841 }
|
|
324
|
13842 else if (vim_iswhite(c))
|
|
|
13843 {
|
|
|
13844 c = ' ';
|
|
|
13845 k = 1;
|
|
|
13846 }
|
|
323
|
13847
|
|
|
13848 if (z0 == 0)
|
|
|
13849 {
|
|
344
|
13850 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
13851 && (!slang->sl_collapse || reslen == 0
|
|
|
13852 || res[reslen - 1] != c))
|
|
323
|
13853 /* condense only double letters */
|
|
372
|
13854 res[reslen++] = c;
|
|
323
|
13855
|
|
|
13856 i++;
|
|
|
13857 z = 0;
|
|
|
13858 k = 0;
|
|
|
13859 }
|
|
|
13860 }
|
|
|
13861
|
|
344
|
13862 res[reslen] = NUL;
|
|
323
|
13863 }
|
|
|
13864
|
|
372
|
13865 #ifdef FEAT_MBYTE
|
|
|
13866 /*
|
|
|
13867 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
13868 * Multi-byte version of spell_soundfold().
|
|
|
13869 */
|
|
|
13870 static void
|
|
375
|
13871 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
13872 slang_T *slang;
|
|
|
13873 char_u *inword;
|
|
|
13874 char_u *res;
|
|
|
13875 {
|
|
375
|
13876 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
13877 int word[MAXWLEN];
|
|
|
13878 int wres[MAXWLEN];
|
|
|
13879 int l;
|
|
|
13880 char_u *s;
|
|
|
13881 int *ws;
|
|
|
13882 char_u *t;
|
|
|
13883 int *pf;
|
|
|
13884 int i, j, z;
|
|
|
13885 int reslen;
|
|
|
13886 int n, k = 0;
|
|
|
13887 int z0;
|
|
|
13888 int k0;
|
|
|
13889 int n0;
|
|
|
13890 int c;
|
|
|
13891 int pri;
|
|
|
13892 int p0 = -333;
|
|
|
13893 int c0;
|
|
|
13894 int did_white = FALSE;
|
|
|
13895
|
|
|
13896 /*
|
|
|
13897 * Convert the multi-byte string to a wide-character string.
|
|
|
13898 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
13899 * But keep white space.
|
|
|
13900 */
|
|
|
13901 n = 0;
|
|
|
13902 for (s = inword; *s != NUL; )
|
|
|
13903 {
|
|
|
13904 t = s;
|
|
474
|
13905 c = mb_cptr2char_adv(&s);
|
|
372
|
13906 if (slang->sl_rem_accents)
|
|
|
13907 {
|
|
|
13908 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
13909 {
|
|
|
13910 if (did_white)
|
|
|
13911 continue;
|
|
|
13912 c = ' ';
|
|
|
13913 did_white = TRUE;
|
|
|
13914 }
|
|
|
13915 else
|
|
|
13916 {
|
|
|
13917 did_white = FALSE;
|
|
376
|
13918 if (!spell_iswordp_nmw(t))
|
|
372
|
13919 continue;
|
|
|
13920 }
|
|
|
13921 }
|
|
|
13922 word[n++] = c;
|
|
|
13923 }
|
|
|
13924 word[n] = NUL;
|
|
|
13925
|
|
|
13926 /*
|
|
|
13927 * This comes from Aspell phonet.cpp.
|
|
|
13928 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
13929 * Changed to keep spaces.
|
|
|
13930 */
|
|
|
13931 i = reslen = z = 0;
|
|
|
13932 while ((c = word[i]) != NUL)
|
|
|
13933 {
|
|
|
13934 /* Start with the first rule that has the character in the word. */
|
|
|
13935 n = slang->sl_sal_first[c & 0xff];
|
|
|
13936 z0 = 0;
|
|
|
13937
|
|
|
13938 if (n >= 0)
|
|
|
13939 {
|
|
375
|
13940 /* check all rules for the same index byte */
|
|
372
|
13941 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
13942 {
|
|
|
13943 /* Quickly skip entries that don't match the word. Most
|
|
|
13944 * entries are less then three chars, optimize for that. */
|
|
375
|
13945 if (c != ws[0])
|
|
|
13946 continue;
|
|
372
|
13947 k = smp[n].sm_leadlen;
|
|
|
13948 if (k > 1)
|
|
|
13949 {
|
|
|
13950 if (word[i + 1] != ws[1])
|
|
|
13951 continue;
|
|
|
13952 if (k > 2)
|
|
|
13953 {
|
|
|
13954 for (j = 2; j < k; ++j)
|
|
|
13955 if (word[i + j] != ws[j])
|
|
|
13956 break;
|
|
|
13957 if (j < k)
|
|
|
13958 continue;
|
|
|
13959 }
|
|
|
13960 }
|
|
|
13961
|
|
375
|
13962 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
13963 {
|
|
375
|
13964 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
13965 while (*pf != NUL && *pf != word[i + k])
|
|
|
13966 ++pf;
|
|
|
13967 if (*pf == NUL)
|
|
|
13968 continue;
|
|
|
13969 ++k;
|
|
|
13970 }
|
|
|
13971 s = smp[n].sm_rules;
|
|
|
13972 pri = 5; /* default priority */
|
|
|
13973
|
|
|
13974 p0 = *s;
|
|
|
13975 k0 = k;
|
|
|
13976 while (*s == '-' && k > 1)
|
|
|
13977 {
|
|
|
13978 k--;
|
|
|
13979 s++;
|
|
|
13980 }
|
|
|
13981 if (*s == '<')
|
|
|
13982 s++;
|
|
|
13983 if (VIM_ISDIGIT(*s))
|
|
|
13984 {
|
|
|
13985 /* determine priority */
|
|
|
13986 pri = *s - '0';
|
|
|
13987 s++;
|
|
|
13988 }
|
|
|
13989 if (*s == '^' && *(s + 1) == '^')
|
|
|
13990 s++;
|
|
|
13991
|
|
|
13992 if (*s == NUL
|
|
|
13993 || (*s == '^'
|
|
|
13994 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
13995 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
13996 && (*(s + 1) != '$'
|
|
376
|
13997 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
13998 || (*s == '$' && i > 0
|
|
376
|
13999 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
14000 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
14001 {
|
|
|
14002 /* search for followup rules, if: */
|
|
|
14003 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
14004 c0 = word[i + k - 1];
|
|
|
14005 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
14006
|
|
|
14007 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
14008 && p0 != '-' && word[i + k] != NUL)
|
|
|
14009 {
|
|
375
|
14010 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
14011 * all entries with the same index byte. */
|
|
372
|
14012 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
14013 == (c0 & 0xff); ++n0)
|
|
|
14014 {
|
|
|
14015 /* Quickly skip entries that don't match the word.
|
|
375
|
14016 */
|
|
|
14017 if (c0 != ws[0])
|
|
|
14018 continue;
|
|
372
|
14019 k0 = smp[n0].sm_leadlen;
|
|
|
14020 if (k0 > 1)
|
|
|
14021 {
|
|
|
14022 if (word[i + k] != ws[1])
|
|
|
14023 continue;
|
|
|
14024 if (k0 > 2)
|
|
|
14025 {
|
|
|
14026 pf = word + i + k + 1;
|
|
|
14027 for (j = 2; j < k0; ++j)
|
|
|
14028 if (*pf++ != ws[j])
|
|
|
14029 break;
|
|
|
14030 if (j < k0)
|
|
|
14031 continue;
|
|
|
14032 }
|
|
|
14033 }
|
|
|
14034 k0 += k - 1;
|
|
|
14035
|
|
375
|
14036 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
14037 {
|
|
|
14038 /* Check for match with one of the chars in
|
|
375
|
14039 * "sm_oneof". */
|
|
372
|
14040 while (*pf != NUL && *pf != word[i + k0])
|
|
|
14041 ++pf;
|
|
|
14042 if (*pf == NUL)
|
|
|
14043 continue;
|
|
|
14044 ++k0;
|
|
|
14045 }
|
|
|
14046
|
|
|
14047 p0 = 5;
|
|
|
14048 s = smp[n0].sm_rules;
|
|
|
14049 while (*s == '-')
|
|
|
14050 {
|
|
|
14051 /* "k0" gets NOT reduced because
|
|
|
14052 * "if (k0 == k)" */
|
|
|
14053 s++;
|
|
|
14054 }
|
|
|
14055 if (*s == '<')
|
|
|
14056 s++;
|
|
|
14057 if (VIM_ISDIGIT(*s))
|
|
|
14058 {
|
|
|
14059 p0 = *s - '0';
|
|
|
14060 s++;
|
|
|
14061 }
|
|
|
14062
|
|
|
14063 if (*s == NUL
|
|
|
14064 /* *s == '^' cuts */
|
|
|
14065 || (*s == '$'
|
|
376
|
14066 && !spell_iswordp_w(word + i + k0,
|
|
|
14067 curbuf)))
|
|
372
|
14068 {
|
|
|
14069 if (k0 == k)
|
|
|
14070 /* this is just a piece of the string */
|
|
|
14071 continue;
|
|
|
14072
|
|
|
14073 if (p0 < pri)
|
|
|
14074 /* priority too low */
|
|
|
14075 continue;
|
|
|
14076 /* rule fits; stop search */
|
|
|
14077 break;
|
|
|
14078 }
|
|
|
14079 }
|
|
|
14080
|
|
|
14081 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
14082 == (c0 & 0xff))
|
|
|
14083 continue;
|
|
|
14084 }
|
|
|
14085
|
|
|
14086 /* replace string */
|
|
|
14087 ws = smp[n].sm_to_w;
|
|
|
14088 s = smp[n].sm_rules;
|
|
|
14089 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
14090 if (p0 == 1 && z == 0)
|
|
|
14091 {
|
|
|
14092 /* rule with '<' is used */
|
|
389
|
14093 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
14094 && (wres[reslen - 1] == c
|
|
372
|
14095 || wres[reslen - 1] == *ws))
|
|
|
14096 reslen--;
|
|
|
14097 z0 = 1;
|
|
|
14098 z = 1;
|
|
|
14099 k0 = 0;
|
|
389
|
14100 if (ws != NULL)
|
|
|
14101 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
14102 {
|
|
|
14103 word[i + k0] = *ws;
|
|
|
14104 k0++;
|
|
|
14105 ws++;
|
|
|
14106 }
|
|
372
|
14107 if (k > k0)
|
|
|
14108 mch_memmove(word + i + k0, word + i + k,
|
|
|
14109 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
14110
|
|
|
14111 /* new "actual letter" */
|
|
|
14112 c = word[i];
|
|
|
14113 }
|
|
|
14114 else
|
|
|
14115 {
|
|
|
14116 /* no '<' rule used */
|
|
|
14117 i += k - 1;
|
|
|
14118 z = 0;
|
|
389
|
14119 if (ws != NULL)
|
|
|
14120 while (*ws != NUL && ws[1] != NUL
|
|
|
14121 && reslen < MAXWLEN)
|
|
|
14122 {
|
|
|
14123 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
14124 wres[reslen++] = *ws;
|
|
|
14125 ws++;
|
|
|
14126 }
|
|
372
|
14127 /* new "actual letter" */
|
|
389
|
14128 if (ws == NULL)
|
|
|
14129 c = NUL;
|
|
|
14130 else
|
|
|
14131 c = *ws;
|
|
372
|
14132 if (strstr((char *)s, "^^") != NULL)
|
|
|
14133 {
|
|
|
14134 if (c != NUL)
|
|
|
14135 wres[reslen++] = c;
|
|
|
14136 mch_memmove(word, word + i + 1,
|
|
|
14137 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
14138 i = 0;
|
|
|
14139 z0 = 1;
|
|
|
14140 }
|
|
|
14141 }
|
|
|
14142 break;
|
|
|
14143 }
|
|
|
14144 }
|
|
|
14145 }
|
|
|
14146 else if (vim_iswhite(c))
|
|
|
14147 {
|
|
|
14148 c = ' ';
|
|
|
14149 k = 1;
|
|
|
14150 }
|
|
|
14151
|
|
|
14152 if (z0 == 0)
|
|
|
14153 {
|
|
|
14154 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
14155 && (!slang->sl_collapse || reslen == 0
|
|
|
14156 || wres[reslen - 1] != c))
|
|
|
14157 /* condense only double letters */
|
|
|
14158 wres[reslen++] = c;
|
|
|
14159
|
|
|
14160 i++;
|
|
|
14161 z = 0;
|
|
|
14162 k = 0;
|
|
|
14163 }
|
|
|
14164 }
|
|
|
14165
|
|
|
14166 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
14167 l = 0;
|
|
|
14168 for (n = 0; n < reslen; ++n)
|
|
|
14169 {
|
|
|
14170 l += mb_char2bytes(wres[n], res + l);
|
|
|
14171 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
14172 break;
|
|
|
14173 }
|
|
|
14174 res[l] = NUL;
|
|
|
14175 }
|
|
|
14176 #endif
|
|
|
14177
|
|
324
|
14178 /*
|
|
344
|
14179 * Compute a score for two sound-a-like words.
|
|
|
14180 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
14181 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
14182 * avoiding checks that will not be possible.
|
|
|
14183 */
|
|
|
14184 static int
|
|
351
|
14185 soundalike_score(goodstart, badstart)
|
|
|
14186 char_u *goodstart; /* sound-folded good word */
|
|
|
14187 char_u *badstart; /* sound-folded bad word */
|
|
344
|
14188 {
|
|
351
|
14189 char_u *goodsound = goodstart;
|
|
|
14190 char_u *badsound = badstart;
|
|
|
14191 int goodlen;
|
|
|
14192 int badlen;
|
|
344
|
14193 int n;
|
|
|
14194 char_u *pl, *ps;
|
|
|
14195 char_u *pl2, *ps2;
|
|
351
|
14196 int score = 0;
|
|
|
14197
|
|
|
14198 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
14199 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
14200 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
14201 {
|
|
625
|
14202 if (badsound[1] == goodsound[1]
|
|
|
14203 || (badsound[1] != NUL
|
|
|
14204 && goodsound[1] != NUL
|
|
|
14205 && badsound[2] == goodsound[2]))
|
|
|
14206 {
|
|
|
14207 /* handle like a substitute */
|
|
|
14208 }
|
|
351
|
14209 else
|
|
625
|
14210 {
|
|
|
14211 score = 2 * SCORE_DEL / 3;
|
|
|
14212 if (*badsound == '*')
|
|
|
14213 ++badsound;
|
|
|
14214 else
|
|
|
14215 ++goodsound;
|
|
|
14216 }
|
|
351
|
14217 }
|
|
|
14218
|
|
|
14219 goodlen = STRLEN(goodsound);
|
|
|
14220 badlen = STRLEN(badsound);
|
|
344
|
14221
|
|
|
14222 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
14223 * changes. */
|
|
|
14224 n = goodlen - badlen;
|
|
|
14225 if (n < -2 || n > 2)
|
|
|
14226 return SCORE_MAXMAX;
|
|
|
14227
|
|
|
14228 if (n > 0)
|
|
|
14229 {
|
|
351
|
14230 pl = goodsound; /* goodsound is longest */
|
|
344
|
14231 ps = badsound;
|
|
|
14232 }
|
|
|
14233 else
|
|
|
14234 {
|
|
351
|
14235 pl = badsound; /* badsound is longest */
|
|
344
|
14236 ps = goodsound;
|
|
|
14237 }
|
|
|
14238
|
|
|
14239 /* Skip over the identical part. */
|
|
|
14240 while (*pl == *ps && *pl != NUL)
|
|
|
14241 {
|
|
|
14242 ++pl;
|
|
|
14243 ++ps;
|
|
|
14244 }
|
|
|
14245
|
|
|
14246 switch (n)
|
|
|
14247 {
|
|
|
14248 case -2:
|
|
|
14249 case 2:
|
|
|
14250 /*
|
|
|
14251 * Must delete two characters from "pl".
|
|
|
14252 */
|
|
|
14253 ++pl; /* first delete */
|
|
|
14254 while (*pl == *ps)
|
|
|
14255 {
|
|
|
14256 ++pl;
|
|
|
14257 ++ps;
|
|
|
14258 }
|
|
|
14259 /* strings must be equal after second delete */
|
|
|
14260 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
14261 return score + SCORE_DEL * 2;
|
|
344
|
14262
|
|
|
14263 /* Failed to compare. */
|
|
|
14264 break;
|
|
|
14265
|
|
|
14266 case -1:
|
|
|
14267 case 1:
|
|
|
14268 /*
|
|
|
14269 * Minimal one delete from "pl" required.
|
|
|
14270 */
|
|
|
14271
|
|
|
14272 /* 1: delete */
|
|
|
14273 pl2 = pl + 1;
|
|
|
14274 ps2 = ps;
|
|
|
14275 while (*pl2 == *ps2)
|
|
|
14276 {
|
|
|
14277 if (*pl2 == NUL) /* reached the end */
|
|
351
|
14278 return score + SCORE_DEL;
|
|
344
|
14279 ++pl2;
|
|
|
14280 ++ps2;
|
|
|
14281 }
|
|
|
14282
|
|
|
14283 /* 2: delete then swap, then rest must be equal */
|
|
|
14284 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
14285 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
14286 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
14287
|
|
|
14288 /* 3: delete then substitute, then the rest must be equal */
|
|
|
14289 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
14290 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
14291
|
|
|
14292 /* 4: first swap then delete */
|
|
|
14293 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
14294 {
|
|
|
14295 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
14296 ps2 = ps + 2;
|
|
|
14297 while (*pl2 == *ps2)
|
|
|
14298 {
|
|
|
14299 ++pl2;
|
|
|
14300 ++ps2;
|
|
|
14301 }
|
|
|
14302 /* delete a char and then strings must be equal */
|
|
|
14303 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
14304 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
14305 }
|
|
|
14306
|
|
|
14307 /* 5: first substitute then delete */
|
|
|
14308 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
14309 ps2 = ps + 1;
|
|
|
14310 while (*pl2 == *ps2)
|
|
|
14311 {
|
|
|
14312 ++pl2;
|
|
|
14313 ++ps2;
|
|
|
14314 }
|
|
|
14315 /* delete a char and then strings must be equal */
|
|
|
14316 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
14317 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
14318
|
|
|
14319 /* Failed to compare. */
|
|
|
14320 break;
|
|
|
14321
|
|
|
14322 case 0:
|
|
|
14323 /*
|
|
|
14324 * Lenghts are equal, thus changes must result in same length: An
|
|
|
14325 * insert is only possible in combination with a delete.
|
|
|
14326 * 1: check if for identical strings
|
|
|
14327 */
|
|
|
14328 if (*pl == NUL)
|
|
351
|
14329 return score;
|
|
344
|
14330
|
|
|
14331 /* 2: swap */
|
|
|
14332 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
14333 {
|
|
|
14334 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
14335 ps2 = ps + 2;
|
|
|
14336 while (*pl2 == *ps2)
|
|
|
14337 {
|
|
|
14338 if (*pl2 == NUL) /* reached the end */
|
|
351
|
14339 return score + SCORE_SWAP;
|
|
344
|
14340 ++pl2;
|
|
|
14341 ++ps2;
|
|
|
14342 }
|
|
|
14343 /* 3: swap and swap again */
|
|
|
14344 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
14345 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
14346 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
14347
|
|
|
14348 /* 4: swap and substitute */
|
|
|
14349 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
14350 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
14351 }
|
|
|
14352
|
|
|
14353 /* 5: substitute */
|
|
|
14354 pl2 = pl + 1;
|
|
|
14355 ps2 = ps + 1;
|
|
|
14356 while (*pl2 == *ps2)
|
|
|
14357 {
|
|
|
14358 if (*pl2 == NUL) /* reached the end */
|
|
351
|
14359 return score + SCORE_SUBST;
|
|
344
|
14360 ++pl2;
|
|
|
14361 ++ps2;
|
|
|
14362 }
|
|
|
14363
|
|
|
14364 /* 6: substitute and swap */
|
|
|
14365 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
14366 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
14367 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
14368
|
|
|
14369 /* 7: substitute and substitute */
|
|
|
14370 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
14371 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
14372
|
|
|
14373 /* 8: insert then delete */
|
|
|
14374 pl2 = pl;
|
|
|
14375 ps2 = ps + 1;
|
|
|
14376 while (*pl2 == *ps2)
|
|
|
14377 {
|
|
|
14378 ++pl2;
|
|
|
14379 ++ps2;
|
|
|
14380 }
|
|
|
14381 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
14382 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
14383
|
|
|
14384 /* 9: delete then insert */
|
|
|
14385 pl2 = pl + 1;
|
|
|
14386 ps2 = ps;
|
|
|
14387 while (*pl2 == *ps2)
|
|
|
14388 {
|
|
|
14389 ++pl2;
|
|
|
14390 ++ps2;
|
|
|
14391 }
|
|
|
14392 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
14393 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
14394
|
|
|
14395 /* Failed to compare. */
|
|
|
14396 break;
|
|
|
14397 }
|
|
|
14398
|
|
|
14399 return SCORE_MAXMAX;
|
|
|
14400 }
|
|
324
|
14401
|
|
323
|
14402 /*
|
|
|
14403 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
14404 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
14405 *
|
|
485
|
14406 * The algorithm is described by Du and Chang, 1992.
|
|
|
14407 * The implementation of the algorithm comes from Aspell editdist.cpp,
|
|
|
14408 * edit_distance(). It has been converted from C++ to C and modified to
|
|
|
14409 * support multi-byte characters.
|
|
323
|
14410 */
|
|
|
14411 static int
|
|
625
|
14412 spell_edit_score(slang, badword, goodword)
|
|
|
14413 slang_T *slang;
|
|
323
|
14414 char_u *badword;
|
|
|
14415 char_u *goodword;
|
|
|
14416 {
|
|
|
14417 int *cnt;
|
|
372
|
14418 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
14419 int j, i;
|
|
|
14420 int t;
|
|
|
14421 int bc, gc;
|
|
324
|
14422 int pbc, pgc;
|
|
|
14423 #ifdef FEAT_MBYTE
|
|
|
14424 char_u *p;
|
|
|
14425 int wbadword[MAXWLEN];
|
|
|
14426 int wgoodword[MAXWLEN];
|
|
|
14427
|
|
|
14428 if (has_mbyte)
|
|
|
14429 {
|
|
|
14430 /* Get the characters from the multi-byte strings and put them in an
|
|
|
14431 * int array for easy access. */
|
|
|
14432 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
14433 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
14434 wbadword[badlen++] = 0;
|
|
324
|
14435 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
14436 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
14437 wgoodword[goodlen++] = 0;
|
|
324
|
14438 }
|
|
|
14439 else
|
|
|
14440 #endif
|
|
|
14441 {
|
|
|
14442 badlen = STRLEN(badword) + 1;
|
|
|
14443 goodlen = STRLEN(goodword) + 1;
|
|
|
14444 }
|
|
323
|
14445
|
|
|
14446 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
14447 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
14448 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
14449 TRUE);
|
|
324
|
14450 if (cnt == NULL)
|
|
|
14451 return 0; /* out of memory */
|
|
323
|
14452
|
|
|
14453 CNT(0, 0) = 0;
|
|
|
14454 for (j = 1; j <= goodlen; ++j)
|
|
625
|
14455 CNT(0, j) = CNT(0, j - 1) + SCORE_INS;
|
|
323
|
14456
|
|
|
14457 for (i = 1; i <= badlen; ++i)
|
|
|
14458 {
|
|
625
|
14459 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL;
|
|
323
|
14460 for (j = 1; j <= goodlen; ++j)
|
|
|
14461 {
|
|
324
|
14462 #ifdef FEAT_MBYTE
|
|
|
14463 if (has_mbyte)
|
|
|
14464 {
|
|
|
14465 bc = wbadword[i - 1];
|
|
|
14466 gc = wgoodword[j - 1];
|
|
|
14467 }
|
|
|
14468 else
|
|
|
14469 #endif
|
|
|
14470 {
|
|
|
14471 bc = badword[i - 1];
|
|
|
14472 gc = goodword[j - 1];
|
|
|
14473 }
|
|
323
|
14474 if (bc == gc)
|
|
|
14475 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
14476 else
|
|
|
14477 {
|
|
|
14478 /* Use a better score when there is only a case difference. */
|
|
324
|
14479 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
14480 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
14481 else
|
|
625
|
14482 {
|
|
|
14483 /* For a similar character use SCORE_SIMILAR. */
|
|
|
14484 if (slang != NULL
|
|
|
14485 && slang->sl_has_map
|
|
|
14486 && similar_chars(slang, gc, bc))
|
|
|
14487 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1);
|
|
|
14488 else
|
|
|
14489 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
14490 }
|
|
323
|
14491
|
|
324
|
14492 if (i > 1 && j > 1)
|
|
323
|
14493 {
|
|
324
|
14494 #ifdef FEAT_MBYTE
|
|
|
14495 if (has_mbyte)
|
|
|
14496 {
|
|
|
14497 pbc = wbadword[i - 2];
|
|
|
14498 pgc = wgoodword[j - 2];
|
|
|
14499 }
|
|
|
14500 else
|
|
|
14501 #endif
|
|
|
14502 {
|
|
|
14503 pbc = badword[i - 2];
|
|
|
14504 pgc = goodword[j - 2];
|
|
|
14505 }
|
|
|
14506 if (bc == pgc && pbc == gc)
|
|
|
14507 {
|
|
|
14508 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
14509 if (t < CNT(i, j))
|
|
|
14510 CNT(i, j) = t;
|
|
|
14511 }
|
|
323
|
14512 }
|
|
|
14513 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
14514 if (t < CNT(i, j))
|
|
|
14515 CNT(i, j) = t;
|
|
|
14516 t = SCORE_INS + CNT(i, j - 1);
|
|
|
14517 if (t < CNT(i, j))
|
|
|
14518 CNT(i, j) = t;
|
|
|
14519 }
|
|
|
14520 }
|
|
|
14521 }
|
|
344
|
14522
|
|
|
14523 i = CNT(badlen - 1, goodlen - 1);
|
|
|
14524 vim_free(cnt);
|
|
|
14525 return i;
|
|
323
|
14526 }
|
|
307
|
14527
|
|
625
|
14528 typedef struct
|
|
|
14529 {
|
|
|
14530 int badi;
|
|
|
14531 int goodi;
|
|
|
14532 int score;
|
|
|
14533 } limitscore_T;
|
|
|
14534
|
|
|
14535 /*
|
|
|
14536 * Like spell_edit_score(), but with a limit on the score to make it faster.
|
|
|
14537 * May return SCORE_MAXMAX when the score is higher than "limit".
|
|
|
14538 *
|
|
|
14539 * This uses a stack for the edits still to be tried.
|
|
|
14540 * The idea comes from Aspell leditdist.cpp. Rewritten in C and added support
|
|
|
14541 * for multi-byte characters.
|
|
|
14542 */
|
|
|
14543 static int
|
|
|
14544 spell_edit_score_limit(slang, badword, goodword, limit)
|
|
|
14545 slang_T *slang;
|
|
|
14546 char_u *badword;
|
|
|
14547 char_u *goodword;
|
|
|
14548 int limit;
|
|
|
14549 {
|
|
|
14550 limitscore_T stack[10]; /* allow for over 3 * 2 edits */
|
|
|
14551 int stackidx;
|
|
|
14552 int bi, gi;
|
|
|
14553 int bi2, gi2;
|
|
|
14554 int bc, gc;
|
|
|
14555 int score;
|
|
|
14556 int score_off;
|
|
|
14557 int minscore;
|
|
|
14558 int round;
|
|
|
14559
|
|
|
14560 #ifdef FEAT_MBYTE
|
|
|
14561 /* Multi-byte characters require a bit more work, use a different function
|
|
|
14562 * to avoid testing "has_mbyte" quite often. */
|
|
|
14563 if (has_mbyte)
|
|
|
14564 return spell_edit_score_limit_w(slang, badword, goodword, limit);
|
|
|
14565 #endif
|
|
|
14566
|
|
|
14567 /*
|
|
|
14568 * The idea is to go from start to end over the words. So long as
|
|
|
14569 * characters are equal just continue, this always gives the lowest score.
|
|
|
14570 * When there is a difference try several alternatives. Each alternative
|
|
|
14571 * increases "score" for the edit distance. Some of the alternatives are
|
|
|
14572 * pushed unto a stack and tried later, some are tried right away. At the
|
|
|
14573 * end of the word the score for one alternative is known. The lowest
|
|
|
14574 * possible score is stored in "minscore".
|
|
|
14575 */
|
|
|
14576 stackidx = 0;
|
|
|
14577 bi = 0;
|
|
|
14578 gi = 0;
|
|
|
14579 score = 0;
|
|
|
14580 minscore = limit + 1;
|
|
|
14581
|
|
|
14582 for (;;)
|
|
|
14583 {
|
|
|
14584 /* Skip over an equal part, score remains the same. */
|
|
|
14585 for (;;)
|
|
|
14586 {
|
|
|
14587 bc = badword[bi];
|
|
|
14588 gc = goodword[gi];
|
|
|
14589 if (bc != gc) /* stop at a char that's different */
|
|
|
14590 break;
|
|
|
14591 if (bc == NUL) /* both words end */
|
|
|
14592 {
|
|
|
14593 if (score < minscore)
|
|
|
14594 minscore = score;
|
|
|
14595 goto pop; /* do next alternative */
|
|
|
14596 }
|
|
|
14597 ++bi;
|
|
|
14598 ++gi;
|
|
|
14599 }
|
|
|
14600
|
|
|
14601 if (gc == NUL) /* goodword ends, delete badword chars */
|
|
|
14602 {
|
|
|
14603 do
|
|
|
14604 {
|
|
|
14605 if ((score += SCORE_DEL) >= minscore)
|
|
|
14606 goto pop; /* do next alternative */
|
|
|
14607 } while (badword[++bi] != NUL);
|
|
|
14608 minscore = score;
|
|
|
14609 }
|
|
|
14610 else if (bc == NUL) /* badword ends, insert badword chars */
|
|
|
14611 {
|
|
|
14612 do
|
|
|
14613 {
|
|
|
14614 if ((score += SCORE_INS) >= minscore)
|
|
|
14615 goto pop; /* do next alternative */
|
|
|
14616 } while (goodword[++gi] != NUL);
|
|
|
14617 minscore = score;
|
|
|
14618 }
|
|
|
14619 else /* both words continue */
|
|
|
14620 {
|
|
|
14621 /* If not close to the limit, perform a change. Only try changes
|
|
|
14622 * that may lead to a lower score than "minscore".
|
|
|
14623 * round 0: try deleting a char from badword
|
|
|
14624 * round 1: try inserting a char in badword */
|
|
|
14625 for (round = 0; round <= 1; ++round)
|
|
|
14626 {
|
|
|
14627 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS);
|
|
|
14628 if (score_off < minscore)
|
|
|
14629 {
|
|
|
14630 if (score_off + SCORE_EDIT_MIN >= minscore)
|
|
|
14631 {
|
|
|
14632 /* Near the limit, rest of the words must match. We
|
|
|
14633 * can check that right now, no need to push an item
|
|
|
14634 * onto the stack. */
|
|
|
14635 bi2 = bi + 1 - round;
|
|
|
14636 gi2 = gi + round;
|
|
|
14637 while (goodword[gi2] == badword[bi2])
|
|
|
14638 {
|
|
|
14639 if (goodword[gi2] == NUL)
|
|
|
14640 {
|
|
|
14641 minscore = score_off;
|
|
|
14642 break;
|
|
|
14643 }
|
|
|
14644 ++bi2;
|
|
|
14645 ++gi2;
|
|
|
14646 }
|
|
|
14647 }
|
|
|
14648 else
|
|
|
14649 {
|
|
|
14650 /* try deleting/inserting a character later */
|
|
|
14651 stack[stackidx].badi = bi + 1 - round;
|
|
|
14652 stack[stackidx].goodi = gi + round;
|
|
|
14653 stack[stackidx].score = score_off;
|
|
|
14654 ++stackidx;
|
|
|
14655 }
|
|
|
14656 }
|
|
|
14657 }
|
|
|
14658
|
|
|
14659 if (score + SCORE_SWAP < minscore)
|
|
|
14660 {
|
|
|
14661 /* If swapping two characters makes a match then the
|
|
|
14662 * substitution is more expensive, thus there is no need to
|
|
|
14663 * try both. */
|
|
|
14664 if (gc == badword[bi + 1] && bc == goodword[gi + 1])
|
|
|
14665 {
|
|
|
14666 /* Swap two characters, that is: skip them. */
|
|
|
14667 gi += 2;
|
|
|
14668 bi += 2;
|
|
|
14669 score += SCORE_SWAP;
|
|
|
14670 continue;
|
|
|
14671 }
|
|
|
14672 }
|
|
|
14673
|
|
|
14674 /* Substitute one character for another which is the same
|
|
|
14675 * thing as deleting a character from both goodword and badword.
|
|
|
14676 * Use a better score when there is only a case difference. */
|
|
|
14677 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
|
14678 score += SCORE_ICASE;
|
|
|
14679 else
|
|
|
14680 {
|
|
|
14681 /* For a similar character use SCORE_SIMILAR. */
|
|
|
14682 if (slang != NULL
|
|
|
14683 && slang->sl_has_map
|
|
|
14684 && similar_chars(slang, gc, bc))
|
|
|
14685 score += SCORE_SIMILAR;
|
|
|
14686 else
|
|
|
14687 score += SCORE_SUBST;
|
|
|
14688 }
|
|
|
14689
|
|
|
14690 if (score < minscore)
|
|
|
14691 {
|
|
|
14692 /* Do the substitution. */
|
|
|
14693 ++gi;
|
|
|
14694 ++bi;
|
|
|
14695 continue;
|
|
|
14696 }
|
|
|
14697 }
|
|
|
14698 pop:
|
|
|
14699 /*
|
|
|
14700 * Get here to try the next alternative, pop it from the stack.
|
|
|
14701 */
|
|
|
14702 if (stackidx == 0) /* stack is empty, finished */
|
|
|
14703 break;
|
|
|
14704
|
|
|
14705 /* pop an item from the stack */
|
|
|
14706 --stackidx;
|
|
|
14707 gi = stack[stackidx].goodi;
|
|
|
14708 bi = stack[stackidx].badi;
|
|
|
14709 score = stack[stackidx].score;
|
|
|
14710 }
|
|
|
14711
|
|
|
14712 /* When the score goes over "limit" it may actually be much higher.
|
|
|
14713 * Return a very large number to avoid going below the limit when giving a
|
|
|
14714 * bonus. */
|
|
|
14715 if (minscore > limit)
|
|
|
14716 return SCORE_MAXMAX;
|
|
|
14717 return minscore;
|
|
|
14718 }
|
|
|
14719
|
|
|
14720 #ifdef FEAT_MBYTE
|
|
|
14721 /*
|
|
|
14722 * Multi-byte version of spell_edit_score_limit().
|
|
|
14723 * Keep it in sync with the above!
|
|
|
14724 */
|
|
|
14725 static int
|
|
|
14726 spell_edit_score_limit_w(slang, badword, goodword, limit)
|
|
|
14727 slang_T *slang;
|
|
|
14728 char_u *badword;
|
|
|
14729 char_u *goodword;
|
|
|
14730 int limit;
|
|
|
14731 {
|
|
|
14732 limitscore_T stack[10]; /* allow for over 3 * 2 edits */
|
|
|
14733 int stackidx;
|
|
|
14734 int bi, gi;
|
|
|
14735 int bi2, gi2;
|
|
|
14736 int bc, gc;
|
|
|
14737 int score;
|
|
|
14738 int score_off;
|
|
|
14739 int minscore;
|
|
|
14740 int round;
|
|
|
14741 char_u *p;
|
|
|
14742 int wbadword[MAXWLEN];
|
|
|
14743 int wgoodword[MAXWLEN];
|
|
|
14744
|
|
|
14745 /* Get the characters from the multi-byte strings and put them in an
|
|
|
14746 * int array for easy access. */
|
|
|
14747 bi = 0;
|
|
|
14748 for (p = badword; *p != NUL; )
|
|
|
14749 wbadword[bi++] = mb_cptr2char_adv(&p);
|
|
|
14750 wbadword[bi++] = 0;
|
|
|
14751 gi = 0;
|
|
|
14752 for (p = goodword; *p != NUL; )
|
|
|
14753 wgoodword[gi++] = mb_cptr2char_adv(&p);
|
|
|
14754 wgoodword[gi++] = 0;
|
|
|
14755
|
|
|
14756 /*
|
|
|
14757 * The idea is to go from start to end over the words. So long as
|
|
|
14758 * characters are equal just continue, this always gives the lowest score.
|
|
|
14759 * When there is a difference try several alternatives. Each alternative
|
|
|
14760 * increases "score" for the edit distance. Some of the alternatives are
|
|
|
14761 * pushed unto a stack and tried later, some are tried right away. At the
|
|
|
14762 * end of the word the score for one alternative is known. The lowest
|
|
|
14763 * possible score is stored in "minscore".
|
|
|
14764 */
|
|
|
14765 stackidx = 0;
|
|
|
14766 bi = 0;
|
|
|
14767 gi = 0;
|
|
|
14768 score = 0;
|
|
|
14769 minscore = limit + 1;
|
|
|
14770
|
|
|
14771 for (;;)
|
|
|
14772 {
|
|
|
14773 /* Skip over an equal part, score remains the same. */
|
|
|
14774 for (;;)
|
|
|
14775 {
|
|
|
14776 bc = wbadword[bi];
|
|
|
14777 gc = wgoodword[gi];
|
|
|
14778
|
|
|
14779 if (bc != gc) /* stop at a char that's different */
|
|
|
14780 break;
|
|
|
14781 if (bc == NUL) /* both words end */
|
|
|
14782 {
|
|
|
14783 if (score < minscore)
|
|
|
14784 minscore = score;
|
|
|
14785 goto pop; /* do next alternative */
|
|
|
14786 }
|
|
|
14787 ++bi;
|
|
|
14788 ++gi;
|
|
|
14789 }
|
|
|
14790
|
|
|
14791 if (gc == NUL) /* goodword ends, delete badword chars */
|
|
|
14792 {
|
|
|
14793 do
|
|
|
14794 {
|
|
|
14795 if ((score += SCORE_DEL) >= minscore)
|
|
|
14796 goto pop; /* do next alternative */
|
|
|
14797 } while (wbadword[++bi] != NUL);
|
|
|
14798 minscore = score;
|
|
|
14799 }
|
|
|
14800 else if (bc == NUL) /* badword ends, insert badword chars */
|
|
|
14801 {
|
|
|
14802 do
|
|
|
14803 {
|
|
|
14804 if ((score += SCORE_INS) >= minscore)
|
|
|
14805 goto pop; /* do next alternative */
|
|
|
14806 } while (wgoodword[++gi] != NUL);
|
|
|
14807 minscore = score;
|
|
|
14808 }
|
|
|
14809 else /* both words continue */
|
|
|
14810 {
|
|
|
14811 /* If not close to the limit, perform a change. Only try changes
|
|
|
14812 * that may lead to a lower score than "minscore".
|
|
|
14813 * round 0: try deleting a char from badword
|
|
|
14814 * round 1: try inserting a char in badword */
|
|
|
14815 for (round = 0; round <= 1; ++round)
|
|
|
14816 {
|
|
|
14817 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS);
|
|
|
14818 if (score_off < minscore)
|
|
|
14819 {
|
|
|
14820 if (score_off + SCORE_EDIT_MIN >= minscore)
|
|
|
14821 {
|
|
|
14822 /* Near the limit, rest of the words must match. We
|
|
|
14823 * can check that right now, no need to push an item
|
|
|
14824 * onto the stack. */
|
|
|
14825 bi2 = bi + 1 - round;
|
|
|
14826 gi2 = gi + round;
|
|
|
14827 while (wgoodword[gi2] == wbadword[bi2])
|
|
|
14828 {
|
|
|
14829 if (wgoodword[gi2] == NUL)
|
|
|
14830 {
|
|
|
14831 minscore = score_off;
|
|
|
14832 break;
|
|
|
14833 }
|
|
|
14834 ++bi2;
|
|
|
14835 ++gi2;
|
|
|
14836 }
|
|
|
14837 }
|
|
|
14838 else
|
|
|
14839 {
|
|
|
14840 /* try deleting a character from badword later */
|
|
|
14841 stack[stackidx].badi = bi + 1 - round;
|
|
|
14842 stack[stackidx].goodi = gi + round;
|
|
|
14843 stack[stackidx].score = score_off;
|
|
|
14844 ++stackidx;
|
|
|
14845 }
|
|
|
14846 }
|
|
|
14847 }
|
|
|
14848
|
|
|
14849 if (score + SCORE_SWAP < minscore)
|
|
|
14850 {
|
|
|
14851 /* If swapping two characters makes a match then the
|
|
|
14852 * substitution is more expensive, thus there is no need to
|
|
|
14853 * try both. */
|
|
|
14854 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1])
|
|
|
14855 {
|
|
|
14856 /* Swap two characters, that is: skip them. */
|
|
|
14857 gi += 2;
|
|
|
14858 bi += 2;
|
|
|
14859 score += SCORE_SWAP;
|
|
|
14860 continue;
|
|
|
14861 }
|
|
|
14862 }
|
|
|
14863
|
|
|
14864 /* Substitute one character for another which is the same
|
|
|
14865 * thing as deleting a character from both goodword and badword.
|
|
|
14866 * Use a better score when there is only a case difference. */
|
|
|
14867 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
|
14868 score += SCORE_ICASE;
|
|
|
14869 else
|
|
|
14870 {
|
|
|
14871 /* For a similar character use SCORE_SIMILAR. */
|
|
|
14872 if (slang != NULL
|
|
|
14873 && slang->sl_has_map
|
|
|
14874 && similar_chars(slang, gc, bc))
|
|
|
14875 score += SCORE_SIMILAR;
|
|
|
14876 else
|
|
|
14877 score += SCORE_SUBST;
|
|
|
14878 }
|
|
|
14879
|
|
|
14880 if (score < minscore)
|
|
|
14881 {
|
|
|
14882 /* Do the substitution. */
|
|
|
14883 ++gi;
|
|
|
14884 ++bi;
|
|
|
14885 continue;
|
|
|
14886 }
|
|
|
14887 }
|
|
|
14888 pop:
|
|
|
14889 /*
|
|
|
14890 * Get here to try the next alternative, pop it from the stack.
|
|
|
14891 */
|
|
|
14892 if (stackidx == 0) /* stack is empty, finished */
|
|
|
14893 break;
|
|
|
14894
|
|
|
14895 /* pop an item from the stack */
|
|
|
14896 --stackidx;
|
|
|
14897 gi = stack[stackidx].goodi;
|
|
|
14898 bi = stack[stackidx].badi;
|
|
|
14899 score = stack[stackidx].score;
|
|
|
14900 }
|
|
|
14901
|
|
|
14902 /* When the score goes over "limit" it may actually be much higher.
|
|
|
14903 * Return a very large number to avoid going below the limit when giving a
|
|
|
14904 * bonus. */
|
|
|
14905 if (minscore > limit)
|
|
|
14906 return SCORE_MAXMAX;
|
|
|
14907 return minscore;
|
|
|
14908 }
|
|
|
14909 #endif
|
|
|
14910
|
|
714
|
14911 /*
|
|
|
14912 * ":spellinfo"
|
|
|
14913 */
|
|
|
14914 /*ARGSUSED*/
|
|
|
14915 void
|
|
|
14916 ex_spellinfo(eap)
|
|
|
14917 exarg_T *eap;
|
|
|
14918 {
|
|
|
14919 int lpi;
|
|
|
14920 langp_T *lp;
|
|
|
14921 char_u *p;
|
|
|
14922
|
|
|
14923 if (no_spell_checking(curwin))
|
|
|
14924 return;
|
|
|
14925
|
|
|
14926 msg_start();
|
|
|
14927 for (lpi = 0; lpi < curbuf->b_langp.ga_len && !got_int; ++lpi)
|
|
|
14928 {
|
|
|
14929 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
14930 msg_puts((char_u *)"file: ");
|
|
|
14931 msg_puts(lp->lp_slang->sl_fname);
|
|
|
14932 msg_putchar('\n');
|
|
|
14933 p = lp->lp_slang->sl_info;
|
|
|
14934 if (p != NULL)
|
|
|
14935 {
|
|
|
14936 msg_puts(p);
|
|
|
14937 msg_putchar('\n');
|
|
|
14938 }
|
|
|
14939 }
|
|
|
14940 msg_end();
|
|
|
14941 }
|
|
|
14942
|
|
625
|
14943 #define DUMPFLAG_KEEPCASE 1 /* round 2: keep-case tree */
|
|
|
14944 #define DUMPFLAG_COUNT 2 /* include word count */
|
|
701
|
14945 #define DUMPFLAG_ICASE 4 /* ignore case when finding matches */
|
|
710
|
14946 #define DUMPFLAG_ONECAP 8 /* pattern starts with capital */
|
|
|
14947 #define DUMPFLAG_ALLCAP 16 /* pattern is all capitals */
|
|
625
|
14948
|
|
351
|
14949 /*
|
|
|
14950 * ":spelldump"
|
|
|
14951 */
|
|
|
14952 void
|
|
|
14953 ex_spelldump(eap)
|
|
|
14954 exarg_T *eap;
|
|
|
14955 {
|
|
|
14956 buf_T *buf = curbuf;
|
|
701
|
14957
|
|
|
14958 if (no_spell_checking(curwin))
|
|
|
14959 return;
|
|
|
14960
|
|
|
14961 /* Create a new empty buffer by splitting the window. */
|
|
|
14962 do_cmdline_cmd((char_u *)"new");
|
|
|
14963 if (!bufempty() || !buf_valid(buf))
|
|
|
14964 return;
|
|
|
14965
|
|
|
14966 spell_dump_compl(buf, NULL, 0, NULL, eap->forceit ? DUMPFLAG_COUNT : 0);
|
|
|
14967
|
|
|
14968 /* Delete the empty line that we started with. */
|
|
|
14969 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
14970 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
14971
|
|
|
14972 redraw_later(NOT_VALID);
|
|
|
14973 }
|
|
|
14974
|
|
|
14975 /*
|
|
|
14976 * Go through all possible words and:
|
|
|
14977 * 1. When "pat" is NULL: dump a list of all words in the current buffer.
|
|
|
14978 * "ic" and "dir" are not used.
|
|
|
14979 * 2. When "pat" is not NULL: add matching words to insert mode completion.
|
|
|
14980 */
|
|
|
14981 void
|
|
|
14982 spell_dump_compl(buf, pat, ic, dir, dumpflags_arg)
|
|
|
14983 buf_T *buf; /* buffer with spell checking */
|
|
|
14984 char_u *pat; /* leading part of the word */
|
|
|
14985 int ic; /* ignore case */
|
|
|
14986 int *dir; /* direction for adding matches */
|
|
|
14987 int dumpflags_arg; /* DUMPFLAG_* */
|
|
|
14988 {
|
|
351
|
14989 langp_T *lp;
|
|
|
14990 slang_T *slang;
|
|
|
14991 idx_T arridx[MAXWLEN];
|
|
|
14992 int curi[MAXWLEN];
|
|
|
14993 char_u word[MAXWLEN];
|
|
|
14994 int c;
|
|
|
14995 char_u *byts;
|
|
|
14996 idx_T *idxs;
|
|
|
14997 linenr_T lnum = 0;
|
|
|
14998 int round;
|
|
|
14999 int depth;
|
|
|
15000 int n;
|
|
|
15001 int flags;
|
|
381
|
15002 char_u *region_names = NULL; /* region names being used */
|
|
|
15003 int do_region = TRUE; /* dump region names and numbers */
|
|
|
15004 char_u *p;
|
|
500
|
15005 int lpi;
|
|
701
|
15006 int dumpflags = dumpflags_arg;
|
|
|
15007 int patlen;
|
|
|
15008
|
|
710
|
15009 /* When ignoring case or when the pattern starts with capital pass this on
|
|
|
15010 * to dump_word(). */
|
|
|
15011 if (pat != NULL)
|
|
|
15012 {
|
|
|
15013 if (ic)
|
|
|
15014 dumpflags |= DUMPFLAG_ICASE;
|
|
|
15015 else
|
|
|
15016 {
|
|
|
15017 n = captype(pat, NULL);
|
|
|
15018 if (n == WF_ONECAP)
|
|
|
15019 dumpflags |= DUMPFLAG_ONECAP;
|
|
|
15020 else if (n == WF_ALLCAP
|
|
|
15021 #ifdef FEAT_MBYTE
|
|
714
|
15022 && (int)STRLEN(pat) > mb_ptr2len(pat)
|
|
710
|
15023 #else
|
|
714
|
15024 && (int)STRLEN(pat) > 1
|
|
710
|
15025 #endif
|
|
|
15026 )
|
|
|
15027 dumpflags |= DUMPFLAG_ALLCAP;
|
|
|
15028 }
|
|
|
15029 }
|
|
351
|
15030
|
|
381
|
15031 /* Find out if we can support regions: All languages must support the same
|
|
|
15032 * regions or none at all. */
|
|
500
|
15033 for (lpi = 0; lpi < buf->b_langp.ga_len; ++lpi)
|
|
|
15034 {
|
|
|
15035 lp = LANGP_ENTRY(buf->b_langp, lpi);
|
|
381
|
15036 p = lp->lp_slang->sl_regions;
|
|
|
15037 if (p[0] != 0)
|
|
|
15038 {
|
|
|
15039 if (region_names == NULL) /* first language with regions */
|
|
|
15040 region_names = p;
|
|
|
15041 else if (STRCMP(region_names, p) != 0)
|
|
|
15042 {
|
|
|
15043 do_region = FALSE; /* region names are different */
|
|
|
15044 break;
|
|
|
15045 }
|
|
|
15046 }
|
|
|
15047 }
|
|
|
15048
|
|
|
15049 if (do_region && region_names != NULL)
|
|
|
15050 {
|
|
701
|
15051 if (pat == NULL)
|
|
|
15052 {
|
|
|
15053 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
15054 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
15055 }
|
|
381
|
15056 }
|
|
|
15057 else
|
|
|
15058 do_region = FALSE;
|
|
|
15059
|
|
|
15060 /*
|
|
|
15061 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
15062 */
|
|
500
|
15063 for (lpi = 0; lpi < buf->b_langp.ga_len; ++lpi)
|
|
|
15064 {
|
|
|
15065 lp = LANGP_ENTRY(buf->b_langp, lpi);
|
|
351
|
15066 slang = lp->lp_slang;
|
|
500
|
15067 if (slang->sl_fbyts == NULL) /* reloading failed */
|
|
|
15068 continue;
|
|
351
|
15069
|
|
701
|
15070 if (pat == NULL)
|
|
|
15071 {
|
|
|
15072 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
15073 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
15074 }
|
|
|
15075
|
|
|
15076 /* When matching with a pattern and there are no prefixes only use
|
|
|
15077 * parts of the tree that match "pat". */
|
|
|
15078 if (pat != NULL && slang->sl_pbyts == NULL)
|
|
|
15079 patlen = STRLEN(pat);
|
|
|
15080 else
|
|
|
15081 patlen = 0;
|
|
351
|
15082
|
|
|
15083 /* round 1: case-folded tree
|
|
|
15084 * round 2: keep-case tree */
|
|
|
15085 for (round = 1; round <= 2; ++round)
|
|
|
15086 {
|
|
|
15087 if (round == 1)
|
|
|
15088 {
|
|
701
|
15089 dumpflags &= ~DUMPFLAG_KEEPCASE;
|
|
351
|
15090 byts = slang->sl_fbyts;
|
|
|
15091 idxs = slang->sl_fidxs;
|
|
|
15092 }
|
|
|
15093 else
|
|
|
15094 {
|
|
701
|
15095 dumpflags |= DUMPFLAG_KEEPCASE;
|
|
351
|
15096 byts = slang->sl_kbyts;
|
|
|
15097 idxs = slang->sl_kidxs;
|
|
|
15098 }
|
|
|
15099 if (byts == NULL)
|
|
|
15100 continue; /* array is empty */
|
|
|
15101
|
|
|
15102 depth = 0;
|
|
|
15103 arridx[0] = 0;
|
|
|
15104 curi[0] = 1;
|
|
701
|
15105 while (depth >= 0 && !got_int
|
|
|
15106 && (pat == NULL || !compl_interrupted))
|
|
351
|
15107 {
|
|
|
15108 if (curi[depth] > byts[arridx[depth]])
|
|
|
15109 {
|
|
|
15110 /* Done all bytes at this node, go up one level. */
|
|
|
15111 --depth;
|
|
|
15112 line_breakcheck();
|
|
|
15113 }
|
|
|
15114 else
|
|
|
15115 {
|
|
|
15116 /* Do one more byte at this node. */
|
|
|
15117 n = arridx[depth] + curi[depth];
|
|
|
15118 ++curi[depth];
|
|
|
15119 c = byts[n];
|
|
|
15120 if (c == 0)
|
|
|
15121 {
|
|
|
15122 /* End of word, deal with the word.
|
|
|
15123 * Don't use keep-case words in the fold-case tree,
|
|
|
15124 * they will appear in the keep-case tree.
|
|
|
15125 * Only use the word when the region matches. */
|
|
|
15126 flags = (int)idxs[n];
|
|
|
15127 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
500
|
15128 && (flags & WF_NEEDCOMP) == 0
|
|
381
|
15129 && (do_region
|
|
|
15130 || (flags & WF_REGION) == 0
|
|
390
|
15131 || (((unsigned)flags >> 16)
|
|
351
|
15132 & lp->lp_region) != 0))
|
|
|
15133 {
|
|
|
15134 word[depth] = NUL;
|
|
381
|
15135 if (!do_region)
|
|
|
15136 flags &= ~WF_REGION;
|
|
355
|
15137
|
|
|
15138 /* Dump the basic word if there is no prefix or
|
|
|
15139 * when it's the first one. */
|
|
390
|
15140 c = (unsigned)flags >> 24;
|
|
355
|
15141 if (c == 0 || curi[depth] == 2)
|
|
701
|
15142 {
|
|
|
15143 dump_word(slang, word, pat, dir,
|
|
|
15144 dumpflags, flags, lnum);
|
|
|
15145 if (pat == NULL)
|
|
|
15146 ++lnum;
|
|
|
15147 }
|
|
351
|
15148
|
|
|
15149 /* Apply the prefix, if there is one. */
|
|
355
|
15150 if (c != 0)
|
|
701
|
15151 lnum = dump_prefixes(slang, word, pat, dir,
|
|
|
15152 dumpflags, flags, lnum);
|
|
351
|
15153 }
|
|
|
15154 }
|
|
|
15155 else
|
|
|
15156 {
|
|
|
15157 /* Normal char, go one level deeper. */
|
|
|
15158 word[depth++] = c;
|
|
|
15159 arridx[depth] = idxs[n];
|
|
|
15160 curi[depth] = 1;
|
|
701
|
15161
|
|
|
15162 /* Check if this characters matches with the pattern.
|
|
|
15163 * If not skip the whole tree below it.
|
|
710
|
15164 * Always ignore case here, dump_word() will check
|
|
|
15165 * proper case later. This isn't exactly right when
|
|
|
15166 * length changes for multi-byte characters with
|
|
|
15167 * ignore case... */
|
|
|
15168 if (depth <= patlen
|
|
|
15169 && MB_STRNICMP(word, pat, depth) != 0)
|
|
701
|
15170 --depth;
|
|
351
|
15171 }
|
|
|
15172 }
|
|
|
15173 }
|
|
|
15174 }
|
|
|
15175 }
|
|
|
15176 }
|
|
|
15177
|
|
|
15178 /*
|
|
|
15179 * Dump one word: apply case modifications and append a line to the buffer.
|
|
701
|
15180 * When "lnum" is zero add insert mode completion.
|
|
|
15181 */
|
|
|
15182 static void
|
|
710
|
15183 dump_word(slang, word, pat, dir, dumpflags, wordflags, lnum)
|
|
625
|
15184 slang_T *slang;
|
|
351
|
15185 char_u *word;
|
|
701
|
15186 char_u *pat;
|
|
|
15187 int *dir;
|
|
625
|
15188 int dumpflags;
|
|
710
|
15189 int wordflags;
|
|
351
|
15190 linenr_T lnum;
|
|
|
15191 {
|
|
|
15192 int keepcap = FALSE;
|
|
|
15193 char_u *p;
|
|
625
|
15194 char_u *tw;
|
|
351
|
15195 char_u cword[MAXWLEN];
|
|
381
|
15196 char_u badword[MAXWLEN + 10];
|
|
|
15197 int i;
|
|
710
|
15198 int flags = wordflags;
|
|
|
15199
|
|
|
15200 if (dumpflags & DUMPFLAG_ONECAP)
|
|
|
15201 flags |= WF_ONECAP;
|
|
|
15202 if (dumpflags & DUMPFLAG_ALLCAP)
|
|
|
15203 flags |= WF_ALLCAP;
|
|
351
|
15204
|
|
625
|
15205 if ((dumpflags & DUMPFLAG_KEEPCASE) == 0 && (flags & WF_CAPMASK) != 0)
|
|
351
|
15206 {
|
|
|
15207 /* Need to fix case according to "flags". */
|
|
|
15208 make_case_word(word, cword, flags);
|
|
|
15209 p = cword;
|
|
|
15210 }
|
|
|
15211 else
|
|
|
15212 {
|
|
|
15213 p = word;
|
|
625
|
15214 if ((dumpflags & DUMPFLAG_KEEPCASE)
|
|
|
15215 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
389
|
15216 || (flags & WF_FIXCAP) != 0))
|
|
351
|
15217 keepcap = TRUE;
|
|
|
15218 }
|
|
625
|
15219 tw = p;
|
|
351
|
15220
|
|
701
|
15221 if (pat == NULL)
|
|
|
15222 {
|
|
|
15223 /* Add flags and regions after a slash. */
|
|
|
15224 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
15225 {
|
|
|
15226 STRCPY(badword, p);
|
|
|
15227 STRCAT(badword, "/");
|
|
|
15228 if (keepcap)
|
|
|
15229 STRCAT(badword, "=");
|
|
|
15230 if (flags & WF_BANNED)
|
|
|
15231 STRCAT(badword, "!");
|
|
|
15232 else if (flags & WF_RARE)
|
|
|
15233 STRCAT(badword, "?");
|
|
|
15234 if (flags & WF_REGION)
|
|
|
15235 for (i = 0; i < 7; ++i)
|
|
|
15236 if (flags & (0x10000 << i))
|
|
|
15237 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
|
15238 p = badword;
|
|
|
15239 }
|
|
|
15240
|
|
|
15241 if (dumpflags & DUMPFLAG_COUNT)
|
|
|
15242 {
|
|
|
15243 hashitem_T *hi;
|
|
|
15244
|
|
|
15245 /* Include the word count for ":spelldump!". */
|
|
|
15246 hi = hash_find(&slang->sl_wordcount, tw);
|
|
|
15247 if (!HASHITEM_EMPTY(hi))
|
|
|
15248 {
|
|
|
15249 vim_snprintf((char *)IObuff, IOSIZE, "%s\t%d",
|
|
625
|
15250 tw, HI2WC(hi)->wc_count);
|
|
701
|
15251 p = IObuff;
|
|
|
15252 }
|
|
|
15253 }
|
|
|
15254
|
|
|
15255 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
15256 }
|
|
710
|
15257 else if (((dumpflags & DUMPFLAG_ICASE)
|
|
|
15258 ? MB_STRNICMP(p, pat, STRLEN(pat)) == 0
|
|
|
15259 : STRNCMP(p, pat, STRLEN(pat)) == 0)
|
|
701
|
15260 && ins_compl_add_infercase(p, (int)STRLEN(p),
|
|
|
15261 dumpflags & DUMPFLAG_ICASE,
|
|
|
15262 NULL, *dir, 0) == OK)
|
|
710
|
15263 /* if dir was BACKWARD then honor it just once */
|
|
|
15264 *dir = FORWARD;
|
|
351
|
15265 }
|
|
|
15266
|
|
|
15267 /*
|
|
372
|
15268 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
15269 * "word" and append a line to the buffer.
|
|
701
|
15270 * When "lnum" is zero add insert mode completion.
|
|
351
|
15271 * Return the updated line number.
|
|
|
15272 */
|
|
|
15273 static linenr_T
|
|
701
|
15274 dump_prefixes(slang, word, pat, dir, dumpflags, flags, startlnum)
|
|
351
|
15275 slang_T *slang;
|
|
|
15276 char_u *word; /* case-folded word */
|
|
701
|
15277 char_u *pat;
|
|
|
15278 int *dir;
|
|
625
|
15279 int dumpflags;
|
|
351
|
15280 int flags; /* flags with prefix ID */
|
|
|
15281 linenr_T startlnum;
|
|
|
15282 {
|
|
|
15283 idx_T arridx[MAXWLEN];
|
|
|
15284 int curi[MAXWLEN];
|
|
|
15285 char_u prefix[MAXWLEN];
|
|
455
|
15286 char_u word_up[MAXWLEN];
|
|
|
15287 int has_word_up = FALSE;
|
|
351
|
15288 int c;
|
|
|
15289 char_u *byts;
|
|
|
15290 idx_T *idxs;
|
|
|
15291 linenr_T lnum = startlnum;
|
|
|
15292 int depth;
|
|
|
15293 int n;
|
|
|
15294 int len;
|
|
|
15295 int i;
|
|
|
15296
|
|
626
|
15297 /* If the word starts with a lower-case letter make the word with an
|
|
455
|
15298 * upper-case letter in word_up[]. */
|
|
|
15299 c = PTR2CHAR(word);
|
|
|
15300 if (SPELL_TOUPPER(c) != c)
|
|
|
15301 {
|
|
|
15302 onecap_copy(word, word_up, TRUE);
|
|
|
15303 has_word_up = TRUE;
|
|
|
15304 }
|
|
|
15305
|
|
351
|
15306 byts = slang->sl_pbyts;
|
|
|
15307 idxs = slang->sl_pidxs;
|
|
|
15308 if (byts != NULL) /* array not is empty */
|
|
|
15309 {
|
|
|
15310 /*
|
|
|
15311 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
15312 * When at the end of a prefix check that it supports "flags".
|
|
351
|
15313 */
|
|
|
15314 depth = 0;
|
|
|
15315 arridx[0] = 0;
|
|
|
15316 curi[0] = 1;
|
|
|
15317 while (depth >= 0 && !got_int)
|
|
|
15318 {
|
|
390
|
15319 n = arridx[depth];
|
|
|
15320 len = byts[n];
|
|
|
15321 if (curi[depth] > len)
|
|
351
|
15322 {
|
|
|
15323 /* Done all bytes at this node, go up one level. */
|
|
|
15324 --depth;
|
|
|
15325 line_breakcheck();
|
|
|
15326 }
|
|
|
15327 else
|
|
|
15328 {
|
|
|
15329 /* Do one more byte at this node. */
|
|
390
|
15330 n += curi[depth];
|
|
351
|
15331 ++curi[depth];
|
|
|
15332 c = byts[n];
|
|
|
15333 if (c == 0)
|
|
|
15334 {
|
|
|
15335 /* End of prefix, find out how many IDs there are. */
|
|
|
15336 for (i = 1; i < len; ++i)
|
|
|
15337 if (byts[n + i] != 0)
|
|
|
15338 break;
|
|
|
15339 curi[depth] += i - 1;
|
|
|
15340
|
|
455
|
15341 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
15342 if (c != 0)
|
|
351
|
15343 {
|
|
376
|
15344 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
701
|
15345 dump_word(slang, prefix, pat, dir, dumpflags,
|
|
455
|
15346 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
701
|
15347 : flags, lnum);
|
|
|
15348 if (lnum != 0)
|
|
|
15349 ++lnum;
|
|
351
|
15350 }
|
|
455
|
15351
|
|
|
15352 /* Check for prefix that matches the word when the
|
|
|
15353 * first letter is upper-case, but only if the prefix has
|
|
|
15354 * a condition. */
|
|
|
15355 if (has_word_up)
|
|
|
15356 {
|
|
|
15357 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
15358 TRUE);
|
|
|
15359 if (c != 0)
|
|
|
15360 {
|
|
|
15361 vim_strncpy(prefix + depth, word_up,
|
|
|
15362 MAXWLEN - depth - 1);
|
|
701
|
15363 dump_word(slang, prefix, pat, dir, dumpflags,
|
|
455
|
15364 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
701
|
15365 : flags, lnum);
|
|
|
15366 if (lnum != 0)
|
|
|
15367 ++lnum;
|
|
455
|
15368 }
|
|
|
15369 }
|
|
351
|
15370 }
|
|
|
15371 else
|
|
|
15372 {
|
|
|
15373 /* Normal char, go one level deeper. */
|
|
|
15374 prefix[depth++] = c;
|
|
|
15375 arridx[depth] = idxs[n];
|
|
|
15376 curi[depth] = 1;
|
|
|
15377 }
|
|
|
15378 }
|
|
|
15379 }
|
|
|
15380 }
|
|
|
15381
|
|
|
15382 return lnum;
|
|
|
15383 }
|
|
|
15384
|
|
498
|
15385 /*
|
|
626
|
15386 * Move "p" to the end of word "start".
|
|
|
15387 * Uses the spell-checking word characters.
|
|
498
|
15388 */
|
|
|
15389 char_u *
|
|
|
15390 spell_to_word_end(start, buf)
|
|
|
15391 char_u *start;
|
|
|
15392 buf_T *buf;
|
|
|
15393 {
|
|
|
15394 char_u *p = start;
|
|
|
15395
|
|
|
15396 while (*p != NUL && spell_iswordp(p, buf))
|
|
|
15397 mb_ptr_adv(p);
|
|
|
15398 return p;
|
|
|
15399 }
|
|
|
15400
|
|
475
|
15401 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
15402 /*
|
|
626
|
15403 * For Insert mode completion CTRL-X s:
|
|
|
15404 * Find start of the word in front of column "startcol".
|
|
|
15405 * We don't check if it is badly spelled, with completion we can only change
|
|
|
15406 * the word in front of the cursor.
|
|
475
|
15407 * Returns the column number of the word.
|
|
|
15408 */
|
|
|
15409 int
|
|
|
15410 spell_word_start(startcol)
|
|
|
15411 int startcol;
|
|
|
15412 {
|
|
|
15413 char_u *line;
|
|
|
15414 char_u *p;
|
|
|
15415 int col = 0;
|
|
|
15416
|
|
498
|
15417 if (no_spell_checking(curwin))
|
|
475
|
15418 return startcol;
|
|
|
15419
|
|
|
15420 /* Find a word character before "startcol". */
|
|
|
15421 line = ml_get_curline();
|
|
|
15422 for (p = line + startcol; p > line; )
|
|
|
15423 {
|
|
|
15424 mb_ptr_back(line, p);
|
|
|
15425 if (spell_iswordp_nmw(p))
|
|
|
15426 break;
|
|
|
15427 }
|
|
|
15428
|
|
|
15429 /* Go back to start of the word. */
|
|
|
15430 while (p > line)
|
|
|
15431 {
|
|
|
15432 col = p - line;
|
|
|
15433 mb_ptr_back(line, p);
|
|
|
15434 if (!spell_iswordp(p, curbuf))
|
|
|
15435 break;
|
|
|
15436 col = 0;
|
|
|
15437 }
|
|
|
15438
|
|
535
|
15439 return col;
|
|
|
15440 }
|
|
|
15441
|
|
|
15442 /*
|
|
|
15443 * Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
15444 * expand_spelling() is called. Therefore the ugly global variable.
|
|
|
15445 */
|
|
|
15446 static int spell_expand_need_cap;
|
|
|
15447
|
|
|
15448 void
|
|
|
15449 spell_expand_check_cap(col)
|
|
|
15450 colnr_T col;
|
|
|
15451 {
|
|
475
|
15452 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
15453 }
|
|
|
15454
|
|
|
15455 /*
|
|
|
15456 * Get list of spelling suggestions.
|
|
|
15457 * Used for Insert mode completion CTRL-X ?.
|
|
|
15458 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
15459 * allocated strings.
|
|
|
15460 */
|
|
|
15461 /*ARGSUSED*/
|
|
|
15462 int
|
|
|
15463 expand_spelling(lnum, col, pat, matchp)
|
|
|
15464 linenr_T lnum;
|
|
|
15465 int col;
|
|
|
15466 char_u *pat;
|
|
|
15467 char_u ***matchp;
|
|
|
15468 {
|
|
|
15469 garray_T ga;
|
|
|
15470
|
|
625
|
15471 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap, TRUE);
|
|
475
|
15472 *matchp = ga.ga_data;
|
|
|
15473 return ga.ga_len;
|
|
|
15474 }
|
|
|
15475 #endif
|
|
|
15476
|
|
236
|
15477 #endif /* FEAT_SYN_HL */
|
